Despite having very different pathologies, MS, leukoaraiosis, and NPH all share some common characteristics. In all three conditions, cerebral blood flow (CBF) is reduced [
18‐
21]. Both MS [
6,
10] and leukoaraiosis [
13,
14,
22,
23] are characterized by changes in the white matter (WM) in the periventricular region, and enlarged lateral ventricles are associated with both MS [
24,
25] and NPH [
26,
27]. Furthermore, some clinical characteristics are also shared. Gait disturbances [
28‐
31], reduced cognitive ability [
30,
32,
33], and loss of bladder control [
30,
34,
35] have been reported for all three conditions. This suggests that the pathophysiology of these disparate conditions might share a common feature. Having said this, all three diseases display marked pathophysiological differences. For example, MS is an autoimmune disease, characterized by brain atrophy [
36,
37], and it is thought that this is primarily responsible for ventricular enlargement [
24,
25]. Although the ventricles are also enlarged in NPH, brain parenchymal atrophy is not evident [
26], and a measure of ventricular normality can sometimes be restored by the surgical insertion of a shunt to drain away excess CSF [
27,
38,
39]. Although MS and leukoaraiosis both exhibit periventricular WM changes, leukoaraiosis is thought to be one presentation of cerebral small-vessel disease [
40], whereas MS is a chronic inflammatory demyelinating disease of the central nervous system (CNS) [
41]. Consequently, MS therapies focus on preventing axonal demyelination and promoting remyelination [
42,
43], whereas vigorous treatment of cardiovascular risk factors has been advocated to prevent the development of leukoaraiosis, and to reduce the attendant risk of stroke and dementia [
44].
Multiple sclerosis
Since the earliest years of research into MS, there has been suspicion that the venous system might be involved in its etiology, with Dawson [
10], Putnam [
6,
11] and others [
45‐
49] all implicating veins in the pathophysiology of the disease. MS plaques are often venocentric, and frequently form around the periventricular veins [
6]. Dawson [
10] reported that finger-like plaques form at the junction of the subependymal and medullary veins in the periventricular WM. Putnam and Adler [
6], commenting on the appearance of these ‘Dawson’s fingers’, observed that the medullary veins were enclosed in a sleeve of plaque, and that, adjacent to the plaques, the veins were grossly distorted and distended. Others [
45,
50‐
52] have also shown that inflammatory lesions tend to form axially around veins in the WM, with Tallantyre
et al. [
53] finding 80% of MS lesions to be perivenous in nature. Kidd
et al. [
50] showed that lesions in the grey matter (GM) are also associated with veins, with the majority of cortical lesions arising within the territory of the principal vein, V
5, whose course begins in the WM [
54], and the remaining cortical lesions forming in the region drained by its branches or those of the superficial veins. Others have confirmed these observations, finding intracortical [
55‐
57], leucocortical [
55], and sub-cortical [
52] lesions to be perivenous. However, the connection with the venous system has largely been ignored, with the focus of MS research turning instead towards the involvement of the immune system in the disease [
58,
59].
Recently, there has been renewed interest in studying vascular changes associated with MS [
60‐
62]. This has been precipitated by the controversial finding that abnormalities in the extracranial venous system appear to be associated with the disease [
4,
7,
8,
63]. This condition, known as chronic cerebrospinal venous insufficiency (CCSVI), is characterized by multiple intraluminal stenotic malformations of the principal venous-drainage pathways, particularly in the internal jugular veins (IJVs) and the azygos vein, and has been shown to be associated with impaired blood flow from the brain to the heart in patients with MS [
8], with the hydraulic resistance of the cerebral-venous drainage system being on average 63.5% greater in CCSVI-positive individuals [
64]. CCSVI also appears to be associated with changes in the intracranial vasculature, with a strong correlation shown between CCSVI severity and decreased CBF in both the WM and GM of patients with MS [
65]. In addition, Zivadinov
et al. [
9] reported a marked reduction in venous vasculature visibility (VVV) on susceptibility-weighted imaging (SWI) for cerebral veins of less than 0.3 mm diameter in patients with MS compared with controls, a phenomenon that is strongly statistically associated with CCSVI (
P < 0.0001). This finding appears to corroborate the work of Ge
et al. [
66]. However, unlike Ge
et al., who attributed the reduction in VVV to hypometabolic status in the brain parenchyma of patients with MS, Zivadinov
et al. performed a pre-contrast and post-contrast SWI venography experiment, which indicated the reduction in VVV to be due to morphological changes in the cerebral veins of patients with MS. Indeed, such was the clear-cut nature of these venous changes that Beggs
et al. [
67] were able to distinguish between patients with MS and healthy controls with 100% accuracy using cerebral-venous data alone.
