Document N21wrG7ekqpnqRmn1Nop0VQLg

REVIEW Herbicides and Cancer Howard I. Morrison, * Kathryn Wilkins, Robert Semenciw, Yang Mao, Don Wigle and the chemically related chlorophenols. Some commercial Herbicides are a heterogeneous class of chemicals used in agriculture, forestry, and urban settings to kill weeds, shrubs, and broad-leaved trees. The role of herbicides in the etiology of cancer is controversial. Potential studies for review were identified through a MEDLINE' search and from a check of references in related review articles. This review of the literature shows reasonable evidence suggesting that occupational exposure to phenoxy her- preparations of both phenoxy herbicides and chlorophenols have been contaminated with dioxins and furans (4). Although non-Hodgkin's lymphoma and soft-tissue a. comas have been studied the most, possible links with herbicide exposure have been investigated for many other types of cancer, including Hodgkin's disease, multiple myeloma, leukemia, and cancers of the brain, lung, ovary, testis, and prostate. bicides results in increased risk of developing non- Hodgkin's lymphoma. Several studies have noted large Methods increases in risk of soft-tissue sarcomas with phenoxy herbicide exposure. In contrast, others have failed to Study Criteria observe increased risks, and evidence of an exposure-risk relationship is lacking. Although there have been too few appropriate studies for adequate assessment of risk of cancer at other sites, some findings have linked herbicide exposure with cancers of the colon, lung, nose, prostate, and ovary as well as to leukemia and multiple myeloma. Future studies must better identify and quantify the nature of herbicide exposures. In the interim, it seems only prudent to monitor and promote safety practices among persons occupationally exposed to phenoxy herbicides, particularly farmers and professional sprayers. [J Natl Cancer Inst 84:18661874, 19921 For this review, studies were identified through a MEDLINE' search and from a check of references in related review articles. Our inclusion criteria required that a study report individual herbicide exposures in relation to risk of any cancer. Case reports and studies in which individual herbicide exposures were not estimated were excluded. For a number of reasons, studies of civilian and military populations who may have been exposed to herbicides during the Vietnam war were not examined. Most of these studies have not been based on individual exposure data because accurate estimation of doses to which military personnel were exposed has proven problematic (5). In addition, it is only now that the likely minimum latent period for cancer has elapsed. Also, the nature of herbicide The role of herbicides in the etiology of cancer has been exposure Vietnam veterans received was different from that the s u b p t of controversy for the last decade. The welldocumentetiirrcreased risk of certain types of cancer among of herbicide applicators and employees manufacturing farmers (I), coupled with the widespread use of herbicides, has raised concerns about the long-term safety of herbicide use. This interest has resulted in a number of recent review articles dealing with herbicides and long-term health effects (23). Herbicides comprise heterogeneous classes of chemicals used in agriculture, forestry, and urban settings to control Received May 14, 1992; revised October 2, 1992; accepted Octokr X. 1992. Bureau of Chronic Disease Epidemiology. Health and Welfare Canada. Tunney's Pasture. Ottawa, Ontario, Canada. We acknowledge the helpful assistance of Paul Villencuve. weeds, shrubs, and broad-leaved trees. In addition to *Correspondence ro: Howard 1. Momson, Ph.D., Bureau of Chronic phenoxy herbicides, they include triazines, amides, benzoics, Disease Epidemiology, Health and Welfare Canada, Tunney's Pasture. carbamates, trifluralin, and uracils. Most research of the Ottawa, ON K1A OL2, Canada. - effects of herbicides on human health has focused on 'Ed. note: MEDLINE is the National Library of Medicine's bibliog@k phenoxy herbicides, primarily 2,4-dichlorophenoxyacetic database covering the fields of medicine, nursing, dentistry, VeterinaV acid (2,4-D) and 2,4,5-trichlorophenoxyaceticacid (2,4,5-T), medicine, and the preclinical sciences. 1866 Journal of the National Cancer Institute herbicides (3). Vietnamese studies on this subject were excluded because they have not appeared in peer-reviewed Western journals and hence have been somewhat inaccessible (6). Fpidemiologic evidence of health effects of herbicide -sure is available from case-control studies or cohort studies of workers employed in either the manufacture or spraying of herbicides. The designs of the studies reviewed are summarized in Tables 1 and 2. Tables 3 and 4 show the estimates of risk of cancer, by cancer site; these estimates were obtained from the case-control and cohort studies, respectively. Table 5 summarizes the findings from these studies. Epidemiologic Evidence Cancer Some reports have linked herbicide exposure with colon cancer. A case-control study by Hardell (10) noted an odds Table 1. Design of case-control studies of herbicide exposure and selected cancers. Study Hardell and Sandstrom E r i n et al. Hardell et al. Hardell Hardell et al. Donna et al. Smith et al. Hoar et al. Smith and Pearce Hoar et al. Pevcc et al. Pevce et al. P e a et al. Woods et al. Cbeckoway et al. Vineis et al. Hardell and Eriksson MuJicco et al. Dolloa et al. -son et al. >ji-Garin et al. McDuffie et al. zlhm et al. Hudell et al. B m et al. Exikon et al. Cantor et