Figure 1. Description of sample sizes from the original National Alzheimer Coordinating Center (NACC) 2011 data set to our final sample of 890 hypertensive participants. ARBs indicates angiotensin receptor blockers; NP, neuropathology; UDS, Uniform Data Set.
Figure 2. Graphs of Alzheimer disease neuropathologic scores (A and B), vascular brain injury measures (C), and atherosclerosis/arteriosclerosis (D) in the 3 groups (those treated with angiotensin receptor blockers [ARBs], other antihypertensive medications, or no antihypertensive medications). CERAD indicates the Consortium to Establish a Registry of Alzheimer Disease.
Hajjar I, Brown L, Mack WJ, Chui H. Impact of Angiotensin Receptor Blockers on Alzheimer Disease Neuropathology in a Large Brain Autopsy Series. Arch Neurol. 2012;69(12):1632-1638. doi:10.1001/archneurol.2012.1010
Author Affiliations: Division of Geriatric, Hospital, and General Internal Medicine, Departments of Medicine (Dr Hajjar and Ms Brown), Preventive Medicine (Dr Mack), and Neurology (Dr Chui), University of Southern California, Los Angeles.
Background Angiotensin II may be involved in amyloid metabolism in the brain. Angiotensin receptor blockers (ARBs) may also prevent cognitive decline.
Objective To evaluate the impact of treatment with ARBs on the neuropathology of Alzheimer disease (AD) in the National Alzheimer Coordinating Center database, which includes aggregated data and brain autopsies from 29 AD centers throughout the United States.
Design Multiple logistic regression was used to compare the pathologic findings in hypertensive subjects taking ARBs with those taking other antihypertensive treatments as well as with hypertensive subjects who did not receive antihypertensive medications.
Setting Neuropathologic data included neuritic plaque and neurofibrillary tangle measures and vascular injury markers.
Patients Data were collected from participants who were self-referred or provider-referred and included those with and without cognitive disorders. Our sample included only hypertensive participants and excluded cognitively and neuropathologically normal participants (N = 890; mean age at death, 81 years [range, 39-107 years]; 43% women; 94% white).
Results Participants with or without AD who were treated with ARBs showed less amyloid deposition markers compared with those treated with other antihypertensive medications (lower Consortium to Establish a Registry of Alzheimer Disease score: odds ratio, 0.47, 95% CI, 0.27-0.81; Alzheimer Disease and Related Disorders Association score: odds ratio, 0.43, 95% CI, 0.21-0.91; Braak and Braak stage: odds ratio, 0.52, 95% CI, 0.31-0.85; neuritic plaques: odds ratio, 0.59, 95% CI, 0.37-0.96). They also had less AD-related pathology compared with untreated hypertensive subjects. Participants who received ARBs were more likely to have had a stroke; hence, they had more frequent pathologic evidence of large vessel infarct and hemorrhage.
Conclusion Treatment with ARBs is associated with less AD-related pathology on autopsy evaluations. The effect of ARBs on cognitive decline in those with dementia or AD needs further investigation.
Observational studies suggest that angiotensin receptor blockers (ARBs) may have superior effects on cognitive function compared with other antihypertensive medications. In a recent large cohort of American veterans, Li et al1 reported that treatment with ARBs was associated with lower risks for dementia and Alzheimer disease (AD) compared with other antihypertensive agents, including angiotensin-converting enzyme inhibitors (ACEIs). The association between concurrent elevation in blood pressure and AD is not consistent.2- 4 This may be in part related to the inconsistent processes by which AD was adjudicated and confirmed. A pathologic examination is the most accurate process to confirm AD but is not feasible in many observational studies and clinical trials.
Alzheimer disease is related to increased amyloid deposition in the brain.5 The impact of antihypertensive medications on amyloid metabolism is not known. Animal studies have suggested that angiotensin II promotes production of amyloid centrally. Therefore, we hypothesized that treatment with ARBs would be associated with decreased amyloid deposition.6 Animal studies have also suggested that ARBs may decrease amyloid-β(Aβ) oligomerization.7 Additionally, a recent study in APP/PS1 mice showed that treatment with intranasal losartan decreased Aβ plaques by 3.7-fold compared with saline.8 In humans, it is not known whether ARBs are also related to decreased AD pathology and amyloid deposition.
