Vascular Contributions to Alzheimer’s Disease

Abstract

Alzheimer’s disease (AD) is the most common cause of dementia and is characterized by progressive neurodegeneration and cognitive decline. Understanding the pathophysiology underlying AD is paramount for the management of individuals at risk of and suffering from AD. The vascular hypothesis stipulates a relationship between cardiovascular disease and AD-related changes although the nature of this relationship remains unknown. In this review, we discuss several potential pathological pathways of vascular involvement in AD that have been described including dysregulation of neurovascular coupling, disruption of the blood brain barrier, and reduced clearance of metabolite waste such as beta-amyloid, a toxic peptide considered the hallmark of AD. We will also discuss the two-hit hypothesis which proposes a two-step positive feedback loop in which microvascular insults precede the accumulation of Aß and are thought to be at the origin of the disease development. At neuroimaging, signs of vascular dysfunction such as chronic cerebral hypoperfusion have been demonstrated, appearing early in AD, even before cognitive decline and alteration of traditional biomarkers. Cerebral small vessel disease such as cerebral amyloid angiopathy (CAA), characterized by the aggregation of Aß in the vessel wall, is highly prevalent in vascular dementia and AD patients. Current data is unclear whether cardiovascular disease causes, precipitates, amplifies, precedes, or simply coincides with AD. Targeted imaging tools to quantitatively evaluate the intracranial vasculature and longitudinal studies in individuals at risk for or in the early stages of the AD continuum could be critical in disentangling this complex relationship between vascular disease and AD.

Introduction

Alzheimer’s Disease (AD) is the most common cause of dementia and the sixth leading cause of death in the United States [1]. AD is characterized by slowly progressive neurodegeneration and cognitive decline that start many years before symptoms appear. The course of the disease corresponds to a continuum including three phases [1] : a preclinical phase, characterized by the absence of symptoms but showing early alterations of AD biomarkers; a prodromal phase, also known as mild cognitive impairment (MCI) and characterized by cognitive decline that has no or little impact on everyday activities; and finally dementia, the final phase of the disease in which individuals are unable to function independently due to moderate to severe cognitive impairment. In addition to the direct impact on affected individuals, AD is associated with high societal and economic impacts. In 2022, over 55 million people suffer from Alzheimer’s disease and related dementias (ADRD) worldwide. Given the changes in demographics and the global aging of the population, this number is predicted to reach 78 million in 2030 and 139 million in 2050 if no significant curative or preventive solutions are discovered [2].

While the clinical manifestations and epidemiology of AD are generally established, the exact cause and pathophysiology of AD remains elusive [3–5]. Numerous hypotheses have attempted to explain the factors underlying the pathogenesis of the disease, yet none of them has been fully conclusive. The most traditional cascade suggested to explain AD pathogenesis involves the extracellular deposition of an insoluble isoform of the protein amyloid-ß (Aß) into so-called senile plaques and the intracellular aggregation of hyperphosphorylated tau protein in neurofibrillary tangles (NFT) [6]. The accumulation of Aß has been initially associated with AD-related dementia but is actually found throughout the course of the disease, including in the preclinical phase [7]. According to the Aß hypothesis, deposition of abnormal Aß in the brain is thought to be the cause of AD-related changes, including NFT formation, neuronal loss, and cognitive decline [8–11]. The description of early-onset familial forms of AD supports this hypothesis. For instance, mutations of proteins involved in the Aß cascade such as amyloid precursor protein (APP), presenilin-1 (PSEN1), and presenilin-2 (PSEN2) are responsible for autosomal-dominant AD by the intermediate of increased Aß deposition that in turn triggers events leading to cognitive decline [12]. Though Aß was traditionally thought to drive the pathogenesis of AD, thus far treatment options aiming to reduce the production or deposition of Aß have failed to affect disease progression and symptoms compared with placebo [13]. For instance, the recent Food and Drug Administration (FDA)-approved anti-Aß monoclonal antibody, Aducanumab, did not demonstrate any significant efficacy in two different clinical trials [14].

Although the Aß/tau pathology is now considered the hallmark of AD, the exact causes and consequences of their accumulation in the brain remains unclear. It is unlikely that Aß is the sole contributor to AD pathogenesis [15]; instead, a multifactorial mechanism is most likely. In that regard, several other hypotheses have been suggested, though most are still debated [16–18]. Vascular contributions to cognitive impairment and decline (VCID) is the second most common cause of dementia [19]. Cerebrovascular disease is present in most individuals with dementia, though rarely a cause on its own. Vascular components like atherosclerosis, arteriolosclerosis, microinfarcts, and amyloid angiopathy play an important role alongside markers of neurodegeneration. It has been suggested that current limitations of nosology may be alleviated by addition of a vascular component to the amyloid/tau/neurodegeneration etiological classification system for dementia [20].

A vascular hypothesis for AD (VHAD) drew the attention of the scientific community in the last two decades. The VHAD postulates that there is a relationship between AD and cardiovascular disease, at both the systemic and cerebrovascular disease (CVD) levels [21–23]. Numerous studies have been carried out in an effort to clarify the nature of this relationship as there is an important overlap between AD and CVD in older patients [24–28]. A large body of evidence now indicates that vascular pathology is a major risk factor for AD dementia [29–35] and is associated with low scores in most cognitive domains [36]. Moreover, CVD contributes to neuronal loss in AD and AD-related Aß and tau pathology [28, 37–45] and has been shown to promote the conversion from MCI to dementia [46]. Despite all that evidence, the mechanisms underlying these associations are unknown and it remains unclear whether CVD causes, precipitates, amplifies, precedes, or simply coincides with AD [17, 47, 48]. The mechanisms underlying the VCID and VHAD hypotheses for the development of Alzheimer’s are depicted in Figure 1.

Figure 1-. Summary of the two prevailing hypotheses for the development of dementia.

