Abstract
Post-traumatic hydrocephalus (PTH) is a particularly challenging complication of traumatic brain injury (TBI). The primary treatment for PTH is placement of a ventricular shunt. However, shunts are associated with high complication rates regardless of placement indication. PTH is understudied compared with other TBI sequelae, and long-term outcomes remain poorly characterized. We report our institution’s experience with shunted PTH over the last decade. Fifteen patients presented with TBI between January 2014 and April 2024 and underwent at least one shunt placement for diagnosed PTH. Patients’ demographics, injury characteristics, clinical courses, and outcomes were manually extracted from electronic medical records. Most patients were male (86.7%), White (33.3%), and suffered a severe TBI (Glasgow Coma Scale score 3–8). Four (26.7%) patients underwent shunt placement within 1 month of TBI, 9 (60.0%) within 3 months, 13 (86.7%) within 1 year, and all patients within 2 years. At least one shunt failure occurred in 53.3% of patients, and multiple failures requiring revision surgery occurred in 46.7%; all but one patient had their initial shunt failure occur within 1 year of placement. The most common reasons for shunt failure were catheter obstruction (26.3%) and infection (26.3%). Patients had a median follow-up of 2.8 years and an overall mortality rate of 13.3%. Of the surviving patients, a favorable long-term outcome (Glasgow Outcome Scale 4–5) was achieved in 26.7%. Notably, 6/15 (40.0%) patients experienced early post-traumatic seizures (ePTS). Only 4/10 patients who developed post-traumatic epilepsy (PTE) had experienced ePTS, challenging existing literature’s suggestion that ePTS most reliably predicts PTE development. Combining our cohort with that of an underrecognized report from 2000, we report a cumulative 61.9% incidence of PTE in shunt-dependent PTH, which, to our knowledge, is the highest reported incidence of PTE in current literature. The presence of shunted PTH following TBI may thus increase the risk profile for developing PTE.
Introduction
Traumatic brain injury (TBI) is annually responsible for over seven million lived years of disability worldwide, and neurotrauma comprises the majority of global neurosurgical burden.1–3 While TBI typically results from acute injury, its recovery course is neither finite nor entirely predictable and is better conceptualized as a chronic condition impacting multiple domains of physical and cognitive function. 4 Direct sequelae of TBI may include psychiatric problems, personality changes, headaches, neurodegenerative diseases, sleep disorders, and seizures—all of which may arise at unpredictable timepoints following the initial insult. 5
One particularly morbid complication of TBI is post-traumatic hydrocephalus (PTH), a syndrome first described in Walter Dandy’s 1914 report delineating hydrocephalus etiologies. 6 PTH is defined as excessive accumulation of cerebrospinal fluid (CSF) in the ventricular system due to post-traumatic alterations in CSF dynamics. 7 Although the pathophysiology of PTH is not fully understood, one theorized mechanism postulates that head trauma-induced damage to neurons, glial cells, and blood vessels may lead to the impaired production, flow, or drainage of CSF.8,9 Acute PTH typically presents with headache, nausea and vomiting, slowed cognition, and visual impairment; diagnosis is confirmed with computed tomography or magnetic resonance imaging. 10
While post-traumatic ventriculomegaly may occur in roughly half of TBI patients, the development of symptomatic PTH is rarer, with its reported incidence varying widely from 0.7% to 29%.11–13 Post-traumatic ventriculomegaly refers only to enlarged ventricles and may occur as the brain compensates for post-injury atrophy, though it does not involve impaired CSF flow or increased intracranial pressure. 12 Post-traumatic ventriculomegaly typically does not require intervention unless it progresses to PTH, for which the definitive treatment is surgical placement of a shunt system to carry CSF from the ventricular system or spinal canal to an area of the body where it may be more readily absorbed. Shunt types may include ventriculoperitoneal shunts, ventriculoatrial shunts, or lumboperitoneal shunts. Prior studies have largely focused on identifying factors that increase the risk of PTH development and the need for shunting.13–15
While several studies have investigated predictors of PTH development, PTH itself remains an understudied subject (especially relative to other TBI sequelae), and literature reporting on surgical treatment and long-term outcomes of shunted PTH is scant.16,17 This is particularly important considering shunts for all hydrocephalus subtypes have a notoriously high rate of failure, with 29% of adult shunts failing within 1 year of placement and 45–81% of adult patients requiring at least one surgical shunt revision within their lifetime. 18 We report our institution’s experience with shunted PTH over the last decade, including patient and injury characteristics, surgical intervention, and long-term patient outcomes.
