Neurovascular and cognitive dysfunction in hypertension: epidemiology, pathobiology and treatment

Abstract

Hypertension has emerged as a leading cause of age-related cognitive impairment. Long known to be associated with dementia caused by vascular factors, hypertension has more recently been linked also to Alzheimer’s disease, the major cause of dementia in older people. Thus, while midlife hypertension is a risk factor for late-life dementia, hypertension may also promote the neurodegenerative pathology underlying Alzheimer’s disease. The mechanistic bases of these harmful effects remain to be established. Hypertension is well known to alter in the structure and function of cerebral blood vessels, but how these cerebrovascular effects lead to cognitive impairment and promote Alzheimer disease pathology is not well understood. Furthermore, critical questions also concern whether treatment of hypertension prevents cognitive impairment, the blood pressure threshold for treatment, and the anti-hypertensive agents to be used. Recent advances in neurovascular biology, epidemiology, brain imaging, and biomarker development have started to provide new insights into these critical issues. In this review, we will examine the progress made to date and, after a critical evaluation of the evidence, we will highlight questions still outstanding and seek to provide a path forward for future studies.

Introduction

Elevated arterial blood pressure (BP) contributes significantly to the global burden of disease, its harmful effects on the brain being a major culprit.¹Well established as the leading risk factor for stroke, hypertension has also emerged as a pathogenic factor in cognitive impairment with a vascular basis as well as Alzheimer’s disease (AD), the major causes of dementia in the elderly.² The harmful effects of hypertension on cognitive function were already noted at the end of the 19^th century, in patients who later developed massive cerebral hemorrhages, but systematic investigations on the phenomenon did not appear until the 1950-60s. In particular, pioneering observations on the neuropsychological performance of air traffic controllers with hypertension by Spieth³ and the longitudinal study on cognitive function in hypertensives of Wilkie and Eisdorfer⁴ provided initial evidence that hypertension deteriorates cognition. The introduction of methods to measure cerebral blood flow (CBF) let to the discovery that individuals with hypertension have increased cerebrovascular resistance, which correlated with the degree of hypertensive retinopathy⁵, substantiating earlier neuropathological findings linking cerebrovascular damage with the effects of hypertension on brain health. However, the prevailing belief at the time was that the elevation in BP was a beneficial compensatory mechanism aimed at maintaining normal cerebral perfusion, and attempts to lower BP were considered dangerous.⁶

Subsequent population and prospective studies clearly established the relationship between elevated BP and cognitive impairment,⁷ and unveiled an unanticipated link between hypertension and AD⁸. With the development of safe and effective antihypertensive drugs and their widespread use, over the past several decades a dramatic reduction in the morbidity and mortality caused by cardiovascular pathologies, stroke in particular, was observed in a number of large clinical trials⁹. Surprisingly, however, similar beneficial effects on cognition were not uniformly noted¹⁰. Owing to the rapid rise in the prevalence of age-related dementia worldwide, hypertension has received increasing attention as a treatable condition that, if properly controlled, could potentially stave off the onset of the cognitive deficits. Consequently, an increasing number of studies have focused on the mechanisms by which hypertension impairs cognitive function, on the temporal relationships between hypertension and cognitive deficits over the life course, and on the ideal BP targets to achieve in order to minimize its impact on the brain. While this body of new knowledge has provided a deeper insight into the relationship between hypertension and the brain, new questions have emerged that need to be addressed.

The present review seeks to provide an assessment of the state-of-the-art of the effects of hypertension on cognitive function, on the underlying structural and functional alterations and their mechanisms, and on potential preventive and therapeutic strategies to minimize them. After a critical review of the evidence, we will attempt to identify outstanding questions and knowledge gaps that remain to be addressed.

Impact of hypertension on cognition

A growing body of evidence supports the role of hypertension as a risk factor for adverse cognitive outcomes. This evidence primarily comes from epidemiologic studies, which consider hypertension in communities as a risk factor for a range of adverse outcomes including cognitive decline (worsening of cognitive function over years to decades, steeper than expected due to age alone), mild cognitive impairment (MCI) (reduced function in memory, thinking, and other cognitive domains, but not impacting daily functioning), and dementia (impairments in cognition, including memory and other cognitive domains, but with adverse impacts on daily functioning). Studies considering cognitive decline allow evaluation of change, which captures more subtle cases of cognitive dysfunction, without requiring frank thresholds for dementia or MCI to be met, but also may be a less confounded measure of cognitive status, allowing a more pure evaluation of the association between hypertension and cognition.¹¹ In contrast, studies considering MCI and dementia as outcomes have a stronger public health message, as these concepts are easier to define and understand, and cases of preventable disease are easily embraced, as opposed to changes in steepness of cognitive slopes. For hypertension, evidence from epidemiologic studies supports all of these outcomes: hypertension is clearly associated with steeper cognitive decline,^{12, 13} poor cognitive performance,¹⁴ as well as incident MCI¹⁵ and dementia.^16–19 Furthermore, among adults with MCI, elevated BP is associated with increased progression and worsening of cognition.²⁰ Prehypertension, previously considered as a systolic BP (SBP) <140mmHg, has even been reported as a risk factor for dementia in several studies, including the Atherosclerosis Risk in Communities (ARIC) cohort,¹⁸ in which both midlife hypertension and prehypertension conferred a similar risk for dementia, at about 40% higher than among individuals with normotension. New definitions of hypertension now consider SBP in this range consistent with stage 1 hypertension.²¹

Although epidemiologic data emphasizes the importance of hypertension in the development of stroke, the impact on cognitive outcomes appears to be independent of stroke. This is the case when stroke is defined clinically, with similar associations when individuals with incident stroke are excluded from analyses on cognition,^{7, 18} but fewer studies have systematically excluded subclinical stroke or small vessel disease (SVD)(including white matter lesions, lacunes and enlarged perivascular spaces), important mediators of the impact of hypertension on cognitive outcomes.

The primary focus of this review is hypertension and its components, systolic (SBP) and diastolic blood pressure (DBP), but different components or aspects of BP may also play an important role in dementia. Although pulse pressure was not significantly associated with dementia in the Honolulu-Asia Aging Study (HAAS),¹⁷ it was associated with cognitive decline among APOE ε4 carriers in the Framingham Offspring Study.²² Most human studies do not differentiate between primary and secondary causes of hypertension; although most studies primarily enroll individuals with essential hypertension, rodent models considering several mechanisms of hypertension (including modeling renovascular hypertension along with spontaneously hypertensive rats) have demonstrated similar cognitive deficits and white matter hyperintensities. Orthostatic hypotension, which occurs more frequently in individuals with hypertension, was associated with a 54% higher risk (95% CI 1.20-1.97), with relationships persisting after censoring for ischemic stroke.²³ In Uppsala, Sweden, using ambulatory BP monitoring, nondipping circadian BP patterns as well as persistently high 24-hour BP (at age 70) were associated with worse cognitive performance.²⁴ BP variability has also been described in several studies as an important risk factor for reduced cognitive function.²⁵

Age-dependent risk (midlife vs late-life):

Hypertension’s impact on late-life cognitive outcomes appears greatest when considered in middle age. Several studies have demonstrated cognitive decline in the 70’s or older when BP is elevated in the 4th or 5^th decade of life. In the HAAS, risk of reduced cognitive function in late-life increased as midlife SBP increased,¹⁴ and elevated SBP (>140 mm Hg) in midlife was associated with a 1.77-times higher risk of dementia at ages 71-93.¹⁷ Furthermore in HAAS, elevated midlife BP (DBP, especially) and decrease in plasma amyloid-β (Aβ) 1-40 interacted to increase risk of late-life dementia.²⁶ In the National Heart, Lung, and Blood Institute (NHLBI) Twin study, higher midlife SBP was associated with steeper 10-year cognitive decline.¹³ In the ARIC cohort, hypertension in midlife was associated with steeper 20-year cognitive decline on three repeated cognitive tests, without similar associations with late-life hypertension, which was not associated with cognitive decline over the preceding 20 years.¹² Other studies have failed to find an association between BP and dementia or cognitive decline,²⁷ but this may be because they primarily focused on BP in older adults and with a shorter followup period, without evaluation of hypertension during its maximal risk period (middle age).