The Vascular Cognitive Impairment and Dementia (VCID) hypothesis postulates that peripheral etiologies, such as cardiovascular factors, result in direct cerebral damage via strokes or cerebral hypoperfusion. Other peripheral etiologies that result in arterial stiffening are thought to lead to endothelial damage, BBB dysfunction, pericyte injury, and ultimately increased Aß accumulation and disruption in neuronal activity, leading to dementia. On the other hand, the Vascular Hypothesis for Alzheimer’s Dementia (VHAD) hypothesis proposes that underlying cardiovascular and cerebrovascular causes result in dementia. However, is unclear if both of these factors result in or are a consequence of the dementia.

(Adapted from : Zlokovic BV, Gottesman RF, Bernstein KE, Seshadri S, McKee A, Snyder H, et al. Vascular contributions to cognitive impairment and dementia (VCID): A report from the 2018 National Heart, Lung, and Blood Institute and National Institute of Neurological Disorders and Stroke Workshop.

The challenges are plentiful when it comes to testing the VHAD hypothesis. First, it is unclear whether CVD is the primum movens of AD pathogenesis (in which case a causal relationship might be postulated), or whether CVD, as observed in AD, is an inevitable consequence of the pathological features of AD, such as Aß deposition [49, 50]. A great example of the latter would be cerebral amyloid angiopathy (CAA), a small vessel disease often seen in AD and characterized by vascular insults due to Aß deposition in the vessel wall. Second, it is important to differentiate the role played by CVD in AD specifically from the more general implication of vascular dysfunction in dementia, a concept referred to under the umbrella term VCID [50]. Although an overlap might exist between these syndromes, there is a need to highlight the role of CVD in AD, specifically. An important step in testing the VCID hypothesis is to develop biomarkers for early diagnosis. Third, while cross-sectional and retrospective studies are important to demonstrate associations between vascular dysregulation and AD-related outcomes, there is a serious need for longitudinal studies aiming to find vascular biomarkers predictive of disease progression. There is evidence that vascular injuries are present early in the development of AD, even before the classical biomarkers of AD are altered [38, 48, 51]. One of the greatest challenges in the future decade will be to discover a vascular biomarker capable of identifying individuals at risk for AD, risk-stratifying patients already with cognitive impairment, and providing tools for uncovering AD pathogenesis and new treatment targets.

Despite the uncertainties surrounding VHAD (and more generally the vascular contributions to AD onset and progression), a substantial number of findings support its validity. In that regard, it has recently been suggested that biomarkers of vascular dysregulation be included in all AD-related observational and interventional research studies, in addition to the biomarker system proposed in the 2018 National Institute on Aging-Alzheimer’s Association (NIA-AA) Research Framework [52, 53]. This review aims to illustrate the vascular contributions to AD at three different levels: the cell, the brain, and the patient. Specifically, we will review the physiology of the neurovascular unit and explain how vascular disease can affect its function, cause neurodegeneration, and impede metabolite waste clearance. Second, we will explore the relationship between CVD and AD at the brain level and describe potential preclinical vascular biomarkers for AD. Finally, the third and final part of this paper will focus on the patient and what clinical studies still need to be performed to help disentangle these complex relationships, leaving room for discussion on the future directions in the field of AD prevention and risk assessment based on vascular biomarkers.

At the Cellular Level: From Physiology to Pathology

The cerebral vasculature plays an essential role in the physiology of the central nervous system (CNS). It is responsible for the delivery of oxygen and nutrients to the brain, the clearance of metabolic waste, and the exchange of cells and solute substrates between the intravascular compartment, the interstitial compartment, and the cerebrospinal fluid (CSF) [54, 55]. The neurovascular unit (NVU) is a complex functional structure located at the center of these physiological processes. It is composed of neurons, astrocytes, microglia, perivascular macrophages, endothelial cells, pericytes, vascular smooth muscle cells (vSMC), extracellular matrix, and immune cells [54] (Figure 2). These elements are linked together to form a coherent unit that regulates the interactions between the vascular, glial and neuronal components of the NVU. Some of the primary roles of the NVU are to ensure brain homeostasis, maintain the integrity of the blood brain barrier (BBB) and to regulate the cerebral blood flow (CBF) [56]. From a pathological perspective, the VHAD postulates that there is a relationship between vascular insults, neurodegeneration, neurovascular uncoupling, reduced CBF, and disrupted BBB. All these processes are intertwined in their contribution to the development and progression of AD. The pathological implication of CVD in AD and Aß-related disease is multifactorial; not one but multiple interconnected pathological pathways are involved. A feature shared by these pathways is the NVU. For instance, damage to the NVU results in a decrease in CBF, leading in turn to the depletion of oxygen and nutrients in the brain parenchyma, oxidative stress on brain cells, increased permeability of the BBB, and impaired clearance of neurotoxic metabolites such as Aß [56]. In this section, we will review the architecture and physiology of the cerebrovasculature and the NVU, explore involved pathological pathways, and outline their contributions to the pathogenesis of AD at the microscopic level.

Figure 2-. Structure of the neurovascular unit.

Penetrating cerebral arteries arise from pial arteries and dive into the brain parenchyma. They further divide into smaller arterioles that in turn give rise to brain capillaries, the smallest vascular component ensuring blood supply. Arterioles and capillaries interact with local cells of the central nervous system to form a complex functional system called the neurovascular unit (NVU). Structurally, the NVU is composed of neurons, astrocytes (in particular the end-feet), endothelial cells, basement membrane, and either vascular smooth muscle cells (in arterioles) or pericytes (in capillaries). All these elements are linked together to form a coherent unit that regulates closely the interactions between the vascular and nervous systems. The two primary roles of the NVU are to ensure the integrity of the blood brain barrier (BBB) and to regulate the cerebral blood flow (CBF). The blood brain barrier is an essential part of the NVU and plays a critical role in brain homeostasis. It is a semi-permeable, continuous, dynamic interface that separates the brain parenchyma from the circulating blood. In the brain, endothelial cells are non-fenestrated unlike in peripheral capillaries. They are also connected by tight junctions and adherens junctions, two important structural parts of the BBB that prevent undesirable exchanges of large molecules between bloodstream and brain parenchyma. Additionally, astrocytic end-feet cover the circumference of the luminal surface of vessels, where they act as an additional barrier and help in signaling pathways between local neurons and vascular structures. Vascular smooth muscle cells are only found in arterioles, whereas pericytes are found in capillaries. Both present contractile properties by which they contribute to the regulation of CBF and the phenomenon of neurovascular coupling. Pericytes also play a role in metabolite clearance and structural integrity of tight junctions.

