A conserved neurometabolic ratio architecture across regions, field-strength and species

Proton magnetic resonance spectroscopy (¹H‑MRS) increasingly allows simultaneous quantification of multiple neurometabolites in vivo, yet most work still relies on isolated metabolite levels or a small set of pre‑selected ratios. Treating the full profile of inter‑metabolite ratios as the unit of analysis may reveal a neurometabolic backbone that can serve as a systems‑level reference for neuropsychopharmacology.

We aimed to (i) establish a normative pairwise metabolite ratio architecture in the human brain, (ii) determine its stability across scanners, field strengths, regions and species, and (iii) characterise its internal organisation in terms of concentration tiers that may constrain how neuromodulatory or metabolic challenges can reconfigure neurometabolic state.

We analysed ¹H‑MRS data from multiple datasets comprising healthy humans scanned at 7T and 3T in posterior and mid‑cingulate cortices, insula and hippocampus, as well as rat hippocampus at 9.4T. Across datasets, up to 12 metabolites were available, including glutamate (Glu), glutamine (Gln), gamma‑aminobutyric acid (GABA), aspartate (Asp), N‑acetylaspartate (NAA), N‑acetylaspartylglutamate (NAAG), total creatine (tCr), myo‑inositol (Ins), choline‑containing compounds (tCho), glutathione (GSH), lactate (Lac) and taurine (Tau). From water‑referenced concentrations in each dataset, we computed all pairwise ratios and assembled metabolite‑by‑metabolite ratio matrices, where cell values quantify how many times more (or less) concentrated one metabolite is relative to another. We visualised these matrices as heatmaps and quantified pattern similarity across datasets using Pearson correlation, cosine similarity and structural similarity index, and by normalising absolute ratio differences to the median absolute deviation and standard deviation of inter‑individual variation.

The normative architecture revealed a pronounced tiered hierarchy of metabolite pools rather than a continuous spectrum of concentrations. Glu, NAA and tCr formed a high‑abundance plateau, with near‑equal concentrations across individuals. Ins occupied an intermediate tier at roughly half the level of the top plateau, while Gln and Asp formed a lower tier at approximately one‑third of Glu. GABA, NAAG, Tau, GSH and tCho consistently occupied the lowest tier, with ratios below 0.15 relative to Glu, and narrow interquartile ranges indicated that tier boundaries were stable across participants. This tiered pattern was conserved across independent 7T scanners and across cingulate and insular cortex, with near‑perfect preservation of ratio ordering and proportional structure. Despite expected scaling differences, the same ordering was observed when comparing 7T cortical data to 3T hippocampal data and to 9.4T rat hippocampus, where cross‑species pattern correlations remained high, even when absolute deviations exceeded several units of typical inter‑individual variability.

These findings indicate that the brain maintains neurometabolite pools at discrete concentration set points that are conserved across regions, acquisition settings and species. Considering the entire metabolite ratio profile respects the complex metabolic interconnections and thus reduces the risk that subtle shifts remain invisible when only a few hand‑picked metabolites are examined. A ratio‑based profile may therefore serve as a robust, translationally useful scaffold for detecting subtle, condition‑ or drug‑induced deviations in neurometabolic equilibrium in neuropsychopharmacological research.

References

No references