Abstract

Introduction

Age-related dementias are rapidly becoming one of the largest public health problems of our time. As baby boomers age - with more access to healthcare than ever before - Americans face a demographically larger population of elderly individuals. Alzheimer’s disease (AD) is the seventh leading cause of death in the United States, with numbers rising each year. In the absence of effective therapeutics, the population affected by AD is projected to triple, from 5.1 million today to 13.8 million people by 2050, with an estimated healthcare cost of more than $1 trillion [1]. Despite considerable efforts, the mechanism behind neurodegeneration remains unknown, driving a need to better understand not only AD, but all forms of dementia.

The first detailed description of dementia in the literature was at the end of the 19th century, as biological and medical knowledge was expanding [2]. It was originally thought that arteriosclerosis and chronic cerebral ischemia caused dementia [3]. This view changed with the discovery that infarcts, not chronic ischemia, were causing what eventually became known as multi-infarct dementia [4]. This evolved into the term vascular dementia, as multiple other pathological features of the disease became known, such as white matter hyperintensities, single infarcts, hemorrhages, and others. However, vascular dementia became overshadowed by the discovery of AD in 1898.

The German psychiatrist Alois Alzheimer was the first to describe senile plaques and neurofibrillary tangles in patients with dementia [5, 6]. The invention of the election microscope eventually helped elucidate the ultrastructure of neurofibrillary tangles and identified the amyloid core of the previously described senile plaques [7, 8]. As the field advanced with the discovery of amyloid-beta (Aβ) in the 1980s along with identification of APP mutations that cause familial AD, the focus of dementia studies shifted from examining vascular contributions to AD to the amyloid cascade hypothesis [9–11].

Currently, AD is the most common form of dementia, followed closely by vascular dementia. Vascular dementia, when it is thought of as a somewhat distinct entity, accounts for about 20% of all age-related dementias. A diagnosis of vascular dementia is commonly associated with certain risk factors such as obesity, hypertension, cardiac disease, and type 2 diabetes mellitus (T2DM). Over 40 million Americans aged 70 years or older have at least one of these risk factors, yet we know relatively little about how these factors contribute to cognitive decline [12]. Recent debate has centered on the role of cerebrovascular disease in dementia, both as a primary cause of cognitive impairment, and also as a contributing factor to dementia in combination with other pathologies. This has led to the adoption of a range of new terminologies in the field of dementia research, one of which is the umbrella term of vascular contributions to cognitive impairment and dementia (VCID).

VCID is ambiguous in that it can describe any clinical cognitive disorder of cerebrovascular origin. VCID therefore does not denote a specific disease, but rather a heterogeneous disease state under the larger umbrella of cerebrovascular disease [13, 14]. Past definitions of VCID (which has also been called VCI) used multi-infarct dementia or vascular dementia constructs to define a tentative diagnostic threshold [15]. Recent definitions have expanded to cover a continuum of the interactions, from “pure” AD pathology all the way to “pure” vascular dementia [16] (Figure 1).

Open in a separate window

Figure 1

A large majority of cases of dementia can be attributed to AD, cerebrovascular pathology, or a combination of the two. Although the relative importance and placement of different aspects of pathology along this continuum are debatable, it is clear that cases of both pure AD and vascular dementia are relatively rare, and that most cases of dementia will display elements of both.

The wide umbrella of definitions combined with the multiple dimensions of vascular injury leaves a large amount of ambiguity for what does and does not constitute VCID. For example, there is a controversy in the field over which types of vascular lesions contribute to cognitive impairment, including large cortical infarcts, lacunar infarcts, subcortical white matter disease, subcortical infarcts, or any combination of these [15]. This is further complicated by the presence of AD pathology, which is thought to lead to dementia more quickly in the presence of certain types of strokes [17]. In fact, it is very rare for an aged subject to not have any AD or cerebrovascular pathology. The two main pathological hallmarks of AD, amyloid plaques and neurofibrillary tangles, are present with overlapping cerebrovascular lesions in up to 50% of dementia cases [18]. However, the balance between the pathology of these diseases may be the determining factor for displaying clinical symptoms [19].

