Post-Traumatic Hydrocephalus after Decompressive Craniectomy: An Underestimated Risk Factor

Abstract

The incidence of post-traumatic hydrocephalus (PTH) has been reported to be 0.7–51.4%, and we have frequently observed the development of PTH in patients undergoing decompressive craniectomy (DC). For this reason we performed a retrospective review of a consecutive series of patients undergoing DC after traumatic brain injury (TBI). From January 2006 to December 2009, 41 patients underwent DC after closed head injury. Study outcomes focused specifically on the development of hydrocephalus after DC. Variables described by other authors to be associated with PTH were studied, including advanced age, the timing of cranioplasty, higher score on the Fisher grading system, low post-resuscitation Glasgow Coma Scale (GCS) score, and cerebrospinal fluid (CSF) infection. We also analyzed the influence of the area of craniotomy and the distance of craniotomy from the midline. Logistic regression was used with hydrocephalus as the primary outcome measure. Of the nine patients who developed hydrocephalus, eight patients (89%) had undergone craniotomy with the superior limit <25 mm from the midline. This association was statistically significant (p = 0.01 - Fisher's exact test). Logistic regression analysis showed that the only factor independently associated with the development of hydrocephalus was the distance from the midline. Patients with craniotomy whose superior limit was <25 mm from the midline had a markedly increased risk of developing hydrocephalus (OR = 17). Craniectomy with a superior limit too close to the midline can predispose patients undergoing DC to the development of hydrocephalus. We therefore suggest performing wide DCs with the superior limit >25 mm from the midline.

Introduction

O ver the past decade, there has been renewed interest in decompressive craniectomy (DC), a procedure performed by the senior author (C.A.) since the 1980s. In fact, there has recently been a dramatic increase in the number of studies evaluating the therapeutic benefits of DC for otherwise uncontrollable intracranial hypertension (with more than 500 articles recently published, compared with about 150 articles published from 1970 to 1999). Currently two trials are in progress (Cooper et al., 2008; Hutchinson et al., 2006).

The incidence of post-traumatic hydrocephalus (PTH) has been reported to be 0.7–51.4%, with the wide variation being due to different evaluation criteria (Choi et al., 2008; Guyot and Michael, 2000; Mazzini et al., 2003; Tian et al., 2008).

Increased age, cerebrospinal fluid (CSF) infection, intraventricular hemorrhage, thickness and distribution of subarachnoid hemorrhage (SAH), and DC have been reported to increase the risk of developing PTH (Choi et al., 2008; Jiao et al., 2007; Kan et al., 2006; Tian et al., 2008; Yang et al., 2008).

We saw that a large proportion of our patients who underwent DC after traumatic brain injury (TBI) developed PTH within a short time after DC, and that most of these patients had craniotomies close to the midline, where the bridging veins enter the sagittal sinus. This region represents a crucial site in the intracranial system; cerebral blood flow and CSF circulation are closely intermingled here. In a mathematical model of hydrocephalus, we suggested that a selective increase in the venous outflow during the diastolic phase of each cardiac cycle may induce ventricular dilation, and that this effect may be produced in some circumstances by a DC (Anile et al., 1999).

We therefore hypothesized that a too medial craniotomy may have a potentially pathogenetic role in the development of PTH. This induced us to perform a retrospective review of a consecutive series of patients undergoing DC after TBI.

Methods

Patients

From January 2006 to December 2009, 41 patients underwent DC for the treatment of otherwise uncontrollable intracranial pressure after closed head injury. The criteria for decompressive surgical intervention have been described previously (Pompucci et al., 2007). Fourteen patients (34%) died within 1 month after surgery. Of the remaining 27 patients, one patient was transferred to another institution and was lost to follow-up in the early postoperative period, before bone flap replacement. Thus the final cohort included 26 patients.

