Utilizing conventional MRI to predict molecular subtypes of breast cancer.

Objective: The study aimed to investigate the correlation between conventional magnetic resonance imaging (cMRI) features and molecular subtypes of breast cancer (BC) to enhance non-invasive diagnostic accuracy. Materials and methods: A total of 153 patients with breast cancer were included in this retrospective analysis. Tumors were classified into molecular subtypes based on immunohistochemical (IHC) criteria. cMRI findings were analyzed according to the 2013 BI-RADS system. Univariate and multivariate statistical analyses were performed to identify the MRI features most predictive of each molecular subtypes. Results: This study analyzed cMRI features in 153 breast cancer patients, classified into five molecular subtypes: Luminal A (LA, 39.9%), Luminal B HER2-negative (LB (-), 22.9%), Luminal B HER2-positive (LB (+), 9.2%), HER2-enriched (HER-2, 17.6%), and triple-negative (TNBC, 10.4%). LA was associated with absence of lymph node involvement (OR = 2.97, p = 0.039), spiculated margins (OR = 2.24, p = 0.009), no surrounding edema (OR = 4.65, p = 0.008), and low T2 signal intensity (OR = 6.24, p = 0.001). HER-2 tended to present as non-mass enhancement lesions (OR = 6.7, p = 0.001) and washout kinetic patterns (OR = 2.37, p = 0.001). Circumscribed margins (OR = 3.38, p = 0.001) and rim enhancement (OR = 2.97, p = 0.008) were independent predictors of TNBC . There were no associations between cMRI features and LB (-), LB (+). Conclusion: This study suggests that cMRI features can help differentiate breast cancer subtypes, such as LA, HER-2, and TNBC, serving as useful supplements to imunohistochemistry and clinical data.