Pathological Continuum From the Rise in Pulse Pressure to Impaired Neurovascular Coupling and Cognitive Decline

Abstract

The “biomechanical hypothesis” stipulates that with aging, the cumulative mechanical damages to the cerebral microvasculature, magnified by risk factors for vascular diseases, contribute to a breach in cerebral homeostasis producing neuronal losses. In other words, vascular dysfunction affects brain structure and function, and leads to cognitive failure. This is gathered under the term Vascular Cognitive Impairment and Dementia (VCID). One of the main culprits in the occurrence of cognitive decline could be the inevitable rise in arterial pulse pressure due to the age-dependent stiffening of large conductance arteries like the carotids, which in turn, could accentuate the penetration of the pulse pressure wave deeper into the fragile microvasculature of the brain and damage it. In this review, we will discuss how and why the vascular and brain cells communicate and are interdependent, describe the deleterious impact of a vascular dysfunction on brain function in various neurodegenerative diseases and even of psychiatric disorders, and the potential chronic deleterious effects of the pulsatile blood pressure on the cerebral microcirculation. We will also briefly review data from antihypertensive clinical trial aiming at improving or delaying dementia. Finally, we will debate how the aging process, starting early in life, could determine our sensitivity to risk factors for vascular diseases, including cerebral diseases, and the trajectory to VCID.

“Without a healthy brain, little else is worthwhile, and without brain health, there is no health.” ¹

Dementia affects 50 millions people worldwide and this number is expected to triple by 2050 at an estimated cost for the societies of $4 trillion.² Like cancer, dementia is multifaceted. One important cause of dementia is cerebrovascular diseases. The connections between the brain vessels and the cognitive functions are obvious when they fail: in subjects at risk, initial mild vascular cognitive impairment may tilt toward vascular dementia (VaD). An understanding of the basic molecular links of the neurovascular connection is, however, lacking and translates into the absence of therapeutic options for patients with vascular cognitive impairment and VaD. So far, only a healthy lifestyle can protect the brain through cerebrovascular health, an approach advocated in the 2018 World Alzheimer’s Report.² As an example, in a 44-year longitudinal follow-up study of 191 Swedish women, a high cardiopulmonary fitness in midlife (50 years old) was associated with a decreased risk of subsequent dementia.³ In this review, we address the basis of the connectivity between the brain vessels and cognition, focusing on the potentially deleterious effect of central (aortic) pulse pressure on the cerebral microcirculation interconnected with brain cells, and attempt to formulate outstanding questions in the field of cerebrovascular health and the pathogenesis of dementia.

MAJOR TYPES OF DEMENTIA

Alzheimer’s disease (AD) is the most prominent type of dementia and represents 60–80% of all dementia cases. In one of his first reports, Aloïs Alzheimer noticed the presence of cerebrovascular lesions in postmortem brain tissues of demented patients and he proposed that these lesions could contribute to the formation of senile plaques.⁴ Eight decades later, amyloid β (Aβ) was identified as the essential constituent of the amyloid plaque⁵ and scientific research in the 20th century massively focused on the “amyloid hypothesis,” in which the accumulation of Aβ leads to tau-associated neurofibrillary tangles (NFTs), neuroinflammation, neuronal death, cognitive decline, and dementia.^6,7 Although this amyloid hypothesis was largely accepted and was used to conceive treatments for AD, clinical trials targeting the Aβ peptide failed in phase 3: inhibition of the production of Aβ worsened cognitive defects in patients,⁸ while inhibition of the oligomerization of the monomers of Aβ ⁹ or the immunization strategy against Aβ ¹⁰ was not successful.

Today, the accumulation of tau in NFTs and the presence Aβ load are the two major biomarkers of AD (for review¹¹). When tau is abnormally hyperphosphorylated (p-tau), this causes its accumulation in the form of NFTs in AD brains.¹² Rare familial tauopathies (mutations in the microtubule-associated protein tau (MAPT) gene) lead to frontotemporal lobar degeneration with aberrantly phosphorylated, inclusion-forming tau protein.¹³ In many autopsy brain studies, NFTs are seen in all aged brains (>65 years) independently of Aβ, and symptoms in these subjects range from normal to amnesic cognitive changes, rarely profound changes.¹⁴ Three conditions are known to favor p-tau accumulation: (i) chronic traumatic encephalopathy (for review¹¹); (ii) metabolic syndrome, through dysregulation of the insulin pathway (for review¹¹); and (iii) chronic brain hypoperfusion.¹⁵ Whether aberrant tau accumulation targets cerebral vessels is, however, not so clear. Following a brain pathological study in AD patients (Braak tau stages IV–VI) and nondemented control subjects (Braak tau stages I–IV), Merlini and colleagues¹⁶ proposed a model whereby the loss of smooth muscle cells and degradation of elastin seen in penetrating leptomeningeal arterioles are related to early Braak tau pathology and unrelated to cerebral amyloid angiopathy and AD. This tau accumulation could then impair Aβ drainage from the parenchyma and therefore promote AD.¹⁶ Another study also suggested a direct early tau-related impairment of the homeostasis of the brain that is normally maintained by the blood–brain barrier (BBB).¹⁷ Higher cerebrospinal fluid total (t-)tau levels have been related to the presence of microbleeds in control participants, non-AD and non-VaD patients, but not in AD and VaD patients where microbleeds were associated with lower cerebrospinal fluid Aβ ₄₂ levels (reflecting poor drainage).¹⁸ Other studies, however, found a significant association between cerebrospinal fluid tau and microbleed load in AD patients.¹⁹ Likewise, white matter (WM) hyperintensity, a marker of cerebral small vessel disease, has been related to tau pathology in some²⁰ but not all studies.²¹ McAleese and colleagues reported that p-tau was the only independent predictor of age-related WM change scale in postmortem fixed cerebral hemispheres of patients with AD and controls,²⁰ indicating the potentially important role of p-tau in the pathogenesis of WM damage. This would corroborate the model that tau pathology stems for microvascular damage and subsequent AD progression in subjects at risk.

