Venous hemodynamics in neurological disorders: an analytical review with hydrodynamic analysis

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].

Although there are clear differences in the pathologies of MS, leukoaraiosis and NPH, there are also striking similarities. All three are characterized by: 1) WM changes in the periventricular region; and 2) reduced CBF. The lesions associated with both MS and leukoaraiosis tend to be perivenous in nature, and the changes in CSF dynamics associated with NPH and MS also reveal similarities. This raises intriguing questions as to why these similarities exist. Are there some underlying physical mechanisms that are common to all these conditions?

The proximity of immune-cell aggregations to the vasculature is a hallmark of MS [135]. Whereas much attention has been paid to the biological mechanisms involved in the formation of MS plaques, the implications associated with their spatial arrangement have largely been ignored. Why do MS plaques form next to veins rather than capillaries, and why do they consistently occur in some places and not others? In nature, when a process is truly random, events tend to be widely distributed, with no underlying pattern. Conversely, if there is an underlying phenomenon, then the events will tend to cluster in both time and space. The fact that MS plaques consistently form around the periventricular [136] and cortical [50] veins indicates that the latter, rather than the former process must be taking place. If this were not so, then lesions would be randomly distributed throughout the brain parenchyma. From this, it can be concluded that in MS some unknown, but consistent, phenomenon is at work, which causes plaques to form around certain cerebral veins. Although historically considered a disease primarily affecting the WM, it is now known that cortical demyelination is common in MS and more extensive than previously appreciated [78, 137]. Although it has been reported that many intra-cortical and sub-cortical lesions are perivenous in nature [50, 52, 55‐57], it is not known why this is so. However, the fact that MS plaques form in the vicinity of veins in both the deep and superficial systems suggests that the pathophysiological mechanisms at work are extensive and not confined to a focal region.

Another universal principle found in nature is that of mass transfer. In simple terms, in order for matter to move from one place to another, it must be transported by some mechanism. In biology, the transport of cells and chemicals generally occurs either by: diffusion, by active transport (in the case of ion transport across the cell membrane), or through transport in a bulk fluid such as blood. If diffusion or active transport are the mechanisms at work, then there is a tendency towards higher concentrations of the transported substance near its source and lower concentrations further away. If this simple logic is applied to the formation of perivenous MS lesions, it would suggest that the plaque formation emanates from the blood vessels, rather than the other way round. Indeed, the current thinking appears to support this, suggesting that in MS, plaque formation is precipitated by breaching of the blood–brain barrier (BBB) [51, 138, 139]. If the diffusion principle is applied to the observation by Momjian et al. [122] that in patients with NPH the CBF steadily increases the further away from the lateral ventricles, then it suggests that the unknown factor inhibiting blood flow is emanating from the lateral ventricles, which suggests that ventricular fluid reflux might be involved.

The mass transport associated with bulk fluids also appears to offer insights into the spatial arrangement of ischemic WM changes, such as those found in leukoaraiosis. Considering oxygen transport in the blood through the cerebral vascular bed, the law of mass transport dictates that as oxygen is supplied to the brain parenchyma, so the oxygen levels in the blood will decrease. Consequently, the oxygen tension in the cerebral arteries will be higher than that in the cerebral veins. Under normal circumstances, this should not cause any problems, but when CBF is greatly impaired, as in both leukoaraiosis [23, 83, 90, 91] and MS [18‐21], then this might create pathogenic conditions in the distal veins. If the oxygen consumption of the endothelia and brain parenchyma surrounding the arterioles and capillary bed is not downregulated, then the oxygen tension in the veins might become so low that ischemic damage could occur in these vessels. With respect to this, the periventricular WM, being at the distal end of the circuit [88], appears to be particularly vulnerable to ischemic damage when blood flow is reduced. If ischemic damage due to hypoperfusion is involved in the formation of MS lesions, as some have suggested [60, 61], then this might explain why plaques tend to form around the veins, rather than the capillaries and arterioles.

