Abstract 13113: Fragility of Functional/Symptomatic Endpoints to Assess the Efficacy of Drugs for Pulmonary Arterial Hypertension

Abstract

Introduction: Pulmonary arterial hypertension (PAH) drugs have historically been evaluated for the effects on functional capacity and symptoms. However, these assessments are approximate and the ensuing therapeutic choices may be aleatory. Methods: We reviewed the articles describing phase 3 PAH randomized controlled trials (RCTs) and calculated the fragility index (FI) for the outcomes exploring functional capacity and symptoms. For RCTs with a significantly different outcome rate between the investigational drug (ID) and placebo/comparator arms, the FI corresponded to the number of events that needed to be added to the arm with the smallest number of events to made the RCT result non-significant. For RCTs with non-significant results (neutral), the FI was the number of events that needed to be removed from the ID group to reach a p value <0.05. In general, the lower the FI, the fragile the RCT with respect to the endpoint examined. When possible, we also computed the FI for PAH hospitalization. Results: Data about the rate of functional/symptomatic improvement were available for 22 RCTs (Table). The ID was superior to placebo/comparator with p <0.05 in 10 (45%) of these studies. The median FI was 2 (IQR 6.5), with 4 RCTs having a FI = 1 and only 2 a FI >10. For the 12 RCTs in which the effect of the ID was neutral, the median FI was 6 (IQR 4.25). No correlation was found between FI and sample size or year of publication for either significant (R = 0.57, p = 0.08 and R = 0.23, p = 0.52, respectively) or neutral (R = 0.51, p = 0.09 and R = -0.18, p = 0.58) RCTs. The hospitalization FI was determined for 17 (77%) of the 22 RCTs and was overall higher than the one for functional/symptomatic outcomes (median 6, IQR 8) (Table). Conclusions: Very few events in one arm could have flipped the results of PAH RCTs from non-significant to significant or, more remarkably, from significant to non-significant. The evidence supporting the reduction of PAH hospitalizations as a treatment goal appears to be more robust.