The terms pressure and volume may often be used interchangeably when describing CNS pressure, because pressure depends upon volume 5. The hypothesis depends upon the relationship between pressure and volume, which includes the phenomenon of compliance. Compliance enables a volume increase to occur in the intrathecal CNS space without causing a pressure increase 5, and occurs when a corresponding amount of venous blood is displaced. Compliance affects the ability of the CNS to accommodate volume fluctuations that occur with movement and so avoid ischemia.

Blood from the CNS is drained by a venous plexus that is extensive, anastamotic and valveless. It allows blood to flow in a retrograde direction, away from the heart and into the CNS with postural movements 67. CSF pressure fluctuations with movement have been observed directly 8 and small movements such as occur during speech, may cause detectable CNS pressure fluctuations 5. Veins situated between the dura and surrounding bone are highly compressible and capable of large volume fluctuations 7. Such veins are prominent in the spine, and are susceptible to volume fluctuation with alterations in body cavity pressure 7. As CNS venous volume increases with retrograde flow veins may be distended due to 'back pressure' or may be compressed due to elevation of overall pressure, resistance to venous outflow then increases, outflow is reduced, and pressure will tend to rise. Pressure increase in response to increasing intrathecal volume corresponds to a phase of reduced compliance 5. If volume increases further, a pressure may be reached which compromises arterial flow, resulting in ischemia of neural tissue 5. The relationship between overall CNS pressure and added volume is shown in Fig. 1. The pressure volume index (PVI) is the volume that when added to the CSF space results in a ten-fold increase in pressure. The normal adult range is 13–26 ml 9. Adults will have a larger PVI than infants because of a larger central nervous system with a correspondingly greater venous volume.

CSF flow contributes to compliance by regulating the transluminal pressure of veins and therefore their patency and flow. Poiseuille's law states that laminar flow in a tube is inversely proportional to the fourth power of the radius. Hence a decrease in vessel diameter will cause a four-fold increase in resistance. If any pressure increase associated with increase in CNS volume is distributed throughout the intrathecal space, there will be a small average reduction in the diameter of many veins. If the whole CNS venous system is affected by a CNS volume increase, the overall reduction in venous outflow will be minimised, thus free flow of CSF facilitates venous drainage. Veins inside the head do not collapse when moving to an upright position because the movement of CSF with gravitational force allows both fluids to experience the same pressure gradient 10, maintaining the patency of veins. Obstruction to CSF flow at the foramen magnum is an important cause of reduced compliance. Movement of the body appears to contribute to foramen magnum obstruction, so that CSF flow may be affected by posture 8. The head and spine may be transiently separated into compartments with reduced pressure volume indices. Evidence of obstruction may be present without an obvious hindbrain hernia 1112 or with cerebellar deformity. Pressure gradients between the CSF spaces of the head and spine are normally those due to gravity 13. In the presence of an obstruction associated with spina bifida or syringomyelia, pressure may fluctuate independently in the head and spine 81415. This phenomenon has been termed 'craniospinal pressure dissociation'. With normal CSF pathways at the foramen magnum, retrograde flow of venous blood into the CNS may be compensated by compression of distant veins. If a volume increase occurs in either the head or the spine in the presence of foramen magnum obstruction, it will, according to this hypothesis cause a pressure response which will be greater than if CSF pathways were unobstructed. It is proposed therefore that with Chiari malformation retrograde venous flow causes an enhanced pressure response, as illustrated in Fig. 2.

In a relaxed subject with normal compliance, small volume fluctuations caused by blood flow will not be detected by overall pressure measurement. As compliance is reduced, minor volume fluctuations may result in measurable pulsations and physical movement will cause the greatest pressure fluctuations and peaks. The elevation of CSF pressure in normal pressure hydrocephalus 16 and in hydrocephalic children observed during continuous pressure monitoring 17 supports the hypothesis. Reduction of pulsatility of CSF pressure following cranial expansion as a treatment for hydrocephalus 18 is also consistent with the proposed theory.

It is proposed that when compliance is reduced:

• Venous volume increase with postural movements will cause delay of venous and arterial flow in the parenchyma.

