Neuroimaging in Lewy body dementia

MRI has been widely used to investigate patterns of gray matter (GM) atrophy. Advances in image processing have enabled automatic extraction of whole brain cortical thickness, which is retrieved from structural MRI. Although DLB has demonstrated some overlap with the cortical atrophy patterns seen in AD, atrophy is generally less diffuse in DLB with moderate preservation of the medial temporal lobe structures [14‐16].

Cortical thickness assessment has also been shown to have high precision and sensitivity in identifying morphological changes, which arise from neuropathological changes. This method has, therefore, been employed in several studies as a way to differentiate DLB from AD, PDD and healthy controls. Investigations into cortical thickness alterations in DLB revealed relatively small GM change, primarily affecting the posterior parietal areas, as opposed to the patterns of GM change affecting the temporoparietal association cortices in AD [17]. These findings are in corroboration with the notion that DLB is a result of neuronal synaptic dysfunction, not neuronal loss. Through carrying out a multivariate classification study of cortical thickness, Lebedev and colleagues demonstrated that this method has the ability to differentiate DLB from AD with 82.1% sensitivity and 85.7% specificity [18]. Specifically, AD was characterized by patterns of cortical thinning within the temporal pole, subgenual cingulate regions and the parahippocampus, whereas regional thinning was localized to the superior temporo-occipital and lateral orbito-frontal regions, as well as the middle and posterior cingulate in DLB [18]. The finding of AD exhibiting greater temporal involvement compared to DLB has been a homogenous result across several structural imaging studies [19, 20].

Although investigations into hippocampal atrophy have revealed that DLB patients have less severe atrophy compared to AD patients [21, 22], with the entorhinal cortex, CA1 and subiculum areas of the hippocampus being most affected in AD [23], recently, Delli Pizzi et al. explored the differential contribution of hippocampal subfields and adjacent extrahippocampal structures to the pathophysiology of AD and DLB [24]. They reported that the cornu ammonis and subiculum were preserved in DLB, but the perirhinal cortex and parahippocampus were damaged, highlighting the differential alteration of hippocampal subfields and adjacent extrahippocampal structures in DLB and AD [24].

Studies have also demonstrated greater atrophy in the substantia innominate, with increased dorsal mesopontine GM atrophy distinguishing patients with clinically diagnosed DLB from AD [25]. These findings may suggest a greater cholinergic dysfunction in DLB, perhaps related to the presence of midbrain synuclein pathology.

Using diffusion tensor MRI, Watson et al. revealed that the parieto-occipital white matter tracts were preferentially affected in DLB, though this appears to be an early phenomenon, as AD demonstrated a greater longitudinal increase in mean diffusivity in parietal and temporal regions compared to DLB, with no evidence of longitudinal changes in mean diffusivity or fractional anisotropy in DLB relative to controls [26]. However, DLB was differentiable from AD given that it was associated with reduced fractional anisotropy in the pons and left thalamus, highlighting that, despite similar levels of dementia severity, patterns of DTI changes in DLB and AD varied [26].

A recent study by Shams et al. demonstrated that MRI of the swallow tail sign may have diagnostic potential in DLB [27], given that the largest dopamine-containing cluster within caudal and posterolateral part of the substantia nigra (nigrosome 1) is highly affected in parkinsonian syndromes. More specifically, Shams et al. reported that a hypointense nigrosome 1, as visualized on iron-sensitive susceptibility-weighted imaging (SWI), was more common in DLB compared to AD, frontotemporal dementia and controls. This was in corroboration with Kamagata et al. who reported that measuring nigrosome 1 hypointensity with SWI achieved 90% diagnostic accuracy (93% sensitivity and 87% specificity) in DLB [28].

Attempts to compare GM loss between DLB and PDD have revealed a pattern of more pronounced GM loss in DLB compared to PDD, which is in line with the fact that DLB encompasses greater amyloid burden [29]. It is important to note, however, that localisations of GM reductions in DLB relative to PDD vary amongst different studies. For example, Burton et al. were unable to identify distinct cortical atrophy profiles of DLB and PDD [30], but Beyer et al. reported GM reductions in the temporal, parietal and occipital lobes in DLB using a voxel-based morphometry (VBM) approach [31]. Alongside the temporal and parietal atrophy, Lee et al. also reported occipital and striatal GM reductions in DLB [32]. Studies investigating correlation patterns between brain structure and clinical and neuropsychiatric manifestations of the disease, revealed that decreased GM volume of the anterior cingulate, right hippocampus and amygdala were associated with cognitive performance [33], whilst reduced GM volume in the left precuneus and inferior frontal lobe correlated with visual hallucinations in DLB, but not in PDD [34].

