Just a moment, the page is loading...

Development of an accurate statistical method for evaluating region-specific treatment effects in multiregional clinical trials

Development of an accurate statistical method for evaluating region-specific treatment effects in multiregional clinical trials

Hisashi Noma

The Institute of Statistical Mathematics, Research Organization of Information and Systems, Tokyo, Japan

This study is supported by Grants-in-Aid for Scientific Research (15K15954, 26285151) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

A research project of Hisashi Noma was supported by GSK Japan Research Grant, but the project was completed, and it is not related to the current proposal.    
Takahiro Hoshino has been funded by FUJITSU LABORATORIES Ltd, which is not related to the current proposal.

26 May 2017

Multiregional clinical trials have been widely adopted in drug developments. The treatment effects are commonly assessed focusing on the difference in the averaged effects between investigational and control groups. However, as mentioned in recent medical statistics papers (e.g., Cole and Stuart, 2010, Am J Epidemiol 172: 107-115; Hartman et al., 2015, JRSS A 178: 757-778), various demographical, environmental, genetic, and socioeconomic factors are generally heterogeneous across the regions, and they are possibly influents to the treatment effects. Therefore, they often complicate the evaluations of the effect of investigational drug. If there is systematic heterogeneity of treatment effects across the regions, their accurate evaluations are quite important for public health and would be relevant information for regulatory agencies.
The existing methods based on regression models can remove differences of pretreatment variables between regions or institutions to estimate heterogeneous average treatment effects. However, they can result in severe biased estimates if regression models of pretreatment variables are mis-specified. A few methods dealt with this problem but failed to remove the bias correctly, because they neglect the unobservable shared effects between individuals in each region.

This proposal involves a methodological study conducted by our research team. We have developed a statistical method to correctly estimate the heterogeneous average treatment effects as functions of institutional or regional variables, without correct specification of regression models of pretreatment variables. The proposed method is developed based on a potential outcome approach and statistical theories for missing data problems. The proposed estimation method is general in that it is applicable to various dependent variables such as categorical variables, continuous variables, and survival time with censoring. This model and estimation method will shed light on the mechanism why treatment effects vary according to regions or institutions, and enable decision makers to decide which medical intervention is optimal in their region or institution. We will apply the proposed novel method to data sets from the selected studies for which we request access, to illustrate the usefulness of our approach.

Expected results
The potential benefit of the proposed method for public health would be relevant in that it enables researchers to identify the mechanism underlying heterogeneity and also enables clinical decision makers to choose better treatments for their own institutions or regions.

[{ "PostingID": 1362, "Title": "ROCHE-BO18255 (ToGA)", "Description": "An open-label randomized multicentre Phase III study of trastuzumab in combination with a fluoropyrimidine and cisplatin versus chemotherapy alone as first-line therapy in patients with HER2-positive advanced gastric cancer.

Medicine: trastuzumab, Condition: HER2-positive metastatic breast cancer, Phase: 3, Clinical Study ID: BO18255 (ToGA), Sponsor: Roche" },{ "PostingID": 3204, "Title": "LILLY-H3E-MC-JMEN", "Description": "A Phase 3, Double-Blind, Placebo-Controlled Study of Maintenance Pemetrexed plus Best Supportive Care versus Best Supportive Care Immediately Following Induction Treatment for Advanced Non-Small Cell Lung Cancer

Medicine: Pemetrexed, Condition: Non-Small Cell Lung Cancer, Phase: 3, Clinical Study ID: H3E-MC-JMEN, Sponsor: Lilly" },{ "PostingID": 3744, "Title": "LILLY-I4T-IE-JVBD", "Description": "A Phase 3, Randomized, Double-Blinded Study of IMC-1121B and Best Supportive Care (BSC) Versus Placebo and BSC in the Treatment of Metastatic Gastric or Gastroesophageal Junction Adenocarcinoma Following Disease Progression on First-Line Platinum- or Fluoropyrimidine-Containing Combination Therapy

Medicine: Ramucirumab, Condition: Gastric Cancer, Phase: 3, Clinical Study ID: I4T-IE-JVBD, Sponsor: Lilly" },{ "PostingID": 3767, "Title": "ASTELLAS-OSI-774-302", "Description": "A multicenter, randomized, double-blind, placebo-controlled, phase 3 study of single-agent Tarceva® (erlotinib) following complete tumor resection with or without adjuvant chemotherapy in patients with stage IB-IIIA non-small cell lung carcinoma who have EGFR-positive tumors

Medicine: erlotinib, Condition: Non-small cell lung cancer, Phase: 3, Clinical Study ID: OSI-774-302, Sponsor: Astellas" },{ "PostingID": 4204, "Title": "NOVARTIS-EGF102988", "Description": "A Randomised, Double-Blind, Placebo-Controlled, Multi-centre, Phase III Study of Post-Operative Adjuvant Lapatinib or Placebo and Concurrent Chemoradiotherapy Followed by Maintenance Lapatinib or Placebo Monotherapy in High-Risk Subjects With Resected Squamous Cell Carcinoma of the Head and Neck (SCCHN)

Medicine: lapatinib, Condition: Neoplasms, Head and Neck, Phase: 3, Clinical Study ID: EGF102988, Sponsor: Novartis" }]

Statistical Analysis Plan

The publication citation will be added after the research is published.