Impact of sex and APOE4 on cerebral amyloid angiopathy in Alzheimer’s disease

Abstract

Cerebral amyloid angiopathy (CAA) often coexists with Alzheimer’s disease (AD). APOE4 is a strong genetic risk factor for both AD and CAA. Sex-dependent differences have been shown in AD as well as in cerebrovascular diseases. Therefore, we examined the effects of APOE4, sex, and pathological components on CAA in AD subjects. A total of 428 autopsied brain samples from pathologically-confirmed AD cases were analyzed. CAA severity was histologically scored in inferior parietal, middle frontal, motor, superior temporal and visual cortexes. In addition, subgroups with severe CAA (n=60) or without CAA (n=39) were subjected to biochemical analysis of amyloid-β (Aβ) and apolipoprotein E (apoE) by ELISA in the temporal cortex. After adjusting for age, Braak neurofibrillary tangle stage and Thal amyloid phase, we found that overall CAA scores were higher in males than females. Furthermore, carrying one or more APOE4 alleles was associated with higher overall CAA scores. Biochemical analysis revealed that the levels of detergent soluble and insoluble Aβ40, and insoluble apoE were significantly elevated in individuals with severe CAA or APOE4. The ratio of Aβ40/Aβ42 in insoluble fractions was also increased in the presence of CAA or APOE4, although it was negatively associated with male sex. Levels of insoluble Aβ40 were positively associated with those of insoluble apoE, which were strongly influenced by CAA status. Pertaining to insoluble Aβ42, the levels of apoE correlated regardless of CAA status. Our results indicate that sex and APOE genotypes differentially influence the presence and severity of CAA in AD, likely by affecting interaction and aggregation of Aβ40 and apoE.

Introduction

Alzheimer’s disease (AD) is a progressive neurodegenerative disease and the most common form of dementia in the elderly [2]. Epidemiologically, the prevalence of AD and other dementias is higher in females than males in the aged population. At 65 years old, the future risk of developing AD for females is 17.2%, compared with 9.1% chance for males [2]. While senile plaques, due to brain amyloid-β (Aβ) deposition in the parenchyma, and neurofibrillary tangles are the two major histological hallmarks for AD [41], the vast majority (>80%) of autopsy-confirmed AD cases also exhibit at least a mild degree of cerebral amyloid angiopathy (CAA) [12]. In CAA, Aβ deposits, predominantly in vascular smooth muscle layers, are present in leptomeningeal, cortical arteries and capillaries comprising the blood-brain barrier [3, 37, 38, 48]. Importantly, CAA is a major cause of intracerebral hemorrhage [16, 18, 27, 49] and is responsible for ~20% of cerebral hemorrhage cases in the elderly [12]. In addition, severe CAA often associates with microangiopathies (e.g. microaneurysm formation, fibrinoid necrosis) [28] and cerebral microinfarcts [44]. Although various hypotheses have been proposed to explain the increased prevalence of AD in females, whether sex-dependent differences exist in CAA remains unclear.

Genome-wide association studies have confirmed that the ε4 allele of the APOE gene encoding apolipoprotein E (apoE) is the strongest genetic risk factor for AD [19, 25]. Meta-analysis has also shown association of APOE4 with severe CAA [36]. Furthermore, APOE4 is associated with increased risk of age-related cognitive decline as well as vascular cognitive impairment during normal aging [26]. Among the three human apoE isoforms, apoE4 is not only inhibitory to brain Aβ clearance, but is also less efficient in supporting synapses and cerebrovascular functions [8, 22]. Importantly, while APOE4 carriers, in general, have a higher risk of conversion to mild cognitive impairment (MCI) or AD among cognitively normal individuals, the APOE4 effect is considerably stronger in females compared to males [1].

In this study, we investigated how APOE genotype and sex influence the risk of CAA in pathologically-confirmed AD cases using a large sample cohort of postmortem brain tissue (n=428). Additionally, the levels of CAA-related molecules were biochemically assessed by ELISA in the temporal cortex of a subset of cases with severe CAA (n=60) or without CAA (n=39). Our results demonstrate that in AD, male sex and APOE4 are significantly associated with CAA severity. CAA incidence, sex and APOE4 influence brain levels of Aβ and apoE, and their association in AD. Our findings provide novel insights into the pathogenesis of CAA in AD.

