Document oejqoYKrXyzyz8nbY1g9ZVR8

OF EPIOEMIOLOGY . Thc Johns Hopkina University .%hod ofHwicnc and Public Health "--_- -- VOL 121. No. 5 Printed VI U.S.A. A CASE-CONTROL INQUIRY INTO THE ETIOLOGY OF HAIRY CELL LEUKEMIA Oieske, 0. (Hematology/Oncology Section, U. of Chicago, Chicago, IL 60637). H. M.Gdomb, M.0. F W ,md P.S. Levy. A cara-eontrdinquiry intothe etiology of hairy cell leukemia. Am J pidemkl19(55;121:67C83. A casecontrol study was codwted to investigate the etiology of hairy cell kukemia. Forty-five individuals with ktoto@hRyconfirmed hairy cell kukemia Oomprised the cases Tlweo controls were selected on the basis d age (fS years), race, sex, md rstidanceat the time of the intemiew for each case. There was no association f d for cigarette smokiig, alcohol or coffee consumption and hairy cell leukemia. With respect to occupational risk factors, memployment in woodworking or h farmi- was of borderline significance. Re- portedexposure to chemicals hthe workplace was signiiantly greater mang both sexes ef the cases than among their respective matched controls (relalive risk (RR) = 3.10). Other vutsbles found to be significantly associated with hairy d leukemia were farm (RR = 4.20). anemia (RR = 4.29), m b r e h (RR = 4.80h intedious- w ' (RR = 9.00). and routine use of aspirin(RR = 3.41) or tranquilizers(RR = 4.50). No association between radiation and hairy cell leukemia was detected. hematologic diseases leukernla;kukemia, hairy cell Over the past three decades, only small strides have been made in unraveling the causal pathway of leukemia. In reviewing the extensive literature on the epidemiology of this disease, Szklo (1)states that one of the main reasons for the slow progress Received for publintion July 6. 1984, and in final for'mHSeep~te/mObenr 1 c 1 o . l 1 o ~ g . y Section, Univeraity of Chicago, chicago,IL. '-nt of Ophthalmology. University of I11i3yi& C h i , IL. EQidcmidogy and Biometry Program,School of public Health, University of Illinois. Chicago, IL. h l i a t requests to Dr. Denise Oleske. Hematol- Ogy/OmlwSection. Universityof Chicago,Box 420. S. Maryland, Chicago, 1L 60637. %is study was completed as part of the require- ments for the Doctor of Philosophy degree, University of Illinois, Sehool of Public Health (D.0.). ' h s investigation was supported in part by devel- Wmntal funds from the National Cancer Institute, DHHS (CA 'BOX),awarded through the Illinois Cancer-~. . Thc authors wish to acknowledge the assistance of A n d m Kubilius, Levis Segal, and Mary Shepherd in Programming of the data for analysis. in understanding the etiology of leukemia is that most studies fail to examine risk in terms of exact cell type. Indeed, when leukemia risk factors are examined by histologic type, clearer and more precise relationships have been noted (2). Thus, epidemiologic studies of leukemia would be strengthened by further investigation of its subtypes. One histologic type for which there is little epidemiologic information is hairy cell teukemia, a rare and, typically, chronic form of leukemia. Despite the fact that it was fully described in the late 1950s,few risk factors for it are known (3). The only report of the relative frequency of the disease is by Bouroncle (41, who estimated that hairy cell leukemia comprises 2 per cent of all leukemias seen, based on the experience of a single hospital. The literature cites the age range of patients to be from the early twenties to the eighties. The 675 i i ,' f b 676 OLESKE ET' A L percentage of females in case series has leukemia registry which contained most of been observed to range from 18 per cent to the cases in our target area. 38 per cent (5-7). With respect to the etiology of hairy cell MATERlALS AND METHODS leukemia, only two suspect factors were discerned from a review of the literature: ionizing radiation and heritability. In a study of a cohort of individuals exposed to an atmospheric nuclear test, hairy cell leukemia was one of six histologic types of leukemia found to occur (8).Although the number of cases was greater than that expected(with all histologic types combined), only 60 per cent of the original cohort had been located at