Received: 10 March 2000 Abstract This study is a retrospec-tive analysis of intracranial pressure(ICP) and cerebral perfusion pres-sure (CPP) data from 56 childrenwith active hydrocephalus and cere-brospinal fluid (CSF) shunt mal-function. The pressures were mea-sured from a separately sited CSFaccess device placed in the frontalhorn of the lateral ventricle. Of thepatients, 79% had an elevated ICP(mean 20±12 mmHg). A subgroupof patients demonstrated ten differ-ent forms of CSF-filled swelling.This group had significantly lowerICP recordings (P=0.000075) with amean ICP of 8.5 mmHg comparedwith the remainder (22.9 mmHg).This pressure ‘compensation’ wasbecause of additional nonphysiolog-

ical accommodation of CSF volume.Overall the CPP was normal in 35% of cases despite normal ICP oc-curring in only 11% of cases. TheCPPs were not significantly differ-ent in those with and without com-pensation. Measurement of ICP may not always be a reliable indica-tor of shunt malfunction in shunt-dependent children who present with compensatory CSF-filled spaces.

Keywords Hydrocephalus · Shuntmalfunction · Intracranial pressure · Cerebral perfusion pressure · Pressure compensation

Child’s Nerv Syst (2001) 17:52–57© Springer-Verlag 2001 O R I G I N A L PA P E R

C.E. GilkesA.J.W. SteersR.A. Minns

Pressure compensation in shunt-dependenthydrocephalus with CSF shunt malfunction

Introduction

The intracranial pressure (ICP) is often used as the mostreliable indicator of ventriculoperitoneal shunt malfunc-tion in active hydrocephalus [2, 6].

In reviewing our cases of shunt malfunction over a 3-year period we found that a third of the patients had anextracranial fluid collection. Our theory is that such acollection acts as a buffer and tends to normalise the ICP.Therefore, we aim to describe the ICP and cerebral per-fusion pressure (CPP) during shunt malfunction with andwithout coexisting clinical evidence of pressure compen-sation.

Patients and methods

Patients

It is the practice at our unit to insert separate reservoirs in thefrontal horn of the lateral ventricle in conjunction with CSF shunt-ing in the management of progressive hydrocephalus. This pro-vides independent access to the CSF for ICP measurement andCSF evaluation to elucidate shunt blockage or infection.

There are situations where the ICP measurement is known tobe less useful in deciding on shunt patency e.g. IV ventricularshunt malfunction which may not be reflected in pressure mea-surements from the lateral ventricles [1], and therefore we haveincluded only those cases of hydrocephalus with a ventriculoperi-toneal shunt where the measured ICP would normally be reliedupon to decide if the shunt were functioning. This applied to 56 cases over a 3 year period, with pressure recordings and addi-tional blood pressure measurements (Dynamap) being availablein 48 of these patients, allowing the calculation of the CPP to bemade.

C.E. Gilkes · R.A. Minns (✉ ) Department of Paediatric Neurosciences,Royal Hospital for Sick Children, 9 Sciennes Road,Edinburgh EH9 1LF, UKe-mail: Robert.Minns@ed.ac.ukTel.: +44-131-5360631/7Fax: +44-131-5360635

A.J.W. SteersDepartment of Clinical Neurosciences,Western General Hospital, Edinburgh, UK

53

Definitions

A shunt malfunction was defined not only by raised intracranialpressure measurement but also by clinical and radiological fea-tures and confirmation of the need for surgery i.e. these cases wereshunt-dependent with acute shunt malfunction requiring urgentsurgical revision.

We use the term ‘compensation’ to refer to cases where an al-ternative nonphysiological potential space is accommodating CSFvolume and pressure.

Operating details

The shunt insertions and revisions were mostly all performed byone neurosurgeon. A parietal approach was routinely used and theadministration of three perioperative doses of Flucloxacillin wasstandard practice. Of the shunts used, 93% were medium pressureMedtronic unitised systems.

Statistical methods

The Mann-Whitney U-test was used for statistical comparison.

Results

Compensatory mechanisms

Ten different compensatory mechanisms were recogni-sed amongst the 56 cases of shunt malfunction. Thesemechanisms were a mixture of intracranial and extracra-nial buffers as listed in Table 1 and illustrated in Figs. 1,2, 3 and 4.

