Microvascular Pathology and Morphometrics of Sporadic and Hereditary Small Vessel Diseases of the Brain

Mural cells

VSMCs within arteriolar walls and pre‐capillary arterioles serve as contractile elements and control blood flow responses in times of increased parenchymal demand 6, 127. Progressive pathological changes in VSMCs were described in both sporadic SVD 19 and hereditary SVDs 122. The degenerative process in CADASIL appears most aggressive almost irrespective of genotype. Loss of arterial VSMCs is followed by fibrosis of the tunica media in small‐ and medium‐sized penetrating arteries 64; arteriosclerotic changes are concomitant with stenosis, especially at the arteriolar level, through intimal thickening and wall expansion with extracellular matrix components such as collagens, laminin and fibronectin 75, 79, and compounded by altered protein–carbohydrate interactions 12, 16, 33. It is not clear whether failure in NOTCH3 signaling is also responsible for the gradual degeneration of VSMC in sporadic cases of Binswanger‐type or in hypertensive disease 86, 107. Quantitative VSMC numbers per vessel segment length in SVDs have not been fully explored, but a study in CADASIL has suggested that VSMCs undergo apoptosis akin to neurons in the neocortex 114.

Pericytes

Pericytes juxtaposed to cerebral microvessels, most prominently wrapped around capillaries have received recent attention 22, 74. They interact with other cells within the neurovascular unit by direct contact or cell signaling mechanisms to regulate microcirculatory functions. Pericytes also likely contribute to pathogenic mechanisms in the smallest arterioles and capillaries. For example, in CADASIL, we and others 28 have observed significant abundance of granular osmiophilic material (GOM) deposits positive for NOTCH3 extracellular domain (N3ECD) around capillaries within CADASIL brains 122. Platelet‐derived growth factor receptor‐β positive pericyte‐like cells are increasingly evident in sclerosed arterioles in both CADASIL and sporadic SVD (Figure 1 and L.J.L. Craggs and R.N. Kalaria, unpub. data).

Cerebral endothelium

Endothelial cell abnormalities and BBB dysfunction may further contribute to WM damage. Endothelial changes have been previously described in SVDs with particular reference to “blebbing,” change in volume of the cytoplasm and the presence of compact bundles of microfilaments within the cytoplasm of endothelial cells in CADASIL 75, 95, 121. Neuroimaging investigations tracking signal enhancement after gadolinium suggest that breakdown of the BBB (see previous discussion) occurs in areas of leukoaraiosis and may mediate subsequent cellular changes 117, 118. These changes are associated with chronic prothrombotic endothelial dysfunction in cerebral SVD 48 also involving the WM 11. There appears also to be a cerebral response to the SVD, both sporadic and CADASIL, by increasing endothelial thrombomodulin 38.

BBB disruption 1 can cause osmotic demyelination and result in increased permeability of the vessel wall and mobilization of inflammatory factors, such as macrophages, lymphocytes and complement, which also causes myelin damage 5. However, the projected meager perfusion due to capillary loss or abnormalities occurring prior to leukoaraiosis corroborates the finding of a chronic hypoxic state in the deep WM 32, which also releases certain growth‐promoting factors 101. It is not clear if microvascular length density increases in SVD in the most vulnerable subcortical gray or WM regions as it does in post‐stroke dementia (PSD), which suggests either an increase in angiogenesis or the formation of newer microvessel loops in response to cerebral hypoperfusion 14. However, it is likely that microvascular diameters are decreased in SVD as found in VaD, suggesting increased vasoconstriction.

