Invited Session 3: Integration of EHR and Other External Data in the Analysis of Clinical Trials
Chair: Tianjian Zhou, Colorado State University
Leveraging External Data for Testing Experimental Therapies with Biomarker Interactions in Randomized Clinical Trials
Speaker: Boyu Ren, Harvard Medical School
Dr. Ren is a biostatistician at McLean Hospital and an assistant professor of Psychiatry at Harvard Medical School.
Dr. Ren's research mainly focuses on developing machine learning and Bayesian statistical methods that extract reproducible and generalizable information from the rich high dimensional data sources available in clinical and medical research.
Abstract
In oncology the efficacy of novel therapeutics often differs across patient subgroups, and these variations are difficult to predict during the initial phases of the drug development process. The relation between the power of randomized clinical trials (RCTs) and heterogeneous treatment effects (HTEs) has been discussed by several authors. In particular, false negative results are likely to occur when the treatment effects concentrate in a subpopulation but the study design did not account for potential HTEs. The use of external data (ED) from completed clinical studies and electronic health records has the potential to improve decision-making throughout the development of new therapeutics, from early-stage trials to registration. Here we discuss the use of ED to evaluate experimental treatments with potential HTEs. We introduce a permutation procedure to test, at the completion of a RCT, the null hypothesis that the experimental therapy does not improve the primary outcomes in any subpopulation. The permutation test leverages the available ED to increase power. Also, the procedure controls the false positive rate at the desired α-level without restrictive assumptions on the ED, for example, in scenarios with unmeasured confounders, different pre-treatment patient profiles in the RCT population compared to the ED, and other discrepancies between the trial and the ED. We illustrate that the permutation test is optimal according to an interpretable criteria, and discuss examples based on asymptotic results and simulations, followed by a retrospective analysis of individual patient-level data from a collection of glioblastoma clinical trials.
Regulatory and Statistical Considerations for Externally Controlled Trials
Speaker: Somak Chatterjee, FDA
Dr. Somak Chatterjee is a senior statistical reviewer in the Office of Biostatistics under the Center for Drug Evaluation and Research at the FDA. He has extensive experience in reviewing oncologic drug and biologic products while supporting the Division of Oncology 2. His regulatory expertise includes specialized areas such as innovative trial designs in oncology and use of real-world evidence to support regulatory decision-making. Prior to joining the FDA, he completed his Ph.D. in statistics from The George Washington University.
Abstract
In 2023, the FDA released a related to the design and conduct of externally controlled trials (ECT). ECTs represent an interesting but highly complex design which is considered appropriate in situations where concerns exist in relation to the feasibility, equipoise or ethical conduct of clinical trials. However, it is critical for stakeholders to understand the challenges of ECTs in the context of regulatory decision-making. This presentation will provide an overview of the latest FDA guidance related to ECTs.
Leveraging External Data in Oncology Trials in a Rare-Disease Setting
Speaker: Chenguang Wang, Regeneron
Dr. CG Wang is the Head of Statistical Innovation in Regeneron. Previously, Dr. Wang was an Associate Professor at Johns Hopkins University. He also worked as a Mathematical Statistician at the CDRH, FDA. Dr. Wang has extensive experience in clinical trial design and analysis, especially in regulatory settings, and in statistical software development.
Abstract
Oncology therapeutic development in a rare-disease setting presents unique challenges, not least of which is the limited sample size available for clinical trials. This poses significant difficulties for statisticians tasked with designing trials with reasonable study operating characteristics. In this presentation, we will specifically address trials with extremely small sample sizes, often around 20, which is common in ultra-rare diseases. We will explore the following issues related to the use of external data in these trials: 1) the impact of normality approximation on study operating characteristics, 2) the potential for improved covariate balance between current and external data to enhance study operating characteristics, and 3) the performance of various effective sample size evaluation methods in this context.
Propensity score weighted multi-source exchangeability models for incorporating external data in oncology drug development
Speaker: Wei Wei, Yale School of Medicine
Dr. Wei is an assistant professor at the division of Medical Oncology, Yale School of Medicine.
Dr. Wei's current research focuses on the development of Bayesian statistical designs for master protocols and the leveraging of external data to improve the design and analysis of cancer clinical trials. Dr. Wei has extensive experience in drug development from early phase to late phase trials and has served as the principal statistician on numerous investigator-initiated oncology trials.
Abstract
Among clinical trialists, there has been a growing interest in using external data to improve the accuracy and efficiency of oncology clinical trials. This talk introduces a two-step procedure that combines the propensity score weighting (PW) method and the multi-source exchangeability modelling (MEM) approach to augment the control arm of a RCT. The amount of external data to borrow is determined by the similarities in both pretreatment characteristics and outcome distributions. The proposed approach can be applied to binary, continuous and survival data.