Outline ICH E17 General Principles for Planning and Design of Multi-Regional Clinical Trials Future MRCTs Based on E17 Guideline Background and Key Principles in E17 Some case studies of Gastric Cancer in MRCT conducted by Roche Impact of ICH-E17 Guideline on Global Industry and Biostatisticians - Definition of Region & Consideration of Ethnic Differences - Hideharu Yamamoto, PhD E17 JPMA EWG Chugai Pharmaceutical Col, Ltd 生物統計学シンポジウム Mar.1, 18 Outline Regulatory Guidelines and ICH E17 1998 ICH E5 bridging Ethnic Factors in the Acceptability of Foreign Clinical Data Background and Key Principles in E17 ICH E5 Q&A No.11 Some case studies of Gastric Cancer in MRCT conducted by Roche 7 Japan PMDA/MHLW guidance: Basic Principles on Global Clinical Trials 9 EU Reflection paper on the extrapolation of results from clinical studies conducted outside of the EU to the EU population Impact of ICH-E17 Guideline on Global Industry and Biostatisticians 15 CFDA IMCT guidance Guidance for International Multicenter Clinical Trials (IMCT) Coming ICH E17: To provide common points to consider in planning/designing MRCTs and minimizing conflicting opinions from regulatory bodies 3 1
Possible causes of difference among regions How to look at the results from MRCT Non-compliance with GCP and/or protocol Apparent Differences (Play of chance) Real differences 5 What makes differences? Our Question is... Apparent Differences (Play of chance) Non-compliance with GCP and/or protocol The smaller the local sample sizes, the more likely false signals are observed We don t want to put those causes on the table we are going to discuss So, E17 guideline emphasizes the importance of [Planning] Risk assessment, Common training etc. [Conducting] Monitoring, Quality Management System Real differences Real differences are foreordained by influential ethnic factor(s) or those distribution But we always need to keep in mind this possibility 7 8
Real differences illustrated in Fig. Forest Plot: Summary of subgroup analyses Overall result The overall results belong to everyone. However, the meanings of the results may be different among regions. By region/country Country A Country B Regulatory Region C (e.g., EU) Pooled Region D (Country P+Q+R+...) influences By gender Female Male By severity influences Severe Mild Moderate Yes By a specific med.history No Intrinsic Factor Tend to be milder in countries WITH universal health insurance coverage Tend to be severer in countries WITHOUT universal health insurance coverage Test drug worse Extrinsic Factor Test drug better 9 If you found influential factor(s)... 1 If you found influential factor(s)... Can we explain like this? Try to explain by different distribution of influential intrinsic ethnic factor(s) among regions 11 1 3
Structured exploration Outline of regional differences Preplanned analysis for Known factors Background and Key Principles in E17 Some case studies of Gastric Cancer in MRCT conducted by Roche Impact of ICH-E17 Guideline on Global Industry and Biostatisticians Ad hoc analysis for possible factors Prognostic factors of the disease Further Ad hoc investigation Predictive factors (subgroup analysis, info. outside the MRCT) of the treatment outcome 13 1 ToGA Sample Size in Japan Example ToGA Trial (5-8) 5-FU or capecitabine + cisplatin (n=9) 387 patients screened 81 HER-positive (.1%) HER-positive advanced GC n=58 R MST: 1 months vs 13 months (HR=.77) events (a=5% 1-b=8%, log-rank test), 58 pts Japanese Sample Size 5-FU or capecitabine + cisplatin + trastuzumab (n=9) ToGA was an open-label, international, phase 3, randomized controlled trial undertaken in countries in Asia, Europe and Central and South America. Patients with gastric or gastro-oesophageal junction cancer were eligible for inclusion if their tumors overexpressed the HER receptor Primary endpoint was OS (overall survival) ToGA Overall Sample Size Pre-specification of Japanese sample size Signal of efficacy in Japan is at least needed when the overall population demonstrates a significant difference P1=Pr( HRJapan <.88 true HRJapan=HROverseas=.77) P=Pr( HRJapan <.88 true HRJapan=1.) #event P1 P Sample size.97.31 7 7.71.9 89 8.75.8 1 9.737.7 11 1.78.1 17 1
