Histologic Spectrum of Breast Imaging Reporting and Data System-4 (BIRADS- 4) Category Lesions – A Retrospective Study of 2015 Cases.
Dhananjay A Chitale, Azadeh Stark, Richard J Zarbo, David Nathanson, Mathew Burke, Naimei Tang, Felix Fernandez-Madrid. Henry Ford Hospital, Detroit, MI; Henry Ford Hospital, Detroit; Wayne State Univ., Detroit; Karmanos Comprehensive Cancer Institute, Detroit
Background: Annually about 1,700,000 women undergo breast biopsies with only 20% pathologically malignant. The economic cost of biopsies is estimated at $ 3.5 billion per annum. In addition, the psychological impact on women and possible complication of radiographic evaluation of future mammograms due to prior biopsy are the other significant concerns. Optimal clinical work-up and the need for a more standardized guideline for radiographically suspicious lesions remain contentious. We report the histologic spectrum and findings on positive predictive value (PPV) of sub-classification of mammographically suspicious lesions (BIRADS-4).
Design: We identified sequential cases of BIRADS-4 (n=2015) the radiology database that was then linked to pathology database. Sub-classification of BIRADS4: a/Low suspicion for malignancy, b/Intermediate suspicion for malignancy, c/ moderate concern but not classic for malignancy, were recorded. Cases were broadly stratified by their pathologic diagnoses into 4 groups: benign non-neoplastic, benign neoplastic, borderline [atypical duct/lobular hyperplasia] and malignant. Only the biopsy with final diagnosis was considered where more than one biopsy was performed. All statistical analyses were performed using SAS v. 9.1
Results: 221 women did not undergo biopsy and were excluded from analysis. Of 1794 biopsies, 984 (54.8%) were benign non-neoplastic, 403/1794 (22.5%) benign neoplastic; 41/1794 (2.3%) borderline histology, 366/1794 (20.4%) malignant. Only 435 (24.2%) of BIRADS- 4 were further sub-classified: 4a (69.1%, n=301), 4b (13.9%, n=60), 4c (17%, n=74). 61/74 (82.4%) patients with 4c score were diagnosed with either borderline or malignant lesions compared to 21/301 (7%) of 4a. PPV for BIRADS 4b was 31.7% in contrast to 82.4% of 4c (p<0.0001). Both these PPV were statistically significant compared to BIRADS-4a (p<0.0001).
Conclusions: Based on the present method of sub-classification of BIRADS4 lesions, BIRADS 4c has the highest PPV. BIRADS 4b provides added diagnostic value compared to BIRADS 4a, however with much lower PPV than 4c. Correlation of histologic features and clinical information with mammographic findings may help fine tune the radiologists' interpretations to reduce false positive results in BIRADS 4a/4b scores and thereby reduce the rate of unnecessary biopsies.
Tuesday, March 1, 2011 9:30 AM
Poster Session III # 34, Tuesday Morning