Although only a few fMRI studies have examined BOLD signal in DLB, differential patterns of functional connectivity in DLB compared to AD have been reported (Table 2). Using the precuneus as the seed region, Galvin et al. reported that DLB patients exhibit increased connectivity in the inferior parietal cortex and putamen, and decreased connectivity in the fronto-parietal operculum, medial prefrontal cortex and the primary visual cortex compared to AD, whilst a reversal of connectivity was observed in the right hippocampus [35]. Independent component analysis (ICA) has demonstrated that DLB display greater connectivity in the default mode network compared to AD [36], which contrasts with previously reported connectivity dysfunctions between anterior and posterior segments of the default mode network in AD, when compared to healthy controls [37]. Furthermore, increased connectivity between the putamen and frontal, temporal and parietal regions has been illustrated by DLB patients in comparison to AD patients, with the authors suggesting that this may be related to the prominent parkinsonian features in DLB [38]. Consistent with the moderate preservation of memory function observed in DLB as opposed to AD, hippocampal connectivity has not been shown to differ in DLB compared to healthy controls, though the left hippocampal connectivity was identified to be higher in AD compared to controls, potentially reflecting a compensatory mechanism [38].

A recent study by Schumacher et al. aimed to explore within- and between-network connectivity in a range of resting state networks, being the first to investigate how DLB affects connectivity between these resting state networks [39]. DLB patients displayed more decreases in within-network connectivity compared to controls, primarily in temporal, motor and frontal networks. In contrast, long-range functional connectivity appeared to be intact in DLB, with increased connectivity only identified between a frontal and a temporal network [39]. Only subtle differences were observed when AD and DLB were compared, suggesting a potential overlap in their resting state functional connectivity.

Given the prominent prevalence of visuoperceptual impairments in DLB, a task-based fMRI study employed visual presentations of motion, color and face paradigms to explore the functional integrity of the visual system in DLB. They discovered that DLB patients exhibited greater activation in the superior temporal sulcus compared to AD, specifically during the motion task [40]. However, these findings were not replicated by Taylor et al., who reported that DLB patients did not exhibit any significant differences in functional response to objects, motion stimuli or checkerboard in V1 and V2/V3 compared to controls [41], proposing that function in the lower visual areas is relatively preserved. Interestingly, however, ROI analysis demonstrated that the DLB group had a reduction in V5/MT (middle temporal) activation when responding to motion stimuli [41]. Taken together, these results imply that, in DLB, functional abnormalities affect the visual association areas, as opposed to the primary visual cortex, though it is difficult to decipher whether deviations at higher levels of the visual system contribute to the hallmark visuoperceptual impairments and visual hallucinations seen in DLB.

Although studies have demonstrated alterations in functional connectivity in PDD [42‐45] and DLB [35, 38, 46‐48], these were reported when comparing these disease groups against healthy controls. One study has, however, compared DLB and PDD directly with the aim of identifying disease-specific functional connectivity patterns (Table 2). Peraza et al. reported that, for seeds situated within the fronto-parietal network, DLB patients exhibited greater alterations in functional connectivity than PDD when compared to healthy controls, predominately at the precentral and postcentral gyri, cerebellar, occipital and temporal regions, whilst in PDD, changes in functional connectivity were limited to the frontal cortices and precuneal [49]. Interestingly, although the supplementary motor area seed revealed similar regional functional connectivity alterations in the pre- and postcentral gyri, cerebellar, temporal, precuneal and occipital regions, these alterations were more apparent in PDD than in DLB, potentially reflecting the prominent parkinsonism and motor dysfunction in PDD compared to DLB [49]. However, Peraza et al. reported that no significant differences were found when DLB and PDD groups were compared to each other. Taken together, these results suggest that there are subtle functional differences between both diseases, which may be driven by their distinct pathological trajectories, thus potentially reflecting the chronological manifestation of cardinal symptoms in the Lewy body dementias.

[¹⁸F]FDG PET is used in detecting cerebral glucose metabolism, which is impaired in cases of neuronal degeneration and synaptic pathology. It has been widely used in assessing dementias, and has been proven to be an effective tool in aiding the diagnosis of AD and monitoring its progression [64‐66].

