Comorbid cerebral amyloid angiopathy in dementia and prodromal stages—Prevalence and effects on cognition

To determine the contribution of cerebral amyloid angiopathy to cognitive impairment in MCI and dementia.


| INTRODUCTION
CAA, a highly prevalent cerebrovascular disease in older age, is characterized by β-amyloid deposit in the wall of cerebral vessels. 1,2 the elderly population CAA is a common cause of intracerebral bleeding and cognitive dysfunction. 3In accordance with AD, incidence of CAA increases with age. 2 In patients with dementia up to 60% suffer from comorbid CAA and in AD histopathological evidence of CAA is found in more than 80% of patients. 4,5Previous studies have shown variable CAA prevalence rates due to methodical reasons.Therefore, one aim of the study was to investigate the prevalence of CAA-related magnetic resonance imaging (MRI) biomarkers across different diagnostic groups in a memory clinic sample.Despite a distinct pathophysiology CAA and AD show strong associations and complex interactions.With a histopathological prevalence of almost 80% in AD, CAA is a common comorbidity which is also related to AD pathology and severity of clinical dementia syndrome. 6,7CAA-related small vessel disease burden was shown to be an independent predictor for dementia conversion. 8dependently from non-CAA-related cerebral small vessel disease, CAA in AD seems to contribute to an accelerated global cognitive decline and deficits in perceptual speed, episodic and semantic memory. 6After controlling for parenchymal amyloid and general vascular pathology on a histopathological level moderate-to-severe CAA seems to be associated with lower perceptual speed and episodic memory, but not with deficits in semantic memory, working memory nor visuospatial skills. 7Other investigations show independent effects of CAA on memory, executive functioning and perceptual speed, whereas about half of these cognitive CAA effects seem to be mediated by white matter disruption. 9However, study data investigating the influence of CAA on the clinical phenotype and the rate of cognitive decline in AD are inconsistent 10 so that the contribution of CAA to cognitive decline in AD is incompletely understood.
3][14] But beside the effects on cognition mediated by vasculopathic lesions and accelerated AD pathology it is hypothesized that comorbid CAA in AD harbors autonomous effects on cognition which might be related to the predominating CAA subtype. 1,3,11,13,15,16Especially, CAA-specific allocortical microinfarcts in the capillary subtype of CAA were shown to be associated with cognitive decline. 16Thus, CAA-mediated cognitive deficits in AD seem to be partly independent from AD pathology 14,[17][18][19] suggesting a rather multifactorial etiology of cognitive decline in AD with comorbid CAA.Early memory deficits mediated by CAA may be linked to early hippocampal pathology in the capillary CAA subtype 1,16 whereas the expansion of CAA pathology to subcortical structures and consecutive vascular pathology may subsequently result in perceptual slowing and executive dysfunction in the later course of the disease. 1 We therefore hypothesize that comorbid CAA in dementia and prodromal stages modulates cognitive impairment in a stagedependent manner and independently of non-CAA-related cerebral angiopathy.Therefore this study investigates differences in neuropsychological test performance in subjects with SMI, 20 naMCI and aMCI, 21,22 AD, 23 MD and VD 24 with or without MRI biomarkers of CAA and aims to determine the impact of comorbid CAA on cognitive performance within these diagnostic categories and cognitive subscales controlling for age and non-CAA-related cerebral small vessel disease We further hypothesize that, due to the complex interplay of CAA and AD on cognitive impairment, psychometric group differences between MCI, AD and MD are mediated by CAA-associated small vessel disease burden.Therefore, we investigate whether CAA-related small vessel disease burden measured by Total-MRI score 25 modulates differences in cognitive test performance between diagnostic groups.

| Participants
In this retrospective data analysis we analyzed patients from our memory clinic who underwent first diagnostic approach for suspected cognitive decline including neuropsychological testing and cranial MRI including a standard protocol. 26We included all patients who underwent neurocognitive evaluation between February 2016 and December 2020 with the diagnosis of SMI, 20 naMCI, aMCI, 21,22 AD, 23 MD and VD. 24

