Journal of Neuro-Ophthalmology, June 2004, Volume 24, Issue 2

Cerebrospinal Fluid Diversion Procedures

Cerebrospinal fluid (CSF) diversion procedures remain the principal method of treatment of hydrocephalus and an important option in treating idiopathic intracranial hypertension. Recent advances in CSF shunt hardware offer some promise in reducing the rate of complications. Third ventriculostomy has become an increasingly practiced alternative to conventional shunting in an ever-widening patient population. Long-term follow-up studies have identified complications of lumboperitoneal shunt placement. Advances in surgical navigation suggest that ventriculoperitoneal shunting may be a viable alternative in patients with idiopathic intracranial hypertension.

STATE OF THE ART Cerebrospinal Fluid Diversion Procedures Hugh J. L. Garton, MD, MHSc Abstract: Cerebrospinal fluid ( CSF) diversion procedures remain the principal method of treatment of hydrocephalus and an important option in treating idiopathic intracranial hypertension. Recent advances in CSF shunt hardware offer some promise in reducing the rate of complications. Third ventriculostomy has become an increasingly practiced al­ternative to conventional shunting in an ever- widening pa­tient population. Long- term follow- up studies have identi­fied complications of lumboperitoneal shunt placement. Advances in surgical navigation suggest that ventriculo-peritoneal shunting may be a viable alternative in patients with idiopathic intracranial hypertension. ( JNeuro- Ophthalmol 2004; 24: 146- 155) Diseases of disordered cerebrospinal fluid ( CSF) me­chanics, including hydrocephalus and idiopathic intra­cranial hypertension ( IIH), are often treated with CSF di­version procedures, accounting for nearly 70,000 hospital admissions in the United States alone ( 1). CSF FLOW DYNAMICS Although often thought to have unidirectional bulk flow from choroid plexus through ventricles to subarach­noid space, CSF movement on magnetic resonance imaging ( MRI) is oscillatory and pulsatile ( 2). Egnor et al ( 3) have suggested that these observations should lead to the concept of the CSF as a harmonic resonator, buffering the arterial pulsations and transferring them to the venous system, while shielding the capillaries from pulsatile flow. These authors suggest that communicating hydrocephalus results not from mechanical obstruction of CSF outflow, but from a loss of arachnoid buffering capacity to arterial pulsations, resulting in a relative shunting of blood flow to the choroi­dal vasculature. This shunting is said to lead to increased pulsations of the choroid plexus and to increased local in­traventricular pressure as measured across the cortex. As a result, the ventricular system dilates, with the accumulation of CSF passively filling the newly available space. Department of Neurosurgery, University of Michigan School of Medi­cine Ann Arbor, Michigan. Address correspondence to Hugh J. L. Garton, MD, Taubman 2128- 0220, 1500 E. Medical Center Drive, Ann Arbor, MI 48109; E- mail: hgarton@ umich. edu These theories, which have been questioned ( 4) and remain to be validated, have not yet been extended to IIH. Nevertheless, sagittal sinus venous pressures have been shown to be increased in case series of patients with IIH even when no anatomic obstruction is identified ( 5). Thus, if the primary obstruction is at the level of venous outflow, the CSF would be unable to transfer arterial pulse pressure to the venous system and result in an increase in intracranial pressure ( ICP). The ventricular system in this instance would not dilate because both subarachnoid and intraven­tricular spaces are affected equally with no preferential flow to the choroid plexus vasculature and no transmural pressure gradient. The lack of ventriculomegaly would be further explained by the decrease in the brain's compliance to compression as the result of increased resistance to ve­nous outflow. Impairment in the buffering capacity of CSF is also part of the proposed pathophysiology of syringomyelia oc­curring in the context of the Chiari I malformation. Loss of communication between the cranial and spinal subarach­noid space leads to inappropriate elevations in pulse pres­sure in the spinal subarachnoid space caused by the piston effect of the cerebellar tonsils. This leads to syrinx forma­tion, probably through a combination of CSF being forced into the spinal cord through perivascular channels in the cord itself and the forced rostral and caudal movement of fluid within the cord as a result of the spinal subarachnoid pressure waves ( 6). Debate continues to surround basic concepts of CSF reabsorption. Although traditional teaching is that CSF re-absorption occurs at the arachnoid granulations, experi­mental animal evidence suggests that at least a portion of CSF may drain through the lymphatic system of the head and neck, particularly in the region of the cribriform plate and the olfactory bulb, as well as through the cranial and spinal dural root sleeves ( 7). However, the substantial vari­ability in mammalian CSF physiology has cast doubt on the applicability of animal models to humans ( 8). TREATMENT OF HYDROCEPHALUS The CSF shunt remains the most used treatment op­tion for hydrocephalus. It is relatively safe with a proce­dural mortality of 0.1% ( 9) and 0.13% ( based on death of one of 731 patients treated in two recent randomized trials) 146 J Neuro- Ophthalmol, Vol. 24, No. 2, 2004 Cerebrospinal Fluid Diversion Procedures JNeuro- Ophthalmol, Vol. 24, No. 2, 2004 ( 10,11). However, the complication rate, in terms of shunt failure, remains high. For children undergoing first shunt insertion, one- third can expect shunt failure within the first year, and just more than half will require shunt revision within two years ( 10,11). Those with primary aqueductal stenosis may fare better ( 12). Endoscopic Third Ventriculostomy To avoid the problems of shunt failure, endoscopic third ventriculostomy ( ETV), in