Blood Brain Barrier

The BBB is an essential part of the NVU and plays a critical role in brain homeostasis [44]. Structurally, it is comprised of endothelial cells, pericytes, astrocyte end-feet, and the basement membrane. Together, these elements form a semi-permeable, dynamic interface that separates the brain parenchyma and the circulating blood while selectively controlling the exchange of substrates between the two compartments [49, 55]. The BBB prevents blood-borne inflammatory, toxic, or infectious compounds from entering the brain where they could cause deleterious effects and participate in the clearance of waste metabolites (e.g., Aß) from the brain [9, 57, 58]. Consequently, BBB breakdown results in damage on multiple levels and is associated with impaired cerebrovascular function and neurodegeneration [17, 45, 59–61].

Endothelial cells in the brain capillaries are non-fenestrated unlike peripheral endothelial cells, which means that the vascular wall is continuous. Endothelial cells are interconnected by tight junctions that guarantee a low paracellular permeability. Tight junctions are formed by transmembrane proteins such as claudin, occludin, and zona occludens-1 [62, 63]. Deficiency in claudin-5 (the most prevalent claudin isoform in the brain) was associated with BBB impairment in murine models [64]. Cell-cell adherens junctions are another type of connection existing between endothelial cells. They participate in the regulation of BBB permeability through their dynamic opening/closure, allowing the conditional passage of plasmatic compounds such as leukocytes [65, 66]. Adherens junctions are mainly formed by an endothelium-specific protein, vascular endothelial-cadherin, whose dysfunction leads to BBB impairment [67, 68]. Endothelial degeneration was reported in AD patients, in whom reduced capillary length and reduced expression of tight junction proteins were observed [44]. A decrease in the expression of tight junction proteins was also observed in CAA and coincided with the abnormal extravasation of fibrinogen in the brain compartment, a sign of BBB rupture [69, 70]. There is also evidence suggesting that the accumulation of Aß itself disrupts the BBB by altering the structure or expression of tight and adherens junctions [71, 72].

Transcellular and paracellular transport is low in the brain compared to other organs [73, 74]. Small molecules (<400 Da), including oxygen and carbon dioxide, can cross the BBB by transmembrane diffusion following their concentration gradients [75]. Specific transporters ensure the passage of larger molecules, such as glucose, amino acids, and nucleotides [76]. Glucose uptake in the brain relies on glucose transporter 1 (GLUT1), a transmembrane protein specifically located in endothelial cells of brain capillaries [63, 77]. The expression of GLUT1 is reduced in the brain of AD patients and mice models [78, 79]. Moreover, in mice overexpressing Aß precursor protein (APP), GLUT1 deficiency was associated with microvascular damage, BBB disruption, and worsened Aß pathology [79]. In addition to regulating the influx of plasmatic substrates, the BBB contributes to the efflux of waste products from the brain. Importantly, it participates in Aß clearance via selective transporters such as LDL receptor-related protein (LRP1) and receptor for advanced glycation end products (RAGE) [9, 57, 58]. Aß clearance across the BBB is a key concept in AD pathogenesis [63, 80, 81], as it was shown that Aß accumulation results from a deficiency in clearance rather than an overproduction of Aß [82]. Receptor-mediated transcytosis is the main drainage route for brain metabolites and ensures approximately 85% of Aß removal. LRP1 is one of the main pathways responsible for Aß clearance and is significantly reduced in AD [83, 84]. Other drainage pathways (such as the glymphatic hypothesis or the intramural peri-arterial drainage (IPAD) theory) have been described and may play a role in Aß clearance. However, their exact mechanism remains debated [85].

Pericytes are specialized, heterogeneous mural cells present in the basement membrane of capillaries [86]. The density of pericytes in the CNS is higher than in any other part of the body, indicating their unique role in the NVU [87, 88]. Through their close interactions with endothelial cells [89, 90] and contractile properties [91], pericytes contribute to the regulation of multiple microvascular functions, including the formation and maintenance of the BBB [92–94]. Pericyte migration, proliferation and recruitment are dependent on endothelium-secreted platelet-derived growth factor B (PDGF-B) which binds pericyte-specific platelet-derived growth factor receptor beta (PDGFRß). In transgenic mice with impaired PDGF-B/PDFFRß signaling, endothelial hyperplasia and higher accumulation of circulating macromolecules in the brain are observed [95]. In addition, pericytes contribute to the structural integrity of tight junctions [96–98]. Pericyte deficiency correlates with reduced expression of tight junction proteins resulting in BBB breakdown [95]. A recent study showed that pericyte-deficient mice had BBB disruption and white matter neurodegeneration [99]. Pericyte loss has also been associated with Aß-related pathology, including AD [63, 77, 100, 101] and CAA [102, 103]. AD patients show a lower abundance of brain pericytes compared to age-matched non-demented controls [104]. Moreover, there is a correlation between pericyte loss, Aß deposition, and BBB disruption in the cortex and hippocampi of AD patients [104]. A potential route explaining these phenomena is the toxicity of Aß on pericytes. Due to the presence of the receptor LRP1 on their surface, pericytes participate in Aß clearance and degradation under physiological conditions. However, at pathological concentrations of Aß, this system saturates and pericytes degenerate [97, 105]. Furthermore, due to their implication in Aß clearance, pericyte degeneration results in Aß accumulation, leading to a vicious circle where more Aß accumulates, causing more pericyte loss [97].

