Document NG0b2NggEp9mM8MDQMrG312mw

_r 10 diagnose cpilI i tumors at a time when they are still confined to pelvis and highly curable. i paliehts with advanced cancer, treatment approaches are also changing, v combination chemotherapy regimens are under evaluation in clinical (rials vhich issues such as high-dose chemotherapy and dose intensity are being lied. It is our hope that this issue will provide to the reader an overview of current approach to the management of this disease. In addition, we have d to provide a Framework 1u~hetp~tmderstand the forthcoming new theraitic developments that are based on an ever-increasing understanding of the logy of this complex disease. Rouekt F. O zols, M D , Ph D G iicsI E d ito r larlment of Medical Oncology . Chase Cancer Center 1 Uurholme Avenue ladelphia, PA 19111 iv t ' K c i A c r ; i WAN CANCER 0889-8588/92 $0.00 THE EPIDEMIOLOGY OF OVARIAN CANCER Mary B. Daly, MD, PhD Ovarian cancer is (lie leading cause of dealh from a gynecologic malignancy among women in the United Slates and accounts for approximately 12,500 deaths per year.61 Although the etiology and pathophysiology of ovarian carcinogenesis are unknown, there is a growing body of epidemiologic literature that has identified several potential contributing factors. This article explores these findings, with a particular emphasis on the genetic and reproductive features characteristic of the disease. GENERAL FEATURES Ovarian cancer is a disease of older age, with incidence rales increasing with each decade of life and peaking in the mid to late 70s" (Pig. 1). The rale of rise in incidence appears to accelerate coincident with the time of natural menopause.1"' w Malignant ovarian tumors do occur in childhood and represent 1% to 3% of all pediatric malignancies, but approximately two thirds of these are of the germ cell type.3- ,7- ' Age-adjusted rates for ovarian cancer have remained relatively stable over the last two decades (Pig. 2), although mortality rates have climbed as a result of the aging of the population, and possibly owing to changing reproductive patterns.16 Significant geographic and ethnic variations in rates have been observed. Kates are highest for while women in the industrialized countries of northern and western Europe and North America, and lowest in India and Asia.5* The wide geographic variation observed is roughly correlated with average family size.1* W ithin the United Stales, incidence rates for black, I lispanic, and American Indian women are approximately 40% lower than those for w hile wom en,61 and among whites, rates are highest for Jewish From the Cancer Control Science Program, Fox Chase Cancer Center, Philadelphia, Pennsylvania 1IEMATOI.OCY/ONCOLOGY CLINICS OF NORTH AMERICA volum i: 6 nummi: 4 august 1992 729 c o - v j- r c 730 , DAI 1 1973 1977 1978 1981 age Figure 1. Ovarian cancer rales by age lor 1973-1977 and 1978-1981. (Data Irom Ihe SEER program: Cancer Incidence and Modality in the United States. 1973-1981. US Depadment of Health and Human Services, 1984.) w o m e n 1" 11 These ethnic differences in rates may be partially explained by variations in reproductive factors. ENVIRONMENTAL FEATURES Both the observed international and ethnic variations, and the shift in risk to that of the host country among migrant populations," suggest a role for an environmental component such as diet in the etiology of ovarian cancer. Numerous case control studies, however, have failed to show an association between either weight or dietary fat intake and the risk for ovarian can cer.' 11 " M A positive association between ovarian cancer risk and lactose consumption has, however, been found in a number of studies. Cramer" correlated ovarian cancer incidence with both per capita milk consumption and lactase persistence worldwide and found both to be significantly positively associated. A case-control study conducted in 10 Boston-area hospitals dem onstrated both higher levels of lactose consumption and lower levels of galactose converting enzymes in cases than in controls.16 27 Although these data are very preliminary, there is theoretic support from the observation that women who suffer from galactosemia, an inborn error of metabolism resulting in elevated levels of serum galactose, often develop premature ovarian failure and hyper gonadotropic hypogonadism." A series of animal studies has also documented ovarian failure in rodents fed a high galactose diet.65 16 -i 14 - 12 - 10 oo o oo 8 - & UO)) 6 - 55 4- all tmalas while tamales biadi tamales 0 -------- 1---------' ---------1-------- >---------1---------1-------- 1------------------1-------- 1---------1---------1------- 1-- 1972 1974 1976 1978 1980 1982 1984 1986 year Figure 2. Ageadjusted ovarian cancer incidence by year and race lor 1973-1986. (Da(a Irom the SEER program: Cancer Statistics Review, 1973-1986. US Department ot Health and Human Services, 1989.) Other environmental factors that appear to be unrelated to risk for ovarian cancer are occupation, alcohol, and smoking history." " The noted increased incidence among women of higher socioeconomic status most likely reflects the combined effects of reproductive history and differential access to health care."1 O f all the environmental factors implicated as causative agents in ovarian cancer, none has generated as much interest as talc and asbestos particles. Tal. is a specific hydrous magnesium silicate that is chemically related to asbestos Asbestos fibers are often found as contaminants of talc consumer products 1 Interest in these compounds as causative agents for ovarian cancer stems from the recognition of asbestos as a cause of pleural and peritoneal mesothelioma a disease that shares a common embryologic origin with ovarian cancer. I lenderson et al" first reported finding talc particles in ovarian tissue obtaino from nine women, three of whom had ovarian cancer. Cosmetic talc is though to gain access to the ovaries by absorption through the vagina or cervix. In case-control study in Boston, Cramer et al' reported a relative risk of 1.`) among individuals who used talc directly on the perineum or on sanitary pail: Whittemore et al" also showed a weak association between talc use and risl for ovarian cancer, but there was no dose-response relationship. Allhoitgl these data are provocative, they offer only indirect evidence for an associalioi between these products and ovarian cancer. With the removal of talc froi most cosmetic preparations over the last decade, the issue may remain unr< solved. Likewise, with the introduction of widespread vaccination among childn 732 l/Al.Y t against mumps, the causative role of this virus may never be clarified. First thought of as a potential eliologic agent for ovarian cancer because of its known association with clinical orchitis in males, mumps also belongs to the para myxovirus family, which is associated with a number of rare but fatal chronic diseases.4'T w o case-control studies have found a lower self-reported history of mumps infection among cases'1 however, when scrum antibody levels were measured, they were similar for cases and controls, suggesting that ovarian cancer patients with mumps were more likely to have experienced subclinical disease, which may have led to subsequent chronic damage to the ovaries, depletion of oocytes, and alterations in gonadotropin levels.4' Finally, there is evidence from the survivors of the atomic bomb experience that high doses of radiation increase the risk for ovarian cancer, particularly among those women who were less than 20 years of age at the time of exposure, with a latent period of approximately 20 years.4* GENETIC FACTORS Although the recognition of a hereditary component to ovarian cancer is very recent, heredity is emerging as one of the most important aspects of (he disease. The number of known multicase ovarian cancer families has increased several fold over the last 10 years. Extensive pedigree analysis by Lynch et al" 41 has led to the recognition of three patterns of familial ovarian cancer: (1) sitespecific ovarian cancer only; (2) families in whom both breast and ovarian cancer predominate; and (3) the cancer family syndrome characterized by multiple cases of ovarian, breast, colon, and uterine cancer. Through careful analysis of these pedigrees, an autosomal dominant mode of inheritance with variable penetrance is postulated. A subset of approximately 5% to 10% of all ovarian cancers are thought to fit this genetic pattern." The most common relationship reported among cases is sibship, followed by mother-daughter.11 M Among these pedigrees, there is a significant reduction in risk for breast/ ovarian cancer for relatives as their levels of coancestry decrease." Other characteristics of familial ovarian cancer include an earlier age at onset and evidence of both paternal and maternal transmission41- 4>; however, there does not appear to be a significant difference in stage at presentation, histologic type or grade, dr bilaterally between familial and sporadic cases." " The familial aggregation of ovarian cancer demonstrated by these pedigree analyses has been confirmed in numerous case-control studies. Data from the Alberta Cancer Registry have documented a statistically significant increase in affected family members among cases of ovarian cancer." Tzonou et al4' found a familial aggregation of ovarian cancer among women with breast, ovarian, and uterine cancer in Greece. In japan, M ori et al41 found a family history of breast, uterine, and ovarian cancer to be more common in cases than in controls. Using the mullicentcr data base assembled by the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute, Scluldkraut and Thompson" demonstrated an odds ratio for ovarian cancer of 3.6 in first-degree relatives, and 2.9 in second-degree relatives, as compared with relatives of controls. The excess of affected relatives was seen in both maternal and paternal lines O f interest is the fact that neither the SEER data nor a case-control study conducted by I larlow el aP` in Washington state could demonstrate an increased risk of ovarian cancer for relatives of women with borderline tumors, suggesting a different carcinogenic pathway for this subtype of ovarian cancer. The familial aggregation of breast and ovarian cancer is supported by data from the Cancer and Steroid Hormone (C ASH ) Study,' a population based case control study conducted by the Centers for Disease Control, in which Thompson and Schildkraut*' were able to demonstrate an elevated risk of ' developing breast cancer among first degree relatives of women with ovarian ' cancer. D ually, lending further support to a proposed genetic mechanism for ovarian cancer is the finding of clustering of both breast and ovarian cancer in the patients themselves. Data from the Birmingham Cancer Registry in England found a significant excess of second primaries in premenopausal women with either breast or ovarian cancer. Prior and Waterhouse" and Greggi el al" report a fivefold increase in rates of breast cancer occurring as a second primary among familial ovarian cancer cases, as compared with sporadic ovarian cases. Although the precise genetic defect(s) associated with familial ovarian cancer has not yet been identified, a number of genetic alterations and associations have been described. Several investigators have noted decreased levels of a-L-fucosidase among women w ith ovarian cancer and their firstdegree relatives,' 40 " suggesting close linkage of the genetic loci that control a-L-fucosidase activity and ovarian carcinogenesis. Cytogenetic characterization of surgical specimens from ovarian tumors reveals extensive and complex structural chromosome abnormalities involving chromosomes 1, 3, 6," " chro mosome 14," and chromosome 17.' 1 " Marks et a l reported overexpression of the p53 gene, a gene recently linked to familial breast cancer,44 in 50% of 107 tumor samples. This growing body of evidence points to a mullistep model of ovarian carcinogenesis in which a complex series of genetic events result in the initiation, development, progression, and metastatic potential of the disease. REPRODUCTIVE FACTORS Because the ovary is the major source of reproductive hormones in women, it is only natural that a great deal of attention has been paid to the role of reproductive factors in trying to understand the etiology of the disease. The one overwhelm ing observation that has emerged from almost every study in the literature is the relationship between pregnancy experience and risk for ovarian cancer. W omen who develop ovarian cancer are significantly more likely to be nulliparous and to have had fewer pregnancies.*' 14 44" " The relative risk associated with nulliparity ranged in several case-control studies from 1.5 to 3.2.1- * " " 44 **74 Furthermore, there is a clear dose-response relationship associated with pregnancy history, with most studies reporting a sternly decrease in risk with increased parity.*u' M a-4*-'* When controlled for number of pregnancies, however, age at first pregnancy does not appear to. affect risk.4- **" 74 The role of oral contraceptive pills (O C I's), which mimic some of the physiologic changes induced by pregnancy, has also been carefully studied with regard to ovarian cancer. OCRs were first introduced in the United Stales in the 1960s. Most formulations include estrogen, progesterone, or a combination of the two, and they have been shown to suppress ovulation W omen with a history of OCR use have an overall reduction in risk of ovarian cancer of 40% ." The largest case-control study to document this effect was that conducted by the Centers for Disease Control, using the population-based cancer registry data of the eight SEER regions, and age-matched and area ol resilience-matched community controls. Ever-users of OCRs had a relative lisk for ovarian cancer of 0.6. Risk reduction was apparent with as little as 3 months of use, increased in magnitude with duration of use, and persisted for as long , D i 1>.\l as 10 years after discontinuation of use.' The observed reduction in risk applied to all histologic subtypes, including borderline tumors, and was not dependent on age at use, menopausal status, or measures of fertility or parity, although nulliparous women experienced a greater decrease in risk titan did parous women ' A summary of similar findings from a number of case-control studies is-sluuunJu. Table 1, Note that the majority of these studies failed to show a relationship between replacement hormone therapy for menopausal symptoms and risk for ovarian cancer.' * " ** One exception is a case-control study in Washington slate in which replacement hormone use was associated with a relative risk of 3.1 for the histologic subset of endometrioid tumors.71 A similar trend was noted in a study in Boston by Cramer et a l,1' and in a case-control study in Washington, D C ." Endometrioid tumors of the ovary are similar in appearance to tumors of the endometrium and may share common risk factors. No clear-cut relationship between age at menarche or age at menopause and risk for ovarian cancer has emerged from these studies. Another series of studies investigated the role of sterilization procedures on subsequent risk of ovarian cancer. Irwin et al" examined the data from the Centers for Disease Control study to compare histories of tubal sterilization and hysterectomy with or without unilateral oophorectomy between cases and controls. A significantly reduced risk for ovarian cancer was seen in women who had undergone these procedures, particularly if they were performed at younger ages. Data from a case-control study in Washington, DC, demonstrated a 30% risk reduction in women undergoing hysterectomy with preservation of the ovaries." Similarly, a case-control study performed in japan, where inci dence of ovarian cancer is low, found that a significantly lower proportion of ovarian cancer patients than controls had undergone permanent sterilization." O f note is the observation by Cattanach of decreased estrogen excretion in women who had undergone tubal ligation, a finding that he suggested may be related to localized procedure-induced vascular changes.7 Two alternative hypotheses have been advanced to explain the findings related to reproductive history that were described previously. One theory, first proposed by Falhalla,70 contends that constant, uninterrupted ovulation leads to an increased risk for ovarian cancer by virtue of the repetitive trauma to the ovarian epithelium and resultant cellular proliferation associated with the ovulatory cycle, the potential for entrapment and transformation of epithe lial inclusion cysts, or the repetitive exposure of the epithelium to steroid-rich follicular fluid or pituitary gonadotropins '* 7` Thus, events such as pregnancy and O C I' use, which suppress ovulation, would decrease this risk. A number of investigators have actually calculated measures of duration of ovulation by subtracting anovulatory time during pregnancy and lactation and while taking OCI's from time at risk for ovulation, and found a significant correlation Table 1. RELATIVE RISK OF OVARIAN CANCER BV ORAL CONTRACEPTIVE USE Cases/Controls Reference (No.) Users Nonusers (95% Confidence Interval) CDC Weiss" Rosenberg1* Wu" Cramer" Harlow" LaVecchia" 197/439 21/112 33/136 188/299 34/144 66/194 39/178 2335/3867 207/552 187/539 619/1011 48/139 82/193 367/1104 0 6 (0.4-0.9) 0.57 (0 05-1.92) 0.6 (0.4-09) 0 74 (0 52-1 06) 0.38 (0 15-0.96) 0.7 (0.4 --1.2) 0.68 (0 48-0.97) I I between ovarian cancer risk and duration of ovulation.6 76 Support for this theory also comes from a little-known experiment in which epithelial ovarian cancers were induced in hens by using prolonged fluorescent lighting to indue; ovulation and maximize egg production.75 The alternative hypothesis contends that nulliparity, rather than failing to offer protection against ovarian cancer, is actually a manifestation of an underlying reproductive disorder, 'which also predisposes to'the development' of the malignancy. A t least four case-control studies have found a significantly increased prevalence of infertility among cases as compared with controls, whether infertility is measured by self-reporting of difficulty in conceiving," * ?` physician-diagnosed infertility,7* or number of years of unprotected intercourse without conception.*6 *' Incidence of abortion, stillbirth, and congenital malfor mation do not appear, however, to be related to ovarian cancer risk.*6 " A role for altered pituitary gonadotropin levels, which may be the basis for subferiility in many women, is supported by the epidemiologic observation that incidence rates for ovarian cancer accelerate around the time of menopause, when gonadotropin levels also show a sharp rise.23- "* It is also consistent with the previously described apparent protection afforded by O C I3 use, which is associated with suppression of pituitary gonadotropins.* In fact, a case-control study performed by H arlow et al27 described an interactive effect between high lactose intake and the protection afforded by oral contraceptive use They propose that it is in the setting of a hypergonadotropic state induced by high galactose levels that OCPs exert their protective, gonadotropin-lowering effect. The altered gonadotropin theory is also consistent with the observed reduction in risk following sterilization procedures, which have been shown to alter ovarian blood flow and hormonal activity.1 " There are also some data from murine models in which elevated gonadotropin levels are associated with ovarian tumor promotion, but the applicability of this model to human carci nogenesis is questionable.61 SUMMARY The epidemiologic data collected to date have provided some important and provocative clues as to the etiology of ovarian cancer. The recognition of familial clustering of this disease has led to exciting advances in understanding the genetics involved. The contribution of endocrine factors has been well documented in the epidemiologic literature, leading to important insights into the process of carcinogenesis. Clearly, additional information is needed to further explain the interplay of genetic, physiologic, and life style factors, to lead to a better understanding of the disease, and ultimately to a means of prevention and control. References I. Alvarez-Sanchez F, Segal S, Brache V, et al: Pituitary ovarian function after tubal ligation. Ferlil Steril 36:606, 1981 2 Annegers J, Strom II, Decker D: Ovarian cancer: Incidence and case-control study Cancer 43:723, 1979 3. Anleby S, Mor-Josef S, Schenker J: Ovarian cancer in the young. Fur ) Gynaecol Oncol 6:41, 1985 4. lierai V, Fraser P, Chilvers C: Does pregnancy protect against ovarian cancer? Lancet 1:1083, 1978 5. Barlow J, DiCioccio R, Dillard P, et al: Frequency of an allele for low activity of 736 1>A| u-L-fucosidase in sera: Possible increase in epithelial ovarian cancer patients. / Natl Cancer Inst 67.1005, 1981 6. Casagrande ], Pike M, Ross R, et al: "Incessant ovulation" and ovarian cancer, l ancet 2:170, 1979 7. Cattanach ): Oestrogen deficiency after tubal ligation. Lancet 1:847, 1985 8. Centers for Disease Control Cancer and Steroid Hormone Study: Oral contraceptive use and the risk of ovarian cancer. JAMA 249:1596, 1983 9 Centers for Disease Control: The reduction in risk of ovarian cancer associated with oral-contraceptive use. N Engl J Med 316:650, 1987 10. Crainer D, Welch W, Scully R: Ovarian cancer and talc: A case-control study. Cancer 50:372, 1982 11. Cramer D, Norio W, Hutchinson G, el al: Factors affecting the association of oral contraceptives and ovarian cancer. N Engl J Med 307:1047, 1982 12. Cramer D, Hutchinson G, Welch W, et al: Determinants of ovarian cancer risk. I. Reproductive experiences and family history. J Natl Cancer Inst 71:711, 1983 13. Cramer D, Welch W, Cassells S, et al: Mumps, menarche, menopause, and ovarian cancer. Am | Obstet Gynecol 147:1, 1983 14. Cramer D, Welch W: Determinants of ovarian cancer risk. II. Inferences regarding pathogenesis. J Nall Cancer Inst 71:717, 1983 15. Cramer DW: Lactase persistence and milk consumption as determinants of ovarian cancer risk Am J Epidemiol 130:904, 1989 16. Cramer DW, Willett WC, Bell DA, el al: Galactose consumption and metabolism in relation to the risk of ovarian cancer. Lancet 2:66, 1989 17 Cronen P, Nagaraj H: Ovarian tumors in children. South Med J 81:464, 1988 18. Demopoulos R, Seltzer V, Dubin N, et al: The association of parity and marital status with the development of ovarian carcinoma: Clinical implications. Obstet Gynecol 54 150, 1979 19. Diamond M, Baxter I, Peerman C: Occurrence of ovarian malignancy in childhood and adolescence: A community-wide evaluation. Obstet Gynecol 71:858, 1988 20. Fathalla M: Factors in the causation and incidence of ovarian cancer. Obstet Gynecol Surv 27:751, 1972 21 Ferrell R, Anderson D, Chidambaram A, et al: A genetic linkage study of familial breast-ovarian cancer. Cancer Genet Cytogenet 38:241, 1989 22. Franchesi S, LaVecchia C, Mangioni C: Familial ovarian cancer: Eight more families. Gynecol Oncol 13:31, 1982 23. Franceschi S: Reproductive factors and cancer of the breast, ovary and endometrium. Eur J Cancer Clin Oncol 25:1933, 1989 24 Greene M, Clark J, Blayney D: The epidemiology of ovarian cancer. Semin Oncol 11:209, 1984 25. Greggi S, Genuardi M, Benedetli-Panci P, et al: Analysis of 138 consecutive ovarian cancer patients: Incidence and characteristics of familial cases. Gynecol Oncol 39:300, 1990 26 Harlow BL, Weiss NS, Roth GJ, el al: Case-control study of borderline ovarian tumors: Reproductive history and exposure to exogenous female hormones. Cancer Res 48:5849, 1988 27 I larlow B, Cramer D, Geller J, el al: The influence of lactose consumption on the association of oral contraceptive use and ovarian cancer risk. Am J Epidemiol 134:445, 1991 28 llartge P, Hoover R, McGowan L, el al: Menopause and ovarian cancer. Am J Epidemiol 127:990, 1988 29. Ilcintz P, I lacker N, Lagasse L: Epidemiology and etiology of ovarian cancer: A review. Obstet Gynecol 66:127, 1985 30. Henderson W, Hamilton T, Griffiths K: Talc in normal and malignant ovarian tissue. Lancet 1:499, 1979 31. Hildreth N, Kelsey j, LiVolsi V, et al: An epidemiologic study of epithelial carcinoma of the ovary. Am J Epidemiol 114:398, 1981 32. Irwin K, Weiss N, Lee N, et al: Tubal sterilization, hysterectomy, and the subsequent occurrence of epithelial ovarian cancer. Am J Epidemiol 134:362, 1991 33. July O, l.ilienfeld A, Diamond E, el al: An epidemiologic study l. e relationship oi reproductive experience to cancer of the ovary. Am J Epidemiol 99:190, 197-1 34. Koch M, Jenkins II, Gaedke 11: Risk factors of ovarian cancer of epithelial origin: A case control study. Cancer Detect Prev 13:131, 1988 35. Koch M, Gaedke II, Jenkins II: Family history of ovarian cancer patients: A casecontrol study. Inti J Epidemiol 18:782, 1989 36. LaVccchia C, Franchcschi S, Liberati A, et al: The clinical relevance of the epide miology of.ovarian cancer. Eur J Cancer Clin Oncol 20:175, 1984 37. LaVecchia C, Decarli A, Fasoli M, et al: Oral contraceptives and cancers'ofthe-breasi and of the female genital tract. Interim results from a case-control study Ur J Cance-i 54:311, 1986 38. I.esher L, McGowan L, liartge P, et al: Age at first birth and risk of epithelial ovariai. cancer. J Natl Cancer Inst 74:1361, 1985 39. Lynch H, Harris R, Guirgis II, et al: Familial association of breast/ovarian carcinoma Cancer 41:1543, 1978 40. Lynch M, Schuelke G, Wells I, et al: I iereditary ovarian carcinoma: Biomarker studies Cancer 55:410, 1985 41 Lynch II, Uewtra C, Lynch J: Familial ovarian carcinoma; Clinical nuances Am Med 81:1073, 1986 42. Lynch H, Fitzsimmons M, Conway T, el al: Hereditary carcinoma of the ovary an. associated cancers: A study of two families. Gynecol Oncol 36:48, 1990 43. Lynch II, Watson P, Bewtra C-> et al: Hereditary ovarian cancer: Heterogeneity in ag al diagnosis. Cancer 67:1460, 1991 44. Malkin D, Li F, Strong L, et al: Germ line p53 mutations in a familial syndrome >. breast cancer, sarcomas, and other neoplasms. Science 250:1233, 1990 45. Marks J, Davidoff A, Kerns B, el al: Overexpression and mutation of p53 in epithcli. ovarian cancer. Cancer Res 51:2979, 1991 46 McGowan L, Parent L, Lednar W, et al: The woman at risk for developing ovaria cancer. Gynecol Oncol 7:325, 1979 47. Menczer j, Modan M, Ranon L: Possible role of mumps virus in the etiology < ovarian cancer. Cancer 43:1375, 1979 48. Mori M, liarabuchi I, Miyake H, et al: Reproductive, genetic, and dietary risk facto, for ovarian cancer. Am J Epidemiol 128:771, 1988 49. Nasca P, Greenwald P, Chorost S, et al: Epidemiologic case-control study of ovari.. cancer and reproductive factors. Am J Epidemiol 119:705, 1984 50. Newhouse M, Pearson R, Fullerton J, et al: A case control study of carcinoma of tl ovary. Br J Preventive Social Med 31:148, 1977 51. Okamota A, Sameshima V, Yokoyama S, et al: Frequent allelic losses and mulalioi of the p53 gene in human ovarian cancer. Cancer Res 51:5171, 1991 52. Panani A, Ferti-Passantonopoulou A: Common marker chromosomes in ovari. cancer. Cancer Genet Cylogenel 16:65, 1985 53. Parkin M: Cancers of the breast, endometrium and ovary: Geographic correlation Eur J Cancer Clin Oncol 25:1917, 1989 54. Piver S, Metllin C, Tsukada Y, el al: Familial ovarian cancer registry. Obstet Gyms 64:195, 1984 55. Piver S, Baker T, Piedmont M, et al: Epidemiology and etiology of ovarian cam. Semin Oncol 18:177, 1991 56. Prentice R, Thomas D: On the epidemiology of oral contraceptives and disease. A Cancer Res 49:285, 1987 57. Prior P, Waterhouse JAH: Multiple primary cancers of the breast and ovary. It. Cancer 44:628, 1981 58. Radwanska E, Headley S, Dmowski P: Evaluation of ovarian function after lul sterilization. J Reprod Med 27:376, 1982 59. Rosenberg L, Shapiro S, Slone D, et al: Epithelial ovarian cancer and combinati oral contraceptives. JAMA 247:3210, 1982 60. Schiidkraut J, Thompson W: Familial ovarian cancer: A population-based case-coni study. A m ) Epidemiol 128:456, 1988 61. Shimkin II: Hormones and neoplasia. Cancer 1:161, 196(1 62. Silverberg E, Boring C, Squires T: Cancer statistics, 1990. Cancer 40:9, 1990 ,7*8 V 63. Sukhoi S: Ethnic-related incidence of ovarian carcinoma in New York Cily. Isr \ Med Sci 14:363, 1978 64 Sladel B: The etiology and prevention of ovarian cancer. Am J Obslct Gynecol 123:772. 1975 65. Swartz W|, Madison DR: Galactose inhibition of ovulation in mice. Fertil Sleril 49:522, 1988 66 Szamhorski |r Czerwinski W^Gadomska II, et al: Case control study of high-risk factors in ovarian carcinomas. Gynecol Oncol 11:8, 1981 67. Thompson WD, Schildkraul JM: Family history of gynaecological cancers: Relation ships to the incidence of breast cancer prior to age 55. Int | Epidemiol 20:595, 1991 68 Tokuoka S, Kawai K. Shimizu Y. el al: Malignant and benign ovarian neoplasms among atomic bomb survivors, Hiroshima and Nagasaki, 1950-80. | Nall Cancer Inst 79.47, 1987 69 Tzonou A, Day N, Trichopoulos D: The epidemiology of ovarian cancer in Greece: A case-control study. Eur J Cancer Clin Oncol 20:1045, 1984 70 Wake N, I Ireshchyshyn M, Piver S, el al: Specific cytogenetic changes in ovarian cancer involving chromosomes 6 and 14. Cancer Res 40:4512, 1980 71, Weiss N, l.yon |, Krishnamurthy S, et al: Noncontraceptive estrogen use and the occurrence of ovarian cancer. J Natl Cancer Inst 68:95, 1982 72 Whang-Peng |, Knutsen T, Douglass E, et al: Cytogenetic studies in ovarian cancer. Cancer Genet Cylogenel 11:91, 1984 73. Whittemore A, Wu M, Paffenbarger R, el al: Personal and environmental character istics related to epithelial ovarian cancer. II. Exposures to talcum powder, tobacco, alcohol, and coffee. Am J Epidemiol 128:1228. 1988 74. Whittemore A, Wu M. Paflenbarger R, et al: Epithelial ovarian cancer and the ability to conceive. Cancer Res 49:4047, 1989 75. Wilson j: Adenocarcinomata in hens kept in a constant environment. Poult Sci 37:1253, 1958 76. Wu M, Whittemore A, Paflenbarger R, et al: Personal and environmental character istics related to epithelial ovarian cancer. I. Reproductive and menstrual events and oral contraceptive use. Am J Epidemiol 128:1216, 1988 77. Yancik R, Gloeckler Ries L, Yates J: Ovarian cancer in the elderly: An analysis of surveillance, epidemiology, and end results program data. Am j Obstet Gynecol 154:639. 1986 Address reprint requests to Mary B Daly. MD, PhD Fox Chase Cancer Center 510 Township Line Road Cheltenham, PA 19012 OVAKIAN CANCER PATHOLOGY EPITHELIAL TUMO R obert H . Y oung, M D , and Robert E. Scull) This article reviews the pathology of epithelial tumors of the ovar\ particular emphasis on newer information in the literature. Most discussion is devoted to the most common category of malignant < tumors, which have usually been referred to in recent years as "c<. epithelial" or "surface epithelial" cancers.*6 47 These are designated " epithelial-stromal tumors" in the revised classification of ovarian tumor: W orld Health Organization (W H O ) (Table I) that is currently being fin Also discussed briefly is a recently described highly malignant epithelial that does not appear to belong in this category, but that is import oncologists to recognize, the small cell carcinoma.14 Although almost . types of surface epithelial-stromal tumor include those of borderline nancy, those of endom etrioid,7 " clear cell,4 *" and transitional cell'1 are rare and w ill not be discussed. Some particularly important probl differential diagnosis of epithelial cancers of which the clinician sin aware w ill be mentioned; these problems have been discussed in great elsewhere, along with others." Surface epithelial-stromal tumors account for approximately two II all ovarian neoplasms, and their malignant forms for about 90% of cancers.46 In the designation given to a neoplasm in this category, criteria are considered. The two most important are the cell type, i.e., mucinous, and so o n -- and whether the tumor is benign, of so-called bo. malignancy, or frankly malignant. The final two criteria are architectui pertains to whether the tumor is exophytic, endophytic, or both exophytic growth is present, the word "surface" is added to the desij e g., serous surface papillary carcinoma. The second relates to the proportions of the epithelial and stromal components of the neoplasi, surface epithelial-stromal tumors are predominantly epithelial, with minor portion derived from the ovarian stroma; when a tumor has a p From the Department of Pathology, Harvard Medical School; and James Horn Pathology Laboratories, Massachusetts General Hospital, Boston, Massadn HEMATOLOCY/ONCOLOCY CLINICS OF NORTH AMERICA VOLUME 6 NUMBER i AUGUST 1992 Supplantant lo u a Included in subscription Horm on Research (ISSN 03 01-0 163 ), Voi. 3 8 , Suppl. 2 (1 992) Editor: J. Girard. Basel V III International Symposium on Endocrinology and Development: Multiple Endocrine Diseases: Growth Hormone Action: Editor: A. Attanasio, Bad Homburg Intersexuality Telfs, October 11-12, 1991 This publication presents selected contributions from three work shops relating to endocrinology and development. The first sec tion aims to help clinicians and basic scientists understand the pathogenesis and clinical features of different polyendocrme syn dromes. A comprehensive review of the anabolic and metabolic ef fects of growth hormone action is provided in the second section. Finally, the last section is devoted to the interesting mechanisms of sexual differentiation in man. Readers will find that the publica tion offers challenging perspec tives and suggests exciting new directions for further research. Content* In tro d u cto ry L ac turm Lefforth. D . Shemer. J , Roberts. C T . Jr Evolutionary Origin* of Intercellular Com muncation System j Implications for Mam maiian Biology Altd tip U tn d o crin a D haaeea Scherbaum. IV A Introduction Riley, W J . Autoimm une Polyglandular Syn dromes Calender. A . Schuffenecker. /., Lenoir. G M The Genetics o f Multiple En docrine Neoplasia (MEN) Padberg. B C . HoM. K ; Schrder. S Pathology o f Multiple Endocrine Neoplasias 2 A and 2B A Review Raue.F Zink. A Clinical Features o f Mulliple Endocrine Neoplasia Type la n d Type 2 A n a b o lic an d M e ta b o lic i f f e c u o f G ro w th Horm one A ctio n Saenger. P Introduction Clemmons. D R . Underwood. L E. : Role of Insulin-Like G rowth Factors and G rowth Hormone in Reversing C a ia b o k States Germer, J M G rowth Hormone A ctions on Fat Distnbu*IV Don and Metabolism Haymond. M W ; Horber. F F The Effects o f Human G rowth Hormone and Prednisone on Whole Body Estimates o f Proietn Metabolism Capno. S Boolware. D Tamborlane. V Growth Hormone and Insulin Interactions In te ra e a u a U ty Suonenko. P C : Introduction Swaab. D F . Gooren. L J G ; Holman. M A Gen der and Sexual Orientation in Relation to Hypothalamic Structures Hawkins. J R Genetic Determinants of Testis Develop ment in Normal and Abnormal In d iq u a is Vouvlamen. R : Differentiation o f the Fetal Gonad Josso. N.: A nti-M uileni Hor mone and Sertoli Cell Function Hughes. I A Williams. D M ; Batch, J A et al Male Pseudohermaphroditism: Clincial Manage ment. Diagnosis and Treatment Aarskog. D : Syndromes and Genital Dysmorpho lo g y F ield o f In te re s t Endocrinology K19X-60 Im e m e tlo n e l S ym posium on E ndocrinology and D evelopm ent, V III Multiple Endocrine Diseases G rowth Hormone A ction Inter sexuality Teffs. October 11- 12. 1991 Hormone Research . V d 38.Suppl 2(1 99 2) Editor Ananasio. A (Bad Homburg) IV 8 8 p .41 fig , 14 tab . soft cover 1992 SFr 46 - / D M 5 5 - / C 2 3 0 0 /U S S 3 7 .0 0 Prices subject to change DM price for FRG. price for UK. US $ price for USA only ISBN 3 -8 0 5 5 - 5 7 4 0 -X KARGER Review Gynecol Obstet Invesi 1993:35:129-1.15 Johannes Diel I KlaiLs Marziisch Department o f O hsieirics and Gynecology. University o f Tbingen School o f M edicine, Tbingen. FRG Ovarian Surface Epithelium and Human Ovarian Cancer Keywords Ovarian carcinoma Surface epithelium Ovulation Oncogenes Abstract Approximately 80-90% o f adult ovarian cancers are assumed to originate from ovarian surface cells. The morphology o f the ovarian surface epithelium changes constantly, exhibiting features such as crypts, inclusion cysts, villous processes and different forms o f miillerian epithelium. The unique nature u f ovarian surface features and their absence in the immediately adjacent perito neal mesothelium suggest that local factors may play an im portant part in m odifying the growth and morphology o f the ovarian surface epithelium Recent studies tend to emphasize oncogene activation, along w ith concom i tant cytogenetic changes, in the development o f ovarian cancer. This review focuses on the roles o f exogenous and endogenous factors and ov ulation in the pathogenesis o f ovarian cancer. Introduction Ovarian cancer is a disease that is s till largely uname nable to curative treatment. Although accounting for only 4 % o f all cancers, ovarian cancer causes an estimated 6% o f deaths from all types o f malignancies in women |l) . Every year, more women die in the industrialized nations from this cancer than from all types o f uterine cancer, which is three times as frequent [2|. The main reason for this discouraging situation is the fact that over 70% o f all ovarian cancers are first discovered at a stage that permits practically no curative therapy and no effective screening (.1). Less than 25% o f women w ith ovarian cancer survive the initial diagnosis by more than 5 years |4], Epithelial tumors account for 80-90% o f all ovarian cancers and are presumed to originate front the surface epithelium (5-7). They rarely appear before the age o f 10; .their frequency then increases exponentially until the age o f 40. and the highest incidence is in women in their early 60s. Dynamic Morphology ol the Ovarian Surfaca Epithelium The serosa-like cell layer o f the ovary develops front the celomic epithelium. D uring the fetal period, the sur face epithelium is subject to constant morphological changes. A fter approximately the 10th fetal week, a m u lti layered epithelium w ith occasional papillary projections and a basement membrane separating it from the neigh boring stroma develops out o f a single-layered epithelium w ith no basement membrane (8). The increase in prolifer alive a ctivity is associated w ith a high nucleo-cytoplasmic ratio, nuclear polymorphism, prominence o f nucleoli, and abundance o f free and bound ribosomes (9. 10). T he sepa ration o f the superficial epithelium from the cortical stroma by a basement membrane is not yet complete, since islands and strands o f surface epithelium are present in the cortex during the entire fetal and perinatal period (5. I0 | For this reason, conical inclusion cysts, which h> Kprfi*fdJune . 1992 Accepted: Scpirm he. Ml. 1992 JtW inim I H ill. M il Departm ent o f f Rviri n n and (<>i*ca*lofy S d llrtd stirasse 4 0-W -74M1 I u h in jtrn lf K (il o 1991 S X j r r * *.i ; R i OJ' l- I U W . l'n ill- O i. ' J * I/O stratification and papillary projections |27], It remains to be seen whether these morphological changes, which coin cide w ith the peak age o f ovarian carcinoma [28]. arc compensatory to cortical shrinkage o r secondary to other stim uli |29). Fifl. 1. O varian surface e pith eliu m w ilh C ilia . X - 3.420 Fig. 2. C>ioplasm tacu o lc* ( * ) and m acropinocxlosis (a rrow ) n an ovarian surface-cell. X - 14.400. iheir nature have (he same proliferative activity as the superficial epithelium, arc also found in the fclal ovary til] The surface morphology o f the ovary changes con stantly during the whole course o f life (9|. A single-layered ovarian surface epithelium with a regular cuboidal cell structure is usually found resting upon the lunica albugi nea and the stroma below it. Since this epithelium pos sesses the characteristics o f the mullerian duct, endome trioid and endoccrvical cell forms can also develop from it |5. 8|. At various points, the ovarian surface cells exhibit not only m icrovilli but also numerous cilia (fig. I ), whose surfacedistribution is regionally diverse in one and the same ovary and whose emergence is both cyclical and age related (12). Frequent caveolae also indicate that the ovarian surface cells have high absorptive activity [l.J|. These caveolae can be regarded as morphological evi dence o f the endocytosis o f material from the abdominal cav ity (14] It is possible that the uptake o f substances by the ov arian surface cells is further aided by macropinocytosis. in which cytoplasmic evaginations entrap the mate rial to be incorporated, forming small vacuoles (fig 2). Numerous crypts are present on the surface o f the ovary . The epithelium at the base o f these crypts occasionally extends even deeper to form small cell nesls. which can give rise to inclusion cysts, probably due to sealing o ff at the surface epithelium. IT there is proliferation in the opposite direction, papillary excrescences form on the surface. Also biochemically detectable in the superficial epithelium o f the mature ovary is a characteristic pattern o f four polypeptides o f the intermediate filament cytokeratin. w hich is also present in M uller's tissue and in epithe lial ovarian cancers [15-17], It is noteworthy that M iel tinen et al. [18] found the intermediate filament prekera tin in normal superficial epithelium, in inclusion cysts and in epithelial ovarian tumors. Ullrastructurally, scat tered desmosome-tonofilament complexes (fig. 3) are found in the superficial epithelial cells o f the mature ova ry According to studies by Fcrcnzy and Roth [19], such complexes appear w ith predilection in endometrioid ovarian carcinomas and Brenner tumors: moreover, these tum or forms have very dense intermediate filaments. Blaustein [5. I I ] distinguishes two types o f inclusion cysts: an immature type with a prim itive celomic epithe lium that emerges during fetal development, as men tioned above, and a mature type that possibly forms around the time o f ovulation and in which the epithelium is quite variable. Cysiadcnomas and cystadcnocarcinomas arc believ ed to develop from inclusion cysts [20], and certain tum or markers have been found in both inclusion Exogenous Factor* F ig.1. Desm osom c-ionofilam enl complex in a human ovarian surface cell X - 16.400. cysts and serous ovarian tumors [2 11. Blaustein el al. [2 1] found carcinocmbryonic-anligcn-posilive cells in some inclusion cysts, which can be taken as a sign o f the im m a turity and undifferentiation o f certain cortical inclusion cysts. Elevated levels o f human chorionic gonadotropin and carcinoemhryonic antigen have been noted in the fluid o f some inclusion cysts, cystadenomas and cysladenocarcinomas [5|; this suggests that all three have the prim itive embryonic celomic epithelium as Iheir origin. It has also been reported that CA 125. an antigen associated with ovarian carcinoma and present in nearly all serous tumors o f the ovary, is also to be found in inclusion cysts, but not in normal ovarian surface cells (22. 23]. Inclusion cysts form an actively proliferating cell compartment that is responsive to hormonal stim ulation and is capable o f neoplastic change under certain conditions However, not only do inclusion cysts develop around the time o f ovula tion. but prominent papillae also develop - at least in ani mal experiments - al the site o f the postovulalory defect |24|. These excrescences can result, under both physiolog ical and pathological conditions, in the form ation o f superficial structures that resemble neoplastic lesions in most respects |2 5. 26]. Inclusion cysts derived from invag ination and entrapment o f ovarian surface cells in the cor tex are more frequently observed in the later reproductive years and after the menopause [27], Indirect evidence o f increased proliferative activity in the senescent ovarian surface epithelium is also provided by the presence o f cell Ovarian carcinoma exhibits marked differences in fre quency throughout the world. For example, it is three to seven times more common in western countries, such as the USA. Great Britain, and Scandinavia than in Japan [30]. The incidence o f ovarian cancer is very low in Asia, particularly in Japan, which has the lowest age-adjusted incidence rale (2.8/100.000) [31], T his lack o f uniform ity in incidence is most marked among women over the age o f 50. However, in women who emigrate from Japan to the USA. the risk o f contracting ovarian cancer increases above that o f their relatives who remain in Japan. In fact, the incidence among postmenopausal Japanese women living in the USA is practically the same as that o f post menopausal women born in the USA [32|. W ood ruff [3 3 1 believes that the greater incidence o f ovarian cancer noted in many industrialized countries 130. 34) is attributable to exogenous factors. Exogenous substances can ascend into the abdominal cavity via the genital tract in less than 30 m in [35], and there have been a few reports that the risk o f ovarian cancer is lower after hysterectomy [36. 37) and tubal ligation [38], but other studies have been unable to confirm this [39]. Talc granules have frequently been found in epithelial ovarian cancers. Cramer et al. [40] found that the risk o f ovarian cancer was three times higher among women who regularly dusted Iheir perinea or underclothing w ith talcum powder for cosmetic rea sons than among women who did not do so. Keal [41] found a tenfold increase in incidence o f ovarian cancer among women working in the asbestos industry, but Newhouse el al. [42] were unable to con firm this in a pro spective study. Talc and asbestos are chemically similar magnesium silicate compounds, and talcum powder is often slightly contaminated w ith asbestos [43|. Pleural and peritoneal mesotheliomas have repeatedly been ob served in persons exposed to asbestos or talcum powder [44,45], On the other hand, reports rejecting a connection between talc and epithelial ovarian tumors have also appeared (46. 47|. Ovarian tumors have been induced in mice by the application o f carcinogenic polycyclic aro matics. e g . 7,12-dimelhylbenz|a]anlhracene [48], which undergoes hormone-dependent metabolism to reactive r>i erI ' M a m iv h O varian Surface fp ilh c liu m and ( anccr I >1 intermediates in the rat ovary |49). In vivo, the activity o f this carcinogen was found to increase livefold during stimulated cycles in human granulosa tissue compared with human granulosa cells from natural cycles [50J. The possibility that carcinogenic substances enter the ovarian surface epithelium via the genital tract should he seriously considered [51]; however, exogenous factors are difficult to record in a cancer register, and investigations o f this suhject are often speculative in nature. Although both ovarian and peritoneal surface cells are exposed to exogenous factors, peritoneal mesothelioma is much less common than epithelial ovarian carcinoma |52|. Also extremely rare are the so-called 'extraovarian serous carcinomas' [26. 53], What is it, then, that mates the ovarian surface epithelium so much more susceptible to malignant transformation than the peritoneal mesolhe- liu m ? 65). Various findings argue in favor o f the latter possibili ty: the increased incidence o f ovarian neoplasms in the postreproductive years, the proliferative response o f ovar ian superficial epithelial cells to gonadotropins in culture [66], and the presence o f gonadotropin binding sites in ovarian neoplasms [67], Tw o studies published in 1942 showed an increased proliferation rate o f the superficial epithelium o f the mouse ovary in the pcriovulatory period and after estrogen application [68,69). The high prolifera tive potential o f the ovarian surface cells has also been noted in in v itro cell cultures [70], During proliferation. Structures such as intracytoplasmic lumina and cysts ap peared. which are frequently found in epithelial ovarian tumors [71], Oncogenes, including those o f the ras fam ily, are highly regulated to maintain and control cellular prolifer ation (72). Oncogene activation can occur through point mutations, gene rearrangements or am plification and can Endogenous Factor* result in the production o f increased or altered gene prod ucts w ith lumorigcnic potential. A single case o f serous The constantly changing external morphology o f the ovarian surface epithelium contrasts markedly w ith the consistent structural uniform ity o f the bordering perito neal mesothelium (54. 55|. This strongly suggests that local factors govern the morphogenesis o f the ovarian sur face epithelium Some authors believe that the superlicial epithelium o f the ovary reacts like the endometrium to steroid hormones and may tend towards hyperplastic or even neoplastic reactions [8J. Proliferative lesions (papil lary adenomas and adenocarcinomas) have been pro duced in animal experiments by the long-term application o f dicthylstilbestrol [56], It has been shown that the ovarian surface cells possess 17-(3-hydroxysieroid dehydrogenase activity [ 5. 57] and hormone receptors [58. 59). In addition, estrogen and progesterone receptors have been found in certain ovar cystadenocarcinoma w ith known c-k-ra.s amplification was followed over a period o f several months [ 73). c-k-ra.t am plification did not change with tum or progression over this period, and the authors concluded that ras p2 l was most likely involved in the early stages o f carcinogenesis K otyloe t al. [74) found relatively higher Ibvels o f ras p21 in ovarian tumors o f low malignant potential and serous cystadenocarcinomas. in contrast to normal ovaries and serous cystadenomas. No significant differences in ras p 2 l expression distinguished tumors o f low malignant potential from serous cystadenocarcinomas. The overlap in p 2 l expression between these two groups is interesting and indicates that ras p 2 l might be involved at an early stage o f tum or growth or carcinogenesis. Probably onco gene alteration or activation arc not sole determinants of tumor progression. ; ian superficial neoplasms [60. 61). Ham ilton el al. [ 59] hare reported that the growth o f ovarian surface cells in cell cultures is stimulated by androgens and inhibited by Ovulation and tha Ovarian Surface Epithelium antiestrogens. However, treatment o f advanced ovarian carcinoma with antiestrogens has rarely been successful I ] . The formation o f so-called papillary excrescences' appears to be a hormone-dependent process: such papil lary structures in man are more often found in polycystic ovaries and after the menopause, i.c. in circumstances where there are high gonadotropin levels [9. 63. 64] The genesis o f papillary lesions appears to be enhanced by dielhylstilbestrol but suppressed by oral contraceptives, cither directly or via negative gonadotropin feedback (56. I be fact that pregnancy and hormonal contraceptives (75| reduce the risk o f developing ovarian carcinoma suppons the theory that continuously repeated ovulation is an important factor in the pathogenesis o f this neoplasm [76, 77J. It may be the repeated rupture o f the superficial epithe lium o f the ovary during ovulation that increases the risk o f neoplasia. The development o f crypts and inclusion cysts in the ovarian stroma may be due to continuously repeated periovulalory traumatization o f the surface epi- | * Tabt* 1. Follicular s tim u la tion and developm ent o f ovarian cancer in 4 patients Cue No. Age, years O vulatio n in du ctio n regime O varian carcinom a 1 26 2 32 3 25 4 34 e lom iphene. I I C G ( I cycle) FSH. L .ll (8 cycles) . 1-year lapse FS H . L H (3 cycles) elom iphene. F S H rL H (Fcyclc) elom iphene, FSH. L H (1 cycle) b o rde rlin e p a p illa ry serous carcinom a (F1G O stage lb ) e n d o m e trio id carcin om a (F IG O stage lla ) p a p illa ry serous carcin om a (F IG O ^ ta g e lll b ) p a p illa ry serous carcin om a (F IG O stage I I lb ) Reference* 86 87 8889 Ihelium and repair o f the surface defect [78. 79). These ruptures are repaired by cell division, and growth cases when repair is complete. It is proposed that these repeated episodes o f proliferation allow the promotion o f cells already hearing allele loss and those carrying inactivated tumor suppressor genes. This may lead to uncontrolled cell division and malignant transformation |80, 8 1]. fhe suppression o f ovulation and the inhib ition o f gonadotropin secretion have been interpreted as protec tive mechanisms, since oral contraceptives have a protec tive effect against the development o f ovarian carcinoma (75). It remains to be established whether ovulation also has an effect on the form ation o f germ cell tumors or sex cord stromal tumors [65. 82). The technique o f in v itro fertilization has steadily improved in recent years, w ith the goal o f achieving a pregnancy rate as high as possible. The fact that the prob ability o f pregnancy rises in proportion to the number o f embryos implanted has led to the stim ulation o f follicle maturation in the hope o f obtaining many follicles and many conceptions [83|. Due to improvements in freezing techniques, follicle stim ulation is used in grand style in order to provide a certain supply' o f embryos. Gonado tropins. either w ith or without anliestrogens (clomiphene), are used in the first line for follicle stimulation. Excessive estrogen secretion has been found in associa tion with ovarian, endometrial and breast carcinomas, and excessive gonadotropin secretion has been noted w ith ovarian cancer [84). Persistent stim ulation o f the ovary with gonadotropins may have a direct carcinogenic effect or may act in conjunction w ith elevated estrogen concen trations (2. 85). Since currently employed stimulation protocols have been in use for only a few years, it is obvious that nothing conclusive can yet be said about the possible later consequences o f excessive ovarian stimula tion. Three cases reported in the literature (86- 88) and I case o f our own (89), in whom there was a direct chrono logical relationship between hormonal stim ulation o f the ovanes and the development o f ovarian carcinoma, are described in lahle I. Even though these individual reports are in no way proof o f a direct connection between hyper stim ulation and the induction o f ovarian neoplasia, it is nevertheless quite conceivable that the growth o f a preex isting tum or is substantially accelerated by such hormonal therapy. It is a matter o f particular concern that the tumors described developed w ith remarkable rapidity in all 4 cases. On the other hand, it may he coincidental that ovarian carcinoma developed in this subgroup after in duction o f ovulation, and the natural history o f the dis ease may have been interrupted w ith earlier intervention and diagnosis during the in fe rtility treatment [90). Conclusion The ovarian surface epithelium is a dynamic tissue w ith considerable morphological variety. It can grow ver tically downwards to form crypts and inclusion cysts and can also form superficial papillary excrescences. M ore over, metaplaslic transformation to the various types o f m iillerian epithelium can occur. These growth processes appear to be affected by endogenous and/or exogenous factors, and it is conceivable that these factors also induce neoplastic lesions on the ovarian surface epithelium. W hat effect exogenous stimulation o f the ovaries, e.g. as used in in vitro fertilization, may have, is unclear. It remains for future research to identify endogenous and/or exogenous factors responsible fo r triggering neoplastic processes in the ovarian surface cells. 132 Dielt/M arzusch O varian Surface F .piltielium and Cancer Reference! 1 Silvciherg I : Cancer statistics. 1984 ( A 1984. .14 7- 2 l 2 Kulh I M. C/ernobikky B General aspects of ovarian cancel, in Rolh 1 M. Czernobilsky B (eds): lumors and lum orlike Conditions o f the Ovarv New York. Churchill l.ivlngsYdriel 1985. pp 1-9 1 C'rrasman W T. DiSaia PJ: Screening in ovar ian cancer. Am J Obslet Gynecol 1991.165:7 10 4 Baylis MS. Henderson WJ. Pienrepoinl CG. GrifTnhs K: The aetiology o f ovarian cancer; in Morrow CP. Smart GE (eds): Gynaecological Oncology Berlin. Springer. 1986. pp 157-165 5 Blaustein A Surface (germinal) epithelium and related ovarian neoplasms. Pathol Annu 1981; 16 247-254 6 Cavanagh D. Marvien DE. RulTolo E ll. McDonald I W; Epithelial carcinoma o f the ovary Ohstet Gynecol Annu 1983; I 2:31I .119 7 Diet! J. Ruchholz F. Stoll P: Das ovarielle Deckepiihel und seine histogenetische Bezie hung yum (K arialkam nom (iehurtshilfe Frauenheilkd 1986.46561-56*. 8 Gondns B: Surface epithelium o f the developingovary Am J Paihol 1975:81:30.1-312 9 Nicosia SV: Morphological changes o f the hu man ovarv throughout life; in Scrra GB (ed): The Ovary New York. Rsen Press. I9g.l. pp 57-81. 10 Mona PM. Makahe S: Development of the ovarian surface and associated germ cells in the human fetus. Cell I issue Res 1982.226 49.1 510. 11 Blaustein A: Surface celLs and inclusion cysIS in fetal ovaries. Gynecol Oncol 1981;12:2222.1J 12 Motta PM. van Blerkom J. MakaheS: Changes in the surface morphology o f ov arian germinal epithelium during the reproductive cycle and in some pathological conditions. J Suhmicrosc Cytol 1980:12 407-425. 1.1 Died J. Ilrn rich J. Buchhnlz F: Suiface mor phology of the perfused rabbit ovary. Arch Gynrcol 19R7;240;.l 1-4.1. 14 Anderson F.. Lee G. Leiournean R. Alhertini DF. Metier SM: Cytological observ ations o f the ovarian epithelium in mammals during the re productive cycle J Morphol 1976.150.135 166. 15 Czernobilsky B. Moll R. Franke WW. Dallenhach-Hellweg G. Hohlweg-Mayen P: Interme diate filaments of normal and neoplastic tis sues o f the female genital tract with emphasis on problems o f differential tumor diagnosis. Pathol Res Pracl 1984;179:11-34 16 Moll R. le v y R. Cremohdsky B. Hohlweg Mayen P, Dallenbach-Hellweg G. Franke WW: Cytokeratins o f normal epithelial and some neoplasms of the female genital tract I .ah In vest 198.1.49 599-610. I 7 Wu YJ. Parker I.M. Binder NF. Beckett MA. Sinard III, G riffiths C f . Reinwald J(i: I he msothlial krratiru: A new fam ily o f cytoskeletal proteins identified in cultured msothlial cells and non-keratinizing epithclia. Cell 1982: "31:693-699. 18 Mieltinen M. le h to VP. Vjnanen J: Expres sion o f intermediate filaments in normal ova ries and ovarian epithelial, sei cord-stromal, and germinal tumors. In i J Gynecol Pathol 1983;2:64-71. 19 Fercnzy A. Rolh LM : The ullrasiructure of ovarian neoplasms; in Roth LR. Czernobilsky B (eds); Tumors and Tumorlike Conditions of the Ovary New York, Churchill Livingstone. 1985, pp 175-205. 20 Sadly RF: Ovarian tumors. A review. Am J Pathol 1977;87:686-720 21 Blaustein A. Kaganowicz A. Wells J: Tumor markers in inclusion cysts o f the ovary. Cancer 1982:49:722-726. 22 Kabawat S. Bast RC Jr. Bhan AK. Welch WR. Knapp RC. Colvin RB: Tissue distribution of a coelomic cpithelium-related antigen recog nized by the monoclonal antibody O C 125. Ini J Gynecol Pathol 1983.2:275-285. 23 Kurman KJ. Nadji M: Immunocytochemistry o f ovarian neoplasms: in Roth LR. Czemobilsky R (eds). Tumors and lum orlike condi tions o f the ovary New York. Churchill Living stone. 1985. pp 207-221. 24 Mona PM. Van Blerkom J; Scanning electron microscopy o f the mammalian ovary; in Mona PM. Hafer ESF (eds): Biology o f the Ovary. 1 he Hague. NijhofT. 1980. pp 162 -178. 25 Melis M . Carpino F; Ovarian tum orv in Motta PM. Ilafez FSE (eds): Biology o f the Ovary. The Hague. NijhofT. 1980. pp 280-291 26 Gooncranle S. Sassone M. Blaustein A. Talerman A: Serous surface papillary carcinoma of the ovary: A clinico-paihologic study o f 16 cases. Int J Gynecol Pathol 1982:1:258-269. 27 Nicosia SV. Ovarian changes during the c li macteric. in Mastroianni I. Jr. Paulsen CA (eds): Aging. Reproduction, and the Climacter ic. New York. Plenum. 1986. pp 179-199. 28 Nicosia SV. Nicosia RF: Neoplasms of the ovarian msothlium; in Azar HA (ed): Pathol ogy o f Human Neoplasms. New York. Raven Press. 1988. pp 435-486. 29 Nicosia SV. Saunders BO. Accredo-Duncan ME. Setrakian S. Degregorio R: Biopathology o f ovarian msothlium: in Familian G. Ma kahe S. Mona PM (eds): Ulirastructure o f the Ovary. Boston. Kluwer Academic. 1991. pp 287- 110. 30 Barher HRK: Ovarian Carcinoma. Etiology. Diagnosis and Treatment, ed 2. New York. Masson. 1982. 31 Schenker JG. Joseph SM: Epidemiology o f ovarian cancer; in Dallenbach-Hellweg G (ed): Ovarialtumoren. Berlin. Springer. 1982, p p 9 21. 32 lle in iy AP. What causes ovarian cancer? Contemp Oh/Gyn 1987:29-25-30. 11 W oodruff JI>. The pathogenesis of o*ana neoplasia. Johns Hopkins Med J 1979 iu 11 7-120 .14 M uir CS. Necloux J: Ovarian cancer Sr* epidemiological features. World-Health Sun] 1978.31:51-58. 35 De Boer CTI: Transport o f particulate m through the human female genital tract. J I * prod Fenil 1972.28:295-297 36 Annegers JE. Slrom H. Decker DG. Dodaig MB. O'Fallon W M : Ovarian cancer incid and case control study. Cancer I979;43:723 729. 17 Whitelaw RG: Pathology and the consent ovary. J Obstet Gynaecol Br Empire 1959.14 413-416. 38 Mumell EW; Ovarian carcinoma (prdisposai factors, diagnosis and management). Cam* 1952.5:1128-1131 39 McGowan I.; Ovarian cancer after hysterics my. Ohstet Gynecol 1987;69:386-389. 40 Cramer DW . Welch WR. Scully RE. Wn*x chowski CA: Ovarian cancer and tak. A c control study. Cancer 1982:30:372-376. 41 Keal G: Asbestosis and abdominal ncopiaut Lancet I9 6 0 ;ii:l 211-1216. 42 Newhousc M. Pearson R. Fullherton J. Bor G. Shannon M: A case control study o f cam noma o f the ovary. Br J Prrv Soc Med 1977.)| 148-153. 41 Rohl AN. Langer A M . Selikoff IJ. Tordini K K lim eniiclis R: Consumer talcums and po ders: Mineral and chemical characterization J Toxicol Environ Health 1976;2:255-258. 44 Kleinfeld M. Messile J. Zaki M H Monthly experience among talc workers: A folk** study. J Occup Med 1984;16:345-348. 45 Selikoff IJ. Hammond EC: Health hazards of asbestos exposure Ann NY Acad Sci 1979.38) 1- 6 . 46 NN: Cosmetic talc powder, lancet 1971; 13 4 8 -1349. 47 Roe FJC: Controversy: Cosmetic talc and ov ian cancer, lancet I979;ii;744. 48 Rao AR Effects o f carcinogen and/or muiagro on normal and gonadotrophin-primed ovam of mice. In i J Cancer 1981;28:10 5 -MO. 49 Bengisson M, Rydstrom J: Regulation of car cinogen metabolism in the rat ovary by l esirnus cycle and gonadotrophin. Science 1981 219:1437-1438 50 Fishel S. Jackson P: Follicular stimulation f high tech pregnancies: Are we playing it sa'c' BMJ 1989;299:309-311. ' I Venter PF: Ovarian epithelial cancer aid chemical carcinogenesis. Gynecol Oncol I9H. 12 281-285. 52 Enzinger FM . Weiss W: Soil Tissue Tumoun Si. Louis. Mosby. 1983. 53 August CZ. Murad TM . Newton M: Multiple focal extra ovarian serous carcinoma Ini J (nevol Pathol 1985;4 11-15. 54 Blaustein A: Peritoneal msothlium and ovar ian surface cells - Shared characteristics Ini J Gynecol Paihol 1984;1:361-375. 78 Radislavjevic SV: The pathogenesis o f ovan. x NioiMi SV. Johnson J it: Surface morphology of ovanan mesothelmm (surface epithelium) and ofother pelvic and extrapelvic mesolhrlia! 65 Cramer DW . Welch WR: Determinants of ovarian cancer risk II. Inferences regarding pathogenesis- J N a ll Cancer Insl 1983.71:717 inclusion cysts and cystomas Obstet G yne* 1977;49:424-429 79 Zajicck J: Prevention o f ovarian cystomas * tiles in the rahhit Int J <iynccnl Pathol 1984:3: 249-26 . * t>rn NV. Pierce EC. Anderson R( : Papillo matous growths on interna! genitalia o f bitches aJministered the synthetic estrogen trans-4.4'- Jimcthyl-aa'-diethylsiilbene lo xico l Appl Pharmacol 1972:21:582-585. 1 Adams AT. Auersperg N: Transformation of cwhured rat ovarian surface epithelial cells by 721. . c tl 66 Oslerholzer HO. Streibel FT. Nicosia SV: G rowth effects o f protein hormones on cul lured rabbit ovarian surface epithelial cells Biol Reprod 1985;33:247-258. 67 Raganemi H. Kauppila A. Ronnberg L M a n dez K. Pystynen P: LH (HCG ) receptors in benign and malignant tumours o f the human ovary Acta Obstet Scand 1981 ;lO I|suppl) 83 inhibition o f ovulation: A new concept. J R prod Med 1978;2: 114 80 Russell SEN. Hickey GJ. Eowry WS W hin Atkinson RJ: Allele loss from chromosome in ovarian cancer.-Oncogene 1-990:5 158-* 1583. 81 Lowry S. Russell H. Hickey J. Atkinson i I norssani ovulation and ovarian cancer. Lan* 991.337:1544-1545. Kmten murine sarcoma virus. Cancer Res 1911;41:2063-2072. % Adams AT. Auersperg N: A u to ra d io g ra p h ic in 86. . . 68 BulloughWS: Oogenesis and its relation to ocs- trous cycle in the adult mouse J Endocrinol 82 D ietl J. H om y IIP ; Imponance o f ovulation ovarian cancer. BMJ 1990:301:9JI 83 Van den Abbecl E. Van der Elst J. Van Wa vestigation of estrogen receptors in cultured rat ovanan surface epithelial (ROSE) cells. J Cell gioi 1981;91:209a. M Hamilton TC. Davies P. C riffilhs K Steroid hormone receptor status o f the normal and Moplastic ovarian surface germinal epithe- i,ym; in (ireenwald OS. Terranova PF (eds): 1942.3:141-151. 69 Stein KF, Allen E: Attempts to stim ulate prohl eration o f the germinal epithelium o f the ovary. A n a tR e c 1942;82:1-9. 70 Nicosia SV. Johnson JH. Streibel EJ: Growth characteristics o f rabbit ovarian msothlial (surface epithelial) cells. In t J Gynecol Pathol berghe L. Camus M. Devroy P. Khan J. Sm J. Staesen C. Wisanto A. Van Steirteghem Hyperstimulation: T he need for cTyopresei tion o f embrvos. Hum Reprod 1988;3:53-5 84 Henderson BF. Ross R. Bernstein L: Estrog* as a cause o f human cancer The Richard a Hmda Rosenthal Foundation Award Lecm factors Regulating Ovarian Function New knrk. Raven Press. 1983. pp 81-85. aA Al-Timimi Buchley CTI. Fox H. An im m uno 1985;4:58-74. 71 Warhol MJ. Hunter NJ. Corson JM: An ultra structural comparison o f mesotheliomas and adenocarcinomas o f the ovary and endometri Cancer Res 1988;48:246-253 85 Stadel V: The etiology and prevention o f o* an cancer Am J Obstet Gynecol 1975:1 histochemical study o f the incidence and sig nificance of sex steroid hormone binding sites i t normal and neoplastic human ovarian tis sue. Ini J Gynecol Paihol 1985;4:24-31. um. In i J Gynecol Paihol 1982;1:125-134. 72 Saw no H. G an e ti CT: Oncogenes in gyneco logical tumors; in Sasano H (ed): Gynecological T urnon Berlin. Springer. 1992. pp 362-372. 772-774. 86 Allas M . Merc7er J: Massive hypcrslimulai, and borderline carcinoma of the ovary A p sible association. Acta Obstet Gynecol S<.. si Schwartz PE. I ivolsi VA. Hildreth N. MacI inky NJ. Naflolin I N. Eisenfeld AJ Estrogen receptors in ovarian epithelial carcinoma. Ob stet Gynecol 1982;59:229-231 73 Filmus J. Buik R: Slability o f c-k-ras amplifica tion during progression in a patient with ade nocarcinoma o f the ovary. ( ancer Res 1985. 1982;61:261-263 87 Bamford JM . Steele SJ: Uterine and ova. carcinoma in a patient receiving gonadoi phin therapy - Case report. Br J Ohstet G> 42 Schwartz PE Steroid hormone receptors and nti-estrogen therapy in ovarian neoplasms; in Roth LM. Czemobilsky B (eds): Tumors and Turoorlike Conditions o f the Ovary. New 3ork. Churchill Livingstone. 1985. pp 233 45:4468-4472. 74 K olylo PK. Michael H. Finebeig N. Sutton G. Roth L: Flow cytometric analysis o f D N A con tent and ras p 2l expression in ovarian cancer neoplasms. Int J Gynecol Pathol 1992.1 l.3 ri- col 1982.89:962-964 88 Carter MF. Joyce ON: O varian carcinoma . patient hyperstimulaied by gonadolrop therapy for in v itro fertilization: A case rep J In V itro Ferl Embryo Transf 1987 4 I 239. 41 Makahe S. Iwaki A. Hafez ES. M otta PM. Physiomorphology of fertile and infertile hu man ovaries; in Motta PM. Ilafez ESE (eds) Biology o f the Ovary 13ie Hague. NijhofT. 1980. pp 279-290. M Sommers SC: Polycystic ovaries revisited, in progress: in Fenoglio CA. WolIT M (eds): Surgi cal Pathology. New York. Masson. 1980. pp 221-234. 75 Centers for Disease Control Cancer and Ste roid Hormone Study: The reduction in nsk of ovarian cancer associated with oral contracep tive use. N Eng) J Med 1987;316:650-655. 76 Fathalla MF: Incessant ovulation - A factor in ovarian neoplasia? Lancet 197|;ii:l63- 77 Cruickshank DJ: Aeliological importance o f ovulation in epithelial ovarian cancer: A popu lation based study. BMJ 1990;301:524-525. 128. . 89 D ietl J: Ovulation and ovarian cancer l ai 1991:338:445. 90 Goldberg G L. Runowicz CD: Ovanan ca noma o f low malignant potential, infertil and induction o f ovulation - Is there a h A m J Obslet Gynecol 1992:166:853-854 The Biology of Ovarian Cancer Development Antlrcw K Godwin. P hD .. foseph R. Testa. Ph D., ami Thomas C. H am ilton, Ph D. Ila c k g ro u n d . In th e o ry , a ll Ihe c e ll type s th a t c o m p rise the h u m a n o v a ry have the p o te n tia l fo r m a lig n a n t tra n s fo rm a tio n . The vast m a jo rity o f m a lig n a n t o v a ria n tum ors iu the h um an , how ever, a rise fro m the o v a ria n surface e p ith e liu m . These ce lls have im |> o rla n t fu n c tio n s d u rin g re p ro d u c tiv e life : the y c o n trib u te to fo llic u la r ru p tu re and hy c e ll d iv is io n re p a ir the w o u n d tha t a ccom panies o v u la tio n . T h e re has been m u c h s im u la tio n tha t the ra p id cycle s o f c e ll d iv is io n a ssociated w ith w o u n d re p a ir c o n trib u te s ig n ific a n tly to the d eve lo pm en t o f o v a ria n c a n c e r. S u i li .s|ecululion is based o n th e o b s e r va tio n th a t o v a ria n cancer occurs m ost fre q u e n tly at the end o f a w o m a n 's re p ro d u c tiv e life a n d is a sso cia te d w ith n u llip a rity . It is o f |M ile n lie l s ig n ific a n c e th a t, u n lik e m ost e p ilh e lia , these ce lls a re not re p lace d th ro u g h re p le n is h m e n t stem ce lls w ith the d e v e lo p m e n t o f one endstage c e ll a nd o ne c e ll w ith c o n tin u e d g ro w th p o te n tia l. R ather, the d iv is io n o f an o v a ria n su rfa c e e p ith e lia l c e ll y ie ld s tw o d a u g h te r ce lls w ith e qu al p o te n tia l fo r subse quent g ro w th . Thus, a ll |x ile n tia l m u ta tio n s as the y a c c u m u la te are passed on to n e a r-e x p o n e n tia lly e x p a n d in g subsequent g e n eratio n s o f ce lls th a t (a n a c q u ire a d d i tio n a l m u ta tio n s tha t co u ld co nfe r the m a lig n a n t p h e n o ty p e . M ethods. W e have developed a m o d e l to le s l the h y pothesis tha t repealed ce ll d iv is io n by o v a ria n s u rfa c e e p ith e lia l ce lls c o n trib u te s to d e v e lo p m e n t o f m a lig nan cy. In th is m odel, ra t o v a ria n s u rfa c e e p ith e lia l ce lls a rc iso la te d and su bje cted in v itro to re p e titio u s c e ll d iv i sion In in (in i( in a s im p le w ay g rn w lh n f the s u rfu c e e p i th e liu m in v iv o . Ilt'.sulis. i hese ce lls deve lo p u m a lig n a n t p h e n o typ e based on loss o f c o n ta l I in h ib itio n , th e a b ility fo r sub- P iv -v iiie ii .ii tin- IV rq H -tln e s tut O i .h m h C .iu-r in O IJ e r-A g e d IV iii if ii l m i,-n i K n i\K -tlg f .m il K e n ifn in m J .ilk iiis for Kf.-n.-arch W inking l o iifin -iu .- Ilfilif-it,i M.irvl.im l. N ovem ber 2ll 21. IW J I lu lu (III- | >|-p.ltlMtf!ll o M f lt ll.il P ill lili Igv I 111 l ll.lM' ( . n i l f l ( `iilc t. I'IhU I. Ipl.i.i I't n iis \l\,n iM S n p |k iilftl in p.nl hv g i.m l ( A Ur>V27 Iio m (In- N ,iim n .il C .mcer Insum if H fltifstlj. Maivl.nul Aililrfvs fui u-piiiii> Itm m .iNl ll.um lton. I'li I) . I Vparlm eni ol M fitu .iI l >iiiolog\ . I o \ ( }ia>f l .m u i l t'lilt- j. 7?ll I Hurlinlin' A v f nut- I'h il.u tflp iu .i I*A I `J i l l A t if p le il t<>i p u l'lii.ilion lu ll In I*1? s lra te in de p en d en t g ro w th , tu m o rig e n ic ity in a th ym ic m ice, a nd cytog e ne tic changes. C o nclusio n s. O u r data s u p p o rt th e in v o lv e m e n t of tu m o r su pp re ssor genes in the d e ve lo p m e n t o f o varian c a n c e r. C a n c r .r 199.'