Development and validation of prognostic model based on extragastric lymph nodes metastasis and lymph node ratio in node-positive gastric cancer: a retrospective cohort study based on a multicenter database

Background: Regional lymph node metastasis (LNM) is a competent and the most intensive predictor for the prognostic evaluation of patients after curative surgery. This study is based on the databases of two large medical centers in North and South China. It aims to establish a prognostic model based on extragastric LNM (ELNM) and lymph node ratio (LNR) in node-positive gastric cancer (GC). Methods: Clinical data of 874 GC patients with pathologically confirmed LNM in a large medical center in southern China, were included as the training cohort. In addition, the clinical data of 674 patients with pathologically confirmed LNM from a large medical center in northern China were used as the validation cohort. Results: In the training cohort, a modified N staging system (mNstage) based on ELNM and LNR was established; it has a significantly higher prognostic accuracy than the pN, LNR and ELNM staging system (Akaike Information Criterion, pN stage vs. LNR stage vs. ELNM stage vs. mN stage=5498.479 vs. 5537.815 vs. 5569.844 vs. 5492.123; Bayesian Information Criterion, pN stage vs. LNR stage vs. ELNM stage vs. mN stage=5512.799 vs. 5547.361 vs. 5574.617 vs. 5506.896; likelihood-ratio χ 2 , pN stage vs. LNR stage vs. ELNM stage vs. mN stage=177.7 vs. 149.8 vs. 115.79 vs. 183.5). In the external validation, mNstage also has higher prognostic accuracy than the pN, LNR and ELNM staging system. Cox multivariate regression analysis showed that age, mNstage, pT stage, and perineural invasion were independent factors. A nomogram model was established according to the four factors (age, mNstage, pT stage, and perineural invasion). The nomogram model was greater than the traditional tumor–node–metastasis (TNM) staging in the training cohort [1-year area under the curve (AUC), American Joint Commission for Cancer (AJCC) 8th TNM vs. nomogram=0.692 vs. 0.746, 3-year AUC: AJCC 8th TNM vs. nomogram=0.684 vs. 0.758, 5-year AUC: AJCC 8th TNM vs. nomogram=0.725 vs. 0.762]. In the external validation, the nomogram also showed better prognostic value and greater prediction accuracy than the traditional TNM staging. Conclusion: The prognostic model based on ELNM and LNR has good prognostic prediction in patients with node-positive GC.