Time-Dependent Diffusion MRI-Based Microstructural Mapping for Characterizing HER2-Zero, -Low, -Ultra-Low, and -Positive Breast Cancer.

Background: With breast cancer treatment advances, accurate non-invasive methods are needed to distinguish its human epidermal growth factor receptor 2 (HER2) subtypes. Recently developed time-dependent diffusion MRI (t_d-dMRI) has potential in characterizing cellular tissue microstructures in breast cancer. However, its role in identifying HER2 subtypes is unknown.

Purpose: To investigate the feasibility of t_d-dMRI-based microstructural histogram parameters for characterizing properties of four HER2 subtypes in breast cancer.

Study type: Prospective.

Population: Four hundred ninety-five participants with invasive breast cancer (18 HER2-zero, 49 -ultralow, 243 -low and 185 -positive).

Field strength/sequence: 3-T, oscillating gradient spin-echo (OGSE) and pulsed gradient spin-echo (PGSE) sequences for t_d-dMRI.

Assessment: The HER2 status was categorized as HER2-zero, -ultralow, -low, or -positive by immunohistochemistry and fluorescence in situ hybridization. The t_d-dMRI data were fitted using the IMPULSED method. Tumors were identified on dynamic contrast-enhanced MRI and delineated on the PGSE image (b = 0 s/mm²). Forty-nine histogram parameters were extracted from the tumor, including four microstructural maps (diameter, intracellular fraction, extracellular diffusivity, cellularity) and three apparent diffusion coefficient maps.

Statistical tests: Histogram parameters were analyzed via one-way analysis of variance followed by pairwise t tests with Bonferroni correction. The Boruta method selected the significant parameters for each HER2 subtype. The predictive performance was assessed through area under the curve (AUC). A p value < 0.05 was considered statistically significant.

Results: Thirty-two histogram parameters showed significant differences among the four HER2 subgroups. Four models were constructed, which achieved high performance for distinguishing HER2-positive versus negative (AUC of 0.85), HER2-positive versus low (AUC of 0.87), and HER2-low versus immunohistochemistry 0 (AUC of 0.81), along with moderate performance for distinguishing HER2-zero versus -ultralow (AUC of 0.77).

Data conclusion: Selected t_d-dMRI-derived histogram parameters may be applicable for identifying HER2 subtypes in breast cancer.

Level of evidence: 1:

Technical efficacy stage: 2.