(S-053) Leveraging historical clinical trial data using Model-Based Meta-Analysis (MBMA) to investigate the impact of prior biologic therapy on the efficacy of treatments in Psoriatic Arthritis

Anna Beutler – Johnson & Johnson; Edaire Cheng – Johnson & Johnson; Mehrdad Javidi – Johnson & Johnson; Chandni Valiathan – Johnson & Johnson; Qingmin Wang – Johnson & Johnson

Objectives: Biologics are commonly used to manage chronic inflammatory arthropathies, psoriasis and inflammatory bowel diseases. When studying the efficacy of new compounds for these indications, the prior biologic therapy status (i.e. biologic-naive vs. biologic-experienced) of trial participants could be an indicator of treatment-refractory disease that can affect the clinical response to a new compound. Therefore, prior biologic therapy status is considered when designing clinical trials. Published clinical trial data for Psoriatic Arthritis (PsA) trials were analyzed using Model-Based Meta-Analysis (MBMA) to quantify the relationship between prior biologic therapy status and two clinical outcomes, ACR20 and ACR50 which are frequently utilized as primary endpoints in PsA trials.

Methods: A database was developed using systematic literature review resulting in review of 89 trials published from 2002 to 2023 and investigating 39 treatments including placebo in PsA. Clinical endpoint values in the database are reported at aggregate level across time and study arms. Data from a mixed population containing both patients naive and exposed to prior biologic therapies were included. A detailed review of studies was performed to identify studies, arms or stratified arms that use the relevant populations with prior biologic therapies for the final analysis of ACR20 or ACR50. Nonlinear mixed-effects, longitudinal, parametric placebo, dose response models were developed separately for ACR20 and ACR50 endpoints to estimate endpoint values across treatments, timepoints, and populations. Random effects were defined at study level. Study start year, study phase and percent of population with prior biologic therapies were included as covariates for both models.

Results: For the final analysis, 52 and 46 studies respectively reporting ACR20 and ACR 50 by prior biologic therapy exposure, were identified including 15 unique study treatments for up to 24 weeks. Model results showed that at Week 24 the absolute ACR20 and ACR50 responses in the biologic-naïve population were higher than in the biologic-experienced population for both active and placebo arms. However, while the treatment effect (i.e., placebo-corrected) at Week 24 for ACR50 was higher in the biologic-naïve population compared to biologic-experienced population, it was similar for ACR20. Treatment effect for ACR20 were in general similar between naïve and experienced populations at later time points beyond 8 Weeks. In contrast, treatment effect for ACR50 were observed to be higher in biologics naïve populations compared to the biologics-experienced population at all timepoints through week 24.

Conclusions: MBMA models leverage the totality of available historical data from several compounds with different mechanism of action to provide quantitative and data-informed insights on clinical efficacy as measured by ACR20 and ACR50 in PsA for different populations based on prior biologic therapy experience. These quantitative projections can guide the clinical development strategy and help with dose-selection or sample-size calculation with data supported estimates.

Citations: NA

Keywords: MBMA, Psoriatic Arthritis, Prior Treatment