Linking To And Excerpting From “The use of neuroimaging techniques in the early and differential diagnosis of dementia”

Posted on December 19, 2024 by Tom Wade MD

Today, I review, link to, and excerpt from The use of neuroimaging techniques in the early and differential diagnosis of dementia [PubMed Abstract] [Full-Text HTML] [Full-Text PDF]. Mol Psychiatry. 2023 Oct;28(10):4084-4097. doi: 10.1038/s41380-023-02215-8. Epub 2023 Aug 22.

There are 101 similar articles in PubMed.

The above article has been cited by 9 articles.

All that follows is from the above article.

Abstract

Dementia is a leading cause of disability and death worldwide. At present there is no disease modifying treatment for any of the most common types of dementia such as Alzheimer’s disease (AD), Vascular dementia, Lewy Body Dementia (LBD) and Frontotemporal dementia (FTD). Early and accurate diagnosis of dementia subtype is critical to improving clinical care and developing better treatments. Structural and molecular imaging has contributed to a better understanding of the pathophysiology of neurodegenerative dementias and is increasingly being adopted into clinical practice for early and accurate diagnosis. In this review we summarise the contribution imaging has made with particular focus on multimodal magnetic resonance imaging (MRI) and positron emission tomography imaging (PET). Structural MRI is widely used in clinical practice and can help exclude reversible causes of memory problems but has relatively low sensitivity for the early and differential diagnosis of dementia subtypes. 18F-fluorodeoxyglucose PET has high sensitivity and specificity for AD and FTD, while PET with ligands for amyloid and tau can improve the differential diagnosis of AD and non-AD dementias, including recognition at prodromal stages. Dopaminergic imaging can assist with the diagnosis of LBD. The lack of a validated tracer for α-synuclein or TAR DNA-binding protein 43 (TDP-43) imaging remain notable gaps, though work is ongoing. Emerging PET tracers such as 11C-UCB-J for synaptic imaging may be sensitive early markers but overall larger longitudinal multi-centre cross diagnostic imaging studies are needed.

PubMed Disclaimer

Introduction

Dementia, the umbrella term for global cognitive decline causing functional impairment, affects ~55 million people worldwide []. The most common types of dementia are Alzheimer’s disease (AD), Lewy Body dementia (LBD, a term which includes both dementia with Lewy Bodies (DLB) and Parkinson’s disease dementia), vascular dementia (VaD) and Frontotemporal dementia (FTD). There are several other less common forms of dementia such as Progressive supranuclear palsy (PSP), Corticobasal degeneration (CBD), Huntington’s disease, hippocampal sclerosis, prion disease and many others []. In all the degenerative dementias the onset of symptoms is associated with already established brain pathology, which develops many years before symptom onset [, ].

Early and accurate diagnosis of the cause of dementia is important for a number of reasons, including optimising clinical management, offering opportunities for secondary prevention, increasing prognostic accuracy and the identification of the right people to benefit from disease modifying therapies, once these become available [–]. Brain imaging by means of structural magnetic resonance imaging (MRI), at a single time point or performed serially, along with positron emission tomography (PET) imaging using 18F-fluorodeoxyglucose PET (FDG-PET), dopaminergic single-photon emission computerised tomography (SPECT) and PET imaging (for DLB) and PET ligands for amyloid and tau are the best established and validated imaging methods for both early and specific diagnosis of dementia subtype. Brain imaging along with cerebrospinal fluid (CSF) biomarkers have helped to establish the ATN (A = amyloid, T = tau, N = neurodegeneration) framework for the diagnosis of AD which is being used to define the presence of AD pathology at preclinical and prodromal stages []. This is of particular importance for the use of disease modifying treatments. This review summarises findings from diagnostic brain imaging studies, discusses novel developments in molecular imaging and outlines important future directions in the field.

Structural MRI

Structural imaging with computerised tomography (CT) or MRI is widely used in clinical practice and recommended by several diagnostic and research guidelines for the assessment and diagnosis of people with dementia [, ]. MRI is preferred over CT when available. Structural imaging can exclude conditions such as space-occupying lesions, stroke, normal-pressure hydrocephalus, as well as many other pathologies and can also help with the differential diagnosis of dementia based on characteristic patterns of atrophy, white matter changes and the presence or absence of cerebrovascular disease []. The changes can be summarised in radiological reports and quantified using a variety of methods including visual assessment with validated rating scales, volumetric assessment using a region of interest approach or more detailed quantification using methods like voxel based morphometry or assessment of cortical thickness. It is however important to note that in these cases the role of neuroimaging is to corroborate a diagnosis based on identification of a clinical syndrome.