the time of publication. Stewart and Keating (9) found a higher frequency of radiation exposure among cases with hairy cell leukemia than among a control group comprised of individuals with solid tumors. Two case reports comprise the evidence for suggestingsome heritable component in Cases were obtained from the Hairy a l l Tumor Registry and Treatment Center at the University of Chicago, a unique regional resource. Individuals with hairy cell leukemia come to the center's attention from physicians seeking consultation confirm the diagnosisor for assistance with treatment and follow-up. The majority of these cases are referred from hematologists throughout Illinois. The diagnosis of hairy cell leukemia based upon clinical and laboratory param. eters. The most common initial clinical sign is splenomegaly, frequently massive, in as many as 93 per cent of the cases reported Lymphadenopathyand/or hepatomegaly is uncommon in this neoplasm. The onset of the disease is insidious, usually first iden- the development of hairy cell leukemia. tified as a nonspecific complaint of fatigue Cohen et al. (10) report the identification or weakness, with pancytopenia present in of hairy cell leukemia in association with a two thirds of patients (12). The diagnosis family "cluster"of lymphoproliferativedis- can be made from the peripheral blood, orders. In 1981, Ramseur et al. (11) pub- spleen, liver, bone marrow, or involved lished a report of a father and son who were lymph node, because there are no differ- diagnosed with hairy cell Ieukemia. Several ences in the morphologic features of the aspects of the clinical history of these two hairy cell at these sites. However, a bone individuals implicate a familial tendency. core biopsy is the diagnostic procedure of Despite the limited amount of epidemi- choice (12). The fmely diffuse network of ologic information on this disease entity, it hairy cell growth in the bone marrow is appears from a review of the literature that characteristic of this disease (13).In addi- hairy cell leukemia could share some risk tion, pseudosinuses distended with eryth- factor characteristics common to other rocytes in the spleen and hepatic angio- forms of leukemia. In an attempt to under- matous lesions are pathognomonic features stand more about hairy cell leukemia, the of hairy cell leukemia (14). strategy of this investigation was to deter- Cases for this study consisted of those mine which (if any) known risk factors individuals histologically confirmed with associatedwith the other forms of leukemia hairy cell leukemia at the University of were also associated with hairy cell leuke- Chicago between January 1,1975 and July mia. From a review of the literature, the 31, 1981.In addition, cases were limited to major risk factors for the various types of those who were residents of Illinois or leukemia, including those suggested for northern Indiana, because it was deter- hairy cell leukemia, were identified. These mined that the majority of registry cases factors served to guide in the formulation were referred from these areas. In an effort of this study. This investigation was made to identify all cases of hairy cell leukemia feasible by the existence of a hairy cell in the region diagnosed during this period. an informational packet on hairy cell leu- kemia was developed and mailed to a total of 1,100 hematologists, pathologists, and medical oncologists in n l i o i s and to those l o c a d in adiacent mm of Wisconsin, Iowa, Missouri, Indiana, and Michigan. Through this mechanism, 10 additional cases were located, c o n b e d as having hairy cell leukemia, and added to the regis- cases receiving appropriate initial therapy (5) (table 1). Three controls were selected on the basis of age (+5 years), sex, and race for each case. Neighborhood controls were used in an attempt to compensate for any biases possibly operating in the referral of the cases.Controlswere selected from lists constructed from city directories, produced lo- I - _- .. ..