Non-inferiority trials: methodological and regulatory challenges

Abstract

A randomized clinical trial (RCT) is the gold standard to evaluate the intended effects of drugs. In such trials a drug can be compared with a placebo or with another active compound for the same indication. RCTs can be used to demonstrate that a drug is superior to placebo or an active comparator (superiority trial) or that a drug is not worse than an active comparator (non-inferiority (NI) trial). NI trials can be used in a situation when a new drug is expected to have a similar efficacy as its comparator but may offer other advantages over the existing drug such as a more convinient method of administration (for example oral drugs versus parenteral route) or has less side effects. The objective of this thesis was to look deeper into the challenges of the methodology of NI trials, and the role of regulatory guidelines in it. From a methodological perspective, NI trials have special challenges in design and analysis that can influence proper interpretation of its result. First, there are different methods to determine the limit where we can say that the drug is not worse than its active comparator (NI margin). Second, there is a difficulty in interpreting NI trials because of their lack of ability to distinguish an effective drug from an ineffective drug i.e. assay sensitivity, without relying on evidence outside the trial. Third, the validity of the historical data that were used to base the NI margin on, i.e. constancy assumption sometimes is questionable. In the thesis, we discussed the advantages and disadvantages of NI trials based on ethical, methodological and regulatory arguments. We suggest that the non-inferiority aim of showing efficacy of a drug in NI trials should be studied together with aims directed at showing superiority of aspects of the drug over its active comparator e.g. on side effects, or ease of use of drugs. Furthermore, these other study aims should be studied in a scientific sound way (e.g. with adequate statistical power). When these conditions have been met we think it is not unethical to perform NI trials.