which the floor of the third ventricle is surgically opened using an endoscope placed within the ventricular system through a burr hole, has be­come the initial treatment of choice for certain forms of ob­structive hydrocephalus, such as aqueductal stenosis. It is being attempted with varying success in conditions often not thought to be based on intraventricular obstruction. These conditions include CSF shunt obstruction and infec­tion ( 13,14) and hydrocephalus in which the obstruction is considered either at the outflow of the fourth ventricle or at the arachnoid spaces around the brainstem caudal to the prepontine cistern to which the third ventricle is being opened. In either of these two situations, ETV may be indi­cated even though all four ventricles are dilated, suggesting communicating hydrocephalus ( 15). Other conditions for which ETV has been performed include hydrocephalus as­sociated with Chiari I malformation ( 16) and normal pres­sure hydrocephalus ( 17,18). Success rates for ETV depend, predictably, on the mechanism of the hydrocephalus, with rates as high as 65% to 70% for patients undergoing the procedure for aqueduc­tal stenosis ( 19- 21), but only 23% o for neonates with hydro­cephalus from an intraventricular hemorrhage ( 22). Most failures occur soon after the procedure is performed, al­though there has been at least one reported failure six years after the procedure ( 21). The mortality rate after ETV has been estimated at l% o ( 23). Complications include sub­arachnoid hemorrhage, basilar artery injury, and hypothala-mic/ pituitary injury. There is yet no solid evidence of sub­stantial benefit over conventional shunting in the initial few years after treatment; longer follow- up studies will be nec­essary to prove long- term benefit ( 24,25). Despite the promise of ETV, many patients are not candidates for the procedure. Strenuous efforts continue to be aimed at improving the longevity of conventional CSF shunt systems. Endoscopic Shunt Insertion The neuroendoscope has also been applied to CSF shunt procedures in an attempt to position the intraventricu­lar catheter at a distance from the choroid plexus to avoid proximal catheter obstruction. Kestle et al ( 17) tested this proposition in a controlled trial in which 393 patients were randomized to undergo their first shunt placement with or without the neuroendoscope. The neuroendoscope patients had a 42% o one- year shunt failure rate as compared with 34% among controls. On follow- up computed tomography scanning, the two groups had similar rates of choroid plexus avoidance. Thus, although the premise that choroid plexus avoidance will improve shunt survival remains untested, neuroendoscopy evidently does not provide a simple av­enue to achieve this goal ( 11). CSF Shunt Valves CSF shunt systems usually include a valve to regulate the drainage of CSF. Several innovative shunt valves have come into more widespread use in the past five years. Al­though the most common finding at surgical exploration of a failed shunt is a blocked ventricular catheter, this often appears to be the result of excess CSF drainage and collapse of part of the ventricular system around the catheter. Alter­natively, overdrainage of CSF can lead to a uniformly small ventricular system and symptoms related to insufficient CSF volume. The valve in the CSF shunt system plays the critical role in regulating CSF outflow. Shunt vale design modification has thus been focused on limiting these com­plications. Valve systems are grouped within four broad catego­ries: 1) differential pressure valves ( often called " standard" valves); 2) antisiphon valves; 3) flow- regulated valves; and 4) adjustable valves. Differential Pressure Valves These function by requiring a pressure gradient across the valve to prompt CSF drainage. Different mecha­nisms have been used to achieve this. For example, in one design, CSF must push apart a thin slit in a silicone mem­brane to pass through the valve. In another, CSF must push up a small ruby ball against a spring to pass the valve ( Fig. 1). When tested on the bench top, these mechanisms main­tain a stable pressure with increasing flow rates through the valve. In the clinical situation, in which the ICP is variable, flow rates through these valves increase with increasing ICP. These valves are supplied in varying fixed opening Valve open FIG. 1. Standard differential pressure shunt valve, ball-spring mechanism. The valve pressure is determined by re­sistance of the spring to deformation. Reprinted with per­mission. 147 JNeuro- Ophthalmol, Vol. 24, No. 2, 2004 Garton pressures. Crudely, the higher the opening pressure of the valve, the less the drainage for a given range of ICPs. Antisiphon Valves Overdrainage of CSF is caused in part by a siphoning effect through the shunt system. When the patient is not supine, the distal portion of the ventricular CSF shunt sys­tem is almost always below the level of the ventricles. This creates the potential for a siphon effect, increasing the flow through the shunt system even at low ICPs. Antisiphon valves use a differential pressure valve system but add an additional antisiphon component. This can involve a sili­cone membrane that collapses into and blocks the distal tub­ing whenever atmospheric pressure ( transmitted through the skin) is sufficiently higher than pressure in the distal tubing ( Fig. 2). A second device uses a small weight and gravity to rotate the internal parts of the valve when the patient's position changes from horizontal to vertical, di­recting CSF flow through different pressure components of the valve for the two positions. Flow- Regulated Valves CSF production does not vary with ICP, at least at typical ICP values. One way to reduce the overdrainage seen with differential pressure valve systems is to drain CSF at a constant flow rate regardless of pressure. This may be more analogous to normal drainage physiology, in which certain events that increase ICP ( coughing, straining) also increase