Synopsis

Keywords: Non-Proton, Metabolism, 13C

The aim of this study is to assess the reproducibility of a selective proton-observed, carbon-edited (selPOCE) MRS sequences for the detection of glutamate/glutamine cycling at 7T in healthy participants. Participants were scanned twice while undergoing [U-¹³C] glucose infusion. The time course of Glu C45 and Gln C45 labeling were similar for test-retest measurements. This supports the application in future studies on measuring neuroenergetics and neurotransmitter cycling.

Introduction

Glutamate is the major excitatory neurotransmitter in the central nervous system¹. During the glutamate/glutamine (Glu/Gln) neurotransmitter cycle, excitatory glutamatergic neurons can take up glucose and oxidize it via the tricarboxylic acid (TCA) cycle^2-4.
Glu/Gln neurotransmitter cycling in the brain can be measured by the detection of ¹³C labeling of Glu and Gln in the brain during the infusion of ¹³C-labeled glucose using ¹³C magnetic resonance spectroscopy (MRS)^2-4. While the sensitivity of ¹³C MRS is low, it can be enhanced using indirect detection using spectral editing. Compared to direct ¹³C MRS, indirect ¹³C MRS has a lower spectral resolution and Glu and Gln are less well resolved. The latter can be overcome by using selective proton-observed, carbon-edited (selPOCE) MRS⁵.
Monitoring Glu/Gln metabolism may provide important biomarkers for studies on various brain diseases. The reproducibility of determining this metric using selPOCE MRS in humans is not yet known. Here we assess the reproducibility of selPOCE MRS for detecting of ¹³C-labeled Glu and Gln at 7T in the brain.

Methods

Experimental protocol
Four healthy male adults (age:23 ± 2.2 years) underwent ¹H-[¹³C]) MRS in a 7T MR-system (Philips, Best, NL) on two occasions, at least 3 weeks apart. A ¹³C transmit birdcage head coil, combined with 8 transmit-receive ¹H dipole antennas and a 32-channel ¹H receive array (Nova Medical, Wilmington, USA) was used⁶. One participant was excluded due to significant head movements. Participants were fasted overnight. After IV-catheter placement in both arms, participants were positioned in the MR system. T₁-weighted images were acquired for voxel positioning and B₀ and B₁ shimming in a manually drawn ROI in the frontal lobe was performed. The infusion protocol started with an initial, exponential bolus of 99% enriched [U-¹³C] glucose lasting 20 min, followed by 100 min of continuous infusion of 66% enriched [U-¹³C] glucose and 33% unlabeled glucose, allowing the concentration of plasma glucose to fall slowly while maintaining the fractional enrichment (FE) constant.

A selPOCE STEAM MRS sequence was used for dynamic (6:40 min scan interval) detection of the H4 of Glu and Gln signals, labeled at carbons 4 and 5 (referred to as Glu C45 and Gln C45) in a voxel in the frontal lobe (minimal voxel size 3x5x3 cm; TE 7.8ms; TM 30ms; TR 3000ms; NSA 4x32; VAPOR water suppression). The selPOCE STEAM sequence consisted of four acquisitions: (1) conventional STEAM; (2) ¹³C inversion, using a composite ¹³C 90°-180°-90° block pulse set to 32.95 ppm, i.e. in between the C4 resonances of Glu and Gln; (3 and 4) ¹³C Glu and Gln phase discrimination achieved with ¹³C 90°(+x)-^1­H 180°-¹³C 90°(+/-x) block pulses, also set to 32.95 ppm (figure 1). A crusher coil was used to suppress residual signals from extracranial lipids.
Blood samples were collected every 5 minutes to determine plasma glucose levels using a benchtop glucose analyzer (YSI® 2500 series) and for later determination of ¹³C FE of plasma glucose by ¹H MRS at 500 MHz.

Post-Processing
Coil combination was performed using a generalized-least-squares algorithm, using the unsuppressed-water signal to compute the coil sensitivity of each channel. Spectra were frequency and phase aligned, and subtracted per fourfold for selPOCE⁷. The Glu C45, Gln C45 and Glx C3 signals in the difference spectra were fitted with a home-written LCModel approach in MATLAB (MathWorks, USA).
The time course of Glu C45 and Gln C45 signals were normalized to the average amplitude of Glu C45 of the final ~30 minutes of each experiment (steady state).
The test/retest variability of Glu C45 and Gln C45 was defined as: $$ Variability(T/RT)= \frac{R_{\frac{Early}{SS}_{Retest}} - R_{\frac{Early}{SS}_{Test}}}{(R_{\frac{Early}{SS}_{Retest}}+R_{\frac{Early}{SS}_{Test}})/2} $$
Where R_Early/SS is the Glu C45 (or Gln C45) signal over early time points (i.e. ~20-45 min after start infusion) normalized to the Glu C45 signal during steady state.

Results

There was a rapid increase of plasma glucose ¹³C FE and plasma glucose concentrations during the first 20 min of infusion, followed by a steady state of ¹³C FE until the end of the infusion. The time courses were similar for T/RT (Figure 2).  FE of plasma glucose in participant 1-Test and the first datapoints of participant 3-Retest could not be determined, due to insufficient water suppression.
Figure 3 shows the selPOCE difference spectra from one volunteer. The Glu C45 and Gln C45 signals can be observed, as well as the overlapping resonances of Glx C3. The time-dependent increase of Glu C45 and Gln C45 signals for all participants can be seen in figure 4. As expected, labeling of Glu C45 was faster compared to Gln C45 labeling and the steady-state ratio Glu C45 / Gln C45 was on average 0.31±0.06.
R_Early/SS for Glu test and retest was similar (mean 0.51±005 vs. 0.52±0.01) with a test/retest variability of 9.7%. For Gln R_Early/SS test and retest was similar as well (0.16±0.02 vs. 0.13±0.04); however, test/retest variability was higher compared to Glu (32.4%).

Discussion and conclusion

We have investigated the reproducibility of selPOCE MRS at 7T for detection of labeled Glu and Gln. Time courses of ¹³C labeling of Glu and Gln labeling were similar for test retest. Knowing the variability of these readings supports the application to future studies on for example disease specific alterations in Glu/Gln cycling.

Acknowledgements

No acknowledgement found.