• At peaks of pressure, vessels in the parenchyma of the brain or spinal cord may be sufficiently compressed to result in ischemia.

Continuous unidirectional flow in venous plexus vessels is not essential, venous blood may be diverted via anastamoses to lower resistance vessels whereas continuous flow in parenchymal vessels is necessary to provide oxygen. Full compliance benefits flow in all vessels 519. The autonomic nervous system, by controlling vessel diameter and flow, is known to influence extracellular fluid volume in many tissues, with filtration or absorption taking place in venules depending on local circumstances 10. Autonomic regulation of flow in CNS vessels is likely to influence the rate of filtration into the parenchyma. If, however, autoregulation is unable to compensate for the effect of venous pressure fluctuations, extracellular fluid will accumulate causing reduction in compliance. Consequently, venous outflow will be further compromised and hydrocephalus will progress. Ischemia is likely to be a part of the pathological process with a hyperemic response enhancing filtration of fluid into the parenchyma. Volume increases that do not cause ischemia are accommodated, but progressive volume increase will cause ischemia at peaks of CNS pressure.

The observation that removing CSF from patients with symptomatic normal pressure hydrocephalus improves both arterial supply and venous drainage 19 supports the hypothesis proposed here. It is suggested that artificial increase in cranial pressure by balloon insufflation inside a lateral ventricle mimics the mechanism for progression of hydrocephalus. In this model, intracranial pressure is intermittently raised 20 in a manner that is comparable to that which will occur with movement in the presence of any cause of loss of compliance. Signs of a hyperemic response may have been observed during direct measurement of CNS pressure following pressure increase with movement in Chiari I 21. Experimentally induced foramen magnum blockage has lead to subjective observations of venous insufficiency in the cord 22, and the effect of pressure on venous flow, causing localised oedema that resolves with improved perfusion of veins may be seen during surgery 23.

Chiari I and II malformations are characterised by degrees of hindbrain herniation. Adults with Chiari I malformation have a reduced posterior fossa size, relative to head size 12242526, providing evidence for posterior fossa restriction as a cause of cerebellar herniation. Cerebellar herniation is a feature of posterior fossa restriction with known causes, for example craniosynostosis 27. The familial tendency for Chiari I indicates that there are genetic determinants for posterior fossa growth 2829. Chiari II is characterised by a hypoplastic posterior fossa with compression of hindbrain structures 30. Associated bone abnormalities in the face, provide additional evidence of a primary maldevelopment of bone as part of the malformation 30. Facial and posterior fossa bones have common embryological origins and their responsiveness to fibroblast growth factor suggests that they have related growth mechanisms 31. The greater incidence of Chiari I and II in females supports the view that posterior fossa size is genetically determined, with males having larger posterior fossa CSF spaces than females, so that restriction of posterior fossa size leads to hindbrain herniation more readily in females than males.

Spinal CSF and venous blood provide buoyancy to the central nervous system 9, maintaining the position of the hindbrain. The phenomenon of coning in response to lumbar puncture illustrates how hindbrain herniation can occur in response to loss of spinal fluid volume. The use of lumboperitoneal shunts is associated with cerebellar herniation 32, and is likely to be related to loss of CSF from the spinal compartment. Reduction of posterior fossa CSF space may also cause an abnormal cranio-spinal pressure gradient by retaining CSF in the head, which would otherwise be free to move through the foramen magnum. Direct measurement of pressure in the presence of Chiari I support the hypothesis that low CSF pressure may cause the cerebellar tonsils to be drawn into the spine 33. It is suggested here that this may occur when venous blood leaves the spinal venous plexus so that excessively free drainage from the spinal plexus may also contribute to hindbrain herniation. In these instances, hindbrain symptoms may predominate over those caused by raised spinal pressure. In animal studies the hindbrain descends following creation of an artificial spina bifida lesion 34, and may elevate following intrauterine repair 34. This allows for the common assertion that low spinal pressure contributes to the hindbrain herniation of Chiari II.