venous pressure, narrowing the pressure difference and reducing CSF drainage. Flow- regulated valves use this principle by allowing ventricular pressure changes to vary the size of the drainage aperture of the shunt valve inversely with respect to pressure. The larger the pressure, the smaller the aperture ( Fig. 3). Balanced properly, the flow rate re­mains constant over the range of physiologic ICP. At very high pressures of CSF, these valves convert to differential pressure systems as a safety mechanism. Adjustable Valves One of the challenges in CSF shunting is determining which pressure will best suit a patient's needs. For example, . Silicone Diaphragm i - i ; I i ~ z ^ ^ \ . Silicone Diaphragm closed by negative distal pressure FIG. 2. Antisiphon system. Deformable silicone diaphragm closes when the distal pressure in the tubing drops below the pressure outside the valve housing. Reprinted with per­mission. Diaphragm valve Low Pressure High Pressure Higher Pressure Low Resistance High Resistance |_ ow Resistance ( safety mechanism) FIG. 3. Flow- controlled valve ( Orbis, Sigma). Variations in pressure deform a membrane that moves with respect to a wedge- shaped pin. At low pressures, a large aperture for CSF flow is present. At higher pressures the membrane moves to narrow the aperture, limiting the increase in flow that would otherwise occur with the increased pressure. At even higher pressures, the membrane deflects beyond the pin, increasing the aperture and flow, active as a safety release mechanism. Reprinted with permission. a relatively low pressure valve might be desirable in an in­fant, where normal physiologic ICP values are in the 3 to 5 mm Hg range. With the child's growth, a higher pressure valve would be better to avoid overdrainage. Reoperation to change valve pressure carries a morbidity of shunt infection and malfunction. Adjustable shunt valves allow for differ­ential pressure adjustments transcutaneously, usually with the use of a magnet to rotate internal components of the valve mechanism to a new position that increases or de­creases resistance ( Fig. 4) One potential complication of this type of shunt valve is accidental valve adjustment caused by environmental exposure to magnetic fields. Ret­rospective evidence suggests that this phenomenon, al­though rare ( 26), may occur even with household magnets in the range of 0.1 tesla if the magnet is placed within 1 cm of the valve ( 27). Despite solid theory and bench- top data, there is still no class I evidence that these new valves accomplish then-stated purpose. Drake etal ( 10) were the first to compare the different categories of CSF shunt valves in a randomized fashion. Their study found no differences in shunt survival between patients treated with a standard differential pres­sure valve, an antisiphon- equipped valve, or a flow- regu­lated shunt valve. The study enrolled patients undergoing first shunt placement, and therefore involved a high per­centage of young children. Even at four years after shunt insertion, there was no difference in outcomes between the 148 © 2004 Lippincott Williams & Wilkins Cerebrospinal Fluid Diversion Procedures JNeuro- Ophthalmol, Vol. 24, No. 2, 2004 Top View Q ^ Ruby Ball Rotation alters spring configuration and resistance FIG. 4. Adjustable valve ( Codman- Medos). Adjustable cam is turned by externally applied magnetic force, al­tering the position of the spring on the stepper mechanism, increasing or decreasing tension within the spring, altering the resistance of the ruby ball to movement by CSF pressure. Re­printed with permission. Side View three types of valves ( 28). In a large prospective study of children and adults with hydrocephalus treated with the OSV II flow- regulated valve ( 29), the shunt failure rate at two years was only 33%, considerably lower than the rate typically seen in patients with standard differential pressure valves. However, this study ( 29) lacked a control group and used different outcome measures than have been used in the more robust clinical trials. Among the many commercially available adjustable valves ( made by Strata, Sophy, and Codman- Medos), only the Codman- Medos valve has been subjected to a random­ized comparison to other valve types ( 12). In that random­ized study of 235 patients undergoing first implant and 142 patients undergoing revisions, the adjustable valve and con­trol group experiences were similar over the two- year study period. Valve removals occurred in 32% versus 39%, and overall shunt survival rates were near 50% for initial shunts and 43% for revision shunts, regardless of treatment arm. The overall infection rate was 9.8%. Other outcomes that were similar between the treatment groups were patient sense of wellness and persistent ventriculomegaly ( 24% with adjustable valves, 18% with standard valves) ( 12). Case series reports in children have also yet to demonstrate an advantage of the adjustable devices over older, less ex­pensive designs ( 30). Given these negative randomized trials, what prin­ciples can be applied to the selection of the CSF shunt valve for a given patient? If one is using a differential pressure system, the highest pressure valve that still improves the patient's condition is generally the most desirable. Many of the long- term complications of CSF shunt placement ap­pear to relate to overdrainage of CSF, including restricted head growth with cranio- cerebral disproportion, slit ven­tricle syndrome, and overdrainage headache. Knowledge of the patient's typical ICPs may help in judging how much pressure will be tolerated. Patients with traumatic brain in­jury and subarachnoid hemorrhage may require relatively less resistance to achieve adequate drainage. Patients with venous outflow obstruction in achondroplasia, and those with IIH might benefit from high- pressure valves to retain sufficient intracranial CSF for compensation to changing intracranial volume conditions and to limit collapse of the ventricular system around