Article Text

Idiopathic normal pressure hydrocephalus
1. Jan Malm1,
2. Anders Eklund2
1. 1Associate Professor of Neurology, Department of Clinical Neuroscience, Umeå University, Umeå, Sweden
2. 2Associate Professor of Biomedical Engineering, Department of Biomedical Engineering and Informatics, Umeå University, Umeå, Sweden
1. Correspondence to:  Jan Malm PhD, Department of Clinical Neuroscience, Umeå University 901 85 Umeå, Sweden;  jan.malm{at}neuro.umu.se

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Hydrocephalus can present with acute or chronic symptoms at any age. Obstructive and communicating hydrocephalus are the two main types. In adults, the most common form of the latter is normal pressure hydrocephalus (NPH) with a mean age at onset of about 70 years, equally common in both sexes. This review will focus on the idiopathic variant of the disease (INPH), the typical features of which are imbalance, gait disturbance, urinary symptoms, and cognitive decline. Although the treatment is surgical, by placing a cerebrospinal fluid (CSF) shunt, neurologists are essential for diagnosing patients with gait disturbance and/or ventriculomegaly, for preoperative selection of shunt candidates, and to optimise the CSF shunt system postoperatively.

## EPIDEMIOLOGY

Very little is known about the incidence and prevalence of INPH. In a door to door survey in people older than 65 years, 0.7% of the population had Parkinson’s disease and 0.4% NPH.1 In Sweden, between three and six shunt operations in adults per 100 000 inhabitants are done every year at the six neurosurgical centres, about 30% of which are for INPH and 20% for secondary NPH.2 The pre- and postoperative investigation of INPH is the fourth most common diagnosis after stroke, epilepsy, and headache at the neurology inpatient service at our hospital.

## AN IMPORTANT ISSUE FOR NEUROLOGISTS

There are few neurological conditions where treatment has such a large impact on outcome as INPH. The patients are elderly and even a small improvement in their gait or activities of daily living has the potential to substantially increase their independence. In most countries hydrocephalus is considered a “neurosurgical problem” without any need to consult the neurologist. But we believe that a multidisciplinary approach with the neurologist and neurosurgeon in fruitful cooperation is what is needed.

An important issue for the general neurologist is the evaluation of patients referred because of gait disturbance and/or ventriculomegaly. It is of vital importance to perform any shunting as soon as possible because the chance of a good outcome decreases with progression of disease.3 A dedicated neurologist is needed for the preoperative selection of whom to operate on, while the neurosurgeon is responsible for the surgical procedure and the postoperative care, as well as any shunt revisions. At follow up visits, the neurologist should evaluate the success—or not—of surgery, exclude shunt complications, and if necessary adjust the opening pressure of the CSF shunt.

## IDIOPATHIC NORMAL PRESSURE HYDROCEPHALUS: MYTH OR REALITY?

The silicone CSF shunt valve was introduced in 1952 and six years later the first operations on hydrocephalic patients with normal intracranial pressure were performed, although not published until 1965.4 The successful results led to the recognition of normal pressure hydrocephalus. A distinction between idiopathic and secondary NPH was suggested in 1975.5 Because of the name, it is often assumed that INPH patients have normal intracranial pressure (ICP). However, the term normal pressure hydrocephalus is misleading because the ICP is on average slightly higher than normal (fig 1).6 We prefer the term idiopathic adult hydrocephalus syndrome, which was introduced by Ekstedt in 1989, as a more correct description7: the ICP is slightly increased, it is a syndrome with a specific definition, and it affects adults.

Figure 1

Intracranial pressure (ICP) in INPH and healthy elderly, calculated at equilibration, about 30 minutes after insertion of the needle.6 The boxes show the median values and the interquartile range, the whiskers the outermost data points.

## A TREATABLE DEMENTIA?

INPH was initially described as a treatable dementia and many assumed that CSF shunting would ameliorate the symptoms in most patients. Shunt surgery became the accepted wisdom despite the lack of any scientific evidence. However the diagnosis, as well as the effectiveness of surgery, has been questioned and a Cochrane Review concluded “there is no evidence to indicate whether placement of a shunt is effective in the management of NPH”.8 On the other hand, a recent guideline committee states “surgical diversion of CSF is recommended for INPH patients in whom there is a favourable risk-benefit ratio”. 9 The reason for this discrepancy is that the outcome of shunted and non-shunted INPH patients has never been compared in a randomised controlled trial. However, we feel that with a clear definition of INPH followed by robust investigation, doubts about the usefulness of CSF diversion can be removed.