Materials and Methods
Study population
After Institutional Review Board approval was obtained, patients were identified via NYU Langone Health Neurosurgery Epic Datacore. Patient inclusion criteria were (1) initial presentation at NYULH between January 2014 and April 2024, (2) TBI occurring between January 2014 and April 2024, (3) clinical diagnosis of PTH, and (4) at least one shunt placement or revision surgery at NYULH. All diagnoses of PTH were made by neurosurgeons. Patient electronic medical records were manually reviewed to extract clinical and demographic variables for our analysis.
Variables and data analysis
Demographic variables included were age at injury, sex, and race. Clinical variables included Glasgow Coma Scale (GCS) score at presentation, mechanism of injury (MOI), number of cranial fractures, presence of intracranial bleed or contusion, post-traumatic seizures, time to PTH determination, Glasgow Outcome Scale (GOS) score (1 = death, 5 = good recovery), number of antiepileptic drugs (AEDs), and follow-up time. TBI was categorized as mild (GCS 13–15), moderate (GCS 9–12), and severe (GCS 3–8). Early post-traumatic seizures (ePTS) were defined as seizures occurring within 1 week of TBI, and post-traumatic epilepsy (PTE) was defined as at least one seizure occurring more than 1 week after TBI in line with classification in the literature. 19 Surgical variables included need for and timing of decompressive craniectomy or craniotomy, cranioplasty timing, shunt type (ventriculoatrial [VA], ventriculoperitoneal [VP], etc.), length of stay (LOS), perioperative complications, and the number, timing, and cause of shunt revisions.
Results
Demographics and injury characteristics
Fifteen patients met the inclusion criteria, most of whom were male (13/15, 86.7%) and White (5/15, 33.3%). Patients’ median age at the time of injury was 33.0 years (range, 19–77 years), and most TBIs were severe (9/15, 60.0%), although two patients’ GCS scores on initial presentation were not recorded (Table 1). The most common MOI was motor vehicle accident (5/15, 33.3%), and all patients suffered at least one skull fracture, with 12/15 (80.0%) suffering multiple fractures. All patients experienced multiple forms of intracranial bleeding; all 15 patients had subarachnoid hemorrhage (SAH), 14/15 (93.3%) had intraventricular hemorrhage (IVH), 12/15 (80.0%) had a subdural hematoma, 11/15 (73.3%) had hemorrhagic parenchymal contusions, and 10/15 (66.7%) had traumatic intracerebral hemorrhage (Table 2).
Patient Characteristics, Seizures, and Shunts
GCS, Glasgow Coma Scale; TBI, traumatic brain injury.
Traumatic Brain Injury and Sequelae
This patient’s decompressive craniectomy was delayed due to pregnancy at the time of injury requiring an emergent cesarean section and intraoperative external ventricular drain placement. Post-traumatic hydrocephalus development led to a Chiari malformation, requiring decompressive craniectomy.
Patient was lost to follow-up for a time but was still having occasional seizures 910 days postinjury.
AOD, atlanto-occipital dislocation; GCS, Glasgow Coma Scale; ICH, intracerebral hemorrhage; IVH, intraventricular hemorrhage; LOS, length of stay; MVA, motor vehicle accident; PTE, post-traumatic epilepsy; PTS, post-traumatic seizures; SAH, subarachnoid hemorrhage; SDH, subdural hemorrhage.
All but 2 patients (13/15, 86.7%) received a decompressive craniectomy or craniotomy (DC), and 12/13 (92.3%) of those underwent the surgery within 1 day of injury. The one patient with delayed DC was pregnant on presentation and underwent an emergent cesarean section with concurrent external ventricular drain (EVD) placement. The patient then received a DC 167 days post-TBI in the context of post-traumatic Chiari malformation and syringomyelia. Of the two patients who did not receive a DC, one suffered an atlanto-occipital dislocation and underwent occipital to C4 fixation 30 days post-TBI, and the other was a geriatric patient (age 77) with a declining GCS score determined to be a poor surgical candidate. The median LOS for initial hospitalization was 30 days (10–116 days). Ten (66.7%) patients later underwent cranioplasty with a median time from TBI to cranioplasty of 80.5 days (16–1026 days) and median time from DC to cranioplasty of 79.5 days (15–1025 days).
