Objective To investigate the predictive value of ultrasound (US), mammography (MG), and magnetic resonance imaging (MRI) for histological grading and molecular subtyping of ductal carcinoma in situ (DCIS) of the breast.

Methods A retrospective analysis was conducted on the clinical data of 379 patients with DCIS in the Yuebei People's Hospital Affiliated to Medical College of Shantou University. All patients were diagnosed with surgery or biopsy and underwent preoperative US, MG, and MRI examinations. Patients were categorized into Luminal A (n=174), Luminal B (n=114), human epidermal growth factor receptor-2 (HER-2)-positive (n=76), and triple-negative (n=15) subtypes based on molecular subtyping. According to histological grading, patients were divided into high-grade (n=213) and non-high-grade (n=166) groups. The predictive values of US, MG, and MRI for histological grading and molecular subtyping of DCIS were analyzed by receiver operating characteristic (ROC) curve.

Results The proportions of high-grade patients with non-mass-like features, microcalcifications on US, calcifications on MG, and type Ⅲ time-signal intensity curve(TIC) classification on MRI were higher than those in non-high-grade patients. The proportions of high-grade patients with spiculated margins on MG and apparent diffusion coefficient (ADC) values on MRI were significantly lower than those in non-high-grade patients (P < 0.05). Significant differences (P < 0.05) were observed among the four molecular subtypes in terms of lesion shape, lesion morphology, lesion boundary, spiculated margins, posterior echo, microcalcifications, and blood flow Adler grading on US (P < 0.05), the lesion shape, spiculated margins, calcifications, calcifications with mass, and architectural distortion on MG (P < 0.05), and lesion margin, lesion morphology, internal enhancement characteristics, ADC values, volume transfer constant (K_trans), reverse reflux rate constant (K_ep), and extracellularextravascular volume fraction (V_e) on MRI (P < 0.05). ROC curve analysis revealed that the combined use of US, MG, and MRI for predicting histological grading of DCIS exhibited higher sensitivity and area under the curve (AUC) than individual modalities. Similarly, the combined prediction of Luminal, HER-2-positive, and triple-negative DCIS subtypes demonstrated higher sensitivity and AUC than individual predictions.

Conclusion The combined use of US, MG, and MRI exhibits favorable predictive value for histological grading and molecular subtyping of DCIS. Analyzing the predictive effects of multimodal imaging on histological grading and molecular subtyping of DCIS can provide a reference basis for clinicians to assess disease severity and specific disease types.