“The main message of this paper is the mixed pathology as we call it—microvascular disease and Alzheimer’s—is associated with more brain damage, more white matter damage and more inflammation,” says Dr. Zsolt Bagi, vascular biologist in the Department of Physiology at the Medical College of Georgia at Augusta University.
Theirs and other recent findings suggest that some people with Alzheimer’s who have brain changes widely associated with the condition, like amyloid plaques, may not develop dementia without this underlying vascular dysfunction, the researchers write in the journal GeroScience.
“We are proposing that if you prevent development of the microvascular component, you may at least add several years of more normal functioning to individuals with Alzheimer’s,” Bagi says.
He and by Dr. Stephen Back, pediatric neurologist, Clyde and Elda Munson Professor of Pediatric Research and an expert in white matter injury and repair in the developing and adult brain at Oregon Health & Science University, are co-corresponding authors of the new study.
The good news is that vascular disease is potentially modifiable, Bagi says, by reducing major contributors like hypertension, obesity, diabetes and inactivity.
The scientists looked at the brains of 28 individuals who participated in the Adult Changes in Thought Study, or ACT, a joint initiative of Kaiser Permanente Washington Research Institute and the University of Washington, whose scientists also were collaborators on the new study.
ACT is a longitudinal study of the cognitive health of community volunteers from the Seattle, Washington area with the goal of finding ways to delay or prevent memory decline. Participants age 65 or older with no cognitive problems upon enrollment are followed until their death, and about 25% agree to autopsy and making genomic DNA from their blood and/or brain tissue available to scientists.
The individuals that served as controls for the study had no indication of Alzheimer’s or vascular disease in their brain. Other groups had Alzheimer’s without vascular disease, vascular disease without indicators of Alzheimer’s or both Alzheimer’s and vascular disease.
Their focus in the studies was the white matter, which accounts for about 50% of the brain mass, enables different regions of the brain to communicate and is packed with long arms called axons that connect neurons to each other and to other cells across the body like muscle cells; and, the microscopic arterioles that directly feed white matter with blood, oxygen and nutrients.
They wanted to test their theory that when these hair-thin arterioles had difficulty dilating and so supporting this part of the brain, it resulted in changes to the white matter that were evident on sophisticated MRIs, especially when microvascular problems coexisted with the more classic brain changes of Alzheimer’s.
They found that the arterioles of those who had been diagnosed with Alzheimer’s and dysfunction of these tiny arteries did have an impaired ability to dilate in response to the powerful blood vessel dilator bradykinin, compared to those without obvious microvascular dysfunction. Problems with dilation were associated with white matter injury and changes to the white matter structure that were visible on MRI.
Expression of the precursor for the also-powerful blood vessel dilator nitric oxide also was reduced in these individuals with both conditions while the expression of superoxide generating NOX1, which damages blood vessels, was increased.
Arteriole dysfunction also was associated with more white matter injury based on what was visible on those sophisticated MRI scans and the increased number of brain cells, called astrocytes, which support neurons.
The investigators had previously reported an increase in these astrocytes in brains with the microvascular changes. This time they saw that when Alzheimer’s and the microvascular changes were both present, the astrocytes became more reactive, inflammatory and damaging.
Colleagues at Oregon Health & Science University, led by Back, looked at the same brain tissue with a sophisticated MRI technique called diffusion tensor imaging, that uses water diffusion between cells to look at the microstructure of white matter and its connectivity.
They could not visualize individual arterioles because they are too small – about 30 microns or .0011811 inches – to see without a microscope. But they could see the white matter damage that resulted from arteriole disease, and again found the correlation between the vascular impairment and tissue damage that Bagi described from directly visualizing the tissue. This type of blood vessel disease was present in 50% of the brains they studied, and other autopsy studies have indicated a similarly high rate.
In those with less indicators of brain changes, they found the arterioles were better able to dilate, that area of the brain had better connectivity and less damage apparent on the postmortem MRI.
Impaired ability of these small vessels in the white matter to dilate is known to be associated with white matter injury, like that visible on the specialized MRI scans. And there is evidence in both laboratory studies and humans that this vascular dysfunction does not just worsen but plays a role in the development of cognitive decline and dementia in people with Alzheimer’s, the investigators write.
