Construction And Validation Of A Prediction Model Of Non-Sentinel Lymph Node Metastasis In Early Breast Cancer Patients

ObjectiveIn the era of precision medicine,accurate assessment of the status of axillary nonsentinel lymph nodes(NSLN)is an important part of achieving individualized treatment,minimizing harm and maximizing benefit for patients with early-stage invasive breast cancer.The aim of this study is to construct a risk prediction model for positive NSLN in patients with early-stage invasive breast cancer based on the independent predictors obtained by statistical analysis and screening-the Alignment diagram,and then to verify the model internally and externally and discuss its clinical application prospect.MethodsA total of 337 eligible early-stage invasive breast cancer patients with positive sentinel lymph nodes were included in this study.These cases were diagnosed between January 1,2017 and December 1,2022,and underwent axillary lymph node dissection(ALND)and Total mastectomy,which were divided into training set(237 cases)and test set(100 cases)in a ratio of 7:3.Binary logistic regression statistical analysis was used to screen out the independent predictors of NSLN positive in the training set.R language software was used to visualize the binary logistic regression model.Internal and external validation of the model was carried out.The Bootstrap self-sampling number was set to 1000,and draw ROC curves of training set and test set respectively to judge the accuracy of model prediction.The calibration curve combined with Hosmer-Lemeshow test was used to determine the fitting of the model.DCA chart was drawn to evaluate whether the the model could improve clinical outcomes.ResultsAmong 337 patients with early invasive breast cancer,195(57.9%)had positive NSLN.Statistical analysis showed that Ki67(P=0.042,OR:2.046,95%CI:1.0254.084),the number of positive SLN(P=0.007,OR:9.061,95%CI:1.803-45.543),the number of negative SLN(P<0.001,OR:0.078,95%CI:0.026-0.237),SLN tumor budding(P=0.019,OR:3.184,95%CI:1.206-8.403),SLN extra-nodal extension(P=0.030,OR:2.862,95%CI:1.109-7.387)were independent predictors of positive NSLN.The P value of the likelihood ratio test was less than 0.001,indicating that the model was successful.The AUC of the training set and the test set were 0.851 and 0.863,respectively,indicating that the model had good discrimination.The HosmerLemeshow test(P=0.791)and calibration curve indicated that the model had a good degree of fit.DCA threshold probability is in the range of 5%-96%,and This prediction model will obtain net benefits,otherwise the model is invalid.Conclusion1.Nerve invasion,lymphovascular invasion,Ki-67 proliferation index,SLN tumor budding,SLN metastatic size,SLN extra-nodal extension,SLN lymphovascular invasion,SLN negative and positive number are significantly correlated with the positive rate of NSLN in patients with early invasive breast cancer.2.Ki-67 proliferation index,SLN tumor budding,SLN extra-nodal extension,the number of negative and positive SLN are independent predictors of the positive of NSLN in patients with early-stage invasive breast cancer.3.The prediction model of positive NSLN in early-stage invasive breast cancer patients was constructed and verified.The results shows that the model has excellent discrimination and fitting degree,and had potential application prospects to assist clinical decision-making.