| MRI re-evaluation, CAA diagnosis & scoring
Magnetic resonance imaging data sets were analyzed by an experienced neuroradiologist regarding the presence and extent of the following MRI biomarkers of CAA: lobar intracerebral macrohemorrhages (ICH), cortical/subcortical cerebral microbleeds (CMB), cortical superficial siderosis (cSS), cortical microinfarcts, severe (>20) perivascular spaces in the centrum semiovale, and white matter hyperintensities (WMH) of presumed vascular origin.In addition, prevalence of deep ICH and deep CMB was also noted.
Lobar ICH, deep ICH and cerebral microinfarcts were categorized as absent (0) or present (1 or more).CMBs were categorized according to their numbers (0, 1, 2-4, 5-10.10-50,>50), cSS was categorized as absent, focal or disseminated.To assess the severity of WMH Fazekas score was used. 279][30] With regard to the presence of deep CMB, a maximum of two deep CMB was allowed if the respective patients had at least twice the number of lobar CMB and/or a history of lobar ICH or cSS.To capture total brain burden of small vessel disease in CAA Total-MRI-score was applied. 25

| Neuropsychological testing
In the psychometric tests, we investigated intelligence (Horn's LPS reasoning test (the German Leistungs-Prüf-System; LPS), 31 Multiple Choice Word Test-B (MWT-B)), 32  In the subtests of the WMS and in the repeating numbers subtest of the NAI we used percentile ranks.Within the color word and the labyrinth subtest of the NAI C-scores were obtained.

| Statistical analysis
Demographic and clinical characteristics of the participants are reported as means and SDs for continuous variables and percentages for categorical variables.Depending on data level analysis, we conducted t-tests, chi-square tests or univariate analysis of variance (ANOVA) for the identification of covariates.In multiple comparisons, p values were adjusted according to Bonferroni.To investigate the stage-dependent and independent contribution of CAA to cognitive impairment within cognitive subscales and diagnostic groups we used an ANOVA model (ANOVA) categorizing diagnosis of CAA (none vs. possible and probable) as an independent variable and the cognitive measure in each cognitive subtest and diagnostic entity as a dependent variable.Age and Fazekas score were used as covariates.To examine whether differences in cognitive test performance between the diagnostic groups are mediated by CAA-associated small vessel disease burden we used an ANOVA model (ANOVA) using cognitive measure as an independent variable, diagnostic entity as a factor and Total-MRI score, Fazekas score and age as covariates.Statistical computations were conducted using IBM SPSS statistics for Windows, version 25.

