An Updated Evaluation of Atrazine-Cancer Incidence Associations among Pesticide Applicators in the Agricultural Health Study Cohort

Abstract

Background:

Atrazine is a common agricultural herbicide in the United States. Few epidemiologic studies have evaluated cancer risks. Previous analyses within the Agricultural Health Study (AHS) have found some evidence of associations with cancer at some sites.

Objective:

We updated exposure information, incident cases, and follow-up time to assess the associations between atrazine use and cancer at specific sites in the AHS.

Methods:

Information about lifetime pesticide use was reported at enrollment (1993–1997) and follow-up (1999–2005). Among 53,562 pesticide applicators in North Carolina and Iowa, we identified 8,915 incident cases through cancer registry linkages through 2014 (North Carolina)/2017 (Iowa). We used Poisson regression to evaluate the association between ever/never and intensity-weighted lifetime days of atrazine use and incident cancer risk controlling for several confounders. We also evaluated lagged exposures and age-stratified risk.

Results:

Approximately 71.2% of applicators reported ever using atrazine, which was associated with lung cancer [$\text{rate ratios\hspace{0.17em}}\left(\text{RR}\right)=1.24$; 95% confidence interval (CI): 1.04, 1.46]. Aggressive prostate cancer risk was increased in the highest quartile (${\text{RR}}_{\mathrm{Q}4}=1.20$; 95% CI: 0.95, 1.52; $p\text{-trend}=0.19$), particularly among those $<60$ years old (${\text{RR}}_{\mathrm{Q}4}=3.04$; 95% CI: 1.61, 5.75; $p\text{-trend}<0.001$; $p\text{-interaction}=0.04$). Among applicators $<50$ years of age, ever-atrazine use was associated with non-Hodgkin lymphoma (NHL) ($\text{RR}=2.43$; 95% CI: 1.10, 5.38; $p\text{-interaction}=0.60$). For soft tissue sarcoma, there was an elevated risk in the highest tertile of exposure (${\text{RR}}_{\mathrm{T}3}$: 2.54; 95% CI: 0.97, 6.62; $p\text{-trend}=0.31$). In analyses with exposure lagged by 25 years, there was an elevated risk of pharyngeal (${\text{RR}}_{\mathrm{T}3}=3.04$; 95% CI: 1.45, 6.36; $p\text{-trend}=0.07$) and kidney (${\text{RR}}_{\mathrm{Q}4}=1.62$; 95% CI: 1.15, 2.29; $p\text{-trend}<0.005$) cancers.

Discussion:

We observed suggestive associations with some malignancies in overall, age-specific, and lagged analyses. Associations with aggressive prostate cancer and NHL were apparent among those diagnosed at younger ages and with cancers of the pharynx and kidney, and soft tissue sarcomas were observed in lagged analyses. Further work is needed to confirm these observed associations and elucidate potential underlying mechanisms. https://doi.org/10.1289/EHP13684

Introduction

Atrazine is a chlorinated-triazine herbicide used to control broadleaf and grassy weeds in various crops such as corn, sorghum, sugar cane, conifers, turf grasses, and bioenergy plants.¹ According to US Environmental Protection Agency (USEPA) pesticide market estimates from 2008 to 2012, atrazine was the second most commonly applied conventional pesticide active ingredient in the US within the agricultural sector, with an average of $72\text{\hspace{0.17em}million}$ pounds applied annually.^1,2 Atrazine has been documented to cause adverse developmental toxicity and endocrine disruption within ecological and human systems,^3,4 and atrazine-contaminated runoff and sediment from applied agricultural soils have become a source of environmental degradation.^5,6 The USEPA regulates atrazine through the Safe Drinking Water Act with an enforceable maximum contaminant level of $0.003\mathrm{mg}/\mathrm{L}$ for drinking water.⁷ This level is assumed to safeguard against potential cardiovascular and reproductive health effects from elevated long-term exposures. However, privately owned wells are not regulated by the USEPA and are not subject to drinking water quality standards.⁸ Elevated concentrations of atrazine in surface and groundwater drinking sources have been observed in rural and farming communities in the US.^9–12 Since 2004, the European Union has banned atrazine due to concerns about atrazine’s prevalence in groundwater and environmental effects.¹³

According to the most current International Agency for Research on Cancer (IARC) review of atrazine in 1999, atrazine is considered a Group 3 agent: “not classifiable as to its carcinogenicity in humans.”¹⁴ In 2020, the USEPA released an Interim Registration Review Decision for Atrazine, which classified atrazine as “not likely to be carcinogenic to humans” but recommended personal protective equipment and restricted atrazine application practices to address the risk of luteinizing hormone surge suppression among occupational applicators and handlers.¹

Previous evaluations of atrazine within the Agricultural Health Study (AHS) cohort have found adverse health outcomes, including potential carcinogenic effects. A 2011 AHS study of atrazine and cancer found no significant associations with studied sites,¹⁵ except with thyroid cancer among the highest atrazine users. However, a more recent AHS study of incident thyroid cancer in males found no association with atrazine with additional follow-up.¹⁶ Although there was no reported relationship between atrazine use and lung cancer risk in an analysis of the full AHS cohort,¹⁷ an evaluation of atrazine-acetochlor mixtures in a subset of the cohort found that atrazine and acetochlor mixtures may confer higher lung cancer risk than pesticides that include only acetochlor.¹⁸ The same study also observed nonsignificant positive associations between atrazine-alone applications and all cancer sites, lung cancer, and melanoma. A subsequent 2020 AHS evaluation found a significant exposure–response association between atrazine use and the risk of renal cell carcinoma when exposure was lagged by 20 years.¹⁹

In this work, we updated the 2011 analysis of atrazine and the incidence of cancer at multiple sites by incorporating an additional 8.5 average years of follow-up, which doubled the number of incident cancer cases among pesticide applicators, and updated atrazine use information.

Methods

Study Design

The AHS is a prospective cohort that includes 57,310 licensed pesticide applicators enrolled in Iowa or North Carolina between 1993 and 1997.²⁰ Each participant completed a self-administered questionnaire on detailed information about pesticide use, agricultural practices, demographic characteristics, and personal and family medical history. A follow-up telephone interview between 1999 and 2005 was completed by 63% of the participants. The AHS questionnaires can be accessed at https://aghealth.nih.gov/collaboration/questionnaires.html. The initial protocol was approved by the National Cancer Institute (NCI) Special Studies institutional review board (IRB) and NCI contractors Westat, the University of Iowa, and Battelle. All participants implied consent by completing the enrollment questionnaire, consistent with IRB rules in effect at the time.

Incident cancers diagnosed between enrollment and end of study follow-up were identified through linkage with state cancer registries in Iowa (enrollment through 2017) and North Carolina (enrollment through 2014). Cases were classified using the International Classification of Disease- Oncology, Third Revision (ICD-O-3)²¹; lymphoma recodes were based on the Surveillance, Epidemiology, and End Results Program Lymphoma Subtype Recodes (http://seer.cancer.gov/lymphomarecode/lymphoma-orig.html). For this study, we defined aggressive prostate cancer as any prostate cancer case with one or more of the following tumor characteristics: a Gleason score at or above 8, a diagnosed grade or stage at or above 3, or where prostate cancer was reported as the cause of death. We selected a more stringent Gleason score of $\ge 8$ to represent more aggressive tumors that are poorly differentiated, which differs from previous AHS studies,^22,23 given the additional number of exposed cases and follow-up person-time. We linked to regularly updated state mortality registries and the National Death Index through 2017 to determine vital status of participants.

Exposure Assessment

Self-reported use of atrazine and 49 other pesticides during the course of a participant’s lifetime was obtained in the enrollment questionnaire (1993–1997) and the follow-up interview (1999–2005). At enrollment, applicators reported the number of years and days per year each pesticide was used. At follow-up, applicators reported the number of days each pesticide was used in the most recent year farmed. We combined this lifetime use data with an intensity score derived from an algorithm based on literature-based measurements and information provided by the applicator, specifically whether the participant mixed or applied pesticides, repaired pesticide-related equipment, and used personal protective equipment.²⁴ Information on atrazine at enrollment and follow-up were combined to generate estimates of cumulative lifetime days and intensity-weighted lifetime days (IWLD) of use.

Covariates representing demographic information, medical history, farm characteristics, and reported pesticide use at enrollment were used to impute missing pesticide use data at follow-up. A data-driven multiple imputation procedure was used to impute pesticide use since enrollment.²⁵ Previous analyses found the adopted imputation approach yielded similar cancer risk estimates between participants with completed follow-up information and participants with completed and imputed data, and consequently, the imputation did not introduce meaningful bias.²⁶ Cumulative IWLD were categorized into quartiles based on the exposure distribution among all exposed cancer cases with the nonexposed cases serving as the referent group in the primary analyses. When quartiles resulted in cell sizes of less than five, we used tertile or median cut points for unlagged and corresponding lagged models. In sensitivity analyses, we used the lowest exposed quartile as the referent group.

Statistical Analysis

Participants with missing or zero follow-up time ($n=341$), cancer diagnoses before enrollment ($n=\mathrm{1,096}$), and missing atrazine use information ($n=\mathrm{2,311}$) were excluded. Accumulated person-time for each participant was calculated from enrollment until the earliest movement out of state, cancer diagnosis, death, or the end of the follow-up period (2014 for North Carolina, 317,604 person-years; 2017 for Iowa, 693,847 person-years).

