(007) Beyond external control arms: utility of real-world data in oncology regulatory applications

Background: For approvals of oncology therapies in the US, many of which are based on single-arm trial results, real-world external controls have not contributed to substantial evidence of effectiveness (unlike other therapeutic areas). As many authors have noted, regulator concerns over external control arms are largely related to insufficiently comparable data collection between trial protocols and routine clinical practice/real-world data (RWD) sources. However, case studies show that RWD support oncology regulatory applications in other ways.

Objectives: To summarize learnings from use of RWD in oncology applications, beyond external control arms for comparison with single-arm trials.

Methods: We reviewed all publicly available Food and Drug Administration (FDA) Center for Biologics Evaluation and Research and Center for Drug Evaluation and Research approvals of oncology products from 2019 to 2023. A structured abstraction form captured information on real-world data (RWD) use described in approval documentation, including the type of RWD included (previously published RWD analysis, which may or may not have been done by the sponsor, vs. new RWD analysis) and the intended use of RWD (to support safety, effectiveness, or to provide contextual evidence). For RWD that were used to support effectiveness, we recorded how the data were used (external control arm, external control benchmark, or both treated and comparator groups sourced from RWD) and whether FDA explicitly referred to the RWD as contributing substantial or supportive evidence.

Results: Nearly all (93%) approval packages included RWD. Most commonly, previously published RWD provided contextual evidence (primarily to provide background on the disease and treatment options). Previously published RWD was also used to inform endpoint thresholds in the pivotal trial design (e.g. belantamab mafodotin-blmf (Blenrep)). Less commonly, new RWD analysis was used, primarily to support effectiveness via external control arms or benchmarks (though none of these contributed to substantial evidence for FDA approval). Our review of use cases demonstrates that RWD can provide important contextual evidence. For example, tisotumab vedotin-tftv, granted accelerated approval in 2021 for recurrent or metastatic cervical cancer, was supported by SEER data and a claims-based real-world study showing that the pivotal single-arm trial population was comparable to the general indicated population in terms of baseline disease and treatment characteristics. Another example is capmatinib, granted accelerated approval in 2021, where a global chart review study describing the natural history of advanced mesenchymal-epithelial transition (MET)-mutated non-small cell lung cancer provided an external control benchmark, but also demonstrated unmet need.

Conclusions: Using real-world external control arms for comparison of effectiveness endpoints has not contributed to substantial evidence in oncology FDA approvals. However, our review demonstrates that RWD on the baseline characteristics, treatment patterns and disease natural history of the indicated population is valuable to various needs in applications. These uses include rationalizing trial design (e.g., sample size requirement) and accelerated approval (e.g., unmet need with current treatments) and supporting trial interpretation (e.g., representativeness and external control benchmark), which may ultimately increase the likelihood of approval.