Comorbidity of Cerebrovascular and Alzheimer’s Disease in Aging

Abstract

Background:

Cerebrovascular disease often coexists with Alzheimer’s disease (AD). While both diseases share common risk factors, their interrelationship remains unclear. Increasing the understanding of how cerebrovascular changes interact with AD is essential to develop therapeutic strategies and refine biomarkers for early diagnosis.

Objective:

We investigate the prevalence and risk factors for the comorbidity of amyloid-β (Aβ) and cerebrovascular disease in the Australian Imaging, Biomarkers and Lifestyle Study of Ageing, and further examine their cross-sectional association.

Methods:

A total of 598 participants (422 cognitively normal, 89 with mild cognitive impairment, 87 with AD) underwent positron emission tomography and structural magnetic resonance imaging for assessment of Aβ deposition and cerebrovascular disease. Individuals were categorized based on the comorbidity status of Aβ and cerebrovascular disease (V) as Aβ–V–, Aβ–V+, Aβ+V–, or Aβ+V+.

Results:

Advancing age was associated with greater likelihood of cerebrovascular disease, high Aβ load and their comorbidity. Apolipoprotein E ɛ4 carriage was only associated with Aβ positivity. Greater total and regional WMH burden were observed in participants with AD. However, no association were observed between Aβ and WMH measures after stratification by clinical classification, suggesting that the observed association between AD and cerebrovascular disease was driven by the common risk factor of age.

Conclusion:

Our observations demonstrate common comorbid condition of Aβ and cerebrovascular disease in later life. While our study did not demonstrate a convincing cross-sectional association between Aβ and WMH burden, future longitudinal studies are required to further confirm this.

Keywords

Alzheimer’s disease amyloid cerebrovascular disease white matter hyperintensities

INTRODUCTION

Alzheimer’s disease (AD) is the most common form of dementia and is associated with the hallmark pathologies of extracellular amyloid-β (Aβ) plaques and neurofibrillary tangles of hyperphosphorylated tau in the cerebral cortex. Cerebrovascular disease is the most common coexisting brain pathology with AD [1, 2]. Cerebrovascular disease and AD share a number of risk factors including age, midlife hypertension, and diabetes mellitus, and are both associated with cognitive impairment [3]. In recent years, an increasing number of studies have investigated the association between these processes and attempted to clarify how their interaction modifies the risk of cognitive decline and dementia [4 –8]. Some studies suggest that the vascular pathway linking vascular risk factors and cognitive decline is independent from the amyloid pathway. However, no consensus has been reached to date.

On conventional clinical MRI, cerebrovascular disease is usually visualized as white matter hyperintensities (WMH), lacunar infarcts, cortical infarcts, and cerebral microbleeds [9]. Several studies suggest that WMH are associated with cognitive decline, especially within episodic memory and executive function [10 –12], and an increased risk of AD [13 –15]. The association between regional characteristics of WMH and AD have also been investigated. For example, periventricular WMH may be associated with AD pathology, whereas deep WMH is more likely to be associated with small vessel pathology [16, 17]. Silent cerebral infarcts are another common finding in older adults, the number and distribution of which may be associated with the likelihood of cognitive impairment and dementia [18, 19].

Recent studies have focused on the association between WMH burden and Aβ deposition in order to understand the interrelationship between cerebrovascular and AD. Most studies using cross-sectional designs observed no associations between WMH and Aβ burden [6]. However, several recent studies have shown positive correlations between amyloid burden and WMH volumes [20, 21]. These discrepant results might reflect variable study populations and methodology for assessing Aβ and WMH burden. The Australian Imaging, Biomarkers and Lifestyle (AIBL) Study of Ageing assembled a large cohort of older individuals including cognitively normal (CN) individuals as well as participants with mild cognitive impairment (MCI) and AD dementia. This cohort includes a comprehensive database of clinical, cognitive and imaging data and facilitates a systematic examination of the association between WMH and Aβ burden in a large and well characterized group.

The aims of this study are 1) to assess the prevalence of the comorbidity of Aβ and cerebrovascular disease across the older adult lifespan in the AIBL cohort, 2) to identify various risk factors that are associated with this comorbidity, and 3) to examine the cross-sectional association between Aβ deposition and WMH burden. Given the different cognitive, microstructural, and clinical correlates reported for periventricular and deep WMH [22], WMH regional characteristics were also considered in the analysis.

MATERIALS AND METHODS

Participants

Five hundred and ninety-eight participants from the AIBL cohort who aged over 60 years at baseline and had both Aβ-PET and structural MRI were included in the present study. Participants were recruited in two waves: an inception cohort (N = 378) and an enrichment cohort (N = 220). The inclusion criteria, recruitment and diagnostic classification used in the AIBL study have been described elsewhere [23]. Approval for the study was obtained from the institutional ethics committees of Austin Health, St Vincent’s Health, Hollywood Private Hospital and Edith Cowan University. Written informed consent was obtained from all volunteers before participation.

AIBL participants were recruited from volunteers who were aged over 60 years. Participants underwent comprehensive imaging, biomarker, and clinical assessment at 18-month intervals. The clinical status of participants (CN, MCI, or dementia) is determined at each visit by an experienced panel, which is chaired by a consultant psychiatrist. Of the 598 participants, 87 met National Institute of Neurological and Communicative Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association criteria for AD diagnosis (73 probable and 14 possible) [24], 89 had MCI [25] (47 amnestic multi-domain, 27 amnestic single-domain, 8 non-amnestic multi-domain, 5 non-amnestic single-domain and 2 unknown) and 422 were cognitively normal (CN, 265 memory complainers and 157 non memory complainers) (Table 1).

