Read the Full Video Transcript

Rashid Sayyid: Hello everyone. This is Rashid Sayyid, I'm a Urologic Oncology Fellow at the University of Toronto, and along with Zach Klaassen, Associate Professor and Program Director at Augusta University, we'll be discussing the recently published article looking at exposure to Agent Orange and risk of bladder cancer among US veterans. This study was recently published in the JAMA Network Open Journal, with Dr. Steven Williams as the first author. Just a bit of a historical background on this, so the Vietnam War was essentially a conflict in Vietnam, Laos, and Cambodia between 1955 and 1975. Obviously, these dates are going to differ a bit depending on the reference that you look up. Essentially, the US was involved between 1964 and 1973, with a quota peak in 1969. Agent Orange was used in this conflict as a tactical herbicide by the US military for vegetation control. Essentially, veterans who served in Vietnam, the Korean de-militarized zone or the Thai Air Force bases, among other locations of course, may have been exposed.

The active ingredients in this Agent Orange are two phenoxy herbicides and, without going into details, they are listed here. They have been associated with numerous medical comorbidities that may have emerged afterwards. And so, right now the US Department of Veterans Affairs or the VA has on their website the following diseases that are likely associated with Agent Orange, and this information is off the VA website. For the purposes of our listeners, bladder cancer and prostate cancer have both been implicated with exposure to Agent Orange, but it's not quite so clear. There's conflicting data regarding the association of Agent Orange exposure with the subsequent risk of bladder cancer development and mortality. There's been some studies that have shown that there's a positive association, others a negative association so it's quite unclear. And so, based on this confusion, the Institute of Medicine concluded that the association between Agent Orange exposure and bladder cancer outcomes is an area of needed research.

And so, the association between Agent Orange exposure and the potential aggressiveness of subsequent bladder cancer also remains unknown. The study objective was to evaluate the association of Agent Orange exposure with the risk of bladder cancer diagnosis, yes, no, and aggressiveness as well, using data from the VA Health System, which is the largest integrated health system in the United States. For the purposes of this study, the authors used the VA informatics and computing infrastructure, or the VINCI, to identify all veterans that were seen at any VA health system site between January 2001 and December 2019. This dataset uses or includes integrated inpatient and outpatient data in addition to fee-based or community care claims, which essentially means because of potential geographic distances between where the veteran resides and the hospital, the VA system will send a referral to the local community to facilitate care for these patients.

This dataset also captures these consults and claims, given that the VA paid for them themselves and so was able to capture them as well. Another thing is that they included patients from 2001 to 2019 and didn't include patients before that because data from VINCI was not reliable prior to 2000 so that was why the study started in 2001 was chosen. This database includes 25 million veterans across the United States and the eligibility criteria were male active users VA system with at least two medical office visits within five years and served in the Vietnam conflict in the military branch of the Air Force, or the Army or the Coast Guard, the Navy or the Marine Corps. And so, the authors identified 868,912 veterans who were exposed to Agent Orange and, conversely, about two and a half million potential controls. Controls were matched to the cases in a three to one ratio using a stepwise fashion based on factors that were associated with Agent Orange exposure, so the age at service entry, military branch, race, ethnicity, and exact year of service entry.

The way this case-control matching happened in a stepwise fashion is they would start with certain criteria based on these four different factors and then loosen the inclusion/exclusion criteria to allow more and more controls to be selected, so they do it in a stepwise fashion to reach the final cohort. And so, in this analysis the final cohort, as we see in the study flow chart to the right, included just over 600,000 veterans who were exposed to Agent Orange. All patients in the Agent Orange exposed group had documented exposure to Agent Orange containing herbicide agents during active military duty. VINCI specifically has an Agent Orange variable exposure yes, no, so that makes it very easy for the purposes of this analysis. The way that this exposure was verified by VINCI was based on submitted documents fulfilling these requirements as verified by a select VA committee.

