The Research Of Testing Criterion In Targeted Regions Based On MRCT Of Superiority Design About Survival Endpoint

Background In the recent years, drug development is becoming globalized. Through participating in Multi-regional clinical trials(MRCT), diffident countries or regions can share MRCT-related data, as well as the resources of the drugs' development. MRCT is a single study conducted simultaneously across several regions under a common protocol. However, MRCT faces a challenge: if a new drug, which have been registrated, approved and marketed in original region, wants to be approved in other regions, it will be nessary to launch a new trial to evaluate the safety and efficacy in that. Due to the differences in ethnicity, culture and clinical practice between regions may have impact on efficacy, safety and dosing regimen of a drug, lots of resources are wasted up and the intervals to market approval are delayed. As a result, drugs can not be administrated with the patients in need of durg as soon as possible. Moreover, in the targeted regions of MRCT effective criteria and details on the sample size estimation and scientifical statistical significant level are not be established now. To encourage efficient and rapid developments of new drugs in global clinical trials, The guidelines ICH-E5 put forward the concept of bridging study. Nevertheless, due to its proven efficacy and safety, a bridging study is usually conducted in the new region after the product is approved for commercial marketing in the original region. Even through the “drug lag” problem still exists. The purpose of our study was to explore which criteria can make the target area synchronized with the international multi center clinical trial.Objective and Methods A study of superiority design was simulated to explore the statistical decision rules for survival datasets in the target region of MRCT. Two survival datasets were generated by Monte Carlo simulation, the superiority in target region was tested by log-rank under the established superiority in MRCT to evaluate adjusted statistical significant level ?' and estimate the minimum sample size and the events according to the conditional false positive rate(CFPR) and conditional power(CP) to provide a reference for target region of MRCT.This study constructed a two-arm superiority design of MRCT about survival data, including the targeted ethnic region and nontargeted ethnic region.The proprotion of subjects in the treatment group and the control group is 1: 1, the primary efficacy endpoint is OS(overall survival- months), the efficacy of the test group were significantly superior to the control group based on the total significance level ?= 0.05.All data of this study were generated and analyzed by SAS 9.1.3 software. Two kinds of distribution-exponential distribution and Weibull distribution Survival data were generated. Superiority of the target area was tested under global superiority assumption. The proportion of superiority in the target ethnic rigion was counted according the adjusting significance level ?'. The simulation was performed 10,000 times for every condition.Results and Conclusion In our study, survival data for exponential distribution and Weibull distribution were investigated, respectively. Under the precondition of general effectiveness(?= 0.05 and the median survival time of the test group is higher than that of the control group), we conducted the study in the target rigion. while the median survival time of the treatment group longer than that of the control group in the target rigion, through comparing target regions and setting CFPR and CP in the different levels, The feasibility significance level ?' in target areas can be investigated and sample size and events can be caculated. From the simulation study, we draw the conclusions: nomatter what distribution the survival data obeyed, the CFPR and CP increased as the ratio of sample size K and a level in the target region become higher. when survival data was exponentially distributed,setting ?'?0.5, the CFPR can be controlled within 50%. When K ? 30%, setting ?'=0.3 and ??0.9, then the CP in the target regions will larger than 72.1% in the target regions; setting ?'?0.4, and ensuring ? ?0.8, CP will exceed 72.1% in the target regions. When ?'?0.5 when possible to ensure CFPR controlled within 0.5; when K?30%, setting ?'= 0.3, the need to ensure ??0.9, target areas CP can ensure that more than 72%; Based on the assumptions that " ninety percent probable that drugs' treatment effects are shown to be significant based on MRCT data are actually effective in the target region" and the reasonable parameters drew from the study, CP after adjustment is more than 64%, CFPR after adjustment is controlled under 0.05 or less. When the survival data obeys the Weibull distribution, the variation trend of CFPR and CP were similar to the variation trend of exponential distribution in target area, but CFPR and CP less lined with shape parameter ?. when K?30%, setting ?'? 0.4,the need to ensure ??0.9, target areas CP can ensure that more than 76%;We concluded that the sample size of targeted region of MRCT should be larger than 30% of total sample size and ?'should be lower than 0.5. The criterion can be applicated in the situation that treatment effect of the target region is noninferior to that of MRCT(0.8???1).Once same treatment effect for target region and MRCT was assumed, ?' should be within 0.3 and 0.5. If the treatment effect in the target region is far less than that in MRCT(?<0.8), the sample ratio in the target zone will increase(K >0.5). When K>0.5, conditional I type error inflates dramatically, cautious should be taken at this situation.