al. Ref. No. Cancer Location of study Years No. of cases (7) STS Sweden 1970-1977 STS Sweden 1974-1978 NHL. HD Sweden 1974-1978 Colon Sweden 1978-1979 Liver Sweden 1974-1981 Ovary Italy 1974-1980 STS New Zealand 1976-1980 Colon Kansas 1976-1982 STS New Zealand 1976-1980 STS, HD, NHL Kansas 1976-1982 NHL New Zealand 1977-1981 MM New Zealand 1977-1981 NHL STS, NHL New Zealand Washington 1977-1981 1981- 1984 Prostate North Carolina 1984-1985 STS ltaly 1981-1983 STS Sweden 1978-1983 Brain ltaly 1983-1984 Ovary NHL, HD Italy Sweden 1980-1985 1964-1986 Leukemia France 1984-1988 Lung Saskatchewan 1983-1986 NHL Nebraska 1983-1986 Nasal Sweden 1970-1979 Leukemia Iowa, Minnesota 1981-1984 STS Sweden 1978-1986 i33j NHL Iowa. Minnesota 1980-1983 52 110 169 154 98 60 82 57 51 424 83 76 183 704 40 68 54 240 65 160 185 273 201 71 578 237 622 No. of controls 205 220 338 541 200 127 92 948 315 948 396 315 338 694 64 168 321 742 137 275 513 187 725 541 1245 237 1245 Source of controls . Herbicide exposure NPR, DC NPR, DC NPR, DC NPR. DC NPR CR CR RDD. Medicare, DC CR RDD, Medicare, DC CR, electoral rolls CR CR RDD, Medicare, DC Hospital Electoral rolls CR, NPR Hospital Electoral rolls NPR Hospital Community health plan RDD, Medicare, DC NPR, DC RDD. Medicare, DC NPR RDD, Medicare, DC Phenoxys,-c*omphcnols Phenoxys, chlorophenols 2,4,5-T, 2.4-D, MCPA Phenoxys, chlor Phenoxys ?hcnols Herbicides ~ Primarily 2.4.5-T Phenoxys, uracils Primarily 2,4,5-T Primarily 2.4-D Primarily 2.4.5-T Primarily 2.4.5-T Primarily 2.4.5-T Phenoxys, chlorophenols Herbicides Phenoxys Phenoxys, chlorophenols Herbicides Triazines Phenoxys Herbicides Phenoxys 2.4-D Phenoxys, chlorophenols 2.4-D, 2.4.5-T Phenoxys, chlorophenols Phenoxys. triazines *SI5 = soft-tissue sarcoma, NHL = non-Hodgkin's lymphoma, HD = Hodgkin's disease, MM = multiple myeloma, NPR = national population registry, DC = death certificates or death registry, CR = cancer registry, RDD = random digit dialing. Table 2. Design of cohort studies of herbicide exposure and selected cancers I .study Axelson et al. Riihimaki et al. and Gaffey Lyagc Lme .coot et al. .%n et al. on et al. ~ K etI al. Wide et al. $ b i s o n et al. :%en et al. n et al. et al: Location of study Sex Occuoation Sweden Male Railroad worker Finland Male Applicator West Virginia Male Manufacturing Denmark Male Manufacturing Denmark Female Manufacturing Michigan Male Manufacturing Britain Male Manufacturing Michigan Male Manufacturing Michigan Male Manufacturing Canada Male Farmer Canada Male Farmer Canada Male Forestry worker Britain Male Manufacturing Canada Male Farmer 10 nations Male, Manufacturing, Female sDraYer (48) USA Male Manufacturing Years 1957-1978 1972-1980 1955-1977 1947-1982 1947-1982 1940-1979 1947-1983 1940-1982 1945-1982 1971-1985 1971-1987 1950-1982 1963-1987 1971- 1987 Various 1942-1987 No. of exposed workers 348 1971 884 3390 1069 2189 5784 2187 878 34 637 76 981 1222 2239 76981 18910 Exposure based on Job title Job title Job title Job title Job title Job title Job title Job title Job title Census records Census records Job title Job title Census records Various Herbicide exposure Phenoxys. amitrol 2.4-D,2,4,5-T 2.43-T Primarily MCPA Primarily M O A 2,4,5-T Primarily MCPA 2,4,5-T, chlorophenols 2.4-D Primarily phenoxys Primarily phenoxys 2,4-D, 2.45-T Phenoxys, chlorophenols Primarily phenoxys Phenoxys, chlorophenols 5172 Job title 2.4,s-T cludes results for some other listed cohorts (references 35,37,42,43), but for different follow-up periods. 84, No. 24, December 16, 1992 REVIEW 1867 Vol. 84, No. 24, December 16, 1992 - n E.e- a m Y Y REVIEW 1869 - Table 5. Summary of epidemiologic studies of herbicide exposure and-relative risk of cancer by cancer typc Cancer sitenype Stomach Colon Liver Nasal Lung Prostate Testis Ovary Brain Soft-tissue sarcoma Non-Hodgkin's lymphoma Hodgkin's disease MultiDle mveloma a- Leukemia No. of study group 12 9 9 9 14 10 9 3 9 21 19 10 9 12 Range of relative risk 0.0-3.1 0.0-2.1 0.0-1.7 0.0-4.9 0.0-2.1 0.6-1.9 0.0-4.6 1.0-4.4 0.0-1.6 0.0-9.2 0.0-6.0 0.0-3.8 0.0-5.0 0.0-4.0 No. with risk >1.0 7 7 1 3 10 9 3 3 3 9 12 3 5 8 *P<.OS. No. with statistically significant. Elevated risk Decreased risk 10 00 00 10 10 00 00 20 01 50 40 00 00 10 Ref. Nos. (34-41.44,47,48) ( I 0,14,34,35,39-41.47,48) ( I 1.34-36,40,41,44,.(7,48) (30,34-36.39,41,44,47,48) (28,3441,44,45,47,48) (21,35,37-4I ,43,47.48) (34,35,.?841,44,47,48) (12,2527) ( 2 4 , 3 5 3 9 4 1 ,4 4 , 4 6 4 8 ) (7,8,13,15,16,20,22,23,32,3~.35,37,39I ,-444,45,47,48) (9.I6,17,19,20,26,29,33-35,38-42,44,45.47,48) ( I 6,26,35,39-414, 4,45,47,48) (18.34,35.39-4l,JJ,47.48~ (27,31,34,35,37,39JI4,4j47.48) On the other hand, several cohort studies of herbicide workers (37,48) have noted an increased risk of lung cancer, but these risk estimates were not adjusted for smoking. In a case-control study in Saskatchewan in which smoking was controlled for (28), there was a decreased risk of lung cancer among those reporting herbicide use. In a cohort study of Saskatchewan farmers (42), no increase in lung cancer risk with increasing number of acres sprayed with herbicides was noted. Son-Tissue Sarcoma The role of herbicides in the etiology of soft-tissue sarcomas is unclear. A number of Swedish case-control studies (7,8,23)have found a statistically significant increase in risk of soft-tissue sarcoma after exposure to phenoxy herbicides, as has a Danish study (37). An increased risk of soft-tissue sarcoma was observed among female, but not male, rice weeders exposed to phenoxy herbicides (22). A review by Honchar and Halperin (50) of four U.S. cohort studies of