Need Help?
Copied to clipboard!

Proteogenomic Studies to Understand Mechanisms and Drivers of Resistance to Immunotherapies

Immune checkpoint inhibitors (ICIs) targeting programmed death (PD-1) and cytotoxic T lymphocyte-associated protein 4 (CTLA-4) produce durable responses in advanced cutaneous melanoma, yet reliable predictive biomarkers remain elusive. We performed a multi-cohort proteogenomic investigation of pre-treatment formalin-fixed paraffin-embedded (FFPE) biopsies from patients with metastatic cutaneous melanoma who later underwent either dual PD-1+CTLA-4 ICI therapy or PD-1 ICI monotherapy. Melanoma metastases of the lung, lymph node, and skin from the 3 cohorts were collected from 12 study sites in the United States, Europe, and Australia, comprising two separate PD-1+CTLA-4 ICI-treated discovery and validation cohorts ("Discovery Cohort" and "Validation Cohort 1", n=164 samples), and one PD-1 ICI monotherapy-treated validation cohort ("Validation Cohort 2", n=44 samples).

Comprehensive genomic profiling was performed across the Discovery and two Validation Cohorts (across 180 individual samples) using Ribo-Erase total RNA sequencing (RNA-seq) (data generated for 173 samples), low-pass whole-genome sequencing (lpWGS) for the purpose of detecting somatic copy-number alterations (data generated for 177 samples), and a custom melanoma-focused targeted DNA sequencing panel (comprised of 86 genes, ~0.75Mb) designed to maximize the sensitivity of detecting recurrent and clinically actionable mutations (data generated for 175 samples). Pathway-level proteomic clustering identified three tumor subtypes; ICI-sensitive (immune-inflamed), ribosome-high (translationally active), and epithelial-mesenchymal transition (EMT)-like (stromal-enriched) that were reproducibly validated and showed significant PFS differences (p = 0.02–0.03) across cohorts. Subtype–outcome associations were consistent with previous immunotherapy datasets, with pooled hazard ratios ~0.5–0.8 for immune-enriched tumors. Additionally, we identified a 127-protein classifier that outperformed previously reported genomic signatures, suggesting that proteomic subtyping captured clinically relevant biology not accessible via genomics or transcriptomics alone. Our results identify potential new therapeutic vulnerabilities and lay the foundation for biomarker-driven patient stratification in combination ICI trials. Proteomic data will be made available through the National Cancer Institute's Proteomic Data Commons repository.