Erschienen in:
01.09.2013 | Breast Oncology
A Model to Predict the Risk of Upgrade to Malignancy at Surgery in Atypical Breast Lesions Discovered on Percutaneous Biopsy Specimens
verfasst von:
Catherine Uzan, MD, PhD, Chafika Mazouni, MD, PhD, Malek Ferchiou, MD, Laura Ciolovan, MD, Corinne Balleyguier, MD, Marie-Christine Mathieu, MD, Philippe Vielh, MD, Suzette Delaloge, MD
Erschienen in:
Annals of Surgical Oncology
|
Ausgabe 9/2013
Einloggen, um Zugang zu erhalten
Abstract
Background
When any atypical feature is identified on a percutaneous biopsy specimen of a suspicious breast lesion, surgical excision is mandatory, leading to unnecessary surgeries in 70–90 % of the cases. The purpose of this study was to develop a model to predict the presence of cancer at surgery that would be applicable to all atypical lesions.
Methods
We collected complete clinical, radiological, and double-reading histological data concerning all patients with a diagnosis of a pure atypical lesion on image-guided biopsy performed at the One-Stop Breast Care Unit between 2004 and 2011.
Results
Among the 204 eligible patients, 49 cancers (24 %) had been diagnosed at definitive surgery (20 ductal carcinoma in situ, 20 invasive ductal, and 9 invasive lobular carcinoma). The univariate analysis retrieved age (p = 0.03), the focus size in mm (p = 0.02), the number of biopsy cores (p = 0.02), the disappearance of radiological anomalies after biopsy (p = 0.05), the mean number of atypical foci (p = 0.05) and the percentage of atypical lobules and ducts for lobular neoplasia (p = 0.04) as factors associated with cancer at surgery, whereas neither Ki67 nor ALDH1 expression was significantly correlated. The final most informative nomogram comprised information on patient age, the disappearance of radiological anomalies after biopsy and a focus size >15 mm. For the optimal threshold (risk of cancer = 21 %), sensitivity, specificity, positive predictive value, and negative predictive value were 78, 66, 36, and 90 %, respectively.
Conclusions
After validation, this model could help to identify a subset of patients with premalignant disease who could be spared surgery.