Molecular Attribution and Assay Replacement Readiness of Multi-Attribute Methods in Biosimilar Development

Background/Objectives: Multi-attribute method (MAM) provides early molecular attribution in biosimilar development, but assay replacement requires separate evidence. We evaluated MAM integration and replacement readiness in two IgG1 biosimilar programs. Methods: Conventional charge and released-glycan profiles of US-licensed and EU-approved reference product (RP) lots were compared with site-specific MAM measurements. Carboxypeptidase B (CpB) treatment, orthogonal species characterization, intact-mass analysis, and evaluation across alternative analytical configurations were used to assess attribution, coverage, and replacement readiness. Results: MAM assigned specific contributors within conventional charge and glycan profiles, whereas intact methods retained information beyond monitored attributes. For mAb-Cys, 5–7 percentage points of the acidic cation exchange chromatography (CEX) region remained unassigned, and candidate hinge-clipped and non-native disulfide species could not be quantified as contributors to RP residuals. In both molecules, MAM C-terminal lysine values were approximately twice the per-chain occupancy implied by CpB-responsive CEX measurements, demonstrating that peptide- and intact-level measurements are not numerically interchangeable. Across analytical configurations, C-terminal lysine was consistently quantifiable, but reported levels differed; glycan-class reporting depended on processing software whereas deamidation depended on both software detection and chromatographic separation. Common-software reprocessing enabled full-panel quantitation in the evaluated datasets. Conclusions: MAM should be integrated early to provide molecular attribution, but attribution alone does not establish assay replacement readiness. Replacement requires three evidence pillars: (1) reconciliation with conventional measurements and coverage of relevant process-related species, (2) MAM-specific RP distributions, and (3) complete attribute coverage with fit-for-purpose performance on the intended integrated analytical configuration.