Clinicopathological characteristics and lymph node metastasis in patients with early gastric cancer
10.3760/cma.j.cn113855-20210831-00525
- VernacularTitle:早期胃癌患者临床病理特征与淋巴结转移因素分析
- Author:
Qingwei LIU
1
;
Yong LI
;
Bibo TAN
;
Liqiao FAN
;
Qun ZHAO
;
Qiang JI
;
Zhaoxing LI
;
Ming TAN
;
Yijie ZHAO
;
Xinyu YUAN
Author Information
1. 河北医科大学第四医院外三科,石家庄 050000
- Keywords:
Stomach neoplasms;
Lymphatic metastasis;
Risk factor;
Predicting
- From:
Chinese Journal of General Surgery
2022;37(4):255-259
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To explore the risk factors of lymph node metastasis (LNM) in early gastric cancer (ECG), and establish a risk-prediction model based on LNM.Method:Four hundred and twenty-seven EGC patients undergoing curative radical gastrectomy were enrolled in this study. The risk factors for LNM of ECG were analyzed with Logistic regression. LNM risk was stratified and risk-predicting model was established. The risk-predicting model was measured by area under ROC curve. According to the same standard, clinical data of 133 patients with EGC who underwent radical surgery were selected for external verification of the model.Results:The frequency of LNM was 13.3% (32/427) in EGC patients. The LNM ratio of intramucosal carcinoma and submucosal carcinoma was 1.3% (3/237), 15.3% (29/190) respectively. Ulcer presence, tumor size >2 cm, undifferentiated tumor, submucosal invasion, neural invasion, and vascular tumor thrombus were significantly associated with LNM in EGC patients ( χ2=3.408, 16.379, 4.808, 29.804, 25.305, 47.120, respectively P<0.05). Multivariate analysis suggested that ulcer presence, tumor size >2 cm, depth of invasion, neural invasion, and vascular tumor thrombus were independent predictors of LNM in EGC patients, ( OR=0.326, 2.924, 11.824, 13.047, 7.756, respectively P<0.05). LNM predicting model is established, P=e^x/(1+e^x),x=-4.792-1.122 ulcer presence+1.073 tumor size+2.470 depth of invasion+2.569 neural invasion+2.048 vascular tumor thrombus,ROC-AUC of risk-predicting model was 0.845, the best cut-off was 0.094, the sensitivity was 72.70%, the specificity was 77.20%. The external verification result revealed the AUC of ROC was 0.840. The four-grid table is constructed by predicting model results and the postoperative pathological examination. The sensitivity and specificity of the model are calculated to be 82.35% and 68.96%, respectively. Conclusions:EGC patients with ulcer presence, tumor size >2 cm, depth of invasion, vascular tumor thrombus, and neural invasion have higher risk of LNM, the risk-predicting model can identify the high probability of LNM .