Statistical Considerations On Regional Equivalence Tests In Multiregional Clinical Trials

BackgroundMultiregional Clinical Trial(MRCT)may incorporate subjects from multiple regions around the world under a common protocol simultaneously,it could accelerate the availability of important medical products to the needed patients worldwide and increase efficiency in clinical development.The primary objective of MRCT is to assess the efficacy of all participating regions and evaluate the probability of applying the overall results to a specific region.The differences on race,diet,environment,culture,and medical practice among regions might cause impact upon a medicine's effect.How to measure regional treatment effect as well as assure the probability of consistency between regional results and overall results(regional consistency)become an import issue when accessing MRCT.Two major criteria are Method 1 and Method 2 proposed by Japanese MHLW guidance,namely “effect consistency” and “directional consistency”,probability of consistency can be assured if these criteria were used.However,most current discussions on regional consistency are focused on superior designed MRCT,no ideal regional equivalence evaluating methods have been proposed in equivalence designed MRCT.Equivalence trial is aimed to prove the difference between two treatments is not clinically significant,commonly used method is to test whether treatment difference lies in an acceptable equivalence range,borders of which are equivalence margins.Compared with superiority trials,equivalence trials are more cater to ethic requirements due to it uses active control.Thus exploring how to test regional equivalence appropriately to reach expected probability of regional consistency is vital to relevant drug development and registration.PurposeThis article is aimed to explore appropriate methods to reach adequate regional consistency probability by modifying regional test criteria(significance level,equivalence margin)and sample size when regional equivalence tests are required in MRCT,then explore the effectiveness of trial parameters on adjusted results and propose relative suggestions.MethodTheoretical method is applied in our research,based on whether considering the condition that overall null hypothesis is rejected,the probability of regional consistency is defined as product of regional powers of tests or conditional regional powers of tests of all the regions,influence of these two definitions of regional consistency probabilities will be compared.Firstly,the relationship between power of test and regional sample fraction and regional significance in TOST(Two One-Sided Tests)is derived.Secondly,derive how to adjust regional significance level or regional equivalence margins in order to reach adequate probability of regional consistency,calculate and compare trends of adjusted regional test significance and regional equivalence margins under circumstances of different region numbers and regional sample fractions in order to meet expected overall power and probability of regional consistency.Then derive sample size formula to meet expected test power when regional significance level was already set,calculate and compare required sample size under different circumstances.ResultsWe derived formulas for adjustment of regional significance level,regional equivalence margin and sample size calculation to reach expected probability of regional consistency while testing regional equivalence in MRCT.After calculation and comparison,we concluded that adjusting regional test significance level is equivalent to adjusting regional equivalence margin,both of them will expand with the increasing of region number,decreasing of regional sample size proportion,increasing of regional consistency probability and decreasing of overall power.The adjusted regional significance level will become more conservative when regional consistency probability is defined by conditional power.Thus,we recommend adjusting regional test criteria based on conditional power.To avoid excessive expansion of regional test criteria,regional sample size proportion should be greater than 0.4,region number should be greater than 3 and overall power should be adequate.When calculation sample size,small sample size proportion,smaller regional test significance and larger regional consistency probability require more sample size.The sample size will be reduced if sample size proportions of all regions tend to be allocated equally.ConclusionThis research has established suitable methods to adjust regional test criteria and sample size while testing regional equivalence in MRCT,at the same time we compared characteristics of different adjusting methods under common trial parameters.Using results of this study,it is able to test regional equivalence and meet regional consistency simultaneously,so as to provide statistical theoretical basis for designing and evaluation of such trials.