workers employed in the manufacture of phenoxy herbicides noted that three (2.9%) of the 105 deaths observed were from soft-tissue sarcomas; only 0.07% of deaths among U.S.males aged 20-84 years were from soft- tissF-_sarcomas. Fingerhut et al. (48) observed a large increase- risk of soft-tissue sarcomas among workers exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin(TCDD) (SMR = 9.22;95% CI = 1.9-26.95). Case-control studies in North America (16,20) and New Zealand (13,15) and recently in Sweden (32) have shown either very small increased risks or no increased risk of softtissue sarcoma among persons exposed to herbicides. Prostate Cancer Farmers have been observed to be at increased risk of developing prostate cancer ( I ) . Limited evidence suggests that herbicide exposure may increase risk of prostate cancer. A cohort study of western Canadian farmers (43) noted a significant dose-response relationship between risk of dying of prostate cancer and the number of acres sprayed with herbicides. Nine of the 10 studies reviewed (Table 5 ) noted an increased risk of prostate cancer with herbicide exposure; however, only the Canadian study observed a statistically significant trend in risk. Cancer Both Bond et al. (41) and Coggon et al. (39) noted an increased risk of testicular cancer among employees manufacturing phenoxy herbicides. However, Bond et al. observed only one case, and Coggon et al. observed only four. Ovarian Cancer Two Italian case-control studies have linked an increased risk of ovarian cancer with exposure to herbicides. One study by Donna et al. (12) revealed a significantly increased risk of mesothelial ovarian tumors (OR = 4.38;95% CI = 1.90-16.07)corresponding to reported herbicide exposure. In a similar study by the same investigators (25), a significantly increased risk of epithelial ovarian tumors was associated with triazine herbicides (OR = 2.7; 95% CI = 1.0-6.9). However, a study in Denmark by Lynge (37) showed no significant excess risk associated with exposure to phenoxy herbicides (SMR = 1.02). Brain Cancer The available evidence does not substantiate any association between herbicide exposure and brain cancer. Only two of eight occupational cohorts reported an excess of brain cancer, and the excess was not statistically significant in either of the two cohorts. An Italian case-control study (24) reported an increased risk of brain gliomas associated with use of agricultural chemicals. An OR of 1.6 (95% CI = 0.6-4.4), which was not statistically significant, was 1870 Journal of the National Cancer .Institute -c Table 3. Odds ratios from case-control studies of bcrbicide exposure by cancer type Study Ref. No. Colon Liver Nasal Lung Softtissue sarcoma Odds ratio for cancer type Prostate Ovary Brain NanHodgkin's Hodgkin's Multiple lymphoma disease myeloma - Leukemia I Hardell and Sandstrom Eriksson et ai. Hardell et al. Hardell Hardell et al. Donna et ai. Smith et al. Hoar et al. Smith and Pearce Hoar et al. Pearce et al. Pearce et ai. Pearce et al. Woods et al. Checkoway et al. Vineis et al. Vincis et al. Hardell and Eriksson Musicco et al. . Donna et al. Penson et al. Bastuji-Garin et al. McDuffie et al. Hoar Zahm et al. Hardell et al. Brown et al. Eriksson et al. Cantor et al. 1.3b 1.P 1.9 0.P 0.8c.f 6.W.c.' 1.W.' 1.3b 1.3b 1.Ob 0.8b l.lb 1.7d 0.9b.g 2.7a.bb 3.31.b 1.6b 2.71.' 49.1 3.Sb 3.6a.b 0.6b 3.le.f 2.lb 1.2' 1.3b 1.2b =pc.05. bPhenoxy herbicides. =Hodgkin's disease + non-Hodgkin's lymphoma. dHerbicides, not specified. *Based on highest exposure category. '2.4-D. gMales only. bAmong living females only. 'Triazines. jPcc.10. ratio (OR) of 1.3 (95% confidence interval [CI] = 0.6-2.8) Nasal Cancer for exposure to phenoxy herbicides. Although a study by hoar et al. (14) observed an increased risk of colon cancer associated with phenoxy herbicide use (OR = 1.9), there was no.dose-response relationship, and the authors concluded that t l i e w a t a did not support an association. Seven of the nine studies of colon cancer we reviewed (Table 5 ) reported a risk greater than one. A Swedish study by Hardell et al. (30) revealed a twofold increase in risk of nasal cancer for persons exposed to phenoxy herbicides. A cohort of workers exposed to 2,4-D experienced a significant excess of deaths from nasal cancer (standardized mortality ratio [SMR] = 4.93; 95% CI = 1.0-14.4) (39). Several occupational cohort studies have noted excess risk of stomach cancer. In an investigation of Swedish railway workers, a significant excess of stomach cancer among Lung Cancer workers exposed to phenoxy acids was observed, after an Many studies have reported significantly lower lung adjustment for a 10-year induction-latency period (34). cancer mortality risks among farmers (I). However, the However, in a recent study of Saskatchewan farmers (42), no prevalence of cigarette smoking among farmers is typically relationship emerged between risk of stomach cancer lower than that among the general population (1,49); this mortality and acreage sprayed with herbicides. Only seven of situation makes it difficult to assess any association of jr the 12 study groups of patients with stomach cancer herbicides with development of lung cancer for farmers reviewed showed an excess risk, and only the Swedish study compared with occurrence of lung cancer in the general 1 ' results were statistically significant (Table 5). population. t I - 1868 - Journal of the National Cancer Institute associated with herbicide use. However, the risk among farmers who reported exposure to herbicides andlor fertilizers in the absence of insecticides was not elevated (OR = 0.9). In addition, a study of brain cancer among Canadian prair' -3nners (46) suggested an excess risk associated with insei.