A clinical and genomic landscape analysis of GI stromal tumors.

Abstract

11512 Background: Gastrointestinal stromal tumors (GIST) are malignant mesenchymal tumors with distinct histogenesis and behavior from other sarcomas. Treatment of metastatic GIST involves receptor tyrosine kinase (RTK) inhibition. Primary KIT or PDGFRA mutations are established drivers of tumorigenesis via constitutive RTK activation. Growing attention is now focused on how secondary alterations impact clinical outcomes. Methods: The open-source platform cBioPortal was queried for GIST patients. Sociodemographic and clinicopathologic features, including KIT / PDGFRA mutation prevalence, were described using univariate statistics. Primary KIT mutations (exon 11: coding the receptor juxtamembrane domain) and secondary KIT mutations (exon 13/14: coding ATP-binding domain; or exon 17/18: coding activation loop) were assessed. Co-occurrence/exclusivity of altered genes was analyzed using two-sided Fisher’s Exact (p-values) and Benjamini–Hochberg FDR correction (q-values) – pathway pairs deemed significant at p < 0.05 and q < 0.05. Log2 odds ratios (OR) quantified association strength. Differences between KIT -mutated ( KIT -mut) vs. KIT -wild type ( KIT -wt) groups were assessed via chi-squared. Kaplan-Meier modeling evaluated overall survival (OS) by co-occurring alterations, controlling for metastatic status and stratifying by primary site. Results: 499 GISTs were analyzed: 214 gastric (42.9%), 137 small bowel (27.5%), 22 colorectal (4.4%), and 9 esophageal (1.8%).Most tumors harbored KIT mutations (n = 382, 76.6%), which co-occurred with CDKN2A (log2 OR 3.04) and CDKN2B (log2 OR 2.77) deletions (p < 0.001; q < 0.001). 91 (23.8% of KIT -mut) patients harbored both ( CDKN2A/2B -del). Secondary KIT mutations in exons 13/14 or 17/18 were associated with CDKN2A/2B -del (p < 0.001). CDKN2A/2B -del was rare in KIT -wt ( < 5%). KIT mutations were mutually exclusive with PDGFR, NF1 (log2 OR < -3, q < 0.001 for both), and SDHA mutations (log2 OR < -2.67, q = 0.038); prevalent in 9.8%, 4.0%, and 2.0%, respectively. PDGFRA / NF1 mutations showed no significant co-occurrence. SDHA alterations were associated with TERT and RICTOR alterations. GISTs were more likely KIT -wt in women vs. men (OR 1.76, p = 0.012), and in gastric vs. small bowel tumors (OR 2.79, p < 0.001). Tumor mutational burden was similar between groups (mean 2.12 vs. 1.27). No tumors were microsatellite unstable.OS did not differ between KIT -, PDGFRA -, NF1 -, or SDHA -mut (log-rank p = 0.453). In metastatic KIT -mut, CDKN2A/2B -del was associated with poorer OS (log-rank p < 0.001). Median OS for small bowel KIT -mut + CDKN2A/2B -wt disease was not reached, 24.3 months in small bowel KIT -mut + CDKN2A/2B -del, 52.5 in gastric KIT -mut + CDKN2A/2B -wt, and 22.4 in gastric KIT -mut + CDKN2A/2B -del. Conclusions: CDKN2A / CDKN2B deletions co-occur with advanced, imatinib-resistant, KIT -mut GIST. They prognosticate poorer OS but suggest therapeutic potential for cyclin dependent kinase inhibition.