(SLA-037) Can observational studies replace experimental transition (or switching) studies for evaluating the interchangeability of biological drug reference and biosimilars? Methodological challenges and considerations

Background: European Medicines Agency (EMA) and Food and Drug Administration (FDA) have established regulatory frameworks for biosimilar approval, although key differences exist. EMA states biosimilars are interchangeable with the respective reference product; conversely, FDA requires a separate designation to grant interchangeability of biosimilars upon additional evidentiary requirements that may include ad-hoc randomized clinical trials (RCTs). Due to practical challenges, however, such trials have been rarely conducted. The 20th Century Cures Act in the US requires the FDA to identify how real-world evidence (RWE) could be incorporated into regulatory decision-making. One potential application is in demonstrating biosimilar interchangeability. A promising approach is conducting observational studies applying principles from RCTs, by a target trial emulation (TTE) in large-scale databases. Currently, no TTE for assessing the interchangeability of reference products and biosimilars is available in the literature.

Objectives: To explore methodological challenges for conducting TTE studies for specifically evaluating the effects of switching from reference biologics to biosimilars using population-based healthcare databases from Denmark, Italy, and the US in the context of FDA-funded international collaboration.

Methods: We designed a TTE on biosimilar interchangeability using different population-based healthcare databases. Two different test cases were selected: 1) switching from insulin glargine reference and biosimilars in three settings; 2) switching from adalimumab reference and biosimilars in Italy and Denmark. The TTE protocol elements for the three settings were specified, and the main analytical considerations and challenges were discussed.

Results: The following considerations emerged from the test cases: a) only subjects who stay in stable treatment with reference product for a prespecified time period should be included; b) assigning time zero among non-switchers can be complex and different approaches can be applied, such as the clone-censor-weight approach, the active comparator paradigm, and the new-switcher design; c) for causal contrast, only the “as-treated” approach must be considered; d) analytic approaches including propensity scores, inverse probability weighting, and time-varying Cox model approach can be considered.

Conclusions: Through this international collaboration, we provide the application of strategies to perform a TTE on switching studies. This approach could facilitate timely investigation of future research across different settings. Finally, findings from data outside the US with this study design may have the potential to inform FDA for novel ways to apply RWE to regulatory decisions.