Background: Staging breast MRI can detect occult breast cancers in 1–10% of newly diagnosed patients. The performance of MRI, impact of patient factors, and resulting surgical management in underserved and diverse populations has not been well studied.

Methods We performed a retrospective analysis of consecutive patients with newly diagnosed in situ or invasive breast cancer (2006–2011) who had a preoperative staging MRI seen at our Centers that cares for an underserved and minority population. IRB approval was obtained to extract demographic and clinicopathological data, imaging studies and surgical treatment. Non-index lesions (NIL) were defined as those in breast or axillary nodes not known to be malignant or presenting with clinical, mammographic or ultrasound findings, in a different quadrant and given an MRI BIRADS score of 4 or 5. Occult cancers (OC) were those lesions found by biopsy or surgery to be invasive or in situ cancer. Logistic regression was used to examine the association between the probability of a NIL or OC and patients characteristics. Statistical computation was performed using STATA.

Results: Among 678 patients, 144 (21%) had a total of 171 NIL identified by MRI; 67 (9.8%) were ipsilateral, 44 (6.5%) contralateral and 24(3.5%)axillary nodes. 57 patients (8.4%) had a total of 62 OCs detected −34 (5%) ipsilateral, 8(1.2%) contralateral, and 9(1.3%) axillary nodes of which 49 (7.2%) were invasive and 8 (1.2%) in situ. 40% of patients with NILs were confirmed as OCs.

Analyses showed that age was marginally significantly associated with the probability of having OC(OR = 1.5 for each 10 years increase, p = 0.080). Patients with no BRCA mutation had significantly higher chance of having NILs tested with MRI (p = 0.003), but the probability of having OC did not significantly differ between those tested. Ipsilateral compared to contralateral NILs had a 4-fold higher chance of being OCs(p < 0.001). For NILs (n = 157), 59 (37.6%) underwent biopsy, 11 (7%) biopsy followed by lumpectomy, 10(6.4%) had lumpectomy, 24 (15.3%) mastectomy, 15 (9.6%) biopsy followed by mastectomy and (15)9.6% node dissection. Amongst OCs (n = 62) 6 (3.8%) underwent biopsy, 6 (3.8%) biopsy followed by lumpectomy, 6 (3.8%) lumpectomy, 20(12.7%) mastectomy, 12 (7.6%) had biopsy followed by mastectomy and 10 (6.4 %) node dissection.

Conclusions: In this diverse population, staging MRI detected NILs and OC in 21% and 8.4% of patients, respectively. Laterality and age seemed to be associated with chance of having OC. Interestingly, none of the OCs detected were among the known 32 mutation carriers. MRI staging appears to identify OC to an equal extent in a largely Hispanic and younger underserved population as reported in the literature for populations of mostly insured patients.

Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P4-01-11.