Astrocytes are star-shaped cells representing the most predominant type of glial cells. Astrocytes play a crucial role in the relationship between neurons and endothelium, and participate in the regulation of microvascular features, including the BBB [106–108]. They interact with the brain microvasculature through the tip of long cellular processes emanating from the cell body, called astrocytic end-feet [109–111]. These structures are a major component of the BBB: they are innervated by local neurons and surround the abluminal surface of capillaries [60]. Astrocytes contribute to the conformation of tight junctions and regulate transport of diverse molecules across the BBB [110, 112, 113]. In particular, the surface of astrocytic end-feet is populated by aquaporin 4 (AQP4), a transmembrane protein that promotes water transport and interstitial fluid/cerebrospinal fluid exchange and partially regulates the removal of metabolite waste such as Aß [114]. Deficiency of perivascular AQP4 is associated with AD and AD-related pathology such as Aß burden and CAA [115, 116]. Aß deposition also showed a negative effect on astrocytic function; for example, leading to abnormal distribution of AQP4 towards the cell body instead of by the end-feet [117].

Neurovascular Coupling

The cerebrovasculature provides the brain with the constant delivery of oxygen and nutrients required for normal neuronal function and ensures the removal of metabolic waste products such as carbon dioxide. The brain represents about 2% of the body weight but consumes close to 20% of the oxygen available [60]. The regulation of CBF is crucial to proper neuronal function, as illustrated by the fact that when the CBF stops, neurons die and brain damage occurs within minutes [118]. During neuronal activity, the NVU integrates signals issued by neurons and astrocytes and delivers them to the vascular system in order to vasodilate arterioles and capillaries, increase local CBF, and thus local distribution of oxygen and glucose [107]. This phenomenon is called functional hyperemia and relies on a broader physiological process, namely neurovascular coupling. The purpose of neurovascular coupling is to match the blood supply with the metabolic demand associated with neuronal activity [119]. Neurons, astrocytes, endothelial cells, vSMC, and pericytes all play a role in this process through multiple physiological mechanisms described in detail elsewhere [120]. One of the main signaling pathways responsible for neurovascular coupling involves nitric oxide (NO), a biological vasodilatory agent released by endothelial cells, neurons and astrocytes which acts on the endothelium and vSMC. Deficiency in this system leads to poor or nonexistent neurovascular coupling. As an example, the specific blockade of neuronal nitric oxide synthase (nNOS), the enzyme that produces NO, accounted for a decrease of 64% in neurovascular response compared to controls [60, 107, 121]. Neurovascular uncoupling (i.e., impairment of neurovascular coupling) has been widely reported in AD [119, 122–125]. As a proper flow regulation is crucial to neuronal function and brain homeostasis, it is not surprising to observe that the dysfunction of this regulatory system leads to neurodegeneration. These findings align perfectly with the VHAD and the measurement of CBF autoregulation is a primary outcome of numerous AD-related studies [126, 127]. Despite the evidence describing this relationship, the exact mechanisms that link neurovascular uncoupling with cognitive impairment remain unclear to date [128]. Many pathways have been suggested, most of them involving damage to the NVU and its individual components [120].

In vivo studies showed that pericytes maintain the vascular tone of capillaries by both vasodilatory and vasoconstrictive properties [119, 129–131]. By this function, pericytes (especially those located closer to the arterioles) contribute to CBF regulation and neurovascular coupling [132–134]. Aß deposition was associated with excessive pericyte constriction via oxidative stress in AD mouse models and human brain slices [135]. This could be a pathway explaining chronic hypoperfusion and neurovascular dysregulation in AD [86].

Arteries and arterioles (providing the blood supply upstream to the capillary bed) are supported by vSMC, a type of contractile cell essential to hemodynamic regulation. Both pericytes and vSMC interact closely with the other components of the NVU to control vascular tone and diameter [120]. Activated neurons and neighboring astrocytes trigger the release of a series of signaling molecules (such as NO) that cause vSMC to relax and arterioles to dilate in response to neuronal activity, therefore increasing CBF [60, 120]. Aß exerts deleterious effects on vSMC, resulting in a state of hypercontractility that forces arterioles into excessive vasoconstriction, therefore reducing their capacity to vary their diameter in order to increase CBF [136, 137].

Astrocytes also modulate blood flow through dynamic intracellular Ca²⁺ cascades occurring in the astrocytic end-feet. Increase in intracellular astrocytic Ca²⁺ in response to neural activation contributes directly to neurovascular coupling by modifying vascular tone [138]. In murine models of AD, astrocytes exhibited increased levels of resting calcium [139] and abnormal calcium response after electric field stimulation [140], suggesting a dysfunction of the neurogliovascular tripartite system.

Two-Hit Vascular Hypothesis

In this section, we described many biological pathways that may play an important role in the VHAD. However, to date, not one study was able to pinpoint a sole pathological cascade and prove that it was the cause of AD. The two-hit vascular hypothesis (Figure 3) proposes an overall simplified mechanism by which AD and CVD are intertwined. According to this hypothesis [45], there is a two-step positive feedback loop involving Aß that could explain AD pathogenesis and progression. The first step (hit 1) is Aß-independent and consists of microvascular insults, such as BBB disruption and CBF reduction, resulting in vascular-mediated neuronal dysfunction, hypoxic lesions, and leakage of deleterious waste metabolites, such as Aß. The second step (hit 2) is Aß-dependent and starts with the pathological accumulation of Aß, as a result of hit 1. Since Aß exhibits vasculotoxic properties, its pathological accumulation in the brain and the vascular system leads to more neuronal dysfunction and microvascular insults. This second hit thus results in more vascular damage, and the vicious circle goes on. It is possible that the positive feedback loop described above is the reason for rapid progression from MCI or early AD to AD-related dementia [141]. In an animal study, BBB disruption was found to precede formation of senile plaques and cognitive impairment, supporting the two-hit hypothesis [142]. In addition, according to this hypothesis, tau pathology occurs after the vascular insults and the amyloidogenic pathway [45]. More longitudinal studies with vascular specific biomarkers are needed to further elucidate these complex relationships.

Figure 3-. Summary of the two-hit hypothesis.

