Emulating Randomized Controlled Trials with Hybrid Control Arms in Oncology: A Case Study

Area of Science:

• Clinical Trials
• Real-World Data Analysis
• Oncology

Background:

• Randomized controlled trials (RCTs) are the gold standard for evaluating cancer treatments.
• Real-world data (RWD) offers a valuable supplementary data source for clinical research.
• Hybrid control arm designs combine RCT and RWD to potentially improve trial efficiency.

Purpose of the Study:

• To assess the feasibility of a hybrid control arm design using RWD to emulate a completed phase III RCT.
• To evaluate the performance of frequentist statistical borrowing methods in this hybrid design.
• To determine the potential reduction in study duration compared to a traditional RCT.

Main Methods:

• Retrospective analysis of a completed phase III RCT (CheckMate-057) in advanced non-small cell lung cancer.
• Emulation of a trial with an experimental arm from the RCT and a hybrid control arm using both RCT and RWD control cohorts.
• Application of three frequentist statistical borrowing methods: two-step Cox model, fixed Cox model, and propensity score-integrated composite likelihood.
• Evaluation of experimental treatment effect using hazard ratios (HRs) and 95% confidence intervals (CIs), and assessment of study duration reduction.

Main Results:

• All three statistical borrowing methods yielded comparable experimental treatment effects to the original RCT.
• Hazard ratios (HRs) were consistently estimated across methods (0.73-0.74).
• Study duration could be reduced by 99-115 days by borrowing 30-38 events.

Conclusions:

• Emulating an RCT using a hybrid control arm design with RWD and frequentist borrowing methods is feasible.
• The choice of a suitable, fit-for-purpose RWD cohort is crucial for minimizing bias.
• Hybrid designs show promise for increasing the efficiency of clinical trials in oncology.