!; 7 l:S ;i( ) - ti. Key w o rd s: o v a ria n c a n te r, m a lig n a n t tra n s fo rm a tio n , tu m o r su p p re sso r genes, cytog e ne tics, su rface epi th e liu m . The etiology of ovarian cancer is unknow n,12 but there are some data that suggest that m ultiple factors may contribute to the initiation and progression o f the dis ease. The vast majority of ovarian carcinomas arise as a result o f malignant transformation of the ovarian sur face epithelium 3 * This simple epithelium, composed of a single layer of cells, often is described as modified peritoneal mesothelium. This paper is divided into two parts. The first part summarizes some of the inform ation available on the role of hormones and growth factors in regulation of normal and malignant ovarian surface epithelium in the context of how regulatory substances may affect disease etiology and progression The second part con tains experimental data supportive of the long-standing hypothesis that incessant or repetitious ovulation con tributes to the initiation of ovarian cancer. Features of the Norm al and M a lig nan t Ovarian Surface Epithelium In adult life, the ovarian surface epithelial cells are a single layer of nondescript cells that vary from squa mous to cuboidal and low- columnar * 7,8 The generally unremarkable features of these cells suggest that they might have little role in normal ovarian function, but this does not appear to be the case. The ovary has two major functions: (1) the production of steroid hor mones, and (2) the timely release of ova. Based on data from several species, ovarian surface epithelial cells are believed lo have a pivotal role in the latter process 2810 As the time of ovulation approaches, ovarian sur- Ovarian Cancer Biology/G odw in rl al. 531 face epithelial cells (in the area of the ovulation fossa) produce large lysosomal bodies that migrate to the ba sal region of the cells. These large bodies are excreted from the cells' basal surfaces before ovulation. It is be lieved the lysosomal bodies con tain enzymes that partic ipate in the breakdown of the tunica albuginea and thus aid in follicular rupture.1' The process o f ovulation re sults in a wound at the ovarian surface that is repaired by rapid growth o f ovarian surface epithelial cells.11 These surface epithelial cells also may participate in the production of a new tunica albuginea.12 The require ment for regulation of both processes (i.e., iysosome production and wound repair) suggests that the func tions o f ovarian surface epithelial cells are controlled in complex ways. There is potential for a diverse range of regulatory molecules to influence function and grow th of ovarian surface epithelial cells. These signals may reach their target by autocrine, endocrine, or paracrine routes. The latter o f these mechanisms (i.e., paracrine) may have special significance to the surface epithelium in that the ovarian surface is bathed by follicular fluid at the time of ovulation. Direct experimental evidence on the roles of various hormones and growth factors in control of ovarian surface epithelial cells, however, is limited. The potential for hormones and grow th factors to regulate growth and function of these cells is suggested by the presence of appropriate receptors for many of these reg ulatory substances in both the normal and malignant ovarian surface epithelial cells. Estrogen receptors, for example, have been detected in both normal and m alig nant ovarian surface epithelial cells." ,3*,s Independent laboratories have reported estrogen receptors in the normal rat ovarian surface epithelium, whereas there is less evidence for the presence o f estrogen receptors in human ovarian surface epithelial cells." There have been numerous studies, however, showing the pres ence of estrogen receptors in malignant tumors o f ovar ian surface epithelial cell origin.13,6` ,# The presence of estrogen receptors in the malignant counterpart o f a given cell type generally is considered to result from retention of a differentiated feature of the normal pro genitor cell, rather than de novo synthesis as a result of malignant transformation. The complexity of ovarian surface epithelial cell regulation is indicated by the fin d ing of estrogen production by ovarian cancer ceil lines.1920Thus, estrogen, which generally is considered to mediate its effects through the endocrine route, may also have effects by both paracrine mechanisms (i.e., estrogen in follicular fluid) and autocrine mechanisms. The exact function of estrogen in ovarian surface epithe lial cell regulation is not known. Limited evidence in the rat and mouse suggest estrogen is a weak mitogen.1421 In some estrogen receptor-positive human ovarian cancer cell lines it recently has been shown that estro gen is a strong m itogen17,2223and promotes the capacity for substrate-independent grow th.22 it is possible that these effects are mediated indirectly by transforming growth factor o and (1production.24 In the PE04 ovarian cancer cell line, in which estrogen stimulated growth, it was observed that the hormone did not induce proges terone receptor synthesis,22 whereas in the OVCAR-3 ovarian cancer cell line estrogen induced progesterone receptor but not growth.2225Thus, it is possible that the divergent signals that mediate these tw o markers of es trogen action are uncoupled readily in ovarian cancer. This could explain why antiestrogen therapy is effective only in a small subset of patients w ith ovarian cancer.26 The significance of progestagens and androgens to the regulation of normal and malignant ovarian surface epithelial cells is lim ited essentially to the findings of receptors for these tw o sex steroid hormone classes in ovarian tumors.27O nly very lim ited data suggestive of a role for testosterone in stimulation23-2* and antiandro gens on inhibition of g row th162,1of ovarian cancer cells are available. It is noteworthy that the postmenopausal ovary is a net producer of androgens due to the contin ued formation of functional secondary interstitial tissue in the organ,3031 and thus, if androgens stimulate grow th of either normal or malignant ovarian surface epithelial cells, they may well contribute to disease etiology or progression (see below). The roles of peptide hormones and growth factors in the regulation of the ovarian surface epithelium are not completely defined. It has been shown that the agespecific incidence of ovarian cancer increases abruptly at the time of the menopause.3233 The age-related change in ovarian function results in loss o f feedback inhibition on pituitary peptide hormone synthesis such that the pituitary produces constant increased amounts of follicle-stimulating hormone (FSII) and luteinizing hormone (LH) (15- and 5-fold normal levels, respec tively) for a number of years after the climacteric.34This suggests that FSH and LH may somehow contribute to the development of ovarian cancer. The potential for direct effects of peptide hormones on the surface epithe lium is shown by the existence of FSH and LI I receptors in ovarian cancer specimens.3536 In addition, peptide hormones have been shown to stimulate grow th of normal rabbit ovarian surface epithelial cells in v itro 37 and cell lines derived from ovarian cancer specimens.26 Receptors for epidermal grow th factor (EGF; the best known of the peptide growth factors) have been found in both the normal and malignant human ovar ian surface epithelium .3639 Furthermore, EGF has been shown to be weakly to moderately mitogenic in rat40 and human ovarian surface epithelial cells 41 It is of interest to consider how mitogens may come r;i2 CANCKK S u pplem ent fiu iN iirif 15, 199). V o lu m e 71, No. 2 info the proximity of the surface epithelium. This tissue is avascular .uul normally is separated from the ovarian cortex hy a well defined tunica albuginea Ovarian cancer is believed to arise most frequently in surface epithelial inclusion cysts. These cysts result from the proliferation of surface epithelial cells that either in wide tlu* ovarian cortex or are trapped w ithin the cortex during ovulation or remodeling of the ovarian surface " 1 Such cysts increase in number in relation to ovulatory activity, and thus are more prevalent in oldei aged women ,M * It should be noted that the surlaic epithelial cells pning these inclusion cysts in the ovarian cortex are theoretically in close proximity to the mitogen producing components of the ovary, including the functional seiomlarv interstitial tissue of the older aged women. I here is a reasonable amount o f information sug gestive that diverse regulatory molecules have the p o tential to stimulate growth of both normal and m alig nant ovarian surfa* e epithelial cells. The question arises .is io how growth stimulatory effects might relate to disease etiology or manifestation The potential for growth-stimulatory factors to influence disease manitestation appears straightforward. An alteration in the concentrations or patterns of production of grow th-re gulatory suhstanccscould stimulate slowly growing ma lignant cells mk'Ii that tumor rapidly becomes manifest clinically. Alternatively, malignant transformation might include an altered responsiveness to such sub stances and thus more rapid growth. The issue of how hormones and growth factors may duectlv influence the etiology of ovarian cancer is mote complex. The simplest explanation is that any in crease in DNA replication increase's the likelihood of a mutation. Such mutations could be dominant, activat ing a prolooiK ogene or producing an altered antionco gene product. Alternatively, recessive mutations could inactivate antioiuogenes. Stimulation o f mitotic activity may be ol special significance in the ovarian surface epithelium. Ovarian surface epithelial cells, unlike most epilhelia, behave as generative stem cells.2 Thus, the division of one of these cells is believed to yield two cells with equivalent potential for future growth. This is m contiast to replicative stem cells that simplistically yield two cells, one with loss of growth potential (i.e., destined for iliffeienliation) and one stem cell w ith con tinued replicative potential. In the case of ovarian sur face epithelial cells, a single mutation could be passed on to exponentially expanding progeny, and since transformation is essentially a game of numbers, the larger the number of cells available for additional m u tations. the more likely it is that a mutation of signifi cant e to transformation w ill (ccur. It is straightforward to apply this pattern of replication It the simple model o f antioncogene inactivation, where both alleles of the antioncogene must be inactivated to achieve malignant transformation.44 If replication increases the pool of cells w ith one inactivated allele (first hit) and also in creases the likelihood of additional mutations (second hit), the process of replication in generative stem cells becomes doubly (or perhaps exponentially) dangerous Thus, any grow th-prom oting stim uli could increase the possibility to accumulate combinations o f potentially devastating mutational events that eventually result in malignancy. Repetitious Ovulation and Ovarian Cancer Etiology The observation that ovarian cancer is more prevalent in nulliparous women served as the basis for the devel opment o f a hypothesis that repetitious ovulatory activ ity is a factor in malignant transformation of the ovarian surface epithelium 45 Alternatively, it has been sug gested that an underlying hormonal imbalance might be responsible for in fe rtility and that this, rather than ovulatory activity, may be responsible for the increased incidence of ovarian cancer observed in nulliparous women.46 The fact that birth control pill use reduces ovarian cancer risk supports the former, more widely touted hypothesis.47 50It also is noteworthy that tumors of the ovary derived from the surface epithelium are rare in animals Germane to this, in nature animals ovulate seasonally and most frequently become preg nant shortly after ovulatory activity commences in the spring. Thus, the number of ovulations during life is quite low The only exception to this is the incessantly ovulating domestic hen. which has a high incidence of ovarian cancer.51,52 It is tempting to speculate how ovulatory activity could contribute to ovarian cancer etiology If surface epithelial inclusion cysts are im portant in disease etiol ogy, it is obvious that there would be a greater opportu nity for such cysts to form in the more frequently ovu lating person In addition, the surface epithelial cell mi totic activity required to repair the wound created by the follicular rupture accompanying ovulation may be a factor in disease initiation. As noted above, mutations are more likely to occur in dividing as opposed to quies cent cells Furthermore, because grow th increases the pool o f cells carrying a potential first mutation (first hit), the probability of an additional mutation occurring in a cell w ith a preexisting mutation is increased It is of interest to speculate whether certain mechanisms to vig orously assure the fidelity o f DN A replication, repair, and of chromosomal segregation might be inactive in surface epithelial cells because these cells would have been under little selective pressure from the standpoint Ovarian Cancer Biology/G odw in et at. 533 of evolution. In nature, such cells have a very lim ited requirement for replication, in contrast to the selective pressure placed on the skin or gastrointestinal mucosa, which during evolution might be expected to have been rigorously selected for the capacity for wound repair and lumenal replenishment, respectively. In fact, pro teins have been identified that are believed to have a role in the fidelity of D N A replication and chromo somal segregation and, thus, act to reduce the possibil ity of recombination errors and malignancy.53"56 Over the years, there has been a rather strong case made for the role of incessant ovulation in ovarian cancer etiology. Beside the recapitulation of the original hypothesis, however, previously there has been little more supportive epidemiologic data and even less bio logic evidence. Recently, we have described a model in which ovarian surface epithelial cells from the rat u n dergo malignant transformation in association w ith prolonged exposure to a growth promoting environ m e n t . S u r f a c e epithelial cells isolated from the rat ovary by selective trypsin treatment are placed in c ul ture These cells rapidly grow to cover the culture sur face as if to cover the wound created by follicular ru p ture and, at confluence, mitotic activity ceases. These cells may then be repetitiously subcultured to mimic in vitro in a simple way the repeated requirement for growth produced by incessant ovulation in vivo. This / 1 VA 1 II ff (1 if If ii n 12 3 45 6 M II >\ V H >: i i 7 6 9 10 " . '. 1 2 )I >r (f Il II 13 14 15 16 1 7 II 16 1 1 19 20 1 mar / K ., It K Figure 2. Karyotype o f a C'.icmsa banded m rlap h ab i1(.ell from late passage clone 7. Clonal lunges in clude an unbalanci-d 5; 12 derivative chromosome (arrow) and an u n id e n tifie d m arker (mar). Inset: chromosome pairs 5 and 12 fro m another celt; der(S ;!2 ) is arrowed X x) )j X I 12 3 45 II 6 Il M 7 69 If 1 10 1 1 II 12 7/ 71 II If II 13 14 15 16 17 16 U tl 19 20 Figure I Karyotype o f a Ciemsa banded ntelaphase cell fro m early passage clone 7, Clonal changes include loss o f one chromosome 5 and partial deletions of chromosomes 4 and X. Note that the remaining copy o f chromosome 5 appears norm al cytogenetically, but molecular analyses indicate that a small segment, in clu din g the in terfe ro n -> locus, is deleted from this chromosome (see Fig. 3). Loss o f a chromosome 11 is a nonclonal lun g e seen in this cell inly Arrows indicate structural alterations ol 4 aiul X. prolonged growth results in the spontaneous acquisi tion of multiple features consistent w ith malignant transformation. The cells show loss of contact in h ib i tion, acquisition o f the capacity for substrate indepen dent growth, and. when injected intraperiloneally or subcutaneously into nude athymie mice, form tumors consistent with serous cystadcnocarcmoma of ovarian origin. We have examined the karyotypes of two spontane ously transformed clonal cell lines derived from ovarian surface epithelial cells and one spontaneously trans formed mixed population surface epithelial cell line. In each cell line, the karyotypes revealed several cytoge netic abnormalities that could be consistent w ith onco gene activation or tumor suppressor gene inactivation. The most frequent changes involved chromosome 5 and chromosome l . <0 One of the clonal rat ovarian surface epithelial cell lines, clone 7, compared for cytogenetic changes be tween early (normal phenotype) and late passage (transformed phenotype), showed cytogenetic changes of theoretical interest (Figs. I and 2) This cell line in early passage showed monosomy for chromosome 5 (Fig. 1), whereas late passage cells contained tw o copies 534 CANCKR Supplem ent fn n tu iri/ IS. 1991, V o lu m e 71. No. 2 of cliriMnoMimi' 3. one apparently normal homologue and one defective copy consisting o f an unbalanced de rivative between chromosomes 5 and 12 (Fig. 2). We interpret thi* observation to indicate that there has been 5 /5 ; i to o * < J / J > 100* ( o / o- > to o * < */*) jo A /- / / XQ-. *Q- o------ o `5 -/ - 5' / d*r< -5' m ar 50* <J / - > 50* <a / - ) 0* (-t -) 1 100* ( J / J) 100* (0 /8 1 0* ( -/ - ) Figure \ t>NA hint analysis o f carlv and late passage ral ovarian germinal ip id u liu m (RlYVCiF) clone 7 evil lines fur loss ot .specific Ilki >( genes lo ta ti'd on ilirom osonu' 5 High m o lcm lar l) N A (10 vg per lam-) irnm iht- carlv and la it* passage t loin* 7 tells (indicated above) ivjs digested w ith 1.1oKI I'U'Cirnphon-si'd on a 1% agarose gel and Ir.m s lrfin l in nylon m rin h i.iiii- fillers fille rs were liy hi n il/is l s( .|<ii-i i Im U\ w ith IIn1 lollow ing ( ,,|'| l.ib v liil im k lia rts l.iin i t IN A p m tvs. (A) inii ii'io n it, (H> interferon < ((.') c tun, and ( l) | I (.( iis I'ptoi I In- ral fcll; receptor i |)N A probe detects i l i n r hands (M h m 2 and 4 0 kdebase pairs |kb pj). Ihe 4 0 kbp b<\pd (1*1h. iv tm h is i Insi st ni si/e lo inter ferotw f and c /nn, and the 9 b and M 2 hands p ig ltl) an- shovsii lo t comparison ul U N A loading and lia iis lc i lln id III ilig i-s ifd A phage U N A was iim -J as in o le iu l.tr si/e m aikeis ( h r (ragmcnl si/i- is mdu ated in kiluhase pairs lo the left U N A from normal tat spleen ami lung is u uInded as ,i eonlrol lor normal gene dosage Normal Surloco Epunaiium Hypomoucol Clono 7 forty Clono 7 Lata Figure 4. Schematic diagram ilInNtrating promised clonal evolution o RO VGI: clone 7 cells. N orm al surface epithelial cells (N ) conlam tw o norm al copies o f chromosome 5 We hypothesize that on in vitro culture, a single cell loses one copy of chromosome 5; the second copy (" 5 " ) appears norm al cytogenetically hut has a microscopic deletion of a segm m i that in clude the interferon-n locus ta r ly passage clone 7 cells also display deletions of X and 4 (Xq and 4q ). Late passage clone 7 cells have duplicated the defective chromosome "5 ." and one o f these copies participates in a rearrangement w ith a chromosome 12. der("S " ;12) The P N A signal in tensity and allelic constitution o f three different genes residing on chromosome 5 (i e . * -put alleles J and I*, the inlerferon u gene locus |A /A *J . and in te rfe ro n -^ |B/B*J) are indicated Both alleles o in te rfe ro n - are lost in early and late passage clone 7 cells. Percent sign indicates the approximate gene dosage o f the three genes as determined hy Southern blot analysis, mar indicates the presence of an unkn ow n marker chromosome. The existence of the hypothetical suIpopulalio n o f cells noted above is suggested hv the presence of late passage clone 7 cells that do nol contain cytogenetic deletions o f chromosome X and chromosome 4 This would indicate that the hypothetical population (' V / ) gave rise to the early passage clone 7 cells ( 5 " / -, Xq , 4q ) and to another subpopolation w ith the features of late passage clim e 7 This latter subpopolation (at the lim e o f early passage cytogenetic analysis) must have been a very small component o f the culture, and thus was not observed in the chromosomal preparations examined and did not influence gene dosage This latter subpopulation. however, apparentlv had a distinct growth advantage, and by late passage became dom inant if not the entire population duplication of the single chromosome 5 of early passage cells. This view is supported by molecular data (see be low). It has been hypothesized by others that one mech anism by which predisposing recessive somatic or germ line mutations could be unmasked is by the loss of a normal (w ild type) chromosome (by mitotic nondis junction) and reduplication of the remaining homo- Ovarian Cancer Biology/G odw in et al. 535 logue containing the aberrant allele.5758In our example, we examined the D N A o f early and late passage clone 7 cells for three genes that reside on the long arm of chro mosome 5: the interferon- family, interferon-0, and c-/um.4059 It would be anticipated in early passage clone 7 cells, which had one cytogenetically normal chromo some 5, that there would be one-half the diploid dosage for these genes This was the case for c-juti and inter feron-0 (Fig 3). Unexpectedly, all members of the inter feron-or fam ily were totally absent (Fig. 3). This in d i cates that there was a submicroscopic deletion in this apparently "normal ' chromosome 5. A strong indica tion that the chromosome 5 portion of the unbalanced 5; 12 derivative seen in late passage cells originated from the defective chromosome 5 o f early passage cells is given by a doubling of the gene dosage for interferon-0 and c-jun and the continued total absence of the interferon- fam ily in D N A of late passage cells (Fig. 3). This is perhaps the first experimental proof by a combination of cytogenetic and molecular methods, that the process of reduplication that returns a cell to disomy does indeed occur in transformed cells, and the process is summarized in Figure 4 along w ith the gene dosage evidence. The cytogenetic and molecular changes in rat chro mosome 5 we observe are in an area where a yet-to-bedoned gene for the suppression of anchorage-depen dent (SAI) growth is proposed to reside.60 Furthermore, this portion of rat chromosome 5 has synteny w ith the short arm of human chromosome 9,59w hich reportedly shows frequent cytogenetic changes in human ovarian cancer61 and has been proposed as the site of a tumor suppressor gene.62-64 In conclusion, much w ork remains to be done to determine not only those genes involved in the develop ment of the growth-dependent acquisition of the trans formed phenotype in our rat model, but also to show the relationship o f those findings to human ovarian neoplasms. Nonetheless, the development of this model has provided the first experimental evidence to support a role for repetitious ovulation in ovarian cancer etiology, and has proved a starting point to ex perimentally unravel potential genes involved in dis ease etiology. References 1. Piver MS. Baker TR. Piedmonte M. Sandecki AM . Epidemiology and etiology o f ovarian cancer Sr/nm On/ 1991; 18:177-85. 2. Ham ilton, TC O varian cancer: part I: biology. Curr Prohl Cancer 1992; 16:1 57 3. Scully RE O varian tum ours: a review Am / Pathol 1977; 87:686-720. 4. W oodruff |D. I listo ry of ovarian neoplasia: facts and fancy. O b stetrics and Gyiiivofiiffy Annual 1976; 5:331-44 5 Nicosia SV, Nicosia RE. Neoplasms o f the ovarian mesothelium In: Azar HA, editor. Pathology of human neoplasms. New York: Raven Press. 1988:435-86 6 Bell DA. O varian surface e pith elia l-stro m a l tum ors H u m P athol 1991; 22:750-62. 7. Papadaki L. Beilby JOVV. The fine structure o f the surface epithe lium o f the hum an ovary. / Cell Sci 1971; 8:445-65. 8 Blauslein A, Lee II. Surface cells of the ovary and pelvic perito neum: a hiMochemical and ultrastruc lure comparison. Gytietol O m ni 1979; 8:34 43 9. Bjcrsing L. Cajander S. O vulatio n and the role of the ovarian surface epithelium , l.sp tricn lia 1975; 31:605-8. 10 M olta P, Van Blcrkom ). Makabe S Changes in the surface m o r phology o f ovarian germ inal" epithelium during the repn>ductive cycle and in some pathological conditions / Submit rose Cy la l 1980; 12.407-25. I ! Nicosia SV, Saunders BO, Acevedo-Duncan ME, Setrakian S, DeGregoria R. Biopathology of ovarian mesolheiium. In: FamiItari G, Makahe S. M otta PM. editors. Ultraslructure o l the ovary Boston: Kluwers Academic Publishers. I99 I.2 8 7 -3 IO . 12 Auersporg N, Maclaren IA. Kruk PA O varian surface e pith e lium : autonomous production o f connective tissue type extra cellular matrix. Biol Repn*d 1991; 44:717-24. 13. H am iltun TC, Davies P, G riffiths K Androgen and oestrogen bind in g o f cytosols o f hum an ovarian tumors. / TiiJei rinol 1981; 90:421-31 14 Adams AT. Auersperg N Autoradiographic investigation o l es trogen b ind in g in cultured rat ovarian surface epithelial cells. / Hiflochem Cylochem 1983; 31:1321 5. 15. Ham ilton TC, Davies P. G riffiths K. Oestrogen receplor-like b ind in g in the surface germ inal epithelium o f the rat ovary. / Endocrinol 1982;95:377-85 16 Slotman B. Rao B O varian carcinoma |review |. A nth aute r Rei 1988; 8.417 34 17 Langdon SP, Hawkes M M , l^n vne SS. H awkins RA. Tesdale A L, C rew AJ, et al. Oestrogen receptor expression and the ef fects o f oestrogen and tam oxifen on the g ro w th o f hum an o var ian carcinoma cell lines Br J Cancer 1990; 62:213-6. 18. H am ilton TC, Davies P. Horm onal relationships in ovarian cancer Rrt>iVu>s an Endocrine-Related Cancer 1983; 14:19-22. 19 Wimalasena J, Meehan D. Cavallo C. Hum an e pithelial ovarian cancer cell steroid secretion and its control by gonadotropins. Gynecol Oncol 1991; 41.56-63 20 Poels LG, |ap PHIC, Ramaekers FFCS, Scheres JMJC, Thomas C M C . Vooijs PC. et al Characterization o f a horm one produc ing ovarian carcinoma cell line. Gynrcu/ One/ 1989; 32:203-14. 21 Stein KF, A llen S. A ttem pts to stimulate p roliferation o f the g e r m inal epith eliu m o f the ovary. Anal Ref 1942; 82:1-9. 22. Nash JD. Ozols RF, Sm yth JF, H am ilton TC. Estrogen and antiestrogen effects on the g ro w th o f hum an epithelial ovarian cancer in vitro. O bsir/ Gynecol 1989; 73:1009-16. 23. Sawada M, Terada N , Wada A, M on Y, Yamasaki M. Saga T. et al. Estrogen- and androgen-responsive grow th o f hum an o v a r ian adenocarcinoma helerotransplantcd into nude mice, lo t / Cancer 1990; 45:359 63. 24 Nash J. I lu ll I., O zols R, Young R. Smyth J. I la m ilto n TC. Estro genic regulation and g row th iactor expression in hum an ovarian cancer in v itro |abstract| Prec Am As>cr Canter Res 1^89; 30:1189 25 H am ilton TC. Behrens BC, Louie ICC.. Ozols KF. Ind u ction of progesterone receptor w ith 17/t estradiol in hum an ovarian cancer / C/m LndtH rinol M rtah 1^84. 59.561 3. 26 Hatch KD. Beecham |B. Blessing )A. C'reasman WT. Responsive ness o f patients w ith advanced ovarian carcinoma to tam oxifen. Cancer 1991; 68:269-71 r;iH CANCER S upplem ent fu ttu o ry 5, 199J, V olum e 71, No. 2 .7 . Cei>inger KK. Berens M L , D uckett 3 , M o rg an T M , K ute TE, W e la nd e r CE 1he c ffe i Is o f estrogen, p ro g e ste ro n e , a nd ta m o x ife n alone and in c o m b in a tio n w ith c yto to xic agents against h u m a n ovarian carcinom a in v itro Cancer 1990; 6 5 :1 05 5-61 . 28 S im o n W L , A lb re c h t M , 11arise I M . D ie te l M , H blz.el F. C e ll lin e s derri ed from hum an ovanan ca rcin om as:j;ro w th s tim u la tio n by gonadotropic and steroid horm ones. / N a tl C om er fust 1983; 70:8 39 45. 29 S lo tin a n Bl. K ao BR Response to in h ib it io n o f a n d ro g e n a c tio n o f h um an o \a ria n cancer cells in v itro . Cancer Lett 1989; 4 5 :2 1 3 20. 30 (iiira v a SS. In te rs titia l g la n d tissue o f m a m m a lia n o v a ry . Acta 1ndocnnol [5up//| |9 7 2 ; 171:5 27. 31 C .urava SS. I lis to c h e m ic a l o b s e rv a tio n s in th e c o rp u s lu te u m jtre tic n m o f the hum an postm enopausal ovarv w ith reference to steroid h o im o n e svnthesis .\rc fi Ita l A nat Lm hriol I9 7 6 ;8 I:I8 9 202 '2 A n th o n y I I M A ge spe c ific in c id e n c e o f ca n ce r o f the e n d o n u *trn nn . o v a rv a nd breast in the U n ited K in gd om and U n ited S lati-s J:t / /./i i /cimi/ l y 7h; 5 :2 3 1 - 6 33 B un io n I.A , H o ove r RN. E pidem iology o f gynecologic cancers. In . H o s k in s W |, I ' e re / C A , Y o u n g RC*. e d ito rs . G y n e c o lo g ic o n * cologv: p riiu ip le s ,uu l practice P hilad e lph ia : |B l.ip p in c o tt, 1992:3 26 34 M o n ro e SE. M e n o n KMJ. Change's in re p ro d u c tiv e h o rm o n e se cretion d u rin g the clim acteric and postm enopausal periods. C lin I'L stct C vn cii'l 1977; 20:113-22 35 k a ja n ie m i 11, K a u p p ila A , R o n n h c rg L, S e la n d e r K, P v s ty n e n P. LI l(hC G ) receptor in benign a nd m a lig n a n t tum ors o f h u m a n o v a ry . A n a ( 0'>/7 Ci/uc<i/ 5iiu/ |Snp/>/| 1981; 101:83. 36 S to u ffe r KL. G ro d in MS. D avis )K, S u r iv it E A . In v e s tig a tio n o f b in d in g sites lo r to liic le s tim u la tin g h o rm o n e and c h o rio n ic g o n adotropin in h um an o varian cancers. I C lin Endocrinol Metal* 1984, 59:441 6 37 N ico sia SV. Jo hn son I I I , S tre ib e l EJ. G r o w th c h a ra c te ris tic s o f r.ib h ii o varian m esothehal (surface e p ith e lia l) cells, h it / G ynecol P athol 1985. 4.58 74 38 B e rc lu u k A . R o d rig u e / C G , K a m e l A , D o d g e RK, S o p e r JT, L la rk Pearson D l.. el al E piderm al g ro w th facto r receptor e x pression in n o tm a l o varian e p ith e liu m and o v a ria n cancer A m / O b-tC t CvIC..J 1991: 164:669 74. 39 B auknecht 1 K un g e M . S c h u a ll M , P fe id e re r A . O c c u rre n c e o f epiderm al g ro w th facto r receptors in hum an adnexal tum ors am i th e ir pro gn osiit value in advanced o va ria n carcinom as. C v m io l c h ie f 1988, 29:1 47 57 10 G o d w in A K . le s t a JR, H a n d e l L M . L iu Z. V a n d e rv e e r L A , T ra cev PA, et al S pontaneous tra nsfo rm a tio n o f rat o va ria n surface e p ith e lia l cells: association w ith cytog e ne tic changes a nd im p lic a tio n s o f repeated o v u la tio n in the e tiolo g y o f o varian cancer. / N a il Cancer Inst 1992; 84:592 6 0 1 4 1 Rodrigue/. C C . Berchuck A , W h ita k e r RS, Schlossm an D, C la rk Pearson D L Bast RC. E piderm al g ro w th facto r receptor expres sio n hi n o rm a l o v a ria n e p ith e liu m a nd o v a ria n cancer: II re la tio n s h ip b etw een receptor expression a nd response to e p id e rm a l g ro w th fa ito r Am / ( thstet Cvneeol 1991. 164 :74 5-50 . 42 R adisavljev i i SV I'he p a th o g e n e sis o f o v a ria n in c lu s io n cysts a n d ly s lo m a s (>(>// C iyiii'n/ 1976; 4 9 :4 2 4 -9 . 43. N icosia SV. M o rp h o lo g ic a l changes o f the h u m a n o v a ry th ro u g h o u t lile . In : Serra GB. e d ito r. T l'lie O v a r y . N e w Y o rk , Raven Press, 1983:57 8 1 H G od w in AK. H a m ilto n TC. K nudson A C . Oncogenes and a n ti oncogenes In. I loskins W l, Perez C A , Y ou n g RC, e ditors. C y n e - c o lo g ic o n c o lo g y : p rin c ip le s a n d p ra c tic e . P h ila d e lp h ia : JB L ip p in c o tt, 1 992:87 I 16. 45. Fathalla M F Incessant o v u la tio n : a fa cto r in o v a ria n neoplasia? Lancet 1971; 2:163 46 W h itte m o re AS. W u M L. P affen b arge r RS, Sarles D L. K am pert JB, G ro sser S. et a l E p ith e lia l o v a r ia n c a n c e r a n d th e a b ility to c o n ce ive . G iiic r r Res 1989; 4 9 :4 0 4 7 -5 2 . 47. N e g ri E, F ra n c c s c h i S, T z o n o u A , B o o th M . L a V e c c h ia C , P araz z in i F, et al. P o o le d a n a ly s is o f 3 E u ro p e a n c a s e - c o n tro l studies: I. re p ro d u c tiv e facto rs a nd risk o f e p ith e lia l o v a ria n cancer. In t / G u irc r 1991; 49:50-6 48 Francesrhi S, l^iV e cchia C , Booth M , T zo no u A, N e gri E. Paraz 7 in i F, et al. P oo le d a n a ly s is o f 3 E u ro p e a n c a s e - c o n tro l stud ie s u f o varian cancer: II age at m enarche and at m enopause. Int / Cancer 1991; 49:57 60 49. F ra n cesch i S. P a r a //.in i F, N e g ri E, B o o th M , L a V e c c h ia , Beral V, e l al Pooled a nalysis o f 3 E uropean ca s e -c o n lro l studies o f e p i th e lia l o va ria n cancer: III oral co n tra ce p tive use In t / Cancer 1991; 49:61-5 50 L o w r y S, R ussell H , H ic k e y I, A tk in s o n R. In ce ssa nt o v u la tio n and ovarian cancer. Lancet 1991; 1:1544-5. 51. W ils o n |E. A d e n o -c a rc in n m a la in hens ke p t in a c o n sta n t e n v i ro n m e n t. P oult Sci 1958; 37:1253. 52 F re d ric k s o n T N O v a r ia n tu m o rs o f th e h e n . E n v iro n H e a lth Persped 1987; 73:35-51. 53 W e in e rt T A , H a rtw e ll L H C h a ra c te riz a tio n o f R A D 9 o f Sacc/mroniyces cerevisine a n d e vide n ce th a t its fu n c tio n acts posttranslatio n a lly in c e ll c y c le a rre st a fte r D N A d a m a g e Mi/ C e ll Riot 1990; 10:6554-64 54 G e r rin g SL, S p e n ce r F, H ie te r P T h e C H L 1 (C T F I ) g en e p ro d u c t o f Sflic/idrOMiyiTS r r r r i ' is ia r is im p o r ta n t f o r c h ro m o s o m e transm ission and n o rm a l ce ll cycle p rogression in G 2 /M EA480 / 1990; 9:4347 58. 55. M u rra y A W . K irschner M W . D om inoes and clocks: the u n io n of tw o v ie w s o f th e ce ll cycle . Science 1989; 2 4 6 :6 1 4 -2 1 56 M u rra y AVV, K irs c h n e r M W . C y c lin sy n th e s is d riv e s th e early e m b ry o n ic cell cycle. N ature 1989; 3 3 9 :2 7 5 -8 0 . 5 7. C a v e n e e W K , P rv ja TP, P h illip s R A , B e n e d ic t W F , G o d b o u t R, C a llie BL. et al. Expression o f recessive a lle les b y c h ro m oso m a l m echanism s in re tinoblastom a. N ature 1983; 3 05 :77 9-84 58 I lansen M F, Cavenee W K . R etinoblastom a and the progression o f tu m o r genetics. 1rends in G rif/rs 1988; 4 :1 2 5 -8 . 59 L e v a n G , S z p ire r J, S z p ire r C , K lin g a K, H a n s o n C , Is la m Q The gene m ap o f the N o rw a y rat (K a/lus im n'cgici/s) a nd com parative m a p p in g w ith m ouse and h um an . G ernw iirs 1991, 10:699-717 6 0. Is la m M Q , S z p ire r J, S z p ire r C , Is la m K, D a s n o y J F. L e v a n G A gene fo r the suppression o f a n ch orag e in de p en d en ce is located in rat chrom osom e 5 bands q 2 2 -23, a nd the rat alpha in terfe ro n locus maps at the sam e region / C ell Sci 1989; 9 2 :1 4 7 -6 2 . 61 B e llo MJ, M o re n o S, Rey JA. In v o lv e m e n t o f 9 p in m e ta s ta tic o v a ria n adenocarcinom as. Cancer Genet Cytojfenet 1990; 45: 223-9. 62 D ia z M O . Z ie m in S, Beau M M , P it ha P, S m ith S D , C h ilc o te RR, et al. I lo m o /y g o u s d e le tio n o f the n - a nd /3 -in tc rfe ro n genes in h u m a n le uke m ia and d e riv e d cell lines. Proc N a il Acad Sci USA 1988; 85:5259-63 63 D ia z M O , R u bin C M , H a rd en A , Z ie m in S, Larson RA, LeBeau M M , et al D eletions o f in te rfe ro n genes in acute lym ph ob la stic leukem ia N En* / / M ed 1990; 3 2 2 :77 -8 2. 64 O lo p a d e O l, Je nkin s R. L in n e n b a c h AJ. C o w a n JM , P o m yka la I I . R o w le y J, e l a l. M o le c u la r a n a ly s is o f c h ro m o s o m e 9 p d e le tio n in h u m a n s o lid tu m o rs . P roc A m Assoc C ancer Res 1990; 31:318. Age Contrast in Ovarian Pathc Maria ]. Merino, M.D., and Gitie Jaffe, M.D. Background. Ovarian cancer accounts for approxi mately 23% of all gynecologic tumors and is the most common fatal gynecologic malignancy. These tumors can occur in women of all ages, but there are differences in the histologic types during various decades of life. Conclusions. During infancy and childhood, the predominant type of neoplasms are those of germ cell ori gin, such as teratomas, dysgerminomas, and endodermal sinus tumors. In adults, epithelial neoplasms, or tumors that originate from the epithelium that covers the ovar ian surface, are the most common, accounting for almost 85% of all neoplasms after the age of 50 years. The peak incidence of benign epithelial tumors occurs between the ages of 20 and 40 years. Young women (30 to 40 years of age) are frequently affected by the so-called "tumors of low malignant potential," which have excellent progno sis. Older women, on the other hand, usually have the most aggressive forms of ovarian cancer, present with advanced disease, and have a dismal prognosis. Cancer 1993; 71:537-44. Key words; ovarian cancer, germ cell and epithelial tu mors. Ovarian cancer is the fifth leading cause of death in women in the United States, with an estimated inci dence of 50 cases per 100,000 women and nearly 12,000 deaths annually.1,2 Approximately 25% of all gynecologic malignant tumors are of ovarian origin, but ovarian cancer is the most common fatal gynecologic malignancy. Unfortunately, these neoplasms remain asymptom atic until massive ovarian enlargement causes compres sion of pelvic structures, ascites, abdominal distention, or distant metastasis. Only in a small percentage of pa tients will the tumors be found at an early stage, con fined to the ovaries.3*i I' re s e n lrd at the Perspectives on O v a ria n C ancer in O ld e r-A g e d W om en: C u rre n t K no w le d ge a nd R ecom m endations fo r Research W o rkin g C onference. Bethesda, M a ryla n d , N o vem be r 2 0 -2 1 , 1991. i ron: the L aboratory o f P athology, N a tion a l Cancer Institute. N a tion a l Institu tes o f H e alth . Bethesda. M a ryla n d. A ddress fo r re p rin ts: M a ria |. M e rin o , M l ) . L a b o ra to ry o f P a th o lo g y . B u ild in g 10. R oom 2 N 2 I2 , N a tio n a l C ancer In s titu te , N a tional Institu tes o f H e a lth , Bethesda. M O 20892. A ccepted fo r p u b lic a tio n S eptem ber 29. 1992. Ovarian cancer may occur fancy and childhood. The net creases with age, with the grcai being diagnosed in the fifth a ings et al.4investigated the dis plasms by decades of life and mors were the most common more than 40% of all ovarian chance of a tumor being malig with an overall risk of maligna pausal women and 45% in eld' tients. Predisposing Factors Predisposing factors include: Tumors occur predominantly t years of age Family history Mumps parotitis Low parity Environmental factors such as Persistent ovulation The ovary is a complex s present from different germ cel lignant neoplasms can develc ments, and therefore ovarian i cording to the cell of origin. I described in Table I. There al enegs in the type of tumors foi older populations; for exampl that germ cell neoplasms pred children and young adults, w in origin are rare in this age grou derly postmenopausal women Tumors Occurring in Young Premenopausal Women Germ Cell Neoplasms These neoplasms account for ovarian lesions and 15-20% Volume 161 Number t C9*-oontroi study of pithoJial ovviar canc*r 11 A case-control study of epithelial ovarian cancer Patricia llartge . ScD,` M ark H. Schiffm an, M D,` Robert H oover, M D,` Larry McGowan, M D,` Lin d a Lesher, MHS,* and H e nry J. N o rris , M D ` H iih ru ltt. M iiiy ftin /i. fin d W ashington. D C . Wuh data from a sludy ot 296 patients with primary epithelial ovarian cancer and 343 patients hospitalized because ot other conditions, we estimated ovarian cancer risk in accordance with reproductive and other (actors. Risk was greatest among women of lower parity, especially among women who said they planned to have children but could not. The protective effect of oral contraceptives seen In other studies was observed only m subgroups ot our study population. Women who had breastfed their children had decreased risk, but the number ot months of breastfeeding was not related to risk. Incomplete pregnancies did not provide the protection seen tor live births A family history of ovarian cancer and a medical history of breast cancer were both strong risk factor None of the nonreproductive factors thet we examined, including childhood illnesses, tobacco and alcohol consumption, obesity, and selected adult diseases, was convincingly associated with risk. (Am J O bstct Gvneccx 1989;161:10-6.) Key w ords: <K .m u n u n ic i . c u rin g rtim is estrogens, in fe rtility , o ra l contraceptives, rp iilrm io liig v I In* i .iiim 'S o l m . ii i.m ca nce l .ire m a in ly u n kn o w n 1 ' ` k n o w n ris k I a cto rs a re age (in c id e n c e rises u n i 11 a lx iiii age t ill se als), race (w h ile w o m e n are at highet risk ih .in iio n w h iir w om en), low parity, in fe r* liliiv . a history o l e n d o u ie li ial o r Ineasl cancer, a fam ily histois ol ovarian c a m e l, and exposure to radiation. |*in h.ih le prntcc m e I.utens a re o ra l co ntra ce ptive s and s m g n .il m enopause. A lth o u g h the im iM iria n rc o l rep in c h ii Inc- am i co n li.ic e p liv e hisicnv lias lieen (iim ly estahlished. questions ir m a iii. Mow d o p a rity, p re g nane v. a n d I n a l loss jo in tly a lie n risk? Does the tim in g o l ih e h iM h iri h a lle c i ris k ' Is in fe r tility an in de p en d en t risk la cto i. separate Iro m the el le n s o f low parity? In add itio n, m any n o n re prod u ctive (actors have been sug gested. W ith in terview and m edical records data fro m a c .is e -c o n ito l si in I v. we have e stim a te d o v a ria n ca n ce r nsk a u e n d in g to various re p ro d u c tiv e and o th e r sug gested risk la c to is to a d d to in kn o w le d g e w ith re g a rd to these |K T sisie iu issues. M ethods Stuck m ethods aie dcst r ilx d in d eta il elsew here ' A ll e lig ib le cases in v o lv e d w o m e n a ge d 2d to 71) years w h o resided m the W ashington. I m e t l o p o l i t a n area and who w e ir lu st diagnosed In o p e ra !io n w ith m icroscop ic allv <n iilir m c d p i n n a i \ e p ith e lia l o v a ria n ca n c e r d u r in g the pet tod A ugust l!)7K to |u u e JUKI. Cases in- / n . ; , th e 1' u r llim itir r i t,, l t /.l.lrm u .lhifTS H in n r h . S u t t o n a l C .a n r r r I n iIttn le . th e lit flM .ll . ,,1 ( ; y ,,,, .,l u g i t O n to lo g y . D e fitti t i n n i i o f O h- O r lili y .iti .l ( .y u r t ,.-/CV (<" ' ' lY n x I n u g t.ii i I l in v e r s ity M r i l u t t i C e n tr i : m i t i liir .1 r n i n i l o t t i i / iix titn lr n f r 'tilh o ln g y .' X n f f h a l r , 1 Irt th r S u i t ..H u i C u m f t 1 i n ti tu le . Ci.o l i t a t i S a . N C .t- C .P - S i n 'l l l l r f i n i 't i.r r /u r ,l\ M m k 11 S th if ftn m t . S C I . Aiu't u l t v r I 'ln u t S t i t l h , H...,, I D . A h.Miti'tr* .,/ i. I letIteuhi. M l) 2<>M2. e lu d e d tu m o r s o f lo w m a lig n a n t p o te n tia l as w e ll as fra n k ly m alig n an t tu m o rs because risk factors fo r the tw o g roups were sim ila r.' For all p articipants w ith ovar ian cancer, we obtained ihe diagnostic m icroscopic slides and |> e rtin e ni m edical records. A fte r review we excluded 30 w om en fo u n d not to have d e fin ite p rim a ry o varian cancer o f the e p ith e lia l type by m icroscopic (H .J.N .) o r clin ica l (I..M .) evalu a tion . Several o f these cases h ad been cla ssifie d as m u c in o u s o v a ria n cancer hut actually were o f colorectal o rig in . We id e n tifie d 4 00 cases a nd in te rv ie w e d 2 96 w om en (74% ). Losses w ere a result o f death (n = 44). disab ility (n = 12), physician s refusal (n = 8). p a tie n t's refusal (n = 33), o r o th e r reasons (n = 7). As show n in Table I, serous and e nd om etrioid types predom inated. fControl p articipa n ts w eie id e n tifie d fro m hospital discharge lists and were m atched w ith w om en w ith e p ith e lia l o varian cancer a ccording to hospital, age. and race (T a b le 11). A w o m a n w as n o t e lig ib le to be a c o n tro l i f Ite r discharge diagnosis was psychia tric o r was po ten tia lly related to the m a jo r exposures o f interest. For all p ote ntia l c o n tro l p articipa n ts, we c o n firm e d in (he p h ysicia n a n d h o s p ita l r e t o id s th a t th e w o m e n h ad at least o n e o v a ry in ta ct a n d e x c lu d e d those w h o d id n o t l>ecause th e y w e re m u a t ris k o l o v a ria n cancer. We id en tifie d 439 c o n tro ls o l w hom wc interview ed 343 (7H%). Losses w ere a result o f death (n = 13). d is a b ility (n = 8). physician's refusal (n = 11), patie nt's refusal (n = 5 0), o r o th e r reasons (n -- 14). T ra in e d , experienced m edical in terview ers obtained in fo rm e d consent and a dm iniste re d a standardized q uestionnaire in the stud y p articipa n t's hom e shortly a fte r d ia g n o sis. (Q u e s tio n n a ire is a va ila b le o n re q u e s t.) E ffects o f o varian ta m e r risk w ere m easured by the Table I . O v a ria n cancer cases a cco rd in g to h istologic type Hulologv typt Number Percentage Scrou Serous, low malignant potential M u c in o u s Mucinous, low m alignant potential E n d o m e trio id E ndom etrioid, low malignant potential Clear ceU Mixed epithelial Undifferentiated TOTAL 84 35 17 9 76 8 12 25 30 296 28 12 6 3 26 3 4 8 10 100 estimated rate ratio-- the ratio o f ovarian cancer inci-i dence in an exposed group to that in the corresponding unexposed group. A rate ratio o f 2.0 for nulliparity, i for example, indicates that the incidence rate is twice as great among the nulliparous women as among the parous women; conversely, a rate ratio of 0.5 indicates that the incidence is half as great. The rate ratio esti mates were adjusted for the effects of confounding variables by stratified contingency table analysis (pre sented in the tables) and by logistic regression models, which gave similar results.' All trend tests used scored categorized datst and were two-tailed. For each estimate presented we adjusted for age and race. We assessed whether there were confounding ef fects o f parity, difficulty conceiving, recent administra tion o f oral contraceptives, surgical menopause, meno pausal estrogen use, or family history o f ovarian cancer, and we adjusted the estimates accordingly. I f the num ber o f eonfounders held constant exceeded three, we used a logistic regression model to produce smoothed ; estimates. To derive unbiased estimates o f the efTects of to . bacco, alcohol, and adult medical conditions, it was nec essary to include in the control group only women who were admitted for conditions not caused by (or pro tected by) the exposure under assessment. Without such exclusions, a hospital-based control series would produce spurious results because the population o f hos pitalized women generally includes, for example, more smokers than the whole population o f women at risk for ovarian cancer. -- Result - A total of 12% o f the patients with ovarian cancer and 14% o f the controls were black, and the remainder were white. T h e mean age at diagnosis was 54.4 years among women with cancer^nd 54.7 years among eontrols. Because the control participants were chosen from the same hospitals as were the women with cancer, they were similar in social dais and religion- ' `f* Greater parity was related to reduced risk in our study (Table H I) , and pregnancies that ended in spon Table I I . D ischarge diagnoses am ong control gro up Disease category Number Percentage Infectious disease Neoplasm Endocrine-metabolic diseases Disease o f blood o r blood-form ing organa Disease o f nervous syilcm Disease o f eye. ear, o r mastoid Varicose veins, hem orrhoids Respiratory diseases Digestive system diseas U rin a ry diseases Skin disorders Musculoskeletal diseases Congenital anomalies Ill-defined conditions Fractures and o th e r in juries TOTAL 6 22 19 4 20 26 5 27 55 18 6 75 4 18 38 343 2 5 5 1 8 8 1 8 16 5 2 22 1 5 1) 100 taneous or induced abortion or stillbirth did not, as a group, confer any protection. When the effects o f the different pregnancy outcomes were further separated, stillbirths appeared to increase risk, induced abortions appeared to slightly decrease risk, and spontaneous abortions appeared not to affect risk, but the estimates for the specific outcomes were rather unstable. T he re was also no clear pattern o f risk when women were categorized according to the number o f months preg nant with the different outcomes, adjusted for the ef fects o f parity. It was necessary to adjust for the effects of parity in the examination of fetal loss because the two were correlated: about one fifth o f the nulliparous women had some fetal loss, whereas about one half of the women with five or more live births had fetal loss. T h e strong protective effect o f parity was seen in almost all subgroups and was strongest among women aged 20 to 39 years and weakest among those aged 60 to 79 years. T h e protection afforded by greater parity was not explained by an effect o f earlier first birth. O n the other hand, as shown in Table IV, the apparent protection associated with earlier first birth, when this variable was considered alone, was eliminated by adjusting fo r par ity. Also shown in Table IV is reduced risk among women who had breastfed their children, but an erratic relationship is shown between lifetim e months o f breastfeeding and risk. T h e total group o f women who had taken oral con traceptives showed no altered risk (Table V), but women who had taken them as3 years were at slightly reduced risk, as were women who had u k e n them re cently. Women who took compounds with high levels o f progestogen showed no excess ovarian cancer risk (data not shown). Those who quit after only 1 or 2 months, usually because o f side effects, were not at 12 Hange et al Inly 19X9 Ain | <lim n (ynrinl V o lum e 161 Number I Cae-corn roi study of pithaMal ovarian canoar 13 T a b le I I I . f iin i:ile l i. iic i alio s . m o ld in g io re pliai n ie s , b in lis . a n d l ta l lm.se* (roup with r tin ie r ('ontrol Croup Estimated rale ratio* l" a l |M KM.m< it II 1 2 3 1 1 M in i 1M l . l 'f till ll,\ (1 1 2 t 1 le n d l.-t 1 nr. adjuU d lo i live t ill ill II 1 2 :i I n - iiil i i 70 57 19 49 35 35 89 51 71 43 39 206 5-t 24 13 71 45 71 65 37 54 p (trend) 0.10 HI 53 H5 1.0 61 p (ire iid ) 0.03 238 62 29 14 p (trend) 0 70 1.0 14 0.7 0.8 1.0 (1.7 1.0 III 0.8 0.7 <16 1.0 1.0 10 1.2 * \d|Hld I..I ,.K< .111.1 |.,,i 95% Confidence inter t i l l OH-2 4 0 4 -1 .2 11.5-1.4 0.4-1.2 0 4-1.2 0.6-1.7 0 5 13 0 412 0 4-11 0 6-16 0.5-1 H 0.5-2.7 T a ble IV . h M im .ilcri i . i l r ratios unnng | linns w om en a c c o rd in g in age at firs t b irth and m o n th s n i l>rr.ii fee d ing Age al Inst I>ilit. adjusted lo i |i.u ily ,,'(I 211-VI .itl T ie n d test Months o| IxeaS lIreiling. ,id|iitrd lo i parilv I! 10 IK 19-111 I le n d test i rouf) ii'l/J confer 12 72 5H 34 112 62 lb ii Control group 52 108 62 55 p (trend) - 0.22 121 81 37 I5 p (tre n d) - 0,11 EttimiUed rate ratio* I0 0.7 I.O I0 1.0 OH 0.5 1.1 VJ1 Confttinic interval 0 112 0 517 0.5-2 0 0 5-12 0 .2-0 9 0.5-2 6 ` \ 'l | " d l..i age and i.u e In logisn. irg ie ssiu ii im le ased risk. I he r f f e i l n | o ra l <m ill; e p liv rs was sim ii.ii fnt p a io n s a n ti n iilh p a in tis wom en, b ill ii varied som e* lr.it w ith age at diagnosis. A m o n g w om en y o iillgei th a n 10 ve .ii o ld . o ra l c o u tia t e p liv rs w e re w e akly prntei Ii\e (rate ra tio ~ 0 7). hut am ong w om en aged -10 to 59 seats, ilies* w e ie n ot ( la ie ra tio - 1.1). O n ly o ne w o m a n n ld e i th a n f*l ve.n o l age h ad ta ke n o ia l i u i i i i .ii eptive l lie c lle c is o l in le it ilit s w e re ;ieed in several wa\s M.n ried n n lln M id tis wom en had a 70% ln g lie r risk 11i.i11 m im a i'i ietl in illip a c o u s w o m e n , an excess r e l ative risk that was Ihniied to those w ho re p o rte d d if fic u lt* <o n e is in g ( L ilile V I) . A m o n g the m a rr ie d n u lliparm is patlie tp,n its t lin e was little d iffe re n ce in risk Ik-I vsecu those svho had I k v ii p ie g n a n t a nd tho se w h o had not. in a separate c o n firm a to ry question, we asked m a rrie d w om en w ho had never conceived w h y the y had never been pregnant W omen w ho planned not lo have c h ild re n showed n o increased risk c o m p are d w iili n o n m a rr ie d n u llip a ro u s w o m e n (ra te ra tio =* 1.0 .9 5 % c o n fidence in te rv a l = 0.4 to 2.4) h u t w om en w ho were u n able to lie ro m r p re gn an t had a c o rre s p o n d in g rate ra tio o f 2 .8 (9 5 % c o n fid e n c e in te r v a l = 1.1 to 7.3 ). So few w om en re p o rte d each sjie rific cause o f in fe rtility w ith c o n firm a tio n by a p hysician tha t cause-specific estimates were q u ite unstable. A m o n g p ;im u s w om en, a history o f tro u b le conceiving was unrela ted to risk (T able V I). P articipants were asked about a va rie ty o f m edical factors fo u n d in p revious studies to be related to risk o f o varian cancer O u r data o n iiie n o p a iis;il factors. Table V. Estimated rale ratio according to oral contraceptive use f>r women younger than age 60 years Never look o ra l contraceptives . Ever took oral contraceptives ' Duration (months) i III 12-35 36-59 . 60 + T re n d test i Latency (years since first taken) ' 0-5 5-9 10 14 . 15 + 1 rend test i Recency (years since last taken) 10 + 1-9 . <1 T re n d test Group cancer 115 74 23 16 >0 25 2 12 29 30 28 38 7 Control group 131 78 16 19 12 31 p (trend) - 0 76 3 15 31 29 p (trend) - 0.72 24 41 13 p (trend) - 0.75 Estimated rate ratio 1.0 1.0 1.6 1.0 0.8 0.8 0.4 0.7 10 II 14 09 0.5 95% Confidence in te r v a l 0 7-1.7 07 34 0 4 2.2 0 3-2.3 0.4-1.5 0 0-4.3 0.2-2.2 0.3-2.3 0 6-2.1 0 7 -2 6 0.5-1.6 0 1-1.6 * Adjusted fo r age and race; referent g ro up fo r all comparison* a women who never took oral contraceptive*. : Table V I. Estimated rate ratios according to reported difficulty conceiving* Nulliparous women Never married Ever married No trouble conceiving o r never trie d Had trouble conceiving Parous women No trouble conceiving Had trouble conceiving Group with cancer 30 59 32 26 164 42 Control group 37 47 32 13 209 50 Estimated rate ratiot 1.0 1.7 1.3 2.8 1.0 10 95% Confidence in te rv a l 0.9-3 4 0.6-3.4 1.1-7.3 __ 0.6-1.7 'W om en w ith incom plete responses excluded, t Adjusted fo r age and race. including protective effects o f hysterectomy and meno pausal estrogen, are detailed elsewhere. The results for selected childhood and adult diseases are given id Table V II. None o f the childhood diseases influenced risk, regardless o f whether it occurred before, after, or around the lime o f menarche. A n extensive analysis o f mumps revealed only that the protective effect o f parity was absent among women who did not recall having had mumps. .; Among the adult diseases, shingles was significantly protective but chickenpox was not. H igh blood pressure was weakly associated with increased risk. Thyroid disease was apparently protective, and hyperthyroid ism was more protective than hypothyroidism (rate ratio -- 0.5 and 0.9, respectively), but patients could not reliably make this distinction. Adrenal disease was reported by five women in the control group but no patients with cancer reported it. Diabetes was weakly protective. A history 6 f asthma was also protective, but the effects of the illness and the asthma medication could not be distinguished because nearly all women with asthma had taken multiple medicines for their condition. Previous breast cancer was strongly related to risk but nulliparity, a strong risk factor for breast cancer and ovarian cancer, was present in only one of the 17 cases with previous breast cancer. Th e re were only two histories o f endometrial cancer among pa tients with cancer compared with four in the control group, a difference that was not statistically significant. A family history o f ovarian cancer in a sister, daugh ter or mother was reported by 13 patients with cancer and by only five women in the control group, which equals an elevated rate ratio o f 3.3 (95% confidence in terval - 1.1 to 9.4). N o excess risks were seen for firstdegree family histories o f breast or endometrial cancer. ' Obesity was not related to risk. W e considered usual adult weight adjusted for height using Quetelet's index (weight height*) divided into quaniies. T h e rate ra- W -H a fig e et al July 1989 A m J ( M iri Cynccol CM-controJ study of plthnllaJ ovarian caricar 15 T a b le V I I . Estim ated ra le ra tio s . u r o r d in g 10 h is to ry o f selected c h ild h o o d a m i a d u li diseases M u nher of porlinpanli who ret tilted disease* IJivnw (roup irilh (tim er M tlllllH No Ve (iT in.in mc.iMis Nu Yet Measles No Yes ( In, L< iiix n No Yes Shingles No Yes lli^ h l.lixxl (M osu lr No Yes 1lo 101,1 il o r. isr No Ws D i.iIm'Ics No No A 'llnn.i No Ys H i, .......... , i N Yes H7 192 ? 112 II 1H1 4!> 'X 273 18 18'. '.Hi 237 51 271 21 27'. 13 278 17 * l '.u iu i|M n i\ wnli uiK O iiaiu ie '|M iiict excluded. t.\d |u sir<l lo r age and I'.nr. Control group '.2 22 124 136 47 222 59 257 28-1 38 213 97 236 78 274 27 260 26 337 A Estimated rule ttilioT 1.0 0.9 10 1. 1.0 0.8 1.4 1.0 11 05 1.0 1.2 1.0 0.7 10 OH It) 04 1.0 42 95% Confidence intenti! 0 6 14 _ 0.7 1.5 0513 0 716 _ 0309 _ 0 .8 1 .8 0.5-1.1 _ 0.4-15 _ 0.2-0,9 _ 1.4-19.2 . iio s in im le a sin g (ju a u ile s w e re 1.0. I .<1. 1.0, a n d 1.1, le spe n ivd y. As show n in l.ih le V I I I . ovarian earner risk was rc- l.n rd io d rin k in g Inn was m n e la te d to c ig a re tte s m o k in g . A lc o h o l use was m e a s u re d as th e ty p ic a l w e e kly n u in h e i o l shot glasses o l a li o lm i, ta il (It's n i Ite c i, a n d glasses n f w in e lo iis iim e d o v e i th e past 10 years, w n h to ta l .ilc o h o l t o n s in iip iin ii lo m p iiie d as th e sum o l the lin e e snhintals. T h e iu te ra tio rose slig h tly w ith ine te a sin g m ia i al o lm i c o n s n iiip tio ii. a n d w ith e.K h o l I he lin e e kin d s >l a lt o lm i o n siI r re d se pa ra tely. I lie iute ratios weie highest fo r w hiskey (rate ra tio = !.(, 059f co n fid e n ce in te r v a l = l . l to 2 3), f o l io w-d by w in e (ate rati) = 1.3. 0 5 % c o n fid e n c e in t e r sal = O.y to 1.0) a nd l> err (ra te iati = I I , 9 5 CA c o n fid en ce in te rv a l = 0.7 to I t>). Comment I h' results o l iliis in v e s iig .itio n t o n lir m th e in i|> o rU i i i r id p a rilv. in le iiiliiy , la in ily h is io i y. a nd p revious breast i n ice r in th e cause o f e p ith e lia l o v a ria n cancer. I h r le s iilts also suggest tlie i e lative iin iu i|M > na ticc of some o th e r factors associated w ith risk in pre viou s e p idem iologic investigations. Several explanations o f the pro te ctive effect o f c h ild b e a rin g have l>een p i o f fe re d , a m o n g th e m th a t p re g nancy p re ven ts o v u la tio n , th a t it in h ih its p itu ita ry se c re tio n o f g on adotropins, o r that an u m ic i lyin g prob le m leads b o th to lw p a r ity a n d in cre ase d risk. E stim ation o f the effects o f d iffe re n t pregnancy o u t comes m ay h e lp reveal h ow <h ild l> c a rin g reduces risk o f o varian cancer. II p a iity p ro te cted against ovarian cancer sim ply by the in h ib itio n o f o vula tio n, then a pregnancy w ould reduce risk a p p io x itn a te ly in p ro p o rtio n to its le n g th , regardless o f outco m e . O u r data suggest that pregnancies tha t lead to live b irth s o ffe r p ro te c tio n b u t tho se th a t lead to losses m ay n o t. Ina s m u c h as w o m e n w ith m o re liv e b irth s te n d to have had m o re losses, a d ju s tm e n t f o r p a r ity is c ritic a l. O n e p r e vious e stim ate o f th e rate ra tio a djusted fo r p a rity was 0.8 p e r m is c a rria g e .1 O t h e r in v e s tig a tio n s o f fe ta l loss are not d ire c tly co m parable, e ith e r because they were n ot adjusted fo r the effect o f p a rity o r because they m easured (he fra c tio n o f pregnancies (hat ended in Table V I I I . Estimated rate ratios according to cigarette and alcohol use* Group uitk cancer Con&ol group Giga re lic smoking Never smoked Q uit Current Cigarette smoking (d uration) Never smoked 10 years o r le u 11-30 years 31 + years T re n d test A lcohol (average weekly consum ption) 0 Occasional drink 1-6 d rin ks 14 d rin ks T re n d test 128 130 91 69 70 83 128 130 28 23 65 66 66 64 p (trend) - 0.87 109 122 49 49 63 54 36 32 34 26 p (tre n d) - 0.14 T h re e participants w ithout d a u on smoking d uration o r currency were deleted. tA d ju sie d fo r age and race. Estimated rate ratiot 1.0 IS 08 I.U 1.1 0.9 11 I.U II 1.4 1.2 1.3 93% Confidente in te rv a l -- 0.9-2.0 0.6-1.3 -- 06-2.2 0.6-1.5 0.7-1 8 -- 0.7-1.9 0 .8 2 .3 0.7-2.2 0.8-2.8 miscarriage rather than the number or months o f var ious pregnancy outcomes. Apart from the reduced risk associated with higher parity, there clearly seems to be a separate link between infertility and risk evident among the nulliparous women in our investigation. Previous studies have shown that married nulliparous women have a higher risk than single nulliparous w o m e n W e clarified this observation by showing that married nulliparous women who chose not to become pregnant were not at increased risk, but women who tried and failed defi nitely were at increased risk. Case-control studies o f ovarian cancer cannot easily assess infertility as a risk factor, because older women must recall often illdefined diagnoses from decades earlier. One follow-up study o f infertile women found increased ovarian can cer risk among subjects with nonhormonal infertility (rate ratio - 3.2. 95% confidence interval 0.3 to 32.9).* but another did not.* Both studies observed very few cancers and both derived the expected rates o f .7 ovarian cancer from the general population o f parous mand nulliparous women, so the effeett o f parity and Infertility could not be separated. A long-term follow up study o f a large cohort o f infertile women could yd. discriminate between the effects o f different kinds o f Infertility and could provide useful tiologie dues. For example, anovulation that leads to infertility should protect against ovarian cancer if Incessant ovulation is a key mechanism. Several reproductive/factors probably are not strongly related to risk. A majority o f studies indicate that the num ber o f live births and not their timing is the key factor. O u r finding o f reduced risk to women who breastfed their children confirms a previous observation1* and supports the hypothesis that anovu lation reduces risk. T h e absence o f dose response may reflect the poor correlation between the number of months a woman breastfeeds and the months o f ovu lation suppression because o f other' factor such as body mass and variability in breastfeeding regimens. Men strual factors such as age at menarche, menstrual symp toms. and age at menopause do not appear to be strongly related to risk o f ovarian cancer in these or other d a u . although the plateau in age-spedfic inddcnce around menopause strongly suggests a change in risk broadly related to cessation o f ovarian function. It is not dea r whether menopausal estrogen use affects risk. Hysterectomy does appear to decrease risk, but the association may reflea increased clinical surveil lance radier than decreased risk.11 An anomalous finding in this study was that women who u k e bral contraceptives had only a very slight re duction in ovarian cancer risk. T h e estimated rate ratios from many studies have been in the protective range o f 0.5. to 6 .1 We know o f no biases in the current study , that would diston our estimate o f the efTecis o f oral contraceptives. O ther known risk factors showed the same patterns in this study as in other studies, and women with diagnoses potentially related to oral con traceptives were not selected as control participants. From all o f the available d a u we infer that oral con traceptives prote a some women from ovarian cancer. A history o f mumps infection has been found to be a weak protective fa a o r in several previous investiga tions.' * This efTect, i f real, is certainly not strong. T h e biologic meaning is in doubt because recall o f mumps 1 i ? 16 Harlge e( al. July 1989 Am J OI,el (lyntn tl lih ln iv has I x e n ( l i i i h i i i i In- ,i p o o l in d ic a to ! o l u n t il hmIv i ile i. 1' A .u r l i l i e x a m in a tio n o l m u m p s rx|M>- M ito. s e io lo g u a llv . w o u ld I k* w o n h u ll ill*. Most o l (he o th e r <h ilI h o tid a m i a d u li diseases th a t w r t o n - siIr i o d p io h a h lv 1 n ot iiillu c 'iiio ilio d e v e lo p m e n t >f o v jiia u ta m e r. T he a p p a im i p io ic c tio n c o nfe rre d by a h istory o l shingles. in ih r .iI is c ik r o l an e lic c i o f <h u k r n p o x . lias n o ir . id v c x p l.in a lim i N o r does d ie e lle l >1 a s ilu u a . I In I n k *i| o llr t o l I i v | h -i ir n s io ii. n lx -s iiv aii dialM'Irs . . u c o id ' n id i d i r in m sislrn i liu d - ings o l p ie v io n s sim lio s In a i medic al c o n d itio n * th a t n a ir a n t h ir d io t M inis a i r d iv io id disease a n ti a d im a l disease. I h e e x a m in a lio n o l a d u lt disease as ris k l a t in i s in a hosp ital-b ase d ta s r - c o m io l s tu d y is d illic u lt In t u i i *? d ie t o n i m i g io u p "-hnuld in liu ie o n ly p a tie m s whose a dm issio n d ia g n o sis is n o i re la te d (|> osi(isely o r negai n e h | lo d ie a d u li disease lis k la tto r. Data fro m p o p u l.il lam-I taso I sin lies n id i w e ll-lo u n ie n le tl m r d i- al I i M o i ir s in n i a l.u ilv vvlie d h i ille s e tw o associations aie leal. Women n id i a pievious h is io iv o l Incast ta m e r w eie at su h sia n lia llv im le ased u s k o l o v a ria n Cancer. T h is alisei\aliaH i is m il n e w " a m i ap|M *ais in le p ie s r n t a t in e asso i.iiia iii. t >iit- goal o l ami p a in s ta k in g p ath olo g ic a n d Im ita i le s ie w was io ex In d e m e tasta tic breast a m i c o lon ta m e r; we lieliese m isdiagnosis does not e xpla in die liu d in g . A la m ilv h is to iy o l o varian ca n te r seems to pose at least a th re e - to h m r io ld in cre ase d ris k , as indicated bv o u r results a m i die results o f oth e rs.7 S m o k in g was jm i ic l.iie d to tv.n i.u i c a n te r in th is in v e s tig a iio ii. as in m ost p re v io u s w o rk , b u t th e e l le fts I al o lm i ic m iu iii u n cle a r. O u t s tu d y a n d ilh ers (e g ., I-asoli a n d Fi am est h i ) ' e stim ate s lig h tly in creased incidence ol o varian earne r am ong d rin k e rs , Inn m h ri studies ol m a iia n c a nte i Id illic i reduced in- cid e iu r a m o n g d i in ke rs (e.g., C iu in ii el al. a n d B yers r i al ) I he a vailable d a ta suggest d ia l il d i n e is an association, ii is weak. D in in g d ie past decade, t a%e-t o lili o l in te rv ie w s tu d ies w ith LMJO to rim i w o m e n w ith o v a ria n ca n ce r have o lle ic d n u m e ro u s leads, ( in ip a iisous a m o n g e x is tin g Studies suggest d ia l some o l d ie leads w a r ia n i lu r lh e r p in suit. Despite ih c ii cost. I.u g e i studies are needed to give the siatisiical |x*w ci to disem angle the effects o f S ic ilie causes o f in fe rtility , va ryin g p ale rn s o f o ra l con tra ce p tive s. a n d s|>ecific p i eguali y o u tco m e s. R EFER EN C ES 1. W rits N. Ovary. In ; S c h n u riifrld 1>. Fraumeni |F Jr. eds. ( a r n r r epidem iology anil irvc-miou. Philadelphia: WB Sam,clcis. 1982:871-811 2. ('ire n r M il, C l.u k JW . Blayney HW. The epidemiology o f varan cancer. Seminars in oncology 1984:11:209-26. 3 Kit haidvon (IS. Scully RK. N ik ru i N. Nelson j I I J r (aim mon e pillie lia l cancer o i die ovary. N Engl J Med 1983; 312:415-24 4. I la ng e P. lIcKiver R. McGowan I., (jcslicr L. N orris H j. Menopause and ovarian cancer. A m J Epidem iol 1988; 127:990-8. 5. Mc(rowan L. N o rris 111. Harlge P. Hoover R. Lcsher L. Risk lac io n lo r ovarian cancer. Eur J Gynecol Oncol 1988 ;9 :195 9 6. Roll,man K. M cx in n epidemiology. Boston: L illie , Brown, and Co.. 1986. 7. Riscli H A . Weiss NS, Lyon JL, Haling JR, L iffJ M . Events ol ir p ro d u i live lile and ihc me id riice o f epithelial ovarian cancer. A m J E pidem iol 1983; I I 7:128-39. H. Ron E. L u iirn le ld B, M e nczcrJ. et al. Cancer incidence in a cohori ol m le i tile women. A in J Epidem iol 1987; 125:780-91). 1 9. B riiilo n LA. M ellon LJ. Malkasian ( i l) , Bond A. Hoover R. Cancer risk billo w in g evaluation fo r in le riility . A in J Epidemiol 1989;129:712-22 10. Schneider AP. Risk la cio r fo r ovarian cancer. N Engl J Med 1987:317:508 9. 11. Weiss NS. H arlow BL. Why does hysterectomy w ithout bilateral oophorectom y inlluenee ihe subsequent inci dence o f ovarian cancer? A m | Epidem iol 1986; 124: 856-8. 12. C ram er DW. Welch WR, Cassells S. Scully RE. Mumps, mena relie, menopause, and ovarian cancer. A m J O u i ET (YN i.o i. 1983.147:1-6. 13. M e n rrc r |. MckIu ii M. K.mon L. (man A. Passible role o f im iiiip s vim s in ih e etiology o f ovarian earner. Cancer 1979;43:1375-9. 14. Mulli|>le prim ary ram e n in C onn e ciiru l and Denmark. Beihesda. Maryland: National Cancer [n s liu u r. 1985; D H H S pnb lica iio ii no 68. 15 Favdi M. Franceshi S. Dietary faciors and the risk o f epnhelial ovarian tam e r. JN C t 1987:79:663-9 16. ( m iu i M L . IW Iw iei I .A. Lee NC. l-aydr PM. Kuhin GL. Alcohol consum ption and ovarian cancer risk. A m | Ep idem iol 1986; 123:759-66. 17. Byers T . M arshall J. Graham S. M ettlin C, Swanion M. A case-coni m l study o f dietary and n oiK liciary faciors in ovarian tancer. J N C I 1983.71:681-6. l m m x. III * :iNMKS mm 41 u .... ..............- ..... il 'dvM,.g,1.,IK. i.>sUl.^ nV,,`,,,,,',|,|`'Sl 'S,M| U,, M" s|,,d v In ion 1 ............. i, , i i i , s i ,, i , , f .u ju d ., .......... i'1-'...... .................... ' " r ' v r," ` " 1r u l ` ,< lllll,llli M o | | | , )n | . | l |.| , , ...... ........................ ... 'V;.............. ............ ...... ..................................................................................... ..............-- K i'lr ir iu fs , :1.. . ....................... . ' ........... .............. IM ......... . ,, "r" <r vArl\ nCl.,.,.,.,,...........,.,.,, ,,,...x............ I*l.iti,"iishiji be 1 ...... .... iii'iif H 144 | ,, m iiv i i , i Ms ' i ,Mf\ || "I'.nirtiin .4 1 Al\ ('Linn A .l.lirs , I " , S ,i. , >/-,.., A I / ) , v i ' " " , i / r . n M . - w <..1 tn w u in h n i / 'V 1V l `>nliin.l. ( >/ k* t i \ t,4 r, i \ - .... ....,, , , SJ R eview Northwestern Europe and North America1* ( I able 2). The age-specitic incidence rate increases Iron* two per 100,(MX) lor women in their 20s to 5^ per 100.000 at the age of 70. ` The greatest nuinher ol new patients urs in the age group SOlo 5`i v*ars with a median age *1 y*ars The mortality for ovarian earner h i Ni'thuestrn Europe and the United Stales i*. about 7.T lo I.T per 100,(KH1, making ovarian cancer the most Ireipienl cause of death from gynecologic inahgnaiu ies. aiul the fifth leading cause ol death from cancer in women (Table 1). The age-adjtisled death rale show*, an in crease from 0.5 per 100.000 at age 20 to 45 per 100,000 at 75.' Since 19.ltI there has been a steady increase in both the incidence and mortality ol ovarian ta m e r.1 Epidemiologic investigations have not produced an adequate explanation for this phenomenon. ... ....... '1-,ml.... .,, R J ,, . "r'' ' ' " ' I ' l m n l l , , ,I,m k" X` ` '' < rn iy jn . <ivil.'i iilii^*isls / V ,.-frfvr I J Jw ,.v (|m> A | < .-Hr. Epid em io lo g y a n d etiology of ovarian c a n c e r : a review A. Peter M. Hciitlz. MO. Neville F. Under, MO. and Iwo C). Layaasv, MD E n viro n m e n ta l F a it ora That an environmental factor might b* of importance in ovarian cancer is suggested by the tact that the highest incidence of ovarian earner incurs in industri alized countries, the only exception to this observation being Japan.0 *' The hypothesis that environmental factors play an important role is strengthened bv the observation that the incidence of ovarian cancer in Japanese women born in Japan and emigrating to the United Stales increases. In the postmenopausal age group it approxi mates the incidence of American women.1 1,1A similar observation has been made lor |e\vish women in Israel: Ovarian cancer is he m ost fre q u e n t cause o f death fro m The incidence of ovarian earner in Jewish women horn i gynecologic m alignancies in the western w o rld . M uch e ffo rt in Europe or the United Stales is 14.1 per IUI.000 and has been p u t in to a tte m p ts to correlate d iffe re n ces in in c i dence rales w ith e n v iro n m e n ta l, e n d o c rin o lo g ic , and genetic much higher than that reported for Jewish women $1 factors. A re v ie w o f the lite ra tu re reveals tha t there is born in Africa In several studies, a higher im idem v currently no evidence to in crim in a te any single etiologic has been reported among women with higher educa factor fo r th is g ro u p o f tu m o rs. Th ere is g ro w in g e vidence o f tion, which correlates with higher incomes and differ fam ilial p re d is p o s itio n in a sm a ll g ro up o f patients and o f a ent dietary habits ' ,f' This, in conjunction with the relationship w ith re p rod u ctive h is to ry . I f current k n o w l higher incidence in industrialized countries and tin- edge of the e p id e m io lo g y o f o varian cancer is to be tra nslat increased incidence among certain racial groups alter ed in to disease p re v e n tio n , m ore a tte n tio n s h o u ld be p aid to women at ris k because o f the ir fa m ily history, and more awareness s h ou ld be m ade o f the protective effect o f oral contraceptives. (O bster G y m x o l 66.127, J9b5J migration to a country with a high incidence ol the disease, is also strongly in support of an environmon tal influence. It is clear that not everyone is equally susceptible lo these environmental factors. This variation m suscepti The* American Cancer Society indicates that cancer of bility mav be dependent on genetically controlled the ovary is the sixth common malignancy in women factors. Most of the investigations in this field have in the United Stales (Table l) Results ol Ihe third been focused on alimentation, products of industry, Nation.il Cancer Survey of the United States indicate and extrinsic hormones. Alimentation research has that one if every 70 newborn females will develop the mainlv been foa isnl on diets with a high intake I disease.1 The highest incidence rates are reported in meat and animal fat, because these aie the ino**t significant areas in which di*ts of indusliiali/cd ami / friiwi ihr l)in<inn nf C v > ( Hu iduyv- fV/viriMirif./1Wsrrfrii-s and (ivN.'o'f.'XV- U O A N fii'ni i*/ AIi 'i/ i. me 1w A . `<ifif.mn nonindustriali/ed countries differ, flu* results of these investigations are contradictory.4 1 1 livers el ar V j,, ,J l/ o n )..ii Vn /.