The inherent heterogeneity of VCID makes it difficult to develop representative models. VCID is not a complication of AD nor simply a form of stroke, but may encompass these etiologies as well as others. At our current level of knowledge, VCID is the best term we have to represent how vascular issues contribute to dementia. However, as the field expands, the term VCID may become too vague and could evolve into more specific definitions of particular disease states. The purpose of this review is to discuss the strengths and weaknesses of current models of VCID, including an overdue discussion on models of mixed dementias. While there is no current model that encompasses all aspects of VCID, there are ways to examine aspects of the disease separately, or in combination with different facets of neuropathology. These approaches encompass a range of strategies, from cell culture systems to a number of animal models with varying degrees of complexity.

The Neurovascular Unit and Cell Culture Models

Within the past few years, we have gained a larger understanding of the synergistic roles of the cell types encompassing the blood brain barrier (BBB) [20, 21]. This interaction, known as the neurovascular unit, provides an entirely different framework for examining how cerebrovascular disease contributes to cognitive impairment.

The neurovascular unit is composed of endothelial cells, myocytes, neurons and their processes, astrocytes, perivascular cells, and other supporting cells (microglia and oligodendroglia) [22, 23]. These cells work together to coordinate cerebral blood flow and exchange across the BBB. A functioning neurovascular unit is important for mediating blood flow in order to meet the metabolic demands of the brain [20]. Astrocytes, which line the outer walls of cerebral microvessels, are responsible for regulating blood flow to an area of high activity in the brain [24]. If there is insufficient blood flow to an area of metabolic demand, a cascade of rapid responses to the hypoxia stimulates angiogenesis, resulting in increased blood flow to the area of need [25]. Most models of the neurovascular unit use in vitro tissue culture with rodent cells to better understand all of the interacting components. However, there is a general lack of microvascular models using human cells.

Many in vitro BBB models rely on using endothelial cells, as they are the principal cellular component of the BBB. Primary endothelial cells isolated from rat, pig, or cow [26, 27], or human endothelial cell lines that are not of cerebral origin, such as human umbilical vein endothelial cells (HUVECs) are often used for BBB studies [28]. However, there is a large amount of heterogeneity within endothelial cells from different vascular origins which should be taken into consideration when using these cells as a model of the BBB [29]. To examine interacting cell types, endothelial cells are grown alongside astrocytes, pericytes, or a combination of the two using a co-culture system [30, 31]. Co-culture systems have high transendothelial electrical resistance and low permeability coefficients, indicating the presence of a tight barrier similar to the BBB [27, 32, 33]. However, these systems do not examine all aspects of the neurovascular unit simultaneously and are therefore better for understanding the role of a specific factor rather than the interacting cell types that work together to coordinate blood flow.

One of the newer BBB models uses synthetic microvessels for an in vitro model of the microvasculature. This model involves growing endothelial cells in collagen channels to form a microstructure. The endothelial cells form continuous junctions between cells and eventually form complex adherence junctions accompanied by slight re-structuring of channels [34]. Additionally, the collagen matrix can be remodeled to promote cell growth and angiogenesis. Vessel wall conditions and blood flow can be mimicked by seeding the endothelial cells in the matrix with pericytes and platelets [35]. Synthetic microvessels are currently limited to growth within a single plane, but three dimensional models using 3D printing of carbohydrate-glass lattices may be able to solve this problem. This 3D structure encourages endothelial growth in all directions and is a promising model of the microvasculature [36].

Information on the microvasculature has often lagged behind the wealth of information on large vessels despite the growing knowledge on its contribution to disease. In vitro models of the microvasculature are new and exciting tools to study its role in disease states such as VCID. Co-culture systems have taught us a lot about the BBB, from permeability studies of drugs [37, 38] to how Aβ crosses the BBB [39]. Little has been published thus far on the applications of synthetic microvessels, as it is a relatively new technique but the potential applications of the model are broad and range from better understanding the BBB, to having a more clear understanding the roles of individual cell types in the neurovascular unit, and eventually understanding the specific role of the microvasculature in certain diseases. However, these models are limited by the fact that they do not involve a physiological system to study the complex interactions of VCID. Therefore, while tissue culture models are useful tools, animal models are needed for studying the interacting players of VCID.

