Using Data on Survival with Idiopathic Pulmonary Fibrosis to Estimate Survival with Other Types of Progressive Fibrosis Interstitial Lung Disease: A Bayesian Framework

Abstract

Introduction: Among the various types of progressive fibrosing interstitial lung diseases (PF-ILDs), substantial survival data exist for idiopathic pulmonary fibrosis (IPF) but not for other types. This hinders evidence-based decisions about treatment and management, as well as the economic modelling needed to justify research into new treatments and reimbursement approvals. Given the clinical similarities between IPF and other PF-ILDs, we reasoned that patient survival data from four major IPF trials could be used to estimate long-term survival in other PF-ILDs. Methods: We used propensity score matching to match patients with IPF taking either nintedanib or placebo in the TOMORROW, INPULSIS-1, INPULSIS-2 and INPULSIS-ON trials to patients with PF-ILDs other than IPF in the INBUILD trial. Seven models were fitted to the survival data for the matched patients with IPF, and the three best-fitting models were used to generate informative priors in a Bayesian framework to extrapolate patient survival of the INBUILD population. Results: After propensity score matching, the analysis included data from 1099 patients with IPF (640 nintedanib patients; 459 placebo patients) and 654 patients with other PF-ILDs (326 nintedanib patients; 328 placebo patients). Gamma, log-logistic and Weibull models best fit the survival of the matched patients with IPF. All three models led to consistent Bayesian estimates of survival for the matched patients with other PF-ILDs, with median rates of overall survival ranging from 6.34 to 6.50 years after starting nintedanib. The corresponding control group survival estimates were 3.42 to 3.76 years. Conclusion: We provide the first estimates of long-term overall survival for patients with PF-ILDs other than IPF, and our analysis suggests that nintedanib may prolong their survival. Our Bayesian approach to estimating survival of one disease based on clinical trial data from a similar disease may help inform economic modelling of rare, orphan and newly defined disorders.