Materials and Methods

Human autopsied brain samples

The Mayo Clinic brain bank for neurodegenerative disorders at Jacksonville was queried for neuropathologically-confirmed AD cases with available frozen tissue, known APOE genotype, and absence of the following; TDP-43 proteinopathies (hippocampal sclerosis, frontotemporal lobar degeneration, amyotrophic lateral sclerosis), primary tauopathies (progressive supranuclear palsy, corticobasal degeneration), argyrophilic grain disease, and α-synucleinopathies (Lewy body disease and multiple systems atrophy). The presence of cerebrovascular lesions was not considered as an exclusion criteria. A total of 428 AD cases were identified for inclusion in this study. Experimental procedures were conducted in strict accordance with protocol approved by the Mayo Clinic Institutional Review Board. All cases previously underwent standardized neuropathological assessments, as previously described [31]. Thioflavin S fluorescent microscopy was used to assign a Braak neurofibrillary tangle stage, Thal amyloid phase and CAA score (Table 1). APOE genotype distributions were as follows: ε2/ε3 (n=12), ε2/ε4 (n=9), ε3/ε3 (n=159), ε3/ε4 (n=188) and ε4/ε4 (n=60) (supplemental Figure 1).

Table 1.

Subject characteristics of AD brain samples

	APOE4 non-carrier		APOE4 carrier
	Female (N=103)	Male (N=68)	Female (N=143)	Male (N=114)
Age	81 (75, 88)	80 (73, 85)	83 (78, 88)	79 (73, 83)
Braak Stage
<5	12 (12)	8 (12)	11 (8)	10 (9)
5-5.5	39 (38)	27 (40)	39 (27)	45 (39)
6	52 (50)	33 (49)	93 (65)	59 (52)
Thal Stage
Missing	1	0	2	2
<4	11 (11)	6 (9)	8 (6)	11 (10)
4	8 (8)	7 (10)	11 (8)	10 (9)
5	83 (81)	55 (81)	122 (87)	91 (81)

Median and interquartile range (IQR) shown for age; number (%) shown for Braak and Thal stages.

CAA assessment

CAA severity was scored in the inferior parietal cortex, middle frontal cortex, motor cortex, superior temporal cortex and visual cortex using a semi-quantitative method through thioflavin-S staining as follows; 0 = no amyloid positive vessels; 0.5 = scattered amyloid deposition only in leptomeninges; 1 = scattered amyloid deposition in both leptomeningeal and cortical vessels; 2 = strong circumferential amyloid deposition in multiple cortical and leptomeningeal vessels; 3 = widespread strong amyloid deposition in leptomeningeal and cortical vessels; 4 = more severe CAA than score 3 (Fig. 1).

Figure 1. Morphologies of CAA with different scores in AD brains.

Cerebral amyloid angiopathy (CAA) was evaluated using thioflavin-S fluorescent microscopy. (a) Cases with no CAA were given a score of 0. Cases found to have CAA only in leptomeninges were given a score of 0.5. (b) Scattered amyloid deposition in both leptomeningeal and cortical vessels was given a score of 1. (c) Strong, circumferential amyloid deposition in some, but not all vessels was given a score of 2. (d) Widespread, strong amyloid deposition in leptomeningeal and cortical vessels was given a score of 3. (e) Cases with more severe CAA than score 3 were given a score of 4. (a'-e ') Higher magnification views from each image are shown. Scale bar= 200 μm.

Sample preparation

Temporal cortex tissues were dissected from AD subjects with severe CAA (n=60) or without CAA (n=39), and kept frozen until extraction. Brain lysates were prepared by a three-step extraction method based upon differential solubility in detergents (Triton X-100) and the chaotropic agent guanidine hydrochloride (GuHCl) as previously described [42, 43]. In brief, after removal of meninges and blood vessels, 100–200 mg of frozen brain tissue were homogenized in 10-15 volumes (w/v) of ice-cold TBS-containing protease inhibitor cocktail (Roche Diagnostics) by Polytron homogenizer (KINEMATICA). After centrifugation at 100,000 × g for 60 min at 4°C, the supernatant was aliquoted and stored at −80°C (referred to as TBS-soluble fraction or TBS). The residual pellet was re-homogenized in TBS plus 1% Triton X-100 with protease inhibitor cocktail, incubated with mild agitation for 1 h at 4°C and centrifuged as described above. The resulting supernatant was aliquoted and stored at −80°C (referred to as TX-soluble fraction or TX). The residual pellet was re-homogenized in 5 M GuHCl (pH 7.6), and incubated with mild agitation for 12–16 h at 22°C. After centrifugation as above, the resultant supernatant (referred to as the insoluble or GuHCl fraction) was diluted with nine volumes of TBS, aliquoted and stored at −80°C.