This analysis includes time-varying covariates such as cumulative IWLD, calendar year of follow-up, and attained age from cohort participants. These time-dependent covariates were assumed constant over fixed 2-year time intervals from enrollment until a cancer diagnosis or censored event for each individual. We estimated accrued person-time within each interval and calculated time-dependent covariate values at the start of each interval and incorporated these data into Poisson regression models.^27–29 Models calculated incidence rate ratios (RR) and 95% confidence intervals (CIs) for each quartile of atrazine use using the unexposed as the referent group (Q0). Results from imputed data sets ($n=5$) were pooled into final point estimates and standard errors by Rubin’s rules³⁰ using the mice package.³¹ A $p$-value for trend ($p$-trend) was found using the median of each exposure category as a continuous variable and a Wald test statistic. Analyses using the lowest quartile group as a referent group (Q1) are found in the supplemental material for comparison to the 2011 work.¹⁵

In addition to the main exposure–response analysis, we also determined cancer risks associated with ever/never atrazine use to thoroughly estimate potential risks. We also evaluated potential latency from cumulative atrazine IWLD exposures through lagged analysis by subtracting duration time at varying 5-year intervals (5 to 25 years) prior to cancer diagnosis or a censoring event (death, state relocation, or end of follow-up). All models were adjusted for attained age, ${\log }_{10}$ of attained age, sex (male/female), race (white, black, American Indian or Alaskan Native, Asian or Pacific Islander, and other), state of residence at enrollment (Iowa, North Carolina), applicator type (private, commercial), education status (less than high school, high school or equivalent, beyond high school), family medical history of any cancer (yes, no, missing), alcohol consumption within the last year (ever, never, missing), pack-years of cigarette smoking (never, tertiles among former and current smokers, missing), and ever use of five pesticides most correlated with atrazine based on IWLD ($r>0.35$: 2,4-D, alachlor, cyanazine, metolachlor, and glyphosate). Respondents who identified as American Indian or Alaskan, Asian or Pacific Islander, or other comprised $<1\%$ of AHS participants. We collapsed these groups into the other category for analysis. All covariate data were obtained from self-administered questionnaires. In a sensitivity analysis for lung cancer, we also adjusted for endotoxin exposure based on self-reported farm tasks associated with elevated endotoxin levels, including contact with livestock and poultry, as well as stored hay and grain, to explore a potential reduction in lung cancer risk.^32,33 A previous AHS analysis on standardized incidence rates for cancer sites found an overall deficit of lung cancers among endotoxin-exposed participants.³⁴

We also conducted age-stratified ($<50$, 50–59, 60–69, $\ge 70$) analyses where numbers permitted (i.e., sites with at least 20 exposed cases per age strata) using the same covariates as in the main models. Sites with $<20$ exposed cases in younger strata were collapsed. As an example, we could combine the $<50$ and 50–59 age categories to $<60$ years as recategorized age groups for cancer sites to increase the number of exposed cases. Tests for interaction between age and atrazine as categorical variables were performed using cross-product interaction terms in regression models and type III Wald test (p-interaction). In addition, in our main models, we conducted a sensitivity analysis in which we used 55 years of age at diagnosis as a clinically meaningful age for early onset prostate cancer.³⁵ All analyses were performed using R (version 4.0.5; R Development Core Team). All statistical tests were two-sided with an alpha of 0.05.

Results

Among 53,562 participants, 38,383 (72%) reported using atrazine at enrollment or follow-up. The cohort is comprised primarily of males (97%), white (97%), and private applicators (91%) as seen in Table 1. Most applicators that reported higher atrazine use (Q2–Q4) were from Iowa and had no family history of cancer. Among incident cancer cases diagnosed during follow-up that reported atrazine use ($n=\mathrm{6,407}$), the median lifetime days of atrazine use was estimated 64 d (25th to 75th percentile $\text{range}=25–189$ days).

Table 1.

Selected characteristics of AHS pesticide applicators by atrazine use category from enrollment questionnaire data, 1993–1997.

Characteristic	Overall, $n=\mathrm{53,562}$ [$n$ (%)]	Q0, $n=\mathrm{15,413}$ [$n$ (%)]	Q1, $n=\mathrm{10,045}$ [$n$ (%)]	Q2, $n=\mathrm{9,517}$ [$n$ (%)]	Q3, $n=\mathrm{9,624}$ [$n$ (%)]	Q4, $n=\mathrm{8,963}$ [$n$ (%)]
Incident cancer cases	8,915 (17%)	2,508 (16%)	1,591 (16%)	1,610 (17%)	1,605 (17%)	1,601 (18%)
Attained age
$<50$	5,405 (10%)	2,357 (15%)	1,071 (11%)	776 (8%)	719 (8%)	482 (5%)
50–59	13,100 (24%)	3,616 (23%)	2,637 (26%)	2,379 (25%)	2,413 (25%)	2,055 (23%)
60–69	17,023 (32%)	4,235 (27%)	3,132 (31%)	3,131 (33%)	3,258 (34%)	3,267 (36%)
70 and over	18,034 (34%)	5,205 (34%)	3,205 (32%)	3,231 (34%)	3,234 (34%)	3,159 (35%)
Sex
Female	1,458 (3%)	1,122 (7%)	151 (2%)	77 (1%)	61 (1%)	47 (1%)
Male	52,104 (97%)	14,291 (93%)	9,894 (98%)	9,440 (99%)	9,563 (99%)	8,916 (99%)
Race
Black	1,070 (2%)	536 (4%)	183 (2%)	111 (1%)	116 (1%)	124 (1%)
White	52,043 (97%)	14,676 (96%)	9,792 (98%)	9,338 (98%)	9,458 (98%)	8,779 (98%)
Other	313 (1%)	143 (<1%)	47 (<1%)	41 (<1%)	38 (<1%)	44 (1%)
Missing	136	58	23	27	12	16
State residence
Iowa	34,986 (65%)	7,075 (46%)	7,468 (74%)	7,205 (76%)	7,220 (75%)	6,018 (67%)
North Carolina	18,576 (35%)	8,338 (54%)	2,577 (26%)	2,312 (24%)	2,404 (25%)	2,945 (33%)
Alcohol use
Never	16,439 (32%)	5,784 (40%)	2,805 (29%)	2,594 (28%)	2,660 (29%)	2,596 (30%)
Ever in last year	34,595 (68%)	8,554 (60%)	6,739 (71%)	6,587 (72%)	6,633 (71%)	6,082 (70%)
Missing	2,528	1,075	501	336	331	285
Smoking status and pack-years in tertiles
Never	28,130 (54%)	7,433 (50%)	5,611 (58%)	5,285 (57%)	5,240 (56%)	4,561 (53%)
Former, $<3.75$	5,366 (10%)	1,399 (10%)	1,054 (11%)	1,006 (11%)	1,002 (11%)	905 (10%)
Former, 3.75–15	4,830 (9%)	1,446 (10%)	865 (9%)	850 (9%)	849 (9%)	820 (10%)
Former, $>15$	4,598 (9%)	1,424 (10%)	735 (8%)	767 (8%)	816 (9%)	856 (10%)
Current, $<3.75$	3,057 (6%)	1,126 (8%)	533 (6%)	494 (5%)	474 (5%)	430 (5%)
Current, 3.75–15	2,819 (6%)	940 (6%)	503 (5%)	427 (5%)	469 (5%)	480 (6%)
Current, $>15$	2,896 (6%)	970 (7%)	441 (5%)	410 (4%)	481 (5%)	594 (7%)
Missing	1,866	675	303	278	293	317
Attained education
Less than HS	4,865 (9.3%)	2,039 (14%)	804 (8%)	727 (8%)	674 (7%)	621 (7%)
HS/GED	24,947 (48%)	6,637 (44%)	4,735 (48%)	4,555 (49%)	4,630 (49%)	4,390 (50%)
Beyond HS	22,613 (43%)	6,325 (42%)	4,311 (44%)	4,059 (43%)	4,153 (44%)	3,765 (43%)
Missing	1,137	412	195	176	167	187
Family cancer history
No	29,175 (58%)	8,546 (61%)	5,503 (59%)	5,181 (57%)	5,137 (56%)	4,808 (57%)
Yes	20,732 (42%)	5,385 (39%)	3,841 (41%)	3,863 (43%)	3,968 (44%)	3,675 (43%)
Missing	3,655	1,482	701	473	519	480
Applicator type
Commercial	4,581 (9%)	1,994 (13%)	494 (5%)	420 (4%)	626 (7%)	1,047 (12%)
Private	48,981 (91%)	13,419 (87%)	9,551 (95%)	9,097 (96%)	8,998 (93%)	7,916 (88%)
Cyanazine ever use
No	29,212 (59%)	12,383 (90%)	5,465 (59%)	4,427 (49%)	3,777 (41%)	3,160 (37%)
Yes	20,453 (41%)	1,373 (10%)	3,784 (41%)	4,565 (51%)	5,353 (59%)	5,378 (63%)
Missing	3,897	1,657	796	525	494	425
Metolachlor ever use
No	26,666 (54%)	11,390 (83%)	4,883 (53%)	4,057 (45%)	3,556 (39%)	2,780 (33%)
Yes	23,006 (46%)	2,347 (17%)	4,380 (47%)	4,943 (55%)	5,586 (61%)	5,750 (67%)
Missing	3,890	1,676	782	517	482	433
Alachlor ever use
No	23,783 (48%)	10,868 (79%)	4,308 (47%)	3,465 (38%)	3,016 (33%)	2,126 (25%)
Yes	26,011 (52%)	2,872 (21%)	4,956 (53%)	5,542 (62%)	6,162 (67%)	6,479 (75%)
Missing	3,768	1,673	781	510	446	358
Glyphosate ever use
No	12,932 (24%)	4,821 (32%)	2,525 (25%)	2,220 (23%)	1,921 (20%)	1,445 (16%)
Yes	40,358 (76%)	10,477 (68%)	7,474 (75%)	7,253 (77%)	7,663 (80%)	7,491 (84%)
Missing	272	115	46	44	40	27
2,4-D ever use
No	13,561 (26%)	7,454 (49%)	2,144 (22%)	1,633 (17%)	1,317 (14%)	1,013 (11%)
Yes	39,502 (74%)	7,747 (51%)	7,805 (78%)	7,810 (83%)	8,251 (86%)	7,889 (89%)
Missing	499	212	96	74	56	61

Note: Quartile cut points based on cumulative intensity-weighted lifetime days quartile cut-points: Q0: 0 (no reported use); Q1: 2–918; Q2: 919–2,634; Q3: 2,635–7,370; Q4: $\ge \mathrm{7,371}$. Among applicators reporting atrazine use, 234 participants reported atrazine use at enrollment or follow-up had missing frequency information. Consequently, cumulative intensityweighted life days for those persons were not estimated and not assigned exposure use categories. AHS, Agricultural Health Study; GED, General Educational Diploma; HS, high school.

