Beta-amyloid, blood vessels and brain function

Cerebral amyloid angiopathy

Initially recognized in the context of co-existing AD, it was not until the 1970s that CAA was recognized as a major cause of lobar intracerebral hemorrhage.^{16, 17} The in vivo diagnosis of CAA rests on the demonstration of single or multiple lobar hemorrhages or microbleeds without other evident cause.¹⁸ Sporadic and some forms of familial CAA are characterized by deposition of beta-amyloid in the media and adventitia of small arteries and capillaries of the leptomeninges and cerebral cortex. White matter vessels are much less affected, and the occipital regions are heavily affected, for unclear reasons.¹⁹ In contrast to AD, the predominant ABeta species in vascular deposits is the relatively more soluble Aβ40. Although the exact source of vascular amyloid is unknown, it is believed that ABeta is predominantly generated by neurons and then deposited in the vessel wall.²⁰

Numerous pathological studies have demonstrated histological¹⁹ and ultrastructural²¹ abnormalities of the vessel wall in CAA. Loss of smooth muscle cells occurs with replacement of the vascular media by amyloid. There is a direct toxic effect of wild-type and mutant forms of ABeta, as demonstrated in cell culture studies.^{20, 22} In addition recent research using animal models suggests that vascular amyloid decreases adhesion of vascular smooth muscle cells to the basement membrane²³ and that capillary amyloid deposition may be associated with capillary occlusion.²⁴

These observations of smooth muscle degeneration and capillary occlusion raise the question of whether cerebral blood flow regulation, dependent on normal vascular endothelial and smooth muscle activity, is impaired in CAA. Indeed, an animal model of CAA shows decreased cortical vascular reactivity to carbon dioxide and whisker stimulation in proportion to the severity of vascular amyloid.²⁵

The concept that CAA alters blood flow is further supported by pathologic studies showing an increased prevalence of ischemic lesions in brains with CAA and clinical studies showing that CAA is associated with cognitive impairment independent of the presence of brain hemorrhage, at least partly attributable to ischemia.¹⁰ Case-control pathology studies, predominantly performed in brains with a concomitant clinical diagnosis of AD, suggest that severe CAA is associated with small cerebral infarcts independent of age, hypertension or apolipoprotein E genotype.^26-29 The association is particularly strong for microinfarctions, consistent with the involvement by CAA of small arteries and capillaries. In addition, case series show that white matter demyelination, presumably ischemic, may occur.^30-32 The severity of CAA has been associated with increased odds of antemortem dementia,³³ and worse performance on cognitive tests in persons with concomitant AD,³⁴ in population-based autopsy studies that simultaneously controlled for severity of AD pathology.

MRI and CT are insensitive to microinfarctions but are arguably more sensitive than autopsy to the presence of white matter lesions. Studies of consecutive patients with CAA presenting with brain hemorrhage show that CT or MRI white matter lesions are common,³⁵ of greater severity than in healthy aging,³⁶ and are associated with cognitive impairment independent of the effects of the brain hemorrhage.³⁵ Familial forms of ABeta CAA also show extensive white matter lesions.³⁷ Newer MRI measurements of water proton diffusion are sensitive to subtle alterations in tissue microarchitecture and show more widespread abnormalities in CAA that represent the combined effects of tissue pathology (e.g., white matter lesions and microinfarctions) and its consequences such as demyelination and axonal degeneration.³⁸ In CAA patients a whole-brain measure of mean apparent diffusion coefficient (ADC) may correlate better with cognitive impairment than other MRI markers.³⁹

Altered vascular function has recently been demonstrated in humans with CAA.⁴⁰ In this pilot study 11 CAA patients presenting with stroke or memory symptoms, and diagnosed as having probable CAA using the validated Boston Criteria,¹⁸ were compared to 9 healthy controls with similar ages and distributions of vascular risk factors. Evoked blood flow velocity in the posterior cerebral artery in response to a visual stimulation task was measured by transcranial Doppler ultrasound and was reduced in the CAA cases compared to controls (Figure 2). Among CAA patients lower evoked flow velocity response correlated with a higher prevalence of MRI white matter lesions and MRI microbleeds. These data are consistent with observations in a mouse model²⁵ with the exception that in humans, in contrast to the animals, there was little effect of CAA on the evoked response to CO₂, measured in the middle cerebral artery. The reason for the differential effects of CAA on visual evoked flow in the posterior cerebral artery, compared to CO₂ evoked flow in the middle cerebral artery, could not be determined but could reflect relatively less CAA involvement in the middle cerebral artery territory compared to the posterior cerebral artery territory, or the different molecular mechanisms for vasodilation.⁴⁰

