The use of positron emission tomography/magnetic resonance imaging in dementia: A literature review

Positron emission tomography–magnetic resonance imaging (PET/MRI) is an emerging hybrid imaging system in clinical nuclear medicine. Research demonstrates a comparative utility to current unimodal and hybrid methods, including PET‐computed tomography (PET/CT), in several medical subspecialities such as neuroimaging. The aim of this review is to critically evaluate the literature from 2016 to 2021 using PET/MRI for the investigation of patients with mild cognitive impairment or dementia, and discuss the evidence base for widening its application into clinical practice.

recently in the literature, initially assessing feasibility. 5,6 Multiple considerations are needed when combining the two methods, such as the effect on electron flow in PET as a result of both static and gradient magnetic fields, as well as the disruption to the uniformity of the magnetic field resulting from PET components being placed in the MRI scanner bore. 7,8 Methods developed to overcome such technical issues are highlighted by the presence of three whole-body PET/MRI systems utilising different approaches created by GE Health Care, Philips Healthcare, and Siemens. 9 Despite the significant economic and technical barriers, the motivation to advance the use of PET/MRI within routine clinical practice stems from the superior structural information provided by MRI compared to CT, as well as the technical advantages of improved coregistration, and service benefits associated with reducing scanning time. Using radiotracers, PET relies on the production of gamma rays from positron decay to provide functional information based on blood flow, glucose or oxygen consumption, protein synthesis, or neuroreceptor uptake. 10 Whilst PET has high specificity, anatomical information varies according to tracer used. MRI utilises a magnetic field and radiofrequency pulse to influence the axial spin of hydrogen protons, abundant in water and fat, with the energy emitted during proton relaxation used to produce highly sensitive structural images. 11 Compared to PET/CT, the soft tissue contrast, multiplanar capability, and reduced radiation dose offered by PET/MRI makes it an increasingly attractive tool, particularly within neuroimaging.

| PET/MRI in neuroimaging
Whilst used clinically in cardiology, 12 orthopaedics 13 and oncology, [14][15][16][17] the first human research PET/MRI images were of the brain. 6 PET imaging in the examination of neurotransmitter activity, neuroinflammation and haemodynamic compromise in ischaemic stroke is well-established. [18][19][20][21][22] Conversely, MRI is the gold standard in the assessment of brain tumours, 23 as well as in the detection of white matter (WM) damage in multiple sclerosis. [24][25][26] Combined, the advantages of PET/MRI in neurological investigation are increasingly documented, including the increased sensitivity of 18 F-fluorodeoxyglucose (FDG) PET/MRI in differentiating posttherapy tumour recurrence from fibrosis, in addition to an improved accuracy in targeting radiotherapy volumes, and overall superior image quality and artefact reduction. [27][28][29] Additionally, studies investigating epileptic seizure focus have shown PET/MRI to be a valid alternative to PET/CT in paediatric populations, particularly due to the reduced radiation exposure. 30

| PET/MRI in people with dementia
Patients undergoing cognitive assessment often require routine MRI and PET imaging. MRI techniques such as diffusion tensor imaging (DTI) and functional MRI (fMRI) assist in examining functional connectivity networks implicated in dementia, such as the default mode network (DMN) composed of regions including the medial prefrontal cortex, angular gyrus and posterior cingulate cortex. 31,32 Conversely, PET has been used in differentiating clinical subtypes of dementia based on patterns of brain metabolism. 33 Recent amyloid PET radiotracers such as [ 11 C]-Pittsburgh compound-B (PIB) and fluorinated tracers which have now become licensed for clinical use provide additional information on β-amyloid plaque pathology associated with dementias such as Alzheimer's disease (AD), improving understanding of disease prevention and progression. 34 Integration of these methods to assist diagnosis and treatment of dementia is subsequently of great interest to clinicians, with the added benefit of single session multimodal image acquisition being an important consideration for patients who are often elderly or frail. 35 Currently, PET/MRI use in dementia is predominantly research-based,   with the Dementias Platform UK establishing an innovative dementia   imaging network linking PET/MRI scanners at seven UK universities, with an aim to support large scale research and clinical trials. 36 The purpose of this review is to critically examine the most recent literature published between 2016 and 2021 on the use of PET/MRI in dementia, evaluating how it compares to unimodal approaches, and discuss the evidence base for widening its application into clinical practice.

| Literature search
A literature search on the use of PET-MRI in dementia was performed in the PubMed database. The MeSH terms used to search the database were ("PET-MR" OR "PET-MRI") AND ("Dementia" OR "Mild cognitive impairment" OR "Alzheimer"). The search was undertaken on 15th January 2021 and included all papers published from 1st January 2016 until then. The search was designed to ensure that relevant papers on all dementias would be captured, as well as papers which used the term AD (rather than Alzheimer's dementia) and studies of those with mild cognitive impairment.

| Study selection
The inclusion criteria used to screen for eligibility required papers to have been published in the last 5 years, in English, with a minimum of 10 human subjects. Additionally, only papers employing simultaneous PET/MRI imaging relating to the topic of dementia or mild cognitive impairment were selected. Papers comparing the utility of different PET radiotracers, using PET or MRI imaging separately, reports of animal studies, or papers relating to other neurological or neurodegenerative diseases were excluded from selection. Using these criteria, two researchers separately reviewed the titles and abstracts for relevance to agree a final list for inclusion.