the shunt catheter. Flow- regu­lated valves may be appropriate in any of these conditions and generally will drain less CSF than medium to low pres­sure differential pressure systems. Antisiphon systems are also routinely used by some practitioners; they are appro­priate when the siphoning pressure will be high ( tall adults) or when the distal pressure will be low ( ventriculopleural shunts, because the pleural space has subatmospheric pres­sure during inspiration). The ability of adjustable valves to modify CSF outflow has advantages in certain other situa­tions. For example, very young children undergoing shunt insertion are at higher risk for shunt infection than are older patients ( 29,31). One potential reason for this is that chil­dren with very thin skin are more likely to leak CSF out around a shunt system that puts up resistance to CSF drain­age. To avoid this early after surgery, heavy drainage pro­moted by a low- pressure valve setting can keep ICP very low, whereas more standard drainage patterns can be achieved by raising the shunt valve pressure once incisions have healed ( 32,33). If no other factors are considered, dif­ferential pressure valves are clearly the least expensive, be­ing nearly 10 times lower in cost than the current generation of adjustable valves. Antibiotic- Impregnated Catheters Antibiotic- impregnated catheters have recently been used to reduce the rate of infection that occurs in 7% to 12% 149 JNeuro- Ophthalmol, Vol. 24, No. 2, 2004 Garton TABLE 1. Outcomes after optic nerve sheath fenestration Corbett, 1988 ( 48) Sergott, 1988 ( 64) Brourman, 1988 ( 50) Pearson, 1991 ( 39) Kelman, 1992 ( 52) Spoor, 1993 ( 47) Acheson, 1994 ( 46) Herzau, 1998 ( 49) Goh, 1997 ( 53) Banta, 2000( 51) Pts 28 23 6 9 17 54 11 14 19 86 Eyes 40 29 10 14 21 75 15 23 29 158 Mean follow- up ( mo) 27 24 < 12 11 17 22 32 62 16 20 Visual acuity improved post- op (* f J) ( by # of eyes) 12/ 21 ( 57.1%) 19/ 21 ( 90.5%) 3/ 4 ( 75%) NR 12/ 14 ( 85.7%) 14/ 22 ( 63.6%) 7/ 10 ( 70%) 5/ 18( 27%) 4/ 13 ( 30.8%) NR Visual acuity improved or stable post- op (*}) ( by # of patients) 23/ 28( 82.1%) 21/ 23 ( 91.3%) 6/ 6 ( 100%) NR 15/ 17( 88.2%) 52/ 54 ( 96.3%) 9/ 11( 81.8%) 12/ 14 ( 86%) 17/ 19 ( 89.5%) NR Visual acuity improved or stable post- op (*}) ( by # of eyes) 32/ 40 ( 80%) 28/ 29 ( 96.6%) 10/ 10 ( 100%) NR 19/ 21 ( 90.5%) 73/ 75 ( 97.3%) 13/ 15 ( 86.7%) 20/ 23 ( 87%) 26/ 29 ( 89.7%) 148/ 153 ( 93.7%) Visual field improved post- op (* fj) ( by # of eyes) 22/ 28 ( 78.6%) 29/ 29 ( 100%) 10/ 10 ( 100%) 6/ 14 ( 42.9%) 20/ 21 ( 95.2%) 33/ 68 ( 48.5%) 9/ 15 ( 60%) 10/ 21 ( 48%) 16/ NR NR Visual field improved or stable post- op (*}) ( by # of patients) 23/ 28( 82.1%) 23/ 23 ( 100%) 6/ 6 ( 100%) 8/ 9 ( 88.9%) 16/ 17( 94.1%) 48/ 51( 94.1%) 9/ 11( 81.8%) 12/ 14 ( 86%) NR 71/ 81 ( 87.7%) of shunt operations ( 34,35). Perioperative antibiotics re­duce the rate of infection only when the native infection rate is at the upper limit of the usual range (> 15%>) ( 35). Randomized testing of the use of antibiotic- impregnated catheters for CSF shunts has not yet been reported. How­ever, a recent trial of their use in external ventricular drain­age demonstrated a significant ( 18% vs 37%>) reduction in bacterial catheter colonization and in positive CSF culture rates ( 1.3%> vs 9.4%>) ( 36). Controlled studies are underway. TREATMENT OF IDIOPATHIC INTRACRANIAL HYPERTENSION Intracranial Pressure Monitoring The diagnosis and treatment of IIH generally requires that the ICP as measured by lumbar puncture be more than 25 cm H20. There are patients whose symptoms and signs are suggestive of elevated ICP but whose lumbar CSF pres­sures are within the normal range, borderline, or variable, or in whom lumbar CSF pressure measurement is technically difficult. In other cases, the clinical manifestations are equivocal, yet lumboperitoneal ( LP) opening pressures are elevated. In these situations, direct ICP monitoring may be considered. The procedure allows continuous telemetry and may identify periodic elevations in pressure not identified on a one- time assessment. Using a burr- hole momtor that did not require cannulation of the ventricle, Gucer et al ( 37) demonstrated pressures ranging from 100 to 500 mm H20 in IIH patients. Although clearly more invasive than the lumbar puncture, the risk of clinically significant hemor­rhage with current similar techniques has been reported to be in the range of 0.38% o to 0.7% o ( 38), offering a favorable risk- benefit ratio in challenging cases. Surgery Although medical therapies are generally successful, surgical therapies are indicated for patients with symptoms TABLE 2. Outcomes after CSF shunt Pts Eyes Mean follow- up ( mo) 36 31 77 35 9 Visual acuity improved post- op (* f J) ( by # of eyes) 1/ 5 ( 20.0%) 28/ 68 ( 41.2%) b 16/ 28 ( 57.1%) b 10/ 14( 71.4%) 8/ 12 ( 66.7%) Visual acuity improved or stable (*}) ( by # of patients) NR 29/ 37 ( 78.4%) 14/ 14 ( 100.0%) b 16/ 17( 94.1%) 7/ 7 ( 100.0%) Visual acuity improved or stable post- op (*}) ( by # of eyes) 27/ 27( 100.0%) 60/ 74 ( 81. l%) b 28/ 28 ( 100.0%) b 33/ 34( 97.1%) 12/ 12( 100.0%) Visual fields improved post- op (* fj) ( by # of eyes) 2/ 24 ( 8.3%) b b 18/ 28 ( 64.3%) NR Visual fields improved or stable post- op (*}) ( by # of patients NR b b 17/ 17 ( 100.0%) NR Cornblath, 1989 ( 41) 18 36 Rosenberg, 1993 ( 43) 37 74 Eggenberger, 1996 ( 44) 27 54 Burgett, 1997 ( 40) 30 60 Tulipan, 1998 ( 45) 7 14 150 © 2004 Lippincott Williams & Wilkins Cerebrospinal Fluid Diversion Procedures JNeuro- Ophthalmol, Vol. 24, No. 2, 2004 TABLE 1. Continued Visual field improved or stable post- op (*}) ( by # of eyes) 31/ 38( 81.6%) 29/ 29 ( 100%) 10/ 10 ( 100%) 13/ 14 ( 92.9%) 20/ 21 ( 95.2%) 68/ 75 ( 90.7%) 13/ 15 ( 86.7%) 20/ 23 ( 87%) 23/ 28( 82.1%) 71/ 81 ( 87.7%) Delayed visual deterioration (*}) ( by # of patients) 1/ 28 ( 3.6%) 2/ 23 ( 8.7%) 2/ 6 ( 33.3%) 0/ 10 ( 0%) NR 20/ 54 ( 37%) 3/ 11( 27.3%) 2/ 14 ( 14.3%) 2/ 9 ( 22.2%) 11/ 81( 13.6%) Delayed visual