Ever use of atrazine was not associated with cancer overall and for most specific cancers except for lung cancer ($\text{RR}=1.24$; 95% CI: 1.04, 1.46) (Table 2). Exposure–response analyses for lung cancer revealed elevated RRs for all quartile categories but no suggestion of a linear trend ($p\text{-trend}=0.18$) (Table 3). Results for lung cancer in the 10- and 25-year lagged analyses were similar with a pattern of elevated but not statistically significant associations. Results for five- (Lag 5), fifteen- (Lag 15), and twenty (Lag 20) year lagged exposures, presented in Excel Table S2, demonstrate consistency in the positive direction as well.

Table 2.

Adjusted RRs and 95% CIs from Poisson regressions for atrazine exposure (ever/never used) and selected cancers among AHS applicators, 1993–2017.

Cancer sitea	Never casesb	Ever casesc	RR	95% CI
All	2,508	6,534	1.01	0.95	1.06
Organ-specific
Oral Cavity	58	161	0.93	0.66	1.31
Lip	14	55	1.07	0.56	2.03
Tongue	13	27	0.59	0.27	1.26
Pharyngeal	15	37	0.91	0.45	1.85
Esophagus	27	98	1.39	0.85	2.27
Stomach	37	79	0.66	0.42	1.04
Small intestine	15	37	0.76	0.38	1.52
Colon	184	430	0.96	0.78	1.17
Rectum	80	197	1.08	0.79	1.46
Liver	19	49	1.16	0.62	2.15
Gall biliary	14	24	0.90	0.40	2.00
Pancreas	64	145	0.99	0.70	1.40
Laryngeal	28	49	0.64	0.37	1.10
Lung	267	581	1.24	1.04	1.46
Soft tissue sarcoma	8	35	1.84	0.8	4.22
Cutaneous melanoma	112	332	1.05	0.82	1.34
Female breast	46	15	0.97	0.45	2.08
Prostate	876	2,575	0.99	0.91	1.08
Aggressive prostated	196	564	1.03	0.86	1.24
Testis	13	39	1.17	0.57	2.42
Kidney	93	240	0.93	0.7	1.23
Bladder	123	331	1.07	0.84	1.35
Brain	29	79	0.92	0.55	1.52
Thyroid	27	67	0.96	0.56	1.64
Blood-specific
All lymphohematopoietic	241	667	0.98	0.83	1.17
Non-Hodgkin lymphoid malignancies	192	557	0.97	0.81	1.18
Non-Hodgkin lymphoma, B-Cell	181	514	0.96	0.79	1.17
Mature B-cell lymphoma	176	493	0.95	0.78	1.16
CLL/SLL	43	128	0.86	0.58	1.27
Diffuse large B-cell lymphoma	43	120	0.98	0.66	1.46
Follicular lymphoma	29	64	0.58	0.35	0.97
Multiple myeloma	45	114	1.02	0.67	1.54
Non-Hodgkin lymphoma, T-cell	5	27	1.12	0.43	2.93
Leukemia	43	95	1.02	0.68	1.55
Myeloid leukemia	37	79	0.94	0.6	1.47
Acute myeloid leukemia	25	59	0.97	0.57	1.67
Chronic myeloid leukemia	10	18	0.93	0.38	2.23

Note: AHS, Agricultural Health Study; CI, confidence interval; CLL, chronic lymphocytic leukemia; RR, rate ratio; SLL, small lymphocytic lymphoma.

^a Main models were all adjusted for attained age, logarithmic age, year of enrollment, biological sex, race, state of residence, pack-years of cigarette smoking, attained education, family history of cancer, and correlated pesticides (2,4-D, alachlor, cyanazine, metolachlor, and glyphosate). Reproductive-related cancer sites were restricted to biological sex, and sex covariate was not included.

^b Never exposed cases are defined as participants that have not reported atrazine use at enrollment or at follow-up. Never exposed cases contributed to a total of 275,521 person-years.

^c Ever cases are defined as participants that have reported atrazine use at enrollment or at follow-up. Ever exposed cases contributed to a total of 735,930 person-years.

^d Aggressive prostate cancer outcome based on one of more of the following: Gleason score $8+$, stage 3 diagnosis, or death from prostate cancer.

Table 3.

Unlagged and lagged adjusted RRs and 95% CIs from Poisson regressions for cancer incidence associated with cumulative atrazine intensity-weighted days of exposure compared to reported no atrazine use among agricultural health study cohort from 1993 to 2017 ($n=\mathrm{53,562}$ total participants and 1,011,451 total person-years).

