The connecticut experiments second year: ultrasound in the screening of women with dense breasts.

To determine if the addition of screening breast ultrasound in women with mammographically normal but dense breasts improves breast cancer detection. The study utilized a retrospective chart review. Data collected included: (a) total number of screening mammograms; (b) total number of dense breast screening ultrasounds; (c) screening ultrasound Breast Imaging Reporting Data System (BI-RADS) code results; (d) biopsy results; and (e) demographic data on women with malignant biopsies. Data were obtained from sites throughout Connecticut from November 1, 2010 to October 31, 2011. Data from 5 Connecticut radiology practices covering 10 sites were collected. Sites conducted a total of 57,417 screening mammograms and 10,282 dense breast screening ultrasounds. Of the screening ultrasounds, 87% (8,972/10,282) were BI-RADS 1 or 2, 9% (875/10,282) were BI-RADS 3, 4% (435/10,282) were BI-RADS 4 or 5, and 39 were found to have a cancer or high-risk lesion on biopsy. This correlates to 3.8 cancers or high-risk lesions per 1,000 women screened. If high-risk lesions are excluded, there are 24 cases of biopsy proven malignancy corresponding to 2.3 cancers per 1,000 women screened. In this study, screening breast ultrasound in women with mammographically normal but dense breasts demonstrated a positive predictive value of 9% (39/435) and specificity of 96% (8,972/9,368). Based on the data collected from sites throughout Connecticut, screening breast ultrasound in women with dense breast parenchyma detects mammographically occult malignancy and high-risk lesions. The results are especially significant given recent studies suggesting that breast density is an independent risk factor for breast cancer and that mammography is less effective in detecting cancer in dense breasts. The improved specificity and sensitivity between the 1st and 2nd years' suggests there is a learning curve that may continue to improve the results.