CK19 mRNA in peripheral blood combined with contrast-enhanced ultrasound: the predictive value of axillary lymph node involvement in breast cancer

Abstract

Objective To evaluate a model for axillary lymph node involvement combining CK19 mRNA with contrast enhanced ultrasound sonography (CEUS) score in operable breast cancer. Methods Operable breast cancer patients planned for sentinel lymph node (SLN) biopsy were enrolled. Preoperative CK19mRNA expressions in peripheral blood and CEUS score of axillary lymph nodes were tested before surgery. In the training set, postoperative sentinel lymph node (SLN) and non-sentinel lymph node (nSLN) pathological results were taken as the gold standard, effective modeling variables were screened, logistic regression was used to establish the prediction model.Parallel control studies were conducted between the validation set and the MSKCC model to evaluate the prediction accuracy and prediction efficiency. Results From Oct 2015 to Nov 2016, 359 cases (training set) were enrolled and mathematical formulas for predicting SLN and nSLN were established, respectively. The sensitivity, specificity and AUC of predicting SLN involvement were 91.36%, 94.92% and 0.979 respectively. The sensitivity, specificity and AUC of predicting nSLN metastasis were 91.04%, 90.53% and 0.932 respectively. From Dec 2016 to Jul 2017, 219 cases (verification set) were included. The sensitivity of SLN metastasis predicted by the model was 91.84%, the specificity was 96.69%, and the AUC was 0.979, significantly superior to the MSKCC model (0.739). The sensitivity, specificity and AUC of predicting nSLN metastasis were 95.35%, 92.73% and 0.945 respectively, significantly superior to the MSKCC model (0.873). Concolusions Combined with peripheral blood CK19 mRNA and CEUS score, the prediction model for axillary lymph node involvement for operable breast cancer, SLN/nSLN involvement probability can be calculated and qualitative judgment can be made. The overall accuracy and AUC of this model are better than the prediction model of MSKCC. Key words: Breast neoplasms; RNA, messenger; Contrast-enhanced ultrasound; Lymph node involvement; Prediction