Illustrating a multidimensional roadmap using biomarkers and therapeutics to reshape outcomes in gastroesophageal cancer

Introduction

Gastroesophageal cancers (GEC) present a significant global health challenge, with nearly 1.5 million new cases diagnosed annually, resulting in over 1.1 million deaths worldwide. ¹ There is substantial geographic variation in the incidence of GEC and a concerningly rapid increase in early-onset cases among people younger than 50 years of age.^1,2 GECs represent a spectrum of anatomically and molecularly distinct diseases. Esophageal squamous cell carcinoma (ESCC), which accounts for 85% of esophageal cancers (EC) globally, arises from chronic tobacco and alcohol exposure with key molecular alterations including TP53 and CDKN2A mutations. ³ By contrast, esophageal adenocarcinoma (EAC) develops from Barrett’s esophagus—intestinal metaplasia caused by chronic gastroesophageal reflux—with characteristic TP53 mutations, cell cycle pathway dysregulation, and ERBB2 amplification in 15%–20% of cases. ⁴ Gastroesophageal junction (GEJ) adenocarcinomas occupy an intermediate anatomical position, classified by the Siewert system based on tumor epicenter relative to the junction. Siewert types I and II share molecular features with EAC, predominantly displaying the chromosomal instability (CIN) subtype with TP53 and cell cycle alterations, while Siewert type III tumors are more heterogeneous and genomically resemble gastric cancers (GCs). GCs, most associated with Helicobacter pylori infection and dietary factors, are classified into four molecular subtypes: Epstein–Barr virus-positive (characterized by extreme hypermethylation and immune checkpoint expression), microsatellite instable (microsatellite instable-high with elevated mutation burden), genomically stable (enriched for diffuse histology with RHOA and CDH1 alterations), and chromosomally unstable (CIN, with frequent TP53 mutations and receptor tyrosine kinase amplifications). ⁵

Recent advances in GEC treatment have established biomarker-directed therapy as standard care, significantly improving survival across disease settings through perioperative immunotherapy, first-line chemoimmunotherapy stratified by PD-L1 and HER2 status, and novel targeted agents for specific molecular subsets. In the perioperative setting, adding durvalumab to FLOT chemotherapy improved event-free survival and overall survival (OS). ⁶ For advanced HER2-negative disease, combining PD-1 inhibitors with chemotherapy became standard first-line therapy for patients with PD-L1 Combined Positive Score (CPS) ⩾1. HER2-positive disease benefits from dual targeting: KEYNOTE-811 established pembrolizumab added to trastuzumab and chemotherapy as first-line therapy for CPS ⩾1 tumors, while trastuzumab deruxtecan provides effective second-line therapy. ⁷ Biomarker-selected therapies have expanded treatment options: zolbetuximab for CLDN18.2 (an important molecule in tight junction assembly in normal gastric epithelium)-positive tumors improved median OS when combined with chemotherapy. ⁸ These precision approaches enable personalized treatment selection that meaningfully extends survival—with median OS now exceeding 18 months in biomarker-selected populations.

Despite these advances in therapeutics and understanding of the disease pathology, several challenges remain, including the emergence of treatment resistance, tumor heterogeneity, and limitations to access to these innovative therapies. These limitations highlight the need for a multidimensional approach to precision oncology to incorporate functional biomarkers, use real-world clinical data to both guide treatment selection and predict toxicities, and ultimately improve clinical outcomes for patients with GC and EC.

Multidimensional model of precision oncology

Precision oncology is now expanding beyond the focus on just genetic mutations to include biomarkers that incorporate tumor biology, expression of proteins, and real-time indicators of treatment response. These studies illustrate how biomarkers are refining and reshaping patient selection, including CLDN18.2 and other surface targets, to using circulating tumor DNA (ctDNA) for monitoring treatment response to real-world data.

