Regulatory challenges in the design of non-inferiority trials: Evaluating scientific advice letters from the European Medicines Agency (EMA)

Background: Non-inferiority trials are associated with methodological challenges. The European Medicines Agency (EMA) does not have a guideline on designing non-inferiority trials and recommend to define the non-inferiority margin based on clinical and statistical considerations. However, they do not recommend a specific method to determine the margin. Objectives: To assess the challenges in designing non-inferiority trials for drugs intended to be marketed in Europe. Methods: Using the database of the Dutch Medicines Evaluation Board (MEB), a search in recent (2014 and 2015) final EMA scientific advice letters was conducted to identify design proposals that were sent by pharmaceutical companies to the EMA about non-inferiority trials. Each scientific letter is for one drug, and it includes proposals for different aspects of the trial with a response from the EMA to each proposal. The proportion of the accepted proposals by the EMA was assessed taking into account the therapeutic class and the type of the drug application (orphan vs other drugs) using generalized estimating equations with an exchangeable correlation matrix to account for clustering of proposals within letters. Results: The EMA accepted 142 of 232 (61%) of the total proposals. Almost 65% of the proposals were for three therapeutic classes: anti-infectives (most common), drugs for endocrine disorders (mainly anti-diabetics), and oncology drugs. The EMA acceptance did not differ between proposals for endocrine drugs vs anti-infectives (OR: 1.30, 95%CI 0.52 to 3.24) and between oncology drugs vs anti-infectives (OR: 0.54, 95%CI 0.12 to 2.47). The EMA acceptance also did not differ between orphan vs other drug applications (OR: 0.47 95%CI 0.19 to 1.14). The non-inferiority margin was the main challenge, only 25 of 61 (41%) proposals for the choice of the margin were accepted. There was no common approach proposed by pharmaceutical companies to define the margin (the recommended approach by the EMA was proposed for only 18 of 61 margins) nor a common method of the recommended approach. Conclusions: There are many questions about the design of non-inferiority trials with the choice of the inferiority margin as the main challenge. We did not find that the challenge was related to one of the three most common therapeutic classes or to a type of drug applications. This study shows that more explicit guidance from the EMA on the rationale for choosing different approaches to define the margin is needed