P-140-R: From Spectrum to Validated Risk Call: Integrating Automated LC-MS Host Cell Protein Analysis with Peer-reviewed Risk Intelligence

LC-MS has become a routine analytical approach for residual host cell protein (HCP) characterisation in biopharmaceutical products, with modern instrumentation regularly identifying hundreds of HCPs in a single bioprocess sample. The analytical bottleneck has shifted downstream of the spectrum: detection is broadly solved, but triaging identified proteins by risk, justifying those calls with peer-reviewed evidence, and producing audit-ready documentation remain predominantly manual tasks. The introduction of USP General Chapter <1132.1>, effective May 2025, has made regulatory expectations on MS-based HCP analysis explicit. Reviewers expect manual peptide-level inspection of raw, MS, and MS/MS evidence, traceable rationale for risk calls on low-abundance HCPs, and citation-backed impact assessments. Operationalising this guidance across analyst teams is non-trivial, and the handoff between automated identification tools and curated risk databases has historically relied on spreadsheets, manual lookups, and ad hoc citation searches. We describe a workflow integration between SpotMap MS, an automated LC-MS HCP analysis platform implementing a Protein Verification Loop aligned with USP <1132.1>, and the BioPhorum HCP Data Platform, a peer-reviewed risk and impact database curated by an industry Scientific Expert Review Committee. The two systems are connected via an authenticated API: every protein identified and Hi3-quantified by SpotMap MS is cross-referenced against the BioPhorum database at analysis time. Risk classification, impact category (product quality, formulation, biological activity, immunogenicity), and citation lists are returned per HCP and embedded directly in the analytical report, with a full 21 CFR Part 11 audit trail. Initial benchmarking on CHO-derived monoclonal antibody harvest matrices indicates substantial reductions in time spent per protein on manual cross-referencing. We present the technical architecture, illustrative output from a repr