Critical thinking of scientific knowledge has prompted revisiting the possible contribution of cerebrovascular lesions in the pathogenesis of AD as originally proposed by Alzheimer himself.⁴ First proposed under the name CATCH (Critically Attained Threshold of Cerebral Hypoperfusion), the “vascular hypothesis” ^22–25 stipulates that cerebrovascular dysfunction, and more specifically chronic episodes of cerebral hypoperfusion are prodromal to AD.²⁶ Importantly, the consideration of the vascular pathological events in AD does not deny the neurodegenerative markers and mechanisms involved in AD; on the contrary, it has been proposed a decade ago that vascular and neurodegenerative events are interconnected and develop in parallel.²⁷ As such, a synergism between vascular defects and neurodegeneration could explain the faster cognitive decline observed in patients with both higher cardiovascular risks, and higher Aβ and tau burden, in clinically normal old subjects.²⁸ In fact, vascular function has been proposed to be the first to decline and to be altered in asymptomatic patients; cognitive failure then occurs while vascular functions continue to deteriorate; in the last phase of the disease, cerebrovascular dysfunction is maximal.²³ The pertinence of this sequence of events putatively leading to dementia has been recently strengthened by a multifactorial data-driven analysis of brain images, in which a tentative temporal ordering of disease progression in late onset AD patients was performed.²⁹ Imaging data suggest that intrabrain vascular deregulation is an early pathological event during AD development.²⁹ The importance of the vascular dysfunctions in the pathogenesis of AD starts to be recognized and accepted, although it remains to include in guidelines biomarkers of vascular dysfunction for the clinical assessment of the patients.³⁰ Consequently, associations between vascular and neurodegenerative pathologies are still described as “intriguing” associations.³¹ Nonetheless, vascular risk factors predict the susceptibility of an individual to develop AD: after analyses of systematic reviews and meta-analyses, it has been estimated that 50% of AD cases worldwide could be attributable to modifiable risk factors such as diabetes, midlife hypertension, obesity, smoking, depression, low education, and physical inactivity.^31–33 The vascular hypothesis is supported by epidemiological studies suggesting that the risk factors for AD such as hypertension, diabetes, and hypercholesterolemia (for review³⁴) are vascular.²² Hypertension, particularly midlife hypertension, is now recognized by some as one of the major contributors of the pathogenesis of AD, as well as of VaD (for review^35,36). For others, low education,³³ physical inactivity,³⁷ or diabetes,³⁸ among various interrelated factors, could explain most of the cases of dementia or AD.

The second most prominent dementia is VaD, independently of Aβ and tau accumulation. VaD corresponds to a group of different cerebral disorders in which cognitive failure is caused by lesions in the cerebral vasculature, such as microhemorrhages and strokes,²⁴ and has been recently defined through specific guidelines.³⁹ Traditionally, VaD (a vascular disorder) was separated from AD (a neurodegenerative disorder), but, as stated above, there is growing recognition that both dementias can in fact be mixed dementias: they share the same risk factors, and both neuropathy and vascular dysfunctions can be observed in both types of dementias.^24,40 The term Vascular Cognitive Impairment and Dementia (VCID) was introduced to characterize this mixed dementia.^41–43 VaD, however, as reviewed recently,⁴⁴ specifically develops after strokes (within 6 months), after the accumulation of multiple subcortical infarcts and in patients with cerebral small vessel disease.⁴⁵ A rare disease named Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy (CADASIL), a hereditary cerebral small vessel disease caused by mutations in the NOTCH3 gene mainly affecting vascular smooth muscle cell function,^46,47 has been instrumental at validating the concept that cerebral small vessel disease could lead to dementia. CADASIL affects middle age adults with small subcortical infarcts and incomplete ischemic lesions of the WM and subcortical gray matter. Because of the younger age of the patients, however, associated traditional risk factors for vascular diseases are unfrequent. Thus, besides CADASIL, the definition of a cerebral small vessel disease is complex; it should affect the vascular endothelium and the smooth muscle, be associated with a defect in BBB function and with small vessel wall remodeling and declining capillary density leading to a reduced resting cerebral blood flow (CBF) and white matter hyperintensity. Hypertension is the leading cause of VaD, but other risk factors such as large artery stiffness-induced high brain pulse pressure, atherosclerosis-induced microemboli, or diabetes affecting endothelial integrity, are cofactors determinant in the evolution toward cerebral microvascular diseases and VaD.⁴⁸

THE NEUROVASCULAR UNIT

In the brain, vascular cells (endothelial, smooth muscle cells, and pericytes) share a very close proximity with parenchymal cells (neurons, glial cells, and astrocytes). This cellular setting is known as the neurovascular unit (NVU)^23,49 and is characterized by complex functional and dynamic interactions between the vascular and neuronal components that maintain cerebral homeostasis.²³ Astrocytes and pericytes are considered the key cellular elements in the NVU (see below). Both vascular and neuronal components of the NVU communicate in a bidirectional way; this has been demonstrated in an ischemic context where both microvascular and parenchymal (mostly neuronal) responses were observed in the same time frame and in the same cerebral ischemic regions.⁵⁰

NEUROVASCULAR COUPLING

Neurons have substantial energetic needs without the capacity to store their energetic substrates; they rely on the neurovascular coupling, a system that coordinates metabolic demands of the neurons and other parenchymal brain cells with local CBF. Thus, in response to neuronal stimulation, a local cerebral vasodilation increases blood delivery permitting adequate nutrition of the neurons.⁵¹ Neurovascular coupling is realized by the release of neurotransmitters, such as glutamate, inducing the liberation of relaxant factors from neurons, astrocytes, endothelial cells, or pericytes, leading to the local vasodilation and thus, a local increase in CBF (for review⁵²). Collectively, the different elements of the NVU orchestrate prompt and continuous adjustment of local CBF according to neuronal activity and metabolism.^49,53,54 The neurovascular coupling is therefore a critical mechanism that matches oxygen and nutrient delivery with the increased demands in active brain regions. Because the neurovascular coupling is coordinated by an intercellular signaling network, failure of one of the cellular players should induce CBF maladaptations, compromising metabolic supply to the neurons and unbalancing brain homeostasis (Figure 1). Ultimately, repeated periods of hypoxia due to CBF/metabolic mismatch can translate into neuronal stress and damage, promoting cognitive decline, a precursor of dementia.⁵⁵ It is therefore conceivable that with the exception of a traumatic injury, epilepsy, or a neuronal degenerative disease, stressors to the brain parenchyma will come from the circulation, i.e., due to a defect in the BBB made of the endothelial cells lining the cerebral microcirculation and linked to the astrocytes and the pericytes,⁵⁶ and/or from a defect in the cerebral microvascular reactivity and structure.⁵⁷

Figure 1.

