Real-World Evidence in Biologic Therapies

Abstract

Biologic therapies -- monoclonal antibodies, fusion proteins, and cell-based treatments targeting specific immunological pathways -- have transformed the management of inflammatory, autoimmune, and oncological conditions over the past two decades. Their pivotal trial populations, however, systematically exclude patients with moderate-to-severe comorbidities, prior biologic exposure, and the age extremes most commonly encountered in clinical practice. Real-world evidence from registry and claims data is essential for characterising biologic effectiveness in these under-represented groups, but the observational methods used to generate such evidence vary widely in their ability to address the confounding by indication that shapes biologic prescribing. We evaluated five real-world evidence frameworks applied to 96,284 patients receiving biologic therapy at 24 European rheumatology, dermatology, and gastroenterology centres in Rome, Stockholm, Zurich, Copenhagen, and Warsaw between 2012 and 2023, covering rheumatoid arthritis, psoriatic arthritis, inflammatory bowel disease, and psoriasis. Frameworks ranged from administrative claims analysis through ML-enhanced target trial emulation. Performance was assessed by our Biologic RWE Validity Score (BRVS), integrating causal validity, data completeness, population representativeness, clinical translation utility, and cross-registry replication. ML-enhanced target trial emulation achieved the highest BRVS (0.931), generating effectiveness estimates within 3.8% of trial benchmarks for TNF inhibitors while extending findings to biologic-experienced patients and those with moderate renal impairment excluded from trials. Drug survival -- the proportion of patients remaining on first biologic therapy at 24 months -- was the outcome showing the greatest framework sensitivity, ranging from 48.2% to 72.4% depending on analytical approach, with confounding by indication the primary driver of this 24.2 percentage point range.