-i ,,, s VOI NO 1able I. C.u iiv r lin 'h ii'iiii` .m il I V iilh Kate nt U \>m ni l .HH. r In . itU -iit (: i IV .ilfl M ir l". 1 N fr.isl I '.'l.-n u t I n i n 1'iriu s 1 HOK 1 r iik n iH .i .ml lv m p h .> m .is <Ivarv 1>IHMl V r .im ii-.s O i.il S kill i rilirts 1*18.1 i s i im .i lr .1 i .i i k .'i m . i.U-iii w einvo IK is JS nS 17 7 * 1 r> -I 1 .1 21 21 is 2i .m | i .iiim 'i .t<M lh h i s it r m w h ic h also has the h igh e st rale ot co lle e ron sum p tii*n A s im p le re la tio n s h ip b e tw e e n co flee c o n s u m p tio n and o v a r ia n c a n c e r is c o n f o u n d e d b y th e f a d th a t coffee ic a m ix tu re o f m a n y d iffe re n t n a tu ra l ch em icals and many p e o p le a d d p r o d u c ts s u c h as a r t if ic ia l sw e e te n e rs and a r t if ic ia l c re a m to t h e ir c o ffe e b e fo re c o n s u m p tio n . The c a rc in o g e n ic ity o f the se in d iv id u a l chem icals has not been in v e s tig a te d . A t th is tim e th e re is n o conclusive e v id e n c e f o r a c a u s a l r e la t io n s h ip b e tw e e n caffeine c o n s u m p tio n and o v a ria n cancer. A n in terestin g find in g is th a t o t B yers et a l,' w h o re p o rte d an increased ris k fo r o v a ria n ca nce r in a sso cia tio n w ith decreased in g e s tio n o f v it a m in A f r o m f r u it a n d vegetables s o u rc e s , t h is f in d in g o c c u r re d o n ly in th e 3(1 to 49 year age group. Im in d no re la tio n sh ip betw een l.it in ta ke a nd ovarian cancer risk. C M in e r et a l."* h o w e v e r. fo u n d tha t w ornen above (ho h igh e st qua ru le o l a m n i.il l.n c o n siu n p - n o n h a d .1 t w o fo ld in c re a s e in ris k c o m p a re d w it h those b elo w (ho low est q u a rh le . T h e m echanism s hv w h ic h inini.il i.il c o u ld c o n trib u te to the genesis of o v a r ia n c a n c e r m ig h t In* d in v ( . e ia c a rc in o g e n ic o >nt.u n in .iiH s in (lie fa t. o r in d ire c t via th e ir a h ilitv to co n trib u te to e xtrag e nita l estrog e n p ro d u c tio n fro m g u t b a c te ria .1" A re la tio n s h ip b e tw e e n s m o k in g a n d 'o r alcohol c o n s u m p tio n a n il o v .iri.in cancer has n ot h o rn ro u n d .1 There are several re p o rts C oncerning the re la tio n s h ip betw e en o v .iri.in cancer .u u l carteine consm n p d o n . T h is possible re la tio n s h ip w a s firs t re p o rte d hv S i.H k s - 1 h i 1970 .u u l la te r c o iilir m e d h v I ric h o p o u los et a l . " liv e rs e l .I,* h o w e v e r, c o u ld n o t lin d a n y re la tio n s h ip betw een ca tlein e c o n s u m p tio n a n il o v a ri an cancer risk I he possible re la tio n s h ip has been based on the fact tha t the lu s h e s t in cid e n ce rales a nd m o rta lity lo r ovarian cancer are to u n d in S w eden, Table 2. In ic m a tio n In u d i'iic c and D eath Kale <f O v .iri.n i C a ilie r' * " III. Iiil'lu t' |f l tun. u rn 1 U-.ilK .in i'* ' IlNl HU S h .-.i.-li K..IU 41 U S,\ m lu i.- i Im o . I l -m i.ii 1 *.-11. . x r.iln l\,-|'iiM u \ \ .".I 1 Ill.tliV 1*11.I r a im iii;.I> >m S.v ll/.'ll.ll.il t- M ll .. H .I H i.i/il I ih I i.i l.lp.M l N i-lh. il.in .ts Hv u: n u .* 11 11 '< ii i in n 7 ' ii 1 1 n 2 12 * VS 71 11 > sI 21 12 I E xogenous C heh u m I> The h ig h e s t in c id e n c e of o v a ria n cancer in industnal i/e d co u n trie s suggests a re la tio n s h ip betw een thi* t u m o r a n d p r o d u c ts o f in d u s t r y U n t il n o w th e re has been little e vid e n ce to in c rim in a te sp e cific products. C u rre n tly , m u c h a tte n tio n is b e in g focu se d on a possi ble re la tio n s h ip b e tw e e n o v a ria n ca n ce r a n d materials o f the talc-asbestos g ro u p (h y d ro u s m agnesium sili c a t e s ) . '* T h e cle ar re la tio n s h ip b e tw e e n asbestos e x p o s u re a nd p le u ra l a n d p e rito n e a l m esotheliom as a n d th e s im ila rity o f o v a ria n ca nce r to m esotheliom as a re in s u p p o r t o f t h is th e o r y . G ra h a m a n d G ra h a m 2* (1967) fo u n d p a rtic le s s im ila r to talc in ovaries w ith o v a ria n c a rc in o m a . N c w h o u s e ot a llv fo u n d a slightly e le v a te d fre q u e n c y o f o v a ria n ca rc in o m a in female a sb e s to s w o r k e r s . O t h e r s * 4 27 h a v e n o t c o n fir m e d this o b s e rv a tio n C ra m e r e t a l"*1 f o u n d f h . it 42.8*3 o f ovanan ca n c e r p a tie n ts re g u la rly u sed ta lc p o w d e r in the p e rin e a l area, in c o n tra s t to 2K.4*/r o f a c o n tro l g ro u p of w o m e n . H e c o n c lu d e d th a t users o f talc p o w d e r were at in cre ase d ris k to d e v e lo p o v a ria n cancer. It is of sig n ific a n c e th a t in th is s tu d y th e m a jo rity o f patients w h o s u bse qu e ntly d e ve lo pe d o varian cancer had no h is to ry o f e xpo sure to ta lc ,* 1 s u g g e stin g at best a possible, b u t sm a ll, ro le as a p ro m o tin g a gent. Presum a b ly the talc co nie s in co n ta ct w ith th e ovaries after passage th ro u g h the va gin a , u te ru s, a nd fallopian tu b e s ( p e lv ic c o n t a m in a t io n t h e o r y ) . 2'4 T h a t th is route o l c o n ta m in a tio n is fea sib le w a s s h o w n in I9 M b y Kgli a n d N e w t o n '" u s in g c a rb o n p a r tic le s a n d in 1981 by V e n tc i" u s in g labeled h u m a n a lb u m in m icrosp h rre s. I la rlg e et a l.'" h o w e ve r, lo u ld n ot dem on strate anv effect o f the use o t talc o n th e re la tiv e risk o t ovarian i .nicer, liv e n p a tie n ts w h o used talc on th e ir dia p h ra g m d id n o t s h o w a nv increased risk com pared w ith c o n tro ls . T h is o b s e rv a tio n n o tw ith s ta n d in g , talc lia s n o t b e e n s h o w n lo in d n c e t u m o r s in o t h e r a n im a ls . 14 I2N t l i ' i i . t / e l ..I / i>( t \ , i n , m I i J t l i i'f 1 ,/M-fri.> {* ( u /fio e/nyy * I talc p la y s . in v r o le in llu * c a u se o f c a n c e r, i i s t ill is n o t 7 dear w h e t h e r th e a c tiv e a g e n t is th e ta lc p e r sc o r th e ; * asbestos p a r tic le s th a t a re p re s e n t in ta lc p o w d e r . T h is } possible r o le b e c o m e s e v e n less im p o r t a n t b e ca u se o f the d ecre a se d in c lu s io n o f ta lc in c o s m e tic p ro d u c ts alter 1977. Genetic ami Familial Factors > Several s tu d ie s re p o r t fa m ilie s in w h ic h m o re th a n one m e m b e r d e v e lo p e d a m a lig n a n t t u m o r o f th e , ovary T h is p h e n o m e n o n , h o w e v e r, is m o re the ` exception th a n th e ru le M o s t w o m e n w ith o v a ria n carcinom a h a v e n o fa m ily h is to ry o f th e d isea se , a l though accurate d ata o n fa m ily h is to ry fre q u e n tly are not a v a ila b le . A h ig h e r tha n expected in cid en ce o f o v a ria n tu m o rs is fo u n d in w o m e n w it h IV u lz - J o g h e r s s y n d r o m e , 4* basal n a e v u s s y n d r o m e , 4' a n d g o n a d a l d y s g e n e s is 4" 4* i (Table 3 ). M o s t o f th e s e t u m o r s a re n o t o f e p ith e lia l origin, h o w e v e r. O n e h u n d re d fo rty fa m ilie s have been re p o rte d w it h f a m ilia l o v a r ia n c a n c e r . O f these, 9*1 w e re p u b lis h e d re c e n tly b y th e F a m ilia l O va ria n C a n c e r R e g is tr y .4' T h e n u m b e r o f o b s e rv e d g e n e ra tio n s v a rie s f r o m t w o to f o u r p e r f a m ily . In 131 of the se N O fa m ilie s , 321 w o m e n d e v e lo p e d o v a r ia n cancer There are several features th a t are ch aracte ristic o f fam ilial o v a ria n ca rcin o m a s. T h ere is a te n d e n c y to ward e a rlie r o nse t, m ean age b e in g 47.7 years, c o m pared w ith 59 years fo r e p ith e lia l o v a ria n cancer in g en eral." T h e se ro us ty p e o f tu m o r p re d o m in a te s (90 to 97'*.), a n d th e re is a tre n d to w a r d m o re p o o rly d iffe re n tia te d a d e n o c a rc in o m a s . " 4S A h ig h p e r c e n t age o f cases is b ila te ra l (70 to 8 0 % ) ." T h e h e r e d ita r y pattern in these fa m ilie s is c o n s is te n t w ith an a u to s o mal d o m in a n t gene w ith va ria ble p e n e tra tio n , tra n s m itte d th r o u g h m e n o r w o m e n . I I ' 4 M,4J l y n c h e l a l'4 **' h a v e p e r fo r m e d th e m o s t c o m p r e h e n s iv e re search in th is fie ld , A fte r a n a ly s is o f ten fa m ilie s in w hich 25 w o m e n d e v e lo p e d o v a ria n ca nce r o v e r three or fou r g en eratio n s, the y co nclu d ed tha t three d iffe re n t h e re d ita ry s y n d r o m e s e x is te d : 1) s ilc - s p e c itic fa m ilia l ovarian ca nce r s y n d ro m e -- w o m e n in these fa m ilie s are at r is k o f d e v e lo p in g o v a r ia n c a n c e r o n ly . H e n c e , a Table 3. G enetic S yn d ro m e * Associated W ith O varia n Tumors Syndrome Tumor Peni/-je^hers syndrome** Multiple basal cell carcinoma4' Gonadal dysgenesis*"'4'* Sex cord tumor O varian fibroma D y s g e rm in o in a (ionadoblasioma in streak (xy) gonads w o m a n m ay he at ris k if h e r m o th e r o r siste r has ovarian c a tu v r; 2) b re ast-o varia n cancer s y n d ro m e -- th e p e d ig re e o f these* fa m ilie s is c h a r a c te riz e d b y th e presence o f fem ale m e m b ers w ith breast a nd o v a ria n cancer, a lone o r in c o m b in a tio n . I lence, a w o m a n m a y have a 50% risk fo r o v a ria n cancer if h e r m o th e r o r sister had breast ca nce r a n d /o r o v a ria n cancer; a n d 3) the cancer fa m ily s y n d ro m e -- in the p e d ig re e o f these fa m ilie s, m ales a n d fem ales are at ris k o f d e v e lo p in g co lon cancer and , to a lesser e x te n t, o th e r a d e n o c a rc in om as (o vary, e n d o m e triu m ). C o n s is te n t w ith an a u to som al d o m in a n t in h e rita n c e fa cto r fem ales in these fam ilie s sh o u ld also h ave a 50% ris k fo r o v a ria n cancer. Hence, a w om an m ay be al ris k fo r o v a ria n cancer if one o f h e r re la tives has one o f the cancers o f th e cancer fam ily syndrom e. The risk o f fem ale o ffs p rin g d e v e lo p in g o v a ria n cancer in these a uto som al d o m in a n t fa m ilia l s y n d ro m e s m a y he* as h ig h as 50% in th e f ir s t- d e g r e e fem ale re la tives. A lth o u g h th e exact p ath og en e sis of these s y n d ro m e s is n o t u n d e rs to o d , th re e th e o re tic a l expla n atio n s are p o s s ib le u : d iffe re n t g c n e iit su s c e p ti b ility o f o v a ria n tissue to c a rcin og e nic in flu e n c e s , a gen etic d efect th a t in d u ce s a h o rm o n a l im balance w h ic h m ay be p o te n tia lly ca rcin o g e n ic, o r a genetic defect ca usin g an a ltered im m u n o lo g ic response re s u lt in g in ovarian cancer. O f m a jo r s ig n ific a n c e fo r le m a le o ffs p rin g in these fa m ilie s is th e q u e s tio n o f p ro p h y la c tic o o p h o re c to m y A ll a u th o rs a dvise th a t c o n s id e ra tio n s h o u ld be g iv e n to th is p ro ced ure as so on as ch ild be d rin g is c o m p le te d . H o w e v e r, it is e v id e n t fro m the w o rk of T o b a c h m a n et a l'1that p ro p h y la c tic o o p h o re c to m y does n ot g iv e to ta l p ro te c tio n against su bse qu e nt d e v e lo p m e n t o f o v a ria n cancer. T hey d escribe lb fa m ilie s w ith se con d-de g re e re la tives w h o had d o c u m e n te d o v a ria n cancer. T w e n t y - e ig h t w o m e n in th e s e 16 fa m ilie s u n d e r w e n t p r o p h yla ctic o o p h o re c to m y a n d w ere o bserved fo r p e ri o ds ra n g in g fro m o ne to 20 years. T h e re m o v e d ovaries w ere h is to lo g ic a lly n o rm a l. D u rin g the fo llo w u p p e ri o d , th re e o f th e 2H w o m e n d e v e lo p e d in lia a b d o m m a l ca rcin om ato sis H is to lo g ic e x a m in a tio n revealed le sions in d is tin g u is h a b le Iro m o v a ria n cancer These cases are C onsistent w ith the h y p o th e s is th a t e p ith e lia l o v a ria n ca rcin om a m ay a rise fro m a n y tissu e o rig in a lly d e r iv e d f r o m c o e lo m ic e p it h e liu m . -N G ra ha m e l a l'* w e re th e firs t to describe h y p e rp la s tic e p ith e lia l foci in the o varies o f an a s y m p to m a tic w o m an, re m o ved p ro p h y la c lic a lJ v because o f a l.u n ilv his lo ry o f o v a ria n cancer. S im ila r changes h ave U -en re p o rte d b y 1`T a u m e n i e t a l " w h o lo u i u l th e s e lo c i in three o f e ig h t a s y m p to m a tic w o m e n o l c a in c r p ro itc fam ilies. R ecently, G u s b e rg a n d D e lig d is c h '' re p o rte d a s im ila r fin d in g in g ro ssly n o rm a l ovaries re m o ved VC` I m .. n o i . | ( )I Y 1*IH' I fe in t/ et I I of ( In n M il ( i 12** Iro m three nn>H O /vj{ous sets o f tw in siste rs o j th re e p a tie n ts w ii h in v a s iv e c a rc in o m a >1 lln * iv a rv . T h e v lo m u l h y p e rp la s tic .i i h I tlv s p la * * lic 11u'i i l t . i l w e re n o t Um iiu I in c o n tr o l varies Iro n o th e r w o m e n a m i s u g g e s te d d ia l th e s e le g io n s m a v re p re s e n t a p r e c u rs o r >f o v a ria n c a n c e r. Unis w o u ld s u p p o r t th e e o n e e p t o f p io p h v la e lie re m o val o t the ovaiw *s le w is a m i O avis o n . 1' l. v r u h e l a l . '4 T h o r e l a l . " ' S k in n e r e l a l . ,/ a m i 1' iv e r et a l4' re p o rte d .1 to ta l o l 18 a s v in p to n ia tie w om en w h o u n d e rw e n t p ro p h yla ctic oop ho re ctom y. N one o l the m re p o rte il h is lo lu g u a h n o i nn.ililies m the le m o v e j varies. 1 V i a s i f i i a l l v . on an lu u l b e n ig n im 'sothelial p ro life ra tio n in the p e rito n e a l c a v itv o f h e a llln wim en. T h is m a ke s it e ve n m o re lillie u ll to le te rn n n e th e s ig n iltc a fn v o i th e s e lesins in a w o m a n w h o is p ro b a b le at h ig h ris k lo r o v a ria n e a rn e r. Im p o r tan t. h o w e v e r, is the fin d in g o l l-ra u m c m et a l " a n d ( .u s b iT g a n d P c h g ilis c h '' th a t th e re m o v e d varies w e ie in acrosco p icallv n o rm a l. It th e le sio n th e v dese nb e prves t> I v e h a ra c te ris tic >| th e d y s p la s ia th a t p i `ce de s o v a r ia n c a rc in o m a , th e n lh` o n lv w a v to liagn*se th e le s io n is t re n u iv e b o th o va ru *s. It is n o t `a s \- ti le te rm in ' h o w m a n v w o m e n w ill benefit Iro m p ro ph vK iciu * oophore lo m v . If the m o th e r o l a la n n lv lia s a n 111U 1s o m .1l lo n im a n t o v a r ia n 'aticer, the chance o l her d aughters in h c iitin g the gene to r this tu m o r is S ib ;. In th e absence o l a g e n e tic m a rk e r, th e re is n o w a v to s h o w if it is a g e n e tic o r randnn p h e n o m e n on it tw o n u 'in b e rs m o ne la n n lv are alfectei. It the ham'* o l o v a r ia n e a rn e r h ir in g a lile t im e is o n e in 70, th e h .in c e o l (h is h a p p e n in g t w u v in o n e la m ilv is a lw iu t (o n e in 7 0 ) \ ir o n ly 0 .0 2 ` `i . C u ven th e lw m rbiditv o l p ro p h y la c tic o o p h o re c U u n y a nd th e h ig h m o rta tity o l o v a r ia n ca n c e r, it w o u ld se e m p r u d e n t l iisider th is o p e ra tio n a t th e c o m p le tio n o l c h ild b e a r in g m a n v w t n i.in w h o h a s Iw* r m o re firs t-d e g r e e re la tiv e s a h e e le d w it h c a rc in o m a o l th e o v a r y , r in w h o s e la rm lv th e re is s u iln ie n l e v u le n c e I th e b re a sto v a ria n 'a n c e r s v n d r o m e o r tin * a im - r l. n m lv s y n drom e. H ow ever, tile w o rn ,in oncem cd has to be a w a re it th e in c o m p le te lata e x is tin g a t th is tim e . I.iuloi rinolo^ic /'lUfo/s M a n v a u th o rs r*port hlte re m e s m e n d o rn n e -re la te d cu nd itio ns m w o m en w ith o varian cancer com pared w it h n o r m a l m tro l s u b |e c ts . M,l4,,% " ' 4'< ' ` |lv *t a l'" a n a lv / i'i l .T9U w o m e n w it h o v a r ia n e a r n e r a n d fo u n d th a t w 'o m e n w it h th e liseas* hal .111 im r c a s c d in c id e n c e o l m illip a n t v . .1 lo w e r m e a n n u m b e r o l p re gn an cie s. <md an unieas-| incidcn- o l misMr- nag` Mc< .o w .u i *1 a l . '1' o n th* b a s is o i ,1 s iu d v I 197 w o m e n w ith o va ria n e arner, estim ate! tha t n u llig ra v i- HO I le in tz el al / / /i/rw iii/eyi/ .</ 1 K'iirn/11 ( tl/h t'i d as w e re 2.4S tim e s m o re lik e lv to d e v e lo p malignant o v a ria n tu m o rs a n d 2 .9 tim e s m o re lik e ly to develop o v a r ia n c a rc in o m a o f lo w p o t e n t ia l m a lig n a n c y than w e re w o m e n w h o h a d b e e n p r e g n a n t th r e e o r more tim e s . T he r is k o f o v a r ia n c a n c e r in t h e ir series was re d u c e d to l 27 am>ng w o m e n w h o h a d at least one p re g n a n cy. T h is lig u re w a s n o t a n a lv /e d fo r full-term p re g n a n c ie s , p r e m a t u r e b ir th s , a n d a b o rtio n s . In g'nera l in h is series, w o m e n w ith o v a ria n cancer dem on s tra te d a m a rk e d d ecrea se in th e n u m b e r o f liv e births S im ila r e xpe rie n ce has been re p o rte d bv o th e rs .7 " U n to rtu n a le lv . in th e s tu d y o l IV m o p o u lo s et a l," p a r itv w a s c a lc u la te d as th e s u m o f a ll fu ll- t e r m gesta lio n s , p r 'm a tu re d e liv e r ie s , a n d a b o rtio n s d iv id e d hv th e to ta l n u m b e r o f p a tie n ts . W h e th e r o r nil the p re g n a n c ie s in th e s t u d y o f H ild r e t h 't a l4" w e re fullte r m is nit m e n tio n e d . K is c h e t a lMI a n a ly z e d the re p r o d u c tiv e h is t o r y o f t h e ir p a tie n ts in m o re detail a n d r e p o r t e il th a t p a t ie n t s w it h o v a r ia n ca nce r, in a d d it io n to le ss f u l l te r m p re g n a n c ie s , h a d fe w e r misc a rria g e s a n d le ss to ta l t im e b re a s t-fe e d in g . Casag ra nd e et a l' re p o rte d s im ila r re su lts fo r liv e births, but a ls o r e p o rte d a d e c re a s e d r is k in a s s o c ia tio n w it h an in cre asing n u m b e r o f in c o m p le te pregnancies. Cramer et a l'v fo u n d th a t th e r e la tiv e r is k ( o r o v a r ia n canciT decreased s ig n ilic a n tly w ith in cre a sin g p a rity . Thev reported tha t co m p are d w ith n u llip a ro u s women, w o m e n w ith o n e o r tw o c h ild re n h a d o n e -h a lf the risk, w o m e n w it h th r e e o r lo u r c h ild r e n o n e - t h ir d th e risk, and w o m e n w ith fiv e o r m o re ch ild re n had one-fourth the risk fo r o v a ria n canvr. I 11 c o n tra s t w it h th e s e re s u lts . W ijn d e r e t a l' and W e s t1,1 f o u n d n o d iffe re n c e s in p a r it y o r m a rita l status b e tw e e n o v a ria n c a n c e r p a tie n ts a m ! c o n tro l subjects. O n e e x p la n a tio n f o r th e s e d if fe r e n t f in d in g s m av be th e n a tu re o f th e c o n t r o l g r o u p s . l ;o r e x a m p le . West use! p a tie n ts w ith b e n ig n o v a ria n tu m o rs as controls. A ls o , th e r is k o l o v a r ia n 'nicer is r e p o r t e il b y I lild re th e t a l4" to in c re a s e w it h in c re a s in g a g e a t lir s t p re g n a n cy H o w e v e r, m th e ir s tu d y it is n o t clear if this in cre ase d age is a n in d e p e n d e n t ris k fa c to r o r a result I th e o fte n a s s o c ia te d lo w e r p a r it v . In a ll re p o rte d series the n s k o f o v a ria n a iu e r fo r w o m e n wh* had d iffic u ltie s w ith c o n c e p tio n w a s g re ate r. Joly et a fp fo u n d th a t the ris k a lso w a s in cre ase d in m a rried but never p re gn an t w o m e n in co m p arison w ith women w h o n e v e r m a rrie d a n d w e re n e v e r p re g n a n t, a nd that th is la tte r g ro u p in so m e in s ta n c e s h a il a sm aller re la tiv e ris k th a n w o m e n w ith o n e o r tw o pregnancies, jo lv 's re s u lts s u g g e s t .1 r e l. it io n - h ip In -tw e e n th e in a b il it y to g e t p r e g n a n t a n d o v a r ia n c a n c e r O n e pssible in te rp re ta tio n w o u ld he th a t th e c n d o c rm o lo g ic status in p re g n a n t v p ro te c ts a g a in s t o v a ria n cancer, a nd that kth e lack o l th is p r o tc c tu m m a k e s in f e r tile w o m e n a ( V '-fcbiC '* ?* ("VMCu'/i'yv higher ris k fo r o v a ria n cancer. A second e x p la n a tio n could be th a t in fe r tility a nd o v a ria n ca nce r a re the result o f th e sam e a b n o rm a l g o n a d a l s ta tu s . T h is theory c o u ld e x p la in w h y in fe rtile w o m e n arc m o re at risk th a n n e v e r m a rrie d a nd n e v e r p re g n a n t w o m e n . Because th e p a tie n ts s tu d ie d w e re d ia g n o se d b e tw e e n 1957 a n d 1966. o r a l c o n tra c e p tiv e s c a n n o t h a v e b e e n o f any im p o rta n c e in th is g ro u p . H o w e v e r, these p re g nancy-associated h o rm o n a l factors a nd a b n o rm a l g o nadal status have n ot been id e n tifie d . C ra m e r et a l'v could n o t d e m o n s tra te a n y d iffe re n c e in re la tiv e ris k o f ovarian ca nce r a m o n g in fe rtile w o m e n a nd n u llip a ro u s wom en in general. T h ey h yp o th e size th a t p regnancy may in d u ce changes in the p itu ita ry g la n d re s u ltin g in a lterations o f the se cre tio n o f tro p ic h o rm o n e s a n d so in s tim u la t io n o f th e o v a r ie s . T h e e ffe c t o f n u llig r a v id ity o n th e ris k o f o v a ria n cancer is s u m m a riz e d in Table 4. H ild r e t h e t a lJO r e p o rte d a n in c re a s e d ris k o f o v a r ia n cancer in w o m e n w ith a late onset o f m e n op au se a nd decreased ris k in w o m e n w ith a h is to ry o f hea vy m enstrual flo w . These re su lts are in c o n tra st w ith the fin din g s o f W ijn d e r et a l,' w h o re p o rte d an increased risk o f o v a ria n ca nce r associated w ith e a rly m e n op au se and a h is t o r y o f h e a v y m e n s tr u a l flo w . W e s t , 1* N e w house e t a l, ,v C r a m e r e t a l , 'v a n d A n n e g e rs e t a l* 1 found no re la tio n s h ip b etw e en the age at m enopause and he o c c u rre n c e o f o v a ria n ca nce r. Exogenous Hormones Most o f the stu d ie s re la tin g to exogenous h orm on e s and o varian cancer have fo u n d no sig n ific a n t e tio lo g ic re la tion sh ip b e tw e en the tw o .4 I lo o v e re t a l'* o bserved 900 w o m e n w h o u sed c o n ju g a te d e stro g e n fo r six m onths o r m ore. T h e y fo u n d fo u r cases o f o va ria n cancer, w h erea s th re e w o u ld have been expected. There is n o e v id e n c e o f a re la tio n s h ip b e tw e e n d ie th y lsiilbe stro l a nd o v a ria n c a n c e r.'* * 1 S e v e ra l a u t h o r s 7 'MO * ' h a v e re p o rte d a lo w e r fr e quency o f o v a ria n cancer a fte r the use o f o ral c o n tra c e p tiv e s , a lt h o u g h o n ly f o u r s t u d ie s " '* * 2 *4 re a c h e d Table 4. Fa m ilial N e op la sm s o f the O vary* Cysuden<K.jrcinuma Sile-apecific ovarian cancer Breast ovarian earner syndrome Cancer family syndrome Dermoid lum or or teratoma Sometimes associated with tw inning D ytg cn n m o im Arrlienoblanloma Sometime* associated w ith mucinous cyslaJedoma and thyroid adenoma T a ble 5. Effect o f N u llig ra v id ily o n th e Risk o f O v .tri.tn Cancer A u llio i Year Ki'l.iliv r risk ( N iillig r.iv id *) '. ia v id women' July el N e w t h im - .1 .i*Annegers el .it*'1 CjKsagrjmlr el .it' Demopotilos el a l" McCiwan r l I1* Hiltlrelh ct al*" 1*74 1*7/ N7V N7V |S7V 1*7* NX) 1H 2V 1X 1 X* 23 13 2 4 17 * N ulliparity versus paniv sta tis tic a l sig n ific a n c e (T able 5). T h e m ost im p re s s iv e s tu d y w a s re p o rte d b y I he C e n te rs fo r Disease C o n tro l C ancer a nd S te ro id I lo rm o n e S tu d y .M In th e ir case c o n tro l s tu d y th e y s tu d ie d 179 w o m e n w ith o v a ria n cancer. T h e ir m a jo r fin d in g w as th a t w o m e n w h o had e ver used o ra l c o ntra ce ptive s had a s ig n ific a n tly lo w e r risk o f o v a ria n cancer. T he decrease in risk was d ire c tly related to th e d u ra tio n o f o ra l c o n tra c e p tiv e use: a fte r a m in im u m o f th re e m o n th s , th e re la tiv e ris k is d e c lin e d fro m 1.0 to 0 .6 , a n d it d e c lin e d fu r th e r to 0.4 lo r w o m e n w h o used o ral co ntra ce ptive s fo r m o re than fiv e years. T h is lo w e r re la tive ris k p ro v e d to U* a lo n g te rm e ffe ct. T h is is in a g re e m e n t w ith e a rlie r le jM irls .*4 T h ey e stim a te d tha t the in cid en ce o f o v a ria n cancer a m o n g w o m e n b e tw e e n 2(1 a n d 54 y e a rs o f a g e in th e U n ite d S ta le s m ig h t be TO'J. h ig h e r if o ra l c o n tr a c e p tive s h ad n o t been used. A n o th e r in te re s tin g lin d in g in the s tu d y w as that the use o f o ra l co ntra ce ptive s by n u llip a ro u s w o m e n decreased th e ir re la tiv e ris k b elo w tha t lo r p arou s w o m e n . A p ro te c tiv e e lle ct o l o ra l co n tra ce p tive s a m o n g n u llip a ro u s w o m e n also w as fo u n d b y R osenberg et al*4 a nd C ra m e r et a l/' b u t o n lv in ve ry sm a ll sam ples. C o n tra ry to th e C e nte rs fo r Disease C o n tro l s tu d y , R o s o n lv rg et a I*4 a n d C ra m e r e l alv fo u n d a fu rth e r decrease o f the re la tiv e ri-k in o ra l c o n tra c e p tiv e s users w ith incre ase d p a rity I Ins p ro te c tiv e effect c o u ld n o t be c o n tin u e d by Vessev et a I** a nd K o m c h a rs o n et a l.* 7 U n til n o w m e n o p a u s a l e stro g e n use has n o t ln-en s h o w n to have a s ig n ific a n t in flu e n c e in o v a ria n ca n ce r. H o w e v e r , W e is s e l al*"' f o u n d m o re m e n o p a u s a l estrogen users a m o n g w o m e n w ith e n d o m e trio id ca n cers o f th e o v a ry . C ra m e r et a lv fo u n d the sam e lo r e n d o m e trio id cancers a nd cle ar-ce ll cancers. C ra m e r also o bse rve d a n e a rly s ta tis tic a lly s ig n ilic a n t in cre ase d risk in w o m e n w h o had used m e n op au sa l e stro g e n in a cyclic ra th e r tha n c o n tin u o u s fa s h io n , a nd in w o m e n w h o had used these d ru g s fo r m o re th a n liv e years Because o l the c o n llittm g le s u lts tin the t.m m o g e n ic it y o l lo n g - a c tin g p ro g e s te ro n e s , l. ia n g e t a l'-* p e r fo r m e d a f o llo w - u p in v e s t ig a tio n a m o n g 500(1 w o m e n V O I. aa. N O 1. IU I.Y I vhs llt r in U et al / pi. Iran.'/ay i/ ,if fh u iru in < iiih c l i t T a b i, b I H,M ,,1 O ra l l..n ira .V |.||> v U ....... ,, O r a n . . , . I -H U i T U lsk >iiM- ! . . r -' li.nutv .1 a l' I .)' ; 11. . r , .|" ` i -i . . r U.-MI, 1.r4 1 -i .a 1 r l IX ' * S anili, .im t'n lv s ig n ili . 1 i 1-: i*rv |V h| |VK| l*M lux; l **.*0 I'M I |VHl .-kilter nsk II V,i-i (is-a in 'v ,'i i .li r.* 11 7 -i" < Il M ll * H bi ll HV Un Id ri Ih.m 4M V* h`l1' r`,a',u '1' |'r.>tii>u.r..iil . Inr ,,,,H r. ,,. ..................M m lu x l lo r lo u r lo n r v . ir s M ,, 'P" ' L " " ' - i r i . m . u n J o iiiu ir i. il. ,,r | . ,, . . , s l , . ,, H l.r l'*S IlM llllI o o ,.|,o r,t,s o f o i.lv 1 to < l> ; ' U n io rlu n a lu lu . ih , I r ilu n a lo r ll i u d ia g n o s is o l o o p lio r il is a ru n o i Kiv m in v i l l i u r M o d y . T ills f iu v in s lli . u w i l l , a n i i w i d . i i . m u m p s Ol 6.11 p u r IO.ll) y u .ir. II.u in o d u n u u o i is ip h o rih s m u s i b e .ih o c ji 1.2 p e r HNI.iu k i. Cramur ul a l" proposud llu l ihu mumps uiru. dainajus Ilio ovarios and ll.urol.v causus prum.iiui, Juplution ol oo.vl.-s, u liu li rosoli in in.reas.-d < , K "n-` l"'ro p i.. Oowto loss in iho animal modol is ass.Kialod .s ill. tumor growth.1* A s Ihe avail.iblo data rovual an inoi.l.-nto ol oophoritis dial n mu I. lower than llu- moldo,,.,, ol ovarian .amor ih,. tlioorv romains highly spot .ilativo. M cC -.nvan ut a l'" ru p o rlo d III. ru h u lla mluulinn b u tw o o n ih o 1211, a n d tin - IMO, v .-a r in c re a s e s the risk,a o v a ria n .a r .in o n ia lo u r lo ld . II.- .- ,, u ld n o i h n d this .......... . **' a il i n l i r l i . i l . in c h ild h o o d . T h is obser v a tio n has n o t been e o n firm e d in p u b lic a tio n s o t ulh l m i ' In h 'i li, iii ........... .... , ', , n , v r bas b oo n s lu d u -.l m o s t in o o n n o c lio ,,,U " 'I 'S v ir u s VV.-sl1'- w a s Ih u l. r s l ,, - p o r t Ih, m a .ia n e arne r p a t,u n is had a ducr.-asod ,n .,d ..,,, .. , -h n n a l m u m p s ,, the n h is lo rv , w h u n u o m p a ro .l w ill . 0,11,01 s u h i.v ts H o suW -slud p o s s .h lu p r o lu . t.v U llu, t ir o n , m u m p s v ir u s in to , lio n M u ii. v o r u t aT' I.l.n iiu d lh . il th e in v i.lo n c .- u l , .................... n ip s a m o u , o '-u .-m .a m o r p a tio n is was lo w u r, h u t ru p o rlo d i|,a Ih u lr u . |,, u n o v o l s iiK lu . ic a l .......... ... w a s lu y h . - r L v i d .-n .u lo , Ihu In llu r is basu.l ,, ,, prusun..- ,.| p ,,s ,,iv , s u n -lo y ,. O vid.-n.-u in ........... is w i l l , a n u y a liv u m u m p - h 'M o rv l l u a ls o to o , u l ih .,1 m u m p s , o n ,p lu n . u n M is in . ......... .. ............ ... . Men.p a l,u n is w .-lu lo w e r. Co, V - I h o p ,ilh e s ,/o d I h .,1 II,u se r.-s o ils ,,H u e . an u s is im y iin m i . i i. iin . o n i| iu io n . u ih .,1 i.i.-,lu ,,iu s th e du- , e li'p in i- i" Ol o v a ,.J ,, cancer p ro b a b le lliro u g h a d in v l O holo^ic ro le o l m u m p s v iru s (,, .In n .-i n l'" c o u ld n ot . . .,, in n , ihoso re su lts I ,u p .,,,.,,,' s,w .-s,u,l ,, m u m p s V ir u s is in v o lv e d in ih u cause o l o v a ria n . . " ,, e , a d e c re a s e ih o ,, ,, i. l. n , ,- o i th is d is e a s e a lte r ........ .. " 'V w in d .- p o p iil. il io il 0,0 be u s p o c lo d " a t o n c u r n in y m u m p s u llu c lio n m c o u n ln u s w i l l , ., " lv ,u u l , , u 'den e o l o v a ria n line r .ire n ot k ,` ' vn M l,m Ps '' n r . has m .im a tlm .tu s Ik s n U 's llu - p a ro tid A land. ,l n,.,v a llo ol u .h u r g l.,.,,K , ,, . | , , . li,, g .he y o n a .ls M u m p s o r.-l.ih s ,s k n o w n lo i n u r in a b o u t -> o l u lt.-o tu d m a le s I h e in o iilu n c u o l s im ila r m e o lv e - " w V '-a n rs is lo w e r ,, ,, . I u s lim a l.-.l lo he S-; . " > t. o lic ^ c o f ( . t - in - i.il I 'r.u t ilio n e r s uni llu - ................... l l , r S ll" lv Ol In lu u h o u s D ise a se s, h o w - ."ui p u b lis h e d a n lli.i.le r ,,.- ra le o l ,, lu m p s ' l " {" 1 and an iiie id e u .v ,,1 m u m p s l i l i l III (III A llh n u K h a n o th e r p ossible reason lo r p rem ature n o CM.- lo s s a n d o v a r ia n fa ilu r e is ir r a d ia liu n o i pelvic o rn -in s . n o in c re a s e d in e ,d e n .-, o f o v a ria n carcinom a has K -en re p o rle d in w o m e n w h o u n d e rw e n l Ihu the ra p y.4 I t in itia l/ ,,i O iu im in C m ii i r I." India w-,,s the first lo propose Ilial Ihe elioloev.if ovarian eancer was relai.il i ,, ioeessanl ..vulnti..,, Ih, hvpolhesis is Ih.,1 the ovarian surl.l.e incurs mm,a ua ,,,,,a w ill, each ov.ilaiu,,,. wlu. I , . .imulaliv. lv could eoi.lnbuie to llu- .l.-velopmen, ol ovarian neopl.ism >" uvul.ili.in by preyi,an,v or oral ,-,,nlr.,,,-p. live use mierrupis this process ol repealed traun,., and mav thus pr.ilecl ayamsl ovarian earner I'rcyiumcv m.,1 the use of oral contra,-cplives also are ass.Kialed " " 1 Aonadolropin levels 1 1,,- Ingl, g.inad.ilronin levels I I I II,e carlv p.islnie,,op,,us.,l years have K-en snyyesk-d i ,, K- ol |mssible siynic,m.e I,, the espeninental model, excess yonadolropm secrelion seems to be relalert lo .he develop,,,..,,. ,,1 ovaoan lumors ll.e- meessant ovulalion the,in- is also in accordance with those who have reported a higher nsk ol ovarian cancer m women with a laic onset ol menopause and a lower risk in ihose with an carlv onset II ,s ,,01 in aivorda.Hv w ill, a high.-r risk of ovarian eancer in palienis w ill, eieessive ulerine bin d mp insolar as lh,s hkv.h.ig ,s ., resull of am.e,.lali..n I he iJieones ol .ll.essaol ovolalj,,,, ,,nd ,.,,-,.ssive g,,. nadoiropm seerelion are m .iyr.i-.m-ol will, the lower risk ol ovarian eancer in p.y,,,,.. Women, and wi,h the l* e r risk in w o m e n w h o re n u llip a ro u s b y d e sire a nd N*ve used o ra l c o n tra c e p tiv e s fo r lh . it re a s o n . | io w e v it. the th e o rie s a re a ls o in a g re e m e n t w ith th e h ig h e r in i observed in u n d e s ire d in fe rtile w o m e n w h o are nullipari n is . Based o n th e s e e n d o c r in o lo g ie d a ta , th e a u th o r s must c o n c lu d e th a t p re g n a n c y a n d th e u se o f o ra l (vnlraceptives m a y p ro te c t a g a in s t o v a ria n cancer, ilthough u n d e s ire d in fe r tility in w o m e n w h o o v u la te n'gularly in c re a s e s th e r is k o f o v a r ia n c a n c e r. 1` rc g n a n c\ and th e u se o f o r a l c o n tra c e p tiv e s d if fe r Iro m vndcsired in f e r tilit y to th e c e ssa tio n o f o v u la tio n a n d the s u p p re s s io n o f g o n a d o tr o p in re le ase . pjlhotfcncfiis of Ovarian Cancer Cramer a n d W e lc h 7' h a v e t r ie d to u n it e th e d if fe r e n t riHilugic fa c to rs in o n e th e o ry . In th is th e o ry th e firs t tiep to w a r d m a lig n a n t g r o w t h is th e f o r m a tio n o f ndusum cysts by e n tra p m e n t o f surface e p ith e liu m in thr o v a ria n s tro m a , l lu* se co n d ste p is d ire c t o r in d i rect s t im u la t io n o f t h e in c lu d e d e p it h e liu m b y g o n a d otropins a n d e x tra - o r in tra g la n d u la r estrt>gen, re s u lt ing in d if fe r e n tia t io n , p r o lile r a tio n , a n d s o m e tim e s m alignant tra n s fo rm a tio n . H o w e v e r, th e fin a l s tim u lu s lo m a lig n a n t t r a n s f o r m a t io n re m a in s u n k n o w n ( f ig u r e 1). In t h is t h e o r v a ll fa c to rs th a t c a u s e in c lu s io n c y s t form ation, h ig h e xtra- o r in tm g la n d u la r e stro g e n and factors th a t ca use h ig h g o n a d o tr o p in le v e ls , m a y m a k e i w om an at risk fo r o v a ria n eancer. C o n se q u e n tly, factors th a t p re v e n t o n e o r m o re o f th o s e s tim u li m a y protect a g a in st th is disease. Prophylaxis for O varian Carcinoma T h e g r o u p o f p a tie n ts w h o a re it h ig h ris k f o r o v a r ia n carcinom a can o n ly he d ete cte d hv a c a re fu l fa m ily h is to ry . The use ol o ra l lo n tr.u e p liv e s fo r b irth c o n tro l m ay h e lp to decrease th e ir risk, a n d p ro p h y l.ie tii o o p h o re cto m y a fte r c h ild b e a rin g appears to be ju s ti fie d. H o w e ve r, the a u th o rs agree w ith C.usherg and D o lig d is c h ': that the w o m e n conce rn ed have lo he awire o f the in c o m p le te d ata e x is tin g at th is tun e . D ie tary m easures lo p re v e n t o v a ria n c .in in o n i.i c o u ld in c lu d e a lo w m t ike o f a n im a l t .it, a n d >i h ig h in take o f v ita m in A . iltho u gh tlu - e vid e n ce to s u p p o rt these re c o m m e n .la lio n s is m iu II.1 |*o r th e g e n e r .il p o p u la t io n . 1 p r o p h v l. ic t ic a ffe c t o n th e in c id e n c e *f o v a r ia n c a n c e r m ig h t b e e x p e c te d sh o u ld b ila te ra l o o p h o rc c to n tv be p e rfo rm e d in all p atie nts 40 years o f age o r o ld e r re ce ivin g p e lv ic la p .ir o tn m v f o r In - iu g n le s io n s . *' lln w c v e r, as S ch w e p p c- a n d I V t l e r " calc iila le d . to preve n t o n e o v a r ia n c a rc in o m a 70 0 w o m e n w o u ld h a v e to Ih * c a s tra te d . So d e c is io n m a k in g a b o u t th is su bje ct ln -co m e s a m a tte r o f p h ilo s o p h y a n d |H*rsonal a lt it u d e . " In the a u th o rs' o p in io n the d e cisio n sh o u ld Ih- w e ig he d aga in st the risks a nd th e p e rso n a l a n d gen eral p ro b lem s a ris in g o f th e im p o s itio n o f necessary lo n g -te rm estrogen s u b s titu tio n . Conclusions lip id e m iolo g ic a nd eno lo gie research on o v a ria n cancer lia s h ith e rto in flu ite ti a ll te ll iv p c s . A s the large m a jo r it o f d ie Uiiiics ire o l i- p it h e li. il o r ig in , th e HI CHI GONADOYMOVIN LKVKL9 - r e t e C1v i o f p e r tu r o varian fa ilu r e - r e a c t i v e Oi. I n c r t s a t d a c r o 9n d q o c i U o ri in th e l i v e r , ca im e d by c h e m c a la o r aru qa - d i r e c t a t im o l a t I o n (C lo m lp h e n e I ? . HIGH IN T R A GLANDULAR CSTROGLN IN C LU SIO N CYSTS ------------- > OVARY M ITH D IT fT R C N T IA T ION AND P R O L IF tR A T IO N OF --------- > M A LIG N A N T t - xoqcnoui - d ie t - o b e s ity Figure I. llu - >1 o v a ria n i. iiu e r a u o r tlm g u> I r.nu i anl V .V L Ii ' H2 h e in i/ i t j | / /"./ewialayil < V's/l /fltN V <'|/Oi>/|yy V O I i. N i ) I. I I.Y I U n n i / r l ^1 I /'n il m i i ' / i v v < ,r , n i , i n l . i n i 1 1.1 in c id e n c e o l e p ith e lia l o v d ri.m c a rc in o m a is close to th a t o t o v a r ia n c a n c e r in g e n e ra l. l :.p id e n iin lo g ic a n d tiologie studies have id e n tifie d a g ro u p o t risk (actors that appears to ho d ire e tlv o r m d ire c llv related to o varian cancer. These lactors can he g ro u p e d in lo u r m am categones, genetic, e n v iro n m e n ta l, h o rm o n a l, and vira l In the p ie se n l slate o t k n o w le d g e , it appears (li t the e tio lo g y ot o v a ria n cancer is n u iltifd c to ri.il. I (torts to r e arlv d iag n osis a nd p ro p h y laxis, w h ic h seem to he expensive (or the w h o le p o p u la tio n , are jvistihed in the sm all g ro u p o i w o m e n w h o m .iv he at h ig h iis k to r tin s le th a l disease. F ro m a c lin ic a l v ie w p o in t. it is im p o rta n t to re a liz e th a t a c a re fu l fa m ily lu sto rv sh o u ld he taken to id e n tity d a u g h te rs w h o are at h ig h ris k o t d e v e lo p in g o v a r ia n c a n c e r l ;o r th is g ro u p ot p atie nts the use o l m a l to n lr.ite p liv e s lo r I'n Ih lo n iro l n iav help todecreas* the risk .iiu f p ro p h v lactic o o p h n iv d iim v a lte r c o m p le tio n o l c h ild K -a rin g a p p e a rs to hi* |u s tilie d , a lt h o u g h e v e n w it h o o p h o re c lo tiiv these p a tie n ts re m a in at ris k lo r c a rcin o m a s o f extragonadal coelonuc e p ith e liu m References 1 i i i 'li'f S| y m in ^ || i-iK Ih ird N<ilion,il l iihci ' u n r v lii.td e iitV il.ilJ , fs.nl l .11.er In s l n .i - tl. I >111 IV p til'ln .iiio n ( N I I I ) 7S -'S7 \\a s h in g lu ii L1 S g o v e in ilie n l p ilo tin g o ftn e . I**7S 2 S il rt(v i|; I . U rh fw |A A review <>l Am crnan l .im .'i N v ie lv estimates .I . Jm>*i 1,1m s iti i .1 deaths ( A A l .inter I C ini h ). |*IK1 t KoUi.ul I'. Jleetham K' I pideiiuologv <>1 ovarian neoplasia IV.s cedtiigs .I ihe Ann-iii ai-I tuope.rn ( onleieruv on |ln- ovarv M .xm eiix StMl/a-rl.iiul. I xrrpia M tslu.i In i Congrrss N-nes V a 1^7-t -t I ingeman ( M I m ii.irtnionl.il la. lois in lit.- etiologv ol . .ir.mtiiii.i .*1 lin* lui mail .it are *\rn I Indnsl Med 4 ,V5, Jus l s V\i|i)i|i>i | ( , M.ids I I liai her 1IKK. I p ije n tii 'l,.g\ olt.im e ol ihe i-v.iit l ,m.er 2 1 4SJ l*. 1 ingom.iii A M J ri*>1. <>| i .i i u . i ol lIn- luim .in ovarv A .re v ie w | N all t ain.-i Invi s i Jm O. |x: l ? I .i-..i)-r.ui.lt- | | . 1'iki- M l . K o " KK i*i .il ` hwessanl m nl.iiion" in.l .it.iii.in .arn.-r I aiu.-l n J,`ri I'*.*'* 8 ( i-nlrrs loi | htiMM-1 oiiiiol l an. -r amt 'loroiil I l.irnione S im |\ Oi.il in iilt ie p tn e use .nut the risk ol ovan.ni ta n .e i IA M A 249 l'-I'M. |m i ( ramer IHV. H un h in t.<n (.IV W elth Wtt. ei al Factors allot img ihe iisstMaiinn t>l t>ral ttin ii.u epirt es amt ovanan. aiu.-r N I ngl | Me. I VJ7 Ii 147. 1VH2 Ml l ullei SJ. 3 oung }i eds 1hint iialloil.il <arivei suivov liH i.tem e tl.it,i Nail la iu e r Irisl Monog 41 1. I `*75 il I Vnio|milos U|. Neliyyr |. I iiihin N .-i ,il Hie .issm i.ition ol p.i i <\ .In.l maniai status t%ill. the development ol ovarian t .iit i rmma t Im itai implu alunis ( l|>siei i.v n e m l S.) |Su J>i79 17 M a in e -ih i S. I a Ve.tln a ( lle lin ri.li SI', el al Kisk latl..i> lt>r epilheti.il in .l nan tam er in ll.ilv Am I I puliim ol IIS 714. 1*187 P I-r.iirmem II- l.iu iu l.iv l.W . I icagan I I el al S\ laimlit's probe lo ovarian . .inter l .nu.-r li. l , |m?S 14 I lil.li. k Sunlit <A Hie ISiHogV o f lait Mi | linlusl M n l M J |7 . 1*0. |S W eiss NS ( .,, I e p u te im o lo g V a n d p n - t I d ried hv I 1 IJ4 H e in tz el al / /'iJiw io /M yi/ o/ ( )n iru n i C u no 'i Stlii HenIel-1 and |K Frau mem. Philadelphia. W it Saunders. I98J In West R l)- F.pideiniologtc sttidv ol malignancies ot Ihe ovary ( aiuer 1*4 UKU. I*4r# |7 livers Marshall | (Iraham S. el al A vase-vonIn4 study ot drelarv anti norulielarv lactors in ovarian cancer |N ( 'I 7 M l. |W i IS l tamer l)W . Weleli KW , Huivhmsi>n (>|t. el al Ihelarv animal lit n relation i. ovarian eancei risk O l'.tlri (.v n e to i r>3 833, I1K4 I** Ne.vhouse M l . Pearson KM . Fullerlon |M A vase-conin4 siudv ol .artultim a *1 the tv.irv I'r I 1`rev So. hied 4| |4H. |*>77 20 Slade! BV. Ih e etiology ,nu| prevent ion t>l ovarian cancer Am I Ohsiel C.vnecol 12.4-772. I*#7* 21 Sucks P. Cancer morlalilv m relation in naimnal consumption <4 ugaretle. solid luel. lea and coltee. Hr I Cancer 24 215. 1*4711 22 IrnImpoulos l). Papaposioloti M . Pnlychrtnnupoulou A: CnMrt and ovarian cancer Ini | C .liner 28-fV|. 1*481 24 C ramer DW . Welt h W it. Scully KK. el al. Ovarian cancer and lair ( aiuer Vi 472. 1*482 24 Heiiderstin W |. lianullon 1C. l.rirtiih s K: Talc in normal anJ malignant ovarian tissue l..meet t 4*4*1. 197*4 2** I oiigon I >1 . 4 oung K l` C osnielu' talc and ovarian earner l.anrt-1 n .44*1. 1*17*4 2t> (a a h a in |. (irdhani K Ovarian cancer and asbe>Us Hnviron Rn I 115. I*4h7 27 Km- KK Conlrt*versv. Cosmetic laic and ovarian cancer lanrri ii 744. 197V 28 li.irtge P. Utm vei K. I.eslier I P. el al: late and ovarian cancer |A M A 25U 14. IVM1 2V I'armlev I'll. Woidrulf |(). The ovarian mesolhelioma Am | (4hsl,l C.vneCot 120 214. |V?4 10. l.gti C*K. Ncw lon M. Ih e lra n > p n l ol carbon particles in ihe human (eniale reprtKlticlive tr.uA. Feriil Slenl 12151. l*4t*l .41 Venler I'K Ovanan epnhehal canrer and chemit al carcinogenesis Gynecol Oncol 12 2HI. 1981 .42 l-etvis A O .. Davison BC( : Familial ovarian cancer, l-ancel 2.4$. I*M9 1.4 l.i I P. Kapoporl A M . Kiaunieni )F. el al Kamiiial ovarian caronn- iii.i JAMA 214 ISS9. J97II 44 I vnvh M l . (hums M A , Alherl S Familial asstK'iation ol carcino ma t>i the hieast and ovarv Surg (jynet'ol Ohsiel 1.18 717. |iT4 4^ McCT.tnn I) , MartTi-irl l) |. B.irtl.iwil 4VA Ovarian carcinoma in iliree teen age siblings ( U^lei (iynecol 43 132. I*J7t> ,V I hor I . IVrsson MIL hit-ssli-ill I annhan ovarian canrer Upsala) Med Sc Ml IHV. 1*47#. 37 Skinner jl . Oats JIN. Svnionds |-KI. Familial ovarian carcinoma J K C.ill Cien I'rar 27 U.9. Jx77 3x l.uiain |R. I'ivei MS l annli.il i.varian carcinoma ( iynecol OnoT K18S. |979 .I** Philipp IT Familial can inoina of ihe ovarv. C .ise repri Hr J ( >*siet Cvnaettil 86-1^2. IV7V 40 Hildreth NCi. Kelsev 1C, l.iVolsi VA. el al An .epidemiologic siutlv tl epithelial .arcinoina ol ihe ovary Am J F.pidemict 114 3*4H. 19 8 1 41 Franct-schi S. I.a Veccliia C . Mangiom ('- Familial ovarian cancer Light more families Cyynerol Onc*l 13 31. P482 42 l.viu h MT, Albano 4VA, J vm h JF. el al Surveillance and man agement of patients at hi>;h genetic risk lor ovarian taicinoma Ohsiel ( ivne ol S9 S84. |9K2 44 Piver M s, Harltiw |J I .until.<1 ovarian lancet hit teasing in Iret|tivn tv> CK'sit-i (.v n e ttil Wi 497. 1982 44 lin k e r M IIM . l-rver |A . S nnulers D M Ovarian cartinoma w nleniit.il lu-ins Ansi N:7 I Ohsiel C vn.nvtil 22 113. |9H2 4** Piver MS. M elllm C |. Isuka.ia >. el al lannli.il ovarian tainn legistIV llhslet (.Mkt-tol rd |9S |9K4 4n l>n/ mis KK. Kempers Kl> M .iM m IK ', el al ( >vauan liimi.rn associated with IV n l / J i . g l i e r s svV' lin lm n ie A n n S u rg 172 2 4.4, 19711 i (1O lK tc trn s Gi/MCrVi/iiyv I A V . Berlin N l. Van Scon F.J, Clendenning W E. f t al: Basal cell nevus syndrome. A nn Intern M ed 64 403. |966 8. letor |, Hoc/kowski K: Occurrence of tumors in dysgenetic gonads. Cancer 20:130!, 1967 8 Goldberg M B . Sculby A l.. Cionad.il malignancy in gonadal dys genesis. Papillary pseudomacinous cvsiadciiocnrcinonia in a pa tient w ith Turners syndrome I O in Endocrinol 27.341. 1967 50 McGowan I.. I'areni I., l.ednar VV. el al The woman at risk (or developing ovarian cancer, (ivnecol Oncol 7:325. 197* 51 lo bachm an |K, Tucker M A . Kane K. el al: Intraabdominal carci nomatosis alter prophylactic oophorectomy in ovanan cancerprone families, lancet it-795, 19H2 S2- Gusberg SB. Defigdisch I.: Ovarian dysplasia, a study of indentical twins. Cancer 54 l. 1984 S3 Berg JW. Bavlor SM: lh e epidemiologic pathology of ovarian cancer. H u m Pathol 4:537. 1973 Sf Cramer D W , Oevesa SS, Welch WR. Trends in the incidence of endometrioid and clearcell cancers of lhe ovary in the United Slates Am J Epidemiol 114:201. I9M1 55 Cramer i)W . Welch W R. C assells S. el a)' Mumps, mrnarche. menopause and ovarian cancer. Am I Obsiet Gynecol 147.1. I9H3 56. Hoover H, Gray l-A. Fraumeni JF: Stilhtiestrul and lhe risk of ovarian cancer. Lancet u 5.3.3. 1977 57. )ames PD: Epidemiology of ovarian cancer. I-m cel i:4!2. 1976 5t* Jolv D |. I.ilenfield A M . Diam ond FI.: An epidemiologic study of the relationship of repioductive experience In cancer or the ovary. Am | Epidemiol 99 190. 1974 59. Cramer D W . Hutchinson GB, Welch WR. et al. Determinants of ovarian cancer risk. Reproductive experience and lamily history. JN'Cl 71:711. 1983 60. Risch M A . Weiss NS. Lvon |F. et al: Events of reproductive life and the incidence of epithelial ovarian cancer. A m J Epidemiol 117.128. 19K3 61. Annegvrs JF. Strom H . Decker D C . ei al: Ovanan cancer: Inci dence and case-control study. Cancer 43:723. 1979 62. Weiss NS. Lvon JL. Liff JM, et al: Incidence ol ovarian cancer in relation to the use of oral contraceptives. Ini | Cancer 2 669. I9N| 63 Hulka HS: O ral contraceptives- the good news Editorial. JAMA 12.1624. 1V83 M Rosenberg l,, Shapiro S. Slone D. et al: Epithelial ovarian cancer and combination oral contraceptives |A M A 247 3210, 1982 65 Willel W C, Bam C, Hennekes C M , el al. O ral contraceptives and risk ol ovarian cancer Cancer 4K 1684. 1^81 66 Vessev M l . D ull R. Pelo R. et al. A long term follow-up study ol women using different methods of contraception: An interim report. J Biosoc H:375. 1976 67 Romcharson S. Pellegnn f A . Rav R. el al: The Walnut Creek contraceptive drug study. A prospective study ol the side eflexls of oral contraceptives. N IH publication 81.564, |VH| 66 Liang A l\ Gieenspan-Levenson A, I avde I 'M , el al Risk of breast, uterine corpus and ovanan cancer in women receiving medroxyprogesterone iii|ections JAMA 249 29(19. 1*90 69. M r n r/er J. Modau M . Kanon I . et al Possible role of mumps virus m lhe etiology ol ovarian cancer. Cam er 43 1375. 1979 70 Ciolan A. Jousting A C C . Orchard M l- M um ps virus and ovarian cancer S A ir Med I 56 20. 1979 71. M arry SM , Kilbrick S: M um ps in inlectiotis diseases l.diled bv PD lloperich. Philadelphia. Harper & Row, |9Mt. p 228 72 Research Unit ot the Royal College of (General Prat im onrrs I lie incidence and t(implications of mumps |K Coll G en I'rac 24 545, 1974 73. Association tnr the sludy of Infectious Disease. A retrospet live survey ot the complicalions ol mumps. J K Coll O n Prac 2* 552. 197 74. Falhalla M R Incessant ovulation. A lactor in ovarian neoplasia? Lancet it. 16.3. 1971 75 Cramer DW , W eh h WR: Determinants ol ovarian cancer risk I I lliterleteni.es regarding palhogenis JNCl 71 717. 198 3 76. Te r/ JJ. Harber I IKK. Ilrunsi hwig A Ini idem e of <an moina in the retained ovary. Am J of Surg I l 3 -5 | l , 1967 77. Schweppe KW . Beller EK: Prophylactic oophorectomy. Geburlsh Frauenheilk D 39 1024. 1979 7 Horvath 11. Biedermann II. Schindler Al liin io r of the ovarv alter hysterectomy. Geburlsh Erauenhnlk I ) 40 1987. |9Ktl 79 v Hall EV: Hysterectomy. Ib e controversial climacteric Edited bv PA v Kepp. W II Utian. A VermeuJen. lh e liaque. M i l 1press ltd. 1982. pp 73-79 A dd re ss re p rin t requests to. A . Peter M . Ile tn lz . M L ) 7 /ir N etherhviib Cncer Institute Plesitianlaan 121 1066 C X Amsterdam The Netherlands S u b m i t t e d for piW'fn ir/nur A p r i l 19. 1984. R r n iH 'd /V ivw ifyr 1 0 . I 984 A u e p t e d lo r p u N u n l i o n l u n m i n i 18. 1985 * Copyright *' 19M5 bv The American 1 1.liege ol Ohstelruians and Gynecologists. i! 9 ' { I - VO I 6 *. N O l. JUI Y lTK5 M e in ! / e t a l / {'id c rm o /o y t/ of f fpor run ( r in ic i 35 Ini I 5-1. n ' s .n M i n i l i 1 ,s- I ik Ulc, an /.v n in o Moni i. or ovarian eimtiihliai. carcinoc.enesis: ( H A N O I S IN C El.l.-C E Lf. C O M M U N IC A T IO N A N D A D IIH S IO N (H 'CU R RING DURINO N liO I' LASTIC RROGRLSSION A rili>n ( > II I i vi \ n"` . U n iv ii ( t f i u . i i v u o i 1 '."R o jjc P T T iT T S r *~aml N c lly A p i KM'i l o . ' 'P t /h u im i tu >/1 <i u m ili in a lin ii\ m u l 1`uhht H ca lih , lc \ a \ .IA A / l'n n v r.u iv . ( iillc iji' S iaim n. T X 77S4 I. US. 1. a ih l :lK '/u iilim -n t <! Inaloniw U n ii a u l o f H n li\h C o lu m b ia, l i m e n in e ;. H.C.. C iin iiih i. To investigate the cellular mechanisms of ovarian epithelial carcinogenesis, a series of progressively transformed rat ovarian surface epithelial (ROSE) cell lines were developed and studied. Transfection of primary ROSE cells and an im m ortalized ROSE line (ROSE 199) with the pSVlneo plasmid (SV40 T-antigen) yielded transformed lines which retained epithelial morphology. In vivo selection of these pSVlneo cell populations resulted in further phenotypic transformation. Transfection of ROSE 199 with pSV2neo/c-H-rot *' (ra t11p 2 l) resulted in a malignant line which appeared fibroblast-like and formed invasive sarcomas both in athymic mice and in im munocompetent rats. Gap junctional intercellular communication (G|IC) and cell-cell adhe sion were studied in this series of ROSE lines. Both c -H -ro tC|transformation and in vivo selection resulted in a significant reduction of GJIC between adjoining cells and a transition of in vitro migration as continuous epithelial sheets to the dissocia tion of individual cells. This apparent shift in cell adhesiveness was associated w ith reduced expression of the E-cadherin adhesion molecule. Our data suggest that neoplastic progres sion of the ovarian surface epithelium may be associated w ith concomitant reductions in GJIC, E-cadherin expression and functional adhesiveness. /W l II//. i /h v hi, O v.i M.i H c.meei is I lie lead in g s',him / o i d e a th slue to iii.ilip i.m e v i>! ih e g v in t.il iriiv 'l .nisi ih e t ilth m ost co m m o n e.mse h i cancel ile .iih m e i. ilt in A n ic iic .in w o m e n (Ito r m g e f a l. l*/*0). A mjjiiiIu .m i l.is im c o n iiib n im g to the high m o rta l ity i a le is l lie lack >1 specific early sym ptom s m .u n le s lc il by I lie ilise.ise ( 'o iis c ijin nils, niosi p .ilic u ls aie diagnosed w ith slagc- I I I in sta g e -IV disease w hen m eiasiases beyo nd I lie pelvic easily .lie co in in o n .u n i SO in 9 .V , ol ilie se vvoinen w ill she vviilim ve.ns ( I ' e iie is o n c l a l . 1` isS). I `n lo iIn n a te ly , f ills la ic p ie s e iil.iliu n lias liim ic t l u n i iin d c i sta n d in g I the c e llu la r m eih.inisnis o l ov.ui.m carcinogenesis and metastasis ia m u O l ihe diverse cell types whieh com prise ih e ovary, all have lhe capacity io u n d c ic o neoplastic ir.in s lo rm a iio n : however, ihe eonniion epiihe h .il m inors are the most prevalent o l ihe histologic l v|x:s o l o v.m a n Illin o is w in ch have Itc c n desenbesl. accounting lo i Sii io 1 o l a ll ovatian m alignancies (S cully. W?`J| These c p n h e lia l tu m o rs are th o u g h t to arise fro m the neoplastic ir.in s liu m a iio n o l the ovarian surface e pith eliu m (O S I.) (S cu lly. l*7/j, a d e riva tive o l the e c lo m ic e p ith e liu m winch covers the em bryonic gonadal ridge d uring develop m ent D espite its im p o it.in e e . the n u m b e r o f stu d ie s in v e s tig a t ing Ihe in H in t s iillm e ! ih e in a iiiiu a lia n ( )S1 * is lim ite d (11.1111111011 cl a l. in s il; A d .m is a n il A n e is p e rg . I*>K. |9.H$; (o.lw m , i a l . |'i'i_ '|. W InJc 11i t s i si tid ie s have p io v iile d v.iluable iiisighls in to our evolving knowledge o l ih e biology o l Ihe O SI . lew li.iw in v is iig .iie d ihe n eo p l.isiie li.u t s lo r n i.ilio il ol the t >S1- in ia n i (A d .m is and A iie ts p e ig . | W | ; ( io d w iu .7 a l. I `` 2.1lo in In |I92|. Most, ti not .ill hm n.in and anim al cancers develop hi a sicpw ist in.um ei hom ihe aecuniulaiion o l m u ltip le genetic lesions w it Inn the i a a c tio n s c ells, thus e o n le rriu g several nevy phenoiypieehaiaeiensties which distinguish them Irom noi mal cells ( K le in am i K le in . |9K>). |*ei haps the m ost serious o l these is the a bility ol in d iv id u a l cells to detach Iro n ) a p iim a ry tu m o r and spread to d is i.m l Mies; the process o l metastasis. It has K e n suggested i l u l loss u|' gap jm ic iio n s a nd o th e r ce ll- adhesion mechanisms between tum or cells may p erm it deta ch m ent a iu l dissem ination d u rin g m alignant progression inwards metastasis ( Larsen r ia l.. I9K7; la k c id ii. 1991) (a p ju nctio n s, w hich consist o l m em brane channels w hich connect the cyto plasm ic co m partm ents o f a djo in in g cells, are thought to fun ctio n p i m anly in the m aintenance o f tissue hom eostasis and the exchange o l re g ula io iy m olecules necessary fo r the c o u iio l o f n o rm a l c e llu la r g ro w th ( I o c w c n s tc iii. 1979). I i lias thus b een h yp o the size d th a t loss o f gap ju n c tio n a l in lc r c c llu la r eonm um icalion ((JIC ) may in ie ile re vvitli n o in ia l c e llu lar g lo w th re g u la tio n (l.n c w c n s ic in . 1979). In d e e d th is th e o ry has been su pported by o bseivalions that many, but not all. tum or c e lls d is p la y re d u ce d f iJ I C ( I .o c w c n s lc iit. 1979; K la n n <7 a l. |9X;j. I io w e ve r. stud ie s w h ich ile in o iis ira ie re d u ce d ( J IC in cells that are already neopl.isiie may n oi be sulDcieni lo r precise u iid c rs ia n d in g o l bow e o m m im ie .iiiiu i is u ltc ic d d u rin g Ilie n atu ra l carcinogenic progression fm m n orm al to neoplas tic cells. A nalvsis o l ( i J I ( ` characteristics am ong piogressively tra n s lo riu c d cell lines niuy p io v u le u seltil in lo im a iio il reg ard ing how . . iiii 1 at vvli.it stage, c o iiim im ic a iio n is a lte r e il d in in g neoplastic p iogicssion. C e ll-cell adhesion plays an m ip n iia n l p a ri m m o ip ho g ciie sis. m a in te n a n ce o l tissue n r c h ilc c iu ie a nd in u m m o lo g ic fu n c tio n (S h im o yam a c l a t. 19X9; T .ik e ic h i. 1991). A s n e o p la stic pop ulatio n s p iog ie ss tow ards m alignancy, in d iv id u a l cells acipure ihe a b ility to detach lo m i the p rim a ry tu m o r and metastasize. D isturbances in adhesion-m olecule a ctivity may play a sig n ifica n t ro le in the release o f tu m o r ce lls (T .ik e ic h i. 19U | j. O n e m a jo r class o l c a lc u m i-d c p e m le n i lio m o p liih c cell adhesion m olecules, ihe e.ulhcim s, have recently received e o iisid cia h le al le n t ton legal d ing th e ir role in tu m o r piogiession. l ;.-e a d h e rin ( u v o m o iu lin ) . a tra iis n ie m b ra n e 120-kDa p ro te in expressed p rim a rily in e p ith e lia l tissues, m ediates early events in c e llu la r adhesion necessary fo r the fo rm a tio n o f ju n c tio n a l com plexes, in c lu d in g gap ju n c tio n s (T a k e ic h i. 1991; Jongcn <7 a l. 1991). R e du ce d li- c a d h c rm expression may be associated w ith enhanced ttim o iig e n ie iiy ( N avarro cl a t. 1991) am i invasiveness( Itc h rc n s e /at.. 9X9). T h e d ete ction o l m u ltip le genetic d elects w ith in Ihe cells o f ovarian tum ors supports the hypothesis d ia l ovarian cancer may develop in a m u ltis tc p m anner (l.e e c l al.. 1990). The developm ent o l in ta n / m u ltis tc p m odels has been a recent a pp io .ich used by investigators to study the biology o f hum an e p ith e lia l c e ll ca rcin og e ne sis ( K h u n i t al.. I*i9 f)j. Such m o d els a ie genet ally based upo n d ie in iit m m u n o iia liz a lio n and ii.iiis lo i in.it inn o l n o im a l e p ilh e li.il cell c u ltu ie s by co m b ina tions o l oncogenic Ir.m s lc c iio ii ju o c e ilu lc s ailvl cvposm c Io la u iiio g e n s (K h im . 1**92) In th is stud y, a se iie s o l p ro g ic s sivelv ira iis lo im e d i.tl veil lines was developed to si in ly ovarian '( in re'll! .itkliess' I )epai im eiil o l Itmlugv. A im siio n g Si.iie ( 'allege. S .i\.iin ia li.<A J U l'i. lSA ' I n whom coiies|>iiidciice and ic p in il ici|iiests should he ad di essed K eieivcd: Jaiiuary I.V 199.1 and in icviscd lo im March 22. 199.1. ( i \ A i d VN I I ' l l I II I I M < MU I M K >1 M M s M l H >1 I C.llcillOgCIICsisf/l l /l/i* III lll/ing O IIC O g C IlC -IU ed i.Ilcil tia n s lo rilia turn. SV4U r . ih t ig c n ( I - A g ) a m i c - I I - m/ v1 1 ineocenes were e m p lo ye d , based o n the lin d m g s d ia l T - A g is a c o m p le te tra nsfo rm ing p io le in mi lo d c itl fe lls , that ll-m v lias been im p lic a te d m th e path og en e sis o f o v a ria n ca nce l tl.e e c l ul.. I 11)), and llia i b o th have ic c c tid y been e in p lu y e d in liu n ia n e p ith e lia l ce ll systems to m o d e l c a ic m o g e iic s is ( K liim . I 'W2). C e ll lin e s, w h ich range Iro m n o m ia l to m a lig n a n t. vveie all derived lio m the n a in i.il t.u g c l tissue lo i most ovanan cancels, the O S F . T o address die to le o l ce ll-ce ll m tc ra c lio iis m n e o p la stic p ro g re ssio n o l d ie O S I;. lin te l io n aI (.11C. ce ll adhesion and P.-cadherm cvp icssio n w ere analy/.cd. T he m odel may piovc' useful in studies focusing on the mechanisms ol m etastatic progression d u rin g ovarian caicinogcnesis. M A I I K IA I A N D M l. I l i t IDS ( e ll lines iiiu l culture c o riililio il'i T is s u e -c u ltu re lla s k s w e ie purcha sed lio m C o rn in g ((>110011 ta. N Y ): c u llu re dishes w c ie o bta ine d lio m U ccton D ickinso n (H o u s to n . T X ) a n d l.al> I'ek slides w e re o b ta in e d Iro n i N u n c (N a p e rv ille . II ). C u ltu r e m e d ia , s e n n it, g e n e ra l ch e m ica l reagents and o ilie r m edia additives were purchased front Sigma (St. l.m iis . M O ), u n le s s o tlic iw is e sp ecifie d . Prim ary rat ovarian suilaee e pith eliu m (KO SP.) cultures were isolated fro m n orm al Fisher 344 rats ( 2s to 42 d ) u tiliz in g the e u /y m a lie p ro c e d u re d e scrib e d by I la m illo n c l ul. ( l l>Nd). A spontaneously im in o rta li/e d ce ll lin e d erived lio m the c o n tin u o u s passage o l a p iim a ry K O S P c u ltu r e . K O S P l `>`>. has been d escrib e d (A d a m s a n d A u e is p e rg . I ` x5). ( 'O S -1, an SV,4 (t-lra n s lo n n e d sim ia n kid n e y ce ll lin e . llA l. H / . V n . a c o il- lin u o u s mouse lib io b la s i ce ll hue. a m i 124. .1 hum an ii.u iM iio n a l-e e ll-e a re iu o iu a lin e c o n ta in in g d ie a c tiv a te d c-11 o n 11( T a h in c l ul.. IW 2 ) o ncogene vveie o b ta in e d Iro m d ie A T O C ( R o c k v ille . M O ). COS-1 and T24 w ere utih/.cd as positive e o n lro l ce ll lines lo r T -A g and ras11 p 2 l expression ics|>cclivcly. H A I.H /3 T 3 was used as a negative c o n tro l cell lin e lo i expression ol both oncogenes. Piunary KOSP c tiliiu e s w c ie grow n m D u llv c c o 's m o d ilicd M agic's m e d iu m n u ll le n t n m iu ic I 12 ( D M F M T I 2 ) c o n ta in in g I fe ta l b ovin e s c itim (P U S ) a n d s iip p lc in e n tc d w ith 100 ID m l 1 p e n icillin ( i. K ill pg m l 1 s lic p lo m v c iii am i 0.2^ pg m l 1 a m p lio le n c m 11 ( I ' * P S A ) . ( '< >S I. U A l .U /3 F*. and 124 w e re a ll m a in ta in e d in D M F .M I 12 w ith ICJ'V I IIS and IS PSA. A ll o th e r c e ll m im ic s w e ie g ro w n in D M P M T ' 12 w ith S 'i new born c a lf scrum (N C S ) and IS l*SA. C u ltu re s were in c u b a te d at 37" C in 5r i C ( )2 :l >51 a ir at 100' Im m id ity and lo iiiin e ly passageil 1:1(1 p iio r to re a ch in g c o n llu c n e y . I ivin g c u ltu re s w e ie e xa m in e d by phase e o n iia s i m icroscopy o n an in v e rte il m icroscope (C a ll Zeiss. New Y ork. N Y ) .m il p h o to graphed on technical pan lilm (P astm an k o d a k . Rochester. N Y ). (e n c n ilio n o f tu im fc c iC il ce ll lines The P S V .W .* p la sm id ( A P( '( ) was g c iic io u s ly p ro v id e d by l) i. I). Husbec (Texas A .V M l Inivcrstly. C ollege S tation. T X ). Phis plasnm l is a d c itv .iiiv v o l d ie l*SV2/co plasm id, w hich c o n ta in s the iic o n iv c iu a n tib io tic resistance ( / ir o ) gene as w il l as scvcia l ic s iiie lio ii e ii/v m e sites, in to w h ich d ie SV4II c.uly-icg io n genes have been im o rp o ia ie d (S oulhem ami U cig. I `W2). The pSV h im plasm id was tra uslecle d (via electrop o ia iio n ) in to K O S P I'm and early passage R O S P cu ltuie s. B ric lly . 1O'* c e lls w e ie ro u tin e ly c o lle c te d Iro n ) m o n o la y e r c u ltu re s by iry p x iiii/a iio n , tia iis le rre d to a s te rile c u v c lle . and c h ille d o n ice hr Id n un in I m l ice -c o ld C a - '/ M g ' lic e D u lb e c c o 's p ho sph a te b u lte r c il salin e (D P B S ). T w e n ty m ie ro grants o l plasm id were added to the cells and transfected by e le c tro p o ra tio n p e rfo rm e d w ith a Ira n s le e to r 3(H) system (B io tech no lo g ies & P x p e rim c n ia l Research. San D iego. C A ) at 300V. 2dd pF . Cells expressing the nco and SV40 genes were selected m m edia siip p K nu n ii il vv iili 4(KI (ig m l ` I lln am inoglycoside .m lib io lu ( ic iie lu in ((> 4 IS ) (C IU C l) . ( ii.u u l Isla nd . N Y ). P o lliiw m g 21 days o l s e le e tio u . c u ltu re s w e re m a in ta in e d in Idd p g m l 1( D IN a nd 1 lo n e il w ith e lo u u ig 1mgs. T h e a ctiva te d m \ megene. i - H / i n 11. lia s been ile n v i- il lio m the l l( 124) hum an b la d d e i-ia ii u ioni.t cell Ime ( I a bili cl u l . I*>K2) p S V 2/nvi.` e I I /./ ' a pSV 2/ir* p la s n m l in to w h ic h ib is ,ie liv .ilc il e l V / i o r i o ncogene lias h c c if" in s c ilcd7~\v",is p ro v id e d by |) r . (. N ie o lson ( M .D A m ic i sun C a m e l C e n ic i H o usto n, T X ). T h e p lasm ili was ii.n is lc ite d in to R O S I l*m c e lls is d e scrib e d above A s a c o iilio l. K O S P l ` >`> t e lls w e ie ilso tra u s le c le d w ith d ie pSV2iv> p la s m ili t A 1( '( d o n a te d by D r. I). Bushce) w inch contains no oncogenes. ( nncih fitininiclcrs Po assess d ie ta lc o l c e llu la i p io lilc ia lio n . p o p u la tio n ilo u h lin g tim es were d c le im in c d lo i a ll ce ll lines. C u ltures w e re s e e ile d at lo w d e n s ity ( I d 1c e lls /w e ll) in each w e ll ( Ih m m ilia ) o l a 24-well plate. T iip lic .itc cell counts w eie then d ele im in e d ila ilv lo r S days w ith a C o u lie i co un ici a p p a iaius (( m ille r . I h a lc a li. I I ). .Iv.vuv oru n i h o n ij'c -itu lc /fi'n iic n i ^ lim ili T he p ote ntia l o l cells to lo rm colonics in sem i-solid m edium was m ea sure d by s ta n d a rd p r o c e d ilics 111 ic ily , e s p o n e n ti.dly g ro w in g ce lls w e ie lo u lin c ly h arveste d w ith tix p s iu a m t mis p e m le d ( I d 1 c e lls m l ') in a lo p agar co n sistin g o l 11..(.s'. Haclo-agar ( I.akew ood B iochem ical*. Irvin g . T X ) in D M i M P I2 w ith I d '. 1`liS . T o p agar (2 m l) was p la te d o n to 3 m l o l s o lid ified b ottom agar, consisting u l 0.75' lia e tu -a g iir m D M P .M -F I2 . T rip lie ;iie dishes w ere sei u p lu i each ce ll line exam ined. A lle i III days, die dishes weie cva im iied am i c o lo n ie s - S(t fu n in d ia in e le i w c ie v o m ite d < to n in g e lh iic u eies w e ie d e lc iu iu ie d by d ie ra tio : (c o lo n ie s >0 |iin dia n u iiih e i o l cells seeded) x Id ll' ; iu n n in i;c n n ilv < m im u n ii ilc c c lo /m ic n i **/ Iio m u - J c in c il ce ll Hues C ongenitally adiyuiie in n h m ) mice (H a i Ian. Indianapolis. I N ) w e re in ic c le il s.e. w ith Id " c e lls h arveste d lio m sul c o n tin e n t m o n o la ye rs su spended in D M I M - P 12 w ill lo u t p lu m il red. T h e mice w e ie m oni lo u d daily lo i c v u lc m c ol p alp a ble tu m o rs, ( `e ll lin e s w e re c o its id c ic d n on iim io n g c im ii they la ilc d to lo rm p alp a ble Illin o is vv11Inn 4 m o n th s a lle i injection. T m iio is w hich lo u n c il w e ie allow ed lo g io w in | m i d ia m c ic i, .11 w h ic h tim e th e a n im a ls w e re h um an ely k ille d : d ie tu m o rs w e re excised, riu s c ii s c v c ia l lim e s in I >PUS a nd cu t in to sm all pieces Several pieces lio m each tu m o r w eie live d in .V, g lu tu ia ld e h y d e . 2'< p .11atm m a h lc h v d c in Id d 111M so d iu m caeoilyl.ilv' (F le c tio n M iero sio pv S uciiccs. f o i l W ashington. I'A ). sectioned at I urn and stam eli w ith lo liin lin e blue loi evaluation o l m m oi histology. T u m o i-d e n ie d ce ll lines w eie e sta b lis lie d ly e v p l.m i c u lt m e o l th e re m a in in g iiu n o i pieces, fo llo w e d by s e le ctio n w ith ( i4 lN (4(HI u g m l ') . A ll c e ll lines d ia l form ed Illin o is in adiym ic n in e , and the re su lt.m i ttn n o i- d c liv e d 1e ll lines, w e ie s iim i.u lv iii|e e te tl in to P is lic i *44 ia ts to assess n e o p la stic p o te n tia l m u itiiu u ii c o iit|> c ic n i hosts /tn in u iu 1 \ u u h c m i\li\ P un.us a iitilu tily used lo i the in d iie e l lu u n iin o llu o ie s ic m d ete ction o l k e ia lin im e im e d i.iie lila n ie iiis was |>uieli.ised lio m S ig im i (c lo n e KN. I.*|. P itm a n a n tib o d y to r the ile i c it im i o f f Ag was ile i n v d Iro m a m o noclonal ce ll Im e. Pah Ids. a giti fro m D i. V. W ilson (Texas AiViM U n ive rsity. College S tation. I X ) C ell e u lliirc s were grown at low ile n sity on la b le k cham ber slides p rim to lix a iio u . C u liu rc s e v a iiu u c il Im T -A g were lixe il Im Id m in w ith 2' i paraN nnt.ildchydc ut 2d niM PUS, p H 7.3. washed in PUS a lo n e , a m i th e n p e iiu e a b ili/e d w ith P i T rito n X -100 in PUS. C u ltu re s exam ined lo r ke ra tin in te rm e d ia te fila m e n ts vveie s im ila rly tre a te d , c v c c p l d ia l they w e re fixe d in I d d '* m e th a n o l ( - 2 d 'C ) a n d d ia l th e p e rm e a b ili S i l l H u l l MAN / / 1/ /.11!mi step .!' mined hashing and iihlibation steps weie p c i I>i n il'll .is destiilvd In Su n) t i <tl ( |W |). Slides wi'ic mounted in ghieiol vsiili p phemlenedi.immc (`*.25 niM pplu'iivleiieJiaiimu. 'Mi's glveetol id rilS. pi I X.ll). and lho s.lllipU's M ill V|s|l.||l/0il Will) .1 /oiss I'ilOlOiniciOsCOpC III equipped loi opiiUimoNicute iliimiiii.iliit. S/)v / ' U il. i/iii/im rn n n iib lo llm ii ( o i l lines ev iiiu u e d lm ill, pi esc lice c l i .is p 2 l s|)ocicn and I i. iil lio m i w iie s m i .i t le d .is described by ( iu iiie i lo ck c l n l. ( I 'ls n j .m il N .iv .iim / ill ( l ' >` M j rcs|V C iivclv A lt e r w ashing iho cultures u n it D I'IIS . ihe cells w eie seiapeil Iron) ihe dish s u ih u e w ill) .i ru l'b e i p olicem an a n d c e n in lu g c d ( I.SIMli). ( lie cell p elleis w e ie Ivsed in Ivsis b u tle r lo r 15 m in on ice w ith oiC .ision.il vo ricM iic. F xlt.iC ls were then c c n irilu g c d (ii.lHMl g ) lo i 15 nun at 4 ( to p e lle t in so lu b le m a te ria l The .soluble p io lc in evtr.ict ( s u p e in a ta iit) was rem oved and ip ia n iila te d w ith a in o d ilic d I h a d lo r d anal>sis (P ie rce . K o c k lo rd . II.) . I.viracis w eie unveil w ith an ei(ual volum e o l sample bn Her. b o iled lo i 2 nun a nd lio / c n at -SO ( ' u n til a n a ly /c d . Sih Iu iiii dod ccvl s u li.n o |n>lvaer>l.nmde gel e le c tro p h o re s is (S i > V |' .-\( .1 ) .h i.i l i ms o l p io lc in samples was co n d u c te d w ith a inuiigel app.ti.nils I llo c lic i. San It .incisco. ( A ). F -ca d hcrin and las p 2 l p io te n ts were sc p a ia tc d on N' / a m i I5 r i gels U' s p cciivcly They w e ie the n elect to b lo i te d w ith a TF.42 T i.m sphor e le e tio p h o ie si' unit (llo e ite r) o nto niliooclhilo.se niem biaiies ( llo e lle i) at a constant dll V overnight Memhi aiios were b locke d o ve rn ig h t at lo o m ie m |v r , iiu r e in " b lo t t o " (5(1 in Nf l'ris-11( 1. 2 n iM ( 'a( l>. X m M N a ( I. p i I H.(l w ith 5r < ( a rn a tio n n o n -i.it dry m ilk . (1.2*/ N I M t t a nd HU5'` N a a / id c ) . The n iem biaiies u h ii li were probed bn ras p 2 l species weie ineiibated w ith a bio.nllv icaetive a iiti rasaiuilnK ly (pai) ras(Ab2). O ncogene Science. I Im u n d .ile . NN ). M e m ln a n e s w h ich weie probed lo i I cadhcim were liis l cut h o ii/o n l.illy approx im .ilelv at the l o c i o l 7> k l >a. y ie ld in g high a m i lo w -m o lc c u la rvieight sections o l the blots. The h ig h -m o le c u la r-w e ig h t sec tions weie incubated w ith a m onoclonal antib od y ( | ) I : ( 'M A - I , Sigma) w hich selectively hinds F -cadhcrm (u v o n io ru lin ). The low -uioleeui.ii-iveight sections were p tohed w ith an a nti-u(iilu iln i am iboilv (A h I. O ik o g c n e S eieiue) to ensure equal p ro te in lo a d in g A lt e r in c u b a tio n w ith a lk a lin e p lio s p h .it use eoii|iigated secondary antibody (H io -K .id . K ie hm o n d. ( `A ) the p ro te in s w ere v is u a li/e d by th e a d d itio n o l the c liio n io g c n ic substrate S-bioino-4 -chloio-5-indoly I phosphate/ ilitr o blue le lra /o ln iiu R e sulting bands were ip ia n tilie d by re lle e tive seanm iig detiM tonieiiy . l o i 'iiu ri'/i/ nf Inn. im riiil fiu /i tint* lu n n il in u rt c llu la i ( uininunu'iiiitm i ( J H ) CiJIC' was m easured by lluoreseence re co ve ry liter p h o io b le a ih in g (gap I R A F ) te ch n iq u e (W ad e cl i l l . l*)Kb) w ith a M e ridian A( AS >70 w ork station (O kcntos. M l) A p p ro x i mately I x |n .ells were p la te d in to 55-m ni dishes and ineu h aied lo i 4N h r p r io r to ana lysis T h e c e lls w e re washed several tim es w ith DIM1S p rio r to la b e lin g w ith It) p g m l 1 >(b)-eaibovylluoresceiu d i.ic c l.itc (C 'I'D A ) (M o le c u la r Probes, f.ugeiie. O H ) in seium lree. phenol red -free D M F M -F I2 lo r 10 nun at 17 ( I lie cells w ere th e n washed 5 tim e s w ith the same media w ith o u t ( 'I I ) A . p rio r to gap I K A P analysis. M ie io s io p ii liclds lo iil.iim n g several a dg n e iii cells were ch o sen lo r analysts o l (.IK An argon lasci beam (4NS m il) was focused through the objective lens o l ilie m v e ile d e p illu o rcscent iiiic ro s io p c <>n to a I -p m * area Scanning p a ttim e le rs were o p n iiii/e d lo r ni.ivim .il lluoieseence detection and m iuiu ia l ce lliil.n photoble.il llin g Selected cells were m arked and pholo h lc a c h c d w ith a series o f laser |x tim b leaches ai s u llie ie n t in ie iis iiy to re tim e c c llu l.u tUioresccncc to 20 to 5IP7 o f pre-hieach. but vnthout causing visible c e ll ttaniagc at the m icrosco p ic level Isolated ce lls were bleached as negative co n tro ls; isolated cells and cells c o nta cting o th e rs w ere m a rke d a m i le ll u n b le ach e d as p ositive c o n lio ls . A lt c i h le.ie h in g, Ilu o ie s ie iie e recovery in the pltoiobleaehed cells was nioni- lo ie d lo r 5 secpiential scans hegm nm g im m ed ia tely postbleaeh. am i then at I m in intervals K esulis o l gap I-H A P experim ents w eie expiesseil as the avciage (pet c e n t) ice ove ry o l llu o rcs- eence s la iu la iil e rio r o l the m ean (S I' M ). S tatistical sig nifican ce -. i)J t5 ) h c .lM x x u J h e jiie .jiu c c o v c ry ta le s o l the K O S I: ce ll lin e s was assessed w ith D u n e .id's m u ltip le i.u ig c lis t. In vin n t:\/>liiiil a nti nuam iion win The h i n ir o m igration am i ce llu lar detachm ent patterns ol the K O S I: cell lines was c h a ia c ic rt/c d w ith a head cxp h iiit assay (S te in <7 n l., l*J`>2) a nd a w o u n d in g assay ( Boyer i7 til., I W ` >). a c co rd in g to p re vio u s d e s c rip tio n s . In the bead e x p la in assay. ( y to d e x 5 m ic ro c a ir ie r heads (P lia rm a e ia . PiscMtaway. NJ). covered w ith cells, were suspended m c u ltu re m edium h c lo rc a d d itio n to 55 m m c u ltu re dishes at low d e n sity (5(1 to HKl be tils to each o f 10 dishes p e r c e ll lin e ) D ish e s were re tu rn e d to the in e u h ;ilo r a nil m ig n ih o n o l cells o il the he.ids o n to the snrlaee o l ih e c u ltu re dish was m o n ito re d it (. 