Vascular dementia and related vascular brain injury are also common in hypertension, and the impact of antihypertensive medications to treat these conditions is unknown.9,10 Few longitudinal studies have included brain autopsy accompanied with medication and blood pressure data. The National Alzheimer Coordinating Center includes brain autopsy series; therefore, we aimed to investigate the impact of treatment with ARBs on the neuropathologic changes related to amyloid and vascular pathology.
Data from the National Alzheimer Coordinating Center, which included aggregate data from 29 AD centers across the United States collected between September 2005 and June 2011, were used for this analysis (details described elsewhere).11 Participants were either provider-referred or self-referred and were with or without cognitive disorders. Informed consent was obtained from all subjects including an optional brain autopsy. Data were collected annually using the National Alzheimer Coordinating Center's Uniform Data Set data collection protocol and included demographic information, education level, marital status, tobacco use, family history of dementia, and body mass index (calculated as weight in kilograms divided by height in meters squared). Medical information included medical history (ie, cardiovascular disease, stroke/transient ischemic attack, Parkinson disease, brain trauma, hypercholesterolemia, hypertension, diabetes mellitus, thyroid disease, and B12 deficiency) and current medications (taken within the previous 2 weeks including antihypertensive, lipid-lowering, anticoagulant [such as warfarin or aspirin], antidiabetic, antidepressive, and antipsychotic medications). Neuropsychiatric data elements included Geriatric Depression Scale; Mini-Mental State Examination; Trail Making Test; logical memory test; digit span test; and Clinical Dementia Rating scale (a global measure of dementia stage) scores.11 If more than 1 evaluation was done, we used the last one before death in our analyses. Data collection and recording were performed by trained personnel and clinicians. Cognitive disorder diagnoses were made by either a consensus team or a physician who performed a detailed examination (http://www.alz.washington.edu/WEB/study-pop.html). Clinicians reviewed all available information and provided a clinical diagnosis for each participant, categorizing them as having normal cognition, mild cognitive impairment, AD, vascular dementia, mixed dementia, or other types of dementia. A pathologic diagnosis was ascertained by the trained neuropathologist, who classified participants as having normal pathology, AD, vascular dementia, mixed dementia, or other dementia. We restricted this analysis to hypertensive subjects with available brain autopsy data, medication information, and blood pressure measurements from at least 1 visit. We excluded those with normal cognitive function as assessed by the clinician on the last assessment as well as those with a pathologic diagnosis of normal brain as assessed by the neuropathologist.
Subjects meeting criteria for hypertension included self-reported hypertension, systolic blood pressure of 140 mm Hg or greater or diastolic blood pressure of 90 mm Hg or greater, or receiving antihypertensive medication during at least 1 visit. Antihypertensive medications reported at each visit were identified and categorized into their broad corresponding classes: ARBs, ACEIs, β-blockers, calcium channel blockers, diuretics, anti-adrenergic agents, and aldosterone antagonists. Participants receiving more than 1 antihypertensive medication were categorized into more than 1 class. Participants receiving ARBs at any visit were grouped into the ARBs group, including those on more than 1 class. Our comparison groups were those treated with antihypertensive medications other than ARBs and untreated hypertensive subjects.