Post-traumatic seizures
Six (40.0%) patients experienced at least one ePTS, and four (66.7%) of these went on to develop PTE (Fig. 1). Of the 9 patients who did not experience ePTS, 6 (66.7%) went on to develop PTE for a total of 10/15 (66.7%) patients in the cohort developing PTE. All but two patients who developed PTE had their first post-traumatic epileptic seizure after shunt placement. All patients with PTE were maintained on AEDs. Of the 10 patients with PTE, at their most recent follow-up, 6/10 were maintained on one to three AEDs with no breakthrough seizures, 3/10 were maintained on one to two AEDs and were still experiencing breakthrough seizures, and one had developed drug-resistant epilepsy on four AEDs.
Days to first post-traumatic epileptic seizure and shunt. When applicable, darker colored bars show the days from trauma to shunt placement, and lighter colored bars show the days from initial injury to the first post-traumatic epileptic seizure. All but two patients who developed PTE experienced their first epileptic seizure after shunt placement. PTE, post-traumatic epilepsy.
Hydrocephalus diagnosis and treatment
Patients in our cohort had a median time from injury to PTH diagnosis of 42.5 days (range, 0–669 days), though, in one case (Patient 1), the time to initial PTH diagnosis was unclear. The median time from PTH diagnosis to shunt placement was 20 days (5–178 days), with the median time from initial TBI to shunt placement being 85 days (range, 15–738 days; Fig. 2). Four (26.7%) patients underwent shunt placement within 1 month of TBI, 9/15 (60.0%) within 3 months, 13/15 (86.7%) within 1 year, and all patients within 2 years. Most patients (13/15, 86.7%) initially underwent placement of a VP shunt, and 12/15 (80.0%) had a programmable valve (Table 3). One patient’s hydrocephalus was initially treated with an endoscopic third ventriculostomy (ETV), but the hydrocephalus did not resolve, and the patient’s first shunt was placed 7 days later.
Days to PTH diagnosis and subsequent shunt placement. Darker colored bars show the days from trauma to the diagnosis of post-traumatic hydrocephalus, and lighter colored bars show the subsequent wait from diagnosis of PTH to the placement of a shunt system. Note that the precise time to PTH diagnosis for Patient 1 is unclear but occurred shortly after the initial trauma, and their first shunt was placed within 30 days of injury. PTH, post-traumatic hydrocephalus.
Treatment and Outcomes
GOS, Glasgow Outcome Scale; LP, lumboperitoneal; NR, not recorded; PTH, post-traumatic hydrocephalus; VA, ventriculoatrial; VP, ventriculoperitoneal.
Seven (46.7%) patients experienced no shunt failures. At least one shunt failure requiring revision occurred in 8/15 (53.3%) patients, and multiple failures requiring revision occurred in 7/15 (46.7%) for 19 total shunt failures in our cohort (Table 4). All but one patient (7/8, 87.5%) experienced their first shunt failure within 1 year of initial placement, with a median time to failure of 37.5 days (about 5 weeks). For the seven patients with multiple shunt failures, the median time from first to second failure was 23 days (range, 1–270 days). Notably, one patient was lost to follow-up after initial shunt placement, and of the other six patients who required no shunt revision, three had follow-up periods of less than a year. The most common reasons for shunt failure were catheter obstruction (5/19, 26.3%) and infection (5/19, 26.3%). Of the 19 cases of shunt failure, seven (36.8%) required initial externalization with EVD placement and then a subsequent revision surgery days later to replace the EVD with a permanent shunt. Thus, there were a total of 26 revision surgeries in the eight patients who experienced shunt failure, 14/26 (53.8%) of which were changes from a permanent shunt to a temporary EVD or subsequent replacement of the EVD with a permanent shunt.
Shunts Failures and Revisions
EVD, external ventricular drain; LP, lumboperitoneal; VA, ventriculoatrial; VP, ventriculoperitoneal.
Among the 14 patients with follow-up, median follow-up time was 2.8 years (0.5–7.0 years). Two (13.3%) patients died, both of whom received no shunt revisions with a time from initial shunt to death of 6 months and 3 years, respectively. Both patients who died were quadriplegic, a consequence of their initial injury, and died from systemic and respiratory infections. A favorable long-term outcome (GOS 4 or 5) was achieved in 4/15 (26.7%) patients, with 6/15 (40.0%) remaining dependent on others for daily living (GOS 3) and 3/15 (20.0%) remaining minimally responsive (GOS 2).