*-- r ., I try. There were a total of 53 cases who met d l y (homeowner's associations, chambers of commerce),or from field listing (for rural the eligibility criteria regarding date of di- agnosis and residence at time of diagnosis. The residence distributionof these cases at the time of their diagnosis reflected the population distribution of the area. Six areas). The starting point for attempting to identify eligible controls was the entry on the liit representing an address consecutive to the index cases for which makhed controls were sought. A phone call was made to individuals on the lists for the cases were dropped from the study because purpose of identifying eligible respondents. no relative could be located to respond on . At the time of the phone call, a screening behalf of the case (two cases who were interview was conducted with any identifi- mentally incompetent and four deceased cases);two refused to be interviewed (one able adult who answered. Individuals not at home were retelephoned up to two more living case and a spouse of a deceased case). times at hours and days different from the The remaining 45 persons comprised the initial attempt, before the next entry was study cases. Of those participating, 36 were selected from the list. Of the total p l of individuals with hairy cell leukemia and controls found to be eligible for the study nine were proxies for the cases. Proxies by the telephone screening interview, 18.6 we= either a spouse or an offspring, per cent refused to participate in the per- whoever was felt to be the most reliable sonal interviews. An average of 14 screen- I mpondent. The high percentage of cases ing phone contacts were made to obtain repOrting for themselves (80per cent) re- one eligible control who consented to par- flects the high survival rates found among ticipate in a personal interview. TABLE 1 Dktribution 4eligiblc cases at the Hairy Cell Tumor Registry, Januoy 1. 19fSJuly31. 1981 A 196-item structured questionnaire was administered to cases and controls. In the questionnaire, the period of inquiry for exposure factors of the cases began one year Cabnr No. % before the diagnosis of hairy cell leukemia Alive and went back in time. All variables con- Participated in interview Mentally incompetent/no next of kin available Mentally incompetent/spouse interviewed 36 90.0 2 5.0 1 2.5 sidered lifetime exposure, with the exception of drug usage and exposure to diagnostic radiation procedures which only covered the 10-year period beginning one calendar interview 1 2.5 i6zE year before the diagnosis of hairy cell leukemia. To have the cases and controls com- Relative participated in interview Not abk to locate next of kin hlative refused interview Tohi eligible c a w 8 61.5 4 30.8 - -1 7.7 13 100.0 53 100.0 parable with respect to any recall biases concerning drug usage or diagnostic x-ray exposure, the interviewers were given written instructions to cover the same 10-year interval for the matched controls as was I 678 OLESKE ET AL. TABLE 2 -Demographic chamcteristics among h a i y cell leukemia cases and mntrds. 1975-19S1 Cases ( n = 451 Controls ( n 1 3 0 Charactenstic NO. t No. % Racelaex White malea White females Nonwhite males Nonwhite females Marital status Mamed Unmamed Religion Catholic Jewish Protestant Other None Income S9.999 t 1o.OOo-519.999 t20,000-t29,999 r$30.000 Refused Median Education High school graduate or leas Some college college graduate Median Mean age at interview (years) Mala Females Combined 33 73.3 10 222 2 4.5 0 0.0 40 88.9 5 11.1 16 35.6 4 8.9 22 48.9 1 2.2 2 4.4 9 20.0 13 28.9 10 22.2 11 24.4 2 4.4 $19,722 22 48.9 9 20.0 14 31.1 13 years 56.6 53.7 55.9 98 73.1 30 22.4 6 4.5 0 0.0 118 88.1 16 11.9 52 38.8 4 3.0 66 49.3 3 2.2 9 6.7 23 17.2 31 23.1 30 22.4 37 27.6 13 9.7 522.321 64 47.8 35 26.1 35 26.1 12 years 56.5 53.7 55.9 covered for their respective cases. A complete work history was obtained from each study subject and recorded on the interview form. To be classified into an occupational category, the individual must have received payment for services or products produced and must have worked a minimum of 20 hours per week in the position for at least six months. Each employment situation meeting these criteria was assigned a code number corresponding to the appropriate job title as defined in the Dictionary of