Exposure to Agent Orange and Risk of Bladder Cancer Among US Veterans

Key Points Question Is there an association between Agent Orange exposure and risk of bladder cancer and/or aggressiveness of bladder cancer? Findings In this cohort study of more than 2.5 million male Vietnam veterans, there was a modestly increased risk of bladder cancer—but not aggressiveness of bladder cancer—among veterans exposed to Agent Orange. Exposure to Agent Orange was associated with a significantly increased risk of bladder cancer, although the association was very slight (hazard ratio, 1.04; 95% CI, 1.02-1.06). Meaning These findings suggest an association between exposure to Agent Orange and bladder cancer, although the clinical relevance of this was unclear.


Introduction
There will be an estimated 81 180 new cases of bladder cancer and 17 100 deaths from bladder cancer in the US in 2022. 1 Although most patients with bladder cancer receive a diagnosis of non-muscleinvasive disease (clinical stage T1 or less), 2 approximately 20% to 40% of patients either present with or later develop muscle-invasive disease. 3though smoking is one of the strongest bladder cancer risk factors, other risk factors include older age, male sex, and exposure to carcinogens, such as aromatic amines, inorganic arsenic, and polycyclic aromatic hydrocarbons. 4,5However, many bladder cancer cases cannot be explained by the known risk factors, leading to a need to identify new risk factors.One factor recently implicated is Agent Orange.][8][9][10][11][12][13] The Institute of Medicine concluded that the association between Agent Orange exposure and bladder cancer outcomes is an area of needed research. 6Based on this and the aforementioned studies, the US Department of Veterans Affairs (VA) recently designated bladder cancer as a cancer caused by Agent Orange exposure. 6,14To provide a true measure of aggressive bladder cancer, it is crucial to study not just the diagnosis of bladder cancer but the aggressiveness of disease when diagnosed.However, to our knowledge, such a comprehensive analysis of Agent Orange and bladder cancer has not been done to date, mainly due to the small numbers of exposed cases, the lack of ability to identify Agent Orange exposure, and the lack of an ability to control for known confounders, particularly smoking, a major bladder cancer risk factor.To address these issues, we examined bladder cancer according to Agent Orange exposure using data from the largest integrated health system in the country, the VA health system.

Data Source
Using inpatient and outpatient data and fee-based or community care claims (ie, care provided outside the VA system for which the VA paid), we queried data from the VA Informatics and Computing Infrastructure (VINCI) to identify all veterans seen at any VA health system site from January 1, 2001, to December 31, 2019.Key to this proposal is the fact that all data from all VA sites are integrated into a common electronic medical record collated in VINCI.This study was approved by the institutional review boards at the Durham VA Medical Center, The University of Texas Medical Branch, and the US Army Medical Research and Development Command Human Research Protection Office.This study was conducted using a waiver of informed consent because the study met the following criteria: (1) the use or disclosure of the protected health information involves no more than a minimal risk to the privacy of individuals, (2) the research could not practicably be conducted without the waiver or alteration, and (3) the research could not practicably be conducted without access to and use of the protected health information.This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. 15

Study Cohort
A total of 25 144 713 veterans enrolled in the VA nationwide from 2001 to 2019 were identified.This cohort study included 3 520 249 veterans who met the following criteria: active users of the VA system (defined as Ն2 medical office visits within 5 years); served in the Vietnam conflict; served in the military branch of Air Force, Army, Coast Guard, Navy, or Marine Corps; not transgender; 18 to 75 years of age at enlistment; and not deceased prior to January 1, 2001.To create a more homogeneous cohort that had opportunity for Agent Orange exposure, we then limited the study to men (as very few female veterans among those enrolled in the VA nationwide were exposed to Agent Orange) and excluded veterans missing pertinent data for matching, including race and ethnicity, date of birth, and military branch.After inclusions and exclusions were applied, the cohort consisted of 868 912

JAMA Network Open | Urology
Exposure to Agent Orange and Risk of Bladder Cancer Among US Veterans

Exposure Data
The group exposed to Agent Orange consisted of veterans with documented exposure to Agent Orange-containing herbicide agents during active military duty.VINCI has an Agent Orange variable that designates exposure status as yes or no.Submitted documents fulfilling these requirements are then verified by a select VA committee. 16 b Included based on the following criteria: active users of the VA system; served in the Vietnam conflict; served in the military branch of Air Force, Army, Coast Guard, Navy, or Marine Corps; not transgender; age at enlistment 18 to 75 years; and not deceased prior to 2001.c Men aged 18 years or older at service entry who were not missing data on race and ethnicity were eligible for matching.
d Matched on race and ethnicity, age groups at service entry, military branch, and year of service entry.For year of service entry, 86% of matched controls were matched on the exact year, and 14% were matched in subsequent steps using loosened criteria (±1-4 years).
e Patients were excluded if they were missing follow-up data or had a VA entry date the same as the most recent follow-up date.
linking each patient's zip code of primary residence to 2014 American Community Survey 17 estimates of the poverty rate and was categorized into quartiles.Comorbidities were assessed from claims data using the Klabunde modification of the Charlson Comorbidity Index (CCI) summed to the year after VA enrollment. 18

