Application value of prediction model based on multi-dimensional indicators in axillary lymph node metastasis in breast cancer patients

ObjectiveTo explore the application value of establishing a prediction model based on multi-dimensional indicators in axillary lymph node(ALN) metastasis in breast cancer patients.

MethodsA total of 242 female patients with unilateral breast cancer were retrospectively selected as the research subjects.The presence or absence of ALN metastasis were confirmed by pathological test results.Patients were divided into metastatic group(n=66) and non-metastatic group(n=176).The clinical data of two groups were compared.Multivariate logistic regression analysis was used to analyze the risk factors of ALN metastasis in breast cancer patients and establish a prediction model.ROC curve was used to analyze the discrimination of the model, and goodness of fit was used to test the calibration of the model.Another 93 female patients with unilateral breast cancer were selected for clinical validation of the model.

ResultsMultivariate logistic regression analysis showed that blurred tumor margin(OR=1.912), thick ALN cortex(OR =2.789), short ALN diameter(OR=2.280), large ratio of short ALN diameter to long ALN diameter(OR=3.773), high grade of pathological(OR=2.101), and high expression of miRNA-203(OR=3.529) were risk factors for ALN metastasis in breast cancer patients(P < 0.05 to P < 0.01).The prediction model equation derived from the risk factors was Prob =1/(e^-Y), Y=-30.724+0.648×tumor margin blur+1.026×ALN cortical thickness+0.824×ALN short diameter+1.328×ALN short diameter/long diameter ratio+0.742×pathological grade +1.261×miRNA-203 expression level.ROC curve analysis showed that the AUC area of ALN metastasis in breast cancer patients predicted by the model was 0.889(95%CI: 0.840-0.934), indicating that the model had good discrimination.Goodness of fit test was χ²=2.06, P>0.05, indicating that the model had no overfitting phenomenon.The sensitivity, specificity and accuracy of the prediction model were 88.46%, 83.58% and 84.95%, respectively.

ConclusionsThe prediction model of ALN metastasis in breast cancer patients constructed by serum miRNA-203 expression, tumor pathological grade, tumor margin and cortical thickness of ALN, short diameter, and short diameter/long diameter ratio has certain value.