Table 1

Cohort characteristics

	CN	MCI	AD	p
No. of participants	422	89	87
No. of females (%)	239 (56.6)	38 (42.7)	47 (54.0)	0.056¹
Age, y	74.1±6.1	75.2±6.4	76.8±6.8	0.001²
MMSE	28.7±1.3	27.1±2.1	20.2±5.8	<0.001²
CDR [min-max]	0.0 [0.0–0.5]	0.5 [0.0–0.5]	1.0 [0.5–3.0]	<0.001²
PACC	–0.06±0.74	–1.48±0.87	–3.38±0.86	<0.001²
% APOEɛ4+	24.2	43.7	70.9	<0.001¹
Centiloid	19.6±32.9	49.7±46.4	93.5±38.2	<0.001²
% Amyloid positive, Aβ+	31.3	57.3	90.8	<0.001¹
No. with subcortical infarcts (%)	33 (7.8%)	6 (6.7%)	9 (10.3%)	0.652¹
No. with large cortical infarcts (%)	6 (1.4%)	0 (0.0%)	0 (0.0%)	–
WMH/TIV, %	0.54±0.83	0.78±1.14	1.43±1.64	<0.001³
Periventricular WMH/TIV, %	0.32±0.51	0.47±0.70	0.91±1.03	<0.001³
Deep WMH/TIV, %	0.12±0.29	0.19±0.47	0.31±0.57	0.002³
% Cerebrovascular positive, V+	24.9	29.2	59.8	<0.001¹

All the variables are presented as mean±standard deviation, except for CDR which is listed as medium (min-max). ¹Chi-square test for categorical variable, ²Kruskal-Wallis rank sum test, ³analysis of covariance, adjusting for age and MR acquisition site.

Neuropsychological evaluation

The full AIBL neuropsychological test battery was constructed as previously described [23]. In this analysis, we report cognitive function based on the Mini-mental State Examination (MMSE), Clinical Dementia Rating (CDR) as well as a composite score of tests, i.e., the Preclinical Alzheimer Cognitive Composite (PACC). The PACC was calculated as a mean of z score performances on four tests including MMSE, the Digit Symbol Coding, the California Verbal Learning Test delayed, and Logical Memory delayed [26].

Brain imaging

Aβ PET imaging was performed using three different radiotracers: ¹¹C-Pittsburgh compound-B (PiB), ¹⁸F-flutemetamol (FLUTE), and ¹⁸F-florbetapir (FBP). The PET methods for each of the tracers have been previously described [27 –29]. PET data were acquired 40–70 min post injection of ¹¹C-PiB, 90–110 min post injection of ¹⁸F-FLUTE, and 50–70 min post injection of ¹⁸F-FBP.

MRI scans were acquired at three different scanning centers, two in Melbourne using a Siemens 3T Trio (∼67%) and one in Perth using a Siemens 3T Verio (∼33%). A 3D T2-weighted fluid attenuation inversion recovery (FLAIR) sequence was included in the image acquisition protocol, which was acquired using two different sets of parameters: 1) in-plane resolution 0.98×0.98 mm, slice thickness 0.9 mm, repetition time/echo time/inversion time = 6000/420/2100, flip angle 120°, field-of-view 240×256, and 176 slices; 2) in-plane resolution 0.5×0.5 mm with in-plane interpolation (factor of 2) enabled, slice thickness 1.0 mm, repetition time/echo time/inversion time = 5000/355/1800, flip angle 120°, field-of-view 256×256, and 160 slices.

Image analysis

Aβ PET assessment

Aβ burden was automatically quantified from each PET scan using CapAIBL, a PET-only approach [30], allowing quantification of Aβ burden without any bias from MRI features. Briefly, an adaptive atlas was automatically fitted to each image to match its PET retention pattern. Each PET image was spatially normalized to the best fitting atlas in order to calculate a standardized uptake value (SUV) for all brain regions examined. SUV ratios (SUVR) were generated by normalizing the regional SUV using the tracer-specific reference region. Aβ burden was then estimated using the standard SPM centiloid (CL) cortical mask [31], which was the average SUVR of the area-weighted mean of frontal, superior parietal, lateral temporal, lateral occipital and anterior and posterior cingulate regions of the brain. For PiB, the cerebellar cortex was used as the reference region, whereas the pons and the whole cerebellum were used as reference regions for FLUTE and FBP, respectively.

CL values were generated in order to transform the SUVR data from different tracers to a universal scale for subsequent analysis, where 0 and 100 CL represent the typical Aβ burden in young controls and mild AD participants, respectively [31, 32]. The CL values derived from CapAIBL have been shown to strongly agree with the values computed using the standard SPM pipeline with R² > 0.97 across all tracers from the AIBL study [32]. A binary PET-Aβ status for all participants was derived using a CL threshold of ≥20 as Aβ+ and <20 as Aβ- [33].

Cerebrovascular disease assessment

Cerebrovascular disease was assessed on 3D FLAIR images to identify subcortical (lacunar) and large cortical infarcts and to quantify the volume of white matter hyperintensity.

Subcortical infarcts were manually identified by a stroke neurologist and defined as discrete subcortical lesions >3 mm in diameter with intensity that is equivalent to CSF on FLAIR, and an accompanying hyperintense gliotic rim. These infarcts were located in the white matter, deep grey matter nuclei, cerebellum and brain stem [34].

Large cortical infarcts were manually rated on FLAIR by a stroke neurologist. These large infarcts were defined as areas of elevated signal intensity involving the cortical grey matter and immediately subjacent white matter >10 mm in diameter [34].

White matter hyperintensities were automatically segmented using the HyperIntensity Segmentation Tool based on an ensemble of pre-trained neural network classifiers [35, 36]. Periventricular and deep WMH were classified using the distance-based criteria described in Kim, et al. [17]. WMH volumes were quantified in the common Montreal Neurological Institute space and normalized by the total intracranial volumes (TIV), i.e., WMH/TIV (%).

