Prevalence, distribution, and severity of cerebral amyloid angiopathy differ between Lewy body diseases and Alzheimer’s disease

Neurodegenerative diseases of the ageing brain are defined neuropathologically by their most prevalent pathology, using internationally recognised staging criteria. However, it is rare that pathologies exist in isolation, with pure pathology only seen in 22.7% of post-mortem cases in a large neuropathological study consisting of 670 brains [40]. Although low/intermediate or indeed severe/high levels of concomitant pathology are present across numerous diseases, they are particularly evident across Lewy body diseases (LBDs) including dementia with Lewy bodies, Parkinson’s disease (PD), and Parkinson’s disease dementia (PDD), where in addition to hallmark inclusions of α-synuclein, concomitant AD related pathology (intracellular tau-immunoreactive neurofibrillary tangles and extracellular amyloid β (Aβ) plaques) is a prominent feature [17, 35, 39, 61, 62].

Cerebral amyloid angiopathy (CAA), observed in 20–100% of AD cases [20, 32, 38, 48], is defined by the deposition of Aβ (predominantly Aβ_1-40) in the walls of meningeal vessels, cerebral arteries, arterioles, and less commonly in the capillaries and vein vessel walls [28, 49, 67]. It exists in two forms, the first (type 1 CAA) affects the capillaries, with or with-out involvement of cortical or leptomeningeal vessels, the second (type 2 CAA) includes Aβ deposition restricted to leptomeningeal and cortical arteries, arterioles, and rarely veins without capillary involvement [54].

The association of CAA with clinical dementia has been investigated, 87.5% of individuals with dementia have CAA, whilst only 55.6% of cognitively normally individuals exhibit CAA at post-mortem examination [4]. CAA has also been identified as a contributor to cognitive decline independent of other AD related neuropathology [7, 56]. Furthermore, type 1 CAA was found to mildly correlate with clinical dementia and AD neuropathology, whereas type 2 CAA did not [4].

Although a frequent observation in AD at post-mortem examination, CAA has also been reported in LBDs with recent studies reporting between 91 and 100% of DLB cases, 50–63% of PDD cases, and 13% of PD cases displaying CAA pathology [25, 29]. The increased presence of type 1 CAA has also been reported in LBD cases with dementia as demonstrated by a series of 88 cases including PD, PDD and DLB cases. Whilst only 25% of PD cases exhibited type 1 CAA, DLB and PDD cases exhibited higher rates of type 1 CAA deposition (90% and 85% of cases respectively) [30].

Studies suggest that neurodegenerative pathologies spread in a predictable stereotypical manner i.e. tau pathology originates in the entorhinal cortex and progresses to the neocortex [1, 8] and Aβ starts in the neocortex and advances through the limbic system and brainstem to the cerebellum [55]. α-synuclein originates in the brainstem, progresses through the limbic regions and to the neocortex in PD/PDD [9, 42], with a limbic -predominant profile being more associated with cognitive decline in DLB [58]. However, in the presence of multiple pathologies typical topographical patterns of distribution may be altered. Lewy related pathology in DLB cases with significant concomitant AD related pathology shows a different distribution to DLB cases with minimal AD related pathology, and both groups differ from PD cases [58, 61]. With respect to CAA in AD, the occipital cortex is the most commonly and severely affected brain region in AD, with frontal, parietal, and temporal lobes less affected [3, 59, 64]. However, distribution patterns of CAA have yet to be investigated in LBDs, and as DLB cases can exhibit a greater Aβ burden compared to PDD and PD cases [18, 24, 31] whether LBDs have a similar distribution of CAA is currently unknown.

Therefore, the aim of this study was to investigate differences in type, topographical distribution of CAA, and association with dementia, across the spectrum of LBDs and determine if this differs from AD cases.

Mixed pathologies, in particular AD related pathology is a frequent finding in LBDs, and data from this study adds to growing evidence to suggest CAA may be a contributing pathological substrate in LBDs, in particular in DLB. Building on previous studies that have demonstrated differences in the prevalence of CAA in DLB compared to PDD [25, 29, 30], here we report differences in the severity and the topographical distribution of CAA across the LBD spectrum.