## PROGNOSIS

Most often the outcome of shunting is reported as a comparison between the pre- and postoperative functional level of individual patients. In a large, retrospective study of 127 patients, only 15% markedly improved.10 On the other hand, in recent prospective studies from centres specialising in hydrocephalus treatment, about 70% of patients improve within the first year after surgery.3,6,1113 Usually balance and gait get better, and there is probably also a slowing down of cognitive decline after shunting. Using the Mini-Mental State Examination (MMSE), two thirds of demented patients (that is, <25 points) had at least a two point improvement three months postoperatively.6 If more sophisticated neuropsychological tests are used, different studies contradict each other with limited or no improvement in cognitive function following shunt diversion in some3,6,14,15 and postoperative improvement in others.1618

Of course, as the patients get older, the long term results will be influenced by age related comorbidity. In one study of 42 shunted patients (mean age 72 years) 64% had improved three months after the operation and 26% remained much improved in gait and activities of daily life at three year follow up.19 However, the mortality among INPH patients was about three times greater than a general elderly population, the deaths mostly being the result of vascular disease, although a small number of patients experienced serious shunt related complications.

## LIMITATIONS OF THE RESEARCH

Unfortunately, many conclusions are vague or unreliable in hydrocephalus research because of methodological limitations. Idiopathic cases often are lumped together with secondary cases despite their different aetiology, age, and prognosis. The studies are often observational and more often retrospective than prospective. Inclusion and exclusion criteria differ between different studies. There is a need for a widely accepted definition of the syndrome in order to include homogeneous patients in future study populations. The outcome measures have always been focused on absolute improvement from shunting; little is known about whether surgery can affect the progress of disease in shunted patients without postoperative improvement, or whether a shunt in a patient with minimal symptoms can prevent further deterioration. This kind of research requires control groups. Another important issue is to increase the objectivity of the pre- and postoperative assessments.

## THE RELATION BETWEEN CSF DYNAMICS AND SYMPTOMS

The CSF is a dynamic system: the brain environment is affected by the formation rate of CSF, the outflow resistance, venous pressure, compliance of the craniospinal system, and variations in ICP. When these are not in balance hydrocephalus results and the clinical syndrome of INPH (fig 2). The causal relation between abnormal CSF dynamics and the symptoms is supported by the observation that patients improve after shunting. Therefore, clinical improvement can be accomplished by changing the brain’s environment—that is, the CSF dynamics.

Figure 2

INPH can be explained by the hydrodynamics of the CSF system (including any shunt) which create the working environment of the brain; deteriorating brain function due to the disturbed environment; and so reversible clinical symptoms due to impaired brain function. The question mark indicates the missing link between the hydrodynamic environment and brain dysfunction. Most of the research in this field has followed the curved arrow as it has been shown to develop predictive tests for a good outcome after shunting.

Let us therefore look more closely at the CSF system. The brain and spinal cord are contained in a cavity of approximately 1.5 litres. Using MRI, the mean intracranial CSF volumes in INPH and controls are 280 ml and 195 ml, respectively.20 The most important function of the CSF is probably physical support of the brain.21 It also acts as a volume reservoir, vital to accommodate intracranial volume expansion. Furthermore, CSF is necessary for ionic homeostasis and clearance of metabolites formed in the brain. The main production of CSF takes place in the ventricles by secretion from the choroid plexus, and possibly in other parts of the craniospinal space, for instance as metabolic water in the parenchyma.22 The formation rate is believed to be independent of ICP and is about 500 ml/day. The main site of CSF absorption is via the arachnoid villi into the dural venous sinuses, and also into the spinal subarachnoid space.23

The rate of CSF absorption is proportional to the difference between the ICP and the venous pressure in the dural sinuses. At steady state equilibrium the formation rate is equal to the absorption rate. Therefore, the pressure difference is set by the formation rate and the outflow resistance through the arachnoid villi. Steady state ICP will thus be regulated by dural sinus venous pressure (Pds), outflow resistance (Rout), and CSF formation rate (qf):

$Math$

Intracranial pressure is the most easily accessible marker of the balance produced by these properties, and is defined as the counter-pressure that has to be applied to a needle placed in the CSF system in order to stop CSF flow. When clinicians are discussing an ICP value, measured intracranially or via the lumbar space, they are usually referring to a static pressure—that is, the mean ICP. However, the ICP varies continuously around the mean due to physiological periodic variations, for example originating from changes in blood volume resulting from the heartbeat, and also from vasomotoric waves with a duration of minutes. The magnitude of the ICP fluctuations in relation to the volume variations is described by the compliance of the craniospinal system.

