Abstract Background: Accurate assessment of the extent of breast cancer with breast MRI (BMRI) yields additional findings (AF) that must be characterized in order to guide treatment. The aim of our study was to analyze the work-up of AF seen in pre-therapeutic BMRI and their impact on therapeutic approach and re-excision rates.Methods: From 07/2002 to 04/2007 we studied prospectively 465 consecutive patients with a diagnosis of breast cancer that underwent a pre-treatment BMRI to evaluate disease extent. BMRI studies were performed with a 1,0 and a 1,5 T magnet and coronal T2-weighted fast spin-echo and contrast-enchanced T1-weighted 3D gradient-echo sequences were evaluated. Post-processing included parametric, multiplanar reconstructions and maximum intensity projections. Fischer's scale was used to characterize additional lesions (AL). AL >5 mm were classified as: increase in tumor size, multifocal, multicentric and contralateral disease. AF that would potentially change therapeutic approach were re-evaluated with 2nd-look ultrasound (US), biopsied if found and/or marked with clips or with a radioisotope (ROLL). BI-RADS 3 lesions were followed up with BMRI. Gold standards were pathology reports and follow-up >2 years for benign lesions. Therapeutic change based on MRI findings was deemed correct in malignant AF and incorrect if pathology was benign. Re-excision rates and disease-free intervals were calculated.Results. A total of 280 AF were found in 222 (47,7%) patients. US studies were performed in 111 patients, in 99 of which the AL (89,1%) was found. 63 patients underwent biopsies and 48 of them (76,1%) were malignant. ROLL procedures were performed in 32 patients and malignancy rate was 65,6%. Follow-up BMRI studies for BI-RADS 3 lesions were performed in 42 patients (9%). BMRI found index lesions seen by mammography or US in 98,9% of the patients and did not find additional multifocal or multicentric disease in 10 patients, yielding a total negative predictive value of 96,7% for BMRI (6 invasive and 9 intraductal cancers). Therapeutic approach was changed in 107 patients (23%) and considered correct in 92 (86%) representing 19,7% of all patients and incorrect in 15 (14%) corresponding to 3,2% of the total. Re-excision due to positive margins during the first 6 months was performed in 39/323 patients treated with conservative surgery (re-excision rate of 12%). Mean follow-up was 48,4 months (m) (6,7-81,4 m) and mean disease-free interval was 45 m (6,7-81,4 m).Conclusion: Work-up of AF diagnosed in BMRI allows accurate treatment adjusted to disease extent in 96,8% of the patients. Additional work-up using US and close correlation with the rest of conventional modalities is the mandatory behaviour to ensure an integrated evaluation of AF. This policy allows a correct treatment in 86% of the patients with potential therapeutic change. Overtreatment was induced in 14% of these patients. Regarding the overall series of 465 patients submitted to BMRI, the 15 patients who were overtreated represent 3,2% of the total. The precision level of preoperative MRI in this scenario is clearly superior to conventional modalities when combined in an integrated multimodality approach. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 4023.