Intracranial pressure measurements

The ICP measurements (opening pressure) from 56 eval-uations on patients aged between 2 weeks and 17 yearsare shown in Fig. 5. The upper limit of normal ICP [8] isadded for reference. The overall mean opening ICP was20±12 mmHg. Of the patients, 78.6% had elevated ICPoverall. Of those cases with malfunction and no compen-sation, 84.4% had an elevated ICP. It can be seen that 12measurements (21.4%) were on or below the ‘normal’

line despite acute shunt dysfunction. Of these five hadclinical evidence of a compensatory mechanism. A fur-ther four patients with compensation had opening pres-sures marginally above the normal upper limit of ICPand two had definite pressure elevation.

The mean ICP for the group without compensationwas 22.9 mmHg (n=45) and for the group with compen-sation was 8.5 mmHg (n=11). There was an extremelysignificant difference between the pressures in the twogroups (P=0.000075), Mann-Whitney U-test.

Pathological pressure waveforms were documented innine cases (five with A-waves, three with B-waves andone with C-waves). No such waveforms were evident inthe group with compensation but of the seven individualswith near-normal pressure and no evidence of compensa-tion, four had pathological waveforms.

Evidence for dynamic compensation is seen in the se-quential ICP measurements from a child with a large po-tential compensatory subgaleal space (Fig. 6a). Thisspace expanded over the time period with a correspond-ing fall in the opening pressures (Fig. 6b).

Cerebral perfusion pressure measurements

Simultaneous ICP and blood pressure measurementswere available in 48 cases allowing the calculation ofthe CPP (CPP = mean arterial pressure (MAP)–ICP) tobe made. Figure 7 illustrates the calculated CPP relativeto the estimated normal CPP for age, calculated fromthe physiological measurements of MAP over this age-range [4, 10], minus the estimated upper limit of ICPover the same age-range (P.A Jones, V.J. Easton, P.J.D.Andrews, S. Ali, R.A. Minns, manuscript submitted),[8].

The mean CPP was 54.3±12.75 mmHg and 65% ofpatients had a low CPP overall. The mean CPP in thenoncompensated group was 53.7±12.9 mmHg and the mean CPP in the compensated group was 57.4±12.5 mmHg. In those cases with compensation, 60%,(n=5) had preservation of CPP. In those cases withoutclinical compensation, 30% (n=12) had unimpaired

Table 1 Compensatory mecha-nisms (CSF cerebrospinal fluid)

Compensatory mechanism

1 Abdominal pseudocyst2 Abdominal pseudocyst with proximal CSF tracking around distal tubing3 Subcutaneous swelling around reservoir and shunt4 CSF collection and leak in subcutaneous tissue over craniotomy site5 Subcutaneous postauricular CSF collection around shunt x2 cases6 Tubular CSF swelling along the length of the extracranial portion of the proximal catheter (Fig. 1)7 Reexpansion of a previous postoperative occipital cyst (Fig. 2)8 CSF pressure expansion of hydromyelic cavitation of cervical cord (Fig. 3)9 Intracranial (subdural space) and extracranial (subcutaneous) compensation (Fig. 4)

10 Tracking of CSF around the reservoir into a previously damaged subgaleal space with nonaccidental head injury (Fig. 6)

54

Fig. 1a, b This patient was a 2-month old boy with shunted con-genital porencephalic cysts. a Illustrates how his shunt became ob-structed by omentum at the distal end. This caused an increase in in-tracranial pressure (ICP), forcing cerebrospinal fluid (CSF) to drainalongside the proximal catheter to the level of the neck. Into the res-ervoir 18 Mbq of Tc DTPA were injected. At 5 min after injection,activity was confined to the ventricles and the region of the neckswelling. b By 1 h after injection, there was also some activity inthe bladder indicating some reabsorption of CSF by the normalphysiological means; however, there was no activity in the peritone-um or along the shunt tract, indicating an obstruction to CSF flow

Fig. 2 A schematic diagram illustrating overdrainage by the ven-triculoperitoneal shunt in a 5-year old girl with previously re-moved cerebellar pilocytic astrocytoma. Subsequent ventricularpressure increases were transmitted through the IV ventricle toform a protruding collection beneath the occipital wound site