In DLB, the topographical pattern of hypometabolism includes mainly the occipital areas, visual association cortices and the posterior parietotemporal areas [67‐69], though in AD, decreased cerebral metabolism tends to involve other areas as well [70]. In a recent multimodal PET study assessing amyloid-β deposition and cerebral glucose metabolism, with [¹¹C]PiB and [¹⁸F]FDG, respectively, Chinese patients with probable DLB exhibited cortical amyloid-β deposition, as well as hypometabolism in the temporo-parieto-occipital region, insular, precuneus, frontal lobe, posterior cingulate and caudate nuclei [71].

Another characteristic feature of DLB is preserved metabolism in the posterior cingulate area when compared to the precuneus and cuneus [72]. This is called the cingulate island sign and can be related to the common visual hallucinations in patients with DLB. Furthermore, it harbors a notable sensitivity and specificity [66, 73]. The cingulate island sign has been inversely correlated with neurofibrillary tangle pathology in autopsy studies [73]. A recent study has also reported association of cingulate island sign, not only with medial temporal lobe atrophy, but with clinical symptoms (cognitive impairment, visual hallucinations) of DLB patients as well [73].

Dopamine transporter (DAT) imaging with SPECT using as a radiotracer [¹²³I]FP-CIT-SPECT has been a valuable tool in assessing dopaminergic function in vivo. Decreased DAT uptake in basal ganglia is considered a supportive diagnostic feature according to current consensus diagnostic criteria [74, 75]. The diagnostic accuracy is even higher when applied in autopsy-proven cases of DLB [76‐78]. Yielding a sensitivity of 88% and a specificity of 100% over non-DLB cases, [¹²³I]FP-CIT-SPECT is a highly useful diagnostic tool [74]. A meta-analysis referring to 419 patients enrolled in 4 studies, showed a remarkable diagnostic accuracy, with a mean sensitivity of 86.5% and a mean specificity of 93.6% [79]. When comparing pathologically proven cases to clinical diagnosis, [¹²³I]FP-CIT-SPECT has demonstrated increased accuracy in differentiating DLB from AD [80, 81]. In DLB, there is a decreased level of DAT, which is helpful in differentiating from AD where DAT is preserved [82, 83]. On the other hand, DAT imaging is not useful in discriminating DLB from PD-MCI and PDD, where there is a profound loss of DAT in the striatum [84]. Although DAT imaging possesses an inability to distinguish between parkinsonian syndromes, a recent study by Takaya et al. revealed that a combination of disease-specific perfusion patterns and striatal DAT activity accurately differentiates between atypical parkinsonian syndromes and Lewy body dementia [85]. However, in the rare cases of DLB where nigrostriatal degeneration is minimal and cortical pathology is the prominent feature, false negative results might occur. As for discriminating DLB patients from frontotemporal dementia or atypical parkinsonian syndromes (i.e. progressive supranuclear palsy (PSP), corticobasal degeneration (CBD)), [¹²³I]FP-CIT-SPECT should not solely be accounted for as a reliable method of investigation [86]. The clinical phenotype should always be considered when interpreting findings regarding the above-mentioned conditions.

[¹²³I]MIBG cardiac scintigraphy is widely used to assess cardiac postganglionic sympathetic degeneration, which is a common feature in neurodegenerative diseases with Lewy Bodies pathology. [¹²³I]MIBG is a promising biomarker with the ability of excluding AD and predicting conversion of possible to probable DLB [87‐89]. A large multicenter study including 133 patients, diagnosed according to the consensus criteria, highlighted similar sensitivity and specificity to [¹²³I]FP-CIT-SPECT [90, 91]. Although the [¹²³I]MIBG is a credible modality, certain pitfalls should be considered. The presence of diabetes mellitus or cardiac disease might provide false positive results [92]. Thus, such patients should be excluded from undergoing cardiac scintigraphy for diagnostic purposes. When comorbidities are taken into account, [¹²³I]FP-CIT-SPECT may have a distinguishable diagnostic significance in the clinical setting [93].

Positive amyloid imaging is a classic feature of AD, with plaque deposition becoming apparent years after clinical symptomatology. Incorporation of amyloid imaging in AD consensus diagnostic criteria highlight the importance of such findings [94]. Furthermore, it may be proven elusive in early detection of disease pathology, disease monitoring and as a biomarker in disease-modifying trials with treatment targeting amyloid deposition. In DLB apart from α-synuclein aggregation, in some cases, pathology is also characterized by amyloid-β and tau deposition [95, 96]. The concurrence of the above-mentioned events leads to greater cognitive impairment [97].