That the degree of hindbrain herniation does not correlate with spinal injury in syringomyelia 35 and spina bifida, and posterior fossa size does not correlate with the degree of cerebellar herniation in Chiari I 111236, indicates that factors other than posterior fossa hypoplasia are influencing the degree of neural injury. Herniation will relate to the degree to which skeletal defects lead to pressure changes and the extent to which low spinal pressure causes hindbrain herniation in any individual case. Both herniation and neural injury will depend upon an interaction between the anatomy of the posterior fossa and physiological influences on CSF pressure and flow. Although low pressure may contribute to the cause of hindbrain herniation, once herniation becomes more established CNS pressure will tend to vary more widely as the head and spine become separate compartments. Pressure in the head may tend to rise 2 due to a larger volume of neural tissue in the head than the spine, with a greater arterial supply, as illustrated by the caudal flow of CSF across the foramen magnum in systole and the rostral flow in diastole 37. The potential range in pressure is arguably greatest in the spine because of the capacity and compressibility of the intraspinal venous plexus vessels. This theory asserts that the highest spinal pressures lead to cord ischemia and the lowest pressures contribute to maintaining the hindbrain herniation. Loss of neural tissue by ischemic atrophy will tend to benefit remaining cord or brain, whereas growth of neural tissue may perpetuate reduced compliance. A syrinx cavity forms a space-occupying lesion and so reduces compliance. The effect of a space-occupying lesion on spinal pressure will be as in Fig. 3. Benign lesions may be accommodated without pressure increase if neural or CSF volumes fall. The combination of pressure effects illustrated in Figs. 2 and 3 will be a left shift in the pressure volume curve in hydrocephalus.

It is proposed that hydrocephalus is an oedema of the central nervous system. Rapidly developing venous insufficiency leads to increase in parenchyma water content and slower processes lead to increase in water that is distributed between the parenchyma and larger extracellular spaces. Symptoms, signs and morphology of hydrocephalus will depend upon the pathology that causes the pressure increase, the rate of fluid accumulation, and the developmental stage at which hydrocephalus develops. Compliance and therefore hydrocephalus, is influenced by:

• The internal dimensions of CNS bones

• Fontanelles and surgically created bone defects

• The presence of any space occupying lesion

• CSF volume

• Free flow of CSF around the CNS

• The integrity of CNS veins and central venous pressure

• Growth and atrophy of neural tissue

• Autonomic regulation of blood pressure and flow

Fluid flows down pressure gradients from areas of higher to lower pressure 1. The passage of fluid into the ventricles and subarachnoid spaces represents flow to an area of lower mean pressure than the parenchyma. Maximal CSF pressures are achieved during physical movement and arterial pressures will tend to exceed this. Movement of tracers from the subarachnoid space to the parenchyma has been demonstrated to occur rapidly 38 and might appear to contradict this theory. If movement of fluid occurs along perivascular channels, such movement might depend upon the amplitude of the arterial pulse and the energy that it imparts 39. According to the proposed hypothesis rapid accumulation of fluid would tend to remain in the parenchyma. Slow accumulation that occurs with intermittent periods of relaxation, when CSF pressure falls, would encourage fluid flow into CSF spaces.

Ventricle size is not a good indicator of intracranial pressure. Lateral ventricle size is determined by ventricle wall tension, brain turgor and the ability of CSF to flow through the aqueduct and fourth ventricle. Small ventricles may be found with raised, normal, or low intracranial pressure if CSF can exit the head 40. Brain oedema may lead to small ventricles if CSF can pass into the spine. Fluid accumulation in the parenchyma will increase brain turgor and cause it to resist ventricle enlargement 41 whereas tension in the ventricle walls will exert a force on brain tissue, favouring enlargement of the ventricles. Wall tension is proportional to the internal radius of the cavity 10, as well as fluid pressure. Pressure waves generated by venous volume fluctuation in the spine may influence lateral ventricle wall tension without the requirement for flow 42, so physical movement may maintain or increase ventricle size when posterior fossa CSF pathways are patent. In chronic cases of raised intracranial pressure there will be atrophy related to ischemia, causing the brain to shrink. The posterior fossa CSF spaces and the aqueduct will remain open, giving the clinical presentation of normal pressure hydrocephalus. Rapid pressure increase will tend to cause obstructions to CSF flow. When the overall volume occupied by the brain increases with space occupying lesions, enlarged ventricles, or oedema, there may be obstruction of CSF flow at the aqueduct 43. This is due to the narrowness of the aqueduct and its compressibility in comparison to the ventricles. As overall brain (including fluid) volume increases, CSF may be displaced from the subarachnoid spaces into the spine and the hindbrain may be displaced towards the foramen magnum. This may enhance the rate of intracerebral pressure increase by preventing the normal to and fro flow movement of CSF across the foramen magnum 44 that facilitates CNS venous drainage. These two major obstructive effects on CSF flow will contribute to the self-perpetuating nature of hydrocephalus, minor obstructions will also be detrimental.