Impact of vascular stresses on the neurovascular unit. The neurovascular unit, composed of astrocytes, neurons, and vascular cells including endothelial cells and pericytes, permits fine control of brain perfusion partly by regulating endothelial dilatory function of the microcirculation through the release of vasoactive molecules targeting precapillary arterioles. In the presence of abnormal vascular stresses such as high pulse pressure, hypertension, and/or dyslipidemia-induced vascular remodeling, the endothelium is the first target and damage responses occur: dysfunction of endothelial cells, disruption of the tight junctions of the BBB, increased permeability of the BBB leads to microhemorrhages, is associated with brain inflammation and senescence, inducing cerebral hypoperfusion and neuronal injury. This disruption of the neurovascular unit ultimately leads to cognitive decline and dementia. Abbreviations: BBB, blood–brain barrier; MCI, mild cognitive impairment; NO, nitric oxide; VaD, vascular dementia; VCID, vascular cognitive impairment and dementia.

GLIAL CELLS AND PERICYTES IN NEUROVASCULAR COUPLING: CONTRIBUTION TO BRAIN DISEASES

The NVU integrates brain parenchymal cells, including neurons, interneurons, astrocytes, and microglia, all interacting closely with vascular cells including pericytes (in capillaries), vascular smooth muscle cells (in arterioles and arteries), and endothelial cells. Altogether, they communicate to finely regulate local CBF according to neuronal activity and to form the BBB.⁴⁹

Brain glial cells are composed of astrocytes, microglia, ependymal cells, and oligodendrocytes; these cells are intermediaries between the neurons and the microvasculature.^58,59 The most important and abundant players are probably the astrocytes, with their vascular end-feet wrapped around microvessels in the brain, facilitating neurovascular coupling.⁶⁰ The astrocytic end-foot is a specialized unit that maintains homeostasis of the brain, playing several key roles in regulating the neurons: they maintain synapses through the regulation of extracellular glutamate levels, and also maintain ionic (regulating neuronal resting membrane potential by controlling the extracellular K⁺ concentration) and osmotic homeostasis in the brain.⁶¹ Hence, astrocyte malfunction likely has significant consequences on neuronal integrative work and in the pathophysiology of VCID.⁵⁹ Degenerated astrocytes have primarily been reported in Binswanger’s disease, a form of VCID characterized by diffuse WM lesions.⁶² Astrocytic end-foot disruptions have been observed in mouse models of AD and of VCID,^63,64 that lead to impaired neurovascular coupling, altered neuronal excitability and ultimately to cognitive deficits. The astrocytic changes (called astrogliosis) were preceded by a proinflammatory phenotype in the brain and microglial activation, at least in a mouse model of VCID.⁶³ Thus, the cellular components of the gliovascular unit (astrocytes juxtaposed to capillaries) respond to ongoing pathological insults leading to disruption of the unit. The most robust modifications involve oligodendrocyte loss and astrocytic clasmatodendrosis (beading and disintegration of astrocyte projections, along with cytoplasmic vacuolization and swelling) with displacement of the water channel protein, aquaporin 4.⁵⁸ These modifications likely precede capillary degeneration and involve tissue edema, disturbing the integrity of the BBB and inducing a chronic hypoxic state in the deep WM.⁵⁸ Interestingly, some studies report that aging astrocytes acquire a senescent phenotype both in vitro and in vivo, with secretion of neurotoxic cytokines (for review⁶⁵). The very first study that demonstrated a causal link between glial senescence and neurodegeneration was performed in mice: the clearance of the senescent cells prevented deposition of tau and degeneration of cortical and hippocampal neurons.⁶⁶ The role of astrocyte senescence and/or dysfunction in neurodegenerative diseases clearly deserves future investigations.^54,65

Similar to astrocytes, pericytes are central to the NVU function as they are located at the interface between the brain parenchyma and the brain microvasculature. Pericytes are specialized mural cells covering capillaries, where vascular smooth muscle cells are no longer present. There are in fact different types of pericytes and they could be present in various blood vessels, and not be restricted to capillaries, both in the brain and in the periphery.^60,67,68 Considering the lack of specific markers that distinguish between smooth muscle cells and pericytes, and the technical difficulty of assessing capillaries, the reported roles of these two cell populations in the healthy brain and under pathological conditions are quite controversial (for review^60,69,70). The functions of pericytes in the brain microvasculature are diverse: in addition to be supportive cells for the endothelium and thus, to participate in the control of the BBB integrity like astrocytes, pericytes are able to clear cell debris by phagocytosis, and can interact with endothelial cells to promote angiogenesis and neovascularization in a context of stroke (for review⁷⁰). The control of capillary blood flow by pericytes is, however, more controversial: previous reports suggested that capillary pericytes could contract or dilate capillaries to regulate brain blood flow,^71,72 but recent studies revealed that arteriolar smooth muscle cells, and not pericytes, regulate regional blood flow.^73,74 More recent evidences indicate that capillary pericytes maintain optimal flow through the capillary bed without active vessel diameter changes: using novel tools, including high-resolution intravital optical imaging, pericyte-specific dyes, calcium imaging, and single cell transcriptomic analysis, the analyses showed that brain pericytes are a distinct mural cell type and do not express the contractile protein smooth muscle actin (αSMA).⁷⁵

Recently, the dysfunction of pericytes in diseases such as stroke and AD has been highlighted (for review⁷⁶). In addition, pericytes (not necessarily brain pericytes) may play a key role in the pathogenesis of vascular disorders, including atherosclerosis: they could accumulate lipids, promote atherosclerotic plaque growth and vascularization, they participate in vascular remodeling, calcification, and thrombosis.⁶⁷ Pericytes may also express proinflammatory molecules and thus promote local inflammation, which plays a key role in atherosclerosis and stroke. On the other hand, it has been reported that in ApoE^−/− mice, hypercholesterolemia can affect pericytes and their interactions with endothelial cells.⁷⁷ Indeed, hyperlipidemia dose-dependently attenuated pericyte coverage of brain endothelial cells, abolishing the improvement of CBF during subsequent stroke and increased brain infarction.⁷⁷ The impact of lipids and/or atherosclerosis on cerebral pericytes, however, deserves further investigations. Finally, some studies reported that pericytes can express adhesion and structural proteins and that changes in their expression may be related to the development of hypertension.⁷⁸ However, the exact role of brain microvascular pericytes in the pathogenesis of hypertension remains to be elucidated. On the other hand, it is possible that hypertension affects brain pericytes: an increase in blood pressure could evoke changes in the number, morphology, and function of brain pericytes (for review⁷⁹).