Some studies have suggested a U-shaped relationship with BP and cognition, but these studies seem isolated to studies of older adults, further reflecting the potential detrimental effect of midlife hypertension but perhaps even a protective effect of late-life hypertension. There is no convincing evidence that low BP, in younger people, is associated with bad cognitive outcomes, but several studies do suggest worse outcomes in older individuals with low BP. In one study, cognitive performance was worst among individuals with either SBP<130 mm Hg or with elevated SBP (≥160 mm Hg), compared to persons at 130-139 mm Hg, but this cohort consisted of adults aged 65-102.²⁸ In an analysis including data from the Rotterdam study and the Göteborg H-70 study, hypertension in old age appeared protective: elderly adults had an inverse association between BP and dementia, with a reduced relative risk for dementia of 0.93 (95% CI 0.88-0.99) per 10 mm Hg higher SBP – but only among elderly adults taking antihypertensive medications.²⁹ In the Bronx Aging Study, adults over age 75 with SBP between 140 and 179 mm Hg had a lower risk for AD (HR 0.55, 95% CI 0.32-0.96), and risk for dementia was especially high in individuals with low DBP at baseline (age>75).³⁰ A U-shaped relationship between BP and cognition was also reported in the Baltimore Longitudinal Study on Aging, among adults over age 80 who had decline at both extremes of BP.³¹

Although most observed associations are with dementia broadly, several studies have attempted to capture dementia subtype. Given the lack of gold standard for definitively subtyping dementia, it is certainly plausible that these etiologic classifications actually represent mixed etiology, so still may point to a role of hypertension in dementias of other types. In the Hisayama study, midlife (more so than late-life) hypertension was associated with vascular dementia, but not AD,³² with similar findings in the Adult Health Study in Hiroshima, which evaluated SBP in association with subsequent 25-30 year development of vascular dementia.³³ In a large Finnish study, midlife SBP≥160 mm Hg was associated with an over two-fold higher risk of AD, independent of other vascular and demographic risk factors.¹⁹

Studies using imaging surrogates for cognitive or dementia status also support the importance of midlife hypertension. In the NHLBI Twin study, midlife SBP was associated with not only more white matter hyperintensities in later life but also smaller brain parenchymal volumes.¹³ Furthermore, BP appears particularly related to atrophy in regions highly relevant for AD (the amygdala and the hippocampus, in this case).³⁴ Other studies have also reported strong associations with hypertension for SVD, including white matter hyperintensities³⁵ and their progression.³⁶

Hypertension over the life course:

Although the above studies emphasize the impact of midlife hypertension as a risk factor for cognitive decline and dementia, change in BP may be relevant as well. As discussed later in this review, BP changes are particularly important when considering shifts that occur in an individual’s autoregulatory curve in response to that individual’s baseline value. In the HAAS study, men who developed dementia had a gradual increase in SBP from mid- to late-life, with a subsequent decrease of 1.36 mm Hg/year in late life; in addition, of the 109 participants in this subcohort who developed dementia, the majority (58%) had had a recent fall in SBP over the preceding 6 years.³⁷ Similar patterns were shown in the Göteborg Longitudinal Population Study, where 70 year olds with elevated BP had more dementia at age 79-85, but many also experienced decline in BP in the years more proximal to a dementia diagnosis.³⁸ In the Framingham Offspring Cohort, elevated midlife BP was an important risk factor for dementia, but a steep decline in BP from mid-to- late-life was associated with a 2.4-times higher risk of dementia in later-life (95% CI 1.4-4.2).³⁹

One complicating factor in studies of BP trajectories is the fact that BP often falls in response to incident cognitive impairment and dementia, making it difficult to determine whether the apparent relationship with recent decline in BP is actually due to reverse causation. Studies support lower and declining BP in elderly individuals with lower cognitive performance, some of whom went on to develop dementia,⁴⁰ consistent with this concept of reverse causation possibly playing a role. This may also explain the above-cited Rotterdam/ Göteborg H-70 result,²⁹ where hypertension in old age appeared “protective”. Studies considering brain volumes as an outcome have also found evidence that a late-life decline in BP may be a marker for brain volume loss, with more cortical atrophy in individuals with a drop in DBP of more than 10 mm Hg over 20 years in one study.⁴¹ In the ARIC study, the pattern of midlife hypertension with subsequent late-life hypotension was associated with smaller brain volumes in regions typically affected in AD.⁴²

Persistently elevated BP, from midlife and continuing through later life, does appear to be consistently associated with worse cognitive outcomes: Yaffe et al., evaluated cumulative SBP and DBP level over 25 years, finding worse cognitive performance on several tests among individuals with higher area-under-the-curves based on cumulative BP.⁴³ However, using a Mendelian Randomization design, genetically-predicted systolic hypertension was associated with lower risk of AD, possibly indicative of subsequent changes in lifestyle or management in response to this elevated risk.⁴⁴

Cognitive domains affected:

Although several studies have suggested that the impact of hypertension on cognition is global, this is mostly the case because not all studies broadly consider several distinct cognitive domains, or because the most impacted domains in hypertension-associated brain disease are also represented in global cognitive measures. When more detailed neuropsychological batteries are performed and hypertension-specific effects are considered and compared, the biggest impact of hypertension appears to be on executive function,¹² motor speed, and attention,⁴⁵ which are classically considered domains involved with subcortical disease, such as typical vascular disease or pure vascular dementia.⁴⁶ Even scores on the Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA), two global cognitive measures, although lower in individuals with more vascular risk factors, including hypertension, are primarily driven by impairments in items related to attention (for the MMSE) and visuo-executive function (for lower scoring individuals on the MoCA).⁴⁷

The role of memory impairments in hypertension-related dementias is complicated by the likely overlap between etiologies and the high prevalence of mixed dementias. Although memory is more typically involved in AD-related cognitive impairments than in vascular types of dementia, it has been noted to have impairments in individuals with a combination of hypertension and elevated brain amyloid by PiB-PET.⁴⁸ However, many of these studies of domain-specific cognitive deficits are influenced by misdiagnosis or difficulty determining etiology during life. Using a study neuropathologic data, domains did not differ by etiology as might be expected: individuals with pure AD neuropathology had reduced memory scores, more so than executive function, but individuals with cerebrovascular disease had similarly reduced premorbid cognitive function in executive function and verbal and nonverbal memory.⁴⁹

In summary, epidemiologic data strongly support that hypertension is associated with worse cognitive outcomes, especially when BP is considered in midlife. A pattern of midlife hypertension followed by later-life hypo-or normotension may be especially problematic, and consideration of BP variability not only over the course of a day but over years and decades may be important in understanding what components of BP are most important in increasing risk of cognitive decline, MCI, and dementia. Although hypertension may impact all aspects of cognition, its impact is greatest on motor speed, attention, and executive functioning. The likely overlap with AD pathology, however, in the large group of individuals with mixed pathology of MCI and dementia, makes differentiation of dementia subtype on the basis of patterns of impaired cognitive domains challenging.

Pathobiology of the effects of hypertension on the brain

The cerebral vasculature is the prime target of the deleterious effects of hypertension on the brain. In this section we will briefly review basic concepts on the blood supply of the brain and its regulation, and examine how chronic hypertension affects the structure and function of cerebral blood vessels.

The blood supply of the brain

The brain’s blood supply is provided by the internal carotid arteries and the vertebral arteries. These extracranial vessels enter the skull and merge the base of the brain to form the circle of Willis, from which the major intracranial vessels supplying the brain originate (anterior, middle and posterior cerebral arteries)(Figure 1).⁵⁰ Branches of these arteries travel along the brain surface giving rise to a highly anastomotic vascular network (pial arteries)⁵⁰. Branches of the pial arteries penetrate the substance of the brain (penetrating arterioles) surrounded by a virtual space, the perivascular space or Virchow-Robin space⁵⁰, separating the vascular basal lamina from the astrocytic processes enveloping the brain parenchyma (glial limitans). As the vessels get smaller, the perivascular space disappears and the vascular basal lamina enters in direct contact with the astrocytic foot processes, which completely envelop the vessel’s outer surface⁵⁰. At this level, the vessel wall includes a single layer of smooth muscle cells resting on the endothelial basement membrane⁵⁰. As the vessels transition into capillaries, the smooth muscle cell layer becomes discontinuous and is gradually replaced by pericytes covering ≈30% of the abluminal capillary surface.⁵¹ Venules are endowed with a discontinuous smooth muscle cell layer, which becomes more pronounced in veins, although not as marked as in the media of arteries. Recent single cell RNA sequencing studies have started to unveil the molecular bases of the diversity of the cellular composition of the cerebrovascular tree⁵². These efforts, still in an initial stage, will provide critical insight into the cellular, molecular and signaling mechanisms underlying the segmental susceptibility to hypertension of the cerebral vasculature.