Animal Models of Altered Blood Flow

Cerebrovascular changes alter the macro and microvasculature, leading to both structural and functional brain damage. The development of new neuroimaging techniques has revolutionized our ability to examine these cerebrovascular changes [19]. One of the most important neuropathological markers of cognitive decline due to cerebrovascular dysfunction is cerebral infarcts [15]. There is a strong association between increased number of macroscopic infarcts and increased likelihood of dementia, but the relationship is not a simple one, and there is currently no defined volume or number necessary for a diagnosis of VCID [40]. One of the reasons for this is that infarct severity has varied effects on cognition, depending on the individual. Infarct location may determine the impact on dementia, with infarcts in regions such as the thalamus, angular gyrus, and basal ganglia more likely to lead to dementia [41, 42]. In other words, a single strategically placed infarct can be just as cognitively devastating as many smaller ones scattered throughout the brain.

Animal models have helped us understand how infarcts contribute to VCID. Chronic cerebral hypoperfusion (CCH) surgery is a good way to study infarcts in rodent models. CCH is one of the major causes of vascular - related dementia and is a result of various diseases, such as obstructive sleep apnea, congestive heart failure, and cardiac arrhythmias, that cause reduced blood flow to the brain [43, 44]. CCH typically develops as a result of vascular lesions caused by artery stenosis or occlusion, cerebral hemodynamic changes such as prolonged hypotension and reduced cardiac output, or by a change in blood viscosity, commonly associated with hyperlipidemia or elevated homocysteine levels [45]. Over time, these changes can decrease blood flow to the brain, causing increased neuroinflammation and oxidative stress, neuronal energy failure, and white matter lesions, all of which lead to cognitive impairment.

One of the more common CCH surgeries performed is the occlusion of the bilateral common carotid arteries (CCA). In rats, both the left and right CCAs are occluded, causing hippocampal and neuronal damage, striatal infarcts, white matter lesions, increased neuroinflammation, increased oxidative stress, and BBB disruption [46, 47]. Additionally, these rats perform poorly on several tests of cognition, suggesting cognitive impairment. However, this model is strictly feasible in animals with a complete circle of Willis (this excludes mice), which allows for continued, decreased blood flow to the brain via the basilar artery [48]. Additionally, there is a high amount of variability in the number of infarcts the rats develop and amount of hippocampal damage among animals from different vendors [47, 49]. Similar to the bilateral CCA occlusion model, the four vessel occlusion (4VO) model involves blockages of both vertebral arteries in addition to the CCAs. These animals have a low incidence of seizures and develop predictable ischemic neuronal damage [50]. However, many of the same drawbacks from the bilateral CCA model are present in the 4VO, with high variability among species and differences in CCA occlusion times reported in the literature [50, 51].

Bilateral CCA stenosis (BCAS) may be a more disease-relevant variant of the CCH models, as there is simply a reduction of blood flow rather than a total occlusion. BCAS is done by placing micro-coils consecutively around the CCAs, causing around an 80% decrease in cerebral blood flow [52]. BCAS works well in mice, causing a decrease in brain metabolism, increased neuroinflammation, and cognitive impairments such as decreased working and reference memory [53]. However, due to the small size of the coils, BCAS is a technically challenging procedure and very few labs have been successful in performing the surgery. Therefore, despite its efficacy as a rodent model to study CCH, there is currently little published literature on the technique.

Animal Models of Small Vessel Disease

Small vessel disease (SVD) causes nearly a fourth of all ischemic strokes and is a leading cause of vascular dementia. People with SVD often have cerebral amyloid angiopathy (CAA) and display deficits in information processing and motor function [54]. These cognitive impairments are often due to cerebral white matter lesions and subcortical lacunar infarcts [55]. Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is a form of SVD and the most common hereditary cause of vascular dementia [56]. CADASIL causes progressive white matter degeneration and ischemic strokes and can be exacerbated by vascular risk factors, such as high cholesterol, smoking, and hypertension [57]. Nearly all CADASIL cases are caused by mutations in Notch homolog 3 (NOTCH3). NOTCH3 is required for the maturation and function of small vessels and is primarily found in vascular smooth muscle cells [58]. The mutations in NOTCH3 cause accumulation of granular osmiophilic material (GOM) and a NOTCH3 ectodomain on vascular smooth muscle cell membranes. These vascular smooth muscle cells eventually die, causing enlarged perivascular spaces. This in turn causes stenosis of penetrating arteries, leading to strokes and white matter degeneration [59, 60].

