Prediction of Disease Progression and Clinical Response in Systemic Sclerosis: Experience From a Proof-of-Concept Trial.

Abstract

Using the modified Rodnan skin score (mRSS) as a surrogate for disease activity, a phase 2a study in patients with systemic sclerosis (SSc) measured efficacy of the autotaxin inhibitor ziritaxestat. Mathematical modeling of mRSS was used to predict disease progression, examine candidate trial designs, and predict the probability of successfully discriminating treatment effect. Patients with SSc receiving 600 mg of ziritaxestat or placebo for 24 weeks were included, in addition to data up to week 52 of the open-label extension (OLE). Longitudinal mRSS data were described using a disease progression model; drug effect was a binary variable. Parameters used to predict the OLE mRSS outcome were estimated using data from the 24-week double-blind phase and validated with observed data. Three trial designs were simulated to identify which had the highest probability of detecting a treatment effect. Power to detect a treatment effect was quantified using the simulations. Maximum decreases from baseline in mRSS were 50.4% (ziritaxestat) and 34.7% (placebo). Study designs based on 300 patients randomized 2:1 or 1:1 to 600 mg of ziritaxestat or placebo had similar probabilities of detecting a significant treatment effect. Power to detect a treatment effect was >80% for all simulations. Disease progression and drug effect could be predicted beyond the range of observed data. This modeling and simulation approach may inform future trial design, including study duration, and predict the probability of success.