The two-hit hypothesis proposes to explain the relationship between cardiovascular disease and Alzheimer’s disease (AD). It claims that there is a two-step positive feedback loop involving beta-amyloid (Aß) and attempts to explain how it plays a role in AD pathogenesis and progression. The first step (hit 1) is Aß-independent and consists of microvascular insults, such as blood brain barrier (BBB) disruption and cerebral blood flow (CBF) reduction, resulting in vascular-mediated neuronal dysfunction, hypoxic lesions, and leakage of deleterious waste metabolites such as Aß. The second step (hit 2) is Aß-dependent and starts with the pathological accumulation of Aß, as a result of hit 1. The toxic effect of Aß on brain tissue results in neurodegeneration, translated clinically in progressive cognitive decline and eventually, dementia. In addition, since Aß also exhibits vasculotoxic properties, its pathological accumulation in the brain and the vascular system leads to more microvascular insults. This positive feedback loop therefore leads to more vascular dysfunction, itself causing more Aß accumulation, and so forth.

At the Brain Level: Imaging Early Vascular Biomarkers

The cellular and microvascular mechanisms discussed above are the main pillars supporting the VHAD. However, foundational research, especially animal studies, must be translated into humans to be applicable to clinical practice. Importantly, human brain imaging is a valuable asset in the arsenal of methods available for clinical investigations. In particular, neuroimaging including advanced magnetic resonance imaging (MRI) techniques, which can detect location-specific atrophy associated with Alzheimer’s, and amyloid- specific positron emission tomography (PET), which can detect fibrillar amyloid-β in neuritic plaques, in vivo, are valuable when it comes to the exploration of AD-related cerebrovascular dysfunction [143, 144]. Both structural and functional MRI have discovered interesting findings in the AD/CVD relationship, thereby improving our understanding of the pathological pathways introduced in the previous section. The recent years have been marked by an increased interest in vascular biomarkers for AD, with some authors suggesting vascular biomarkers be added to the Amyloid/Tau/Neurodegeneration (ATN) biomarker system [53]. In many respects, human brain studies have challenged the VHAD while bringing considerable momentum towards the development of early AD vascular biomarkers [21]. In this section, we will explore neuroimaging-based findings related to VHAD and bridge concepts from physiological, preclinical, and clinical studies.

Cerebral Hypoperfusion

Chronic cerebral hypoperfusion (CCH) is considered the earliest and most prominent pathological change related to the development and the progression of AD [21] (Figure 4). The relationship between AD-related neurodegenerative changes and CCH was first hypothesized approximately three decades ago [22]. Whereas causality is hard to establish, it is reasonable to say that CCH represents an early preclinical biomarker of AD that shows promising results in predicting the trajectory of the disease. The lower brain metabolism in AD might contribute in part to the reduction of CBF [145], but the fact that vascular dysfunction precedes neurodegeneration [36] indicates that brain atrophy and reduced cerebral metabolism are not the sole contributors to reduced CBF [120].

Figure 4-. Example of cerebral hypoperfusion.

Reduction of the cerebral blood flow (CBF), also called hypoperfusion, is one of the most prominent and earliest signs of vascular dysregulation in Alzheimer’s disease (AD). In this patient with AD-related dementia, arterial spin labeling (ASL) magnetic resonance imaging (MRI) shows bilateral occipital hypoperfusion (arrows).

In a large-cohort longitudinal study using arterial spin labeling (ASL) MRI in AD, decrease of CBF was shown to precede detectable Aß and tau proteins [38]. Similarly, in a large population-based study of AD-type dementia, greater CBF was associated with lower risk of dementia [36]; and among non-demented subjects, the risk of cognitive decline was significantly lower in subjects with higher CBF, and subjects with higher CBF had higher hippocampal and amygdala volumes [36]. Macrovascular hemodynamics, i.e. volumetric flow rate in cerebral arteries, exhibits a similar relationship with AD-related changes. In a 4D MRI study, flow in internal carotid arteries (ICA) and middle cerebral arteries (MCA) predicted a higher degree of atrophy and amyloid deposition was associated with decreased flow [146]. On the contrary, no correlation was found between blood flow and tau tangle load [146]. These findings have been replicated in a second 4D flow MRI study using AD compared to age-matched and middle-aged control groups [147]. The early alteration of CBF in AD makes it an excellent biomarker to identify patients at increased risk of AD or to predict the risk of conversion from MCI to dementia [46, 148, 149]. In a follow-up study of participants suffering from MCI, lower baseline cerebral perfusion in the right inferior parietal and the right middle frontal lobes was associated with a higher risk of conversion to dementia [150]. Additionally, the comparison of MCI patients with middle-age controls suggested reduced CBF in the MCI group, highlighting the early alteration of vascular functions in AD [147].

Cerebral Small Vessel Disease

In both vascular dementia and AD, cerebral small vessel disease (CSVD) is highly prevalent and is thought to contribute to cognitive impairment [151–157]. Signs of CSVD on MRI include white matter hyperintensities (WMH), microbleeds, microinfarcts, lacunes, and enlarged perivascular spaces [158]. A type of CSVD associated with aging and cardiovascular disease risk factors, as well as closely associated with AD, is cerebral amyloid angiopathy (CAA), the vascular form of amyloidopathy. Its prevalence in older adults without dementia is around 30–40% and increases with age after 65 years. Signs of CAA include peripheral parenchymal microhemorrhages/hemorrhages, subarachnoid hemorrhage, and leukoencephalopathy [159, 160]. In AD populations, over 80% of patients show signs of CAA, suggesting a strong association between the two pathologies [161–165]. CAA is characterized by the deposition of Aß in the vessel wall, mainly in cortical and meningeal microvessels, where the abnormal protein exerts deleterious effects and causes hypoperfusion and BBB disruption [166, 167]. The presence of advanced CAA in AD correlates with greater cognitive impairment [168] and faster cognitive decline [169]. According to the two-hit hypothesis, CAA corresponds best to a comorbidity of AD that occurs after the advent of cardiovascular risk factors and the failure of Aß clearance. It would therefore contribute to neurodegeneration only in the late stage of the disease [170]. To support this claim, it was shown that CAA-related damages mostly occur in the neocortex and tend to spare the hippocampus in the early phase of AD [171].

