Metastatic melanoma remains a lethal malignancy, with durable responses achieved only in a subset of patients despite the advent of immune checkpoint blockade. While genomic alterations in melanoma have been extensively catalogued, the proteomic architecture of advanced disease—and particularly the degree of molecular divergence among multiple metastases within the same individual—has not been systematically explored.

We assembled a unique rapid-autopsy cohort comprising 83 distant metastases derived from 24 patients, representing 19 anatomical sites and largely predating the routine use of immunotherapy. Each lesion underwent standardized histopathological assessment, transcriptome profiling by RNA sequencing, and quantitative mass spectrometry–based proteomics. Proteogenomic integration enabled the identification and validation of non-reference peptides corresponding to expressed sequence variants.

Across the cohort, mutation-associated non-reference peptides were recurrently detected in proteins implicated in focal adhesion, vesicular trafficking, MAPK signalling, and immune regulatory pathways. Conventional parameters—including morphology and canonical driver mutations—showed limited variability within individual patients. In contrast, integrated proteogenomic analysis uncovered pronounced intrapatient heterogeneity at the protein expression and variant level. This divergence was largely independent of metastatic site, although liver metastases displayed a distinct metabolic signature. Unsupervised clustering defined four major proteomic phenotypes: (C1) enrichment of cell cycle and RNA processing pathways; (C2) extracellular matrix and immune-related signalling; (C3) mitochondrial and metabolic programs; and (C4) extracellular matrix and vesicle-mediated transport processes. Approximately two-thirds of patients harboured metastases spanning markedly different proteomic states. Clinically, patients whose lesions predominantly mapped to C1 or C3 experienced inferior overall survival compared with those enriched for C2 or C4 phenotypes.

This multi-lesion proteogenomic analysis of immunotherapy-naïve metastatic melanoma reveals substantial intrapatient molecular divergence that is not apparent at the genomic or histological level. The findings underscore the complexity of metastatic evolution and provide a framework for identifying context-specific therapeutic vulnerabilities that may enhance current treatment strategies.

EU's Horizon 2020 Research and Innovation Program with [No. 739593], the Ministry of Culture and Innovation of Hungary from the National Research, Development and Innovation Fund, TKP2021-EGA funding scheme [TKP-2021-EGA-05 ], and, 2022-2.1.1-NL funding scheme, [2022-2.1.1-NL-2022-00005].