Feasibility Study Of Predicting Axillary Lymph Node Metastasis In Patients With Early Invasive Breast Cancer Based On Nomogram

Objective:A prediction model was established based on clinicopathological features combined with MRI imaging features to predict ALNM in patients with early invasive BC,and a nomogram was drawn to verify and evaluate the prediction model.Methods A total of 359 female patients admitted to the First Affiliated Hospital of Hebei North University were selected and randomly divided into modeling group(n=269)and verification group(n=90)at a ratio of 3:1.Clinicopathologic features(including:patients age,histological grade of primary tumor,ER,PR,HER-2 status,Ki-67 expression,LVI and molecular subtypes)and MRI imaging features(including:primary tumor size,location,multifocal,lymph node status reported by MRI,ADC value,TIC type,peritumoral edema,and glands type)were evaluated in the modeling group for univariate analysis.Multivariate Logistic regression analysis was conducted for the variables with statistical significance in univariate analysis.Based on the results of multivariate Logistic regression analysis,a prediction model was established and a nomogram was drawn.The ROC,DCA,AUC,c-index,specificity,sensitivity,Hosmer-Lemeshow test and Delong test of the two groups were used as indicators to judge the prediction ability and goodness of fit of the model.P<0.05 was statistically significant.Results:There were no significant differences in all clinicopathological and MRI features between the modeling group and the validation group.In the univariate analysis of the modeling group,the high expression of Ki-67,LVI positive,tumor size,multifocal,MRI reported ALN positive and peritumoral edema were statistically significant.Multivariate Logistic regression analysis showed that LVI positive,tumor≥2cm,multifocal,MRI reported ALN positive,and peritumoral edema were independent risk factors for ALNM in BC patients.A nomogram was drawn based on the influence of these five factors on ALNM.The AUC,c-index,specificity and sensitivity of the modeling group were 0.904(95%CI:0.863-0.944),0.864,0.809 and 0.861,respectively.The AUC,c-index,specificity and sensitivity of the validation group were 0.894(95%CI:0.823-0.966),0.862,0.682 and 0.909,Hosmer-Lemeshow test(modeling group:χ ~2=2.069,P=0.979,validation group:χ ~2=5.287,P=0.727),Delong test(P=0.823).Conclusions:In this study,LVI positive,tumor≥2cm,multifocal positive,MRI reported ALN positive,peritumoral edema positive are independent risk factors for ALNM in patients with BC.The prediction model and various indicators proved that its prediction effect is good,which can provide a reference for the development of treatment plan and prognosis of BC patients.