In fact, the vascular dysfunction may be present before the damage to the brain tissue and cognitive dysfunction is apparent. In research animals bred to develop Alzheimer’s, for example, there is evidence of problems with the microvasculature in areas of the brain associated with Alzheimer’s, like the hippocampus, a center of learning and memory, at a very young age.
The new work confirms the growing concept that small blood vessel disease may help predict the severity of dementia and/or use of DTI MRI may help identify those patients with early enough disease that strategies to reduce or slow small blood vessel disease may help delay or reduce their cognitive loss. The technique might also help assess the potential benefit of intervention.
“These individuals might especially benefit if they would exercise, control blood sugar level and control their blood pressure,” Bagi says.
Some patients with Alzheimer’s disease are known to have white matter hyperintensities on MRI scan, basically damaged areas that show up particularly bright on the scan and are associated with problems like dementia. A significant proportion of individuals with Alzheimer’s also have conditions like high lipid levels in their blood and hypertension that are known to impair blood vessel function, including the smallest vasculature, Bagi notes.
Small blood vessel disease in the brain also is common in aging and may indicate an increased risk of problems like stroke or dementia. Sophisticated brain scans also often indicate microinfarcts, essentially microscopic strokes, which also tend to increase with age and are associated with memory impairment.
Age and a family history are major risk factor for Alzheimer’s and there are two categories of genes associated with an increased risk, including risk genes, like APOE-e4, the first gene identified and the one that has the strongest impact on risk, according to the Alzheimer’s Organization.
Then there are those genes that can directly cause Alzheimer’s, called deterministic genes, which impact production or processing of beta-amyloid, the main component of the plaque associated with Alzheimer’s, but even having these rare genes are not a guarantee of disease.
“You have some genetic predisposition but people realize that not everybody develops memory decline or cognitive deficits unless something else is coming in,” Bagi says. He notes that they have not yet analyzed the genes for this study.
Next steps include studying the associations they found in more human brains and more studies to better understand exactly how the small blood vessel disease happens, which could point toward new targets to intervene.
Alzheimer’s disease (AD) is a chronic neurodegenerative disorder and the predominant form of dementia [1,2]. Dementia was estimated to affect approximately 50 million people worldwide as of 2018 and this figure is expected to triple by 2050, the majority of cases being of the Alzheimer’s type .
AD presents with an insidious onset, with progression of symptoms over years to decades . These may include the loss of memory, cognitive decline, emotional and behavioural changes, psychological impairment and loss of motor coordination . From a neuropathological perspective, AD is associated with several characteristic features, the most important being progressive and extensive atrophy of the cortex and hippocampus, the deposition of insoluble β-amyloid (Aβ) within extracellular neuritic plaques and the appearance of intracellular neurofibrillary tangles (NFTs), composed of hyperphosphorylated tau protein [1,2].
AD is clinically differentiated from several other forms of dementia and graded by the appearance of the latter two pathological features. Due to an incomplete understanding of the factors underlying AD pathogenesis, a cure for the condition has been elusive. To date, only four drugs have been approved by the US Food and Drug Administration (FDA) for clinical use, but these are purely symptomatic rather than disease-modifying therapies [5,6]. Given the diversity of changes within the AD brain, it is clear that alternative mechanisms of neurological dysfunction must be considered in the design of combinatorial therapies to better address the complexity of AD pathogenesis.
In attempting to ascertain the underlying etiology of AD, a commonly overlooked pathological aspect of the disease is the occurrence of extensive vascular dysfunction—the most apparent anatomical signs being the appearance of cerebral amyloid angiopathy (CAA) and vascular morphological and degenerative changes in affected parts of the brain . Indeed, neurovascular dysfunction is ubiquitous within the AD brain and a “vascular hypothesis” of AD was suggested a quarter of a century ago based on observations of cerebral perfusion and metabolic deficits in AD patients [8,9].
In addition to these gross anatomical and physiological changes, numerous studies have reported diverse correlates of vascular cell dysfunction, including Aβ-mediated cytotoxicity, deficits in Aβ clearance, the weakening of the blood-brain barrier (BBB), aberrant immune cell recruitment and a direct vascular contribution to the pro-inflammatory state in vulnerable brain regions  (Figure 1).