Surgical technique

In cases in whom surgery was performed to treat diffuse cerebral edema without a mass lesion, a wide unilateral hemicraniectomy was performed, including removal of bone in the frontal, temporal, and parietal regions ipsilateral to the hemisphere in which the greatest swelling was observed on CT scans. The craniectomy was performed from the temporal bone to the floor of the middle fossa to maximize the extent of decompression at the level of the basal cisterns. The dura mater was widely opened in a stellate fashion to the extent of bone decompression, and a dural graft was placed to increase the available volume before closure. The bone flap was stored in sterile conditions with cryopreservation until cranioplasty. Four patients received computer-guided synthetic flaps (Synthes Inc. s.r.l Opera [MI] or Finceramica-Codman, Faenza, [RA]). When a mass lesion was present, a similarly wide craniectomy incorporating the mass lesion was performed. The craniectomy limits were: anterior, frontal to the midpupillary line; posterior, 3–6 cm posterior to the external acoustic meatus; superior, 1–4.5 cm from the midline; and inferior, at level of the middle cranial fossa floor at the origin of the zygomatic arch. The mean anteroposterior diameter of the bone flap was 15 cm (range 13.5–16.5 cm). The mean area of the craniotomy was calculated by assuming the craniectomy area as a hypothetical circle with diameter = d (i.e., the largest diameter calculated on CT scans), as shown by Münch and associates (Münch et al., 2000): the area (A) of the segment of a sphere above the craniectomy area was therefore calculated with the formula: A = π[(d/2)² + h²], where h represents the perpendicular line to d with the longest distance from d to the dural flap (Fig. 1).

FIG. 1.

Axial computed tomography scan showing the method used for calculation of the craniectomy area (see text for details).

Definition of hydrocephalus

For the purposes of this study, post-traumatic hydrocephalus was defined as radiographic evidence of progressive ventricular dilatation, with an Evan index >0.3, associated with narrowed CSF spaces at the convexity on serial CT imaging.

For patients with poor baseline neurological status (i.e., comatose patients), clinical examination results were not included as a defining element in the determination of hydrocephalus. For patients showing an initial improvement of clinical condition, impaired consciousness, or a worsening neurologic status (not due to infections or other medical causes), this was included as a defining element in the determination of hydrocephalus. Intracranial pressure measurement was not included in this definition.

Statistical analyses

The statistical package used for the analyses was SPSS for Windows version 13.0 (SPSS Inc., Chicago, IL). Study outcomes focused specifically on the development of hydrocephalus after DC and the need for permanent CSF diversion. Variables described by other authors to be associated with PTH were also studied, including patient age, the timing of cranioplasty, distribution and thickness of SAH (Fisher grading system), post-resuscitation Glasgow Coma Scale (GCS) score, and CSF infection (Table 1). We also analyzed the influence of the distance of craniotomy from the midline, and the area of the craniotomy on the development of hydrocephalus. Fisher's exact test was used to evaluate associations between these factors and the development of hydrocephalus. As the outcome was dichotomous (hydrocephalus versus no hydrocephalus), a logistic regression model was used to assess the independent contributions of the predictive factors to the development of hydrocephalus. Variables with p values <0.05 were considered statistically significant.

Table 1.

Characteristics of the Study Population

Factors	Hydrocephalus (9 patients)	No hydrocephalus (17 patients)	p Value ^a
Distance
<25 mm	8	5	<0.01
≥25 mm	1	12
Timing of cranioplasty
<30 days	3	6	0.92
≥30 days	6	11
Fisher grade
Grade 1	1	1	0.8
Grade 2	3	8
Grade 3	4	5
Grade 4	1	3
Age
<40 years	7	11	0.5
≥40 years	2	6
Post-resuscitation GCS score
3–5	4	11	0.16
6–8	4	2
9–15	1	4
Craniectomy size
≥162 cm²	4	8	0.09
<162 cm²	5	9
CSF infection
Yes	1	0
No	8	17

p by Fisher's exact test.

GCS, Glasgow Coma Scale; CSF, cerebrospinal fluid.