deterioration (* § ) ( by # of eyes) 1/ 40 ( 2.5%) 2/ 29 ( 6.9%) 2/ 10 ( 20%) 0/ 14 ( 0%) NR 24/ 75 ( 32.0%) 3/ 15 ( 20%) 3/ 23 ( 13%) 2/ NR 16/ 158( 10.1%) Shunt rate ( 1) 3.6% NR NR NR 0.0% 7.4% 36.4% 14.2% NR 16.3% Improved headache 11/ 17( 64.7%) 9/ 19 ( 47.4%) 2/ 3 ( 66.7%) NR 9/ 10 ( 90%) NR NR 6.9 ( 66%) NR 8/ 61 ( 13.1%) Optic nerve injury (*) ( by # of eyes) 7.5% 0.0% 0.0% 0.0% 4.8% 0.0% 0.0% 0.0% 0.0% 0.6% Diplopia 3.6% 0.0% 0.0% NR 0.0% NR NR 0.0% NR 4.7% Pts, patients; NR, not reported; post- op, postoperative; #, number. * Data are presented both by eyes and by patients to allow for comparison with CSF shunt outcome data. f Denominator refers to patients with abnormal visual acuity/ visual field before ONSF; numerator refers to patients who improved postoperatively. % For patients undergoing multiple ONSF, final outcome is used regardless of the number of procedures performed. § Delayed deterioration includes all patients deteriorating after apparent period of stability, including those who subsequently improved with additional procedures. 1 Shunt rate = % of patients undergoing CSF shunt placement after ONSF. refractory to medical management. This occurs in an esti­mated 18% to 22% of IIH patients ( 39- 41). Surgical options include optic nerve sheath fenestration ( ONSF) and CSF shunt. A recent Cochrane systematic review identified no randomized controlled trials comparing these treatments to each other or to medical management ( 42). The CSF shunt appears to come closer to correcting the root cause of the problem, whereas ONSF focuses on protecting the vulnerable optic nerve head. Clinical out­come data comparing CSF shunt and ONSF are limited to competing case series ( Tables 1- 3). Comparisons are diffi­cult because the outcome measures are often generalized to " visual symptoms" in case series of treatment with CSF shunts, while being quite specific about visual function measures in ONSF case series. ONSF series typically report results in terms of the number of eyes treated, whereas shunt series report number of patients treated. Better outcomes tend to be reported after both types of surgeries in which follow- up is short. ONSF and shunt procedures have changed over time and complication rates in early series may not reflect current results. Shunt treatment appears to be successful in halting deterioration or improving visual manifestations in 78% to 100% of patients ( Table 2) ( 40,43- 45). This is at least com- TABLE 2. Continued Visual fields improved or stable post- op (*}) ( by # of eyes) 36/ 36 ( 100.0%) b b 34/ 34 ( 100.0%) NR Pts, patients; b, Delayed visual deterioration (*}) ( by # of patients) NR 11/ 37( 29.7%) NR NR 0/ 7 ( 0.0%) reported as " visual change,' By eye delayed visual deterioration (* § ) ( by # of eyes) NR 20/ 74 ( 27.0%) NR NR 0/ 14 ( 0.0%) Improved headache NR NR 18/ 18( 100.0%) 3/ 4 ( 75.0%) 6/ 7 ( 85.7%) ' designating a combination of visual acuity and Optic nerve injury NR NR 0% 0% 0% visual field outcomes; Diplopia NR NR 0% 0% 0% NR, not reported. By person abducens palsy resolved NR NR 5.5 ( 100%) NR NR 151 JNeuro- Ophthalmol, Vol. 24, No. 2, 2004 Garton TABLE 3. CSF shunt- related Comblath, 1989 ( 41) Rosenberg, 1993 ( 43) Eggenberger, 1996 ( 44) Burgett, 1997 ( 40) Tulipan, 1998 ( 45) Pts 18 37 27 30 7 complications Eyes 36 74 54 60 14 Mean follow- up ( mo) 36 31 77 35 9 Shunt procedures performed Lumbar: 34 Ventricular: 0 Lumbar: 73 Ventricular: 10 Lumbar: 93 Ventricular: 0 Lumbar: 156 Ventricular: 0 Lumbar: 0 Ventricular: 7 Average survival of first shunt ( mo) NR Unclear 11 18 NR Mean# operations ( including placement) 1.9 2.2 3.4 5.2 1.0 Patients requiring at least one reoperation 38.9% 51.4% 55.6% 60.0% 0.0% Revisions per shunt-year exposure 0.30 0.48 0.38 1.44 0 parable with the case series success rates of 73% to 100% for ONSF ( Table 1) ( 46- 53). Failure rates up to 30% have been reported for shunt ( 40,43,44) and ONSF treatment ( 46- 51,53,54). Control of headache appears to be slightly better with shunt treatment, although Kelman et al ( 52) re­ported a 90% rate of improvement with ONSF ( Table 1). The clear disadvantage of CSF shunt treatment is the incidence of shunt failure. Case series of patients managed with LP shunt placement report reoperation rates of 44% to 63% over variable time periods ( Table 3) ( 40,43,44). How­ever, on closer review, several patterns with respect to shunt revisions in these patients are apparent. First, relatively few patients are responsible for the majority of revisions. For example, in the series of Burgett et al ( 40), two ( 7%) of 30 patients accounted for more than 50% of revisions. Thus, the average number of procedures is poorly reflective of the typical experience. Still, more than half the patients appear to require at least one reoperation. Another confounding factor is that the indications for shunt revision are variable. Persistent or recurrent headache has often been an indica­tion for shunt revision. Many clinicians would reserve sur­gical revision for patients with evidence of intracranial hy­pertension and visual deterioration. In case series, persistent headache appears to account for up to 21% of " shunt failure" and for 1.6% to 28% of revisions ( 43). Overdrainage may explain some of these headaches. Newer valve technology may reduce the need to reoperate for overdrainage. Whether patients with prominent headaches and IIH are better candidates for shunt insertion than ONSF, as suggested by Banta ( 51), remains unproven. Whereas ONSF avoids long- term implant placement and its attendant problems, ONSF has its own acute com­plications. Optic nerve injury, retinal artery occlusion, and diplopia appear to be the main reported surgical complica­tions, occurring at rates of 1% to 10% and 3% to 5%, respec­tively ( Table 1). However, more recent ONSF case series generally report lower rates of complications, suggesting that procedure modifications or a learning curve are at play. Lumboperitoneal Shunts The LP route has