A balance between brain turgor and ventricle pressure may occur, this is a feature of normal health and is also illustrated by cases where normal lateral ventricle size may occur with raised pressure and papilloedema. Alternatively, very rapid increase in ventricle pressure may cause symptoms of raised CNS pressure, with little time for the development of clinically detectable oedema of the optic nerve. A phenomenon of reactive enlargement of the ventricles following lumbar puncture has been observed in the presence of cerebral oedema 23. In these cases a decrease in overall CNS pressure with lumbar puncture allows improved venous drainage, water passes out of the oedematous brain and CSF may then move out of the spine into the cranial cavity as venous blood enters the spinal venous plexus. If the ventricles are small or obstructed and intracranial pressure is high the patient's condition will be critical. This will be indicated by obliteration of subarachnoid CSF spaces.

There are many disease processes that can lead to a state of CNS venous insufficiency by causing an increase in arterial supply, filtration of fluid into the parenchyma or venous pressure. An incomplete list includes CNS infection, hypoxia with altitude, carbon monoxide poisoning, water intoxication, renal impairment, obesity, and anaemia. Hydrocephalus may be caused by abnormalities of CSF production in the choroid plexus or absorption at the arachnoid granulations. That cardiac failure is an infrequent cause of hydrocephalus 45 may be an illustration of the efficiency of normal autoregulation of CNS extracellular fluid volume. Restriction of internal bone space through multiple suture craniosynostosis with a normally developing brain will tend to cause hydrocephalus 46 by compressing venous channels and CSF spaces. Single lambdoid suture fusion will be associated with raised pressure 46, because of the relative importance of posterior fossa size to ventricle emptying and foramen magnum flow. Direct and indirect measurements of venous pressure demonstrate variable correlation with hydrocephalus because pressure 5 and venous drainage 47 are dynamic processes. Current methods of quantifying these pressures are not representative of fluctuation with time.

Syringomyelia and spina bifida have been described as part of a disease continuum, with more severe manifestations in the fetus 48. The mechanism for neural injury in the original theory had a requirement for hydrocephalus with CSF flow from the head into the cord parenchyma via the fourth ventricle. Hydrocephalus is not always present in the two conditions and fluid flow from the fourth ventricle into spinal cord cavities in syringomyelia is uncommon. It is proposed that the disease continuum relates to posterior fossa hypoplasia that causes reduced CNS compliance before or after birth.

Features of spina bifida that need to be addressed for a unifying hypothesis to be plausible include:

• The nature of progressive neural injury

• An association with hydrocephalus

• A higher incidence of Chiari malformation in females

• The relationship between the level and severity of the lesion

• Small head size in the fetus

• Intrauterine growth retardation

• Syringomyelia after birth

Animal models indicate that spina bifida may result from failure of neural tube closure 49. Evidence suggests that mesodermal growth supports and shapes the neural tube to facilitate the closure process 4950. Vitamin A impairs growth of the mesoderm and may induce dysraphic disorders by this mechanism 51. Bones of the basichondrium arise from mesoderm early in fetal development, prior to completion of neural tube closure 52. Neural tube closure in humans proceeds in a rostral direction from the hindbrain region 49. It is proposed that genetic and environmental influences have variable effects on mesodermal growth influencing posterior fossa size and the ability of the para-axial mesoderm, which forms vertebral bone, to facilitate neural tube closure. Growth reduction in the posterior fossa will tend to increase CNS pressure which opposes neural tube closure. A fine balance between growth and pressure may be required from the earliest stages to achieve normal development of central nervous system and surrounding bone. Where raised pressure interacts with mesodermal restriction a failure of neural tube closure will occur. A simplified representation of the mechanism for dysraphism is shown in Fig. 4.