Vascular changes are an early preclinical feature of AD pathology, with changes in cortical blood flow beginning years to decades prior to the onset of clinical symptoms [11,12]. Focal decreases in blood flow in turn have an impact on amyloid clearance and neuronal metabolism [13,14]. It has been suggested in recent years that AD vascular biomarkers be incorporated into current research frameworks for the improvement of clinical AD diagnosis .
The symptomatic overlap between AD and vascular dementia (VD) has also been noted and both conditions respond similarly to some pharmacotherapeutic strategies [16,17]. Indeed, it has been suggested that the majority of older patients present with a mixed-dementia with characteristics of both conditions .
In the seminal Nun study published by Snowdon et al. in 1997, it was demonstrated that the presence of lacunar infarcts in the basal ganglia, thalamus or deep white matter causes a reduction in the neuropathological threshold (i.e., senile plaque and NFT load) required for any given grading of AD dementia . These results have since been confirmed, with the additional finding that pathological comorbidity with cerebrovascular disease determines clinical presentation. It appears that senile plaque load is a predictor of cognitive deficit in combination with cerebrovascular disease, but NFT load does not appear to be a predictor of dementia severity when combined with cerebrovascular disease; NFT load is a known indicator of cognitive decline in cases where cerebrovascular disease is absent . The clinical expression of AD thus appears to be significantly modified by the presence of cerebrovascular abnormalities. Given the strong association between amyloid angiopathy and atherosclerotic changes in the AD brain [21,22], it is likely that the impact of arterial amyloid deposition on vascular hemodynamics and vessel rigidity plays a significant role in arterial narrowing, increased blood pressure and the weakening of the arterial wall . There is thus a growing appreciation for the link between AD risk and pre-existing cardiovascular conditions.
There is also a significant interaction between AD pathological development, systemic vascular risk factors and genetic risk factors—the most predictive being apolipoprotein E (APOE) genotype. APOE is a protein which is important in both neural and vascular health. It is a multi-functional protein, which is involved in cholesterol transport, lipid metabolism and clearance of Aβ, among other functions [23,24]. There are 3 isoforms, APOE epsilon2 (APOE2), APOE epsilon3 (APOE3) and APOE epsilon4 (APOE4). APOE3 is the most common isoform . APOE2 has been shown to be protective against AD, while APOE4 significantly increases AD risk [25,26,27] without altering the rate of cognitive decline after onset . This association is even more pronounced in the presence of pre-existing chronic risk factors like diabetes [29,30,31].
Taken together, it is clear that cerebrovascular dysfunction is an area deserving of further consideration in the formulation of theories of AD pathogenesis. In this review, we present a summary of the current state of knowledge in this area and critically discuss implications for primary AD research and the search for new clinical therapies.
.Neurovascular Coupling Deficits and Metabolic Dysfunction
A link between impaired neurovascular coupling and AD pathogenesis was first formally proposed by de la Torre and Mussivand in 1993 . Although global changes in cerebral blood flow (CBF) and glucose metabolism have been reported in the AD brain , several single photon emission computed tomography (SPECT) studies and arterial spin-labeling (ASL) magnetic resonance imaging (MRI) studies have demonstrated that reductions in CBF in AD are most pronounced in the temporal and parietal cortices [32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47].
These findings complement [18F]fluorodeoxyglucose positron emission tomography (FDG-PET) studies that reveal AD-associated reductions in glucose metabolism to be most severe in the temporoparietal cortex [48,49,50,51,52,53,54,55]. Indeed, temporoparietal CBF reduction is often considered to be a defining pathological features of AD and the utility of FDG-PET [56,57,58], SPECT [45,59,60], ASL MRI [47,61,62] and functional and perfusion MRI [63,64] in the diagnosis of AD and its differentiation from other dementias has been established. Within the parietal cortex, hypoperfusion and hypometabolism have been reported to be especially pronounced in the angular gyri [47,61] and posterior precuneus [12,47].
Reduced CBF has been noted in the frontal cortex in some studies [32,38,40,46] and hypometabolism has also been observed in this region [52,53,54], although other studies appear to dispute the occurrence of frontal perfusion deficits . Interestingly, marked temporoparietal metabolic dysfunction is observed in both mild and severe AD patients, indicating that these changes may precede clinical symptoms .