There is increasing evidence that hypoxia-like metabolic injury may be a pathogenic component in the formation of MS lesions [62, 86]. Wakefield et al. [140] found morphological changes in the venous endothelia, which progressed to occlusive vascular inflammation. They proposed that these changes were the precursor to lesion formation, and suggested that demyelination in MS may have an ischemic basis. Aboul-Enein and Lassmann [141] reported similarities between the tissue injury found in inflammatory brain lesions and that found in hypoxic conditions of the CNS. Ge et al. [142] identified subtle venous wall signal changes in small MS lesions, which they interpreted as early-stage vascular changes. These changes may be the result of early ischemic injury, marking the beginning of trans-endothelial migration of vascular inflammatory cells, before any apparent BBB breakdown. Further evidence that focal inflammatory BBB leakage may not be the initiating event in MS plaque formation comes from Werring et al. [143], who measured random motion of water molecules (apparent diffusion coefficient; ADC) in the NAWM of patients with MS. They found that the formation of lesions was preceded by subtle progressive alterations in tissue integrity. Similarly, Wuerfel et al. [144] found that changes in perfusion parameters (CBF, CBV and MTT) were detectable not only prior to BBB breakdown, but also prior to increases in the ADC. They concluded that in MS, inflammation is accompanied by altered local perfusion, which can be detected prior to permeability of the BBB. Commenting on this, D’haeseleer et al. [60] concluded that ‘focal ischaemia might play a part in the development of a subcategory of focal MS lesions’. Lochhead et al. [145], using a rat model, showed that hypoxia followed by re-oxygenation altered the conformation of the occlusion in the tight junctions between the endothelial cells, resulting in increased BBB permeability. In doing so, they confirmed the findings of earlier studies undertaken by the same team [146, 147]. The earliest detectable event in the development of WM lesions is thought to be an increase in the permeability of the BBB [51], followed by inflammation and demyelination. Others have implicated tight-junction abnormalities in increased BBB permeability and lesion formation in MS [139, 148, 149].

Several researchers have found similarities between leukoaraiosis and MS [60, 61]. Leukoaraiosis is characterized by periventricular hyperintensities [80] and reduced CBF in the WM [150]. Its clinical symptoms include cognitive features that are similar to those associated with MS [151]. Graumann et al. [152], investigating gene expression in the NAWM of patients with secondary progressive MS and healthy controls, showed that the patients with MS exhibited consistent differences in the expression of hypoxia-inducible factor (HIF)-1a compared with controls. Similar upregulation of HIF-1a in cerebral WM was found by Fernando et al. [153] to be associated with leukoaraiosis, which they attributed to WM hypoperfusion. Leukoaraiosis is associated with significantly decreased CBF in the deep WM [150], and it is thought that ischemia, resulting from poor perfusion, is a major contributing factor [12, 13, 22]. The condition is characterized by non-inflammatory PVC, resulting in thickening of the vessel walls and narrowing of the deep cerebral veins [13, 22], which will inevitably increase the hydraulic resistance of these pathways. Although lumenal narrowing of the periventricular veins has not been reported in patients with MS, Putnam and Adler [6] reported that the periventricular MS plaques resulted in gross distension of the medullary veins upstream of the lesions, suggesting that venous stenosis is occurring. This would inevitably increase the hydraulic resistance of these vessels and promote hypoperfusion. Given that the perfusion pressure, which promotes blood flow, is relatively low in the periventricular veins, the WM in this region is particularly sensitive to fluctuations in total CBF [154]. Any increase in the hydraulic resistance of the periventricular veins might cause shunting of blood away from these vessels [22, 155], an action that would also tend to promote ischemia.