Figure 1:

Cerebrovascular anatomy and segmental pathology induced by hypertension. Large extracranial cerebral arteries (internal carotid arteries and vertebral arteries) enter the skull and converge at the base of the brain to form the circle of Willis from which the major intracranial cerebral arteries emanate. Branches of these arteries form a dense anastomotic network (pial arteries ad arterioles) which give rise to arterioles that dive into the substance of the brain (penetrating arterioles). Some vessels, like the lenticulostriate arteries, arise from the circle of Willis and proximal branches and vascularize the deep brain structures including the white matter. The predominant pathology caused by hypertension in the different segment of the cerebrovascular tree is indicated. Due to its precarious blood supply from terminal arterioles particularly vulnerable to hypertensive damage, the subcortical white matter is highly likely to be harmed by hypoxia-ischemia. Abbreviations: ICA, internal carotid artery; MCA, middle cerebral artery; WM, white matter.

Effects of hypertension on cerebrovascular structure

Chronic hypertension exerts a profound influence on the structure of the cerebral vasculature. These structural changes reflect an adaptive response to protect downstream microvessels from the increased transmural pressure. However, this process over time becomes maladaptive resulting in different pathologies depending on the segment of the cerebrovascular tree involved (Figure 1).

Large cerebral arteries:

Hypertension is a leading risk factor for extracranial and intracranial atherosclerosis.⁵³ Extracranial lesions are characterized by lipid accumulation in carotid and vertebral arteries, often associated with ulceration and atheroma instability and protrusions, all linked to artery-to-artery embolism⁵⁴. Intracranial lesions, affecting the circle of Willis and its major branches, have a more fibrous component, more likely to result in local vascular occlusions and focal ischemic stroke⁵⁵.

Remodeling is an important structural change induced by hypertension and other vascular risk factors in extracranial and intracranial cerebral arteries⁵⁶. Hypertrophic remodeling leads to increased wall thickness and reduction in vascular lumen, with an increase in media-to-lumen ratio and in the total volume of wall tissue⁵⁶. Eutrophic (inward) remodeling consists of an increased media thickness, a reduced lumen and external diameter, without changes in the total amount of wall tissue⁵⁶. In a recent study assessing remodeling in the carotid arteries using MRI, the pattern of remodeling varied according to the arterial segment⁵⁷. Whereas the common carotid artery demonstrated hypertrophy, the carotid bifurcation exhibited eutrophic remodeling. In contrast, the internal carotid artery exhibited a mixed pattern of inward and outer remodeling⁵⁷. Sex and comorbidities had a significant effect on the pattern of remodeling⁵⁷.

Hypertension is also associated with stiffening of large arteries. In humans as in animal models, stiffening precedes the development of hypertension, raising the possibility that it may contribute to the BP elevation by reducing vascular compliance and increasing pulse pressure⁵⁸. At the same time, stiffening may also reduce CBF⁵⁹, deplete cerebrovascular reserves⁶⁰, and increase the hydrodynamic impact of the elevated pressure (pulsatility) on the cerebral microvasculature⁶¹. Perhaps due to these deleterious microvascular actions, arterial stiffening is a sensitive predictor of subsequent white matter lesions and cognitive impairment.

Pial and penetrating arteries and arterioles:

These vessels are responsible for a sizable component of the pressure drop observed between large arteries and the cerebral capillaries^{50, 62}, and, as such, are a prime target of the effects of chronic elevations in BP. Animal studies suggest that hypertrophic and eutrophic remodeling are the predominant alterations in pial and penetrating vessels, stiffening being less common than in large arteries.⁵⁶ These vessels are often surrounded by enlarged perivascular spaces, a hallmark of SVD.⁶³ Particularly vulnerable are small arteries and arterioles of the deep cortical, periventricular and basal ganglia white matter. These vessels arise either from the first segment of the middle cerebral artery (MCA), e.g., lenticulostriate arteries, or from terminal branches of the pial arteries, and converge on the white matter deep into the hemisphere (Figure 2).⁶⁴ Due to the limited anastomosis and collateralization among penetrating arterioles, their dysfunction or occlusion is particularly damaging to the white matter.^{62, 65, 66} Hypertension also induces degenerative changes of the vessel wall, including accumulation of an amorphous material consisting of degenerated smooth muscle and collagen (lipohyalinosis) or infiltration of fibrin and fibrin degradation products (fibrinoid necrosis).⁶⁷ Possibly due to their shorter length and higher pressures, fibrinoid necrosis is more common in basal ganglia arterioles, and is frequently associated with cerebral hemorrhage.⁶⁷ Microatheromas can also be observed in larger arterioles, which can lead to vascular occlusion⁶⁴. Alterations are also observed in capillaries, with reduction in their number, swelling and loss of endothelial cells and pericytes (string vessels), tortuosity, increased thickness of the basal lamina, and fibrin deposition.⁶⁸

Figure 2:

Cerebrovascular autoregulation and hypertension. Autoregulation keeps CBF relatively stable during fluctuations in BP. Autoregulation was first examined by measuring CBF during stepwise changes in BP, once a steady state is reached (static autoregulation) (left). Based largely on animal data, chronic hypertension was found to shift the pressure-flow curve to the right, such that the autoregulated range moves towards higher BP. The development of methods to assess CBF dynamically like transcranial doppler flowmetry, enabled to study the flow velocity changes in large cerebral arteries during BP transients induced by different maneuvers, e.g., standing from a sitting position. As illustrated in the right panel, MCA flow drops in sync with AP. But, owing to an autoregulatory drop in cerebrovascular resistance (CVR), it recovers faster. Evidence suggests that hypertension does not impair dynamic autoregulation, except in severe hypertension (see text for further details). Right panel data from Aaslid et al. ¹⁹⁵

What causes remodeling and stiffening?

Remodeling and stiffening result from interacting mechanical, humoral and cellular factors including hemodynamic stress, endothelial dysfunction, vascular inflammation, immune cell infiltration, and calcium deposition^{69, 70}. Both processes are driven by overlapping mediators including inflammatory cytokines, sex hormones, angiotensin-II (AngII), endothelin, aldosterone, and oxidative stress⁷¹. In addition, modifications of the extracellular matrix plays an essential role in stiffening. Accumulation of collagen and fibronectin, elastin fragmentation by matrix metalloproteases, and the pro-fibrotic effects of transforming growth factor-β are critical factors in the process⁶⁹. A deficit in endothelial nitric oxide (NO) and free radicals may play an important role in remodeling.^{56, 71} In the aorta perivascular macrophages (PVM) expressing the lymphatic hyaluronan receptor LYVE1 safeguard the elasticity of the vascular wall by governing collagen content in smooth muscle cells through hyaluronan and pericellular matrix metalloprotease-9.⁷² Although LYVE1+ PVM are present also in cerebral arteries,^{72, 73} it remains unclear if they serve a similar role in the homeostasis of the cerebrovascular wall.

In summary, stiffening, hypertrophic and eutrophic remodeling result from the concerted action of mechanical, cellular and molecular factors that converge on the vascular wall to alter its structure and composition, as well as the balance between collagen and elastin. Intended to protect the downstream microvessels from the mechanical stress associated with elevated pressure, these responses become maladaptive and, in concert with the functional alterations described in the next section, contribute to the harmful effects of hypertension on the brain.

Impact of hypertension on cerebrovascular function

Endowed with limited energy reserves, the brain is highly dependent on the timely delivery of oxygen and glucose through blood flow to maintain its functional and structural integrity⁵⁰. To this end, sophisticated neurovascular regulatory mechanisms assure that the blood supply is well matched to the brain’s regionally and temporally diverse metabolic requirements, balancing energy delivery with removal of unwanted byproducts of brain activity⁵⁰. In this section, we will first review these mechanisms and then examine their disruption by hypertension.

Neurovascular coupling and endothelial regulation:

Neurons, astrocytes and vascular cells work as a single functional unit (neurovascular unit) to maintain the homeostasis of the brain’s internal milieu⁵⁰. Brain activity is a major regulator of local cerebral perfusion, such that increased synaptic activity leads to increases in CBF highly localized to the activated area⁵⁰. This fundamental property of the cerebral circulation (neurovascular coupling) results from the close interaction among cells of the neurovascular unit at all levels of the cerebrovascular tree. Although the mechanistic bases of the hemodynamic response remain to be completely elucidated, it is now clear that neurons, astrocytes, endothelial cells, vascular smooth muscle cells, and pericytes all participate in this carefully orchestrated process through multiple diffusible mediators (NO, prostanoids, adenosine, ions, etc.), ion channels and segment-specific signaling mechanisms⁵⁰. Thus, activation of neurons deep in the substance of the brain initiates a series of highly coordinated vascular changes that start at the level of the capillary endothelium and are transmitted to upstream arterioles through intramural vascular signaling resulting in their relaxation. The retrograde propagation of the vasodilation ultimately involves larger vessels at the brain’s surface, which is required to induce a vigorous increase in flow to the activated area^{50, 74}. Endothelial cells are important regulator of cerebral perfusion by producing powerful vasodilators, e.g., NO, prostacylin, or vasoconstrictors, e.g., endothelin-1.⁷⁵ These vasoactive mediators are released in response to mechanical or chemical stimuli resulting in increased or decreased in blood flow (endothelium-dependent vasodilatation or vasoconstriction).⁷⁵ The endothelium is also the site of the blood-brain barrier (BBB) governing the bidirectional molecular exchange between brain and blood.⁷⁶ Furthermore, the endothelium exerts vital trophic effects on brain cells and contribute to maintain the health of neurons, glia, and oligodendrocytes^{77, 78}.