In addition, it has been demonstrated that disease severity is associated with worsening deficits in CBF in the frontal and parietal cortices but not in the temporal cortex, suggesting a progression of blood flow abnormalities from the temporal to the frontoparietal cortices with disease progression . In more established or severe AD, there is additionally greater occipital involvement [12,40,54]. Recently, 4D flow MRI studies have revealed a general reduction in total mean arterial blood flow in several major cerebral arteries as well as an increase in arterial pulsatility in AD patients [65,66].
Specific regional perfusion deficits are associated with different clinical and functional outcomes in AD patients. Cognitive performance has been correlated with CBF reductions across multiple cortical regions . Additionally, asymmetric alterations in CBF have been demonstrated in numerous studies, with relevance to specific functional deficits. For instance, reduced posterior parietal CBF is correlated with the clinical presentation of apraxia in a subset of AD patients, while reduced temporal lobe CBF is correlated with memory deficits .
Left temporal lobe CBF appears to be more correlated with memory deficits than right temporal CBF and reductions in the left frontal, left lateral temporal and left posterior parietal regions are correlated with aphasia . Results such as these underscore the importance of early vascular changes in the cortex in determining the presentation and course of AD and the contribution of asymmetric deficits to functional outcomes.
While cortical patterns of CBF dysfunction are well-attested, there is less consensus regarding subcortical structures. Some studies reported the preservation of hippocampal and subcortical blood flow and metabolism in prodromal and established AD [40,61], interesting given the centrality of hippocampal cell loss to AD pathogenesis. Posterior cingulate cortex hypoperfusion [12,61,67,68] and right anterior cingulate hyperperfusion  have been demonstrated in AD patients using ASL imaging and H215O PET, but ASL and SPECT studies have inconsistently demonstrated either hypoperfusion or hyperperfusion in the hippocampus and amygdala [38,42,67,69].
Posterior cingulate hypoperfusion, in particular, is well-attested and metabolic deficiencies in this region are a feature of very early AD . Hyperperfusion has been reported in the left thalamus and in limbic structures such as the right striatum, hippocampus and right amygdala in mild AD patients, potentially suggestive of compensatory effects in early AD or perhaps inflammation or some other vasodilatory stimulus [38,42]. One ASL study reported significant hippocampal hyperperfusion in early AD after accounting for grey matter loss . An earlier FDG-PET study, however, reported modest decreases in caudate and thalamus glucose metabolism . Cerebellar CBF is reportedly unaffected, even in later AD stages .
Differences in sub-regional perfusion deficits within these structures and hemispheric asymmetries in pathology likely contribute to some of the inconsistencies in the literature. A general weakness of such studies is in the range of disease stages represented by participants, even amongst early AD patients—some patients with preclinical signs of AD may have more severe pathology than others prior to formal diagnosis or may develop other forms of dementia.
The well-established changes in temporoparietal and posterior cingulate perfusion and metabolism are likely the most reliable vascular biomarkers of risk and progression of preclinical/early AD. Another factor of concern is the lack of appropriate atrophy correction in many imaging studies, which can have significant bearing on the results of these studies [42,44]. Indeed, asymmetric perfusion deficits in the AD brain appear to closely mirror asymmetric atrophy . Partial-volume corrected CBF values from many cortical regions, as measured by ASL, are much higher than uncorrected CBF values from the same brain regions , and appropriate volume correction may thus reduce or eliminate the statistical significance of reported changes in some sub-regions . This indicates the possibility of bias in many older studies.
The elucidation of AD risk based on preclinical changes in the brain is a matter of growing interest, but relatively little is known about the earliest phases of AD pathogenesis. Given the consistency of cortical patterns of CBF and metabolic disruption in the AD brain, particularly in mild AD, and the correlation with disease severity, measures of CBF and metabolism could potentially help predict predisposition to AD prior to diagnosis. With this in mind, Okokwo et al. conducted a comprehensive study into such perfusion changes in patients with AD, as well as asymptomatic individuals with a family history of AD and those with mild cognitive impairment (MCI) with amnesia. Intriguingly, they showed that cognitively normal patients with a history of AD in the maternal line showed greater parietofrontal and hippocampal perfusion deficits than cognitively normal patients with AD in the paternal line or without a family history of AD, suggesting that some maternally-inherited factor may confer AD risk .
