Prognostic factors and the necessity of chemotherapy for stage II gastric cancer: a model based on multicenter retrospective study.

Background: This study aimed to construct a prognostic model for prognosis prediction and assess the response to adjuvant chemotherapy (ACT) of stage II gastric cancer (GC) patients on high and low survival risk stratifications.

Methods: We retrospectively reviewed 547 stage II gastric cancer patients who underwent D2 radical gastrectomy from January 2009 to May 2017 in Sixth Affiliated Hospital of Sun Yat-Sen University (SAH-SYSU), the Fujian Medical University Union Hospital (FJUUH), and the Sun Yat-Sen University Cancer Center (SYSUCC).The propensity score matching (PSM) of all variables was performed to balance selective bias between ACT and surgery alone (SA) groups. Kaplan-Meier survival and multivariate Cox regression analyses were carried out to identify independent prognostic factors. Independent factors selected by the Cox regression were integrated into the nomogram. The nomogram points stratified patients into high-risk and low-risk groups by the optimal cut-off value.

Results: 278 patients were selected after PSM. Age, tumor site, T stage and lymph-nodes-examined (LNE) selected by Cox regression as independent prognostic factors were integrated into the nomogram. The nomogram performed well with a C-index of 0.76 and with C-indexes of 0.73 in and 0.71 in two validate cohorts. AUCs of the 3 year and 5 year ROC curves were 0.81 and 0.78. High- and low-risk groups stratified by the cut-off value demonstrated different responses to ACT.

Conclusions: The nomogram performed well in prognosis prediction. Patients in high- and low-risk groups demonstrated different responses to ACT, and high-risk patients might need ACT.