Hypertension reduces resting CBF, suppresses neurovascular coupling and endothelium dependent responses. Contrary to early observations⁵, several studies have shown focal or global reductions in resting CBF in individuals with hypertension^{79, 80}. In a large scale longitudinal study of hypertensives with ischemic lesions and other vascular risk factors, a reduction in global CBF was observed with uncontrolled or untreated hypertension, but not with satisfactory BP control⁸¹. In cognitively normal people with hypertension, but no other risk factors or evidence of brain damage, focal reductions in CBF were observed in the prefrontal, cingulate, temporal, and occipital cortex⁸², suggesting that the reduction in CBF may precede the development of brain pathology. It remains unclear if these reductions in CBF are a consequence of reduced brain energy metabolism or of a direct effect of hypertension on the cerebral vasculature. Although reductions in glucose utilization, an index of energy metabolism, have been reported in neocortical and subcortical brain regions of patients with well-controlled longstanding hypertension⁸³, it has not been established whether CBF is also reduced in these regions. However, in patients with severe hypertension oxygen consumption is reduced in regions with reduced CBF⁸⁰, suggesting that with disease progression the vascular effect could be related to reduced metabolic demands possibly related to brain dysfunction and damage.

Relatively few studies have investigated neurovascular coupling in patients with hypertension. The cerebrocortical CBF response produced by cognitive tasks (memory and attention), assessed by O¹⁵ positron emission tomography (PET), was reduced in patients with untreated hypertension.⁸⁴ Furthermore, in patients with repaired aortic coarctation, in whom only pulse pressure was significantly elevated, the flow velocity increase produced by photic stimulation in the posterior cerebral artery, which supplies the visual cortex, was suppressed.⁸⁵ Consistent with altered neurovascular coupling, the increase in retinal blood flow induced by light flickering was absent in individuals with hypertension.⁸⁶ These findings, albeit limited, support the view that neurovascular uncoupling may compromise substrate delivery to the brain and render the brain more vulnerable to vascular insufficiency.

Endothelial dysfunction has been studied extensively in peripheral arteries and was found to precede the elevation in BP and, once developed, to correlate with its severity⁸⁷. Although direct assessment of cerebral endothelial function in humans is not feasible, inhibition of NO synthesis does not reduce retinal flow in patients with hypertension, indicating a deficit of NO and endothelial dysfunction⁸⁶. Furthermore, post-mortem analysis of arterioles of patients with SVD, most often caused by hypertension, showed a reduced response to the endothelium-dependent vasodilator acetylcholine⁸⁸. Independent evidence for endothelial dysfunction is also provided by the observation that hypertension alters the BBB⁸⁹, a property of the cerebral microvasculature attributable to endothelial cells.⁷⁶

The implications of cerebral endothelial dysfunction for the effects of hypertension on cognition are paramount. Endothelial dysfunction is a prelude to atherosclerosis⁷⁵ and, as mentioned above, contributes to vascular remodeling and stiffening. Failure of endothelium-dependent vasodilation in peripheral arteries correlates with white matter lesion burden⁹⁰ and microbleeds⁹¹, whereas in a population of elderly hypertensives with subjective memory complaints, white matter lesion burden was associated with circulating levels of von Willebrand factor, an index of endothelial dysfunction.⁹² The BBB dysfunction is also likely to have a devastating effect on the brain especially in the white matter, wherein increased BBB permeability has been implicated in the damage produced by hypertension.⁸⁹

Cerebrovascular autoregulation:

Cerebrovascular autoregulation refers to the ability of brain vessels to maintain a relatively stable CBF despite changes in BP within a certain range.⁹³ Owing to cerebrovascular autoregulation, CBF does not follow passively the BP swings that occur during activities of daily living, protecting the brain from excessive CBF fluctuations. Autoregulation has traditionally been assessed using stepwise changes in BP and measuring the corresponding changes in CBF when a steady state is reached (static autoregulation)^{94, 95}. In the normal state, plotting the CBF change as a function of the BP changes leads to an italic S-shaped curve with a plateau in which CBF remains relatively stable despite changes in BP (the autoregulated range: 60-150 mmHg mean BP) (Figure 2)⁹⁴. More recently, methods have been developed to assess cerebrovascular perfusion dynamics in response to fast changes in BP (dynamic autoregulation)(Figure 2)⁹³. These studies have shown that during pressure changes on a time scale of seconds, blood flow does not remain stable, but follows BP until the autoregulatory adjustments in vascular resistance counteract the BP change (Figure 2). In general, the faster the change in BP the less efficient is autoregulation⁹⁶.

The cellular bases of autoregulation rest in part on the intrinsic property of the smooth muscle cells to constrict in response to increases in transmural pressure (the myogenic response of Bayliss)⁹⁵. Accordingly, if BP increases, the smooth muscle cells constrict the vessels and increase vascular resistance, thereby preventing the CBF increase resulting from the increased pressure. Conversely, decreases in BP relaxes the smooth muscle, reducing vascular resistance and counteracting the anticipated CBF decrease. Multiple redundant mechanisms underlie the myogenic response including: (a) mechanosensors that modulate intracellular Ca²⁺ levels in response to changes in transmural pressure; selected candidates include: transient receptor potential channels⁹⁷ and G-coupled receptors, including AngII type-1 receptors (AT1R)⁹⁸ and, possibly, tissue necrosis factor receptors⁹⁹; (b) amplification of the Ca²⁺ signal by depolarization and opening of voltage-gated Ca²⁺ channels leading to further Ca²⁺ entry, as well as intracellular Ca²⁺ release¹⁰⁰; (c) modulation of the Ca²⁺ sensitivity of the contractile apparatus by regulating the balance between kinases and phosphatases responsible for myosin light chain phosphorylation and involving Rho kinase and protein kinase-C¹⁰¹. The mechanisms regulating the vasodilatation that occurs during BP lowering are less well understood. It has been suggested that the vasodilatation occurring with reduced BP may be an active phenomenon driven by metabolic byproducts, such as adenosine, which accumulates as a result of the reduced tissue oxygen tension caused by the hypoperfusion¹⁰².

Hypertension results in the chronic adaptation of the cerebral circulation to higher levels of BP, such that the autoregulated range of CBF is higher than in normotensive individuals¹⁰³. The implication of this right shift of the curve is that, if the BP is lowered to a level that would be safe in non-hypertensive individuals, the brain would be more susceptible to hypoperfusion⁹⁴. The shift of the autoregulatory curve, first described in a baboon model¹⁰⁴ and verified in a limited number of hypertensive individuals, has been attributed to the increase in vascular resistance induced by remodeling in cerebral resistance vessels.⁹⁴ More recent studies of dynamic autoregulation have failed to report major alterations in the moment-to-moment adjustments of CBF to BP changes.^105–107 Rather, in patients with moderate hypertension (mean systolic BP: 163±11mmHg), an enhancement of dynamic autoregulation was found.¹⁰⁸ Furthermore, lowering BP with anti-hypertensive therapy did not reduce flow velocity,^{107, 108} but improved carotid artery distensibility after 6 months of treatment.¹⁰⁷ However, in patients with malignant hypertension (SBP: 180-260 mmHg) dynamic autoregulation was altered and flow velocity followed passively the reduction in BP during treatment with sodium nitroprusside,¹⁰⁹ suggesting that severe disease can have a profound impact on both static and dynamic autoregulation.

These observations, collectively, indicate that while hypertension may shift the autoregulatory range towards higher BPs, the compensatory adjustments taking place during BP variations within this range are not impaired, except in malignant hypertension. In addition, in moderate hypertension, lowering BP does not seem to lead to cerebral hypoperfusion. However, autoregulation data over the life course and in the presence of vascular comorbidities are not available. Therefore, it remains unclear how magnitude and duration of the BP elevation influence dynamic autoregulation and what is the impact of antihypertensive therapy on CBF.

