The International Conference on Harmonization represents an important step toward more efficient and coordinated global development and approval of new pharmaceutical products. One of the challenges of global drug development is demonstration of efficacy in each of several regions. When a substantial body of data, including Phase III studies, has been gathered in one or more regions, drug development could be accelerated if that evidence of efficacy and safety could be extrapolated (bridged) to another region without repeating Phase III studies on the same scale in the new region.

One class of approaches to the problem of bridging is based on conventional methods of hypothesis testing or Bayesian inference at the population level. Methods within this class include tests for either a) equivalence of treatment effect between regions or b) interaction between region and treatment, and Bayesian models for treatment by region interaction. When such methods are applied to a data set with limited Phase III date in the new region, the result is a tradeoff between increased Type I and increased Type II error in the assessment of the efficacy of a drug in a new region. For example, tests for equivalence designed to have high probability of accepting equivalence of treatment effect in two regions when the treatment effects are truly equivalent will inevitably have low power to detect nonequivalence of treatment effects. This paper will characterize this tradeoff. One more promising approach would rely on biological models that predict the efficacy of treatment as a function of individual characteristics. The growing body of information on the genetic basis of drug efficacy creates potential for bridging of evidence of safety and efficacy based on models for efficacy at the individual level. A second approach would utilize data on drugs within the same class to borrow strength in the assessment of the efficacy of a new drug. This talk will explore both of these approaches.