In our study we found CAA is more common in cases with clinical dementia, as significantly more AD, DLB, and PDD cases had CAA, and there was no difference in the number of PD cases with and without CAA. This is in agreement with other studies, as evidence from a longitudinal study using data from 1100 well characterised older adults suggests that CAA is associated with faster rates of decline in global cognition, perceptual speed, and episodic and semantic memory over a 19 year period [7]. Furthermore, both large population and community-based studies have reported individuals with moderate to severe CAA perform worse on cognitive tasks [2, 10]. The disease groups with the highest CAA scores (Olichney grade 4) in our study are AD and DLB, with none of the PD or PDD groups (which originally start out as motor disorders, and potentially progressing to dementia) exhibiting the highest CAA score available. When considering the LBD cases only this finding is in alignment with a study by Jellinger where only DLB cases exhibited the highest CAA score, with neither PD or PDD cases displaying the highest severity level [29]. In the current study the cumulative cortical CAA scores were significantly higher in AD compared to DLB and PD, and there was no difference observed between AD and PDD. However, in the PDD group there was one case that exhibited severe CAA in all cortical regions which could be driving this. Type 1 CAA is more common in DLB cases compared to PDD, which is interesting given the association between type 1 CAA and cognitive decline [7, 56], and not present at all in our PD cohort. The relationship between type 1 CAA, AD related pathology and dementia has been reported in a population-based study using the Vantaa 85 + cohort, where the authors found type 1 CAA was associated with severity of generalised CAA and the presence of dementia [37]. Furthermore, in agreement with a previous neuropathological study DLB and PDD cases are more likely to have type 1 CAA compared to PD without dementia [30]. Although it is difficult to neuropathologically distinguish DLB from PDD, DLB cases have been shown to exhibit increased AD related pathology compared to PDD cases [31, 62]. Therefore to investigate if the differences in CAA severity between DLB and PDD are evident in the absence of abundant amyloid plaques in the parenchyma, this study was designed to exclude all LBD cases with high AD neuropathologic change. There were no differences in Thal phase, Braak neurofibrillary tangle stage or CERAD score between DLB and PDD groups, which is also in agreement with a study by Hansen and colleagues. They found no difference in Thal phases between DLB and PDD groups, however CAA in DLB was more severe compared to PDD [25]. The relationship between parenchymal amyloid load and CAA severity has been investigated in AD [57]. No association was observed between amyloid plaque load and CAA severity in individual brain regions, and it is speculated that differing pathogenic mechanisms underpin the deposition of amyloid in plaques and blood vessels. This is not surprising considering the predominant amyloid species in plaques are mostly peptides that are 42 amino acids in length whilst amyloid deposits in CAA mainly comprise of shorter peptides of 40 amino acids.

In the current study the topographical pattern of distribution of generalised CAA in DLB is similar to AD, and this differs from the distribution pattern observed in both PD and PDD. The most commonly affected brain region across all diseases was the occipital cortex. We did not look at differences in specific Brodmann areas in the occipital cortex i.e. Brodmann areas 17 and 18, as unlike tau pathology in Braak neurofibrillary tangle staging, where this delineates Braak stages V/VI, there is no protocol for this when assessing amyloid pathology. In AD and DLB the next most commonly affected brain region is frontal cortex whilst this is the least affected brain region in PD and PDD cases. Interestingly other reports have described concomitant pathology altering the topographical distribution of pathological protein aggregates in patients with multiple pathologies. Toledo and colleagues describe different clusters patterns of pathology in patients with clinical dementia and AD and Lewy related pathology compared to PD patients without AD related pathology [58], whilst a study from our group demonstrated the spread of hyperphosphorylated tau pathology in cases with mixed AD/DLB differs to that in ‘pure ADs [61]. It has been suggested that multiple pathologies promote suitable conditions for a synergistic relationship between proteins that results in cross seeding and exacerbation of overall pathology burden and accelerating cognitive decline [6, 13, 14, 21, 27]. To our knowledge this is the first study demonstrating divergence of pathology patterns in cerebral amyloid deposition in LBDs, and could, in part, be a product of increased synergistic relationships between cortical pathologies in DLB. From a clinical perspective the finding of significant involvement of CAA in the frontal cortex of patients with AD and DLB (predominantly cognitive neurodegenerative diseases) compared to patients with PD or PD (primarily diagnosed as motor disorders) is interesting. Two previous studies have suggested the presence of CAA is associated with impairments to executive function which is controlled by the frontal cortex [12, 63]. This raises questions regarding the potential contribution of CAA to specific cognitive domains in neurodegenerative diseases.

One of the strongest genetic risk factors for increased CAA scores in AD is APOE ε4, [15, 26, 52, 66]. This has also been studied in LBD cases, as APOE ε4/4 and ε2/4 genotypes exhibit the highest general CAA scores [25]. In the current study LBD cases with ε4/4 and ε2/4 genotypes did all exhibit CAA, although this was only a small number of cases (n = 3). Although it is worth noting that we did not include LBD cases with high AD neuropathologic change (that would be neuropathologically classified as mixed DLB/AD) to avoid a masking affect from abundant amyloid β pathology. Previous studies have discussed the masking effect of abundant amyloid pathology on an association between APOE ε2 and CAA severity [66], and it has been suggested that whilst APOE ε4 promotes vascular amyloid deposition, ε2 promotes progression to severe CAA and associated vasculopathic changes [22]. Interestingly APOE ε2 has been clearly associated with CAA related haemorrhage as the frequency of the ε2 allele is high regardless of whether significant AD related pathology was present. Aditionally in the group where significant AD pathology was present the APOE ε2 frequency is 4 times higher than in patients with AD without haemorrhage [46]. The mechanisms behind the influence of APOE ε2 on increased risk of cerebral haemorrhage are still yet to be elucidated, however it has been suggested that fibrinoid necrosis caused the breakage of amyloid laden vessels though its association with APOE ε2 [41].

