Molecular profiling of a gastroesophageal adenocarcinoma (GEA) multicohort using next-generation sequencing (NGS).

Abstract

494 Background: The mutational landscape of GEA is rapidly evolving with the identification of promising molecular targets. However, histological subtypes are insufficient to capture the molecular heterogeneity in these patients (pts). We aim to assess the histopathological and molecular features in a GEA cohort using NGS techniques, and to evaluate the mutational frequency and its prognostic correlation. Methods: We conducted a retrospective study on GEA from 2019 to 2024. Immunohistochemistry (IHC) was used to evaluate HER2 expression, microsatellite instability (MSI), Epstein-Barr virus (EBV) and PD-L1 status. In-house and commercial NGS platforms were used to detect molecular alterations from tumor samples. Cox regression model was used to analyze overall survival (OS) based on the clinicopathological and molecular subgroups. Results: Using NGS techniques, we identified the following alterations with potential clinical relevance in the 115-pts included: pathogenic mutations in PIK3CA (5.2%), BRCA1/2 (3.5%) and POLD1 (0.9%), along with copy number alterations (CNA) in ERBB2 (10.4%), EGFR (8.7%), MET (5.2%) and FGFR2 (4.3%) genes, among others. Additionally, a high tumor mutational burden (≥ 13 Mut/Mb) was observed in 6 pts (5.2%). Clinical characteristics are detailed in the table. We detected a higher rate of alterations in CDH1 (10.42% vs 4.48%; p=0.22) and PIK3CA (8.33% vs 2.99%; p=0.20) genes in early (<50y) vs late-onset (≥50y) GEA. Additionally, we identified a significantly higher mutation frequency in diffuse vs intestinal tumors; ARID1A (17.6% vs 2.2%; p=0.01) and CDH1 (11.8% vs 0%; p=0.02) genes. Also, pts with peritoneal involvement showed a greater prevalence of mutations in CDH1 (12% vs 0%; p=0.04) and RHOA (10% vs 0%; p=0.06) compared to those with liver disease. We observed a significantly worse OS in pts with diffuse vs intestinal tumors (15.7m vs 19.1m; p=0.04) and a trend toward shorter survival in pts with CDKN2A mutations (11.97m vs 17.83m; p=0.42). Although not significantly, CNA in ERBB2 were associated with a better prognosis (22.2m vs 16.2m; p=0.09). Conclusions: Our results support the utility of NGS platforms in detecting novel molecular targets with prognostic and clinical impact, beyond ERBB2 . Additionally, we observed poorer prognosis in diffuse subtype tumors, which exhibited a higher frequency of mutations in ARID1A and CDH1 genes. Clinicopathological characteristics of the 115-pts with GEA. N=115 (%) AgeMedian years [range] 54 [18-83] Sex Male 69 (60.0) Tumor site Gastric 72 (62.6) Gastroesophageal junction 37 (32.2) Esophagus 6 (5.2) Histological subtype Diffuse or pooly cohesive 51 (44.3) Intestinal or tubular 46 (40.0) Others 18 (15.7) Stage at diagnosis Advanced 74 (66.9) Localized 41 (33.1) Metastatic site Peritoneum 53 (46.1) Liver 34 (29.6) Molecular profile (IHC) HER2 (+) 20 (17.4) MSI 5 (4.3) EBV (+) 2 (1.8) PDL1 (CPS) ≥1 35 (30.4)