Neuropathologic data were collected using a standard protocol by trained neuropathologists. Alzheimer disease–related pathology was assessed using several measures. (1) The Consortium to Establish a Registry of Alzheimer Disease (CERAD) score measured neuritic plaque counts, with adjustments for age and clinical history in the most densely populated 1-sq-mm area of each brain region. The neuritic plaque count was measured in the most severely affected cortical region and was rated as none, sparse, moderate, or frequent.12 Neuritic plaques were identified as plaques with argyrophilic, thioflavin-S-positive, or tau-positive dystrophic neuritis with or without dense amyloid cores. (2) Khachaturian criteria (National Institute on Aging) measured the density of the neocortical plaques per unit area, corrected for age (age <50 years: neurofibrillary tangles and senile or neuritic plaques in the neocortex, >2 to 5/field; 50-65 years: tangles/plaques, ≥8/field; 66-75 years: tangles/plaques, >10/field; >75 years: tangles may or may not be found and senile plaques, >15/field.13 (3) The Braak and Braak neuritic plaque score was classified as either no evidence of neurofibrillary degeneration; stages 1 and 2 (the neurofibrillary tangles preferentially involve the entorhinal-perirhinal cortex); stages 3 and 4 (tangles also accumulate in hippocampus and other limbic regions with limited neocortical involvement); stage 5 (neurofibrillary changes occur in association cortices); or stage 6 (neurofibrillary changes occur in primary sensory cortex).14,15
Vascular-related measures included the presence or absence of large artery cerebral infarcts (infarcts >1 cm in diameter in the distribution of large-sized and medium-sized meningocerebral vessels) and microinfarcts or lacunes (≤1 cm detected in the cortical areas microscopically). Large infarcts and microinfarcts were included regardless of the histologic age and included acute and chronic cystic lesions. Cerebral hemorrhages were also reported, irrespective of size and brain region. Subcortical arteriosclerotic leukoencephalopathy was defined as multifocal or diffuse white matter pathology due to arteriosclerotic small vessel disease.16 Measures of the severity of atherosclerotic disease of the circle of Willis and of arteriosclerosis in the small parenchymal and/or leptomeningeal vessels were scored subjectively as none, mild, moderate, or severe.
We classified participants into those who were treated with ARBs, those treated with other antihypertensive medications, and those who were not treated with antihypertensive medications during the follow-up. We used χ2 or analysis of variance to compare these 3 groups on demographic, social, cognitive, medication, and clinical characteristics including clinical diagnosis for the cognitive disorder. Neuropathologic outcomes were grouped into dichotomous dependent variables (absent vs present) to simplify the interpretation/presentation of results. We used multiple logistic regression with the primary independent variable being treated with ARBs compared with other antihypertensive therapies or untreated hypertensive subjects. To address the issue of treatment with multiple antihypertensive medications, we initially restricted our sample to those treated with only 1 class of antihypertensive medication; because the results did not significantly change, we present the results for the full sample. All models were adjusted for covariates that may impact cognitive function: age at death, sex, body mass index, clinical stroke, apolipoprotein E (APOE) ϵ4 alleles (included as with or without the ϵ4 allele), and systolic blood pressure (mean during the evaluation period). All analyses were conducted using SAS version 9.2 (SAS Institute).
As shown in Figure 1, of the 1275 subjects who had brain autopsy and available pathologic and clinical data, 339 (27%) did not have hypertension and 46 (4%) were clinically and neuropathologically diagnosed as normal and were excluded, leading to a final sample size of 890. Among our analysis sample, 710 (80%) were treated with antihypertensive medications (133 [15%] were treated with ARBs and 577 [64%] were treated with other antihypertensive agents) and 180 (20%) were untreated. Of the 133 who reported using ARBs, 56% reported using ARBs during only 1 visit, 27% during 2 visits, and 17% during more than 2 visits. The median age at death was 83 years (25th percentile: 75 years; 75th percentile: 90 years). Of the total sample, 94% were white and 43% were women. Table 1 presents the demographic, cognitive, and clinical characteristics of the 3 groups. When comparing all 3 groups, those who received ARBs were slightly older at death (P < .001); had significantly higher body mass index (P = .02); and had higher scores on the Mini-Mental State Examination (P = .001) and logical memory test (P = .01) and a lower score on the Clinical Dementia Rating scale (P = .01) relative to the other 2 groups. Median time between enrollment and death was 2 years (25th percentile: 1 year; 75th percentile: 3 years) and the median number of visits was 2 (range, 1-6). Participants who were treated with ARBs were less likely to have a clinical diagnosis of AD compared with those treated with other antihypertensive medications or untreated hypertensive participants (P = .001).