For example, in AD there is generalised atrophy with focal changes in the temporal lobe, especially the hippocampus. Early onset AD may be associated with more posterior and less temporal lobe atrophy. FTD is associated with anterior temporal pole and frontal atrophy, with semantic dementia subtype associated with asymmetric temporal atrophy. DLB shows relative preservation of the hippocampus and occipital and subcortical atrophy while vascular cognitive impairment and VaD are associated with cortical and subcortical vascular changes (latter including white matter hyperintensities (WMH), lacunes, enlarged perivascular spaces and microbleeds) [, ].

Volumetric MRI

Visual assessment of scans using rating scales are reliable and offer diagnostic accuracy equivalent or better to unstructured scan evaluation by expert raters with an area under the curve (AUC) ranging from 0.67 to 0.97 for the differential diagnosis of different types of dementia [, ]. Considering automated volumetric analyses, structural MRI atrophy maps to identify patterns characteristic for AD, LBD or FTD showed that these atrophy maps had 90% sensitivity and 84% specificity for AD, 78.7% sensitivity and 98.8% specificity for LBD and 84.4% sensitivity and 93.8% specificity for FTD []. A study that included 504 individuals with AD, FTD, VaD, DLB and control subjects, quantifying volumetric, morphometric and vascular characteristics showed that MRI was accurate in 70.6% of cases. VaD groups were detected with 96% sensitivity, controls with 82%, and AD with 74%. DLB was the most difficult for detection with 24% sensitivity []. Koppel et al. compared 134 cases with AD, FTD, LBD and MCI and were able to separate healthy elderly from patients with dementia with an AUC of 0.97 []. Ma et al. found that a proposed deep learning framework achieved an overall accuracy of 88.28% in differentiating AD from FTD [] while Yu et al. showed that an AD atrophy index could identify AD and FTD from controls with an AUC of 0.88 and an FTD atrophy index could identify FTD from AD with an AUC of 0.93 [].

In terms of structural changes in LBD, brain atrophy is seen but the relative preservation (compared to the marked atrophy in AD) of cortical structures and the medial temporal lobe is well established (Fig. 1) and is a supportive feature in the diagnostic criteria for DLB [–]. Mak et al. compared 35 DLB, 36 AD and 35 controls and suggested that the relative preservation of the hippocampus in DLB is characterised by preservation of the cornu ammonis (CA) 1, fimbria and fissure while all other hippocampal subfields had comparable atrophy in both AD, DLB and control groups [].

Fig. 1. Representative MRI scans in different types of dementia.

Fig. 1

The figure shows representative MRI scans from a non-demented control and from patients with Dementia with Lewy Bodies (DLB), Alzheimer’s disease (AD) and frontotemporal lobe degeneration (FTLD). It highlights the characteristic patterns of atrophy with relative preservation of the hippocampus in DLB, severe hippocampal atrophy in AD and temporal pole atrophy in FTLD. These scans are from the Neuroimaging of Inflammation in Memory and Other disorders (NIMROD) study cohort. Images are courtesy of Dr Elijah Mak, University of Cambridge, UK.

Nevertheless, assessment of atrophy in older populations can be more challenging. Barkhof et al. carried out post-mortem MRI in 132 autopsy brain tissues from the Vantaa 85+ community study and compared visual ratings of medial temporal lobe atrophy (MTA) to neuropathological findings []. Overall, high MTA scores were associated with clinical dementia with sensitivity of 63% and specificity of 69% for AD [].

In terms of use of MRI in earlier dementia stages, a review summarising 33 studies in structural neuroimaging for the early diagnosis of AD in people with MCI (including 3935 participants) concluded that there is lack of systematic approach in data collection, analysis and interpretation [24]. Using machine learning to quantify neurodegeneration patterns in structural MRI, studies have managed to predict MCI conversion to AD with modest accuracy ranging from 63 to 85% depending on the cohort, imaging modality and models used [25–27].

