Multi-indication evidence synthesis in oncology health technology assessment: case-study on bevacizumab

Abstract: 

It is increasingly common for cancer drugs to have their license extended to other indications as trial evidence accumulates over time. Reimbursement agencies seek to make a decision soon after a drug is licensed, leading to large uncertainty in relative treatment effect (RTE) estimates on the final clinical outcome (e.g., overall survival (OS)), or RTE estimates restricted to a surrogate endpoint (e.g., progression-free survival (PFS)), at the time of a technology appraisal (TA). Currently, TAs conducted on behalf of the National Institute for Health and Care Excellence (NICE) restrict their scope to a single target indication despite the availability of substantial evidence, including more mature OS data, from non-target indications. Panoramic meta-analysis methods have been proposed to share information on RTEs across indications, and multivariate meta-analysis methods have become an established way to share information between RTEs on a surrogate endpoint and a final outcome.

This talk will focus on the application of multi-indication meta-analysis methods to survival data from randomised controlled trials (RCTs) assessing bevacizumab vs. comparator therapies across multiple cancer indications, in the context of NICE TAs. Compared to an analysis of RCTs within the target indication alone, methods which allow for sharing of information across indications can provide a gain in precision in the RTE estimates for the target indication. Similar results are seen for estimates of parameters describing the surrogate relationship, suggesting that the methods would be particularly applicable for predicting a RTE in TAs where there are little or no data on the final outcome in the target indication.