Another study utilized ASL and blood-oxygen-level-dependent functional MRI (fMRI) to show that resting CBF is elevated in the middle temporal cortex in cognitively normal APOE4-positive patients with a family history of AD and that CBF responses in these patients are diminished during functional encoding tasks compared with low-risk individuals . Thus, regional CBF and glucose metabolism measured by standard medical imaging techniques could be useful as biomarkers of AD risk or early pathological changes in the disease prior to the onset of clinical symptoms, and this could perhaps aid in the assessment of potential preclinical therapies. Indeed, at-risk individuals may show metabolic deficits in temporoparietal cortical regions .
FDG-PET has been used successfully to track regional declines in metabolic rate over time, mirroring cognitive decline, and changing glucose metabolism in AD-affected brain regions may thus be a good biomarker of AD progression over time . With regards to CBF changes, one SPECT study was able to differentiate between healthy controls, patients who presented with signs of impending AD, patients who had just been diagnosed with AD and patients with established AD . It was shown that prior to clinical AD onset, hypoperfusion in the hippocampus, amygdala, posterior cingulate gyrus and left anterior thalamus precedes deficits in the temporoparietal cortex .
SPECT has also been used to correlate the Braak stage progression of AD with perfusion changes in various brain regions, revealing the appearance of deficits in anterior middle temporal, subcallosal, posterior cingulate and cerebellar perfusion between the entorhinal and limbic stages, and then large deficits in posterior temporoparietal perfusion between the limbic and neocortical stages of the disease . The progression of deficits has also been correlated with decreases in cerebrospinal fluid (CSF) Aβ and increases in CSF tau prior to the onset of AD. Abnormal Aβ downregulation in the pre-AD CSF appears to be associated with the onset of temporoparietal CBF reduction and this hypoperfusion then worsens with the onset of abnormal CSF tau upregulation .
High CSF phospho-tau (p-tau) and total tau in the healthy brain have been correlated with reduced CBF in frontotemporal regions . The elucidation of such stereotyped and progressive changes in pre-AD patients and through the course of the disease could aid in differential diagnosis at an early stage of disease progression and aid in preclinical prediction of disease risk, potentially allowing for early intervention. It has been shown that brain Aβ load in early AD is associated with decreased blood flow in various brain regions . Aβ1-40 application reduces resting CBF in APP transgenic mice  and vascular Aβ deposition in APP/PS1 mice brings about regional CBF reductions . Inter-individual variations in Aβ load may have a significant effect on patterns of perfusion changes in AD, with high Aβ load being associated with longitudinal increases in perfusion in some brain areas and decreases in others . Regional CBF even in the healthy brain may show reductions in the presence of Aβ deposition  and metabolic disturbances may be observed in these brain regions [82,83,84]. Thus, correlations between CBF reduction and other molecular biomarkers of AD are well attested.
Although APOE genotype is considered to be the most significant genetic risk factor for AD, the link between APOE genotype and CBF dysfunction is controversial. Many studies have failed to show a correlation between this factor and the severity or distribution of perfusion or metabolic deficits in AD [55,67,75,85,86], with one study even demonstrating dose-dependent increases in frontotemporal metabolism . This may be consistent with the finding in previous studies that APOE status, while modifying AD risk, does not affect the rate of disease progression following AD diagnosis .
However, many others have demonstrated reduced CBF and metabolism across multiple cortical regions in APOE4-positive AD patients compared with non-carriers [88,89,90,91,92,93]. Once again, it is important to consider the impact of AD stage and the evolving nature of these deficits, the importance of volumetric correction and appropriate analysis in imaging studies , the use of different imaging techniques, the effect of APOE4 gene dose, correction for confounding factors like age and sex, and the impact of patient sample sizes. APOE4 homozygous patients in some studies reportedly present with greater perfusion and metabolic dysfunction than heterozygotes [87,93] but also with greater volume loss in the temporal cortex .
Given the inconsistencies between studies examining patients with established AD, it is perhaps more instructive to look at the contribution of APOE4 to pre-clinical risk of AD. Individuals with a family history of AD and positive for APOE4 display pronounced metabolic deficits in several cortical regions, in a pattern similar to that observed in AD patients [67,95,96]. Cognitively normal APOE4 carriers display more significant perfusion deficits with age across multiple sub-regions of the frontal, temporal, parietal and cingulate cortices compared with non-carriers, which could contribute to increased AD risk with age [71,84,97].