Vascular basement membrane

The basement membrane of cerebral vessels comprises several proteins, of which laminins, collagen (COL), nidogens and perlecans are the major constituents 49. The collagens are the most often investigated in the context of pathological changes in basement membranes. They are increased in hypertensive disease and associated with SVD pathology, which often reveals concentrically thickened vessel walls or hyaline arteriolosclerosis 37, and ultimately decreases the lumen diameter, impinging auto‐vasoreactivity of the vessels described as an “Earthen pipe state” 86. We evaluated COL4 staining in our cohort of hereditary and sporadic SVDs and found region‐specific differences in COL4 staining in sporadic SVD, PADMAL and Swedish hMID, where we observed increased COL4 in the frontal lobe of sporadic SVD cases, whereas the Swedish hMID cases showed increased COL4 within the basal ganglia 122. Comparisons of the different types of collagens, for example, COL3 and COL4, show the progressive changes that occur in the mobilization and restructuring of collagen‐containing vascular components (Figure 2). The pathophysiological significance of basement membrane changes is also likely to impact on clearance of fluids, solutes and toxins along the perivascular drainage route 9, 49. Enhanced COL1 and COL3 deposition has been observed in concentric rings around veins, too, termed as periventricular venous collagenosis 11, 80.

The type 4 collagens, encoded by multiple genes of which COL4A1 and COL4A2 are highly conserved, have gained particular importance in basement membrane changes in SVDs but also impact on large vessel disease 68. Genetic variation within both the COL4A1 and the COL4A2 genes have been linked to cerebral SVDs, which usually occur with systemic vascular abnormalities, causing a wide range of clinical phenotypes (Table 1). Thus, COL4A1‐related disorders show a range of phenotypes with overlapping features, including autosomal dominant type 1 porencephaly 40, brain small vessel disease with hemorrhage 41, brain small vessel disease with Axenfeld‐Rieger anomaly 99, and hereditary angiopathy with nephropathy, aneurysms and muscle cramps (HANAC) syndrome 92, 109. The consequence of altered collagen IV proteins within vessel walls is clearly catastrophic; infantile intracerebral hemorrhage and hemiparesis are common with COL4‐affected families, and studies in genetically modified mice expressing mutant COL4A1 have reported 50% mortality of pups at birth, a problem that was overcome by surgical delivery 41. In adulthood, subjects with COL4 mutations commonly have aneurysms within the carotid and cerebral arteries. Segments of the most commonly affected arteries include the intracranial carotid artery, middle cerebral artery and, least frequently, the basilar artery 71, illustrating the vital role collagens play in vascular wall structure and support.

Lacunar infarcts

Lacunar infarcts, occurring as complete or cavitating lesions frequently in both subcortical gray and WM, are the hallmark parenchymal lesions for SVD. Neuroimaging studies suggest that lacunes occur in greater frequency in CADASIL 90 and possibly other disorders such as CARASIL. They increase with age and are associated with cognitive impairment 73, 115, but are indistinguishable between sporadic or hereditary SVD disorders. Lacunar infarcts are frequently multiple and bilateral, and often coexist with other vascular lesions, for example, large infarcts or diffuse WM damage. Interestingly, the majority of incident lacunes develop proximal to WM hyperintensities along the course of perforating vessels supplying the respective brain region. Lesion prevalence maps in different stages of disease show that lesions spread toward the subcortical regions in both sporadic SVD and CADASIL. Whether single or multiple, they may be asymptomatic, depending on their location and the loss of volume of normal brain tissue. Lacunar volumes may vary widely and do not appear to be related to age or associated cerebral lesions such as WM hyperintensities in CADASIL. However, volumes of lacunar lesions rather than that of WM hyperintensities affect cortical depth and other structural changes, supporting the role of the neocortex in subcortical ischemic VaD 58. Lacunar infarcts and related microstructural alterations may also affect global brain volume or atrophy. To distinguish perivascular cavities or spaces, it has been suggested that lacunes be classed into three subtypes: lacunar infarcts, lacunar hemorrhages and dilated perivascular spaces 123. Lacunar infarcts usually result from progressive SVD manifested as hypertensive angiopathy or possibly microthrombi that may involve stenosis caused by hyalinosis. Apart from critical lesions occurring often in the internal capsule or caudate nucleus, there appear to be no pathological differences between symptomatic and asymptomatic patients. Perivascular edema and thickening, inflammation and disintegration of the arteriolar wall were common, whereas vessel occlusion was rare 4. Occasionally, lacunes may represent small hemorrhages or dilated perivascular spaces without infarction or hemorrhage. Microlacunes have also been described in SVD that essentially should be thought of as large cystic microinfarcts.