| RESULTS
475 patients (222 male, 253 female) were included in this retrospective analysis.47 patients were diagnosed with SMI, 41 with naMCI, 189 with aMCI, 9 with early onset AD, 114 with late onset AD, 71 with mixed dementia and 4 with VD.Mean age was 73.2 � 9.9 years at time of presentation and diagnostic procedures in our memory clinic (Tables 1 and 2).
116 patients fulfilled the modified Boston diagnostic criteria for CAA (43 patients with possible and 73 patients with probable CAA) (Table 1).Patients with possible or probable CAA tended to be older than patients without CAA (Table 1).CAA prevalence in the different diagnostic groups was 14.9% in SMI, 14.6% in naMCI, 24.3% in aMCI, 22.2% in early onset AD, 18.4% in late onset AD, 46.5% in MD and 25% in VD.Especially patients with aMCI, early onset AD, MD and VD exhibited high CAA prevalences (Table 3).4).
Differences in cognitive test performance between the different diagnostic groups of aMCI, naMCI, AD and MD were mediated by total CAA-related small vessel disease burden regarding speech and word fluency.Total-MRI score mediated psychometric group differences in the AAT simply nouns subtest (F (2.39) = 4.059; p = 0.025; partial etasquare = 0.172) and the CERAD verbal fluency s-words subtest (F (3.129) = 3.544; p = 0.017; partial eta-square = 0.076) (Table 5).
According to population-based studies CAA prevalence is consistently higher in patients with dementia compared to non-demented individuals (55%-59% vs. 28%-38%). 4Especially in AD up to 80% of patients show histopathologic evidence of CAA. 5 Current metaanalytic data analyzing the prevalence of CAA in AD demonstrate a histopathologically-based prevalence of 48%, 34 but estimates of CAA prevalence vary widely.In this respect, Jäkel et al. referred to an enormous heterogeneity among neuropathology and MRI protocols in CAA recognition. 34Since CAA is linked to side-effects in upcoming immunotherapies in AD this aspect calls for standardized assessment and reporting of CAA. 34king the lower diagnostic sensitivity of cerebral imaging compared to histopathological approach into account, CAA prevalence estimates in our cohort seem to be widely in line with current data.But regarding CAA prevalence estimates of 2%-3% between the ages of 65 and 74 years 2 the relatively high CAA prevalence in our subjective memory impairment (SMI) group (mean age 69.4 (10.0) years; prevalence possible þ probable CAA: 14.9%) is suggestive of a high rate of underlying neurodenerative SMI etiology in our memory clinic cohort. 35,36The relatively high CAA prevalence in individuals with aMCI (mean age 70.4 (10.4) of 24.3% and early onset AD (mean age 59.2 (6.0) years) of 22.2% (0% possible, 22.2% probable CAA) may be driven by the close relations between capillary cerebral amyloid angiopathy, the APOE epsilon 4 allele and early onset of AD. [37][38][39] Since vascular pathology with white matter disruption is known to account for about half of CAA-specific effects on cognition the high prevalence rates of CAA in MD and VD in our cohort are not surprising and seem to be attributable to overlapping pathomecanism. 9garding the objectified influence of comorbid CAA on cognitive test results in patients with aMCI our findings with better short and working memory test performance in aMCI patients with CAA compared to aMCI patients without CAA are rather counterintuitive with respect to current knowledge on CAA distribution in the brain. 1 Like AD, CAA involves different brain areas following a hierarchical sequence starting with a capillary pathology in leptomeningeal or parenchymal vessels or neocortical areas spreading to allocortical and, thus, hippocampal areas in a second step before involving basal ganglia, thalamus and brainstem. 1Against this background, early disadvantageous effects of comorbid CAA on short term memory would have been expected.This surprising finding might be partly attributable to the heterogenous etiology of aMCI which might also be due to medication side effects, depression or other somatic factors instead of beginning neurodegenerative disease. 40It might be possible that the aMCI cases within the no-CAA group were more frequently of neurodegenerative origin resulting in more prominent short and working memory deficits due to AD pathology overreaching and probably masking CAA-related cognitive effects.
T A B L E 3 CAA prevalence within diagnostic groups.Furthermore, the relatively high percentage of possible CAA cases in the aMCI group (12.2% possible vs. 12.2% probable CAA) and corresponding diagnostic sensitivity and specificity issues should be taken into consideration when discussing this implausible finding.

Diagnosis
Beyond, it should be discussed that controlling for general cerebral microangiopathy might also weaken CAA-associated effects on cognition mediated by white matter pathology.But according to our data set CAA alone does not seem to have detrimental effects on memory in aMCI patients.
Investigating the effects of comorbid CAA on cognitive impairment in patients with AD the strong association of CAA with parenchymal amyloid pathology, severity and accelerated progression of clinical dementia syndrome has to be considered. 7,33,41Since CAA significantly interacts with AD pathology in terms of Braak stages 41 a parallelism in the evolution of AD and CAA was expected so that poorer memory functioning in AD patients with CAA is in line with our expectations and current data in this field.Capillary CAA pathology starts early in AD and develops simultaneously with progressing parenchymal amyloid pathology 1 which results in a CAA-mediated contribution to neurodegeneration with a global cortical thickness reduction in patients with CAA regionally overlapping with an AD cortical atrophy signature. 14Beyond, new data show diminished functional brain connectivity in patients with CAA with an impairment of visual, executive control and bilateral frontoparietal networks. 18As a consequence of these far-reaching functional effects, CAA per se is known to be associated with progressive cognitive deficits in multiple domains. 19Even in non-demented individuals CAA was shown to mediate a cognitive decline in episodic memory, executive function, semantic fluency and attention 19 whereas about half the effect of CAA on cognition is mediated by vascular pathology and white matter disruption. 9Taken together, of comorbid CAA on cognitive performance in AD can be interpreted as an interplay of advanced parenchymal, CAA-related-and non-CAArelated vascular pathology in multiple areas of the brain.Thus, our measured effects of comorbid CAA on memory in patients with AD might be best understood as a cumulative effect of CAArelated and AD-associated hippocampal pathology. 1The fact that CAA-specific effects on cognition were not demonstrated in other cognitive domains and that no significant effects on memory functioning were measured in the MD group might be due to methodical aspects in terms of controlling for Fazekas score which seems to underpin the functional relevance of white matter disruption in CAA-mediated cognitive effects in AD.
Our data further demonstrate that CAA-related small vessel disease burden contributes to psychometric group differences between naMCI, aMCI, AD and MD within the cognitive domains of word finding and verbal fluency.But in contrast to our categorical and cross-sectional approach when measuring effects of CAA on cognitive performance within diagnostic groups Total-MRI score did not contribute to psychometric group differences with regard to memory functioning.Dimensional CAA detection using the Total-MRI score is a valid approach. 25 After controlling for age and Fazekas-score CAA diagnosis influenced cognitive test performance in short term and working memory within the aMCI group.There was a significant effect of CAA diagnosis on the WMS digit span forward (F (1.93) = 9.334; p = 0.003; partial eta-square = 0.091) and backward (F (1.93) = 6.303;HAUSSMANN ET AL. p = 0.014; partial eta-square = 0.063) subtests with better test results in the CAA group.In addition, we found a trend in the NAI color word test (F (1.159) = 3.394; p = 0.067; partial eta-square = 0.021) within aMCI group.In AD with comorbid CAA we detected poorer memory functioning with significant effects of CAA on test performance in CERAD word list recall (F (1.78) = 4.505; p = 0.037; partial eta-square = 0.055).In the naMCI group there was a trend within memory function in CERAD word list recall (F (1.34) = 3.188; p = 0.083; partial eta-square 0.086).In MD we found a trend regarding CAA-mediated effects on speech within CERAD Boston naming test (F 1.44) = 3.389; p = 0.072; partial eta-square = 0.072) (Table