Natural Language Processing
Given the large cohort and the inability to capture disease stage information for all veterans with bladder cancer, along with the fact that the VA tumor registry lacks disease stage information for approximately two-thirds of veterans, 19 we applied a natural language processing model that was previously developed, based on pathology reports to determine muscle invasion (yes or no) at the time of diagnosis, which had 94% accuracy in a validation set to separate veterans into groups according to whether they had muscle invasion (yes or no). 20

Outcomes
Bladder cancer incidence was defined as the total number of cases within the study period divided by the summed person-years and was multiplied by 1000 to get incidence per 1000 person-years.The primary outcome of interest was time to bladder cancer diagnosis.The secondary outcome was depth of invasion (muscle-invasive vs non-muscle-invasive bladder cancer) at the time of diagnosis among those with a diagnosis of bladder cancer.

Statistical Analysis
Statistical analysis was performed from December 14, 2021, to May 3, 2023.Differences in demographic features by Agent Orange exposure were examined using the t test for normally distributed continuous variables, the Wilcoxon rank sum test for nonnormally distributed continuous variables, and the χ 2 test for categorical variables.With a very large sample size, we had very high power to detect small differences as statistically significant.As such, standardized differences (d)   were computed between Agent Orange groups, defined as the difference in mean values or proportions divided by the SE and interpreted as clinically relevant if d Ն 0.10.Univariable and multivariable Cox proportional hazards regression models were used to test whether Agent Orange exposure was associated with risk of bladder cancer.Although we matched veterans on variables such as age at time of Vietnam service entry, or when they would have been exposed to Agent Orange, most data queried from VINCI (such as diagnoses and laboratory test results) did not become reliably available until the year 2000. 21This study thus used a landmark analysis whereby veterans were excluded if they received a diagnosis of bladder cancer, were deceased, or lost to follow-up prior to January 1, 2001.The first contact with the VA system starting in 2001 or after (VA entry) served as time zero in our models.Multivariable models were adjusted for age at VA entry, year of VA entry, race and ethnicity, BMI at VA entry, military branch, smoking status at VA entry, CCI, and socioeconomic status.Interactions were tested between Agent Orange exposure and year of service entry (tertials), smoking status (current, former, never, or unknown), age (younger than vs older than the median), BMI (<25, 25-29.9,Ն30, or unknown), and CCI (0, 1, 2, or Ն3) by including a crossproduct term in the model.If no significant interactions were found, results were adjusted for these factors.If significant interactions were found, results were stratified by the effect modifier and reported in eTable 4 and eTable 5 in Supplement 1.Among veterans with a diagnosis of bladder cancer, we used univariable and multivariable logistic regression to test whether Agent Orange was associated with muscle invasiveness at diagnosis.Multivariable models were adjusted for the same covariates listed but with baseline variables at bladder cancer diagnosis as opposed to VA entry.All P values were from 2-sided tests and results were deemed statistically significant at P < .05,and all analyses were performed using SAS Enterprise Guide, version 8.2 software (SAS Institute Inc).

JAMA Network Open | Urology
Exposure to Agent Orange and Risk of Bladder Cancer Among US Veterans

Results
Among 2 517 926 male veterans (median age at VA entry, 60.0 years [IQR, 56.0-64.0years]) who met inclusion criteria, there were 629 907 (25.0%) veterans with Agent Orange exposure and several variables were statistically different between veterans exposed to Agent Orange and unexposed veterans at baseline, the differences were small and did not meet our criterion for clinically meaningful differences.Further analyses were performed to evaluate whether interactions were present between Agent Orange exposure and age, year of service entry, BMI, smoking status, and CCI for estimating bladder cancer risk (Table 3).There was no significant interaction according to smoking, BMI, and CCI.Agent Orange exposure was associated with an increased risk of bladder cancer for veterans younger than the median age (HR, 1.07; 95% CI, 1.04-1.10),but not for veterans older than the median age (HR, 1.03; 95% CI, 0.99-1.05;P = .04for interaction).Agent Orange exposure was associated with an increased risk for bladder cancer for veterans who entered the military between the years 1969 and 1975 (HR, 1.08; 95% CI, 1.04-1.12),but not for the earlier 2 tertials (P = .04for interaction).
Given significant interactions between median age and Agent Orange exposure and between year of service entry and Agent Orange exposure, we conducted separate analyses for both age groups (eTable 4 in Supplement 1) and the year of service entry groups (eTable 5 in Supplement 1).