Analysis of the Ni-High phase Ib clinical trial of chemotherapy plus dual-targeted therapy (trastuzumab + nivolumab) in patients with HER-2 positive advanced GC by Osumi et al. ⁹ revealed that plasma ctDNA was detectable in 20 out of the 21 patients and also showed an ERBB2 amplification in 12 out of the 21 patients. The analysis also showed that patients without ERBB2 single-nucleotide variants (SNVs) had both longer median progression-free survival and OS than patients without these alterations. Interestingly, patients with a lower maximum mutant allele frequency at cycle 2 had a better response to chemotherapy with the dual-targeted treatment. Overall, Osumi et al. ⁹ demonstrated that an early drop in ctDNA burden and focal ERBB2 amplification identified a cohort of patients who had durable treatment benefit from combined PD-1 and HER2 blockade compared to those with ERBB2 SNVs who had poorer responses.

The review by Yamamoto et al. ¹⁰ highlighted the emergence of non-oncogenic drivers like CLDN18.2 as actionable targets for therapeutics. They distinguish non-oncogenic targets from oncogenic drivers in that the former targets are involved in all of proliferation, survival, or adaptation without directly causing tumorigenesis. Specifically, zolbetuximab is now approved as a first-line option in the management of HER-2-negative, CLDN18.2-positive patients. The review by Yamamoto et al. ¹⁰ also highlights other non-oncogenic targets such as Caprin-1, TROP2, and Nectin-4, all molecules that are overexpressed in the tumor cells and possible therapeutic targets in those with advanced GC.

Marchesi et al. ¹¹ highlight the central role of molecular profiling in the rapidly evolving management of advanced GC and GEJ adenocarcinomas. The recent advances in immune checkpoint inhibitors (ICIs) and targeted therapeutics in biomarker-selected patient subgroups have resulted in clinically meaningful survival benefits. Specifically, key biomarkers such as HER2, mismatch repair (MMR) deficiency, elevated PD-L1, CLDN18.2, and FGFR2b now enable more personalized treatment strategies for patients. ¹¹ Cutting-edge approaches such as bispecific antibodies, chimeric antigen receptor T-cell (CAR-T) therapy, and antibody–drug conjugates further highlight the challenges in optimizing treatment for those with overlapping biomarkers and tumor heterogeneity but nonetheless underscore the central role of precision oncology in improving outcomes in this aggressive disease.

Furthermore, Shimozaki et al. and Lin et al. showed the clinical benefit of ICI therapy, although highlighting the need for more real-world data to monitor treatment effectiveness and regular use of predictive biomarkers to appropriately select patients who gain the most therapeutic benefit.

Shimozaki et al. ¹² retrospectively evaluated outcomes in HER2-negative advanced GC patients who received first-line platinum-based chemotherapy across three distinct periods (pre-immunotherapy vs immunotherapy approved for third-line or later vs immunotherapy as first-line). This analysis demonstrated that first-line immunotherapy + chemotherapy resulted in significantly higher OS compared to patients who did not receive immunotherapy. ¹² Furthermore, patients receiving immunotherapy at any line were associated with improved outcomes. Thereby, the improved survival in HER2-negative advanced GC patients highlights the need for improved biomarker identification.

Lin et al. ¹³ evaluated 202 patients with advanced/recurrent metastatic ESCC who received first-line pembrolizumab + chemotherapy and found that this combination had clinical benefit with an acceptable safety profile. The study findings note similar OS for both treatment-naïve and recurrent disease patients. Patients with oligometastases also had favorable clinical outcomes. ¹³

These studies collectively signal a new phase in precision oncology where these diverse biomarkers ultimately help inform a more tailored therapeutic approach.

Anticipating treatment toxicities using clinical and functional metrics

Using real-world data, the development of practical clinical decision tools that provide individualized risk stratification can optimize patient safety and help reduce treatment-related toxicities.

A retrospective study of 435 patients with EC patients who received radiation therapy by Yang et al. ¹⁴ presented a nomogram to predict chemoradiotherapy-associated thrombocytopenia in EC patients receiving radiotherapy. The study showed that among the 435 EC patients who received radiotherapy, 23.91% developed thrombocytopenia during or after their treatment. Yang et al. ¹⁴ identified nine independent predictors of patients developing chemotherapy-associated thrombocytopenia and incorporated these into developing a nomogram, which ultimately had a high prediction accuracy. Ultimately, this study showed that real-world data sets can help develop practical clinical support/decision tools that can be used to enhance treatment safety.