Perivascular spaces, clearance, neuroimmune trafficking:

The brain vascular system and surrounding structures are also responsible for the clearance of waste products generated by brain activity, including toxic proteins such as amyloid-β and tau.¹¹⁰ A transvascular pathway utilizes abluminal transporters that carry solutes through the vascular wall and drain into the circulating blood.⁷⁶ Other pathways are thought to use the perivascular space, a site of exchange between the interstitial fluid (ISF) and the cerebrospinal fluid (CSF), as a conduit to carry solutes to the brain surface. In particular, a perivascular pathway has been proposed to travel retrogradely within the arterial wall¹¹¹ whereas a paravascular “glymphatic” pathway, involving aquaporin-4 in astrocytic end-feet, reaches the brain surface through perivenous spaces.¹¹² However, this field is still in evolution and additional evidence is needed to assess the relative weight of these clearance pathways in health and disease. In the end, solutes drain into the cervical lymphnodes, probably through dural lymphatics.¹¹³ Cerebral blood vessels and perivascular spaces are also critical for the trafficking of immune cells in and out of the brain required to maintain immune homeostasis and protect the brain from invading pathogens.¹¹⁴ Innate and adaptive immune cells enter the brain through transvascular, meningeal, choroid plexus routes and play a critical role in immune surveillance and in a wide variety of pathologies.¹¹⁴ In particular, as described in the next section PVM located in the perivascular space have emerged as a critical source of reactive oxygen species (ROS) in hypertension with a profound impact on neurovascular function.¹¹⁵

How does hypertension cause neurovascular dysfunction?

Several animal models have been used to investigate the neurovascular effects of acute or chronic elevations in BP.¹¹⁶ Better-studied models involve administration of pressor doses of AngII, which induce acute elevations in BP, or administration of sub-pressor doses (“slow pressor” AngII hypertension) which produce delayed elevations in BP over several days. The slow pressor model has gained popularity since is thought to reflect the progressive increase in BP observed in essential hypertension, although with a more compressed time scale (2-4 weeks)¹¹⁷. In hypertension induced by acute or chronic AngII administration there is attenuation of neurovascular coupling and endothelium-dependent responses.^118–120 Interestingly, the mechanical effects of the elevation in transmural pressure are not required for the cerebrovascular dysfunction, at least in the short term.^{118, 119} Neurovascular dysfunction and cognitive deficits have also been reported in other models of hypertension, including life-long hypertension in spontaneously hypertensive rats (SHR) or mice (BPH mice)⁷³, intermittent hypoxia (a model of sleep apnea),¹²¹ hypertension induced by administration of deoxycorticosterone acetate+salt (DOCA-salt)¹²², eNOS inhibition, etc. (see also^{116, 123}).

The pathways mediating the neurovascular effects of hypertension induced by slow-pressor AngII administration have been studied in some detail and involve both central effects on the brain and peripheral effects on cerebral blood vessels. The evidence suggests that circulating AngII acts on AT1R in the subfornical organs, one of the circumventricular organs, leading to ROS-mediated production of vasopressin from the hypothalamic paraventricular nucleus. Vasopressin, in turn, acts on cerebral blood vessels to trigger local endothelin-1 production which, in concert with circulating AngII, contributes to the neurovascular dysfunction by inducing vascular oxidative stress.¹²⁴ Evidence from systemic vessels also indicates a critical role of vascular oxidative stress.¹²⁵ However, how oxidative stress alters vascular function remains to be defined. In AngII hypertension, peroxynitrite, the reaction product of NO and superoxide, is required for the neurovascular dysfunction,¹²⁶ but whether the effect depends on reduction of NO bioavailability by the reaction with superoxide, or on downstream effects of peroxynitrite, remains to be established. Evidence from the AngII hypertension model indicates that the latter possibility is more likely, since a peroxynitrite decomposition catalyst rescues the neurovascular dysfunction.¹²⁶ Peroxynitrite could alter vascular function through several mechanisms.¹²⁷ In the neurovascular dysfunction produced by Aβ, peroxynitrite-induced DNA damage activates the DNA repair enzyme poly-ADP ribose polymerase and leads polyADP ribose production, which, in turn, opens transient receptor potential melastatin-2 channels resulting in endothelial Ca²⁺ overload and dysfunction¹²⁸. However, it remains to be established if this mechanism is also involved in the neurovascular effects of AngII hypertension.

There are several potential enzymatic sources of vascular oxidative stress, including mainly mitochondrial complexes, xanthine oxidase, and eNOS uncoupling, a condition in which the enzyme generates ROS instead of NO.^{129, 130} However, a NOX2-containing NADPH oxidase has emerged as a critical source of the ROS involved in neurovascular dysfunction especially in AngII hypertension. NOX2 is present in vascular and perivascular cells in close association with AT1R.^{73, 119} Furthermore, pharmacological inhibition or genetic deletion of NOX2 rescues the neurovascular dysfunction in models of hypertension.^{119, 120, 124} Other isoforms of NOX, particularly NOX1 and NOX4, may also present in the cerebral vasculature, but their cellular localization and role in neurovascular dysfunction have not been elucidated.¹²⁹

As for the cellular source of ROS, endothelial cells, smooth muscle cells, and pericytes have the potential to produce ROS. However, recent data suggest that PVM are a critical source of the ROS initiating the vascular dysfunction in AngII hypertension. PVM are endowed with AT1R and, as phagocytic cells, express large amounts of NOX2 and have a greater potential for ROS production than vascular cells¹¹⁵. In slow pressor AngII hypertension and in BPH mice, a genetic model of life-long hypertension, PVM depletion prevents vascular oxidative stress, neurovascular dysfunction and cognitive impairment, as well as MCA remodeling in hypertensive rats^{73, 131}. Using bone marrow transplantation to replace PVM with bone marrow-derived macrophages lacking the AT1R or NOX2 abrogates vascular oxidative stress and neurovascular dysfunction, pointing to PVM as the cellular site of the AT1R and NOX2.⁷³ Since in BPH mice and in the slow pressor model AngII is elevated in the circulation but not in brain, circulating AngII was found to cross the BBB and gain access to PVM on the abluminal side of the vessels (Figure 3).⁷³ Although PVM are unlikely to be the sole source of vascular oxidative stress, the observation that their elimination or genetic modification blocks vascular ROS production, suggests that they are the primary source, which may trigger oxidative chain reactions in adjacent vascular cells. Whether PVM are also involved in the neurovascular dysfunction in human hypertension remain to be established. However, as discussed in the next sections, the observation of perivascular space abnormalities, vascular oxidative stress and inflammation in SVD is consistent with findings emerged from animal models implicating PVM.

Figure 3:

Putative mechanisms of neurovascular dysfunction in slow pressor AngII hypertension, BPH mice and DOCA-salt hypertension. Circulating or brain AngII reaches AT1R in PVM wherein it activates Nox2 leading the vascular oxidative stress and neurovascular dysfunction. AT1R and Nox2 are also present in other vascular cells, but they do not seem to play a primary role in these hypertension models. The cartoon depicts a penetrating arteriole, but brain macrophages are also present in the meninges, wherein they may play a role in pial and meningeal vessel vasomotor function and permeability. Whether PVM have a similar role in models of AngII-independent hypertension remains to be established. Abbreviations: AT1R, AngII type 1 receptor; PVM, perivascular macrophages; SMC, smooth muscle cells. (Illustration Credit: Ben Smith).

In summary, experimental studies and human data indicate that oxidative stress and inflammation are critical factor in the alterations in neurovascular and endothelial function produced by hypertension. Although other enzymatic sources cannot be excluded, NOX2 has been identified as a major source of ROS in hypertension involving AngII. Furthermore, innate immune cells, the PVM in particular, have emerged as a powerful source of vascular ROS production.

Brain lesions induced by hypertension and their impact on cognition

The alterations in cerebrovascular structure and function induced by hypertension described in the previous sections predispose the brain to dysfunction and damage, which, in turn, alters cognition. Chronic hypertension is the major risk factor for ischemic and hemorrhagic stroke, which is associated with a 3-6 fold increase in cognitive impairment, especially when multiple strokes are involved (multi-infarct dementia).⁶⁶ Additional lesions associated with hypertension include microinfarcts and microhemorrhages, microscopic ischemic or hemorrhagic lesions most common in the cerebral cortex, which portend cognitive deficits.^132–134 Hypertension is also associated with enlarged perivascular spaces, which also correlate with cognitive impairment.⁶³ The expansion of the perivascular spaces may be due to fluid stagnation and protein accumulation reflecting impaired ISF/CSF clearance, as well as the mechanical stresses on the perivascular tissue by the increased pulsatility and stiffening⁶³. Of note, enlarged perivascular spaces can be confused with small cavitated lesions (lacunae) on conventional brain imaging, and it has been proposed that lacunes rather than enlarged perivascular spaces correlate with cognitive impairment.¹³⁵ The mechanisms of the brain atrophy induced by hypertension remain unclear, but (a) chronic hypoperfusion leading neuronal loss⁷⁹, (b) retrograde neurodegeneration from deafferentation due to lesions in other brain regions¹³⁶, or (c) vascular oxidative stress and inflammation,¹³⁷ leading to neurovascular trophic failure¹³⁸ are potential contributing factors.