Discussion
In 2014, the committee to review the health effects in Vietnam veterans of exposure to herbicides (Tenth Biennial Update) upgraded Agent Orange and bladder cancer risk from "no association" to "limited evidence." 6Herein, we developed the largest nationwide cohort ever assembled to address bladder cancer risk and aggressiveness of bladder cancer according to Agent Orange exposure.
Our study revealed several important findings.First, we found that Agent Orange was associated with a 4% increased risk of bladder cancer and that this association depended on age and

JAMA Network Open | Urology
Exposure to Agent Orange and Risk of Bladder Cancer Among US Veterans year of service entry.We found that Agent Orange was associated with a 7% increased risk of bladder cancer for veterans younger than the median age at VA entry, but no significant association was seen for veterans older than the median age.Thus, younger Vietnam veterans exposed to Agent Orange with potentially more life-years to develop bladder cancer were at greatest risk.This finding is critical because bladder cancer is more commonly diagnosed in older individuals, corresponding to the age of most Vietnam veterans who may have been exposed to Agent Orange. 6Given that the VA has recently designated bladder cancer as a cancer caused by Agent Orange exposure, veterans who were exposed to Agent Orange during military service may be eligible for a variety of VA benefits, including disability compensation for diseases associated with Agent Orange exposure.These benefits may extend to dependents and survivors of exposed individuals. 6,14Targeted screening could be implemented among people exposed to Agent Orange to detect bladder cancer earlier when it is more treatable.These findings persisted after controlling for numerous confounders using data from the largest equal-access system in the US and 28 672 655 person-years of follow-up.
Second, in the present study, other risk factors for bladder cancer were also tested, including race and ethnicity and smoking.3][24][25] Consistent with this statistic, we found a 35% decreased incidence of bladder cancer among Black veterans when controlling for other important confounders, including age and smoking status.Moreover, current smokers had up to a 2-fold increased risk of bladder cancer compared with never smokers, in accordance with prior findings. 4,5,26The fact that our findings mirror well-known population data for established bladder cancer risk factors lends credibility to our results.
Third, we attempted to control for social determinants of health, including poverty and access to care.The present findings were derived from the largest equal-access health system in the US.0][11] However, those studies were limited by either small numbers of patients, inadequate follow-up, and/or lack of  Fourth, we found decreased odds of muscle-invasive disease among veterans with Agent Orange exposure.Although we cannot determine causality given the retrospective nature of our study design, this observation may be due to earlier bladder cancer detection in the group exposed to Agent Orange.Moreover, given the increasing number of Agent Orange-related diseases, routine urinalysis and subsequent workup and/or screening may be initiated and may lead to earlier detection and less-advanced bladder cancer disease.We were unable to control for a possible doseresponse relationship regarding Agent Orange exposure; prior studies have shown that an increased risk of bladder cancer aggressiveness may be associated with a higher concentration of and/or a close proximity to toxin exposure. 27,28Thus, the latency period (ie, Vietnam era) and weak dose-response relationship may be associated with our observed decreased likelihood of aggressive bladder cancer among veterans exposed to Agent Orange.Despite being in an equal-access setting, Black veterans had increased odds of muscle-invasive bladder cancer at diagnosis compared with White veterans.
These findings mirror prior studies noting that Black vs White individuals have a decreased incidence of bladder cancer but more aggressive disease when diagnosed. 4,5,26Race-based differences in bladder cancer biology have been previously described wherein Black individuals may have more advanced disease at diagnosis compared with White individuals. 29Further investigation into bladder cancer incidence according to race and ethnicity are needed to better understand the interplay between biology, social determinants, and association with bladder cancer incidence and disease severity. 30,31

Limitations
This study has some limitations, and our findings must be considered within the context of the study design.First, veterans enrolled in the present study were male, with most having no comorbidities at the time of VA entry.Due to the limited numbers of women enrolled in the VA and exposed to Agent Orange during the Vietnam conflict, we excluded women to create a more homogeneous cohort.Given that most veterans with bladder cancer are diagnosed in their sixth decade of life or older (ie, the current age of most Vietnam veterans), our findings may not be generalizable to younger veterans.Second, while we performed a match design to control for several confounders, we cannot exclude potential selection bias and misclassification bias (ie, Agent Orange exposure due to lack of Agent Orange dosing or lack of exposure information, or CCI classification at VA entry) when using large population-based data.We used prior accepted methods and Agent Orange designation as defined by the VA. 27Moreover, we used year of service entry and year of VA entry variables to control for latency between Agent Orange exposure and bladder cancer diagnosis.We understand the inherent limitations in that there may have been a cohort of men with moreaggressive bladder cancer who died prior to 2001, thus confounding the study outcome.A further biological assessment of Agent Orange exposure and bladder cancer risk may support our findings. 32,33Third, while we attempted to control for the inherent selection bias and other confounders in this retrospective study, we acknowledge the limitations in using such a study design, including but not limited to unmeasured confounding.

