Objective: The purpose of our study was to describe the characteristics of probably benign breast MRI lesions and determine how these characteristics could be used to define the MRI BI-RADS 3 category.
Materials and methods: We prospectively collected morphology and kinetic data on lesions assessed as BI-RADS 3 in 2,569 consecutive breast MRI examinations from January 2003 through November 2006. The clinical indications for MRI, follow-up assessments, and pathology findings through May 2008 were collected from clinical records. Data were linked to the regional tumor registry to identify cases of malignancy in patients who did not follow-up at our institution. Frequency of BI-RADS 3 and cancer yield were calculated. Characteristics of probably benign lesions were analyzed for predictors of malignancy.
Results: Three hundred sixty-two lesions were assessed as BI-RADS 3 in 260 (10.1%) of 2,569 examinations in 236 patients. The 362 lesions included 168 (46%) foci, 132 (36%) nonmasslike enhancements, and 62 (17%) masses. Delayed kinetic information was available in 275 lesions. The most suspicious delayed kinetic enhancement was persistent in 164 (60%) of 275, plateau in 47 (17%) of 275, and washout in 64 (23%) of 275. The cancer yield in patients with a BI-RADS 3 assessment was two (0.85%) of 236; both were ductal carcinoma in situ lesions. There were no malignancies in the 69 foci with 100% persistent enhancement.
Conclusion: The characteristics of BI-RADS 3 lesions were highly variable in our population, and the risk of malignancy was low (0.85%). Assigning foci with 100% persistent enhancement to the BI-RADS 2 category can decrease the frequency of BI-RADS 3 assessment and maintain a likelihood of malignancy in less than 2% of cases.