## PATHOPHYSIOLOGY OF INPH

A grand unified theory explaining the pathophysiology of INPH has yet to be proposed and validated (fig 2). It must explain the development of hydrocephalus (the working environment of the brain), the symptoms of the patient (their brain function), as well as why these symptoms are improved by a CSF shunt. A hallmark is the disturbance in the CSF dynamic system and one of the most concordant findings is high resistance to CSF outflow (Rout).6, 12, 24 A relation of Rout to the ventricular enlargement is plausible, although the mechanism has not been clarified. It has however been shown that a CSF shunt converts the pathological high Rout into an extremely low value.25 Other disturbances in the CSF dynamic system are listed in table 1.

TABLE 1

Disturbances in the CSF dynamic system (“the working environment of the brain”) in INPH

A disturbance in CSF dynamics cannot by itself explain the clinical features. For this, involvement of the brain is necessary (fig 2). Current theories include:

• mechanical compression of the brain parenchyma caused by the dilated ventricles26

• ventricular reflux and periventricular absorption of CSF27

• inability to clear potentially toxic metabolic products from the CSF28

• non-symptomatic congenital hydrocephalus that becomes symptomatic with increasing age29,30

• increased amplitude of the arterial pulsations during systole inducing large ICP pulsations (waterhammer) leading to secondary damage to the veins in the periventricular region.3133

The symptoms of INPH are more and more recognised as being caused by a complex interaction between a CSF dynamic disturbance and cerebrovascular disease, mainly affecting the periventricular deep white matter.34 In INPH, the frontal horns are pathologically widened, and there are axons in the white matter in surrounding parts of the brain.35 This area is supplied by the small perforating end-arteries creating boundary zone territories sensitive to ischaemia.36 Thus, the predisposing causes of the syndrome include vascular disease, arterial hypertension,36 decreased cerebral blood flow, and ischaemia3641 (table 2).

TABLE 2

Evidence of vascular white matter damage in idiopathic normal pressure hydrocephalus

In the last year, two studies to elucidate “the missing link” (fig 2)—that is, how the brain parenchyma interacts with the CSF hydrodynamic disturbance—have been published. In both, CSF infusion has been used to manipulate the ICP. In the PET study, cerebral blood flow was reduced in the white matter adjacent to the ventricles and progressively normalised with distance from them.42 In the microdialysis study, the results indicated low grade ischaemia in the periventricular white matter which was ameliorated after CSF removal.43

## WHAT SHOULD THE NEUROLOGIST LOOK OUT FOR?

Because decline in balance, gait, or memory are some of the most common neurological complaints in the elderly population, it is a challenge to diagnose the occasional case of INPH. The diagnosis (table 3) is based on the clinical symptoms, radiological features, and the exclusion of other diagnoses. Gait disturbance is the cardinal feature and a mandatory symptom, almost always preceding the development of other symptoms.6,44 The patient may or may not have mental impairment and/or urinary symptoms. If the history and the neurological examination are more or less consistent with the diagnosis, the patient should have a CT scan, and if this reveals communicating hydrocephalus the patient should be referred to a hydrocephalus group or a neurosurgical department. In table 3, we have also brought together signs and symptoms that make the diagnosis less likely in certain patients. However, many of these patients may still have the potential to improve following surgery.

TABLE 3A

Criteria for the diagnosis of INPH

TABLE 3B

Negative prognostic factors, indicating a less likely probability of postoperative improvement

Historically one diagnostic criterion for INPH has been improvement after shunting. In addition to being impractical, this criterion can be wrong because the disease may have reached an irreversible stage with no chance of postoperative improvement despite a correct diagnosis and a functioning CSF shunt. The diagnosis must be based on preoperative evaluation, not the result of surgery.