Alzheimer’s disease

AD is characterized pathologically by senile plaques, neurofibrillary tangles, and synaptic and neuronal loss. Studies demonstrate that vascular abnormalities also accompany AD, however, even in the absence of concomitant atherosclerotic vascular disease. The most frequent vascular abnormality seen in AD is CAA.⁴¹ Recent careful studies of capillary morphology and density suggest that a broader spectrum of vascular abnormalities may accompany AD and may be partly independent of CAA. Some have even proposed that vascular dysfunction is a critical early component of Alzheimer pathology.⁹

Severe Alzheimer’s pathology is usually accompanied by some degree of CAA.^{41, 42} These autopsy studies are supported by MRI studies showing that lobar microbleeds, associated with CAA, are frequently present in the brains of persons with AD even though symptomatic lobar ICH is rare.^43-45 The pathology of CAA associated with AD shows differences from the pathology of CAA in the absence of AD, with more frequent capillary relative to arterial CAA.^{41, 42} In one study capillary CAA was predominantly composed of Aβ42 deposits in the glia limitans rather than deposits in the vessel wall, in contrast to arterial CAA in which Aβ40 deposits predominated and typically involved the arterial wall.⁴⁶ The authors interpreted their data to suggest that capillary CAA has a neuronal origin and travels along peri-capillary interstitial pathways before depositing in the capillary adventitia. Data from a transgenic mouse model suggests that capillary CAA is associated with capillary occlusion and decreased blood flow.²⁴ Other authors have also documented significant arteriolar vascular deposition in addition to capillary deposition, with loss of smooth muscle actin, in AD compared to controls.⁴⁷ These CAA-related vascular abnormalities may contribute to impaired cognition in AD. In the Honolulu-Asia Aging Study the combined presence of both moderate-severe CAA and AD on autopsy was associated with greater cognitive impairment in life than the presence of AD alone.³⁴

The microvasculature of patients with AD exhibits morphological changes in addition to vascular or perivascular amyloid deposits. Careful autopsy studies show that capillary density, length, and mean diameters are decreased in AD in comparison with similar-age controls, sometimes with segmental narrowing or dilation.^48-51 An initial quantitative study of hippocampal capillary characteristics between AD and controls did not find overall differences, but suggested that associations between AD pathology and capillary morphology may be restricted to the hippocampal subregions most affected by AD.⁵² A subsequent study compared hippocampal CA1 and entorhinal cortex capillary parameters with the regional burden of Alzheimer pathology and found that lower total capillary number and lower mean capillary diameter correlated with higher burden of senile plaques and neurofibrillary tangles.⁴⁹ Lower mean capillary diameter was independently associated with greater antemortem cognitive disability, as measured by the Clinical Dementia Rating scale, even after controlling for the burden of senile plaques and neurofibrillary tangles.⁴⁹ Similar observations have recently been published in a transgenic mouse model of AD, where a vascular corrosion cast was produced by intravascular infusion of a resin followed by removal of the intervening tissue, allowing examination of the 3-dimensional architecture of the microvasculature.⁵³ Electron microscopy of the cast showed “holes” of decreased capillary density alternating with regions of increased density of morphologically abnormal capillaries.⁵³

