Construction of risk model for lymph node metastasis of breast cancer based on dynamic contrast-enhanced magnetic resonance imaging

YANG Yan; JIA Min; PENG Chuanyong

doi:10.3969/j.issn.1009-0460.2025.10.008

Abstract：Objective To construct a risk model of lymph node metastasis of breast cancer based on dynamic contrast-enhanced magnetic resonance imaging(DCE-MRI).Methods A total of 81 breast cancer patients who underwent surgical treatment at Lu'an People's Hospital from July 2021 to December 2024 were retrospectively enrolled.According to a 7:3 ratio,the patients were randomly allocated into a training set(n=57)and a validation set(n=24).Based on postoperative pathological results,patients in each dataset were further categorized into a lymph node metastasis group and a non-metastasis group.DCE-MRI results from preoperative examinations of all enrolled patients were collected.Binary logistic regression analysis was employed to identify DCE-MRI findings associated with breast cancer lymph node metastasis.Based on these findings,a risk prediction model was developed.The predictive performance of the model for evaluating lymph node metastasis in breast cancer was assessed using statistical methods including receiver operating characteristic(ROC)curve analysis,calibration curve evaluation,decision curve analysis,and external validation.Results In the training set,29 patients(50.88%)were pathologically confirmed to have lymph node metastasis,while 12 patients(50.00%)in the validation set had lymph node metastasis.Univariate analysis revealed no statistically significant differences in baseline characteristics between the two groups(P>0.05),while significant differences were observed in multiple DCE-MRI parameters(all P<0.05).).Specifically,the metastatic group exhibited predominant irregular/round lymph node morphology(58.62%vs.32.14%),reduced long-to-short-axis ratio(1.74±0.43 vs.2.08±0.24),increased cortical-medullary thickness ratio(1.15±0.35 vs.0.87±0.21),higher incidence of hilum absence(58.62%vs.25.00%),and elevated signal enhancement ratio(SER)values(185.01%±8.16%vs.176.94%±8.02%).Regression analysis identified the following as independent influencing factors for lymph node metastasis,including lymph node morphology(OR=13.173,95%CI:1.069-162.335,P=0.044),long-to-short-axis ratio(OR=0.024,95%CI:0.002-0.380,P=0.008),cortical-medullary thickness ratio(OR=1.927,95%CI:1.205-3.083,P=0.006),absent lymph node hilum(OR=18.241,95%CI:1.476-225.458,P=0.024),and SER(OR=1.205,95%CI:1.050-1.383,P=0.008).A risk scoring model constructed based on these factors demonstrated that a higher total score was associated with an increased risk of metastasis.Model validation in the training set showed strong agreement between the calibration curve and the ideal curve,an area under the ROC curve of 0.953(95%CI:0.905-1.000),and a positive net benefit in the decision curve within the threshold probability range of 0.03-0.99,with a maximum net benefit of 0.509.External validation further confirmed the model's generalizability,with an area under the ROC curve of 0.946(95%CI:0.894-0.998)in the validation set.Conclusion A DCE-MRI-based risk model for breast cancer lymph node metastasis demonstrates satisfactory performance in predicting nodal involvement and may serve as a non-invasive method for preoperative assessment of lymph node status in breast cancer patients.