### History

A careful patient history is of utmost importance and if there is any cognitive decline, the relatives should be interviewed as well. An INPH patient almost always begins with balance and gait disturbance44 whereas in other dementias similar symptoms are preceded by pronounced cognitive decline. A common opinion is that the results of surgery are less favourable the longer the duration of symptoms.19 For instance, dementia lasting for more than two years has been considered a bad prognostic sign.45 Patients with NPH caused by subarachnoid haemorrhage, traumatic brain injury, and bacterial meningitis should be excluded by the history, and diagnosed as secondary NPH.

### Gait disturbance

The cardinal feature is disturbance of balance and gait which develops insidiously over many months, sometimes years. At first, the patient complains of problems when walking in woods and fields, but later also on flat surfaces. At this stage, falls are common. The patient starts to use a stick, then a wheeled walker, and finally a wheelchair. On examination, the gait disturbance is symmetrical. The stride length and step height are decreased and the patient walks slowly with an increased number of steps.46,47 The gait may be mildly or moderately broad based. Turning generally includes multiple unsteady steps, and there is often sway and a tendency to fall. Increased muscle tone (Gegenhalten) in the lower extremities is sometimes noted. On the bed, the patient can usually perform simple motor tasks such as the heel-shin test, even if they are unable to walk. With Romberg’s test, or “standing on one foot”, there is a tendency to sway or fall. Bradykinesia in the upper extremities is a common finding, as well as other extrapyramidal features, but these are usually of minor importance to the patient compared to the gait disturbance, and also these signs are symmetrical in contrast to Parkinson’s disease.48,49

### Cognitive dysfunction

The degree of cognitive decline varies: using the MMSE, about 60% of patients score less than 25,6 and with an extensive neuropsychological battery most will probably show cognitive impairment. It is not possible to differentiate between INPH and Alzheimer’s disease on the basis of the the different domains of the MMSE.50 The guiding rule for the clinician is that severe or longstanding dementia probably predicts a poor response to shunting.3,45 Severe cognitive disturbance is also correlated with neuritic Alzheimer plaques obtained at operative biopsy.51

Few studies have compared INPH with healthy elderly, Binzwanger’s disease, or Alzheimer’s disease patients.50,52 Thus, it is difficult to provide a specific neuropsychological profile53 but INPH is often classified as a subcortical dementia. Characteristic cognitive features are memory loss, slowness in thought and in processing information, difficulty in planning, and mental shifting.54 There is also decreased attention and visuospatial disturbances. Frontal lobe dysfunction50,55,56 has been suggested. Most neuropsychological data come from before and after shunting but of course these cannot be used to predict the surgical outcome.

### Urinary symptoms

Although urinary frequency and urgency are common problems and one of the main features, these are very common in the elderly and anyway incontinence is a late symptom in INPH. Thus, these symptoms are too non-specific for standalone diagnosis. Nonetheless, because urinary symptoms often improve following CSF diversion, they are important to ask about in the preoperative evaluation.

## NEUROIMAGING

A CT or MR brain scan should be performed in any patient with symptoms and signs compatible with the diagnosis of INPH. Although a CT scan will reveal the ventriculomegaly, an MR scan is mandatory before surgery57 to exclude obstructive hydrocephalus. Ventriculomegaly is a cardinal feature of INPH and the degree can be defined by an Evan’s index greater than 0.3 (fig 3). Other important findings are flattened cortical sulci in the top slices of the scan, and widened temporal horns (fig 3).58 Sometimes there is focal dilatation of the cortical fissures and sulci.59 CT scan can also reveal the degree of any cortical atrophy and previous cerebrovascular events. Cortical atrophy is not an absolute contraindication to surgery, but should raise the question of Alzheimer’s disease.

Figure 3

Preoperative (left) and postoperative (right) CT brain scans. The ventricles are widened and the cortical sulci in the top slices are flattened, less postoperatively. Evan’s ratio is calculated as the ratio of the greatest width of the frontal horns of the lateral ventricles (A) to the maximal internal diameter of the skull (B). Observe the tip of the shunt catheter in the postoperative images.

## DIFFERENTIAL DIAGNOSIS

Among patients admitted to our hospital because of communicating hydrocephalus on CT scan, or clinical suspicion of INPH, 60% do not have INPH and 20% are diagnosed with INPH but not operated on because of minor disability, operation is refused by the patient, or severe dementia.6 Thus, only one in five admitted patients is suitable for surgery. As most of the differential diagnoses are non-surgical, a neurologist is probably best suited to select the proper surgical candidates.

