Synopsis

Keywords: Deuterium, Metabolism

We present an estimation of the minimal dose for D-glucose needed for evaluation of hepatic glucose dynamics by deuterium metabolic imaging (DMI) and investigate whether a subsequent change in hepatic glycogen content is observable by ¹³C-MRS at 7 T. Six healthy subjects received an oral glucose load of 60g, 20g or 10g and were examined with interleaved ²H-MRSI/¹³C-MRS scans for 150 min. Results suggest the feasibility of reducing D-glucose loads to as low as 10g per subject for hepatic DMI with several benefits ‑ including cost, but precision of determination of changes in hepatic glycogen content is substantially reduced.

Introduction

Quantitative non-invasive exploration of hepatic glucose metabolism is of high importance due to its central role in glucose homeostasis and relevance for various metabolic diseases¹. Deuterium metabolic imaging (DMI) has emerged as novel tool to dynamically assess the metabolic fate of administered deuterated glucose (and other substrates) in animals² and humans³. To obtain a comprehensive picture of hepatic carbohydrate metabolism, ¹³C-MRS has been used to follow glycogen kinetics⁴. For mapping of orally administered [6,6′-²H₂]-glucose (D-Glc) in brain and liver, initial work suggested doses of 60g or 0.75g/kg body weight³. High doses of D-Glc are expensive, potentially challenging to digest (in particular for post-bariatric-surgery patients) and extend the time for gastric emptying (and hence the scan time to cover the entire metabolic processing). Thus, the feasibility of using lower doses of D-Glc for dynamic hepatic glucose mapping is evaluated. Here, we estimate the minimal dose of D-Glc needed for evaluation by hepatic DMI while still allowing determination of ensuing variations of glycogen levels by ¹³C-MRS. Our project builds on previous work⁵ focusing on glucose and glycogen turnover using DMI and ¹³C-MRS with standard doses of D-Glc.

Methods

Exams are performed at 7T (Terra, Siemens) with a triple-tuned surface coil (¹H: quadrature-driven dual loop, ²H and ¹³C: linearly driven single loops, outer dimensions: 30x10x30 cm) from Rapid Biomedical. Chromium-doped acetone (8 mL, 1% deuterated) inserted in a vial near the coil center served as external reference.
- DMI: conventional 3D-MRSI (0.50 ms rectangular excitation pulse, 0.35 ms phase-encoding gradient, TR 500 ms, 4 acquisitions with acquisition-weighting, 12x12x8 phase encodings, nominal resolution of 18.3x18.3x27.5 mm³, 1000 Hz spectral width, 4:08 min acquisition time). MRI-visible fiducial markers on the coil.
- ¹³C-MRS: pulse-and-acquire sequence (2 ms hyperbolic-secant-pulse excitation, TR 600 ms, 512 acquisitions, Nuclear-Overhauser-irradiation (fixed voltage, 510 ms), acquisition time 5:08 min).
Data processing and fitting was performed in jMRUI using AMARES⁶.
Study population according to D-Glc intake: six healthy subjects each received an oral glucose load of either 60 g (0.74 g/kg, 0.73 g/kg), 20 g (0.29 g/kg, 0.21 g/kg) or 10 g (0.16 g/kg, 0.15 g/kg) of D-Glc (in 200 ml of water) to ingest in supine position (i.e. two subjects per dose). Scanning was performed until 150 min after D-Glc intake. Blood sampling was performed for plasma concentration of (enriched) glucose, insulin, glucagon and C-peptide. Subjects underwent 48h of standardized diet and withdrawal from strenuous exercise prior to scanning.
The time courses of D-water and glycogen signals were represented by a linear model, where the starting point was set when the related D-Glc signal reached 25% of the baseline natural abundance water signal (P-values for non-zero slope).

Results and Discussion

Fig-1 shows the spectral quality of DMI at different time points (T0, T30, T60, T150) for three subjects receiving different D-Glc doses. Each spectrum represents the average of 6 voxels in the liver, manually selected close to the coil for optimal signal. The 60 g dose resulted in a hepatic D-Glc signal (3.7 ppm) comparable to the amplitude of water (4.7 ppm) around T60 and remaining visible until the end of the session. For 20 g and 10 g, the maximum D-Glc signal was lower and strongly decreased towards the end of the exam, suggesting a more substantial coverage of Glc digestion. SNR of water of a single voxel spectrum near the center of the coil was ~34 and the linewidths ~21 Hz.
Fig-2 shows the time course of the D-Glc signal with a time resolution of 5-10 minutes for D-Glc and D-water for all subjects and doses. Even for 10 g, the signal was well defined and adequate for fitting. For 60 g, a strong increase in the water signal was observed in the second half of the scan time, indicating metabolization into end products of the TCA cycle. For 20 g and 10 g the increase is progressively lower, but still significant even for single subjects.
Fig-3 illustrates the glycogen signal kinetics from ¹³C-MRS for different doses. While for the 60 g cases we can detect an increase of ~10-15% at the end of the scan⁵, no significant signal increase is found for individual subjects for lower doses. Quantitative evaluations of the time courses for D-water and glycogen signals are reported in Table-1.

Conclusions

These initial results suggest the feasibility of lowering D-Glc loads to as low as 10 g for kinetic evaluation of hepatic D-Glc by DMI. The benefits are manifold: 1) more complete observation of the whole dynamic of increase and decay to baseline of D-Glc (beneficial for metabolic modeling); 2) reduced cost for agents (from ~1000$ to ~200$); 3) ease of intake (most relevant for post-bariatric surgery patients). However, the observation of the concomitant conversion of glucose to glycogen by natural abundance ¹³C MRS, necessary to fully understand the metabolic fate of D-Glc, is limited at such low doses to evaluations of larger cohorts. Amelioration of this problem might be possible with the use of denoising for ¹³C-MRS (potentially also beneficial for DMI) or with substituting some of the deuterated by unlabeled glucose.

Acknowledgements

This is project is supported by the Swiss National Science Foundation (PCEGP3_186978) and Diabetes Center Bern. We further acknowledge the support received from the study nurse team (Valérie Brägger, Joana Filipa Rodrigues Cunha Freitas, Sandra Tenisch), Andreas Melmer and Laura Goetschi from the Department of Diabetes, Endocrinology, Nutritional Medicine and Metabolism, Inselspital, Bern University Hospital.