Subsequently, imaging amyloid-β and tau deposition could potentially elucidate the association between AD-related pathology and α-synuclein aggregation. [¹¹C]-Pittsburgh compound B ([¹¹C]PiB) has been the most used radioligand to assess amyloid-β deposition in patients with DLB. Patients with DLB have shown increased [¹¹C]PiB retention when compared to patients with PD or PDD and reduced retention when compared to patients with AD [67]. However, although the load of amyloid-β deposition cannot distinguish DLB from AD, it can be associated with the pace of cognitive decline in DLB patients [98, 99]. Other studies have associated amyloid pathology to the time-onset of cognitive features when related to parkinsonism [62]. Meta-analyses highlighted that 68% of patients with a diagnosis of probable DLB harbor abnormal [¹¹C]PiB retention [67, 100]. Regarding differences between DLB and PDD, it has been demonstrated that cortical amyloid-β burden is significantly high in DLB patients, which is comparable to amyloid-β retention in AD, but conversely to PDD patients where amyloid-β pathology is scarce [62]. Dementia severity has been to shown to be trivial in the differential load of amyloid-β between DLB and PDD, with amyloid deposition possessing the ability to differentiate DLB and PDD, despite their overwhelming overlap in clinical, neuropathologic and neuropsychologic features [94]. Gomperts et al., through measuring [¹¹C]PiB retention in Lewy body diseases such as DLB, PDD, PD and PD-MCI, found that amyloid-β burden was higher in DLB subjects compared to the other groups, with amyloid deposits being associated to cognitive impairment exclusively in DLB [97]. The early amyloid burden in DLB, comparative to PDD, may account for the variability in onset of dementia and parkinsonism between the two conditions. However, it is important to note that [¹¹C]PiB binds to amyloid fibrils, but not soluble amyloid oligomers, thus the possibility remains that both DLB and PDD have high levels of toxic amyloid oligomers, which could potentially underlie cognitive impairment in both conditions. Notably, [¹¹C]PiB retention patients with probable DLB or PDD, tend to have a similar pattern of cortical atrophy in MRI to patients with AD [101]. A recent study comparing [¹¹C]PiB binding to GM atrophy rates concluded that higher retention at baseline was correlated to increase loss of GM, greater ventricular expansion and cognitive impairment [101]. In concordance with novel therapeutic strategies in AD, where amyloid pathology is targeted, amyloid imaging will have an upgraded role when anti-amyloid treatments are available for DLB patients as well.

The in vivo evaluation of tau pathology in DLB has been lacking until recently. The radioligand fluorine 18-labeled AV-1451 ([¹⁸F]AV-1451), also known as [¹⁸F]T807, has been proven suitable to assess tau deposition. Pathological studies have confirmed the predisposition of [¹⁸F]T807 for tau protein in neurofibrillary tangles instead of amyloid-β plaques or α-synuclein in Lewy bodies [102]. A recent study highlighted that cortical [¹⁸F]AV-1451 uptake was highly variable and greater than in the controls, especially in the inferior temporal gyrus and precuneus [102]. Furthermore, increased binding in these regions was found to be associated with cognitive impairment, as measured by the mini-mental state examination (MMSE) and the Clinical Dementia Rating scale [102]. These finding indicate a role for tau pathology in DLB pathogenesis. A subsequent larger [¹⁸F]AV-1451 PET imaging study reported that [¹⁸F]AV-1451 uptake was substantively more extensive and severe in AD compared to DLB patients [103]. [¹⁸F]AV-1451 uptake within the medial temporal lobe completely discriminated AD dementia from probable DLB, with AD exhibiting highest medial temporal uptake and DLB exhibiting the lowest. Probable DLB subjects had higher [¹⁸F]AV-1451 uptake in the posterior temporoparietal and occipital cortex compared to healthy controls, though no correlations were found between uptake in these regions and clinical measures such as motor parkinsonism, visual hallucinations, cognition or the presence of REM sleep behavior disorder (RBD). Global cortical [¹¹C]PiB uptake, a marker of amyloid-β, was associated with elevated posterior temporoparietal and occipital [¹⁸F]AV-1451 uptake, indicating an atypical pattern of tau deposition in probable DLB [103]. Generally, there appears to be a gradient of increasing tau binding: from absent to minimal tau binding in cognitively normal PD, to low tau binding in PD patients with cognitive impairment, to intermediate tau binding in DLB and very high tau binding in AD [102, 104].

