(A-046) A Study of Granulocyte-Colony Stimulating Factor (G-CSF) Reference and Biosimilar Products using Real World Data to Emulate a Target Trial for Switching: A Biologics and Biosimilars Collective Intelligence Consortium Experience

Background: US biosimilar interchangeability requires additional evidence indicating no increased safety risk or loss of efficacy when switching between a reference biologic and biosimilar. Sufficiently powered clinical trials to examine switching are often infeasible. The 20th Century Cures Act requires the FDA to identify how real-world data (RWD) could be incorporated into regulatory decision-making. One promising application is demonstrating interchangeability. A major obstacle is understanding RWD quality and relevance for these decisions.

Objectives: To assess the extent to which RWD (electronic health records and administrative claims) are suitable to emulate clinical trials targeting biosimilar interchangeability.

Methods: We conducted a target trial emulation of pegfilgrastim in breast cancer to assess RWD availability and completeness for FDA regulatory tests for interchangeability across three research partners: a large, national multi-payer claims database, a national insurer, and an integrated healthcare system. A hypothetical trial was designed based on a switching study evaluating filgrastim products, modified to assess pegfilgrastim products for febrile neutropenia (FN) prophylaxis and adverse events within the first 6 chemotherapy cycles. Variables included chemotherapy regimen, administration, and discontinuation, pegfilgrastim utilization, switching, and discontinuation, adverse events, and laboratory results. Sites independently examined outcomes among switchers between reference pegfilgrastim and biosimilars compared with non-switchers to assess the ability to inform interchangeability. Outcomes for switching and non-switching patients were compared following inverse probability of treatment weighting (2 sites).

Results: Across data sources, we identified 17,964 patients (99% female) with breast cancer receiving high or intermediate FN risk chemotherapy treated with prophylactic pegfilgrastim. Most (76.2%-86.6%) initiated reference pegfilgrastim. Overall, 555 patients switched between reference pegfilgrastim and an available biosimilar. FN event rates were low with some variability (1.6%-3.5%) across sources. All sources readily identified patients, treatment, and demographics. Some variability in data availability across sources exists, particularly in laboratory results and some disease measures. Outcomes did not differ for switching and non-switching patients.

Conclusions: We demonstrated three independent data sources successfully conducted this emulation study with similar results. Strengths and limitations of different RWD sources led to variations in analytic approaches that should be considered when designing real-world studies.