Quantification of CAA-related molecules

Levels of Aβ40, Aβ42 and apoE were determined by ELISA as previously described [42, 43]. Colorimetric quantification was performed on a Synergy HT plate reader (BioTek) using horseradish peroxidase (HRP)-linked streptavidin (Vector) and 3,3′,5,5′-tetramethylbenzidine substrate (Sigma). All measured values by ELISA were normalized against total protein measured by BCA assay.

Data analysis

Data were descriptively summarized using median and interquartile ranges for continuous variables and frequencies and percentages for categorical variables. The overall CAA score for the primary analysis was defined as the mean of the region specific scores. Associations of sex and APOE4 with CAA score were examined using linear regression models with adjustment for age, Braak stage, and Thal amyloid phase. Associations between CAA, sex, and APOE4+ with log-transformed Aβ and apoE levels were examined using multivariate logistic regression. The parameter estimates were exponentiated to obtain estimates of relative levels in one group compared to another. Age, Braak stage and Thal amyloid phase were adjusted for all models. All statistical tests were two-tailed. Analyses were primarily performed using SAS statistical software (SAS Institute Inc, Cary, NC).

Results

APOE genotype, sex and CAA severity in AD

Of the 428 pathologically-confirmed AD cases, we found that 374 AD subjects (88.4%) had CAA pathology in at least one of the five cortical regions analyzed. Visual cortex had the highest frequency of CAA with a score of 2 or more (36.7%), followed by the middle frontal cortex (35.0%), motor cortex (28.0%), inferior parietal cortex (27.8%) and superior temporal cortex (8.9%) (Fig. 2 and supplemental Table 1). There was evidence that males and those with one or more APOE4 alleles had higher overall CAA scores, with or without adjusting for age, Braak stage and Thal amyloid phase (Table 2). In particular, male sex showed higher CAA scores than females in every cortical region (supplemental Table 2). The associations between APOE4 and average CAA score were statistically significant only in females (n=246), but not in males (n=182), although there was no evidence of an interaction between APOE4 and sex (p=0.84) (Fig. 3 and Table 2). Increasing APOE4 allele number was associated with higher CAA score regardless of sex (Table 2 and supplemental Figure 1).

Figure 2. CAA severity in different cortical regions.

The distribution of CAA scores in inferior parietal cortex, middle frontal cortex, motor cortex, superior temporal cortex and visual cortex are shown as they relate to sex and APOE4 status.

Table 2.

Association of Sex and APOE4 with CAA Score

		Sex and APOE4 in model c		Additionally adjusting for age, Braak stage and Thal stage c
Covariate	Subgroup	Effect (95% CI)	p-value d	Effect (95% CI)	p-value d
Male	All a	0.25 (0.10, 0.40)	<.001	0.27 (0.12, 0.42)	<.001
	APOE4− b	0.27 (0.03, 0.51)	0.027	0.29 (0.05, 0.53)	0.017
	APOE4+ b	0.24 (0.05, 0.43)	0.014	0.26 (0.06, 0.46)	0.010
APOE4+	All a	0.23 (0.08, 0.38)	0.003	0.22 (0.06, 0.37)	0.005
	Females b	0.24 (0.04, 0.44)	0.017	0.23 (0.03, 0.43)	0.024
	Males b	0.21 (−0.02, 0.45)	0.074	0.20 (−0.03, 0.43)	0.095
APOE4 allelic dose	All a	0.30 (0.20, 0.41)	<.001	0.30 (0.19, 0.40)	<.001
	Females b	0.30 (0.16, 0.44)	<.001	0.29 (0.15, 0.43)	<.001
	Males b	0.30 (0.15, 0.46)	<.001	0.31 (0.15, 0.46)	<.001

^aResults from models with no interaction terms.

^bResults from models with interaction of APOE4 and sex.

