Abstract 5770: Single-cell Raman-based omics of the melanoma tumor-immune microenvironment for precision therapy

Abstract Immunotherapy and targeted therapy have altered the standard of care for cancer patients, improving survival rates and cancer control, even in intransigent metastasis. Yet, only 20-40% of patients respond to immunotherapy, and up to 50% experience severe adverse events, limiting therapy application. Notably, cellular heterogeneity and diversity in the makeup of the tumor immune microenvironment (TiME) present extraordinary challenges in finding predictive biomarkers correlating to therapeutic response. Existing profiling methods, even with highly resolved, cutting-edge transcriptomic and proteomic mapping methods remain costly, time-intensive, and do not faithfully capture structural or post-translational modifications actively impacting tumor behavior. Here, we present a high-fidelity, single-cell omics approach based on Raman spectroscopy and machine learning to provide rapid, cellular profiling that enables non-destructive therapeutic response characterization. Using Raman, we first demonstrate single-cell differentiation and functional state phenotyping of various melanoma and immune cells found in the TiME. Our ML models yield a classification accuracy of >94% across all cell types and parent and derivative cancer lines. We then extend our characterization to track temporal changes across multicellular systems; by investigating cell growth in normal and apoptotic-inducing environments, we show how melanoma-immune cell interactions preserve certain spectral features while introducing new features via cell fusioning under selective pressures. By stimulating cells with small molecule therapeutics serving as growth and inhibiting factors, we demonstrate the potential to monitor melanoma cell plasticity and macrophage polarization changes. Finally, we illustrate the potency of Raman phenotyping by examining the collective drug response in clinically relevant patient melanoma samples and extract pertinent biomarkers associated with small molecule and immunotherapy treatment outcomes. In combination with neural networks, our approach correctly predicts therapy response with >75% accuracy for unseen melanoma patient samples, demonstrating Raman as a powerful rapid, label-free omics approach with groundbreaking potential to advance clinical biomarker efforts and impact the precision treatment decisions of cancer patients. Citation Format: Kai Chang, Mamatha Serasanambati, Baba Ogunlade, Ariel Stiber, Fareeha Safir, Hamish Carr Delgado. Single-cell Raman-based omics of the melanoma tumor-immune microenvironment for precision therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 5770.