We observed 2,960 cancer cases, 1,196 of which were prostate, among the 44,133 men who reported no use of coumaphos, and we observed 258 cancers cases, 115 of which were prostate, among 3,689 men who reported using coumaphos. Among the 1,311 prostate cancers, 504 occurred early in follow-up (1993–1999), and 807 occurred later in follow-up (2000–2005). More than one-half of the prostate cancers identified were local stage (55%), 13% were regional or distant, and 32% were not classified as to stage at diagnosis, possibly because the “watchful waiting” treatment strategy was recommended for some cases. Most prostate tumors diagnosed in the cohort were moderately well or well differentiated (70%), 26% were poorly differentiated, 0.3% of tumors were undifferentiated, and the remaining 3.3% of tumors were not determined as to histologic grade. Thirty-eight percent of the cohort reported a family history of any cancer.
At enrollment, 7.7% of the study population reported using coumaphos, and the range of annual exposure days reported by coumaphos users was between 2.5 and 200 days per year. displays the frequency of selected demographic characteristics of the study population by categories of lifetime exposure days of coumaphos use. The participants were predominantly non-Hispanic white (97%) male private pesticide applicators. Compared with the low- and high-exposure groups, the unexposed group was generally younger and included a higher proportion of commercial applicator and nonwhites. Men in the unexposed group were less likely to have attained education beyond high school or to report a first-degree family member with a cancer diagnosis (). The mean (± SD) number of coumaphos exposure days over the lifetime was 7.5 ± 2.5 among men in the lowest exposure group and 167.1 ± 402.1 among men in the highest exposure group. Those with a family history of prostate cancer had twice the risk of prostate cancer (OR = 2.02; 95% CI, 1.73–2.37) (data not shown).
Characteristics of male AHS farmers and commercial applicators by coumaphos cumulative exposure category, 1993–2005 [no. (%)].
RRs and 95% CIs for selected cancers by categories of coumaphos lifetime exposure days are shown in . Risk estimates for all cancers combined were not significantly different from the null, regardless of whether the reference group was men never exposed to coumaphos or men with low exposure. No cancer site displayed evidence of an exposure–response gradient. Coumaphos lifetime exposure days were not related to the risk of other major cancer sites examined including prostate overall or when stratified by stage and grade of disease. Results were similar for analyses based on intensity-weighted lifetime exposure and are not shown here.
RRs for total and major cancers by coumaphos lifetime exposure days among males, AHS 1993–2005.
shows the association between coumaphos exposure and prostate cancer risk among men with and without a family history of prostate cancer. Results are presented for later in follow-up (2000–2005) and across the entire follow-up period by ever/never use of coumaphos. For comparison, results are also presented for early in follow-up (1993–1999), which was reported previously (Alavanja et al. 2003
). At baseline, approximately 8% of men not exposed to coumaphos reported a family history of prostate cancer, whereas 11% of coumaphos-exposed men reported a family history of prostate cancer. Among men without a family history of prostate cancer, there was no association between coumaphos and prostate cancer early, later, or across the entire follow-up period. However, as observed early in follow-up, among men with a family history of prostate cancer, men who ever used coumaphos had a nonsignificantly higher risk of prostate cancer in follow-up years 2000–2005 than men who did not use the chemical (RR = 1.46; 95% CI, 0.89–2.40). Across the entire 12 years of follow-up, the RR of prostate cancer comparing ever versus never users among those with a family history of prostate cancer was statistically significant (RR = 1.65; 95% CI, 1.13–2.38), as was the test for interaction between coumaphos exposure and family history (p
-interaction = 0.004).
Association between ever use of coumaphos and prostate cancer RRs (95% CI) by family history of prostate cancer, early (1993–1999) and later (2000–2005) in cohort follow-up, AHS.
In a joint-effects analysis of the effect of both a positive family history of prostate cancer and self-reported ever use of coumaphos in association with prostate cancer (age-adjusted only), we observed the joint effect of both exposures in excess of the expected joint effect of the two independent exposures, measured on the multiplicative scale. Compared with the referent group of no reported family history of prostate cancer and no self-reported use of coumaphos, those who report use of the chemical but no family history have no increased risk of prostate cancer (RR = 0.86; 95% CI, 0.69–1.10); those who report a family history but no use of coumaphos have an almost 2-fold increased risk of prostate cancer (RR = 1.75; 95% CI, 1.49–2.05), and those who report both a positive family history of prostate cancer and self-report use of coumaphos have a nearly 3-fold increased risk of prostate cancer (RR = 2.89; 95% CI, 2.04–4.09) compared with the expected joint effect of 1.5. Therefore, those who report both exposures have a nearly 2-fold increased risk beyond the expected joint effect of the two exposure factors if they were independent. Results of the joint effect model are similar in both periods of prostate cancer diagnosis—that is, there is statistically significant effect modification of prostate cancer risk in both the later and earlier follow-up periods.
We observed similar patterns of effect modification by family history of prostate cancer in the relation between lifetime exposure days of coumaphos use and prostate cancer across the entire follow-up period (p-interaction = 0.02) (data not shown). Adjusting for age and using measures of cumulative exposure collected at enrollment, we observed increased risk when comparing the highest exposure group (i.e., upper half of exposed group) (n = 16 prostate cancer cases) with the nonexposed group (n = 184 prostate cancer cases) among men with a family history of disease (RR = 1.36; 95% CI, 0.81–2.26, p for trend = 0.04) and observed no increased risk between the top tertile (n = 37 prostate cancer cases) and the nonexposed group (n = 935 prostate cancer cases) among men without a family history of disease (RR = 0.78; 95% CI, 0.56–1.08, p for trend = 0.15). However, the risk of prostate cancer by lifetime exposure day category for men with a family history of disease was not monotonic. There were insufficient data available to investigate the exposure–response relation between coumaphos exposure and prostate cancer risk by presence of family history (yes/no) and by time period of follow-up (early and later periods).