A Cautionary Note on Regional Consistency Evaluation With Multiple Primary Endpoints in Multiregional Clinical Trials

Area of Science:

• Clinical Trials
• Biostatistics
• Regulatory Science

Background:

• Multiregional clinical trials (MRCTs) require evaluation of regional consistency probabilities (RCPs) alongside overall trial power.
• Existing guidance, like the Japanese Ministry of Health, Labour and Welfare's Method 1 and Method 2, addresses regional consistency criteria.
• The impact of multiplicity adjustments on RCP evaluation in MRCTs with multiple primary endpoints remains underexplored.

Purpose of the Study:

• To investigate the inflation of null RCPs when standard multiplicity adjustments (e.g., Bonferroni) are applied to regional consistency evaluation in MRCTs.
• To propose and validate adjusted consistency thresholds that maintain the null RCP at its single-endpoint level.
• To assess the performance of these adjusted thresholds under various multiplicity adjustment procedures.

Main Methods:

• Demonstrated the monotone inflation of null RCPs with increasing endpoints under unadjusted Bonferroni correction for both Method 1 and Method 2.
• Developed and proposed adjusted consistency thresholds using a root-finding algorithm to restore null RCPs.
• Conducted numerical studies to evaluate the proposed thresholds' impact on RCP under alternative hypotheses and their compatibility with step-down procedures (Holm, Hochberg, Hommel).

Main Results:

• Unadjusted consistency thresholds combined with Bonferroni correction lead to inflated null RCPs, especially with more endpoints.
• The proposed adjusted consistency thresholds effectively restore the null RCP to the single-endpoint level.
• Adjusted thresholds cause only a minor decrease in RCP under the alternative hypothesis.
• Bonferroni-derived thresholds also control null RCP inflation when used with Holm, Hochberg, and Hommel procedures.

Conclusions:

• Standard multiplicity adjustments can inflate null RCPs in MRCTs with multiple endpoints, potentially leading to erroneous conclusions about regional consistency.
• The proposed adjusted consistency thresholds provide a statistically sound method to control for multiplicity in RCP evaluation.
• The developed methodology and available R package ('rcpmcp') offer practical tools for designing MRCTs with multiple primary endpoints.