The most common brain lesions associated with hypertension are white matter lesions, especially in the frontal cortex, which appear as areas of hyperintensity on T2-weighted MRI.¹³⁹ SVD is considered the major cause of white matter disease, but the mechanisms have not been completely elucidated.¹⁴⁰ Cerebral hypoxia-ischemia due to stenosis or occlusion of vessels feeding the white matter, which are particularly susceptible to the effect of hypertension, is considered to be a major contributor.⁶⁶ In support of this hypothesis, reduction in CBF in normal appearing white matter and expression of hypoxic markers have been reported.^{141, 142} Other studies have not supported this view^{143, 144}, and have suggested that hypoperfusion is not the cause but the consequence of the white matter lesion.¹⁴⁵ Inflammation, upregulation of metalloproteases, and increased BBB permeability in the white matter may lead to water shift and local edema resulting in microvascular compression and occlusion.^{145, 146} A link between vascular health and white matter maintenance and repair by oligodendrocyte has also been established. A population of oligodendrocytes associated specifically with cerebral blood vessels has been identified¹⁴⁷ and oligodendrocyte proliferation and survival requires vascular growth factors.¹⁴⁸ These observations have suggested an “oligovascular niche” essential for the maintenance of white matter health. Indeed, arrested development of oligodendrocyte precursors has been observed human white matter in SVD resulting in faulty remyelination of damaged white matter (Figure 4).¹⁴⁹

Figure 4:

Potential mechanisms of WM damage by hypertension. Vascular oxidative stress and inflammation disrupt the BBB, induce NVU dysfunction and damage, and impair oligodendrocyte development and function. The resulting alterations in tissue homeostasis, hypoxia-ischemia, reduced brain clearance and impaired remyelination lead to WM damage. Abbreviations: NVU, neurovascular unit.

In summary, oxidative stress, hypoxia-ischemia, inflammation and BBB dysfunction are critical vascular factors threatening the health of the subcortical white matter. There might be regional differences on the impact of hypertension on white matter, and the white matter of the frontal lobe may be more susceptible.^{133, 139} However, the mechanistic bases for such predilection remain to be established.

How does hypertension cause cognitive dysfunction?

Atrophy, brain damage caused by macro- and micro-infarcts and hemorrhages are important determinants of cognitive impairment. Furthermore, infarcts strategically placed in brain regions involved in cognition, e.g., hippocampus, medial thalamus, frontal lobe, etc., can produced cognitive dysfunction despite relatively small volume of damage,^{66, 133} and the estimated functional impact of microinfarcts is greater than anticipated by the volume of injury.¹⁵⁰ The volume and location of white matter lesions also correlate with the degree and temporal evolution of cognitive dysfunction.¹³³ These white matter lesions may affect cognition by impairing the connectivity between the anterior thalamus to the frontal cortex.^{133, 151} Indeed, studies using white matter tractography and resting state MR imaging have provided evidence for loss of connectivity and network efficiency degradation, which correlate with a reduction in processing speed, a typical feature of vascular cognitive impairment.¹⁵¹ In addition, brain atrophy could also be a contributing factor by reducing gray matter in critical brain regions such as hippocampus and prefrontal cortex.¹⁵²

The impact of enlarged perivascular spaces on cognitive impairment of hypertension remains uncertain. The morphological alterations of the perivascular space raise the possibility that perivascular and paravascular clearance systems may be altered and have a role in the white matter damage^{63, 140}. Perivascular spaces are enlarged and distorted in hypertension⁶³, an alteration that could hamper the disposal of potentially toxic byproducts of brain activity. Hypertension is associated with amyloid-β and tau pathology (see section on overlap with AD), suggesting that clearance of these proteins may be reduced. Hypertension-induced large artery stiffening increases pulsatility in microvessels, which has been proposed to alter perivascular spaces and reduce ISF/CSF clearance⁶³. However, a transient increase in pulsatility induced by administration of dobutamine increases CSF clearance in mice¹⁵³, suggesting that hypertension could increase ISF/CSF clearance. In support of this hypothesis ISF flow is increased in spontaneously hypertensive rats, but such increase is associated with greater retention of tracers at the ISF/CSF interface, indicating a reduction in clearance¹⁵⁴. On the other hand, carotid ligation reduced CSF clearance¹⁵³, suggesting a negative impact of cerebral hypoperfusion and/or reduced pulsatility, which is relevant to large artery stenosis or occlusion often associated with hypertension. Therefore, direct evidence linking hypertension to brain clearance pathways has not been provided, in humans as in animal models, and additional work is needed in this area.

These observations, collectively, suggest that hypertension may lead to cognitive impairment through several pathogenic factors (Figure 5). Gray matter loss, loss of connectivity and network efficiency from white matter lesions, reduced perivascular clearance and neurovascular dysfunction all have the potential to impair brain function, but their relative impact on cognitive processes remain to be established.

Figure 5:

Brain lesions produced by hypertension that underlie cognitive impairment. Brain atrophy, microinfarcts and microbleeds cause neuronal loss and brain dysfunction. In addition, microinfarcts and microbleeds disrupt brain connectivity and reduce network efficiency Damage to WM tracts (WML) also degrades connectivity, especially in thalamo-cortico networks. Alteration in PVS impair brain clearance and may promote protein accumulation in brain and vessels, and dysfunction of the NVU leads to vascular insufficiency and BBB damage. Potential interactions between different pathogenic mechanisms are indicated by the dashed arrows. Abbreviations: NVU, neurovascular unit; PVS, perivascular spaces; WML, white matter lesions.

Hypertension and Alzheimer’s pathology

A large body of evidence indicates that hypertension is a risk factor for AD, a condition once considered purely a neurodegenerative disease. AD is characterized neuropathologically by extracellular accumulation of amyloid-β (amyloid plaques) and intraneuronal aggregates of hyperphosphorylated tau (neurofibrillary tangles).² Aβ is derived from the proteolytic cleavage of the amyloid precursor protein (APP) by β- and γ-secretase enzymes.² Vascular brain lesions similar to those associated with vascular cognitive impairment are also present in AD, such that mixed AD-vascular pathology is present in the majority of AD cases diagnosed clinically.¹⁵⁵ The recent availability of imaging and biochemical biomarkers of AD has provided the opportunity to investigate the relationship between AD pathology and hypertension in vivo, in an attempt to explore the nature, causal or incidental, of their association.

Clinical-pathological biomarkers of AD in hypertension:

Although hypertension’s impact on cognition is fairly clearly established, whether this acts directly on AD neuropathology or simply as a parallel contributor to cognitive impairment and dementia is still somewhat controversial.⁸

In the ARIC-PET study, nondemented participants with a greater number of vascular risk factors in middle age, including hypertension, had more elevated brain amyloid in late-life, suggesting a possible direct impact of these risk factors on amyloid; hypertension by itself, however, was not significantly associated with elevated amyloid so may not have an independent role.¹⁵⁶ Other studies have failed to find strong relationships between hypertension and brain amyloid, although many of these studied hypertension in older age. Elevated vascular risk, including hypertension, was associated more with atrophy and neurodegeneration than with brain amyloid, among 430 adults studied at >60 years of age with both tau and amyloid PET as well as brain MRI.¹⁵⁷ In a neuropathologic sample, late-life BP was not associated with brain amyloid, but SBP was associated with neurofibrillary tangles.¹⁵⁸

Using CSF biomarkers, the potential interaction of BP and APOE genotype was identified on tau levels, in one memory clinic cohort, in which hypertension was not associated with amyloid but rather was related to tau among APOE ε4 homozygotes.¹⁵⁹ Although not specific for AD, associations between hypertension and regional brain atrophy also support a potential role of elevated BP in Alzheimer’s neurodegeneration. In addition to the above cited studies of elevated BP and brain atrophy,⁴² midlife BP has also been associated with hippocampal atrophy by MRI, with strongest associations among untreated hypertensives.¹⁶⁰