We have learned a lot about the role of NOTCH3 in cerebrovascular disease from NOTCH3 knockout mice. These mice are viable and develop impaired cerebrovascular reactivity, reduced myogenic tone, and structural arterial defects [58, 61]. Interestingly, NOTCH3 knockout mice have more than two-fold larger infarcts when compared to controls after middle cerebral artery occlusion. Restoring NOTCH3 expression using a ROSA NOTCH3 mouse crossed to an appropriate Cre line (SM22-Cre) restored the stroke phenotype by reducing infarct volume [62]. However, NOTCH3 knockouts do not develop CADASIL pathology such as white matter degeneration and lacunar strokes [58]. Additionally, these mice do not develop GOM or NOTCH3 accumulation. Interestingly, a knock-in mouse model using the C455R mutation from a large Colombian CADASIL family causes a CADASIL phenotype with more severe stroke pathology than the NOTCH3 knockout mice [63]. This tells us that loss-of-function NOTCH3 mutations do not solely cause CADASIL, but may play a larger role in stroke pathology.

CADASIL transgenics, such as the R90C mouse, express a human NOTCH3 mutation which causes early CADASIL onset. These mice show age-associated vascular smooth muscle cell loss, as seen in humans, and accumulation of the NOTCH3 ectodomain occurs around 10 months [64]. Additionally, R90C mice develop diffuse white matter degeneration and subcortical infarcts in the basal ganglia and white matter [65]. However, these mice display vascular smooth muscle cell changes prior to any NOTCH3 accumulation, suggesting that NOTCH3 accumulation triggers but does not cause vascular dysfunction [64, 66]. Although these models are helpful in understanding how vascular dysfunction occurs in people with CADASIL, and may have some role in elucidating broader mechanisms involved in other SVDs, they are somewhat limited in scope. This is a common problem with mouse models based on rare familial mutations, a point which we will return to below.

Animal Models of CAA

Cerebral amyloid angiopathy is an important contributor to age-related cognitive decline. The main hallmark of CAA is the buildup of Aβ deposits in the penetrating arterioles and capillaries of the leptomeninges and cortex. APP gets cleaved into Aβ peptides of differing length, with senile plaques primarily composed of Aβ42 and cerebrovascular Aβ mainly consisting of Aβ40 [67, 68]. When neurons release Aβ, it is thought that Aβ42 sticks together and aggregates, while Aβ40 is flushed out of the brain via interstitial fluid drainage pathways [69]. While further discussion of Aβ is outside the scope of this review, there are many outstanding reviews of Aβ production and clearance [70–73]. Over time, Aβ intravessel accumulation can lead to necrosis, perivascular leakage of red blood cells, and eventually intracerebral hemorrhages and microbleeds [74–76]. Additionally, CAA contributes to cognitive decline and is the most common vascular pathology associated with AD, present in up to 90% of AD cases [77, 78]. CAA is most commonly seen as an underlying cause of intracerebral hemorrhages, but studies show that it also plays a major role in age-related cognitive decline, even when subsequent AD pathology is not present [79]. However, this mechanism is not well understood. CAA has been studied in several model systems over the years, and there are many excellent animal models of this disease pathology.

Canines provide a unique resource for studying aging and dementia. Dogs show age-associated cognitive decline with many similarities to humans. Canines accumulate Aβ in both plaques and the cerebral vasculature and develop neurodegeneration from oxidative stress, much like humans. Additionally, they are a good model for studying possible therapeutics for dementia, as they share similar pharmacokinetic and pharmacodynamic profiles with humans. One of the largest advantages to using a canine model is that, unlike most animal models, they often share a common environment and diet with humans [80]. In 1956, Anton von Braunmuhl first observed that canines develop CAA [81] and several studies have since confirmed this finding [82, 83]. Cognitive dysfunction and incidence of intracerebral hemorrhage correlates strongly with severity of CAA in both canines and humans [84, 85]. Furthermore, amyloid deposits in canines are primarily found in the intracellular spaces of the tunica media, similar to human CAA [86, 87]. Though canines are good models of CAA, there is considerable individual variability in the extent of pathology. Canines develop CAA by about the age of 13, but the severity of CAA varies largely, much like humans. Therefore, it is important to have large groups of subjects when using canine models [87, 88].