Cognitively normal APOE4 carriers also appear to display increased CBF in the left lingual gyrus and the cuneate nucleus, particularly in older carriers . Interestingly, hyperperfusion is observed in the left anterior cingulate cortex in younger APOE4 carriers, while hypoperfusion occurs in older carriers , potentially indicating very early compensatory changes followed by the development of more pronounced deficits and CBF reduction. Indeed, younger APOE4 carriers appear to display greater activation in hippocampal and default network circuits during memory encoding, reflected in BOLD fMRI measurements from these regions, and this over-activation gives way to significantly reduced activation with age [98,99]. Thus, the very early contribution of APOE4 to neuropathological processes may be recapitulated by CBF and cerebral metabolic changes through the lives of carriers prior to the clinical appearance of AD. It is also likely that APOE4 status contributes to the heterogeneity in imaging-based measures of early CBF changes in AD.
The causes of perfusion deficits in AD are likely numerous. However, changes in blood flow are believed to be due in large part to changing patterns of vascular innervation with neuronal loss . Cognitive reserve may result in the preservation of cognitive function until a critical mass of neuronal loss has been achieved . Deficits in neurovascular coupling may thus become pronounced prior to the onset of clinical symptoms. The cholinergic hypothesis is based on the observation that post-mortem AD brains exhibit severely decreased cholinergic innervation in several brain regions, in particular the temporal cortex and hippocampus .
This is believed to play a role in the cognitive and behavioral deficits characterizing the disease, alongside several other pathological contributors. The cholinergic-vascular hypothesis posits that the significant cholinergic neuron loss observed in AD results in decreased vasodilatory tone in innervated vessels, due to the fact that cholinergic neurons play a key role in the maintenance and control of vascular tone in affected brain regions [103,104].
Cholinergic nerve terminals projecting from the basal forebrain are closely associated with arterioles in the frontal, parietal and temporal cortices and cholinergic innervation to cortical vessels is greatly reduced in AD . Acetylcholine (ACh) functions as a vasodilator through muscarinic receptors on vascular smooth muscle cells (vSMCs) and decreased ACh tone likely contributes to greater baseline vasoconstriction in affected areas . It is important to note that other neurotransmitters like glutamate and noradrenaline have also been demonstrated to act as vasoactive agents, either indirectly through intermediate vasoactive factors or directly by acting on contractile vascular cells like pericytes and vSMCs [106,107,108].
Thus, the loss of glutamatergic and noradrenergic innervation in AD [109,110] could similarly disrupt neurovascular coupling mechanisms. Many excitatory neurotransmitter systems, including the cholinergic system, may be coupled to vascular changes either directly or indirectly through GABAergic interneurons or glial cells [105,111,112,113]. Glial cell dysregulation is a key pathological feature of AD  and this likely also plays a role in neurovascular coupling deficits. Thus, preclinical perfusion deficits are likely the result of early loss or remodeling of innervation in the presenile brain, resulting in metabolic dysfunction and potentially contributing to a feedback loop involving neuronal death in affected brain areas. Reduced blood flow in vulnerable areas is associated with reduced clearance of Aβ, potentially resulting in enhanced neurotoxicity and deposition in these regions .
Numerous other vasoactive signaling mechanisms are believed to become dysfunctional in AD but these will not be reviewed exhaustively. The vasoactive effects of Aβ and the effects of neuroinflammatory mediators on vasoactive changes will be covered in brief in later sections. We will also review the possible mechanisms of vascular amyloid deposition and touch on how this may contribute to impaired vascular contractility and reductions in vSMC and pericyte number, which may have important implications for perfusion in the AD brain.
Vascular Morphology and Angiogenesis
It is well established that vascular abnormalities are common in the AD brain at the macrostructural level. Several early qualitative studies reported distortions in small arterioles and capillaries in the AD brain, particularly in the hippocampal region and temporoparietal cortex. These vessels were often described as being tortuous, kinking, looping, twisting, spiraling or forming bundles and knob-like structures [115,116,117,118,119]. Such dementia-related changes in arteriole structure may be accompanied by the thinning of vessel walls and the loss of smooth muscle and elastic tissue . While changes in brain volume and parenchymal tissue loss may influence vascular architecture, features like vessel lengthening and the formation of “wicker-like” networks point to the active influence of other factors on angiogenesis .