The first diagnostic issue concerns other forms of hydrocephalus. All types of hydrocephalus, irrespective of aetiology, may present with disturbance of balance, gait, and cognition. Patients with aqeduct stenosis may have the same clinical features as INPH.60 Of course, many of these patients should have surgery, but the aetiology, investigational methods, as well as treatment options and prognosis, are different from INPH.

The second issue is to exclude other types of dementia such as subcortical or vascular dementia (Binswanger’s disease) and Alzheimer’s disease. Keep in mind that although gait disturbance often is the first symptom of INPH and that other forms of dementia can be accompanied by balance and gait disturbances, these are generally not until the advanced stages. Most clinical features as well as MRI findings of a patient fulfilling the diagnostic criteria for vascular dementia can also apply to a patient with INPH and “mixed” cases further complicate the diagnostic problem. A diagnostic rule is that extensive leukoaraiosis is more common in vascular dementia, and dilated ventricles are the hallmark of INPH. In addition, patients with vascular dementia often have a history of stroke and focal neurological signs such as hemiparesis.

The third diagnostic issue is to exclude gait disturbance caused by other problems which are common in the elderly such as pain, polyneuropathy, and myelopathy. Asymmetrical gait is the hallmark in arthralgia, hip prosthesis, or hemiparesis. Parkinson’s disease gait disturbance may resemble the INPH gait.

## WHAT ARE THE SPECIALISTS DOING?

### Predictive tests

After confirmation of the diagnosis, the next question is whether or not the patient is likely to benefit from a shunt. Few neurosurgeons base their decision on just the clinical picture—predictive tests are used in patients with the clinical diagnosis of INPH and are also helpful in selecting borderline cases not entirely fulfilling the diagnostic criteria. The tests can roughly be divided into two categories: those that aim to measure parameters characterising the CSF dynamic system, and CSF removal tests that simulate shunting. Unfortunately many predictive tests lack validation, and an important problem in some studies is that the patients were included on the basis of the predictive test under study. Furthermore, many studies included only clinically typical cases whereas most of the patients admitted for assessment are less typical.

A radioactive isotope is injected via a lumbar puncture into the subarachnoid space and using a gamma camera the speed and flow direction of the isotope are observed. Normally, the isotope flowing with the CSF accumulates over the cortical surface. In INPH, where there is thought to be retrograde flow, accumulation of the isotope in the ventricles is anticipated. This method was once very popular but in 1992 a review concluded that “cisternography does not improve the diagnostic accuracy”61 and there has been little interest since then.

#### Infusion methods

CSF infusion tests are designed to reveal the dynamics of the CSF system. During infusion of artificial CSF via the lumbar or ventricular route, the resulting ICP response is recorded and parameters such as outflow resistance, compliance, formation rate, and dural venous pressure are calculated. However, these tests are carried out, analysed, and interpreted in different ways in different centres, which is a problem when comparing study results. Another problem is the range of physiological variation—that is, small or large amplitude changes of the ICP, making an investigation of a patient with small fluctuations more reliable than that of a patient with larger variations. Thus, standardisation of the infusion tests is needed and the precision of the measurements should be defined.62

It is generally believed that the outflow resistance (Rout) (or the inverse, CSF outflow conductance) is the clinically most important CSF dynamic parameter. Determination of Rout can be considered as a “three-in-one investigation”: as a diagnostic tool, a prognostic test, and a technique to check CSF shunt function. The Rout is increased in INPH patients compared to healthy adults6 and to patients with Alzheimer’s disease or vascular dementia63 (fig 4). It has been stated that if the outflow resistance exceeds a certain threshold, this is an excellent predictor of surgical outcome.24 However today there is a lack of consensus concerning the usefulness of this measure, with some studies supporting Rout3, 12, 64 and others finding it less useful in the selection process.6 The methodological issues discussed above can explain some of this discrepancy. Furthermore, an INPH patient in an irreversible stage of disease can have pathological resistance but not be shunt responsive.