These findings reinforce a large body of evidence connecting MS with alterations in the cerebral vascular bed. Using tomography, a number of early investigators [
68‐
71], found reduced CBF in the GM and WM of patients with MS. However, this work received little attention and it was not until the advent of magnetic resonance imaging (MRI) that interest was renewed [
61]. Using dynamic susceptibility contrast-enhanced MRI, Law
et al. [
18], identified a 53.4% decrease in CBF throughout the normal-appearing white matter (NAWM) in patients with relapsing–remitting (RR) MS compared with controls. This was accompanied by a twofold increase in vascular mean transit time (MTT), and a 13.6% decrease in WM cerebral blood volume (CBV). Adhya
et al. [
21] studied tissue perfusion in the NAWM of patients with primary progressive MS, relapsing-remitting (RR) MS, and healthy controls. They also found CBF and CBV to be significantly decreased in all NAWM regions in both forms of MS compared with controls. Similarly, Ge
et al. [
19] found reduced CBF with significantly prolonged MTT in the NAWM to be a feature of MS. Varga
et al. [
20] reported blood flow to be particularly low in the periventricular region, with CBF in the NAWM in this region being significantly lower in patients with MS compared with controls. Interestingly, they also found CBF to be decreased in the sub-cortical normal-appearing grey matter in patients with RR MS, suggesting a continuum of decreased tissue perfusion, beginning in the WM and spreading to the GM as the disease progresses [
60]. Collectively, these findings indicate that hypoperfusion of the WM is a consistent phenomenon in MS, whatever the disease subtype [
61]. Several researchers have also found MS to be associated with reduced CBF in the GM. Rashid
et al. [
72] found hypoperfusion in several cortical areas of patients with RR and progressive MS. Investigating tissue perfusion in the thalamus, putamen, and caudate nuclei of patients with MS, Inglese
et al. [
73] found a decrease in CBF in the deep GM, the magnitude of which increased with the severity of the disease. These findings, along with those relating to the WM, suggest that MS is associated with systemic changes in blood flow through the cerebral vascular bed, something highlighted by Mancini
et al. [
74], who found the mean tissue-vein transit time to be 3.2 seconds in patients with MS, compared with only 2.9 seconds in healthy controls.
Venous hypertension in the dural sinuses is known to inhibit absorption of CSF through the arachnoid villi (AV) [
75,
76]. Zamboni
et al. [
4] reported reduced CSF net flow and increased CSF pulsatility in the aqueduct of Sylvius (AoS) in patients with MS, and found this to be strongly associated with CCSVI severity. Magnano
et al. [
77] also found MS to be strongly associated with increased aqueductal pulsatility and reduced CSF net flow. Although Magnano
et al. did not specifically consider CCSVI, their findings are consistent with those of Zamboni
et al., and suggest that venous hypertension may be a feature of MS. Abnormal CSF hydrodynamics have also been implicated in the formation of cortical lesions in MS. Sub-pial lesions, which appear not to be perivenous, cover extensive areas of the cortex, and extend from the surface into the brain [
55]. They appear to be mediated by infiltrates, generated by inflammatory cells in the meninges or the CSF, which diffuse inwards from the surface of the brain [
55,
78]. Kutzelnigg
et al. [
79] found sub-pial demyelination to be most pronounced within deep invaginations of the cortex, and suggested that this reflected regional differences in CSF flow, with extensive demyelination occurring in areas of CSF stasis.
Leukoaraiosis
Leukoaraiosis is a radiological finding, characterized by WM hyperintensities in the periventricular region on T2-weighted MRI scans [
80], which is associated with diverse clinical symptoms, including cognitive impairment [
81], vascular dementia [
82,
83], gait disturbance [
30], and enhanced risk for stroke [
84]. Although leukoaraiosis is primarily a pathology associated with aging [
83,
85], it shares several similarities with MS [
62,
86,
87]. Both diseases affect the WM and are associated with demyelination [
13,
82]. In a similar manner to MS, leukoaraiosis is characterized by WM morphological changes around the periventricular veins [
13,
14,
22,
23]. Although not fully understood, leukoaraiosis is thought to be associated with chronic cerebral ischemia [
88]. In cases of hypoxic/ischemic injury, histological changes of the WM can range from coagulative necrosis and cavitation to non-specific tissue changes such as sponginess, patchy demyelination, and astrocytic proliferation [
88]. Such changes are consistent with the lesions seen in patients with leukoaraiosis [
89], suggesting that ischemia is closely associated with the condition [
88]. In particular, leukoaraiosis is characterized by non-inflammatory collagenosis of the periventricular veins [
13,
22], resulting in thickening of the vessel walls and narrowing, or even occlusion, of the lumen [
13]. Moody
et al. [
13] found a strong association between the probability of severe leukoaraiosis and periventricular venous collagenosis (PVC).
