Biology before stage in advanced gastric cancer: converting to cure in biomarker-selected patients

Metastatic and initially unresectable gastric cancer has long been managed with palliative intent. As biomarker-directed therapy matures, a small but meaningful subset now appears to defy that rule. Park et al. present a multi-institution series describing four patients in whom tumor biology rather than anatomic stage determined the clinical path. Three tumors were human epidermal growth factor receptor 2 (HER2) positive and received a combination of trastuzumab, a programmed cell death protein 1 (PD-1) inhibitor, and chemotherapy. One tumor was microsatellite instability-high (MSI-H) with high programmed cell death ligand 1 (PD-L1) expression and responded to chemotherapy in combination with nivolumab. All four patients, aged between 54 and 80 years, were subsequently evaluated for resection after deep systemic responses, two achieved a pathologic complete response, and the others had substantial downstaging with negative margins and extensive nodal sterilization. These experiences do not yet represent a new standard of care, but they challenge the assumption that stage at presentation solely determines prognosis. In select molecular contexts, underlying biology can create opportunities for cure. The individual narratives illustrate how conversion can be accomplished in practice. In the first case, a woman with microsatellite stable, HER2 3+ disease progressed on perioperative chemotherapy, then received FOLFOX (fluorouracil, leucovorin, and oxaliplatin) with pembrolizumab and trastuzumab. Cross-sectional imaging showed disappearance of the primary mass and resolution of nodes, laparoscopy excluded peritoneal disease, and total gastrectomy with D2 dissection yielded a pathologic complete response with negative margins and nodes. She has remained disease free since 2020, which is a strong curative signal. The second case began as de novo stage IV disease dominated by multistation nodal metastases from the neck to the pelvis. First-line therapy with pembrolizumab, trastuzumab, and CAPOX (capecitabine and oxaliplatin) produced a clinical complete response, followed by localized relapse at the primary site after de-escalation. Rechallenge consolidated control, laparoscopy again excluded peritoneal spread, and total gastrectomy produced ypT3N0 with 0 of 64 nodes involved and no radiographic evidence of disease at 1 year. The third case presented with malignant ascites and peritoneal carcinomatosis, traditionally a poor prognostic setting. After 1 year of trastuzumab, pembrolizumab, and fluoropyrimidine-based therapy, repeat laparoscopy and positive emission tomography/computed tomography (PET-CT) documented clearance of peritoneal disease and distal gastrectomy achieved ypT3N1 with negative margins, followed by 2 years of adjuvant pembrolizumab and trastuzumab and ongoing absence of detectable disease on imaging. The fourth case, an MSI-H, PD-L1, combined positive score (CPS) 70, HER2-negative tumor in an older adult with Eastern Cooperative Oncology Group performance status (ECOG PS) 2, achieved pathologic complete response after FOLFOX plus nivolumab, with negative margins and nodes at partial gastrectomy (1).

Placed in the context of contemporary evidence, Park et al. align with the ongoing shift toward biomarker-integrated first-line therapy in unresectable or metastatic gastric and gastroesophageal junction adenocarcinoma. For HER2-positive disease, adding a PD-1 inhibitor to chemotherapy and trastuzumab deepens responses, this is further supported with findings from the Keynote-811 trial (2). For MSI-H disease, checkpoint blockade has transformed expectations across gastrointestinal malignancies and is increasingly considered in the neoadjuvant setting (3). Park et al. show how these regimens can move selected patients from systemic control to operability when restaging is disciplined and multidisciplinary.

Why do outcomes differ among patients who receive similar backbones of therapy? The variability observed here likely reflects differences in target expression, PD-L1 combined positive score, MSI status, immune infiltration, stromal features, and baseline disease distribution, including peritoneal involvement. Two practical rules follow. First, test early and broadly at diagnosis. At a minimum, determine HER2 by immunohistochemistry with in situ hybridization confirmation when indicated, PD-L1 by combined positive score, and MSI or mismatch repair by validated assays. Where resources and local practice support additional targets, claudin 18.2 (CLDN18.2) and fibroblast growth factor receptor 2 (FGFR2) or mesenchymal-epithelial transition (MET) amplification can be considered (4,5). Second, restage with surgical intent once deep systemic control is suspected. For patients who presented with peritoneal disease or where occult peritoneal spread is a concern, repeat diagnostic laparoscopy and peritoneal assessment should precede any commitment to gastrectomy, since negative imaging alone may not exclude microscopic peritoneal involvement.

Two themes from Park et al. deserve emphasis for day-to-day practice. The first is receptor plasticity under treatment pressure. In two HER2-positive cases, HER2 expression on the resection specimen was markedly reduced or lost. This observation supports routine re-biopsy and retesting at conversion and at recurrence, since ongoing HER2-directed therapy may not remain relevant after surgery. Where tissue sampling is limited or risky, radiogenomics may offer a noninvasive adjunct to infer target expression and immune contexture. Emerging radiomic signatures derived from routine CT or PET have been reported to correlate with HER2 status, PD-L1 expression, and MSI phenotype, and longitudinal delta radiomics may capture treatment-induced shifts in tumor biology (6-10). These methods are not a substitute for histology when tissue is available, but they can aid in prioritizing patients for re-biopsy, detecting potential loss of HER2 during therapy, and guiding the timing of conversion surgery. Prospective validation with standardized image acquisition and external testing sets is needed before routine clinical adoption.