As shown in Table 2, participants exposed to ARBs were less likely to have received a neuropathologic diagnosis of AD using the various pathologic criteria: CERAD (P = .005), Alzheimer Disease and Related Disorders Association score (P = .04), and the overall neuropathologist diagnosis (P = .06). After adjusting for covariates, the association between ARBs and a lower likelihood of AD diagnosis remained significant, as shown in Table 3. Compared with other antihypertensive medications, treatment with ARBs was associated with a 32% to 35% lower likelihood of AD diagnosis, depending on the criteria used. This was also true when we compared subjects treated with ARBs vs untreated participants, as shown in Table 3.
Those treated with ARBs, with or without a diagnosis of AD, also had less amyloid deposition compared with both untreated participants and those treated with non-ARB antihypertensive medications. As shown in Figure 2A-B, both Braak and Braak score (P < .001) and the CERAD neuritic plaque count (P = .003) were lower in the ARB group compared with untreated participants or those treated with other antihypertensive agents. Odds ratios remained significant in the multivariate models, as shown in Table 3, even after controlling for APOE genotype.
Participants who took ARBs were more likely to have had a stroke (P = .03). Consistent with this observation, those taking ARBs were more likely to show large artery infarcts and hemorrhage on autopsy, as shown in Figure 2C-D. Angiotensin receptor blockers–treated participants also had greater degrees of atherosclerosis and arteriolosclerosis at autopsy. However, these associations were not significant after adjusting for age, sex, systolic blood pressure, body mass index, stroke, APOE ϵ4 alleles, and prior use of anticoagulants (Table 3), except for higher prevalences of large artery infarct, hemorrhage, and arteriolosclerosis.
When we compared those who took ARBs vs those who took ACEIs (n = 362), ARB therapy was associated with lower amyloid deposition (CERAD: adjusted odds ratio, 0.43; 95% CI, 0.21-0.86; P = .02) and neuritic plaque count (odds ratio, 0.50; 95% CI, 0.28-0.89; P = .02) compared with ACEIs. Finally, there was no significant difference between untreated hypertensive participants and those treated with non-ARB antihypertensive medications in either the vascular-related pathology or AD pathology, except for atherosclerosis as shown in Table 4.
These autopsy data aggregated across the US AD centers demonstrate that hypertensive individuals treated with ARBs are less likely to have a neuropathologic diagnosis of AD post mortem, independent of blood pressure, APOE status, and prior strokes. Angiotensin receptor blocker therapy was also associated with fewer amyloid plaques compared with treatment with other antihypertensive medications or with no antihypertensive medications in those with or without AD. Furthermore, ARBs were associated with less AD pathology compared with ACEIs.
In a study of 291 post mortem brains, Hoffman et al17 showed that antihypertensive therapy was associated with lower AD neuropathology, even compared with normotensive individuals. In this current study, to our knowledge, we provide the first autopsy evidence suggesting that ARBs, rather than other classes of antihypertensive medications, may be associated with reduced amyloid accumulation and AD-related pathologic changes. This study provides pathologic confirmation to the observation that ARB use is associated with lower AD risk.1
Animal studies have demonstrated that ARBs are associated with decreased amyloid deposition in the brain compared with other antihypertensive agents.16 Using high-throughput drug screening in AD mouse models, Wang et al7 showed that only valsartan treatment attenuated the oligomerization of Aβ peptides into the high molecular weight peptides, which are known to be associated with cognitive deterioration. None of the other antihypertensive classes had this specific positive effect. More recently, Danielyan et al8 showed that treatment of APP/PS1 mice with intranasal losartan decreased Aβ plaques by 3.7-fold. To our knowledge, our study provides the first clinical translational evidence for an association of ARBs with reduced amyloid accumulation in humans.