12. 24 a m i 4H h r by p ha se-eo n lH ist m icroscopy- In the w o u n d in g iss;iy. ce lls w e re p la te d on 55 nun e u liu ie dishes uul a llo w ed to ieach confluence p rio r to "w o u n d in g " w ith a sterile 100-pl rp p c n d o rl pipette u p to create a u iiilo rn i lin e a r ce ll-lree area. M ig ra tio n o f the cells o n to the aeellulai siib stra lc Iron) the w o u n d e d edge was th e n m o n ito re d at 12. 24 and 4H h r. K IM I I IS ( in m ill n i/H iin s n fS l '40 a n il t - l l - r;is/ 'tn itis lim u c il R ( )S K f i l l Illti'S P rim ary K ( >Si: cu ltures w hich were e n /y m tilie iilly generated I m m ju ven ile Fisher 544 n il ovaries lo n tic d sm all colonies o f relatively Hal |xilyg o na l cells w ith debited borders a fte r 2 o r 5 days incubation, typical o f p im ia iy K O S I: c iillm c s (Adam s and A uetspcrg. I `W I). ( o n tm iie d m eub.ilion resulted in "c o b b lesto ne -like" m onolayers w ith few co nta m in atin g strom al cells, w h ich w e re e lim in a te d Iro m K O S I: ce lls it lirs i passage is d e scrib e d (A d a m s md A u e is p e ig . I ` W |). I:arly-passage K O S I: cu ltures grew relatively slowly, w ith an a pp m xiina le p o p u la tio n d o u b lin g tim e o l 2`t h r d u r in g lo g-phase g ro w th (Table I). reaching a saturation density o f approxim ately 5 x 10* c c lls /c n iv In d ir e c t im m u n o flu o re s c e n c e w ith a b ro a d ly reactive a n ti k c n iim a ntib od y revealed strong perinuclear and cytoplasm ic lil.m ie n t netw orks m the m a jo rity ( > W i ) o f p iim a ry K O S I: cells (d iita not show n), su bsta ntia ting the s u rla c e -e p ith e li.il o rig in o f the p rim a ry c u ltu re s ( I lo rn h y <t til., I*w 2 ). ( e lls in p rim a ry c u ltu re are co n sid e re d n o rm a l: how e ve r, as they develop in to n eo plastic;illy lia n s lo rm e d cells they acquire several a d d itio n a l p ro p e l lie s o l w h ic h iin m o r ta li/a tio n is co n sid e re d m ost im p o rta n t. K O S I: I'W is a spon tan eo u sly iin m o rl.th ic d c e ll lin e developed fro m (he co ntin uo u s passage o l K O S I: c u ltu re s w h ich e xh ib its e p iib e lia l-lik c m orphology in culture (A dam s and A uerspcrg. |W 5 ). In developing a m illlis te p ira n s fo rm a iio n m odel o f the O S f:. tra n s fo m ie il cell lines were generated hv in eo rp o riitiiig viral (pS V5m v/) and activated c e llu la r (pSV2nr> e -l l- r < / \ '') oneogenes in to the im m o r la h /c d K O S I: I `W cell line. A d d itio m tlly . an e a rly passtige ( p 2 ) K O S I: c u ltu re was ira n slce icd w ith p S V .W n . H ie ce ll fines resulting from pSX'.V/co (ra n s ie e iio n o l K O S I: PW and K O S f: w ere d esig nated s v-K O S I: IW and sv-l< ()S I: respectively. A c o n tio l cell lin e . n c o - K O S f: l 4W. vvas d e v e lo p e d fro m p S V iic 'o tra n s fe c tio n if K O S F. IW . S v -K O S F )` W. sv K O S I: a nd n eo K O S F 1` PI a ll e x h ib ite d e p ith e lia l m o rp h o lo g y in c u ltu re (F ig . I). closely resem bling the p arental ce ll lines Iron) w hich they were d e riv e d . In c o rp o ra tio n o f SV'41) e a rly -re g io n genes md subse quent expression o l T -A g in the p S V 5 nco cell lines was i 4 i 4 I I >\ , \ K I \ N I n I III I I \ l I \ | t < I N i K . I M M s M i l l KOSI KO S I: I'M m'n K O S I X. K O S I T o KOSI o K O S I I'M 1 xi-K O S I: I'M n o K O S I MM T n o KOSI I'M 2' it T J X 17.1 . 114"" 1 7 'I ! 1 7" |S u 2 V12 4 (l 4 ' 1 s l 12" lit (1 i 1.2"* Hi o . l ' 'H I T O K*' I f O' N D r 11.0' 0 4 IM1* 0 /4 ll 7 0 1' 4/4 ( I I I ) ' llx It v N A " 4.M 0.2" 4/4 (() M il) t 0.2- N A 47 V 11.2* T 4 (1) 4K 5 (I..V N A N il N l) N l) 04 0.4 tl.-l 0/4 /!( 1) 4.4 (1) 'A ll ce ll lines v ie analyzed it com parable passages except K O S I: (lirs i passage). T o p u lu lu m d ou blin g tunes w e ic d e te r m ined lim in g log-phase grow th am i arc expressed as mean h r SI M lo r trip lic a te cultures, f e l l lines xxith the same le tie t w e ic not significantly d ille ie itl | /< "* (MIS. D uncan's m u ltip le lunge tesi.)- t lonin^ eilicicucv m soli agar - (num ber o i clones >50 tu n /n iu n h c i o l cells seeded) x HIM'. expressed as mean p erce n t age - SI M lo r tiin ln .a te assays C e ll lines w ith the same le tte r were not significan tly d ille ie n l (/> > H.llS. I h m ca ii's m u ltip le range lest | - 4N I). not done. ( e ll lines w hich la ded to lo rm tum ors w ith in `JO days in aihym ic mice were not tested lo r liim o iig e n ic ily in lis h e r .144 rats. - 'N u m h e is in parenihcscs = la ten t pcriiM l (weeks) until tum ors were palpable.-"N A . not applicable. conlirm e d by iiiiim in o llu o ie s c c iii detection o f the viral p rotein in lOO'-r o l lh e ce lls (d a ta n ot sh o w n ). T h is expression appeared stable, since there were no changes in la be lin g u n ifo r m ity o r in ie u M ly in In te r passages (15 10 2d passages). T -A g was not d ele cted in n e o -K O S I: IW o r K O S I: l `M. o r in K O S li lines M ihiccicd to sim ila r in m m n o cyio ch cn iisiry (data n ot sh o w n ). I be v e il lin e d e ve lo p e d f it m p S V iie o / c - l l-/ o l 1 ira n s lc e lio n o l K O S I: I'M was designated ra.s-KOSf- l `M. U n lik e K O S li I'M , th is p o p u la tio n gre w in c u ltu re is ra n d o m ly o rie n te d lib ro h last lik c cells lacking extensive ce ll c e ll associa tio n s ( l ig I ) a n d e x h ib ite d loss o l c o n ta c t in h ib itio n ( i . r , ce ll p ilin g ) p rio r to i caching co niine lice. In u m m o h lo i analysis o l c -ll-m x * 1e x p lo s io n in ra s -K O S li I'W d e in o tis ira ie d an ahnnd an i level o l the slow er m ig ra tin g ras11 p 2 l species ((iu m c rloek ct u i. IMH`>) in ra s -K O S li I'M to ta l c e llu la r e xtra cts (b ig . 2). P opulation d o u b lin g tim es d u rin g log-phase gro w th o f the tra n s fe c te d c e ll lin e s (s v - K O S li I *>*>. ra s -K O S li MM. neoK O S I7. MW. a nd s \ - K O S li) . as w e ll as th e n o rm a l ( K O S I . ) and im m o rla li/c d (K O S li l `M j lines, were dete rm in e d (T able I). N o sig n ilie a n t cl d lo r once in g ro w th ra le s was d e le c te d b etw een K O S n MM and s \ - K O S li MM. im p ly in g n o e ffe c t o n in M in i grow th rale due to T -A g expression in im n io ria ii/e d K O S li cells. O n the o lh c i hand, s liib le expression o f c -ll-n ix * 1 in K O S li I'M resulted in a sigm ltcant increase in g ro w th tale. The d o u b lin g lim e o l sx K O S li was s ig n ilic a n tly s h o tlc n e d as com pared w ith K O S I: cultures. T o dale, s v -K O S li has u n d e r gone o ver Mil p o p u la tio n d o u b lin g s w ith o u t e x h ib itin g crisis o r a decrease in grow th tale, and retpm es much low ci scrum c o n c e n tra tio n s th a n KO.SI- c u ltu re s ( 5 ' i N C S r%. 15'< l ; HS respectively), in d ica tin g that p S V .W / is capable o l u im iu ilu li/mg early-passage K O S I- cultures. The K O S li c e ll lines were evaluated lo r th e ir a b ility to form c o lo n ie s in so ft-a g a r suspension c u ltu re s , in in Mtn> iiu lie a lo r o f transform ation (Table ll. K O S I: and K O S li I'M failed Iti fo rm co lon ie s a lte r Ml davs. re m a in in g as single-cell suspen sions. S table e xpression o l T - A g m K O S li c o n f in e d an a n ch o ra g e -in d e p e n d e n t p h e n o typ e as s v -K O S li lo r m e il c o lo nics in soft agar; however the clo ning ellicieney was not signilica ntly d ille re n l fro m the iio n -tra n s fc c tc d K O S I: lines. Sv-K C )S li IW e xhibited a significantly higher cloning ellicieney than ils parental line, thus tie in o iis tra tin g the a b ility o f T -A g expiessiim to co u le i anchorage`iiitlc p c iiilc iic c on iiiiin *ila l i/c d K O S I: cells. A lic i III tl.ixs. ra s -K O S I: I'M e x h ib ite d a .significantly g ie a ltT e liu iu ig c llim - iic y ili. u i a ll i thet lin e s N e o -K O S I- MM l.u le d i o b u m c o lo n ie s ui s o li .ig .u . it ilin g o u i the possibility that the n n ii u iu o g c n n e om pim eiits o l the p S V .W o and p S V iim 'c l l i* id 1 p lasm ids o r . . p.NV2/ii-o) were icspoiisihlc fn rrh c acquirer! ch r.na e tciisikvn i-xx K O S I . sv K O S I MM and i as K O S I MM. The lia llm a ik o f in M ini neoplastic li.m s lo im .ilio n is die a b ilily o l cells to lo t UI Illin o is w h en i l l | t t ie d in to a m n ia ls I lie capacity ol the continuous K O S I cell lines (K O S I. I'M . n e o -K O S I MM. sv K O S I: MM. ras K O S I I'M a n d sx K O S I t to lo rm tum ors in aihym ic mice was investigated (T a b le I) W h ile K O S I MM ain I n e o -K O S I: MM fa ile d to h u m tu m o rs at any o l the in je ction sites. s v -K O S li am i sx-K O M - I'M both fo rm e d p a lp a b le H u m u s it a ll sites a lic i la te n t p c i im ls o l Ml weeks a n d 1 weeks respectixelv. lir o s s ly . the m m o is w e nsolid, rounded and sm ooth on the surbice. w ith no ohxio iis a tta c h m e n ts to the skin o r m n le rly m g body xxall. K .is -K O M MM lo rm e il ra p id ly g ro w in g I llin o is w in c h w c ie in e g iila i. h ig h ly v a s c u la n /e d a nd lin u ly a tf lie ie iil to the ox e i lx lu g -b in a nd u n d e rly in g th o ia e ic m u s c u la tu re . Most m o iie m e x .u iu ii.ilio n revealed no visible niciastases to d u n n in g ly in p h o c c n tc is o r in ie rn a l organs second.uy to any tun io is. Ilisto lo g ie a llv. the neoplasm s fo rm e d by s v -K O S li MM xxcrc coni|>oscd o l c o o ls and large nests o f lig h tly packed e u lx m l.il to ox aI cells, xxnli scant am ounts o f cytoplasm interspersed w ith num erous c a p il la rie s ( Tig. 5)- Th ese p oo rly d illc ic o lla t e d e a rcin o m a s h is to lo g ic a lly re se m bled o v a ria n serous p u p illa ry a d c n o e a rc im u n .i. as w e ll as the m id illu e u M a te d o v a ria n e a rcin o m a s (S c u lly . M*7` >) R a s -K O S li MM fo rm e d u m lillc ic n iia ic d sarcom as com p ose d o f lig h tly packed lu s ilo rm cells xxith n io d e ta ie a im m n is o l cytop la sm a nd e x h ib itin g a b u n d a n t m ito tic lig in e s ( l ig 5). T h is n e o plasm xvas ca pa ble o f in v a d in g host tissue, as ex i deuced by num erous a icas o l ske leta l muscle xxiihin llu tiu iio i sections. S v-K O S I: also lo iiiie d p o o ilv d illc r c iiliu lc J s.ueo mas. m ade up o f xvhoils o l lu s ilo rm cells xxith s ta in to m oderate am ounts o f cytoplasm ( lig . T u m o i-d c iiv c d e xpla in c tilliu c s xxeic g i-in -ia lc d lio m c.n h o f th e I llin o is ile s c iilv d ilxoxe. lo llo xxm g s c lc th o u ill ( i4 | S (4IH) p g / m l) lo r 21 days, in in a tte m p t to d e v e lo p K ( )S 1-. ce ll lines w hich were fm tlie r ira n s lo rin e il. T u m o i -d e n ie d ce ll lines w e re su bse qu e ntly c lo n e d a n d xlcsig n uicd T -s v -K O S I MM. T -i;is -K O S I : I 'M a n d l-s v -R O S I ( I ', " m ir o r - d c n x e d " ) K e le n lio n o l T -A g e xpie ssio n in th e T -s x - K O S I. I 'M a n d I sx R O M lilie s xvas c o n lirm e d by in d ire c t im n iin io llu o ic s c c iic c ( ila t . i n ot slioxxn); ia s * 1 (21 e x p ie s s io n in I '-i .is K O S I I 'M xs.is u r n firm e d by im im iu o b lo llin g ( I 'ig . 2). U n lik e th e n p a ie n ta l ce ll lines. Im th T-s\-K (> S J M M .nul T -s v -K ( JSI: e xh ib ite d a d istin ct fibroblast like m o rpliology in c u lim e . grow ing in a xlisoiga n i/ c d fash io n xxith less e x le iis ix e c e ll-c e ll usm* iu lio n s ( I ig . I ) N c illic i cell line a c i|iiiic d a sig nilit u n til lo gin i g io w ili u ilc as a result ol tum origcm c selection; hoxxcvcr. lo iiiia tio n ol m ote fre x |u c iit HI4.I h u g e r c o lo n ie s in s o li a gai was o b se rve d (T a b le I ). T - ia s - K O S r I'M ixas p h e iio iv p ie a llx in d is tin g u is h a b le fro m the p a ie n ta l ras K O S I: I 'M p o p u la tio n , w ith v ir tu a lly th e same m orphology (ila i.i not shown), g io w ili ta lc am i clo ning c lli cicncy in s o li agai (T a b le 1). W h ile iiim o r b n m illio n in m iim in c d e lic ie n t lo d v n is is a va lid in d ic a io i o| m i nm iia n s jo rm .iu o n . the a b ility ol a cell lin e to lo rm Illin o is h i im im in o c o in p e ie n l hosts is u m o te s tiiiig c n l iiu lie a lo r o f true n ialigiia nt p o le n lia l A ll ol the cell lines w hich lo in ic d tum ors in a iliy in ic m ice, and the lines derived Ih e ie lio n i. were assayed lo r th e ir lin iio iig e n ic capitcity in syngeneic l isher 544 rats (T a b le I). K a s -K O S I: MM am i T -ras K O S I: I'M were Ih c o n lv c e ll lines w hich lo rm e il tu n io is . w iih a la ten t |X*riiHl s im ila r to th .it seen in a ih y m ic m ice. These Illin o is xxeic gm ssly and histologieally in disting u isha b le Iro in those fo rm e d in ilhym ie m ice (d a ta not show n). l i n i I MAN / / 1/ t ic.t it i - M*i pin ili gic ili;ui|!CN ;i\ m k u U'(I w ith progressive tra its lo rm a tio n oi K O S tl ce ll lines. Phase m icroscopic images o f ' V ROSE t i l t . sx-KO SI- lw ( h i nco K O S i: I W fJ. I s v R O S l u t l f sv-R O S h | W t e l and ras R O S I: I W ( f ) cultures, e-11-fin 1' -ir jiis iiH in .iiio n n l R( )SI-. I `W aiul tum orige n ic selection o f ihc S V4U-iranslorincxl lines resulted in a s h ill in m o rphology from c p n ltcli.il in lihroM .isi hive ( 'mcnmiiam w ith ih is s h ill in m orphology xx.is an apparent reduction in celi celi associations. Scale har, 100 tint. 12 34 5 Fit a nr 1 - liiiim in o H .ii analysis o i ras p 2l expression in c-H r io 1' li.uislccied ami co ntro l cell lines, liip ia l .m iounls o f c e llu la r extract Iront the I 24 Madder carcinoma (lane I ). B A I.H /J r id a n e ). ras ROSI l `*tlane )). R O S I' I'<| l.inc 4>. and I ras RO SI i I w lia n e 5) cell lines xxetc separateli ( l * ' i M)S-PA< il. ) . electroH o tte d and probed with a pan-re.u tixe .llti-r.is antibody. / I n o w / liv ii <>j JitiH In m o l im > ju m lu m a l tn lc rt'tilu lu r (o m n u o iu a tu m f(7/ ( ') (la p T R A P , the a b ility o f cells to recover carhoxy fluorescein ( ( T ) llu o rc s c e iu e a lte r phoio hlcnch in g via gap junctio n s (W a d e <7 a!.. I*>8h). was e m p lo ye d 1o M u d y ih c re la tio n s h ip between progressive tra nsfo rm a tio n o f R O SE cells and p e rtu r bations in fun ctio na l (JJIC . (iJ IO in R O S E c u ltures was first exam ined lo appraise recovery characteristics in the most norm al ce ll type o f those const it u tin g the g ro up o f p rogres s e d ) tra nsfo rm e d R O M : ce ll lines. A ll p h o ln h lc a c h c d R O S E ce lls n co m act w ith n e ig h b o rin g ce lls e x h ib ite d ( ' F llu o rc s cence recovery; th is re co very xvas ra p id , w ith m ost o c c u rrin g in the lirs t I to 2 m in a fte r b le a ch in g (F ig . -1). T h e m ean recovery ra te a l I m in was 21.72 r I .(WO (S E M ). T he co ntinuous R O S E ce ll lines were sim ila rly subjected to gap F R A P analysis (F ig. 5). P S V .W o tra n s fo rm a tio n o f R O S E d id not result in any appreciable reduction o f G JIO , as sx-R O S li recovered al rates indistinguishable Iro m norm al R O S E cultures. The in im o ita lt/c d R O SE IW line exhibited recovery rales s ig nifican tly loxxcr than R O S E cultures, axerag- l ) \ A K IA N I I ' l l Ml I IA l I A III IM H .I M M s M i HU h i.i'K i 4 - ( ir a p h ic display o f gap FR A P m p i unary R O M eells. P rim ary ROSLi c u llu rcs were llu o re secn ily labeled w u li ( 'F D A and subjected (o gup FR AP. The recovery characteristics displayed in I his figure arc typical lo r n orm al K O S F cells hi p rim a ry e u llu rc . T w o KO S I cells (d ia m o nd and c itc lc ) w e ic p lio ioN ea e he d lo a level o l 2(1 lo 5(1' o l p ic h lc .u h levels A control KOSF! cell (square) was also m .ukcd and m o m lo ie d m e i I lie *1-01111 scanning period. in g h.55 r O.M2'V ( S I-!M ) per m im ile . S im ila r i). p S V .W n ir a n s lo rm a iio n o li lie R O S I: I n l in e d id noi appreciably a ile d C iJIC . as sv-R O S F! | W and R O S I' IMM re c o vere d al m i i i i I.ii rales. H ow ever, (lie highly tra nsfo rm e d ras R O S I! I'M Ime exhibited a significantly low er ra le o f C I: dye iran sle r. T o de le n il ine die d ie c i o f I'urdier neoplasiic piogressum upon CiJIC'.g a p F IC A I' was p e r lo n iie d o n d ie m in o r d e riv e d po(>ula lio n s. Passage o l the p S V .W n -lia ils Io rm c d c e ll lines ih io n g h a lh y n iic m ice (s v -R O S f: -- T -s v -K O S f! a n d s v -R O S I! I MM I s v -K O S I! I MU) le d lo a s ig n ilie .in i re d u e iio n in lu n c iio n a l co m m u n ica tio n . S im ila i passage o l ia s -R O S I. I'M d iio n g lT a lh y n iic m ice ( i .in-ICC>SI: I 'M T -ia s -R O S f*. I'M ) i.-s id ic d in a p o p o la i io n w ith lo w e r gap F R A P a le s; h o w e v e r, i Indecrease was n o i s ta tis tic a lly s ig m lica n t \j> -- O.dfi). KiM 'K t: 3 - I liM olog ic appearance of- (he lu n iiirs form e d l>> sv R O S I. I'W , sv-R O S f. .uni ras-K O SF I'W in ;iih )m ie miee. Tu m o rs re s tiliiii|* fro m sv-R O S F I MM (a t were poor!) d iiic r c n ii ;dcd carcinomas. lin d i sv-KO SF (bt :md ras-KO SF IM M pj lorm cd p oorly d illc r c n iia ic il s;ireom;is. Note (lie illu sion o l skeletal muse'll' lissue e xh ib ite d hy ras-KOSF. I MM Se:de h:ir. 5i) fin i. I f f , o f p u tir e .w ill' i i t t i i \ f o n n t i l i t n i <m c llu lu r < n ilh \ in n a m i n in fra lii m ih a r a v W n \U i\ in v itio Necessary stc'ps in the m e ta sta tic cascade a ie d e ta c h m e n t and dissem ination ol in d iv iilu .il cells horn die p rim a iy lu n io i T w o m ig ra i ion assays were used lo evaluate th e a b ility o l progressively transform ed KOS!-, cells to dissem inate Iro m a p rim a ry p o p u la tio n . In the lirs t assay, ce lls w here gro w n on sp h e ric a l C'y to d c x heads a m i tra n s fe rre d lo 3 5 -m m c u llin e dishes w h e re th e y al la d le d lo I hi* surface a n d m ig io lc d away tro n i d ie p n m a iy im p u ta tio n (F ig . (>). T h is assay has b e e n used lo investigale ce ll d iia c h m c iii and m ig ra tio n p a iic in s in t ia iis fo r m e ii g ia m tlo s a eel! lin e s (S ie iu <7 a l.. I MM2). S \ -R ( )SI cullurcs grew Irom die Cyiodcx beads a s e o n iin iio n s e p iih e li.il sheets m a fashion iile n lic a i to d ia l e xhib ite d by m im .il R O SI e u lliire s. S im ilarly, both die R O S I! I'M and m - R O S I! I 'i't a ll lilie s m ig ra te li as com m um s sheets Iro m the c e ll-co a te d beads. I ll c o n tra s t. ras-R O S F ! I MM d eta che d and m ig ra te d as indivie! ual cells away Iro m the p rim a ry head cell p op ulatio n . As a re s u lt o l in r iv o passage. T -s v -R ( )S I: I MM a c i|iiir e il d ie a b ility lo dissociate and m igrate as in d iv id u a l cells. M u ch lik e is p a re n ta l sv-R O S f: line. T-sv-RC )SI ! co n tin u e d in m ig ra te as an e p ith e lia l sheet: however, glow ill was m ote disorganized and A 3A 003 iN 3 D 3 d A 3A 003d IN 3 0 iJ3 d A d3A 003 IN 33 3d M ill I MAN I I I I . r . ( \ * 1 11.in \io i iiu 'il R O S E ce ll lines fc p ic s c n l la ic si ages o l n e o plastic p iog icsM on o l I lie ( )SI The Illin o is (limed hs ns-R O S E I 1* ) * ere h is to lo g ic a lly eoiiM sfeni 'M ill seious .ulen oe .iie in on ia o l the os ary T u m o rs I M in il.ii liis|nlii!\ li.ive been re p o ile il lo d e ve lo p in a lh ym ie n in e iii ik ill.iii il " i i l i s|tn iii.n ie o iis l) li.n is lo rn ic d U( )SE eells ('fto il'i'iiiiV i* / f f i i r t i i m h i t i i y o H rn n stn rm e d R O M : eells to lo im lu m o is ic tc iiih liiig hum an osariati m alignancies o l clin ie.il significance sup|>orls (lie iiscluln css o l the in M in t K O S I: m odel system. The ras-R O S E IW p o p u la lio n in du e ed h ig h ly in .then.m i sarcomas m a th sm ic m ice a n il n o rm a l rais. k i- M S V ira n s l'o rn ie d R O S E eells have also hecn show n (o fo rm sareonias jn in in m no.suppressed ro d en ts (A d a m s and A u e rsperg. IW I) . The mesenchymal nature o f the tum ors form e d hs n /'- lra iis |o im e d R O S H cells is eonsistent w ith the E M T observed in c u ll it i e . A s ra s-R O S E I W e x h ib its a m e senchym al phenolspe in n u o . n is not su rp risin g that it fo rm s sarcom as in i m i ' I lie in m o i h is toliigs o l the s'ROSE, tu m o rs was not to u s is tc n i '-iili the assum ptio n that iia iis lo r m c d e p ith e lia l eells beget e p ith e lia l tum ors. S ' -R ( )SE. " I ii e h app ea rs e p ilh e lu l m culture, lailed to lo rn i e pithelial tum ors in alhym ie m ice. This ability ol iia iis lo im c d ROSE eells to lo rm tum ors o l ineseuehw u.il e h a ia e ie i mas he related to the e m b ryo n ic inesoeleim.il o iig m o l the O SE. A lih ou g h the OS I d illc r c u li. U s iiiio e p ith e lia liis s u e .il mas re ta in the a b ility t o re ve rt to a nioie p rn m ii'c mesenchymal hllerenitaiioii ptogram when p ro gicssi'cls ii.n is io im e ii The plastic nature exhibited by the <)S l mas account lo r the histogenesis o f the m esenchym al c o m p o n e n t' o tte ii observed in osarian Illin o is (S cu lly. I ll7 `l). I he le suh s ot the e h .ira e ie ii/.tiio n c x p c iim c n is o l the U( >SE m odel system suggest that neoplastic I r a tis lo iilla tio n o l the cell t>|K' responsible lo r the m a jo rity o l osarian m a lig n a n cies may deve lo p in a stepw ise m anner. Progressive in i n ro i rails Ior nwmon ol ihc O SE was accom plished through oneogcne-niediaied ti.u is lo rn ia iio n and in vivo selection proce dures W h e n the e n tire series o f R O S E c e ll lines is vie w e d as a progression Irm n norm alcy lo malignancy, some degree o f older becomes apparent. Rased upon their g row th charactcrisiics. K it h in iv/io and in m o . the ROSE, ce ll lin e s c a n be ordered js lollow s norm al (R O S E ) -- im m o rta liz a tio n (ROSE, p in ) -- e.uls t ia n s io in u tio n (s ' RO S E a m i s '-R O S E I > -- in ie im e diaie tr.m slo m iaiion (T-sx-RO SE anil T-sv-KO S E | `/>i) -- late lia n s lo r m a lin n (ra s -K O S E I W and T -ra s -R O S E r> ll|. This o id e n n g is based ii|>oii die p henols pc o f the ROSE, cell lines and is e m p loye d In illu s tia le the levels o l progressive ii. iiis h iiin a iio ii w h ich was ulserved in v in o T h e speeilie cell lines c ite d p a re n th e tica lly a ie exam ples o l d ie p a rtic u la r levels ol ii.iiislo rm a tio u A lthough studies eontparing CiJIC o l norm al and fully tra nsfo rm e d Cells suggest tha t a lte ra tio n s in CiJIC* arc associ ated w ill] neoplastic tra ris lo rn ia lio n (Y a m a saki. I WO; Laiewenstein. 1^7^). in fo rm a tio n available on ( iJ IC 'iliirm g the p rogres sive stages o f neoplastic d eve lo pm en t is lim ite d ( k la n n vl ill., IW M Studies fo cu sin g on (JIC d u rin g m ouse-skin c a rc in o g e n esis have d e m on strate d an association b etw een n e o p la stic progression m m o and the progressive loss o f h om olo g ou s GJIC I k la n n i-r u l . I'W M) R educed CiJIC- has also been reported d u n iig progiessive tra iis lo riiia lio n in i nro (S tein cl i l l . 1*1*111. 1`iim a iv R O M ce lls were ju n c iio iia ll) c o u p le d in c u ltu re , as dem on strate d by i. ip id dye tra n s fe r betw een c o n ta c tin g cells " i ih in the first lew m in u te s a lte r p ho tob le ae h ing . T h e ra te o f dye tia n s le r between coupled R O SE cells was com parable to rates re p o rte d lo r o th e r high ly co up le d cells (S tem ci ul.. | W | ) I he ra le o f dye co u p lin g in sv-R O S E was n ot s ig n ilic a n tly d ille ie n l from that observed in ROSE cultures, in dicating retention o l a high level o f CiJK . S im ila r!', dye c o u p lin g in ns-R O SE IW w.is not sig nilic a n tly d i lie rent fro m that meas u ic d in RO SE IW cells In a recent re p o rt. pSV .^ii-o ira u s lo rn ia tio n o f p rim a ry rat granulosa cell cu ltures was not associated w ith any significant re d u ctio n s in C iJIC in the tia nsfo rm e d populations (S tein ct u l . IW 2 ). consistent w ith the obse rva tion s n o te d w ith the p i H ilary U( ESI |x> p u lalinn s. In i ih - passage-*! b o th p S V .W -f^-tra n v io rm e d -R C JSE-Cell-Hues th ro u g h a lh ym ie m ice (s v -R O S E -- T -s v -R O S E and sv-R O S E |W -- T-ss-R O S E IW ) resulted in ce ll lines w hich were U irthcr tra nslorn ie d am i exhibited reduced C iJIC These re sults suggested that p S V .W u Irans fo rm a l ion o f R O S E cells had no im m ediate c llc c i on G JIC . but d ia l fu rth e r neoplastic progre ssio n was associated w ith a s ig n ific a n t re d u c tio n in CiJIC'. S im ila r o b se rva tion s o f a re d u c tio n in G J IC u p o n in vivo passage o f p S V .W o -tra n s fo rm e d cell lines have been re p orted (S tein f t at., I W I . IW 2 ). c - ll- r u s * 1 tra n s fo rm a tio n o l ICC)S E l ` W in d u ce d a sig n ifica n t decrease in the ra te o f dye tia n s le r b e tw e e n a d jo in in g ce lls in culture. The c o rre la tion between ll-o /s expression in tra n s form ed cells and decreased C iJIC has been reported by o th e r in vestig a to rs ( I ) e P c ijic r c r j/ . . (W O ). Passage o f ra s -R O S E IW in vivo (ras-RO SE. IW - * T -ras-R O S E IW ) a ppeared to som e w h at reduce CiJIC'. h ow e ve r, this d ro p was n ot s ta tis tic a lly sig n ifica n t. C o n s id e rin g th e losv ra te o f d ye tra n s fe r d isp la ye d hs d ie ras-RO SE. I W p o p u la tio n , any lu r t lie r re d u c tio n s may he ra th e r su btle . Passage o l ra s-R O S E l `W in vivo d id not appear to fu rth e r tra nsfo rm the resultant p o p ulatio n a pprecia bly. since ras-R O S E IW and T -ras-R O S E IW were p he no l yp- ieally indistinguishable. A s in iila i o bservation lias been rc- |wtrie d lo r c -II-/m -tra n s fo rm e d rat liv e r e p ith e lia l cells passaged in vivo; U iinoi -d ci tved pop ulatio n s displayed gro w th eharaeter- isties a m i C iJIC s im ila r lo th a t b e fo re in vivo passage (D e E c ijlc r c l ul.. I WO). Thus, the observation that in vivo passage o f ra s-R O S E I W d id s ta tis tic a lly decrease C iJIC is n ot in c o n lh c i w ith the o b se rva tion s n o te d w ith s v-R O S E a nd sv- ROSE tha t neoplastic progression in vivo was accom pa nied by a reduction in G JIC . Taken together, the results o f the gap T R A P studies perform ed on R O S E cell lines strongly suggest tha t progressive tra n s fo rm a tio n o f th e C)SP. is a ccom pa nied by a progressive decrease in C iJIC and d ia l this re d uctio n may occur very early, before the appearance o f anchoragc- im lepem lenec. lu in o rig c n ic ily and E M I . as evidenced by the d ille re nce in C iJIC between RO SE and R O S I: IW . Results o l in vitro ce ll-a illiesiveiiess assays dem onstrated d ia l, whereas sv-R O S E a m i sv-R O S E I W m ig ra te d as c o n tin u ous e p ith e lia l sheets, die in i'iVn-selected T -sv-R O S E and I sv-R O S E I W lines dissoe ia ted a m i m ig ia lc d m u ch n io ie as in d iv id u a l cells. A s im ila r tra n s itio n in g ro w th p a tte rn was o bserved w ith e -ll-n i.v 11 tra n s fo rm a tio n o l ROSE! IW . T h is tra nsition in ce ll adhesiveness may represent progression tow ards a m etastatic phenotype, as a te d u c tio n in lu n io r-c c ll adhesiveness in viv o is th o u g h t to c o n trib u te to c e llu la r d e ta ch m e n t Iro n ) the p rim a ry tu m o r, a necessary step in metastasis (T a ke ieh i. IW I) . S im ila r tra nsition s o f in vitro grow th patterns fro m adherent e pith elia l m orphology to disso ciated fibroblast-like m orphology hase been observed d uring tra nsfo rm a tio n and neoplastic p ro gression'ol o th e r e pith elia l c e ll types (fic iiie n s et u l. IWW; De E e ijte r c i u l. IWO; S tein cl u l . IW 2 ) a nd are th o u g h i to re p re se n t a m o d u l.itio n in d illc r c n t ia lio n ( I lo y e r /<//.. IW 9 ; F rix e n et ul.. I W l ; T a k e ie h i. IW I). Co n s id e ra b le evidence has a e e u im ila le d d ia l c e llu la r adh c sion o l e p ith e lia l ce lls, b o th in vitro a nd in vivo, is m e d ia te d by the E -cadherin cell-adhesion m olecule (T a ke ieh i. IW I; S liim oyauia cl ul., IW V ). M o d u la tio n o f e a d h e rin -n ie d ia le d in te r c e l lu la r a dh esion is h y p o th e size d to o c c u r d u r in g the n e o p la stic progression o f e p ith e lia l cells in u u /, re su lting in m alignant O VA K I A N I-1*1 IM I.I.IA I ( A lU IN tX il N i MS M i l l I ce lls c ;ip .ih lc o l d e ta ch m e n t .n u l m etastasis (Itc h r c n s et a t, IW 9 ; T a k c ic h i. I W |) . Several studies have indeed dem onslra ie lh a l re duced F.-eadhcrin expression may play a s ig nilicanl rle in th e p ro g re ssio n o f c a ic in o n ta ce lls tow a rd s a m a lig n a n t p h e n o ty p e ( l ;ri\e il 7 t i l . I I : N a v a r ro el a i. I ` W I). lit I he p re sen t stud y, re d u c tio n in ce ll adhesiveness observed ih trin g n eo plastic p ro g i e ssion u l lt< )SH eelH m es was aecoinpamed hy a re d uctio n in im im m o lo gically detectable f. ca dh eiin expression on blots. T aken together w ith the results o f p re vi ous studies, the results o f o u r study suggest that neoplastic progression o f the OSH may involve m o d ula tio n o f L -ca dh crin m ediated ce ll adhesion. The observation o f a concom itant reduction in R-cadhcrinm c d ia lc d ce ll adhesiveness a nd ( i J I C in the R O S E system is o f p a rticu la r in terest. Recent studies have suggested a lin k between cadhcrin-m ediated cell adhesion and the form ation o f ju n c tio n a l com plexes, inclu din g gap ju nctio n s, re p ortin g that transfection o l cadhcrin cO N A in to com m unicationincom petent cell lines results in the fo rm a tio n o l fun ctio na l gap ju n c tio n s b e tw e e n ira n s le c lc d ce lls (M u s il et a i. I `WO; Jong cn e l at.. I `W I) F u rth e rm o re , the fo rm a tio n o l gap ju nctions induced hy R -cadhcrin expression appeared to in volve p ost-tra n slatio n al m o d ilic a lio n (p h o s p h o ry la tio n ) o f ihe gap-junction p ro te in to a h u m capable o f assem bling into fu n c tio n a l ju n c tio n s (M u s il et ut.. IW II). T h e o b se rva tio n lh a l re duced ({-c a d h c rin e xpressio n in K O S F c e lls is associated w ith a re d uctio n in ( iJ IC s tip p o its the results o l p i o ions studies suggesting that specific c e lic e li adhesion may m o d u late gap-junction form ation and thus lu n c lin n a l co up lin g id cells. H ow ever. ( iJ IC may not always co rre la te w ith I! ca d h e iin e x p ie ssim i, since l : -c .id h c iiii a pp e a re d s u n ila i l*o th in lt ( ) S I { and in ROS|-{ IW . b ut ( . l i e Was d is s iim l.ii m the 2 popola rions \ ( K N t m i I IX .I M l N IS W e thank l) r . ( Stoiea lo r h e lp fu l p a th o lo g ic in te rp re ta tions; D r. R. Ila rh ou m i lo r co ntribu tion s to the gap F R A P studies; D r. I). Ilusbee for assistance w ith tra n s fe c tio n p ro c e d ures and p ro v id in g the pSV2rivi a nd p S V .W o p la s m id s : l) r . Ci. N ie o lson fo i p ro v id in g the p S V '2 /irn .'c l \-ru \ 1 p la s m id : l ) i . V . W ils o n fo r p to i tilin g the Pub 4(15 a n ith o d v : D r I I I )o o k w .ih fo r his assistance w ith p iim .u y c u llm e is o la tio n s . M i. I ' D o gg e tt lo r assistance m the p ic |) . u a lio ii o l lig u ie s ; a n d M i . It. I.e w is fo r le chiue a l assistance. T h is w o rk was s u p p o rte d in p.ri hy g ra n ts fro m the Texas A g r ic u ltu ra l l :. \ |) c iin ic n i S ta tio n a m i the Texas A & M College o l V eterinary M e d icin e l.n h.m e e m ent Program (U .C M i.). These studies com prise a p o rtio n ol an investigatiim subm itted in p a rtia l liillillm e n t lo r ilie D o c to i o f P hilosophy degree in the D c p a m n e n i o l V e le rm .iiy A n a lo m v a n il P u b lic Ile a lth . T exas A \ M l h m e is iiv ( A . ( i . l I ). Iti 11 ItliN t i s A dams. A .T . and A h ik m 'ik o . N . Iransiorm aiion ol cultured ia i ovarian surface e pithelial cells hy k iisicn nm riiic saretnna virus ('inn er / h r . 41. 2063-21172(1 981). A dams. A T. and A di km*i ho. N . A cell line. ItOSP. IW . derived I rum normal ral ovarian Mrlate epithelium, /.'./ Cell Hint. 53. IKI-JKK (1985). Hi iim i ns. J.. M a k i i .i . M .M .. V an K m . and Hik c iim i-ii k . W.. Dissecting tum or cell invasion: cpillu-linl cells aci|iiire invasive proper ties alter the loss ol' u vo m o n iiin -n iciliaicil cell-cell adhesion. J. Cell Hull.. IIUJ. 24.15-2447 <I in#). Ho k in o . Sot mu s. T.S and T ono, T.. Cancer statistics. I*W2. Caneer. 42. I 9 - 3 8 ( I ` 2). llo v H. Ik. T t't Ki m. ( i.( . V ai i I s. A M . I 'hanki . W.W. and T im to . J.P.. Kcarrangemenis o l dcsinosonial and cyloskclelal proteins lim in g the iransition fto n t epithelial to lihrohlaslonl o rg an i/alion in eullured ia i bladder-ca i entonta cells. 7 Cell lh a l. 109. |4`>5 l) i l*i m i m. A.W .. H an. J S.. W h .iiu k s i. C M . Ki adnio . .1.1'.. t it mHIM a n , J I . ( 'iia n o . ( '.( '.. |ti> ii. it J. and I hosKo . J l . Iiile c lio n ol rat liver e pithelial cells with v-l lu-rui currelalm n between oncogene expression, gap juucliunal co iiiniiim calion. and iiiiiH ingeincily. Mnl. Curenit >f;.. 5.54-67 ( |*Wti). I'KIXI.N. m i . . I ll i iki ns. J . S \ i lis. M . t i n hi i . ( i . Voss. Ik. W a k d a . A . l.iH UNi.K. I), and IliKc iim i ii m. W . I tailhe rin-n ie ih aieil cell-cell adhesion prevents inv.isiveneSs o f liunian carcinoma cells. J. Cell Hi,>1.. I D . 173-185(1991) (o im iN . A K . 1 1 s ia . J.K.. I I andi i . I..M . I.m . /... V aniii k m i k . I A.. T h ai i v , P.A. and I Ia m ii io n . T .< S p o n ta n e o u s iranstorm .ilion o f ral ovarian surlace epithelial cells: association w ith cytogenetic changes and im plications o l repealed ovulation in the etiology ol ovarian cancer. J. nut Can, er In s i. 84. 592-601 ( 1992). ( i t >mi k i i k 'K, (M l.. M i vi its. I' J.. Ko k o kis, S.P . Wd no . < i . M i Convim K, l-'.P. and W in I i . R.W n.. Ka\ cn/ynic-linked tmniunohkU assay discrim nales p 2 l sjicccn: a lechniipie to disseel gene lainily expres sion Anal. H uh hem.. I HO, DM - 168 ( I9K9J. I I amii io n . K C . I l l n i* ksiin. W.J ami Ha io n . ( . Isolation and growth o f the rat ovarian germinal epithelium . In. It.J. Richards and k .T. Rajan (cils ). Dwite euhure in me,lh a l research III), pp. 237 244. Pcrgamun Press. Oxford ( IMHO). Ih ik n h i, A.H .. Pa n . i and A di ksh k i.. N.. Inlerm etliale lilainenls hi rat ovarian surlace epuheli.il cells: changes with neoplastic progression in culture. Itiaehem ( ell Hint.. 7(1. 16-2.5 (1992). JoNlilN , W .M.K.. I ' l l / l . l MAI D. D.J.. A SAMOID. M.. I' ll <I ill . ('.. Si a d a . 1' J . fin o s . I).. 'I a m ii h i. M. and Y a m a s a k i. Ik . Regulation ol conncxin -J.t-mediated gap junciion.il iiiii-u c llu l.u coiuinuii.canon hi C A * ' in mouse epulerm.il cells is co n lio llcd hy I -cadheiin. J ( e)l Hint. 114. 545 555 ( PIVI ) Ki Ann. R.( ' . I ' l i / i . i mai D. I ). I . I 'm ni i. ( Si m . \. T.J .nul Y xM xSAM. Ik . (iap-junctional iiili-n e llu la r e o n iiiiiiu ii.ilio n in e p.ile rin.il ie ll lines lu m i selected stages ol S I.N ( A lt niouse-skin eaiiinogenesis Cam it Kes.. 4*. 699- 7115 (I'W M Ki MN. ( i. and Ki I in . H . ({solution ol luinom s .nut the impact ol molecular oncology. Nature II.am i ). 315. 1h. 195 ( 19851 I.AKSIN. W J . I .iiMNi ih . ( S and < m m in . C P ., (ia p p illi lio n m oihilalion: pissihle role in iunior cell hch.isioi In I I A. M ilm .in .owl I-. t imore (e ib -l. Hu hemnul me, hum*m\ ami legal,Hum a l m iei, ella la r eainm nnit alum, pp. | 5 | | 7 9 . Pi in. elou U nisci s iti Press. Pintee Ion. NJ ( Pk7) 1.1 i . I I I . K w an \ ia i. J J . Wn uh k . D M . W o mi d in . I I .uni I t ili k. M I le.pien! loss ol liete to /ig o s i l i on chioinosom i s (n|. l i . and 17 ni human ovanaii la niiHUiias. ( ane,r Ke\.. .Mi. 2724 27*8 ( l ` ll|. I n i \ \ i ns I IN. W R . Juin Donai m iei, ellol.o io iim in n ie ..lio ii and the co iiu o l >1growth. H u i hun (wu/'/io .-l.ru. 560.1 ( 5 |l, i7 'i| M i su . H.S.. ( t'N NiNi.il \M. H.A.. I di i m a s . ( ..M, and ( im >oi s n r ia i. D A . D iirereniial plio splioiyl.iii.m ol lite gap junciion p iolem conneon 43 in im iciional com iniinic.ilioii-coinpclcni and -delicicni celi lines. J ( eli llia l . I I I . 2H77-2IIMI ( IW II) N a XAKKO. P.. (.IM I / . M . Pl/.AKKD. A .. (AMAI IO. ('.. (Jl IM XS|| | M. and C aso. A . A role lo r Ihe I . c.idliei in c e lic e li .i.IIicmoii molecule ilm m g iunior progiession o l mouse cj>idcrinal caicinogcncMs. J ( ell Hint. 115. 5 17- 533 ( IW II Pi n iK s o N . I . Km s i a d . P. .im i l.t'im n ,. II., .in n u a l re fi ai an the reuiltx of treatment ni gv>nvi/iei<a l earner, liilc rn a iio ii.il le d c r .ilio ii ol (iynecologv ami O hslcln. s. V ol 2U. Stockholm. Sweden ( l'JS.sj K iiim . J S . M .iliisU 'p c.iiciiHigencsis ami human e p ilhe li.il cells In ( i . l .. M ilo. M.1 '. <'asn ami I I Shuler teils.j. in im fa imolan o /lm m un e/nihehal eell\- male, alar am i i u u a ^ neii, meth,im.\m\. pp li` l 2III. ( I t e Press, lim a Raion. I l ( l'W2) K iiim . J.S.. Ym. J.H.. I 'a k k . J.H.. I d k s s is . P . S vi i d i . / ami D k iis i m io . A . (.videm e l'or th mullsteis nature >l m r,, human cpilhehal-celt carcinogenesis ( amer Key iS ti/yil ). 50, 5h53s-5r57s <t o m i S un I 1 . I t . I ... ( 'ommoil "e p ith e lia l" tumors. In W .ll I l.o lm .ih <c.< I. .-I/Ars of mmar /iihaloiy : liarais </ the m urv tim i m ahleie la fi tl .g.nu./s. pp. 53-151. Arm ed Forces Insti tuie o f Pathology. Washington. DC (b7V). HOI I M A N / i M. Si ii m iO sM s . I Inn .ii\M ii. N. IliK W o . S . N u i. m ii . M.. Siiim o w it . Y . I am n in. M and Am t ) . C.tdhcnn cell-adhesion mole cule'. in huiii.ni cpuhch.il tissues and carein*mus. `ricer H a ., 4V, J | : s 2 lM<IVNV| Sm iiiiK N I' ..ml III h i.. I1. M.nmn.ili.tii cell n .iiistorm .ilio ii with S\ in In l'ii.l |>l.i'ii)i.l Vcdrs In 'I. (ilu/n m n led.), .ukur\otH n n tl pp 4 1 I', l old Spring H.irtxu l.,iN n ;iio iy. N Y | I `)K). S1 1IN i S . Si* <|. V. ( I ( i l l I 1. K .111*1III -K<il* ahi >i . r t f . . Ral os.iri.m (,'i.im ilovi ell m il hre .> model -oslem lor the siudy ut cell-cell Coiiiiiniiiu.iltou during m uliislrp n.nivfoiiu.itton ( m u e r He\ . $1, f.1. 7 u fi(i*m ij S1 1 r v I .s . Wi i m i. I l l Jk . W ii v i v V.< ami Iti'K i.m a x im . K.C.. I LII lo-cell (o iiin iiiiiK .iiio ii compeicnee in siniian-vitusM Il-irjiisIcclcd m l ts.in.in cells is icditced follow ing tum or vclcciiotl. In l'u ro Cell. I \ r Huit.. 28A.436-444 T m iin . C'J . Bh a im i V. S.M .. Ha k m m an n . CM anil W i in iiix h . K A , Mechanism o f a c liv .iiio ii o l a hum.in oncogene. N u twe (l.n tiJ I. 3Oh, 143 14* (I*' ). I v m k i i i . M . C adherin ecM-.idhesion lecepluis as a morphogenetic rcgnhiior:;Vrrt-if,T251;4 45t--r455Tl *W T W a ii . M i l . T k<m *o . J.I:. and St m in im i k . M.. A lluoicsecnec phoiohienching away o f gap-pinci io n-m cdi.ilcd communie.ilion li is* een human cells. Yt ii-titf. 32. .S25- ( lOKr). Y am asaki. II.. Ciap junciional inicreellular eommunicalion and earei nogene.Ms. CYircirtogivicw.s. 11, K lS l- |l)5k ( t . Ini. J. I tim e r: 5-f. H.*9~H45 ( |W . l) * 1997 W iley-1.is*. In c. ^IC c Puljl>Cdliu(i ol Die inlufnaiiOiMi (Jim m i Ag^,n>i Cfl'H'ci Publication do I Urnon Irue11141o n a to Corme le C.mcei D IF F E R E N T IA L M ETABO LISM AN D TR A F F IC K IN G OF SPHINGOLIPIDS IN D IFFER EN TIATED VERSUS U N DIFFEREN TIATED HT29 CELLS T e re sa B a h ia 1. Jan W ille m Kk ', C aesar M u i s i a i .k i*`. I le n k 1)1 W i i k d ' .iiu I D ic k i 1>i k .M U A 1 ` 1U niversity o fU fiin in ^ e n . f.a h tirtiltin - o f f'h v u o lo tiie til C h e n u \tn . /tlo e tn snivel 10. 0712 K /. i in>nini;en; a n ti :t 'n n e r s ii\ o j ( ironingen. L a b o ra to ry o f ll i s l o l t w a m ! i e ll lim /n ifi, ( )oslerxingt`l 6 9 / / . 9 7 /7 ! . / . ( toniitfien. The N etlierlam ls. Trafficking and m etabolitrn of sphingolipids were examined in undifferentiated (G *) and differentiated (G* reversed) H T I9 human colon adenocarcinoma cell lines. M etabolic experiments employing a fluoresce ntly labeled sphingofiptd precursor, 6-[N(7 -n itro -I, I ,)-benzoxadiaxol-4-yl)amino]hexanoylceramide (C,NBO-ceram ide) revealed that both qualitative and quantitative differences exist in sphingolipid synthesis between the 2 cell lines. One o f the C4-NBO-sphingolipids synthesized in G * cells is not found in the G ' reversed cells. Furtherm ore, the ra tio of the 2 main products, C4-NBO-glucosylceramide and C4-NBD sphingomyelin, differs: in G * cells glucosylceramide is by far the main product, whereas G ` reversed cells synthesize C*NBD* sphingomyelin in slight excess. Once established, these ratios of sphingolipids are quickly restored metabolically when distortion of the ra tio caused by experim ental manipulation. This indicates th a t they represent a true metabolic equilibrium situation o f th e 2 sphingolipids in these cells, while the distinct ratios are mainly determ ined by the NBD-Itpld pool in the plasma membrane. Preferential synthesis and transfer of glucosylceramidc from its site of synthesis to the cell surface do not occur when the plasma membrane pool of glucosylceramide is selectively removed. This suggests th a t instantaneous replenish m ent via specific signalling is probably not involved as a mechanism in re-establishing perturbed lipid pooh. In conjunc tion w ith observations on distinct lipid trafficking pathways of glucosylceramide in C ' and G* reversed cells, the present m etabolic studies emphasize a relation between the expression of this glycolipid and the state of differentiation of HT29 ceils, o /9V.? W tky-Liss. h u . G ly c o s p h in g o lip id s arc in v o lv e d in c c ila in pr*ccsscs such as cell U itfc rc n iia lio n and tra n sfo rm a tio n (M a k o m o ri. 1981; Mannun and Bell. 1989). These processes m ay he accom panied by changes in g ly c o lip id c o m p ositio n . Q u a n tita tiv e ly , changes in ihc relative am ounts o f various g lycolipid* may occur. In a q ua lita tive sense, a ce rtain type o f glycolipid may (dis)appcar. w hich can then be used as a d ill c rc n iia l io n m arker, h rem ains to be e sta b lish e d , how e ve r, w h e th e r these changes a rc It be considered as a cause o r a consequence o l the d iffe re n tia tio n process. In the present study, scvcial aspects o f g lyeo sp hing olipiil tra ilie k in g and m etabolism were exam ined in the hum an colon adenocarcinom a c e ll line IIT 2 9 . F ro m the H T 2 9 G " cell p op ulatio n , a w e ll-d iffe re n tia te d c n ic ro c y lc -lik c cell pop ula tio n d e fin e d as H T 2 9 Ci can he selected b y c o m p le te ly w ith h o ld in g sugars fro m the gro w th m edium (Z w c ih a u tn c f a l, 1985). W h e n th is I I T 2 9 Ci ce ll lin e is again s u p p lie d w ith glucose, a no the r d iffe re n tia te d c e ll line. C r reversed, can be obtained (Z w eibaunt; personal co m m unication). A s such, this cell lin e has been used to study various processes in re la tio n to d iffe re n tia tio n (G a rn e t ct al.. 1991; V iu lla rd et al.. 1991). W e have em ployed the la tte r system to study tra ffickin g a nil m etabolism o f (glyco)sphingoljpids. am i addressed the ques tion w h eth er d iffe re n tia tio n oi the cells co uld affect these processes. In studies concerning (lie tra ilie k in g o f lip id s in ic la lio ii to ih e ir m e ta b o lis m , the use o l llu o rc s c e n tly lagged lip id s has p ro ve d to he o f g re at v a lu e (P a g a n n a nd S le ig h t, 1985; ilo c k s lra and K ok. 1992). W e have em ployed ( V N B D - lip iils m a m in th cr o f lip id transport and m e la lxilism studies (K o k ft al.. 1989, 199(1. 1991. 1992). in a previous study (K o k f t t il. 1991) we have shown that d iffe re n c e s exist in e n d o e ylie tr a ilie k in g o f ( ( ,. N B D -)g lu c o s y lc c ra m id c b etw e en Ci* and ( ' reve rsed ce lls, as th is g ly c o lip id is s pecifically salvaged and recycled via the G o lg i app aratu s to the cell surface in G ' cells, hut not in (* reversed cells. T he pre sen t s tu d y shows th a t th e re is also a p ro n o u n c e d d iflc re n e e in the synthesis o f glucosylceram ide in G * re ru n G ' re vciscd cells, as reve ale d hy m o n ito rin g s p h in g o lip id m e ta b o lis m in cells in cub a te d w ith a flu oresce nt e cra nm le d e rn a liv e ( ( ',, - N B D eeram idc). a com m on sphingolipid precursor. It appeals that long-term m etabolic differences are reflected m sh ort-te rm d ifferences in tra ilie k in g o f (N ltl)-)s p h iu g o iip id s in the e n d o eylie pathway. T aken together, life results indicate that the expression and tra ilie k in g o f g lu co sylce ra m id e m ay he re la te d t o the stale o l c e ll d iffe re n tia tio n , as. in c o n tra s t to its fate in the C i' reve rsed cells, the lip id is n ot o n ly specifically so rted b u t also specifically m etabolized in the H T 29 G ' ce ll line. Mu tfrial M /M T U IA I AND M l I HODS 1-JJ-D -glucosylsphingosine. e e rcb ro siile type II, N -p a ln u lo y lD L -d ihydrolaclocerebroside. sphingosylphosphorylcholinc and D -sphingosm c were purchased fro m Sigma (S i L ouis. M O ). 6-|A/-(7-niiro-2.1.3-benzoxadiazil-4-y|)am ino|hexanoie acid was obtained fro m M olecular Probes (F.ugenc. O R ). U rid in e d ip h o s p h o - 0 - |/ /- N C'lglueose, a m m o n iu m sail ( > 2(N) m C i/ m m ol) was supplied by A m crsh a ni (A ylesbury. U K ). Synthesis o f i N H D - lo M e tl .\y>/iui,i;r/r/;ir/v C.'(,-N B I)-g lu c o s y le e ra m id e , C',, - N B I ) g a la e to s y le e ra m id e . C h-N B |)-la c to s y le e ra n id e . C .-N B D -s p h in g o m y e lin a nd N B D -c e ra m id c w e re syn th e sized fro m ( ',,-N H I) a n d I g lu e o s y lsp h in g o sin e . 1-(J -U -g a la c io s y ls p h in g o s m e . I - J i- |) |. laetosyldihyiirosphingosine. sphingosylphosphoiyleholine am i D -sphingosine. respectively, as described (K is h irm tio . 1975). I B -D -galactosylsphingosincam i I- |M ) I -laciosyldihydiosphirigosine w e re p re p a re d (C io ila et a l. 1987) h o rn c c ic h io s id c (type I I ) ami N -pn lm iloyl-l)I.-d ih ydroi:icloccich ro sid c.re spe etively. T h e C V N H D -lip k ls were q u a n tifie d s p e e iro p lio to n ie trieally in a flu o ro m c ie r ( P e rk in -f:lm c r-M P I: -47) w ith an cxcita (io n w a velen g th o(' 4(>5 n m and an e m issio n w a ve le n g th o l 5711 nm, by reference to known am ounts o f <V N H !)-p h o s p h a tid y l c h o lin e . C ell fa ilu re M o n ola ye r cu ltures o l H T 2 9 cells w ere g ro w n in D M IiM co n ta in in g 25 m M glucose, su pplem ented w ith 1 0 'i (v o i/v o i) *'l u whom correspondence amt reprint requests should he sent A h h in im in to USA. In nine scrum albumin: ( ,, N l l l ) . (7 -iiu ro 2.1.7-bcn/oxadi.i2ol 4-y|),iiiiiiioJhex.movl oi -lievanoic acid. | ) M ! M . Dulhccco's modification o f minim al essential inediiiin; C i'A il u c ' . (IIT 2 V ) cells growing in the presence o f glucose; ( i /G lue . (H T 2 `>) cells growing hi the absence o f glucose; (I N * )T I( h ig fi- p e r lo iin a n c e ) lliin-H iver chrom .ilogiaphy; !'( Si. fetal cell h ' i i i i i i . Received: January 15. 1997 and in revised lo n n M aieh 15. |9H 8 8 I.jI.w* O M EN S HEAl IH A f f V d . 4. No 2 M a n h 994 84-88 m a n y in d iv id u a ls a n d o rg a n iz a tio n s i f we are to meet o u r u ltim a te g o a l-h e a lth fo r a ll o u r citizens. REFERENCES 1 Ri>v. JM Surgical jt\ncn'l.j;y la : A pple R A , cd Health, tin./ M rJiiin e ii .Am riiiii A I lii|i>n.,il I lundfvx'l N ew Y o lk , (ia r la n d P u b lid iin g . l<WO.I7t- fJS. 2 U S IV p a im ie n i tf H ealth a n J I lu m s n S e r v ito R ep o rt o f th e Public M c.ilih Sei M ie Tm k l o n e . W om en' H ealth Issues, V o lu m e II D i l l i > Pub N o (PUS) 88 V-2tYi. tX io h e i I8?. I N adel. M V . N ational Im unite of H ealth: Problems m Im plem enting P o lio on W om en m Study Populations. U nited State Ciencral A c c o u n tin g O rtu e . A u :n h m e m l , lu n e 18, I'WO. ^ U S IV p .ittinriit of H ralih a n J H um an Services. Public H ealth Service, N ational Institutes of H ealth. NIH Instruction and Inform ation M emo ran d u m o n O ER 00-5 - Inclusion of M inorities and W om en in Study P o pulation*. N IH M a n u a l 4000 an d 6000, 1990. 5. US D epartm ent of H ealth and H um an Services, Puhlic H ealth Service, N ational Institutes of H ealth. Report of the N ational Institutes of H ealth : O p p o rtu n itie s for R esearch o n W o m en 's H ealth- N IH Publica tio n N o . 92-3457A , Septem ber 1992 6 . N ational Institutes of H ealth R e v a lu a tio n A ct o f 1993, Public Law 103-4). June 10, 1993. 7. M c rk a tt RB, T em ple R, Sorel S , F te id e n K. Kessler D A . W o m en in clinical trials o f new drugs. N Engl J Med. 1993;329:292-296. 8 . Federal Refiner V ol $8 (1)9), T hursday, July 22, 1993. 9. Institute of M edicine. W om en and H ealth Research: T he Ethical and Legal Usues of Including W om en in C linical Studies. W ashington, D C ; N atio n a l A cadem y Pres*. 1994 (in press). 10. N atio n a l In stitutes o f H ealth: R ecru itm en t, R e te n tio n . R e c n n y . an d A d v a n c e m e n t o f W om en In Biom edical S cience. 1994 (in press). Epidemiology and Primary Prevention of Cancers of the Breast, Endometrium, and Ovary A Brief Overview JENNIFER L. KELSEY, PhD, AND ALICE S. WHITTEMORE, PhD ABSTRACT G m c m o f th e b rea st, e n d o m e tr iu m , a n d o va ry account fo r 4 J% o f in c id e n t co n c ert a m o n g w o m e n . M a n y risk fa c to r s fo r b re a st C o n o r h a v e b e e n id e n tifie d , b u t m o a r e a s s o c ia te d w ith only m o d e s t e le v a tio n s in risk. A ls o , o f a ll th e ru k fa c to rs id e n tifie d , f e w a r e lik e ly to he a ffe c te d b y intervention p r o g r a m s . T h e p a th o g e n e sis o f b re a st cancer is not w e ll u n d e r s to o d , b u t estro g en a n d p o s s ib ly estrogen />1u j p ro g e ste ro n e a r e lik e ly to he etioJogicaUy involved. F o r e n d o m e tr ia l c a n c e r , a m a jo r etio lo g ic pathw ay is exposure to estro g en w ithout cy c lic exposure to p ro g e ste ro n e . M ost o f the e s ta b lis h e d risk fa c to r s fo r e n d o m e tr ia l c a n c e r a p p e a r to a ffe c t risk a t le a s t in p a r t th r o u g h th is p a t h w a y . O n ly a f e w risk fa c to r s fo r o v a r ia n cancer have been id e n tifie d . T h e tw o most commonly su g g e ste d e tio lo g ic m e c h a n is m s fo r ovarian c a n c e r a r e (a) t h a t s u p p re ssio n o f ovulation reduces risk and (b ) that suppression o f p itu ita r y g o n a d o tro p in s reduces risk. Each h yp o th e sis is c o n sisten t w ith so m e, bsU n o t a ll, o f th e d a ta . P ro sp ec ts fo r the prim ary prevention o f th e s e cancers a r e d is c o u r a g in g a t present because f e w o f rhe risk fa c to r s id e n tifie d to date are readily s u b je c t to m o d ific a tio n , e s p e c ia lly fo r b re a st cancer, and the underlying etiologies o f th e se ca n c e rs a r e n o t w e ll understood, p a r tic u la r ly fo r b r e a s t a n d ovarian cancers. Ann E pidem iol 1994;4:89-95. k by w o r d s : Breast cancer, endom etrial cancer, ovarian cancer, epidem iology, prevention. IN T R O D U C T IO N Cancers o f the breast, corpus u te ri, and ovary account for abo ut 41% o f new ly diagnosed cancers and abo ut 26% o f cancer deaths am ong w om en (1). T h us, id e n tific a tio n o f risk factors in the hope th a t means o f p re ve n tio n can be suggested is o f consid e ra ble societal im p o rta n c e . T h is review sum m a rize s w h a t c u r r e n tly is k n o w n o f th e e p id e m io lo g y o f cancers o f rhese th re e sites, discusses th e possible roles o f h orm on e s in th e ir e tio lo g y , a nd b rie fly describes o n g o in g and p lanned ra n do m ize d tria ls to evaluate possible p reven tiv e measures. S ince far m o re e pid e m io lo gic research has focused o n b re ast ca n ce r th a n ca nce rs o f th e o th e r tw o sites, breast cancer receives th e greatest emphasis. M o s t o f the m a terial o n breast cancer epidem iology can be fou n d in m o re d e ta il in a re c e n r issue o f E p i d e m io lo g ic R e v ie w s (2 ), w h ile m uch o f the d escriptio n o f o varian cancer epidem iol o g y Is d e riv e d fro m a re ce n t series o f a rtic le s b y W h itte m o r e and colleagues (3 -7 ). W ith th e e xcep tion s o f age a nd In te rn a tio n a l v a ria tio n , descriptive e pidem iologic characteristics are n o r considered because o f space lim ita tio n s . From the D epartm ent of H ealth Research and Policy, Stanford U niverlity School of M edicine. Stanford, C A . A d d re te rep rin t, r e q u e m to : Je n n ifer L. Ke1*ey, P hD , D ivision o f E pide miology, D epartm ent of H ealth and Reaearch Policy, Stanford U nlvetilty School of M edicine, S tanford, C A 94305-5092. R eceived A pril 14, 1993; re v lie d S e p tem b er 16, 1993. IW l by Ebrvitr Scimic Int BREAST CANCER E p idem iology In the U n ite d States, breast cancer incidence rates increase steeply w ith age th ro u g h the 40s, th e n c o n tin u e to increase, b u t at a slow er ra te, th ro u g h the rest o f th e life span (8). G e n e ra lly accepted risk factors fo r breast cancer (8 -1 6 ) are sum m arized in T a ble 1 a ccording to the a pp ro xim a te m a g n i tu d e o f th e re la tiv e risk associated w ith th e m (9), M o s t o f the c h aracte ristics associated w ith re la tive risks h ig h e r th a n 2.0 are re la tiv e ly u n c o m m o n . A ls o , o f a ll these g enerally accepted risk factors, o n ly o besity m ig h t be affected b y in te r v e n tio n p ro g ra m s , a n d an effe ct fro m w e ig h t r e d u c tio n is u n lik e ly to be large. O t h e r a ttrib u te s m ig h t be ca te g orized as " p r o b a b le " ris k factors fo r breast cancer. A lth o u g h they have been id e n ti fie d in several stud ie s, e ith e r th e s u p p o r tin g e v id e n c e is s o m e w h a t in c o n s is te n t o r th e re is c o n c e rn t h a t u n id e n tifie d c o n fo u n d in g m ay account fo r the association. T allness, fo r in stan ce, has been associated w ith increased ris k in m a n y studies, b u t n o t a ll (16). M o s t studies, a lth o u g h the re are exceptions, suggest th a t h ig h p a rity p rotects against breast c a n c e r d ia g n o se d a fte r th e age o f 50, e ve n a fte r t a k in g in t o accou nt the association betw een h ig h p a rity and e arly age at first fu ll-te rm pregnancy (15). H o w e ver, m u ltip a rity may be associated w ith an Increased risk fo r breast cancer d ia g nosed b efo re th e ages o f 40 to 50 (15). A re la tiv e ly s h o rt l(M7-2W/lX/K'7.00 9 0 k f lx v and Whmrmnre HREAST. ENPOMETRIAI . ANO OVARIAN CANCERS A f P Voi 4. No 2 March 1994: 89-95 T A B L E I. Established risk factors for breast cancer1 l\h ! Ur f.rriH c ) Age (Si t leogt -I'lu. -irr.i (JO M o llic i .<J MctCT u n i . h , cM fi o f h rc .i'i m in e r , p e n a lly il diagli need al :m carle ace |9 l illu m ed Iv iiir ii proliferane e 1 rea J ic c a x u f i a u p i.. i Ic'l A i\ pi .I ..jn f.c li..f .clU in1 nipple .o p in ile filliJ (111 N u.lnl.ir .j. IIMI|C, w iiW M ntl?! M i.lorc o'1 . .iik ci in o n e 1'r e a ( 1 1 M->li- 1 ...' Mclcr U.11H Kim.tv o f b r e a d m i n e r (9) .......... u lu lim i benign | >io liler.itivc b re a d di>c:ise Mlll.m a, p ia tii 1 lv|<rf|.| i.. r p i l h r l u l .r l l . iecu in mi at j pia in m jiplc n p i r a i r fini,1 III1) I U I . . I i . f l I...I.C M ... n m lletale lo h ig h d o c ( | 4) Krmiic.il >1 oc ance hetnre *ee -fi' (I5 l NufllpIMI v. Ihcim cancer al age > 40 v (IS) A rc m iic-< full i n m prem am v I I M A rc un ii.li lie 113) A rc al ihi-||op.HlC (I*) H imoiv ,-f prim ary ea rn e r hi ciidoiiH 'iriiim . o v a ry (1 ll fkvjy n u v tu ra s i i nccr ai age > W y (lt>) K J y m a, b ic a e a rn e r at age < 50 y (16) AJ..,.. n t f.om K rhrv (9) High-risk group Relative risk > 4 .0 > 50 Y N orth America, northern Europe Y Ye* Ye* R elative ride o f 2 . 1 -4.0 Densities occupying < 75% of breast volume Ye* Ye* Ye* Yc* Ye* Relative risk of 1 .5-2.0 No Ye* > 10 y < II y > 55 V Ye* Obese T h in Low-risk group < 30 y A*ia, A frica No No N o fluid produced Parenchyma composed entirely o f fat No No No N o fluid produced No Ye* No < 20 y > 15 y < 45 y No T h in Obese num be r f days hoc w een m e n stru a l p e rio Js appears ii he associated w ith an in creased ris k , a lth o u g h m o re s tu d y o f th is i>Mie is needed b efore c o n clu sio n s are re ached (15). M o st studies th a t in c lu d e d re la tiv e ly large n u m b e rs o f y o u n g w o m e n w h o used o ra l co n tra c e p tiv e s fo r 10 t o 15 years o r lo ng e r suggest th a t th e risk fo r breast ca nce r is s lig h tly increased a m o n g these lo n g te rm users (17). L ik e w is e , m ost of the relatively sm all num be r o f studies th a t in c lu d e d sulfi<lent n um be rs o f w o m e n w h o used e strogen re p lace m e n t th e ra p y lo r 10 t o IS years o r lo ng e r in d ic a te th a t su ch lo n g te rm ns- is associated w ith a s lig h tly e leva ted risk (IS ). 1 he lim ite d data presently available suggest th a t use o f h o rm o n e replacement therapy w ith b o th estrogen a nd p ro ge stin in crease^ risk at least as m u ch as use o f e strog e n a lo n e (19). W o m e n w h o used d ie th y 1stilh c s tro l (D E S ) in th e 1940s th ro u g h i he 1960s to p re v e n t e a rly te r m in a tio n o f p re g nant v appear to have an elevated risk, b u t the p o s s ib ility th a t tin - reasons fo r its use were the im p o rta n t fa c to r ra th e r th a n the lM ' S itse lf a n n u l be ruled o ut (20). M o st studies in d n ate th a t a lc o h o l C o nsu m p tio n is associated w ith a m o d estly i i u teased ris k , b u t in th e abseiue o f a g e n e ra lly nc le p tc d t .nival m echanism , co n fo u n d in g b y a h ith e rto u n id e n tifie d nsk facto r ca n n o ! he e \ lu d e d (21). S om e e vide n ce, h u t n o i a ll, suggests th a t la c ta tio n , p a r tic u la rly if it is p ra c tice d fm m a n y years as m ce rta in A sia n c o u n trie s , p ro te c ts against breast cancer, in d e p e n d e n tly o f the association o f d u ra tio n o f la c ta tio n w ith early age at first b irth and m u ltip a rity (15). M o st o f these p o te n tia l e tio lo g ic factors appear to confer o n ly m odest changes in risk, w ith re la tive risks in the h igh-risk g roup com pared to the low -risk g roup gener a lly less th a n 2.0. F in a lly , expo sures th a t m ig h t be c a te g orized as " possib le " etiolo g ic factors, and th a t arc p ro b a b ly at m ost w eakly asso ciated w ith breast cancer risk, in clu de lack o f physical a c tiv ity (22), exposure to e lectrom a g ne tic fields (22), and ce rtain c o m p o n e n ts o f d ie t, su ch as a lo w in ta k e o f v ita m in A a n d a h ig h in ta k e o f calories d u rin g c h ild h o o d a nd adolescence (16). A lth o u g h an increased risk fro m a h ig h -fa t d iet has been h yp o th e s iz e d o n th e basis o f a n im a l e x p e rim e n ts a n d v a ria tio n in incidence rates a m ong co u n trie s , m ost evidence fro m e pidem iologic studies does n o t su pp ort th is hypothesis (16), at least in respect to a h ig h -fa t d ie t in a d u lth o o d . R e la tiv e ly few studies have evalu a te d the e tiolo g ic role o f these "possible" risk factors o th e r th a n d ie t, and no firm c o n c lu sions can lie reached at th is lim e . P ossible E tio lo g ic P a th w a y s T h e b io lo g y o f bre ast ca n ce r is n o t w e ll u n d e rs to o d a n d is u nd o u b te d ly co m p lex, b u t evidence stro n g ly indicates th a t Ai A E P Vol. 4, No 2 Morck 994 8 9 -9 5 Kd*ry and Whkrrmui* BREAST. ENDOMETRIAL. AND OVA RIA N CANCERS 91 ovarian hormones are important in its pathogenesis (23). The increased risks associated with early menarche and late menopause and the protection afforded by early removal o f the ovaries suggest that longer exposure to ovarian hor mone* increases risk. The elevation in risk from postmeno pausal obesity is consistent with involvement of estrogen, since most estrogen in postmenopausal women is derived from conversion from androstenedionc in adipose tissue, and the greater the amount of adipose tissue, the higher the concentration of circulating estrogen. Also, in obese individuals, less estrogen is bound to sex horm one-binding globulin and is more available to target tissue than in thin ner women. Possible associations o f breast cancer risk with shorter length of time between menstrual periods, long-term use of hormone replacement therapy, long-term use of oral contraceptives, parity, and lactation also suggest hormonal involvement. It is believed that ovarian hormones affect breast cancer risk by increasing mitotic activity, which in turn increases the likelihood that D N A damage will be convened into a stable genetic error (24). Estrogen alone induces some cell division, but estrogen and progesterone together (as during the luteal phase of the menstrual cycle) may induce more cell division. Thus, exposures associated with elevated estrogen levels, and possibly elevated levels o f both estrogen and progesterone, are at present o f considerable Interest as po tential risk factors for breast cancer. Evidence for a role of other endogenous hormones, such as prolactin and andro gens, is less compelling, but further study is needed. Heredity appears to be important in the etiology of some breast cancers, especially breast cancers diagnosed at early ages (9). As breast cancer susceptibility genes are identified with new technologies in molecular biology, the study of gene-environment interactions will be o f great interest. L it tle is known about etiologic pathways for increased risk related to disease susceptibility genes. E N D O M E T R IA L CA NC ER E p id e m io lo g y Endometrial cancer is rare before the age of 40 years, after which incidence rates increase steeply w ith age until 60 to 64 years and then decrease somewhat (25). Several risk fac tors for endometrial cancer, shown in Table 2, are well established, and relative risks are generally higher than 2.0 (13, 25-32). Although estrogen replacement therapy w ith out progestin increases the risk for endometrial cancer (see Table 2), risk appears not to be increased if progestin is added cyclically for 10 days or more per m onth to estrogen as a hormone replacement regimen (33). Use o f sequential oral contraceptives (34), early age at menarche (35), and tallness (35) are probably also risk factors, while cigarette smoking is probably protective (35). Some evidence suggests that a family history of endometrial cancer (29) and a history of diabetes increase risk (25), while late age at last pregnancy may be protective (36). However, studies of these factors have either been few (family history o f cancers, late age at last pregnancy) or inconsistent (diabetes), and no conclu sions can be reached at this time. Etiologic P athw ay o f Several R isk Factors Most of the risk factors identified for endometrial cancer involve exposure to estrogen w ith insufficient cyclic expo sure to progesterone, and this explanation is generally ac cepted as a major etiologic pathway for the development of endometrial cancer. Unopposed estrogen increases the frequency of mitosis in the endometrial cells, while progestogens markedly reduce mitotic activity (37). A ll o f the gener ally accepted factors associated w ith an increased risk (ex cept pelvic irradiation and possibly pregnancy) are believed to contribute to the development of endometrial cancer at least in pan through this pathway, while combination oral contraceptives and particularly exposure to depot-medroxyprogesterone acetate (D M P A ) provide greater exposure to progesterone and are thereby protective. Cigarette smoking may be protective because of a possible association with lowered circulating estrogen levels (37). A late age at last pregnancy could be a marker for decreased risk; being able to become pregnant at a late age may indicate that a woman has more normal ovulatory function and thus sufficient exposure to progesterone compared to a woman who is unable to become pregnant relatively late in her reproduc tive life (36). Hereditary predisposition to endometrial can cer has not been well studied. E P IT H E L IA L O V A R IA N C A N C E R E p idem iology Cancer of the ovary has been less well studied than breast and endometrial cancers, and only a few risk factors have been identified with any degree o f certainty (3). Epithelial ovarian cancer, the most common histopathologic type, Is uncommon before the age of 40, after which incidence rates increase until the ages o f 70 to 79 and then decrease some what (25). Established risk factors for epithelial ovarian can cer are listed in Table 3 (4. 13. 25. 26, 36). T h e femilial aggregation is believed to be in part attributable to heredi tary susceptibility (39, 40). A gene located to chromosome 17q21 has been found to predispose to breast and ovarian cancer in some families as well as to some early-onset breast cancers (41). Possible risk factors that have not been exten sively studied or for which the evidence is inconsistent or difficult to interpret include perineal talc use (42), fertility drugs (4), galactose from lactose in milk (43), caffeine con sumption (44), and a history of mumps (45), while tubal 92 Kebev and Whitremorr BREAST. H N lW M tl RIAL. AND OVARIAN GANGERS AP Voi 4. No 2 Mare* 1994 89-95 T A B L E 2. Established risk factors for endom etrial cancer r-c i... (iclcrerur) High-risk group A rc (25) G eo g ra p h ic area (26) U< of c m r-urn replacem ent th e ra p y (27) Body m a n (28) N o. of pregnancies (2) Age at m .-iK ip a u x ()0) l'<< of c o m b in a tio n o ral c o n tr a iepcives 0 1 ) I 1* of P M PA 021 M otori ot primary carnet in bieast (H ) Relative risk > 4.0 60-64 y N orth A m erica, n o rth ern Europe >5y Relative risk of 2.1-4.0 Obese >4 > 55 y No No Relative risk of 1.3-2.0 Yes A w xiation* that appear strong, from clinical observations, but are difficult to quantify exactly lin .lx n iru i.il liv|v-rpl.iM j |2S) F 'li i i t r i i p r.d u e r n a ov,liuti tu itio n (?5) Stem L c irn c lia l syndrom e (2S> IV Im it c u lla tim i (25) P M P .