Real-World Data to Support Surrogate Endpoint Evaluation for Health Technology Assessment

OBJECTIVES: To identify levels of validity of a surrogate endpoint (SE), hurdles to their acceptance and how Real World Data (RWD) can support SE evaluation in the context of Health Technology Assessment (HTA).

METHODS: Acceptability of SE by HTA agencies depends mainly on the level of SE validity when evidence on the gold-standard endpoint (GSE) is lacking. The validity depends on SE clinical relevance, correlation with GSE and ability to predict treatment effect size on GSE. Demonstration of validity, often performed by meta-analysis on clinical trials, is time consuming, costly and leads to uncertain conclusions.

RESULTS: In early phase oncology, validation of SE is challenging, often affected by the delayed evaluation of GSE (Overall Survival), as well as the lack of data available for the related indication or therapeutic class. In certain neurodegenerative diseases, the innovative aspect of SE (as cognitive score) or the multiplicity of GSE (as institutionalization, death or other cognitive score) affect the assessment of SE validity. In such cases, RWD can help to evaluate causality between SE and GSE before HTA assessment and confirm effectiveness on GSE after assessment. In France, a lot of qualitative RWD are available (including cohorts, claims database & chart reviews) allowing to support SE assessment. We applied our proposal by implementing two French RWD studies initiated in early Renal Cell Carcinoma and Alzheimer disease designed to demonstrate SE causality relation with GSE.

CONCLUSIONS: Strict methodological validation of SE is rarely feasible, making the demonstration incomplete for industry and decision making complex for HTAs. RWD are a lever of particular interest to better support the demonstration of SE validity and to reinforce the evidence before HTA dossier submission or confirm the long-term treatment effect on GSE. RWD potential should be unleashed to generate evidence and reduce uncertainties, time and cost of demonstration.