CONSEQUENCES OF ARTERIAL CHANGES ON CEREBRAL FUNCTION

While it is easily conceivable that strokes can lead to dementia because of the neuronal loss, similar to a myocardial infarction can lead to heart failure because of the loss of cardiomyocytes, the impact of a chronic cerebrovascular dysfunction on brain functions is much more difficult to quantify. Recently, however, the negative impact of age-related macro- and microvascular endothelial dysfunctions, and large artery stiffness (estimated by pulse wave velocity measurements) on cognitive performance was reported.⁸⁰ Arterial blood pressure and blood flow oscillate at each heartbeat, and the pressure gradient, i.e., the pulse pressure, translates into a pressure wave that propagates along the circulation from the aorta and carotids to the brain microcirculation (recently reviewed^81–83). To protect the microcirculation, pulsatile blood flow is normally converted into a low pressure and relative continuous capillary flow. The efficiency of this conversion is mostly ensured by the compliance of large peripheral elastic arteries that absorb the pulsatility generated by the cardiac cycle. The amplitude of the pulsatile pressure cycle is dampened starting in the aorta and carotid arteries,⁸⁴ followed, at the level of the resistance arteries, by the pressure-dependent myogenic response⁸⁵ that further reduces the pulse pressure gradient transferred to the microcirculation. With aging, however, the cerebral circulation becomes vulnerable to the pulse pressure: vascular stiffening of the large peripheral elastic arteries occurs,⁸⁶ impaired myogenic contractions⁸⁷ and endothelial dysfunctions arise.⁸⁸ The cerebrovascular endothelium likely takes center-stage in these age-related processes: it senses pulse pressure, regulating flow-mediated dilation and myogenic tone, at least ex vivo⁸⁹; any impairment in endothelial function will thus reduce both flow-mediated dilation and autoregulation of blood pressure, while, on the other hand, pathological increase in pulse pressure will compromise endothelial function. With aging, and precociously in the presence of risk factors for vascular diseases, pulse pressure increases and creates more mechanical stress as the pulse pressure wave propagates further along the fragile brain microcirculation.^82,90,91 The penetration of pulse pressure in the cerebral microvasculature is at the basis of the “biomechanical hypothesis” of brain damage,⁹² stipulating that cognitive decline originates from the pulse pressure-induced microvascular damage. A systematic review of 23 studies with a total of 15,666 patients supports this hypothesis showing that greater arterial stiffness was associated with markers of cerebral small vessel diseases.⁹³ More specifically, heart-to-carotid pulse wave velocity has been shown to be associated with greater Aβ deposition and this association was strongest in mild cognitive impairment patients.⁹⁴ On the other hand, carotid-to-femoral pulse wave velocity was not associated with AD or all cause dementia in a cross-sectional study⁹⁵ and not related to Aβ deposition.⁹⁴ Data from the Atherosclerosis Risk in Communities-Neurocognitive (ARIC) study showed that carotid-to-femoral pulse wave velocity is more associated with VaD than with AD,⁹⁶ while middle cerebral artery pulsatility at baseline is a significant and independent predictor of conversion to AD.⁹⁷ Consistent with these findings, carotid or intracerebral stiffness, distensibility, and pulsatility are more important for brain health than central aortic stiffness.⁹⁸ Altogether, the results of the ARIC study strongly suggest that carotid stiffness is more strongly related to cerebrovascular damage and dementia, whereas femoral stiffness is more related to coronary artery disease.

Although there are only indirect evidence that pulse pressure extends up to the penetrating arterioles, capillaries, and venules in the context of arterial stiffening, experimental evidence supports the deleterious impact of an increase in central pulse pressure on cerebrovascular and cognitive functions. A rise in pulse pressure is associated with hypertrophy of the cerebral vascular arterial wall,^99,100 and this vascular remodeling could contribute to the disruption of the neurovascular coupling (reviewed in¹⁰¹). One study reported an increase in the cerebrovascular resistance index in AD patients compared to mild cognitive impairment and noncognitively impaired patients, while in mild cognitive impairment patients that converted to AD, the cerebrovascular resistance index was higher.¹⁰² However, while vascular remodeling could contribute to cognitive decline, the increase in vascular resistance resulting from inward remodeling has several physiological effects, particularly in hypertension, as distal vessels and the BBB are, at least initially, protected from the increase in upstream pressure (for review⁷⁰).

Evidence for a pulse pressure penetration is also given by a pulsatile CBF and WM structure abnormalities observed in an elderly population,^103–105 and by the development of microhemorrhages in hypertensive old subjects.¹⁰⁶ As noted in humans, this chronic rise in pulse pressure is strongly associated with cognitive decline and dementia (reviewed in⁸¹). This association was experimentally demonstrated in mice.¹⁰⁷ Importantly, the disruption of the barrier function of the endothelium in the brain, the BBB, appears to be an early marker of cognitive impairment as recently reported by Zlokovic’s¹⁰⁸ and Iadecola’s groups.¹⁰⁹ Disturbed BBB function will favor parenchymal inflammation and inevitably, neuronal loss with time.¹¹⁰