\. .I-|j* nirilin>v|'i||r>lriuii( k r u i r Ye. Yes Yes Yes Low-risk group < 40 y A sia, Africa None T h in 0 <45 y >4 y Yes No No No No No lig a tio n and hysterectom y possibly are p ro te c tiv e (4). A ge at m c n a rc h c , age at m enopause, age at first b ir t h , a n d e strogen replacem ent therapy appear to have little o r n o effect o n risk (4). P ossible Etiologic P a th w a y s o f S om e Risk Factors T w o m a in biologic explanations have been put forw a rd to explain the associations between reproductive variables and risk o f e p ith e lia l o v a ria n cancer. T h e firs t is th a t s u ppression o f o v u la tio n reduces risk. Fathalla (46) hypothesized th a t o v a ria n carcinogenesis involves some m echanical sequelae o f o v u la tio n , such as tra u m a o r m ito tic s tim u li to th e o v a r ia n e p ith e liu m . T h is h ypothesis suggests th a t p regnancy and o ta l contra ce ptive use p rotect against o v a ria n cancer b y in h ib it in g o v u la tio n . T h e second is th a t su pp re ssio n o f secretion o f p itu ita ry g on ad o trop in s decreases risk. G a rd n e r (47) and St a d d (48) hypothesized tha t exposure o f the o v a r ian e p ith e liu m to |K*rsistently h igh circ u la tin g levels o f p itu ita ry g on ad o trop in s increases the lik e lih o o d o f m a lignancy. T h i* h y p o th e s is suggest, th a t pre gn an cy a n d o ra l c o n tra c e p tive use p ro te c t against o v a ria n cancer b y in h ib it in g p it u ita ry secretion o f gonadotropins. W h ile a num ber o f epidem iologic observations support b o th hyfxithcscs, there are exceptions. D iffe re n tia l risk re d u c tio n associated w ith pregnancy and o ra l co n tra c e p tiv e use (p re g n a n cy a p p ea ring to be th e m ore e ffe ctive , p a rtic u la rly in y o u n g w o m e n ) (4) c o n flic ts w ith che firs t h yp o th e sis; this hy|xxhcsis predicts equal pro te ctio n per year o f preg n an cy and per year o f o ra l co ntra ce ptive use, since these c o n d itio n s arc equally effective in suppression o f o v u la tio n . T h is d iffe re n ce is m o re c o nsiste nt w ith th e second h y p o th e sis, because th e lo w -p o te n c y o ra l c o n tra ce p tive s m a y he less effective than pregnancy in inhibiting pituitary secretion of gonadotropins. O n the other hand, observations of reduced ovarian can cer risk associated with lactation conflict with the second hypothesis, because lactation increases secretion of folliclestimulating hormone. This result supports the first hypothe sis, since lactation induces partial inhibition o f ovulation. Also in conflict with the gonadotropin suppression hypoth esis is the absence of protection associated with estrogen replacement therapy, since conjugated estrogens suppress pituitary hormones in postmenopausal women, although not to premenopausal levels. Some findings, including the observations that age at menarche and age at menopause are not at all or only weakly T A B L E 3. Established risk factors for epithelial ovarian cancer Factor (reference) H ig h -risk group Low-risk group A ge (25) G eo g rap h ic area (26) Relati vc risk > 4.0 70-79 y N orth America, northern Europe R elative risk o f 2 .1 -4 .0 Family hiitory uf ovarian Yes cancer (38) N o. o f p regnancies (4) 0 O ta l c o n tracep tiv e use (4) None Relative risk of 1 .3-2.0 B reast-feeding (4) None H istory o f p rim ary cancer in Yes b re a st (1 3) or endom etrium (25) < 40 y A sia, Africa No >4 > 6y > 1V No AP Vol. 4, No. I March 1994- 89-95 Kelsey and Whinctnorc BREAST, ENDOMETRIAL, AND O VA RIA N CANCERS 93 related to risk, are inconsistent w ith both hypotheses. Odds ratio attenuation resulting from error* in reporting personal characteristics may be responsible for some of these inconsis tencies. Further research is needed to clarify the tiologie roles of ovulation and gonadotropin stimulation, both of which may enhance carcinogenesis in the ovarian epithe lium. Identification of interactions between genetic and en vironmental factors will also be of great interest. A C O M P A R IS O N O F SE LEC TED RISK FA CTO RS T h e incidence rates of cancers of the breast, endometrium, and ovary in various countries around the world are rela tively highly correlated (26), and the demographic charac teristics of high-risk women are somewhat similar for the three cancer* (white women after ages 40 to 50, nevermarried women, and probably Jewish women, and women of high social class) (25). In addition, multiple primary can cers involving these three sites occur somewhat more often than would be expected by chance (13). It therefore used to be thought that they might to some extent have similar etiologies. However, comparison o f other risk factor* indi cates that there arc in fact many differences. Whereas long term use o f oral contraceptives may slightly increase the risk for breast cancer, use o f combination oral contraceptives decreases the risk o f endometrial and ovarian cancers. W hile use of estrogen replacement therapy strongly increases the risk for endometrial cancer, long-term use is probably associ ated with only a slightly increased risk for breast cancer and no alteration in risk for ovarian cancer. Although data are limited at this time, hormone replacement therapy with both estrogen and progestin is probably associated with an increased risk of breast cancer, but no alteration of risk for endomentrial cancer. After the age o f 50 years, obesity is associated with a modestly increased risk for breast cancer, a more marked increased risk for endometrial cancer, and no alteration in risk for ovarian cancer. Below the age of 50 obesity appears to be protective against breast cancer, increases the risk for endometrial cancer, and has no effect on ovarian cancer risk. Thus, in formulating ideas for preventive measures, care must be taken that while attempting to reduce the frequency of one of these cancers, the incidence of another is not increased. Also, preventive measures for one o f these can cers could, o f course, have an adverse effect on other body organs. PROSPECTS FOR P R IM A R Y P R E V E N T IO N Prospects for primary prevention of these cancers are dis couraging at present. Most attention has focused on the prevention of breast cancer. As noted above, of all the established risk factors for breast cancer, only postmeno pausal obesity might be amenable to intervention efforts, and it is associated with only a modest elevation in risk. If such exposures as long-term use o f hormone replacement therapy, long-term use of oral contraceptives, and alcohol consumption are firmly established as ruk factors, then some preventive efforts may be applied, although other ad verse and beneficial effects of these agents, including those related to risk of endometrial and ovarian cancer, will need to b e considered. Use o f oral contraceptives and o f hormone replacement therapy that contains progestin as well as estro gen could reduce the incidence o f endometrial cancer com pared to the incidence when replacement estrogen alone is used, but use of such a combined regimen could adversely affect risk for other diseases, such as breast cancer and coro nary heart disease. Sufficiently little is known o f the epidemi ology of ovarian cancer that it is difficult to suggest many preventive measures-, use of oral contraceptives provides some protection and could be recommended for women at very high risk. Despite the apparent lack of primary prevention mea sures suggested by analytic epidemiologic studies, three pri mary prevention trials in which breast cancer is a major end point are now underway or are about to start in the U nited States. T h e randomized trial component o f the Women's Health Initiative will have as its primary objective to determine whether a low-fat diet, hormone replacement therapy, and a calcium /vitam in D regimen in adulthood affect the risk for coronary heart disease, breast cancer, colon cancer, and hip fracture. However, observational epi demiologic studies provide little evidence that a high-fat diet in adulthood increases the risk for breast cancer, so it is unlikely that a low-fat diet in adulthood will decrease risk. Pharmacologic interventions also are being tried, in part because o f the lack of feasible methods of primary pre vention by environmental or behavioral change. A large randomized trial, the Breast Cancer Prevention T rial, has already begun in the U nited States and Canada; in this trial the chemotherapeutic agent tamoxifen, which has been shown to reduce the frequency o f contralateral breast cancer and to increase the length o f disease-free survival in women w ith cancer in one breast, is being administered to appar ently healthy women 35 years and older, exclusive of women at very low risk of breast cancer. There is some evidence that tamoxifen also may favorably affect mortality from coronary heart disease and loaso f bone mass in the vertebrae (but not the forearm) o f postmenopausal women with breast cancer. However, little is known of the effects o f tamoxifen in healthy women; potential side effects o f concern Include such rare but serious complications as thromboembolism, endometrial cancer, liver toxicity, and ocular changes, and such common but lets serious tide effects as hot flashes, vaginal dryness, and irregular menstrual periods (49). Also, there U evidence that tamoxifen may be associated w ith a 94 k flx y and V'huim<<(( HKbAST. ENDOMETRIAL. A M -O V ARIAN CANCERS A P Vof. 4. No 2 Merrh 1904 89-9J le w o f hone mass in premenopausal w o m e n, and (here is conce rn ih a i its effect o n ca rd iova scu lar disease risk in pre menopausal wom en, in w hom endogenous estrogen levels are h ig h , c o u ld be d ifferent fro m its effect in p ostm enopausal w om en, in w h o m endogenous estrogen levels are lo w . F in a lly , a small random ised tr ia l in Los A ngeles (50) is evaluating a prototype h orm on a l contraceptive regimen that contains (I) a gonadotropin-releasing agonist to sup press o v u la tio n , (h) a low-dose estrogen, a nd (c) a sh ort course o f progestogen every fo u rth m o n th , in the hope that this co m b ina tio n of agents w ill decrease the risk for breast, e n d om etrial, and o varian cancers. T h is fo rm u la tio n is based o n the assumptions th a t the " u nopposed estrogen h y p o th e sis" account for e n d o m e tria l cancer risk facto rs, the to ta l ftri|u e iu v o f o vu la tio n is an itn|>ortaiit d eterm inant o f o v a r ian c a m rr risk, and an " estroge n-p lu s-p ro grsiog e n h y p o th e sis" explains breast cancer risk. H o w e ve r, w-ith th e e xception o f the "unopposed estrogen h yp o the sis," the re is co nsid e r able u n ce rta in ty a bout the o th e r e tio lo g ic parhw ays, and also about th e effects o f th is re g im e n o n o th e r b o d y systems, such as hone and the ca rd iova scu lar system. In sum m ary, fo r th e m a jo rity o f w o m e n the re is n o t m uch cause for o p tim is m at present fo r p rim a ry p re v e n tio n o f these cancers by m o d ific a tio n o f th e e n v iro n m e n t o r life style. except fo r w e ig ht loss in p ostm enopausal w om en. P ri m ary p re ven tion s b y p ha rm a co lo gic agents are bein g e va lu ated. h ut such in te rv e n tio n s are risky m view o f th e ir la rgely u n kn ow n mechanisms o f a ction and u n k n o w n effects on o th e r bod y systems. REFERENCES 1 Boring (.`C . Squires T S, T u n a T. < an* f t m .Kim i. s. 1^92, C A C ancer j n . n 1^2:42:10.18 2 K elxy |I.. r J Epidemiology o i hrciM canter. Epidemiol Rev 199); IV I JM ' W h m o n o re AS, I lam * R. Iinyre I. I lalp ctn J. I -ollahoracive O v arian Cancer ( roup C liaracteriM n v relating lo o v arian can t er ri.*V: C o llab o r a to r .-nialv*i* o f iw c ltr U S t a x c o n tro l Mitdic* I. M ethod for analvtm g d.ua from x v rt.il ipiJcm iologi. stodic*. A m J Epidemiol 1 ^2 :1 ft I I 75 ||( U . 4. W hiricm orc AS, H a rm R, Iinyre ). C ollaborative O v arian C ancer Group. C haracteristics relating to ovarian cancer iisk: C ollaborative anaW'i * of **el>,-e U S c a x lo m ro l studies. II In' avive epithelial can err* m while wom en, A m J Epidemiol. |9 9 2 ;lif o S H arris R. W hit remote A S . Iinyre ). c nllal*.it all vc O v arian C ancer G roup t Tiara* leriMi. * tria lin g to <van a n :hh cir n>k: C ollaborative analv'i * nl lu rl *c U S . n*c oiitro ! u n,die* III 1:1<ii licliiil <:iii>t-rs of III.lllMCH-ltSl |M e n u :il in u lu ii **oil >1-1., A m 11 ...............1 1' 2 .1 36. 1-04 1: n f. VIIHire'ni,He a :S, 1l.irn* R . Iinyre 1. C. Hal.. ..n vc O v a ria n (Tan er G rotip C hat.li ter lit li teladiiK to i van a n i .un ci: rn k : C ollaborative a*M\*i * of . el ve U S .S K .o m r o 1 *fuJlCS IV 1h e paih<<gencsis of cpnlirli.il ovarian tancer, Am | Epuitmiol. ln: 1212-1220. 7. John EM. 'X IM emoir A S. I lam * R. Iinyre I. f ol labor at ive O varian Can* rr t loup i a m i r m i i o rel.iiing in o v arian i ancer rij.lt: C ollabo jti\ean alv M fi x * c n l 1S c a v <o titto l ttm lir* V . E pithelial earn er ... H a. k w.niien. I S a ri ( a m er ln*t l*N2.HSN> H 7. 8. Kelsey JL. Horn-RoM I'L. Breast cancer: M agnitude of th e problem an d descriptive epidem iology, Epidem iol Rev. I993-.I5.7-16. 9. Kelsey JL. Breast cancer epidemiology: S um m ary an d future directions. Epidemiol Rev. 199); 15:2*6-263. 10. B o d ian C A . Benign breast d iicaset. carcinom a in situ a n d h rra st cancer risk, Epidemiol Rev. |9 9);|$:I77-187. 11. Petrakis N L. N ipple aspirate fluid in epidemiologic studies of brcsM cancer, Epidem iol Rev. 19 9 );I5 :1 8 8 -195. 12. O ia A M . Boyd N F. M am m ographie parenchym al p atterns: A m arker of breast cancer risk, Epidemiol Rev. |9 9 ) ,|5:196-208. 1). H orn-R oat PL. M ultiple prim ary cancers involving ih r breast, Epide miol Rev 1993;15:169-176. 14. Jo h n EM , Kelsey JL. R adiation an d oth er environm ental exposures an d breast cancer, Epidem iol Rev. 1993;15:157-162. 15. Kelsey JL, G a m m o n M D . Jo h n EM Reproductive factors an d btcasi cancer, Epidemiol Rev. 1993;15:36-47 16. H u n ter DJ. W illett W C . D iet, body sue an d breast cancer. Epidemiol Rev 1993; 15:110-132. 17. M alone KE. D ating JR. Weiss N S. O ral contraceptives an d breast cancer risk. Epidemiol Rev. 1993;15:80-97. 18. B rin io n L A , Schairer C - Estrogen replacem ent therapy and breast cancer risk, Epidemiol Rev. 1993,15:66-79. 19. S tanford JL, T hom as DB. Exogenous progestin* an d breast cancer, Epidemiol Rev. 1993.15:98-107. 20. M alone KE, DicthylsroJIeurol (DES) and breast cancer, Epidemiol Rev. 1993;15:108-109. 21. Roaenberg L, M et get LS, Palmer JR A lcohol consum ption an d risk of breast cancer: A review of the epidemiologic evidence. Epidemiol Rev I99);15;l 33-144. 22. G am m on M D, John EM. Recent etiologic hypotheses concerning breast cancer, Epidemiol Rev. 1993;15:163-168. 23. B ernstein L. Rosa RK. Endogenous horm ones a n J breast cancer risk, Epidemiol Rev. 1993;15:48-65. 24. Pike M C . Spicer D V , D ah m o u th L. Pres* M F. Estrogens, progesto gen*. n orm al breast cell p roliferation, and h restr cancer risk, Epidemiol Rev 199); ISt 17-35. 25. Kcliey JL, H iU retb N G . Beai a n J G ynecologic C ancer. Epidem iol ogy Boca R atun, FL: C R C Tress; 1^8 3. 26. Parkin DM . C ancers o f the breast, endom etrium and ovary: G eo graphic correlarioru, Eur J C ancer. 1989;25:1917-1925. 27. Weiss N S. Srekely D R , E ngliib D R , Schw cid AJ. E ndom etrial cancer in relation to p atterns of m enopausal estrogen uic, JA M A I979;242: 261-264 28 L aV ecchia C . Decarll A . Fasoli M . G entile A. N u trition an d diet in th e etiology o f e n d o m etrial c a n te r. C a n .e r. 1986;57:1248-125). 29. Kcliey JL, LiVolsi V A , H olforJ T R . al. A caie-com rol study of cancer of the endom etrium . Am J Epidemiol. I962;II6:)3)-J42. 30. L aV ecchia C , Franceschi S, Decarli A , G allus G . T ognonl G . Risk factors for endom etrial cancer at different ages. J N atl C ancer lrw . 1984.73:667-671. il . S cbrsselm an JJ. O ra l contraceptives and neoplasia of th e uierine cor pus, C ontraception. 1991;43:557-579. 32. T he W H O Collahorarive study of neoplasia a nd ircioid *ourriw.cptivc*. D epot -m edroxyprogesterone acetate (D M PA ) a m i risk nf endom etrial laiH ei. Ini J C anter. !W I;49.I86 |00 3). Voigt LF. Weiss N S. C h u J. D ating JR. M cKnight R. van Belle G . Progestagen supplem entation of exogenous oestrogen and risk of en dom etrial canter. Lancet. IW I;))8:274-277- )4. W eiu NS. S ay v eu T A . Incidence o f endom etrial cancer in relation to the u k of oral contraceptives. N Engl J Med. I'*80;)02:55I-5S4. 35. K oum antaki Y, T io n o u A . K oum anrakis E, K aklam ani E, A ravanlinos D , Trichopoulos D . A case<onrrol study of cancer o f endome triu m in A th en s, Int J C a n te r. 1989.43:795- TW. A P Vol. 4. N o 2 Mardi 1994 09-95 Kciaev and Whiltcmorc BREAST. ENDOMETRIAL. AND OVARIAN CANCERS 9$ 36. Leako SM , Roaenberg L, K aufman D W , ct iL Endom etrial cancer and age a t laM delivery: Evidence for a n aaaod a(io n . A m J Epidem iol. 1991; U 3:S54-s6o. 37. Key T1 A , Pike M C . T h e doac-effect teU tionahlp br c w i *unoppoaetf* oeurogen* a n d endom etrial m itotic rate: Ira centra) rote In explaining an d predicting endom etrial cancer halt, Be J C an cer. 1960;57:2O5211 36. Kcrlikowake 1C, B row n ] S . O r a d r D O . S h o u ld w o m en w ith familial ovarian cancer undergo prophylactic oophorectom y? O batet Gynecol 1992i80-.700-707. 39. SchikJkraut JM. Thom paon WD. Familial o varian cancer; A popula tion baaed care-control itudy, A m J Epidemiol- 1980;120:456-466. 40. P onder BAJ, E aaion D F , Pcto J. Riak o f o v arian cancer aaaoc iared w ith a family hidory. Preliminary report of th e O PC S atudy. In: Sharp F, M aaon W P, Leake RE, e d i. O v a ria n C ancer: Biologic a n d T herapeutic Challenge. Cam bridge: C hapm an and Hall; 1990:3-6. 41. Hall JM , Lee M K . N ew m an B, er al. Linkage o f early onaer familial bread cancer to chromoaoanc I7q21, Science. 1990,250:1664-1609. 42. Harlow BL. C ram er DW . Bril D A . W kh W R. Perineal expoaure to talc a n d o v arian cancer riak, O b d e t G y n e c o l 1992;00:19-26. 43. C ram er D W , H arlow BL, W illett W C . e t al. G alactoee corw um ption an d m etaboliun in relation to (h e tick o f o v arian cancer. Lancer. 1969; 2:66-71. 44- W hkteroore A S , W u ML, Paflfenbargrr R S, *L Peraonai an d en v iro n m ental ch aracter! lea related to epithelial o v a ria n cancer. D. Expoaure to talcum pow der, tobacco, alcohol a n d coffee. A m J Epidemiol. 1980; 120:1220-124a 45. M eneau J, M odan M , R an o n L, G o la n A . Poaalblc role o f mump* vhua In th e etiology of ovarian cancer. C ancer. 1979;43:1375-1 )79. 46. Farhalla M F. Inceaaant o v u latio n - A factor in o v arian ncoplaaJa? (let ter). Lancer. 1971 ;2:163. 47. G ardner W U. Tumoripeneal In tranaplantcd Irradiated and nonirradiated ovariea. J N a tl C an cer i n d 19 61.26029-054. 40. Stadd BV. T he etiology and prevention of ovarian cancer. A m J O b d et G ynecol 1975;123:772-774. 49. Buah TL. H clrboucr KJ. T am oxifen for th e prim ary p rev ention of b re a d cancer: A rev tew and critique o f th e concept a n d trial, Epidemiol Rev. l99J;15:233-243. 50. Spicer D V , Pike M C. T he preventio n o f b re a d cancer through reduced ovarian ateroid expoaure, A cta O ncol. 1992;31:167-174. 1 .iA ll i H k Dietary Factors and the Risk of Epithelial Ovarian C ancer12 Cvto La V acchl* , 3 A drlano D a c a rtl,4 5 Eva N a g rt,3 8 Fab lo Parazzini, 3 A n lon alla G a n lH a ,3 Qabrtala C a c c h a ttl, 3 M o n ic a F a a o ,3 and S llv la F ra n c a a c h l '* ABSTRACT--D ietary fa cto rs in the e tio lo g y o f ovarian cancer aere investigated w ith the uae o f data fro m a ca se -co n tro l study conducted in N o rth ern Italy o n 4S5 h isto lo g ica lly co nfirm ed aptlhellal carcinom as and 1.385 co n tro l subjects In the hosp ital fo r acuta c o n d itio n s unrela ted to any o f tha kn o w n o r p ote n tia l risk factors lo r cancer o f tha ovary. W om en w ith ovarian cancer reported sig n ific a n tly elevated fre q u e n cy o f c o n su m p tio n o f meat lieleiive risk (RR) * 1. fo r 2 7 vs. < 4 p o rtio n a /w k). ham (RR>=1.9 tor 2 4 va. < 2 p o rtion a/w k). and high e r su b je ctive ecoree o f fat Intake (RR = 2 1 for highest vs low est scores), p a rtic u la rly butter. In contrast, co nsu m p tion s o f hah. g reen vegetables, carrots, and holemeal bread or pesta were less frequent in cases; the co r responding risk estim ates for highest versus low est frequencies ranged between 0.6 a nd 0 6. These results ware n o t m a terially modified b y a djustm ent fo r in d ica to rs o f s o cio e co n o m ic status, parity, and o th e r id e n tifie d d ete rm in a nts o f ovarian cancer. No relation a m erged between a lco h o l co n su m p tio n and ovarian can cer risk. The present stud y provides Inte re sting in d ica tio n s that help to explain the co nsid e ra ble va ria tion s In ovarian cancer rates In d ilfarant p op ulatio n s and. If co nfirm ed , co u ld , in p rin cip le , have Important p u b lic health im plicatio ns. D ue ca utio n, however. Is required in in terpre tin g the present resu lts because of the lim ita tions o f available Info rm a tio n and o f tha u n ce rta intie s o f oth e r p ub lished m aterial conce rn in g d ie t and ovarian cancer --JN C I 1667; 79 663-669 There are considerable variations in in idence and m ortality rates fro m canter of the ovary. O n an in te rn a tional scale, the c u m u lative incidence was sixfold higher in Denm ark than in Japan ( / ) , an d also w ith in Ita ly m ortality rates were about 2.5 tim es h ig h er in N o rth e rn areas than in the South (2, 5). T h e re fo re, the N o rth South gradient in Ita ly was greater for o va rian cancer titan for any other neoplasm in females (2 ). These broad variations can be exp lain ed in p a n in terms of reproduc tive variables; for instance, o v a ria n cancer m o rta lity rates in various Ita lia n regions were strongly and nega tively related w ith the average nu m b e r o f births per woman of reproductive age (r = --0.76) ( ). Strong corre lations, however, were evident also w ith econom ic in d i cators and dietary variables, on an in te rn a tio n a l (5 ) and Italian scale (i.e., --0 .8 1 foi bread, --0.84 for jiasta, + 0 .6 2 for m eal, and + 0 .7 9 for m ilk co n su m p tio n ), w h ic h could not be to m p le ie ly accounted for by c o n fo u n d in g (or reproductive vat tables after co m p u la tio n of firstorder p artia l correlation coefficients (4). Analysis of geographical variations in cancer rates or correlational studies can only provide suggestive hy potheses on u n d erlyin g causes, w h ic h sho uld be th o r oughly tested in ad hoc studies. A few case-control studies conduc ted in N o rth A m erica reported that a n i mal fat con sum ption increased the risk (6 ), whereas beta carotene (7) or elevated a lco h o l (8) con sum ption m ig h t provide some protection against o v a ria n catc b io genesis. H o w ev er, ava ilab le evidence is rather con trover sial. and the results from the dataset analyzed m ore in detail (from the Roswell Park M em orial Institute) were largely inconclusive (7). Using data from a case-control study conducted in N o rthern Ita ly , we decided, therefore, to reconsider the p o te n tia l role o f a few selected dietary item s o r in d ic a tors in the etiology of ep ith elial ovarian cancer. SUBJECTS AND METHODS Since 1979, we have been conductin g a case-control study o f ovarian cancer. T h e general design and m eth odology of the study have been previously described (V). Train ed interviewers identified and questioned w om en adm itted for ovarian cancer and for a w ide spectrum of other conditions to university and general hospitals in the areas under surveillance. O n the average, less than 2% o f the e lig ib le w om en (cases and con trols) refused to be in te rv iew ed A standard q u e stio n n aire was used to o b ta in in fo rm a tion on sociodemographic characteristics and habits; medical histories w ith special emphasis on horm onal and gynecologic features; obstetric data, in clu d in g age at menarche and at menopause and menstrual pattern; num ber o f b irths, miscarriages, and induced abortions, age at first pregnancy, at first and last b irth ; history of A U K Z v ia t io n s u s e d : C l =c o n fid e n c e in te rv a l; R R = re la tiv e r iik . ` Received J a n u a ry 7. 1967, revived M ay H . 1967; ac<e|>ted M ay I V 1987 * T h is w o rk waa conducted w ith in the fram ew ork of the Ita lia n N a tio n a l Reaeatch C o u n c il A p p lie d Project "O n c o lo g y " (contract N o. 65.02209.41) a n d "P re v e n tiv e and R e h a b ilita tiv e M e d ic in e " (c o n tra c u N o. 85.00487.56 a nd 85.00549.56) * la titu to d i R iceiche Farm acolngithe "M a rio N e g ri," V ia Eritrea, 6 2 -2 0 I5 7 M ila n o , Italy. ` lit it u io d i B io m e u ia r Slat i n ic a M edica. U n i ver n i d i M ila n o e la titu to N a iio n a le T u m o ri, V ia Venenan. I. 20IS3 M ila n o . Italy. J la titu to d i S u tiitic a e Ricerca O p e ia liva , U n iv m itk d i T re n to . 38100 T re n to , Ita ly . ` C o n s o riio I n te r u n iv e n ita r io lo m b a d o p e r r E la b o ra r ocie A u to m tica. 20090 Segrate, M ila n o . Italy. TC enuo d i R ile rim rn to O ncolgico, 33081 A viano. Purdenone. Ita ly . * T h e c o n trib u tio n o f the Ita lia n League againat T u m o r*, M ila n . Ita ly , ia g ra te fu lly acknow ledged. Data processing and analyaia were perform ed by meana o f the c o m p u tin g fa cilitie a o f the In te r-U n ive rsity C o n so rtiu m o f L o m bardy (or A u to m a tic Data Processing (O L E A ). We th a n k M i. J u d y B aggott, M t. F la v ia R o n ia td t, and the O. A P ie iflr r M em orial L ib ra ry n a ff (or e d ito ria l aiaiitance 663 \ J N C I . V O L . 79. N O . 1. <X:i O B F R 19*7 664 La Vecchla, D e carll, N e gri, et ai. T a h I.K \ . --D is trib u tio n o f 455 o im * o f e p ith e lia l a r a r ia n ca n cer a n d 1.885 contrle according to aye. M ila n . Italy. 1988-86 Age group, yr Ovarian cancer No. Percent Controls No. Percent ' :to :tii 39 40 40 fo 59 BO 69 711 74 14 3.1 f3 4 5 25 5 5 182 13 1 104 22.9 271 19.6 154 33.8 347 25 1 I I I 24 4 353 2 5 5 47 10.3 169 12 2 lifetime iim* <( oial cmiriaerptivcs ami female hoi inoncs Im eiihe-t indieaiioiiv i n 111 Jaiiuaiv 1983. tin* subjre is writ* also asked about i In*ii f i y of M iiisuinplioii | h -i wrrlc of ten selected loexl items, 11u bid ini' ib r majoi sources ill the Italian iliei of iciuioitts (m ilk ami dairy pioclucis. meat, fish, Iivet. and eggs), and caiolenoids (cauots. green vegeta bles. and fiu ii). Subjec live stores (low. inieimediate, and high} weie used to obtain in fo iu ia lio il on fat intake (b unu. muigatiiie, and oil) and on tniiM im plicm of wholemeal bread or pasta. Thus ihe present lepoit is based on data collected between January 1983 and June 1986. C a s e s .--The <uses studied weie women below the age of 75 with histologically confirmed epithelial ovarian cancel diagnosed w ithin the year preceding the inter view. who bad been admitted to the Obstetiics and Gynecology Clinics of the University, the National (winter Insiituie. and the Osjx-dalr Maggiore of M ilan (which nullities the foul iaigest leaching and general hospitals in the m e riojxdiijn aiea). A total of 455 cases, aged 22-74 (median age = 55 yr) weie included Among them. 81% had scions care inoinas; 3%, endometrioid: 5%. mucinous: 3%, clear cell; and 5%, poorly differentiated epithelial carcinomas. C o n t r o l s . -- The controls were women admitted for ac me conditions to several sjiet iali/ed University Clinics and to the Ospedalr Maggioie of M ilan for diseases oihci than malignant, hormonal, or gynecologic, diag nosed w ith in the yeai befoie the- inte rview. The <*tuhmem aiea of cases and contiols was well coinfrarabk, since over 85% of the subjec is iniei viewed tesided in thr same region--I.oinhaidy Afiei exclusion of women who had undergone oophorectomy, a total of 1.385 control}, aged 21-74, wereconsidcied in the piesent analyses. Thfil median age was 54 years, and iheir age distribution was well compaiable will dial of the cases (table I). Among die coniiol subjects. 31% were admitted (or traumatic conditions (mostly ftacitites and spiains). 25% had non(rautnalic ortho|>edic disorders (mostly low back pain and disk disorders), 16% were admitted for acute sutgicai conditions (mostly abdominal, such as acute appendici tis or strangulated hernia), and 28% had other illness, such as ear. nose, and ilitoat or dental disorders. D a ta analysis. --We computed the mean ftecjuenc ies of consumption of available dietary items among cases and controls, logeihei w ith their standard errors. For the food items showing significant differences between cases and contiols, we calculated the RRs of ovarian cancer, together w ith their 95% apptoxiinate CIs, according to three levels of consumption itu hiding, as far as jxissible, comparable numbers of cases and contiols combined (i.e., approximate let tiles). An index of fat intake was simply derived by adding the scores (low -- I; intermediate = 2; high = 3) of self-reported intake of the three major types of dietary fat (butter, margarine, and oil). Indices of retinol and beta-caroienr (pro vitamin A) intake were computed w ith the use of tables of nutrient values of the Italian Department of Agi icultm e ( 0 ) or. in tire absence of Italian data, standard American tables ( //) . These indices should clearly be interpreted as rela tive estimates of vitam in A intake, since the limited die tary inform ation available did not cover all the possible sourc es of vitam in A In the compulation of RRs, decade of age was con(tolled lot by stiatification and the Mantel-1laenuel pi ex edure (1 2 ). For m u ltip le levels of rx|>osure, signifi cance was assessed by means of the linear trend test given by Mantel (/?). Ter allow simultaneously for the effects of several potential confounding factors, wc used T a M.R 2 . - / 4 rerage freq u en c ies o f in ta k e o f selected fo o d item s a m o n g 455 cases o f e p ith e lia l oca r ia n ca n c er a n d 1.885 co n ln d s, M ilan, Italy. 1983-86 Food item Average frequency of intake, portions per week Cases C o n tro ls D iffe re n ce Cases - controls SR P value4 M ilk Meat L iv e r Carrots Green vegetables Fresh fru it Eggs Ham Pish Cheese 60 5.4 0.6 1.2 7.8 12 1 1.8 2.5 09 4.8 6.2 50 06 1.5 8.5 119 20 2.3 11 51 -0 .2 + 0.4 -0 .1 -0 .3 -0 7 + 0.2 -0 .2 +0.2 -0 .2 -0 .3 032 0.13 0.06 0.10 0.22 0.78 0.13 0.10 0.09 0.16 NS <001 NS <001 <001 NS NS <06 <06 NS * NS = not sig nifican t. j N< |. VC>1 Vi. M -l.lM . il Ht H l*M7 m u ltip le lo gistic, regression, w ith m a x im u m lik e lih o o d liltin g {1 4 ). In c lu d e d in th e re g re s s io n e q u a tio n s , besides the d ie ta ry fa c to rs c o n s id e re d , w e re te rm s fo r a ge ( in c a rdin aI fo r m ) , in te r v ie w e r, m a r it a l s ta in s , s o c ia l cla ss, education, p a rity , age at lirs t b irth , age at m e n a rc h e , m enopausal statu s, age at m e n op au se , body m ass in d e x , and o ra l c o n tra c e p tiv e a n d o th e r fe m a le h o rm o n e use. Dlt and Ovarian Cancar 66 RE8ULT8 T h e average w e e kly fre qu e ncie s o f c o n s u m p tio n o f a v a ila b le d ie ta ry ite m s are re p o rte d in ta b le 2. C o n* s u m p tio n o f m ilk , cheese, liv e r, eggs, a n d fresh fru it d id n o t s ig n ific a n tly d iffe r (o r cases a nd c o n tto ls b u t a l th o u g h d iffe re n c e s in a b s o lu te te rm s w e re s m a ll (e .g ., 5.4 TABLE 3.-- L H strib u tio n o f 655 ca te* o f ep ith e lia l o v a r ia n c a n c e r a n d 1.S86 co n tro l< a cc o rd in g to fr e q u e n c ie t o f c o n s u m p tio n o f *elected food Hem*, M ilan. Italy, 1983-88 Pood item and portions per week Ovarian cancer No. (%) C o n tro ls * No. ( v RR e stim ates (95% C l) M H> M u ltiva ria te' Meat <4 114 (26.1) 466 (32.9) 1' 4-6 131 (28.8) 409 (29 6) 1.30 (0.98-1.73) >7 210 (46.2) 621 (37 6) 1.63 (1.26-2.12) i f (trend) 12.93 (P C 0 0 1 ) Ham <2 136 (29 9) 487 (36.2) I' 2-3 227 (49.9) 706 (6 1 0 ) 1.17 (0.91-1.49) >4 92 (20 2) 192 (13 9) 1.85 (1.36-2.68) if(tre n d ) 12.00 (PC.001) Pish <1 220 (48.4) 691 (42.7) i' 1 162 (36.6) 496 (36.8) 0.87 (0.69-1.10) >2 73 (16.0) 298 (21.6) 0.66 (0.43-0.90) i f (trend) 6.68 (P = .0 1 ) Carrots <1 197 (43.3) 667 (40.2) 14 1 129 (28.4) 319 (23.0) 1 11 (0.86-1.45) >2 129 (28.4) 608 (36.7) 0.70 (0.66-0.91) i f (trend) 6.41 (P = .0 1 ) Green v e g e ta b le s < 7 126 (27.6) 342 (2 4 7 ) 7 185 (40.7) 419 (30.3) 1.18 (0.90-1.66) >8 146 (31.9) 624 (46.1) 0.69 (0.46-0.78) X? (tre n d) 17.01 (PC.001) W holegrain bre a d o r pasta Never 336 (73.8) 64 (69.1) 1* O ccasion ally 82 (18.0) 268 (18.7) 0.88 (0.66-1.16) F re q u e n tly 37 (8.1) 169 (12.2) 0.60 (0.41-0.88) X? ( t r e n d ) 6.62 <P=.01> ll a. ld 1.30 (0.96-1.86) 1.60 (1.21 2.12) 16.76 (P C .O O l) i1.13 (087-1.46) 1.66 (1.11-2.16) 4.67 (P =03) I' 0.94 (078-1.21) 0.84 (0.61-1.16) 0.69 (N S ) 14 1.19 (0.90-1.68) 0.81 (0 62-1.06) 2.29 (N S ) 1' 1.10 (0.83-1.47) 061 (0 48-0.82) 13.99 (P C .O O l) 1' 0.96 (0.71-1.77) 0.63 (0.42-0.93) 4.46 * In some item s, the sum o f the s tra ta does not add up to th e to ta l because o f m issing values. * Mantel-Haensxel estimates adjusted fo r age in decades. ' Estim ates fro m m u ltip le lo g istic regression equations in c lu d in g te rm s fo r age. In te rv ie w e r, m a rita l status, social class, education, p a rity , age at fir s t b irth , age at m enarche, m enopausal status, age a t menopause, body mass index, o ra l contraceptive and o th e r fem ale horm one use, retinol and carotene indices, added score o f fa t co nsum ption and alcohol in take , besides d ie food item s considered in th is tab le . NS = not sig nificant. * Reference category. J N C I. VOL. 79. NO i . O C IO H V X 19*7 . n> ... a . 666 La Vacchla, Dacarll, Negri, el al. vs. 5 tim e's | k t w eek fo r m eat <o n s u m p tio n ) , w o m e n w iili o v a iia n cancel ie |> m ied m o re fre q u e n t c o n s u m p - i i 11 o f m e a l a n d f la m a n d less f r e q u e n t c o n s u m p t io n o f fis h , t a i i o i s . a n d i4ic**n ve g e ta b le s. 1 lie i (iiie s |N > n d iitK R R e s tim a te s a ie s h o w n in ta b le S. I lie |H )iill e sim a tes fo i h ig h e s t v ris u s least fre q u e n t c o n s u m p t io n le v e ls w e ie 1.63 lo t m e a l. 1.85 f o i h a m , 0 (iti fo i fis h . 0.70 fo i ( a llo ts , a n d 0 .5 9 fo r g re en ve ge ta bles, a n d a ll tile a g e -a d ju ste d tests f o i lin e a r tre n d in ris k w eie s ta tis tic a lly s ig n ific a n t. M o te o v e i, fre<|uent c o n s u m p tio n o l w h o le -g ra in h iea d o i pasta was associated w ith s ig n ific a n tly ic d in c d o v a ria n (a n c e i tis k (R R -- OtiO. 95% C l = 0 .1 1 -0 .8 8 , ta b le 2). A llo w a n c e lo t id e n tifie d |x> ie m ia l d ts io itin g factors, as we l l as iec ip io al a llo w a n c e f o r v a iio u s d ie t a r y ite m s in c lu d e d i l l ta b le .1. d id n o t m a t c iia ily m o d if y th e e s t i m a ted R R s io i m eat, g ie e u vegetables, a n d w h o le g ia in lo o d s . r i i c a s s o c ia tio n w it h h a m w as less e v id e n t, a n d associa tio ns w ith fis h a n d c a rto ts w e ie n o lo n g e r s ig n ific a n t aftei im ih iv a i iate analysis. Ind ice s o l m o n th ly in ta k e o f le tin o l (p ie lo n n e d v ita m in A ) a n d heta-c a to te n e were c n in p u te d o n tin - basis o f th e fir c |iir iic ies o f c o n s u m p tio n o f fo o d ite m s lis te d in table 2 (m ilk . m eal. Ii\e t, eggs, fis h , a nd cheese fo r te im o l; ca llo ts , g ire n vegetables, a n d Irt-sh f r u it fo r c a ro tene). In th e u n iv a iia te a na lysis, n o te la tio n em erged betw een in d ice s of ic iin o l in ta k e a nd o v a iia n cancer iis k , fu n a s ig n ific a n t im e is e assm ia tio n was rv id e n t w ith heta-c a to te n e in ta k e (io m p a ie d w it h w o m e n w h o se c a io te n e in d e x w a s n n d r i 150,1)00 f U jk -i m o n t h , the* iis k e stim a te was 0.62 Io i those w hose in d e x was o ver 200,000 l ( *, a n d th e iie n d in ris k w as h ig h ly s ig n i f i c a n t (ta b le 4). H ow ever, a fie r m u ltiv a iia ie ana lysis w ith in c lu s io n o f d ie ta ry fa cto rs js w e ll as o ilie i m a jo r ic le iiiifie d co var- iates. ilie in v e is e re la tio n o f o v a ria n ca nce r ris k w ith beta-carotene was n o lo n g e r e vide n t. In ta b le 5. subjec tiv e scotes o f fa t in ta k e in cases and c o n tro ls a rr reported. F lrv a ie d b u iir i c o n s u m p tio n w u a s s o c ia te d w it h a b o u t t w ic e th e o v a r ia n c a n c e r risk ( R R = 2 . 3 u n iv a r ia te . 1.9 m u ltiv a r ia te ) . T h e la ig c m a jo r ity o f th e su bje cts in te rv ie w e d re|>orted lo w scores ol m a rg a rin e c o n s u m p tio n ; c o n s e q u e n tly , lit t le relation w as e v id e n t b etw e en m a rg a rin e in ta k e a n d ovarian cancer ris k . N o a sso c ia tio n w as observed w ith o il con s u m p tio n e ith e r, even th o u g h o il was ih r fa t ite m with h ig h e s t in t a k e sco re s. C o n s e q u e n tly , th e p o s it iv e asso c ia tio n b etw e en added fa t c o n s u m p tio n score was chiefly a ttrib u ta b le to th e in flu e n c e o f b u tle r o n th e overall in d e x . T h e m u lt iv a r ia t e R R s p re s e n te d in t a b le 5 were d e iiv e d fro m tw o d iffe re n t m o d e ls , o n e in c lu d in g ihr th re e s in g le fa t ite m s a n d i h r o ilie r inc lu d in g th e added in d e x o n ly . T h e d is t r ib u t io n o f o v a r ia n canc er cases a n d th e corn p a iis o n g ro u p a c c o rd in g to to ta l a lc o h o lic le ve ra g e con s u m p tio n (w ine-, here, a n d s p iiiis ) is g iv e n in ta b le d C o m p a r e d to w o m e n w h o d id n o t d r i n k a lc o h o l, the R R s w e re s lig h t l y a b o v e u n i t y f o r d r in k e r s , b u t there w a s n o te n d e n c y fo r (he ris k to u n e a s e w it h higher a lc o h o l c o n s u m p tio n . DISCUSSION f ile fin d in g s o f th e p re sen i s tu d y in d ic a te th a t the din o f I t a lia n w o m e n w iih e p it h e lia l o v a iia n ca n c e r differ in several a q x 'c is fr o m th a t o f a h o s p ita l-b a s e d c o m p a ri son g ro u p o f su bje cts w ith a cute n o n n e o p la s tic condi tio n s . In p a r tic u la r , m o re fre q u e n t m e a t a n d h a m con s u m p t i o n a n d h ig h e r fa t in t a k e ( i n p a r t i c u la r , b u t t e r ) in T aim.k 4. Ih x tr ib u tiim u / .VS oist-s o f r fiilh flia l o i'a r ia n ca n c er a n d i..*8 5 co n tro lit a cc o rd in g tv cut m a ted M o n t h l y in ta k e o f retinol a nd beta-carotene. M ila n . Ita ly , I9H3-M6 Index Ovarian cancer No (%) C o n tro ls No. (%) RR estim ates (95% C l) M H- M u ltiva ria te 6 Relm or <r 1no to n -1 Mi > 1Ml ci (trend) Carotene' - IMI [M l 2cio 200 (tre n d ) 149 M2.7) i: u (29.5) 470 341 172 (37.8) 574 215 (47 3> 508 |C19 (24 0) 377 Ml 128 Ml MM) (33.9) (2 4 .6 ) (41 4) (3C.7) (27.2) CM>. 1 l4 1.23 (0.94-1.61) 0.96 10.75-1.24) 0 18 (N S ) 14 068 (0 52 0.88) 0 62 10 48 -11.80) 14.73 < /'< .0 0 l) l4 1.18 (0 89-1.67) 094 (0.72 1.22) 0.14 (N S ) 10.76 <0 53 1.10) 094 (0 66 -1.61) 0.17 (N S ) " Mann-I llacnftzt-l estim ate adjusted for age in decades. K s lim a te s fro m m u ltip le logistic regression. A llo w a n c e w as m ade fo r id e n tifie d |>olenlial d is to rtin g factors (*e r " S u b je c t* and Method' and footnote r. tab le It). ' In thousand international units rf Reference category. I N" u " ;,` <>< i n lin e l'W7 D M and Ovarian Cancar M 7 T able 6 --D istrib u tio n o f 155 caeca o f e p ith e lia l o v a ria n ca n cer amd 1,385 con tro le a cco rd in g to a b jective corea o f fo i in ta ke, _______________________________ .___ M ila n , Ita ly . 1983-84 ________________________ _____________ consumption score Ovarian cancer No. <%> Controla No. <*> RR estimates (96% C l) M H* M ultivariate* Butter Low Intermediate High 241 (63.0) 179 (39.3) 36 (7.7) X? (trend) Margarine Low 370 (81.3) In te rm e d ia te -h ig h 86 (18.7) Oil Low Intermediate 90 (198) 291 (M 0 ) High 74 (16.3) Xi (trend) CombinedLow Intermediate 198 (43.6) 160 (33.0) High 107 (236) X* (trend) 879 449 67 1,167 228 221 1,002 162 738 469 188 (63.6) (32.4) (4.1) (83.6) (16.5) (16.0) (72.8) (11-7) (63.3) (33.1) (13.6) 1* 1.47 (1.17-1.84) 227 (1.47-3.61) 20.52 (/ C .O O l) 1* 1.16 (0.88-1.62) r 0.73 (0.66-0.96) 1.17 (0.80-1.69) 0.21 (NS) 1* 1.22 (0.96-1.66) 2.12 (1.60-2.82) 28.78 (/*< .001) 1' 1.41 (1.11-1.80) 1.93 (1.20-3.11) 12.61 (/><00l> r l.n (0.83-1.49) 1* 0.71 (0.66-0.96) 1.29 (0.87-1.92) 0.64 (NS) 1` 1.16 (0.90-1.48) 2.14 (1.69-2.88) 18.63 (#<.001) ` Mante)-H senate! estimates adjusted for age in decadea. ' Estimates for m ultiple logistic regression. Allowance was mads for major identified potential distorting factors. (Sss " Subjects and Methods" and footnote c. table 3; two aquations were fitted, one including various types of fat and the other the combined score only.) * Reference category. ` Computed as the simple sum of scores (1-lo w ; 2 = intermediate; 8-h ig h ) o f the various fat items and subdivided as follows: 8-4=*low; 6=intermediate; 6-9=high. seasoning were associated w ith elevated o va rian cancer risk, whereas m ore frequent vegetable con sum ption showed a protective effect. We were unable to exp lain these find ing s in terms of selection bias, since cases and controls were d raw n from similar catchm ent areas, and the p a rtic ip a tio n rate was practically complete. W ith regard to confounding, a l lowance for several identified potential distorting factors, in c lu d in g measures o f body w eight and indicators of socioeconomic status, p arity , oral contraceptive use, and other know n determ inants of ovarian cancer risk failed to e x p la in the observed associations w ith various dietary Table 6.--Lh a trib u lio n o f 155 catea o f c jn tk e iio l o v a ria n ca n cer an d 1,385 co n tro la a cco rd in g to to ta l a lco h o lic beverage con su m p tion , M ila n , Ita ly , 1983-84 ___________ umplion. drinks per day Ovarian cancer No. <*> Controls No. <%) RR estimate (96% C l) M -H* M ultivariate* 0 <2 2-3 >8 x f (trend) 134 (29.6) 606 (36.5) 162 (33.4) 391 (28.2) 138 (30.3) 397 (28.7) 81 (6.8) 91 (86) 1* 1.46 (1.11-1.89) 1.31 (1.00-1.72) 1.30 (0.82-2.04) 8.44 (NS) 1* 1.24 (0.94-1.66) 1.06 (0.80-1.40) 1.32 (0.78-2.37) 0.40 (NS) *Manlel-H senate! estimates adjusted for age in decade. ' ` Estimates from m ultiple logistic regression. Allowance was mad for Idantiflad potential distorting factors (sm "Subjects and Methods" | sad footnote c, table 3). 'Reference category. I NCI. V O I. 79. NO. i . OCTOBER I9S7 38 La Vecchia, Oecaril, Negri, ei al. i. id o is . n a m e ly m e a t, g ie r ii vegetables, w h o le g ra in fo o d s . a n d Ih i i u i . I .ikrw i.se, a m a jo r im p a c t o f in fo r in a lio h h i,is is u n lik e ly , .sim e ih e (rossd ile ir la lio u o f d ie t w ith o s ,u ia n m ik c i risk was p to h a b ly u n k n o w n to the iiia fn iity o f th e subjects in ir i v ic w rtl. I In i('s u its o l th is s tu d y a rc in a g ie e m e n i w ith e v i dence lim n d e s c rip tiv e e p id e m io lo g y s h o w in g , o n b o th an in te r n a tio n a l (9 ) a n d an Ita lia n n a tio n a l (d ) scale, s iltin g jM i.siiive c o n e la lic m s betw e en m e a sure s o f fat in ta k e a n d o v a iia n cancel m o ita lity . In th is sense, they p io v id e u s e fu l <lues to e x p la in re g io n a l a n d in te r n a tio n a l d iffe re n c e s in o v a ria n cancer inc id e n ce a n d m o ita lity ( / - I ) . U re y a ls o c o n fin n th e fin d in g s o f a s tu d y co n d u c ted in N o t i l l A m e i ic a a n d u t i l i z i n g p o p u l a t i o n c o n t r o ls (6 ), w h ic h show ed s ig n ific a n tly liig h e i o v a iia n cancer ris k in w o m e n re p o ttin g m o re fie c p ie itt c o n s u m p tio n o f b u i ld . a nd an a p p ro x im a te ly tw o fo ld e le va te d ris k e s tim a te in the h ig h e s t c ju a itile o f a n im a l (at c o n s u m p tio n . I lie ne g a t iv e l e l a t i o i i w i t h v e g e ta b le s a n d w h o l e g r a in file .u l <ii p a s ta is a ls o o f p o te n tia l iiite re .s l in re la tio n to d e fin in g a lo w -iis k d ie t fo i o v a iia n cancer in c o n tra s t w ith a h ig h tis k o n e chatac le ri/e d by e leva ted m e a l a n d a n im a l h it in ta k e . In th is rc'g a id, data c o lle c te d in a n Ita lia n p o p u la tio n m a y he p a iiit u la r ly h e lp fu l in stud ie s o f d ie t a n d ca nce r, o n a c c o u n t o f th e c o n s ic ie ia b le h e te ro g e n e ity o f these co ni|N M ie m s i ll th e Ita lia n d iet. A p la u s ib le b io lo g ic e x p la n a tio n fo i th e present fin d in g s can he te la ie d to th e lo w e r u iin a iy le vels o f e s tro gens a n d h ig h e r p la s m a s e x -h o n n o iie b in d in g g lo b u lin le v e ls in v e g e ta ria n w o m e n (7 5 ), s in c e fe m a le h o rm o n e s a n d th e ii (u n a v a ila b ility a ie p io h a b ly re la ted to o v a ria n c ate iitn g e u c s is , a lt h o u g h at p re s e n t it is d i f f i c u l t to d e lin e c le a rly the to le o f v a iio tis h o rm o n e s . F u rth e r, stud ie s in ro d en ts sh ow e d p o s itiv e re la tio n s b etw e en h ig h -lu i d ie ts a nd p io m o tio ii fo i a n o th e r fe m a le h o rin o n e ie la te d n e o p la s m , i.e ., hieast ca n ce r, a lth o u g h the le la lio n w as n o t h o m o g e n e o u s fo r v a rio u s ty|>es o f fats, o liv e o il ( w h it It is th e m o st c o m m o n ly used o il in Ita ly ) , a n d o th e r m o iio im s a im a ie d lats in p a n ic tila i s h o w in g lit t le p i o m o t i n g e ffe c t ( h S). T h is e v id e n c e f r o m a n im a l siuclies s h o u ld lie c a ie lu lly c o iis id e ie d in lig h t o f the seveial s iin ila ittie s in th e e p id e m io lo g y o f b ic a s i a n d o v a iia n c a n c e l (V, 17). T h e p irs e n t e p id e m io lo g ic fin d in g s , if accurate, w o u ld have iin p o iia m p u b lic h e a lth im p lic a tio n s , sin ce the d ie t a iy lac t o t s a n a ly z e d a ie , a t le ast i n p r i n c i p l e , m o d i f i a b le . a n d th e m a g n itu d e o l the R R s in th is s tu d y is s iin ila i to tha t o f the best id e n tifie d re p ro d u c tiv e c o rre lates o f ov at ra il c aic b io g e n e s is (9 ). w h ic h a ie . o b v io u s ly , m a ir c iv a m enable to in te rv e n tio n . T h e re su lts o l th is s tu d y , h o w e v e i. s h o u ld be c o n sid e ie d w ith c a u tio n , since a v a ila b le in fo r m a tio n was lim ite d to m e n ly s u b je c tiv e scenes o f fa ts o t w h o le g ra in fo o d in ta k e a nd to the n u m b e r of (to rtio u s p e r w eek o f a re s t11 le d lis t o f d i c i a i y ite m s , w i t h n o a s s e s s m e n t o f ilo ir size. C o n s e c p iriu ly . m a jo r u n c e r ta in tie s a re o b v i o u s ly m h e ie iii in a n y e v a lu a iio n o f s|x*cific in ic ro m iii in n s ilS , 9 ), a nd . p a n ic u la iiy . in d ie n e g a tiv e le la i t io n b e tw e e n in d ic es o f b e ta -c a ro te n e a n d o v a r ia n cancel r is k . F u r t h e r , t h e o b s e r v a t io n t h a t in c l u s i o n o f .specific fo o d ite m s (i.e .. g re e n ve g e ta b le s, c a rro ts , a n d fruit*) s u b s ta n tia lly reduced th e re la tio n b etw een b rta -ca io irn r a n d o v a r ia n c a n c e r r is k s u g g e s ts t h a t l i t t l e r e s id u a l effect f o r t h i s in d e x w a s e v id e n t b e s id e s t h a t a t t r i b u t a b le to iu v a r io u s c o m (K n e n ts . I t is t h u s p o s s ib le t h a t o t h e r com p o n e n t s o f a v e g e ta b le - r ic h d ie t a re in d e e d p r o te c tiv e in th e a b s e n c e o f a n y s |K *(ific p r o te c tio n c o n v e y e d b y lieu ca ro te n e o n o v a ria n ca rc in o g e n e s is . F u r t h e r c a u t io n in d r a w in g in fe r e n c e s f r o m th e find m g s o f t h e p re s e n t i n v e s t ig a t i o n is in d ic a t e d b y a stud) o f d ie ta r y fa c to rs a n d o v a r ia n c a n c e r base d o n th e Rur w e ll P a r k M e m o r ia l I n s t i t u t e d a ta s e t. T h e re s u lts o i that s t u d y , in fa c t, w e re la r g e l y in c o n c lu s iv e in r e la t io n to la ts , v e g e ta b le s , f ib e r s , a n d a ls o v i t a m i n A , w i t h th e sob e x c e p tio n o f a n e g a tiv e a s s o c ia tio n w it h beta-caioiew th a t, h o w e v e i, w as re s tric te d to th e s u b g ro u p o f w o iiki b e lo w age 50 (7). h i i b i s s t u d y , t h e r e w a s n o r e l a t i o n b e tw e e n alcohol c o n s u m p t io n a n d o v a r ia n c a n c e r r i s k . T h u s th e se find m g s d o n o t c o n f i r m t h e s u g g e s t io n o f a lo w e r ovarian ca n c e r ris k in h e a v y d rin k e r s re fx m c d in th e C enters la D is e a s e C o n t r o l s ( l a n c e r a n d S te r o id H o r m o n e Studi a n d te n ta riv e ly e x p la in e d in te rm s o l dep re ssed levelsd g o n a d o t r o p in s a n d h e n c e a n o v u l a t io n in d u c e d b y larp doses o f a lc o h o l (8). I n c o n c lu s io n , a l t h o u g h t h e p re s e n t s t u d y h a s suk s t a n t ia l l i m i t a t i o n s , c h ie f ly in te n n s o f t h e restricted a m o u n t o f in f o r m a t io n a v a ila b le , it o ffe rs interesiinf i n d i c a t io n s t h a t h e lp t o e x p l a i n p e c u lia r fe a tu r e s o f dr d e s c r ip tiv e e p id e m io lo g y o f o v a r ia n c a n c e r, a n d , if cot f ir m e d , c o u ld h a v e i m p o r t a n t p r a c t i c a l p u b l i c heald im p lic a tio n s . REFERENCES (7) D u l l. R . P f i o R I h r t auses o f c a n trr: Q u a n iic a iiv r rsiimam avoidable in k s o f c u iicn in i tic- tin n e d Siair unlay. JNCI IV 66:1191- I SUM (2) M i -Z/a n o i i i G . C im ac.mi c:, O il-AftM A. r i ul C anter mor in Inoad Ita lia n g n iK u it liu u l areas. 1975-1977. T u in u ii 1 72:1 - 1 52 ( I ) Cim a i.i l l C. D m .ski i A. 1 a V m .i mia C. ei al. D an. induaion ntap|Ka di n io iiu litS lu in o ta lc . (Data. iaiiiit and in*|a a n t1 1 m o ria liiy .) lia lia 1975/1977. Bologna. frditrice. 1986. (4 ) D m .a k ii A. I .a VlC i.lH A tliv iio n m e n ta l f u o o it and ( m o iia liiy in Italy. (- o iirl.tiio n a l t*xc*ici\r. Oncology I 116-126 ( i ) A km m ko n c . B. D o n . R k n v iiiiiiin r n ia l la ito i and cancel i d e m r and m o ria liiy in d iiic r r n i to m m ie s. w ith speoul enre in dietary pi unite In i J Cancer 1975; 15:617-631. (6) C r a m ir DW . W li.C il W R . H o l c h im in G B . e ia l. Dietary aiu fai in relation to ovaiian canter iik O lis iri Gynecol 63:833-H38 (7) Bu r s I . M a k m ia i i J. G raham S. ei al. A <ae-tom rol siutfi dleiary and n o n d iria iy fat lor in o vaiian ta u te r. JNCI I 71:681 >686 {8) Cw inn M l.. W i r s i ir I .A. I.t.t NC. ei al. A lc o h o l conjum " and o va iia n t.am ei risk. A m J tp id e m io l 1986; 123:759-766 (V) frRANt.iM .iii S. I .a V ic c iiia C. M m .m ricji SP. ci al. R is k ' for epithelial ovaiian la n tr i in lia ly Am J Kpidemiol I 115:714-719. (W ) CuM>|>eraiiva N iio vj A lin it'iiia z io o r. ABC |>ei I ' Ktiutaiione I NC I \ l It 7*1. NO 1 inline I'ih7 nte n ia re . M a n u a lc . M ila n o : C I.E S A V , 1985. (//) U S. D e p a rtm e n t o f A g m u liu r e A g ric u ltu r e Research Service. C o m p o s itio n o f foods: Raw . piocessed, p re p a re d W a sh in g to n , D C : U S. G o v t P n m O ff. 1972. |/2) M a n im N , H a i n u i i . W . S ta tistica l aspects o f the analysis o f data fro m retros|>eaive studies o f disease. J N a tl C ancer Inst 1959; 22:719-748. ( / / ) M a n i m . N . C h i-s q u a re tests w it h o n e degree o f fre e d o m : E x te n sion o f the M a n te l I f a r n u r l procedure. J A m Slat Assoc 196); 58:690-700 (14) Ba s is R J. N u .ocr JA . T h e C I . I M system . Release 5. O x fo r d : N u m e ric a l A lg o r ith m s C r o u p , 1978. |M ) A r m str o n g B K . B r o w n JB. Cla r k e H T , et a l. D ie t and repco- D M and Ovarian Cancar 669 du ctive horm ones: A study o f vegetarian a n d n onvegetauan p o s tm e n o p a u s a l w o m e n . j N C I 1981; 67:761-767. ( 16) Co k in L A , T hompson D O . M aeura Y . et al. Dietary fat and m am m a ry cancer. I. P ro m o tin g effet it o f d iffe re n t d ie ta ry fats on N -nitrosom cihyiurea-induced tat m am m ary tum urigenesis. JN C 1 1986; 77:55-42. (17) K elsey J L . A review of the epidem iology o f h u m a n breast cancer. E p id e m io l Rev 1979; 1:74-109 ( / i ) C h u SY, K o io n e l L N . H a n k in J H . et al. A co m p a riso n o f fiequency and q u a n tita tiv e dietary m ethods fo r e p id e m io lo g ic studies o f d ie t a n d disease. A m J E p id e m io l 1984; 119:323-334. (19) B locs G . A review o f va lid a tio n o f die ta ry assessment m ethods A m J E p id e m io l 1982; 115:492-505 J N C f. V O I. 79. NO 1. O C TO B K R I987 (C A N C E R R E S EA R C H 50. 2724-2728. M ay I. 1990] Frequent Loss of Heterozygosity on Chromosomes 6q, 11, and 17 in Human Ovarian Carcinomas1 Je H . Lee, John J . Kavanagh, David M . W ild ric k , J . Taylor W harton, and M a r k B lick2 Department o f Medical Oncology J. II. L.. J. J. K.. D. M . W.J. Department o f Gynecology ]J. T. W.J, and Department o f C linical Immunology and Biological The, JM. H ], The University o f Texas M. D. Anderson Cancer Center, Houston. Texas 77030 ABSTRACT aberrations in fresh ovarian tumors has been very limit Recently, tumor-specific allele loss has been shown to be an important characteristic of some tumors. When such loss includes one or more growth-regulatory genes, it may allow the expression of tumorigenicity. Using Southern blots, we analyzed normal and tumor ONA samples from 19 ovarian cancer patients, using a series of polymorphic DNA probes that map to a variety of chromosomal loci. Of 14 informative cases, minor-specific allelic loss was observed in nine (64%) at the estrogen receptor (E S R ) gene locus on chromosome 6q. On chromosome 17p at nonrandom structural or numerical aberrations have been , ported for several different chromosomes (14-17). We, the ` ? fore, analyzed 19 fresh ovarian cancers for possible tum specific allelic loss at 18 different chromosomal loci. Becai of cytogenetic reports of 6q deletions in this disease, we , particularly interested in the 6q region, as well as in otl chromosomal regions that frequently show allelic loss in otl tumors. the D I7 S 2 8 and D I7 S 3 0 loci, allelic losses were also detected in 6 of 8 (75%) and 9 of 14 (64%) cases, respectively. Allelic loss at the H R A S 1 M A T E R IA L S A N D M E T H O D S gene locus on chromosome lip occurred in 5 of 11 (46%) informative cases. The relatively high incidence of these allelic losses observed on chromosome 6q represents the first implication by molecular genetic analysis of this chromosomal region in a human malignancy, and it thus appears to be a genetic change specific to ovarian carcinoma. DNA sequence losses on li p and 17p, also reported for other cancers, may reflect the presence of tumor- or growth-suppressor genes on these chromosomes that are important in the genesis of many tumor types, including ovarian malignancies. Patients and Tissues. Primary and metastatic malignant ovar; tumors from a total of 19 patients at the M. D. Anderson Cam Center (Houston, TX) were obtained from surgical pathology imme ately after surgery. The fresh solid tissues were dissected free of fat a necrotic tissue. The specimens were stored at -70*C until DNA extracted. The histological diagnosis of the tumors studied was esc lished by the Department of Pathology, from paraffin-embedded s< tions. The histological grading was performed according to a systi described by Mauch e ta l. (18). Peripheral blood samples were obtain from the same patients from whom the tumor tissues were procurt INTRO DU CTIO N Stages of disease were assigned according to the classification schet accepted by the general assembly of International Federation of Gyr Although ovarian carcinoma is the most frequent cause of death from gynecological malignancies in the Western W orld (1), the histogenesis and biological characteristics of these tu mors are not well understood. Furthermore, little is known about potential molecular genetic alterations that may contrib ute to development of such tumors or to disease progression. It is now well established that molecular genetic aberrations can alter the expression of genes and may play a role in malignant transformation (2). Recent studies indicate that loss cology and Obstetrics (19). Preparation of High Molecular Weight DNA. Tumor tissue, pulvt ized in liquid nitrogen as described (20), and lymphocytes from perip eral blood samples were lysed in EDTA/sodium dodecyl sulfate sol lion, proteinase K digested, phenol/chloroform extracted, ethanol pr cipitated, resuspended in sterile water, and spectrophotometries] quantified as previously described (20). Allelic Loss Analysis. Fifteen jig of high molecular weight DNA we: digested to completion with the appropriate restriction endonudea: as recommended by the supplier (Bethesda Research Laboratorie or inactivation o f genes at specific chromosomal loci may contribute to the development of certain childhood tumors such as retinoblastoma (3), W ilms' tumor (4), and rhabdo myosarcoma (4). It has been hypothesized that loss of hetero zygosity at a specific locus may indicate loss o f genetic material that suppresses the tumor phenotype. Such negative growth regulating genes have been called "tumor-suppressor genes" or "anti-oncogenes" (5). Because these allelic losses occur not only in hereditary but also in nonhereditary tumors, it has been suggested that negative growth-regulating genes are commonly Gaithersburg, MD), size fractionated in a 0.8-1.2% agarose gel, den: tured, neutralized, and transferred to a nylon filter (21). Hybridizatio to oligo-primed " P-labeled (22) probes was conducted at 42*C for 48 72 h. The filters were then washed at 60"C for 60 min in O.lx salin sodium citrate (SSC = 0.15 m sodium chloride, 0.015 m sodium citrau containing 0.1% sodium dodecyl sulfate and autoradiographed i -70*C using intensifying screens. The following probes and restrictio enzymes were used: chromosome 1, pAT3 (Pstl) (23); chromosome 1 pY NZl 5 ( Taql) (24) and IMR-6 (coRI) (25); chromosome 5, pHGR (B ell) (26); chromosome 6, pH H H 157 (BamHI) (27), pHM 2.6 (EcoRl (28), and pOR3 (PvuII) (29); chromosome 7, HER-A64-1 and HER lost in adult malignancies (6-12). It has been further suggested A64-3 (H in d III) (30); chromosome 11, c-Ha-rasl (BamHI) (31) that the inactivation of a gene is achieved more readily than its activation through specific mutations. Consequently, the loss of an important negative growth-regulating gene may occur even more frequently during tumorigenesis than activation of a positive growth-regulating gene (13). Even though the identification of consistent chromosomal pHINS300 (EcoRl) (32), and SS6 (BamHI) (33); chromosome 12 pSW480-cDNA-5 (Taql) (34); chromosome 17, pYNH37.3 (Taql)(35) pYNZ22 (M spl or BamHI) (36), and cEFD52 (PvuII) (37); chromo some 19, pHP450 (5ocI) (38); and chromosome 22, L335 (H indlll (39). Patients' constitutional DNA samples that were informative, Le.. heterozygous, for a given marker displayed restriction fragments cor Received 6/9/89; revised 12/21/89. The costs o f publication o f this article were defrayed in part by the payment o f page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ' Supported by an Institutional Research Grant, The University o f Texas M . D. Anderson Cancer Center. Houston. T X , and the Bessie Frisch Memorial Fund. 1To whom requests for reprints should be addressed, at Molecular Analysis Incorporated. 8000 El Rio. Houston. TX 77054. responding to each of two alleles in the resultant autoradiograph. Allelic loss from the matched tumor DNA was indicated by the disappearance of restriction fragments specific for either allele. RESULTS Loss of heterozygosity occurred frequently at several chro mosomal loci in ovarian carcinomas. We examined a total of 2724 LOSS OF HETEROZYGOSITY IN OVARIAN CARCINOMAS from 10 different chromosomes. Representative results ^ disp'luyedlin Figs. 1-3. Loss of heterozygosity was observed K ,ci o n seven different chromosomes: chromosomes 2, 6, 7, U 19. and 22. The frequency of allele loss at individual , ranged from one o f eight (13%) at the M Y B locus on A) Ch. 6q Case Case 10 NT 12 NT kb - 1.5 1.0 D ESR (Pvull) C) Ch. 17p Case Case 11 12 NT NT 0.7 kb am -4 B) Ch. l i p Case Case 10 11 N T N T kb - 8.8 M mm * - -- - 6.5 HRAS1 (BamHI) D) C h. 17p Case Case 4 16 N T N T kb D17S28 D 17S 30 (Taql) (BamHI) Tig. I . Southern blot analysis demonstrating loss o f heterozygosity on chro mosomes 6q (A. cases 10 and 12), 1lp ( B . cases 10 and 11), and 1 7 p (C cases 11 mil 12\D. cases 4 and 16) in patients w ith ovarian carcinomas. ON A was isolated iront tumors (T ) and corresponding normal lymphocytes ( .v ). The loci analyzed .mil restriction endonucleases used are indicated under each panel. ESR (Pvull), HR I.S7 (BamHI), D17S28 ( Taql). and D I7S30 (M spl or BamHI), respectively. Mlelic forms are designated to the left o f each panel: a is the longer and b the Muirier fragment. The size in kilobases o f each allele is given to the right o f each I'.mel '. the allelic fragment lost in tum or O N A sample in each panel. In this inure, eases 10. 11. and 12 show loss o f alleles on more than one chromosome. chromosome 6q to six of eight (75%) at the D I 7 S 2 8 locus on chromosome 17p (Table 1). Loss of heterozygosity at one or more loci was detected in 75% (14 of 19) of ovarian carcinomas. The incidence of loss of heterozygosity was remarkably high on chromosomes 6q (64% at the ESR gene locus), 17p (75 and 64% at the D17S28 and D17S30 loci, respectively), and l i p (46% at the HRAS1 gene locus). Allelic Losses on Chromosome 6. To characterize allelic loss on chromosome 6, we used three different polymorphic D N A probes: one for the short arm (D6S29 locus) and two for the long arm (M Y B and ESR loci). Loss of heterozygosity occurred at high frequency in the terminal region of the long arm of chromosome 6, the area to which the ESR gene maps (Table 1 and Fig. 3). The frequencies of loss of heterozygosity were similar in both primary tumors (63%) and metastatic lesions (67%) at the ESR locus. O f six individuals heterozygous at both the ESR and M Y B loci, five (83%) showed an allelic loss at only the ESR jocus. A polymorphic D N A probe (p H H H 157) for the D6S29 locus on the short arm detected nine informative individuals, none of whom revealed an allelic loss. Allelic Losses on Chromosome 17. On chromosome 17p at the D17S28 and D17S30 loci, loss o f an allele was observed in 6 of 8 (75% ) informative individuals and in 9 of 14 (64%) informative patients, respectively (Table 1 and Fig. 2). Among 14 samples that showed loss of heterozygosity on the short arm of chromosome 17 (D17S28 or D I7 S 3 0 ) and that also were informative at the D17S26 locus on the long arm, six (43%) showed concurrent allelic losses. O f eight samples from inform ative individuals who revealed ESR gene allelic loss and also were informative at a chromosome 17p locus (D17S28 or D17S30), five (63%) showed simultaneous allelic losses on both chromosomes. Allelic Losses on Chromosome lip . Among the 19 normal/ tumor D N A sample pairs we examined, 11 (58%) individuals were heterozygous and thus informative. In 5 of these 11 (46%) cases, loss o f heterozygosity occurred. The fact that we did not observe any allelic loss at the IN T 2 locus (on 1lq) suggests that the H R A S I allelic loss was largely due to subchromosomal structural changes. Allelic Losses on Other Chromosomes. To determine whether these restriction fragment length polymorphism allelic losses were specific for chromosomes 6q, l i p , and 17p in ovarian C ase 15 Kb N T FiR. 2. Case 15 has a terminal deletion w ith " f nn allele at D l 7S2S but retained geno- 141 *111DI7S30 in tum or tissue. Sim ilarly, case 14 has an interstitial deletion w ith loss o f one hide at DI7S30 but retained genotype at l>l S28. These two loci are separated by only centimorgans at the tip o f !7p. *. the allelic iragment lost in tum or D N A sample in each panel. NT C ase 19 Kb m , N T N T Kb - 1.1 -0.7 D17S28 D17S30 (Taql) (Bam HI) 2725 D17S28 D17S30 (Taql) (BamHI) LOSS OF HETEROZYGOSITY IN OVARIAN CARCINOMAS C ase 13 C ase 14 Fig. 3. Cases 13 and 14 have partial dele tions With loss o f one allele at ESR locus (6q24T27) in tumor tissue but retained geno type at M Y B locus (6q22-23). *. the allelic fragment lost in tumor D N A sample in each panel. 2 .6 - 1.5- H 1 .1 - t i ' MYB ESR (EcoRI) (Pvull) MYB ESR (EcoRI) (Pvull) carcinomas and not simply a result of widespread random losses indicate that inactivation of a tumor- or growth-suppressor of genetic material, we employed other DNA probes capable of locus located there is an earlier event in ovarian tumorigenesis. detecting restriction fragment length polymorphisms at 14 ad However, this hypothesis needs to be tested by looking for ditional loci on chromosomes 1, 2, 5, 6, 7, 11, 12, 19, and 22 tumor-specific 6q allele loss in the DNA of such early and (see Table 1). Among these, at each chromosomal locus where borderline cancers before trying to classify such genetic changes constitutional heterozygosity was observed, heterozygosity was as early or late events. also maintained in the matched tumor tissue in most cases. The fact that the more distal E S R allelic marker on 6q is lost more frequently (65%), compared to the more proximal MYB DISCUSSION marker (13%), suggests that the E S R locus is closer to the region possibly uniquely lost in ovarian cancer. This notion is The demonstration of loss of heterozygosity at the H R A S1 supported by the fact that the only tumor sample showing loss locus, as well as other 11p markers, in a variety of human of heterozygosity at the M Y B locus had concordant loss of cancers (reviewed in Ref. 40) including ovarian carcinomas (20) heterozygosity at the E S R locus. Of further interest, other suggests that their development or progression may involve loss investigators in our laboratory have analyzed breast cancer or inactivation of normal genes on lip. Previously we reported DNA samples from 15 informative patients and observed an that tumor-specific allele loss occurs at the H R A S 1 locus on E S R gene allelic loss in tumor DNA from only one patient, chromosome 1Ip in 50% of ovarian carcinomas (four primary demonstrating that loss of 6q sequences is not common to all tumors, eight metastases, and five ascites samples) (20). The cancer types.3 current study extends that observation at about the same fre In addition to the cytogenetic reports of chromosome 6q quency (46%) to an expanded sample, cases 1-9 in Table 1, deletions, an isochromosome for the long arm of chromosome which includes five additional primary ovarian carcinomas, and cases 10pl9 in Table 1, which includes two additional metas tases. More significantly, our study clearly shows that tumor-spe cific allele loss occurs in ovarian carcinoma at an even higher frequency for markers on chromosomes 6q (64% at E S R ) and 17p (75% at D 1 7 S 2 8 , 64% at D 1 7 S 3 0 ) than for lip. This is the first report of tumor-specific allele loss from chromosomes 6 and 17 in ovarian carcinomas, suggesting the potential exist ence on these chromosomes of tumor- or growth-suppressor genes whose loss promotes this disease. Deletion of part of chromosome 6q has been observed in several cytogenetic studies of ovarian carcinoma (14-17), which prompted us to investigate the long arm of chromosome 6 at the molecular level. Our finding of such a high incidence of tumor-specific allele loss in both primary and metastatic tumor tissues at the E S R gene locus (6q24-q27) (41) suggests that this loss might be a genetic event specific to ovarian cancer. Because 6q deletions have been observed cytogenetically in early gynecological malignancies, including a noninvasive ovarian carcinoma (42) and an ovarian granulosal cell tumor (43) (a 17 (i[17q]) was also reported in early gynecological malignan cies, indicating possible deletion of the short arm of chromo some 17 (42). Our results suggest that the mechanism of allele loss from chromosome 17p resulted from partial or total loss of chromosome 17, because in 6 of 14 (43%) instances there were concurrent allele losses at both short and long arm loci Because others have shown that tumor-specific allelic loss from chromosome 17 occurs in several different tumor types (12, 44--46) and because our data imply frequent chromosome 17 losses in ovarian cancer, genes on this chromosome might be involved in a cooperative role with other genes on 6q (or 1lp) in mechanisms of tumorigenesis or disease progression. Fur ther, the high incidence of concordant allelic losses from chro mosomes 6q and 17p in 7 of 14 (50%) patients informative for both loci was, in our study, very impressive. Within the limits of the present study, it would be very difficult to determine the timing of these allelic losses from different chromosomes. The target genes in ovarian cancer may be located on both chro mosomes. In conclusion, whereas allele loss from chromosome 1lp has1 borderline malignancy), loss of DNA sequences from 6q might 1 E. H . Koh el a/., personal communication. 2726 -1.5 - 1.0 -0.7 essor lesis. I for and nges lost iYB the >n is loss sc ther icer an ent, all 6q me an- 10- ele iss ere id . )m .2, 17 be P) iroor ts ie ie > LOSS OF HETEROZYGOSITY IN OVARIAN CARCINOMAS in r- rs au ------ft. Q Q- ZZ US S,p. Wx>v, & q <E ss E ,0 I X X X x l nnaa I I sns x\ x^ -Qr x 5? I x v Vx . O7Z h\-- -- Si X fc? X X X --_ been linked to at least eight cancer types and chromosome 17p allele loss to three different neoplasms (see Ref. 40), loss of chromosome 6 sequences has not been associated previously with any cancer. This makes it more likely that the 6q allelic losses uniquely observed here for ovarian cancer indicate the presence on 6q of a putative tumor- or growth-suppressor gene specific to this cancer. ACKNOWLEDGMENTS - ac ^-- - - x. 4 - 3 "" aj r. SC 4^1 'S'. Z *C CO * U. -- C 'O 21 9-i 1 S 5 O 5 Q1' 3* 2Q>S0 T a. E HP E 1- X I I I XX I a n n a I I j cb J cb I 1 cb I a cb I ~ r- t I oC S Uo Pf--E3 -- aE | <0X W We are indebted to Y. Nakamura for kindly providing probes pYNZ15, pYNH37.3, pYNZ22, and cEFD52; to C. Dickson for SS6; to G. I. Bell for pHlNS300; to D. Stehelin for pHM2.6; to M. Hansen for valuable advice and critical review of the manuscript; to Leslie Wildrick for editorial assistance; and to Maricela Malbrough for typing the manuscript. ND a/b a/b ND a/b ND Q Q Q ^Q ^ r-? Z Z Z CBZ CB CBM o H I * REFERENCES C7 0/6 (0%) ' | CB XV n XX w CB CB CB CB CB CB CB CB 0/5 (0%) w<Uea 2w <A E ng as 55 1. Bertelsen. K.. Jakobsen. J. E., Andersen. S.. Ahrons, P., Hjortjaer Pedersen. P.. Kiaer. H ,, AriTmann. E ,, Bichel, P.. Boestofte. E., Stroyer Christophersen, I., Gregersen. E., Kern Hansen, M .. Holund. B ,, Jacobsen. M .. Kaalund Jensen. H., Love Jepsen. F ,, Larsen. G.. Sandberg Neilsen. E., Nyland. M .. Olsen. J.. Panduro. J., Rank. F.. Sell. A., and Sogaard. H. A. A randomized XX X C X CB CB 1 CB I CB iii e 5/11 (46%) PI 2 |g e iC 1o = study o f cyclophosphamide and cu-piatinum with or without doxorubicin in advanced ovarian carcinoma. Gynecol. Oncol., 18: 161-169. 1987. 2. Klein. G.. and Klein. E. Evolution o f tumors and the impact o f molecular oncology. Nature (Lond.), 315: 190-195. 1985. 3. Cavenee, W. K.. Dryja, T. P.. Phillips. R. A.. Benedict. W. F ,, Godbout, R ,, XX a to X n .2? 3 E Gallic. B. L., Murphree. A. L., Strong, L. C.. and White. R. L. Expression CB eCB e O- gS 5f3UEl (<"-J'~ .M5 O= o f recessive alleles by chromosomal mechanisms in retinoblastoma. Nature (Lond.). 305: 779-784, 1983. 4. Koufous. A.. Hansen, M . F ,, Copeland. N. G.. Jenkins. N. A.. Lampkin. B. naa a 9S !?o - a -o J G . and Cavenee. W. K. Loss o f heterozygosity in three embryonal tumors suggests a common pathogenetic mechanism. Nature (Lond.), 31 6:330-334. 1985. 5. Knudson. A. G. Model hereditary cancers o f man. Prog. Nucleic Acid Res. I I cb cb I cb cb cb I -L -L -L e cb cb cb cb 0.2 '35 ^>-#r; 25 0E = M o l. Biol., 29: 17-25, 1983. 6. Fearon, E. R., Feinberg, A. P., Ham ilton, S. H.. and Vogelstein, B. Loss o f genes on the short arm o f chromosome 11 in bladder cancer. Nature (Lond.) 3 1 8 :377-380, 1985. e ^ cb cb i " w 7 \X X X C x X X X CQ CQ tQ 4-- Q f l (S CB Qx Z> 8 n 'r - ao .. E S 2ts <gJ i"O 7. M ukai. S.. and Dryja, T. P. Loss o f alleles at polymorphic loci on chromo some 2 in uveal melanoma. Cancer Genet. Cytogenet., 2 2 :45-53, 1986. 8. Seizinger. B. R ,, Martuza, R. L.. and Gusella, J. F. Loss o f genes on chromosome 22 in tumorigenesis o f human acoustic neuroma. Nature (Lond.), 322: 644-647,1986. 9. Lundberg, C.. Skoog, L., Cavenee, W. K ,, and Nordenskjold, M . Loss of heterozygosity in human ductal breast tumors indicates a recessive mutation 1E C8 4' on chromosome 13. Proc. Natl. Acad. Sci. USA. 84: 2372-2376, 1987. XX X . oQ Z x x x x oo ^ CB CB CB o -- -o --c c --x 10. Zbar, B., Brauch. H., Talmadge, C.. and Linehan. M . Loss o f alleles o f loci on the short arm o f chromosome 3 in renal cell carcinoma. Nature (Lond.), E8 s 327: 721-724. 1987. E-Sa "c 11. Solomon, E., Voss, R., H all, V., Bodmer. W. F ,, Jass. J. R., Jeffreys. A. J.. Lucibello, F. C., Patel, 1., and Rider. S- H. Chromosome 5 allele loss in Z cb cb c i c B s c B n n n t t t cb Z Z o 2- Lo5 *ccw]?^TgoT 12. human colorectal carcinomas. Nature (Lond.), 3 2 8 :616-619. 1987. Yokota, J., Wada. M ., Shimosato, Y ,, Terada, M ., and Sugimura. T. Loss o f e U heterozygosity on chromosomes 3. 13, and 17 in small-cell carcinoma and nZ Q Z x xQ ^T ^ Z x IC Cd Z m ' i= I 1i on chromosome 3 in adenocarcinoma o f the lung. Proc. Natl. Acad. Sci. USA, 8 4 :9252-9256. 1987. a a ZZ ^(8 D Z s i 5 I>osvO2v"> 0 5- 13. Weinberg, R. A. The genetic origins o f human cancer. Cancer (Phila.), 61: 1963-1968. 1988. 14. Woods. L. K-. Morgan, R. T., Quinn, L. A ., Moore, G. E ,, Semple, T. U.. and Stedman, K. E. Comparison o f four new cell lines from patients with adenocarcinoma. Cancer Res., 3 9 :4449-4459,1979. 15. Whang-Peng, J., Knutsen. T ,, Douglass. E. C , Chu, E ,, Ozols. R. F., Hogan. 0> 0> 0> 0> >0 >0 >0>0>0 0E 0E = 0E 0E - oE <x oEoE o H s s E " - 2P W . M ., and Young, R. C. Cytogenetic studies in ovarian cancer. Cancer Genet. Cytogenet., 72:91-106, 1984. 16. Trent, J. M ., and Salmon. S. E. Karyotypic analysis o f human ovarian Ex =j?*.2 o c .y - ,-os E8 e carcinoma cells cloned in short-term agar culture. Cancer Genet. Cytogenet.. 3:279-291, 1981. 17. Sheer. D.. Sheppard. D. M ,, Gorman, P. A., Ward, B., Whelan. R. D. H.. and H ill, B. T. Cytogenetic analysis o f four ovarian cell lines. Cancer Genet. Cytogenet., 2d: 339-349, 1987. 18. Mauch, P. M ., Ehrmann, R. L., G riffiths. C. T ,, et al. Radiation therapy in U ^U U U U^ n ( iU M (j U t" U U l< < < U O < U ` U < ;< < ^ < U < o < h 2 (/)(/3 U U < C /3 < 2 < (/3 (/3 U 2 (/)U ltf3 u I ll i.S a 511 Si stage I I ovarian carcinoma. The influence o f histologic grade. Cancer (Phila.), 4 5 :1344-1351.1980. 19. Classification and staging o f malignant tumors in the female pelvis. Acta Obstet. Gynecol. Scand., 50: 1-7, 1971. 20. Lee, J. H.. Kavanagh. J. J., W harton, J. T ,, W ildrick, D. M ., and Blick. M . Allele loss at the c -H a -ra il locus in human ovarian cancer. Cancer Res.. 49: 1220-1222, 1989. 2727 I LOSS OF HETEROZYGOSITY IN OVARIAN CARCINOMAS 21. Southern. E. M . Detection o f specific sequence among D N A fragments separated by gel electrophoresis. J. M ol. Biol.. 938: 503-517, 1975. 22. Feinberg, A. P.. and Vogelstcin. B. A technique for radiolabeling D N A restriction endonuclease fragments to high specific activity. Anal. Biochem., 3 7: 266-267. 1984. 23. Prochownik. E. V.. Markham. A. F., and Orkin. S. H. Isolation o f a cD N A clone for human antithrombin III. J. Biol. Chem.. 258: 8389-8394. 1983. 24. Nakamura. Y.. Culver. M .. O'Connell. P.. Leppert, M .. Lathrop, G. M ., Lalpuel. J-M .. and White, R. Isolation and mapping o f a polymorphic D N A sequence <pYNZ15) on chromosome 2 (D2S50). Nucleic Acids Res.. 16: 5702, 1988. 25. Shiloh. Y., Kanda. N ,, Kunkel. L. M .. Bruns. G., Sakai. K., and Latt. S. A. Two RFLPS identified by a human chromosome #2 clone at 2p15--2p 16 (H<j3M8 assignment No. D2S5). Nucleic Acids Res., J3: 5403, 1985. 26. Weinberger. C., Evans. R.. Rosenfeld, M . G., Hollenberg, S. M ., Skarecky, D., and Wasmuth. J. J. Assignment o f the human gene encoding the gluco corticoid receptor to the q l 1--q 13 region on chromosome 5. Cytogenet. Cell Genet., 40: 776, 1985. 27. H off. M ., Nakamura, Y.. and Holm , T. Isolation and mapping o f a poly morphic D N A sequence (pHH H 157) on chromosome 6p |D6S29]. Nucleic Acids Res.. 6 : 5217, 1988. 