Results

The study population consisted of 26 patients (16 males, 10 females; age range 16–76 years, mean age 35 years). The mean time from DC to cranioplasty was 44 days (range 16–153 days). The cause of trauma was road accidents (as a pedestrian or as a bicycle/car/motorcycle driver) for 24 patients, and falls for two patients (Table 2). Nine patients (34.5%) presented with post-operative hydrocephalus as defined above. All of these patients presented with progressive ventricular enlargement both before and after cranioplasty, and underwent CSF diversion (with a Codman-Hakim programmable valve in all cases, and with a Codman-Bactiseal shunting system placed in four patients). The development of PTH was not associated with worse outcome, as measured by 6-month Glasgow Outcome Scale score (GOS; p = 0.3 by Fisher's exact test). Mean area of craniectomy was 162 cm² (165 ± 22 cm² in the group that developed hydrocephalus, and 160 ± 28 cm² in those that did not). Patients who underwent craniectomy whose superior limit was ≥25 mm had a mean area of 162 ± 17.8 cm², while the other patients had a mean area of 169 ± 24.6 cm². A significant reduction in mean ICP was obtained for all patients, regardless of craniectomy area (Table 2). These results are comparable to those obtained using unilateral hemicraniectomy or bifrontal decompressive craniotomy by other authors (Olivecrona et al., 2007; Polin et al., 1997; Whitfield et al., 2001). Older age, later timing of cranioplasty, higher score on the Fisher grading system, craniectomy size, and low post-resuscitation GCS score were not significantly associated with the development of hydrocephalus in our series (Tables 2, 3). Only one patient suffered from CSF infection prior to cranioplasty, and developed hydrocephalus after resolution of the infection. Statistical analyses addressing the relationship between CSF infection and the development of hydrocephalus were therefore not possible. Of the nine patients who developed hydrocephalus, eight patients (89%) had undergone craniotomy whose superior limit was <25 mm from the midline (Fig. 2B), and that association was statistically significant (p = 0.01). Logistic regression analysis showed that the only factor independently associated with the development of hydrocephalus was the distance from the midline (Fig. 3). Patients with craniotomy whose superior limit was <25 mm from the midline had a markedly increased risk of developing hydrocephalus (OR = 17.85; Table 3).

FIG. 2.

(A) A patient with a superior border of the craniectomy ≥25 mm. (B) A patient with a superior border of the craniectomy <25 mm.

FIG. 3.

Graph showing the distribution of the patients between groups (those with hydrocephalus and those without), and the distance of the craniotomy from the midline. All patients with hydrocephalus but one belonged to the group with a distance <25 mm from the midline (p < 0.01).

Table 2.

Cause of Trauma and Surgical Details

Patient no.	Craniectomy area (cm ² )	Distance from the midline (mm)	Post-operative intracranial pressure (mm Hg)	Cause of trauma: RA (road accident), F(fall)
1	201	<25	<20	RA
2	150	<25	<10	RA
3	145	<25	<15	RA
4	161	<25	<10	RA
5	153	<25	<15	RA
6	150	<25	<10	RA
7	170	<25	<20	RA
8	155	<25	<10	RA
9	150	<25	<10	RA
10	165	<25	<15	RA
11	170	<25	<20	RA
12	190	<25	<10	RA
13	170	<25	<10	F
14	161	≥25	<20	RA
15	140	≥25	<15	RA
16	192	≥25	<15	RA
17	145	≥25	<10	RA
18	150	≥25	<15	RA
19	185	≥25	<15	F
20	172	≥25	<10	RA
21	160	≥25	<10	RA
22	229	≥25	<10	RA
23	150	≥25	<15	RA
24	192	≥25	<10	RA
25	170	≥25	<10	RA
26	160	≥25	<10	RA

Table 3.