Consequently, the term “pulse wave encephalopathy” is used to describe pulse pressure-induced microvascular dysfunctions implicated in cognitive failure observed in various brain diseases.^111–113 In AD patients, aortic stiffness index and pressure–strain elastic modulus were higher, showing that alteration in the vascular component is associated with the pathogenesis of the neurodegenerative disease,¹¹⁴ as hypothesized by Alzheimer himself.⁴ Transcranial Doppler is used routinely in patients to measure noninvasively mean flow velocity and its derived pulsatility index in the large arteries of the circle of Willis, mostly the middle cerebral artery¹¹⁵; when combined with neuropsychological tests, this technology permits to detect and diagnose dementia (for reviews^116,117). Multiple cross-sectional studies showed that patients with dementia (VaD, mild AD, and established AD) had systematically higher pulsatility indices.^118–121 In addition, few longitudinal studies showed that high pulsatility index at baseline correlated with cognitive impairment and that increased pulsatility index predicted aggravation of the dementia from mild cognitive impairment to AD.^97,122,123 The pulsatility index measured by transcranial Doppler in pial arteries is very useful, but it measures flow pulsatility and not pressure pulsatility. Baumbach and Heistad¹²⁴ followed by Faraci and Mayhan in Iowa¹²⁵ developed an open skull preparation to measure pulse pressure in first order rabbits and cat pial arterioles with a micropipette connected to the servo-null pressure device, that was then adapted to rats,^126–128 and more recently to mice.¹²⁹ In 35 µm internal diameter mouse pial arterioles, pulse pressure averaged 9 mm Hg.¹²⁹ The value of pulse pressure in cerebral capillaries or venules was not assessed. In hypertensive rats, the value of the cerebral arteriolar pulse pressure was increased.¹³⁰ Using optical coherence tomography to produce in vivo 3D angiograms and blood flow maps of cerebral vessels of the mouse brain, we reconstructed the flow profile over a single cardiac cycle in order to evaluate flow pulsatility (not pressure).¹³¹ We observed a trend toward increased flow pulsatility in small arterioles (<80 µm) in an atherosclerotic mouse model (LDLr^−/−; hApoB₁₀₀^+/+ mice) compared to wild-type mice.¹³¹ High-resolution optical imaging techniques such as in vivo two-photon imaging have provided for the first time direct visualization of brain capillary blood flow (not pressure) and its modulation by neural activity (for review⁷⁵). Nevertheless, transcranial Doppler of pial arteries remains the most used technique to measure CBF pulsatility, and a recent study showed that advancing age was associated with an accelerated increase in CBF pulsatility after midlife, and that higher CBF pulsatility was correlated with greater volume of white matter hyperintensity (i.e., the likely severity of cerebral small vessel disease) in older adults.¹³²

Despite these data showing that the diagnosis and even the prediction of cerebral small vessel disease and dementia can be established on the basis of abnormal cerebrovascular hemodynamics, the mechanisms underlying this association are not known.

DEMONSTRATION OF THE BIOMECHANICAL HYPOTHESIS

Clinical studies associate the rise in central pulse pressure^106,133 and in cerebrovascular pulsatility^116,117 with cognitive decline. Nonetheless, the pathological continuum of events from the rise in central pulse pressure to the impaired cerebral microcirculation and brain functions observed with age and with vascular risk factors has never been clearly demonstrated experimentally in one and the same model. Few studies reported a link between an increased pulse pressure and vascular damage, but either they did not include cognitive assessment,¹³⁴ the studies were performed in cultured cells exposed to abnormal flow patterns,¹³⁵ or the data were based on algorithms calculations.²⁹ One study showed that brachial endothelial dilatory function was impaired in patients with AD, but pulsatility was not assessed.¹³⁶ Very recently, we designed two studies in mice to characterize in vivo the impact of an increase in pulse pressure on the structure/function relationship of both the cerebral macro- and microvasculature and the brain, in order to test whether and how pulse pressure caused cognitive decline.^137,138 A unilateral rise in carotid pulse pressure was achieved by transaortic constriction (TAC) surgery, which specifically induces vascular damages in the ipsilateral (right) side of the brain and allows the comparison with the contralateral (left) control side. Other groups previously used this TAC model to primarily induce hypertension, and demonstrated vascular damage (including BBB alteration),^139,140 neuroinflammation,¹⁰⁷ and cognitive decline.¹⁴¹ We added to this knowledge by showing that the cerebral microvasculature was specifically affected by the mechanical stress induced by a unilateral rise in carotid pulse pressure, and that the cumulative damage in both the vascular and parenchymal compartments was associated with a cognitive decline: in wild-type TAC mice, we detected in the ispilateral side only, a severe endothelial dysfunction, a loss in the expression of gap junction proteins supporting endothelial cells disconnection, increased BBB permeability and appearance of microbleeds, and a decrease in capillary density paralleled by a rise in endothelial apoptosis.¹³⁷ In response to this unilateral vascular stress, we measured a cerebral hypoperfusion, parenchymal inflammation in astrocytes and cellular senescence, suggestive of an impaired NVU.¹³⁷ All these TAC-induced events were associated with the induction of a cognitive decline (learning and spatial memories)¹³⁷ (Figure 2). Interestingly, in the AD APP/presenilin-1 mouse model coexpressing the KM670/671NL “Swedish” mutated amyloid precursor protein (APP) and the Leu to Pro mutated presenilin-1, TAC not only exacerbated endothelial dysfunction, microbleeds, and Aβ deposition, but also the cognitive dysfunctions,¹³⁷ demonstrating that the mechanical stress associated with the rise in carotid pulse pressure magnified the cognitive decline induced by the genetic mutations. In a murine model of arterial stiffness induced by carotid calcification, which augmented CBF pulsatility,¹⁴² the impaired function of the BBB and the neurovascular uncoupling preceded cognitive impairment.¹⁴³ In another study,¹³⁸ we used 12-month-old severely dyslipidemic LDLR^−/−;hApoB₁₀₀^+/+ mice, that are also hypertensive, spontaneously develop extracranial but not intracranial atherosclerotic plaques from the age of 6 months,¹⁴⁴ show severe cerebral endothelial dysfunction¹⁴⁵ together with a reduced CBF, and exhibit cognitive decline from the age of 6 months.^146,147 Intracranial atherosclerotic lesions are in fact rarely observed in rodent cerebral arteries,^145,148,149 unlike in humans where intracranial atherosclerosis is present and progresses with age (for review⁷⁰), but at a slower rate than extracranial atherosclerosis.¹⁵⁰ We observed brain atrophy in 12-month-old LDLR^−/−;hApoB₁₀₀^+/+ mice, with cortical microbleeds and cerebral hypoperfusion; TAC worsened all vascular dysfunctions, exacerbated the endothelial dysfunction, the occurrence of microbleeds, neuroinflammation, and oxidative stress, and further impaired cognitive dysfunctions.¹³⁸ Thus, globally, these studies support the concept that vascular dysfunction, possibly triggered by carotid artery stiffening and magnified by risk factors for vascular diseases, is a key primary step in the impairment of cognitive functions, and confirm the multifactorial data-driven analysis demonstrating that vascular dysfunction was the first event leading to disease progression in late onset AD.²⁹

Figure 2.