In a study focusing on prodromal DLB, Kantarci et al. compared 56 patients with MCI and features of DLB to 112 cognitively unimpaired controls. They showed that at baseline prodromal DLB was associated with atrophy in the nucleus basalis of Meynert, measured through region of interest analysis from an in-house atlas of the substantia inominata [].

Overall, a large body of evidence suggests that volumetric analyses on MRI have an important role for the differential diagnosis of dementia with high specificity when changes are present, especially using automated analyses. Volumetric MRI changes however lack sensitivity in early prodromal dementia stages as they mostly correlate with established neurodegenerative processes. It is important to highlight that research studies generally utilise well characterised participants recruited at clinical academic settings enhanced for patients with a more clear-cut diagnosis following certain inclusion and exclusion criteria set out by each study and therefore the quoted calculated sensitivities and specificities are likely to be overestimated with regards to real world patient settings. There is a high likelihood of co-pathology in patients with dementia and often patients with mixed dementias may show neuroradiological features characteristic of more than one type of dementia, e.g. atrophy and infarcts.

White matter hyperintensities and cerebral microbleeds

A substantial burden of WMH, lacunes and strategic infarcts are consistent with vascular cognitive impairment and dementia []. Nevertheless, there is a significant association between WMH, grey matter atrophy and cognitive decline in AD and FTD. Dadar et al. compared 571 normal aging subjects with 551 MCI, 212 AD, 125 FTD and 271 PD from the Alzheimer’s disease neuroimaging initiative (ADNI), the frontotemporal lobe degeneration neuroimaging initiative and the Parkinson’s Progression Markers Initiative datasets []. They found significantly higher WMH loads in MCI, AD and FTD compared to controls. WMH were related to grey matter atrophy in insular and parieto-occipital regions in MCI/AD and frontal regions and basal ganglia in FTD. WMH were associated with more severe cognitive deficits in AD and FTD but had no impact in MCI and PD. Importantly, WMH are associated with higher cardiovascular risk factors in midlife [, ]. However, they have also been linked to tau pathology, a reminder that WMH cannot always be taken to represent vascular disease [].

Cerebral microbleeds are common in people with AD, DLB, stroke and trauma []. They represent iron accumulation in perivascular spaces and are linked with vascular disease and cerebral amyloid angiopathy []. Lobar microbleeds are associated with amyloid pathology while deep/basal ganglia microbleeds are associated with hypertensive small vessel disease [, ]. Their role in the pathophysiology and diagnosis of different types of dementia is not yet clear but they were found to be of similar frequency among patients with AD and DLB, albeit with greater densities in the parietal, temporal and infratentorial regions in AD compared to DLB []. Meanwhile in patients with first episode ischaemic stroke, three or more microbleeds were associated with higher risk of developing vascular dementia []. Studies in younger and presymptomatic individuals showed that cerebral microbleeds are significantly higher in number in APOE ε4 carriers []. Studies in unimpaired populations that were longitudinally followed up showed that high microbleed number (>3–4) is associated with an increased risk of cognitive deterioration and dementia [, ].

Serial MRI

Serial MRI has been used as a measure to improve differential diagnosis of dementias and has often been incorporated as a secondary outcome measure in clinical trials in AD. It is well established that serial atrophy rates are significantly higher, approximately fourfold, in AD compared with similarly aged controls. Rates are also higher than controls in VaD and FTD. One study found people with AD had an atrophy rate of 2.0% per year compared to 1.9% in VaD and 1.4% in DLB []. Further studies have also shown greater atrophy rates in AD compared to DLB which showed similar atrophy rates to controls, a finding in keeping with the lesser overall atrophy in DLB []. In parallel, studies have shown that DLB with co-existing AD pathology is associated with faster rates of progression suggesting that the presence of AD pathology is likely the driver of atrophy [, ]. In a study comparing behavioural variant (bv)FTD, AD and healthy controls with consecutive scans over at least 12 months, Frings et al. showed that annual volume decline was larger in bvFTD, then AD, and then in controls, predominantly in white matter of temporal areas and orbitofrontal grey matter []. In summary, studies in longitudinal atrophy in dementia can support a specific diagnosis but considering the interval required between scans and the lack of sensitivity may not be as useful for early diagnosis.

