Zhengning Lin* and Donald A Be
Abstract Background: Designing a study for independent confirmation of a treatment effect is sometimes not practical due to required large sample size. Post hoc pooling of studies including those for learning purposes is subject to selection bias and therefore not scientifically solid. We propose a Bayesian approach which calibrates the role of prior information from historical studies for learning and confirming purposes. The method is illustrated in the analysis of mortality data for the pirfenidone NDA. Methods: The pirfenidone NDA includes three placebo-controlled studies to demonstrate efficacy for idiopathic pulmonary fibrosis (IPF), a rare and ultimately fatal lung disease with no approved treatment in the US at the time of NDA. The results of two earlier conducted studies PIPF-004 and PIPF-006 suggested that pirfenidone might reduce mortality risk. We used a Bayesian analysis to synthesize mortality results from the subsequent confirmative Study PIPF-016 and the combination of Studies PIPF-004 and PIPF-006. Results: Pirfenidone’s treatment effect on mortality rate reduction for Study PIPF-016 is statistically significant with discounts of historical evidence from PIPF-044 and PIPF-006 for both all-cause mortality and treatment-emergent IPF-related mortality. Conclusions: The Bayesian analysis provides a formal method to calibrate the role of information from historical evidence in the overall interpretation of results from both historical and concurrent clinical studies. The increased efficiency of using all available data is especially important in drug development for rare diseases with serious consequences, where limited patient source prohibits large trials, and unmet medical needs demand rapid access to treatment options.