Fig. 3 A 15-year old girl with meningomyelocele and hydroceph-alus had developed a syringomyelia which required posterior fossadecompression. Later shunt degradation was followed by an ele-vated ventricular pressure being transmitted to the posterior fossawound site and to the cervical cord cavitation where it resulted inadditional signs and symptoms of extension

Fig. 4 A child of 6 weeks of age with a congenital aqueduct ste-nosis and hydrocephalus treated with a ventriculoperitoneal shuntand access device. One week previously he had his first shunt re-vision because of overdrainage and a subdural haematoma. Theventricles were therefore small at revision and easily transfixed bythe new shunt. The subsequent build-up in pressure caused the ce-rebrospinal fluid (CSF) to be diverted around the proximal cathe-ter into the new large subdural space and through the cranial bonesinto the subcutaneous space, causing a swelling in the neck

55

CPP. This difference was not statistically significant(P=0.36).

It can be seen (Fig. 7) that preservation of a normalCPP was more common in those children younger than100 months (45%), compared with older children (24%).

Discussion

In response to raised ventricular pressure a number ofphysiological buffers operate to maintain a normal intra-cranial tension. The first of these occurs in the early neo-natal period, when fluid depletion results in a decreasedbrain volume with resultant craniocerebral disproportion,which may mask early hydrocephalus [12]. Later physio-logical buffers include shunting of the CSF into the dis-tensible spinal theca, transependymal migration of CSFto the periventricular white matter, an increase of CSFabsorption at the sagittal sinus and absorption at alterna-tive sites e.g. paranasal sinuses, spinal nerve roots. Chil-dren with shunted hydrocephalus will exhaust their phys-iological compliance early in response to shunt block.We have identified ten compensatory mechanisms whichmaximise available communications between the intra-

Fig. 5 The intracranial pres-sure (ICP) measurements of 56 children with ventriculoperi-toneal shunt malfunction withrespect to the normal upperlimit of ICP in childhood. Themean ICP was 20±12 mmHg

Fig. 6a, b This shows a a T1-weighted MRI scan of a 7-week oldchild with a congenital aqueduct stenosis and previous nonacci-dental injury with scalp bruising. Upon ventriculoperitoneal shuntblockage there was compensatory cerebrospinal fluid (CSF) locu-lation around the access device and into the previously damagedsubgaleal space. b Serial intracranial pressure (ICP) measurementsfrom the same child showing a steady increase in the ICP until thesubgaleal space was established as a compensatory space. As itexpanded there was a progressive drop in the ventricular pressuremeasurements

56

cranial and extracranial spaces e.g. around the reservoiror the shunt tubing, into the subcutaneous or subgalealcompartments. CSF will follow a pressure gradient,tending to normalise the ICP until this site is also ex-hausted. A similar pressure gradient may occur through askull fracture with an underlying haematoma [7] or intothe spinal dural sac [9, 11].

The mean pressures of our patients with blockedshunts and no additional compensation was 22.9 mmHg,significantly higher (P=0.000075) than for those withcompensation (8.5 mmHg). In those cases where pro-longed pressure-recording was available, episodic wave-forms (A- and B-waves) were found only in those pa-tients with no clinical compensation. It is therefore im-portant to recognise that normal ICP is not incompatiblewith shunt malfunction where there is other evidence ofshunt dysfunction e.g. compensation sites or pathologi-cal pressure waveforms. However, the compensationmechanisms provide only a temporary buffer and onemay see a raised ICP once the buffer has been exhausted,as in two cases in our series (Fig. 5), or a dynamicchange may be observed in the ICP, rising as the bufferis exhausted or falling as the buffer is utilised (Fig. 6b).Such a buffer may delay the signs and symptoms ofraised ICP in hydrocephalus e.g. expanding head circum-ference, tense nonpulsatile fontanel, sutural separationetc. [5]. Therefore, a child with a blocked shunt maypresent initially with only a CSF-filled swelling.

There is no documented critical cerebral perfusionthreshold throughout childhood. We have calculated atheoretical level based on normal blood pressure mea-surements (MAP) and the accepted upper limit of ICPfor children of different ages [8]. The literal thresholdscalculated in this way provide isolated normal values forwhat are continuous variables. Therefore, we have joinedthe points in a continuous fashion as opposed to a step-wise progression.

