Assessing the relationship in relapsed-refractory multiple myeloma between response, progression, and survival between pooled clinical trial subjects and a real-world electronic medical record data source.

Abstract

e20525 Background: Despite new treatments, Relapsed-Refractory Multiple Myeloma (RRMM) remains an incurable disease1. In a recent study by Moreau et al, there was a demonstrated survival benefit to treatment with pomalidomide in RRMM patients2. To better understand how the relationships between initial depth of response and long-term prognosis can inform clinical decision making and guide new compound development, we performed an analysis of standardized clinical trial patient pool data compared with Real-World Data (RWD), across multiple studies. Methods: Pooled clinical trial data was obtained from a Study Data Tabulation Model (SDTM) dataset (n=1,815) from the Medidata Enterprise Data Store. De-identified Oncology EMR data was sourced from the Guardian Research Network™ (GRN)5 of integrated delivery systems from 2010-2018, with robust clinical endpoint extraction and curation to enable outcomes comparisons (n=962). Subject selection was refined based on the inclusion/exclusion criteria from the NIMBUS trial2. Response, Progression-free survival (PFS), and Overall Survival (OS) were extracted. Patients were stratified by age, gender, and prior regimens. Log-rank tests were conducted to compare PFS and OS in patient sub-populations at 90, 180, and 240 days from most recent treatment start. Cox proportional hazard models assessed predictors of survival. Rates were estimated for common adverse events, including leukopenia, neutropenia, and thrombocytopenia. Factors associated with neutropenia were assessed using logistic regression. Results: Within pooled trial and RWD patients, the majority were on regimens with proteasome inhibitors, followed by immunomodulators, and approximately one third on monoclonal antibodies. Within the pooled trial analysis, survival rates were consistent with published literature rates, at ~4 months and ~12 months, respectively. EMR data analysis showed overall longer times to progression with an increased depth of response, with similar associations. Frequency-Factored Time-to-Progression also decreased. Conclusions: The use of SDTM for pooled clinical trial analyses, together with real-world data, can overcome individual trial sample size limitations and biases. Together this approach can expand the range of populations, relative treatment comparisons, and clinical events that can be studied to more comprehensively understand the complexity of the oncology treatment landscape.