P316: Characterizing Regulatory Acceptance of External Control Arms by US FDA and EMA in Oncology
Poster Presenter
Lincy Thomas George
Vice President, Oncology Global Regulatory Affairs
Sanofi United States
Objectives
Describe positioning and regulatory acceptance of external comparators in oncology, detailing strengths and scientific/operational considerations for sponsors.
Method
Eight products that utilized real-world evidence (RWE) as an external comparator were selected to be reviewed for study design and operational considerations. Once selected, a product’s health technology assessments (HTA) in Germany, France and England were evaluated.
Results
For this analysis, eight products that utilized RWE as an external comparator were selected to characterize the factors that led to regulator and payer acceptance. Products approved for oncology were prioritized, but three products were approved for rare diseases. For oncologic products, RWE was used for indications with high unmet patient needs (e.g., lack of standard of care [SOC]) or in rare subpopulations (e.g., rare biomarkers). In our analysis, regulators utilized RWE as part of the product’s benefit-risk assessment in six products. In two products, regulators rejected the RWE.
Of the products analyzed, half used data from electronic medical records (EMR) either from clinical sites or curated EMR databases while others used data from registries, and previous clinical trials. Two of the oncology products, idecabtagene vicleucel (ide-cel) and lisocabtagene maraleucel (liso-cel) supplemented EMR data with data from clinical sites and registries. Use of RWE as primary evidence for a benefit-risk determination was limited by RWD quality. For example, ide-cel used four RWD sources to build a cohort. Differences in follow-up assessment schedules, missing baseline covariates, and more than 90+ captured treatments introduced data heterogeneity and negatively impacted comparability between study populations. As such, the RW study was not used in the final benefit-risk determination by regulators. Conversely, liso-cel utilized similar RWD sources, but conducted a RW study to not only characterize the SOC, but also to characterize differences in outcomes for US and EU patient populations. By limiting use of RWE to contextualization, liso-cel was able to strengthen the rationale for a RW study and ultimately the study was used in the benefit-risk determination.
HTA submissions typically agreed with the findings from regulator assessments with HTA bodies accepting RWE for only one product’s HTA. Guidance for the use of RWE continues to be developed.
Conclusion
Through the case studies, five factors emerged that impacted the acceptability of RWE: context of use; heterogeneity of data; lack of comparability and exchangeability between RW datasets and clinical datasets; missing data; and selection bias. Of these five factors, context of use was the key factor in substantiating use of RWE in evidence packages. The context of indication, unmet need, disease severity, and mechanism of action supported sponsors’ claims that a traditional blinded randomized clinical trial (RCT) could not be conducted. The remaining four factors highlighted limitations in RWD quality as it pertains to missing data, potential for selection bias, and unmeasured confounders in comparative analyses. However, even in case studies, such as eflornithine, where prior RCT data was leveraged for a RW study regulators continued to highlight limitations in accessing data, missing data, and unmeasured confounders. This indicates that even when RWE is generated in scenarios like RCTs or have similar outcome definitions, the evidentiary bar for these studies remains high. Sponsors will need to develop robust justifications that describe the context for the use of RWE and ensure the relevance and reliability of each data source selected if they choose to pursue RW studies.
While three products reviewed through this case study used RWE as part of their primary evidence submitted to regulators, only elivaldogene autotemcel was able to utilize RWE from a registry as an external control arm for a single-arm trial. As such, regulators and payers continue to push RCTs as the primary form of evidence generation, while single arm trials and external controls remain an exception.