-5 of 11 T A B L E 4
Analysis of cognitive performance within different diagnostic groups and cognitive measures divided into no CAA versus possible and probable CAA.
But our study results imply that it T A B L E 4 (Continued) is important to use different approaches when investigating the complex interactions between AD and CAA on cognitive decline.Methodical aspects have to be closely regarded when interpreting effects of CAA on cognition in AD.Considering our study results with different approaches comorbid CAA in MCI and dementia modulates cognitive performance in memory functioning, word finding and word fluency in a presumably stage-dependant manner.Within the AD group prominent CAA-related effects on memory were objectified whereas effects on speech and word fluency were demonstrated across the diagnostic groups of naMCI, aMCI, AD and MD.Our study has relevant limitations.Beside the retrospective and cross-sectional design, not obtaining biomarkers like CSF, amyloid PET or APOE e4 genotyping, especially in the heterogenous constructs of T A B L E 5 Analysis of interaction between diagnosis and Total-MRI score.
Descriptive statistics including age, sex, diagnosis and CAA prevalence.
T A B L E 2 Age distribution within diagnostic groups in means (SD).Total (n = 475) CAA Diagnosis No (n = 359) Possible (n = 43) Probable (n = 73) Possible þ probable (n = 116) Our data show high prevalences of possible and probable CAA especially in patients with aMCI, early onset AD, MD and VD.Controlling for age and non-CAA-related cerebral small vessel disease the conducted analyses suggest that comorbid CAA in AD and its prodromal stages significantly impacts cognitive test performance in the course of AD in a stage-dependent manner.In patients with aMCI short term and working memory were conterintuitively influenced by comorbid CAA in terms of better test performances in patients with CAA compared to the no CAA group.Beyond, there was a trend regarding the influence of CAA on executive functioning and perceptual speed in aMCI patients.In patients with AD comorbid CAA mediated poorer memory functioning.Further trends were demonstrated regarding the impact of comorbid CAA on memory function in naMCI patients and speech in patients with MD.Total burden of CAA-related small vessel disease contributed to psychometric differences between diagnostic groups with regard to speech and verbal fluency.Our results indicate that accompanying CAA in AD and its prodromal stages is a frequent phenomenon, which requires consideration within general medical treatment of elderly patients with dementia and cognitive decline.Furthermore, management of CAA in patients with cognitive impairment, for example, in terms of optimal blood pressure control, might also be of functional relevance since cognitive performance is additionally influenced by CAA.
22te: ANOVA; independent variable: cognitive measure; factor: diagnostic group; covariates: Total-MRI score, Fazekas-score, age.Bold letters within the table mark significant results (p < 0.05) or trends (p < 0.1).SMI42and MCI,22represent relevant constraints impeding the interpretability and generalizability of our study data.But beside the real world character of our data set, the relatively large number of patients as well as the extensive neuropsychological and neuroradiologic assessment strengthen the validity of our study results.
HAUSSMANN ET AL.