Amanda Nizam, MD, Cleveland Clinic, provides an overview of her recent research on biomarkers of treatment-related adverse events (TRAEs) associated with enfortumab vedotin (EV) for advanced urothelial carcinoma (UC). As part of a series of analyses of the UNITE study, Dr. Nizam shares what the identified biomarkers were and whether somatic gene alterations are strongly predictive of treatment-related toxicity for patients receiving this therapy.
Previous research has showcased biomarkers of response to EV in patients with advanced UC in the UNITE study. How strong were the data related to TRAEs in the previous analyses of the UNITE study?
Dr. Nizam: My colleagues, Dr. Koshkin and Dr. Jindal, previously presented data at ASCO GU and at ASCO in 2023 about somatic alterations associated with overall survival with EV. So the generation of this project really came out of my interest in TRAEs, and initially, I wanted to examine germline mutations associated with toxicity, as there is some data in breast cancer with taxane-related neuropathy associated with certain germline SMPs. However, we didn’t have enough of that data available; the UNITE Registry is a retrospective registry. So we decided to examine somatic mutations using multivariate analyses, controlling for pertinent clinical factors.
We started with 607 patients at 16 US sites; 375 of those patients had EV TRAEs. We then narrowed it down further to those who had available NGS data, either liquid or tissue somatic NGS, narrowing the population down to 275 patients. The most common side effects we observed were neuropathy (46% of patients), skin toxicity, fatigue, GI toxicity, cytopenias, and anorexia or weight loss, in descending order of frequency. The median time from EV start to AE onset was about 4 weeks, with a range of 2 to 11 weeks.
For the previous iterations that examined outcomes between survival and somatic alterations, obviously, we had much larger datasets. This was a much more focused dataset given that patients needed to have EV TRAEs and available NGS data.
Why is studying treatment-related toxicity important, particularly for a novel and efficacious therapy like EV?
Dr. Nizam: EV is a potent drug with remarkable response rates, as demonstrated in the ESMO data from EV-302. Now, practitioners dealing with metastatic bladder cancer often encounter dose-limiting toxicity, with the primary reason for treatment discontinuation being the development of neuropathy. Skin toxicity is another factor, but neuropathy tends to be the most prevalent issue. Fatigue and anorexia, leading to weight loss, are also observed in some patients.
The goal of studying toxicity and identifying predictive biomarkers is crucial. Drawing parallels from breast cancer, especially with taxane-related neuropathy, differences in patient characteristics contribute to varying thresholds for toxicity development. Identifying personalized toxicity management and recognizing these issues earlier enables us to adjust doses or schedules appropriately. This approach ensures that patients can continue with the treatment and derive maximum benefit.
Why did you deem it important to investigate biomarkers of treatment-related toxicity in this population?
Dr. Nizam: Certainly, germline factors make sense as they are inherent to us and can impact pharmacodynamics or pharmacokinetics. However, somatic alterations can also significantly influence the tumor microenvironment. It was a logical question to explore whether any of these alterations are linked to variations in either the pharmacokinetics or pharmacodynamics of EV.
What were the identified biomarkers and how did you go about finding them?
Dr. Nizam: We examined a variety of clinical variables and approximately 16 groupings of somatic alterations, each occurring in over 5% of patients. These genomic alterations included DNA damage, response, repair alterations grouped together, and CDK and 2ACDK, 2B, and MTAP alterations grouped together due to their location on the same chromosome. Other alterations were considered individually. We assessed clinical variables to gather relevant data and specifically focused on those somatic alterations. In our initial step, we conducted univariate analyses of all clinical variables and somatic alterations to evaluate their association with specific TRAEs from EV.
In the univariate analysis, we found that diabetes, prior neuropathy, higher hemoglobin and albumin at the start of EV were significantly associated with neuropathy. Lower white blood cell count and neutrophil-lymphocyte ratio, along with GFR greater than 30, were associated with skin toxicity. TERT alteration and lower absolute lymphocyte count were linked to cytopenias. Male sex, liver metastasis, and KDM6A alteration were associated with anorexia and weight loss. Higher BMI, diabetes, lower neutrophil-lymphocyte ratio, and male sex were associated with hyperglycemia.
Although we identified a few random associations in the univariate analyses, it’s important to note the presence of potential confounding factors. To isolate the impact of somatic alterations on EV-related TRAEs, we conducted multivariate analyses (MVAs) that adjusted for pertinent clinical variables. Each MVA model included about five or six relevant variables, such as baseline diabetes, prior neuropathy, BMI, prior platinum chemotherapy, and other significant factors from univariate analyses.
Interestingly, our multivariate analyses did not reveal significant associations except for ML2 genetic alteration being linked to cytopenias and KDM6A alteration being associated with anorexia and weight loss. These analyses focused on each of the 16 alteration groups and their association with specific toxicities like neuropathies, cytopenias, weight loss, hyperglycemia, ocular toxicities, etc. This suggests that somatic alterations may not play a substantial role in the underpinnings of EV-related toxicity.
The next phase of our study involves conducting multivariate analysis models using clinical variables to further explore these associations.
What are the largest takeaways from your study?
Dr. Nizam: An essential takeaway from this analysis is that in the quest to develop predictive biomarkers for TRAEs, there isn’t a single definitive variable. It’s more likely a combination of different variables. This is why our focus is on isolating which variables could be related to treatment-related toxicity and which may not be. It’s important to note the retrospective nature of these analyses, and the attributions of TRAEs were determined by investigators at each site. Despite these limitations, there is value in exploring this prospectively.
Our study suggests that somatic genomic alterations do not appear to be associated with predicting EV TRAEs. However, further investigation into host and tumor-related factors may reveal other underlying aspects of EV toxicity. Emphasizing clinical variables, as we will in multivariate analyses, is crucial. Additionally, collaboration with companies to examine germline alterations in patients using a different dataset and a prospective approach could shed light on specific toxicity associations with certain germline SNPs, similar to what is observed with taxane-related neuropathy in breast cancer.