^cAge is included as a continuous variable, Braak stage as dichotomous (6 vs. <6), and Thal as dichotomous (5 vs. <5).

^dWald p-value.

Figure 3. Sex- and APOE4-dependent differences in CAA severity.

The overall averaged CAA scores from the five cortical regions were plotted as they relate to sex and APOE4 status. Horizontal lines are medians and boxes are interquartile ranges (IQRs).

Biochemical analysis of Aβ and apoE in CAA

To further investigate the effects of APOE4 and sex on CAA, levels of Aβ and apoE in temporal cortex from AD cases (males n=49; females n=50) without CAA (n=39), or with severe CAA (average CAA score >2; n=60) were subjected to biochemical analysis (Table 3). Multiple regression analyses, including variables for CAA, sex and APOE4 status, showed statistically significant evidence that the levels of Aβ40, but not Aβ42, in TBS, TX and GuHCl fractions were higher in the presence of CAA or APOE4 (Fig. 4 and Table 4). Accordingly, in those with APOE4, the ratios of Aβ40/Aβ42 in all fractions were significantly higher in cases with CAA. Aβ40/Aβ42 ratios in TBS and GuHCl fractions were higher by an estimated 1.86 (p=0.003) and 3.62 fold (p<0.001), respectively. Interestingly, male sex was associated with lower Aβ40/Aβ42 ratios in the TBS and GuHCl fractions (Fig. 4 and Table 4). ApoE levels in GuHCl fraction were elevated in cases with CAA or APOE4 (Table 4). There was no compelling evidence of interactions in CAA, sex and APOE4 status among the different parameters (supplemental Table 3). When Thal amyloid phase was used as a variable instead of CAA status, neither the amounts of Aβ40 and Aβ42, nor the ratios of Aβ40/Aβ42 in any fraction were associated with Thal amyloid phase and sex in our AD cohort (data not shown).

Table 3.

Levels of Aβ and apoE in temporal cortex from AD subjects with or without CAA

	Without CAA (N=39)	With CAA (N=60)
Sex; Male	17 (44)	32 (53)
APOE4+	21 (54)	42 (70)
Age; years	80 (76, 85)	80 (75, 85)
Braak Stage
<5	4 (10)	3 (5)
5-5.5	14 (36)	24 (40)
6	21 (54)	33 (55)
Thal phase
<4	2 (5)	4 (7)
4	3 (8)	7 (12)
5	34 (87)	49 (82)
Aβ40 (ng/mg protein)
TBS	0.26 (0.03)	1.43 (0.19)
TX	0.25 (0.03)	1.15 (0.14)
GuHCl	35.1 (9.8)	708.0 (108.4)
Aβ42 (ng/mg protein)
TBS	0.46 (0.11)	0.58 (0.07)
TX	5.16 (0.55)	3.92 (0.27)
GuHCl	687.6 (57.3)	839.3 (65.7)
Aβ40/Aβ42
TBS	2.65 (1.01)	3.89 (0.83)
TX	0.09 (0.03)	0.35 (0.05)
GuHCl	0.06 (0.02)	0.74 (0.09)
ApoE (ng/mg protein)
TBS	1.27 (0.08)	1.37 (0.07)
TX	1.59 (0.08)	1.74 (0.11)
GuHCl	381.4 (78.2)	585.3 (59.0)

Median and interquartile range (IQR) shown for age; number (%) shown for gender, APOE4+, Braak stage and, Thal phase; Mean (SEM) shown for Aβ and apoE levels.

Figure 4. CAA-, Sex- and APOE4-dependent differences in Aβ40 levels and Aβ40/Aβ42 ratios.

The levels of Aβ40 in TBS (a), TX (b) and GuHCl (c) fractions and Aβ40/Aβ42 ratio in GuHCl fraction (d) from temporal cortex of AD cases with or without CAA were plotted as they relate to sex and APOE4 status. Horizontal lines are medians and boxes are interquartile ranges (IQRs).

Table 4.