The more severe the mesodermal impairment, the more hypoplastic the posterior fossa will tend to be, with earlier and greater potential for separation of the spinal and cerebral CSF spaces and subsequent neural injury. Anencephaly represents the highest and most severe lesion. A more normal posterior fossa size will lead to more normal cerebral development. If posterior fossa size is not significantly restricted and a small deficit in the vertebrae occurs, a meningocele will tend to result at any level. The effect of timed dosing of vitamin A in the creation of neural tube defects adds strength to the argument that abnormal mesodermal growth at different stages may result in different morphological features. This teratogen results in more frequent anencephaly when given early in gestation and more frequent spina bifida if dosing is delayed 53. There will, according to this theory, be a tendency for higher lesions to be more severe, which accords with some observations on spina bifida 5455.

In anencephaly and spina bifida, early neural development may be relatively normal 56, with progressive damage during gestation. Posterior fossa restriction occurs at an early stage with progressive hindbrain herniation 5758. It is suggested that any movement may cause mechanical force on neural tissue. Fetal cardiac movement is present from five post-menstrual weeks, trunk movements are detectable from seven weeks and the fetus is active by ten weeks 59. It is proposed that disturbances in blood flow, accumulation of CSF and stretching forces act on tissue, which lacks mechanical support of the mesoderm. As fetal movement becomes stronger forces will be magnified around an open lesion, but the whole CNS may be affected. Widening of the whole vertebral column may be found in anencephaly, suggesting severe distending forces 48 and vertebral widening is found in the cervical spine with syringomyelia suggesting a mild distending force during growth 60.

Spina bifida is associated with reduced head size in the fetus 61. This may be because as gestation progresses skull growth depends upon pulsatility of pressure. This view is supported by the rapidity of response to tensile forces in skull suture fibroblasts 62. Pressure pulsations generated in the spinal venous plexus are normally transmitted to the head 13 but foramen magnum obstruction attenuates pressure transmission across the foramen magnum, between the two compartments 815. Posterior fossa hypoplasia may, by compressing the hindbrain, block transmission of pressure pulsations and so reduce the normal stimulus for skull growth in the fetus. Pressure changes with movement will tend to be confined to the spine, dissipated by the vertebral defect and prevent the vertebral canal from closing. The lowest spinal pressures will occur during relaxation of the fetus and may be abnormally low or low for excessive periods, as a result of the vertebral defect. It is proposed that these mechanisms explains the majority of the progressive cord injury that is observed in spina bifida. Brain growth in the presence of skull restriction would tend to decrease ventricle size and worsen the hindbrain herniation until foramen magnum obstruction impairs fourth ventricle emptying sufficiently that the ventricles enlarge. The skeletal abnormalities and abnormal pressure gradient lead to a progressively worsening hindbrain herniation that tends to become moulded and impacted 63. It is proposed that ventricle size in the affected fetus fluctuates with pressure according to Fig. 5. The phase of small ventricle size represents the early stage of relatively raised intracranial pressure. As hydrocephalus progresses the development of a severe foramen magnum obstruction is potentially lethal before or after birth.

Excessive pressure in the head or spine as a result of posterior fossa hypoplasia may be viewed as a progressive process occurring during or after gestation. In the fetus impairment to CSF flow may adversely influence neuronal migration 64. Animal studies demonstrate that ischemia damages developing neural tissue and neurobehavioural abnormalities, such as memory deficits may result from such injury 65. Abnormalities of migration have been described in human neurospheres transplanted into rat cerebral cortex following ischemic injury 66. Such studies give clues as to the possible origin of complex cerebral abnormalities that may be found in anencephaly and spina bifida.