Figure 4

CSF outflow conductance (that is, the inverse of resistance, Rout) (mean±95% confidence interval) in patients with INPH (AHS), vascular dementia (VD), Alzheimer’s disease (AD), idiopathic intracranial hypertension (IIH), superior saggital sinus thrombosis (SSST), other cases, and normal controls.79

#### Intracranial pressure measurement

Slow and rhythmic oscillations in ICP (fig 5), known as b-waves, have been claimed to be one of the best preoperative predictive factors in INPH.45 Their origin is not fully understood but they may stem from oscillations in the blood volume of the craniospinal space, and a relation to compliance or elastance of the brain has also been proposed.65 However, examination of b-waves is invasive, involving surgical insertion of an intracranial probe followed by pressure measurement lasting for at least five hours,66 and the definition of b-waves varies widely. We recently presented a computerised method for analysis of b-waves but they were only weakly related to post shunting improvement.67

Figure 5

Example of intracranial pressure (ICP) recording. In the 8–15 minute interval typical b-waves are shown.

#### CSF tap tests

Simulation of the effect of shunting by draining CSF is another approach to predict outcome after surgery. This type of test was first described by Wikkelsö 68 and is done by measuring psychometric function and gait pattern before and after draining 40–50 ml of CSF at lumbar puncture. It is easy, safe, and inexpensive, and is probably the most commonly used test to identify patients suitable for shunting.69 But although the positive predictive value is high, there are still a considerable number of patients who can improve after shunting in spite of a negative tap test.64 A European multicentre study is ongoing and will hopefully provide data on the reliability of this test.

False negative CSF tap tests (that is, the patient improves after surgery, despite a negative tap test) are seen in most studies, which has lead to the hypothesis that the test would be more reliable if a larger volume of CSF was removed over several days, imitating a CSF shunt. For continuous CSF drainage a catheter is placed in the lumbar CSF space for 2–4 days and 150–250 ml is removed each day. This is expensive because the patient has to stay in hospital for several days, there is a risk of meningitis,3,70 and catheter related problems are not uncommon (blockage or catheter withdrawn by a confused patient). Although this method has been considered as a valuable predictive test3 the scientific support is weak, with two positive studies3,71 and one negative.70 A well designed multicentre study is needed.

#### Blood flow measurement

There are numerous studies of regional cerebral blood flow with SPECT, PET, and Xenon-CT before and after surgery, all showing decreased rCBF in INPH. However, the areas of impaired flow differ from study to study and the patterns are not useful in predicting response to shunt surgery.72,73

#### MRI

An MRI is necessary to confirm communicating hydrocephalus (that is, to exclude obstruction of the CSF pathways) and to quantify the severity of any white matter lesions and atrophy. These abnormalities do not necessarily predict a poor outcome from surgery38,7476 but if they are very obvious it is our opinion that there must be other strong reasons to justify shunting, for instance high CSF outflow resistance and improvement after CSF drainage.

CSF dynamics can be studied with MR techniques measuring CSF flow in different parts of the CNS. Increased pulsatile CSF flow velocity or flow volume through the aqueduct during each cardiac cycle has been suggested as a predictor of successful surgery.77 But, as is so commonly the case in INPH research, different techniques have been used and the results are contradictory.78,79

### Criteria for surgery

Different centres investigate patients in different ways and the surgical indications vary, but generally clinically typical patients are shunted, while more attention has to be paid to the risk-benefit balance in atypical cases. In summary, the investigative process in our hospital for diagnosing INPH and selecting shunt candidates includes history and clinical examination, MRI scan, CSF outflow resistance measurement including CSF tap test, and in a few patients long term CSF removal (Fig 6).

Figure 6

Flow diagram describing the investigational process used in Umeå for diagnosing INPH and selecting shunt candidates. Boxes with rounded corners indicates endpoints of the INPH investigation.

## WHAT IS A CSF SHUNT DOING?

It is generally believed that the essential and sole role of a CSF shunt in a hydrocephalic patient is to increase CSF outflow and so normalise ICP and improve the patient’s symptoms. However, very little is known about what the optimal physiological profile of the CSF circulation should be postoperatively, and there is little agreement as to which CSF valve system is the most efficient and reliable. No CSF shunt has been proven superior to others, and yet it is estimated that 50% of postoperative shunt complications are directly or indirectly caused by inadequate hydrodynamic performance of the shunt.80

A shunt consists at least of a ventricular (proximal) catheter, a shunt housing with a valve mechanism, and a distal catheter for outlet, most often into the abdominal cavity (fig 7). Most of the CSF shunts currently on the market are of the differential pressure type: the opening pressure (Pop) of a shunt states the differential pressure across the shunt at which the valve opens.

