Repeatability and reproducibility of cerebral 23Na imaging in healthy subjects

Abstract

Background: Initial reports of ²³Na magnetic resonance imaging (MRI) date back to the 1970s. However, methodological challenges of the technique hampered its widespread adoption for many years. Recent technical developments have overcome some of these limitations and have led to more optimal conditions for ²³Na-MR imaging. In order to serve as a reliable tool for the assessment of clinical stroke or brain tumor patients, we investigated the repeatability and reproducibility of cerebral sodium (²³Na) imaging in healthy subjects.

Methods: In this prospective, IRB approved study 12 consecutive healthy volunteers (8 female, age 31 ± 8.3) underwent three cerebral ²³Na-MRI examinations at 3.0 T (TimTrio, Siemens Healthineers) distributed between two separate visits with an 8 day interval. For each scan a T1w MP-RAGE sequence for anatomical referencing and a 3D-density-adapted, radial GRE-sequence for ²³Na-imaging were acquired using a dual-tuned (²³Na/¹H) head-coil. On 1 day, these scans were repeated consecutively; on the other day, the scans were performed once. ²³Na-sequences were reconstructed according to the MP-RAGE sequence, allowing direct cross-referencing of ROIs. Circular ROIs were placed in predetermined anatomic regions: gray and white matter (GM, WM), head of the caudate nucleus (HCN), pons, and cerebellum. External ²³Na-reference phantoms were used to calculate the tissue sodium content.

Results: Excellent correlation was found between repeated measurements on the same day (r² = 0.94), as well as on a different day (r² = 0.86). No significant differences were found based on laterality other than in the HCN (63.1 vs. 58.7 mmol/kg WW on the right (p = 0.01)). Pronounced inter-individual differences were identified in all anatomic regions. Moderate to good correlation (0.310 to 0.701) was found between the readers.

Conclusion: Our study has shown that intra-individual ²³Na-concentrations in healthy subjects do not significantly differ after repeated scans on the same day and a pre-set time interval. This confirms the repeatability and reproducibility of cerebral ²³Na-imaging. However, with manual ROI placement in predetermined anatomic landmarks, fluctuations in ²³Na-concentrations can be observed.

Keywords: 23Na (sodium) imaging; Cerebral magnetic resonance imaging; Oncologic imaging.

Fig. 1

Left: Example of region-of-interest (ROI) placement in the head of the caudate nucleus on a ²³Na-image; right: T1 MP-RAGE as an anatomical reference image

Fig. 2

Top: First examination (V1) on the first examination day with ²³Na-image, T1 MP-RAGE and DWI (from left to right); bottom: Second examination (V2) on the first examination day

Fig. 3

Total ²³Na-concentrations with standard deviations (SD) according to laterality (GM = gray matter; WM = white matter; HCN = head of the caudate nucleus); * indicates significant difference with p = 0.01

Fig. 4

Bland-Altman plot, plotting the ratio of the two readers versus the average of both readers. The lines and given numbers on the right side represent the mean of the ratio and the +/− 1.96 standard deviation (SD) of the ratio