All Times EDT
Keywords: Multiple Myeloma, Modeling and Simulation, Surrogate Endpoint, Causal Inference
In oncology/hematology clinical trials, it’s common to observe early efficacies reflected in response (e.g. response rate, depth, timing to and duration) before the follow-up is long for time-to-event endpoints, such as Progression Free Survival (PFS) and Overall Survival (OS). This places challenges in clinical design since PFS and OS are usually the primary endpoints for registrational trials. Multiple Myeloma (MM) is an incurable disease even though the therapeutic area has relatively plentiful of treatment options. Patients typically go through multiple episodes of remission and relapse, and the duration from initial diagnosis to death could span over 10 years. It’s of great interest to establish a quantitative relationship between early efficacy outcomes (such as responses, minimal residual disease negative responses) and time-to-event endpoints, with such a relationship established, an endpoint may be validated as surrogate endpoint.
We have developed a novel simulation tool that is able to accurately establish the quantitative relationship between responses and PFS and/or OS in MM. We approached it by modeling the disease course with a continuous Markov chain (CMC) model. The transition intensity matrix of the model could be parameterized based on publicly reported aggregate level statistics. Then subject-level data with disease courses comparable as reported could be simulated with the parameterized transition intensity matrix. A R shiny simulation tool PubPredict is in development.
This tool could advise hypothesizing the treatment effect size on PFS and/or OS in phase 3 trial and validating potential surrogate endpoint(s). Both aspects are very relevant for clinical trial designs for multiple myeloma therapy development.