Association between hypertension and AD pathology: experimental studies

Several studies have suggested that experimental hypertension promotes Aβ accumulation and tau phosphorylation. Hypertension, induced either by administration of AngII¹⁶¹, aortic coarctation^{162, 163}, renal artery stenosis¹⁶², or in genetic models¹⁶⁴, increases microvascular deposition of Aβ or neuronal tau phosphorylation. At the same time, administration of RAS blockers or AT1R genetic deletion ameliorates amyloid deposition and behavioral dysfunction in APP-overexpressing mice.^{165, 166} There are several mechanisms by which hypertension could worsen AD pathology (Figure 6). As discussed in a previous section, Aβ is cleared in part through vascular and perivascular pathways, and, consequently, the vascular and perivascular dysfunction and damage induced by hypertension could impair Aβ disposal leading to its accumulation. On the other hand, the loss of NO and prostacyclin bioavailability associated with vascular dysfunction may also contribute to promote amyloidogenic APP processing and increase Aβ. Thus, NO reductions upregulate APP and β-secretase, while reduction in prostacyclin may favor amyloidogenic APP processing.¹⁶⁷ Mediators of hypertension may also promote amyloidogenic APP processing. For example, AngII increases APP cleavage by β- and γ-secretase and increases Aβ production in vitro and in vivo^{161, 168}. Less is known about a potential effect of hypertension on tau phosphorylation. Endothelial NO prevents tau phosphorylation by inhibiting Cdk5, one of the main kinases phosphorylating tau,¹⁶⁹ and consequently the NO deficit associated with hypertension could promote tau phosphorylation. Central administration of AngII induces tau phosphorylation by activation of the tau kinase GSK3-β via AT1R dependent mechanisms.¹⁷⁰

Figure 6:

Potential mechanisms underlying the relationship between hypertension and AD. The vascular damage produced by hypertension leads to brain dysfunction thought hypoxia-ischemia, as well as increase in Aβ production due to increased APP processing by secretase enzymes and reduced clearance of AD-related proteins. In turn, Aβ and tau alter vascular function amplifying the deleterious vascular impact of hypertension (reproduced from ref.⁸ with permission). (Illustration Credit: Ben Smith).

These clinical and experimental observations, collectively, suggest that hypertension has the potential to promote AD pathology by acting at different levels. However, it remains unclear if hypertension is a pathogenic factor in AD, and conversely, if the AD pathology associated with hypertension is a contributor to the cognitive dysfunction.⁸

Prevention and therapy

Given the strong and consistent associations seen in the epidemiologic literature between hypertension and dementia, and given the strong biological plausibility for a link between the two, the inevitable next question is whether treatment of hypertension reduces risk of dementia. Certainly, this is the major reason why the study of hypertension as a risk factor for dementia is of particular interest: in the absence of other ways to treat or prevent dementia and AD specifically, hypertension is an especially appealing target.

Does treating hypertension reduce dementia risk?

Several epidemiologic studies have considered the role of antihypertensive medication treatment, although all remain susceptible to some indication bias: individuals who are prescribed and take antihypertensives are different than people who don’t, in many ways beyond which can be adjusted statistically. Thus, clinical trials would be the ideal forum in which to test this question, but given the above reviewed evidence that relationships between hypertension and dementia are strongest when hypertension is considered in middle age, decades before the development of dementia, and that clinical trials cannot randomize and follow participants for that long duration, some reliance on longer-term observational study designs is needed to consider these lifespan considerations. These studies also allow consideration of age of treatment or duration of treatment which can extend beyond the windows allowed by clinical trials.

In several observational studies, antihypertensive medication use is associated with less cognitive decline; in the ARIC study, participants taking antihypertensive medications had 20-year cognitive decline equivalent to a prehypertensive group (higher than the normotensives but lower than untreated hypertensives).¹² In the Epidemiology of Vascular Aging (EVA) Study Group, treated hypertension was associated with less cognitive decline over only 4 years compared to untreated hypertension.¹⁷¹ Furthermore, duration of antihypertensive therapy appears important in risk reduction. In the HAAS study, per additional year of antihypertensive treatment, dementia risk was lower (HR 0.94, 95% CI 0.89-0.99), with the greatest reduction for individuals with treatment over 12 years in duration, with overall risk nearing that of normotensives.¹⁷² In the Rotterdam study, each additional year of antihypertensive use, before age 75, was associated with an 8% reduced risk of dementia; after age 75 the additional risk was reduced and was no longer statistically significant.¹⁷³

Trial data consists of several randomized clinical trials focusing primarily on antihypertensive therapy, with most trials to date showing negative results, with the exception of Syst-Eur,¹⁷⁴ and the most recent SPRINT-MIND trial.¹⁷⁵ These negative trials include those in individuals with a history of stroke, in which a Cochrane review yielded a pooled RR of 0.88 (NS) for the use of antihypertensives to reduce dementia,¹⁷⁶ as well as studies of individuals without prior known cerebrovascular disease.¹⁷⁷

As one of the few positive clinical trials, the Syst-Eur trial showed that dementia-free hypertensive individuals (SBP 160-219 mm Hg and DBP<95 mm Hg) aged ≥60 years who received medications (a dihydropyridine calcium channel blocker plus other antihypertensive drugs as needed) to lower BP to <150 mm Hg systolic, compared to placebo, had less incident dementia over a median followup of 2 years.¹⁷⁴ The preliminary SPRINT-MIND data,¹⁷⁵ not yet published at the time of preparation of this review, do suggest a benefit in progression of cerebral SVD for individuals treated with intensive BP control vs standard control, as well as for cognitive outcomes, but only for the combination of MCI and dementia, and not for dementia itself.

Importance of other comorbid vascular disease and risk factors:

Although the focus of this review is on hypertension, elevated BP as a risk factor rarely occurs in isolation, and several risk factors co-occurring in the same individuals may further increase accelerated cognitive decline. In the Framingham Heart Study, hypertension was associated with worse decline in the presence of diabetes compared to its effect in nondiabetics,¹⁷⁸ and in the Framingham Offspring Study, cognitive outcomes were worst among participants with hypertension and elevated waist-hip ratio.¹⁷⁹ In 1449 adults aged 65-79 yo in Finland, the combination of SBP at 160 mm Hg or higher and elevated serum cholesterol, each in midlife, was associated with a markedly increased risk of AD (OR 3.5, 95% CI 1.6-7.9), when compared to the presence of either risk factor alone.¹⁹

Similar relationships are seen when consideration is made of any additional risk factor: in 8,945 participants of the Kaiser HMO, an increasing number of midlife vascular risk factors was associated with elevated risk of dementia in late-life, with a HR of 1.27 for individuals with only one midlife vascular risk factor and 2.37 for four midlife vascular risk factors (considering smoking, hypertension, hypercholesterolemia, and diabetes).¹⁶ The CAIDE score, which considers several vascular risk factors as a dementia risk score, including hypertension, has been reliably associated with dementia,¹⁸⁰ with reasonably strong predictive power (area-under-the-curve 0.74).¹⁸¹ In parallel, better Life’s Simple 7 scores, which consider optimal control of vascular (including hypertension) and lifestyle risk factors, were associated with less cognitive decline, in the ARIC study.¹⁸²

As the primary genetic risk factor for AD, carriage of the APOE ε4 allele independently increases risk for AD, but the combination of the ε4 genotype and hypertension may further increase risk of AD.¹⁸³ In the PATH Through Life study, the combination of an APOE ε4 allele with hypertension was associated with greater decline over 8 years in 1,474 cognitively normal adults aged 60-64.¹⁸⁴ In ARIC-PET, as cited above, although no statistically significant difference was found among APOE ε4 carriers vs noncarriers for risk of elevated BP on late-life brain amyloid, there was a suggestion of more brain amyloid in individuals with at least one ε4 allele, as midlife BP increased.¹⁵⁶ This potential interaction was further supported by a study by Rodrigue et al in which the combination of hypertension and at least one APOE ε4 allele was associated with greater brain amyloid, by PiB PET,¹⁸⁵ with a similar pattern with greater cortical thinning among individuals with both hypertension and at least one ε4 allele.¹⁸⁶

Class-specific effects of antihypertensive drugs:

There are conflicting data as to the potential role of distinct antihypertensive medications: in HAAS, the only antihypertensive drug category when considered at mean age 77 which was associated with reduced cognitive impairment was β-blockers (incidence rate ratio 0.69, 95% CI 0.50-0.94).¹⁸⁷ Syst-Eur primarily had a calcium channel-blocker (CCB) based intervention, and showed reduction in dementia rates,¹⁷⁴ and the Perindopril Protection Against Recurrent Stroke Study (PROGRESS) showed a benefit in dementia reduction among individuals with a prior history of stroke or transient ischemic attack who were treated with active therapy (perindopril (an ACE inhibitor) or indapamide (a diuretic)) over 3.9 years, but only among individuals who also experienced recurrent stroke.¹⁸⁸ In the 3-City cohort study, participants who took a combination of selective serotonin reuptake inhibitors (SSRI’s) with CCBs, compared with SSRI’s combined with other antihypertensives, had improved cognitive performance, as well as improved depression scores, at 2-year followup,¹⁸⁹ with a similar beneficial effect on cognitive decline from CCBs noted in the Newcastle 85+ Study of elderly adults, without a similar benefit from other antihypertensives.¹⁹⁰ This is in direct contrast to findings from the Canadian Study of Health and Aging, in which Canadians aged 65 and older who took CCBs had steeper decline, measured by a modified mini-mental state examination repeated serially over 5 years, than did their counterparts who used other antihypertensive drugs (75% on CCBs declined, versus 59% on other antihypertensive drugs).¹⁹¹

A growing body of evidence supports the biological importance of the RAS, and points to the potential importance of drugs within this family for dementia prevention in individuals with hypertension. Intervention trials in spontaneously hypertensive rats have shown reduced post-stroke cognitive impairment in rodents treated with RAS modifiers (specifically candesartan, an AT1R blocker).¹⁹² The ongoing HEART phase 1B randomized controlled trial will evaluate the RAS system specifically, in nondemented adults at risk for AD, in which participants will be randomized to placebo vs one of two doses of telmisartan.¹⁹³

The optimal treatment regimen to prevent future cognitive impairment or dementia is difficult to ascertain, since nearly all antihypertensives have studies supporting their potential benefit. Again, these are still likely confounded by indication bias, similar to the overall issue of indication bias regarding any antihypertensive use vs none, although to a lesser degree (considering one antihypertensive versus another may be influenced by comorbidities or demographic characteristics, but are less profoundly impacted by socioeconomic or other factors influencing access to medical care more broadly.

Outstanding Questions and Future Prospects

Although the body of knowledge relating hypertension to cognitive outcomes, dementia, and AD has expanded tremendously in recent years, several key questions remain. Answers to some of these questions, listed below, will be critical in order to gain a better insight into how hypertension impacts cognitive function and to best recommend prevention and management strategies.

1. Is the neurovascular dysfunction induced by hypertension sufficient to cause cognitive impairment?

   Hypertension induces alterations in neurovascular function, which are thought to induce brain lesions associated with cognitive impairment (Figure 5). However, it remains to be established if the neurovascular dysfunction alone is sufficient to induced cognitive impairment. Reduced cerebral perfusion, alterations in brain clearance and BBB, as well as vascular growth factor deficiency have the potential to alter neuronal function in metabolically-active brain regions involved in cognitive function, such as the hippocampus, entorinal cortex and prefrontal cortex. A better understanding of the natural history of the impact of hypertension on cerebrovascular function, network degradation and cognition would be needed to address this question, which is relevant to the initiation of anti-hypertensive therapy. In parallel, a more nuanced understanding of the signaling mechanisms by which the neurovascular dysfunction interferes with neuronal function would also be desirable. Is CBF insufficiency the major factor? Or, are there other aspect of cerebrovascular biology, such as trophic support by endothelial factors or perivascular clearance, also at play?

2. When does antihypertensive therapy need to be initiated in order to minimize risk of cognitive decline and dementia?

   As summarized above, the benefits of antihypertensive therapy are likely to be greatest when initiated in midlife, and continued over decades, although preliminary results from the SPRINT-MIND trial suggest that benefits may still be possible with a shorter duration of antihypertensive therapy. In this regard, it would be important to assess if the appearance of brain lesions diminish the impact of hypertensive therapy on cognitive health. Although difficult to study in clinical trials, further consideration will need to be made of optimal age and duration of antihypertensive therapy, and correlation with cardiovascular and structural-functional imaging biomarkers, in order to best prevent cognitive decline, MCI, and dementia.

3. Are there subgroups of individuals who are especially susceptible to the cognitive effects of hypertension and in whom more aggressive antihypertensive therapy might be especially warranted?

   Individuals with more vascular risk factors appear especially vulnerable to hypertension, with worse outcomes among individuals with several vascular risk factors. This suggests that individuals with other known risk factors, such as diabetes, may need better screening of BP with a lower threshold for initiation of antihypertensive therapy. This is consistent with the management of vascular risk factors as they relate to other cardiovascular outcomes. Beyond vascular risk, there may be genetic susceptibilities, such as has been seen with the APOE data, leading to greater risk, and therefore greater potential benefit from antihypertensive therapy, in individuals with a known risk allele, or even of a particular race or sex. Focused trials with enrollment of these higher-risk groups could identify a subgroup in whom prevention might be especially effective.

4. What class of antihypertensive is the most effective for prevention of cognitive decline, MCI, and dementia?

   Although ongoing studies are directly addressing the potential benefit of particular antihypertensives, at least on surrogate endpoints, at the time of this manuscript preparation no convincing data point to a clinical difference with the use of particular antihypertensive medications. This question—whether any antihypertensive is as effective as the next—will be critical as evidence is translated into practice. Furthermore, it is possible that the same recommended therapeutics may not be ideal for all individuals. Race-based differences in therapeutic effect for distinct antihypertensives has been noted for other cardiovascular outcomes,¹⁹⁴ and similar differences by race, or other demographic or genetic factors, may be found for reduction of adverse cognitive outcomes.

5. What are the best surrogate endpoints or biomarkers for measuring a cerebral or cognitive benefit in treating hypertension?

   Biomarkers or surrogate endpoints are valuable in studying long-term relationships as seen between hypertension and dementia because they allow for earlier assessment of therapeutic effect, both for research purposes as well as clinically, to evaluate benefits of treatment. Ongoing studies and consortia, including the MarkVCID consortium, are evaluating biomarkers for the vascular contribution to cognitive impairment and dementia, which likely would be candidates for the specific effect of hypertension on the brain as well, and may include serum, CSF, or imaging markers to examine the contribution from AD pathology.

6. Is there a synergistic relationship between hypertension and AD pathology?

   As discussed in a previous section, there is biological plausibility that hypertension and other vascular risk factors may promote AD pathology, and vice-versa. However, the clinical-pathological evidence remains contradictory. The recent development and validation of amyloid and tau PET imaging provides the opportunity to examine in vivo the reciprocal interaction between markers of AD and hypertensive neuropathology and their relative contribution to cognitive impairment. These findings would have important implications for targeting treatments to the relevant pathology responsible for the cognitive deficits in a particular individual. At the same time, in depth investigations on the impact of AD and hypertension on vessel-associated clearance pathways would be invaluable to provide insight into the role of failure of Aβ and tau disposal on disease onset and progression.

These are a few of the outstanding questions that remain to be addressed. Answering these questions will require engaging both the pre-clinical and clinical scientific communities in a concerted effort to advance the understanding of how hypertension affects brain function leading to cognitive impairment. Thus, epidemiological, biomarker, and clinical-pathological studies should be closely related to basic science efforts to unravel mechanisms and provide new targets to be tested in clinical trials. Such cooperation will be essential for the development of new diagnostic tools and treatment strategies for one of the most devastating health challenges of our times.

Non-Standard Abbreviations and Acronyms

Aβ:

amyloid-β

ACE:

angiotensin converting enzyme

AD:

Alzheimer’s disease

ACA:

anterior cerebral artery

AngII:

angiotensin-II

ApoE:

apolipoprotein-E

APP:

amyloid precursor protein

ARIC:

Atherosclerosis Risk in Communities

AT1R:

angiotensin type-1receptor

BBB:

blood-brain barrier

BP:

blood pressure

BPH:

blood pressure high

CBF:

cerebral blood flow

CCBs:

calcium channel blockers

Cdk5:

cyclin dependent kinase-5

CSF:

cerebrospinal fluid

DBP:

diastolic blood pressure

EVA:

Epidemiology of Vascular Aging

GSK3:

glycogen synthase kinase-3

HAAS:

Honolulu Asia Aging Study

ISF:

interstitial fluid

LYVE1:

Lymphatic vessel endothelial hyaluronan receptor-1

MCA:

middle cerebral artery

MCI:

mild cognitive impairment

MMSE:

mini-mental state examination

MoCA:

Montreal cognitive assessment

MRI:

magnetic resonance imaging

NHLBI:

National Heart Lung and Blood Institute

NO:

nitric oxide

NADPH:

Nicotinamide adenine dinucleotide phosphate

NOX:

NADPH oxidase

PET:

positron emission tomography

PiB:

Pittsburgh compound-B

PVM:

perivascular macrophages

RAS:

renin angiotensin system

ROS:

reactive oxygen species

SBP:

systolic blood pressure

SSRI:

selective serotonin reuptake inhibitors

SVD:

small vessel disease