Occupational Titles (15). The questionnaire used in this study was administered by the study principal inves- tigator (D. 0.) and trained interviewers. The investigator interviewed all the cases and 17.2 per cent of the controls. Ninety seven per cent of the interviews were conducted by personal interview; 3 per cent were conducted by telephone. The multiple logistic model for matched data as described by Breslow et al. (16) was used to obtain point estimates of and 95 per cent confidence intervals (CI)for the relative risk (RR) associated with exposure variables. Analysis was univariate, and the resulting estimates of relative risk were controlled only for the design factors (age. sex, race). Analysis was performed sepa- rately, including and excluding the proxy respondents, using two-tailed tests of significance. Since the magnitudes of the es- timates of relative risks did not change appreciably when the proxy respndents were excluded, only the results for the entire study group are reported. HAIRY CELL LEUKEMIA RESULTS were found to have a suggestiveassociation The final sample consisted of 179 in&- '"*45 and 134 As &- played in table 2, cases and controls were similar with respect to major demographic variables. They were also similar with re- rpect to presence of familial consanguinity with hairy cell leukemia (table 5). Exposure to organic chemicals for two years or more was found to be significantly associated with the disease in the study sample as a whole, as well in males and females when analyzed separately. Ofthe previous diseases to Promote genetic Predisposition anemia, migraine, and infectious mononu- to leukemia)and sibship size (small sibship size is thought to be to watevr'l- cleosis Were found to be with hairy cell leukemia (table 6). Concerning nerability in adulthood because of pro&- routine medication use,aspirin (acetylsali- tion from early chilfiood infection). As cylic acid) and tranquilizers were found to shown in table 3, cws and controls were be associated with hairy cell leukemia (h- similar in their intake of alcohol and coffee ble 6). and in their cigarette usage. With respect With respect to ionizing radiation expo- to residence history, 17.8 per cent of all sure,four types were evaluated diagnostic, cases were born on a farm as opposed to military, therapeutic, and occupational. No 8.2 percent of controls (RR = 4.20; 95 per differences were found between cases and cent CI = 1.04-16.95). controls in exposure to ionizing radiation The distribution of occupational groups from diagnostic procedures or from the for the m s and controls of all j& ever three other sources combined. held full time is shown in table 4. In ex- amining the distribution, we found six oc- DISCUSSION cupational categories in which the proportion of cases exceeded that of controls: service occupations, agricultural work, machine trades, benchwork, structural work, and motor freight and transportation. Within each of the categories, the fre9wncy of codes for specific job types was This preliminary investigation into the etiology of hairy cell leukemia represents an attempt to examine the association be- tween this disease and factors that have been found in previous studies to be related to either hairy cell leukemia or other histologic forms of leukemia. Although the Compiled. Among specific job types which relatively small number of cases, use of mntained at least four cases who were ever proxy respondents, and the time interval in that occupation, only the oc- between event and interview pose threats cupations of woodworking and farming to the validity and reliability of the find- TABLE 3 h t r i b u t w n of study shjects according to smoking status and alcohol and coffwconsumption; hoiv cell 1 leukemia rases and control, 1975-1961 -V W b k (high-/low-risk contrast) Pn and no.in high-risk category CuCr Controls NO. % NO. % m' - smoking (ever smoked 10+ cig-tta per day/ never smoked) Aloaonsbumption (daily consump- tbnn/Wcr drank) f4ffaensumption (J+ cups per day/ e i e a I l y never) 21 46.7 80 59.7' 10 22.2 20 14.9. 16 35.6 55 4 1.O' 'P>0.05. 