Cancer sitesa	Pesticide useb,c	Unlagged					10-year Lagd					25-year Lagd
		Cases	RR	95% CI		$p$-trende	Cases	RR	95% CI		$p$-trende	Cases	RR	95% CI		$p$-trende
All	Q0	2,508	1.00	1.00	1.00	—	2,850	1.00	1.00	1.00	—	4,450	1.00	1.00	1.00	—
	Q1	1,591	0.98	0.92	1.05	—	1,541	0.99	0.92	1.05	—	1,147	0.99	0.92	1.06	—
	Q2	1,610	1.02	0.95	1.09	—	1,556	1.00	0.93	1.07	—	1,145	1.00	0.94	1.08	—
	Q3	1,605	1.00	0.93	1.07	—	1,550	1.00	0.93	1.07	—	1,152	1.04	0.97	1.11	—
	Q4	1,601	1.04	0.97	1.12	0.21	1,545	1.05	0.98	1.12	0.21	1,148	1.05	0.98	1.13	0.12
Organ-specific
Oral cavity	Q0	58	1.00	1.00	1.00	—	69	1.00	1.00	1.00	—	109	1.00	1.00	1.00	—
	Q1	36	0.80	0.51	1.25	—	37	0.83	0.54	1.28	—	22	0.79	0.49	1.27	—
	Q2	37	0.87	0.56	1.37	—	31	0.74	0.47	1.17	—	27	1.02	0.65	1.59	—
	Q3	36	0.87	0.55	1.37	—	33	0.79	0.50	1.25	—	23	0.89	0.55	1.44	—
	Q4	49	1.25	0.81	1.93	0.54	49	1.27	0.83	1.92	0.65	38	1.54	1.02	2.32	0.32
Lip	Q0	14	1.00	1.00	1.00	—	18	1.00	1.00	1.00	—	35	1.00	1.00	1.00	—
	Q1	9	0.74	0.31	1.76	—	12	0.80	0.36	1.76	—	7	0.61	0.26	1.40	—
	Q2	21	1.62	0.78	3.39	—	14	1.04	0.49	2.20	—	13	1.14	0.58	2.25	—
	Q3	16	1.29	0.59	2.81	—	16	1.15	0.55	2.42	—	8	0.71	0.31	1.59	—
	Q4	9	0.79	0.32	1.95	0.78	9	0.72	0.30	1.72	0.90	6	0.56	0.22	1.40	0.27
Tongue	T0	13	1.00	1.00	1.00	—	13	1.00	1.00	1.00	—	21	1.00	1.00	1.00	—
	T1	11	0.69	0.29	1.66	—	11	0.90	0.38	2.13	—	5	0.61	0.22	1.67	—
	T2	6	0.40	0.14	1.15	—	6	0.49	0.17	1.40	—	8	1.04	0.43	2.50	—
	T3	10	0.70	0.27	1.79	0.24	10	0.90	0.35	2.31	0.65	6	0.82	0.30	2.19	0.83
Pharyngeal	T0	15	1.00	1.00	1.00	—	18	1.00	1.00	1.00	—	25	1.00	1.00	1.00	—
	T1	8	0.55	0.21	1.44	—	7	0.56	0.22	1.42	—	6	1.03	0.40	2.63	—
	T2	9	0.70	0.28	1.76	—	9	0.73	0.30	1.77	—	5	0.92	0.33	2.55	—
	T3	18	1.35	0.59	3.10	0.64	18	1.50	0.68	3.29	0.61	16	3.04	1.45	6.36	0.07
Esophagus	Q0	27	1.00	1.00	1.00	—	32	1.00	1.00	1.00	—	55	1.00	1.00	1.00	—
	Q1	23	1.39	0.76	2.51	—	23	1.24	0.70	2.20	—	19	1.30	0.75	2.25	—
	Q2	26	1.55	0.85	2.80	—	26	1.32	0.75	2.35	—	15	1.00	0.55	1.84	—
	Q3	23	1.23	0.66	2.30	—	20	1.02	0.55	1.89	—	19	1.27	0.72	2.24	—
	Q4	24	1.31	0.70	2.45	0.65	24	1.22	0.67	2.22	0.85	17	1.10	0.61	2.00	0.52
Stomach	Q0	37	1.00	1.00	1.00	—	41	1.00	1.00	1.00	—	58	1.00	1.00	1.00	—
	Q1	19	0.66	0.37	1.20	—	19	0.73	0.41	1.30	—	11	0.68	0.35	1.32	—
	Q2	18	0.60	0.33	1.12	—	14	0.52	0.27	0.99	—	16	0.97	0.54	1.74	—
	Q3	24	0.83	0.47	1.47	—	25	0.89	0.51	1.56	—	17	1.07	0.60	1.91	—
	Q4	18	0.63	0.34	1.20	0.19	17	0.68	0.37	1.28	0.29	14	0.92	0.49	1.71	1.00
Small intestine	Q0	15	1.00	1.00	1.00	—	15	1.00	1.00	1.00	—	25	1.00	1.00	1.00	—
	Q1	9	0.69	0.28	1.71	—	9	0.92	0.38	2.22	—	5	0.63	0.23	1.70	—
	Q2	6	0.70	0.28	1.77	—	8	1.04	0.43	2.50	—	8	1.05	0.45	2.45	—
	Q3	13	0.95	0.40	2.26	—	12	1.05	0.43	2.58	—	10	1.37	0.62	3.05	—
	Q4	9	0.84	0.33	2.14	0.94	8	0.93	0.36	2.44	0.91	4	0.59	0.20	1.80	0.82
Colon	Q0	184	1.00	1.00	1.00	—	207	1.00	1.00	1.00	—	318	1.00	1.00	1.00	—
	Q1	112	0.97	0.75	1.26	—	106	0.99	0.77	1.27	—	78	1.00	0.77	1.30	—
	Q2	108	1.02	0.79	1.32	—	107	1.05	0.81	1.36	—	79	1.04	0.80	1.35	—
	Q3	94	0.84	0.64	1.11	—	87	0.85	0.64	1.12	—	67	0.91	0.69	1.21	—
	Q4	111	1.11	0.85	1.45	0.85	107	1.13	0.87	1.48	0.69	72	1.01	0.76	1.33	0.79
Rectum	Q0	80	1.00	1.00	1.00	—	85	1.00	1.00	1.00	—	139	1.00	1.00	1.00	—
	Q1	54	1.13	0.78	1.65	—	51	1.17	0.80	1.70	—	35	1.13	0.77	1.67	—
	Q2	41	0.97	0.64	1.45	—	41	1.05	0.70	1.57	—	31	1.06	0.70	1.61	—
	Q3	55	1.24	0.84	1.83	—	56	1.37	0.93	2.00	—	43	1.56	1.07	2.26	—
	Q4	46	1.07	0.71	1.62	0.46	44	1.15	0.76	1.74	0.25	29	1.10	0.71	1.69	0.17
Liver	T0	19	1.00	1.00	1.00	—	22	1.00	1.00	1.00	—	31	1.00	1.00	1.00	—
	T1	8	0.69	0.29	1.65	—	6	0.48	0.19	1.23	—	8	0.99	0.44	2.23	—
	T2	17	1.32	0.63	2.78	—	18	1.34	0.66	2.71	—	12	1.35	0.65	2.78	—
	T3	22	1.51	0.72	3.14	0.39	22	1.51	0.75	3.05	0.46	17	1.74	0.89	3.41	0.09
Gall biliary	M0	14	1.00	1.00	1.00	—	15	1.00	1.00	1.00	—	17	1.00	1.00	1.00	—
	M1	8	0.60	0.23	1.56	—	9	0.67	0.27	1.66	—	8	1.17	0.47	2.90	—
	M2	15	1.22	0.50	2.98	0.73	14	1.15	0.48	2.77	0.73	13	2.12	0.91	4.90	0.30
Pancreas	Q0	64	1.00	1.00	1.00	—	68	1.00	1.00	1.00	—	102	1.00	1.00	1.00	—
	Q1	43	1.20	0.79	1.82	—	44	1.29	0.86	1.96	—	25	0.99	0.63	1.57	—
	Q2	33	0.90	0.56	1.43	—	31	0.93	0.59	1.48	—	30	1.20	0.77	1.85	—
	Q3	37	1.05	0.67	1.66	—	40	1.18	0.76	1.84	—	29	1.18	0.75	1.85	—
	Q4	30	0.85	0.52	1.40	0.74	26	0.82	0.50	1.36	1.00	23	0.94	0.58	1.54	0.97
Laryngeal	Q0	28	1.00	1.00	1.00	—	31	1.00	1.00	1.00	—	40	1.00	1.00	1.00	—
	Q1	12	0.74	0.37	1.46	—	13	0.74	0.37	1.47	—	9	0.96	0.45	2.04	—
	Q2	14	0.64	0.31	1.33	—	13	0.74	0.37	1.50	—	11	1.20	0.59	2.44	—
	Q3	12	0.55	0.26	1.19	—	11	0.62	0.29	1.30	—	9	1.00	0.46	2.15	—
	Q4	9	0.45	0.20	1.03	0.08	9	0.52	0.23	1.17	0.13	8	0.92	0.41	2.06	0.91
Lung	Q0	267	1.00	1.00	1.00	—	310	1.00	1.00	1.00	—	455	1.00	1.00	1.00	—
	Q1	146	1.28	1.03	1.59	—	137	1.20	0.97	1.48	—	96	1.08	0.86	1.36	—
	Q2	133	1.17	0.93	1.47	—	140	1.19	0.95	1.48	—	97	1.07	0.85	1.35	—
	Q3	142	1.24	0.99	1.55	—	129	1.10	0.88	1.38	—	109	1.18	0.94	1.48	—
	Q4	133	1.09	0.86	1.38	0.18	132	1.09	0.86	1.37	0.33	91	0.92	0.72	1.18	0.69
Soft tissue sarcoma	T0	8	1.00	1.00	1.00	—	9	1.00	1.00	1.00	—	20	1.00	1.00	1.00	—
	T1	13	2.23	0.90	5.51	—	12	2.00	0.82	4.88	—	6	0.99	0.38	2.59	—
	T2	8	0.97	0.32	2.97	—	9	1.10	0.39	3.14	—	7	1.20	0.47	3.04	—
	T3	14	2.54	0.97	6.62	0.31	13	2.39	0.93	6.13	0.21	10	1.89	0.80	4.44	0.21
Cutaneous melanoma	Q0	112	1.00	1.00	1.00	—	124	1.00	1.00	1.00	—	212	1.00	1.00	1.00	—
	Q1	83	1.03	0.76	1.39	—	77	1.10	0.82	1.49	—	60	1.17	0.87	1.58	—
	Q2	99	1.23	0.91	1.65	—	96	1.33	0.99	1.79	—	59	1.21	0.89	1.65	—
	Q3	81	1.09	0.80	1.49	—	81	1.19	0.87	1.63	—	59	1.25	0.92	1.71	—
	Q4	67	0.90	0.64	1.26	0.96	66	1.04	0.75	1.45	0.42	54	1.18	0.85	1.64	0.12
Female breast	M0	46	1.00	1.00	1.00	—	48	1.00	1.00	1.00	—	—	—	—	—	—
	M1	9	0.81	0.34	1.95	—	8	0.68	0.27	1.69	—	—	—	—	—	—
	M2	6	1.49	0.53	4.19	0.71	5	1.31	0.43	3.96	0.94	—	—	—	—	—
Prostate	Q0	876	1.00	1.00	1.00	—	1,007	1.00	1.00	1.00	—	1,619	1.00	1.00	1.00	—
	Q1	606	0.93	0.84	1.04	—	602	0.96	0.86	1.07	—	463	0.93	0.83	1.04	—
	Q2	639	1.00	0.90	1.12	—	598	0.95	0.85	1.06	—	450	0.91	0.82	1.02	—
	Q3	620	0.96	0.86	1.08	—	609	0.96	0.86	1.07	—	450	0.93	0.83	1.04	—
	Q4	660	1.07	0.95	1.20	0.26	635	1.05	0.94	1.18	0.48	469	0.99	0.88	1.10	0.42
Aggressive prostatef	Q0	196	1.00	1.00	1.00	—	226	1.00	1.00	1.00	—	348	1.00	1.00	1.00	—
