A validated web-based nomogram for predicting positive surgical margins following breast-conserving surgery as a preoperative tool for clinical decision-making

Abstract

Breast-conserving therapy, consisting of lumpectomy and adjuvant radiotherapy, is considered standard treatment for early-stage breast cancer. One of the most important risk factors of local recurrence is the presence of positive surgical margins following lumpectomy. We aimed to develop and validate a predictive model (nomogram) to predict for positive margins following the first attempt at lumpectomy as a preoperative tool for clinical decision-making. Patients with clinical T1-2N0-1Mx-0 histology-proven invasive breast carcinoma who underwent BCT throughout the North-East region of The Netherlands between June 2008 and July 2009 were selected from the Netherlands Cancer Registry (n = 1185). Results from multivariate logistic regression analyses served as the basis for development of the nomogram. Nomogram calibration and discrimination were assessed graphically and by calculation of a concordance index, respectively. Nomogram performance was validated on an external independent dataset (n = 331) from the University Medical Center Groningen. The final multivariate regression model included clinical, radiological, and pathological variables. Concordance indices were calculated of 0.70 (95% CI: 0.66-0.74) and 0.69 (95% CI: 0.63-0.76) for the modeling and the validation group, respectively. Calibration of the model was considered adequate in both groups. A nomogram was developed as a graphical representation of the model. Moreover, a web-based application (http://www.breastconservation.com) was build to facilitate the use of our nomogram in a clinical setting. We developed and validated a nomogram that enables estimation of the preoperative risk of positive margins in breast-conserving surgery. Our nomogram provides a valuable tool for identifying high-risk patients who might benefit from preoperative MRI and/or oncoplastic surgery. Copyright © 2013 Elsevier Ltd. All rights reserved.