Session 5: Subgrouping Analysis
Organizer: Xuan Liu (Abbvie)
Chair: Xuan Liu (Abbvie)
Martin King (Abbvie) “Identifying Subgroups in Product Labeling: Two Recent Case Studies”
We evaluate 2 recently approved new drugs for which subgroups were identified in product labeling for potentially different treatment. For each case, trial results and labeling decisions are reviewed in light of the EMA draft guideline on subgroups in confirmatory clinical trials. We discuss the relative contributions of various factors, including evidence of heterogeneity, biological plausibility, pre-specification, and risk of misclassification.
Michael Rosenblum (Johns Hopkins University) “Optimal, Two Stage, Adaptive Enrichment Designs for Randomized Trials, using Sparse Linear Programming”
Adaptive enrichment designs involve preplanned rules for modifying enrollment criteria based on accruing data in a randomized trial. Such designs have been proposed, for example, when the population of interest consists of biomarker positive and biomarker negative individuals. The goal is to learn which populations benefit from an experimental treatment. Two critical components of adaptive enrichment designs are the decision rule for modifying enrollment, and the multiple testing procedure. We provide the first general method for simultaneously optimizing both of these components for two stage, adaptive enrichment designs. We minimize expected sample size under constraints on power and the familywise Type I error rate. It is computationally infeasible to directly solve this optimization problem since it is not convex. The key to our approach is a novel representation of a discretized version of this optimization problem as a sparse linear program. We apply advanced optimization methods to solve this problem to high accuracy, revealing new, approximately optimal designs.
Shuai Chen (University of Wisconsin) “A Flexible Framework for Treatment Scoring in Clinical Studies”
To identify subgroups of patients who have different responses to different treatments, one essentially needs to investigate interactions between the treatments and covariates. Instead of using the traditional outcome-modeling approach, we propose two alternative frameworks for treatment scoring in both observational studies and clinical trials. In particular, we construct personalized scores ranking the patients according to their potential treatment effects. In contrast to outcome-modeling, under our framework, there is no need to model the main effects of covariates. The proposed methods are quite flexible and we show that several recently proposed estimators can be represented as special cases within our frameworks. As a result, some estimators which were originally proposed for randomized clinical trials can be extended to observational studies. Moreover, our approaches allow regularization in presence of a large number of covariates. Many powerful M-estimation technologies can be used in estimation.