Putative mechanisms induced by vascular stresses of different amplitudes. In the model of pressure overload (TAC) in mice, in the ipsilateral side of the brain where the increased pulse pressure is the highest, propagation of the pulse wave in the microcirculation of the brain promotes endothelial cells apoptosis and capillary rarefaction; this is associated with the appearance of senescent cells in the parenchyma, possibly as a response to the endothelial damages and the associated dysfunctions. In the contralateral side of the brain, endothelial function is preserved, no apoptosis is observed, but spreading of the (attenuated) vascular stress reaching the microcirculation on the other side of the brain leads to vascular senescence. In the LDLr^−/−;hApoB₁₀₀^+/+ mouse model, characterized by age-dependent stiffened carotids, hypertension, and severe dyslipidemia, the associated vascular stress imposed to the brain microcirculation leads to vascular senescence. Thus, an acute high vascular stress could induce cell death by apoptosis, whereas a chronic (age-related) milder vascular stress could promote senescence; the latter could permit a compensatory, but temporary, response to the stress. TAC mice data were obtained in ref. ¹³⁷ and the data from LDLr^−/−;hApoB₁₀₀^+/+ mice were reported in ref. ¹³⁸. Abbreviations: PP, pulse pressure; TAC, transaortic constriction.

To our knowledge, no study investigated the impact of carotid or cerebral pulse pressure on pericytes. However, since (i) adaptive changes of pericytes could be pressure dependent^151,152 and (ii) pericytes sense applied mechanical force,¹⁵³ it is plausible that pulsatile pressure affects pericytes and thus contributes to neurovascular uncoupling. Collectively, these studies are highly suggestive of a continuum between elevated carotid pulse pressure and cognitive decline, strengthening the biomechanical hypothesis of pulse-induced neuronal damage. Nonetheless, key questions remain to be answered.

OUTSTANDING QUESTIONS

By which mechanism a rise in carotid pulse pressure initiates its pathological effect in the brain?

The vascular endothelium is the first cellular component of the vascular wall exposed to various hemodynamic forces, including cyclic shear stress and pulsatility-dependent stretch. In physiological conditions, the mechanical stimulus inherent to these forces is beneficial and maintains the function of the vascular endothelium.^89,154 However, when hemodynamic forces acutely exceed physiological values, the abnormal mechanical stimulation of the endothelium reduces flow-mediated dilation⁸⁹; chronically, this deleterious hemodynamic environment leads to endothelial dysfunction, inflammation, and cerebrovascular wall remodeling.^{45,89,144,154} In our mouse model of TAC, we noted that the mechanisms of damage are dependent on the level of mechanical stress (Figure 2): in the side of the brain exposed to the abrupt increase in pulse pressure, the capillary loss was dependent on an increase in endothelial cell apoptosis that was paralleled with the appearance of senescence in the brain parenchyma.¹³⁷ In contrast, in the contralateral side of the brain where the hemodynamic stress was mostly contained over the 6-week TAC period (as shown by the preserved endothelial function in large pial arteries), the appearance of senescent cells in the brain microcirculation was noted, while capillary density was maintained (no apoptosis), and no senescence signal was detected in the brain parenchyma.¹³⁷ Thus, dependently on the level of mechanical stress, endothelial apoptosis (high stress, microvascular death) or senescence (low stress, maintenance of microvascular architecture) was observed. The consequences of the TAC surgery observed in the contralateral side of the brain may mimic the slow rise in systolic blood pressure observed with aging and in younger patients with hypertension.¹⁵⁵ Similarly, in naïve 1-year-old hypertensive and dyslipidemic LDLR^−/−;hApoB₁₀₀^+/+ mice, senescence of the brain microcirculation predominates without signs of apoptosis,¹³⁸ demonstrating that the chronic and progressive exposure to risk factors for vascular diseases damages endothelial cells but does not induce their death. Instead, damage endothelial cells likely replicate, and eventually reach replicative senescence with time.¹⁵⁶ Thus, appearance of senescent cells in the microcirculation is not initially associated with a disintegration of the architecture of the microcirculation but likely contributes to maintain local brain perfusion (Figure 2). This is not a sustainable damage/repair response, failure ultimately occurs and dysfunctions appear.

As recently reviewed,¹⁵⁵ the accumulation of senescent endothelial cells has been reported in hypertensive animal models: if initially senescence is likely a defense mechanism protecting the brain as we observed in mice,¹³⁷ the chronic exposure to hypertension in combination with dyslipidemia observed in naïve 1-year-old LDLR^−/−;hApoB₁₀₀^+/+ mice is ultimately deleterious as evidenced by a decrease in endothelial function, impaired BBB and appearance of microbleeds.¹³⁸ Senescent cells remain metabolically active¹⁵⁷ and release numerous factors of the Senescence-Associated Secretory Phenotype (SASP)^158,159 potentially harmful when senescent cells accumulate.¹⁶⁰ SASP factors become deleterious by contributing to chronic inflammation and by inducing senescence of neighboring cells^161,162; conversely, elimination of senescent cells by senolytic (selective therapeutic elimination of senescent cells) approaches are beneficial,¹⁶² including in the brain.⁶⁶ It remains to be tested whether senolytics could slow the damage induced by chronic systolic hypertension by regenerating the cerebrovascular endothelium,¹⁶³ but also glial cells⁶⁶ and astrocytes,¹⁶⁴ and delay VaD in an older population. This would be the ultimate demonstration that accumulation of senescent cells in the NVU is the mechanism by which a chronic rise in pulse pressure initiates its pathological effect in the brain.

How cerebral vascular remodeling and endothelial dysfunction affect cognitive functions?