| RESULTS
The results of the literature search are illustrated in Figure 1. Of the 116 studies identified using the search, 39 papers were included in the final selection, and they can be broadly categorised into 1) review F I G U R E 1 Flow chart demonstrating selection process of retrieved articles from literature search

| Hippocampal studies
MRI evidence of medial temporal lobe atrophy, involving structures such as the hippocampus, entorhinal, and perirhinal cortices is welldocumented in dementia. 53,54 Hippocampal regions including the cornu ammonis, and dentate gyrus are often smaller in AD/MCI patients, 55 an important biomarker for predicting AD onset. 56 were grouped together to increase statistical power, which may influence the applicability of these findings to patients at varying disease stages. Despite this, the authors noted an improved spatial alignment between PET and MRI images due to intrinsic scanner coregistration, reducing issues such as misalignment and spatial intensity distortion associated with coregistering unimodal images. 60 Kang et al. 39 similarly used automated brain segmentation software and [ 11 C]PiB PET/MRI to demonstrate the predictive ability of hippocampal regional volume on the presence of Aβ, an important predictor of dementia development. 61 These findings support the association between early memory deficits and hippocampal atrophy, 55 and further demonstrate the relationship between these structural changes and developing Aβ pathology in MCI. Although MRI is widely used to assess structural abnormalities associated with neurodegeneration, 55  Researchers also reported a significant negative correlation between 18 F-FDG hypometabolism and decreased left cornu ammonis/dentate gyrus-superior medial frontal gyrus connectivity in AD patients, a finding corresponding to previous fMRI studies showing that increased activity in cornu ammonis/dentate gyrus is associated with reduced cognitive function, 63 and can be therapeutically targeted. 64 The relationship between hippocampal functional connectivity and metabolism demonstrated by PET/fMRI in this study improves on previous findings using unimodal techniques, and allows for a more comprehensive understanding of underlying neurodegenerative processes occurring at the subregional hippocampal level.

| Functional connectivity studies
The default mode network is a tightly connected functional brain network important in memory function, studied previously using predominantly fMRI. 31 reported by both Ding et al. 50 and Scherr et al. 52 Moreover, withinsubject PET/fMRI correlations showed a 17% reduction in AD/MCI patients compared to controls. These findings support prior unimodal studies [65][66][67] and go further to provide evidence for the theory that glucose-mediated CBF regulation and glucose/oxygen coupling deteriorates with AD disease progression. 68 Limitations of this study include the finding that DMN hypometabolism in AD/MCI patients was more extensive than patterns of hypoconnectivity. 18 FDG metabolism reflects resting-state glucose mobilisation and is largely unaffected by physiological artefacts, 69 whereas the blood-oxygen level-dependent (BOLD) signal is a haemodynamic measure of oxygen consumption, and can be influenced by altered physiological conditions. 70 These differences could therefore bias the combined used of PET/fMRI metrics when interpreting underlying DMN changes in AD/MCI. However, Göttler et al. 47 found that whilst overlapping changes in BOLD signal, 18 FDG uptake, and CBF were found in the DMN of AD patients, reductions in BOLD signal were correlated with reduced CBF independently from 18 FDG-PET hypometabolism. This suggests that altered BOLD signals found in AD/MCI patients can reflect impairments in haemodynamic processes separately from changes in neuronal activity.
Regional reductions in CBF for AD patients was also found by Okazawa 41 and Riederer. 45 Altogether, the use of both fMRI BOLD signal and 18 FDG-PET data in these studies allows for a unique insight into simultaneous neurodegenerative processes underlying AD in a way that cannot be achieved using unimodal imaging methods.
PET/MRI has also been used to investigate WM tracts forming the basis of functional brain networks such as the DMN, as WM damage is consistently observed in AD postmortem studies. 71 There is also increasing evidence that neurodegeneration spreads via brain networks. 72

| Diagnostic utility
An important consideration for the clinical applicability of PET/MRI in dementia is the perception of clinical utility. Compared to other cerebral disorders (e.g., tumours, stroke), dementia and associated pathology is underemphasised in radiology teaching, with few radiological reports suggesting dementia diagnoses despite identification of characteristic atrophy patterns. 74 In practice, MRI and PET images are often interpreted by different subspecialist radiologists, with integration of these separate findings often left to the referring clinician. 75 Mukku et al. 43 examined the diagnostic utility of 18 FDG-PET/MRI in dementia diagnoses, finding that biomarkers of hypometabolism and structural atrophy showed 90.5% concordance with clinical diagnoses, and further assisted with diagnostic subtyping.
Although offering limited benefit in two patients, the authors note that PET/MRI reduced the need for sedation in distressed cognitively-impaired patients due to its relative simplicity and reduced scanning time. Tahmasian et al. 49 reported a similar benefit of multimodal imaging for differentiating dementia subtypes. Using 18 FDG-PET/MRI, they found classification accuracies to be between 77.5% and 97.5% for subtypes including bvFTD, SD, and PNFA, findings that were significantly superior to unimodal approaches.
Further support is provided by Schutz et al. 48