WMH are an important feature of CSVD and correspond to bilateral, symmetrical hyperintense lesions visible on T2-weighted MRI [172] (Figure 5). White matter is more vulnerable to ischemic conditions compared to grey matter, especially at advanced age [173]. In the case of CSVD, several factors underlie the pathology of WMH, including hypoperfusion, BBB breakdown, gliosis, and demyelination [174, 175]. A higher degree of WMH (particularly periventricular WMH [176]) is associated with loss of cognitive functions and accelerated cognitive decline [177, 178]. In addition, the presence of WMH is associated with a decrease of cognitive functioning in AD and MCI [179], and a higher probability of disease progression [158]. Two recent meta-analyses confirmed this trend and showed that higher WMH burden increased the overall risk of AD [152, 172]. It was reported that WMH were early and independent predictors of AD, with one group estimating onset up to 10 years before symptoms appear [180]. Individual studies also found that a higher WMH burden correlated with worse memory function in MCI patients, independent of Aß load [181], and with a higher degree of atrophy in the temporal lobe and the insula [182].

Figure 5-. White matter hyperintensities.

In this patient, periventricular white matter hyperintensities (WMH) are visible as bilateral, symmetrical hyperintense lesions on fluid attenuated inversion recovery (FLAIR) magnetic resonance imaging. Compared to grey matter, white matter is more vulnerable to the ischemic conditions related to hypoperfusion, especially at advanced age. WMH are associated with multiple causes of cognitive decline, including vascular dementia and Alzheimer’s disease. Pathological changes associated with WMH include gliosis, demyelination, and axonal loss. Non-vascular causes of WMH are also described, such as inflammatory (e.g., multiple sclerosis), infectious (e.g., human immunodeficiency virus, Lyme encephalitis), or toxic (e.g., chronic alcohol abuse) etiologies. Several classifications aim to quantify the burden of WMH, for instance the Fazekas score, the WMH volume (absolute volume of WMH calculated with (semi-)automated imaging tools), and the WMH ratio (relative to total intracranial volume).

Microbleeds are visible on MRI exams and represent small-sized (<10 mm) hemorrhages caused by the rupture of microvessels [183] (Figure 6). The aggregation of Aß in the wall of microvessels compromise their structural integrity and increase fragility [184]. Moreover, as damaged arterioles get stiffer and lose the ability to absorb the energy from the systolic pressure waves, downstream microvessels are confronted by higher pressures and rupture [184]. The prevalence of microbleeds in the AD population varies between studies with a pooled frequency of 23% (95% CI 17% to 31%) [185]. Although there is evidence that microbleeds are contributors to cognitive impairment in the general population across multiple cognitive domains [186, 187], the exact relationship between presence of microbleeds and AD remains unclear. A recent meta-analysis of 3 studies concluded that they do not impact the incidence of AD [188]. Other studies showed a correlation between microbleeds and severity of cognitive decline, atrophy and WMH [168, 189]. The distinction between lobar and deep microbleeds could play a role in the variability of these results. Most microbleeds found in patients with AD have a lobar predominance [190] and may correspond to CAA-related comorbid lesions [191].

Figure 6-. Microbleeds in cerebral amyloid angiopathy.

In this patient with cerebral amyloid angiopathy (CAA), susceptibility-weighted angiography (SWAN) was used to illustrate the extent of peripherally located microbleeds. SWAN is a high-resolution 3D magnetic resonance imaging (MRI) sequence that is highly sensitive to the presence of blood products. Although they represent a hallmark of CAA, their association with cognitive decline remains debated.

Cerebrovascular Reactivity

The ability of the cerebrovasculature to regulate CBF depending on neuronal activity (a phenomenon called neurovascular coupling) is crucial for proper brain function. Neurovascular coupling has been associated with several neurological disorders such as AD, amyotrophic lateral sclerosis, and stroke [60]. In addition, the cerebrovasculature has the ability to autoregulate CBF in response of external stimuli (hypercapnia, supine-to-standing position, pharmaceutical compounds): this phenomenon is called cerebrovascular reactivity (CVR). Most investigations dedicated to the study of CVR utilize carbon dioxide (CO₂) or breath-holding challenges to assess the subsequent increase of CBF. Using different imaging techniques such as MRI and transcranial doppler, studies showed that AD patients have a reduced cerebrovascular reactivity compared to cognitively normal individuals [192, 193]. A greater WMH volume is associated with a greater deficit of the cerebrovascular reactivity [193], highlighting the connection between CSVD-related microvascular insults and the lack of cerebrovascular autoregulation. In addition, aging (the primary risk factor for AD) has a negative effect on CVR as was shown in a recent 4D flow MRI study [194]. These findings could have an etiological significance in AD as CVR impairment with aging could be an early marker of neurological pathology.

Arterial Stiffness

Atherosclerosis and amyloid angiopathy both contribute to increased arterial rigidity, leading to luminal narrowing, microvascular distorsion, and weakening of the vessel wall [195, 196]. Pulsatility index (PI) is a marker of arterial stiffness and reflects the vessel compliance when confronted to the systolic pressure wave. Higher values of PI are associated with more rigid arteries, which is typically seen in normal aging. In a 4D flow MRI study, AD and MCI subjects showed higher PI values compared to age-matched and middle-aged cognitively normal controls [147]. In another study, higher PI in the MCA predicted a higher degree of atrophy, but PI was not significantly associated with CSF biomarkers [146]. Transcranial doppler showed that increased PI was associated with a clinical diagnosis of presumptive AD [197]. Pulse wave velocity (PWV) is another metric of arterial stiffness and describes how fast the pressure wave propagates in the artery after the cardiac systolic pulse (the more rigid, the higher the PWV value). It is typically measured as carotid-to-femoral ratios using applanation tonometry (cfPWV), but recent progress allows direct intracranial PWV measurement using high-temporal resolution 4D-flow MRI [198]. In the first study to use this technique to measure PWV, higher PWV values were observed in AD and MCI compared to age-matched controls, highlighting the higher vascular rigidity present in clinical and preclinical AD. Moreover, PWV was higher in middle-age subjects with ApoE4 genotype (a genetic risk factor for AD) and family history of AD, compared to controls. Findings suggest that arterial stiffness could be a risk factor for, or a preclinical biomarker for AD development in subjects at risk for AD.

