Genomic profiling of gastrointestinal cancers by comprehensive tumor-normal sequencing.

Abstract

109 Background: Gastrointestinal cancers (GIC) account for 26% of global cancer incidence and 35% of cancer-related deaths. We investigated the molecular landscape and therapeutic targets across 18 types of GIC using whole exome (WES) and whole transcriptome sequencing (WTS). Methods: GEM ExTra assay was performed on 603 paired samples (ages 18-90 years, median = 61 years). Targeted sequence coverage was 180X for germline DNA and 400X for tumor DNA. Reportable somatic alterations included single base substitutions, indels, Copy Number Alterations, gene fusions, alternate transcripts, as well as tumor mutational burden (TMB) and microsatellite instability (MSI) status. Germline subtraction identified somatic-specific alterations. Results: Analysis of 603 GIC patient samples, including esophageal, gastric cancer (GC), biliary tract (BT), pancreatic cancer (PC), colorectal cancer (CRC), and other cancers, identified 434 actionable targets. The median number of alterations was 3 per GIC patient. The 5 most common actionable alterations were in APC, KRAS, CDKN2A, ARID1A and PIK3CA. Activation of Wnt signaling was found in 264/603 (44%), with the majority being in CRC cases. Alterations in cell cycle genes including CDKN2A, CDK4/6 and others were noted in 129/603 (21%) cases, with the majority in PC, suggesting benefit from CDK4/6 inhibitors. Activation of PI3K/PTEN/Akt/mTOR pathway was noted in 105/603 (17%), with the majority harbored in CRC, suggesting benefit from targeting this pathway. ERBB2 amplification and mutations were noted in 22/603 (4%) across different GIC tumor types. Alterations in homologous recombination genes predicting platinum and PARP inhibitor response was noted in 181/603 (30%) samples distributed across GIC subtypes. KRAS (G12C) mutation was found in 7% of all KRAS mutations across GIC subtypes, thus allowing patients to enroll in clinical trials with G12C-specific inhibitors. The majority of cases with MSI- and TMB-high status were identified in GC and CRC tumors and may be predictive of response to immunotherapy. WTS identified actionable fusions, including FGFR1/2/3 and novel NRG1 fusions in BT cancers. Conclusions: Our study revealed actionable targets used in patient selection for precision therapies, in addition to other mutational profiles of clinical significance. Overall, comprehensive genomic profiling enabled detection of established and novel actionable alterations, including fusions, which may have gone undetected using hotspot panels.