A preoperative nomogram to predict the risk of synchronous distant metastases at diagnosis of primary breast cancer.

ABSTRACT

BACKGROUND: The detection of synchronous metastases at primary diagnosis of breast cancer (BC) affects its initial management. A risk calculator that incorporates many factors to evaluate an individual's risk of harbouring synchronous metastases would be useful to adapt cancer management. PATIENTS AND METHODS: Patients with primary diagnosis of BC were identified from three institutional databases sharing homogeneous work-up recommendations. A risk score for synchronous metastases was estimated and a nomogram was constructed using the first database. Its performance was assessed by receiver characteristic (ROC) analysis. The nomogram was externally validated in the two independent cohorts. RESULTS: A preoperative nomogram based on the clinical tumour size (P<0.001), clinical nodal status (P<0.001), oestrogen (P=0.17) and progesterone receptors (P=0.04) was developed. The nomogram accuracy was 87.3% (95% confidence interval (CI), 84.45-90.2%). Overall, the area under the ROC curve (AUC) was 86.1% for the validation set from the Institut Curie-René Huguenin, and 63.8% for the MD Anderson validation set. The negative predictive value (NPV) was high in the three cohorts (97-99%). CONCLUSIONS: We developed and validated a strong metastasis risk calculator that can evaluate with high accuracy an individual's risk of harbouring synchronous metastases at diagnosis of primary BC. CONDENSED ABSTRACT A nomogram to predict synchronous metastases at diagnosis of breast cancer was developed and externally validated. This tool allows avoiding unnecessary expensive work-up.