BBB Permeability Imaging

Neuroimaging studies in the living human brain, post-mortem studies, and biomarker studies have demonstrated that BBB permeability increases with age, and that BBB dysfunction is accelerated in microvascular disease and dementia [39, 199, 200]. This breakdown allows for an influx of neurotoxic blood- derived debris, cells, and microbial pathogens initiating inflammatory and immune responses, which can in turn lead to multiple pathways of neurodegeneration [44]. One study reported evidence of BBB breakdown in the hippocampus before hippocampal atrophy [39], which is typically seen early in AD, raising the possibility that BBB breakdown might precede neurodegeneration [201, 202]. 3D diffusion-prepared ASL perfusion MRI is a promising BBB imaging technique that uses a safe, abundant, and endogenous tracer, magnetically labeled water. Unlike positron emission tomography (PET), which is both expensive and uses radiotracers, and dynamic contrast-enhanced (DCE) MRI, which requires gadolinium and suffers from low signal-to-noise ratio (SNR) [203–206], diffusion-prepared ASL may provide a more direct and sensitive biomarker of BBB function at the early stages of disease progression with a higher signal-to-noise ratio and higher sensitivity to small changes in water exchange across the BBB [207]. Diffusion-weighted ASL techniques have been proposed to differentiate the fraction of labeled water in capillary and brain tissue based on their distinctive pseudo-diffusion coefficients (high in capillary and low in tissue) [207–209].

At the Patient Level: A Brief Overview of the Next Steps…

The “AT(N)” biomarker system for AD relies on three components: Aß [A], tau [T], and neurodegeneration [N] [53]. Cardiovascular risk factors (CVRF) such as hypertension, diabetes, and obesity contribute to the clinical expression of AD, increase the Aß/tau burden and promote cognitive decline [210–213]. Given the strong relationship between vascular risk factors and AD, Sweeney et al. recently suggested that biomarkers of vascular disease be added to the AT(N) system [53]. Drawing insights from promising findings such as those summarized in the previous section(s) could be instrumental in the detection and tracking of vascular contributions to AD [53]. Much more research needs to be performed in humans at risk and in the early stages of AD and VCID to better understand potential biomarkers and their trajectory across the heterogeneity of AD.

Future Directions in Research

To advance our understanding of vascular contributions in AD, one step would be to better characterize markers of vascular dysfunction in vivo and ideally in humans, at multiple time points during a study. The present literature contains a dearth of strategic prospective studies using repetitive tracking of select biological, imaging and phenotypic measures. Although illuminating, the majority of research comes from post-hoc analyses of existing data. Though much can be derived from these rich datasets, retrospective study sometimes lacks the precision of measure variability and psychometric evaluation to interpret results. For example, one noteworthy disappointment emerging over the past decade is the poor reliability and irreproducibility of blood pressure measures attained across institutions, clinics, persons and devices [214, 215]. Prospective study to compare the emergence and progression of ADRD will require strategic designs comparing risk factors obtained in a standardized manner with outcomes from neuroimaging also analyzed in reproducible ways to document the cascade of vascular and neuronal changes occurring in VCID and VHAD. Among the many potential candidates for vascular biomarkers of AD and VCID, metrics of arterial stiffness (i.e., PI and PWV) and blood flow autoregulation (CVR) show promising results.

A second important point to consider in the research pipeline is the enrollment of normal older adults and patients in the prodromal phase of the disease, in order to study the earliest manifestations of AD and VCID and how these indicators interact. More than 20 years of exciting results have been collected from multiple preclinical cohorts. Among the main goals that these large studies carry out, there is a tremendous interest in determining the cognitive trajectory of patients during midlife and in differentiating normal aging from disease. The reason these longitudinal studies could have a great impact on how we conceive VCID and vascular contribution to AD is that they investigate sensitive biomarkers and their patterns before the onset of symptoms. Additionally, the conversion from MCI to dementia represents another time point worth investigating. Patients with MCI have limited cognitive decline that minimally impacts their day-to-day activities.

Many studies evaluating VCID (such as Mark VCID [216, 217]) have made great strides in studying the impact of CSVD on cognition, though the field could benefit from targeted studies of persons with autosomal dominant forms of VCID to reduce the extreme heterogeneity found in sporadic large grouped samples of vascular and Alzheimer’s dementias. As suggested by the National Plan to Address Alzheimer’s Disease at the Alzheimer’s Disease-Related Disorder 2019 Summit [218], research to observe monogenic conditions may help synergize efforts to disentangle vascular and AD components, just as other genetic disorders have impacted our knowledge of AD, PD, prion disease, amyotophic lateral sclerosis, Huntington disease [219, 220] and frontotemporal dementia [221–228]. One way to do this is to study genetic forms of CSVD, for instance Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy (CADASIL). CADASIL is caused by a mutation in the gene Notch3, which encodes a receptor protein expressed in vSMC and pericytes [229, 230]. When mutated, expression of the gene leads to generalized degeneration of vSMC affecting small and medium sized arteries. Notch3 mutations in CADASIL are autosomal dominant enabling the identification and study of participants at risk for VCID before the onset of related symptoms or other concomitant age-related pathologies [231–233]. In AD, it is often difficult to determine the independent effects of age-related vascular disease because it is frequently accompanied by confounding comorbidities. Through the study of Notch3 gene mutation carriers, prodromal vascular disease can be investigated and compared to its differential manifestation in prodromal AD (Aß positive, asymptomatic). Exploration of VCID by focusing on a rare heritable monogenic disorder, such as CADASIL, will advance knowledge of the full spectrum of this vascular dementia from asymptomatic gene carriers through dementia and will facilitate studies evaluating vascular contributions in different neurodegenerative processes including AD. Identifying key, targeted studies in VCID, and AD, such as this would vastly advance the field by providing insights into the degree and mechanisms of CVD (and more generally VCID) that lead to parenchymal damage and clinical deterioration, enabling a unique avenue to disentangle the contributions of CVD to neurodegeneration and cognitive impairment in AD-related dementia.

