Malignancy Rate, Number Needed to Treat, and Positive Predictive Value for Breast MRI

Breast MRI is being used more frequently for advanced screening for breast cancer. Patients may be at increased risk, or are symptomatic, with nonsuspicious mammograms. There is little data regarding the likelihood of a recommendation for biopsy, or for detecting a malignancy, in this population. We intended to determine the malignancy rate, number needed to treat, and positive predictive value for patients receiving adjunctive MRI at our institution. A retrospective review of all breast MRIs from 2008 to 2010 was done. Patients with any prior diagnosis of breast cancer, or BRCA+ were excluded. There were 324 patients. Most common reasons for ordering the breast MRI included: abnormal test result 130 (44%), palpable mass 74 (23%), family history 58 (18%), breast pain 47 (15%), and nipple discharge 45 (14%). Breast Imaging-Reporting and Data System score (BIRADS) was 1 or 2 in 36 per cent, 4 or 5 in 18 per cent, 3 in 26 per cent, 0 in 10 per cent, and not given in 9 per cent. Biopsy was recommended in 77 (24%), with biopsy actually performed in 57 (18%). Of the eight cancers identified, four (1.2%) were ductal carcinoma in situ (DCIS) and four (1.2%) were invasive cancer, yielding a true-positive rate of 2.5 per cent. Number needed to treat was 40. Positive predictive value was 14 per cent with a false-positive rate of 86 per cent. In this group of generally higher risk women, typically prescreened with mammography, 1.2 per cent had an invasive cancer, and another 1.2 per cent had DCIS. Those who undergo biopsy are 6.1 times more likely to have benign pathology. The efficacy of adjunctive breast MRI could be improved through refinements in indication, test interpretation, or alternative screening strategies.