An Explorative Research On Observed Regional Difference In Treatment Effects

With the globalization of drug development, regional difference in treatment effect attracts more and more attention. Out of consideration of regional difference and regulatory requirements, clinical trials should be conducted in different regions to obtain more clinical data. Currently, the major paradigms of global drug development are bridging studies and multi-regional clinical trials. When using these global data to register in one country or region, it should be assessed whether result of the specific region is consistent with the overall.Relationship in treatment effects between different regions is divided into homogeneous effect, quantitative interaction and qualitative interaction. Reasons for regional differences include:true inconsistent treatment effect caused by ethnic factors, data quality, and chance factor in statistics. In practical clinical trials, regional difference is common, while qualitative interaction is less. Qualitative interaction can influence drug registration and application. Therefore, it’s the focus of this study.When qualitative interaction is observed, it should be judged from various perspectives. So far, assessment methods for qualitative interaction are limited. In addition, policy on regional sample size has been proposed in some countries, but almost no data on applicability of policy are available.Consequently, the objective of study is to explore observed regional difference in treatment effects, which is achieved from three perspectives.Firstly, this study conducted methodology research by simulated cases. When an inconsistent result is observed, conditional probability, Bayesion factor, and traditional methods are used to analyze. The result showed that new methods and traditional methods were applicable. The study provides new methods to make reasonable judgment on observed qualitative interaction, adds data-based evidence, and helps interpret observed inconsistence more scientifically.Secondly, this study analyzed factors affecting the probability of observing qualitatively inconsistent effects in clinical trials, and provided recommendations on designing more reliable clinical trials. The probability of observing inconsistent effects was calculated using mathematical and numerical (SAS) methods. The result showed that factors affecting the probability of observing inconsistent results were:sample size, regional standardized effect size, regional proportion and number of regions. Increasing sample size, increasing the proportion of patients in region with a smaller effect in absolute value and reducing the number of regions can reduce the impact of chance factors and make any observed result more likely the reflection of true effects.Thirdly, this study explored the applicability of policy on regional sample size calculation. The result showed that in terms of power to detect regional difference, Chinese and Japanese current sample size calculation methods had their own applicability. Regional sample size should be decided by appropriate method based on actual situation and total sample size, in order to ensure a relatively higher power to detect qualitative interaction in treatment effect between different regions in clinical trials.