Using this normal CPP limit, we found that 60% ofthose patients with compensation had maintained a nor-mal CPP compared with only 30% in those cases withoutadditional compensation. Although this did not reachstatistical significance due to the small sample sizes, alarger sample may confirm that children with compensa-tory cysts etc. may be at less risk of secondary brainischaemia. It can be seen (Fig. 7) that regardless of com-pensation, a larger percentage (35%) of children hadmaintained normal CPPs than had maintained normalICPs (11.4%), indicating that the ICP elevation withblocked shunts is often not sufficient to impair perfusion.We have also observed that CPP appears to remain nor-mal in younger children even though the ventricularpressure was often elevated by up to three times the up-per limit of normal for that age. This may reflect either(1) the ability of younger children to elevate their MAPadequately as part of the Cushing response which ispresent from the middle trimester of pregnancy [3] or (2)the relatively larger contribution that systemic pressuremakes to the normal CPP in younger children.

Whilst we still consider the measurement of ICP to bean important investigation in the management of thechild with suspected shunt malfunction, we have high-lighted that this may be normal or lower than expectedwhen the presentation is accompanied by an additionalclinical sign of compensation, such as swellings in theneck or cranium related to the shunt pathway i.e. theremay be clinical signs that account for a normal ventricu-lar pressure in CSF shunt malfunction.

Fig. 7 Cerebral perfusion pres-sure (CPP) measurements from48 children with ventriculoperi-toneal shunt malfunction withrespect to the calculated normallower level for age. The meanCPP was 54.3±12.75 mmHg

57

References

1. Cinalli G, Sainte-Rose C, Simon I, LotG, Sgouros S (1999) Sylvian aqueductsyndrome and global rostral midbraindysfunction associated with shunt mal-function. J Neurosurg 90:227–236

2. Fouyas IP, Casey AT, Thompson DHarkness WF, Hayward RD (1996)Use of intracranial pressure monitoringin the management of hydrocephalusand shunt-related problems. Neurosur-gery 38:726–731

3. Harris AP, Helou S, Traystman RJ,Jones MD Jr, Koehler RC (1998) Effi-cacy of the Cushing response in main-taining cerebral blood flow in prema-ture and near-term fetal sheep. PediatrRes 43:50–56

4. Horan MJ, Falkner B, Kimm SS (1987)Report of the second task force onblood pressure control in children. Pediatrics 79:1–24

5. Kirkpatrick M, Engleman HM, MinnsRA (1989) Symptoms and signs of pro-gressive hydrocephalus. Arch DisChild 64:124–128

6. Leggate JRS, Baxter P, Minns RA,Steers AJWS, Brown JK, Shaw JF, Elton RA (1988) Role of a separatedsubcutaneous cerebro-spinal fluid res-ervoir in the management of hydro-cephalus. Br J Neurosurg 2:327–337

7. Malek AM, Barnett FH, Schwartz MS,Scott RM (1997) Spontaneous rapidresolution of an epidural hematoma as-sociated with an overlying skull frac-ture and subgaleal hematoma in a 17month old child. Pediatr Neurosurg26:160–165

8. Minns RA, Engleman HM, Stirling H(1989) Cerebrospinal fluid pressure inpyogenic meningitis. Arch Dis Child64:814–820

9. Misu T, Takahashi T, Sato S, TateyamaM, Kato H, Itoyama Y (1999) A caseof spontaneous intracranial hypoten-sion with a remarkable leakage andcollection of CSF. Rinsho Shinkeigaku39:948–952

10. Roccella EJ (1996) Update on the 1987Task Force Report on High Blood Pres-sure in Children in Adolescence: aWorking Group Report from the Na-tional High Blood Pressure EducationProgramme. Pediatrics 98:649–658

11. Schievink WI, Meyer FB, AtkinsonJLD, Mokri B (1996) Spontaneous spi-nal cerebrospinal fluid leaks and intra-cranial hypotension. J Neurosurg84:598–605

12. Welch K (1980) The intracranial pres-sure in infants. J Neurosurg52:693–699