traditionally been the method of shunting in IIH. This technique has been preferred because of the perceived difficulty in cannulating a normal or small lateral ventricle and to avoid the small risk of hemorrhage while passing a catheter through the brain parenchyma. As with other aspects of the treatment of IIH, reliable compara­tive data are unavailable. Acknowledging this limitation, the failure rate of LP shunts appears to be higher than that of ventriculoperitoneal ( VP) shunts in the typical adult popu­lation. The CSF pressure transmitted to the LP shunt valve is quite different than that transmitted to a VP shunt valve. CSF pressures measured at a lumbar- positioned valve are much higher in the sitting than supine position ( average, 490 mm Hg sitting, 140 mm Hg supine) ( 55). Siphoning is not an issue in LP shunts. Shunt valves designed to handle these pressures typically require two separate differen­tial pressure mechanisms, one for the sitting or standing position, the other for the supine position, with a gravity-operated ball valve mechanism directing flow to the appro­priate portion of the system. In addition to the usual mecha­nisms of shunt occlusion, these systems can be prone to stick in either the vertical or the horizontal position. Perhaps because of overdrainage common with all shunt systems, LP shunts can be responsible for an iatro­genic Chiari I malformation. In reviewing the pediatric experience with LP shunts ( for varied indications that in­cluded IIH), Chumas ( 56,57) noted a 70% incidence of ton­sillar herniation through the foramen magnum with criteria compatible with a diagnosis of acquired Chiari I malforma- 152 © 2004 Lippincott Williams & Wilkins Cerebrospinal Fluid Diversion Procedures JNeuro- Ophthalmol, Vol. 24, No. 2, 2004 TABLE 3. Continued Shunt revisions for headaches alone Yes No Yes Yes NR Pts, patients; Vision deteriorates despite working shunt NR 10.8% 0.0% 0.0% 0.0% NR, not reported. Revisions induced by low pressure headaches 25.0% 16.9% 15.2% 1.6% 0.0% Revisions not related to visual symptoms 100% 30.1% Unclear 100% 0.0% Patients accounting for 50% of symptoms NR 27.0% 11.1% 6.7% NR Tonsilar herniation NR NR 7.4% 0.0% NR Symptomatic tonsilar herniation NR NR 3.7% 0.0% NR tion. Six ( 4%) of 143 patients in that series became symp­tomatic and one died of hindbrain herniation. However, Re-kate ( 5 8) found no case of acquired Chiari I malformation in a 25- patient pediatric LP shunt case series. This complica­tion has also been reported in adults ( 59- 61). Johnston et al ( 61) detail the incidence of acquired Chiari I malformation in a large population of patients with medically and surgi­cally treated IIH. Two ( 4.5%) of 44 patients treated medi­cally were found to have asymptomatic Chiari I malforma­tion whereas 11 ( 14%) of77 patients treated surgically were found to have Chiari I malformation, three ( 4%) of whom were symptomatic. MRI detection of the malformation oc­curred at three to four years after shunt placement. In pa­tients with acquired Chiari malformation, a variety of treat­ment options are available, but conversion to a ventricular or cisternal shunt is a reasonable first step ( 61). The true likelihood of this complication in the IIH population re­mains unclear without well- designed prospective studies, but vigilance for the symptoms of foramen magnum com­pression and associated syringomyelia are warranted, par­ticularly in children in whom the incidence of an acquired Chiari malformation may be higher ( 56, 62). Ventriculoperitoneal Shunts VP shunts are a viable alternative to LP shunts in treatment of IIH. Operatively, VP shunt placement is easier than LP shunt placement, which requires a lateral position, particularly in obese patients. Cannulation of even a small frontal horn can generally be performed using standard ex­ternal landmarks or with simple assist devices ( 63). If head shape or ventricular anatomy suggest that standard maneu­vers for ventricular cannulation may be insufficient, ad­vances in stereotactic surgical navigation have made it fea­sible to cannulate very small ventricular systems reliably. In a small case series with short follow- up, Tulipan ( 45) re­ported stereotactic VP shunt placement in seven patients with IIH. All had improvement in optic disc swelling, and six had improvement in headache without need for reopera­tion. VP shunts avoid the risk of inducing a Chiari I malfor­mation, and may be less likely to overdrain. A wider range of shunt valve options is available for VP than for LP shunts, with the caveats noted about the lack of proof as to the benefits of these valves. As indicated, IIH patients may be best served by a relatively higher pressure shunt system, possibly with an antisiphon system to limit overdrainage. Adjustable shunts are particularly attractive when head­aches are a prominent symptom because a range of pres­sures can be tested for results without reoperation. These technical factors likely add up to fewer complications with the VP than LP shunt systems. Shunt Failure Regardless of the shunt system used, patients with IIH are no less prone to shunt failure than those with other diagnoses. Shunt obstruction caused by mechanical occlu­sion of the catheter or valve, or to disconnection of the shunt system, is usually met by a return of the presenting symp­toms, with progressive headaches and loss of vision being the two main clinical features reported. Unfortunately, nei­ther of these two manifestations is very specific to shunt failure, and objective measures of shunt function are needed to confirm malfunction. Because patients with IIH typically show no change in ventricular size when ICP increases, brain imaging is not useful in suspected shunt malfunction. A plain x- ray of the shunt can show a disconnection but does little to rule out the possibility of catheter obstruction. ICP measurements are thus often indicated, by shunt tap, 153 JNeuro- Ophthalmol, Vol. 24, No. 2, 2004 Garton lumbar puncture, or, in