WM changes

Diffuse and focal WM lesions are another prominent hallmark of SVD 54. Neuroimaging and pathological studies show that WM changes are invariably associated with cognitive abnormalities 23, 82, 90. WM alterations are much more profound in the hereditary SVDs such as CADASIL and CARASIL. They may occur in the absence of lacunar infarcts and extensive WM hyperintensities (WMH) appear associated with increased brain volume, particularly in CADASIL 125. In this disorder, WMH relate not only to loss of WM components but also to a global increase of water content in the cerebral tissue. WM hyperintensities on T2‐weighted MRI or leukoaraiosis as a decreased signal on CT may not only incorporate WM rarefaction, incomplete infarction, lacunar strokes, perivascular spacing and demyelination, but sometimes also axonal and Wallerian degeneration. Diffusion tensor imaging 89 has demonstrated how tissue microstructural changes in WM tracts and subcortical regions, for example, putamen and thalamus, are related to worsening clinical outcomes in SVD and CADASIL. The widespread WM axonal changes, particularly in the frontal lobe, may arise from differential stenosis and sclerosis of arterioles 20, possibly affecting certain axon bundles connecting to targets in the subcortical structures, specifically degeneration of thalamocortical pathways causing cortical atrophy 59.

There is some controversy as to whether deep or periventricular lesions are of greater importance but this depends on the definition of boundaries between the periventricular and deep WM 66. While lacunar infarcts are produced when the ischaemic damage is focal and of sufficient severity to result in a small area of necrosis, diffuse WM change is considered a form of rarefaction or incomplete infarction, where there may be selective damage to some cellular components. Although the U‐fibers are frequently spared, WM disease may comprise several patterns of alterations including pallor or swelling of myelin, loss of oligodendrocytes, axons and myelin fibers, cavitations with or without presence of macrophages and areas of reactive astrogliosis 100, where the astrocytic cytoplasm and cell processes may be visible with standard stains. Lesions in the WM also include spongiosis, that is, vacuolization of the WM structures and widening of the perivascular spaces 123.

Whereas WM changes focus on narrowing of the arterial system, in many cases, occlusion of veins and venules by collagenous thickening of the vessel walls also occur. The thickening of the walls of periventricular veins and venules by collagen (collagenosis) increases with age, and perivenous collagenosis (see previous discussion) increases in concert with leukoaraiosis 8. The presence of apoptotic cells in WM adjacent to areas of leukoaraiosis suggests that such lesions are dynamic, with progressive cell loss and expansion 7. Vascular stenosis caused by collagenosis may induce chronic ischemia or edema in the deep WM leading to capillary loss and more widespread effects on the brain 9, 10.

Perivascular spaces (PVS)

Dilated PVSs are a frequent finding in the pathology of SVD (Figure 2). Both neuroimaging and pathological studies show that the severity of dilated PVS increases with age regardless of the brain region. Previous neuroimaging studies have indicated that the size of dilated PVS in the basal ganglia and frontal lobe WM correlates with cognitive impairment 21, 91. In CADASIL, the severity of dilated PVS in the temporal lobes or subinsular areas was found strongly and specifically related to the extent of WMH 126. Consistent with this finding, we previously demonstrated that increased volumes of PVS were related to WM myelin protein changes 123, suggesting that reduction in WM volume is one factor that creates PVS. Another important factor involved in the expansion of PVS is likely the lack of drainage of solutes including degraded proteins 15.

Incomplete infarcts

Larger areas of incomplete infarction may extend into the WM 54. These are characterized by mild to moderate loss of oligodendrocytes, myelin and axons in areas where there may be hyalinized vessels 13. The parenchymal changes are normally accompanied by astrogliosis, some microgliosis and macrophage infiltration with usually no quantification of such response. The morphology of incomplete or subinfarctive changes, although suspected to be associated with cognitive function, is not consistently described in SVD. It may variably manifest as tissue rarefaction assessed by conventional stains and revealed as injury response such as microgliosis and astrocytosis, or the presence of other “reactive” cells or surrogate markers of dendritic, synaptic or axonal damage.

