To investigate the application of quantitative sodium (23Na) MRI technique for characterizing breast lesions before biopsy as part of a comprehensive 1H /23Na MRI protocol.

Registered 23Na and 1H images were recorded on a 1.5T GE Signa scanner with a 23Na coil inserted in a 1H phased array breast coil. 1H images consisted of sagittal T1-weighted (T1w), T2-weighted (T2w) images and 3D T1w images, pre- and post-administration of GdDTPA. T1w images were acquired with a Gradient Recalled Echo (GRE) sequence (TR/T1=250/4 ms). T2w images (Fast Spin Echo sequence, TR/TE = 5700/99 ms) and Pre- and post-Gd contrast images (3D-GRE sequence, TR/TE = 23/4 ms) were recorded with fat suppression. 23Na images (TR/TE = 100/0.4 ms) were acquired using adiabatic excitation with twisted projection imaging in 12 min for a0.2 ml voxel size. An embedded coil phantom with 150 mM NaCl served as fiducial marker and external reference for receive sensitivity corrections. ROIs Lesion ROIs were identified from post-Gd contrast images. Glandular and adipose tissue ROIs were identified from T2w and T1w images. TSCs were calculated from mean signals found in patients’ ROIs and signals from identical regions in a concentration reference scan. Eighteen patients (age 54±8, range 33-66) were studied and biopsies were obtained after the MR study.

Sixteen patients had malignant breast lesions with a significantly increased TSC of 51±13 mM, versus 32±12 mM in non-involved glandular tissue (57% higher p < 0.001, N=16). In adipose tissue, TSC was 17±5 mM (N=18). Other patients had fibrocystic changes (n=1) and lymphoma (n=1) and the average TSC in those lesions was 33 ± 4, not significantly different from non-involved glandular tissue.

The TSC in biopsy-proven malignant lesions was consistently elevated compared to non-involved breast tissues and non-malignant breast lesions. Thus, adding quantitative 23Na to a comprehensive 1H MRI protocol could improve specificity of diagnosis of breast cancer and provide molecular information on the complex breast environment. Grant support: NIH/NCI R21CA095907, 1R01CA100184, P50CA103175.