(P 24) COMPREHENSIVE ANALYSIS OF THE GENOMIC LANDSCAPE AND CLINICAL BEHAVIOR IN GIST USING NEXT GENERATION SEQUENCING (NGS, MSK-IMPACT)

Objective: The relationship between the genomic profile in GIST and clinical behavior is relatively unknown. The objective of this study was to describe the genomic landscape in GIST and explore the association with tumor stage, treatment status and clinical outcome

Methods: GIST patients (pts) that underwent NGS using MSK-IMPACT of archival tumor and matched normal DNA, in a prospective IRB-approved study were identified. Clinicopathologic data was collected, and a single sequenced sample was selected based on the most recent date of procurement and highest purity. Descriptive statistics were used to report clinicopathologic attributes, somatic, germline and copy number alterations (CNA), tumor mutational burden (TMB), and fraction of genome altered (FGA). Fishers exact, Chi-squared, or Wilcoxon rank sum tests, logistic regression and Kaplan-Meier analyses were performed as appropriate using R version 4.2.2.

Results: As of April 2020, 491 samples from 411 GIST patients were sequenced. (Table 1) The analysis included 254 primary tumor samples (15% treatment-exposed [TE]) and 157 metastatic/local recurrence samples (86% TE). Median (m)TMB was 1.80. mFGA was 0.19. NGS classified patients into six molecular subtypes [KIT- (76%; exon 11 - 64%, exon 9 - 8.5%, exon 13/14 - 2.4%, exon 17/18 - 1.7%), PDGFRα- (9.7%; D842V mutant- 5.8%), SDH-deficient (7.1%), NF1-related (4.6%), Wild type (1.9%), BRAF-mutant (0.2%)]. In KIT-mutant GIST (n=314) the most frequently altered genes included CDKN2A/B (29%), RB1 (8.0%), CSDE1 (7.0%), MGA (5.7%), SETD2 (5.4%), TP53 (5.1%), MAX (4.8%), TNFRSF14 (3.8%), PTEN (3.5%), GATA3 (3.2%), TERT (2.9%), TSC1 (2.9%), PDGFRA (2.5%), PIK3CA (2.5%). Additional gene mutations (muts) were infrequently observed in other subtypes except TP53 (17%) and TERT (10%) in SDH-deficient and CDKN2A/B (7.5%) in PDGFRα-mutant GIST. A significantly higher frequency of disruption in cell cycle, epigenetic, MYC and PI3Kinase pathways was observed in metastatic compared to primary tumor samples. Muts in CDKN2A/B, RB1, TERT, MGA and TSC1 were significantly more common in TE samples. Co-occurring CDKN2A/B, MAX, TP53, and PIK3CA muts were associated with a poor prognosis in the metastatic setting. In KIT-mutant, TE samples, secondary KIT exon 13&14 and 17&18 alterations were observed in 19% % and 26% of patients. Secondary KIT muts significantly co-occurred with KIT exon 11-inframe deleted GIST but not -missense mutated GIST. In TE samples, KIT exon 17/18 muts (31.9%) were more common in inframe deleted GIST and exon 13/14 muts (15%) were more common in missense mutated GIST. (Fig. 2) Unsupervised clustering was done using CNA segmentation among the entire population. Six distinct genomic clusters emerged characterized by CNAs involving specific genes, clinicopathologic and molecular patterns. (Fig.3)

Conclusion: This study represents a one of the most comprehensive analyses of the genomic landscape and integration of genomic features with clinicopathologic attributes and survival outcome in GIST.