Journal Publication

Meta-analysis ties PD-(L)1 benefit in gastric cancer to metastasis site, histology

Pooling six phase 3 trials and 5,410 patients, the analysis finds checkpoint inhibitors add little survival benefit when peritoneal metastases or diffuse-type histology are present.

By AppliedXLJuly 14, 2026

Executive Summary

Pooled results across six phase 3 trials confirm that adding a checkpoint inhibitor to first-line chemotherapy extends survival in HER2-negative advanced gastric and gastroesophageal junction cancer, but the size of that benefit depends heavily on where the cancer has spread and what it looks like under the microscope.
Patients with peritoneal spread and those with diffuse-type tumors gained less from immunotherapy than patients with liver-only disease or intestinal-type tumors, pointing to biological heterogeneity within a population usually treated as one group.
Higher PD-L1 expression tracked with a larger survival gain, reinforcing biomarker-based patient selection as a lever separate from histology and metastatic pattern.
The results support tailoring expectations, and potentially treatment intensity, by metastatic site and histologic subtype rather than treating checkpoint-inhibitor benefit as uniform across HER2-negative disease.

The question

PD-(L)1 blockade added to platinum-fluoropyrimidine chemotherapy has become a first-line standard in HER2-negative advanced gastric and gastroesophageal junction adenocarcinoma (GC/GEJ), the trial abstract states, but the benefit has appeared inconsistent across patient subsets. The analysis set out to test whether metastatic site, liver versus peritoneal spread, and Lauren histologic subtype, intestinal versus diffuse, explain that inconsistency, rather than treating checkpoint-inhibitor benefit as a single population-wide effect. Impact

How it was done

The authors ran a systematic PubMed, MEDLINE, and Embase search for phase 3 randomized trials published between 2021 and 2025 comparing immunotherapy plus chemotherapy against chemotherapy alone in HER2-negative metastatic or unresectable GC/GEJ. Six trials met criteria, CheckMate 649, ATTRACTION-4, KEYNOTE-859, ORIENT-16, RATIONALE-305, and GEMSTONE-303, covering 5,410 patients. Hazard ratios for overall survival were pooled with random-effects models in the intention-to-treat population and in prespecified subgroups by liver metastases, peritoneal metastases, Lauren subtype, and PD-L1 expression by combined positive score or tumor area positivity score, with heterogeneity assessed by the I² statistic. The review is registered with PROSPERO under identifier CRD420251174893. Impact

The overall result

Across the pooled population, immunotherapy plus chemotherapy cut the hazard of death by 21% versus chemotherapy alone (HR 0.79, 95% CI 0.75-0.84, p<0.001). Grade 3 or higher adverse events were infrequent and predominantly hematologic, consistent with the safety profile chemotherapy alone carries. Impact

Where the benefit splits

The subgroup results are where the finding sharpens. Patients with liver metastases and those without liver metastases both benefited (HR 0.75 and HR 0.82, respectively), but patients with peritoneal metastases showed a hazard ratio of 0.93, indicating limited benefit from adding immunotherapy in that subgroup. Intestinal-type tumors showed a larger survival gain than diffuse-type tumors (HR 0.78 versus HR 0.87). Patients with a PD-L1 combined positive score of 5 or higher had the largest benefit of any subgroup analyzed (HR 0.70). Impact

What it changes

The authors conclude that first-line chemoimmunotherapy prolongs survival in HER2-negative GC/GEJ overall, but that reduced benefit in diffuse-type and peritoneal disease reflects underlying tumor heterogeneity and supports site- and histology-specific treatment strategies rather than a single approach across all HER2-negative patients. Impact

This analysis was produced using AI-assisted reporting systems, AppliedXL data, and official public records. These systems undergo editorial review, quality checks, and regular audits by human experts. Errors may still occur, as with any automated system. Always consult the linked primary sources. Read our AI Editorial Policy.