Integration of spatial single-cell proteomics and spatial metabolomics reveals tumor microenvironment predictive of immunotherapy response in mucosal melanoma

Mucosal melanoma (MuM) is a rare but aggressive malignancy with limited benefit from immune checkpoint inhibition and few predictive biomarkers. We integrated single-cell spatial proteomics (COMET) and spatial metabolomics (MALDI-IMS) to profile 97 tissue cores from 26 patients treated with PD-1/PD-L1 and/or CTLA-4 inhibitors. We profiled 695,444 cells and resolved 25 cell states across eight major cell types. Cellular neighborhood (CN) analysis revealed distinct tumor- and stromal-associated spatial architectures. Responders were enriched for tumor-associated CNs (invasive tumor and tumor boundary) with close spatial proximity among Ki67⁺ tumor cells, CD163⁺ macrophages, and CD11c⁺ dendritic cells (DCs), and increased proliferating/cytotoxic CD8+ T-cell subsets. Non-responders showed stromal CN dominance with reduced immune infiltration. Spatial metabolomics identified lower abundance of indole-derived metabolites and reduced indole/tryptophan pathway activity in responders within tumor and TME regions that tracked with DC/macrophage-enriched spatial contexts. This study advances MuM spatial biology and provides a framework for biomarker-driven immunotherapy strategies. Statement of significance: Mucosal melanoma (MuM) responds poorly to immune checkpoint blockade, and predictive biomarkers are limited. Integrated spatial proteomics and metabolomics reveal response-associated tumor-immune neighborhood architecture, stromal contexts linked to immune exclusion, and altered indole/tryptophan metabolism in the microenvironment. These spatial features nominate biomarkers and therapeutic hypotheses to improve immunotherapy for MuM.