Figure 7

A standard CSF shunt system. The CSF is diverted from the ventricles, through the shunt valve and into the peritoneal cavity.

The CSF flow and differential pressures in a shunted patient in the supine position are shown in figure 7. The differential pressure over the shunt is the difference in pressure between the ICP (in the ventricle) and the pressure at the outlet (that is, in the peritoneal cavity). For differential pressures lower than the opening pressure, the shunt is inactive and will not affect the CSF dynamics. For differential pressures exceeding the opening pressure there will be net flow through the shunt and outflow will take place both through the normal pathways and through the shunt. In practice, the shunt flow dominates, the pressure in the peritoneal cavity (Ppc) will be the reference pressure, and the steady state ICP in the shunted patient can be approximated with:

$Math$

where Rsh is the flow resistance of the shunt system. This relation assumes that the shunt is open—that is, that differential pressure exceeds the opening pressure of the shunt (ICP − Ppc > Pop), and that the Rsh is much smaller than the outflow resistance of the normal pathways.

It is possible to evaluate CSF shunt characteristics postoperatively using a CSF infusion method.81 Postoperative CSF outflow resistance remains remarkably stable, and much lower than at baseline (that is, preoperative investigation showing resistance of normal pathways) in patients with functioning CSF shunts. ICP also remains stable, but similar to the preoperative value.

In the upright position the situation becomes more complicated. There is a hydrostatic siphoning effect from the vertical CSF column between the ventricle of the brain and the outlet at the abdominal cavity (a similar principle as siphoning aquarium water). This siphoning can cause overdrainage and modern shunts therefore often include an anti-siphoning device to eliminate this problem. Another common feature of modern shunts is an adjustable opening pressure valve—that is, an externally applied magnetic field changes a spring mechanism, thus increasing or decreasing the opening pressure of the CSF shunt. By way of this the properties of the shunt system can be tuned postoperatively if the clinical outcome is not satisfactory.

Following CSF shunting, a follow up visit is important (table 4) to evaluate whether an unchanged or worsened patient can be helped by shunt adjustment and to discover any cases of shunt dysfunction requiring surgical revision. It is also possible that even an improved patient could benefit more with better adjustment of the shunt. Gait, cognition, and activities of daily living should be evaluated with standardised scales. Video recording of gait is an excellent method to compare the baseline and post shunting states.

TABLE 4

Postoperative follow up 3–6 months after shunting

A CT brain scan will exclude a subdural haematoma caused by overdrainage, as this can be asymptomatic. Compared with preoperative images, unchanged width of the ventricles despite clinical improvement is a common finding, and it is important to note that the width of the ventricles does not reflect the function of the CSF shunt.82,83 Sometimes widening of the cortical fissures is more pronounced postoperatively (fig 3). The CT scan also reveals the location of the proximal (or ventricular) catheter, which should usually be in the right frontal horn (fig 3).

Blood samples are drawn to check for signs of infection. If CSF is available it should also be analysed. There is no contraindication to lumbar puncture in a patient with a CSF shunt because of communicating hydrocephalus. Puncture of the shunt valve should be avoided, but may be indicated if there is suspicion of CSF shunt infection or dysfunction, but only after discussion with a hydrocephalus specialist.

No clinical improvement despite a functioning CSF shunt may be explained by either misdiagnosis, or by the development of a concurrent disease. In addition, the INPH syndrome at the time of surgery may have reached an irreversible state which cannot be improved by shunting. A non-functioning or suboptimal functioning CSF shunt means that a patient with potential for improvement does not improve, or first improves and later deteriorates. Thus, an important issue is to clarify whether the CSF shunt system is working properly:

• Inspect the CSF shunt valve (usually behind or above right ear) as well as the tubing (fracture? skin infection?).

• Pumping the valve gives a rough estimate of shunt function (fig 8). If the valve does not refill after compression, there is probably shunt dysfunction (usually obstruction in the proximal catheter). On the other hand, “normal” pumping and refilling of the valve does not guarantee a functioning CSF shunt.

