Breast lesions detected with MR imaging: utility and histopathologic importance of identification with US.

PURPOSE

To determine the utility and histopathologic importance of ultrasonographic (US) depiction of breast lesions detected with magnetic resonance (MR) imaging.

MATERIALS AND METHODS

Retrospective review was performed of 654 consecutive breast MR examinations performed from November 1999 to February 2001. This yielded 64 patients with 93 suspicious, nonpalpable, mammographically occult lesions evident on MR images and recommended for biopsy, for which directed US assessment was performed. Images, pathologic records, and medical records of these patients were reviewed for imaging findings, patient demographics, and histopathologic findings from subsequent biopsy. Statistical analysis was performed with the Fisher exact test.

RESULTS

The median size of MR abnormalities was 0.9 cm (range, 0.3-5.0 cm). The pattern of enhancement was a focal mass in 76 (82%) lesions and nonmass in 17 (18%). A US correlate was identified in 21 (23%) lesions detected with MR--19 (25%) of 76 focal mass and two (11%) of 17 nonmass lesions. Carcinoma was found in nine (43%) of these lesions, of which seven (78%) were invasive carcinoma and two (22%) were ductal carcinoma in situ (DCIS). Among the lesions without a US correlate, 10 (14%) yielded carcinoma, of which five (50%) were invasive carcinoma and five (50%) were DCIS. The frequency of cancer was significantly higher for lesions that were detected with MR imaging and had a US correlate than for those that did not have a US correlate (43% vs 14%, P =.01). Overall, 19 (20%) lesions detected with MR imaging and examined with US yielded carcinoma at biopsy, and nine (47%) malignant lesions were seen on US images. US depicted seven (58%) of 12 invasive cancers and two (29%) of seven instances of DCIS.

CONCLUSION

The likelihood of carcinoma was significantly higher among lesions with a US correlate (43% carcinoma) than lesions without a US correlate (14% carcinoma).