Diffusion weighted imaging MRI

Diffusion tensor imaging (DTI) is an MRI technique that provides information on the orientation and integrity of white matter tracts through measuring parameters associated with diffusion of water molecules in the brain. It generates measures of fractional anisotropy (FA) and mean diffusivity (MD) of water molecules in a region of interest, and studies have shown lower FA and higher MD (associated with reduced axonal integrity) in MCI and AD compared to controls [, ]. DTI data are also used for other analytical methods such as tractography to investigate tract integrity. Compared to AD, FTD was associated with lower FA in frontal regions [] while specific tractography of long and short white matter tracts suggested that large scale tracts are particularly vulnerable to vascular disease in FTD and associated with executive dysfunction while short tracts were associated with semantic symptoms []. DTI differences may be able differentiate typical AD from the posterior cortical atrophy variant of AD showing differences in regions including parietal and temporal lobe areas []. Meanwhile, DLB was associated with increased amygdala MD compared to AD []. Spotorno et al. compared 34 LBD patients with 16 PSP and 44 healthy controls using a FA score from a combination of regions sensitive to pathologic features of PSP []. They distinguished PSP from LBD with AUC of 0.97 with sensitivity of 0.94 and specificity of 0.91. They validated these results in a second cohort with 34 patients with PSP, 25 LBD and 32 controls with an AUC of 0.96 [].

In a study using DTI data for tractography analyses in the nucleus basalis of Meynert (NBM), Schumacher et al. compared the cholinergic white matter pathways in 46 AD, 48 DLB 35 MCI-AD, 38 MCI-LB and 71 control participants and found that MD of the lateral pathway was higher in the dementia and MCI groups and that particularly in MCI, loss of integrity of both NBM pathways was associated with an increased risk of progression to dementia []. Recent novel studies assessing cortical microstructure via cortical mean diffusivity (cMD) were found to be more sensitive than macrostructural neurodegeneration. Along with free water fraction (FW), cMD changes have shown that in the AD continuum, microstructural changes show a biphasic trajectory. There is increased cortical thickness and decreased cMD and FW in the initial presymptomatic dementia stages while there is decreased cortical thickness and increased cortical MD and FW in symptomatic changes []. cMD was found to be associated with PET tau in vulnerable to AD pathology regions and predict hippocampal atrophy rate and cognitive decline while cortical microstructure changes in the frontal and parietal areas appeared to be sensitive biomarkers for microstructural alterations in FTLD subtypes [–].

In summary, DTI has been successfully used in research studies to show biologically plausible differences between dementia subtypes and to predict progression from MCI to dementia. However, studies have been modest in size and from single sites, and no clearly established cutoffs or harmonised, validated methods are available, limiting the ability for DTI to be used in clinical practice.

Assessment of blood flow and perfusion

MRI can be used to measure blood flow, either through the use of injected contrast agents or through magnetically labelling blood, a technique known as arterial spin labelling (ASL). Blood flow closely matches the patters of hypometabolism on FDG-PET due to the close coupling between perfusion and metabolism in brain [, ]. ASL was shown to be comparable to FDG-PET in identifying AD compared to controls with an AUC of 0.91 []. However, in a study using PET-MR that compared FDG-PET with ASL, Ceccarini et al. compared a combined group of 27 patients with AD, DLB, FTD and 30 matched controls and found that FDG-PET performed better than ASL []. In keeping with patterns on FDG-PET, DLB has been associated with reduction in cortical perfusion on ASL in higher visual areas compared to AD []. Such findings were similar in a cohort with MCI-LB with reduction in posterior parietal and occipital regions but relatively preserved posterior cingulate [].

Using ASL in 32 early onset AD and FTD patients and 32 controls, ASL achieved an AUC of 86–91% for the correct classification of patients with dementia and potentially adds diagnostic value when combined to structural MRI data []. In an attempt to differentiate early AD from bvFTD, Stekeete et al. compared 13 AD with 19 bvFTD and found that AD was associated with hypoperfusion in the posterior cingulate cortex and this differentiated, to some extent, AD from bvFTD though AUC was modest at 0.74 []. In a comparison of ASL with FDG-PET in ten FTD patients and ten controls, Anazodo et al. found that FDG-PET outperformed ASL in inter-rater reliability as well as sensitivity and specificity in discriminating patients from controls (ASL AUC 0.75 and FDG-PET AUC 0.87) []. ASL findings however in AD and FTD are not consistent, with an earlier study by Du et al. in 21 FTD, 24 AD and 15 controls showing that FTD and AD display different spatial patterns of hypoperfusion on ASL and were able to classify AD from FTD with an AUC of 0.87 [].

