Rethinking the Siewert classification in the era of precision oncology Authors: Lawrence W. Wu, Ryan H. Moy Published: 2026 The Siewert classification is a system for categorizing adenocarcinoma of the gastroesophageal junction (GEJ) by anatomic location (1). Originally conceived in 1987, this system defines Type I tumors as centered 1–5 cm above the gastric cardia, Type II tumors as located 1 cm above or 2 cm below the gastric cardia, and Type III tumors as centered 2–5 cm below the gastric cardia (2). The Siewert classification carries both important prognostic and therapeutic implications, particularly in guiding surgical approach for patients with localized disease (3). The Siewert classification was developed at a time when anatomic considerations predominated and surgery was the principal treatment. However, current management has evolved to employ systemic therapies, biomarker selection, targeted treatments, and immunotherapy. Prior landmark work from The Cancer Genome Atlas (TCGA) characterized esophagogastric adenocarcinoma into four molecular subtypes: chromosomal instability (CIN), genomically stable (GS), microsatellite instability (MSI), and Epstein Barr virus (EBV)-associated (4,5). CIN tumors are marked by extensive aneuploidy and copy number alterations, as well as an enrichment for TP53 mutations and RTK-RAS pathway activation. In contrast, GS tumors lack widespread chromosomal aberrations and are enriched for CDH1 and RHOA mutations with dysregulation of cell-adhesion pathways. MSI tumors exhibit a hypermutated phenotype, and EBV-associated tumors demonstrated recurrent PIK3CA mutations and prominent immune cell signaling. Further work at Memorial Sloan Kettering Cancer Center (MSKCC) identified that TCGA molecular subtypes carry prognostic significance in patients with esophagogastric cancers, independent of anatomic location (6). Patients with EBV and MSI subtype esophagogastric cancers had improved survival compared to those with the GS and CIN subtypes. A similar pattern emerged in exploratory biomarker analysis of the landmark randomized phase III clinical trial CheckMate 649, which evaluated nivolumab plus chemotherapy versus nivolumab plus ipilimumab versus chemotherapy alone in untreated, advanced HER2-negative esophagogastric adenocarcinoma (7,8). In this study, patients with the MSI and EBV tumors derived greater clinical benefit with nivolumab-based regimens compared to those with the GS and CIN subtypes (8). Thus, these studies suggest that TCGA molecular subtypes may inform not only prognosis, but there may be potential therapeutic implications with respect to immune checkpoint inhibition. However, whether these molecular distinctions show key differences across the anatomic regions of the GEJ has remains incompletely defined. To address this question, Nakauchi et al. performed a comprehensive genomic landscape analysis of Siewert type I, II, and III tumors, which was recently published in Annals of Surgery (9). The study authors performed a retrospective analysis of 121 type I, 170 type II, and 59 type III tumors from patients who underwent curative intent surgery and comprehensive genomic sequencing with MSK-Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT) at a large tertiary cancer center (10). Tumors were analyzed for somatic mutations, copy number alterations, and select structural rearrangements. Additionally, pathway analysis of canonical signaling pathways per TCGA PanCancer Atlas Project was performed (11). The GEJ tumors were compared to an esophageal adenocarcinoma (n=293) and a gastric adenocarcinoma (n=100) cohort. Siewert type I and II tumors demonstrated a higher frequency of TP53 mutations and CDKN2A/B deletions, and were also enriched for the CIN TCGA subtype. In contrast, Siewert type III tumors more frequently harbored ARID1A and CDH1 mutations, FGFR2 amplifications, and a higher proportion of MSI-high (MSI-H) and GS TCGA subtypes. The percentage of CIN tumors decreased stepwise and percentage of MSI-H and GS tumors increased stepwise from Siewert I to III. Nonetheless, CIN remains the dominant molecular subtype across Siewert types. The key finding was that Siewert type I and II tumors were molecularly similar to distal esophageal cancer, and Siewert type III tumors were molecularly similar to gastric cancer (Figure 1). Integrated molecular features of gastroesophageal junction adenocarcinoma by Siewert classification. Image adapted from Servier Medical Art (https://smart.servier.com), licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). CIN, chromosomal instability; GEJ, gastroesophageal junction; GS, genomically stable; MSI-H, microsatellite instability-high. We believe that the study by Nakauchi et al. highlights the value of comprehensive genomic profiling and how it can augment the Siewert anatomic classification from both a prognostic and therapeutic perspective. Anatomic labels are likely imperfect proxies, particularly for immunotherapy-sensitive molecular subtypes. When comprehensive genomic profiling is not available, then immunohistochemistry assays such as mismatch repair protein testing can provide pragmatic, cost-conscious surrogate information to inform perioperative management particularly when deficient mismatch repair is detected. This is especially relevant at the GEJ, where there is significant tumor heterogeneity given that it is a transition zone between the esophagus and stomach. Limitations of this study include that data is only from a single institution and there is no transcriptional, epigenomic, or tumor and immune microenvironment data. Thus, future studies should attempt to validate these findings in a larger real-world cohort, and efforts to explore differences in other tumor molecular features by Siewert type may yield important mechanistic insights. Overall, the Siewert classification is still relevant for surgical planning, but molecular subtypes may be helpful for guiding systemic treatments, particularly in the perioperative setting, and clinical trial design. Supplementary The article’s supplementary files as Available at https://tgh.amegroups.com/article/view/10.21037/tgh-2025-172/prf