-de use, but not herbicide use. significantly increased risk of non-Hodgkin's lymphoma was also observed in the Kansas study (26) of farmers exposed to trifluralin (OR = 12.5; 95% CI = 1.6-116.1) or amides, such. as alachlor and propachlor (OR = 2.9; 95% CI = 1.1-7.6). Hodgkin's Disease Non-Hodgkin's Lymphoma An elevated risk of Hodgkin's disease among farmers has been reported by many studies (I), but there is only limited The most convincing evidence suggesting that herbicides evidence linking Hodgkin's disease with exposure to may be human carcinogens arises from studies of non- herbicides. In a Swedish study (26), increased risk of Hodgkin's lymphoma. An increased risk of non-Hodgkin's Hodgkin's disease was associated with phenoxy herbicides lymphoma among farmers has been reported from the United and chlorophenols. However, a case-control study conducted States, Australia, and New Zealand ( 2 ) . Although several in Kansas by Hoar et al. (16) failed to support this finding. studies have noted an increased risk associated with farm Only three of 10 studies reviewed (Table 5) reported risks ---.exposure to phenoxy herbicides (I6,29,42),studies by Pearce greater than one, and these lacked statistical s a f i c a n c e . et al. (19,51) have not. A case-control study conducted in Kansas by Hoar et al. (16 ) reported a statistically significant dose-response rela- Multiple Myeloma tionship between the annual number of days on which a Although several studies have related multiple myeloma tb farmer applied herbicides and risk of non-Hodgkin's exposure to pesticides, there is only limited evidence linking lymphoma. Farmers exposed on 20 or more days per year this cancer with herbicide exposure. Five (18,35,39,40,48)of had a sixfold increase in risk of this disease relative to the nine studies (Table 5) of herbicides and multiple nonfarmers. A case-control study of nomHodgkin's lym- myeloma were positive; however, most of the studies had phoma in eastern Nebraska (29) noted an increased risk (OR low power, and none attained statistical significance. = 1.5) for men who reported mixing or applying 2,4-D; this risk increased to 3.1 for those who reported exposure for 20 or more days per year. Leukemia A Swedish case-control study (26) showed an OR of 4.9 Evidence supporting a possible association between associated with occupational exposure to phenoxy acids for herbicides and leukemia is weak and is limited by the at i 1 year. An analysis of data from Saskatchewan availability of only one study with reasonable power. This farmers (42) showed a dose-response relationship between case-control study (27) linked exposure to herbicides with the number of acres sprayed with herbicides and risk of non- acute leukemia. Some cohorts of workers employed in the Hodgkin's lymphoma. In a casexontrol study of non- manufacture of phenoxy herbicides have been reported to be Hodgkin's lymphoma in Washington state (20), risk of at an elevated risk of leukemia. Eight of the 12 study groups developing the disease was elevated among males potentially reviewed (Table 5 ) showed an excess leukemia risk among exposed in any occupation to phenoxy herbicides for 15 or individuals exposed to herbicides; only one (27) (the only more years (OR = 1.71; 95% CI = 1.04-2.8). case-control study) attained statistical significance. How- Of the 11 occupational cohort studies reporting informa- ever, the cohort studies reviewed had very low power. tion on nomHodgkin's lymphoma, five noted an increased risk, although none of the values were statistically Methodologic Issues significant. However, most of the studies (38,40-42,45)were small, with limited statistical power. Studies of the relationship between herbicides and cancer A small case-control study in New Zealand by Pearce et among farmers often fail to give information about other al. (17) observed a weak positive but not statistically exposures that may confound the herbicide-cancer associa- significant excess in risk of non-Hodgkin's lymphoma with tion. Most farmers apply a wide variety of farm chemicals, exposure to phenoxy herbicides. After the study was such as insecticides, fumigants, and fungicides, and are enlarged, no association was observed (19). Finally, little exposed to others, such as solvents and paints. Farmers may evidence of an association of non-Hodgkin's lymphoma with also be exposed to fuels and exhausts in the process of either duration or frequency of use of phenoxy herbicides applying herbicides. In addition, most studies in this review was observed in a subsequent analysis of the data (51). did not differentiate between specific types of herbicides. Exposure to triazine herbicides has also been associated Studies of cancer usually require some proxy interviews to with an increased risk of non-Hodgkin's lymphoma. ascertain exposure information, and it is often assumed that Analysis of data from Kansas farmers exposed to triazine information supplied by proxies is less reliable than that herbicides showed an increased OR of 2.2 (95% CI = supplied by the subjects themselves. In a study by Hoar et 0 . L ') for nomHodgkin's lymphoma, independent of al. (16) of lymphoma and soft-tissue sarcoma and herbicide expeure to phenoxy herbicides (16). A lower risk was exposure in Kansas, fully one half of the case patients had associated with triazine exposure in an IowaIMinnesota died prior to interview. However, results for proxies were case-control study (33) (OR = 1.1; 95% C1 = 0.8-1.6). A similar to those for living subjects. Other evidence that the Vol. 84, No. 34, December 16, 1993, REVIEW 1871 -_ 8 F i I accuracy of data obtained from farmers themselves and their surrogate respondents on use of pesticides is comparable was reported by Brown et al. (52).If recall of exposure was less accurate among proxy respondents than among exposed subjects, the nondifferential misclassification that would result from the use of proxy interviews should bias any estimates toward the null and therefore would not account for positive results. Case-control studies may be limited, however, by the ability of both case patients and control subjects to recall specific agents used many years previously. In the Kansas study (16), information was obtained from pesticide suppliers to corroborate farmers' reported use. ORs using these data were similar to those derived from farmers' self- reported exposure. A more worrisome concern with a case-control design is that the large amount of recent publicity about herbicides and cancer may increase the possibility of recall bias, i.e., a greater likelihood among case patients than control subjects for reporting of exposure. In this regard, the finding by Hoar et al. (16) of a phenoiy herbicide effect for non-Hodgkin's lymphoma, but not for soft-tissue sarcomas or Hodgkin's disease, is reassuring. Retrospective cohort studies are characterized by the necessity of using pre-existing and limited exposure information. For example, in the study of Saskatchewan farmers (42), which noted a significant dose-response relationship between herbicides and non-Hodgkin's lym- phoma, exposure was defined as the number of acres sprayed with herbicides on each farm, as reported on census forms. No information was available about herbicide application actually carried out by individual operators, much less the use of protective equipment. However, because the normal effect of nondifferential misclassification is to bias any risk estimates toward the null, the true risk of herbicide exposure in the Canadian farm operator study is probably greater than that observed. Because of the many chemicals to which farmers are occupationally exposed, there are theoretical advantages to studying workers employed in the manufacture of herbicides rather than studying farmers. However, most manufacturing workers appear to have been exposed to a variety of phenoxy herbicides or herbicide precursors over the course of their working lives. Workers in herbicide plants are exposed on more days per year to herbicides than farmers. H&he-r>oses may make it easier to detect an effect. However, it is far from clear that exposures among manufacturing workers equal or exceed those among farmers. The main limitations to studying workers employed in manufacturing herbicides are the lack of quantitative exposure information and the small size of study popula- tions, with few cases of cancer available for study. The latter fact has severely limited the usefulness of these studies. A further difficulty is that several of the occupational cohort studies were of workers exposed to 2,4,5- trichlorophenol, a precursor to 2,4,5-T, and not to 2,4-D itself (53). Various cohort and case-control studies have reported on exposures to many different phenoxy herbicides. It is difficult to know 1) if the results of such studies should be combined, to gain greater statistical power, or 2) whether the specific potential carcinogenic effects of 2,4-D, 2,4,S-T. 2-methyl-4-chiorophenoxyacetic acid (MCPA), or other phenoxy herbicides are likely to be so different as to make combining the results of such studies inappropriate. This situation may be further complicated by the potential role of dioxins as the agents responsible for the putative carcinogenic effect of phenoxy herbicides. One study (48) defined exposure on the basis of the dioxin 2,3,7,8-TCDD, a trace contaminant of certain phenoxy herbicides, instead of on the basis of direct exposure to the herbicides. A limitation of occupational cohort studies has been the use of fatal rather than incident cases of cancer. This is particularly problematic for soft-tissue sarcomas, because cancer deaths are coded to anatomic site, not histologic type. Cohort studies that rely on death certificates to determine cancer status will miss many sarcomas, such as those of the gastrointestinal tract. There are no clear reasons why the results of the New Zealand non-Hodgkin's lymphoma studies and the Scandinavian soft-tissue sarcoma studies are at variance with most of the studies from North America. No major methodologic flaws limit the New Zealand and Scandinavian studies, and the level of risk they report is such that chance is not a likely explanation. One possible explanation is the difference in relative exposures to specific phenoxy herbicides or their contaminants. The predominant phenoxy herbicide used in Scan- dinavia was MCPA, whereas 2,4-D was used primarily in North America, and 2,4,5-T was used principally in New Zealand. Scandinavian use of phenoxy herbicides is principally in forestry, while use in North America is primarily in agriculture. Application techniques and exposures in agriculture and forestry are likely to differ. Also, the lifestyle of forestry workers in Scandinavia is likely to differ from that of North American farmers (2). A possible explanation for the failure of the New Zealand studies to observe a positive relationship between nonHodgkin's lymphoma and phenoxy herbicides may relate to how herbicide exposure was measured. The Kansas study by Hoar et