In combining DTI with ASL to differentiate early onset AD with early onset FTD, Bron et al. compared 24 AD and 33 FTD with 34 controls and used support vector classification finding that ASL and DTI combined with structural MRI could differentiate AD from FTD with AUC 0.84 compared to structural MRI alone with AUC of 0.72 []. ASL has further shown some promise in the differential diagnosis between AD and DLB with distinct patterns seen in DLB compared to AD and cognitively normal individuals [, ].

In studies focusing on the blood brain barrier (BBB), dynamic contrast-enhanced MRI with temporal and spatial resolutions to measure BBB permeability have shown a breakdown of the BBB in the hippocampus of patients with early cognitive dysfunction, independent of their amyloid and tau biomarker status and this also occurs in normal aging [, ]. In a follow up study, BBB breakdown in the hippocampus and medial temporal lobe was able to distinguish Apolipoprotein (APOE) ε4 from non-ε4 carriers [].

In summary, studies in blood flow and perfusion in dementia have shown great potential for the early detection of neurodegeneration. However, the differences detected are subtle and multi-centre studies in large cohorts are lacking in order to test their potential use in clinical practice. FDG-PET seems to outperform methods for MRI cerebral blood flow and is more widely adopted.

Functional MRI

Functional MRI measures blood flow changes to determine which parts of the brain are engaged when performing a test or at rest and can study brain networks that are functionally connected []. The default mode network (DMN) activity has been found to be abnormal in AD [–]. Resting state fMRI has been used for the differential diagnosis between AD and bvFTD with decreased connectivity in the lateral visual cortical network, lateral occipital and cuneal cortex as well as the auditory system network and angular gyrus seen in bvFTD compared to decreased connectivity in the dorsal visual stream network and lateral occipital and parietal cortex in AD []. The disrupted functional connectivity seen in AD has been associated with tau burden and neuroinflammation measured through in vivo PET imaging [, ].

In DLB, functional connectivity has been used to study symptoms of cognitive fluctuations. Peraza et al. found that the DMN is unaffected in DLB compared to controls but DLB patients show differences in the left fronto-parietal, temporal and sensory motor-network suggesting a potential al role of attention-executive networks in the aetiology of cognitive fluctuations in DLB []. Recently, it has been suggested that higher physical activity is associated with greater connectivity in the DMN and may be one of the pathways through which exercise promotes resilience to neurodegeneration []. Overall studies in functional MRI in dementia have shown differences in resting state functional connectivity and have pointed to specific networks and regions affected in each dementia, however there seems to be a significant overlap among diagnostic groups and unlikely to be useful clinically.

Magnetic resonance spectroscopy, electroencephalography and magnetoencephalography

Magnetic resonance spectroscopy, an MRI method measuring metabolite levels in the brain, has been explored in dementia research. The present findings suggest the need for larger studies with more consistent methodology before being considered for use in clinical practice (for systematic review of studies please see [] Similarly, studies using electroencephalography and magnetoencephalography for the differential diagnosis of dementia are lacking and more research is needed for these important imaging modalities [, ].

More advanced MRI methods for the diagnosis of dementia

More advanced ways of brain MRI imaging, for example the neurite orientation dispersion and density imaging (NODDI) have shown great promise for the early and differential diagnosis of dementia. NODDI is a DTI technique that derives measures of orientation dispersion index and neurite density index and can detect distinct microstructural features []. NODDI changes have been shown as part of brain aging and seem to complement traditional DTI measures by characterising the cytoarchitecture of brain tissue [–]. In dementia, NODDI has been studied in young onset AD [, ] showing it is affected in regions associated with early atrophy in AD while NODDI measures in animals correlate with tau burden []. NODDI measures seem to be lower in temporal and parietal cortical regions in MCI when compared to controls while they are lower in parietal, temporal and frontal regions in AD []. In a multi-centre study, Raghavan et al. tested the associations between NODDI and neuropathological changes in the Mayo Clinic Study of Aging and the Mayo Alzheimer Disease research centre cohorts and found that cerebrovascular disease, tau and TDP-43 pathologies cause white matter microstructural damage seen with the NODDI methods []. NODDI however as an emerging new imaging method maybe subject to biases, e.g. presence of CSF partial volume in individuals with larger ventricles or atrophy due to degeneration []. While NODDI is a very promising new method of DTI, there are no studies yet looking specifically at its potential at the early presymptomatic diagnosis of dementia or its potential role in the differential diagnosis of the most common types of dementia.