Factors Affecting the Development of Hydrocephalus

			95% Confidence interval
Variables	p value	Odds ratio	Lower	Upper
Hemoventricle	0.877	1.349	0.03	60.433
Age cut-off 40 years	0.847	1.347	0.065	27.8
Preoperatice GCS score	0.441	4.938	0.085	285.847
Cranioplasty timing (cut-off 30 days)	0.715	1.548	0.149	16.132
Craniectomy size (cut-off 162 cm²)	0.86	0.828	0.102	6.739
Distance from the midline (cut-off 25 mm)	0.032	17.850	1.273	250.288

GCS, Glasgow Coma Scale.

Discussion

The main limitation of our study is its retrospective, nonrandomized nature, with a limited number of patients. We cannot confirm with certainty that in our series hydrocephalus was only due to DC as did other authors for two reasons: (1) hydrocephalus developed both after DC and after cranioplasty; and (2) primary and secondary damage and SAH (that was present in all cases) due to TBI may have per se influenced the onset of hydrocephalus (Waziri et al., 2007). Nonetheless, we found a strong independent association between craniectomy limits and the development of PTH.

Several articles have documented a possible correlation between PTH and DC (Jiao et al., 2007; Mazzini et al., 2003; Phuenpathom et al., 1999). The incidence of this correlation appears to be more relevant in the presence of larger craniectomies (Kan et al., 2006). Such a relationship has been already described by Weiford and Gardner (Weiford and Gardner, 1949), who reported a case of cerebral herniation secondary to ventricular dilation after craniectomy for trauma. The possible occurrence of such a condition was experimentally demonstrated in cats by Hochwald and associates (Hochwald et al., 1972). The correlation seen between ventricular dilation and the opening of the intracranial system to the atmosphere is also curiously suggested by the case described by Kaufman and Miller (Kaufman and Miller, 1978), of a patient suffering from marked herniation of cerebral tissue through a large skull defect associated with ipsilateral lateral ventricle enlargement, which was treated by elastic bandages that were rewrapped several times a day until the brain herniation was definitively controlled. This may be the pathogenetic mechanism linking craniectomy and ventricular dilation.

Waziri and colleagues (Waziri et al., 2007) have recently suggested that decompressive craniectomy may play a role in the “flattening” of the normally dicrotic CSF pulse waveform seen in patients who undergo DC, due to the transmission of the pressure pulse out through the open cranium. Because the arachnoid granulations function as pressure-dependent one-way valves from the subarachnoid space to the draining venous sinuses, it is possible that disruption of pulsatile intracranial pressure dynamics secondary to opening the cranial vault results in decreased CSF outflow, thus affecting ventricular enlargement. In the opinion of these authors, this hypothesis appears to be further confirmed by the observation that early cranioplasty should lead to restoration of normal intracranial pressure dynamics and spontaneous resolution of hydrocephalus, as witnessed. In contrast, late cranioplasty, by prolonging this disruption, might be expected to result in permanent dysfunction of the arachnoid granulations, such as that seen in hydrocephalus induced by long-term CSF drainage (Waziri et al., 2007). This hypothesis, however, does not fully explain our data, which clearly show the statistically significant independent correlation between the risk of hydrocephalus and the distance of the superior boundary of the craniectomy from the cranial midline.

Another possible explanation may be found in a new hypothesis about the mechanisms leading to ventricular dilation that we proposed in 1999 (Anile et al., 1999).

In our opinion, the development of hydrocephalus is determined by the association between two main factors: the pulsatile component of ICP arising from the choroid plexuses, and the asymmetric response of the brain parenchyma to this alternate force. This second factor is represented by the non-linear behavior of the venous blood flow passing through the “lacunae laterales” from the bridging veins to the dural sinuses. The anatomo-functional complex constituted by these structures acts as a Starling resistor, determining during each cardiac cycle the exact balance between the production and absorption of extracellular fluid. Extracellular fluid is produced during the systolic phase, and is absorbed during the diastolic phase, and an imbalance between production and absorption in favor of absorption will cause a decrease in the volume of the brain parenchyma, and a consequent increase in ventricular volume, causing hydrocephalus. In DC, when the skull is removed too close to the midline, this reduces the external force compressing the veins mainly during the diastolic phase, thus causing an increase in venous outflow, which in turn produces an increase in extracellular fluid absorption and a decrease in the volume of the brain parenchyma, which causes ventricular enlargement (Anile et al., 2009). Introducing this condition into the previously mentioned mathematical model may cause a more pronounced effect of DC on the venous outflow resistance than on the ICP reduction, and an increase in the ventricular volume is clearly evident (unpublished data presented by C. Anile at the 12th Euroacademia Multidisciplinaria Neurotraumatologica Annual Meeting, held in Rome on June 21–23, 2007).