Cerebrovascular β-amyloidosis is also commonly found in non-human primates (NHPs), particularly in rhesus and squirrel monkeys. Rhesus monkeys commonly develop amyloid deposits in the neural parenchyma at around 25 years old [89] with some developing moderate CAA. This variability is similar to human CAA, though it is probably true that rhesus monkeys develop sporadic CAA more frequently than humans [87]. Squirrel monkeys, on the other hand, develop CAA by age 15. These monkeys more reliably develop CAA than rhesus monkeys, but unlike humans, the CAA is usually found in capillaries. NHPs are physiologically relevant models of human disease, as we are closely related and they mimic complex behaviors seen in humans. However, along with this close relation comes increased ethical consideration for the care and use of NHPs [90], requiring additional levels of scrutiny and justification for approval to ensure that their use is necessary, beneficial, and humane. Additionally, NHPs are costly to breed and house, particularly in aging studies, where animals require housing for nearly their entire lifetime [91].

There are several transgenic mouse lines that are valid models of CAA (for a comprehensive review, see [92]). Transgenic mice with artificial promoters to drive APP overexpression commonly show CAA pathology, with vascular Aβ deposition developing at different ages depending on the mutation. Transgenic mouse models of CAA have taught us a lot about the role of Aβ in the progression of CAA. For example, APPDutch mice, which bears an APP E693Q mutation causing CAA, strokes, and dementia, and APP23xAPPDutch mice, which have an APP KM670/6771NL mutation causing a 7-fold overexpression of mutant human APP, both have a high Aβ40/42 ratio and develop severe CAA, indicating that Aβ40 is the form most found deposited in the vasculature [93].

One of the most common CAA mouse models is the Tg-SwDI mouse, which has the APP KM670/671NL Swedish mutation, the APP E693Q Dutch mutation, and the APP D694N Iowa mutation, and develops extensive amyloid deposition in the cerebrovasculature. These mice start displaying CAA at around 6 months and this pathology increases with age, eventually causing oxidative stress, neuroinflammation, activated astrocytes and microglia, and impairments in learning and memory [94, 95]. However, the Tg-SwDI mice largely display pathology in the microvessels, which is rarely the case in humans with sporadic CAA.

Animal models of CAA are excellent models to study VCID. These studies give us a better understanding of how Aβ in the vasculature contributes to cognitive impairment and cerebrovascular disease. Further, the larger animal models, such as canines and NHPs, allow us to study the disease in mammals more closely related to humans than rodents. These animals not only share more complex physiological systems, but also have similar lifestyles to humans. This interaction sheds some light into how environmental factors contribute to CAA and VCID. However, these large animal models cannot easily undergo genetic modification, and require increased ethical and financial concerns.

Animal Models of Mixed Dementia: Interacting Disease States

Mixed dementia describes the comorbidity of two or more dementias, the most common being the overlap of AD and vascular dementia [96]. There are several risk factors that contribute to this mixed disease state, such as obesity, hypertension, and T2DM [15]. Over 40 million Americans aged 70 years or older have at least one of these risk factors, yet we know relatively little about how these factors contribute to cognitive decline [12].

Cognitive impairment strongly correlates with obesity and T2DM in both rodents and humans. This risk is exacerbated with the presence of AD, forming a unique type of dementia with vascular pathology, small strokes and AD related neuropathology. Interestingly, people with this disease state often have lower plaque and tangle counts. It is thought that the presence of vascular pathology in these cases (mainly subcortical and/or lacunar infarcts) lowers the threshold of AD pathology required for development of dementia [17, 97]. The presence of diabetes, therefore, does not change the amount of AD pathology, but rather increases cerebrovascular pathology leading to dementia [98, 99].

