Abstract Background: Chromosomal instability (CIN) is a dominant molecular feature of gastroesophageal adenocarcinomas and is associated with poor outcomes to standard chemotherapy and chemo-immunotherapy. Given its high prevalence and the absence of standardized, clinically deployable methods to identify CIN, we sought to develop a unified clinical and molecular framework to quantify CIN across real-world and phase III datasets and to build NGS- and histopathology-based tools for routine CIN classification. Methods: We analyzed 3,822 tumors across complementary platforms: 2,613 patients profiled with MSK-IMPACT, 587 RNA-seq and whole-exome sequencing samples from patients treated on Phase III CheckMate-649 (CM649) trial, and 622 TCGA samples. CIN was defined by fraction of genome altered (absolute log2 deviation 0.2) and complemented by arm-level and focal copy-number metrics using FACETS. Clinicopathologic, molecular, and outcome associations were evaluated across CIN strata. CIN will be integrated with CM649 transcriptomic programs and survival to assess predictive value for chemotherapy with and without nivolumab. Results: CIN was present in 63% of tumors within the MSK-IMPACT cohort, including 76% in esophageal, 68% in GEJ, and 47% in gastric cancers. Among 2,445 microsatellite-stable tumors, CIN tumors were enriched for TP53 mutations (79% vs 64%), CDKN2A loss (24% vs 10%), and focal amplifications in CCND1, MYC, EGFR, ERBB2, VEGFA, and CCNE1 (FDR 0.001), whereas CDH1 and RHOA mutations were enriched in non-CIN tumors. CIN exhibited a graded spectrum, with higher FGA associated with increased aneuploidy and a greater number of unique metastatic sites. Deep-learning models trained on H Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 1182.
Xu et al. (Fri,) studied this question.