28. Dozier. C.. Walbaum. S.. Leprince. D., and Stehelin. D. coRI R FLP linked to the human myb gene. Nucleic Acids Res., 4: 1928. 1986. 29. Castagnoli, A.. Maestri. I.. Bernardi, F.. and Del Senno, L. PvuU RFLP inside the human estrogen receptor gene. Nucleic Acids Res., J5: 866, 1987. 30. Lee, J. S., Ro, J. S., Eisbruch, A., Shtalrid. M .. Ferrell. R. E., Gutterman. J. U.. and Blick. M . M ultiple restriction fragment length polymorphisms o f the human epidermal growth factor receptor gene. Cancer Res., 48: 4045-4048, 1988. 31. Chang, E. H.. Gonda. M . A., Ellis. R. W.. Scolnick, E. M ., and Lowy, D. R. Human genome contains four genes homologous to transforming genes o f Harvey and Kirsten murine sarcoma viruses. Proc. Natl. Acad. Sci. USA. 79: 4848-4852. 1982. 32. Bell. G. I., Karam. J. H., and Rutter, W. J. Polymorphic D N A region adjacent to the 5' end o f the human insulin gene. Proc. Natl. Acad. Sci. USA. 78: 5759-5763. 1981. 33. CaSey, G., Smith, R., M cG illivray. D.. Peters. G., and Dickson, C. Charac terization and chromosome assignment o f the human homolog o f in t-2, a potential proto-oncogene. M ol. Cell Biol., 6: 502-510. 1986. 34. Ca^>on, D. J., Seeburg, P. H., McGrath, J. P., Hayflick, J. S.. Edman. U., Levinson. A. D.. and Goeddel, D. V. Activation of Ki-ros2 gene in human colon and lung carcinomas by two different point mutations. Nature (LonH 304: 507-513. 1983. ^ 35. Nakamura. Y ,, Bragg, T., Ballard. I_Leppert. M.. O'Connell. P ,, Lad,, G. M .. Lalouel. J-M .. and White. R. Isolation and mapping o f a polymorph*- DN A sequence pYNG 37.3 on chromosome I7p |D17S28). Nucleic Aci2 Res.. 16: 782. 1988. 36. Nakamura. Y.. Leppert, M ,, O'Connell. P.. WolfT. R.. Holm . T., Culver M M artin. C.. Fujimoto, E.. H off. M ., Kum lin. E.. and White, R. Vaiik number o f tandem repeat (V N TR ) markers for human gene mapping Scienn. (Wash. DC), 2JJ: 1616-1622, 1987. ` 37. Nakamura. Y.. Fujimoto. E.. O 'Connell. P ,, Leppert. M ., Lathrop, G. M_ Lalouel. J-M .. and White. R. Isolation and mapping o f a polymorphic Obf* sequence cEFDS2 on chromosome I7q (D17S26). Nucleic Acids Res, 786. 1988. '* 38. W ainwright. B. J.. Watson. E. K., Shephard, E. K.. and Phillips, I. R. RFXp for a human cytochrome P-450 gene at 19ql3.1-qter (H G M 8 provision,! designation CYP1). Nucleic Acids Res., 13: 461, 1985. 39. Shin. D. M .. lnce. C., Shtalrid, M .. Lee. J. S.. Ro. J. S., Donner. L.. Ferrell. R. E.. Hong, W. K., W ildrick. D ,, and Blick. M. Reduction to homozygosity at the SIS/PDGF-2 locus in human mesenchymal tumors. Biochem. Biophn. Res. Commun., 155:692-699, 1988. 40. W ildrick. D. M . Molecular genetic studies o f colon cancer. Hematol. Oncol. Clin. North Am.. 3: 1-19, 1989. 41. Human Gene Mapping 9.5 (1988): update to the N inth International Work shop on Human Gene Mapping. Cytogenet. Cell Genet., 46: 1-258, 1988. 42. A tkin. N. B.. Baker. M . C., and Ferti-Passantonopoulou. A. Chromosome changes in early gynecologic malignancies. Acta Cytol., 2 7 :450-453, 1983. 43. Teyssier. J. R.. Adnet. J. J.. Pideon, F., and Bajolle. F. Chromosomal changes in an ovarian granulosa! cell tu m o r sim ilarity with carcinoma. Cancer Genet Cytogenet.. 14: 147-152, 1985. 44. Toguchida. J.. lshizaki. K.. Sasaki. M . S.. Ikenaga, M ,, Sugimoto. M_ Kotoura. Y.. and Yamamuro. T. Chromosomal reorganization for the expres sion o f recessive mutation o f retinoblastoma susceptibility gene in the devel opment o f osteosarcoma. Cancer Res.. 48: 3939-3943, 1988. 45. Vogelstcin. B ,, Fearon, E. R ,, H am ilton. S. R., Kern, S. E., Preisinger, A. C.. Leppert. M ., Nakamura. Y., W hite. R., Smits. A. M . M ,, and Bos, J. L Genetic alterations during colorectal tum or development. N. Engl. J. Me<L 3 1 9 :525-532. 1988. 46. Law, D. J.. Olschwang, S.. Monpezat. J. P ,, Lefrancois. D ,, Jagelman, D_ Petrelli. N.. Thomas. G., and Feinberg, A. Concerted nonsyntenic allelic loss in human colorectal carcinoma. Science (Wash. DC), 2 4 1 :961-965,1988. Aims and Scope Hie H n ii. J i J in tn n iln i Cancer w ill accept papers \\h .d u o nia Mi-Mow--iufo n nilio n~ei>ii slrtute a distinct con tribution to knowledge, and arc relevant to the clinical, epidemiological, pathological or molecular aspect*, o l oncology . I lie Journal is published m o nthly o n behalf of* the Cancer Research Cam p aig n by the Scientific and Medical D ivision o f M acm illan Press Ltd. ( op\ right 0 1**91Cancer Research Campaign. ISSN(HIM7 ilD|| M aiuiscnpls plus tw o copies and all editorial correspondence should be sent to: Dr IV R Twenty m ail. E ditor. B ritish Journal o f Cancer. M RC ( limcal Oncology and Radiotherapeiiiics U nit. Hills Road. Cambridge cb: : q h.u k. I clephone: 00 23) 2 IO II5 l av (02:5)215556 Annual subscription prices: EEC 280, elsewhere 310/US $620 (sterling rare is definitive) Orders must be accompanied by remittance. Cheques should be made payable to M acm illan Press and sent to: Subscription Department. M acm illan Press, Brunei Road. H oundm ills. Basingstoke. RG21 2XS. U K . Where appropriate, subscribers may make payments into U K Post Office G iro Account No. 519 2455. L u ll details must accompany the payment. Second class postage permit USPS689 450. Postmaster please send address corrections to M ercury A irfre ig h t International Ltd . 2323 E-F R andolph Avenue. A vend. NJ 07001, USA (US M a ilin g agent). Second class postage paid at Rahway. NJ. Enquiries concerning advertising space or rates should be addressed to: Michael Rowley. Advertisement Manager. M acm illan Press Ltd.. 4 Little Essex Street, London. Telephone: 071 836 6633. Fax: 071 3794204. A ll business eoi rcspoiulenee and reprint requests should be addressed to the Scientific Ac. Medical Div ision. M acm illan Press L u i.. I lo u n dniills. Basingstoke. Hampshire RG2I 2XS. U K I clephone: (0256) 29242 I a v d ) 2 '6 iS 10526 A ll rights o f reproduction arc reserved in respect o f all papers, articles, illustrations, etc., published in this journal in all countries o f the world A utho rization to photocopy items for internal or personal use. o r the internal or personal use o f specific clients, is granted by M acm illan Press Ltd for libraries and other users registered w ith the Copyright Clearance Centre (CCC) Transactional Reporting Service, provided that the base fee o f S3 50 per copy, is paid directly to C CC. 27 Congress St . Salem. M A 01970. USA. 0007 - 0920/91 $3.50 + $00.00 <n\tT t/U ,\it.//it./i The cowl slum J generalised M.lieme lor ihe mechanism ol cell mioxicjtion h> toxins .ml iinmunoioMus Ihe topic of systemic mummoioxin ih e rjp s oi'c.nuer i> reviewed in iliis issue by l)r FJ W .iu r / u ic / .ik 11' N N ) i - * ii*i. - i-, .3- . slr jjiiif a .iij .pani,f i,* <ut(tw M in ihi. uriul. fx-j wish in m i l t iU l l itu i ite t l . u <nj uli-r. jiK rili ! t.NKinol Ak.irjii>|(l). Ite piiM isteis jim\ Hit I cll.--.hip Ihc n lilte u l ..llfnillcf pwn.es i*1 *n mk h n o ., u u lr n m k .iJ .n f .U ij. .i . u k m i i I Whilst rs fts rlT.m i* rru ik li> n i IIif IH-sis hii.I In hili^tes ills<1stu fai Iu f u sjy f <i.l . t e n t e d ululi Ihis l..|.rn jl ahiasilil .ml) te I\|V M -I h \ I In.* Iv jv M lim * I eclituqui'v Siuilhampion m il .I IM r . n . i i llm I'll I l .is s ili U m i l i c i - | i d . I l i e ( i i e s h . u i i l* ie s. OM W i.L io g . Surrec Br J C ancer (1991). 64. 617 62U M acm illan Press Lid G U E S T E D IT O R IA L Epithelial ovarian cancer: a cytokine propelled disease? S Malik & F. Balkwill B io lo g ic a l Therapies lu ih o ra lo ry . Im p e ria l C ancer Research Fond. L in co ln 's Inn Fields. I jm ilo n I f f 2 A .IP X . O v a n a n cancer is the com m o ne st cause o f d eath fro m gynae cological cancer in the developed w o rld (S ilverbcrg e i at., 1990, B o o th Sc B eral. 1985). T h e age a dju sted death rates fro m o v a ria n cancer have c o n tin u e d to increase since 1955 (S ilverbcrg r i aI.. 1990). A m a jo r cause o f the high m o rta lity rate fro m o v a ria n cancer is the late p re sen tatio n o f the disease in o ver 6 0 % o f p atie nts, w ith an associated 5-year s u rv iv a l rate o f a p p ro x im a te ly 5% (O zo ls e l at.. 1980). Several factors have been im plicated in the a etio lo gy o f o varian cancer, in c lu d in g a genetic pre disp o sition , rep rod u c tive history, sm oking, infection w ith m um ps virus, dietary fa cto rs, a nd the use o f ta lc (P ive r, 1987). T h e m ost c o n v in c ing fa cto r to emerge as an influence on the developm ent o f o varian cancer is horm on a l. Several studies have shown that the risk o f o v a ria n cancer in n u llip a ro u s w o m e n is 1 .5 -3 .3 tim es greater than tha t o f parous wom en (S ilverbcrg et at., 1990; Piver, 1987). The incessant o v u la tio n ' hypothesis p ro posed in itia lly by Fath a lla (1971), and expanded by others (Casagrade et aI.. 1979) attem pted to e xpla in the protective effect o f pregnancy as being due to the rest periods afforded b y p re gn an cy fro m incessant o v u la tio n . T h e incessant o v u la tio n hypo the sis is su p p orted b y the m o re co m m o n occurrence o f m a lig n an t change in the righ t ovary w hich ovulates m ore fre qu e ntly (C ruicksha nk. 1990). C ram er and W elch (1983) suggested th a t w h ils t incessant o v u la tio n led to the fo rm a tio n o f in c lu s io n cysts, excessive and co n tin u e d g o n a d o tro p in stim u la tion contributed to m alignant tra nsfo rm a tio n by tro p h ic effects on ovarian e pithelium . T hus, suppression o f g o n a d o tro p in release b y o ra l co ntra ce ptive s c o u ld e x p la in the decreased in cidence o f o v a ria n cancer in w o m e n ta k in g o ra l co ntra ce ptive s. T h is h ypo the sis presents the case fo r the involvem en t o f a no the r g ro u p o f peptide cell regulators, the cytokines, in the pathogenesis o f ovarian cancer. C ytokines are lo w m olecular w eight (< H 0 k D a ) peptide cell re g ula tory factors tha t have w ide ran gin g effects on d iffe re n t cell p o p u la tio n s (B a lk w ill Sc B u rke . 1989). Some cytokines can directly o r indirectly in h ib it tum ou r grow th and have been used in cancer the ra p y (B a lk w ill. 1989). H o w ever. in ap p ro priate cytokine a c tiv ity has been im plicate d in the pathogenesis o f infectious, in fla m m a to ry and a u to im m une diseases, and there is n o w in cre asing evidence tha t cytokine production bv tum ours m ay contribute to the patho p h ysio lo g y o f cancer F o r exam ple, in te rle u k in -6 has been sh ow n to be an ttu to c rin c g ro w th fa c to r in m ye lo m a , a nd o v e rpro d uctio n o f IL -6 causes the systemic sym ptom s and the im m u n o lo g ic a l a b n o rm a litie s in C a s tle m a n s disease (K is h im o to , 1989) T u m o u r necrosis fa c to r ( T N F ) has been im plicated in the suppression o f haematopoesis in h a iry cell leukaem ia, and in bone m a rrow necrosis in cancer patients (L in d e m a n n et a l.. 1989; K n u p p , Pekala & V o rn e liu s. 1988). E xperim ental studies also suggest roles fo r cytokines in cancer cachexia and the hypcrcalcaem ia o f cancer ( D a rlin g et a t . 1990; S ato et a l.. 1987) C o n stitu tive p ro du ction o f several cytokines by hum an o v a ria n cancer cell lines and fresh tu m o u r b io p s y m a te ria l Correspondence: S.T.A. M a lik. Department o f Clinical Oncology. Hammersmith Hospital. Du Cane Road, London W I2 0HS. UK Received 26 Match 1991. and in revised form 10 May 1991 has been dem onstrated. N a y lo r et al. (1990) have shown T N F m R N A and p ro te in is expressed by o v a ria n tu n cells in biopsies fro m patients w ith e p ith e lia l o va ria n cat In contrast, in the n orm al the rut and bovine o\ im m un o rca clivc T N F ' has been detected in granulosa a nd s tro m a l m acrophages, b u t n o t o v a ria n e p ith e liu m (N St T e rra n o va , 1989). C o n s titu tiv e p ro d u c tio n o f m a cro p l co lon y stim u la ting facto r (M -C S F ) and expression o l c-fms oncogene (encoding the M -C S F receptor) has I dem onstrated in hum an o varian cancers and ovan an c.v cell lines (R am akrishnan et al.. 1989; K ascirtski et al.. I `> but not in norm al o varian epithelium . Indeed, v measurements o f scrum M -C S F in p atients w ith ova cancer m ay be useful in m o n ito rin g disease a c tiv ity ( K a c r et a l.. 1989). A m o n o cyte c h e m o ta c tic fa c to r p ro du ced o v a ria n cancer cell lines (B o tta z z i et a l.. 1985), is n o w kn< to be the T N F and I L - I in d u cib le c yto kin e m ono, ch e m o ta ctic p ro te in ( M C P - I ) (B o tta z z i et a l.. l lJ Inte rle u kin-6 p ro du ction is tra nsien tly induced g on adotropin-pnm ed h yperslim ulaied ovaries (M o tro et 1990), can be detected in the a scitic flu id o f o v a ria n c:n patients (E rro i el al.. 1989), and is c o n s titu tiv e ly produced hum an ovarian cancers and o varian cancer cell lines (W ;n et al.. 1990). Interleukin-1 has also been detected in a p ro[ lio n o f human ovarian cancers by in situ h yb rid isa tio n im m unohistochem istry (unpublished observations). The t fo u n d cell re g ula tory effects o f cytokine s at p ic o m o la r i cent radons in vitro, suggest tha t the expression o f c y to k i by hum an epithelial ovarian cancers m ay have an im po rt ro le in th e ir in vivo p a th o p h y s io lo g y . T u m o u r necrosis fa cto r has a n titu m o u r effects in m u i and hum an tu m o u r xenograft models (H ara n aka et 1984) W e have studied the effect o f in tru p c rito n c a l I a dm inistratio n in in tra pe rito ne a l hum an ascitic ov.n cancer xenograft models (M a lik et a!.. 1989). T N F th n prolonged survival in tw o o u t o f three xenograft models. I paradoxically, restored the a b ility o f all three n o n -T N I p ducing ovarian tum ours to form peritoneal tum ou r im pl:n Exogenous T N F adm inistration therefore led to a path oiop picture reminiscent o f m etastatic hum an o varian cancel the xenograft models T o study the possible role o f tuinproduced T N F on peritoneal im p la n ta tio n and invasion, b e h a vio u r o f Chinese H a m ste r O v a ry cells th a t had l> transfected w ith the hum an 1 N F gene ( C H O /T N F cells) ' com pared to C'H O cells that co nta ine d the tra nsfe ction \ to r alone (C H O /N E O cells) (O lilT et a t . 1987). C H O / I ' cells showed enhanced im p la n ta tio n to the surface o f p eriton eu m and live r, and also m etasiasiscd to the Inn Furtherm ore, the metastatic ca p a b ility o f C H O /T N F c o u ld be sp ecifica lly a brog ate d b y in je c tio n o f a n iib o d it-' T N F (M a lik et a l, 1990) O th e r studies in expcrinicn models have shown that T N F can enhance tu m o u r metasi.' (G iavazzi et a l, 1990) P roperties o f TN F" that coi p o te n tia lly co ntribu te to p ro m o tio n o f metastasis inchs tim u la tio n o f angiogenesis (F 'ra te r-S c h ro d e r et a l . 19s enhancement o f tu m o u r cell adhesion to host cells (Rice a l., 1988), and in du ctio n o f tu m o u r cell m e tallop ro tein.i ( l lo et a l . 1990). The kn o w n p ro co a g u la n t a ctiv itie s o f 1 ' (P ober, 1987), and the in d u c tio n o f platelet derived grov fa cto r (H a jja r el a l . 1987). co uld c o n trib u te to the g en eral. f>I s n \ l \ l l k . M H V I K VVII I o f tu m o u r 'ir i'iiu . .in lin p o ita n l >icp in the establishm ent o f tum ou r % m m etastatic 'lie '; (D v o ra k . 1986). I lie o ilie r a i o k i m s lii . i l are pro du ced b> o v a ria n cancers are m iim .ilclv associated w ith T N I; m the cytokine netw ork i Haiku ill \ Dm kc. 1^x9) Interleukin-1 and M -C S F both in duce I N I - m m .'iio ic ie s and m esenchym al cells (P h ilip & f pstcin. 19X6. l e V V ilc c k . 19X7, W a rre n A R a lp h , J 9 8 6 ). I N I im iiiccs the p ro d u c tio n ol I N I ' . M -C S F . I I . -1. 11.-6. and M O ' -1 in several cell p o p u la tio n s ( H a ik u ill. 1989). Ih u s there exists a p o te n tia l sell' p e rpe tua ting c y to k in e in d u c tio n netw ork in the iic in u v ol'. tu n io u r cell o r a norm al host cell I'cncr.ilin ^ iv io k m c s which can iiilh icn cc tiu n o u i biology. I o i exam ple II I c n li.iik c d n ict.isiasis o l n ie l.in o in a cells i ( iiava/ / 1 i / t i l . and p iou io ted the peritoneal im plant .lim n ol h u m a n o v .in a n cancer xe no grafts (M a lik e( ol. u np ub lishe d d a i.it II -n has been re p o rte d to enhance the m o tility of breast cancer cells, a p ro p e rly that is associated w ith enhanced m eiasiatie p otential (T a m m r / ol.. 1989). M - ( S| and M C I' m.iv co ntribu te in d ire c tly to metastasis by te c tlin in g m acrophages to tu m o u r sues. These c o u ld be a fu rth e r source o f metastasis enhancing cytokines such as IL - I and I N I . and secrete o th e r facto rs ih .n have g ro w th stim u- la to iv ellects on o va ria n cancer cells (W cla n d cr et o i . 1982). ( an le .is o ii.ib le lo iie la tio iis lx* m ade between the elfecis o f c y to k in e s m o v.h i.m cancer models w ith some o f the kn o w n ri-k l.ic to is a ih I prognostic indicators m ovarian cancer? I ig u ie I show * ilie p o te n tia l sites o f e yto kin e induced tu m o u r progu-NNion and the cytokine s tha t c o u ld influence these stages (h e damage in volved m o vu la tio n c o u ld lead to the lo cal release o l evtokiues. such as I N T and TC iT -0 hy host m a c ro phages. tha t c o n trib u te to the healing lih ro tic response (Pos- ih c lw .tiie & Sever. l `*9oj B.ill these c y to kin e s are capable o f inducing m ciallopioiem .ise production h \ tu m o u r cells, and iiicie.ising then invasive potential ( lio et o i. I w o . W elch ct . f w n i I here is m i evidence as \ c l tha t c y to kin e s can c.ium * m alignant lu n s lo r m u tio n o f n o rm a l e p itlie lia . a lth o u g h I N l c.m i a use 1 ) \ \ stra nd damage in cells- susceptible to its iv io to v ii e th vis I N I also activates several putative ik ogeiics iii ilii d m i ' t-)u o t llic iu ie i t i o l . 19X9) w h ic h is ovcicvpcesscd in sojuc hum an o va n a n tu m o u rs t l) e Circve et t i l . 19y u O verexpiession o f the c-it A H -2 oncogene is an adveise p io g n o siic in d ic a to r m hum an o v a ria n cancer i.- j:r . 1 .-j-.j-.XTU) C ERB B? UGAND m csr t * iA M i iNDuCitnr. '!( H ASt D AD*t lo ro F INCMf ASC l> MO l u l l r A I ION T*Kir<- I I `o le iiiial m U`s ol cytokine inter.iclious m ovarian (Slam on f t a i. 1989). and the s im ila rity o f the c-erbB -2 oncogene product to the epiderm al gro w th fa cto r receptor suggests th a t an u nid e n tifie d c y to k in e m ay be in v o lv e d in p ro m o tin g o va ria n cancer g ro w th . Levels o f c-e r/jB -2 oncogene expression corre la te w ith resistance to the cyto cid a l effects o f T N I ' ( H u d /ia k c l u i . 1988). A lth o u g h there is no evidence as yet t h a t T N I ' regulates -e r6 B -2 oppression, one c o u ld speculate that c-erAB-2 positive tum ours are T N F resistant tum ours that have emerged a lter prolonged exposure to IN I C u rrent evidence indicates tha t t^e n orm al ovarian surface epithelium , unlike epith elia l ovarian cancer cells, docs not express m R N A o r p ro te in fo r 11.-6, M -C S F , o r T N I* ' (R o b y & fe rra n o v a . 1989: M o tro et o i . 1990) D ysreg ulatcd cytokine production could occur d uring the transform ation o f norm al epithelium lo m alignant cells o r due to oncogene a c tiv a tio n (D e m e iri et a i . 1990). and subsequent c o n s titu tiv e expression o f cytokines and cytokine receptors m ay lead to autocrine g ro w th s tim u la tion o f tum ou r cells o r to m o d u la tion o f their metastatic potential. The presence o f M -C S F and c-fms like receptor tanscripts. p a rticu la rly in p oo rly d iffe re n tia te d o v a ria n tu m o u rs suggest th a t M -C S F m ay be an autocrine gro w th facto r in some ovarian cancers. M -C S F p ro d u c tio n can be increased by g o n a d o tro p in s (B a rto c c i et ol.. 1986). and this cytokine m ay be an im p o rta n t m e d iato r in the p utative tu m o u r p ro m o tin g ro le o f g on ad o trop in s. The p ro te c tiv e effect o f o ra l co ntra ce ptive s c o u ld be m ediated by decreased M -C S F in d u ctio n due to feedback in h ib itio n o f g o n a d o tro p in release R e prod u ctive h o m io n e /c y io k in e in te ra c tio n may also in volve the reported in h ib ito ry effects o f ocNtrogens on T N F secretion (R alston et o i. 1990). Loss o f oestrogen secretion coupled w ith high levels o f g on ad o trop in secretion could perpetuate the in d u ctio n o f T N F in the ovary v ia M -C S F . and e x p la in the increased incidence o f o v a ria n tu m o u rs w ith a d va n cin g age. O f relevance to th is is the observation that a high percentage o f o varian cancers have specific deletions near to. and possibly in volving , the oestro gen re ce pto r lo cus on ch ro m oso m e 6 (Lee e l u i . 1990). C y to k in e s released in the v ic in ity o f the tu n io u r may in du ce tu m o u r c y to k in e p ro d u c tio n . F o r exam ple, T N F has been shown to induce I N F m R N A and protein in epithelial tu m o u rs in v itro and in i*w ( N iils u e t a i.. 1988; M a lik & H a lk w ill. 1990). P ro lon g ed exposure to T N F led n o t o n ly to the d eve lo pm en t o f resistance to T N I -', b u t in du ce d c o n s ti tu tiv e secretion o f T N T by tu m o u r cell lines (Spriggs et u i, 1987). T h is c o u ld be relevant to the o b se rva tio n th a t tu m o u rs fro m the m ore fre qu e ntly o vu la tin g righ t o va ry have a higher rate o f m etastasis at p re sen tatio n (( T u ic k s h a n k . 1990). i.e. these tum ou rs may have acquired the a b ility to produce T N F by m o re fre qu e nt exposure lo T N F released lo c a lly d u rin g the w ound healing process a lte r o vula tio n. A sim ila r phenom enon co u ld expla in the association between the use o f talc and occurrence o f ovarian cancer in some studies, as talc in duces g ra n u lo m a to u s lesions w h ich are a source o f T N F ( K iu d le r et o l.. 1989). A lte rn a tiv e ly , the in v a g in a tio n o f o v a ria n surface e p ith e liu m th a t leads to in c lu s io n cyst fo rm a tio n. m ay lead to the e pithelium being exposed to in tra o varian T N F being produced by macrophages and granulosa cells (R o b v & T e rra n o v a . 1989). In conclusion, epithelial ovarian cancers and ovarian cancer cell lines have been shown to in n a p p ro p ria te ly e lab o r ate a n um ber o f cytokine s. These cytokines can p o te n tia lly perpetuate th e ir o w n b io lo g ica l effects by re cru itin g cyto kin e p ro ducing host cells and in ducing furth e r cytokine p ro d u c tio n Recent experim ental evidence indicates a p ote ntia l role fo r at least tw o c y to kin e s (T N T a nd I L - I ) in the p ro m o tio n o f ovarian cancer metastasis, and tw o others (M -C S F and the lig a n d fo r the c-t7>B-2 p ro to -o n c o g e n e ) as p u ta tiv e a u to crin e g ro w th facto rs in o v a ria n cancer. These cell re g u la to ry m olecules m ay be a c tin g to create a h orm on e m o d u la te d tu m o u r p ro m o tin g e n v iro n m e n t in this disease. I f so. fu tu re therapies fo r o va ria n cancer may involve negating the effects o l these cytokines, either by in h ib itio n o f cytokine secretion, in h ib itio n o f the recruitm ent o f cyto kin e p ro d u cin g cells, o r treatm ent w ith specific cytokine antagonists EPITHELIAL OVARIAN CANC'KR References BALKWII L. f.R (19891. In < yiokine.% in C a n c e r T h e ra p y , Chapter 8. O xford University Press. BALK WILL. I- R A BURKE. F. (1989). The cytokine network. Im m u n o l. T o ila y , 10, 299 BARKKX'I. A . p o n ARI). S3 ANIJEr._E.8-J 1986). Regulation o f colony-stim ulating factor I during pregnancy. J . E x p . M e d ., 164, 956 BOOTH, M . BFRAL. V (1985). The epidemiology o f ovarian cancer: In O v a ria n C a n te r . Hudson. C.N. (cd.). Chapter 12: pp. 22 44. O xford University Press. H O IT A //I. H. GHE77I. I*. IARAHOI I I T I I . G A 5 others (I9H5). Tumor-derived chemotaciic factors Irom huntan ovarian car cinoma. evident for a role in the regulation o f macrophage content o f neoplastic tissues. I n i. J . C a n c e r. 34, 167. BOTTAZ/.I. B COI.LOTTA. I . NOHILI. N . SICA. A A MANTOVANI. A (1990) A chcinoattracant expressed in human sarcoma cells (tumour derived chemotaciic factor, T D C F ) is identical to monocyte ehcmoattracanl protein I/monocyte chcmotactic and activating factor (M C P -l/M C A F ). I n i. J . C a n c e r, 45, 795. BRENNfcR. D A . O'HARA. M . ANGEL. P . CHOJIKbR. M A KARIN. M (1989). Prolonged activation o f ju n and collagenase genes by tum our necrosis factor alpha. N a iu re , 337, 661. CASAGRANDF. J T . LOUIE. F. W . PIKE. M C . ROV. S . ROSS. R K A IIENDFRSON. R.F. (1979) Incessant ovulation and ovarian cancer. I M m e l. II, 170 CRAMER. D W A WELCH. W R (1983). Determinants o f ovarian cancer risk. II. Inferences regarding pathogenesis. J . N a ll C an cer In s t.. 71, 717 CRUICKSHANK. I> J (1990) Actiological importance o f ovulation in epithelial ovarian cancer, a population based study. B r. M e d . J., 301. 524. DAKLINO. ti.. I RAKI R |> | . . CHRISTIAN JENSEN. J . OORSCHBOTH. C M A NORTON. J.A. (1990). Cachectic effects o f recombinant human tum our necrosis factor in rats. C a n c e r R es.. 50, 4008. Dh O RIVE. I . VANDAMME. B , I.ISSFNS. W . BOURGAIN. C . LIFBAERS. I A Hl.RNHETM. J. (1990). Human ovarian cancer cells express multiple proto oncogenes, some at very high level. P rn c . A n n . M e e t. A m . A s s o c . C a n c e r R es., 31, Abst. no. 77. DFMETRI. ( O . IRNS1. I J. PRATT. fc.S . ZF.N/.IE. B W . RHEIN WALD. J.Ci A (K II I-IN. J D. (1990). Expression o f r a t oncogenes in cultured human cells aliers the transcriptional and poftflran- .scriplional regulation ol cytokine genes J. C lin . In v e s t . 86, 1261. DVORAK. HE (1986) Tumours Wounds that do not heal. N e w E n g l. J M e d . 315, 16 SO F.RROI. A . SIRONI. M . CHIAFEARINO. F . 7HEN-G1JO. C . MEN GOZ7I. M A MANTOVANI. A (1989) IL -I and IL-6 release hy tum or associated macrophages from human ovarian carcinoma. I n i J . C a n te r . 44. 795. EATHALA. M R (1971). Incessant ovulalion: a factor in ovarian neoplasia. I m m e t. li. 163. fRATFR-St IIRODER. M . RISAIJ. W . MAI I.MANN. R.. FAUTSOII. P A BOHI.EN. P (1987) T umor necrosis factor type alpha, a potent inhibitor o f endothelial cell growth in v itr o , is angiogenic in v in e P rn c . N a ll A , a d S c i. U S A . R4. 5277. GIAVAZZI. R . (ARO) ALO, A . BAN!. M R A 5 others (1990) Interleukin-1-induced augmentation o f experimental metastases from a human melanoma in nude mice. C a n c e r Res.. 50, 4771. NAJJAR. K A . NAJJAR. D P . SILVtRSTEIN. R I. A NACKMAN. R i (1987) Tumour necrosis factor-media ted release o f platelet derived growth factor Ironi cultured endothelial cells. J . F \p , M e d .. 166, 235 HARANAKA. K . SATOMI. N A SAKURAI. A (1984). A ntitum or activity o f murine tum our necrosis factor (T N K ) against trans planted murine lumours Sc hcieronlranspianted human lumours in nude mice. I n i J C a n te r . 34. 263. HUDZIAK. R M . I.F.WIS. G D . SIIALAHY. M R A 4 others (1988) Amplified expression o f the H F R 2 F .R B 8 2 oncogene induces resistance to tumour necrosis factor a in N IH 3T3 cells. P r t u . N a i l A r a d S t i. U S A . 85. 5102. ITO. A . SATO. T . IGA. f A MORI. Y (1990) Tum our necrosis factor hifunctionally reglales m atrix mclalloprnicinascs and tissue inhibitor o f metalloproteinase* (T IM E ) production hy human fibroblasts F E B S . 269. 93. KACINSKI. H M . CAR IFR. IT. MITTAL. K A 8 others (1990) Ovarian adenocarcinomas express /n/.r-complcmcniary transcripts and fnts antigen, often w ith coexpression o f CSE'-I. A m J P a th o l.. 137. 135. K A C IN S K I. B.M . STANI.EY. E R . CARTER. O A * others (I Circulating levels o f C S K -K M -C S F) a lyntphohaemop. cytokine may he a useful marker o f disease status in patternmalignant ovarian neoplasms. Ini. J. Ratlioiion Oncol B iol / 17, 159 KINDI.ER. V . SAPPINO. A P . (RAU. G F . PKiUF.I. P I * VAS\ \ P. (1989). The inducing role o f tum our necrosis factor in development o f bactericidal granulomas during BCG mfectnfocal autoam plifying process o f tissue organisation. Celt. 56. KISHIMOTO. T. (1989). The biology o f inlcrlcukin-6 . BIihhI. 7-1 K N lll'P . ( . PEKALA. PH A VOKNEI.IUS. p (|9KK| Extensive t marrow necrosis in patients w ith cancer and in in o tii ms factor activity in plasma. Am. J. H aem atol. 29, 215. LF. J A V ll CEK. J. (1987). Tum our necrosis factor and interleuki Cytokines w ith multiple overlapping htological activities I Invest . 56. 234. LEF. J.H.. KAVANAGH. J J. W IID R IC K . D M TAYLOR WIIAK I J. A BUCK. M. (1990). Frequent loss o f heterozygosity chromosomes 6 q. I I . and 17 in human ovarian carcinmi Cancer Res . 50, 2724 LINDEMANN. A.. LUDWIG. W . OSTF.R. W . MERTELSMAN. R . Ii MANN. F. (1989). High level secretion o f tum our necrosis fat alpha contributes to haematopoietic failure in hairy cell Icuk.n Blood. 73. HH0 884 M A l.fK . S.T A . GRIFFIN. U.R.. HERS. W. A HA I KWH.I . J- K (I** Paradoxical effects o f tum our necrosis factor in experun, ovarian cancer. Ini. J. Canter. 44, 918 M A LIK . S.T.A.. NAYLOR. M S , OLIEE. A A BAI KW ILI . I R ( I " Cells secreting tum our necrosis factor show enhanced meia-i in nude mice. Fur. J. Cancer. 26. 1031. MAI.IK. S T A A HALKWII.L. I R (1990) T N F induces T N F niH in human ovarian cancer xenografts in vivo (in preparation MO I RO. B . (TIN. A.. SACHS. I. A K ESI I FT. E. (1990). Pal ten interleukin 6 gene expression in vivo suggests a role lo i cytokine in angiogenesis. Proc. N a il Aeatl. Set. USA. 87, t> NAYLOR. M S . M A L IK . ST A . JOBI.ING. T.. STAMP. G. A H \ W ILL. I; (1990). Demonstration o f tum our necrosis I'ach human ovarian cancer hy in situ hybridisation. Fur. J. Com et 1027. NIITSU. Y . WATANABE. N . NF.DA. II. A 4 other- (1988). Indtu o f synthesis o f tum our necrosis factor in human ami iniimi lines by exogenous rccomhiuaui luini.ni tum our necrosis l.n Cancer Res.. 48, 5407. Ol IFF. A.. DEEEO-JONES. BOYER. M. A 5 others (1987). Turn. secreting human TNF/cachcclin induce cachexia in mice ( ell 555. 0 7 0 1 S. R . GARVIN. A J A COSTA. J (19X0) Advanced ox.i cancers: correlation o f histologic grade w ith response lo lhi and survival. Cancer. 45, 572. PHILIP R A EPSTEIN. L B (1986). Tum our necrosis facto im m unom odulaior anhd mediator o f monocyte cytoloM induced hy itself, 7 -interferon and interleukin-1 Nature, 323. PIVFR. M S (1987). Epidemiology o f ovarian cancer. In (Ji.j Malignancies Diagnostic and Therapeutic Advances. Pivci. N (cd.). pp. I 10. C hurchill Livingstone. POBF.R. JS. (1987). Effects o f tumour necrosis factor and u-l.. cytokines on vascular endothelial cells. Ciha. Found. Svm p . I 17. POSTIIEI. WAITE, A F. A SHYER. JM (1990) Stimulation o f lib, last chcmoiaxis hy human recombinant tum our necrosis fail (T N F-a) and a synthetic T N I'-a 3 l-6 8 peptide. J. F tp M e d . I 1749 RALSTON, S .ll. RUSSELL. R.G.R. A GOWEN. M (1990) OcMr. inhibits release o f tum or necrosis factor fro m peripheral N mononuclear cells in postmenopausal women. J. Bone M Res . 5, 983. RAMAKRISHNAN. S.. XU. F L . BRANDT. S J. NIF.DE1. J.F. . ' R C Jr A BROWN. F..L. (19X9). Constitutive production macrophage-colony stimulating factor hy human bica-i ovarian cancer cell lines. J. Clin. Invest.. 83, 921. RICE. G l i . GIMBOKNE. M A A Bl V|LA( QUA. M E (I9XX). Tun. cell endothelial interactions. Increased adhesion o r hm melanoma cells to activated vascular endothelium. Am J. Pah 133, 204 \ \ \ \ IK A I H \ l k \W I I to k 1 . V I I KK \S i i l V P 1 (I9K9| 1 iv.tliN .iiion o f tumour nCTs|- t.ulor in ihc i.o ilid hivme ovary using im nuinoliisiovlu'llllsl11 V llld cell hlot cv idemx lo i granulosal pr ihI m l mu In (./.>, /J, /.A tor * ,m,l ih. <tw in 1 Inw lifie ld (el 1 pp 27t 27ft PI, mulli I W New V u l III K 11 J1 1 (M l V.1 M 1 M I 'MA It A S H I/1 1 M l . K (1987) P i...Inc I|>>|| .>1 1 1 .1 .it. .it m 1 ll|>ll.llVc I.KlKI and colony* 'liriiu '.i ling'lT ioT I'v .Tv9 1 1 .1 1 mto cell .ircinoiii.i o f the thyr.nd 1 1 i\|.< > dor O\ l Irorn ,i P-H.cn 1 with hvpercakvnii.i and leuko- viox.v < .(. . ! R. . . 47. /474 Ml v| Kill R<>. I BORIS.. i I . MjUIRI S. | s (19901 Cancer sialis- n. v W^u ( . i - \ ( .) d m . 44), 9 m \ m o \ D i i. o i) o i m i s v, m in is . i a ft x others (1989). S(dvlio " I U t fttii proto-oucopciic in iiiim an hica.M and Oi.io.in i.nicei A. I, >u. 244. 7(>7 M,R I(.i,'i I ) . IMAMCKA. K . K( )|>RI< d It-?. p . HORICd'CHI. J A K i l l l> W 1 1*4X71 In.lu d io n o f tumour necrosis (actor expres sion .uni icMNi.mcc in ,i lunn.in hrejst tumour cell line Pro,- A'a il I..../ S.i / \ l. M4. (<() I A M M . I CARDINALE. I . KRUHCifc'R. J . MURPHY. J.S.. MAY. L. A S F iilIAL. PB. (1989). Interleukin 6 decreases cell-cell association and increases m o tility o f ductal breast carcinoma cells. J. Exp. M a t . 170, 1649 WARRJ-N. M K A RALPH. P (1986) Macrophage growth factor ( S I -I stim ulai ex human monocyte pioduclion o f interferon, tumour iiecroMs factor, and c<>lony siin>.ulating. .activ.ij_y.._ J. Iinmunttl., 137, 2281. WATSON. ).M . SENSINTAFFAR. J L . BEREK. J.S. A MARTINEZM A M . O. (1990). Constitutive production o f interleukin- 6 by ovarian cancer cell lines and by primary ovarian tum our cultures. Cancer Res.. 50, 6959. WELANDFR. C E . NATALE. R B A LEWIS. J L. (1982). In vino growth stimulation o f human ovarian cancer cells by xenogeneic peritoneal macrophages. J. N a l l C a n c e r I n s t.. 69, 1039. WELCH. DR.. FABKA. A A NAKAJIMA. M. (1990). Transform ing growth factor fi stimulates mammary adenocarcinoma cell invasion and metastatic potential Paie. N a il AcoJ. Sa. USA. 87. 7678. Bt J Canea (1991). 64. 621 62} G U ES T E D IT O R IA L Liver transplantation for malignancy N.J. London & G.R. Giles U n iv e rs ity D epartm ent o f S urgery. S t James 's U n iv e rs ity H o spita l. Leeds L S 9 7TF. U K . Macmillan Press Lid.. 1991 T u m o u r a b la tio n by resection, when feasible, is the treatm ent o f choice fo r most patients w ith p rim a ry hepato-biliary tum ours and secondary malignancies o f colo-rectal o rigin (Iw a ts u k i Si S tarzl, 1988tf) A s a result o f the use o f tu m o u r m a rke rs in the p o p u la tio n s at risk and an increased use o f accurate non-invasive im aging techniques, im proved detec tio n o f sm a lle r lesions w ith in the p a ren chym a o r h ilu m has o ccurred a nd th is has led to a high e r re se cta b ility rate (O k u d a St Isha k, 1987). a lth o u g h th is ra re ly exceeds 3 0 - 4 0 % (B ism u th e l at., 1986) F u rthe rm ore, a m u ltid iscip lin a ry a pproach, im provem ents in patient selection and manage m ent, and refined techniques in live r surgery, have c o n tri buted both to a low er perioperative m o rta lity and im proved lo n g -te rm s u rv iv a l rates (T a n g , 1985) F o r small p rim a ry hepatocellular carcinom as, fo llo w -u p o f 144 p a tie n ts, w h o und erw e nt resection, show ed a 5-year su r v iv a l o f 6 3 .2 % (T a n g et a l . 1989). T h e best results were fo u n d in p a tie n ts w ith a tu m o u r o f less th a n 2 cm a nd those w ith an encapsulated tu m o u r S im ila r results were fou n d in a F re n ch -lta lia n cooperative study o f 72 patients w ith cirrhosis (F ra n c o et a l . 1990) H o w e ve r, w here the tu m o u r is b ilo b a r o r c e n tra l, where there is vascula r in v a s io n , o r i f the re m a in in g p are n ch ym a is severely co m p ro m ise d fro m c irrh o s is , sub tota l hepaiectom y may not be an o ptio n. In this situation, the m edian post-operative su rviva l is o nly 2 to 4 months. F o r p a tie n ts whose disease is n o t resectable b y co nve n tio na l approaches, a tota l hepaiectom y and subsequent liver replacement m ay o ffer the o n ly true chance fo r survival, although in practice only about one qua rte r o f patients refer red to transplant centres w ill be suitable (Is m a il et al.. 1990). Hepatocellular carcinoma (H C C ) The value o f live r replacement fo r H C C is s till the subject o f much co ntroversy The European L ive r T ransplant registry, w ith results fro m 32 E u rop ea n centres, re p o rte d a 2-year a ctu a ria l su rviva l rate o f 30% in 217 patients w ho underwent liv e r replacem ent fo r H C C (B is m u th e t a l.. 1987). C entres in H annover. K in g 's College/C am bridge and P ittsburgh report sim ila r 3-year su rviva l rates o f a bo ut 20% and the K in g 's C o lle g e /C a m b rid g e g ro u p re p o rt som e lim ite d e a rly success w ith subsequent ch e m o th e ra p y (R in g e et a l.. 1989; O 'G ra d y e t a l.. 1988; K o n e ru . 1988) Iw a ts u k i et al. (1989) have described the p rognostic in d i cators affecting survival and recurrence after transplantation. In a g ro u p o f 80 liv e r re cip ie nts w ith H C C . the tu m o u r size (greater than 5 cm ). m u ltip le nodules, vascular invasion and tu m o u r shape (non-circum scribed) were p o o r prognostic histopathologic;]I factors fo r recurrence in non-ftbrolam ellar H C C . They and others have confirm ed the m ore favourable prognosis associated w ith the fib ro la m e lla r variant o f H C C , where the I-ye ar su rviva l rate was 100% w ith an overall Correspondence: G .R . Giles Received and accepted 3 June I9 9 | recurrence rate o f 5 0 % (O 'G ra d y et a l.. 1988; S tarzl et a l.. 1986). The H annover g ro up re p ort significant differences in median survival tim e fo r different groups classified according to T N M (R in g e et a l. 1989). Stage II (p T 2 pNO pM O) p a tie n ts had a m e d ian s u rv iv a l o f 120 m o n th s co m p are d to 11.8 m o n th s f o r stage I I I ( p T I - 3 p N 0 - l p M O ) and 8 75 m onths fo r stage IV A (p T 4 p N 0 -l pMO). A ll patients g ro u p ed as stage IV B ( p T i- 4 p N 0 - l p M I ) d ied w ith in 2 m o n th s. In th is series p re -o p e ra tiv e serum a lp h a -fe to p ro te in levels d id not have a significant correlation w ith post-operative o u t come. H ow ever tum our-free long-term survivors were noted to have n orm al o r o n ly s lig h tly elevated a lp h a -feto prote in levels b efore liv e r replacem ent. F o r these p a tie n ts, coexistence o f cirrho sis d id n ot influence the lo ng -te rm s u rviva l rale but the 30-day m o rta lity was higher in the c irrh o tic patients How ever, o f the ten patients transplanted fo r coexistent c ir r hosis and H C C in B irm ingham , none survived m ore than I year (Is m a il et a l., 1990). Carcinoma of the biliary tract The prognosis fo llo w in g liv e r replacement fo r tum ou rs o f the b ilia ry tra c t is u n ifo rm ly p o o r. P eriph e ra l c h o lu n g io c a rc in omas appear incurable by this approach ax widespread d is tant metustases developed in all eight patients su rv iv in g m ore then 30 days in th e H a n n o v e r series (R in g e et a l . 1989). S im ila rly six o f seven p a tie n ts in the K in g 's C o lle g c /C a m bridge series deve lo pe d recurrence o f tu m o u r (th e re m a in in g patient died at 4 m onths fo llo w in g a cerebro-vascular acci d en t) (O 'G ra d y et al., 1988). S lig htly m ore encouraging results are fou n d in patients w ith c e n tra l lesions. T h e H a n n o v e r g ro u p re p o rt o n 20 patients and fou n d (he m a jo r influence on s u rviva l tim e was the lym ph node status (R inge et a l., 1989). O f the node negative re cip ie nts, e ig h t o f 13 were a liv e (m e d ia n s u rv iv a l 35 m o n th s ) a nd the o v e ra ll 2-yca r a c tu a ria l su rv iv a l rate was 64.1% . A ll seven patients w ith regional lym ph node metastases, however, had a m uch m ore lim ite d su rviva l span (m e d ia n s u rv iv a l 7 m o n th s). H o w e ve r, o f the 13 liv e r tra n s plants perform ed in K in g 's C ollege/C am bridge fo r central cancers, o n ly seven patients survived 3 m onths and 6 died o f recurrence o f tu m o u r (m edian surviva l 8.5 m onths). The rem aining patient was alive and w ell a l 6.5 years (O 'G ra d y et a l.. 1988). In a series o f n ine p a tie n ts fro m P itts b u rg h , Iw a tsuki reported tha t no patient w ith bile duct cancer had lived 2 years p ost-o p era tive ly (Iw a ts u k i et al.. 19886). Th ere is n o d o u b t th a t accurate stag in g o f the tu m o u r is vita l before tra nsp la nta tion and many centres ro u tine ly per form an e xplo ra tory la pa ro tom y w ith lym ph node sam pling The value o f such a procedure has to be balanced against an increased ris k to the p a tie n t and the disad va nta ge o f in ie rfe r in g at the site o f tra n s p la n t. Some hope fo r these patients can be dra w n fro m reported results o f a senes o f six bile d uct ca rcin o m a s treated by 'abd o m in al clu ster' operations (Starzl el al.. 1989) T h e oper a tio n consists o f the re m o va l o f the liv e r, stom ach , spleen, pancreas, duodenum , p ro xim a l je ju nu m , term in al ileum , and 0 Mi INTOSII KT AL. 1983. Colton T. Statistics in medicine. Boston: Little, Brown anti Company, 1974. Rothman KJ. Boyce JL). Epidemiologic analysis with a programmable calculator. Washington. DO. US C iP a 1979. R ural a TTCKIashoff JD. Grossman M l. Inade quacy o !t he literature on the relationship between drug>. ulcers, and gastrointestinal bleeding. { E di torial.) Gastroenterology 1982;82:373-6. Hichardson OT. Gastric ulcer. In: Sleinenger MM. Kordiran JS. eds. Gastrointestinal disease. Patho physiology. diagnosis anti management. 3rd ed. Philadelphia, PA: WB Saunders Co.. 1983:67293. Domschke S, Domschke W. Gastroduodenal dam age duo to drugs, alcohol and smoking. In: Isenla-rg J I, -Johansson C. eds. Clinics in gastroenter ology. Vol. 13. Philadelphia, PA: WB Saunders Co.. 1984:40.r 30. Boyd EJS, Worinsley KG. Aetiology anil patho genesis of peptic ulcer. In: Berk .IK. ed. Bockus gastroenterology. 4lh ed. Vol 2. Philadelphia, PA: WB Saunders C o, 1985:1013-59. Rolhermich NO. An extended study of indometh acin. 1. Clinical pharmacology. JA M A 1966; 195:531-6. O'Brien W M. Indomethacin: a survey o( clinical trials. Clin Pharmacol Ther 1968;9:94-106. Sun DCH, Roth SI I, M itchell OS, et al. Up|>er gastrointestinal disease in rheumatoid arthritis. Am J Dig l)is 1974;19:405-10. Segre K. Long-term experience with naproxen: open label cohort survey of nearly 900 rheumatoid arth ritis and osteoarthritis patients. Curr Ther Res 1980;28:47-60. Caruso 1, Bianchi Porru (1. Gostroscopic evalua tion of anti-inflam m atory agents. Br Med .) 1960;280:75-8. 32. Neznmis .IE, Lancaster C, Robert A, et al. Indo methacin-induced gastric and duodenal ulcers and intestinal lesions in hamster. (Abstract.) Gastro enterology 1982;82:l 138. _________ 33. Satoh H. Inada I, Hirata T . et al. Indomethacin produces gastric antral ulcers in the refed rat. Gastroenterology 1981;8:719-25. 34. Thompson MR. Indomethacin and perforated duodenal ulcer. (le tte r.) Br Med .1 1980;280:1618. 35. Collier DSl.I, Puin JH . Rectal indomethacin for control o f post operative pain. B r Med J 1983;287:1799-1800. 36. Lanza FGL. Endoscopic studies o f gastric and duodenal injury after the use of ibuprofen. aspirin and other non-steroidal anti-inflammatory agents. Am J Med 1984;77( 1A):19-24. 37. Rover GL, Seek man CE, Welshman IR. Safety profile: fifteen year o f clinical experience with ibuprofen. Am .1 Med 1984;77(1 A):25-34. 38. Graham DY. Sm ith JL, Holmes (SI. et al. Non steroidal anti-inflam m atory effect of sulindac sulfoxide and sulfide on gastric mucosa. Clin Pharmacol T her 1985;38:65-70. 39. Konturek SJ, Oblulowicz W, Sito E, et al. Distri bution o f prostaglandin in gastric and duodenal mucosa of healthy subjects and duodenal ulcer patients: effects of aspirin and paracetamol. Gut 1981;22:283-9. 40. Stolte JB. Gross bleeding from digestive tract: frequency of manifest bleeding in peptic ulcer with regard to duration o f disease and to age of dis eased. Acta Med Scand 1944;116:584-93. 41. PulverUift t ' N. Comments on the incidence and natural history o f gastric and duodenal ulcer. Postgrad Med J 1968;44:597-602. A m e r ic a n .Jo u r n a l tu ' K f i o k m i o i .o g y ('u p y rig h l >ci 198H h y T h e J o h n s H o p k in s U n iv e rs ily S cho o l <!' H y g ie n e a nd P u b lic H e a h h AU rig h ili reserved Voi. l i . I. * t `n n t r d in l I . S A REPRODUCTIVE, GENETIC, AND DIETARY RISK FACTORS FOR OVARIAN CANCER MITSURU M ORI,' IZUM I H ARABUCHI,' HIROTSUCU M IY A K E ,' JOHN T . CASAG R AN DE," B R IA N E. H EN D E R S O N ," a n d R O N A LD K. ROSS' Mori, M. (D e p t of Public Health, Sapporo M edical C ollege, Sapporo 060, Japan), I. H arabu ch i, H. M iy a k e , J. T . C asa g ran d e, B. E. H en derso n, and ft. K. Ross. R eproductive, genetic, and dietary risk factors for ovarian cancer. A m J E pid em iol 1988;128:771-7. A case-control study of 110 women with ovarian epithelial carcinom a and 220 individually ag e-m atched controls w as conducted in Hokkaido, Japan, to identify ovarian cancer risk factors. Both the cases and the m atched controls w ere surveyed either from 1980 to 1981 o r from 1985 to 1986. Ovarian cancer risk was increased in single w om en (p < 0.01), and in w om en with a fam ily history of breast, uterine, or ovarian cancer in a m other or sister ( p < 0.001). Conversely, risk w as decreased in w om en who had experien ced a livebirth ( p < 0.001), an induced abortion (p < 0.05), or who had had perm anent sterilization by tubal ligation ( p < 0.05). Each of the reproductive factors rem ained significant when adjusted for each other using logistic regression analysis. The odds ratio for ovarian cancer decreased significantly with increasing num ber of livebirths (p < 0.001). Furtherm ore, a significant negative association between anovulatory periods and ovarian cancer risk w as noted (p < 0.01). No association was observed for any types of contraceptive m ethods other than tubal ligation, but the prevalence of oral contraceptive use was very low. A significant positive association with daily fish consumption (p < 0.0 5) and a m arginally significant negative association with daily milk consumption ( p = 0.05) w ere also observed. anovulation; diet; neoplastic syndromes, hereditary; ovarian neoplasms; repro duction; retrospective studies The age-adjusted mortality rate of ovar ian cancer has been gradually rising in Ja pan, although it is still among the lowest in the developed countries (1)- Japanese women have a number of different repro ductive experiences and dietary habits com pared with those of other populations. For example, more than a third of Japanese married women have ex|>erieiiced an in duced abortion (2), but less than 5 per cent have ever taken oral contraceptives. Hy contrast, more than 50 per cent of their partners have used a condom for contracep tion (2). The amount of daily meat con- Received for publication July 6 , 1987, and in final form February 16, 1988. ` Department o f Public Health, Sapporo Medical College, South 1 , Went 17, Chuo-ku, Sapporo 060, Japan. (Send reprint request to Dr. M itsuru M ori at this address.) 1 Health Service Division, Sapporo Hospital of.Jap anese Railways Hokkaido, Sapporo, Japan. 3 Department o f Preventive Medicine, University of Southern C alifornia School o f Medicine, Los A n geles, OA. T h ifl aludy was supported in part hy grant 60770429 from the Japuneae M in istry o f Kducal.ion. The inves tigation w h s performed in part under the US Japan Cooperative Cuocer Research Program supported by the US National Cancer Institute and the Japanese Society for Promotion o f Sciences 'H ie authors thank the gynecologists in Hokkaido for the ir cooperation, and in particular Profs. K. Ichinohe o f the U niversity o f Hokkaido. T . Shimizu o f Aaahikawa Medical College, and M. Hashimoto of Sapporo Medical College. M d lll KT Al. sumpljuii per capita is lower in Japan than in any ol ibe Western countries (3). We conducted a case-control study of ovarian cancer in Japan designed to evalu ate reproductive (4-17), genetic (5, 7, 9, 12, 17). and dietary risk factors (17-21), all o f which have been found to be associated with the occurrence of ovarian cancer in Western countries. confirmed primary epithelial ovarian can cer were interviewed at 35 gynecologic wards which cover the entire Hokkaido geo graphic area. I he cases had adenocarcino mas with the following histologia classifi cations: 52 serous, 20 mucinous, three clear cell, four endometrioid, 16 undifferentiated, and 15 unspecified. M a t e r ia ls a n d m e t h o d s Cases Controls Two control series were obtained by matching the cases on year of birth (within All hospitals in Hokkuido which treat three years) and the year of the survey. The ovarian cancer patients were registered for first control group (control group A) com i lie survey. During the period of the survey, prised 110 inpatients from the 37 gyneco ve contacted the chief gynecologist of each logic wards in hospitals in Hokkaido, if these hospitals. When an ovarian cancer mostly from the same wards as the case latient was identified, our interviewer vis- group. The control patients had gyneco ted the hospital within a week. None of logic complaints other than an ovarian tu he doctors in charge or their patients re mor. Forty-two (38 per cent) were admitted eded our request for an interview. The for myoma uteri, 33 (30 per cent) for cer mtients were surveyed from November vical carcinoma, 11 (10 per cent) for various 980 to June 1981 ( n = 45) or from July gynecologic inflammatory disorders, seven 985 to February 1986 (n = 65) with the (6 per cent) for prolapsus uteri, and the se of a uniform questionnaire. All infor- remainder for various other conditions. lation was collected through direct inter The second group (control group B) com lews. The information obtained included prised 110 women who were identified from ie variables shown in table 1. outpatients at the Department of Obstet The 110 patients with pathologically rics and Gynecology, Sapporo Medical Col lege In = 39), from women screened by the T able 1 l ormblcs fur w s m * differences between m n riu n "icer patients and control women, Hokkaido, Japan 1 9 8 0 -1 9 8 6 logit; v a r ia b le s Itody height and weight, Quetelel'a index, blood lype, sihship size. Iiirlh order roily history Papanicolaou test at three anti-cancer as sociations in the cities of Sapporo, Asahikawa, and Kushiro (n = 61), or from women who attended two senile welfare facilities m the cities of Sapporo or Abashiri (n = 10). Women who were diagnosed as having a malignant disorder or ovarian disease had been excluded from this control group. Health siaius of close relatives, occurrence o f can ers among close relatives nstrual profile Oenarclie, menstrual irregularity, dysmenorrhea ut*nopause Fifty-four (49 per cent) were disease-free gyecologically, 51 (46 per cent) had var ious kinds of vaginitis, and the remainder had various benign disorders. i ro d u ciive h is to ry Controls, as with cases, were interviewed Inrital status, pregnancy, parity, lactation, contraepUve usage ary variables) ha t. lish. milk, colfee. alcohol, tobacco er variables $ asidenee. birthplace, sports activity. i,im ,,e iic s 'Kt'. occupational history ^4? in person, and no one who was selected as a control declined our request for an inter view. S tatistical analyses Differences in variables between the cases and the controls were evaluated by RISK FACTORS FOR OVARIAN CANCER 773 the univariate method of Miettinen (22). That is, the maximum likelihood estimate was used for point estimations of odds ra tios, and exact interval estimations of odds ratios were calculated. Multivariate anal yses were performed by the conditional lo gistic regression method (23) to adjust for potentially confounding variables. The paired i test and analysis of variance were used to assess differences between cases and the two control groups for continuous variables. R esults The mean age at interview for the 110 cases was 51.1 years (standard deviation (SD) = 12.2); for the 110 women in control group A it was 50.6 years (SD = 12.0); and for the 110 women in control group B it was 51.1 years (SD = 12.4). Eighty-seven (79 per cent) of the cases were interviewed within a year from the diagnosis. Reproductive variables The average ages at menarche were 14.7 years (SD = 2.2) for cases, 14.8 years (SD 2.1) for control group A, and 14.5 years (SD = 1.7) for control group B. The average age at natural menopause of cases (48.9 years, SD = 3.7 (n = 48)) was younger than that of either control group A (49.2 years, 4JD-- -4.2 (n = 46)) or control group B (49.9 years, SD - 3.6 (n = 55)), but the differ ences were not statistically significant. Seven cases and three controls in group A and five controls in group B had had a hysterectomy. Table 2 shows odds ratios of variables related to marital status and childbearing. The proportion of cases who had never married was significantly larger than that of controls ( p < 0.01). The average age at first marriage of cases (23.6 years, SD = 3.8) was similar to that of both control group A (23.1 years, SD = 3.9) and control group B (23.3 years, SD = 2.8). Analyses for various reproductive char acteristics were restricted to women who had ever married. More cases than controls were nulliparous (p < 0.001). Frequency of miscarriages did not differ between the cases and controls, but significantly fewer cases had had induced abortions (p < 0.05). The average age at first livebirth was sim ilar among cases (24.4 years, SD = 2.8), T able 2 Odds ratios (O R s) (w ith 95 p e r cent confidence in terva ls (C ist) fo r ovarian cancer associated w ith reproductive variables, H o kkaido, Ja p a n , 1980-1986 Reproductive variables Comparative group Cases (%) Comrol Control group A group B (% ) OH* (%> OK* Combined control group A + H OH* 95% C lt Never married Ever pregnant Ever had a livebirth Miscarriage Induced abortion Tubal ligation Condom use by partner Had ever breast-fed Lactation & 6 months per child Anovulatory period & 6 years Anovulatory period > 6 years 8.2 1.8 8.0 1.8 8.0 7.6 1.5-73.4 88.8 94.9 0.4 93.9 0.4 0.44 0 . 1 - 1.2 81.6 93.9 0.3 92.9 0.3 0.24 0. 1 - 0.6 29.6 29.6 1.0 37.8 0.7 0 .8 J 0.5- 1.4 38.8 59.1 0.3 45.9 0.5 0.54 0.3-0.9 9.1 24.5 0.4 18.4 0.4 0.41 0 .1 -0.8 41.8 42.9 1.0 51.0 0.6 0.84 0.4-1.4 86.3 89.0 0.8 90.0 0.7 0.7 0.3-1.9 71.2 67.1 1.3 60.3 1.7 l . t il 0.7 3.4 6.4 20.4 0.1 17.3 0.3 0.2 0.1 -0.7 9.6 26.0 0.1 19.2 0.4 0.3 0.1 0.8 Maximum likelihood estimates o f odds ratio, t 95% exact confidence intervals o f odda ratio. t Baaed on 98 seta o f a caae and two controls excluding never married women. Based on 73 Beta o f a caae and two controls excluding non-parous women. 774 MOW KT AL. KISK FACTORS FOR OVARIAN CANCKR 775 control group A (24.4 years, SD = 3.1), and control group B (24.9 years, SD = 3.2). Table 2 shows that a significantly lower proportion of cases than controls had per manent sterilization by tubal ligation (p < 0.05). The average age at tubal ligation of cases (31.5 years, SD = 4.1) was sim ilar to control group A (31.2 years, SD = 4.3) and control group B (32.3 years, SD = 4.2). T he proportion of cases whose partners used condoms was not significantly different from controls. Only two cases and four controls had ever taken oral contraceptive pills. No difference was observed in the frequency of use of any other types of con traceptive methods. Nulliparous subjects were excluded when comparisons were made concerning breast feeding (table 2). There was no significant difference in either the proportion of women who had ever breast fed or the period of breast feeding per child. As shown in table 3, the odds ratio for ovarian cancer decreased significantly with increasing number of livebirths, either crudely (linear tend, p < 0.001) or after adjusting for marital status and age at first livebirth (adjusted linear trend, p < 0.05). Compared with nulliparous subjects, women with one or two children had a third of the risk of ovarian cancer, women with T a h i.k :i (kid s ratios tO llsl for ovarian cancvr associated w ith n u m b e r o f Iw c b o rn c h ild r e n . H o kka id o . J a fx in , IUHO- Itm ti N<>. <>f liveborn children Cases Control group A Control sroup 11 O 1 2 :\ 4 5+ 29 45 32 9 52 39 10 5 0 31 4 It) 13 Unadjusted O K ( Linear i rend, p < 0 .0 0 1 ) 1.00 0 .1 9 0 .1 7 0 .0 3 A d ju ste d O K * (L in e a r I rend, p < LUO o.:u 0 .2 8 0 .0 5 0 .0 5 ) L * Adjusted for marital status and i birth first liv e - three or four children, a fourth the risk, and women with five or more children, a 20th the risk. T o assess the "anovulatory years" model of Casagrande et al. (5), an anovulatory period (in m onths) was individually calcu lated as follows: anovulatory m onths = (to tal m onths of pregnancies) + (total months of oral contraceptive use) + (total anovu latory m onths caused by lactation). Total m onths of pregnancies in this formula was calculated as follows: (parity * 9) + (total m onths of each abortion, miscarriage, or ectopic pregnancy). T otal anovulatory- ; m onths caused by lactation were calculated by the following formula provided by Risch et al. (14): total anovulatory months by lactation = (1.5 + 0.55 * mean lactation m onths) * (num ber of livebirths). A signif icant negative association between anovu latory periods and ovarian cancer risk was observed whether it was examined in the entire group (p < 0.01) or restricted to subjects who had ever had a livebirth (p< 0.01) (table 2). Conditional logistic regression analysis was performed using th e reproductive vari ables identified as risk factors in the uni variate analyses; they included marital sta tus, num ber of livebirths, experience of abortion, and tubal ligation. T able 4 shows th a t the num ber of livebirths (odds ratio 1 per child = 0.7, p = 0.002), experience of : ever having an abortion (odds ratio = 0.6, p = 0.04), or tubal ligation (odds ratio = 0.5, p = 0.05), all remained significantlyassociated with ovarian cancer risk. Genetic variables t T he occurrence of various cancers among parents and siblings of the cases and con trols was examined. As shown in table 5, there were no differences in the occurrence of cancers in parents and siblings of both sexes combined. Although not significant, more cases reported a cancer history in the m other or a sister. Moreover, a family his tory of breast, uterine, or ovarian cancer in a m other or sister was significantly more T a b i.e 4 A d ju ste d effects v t rep ro d u ctive variables on ovarian cancer, H okkaido, Ja p a n , 1980-1986 Logistic regression Unit Coefficient SR* p value Never married No. o f livebirths Induced abortion Tuba) ligation Yes/No No. Yea/No Yes/No 0.943 0.725 2.6 0.19 -0.365 0.117 0.7 0.002 -0.544 0.270 0.6 0.04 -0.764 0.396 0.5 0.05 * SE, standard error o f logistic regression coefficient, t Two-sided probability. T able 5 Odds ra tio s (O H s) (w ith 95 p e r c e n t co n fid e n c e in te r v a ls (C ls )i fo r o v a r ia n ca n c e r a sso c ia te d w ith g e n e tic fa c to rs, H okkaido. Japan, 1980-1986 Comparative group Genetic factor Cases (%) Control group A Control group it (%) OR* <%) OK* Combined control groups A + B O R * 95% Cl* Cancer in parents or siblings Cancer in a mother or sister Breast, uterine, or ovarian cancer in mother or sister Blood type A Blood type O 34.5 30.9 1.2 38.1 20.9 9.1 3.2 16.4 0.9 1.0 0 .6 - 1.6 1.3 1.9 0.9-3.7 10.9 3.6 3.7 1.8 1 1 .0 5.3 1.6-22.9 45.5 37.3 1.4 34.5 1.6 1.6 0.9-2.5 20.9 30.9 0.7 32.7 0.6 0.6 0.3- 1.0 Maximum likelihood estimates of odd ratio, t 95% exact confidence intervals o f odds ratio. common in the cases (p < 0.001). T hree breast cancers, eight uterine cancers and one ovarian cancer were reported by cases; one breast cancer, three uterine cancers, and no ovarian cancers were reported by control group A; and no breast cancers, two uterine cancers and no ovarian cancers were reported by control group B. It was not specified w hether the site of uterine cancers was the cervix or endom etrium . A larger proportion of cases had blood type A (odds ratio = 1.5), and a smaller proportion had blood type O (odds ratio = 0.6); however, these results were not signif icant. erage consumption, or smoking history. However, the association between daily fish consumption and the risk of ovarian cancer was positive and significant (p < 0.05). Conversely, there was a marginally signifi cant negative association between daily milk consum ption and ovarian cancer (p = 0.05), but a dose-response relation was not observed. The odds ratio of ovarian cancer for Q uetelet's index (weight(kg)/ height(m)*) less than 20 was 1.7, but this difference was not significant. No additional medical history variables investigated in this study revealed signifi cant associations with ovarian cancer. D ietary variables Table 6 shows the distribution of cases and controls by daily consum ption of meat, fish, milk, coffee, alcohol, and smoking prior to diagnosis. T here were no signifi cant differences between cases and controls observed in meat, coffee, and alcoholic bev D is c u s s io n In this study, as with numerous other case-control studies (4-12, 14, 16. 17), (he number of livebirths was found to have a strong negative association with the occur rence of ovarian cancer. We have proposed previously (5) th a t 776 ..... ' - o rtiW A -ii MOKI KT A I.. T a h i.E 6 ( M d s r a tio s (t> H s ) l ii i t f i .95 p e r c e n t c o n f i d e n c e i n t e r v a l s ( C I s J J f u r o v a r i a n c a n c e r a s s o c i a t e d w i t h d ie ta r y variables. H o kka id o , J a p a n , 1 9 H 0 -Ilt8 6 C omparative group 1hi-liirv vttrinlilf --------------------- '--------------------- ( '|>J4*H ( ft group A group H <%> OK* Co) OR* Combined control OR* 9S%Clt Eats meat daily 27.3 20.9 1.4 2 1 .8 1.3 1.4 0.8-2.5 Eats fish daily 63.6 60.9 19 50.9 1.6 1.7 1.0-2.9 Drinks milk daily 35.5 50.0 0.6 46.4 0.7 0.6 0.4-1.0 Drinks < 20 0 ml milk daily 20.9 20.9 1.0 19.1 1.1 1.1 0.6-1.9 Drinks coffee daily 41.H 32.7 1.6 36.4 1.3 1.4 0.8-2.5 Drinks alcoholic beverages at least once a u-eek 21.8 23.6 0.9 19.1 1.2 1.0 0.6-1.9 Smokes daily 14.5 16.4 0.8 10.9 1.3 1.1 0.5-2.3 Quetelet index < 20$ 22.7 17.3 1.5 12.7 2.0 1.7 0.9-3.3 * Maximum likelihood estimates o f odds ratio, t 95% exact confidence intervals o f odds ratio X Weight (kgl/hpighi tin 1'). (incessant) ovulation provides the stimulus for division of epithelial ovarian cells, and th at ovarian cuncer risk is directly related to the frequency of such periods of active cell division. Under this hypothesis, total duration of anovulatory periods would be expected to reduce the risk. T he total of anovulatory periods was shorter among ovarian cancer patients than among the controls (even if the analysis was restricted to parous women). T his finding, in con junction with other reports (S, 11, 13), which show a sim ilar association with the total duration of anovulatory period, would suggest that number of livehirths directly reduces the incidence of ovarian cancer. A tubal ligation was found to be nega tively associated with the occurrence of ovarian cancer, even after controlling for marital status, number of livehirths, and experience of induced abortion. A tubal li gation has been reported to reduce serum levels of gonadotropins (24), and H eintz et al. (25) have suggested that inclusion cysts, produced by incessant ovulation (25-28), may be induced to proliferate and tra n s form by high gonadotropin levoLs. Therelore, it is possible that a tubal lijsilion may indirectly reduce ovarian canfgr^risk by diminishing the opportunity of malignant transform ation of inclusion cysts via n duced gonadotropin stimulation. Family history of breast, uterine, and ovarian cancer was more frequently ob served in cases in th is study. T his result it consistent with the reports of other casecontrol studies (5, 7, 9) as well as with the known familial aggregation of these cancers (29, 30). A ttention m ust be paid to the possibility th a t recall bias may lie an expla nation for this finding. T his seems unlikely, given the site-specificity of the results. We could not confirm the positive asso ciations of either alcohol consumption (17, 21) or coffee consum ption (18) with ovarian cancer, which have been reported previ ously. Studies on the role of dietary fat have been inconclusive; one report showed a significant association (20) while another did not (19). Our study suggests no strong association. T h e marginally significant as sociations of higher fish consumption and less frequent milk consumption observed here require further investigation, but we can propose no mechanism to explain the associations with these specific food items. Despite the markedly dissimilar sources used for identifying control groups A and B, Lhe distribution of most variables of interest was quite sim ilar between the two Z> RISK FACTORS FOK OVAKIAN CANCER 777 control groups and risk factors tended to be the same using either group for compar ison. Since the major sources of both con trol groups were facilities that diagnosed nd treated gynecologic complaints, biased risk estimates might exist for some vari ables, especially the reproductive factors. Population-based or neighborhood controls are difficult to identify in Japan, because a close protection of privacy is still common among Japanese women. The similar re mits obtained using either control group offer some reassurance of the validity of our findings. R eferenc es 1. M uir CS, Nectoux J. Ovarian cancer. Some epi demiological features. W orld Health Stat Rep 1978;31:51-61. 1 Committee on Population Problems. Reports on the 17th N ational Survey o f Planned Parenthood. Tokyo: M ainichi Newspaper Co., 1984. (In Japa nese.) 3 Armstrong B, Doll R. Environm ental factors and cancer incidence and m ortality in different coun tries. w ith special reference to dietary practices. Int J Cancer 1975;15:617-31. 4. Joly DJ, Lilienfeld A M , Diamond E L, et a!. An epidemiologic study of the relationship o f repro ductive experience to cancer o f the ovary. Am J Epidemiol 1974;99:190-209. 5. Casagrande J T , Pike MC, Rose RK, et al. " inces sant ovulation" and ovarian cancer. Lancet 1979;2:170-3. 6. Annegera JF, S trom H, Decker DG, et al. Ovarian cancer. Incidence and caae-control study. Cancer 1979;43:723-9. 7. McGowan L, Parent L, Ledner W, et al. The women at risk for developing ovarian cancer. Gynecol Oncol 1979;7:325-44. 8. Demopouloa RI, Seltzer V, Dubin N, et al. The association o f parity and marital status w ith the development o f ovarian carcinoma. C linical im plications. Obstet Gynecol 1979;54:150-5. 9. Hildreth NG, Kelsey J L . LiV olsi VA, et al. An epidemiologic study o f epithelial carcinoma o f the ovary. Am J Epidemiol 1981;114:398-405. 10. Szamborski J, Czerwinski W, Gadomska H, et al. Case control study o f high-risk factors in ovarian carcinomas. Gynecol Oncol 1981;11:8-16. 11. FrsnceBchi S, La Vecchia C, Helmrich SP, et al. Risk factors for epithelial ovarian cancer in Italy. Am J Epidemiol 1982;115:714-19. 12. Cramer D W , Hutchison GB, Welch WR, et al. Determinants o f ovarian cancer risk. I. Reproduc tive experiences and fam ily history. JN C I 1983;71:711-16. 13. La Vecchia C, Franceschi S, Callus G, et al. In cessant ovulation and ovarian cancer; a critical approach. In t J Epidemiol 1983;12:161-4. 14. Risch HA, Weiss NS, Lyon JL , et al. Events of reproductive life and the incidence o f epithelial ovarian cancer. Am J Epidemiol 1983; 117:128 -39. 15. La Vecchia C, Decarli A, Franceschi S, et al. Age at fira t birth and the risk o f epithelial ovarian cancer. JN C I 1984;73:663-6. 16. Nasca PC, Greenwald P, Chorost S, et al. An epidemiologic case-control study o f ovarian cancer and reproductive factors. Ain J Epidemiol 1984;119:705-13. 17. Tzonou A, Day NE, Trichopoulos I), et al The epidemiology o f ovarian cancer in Greece. A case control study. Eur J Cancer C lin Oncol 1984;20:1045-52. 18. Trichopoulos D, Papapostolou M , Polychronopoulou A. Coffee and ovarian cancer. In t J Cancer 1981;28:691-3. 19. Byers T , Marshall J, Graham S, et al. A casecontrol study o f dietary and nondietary factors in ovarian cancer. JN C I 1983;71:681-6. 20. Cramer DW , Welch WR, Hutchison GB, et al. Dietary animal fat in relation to ovarian cancer riak. Obstet Gynecol 1984;63:833-8. 21. Gwinn M L , Webster LA, ^ee NC, et al. Alcohol consumption and ovarian cancer risk. Am J E pi demiol 1986;123:759-66. 22. M iettinen OS. Estimation o f relative risk from in d iv id u a lly matched series. B iom etrics 1970;26:75-86. 23. Breslow NE, Day NE. Statistical methods in can cer research. Vol. 1. The analysis o f case-control studies. 1ARC scientific publication no. 32. Lyon: 1ARC, 1980. 24. Alvarez-Sanchez F, Segal SJ, Brache V, et al. Pituitary-ovarian function after tubal ligation. Fertil Steril 1981;36:606-9. 25. Heintz A P I', Hacker NF, Lagasae LD . Epidem iol ogy and etiology o f ovarian cancer. A review. Obstet Gynecol 1985;66:127 -35. 26. Fathalla M F. Incessant ovulation. A factor in ovarian neoplasia? (le tte r.) Lancet 1971;2:163. 27. Radiaavljevic SV. The pathogenesis o f ovarian inclusion cysts and cystoma. Obtet Gynecol 1977;49:424-9. 28. Zajicek J. Prevention o f ovariun cystoma by in h i bition o f ovulation. A new concept. J Reprod Med 1978;20:114. 29. Lynch H T , Guirgis HA, AH>ert S. et al. Familial association o f carcinoma o f the breast and ovary. Surg Obstet Gynecol 1978; 138:717 24. 30. Lynch H T , Harris RE, Guirgis HA, et al. Familial association o f breast/ovarian carcinoma. Cancer 1978;41:1543-9. Jpn. i. Cancer Res. (Gann), 79, 997- 1004 ; Scplemher. 1988 Dietary and Other Risk Factors of Ovarian Cancer among Elderly Women M itsuru M ori and Hirotsugu M iyakl D e p a r t m e n t o f P u b l ic H e a l th . S a p p o r o M e d i c a l C o lle g e . S o u t h 1. W e n t 17, C h u o - k u . S a p p o r o 0 6 0 The age-specific mortality rate of ovarian cancer is increasing among women over 50 years of age, but remaining at a stable level among women under 50. This case-control study of ovarian cancer was undertaken to assess the environmental factors which may increase the mortality of the disease in the elderly. Fifty-six women with primary epithelial ovarian cancer whose ages were 50 years old or over were compared with two age-matched control groups. The results of the Mantel-llaenszel analysis were as follows. More cases were found to have never married than controls (P < 0.051, a larger proportion of cases were nulliparous (/*<0.05), a smaller proportion of cases had experienced an induced abortion (/*<0.05) or had undergone permanent sterilization by tubal ligation </*<0.05), the occurrence of breast or uterine cancer was more common in the mother or sisters of cases (P < 0 .0 l), a larger percentage o f cases used to eat meat daily (/'<0.01> or used to eat fish daily </` <0.Q5), and a larger proportion of cases weighed under 40 kg (J<0.05). Daily meat consumption was significantly associated with the occurrence of ovarian cancer even after adjusting for reproductive and other risk factors by conditional logistic regression analysis. The attributable risk for ovarian cancer in the elderly was 19.2% for daily meat consumption. The recent change in dietary habits might in part explain the rise of the mortality rate among the Japanese elderly. Key words: Ovarian neoplasms Retrospective study Risk factor Diet The age-standardized m o rta lity rale o f ovarian cancer is gradually rising in Japan. The age-specific m o rta lity rate o f the disease is increasing year after year among women over 50, but this rate remains stable among women under 50 (F ig. I) . T his divergence could not be explained only by the difference in the histological origins o f p rim ary ovarian cancer in the different age groups. We suspected that some environm ental factors m ight contribute to the increased incidence o f ovarian cancer among the elderly in Japan. We have conducted a case-control study o f ovarian cancer in H okkaido, and reported the reproductive, genetic, and dietary risk factors when all age groups were com bined and analyzed.0 In the present paper, we restrict the analyses to the ovarian cancer patients who were diagnosed at the age o f 50 or above in order to better assess the dietary risk factors w hich m ight explain the recent in crease o f the disease among the elderly in Japan. M a ter ia ls a n d M ethods One hundred and ten cases with pathologically conArmed primary epithelial ovarian cancer were 79(9) 1988 interviewed at 35 gynecological wards, covering the entire Hokkaido geographical area The cases were surveyed using a uniform questionnaire from November, 1980, to June, 1981 (N 45), or from July, 1985 to February, 1986 (N 65). because the main author had gone abroad from 1982 to 1984 A ll information was collected through direct inter views. The information obtained included the vari ables shown in Table 1. The questions asked about dietary habits and consumption o f tobacco and alcoholic beverages are presented in Table II. The analyses were restricted to 56 cases (50.9%) o f 110 ovarian cancer patients, who were diag nosed at or over 50 years o f age. Histologically, all the cancers were adenocarcinomas o f common epithelial origin; 27 serous, 11 mucinous. I clear cell, 1 endometrioid, 8 undifferentiated, and 8 unspecified. Cases were compared to two control series, indi vidually matched with cases by age (w ithin 3 years) and the year o f the survey. The lirst control series (control group A ) consisted o f 56 in-patients from the gynecologic (N - 53), ophthalmologic (N = 2), and dermatologic wards (N 1). Most of control group A were selected from the hospitals where the cases were identified and interviewed, except for 4 women who were selected from two other hospitals. Their primary diagnoses were 25 cervical carcinoma, 12 myoma of uterus, 7 \ 997 M MOKI ANO II MI YAKb Fig. I. The age-specific mor tality rate o f ovarian cancer per 1 ,000.000 population from 1971 to 1983 in Japan ' . 