The review by Narita and Muro ¹⁵ demonstrates the utility of a comprehensive geriatric assessment in providing a framework to optimize systemic treatment of elderly patients with GC. The review highlights the gap in current trials that historically underrepresent individuals aged >65 years, who comprise over half of GC cases. For “fit” elderly patients, doublet therapy provides efficacy comparable to that of younger patients; however, frailer patients may benefit from either a de-intensified or monotherapy regimen to reduce toxicity while maintaining efficacy. ¹⁵ Interestingly, subgroup analyses also show that elderly patients derive similar benefits from targeted agents and ICIs compared to younger patients. ¹⁵ Thereby, the review provides ample discussion on using practical, real-world strategies to tailor systemic therapy that enable fit elderly individuals to benefit from modern therapeutics but spare frailer patients from excess toxicity.

Overall, these studies demonstrate the utility of real-world data to develop models that predict treatment-associated adverse reactions and guide therapy intensity to balance treatment efficacy and toxicity.

Tailoring treatment for high-burden peritoneal disease

Peritoneal metastases (PM) in GC occur in nearly a third of the patients at diagnosis and represent a dismal prognostic finding, with a median survival of less than a year. ¹⁶ GC related PM is increasingly viewed as a clinically distinct entity with its biological and molecular nuances. ¹⁶ A deeper understanding of the unique and complex genetic architecture of peritoneal tumors is thereby warranted to develop therapeutics directed toward peritoneal-specific tumor biology. Several studies analyzed the role of ascites as a biomarker and peritoneal-directed treatment strategies given the paucity of data evaluating the relationship between ascites, peritoneal disease, and survival outcomes in patients with metastatic colorectal (mCRC) and metastatic GCs (mGC).

Provenzano et al. ¹⁷ highlighted malignant ascites as a marker of peritoneal carcinomatosis burden as well as poor prognosis in advanced mCRC and mGC. Specifically in GC patients with PM, those with ascites had a significantly lower OS (13 vs 21 months) and a higher Peritoneal Cancer Index. ¹⁷ This pooled analysis of clinical trial data thereby highlights a specific subset of patients with advanced peritoneal metastatic disease who may benefit from tailored treatment.

Furthermore, Filho et al. ¹⁸ evaluated the role of neoplastic cells’ (NC) CD44+ and CD326+ levels as a dynamic biomarker in peritoneal fluid. This prospective cohort study included patients with advanced GC with peritoneal involvement and ascites who then underwent repeated intraperitoneal perfusion normothermic chemotherapy. Quantification of NC in peritoneal fluid by markers CD44+ and CD326+ by flow cytometry as a surrogate for peritoneal tumor burden. The study noted that the median OS was 22.6 months among patients who underwent conversion surgery after achieving negative peritoneal lavage cytology compared to 14.6 months among those who did not undergo resection. ¹⁸ Continued monitoring/tracking changes in the CD44+ and CD326+ NC populations during treatment allowed for early identification of those who were responding to current treatment. ¹⁸ Overall, the concept of a liquid biopsy to assess for peritoneal tumor burden allows a more precise method to stratify patients and monitor treatment response.

Together, these findings highlight malignant ascites and peritoneal-directed biomarkers as important tools to predict treatment outcomes and guide clinical management in GC and CRC patients with peritoneal disease.

Overall, the modern approach for management of GEC is now increasingly guided by a more sophisticated array of tools to integrate tumor biology, biomarkers, and real-world data. From using real-time testing to confirm treatment response, to using profiling to identify targets such as CLDN18.2 or HER2, to developing clinical monograms that predict treatment-related toxicities, cutting-edge research is ultimately providing a new roadmap toward a future in precision oncology where treatment regimens evolve to best treat the individual patient.