Abstract 158: Spatial heterogeneity of tumor cell states in primary cutaneous melanoma

Abstract Cutaneous melanoma develops from precursor cells that progress to melanoma in situ (MIS) and eventually to invasive disease. In metastatic tumors, cells switch from a melanocytic to an undifferentiated, and neural crest (NC)-like state, which is associated with metastasis and therapeutic resistance. However, it is unclear whether these cell states or gene expression changes occur at the premetastatic stage or if they are associated with features of the local tumor microenvironment (TME). We analyzed a cohort of 62 patients with primary cutaneous melanoma (stages 0-III), using cyclic multiplex immunofluorescence imaging (CyCIF), and spatial transcriptomics to assess the molecular and TME changes that occur during the transitions from normal to precursor lesions, MIS and invasive melanoma. The loss of MART1 was associated with increased NGFR expression and elevated proliferation (KI67) in thicker tumors compared to thin (<1mm) lesions. Various tumor cell states coexisted in close proximity within early-stage, untreated melanomas. These tumor states were not arranged by depth of invasion but were associated with specific immune and stromal cell niches. Using CyCIF, we quantified the spatial heterogeneity of tumor cell states at whole-slide, microregion, and single-cell levels by calculating spatial entropy. We found that samples with a range of spatial entropy values contained both regions of high and low local neighborhood Shannon entropy at the single cell level. High spatial heterogeneity at the protein level was associated with enrichment of interferon gamma and alpha pathway transcripts. We also identified that transcriptomic heterogeneity within a single primary melanoma sample could be just as varied as across the entire cohort, with microregions showing distinct morphologic and spatial features. To understand the gene expression changes leading to intratumoral heterogeneity, we applied a self-organizing map to the spatial transcriptomic data to determine a developmental trajectory from melanocytic atypia to invasion. We discovered that tumor cells in invasive regions followed two developmental branches, characterized by epithelial-to-mesenchymal transition, angiogenesis, interferon response pathways, and dedifferentiation signatures. These developmental branches also showed differences in the local TME composition and the depth of invasion. Our multi-omic analyses reveal that a spectrum of tumor states exists even in early-stage primary cutaneous melanomas, suggesting an association between these tumor states and TME niches. Citation Format: Tuulia Vallius, Yingxiao Shi, Edward Novikov, Ajit Nirmal Johnson, Roxanne Pelletier, Shishir Pant, Zoltan Maliga, David Liu, Guihong Wan, Yevgeniy Semenov, Christine Lian, George Murphy, Sandro Santagata, Peter Sorger. Spatial heterogeneity of tumor cell states in primary cutaneous melanoma [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 158.