Authors
Alavanja MC, Samanic C, Dosemeci M, Lubin J, Tarone R, Lynch CF, Knott C, Thomas K, Hoppin JA, Barker J, Coble J, Sandler DP, Blair A. .

Title
Use of agricultural pesticides and prostate cancer risk in the agricultural health study cohort.

Source
American Journal Epidemiology. 157(9):800-14. 2003

Summary
Many factors have been linked to increased risk of prostate cancer including age, family history, African-American ethnicity, hormonal factors and diet. The most consistent occupational risk remains farming and agricultural practices due to exposures to insecticides, fertilizers, herbicides and other chemicals. The role of specific agricultural chemicals in prostate cancer has not been explicitly established because of lack of precise exposure data. Here the authors examine the exposure-response relationship between 50 agricultural pesticides and prostate cancer incidence in the Agricultural Health Study cohort.

The Agricultural Health Study cohort is a prospective cohort study of 89,658 people and includes 52,395 private and 4916 commercial pesticide applicators. Study participants were given questionnaires to determine use of 50 pesticides, crops grown, livestock raised, protective equipment used, pesticide application methods used, other agricultural activities, nonfarm occupational exposures, smoking, alcohol consumption, fruit and vegetable intake, multiple vitamin use, medical conditions and family history and basic demographic data (questionnaire: www.aghealth.org). For 22 of the 50 pesticides used, information was obtained regarding duration of use and frequency of use. An exposure intensity index "I" was computed based on application methods and protective equipment. Cohort members were matched to cancer registry files in Iowa and North Carolina for case identification and to the state death registries and National Death index to ascertain vital statistics. Prostate cancer cases diagnosed prior to enrollment were excluded. Incident cases were identified from enrollment (1993-1997) through December 31, 1999.

The analysis was restricted to 55,332 male private and commercial applicators with no history of prostate cancer at enrollment. During the follow-up period (4.3 years) 566 prostate cancer cases were observed, greater than the total number of prostate cancer cases expected (494.5) based on state age-adjusted incidence rates. Prostate cancer incidence was slightly higher among commercial applicators (standard incidence ratio (SIR)=1.41; 95% CI, 0.89, 2.11) than among private applicators (SIR=1.13; 95% CI, 1.04, 1.24). Geography appeared to play a role as prostate cancer incidence was higher among Caucasian Iowa men (SIR=1.27, 95% CI: 1.13, 1.27) compared to Caucasian North Carolinians (SIR=1.10, 95% CI: 0.99, 1.21). The number of non-white participants was too low to provide meaningful standardized incidence ratios.

Prostate cancer incidence increased with age and was more common among men with a family history of prostate cancer. Pesticide use (50 individual pesticides examined) tended to aggregate into one of three groups which contributed to the variance in pesticide use observed. Factor 1 chemicals were pesticides used primarily on corn, soybeans and other grain crops important in Iowa (herbicides atrazine, dicamba, cyanazine, metolachlor, S-ethyl dipropylthiocarbamate (EPTC), alachlor, imazethapyr, 2,4-dichlorophenoxyacetic acid (2,4-D), trifluralin, chlorimuron ethyl, metribuzin, petroleum oil, pendimethalin, and butylate and with the insecticide terbufos). Factor 2 chemicals included pesticides used commonly on cotton, tobacco, vegetables and fruit crops raised in North Carolina (herbicide (paraquat), insecticides (parathion, carbaryl, aldicarb), fumigant (methyl bromide), and fungicides (benomyl, chlorothalonil, maneb/mancozeb, and metylaxyl). There was no significant association between Factor 1 and 2 chemicals and prostate cancer incidence. Only Factor 3 chemicals were significantly associated with increased incidence of prostate cancer. These chemicals included chlorinated insecticides (aldrin, chlordane, dieldrin, dichlorodiphenyltrichloroethane (DDT), heptachlor, and toxaphene; and chlorinated phenoxy herbicides, 2,4,5-trichlorophenoxyacetic acid (2,4,5-T) and 2,4,5-trichlorophenoxypropionic acid (2,4,5-TP) which were primarily used by men over the age of 50 and are no longer registered for use in the US.

Analysis of individual pesticides revealed a linear trend between prostate cancer incidence and methyl bromide exposure, a fumigant used by about 12% of the cohort. Both well-differentiated and poorly differentiated tumors demonstrated a significant linear trend with methyl bromide exposure. Methyl bromide exposure was significantly associated with increased risk of prostate cancer in both Iowa and North Carolina and among private and commercial pesticide applicators. Methyl bromide is an alkylating agent and is considered to be a potential occupational carcinogen.

Three of the chlorinated insecticides (aldrin, DDT, and heptachlor) were associated with a significant excess risk of prostate cancer in ever use/never used analysis, however, no exposure-response pattern was observed. Lack of exposure-response pattern suggests that increased risk may be due to other exposures not identified in this analysis.

A family history of prostate cancer among first-degree relatives increased risk of prostate cancer two-fold. Pesticide exposures linked to prostate cancer incidence tended to occur in men with a family history of prostate cancer. This suggests that some individuals share familial genes that enhance susceptibility to environmental exposure or that families share environmental risk factors for prostate cancer.

The strengths of this study are numerous. The cohort represents a substantial number of study participants and a well-studied group of individuals which enabled analyses of significant statistical power to be conducted. Data was obtained prior to prostate cancer diagnosis which precludes bias and cancer outcomes were confirmed using cancer registries, thereby eliminating survival problems. Significant data were obtained from detailed questionnaires regarding lifestyle, sociodemographic data, occupational history and family medical history in addition to pesticide use, which enabled the analysis to control for confounding factors. However, data obtained from questionnaires was subject to recall bias, in some instances patterns of pesticide use were reconstructed from many years prior to the study. Estimates of exposure were used based on methods for pesticide application, days of use per year, years of use and protective equipment. While this multivariate approach strengthened the estimation of exposure, no direct measurements of pesticide exposure were obtained for the study. Finally, the follow-up period for the study was relatively short (4.3 years) which did not enable the evaluation of time-dependent exposures and risk.

The results obtained here are consistent with the literature linking farming and agricultural occupations to increased risk of prostate cancer. The study has identified methyl bromide exposure to increased risk of prostate cancer and suggests that chlorinated pesticides may also be related to prostate cancer incidence. Additional follow-up studies for this cohort are anticipated and should further strengthen the association between farming and prostate cancer risk. Further studies are required to establish the biological mechanisms by which pesticides cause prostate cancer. Some pesticides, such as methyl bromide, have been suggested to act as carcinogens, while others have been implicated as endocrine disrupters. The vast array of pesticides, comprising numerous chemical classes with different biochemical properties may represents several different mechanisms of action with respect to the etiology of prostate cancer. Identification of pesticides that have shown to be linked to prostate cancer, such as methyl bromide, should be followed by studies to elucidate the mechanisms of action.