	Q1	136	0.98	0.78	1.24	—	134	0.99	0.78	1.24	—	95	0.90	0.71	1.14	—
	Q2	139	1.03	0.81	1.30	—	123	0.94	0.75	1.19	—	114	1.09	0.87	1.37	—
	Q3	126	0.97	0.76	1.24	—	132	0.97	0.76	1.23	—	93	0.91	0.71	1.16	—
	Q4	151	1.20	0.95	1.52	0.19	145	1.16	0.92	1.47	0.39	110	1.08	0.86	1.37	0.75
Testis	T0	13	1.00	1.00	1.00	—	20	1.00	1.00	1.00	—	42	1.00	1.00	1.00	—
	T1	16	1.31	0.59	2.91	—	16	1.42	0.68	2.96	—	7	1.45	0.60	3.50	—
	T2	17	1.47	0.64	3.35	—	11	1.12	0.48	2.60	—	2	0.51	0.12	2.24	—
	T3	6	0.58	0.20	1.70	0.62	5	0.62	0.21	1.83	0.33	1	0.31	0.04	2.42	0.33
Kidney	Q0	93	1.00	1.00	1.00	—	100	1.00	1.00	1.00	—	145	1.00	1.00	1.00	—
	Q1	52	0.82	0.57	1.19	—	51	0.92	0.64	1.31	—	38	1.06	0.73	1.53	—
	Q2	62	1.00	0.70	1.43	—	59	1.05	0.73	1.49	—	49	1.41	1.00	2.00	—
	Q3	65	1.04	0.73	1.49	—	63	1.16	0.82	1.66	—	48	1.43	1.00	2.03	—
	Q4	60	0.98	0.67	1.41	0.79	60	1.09	0.76	1.58	0.33	53	1.62	1.15	2.29	0.004
Bladder	Q0	123	1.00	1.00	1.00	—	139	1.00	1.00	1.00	—	220	1.00	1.00	1.00	—
	Q1	89	1.19	0.89	1.59	—	84	1.16	0.87	1.54	—	59	0.95	0.70	1.28	—
	Q2	85	1.13	0.83	1.52	—	80	1.05	0.78	1.42	—	54	0.85	0.62	1.16	—
	Q3	77	0.97	0.71	1.33	—	74	0.95	0.69	1.30	—	53	0.84	0.61	1.16	—
	Q4	77	1.02	0.74	1.40	0.67	77	1.04	0.76	1.42	0.64	68	1.08	0.80	1.46	0.67
Brain	Q0	29	1.00	1.00	1.00	—	32	1.00	1.00	1.00	—	50	1.00	1.00	1.00	—
	Q1	18	1.03	0.56	1.91	—	16	0.91	0.48	1.71	—	18	1.53	0.86	2.72	—
	Q2	20	0.92	0.48	1.76	—	22	1.15	0.63	2.11	—	10	0.88	0.43	1.81	—
	Q3	20	0.94	0.49	1.80	—	18	0.99	0.52	1.88	—	14	1.30	0.68	2.46	—
	Q4	14	0.74	0.36	1.51	0.36	14	0.84	0.42	1.70	0.47	10	0.98	0.47	2.03	0.79
Thyroid	Q0	27	1.00	1.00	1.00	—	33	1.00	1.00	1.00	—	60	1.00	1.00	1.00	—
	Q1	17	0.89	0.46	1.74	—	15	0.87	0.46	1.67	—	11	0.90	0.46	1.77	—
	Q2	17	0.98	0.49	1.94	—	19	1.00	0.52	1.93	—	8	0.71	0.33	1.55	—
	Q3	14	0.68	0.32	1.46	—	14	0.85	0.42	1.72	—	10	0.95	0.46	1.96	—
	Q4	15	0.91	0.43	1.90	0.67	13	0.88	0.42	1.84	0.80	5	0.51	0.20	1.34	0.42
Blood-specific
All lymphohematopoietic	Q0	241	1.00	1.00	1.00	—	280	1.00	1.00	1.00	—	445	1.00	1.00	1.00	—
	Q1	158	0.93	0.75	1.15	—	147	0.88	0.71	1.09	—	123	1.03	0.83	1.27	—
	Q2	160	0.98	0.79	1.22	—	168	1.02	0.82	1.25	—	124	1.06	0.86	1.31	—
	Q3	176	1.04	0.84	1.30	—	166	1.02	0.83	1.27	—	109	0.96	0.76	1.20	—
	Q4	159	0.99	0.79	1.25	0.79	147	0.95	0.76	1.19	0.92	107	0.97	0.77	1.22	0.94
Non-Hodgkin lymphoid malignancies	Q0	192	1.00	1.00	1.00	—	222	1.00	1.00	1.00	—	361	1.00	1.00	1.00	—
	Q1	124	0.89	0.70	1.13	—	119	0.87	0.69	1.11	—	99	0.96	0.76	1.21	—
	Q2	139	1.01	0.79	1.28	—	142	1.03	0.82	1.30	—	103	1.01	0.80	1.27	—
	Q3	151	1.05	0.82	1.33	—	145	1.05	0.83	1.33	—	97	0.97	0.76	1.23	—
	Q4	131	0.97	0.76	1.25	0.80	121	0.94	0.73	1.21	0.79	89	0.92	0.71	1.18	0.75
Non-Hodgkin lymphoma, B-Cell	Q0	181	1.00	1.00	1.00	—	202	1.00	1.00	1.00	—	335	1.00	1.00	1.00	—
	Q1	116	0.88	0.69	1.13	—	107	0.85	0.66	1.09	—	92	0.95	0.75	1.21	—
	Q2	129	1.00	0.78	1.28	—	127	1.01	0.79	1.29	—	99	1.03	0.81	1.31	—
	Q3	135	1.00	0.78	1.28	—	127	1.00	0.78	1.28	—	86	0.91	0.71	1.18	—
	Q4	123	0.97	0.75	1.26	0.95	106	0.89	0.69	1.17	0.90	83	0.91	0.70	1.18	0.70
Mature B-cell lymphoma	Q0	176	1.00	1.00	1.00	—	202	1.00	1.00	1.00	—	325	1.00	1.00	1.00	—
	Q1	112	0.86	0.67	1.12	—	107	0.85	0.66	1.09	—	90	0.95	0.74	1.21	—
	Q2	123	0.99	0.77	1.27	—	127	1.01	0.79	1.29	—	94	1.00	0.78	1.28	—
	Q3	131	0.99	0.77	1.28	—	127	1.00	0.78	1.28	—	84	0.91	0.71	1.18	—
	Q4	116	0.94	0.72	1.22	1.00	106	0.89	0.69	1.17	0.90	76	0.85	0.65	1.11	0.46
CLL/SLL	Q0	43	1.00	1.00	1.00	—	49	1.00	1.00	1.00	—	83	1.00	1.00	1.00	—
	Q1	26	0.70	0.41	1.18	—	32	0.89	0.55	1.44	—	30	1.06	0.68	1.64	—
	Q2	40	1.13	0.70	1.81	—	37	1.05	0.65	1.67	—	28	0.99	0.63	1.57	—
	Q3	36	0.94	0.57	1.55	—	33	0.94	0.57	1.54	—	17	0.61	0.35	1.06	—
	Q4	23	0.67	0.38	1.17	0.47	20	0.62	0.35	1.10	0.29	13	0.49	0.26	0.91	0.04
Diffuse large B-cell lymphoma (DLBCL)	Q0	43	1.00	1.00	1.00	—	50	1.00	1.00	1.00	—	84	1.00	1.00	1.00	—
	Q1	31	0.95	0.58	1.57	—	29	1.00	0.61	1.63	—	18	0.83	0.49	1.41	—
	Q2	30	0.99	0.59	1.64	—	31	1.13	0.69	1.84	—	23	1.07	0.65	1.75	—
	Q3	34	1.13	0.69	1.87	—	32	1.20	0.74	1.97	—	24	1.16	0.71	1.90	—
	Q4	22	0.76	0.43	1.35	0.87	21	0.80	0.45	1.42	0.78	14	0.71	0.39	1.29	0.96
Follicular lymphoma	Q0	29	1.00	1.00	1.00	—	32	1.00	1.00	1.00	—	51	1.00	1.00	1.00	—
	Q1	18	0.65	0.34	1.25	—	16	0.69	0.36	1.30	—	11	0.73	0.37	1.43	—
	Q2	12	0.48	0.23	0.98	—	11	0.47	0.23	0.98	—	9	0.61	0.29	1.27	—
	Q3	15	0.53	0.26	1.06	—	16	0.63	0.32	1.24	—	8	0.55	0.25	1.21	—
	Q4	19	0.73	0.38	1.42	0.23	18	0.79	0.41	1.53	0.19	14	0.99	0.52	1.89	0.41
Multiple myeloma	Q0	45	1.00	1.00	1.00	—	53	1.00	1.00	1.00	—	75	1.00	1.00	1.00	—
	Q1	26	1.00	0.59	1.67	—	20	0.71	0.41	1.22	—	20	0.95	0.57	1.59	—
	Q2	23	0.83	0.47	1.44	—	29	0.92	0.55	1.52	—	21	0.98	0.58	1.64	—
	Q3	29	0.99	0.58	1.69	—	28	0.89	0.53	1.50	—	21	0.98	0.58	1.66	—
	Q4	32	1.08	0.64	1.84	0.92	29	0.96	0.57	1.61	0.85	22	1.01	0.60	1.71	0.74
Non-Hodgkin lymphoma, T-cell	T0	5	1.00	1.00	1.00	—	8	1.00	1.00	1.00	—	17	1.00	1.00	1.00	—
	T1	7	1.02	0.34	3.12	—	11	1.20	0.43	3.33	—	5	0.52	0.18	1.52	—
	T2	12	1.41	0.48	4.16	—	13	1.55	0.54	4.44	0.42	10	1.15	0.46	2.84	0.94
	T3	7	1.05	0.32	3.46	0.54	—	—	—	—	—	—	—	—	—	—
Leukemia	Q0	43	1.00	1.00	1.00	—	49	1.00	1.00	1.00	—	69	1.00	1.00	1.00	—
	Q1	26	1.05	0.62	1.76	—	23	0.94	0.55	1.59	—	20	1.25	0.74	2.11	—
	Q2	20	0.87	0.49	1.54	—	23	0.97	0.57	1.66	—	19	1.25	0.73	2.14	—
	Q3	21	0.98	0.56	1.73	—	19	0.89	0.51	1.58	—	11	0.75	0.38	1.46	—
	Q4	26	1.17	0.67	2.05	0.89	24	1.07	0.61	1.88	0.80	19	1.34	0.77	2.33	0.72
Myeloid leukemia	Q0	37	1.00	1.00	1.00	—	43	1.00	1.00	1.00	—	61	1.00	1.00	1.00	—
	Q1	23	0.96	0.55	1.70	—	20	0.83	0.47	1.48	—	16	1.08	0.61	1.92	—
	Q2	16	0.76	0.41	1.43	—	19	0.86	0.48	1.55	—	15	1.06	0.58	1.93	—
	Q3	15	0.78	0.41	1.48	—	13	0.67	0.35	1.29	—	7	0.52	0.23	1.16	—
	Q4	23	1.17	0.65	2.12	1.00	21	1.04	0.57	1.89	0.60	17	1.29	0.72	2.32	0.88
Acute myeloid leukemia	T0	25	1.00	1.00	1.00	—	28	1.00	1.00	1.00	—	42	1.00	1.00	1.00	—
	T1	22	0.99	0.53	1.85	—	22	1.05	0.57	1.94	—	16	1.16	0.63	2.14	—
	T2	16	0.80	0.40	1.60	—	14	0.71	0.35	1.43	—	9	0.67	0.31	1.43	—
	T3	20	1.09	0.56	2.14	0.74	20	1.15	0.60	2.22	0.79	17	1.32	0.71	2.49	0.87
Chronic myeloid leukemia	M0	10	1.00	1.00	1.00	—	12	1.00	1.00	1.00	—	16	1.00	1.00	1.00	—
	M1	9	0.88	0.33	2.36	—	10	0.89	0.35	2.25	—	6	0.92	0.34	2.51	—
	M2	8	0.90	0.31	2.61	0.98	6	0.63	0.21	1.88	0.71	6	1.04	0.37	2.91	0.65

