Predicting targeted- and immunotherapeutic response outcomes in melanoma with single-cell Raman Spectroscopy and AI
Journal:
bioRxiv
Published Date:
Mar 12, 2026
Abstract
Identifying predictive biomarkers of immunotherapeutic response in melanoma remains an outstanding challenge. Existing transcriptomic and proteomic profiling methods of the tumor-immune microenvironment are costly and may not faithfully capture modifications actively impacting tumor behavior. Here, we present a non-destructive, single-cell approach combining Raman spectroscopy and machine learning (ML) that enables rapid cell profiling and therapeutic response prediction. We tested mouse and human melanoma cell lines alongside nine melanoma patient-derived samples. Each sample had known resistance profiles to a panel of targeted and immunotherapeutic inhibitors, including bemcentinib, cabozantinib, dabrafenib, nivolumab, and a combination of nivolumab and relatlimab. In cell lines, our single-cell Raman and ML approach achieved >96% differentiation accuracy across tumor microenvironment cell types and functional phenotypes. Formations of subclusters for persistent (e.g. drug-resistant) cells were observed based on genetic mutations rather than sample origin, with Raman signatures reflecting biochemical changes relevant to various therapeutic pathways. For patient samples, we constructed a two-stage evaluation workflow to assess clinical drug resistance. Using therapy-specific random forests, our workflow correctly inferred resistance likelihoods for 30 of 33 clinically relevant patient-drug combinations (91% accuracy) unseen by our model with optimized labeling thresholds. Our scalable, prognostic model using single-cell Raman offers potential to advance clinical, multi-omic biomarker efforts and impact first- and second-line therapy selection assessments for precision medicine.
