# Genomic Landscape of Poorly Differentiated Gastric Carcinoma: An AACR GENIE® Project

## TL;DR

This study explores the genomic features of poorly differentiated gastric carcinoma, identifying key mutations and pathways that could guide future precision treatments.

## Contribution

The paper provides a detailed genomic characterization of poorly differentiated gastric carcinoma using the AACR GENIE® database.

## Key findings

- TP53, CDH1, ARID1A, and CCNE1/FGFR2 amplifications are key genomic features in poorly differentiated gastric carcinoma.
- Co-occurrence of POLD1, ARID1A, and KMT2D suggests disruption of DNA repair and epigenetic regulation.
- Metastatic tumors show enrichment of CDH1 and MLH1, indicating roles in invasion and resistance.

## Abstract

Poorly differentiated gastric carcinoma (PGC) is aggressive, yet subtype-specific genomics are under-characterized. We queried AACR Project GENIE® (cBioPortal v18.0-public; 12 August 2025) for PGC and analyzed somatic alterations from targeted panels (depth ≥ 100×; variant allele frequency ≥ 5%). Mutation and copy number frequencies were summarized, co-occurrence and exclusivity were tested, and primary versus metastatic tumors were compared using chi-square with Benjamini–Hochberg correction. The cohort included 189 tumors from 188 patients (71% primary; 25% metastatic), with primary and metastatic tumor samples being collected from different patients. Recurrently mutated genes were TP53 (48.7%), CDH1 (31.2%), ARID1A (21.2%), KMT2C (8.5%), and POLD1 (7.4%); additional alterations involved ERBB3, KMT2D, KEL, CDKN2A, and FAT1 (≈1–7%). Amplifications in CCNE1 (8.2%) and FGFR2 (7.6%) were common, alongside gains in MET, MYC, KRAS, and ERBB2 and losses in CDKN2A/CDKN2B, CDH1, and PTEN. Significant co-occurrence was observed for POLD1–KMT2D (p < 0.001), POLD1–ARID1A (p < 0.001), and ARID1A–KMT2D (p < 0.001), while TP53 was mutually exclusive with ARID1A (p = 0.029) and CDH1 (p = 0.041). CDH1 (48.9% vs. 29.6%; p = 0.021) and MLH1 (8.5% vs. 1.5%; p = 0.040) were enriched in metastases, and CCNE1 alterations showed female predominance (p = 2.83 × 10−4). Several “primary-only” findings likely reflect small denominators and require replication. PGC demonstrates a mutational framework dominated by TP53, CDH1, ARID1A, and recurrent CCNE1/FGFR2 amplifications, underscoring dysregulation of cell cycle and chromatin-remodeling pathways as key drivers. Co-occurrence of POLD1, ARID1A, and KMT2D suggests coordinated disruption of DNA repair and epigenetic regulation, whereas mutual exclusivity of TP53, ARID1A, and CDH1 indicates distinct tumorigenic routes. Metastatic enrichment of CDH1 and MLH1 supports their roles in invasion and therapeutic resistance. Together, these findings highlight candidate biomarkers and actionable pathways warranting validation in larger, multi-omic cohorts to refine precision treatment strategies for this aggressive gastric cancer subtype.