A cut-off for WMH was estimated in the non-demented sample (298 CN and 66 MCI participants, aged over 70) in AIBL cohort that was consistent with previous criteria [8], assuming a prevalence of cerebrovascular disease in such an elderly population to be approximately one-third. The WMH/TIV cut-off was determined as WMH burden at the top 33% of the AIBL sample ordered by the number of identified brain infarcts and WMH/TIV measures, which corresponded to 0.65%. Moreover, this was validated using a statistical approach [8] that demonstrated similar cognitive decline trajectories in participants with a brain infarct and participants with no brain infarct but large WMH burden (WMH/TIV ≥0.65%) (shown in Supplementary Material). Therefore, the cerebrovascular-positive (V+) group was defined as subjects with a brain infarct and/or WMH/TIV ≥0.65%.

Data analysis

All individuals were classified into four categories, i.e., Aβ–V–, Aβ–V+, Aβ+V–, Aβ+V+, based on the presence or absence of Aβ and cerebrovascular disease. Group-wise differences in demographic variables and neuroimaging measures were examined using one-way ANOVA with Tukey’s post hoc test or Kruskal-Wallis rank sum test for continuous data, and the Pearson’s χ ² test for categorical data.

To investigate the risk factors for the comorbidity of Aβ and cerebrovascular disease, multinomial logistic regression (MLR, using multinom function from the nnet package) was used to examine the odds of the classified categories as a function of demographic variables including age, sex, education level, Apolipoprotein E (APOE) ɛ4 genotype (determined as previously described [37]) as well as vascular risk factors of hypertension, diabetes, and cardiovascular and cerebrovascular history (collected from a detailed questionnaire regarding personal medical history and medication use [23]). The Aβ–V–category served as the reference group and odds ratios were calculated comparing each level back to the referent category.

Associations between Aβ load and WMH burden were firstly assessed in terms of continuous variables, i.e., CL and WMH/TIV, stratified by clinical classification using Generalised Linear Modelling (GLM, Gaussian model), adjusting for age. Secondly, the effect of high cerebrovascular burden (V+) on Aβ load (via the CL) was examined using GLM, adjusting for age, where V–was the referent category. The effect of high Aβ load (Aβ+) on WMH burden was likewise examined only in CN and MCI participants as the AD group had the overwhelming majority of Aβ+. The same analyses were also performed using measures of periventricular and deep WMH. All measures for the total, periventricular and deep WMH volume were log10 transformed before statistical analysis to approximate normality.

For all analyses, a p value of < 0.05 was used to indicate statistical significance.

RESULTS

Participant characteristics

Characteristics of 598 participants included in the analyses are shown across CN, MCI and AD groups in Table 1. As expected, significant group-wise differences were observed in age, APOE ɛ4 status, cognitive performance (e.g., MMSE, CDR, and PACC) and Aβ burden, both as a dichotomous measure as well as using CL values. Significantly greater total, periventricular and deep WMH burden were observed in participants with AD (all p < 0.05). Figure 1 illustrates the volumetric measures of total WMH across the three clinical classification groups. There was no significant difference between CN and MCI participants in volumetric measures for any of the three WMH types. No significant association (χ²(2, N = 598) = 0.857, p = 0.652) was observed between the incidence of subcortical infarction and clinical classification. Based on WMH volume and presence of brain infarcts, the prevalence of cerebrovascular disease in this study cohort was significantly higher in the AD group (59.8% versus 24.9% in CN and 29.2% in MCI, p < 0.001, Table 1).

Fig. 1

WMH burden among CN, MCI, and AD participants. The y-axis is on a base 10 logarithmic scale, and p values were calculated using analysis of covariance for log10-transformed WMH measures, adjusting for age and MR acquisition site.

Table 2 summarizes the characteristics of the four categories classified according to comorbidity status of Aβ and cerebrovascular disease, in the non-demented participants only; Aβ positivity was high (91%) in AD participants, and data from this group was not analyzed here. Among 511 CN (N = 422) and MCI (N = 89) participants, 261 (51.1%) were assigned to the Aβ–V–category, 67 (13.1%) to the Aβ–V+ category, 119 (23.3%) to the Aβ+V–category and 64 (12.5%) to the Aβ+V+ category. For MCI participants, the prevalence of V+ was higher in those who were Aβ+ as compared to those who were Aβ–(Aβ+V+: 37.3% versus Aβ–V+: 18.4%, p = 0.05). Significant associations across the four categories were found for age, APOE ɛ4 carriage, MMSE and PACC scores (Table 2). Investigating the pair-wise relationships, participants in the Aβ–V–category were significantly younger than those in the other three categories (all p < 0.05), whilst participants were more likely to carry an APOE ɛ4 allele if they were Aβ+ as compared with the Aβ–V–category. No group difference was observed between Aβ+V–and Aβ+V+ in APOE ɛ4 carriage.

Table 2

Characteristics of four group categories defined by the assessment of amyloid and cerebrovascular disease among CN (N = 422) and MCI (N = 89) participants

	Aβ–V–	Aβ–V+	Aβ+V–	Aβ+V+	p^*
No. of participants	261	67	119	64
No. of MCI (%)	31 (11.9)	7 (10.4)	32 (26.9)	19 (29.7)	<0.001¹
No of females (%)	141 (54.0)	40 (59.7)	66 (55.5)	30 (46.9)	0.516¹
Age, y	72.4±5.7	77.9±5.7	74.4±5.6	78.5±6.2	<0.001²
MMSE	28.8±1.3	28.5±1.3	28.0±1.7	27.4±2.3	<0.001²
PACC	–0.01±0.76	–0.38±0.78	–0.58±1.06	–0.88±1.03	<0.001²
% APOEɛ4+	16.2	9.0	51.7	48.4	<0.001¹
Centiloid	1.4±8.6	1.0±10.1	66.1±30.9	68.8±32.1	–
No. with subcortical infarcts (%)	0 (0.0)	17 (25.4)	0 (0.0)	22 (34.4)	–
No. with large cortical infarcts (%)	0 (0.0)	1 (1.5)	0 (0.0)	5 (7.8)	–
WMH/TIV, %	0.22±0.16	1.50±0.98	0.25±0.17	1.69±1.55	–
Periventricular WMH/TIV, %	0.11±0.09	0.94±0.64	0.14±0.11	1.03±0.90	–
Deep WMH/TIV, %	0.05±0.04	0.35±0.36	0.04±0.04	0.45±0.72	–

All the variables are presented as mean±standard deviation. ^*p-values computed using either Chi-square or Kruskal-Wallis test. ¹ Chi-square test for categorical variable, ² Kruskal-Wallis rank sum test.

