Modeling and Simulation of Pivotal Clinical Trials Using Linked Models for Multiple Endpoints in Chronic Obstructive Pulmonary Disease With Roflumilast

Roflumilast is a selective phosphodiesterase 4 inhibitor (PDE4i) for the treatment of severe chronic obstructive pulmonary disease (COPD). In 2 large phase 3 trials in a broader population of COPD patients (BY217/M2‐111, ClinicalTrials.gov: NCT00076089 and BY217/M2‐112, ClinicalTrials.gov: NCT00430729), treatment with roflumilast reduced the rate of exacerbations; however, the reduction did not reach statistical significance. Two linked dose‐response models for the primary (annualized COPD exacerbation counts) and secondary (change from baseline in forced expiratory volume in 1 second [FEV1]) end points were therefore developed to characterize and quantify effect sizes and the patient characteristics influencing them. The models showed that disease severity and bronchitis, particularly the severity of bronchitis expressed in cough‐and‐sputum scores, were good predictors of exacerbation rates and differential benefit of roflumilast in exacerbation reduction. The models were used to support the rational design of 2 phase 3 randomized, placebo‐controlled clinical trials (BY217/M2‐124, ClinicalTrials.gov: NCT00297102 and BY217/M2‐125, ClinicalTrials.gov: NCT00297115) by identifying the most appropriate patient population using clinical trial simulations. Model predictions for both end points were found to be highly accurate — as confirmed by the results from these trials, which led to the approval of roflumilast as the first oral PDE4i for the treatment of COPD in patients associated with chronic bronchitis and a history of exacerbations.