Pathological SNCA is detected in various forms, such as fibrils, Lewy bodies and oligomers. Moreover, SNCA deposits are abundant in other misfolded proteins, including tau and amyloid [105]. Thus, a radiotracer with high selectivity for SNCA over tau and amyloid is required to provide adequate accuracy. Other key features of a potential radiotracer include high affinity for SNCA aggregates, high penetration in the brain and prompt clearance.

Several potential compounds with a desirable profile and acceptable characteristics have been identified and the production of an accurate radiotracer for SNCA remains the greatest challenge of the neuroimaging community in movement disorders [106]. One of the first compounds that was tested in vitro was the benzoxazole BF227 [107]. Although [¹⁸F]BF227 harbored high affinity for amyloid and low affinity for SNCA in brain tissues, it was also evaluated in vivo in a cohort of MSA patients without fully overcoming interpretation issues [108]. A group of phenothiazine derivatives has also been investigated in animal studies, as potential compounds, due to their moderate selectivity for SNCA in PD brains [109]. [¹⁸F]WC-58a harbored a promising selectivity and affinity for synthetic SNCA fibrils; however, it proved to be too lipophilic with a slow clearance [110].

The development of a reliable SNCA radioligand is an unmet need regarding in-depth cohort stratification, monitoring disease progression and designing experimental treatments for synucleinopathies. The presence of incidental Lewy body disease among elderly is a caveat regarding the diagnostic utility of SNCA-PET. However, the capability of in vivo quantification could provide a valuable tool, especially when combined with other modalities to understand the full spectrum and progression of overlapping proteinopathies.

The importance of [¹²³I]FP-CIT-SPECT in the differential diagnosis of DLB and non-Lewy body dementias has been extensively elucidated and is appreciated in the clinical setting [111].

Class I evidence have been provided regarding the application of [¹²³I]FP-CIT-SPECT in discriminating DLB patients [112]. However, results should be replicated with patients recruited from different clinical settings. Reduced uptake yields a respectable diagnostic accuracy in discriminating DLB from AD. Alas, regarding differential diagnosis with atypical parkinsonian syndromes and frontotemporal dementia, the utility of [¹²³I]FP-CIT-SPECT is limited [86, 113, 114].

There are scarce studies evaluating [¹²³I]FP-CIT-SPECT alongside post-mortem tissue in DLB (Table 4). Among all cohorts, [¹²³I]FP-CIT-SPECT exhibits higher sensitivity and specificity when compared to clinical diagnosis. Vascular lesions in the substantia nigra have been reported as a cause of false positive results [115]. Although positive scans have been reported in PSP, FTLD and CBD, diagnosis can typically be made on distinct clinical characteristics. However, Thomas et al. have reported two false positive cases with features of parkinsonism and a clinical diagnosis of DLB; post-mortem diagnosis revealed either AD or FTLD features without evidence of SNCA pathology in the substantia nigra [112]. The authors also identified six cases of false negative scans; three of the cases had a clinical diagnosis of AD at baseline without any signs of parkinsonism. At post-mortem examination, they harbored a mixed picture of AD and DLB features. The other three cases were retrospectively reassessed and actually fulfilled criteria for probable DLB. Hence, although [¹²³I]FP-CIT-SPECT harbors a suitable accuracy, absence of an abnormal scan cannot fully exclude the presence of DLB. This discrepancy could be explained either by the fact that [¹²³I]FP-CIT-SPECT measures the effect of SNCA in neurons and not the deposition of SNCA per se; thus cortical and striatal pathology might be evident without substantial nigrostriatal neuronal degeneration.

A study by Albin et al., combining amyloid and dopamine terminal PET imaging, revealed that imaging classifications were concordant with neuropathological diagnostic classifications in 33/36 cases (91.7%) [116]. Of three cases with discordant imaging-pathological classification, one had a clinical and imaging diagnosis of DLB, but a pathological diagnosis of AD. However, alpha-synuclein immunoreactive Lewy body inclusions were present in the midbrain, thus suggestive of mixed AD-DLB pathology. The other discordant subject was classified as DLB via imaging, but within the frontal cortex and hippocampus, had transactive response DNA binding protein 43 kDa (TDP-43)-immunoreactive neurites. This was particularly unusual given that this case exhibited unilateral striatal loss of [¹¹C]DTBZ, a marker of striatal dopamine terminal integrity. Although 8.3% of cases differed in diagnostic classifications based on neuroimaging and histopathology, this was an improvement compared to their previous studies, which demonstrated that ~ 35% of cases had discordant expert clinical consensus and imaging classifications [117, 118]. Therefore, this combined imaging approach may be useful in establishing more accurate markers for differentiating dementias.

