Tree-based exploratory identification of predictive biomarkers in non-randomized data

Résumé fourni par la source

Abstract The idea of “stratified medicine” is an important driver of methodological research on the identification of predictive biomarkers. Most methods proposed so far for this purpose have been developed for the use on randomized data only. However, often data from clinical registries or non-randomized studies are the only available data source. For such data, methods for estimation of the average treatment effect are well established. Research on confounder adjustment when investigating the heterogeneity of treatment effects and the variables responsible for this is usually restricted to regression modelling. In this paper, we demonstrate how the predMOB, a tree-based method that specifically searches for predictive factors, can be combined with common strategies for confounder adjustment. In an extensive simulation study, we show that the proposed strategies allow for the correct identification of predictive factors in the presence of confounding. It is demonstrated that the most robust method is a combination of both covariate adjustment and Inverse Probability of Treatment Weighting (IPTW). Applications to the German Breast Cancer Study Group (GBSG) trial 2 and the Acute Myeloid Leukemia Study Group (AMLSG) 16-10 study illustrate these conclusions.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.