Background: For ductal intraepithelial neoplasia, grade 1B, studies that predict breast cancer risk after an 11-gauge vacuum-assisted breast biopsy have yielded contradictory results. In order to identify a predictive model of breast cancer risk, we assessed the underestimation rate according to radiological and clinical findings.
Patients and methods: Our study involved 212 patients. We compared the area under the receiver operating characteristic curves and the clinical utility of a logistic regression and partitioning model.
Results: Overall upgrade to malignancy occurred in 42 (19.8%) out of the 212 cases. The area under the curve for the logistic regression and partitioning model were 0.65 (95% confidence interval=0.61-0.70) and 0.58 (95% confidence interval=0.54-0.62), respectively. The lowest predicted underestimation rate obtained with the logistic regression model was 9.5%.
Conclusion: From this large series, we were unable to define any accurate safety model for breast cancer. Surgery should be thus recommended.