Conclusions
In a large population-based cohort, we found a slight increased risk of bladder cancer among veterans exposed to Agent Orange.In contrast, we found that Agent Orange exposure was associated with a decreased risk of muscle-invasive bladder cancer.These results support prior investigations and further support bladder cancer to be designated as an Agent Orangeassociated disease.

JAMA Network Open | Urology
Exposure to Agent Orange and Risk of Bladder Cancer Among US Veterans to Agent Orange and 2 427 677 controls.Next, we matched each veteran exposed to Agent Orange to 3 controls on factors associated with Agent Orange exposure using a stepwise algorithm, broadening the matching criteria at each stage.In the first step, we matched 67% of the controls to veterans exposed to Agent Orange on age at service entry (18-20, 21-22, or Ն23 years), military branch (Air Force, Army, Coast Guard, Navy, or Marine Corps), race and ethnicity (Black, White, or other [American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and biracial]), and exact year of service entry.Subsequently, we matched 11% more controls to veterans exposed to Agent Orange following the same criteria but loosening year of service entry by ±1 year JAMA Network Open.2023;6(6):e2320593.doi:10.1001/jamanetworkopen.2023.20593(Reprinted) June 27, 2023 2/13 Downloaded From: https://jamanetwork.com/ on 09/28/2023 veterans exposed

Table 1 .
Male Patient Characteristics by Agent Orange Exposure a The standardized difference or effect size.b Matching variables.c Included American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and biracial individuals.
During a follow-up of 28 672 655 person-years, there was a total of 50 781 bladder cancer diagnoses (2.0%) (2.1% vs 2.0% among those exposed to Agent Orange vs those unexposed) recorded, for an overall bladder cancer incidence of 1.77 cases per 1000 person-years (eTable 3 in HR, 2.21; 95% CI, 2.16-2.26).Veterans in the highest socioeconomic quartile and least likely to live in poverty were most likely to receive a diagnosis of bladder cancer (HR, 1.07; 95% CI, 1.05-1.10).

Table 4 .
There was a decreased odds of diagnosis of muscle-invasive bladder cancer among those with Agent Orange exposure (odds ratio [OR], 0.91; 95% CI, 0.85-0.98).We observed increased odds of muscle-invasive bladder cancer among Black vs White veterans (OR, 1.16; 95% CI, 1.05-1.28).Veterans in the highest socioeconomic quartile and least likely to live in poverty had the lowest odds of receiving a diagnosis of muscleinvasive bladder cancer (OR, 0.88; 95% CI, 0.80-0.96).Current smokers (OR, 1.10; 95% CI, 1.00-1.21)and former smokers (OR, 1.08; 95% CI, 1.00-1.18)had higher odds of receiving a diagnosis of muscle-invasive bladder cancer compared with veterans who never smoked.Age at bladder cancer diagnosis, year of bladder cancer diagnosis, and branch of service were not associated with diagnosis of muscle-invasive bladder cancer.No significant interactions were observed between Agent Orange exposure and age, BMI, smoking, or CCI for estimating muscle-invasion status (eTable 6 in Supplement 1).

Table 3 .
Hazard Ratios for Agent Orange Exposure and Time to Bladder Cancer Diagnosis Stratified by Median Age, BMI Group, Smoking Status, and CCI at VA Entry Corresponds to P value from interaction term between Agent Orange exposure and the given variable in multivariable model adjusted for the covariates in Table 2.controlling for important confounders (ie, smoking).Our study included more than 2 million matched veterans with long-term follow-up and adjusted for important confounders (ie, age, smoking, and BMI) in an equal-access setting to assess whether there is an association between Agent Orange and bladder cancer.Moreover, we found increased risk of bladder cancer among veterans who were least likely to live in poverty, suggesting that improved access to care and other improved social determinants of health may result in improved screening and, thus, detection of bladder cancer.This possibility is further supported by our findings noting decreased risk of muscle-invasive bladder cancer among those least likely to live in poverty, which would support the fact that these patients had earlier detection. b

Table 4 .
Univariable and Multivariable Associations Between Agent Orange Exposure and Muscle-Invasive Bladder Cancer Assessed From Diagnostic Transurethral Resection of Bladder Tumor Among Men With Bladder Cancer Diagnosis Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CCI, Charlson Comorbidity Index.aIncluded American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and biracial individuals.