This new hypothesis may explain the high incidence of hydrocephalus (29%) seen in the series of Polin and associates (Polin et al., 1997). Those authors performed a decompressive bifrontal craniectomy whose posterior limit was between 3 and 5 cm posterior to the coronal suture, thus interfering with venous outflow. The very high incidence (88% of patients, half of whom had persistent hydrocephalus after cranioplasty, and required a ventriculoperitoneal shunt) of post-DC hydrocephalus reported in the article by Waziri and colleagues (Waziri et al., 2007) may be explained by this same effect. In fact, the superior margin of the craniotomy used by these authors was within 1–2 cm of the sagittal sinus. Moreover, in that article the authors had no confounding factors (e.g., traumatic brain injury or SAH), so they probably observed a more pure influence of DC on the development of hydrocephalus. Nonetheless, the aforementioned hypothesis does not explain why we observed a resolution of hydrocephalus after bone flap replacement only in some patients (50% in the series by Waziri and colleagues). Those authors suggest that time might play a role by prolonging the pathogenetic mechanisms such that hydrocephalus becomes irreversible, even after cranioplasty (Waziri et al., 2007).

Conclusion

Craniectomy whose superior limit is too close to the midline can predispose patients undergoing DC to the development of hydrocephalus. We therefore suggest the performance of wide DCs, with the superior limit >25 mm from the midline.

Footnotes

Acknowledgment

The authors wish to acknowledge D. Petricca for reviewing the figures.

Author Disclosure Statement

No competing financial interests exist.

References

Anile

, De Bonis

, Di Chirico

, Ficola

, Mangiola

, Petrella

2009. Cerebral blood flow autoregulation during intracranial hypertension: a simple, purely hydraulic mechanism? Childs Nerv. Syst., 25:325–335.

Anile

, Fravolini

M.L.

, Ficola

1999. A Dynamic Model of the Cerebrospinal Fluid System for the Study of Hydrocephalus. European Control Conference ECC ‘99, August–3 September 31, 1999, Karlsruhe, Germany.

Choi

, Park

H.K.

, Chang

J.C.

, Cho

S.J.

, Choi

S.K.

, Byun

B.J.

2008. Clinical factors for the development of posttraumatic hydrocephalus after decompressive craniectomy. J. Korean Neurosurg. Soc., 43:227–231.

Cooper

D.J.

, Rosenfeld

J.V.

, Murray

, Wolfe

, Ponsford

, Davies

, D'Urso

, Pellegrino

, Malham

, Kossmann

2008. Early decompressive craniectomy for patients with severe traumatic brain injury and refractory intracranial hypertension—a pilot randomized trial. J. Crit. Care., 23:387–393.

Guyot

L.L.

, Michael

D.B.

2000. Post-traumatic hydrocephalus. Neurol. Res., 22:25–28.

Hochwald

G.M.

, Epstein

, Malhan

, Ransohoff

1972. The role of the skull and dura in experimental feline hydrocephalus. Dev. Med. Child Neurol. Suppl., 27:65–69.

Hutchinson

P.J.

, Corteen

, Czosnyka

, Mendelow

A.D.

, Menon

D.K.

, Mitchell

, Murray

, Pickard

J.D.

, Rickels

, Sahuquillo

, Servadei

, Teasdale

G.M.

