New 3D Staging Refines Risk Prediction After Neoadjuvant Therapy in Gastric Cancer
A multicenter real-world study in locally advanced gastric cancer suggests that postoperative prognosis is better captured when tumor regression is folded into traditional pathologic staging. The new TRG-ypT-ypN three-dimensional staging system was built to address a weakness in the 8th edition AJCC ypTNM framework: after neoadjuvant therapy, residual tumor burden alone does not fully describe how much biologic damage treatment has already done to the cancer.
That omission matters in the clinic. Two patients can emerge from neoadjuvant treatment with the same ypT and ypN categories, yet one may have marked tumor regression and the other may have little or no treatment effect. A staging model that ignores regression can flatten those biologic differences into a single postoperative label, even though the underlying prognosis may diverge sharply.
The study included 1270 patients who underwent radical gastrectomy after neoadjuvant therapy between January 2020 and January 2025 across twelve institutions. Of these, 763 received neoadjuvant immunochemotherapy and 507 received neoadjuvant chemotherapy. The cohort was split by institution into a training cohort of 843 patients and a validation cohort of 427 patients, allowing the investigators to construct the model in one set and test it in another. The staging system was developed using a Cartesian coordinate framework that incorporated tumor regression grade, ypT stage, and ypN stage into a single three-dimensional structure.
Survival Separated More Cleanly Across the 3D Stages
The resulting 3D system divided patients into stages I, II, IIIA, and IIIB with clear survival separation. In the overall neoadjuvant-treated cohort, the 3-year overall survival rates were 91.9% for stage I, 78.1% for stage II, 54.9% for stage IIIA, and 45.8% for stage IIIB. Those stepwise declines suggest the model tracks a more biologically coherent spectrum of postoperative risk than ypTNM alone.
The separation was not subtle. As stage burden increased, survival dropped in a graded fashion rather than clustering into overlapping groups. In gastric cancer, where occult nodal spread and incomplete pathologic response can coexist after preoperative treatment, a model that combines local invasion, nodal status, and regression is intuitively better aligned with what surgeons and pathologists actually see in resection specimens.
The 3D staging system outperformed ypTNM in both the training and validation cohorts. Its advantage was especially evident in the neoadjuvant immunochemotherapy subgroup, where prognostic performance improved more clearly. In the neoadjuvant chemotherapy subgroup, the improvement was only a non-significant trend, suggesting that the biological information captured by regression may be more valuable when immune modulation is part of the preoperative regimen.
Why Tumor Regression Changes the Prognostic Equation
Tumor regression grade reflects the extent to which therapy has eliminated viable cancer cells and replaced them with fibrosis, acellular mucin, or other treatment-related changes. In practical terms, it serves as a readout of chemosensitivity or immunochemotherapy sensitivity, not merely of residual anatomy. A small residual tumor after a deep pathologic response may carry a different metastatic potential than the same-sized residual tumor after poor response, because the treatment has already filtered out more aggressive clones.
This is particularly relevant in neoadjuvant immunochemotherapy, where immune activation may alter the tumor microenvironment in ways that a pure anatomic staging system cannot register. The stronger performance of the 3D model in that subgroup fits the idea that response depth is not just a surrogate for shrinkage; it may be a surrogate for biologic vulnerability. In other words, the cancer’s behavior after treatment can be more informative than its size at the moment of surgery.
Relapse-free survival showed a similar pattern. The 3D system demonstrated modestly improved prognostic performance there as well, reinforcing the idea that postoperative recurrence risk is not captured fully by ypTNM alone. Recurrence in gastric cancer often depends on the interaction of residual primary disease, nodal involvement, and the tumor’s response to treatment. A model that integrates all three may be better suited to the post-neoadjuvant setting than one that was designed for broader pathologic staging.
What Pathologists and Surgeons May Change at the Bedside
For clinicians, the most immediate implication is risk stratification after gastrectomy. Patients who appear similar under ypTNM may not be equivalent once regression is considered, and that difference could affect decisions about surveillance intensity, adjuvant planning, and prognostic counseling. The study does not establish a new treatment algorithm, but it does provide a stronger framework for identifying which postoperative patients are likely to do well and which remain at higher risk despite curative-intent surgery.
The multicenter design gives the findings added weight because the model was tested across twelve institutions rather than in a single highly specialized center. Real-world cohorts are messier than tightly curated trial populations, and staging systems that remain useful in routine practice need to survive that complexity. By separating the cohort into training and validation sets, the investigators also reduced the chance that the model worked only because it was fitted too closely to one dataset.
Still, the data stop short of proving that the 3D framework should replace ypTNM. The study shows better prognostic discrimination, not improved patient outcomes. Whether this model can be embedded into future decision pathways, or whether it will be adopted alongside ypTNM as a complementary tool, will depend on broader validation and clinical utility beyond survival curves.
A Better Fit for the Post-Neoadjuvant Era
The core message is straightforward: after neoadjuvant therapy, residual anatomy is only part of the story. TRG-ypT-ypN staging attempts to capture the biologic residue of treatment itself, and that appears to sharpen prognostic precision in locally advanced gastric cancer, especially after neoadjuvant immunochemotherapy.
The current evidence still leaves practical questions open. The model’s role in guiding adjuvant treatment selection, its reproducibility across pathology workflows, and its performance in non-Asian cohorts all remain untested in this report. For now, the study argues that postoperative staging in gastric cancer should reflect not only what remains after therapy, but also how much the tumor has already yielded to it.