The second theme is the need to define remission with more than morphology. Radiologic complete response is not equivalent to cure, and even a pathologic complete response is not a guarantee of long-term freedom from disease in initially metastatic presentations. Molecular remission, defined by persistently undetectable circulating tumor DNA using a validated assay, has the potential to serve as a more reliable stop rule and as an early surrogate for durable cure. None of the four cases reported circulating tumor DNA (ctDNA) assessments. Future protocols should integrate serial ctDNA before and after surgery to confirm clearance of disease and to guide duration of therapy (11-13).

Duration of post-operative systemic therapy remains variable. In the series by Park et al., one patient continued pembrolizumab while de-escalating chemotherapy and discontinuing trastuzumab, another completed 2 years of adjuvant checkpoint and HER2-directed therapy after resection, and others stopped therapy after pathologic complete response and surgery. These pathways represent reasonable clinical judgment in an evidence gap. The field would benefit from prospective, biomarker-anchored protocols that test fixed-duration adjuvant strategies once molecular remission is achieved, with prespecified criteria for stopping. Until such data arrive, multidisciplinary review at each inflection point remains essential, including explicit discussion of residual risk, toxicity, and patient preference.

Safety and feasibility matter as much as efficacy when contemplating conversion. Case 1 achieved pathologic complete response despite ECOG 2 and serious intercurrent illness during initial chemotherapy. Case 4 involved an older patient with performance limitations who nonetheless derived substantial benefit from chemo-immunotherapy. These examples suggest that conversion pathways can succeed outside ideal conditions when comorbidity is carefully managed and restaging is methodical. At the same time, patient selection must remain disciplined. Not every deep response should lead to surgery, and not every radiographic improvement represents a durable biological shift. Negative staging laparoscopy, negative margins, and a convincing assessment that all sites of disease can be addressed are prerequisites for attempting curative intent.

Several cautions are warranted. Selection bias is inherent in a small case series, and follow-up remains modest in two of the four patients. Assay heterogeneity across institutions and over time complicates interpretation, particularly for HER2 and PD-L1, and differences in testing methodology should be made explicit whenever possible. The loss of HER2 expression at surgery introduces uncertainty about the value of continued HER2-directed therapy in the adjuvant setting, a question that only prospective data can resolve. Finally, while ethical publication without written consent is permitted under specific circumstances when exhaustive attempts are documented, broader inclusion of patient perspectives would strengthen future reports and inform shared decision making about conversion strategies.

What should clinicians implement now? The most immediate step is to operationalize a conversion-intent pathway at the institutional level. This pathway should be activated at diagnosis through universal biomarker testing, incorporate predefined radiologic and surgical restaging milestones, mandate reassessment of target expression at the time of conversion, and integrate molecular surveillance where feasible. Multidisciplinary tumor boards should review each candidate at the points of initiation, response assessment, and preoperative decision making. For HER2-positive disease, checkpoint-augmented trastuzumab plus chemotherapy is a rational default in appropriate candidates. For MSI-H disease, early incorporation of a PD-1 inhibitor is reasonable, and in selected patients neoadjuvant immune checkpoint inhibition without chemotherapy may be an option, particularly for older adults or those with limited functional reserve.

Where should research focus next? A pragmatic, multi-center registry would allow rapid accumulation of outcomes for biomarker-selected advanced gastric cancer treated with conversion intent, with standardized definitions for response, operability, pathologic assessment, and molecular remission. Within such a framework, prospective conversion pathways should pair serial ctDNA with radiogenomic modeling to define molecular and imaging milestones, using harmonized imaging protocols and prespecified thresholds for action. This integrated approach could reduce unnecessary invasive procedures, identify early biological escape, and provide reproducible stop rules after R0 resection. Specific questions that merit study include the optimal duration of preoperative therapy, the timing of surgery relative to peak response, the utility of ctDNA to guide stopping or escalation, and the value of continuing or discontinuing HER2-directed therapy when expression is lost at resection.

In selected patients with advanced gastric cancer, tumor biology may reopen therapeutic opportunities that stage alone seemed to foreclose. Park et al. show that chemotherapy combined with PD-1 blockade and HER2-targeted therapy, or PD-1 blockade for MSI-H disease, can downstage tumors to permit curative-intent surgery with negative margins and sustained disease control. The path forward is to standardize how we identify the right patients, how we confirm remission, and when we stop therapy, ideally with molecular criteria in addition to imaging and pathology. If the community can turn these principles into a reproducible pathway, conversion to cure will become less the exception and more an expected outcome in biomarker-selected disease.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.