, Timofeev

, Unterberg

, Kirkpatrick

P.J.

2006. Decompressive craniectomy in traumatic brain injury: the randomized multicenter RESCUEicp study. www.RESCUEicp.com. Acta Neurochir. Suppl., 96:17–20.

Jiao

Q.F.

, Liu

, Li

, Zhou

L.X.

, Li

S.Z.

, Tian

, You

2007. Influencing factors for posttraumatic hydrocephalus in patients suffering from severe traumatic brain injuries. Chin. J. Traumatol., 10:159–162.

Kan

, Amini

, Hansen

, White

G.L.

Jr. , Brockmeyer

D.L.

, Walker

M.L.

, Kestle

J.R.

2006. Outcomes after decompressive craniectomy for severe traumatic brain injury in children. J. Neurosurg., 105:337–342.

10.

Kaufman

H.H.

, Miller

M.H.

1978. Cerebral herniation from ventricular dilatation after craniectomy: a method of treatment. Neurosurgery, 2:136–137.

11.

Mazzini

, Campini

, Angelino

, Rognone

, Pastore

, Oliveri

2003. Posttraumatic hydrocephalus: a clinical, neuroradiologic, and neuropsychologic assessment of long-term outcome. Arch. Phys. Med. Rehabil., 84:1637–1641.

12.

Münch

, Horn

, Schürer

, Piepgras

, Paul

, Schmiedek

2000. Management of severe traumatic brain injury by decompressive craniectomy. Neurosurgery, 47:315–322.

13.

Olivecrona

, Rodling-Wahlström

, Naredi

, Koskinen

L.O.

2007. Effective ICP reduction by decompressive craniectomy in patients with severe traumatic brain injury treated by an ICP-targeted therapy. J. Neurotrauma, 24:927–935.

14.

Phuenpathom

, Ratanalert

, Saeheng

, Sripairojkul

1999. Post-traumatic hydrocephalus: experience in 17 consecutive cases. J. Med. Assoc. Thai., 82:46–53.

15.

Polin

R.S.

, Shaffrey

M.E.

, Bogaev

C.A.

, Tisdale

, Germanson

, Bocchicchio

, Jane

J.A.

1997. Decompressive bifrontal craniectomy in the treatment of severe refractory posttraumatic cerebral edema. Neurosurgery, 41:84–92.

16.

Pompucci

, De Bonis

, Pettorini

, Petrella

, Di Chirico

, Anile

2007. Decompressive craniectomy for traumatic brain injury: patient age and outcome. J. Neurotrauma, 24:1182–1188.

17.

Tian

H.L.

, Xu

, Hu

, Cui

Y.H.

, Chen

, Zhou

L.F.

2008. Risk factors related to hydrocephalus after traumatic subarachnoid hemorrhage. Surg. Neurol., 69:241–246.

18.

Waziri

, Fusco

, Mayer

S.A.

, McKhann

G.M.

2nd , Connolly

E.S.

Jr.

2007. Postoperative hydrocephalus in patients undergoing decompressive hemicraniectomy for ischemic or hemorrhagic stroke. Neurosurgery, 61:489–493.

19.

Weiford

E.C.

, Gardner

W.J.

1949. Tantalum cranioplasty: review of 106 cases in civilian practice. J. Neurosurg., 6:13–32.

20.

Whitfield

P.C.

, Patel

, Hutchinson

P.J.

, Czosnyka

, Parry

, Menon

, Pickard

J.D.

, Kirkpatrick

P.J.

2001. Bifrontal decompressive craniectomy in the management of posttraumatic intracranial hypertension. Br. J. Neurosurg., 15:500–507.

21.

Yang

X.F.

, Wen

, Shen

, Li

, Lou

, Liu

W.G.

, Zhan

R.Y.

2008. Surgical complications secondary to decompressive craniectomy in patients with a head injury: a series of 108 consecutive cases. Acta Neurochir. (Wien.), 150:1241–1247.