Associations of CAA, sex, and APOE4 with levels of Aβ and apoE in temporal cortex

	TBS		TX		GuHCl
	Relative Level a (95% CI)	p-value b	Relative Level a (95% CI)	p-value b	Relative Level a (95% CI)	p-value b
Aβ40
CAA+	3.50 (2.45, 5.00)	<.001	3.64 (2.37, 5.59)	<.001	17.55 (10.17, 30.29)	<.001
Sex; Male	0.71 (0.50, 1.00)	0.053	0.77 (0.51, 1.16)	0.213	0.68 (0.40, 1.17)	0.161
APOE4+	1.71 (1.20, 2.46)	0.003	1.61 (1.06, 2.47)	0.027	3.99 (2.30, 6.92)	<.001
Aβ42
CAA+	0.88 (0.67, 1.17)	0.377	1.74 (0.99, 3.07)	0.053	1.22 (0.98, 1.51)	0.077
Sex; Male	1.10 (0.83, 1.45)	0.513	0.99 (0.57, 1.73)	0.978	1.15 (0.93, 1.42)	0.208
APOE4+	0.92 (0.69, 1.22)	0.575	1.26 (0.72, 2.22)	0.423	1.10 (0.88, 1.37)	0.387
Aβ40/Aβ42
CAA+	3.97 (2.67, 5.92)	<.001	2.30 (1.43, 3.69)	<.001	14.43 (8.74, 23.83)	<.001
Sex; Male	0.64 (0.43, 0.96)	0.029	0.91 (0.57, 1.44)	0.676	0.59 (0.36, 0.97)	0.038
APOE4+	1.86 (1.24, 2.78)	0.003	1.13 (0.70, 1.81)	0.624	3.62 (2.18, 6.01)	<.001
ApoE
CAA+	1.08 (0.94, 1.25)	0.275	1.07 (0.92, 1.24)	0.409	1.69 (1.15, 2.48)	0.007
Sex; Male	1.09 (0.95, 1.26)	0.231	1.15 (0.99, 1.34)	0.068	0.74 (0.51, 1.08)	0.119
APOE4+	0.80 (0.69, 0.93)	0.003	0.96 (0.83, 1.13)	0.649	1.69 (1.15, 2.49)	0.008

^aAnalyses use logarithm of biochemical measures as response variable in linear regression analyses including variables CAA group (severe versus none), sex (male vs. female) and APOE4 (presence vs. absence) and also adjusting for age; estimated effects are exponentiated to provide effect expressed as a relative level or fold change.

^bWald p-value.

Next, potential association between the levels of Aβ and apoE was assessed by linear regression models; apoE levels were positively correlated with both Aβ40 (R²=0.100, p=0.001) and Aβ42 levels (R²=0.377, p<0.001) in GuHCl fractions, but not in TBS (Aβ40: R²=0.008, p=0.376; Aβ42: R²=0.001, p=0.709) and TX (Aβ40: R²=0.005, p=0.495; Aβ42: R²=0.011, p=0.301) fractions. Aβ42 levels were also positively correlated with Aβ40 levels in TBS (R²=0.290, p<0.001) and GuHCl fractions (R²=0.357, p<0.001), but not in TX (R²=0.003, p=0.569) fractions. ANCOVA demonstrated a strong interaction effect of CAA on insoluble apoE levels and insoluble Aβ40 levels (p<0.001) (Fig. 5a), while the apoE levels correlated with insoluble Aβ42 levels regardless of CAA status (Fig. 5b).

Figure 5. Associations between the levels of apoE and Aβ as a function of CAA.

The regression plots for insoluble levels of apoE vs. Aβ40 (a) and apoE vs. Aβ42 (b) were shown.