So why should some regions of the brain be more vulnerable than others to damage? Perhaps the architecture of the cerebral-venous system provides some clues? While the distal venous regions may be prone to hypoxic stress, the spatial arrangement of the veins may also contribute to their vulnerability. Evidence in support this opinion comes from Schlesinger [155], who forced hot carmine–gelatin solution, under high pressure, into the vein of Galen in human cadaver brains. The extravasations that were produced, chiefly in the region of the angle of the lateral ventricle, ‘closely resembled the distribution and shape of plaques in advanced cases of MS’. From this, Schlesinger concluded that: ‘it seems possible that the plaques may only be found in this area of the ventricular wall because they have a definite topographical relationship to the veins which are crowded together in the region of the lateral ventricular angle.’ Although the physiological implications of Schlesinger’s experiment are debatable, his results are clear and unambiguous from a fluid-mechanics point of view. When the gelatin solution was forced into the deep venous system, it divided and flowed up the two internal cerebral veins, so that both hemispheres of the brain were affected. Furthermore, the fluid flowed relatively easily until it came to the junction between the medullary and sub-ependymal veins, where the resistance was so great that the pressure built up to such an extent that the fluid burst through the vessel wall. The experiment therefore indicates two things: first, that the junction between the medullary and sub-ependymal veins has a much higher resistance to fluid flow than the downstream veins; and second, that the vessel walls at this junction are susceptible to rupture if the pressure gets too high.

The finding that the junction between the medullary and sub-ependymal veins has a high resistance to fluid flow is no surprise. The sub-ependymal veins are collecting vessels, which receive venous blood from a large number of the smaller medullary veins that enter the sub-ependymal veins at approximately 90 degrees. From a fluid-mechanics point of view, this is not a very streamlined configuration, and will result in relatively large pressure drop across this junction. Any stenosis at this junction would therefore greatly increase its resistance, possibly leading to distension of the upstream medullary veins, as Putnam and Adler reported [6]. Consequently, the periventricular veins share characteristics normally associated with developmental venous anomalies (DVAs). DVAs are a venous confluence in which a single collecting vessel drains an abnormally large venous territory, resulting in a relative volume overload. This anatomic configuration, as San Millán Ruíz et al. [156] pointed out, is similar to that encountered in the periventricular region. In addition, DVAs have been shown to have thickened walls [156], similar to those associated with PVC [13, 22], with stenosis of the collecting vein reported in 13.1% of patients [156]. Stenosis of this kind invariably increases the hydraulic resistance of the vein, so that the upstream pressure is greatly increased, as was shown by Dillon [157], who measured a 15 mmHg pressure gradient across a stenosis of the collecting vein of a DVA in one patient.

Unlike the deep venous system, the superficial system has thin-walled cortical bridging veins that traverse the SAS. Blood flow through these compliant vessels is controlled by sphincters, which regulate discharge into the SSS [158, 159]. This means that these vessels possess characteristics similar to those of a Starling resistor [160‐163], and these collapse, occluding the blood flow, when the transmural pressure reaches a certain threshold [164]. The cortical bridging veins are very sensitive to small changes in transmural pressure. Indeed, because they are required to ‘open’ and ‘close’ to regulate blood flow from the cortex, the cortical venous pressure is only about 2 to 5 mmHg higher than the ICP [164]. This means that small changes in ICP or venous pressure can have a substantial effect on the behavior of blood flow from the cortex. Indeed, it has been estimated that a change of as little as 1.5 mmHg in the difference between ICP and the pressure in the bridging veins could be responsible for the difference between severe hyperemia (CBF = 1000 ml/min) to serve ischemia (CBF = 300 ml/min) [164]. Given that MS may be associated with venous hypertension in the dural sinuses of greater than 2 mmHg [165], it can be hypothesized that this could have a profound effect on blood flow in the cortex. Although it is difficult to say how this might influence hemodynamic behavior in the cortex, it is notable that Kidd et al. [50] found GM lesions in patients with MS to be exclusively located adjacent to cortical veins. Drawing an analogy with WM lesions, they stated; ‘We have been able to show that there is a clear relationship between the site and characteristics of cortical lesions and the five different types of cortical vein, just as Dawson’s ‘fingers’ arise adjacent to veins in periventricular WM’ [50].