al. (16) observed that average annual number of days of exposure was the critical variable in risk determination. It may be that studies not using this variable cannot be expected to observe risks as large as those observed in the Kansas study. Uncontrolled confounding is not a likely explanation. Many studies were able to control for known risk factors, and no known common sources of exposure are strong risk factors for these cancers. Animal Studies Some animal studies suggest that phenoxy herbicides may be carcinogenic. A New Zealand study (54) noted a statistically significant relationship between adenocarcinoma of the small intestine in slaughtered sheep and exposure to phenoxy herbicides. Dogs exposed in residential settings to 2,4-D appear to be at an increased risk of developing malignant lymphomas (55). A statistically significant in- 1872 Journal of the National Cancer Institute _-- crease in astrocytomas was observed in rats fed bigh doses Some of the conflicting findings in the literature may of 2,4-D (41). However, several 2-year animal bioassay reflect the effect of exposure to different herbicides, only studies have failed to note any carcinogenic effects (56). some of which may increase the risk of cancer. In addition, E++nce of a genotoxic effect is equivocal (57). Both 2,4- issues of latency must be better addressed. Risks should be D h 2,4,5-T are able to inhibit intercellular communica- examined with regard to various windows of exposure. tion. c e r e is evidence that 2,4-D increases the rate of sister Almost all previous studies have examined only males chromatid exchanges (58). In addition, 2,4-D has been exposed to herbicides. Future studies should also examine reported to induce mitotic and meiotic chromosomal exposed female populations. aberrations in barley (59). Exposure to commercial 2,4-D One useful approach would be to study farmers or other resulted in a statistically significant increase in the number exposed populations prospectively. The U.S.National Cancer of chromosomal aberrations in cultured human peripheral Institute, in conjunction with the U.S. Environmental lymphocytes (60). Protection Agency, is presently assembling a large prospec- The International Agency for Research on Cancer (IARC) tive cohort of pesticide-exposed farmers (65). This cohort has concluded that there is limited evidence that phenoxy study should alleviate concerns of recall bias that may affect herbicides are carcinogenic in humans and that there is case-control studies and will provide much mor.e-&ccurate -inadequate evidence of carcinogenicity in animals (57). Wild and detailed exposure information than is available -from1 oat herbicides, such as diallate and triallate, have been retrospective cohort studies. shown to possess mutagenic potential in in vitro tests (61), In the interim, it seems only prudent to monitor and but the IARC considers that there is limited evidence that promote safety practices of persons occupationally exposed diallate is a carcinogen in animals (62). to phenoxy herbicides, particularly farmers and professional Herbicides used for corn production, including alachlor sprayers. Public education with respect to safety practices is and atrazine, have tested positive in Ames tests (63). The also required, given the general population's increasing use IARC has concluded that there is limited evidence that of lawn-care chemicals. atrazine is carcinogenic in animals and that atrazine is a possible human carcinogen (64). The known effects of atrazine on the hypothalamic-pituitary-gonadal axis are consistent with the types of tumors observed in both animals References and humans (64). Cc usions and Recommendations There is reasonable evidence to suggest that phenoxy herbicide exposure results in an increased risk of developing non-Hodgkin's lymphoma. Most studies have revealed an elevated risk, and those that have examined dose-response relationships have usually noted statistically significant trends. Evidence regarding soft-tissue sarcomas is equivocal. Although a number of studies have noted large increases in risk with phenoxy herbicide exposure, evidence of a doseresponse relationship is lacking. There have been too few studies appropriately designed to properly assess other sites, but some data suggest a link between herbicide exposure and cancers of the colon, lung, nose, prostate, and ovary as well as leukemia and multiple myeloma. Studies of the effects of herbicide exposure have typically suffered from two major limitations: inadequate sample size and poor determination of specific herbicide exposures. In an attempt to overcome the problem of lack of statistical power, the IARC has established a register of over 18000 workers exposed to either phenoxy herbicides or chlorophenols (4). The first report of the mortality in these cohorts has been recently published (47). Future studies must better identify and quantify the nature of 5icide exposures. Most case-control studies have had to . __ on proxy interviews. Those that do should analyze and present data obtained from interviews with living patients and proxies for deceased patients separately to determine if results from the two groups differ. ( 1 ) BLAIR A. HOAR ZAHMS: Cancer among farmers. Occup Med 6:335354, 1991 (2) LILIENFELDDE, GALLOMA: 2.4-D, 2,4.5-T. and 2,3,7,8-TCDD: An overview. Epidemiol Rev 11:28-58, 1989 (3) BLAIR A. HOAR ZAHMS: Herbicides and cancer: A review and discussion of methodologic issues. Recent Results Cancer Res 120:132-145, 1990 ( 4 ) 1NERNATIONAL AGENCY FOR RESEARCH O N CANCER WORKING GROUP: Phenoxy acid herbicides and contaminants: Description of the lARC international register of workers. Am J Ind Med 18:39-45, 1990 (5) KELLY SJ, G v r w m TL: Phenoxyacetic acid herbicides and chlorophenols and the etiology of lymphoma and soft-tissue neoplasms. Public Health Rev 17:l-37, 1989-1990 (6) STERLING TD. ARUNDEL A. Review of recent Vietnamese studies on the carcinogenic and teratogenic effects of phenoxy herbicide exposure. lot J Health Sew 16:265-278, 1986 HARDELL L, SANDSTROMA: Case-control study: Soh-tissue sarcomas and exposure to phenoxyacetic acids or chlorophenols. Br J Cancer 39~711-717, 1979 ERIKSSONM, HARDELLL,BERG NO, ET AL: Soft-tissue sarcomas and exposure to chemical substances: A case-referent study. Br J Ind Med 38127-33. 