Discussion

CAA causes cerebrovascular lesions including intracerebral hemorrhage, cerebral ischemia and leukoencephalopathy. In the elderly, CAA pathology is often accompanied by psychiatric symptoms, such as personality change and depression [16, 18]. Currently, CAA is pathologically diagnosed by deposition and accumulation of amyloid peptides, mostly Aβ, in cerebral vessels [3, 37, 38, 48] . Recent findings have shown that Aβ in the brain interstitial fluid is eliminated through the lymphatic drainage pathway flowing from capillaries to cerebral arteries along the basement membranes of cerebrovasculature [4, 30]. Thus, the failure of this drainage pathway might contribute to CAA pathogenesis [20, 50]. The deposition of Aβ in vascular walls likely leads to their damage and subsequent degeneration, which compromises the integrity of the neurovascular units. In this study, we investigated CAA in a large cohort of postmortem AD brain samples. We unequivocally demonstrated a sex-dependent difference in CAA among those with AD, with men exhibiting higher CAA scores than women. Importantly, this correlation persisted even after adjusting for age, Braak stage, Thal amyloid phase and APOE genotype. Recently, an Oxford Project to Investigate Memory and Ageing (OPTIMA) study also reported higher CAA scores in men compared with women in AD (n=154), although it was not significant (p=0.064), potentially due to a relatively small sample size [13]. Sexual dimorphism has been observed not only in AD risk, but also in the pathologic indices [5, 10]. Among clinically diagnosed AD patients, more severe AD pathology including neuritic plaques, diffuse plaques, and neurofibrillary tangles was detected in postmortem brains from female cases than those from male cases [5]. In addition, women had more extensive senile plaque depositions throughout the brain compared to men at each early Braak stage I - III [10]. These results suggest that sex impacts pathogenic pathways for Aβ deposition as senile plaques in brain parenchyma and as CAA in cerebrovasculature. Interestingly, a sex-dependent difference has also been observed in the development of intracranial atherosclerosis, where men are more susceptible than women between 30-60 years of age [35]. Stroke incidence is also higher in men than in women among those aged 45 to 84; however, the sex-dependent effect reverses in the oldest group (aged 85-94) [34]. Importantly, functions of the cerebral vasculature are affected by sex steroid hormones. For example, estrogen protects endothelial cells through diverse pathways by ameliorating mitochondrial functions and reducing reactive oxygen species [23]. Furthermore, chronic estrogen exposure has been shown to decrease cerebral vascular tone and increase cerebral blood flow, while testosterone has the opposite effects [23]. Thus, men may be at increased risk of chronic cerebrovascular damage compared with women due to a lack of estrogen-mediated protection. We propose that this damage may increase the risk of having severe CAA in AD.

Furthermore, accumulating evidence has revealed the strong contribution of APOE4 to CAA development [13, 36, 39, 40]. Consistent with these data, our results also demonstrate that severity of CAA in AD is higher in those with APOE4 in an allelic dose-dependent manner. APOE4 has been shown to increase the risk of cognitive decline in healthy older women, but not men [1]. The elevated risk for the conversion of MCI to AD in APOE4 carriers is also higher in women than men [1]. However, the difference between men and women in their association of APOE4 with CAA was not evident in our cohort likely because of insufficient sample size for detecting a sex-dependent interaction. Further studies are desired to examine if there is a sex-dependent association of APOE4 with AD and CAA pathogenesis. Although APOE2 is protective against AD development, it correlates with an increased risk for CAA-related hemorrhage [7, 33]. However, the potential association of APOE2 with CAA risk is somewhat controversial [32, 36] likely because the high risk associated with APOE4 [36] and the low frequency of APOE2 genotype making assessment relatively difficult. Our current study also did not detect a statistically significant effect of APOE ε2/ε3 genotype on average CAA score compared to APOE ε3/ε3 (supplemental Figure 1). Nonetheless, a better understanding on the contribution of APOE2 to CAA risk and CAA-related cerebrovascular pathology is critical for addressing the roles of sex and apoE isoforms in these pathogenic conditions.

Whereas Aβ42 is the major species that accumulates in senile plaques [29], we found that both soluble and insoluble levels of Aβ40, but not Aβ42, were increased in the temporal cortex from AD patients with severe CAA compared to those without CAA (Fig. 4), which is consistent with previous reports [17, 45]. Furthermore, the ratio of Aβ40/Aβ42 was elevated in AD brains with severe CAA. Since Aβ40 is diffusible compared to Aβ42, more Aβ40 is predicted to traverse through cerebrovasculature than Aβ42. Thus, increased levels of parenchymal Aβ40 might promote its diffusion into perivascular regions, resulting in CAA formation. On the other hand, the ratio of Aβ40/Aβ42 was negatively associated with male sex despite the fact that CAA was exacerbated in male AD patients. Thus, increasing Aβ40 levels and the Aβ40/Aβ42 ratio are not the only factors facilitating CAA formation during AD pathogenesis. In contrast, it is possible that cerebrovascular damage through CAA and/or CAA progression, per se, predominantly interferes with clearance of Aβ40, which causes Aβ40 accumulation in the brain parenchyma. The outcome of a current phase II clinical trial in probable CAA patients will assess the therapeutic efficacy of an Aβ40-targeted immunotherapy using Ponezumab (PF-04360365) [24] (ClinicalTrials.gov Identifier number: NCT01821118), and will hopefully shed light on the pathogenic relationship between CAA and Aβ40.