680 OLESKE ET A L TABLE 4 D i r t a w n Of occupatw~eLvr held' fd time by hauy C d bukpmio eoseS and CQardS1.975-1981 Ocrup.tional poup (DOT+codes) Professional. technical. and managerial 0rmp.tiOns (001-199) Clerical and des occupations (210299) !Service OCCUprtiOM (300-389) Agricultural, fuhcry,and forestry occupations (401-461) Processing occupations (500-559) Machine trades occupations (600- 699) Benchwork occupations (700-795) Structural work occupations (800- 899) Motor freight and transportation oc- cupations (900-919) Miscellaneous occupations (920-979) CvSar No. x 15 333 16 35.6 5 11.1 9 20.0 3 6.7 8 17.8 8 17.8 10 22.2 4 a9 3 6.7 Controls No. R 59 11.0 53 39.6 11 a 2 17 127 20 14.9 18 13.4 18 13.4 21 15.7 9 6.7 12 9.0 - Individuals may have held one or more different occupations. t DOT, Dutiomry of occupational Titles. TABLE 5 Distribution of select occupational facton among hairy cell kukemio cases and contrdr. 1975-1981 Occupational factor C.ra No. R COntroL No. % Relrtin risk (95% ConfIdtnce intml) Occupational subcategory (DOT codes)t Farming (401.410,421) 9 25.7 16 15.4 2.94 (0.80-9.87)$ 4 11.4 3 2.9 4.00 (0.90-17.87)$ t Females were excluded fromthese analyses because there were none who reportedemployment in hrming or woodworking. $ 0.05 <p c 0.10. f p e 0.05. ings, this report represents the first attempt to examine this specific histologic form of leukemia in a study sample comprised of incident cases in a defined geographic area and of matched neighborhood controls. Exposure to radiation, coffee drinking, smoking, and alcohol consumption have been found in previous studies to be associated with certain types of leukemia (9,17, 18). In this study, however, there were no indications that their frequency was greater in cases than in controls. Although acute leukemia is known to be associated with ionizing radiation (19,201, findings from studies investigating the relationship between ionizing radiation and chronic leu- kemia forms, including hairy cell leukemia. have been either negative or unconvincing (9, 21, 22). This could be because of the difficulty in measuring exposure to ionizing radiation retrospectively through inter. views or through review of medical charts. It may also indicate no association, or at most a weak one. P m i w r disease diagnosed by a medical doctor Allergies Anemia Migraine mbem Tube- 9 20.0 18 13.4 1.79 (0.674.77) 13 28.9 13 9.7 4.29 (1.67-11.03)$ 8 17.8 5 3.7 4.80 (1.57-14.67)# 3 6.7 5 3.7 1.80 (0.43-7.53) 0 0.0 2 1.5 0.00 Cancer (excluding hairy cell leukemid Rheumatic fever Hypertension 0 0.0 4 3.0 0.00 1 2.2 5 3.7 0.60 (0.07-5.14) 9 20.0 29 21.6 0.91 (0.39-214) Cudiac disease (other than hy- pcrttnrioa) Stroke Chronic gallbladder problems Stomach ulcer Arthritis (dlforms) Thyroid disorder Chronic skin trauMe Inftctiousm0M)uucIeosis 5 11.1 15 11.2 1.00 2 4.4 6 4.5 1.00 6 13.3 7 5.2 2.89 (0.86-9.69)f 3 6.7 9 6.7 1.00 7 15.6 22 16.5 0.93 (0.38-2.30) 4 a9 io 7.5 1.25 (0.34-4.59) 8 17.8 10 7.5 2.63 (0.97-7.1714 6 13.3 2 1.5 9.00 (1.82-1.(.59)$ Medications routinely used' Vitamins 8 17.8 37 27.6 0.56 (0.24-1.33) Thymid medicine 3 6.7 6 4.5 1.56 (0.36-6.67) Aspirin 12 26.7 12 9.0 3.41 (1.42-8.21)$ Other nonaspirinpain medicine 2 4.4 6 4.5 1.00 Blood pressure pillst 8 17.8 16 11.9 1.59 (0.64-3.98) Tmnquilizcrs Diet pills 6 13.3 4 3.0 4.50 (1.27-15.95)$ 3 6.7 6 4.5 1.63 (0.35-7.63) Honnone pills 3 6.7 9 6.7 1.00 ~~ ~~ Rautinely wed hdefined as consumption tbmtimes a week or more for more than two months. t Includes mcdiclltion~with a combination of pbarmacobgicproperties. # P c 0.05. 5 0.05 <p c 0.10. E x p u r e to agricultural occupations, identified in previous studies as associated with leukemia (23, 24), was found in this study to be marginally associated with the hairy cell type (i.e., the 95 per cent dence intervals for the relative risk ov unity, but the 90per cent confidence i vals do not). Exposure to woodwo also found in this study to be marginally %~ociatedwith hairy cell leukemia, has been linked in previous studies to Hodg- kin's disease, cancer of the lung, and carcinoma of the nasal mucosa (25, 26). Another study suggesting an association between woodworking and leukemia must be viewed as suspect because of small sample size (27). t 