One of the main models for studying these interacting disease states is through treatment with streptozotocin (STZ), a pancreatic islet toxin. STZ damages pancreatic β cells, causing hypoinsulinemia and hyperglycemia [100]. However, STZ is mostly used as a model for type 1 diabetes and does not address the issue of obesity [101]. Transgenic mice are a common tool for studying diabetes, but are limited in scope. When ob/ob mice (which are leptin deficient) are crossed with APP23 mice (which overexpress APP KM670/6771NL under a Thy1 promoter), the mice show early cognitive deficits (2–3 months) independent of amyloid pathology [102]. While the oldest animals (12 months) did not show any plaque pathology, a small number (n=3) showed significant levels of Aβ in the blood vessels. It is important to note, however, that in a separate study in CRND8 mice (which contain both the APP double Swedish mutation and the Indiana mutation), short-term leptin administration caused a reduction in Aβ deposition and improvements in cognitive function and it is unclear how to reconcile these results with the ob/ob cross study [103]. The db/AD mouse, a cross between the obese and diabetic db/db mouse and the APP Swedish x PSEN1 L1660 knock-in mouse model of AD, is one such model of a mixed dementia state [104]. These mice are diabetic, develop amyloid deposits with increasing age, have ischemic strokes and increased neuroinflammation, and display profound cognitive impairments at a much younger age (12 months) than the APP Swedish x PSEN1 L1660 knock-in mice alone. However, these mice show no signs of CAA or hypertension, which is unlikely in a human with mixed dementia, although this may also suggest that CAA and hypertension are not necessary for strokes to occur in an aging brain with AD pathology.

Neuroinflammation is thought to contribute largely to AD progression and cognitive decline. There is an established link between activated microglia and AD [105]. This is further complicated by the presence of proinflammatory cytokines, which are known to contribute to neuronal loss [106]. Increased inflammation is thought to accelerate cognitive decline and is often used as a hallmark of neurodegeneration. A/T transgenic mice, a cross between an APP overproducing mouse (APP Swedish, Indiana) and the constitutively-active TGF-β1 mouse (TGF mice, line T64), is a mouse line that combines AD and cerebrovascular pathology. These mice have increased cerebral and cerebrovascular Aβ deposition, reduced neurovascular and neurometabolic coupling, astrocyte activation, and display cognitive impairment by decreased water maze performance [107]. However, these mice show delays in cognitive decline compared to the APP overexpressing mice alone, indicating that TGF-β may play some sort of neuroprotective role. Additionally, these mice develop cerebrovascular pathology that is unique to the increased activity of TGF-β and the mechanism behind this is not fully understood [108].

It is now widely accepted that there is a link between high fat diets and cognitive decline in the elderly population. The Rotterdam study, a population-based cohort recruited to study diseases in the elderly, showed a strong link between dementia with a vascular component and total and saturated fat levels, also confirmed in rodent models [109]. Mice fed high fat diets have expected metabolic issues in addition to high oxidative stress, impaired cognition, increased inflammation, and decreased BDNF levels [110, 111]. Rats fed diets high in saturated fats and sugar showed cognitive deficits accompanied by increased BBB permeability [112].

Studies show changes in the cerebrovasculature of animals fed a high fat diet [113]. However, there is a discrepancy in the field for the percentage of lard used in a high fat diet. The typical western diet consists of 40% lard, but studies have shown that cerebrovascular changes only occur when a 60% lard diet is used [110, 111]. Additionally, these models are independent of amyloid pathology and only account for a specific lifestyle risk of dementia. While there is a large amount of literature on obesity and diabetes in the context of high fat diets, there is little available on the effect of high fat diets on brain aging [114]. The literature contains conflicting reports on whether high fat diets actually promote or accelerate brain aging and there are currently no comprehensive studies on what metabolic parameters promote brain aging.

Homocysteine (Hcy) is a methionine-derived amino acid that is linked with cardiovascular disease. Methionine synthase maintains normal Hcy levels and uses vitamin B12 and folate as cofactors to remethylate Hcy back to methionine [115]. Elevated levels of Hcy, known as hyperhomocysteinemia, are strongly associated with cardiovascular and various neurologic diseases [116]. Studies suggest that these elevated levels are toxic to endothelial cells and cause other disruptions, such as platelet adhesion, suppression of heparin sulfate expression, and several others [117]. Dietary intake of methionine, folate, and vitamin B12 determine levels of Hcy, so hyperhomocysteinemia is modifiable by diet [118]. Deficiencies in folate and vitamin B12 are known to be a cause of stroke and data shows that dietary folate fortification reduced levels of stroke in the United States and Canada [119].