Primary Endpoint: OS OS for Japanese Population erated 11 Kaplan Meier Curve for Overall Survival Protocol: BO1855 (ToGA): Trastuzumab in HER Positive Advanced Gastric Cancer Analysis: Full Analysis Set - Country: Japan Event 1..9.8.7..5..3..1. Events FP + T FP 17 18 Median HR OS 13.8 11.1 95% CI XP + T XP p value.7.,.91 Japanese Result Events 8 7. Log-Rank Test P =.987 HRJapan=1., p=.98 11.1 13.8 8 1 1 1 1 18 8 3 3 3 3 Time (months) No. at risk 9 77 9 173 17 113 9 9 3 185 13 117 9 71 7 5 3 3 3 1 1 1 13 7 1 5 1 17 Fluoro/Cisp: Fluoropyrimidine/Cisplatin Tras/Fluoro/Cisp: Trastuzumab/Fluoropyrimidine/Cisplatin Fluoropyrimidine: Investigator preference of Capecitabine or 5-FU 18 Program : $HOME/cdp13/bo1855/erated_.sas / Output : $HOME/cdp13/bo1855/reports/erated 11.cgm APR9 19:11 Pre-planned Analysis in Japanese Population OS Subgroup Analysis Category All Subgroup All N 58 HR.7 95% CI.,.91 Region Asia Europe C/S America Other 319 5 19 3.8..3 1..1, 1.11.1,.9.,.89.8, 3.8 8.7., 1.9 1 11.79 1..7, 1.33.59, 1.9 China Korea Japan. Favors T.. 1 Risk ratio 3 5 Favors no T Document similar to the protocol that addresses particular matters concerning Japan (additional information for conducting ToGA study in Japan) Possibility that some demographic or baseline factors might be imbalanced because of the small number of Japanese patients Pre-planning to use an adjustment HR in the Japanese subgroup using multivariate Cox regression with 15 factors All factors were pre-specified in the ToGA study protocol No benefits of Trastuzumab for Japanese patients? 19 5
Preplanned & Post hoc Analysis for Japanese Population As a result of Multivariate analysis, adjusted HR for Japan subgroup changed to.8 from 1. Estimates of effects were extremely unstable for covariates that contained a category which include only one patient To ensure stability of the model, a post hoc analysis was conducted OS and PFS in Preplanned & Post hoc Analysis for Japanese Population HER status was divided into two categories: high vs low expression Covariates that contained a category with only one patient were excluded from the model Adjusted HR=.8 (95%CI:.5-1.5) PFS had a similar result to OS Adjusted HRs for Korea and China subgroup were not changed so large These results suggest that the same benefit was obtained in the Japanese subgroup as in the overall population 1 Conclusion: ToGA Trial Why were Different HRs Observed? Some factors with imbalanced in the ratio of enrollment to each arm in the Japanese population (GC type, prior gastrectomy, PS vs. 1, No. of metastatic sites) Chemo arm was imbalanced towards a better prognosis for these factors compared with Herceptin arm To confirm that the HR is robust, it is necessary to analyze different combinations of factors We found that the HRs were approximately.7 for all combinations of factors, supporting the robustness of our results (Sawaki, et.al 11) 3 Pre-planned and post-hoc analyses suggest that the benefits of Herceptin are of the same magnitude in Japanese patients Herceptin can be considered a new standard therapy for Japanese patients with HER + mgc Herceptin for mgc was approved in Japan in March 11
Examples: AVAGAST Trial (7-8) Phase III trial assessing the clinical efficacy and safety of bevacizumab added to chemotherapy for first-line treatment of advanced gastric cancer AVAGAST was a prospective, randomassignment, double-blind, placebocontrolled global phase III clinical trial. 77 Pts with previously untreated advanced gastric cancer J Clin Oncol. 11: 9: 398-7 Pan-America (19%) Europe (3%) Bevacizumab + Standard Chemotherapy R 1:1 Placebo + Standard Chemotherapy Primary endpoint: Overall survival Stratification factors: Geographical region (Asia, Europe, Pan-America) PMDA-ATC MRCT Seminar 18 (APEC Center of Excellence Workshop) より引用 Asia-Pacific (9%) 5 Subgroup analysis according to region Examples: AVAGAST Trial Hazard ratio.87 (95%CI,.73-1.3, P =.1).97 (95%CI,.75-1.5).85 (95%CI,.3-1.1).3 (95%CI,.3-.9) Bevacizumab seems to be effective in Pan-American patients, but not in Asian patients. PMDA-ATC MRCT Seminar 18 (APEC Center of Excellence Workshop) より引用 Examples: AVAGAST Trial Examples: AVAGAST Trial Author s explanation about the inconsistent result on overall survival among populations (J Clin Oncol. 