1 c I P ~ 682 OLESKE ET A L anulocyte progenitor Cells and in the gnctional capabilities of leukocytes (37) Thus, aspinn ingestion may alter th; .body's ability to recognize and eliminate aberrant cell lines, such as hairy cell leu. mononucleosis to be a significant risk factor among the cases. From a biologic perspective, certain strains of Epstein-Barr virus (the causative agent in infectious mononucleosis) are capable of in vitro transformation of lymphocytes into rapid and persistent growing cells which interact with malarial parasites to produce malignant lymphoma (31). Infectious mononucleosis has also been associated with Hodgkin's disease and acute leukemia (32, 33). Because infectious mononucleosis has been noted to occur before and after the diagnosis of a hematologic malignancy, the significance of Epstein-Barr virus-induced infection is unclear. The following are possible explanations for the association between hairy cell leukemia and infectious mononucleosis detected in this study: a `pre-leukemic" state may promote infection; infectious mononucleosis shares some etiologic factor(s) with hairy cell leukemia such as Epstein-Barr virus; or, early hairy cell leukemia may have been misdiagnosed as infectious mononucleosis. Future study of the association between infectious mononucleosis and hairy cell leukemia would kemia. It is also possible that the signifi. cance of aspirin among the cases may have emerged because of its use to manage s n p toms precipitated by exposure to noxious insults in the work environment (viz., or. ganic chemical exposure). The role of tranquilizers as an etiologic factor for neoplastic growth has been pre. viously identified. Diazepam has been found to be a tumor enhancer in laborabQ animals (38).Friedman and Ury (39) re- ported the association of diazepam with lymphomas and leukemias. Both the mahr and minor tranquilizers may induce blood dyscrasia (40); thus, their role in the &vel. opment of leukemia is biologically plaui. ble. Further study of the significance of specific drugs would require each to & named by the respondents along with the condition(s) for which they were used and would require exploring the possibility of substance abuse (frequent medication in the absence of a pathologic condition). Some of the same disease and drug risk factors may also be indicative of a long subclinical state of hairy cell leukemia or early undiagnosed disease. For example,aspirin usage may be related to the onset of require a review of respondents' medical symptoms such as splenomegaly, an early records to ascertain laboratory confirma- sign of hairy cell leukemia, and may con- tion of infectious mononucleosis. tribute to the development of anemia Un- Regarding drug usage, cases reported sig- fortunately, it was felt that the number of nificantly more routine use of aspirin and available cases at the time of the study was tranquilizers than did controls. With re- not sufficient to examine the significance spect to aspirin (acetylsalicylic acid), most of these subgroup associations. laboratory studies do not find this sub- An exploration into childhood lifestyle stance to be either a primary or cocarcino- appears to be warranted in view of the gen (34.35).In fact, aspirin has been noted significant association of hairy cell leukt. to delay tumor growth and metastasis (35). mia with farm birthplace detected in thi? However, aspirin is also noted to inhibit study. Factors related to this agricultum! the immune system by decreasing normal environment need to be identified and isokiller cell activity (36). Aspirin is also lated for future study. known to interfere with prostaglandin syn- REFERENCES thesis, an essential component for the mat- 1. Szklo M. Are funher epidemiologic studiesOf uration and differentiation of macrophage- kemia needed? Am J Epidemiol 1980,112:-1 i ! t i ! HAIRY CELL LEUKEMIA I 2 Donham KJ, Berg JW, Sawin RS.Epidemiologic 21. Linoa A, Gray J, Orvis A, et al. Lowdose radiation b relationships of the bovlne population and human and leukemia N Engl J Med 1980;302:1101-5. leukemia in Iowa. Am J Eoidemiol lW1121K)- 22. Gibson R Graham S,Lilienfeld A. et al. Irradia- 92. tion in the epidemiology of leukemia among adults. 