Construction And Evaluation Of A Predictive Model For Sentinel Lymph Node Metastasis In Breast Invasive Ductal Carcinoma

Objective: Based on logistic regression to explore the risk factors of sentinel lymph node（SLN）metastasis in patients with invasive ductal carcinoma（IDC）of the breast,and provide a basis for the construction and evaluation of the SLN metastasis prediction model for breast IDC patients.Methods: 210 IDC patients who were hospitalized in the Second Affiliated Hospital of Nanchang University from April 2019 to January 2022 were divided into two groups,namely the SLN negative group and the positive group.Analyze the clinical characteristics of the study subjects and compare whether the differences in age,maximum tumor diameter,SBR grade,ER or PR status,Ki-67 status,Her-2status,intravascular tumor thrombus,and nerve invasion between the two groups of patients are statistically significant.Binary logistic regression was used for multivariate analysis to find independent risk factors for IDC axillary SLN metastasis and establish a regression prediction model.The receiver operating characteristic（ROC）curve was used to evaluate the predictive performance of the regression model,and then the predictive model was externally validated.Results: Among 210 subjects,102 were SLN-positive patients（approximately48.6%）.There were 25 cases of Luminal A type,106 cases of Luminal B type,51 cases of Her-2 type,and 28 cases of triple negative;SLN positive patients were 11cases（about 44.0%）,45 cases（about 42.5%）,and 31 cases（about 42.5%）,respectively.60.8%),15 cases（about 53.6%）.There was no significant difference between SLN metastasis and tumor molecular type（χ2=5.123,P>0.05）.Maximum tumor diameter（χ₂=12.126,P<0.001）,ER or PR status（χ2=6.048,P=0.014）,Ki-67status（χ2=4.039,P=0.044）,intravascular tumor thrombus（χ2=69.159,P<0.001）and nerve invasion（χ2=16.139,P<0.001）between the two groups had statistically significant differences.Multivariate analysis of binary logistic regression showed that the maximum tumor diameter was greater than or equal to 2 cm（OR=2.488,95%CI:1.169-5.294,P=0.018）,Ki-67 was greater than or equal to 14%（OR=3.333,95% CI:1.110-10.008,P=0.032）,intravascular tumor thrombus（OR=13.382,95%CI:6.134-29.197,P<0.001）and nerve invasion（OR=5.608,95%CI:1.653-19.025,P=0.006）are independent risk factors for IDC axillary SLN metastasis.Therefore,a logistic regression model is established: logit（P）=-3.326 + 0.912X1-0.476X2 + 1.204X3 + 0.961X4 + 0.401X5 + 1.724X6 +2.594X7.The area under the ROC curve AUC was 0.847,the standard error was0.028,95%CI:0.739⁰.901,P<0.001;External validation shows that the specificity of the prediction model was 88.9%,the sensitivity was 71.6%;the positive predictive value was 85.9%,the negative predictive value was 76.8%;the accuracy was 80.5%;the Kappa value was 0.607,the standard error was 0.054,P<0.001.Conclusion: Molecular typing is not associated with SLN metastasis,the risk of SLN metastasis is higher in breast IDC patients with large tumor diameter,intravascular tumor thrombus,nerve invasion,and high Ki-67 expression.This logistic regression model has high predictive performance,and has a certain reference value for breast surgeons to determine whether SLN metastasis in breast IDC patients,and can help breast surgeons to determine whether to perform axillary lymph node dissection（ALND）.