Preoperative Magnetic Resonance Imaging Evaluation Improves Breast Conserving Surgery Rate but Not Local Disease Control in Patients with Ductal Carcinoma in Situ

Abstract

Abstract Background: The role of preoperative evaluation by magnetic resonance imaging (MRI) in breast ductal carcinoma in situ (DCIS) management remains controversial. The study aims to assess the associations of preoperative MRI, surgery type, and disease outcome in DCIS patients. Methods: Patients with DCIS who received primary surgery in two breast centers between Jan. 2009 and Jan.2021 were retrospectively included and subdivided into MRI and non-MRI cohorts. The target lesion’s largest diameters (LD) measured on ultrasound, MRI, and histopathology were compared. Associations between MRI, clinicopathological characteristics, surgery type, and prognosis were evaluated after propensity score matching (PSM). Results: A total of 1351 patients were eligible and included. Among 708 cases with complete LD data, the mean of LD measurements for histopathology, ultrasound, and MRI was 19.6±14.0 mm, 21.7±13.0 mm, and 21.9±13.9 mm, respectively. The overall correlation for LD measurements between MRI and histopathology (ρ=0.657) was stronger than that between ultrasound and histopathology (ρ=0.574; P<0.001). LD measurements were significantly more accurate measured by MRI than by ultrasound compared with histopathology (51.3% vs. 45.0%; P=0.033). After PSM, 742 patients were further analyzed. Patients in MRI cohort were more likely to receive breast-conserving surgery (BCS) than those in non-MRI cohort (28.6% vs. 19.1%; P=0.003). For 227 patients planning to receive BCS as primary surgery, initial negative surgical margin was similar between MRI and non-MRI cohorts (73.6% vs. 71.9%; P=0.485). There was no difference in second mastectomy rate between MRI and non-MRI cohorts (22.0% vs. 23.7%; P=0.433). Locoregional recurrence and cumulative breast cancer incidence were similar between two cohorts. Conclusion: Preoperative MRI evaluation for DCIS patients provided a more accurate estimation of lesions than ultrasound. It was associated with a higher BCS rate but no improvement of positive surgical margin.