From the Lab to the Doctor’s Office

The concepts described in this review have not entirely made their way into the routine care of healthcare providers. Vascular contributors to AD seem to attract more and more interest from a research perspective, yet clinical implementation remains limited other than recommending a lifestyle to reduce vascular risk factors. Although most of the biomarkers discussed above have been associated with CSVD, AD, VCID, and other pathological entities, they are often not specific for blood vessel changes but are rather systemic vascular risk factors, and their sensitivity and specificity are relatively low in the prediction of AD development and progression. In line with that limitation, it is paramount to continue the efforts in the search for stronger vascular specific biomarkers for AD. Vascular specific biomarkers (such as pulse wave velocity and low frequency oscillations) are potential candidates as they are altered even prior to cognitive impairment. They could help in two ways: first, to risk stratify asymptomatic individuals with family history of AD or ApoE4 genotypes; second, to predict the disease trajectory of subjects suffering from MCI, in particular the conversion to dementia. Future efforts should be focused on establishing what biomarker would be the most sensitive and specific for these purposes.

Meanwhile, practitioners have a few opportunities in terms of advising patients on ways to mitigate vascular disease and potentially AD. For instance, strong control of modifiable cardiovascular risk factors and healthy lifestyles are associated with better cognition, preserved brain structure, and lower AD pathological burden [234]. Healthy lifestyle habits like physical activity and Mediterranean style diet have been associated with reduced risk of cognitive decline in aging adults [235–239], although recent studies failed to demonstrate this association in middle age [240, 241]. Modifiable risk factors (including hypertension, diabetes, dyslipidemia, and obesity) have been associated with cognitive decline and AD [242]. A strong relationship exists between type 2 diabetes and AD, with a higher risk of developing AD among the diabetic population (RR 1.43, 95% CI 1.25–1.62) [243]. Hypertension is a risk factor for both AD and cardiovascular disease. High blood pressure is the cause of a type of CSVD called hypertensive microangiopathy, characterized by similar imaging and pathological findings as CAA (i.e., microbleeds and WMHs) (Figure 7). A recent study indicated that hypertension was associated with an increased risk of dementia both at midlife and late-life [244]. The same study showed that, in normotensive middle-aged adults, a steep decrease of systolic blood pressure between mid-to-late life correlated with a two-fold higher risk of dementia. These results highlight the necessity to (1) maintain normal blood pressure during midlife and (2) avoid large variability of blood pressure/overtreatment in older adults [244]. Targeting the vascular contributors to AD through the strict management of these risk factors is an attractive and potentially useful way to potentially alter the risk of AD (in normal subjects) and the outcomes of the disease (in AD subjects) [50].

Figure 7-. Cerebral small vessel disease.

In both vascular dementia and Alzheimer’s disease (AD), cerebral small vessel disease (CSVD) is highly prevalent and is thought to contribute to cognitive impairment. In this example, this subject with diagnosed AD dementia has extensive, confluent white matter hyperintensities in addition to hippocampal volume loss. There is one microhemorrhage on SWI that is more centrally located in the white matter, more consistent with a microhemorrhage associated with hypertension than the peripherally located microhemorrhages more commonly seen in cerebral amyloid angiopathy.

Conclusion

Though the exact nature of this relationship remains unclear, vascular dysregulation and AD are interconnected in terms of physiology, pathology, imaging findings, and risk factors. Several pathways, including microvascular insults and dysfunction of the NVU, are hypothesized to contribute to AD-related vascular dysfunction. The identification of sensitive vascular biomarkers is a priority to identify patients at risk of cognitive decline and dementia. Chronic cerebral hypoperfusion and other metrics of cerebrovascular health are showing promising results, although more evidence needs to be obtained before drawing conclusions. Targeted, longitudinal studies evaluating vascular specific markers in at-risk, or early disease VCID, and AD populations are necessary to disentangle the complex relationship of vascular disease to dementia, and more specifically AD. However, current evidence argues for control of vascular risk factors as a potential first step in the prevention of AD-related changes and could be further implemented in a broader clinical system that aims to slow down AD progression.

Abbreviations

Aß

Beta-Amyloid

AD

Alzheimer’s Disease

APP

Amyloid Precursor Protein

AQP4

Aquaporin-4

ASL

Arterial Spin Labeling

BBB

Blood Brain Barrier

CAA

Cerebral Amyloid Angiopathy

CADASIL

Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy

CBF

Cerebral Blood Flow

CCH

Chronic Cerebral Hypoperfusion

cfPWV

Carotid-to-Femoral Pulse Wave Velocity

CSF

Cerebrospinal Fluid

CSVD

Cerebral Small Vessel Disease

CVD

Cerebrovascular Disease

CVR

Cerebrovascular Reactivity

CVRF

Cardiovascular Risk Factors

DCE

Dynamic Contrast Enhanced

GLUT1

Glucose Transporter 1

ICA

Internal Carotid Artery

MCA

Middle Cerebral Artery

MCI

Mild Cognitive Impairment

MRI

Magnetic Resonance Imaging

NFT

Neurofibrillary Tangles

NO

Nitric Oxide

NVU

Neurovascular Unit

PET

Positron Emission Tomography

PSEN

Presenilin

PWV

Pulse Wave Velocity

RAGE

Receptor for Advanced Glycation End Products

VCID

Vascular Cognitive Impairment and Dementia

VHAD

Vascular Hypothesis for Alzheimer’s Disease

vSMC

Vascular Smooth Muscular Cells

WMH

White Matter Hyperintensity