In terms of cognitive performance, participants who were Aβ–V–showed significantly higher PACC scores than those participants who were Aβ+ (p < 0.001, adjusting for age and education level). Participants in the Aβ–V+ category had higher PACC values than those participants in both Aβ+ categories (p = 0.022 for Aβ+V–, p = 0.001 for Aβ+V+). However, no significant difference in PACC was observed between the Aβ+V–and Aβ+V+ groups (p = 0.283). A similar trend was also observed in MMSE, adjusting for age and education level.

Association between baseline risk factors and comorbidity

Table 3 shows the vascular risk factor profile for each comorbidity categories of Aβ and cerebrovascular disease in the study cohort. Significant difference was observed in the history of TIA (p = 0.005) with a higher prevalence of TIA history in the Aβ+V+ category. There were no differences in other vascular risk factors including hypertension and diabetes (all p > 0.05).

Table 3

Summary of vascular risk factor profile in the study cohort

	Aβ–V–	Aβ–V+	Aβ+V–	Aβ+V+	p ¹
No. of participants	267	69	148	114	–
Hypertension, %	92 (34.5)	34 (49.3)	53 (35.8)	44 (38.6)	0.139
Diabetes, %	17 (6.4)	4 (5.8)	14 (9.5)	11 (9.6)	0.435
Angina, %	15 (5.6)	4 (5.8)	4 (2.7)	3 (2.6)	0.430
AMI, %	6 (2.2)	2 (2.9)	3 (2.0)	3 (2.6)	0.970
AF, %	21 (7.9)	4 (5.8)	10 (6.8)	7 (6.1)	0.962
Stroke, %	2 (0.7)	1 (1.4)	2 (1.4)	3 (2.6)	0.501
TIA, %	8 (3.0)	2 (2.9)	2 (1.4)	10 (8.8)	0.005
Cardiovascular², %	31 (11.6)	9 (13.0)	14 (9.5)	12 (10.5)	0.855
Cerebrovascular³, %	10 (3.7)	3 (4.3)	4 (2.7)	13 (11.4)	0.006

AMI, acute myocardial infarction; AF, atrial fibrillation; TIA, transient ischemic attack. ¹p-values computed using either Chi-square test. ² The history of cardiovascular disease was determined from medical history information of angina, AMI and AF. ³ The history of cerebrovascular disease was determined from medical history information of stroke and TIA.

Table 4 reports the odds ratios (OR) with 95% confidence intervals (CI) from the MLR model, which demonstrates the association of demographic variables and vascular risk factors with the comorbidity categories of Aβ and cerebrovascular disease. Age was found to be a common significant risk factor for all three categories as compared with the Aβ–V–category (all p < 0.001), while APOE ɛ4 carriage was a significant risk factor only for the Aβ+ categories as compared with the Aβ–V–category (all p < 0.001). A history of cerebrovascular disease (stroke or TIA) was significantly associated with the Aβ+V+ category (OR = 3.04, p < 0.05). Although p-value did not reach the significance level, an odd ratio of 1.40 was observed for the Aβ–V+ category with a history of hypertension. The history of diabetes increases the risk of being Aβ–V+ or Aβ+V+ with OR of 1.83 and 1.93, respectively. No other associations were observed between the comorbid categories and other risk factors including sex and history of cardiovascular diseases.

Table 4

Odds ratio (95% confidence intervals) of being classified as Aβ–V+, Aβ+V–and Aβ+V+ associated with demographic variables in the entire study cohort¹

	Aβ–V+	Aβ+V–	Aβ+V+
	(N = 69)	(N = 143)	(N = 107)
	OR (95% CI)	OR (95% CI)	OR (95% CI)
Age	1.17 (1.12–1.23)^***	1.09 (1.04–1.13)^***	1.22 (1.16–1.28)^***
Sex, Male	0.73 (0.41–1.30)	1.10 (0.71–1.71)	0.79 (0.46–1.34)
APOEɛ4+	0.65 (0.26–1.62)	6.80 (4.20–11.02)^***	12.69 (7.04–22.86)^***
Hypertension	1.40 (0.79–2.49)	0.98 (0.62–1.56)	0.95 (0.55–1.64)
Diabetes	0.80 (0.24–2.63)	1.86 (0.82–4.23)	1.93 (0.75–4.96)
Cardiovascular	0.64 (0.33–1.24)	0.67 (0.40–1.11)	0.67 (0.36–1.27)
Cerebrovascular	0.81 (0.20–3.18)	0.80 (0.23–2.75)	3.04 (1.08–8.54)^*

¹N = 14 subjects with missing information were excluded in this analysis. The Aβ–V–category is the reference group. Significance levels: ^*p < 0.05; ^**p < 0.01; ^***p < 0.001.

