Prediction of axillary lymph node status in invasive breast cancer with dynamic contrast-enhanced MR imaging

PURPOSE: To determine if magnetic resonance (MR) imaging can be used to predict axillary lymph node status in patients with breast cancer. MATERIALS AND METHODS: Fifty-one women with primary invasive breast cancer underwent dynamic contrast material-enhanced MR imaging of the breast Region-of-interest (ROI) analysis was performed on parametric images obtained with kinetic modeling of the data. Large and automated ROIs were selected. Typical enhancement ratios that represented the relative increase in mean pixel signal intensity were calculated for each ROI. Stepwise logistic regression analysis was applied to identify prognostic factors of axillary node status. Receiver operating characteristic analysis was performed and a Prier score and calibration curve were calculated to assess the diagnostic efficacy and predictive capability of the logistic regression model. RESULTS: The maximum enhancement ratio of the automated ROI was found to be the strongest predictor of node status (P < .001). Patient age (P = .007) and ROI size (P = .045) were also significant predictor variables. The model showed good accuracy (area beneath the fitted binormal receiver operating characteristic curve [A(z)] = 0.90; Prier score, 0.133). In 12 (24%) of the patients, a less than 5% or greater than 95% probability of positive-node status was correctly identified. CONCLUSION: The Suggested predictive model may decrease the need for surgical staging of the axilla in patients with breast cancer.