Orchestrating the Data Symphony: Harmoniously Combining Diverse Datasets in External Control Arm Studies in Oncology

Background: External control arms (ECAs) based on real-world data (RWD) have increasingly been used in regulatory submissions and payor negotiations to inform decisions about drug effectiveness and patient access, particularly in oncology where randomized trials can be challenging. Generating robust real-world evidence using an ECA can require aggregating and harmonizing diverse datasets, e.g. to achieve geographic distribution targets and sufficient sample size. However, heterogeneity in data quality, availability, and composition can pose challenges to validity. Understanding the practical implications of these issues and how to mitigate them will help ensure that ECAs constructed from multiple RWD sources meet decision-makers’ evidentiary requirements.

Objectives: To discuss challenges, mitigations, and regulatory perspectives for designing and executing ECAs from multiple datasets to contextualize efficacy findings from clinical trials, drawing from case examples across hematologic malignancies and solid tumors.

Description: Christian Hampp will provide an Introduction (5 min) and will moderate a panel discussion and Q&A session (25min) after the speaker presentations.

1) Christian Hampp: Set the stage by discussing how the need for fit-for-purpose data for an ECA can necessitate the evaluation, aggregation, and harmonization of diverse datasets, and the challenges involved in doing so. 5 min

2) Nicolle Gatto: Describe an objective, phased approach to ECA design and implementation that relies on published frameworks to assess the suitability of heterogeneous datasets, and allows data-driven decisions during protocol design and implementation, while maintaining objectivity. 15 min

3) Uwe Siebert: Provide a health decision scientist perspective and present approaches to causal inference and target trial emulation, with the aim of avoiding self-inflicted bias (i.e., bias introduced by errors in study design or analysis) and minimizing confounding. Discuss the importance of an independent data review committee guiding stepwise decisions when using multiple ECAs. 15 min

4) Olivier Humblet: Discuss threats to ECA study validity when integrating diverse datasets, such as registries and EHRs, or leveraging datasets from different countries. Approaches for handling data heterogeneity based on a case example in multiple myeloma will be discussed, covering dataset selection; application of I/E criteria; patient deduplication; and analytical approaches. 15 min

5) Patrice Verpillat: Provide a regulatory perspective on the use of ECAs in the EU regulatory context, as well as the challenges involved in making regulatory decisions based on single arm trials. 15 min