Regional efficacy assessment in multiregional clinical development

ABSTRACT

It is common in multiregional clinical development that data from a global trial and a local trial (in a target country) together will be used to support local filing in the target country. This approach is considered efficient drug development both globally and in the target country. However, it remains a challenge how to combine global trial data and local trial data toward local filing. To address this challenge, we propose an “interpretation-centric” evaluation criterion based on a weighted estimator that weights data from the target country and outside of the target country. This approach provides an unbiased estimate of a global treatment effect with appropriate representation of the target country patient population, where the “appropriate representation” is the desired proportion of the target country participants in a global trial and is measured by the weight parameter. This natural interpretation can facilitate drug development discussion with local regulatory agencies. Sample size of the local trial can be determined using the proposed weighted estimator. Approaches for weight determination are also discussed.

KEYWORDS:

• Global trial
• local trial
• multiregional clinical development
• target country
• weighted estimator

Declaration of interest

The design, study conduct, analysis, and financial support of the analysis were provided by AbbVie. AbbVie participated in the interpretation of data, writing, review, and approval of the content.

Appendix A

Local trial sample size derivation based on the weighted estimator:

→

→

→

→

Appendix B

Back-calculation of weight parameter value p for a given local trial sample size of m per arm, keeping α and β fixed:

Rewrite it as a function of p. When , it becomes:

thus:

Note that for a quadratic function the two solutions take the form of:

and thus:

Therefore, the bigger of the two solutions for p is that:

When , the quadratic function will have:

and thus there is not a guaranteed relation between p and total T data fraction