(M-004) MBMA Bridging Models as A Tool for Exploration of Clinical Endpoints in Unstudied Indications.

Teddy Kosoglou, N/A – Consultant, Janssen Research & Development, Spring House, PA, USA.; An Vermeulen, N/A – SR. DIRECTOR, Janssen Research & Development, a division of Janssen Pharmaceutica NV , Beerse, Belgium; Chandni Valiathan, N/A – Director, Janssen Research & Development, La Jolla, California, USA.

Objectives: In many instances, drugs used for one indication may be viable for a different indication. However, a randomized controlled trial (RCT) in the new indication of interest is often needed to determine the efficacy or safety of the treatment profile in another disease area. One approach to accelerate clinical development for treatments in a new indication is to use Model-Based Meta-Analysis (MBMA). MBMA is a quantitative approach of utilizing aggregate data from RCTs to develop modeling tools that incorporate properties of clinical data such as drug, dose, time, and other covariates. This approach leverages mathematical models using information-rich datasets to facilitate bridging of clinical endpoints across indications to help determine how novel treatments could behave in a new indication. An example from immunology is used to illustrate the methodology.

Methods: Multi-Center, randomized placebo controlled and parallel design studies investigating the treatments of psoriasis (PsO) and psoriatic arthritis (PsA) published between 2000 and 2023 were identified using a systematic literature review from PUBMED, clinicaltrials.gov and other sources (e.g., FDA clinical review). The bridging approach combines multiple datasets to create a joint non-linear model across different indications/endpoints and using data from the first indication the model generates clinical efficacy in new indications. The longitudinal, non-linear mixed effects parametric placebo bridging model with Emax dose response, was developed to bridge between PsO and PsA. PASI75 scores for both indications were used to bridge across indications and additional shift parameters were used to bridge to the ACR20 endpoint commonly used in PsA. A leave-one-out cross-validation (LOOCV) analysis was performed, where a deucravacitinib study in PsA for the ACR20 endpoint was removed from the analysis and compared using the model.

Results: The total of 85 studies were identified for PsO and PsA combined, with 49 studies in PsO. This large number of studies and datapoints allowed us to capture drug performance from one endpoint to the next with good precision. In the LOOCV, the models mean [90% CI] ACR20 score of 51.7% [39.1%, 63.8%] for deucravacitinib 6 mg QD at week 24, which closely matched the observed value of 52.9%. Similarly, projections or efficacy in PsA can be estimated using data from PsO.

Conclusion: There is often an interest in understanding potential clinical efficacy or safety of treatments in new indications. The combination of datasets across indications facilitated the use of information about the population and therapeutic landscape for cross indication and endpoint analysis using MBMA. The results of the cross-validation showed that efficacy of a new treatment in PsA can be assessed using the MBMA bridging method using data from another indication like PsO. Such assessments can be utilized to explore new treatments for new indications.

Citations: N/A