Integrity of Neurocognitive Networks in Dementing Disorders as Measured with Simultaneous PET/Functional MRI

Isabelle Ripp, Thomas Stadhouders, Alexandre Savio, Oliver Goldhardt, Jorge Cabello, Vince D. Calhoun, Valentin Riedl, Dennis M. Hedderich, Janine Diehl‐Schmid, Timo Grimmer, Igor YakushevJournal of Nuclear Medicine 2020 · DOI

Abstract

Functional MRI (fMRI) studies have reported altered integrity of large-scale neurocognitive networks (NCNs) in dementing disorders. However, findings on the specificity of these alterations in patients with Alzheimer disease (AD) and behavioral-variant frontotemporal dementia (bvFTD) are still limited. Recently, NCNs have been successfully captured using PET with $^{18}$F-FDG. This study measured network integrity in 72 individuals using simultaneous resting-state fMRI and $^{18}$F-FDG PET. Results showed that while fMRI integrity was reduced across multiple networks in both patient groups, PET data revealed highly disease-specific patterns. Specifically, PET-derived integrity of the anterior default-mode network (DMN) accurately differentiated AD from bvFTD with 94% accuracy, suggesting metabolic connectivity as a superior diagnostic tool.

Introduction

Distinguishing between Alzheimer’s Disease (AD) and behavioral-variant frontotemporal dementia (bvFTD) is a significant clinical challenge, particularly in the early stages where symptoms overlap. Functional MRI (fMRI) has been the primary tool for studying neurocognitive networks (NCNs), but its results in dementia are often inconsistent due to low signal-to-noise ratios and the indirect nature of the BOLD signal, which measures hemodynamic changes rather than direct neural activity.

This study investigates whether "metabolic connectivity"—measured via $^{18}$F-FDG PET—provides a more specific and reliable diagnostic fingerprint for these disorders than traditional fMRI-based functional connectivity.

Methods

The researchers utilized a simultaneous PET/MRI scanner to capture metabolic and functional data under identical physiological conditions in 72 subjects (29 AD, 21 bvFTD, and 22 healthy controls).

Independent Component Analysis (ICA) was used to decompose the complex brain signals into distinct spatial networks. The analysis focused on four major hubs: the anterior Default Mode Network (DMN), posterior DMN, Salience Network (SN), and the right Central Executive Network (CEN). Integrity metrics were calculated for each subject to quantify how well their brain activity matched healthy network templates.

Spatial maps of neurocognitive networks identified via simultaneous fMRI and PET.

Key Results and Insights

The Modality Gap

A striking finding was the low correlation between fMRI and PET network indices, suggesting that functional and metabolic connectivity capture distinct biological processes. While fMRI showed reduced integrity across almost all networks in AD patients, PET data revealed much cleaner, disease-specific patterns.

Disease-Specific Metabolic Signatures

PET imaging demonstrated that AD primarily affects the posterior DMN, whereas bvFTD is characterized by significant hits to the anterior DMN, Salience Network, and Central Executive Network. This specificity allowed for high diagnostic accuracy:

  • Classification Accuracy: PET-derived integrity of the anterior DMN alone distinguished AD from bvFTD with 94% accuracy (97% sensitivity).
  • Combined Models: Adding the Salience Network to the PET model increased diagnostic accuracy to 96%.
  • fMRI Performance: In contrast, the best fMRI-based predictor (posterior DMN) achieved only 64% accuracy.
Comparison of network integrity across AD, bvFTD, and healthy controls for PET and fMRI modalities.

Vascular Interference in fMRI

Post-hoc analysis suggested that the global impairment seen in fMRI for AD patients might be due to vascular issues, such as white matter hyperintensities, which "muddy" the BOLD signal. Metabolism, as measured by PET, appears to be a more stable and specific marker of network health in neurodegenerative contexts.

Conclusion

Simultaneous PET/fMRI reveals that metabolic connectivity is a more effective biomarker for differentiating AD and bvFTD than functional connectivity. The high disease specificity of PET-derived network integrity supports its potential as a powerful diagnostic tool. Future work should focus on validating these findings in larger cohorts and developing a standardized atlas for PET-based metabolic networks to facilitate clinical adoption.

Summary updated

Abstract

Functional MRI (fMRI) studies have reported altered integrity of large-scale neurocognitive networks (NCNs) in dementing disorders. However, findings on the specificity of these alterations in patients with Alzheimer disease (AD) and behavioral-variant frontotemporal dementia (bvFTD) are still limited. Recently, NCNs have been successfully captured using PET with 18 F-FDG. Methods: Network integrity was measured in 72 individuals (38 male) with mild AD or bvFTD, and in healthy controls, using a simultaneous restingstate fMRI and 18 F-FDG PET. Indices of network integrity were calculated for each subject, network, and imaging modality. Results: In either modality, independent-component analysis revealed 4 major NCNs: anterior default-mode network (DMN), posterior DMN, salience network, and right central executive network (CEN). In fMRI data, the integrity of the posterior DMN was found to be significantly reduced in both patient groups relative to controls. In the AD group the anterior DMN and CEN appeared to be additionally affected. In PET data, only the integrity of the posterior DMN in patients with AD was reduced, whereas 3 remaining networks appeared to be affected only in patients with bvFTD. In a logistic regression analysis, the integrity of the anterior DMN as measured with PET alone accurately differentiated between the patient groups. A correlation between indices of 2 imaging modalities was low overall. Conclusion: FMRI and 18 F-FDG PET capture partly different aspects of network integrity. A higher disease specificity for NCNs as derived from PET data supports metabolic connectivity imaging as a promising diagnostic tool.
BibTeX
@article{ripp2020integrity,
  author = {Isabelle Ripp and Thomas Stadhouders and Alexandre Savio and Oliver Goldhardt and Jorge Cabello and Vince D. Calhoun and Valentin Riedl and Dennis M. Hedderich and Janine Diehl‐Schmid and Timo Grimmer and Igor Yakushev},
  title = {{Integrity of Neurocognitive Networks in Dementing Disorders as Measured with Simultaneous PET/Functional MRI}},
  journal = {Journal of Nuclear Medicine},
  year = {2020},
  doi = {10.2967/jnumed.119.234930},
}
Figure