Non-invasive measurement of neurotransmitter-specific glucose metabolism in the human brain using proton-observed proton-edited ¹³C-MRS (POPE¹³C-MRS)

Reviewer #1 (Public review):

Excitation/inhibition (E/I) balance between excitatory (glutamate) and inhibitory (GABA) neurotransmission is being increasingly studied using magnetic resonance spectroscopy (MRS), for example in autism spectrum disorder, schizophrenia and attention deficit/hyperactivity disorder. These are typically measured using standard single-voxel MRS methods (eg PRESS, sLASER) to measure glutamate/glutamine or "Glx" and spectral editing methods (eg MEGAPRESS, MEGA-sLASER) techniques to measure GABA. Such methods only give a measure of the total MR-visible metabolite concentration in the voxel. That is, they don't distinguish between glutamate/GABA involved in neurotransmission or in other metabolic processes.

Cherix et al present a method for measuring glucose metabolism, with the potential to be used on a standard clinical MRI scanner. This proof-of-concept study focused on measuring proton signals from glucose metabolites, including lactate, glutamate, GABA & Glx. The method works by administering 13C universally labelled glucose (where all six carbon atoms are substituted with 13C). When the glucose is metabolised, 13C label is incorporated into specific positions within its metabolites. Protons attached to 13C don't produce a signal in 1H-MRS in a subsequent MEGA-sLASER scan, leading to a drop in the signal as the labelled metabolite concentration builds up. At the same time, "satellite resonances" appear for protons coupled to 13C, which increase as the labelled metabolite concentration builds up. Metabolite concentrations were inferred using a simple dynamic model.

The main strength of the method is that it enables dynamic metabolic information that would typically only be available with multi-nuclear MRS capability (13C or 2H) to be achievable using standard preclinical or high field (>= 7T) human MRI systems, with widely available spectral-editing acquisitions.

The results in the mouse spectra seem very convincing for lactate and GABA/Glx. For the human scans, changes in lactate weren't detectable, which is not surprising given how little lactate appears in the normal brain. In the discussion, the authors argue the method can potentially be used in a standard, 3T clinical scanner. It may be too soon to conclude that, as it's not yet clear there would be sufficient SNR in spectra at that field strength. Additionally, the heteronuclear coupling constants are quite high. The authors recognise that this may complicate detection of satellite resonances due to signal dephasing. Another potential complication at 3T (or 2.9T) is the potential for the satellite resonances to come close to the GABA peak at 3 ppm. More accurate measurements of coupling constants will allow that to be determined. Another potential limitation of the method is macromolecule contamination of the 3 ppm GABA peak. That may be overcome by using macromolecule-nulled MEGA-editing, though frequency navigators may be necessary to overcome the increased sensitivity to frequency drift.

The authors achieved their aims of showing that imaging glucose metabolites was possible using standard proton-only MRI systems, without the need for additional multinuclear coils, transmitters and receivers. The evidence if very compelling for the mouse scans but only incomplete for the human scans.

The ability to quantify metabolites involved in E/I balance has the potential to revolutionise studies into disorders where changes in E/I balance are implicated. This is especially the case for preclinical models. Such studies may be less feasible in clinical studies due to the high cost of universally 13C-labelled glucose, but this proof-of-concept is a promising start.

Reviewer #2 (Public review):

Summary:

The main aim of the presented manuscript was to test and validate proton observed proton edited 13C MRS and track the 13C label from uniformly labeled U-13C-glucose into glutamine, glutamate, GABA and lactate in rodent and human brain in vivo.

Strengths:

In contrast to already established methods of 13C and 2H MRSI this method applies only proton RF and thus can potentially be implemented on a standard clinical scanner.

Weaknesses:

The validation of the method in a human setting is rudimentary. First, even at ultra-high field strength of 7T, the authors did not reach sufficient SNR to detect and quantify GABA with good CV, and further, the experiment needs optimisation to reach metabolic and fractional enrichment steady state and/or for additional conditions to show the possibility of lactate detection.

Reviewer #3 (Public review):

Summary:

In their paper, the authors propose using dynamic MEGA-sLASER acquisitions to track the incorporation of ¹³C from uniformly ¹³C-labeled glucose into glutamate(+glutamine) and GABA pools. This approach enables the direct and simultaneous investigation of excitatory and inhibitory neurotransmission and metabolism without requiring a ¹³C radiofrequency coil. While the authors' goals and efforts are commendable, the work has several significant limitations.

Strengths:

Use of excellent hardware (¹H cryoprobe, ultra-high-field MRI scanners); Ambitious objectives.

Weaknesses:

MRS data analysis requires improvement; Small cohort sizes; No demonstration that the proposed acquisition scheme is superior in terms of robustness, accuracy, or performance.