Most MRI studies today have been done in the widely available 1.5 and 3 Tesla scanners. New technologies allow for higher power magnet and 7 Tesla MRI (7T) scanners are now becoming more widely used and allow higher signal-to noise resolution. 7T studies have measured hippocampal subfield volumes in AD and imaged the substantia nigra in PD []. Van Rooden et al. showed that increased cortical phase on 7T may reflect early stages of amyloid beta (Aβ) pathology in AD [] while Theyshon et al. found that the use of 7T in vascular dementia may be more sensitive in the detection of cerebral microbleeds []. 7T MRI imaging has a huge promise in research and clinical practice but there are still challenges with regards to the costs, operating complexity, and availability, while diagnostic superiority for dementia over lower field strength MRI remains to be shown [].

F-Fluorodeoxyglucose (FDG) PET

FDG-PET changes are a supportive feature in the AD and DLB diagnostic criteria and FDG-PET is widely used clinically for the diagnosis of AD and the differential diagnosis of different subtypes of dementia [, , ]. FDG-PET is a readout of the local cerebral metabolic rate of glucose consumption []. Reduced uptake of the radioactive compound is suggestive of hypometabolism which in the brain correlates with reduced synaptic activity and evidence of neurodegeneration, correlating with brain atrophy and tau pathology [, ]. It is analysed either using expert visual rating or specialised quantitative analytical software [–]. Meta-analytic evidence suggests that FDG-PET has 90% sensitivity and 89% specificity in diagnosing AD from controls []. FDG-PET was found to have superior diagnostic accuracy in AD and DLB compared to hexamethylpropyleneamine oxime (HMPAO) SPECT with an AUC of 0.93 for FDG-PET compared to 0.72 for HMPAO SPECT []. A recent systematic review by Fink et al. analysed the accuracy of FDG-PET comparing AD to non-AD dementias and showed a median sensitivity of 0.89 and specificity of 0.74 [].

The patterns seen in AD involve hypometabolism of the temporal and parietal lobes (Fig. 2). In dominantly inherited AD, hypometabolism on FDG-PET can be detected as early as 10 years before symptom onset []. There is evidence that FDG-PET may also predict conversion from MCI to dementia however limitations relating to individual studies with small sample sizes do not allow reliable meta-analyses of such studies and pooled results show large range in the sensitivity (56–100%) and specificity (24–100%) for the role of FDG-PET in predicting conversion from MCI to dementia [, ]. Considering the variability of FDG-PET it is therefore not recommended for clinical use at the MCI stage [].

Fig. 2. 18F-Fluorodeoxyglucose (FDG) PET in Alzheimer’s disease (AD) and Dementia with Lewy Bodies (DLB).

Fig. 2

FDG PET representative images showing reduced local cerebral metabolic rate of glucose consumption in cases of AD and DLB compared to a non-demented control study participant. The white arrows highlight the relative preservation of the hippocampus and posterior cingulate gyrus in DLB compared to AD and the occipital hypometabolism in DLB. These are FDG PET scans from the Study of the clinical utility, patient preference and cost benefit of SPECT and PET-CT brain imaging in the evaluation and diagnosis of Alzheimer’s Disease (Suspected-AD). Images are courtesy of Dr Michael Firbank, Newcastle University, UK.

Using FDG-PET data from the ADNI, Blazhenets et al. showed that FDG-PET in combination with amyloid PET and non-imaging variables may improve the prediction of conversion from MCI to AD and support the stratification of patients according to their conversion risks []. Levin et al. used FDG-PET in the ADNI dataset to subtype AD in ‘typical’, ‘limbic-predominant’ and ‘cortical-predominant’ types that correlate with the brain atrophy subtypes and the different clinical trajectories [].

 

This entry was posted in Dementia, Dementia Imaging, Neuroradiology. Bookmark the permalink.