Additionally, we found that insoluble apoE levels were elevated in CAA cases, which were positively correlated with insoluble Aβ levels. Of note, there was a strong association between apoE levels and Aβ40 levels for those with severe CAA, but not for those without CAA. The same pattern was not apparent for Aβ42, with a similar degree of association between apoE levels and Aβ40 levels for those with and without CAA. These results indicate that apoE may co-deposit with Aβ40 in CAA, which could contribute to the pathogenic process. In AD cases without CAA, the interaction between apoE and Aβ40 might not be as prevalent. It is also possible that the clearance pathways for Aβ40 are more efficient in the absence of CAA, thereby lowering its chance to encounter apoE. Further studies are needed to determine if apoE promotes the aggregation of Aβ40 or whether apoE is merely captured by Aβ40 aggregates during CAA development.

APOE4 has been shown to be strongly associated with increased Aβ depositions in both senile plaques and CAA [40]. Given that apoE4 accelerates Aβ aggregation and suppresses Aβ clearance as compared to other apoE isoforms [22], the exacerbated effects of APOE4 on CAA are also predicted to be mediated by Aβ. In a transgenic amyloid AD model with apoE4-targeted replacement (TR) background in which mice express human apoE4, the brain ratio of Aβ40/Aβ42 is elevated in soluble fractions prior to Aβ deposition compared to brain fractions from apoE3-TR mice, leading to exacerbated CAA formation in aged mice [15]. Since we found that APOE4 is correlated with an increase in both detergent-soluble and -insoluble Aβ40 levels, apoE4 may directly influence the metabolism of Aβ40 by altering its clearance and/or brain distribution. In addition, APOE4 is significantly associated with increased insoluble apoE levels. Thus, the binding properties of apoE4 may also be involved in the mechanism underlying the risk of CAA in APOE4 carriers. On the other hand, APOE4 may also contribute to CAA development through Aβ-independent pathways. For example, apoE4 negatively affects brain homeostasis by influencing neuronal functions, lipid/glucose metabolism, cerebrovascular system and neuroinflammation compared to apoE3 [8, 21]. In particular, reduced cerebral blood flow has been described in elderly APOE4 carriers relative to non-carriers [14, 47]. Furthermore, apoE4-TR mice exhibit dysregulated blood-brain barrier integrity and reduced brain microcirculation compared to apoE2-TR and apoE3-TR mice [6]. Importantly, the detection of regional cerebral hypoperfusion through neuroimaging techniques can identify individuals at risk for AD since cerebral hypoperfusion precedes hypometabolism, cognitive decline and neurodegeneration in AD [11]. Thus, apoE4-related cerebrovascular dysfunction may trigger the pathogenesis of both AD and CAA. Since the cerebrovascular system critically mediates brain Aβ clearance [51], CAA-mediated disturbances may facilitate further Aβ deposition.

In summary, using a large cohort (n=428) of AD patients, our current study has shown that the majority (88.4%) of AD cases has concurrent CAA, where APOE4 and sex significantly influence its severity. One limitation of our study is that capillary CAA was not specifically addressed. Because AD cases with capillary CAA are likely associated with APOE4 to a greater extent than those lacking capillary CAA [46], future studies should be designed by addressing CAA types as well as its severity. CAA data was obtained from our databased reports, which may have suffered from ‘evaluation drift’ over the years. The CAA data, however, was scored by a single neuropathologist (DWD) on the same microscope limiting variability to some extent. Our data further demonstrate that severe CAA in AD is more frequent in men than women, regardless of APOE4 genotype. In addition, Aβ40 levels and Aβ40/Aβ42 ratio in temporal cortex are elevated in CAA cases and APOE4 carriers. Intriguingly, the association between apoE levels and Aβ40 and Aβ42 levels are differently influenced by CAA status. Despite increasing evidence that shows sex-dependent differences in AD [9], the interaction between APOE and sex has received limited attention in studies of CAA and cerebrovascular diseases. Future studies that focus on sex, apoE and the cerebrovascular system may discover new genetic or environmental factors for CAA that would aid in better understanding of the mechanisms underlying pathogenesis of AD and CAA.