Simultaneous resting-state FDG-PET/fMRI in Alzheimer Disease: Relationship between glucose metabolism and intrinsic activity

Highlights

• Simultaneous estimates of glucose metabolism and intrinsic connectivity are both sensitive to aMCI/AD in impacted brain regions.

• The relationship between FDG-PET and fMRI metrics is reduced in aMCI/AD patients.

• Resting-state fMRI metrics are related to fluorodeoxyglucose metabolism in different regions of the brain.

Abstract

Simultaneously evaluating resting-state brain glucose metabolism and intrinsic functional activity has potential to impact the clinical neurosciences of Alzheimer Disease (AD). Indeed, integrating such combined information obtained in the same physiological setting may clarify how impairments in neuroenergetic and neuronal function interact and contribute to the mechanisms underlying AD. The present study used this multimodality approach to investigate, by means of a hybrid PET/MR scanner, the coupling between glucose consumption and intrinsic functional activity in 23 patients with AD-related cognitive impairment ranging from amnestic mild cognitive impairment (MCI) to mild-moderate AD (aMCI/AD), in comparison with a group of 23 healthy elderly controls. Between-group (Controls > Patients) comparisons were conducted on data from both imaging modalities using voxelwise 2-sample t-tests, corrected for partial-volume effects, head motion, age, gender and multiple tests. FDG-PET/fMRI relationships were assessed within and across subjects using Spearman partial correlations for three different resting-state fMRI (rs-fMRI) metrics sensitive to AD: fractional amplitude of low frequency fluctuations (fALFF), regional homogeneity (ReHo) and group independent component analysis with dual regression (gICA-DR). FDG and rs-fMRI metrics distinguished aMCI/AD from controls according to spatial patterns analogous to those found in stand-alone studies. Within-subject correlations were comparable across the three rs-fMRI metrics. Correlations were overall high in healthy controls (ρ = 0.80 ± 0.04), but showed a significant 17% reduction (p < 0.05) in aMCI/AD patients (ρ = 0.67 ± 0.05). Positive across-subject correlations were overall moderate (ρ = 0.33 ± 0.07) and consistent across rs-fMRI metrics. These were confined around AD-target posterior regions for metrics of functional connectivity (ReHo and gICA-DR). In contrast, FDG/fALFF correlations were distributed in the frontal gyrus, thalami and caudate nuclei. Taken together, these results support the presence of bioenergetic coupling between glucose utilization and rapid transmission of neural information in healthy ageing, which is substantially reduced in aMCI/AD, suggesting that abnormal glucose utilization is in some way linked to communication breakdown among brain regions impacted by the underlying pathological process.

Introduction

Alzheimer Disease (AD) is a chronic neurodegenerative disorder that affects approximately 9 million people in Europe (Diaz-Ponce et al., 2016). This disease is characterized by asymptomatic onset (Dubois et al., 2014) followed by cognitive decline that worsens with disease progression (Bouwman et al., 2010; McKhann et al., 2011). AD affects memory and the ability to carry out voluntary and purposeful actions, inducing difficulties in language production and comprehension as well as disorientation in time and space.

Novel biomarkers have been developed to support the diagnosis of AD (Dubois et al., 2014). These are primarily derived from structural magnetic resonance imaging (MRI) of the hippocampus (Sarazin et al., 2010) and from analysis of cerebral spinal fluid (CSF) amyloid β (Aβ) or tau protein concentrations (Nisbet et al., 2015; Nordberg, 2010). In addition, molecular neuroimaging biomarkers by means of positron emission tomography (PET) using 18F-2-fluoro-2-deoxy-d-glucose (FDG-PET), which reflects glucose metabolism mainly from neurons (Dennis and Thompson , 2014), provides an accurate predictor of AD progression in the form of temporo-parietal, posterior Cingulate and Precuneus hypometabolism (Chételat et al., 2003; Ito et al., 2015; Minoshima et al., 1997; Mosconi, 2005; Scheltens et al., 2016).

In these regions, resting-state functional MRI (rs-fMRI), which exploits the blood-oxygen level-dependent (BOLD) endogenous contrast (Buckner et al., 2008; Dennis and Thompson , 2014; Greicius et al., 2004; Supekar et al., 2008), has disclosed the physiological presence of intrinsic functional architectural patterns in the low-frequency (0.01–0.1 Hz) oscillations, including the default mode network (DMN) (Gusnard and Raichle, 2001), which are sensitive to the AD process (Greicius et al., 2004; Hafkemeijer et al., 2012; Klaassens et al., 2017; Zhang et al., 2010). rs-fMRI therefore has potential clinical relevance, especially as it is non-invasive, relatively inexpensive and easy to acquire in comparison to PET-FDG.

The recently developed hybrid PET/MRI scanners now allow one to simultaneously acquire glucose metabolism and rs-fMRI under the same physiological condition (Wehrl et al., 2015). This constitutes an exciting opportunity to investigate the relationships between intrinsic metabolic and functional brain changes (Aiello et al., 2016; Cecchin et al., 2017; Tahmasian et al., 2015). To date, published studies that have integrated these two imaging modalities in a simultaneous acquisition setup have involved healthy adults using seed-based methods for rs-fMRI network analysis (Riedl et al., 2014), or healthy elderlies using local intrinsic functional activity metrics (Aiello et al., 2015). However, to date the intimate coupling between FDG-PET hypometabolic pattern and rs-fMRI network disruption in AD-related brain damage has been little addressed.

To the best of our knowledge, no simultaneous PET/fMRI study published to date has investigated how early AD-related local changes in resting-state glucose consumption relate to the changes in functional activity observed in stand-alone rs-fMRI studies. Such knowledge is likely to provide novel insights towards the development of clinically relevant tools (Drzezga et al., 2014; Tahmasian et al., 2015). In the present simultaneous resting-state FDG-PET/fMRI study we investigated the relationships between glucose consumption and intrinsic functional activity in healthy aged individuals and patients with mild AD on a voxel-by-voxel basis. Intrinsic functional activity was assessed by means of three metrics that differ in spatial-extent definition and relevance to AD process (Aiello et al., 2016; Cha et al., 2015; Damoiseaux et al., 2012), namely i) fractional amplitude of low frequency fluctuations (fALFF), a single-voxel indicator of BOLD signal frequency power independent from brain connectivity thereof (Zou et al., 2008; Zuo et al., 2010); ii) regional homogeneity (ReHo), a measure of local functional connectivity (Zang et al., 2004); and iii) group-independent component analysis with dual regression (gICA-DR), which identifies common spatiotemporal patterns and project them back to each individual (Beckmann et al., 2009; Beckmann and Smith, 2004).