Assessment of Overall Treatment Effect in the Presence of Inconsistent Regional Effects in Multiregional Clinical Trials

Abstract

The use of multiregional clinical trials (MRCTs) in clinical developments increased to accelerate patients’ access to new drugs globally. MRCTs inherently assume that the treatment effect is consistent across regions; therefore, we generally estimate the overall treatment effect using a weighted mean of regional estimates with the proportions of enrolled patients as a weight. However, recent MRCTs occasionally encountered an inconsistency in treatment effect among regions. This study was motivated by two actual MRCTs showing the inconsistency of the treatment effects. In these trials, inconsistent treatment effects were observed in a few regions and the estimates in these regions impacted the overall treatment effect estimate. In this study, we propose an exploratory analysis to estimate the overall treatment effect based on robust estimation. The proposed estimator adaptively changes the weights for each region by minimizing the weighted jackknife-based mean squared error of the proposed estimator. We examined the utility of the proposed method through simulation studies. We found that the proposed estimators were less biased and more statistically powerful compared with other ordinary estimators when inconsistent regions existed. We also illustrated estimation of the overall treatment effect by applying the proposed estimators to the two MRCTs.

Keywords:

• Multiregional clinical trials
• Overall treatment effect
• Regional estimate
• Robust estimation

Supplementary Materials

Title:R code used in the illustrations of our proposed method in Section 4 (Illustration.R)

Disclosure Statement

SK is an employee of Novartis Pharma K.K., Tokyo, Japan.

Additional information

Funding

This work was partially supported by Japan Society for the Promotion of Science grant 17K00045 (“Grant-in-aid for scientific research C”).