Cerebral microinfarction

The accumulation of small, even miniscule, ischemic lesions is an important substrate of SVD 60. Microinfarcts may or may not involve a small vessel at its center but are foci with pallor, neuronal loss, axonal damage (WM) and gliosis. They are estimated to occur in their thousands and described as attenuated lesions of indistinct nature occurring in subcortical regions in sporadic SVD 24 and hereditary SVDs such as CADASIL 121. Microinfarction in the subcortical structures has been emphasized as a substrate of cognitive impairment 3, 60 and correlated with increased Alzheimer type of pathology, but cortical microinfarcts also appear to contribute to the progression of cognitive deficits in brain aging 67. Furthermore, microinfarcts even in border‐zone (watershed) regions may aggravate the degenerative process as indicated by worsening impairment in AD 106.

Neocortical microinfarcts are increased in the presence of CAA 84, 103, but are rarely observed in subcortical VaD linked to SVD 83, 106 or in CADASIL. However, cortical microinfarcts and, to lesser extent, periventricular demyelination were associated with cognitive decline in individuals at high risk for dementia 39. It is proposed that the changes in hemodynamics, for example, hypotension and atherosclerosis, may play a role in the genesis of microinfarcts in watershed regions.

Cerebral microbleeds and hemosiderin

Cerebral microbleeds are an imaging construct to represent ferromagnetic hemosiderin iron derived from extravasation of erythrocytes. Cerebral microbleeds detected by T2‐W* or echo gradient MRI are also associated with histopathological evidence of lipohyalinosis and CAA 31. They are likely a surrogate marker of SVD evident on MRI along with lacunes and WM changes 111. The prevalence of radiological microbleeds in SVD ranges from 35% to 85%. Both radiological cerebral microbleeds and foci of hemosiderin containing single crystalloids, or larger perivascular aggregates, are found in the brains of older subjects including those diagnosed with VaD and CADASIL 72, but the radiological and pathological relationship between these findings has not been entirely clear. Recent evidence suggests that cerebral microbleeds detected by MRI are a surrogate marker for ischemic SVD rather than exclusively hemorrhagic diathesis 55. Greater putamen hemosiderin was significantly associated with indices of small vessel ischemia, including microinfarcts, arteriolosclerosis, perivascular spacing and lacunes in any brain region, but not large vessel disease or whole brain measures of neurodegenerative pathology. It appears that brain iron homeostasis and small vessel ischemic change are responsible for these rather than minor episodes of cerebrovascular extravasation.

Cerebral vessel sclerosis

The understanding of the cellular mechanisms on how cerebral arterial vessels alter with age or disease in the absence of atherosclerosis is still a matter for investigation. However, the degree of vessel wall changes is commonly reported in pathological assessments as a way of assessing extent of SVD (Figure 3). This tends to be qualitative and does not attempt to quantify the true burden. Nevertheless, some studies have attempted to quantify the pathological changes observed in several cerebrovascular diseases and provided insights as to how the extent of SVD burden varies in relation to hypertension, CADASIL and sporadic SVD (Table 2). The quantitative assessment of vascular wall changes provides a more sensitive method for comparing between diseases, which may appear qualitatively the same 19. Additionally, it could also be presumed that morphometric techniques for the purpose of measuring vessel wall thickness are less likely to encounter inter‐rater variations. Despite this, there still remains considerable variation in morphometrical methods used for assessing vessel wall stenosis, where different investigators use different section thickness, stains and equations to derive the final ratio of vessel wall to lumen changes (Table 2). The earliest methods on quantifying cerebral arterial thickening were derived by Furuyama 36 for assessing the effects of hypertension on systemic arteries. Furuyama's method required the radius of an artery's lumen to the midpoint of the medial layer “R” and the thickness of the medial layer “D” were used to calculate the degree of medial layer thickening. This equation was based on the assumption that the ratio between the radius, that is, distance from lumen center to central point of the media, and the thickness of medial layer, would increase as the medial layers become stretched and expanded with progression of disease, mostly due to uncontrolled hypertension. Okeda et al 87 utilized this method to assess medial layer thickening in the arachnoid and medullary arteries in several disorders including Binswanger's encephalopathy, CADASIL and hypertension. Despite some limitations of the study, for example, low sample numbers, their results provided clear evidence of the subtle differences in the extent of vessel wall changes between disorders, most notable that CADASIL vessel wall changes were more severe than those observed in Binswanger's disease 86. Prior to these studies, Lammie et al 70 developed the “sclerotic index” (SI) ratio, where the proportion of external and internal (luminal) diameter of the vessels is converted to a simple ratio to represent the extent of lumen narrowing. Using this system, three subclasses of SI were formulated, where a ratio of 0.2 to 0.3 denoted healthy vessels, a ratio of 0.3 to 0.5 was considered to be diseased and a ratio of over 0.5 was denoted as severe. Importantly, this study examined arterioles across the whole of the cerebrum and provided qualitative meaning to the SI ratio, using the three tier scale. They were also able to identify subtle differences to changes in vessels between the cortical and subcortical regions and compare the deep gray matter and WM. The SI ratio has since been widely used to estimate the extent of vessel wall degeneration and stenosis in morphometric studies. However, there is still no consensus on the most appropriate section thickness or size of vessels to be assessed. It is possible that the variations in the above parameters may explain the varying results. For example, while focusing on CADASIL, some studies using 5‐µm‐thick sections have reported mean SI to be in the region of 0.6–0.7 and for control cases as 0.4, whereas our studies using sections of 10 µm thickness indicate estimates comparable to Lammie et al 70: mean SI of 0.4 for CADASIL and 0.3 for controls.