1981 HARDELLL. ERIKSSONM. LENNER P, ET AL: Malignant lymphoma and exposure to chemicals. especially organic solvents, chlorophenols and phenoxy acids: A case-control study. Br J Cancer 43:169-176, 1981 HARDELLL: Relation of soti-tissue sarcoma. malignant lymphoma and colon cancer to phenoxy acids. chlorophenols and other agents. Scand J Work Environ Health 7:119-130, 1981 HARDELLL, BENGTSSONNO, JONSSON U, ET AL: Aetiological aspects on primary liver cancer with special regard to alcohol, organic solvents and acute intermittent porphyria-an epidemiological investigation. Br J Cancer 50:389-397, 1984 DONNA A, BEITA PG, ROBWIT F, ET AL: Ovarian mesothelial tumors and herbicides: A case-control study. Carcinogenesis 5:941-942, 1984 SMITH AH. PEARCE NE, FISHER DO, ET AL: Soft tissue sarcoma and exposure to phenoxyherbicides and chlorophenols in New Zealand. JNCl 73:1111-11 17. 1984 HOAR SK, BLAIRA. HOLMES FF,ET AL: Letter: Herbicides and colon cancer. Lancet 1:1277-1278, 1985 SMITH AH. PEARCE NE: Update of soft tissue sarcoma and phenoxyherbicides in New Zealand. Chemosphere 15:9-12, 1986 Vol. 84. No. 24, December 16, 1992 REVIEW 1873 I (16) HOAR SK, BLAIR A, HOLMES FF. ET AL: Agricultural herbicide use Environ Health 12:448-454, 1986 and risk of lymphoma and soft-tissue sarcoma [published erratum (40) Orr MG. OLSON RA. COOK RR, ET AL: Cohort mortality study of appears in JAMA 256:3351]. JAMA 256:1141-1147, 1986 chemical workers with potential exposure to the higher chlorinated (17) PEARCE NE. SMITH AH, HOWARD JK. ET AL: Non-Hodgkin's dioxins. J Occup Med 29:422429. 1987 lymphoma and exposure to phenoxyherbicides. chlorophenols, fencing (41) BOND GG. WETERSTROEM NH, ROUSH GJ. ET AL: Cause specific work, and meat works employment: A case-control study. Br J Ind mortality among employees engaged in the manufacture, formulation. Med 43:75-83. 1986 or packaging of 2.4-dichlorophenoxyacetic acid and related salts. Br J (18) PEARCE NE, SMITH AH, FISHER DO: Malignant lymphoma and Ind Med 45:98-105. 1988 multiple myeloma linked with agricultural occupations in a New (42) WIGLE DT, SEMENCIWRM, WILKINSK, ET AL: Mortality study ot Zealand Cancer Registry-based study. Am J Epidemiol 121:225-237. Canadian male farm operators: Non-Hodgkin's lymphoma mortalit! 1985 and agricultural practices in Saskatchewan. J Natl Cancer Inst 82575- (19) PEARCE NE, SHEPPARD RA, SMITH AH, ET AL: Non-Hodgkin's 582. 1990 lymphoma and farming: An expanded case-control study. Int J Cancer (43) MORRISON HI, SAVR D, SEMENCIW R, ET AL: Farming and prostate 39:155-161. 1987 cancer mortality. Am J Epidemiol. In press (20)WOODS JS. POLISSARL, SEVERSORNK, ET AL: Soft tissue sarcoma (#4) GREEN L: A cohort mortality study of forestry workers exposed IO and non-Hodgkin's lymphoma in relation to phenoxyherbicide and phenoxy acid herbicides. Br J Ind Med 48:234-238, 1991 chlorinated phenol exposure in western Washington. JNCI 78:899- (45) C ~ G G ~DN, PANNETTB. WINTER P: Mortality and incidence of cancer 910. 1987 at four factories making phenoxy herbicides. Br I Ind Med 48:173- (21) CHECKOWAYH. DIFERDINANDGO. HULKA BS. ET AL: Medical. life- 178. 1991 style. and occupational risk factors for prostate cancer. Prostate 10:79- (46) MORRISON HI,SEMENCWRM, MORISOND. ET AL: Brain cancer and 88. 1987 farming in western Canada. Neuroepidemiol. In press (22) VINEISP. TERRACINBI. CICCONEG, ET AL: Phenoxy herbicides and (47) SARACCIR, KOGEVINASM. BERTAZZI PA. ET AL: Cancer mortality in soft-tissue sarcomas in female rice weeders. A population-based case- workers exposed to chlorophenoxy herbicides and chlorophenols referent study. Scand J Work Environ Health 139-17. 1986 Lancet 338:1027-1032. 1991 (23) HARDELLL. ERIKSSON M: The association between soft tissue (48) FINGERHUTMA, HALPERIN WE, MARLOW DA, ET AL: Cance. sarcomas and exposure to phenoxyacetic acids. A new case-referent mortality in workers exposed to 2.3,7.8-tetrachlorodibenzo-p-dioxin.;I study. Cancer 62:652-656. 1988 Engl J Med 324:212-218, 1991 (24) Muslcco M, SANTM, MOLINARI S, ET AL: A case-control study of (49) MORRISON HI, MAO Y. RITI-ERL, ET AL: A census based recorc brain gliomas and occupational exposure to chemical carcinogens: The linkage study of Canadian farmers. In Health and Safety ir. risk to farmers. Am J Epidemiol 128:778-785, 1988 Agriculture (Dosman J. Cockcroft D, eds). Boca Raton. Fla: CRC (25) DONNA A, CROSIGNANPI, R o e u r n F. ET AL: Triazine herbicides and Press. 1989, pp 280-282 ovarian epithelial neoplasms. Scand J Work Environ Health 15:47-53. (50)HONCHAR PA, HALPERINWE: Letter: 2,4.5-T. trichlorophenol, anc 1989 soft tissue sarcoma. Lancet 1:268-269, 1981 (26) PERSON B. DAHLANDER AM. FREDRIKSSONM, ET AL: Malignant (51) PFARCE N: Letter: Phenoxy herbicides and non-Hodgkin's lymphom, lymphomas and occupational exposures. Br J Ind Med 46516-520, in New Zealand: Frequency and duration of herbicide use. Br J InL 1989 Med 46:143-144, 1989 (27) BASTUJI-GARINS, RICHARDSON S, ZI-ITOUN R: Acute leukaemia in (52) BROWN LM, DOSEMECIM. BLAIR A. ET AL: Comparability of dar. workers exposed to electromagnetic fields. Eur J Cancer 26:1119- obtained from farmers and surrogate respondents on use of agricultura. 