An integrated pathological-radiological downstaging score for prognostic classification in gastric cancer after neoadjuvant chemotherapy: A multicenter retrospective study.

Background

Neoadjuvant chemotherapy (NAC) followed by surgery is the standard treatment for locally advanced gastric cancer (LAGC). However, post-therapy staging based on ypTNM mainly reflects residual pathological status and does not capture pretreatment disease burden or the extent of tumor regression after NAC, limiting postoperative prognostic stratification. To address this limitation, we developed a Pathological-Radiological Integrated Model (PRIM) that combines baseline stage, post-treatment pathology, and radiologic response.

Methods

We enrolled 414 patients with LAGC who underwent radical gastrectomy after NAC, including 348 in a development cohort and 66 in an external validation cohort. A Down-stage System (DS) was constructed to quantify tumor stage changes after NAC by weighting cT/cN and ypT/ypN categories using multivariable Cox regression. DS was then combined with RECIST response to establish the PRIM system. Prognostic performance and clinical utility were evaluated and compared with conventional ypTNM staging.

Results

PRIM stratified patients into four distinct risk groups with significantly different 3-year overall survival rates (86.9%, 69.9%, 40.0%, 21.2%; P < 0.001). Compared with ypTNM, PRIM demonstrated improved prognostic performance, with a higher C-index (0.728 vs. 0.696, P < 0.001), AUC (0.775 vs. 0.731, P < 0.001), and better model fit (AIC: 1471.62 vs. 1509.33; BIC: 1474.61 vs. 1512.32). Risk reclassification was also improved (NRI 14.49%; IDI 8.9%). Time-dependent ROC analysis showed consistently higher predictive accuracy for PRIM over 5 years.

Conclusion

By integrating pathological and radiological response information, the PRIM classification provides improved postoperative prognostic stratification compared with ypTNM alone in patients with LAGC treated with NAC.