0 9 years; i i, 10 19 years; . 20 29 yean; . 30 39 years; . 40 49 years; 50 59 years; . 60-69 years, a , 70 79 years; *, 80 years. (Data source: Vital Statistics, Japan. Volume 3). 1971 1973 1975 1977 1979 1981 1983 ' Table I. Variables Surveyed Biological factors Age, blood group, height and weight Family history Mothers's parity, cause of death of close relatives Menstrua] profile Menarchc, general menstrual, difficulties with menstruation, premenopausal symptoms, menopause Marital status Reproductive history Gravidity, parity, lactation Contraceptive usage Medical history Chronic disease, surgery, X-ray examination, hormonal treatment Social factors Address, occupation, tobacco and alcohol, cosmetic usage, food intake 998 Jpn. J. Cancer Kcs t j t M B kM M M kd DIETARY RISK FACTORS OF OVARIAN CANCER Table II. Questions about Dietary Habits and Consumption o f Tohacco and Alcoholic Beverages Could you lell me your dietary habits and consumption o f tobacco and alcoholic beverages in several years before you had been notified o f the disease ? A . Did you eal meal almost every day 7 I. yes 2. no B Did you eat fish almost every day ? 1. yes 2. no C . Did you drink milk almost every day 7 1. yes ( ) ml/day 2. no D . Did you drink coffee or tea almost every day 7 l.cofTee 2. lea 3 neither E Did you smoke every day ? 1. yes ( ) cigarettes/day 2. no F . Did you drink alcoholic beverages at least once a week? 1. yes What kind o f beverages ? ( ) 2 no gynecological inflammation, 7 prolapsis o f uterus, 2 cataract, I cytocele, 1 tumor o f palate, and I tumor of vagina. The second comparative group (control group B) consisted o f 56 women, who were identified from among women screened by means of Papanicolou's test at three Anti-Cancer Associa tions in Sapporo. Asahikawa, and Kushiro cities (N -4 2 ); from among outpatients at the Depart ment o f Obstetrics and Gynecology, Sapporo Med ical College (N 9); or from among women who attended two Senior Citizen Welfare Facilities in Sapporo or Abashiri cities (N = 5). We thus elected control group B from different institutions fa Hokkaido. As a result, 34, 8, and 14 women o f control group B were from Sapporo and its adja cent areas, northern Hokkaido, and eastern or southern Hokkaido, respectively, as compared with 33, 10, and 13 women o f the case group who were from the respective areas. The disease status o f control group B was 29 with no gynecological disorder, 26 with gynecological inflammation, and 1 with cervical polyp. The controls, like the cases, were interviewed in person. A ll cases and most controls were inter viewed by two medical doctors who had been trained for interviewing, except for 8 women o f con trol group B who were interviewed by public nurses. Differences in variables between the cases and the controls were evaluated by the univariate method of Mantel and Haenszel." Multivariate analytes were performed by the conditional logistic regression method" to adjust for potentially conbunding variables. Population attributable risks o f be main risk factors were calculated by the ; method o f Fleiss." R esults The mean age at the tim e o f the interview ^forthe 56 cases was 60.2 years (S D = 7 .6 ); for 1 *( ) 1988 the 56 women in c o n tro l group A it was 59.6 (S D --7.6); fo r the 56 women in co n tro l group B it was 60.8 years (S D 7.6). The mean age at diagnosis fo r the cases was 59.7 years (S D = 7.2), and 48 cases (8 5 .7 % ) were interview ed w ith in a year after the diagnosis. R eproductive C haracteristics Table 111 shows the odds ratios (the estimated relative risks) o f variables related to m a rita l status and child-bearing. The proportion o f cases w ho had never m arried was significantly larger than that o f the con trols (P < 0 .0 5 ). Analyses fo r various reproductive characteristics were restricted to women who had married. M ore cases than controls were nulliparous (the odds ratio, O R --3.75, P < 0.05). Frequency o f m is carriages d id not d iffe r between the cases and the controls, but significantly fewer cases had experienced an induced abortion (O R --0.39, F < 0 .0 5 ). Table I I I shows that a significantly low er proportion o f cases had had permanent sterilization by tubal ligation than controls (O R = 0.39, P < 0.05). The average age at the tim e o f tubal ligation fo r the cases (33.2 years, S D = 4 .0 ) was sim ilar to that o f control group A (33.0 years, SD = 4.7) and co n tro l group B (33.1 years, SD --4.7). There were no sig nificant differences between the cases and the control groups w ith regard to other contra ceptive methods. The prevalence o f taking ora l contraceptives was rare in both the cases ( N = 1) and the con trols ( N -- 1). Genetic Factors The occurrence o f various cancers among parents and sibling o f the cases and the controls was examined. As shown in Table IV , there were no differences in the occurrence o f cancers in parents and siblings \ M MORI AND H M IYAKE Table III Reproductive Factors Cux Control Item group group A Never married Never pregnant*1 Never had a live birth*' Live births ' 3*' Miscarriage*' Induced abortion*' Tubal ligation*' Condom use*' 3.6% 93 14 8 500 25.9 31 5 III 25 9 00% 56 56 61.1 29 6 500 22 2 24 1 a ) P 005 b ) Analyzed 54 iriplcs excluding never-married women. OK. odds raiio. Compurvlivc group Com ml OR group U OR 1 33 2.67 0 50 0.83 0 38 0 46 0 91 00% 1.9 37 630 35 2 48 1 25 9 29.6 7.00 0 65 0.67 0 40 033 082 Control A It OR 3.00 3.75" 0 59 0.74 O 39" 0.39" 0.79 Table IV Genetic Factors Case Control licm 8roup group A Cancer in parent or sibling Cancer in mother or sister 37 5% 28.6 37.5% 10.7 Breast or uterine cuncer in mother or sister 10.7 / , / 1.8 a ) P 005 b ) P 001 OR odds ratio. Comparative group Control OR group B OR 1.00 44 6% 0.77 Control A i b OR 085 4 33"* 214 1 44 2.17 6.00 18 II 00" when a combined analysis was performed. A llhough nol significant, more cases had reported a cancer history in a m other o r sister Moreover, a fam ily history o f breast o r ute r ine cancer among a m other o r sister was significantly more common in the cases than in the controls (O R 11.0, P- 0 01) T w o breast and 5 uterine cancers were reported by families o f cases, 0 breast and I uterine cancer by families o f control group A ; and 0 breast and I uterine cancer by fam ilies o f control group B It was not specified whether the site o f uterine cancer was the ce rv i* or the en d o m e triu m . D ietary Factors Table V shows the d is trib u tion in cases and controls o f the daily con sumption o f meal, fish, m ilk, coffee, alcohol, and smoking prio r to diagnosis. There were no significant differences observed in m ilk, coffee, and alcoholic beverage consum ption, 10(10 or even in the smoking history. However, (he positive association between daily meal con sum ption and (he risk o f ovarian cancer wu found to be significant (O R 3.11, P 0.01). Likewise, the positive relation o f daily fish consum ption to the disease was significant (O R 2.54, P 0.05). A significantly larger proportion o f cases weighed under 40 kg than that o f the controls (O R 3.50, P < 0.05) Table V I indicates the correlation matria o f food consum ption frequency among the women surveyed. There was a significant pos itive relationship o f d a ily meat consumption to daily fish consum ption (Pearson's correla tion coefficient r 0.25, P * 0.01). A positive correlation was also observed between daily coffee d rin k in g and daily m ilk drin king (r= 0.22, P < 0.01), as well as between daily coffee d rin k in g and weekly alcoholic beverage con sum ption ( r 0.28, P < 0.001). Jpn. J. C a n c e r K c v (Gann) * > s l * i d f c J i s i __* _ A DIETAR Y RISK I ACTORS OF O V A R IA N CANCER Table V Dietary Factors Case lic in group Ealing meal daily Ealing fish daily Drinking milk daily Daily milk i: 200 ml Drinking coffee daily Drinking alcoholic bever- ages al least once a week Smoking daily Body weight < 40 kg 28 6% 73 2 37.5 8.9 23.2 16.1 16.1 12.7 o) P< 0.05. b ) P< 0 01 OR. odds ratio. Control group A 8 9% 60.7 53.6 17.9 26 8 14.3 16.1 3.6 Comparative group Control OR group t OR 4.67" 2.40 0.50 0.50 0 82 1.17 14.3% 500 500 19.6 32 1 12 5 2 33 2 63" 0.65 0.40 0 58 1 29 Control A i H OR 3.11** 2.54" 0.58 0 45 0.70 1.23 1 0 0 5 4 3.00 1 67 3.50 3.6 3 50 3.50" Table VI. Correlation Matrix of Dietary Habits Item l. Eating meat daily 2. Eating fish daily 3. Drinking milk daily 4. Drinking coffee daily 5. Drinking alcohol weekly 6 . Smoking daily 1. 1.00 0.25" 0.04 0.07 0.08 0.16 2. 3. 1.00 0.02 1.00 0.02 0 .2 2 " 0.03 0.09 0.03 -0.03 o) P < 0 0 l 6) !><0.00>. 4. 1.00 0.28" 0.05 5. 1.00 0 15 6. 100 Adjustment of the Dietary Factors To adjust the possible confounding variables wilh each other, a conditional logistic regression analy sis was performed. Six items included in the multivariate formula were all significantly associated wilh ihe disease according to uni variate analyses; number of live births, expe rience of induced abortion, status of tubal ligation, familial history of breast or uterine cancer, daily meat consumption, and a body weight of less than 40 kg. Since daily fish consumption was significantly correlated with daily meat consumption, daily meat consump tion and daily fish consumption could not be simultaneously involved In the regression model." 7 9(9| 1988 Table V II shows Ihe result of ihe logistic regression analysis when daily meal consump tion was included in the model. The associa tion between daily meat consumption and ovarian cancer was still significant ( / >< 0 05) even after adjusting for the other 5 variables, although Ihe estimated relative risk was slightly reduced to 2.7 from 3.1 in the uni variate analysis. Population Attributable Risk of the Main Risk Factors The population attributable risks were calculated for three factors, all of which were found to be significantly associated with the disease by multivariate analysis. As shown in Table V III, the popula tion attributable risk for ovarian cancer \ ' 1001 M MOKI AND tl MIYAKE Table V II Adjusted Risk Factors 1 ogislic regression Relative Item Unit coefficients { ) SE </U risk Number o f live births Induced abortion Tubal ligation Breast or uterine cancer in mother or sister Eating meal daily Body weight 40 kg Number Yes/No Ye*/No Yes/No Ycs/No Yes/No 0.34 0 74 0 69 2 55 0 99 0.03 0.15 047 0.56 1 19 0 49 0.02 0.7" 05 05 1 2 .8" 2 7" 10 u) P 0.05. SK (/): standard error of logisi ic regression coefficient. CT confidence intervals. W , Cl 0.5 0 9 0 8 5.3 0 2 15 1.1 132 8 10 70 0 9 1.0 Table V III Population Attributable Risk of Risk Factors Risk factor Populations attributable risk 95% confidence intervals Never had a live-born child Breast or uterine cancer in mother or sister Eating meat daily 14 0 % 9.1 0 2 24 4% 0 0 17 3 19 2 3 4 32 4 / / among the elderly women was 14.0% fo r never having a live-born child, 9 1% for fa m ilial history o f breast or uterine cancer, and 19.2% for daily meat consumption D is c u s s io n A few case-control studies and cohort studies have provided data on dietary factors in ovarian cancer A ccording to the result o f a case-control study conducted by La Vecchia el a lV in Italy, women w ith ovarian cancer had a significantly elevated frequency o f con sumption o f meal and higher subjective scores o f fat intake The median age o f the ovarian cancer cases was 55 years in their study. Sim ilarly, Cram er el a/.' ' indicated that there was a significant trend fo r increasing risk in ovar ian cancer w ith increasing anim al fat co n sumption. The average age o f the cases was 53.2 years in their study. On the contrary, Byers el a l " reported that there was no sig nificant risk o f ovarian cancer associated w ith the consum ption o f fat 1002 H iraya m a'' reported a significantly positive association between daily meal consumption and ovarian cancer as a result o f a 13-year cohort study o f 142,857 women aged 40 years o r over in Japan Snowdson10' noted the pre lim in a ry result o f a cohort study o f 16,190 persons that women w ho ale fried food five or more days per week had a significantly greater risk (relative risk 2.9). The analysis showed a m arginally sig nificant association w ith daily fish consump tion but no relationship w ith meat consump tion when all age groups o f ovarian cancer patients were com bined 11 However, when the analysis was restricted to the ovarian cancer patients w ho were diagnosed al o r over 50 years o f age, there was a clear positive associ ation o f the disease w ith daily meat consump tion as w ell as w ith daily fish consumpiion. This significant association was noied inde pendently o f nondielary risk factors such as decreased number o f live births and family history o f breast o r uterine cancer, as found by the logistic regression analysis. It is not entirely understood how dietary factors such as high meat consumpiion and high anim al fat intake increase the risk o f the disease. C ram er el al.' 1 have proposed that elevated gonadotropin levels brought about by ingesting high levels o f anim al fal might in crease the ovarian cancer risk. However, this hypothesis is not consistent w ith several sero logical studies o f the dietary-fat effect on gonadotropins. F o r instance. A rm strong ri a l.'" indicated lh a l postmenopausal vegetar ian (lo w -fa t d ie t) women did not have differ ent plasma gonadotropin levels when com pared to non-vegelarian women (high-fat Jpn. J Cancer Rev (Gann) i i * * * DIETAR Y RISK FACTORS OF O VA RIA N CANCER d id ) Furtherm ore, H ill et a l. '" reported that in postmenopausal black women in South A frica a Western diet (high-fat diet) de creased both plasma lute in izing and folliclestim ulating hormones as compared to a lowfat diet. A lternatively, Cramer and W elch'" have postulated that in view o f the presence o f estrogen receptors in epithelial ovarian tumors, diets high in animal fat could pro mote tum or development by their purported a b ility to cause an increase o f extragenital estrogen. T his hypothesis is consistent w ith the serological study w hich has shown that postmenopausal vegetarian women have lower u rin ary estrogen levels in com parison w ith matched nonvegelarian w om e n."' F u r ther investigations are needed to c la rify the mechanism o f the relationship o f dietary factors to ovarian cancer by changing the hormonal milieu. In numerous other case-control studies, like this study, the number o f live births was found to have a strong negative association w ith the occurrence o f ovarian cancer. The references and discussions relevant to this finding appeared in our previous re p o rt." Family history o f breast or uterine cancer was more frequently observed in cases in this study. T his result is consistent w ith the reports o f a case-control study by M cG owan el o/.14' as well as w ith the know n fam ilia l aggregation o f these cancers."' I6) A bout 45% o f ovarian cancers among the elderly women in o u r study was a ttribute d to three risk factors: never having a live-bom child (1 4.0% ), fam ily history o f breast or uterine cancer (9 .1 % ), and the dietary factor o f daily meat consumption (1 9.2% ). In regard to this result, the recent rapid change o f dietary habits among Japanese m ight partly explain the rise o f ovarian cancer m o rta lity . A decreasing trend in the number o f births may also account in part fo r the increasing m o r tality o f the disease. A lth o u g h we asked the patients about the ir dietary history before they had been notified o f the diseases, recall bias m ight exist. F u r thermore. the results concerning dietary his tory should be ca re fu lly interpreted, because the latent period between dietary exposure and the occurrence o f ovarian cancer has not yet been clarified. 79(9) 1988 Since the m ajor sources o f both c o n tro l groups were facilities diagnosing and treating gynecologic com plaints rather than the gen eral population, bias m ight exist. H owever, the sim ila rity o f the results obtained using the tw o c o n tro l groups offers some reassurance as to the va lid ity o f our findings. A cknowledgments This study was supported in pari by a Grant-inAid from the Japanese Ministry of Education, Sci ence and Culture. The authors thank the numerous gynecologists in Hokkaido who participated in this study, in particular Professor K. Ichinohe o f Hokkaido University, Professor T. Shiiui/u o f Asahikawa Medical College, and Professor M Hashimoto of Sapporo Medical College. (Received April 6. 1988/Accepted July IJ. IV88) R eferences 1) Mori, M., Harabuchi, I., Miyake, II.. Casagrandc, J. T , Henderson, B. E. and Ross, R. K Reproductive, genetic, and dietary risk factors for ovarian cancer Am. J. Epidemiol., in press. 2) Kletnbaum, D. G., Kupper, I.. L. and Morgenstern, H "Epidemiologic Research. Principles and Quantitative Methods," pp 377-402 (1982). Life Learning Publications, Belmont 3) Breslow, N E. and Day. N E. "Statistical Methods in Cancer Research. Volume I The Analysis o f Case-control Studies," IAR C Scientific Publications No 32. pp 248 279 (1980) . International Agency for Research on Cancer, Lyon. 4) Fleiss, J. L. "Statistical Methods for Rales and Proportions," 2nd Ed., pp. 93 95 (1981) . John Wiley and Sons, Inc., New York. 5) Johnson, R. A. and Wichern, D. W. "Applied Multivariate Statistical Analysis." pp. 291-358 (1982). Prentice-Hall, Inc , Englewood Cliffs. 6) La Vecchia, C., Decarli, A.. Negri, E., Parazzini, F., Gentile, A , Cecchitti, G., Fasoli, M. and Franceschi, S. Dietary factors and the risk o f epithelial ovarian cancer. J. Natl. Cancer Inst.. 79, 663-669 (1987). 7) Cramer, D W , Welch, W. R . Hutchison, G B., Willett, W. and Scully, R. E. Dietary animal fat in relation to ovarian cancer Obstei. Gynecol., 63, 833-838 (1984). \ IOIIJ  A . VW i i A m * M MORI AND II MIYAKE Byers. T , Marshall. J . Graham. S , Meillin, C and Swanson, M A ease-conlrol study of dietary and nondielary factors in ovarian cancer. J. Nall. Cancer Inst . 71, 681 686 (1983). 9) Hirayama. T An analytical epidemiology of ovarian cancer and possible environmental carcinogenesis. Med. Chugai. 34, 282 286 (1981) (in Japanese) 10) Snowdson, D. A. Diet and ovarian cancer. J Am. Med Assoc.. 254, 356-357 (1985). I I ) Armstrong. B. K . Brown. J B.. Clarke. H. T . Crooke, D K , Hhnel. R , Masarei, J R and Raiajczak, T. Diet and reproductive hormones A study o f vegetarian and non vegetarian postmenopausal women. J. Nall. Cancer Insi.. 67, 761 767 (1981). 12) Hill. P . Garbaczewski, L , Hclman, P . Husktsson, J , Sporangisa. E and Wynder, E L Diet, lifestyle, and menstrual activity. Am. J. Clin. N u ir. 3 3, 1192 1198 (1980). 13) C ra m e r, D W and W e lch , W R n e u n nants o f o va ria n ca nce r risk. II Infcrcnca re g a rd in g p a th o g e n e sis. J. Nall. Cancer Isii. 71, 717 721 (1 9 8 3 ) . 1 4) McGowan, L , Parent, L., l.edner, W u A Norris, H J The women at risk lor develop ing ovarian cancer. Gynecol. Oncol.. 7, 325 344 (1 9 79 ), 15) Lynch. H T , Guirgis. H A . Albert, L, Brennan, M , Lynch, J., Kraft. C., Pocckzy, D . Vaughns, C and Kaplan, A. FamiW association o f carcinoma o f the breast a*4 ovary. Surg. Gynecol. Obstei.. 138, 717 724 (1 9 7 4 ). 16) Lynch, H T . Harris. R E . Guirgis, H A . Maloney. K , Carmody, 1. L and Lynch, J F Familial association o f breast/ovinaa c a rc in o m a . Cancer. 4 1 , 1543 1549 (1971), ' a. J. Cancer Res (Gann), 79, 1005-1016 ; September, 1988 A Comparative Epidemiologic Study on Geographic Distributions of Cancers of the Lug and the Large Intestine in Japan R''.' Motoi Murata, * 1Kimiko Takayama,* 1Seigo Fukuma,* 1Naoyuki Okamoto,*' Ikuko Kato,** Aya Hanai,*5 Hideaki Nakayama,*6 Kouichi Fujiwara,*7Takayoshi Ikeda" and iuburo Fujimoto** Division o f Epidemiology and * 2Emeritus President. Chiba Cancer Center. Nnona 666-2. Chiba 280. Department o f Epidemiology. Clinical Research Institute. Kanagawa Cancer Center. Yokohama 241. y*4Division o f Epidemiology. Aichi Cancer Center Research Institute. Nagoya 464. * 3Department o f Field 5tmearch, Osaka Centerfo r Adult Diseases, Osaka 535. * tDepartmeni o f Hygiene, Toltori University. Tottori MIL Ministry o f Health and Environment. Saga Prefecture, Saga 840 and * JDepartment o f Pathology. Nagasaki University School o f Medicine. Nagasaki 852 To examine what kinds of factora could have caused the geographic variation observed in lung cancer morbidity in Japan, a correlation study was performed comparing various regional traits. The same study was also conducted on large intestinal cancer, aiming to distinguish the possible urbaa factors associated with both cancers. Lung cancer was highly correlated with industrlalizaboa-related factors such as localization of manufacturing industries, automobile traffic and air pollution, whereas colon cancer was correlated with the population density of workers in the tertiary Industries such as services, trade and government. A multiple regression analysis could not detect any single factor with an exceptionally strong influence on either cancer. The present ladings suggest that the hazardous environmental condition of urban areas has, to some extent, toatributed to the recent Increase of lung cancer cases in this country. Key words: Lung cancer Colon cancer - Cancer registry -- Geographic correlation A ir pollution Jpn. J Cancer Res (Ossa) U Japan lung cancer has been rapidly in ' m a tin g in recent years. The age-adjusted mortality rates were 27.4 in males and 7.7 in tics per 10! population in 1985, nearly five la n d fou r times larger, respectively, than e in 1955. It is generally suspected that Ifct growing tobacco consum ption in these i has made a significant co n trib u tio n to tb t past increase o f lung ca n ce r." However, i incidence rate also shows a notable geo . graphic variation w hich m ainly corresponds . la the urban-rural difference. I f a ll geographic ' dbtrict* are classified by the ir population size, lung cancer m o rta lity is highest in the |ts t cities and lowest in the ru ral areas. . t t k urban-rural difference may be. as has ` |a m peculated fro m epidem iologic findings Mported in other co u n trie s," ascnbable to ntial tobacco consum ption rales. Neverk it is quite im probable that other kinds i t f hazardous factors in the urban environacM have nothing to do w ith this disease. fcveral authors in other countries have I that, even a fte r age and sm oking rate , m r) D were standardized, a significant effect o f resi dence s till appeared in (hegeographic variation o f lung cancer m o rta lity .' " The effect o f res idence could be due lo cither some feaiure(s) o f urban life style or environmental pollution. R egarding ihe association o f these latte r factors w ith lung cancer, many epidemiologic studies' have been carried out and provided apparently positive but inconclusive evidence because other factors such as occupation and smoking rate were not thoroughly stan dardized.11 111 The present study attempted to obtain some clues concerning the geographic correlation between lung cancer m o rb id ity and those factors inherent to urbanization, especially environm ental pollution or localization o f its sources. In general, this type o f correlatio n study is rather in fe rio r to the more advanced analytical epidemiologic studies, because it can only demonstrate a correlatio n between the disease and regional traits. S till it should be useful i f the factors to be studied arc o f area-specific rather than person-specific kinds. 324 BER M A N ET A L . lions of solutions will provide maximal nilrogen sparing and caloric supplementa tion. CONCLUSION Some patients who currently are being supported with central venous TPN could benefit from I'NI'V with little attendant morbidity. Similarly, many patients who might benefit from protein sparing therapy and caloric supplementation who do not receive TPN because of risk ol complications, cost, or complexity of the monitoring and delivery system are good candidates for PNPV. Even those patients who appear to be at low risk for postoperative morbidity might benefit from the infusion of amino acid solutions alone because of their nitrogen sparing properties. REFERENCES 1 Dudiick.S J . 1 .01142. J. M.. Steiger. H .a n d Rhoades. J. E. Intravenous hyperalimentation.Med. C lin . \ o . A m e r . 54. 577 <1970). 2 McKIrum. 1) R . Dhopcshwarkar. (i. A . Lin. S..and Smith. M. 1. Essential fully acid deficiency with long la in hyperalimentation. G y n e c o l. O n c o l . 4, 66 (1976). V Aguirre. A.. Fischer. J E . and Welch. C. E. The role o f surgery and hyperalimentation in therapy of gasiiointestinal-culaneoiis fistulae. A n n . S in -ft. 180, 393 (1974). 4 Schwartz. Ci. R .G io e n . II. I... Hendon. M. L.. Graham. W. P. .and Blakemore. W. S. Combined paienici.il h\pet alimentation and chemotherapy in the treatment of disseminated solid tumors. A m , 1 J S i n .e 121. 169(1971). 5. Ford. J. II.. Mtidan. R. C . . Bennett. J. S.. and Averette. H. E. Parenteral hyperalimentation in gynecologic oncology patients. G y n e c o l. O n c o l I, 70(1972). Ryan. J. A. Complications of total parenteral nutrition, in T o t a l p a r e n t e r a l n u t r i t i o n (J E. Fischer. Ed.), l ittle. Brown. Boston, pp 55-100(1976). 7 Coian. A (. I otal intiaxenotts feeding of infants and children without the use of a central venous cnlhclci. A n n . S u m 179. 445 (1974). X Fleck. A., and Munro. II N. Protein metabolism after injury. M e t a b o l i s m 12, 783 (1963). 9. Moore. F. I). Convalescence: The metabolic sequence after injury, in M a n u a l o f p r e o p e r a t i v e and p o s to p e r a t iv e c a r e t Kinney. Egdahl. and Zuidema. Eds.), W. B. Saunders. Philadelphia. 2nd ed.. pp 19-41 1 1971) 10 Siudley. H. O. Pcicent o f weight loss: A basic imlicator of surgical risk in patients with chronic peptic ulcer. J . A m e r . M e ,! . A s s o c . 106. 458 (1936). 11. Bistiian. B R . Blackbuin. G. L... Scrimshaw, N. S., and Flail. J. P. Cellular immunity in semi-starved slates in hospitalized adults. A m e r . J . C lin . N u t r . 28, 1148 (1975). 12. Daly. J. M . Vais. H M.. and Dudiick. S. J. Effects of protein depletion on strength of colonic anastomoses. S u m . ( \ n c c o l . O b s t e t. 1.34, 15 (1972). I V (iieenbcrg. G. K.. Mailiss. F.. H . Anderson. G H . Langer. B.. Spence. W .. Tovee. E. B.. and Jccjechhoy. K. N Pioicin sparing therapy in postoperative patients: EfTect o f added hypocaloiic glucose 01 lipid. A'. T o ta l. J . M e d . 294, |4|2 (1976). 14 Blackburn. G. I.. Pcriphcial amino acid infusions, in T o t a l p a r e n t e r a l n u t r i ti o n (J E Fischer. Ed ). Little. Brown. Boston, pp. 363-393 (1976). 15. Groite. (i . Jacobson. S.. and Wreilind. A. Lipid emulsions and technique of peripheral adminisuation in paiemeial nu iiiiion. in I o t a l p a r e n t e r a l n u t r i ti o n (J. E. Fischer. E d ). Little. Brown. Boston, pp 335-362 (1976). 16 Wadstrom. I . K .. and Wiklund. P. F.. EfTect of fat emulsions on nitrogen balance in the postopera tive pcnotl. , \( ta C h u S i a m i . i S u p p l . l 325, 50 (I9M ). OYNECOLOCIC ONCOLOGY 7, 325-344 (1979) The Woman at Risk for Developing Ovarian Cancer1-2 L arry McGo w a n , M.D., L inda Par ent. Wayne L ednar, Ph.D., and H enry J. N orris. M.D.* Division o f Gynecologic Oncology, Department o f Obstetrics and Gynecology. George Washington University Medical Center. 2150 Pennsylvania Avenue, N. W.. Washington, D. C. 20037. and *Department o f Gynecologic and Breast Pathology. Armed Forces Institute o f Pathology, Washington. D. C. 20305 Received July 18. 1978. Defining the woman at risk for developing ovarian cancer can identify factors that cause the disease and lead to earlier diagnosis and improved survival. Our 3-year case-control study o f 197 women with histologic proven primary ovarian cancer o f the epithelial type reveals the typical patient to have a family history o f gynecologic cancer, be exposed to rubella in the peripubertal period, be nulliparous or of low parity, experience difficulty in conceiving, have certain exacerbated premenstrual symptoms, and to rarely undergo surgi cally induced menopause. In women with carcinomas o f low malignant potential a later onset o f menarche as well as menstrual disorders occur. During the past three decades the persistently poor survival has made ovarian cancer one of the most common causes of cancer deaths in American women. In 1979 there will be approximately 17,000 new cases of primary ovarian cancer with approximately 11,100 deaths. Ovarian cancer is the fourth leading cause of cancer deaths in American women 12). Because of our inability to consistently diagnose the disease in its early stages, the usual ovarian cancer patient has widespread disease, clinical stage III, at the time of operation 11,17,21). Improvements in survival depend, in great measure, upon earlier detection of the disease. Even if we had a near-perfect diagnostic aid for ovarian cancer, we would have to apply it to those women who have the greatest likelihood of developing the disease. In addition, the concept of preventing ovarian cancer or at least recognizing those, causative or contributing factors in the development of the disease is one of our more interesting clinical problems. The purpose of this study was to identify risk factors of presumed etiologic and diagnostic significance for ovarian cancers of the epithelial type. 1 This investigation was supported by Grant R0I-CA-I494I 01-03 and in part by Grant RI0-CA13446 04-05. awarded by the National Cancer Institute, the Department of Health. Education and Welfare. * Presented at The Ninth Annual Meeting of The Society o f Gynecologic Oncologists*. Miami. Fla.. January 18. 1978. 325 0090-8258779/030325-20S01.00/0 Copyright (E) 1979 by Academic Pres. Inc. All right of reproduction in any form reserved. 326 M C G O W A N ET A L . MATERIALS AND METHODS Th present study was designed specifically to address the methodological deficiencies of prior epidemiological studies of cancer of the ovary. Cose selection and putholouic confirm ation. Our case-control study was con ducted from September 1974 to April 1977 in the Washington, D.C., metropolitan area. One hundred ninety-seven cases of pathologically confirmed primary ovar ian cancers of the epithelial type were identified and included in the study. Of these 42 (21%) were carcinomas of low malignant potential (LMP) and were analyzed separately. Operative reports, surgical pathology reports, and representative microscopic slides of the tumor were obtained for each ovarian cancer case and were reviewed by the study's pathologist (H.J.N.). Study eligibility was limited to the epithelialtype primary cancer of the ovary as defined by the World Health Organization 122). The neoplasms included in this study were mainly serous, mucinous, or endomclroid types, but the clear cell, Brenner, mixed, undifferentiated, or un classified epithelial tumors [1.19.22) were also identified. Within each type, tumors of borderline malignancy (carcinomas of low malignant potential) as well as malignant tumors were recognized |22). Control selection. One hundred ninety-seven women were selected as controls and were frequently-matched by age, race, and hospital of origin with the ovarian cancer cases. The control group of women had no history of cancer and were selected from general medical and surgical services so as to exclude obstetric, gynecologic, psychiatric, and cancer patients. Ditto collection. Information was collected through direct interviews conducted by women trained as medical interviewers. In an attempt to minimize interob server variability, only live interviewers were used. The detailed interview was administered in the same manner to all cancer patients and to those women in the control group and took approximately 60 min to complete. The information obtained included the categories of variables seen in Table 1. Differences in risk factors between cases and controls were statistically evaluated using the Students t lest and the for univariate comparisons. The risk of ovarian cancer associated with various reproductive and menstrual characteris tics was calculated and evaluated for statistical significance using the MantelHacnszel method 115.231. RESULTS S ocioilnnonropliii Characteristics Ovarian cancer patients and their controls were comparable in age, race, and sex due to the frequency-matching criteria utilized for control selection. The mean age of women with malignant ovarian tumors was 52.1 years with a mean age of 52.2 years for their controls. Women with carcinomas of low malignant potential and their controls were considerably younger with a mean age of 40.9 and 40.4 years, respectively. These age distributions are expected 11). White women with malignant ovarian tumors comprised 87.1% of the group and blacks 12.9%; identi cal racial proportions were achieved in the control group. Among the women with WOMAN AT RISK: OVARIAN CANCER TABLE 1 C ategories for Ev a l u a t in g Differ en c esjn Ovarian Ca n c e r Pa t ie n t s a n d Co n t r o l W o m e n Sociodemographic data Age. race, religious preference Residence patlems Years of completed education Marital status Socioeconomic status Occupation Family-relatives Anthropometric measures Menstrual profile Menare he General menstrual Difficulties with menstruation Premenstrual symptoms Menopause Reproductive history Sexual history Contraceptive usage Past history Benign cysts Chronic disease history Diet and weight Exposure to loxic substances Family cancer history Hygiene Infectious diseases Medication usage Substances abuse Surgical history 327 carcinomas of low malignant potential and their controls, 76% were white and 24% black. The majority of women (92.9%) with malignant ovarian tumors had been mar ried at least once as compared with 87.1% of their controls. This high percentage of "ever-married'' women was also observed among patients with borderline malignant ovarian tumors and their controls. 78.5 and 83.3% respectively. Women with ovarian cancer and their controls were also observed to be com parable with regard to religious preference, social class (as measured through use of the Hollingshead Social Index), and educational level. The majority of women with ovarian cancer and their controls lived in the metropolitan Washington, D.C., area for at least 10 years prior to their diagnosis or inclusion in this study. Because of the small number (42) of low malignant potential cases included in the study, statistical analyses could not be carried out for all variables. Statistical data for this group will therefore be presented for selected variables only. Reproductive H istory The reproductive experience of the study participants was divided into mea surements of gravidity (the state of a woman after conception and until termina- 328 M C G O W AN ET AL. lion of the gestation) and measurements of parity (the state of having given birth to an infant or infants, alive or dead). Prcuiunuy (y rtn id iiy ). Comparative analyses of women with ovarian cancer and their controls were carried out for the following variables: mean and categori cal number of pregnancies, person-years at risk of conceiving, number of preg nancies per 1000 persons-years at risk of pregnancy, and perceived difficulty in conceiving. Analyses for differences between cases and controls as related to various reproduction characteristics were restricted to women who were ever married (including all subjects who were currently married, widowed, divorced, or sepa rated). This allowed for the comparison of women whose risk of pregnancy was considered to be more homogeneous. Confining the analyses to women who were ever married did not alter the comparability of the case and control groups as this subgroup was noted to be similar with respect to race, religious preference, age, area of residence, and current marital status. Ever-married women with ovarian cancer were found to have significantly fewer pregnancies than their respective controls (Table 2). A comparison of the mean number of pregnancies, although a useful general indicator of decreased gravidity, disguises the fact that a large proportion of ever-married controls had five or more pregnancies, whereas a similarly large proportion of ovarian cancer cases were nulligravidas (see Table 3). The magnitude of difference in pregnancy experience between the women with malignant ovarian tumors and their controls is reflected in the strength of the P value (P = 0.008), a finding not likely to be produced by an artifact secondary to small sample size (Table 3). Calculations of relative risk indicated that nulligravidas were 2.45 times more likely to develop malignant ovarian tumors and 2.9 times more likely to develop carcinomas of low malignant potential than women who were pregnant three or more limes. The risk of ovarian cancer was greatly reduced among women who had at least one pregnancy (see Table 4). Gravidity measurements are, of course, highly influenced by a number of factors in addition to those which have been controlled for in frequency matching. A number of pregnancy risk estimates have, therefore, been developed for pur poses of the present analysis and include: (I) Married risk which refers to a composite person-years at risk of pregnancy while married. This index is com puted by subtracting the subject's age at marriage from the subject's age at termination of marriage, menopause (surgical or natural), or adoption of perma nent birth control measures (e g., tubal ligation in self or vasectomy in husband) TABLE 2 M l A N N u M B I r <11 P R E G N A N C I E S A M O N G E V E R M a R R I E D W O M E N A N D C O N T R O L S M alignant o varian tum or* N = 144 C o n tro ls N = 135 Low malignant potential N = 33 Controls N - 35 2.2 3.0 r o.oi 1.9 2.8 P < 0 05 WOMAN AT RISK: OVARtAN CANCER 329 TABLE 3 G r a v id it y a m o ng Ever: M a r r ie d C ases a n d Controls Times pregnant 0 1-2 3-4 5+ Women with malignant ovarian tumors Control group 35 47 47 15 17 44 43 31 P = 0.008 Total No. o f ever married 144 135 depending on which of these three events occurs first. (2) Adjusted married risk which is computed by subtracting the total number of married years in which birth control methods were used from the married risk index as described above. Analysis of married risk and adjusted married risk indices indicates that the low gravidity reported by both case groups cannot be attributed to fewer years at risk of pregnancy as compared with their controls (see Table 5). Data presented in Table S also indicate that a large percentage of the nulligravidity among controls can, perhaps, be attributed to significantly decreased (as compared with cases) person-years at risk of pregnancy. Furthermore, when two additional indices (number of pregnancies per 1000 years of married risk and number of pregnancies per 1000 years of adjusted married risk) are computed, women with ovarian cancer were found to have significantly fewer pregnancies per 1000 years of married risk and adjusted married risk than their controls (Table 6). The actual degree of difference, however, appears to be stronger in the study group com prised of women with low malignant potential ovarian tumors and their controls. We compared gravidity of cases and controls who were equally "unprotected" from the risk of pregnancy and the number of pregnancies among subjects who reported that they had never used birth control measures of any kind. Thirty-three percent (47/144) of the ever-married patients with malignant ovarian tumors and 28% (38/135) of their respective controls stated that they had never used birth control. The average age of these subjects and years of married risk (or preg nancy) were comparable. Women with malignant ovarian tumors who had never used birth control methods had significantly fewer pregnancies than controls who also abstained from the use of birth control (P < 0.02). TABLE 4 Re l a t iv e Risk A sso ciated w it h N u m b e r of Pr e g n a n c ie s among Ever-M arr ied Women No. of pregnancies Women with malignant ovarian tumors None 1-2 3+ 2.45* 1.27 LOO Significan! at 0.03. Women with carcinomas of LMP 2.9 1.02 1.00 330 M C G O W A N ET A L . Control group r2t o ' O C a -o 2. E 2J So 2 -- sOOO | fX O' | Women with carcinomais of LMP ol p Adjustment o f mean: Married years at risk minus married yean using any form of birth control. o .* sO>s *"OC s ' ss xC O=> U 00 S W i i^ c oS 5 . -t o S * *3 n OJ j r-4 m go | No. of pregnanci V e --7 t'^+ LU -J * ffl 52 <* Q s. ol p 5 a r- o oo " e u" tla/>l r<u o' 5 c ,E SO s _g *\= 1 e lo ^ NN- e acP W <0 ffv :an years of lamed risk 5C e ,,g * sI |.l = ill > os C" J O--' N-- Do V <rr iGCd ' z asS> o 7-- mT + W O M A N AT RISK! O V A R IA N CANCER 331 TABLE 6 N u m b e r o f Preg n an c ies per 1000 Person-Y ears a t Risk o f Pr eg nan cy Risk status Women with malignant ovarian tumors Control group Women with carcinomas of LMP Control group Mamed (Total) (Adjusted)" 126 213 183 130 208 314 208 419 Married years at risk unprotected by any method o f birlh control. The categorical pregnancy history among non-birth control users is presented in Table 7 and further indicates that cases and controls have significantly different pregnancy profiles, a difference which cannot be explained on the basis of unequal pregnancy risk. In order to explore the possibility that infertility per se may be associated with the development of ovarian cancer, the subjects were asked whether or not they: (I) ever had difficulty conceiving. (2) consulted a physician for fertility-related disorders, and (3) whether or not the nature and source of the infertility problem was known. Table 8 reveals that women with ovarian cancer report that they had experienced difficulty in conceiving significantly more often than their controls. Women with malignant ovarian tumors who stated they had difficulty in con ceiving reported significantly fewer pregnancies than their controls (Table 9). The relative risk of developing ovarian cancer based on equal risk for three or more pregnancies among those subjects stating difficulty in conceiving suggests that nulligravidity may be highly associated with ovarian cancer development. Among women with malignant ovarian tumors and their controls who had reported difficulty in conceiving, slightly more controls attributed this difficulty to themselves rather than to their spouses (65 versus 50%). However, this finding may be the consequence that considerably more cases than controls indicated that the source of the difficulty was unknown. Similar proportions of women with tumors of low malignant potential and their controls attributed their apparent infertility to themselves. More controls than cases reported that their infertility had been confirmed by a physician. TABLE 7 N u m ber of Preg nancies a m o ng E ver-M a r r ie d Subjects U sing N o Bir t h Co ntro l No. of pregnancies Women with maJignant ovarian tumors N% Control Group N% Women with carcinomas of LMP N% Control group N% 0 18 38 11 29 4 44 0 -- 1-2 14 30 10 26 4 44 2 29 3-4 13 28 7 18 1 11 3 43 5+ 2 4 10 26 0 -- 2 29 332 MCGOWAN ET AL. TABLE 8 Ac k n o w i e d g e d D if f ic u l t y in B e c o m in g Pr e g n a n t a m o n g W o m e n w it h O varian Cancer an d T heir Controls D iffic u lty in beco m in g pregnant Yes No Unsure or unknown Total Women with m a lig n an t o varian tum ors Conlrol group tV % N % 66 43 35 23 6 43 89 57 21 14 31 20 155 100 155 100 P < 0.005 Women wilh carcinomas of LM P Conlrol group N %N % 19 45 8 19 13 31 29 69 to 24 5 12 42 100 42 100 P < 0.005 Age in pri yiuiru v. A comparison of the mean age of first pregnancy shows women with malignant ovarian tumors to report their first conception at a some what older age than their controls (24.4 versus 23.7 years). Women with car cinomas of low malignant potential demonstrate a similar trend (23.7 versus 22.9 years). The age differences noted above were not statistically significant in either study group. However, a significantly smaller proportion (P < 0.05) of the women wiih malignant ovarian tumors were pregnant for the first time before age 25 (see Table 10). The reported age at first sexual intercourse was approximately the same for cases and controls. Purity. When the mean number of pregnancies producing a live birth was computed for ever-married subjects, women with ovarian cancer clearly demon strated a marked decrease in the number of live births relative to their controls (see Table II). Analysis of the pregnancy history by number of pregnancies reveals that ever-married women with malignant ovarian tumors reported nullipar ity almost twice as often as their control group of women and the overall distribu tion of ihe number of live births is highly significant (see Table 12). This difference in distribution of live birth pregnancies failed (due to small numbers of subjects) to table 9 NllMRfcK OF PREGNANCIES AND RELATIVE RlSKS AMONG EVER-Ma RRIED WOMEN w m i Ackn o w led g ed Dif f ic u l t y i n Co nceiving No. of pregnancies Women wiih malignan! ovarian lumors iV % Control group N% Relative risk Never pregnam 1-2 3-4 5+ Tolal 2i3> 34 8 33.3 8 14 16 24 2 4 5 7.6 8 U6 100.0 34 P = 0 05 23.5 41.2 11.8 23.5 100 0 1.64 0.897 1.00 W O M A N A T RISK: O V A R IA N CANCER 333 T A B LE 1 0 ___ D is t r ib u t io n b y A oe a t F irst Pr e g n a n c y Age at first Pregnancy Less than 20 20-24 25 + Never pregnant Total Women with malignant ovarian tumors N% 16 10 35 23 57* 37 45 30 153 100 Control group N% 18 12 53 35 47 31 34 22 152 100 P < 0.05. reach significance in those women with carcinom as o f low malignant potential and their control group of women although the same trend was present. Relative risks computed on the basis o f equal risk for three or more live births indicated that the risk of ovarian cancer was elevated for nulliparous women (see Table 13). W omen who were married and had at least one live birth were at similar levels of risk of developing ovarian cancer. Women with ovarian cancer did not have more spontaneous abortions, stillbirths, or defective children than their controls. Therefore, although the women with ovarian cancer dem onstrated a significantly lower mean number of live births as compared with their controls, this difference can be attributed to decreased gravidity rather than to decreased fetal viability. There was also no dissimilarity between "ever-pregnant" cases and controls with regard to problems during their pregnancy. We attem pted to determine w hether nulligravity among women with ovarian cancer was in fact planned and therefore a reflection o f conscious preference to remain childless or unplanned and therefore suggestive o f underlying infertility. T w enty-seven percent o f the cases with malignant tum ors (39) and 15% of controls (20) reported that they were childless; o f this group o f subjects, 21% o f the cases and only 11% of their controls stated that their childlessness was unplanned. The difference in unplanned childlessness was statistically significant ( P = 0.02) and indicates that childlessness might be ascribed to fertility-related disorders (in self or spouse) more often among ovarian cancer patients than among their controls. T A B LE 11 M ean N umber of L ive Births am o no E ver-M arried Wom en a n d T heir Controls Women with malignant ovarian tumors /V - 144 1.56 Control group N - 135 P < .05 2.25 Women with carcinomas of LMP N = 33 1.57 Control group N - 35 2.20 334 MCGOWAN t T Al.. T A B LE 12 N u m m i m o i L iv i Hih m i [` h i i .n a Ni i i s a m o n g E v f r -M a r m f d W o m f n No. ol pregnancies 0 1-2 3-4 54 Women with malignant ovarian tumors N% 45 31 55 38 40 29 43 P = .0003 Control group N% 23 17 64 47 29 22 19 14 M c il. \ I i i i i i I 1` m f i l t M c iu m h c The metiii age til m enarche was 12.8 years for women with malignant ovarian tumors and 12.7 years for their controls. For women with carcinomas of low malignant potential and their controls, the mean age at m enarche was 12.7 and 12.2 years, respectively. Thus, the age at m enarche was similar for all study groups and parallels general m enarche statistics available for the United States I 31. When relative risks based on equal risk for menarche between the ages of 11 and 13 were com puted, no increased risk of ovarian cancer was found in subjects who reached menarche at age 10 years or less or 14 o r more years o f age. In the group of women with carcinom as o f low malignant potential, those subjects who re ported iheir m enarche at age 14 or older experienced a 3.8-fold risk of ovarian cancel over subjects who reported m enarche betw een the ages o f 11 and 13 (P = 0 05). ( tent-nil iiieiixiriinl t luinicn-ri.\/ii x. A comparison of a variety of menstrual characteristics revealed that the average length o f the m enstrual cycle among ovarian cancer cases and their controls was quite similar (2 8 .1 versus 27.8 days). Women with malignant ovarian tum ors reported an average of 4.9 days of menstrual (low as com pared with their controls who reported an average duration of 5.2 days of menstrual flow. Mcnxlriml niolinicn. The reported occurrence of specific types of premenstrual symptoms including breast engorgem ent, headache, weight gain, mental cloudi- Ki i a iiv i TA B LE 13 Risk A ssih iatf.d w it h N umber of L ive Births A M O N G F.VI.R-Ma RRIED WOMEN Number of live births 0 1-2 3 * ' Significant al 0.05. Women with malignant ovarian tumors 2.13* 0.937 1.00 Women with carcinomas of LMP 3 00 1.17 1.00 WOMAN AT RISK: OVARIAN CANCER t S g Is x: <3 si a "S 5 !^ C 5s 0 E 5 j J-- 1 30 # = Os V T A B LE 14 Reported I n c id e n c e o f Pr e m e n s t r u a l Sym ptom s o '* 1 || 5 '5; 0 '>3 u 5a * i JC ! 5 80 ` 1 0 * COl --O O V s * N t 2V) Mu 11 !* '3 3 NOn I c a*3 C cc0 i 52 # ^O d V c 0- v<0i 0 ? 5 Q. ' j1 # 3 1 :1 OS. "8 JC 5S e Cfc 0 K- n p -aF 5 O e 2 Ze 4 ONN VI ** O d V C NO 335 - . i 'i ... . 336 M C G O W A N ET A L . ness, and insomnia was similar in women with ovarian cancer a n d in their con trols. Women with ovarian ca nce T , however, reported a history of marked irrita bility and generalized malaise preceding their menstrual flow significantly more often than did their controls (Table 14). These symptoms appeared to be highly correlated as an equal number of subjects reported a history of both conditions. Difficulties with menstruation. Women with ovarian cancer and their controls reported a history of menstrual difficulties such as dysmenorrhea, menorrhagia, and amenorrhea with approximately the same frequency. However, women with carcinomas of low malignant potential who gave a positive history of dys menorrhea and menorrhangia reported that they experienced these disorders for significantly longer periods of time than did the controls who also reported a history of such disorders (see Table 15). Menopause. Examination of data regarding menopause indicated that an equal proportion of women with ovarian cancer and their controls experienced a natural menopause prior to their cancer diagnosis or, in the case of the controls, study interview. The mean age at which they experienced their natural menopause was also very similar. When compulation of the percentage of subjects who were menstruating at the time of cancer diagnosis (or at the lime of the interview) was performed, it was found that significantly more women with ovarian cancer than controls were menstruating despite similarities in age distribution (see Table 16). This difference was due to the fact that 28% of the controls versus 1% of the women with malignant ovarian tumors had undergone surgical menopause. None of the women with carcinomas of low malignant potential versus 21% of their control group hud undergone surgical menopause. The surgical menopause as discussed here means only hysterectomy and not removal of (he ovaries. Exogenous Hormones There were no statistically significant differences in the incidence or duration of use of oral contraceptives between women with ovarian cancer and controls; TA B LE 15 I ncidence a n d D u r atio n of D ysmenorrhea a n d M eno rrhag ia Women with carcinomas of LMP Control group N %N% Percentage p o sitive h is to ry o f m enorrhagia M ean d u ra tio n P ercentage positiveh is to ry o f dysm enorrhea M ean d uratio n 23 55 20 48 16 years 7 years (P < 0.02) 16 39 19 46 17 years 9 years (P < 0.02) i WOMAN AT r i s k : o v a r i a n c a n c e r 337 TA B LE 16 M e n o p a u s a l St a t u s Women with malignant ovarian tumors Control group Women with carcinomas of LMP Control group N % N % N %N % Natural menopause 69 45 64 41 9 21 9 21 Surgical menopause Other menopause 2 , 43 28 0 3 22 , 0 9 21 12 P < 0.01 Still menstrual mg" 81 52 46 30 33 79 23 55 P < 0.005 P < 0.05 ' Percentage o f subjects menstruating at time o f diagnosis (for women with malignant ovarian tumors and women with carcinomas o f low malignant potential) or at lime o f interview (for control group o f women). however, women with ovarian cancer did report use of oral contraceptives less frequently than their controls. Women who did not use oral contraceptives were 1.4 times more likely to develop malignant ovarian tumors and 1.78 times more likely to develop carcinomas of low malignant potential than those women who used oral contraceptives. The mean duration of oral contraceptive use among ever-married birth control users was approximately 40 months for both cases and controls. The reported use of hormones for purposes other than birth control was quite similar between women with ovarian cancer and their controls. Family Pregnancy H istory The average age at pregnancy for mothers of women who subsequently de veloped a malignant ovarian tumor was 27.7 years and for the control group, 28.3 years. Similar means were noted among the mothers of women with carcinomas of low malignant potential (28.3 years) and their controls (27.4 years). To determine whether or not the low gravidity noted among the cases could be attributed to certain sociocultural factors which influence the size of a woman's family, the mean number of siblings was determined for cases and controls. We found that cases and controls came from families of the same average size which indicates that the low-parity cases are not imitating low-parity mothers. Family Cancer H istory A family history of specific cancers such as ovarian, breast, and cervical was not reported with any more frequency among women with ovarian cancer as compared with their control groups. However, women with malignant ovarian 338 MCGOWAN ET AL. tumors did report significantly more cancer of the "female reproductive organs" in their relatives than did -their controls ( 14/T55 versus 2/155). (P = 0.01). Many women could not recall the specific gynecologic organ affected by the cancer, consequently the term "female reproductive organs" was used. Benign Cysts Only 7 out of 155 women with malignant ovarian tumors reported a (nonsurgicully confirmed) history of benign ovarian cysts as compared with 21 of 155 controls. This finding was significant at the 0.005 level. The reported incidence of benign ovarian cysts among family members of women with malignant ovarian tumors and women with carcinomas of low malignant potential was higher than among relatives of their respective controls (e.g., 12/155 versus 5/155 and 6/42 versus 0/42. respectively). In addition, women with malignant ovarian tumors reported a significantly higher prevalence of breast cysts among female relatives than did those women who were their controls (26/155 versus 12/155) (P = 0.01). Surgical History There was no difference between women with ovarian cancer and the control women with regard to a past history of uterine suspension, cervical repair, tubal ligation, cesarean section, or lyroidectomy. Stress Women with ovarian cancer and their controls also reported similar histories with regard to exposure to stressful situations or duration of stress. Pont hint; mil Hygiene Women with ovarian cancer and their controls did not differ in their reported use of douches, consistency of use, average age first began douching, or average number of years of douching (see Table 17). Percemage ever used Consisleill use only Total Average age began douching Average number of years douched T A B LE 17 Do uch in g H istorv Women wiih malignan! ovarian tumors Control group N %N % 97 7K 112 72 61 49 70 45 125 l(M) 155 100 22.4 21.8 23.3 22.8 Women with carcinomas of LMP N% 23 70 14 42 33 100 21.5 17.1 Control group N% 36 86 18 43 42 100 20.0 15.7 W O M A N A T RISK: O V A R IA N CANCER 339 Age (years) Less than 12 12-18 19+ Total P < 0.02. T A B LE 18 A ce of Ru bella Infection Women with malignant ovarian tumors N% 35 54.7 17 26.6* 12 18.8 64 100.0 Control group N% 54 75 6 8.3 12 16.7 72 100.0 No differences with regard to circumcision among male partners were noted among cases and controls. Infections Diseases The subjects' experience with a variety of childhood illnesses and other infec tious diseases was ascertained. Women with malignant ovarian tumors reported having rubella infection during the 12- to 18-year age interval significantly more often than the controls (see Table 18). The study subjects who reported infection with rubella during the ages 12 to 18 were 3.9 times more likely to develop ovarian cancer than women who were infected during their early childhood years (Table 19). The prevalence (i.e., whether or not the subject ever had the disease) of rubella infection is not a distinguishing risk factor but the reported age of the infection does appear to be associated with risk of ovarian cancer development. Perhaps secondary to the uncertainty in distinguishing rubella from measles for these study subjects, measles demonstrates a similar trend of changing risk with age. The reported age of mumps infection was not found to be strongly associated with the development of ovarian cancer but does repeat the age-risk trend noted for rubella and measles. This raises the question as to the specificity of the association. Perhaps more than one virus may be contributing to later ovarian neoplastic development. The potential etiologic significance of the timing of the rubella infection and the menarche to ovarian cancer prompted us to cross .ubulale age of rubella infection and age of menarche (Table 20). Among women who experienced menarche at age T A B LE 19 Risk of M a l ig n a n t O v a r ia n T umors Associated w it h Se l e c t e d I n fe c t io u s D iseases Age Mumps Measles Rubella Significant at 0.05. 6 - II 1.08 0.82 0.02 12-18 1.35 2.67 3.90* 19 + 0.90 0.82 1.38 340 MCGOWAN LI AL. TA B LE 20 ASSOt tA I ION OF AciL OF MENARCHE AN|> RUBELLA WITH -VARIAN G*NCEft Age a! menarche II or younger 12 or older Age of Rubella infection 11 or younger 12 or older Total Women with malignant ovarian tumors M r 'r Control group /V % 9 69 10 83 4 31 2 17 n 100 12 100 p =>0.72 Women with malignant ovarian tumors /V % Control group N% 26 51 44 75 25 49 15 25 51 100 59 100 P = .02 11 or younger, the age of rubella infection was not found to be associated with malignant ovarian tumors [P - 0.72). In women experiencing menarche at age 12 or older, however, the age of rubella infection was associated with ovarian cancer (P = 0.02). In addition to the aforementioned, data analyses found cases and controls similar in general identifying characteristics and in the majority of variables under study (see Table 21). T A B LE 21 Co m p a r a b iih y b e t w i.cn W o m e n w it h O v a r ia n C a n c e r a n d Co n t r o l W o m e n Sex. age. race Religious preference Marital status Socioeconomic status Years of completed education Residence patterns Age at first coitus Age at menarche and menopause Length or menstrual cycle Spontaneous abortions and stillborns Problems during pregnancy Prcgnancy/live birth ratios Defective children Married years at risk o f pregnancy Mother s age when pregnant with her daughter who later developed ovarian cancer Douching history Exposure to stress Operative procedures such as cesarean section, tubal ligation, uterine suspension, cervical repair, and thyroidectomy Circumcision in male partners WOMAN AT RISK: OVARIAN CANCER 341 DISCUSSION Reproductive History Significant differences were consistently noted between cases and controls with respect to both gravidity and parity. Women with ovarian cancer had significantly fewer pregnancies even when controlling for age. race, socioeconomic status, residence, religion, marital status, number of years at risk of pregnancy, and use of birth control measures; were more frequently nulligravidous; reported more difficulty conceiving; and tended toward a later age at first pregnancy than their controls. Calculations of relative risks indicated that nulligravidity per se is more influential in establishing risk than the actual number of pregnancies. In married, sexually active women decreased or zero gravidity could be attrib uted to ovarian-mediated infertility (either relative or complete), nonovarianmediated infertility, the conscious use of contraceptive techniques, or an infertile husband. Certain results of the present study tend to support the hypothesis advanced by Pathalla [5] that "extravagant" and incessant ovulation in the woman may be an inciting factor for ovarian cancer development of the epithelial type. It was noted in our study that: ( I) Cases reported use of oral contraceptives less often than their controls, (2) decreased gravidity in cases seemed not to be attributable to anovula tory cycles, and (3) cases reported a history of surgical menopause significantly less often than the controls. These three observations seems to indicate that the cases ovulate for a greater proportion of their reproductive years than their controls. Fraumeni et al. have observed that in white Catholic nuns (whose nearly universal nulligravidity cannot be ascribed to infertility) there was a higher fre quency of ovarian cancer than in the national white female population |8J. Furthermore, the experimental acceleration of ovulation and egg production in the domestic fowl with the early appearance of ovarian adenocarcinoma and no tumors developed in control hens under normal conditions lends supportive evi dence that continued ovulation may be associated with ovarian cancer develop ment. Women with ovarian cancer demonstrate a significantly lower mean number of pregnancies resulting in live births than their controls. This decreased parity can, however, be attributed to decreased gravidity generally as there were essentially no differences in the number of spontaneous abortions and voluntary interruptions of pregnancies among cases and controls. Based on our study data it would seem that women with ovarian cancer do not demonstrate excessive reproductive handicaps if they do become pregnant. Exogenous Hormones Our study data found that women with ovarian cancer used oral contraceptives less often than their controls. This finding is interesting in relation to the hypothesis suggested by Stadel 124] of an association between increased pituitary gonadotrophins and the subsequent development of ovarian cancer. The incidence of ovarian cancer increases rapidly after age 40 which is also when significant numbers of women began reporting natural menopause (14). The menopause is 142 M C G O W AN F t A I . . associated with a decline in estrogen production and a rise in pituitary gonadotro pins This association has been noted by McGowan and Davis as well as others in the experimental production of ovarian neoplasms in animals 118]. The current widespread use of estrogens for treatment of menopausal symptoms or estrogens in oral contraceptives may decrease the risk of ovarian cancer in women by suppressing pituitary gonadotrophin production. However, the present study did not reveal decreased use of estrogen for menopausal symptoms among cases as compared with their controls. Ftim ily H istory of Cum cr The observation that significantly more cases than controls reported a family history of "gynecologic cancer" may indicate a genetic predisposition toward the development of ovarian cancer of the epithelial type. The literature has estab lished the occurrence of certain "familial cancers,-' among them gynecologic cancers 16.7.11.12. I6|. V ira l Discuses Extensive information regarding childhood diseases was collected in the present study. It must be emphasized that these data represent the patients' subjective reporting of their illness and therefore tire limited by the adequacy of recall, completeness of the woman's knowledge of her disease experiences, and lack of confirmation such ns with serology. A previous study had noted the negative association between mumps and ovarian cancer ti e., ovarian cancer patients reported a history of mumps signifi cantly less often than their controls) 1251. As previously stated, no differences were noted in this study among cases and controls as to the reported incidence of any of the childhood diseases. The strong association between reported age at rubella infection and increased risk of ovarian cancer noted in the present study is, however, of particular interest. Children with congenital rubella harbor virus in the nasopharynx, lungs, and spinal fluid for at least six months after birth. In a few children, the virus persists until they are about a year old. If cataracts are removed from patients with congenital rubella at age 3 to 4. the virus can often be cultured from the lens tissue 120). The rubella virus may lay dormant in the body for 12 to 15 years before it manifests itself as a chronic progressive panencephalitis [4,20], With the ability of the rubella virus to persist for years after the initial infection, one could theorize that a change in ovarian function or morphology is produced which, although it does not interfere with ovulation, might alter cell surfaces making impregnation difficult or impossible. It is also of interest that a (rend toward increased risk of ovarian cancer in association with a reported history of measles and mumps in the 12- to 18-year age range was also present among our study population. The measles virus has also been associated with many years of dormancy |9], Although these findings in our study did not carry the same strength of statistical association noted for rubella infections, the possibility exists that the incidence of a variety of common child hood viruses in the peripubertal years may increase the risk of ovarian cancer. W O M A N AT RISK: O V A R IA N CANCER 343 In this country required immunization for rubella for all children under 12 is being even more rigidly enforced, in Sweden and elsewhere in Europe rubella vaccination is carried out in girls, ages 11 to 14 years, without prior serologic tests 113). So far most of the cases of subacute sclerosing panencephalitis have involved congenital rubella, although a case that was reported by Lebon and Lyon may have resulted from acquired rubella 110). The establishment, maintenance, and periodic review of immunization records by all responsible parties in this very interesting and possibly serious link between rubella infection in the teenage girl and ovarian cancer development are essential. The importance of this study's findings to the clinician lies in our increased ability to define the woman at risk of developing ovarian cancer of the epithelial type. Although some of our results are tentative and require confirmation through further study, a number of potentially important risk characteristics have been identified. These are: A family history of gynecologic cancer and cystic ovaries; exposure to an infectious disease such as rubella during the peripubertal period; zero or low gravidity; stated difficulty in conceiving; premenstrual symptoms; lack of exposure to oral contraceptives; menopause usually not surgically in duced; and for women with carcinomas of low malignant potential, a later onset of the menarche as well as menstrual disorders. Certainly, the era of preventive oncology is upon us. REFERENCES 1. Aure. J. C., Hoeg, K.. and Kolstad. F. Clinical and histologic studies o f ovarian carcinoma, O h slet. Gynecol. 37, I (1971). 2. Cancer facts and figures. Amcr. Cancer Soc. (1979). 3. Catislon. G. Viral infections and birth defects. Acta Ohstet. Gynecol. Scand. Suppl. 67,21 (1977). 4. Cremer. N. E.. Oshiro. L. S.. Weil. M. L .. Lennette. E. H., Itabashi. H. H ., and Camay. A. Isolation o f rubella virus from brain in chronic progressive panencephalitis. J. Gen. Virol. 29, 143 (1975). 5. Fathalla, M. F. Incessant ovulation--a factor in ovarian neoplasia? Lancet 2, 163 (1971). 6. Fraumeni, J. F. (Ed.) Persons at high risk o f cancer: An approach to cancer etiology and control Academic Press. New York (1975). 7. Fraumeni. J. F ., Grundy. G. W.. Creagan, E. T ., and Everson, R. B. Six families prone to ovarian cancer. Cancer 36. 364 (1975). 8. Fraumeni. J. F., Lloyd. J. W., Smith, E. M., Wagoner. J. K. Cancer mortality among nuns: Role o f marital status in etiology o f neoplastic disease in women. J. Nat. Cancer Inst. 42,455 (1969). 9. Krugman. S. Present status o f measles and rubella immunization in the United Stales: A medical progress report. J. Pediat. 90, I (1977). 10. Lebon, P.. and Lyon. G. Non-congenital rubella encephalitis, Lancet 2, 468 (1974). 11. L i. F. P.. Rapoport, A. H.. Fraumeni, J. F.t and Jensen, R. D. Familial ovarian carcinoma. J. Amer. Med. Ass. 214, 1559 (1970). 12. Lynch, H. T.. Guirgis, H. A.. Albert. S., Brennan. M .. Lynch, J., Kraft. C.. Pocekay, D., Vaughns, C., and Kaplan, A. Familial association o f carcinoma o f the breast and ovary. Surg. Gynecol. Obstet. 138, 717 (1974). 13. MacMahon, B. Age at menarche. USDHEW Publ. No. (HRA) 74-1615, National Center for Health Statistics. Rockville. Md.. November (1973). 14. MacMahon, B.. and Worcester. J. Age at menopause. U.S. 1960-1962. National Center for Health Stat.. USDHEW Scries 11, No. 19, U.S. Government Printing Office. Washington, D.C., October (1966). 15. Mantel, N.\ and Haenszel, W. Statistical aspects o f the analysis o f data from retrospective studies o f disease, J. Nat. Cancer Inst. 22, 719 (1959). 344 M C G O W A N F T A I.. 16. McCiunn. I). J.. Marchant. 1) J., and Bardwil. W. A. Ovarian carcinoma in three teen-age siblings. O h * U `t . til. 43, 1-32 t-1974). 17. McGowan. L. (Ed.) G y n e c o h > u n m u o lo .u y . Applelon-Cenlury-Crofts, New York (1978). IS. McGowan. L.. Davis. R H Intrasplenic ovarian grafts in Syrian hamsters and periloneal fluid cellular disiribuiion. P ro t . S i t e . t l x p . B io t. M e d . 134, 507 (1970). 19. P r o c e e d in g c\ o f t h e m u t u a l ( f i n i c a l a m f c r c i u c o n t a m e r i f f h e o v a r y . The Ontario Cancer Treatment and Research Foundation. Univ. of Windsor. Windsor. Ontario. November (1965). 20. Rawls. W H. Viral persistence in congenital rubella. P r tty r . M a i . V ir o l. 18, 273 (1974). 2I . Segi. M.. and Kurihara. M. C o m e r m o r t a l i t y f o r .s e lf( l e d .site s in 2 4 c o u n t r i e s . No. 6 (1966- 1967) Japan Cancer Society. November (1972). -- 22. Serov. S. F . Sobin. 1.. H . and Sculley. R. F. H i s t o l o y i e a l t y p i n g o f o v a r ia n t u m o r s , in te r n a tio n a l h i s h d o y i t a l i o s s ific a tio n o f tu m o r s . No. 9. World Health Organization. Geneva. Switzerland (1973). 23. Snedecor. G. W.. and Cochran. W. G. S t a t i s t i c a l m e t h o d s . Iowa State Univ. Press. Ames. Iowa, 6 th ed. (1967) 24. Sindel. U. V. The etiology and prevention of ovarian cancer. A m e r . J . O h s t e t. G y n e c o l. 123, 772 (1975). 