Bayesian Modeling of Recurrent Cardiovascular Events and Tumor Heterogeneity in Older Adults with Cancer

About This Grant

Project summary Immune checkpoint inhibitors (ICIs) have transformed cancer treatment, but growing evidence indicates significant cardiovascular risks associated with their use. This is particularly concerning for older adults who are at elevated risk due to increased vulnerability and a higher burden of comorbidities. In many cases, patients experience not a single cardiovascular event, but multiple events over time. These recurrent events reflect the chronic nature of cardiovascular vulnerability and often have greater implications for clinical burden and long-term prognosis. As such, modeling the recurrent nature of cardiovascular risk is important for a comprehensive understanding of long-term cardiovascular burden and for informing effective monitoring strategies in geriatric oncology. In this project, our ultimate goal is to develop novel statistical methods and computational tools to model recurrent adverse cardiovascular events (ACEs) in older adults with cancer. We specifically aim to improve adjustment for tumor-related heterogeneity in modeling recurrent ACEs by addressing limitations in existing approaches that rely on coarse diagnostic categories or fail to account for complex disease structures. The specific aims are as follows. In Aim 1, we will develop a Bayesian recurrent survival model that incorporates latent traits derived from high-dimensional cancer profile data to account for tumor heterogeneity more efficiently. In Aim 2, we will identify clinically interpretable subphetnotypes using a Bayesian nonparametric model that jointly models recurrent events trajectories and cancer profiles as the basis for clustering. We plan to analyze the Yale retrospective cohort of older adults treated with ICIs. In addition to generating new clinical insights, we will develop publicly available and user-friendly software for implementing the proposed methods, ensuring broader scientific impact. This proposal, which will result in new methods and software development as well as clinical findings, will provide valuable insights into cardiovascular risk in geriatric oncology populations.