Discussion
Our report details a single institution’s decade-long experience with ventricular shunt-managed PTH. To our knowledge, this is the first study of its kind to describe the rates and specific causes of shunt failure and revision surgery in the context of shunted PTH. Our cohort consisted mostly of patients who suffered severe TBIs, nearly half of whom experienced repeated shunt failure requiring multiple revision surgeries.
Two-thirds of patients in our cohort were aged 35 years or younger. Previous research has demonstrated that pediatric and elderly age groups may be at increased risk for developing PTH following injury. 14 PTH may develop in part due to scarring and fibrosis of arachnoid granulation tissue that forms after injury, which prevents CSF reabsorption into the venous system in animal models. 20 This may explain why older patients have been suggested to be more susceptible to PTH, as meningeal fibrosis, which further inhibits CSF absorption, tends to increase with age. 12 Adolescents and young adults are also more likely to survive severe TBI than older adults, which may help explain the longer period of pathogenesis seen in these individuals. 7 However, surgical intervention for TBI in younger patients may also alter CSF dynamics and increase subarachnoid scarring in the developing brain. This all demonstrates that vigilance for the development of PTH needs to be a clinical priority in the early phases of follow-up care for pediatric neurotrauma. 21
Intracranial bleeding (particularly SAH and IVH) and meningitis have also been consistently reported as risk factors for developing PTH. This is postulated to occur consequent to ventricular system scarring, formation of arachnoid granulations, and scarring in the subarachnoid space, all of which may impede CSF outflow.22,23 Every patient in our cohort suffered SAH, all but one suffered IVH, and 40% had clinical courses complicated by ventriculitis or meningitis prior to shunt placement. The addition of infectious complications to the patient’s hospital course has been associated with a worse clinical outcome after shunt placement. 24 All but two patients in our cohort also underwent decompressive craniectomy or craniotomy prior to shunt placement. Decompressive craniectomy is a well-established risk factor for PTH development and need for shunt, as sudden changes in intracranial pressure distribution through an open craniectomy defect have been modeled to decrease CSF volume reaching arachnoid granulations.7,13,25 This is likely exacerbated by the residual blood mechanically obstructing CSF outflow from common outflow tracts. This all serves as further explanation regarding the severity of outcomes in our cohort.
Most patients in our cohort developed PTH within 1 year of their initial injury and all within 2 years, a finding consistent with Licata et al.’s 2001 report of 98 patients with PTH, all of whom developed hydrocephalus within 1 year of injury. 11 The median 20-day period between PTH diagnosis and initial shunt placement in our cohort may represent a conservative management strategy given the relative uncertainty of shunting for symptomatic relief. More work is required to identify if early shunting at the first signs of PTH improves outcomes and recovery from the original trauma. Anecdotally, two of the longest lead times for PTH diagnosis in our cohort (Patient 8 and Patient 15) eventually required shunt revision. The revision-free prognosis of PTH appears to lag behind that of the hydrocephalus of other etiologies based on our observations. While most studies report that approximately 29% of adult shunts are reported to fail within 1 year of placement, 18 53% of shunts in our cohort failed, with a median time to failure of just over 1 month.
Infection and shunt catheter obstruction were the most common reasons for shunt failure in our cohort. Infection may be a particularly dangerous complication contributing to shunt revision in PTH in the setting of decompressive craniectomy. Surgical site infection occurs following 12.3–29.7% of cranioplasty and can spread to previously implanted shunt systems, requiring shunt externalization, EVD placement, and later permanent shunt replacement. 26 These complications require patients to undergo repeated craniotomies, extend hospital stays, and increase the risk of perioperative complications.
Of the 15 patients in our cohort, only 4 (26.7%) achieved good long-term outcomes, defined as a GOS score of 1 or 2 on follow-up. This is consistent with prior studies that found that requiring a VP shunt for PTH is an independent risk factor for poorer clinical outcomes for TBI patients. 12 One prior study in 2000 conducted follow-up interviews to determine long-term outcomes of shunted PTH, and 19/48 (39.6%) patients were reported to have a GOS score of 1 or 2 at a mean follow-up interval of 3 years. 24 In the same study, 15/48 (31.3%) patients experienced shunt failure requiring revision. 24 Thus, while some patients may have good clinical recoveries, injury severity and GOS score prior to shunt placement are often the strongest prognostic indicators of post-traumatic quality of life. 12
ETV is a minimally invasive alternative to shunt placement that has shown promise in PTH and may avoid the complications of future revision surgery. 27 One patient in our cohort was initially treated with ETV, but hydrocephalus symptoms did not resolve, and shunt placement was required for definitive treatment.