With regards to vascular pathology, it is well known that CAA is associated with ischaemic stroke, cerebral infarction (particularly microinfarcts) in addition to haemorrhages [11, 47, 50, 60]. With increasing severity of CAA, smooth muscle and elastic elements in the vessel walls are replaced by Aβ depositions which results in fragile vessles and subsequent brain bleeds. Another consequence of Aβ in vessel walls is impaired vasoreactivity, which can lead to vessel narrowing/occlusion and hypoperfusion which may lead to ischaemic lesions in the parenchyma. When investigating the effects of severe CAA on vascular insults in the current cohort, of the 49 AD cases with severe CAA,13 were found to have infacts and microhaemorrhages, with no significant haemorrhages seen in any of the case. Out of the 4 DLB cases that exhibited CAA there were no reported vascular lesions in the neuropathological reports.

A neuroimaging study conducted by Gungor and colleagues demonstrated cerebral microbleeds (CMBs) were predominant in the occipital and frontal regions in DLB cases [23], which is in line with the finding that occipital and frontal lobes are the most frequently affected by CAA in DLB cases in the current study. Other groups have investigated the topography of CMBs across the Lewy body disease spectrum. Yamashgiro and colleagues found deep or infratentorial microbleeds were more common than lobar microbleeds (65.5% vs 34.5% respectively) in PD [65]. Although Kim and colleagues found no differences in frequency of deep or infratentorial microbleeds in DLB compared to PDD, they did demonstrate lobar microbleeds were found more frequently in DLB compared to PDD [33]. Also the occipital lobe is the brain region most commonly affected by microbleeds in DLB [53]. Interestingly, a study comparing CMBs between AD and DLB patients found there was no significant difference in the frequency of CMBs between AD and DLB, and the presence of microbleeds in DLB was not associated with amyloid deposition [16]. This suggests other mechanisms may underly the presence of microbleeds outside of general Aβ deposition, potentially the propensity of APOE ε2 carriers to exhibit more CAA vasculopathic changes in DLB. Taken together the results are inkeeping with the hypothesis that CAA is a common finding in DLB and may contribute to other pathological lesions and the clinical phenotype observed in these cases.

When investigating the APOE status in different subtypes of CAA we found APOE ε4 carriers were more likely to have type 1 CAA in the overall cohort. This is not surprising as Thal and colleagues found the frequency of APOE ε4 carriers in type 1 CAA is 4 times higher than in type 2 CAA in AD and controls, and type 2 CAA has a higher APOE ε2 frequency compared to type 1 CAA [54]. Another study investigating the masking effect APOE ε2ss protective association with comorbid AD related pathology also ran path analysis for the presence of type 1 CAA and found no significant associations between type 1 CAA APOE ε2. However, this study excluded all participants with non-AD dementia, therefore the effects of APOE ε2 in DLB warrants further investigation.

A caveat of this study is that all of the tissue used was sampled from the right hemisphere, due to the routine protocols carried out in the Newcastle Brain Tissue Resource. Several studies have observed hemispheric asymmetry in neurodegenerative pathologies that are associated with dementia [19, 34, 51]. In terms of Aβ neuroimaging Frings and colleagues demonstrate PiB retention on average was slightly higher in the right hemisphere compared to the left [19]. Whilst neuropathological studies by King and colleagues show mild asymmetry in 3/20 AD cases (with left and right hemisphere affected differently in different cases) [34]. Stefanits and colleagues demonstrate mild vulnerability of the righ hemisphere in 5/20 AD cases and 3/15 AD/DLB cases suggesting the proportion of asymmetry between cases that have AD or AD and DLB related pathologies was similar and we assume this would be similar in our cohort. We are not aware of studies that specifically investigate asymmetry of CAA, however this would be an interesting line of research.

Data from the current study supports growing evidence that CAA may play an important role in the clinico-pathological phenotype of Lewy body diseases, particularly DLB. We have shown DLB cases can have more severe CAA, have an increased presence of type 1 CAA, and have a different topographical distribution of CAA, which is similar to AD, compared to PDD and PD. In conclusion data from this study suggests DLB is more aligned with AD than PDD/PD with regards to CAA severity and topographical distribution, therefore should be considered when stratifying cases for clinical trials and the design of future disease modifying therapies.