The authors also looked at other available patient-level variables, including smoking status (never, former or current) obviously, highly correlated with subsequent risk of bladder cancer, race and ethnicity as well, which is self-reported or clinician assigned, the military service details, medical comorbidities as quantified by the Charleston Comorbidity Index, and socioeconomic status that was essentially defined as the percentage of people living above the federal poverty level based on the patient's zip code. The primary outcome is bladder cancer diagnosis reported as incidence for 1,000 person-years. And the secondary outcome was depth of invasion, meaning muscle invasive versus non-muscle invasive diagnosis. This was determined using a natural language processing model that uses path reports at diagnosis to determine the depth of invasion and prior validation studies. The accuracy of this system was shown to be about 94%.

Given the very large sample size, we anticipate a very high power to detect statistically significant differences so there needs to be a distinction between what is statistically significant and what is clinically significant or clinically meaningful. And so, a prior the investigators decided, that they would use standardized differences and if the standardized difference was greater than 0.1, then that would be deemed a clinically meaningful difference. The association between Agent Orange exposure and the study outcomes were assessed using both uni-variable and multi-variable Cox proportional hazards regression models, given that this was a time to event outcome and this was adjusted in the multi-variable model for the age of VA entry, the year of VA entry, race, ethnicity, BMI, military branch, smoking status, comorbidities, and socioeconomic status.

They also looked at interactions between Agent Orange exposure and the year of service smoking status, BMI, and comorbidities. That's to see whether they needed to stratify the analysis based on these variables. The time zero or the start of the follow-up was defined as the date of first contact with the VA system and not the date of presumed exposure during service. This also raises some issues with lead time bias but it's essentially unavoidable, given that you cannot ascertain exactly when these patients were exposed. And so, to kind of level the playing field for everyone, the date of the first contact with the VA system was determined to be the time zero. Also, the association between Agent Orange exposure and muscle invasiveness at diagnosis was assessed among veterans diagnosed with bladder cancer so not among the entire cohort, just those diagnosed with bladder cancer. This was looked at using a logistic regression model that looks at the outcome, yes versus no. At this point,Rashid Sayyid I'll turn it over to Zach to go over the results and discussion for this paper.

Zach Klaassen: Thanks so much, Rashid, for the introduction, particularly the fascinating history of Agent Orange with regards to the Vietnam conflict. This is the male patient characteristics by Agent Orange exposure, and we can see here that this is stratified by exposed to Agent Orange on the right and to the left of that is not exposed to Agent Orange. When we look at the age at service entry for both cohorts, roughly two-thirds of patients were less than 21 years of age. When we look at the year of service entry, roughly one-third for each group between the aforementioned time periods that Rashid mentioned, less than or equal to 1965, 1966 to 1968, and 1969 to 1975. As you'd expect from our VA cohort, there was a decent black or African-American representation, 13.3%, 83.3% white, and 3.4% other in each of these groups. With regards to the branch of service, most commonly was Army at over 50% and Navy or Coast Guard was just under 20%.

When we look at the year of VA entry, we see some differences here between these two groups. The not exposed to Agent Orange group was 46.1% between 2001 and 2004, which was their most common group. And we see that among patients that were exposed to Agent Orange, 37.1% between 2001 and 2004. The age at VA entry median was 60 years of age for both groups. BMI at VA entry was most commonly greater than or equal to 30, at 36.2% in the non-exposed, and 38.6% in the Agent Orange exposed. With regards to smoking status at VA entry, the majority of these patients were either current or former smokers at over 50%. Charleston comorbidity Index was most commonly zero at just under 50%, and percent living above the poverty line, median was just around 85% in each of these cohorts.

This table looks at the bladder cancer incidence rates by Agent Orange exposure. We can see that among all patients, over 28 million person years of follow-up, the incidence per 1,000 person years was 1.77, and the Agent Orange exposed was 1.84. And then, the not exposed was 1.75. When we look at the stratification below median age, Agent Orange exposed 1.56, which is higher than 1.44 below the median age in the not exposed. And above the median age, Agent Orange exposed 2.21 compared to very similar 2.20 in the not Agent Orange exposed above the median age. This table looks at uni-variable and multi-variable associations between Agent Orange exposure and bladder cancer diagnosis. I've highlighted in the box the variable of interest so this is Agent Orange exposure, yes versus no. We see that those on multi-variable analysis exposed to Agent Orange had a hazard ratio of 1.04, statistically significant 95% confidence interval of 1.02 to 1.06.