EPTS, or seizures occurring within the first week after injury, have a reported incidence of 2.1–16.9% according to existing literature and are more likely to occur with severe TBI and in the presence of intracranial bleeding.28,29 While 6/15 (40.0%) of the PTH patients in our series experienced ePTS, this increased percentage may be partly explained by injury severity, as four of these patients suffered severe TBI, and all had mixed intracranial hemorrhage. Interestingly, however, one prior study found that TBI patients who suffered in-hospital seizures had increased intracranial pressure relative to TBI patients without ePTS, and ePTS were an independent predictor of longer LOS and worse hospital outcome (such as discharge to a nursing home). 30
PTE, defined as one or more unprovoked seizures occurring more than 1 week after initial injury, is reportedly rarer than ePTS.28,31 One study including 19,336 TBI patients and 540,322 non-TBI controls found that hazard ratios for developing PTE were 5.05 (95% confidence interval [CI]: 4.40 to 5.79) and 10.6 (95% CI: 7.14 to 15.8) in cases of severe TBI and in cases of TBI with concomitant skull fracture, respectively. 32 This may partially explain the high rates of PTE in our cohort, given that all patients suffered skull fractures associated with their injuries.
Meta-analyses and systematic reviews have reported an incidence rate of PTE ranging from 1.9% up to 53%,28,33,34 with the highest cited incidence of 53% coming from a 1985 report of Vietnam veterans who sustained penetrating head injury from ballistic trauma. 35 However, a potentially underrecognized report in 2000 from Tribl and Oder in Vienna, which, to our knowledge, is also the only other previous study reporting exclusively on patients with shunted PTH, found a PTE incidence of 29/48 (60.4%). 24 Notably, ePTS were reported in only 6/48 (12.5%) of these patients with shunted PTH, making PTE five times more common in their cohort than ePTS. 24 In our cohort, PTE developed in 10/15 (66.7%) of patients. The majority (6/10) of patients in our series who developed PTE also did not experience ePTS, contradicting previous research suggesting that ePTS is the most consistent predictor for PTE in the setting of TBI. 31 Further, equal proportions of patients who did and did not experience ePTS in our cohort developed PTE.
This may be partly explained by the fact that most TBI patients in our series received prophylactic levetiracetam in the hospital, and AEDs have been shown to be effective in reducing the risk of early, but not late, post-traumatic seizures. 36 However, in 2003, Mazzini et al. followed 140 patients with severe TBI; 13 patients developed shunt-dependent PTH, and hydrocephalus severity was found to be a prognostic factor for PTE development (p < 0.02). 37 The incidence of PTE in that cohort was not reported. Thus, combining the 15 patients in our series with the 48 in Tribl and Oder’s report, 24 the cumulative reported incidence of ePTS in shunted PTH is 6/63 (9.5%), while that of PTE in shunted PTH is 39/63 (61.9%). In this manner, the reported incidence of PTE in patients with shunted PTH surpasses the highest incidence of PTE previously reported in the literature. This all suggests that the presence of shunted PTH following TBI may fundamentally alter the risk profile for developing PTE, independently of whether ePTS were experienced or not.
This study’s primary limitations are its small sample size and largely descriptive nature. Due to the rarity of PTH, additional reports and further research are necessary to characterize the long-term outcomes and complication rates associated with PTH requiring ventricular shunt placement.
Conclusions
PTH is a clinical entity for which management with conventional shunting has yet to be optimized. PTE often co-occurs with PTH and complicates patient prognosis, justifying further research into management of PTH with and without seizure comorbidity.
Footnotes
Authors’ Contributions
E.A.G.: Conceptualization (lead), methodology (equal), investigation (lead), data curation (equal), and writing—original draft (lead). A.E.: Project administration (lead), methodology (equal), and writing—review and editing (equal). A.P.: Visualization (lead) and writing—review and editing (equal). D.d.S.: Investigation (supporting) and writing—review and editing (equal). M.F.: Data curation (equal) and formal analysis (lead). D.H.H.: Supervision (lead) and writing—review and editing (equal).
Author Disclosure Statement
The authors have no competing interest to disclose.
Funding Information
There was no funding provided for this research.