Delivering Robust Outcomes from Multinational Clinical Trials: Principles and Strategies Andreas Sashegyi, PhD Eli Lilly and Company

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Organization The problem in context Principles and strategies for addressing heterogeneity –Race –Culture –Standard of care –Data quality The argument for multinational trials

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company The Problem in Context Two Issues: 1.) To what extent are data collected globally applicable / relevant to a narrower region? 2.) To what extent can we generalize results to a broader population? Just another example of the need to deal with variability Understand the extent of variability to be expected Identify reasons for extraneous variability and –Plan ahead –Address with appropriate analyses

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Problem in Context – Reasons for Heterogeneity Race –True differences in reaction to drug (ADME) –Most difficult to address, but also unlikely –Should be identified early in the clinical development program Culture –Differences in customs and attitudes may have profound impact on subjective endpoints (e.g. neuroscience) endpoints and AE reporting –Should not affect reporting of hard endpoints

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Problem in Context – Reasons for Heterogeneity (cont’d) Standard of Care –Differences in standard of care may have profound impact on “soft” endpoints (e.g. acute coronary syndrome) –Likely to affect reported incidence rates of events Quality of Data –Potentially of greatest concern, but also most readily (though not necessarily easily) influenced –Global standardization of processes needed

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Principles and Strategies Fundamental Approach (to Issue 1): –Whereas variability among geographies is to be expected, high-quality global clinical trials should be designed in such a way as to deliver results consistent with that expectation –Every attempt should be made to avoid having to explain or justify inconsistent results in retrospect –Consistency  lack of variability  Underdispersion is as much a concern as overdispersion

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Principles and Strategies (cont’d) –Choose objective (“hard”) endpoints whenever possible –Consider blinded adjudication of endpoints –Enroll representative numbers of patients –If possible, enroll sufficient numbers of North American patients to allow country-to-country variation to be compared to regional variation within North America –Primary analysis needs to be driver

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Principles and Strategies (cont’d) Anticipate issues and –Discuss approaches with regulators –Incorporate strategies into analysis plan, e.g.: –AE reporting: event rates may differ, but should relative comparisons? –Expected heterogeneity should translate into larger variance estimates for sample size calculations –Similar approaches as used for subgroup analyses (considerations of power, multiplicity, etc.) –Simulation to help demonstrate consistency?

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Principles and Strategies – On Quality Consistency = Quality Drive toward global standards Standard education / training Clear protocol instructions / definitions / criteria (e.g. study endpoints) Regular Audits NB.: Global (any) clinical trials should be conducted in a well-controlled experimental environment, so as to minimize the impact of extraneous factors and make the results as conclusive as possible.To that end…

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Principles and Strategies – Quality (cont’d) –Results from data collected in a CT setting in any region may not reflect local standard care. Should we be more worried about this? –Possible remedies: –ITT analyses –Augmenting results of RCTs with observational studies

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Principles and Strategies – On Bridging To what extent can we generalize results to a broader population? (Issue 2) Concern about ethnic (racial) differences Definition: “A study performed in the new (local) region to provide PD or clinical data on efficacy, safety, dosage and dose regimen in the new region that will allow extrapolation of the foreign clinical data to the population in the new region”(ICH E5)

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Principles and Strategies – Bridging (cont’d) Local DataForeign Data PK/PD  Similar  PK/PD Surrogate Endpoints  Similar  Surrogate Endpoints Clinical Outcomes/  ExtrapolationClinical Outcomes/ Confirmatory TrialsConfirmatory Trials –Need to control for demographic differences! –Match subjects from large global trial with similar characteristics to those from the local study

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Principles and Strategies – Bridging (cont’d) Use case-control matching utilizing covariates to choose controls from the global population  Application of Propensity Scores: Let Y=1  subject from the local population Y=0  subject from global population Model logit [P(Y=1)] for all patients as a function of demographic and other background variables

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Principles and Strategies – Bridging (cont’d) –Arrange subjects in local population randomly and select one-by-one –Match ≥ 1 subject from the global population based on proximity of propensity score (choose an appropriate tolerance) Principle: Patients with similar propensity scores have similar probability of being in either population based on their covariate profiles, hence differences in treatment effects seen in the local vs the global population are unlikely to be due to differences in controllable factors

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company The Argument for Multinational Trials Given: –acceptable standards of quality (std of care, data quality) –no reason to believe that a drug would not have a similar therapeutic effect in all members of a culturally/racially diverse population (race, culture) multinational trials make sense and in some therapeutic areas are the only feasible way to conduct adequately powered studies in a reasonable time frame

File name/location Company Confidential Copyright © 2000 Eli Lilly and Company Argument for Multinational Trials – Additional Thoughts –True differences between two geographic regions are best assessed in one adequately powered study than in two separate studies, one in each region –If designed appropriately, such a trial should support registration in each region, based on a primary analysis involving all patients NB.: –Multinational trials are efficient and sensible, but typically not the most convenient option for the sponsor