The association between ARBs and decreased AD pathology and amyloid accumulation is unique. None of the other antihypertensive medications, including ACEIs, showed a similar association. Prior evidence suggests that ACEIs significantly inhibits fibril formation and aggregation of Aβ in a dose-dependent manner, suggesting that inhibiting ACEIs may lead to an increase in Aβ deposition.18 In this analysis, we observed greater amyloid content in those treated with ACEIs compared with those treated with ARBs, suggesting that ARBs may have a preferential positive effect on AD pathology. These are in agreement with observational data suggesting that ARBs are superior to ACEIs in preventing AD and other related dementias. To our knowledge, the mechanisms by which ARBs may decrease amyloid accumulation have not been described, but ARB treatment may reduce total Aβ content in the brain in part by facilitating membrane-associated insulin degrading enzyme–mediated proteolytic cleavage of Aβ peptides.7
In contrast to the positive associations of ARBs with amyloid deposition and AD-related pathology, we observed greater vascular-related pathology in the ARB group. This is likely related to confounding by indication where ARBs were more likely to be used in those with increased vascular risk. This is further confirmed by the greater prevalence of stroke and cardiovascular disease compared with those treated with other antihypertensive medications and those who were untreated. Adjusting for covariates including anticoagulation accounted for some of the observed associations between ARBs and increased vascular neuropathology. Nevertheless, caution should be used when interpreting our results owing to the observational nature of this study.
A main limitation of this study was that it included a primarily highly educated white sample who volunteered for this observational study. This limits the generalizability to different populations. Another limitation is the possibility that ARBs were more likely to be prescribed for those with non-AD diagnosis, leading to a greater likelihood of diagnosis of vascular dementia and mixed dementia in the ARB group. Although this is a possible explanation for our findings, the fact that ACEIs, which are commonly used in patients with high vascular disease, did not show this inverse association, making this less likely. Although there may be an intraclass variation in the effects of ARBs on pathology, we were unable to perform subgroup analyses owing to the relatively small number of participants taking ARBs. Finally, owing to the variable number of follow-up visits and that most reported taking ARBs during 1 or 2 visits, we were unable to test whether the duration of the exposure to ARBs had an impact on neuropathology.
The implications of our results are that ARBs may offer an advantage over other antihypertensive agents in regard to AD risk and pathology. The effect of ARBs in AD under randomized conditions needs further investigation. Furthermore, pre mortem assessments of amyloid in the brain should be considered in future studies related to antihypertensive agents and cognitive function, as recent evidence has suggested an increase in amyloid in healthy older adults measured by Pittsburgh Compound-B positron emission tomography.19
In conclusion, this autopsy study suggests that treatment with ARBs is associated with less amyloid accumulation and AD-related pathology independent of other AD risk factors. To our knowledge, it is the first human evidence to suggest that treatment with ARBs may have a selective beneficial effect on amyloid metabolism. Prospective studies are needed to investigate whether ARBs may have an effect on cognitive decline in those with dementia or AD.
Correspondence: Ihab Hajjar, University of Southern California, 2020 Zonal Ave, IRD 320, Los Angeles, CA 90033 (email@example.com).
Accepted for Publication: March 26, 2012.
Published Online: September 10, 2012. doi:10.1001/archneurol.2012.1010
Author Contributions:Study concept and design: Hajjar. Acquisition of data: Brown and Mack. Analysis and interpretation of data: Hajjar, Brown, Mack, and Chui. Drafting of the manuscript: Hajjar, Brown, and Mack. Critical revision of the manuscript for important intellectual content: Hajjar, Mack, and Chui. Statistical analysis: Hajjar, Brown, and Mack. Obtained funding: Hajjar. Administrative, technical, and material support: Brown and Chui. Study supervision: Hajjar and Mack.
Financial Disclosure: None reported.
Funding/Support: The National Alzheimer's Coordinating Center database is funded by grant U01 AG016976 and enables the center's ability to deliver data to researchers. Dr Hajjar's work is supported by grant 1 K23 AG030057 from the National Institute on Aging. This work was also supported by grant P50 AG05142 from the National Institute on Aging to Drs Mack and Chui, as well as grant P01 AG12435 to Dr Chui.