Note: —, no data; AHS, Agricultural Health Study; CI, confidence interval; CLL, chronic lymphocytic leukemia; RR, rate ratio; SLL, small lymphocytic lymphoma.

^a Main models were all adjusted for attained age, logarithmic age, year of enrollment, biological sex, race, state of residence, pack-years of cigarette smoking, attained education, family history of cancer, and correlated pesticides (2,4-D, alachlor, cyanazine, metolachlor, and glyphosate). Reproductive-related cancer sites were restricted to biological sex, and sex covariate was not included.

^b Quantile cut points based on intensity-weighted days among any cancer cases for unlagged exposures. Q0: 0 (no reported use); Q1: 2–918; Q2: 919–2,634; Q3: 2,635–7,370; Q4: $\ge \mathrm{7,371}$. Tertiles: T0: 0; T1: 2–1,343; T2: 1,344–5,207; T3: $\ge \mathrm{5,208}$. Median: M0: 0; M1: 2–2,634; M2: $\ge \mathrm{2,635}$. Cut points for lagged exposures are specific to exposed cases to defined lag years.

^c Person-years for nonexposed (Q0, T0, or M0) and exposed applicators (Q1–Q4, T1–T3, or M1–M2), respectively, included the following: 275,521 and 725,356 (unlagged); 288,375 and 723,076 (10-year lag); and 384,018 and 627,432 (25-year lag).

^d Lag 10 and Lag 25 represent 10 and 25 lagged years of atrazine use subtracted from date of the last follow-up, respectively.

^e Linear trends based on the Wald test by using the median of each exposure category as a continuous variable.

^f Aggressive prostate cancer outcome based on one of more of the following: Gleason score $8+$, stage 3 diagnosis, or death from prostate cancer.

In age-stratified analyses (Tables 4 and 5; Excel Table S3), participants 70 and over exhibited a non-statistically significant positive association with lung cancer at the highest quartile (Q4) of atrazine use compared to Q0 (${\text{RR}}_{\mathrm{Q}4}=1.29$; 95% CI: 0.94, 1.78; $p\text{-trend}=0.17$). In a sensitivity analysis including endotoxin exposure at enrollment (Excel Table S4), we observed no noticeable changes after adjusting for endotoxin exposure within unlagged models except for ever-use results.

Table 4.

Age-stratified ($<60$, 60–69, and $\ge 70$ years of age) and adjusted RRs and 95% CIs from Poisson regressions for cancer incidence associated with cumulative atrazine intensity-weighted days of exposure compared to reported no atrazine use among agricultural health study cohort applicators from 1993 to 2017 ($n=\mathrm{53,562}$ total participants and 1,011,451 total person-years).

Cancer sitesa,b,c	Pesticide used	$<60$					60–69					$\ge 70$
		cases	RR	95% CI		$p$-trende	cases	RR	95% CI		$p$-trende	cases	RR	95% CI		$p$-trende	$p$-interactionf
Esophagus	Never	12	1.00	1.00	1.00	—	7	1.00	1.00	1.00	—	8	1.00	1.00	1.00	—	—
	Ever	25	0.94	0.41	2.15	—	28	0.81	0.30	2.14	—	45	2.85	1.25	6.51	—	—
	T1	7	0.96	0.36	2.54	—	8	0.82	0.27	2.50	—	15	2.94	1.19	7.26	—	—
	T3	6	0.76	0.26	2.26	—	11	0.96	0.32	2.87	—	19	3.36	1.34	8.40	—	—
	T3	12	1.21	0.44	3.30	0.93	9	0.68	0.21	2.16	0.73	9	1.79	0.64	5.05	0.02	0.20
Lung	Never	46	1.00	1.00	1.00	—	87	1.00	1.00	1.00	—	134	1.00	1.00	1.00	—	—
	Ever	91	1.26	0.81	1.96	—	202	1.24	0.92	1.66	—	288	1.22	0.97	1.54	—	—
	Q1	24	1.46	0.86	2.50	—	52	1.18	0.80	1.75	—	70	1.15	0.85	1.57	—	—
	Q2	19	1.02	0.54	1.90	—	50	1.17	0.78	1.76	—	64	1.16	0.84	1.61	—	—
	Q3	25	1.47	0.84	2.59	—	51	1.18	0.79	1.78	—	66	1.16	0.84	1.60	—	—
	Q4	20	0.88	0.46	1.66	0.71	42	0.85	0.55	1.33	0.89	71	1.29	0.94	1.78	0.17	0.85
Aggressive prostateg	Never	17	1.00	1.00	1.00	—	76	1.00	1.00	1.00	—	103	1.00	1.00	1.00	—	—
	Ever	94	2.00	1.18	3.39	—	204	0.87	0.64	1.18	—	266	1.01	0.78	1.30	—	—
	Q1	21	1.41	0.72	2.75	—	40	0.79	0.52	1.20	—	75	1.08	0.78	1.49	—	—
	Q2	23	2.10	1.10	3.98	—	54	0.90	0.60	1.35	—	62	0.96	0.69	1.35	—	—
	Q3	19	1.86	0.96	3.63	—	49	0.78	0.51	1.19	—	58	0.92	0.65	1.31	—	—
	Q4	30	3.04	1.61	5.75	0.001	55	0.95	0.63	1.44	0.90	66	1.13	0.80	1.59	0.98	0.04

Note: —, no data; AHS, Agricultural Health Study; CI, confidence interval; RR, rate ratio.

^a Main models were all adjusted for year of enrollment, biological sex, race, state of residence, pack-years of cigarette smoking, attained education, family history of cancer, and correlated pesticides (2,4-D, alachlor, cyanazine, metolachlor, and glyphosate). Reproductive-related cancer sites were restricted to biological sex, and sex covariate was not included.

^b Some age-stratified sites with $<20$ exposed cases were collapsed into sequential age categories or completely removed due to data spareness.

^c Person-years for nonexposed (Never) and exposed applicators (Q1–Q4, T1–T3, or M1–M2), respectively, included the following age stratification: 109,268 and 237,706 ($<60$ years-old); 77,074 and 247,043 (60–69 years old); and 89,179 and 240,607 ($\ge 70$ years old).

^d Never cases are defined as participants that have not reported atrazine use at enrollment or at follow-up. Ever cases are defined as participants that have reported atrazine use at enrollment or at follow-up. Quantile cut points based on intensity-weighted days among any cancer cases for unlagged and lagged exposures. Quartiles Q0: 0 (no reported use); Q1: 2–918; Q2: 919–2,634; Q3: 2,635–7,370; Q4: $\ge \mathrm{7,371}$.Tertiles T0: 0; T1: 2–1,343; T2: 1,344–5,207; T3: $\ge \mathrm{5,208}$. Median M0: 0; M1: 2–2,634; M2: $\ge \mathrm{2,635}$.

^e Linear trends based on the Wald test by using the median of each exposure category as a continuous variable.

^f Tests for interaction between categorical atrazine exposures and specified age categories using Type III Wald test.

^g Aggressive prostate cancer outcome based on one of more of the following: Gleason score $8+$, stage 3 diagnosis, or death from prostate cancer.

Table 5.

Age-stratified ($<50$, 50–59, 60–69, and $\ge 70$ years of age) and adjusted RRs and 95% CIs from Poisson regressions for cancer incidence associated with cumulative atrazine intensity-weighted days of exposure compared to reported no atrazine use among agricultural health study cohort applicators from 1993 to 2017 ($n=\mathrm{53,562}$ total participants and 1,011,451 total person-years).

Cancer sitesa,b,c	$<50$						50–59					60–69					$\ge 70$
	Pesticide used	cases	RR	95% CI		$p$-trende	cases	RR	95% CI		$p$-trende	cases	RR	95% CI		$p$-trende	cases	RR	95% CI		$p$-trende	$p$-interactionf
All	Never	165	1.00	1.00	1.00	—	446	1.00	1.00	1.00	—	836	1.00	1.00	1.00	—	1061	1.00	1.00	1.00	—	—
	Ever	336	0.98	0.78	1.22	—	1,285	0.97	0.85	1.1	—	2,318	0.99	0.9	1.08	—	2,595	1.06	0.98	1.15	—	—
	Q1	87	0.86	0.65	1.15	—	321	0.93	0.80	1.09	—	545	0.98	0.86	1.1	—	638	1.03	0.93	1.15	—	—
	Q2	78	0.93	0.69	1.25	—	305	0.97	0.83	1.14	—	567	1.00	0.88	1.13	—	660	1.09	0.98	1.21	—	—
	Q3	78	0.92	0.67	1.25	—	319	0.99	0.84	1.16	—	574	0.98	0.86	1.11	—	634	1.06	0.96	1.19	—	—
	Q4	87	1.27	0.93	1.72	0.49	327	1.03	0.87	1.22	0.70	601	1.02	0.9	1.16	0.54	586	1.05	0.94	1.17	0.19	0.53
Non-Hodgkin lymphoid malignancies	Never	8	1.00	1.00	1.00	—	34	1.00	1.00	1.00	—	51	1.00	1.00	1.00	—	99	1.00	1.00	1.00	—	—
	Ever	45	2.43	1.10	5.38	—	116	0.74	0.48	1.13	—	167	1.00	0.70	1.43	—	229	0.94	0.71	1.24	—	—
	Q1	11	2.21	0.88	5.56	—	27	0.67	0.39	1.16	—	33	0.86	0.53	1.41	—	53	0.91	0.64	1.3	—	—
	Q2	14	2.84	1.12	7.21	—	28	0.76	0.44	1.3	—	40	1.05	0.66	1.68	—	57	0.92	0.64	1.31	—	—
	Q3	10	2.73	1.03	7.2	—	28	0.76	0.44	1.32	—	47	1.02	0.63	1.64	—	66	1.08	0.76	1.53	—	—
	Q4	9	2.53	0.88	7.27	0.03	31	0.84	0.49	1.46	0.42	45	1.11	0.69	1.78	0.53	46	0.82	0.55	1.21	0.69	0.60
Non-Hodgkin lymphoma, B-Cell	Never	8	1.00	1.00	1.00	—	30	1.00	1.00	1.00	—	48	1.00	1.00	1.00	—	95	1.00	1.00	1.00	—	—
	Ever	38	2.04	0.89	4.65	—	106	0.78	0.49	1.24	—	158	0.98	0.68	1.42	—	212	0.92	0.69	1.22	—	—
	Q1	10	2.04	0.79	5.3	—	25	0.71	0.40	1.27	—	32	0.88	0.53	1.45	—	49	0.89	0.61	1.28	—	—
	Q2	11	2.14	0.79	5.85	—	26	0.83	0.47	1.48	—	38	1.08	0.67	1.74	—	54	0.91	0.63	1.32	—	—
	Q3	8	2.21	0.79	6.18	—	24	0.75	0.42	1.35	—	44	0.99	0.61	1.62	—	59	1.03	0.71	1.48	—	—
	Q4	8	2.19	0.72	6.63	0.05	30	0.93	0.52	1.66	0.62	42	1.09	0.67	1.79	0.59	43	0.8	0.54	1.20	0.49	0.78
Mature B-cell lymphoma	Never	8	1.00	1.00	1.00	—	29	1.00	1.00	1.00	—	47	1.00	1.00	1.00	—	91	1.00	1.00	1.00	—	—
	Ever	35	1.79	0.78	4.09	—	100	0.77	0.48	1.23	—	155	0.99	0.68	1.43	—	203	0.91	0.68	1.21	—	—
	Q1	9	1.76	0.67	4.66	—	22	0.64	0.35	1.18	—	32	0.9	0.54	1.49	—	49	0.9	0.62	1.30	—	—
	Q2	9	1.61	0.56	4.63	—	25	0.85	0.47	1.52	—	38	1.1	0.68	1.79	—	51	0.9	0.62	1.32	—	—
	Q3	8	2.15	0.77	6.01	—	24	0.79	0.44	1.44	—	42	0.96	0.59	1.59	—	57	1.02	0.71	1.49	—	—
	Q4	8	2.20	0.73	6.65	0.04	28	0.92	0.51	1.67	0.84	41	1.09	0.66	1.8	0.62	39	0.75	0.49	1.13	0.35	0.84

