Reducing false-positive biopsies: a pilot study to reduce benign biopsy rates for BI-RADS 4A/B assessments through testing risk stratification and new thresholds for intervention.

The aim of this study is to evaluate Breast Imaging Reporting and Data Systems (BI-RADS) 4A/B subcategory risk estimates for ductal carcinoma in situ (DCIS) and invasive cancer (IC), determining whether changing the proposed cutoffs to a higher biopsy threshold could safely increase cancer-to-biopsy yields while minimizing false-positive biopsies. A prospective clinical trial was performed to evaluate BI-RADS 4 lesions from women seen in clinic between January 2006 and March 2007. An experienced radiologist prospectively estimated a percent risk-estimate for DCIS and IC. Truth was determined by histopathology or 4-year follow-up negative for malignancy. Risk estimates were used to generate receiver-operating characteristic (ROC) curves. Biopsy rates, cancer-to-biopsy yields, and type of malignancies missed were then calculated across postulated risk thresholds. A total of 124 breast lesions were evaluated from 213 women. An experienced radiologist gave highly accurate risk estimates for IC, DCIS alone, or the combination with an area under ROC curve of 0.91 (95 % CI 0.84-0.99) (p < 0.001), 0.81 (95 % CI 0.69-0.93) (p = 0.011), and 0.89 (95 % CI 0.83-0.95) (p < 0.001), respectively. The cancer-to-biopsy yield was 30 %. Three hypothetical thresholds for intervention were analyzed: (1) DCIS or IC ≥ 10 %; (2) DCIS ≥ 50 % or IC ≥ 10 %; and (3) IC ≥ 10 %, which translated to 22, 48, and 56 % of biopsies avoided; cancer-to-biopsy yields of 36, 47, and 46 %; and associated chance of missing an IC of 0, 1, and 2 %, respectively. Expert radiologists estimate risk of IC and DCIS with a high degree of accuracy. Increasing the cut off point for recommending biopsy, substituting with a short-term follow-up protocol with biopsy if any change, may safely reduce the number of false-positive biopsies.