The proposed hypothesis depends upon restriction of posterior fossa growth as a cause of neural injury with females having a smaller, genetically-determined, average posterior fossa size and cisterna magna CSF space than males. Posterior fossa size will be normally distributed. The genders represented separately will form overlapping curves with males to the right. If genetic or environmental factors that restrict growth of the posterior fossa interact with a normal variation in posterior fossa CSF space, and if the pathogenesis of spina bifida is related to hindbrain compression due to posterior fossa hypoplasia, there will tend to be a difference in severity of lesions between the sexes. Anencephaly will predominate in females, with lower spinal lesions being more common in males. This concept is represented in Fig. 6, and accords with observations on differences between lesion prevalence between the genders 4967. This hypothesis allows speculation that smaller CSF spaces in the female are part of a regulatory mechanism leading to smaller head size. With similar fluctuations in spinal venous volume a more spacious posterior fossa in the male may facilitate greater pressure wave transmission into the head resulting in larger fetal head size 68.

Acute worsening of obstructive hydrocephalus after birth has been observed with spina bifida 69. Repair of the spinal lesion may increase CNS pressure, whereas loss of CSF from a lesion may be beneficial for the fetus as gestation progresses in circumstances where CSF tends to accumulate. Breathing air and expansion of the thorax will increase capacitance in pulmonary vessels and reduce thoracic pressure 70. This will tend to improve CNS venous drainage and lower CNS pressure. Improved venous drainage may contribute to the low CSF pressure found in normal neonates 71. It may also encourage herniation of the hindbrain due to lowering of pressure in the spinal venous plexus. Loss of brain turgor associated with weight loss that tends to occur in the neonatal period 71 may also contribute to ventricle enlargement. These factors in combination will tend to favour foramen magnum obstruction, whereas taping of CSF in the neonatal period may avert an obstruction to CSF flow.

Abdominal growth tends to be reduced in fetuses with open spina bifida as gestation progresses 61. If degrees of CNS hypoxia are a feature of spina bifida that also progresses during gestation then it is possible that this pattern of abnormal abdominal growth would be expected. One mechanism for accommodating a chronic increase in central nervous system pressure may be decreased cardiac output 72 suggesting a mechanism for growth retardation. Areas of brain that are most damaged in anencephaly are also the most susceptible to ischemia in chronic hydrocephalus 7273. The relative preservation of basal brain structures in anencephaly will relate to their blood supply. Cervical meningoceles are sometimes found, but are not associated with significant CSF obstruction at the foramen magnum and neural development appears to be normal 74. Support for a theory of ischemic damage in spina bifida has been obtained from histological observations on the spine, movement analysis of neonates with spina bifida, 7576 and the presence of metabolites associated with ischemia in the CSF of affected neonates 77.

Surgery that improves CSF pathways at the foramen magnum may cause collapse of the syrinx cavity, improve cord blood flow and the clinical features of syringomyelia. This indicates that neural injury results from impaired flow of CSF at the foramen magnum 12. The central canal of the cord forms a potential space, into which fluid from the parenchyma may pass, allowing the formation of separate cavities that characterise syringomyelia 78. During physical exertion, movement of fluid within the syrinx cavity may weaken its walls causing it to have a greater capacity to enlarge 7980 and during relaxation, the cavity will tend to fill. This may be facilitated by a hyperaemic response. Syringomyelia occurs in association with posterior fossa restriction of Chiari I and other causes of reduced CNS compliance, particularly if they directly affect the spine. Tumours, vertebral deformity and arachnoiditis may reduce space and CSF flow and may lead to the onset of cavity formation. Vertebral injury causing abnormal function of the spinal venous plexus may impair spinal venous drainage as a result of autonomic dysfunction 81. Decompressing the cord ameliorates syringomyelia, and reducing the volume of fluid inside a syrinx cavity at operation may enhance the rate of any post-operative improvement 82. Normal pressure hydrocephalus is an analogous disease process to syringomyelia, as has been argued elsewhere, 83 with differing distribution of excess extracellular fluid. Spinal cord oedema would, according to this theory, represent a pre-syrinx state 84. Syringobulbia is the manifestation of raised spinal pressure with the unusual situation of fluid being able to track into the brain stem. Communicating syringomyelia would occur with pressure gradients and anatomy allowing flow from the fourth ventricle into the spine. A simplified representation of the mechanism for hydrocephalus and syringomyelia is shown in Fig. 7. Posterior fossa hypoplasia makes hydrocephalus and syringomyelia features of spina bifida after birth.