Note: —, no data; AHS, Agricultural Health Study; CI, confidence interval; RR, rate ratio.

^a Main models were all adjusted for year of enrollment, biological sex, race, state of residence, pack-years of cigarette smoking, attained education, family history of cancer, and correlated pesticides (2, 4-D, alachlor, cyanazine, metolachlor, and glyphosate). Reproductive-related cancer sites were restricted to biological sex, and sex covariate was not included.

^b Some age-stratified sites with $<20$ exposed cases were collapsed into sequential age categories or completely removed due to data spareness.

^c Person-years for nonexposed (Never) and exposed applicators (Q1–Q4, T1–T3, or M1–M2), respectively, included the following age stratification: 39,862 and 51,495 ($<50$ years old); 69,406 and 186,211 (50–59 years old); 77,074 and 247,043 (60–69 years old); and 89,179 and 240,607 ($\ge 70$ years old).

^d Never cases are defined as participants that have not reported atrazine use at enrollment or at follow-up. Ever cases are defined as participants that have reported atrazine use at enrollment or at follow-up. Quantile cut points based on intensity-weighted days among any cancer cases for unlagged and lagged exposures. Quartiles Q0: 0 (no reported use); Q1: 2–918; Q2: 919–2,634; Q3: 2,635–7,370; Q4: $\ge \mathrm{7,371}$. Tertiles T0: 0; T1: 2–1,343; T2: 1,344–5,207; T3: $\ge \mathrm{5,208}$. Median M0: 0; M1: 2–2,634; M2: $\ge \mathrm{2,635}$.

^e Linear trends based on the Wald test by using the median of each exposure category as a continuous variable.

^f Tests for interaction between categorical atrazine exposures and specified age categories using Type III Wald test.

Soft tissue sarcoma was positively associated with highest use (${\text{RR}}_{\mathrm{T}3}=2.54$; 95% CI: 0.97, 6.52; $p\text{-trend}=0.31$) of atrazine among 35 cases (Table 3). Lagged results maintained a pattern of elevated risks across tertiles showing statistically significant associations with Lag 15 (${\text{RR}}_{\mathrm{T}3}=2.69$; 95% CI: 1.08, 6.73) and Lag 20 (${\text{RR}}_{\mathrm{T}3}=2.74$; 95% CI: 1.09, 6.93) exposures. A significant exposure–response trend was observed at Lag 20 ($p\text{-trend}=0.03$) (Excel Table S2). Significant associations at the highest tertile did not extend to lagged exposures at 25-years (${\text{RR}}_{\mathrm{T}3}=1.89$; 95% CI: 0.80, 4.44).

There was no overall association for ever use with all prostate ($\text{RR}=0.99$; 95% CI: 0.91, 1.08) or aggressive prostate cancers ($\text{RR}=1.03$; 95% CI: 0.86, 1.24) (Table 2). Unlagged intensity-weighted days of atrazine use in the highest quartile were positively associated with all prostate (${\text{RR}}_{\mathrm{Q}4}=1.06$; 95% CI: 0.95, 1.19; $p\text{-trend}=0.49$) and aggressive prostate cancers (${\text{RR}}_{\mathrm{Q}4}=1.17$; 95% CI: 0.93, 1.49; $p\text{-trend}=0.13$), although not statistically significant (Table 3). In a sensitivity analysis, we modified our definition of aggressive prostate cancer to include Gleason scores $\ge 7$ and observed a significant positive association at the highest quartile (${\text{RR}}_{\mathrm{Q}4}=1.19$; 95% CI: 1.03, 1.38; $n=398$ cases) with significant linear trend ($p\text{-trend}=0.007$) (Excel Table S4).

We examined whether the effect of atrazine on aggressive prostate cancer varied by age. Due to a small number of cases, we recategorized $<50$ and 50–59 age groups to $<60$ years and conducted a test for interaction by age. This test for interaction was statistically significant ($p=0.037$) (Table 4), suggesting differential effects of exposure by age. There was a strongly positive effect of exposure on aggressive prostate cancer among participants below 60 years old (${\text{RR}}_{\mathrm{Q}4}=3.04$; 95% CI: 1.61, 5.75; $n=30$ cases; $p\text{-trend}<0.001$), while there was no effect for participants 60 years of age and above. We conducted a sensitivity analysis by using 55 years old at diagnosis as a clinically meaningful age for early onset aggressive prostate cancer and found significant results (test of age by exposure interaction, $p<0.0001$; ${\text{RR}}_{\mathrm{Q}4}=4.05$; 95% CI: 1.09, 15.09) with no effect for individuals older than 55 years of age (Excel Table S5). Using the 55-year cutoff for early onset prostate cancer,³⁵ we found a significant interaction between exposure and age ($p=0.007$). There was a significant effect of exposure on all prostate cancer incidences among applicators below the age of 55 years (${\text{RR}}_{\mathrm{Q}4}=1.77$; 95% CI: 1.14, 2.75; $n=51$ cases; $p\text{-trend}=0.01$) with no apparent effect for older ($\ge 55$) individuals (Excel Table S5).

Other notable findings from the 25-year lagged analyses (Table 3) include significant positive associations with cancer of the oral cavity (${\text{RR}}_{\mathrm{Q}4}=1.54$; 95% CI: 1.02, 2.32; $p\text{-trend}=0.32$; $n=38$ cases), which was partially driven by the association with cancer of the pharynx (${\text{RR}}_{\mathrm{T}3}=3.04$; 95% CI: 1.45, 6.36; $p\text{-trend}=0.07$; $n=18$ cases). We also observed significant increased RR of kidney cancer and linear trend (${\text{RR}}_{\mathrm{Q}4}=1.62$; 95% CI: 1.15, 2.29; $p\text{-trend}<0.005$). In the 20-year lagged analysis (Excel Table S2), pharyngeal (${\text{RR}}_{\mathrm{T}3}=2.15$; 95% CI: 0.99, 4.63; $p\text{-trend}=0.21$) and oral cavity (${\text{RR}}_{\mathrm{Q}4}=1.49$; 95% CI: 0.99, 2.24; $p$-trend 0.29) cancer risks remained elevated at the highest tertile and quartile, respectively. For kidney cancer, the RR in the highest quartile was 1.34 (95% CI: 0.94, 1.90; $p=0.06$).

We did not see an association between atrazine use and esophageal cancers in our non-stratified unlagged analysis. However, among applicators 70 years of age and older, we observed positive associations between ever-use and esophageal cancers (${\text{RR}}_{\text{Ever}}=2.85$; 95% CI: 1.25, 6.51). This cancer site also exhibited significant positive associations in the first and second tertiles and a significant linear trend ($p=0.02$), though, evidence of interaction with age was not observed ($p\text{-interaction}=0.20$).

Among lymphohematopoietic cancers, we saw no overall association of atrazine with non-Hodgkin lymphoma (NHL). However, among applicators diagnosed at $\le 50$ years of age, the risk of NHL was higher with ever-use of atrazine (${\text{RR}}_{\text{Ever}}=2.43$; 95% CI: 1.10, 5.38; $n=45$ cases) (Table 5). In the corresponding exposure–response analyses for applicators below 50 years of age, significant (${\text{RR}}_{\mathrm{Q}2}=2.84$; 95% CI: 1.12, 7.21; ${\text{RR}}_{\mathrm{Q}3}=2.73$; 95% CI: 1.03, 7.20) and nonsignificant (${\text{RR}}_{\mathrm{Q}1}=2.21$; 95% CI: 0.88, 5.56; ${\text{RR}}_{\mathrm{Q}4}=2.53$; 95% CI: 0.88, 7.27) increased NHL risks were observed with a $p$-trend of 0.03. Similar elevations were not observed in the older age categories, although there was no statistically significant interaction between age and atrazine use ($p\text{-interaction}=0.60$). Mature B-cell lymphoma cases diagnosed younger than 50 years of age also exhibited nonsignificant positive associations across quartiles ($p\text{-trend}=0.04$) (Table 5). Among other NHL subtypes, we observed a significant inverse association with follicular lymphoma for ever-atrazine use ($\text{RR}=0.58$; 95% CI: 0.35, 0.97) (Table 3). Among chronic lymphocytic leukemia and small lymphocytic leukemia (CLL/SLL) cases, we detected significant inverse associations at the highest quartile among higher lags with a significant decreasing trend ($p\text{-trend}=0.04$) at Lag 25.

