Abstract 4972: KIT exon 9 versus exon 11 mutations imprint distinct spatial and phenotypic immune microenvironments in gastrointestinal stromal tumors

Abstract Background: The tumor microenvironment (TME) is a key determinant of progression and treatment response in gastrointestinal stromal tumors (GIST). Oncogenic KIT mutations in different exons (exon 9 vs exon 11) are clinically relevant, but their impact on immune and stromal architecture in situ remains insufficiently understood. We hypothesized that KIT exon 9 and exon 11 mutations drive distinct spatial transcriptional programs and immune cell compositions. Methods: FFPE samples from KIT-mutated GIST (exon 9 and exon 11) were profiled using the NanoString GeoMx® Whole Transcriptome Atlas (18,695 protein-coding genes). Eleven areas of illumination per case were selected in tumor and peritumoral regions and segmented into CD45+ (immune-enriched) and CD45− (tumor/stromal) compartments. Cell-type abundances in CD45+ segments were estimated with SpatialDecon. In parallel, three multiplex immunofluorescence (mIF) panels (DOG1/CD11c/CD11b/CD68/CD45/HLA-DR; DOG1/CD45RO/CD56/CD4/CD8/CD20; DOG1/CD45/CD16/CD56) were applied to independent retrospective GIST cohorts (n=170), and cell densities were quantified in tumor and stromal compartments using QuPath-based segmentation and FACS-like phenotyping. Results: GeoMx analysis identified 2315 differentially expressed genes in CD45+ tumor regions between exon 9 and exon 11 tumors, indicating exon-specific immune transcriptional programs. SpatialDecon deconvolution revealed that exon 11-mutant tumors exhibited higher NK-cell scores (adjusted p≈6×10−6), whereas exon 9-mutant tumors showed higher fibroblast (adjusted p≈6×10−6) and CD8+ memory T-cell signals (adjusted p≈1.9×10−4). In the large mIF cohorts, unsupervised analyses of DOG1-gated tumor regions confirmed robust technical performance of all three panels and demonstrated genotype-segregated TME patterns KIT-mutant cases, and within them exon 9 vs exon 11 tumors, occupied distinct regions in PCA space, with exon 11 tumors showing higher CD56+ (NK/NKT) and myeloid (CD11b+/CD68+/HLA-DR+) densities, whereas CD8+CD45RO+ memory T cells were more abundant in exon 9 tumors, independently of PDGFRA-mutant GIST. Across panels, these phenotypic differences aligned with the GeoMx-derived enrichment of NK cells in exon 11 tumors and of and CD8+ memory T cells in exon 9 tumors. Conclusions: Integrated spatial transcriptomics and multiplex IF demonstrate that KIT exon 9 and exon 11 mutations are associated with distinct immune and stromal microenvironments in GIST, characterized by an NK-cell-enriched, myeloid cells , less fibroblast- and CD8+ memory T-cell-dense niche in exon 11 tumors and the converse pattern in exon 9 tumors. These data support the concept that the precise KIT exon mutated imprints the TME and should be considered when designing genotype-adapted combinations of kinase inhibition and immunotherapy. Citation Format: Jean-Philippe Guegan, Jean-Michel Coindre, Carine Ngo, Antoine Bougoüin, Christophe Rey, Catherine SAUTÈS-FRIDMAN, Wolf Herve Fridman, Armelle Dufresne, Alban Bessede, Antoine Italiano. KIT exon 9 versus exon 11 mutations imprint distinct spatial and phenotypic immune microenvironments in gastrointestinal stromal tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4972.