Chronic endothelial dysfunction and the associated BBB defects ultimately weaken the cerebrovascular structure by degeneration and thinning of the arterial wall, leading to the rupture of the vessel and thus to microhemorrhages and microinfarcts^45,154 (Figure 1). Consequently to the vascular damage, chronic cerebral hypoperfusion appears,⁴⁵ which is predictive of dementia.¹⁶⁵ Brain damages are indeed partly caused by hypoxic/ischemic and/or hemorrhagic events, as observed in VaD,^166,167 in AD^26,168 and as estimated by algorithmic modeling.²⁹ The first measurable consequence of the cumulative damage to the brain circulation is possibly a perturbation in the BBB integrity, a function of the cerebral endothelium and interconnected supporting cells. As recently reviewed,^56,110 dysfunction of the BBB in disease states leads to leakages of harmful blood components into the brain parenchyma, cellular infiltration, and aberrant transport and clearance of molecules, which is associated with maladaptation and reduction in CBF that contribute to cognitive deficits. Recently, it has also been shown that individuals with early cognitive dysfunction show signs of brain capillary damage and BBB deregulation (through increased in both cerebrospinal fluid pericyte-derived soluble platelet-derived growth factor receptor-β levels and hippocampal BBB permeability using dynamic contrast-enhanced magnetic resonance imaging) in the hippocampus irrespective of the load in the AD biomarkers Aβ and/or tau.¹⁰⁸ Such initial endothelial permeability defects have also the potential to produce systemic dysfunctions: aberrant signaling may reset autonomic nervous system sensibility, modify blood pressure regulation, generate metabolic unbalance, changes that could all magnify cerebrovascular and neuronal stress. Therefore, early prevention of the loss of integrity of the NVU may be key to the success of future intervention aiming at preserving brain health.

Are all neurological and brain diseases associated with vascular dysfunctions?

In humans, the presence of vascular lesions and abnormalities has been reported in most neurodegenerative diseases and even psychiatric diseases: in frontotemporal dementia, for example, microhemorrhages,¹⁶⁹ BBB rupture,¹⁷⁰ and cerebral hypoperfusion¹⁷¹ have been observed. Similarly, lateral amyotrophic sclerosis,¹⁷² multiple sclerosis,¹⁷³ Parkinson disease,¹⁷⁴ Huntington disease,¹⁷⁵ epilepsy,¹⁷⁶ bipolar disorder,¹⁷⁷ depression,¹⁷⁸ autism,¹⁷⁹ and schizophrenia¹⁸⁰ are all characterized by some cerebrovascular abnormalities. These abnormalities include cerebral infarct and vascular lesions in Parkinsonism,¹⁸¹ BBB dysfunction in epileptic¹⁸² and bipolar¹⁷⁷ patients. Such vascular defects, via inflammatory and hypoxic responses, are deleterious to the neuronal function. Nevertheless, the contribution of these vascular dysfunctions in the incidence of these pathologies is relatively poorly studied. Hence, as for AD, it is possible that the cerebrovascular weaknesses increase the susceptibility of a subject at risk to develop the neurological disease.

Is an increase in pulse pressure associated with various neurological and psychiatric diseases?

Unfortunately, pulse pressure (i.e., central/carotid pulse pressure or any index of cerebrovascular pulsatility) is not usually investigated in patients with neurological or psychiatric disorders; thus, the biomechanical hypothesis, by which the pulse destroys the brain, remains to be validated in these diseases. Nevertheless, it has been suggested that in patients with multiple sclerosis, a pulsatile encephalopathy could trigger the initial degenerative process.¹⁸³ In addition, arterial stiffness measured as pulse wave velocity is associated with an increase in pulse pressure (for review⁸³) and therefore, only indirect conclusion concerning the potential contribution of pulse pressure to various neurological diseases could be made.

Do antihypertensive drugs reduce the risk of cognitive decline?

Excellent reviews on the subject reported that midlife hypertension predicts cognitive decline and the incidence of dementia in later age.^33,36,44 One of the first studies that demonstrated the predictive value of hypertension in VCID (VaD and AD) was the Honolulu-Asia Study, performed in Japanese-American men.¹⁸⁴ Interestingly, in the same cohort, hypertension was reported to be associated with NFTs and neuritic plaques,¹⁸⁵ suggesting that midlife hypertension is a major mediator or promoter of dementia.^36,186

The next most important question is whether the reduction of blood pressure by antihypertensive drugs, and thus diminution of mechanical stress, reduces the risk of dementia. Different clinical studies evaluated the potential of antihypertensive therapies on cognition. Positive studies include the beneficial effects of the calcium channel blocker nitrendipine in the Systolic Hypertension in Europe (Syst-Eur) study, a drug that reduced by 50% the risk of dementia within 2–4 years of follow-up in hypertensive patients with no dementia at baseline.^187,188 In the MOrbidity and mortality after Stroke, Eprosartan compared with nitrendipine for Secondary prevention (MOSES) study, the angiotensin II receptor antagonist eprosartan reduced by 50% the risk of dementia in high-risk hypertensive patients with cerebral events during the last 24 months,¹⁸⁹ and in the ONgoing Telmisartan Alone and in combination with Ramipril Global Endpoint Trial (ONTARGET) study, a combined treatment with telmisartan (angiotensin II receptor antagonist) and ramipril (angiotensin-converting enzyme inhibitor) decreased the risk of dementia by 9% in patients who had vascular disease or high-risk diabetes without heart failure.¹⁹⁰ In the Honolulu Asia Aging Study (HAAS) study, the longer the treatment with a β-adrenergic receptor blocker, the slower the onset of dementia in hypertensive men.¹⁹¹ Recently, the Systolic blood PRessure Intervention Trial-Memory and cognition IN Decreased hypertension (SPRINT-MIND) study, which objective was to evaluate the effect of intensive blood pressure control (target of <120 mm Hg) on risk of dementia in adults aged 50 years or older with hypertension but without diabetes or history of stroke, demonstrated that an antihypertensive intervention (major classes of antihypertensive drugs were used) reduced the risk of occurrence of a composite outcome of mild cognitive impairment and dementia.¹⁹² But negative results have also been reported, including in some randomized clinical trials (for review^36,186,193). The effects of antihypertensive drugs on the risk of cognitive decline, the debate concerning an optimal antihypertensive drug, blood pressure target, and time of treatment, have been discussed in several reviews (for example^36,44). It is important to note, however, that the negative results obtained in some randomized trials questioning the efficacy of blood pressure control to prevent dementia, may have to be reconsidered: multiple bias may explain these negative data, including patients heterogeneity (age, sex, and ethnicity), the choice of the antihypertensive drug, the intensive or standard blood pressure lowering, the duration of the follow-up as well as the sensitivity of the neuropsychological tests, as recently reviewed.³⁶ The question whether blood pressure lowering protects against cognitive decline is nonetheless still open, with encouraging data of the SPRINT-MIND trial.¹⁹²

Of note, even if clinical evidences suggest that some antihypertensive drugs reduced the incidence of cognitive decline, there is no proof that this beneficial effect was associated with a decrease in pulse pressure (carotid or cerebral pulse pressure). Future studies should incorporate measures of pulse wave velocity or pulsatility index, for example, to evaluate the contribution of this important vascular parameter on brain function with age. However, there is no clinically approved drug that “de-stiffens” the aorta and the carotids: the current antihypertensive therapies lower the diastolic pressure (consequently bringing down systolic pressure) through a decrease in peripheral resistance or blood volume. Nevertheless, the demonstration of a reduced incidence of dementia by antihypertensive treatments strengthens the concept that an abnormal mechanical stress causes cerebral dysfunctions, and indirectly validates the biomechanical hypothesis of the damages caused by the pulse pressure to the brain.