3. Bouronck 9. Wiseman B,Doan C. Leukemic re- JNCI 1972;48:301-11. ticuloendotheliosis. Blood 195&13SO9-30. 23. Milham S Jr. Leukemia and multiple myeloma in 4. Bouronck 9. Leukemic reticuloendotheliosis. farmers. Am J Epidemiol 1971;9(:307-10. Blood 197953412-36. 24. Burmeister LF, Van Lier SF,Isacson P. Leukemia 5. Golomb HM, Catovsky D, Golde DW. Hairy cell and farm practices in Iowa Am J Epidemiol leukemia: a clinical review basedon 71 cases. Ann 1982;115:720-8. Intern Med1978;8%77-83. 25. Grufferman S.Duong T. Cole P. Occupation and 6. Jansen J. Hermans J. Splenectomy in hairy cell Hodgkin's disease. JNCI 197657:1193-5. leukemia. Cancer 1981;472066-76. 26. W& J. N a d carcinoma in woodworkers: a re- 7. Katayama I. Finkcl HE. Leukemic reticuloen- view. J Occup Med 1982;24:5&30. Qtheliosk a clinicopathologic study with rwiew 27. Swanson GM. Belle SH. Cancer morbidity among of the literature. Am J Med 1971;57115-26. woodworkers in the US automotive industry. J 8. Center for Divase Control. Leukemia among pr- sons p m n t at an atmospheric nuclear test (Smoky). MMWR 1979;2&361-2. 9. Stewart D, Keating M. Rediation exposure as a possible etiologic factor in hairy cell leukemia (leukemic reticuloendotheliosis). Cancer 1980; 16:1577-80. 10. &hen HJ. Skinner D, Pans SA, et d Hairy cell 30. W leukemia-associated familid lymphopmliierative disorder. Ann Intern Med 1979;903174-9. 11. Ramseur WL, Golomb HM, Vardiman JM, et al. Hairy cell leukemia in father and son. Cancer 1981;48.1825-9. 12. Golomb HM, Vardiman J. Hairy cell leukemia: diagnosis and management. CA 1978;28:265-77. 13. Vykoupil KF, Thiele J. Ceorgii A. Hairy cell leu- kemia: bone marrow findings in 24 patients. Vir- chows Arch A Pathol Anat Histol 1976370273- 89. 14. N ~ b Ka , soban E. Bowling MC, et aL Splenic Pseudosinuses and hepatic angiomatous lesions. Am J Cli Pathol 1975;67415-26. 15. US Dept of Labor. Dictionary of occupational titles. Washington. D C US GPO, 1979. 16. Bredow NE, Day .NE. Hdvorsen KT, et al. Estimation of multiple relative risk hurctionr in matched case-control studies. Am d Epidemiol 1978;108.299-307. 17. Paffenbarger R Jr, Wing AL, Hyde R Chiuacterdits in youth predictive of adult-onset malignant haphornas. melanomas, and leukemias. JNCI 19%$%392-. Williams RR, Horn JW. Association of cancer pational cancer. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory,1981:247-65. 31. Chang RS.Infectious mononuclmsis. Boston: GK Hall dr Co, 1980. 32. Newell CR, Rawlings R, Kinnear 9, et al Case- control study of Hodgkin's disease. I. Results of the interview questionnaire. JNCI 197351:1437- 41. 33. b n e r F,Grunwald H. Infeetiow mononucleosis and acute leukemia. JAMA 1981;2461783-4. 34. Ibkklin AW, S m t RJ. Eighteen month study of aspirin, phenacetin and caffeine in C578B1/6 mice. Drug Chem Toxicol lW,3:135-63. 35. Perkins 'I'M, Shklar C. Delay in hamster pouch carcinogenesis by aspirin and indomethacin. Oral Surg 1982;53:170-8. 36. Grohmann PH. Ponsolt F. Quirt I, et al. Stimu- lation of human NK cell activity of cultured cell. 11. Ingestion of aspirin by blood donors suppresses NK activity. Clin Exp Immunol 1981;44:611-14. 37. Das U. Prostaglandins and drug induced agranu- locytosis. aplastic anemia and leukemia. Prosta- s i t e with tobacco and alcohol consumption and socioeconomic status of patients interview study from the Third National Cancer Survey. JNCI 1977335'25-47. Abbatt JD. Lea AJ. The incidence of leukemia in ankylosing spondylitis treated with x-rays. Lancet .' 1956;21317-20. Bizzozero OJ, Johnson KG.Ciocco A. Radiation- related leukemia in Hiroshima and Nagasaki: 1946-1964.1. Distribution. incidence. and amearance time. N Engl J Med 1966;274:1b95-1lbi. glandins Med 1979;2:235-83. 38. Karmali RA. Volkman A, Muse P, et al. The influence of diazepam administration in rats bear- ing the R39230AC mammary carcinoma. Prosta- glandins Med 1979;3:193-8. 39. Friedman GD. Ury HK. Initial screening for car- cinogenicity of commonly used drugs. JNCI 1980,65:723-33. 40. Swanson M, Cook R Drugs. chemicals and blood -dvscrasias. Hamilton. IL: Drue lntcllieence PubI lications. 1977. I 1