Selected characteristics of the study population are displayed in according to lifetime exposure days to EPTC. The low-exposure group refers to participants in the lowest tertile (< 10 lifetime exposure days), and the high-exposure group is a combination of the top two tertiles (≥ 10 lifetime exposure days). Applicators were primarily white, with > 40% attaining at least a high school diploma. Regardless of exposure, approximately 50% of participants were never smokers, and most reported some alcohol use over the preceding 12 months. In general, the three exposure groups (no exposure, low, and high) were similar with respect to most demographic characteristics, except for the predominance of EPTC use by farmers in Iowa.
EPTC exposure was initially divided into no exposure (never users) and exposed (ever users). For all cancer sites combined, 470 cancer diagnoses were made through December 2004 among the 9,878 applicators exposed to EPTC. In contrast, 1,824 cancers were diagnosed within that same time among those with no exposure (n = 38,500). The mean (± SD) age of cancer incidence in the study population was 58 ± 10.3 years with a range of 23–87 years. A small increase in risk for all cancers was observed [rate ratio (RR) 1.14; 95% confidence interval (CI), 1.02–1.27]. No specific cancer was observed to be statistically associated with individuals ever exposed to EPTC versus those never exposed, but the risks for both colon cancer (RR = 1.35; 95% CI, 0.93–1.97) and leukemia (RR = 1.31; 95% CI, 0.75–2.28) were elevated. The risk for rectal cancer was not significant (RR = 0.80; 95% CI, 0.44–1.42).
summarizes our analysis investigating the association between cancer incidence and EPTC lifetime exposure days (left panel). We observed an increase in risk when comparing the highest level of exposure for all cancer (RR = 1.28; 95% CI, 1.09–1.50), colon cancer (RR = 2.09; 95% CI, 1.26–3.47), and leukemia (RR = 2.36; 95% CI, 1.16–4.84) with subjects with no exposure. We also observed a significant increasing linear trend for the incidence of these cancers (p-trend for all cancer < 0.01, p-trend for colon cancer < 0.01, p-trend for leukemia = 0.02). When the lowest exposure tertile was used as the referent, we observed a slight attenuation of the point estimate for all cancer (RR = 1.13; 95% CI, 0.92–1.39, p-trend = 0.05) and an increase for both colon cancer (RR = 2.76; 95% CI, 1.27–6.00, p-trend = 0.03) and leukemia (RR = 2.91; 95% CI, 0.97–8.72, p-trend = 0.04).
Rate ratios for selected cancer sites by lifetime exposure days to EPTC among male pesticide applicators in the AHS followed through December 2004.
We also analyzed EPTC exposure by tertiles of intensity-weighted lifetime exposure days. The increased risk associated with EPTC lifetime exposure days and colon cancer remained constant when we used the intensity-weighted lifetime exposure days metric (, right panel). The associated risks for all cancer and leukemia were slightly attenuated using the intensity-weighted lifetime days exposure scale. When comparing the highest tertile of EPTC exposure with those with no exposure, we found increased risks for all cancer (RR = 1.16; 95% CI, 1.01–1.35), colon cancer (RR = 2.05; 95% CI, 1.34–3.14), and leukemia (RR = 1.87; 95% CI, 0.97–3.59). When the lowest tertile of exposure was used as the referent group, the risk associated with cancer incidence and EPTC intensity-weighted exposure days remained elevated at the highest tertile of EPTC exposure for all cancer (RR = 1.19; 95% CI, 0.95–1.49), colon cancer (RR = 2.59; 95% CI, 1.13–5.97), and leukemia (RR = 3.93; 95% CI, 0.87–17.67).
We also divided the upper tertile of exposure at its median to expand examination of the association between high EPTC exposure and cancer incidence for those cancer sites that had five or more cases within each of the upper two exposure levels. An increase in risk associated with the highest level of EPTC lifetime exposure days (≥ 110) was observed when the no-exposure group was used as the referent for all cancer (RR = 1.30; 95% CI, 1.03–1.63) and colon cancer (RR = 3.55; 95% CI, 1.97–6.42). The linear trend tests for all cancer (p-trend = 0.01) and colon cancer (p-trend = < 0.01) were also statistically significant. When the low-exposure group was used as the referent, only colon cancer remained significantly elevated (RR = 4.70; 95% CI, 2.03–10.87, p-trend = < 0.01). An increase in risk associated with the highest level of EPTC intensity-weighted lifetime exposure days (≥ 333) was observed for colon cancer when the no-exposure (RR = 2.21; 95% CI, 1.27–3.86, p-trend = < 0.01) and low-exposure (RR = 2.80; 95% CI, 1.13–6.94, p-trend = 0.02) referent groups were used. The risk estimate for all cancer associated with the highest level of EPTC exposure remained elevated using either the no-exposure referent (RR = 1.14; 95% CI, 0.93–1.38, p-trend = 0.11) or the low-exposure referent (RR = 1.15; 95% CI, 0.89–1.50, p-trend = 0.54). However, the CI for the low-exposure referent includes the null, and the linear trend is no longer significant.
We performed stratified analysis of risk for all cancer by state of residence (Iowa vs. North Carolina). No difference was observed in all cancer risk when comparing the highest tertile of EPTC exposure using either lifetime exposure days or the intensity-weighted lifetime exposure days when stratifying by state of residence. When we included all participants with missing information in the referent group, there was no change in the observed risk estimates for all cancers, colon cancer, and leukemia.
Finally, we modeled cancer risk with pesticides previously reported to be associated with colon cancer risk and those found to be most correlated with EPTC use within the AHS cohort. When the five most highly correlated pesticides were added to the lifetime exposure days model as covariates (butylate, trifluralin, imazethapyr, metribuzin, and dicamba), no significant change was observed in our cancer risk estimates for all cancer, colon cancer, and leukemia. Other work from the AHS reported an increased risk of colon cancer associated with increasing exposure to aldicarb (Lee et al. 2007
) and dicamba (Samanic et al. 2006
). To determine that our observed risk estimates were not related to these pesticide exposures, we modeled colon cancer risk including both as covariates in our lifetime exposure days model. Similarly, our risk estimate was not markedly changed for colon cancer.