Multi-omic characterization of gastrointestinal stromal tumor (GIST) in a large real-world patient cohort.

Abstract

11522 Background: Molecular knowledge of GIST is limited due to its rarity, few genes have been identified as relevant determinants of outcomes, tumor evolution and therapeutic targets. Therefore, we aimed to dissect the GIST molecular landscape in the largest series of real-world patients reported to date. Methods: 946 GIST patient samples (536 localized, 369 metastatic, 41 unknown) underwent next-gen sequencing of DNA (592-gene, N = 495; whole exome, N = 451) and RNA (whole transcriptome, N = 592) at Caris Life Sciences (Phoenix, AZ). Gene expression signatures of proliferation (Cristescu, 2021), cell cycle activation (CINSARC; Chibon, 2010), inflammation (T-cell inflamed; Ayers, 2017) and tumor microenvironments (MCP-counter; Becht, 2016) were examined. Statistical significance tested by χ2, Fisher’s exact, or Mann-Whitney U as appropriate. Results: GIST samples were comprised of 80% (N = 758) KITmut, 8.1% (N = 77) PDGFRAmut, and 11.7% (N = 111) KIT/PDGFRA wild-type (WT), with 14.8% (N = 140) samples harboring a secondary KIT variant suggestive of TKI resistance. Overall median TMB was 2 mutations/MB (range 0-13). WT GIST were identified with mutations in NF1 (33.7%), DNA repair genes (16.7%), SDHX (8.2%), BRAF (6.3%), and PTEN (1.9%), along with NTRK3 fusions (3.1%). Primary KIT variants occurred in exons 11 (83.5%), 9 (13.9%), and 13 (2.6%), and secondary KIT variants (14.6% of total KIT mutations) were distributed across the ATP binding pocket (36.8%) and activation loop (63.2%). Primary PDGFRA mutations were in exons 18 (62.4%), 12 (11.0%), 14 (4.6%) and other (16.5%). KIT/PDGFRAmut GIST infrequently harbored RB1, TP53, SETD2, ARID1A, PIK3CA, PTEN, TSC1, BRCA1, or CHEK2 co-mutations (1-5% each) . Copy number amplification (≥6 copies) was overall uncommon (≤2% for all genes). Proliferation and cell cycle activation signatures were higher in KIT exon 11 indels v. missense mut (1.2-fold, p < 0.05) and KIT resistant v. KIT primary (1.2-fold, p < 0.05), but not in KIT exon 11 557/558 v. others, nor between KIT v. PDGFRA v. WT subgroups. Deletion of tumor progression genes MAX (40.0%), CDKN2A (32.3%), and DEPDC5 (33.9%) was associated with increased proliferative gene expression (1.2-, 1.3-, and 1.2-fold, p < 0.05 each), while DMD deletion (52.5%) was not (1.1-fold, p = 0.37). Compared to KITmut and WT GIST, PDGFRAmut had increased abundance of several immune cell populations (range 1.2-3.7-fold, p < 0.05), along with enhanced inflammation signatures (1.1- and 1-2-fold, p < 0.05). Conclusions: This series provides unprecedented resolution of KIT/PDGFRAmut GIST with features of clinical aggressiveness associated with KIT exon 11 indels and resistance mutations, illustrating a specific cytogenetic genotype with more aggressive growth and malignant behavior. Identification of less common molecular alterations that drive kinase activation and impaired DNA damage repair warrant further investigation.