• We do a lumbar infusion test at follow up to check for shunt patency (fig 9). We consider this as the gold standard for testing shunt function in patients with communicating hydrocephalus.8486

• If shunt dysfunction is revealed, a “shunt series” of plain x ray images visualising the entire shunt system should be performed, as well as an ultrasound examination of the abdomen to exclude a cyst or abscess at the distal catheter tip.87

Figure 8

Pumping the valve

Figure 9

Pre- and postoperative evaluation with a CSF infusion method. The slope of the regression lines corresponds to the outflow resistance, Rout. Normal Rout slopes are shown by dotted lines. This INPH patient had a typical pattern with a higher than normal Rout preoperatively, which changed to a lower than normal Rout after shunting.

With modern, adjustable, or manoeuverable CSF shunts, it is possible to modify the opening pressure postoperatively. A new field of research is to evaluate whether patients showing improvement can improve further through lowering or increasing the opening pressure. Also, a small subdural haematoma can be treated by increasing the opening pressure, thus avoiding a neurosurgical operation.88 Adjusting the opening pressure of the CSF shunt requires frequent follow up visits and careful monitoring of the patient in between. Modern CSF shunts are easy to handle and require minimal training to learn how to adjust.

Quality assurance for patients with hydrocephalus is of utmost importance. In Sweden, there is a national hydrocephalus registry, where all units register their CSF shunt operations. The registry is web based, making it possible to present real time analyses. In the UK, there is a similar registry.89

## COMPLICATIONS OF SHUNTING

Shunt related complications are common in patients with INPH. In 147 INPH patients treated with a modern, programmable valve, 23% had a complication that required shunt revision: Underdrainage in 34%, subdural fluid collection (that is, overdrainage) in 20%, and infection in 20%. The five year shunt survival rate (to shunt revision) in this study was 80%,90 other studies have shown survival rates of about 50%.88,91

## UNDERDRAINAGE

Common causes are obstruction (of the ventricular catheter, shunt valve, or abdominal catheter), fracture of the tubing, displacement of the catheter tips, or incorrect setting on an adjustable shunt. In patients with a nonfunctioning shunt the symptoms will reappear or become aggravated but in INPH they are never dramatic, in contrast to obstructive forms of hydrocephalus.

## OVERDRAINAGE

In a standard CSF shunt system, there is always some gravity induced flow (siphon effect). Even if a system with antisiphon properties is used, this mechanism can fail and cause overdrainage. Symptoms related to overdrainage are very variable, from a mild headache to life threatening intracranial bleeding. The most common symptom is a posture related headache (similar to post lumbar puncture headache), worse when the patient is sitting or standing, and relieved when lying down.

## INFECTION

Infections are most commonly caused by Staphylococcus aureus or epidermidis and they usually present during the first year after the operation. The patient has general signs of infection with malaise, nausea, vomiting, fever, leucocytosis, and an increased blood CRP. The symptoms are often vague, and so an infection should be sought in all CSF shunt patients presenting with a non-specific clinical picture. Follow the catheter and look for signs of skin infection or tenderness, or pain in the abdomen. As a result of the infection, the shunt valve can stop functioning, giving the patient symptoms of underdrainage. Besides antibiotics, treatment often requires removal of the CSF shunt system.

### CONCLUSIONS

• Hydrocephalus is a common disorder and its reversibility, by means of a CSF shunt, justifies increased awareness among neurologists. A multidisciplinary approach together with a neurosurgeon is recommended.

• Dilated ventricles, impaired balance, and gait disturbance are the cardinal features. Mental impairment and/or urinary symptoms may also occur.

• INPH is increasingly recognised as caused by a complex interaction between a CSF dynamic disturbance and cerebrovascular disease.

• The most important predictive tests are CSF outflow resistance/conductance and a CSF tap test.

• Today, 70–80% of patients operated on with a CSF shunt improve postoperatively.

• The neurologist should learn how to recognise the signs and symptoms of a shunt complication, as well as how to adjust the opening pressure of the shunt.

• A lumbar infusion test is the gold standard for testing shunt function in patients with communicating hydrocephalus.

## Acknowledgments

The authors wish to thank Bo Kristensen, Lars Johan Liedholm, and Aina Ågren-Wilsson for helpful discussion and comments on the text of this article. The article was reviewed by Professor Ian Whittle, Edinburgh, UK.