Several other notable variables that were statistically significant when we look at race and ethnicity, black versus white, black patients were actually less likely to have bladder cancer diagnosis, hazard ratio of 0.65, 95% confidence interval, 0.63 to 0.67. Those that were in the Marine Corps compared to Army, slightly higher risk of bladder cancer, a hazard ratio of 1.07. Not surprisingly, current smokers compared to never smokers hazard ratio of 2.02, which was statistically significant. And we see basically a stepwise increase in risk for bladder cancer based on Charleston Comorbidity Index, with zero as the reference. You can see the increasing hazard ratio with more comorbidities listed here. What's interesting is that if you look at the percent living above the poverty line, those that were more affluent had a higher risk of bladder cancer, at hazard ratio of 1.07 and a 95% confidence interval of 1.05 to 1.10.

This is the hazard ratios for Agent Orange exposure and time to bladder cancer diagnosis. We see P-values for interaction that were significant with regards to age and year of service entry, so we see that those below the median in terms of age had a hazard ratio of 1.07, 95% confidence interval 1.04 to 1.10. We see that those at year of service entry that were the latest, 1969 to 1975, had a hazard ratio of 1.08, 95% confidence interval of 1.04 to 1.12. This table looks at the uni-variable and multi-variable associations between Agent Orange exposure and muscle invasive bladder cancer diagnosis. We see here with those that were exposed to Agent Orange compared to those that were not, there was a decreased likelihood of muscle invasive bladder cancer if you had Agent Orange exposure odds ratio of 0.91, 95% confidence interval of 0.85 to 0.98.

However, as we saw in the previous tables where black men were less likely to get bladder cancer in this analysis, looking at muscle invasive bladder cancer, they are more likely to get muscle invasive bladder cancer, odds ratio of 1.16, 95% confidence interval of 1.05 to 1.28. Finally, at the bottom we see that those that were more affluent, in terms of percent living above the poverty line, had a decreased likelihood of muscle invasive bladder cancer with an odds ratio of 0.88 and a 95% confidence interval of 0.80 to 0.96. By way of discussion, this is the largest nationwide cohort ever assembled to address bladder cancer risk and aggressiveness according to Agent Orange exposure. As you'd expect, there are several notable findings in this trial. The first is that Agent Orange was associated with a 4% increased risk of bladder cancer, and this association depended on age and the year of service entry.

Agent Orange was associated with a 7% increased risk of bladder cancer for veterans younger than the median age at VA entry, which is critical because bladder cancer is more commonly diagnosed in older individuals, corresponding to the age most Vietnam veterans who may have been exposed to Agent Orange. Other risk factors for bladder cancer were also tested, including race, ethnicity, and smoking. This study found a 35% decreased incidence of bladder cancer among black veterans, when controlling for confounders, and current smokers had up to a twofold increased risk of bladder cancer versus never smokers. Third, the study assessed for the impact of social determinants of health and found an increased risk of bladder cancer among veterans who were least likely living in poverty. This suggests that improved access to care and other improved social determinants of health may result in improved screening and, thus, the detection of bladder cancer. This hypothesis is further supported by decreased risk of muscle invasive bladder cancer among those least likely to be living in poverty.

Finally, the study showed decreased odds of muscle invasive bladder cancer among veterans with Agent Orange exposure, which may be due to earlier bladder cancer detection in the group exposed to Agent Orange. Despite being equal access in the VA, black veterans had increased odds of muscle invasive bladder cancer at diagnosis compared to white veterans. In conclusion, in this large population-based cohort, there was a slight increased risk of bladder cancer among veterans exposed to Agent Orange. In contrast, Agent Orange exposure was associated with a decreased risk of muscle invasive bladder cancer, and these results support prior investigations and further support bladder cancer to be designated as an Agent Orange associated disease.

Thank you very much for your attention and we hope you enjoyed this UroToday Journal Club Discussion.