Furthermore, we used the MLR model to describe the predicted probability of a participant to be in any of the four comorbidity categories [(1) Aβ–V–; (2) Aβ–V+; (3) Aβ+V–; (4) Aβ+V+] across the age span of 60 to 90 years, by APOE ɛ4 carrier status (Fig. 2). Regardless of APOE ɛ4 status, the predicted probability of a participant being Aβ–V–was highest at age 60 (87.7%) and decreased to 14.4% for non-ɛ4 carriers and 2.7% for ɛ4 carriers by age 90. The predicted probability of a participant being Aβ+V+ increased with age and reached a peak of 34.3% for non-ɛ4 carriers and 73.3% for ɛ4 carriers by age 90. For the Aβ–V+ category, the trend started from less than 3% for both non-ɛ4 and ɛ4 carriers at age 60, and the predicted probability subsequently increases to a peak of 35.9% by age 90 in non-ɛ4 carriers while remaining very low (<5%) for ɛ4 carriers. For the Aβ+V–category, ɛ4 carriers started from a higher probability at age 60 than non-ɛ4 carriers (36.7% versus 8.8%) and reached a peak of 44.7% near age 70. In contrast, the probability of a participant being Aβ+V–for non-ɛ4 carriers reached a peak of 19.8% at age 80. However, both trends fell to a probability below 20% at age 90.

Fig. 2

The effect of age and APOE ɛ4 allele carriage on the predicted probabilities of the four classified categories (Aβ–V–, Aβ–V+, Aβ+V–and Aβ+V+) based on Aβ-amyloid and cerebrovascular disease. 95% confidence intervals are marked by the shaded areas.

Cross-sectional association between Aβ and WMH burden

The cross-sectional correlation analysis between Aβ and measures of WMH burden was performed in the entire cohort as well as within the three clinical classification groups. When looking at the overall group, a weak but significant correlation (R² = 0.045, p < 0.001) was observed between the CL values and total WMH volume after adjustment for age. However, when looking at individual clinical classification categories, no significant correlation was noted between CL values and volumetric measures of WMH (all p > 0.05 for total, periventricular and deep WMH) (Fig. 3). When examining the association between overall cerebrovascular disease (i.e., WMH and infarcts) and Aβ load, there were no significant differences in Aβ load between V–and V+ for any of the clinical classification groups (Fig. 4, all p > 0.05). Similarly, Fig. 5 demonstrates that the Aβ status was not significantly associated with total, periventricular or deep WMH burden for both CN and MCI groups.

Fig. 3

Regression plots of Aβ deposition in Centiloid values, and (top) total, (middle) periventricular, (bottom) deep WMH burden in the AIBL cohort across different clinical classification groups. The x-axis is displayed on a base 10 logarithmic scale. The coefficient R and p values were calculated using linear regression, adjusting for age. 95% confidence intervals were marked by the dotted lines.

Fig. 4

Boxplots comparing Aβ burden (Centiloid) between cerebrovascular negative (V–) and cerebrovascular positive (V+) in groups of CN, MCI and AD participants. The p values were calculated using analysis of covariance, adjusting for age.

Fig. 5

Boxplots respectively comparing total, periventricular and deep WMH burden between Aβ negative and Aβ positive subjects in groups of CN and MCI participants. The y-axis is displayed on a base 10 logarithmic scale. The p values were calculated using analysis of covariance, adjusting for age and MR acquisition site.

DISCUSSION

We assessed the comorbidity of Aβ and cerebrovascular disease among CN, MCI, and AD participants in the AIBL cohort. In the cross-sectional analysis, higher total and regional WMH burden was observed in AD participants compared with CN and MCI participants after adjustment for age. This was also observed as a high prevalence of cerebrovascular disease (59.8%) and which was almost double the prevalence in CN and MCI participants. These results are similar to the 57.34% versus 33.2% reported from autopsy-proven AD and age-matched controls in previous studies [38]. In the entire cohort, WMH volume and Aβ burden were associated weakly after adjustment for age; however, the association was lost when stratified by clinical classification. This suggests that the associations observed between WMH and Aβ burden are likely to be driven by the older age of individuals in the AD category with concomitant larger WMH burden. These findings are similar to other reports in the literature [6, 8], which suggest an independent relationship between Aβ deposition and cerebrovascular burden.

Our findings contribute to previous findings examining spatial distribution characteristics of WMH associated with Aβ load as the distinction of periventricular and deep WMH may reflect functional and histopathological features. In our volumetric analysis, no associations between overall Aβ load and regional (periventricular or deep) WMH volumes were observed. Although region-specific WMH patterns have been proposed to be associated with cerebral amyloid in some studies [39, 40], none of these associations were reported from analyses with volumetric measures of regional WMH.

In the subset of CN and MCI participants, our reported prevalence for the comorbidity categories of Aβ and cerebrovascular disease are comparable to previous findings [8, 41] except for a lower prevalence of the A–V+ category observed in our cohort. This discrepancy might be explained by differences in demographic and vascular risk profiles of our study cohort compared to previous studies.

Age and APOE ɛ4 carriage were found to be the main variables that could influence the comorbidity of Aβ and cerebrovascular disease. Age was independently associated with the presence of cerebrovascular disease, high Aβ load, as well as with their combination. Carriage of an APOE ɛ4 allele was associated with Aβ positivity but not with cerebrovascular disease in the absence of high Aβ load (Aβ–V+). This concurs with the recent finding that the APOE ɛ4 allele impacts cognition primarily through its association with Aβ deposition but not with cerebrovascular disease. The complex interplay between age, APOE ɛ4 allele, and comorbidity between Aβ and cerebrovascular disease is well demonstrated across the older adult lifespan in Fig. 2. As expected, the probability of cerebrovascular disease appears to increase with increasing age, with APOE ɛ4 carriers representing the greatest increase without comorbid Aβ, and APOE ɛ4 carriers representing the greatest increase in the presence of comorbid Aβ. The absence of both conditions (Aβ–V–) becomes less likely with advancing age, with APOE ɛ4 carriers being less likely to be in this category throughout most of the older adult lifespan. The Aβ+V–group appears to have the most complex trend with APOE ɛ4 carriers having a peak probability at approximately 70 years of age, and APOE ɛ4 non-carriers having a peak probability beyond the age of 80 years, albeit at a lower absolute probability. This demonstrates that carriage of an APOE ɛ4 allele could lead to the early transit from Aβ–V–to Aβ+V–.