Mirroring the cerebral hemodynamics of MS, several researchers have reported leukoaraiosis to be associated with reduced CBF [
23,
83,
90,
91]. However, unlike MS, a strong epidemiological link exists between leukoaraiosis and cerebrovascular disease [
92‐
94]. Arterial hypertension and cardiac disease are also risk factors that are frequently associated with leukoaraiosis [
88], and these are thought to induce arteriolosclerotic changes in the arteries and arterioles of the WM, replacing the smooth muscle cells by fibro-hyaline material, causing thickening of the vessel walls and narrowing of the vascular lumen [
95]. Indeed, arteriolosclerosis is often present within areas of leukoaraiosis [
96,
97]. Furthermore, the arterioles supplying the deep WM, which are some of the longest in the brain, frequently become tortuous with aging [
23,
98‐
100], with the result that there is a trend towards increased tortuosity in individuals with leukoaraiosis [
23]. This tortuosity usually begins abruptly as the arteriole passes from the cortex into the WM [
23], and greatly increases the vessel length. Given that this will increase the hydraulic resistance of the arterioles [
99], it will tend to inhibit the blood flow to the deep WM. It is therefore perhaps not surprising that the periventricular veins, being a ‘distal irrigation field’ [
88], are prone to ischemic damage under conditions of moderate deficit in blood flow.
Further evidence linking leukoaraiosis with altered venous hemodynamics comes from a series of studies by Chung and co-workers [
15,
16,
101], who investigated jugular venous reflux (JVR) (that is, retrograde flow in the IJVs) in older individuals. They found JVR to be a phenomenon that increased with age, and concluded that it was associated with more severe age-related WM changes (leukoaraiosis) [
16]. In particular, they found that the IJV lumen cross-sectional area increased with age [
101], which suggests dilation of the veins due to increased venous pressure and reduced flow velocity. Chung
et al. [
101] suggested that if the venous hypertension exceeds the ability of the dilation to compensate for the additional pressure, then it would compromise the competence of the jugular venous valves, with the result that the direction of venous flow could be reversed. They further hypothesized that this ‘chronic or long-term episodic elevated cerebral-venous pressure might cause cerebral venule hypertension, resulting in… reduce[d] CBF since elevated cerebral venule pressure would lower cerebral perfusion pressure’ [
15].
In a series of studies, Bateman and co-workers investigated altered venous hemodynamics in a variety of neurological conditions [
2,
3,
5,
17,
102,
103]. In particular, they investigated pulsatile blood flow in leukoaraiosis [
3] and vascular dementia [
5]. In both conditions, they found venous pulsatility to be greatly increased in the straight sinus compared with healthy controls, implying that in individuals with leukoaraiosis and vascular dementia, the blood flow through the WM is highly pulsatile. Given that blood flow through the cerebral vascular bed is generally non-pulsatile in healthy young adults [
3,
104], Bateman’s findings imply marked changes in hemodynamic behavior in individuals with leukoaraiosis and vascular dementia, something that will induce transient shear stresses on the endothelia. Given that vessels experiencing highly oscillatory flows also seem to be at high risk of developing arteriosclerosis [
105], it is perhaps not surprising that leukoaraiosis is associated with morphological changes in the WM vasculature [
13,
14,
22,
23]. Bateman hypothesized that the increased pulsatility exhibited by the CBF was a direct consequence of a dysfunctional windkessel mechanism [
3,
5], implying profound alterations in the dynamics of the CSF system. Indeed, Bateman calculated the CSF pulse volume in severe cases of leukoaraiosis to be 46% greater than that in healthy controls [
3]. Furthermore, he found that the CSF dynamics associated with leukoaraiosis delayed the egress of blood from the cortical veins into the superior sagittal sinus (SSS), inducing a complex pulse wave, which propagated backwards towards the capillaries of the cortex [
3].
