Abstract 6256: Dissecting tumor-intrinsic programs and immune dynamics driving response to immunotherapy in metastatic melanoma through non-negative matrix factorization decomposition

Background: Understanding the biological mechanisms underlying differential responses to immunotherapy is critical for advancing treatment strategies in metastatic melanoma. We analyzed single-nucleus RNA sequencing data from 66 melanoma samples, including 18 with matched spatial transcriptomics data via MERFISH, across metastatic sites such as lymph nodes, colon, skin, lung, brain, and liver. Our study aimed to identify tumor-intrinsic transcriptional programs and their association with durable clinical benefit. Methods: We employed non-negative matrix factorization to delineate distinct tumor programs within tumor cells. To assess the robustness and biological relevance of these programs, we conducted hallmark pathway analyses and utilized spatial transcriptomics to examine their associations with biopsy sites, response, and immune infiltration. Comparisons with published datasets were performed to characterize and ensure the stability and applicability of these programs. Results: We identified five distinct tumor programs with differential enrichment between responders and non-responders. Tumor programs enriched in responders demonstrated strong associations with interferon alpha-beta signaling and antigen presentation pathways. This program was also enriched across diverse lymphocyte populations. Via MERFISH analysis, we classified patients into three categories based on the spatial distribution of immune cells relative to tumor borders: immune-enriched, immune-excluded, and immune-desert. Notably, the program enriched in responders was prevalent in the immune-excluded samples, suggesting that robust signaling pathways can drive therapeutic responses even in the absence of high immune infiltration. In contrast, a mesenchymal-like tumor program was predominant in non-responders, characterized by elevated epithelial-mesenchymal transition activity and diverse tumor-associated macrophage infiltration. This program was particularly enriched in colon metastases. Conclusion: Our findings underscore the importance of interferon signaling in driving therapy response and the role of mesenchymal-like programs in resistance. The unique association of immune-excluded but therapy-responsive samples with interferon-related pathways raises intriguing questions about tumor-immune dynamics. By integrating single-cell and spatial transcriptomics, this study provides insights into tumor heterogeneity and the tumor microenvironment, offering avenues for targeted therapeutic interventions. Future research will focus on spatial interactions, the characterization of immune subpopulations within tumor programs, and the dynamics of cell-cell communication between tumor and immune cells, aiming to refine and advance personalized treatment strategies. Citation Format: Priyanka Solanky, Giuseppe Tarantino, Yiwen He, Xinyu Cui, Kathleen Pfaff, Aaron R. Thorner, Tyler J. Aprati, Boyang Zhang, Timothy Blosser, Jiajia Chen, Junko Tsuji, Elliot Boblitt, Allison Frangieh, Hannah Mae Faulkner, Marta M. Holovatska, Aleigha R. Lawless, Michael Manos, Emily Robitschek, Karla Helvie, Tatyana Sharova, Dennie Frederick, James Liam Fahey, Diego Villamarin, Sachi Krishna, Chanell Mangum, Ajit J. Nirmal, Domenic Abbondanza, Cai McCann, Bruce Johnson, Alex Shalek, Eliezer Van Allen, Xiaowei Zhuang, Ryan J. Sullivan, Barbara Engelhardt, Sami Farhi, Scott J. Rodig, Stephen Hodi, Genevieve Boland, David Liu. Dissecting tumor-intrinsic programs and immune dynamics driving response to immunotherapy in metastatic melanoma through non-negative matrix factorization decomposition [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 6256.