The WMH/TIV threshold estimated in this study was different from that reported in literature for determining the presence of cerebrovascular disease [8 , 42]. The cut-off of WMH/TIV measures particularly for the AIBL cohort was found to be 0.65%, which is higher than a previously used threshold of 0.5% [41, 42], but lower than the threshold of 1.11% from a more recent study [8]. Such differences are likely related to specific inclusion and exclusion criteria used in the AIBL cohort compared to other studies. Specifically, AIBL study participants have a relatively low prevalence of cerebrovascular risk factors. Further differences could arise due to different methods used for segmenting WMH.

Our study has thoroughly examined the comorbidity of Aβ and cerebrovascular disease in the AIBL cohort. A high prevalence of this comorbid condition was confirmed in the AD participants, being approximately twice the prevalence rate in CN and MCI participants. This finding is consistent with other studies [38], although in this context it must be considered that a history of significant vascular disease (such as stroke or uncontrolled diabetes) was an exclusion criterion for the AIBL study. Hence, the prevalence of vascular comorbidity may be an under-estimate of the actual prevalence. We demonstrated the necessity of accounting for the burden of cerebrovascular disease while studying neurodegenerative diseases such as AD. Given that recent studies demonstrate the differential impacts of Aβ and cerebrovascular disease on neurodegeneration (e.g., cortical thinning, hippocampus atrophy) and cognitive decline [43, 44], it is important to accurately evaluate the comorbidity status of Aβ and cerebrovascular disease that could lead to different cognitive trajectories and pathways to dementia. Assessment of comorbidity of Aβ and cerebrovascular disease will also be beneficial to develop prevention and therapeutic strategies with better specificity, which can potentially modify the Aβ and cerebrovascular pathways to cognitive decline.

Previous studies have examined the association between Aβ and cerebrovascular burden [6 , 39]. However, the question has not yet been answered definitively. The discrepant findings in the literature might relate to variable study populations and methodology, different PET radiotracers for Aβ quantification and different WMH assessment methods. The strengths of our study include the comprehensive study design of the AIBL study, which involves a large study cohort with both pre-clinical and clinical AD, standardized Aβ quantification from different radiotracers as well as accurate volumetric assessment of WMH. Moreover, spatial distribution characteristics of WMH were considered in our analyses given that distinctive functional and histopathological features for periventricular and deep WMH have been proposed [22, 40].

This study has several limitations that require consideration. First, the study is cross-sectional in nature. Future longitudinal studies are needed to allow a full examination of the association between the two complex processes of Aβ accumulation and cerebrovascular disease, particularly to understand the patterns through which individuals transit between different comorbidity groups over time. Second, the AIBL study does not represent an epidemiological sample and therefore may not be representative of the general population. For example, in view of the selection criteria for the AIBL study aiming to study AD, a relatively low prevalence of brain infarction was expected and observed, particularly for cortical infarcts in participants with MCI and AD. On the other hand, the rate of APOE ɛ4 carriage and Aβ positivity is likely to be an overestimate of the general population. However, these results from an AD-focused study have important implications that the comorbid cerebrovascular condition may need to be considered in clinical trials of anti-amyloid therapies. Third, measures of WMH and brain infarction that were used to evaluate cerebrovascular disease in this study do not fully measure the spectrum of cerebrovascular disease on MRI (e.g., cerebral microinfarcts, microbleeds, perivascular spaces). Although WMH served as the traditional biomarker for cerebrovascular disease, other MRI-based markers such as cerebral microinfarcts could provide more information for estimating the cerebrovascular burden [44, 45]. However, the optimal combination of these measures for quantifying total cerebrovascular disease burden still remains unclear. Fourth, assessment of the probability of different comorbidity categories may be affected by survival bias. For example, the effects of cerebrovascular disease on mortality may confound the proportion of individuals predicted to be V+ at older age group.

In conclusion, comorbidity of cerebrovascular disease and Aβ is common in later life. The risk of comorbidity increases with ageing, while APOE ɛ4 carriage is only associated with AD risk. While our study did not demonstrate a convincing cross-sectional association between the burden of Aβ and cerebrovascular disease, future longitudinal studies are required to further confirm this.

Footnotes

ACKNOWLEDGMENTS

We thank all those who took part as subjects in the study for their commitment and dedication to helping advance research into the early detection and causation of AD. We kindly thank all AIBL Research Group members ().

The AIBL study () is a consortium between Austin Health, CSIRO, Edith Cowan University, the Florey Institute (The University of Melbourne), and the National Ageing Research Institute. Partial financial support provided by the Alzheimer’s Association (US), the Alzheimer’s Drug Discovery Foundation, an Anonymous foundation, the Science and Industry Endowment Fund (SIEF), the Dementia Collaborative Research Centres (DCRC2), the Victorian Government’s Operational Infrastructure Support program, the McCusker Alzheimer’s Research Foundation, the National Health and Medical Research Council (NHMRC), as well as the Cooperative Research Centre (CRC) for Mental Health. Numerous commercial interactions have supported data collection and analysis. In-kind support has also been provided by Sir Charles Gairdner Hospital, CogState Ltd., Hollywood Private Hospital, the University of Melbourne, and St Vincent’s Hospital. NY is funded by a National Health and Medical Research Council –Australian Research Council Dementia Research Fellowship. SRRS is funded by a BrightFocus Foundation Fellowship.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Petrovitch

, Ross

, Steinhorn

, Abbott

, Markesbery

, Davis

, Nelson

, Hardman

, Masaki

, Vogt

(2005) AD lesions and infarcts in demented and non-demented Japanese-American men. Ann Neurol 57, 98–103.