Discussion

This updated evaluation reports on associations between atrazine use and risk of cancer at multiple sites across varying unlagged and lagged exposures as well as ages of diagnosis. We found a statistically significant association between ever use of atrazine and lung cancer compared to unexposed participants. In addition, we observed a pattern of elevated risks of soft tissue sarcoma. Increased risks were also seen for pharyngeal cancer at the highest use for 25-year lag exposures. A positive exposure–response association for risk of kidney cancer was consistent with previous analyses. We also note an inverse relationship with CLL/SLL particularly among 25-year lagged exposures. There was some evidence of risks varying by age, including increased risks for those diagnosed with aggressive prostate cancer and NHL at younger ages and esophageal cancer at older ages.

Accumulating evidence from animal model studies has provided suggestive linkages between atrazine exposure and potential carcinogenic mechanisms of action that may act at multiple cancer sites. In experimental studies, atrazine can cause genomic damage and oxidative stress.^36–38 Oxidative stress and chronic inflammation are some of the key characteristics of carcinogens.³⁹ A recent molecular epidemiologic study that examined the short-term effects of atrazine among humans found no associations between atrazine mercapturate (an atrazine metabolite) and several oxidative stress biomarkers in overall analyses,⁴⁰ though a statistically positive association with 8-hydroxy-2′-deoxyguanosine (8-OhdG) was observed in analyses restricted to individuals with measures of atrazine mercapturate above the detection limit. A recent toxicology study has characterized mechanistic pathways that induce lung inflammation caused by aerosolized atrazine exposures.⁴¹ Researchers found that changes in the Nrf-2 pathways and Beclin 1/Lc3 expression can increase oxidative damage and inflammation, eventually altering lung tissue. No association between atrazine and lung cancer was reported in a previous exposure–response analysis ($n=358$ cases).¹⁷ Results from this analysis suggest a higher risk of lung cancer related to ever-use of atrazine and positive exposure–response associations but lacks significant evidence of increasing RRs with exposure.

Decreased lung cancer risk has been observed in many agricultural settings, partially attributed to endotoxin exposure.³³ In the AHS, there is a diminished overall risk compared to the general population but more pronounced in those exposed to endotoxin.³⁴ In a previous evaluation within the AHS,⁴² we observed that endotoxin exposure modifies the effect of diesel exhaust, a known lung carcinogen. In this work, we evaluated potential endotoxin-related confounding by conducting sensitivity analyses to determine whether it may have similar attenuating effects on atrazine exposure. Unlike our evaluation of diesel exhaust, we did not observe any evidence of attenuation. One explanation is that atrazine-related carcinogenesis in lung cancer may involve mechanistic pathways not dependent on chronic inflammation, the mechanism by which endotoxin is thought to influence lung cancer risk.⁴³

The association with prostate cancer observed in this study is supported by evidence of the role of endocrine disruption in contributing to prostate cancer risk.^44,45 A 2021 meta-analysis of experimental studies focused on male reproductive system toxicity found atrazine exposure was associated with decreased testosterone production, reduced weight in multiple male reproductive organs, reduced sperm motility, and sperm abnormalities specific to rodent animal models.⁴⁶ In an experimental study, authors described a plausible atrazine-activated pathway for cancer cell proliferation and malignancy using mouse prostate cell lines.⁴⁷ Our analysis found potential evidence of total and aggressive prostate cancer risk, particularly among applicators below 60 years old. Previous evaluations of atrazine and prostate cancer associations in the AHS cohort did not find an exposure–response association with follow-up through 2007.^15,48 In addition to increased follow-up time and cases, we also have used a more stringent definition of aggressive prostate cancer in this analysis than in previous studies. Results were similar with both definitions in this analysis, although the positive association at the highest quartile appeared stronger based on the more stringent definition.

Other studies evaluating atrazine-prostate cancer associations have been mixed. Atrazine and triazine (a broad chemical class of nitrogen-containing heterocycles that includes atrazine) manufacturing plant workers below 60 years of age had a nearly six-fold significantly higher incidence of prostate cancer incidence during a follow-up.^49,50 In 2019, a meta-analysis characterizing the effect of occupational risk factors on prostate cancer did not find an association (${\text{RR}}_{\text{meta}}=1.02$; 95% CI: 0.92, 1.14) with triazine herbicides.⁵¹ In a case-control study among farmers in British Columbia, the authors found an elevated but nonsignificant association between ever-atrazine users and prostate cancer [$\text{odds ratio\hspace{0.17em}}\left(\text{OR}\right)=1.51$; 95% CI: 0.64, 3.35] with nine exposed cases.⁵² Despite the lack of consistency in the epidemiological literature, experimental studies have consistently demonstrated atrazine’s harmful effects on the endocrine and reproductive systems.^3,4,47,53 Also, it is important to note the increased amount of follow-up information and the number of cases that have enhanced the statistical power to detect modest effects in this updated analysis, including power for age-stratification analyses and interaction tests. From this work, we have evidence age may modify the relationship between atrazine use and incidence of total and aggressive prostate cancers. Our findings highlight the potential vulnerability among younger atrazine applicators as a high-risk group.

A recent AHS study with a follow-up time period from 1993 to 2015 reported a significant exposure–response association between atrazine use and risk of renal cell carcinoma (RCC) in 20-year lagged analyses (${\text{RR}}_{\mathrm{Q}4}=1.43$; 95% $\text{CI}=1.00–2.03$, $p\text{-trend}=0.02$).¹⁹ We observed a similar association with kidney cancer in analyses lagged by 25 years (${\text{RR}}_{\mathrm{Q}4}=1.62$; 95% CI: 1.15, 2.29; $p\text{-trend}<0.005$). And, more specifically, RCC cases ($n=332$) make up nearly all total kidney cancers in this study (${\text{RR}}_{\mathrm{Q}4}=1.63$; 95% CI: 1.15, 2.30; $p\text{-trend}=0.006$).

We noted increased RRs for soft tissue sarcoma and pharyngeal cancer at the highest quartiles when considering lagged exposures. In the 2011 AHS atrazine-cancer study, increased oral cancer risk was not observed based on 74 exposed cases when considering cumulative intensity-weighted days.¹⁵

We found a potential inverse association between atrazine use and laryngeal cancer, which remained after adjustment for known risk factors for laryngeal cancer, such as smoking and alcohol use. A biologic rationale for this observation is not currently known. However, similar decreased risks were observed in the previous atrazine analysis in 2011¹⁵ and with half the number of exposed cases.

This study found significant positive associations for total NHL and a significant exposure–response trend for NHL B-cell subtype among applicators below 50 years of age. Earlier non-AHS studies observed significant positive associations between ever atrazine use and incidence of NHL^54,55 and t(14:18)-positive NHL subtypes.⁵⁶ However, applicators below 50 years of age included few exposed cases. We found no associations with other age groups and other hematological outcomes, including leukemia and multiple myeloma. To our knowledge, no other studies have investigated atrazine-NHL effects stratified by age. No other positive associations were found in unlagged and lagged exposure–response analysis within the AHS cohort across selected lymphohematopoietic malignancies apart from the significant inverse associations seen with follicular lymphoma. We did find a potential inverse association between atrazine and CLL/SLL cases. Similar to laryngeal cancer, there was no known biologic rationale for this relationship, and there was no detected association in the 2011 study. However, CLL/SLL was not grouped with mantle cell lymphoma (MCL)—a rare B-cell NHL subtype—in this study.

Atrazine use and thyroid cancer in the AHS was last evaluated in 2021, where an overall null relationship was observed.¹⁶ The study period included a follow-up to 2014 for North Carolina and 2015 for Iowa with 60 exposed thyroid cancer cases. Based on two additional years of follow-up for Iowa and four additional exposed cases, we also observed no association using the nonexposed (Table 3) or the lowest exposed group as a referent (Excel Table S1) in this analysis.

This study is the largest and most comprehensive epidemiological evaluation of atrazine use and cancer risk to our knowledge. The study design involved a long-term prospective cohort with high-quality and self-reported lifetime pesticide use information before cancer diagnosis. Through updated linkages to respective state cancer registries, our analysis included over twice as many cases of atrazine users as the previous evaluation, allowing for evaluation of additional cancer types and age-specific risks. Despite using self-reported pesticide data, past AHS studies have demonstrated validity and reliability in self-reporting exposures from self-administered questionnaires and interviews. For example, researchers found that AHS participants largely reported a valid self-reported duration of pesticide use by cross-checking with information on pesticide active ingredient registrations.⁵⁷ Also, earlier work demonstrated strong agreement on pesticide use with self-completed questionnaires administered 1-year apart, indicating that the pesticide applicators were reliable in their reporting.⁵⁸ This updated analysis had some limitations. We evaluated the potential correlation between atrazine and other pesticides although the correlation between any individual pesticide and atrazine was low ($r=0.35–0.45$). We note most of this cohort comprises white (98%) licensed applicators; however, the cohort likely reflects the 82% of those seeking pesticide licenses enrolled in the cohort. We also did not have information on potential nonoccupational atrazine exposures, such as drinking water consumption, although the evidence for exposure via this route is not well characterized.⁵⁹ While misclassification of exposure is likely, any bias related to this misclassification is likely to be nondifferential given the prospective nature of the study. There is a possibility that significant findings among analyses with smaller number of exposed cases may be driven by chance.

In conclusion, this updated analysis provides evidence of potential associations between occupational atrazine use and cancer risk. We observed potential risk of cancers of the prostate and lung among AHS applicators. Lagged exposures exhibited associations with soft tissue sarcoma and pharyngeal and kidney cancers. Our investigation also found potential differential age-specific cancer risks, particularly for aggressive prostate and NHL. Further work in elucidating biologic plausibility and mechanisms of these findings are needed to investigate the reasons for these observations.

Conclusions and opinions are those of the individual authors and do not necessarily reflect the policies or views of EHP Publishing or the National Institute of Environmental Health Sciences.