25. West. R. O. Epidemiologic study o f malignancies of the ovaries. C a n c e r 19, 1001 (1966). GYNECOLOGIC ONCOLOGY 7, 345-348 (1979) Chemotherapy of Squamous Cell Carcinoma of the Vulva: A Review G u n t e r Deppe, M .D ., C a r m e l J. C o h e n , M .D .,' a n d H o w a r d W. B r u c k n e r , M.D, Division o f Gynecologic Oncology. Deparlmcnl o f Obstetrics and Gynecology, the Mount Sinai School o f Medicine o f the City University o f hie m- York. New York. New York. 10029 Received August 3, 1978 A review o f published experience with chemotherapy for the treatment o f recurrent or metaxtalic squamous cell carcinoma o f the vulva found data for nine different drugs used as single agents. Only 75 patients have been treated. Bleomycin produced responses in 19 out of 31 patients, adriamycin in 4 out o f 6 patients, and methotrexate in 2 out o f 5 patients. Seven patients with squamous cell carcinoma o f the vulva were treated with live different combina tion chemotherapy regimens. Responses only occurred with regimens which contained bleomycin and methotrexate. INTRODUCTION Squamous cell carcinoma of the vulva accounts for 3-4% of all primary malig nancies of the genital tract. Radical vulvectomy and bilateral groin dissection with or without pelvic lymphadenectomy is the standard method of therapy employed in over 80% of all cases. There is to date no published review of results in treating squamous cell carcinoma of the vulva with cytotoxic chemotherapy. This modality has rarely been employed in treating squamous cell carcinoma of the vulva, because patients are often considered to be too old with diminished performance status. Recently new dosage schedules and newer drugs have been employed with a satisfactory risk-benefit ratio for these elderly patients. Chemotherapeutic agents can have a beneficial role in the management of selected patients with recurrent or metastatic squamous cell carcinoma of the vulva, and reexamination of this treatment mo dality by gynecologic oncologists is now appropriate. A controlled comparison of the most effective regimens may be feasible. METHODS A literature review was undertaken to discover which chemotherapeutic agents have been used in the treatment of metastatic and/or recurrent squamous cell 1 To whom requests for reprints should be sent: Division o f Gynecologic Oncology. Department of Obstetrics and Gynecology. Mount Sinai Hospital. 1176 Fifth Avc.. New York. N. Y. 10029 345 0090-8258/79/030345-04SO1.00/0 Copyright 197V by Academic Press. Inc A ll rights of reproduction in any form reserved 8 STRANG fcT AL 2. A tk in . N . and U jk e r. M . N onrandom c h ro m u vim c change in caicinoma o f ihe cervix uteri. I. Nine near-diploid iu n io n . Cancer Genei. Cytogenel 7, 21W-222 (1982). 3 A tk in . N. P rognostic value of cytogenetic studies o f tum ors o f the female genital tract, in Advancer in clinical cytology ( L . G . K o u , and D Colem an, Eds ). Masson. New Y o rk . V o l. 2. pp. 103-121 (1984) 4. Strang. P.. Stendahl. U .. Frankendal. B . and L m d g re n , A . R o w cytom etric D N A patterns in cervical carcinom a, Acta Radiol. Oncol. 25. 249-254 (1986). J. Johnson. T .. A d c lio n , M . Sneige. N . W illia m so n . K .. Lee, M .. and Katz. R. Cervical carcinoma D N A content, S-fraction and m a lignancy grading I- Interrelationships. G ynecol. O ncol. 26, 41-5 6 (1987) 6. A dclson. M . Johnson. T -. Sneige. N . W illia m so n . K .. Freedm an. R . and Peters. L. Cervical carcinoma D N A co n te n t. S -fra ctio n and malignancy grading II Comparison w ith clinical staging. Gynecol. Oncol. 24, 57-7U (1987) 7 Y unis. J. The chrom osom al basis o l hum an neoplasia. Science 221, 227-236 (1983) 8 T rib u k a il. B. C linical flow cytom etry. Med. O ncol. T um or Pharmacoiker. 1. 211-218 (1984). 9. T rib u k a il. B Flow cytom etry in surgical paibohsgy and cytology o f tum ors o f the genito-urinary tract, in Advances in clinical cytology (L . G . Koss. and D Coleman. Eds ). Masson. New Y o rk , V o l. 2. pp 163-169 (1984) ID. Strang. P . fcklund. G ., Stendahl. U . and F ranke n d a l, B. S-phase ra le as a p re d ic to r o f e a rly recurrences in carcinom a o f the uterine c c rv ii. Anlicancer Res 7, 807-810 (1987) 11 A tk in . N. Prognostic significance o l p lo id y level in hum an tum ours. I. Carcinom a o f the uterus. J. Nail. Cancer Inst 54, 9U 9-9IU (1976). 12. R uigcrs, D . van dcr Lin d e n . P.. and van Pcperseel. H . D N A -flo w cytom etry o f squamous cell carcinomas fro m the uterine cervix: Identification o f different subgroups. Radiother Oncol. 7, 248-258 (1986). 13. Jacobsen, A . P loidy level and short tim e prognosis o f early cervix cancer. Radiother Oncol. I , 271-275 (1984) 14 G A pptngcr. A ., Frcundcnbcrg. N , R ost. A , H illcm a n s, H -G .. and H ilg a n h , M . The prognostic significance o f (he D N A d istrib u tio n in squamous-cell carcinomas o f the uterine ce rvix. Anal. Quant Cytol I . 148-151 (1986) 15. Baisch. H .. G ohdc, W . and L in d e n . W A nalysis o f PCP-daia to determ ine the fra ctio n o f cells in various phases o f cell cycle. Radial Environs*. Btophys. 12, 3 1-39 (1975). 16. K o iim c ic r. H. Surgical and ra d ia tio n tre a tm e n t o f invasive carci noma o f the uterine cervix. Experience by current individualized Stockholm Technique, Acta Obstet. Gynecol. Scand. Suppl. 43, 148 (T964). 17 Stendahl. U , E kund, G .. W ill n . H . and W ill n , R. Invasive squamous cell carcinoma o f the ulcrin c cervix. III. A malignancy grading system fo r indication o f prognosis a fte r ra d ia tio n therapy, Acta Radiol Oncol 20, 231-243 (1981) 18. Lawless. J F. Statistical m odels a nd m ethods for life-time data, W iley. New Y o rk . (1982). 19 Stanhope, C . Sm ith. J.. T a y lo r W h a rto n . J . R utledge, F . Fletcher, G .. and G allagher, H . C arcinom a o f th e cervix: The effect o f age on survival. Gynecol. Oncol. I l , 188-193 (1980). 20. K jO n ta d . K C arcinom a o f the cervix in the young panent, Obstet Gynecol 50. 28-30 (1977) 21 Strang, P.. Lm dgren. A . , Frankendal. B .. and Stendahl, U . D N A patterns and aggressive histopathologic features in 159 patients w ith cervical carcinom a, Int. J. Gynecol. Paihol. 7, 56-6 3 (1988). 22 D yson. J ., Juslin. C . R o th w e ll. R .. O u irk e . P , K h o u ry . G .. and B ird , C . F low cylom cinc evidence for the differential radioresponsivcncss o f a ncuploid and d ip lo id cervix tum ors. Radiother. Oncol. I , 263-272 (1987) 23. Rehn, S., G lim elius, B , Strang. P , Sundsirm . C , and T ribukail. B . Prognostic significance o f flow cytom etry studies in B -ccll nonHodgkin lym phom a. Hematol. Oncol. I , 1-12 (1990). 24. Stendahl. LI.. Strang, P., W agcnius, G .. BergstrOm, R . and T ribukait, B. Prognostic ngnihcancc o f proliferation in endom etrial carcinomas A m ultivariate analysis o f clinical and flow cytom etric variables. Int. J. Gynecol. P a ih o l. in press 25. Tubiana. M .. and C o u rd i, A . C e ll p ro life ra tio n kinetics in human solid tumors. Relation to probability o f m etasiatic dissemination and long-term survival. Radiother. O ncol. 15, 1-18 (1989). 26. Ivcrscn, O . R o w cyto m e tric deoxyrib o n u cle ic acid index: A prog nostic factor in endom etrial carcinom a, A m . J. Clin. Pathol. 85, 536-541 (1986) 27. Stendahl. U . W ugcnius, G , Strang. P., and T rib u k a il. B R o w cytom etry in invasive endom etrial carcinom a. In Vivo 2, 123-128 (1988) 4.VNH <i| (H ill un.........K,Y 43, 9 2 ' (1991) 5 7 REVIEW The Epidemiology of Ovarian Cancer' T V .<Fabjo Pa r a z z in i, ' Silvia Fran cesoii. I ' C ari u L a i c iiia , ' and M onica Fa s o ii* '"M a rin N enri" Institute fn r Phurrnucalogitul Research. Vut t.ritrcu ti2. 201^7 Milan. It,ils: I l.pitlennalafn (bat. Asuma ( amet ('enter. l i Pedemontana ( h r . . XiOHi Asiann (PN), Italy. Xt'.urapean Canter Presentan Ornant:altan (l'.CPl. H/nilemialn^v untl Cum ci Warkmu ( rotip. tlruxsels, Belgium: and hInstitute a f Social a nil Preventive Metante. Untccrutv of I amanite. IIMlS l.auutttne. .Vun :erlanil Received December 27. I `1**1 The descriptive and analytical epidemiology o f ovarian cancer is reviewed, s ta rlin g fro m (he substantial geographical differences, w ith high rales in N o rth Am erica and Europe and low rales in developing countries and Japan, a llh ou g h, on a w o rld w id e scale, alm ost 50% o f 140.000 to ta l cases occur in developing countries. Over (he past decades, incidence and m o rta lity rates have re m ained a p p ro xim a tely stable in h ig h -risk areas, but have gen e ra lly tended to rise in lo w -risk areas. In the past, ovarian cancer was more com m on in higher social classes, b ut sociocu ltu ra l d if ferences seem to have fla ttened o ff over recent decades. In e tiolo g ical term s, the protection a fforded by m u ltip a rity and o ra l con traceptive (O C ) use is w e ll established, w ith relative risks (R Rs) o f the ord er o f 0.5 for m u ltlp arae and O C users. There is also consistent evidence tha t risk increases w ith late age at menopause. Less consistent and weak, i f any, arc the effects o f age at menarche and first b irth , although there is h in t tha t a first b irth over age 35 is not protective. It is conceivable tha t diet can play an im p o rta n t role in ovarian carcinogenesis, but o nly scattered data arc available to suggest tha t h igh fat consum ption may represent an indicator o f risk. Available knowledge on ovarian cancer ep idem iology is also discussed In re la tion to models o f carcinogenesis and Im p lica tion s fo r prevention and pub lic health. iw i A>iraik Po.. lac INTRODUCTION In developed areas of the world ovarian cancer is a common neoplasm, being among the five leading sites of incidence and mortality [1,2|. On a worldwide basis the tidal number of eases has been estimated to be approx- 1 Supported hy (he li.ilta n Association for (' lice t Research. M ilan, and the Ita lia n League Against T um our and conducted w ith in the fra m e w o rk o f the C N R (Ita lia n N a tio n a l Research C o u n cil) A p p lie d Project "O ncology" (C ontract Jt7.OI.S44.44) and Prevention and C ontrol o f Disease E d ito rs (C o n tra ct 9l.OU2Jt5.PE4l) T o w hom re p rin t requests should he addressed. imatcly 1-40,000, thus representing over 4r < o f all cancers in women and the seventh loading site o f incidence 13J Besides the observation that ovarian cancer is a disease of afllucnl societies, a few aspects o f its epidemiology are well understood, such as iis inverse relationship with par ity and oral contraceptive (OC) use. These factors c\ plain, however, only a minor proportion o f the substantial differences in worldwide ovarian eaneer incidence and mortality. A role for other menstrual, reproductive, and hormonal factors, as well as the potential influence <>! diet, has been suggested in some investigations, but is fat from established or well i|uaniilicd. In the present review, therefore, available knowlcdgi on the epidemiology o f ovarian cancer is .summarized and discussed. In addition to a brief mention of histopathn logical characteristics, the article includes four subdivt sions: hisiopaihology, descriptive epidemiology, risk lac tors, and conclusions and eliopalhogcnetie factors. III.STOI'ATTIOIXXiY Three chief groups of Illin o is originate fiom the ovaix epithelial, sex-cord mcsenchimc. and germ-cell tumors. According to Scully |4j, epithelial forms represent two thirds of all primary ovarian minors and 90% o f thesi are malignant, (icrm-ccll tumors account lor altotil 25 30% of all ovarian tumors (dermoid cysts and dysgei minomas are Ihe more common varieties in this group) and sex-cord and stromal tumors account for an aildiliona <i%. approximately. The most common epithelial tumors derive from tin surface epithelium of the ovary, the descendant of (hi celomic epithelium of the embryonic gonadal ridge (i e. Mullerian ducts). When a neoplasia develops from tin surface epithelium, it acquires the potential o f its cm U M IK .'S S 'M t l ' >>|i)itglil v |v | | t> A Iil. n ii. 1'n ` v*. 1. A ll lig h t'. >! ic |iiin liii'ii> n i hi .m y l - i m ii-M .it> a 10 I'AKAZZINI ET AL bcyonic neighbor 10 differentiate into epithclia resembling those of various Mullerian duel derivates. Serous turnon hi>tologically resemble fallopian tubes, endometrioid tumors the endometrium, mucinous tumors the endocervix, and clear-cell tumors the endometrium during pregnancy. Since germ cells also are totipotential, a wide variety of ncoplasms'may develop from- them tooHndeed gemT cell tumors recapitulate stages of embryonic development: dysgerminomas (from primitive germ cells), endodermal sinus turnon (from yolk sac development), and various kinds of turnon composed of or containing cytosyncytiotrophoblastic cells. Teratomas also belong to this group of turnon: they are solid and predominantly contain ma ture tissue. Finally, sex-cord mesenchimc tumors derive from the sex cord and stromal component of the developing gonad. The embryonic sex cord develops into Sertoli cells in the testis and granulosa cells in the ovary. The stroma de velops into Lcydig cells in the testis and stromal lutein cells and fibroblasts in the ovary. Population-based data on the histologic distribution of ovarian cancer are scanty. In the United States Third National Cancer Survey (1969-1971) (5) all newly diag nosed cases of ovarian cancer in nine areas of the United Slates were identified and reported according to age and histologic type. Nonepithclial types predominated in younger women and their incidence did not change with increasing age. whereas epithelial types rose rapidly with age. overtaking in frequency nonepithclial types at 1524 years of age. The most common epithelial subgroups were serous tumors (36.3%), followed by mucinous (13.8%), endometrioid (2.0%), and clear-cell tumors (0.6%). Sex-cord mesenchimal and germ-cell tumors ac counted for about 2% of all ovarian tumors. The White/Black ratio was above unity in epithelial types but below unity in nonepithclial tumors. Also in this study, however, a large number of "not otherwise specified" carcinomas, adenocarcinomas, and cystoadenocarcinomas was recorded (i.e., approximately 1/3 of the total). Other published material dealing with the relative frequency of individual histologic types mostly derives from hospitalbased series, where, however, substantial discrepancies may derive from chance or different referral patterns with respect to age. Another source of heterogeneity within epithelial ovar ian cancers is the presence of estrogen receptors (ER) and progestin receptors (PGR). Although the ovary is not a primary target tissue for estrogens. ER and PGR have been found in benign and malignant ovarian neo plasms |6- 8|. Extracts of more than 50% of epithelial ovarian cancers contain at least 20 fmol ER per milligram soluble cytosol protein, or at least 500 fmol ER per gram tissue wet weight, in contrast to the much smaller quan tities of ER observed in normal ovarian tissue (6,7|. The content of ER and PGR in ovarian carcinomas has not been correlated consistently with the histologic type and grade of the carcinoma. High levels of ER and PGR were found to affect favorably survival in some reports [8| but not in others (9). Therefore it has not been established convincingly whether a correlation between ER or PGR content and clinical prognosis comparable to that found in breast cancer exists in ovarian cancer. The considerable variety among and difficulty in di agnosing histologic types of ovarian tumors has always hampered the interpretation of etiological data. Most ep idemiological work, including the present review, how ever, simply refers to ovarian cancer by and large, or, in most recent analytical studies, epithelial ovarian cancer. DESCRIPTIVE EPIDEMIOLOGY Geographical Differences There is a large variation in the incidence of ovarian cancer in different countries (Tables 1 and 2). The highest risk areas are European (especially Nordic countries and the United Kingdom) and North American countries, with a ratio o f approximately 4 between the highest and the lowest incidence countries worldwide. Lower rates are generally observed in South European areas. Devel oping countries and Japan have the lowest incidence rates although registration deficiencies should always be con sidered in the interpretation of these findings (Table 2). With respect to the incidence of ovarian cancer as a percentage of all incident cancer cases in women, less marked differences between various countries are ob served. Ovarian cancer, in fact, accounts for about 4-6% of all incident cancer cases in women in most areas. It seems, therefore, that ovarian cancer is a relevant pub lic health problem, especially in Westernized societies or developed countries, but more than 50% of new cases diagnosed every year worldwide occur in developing countries (3). A possible bias in the interpretation of international difference is the influence of the number of ovaries at risk. Where the rale of hysterectomy is high and bilateral oophorectomy tends to be performed routinely also in relatively young patients, such practice may affect per ceptibly the denominator and lead to a certain degree of underestimation of ovarian cancer rates. infracounfry Variability The suggestion that ovarian cancer risk is directly as sociated with modem Western life-style is further con firmed by the consistent imracountry variation in inci dence rates. Considering the incidence data from 20 regions in the world where it has been possible to distin guish urban and rural areas (l j , higher rales of ovarian V OVARIAN CANCER: DISTRIMI I 1ION AND RISK I AC IO RS II TAIU.K I Incidence o f O varian C ancer (and Percentage o f A ll In cid e n t Cases in W om en) in Selected Developed C o un trie s (W o rld P opulation Age-Standard red Kates). 1 ^ 7 8 -IVH2 Country Percentage o l all incident Incide lice cancer eases ra tc /K X U tJ n in women 3 unity lucilie nee ra le /0 0 . 0 0 0 r clientage ol ail i alicer tas in women Norway Sweden Denmark Iceland U K . North East Scotland FRG. Hamburg UK. Oxford U K . Southeast Scotland GDK Switzerland. Geneva Poland. Cracow Poland. Nowy Sacz U K . Trent UK. South Western U K . South Thames Italy. Parma Netherlands. Heindhoven Switzerland. Neuchtel UK. North Western France. Has Rhin U K . East Scotland U K . Birmingham U K . West Scotland U K . Mersey Switzerland. Zurich Poland. Warsaw Italy. Varese Finland Yugoslavia. Slovenia UK. North Scotland Hungary. Vas Czcchoslavakia. Slovakia Switzerland. Basel FKG. Saarland Europe 15.3 15.2 14.5 13.y J32 12 .8 12 .6 12 1 12 .1 12 .0 1 2 .1 li y II.H II.K 11.7 11.7 11.6 113 11.2 112 III Il I 11(1 10 8 10.5 10.5 10 .2 y8 y.s y.5 y.4 y2 y.i y.i 7.2 6.8 6 .0 6 .0 6 .1 60 6 .2 51 6.5 54 66 y.5 5.7 5.y 6 .2 5.8 5.7 50 5.5 5.5 4.y 5.y 5.0 5.0 5.2 6 .6 5.0 5.6 6 .0 4.6 5.7 5.6 4.4 43 Switzer lami. Vaml Ireland. Southern France. Doni France. Calvados Italy. Ragusa France. Iscrc Romania. County C lii| Spain. Tarragona Spain. Navarra Hungary. S/aholcs Spain, /.aiago/,i 1 II i II y0 K2 7y 7.8 6 .(i 6.5 64 58 54 ( anada Manitoba ( muriti Duchee A lb e rta British Columbia United Stales law Angeles (W hites) New York Stale (except C ity ) Iowa Detroit (W hites) Seattle Conneciiciit (Whiles) Nagasaki Osaka Miyagi Notili Alucina 12 .8 1 2 (1 11.4 H1 10 .6 n .s 12 h 12 4 12 3 12 1 12 o Japan 6 `7 44 4.2 New /ca la m i (Marni) New Zealand (non-M aori) Australia II).) 8y 45 S0 5 51 S5 4 I. 4y 42 4.2 S1 42 4y 50 48 S1 47 4y 5o ** 46 46 46 41 2 .y 2y 3.7 41 Source. M u ir ei ul. | l | cancer in urban populations were reported in 14. The excess was about 20-50%. irrespective of whether the region was in a low- or high-risk area (Table 3). In (he United Stales the incidence of ovarian cancer was approximately 50-70% higher among Whiles than among Blacks in all seven areas covered by the Surveil lance Epidemiology and End Results (SEER) Programs (Table 4) and in both premenopausal and postmenopausal women 11, lOj. Trends in Ovarian Cancer Incidence Trends in ovarian cancer rales over the last 20 years in selected ureas are presented in Table 5. Rales in the highest risk areas, for example, the United Slates, Cun- ada, Scandinavia, and the United Kingdom, lemamcd over the period I95X-I9K2 alinosi constant while rising values were observed in low-risk areas such as Japan, India, and Smg;t|orc (Chinese population) (1.10 13) Besides improvements in diagnostic and registration ac curacy. which might have produced artilactual increases, especially in less developed areas, an increased frequency of oophorectomy (often performed together with hyster ectomy in peri- or postmenopausal women), especially in develop'd countries, should be taken into account in the interpretation of temporal trends An additional difficulty stems from the presence of complex cohort effects. In several developed countries younger women lend to show substantial declines in ovar ian cancer risk, possibly due to the protection alloidcd 12 PAKAZZINI LT AL TABLE 2 * Incidence (World Population Age-Standardized Rates) of Ovarlan Cancer and Percentage of All Incident Cues In Women in Selected Developing Are. 1980 TABLE 3 Ovarian Cancer Incidence Rate/100,000 Women in Urban and Rural Populations in Selected Countries (World Population Age- Standardized Rates) , j Country Incidence rates/1 0 0 . 0 0 0 Percentage o f ail incident cancer cases in women Country 1975 -----------------------Urban Rural I960 ---------------Urban Rural Africa Japan, Fukouka 32 29 Middle Africa Western Africa Northern Africa 5.3 4.6 4.1 5.0 Japan. Miyagi 4.2 2 . 6 4.9 3 3 7.8 Japan, Osaka 3.5 2 . 1 4.7 Czechoslovakia, Slovakia 1 2 . 1 7.7 1 2 2 8 . 8 Eastern Africa 3.6 4.7 FRG, Saarland 1 0 2 6.9 Southern Africa 2.4 3.1 France, Calvados 7.9 8 . 8 Central/South America Temperate South A m end 8.7 Tropica) South America 5.7 Caribbean 5.2 Central America 1.9 3.6 3.5 24 Hungary, Szaboles Romania, County Cluj Switzerland. Vaud U K , England and Wales 7.2 9.0 14.8 12 .6 5.7 6 .0 14.2 149 99 7.2 159 64 6.9 111 71 54 14 8 7.1 10.3 109 Asia Austria. North South West 9.2 79 9.2 7.8 USSR Melanesia South Eastern South Asia 93 55 54 5.2 Finland 5.8 France. Bas Rhin 6 1 Poland, Cleszyn 1 1 . 8 9.8 113 107 9 6 7.7 Micro/Polyncsia 46 5.3 Poland. Katowice 8 0 7.9 Western Asia 4.1 4 5 Poland. Warsaw 11.9 9.2 China Southern Asia 4.0 3.7 Spain, Zaragoza 5.2 5 3 35 4.4 U K . North Western 10 3 6.7 2 8 2.4 -- Source. Waterhouse et al. [10); M uir et al. [ I |. Source. Parkin et a t. (3|. by OCs 114]. Ai older ages, conversely, upward trends, which are probably partly real and partly due to improved Menstrual History diagnosis or death certification, are still evident. Age at menarche. The role of age at menarche in RISK FACTORS ovarian cancer risk has been extensively investigated. The majority of the studies (17, 20-22] found a moderately j Sociocultural Indicators increased risk in women with early menarche. The esti mated RRs were of the order of 1.2 to 1.5 in women i Education, as a sociocultural indicator, is examined in whose menarche occurred below age 12 as compared to Table 6. In 5 out of 10 studies, cases were more educated that of those with menarche at age IS or more (Table 7). than controls, and in 1 study, conducted in China (15), the difference was statistically significant (Table 6). Such a sociocultural gradient seems, however, to have Aattcned off over more recent years, especially in studies from TABLE 4 Age-Adjusted Ovarian Cancer Incidence Rales/100.000 developed areas of the world [ 16,17]. Women in Whites and Blacks in Selected SEER Areas: United In other investigations different social indicators such States (World Population Age-Standardized Rates) as annual income or social class were examined, with rather inconsistent results. For instance, Slattery et al. 1975 1980 [18] reported a nonsignificant decrease in ovarian cancer Area Whiles Blacks Whites Blacks risk in the upper level of income, whereas Beral et al. (19) and Booth et al. (20] found that cases belonged more frequently to the higher social class. Generally the higher proportion of women of upper social class among cancer Alameda Bay Area Los Angeles Connecticut 14.0 14.4 139 8.7 8.5 9.6 12 .1 12.9 13.5 12 .0 141.5 9.U 8 .0 7.3 cases as compared to controls is interpreted in terms of Atlanta 12 2 69 12 .6 84 lower fertility in this group, as the number of children New Orleans 9.3 5.5 ILO 7.2 and (he economic level appear to be in most instances Detroit 1 2 . 8 9 8 12.3 8 . 8 inversely related. Source. Waterhouse et al |10|; M uir et al. |l| . .. Jfifcsk.... OVARIAN CANCER: DISTRIBUTION AND RISK I ACIOKS \.\ TABLE 5 Trends in Incidence o f O varia n Cancer in Selected Areas, 1960-1980 (W o rld P opulation A^e-Stand a rd i red Hates) Calendar period* Country I95H-I962 1963-1967 |9hK 1972 1973 .1977 I97X 19X2 Canada Newfoundland Manitoba Ouebec A lb e rta Saskatchewan British Columbia United States Hawaii Caucasian Japanese Chinese Hawatan F il ip in o Alameda While Black Connecticut Now York State Colombia. Cali Jamaica Puerto Rico Japan. Miyagi India. Bombay Singapore Chinese Israel A ll Jews FRG. Hamburg Slpvcma GDR Hungary Szaboles Vas Poland Warsaw Cracow Denmark Finland Norway Sweden United Kingdom Oxford Liverpool South Metropolitan Birmingham New Zcaiand Non-Maori Maori 9.0 15.4 -- 9.3 1 (1 . 0 -- LI 4 7.5 -- 15.9 -- -- -- 13.2 10 .2 12 .8 7.4 4.9 2 .2 -- 32 119 |7.6 X1 -- -- -- -- -- 13. J 9.0 11.5 13.1 -- 9.7 10 .6 1 (1 . 1 1 2 .6 99 14.3 7.5 10 .0 10.3 -- 14.X 9.4 13 1 (6 . 6 4.X 12 .6 10 4 113 -- 9.2 X.3 51 19 6 .1 -- 11.4 -- 86 11.3 5.6 9.0 11.2 6.3 13 5 10 .1 115 14 4 l(M 10.4 -- 10 .8 11.9 11.7 II 7 116 9.6 9.X II 0 14.0 1 1.7 (1.9 5X 11.6 9.(1 13 5 10 J 12.5 11.5 80 ` X. 6 5.3 2 .8 4.X 6 .0 13.0 11.5 10 .1 12 .1 56 9.0 11.4 9.1 15 1 10 4 14 2 IS 1 II 7 lo i II 1 114 13 .1 9.2 lu x 9.2 10 3 10.7 12 .2 9.4 7.0 7.3 III 4.6 14.(1 X7 12 .1 13.0 10 .1 7.0 S3 34 72 S7 12 6 13.3 10.7 12 .2 66 X6 11.9 11.4 14.6 10 9 14.6 15 9 II 4 99 -- 11.0 II 3 12 7 XX 12 X 11.4 il 1 II 9 III 6 II 0 X II 78 14 1 ss 12 1 10 5 12 .0 12 6 7X SU 42 72 X6 II 9 1 2 .K 9.5 12 .1 5X 94 10 5 12 0 14.5 9X 15 ' 15 2 12 6 III 8 Il 1 10 5 III 9 ` Data arc not necessarily lor the whole 5 years. Source D oll el ut. ( I I ) ; Doll et at. (I2 |; Waterhouse el at. 1131; Waterhouse el tit. |IO |; M uir cl til. | l | A stronger inverse relation emerged from a study in relationship was found in ihrcc other studies from Ihe Shanghai, China (15j, where very lute menarche (i.c., at United Slates 12.1-25j and one from Norway |2h|. age 18 or over) was not unusual. Finally, no apparent Available evidence thus indicates that age at menarche 14 PARAZZINI ET AL. TABLE 6 Percentage Diatrbuon o f Ovarian Cancer C a n o and C ontrols la the Upper Educational Level ( > 12 Y ea n o f E d u ca tion ) in Selected Cma Control Studies Auibor [Ref.j Ovarian cancer Control W o t |44| W yw kr a Ml [23| Frawxachl et d (21) Cramer t* d . (49] Tzonou et d . |29] CASH Study [16) Harto ar d . 153]* Cramer et d . |72| Shu et d . (I5 r Parazzini et d . (I7| 113 (30)23.0 (38) 21.0 (34) 26.0 (38) 17.0 (26) 47.0 (231) 50.9 (59) 60.4 (142) 14.5 (25) 14.7 (93) 8.8 (18) 1511 (44) 20.0 (113) 27.0 (38) 19 0 (48) 52.2 (2207) 61.4 (97) 51.9 (124) 2 9 (5) 15 6 (254) The oorrciponding number of eme and control arc hnwn in parentbeies. * Borderline tumor*. ' f> < 0 01 TABLE 8 Relative Risk Estimates o f O varian Cancer and Age at N atural Menopause Age at natural menopause Author |Ref.) <4iT 41-45 46-49 50-54 >55 Wynder t t d . (23] No association Anncgcn ei al. (28| 1.1 r Hildreth et al. (24] 1* 2.0 23 2.9 Francetchi et ul. (211 1` 2.9 4 6 Tzonou et al. (29) 1* 0.7 1.6 IS Ha rige el al. (30] No association Kvale et al. |26]* r 1.0 10 1.0 Wu el al. |2 2r L I 1.3 1.4 Booth et al. (20) i* 2.0 2.5 Parazzini pr ut. (17] i* 13 IS 17 Shu et at. [IS] I* 28 16 21 ` Reference category. * Cohort study. ' Reference category: premenopausal women. has very little if any effect on ovarian cancer risk in de veloped countries. The possibility that very late menarche may exert some protection deserves further attention in studies from developing countries. Age at menopause. At least 12 studies [15.17,2024,26-30] considered the relationship between age at menopause and ovarian cancer risk (Table 8). In seven (15,17,20-22,24,29] a direct relationship was found. Such direct association with age at menopause tends to be long lasting, being present also among older women (17]. The strength of this association, however, seems to vary across different studies, ranging from 1.4 [22] to 4.6 [21j in the highest category of age at menopause. In four TABLE 7 Relative Risk Estimates o f O varian C ancer and Age at Menarche Age at menarche A ulhur |R cf| <12 12- 14 >14 >17 W ynder et al (23| H ildreth et ut (24| Franceschi a al [2 l| Tzonou pr at. |29| Kvale et at [26|* Wu etaJ |22| Booth pr ul |20| Parazzini et al. 117] Shu et at. 115] G w inn et at. |25| 1* 1.5 I I N o t .igniricani 1.5 1 3 1" r 1.2 1.5 1.9 r 1 o 1.0 I* 0.9 0.6 1.3 0.9 1- 1.1 1- 4 4 3 4 3 2 1* 0 9 1* 1.0 Reference category. * Cohort study other studies (23,26,28,30] no association emerged be tween age at natural menopause and ovarian cancer risk. Such large differences are partially explainable by means of potential methodological problems, including definition of menopause or type of menopause, different age distribution, or adjustment for age and other covar iates in the analysis. The overall evidence tends to weigh in favor of a real, although moderate, direct relationship between age at menopause and ovarian cancer risk. Reproductive Factors Number o f pregnancies. Nulliparity and, even earlier, single marital status (see Wynder et al. (23]) have been very consistently related to high risk of ovarian cancer. A strong inverse association between completed family size and mortality from ovarian cancer in different coun tries and for successive cohorts in the same country has been reported (19]. The most debated issue is whether nulliparity and low purity per se or difficulty in conceiving facilitates the de velopment of ovarian cancer. Unfortunately, infertility cannot be measured accurately in most epidemiological studies [31]. Many infertile women do not seek medical care for this problem, and even more women receive only an incomplete and inconclusive evaluation of the causes of their infertility and/or are unable to report accurately on them. Some studies suggest, however, that pregnancies per se reduce the risk of ovarian cancer. Of al least 20 studies that have examined the question ]21 -27,29.32--43]. 16 121.25-27,29.32,33,35-43] showed a further decline in risk associated with full-term pregnancies beyond the first. OVARIAN CANCER. OISTKHtUIION AND RISK I ACIORS 15 TABLE 9 Relative Risk Estimates o f O varia n C ancer and P a rity and Age at f irst B irth Parity A,:e at lirst birth (years) Author | Kef| Wyndcr e ta t: |23J lo ly et at. |32| Ciiagrande et at. |33| McGowan et at. |34| Hildrclh et at. (24 J Cramer ei at |36| Unadjusted Adjusted* Risch et at |.V7| U Vccchia et at. [3X| Unadjusted Adjusted Nascu et at |39| Tzonou et ut. [29| Lcshcr et at. (40) Unadjusted Adjusted1 Voighl et ut |-411 Unadjusted Adjusted Kvalc et a t J |2h| Unadjusted Adjusted' M ori et at (4.3J Wu et at. (22] Gwinn et at. (251 ' 1* r i r i* r r 1.5 1.4 r i 23 4 OH OvK 1.2 2 1 1.3 r OH i* 0 . 0 1.(1 1.6 r 0.5* i* 0.5* r OH i > 13 1.0 I" 1 ll t7 1.3 0.9 0.7 0.9 0.7 0.9 0.7 1 1 . 0 7 1 o 0.9 I I I 0.9 r 0.9 l.l OK 11.7* OH 0.9 0.9 0.4* ,?5 -.20 o.tc 1- 1" 0.3* 0 5* 1" 1" u.S* r 0 S* r r r II.H n.7 iii 0.6 r ll 5 r UK r till r (1 fl* 0 7 0.7* II.H 0.4 10.4* 1 2u 24 25 29 .30 34 14 2 3 II 7 1 4 1 9 3.0* 2.5* l. l 1.2 ID 1 . 0 2 T 2* 4 O' 3 V 2 K* 3 t" U 7 tin n r 0 H 1.2 OH 1 ll <1K 0 9 (Mi 0.9 0.9 1 0 Cl 9 1.3 1 u O.H 11.7 1.2 11 1.2 III 1.4 1.3 1.3 o5 ' Reference category. * 95% confidence interval dries not overlap I . ' Adjusted for parity or age at first birth, as appropriate. 4 Cohort study 35 11 n *i 10 II X thus suggesting (hut additional risk reduction was con ferred by events accompanying each pregnancy. Table 9 summarizes RRs from studies allowing an as sessment of the effect of number of pregnancies and age at first pregnancy in parous women only 29,2124,26,27,29,31-41.43J. It is apparent that the reduction in risk deriving from pregnancies beyond the first is weak and, partly on account of limited study size, it reaches statistical significance in only four investigations 12426,36). Age at first pregnancy. The suggestion that early age at first pregnancy may also protect against ovarian cancer came from four case-control studies (24,32,34,38). Since multiply pregnant women usually start reproduction early in their life, independent effects of age at first pregnancy and number of pregnancies are difficult to establish. An Italian study (38] found a significant excess risk associ ated wilh age at first birth after adjustment for parity. Conversely, in three case-control studies 136,40.41] and one prospective investigation 126). no relationship wiih age at first birih remained after adjustment for parity. Several other investigations did not find a consistent re lationship between risk of ovarian cancer and age at first pregnancy )25,36,29j, but in at least two further studies 20,29) the RR was elevated in women in whom first birth had occurcd at age 35 or more. Furthermore, in these two studies, nulliparous women were not at increased risk over those who first bore a child at age 35 or more. In conclusion, the effect o f age at lirst pregnancy on ovarian cancer is probably weak, if any, but there are various suggestions that the protective effect of full-term pregnancy depends on the age at which it lakes place, and that this protection may be weak, or absent, above age 35. Incomplete pregnancies. Studies dial provided infor mation on ovarian cancer risk according to different oin comes of pregnancies are shown in Table II). Seven in vestigations 117.2II.26.29.33.37.43| showed that the decrease in ovarian cancer risk deriving from incomplete pregancics is not appeeiahly dillcicnl from that confeired 16 PARAZZINI ET AL TABLE 10 Relative Risk Estimates of Ovarian Cancer and Different Pregnancy Outcomes A b o rtio n s Author 1Ret. | West [44] Wynder et at. |23| Joly et al (321 O im grande et al. (33j McGowan et al. |34| Cramer et at. |3| Risch et al. [37] Nasca et al. |3V| Induced Spontaneous No difference 1.1.1 1:1.8 -- No difference 0:1 1:0 54 2 0 54 -- No difference -- 1:0. - 1.07 0:1 Stillbirths -- -- No difference -- No difference -- -- Tzonou et al (29J Kvale et al. \2t\Mor et at. |43| Wu et at. [ 221 Booth et al. (20) Hsrtgc et ai. |46] P araiiini et al. |l7 j Shu et al |15| Whittcmorc et al. [45] 2:11 3 16 01 10 9 2:0.7 109 1:0.5* 1:0.8 No difference 0:1 1:1.0 2:0.6 0:1 1:1 II 2:1 0 3:1 1 0:1 1:0.9 2:0.7 0:1 0: 1 1:0.7 t 1.3 2:0.6 2. 10 No difference -- -- -- * Cohort study. * 95% confidence interval does not overlap I by full-term pregnancies. The comparability with a few other studies in which no association emerged (15,22.32,34,44-461 and others (23,39], which showed nonsignificant inverse associations, is hampered by sub stantial differences in data analysis, such as inclusion of ever pregnant women only [23,32), choice of hospital con trols with benign ovarian tumors or other gynecological and obstetrical complaint or hormone-dependent cancers (32.44), inclusion of stillbirths [15,39|, and, by and large, intention to assess the role in cancer risk of predisposition to miscarriages as a possible outcome of pregnancy rather than pregnancy per se. Oral contraceptives The protection afforded by com bined OCs against subsequent ovarian cancer risk is one of the best established and most important, on a public health scale, characteristics of epithelial ovarian cancer. Available results on this relationship are shown in Table 11. Nineteen out of twenty-one studies [16,20,22,24,25, 27,29,33-35,47-551 found relative risks below unity, and the sole apparent outlier was a study conducted in China (15) that included only 21-pill-users among the cases and 12 among the controls. The overall estimate of protection is approximately 40% in ever users of OCs and such protection increases with duration of use (56). The favorable effect of OCs on lowering ovarian cancer risk seems to persist for some time (at least 15 years) after OC use has ceased, but the data do not provide definitive RR estimates for longer periods or in relation to latency or recency of use. Nulliparity (and perhaps infertility) increases the risk of ovarian cancer and, there fore, the possibility that nonusers tend to be more fre quently infertile than OC users has been claimed to ham per the interpretation of the inverse association between OC use and ovarian cancer [54|. Along this line, OCs were reported to have a more limited protective effect in nulliparous than in parous women in two investigations [34,51]. However, in six other studies (22,29,46,47,49,55) no relations were found between the magnitude of the protective effect of OC use and strata of parity or marital status. Furthermore, no association emerged in published studies between ovarian cancer and barrier methods of contraception or 1UD use TABLE 11 Relative Risk of Ovarian Cancer and Oral Contraceptives Author | Ref. | Relative risk (95% C I) (ever v t never users) New house et at. (27] Casagrande et al (33| McGowan et al. [34) Hildreth et al. (24) Weiss et al [47) W illett et al. (48) Cramer et al (49) Rosenberg et al. [35] La Vecchia et al. (50) Tzonou et al. [29] Ca s h study (i6| Vcsacy et al. [511 Bcral et al [52|* Harlow et al [53]* Wu et at. [22[ Booth et al (20| Hartgc et al. |46| Shu et al. [ ] S] W H O Study (54] Gwinn et al. 25) Para n i ni et ul. (5S| 0.6(0.3-11) 07(0.4-1 1) 0.7(nonsignifk:ant) 0.5(0.2-1.7) 0 6(0.4-1 0) 0 8(04-1.5) 0 4(0.2-1.0) 0 6 (0 4 -0 9) 0 6 (0 3 -1 0) 04(0.1 -1 1) 0 6 ( 0 4 -0 9) 0 3(0.1-0 7) 0 6(0 1-2.3) 0.4(0.2-09) 0.7(0.5-11) 05(0 3-09) 1 0(0 7-1.7) 1 8 (0 8 -4 1) 0 8(0 6-1 0) 0.5(04-0.7) 0 7(0 5-1 0) ' Relative mortality from a cohort study. * Borderline tumors. OVARIAN CANCI.K- |>IS I Mill I'1 1ON AND RISK I At I OKS I (20.55J (although a nonsignificantly reduced risk in IU I) users was reported in a Chinese study j i 51>. Since the incidence of ovarian cancer is already appre ciable in middle age and survival is low, the protection attributable to OC use corresponds to a far from negli gible number of deaths (i.c , probably several hundred per year in countries like Britain (51j. where OC use has been frequent for a long time). Hormonal replacement therapy. A lter a first report in the late 1970s |57|, several studies examined the rela tionship between hormone replacement therapy (in par ticular estrogens) and risk of ovarian cancer (Tabic 12). Two studies 128,-16] showed a lower but not significant risk in women who have ever used estrogens in meno pause. whereas live studies (20,29.36.58.59) found a mod erate increase in risk that was significant only in subgroups (i.c., women who underwent hysterectomy or in some histological subgroups). Some authors reported a significant increase in risk only in tumors of endometrioid type (58,60), but this finding was not confirmed by others (30,59|. One study ( 151 investigated the association between other female hormones and risk of ovarian cancer: ele vated RRs emerged in women who used medroxypro gesterone (2.8). dictil-stilhcstrol (5.4), or other hormones for the treatment of fertility problems (2. 1), but these associations should again be discussed in relation to the possibility of indication bias. Dietary Factors Most of the links between ovarian cancer and diet arc based on international differences or correlational studies. Positive correlations were observed with fats, proteins, and total calorics and are in the same direction as those of endometrial and breast cancer (6I,62|. Information from follow-up studies in selected popu lations is scanty and controversial: Seventh-Day Adven tists. many of whom are lactovegetarians. showed ovarian cancer mortality 40% lower than that of the population of California |63|, but Mormons in Utah had a standard ized incidence ratio of 1.7 compared to the general pop ulation of the United States (64|. In a cohort of British nuns with very low intake p f meat and fats, no reduction in risk was, however, apparent (65|. The results of case-control studies arc somewhat more consistent (see, as a review. J66|). Among four studies (15,67-69) that considered various measures of fat intake (Table 13), three |!5,68,69J found significant direct as sociations. In one of these |I5 |, which was based on a Chinese population with wide ranges of consumption of various foods and was able to provide estimates of intake of several nutrients, the effect of fats persisted after ad justment for total calories, whereas, after allowance for animal fat intake, no significant associations with calour or proteins remained. Within Seventh-Day Adventists, it has been reporter that women who ale eggs or fried foods more frequently had an approximately threefold elevated risk of ovariat cancer |70|. An inereased-risk also emerged for daily men consumption from a case-control study of 56 eases age above 50 conducted in Japan (43|. Studies in Italy |6'>J and North America (69aJ showci reduced risk with more frequent consumption of greet vegetables and carrots and greater vitamin A intake. *Vi lamin ( ' and total vegetables were protective also in . Chinese study ( 151. Conversely, according to Slatleiy < al. (18), calories, Ini, protein, liber, and vitamin A am C dirt not appreciably alter the risk til developing ovariai cancer, whereas high intake o f /-carotene appeared t< confer protection. The mechanism ol action of these im trienls is. however, far Irom being understood. Since worldwide ovarian cancer risk is strongly cone iated with per capita milk consumption and lactase pci sislcncc (i.c.. the ability to digest lactose alter infancy (711. it has been hypothesized that lactose consumption rather than animal fat, may be a dietary risk for ovariat cancer and low transferase activity a genetic marker fo increased susceptibility (72j. The observation of hypci gonadotropic hypogonadism in rodents fed with Inclose rich diet and in galuctoscmic women or carriers with re duced transferase (721provides some biological basis f> the aforementioned associations. which however was no supported by the results of an epidemiological investi gallon (69a(. Alcohol and coffee consumption. Another componcn of diet which has been found to be protective in a larg ease-control study from the United Stales is ethanol |73| This observation finds a possible biological interpretatio in the reduced gonadotropin levels caused by high dost of alcohol consumption with consequent anovulation an hence protection against ovarian carcinogenesis 133.74 The results of other studies are. however, largely incot sisteni. A similar suggestion of a direct association hi tween alcohol and ovarian cancer emerged from a larg American hospital-based study J75). but in three otlu investigations from the United Slates (44,67,76), no sij nilicani association was found In two European sludii |29.77|. the relationship between alcohol consumplio and ovarian cancer risk was above unity with a significai positive trend with amount o f alcohol ingested. These apparent discrepancies could simply be diseussias the results of random fluctuation, but also could I viewed as reflections of a complex series of a sociocci nomic. reproductive, dietary, or oilier correlates of a cohol drinking in different countries. bindings from epidemiological studies oil the rclalioi i 18 P A R A Z Z IN I E T AL. T A B L E 12 Relative Risk Estimates o f O varian C ancer and Eatrogen Replacement Therapy Author | Ref. | Annegert em f. -|28|Hildreth el at |24| La Vccchi* f t at. [60| Endometrioid Weiss ct at |5| Endometrioid Cramer ft at. |3h| Tionou ei at |29| Postmenopausal women Harlow ct at. | *3| Harigc ct al |.T0| Endometrioid type Wu a at. |22| Booih ct at. 12tlJ Postmenopausal women Women repining hysterectomy Nonhyslercctomizcd women Kaufman ct at. |59| Endometrioid Retali ve rink (*>15% C l) (ever v never users) (J.5(0.3-0.9) 0 9(0 5-1 6 ) 0 8(0 4-1.3) 2 3(1 3-5.3) 1 .3 (0 9 -1 .8 ) 3.1(1 0-9.8) 1 6(0 9-2.9) 1.6 0.9 0.6(0 4 -0 8 ) 0 5(03-08) 0 v (n.s.) 1 5(0 9-2 6 ) 10 9(1 7-69.0) 1 2(0 7-2 3) 1 2 (08-1 9) 0 9(0.3-3 0) Comment No trend in risk with duration of use No trend in risk with duration of use No trend in risk RR for cyclic fashion. 2.5 (95% C l - 1 0 -6 4); RR for duration of use longer than 5 yean, 2.8 (borderline significance) Borderline i um ori; no trend in ru k with duration of use No trend in risk with duration o f use No trend in risk with duration o f use No trend in risk with duration of use ship between coffee consumption and risk of ovarian can cer are shown in Table 14. The risk estimates tended to be systematically (although moderately) above unity, and a pooled analysis based on crude data reached an RR of 1.3, with 95% confidence interval from 1.1 to 1.5, for regular coffee users versus nonusers (78J. Although con founding factors did not seem to be a major problem in these studies, and significance was formally obtained with such large numbers of cases, it is not possible to make an inference of causality with such a moderate risk elevation. Family History There are several clinical observations suggesting fa milial aggregations of ovarian cancer. Almost 150 kindreds in which (wo or more women were affected by T A B L E 13 Relative Risk E stim ile * o f O varian Cancer and Fat Intake Level of fat intake Author |Rcf| 1" 2 3 4 Byers cf at |67| _1 1.3 1 . 2 Cramer et al. |6 8 | 1 14 1.9 18 La Vecchia et at |69| 1 12 2 1 -- Shu et al. [I5J 1 I I 1 . 6 19 epithelial ovarian cancer have been published (79-81). Besides the clustering of ovarian cancer, an excess of T A B L E 14 Relative Risk Estimates o f O varian Cancer and Coffee Consum ption Author |Rcf ] Coffee consumption (cups/day) Relative risks* (95% C l) Harigc et al. |96j B ycn el at (67| Cramer el at. |6 8 | U Vecchia el al. |97| Tzonou ct u l |29| Miller et al |98| Whiltemoie ei at. |K| <2 2-3 4 <3 *3 Regular drinkers <1 2-3 >4 05 1 1.5-2 2 5-3 35 1 2 3 4 5 1 2-3 4 1.0(0 5-2 2) 1 6(0 9 -3 6 ) 1.4(06-3.0) 1.3 10 1.1 l. 5 ( 0 .9 - 2 .5 ) 1 9(12-3.0) 2 2(1 2 -3 9) 0.9 1.6 0.9 15 13 1.2 1.3 1.7 1.1 24(1.2-5 1) 23(1.1-4.7) 2 1(1.0-4.4) * Reference category * Never drinkers were the reference category. Wm w OVARIAN < ANO-iK: DISI RlllU I ION AND RISK I A i 'IOKS T A B U i 15 M a in R esults fro m Selected C a s e -C o n tro l S tudies on F a m ily H is to ry o f O va ria n and lirc a s i N eoplasm s and R isk o f O va ria n C ancer Author | Kef. | Fam ily history o f ovarian cancer Fam ily h is lo iy o f Ih im m ca in cr W yndcr el a t |2Af Casagrandc el ul. |33| M cG ow an ei ul. |34| H ild re th ei ul |241 C ram er el ul. |.V>| M o ri ei ul. |43| S ch ild kra u t and Ifiiiin p s o n |K4| Koch ei ul. 1851 P a ra u im el ul. |M6| N o association 7 cases and no c o n tro l reported a la im ly history ot ovarian cancer 14 cases and 2 c o n tro ls reported a fa in tly h is to ry o l cancer o f the "fe m a le reproductive organs" (one-sided. P = <1.01 > RR IK 2 (9 5 % C l. 4.8-69.11) lo r a fa m ily history of ovarian cancer R R 11.3 (95%. C l. 8 . 9 - ID il) lo r a fa m ily liism ry o f ovarian cancer I case and no c o n tro ls reported a fa m ily history ol ovarian cancer H R 3 6 (9 5 % C l. 1.8 -7 .1 ) fo r a h is lo iy o f ovarian cancer in first- and second-degree relatives RR 2 6 for a history o f ovarian cancel m first- and second-degree relatives RR 1 9 (95% C l. 1.1 -3 .6 | fo r a history ol ovanan cancer in first-degree relatives Ni* association RR I I. lo r a lamily history ol h ic a 'l cancer RR 1 .0 (9 5 % C l. u . 4 - 2 5) Io i a history o f breast cancer in mother or sister RR 1.5 (95% C l. 118-2.9) f.r a laimly hisloiy ol In east eamv i 3 cases ami I eom rol icp orli-tl a l.mnlv history ol In cast caiK-ei (H R ii l. 95% ( i. it t Iti7.4 RR I 5 lor a history ol hicast earner in Inst- anl sccoml-ilegiee lelatives RR I <i (95% C l. 1.1 2.A) lot a hisloiy ol breast vainer in liisi -degiee lelanves breast cancer (breast-ovarian cancer syndrome) or more Despite convincing evidence of an increased risk < general excesses of several cancers (including colon and ovarian cancer in women with a positive lamily hisiorv endometrium, " cancer family syndrome" (K2J) have been the percentage ol ovarian cancer cases explained by lamilx described, also. These hereditary patterns arc consistent history is limited and probably smaller than (lie propoi with an autosomal dominant gene with variable penetra lion for other female tumors. For example, less ihuti \' < tion. transmitted through men and women, probably ac- of ovarian cancers were attributable to the "lam ily risf couniing, however, for an extremely limited proportion factors" in a case-control study in Italy |N6| (as compared of eases |83|. lor instance, with the 4 'r esiimaled lor Ineasl cancer) Most case-control studies that included data on family I'pidemiologieal ilala on the relationship between ovai history have confirmed this association |23,24.33,34, tan cancer risk and family history of breast or endometriai 36.43.84-K6J and have provided a quantitative evaluation cancer arc scanty and not completely consistent. An in of risks related to belonging to families prone to ovarian creased risk o f ovanan cancer in women ream ing familial cancer (Table 15). The estimated KRs range from 2 to bicasl cancer cases was found in tom |36. l 3.K5.N6| om III. It must, however, to he borne in mind that, in the of seven case-control studies |2.3,24,.3.3.36.43.85.S6|. and absence of a reliable gcnctical marker of susceptibility, a an association with a family history of endometrial cancel " positive family history" includes both true genetically was found in iwo |24.43| out ol lour studies 124.36.43.,so | determined familial eases and eases occurring by chance, (Table 15) thus resulting in an underestimation of the actual impor tance o f familial predis|)osilic>n. M n lic a l H isto ry a m i O th e r M ist c lh u ic o u s I in tors Discrepancies in investigations on family history can be The relationship between childhood infectious disease'* at least in part explained by chance (the confidence limits ami ovarian cancer was investigated in several studio largely overlap) and by selective publication o f positive 127.34..'36.441. Their results are not consistent, since some results, particularly in earlier studies. Furthermore, po hut not all. showed, among eases, more frequent pasi tential differences in baseline genetic characteristics of history o f mumps |.36,441or occurrence of rubella in ad various populations and in the prevalence of "hormonal" olescence 1341. risk factors should be considered, fhormonal factors can, The rationale lor suspecting ail association between tali in fact, play a role in (he expression of an inherited en ami ovarian carcinogenesis comes from the similarity ami hanced susceptibility of ovarian (issue toward carcino the frequent contamination of tale with asbestos, an im genesis. Notably, no relationship emerged in a survey portant factor in carcinogenesis ol pleural and periloueal conducted in Japan |43| and in a study on borderline mesotheliomas. Case-control studies have. Itowcvci. malignancies |K4|. failed to provide convincing evidence |X7-SU|. 20 PARAZZlNI FI AL CONCLUSION AND ETIOPATHOGENETIC HYPOTHESES The incidence of ovarian cancer is higher in more de veloped countries, although in the past years steady incrcasesjiavc been observed in the. areas where incidence was low, while a battening off of upward trends in more recent cohorts has been recorded in several developed countries. Many factors may be at the origin of such betweencountry variations and their gradual decrease over the past years. Nulliparity or low parity and late age at first birth, oophorectomy, and OC use arc, for instance, more common in industrialized, highest risk areas, but some of these characteristics or medical procedures are increas ingly being adopted by other countries. Besides OC use and parity (the two best ascertained and quantified protective factors for ovarian cancer) casecontrol and cohort studies have shown, or suggested, that late menopause and lifelong regular menstrual periods 117] increase the risk of the disease. In terms of mech anisms, these factors arc thought to act on ovarian cancer risk by affecting lifetime number of ovulations. Along this line. Casagrande el al. (33). following the suggestion of Fathalla (90), proposed that " incessant ovulation" could be the relevant exposure that defines the incidence of the neoplasm. Ovulation exposes ovarian epithelium to recurrent minor trauma and contact with follicular fluid and, according to Zajicek (91). may induce epithelial in clusions of the surface epithelium, relevant to cancer de velopment. Animal data have shown that ovarian epi thelium proliferates rapidly after ovulation and mitotic figures are concentrated near the site of ovulation (92). The subsequent model based on " ovulatory age" (i.e., period from starting to cessation of menses minus "pro tected time" constituted by pregnancies and OC use) is not. however, at least at its present state (33). satisfactory, perhaps on account of different proportions of ovulatory years at different ages. Others (36) have suggested that pregnancy, as well as OCs, may protect against ovarian cancer by reducing total exposure to pituitary gonadotropins (which stimulate the growth of cell lines derived from human ovarian carci nomas (93J). The lack of protection from estrogen re placement therapy (37) provides, however, evidence against the importance of gonadotropin stimulation. Among the nonreproduclive or hormonal factors, diet is likely to be the most relevant; it is still, however, very poorly quantified. It may well explain part o f the sub stantial international variability in the incidence of the disease and. therefore, remains a priority in epidemiolog ical research on ovarian cancer, especially from a public health viewpoint, since it is more amenable to modifi cation than reproductive and menstrual factors. Despite the large number of epidemiological studies, no clear associations between viral factors, chemical agents, and ovarian cancer risk have emerged, thus con firming that the mechanisms of ovarian carcinogenesis should be essentially searched in the hormonal and dietary (together with genetic) fields. Finally, relatively few epidemiological studies have been able to subdivide reliably epithelial ovarian carci nomas according to histological types. Also, the epide miology of borderline tumors and benign cysts has seldom been considered (53,94,95), suggesting, however, differ ences in risk factors between benign and malignant ovar ian tumors. Future epidemiological research should take advantage of improvements in diagnostic standards in or der to cast more light on potential epidemiological dif ferences between ovarian tumors of different histological types and degree of malignancies and, hopefully, to clarify carcinogenetic pathways. ACKNOWLEDGMENT The authors (hank Mrs. Anna Redivo for editorial assistance. REFERENCES 1. M uir. C. S.. Waterhouse, J., Mack, T .. Powell. J , and Whelan, S. Cancer incidence in five continents, V. ARCSet. Publ. 0 (1 9 8 7 ). 2. Levi, F.. Maisonncuve, P.. Filibeni, R.. La Vecchia, C .. and Boyle. P. Cancer incidence and monality in Europe, Sox. Praevenuvmed. Suppl. 2. 34, S1-S84 (1989) 3. Parkin. D M .. Laara, E.. and M uir. C. S. Estimate o f the w orld wide frequency o f sixteen major cancers in 1980, Int. J. Cancer 41, 184-197 (1988) -4. Scully. R. E. Classification of human ovarian tumors. Environ. Health Penpecl. 73. 15-24 (1987). 5. Weis*. N S . Homonchuk. T .. and Young;. J. L ., Jr. Incidence of the histologic types o f ovarian cancer: The U.S. Third National Cancer Survey. 1969-1971. Gynecol. Oncol. J, 161-167 (1977). 6. V icnkko. P . Kauppiia. A ., and V ihko. R. Cytosol and nuclear estrogen and progesterone receptors and 17-0-hydroxystcroid de hydrogenase activity in nondiscascd tissue and in benign and ma lignant tumors of the human ovary. In i 7. Cancer 32, 413-422 (1983). 7. Richman. C. M . H olt. J. A . Lorincz. M. A ., and Herbst. A . L. Persistence and distribution o f estrogen receptor in advanced epi thelial ovarian carcinoma after chemotherapy. Obstet Gynecol 45, 257-263 (1985) 8. Bizzi, A .. Codegoni. A . M . Landoni. F., M arclli, G ., Marsoni, S., Spina, A . M , T o rn , W . and Mangioai. C. Steroid receptors id epithelial ovarian carcinoma: Relation to clinical parameters and survival. Cancer Res. 0 , 6222-6226 (1988) 9. Kauppila, A . V icn kko, P.. Kivinen. S.. Stcnback. F.. and Vihko. R. Clinical significance o f estrogen and progestin receptors in ovar ian cancer. Obstet. Gynecol. 61, 320-326 (1983) 10. Waterhouse, J . M uir, C , Shanmugarainam. K . Powell. J . Peacham, D . and Whelan. S. Cancer incidence in five continents. IV lA R C Set Publ. 42 (1982). ww W OVARIAN CANCER: DISTRIBUTION AND RISK I 'ACTORS 21 11. D o ll, R.. Payne, P . and Waterhouse. J. Cancer incidence in / n r continent!;, I. A technical report. Springer -Verlag. Berlin (1966). antecedents oil the incidence o l ovarian cancer. Am I hpnlemint 128.451-455 (1988). 12. D o ll. R . M uir, C., and Waterhouse. J. Cancer incidence in /tee continents. II. Springer- Verlag. Berlin (19711). 13. Waterhouse. J.. M uir, C , Correa. P . and Powell. J. Cancer in cidence in five continents. I l l , lA K C Set. P u b l. 15, ( 1 9 7 6 ). H . O o t lr R Progress againsr cancer: A rc wc w inning ih c vratTActa Oncol. 28, 611-621 (1989) 32. July. D J . t.ilic iilc ld . A . M . Diamond. I:. L . .nul Itmsx. I D i A il epidemiologic study o f the relationship o f rcpoHluctivc exjc ricncc to cancer o lite ovary, Am. J Epidemiol. 99. l*W 209(1974) 33. fasagi.indc. J. T .. I antic. E. W . |'ik e ..M C . Roy. S.. Ross K. K .. w h f c n l e n m r It: K. "tncessain ovulation" and ......................... . cancer, lamed ii. I7tt 173 (1979). 15. Shu. W. O .. Brinlon, L A . Oao, Y. T -. and Yuan. J. M. Pop ulation -bawd cane-control study o f ovarian cancer in Shanghai. Cancer Res 49. Vi70- 3674 (1989) 34. McGowan. I... Parent. I... I.c d iu r. W . and N oins, I I J The worn.m ai risk for developing ov.inan cancer, (ivneeol. Oncol. 7. 325*344 (1979). 16. The Cancer and Steroid Hormone Study of the Centers for Disease Control and the National Institute o f Child Health and Human Development The reduction in risk of ovarian cancer associated with oriil-contraccpiive use, N Kit#!. J. Med. 316, 650-655 (1987). 35. Rosenberg, I... Shaptio. S.. Slone. D . Kaufman. I). W . Ilcln in ch. S. P . M icttineii. O. S.. Siollcy. I* D .. Kosenshcin. N. S.. Schoi le ule Id. I) . and Engle, R. I.. Epithelial ovarian canec and com bin.ilion oial contraceptives. JA M A 247, 3210 3212 (1982) 17 P .irauini. K . l a Vecchia. C . Negri. K . and Gentile. A . Menstrual factors and the risk o f epithelial ovarian cancer. J. Clin. Epidemiol. 42. 443-448 (1989) 36. fra m e r. I) W . Iluichiiisnn. ( i It.. Welch. W l< . Scully. K I and Ryan, K. J. Dciciiim ianis ol **anan canec usk I Kcpio ductive experiences and family hislory, JN C I 71. 711-716 (198'). 18 Slattery. M. L , Schuman. K. L .. West. D. W .. French. 1' K , and Robison. L M Nutrient intake and ovarian cancer. Am J. Epi demiol. 130. 497-502 (1989) 37. Risch. II. A .. Weiss. N. S.. l.yon. J. I... Daling. J. R . and L ilt J. M Events of reproductive life and ihc incidence o f epilhchal ovarian cancer. Am. J. Epidemiol. 117, 128-139 (1983). 19. Bcral. V .. Fraser. P . and Chilvcrs. C. Dies pregnancy protect against ovarian cancer'' l.nneet I, 1083-1087 (1978). 2d. Booth. M . Bcral. V.. and Smith. P. Risk factors for ovarian cancer A case-control study. Br. J. Cancer 60, 592-598 (198*7). 21. Franccschi. S.. La Vecchia, C . Helm rich. S. P.. Mangioni. C .. and Tognoni, ( i . Risk factors for epithelial ovarian cancer in Italy. Am. J Epidemiol 115.714-719 (1982) 22. Wu. M L .. Whirtcmorc. A. S., Palfengargcr. R. S.. Jr.. Sarlcs. D l... Kampert, J. H.. Grosser, S.. Jung. D. L., Ballon. S., Hen drickson, M .. and Mohle-Boctani, J. Personal and environmental characteristics related to epithelial ovarian cancer. I Reproductive and menstrual events and oral contraceptive use. Am. J. Epidemiol. 38. I.a Vecchia. Dccarli. A .. Franccschi, S.. Regado. M .. and Tog noiti. G. Age at first Ini Ih and (lie risk of epilhchal ovauan cancer JNCt 73. 66.3-61* (1984). 39. Nasca. P. C . Grccnwatd. IV. fh o io s t, S.. Richaii K . ami Capillo T. An epidemiologic ca.se-co n iro l sim ly o f ov.in.ui cancel and re productive laclofs. Am. J. Epidemiol 119, 705 71.1 (1984). 4H. I vsher. I. . McCiow.in, I. . Haitgc. P . and Hoover. R. Age at Ills! birth and risk o f epithelial ovuiian cancer. JN ( I 74, 1361 1.362 (1985). 41. Voighl. 1.. F . Harlow. It. I... ami Weiss. N S Die nillucncc ol age at hist birth and parity on ovarian cancel risk, Am. J . l-.pnlemnd 124. 4*8)-49I (1986) 12. 1216-1227 (1988). 42. M on. M .. ktyosuwa. II.. and Miyake. I I Case control study ol 23. Wynder, b` I... Dodo. H .. and Barber. II. R. K. Epidemiology of ovarian cancer in Japan. Cuneer 53. 2746 - 2752 (1984) cancer of the ovary. Cancer 23, 352-370 (1969). 43. M ori. M .. Ilarahnchi. I . Miyake. I I .. Casagrandc. J. T .. Ilcn dci 24 H ildreth. N. G . Kelsey. J I... LiVolsr. V. A .. Fisher. D. U . son. It. b .. and Ross. R. K Reproductive, geneIif . and dictate Holford. T. R . Mostow. E. D , Schwartz, P. E.. and White. C. An epidemiologic study ol epithelial carcinoma of the ovary. Am. risk factors for ovarian cancer. Am. J. Epidemiol 128. 771- 777 (1988) J. Epidemiol 114. 398-405 (1981). 25. Gwinn. M l... le e , N. ('.. Khoses. P. If .. I^iydc. P M .. and Rubin, Ci. I.. Pregnancy, breast feeding, and oral conn aceptives and the risk o f epilhchal ovarian cancer, J. Clin. Epidemiol. 6 , 559568 ( I99U). 26. Kvulc, G .. Heuch. I., Nilssen. S.. and Bcral, V. Rcprinluctive 44. West. R I ) Epidemiologic simly o| malignancies of llic ov.iiic . Cancer 19, |H0l 11X17 (1960). 45 W hillcm ore. A . S . Wu. M l . Pallenhargci. K. S , Salles. D I Rani|>crt. J. IE. Grosser. S.. Jung. I) E . Ballou. S.. Hendrickson. N , and Mohle-lhielani. J. Epithelial ovarian earner and (he alulns to conceive. Cuneer Res 49. 4047-4052 (1989) factors and risk of ovarian cancer: A prospective study. Ini. J. Cancer 42. 246-251 (1988) 27. Ncwhouse.M l... Pearson. R M . Fullerton, J. M .. Boesen. E. A . and Shannon. U S A case control study o f carcinoma o f the ovary. Br. J Prev Soc. Med 31. 148-153 (1977) 28. Anncgers. J F , Strom. H .. Decker. D. G ., Dockcrty. M B.. and O'Fallon. W. M. Ovarian cancer: Incidence and case-control study. Cancer 43, 723 - 729 (1979). 29. Tzonuu. A ., Day, N. E.. Trichopouios. D ., Walker. A . Saliaraki. 46. Harlgc. P.. Sclutlmnn. M H.. Hoover. R . McGowan. I. . I .eslei I . .. and Norris. H. J. A case-comml study o f epuhelial ovauan cancer. Am. J. O bi tel. (iyneeol. 161, 10-16 ( 1 9 8 9 ). 47 Weiss. N. S.. l.yon. J I... I .iff. J M . Vollm cr. W M .. and Dating J. R Incidence of ovarian cancer in relation to ihe use of oral conlliiccptivcs. Int. J. Cumer 28. tiW-67| (1981). 48 W illett, W. Bain, Hcmickcns, C. I I . Kosncr. IE. and Spoizer, F. E. Oral contraceptives and risk o f ovarian cancer. Can cer, 48. 1684- 1687 (|9 8 l). M .. Papapostulou. M . and Polychronopoulou. A . The epidemiol ogy o f ovarian cancer in Greece: A case-control study. E ar J. Cancer Clin. Oncol 20, UM5- 11152 (1984). 30. fla ng e. P.. Hoover. R . McGowan. L.. l.esher. I... and Notns. II. J. Menopause and ovarian cancer. Am. J Epidemiol. 127. 990998 (1988) 31. Weiss. N. S Measuring ihe separale effects of low parity and iis 49 fra m e r. D. W.. Hutchinson, ( i IE. Welch. W R . Scully. R. E and Knapp. R. C. Factors .iflccling the association n| oiul conii.i ccptivcs and ovauan cancer. N. In n ! I. Med. 3417, 1047 |05l (1982) 50. I.a Vecchia. C . Franccschi, S.. and D ccaili. A . D ia l coiiiraccpiic. use and ihc risk ot epuhelial ovarian cancel. Hr J C u rtir/ 5<J, 31 34 (1984) 2 2 PARAZZINI ET AL 51. Veey. M . Metcalfe. A .. Well. C . McPherson. K.. W eithotf. 70. Snowdon. D A Diet and ovarian cancer, JAM A 254, 356-357 C.. and Yale, O. Ovarian neoplasm, functional ovarian cysts. and (1985). oral contraceptive, Br. Med. J 294, 1518-1520 (1987). 71. Cramer. D. W. Lactase persistence and milk consumption us de 52. Bcral. V .. Hannaford. P . and Kay. C. Oral contraceptive uie and malignancies of the genital tract. Lancet U, 1331-1334 (1986). terminants o f ovarian canper risk. Am. J. Epidemiol. 130, 904-910 (1989). 53. Marlow. 8 L . Weiss, N $.. Roth, G. J.. Chu, J.. and Dating, J._R.-Cac-control study o f borderline ovanan turnon: Reproduc tive history and exposure to exogenous female hormone*, Cancer Hey 41, 58*9-5852 (1988). 72. Cramer. D. W . Harlow. B L .. W illett. W C .. Welch. W R . Bell. D. A . Scully. R E , Won, G Ng., and Knapp. R . C Gal uosc consumption and metabolism in relation to the risk of ovarian cancer. Lancet U, 66-71 (1989). 54 The WHO Collaborative Study of Neoplasia and Steroid Contra ceptives. Epithelial ovarian cancer and combined oral contracep tive. Ini. J. Epidemiol. I t , 538-545 (1989) 55. Paruzini. F . La Vecchia. C .PNegri. E .. Bocciolone. L . and Fedele, L. Oral contraceptive use and the risk of ovanan cancer: A n other Italian case-control study, Eur. J. Cancer 27, 594-598 (1991). 56. Prentice, R. L.. and Thomas, D. B. On the epidemiology o f oral contraceptive and disease. Ad*. Cancer Hes 49, 285-401 (1987). 57. Hoover, R., Gray. L. A ., and Fraumcni. J F . Jr.. Stilhocttrol (diethylstilbestrol) and the risk of ovarian cancer. Lancet H, 533-- 534 (1977). 58. Weiu. N. S . Lyon. J. L , Knshnamurthy. S . Dictert. S. E . L ift, J. M , and DaJing. J R Nonconi raceptivc estrogen use and the occurrence o f ovanan cancer. JN C I 68, 95-98 (1982). 59 Kaufman, D W . Kelly. J P . Welch, W R . Rosenberg. L.. Stoliey. P D . Warshauct. M. t , Lewi. J . Woodruff. J.. and Shapiro. S. Nonconi receptive estrogen use and epithelial ovarian cancer. Am. J. Epidemiol 134, 1142-1151 (1989). 60 La Vecchia. C . Liberati. A . and Franceschi. S. Non contraceptive ,estrogen use and the occurrence of ovarian cancer. JNCI 69 1207 (1982). 61. Annstrong. B K . and Doll, R. Environmental factors and cancer incidence and mortality in different countries, with pedal reference to dietary practice. Ini. J Cancer 15, 617-631 (1975). 62. Rose. D. P.. Boyar. A . P . and Wyndcr, E. L. International com parisons of mortality rates for cancer of the breast, ovary prostate and colon and per capita food consumption. Cancer S I, 2363-2371 (1986) 73. Gwinn, M. L.. Webster. L. A .. Lee, N. C .. Laydc. P. M .. Rubin. G. L.. and the Cancer and Steroid Hormone Study Group. Alcohol consumption and ovarian cancer risk. Am. J. Epidemiol. 123, 759766 (1986). 74. Mello. N. K .. Bree. M. P.. Mcndclson. J H .. Ellingboe, J.. King. M. W ., and Sehgal. P. Alcohol self-administration disrupts repro ductive function in female macaque monkeys, Science 221, 677-679 (1983). 75. Rosenberg. L.. Shapiro. S.. Slone. D .. Kaufman. D. W .. Ilelm rich. S. P . M tcllincn, O. S-. S lo lky. P D .. Levy. M .. Rosenshein. M B . Schottcnfcld, D , and Eagle, R. L. Breast cancer and al coholic-beverage conkumption. Lancet i, 267-271 (1982). 76. Williams. R. R.. and Horm. J. W. Association of cancer sites with tobacco and alcohol consumption and socioeconomic status o f pa tients: Interview study from the Third National Cancer Survey. JN C I SI, 525-527 (1977). 77. La Vecchia. C.. Negri. E.. Franceschi. S-, Parauini, F.. Gentile. A ., and Fasoli. M. Alcohol and epithelial ovarian cancer. Eur. J. Cancer, in press. 78. IA R C Working Group on the Evaluation o f Carcinogenic Risks to Humans. IA R C monograph evaluation carcinog. risks hum. coffee, tea. mate methylxanthines and methylglyoiul, IA R C , Lyon, Vol. 51. pp. 149-152 (1991). 79. Lurain, J. R . and Piver. M. S. Familial ovarian cancer, Gynecol Oncol 8, 185-192 (1979). 80. Piver. M. S . Barlow, J. J., and Sawyer. D. M Familial ovarian cancer: Increasing In frequency? Obnet. Gynecol. 60, 397-414) (1982). 63. Phillips. R L . Garfinkcl. L.. Kuzma. J. W.. Beeson. W L.. L ou . T.. and Bnn. B. Mortality among California Seventh-Day Adven tists for elected cancer sues. JNC! 65, 1097-1107 (1980). 81. Piver, M. $ . M ettlin, C. J., Tsukada. Y.. Nasca. P., Greenwald. P.. and McPhce, M . E. Familial ovarian cancer registry, Obsiet Gynecol 64. 195-199 (1984). 64 Lyon. J L.. Gardner. J. W . and West. D W. Cancer risk and life-style: Cancer among Mormons from 1967-1975. in Cancer in cidence in defined populations (A Cairns. F. L Lyon, and M. 82. Lynch. H T .. Albano. W.. Black. L . Lynch. J F . Recaharcn. J . and Pierson. R. Familial cicess of cancer of the ovary ami other anatomic sites. JA M A 245, 261-264 (1981). Stolnick Eds ). Cold Spring Harbor Laboratory. Cold Spring Har 83. Lynch. 11. T .. Schuelkc, G. S., Wells. I. O icng. S. C.. Kimhcrling. bor. pp 93-98 (1989). W. J . Biscone. K. A ., Lynch. J F.. and Danes. B S. Hereditary 65. Kinlen. L. J Meat and fat consumption and cancer mortality: A ovarian carcinomas. Biomarker studies. Cancer 55, 410-415 (1985). study of strict religious orders in Britain, Lancet 1, 946-949 (1982). 84 Schildkraut. J. M ., and Thompson. W. D. Familial ovarian cancer: 66 La Vecchia, C Nutritional factors and cancers of the breast en dometrium and ovary. E u i J. Cancer Clin Oncol. 25. 1945-1951 A population-based case-control study. Am J. Epidemiol. 128, 456-466 (1988) (1989) 85. Koch, M.. Gaedkc. H . and Jenkins. I I Family history of ovarian 67. Byers. T . Marshall, J.. Graham. S.. M ettlin. C . and Swanson M. A case-control study of dietary and nondietary factors in ovarian cancer patients; A case-control study. Ini. J Epidemiol. 18. 782785 (1989). cancer. JNC/ 71, 681-686 (1983) 86 Parauim. F.. Negri. E . La Vecchia. C . Rcstclli. C . and Fran 68 Cramer. D W . Welch. W R . Hutchinson. G B W illett. W C., and Scully. R E. Dietary animal fat in relation to ovarian cancer ceschi, S. Familial ovarian cancer: Evidence for an association from an Italian case-control study. Am J Epidemiol, m press. nsk. Obiiet Gynecol 63, 833-838 (1984). 87. Longo, D L . and Young. R. C. Cosmetic talc and ovarian cancer. 69. La Vecchia. C.. Dccarli. A .. Negri. E.. Parazzint. F . Gentile, A .. Lancet U, 349-351 (1979) Cecchetti. G .. Fasoli. M . and Franceschi. S. Dietary factors and 88 Hartge. P . Hoover. R . Leshcr. L. P . and McGowan. L. Talc and the risk of epithelial ovarian cancer. JNCI 79, 663-669 (1987). ovarian cancer, JA M A 250. 1844 (1983) 69a. Engle. A . Muscat, J. E . and Harris. R. E. Nuiational risk factor 89 Whutcmore. A S . Wu. M L . Paffcnbargcr. K S . Sarles, D L.. and ovarian cancer. Nuir. Cancer 15, 239-247 (1991). Kampcn. J B . Grosser. S.. Jung. D. L., Ballon. S . and Men O V A R IA N C ANCER: l>IS I K IIIU 'I IO N A N D RISK I ACTORS drtckson. M. Personal and environmental characteristics related it epithelial ovarian cancer. Am. J. Epidemiol. 128. 1228-1241) (1988). 9U. Fathalla. M. F. Factors in the causation and incidence of ovarian cancer. Ohuel. Gynecol Surv. 27, 751-768 (1972). 91. Zajicck. J. Prevention o f ovarian cystomas by inhibition o f ovu lation: A new concept,. J. Reprod,. M rJ 20. 114 (1978) 92. Stein, K. F.. and A llen. E. Attempts to stimulate proliferation of the germinal epithelium if the ovary. Anal. Rrc. 82. 1-9 (1942) 93. Simon. W. h .. Albrecht. M .. Hansel. M .. Dietel, M .. and llo lze r. F- Cell lines derived from human ovarian carcinomas: G rowth stim ulation by gonadotropic and steroid hormones. JS C ! 70, 839-845 (1983). 94. Koch. M Risk factors in ovarian cancer o f epithelial origin by palhological diagnosis. Hr. J. Cancer 59, 257-258 (1989). 95. P au z/in i. I '., l.a Vecchia. C\. Franccschi. S . Ncgti. I :.. ami ( > chclti. (J. Risk faciors lor endometrioid, mucinous and yrrotis lie nign ivarian cysts. Ini J. Epidemiol 18. 1-5 (1989). 96. I lunge. I.. U sh er. I. IV. McGowan. 1... ami ll.n .ve r. R Colic and ovarian cancer. Ini J. Cancer 3d. 531-532 (1982). 97. I ji Vecchia. ( '.. Fraiiccschi. S.. Dccarli. A .. G entile. A . I.iati. I ' . Regallo. M .. and Tognom. <i. Coffee drinking and the risk of epithelial ovarian cancer. Ini J. Cancer JJ, 559-Sn2 (I9KJ). 98. M iller. D. R.. Rosenberg. L . Kaufman. I) W .. Ilekmrich. S I' . Schotlcnfcld. I).. le w is. J.. Slollcy. P. I).. Roscnshcin, N.. and Shapiro. S Epithelial ovarian cancer and coffee drinking. Ini. J Epidemiol. 16, 13-17 (1987).