Could exhaustion of the lifelong adaptive mechanisms maintaining cerebrovascular function be the primary cause of VCID?

A parallel could be drawn between myocardial infarction-dependent heart failure, renal failure, and VCID: these medical conditions are the consequence of the respective loss of “contractile units,” “filtration units,” and “connecting units.” So far, it is not possible to replace either of these units (except by organ transplantation for the two first medical conditions), and death follows organ failure. It is logical to suggest a common origin of these diseases because they develop over decades and they share similar risk factors for cardiovascular diseases. The failure of the heart function is often qualified by the term “maladaptation,” that is, the consequence of the failure of the molecular adaptive responses over the years to compensate for the primary insult to the cardiomyocytes. Adaptations to environmental events are trade-offs governed by the genetic background, and the consequences of any insult are likely the recruitment of the best possible adaptive molecular pathways by the cells at the moment of this insult. This adaptive response likely becomes itself a causal signal for other molecular adaptions, changing the path of the homeostatic molecular environment necessary to reach a new cellular equilibrium. With time, accumulation of these adaptive molecular trade-offs leads to decompensation and failure (Figure 3). With this simple concept, it is easily understandable that the precocious is the insult, the premature is the fall of the first molecular adaptive domino that initiates the cascade of other adaptive molecular events; in theory, a premature stress, even minor, could hasten decompensation with the premature exhaustion of all possible molecular, structural and neurohumoral compensatory mechanisms, causing premature death.¹⁵⁶ All these adaptive changes initiated by an environmental stressor will, of course, depend on the genetic background of each individual, defining a unique “biological aging” trajectory, a concept in which the environment-to-genetic interaction from the fetal state to postnatal organ maturation determines the development of age-related chronic disease that eventually could lead to organ failure.¹⁹⁴ Barker was the first to demonstrate that the intrauterine environment associated with fetal growth restriction favored appearance of hypertension, obesity, and diabetes in adulthood and premature death by myocardial infarction.¹⁹⁵ In another setting, it has been shown that fetuses develop adaptive molecular responses to the stress of the assisted reproduction technology procedures, with consequences on cardio-metabolic health later in life: metabolic syndrome¹⁹⁶ and endothelial dysfunction, thickening of the carotid artery wall and an increase in pulse wave velocity¹⁹⁷ were observed in assisted reproduction technology teenagers. It is yet too early to assess the impact of assisted reproduction technology on age-related cognitive decline. Mechanistically, these phenotypic changes have been associated with early epigenetic responses presumably to compensate for the stress of the procedure.¹⁹⁸ Thus, to put it simply, early insults stimulate adaptive responses that may ultimately reveal dysfunction as age advances (Figure 3). Similarly, during teenage years, exposition to a poor life style increases the incidence of obesity with adverse health effects in adulthood.¹⁹⁹ With no surprise, middle-life hypertension has thus been shown to be a strong risk factor for older age cardiovascular diseases and VCID, as discussed above.^{38,184,185,200} Hence, failure of the brain (as that of the heart or the kidneys) may reflect its inability to keep functioning at the end of the recruitment of all possible backup mechanisms. We cannot replace the missing “connecting units” in the brain and in the absence of a deep knowledge of the function of the NVU and thus our ability to conceive efficient drugs, primary prevention, starting as early as possible in life, could be the key to slow the expected epidemics of dementia that is envisaged by the World Health Organization.²⁰¹

Figure 3.

Schematization of the concept of age-dependent organ damage. Top: Schematization of the individual, unique, pattern of age-dependent changes in systolic blood pressure, depending on lifetime cumulative high or low stresses. Bottom: Environment-to-genetic interaction from the fetal state to postnatal organ maturation could determine the development of age-related chronic cardiovascular diseases that eventually could lead to VCID. Adaptations to the environmental events are trade-offs governed by the genetic background, and the consequences of any insult is likely the recruitment of the best possible adaptive molecular pathways of the cells at the moment of this insult. This adaptive response becomes itself a causal signal for another adaptive molecular change downstream, changing the path of the cellular homeostatic molecular environment to reach a new equilibrium. With time, accumulation of these adaptive molecular trade-offs leads to decompensation and failure. Lifelong recruitment of adaptive, epigenically regulated, molecular pathways may change the course of cell function and cell interactions unique to each individual, a change that could have been triggered by one minor stressful stimulus in utero. In the case of the brain, failure of the adaptive and compensatory mechanisms could, ultimately, lead to dementia. Abbreviation: Epi, epigenetic.

Preclinical and clinical studies strongly suggest that vascular dysfunction affects brain structure and function. In humans, vascular cerebral lesions appear years or decades before the occurrence of neurological symptoms, indicating that the first step in the prevention of cognitive decline should be to prevent the vascular dysfunction. Targeting the pulse pressure, preventing its rise and penetration in the fragile microvasculature of the brain should delay and protect the brain against dementia. Given the clear deleterious impact of a vascular dysfunction on brain structure and function, the vascular component of various neurodegenerative diseases and even of psychiatric disorders should be better considered. This could allow new preventive approaches to delay the onset of cognitive decline. Ideally, through education programs and by using primary approaches or therapies that would precociously protect the vasculature, brain homeostasis would be preserved longer. In our aging societies, the social and economical burden associated with neurodegenerative diseases shows the absolute necessity to consider all therapeutic avenues to delay age-related cognitive decline.

FUNDING

This research is supported by the Canadian Institutes of Health Research (ET: PJT 166110 and PJT 162446), the Natural Sciences and Engineering Research Council of Canada (ET: RGPIN-2017-04770) and by the Foundation of the Montreal Heart Institute (ET).

DISCLOSURE

The authors declared no conflict of interest.