RESUMO
Objective:To explore the clinical application values of a nomogram based on preoperative breast MRI and axillary ultrasonography imaging parameters for predicting the risk of sentinel lymph node (SLN) metastasis in early-stage breast cancer patients.Methods:Three hundred and ninty-seven female patients (mean age 48.0±10.7 years old, range 25-81 years old) who admitted to Sun Yat-sen University Cancer Center from May 2007 to December 2017 were enrolled in this study. All patients were diagnosed as primary unilateral invasive early-stage breast cancer confirmed by surgical pathology. Preoperative breast MRI, axillary ultrasonography and clinical pathological data of enrolled patients were retrospectively analyzed. According to the pathological results of sentinel lymph node biopsy (SLNB), the cases were divided into negative SLN group ( n=200) and positive SLN group ( n=197). Clinicopathologic data, MRI and axillary ultrasound features were analyzed and compared between two groups. Logistic regression analysis was used to select independent risk factors. Then a predictive model was constructed and a nomogram was made for visualizing the associations between the predictive factors and SLN metastasis. Goodness-of-fit of the model was evaluated by using the Hosmer-Lemeshow test. Predictive performance was assessed based on the receiver operating characteristic (ROC) curves. Bootstrap resampling was performed for internal validation. Results:Significant differences were found in patient age, lymphovascular invasion status, PR status, HER2 status and molecular subtype between negative and positive SLN groups (all P<0.05); MRI features including tumor size, mass margin, long and short diameter, as well as the ratio of long to short diameter of LNs, LN margin, presence or absence of LN hilum, and axillary LNs symmetry were found significantly different between negative and positive SLN groups (all P<0.05); as for the axillary LN ultrasonography parameters, the interface between cortex and medulla, presence or absence of cortical thickening, and LN hilum were significantly different between negative and positive SLN groups (all P<0.05). Logistic regression analysis results showed that several factors could be identified as predictors of SLN metastasis, including patient age, MRI features (lymph node margin, presence or absence of lymph node hilum, and lymph node symmetry), axillary ultrasonography descriptors (presence or absence of cortical thickening) and pathological factors (lympovascular invasion, PR and HER2 status). The nomogram with patient age and the above imaging factors showed good,prediction performance with the area under the ROC as 0.778. Combining with the pathological parameters, the prediction performance of the nomogram model was significantly improved, yielding the area under the ROC of 0.866. Conclusions:The nomogram based on breast MRI and axillary ultrasonography can be applied as a noninvasive quantitative tool to predict the risk of SLN metastasis in early-stage breast cancer, which may facilitate decision-making for axillary treament strategy preoperatively.