Another important obvious factor to consider is the size of the vessels. Okeda et al 86 had first verified that the smaller the arteriole, the more severe the vascular smooth muscle loss. Similarly, Miao et al 79 showed that smaller arterioles exhibited more severe effects of vessel wall thickening on stenosis. In our study of several SVDs 19, we showed that smaller external diameters, followed by location were factors in differential involvement of arterioles between frontal lobe and basal ganglia locations. We also showed that there was an apparent low burden of severely sclerosed arterioles with a SI value of >0.5. Previous studies had reported SI values in CADASIL patients to be well over 0.5 in most cases; however, while we did observe some arterioles with SI considered severe, this was only 7.4% of vessels in the frontal lobe of CADASIL compared with 19.9% of vessels within the basal ganglia. In comparison, the older sporadic SVD subjects had only 6.3% arterioles in the severe category in the frontal lobe and 7.3% in the basal ganglia 19. The large difference between two diseases thought to mirror each other illustrates the power of morphometric assessment of arteriolar sclerosis and provides insights into potential mechanisms for differences in clinical presentations of subtypes of SVD. Our study also provided some potential for a threshold of SVD, which may be related to cognitive function. We found that the proportion of vessels within the healthy SI range was greater than 60% in cognitively normal controls, aged between 49 and 94 years of age, whereas in the SVD subjects, there were 36%–58% in the frontal lobe, and 35%–59% in the basal ganglia, indicating a subtle but potential difference in the cerebral tolerance to burden of vessel stenosis.

WM changes

Another type of measurable cerebral change attributed to SVD is attenuation of the WM. We have previously shown that the greatest loss in myelin staining is encountered in VaD, most of which exhibited SVD 54. Furthermore, axonal abnormalities are also evident in SVD. It may be more useful to fully quantify the extent of WM damage rather than using a semiquantitative 0–3 scoring scale, as in our experience, a WM score of 0–1 is very rare in any cases aged over 40, and the scale of 2–3 does not differentiate between age‐associated changes in WM damage and extensive WM damage associated with dementia. We have previously reported on WM damage in young and old control cases without significant cognitive impairment and found that their WM damage may appear extensive enough to score 3 on the accepted scales 24, 102. While other reliable markers are urgently needed to assess both myelin and damaged axons, we showed that amyloid precursor protein immunoreactivity in axons correlates strongly with severe WM damage albeit in CADASIL 20. Quantitative measures for WM damage, either using histochemical stains or quantification of myelin staining using the myelin index, may allow further differentiation of subtle changes in WM pathology and delineate mechanisms in different disorders.