11: 9: 398-7) Although gastric cancer is a global disease, it is not uniform. There are differences in the presentation and management of gastric cancer patients in different countries and regions. Asian patients more commonly receive second and further lines of therapy more frequently have a prior history of gastrectomy less frequently have liver metastases of proximal or gastroesophageal junction tumors. These difference in extrinsic ethnic factors might have caused the inconsistent result. PMDA-ATC MRCT Seminar 18 (APEC Center of Excellence Workshop) より引用 7 Region PL (months) OS BV (months) HR (95%CI) Asia / Pacific 1.1 13.9.97 (.75-1.5) Japan 1. 15..9 (.7-1.39) Korea 1.9 13.8.89 (.-1.33) Non- Asia / Pacific 8. 11.1.7 (.-.97) Europe 8. 11.1.85 (.3-1.1) Americas.8 11.5.3 (.3-.9) PL (months) PFS BV (months) HR (95%CI) 5..7.9 (.7-1.1) 5.7.8.99 (.73-1.35) 5...79 (.5-1.11)..8.9 (.5-.85)..9.71 (.5-.93). 5.9.5 (.-.93) Especially no benefits of Bevacizumab for Japanese patients? 8 UNAUTHORIZED COPYING AND REPLICATIO 7
Example: JACOB study Example: JACOB study (13-17) 1 JACOB was a prospective, random-assignment, doubleblind, placebo-controlled global phase III clinical trial. 78 Pts with previously untreated advanced gastric cancer R 1:1 Pertuzumab + Trastuzumab + Standard Chemotherapy Overall survival (%) No. at risk P + H + CT 388 PLA + H + CT 39 Placebo + Trastuzumab + Standard Chemotherapy Median, mo 17.5 1..8 (.71 1.).55 P-value (log-rank) P + H + CT (n = 388) PLA + H + CT (n = 39) Censored 8 1 1 1 1 18 33 359 3 339 33 3 97 79 5 3 1 9 175 175 13 19 118 Months 11 95 9 7 7 PLA + H + CT (n = 39) Events, n HR (95% CI) P + H + CT (n = 388) ITT population 8 8 3 3 3 3 38 5 7 3 38 7 31 1 3 1 1 7 3 Primary endpoint: Overall survival Key eligibility criteria: HER-positive mgc/gejc Stratification factors: Geographical region, Prior gastrectomy Asia [excluding Japan] Japan North America/Western Europe/Australia South America/Eastern Europe (ESMO 17) Author s explanation was that OS was generally consistent in subgroups. It is difficult to conclude the Japanese patient has no benefit for gastric cancer. 9 3 Outline Background and Key Principles in E17 Some case studies of Gastric Cancer in MRCT conducted by Roche ICH E17 Think Globally Promote MRCT Promote international harmonization Impact of ICH-E17 Guideline on Global Industry and Biostatisticians A globally harmonized approach to drug development should be considered first Provide better evidences for drug approval in each region 31 Reduce the need to conduct standalone regional or national studies, including bridging studies Build up scientific capabilities and infrastructure globally Encourage better planning and design of MRCTs based on the latest scientific knowledge and experiences 3 8
Significant Factors to Consider for a Successful MRCT From a Global Industry Perspective Big differences between Multi-National Clinical Trials and MRCT under E17 Some points to consider for the global industry Global industry Timing of participation to the MRCT will take care of regional differences and ethnic factors more than before will consider acceptability from every participating region/country more than before should define the region (pooled region) and determine the sample size allocation to each region at the planning stage should obtain HA s agreement with the proposed analysis strategy about regional consistency How to manage the speed of subject enrollment among regions late timing of participation to the MRCT can not keep the prespecified sample size different speed of enrollment among regions could result in difficulties in evaluating the consistency How to manage to obtain the homogeneous baseline characteristics among regions as much as possible different baseline characteristics could influence the consistency of results 33 3 Significant Factors to Consider for a Successful MRCT How to interpret a regional difference in the efficacy among regions important to be able to explain the regional difference Thank you for your attention! Pre-specify in the Protocol/SAP not only subgroup analysis but also model analysis adjusted by prognostic factors ad-hoc analysis or exploratory analysis are very helpful to interpret the regional difference 35 9