Patterns of cerebral glucose metabolism in DLB have been reported to encompass the ability to differentiate DLB from other forms of dementia. Although DLB has shown to exhibit widespread glucose hypometabolism across cortical regions, metabolic reduction has been shown to be most prominent within the visual association cortex. Through looking at metabolism within this region, DLB can be distinguished from AD with a sensitivity of 86% and specificity of 91% [119]. Although these authors only scanned one patient with autopsy-confirmed DLB diagnosis, postmortem results from 17 DLB brains revealed a distinct and extensive white matter spongiform change with coexisting gliosis throughout cerebral white matter. These changes were consistently and pronouncedly observed within the occipital lobe, with the severity of the regional spongiform change mainly corresponding to the regional differences in patterns of reduced glucose metabolism illustrated by living AD and DLB patients [119]. This study is in corroboration with findings reported by Minoshima et al., who revealed that autopsy-confirmed AD and DLB patients exhibited regional metabolic reductions, specifically within posterior cingulate, parietotemporal association, and frontal association cortex. DLB cases, in particular, demonstrated significant metabolic reductions within the occipital cortex, specifically within the primary visual cortex, which had the ability to distinguish DLB from AD with a specificity of 80% and sensitivity of 90% [120]. Furthermore, patients who were initially clinically diagnosed with probable AD, but later fulfilled the clinical criteria for DLB, demonstrated hypometabolism within the primary visual cortex at higher frequencies, which often preceded the manifestation of several DLB symptoms [120]. Although the authors of these studies argue that [¹⁸F]FDG PET may be a useful tool to distinguish DLB from other dementias, Albin et al. demonstrated that ~ 30% of classifications, based on glucose metabolism, differed from final neuropathological diagnoses in a cohort of DLB, AD and FTD who underwent PET imaging and subsequent autopsy [116]. They reported 2 cases where [¹⁸F]FDG PET classification was AD but pathological verification was DLB, though combined amyloid and dopaminergic terminal PET imaging correctly identified the pathological diagnosis [116]. Therefore, the authors argued that classifications based on [¹⁸F]FDG PET are less precise, with misclassifications ascribed to [¹⁸F]FDG PET being due to the absence of occipital metabolic deficits in a substantial proportion of DLB patients [121].

A proposed [¹⁸F]FDG PET imaging feature of DLB is the cingulate island sign, which refers to the sparing of the posterior cingulate relative to the precuneus and cuneus. This sign is said to be useful for an accurate diagnosis of DLB, given that it is specific, with a reasonable sensitivity [66]. Studies assessing this sign in clinically diagnosed DLB have revealed that the cingulate island sign metabolism, as measured by [¹⁸F]FDG PET, is highly specific for detecting DLB, with a specificity of 100% and sensitivity ranging from 62 to 86% [66]. In this study, 4/14 subjects had autopsy-confirmed DLB, with the others being followed clinically for several years and their diagnosis remaining unchanged. Similarly, the cingulate island sign metabolism is higher in DLB patients compared to AD, independent of amyloid load [72]. Patients who exhibited the cingulate island sign were more likely to be classified as having high or intermediate probability of DLB pathology, receiving a clinical diagnosis of DLB. Furthermore, a higher cingulate island sign ratio was associated with a lower burden of neurofibrillary tangles. 2 subjects who had the lowest cingulate island sign ratio where clinically diagnosed with DLB, but at autopsy, exhibited high likelihood of AD pathology without Lewy body pathology. Taken together, these results indicate that a reduction in cingulate island sign ratio is associated with high burden of AD-type neurofibrillary tangles, therefore ‘pure’ DLB would present with the typical cingulate island sign. This is incredibly important, as the convergence and co-occurrence of AD and DLB pathology is common, with ‘pure’ DLB accounting for no more than a third of all DLB cases and possibly 10% of all clinical dementia cases. This was demonstrated by Barker et al., who identified DLB in 14–26% of dementia cases, with the ‘pure’ form of DLB accounting for 0–19% of dementia patients [122]. Therefore, whilst helpful, the cingulate island sign will not very sensitive or specific for DLB in other pathological series.