1120, 1990 pesticides. Am J Epidemiol 134:348-355, 1991 (28) MCDUFFIEHH. KLAASSEN DJ, DOSMANJA: Is pesticide use related to (53) BONDGG. BODNERKM,COOK RR: Letter: Re: "2,4,-D, 2.4.5-T, anc the risk of primary lung cancer in Saskatchewan? J Occup Med 2,3,7,8-TCDD: An overview." Am J Epidemiol 133:1293-1296, 199 32:9961002, 1990 (54) NEWELLKW, Ross AD, RENNER RM: Phenoxy and picolinic acic (29) HOAR ZAHM S, WEISENBURGEDRD, BABBITT PA, ET AL: A case- herbicides and small-intestinal adenocarcinoma in sheep. Lance. control study of non-Hodgkin's lymphoma and the herbicide 2.4- 21301-1305, 1984 dichlorophenoxyacetic acid (2,4-D) in eastern Nebraska. Epidemiology (55) HAYES HM, TARONREE, CANTOR KP, ET AL: Case-control study o! 1:349-356, 1990 canine malignant lymphoma: Positive association with dog owner., (30)HARDELLL, JOHANSSONB, AXELSON 0: Epidemiological study of use of 2,4-dichlorophenoxyacetic acid herbicides. J Natl Cancer Ins: nasal and nasopharyngeal cancer and their relation to phenoxy acid or 83:122&1231. 1991 chlorophenol exposure. Am J Ind Med 3:247-257, 1982 (56) BOND GG, BODNER KIM, COOK RR: Phenoxy herbicides and cancer (31) BROWN LM, BLAIR A, GIESON R, ET AL: Pesticide exposures and Insufficient epidemiologic evidence for a causal relationship. Func other agricultural risk factors for leukemia among men in Iowa and Appl Toxic01 12:172-188. 1989 Minnesota. Cancer Res 50:65854591, 1990 (57) INTERNATlONAL AGENCY FOR RESEARCH ON CANCER: Monographs or. (32) ERIKSSONM, HARDELLL, ADAMIHO: Exposure to dioxins as a risk the evaluation of the carcinogenic risk of chemicals to humans ' factor for soft tissue sarcoma: A population-based case-control study. J Natl Cancer lnst 82:486490, 1990 Supplement 7. Overall evaluations of carcinogenicity: An updating ot lARC monographs volumes 1 to 42. Lyon: IARC. 1987 i(33) CANTOR KP, BLAIR A, EVE RE^ G, ET AL: Pesticides and other (58) TURKUUTE. JALAL SM: Increased rates of sister chromatic! agricultural risk factors for non-Hodgkin's lymphoma among men in exchanges induced by the herbicide 2.4-D. J Hered 76:213-214, 1985 Iowa and Minnesota. Cancer Res 52:2447-2455, 1992 (59) KHALATKARAS, BHARGAVA YR: 2,4-Dichlorophenoxyacetic acid-A (34) AXELSON0, SUNDELLL, ANDERSSONK. ET AL: Herbicide exposure new environmental mutagen. Mutat Res 103:lll-114, 1982 ---- and tumor mortality. An updated epidemiologic investigation on (60)MUSTONENR. KANGAS J, VUOJOLAHTIP. ET AL: Effects o S w a railroad workers. Scand J Work Environ Health 6:73-79, phenoxyacetic acids on the induction of chromosome aberrations ir 1980 vitro and in vivo. Mutagenesis 1:241-245, 1986 (35) RIIHIMAKIV, ASP S, HERNBERG S: Mortality of 2.4-dichlo- (61) SANDHU SS. WATERS MD, MORTELMANSKE, ET AL: Evaluation o` rophenoxyacetic acid and 2,4,5-trichlorophenoxyaceticacid herbicide diallate and triallate herbicides for genotoxic effects in a battery of ir applicators in Finland: First report of an ongoing prospective cohort vitro and short-term in vivo tests. Mutat Res 136:173-183, 1981 study. Scand I Work Environ Health 8:37-42, 1982 (62) INTERNATIONAL AGENCY FOR RESEARCH O N CANCER: Monographs 01 (36) ZACKJA, GAFFEYWR: A mortality study of workers employed at the the evaluation of the carcinogenic risk of chemicals to humans. Vc Monsanto Company plant in Nitro, West Virginia. Environ Sci Res 30. Miscellaneous Pesticides. Lyon: IARC, 1983 26575-591. 1983 (63) PLEWA MJ, GENTILE JM: Mutagenicity of atrazine: A maize-microbc (37) LYNGE E: A follow-up study of cancer incidence among workers in bioassay. Mutat Res 38:287-292, 1976 manufacture of phenoxy herbicides in Denmark. Br J Cancer 523259- (64)1NTERNATIONAL AGENCY FOR RESEARCH ON CANCER: Monographs or 270. 1985 the Evaluation of the Carcinogenic Risk of Chemicals to Humans. Vo (38) COOK RR, OLSON RA, O n MG, ET AL: Evaluation of the mortality 53. Occupational Exposures in Insecticide Application, and Somc experience of workers exposed to the chlorinated dioxins. Chem- Pesticides. Lyon: IARC, 1991 i osphere 151769-1776, 1986 (65) BLAIR A, HOAR ZAHM S: Epidemiologic studies of cancer amon& (39)COGGOND, PANNE'ITB, WINTER PD, ET AL: Mortality of workers agricultural populations. I n Practical Applications for Health an? exposed to 2 methyl-4 chlorophenoxyacetic acid. Scand J Work Safety in Agriculture (Dosman J, ed). In press 1874 Journal of the National Cancer Institutc $fgienists were b w d on thcir general knowledge the codedindusmts and occupationsatherthan iater!view or monitoring dau. This method d evaluation bad a nul&: of liiibtioim (i) cxpo9ure misclrssifiation may haw occurred if oceupdm and industry +dm could not bc Uakd spccificdly lo exposures; (ii) the specificchemicals used by certain industria ma)' not have bccn evident from the industry c o d q and (ii) change, in cxpowreecovettimu couldnot bcrddrdsccf. Despire thenp e n t i d limiraticnls, the Danish Link- age cornputwid system wzis an cxxcelkntnsourdc for identifyingfmnlemyebma casesuld conuohand for invedgrting eli\ploymmr hiswry a d workplace upeurcs as potential Ask hcwra Our findings ?f an increascd myelom tisic amorzg Danish women employed inagrkdtud mdtextileindltetricsarctonristcnt with prtviously augptcd association$ md deserve funhtr investigation, References I. , '! I. j. i I .. ! i I i