MS, leukoaraiosis, and NPH all appear, to a greater or lesser extent, to be associated with marked changes in the dynamics of the intracranial CSF system. This suggests that these diseases might be associated with alterations in the intracranial hydrodynamic regulatory system, which controls the volume and pulsatility of the blood in the cerebral vascular bed [3, 166, 167]. The interactions between the CBF and CSF are illustrated in Figure 1, which shows an idealized model of the principal intracranial fluid pathways. From this, it can be seen that there is a bulk flow of CSF from the choroid plexus to the SSS, via the AV, driven by the pressure gradient between the two. There is also a complex windkessel mechanism that ensures Monro-Kellie homeostasis, which compensates for transient increases in CBV by pushing CSF out of the cranium [168]. This sophisticated windkessel mechanism uses the CSF to dampen the arterial pulse and to ensure, in healthy young adults, the smooth flow of blood through the capillary bed [5]. The energy from the arterial pulse is transferred to the CSF, which pulses backwards and forwards across the foramen magnum. The blood flow through the cerebral capillary bed is normally smooth and free from a pulse, but by the time it reaches the venous sinuses, it once again exhibits pulsatile characteristics [3, 169]. This suggests that energy transferred from the arterial pulse to the CSF is in turn transferred back to the venous-discharge flow.

The various pulses associated with the intracranial hydrodynamic system are illustrated in Figure 2, which shows the transient flow rates of the arterial, venous, and CSF flows in and out of the cranium in a healthy individual [169]. From this, it can be seen that the system is driven by the arterial pulse, which, as it enters the cranium, greatly increases the volume of blood in the pial arteries during systole [1]. Peak CSF flow through the foramen magnum occurs shortly after the arterial peak, which is in turn followed in late systole by a peak in the venous blood flow leaving the cranium. This movement reflects the transfer of kinetic energy from the arterial to the venous pulse via the CSF system. By comparison, the aqueductal CSF pulse, which is not well understood, is much smaller than, and out of phase with, the cervical CSF pulse. In theory, according to the Monro-Kellie doctrine, at any point in time, the volume of CSF leaving the cranium should be equal to the volume difference between the arterial blood entering the cranium and the venous blood leaving it [170]. In reality, however, the Monro-Kellie doctrine is only approximately true [170], and compliance within the intracranial space, together with inertial forces associated with the CSF fluid column, ensure that small transient imbalances occur between the fluid volumes entering and leaving the cranium.

Close inspection of Figure 2 reveals an interesting and poorly understood phenomenon. During diastole, when the excess arterial blood stored by the windkessel mechanism is discharged back into the cerebral vascular bed, the venous blood-flow rate leaving the cranium actually decreases. This implies that during this period, the volume of venous blood in the cranium must be steadily increasing, only to be ejected during systole. Although the physiology associated with this mechanism is not well understood, it seems likely that much of this blood is stored during diastole in the cortical bridging veins that transverse the SAS. Discharge from these veins is controlled by regulatory sphincters [158, 159]. Constriction of these sphincters results in an increase in the transmural pressure of the bridging veins, causing them to engorge and ‘puff out’, before periodically discharging into the SSS. Thus, these sphincters, in effect, create Starling resistors, whose characteristics are wholly governed by the respective venous and CSF pressures. Under normal circumstances, this Starling resistor interacts with the CSF pulse in such a way as to ensure the correct flow of blood into the SSS. However, if the venous pressure in the SSS should increase, due to partial occlusion of the cerebral-drainage pathways, then this would change the functional compliance of both the AV and the cortical bridging veins [2, 102, 115, 116]. Under normal circumstances, the compliant AV dampen the systolic CSF pressure wave as the wave travels along the SAS, thus reducing the pressure that it exerts on the cortical bridging veins. Venous hypertension in the dural sinuses is therefore likely to reduce the time duration between arterial initiation of the CSF pulse and impingement of that pulse on the cortical bridging veins. Furthermore, it is likely to increase the intensity of the CSF pressure wave impinging on the bridging veins, so that the venous blood is expelled more rapidly into the SSS. Consequently, a reduction in the time duration between the arterial and venous peaks would be expected, which is exactly what Bateman found in patients with NPH [2, 102]. However, the fluid mechanics associated with this mechanism are complex and it is difficult to predict how the system would behave under different circumstances. For example, if the hydraulic resistance of the cerebral-venous drainage system is greatly increased, as Beggs et al. [64] reported in patients with MS, then rapid discharging of the contents of the cortical veins might lead to a transient increase in pressure in the SSS. How this would behave in practice is a matter of conjecture. It might result in a complex reverse-pressure wave traveling down the cortical veins, as Bateman observed in an individual with leukoaraiosis [3], or alternatively, it could result in venous reflux traveling up the straight sinus, as has been suggested by Schelling [46].