Normal-pressure hydrocephalus
NPH occurs when there is an abnormal accumulation of CSF in the ventricles, causing them to become enlarged [
27], but with little or no increase in intracranial pressure (ICP). Most adults with the condition experience an ICP that is not unusually high, being generally less than 15 mmHg [
106,
107]. NPH is characterized by gait disturbance, urinary incontinence, and dementia [
108]. Although its pathophysiology is poorly understood, NPH has traditionally been thought to be a form of communicating hydrocephalus, characterized by poor absorption of CSF into the SSS due to defective AV [
109]. However, evidence supporting this opinion is lacking [
109], and several commentators have suggested alternative theories [
2,
102,
110‐
112]. In particular, there is growing evidence that reduced intracranial compliance [
2,
102,
113,
114], induced by venous hypertension, might be involved in the pathophysiology of NPH [
2,
102,
115,
116], although this opinion is disputed by others [
26,
117,
118]. Bateman [
102] found the arteriovenous delay (AVD), a general marker of intracranial compliance, to be 53% lower in patients with NPH compared with healthy controls. A similar reduction in AVD in patients with NPH was reported in a subsequent study [
2]. Mase
et al. [
114] independently confirmed this finding, showing a 64% reduction in intracranial compliance in patients with NPH compared with controls. The fact that an AVD exists at all indicates the presence of compressible material within the intracranial space, which is deformed when the systolic arterial pulse enters the cranium. With respect to this, the cerebral veins are a likely candidate [
115,
116]. Approximately 70% of intracranial blood volume is located within the venous compartment, much of it in thin-walled veins that readily collapse under small changes in transmural pressure. Given that the intracranial veins, particularly those of the superficial venous system, are much more compliant than the arterial vessels, it has been suggested that the change in intracranial compliance seen in patients with NPH may be associated with venous hypertension [
2]. In patients with NPH, cortical-vein compliance is significantly reduced [
102]; however, following shunt surgery, compliance greatly increases, suggesting that the compliance changes associated with these veins are functional and not structural [
2,
102]. NPH has been shown to be associated with venous hypertension in the SSS [
119]. It is therefore plausible that hypertension in the SSS might increase the pressure in the cortical veins, with the result that the functional compliance of these vessels is reduced [
2]. Furthermore, venous hypertension in the SSS would tend to reduce the compliance of the AV, and this, together with reduced cortical-vein compliance, might account for the reduction in AVD seen in individuals with NHP.
CBF has been found to be lower in patients with NPH than in normal controls [
120‐
123]. This is generally thought to be associated with the formation of ischemic lesions, particularly in the deep WM [
118,
122,
124], implying that regional differences in CBF might exist in individuals with NPH. Momjian
et al. [
122] found the distribution of regional CBF in the WM to be different in patients with NPH compared with healthy controls, with a more pronounced CBF reduction adjacent to the lateral ventricles, and a logarithmic normalization occurring with distance from the ventricles. These findings built on an earlier study by Owler
et al. [
121], who reported NPH to be associated with a marked reduction in mean CBF in the deep GM. Momjian
et al. [
122] attributed these phenomena to a combination of factors, including cerebral small-vessel disease; tissue distortion, and reversal of CSF and interstitial fluid flow, resulting in reduced cerebral perfusion pressure (CPP) near the ventricles and resultant ischemia. However, this interpretation was challenged by Bateman [
102], who found blood flow in the straight sinus, which serves the periventricular region, to be unchanged in patients with NPH compared with controls. Having said this, Bateman also reported 27% less drainage from the SSS in patients with NHP compared with healthy controls. Although Bateman’s findings concerning the blood flow through the deep venous system are difficult to explain, those relating to the superficial venous system, might help to explain the formation of cortical infarcts in patients with NPH reported by Tullberg [
124].
A number of researchers have reported marked alterations in CSF dynamics in NPH, with CSF pulsatility in the AoS found to be markedly greater in patients with NPH compared with controls [
112,
125‐
129]. This mirrors the findings of Magnano
et al. [
77], who found a similar phenomenon in patients with MS. By contrast, the cervical CSF pulse was either unchanged [
112] or actually reduced in individuals with NPH compared with controls [
126]. Although the reasons for this apparent paradox are difficult to explain, it suggests that biomechanical changes occur with NPH, which alter both intracranial compliance and pulsatility of the cerebral venous and arterial blood flows. NPH also appears to be associated with significantly reduced CSF resorption into the SSS through the AV [
26,
130], which is a finding consistent with venous hypertension in the dural sinuses. Drainage of CSF into the dural venous sinuses requires a pressure gradient between the sub-arachnoid space (SAS) and the SSS of about 5 to 7 mmHg [
131,
132]. If the pressure in the SSS is increased, then either the ICP must also increase to facilitate CSF absorption through the AV [
117], or alternatively the CSF must be absorbed elsewhere in the intracranial space. Given that ICP does not substantially increase in individuals with NPH, this indicates that CSF is being resorbed elsewhere [
124]. Bateman [
102] suggested that CSF resorption is likely to occur in the sub-ependymal brain parenchyma. Ventricular reflux of fluid has been shown to be a characteristic of communicating hydrocephalus [
133,
134], with the periventricular tissue characterized by disruption of the ependyma, and by edema, neuronal degeneration, and ischemia [
124]. Although the hydrodynamics associated with ventricular reflux are poorly understood, it may be that reduced CSF absorption by the AV in individuals with NPH at least partly explains the increase in aqueductal CSF pulsatility that is associated with the condition [
133].