Comparison of biomarkers among 34,855 GI cancer samples show heterogeneity of tumor types.

e14287 Background: Recent data indicate that biomarker driven use of targeted therapy and I/O-therapy among patients with GI cancer is associated with improved outcome. The presence of biomarkers varies broadly between different GI tumor types, highlighting the importance of comprehensive molecular profiling.To analyze the presence of various alterations in GI cancer samples of a large database, comparing congruency between various tumor types. Methods: A retrospective data analysis of 34855 GI cancer patients profiled at CARIS Life Sciences obtained from Jan 1st2010 till Sep 14th2018 was performed. GI tumors were classified as : CRC, esophageal/gastric/GIST, small intestine, pancreatic/hepatobiliary/liver. Technologies used to analyze the biomarkers: IHC for PD-L1, MMR and Her2, and DNA-NGS for EGFR, BRAF, KRAS, NRAS, MET-CNV, TMB, MSI, POLE and BRCA1/2. Results: Median age was 61 (range 18-89 years). 51.5% was CRC (n = 18047), 15.5 % was esophageal /Gastric/GIST (n = 5470), 3% was small intestinal cancer (n = 886) and 30% was pancreatic/hepatobiliary/liver cancer (n = 10452). Information on biomarkers was available from 2931 cases for MET amplification to 28536 for RAS information. Overall, the most common finding was a pathogenic RAS-mutation (either KRAS or NRAS) in 7650 cases (26.81%), the rarest one was a mutation in EGFR in 12 cases (0.06%). Higher rate of HER2 amplification was observed among pts with esophageal/gastric/GIST tumors (6.5%) in comparison to tumors at other GI sites ( ~1.5%). High TMB was seen among patient with CRC and small intestine tumors (~7%) while it was lowest among pancreatic cancer (1.8%). Compared to other GI sites lower MSI /MMR deficiency rates were observed in pancreatic /hepatobiliary tumors, significantly higher PD-L1 positivity was observed in gastrooesophageal cancer types, increased MET-amplifications in gastrooesophageal and small intestinal cancer types and lower RAS-mutation rate in gastrooesophageal cancer. Conclusions: Molecular profiling analyzing druggable biomarkers can help identify patients with increased likelihood for benefit from immune-checkpoint-inhibitors and targeted therapies. Further investigations are needed to evaluate the different findings in various GI cancer types.