Mice that are put on a hyperhomocysteinemic diet (folate and B12 deficient with excess methionine) show cognitive decline, high microhemorrhage counts, increased neuroinflammation, and elevated matrix metalloproteinase levels, indicative of BBB breakdown [120]. However, B-vitamin deficiencies can cause cognitive impairment, so it is unclear whether the B-vitamin deficiency or the hyperhomocysteinemia itself causes the cognitive decline shown in these animals [121]. Additionally, although hyperhomocysteinemia is a known risk factor of stroke [122] and correlates strongly with AD [123, 124], this is an independent risk factor for disease and by no means represents the majority of VCID cases. Further, hyperhomocysteinemia is toxic to neurons [125, 126], which may argue that rodents put on the diet show cognitive decline from toxicity effects and not from VCID.

Although risk of developing both hypertension and dementia increases with age, hypertension is a major risk factor for dementia independent of age. The Honolulu Asia Aging Study examined 3703 men starting midlife and followed up with them for the next 26 years [127]. This study showed a strong correlation between middle aged men with untreated hypertension and both AD and vascular dementia. Several other longitudinal studies show correlations between high blood pressure and dementia [128–130]. It is thought that chronic high blood pressure causes vessel wall thickening and reduction in microvessel diameter [131]. Additionally, plaques in the larger cerebral arteries can rupture, causing complete blockage of arteries and infarcts in the surrounding tissue [132].

The most popular model for studying hypertension is the stroke prone spontaneously hypertensive rat (SHRSP). These rats are normal at birth and develop high blood pressure as they age. This eventually leads to ischemic lesions in the cortex and basal ganglia [133]. Additionally, these hypertensive rats perform poorly on learning and memory tests and worsen post-stroke. Vessel occlusion surgery in SHRSP has been shown to cause an even more exaggerated vascular phenotype, with white matter lesions, hardening of vessel walls, BBB breakdown, and increased neuroinflammation [134, 135] It is very important to carefully observe SHRSP, as they often develop paralysis due to the ischemic strokes which is easily misinterpreted as muscle weakness or cognitive decline [136]. Overall, these rats are important for studying the influence of hypertension on vascular pathology, a known risk factor of VCID in humans.

Conclusions

As people in the developing world are living longer, our aging population is increasing. Given that cognitive impairment is a common condition in the elderly, the incident rates of dementia will increase drastically within the next 50 years. Understanding of the common causes of dementia, such as AD and VCID, has come a long way in the last 10 years. However, there is still a great deal that we do not know about different types of cognitive impairments. This review has focused on VCID and the current models that we have for understanding this heterogeneous disease state.

Currently, there are no definitive guidelines for diagnosing VCID. While there are several recommendations for physicians, there is a general lack of consistency in stroke counts and type, location of vascular injury, along with several other thresholds to determine if VCID is present. The molecular and cellular basis for how lifestyle factors influence vascular injury, particularly in white matter, remains unknown. Understanding how risk factors influence disease would be helpful for developing potential therapeutics to treat different aspects of VCID. We have a general understanding of the roles that hypoperfusion, the neurovascular unit, and inflammation play in cerebrovascular injury in animal models. Yet this understanding has not yet led to any viable therapeutic targets. As we develop better models of VCID, we will have a more complete understanding of the disease state and the best way to treat it.

While we have several useful animal models to model certain aspects of VCID, none of them are able to fully model VCID, which encompasses several spectrums of pathological markers. This gap in the literature stalls the development of therapeutics and hinders our understanding of VCID. However, the broad definition of VCID will likely mean that there will never be an all-encompassing model of the disease state. Current and future VCID models will likely tackle different aspects independently, resulting in slow pathways to VCID treatments. With better understanding of the factors involved in VCID, perhaps we can create more specific definitions and models for each individual disease state. This would create an opportunity for the development of therapeutics and treatments. In the current absence of therapies, the best option is to recommend reducing lifestyle risk factors for VCID and maintaining general vascular health.

Footnotes

References