Selected demographic characteristics and use of correlated pesticides by atrazine use (none, lowest quartile, highest three quartiles) are reported in . Pesticide applicators in this cohort are overwhelmingly white and male. For many characteristics, such as sex, state, alcohol use, and use of correlated pesticides, applicators in the highest quartiles of use are more similar to applicators in the lowest quartile of use than to the nonexposed. Private pesticide applicators were more likely to have used atrazine than were commercial applicators (70% and 50%, respectively). However, among commercial applicators who reported use of atrazine, the number of days per year and total lifetime days of use were higher than that among private applicators (lifetime days median = 116.0, interquartile range, 49.50–457.25 for commercial; lifetime days median = 56.0, interquartile range, 20.0–178.50 for private applicators).
Selected characteristics of pesticide applicators, by atrazine use category based on enrollment questionnaire data, 1993–1997 [n (%)].
Overall, there were 4,737 incident cancers diagnosed among 53,662 applicators through 2007, including 1,785 who were diagnosed since the previous report on atrazine and cancer incidence from the AHS (Rusiecki et al. 2004
). Among the 36,357 atrazine users, there were 3,146 incident cancer diagnoses. As shown in , no association for cancer overall with either the lifetime days or intensity-weighted metric was apparent. For most cancer sites, there was little evidence for an association between atrazine use and incident cancer, including cancers of the prostate and lung, two of the most common sites diagnosed in this cohort. In general, results for the intensity-weighted and lifetime days of use metrics were similar.
Solid tumors among 36,357 atrazine users in the AHS, 1993–2007.
Based on 29 exposed cases, there was a statistically significant association between thyroid cancer and atrazine intensity-weighted lifetime days of use in the highest quartile of use [RR = 4.84; 95% confidence interval (CI), 1.31–17.93, p
-trend = 0.08] (). For lifetime days of use, there was also a suggestion of an association in the highest category (RR = 2.32; 95% CI, 0.66–8.22, p
-trend = 0.61). However, neither metric exhibited a monotonic trend. Because obesity has been linked with increased risk of thyroid cancer (Kitahara et al. 2011
), we further adjusted these results for body mass index. The RRs for intensity-weighted lifetime days were 4.30 (95% CI, 1.21–15.28), 1.09 (95% CI, 0.22–5.43), and 4.30 (95% CI, 1.19–15.57) for the second, third, and fourth quartiles of exposure, respectively, compared with the lowest quartile (p
-trend = 0.30). We were not able to stratify results by applicator type, because only one case of thyroid cancer occurred in commercial applicators and removing this case did not change risk estimates (data not shown). Twenty-two of the 29 exposed cases occurred in Iowa, and the RRs were 3.88 (95% CI, 1.01–14.90), 0.87 (95% CI, 0.14–5.42), and 4.50 (95% CI, 1.12–18.15) for the second, third, and fourth quartiles. There were no cases in North Carolina in the lowest quartile of exposure.
Esophageal cancer was significantly elevated in the second category of exposure for intensity-weighted lifetime days (RR = 2.83; 95% CI, 1.03–7.82), but not in the higher categories of intensity-weighted lifetime days (RR = 2.33; 95% CI, 0.81–6.67 and RR = 1.78; 95% CI, 0.59–5.30) or for lifetime days (). Similarly, a statistically significant increased risk for cancer of the oral cavity in the second category of atrazine lifetime use compared with the lowest use category was observed (RR = 2.09; 95% CI, 1.09–3.99). However, the RRs in the higher two quartiles were closer to the null (RR = 1.43 and 1.19, respectively), and RRs for intensity-weighted lifetime days were not significantly elevated. RRs for intensity-weighted lifetime days were not elevated. For pancreatic cancer (n = 51), there were decreased risks in all exposure categories, which was statistically significant in the second exposure category for intensity-weighted lifetime days (RR = 0.38; 95% CI, 0.17–0.87, p-trend = 0.18), but no evidence of any association with lifetime days of exposure. Based on 24 cases, there were decreased risks for laryngeal cancer for both metrics of exposure in the second, third, and fourth quartiles, although none were statistically significant (p-trend = 0.05 and p-trend = 0.09 for lifetime and intensity-weighted lifetime days, respectively).
Because of a priori interest in hormonally related cancers, we also evaluated use of atrazine and breast cancer (n = 36) and ovarian cancer (n = 9) among female pesticide applicators, although the number of cases was limited. We used the nonexposed category as the referent (n = 1,154) and evaluated only ever use of atrazine (n = 312). Results presented are adjusted only for age, but results adjusted for other factors, including parity, were similar (data not shown). There was no association between atrazine use and breast cancer based on 9 exposed and 27 unexposed cases (RR = 1.14; 95% CI, 0.47–2.50). Assessing risk for those above the median level of lifetime days of use also provided no evidence of risk. There were only four cases of breast cancer among male applicators, which precluded us from carrying out a separate analysis for breast cancer among men. For ovarian cancer, the RR for ever use of atrazine was 2.91 (95% CI, 0.56–13.60), based on four cases among the women who reported atrazine use and five cases among those who reported no use. All exposed cases occurred among those with less than the median of atrazine use among all cancer cases.
Use of atrazine did not appear to be associated with any lymphohematopoietic malignancy (). For NHL, RRs were near 1.0 for both lifetime days and intensity-weighted lifetime days. Similarly, null patterns were observed for all subtypes. We evaluated potential effect modification by use of chemicals previously reported to interact with atrazine and NHL (alachlor, carbofuran, and diazinon) by conducting analyses stratified by use of these chemicals. We saw no evidence of stratum-specific differences (data not shown). There was also no association with leukemia or multiple myeloma.
Lymphohematopoietic malignancies among 36,357 atrazine users in the AHS, 1993–2007.
Because of the differences in pesticide usage patterns between private and commercial applicators and the higher prevalence of atrazine usage in Iowa than in North Carolina, we also evaluated each cancer site stratified by state and by license type (private vs. commercial). We observed no meaningful differences in the results for any cancer site (data not shown).