certain circumstances, ICP monitor­ing. Shunt infection is usually suspected on the basis of symptoms of shunt obstruction with the addition of fever, meningismus, and abdominal pain ( or other symptoms spe­cific to the terminal site of the shunt system). Infection is most likely to occur within three months of a previous shunt operation, becoming relatively rare beyond 12 months after the procedure. A shunt tap or lumbar puncture can confirm the diagnosis. Overdrainage is another important complication of CSF shunts. Headache is the most common symptom. Clas­sically, there is a strong positional component to the head­ache, akin to that seen after a lumbar puncture. Certain im­aging findings, such as subdural hematoma and Chiari malformation ( in the case of an LP shunt), lead quickly to the diagnosis. In a patient with an LP shunt and a horizon­tal- vertical valve, it is sometimes possible to identify that the valve is stuck in the wrong position. Frequently, how­ever, imaging studies are unrevealing. In this circumstance, ICP monitoring may be of value because it allows a corre­lation between intermittent symptoms and ICP and allows continuous measurement in different positions. Manage­ment options usually revolve around revising the CSF shunt valve to a higher resistance system. REFERENCES 1. Bondurant CP, Jimenez DF. Epidemiology of cerebrospinal fluid shunting. Pediatr Neurosurg 1995; 23: 254- 8. 2. Greitz D, Franck A, Nordell B. On the pulsatile nature of intracra­nial and spinal CSF- circulation demonstrated by MR imaging. Acta Radiol 1993; 34: 321- 8. 3. Egnor M, Rosiello A, Zheng L. A model of intracranial pulsations. Pediatr Neurosurg 2001; 35: 284- 98. 4. Tenti G, Sivaloganathan S, Drake JM. The synchrony of arterial and CSF pulsations is not due to resonance. Pediatr Neurosurg 2002: 37: 221- 2. 5. Karahalios DG, Rekate HL, Khayata MH, et al. Elevated intracra­nial venous pressure as a universal mechanism in pseudotumor cerebri of varying etiologies. Neurology 1996; 46: 198- 202. 6. Oldfield EH, Muraszko K, Shawker TH, et al. Pathophysiology of syringomyelia associated with Chiari I malformation of the cerebel­lar tonsils. Implications for diagnosis and treatment. J Neurosurg 1994; 80: 3- 15. 7. Johnston M, Papaiconomou C. Cerebrospinal fluid transport: a lym­phatic perspective. News Physiol Sci 2002; 17: 227- 30. 8. Bell WO. Cerebrospinal fluid reabsorption. A critical appraisal. 1990. Pediatr Neurosurg 1995; 23: 42- 53. 9. Di Rocco C, Marchese E, Velardi F. A survey of the first compli­cation of newly implanted CSF shunt devices for the treatment of nontumoral hydrocephalus. Cooperative survey of the 1991- 1992 Education Committee on the ISPN. Child's Nervous System 1994; 10: 321- 327. 10. Drake JM, Kestle JR, Milner R, et al. Randomized trial of cerebro­spinal fluid shunt valve design in pediatric hydrocephalus. Neuro­surgery 1998; 43: 294- 303. 11. Kestle JR, Drake JM, Cochrane DD, et al. Lack of benefit of endo­scopic ventriculoperitoneal shunt insertion: a multicenter random­ized trial. J Neurosurg 2003; 98: 284- 90. 12. Pollack IF, Albright AL, Adelson PD. A randomized, controlled study of a programmable shunt valve versus a conventional valve for patients with hydrocephalus. Hakim- Medos Investigator Group. Neurosurgery 1999; 45: 1399- 411. 13. Buxton N, Macarthur D, Robertson I, et al. Neuroendoscopic third ventriculostomy for failed shunts. Surg Neurol 2003; 60: 201- 4. 14. Boschert J, Hellwig D, Krauss JK. Endoscopic third ventriculos­tomy for shunt dysfunction in occlusive hydrocephalus: long- term follow up and review. J Neurosurg 2003; 98: 1032- 9. 15. Kehler U, Gliemroth J. Extraventricular intracisternal obstructive hydrocephalus- a hypothesis to explain successful 3rd ventriculos­tomy in communicating hydrocephalus. Pediatr Neurosurg 2003; 38: 98- 101. 16. Decq P, Le Guerinel C, Sol JC, et al. Chiari I malformation: a rare cause of noncommunicating hydrocephalus treated by third ven­triculostomy. J Neurosurg 2001; 95: 783- 90. 17. Schwartz TH, Yoon SS, Cutruzzola FW, et al. Third ventriculos­tomy: postoperative ventricular size and outcome. Minim Invasive Neurosurg 1996; 39: 122- 9. 18. Meier U, Zeilinger FS, Schonherr B. Endoscopic ventriculostomy versus shunt operation in normal pressure hydrocephalus: diagnos­tics and indication. Minim Invasive Neurosurg 2000; 43: 87- 90. 19. Grunert P, Charalampaki P, Hopf N, et al. The role of third ventricu­lostomy in the management of obstructive hydrocephalus. Minim Invasive Neurosurg 2003 ; 46: 16- 21. 20. Hopf NJ, Grunert P, Fries G, et al. Endoscopic third ventriculos­tomy: outcome analysis of 100 consecutive procedures [ see com­ments]. Neurosurgery 1999; 44: 795- 806. 21. CinalliG, Sainte- Rose C, ChumasP, etal. Failure of third ventricu­lostomy in the treatment of aqueductal stenosis in children. J Neu­rosurg 1999; 90: 448- 54. 22. Buxton N, Macarthur D, Mallucci C, et al. Neuroendoscopic third ventriculostomy in patients less than 1 year old. Pediatr Neurosurg 1998; 29: 73- 6. 23. Schroeder HW, Niendorf WR, Gaab MR. Complications of endo­scopic third ventriculostomy. J Neurosurg 2002; 96: 1032- 40. 24. Barlow P, Ching HS. An economic argument in favor of endoscopic third ventriculostomy as a treatment of obstructive hydrocephalus. Minim Invasive Neurosurg 1997; 40: 37- 9. 25. Garton HJ, Kestle JR, Cochrane DD, et al. A Cost- Effectiveness Evaluation of Endoscopic Third Ventriculostomy. In: Pediatric Sec­tion Meeting of the American Association of Neuological Surgeons. Indianapolis. 1998. 26. Zemack G, Romner B. Seven years of clinical experience with the programmable Codman Hakim valve: a retrospective study of 583 patients. J Neurosurg 2000; 92: 941- 8. 27. Czosnyka Z, Czosnyka M, Copeman J, et al. A randomized, con­trolled study of a programmable shunt valve versus a conventional valve for patients with hydrocephalus. Neurosurgery 2000; 47: 1250- 1. 28. Kestle J, Drake J, Milner R, et al. Long- term follow- up data from the Shunt Design Trial. Pediatr Neurosurg 2000; 33: 230- 6. 