Abstract
OBJECTIVES: Deuterium ( 2 H) metabolic imaging (DMI) is an emerging magnetic resonance technique to non-invasively map human brain glucose (Glc) uptake and downstream metabolism following oral or intravenous administration of 2 H-labeled Glc. The achievable spatial resolution is limited due to inherently low sensitivity of DMI. This hinders potential clinical translation. The purpose of this study was to improve the signal-to-noise ratio (SNR) of 3D DMI via a balanced steady-state free precession (bSSFP) acquisition scheme combined with fast non-Cartesian spatial-spectral sampling to enable high-resolution dynamic imaging of neural Glc uptake and glutamate+glutamine (Glx) synthesis of the human brain at 7T. MATERIALS AND METHODS: Six healthy volunteers (2 f/4 m) were scanned after oral administration of 0.8 g/kg [6,6']- 2 H-Glc using a novel density-weighted bSSFP acquisition scheme combined with fast 3D concentric ring trajectory (CRT) k-space sampling at 7T. Time-resolved whole brain DMI datasets were acquired for approximately 80 minutes (7 minutes per dataset) after oral 2 H-labeled Glc administration with 0.75 mL and 0.36 mL isotropic spatial resolution and results were compared to conventional spoiled Free Induction Decay (FID) 2 H-MRSI with CRT readout at matched nominal spatial resolution. Dynamic DMI measurements of the brain were accompanied by simultaneous systemic Glc measurements of the interstitial fluid using a continuous Glc monitoring (CGM) sensor (on the upper arm). The correlation between brain and interstitial Glc levels was analyzed using linear mixed models. RESULTS: The bSSFP-CRT approach achieved SNRs that were up to 3-fold higher than conventional spoiled FID-CRT 2 H-MRSI. This enabled a 2-fold higher spatial resolution. Seventy minutes after oral tracer uptake comparable 2 H-Glc, 2 H-Glx, and 2 H-water concentrations were detected using both acquisition schemes at both, regular and high spatial resolutions (0.75 ml and 0.36 mL isotropic). The mean Areas Under the Curve (AUC) for interstitial fluid Glc measurements obtained using a CGM sensor was 509 ± 65 mM·min. This is 3.4 times higher than the mean AUC of brain Glc measurements of 149 ± 43 mM·min obtained via DMI. The linear mixed models fitted to assess the relationship between CGM measures and brain 2 H-Glc yielded statistically significant slope estimates in both GM (β 1 = 0.47, P = 0.01) and WM (β 1 = 0.36, P = 0.03). CONCLUSIONS: In this study we successfully implemented a balanced steady-state free precession (bSSFP) acquisition scheme for dynamic whole-brain human DMI at 7T. A 3-fold SNR increase compared to conventional spoiled acquisition allowed us to double the spatial resolution achieved using conventional FID-CRT DMI. Systemic continuous glucose measurements, combined with dynamic DMI, demonstrate significant potential for clinical applications. This could help improve our understanding of brain glucose metabolism by linking it to time-resolved peripheral glucose levels. Importantly, these measurements are conducted in a minimally invasive and physiological manner.