Although the precise behavior of the intracranial hydrodynamic system under conditions of venous hypertension is unknown, there is evidence that occlusion of the venous-drainage pathways causes blood to accumulate within the cranium. In an experiment involving healthy subjects, Kitano et al. [171] showed that compression of the IJVs could result in an increase in intracranial blood volume of 5 to 20%. Frydrychowski et al. [1] also performed bilateral compression of the IJVs on healthy individuals, and found that it caused a reduction in the width of the SAS, a finding consistent with increased CBV. Furthermore, these authors found that during compression of the IJVs, the pulsatility of the pial arteries traversing the SAS increased by 107%. This suggests that occlusion of the venous-drainage pathways reduces compliance of the intracranial space, impairing the windkessel mechanism, with the result that the blood entering the cerebral microvasculature becomes more pulsatile. Frydrychowski et al. [1] concluded that this mechanism potentially linked jugular-outflow insufficiency with arterial small-vessel cerebral disease. Their work seems to corroborate that of Bateman [5], who found vascular dementia to be associated with greatly increased pulsatility in both the basal venous and straight sinus flows.

In 2009, Zamboni et al. published a paper [7] linking a vascular syndrome, CCSVI, with MS. CCSVI is characterized by restricted venous outflow from the brain [8, 64] due to occlusions, which can take several forms, including the presence of intra-luminal septa, membranes, and immobile valves, as well as segmentary hypoplasia of the veins [172]. CCSVI has proven to be a highly contentious issue [173, 174], with a number researchers doubting its validity as a physiological phenomenon [173, 175‐179]. Notwithstanding this, biomechanically, CCSVI will tend to increase the venous pressure in the dural sinuses, which hypothetically could alter the dynamics of the intracranial CSF system and potentially influence CBF.

The results obtained by researchers for CCSVI have been very mixed. For example, some researchers found CCSVI-like venous anomalies to be strongly associated with MS [7, 63, 180‐186], whereas others found no significant difference between the venous characteristics of healthy controls and patients with MS [173, 176, 187‐190]. Furthermore, CCSVI-like abnormalities are not exclusive to MS, and have been found in lesser numbers in healthy controls [74, 180, 191] and in those with other neurological diseases [180]. However, a recent meta-analysis of the published evidence showed a positive association between CCSVI and MS, although poor reporting and marked heterogeneity between studies precluded any definitive conclusions [192].

One possible explanation for the discrepancies between studies is the echo color Doppler sonography (ECDS) frequently used to diagnose CCSVI. The floppiness of the vessels involved and the variability of the venous vasculature can lead to erroneous results if ECDS is not undertaken correctly [193‐195]. In an attempt to avoid these difficulties, Zamboni et al. [8] developed a non-invasive strain-gauge cervical plethysmography technique for characterizing cerebral-venous drainage in patients with MS. In a blinded study involving healthy controls and patients with MS diagnosed with CCSVI, it was found that the hydraulic resistance of the extracranial venous system was on average 63.5% greater in those diagnosed with CCSVI compared with controls (P < 0.001) [64]. This corroborates the work of Monti et al. [196], who found reduced cerebral-venous outflow in the upright position to be strongly associated (P < 0.0001) with MS.