29. Hanlo PW, Cinalli G, Vandertop WP, et al. Treatment of hydro­cephalus determined by the European Orbis Sigma Valve II survey: a multicenter prospective 5- year shunt survival study in children and adults in whom a flow- regulating shunt was used. J Neurosurg 2003; 99: 52- 7. 30. Zemack G, Bellner J, Siesjo P, et al. Clinical experience with the use of a shunt with an adjustable valve in children with hydrocephalus. J Neurosurg 2003; 98: 471- 6. 31. Tuli S, Drake J, Lawless J, et al. Risk factors for repeated cerebro­spinal shunt failures in pediatric patients with hydrocephalus. J Neurosurg 2000; 92: 31- 8. 32. Korinth MC, Weinzierl MR, Gilsbach JM. Experience with a new concept to lower non- infectious complications in infants with pro­grammable shunts. Eur J Pediatr Surg 2003; 13: 81- 6. 33. Rohde V, Weinzierl M, Mayfrank L, et al. Postshunt insertion CSF leaks in infants treated by an adjustable valve opening pressure re­duction. Childs Nerv Syst 2002; 18: 702- 4. 34. Borgbjerg BM, Gjerris F, Albeck MJ, et al. Risk of infection after cerebrospinal fluid shunt: an analysis of 884 first- time shunts. Acta Neurochir ( Wien) 1995; 136: 1- 7. 35. Haines SJ, Walters BC. Antibiotic Prophylaxis for Cerebrospinal Fluid Shunts: A Metanalysis. Neurosurgery 1994; 34: 87- 93. 154 © 2004 Lippincott Williams & Wilkins Cerebrospinal Fluid Diversion Procedures JNeuro- Ophthalmol, Vol. 24, No. 2, 2004 36. Zabramski JM, Whiting D, Darouiche RO, et al. Efficacy of anti­microbial- impregnated external ventricular drain catheters: a pro­spective, randomized, controlled trial. J Neurosurg 2003; 98: 725- 30. 37. Gucer G, Viernstein L. Long- term intracranial pressure recording in the management of pseudotumor cerebri. J Neurosurg 1978; 49: 256- 63. 38. Maartens NF, Lustgarten L, Josan V, et al. How safe is twistdrill craniostomy? Br J Neurosurg 2002; 16: 290- 3. 39. Pearson PA, Baker RS, Khorram D, et al. Evaluation of optic nerve sheath fenestration in pseudotumor cerebri using automated perim­etry. Ophthalmology 1991; 98: 99- 105. 40. Burgett RA, Purvin VA, Kawasaki A. Lumboperitoneal shunting for pseudotumor cerebri. Neurology 1997; 49: 734- 9. 41. Cornblath WT, Miller NR. Pseudotumor Cerebri Treated with Lum­boperitoneal Shunt [ Abstract]. Ann Neurol 1989; 26: 183. 42. Lueck C, Mcllwaine G. Interventions for idiopathic intracranial hy­pertension. Cochrane Database SystRev 2002: CD003434. 43. Rosenberg ML, Corbett JJ, Smith C, et al. Cerebrospinal fluid di­version procedures in pseudotumor cerebri. Neurology 1993; 43: 1071- 2. 44. Eggenberger ER, Miller NR, Vitale S. Lumboperitoneal shunt for the treatment of pseudotumor cerebri. Neurology 1996; 46: 1524- 30. 45. Tulipan N, Lavin PJ, Copeland M. Stereotactic ventriculoperitoneal shunt for idiopathic intracranial hypertension: technical note. Neu­rosurgery 1998; 43: 175- 7. 46. Acheson JF, Green WT, Sanders MD. Optic nerve sheath decom­pression for the treatment of visual failure in chronic raised intra­cranial pressure. J Neurol Neurosurg Psychiatry 1994; 57: 1426- 9. 47. Spoor TC, McHenry JG. Long- term effectiveness of optic nerve sheath decompression for pseudotumor cerebri. Arch Ophthalmol 1993; 111: 632- 5. 48. Corbett JJ, Nerad JA, Tse DT, et al. Results of optic nerve sheath fenestration for pseudotumor cerebri. The lateral orbitotomy ap­proach. Arch Ophthalmol 1988; 106: 1391- 7. 49. Herzau V, Baykal HE. Long- term outcome of optic nerve sheath fenestration in pseudotumor cerebri ( in German). Klin Monatsbl Augenheilkd 1998; 213: 154- 60. 50. Brourman ND, Spoor TC, Ramocki JM. Optic nerve sheath decom­pression for pseudotumor cerebri. Arch Ophthalmol 1988; 106: 1378- 83. 51. Banta JT, Farris BK. Pseudotumor cerebri and optic nerve sheath decompression. Ophthalmology 2000; 107: 1907- 12. 52. Kelman SE, Heaps R, Wolf A, et al. Optic nerve decompression surgery improves visual function in patients with pseudotumor cerebri. Neurosurgery 1992; 30: 391- 5. 53. Goh KY, Schatz NJ, Glaser JS. Optic nerve sheath fenestration for pseudotumor cerebri. JNeuroophthalmol 1997; 17: 86- 91. 54. Mauriello JA, Jr, Shaderowfsky P, Gizzi M, et al. Management of visual loss after optic nerve sheath decompression in patients with pseudotumor cerebri. Ophthalmology 1995; 102: 441- 5. 55. Magnaes B. Body position and cerebrospinal fluid pressure. Part 2: clinical studies on orthostatic pressure and the hydrostatic indiffer­ent point. J Neurosurg 1976; 44: 698- 705. 56. Chumas PD, Armstrong DC, Drake JM, et al. Tonsillar herniation: the rule rather than the exception after lumboperitoneal shunting in the pediatric population. J Neurosurg 1993; 78: 568- 73. 57. Chumas PD, Drake JM, Del Bigio MR. Death from chronic tonsillar herniation in a patient with lumboperitoneal shunt and Crouzon's disease. Br J Neurosurg 1992; 6: 595- 9. 58. Rekate HL, Wallace D. Lumboperitoneal shunts in children. Pedi-atr Neurosurg 2003; 38: 41- 6. 59. Hart A, David K, Powell M. The treatment of ' acquired tonsillar herniation' in pseudotumor cerebri. Br J Neurosurg 2000; 14: 563- 5. 60. Sullivan LP, Stears JC, Ringel SP. Resolution of syringomyelia and Chiari I malformation by ventriculoatrial shunting in a patient with pseudotumor cerebri and a lumboperitoneal shunt. Neurosurgery 1988; 22: 744- 7. 61. Johnston I, Jacobson E, Besser M. The acquired Chiari malforma­tion and syringomyelia following spinal CSF drainage: a study of incidence and management. Acta Neurochir ( Wien) 1998; 140: 417- 28. 62. Cinciripini GS, Donahue S, Borchert MS. Idiopathic intracranial hypertension in prepubertal pediatric patients: characteristics, treat­ment, and outcome. Am J Ophthalmol 1999; 127: 178- 82. 63. O'Leary ST, Kole MK, Hoover DA, et al. Efficacy of the Ghajar Guide revisited: a prospective study. J Neurosurg 2000; 92: 801- 3. 64. Sergott RC, Savino PJ, Bosley TM. Modified optic nerve sheath decompression provides long- term visual improvement for pseudo­tumor cerebri. Arch Ophthalmol 1988; 106: 1384- 90. 155