Schneider

, Arvanitakis

, Bang

, Bennett

(2007) Mixed brain pathologies account for most dementia cases in community-dwelling older persons. Neurology 69, 2197–2204.

Breteler

(2000) Vascular risk factors for Alzheimer’s disease: An epidemiologic perspective. Neurobiol Aging 21, 153–160.

Bos

, Verhey

, Ramakers

, Jacobs

, Soininen

, Freund-Levi

, Hampel

, Tsolaki

, Wallin

ÅK

, Van Buchem

(2017) Cerebrovascular and amyloid pathology in predementia stages: The relationship with neurodegeneration and cognitive decline. Alzheimers Res Ther 9, 101.

Jellinger

, Attems

(2006) Prevalence and impact of cerebrovascular pathology in Alzheimer’s disease and parkinsonism. Acta Neurol Scand 114, 38–46.

Roseborough

, Ramirez

, Black

, Edwards

(2017) Associations between amyloid β and white matter hyperintensities: A systematic review. Alzheimers Dement 13, 1154–1167.

Van Westen

, Lindqvist

, Blennow

, Minthon

, Nägga

, Stomrud

, Zetterberg

, Hansson

(2016) Cerebral white matter lesions–associations with Aβ isoforms and amyloid PET. Sci Rep 6, 20709.

Vemuri

, Lesnick

, Przybelski

, Knopman

, Preboske

, Kantarci

, Raman

, Machulda

, Mielke

, Lowe

(2015) Vascular and amyloid pathologies are independent predictors of cognitive decline in normal elderly. Brain 138, 761–771.

Vemuri

, Knopman

(2016) The role of cerebrovascular disease when there is concomitant Alzheimer disease. Biochim Biophys Acta 1862, 952–956.

10.

Carmichael

, Schwarz

, Drucker

, Fletcher

, Harvey

, Beckett

, Jack

, Weiner

, DeCarli

(2010) Longitudinal changes in white matter disease and cognition in the first year of the Alzheimer Disease Neuroimaging Initiative. Arch Neurol 67, 1370–1378.

11.

, Jagust

(2012) Vascular burden and Alzheimer disease pathologic progression. Neurology 79, 1349–1355.

12.

Maillard

, Carmichael

, Fletcher

, Reed

, Mungas

, De Carli

(2012) Coevolution of white matter hyperintensities and cognition in the elderly. Neurology 79, 442–448.

13.

Boyle

, Yu

, Fleischman

, Leurgans

, Yang

, Wilson

, Schneider

, Arvanitakis

, Arfanakis

, Bennett

(2016) White matter hyperintensities, incident mild cognitive impairment, and cognitive decline in old age. Ann Clin Transl Neurol 3, 791–800.

14.

Brickman

, Provenzano

, Muraskin

, Manly

, Blum

, Apa

, Stern

, Brown

, Luchsinger

, Mayeux

(2012) Regional white matter hyperintensity volume, not hippocampal atrophy, predicts incident Alzheimer disease in the community. Arch Neurol 69, 1621–1627.

15.

Smith

, Egorova

, Blacker

, Killiany

, Muzikansky

, Dickerson

, Tanzi

, Albert

, Greenberg

, Guttmann

(2008) Magnetic resonance imaging white matter hyperintensities and brain volume in the prediction of mild cognitive impairment and dementia. Arch Neurol 65, 94–100.

16.

Debette

, Markus

(2010) The clinical importance of white matter hyperintensities on brain magnetic resonance imaging: Systematic review and meta-analysis. BMJ 341, c3666.

17.

Kim

, MacFall

, Payne

(2008) Classification of white matter lesions on magnetic resonance imaging in elderly persons. Biol Psychiatry 64, 273–280.

18.

Saczynski

, Sigurdsson

, Jonsdottir

, Eiriksdottir

, Jonsson

, Garcia

, Kjartansson

, Lopez

, van Buchem

, Gudnason

(2009) Cerebral infarcts and cognitive performance: Importance of location and number of infarcts. Stroke 40, 677–682.

19.

Schneider

, Wilson

, Bienias

, Evans

, Bennett

(2004) Cerebral infarctions and the likelihood of dementia from Alzheimer disease pathology. Neurology 62, 1148–1155.

20.

Scott

, Braskie

, Tosun

, Thompson

, Weiner

, DeCarli

, Carmichael

, Alzheimer’s Disease Neuroimaging Initiative (2015) Cerebral amyloid and hypertension are independently associated with white matter lesions in elderly. Front Aging Neurosci 7, 221.

21.

Weaver

, Doeven

, Barkhof

, Biesbroek

, Groeneveld

, Kuijf

, Prins

, Scheltens

, Teunissen

, van der Flier

(2019) Cerebral amyloid burden is associated with white matter hyperintensity location in specific posterior white matter regions. Neurobiol Aging 84, 225–234.

22.

Griffanti

, Jenkinson

, Suri

, Zsoldos

, Mahmood

, Filippini

, Sexton

, Topiwala

, Allan

, Kivimäki

(2018) Classification and characterization of periventricular and deep white matter hyperintensities on MRI: A study in older adults. Neuroimage 170, 174–181.

23.

Ellis

, Bush

, Darby

, De Fazio

, Foster

, Hudson

, Lautenschlager

, Lenzo

, Martins

, Maruff

(2009) The Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging: Methodology and baseline characteristics of 1112 individuals recruited for a longitudinal study of Alzheimer’s disease. Int Psychogeriatr 21, 672–687.

24.

McKhann

, Drachman

, Folstein

, Katzman

, Price

, Stadlan

(1984) Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 34, 939–939.

25.

Petersen

, Smith

, Waring

, Ivnik

, Tangalos

, Kokmen

(1999) Mild cognitive impairment: Clinical characterization and outcome. Arch Neurol 56, 303–308.

26.