Such structural changes may exert a significant impact on local blood flow, with looped vessels and abnormal structures contributing to increased vascular resistance and disturbing the overall hemodynamic state of the local vascular network [115,120]. It has also been found that the density of “string vessels”—non-functional capillary remnants mostly composed of connective tissue and lacking in endothelial cells—is significantly increased in the gray matter of the AD brain [121,122,123] and also increased, albeit to a lesser extent, in the gray matter of brains from patients without dementia but exhibiting amyloid pathology .
String vessel density appears to be greatest in brain regions with high Aβ load . A similar increase in string vessel density, greater than two-fold, also occurs in the white matter, along with an increase in string vessel length and string length as a proportion of total vessel length . The exact reason for increased string vessel formation in the AD brain remains unknown. Increases in string vessel density as a result of endothelial degeneration are likely to be closely linked to decreased perfusion and metabolic dysfunction in the AD brain. However, string vessels may also be the result of aberrant angiogenesis and changing patterns of vessel coverage. It has been shown in the healthy brain that transient endothelial tube sprouting and retraction can leave behind these structures and that pathologically affected tissues display more of these outgrowths .
Angiogenic processes may be disturbed in AD, possibly contributing to the process of string vessel formation. Despite reports of sometimes marked reductions in vascular density in the AD cortex, hippocampus and basal forebrain [115,118], there is evidence also of upregulated angiogenesis. It has been reported that vascular sprouts composed of endothelial processes are readily discernible in tissue from advanced AD cases . One study reported increased vascular density within the AD hippocampus, with an increase in angiogenic vessels positive for integrin αvβ3  and another demonstrated a significant increase in vessel density in the AD cortex .
Rather than being contradictory, this may be indicative of the remodelling of surviving vascular networks in the AD brain. Supporting this idea, young APP23 mice display denser vascular networks around amyloid plaques that are associated with truncated blood vessels . Another possibility is that angiogenesis in the AD brain may be aberrant, with new vessels being poorly formed and prone to premature regression due to vascular cell death and growth factor downregulation [129,130].
It has been shown in both AD patients and AD mouse models that angiogenic vessels may differ from patent vessels in several respects, including the presence of abnormal cellular morphology, aberrant branching and disturbed basement membrane and junction formation, and angiogenesis may be localized [128,131,132]. Neoangiogenesis in this context may in fact be detrimental to vascular dynamics and the integrity of the BBB, despite causing increases in vascular density. It is also important to note that many older studies in particular have failed to account adequately for tissue atrophy and AD stage—there are likely changes in the extent and impact of angiogenesis with worsening pathology. Aberrant angiogenesis and endothelial death may be coterminous, with the relative contributions of these processes changing over the course of the disease.
Several markers of angiogenesis, including vascular growth factors, show disturbed expression profiles in AD. Vascular endothelial growth factor (VEGF), a potent and specific mediator of endothelial cell proliferation , is released by numerous cell types, including neutrophils [134,135,136], onto the endothelium to stimulate angiogenesis . VEGF also plays a role in the regulation of BBB permeability, increasing the leakiness of the microvascular wall . VEGF expression changes are well attested in AD patients. Capillary VEGF expression is reduced within the temporal cortex, hippocampus and brainstem in AD patients . Serum levels of VEGF and transforming growth factor-β1 (TGF-β1) are reduced in AD patients [140,141,142], with lower levels correlated with greater cognitive deficits , potentially suggestive of a contribution of angiogenic deficits to the evolution of the disease. VEGF and TGF-β were reportedly found at heightened levels in CSF samples from AD patients .
Lower CSF VEGF levels are correlated with smaller hippocampal volume and ventricular expansion in individuals with high brain Aβ load  and the heightened CSF VEGF levels observed in AD could thus represent a protective response. VEGF release from natural killer (NK) cells and lymphomononuclear cells isolated from AD patients is also impaired and Aβ1-42 suppresses NK cell VEGF secretion . Aβ1-42 also competitively antagonizes VEGF binding to VEGF receptor-2 (VEGFR-2) on endothelial cells , which could contribute to the anti-angiogenic properties of Aβ.
