To establish a predictive model for NO.253 lymph nodes metastasis of rectal cancer around inferior mesenteric artery (IMA) and to verify its accuracy.

The data of 156 patients underwent rectal cancer surgery from January 2014 to March 2019 were analyzed retrospectively.and all of 156 patients underwent NO.253 lymph nodes dissection. Statistical analysis were performed by using SPSS 20.0 software and were examined by using univariate analysis (χ² test) and multivariate logistic regression analysis of clinicopathological factors related to NO.253 lymph nodes metastasis. To establish a prediction model of NO.253 lymph node metastasis, by drawing the receiver operating characteristic curve (ROC) and calculating the area under the curve (AUC) to evaluate the predictive ability of the model. A P value of <0.05 was considered as statistically significant difference . And the prediction model was verified by the postoperative pathological diagnosis results as the gold standard.

Among the 156 patients, 21 cases (13.5%) of NO.253 lymph nodes metastasis ; The average harvested NO.253 lymph nodes was 2.3 (1~9), with a total of NO.359 lymph nodes were dissected, including 72 positive lymph nodes (20.1%). Logistic multivariate regression analysis was performed on clinically significant pathological data with univariate analysis. The results showed that tumor size, pT stage, degree of differentiation, presence or absence of liver metastases, and serum CEA levels were independent risk factors for NO.253 lymph nodes metastasis. According to the logistic regression analysis of the regression coefficients of each variable, a prediction model is constructed, and by plotting the ROC curve, the AUC value is calculated as 0.856 (95% CI 0.825-0.886). When the optimal critical value of Y is 0.49, it has predictive value for NO.253 lymph nodes metastasis. At this time, the sensitivity of the prediction model is 90.21%, with specificity of 86.20%, and the Jordan index of 0.56. By using the postoperative pathological diagnosis as the gold standard to verify the prediction model, with the sensitivity of 85.71%, and the specificity of 99.26%, without significant difference (P>0.05).

This study establishes a prediction model based on Logistic multivariate regression analysis with high sensitivity and specificity, with a high predictive value for NO.253 lymph nodes metastasis of rectal cancer.