MRSI and the Human Metabolic Brain Connectome: Construction, Topology, and Biological Significance
Poster No:
1953
Submission Type:
Abstract Submission
Authors:
Federico Lucchetti1, Edgar Celereau2, Antoine Klauser3, Yasser Alemán-Gómez4, Patric Hagmann5, Camille Piguet6, Arnaud Merglen7, Paul Klauser8
Institutions:
1Lausanne University Hospital and University of Lausanne, Lausanne, Vaud, 2Center for Psychiatric Neuroscience, CHUV / UNIL, Lausanne, VD, 3Advanced Clinical Imaging Technology, Siemens Healthcare AG, Lausanne, VD, 4Centre hospitalier universitaire vaudois (CHUV), Lausanne, VT, 5Lausanne University Hospital and University of Lausanne (CHUV-UNIL), Lausanne, Vaud, 6University of Geneva, Geneva, Geneva, 7Departement de pédiatrie, Hopitaux Universitaire de Genève, Genève, GE, 8Center for Psychiatric Neuroscience, Department of Psychiatry, Lausanne University Hospital and UNIL, Lausanne, Vaud
First Author:
Co-Author(s):
Paul Klauser
Center for Psychiatric Neuroscience, Department of Psychiatry, Lausanne University Hospital and UNIL
Lausanne, Vaud
Center for Psychiatric Neuroscience, Department of Psychiatry, Lausanne University Hospital and UNIL
Lausanne, Vaud
Introduction:
The advent of network science in neurobiology has revealed fundamental principles of brain organization, yet current neuroimaging techniques (e.g., diffusion MRI, fMRI) do not capture the biochemical underpinnings of these networks. Since oxidative glucose metabolism sustains brain function, metabolic imaging can offer deeper insights into the biochemical foundations of network structure. By employing advanced 3D proton magnetic resonance spectroscopic imaging (MRSI), we enable rapid, whole-brain metabolic assessment suitable for large-scale and clinical applications. Variations in brain network organization are linked to key behavioral and cognitive traits and can illuminate the origins of neurological and psychiatric disorders. Many such conditions, or "connectopathies," stem from disruptions in distributed networks rather than isolated regions. Hub nodes, with their high metabolic demands, are particularly susceptible to oxidative stress, and their failure can compromise global network integrity. In this context, metabolic imaging becomes an essential tool for understanding both normal and pathological brain organization.
Methods:
We constructed the brain metabolic connectome using 3D proton MRSI data acquired at 5 mm isotropic resolution in under 12 minutes (Klauser et al., 2022) from 68 healthy adolescents, and validated these findings on an independent cohort (N=13). We resolved the spectral peaks of five standard MRSI metabolites (NAA/NAAG, Cr/PCr, GPC/PCh, mI, and Glu/Gln). The brain was parcellated into 277 nodes, each represented by a z-score normalized metabolite concentration profile. Edges were defined as pairwise Spearman correlations between these profiles, resulting in a 277×277 metabolic similarity matrix (MeSiM) for each individual. This pipeline was tested for reproducibility and robustness across individuals, datasets, MRI scanners, and anatomical parcellations. We then subjected the MeSiMs to network analyses to characterize the global organization of the metabolic connectome. Additional statistical tests were used to assess the influence of spatial factors on the observed topology.
Results:
Our analyses revealed that metabolic networks exhibit natural network properties, including modular organization and robust homotopic patterns (Figure 1A), reflecting functionally integrated yet spatially distinct systems. We found that metabolic correlations are spatially distributed; closer brain regions tend to have more similar metabolic profiles (positive metabolic correlations), distant regions tend to have more dissimilar metabolic profiles (negative metabolic correlations) (r=0.96, p<0.001, N=1000 spatial permutations for positive; r=0.97, p<0.001, N=1000 spatial permutations for negative). Central to these findings is the identification of the "metabolic fiber", (see Figure 1B) a smooth homotopic gradient of metabolic similarity extending from the occipital lobe through the parietal lobe, prefrontal cortex, cingulate cortex, and into subcortical areas (p<0.05 volumetric spin test). Additionally, structural hubs aligned with metabolically active nodes (0.19 < r < 0.25, p<0.001, N=1000 spatial permutations for the five metabolites). Importantly, we showed that metabolic network organization is not rooted in structural connectivity (r<0.05, p<0.001, N=1000 spatial permutations), but aligns closely with genetic coexpression and cytoarchtiecture, including strong correlation with genetic co-expression data from the Allen Brain Institute (r=0.49, p<0.001, N=1000 spatial permutations) and cytoarchitectonic similarity based on the Big Brain similarity matrix (r=0.29, p<0.001, N=1000 spatial permutations).
·Figure 1: Metabolic connectome network topology
·Figure 2: Metabolic fibre
Conclusions:
These associations with genetic co-expression and cytoarchitecture suggest a neurodevelopmental origin, influencing cognitive function and vulnerability to disorders. By integrating metabolic imaging with connectomics, we establish a foundation for exploring the biochemical basis of brain networks in health and disease.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia)
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural) 2
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Anatomy and Functional Systems
Novel Imaging Acquisition Methods:
MR Spectroscopy 1
Physiology, Metabolism and Neurotransmission:
Physiology, Metabolism and Neurotransmission Other
Keywords:
Cortex
Data analysis
Glutamate
Magnetic Resonance Spectroscopy (MRS)
MR SPECTROSCOPY
MRI
Psychiatric Disorders
Schizophrenia
Sub-Cortical
Other - Connectomics
1|2Indicates the priority used for review