Potential of adaptive clinical trial designs in pharmacogenetic research, A simulation based on the IPASS trial

Abstract

Background: An adaptive clinical trial design that allows population enrichment after interim analysis can be advantageous in pharmacogenetic research if previous evidence is not strong enough to exclude part of the patient population beforehand.With this design, underpowered studies or unnecessary large numbers of patients given an inferior/harmful treatment can be avoided. Objectives: Aim of this study was to illustrate the potential benefits of an adaptive trial design using a simulation model based on empirical data. Methods: The simulation was based on data of the IPASS trial, a phase III trial in non-small-cell lung cancer patients, aimed to study the efficacy and safety of gefitinib. A subgroup analysis showed that the tumor response of patients positive for the EGFR mutation was 71.2% with gefitinib versus 47.3% with control, and 1.1% versus 23.5%, respectively, in the mutation-negative subgroup.We simulated two scenarios of an adaptive trial, one using the response rates of the IPASS trial and one assuming an equal response rate (20%) in mutation-negatives, which is a more realistic assumption when planning a trial. The conditional power was calculated to decide at interim analysis whether (1) to continue the trial with the whole population, (2) to continue with only mutation-positive patients, or (3) to stop due to futility. Results: The simulation showed that the overall ±-level was maintained at 2,5% (one-sided). The first scenario resulted in a probability of .91 to continue with only mutation-positive. Continuation with the whole population in this scenario resulted in the largest bias in the treatment effect estimates. Scenario 2 resulted in a probability of .54 to continue with the whole population and .38 to continue with only mutation- positive. Stopping the trial for futility in this scenario lead to the largest bias in the treatment effect estimates. Conclusions: We demonstrated that if an adaptive trial would have been performed in the case of gefitinib, continuation with an enriched population would have been very likely, which would have lead to a smaller trial, with less EGFR mutation-negative patients unnecessarily exposed to the drug.