The VEGF-165 isoform reportedly binds to Aβ with high affinity, resulting in its sequestration into senile plaques in the AD brain and potentially reducing VEGF availability at the vasculature [147,148]. In addition to preventing vessel sprouting, VEGF inhibition in spontaneous and implanted tumours in mice has been shown to cause the regression of existing blood vessels, leaving behind string vessel structures—such a process could also be relevant to AD . Considering all of this and the reported neuroprotective functions of VEGF against hypoxic  and excitotoxic damage  and amyloid aggregation , the stimulation of VEGF synthesis and release could represent a therapeutic strategy in the prevention of neurovascular dysfunction and other pathological processes in AD. Indeed, VEGF supplementation in APP/PS1, PDGF-hAPPV717I and TgCRND8 mice results in cognitive improvement, increased angiogenesis, decreased endothelial apoptosis and reductions in amyloid and p-tau load [152,153,154].
Interestingly, insulin has been shown to counteract the effect of Aβ1-42 on VEGF synthesis in NK cells from AD patients . Given the presence of widespread insulin signaling defects in the AD brain , this could be relevant to neurovascular dysfunction and the heightened AD risk conferred by diabetes. One study reported the upregulation of VEGF in perivascular astrocytes and in extracellular deposits close to the vasculature but this was more associated with vessels exhibiting a high degree of cerebral amyloid angiopathy (CAA) . Aβ injection in mice also stimulates increased VEGF synthesis in astrocytes and microglia . Hypoxia is known to stimulate angiogenesis through the upregulation of VEGF synthesis . Astrocytes, in particular, upregulate VEGF synthesis in response to hypoxia, in a process mediated by hypoxia-inducible factor-1 (HIF-1) . Therefore, high astrocyte VEFG around amyloidogenic AD vessels may be a protective response to more pronounced hypoxia in these tissues. Indeed, HIF-1α expression is upregulated in AD microvessels . In the developing retina, it has been shown that endothelial filopodia migrate along astrocyte mesh networks secreting VEGF and astrocytic upregulation of VEGF in response to hypoxia could thus conceivably cause abnormal vessel branching in AD .
There are several other factors that could potentially contribute to dysregulated angiogenesis in the AD brain. Mesenchyme homeobox 2 (MEOX2), a regulator of vascular cell differentiation during angiogenesis, is downregulated in the AD cerebral vasculature, likely contributing to impaired angiogenesis . Matrix metalloproteinase-9 (MMP-9) expression also appears to be reduced in AD microvessels, alongside the upregulation of tissue inhibitor of matrix metalloproteinase-1 (TIMP-1) . MMP-9 is involved in endothelial migration and basement membrane remodeling during angiogenesis . Transferrin receptor upregulation in the cerebral vasculature in some AD patients may be consistent with its role in cell proliferation .
Thus, there is abundant evidence to suggest that impaired angiogenesis in the AD brain is driven at least in part by defective growth factor availability and signaling mechanisms. The expression of vascular growth factors and the occurrence or suppression of angiogenesis may also be influenced by vascular Aβ load and more pronounced hypoxia due to perfusion deficits might activate perivascular astrocytes and alter their angiogenic signaling . Indeed, it has been hypothesized that hypoxia is a key driver of aberrant angiogenesis in AD and increasing hypoxia with progressive pathological impairment may cause stress-related increases in angiogenesis . A direct role for Aβ has also been confirmed in the regulation of angiogenesis and vessel branching [157,167,168]. The expression of growth factors like VEGF is regulated by a variety of pro-inflammatory cytokines that are themselves affected in AD . Given the complex nature of the molecular pathways associated with angiogenesis and the evolving profile of vascular dysfunction in the AD brain, it is perhaps not surprising that the nature and extent of angiogenic changes in previous studies has been inconsistent. Thus, there is a need for studies clarifying this phenomenon and its importance in the development and progression of human AD.
reference link: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6571853/
More information: Zsolt Bagi et al, Association of cerebral microvascular dysfunction and white matter injury in Alzheimer’s disease, GeroScience (2022). DOI: 10.1007/s11357-022-00585-5