Real-World Survival Outcomes in Patients with Intermediate-to-High-Risk Non-Muscle Invasive Bladder Cancer Utilizing Medicare Claims Data: Estimating Event Free Survival and Evaluating Its Prediction of Overall Survival

OBJECTIVES: To demonstrate estimated event free survival (EFS) as a predictor for overall survival (OS) in patients with Intermediate-to-High-Risk Non-Muscle Invasive Bladder Cancer (IR/HR-NMIBC) who receive adequate BCG maintenance therapy defined as ≥7 installations of BCG within 274 days of BCG initiation.

METHODS: This was a non-interventional retrospective cohort analysis using the complete sample (100%) of Medicare Fee-for-Service beneficiaries with IR/HR-NMIBC. Estimated EFS was captured as a composite measure defined as the time in days from the index date (date of BCG initiation; inclusive) to the first occurrence of events indicating disease progression during the follow-up period. OS was measured as the time in days from the index date to the validated date of death from any cause. Random Survival Forest (RSF) was developed to elucidate variable importance for prediction of OS. Cox regression analysis was generated to investigate the strength of EFS as a predictor of OS. Survival tree analysis, informed by the RSF, was leveraged for EFS classifications for a separate Cox regression.

RESULTS: 19,859 patients with IR/HR-NMIBC and adequate BCG were identified from 2010-2020. During the median follow-up of 46.3 months, 10,008 patients had ≥1 event and 5,573 deaths were captured. The median event-free-survival time was 31.5 months. C-Index for the RSF is 0.81 in the training and the test set, and event-free-survival time and age were the most predictive covariates of OS. Cox regression results showed EFS measured in 3-month periods is positively associated with OS (HR: 0.96 p < 0.01, model concordance 0.67). Cox regression based on survival tree binary EFS classification showed EFS >41.5 months is positively associated with OS (HR: 0.60, p<0.01).

CONCLUSIONS: Results suggest that EFS is a strong predictor of OS in adequately treated IR/HR-NMIBC Medicare patients and should be further investigated as a surrogate endpoint of OS for clinical trials.