These observations have led some authors to hypothesize that cerebral hypoperfusion is an early phenomenon in AD that induces neuronal dysfunction and injury.⁹ Indeed, cerebral hypoperfusion is observed in both the early and late stages of AD,⁵⁴ Although this phenomenon is often attributed to decreased neuronal metabolism or neuronal loss, impaired vascular function may also contribute. Similarly, numerous studies of neuronal activation-associated changes in blood oxygen level dependent (BOLD) response have been interpreted as solely reflecting neuronal function without consideration of possible influences from vascular pathology. There is a relative dearth of physiologic studies in AD that examine the vascular response to other stimuli such as carbon dioxide inhalation. A recent study found that middle cerebral artery evoked flow velocity to carbon dioxide, measured by transcranial Doppler ultrasound, was reduced in AD compared to controls.⁵⁵ By contrast other blood flow studies have failed to find abnormal vascular reactivity to carbon dioxide in AD, although small sample sizes limit the strength of the conclusions.^{56, 57}

Soluble beta-amyloid and blood vessel function

Experimental data shows that soluble ABeta can cause abnormal vascular reactivity in the absence of vascular deposition or vessel wall dysfunction. These experiments raise the possibility that vascular dysfunction could be an early step in beta-amyloid diseases and could even precede significant ABeta deposition. Application of exogenous ABeta to normal blood vessels ex vivo causes endothelium-dependent vasoconstriction.^{58, 59} Likewise application of exogenous ABeta, particularly the ABeta-40 species, to the mouse neocortex induces vasoconstriction in vivo.⁶⁰ Transgenic mice that over-express amyloid precursor protein exhibit reductions in cerebral blood flow at an early age, prior to deposition of ABeta as plaques,⁶¹ and display abnormal vascular autoregulation⁶² and attenuation of functional hyperemia.⁶³ These effects appear to be mediated by endothelial dysfunction with formation of oxygen free radicals⁵⁸ by NADPH oxidase.^{64, 65} The finding that soluble ABeta impairs vasodilatory responses in APP-overexpressing animals has recently been questioned, however,²⁵ and the significance of these experimental findings to human pathophysiology is uncertain at present.

Conclusions and Future Directions

Vascular beta-amyloid deposition is a relatively common pathology of aging and is extremely common in person with AD. Beta-amyloid is directly toxic to smooth muscle cells. Damage to the vessel wall caused by ABeta deposits can result in vascular rupture with intracerebral hemorrhage. ABeta-related vascular dysfunction, manifest as impaired vascular responses to stimuli including changes in systemic blood pressure (impaired autoregulation), arterial partial pressure of carbon dioxide and neuronal activity, seems likely on the basis of experimental evidence but remains underexplored in humans. Vascular dysfunction with reduction in blood flow could be the mechanism by which CAA is associated with microinfaction and small cortical infarcts.

There is a need for further studies to address the relationship between vascular function, brain lesions and clinical function including cognition in persons with beta-amyloid diseases including CAA and AD. Given the high prevalence of CAA in AD, studies are urgently needed to determine whether CAA is an active player in producing cognitive decline, or a bystander. These studies will be aided by careful measurement of beta-amyloid disease burden and vascular reactivity.

Thankfully, recent advances in experimental models and human imaging may aid studies of beta-amyloid and its effects on the vasculature by allowing more careful serial assessments of beta-amyloid burden. The development of mouse models with cranial windows that allow serial amyloid imaging, by multiphoton microscopy,⁶⁶ will enable longitudinal experimental studies of vascular beta-amyloid growth and its determinants. In humans the development of beta-amyloid-binding PET ligands offers, for the first time, the ability to measure regional and global beta-amyloid deposition in a serial longitudinal manner.^67-69 Originally validated as a marker of parenchymal amyloid deposition and AD, it is now clear from autopsy studies that Pittsburgh Compound B (PIB)⁶⁷ also binds to vascular beta-amyloid.^{70, 71} A small case series suggests that occipital PIB retention is relatively greater in CAA compared to AD, consistent with the known topographical distribution in CAA.⁷²

With improved in vivo measurements of beta-amyloid, careful physiologic investigations of vascular reactivity, and longitudinal study designs, it should be possible to more firmly establish the role of blood vessel function in producing impairment in CAA and AD. The impact of vascular dysfunction on age-related changes in cognition may currently be underestimated. Identification of vascular dysfunction as a significant component of CAA and AD may open up new therapeutic approaches for these currently largely untreatable diseases.