Lim

, Snyder

, Pietrzak

, Ukiqi

, Villemagne

, Ames

, Salvado

, Bourgeat

, Martins

, Masters

(2016) Sensitivity of composite scores to amyloid burden in preclinical Alzheimer’s disease: Introducing the Z-scores of Attention, Verbal fluency, and Episodic memory for Nondemented older adults composite score. Alzheimers Dement (Amst) 2, 19–26.

27.

Rowe

, Ellis

, Rimajova

, Bourgeat

, Pike

, Jones

, Fripp

, Tochon-Danguy

, Morandeau

, O’Keefe

(2010) Amyloid imaging results from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging. Neurobiol Aging 31, 1275–1283.

28.

Vandenberghe

, Van Laere

, Ivanoiu

, Salmon

, Bastin

, Triau

, Hasselbalch

, Law

, Andersen

, Korner

(2010) 18F-flutemetamol amyloid imaging in Alzheimer disease and mild cognitive impairment: A phase 2 trial. Ann Neurol 68, 319–329.

29.

Wong

, Rosenberg

, Zhou

, Kumar

, Raymont

, Ravert

, Dannals

, Nandi

, Brašić

, Ye

(2010) In vivo imaging of amyloid deposition in Alzheimer disease using the radioligand 18F-AV-45 (flobetapir F 18). J Nucl Med 51, 913–920.

30.

Bourgeat

, Villemagne

, Dore

, Brown

, Macaulay

, Martins

, Masters

, Ames

, Ellis

, Rowe

(2015) Comparison of MR-less PiB SUVR quantification methods. Neurobiol Aging 36, S159–166.

31.

Klunk

, Koeppe

, Price

, Benzinger

, Devous

Sr , Jagust

, Johnson

, Mathis

, Minhas

, Pontecorvo

(2015) The Centiloid Project: Standardizing quantitative amyloid plaque estimation by PET. Alzheimers Dement 11, 1–15. e14.

32.

Bourgeat

, Doré

, Fripp

, Ames

, Masters

, Salvado

, Villemagne

, Rowe

, AIBL research group (2018) Implementing the centiloid transformation for 11C-PiB and β-amyloid 18F-PET tracers using CapAIBL. Neuroimage 183, 387–393.

33.

Amadoru

, Doré

, McLean

, Hinton

, Shepherd

, Halliday

, Leyton

, Yates

, Hodges

, Masters

(2020) Comparison of amyloid PET measured in Centiloid units with neuropathological findings in Alzheimer’s disease. Alzheimers Res Ther 12, 1–8.

34.

Kantarci

, Petersen

, Przybelski

, Weigand

, Shiung

, Whitwell

, Negash

, Ivnik

, Boeve

, Knopman

(2008) Hippocampal volumes, proton magnetic resonance spectroscopy metabolites, and cerebrovascular disease in mild cognitive impairment subtypes. Arch Neurol 65, 1621–1628.

35.

Manjón

, Coupé

, Raniga

, Xia

, Desmond

, Fripp

, Salvado

(2018) MRI white matter lesion segmentation using an ensemble of neural networks and overcomplete patch-based voting. Comput Med Imaging Graph 69, 43–51.

36.

Manjón

, Coupé

, Raniga

, Xia

, Fripp

, Salvado

(2016) HIST: Hyperintensity segmentation tool. In International Workshop on Patch-based Techniques in Medical Imaging Springer, pp. 92-99.

37.

Porter

, Burnham

, Savage

, Lim

, Maruff

, Milicic

, Peretti

, Ames

, Masters

, Martins

(2018) A polygenic risk score derived from episodic memory weighted genetic variants is associated with cognitive decline in preclinical Alzheimer’s disease. Front Aging Neurosci 10, 423.

38.

Jellinger

, Attems

(2005) Prevalence and pathogenic role of cerebrovascular lesions in Alzheimer disease. J Neurol Sci 229, 37–41.

39.

Graff-Radford

, Arenaza-Urquijo

, Knopman

, Schwarz

, Brown

, Rabinstein

, Gunter

, Senjem

, Przybelski

, Lesnick

(2019) White matter hyperintensities: Relationship to amyloid and tau burden. Brain 142, 2483–2491.

40.

Marnane

, Al-Jawadi

, Mortazavi

, Pogorzelec

, Wang

, Feldman

, Hsiung

G-YR

, Alzheimer’s

Disease Neuroimaging Initiative

(2016) Periventricular hyperintensities are associated with elevated cerebral amyloid. Neurology 86, 535–543.

41.

Marchant

, Reed

, DeCarli

, Madison

, Weiner

, Chui

, Jagust

(2012) Cerebrovascular disease, beta-amyloid, and cognition in aging. Neurobiol Aging 33, 1006.e1025–1006.e1036.

42.

DeCarli

, Murphy

, Tranh

, Grady

, Haxby

, Gillette

, Salerno

, Gonzales-Aviles

, Honvitz

, Rapoport

(1995) The effect of white matter hyperintensity volume on brain structure, cognitive performance, and cerebral metabolism of glucose in 51 healthy adults. Neurology 45, 2077–2084.

43.

Vemuri

, Lesnick

, Knopman

, Przybelski

, Reid

, Mielke

, Graff-Radford

, Lowe

, Machulda

, Petersen

(2019) Amyloid, vascular, and resilience pathways associated with cognitive aging. Ann Neurol 86, 866–877.

44.

Yassi

, Hilal

, Xia

, Lim

, Watson

, Kuijf

, Fowler

, Yates

, Maruff

, Martins

, Ames

, Chen

, Rowe

, Villemagne

, Salvado

, Desmond

, Masters

(2020) Influence of comorbidity of cerebrovascular disease and amyloid-β on Alzheimer’s disease. J Alzheimers Dis 73, 897–907.

45.

Paradise

, Shepherd

, Wen

, Sachdev

(2018) Neuroimaging and neuropathology indices of cerebrovascular disease burden: A systematic review. Neurology 91, 310–320.