Biosimilarity Is a Structured Inference Problem Across Analytical, Functional, and Clinical Evidence

Biosimilarity is commonly described as a totality-of-evidence judgment, but that phrase can obscure a central inferential problem: structural, functional, pharmacokinetic, pharmacodynamic, immunogenicity, and clinical evidence do not interrogate the same uncertainty with equal sensitivity. This Current Opinion argues that biosimilarity is better understood as structured reconciliation across evidence classes. Analytical characterization interrogates product attributes directly; functional assays test whether variation propagates into mechanism-relevant activity; pharmacokinetic and pharmacodynamic studies provide human-level bridges; and immunogenicity and clinical studies address selected residual consequences under defined conditions. Discordance should therefore be localized before escalation. An analytical difference should be interpreted through criticality, function, exposure or response, and clinical context rather than automatically neutralized by a distal endpoint. Conversely, clinical concordance cannot erase unresolved product-level uncertainty when the endpoint is poorly discriminatory. The article proposes an evidential-reconciliation framework in which evidence is weighted by proximity to the uncertainty, sensitivity, mechanistic relevance, independence, convergence, and boundary conditions. The framework is conceptual rather than prospectively validated.