Neonatal Hydrocephalus

1. Hydrocephalus is a significant risk factor for future neurodevelopmental impairment in both preterm and term infants. However, the evidence informing the optimal timing of interventions for infants with evolving hydrocephalus is inadequate.

2. A number of interventions to reduce intracranial pressure exist, but there is practice variation among treating clinicians. Results from recent clinical studies have begun to inform choices among the various interventions.

After completing this article, readers should be able to:

1. Understand the pathophysiology of hydrocephalus.

2. Recognize the clinical symptoms of increased intracranial pressure.

3. Formulate a management approach for a patient with evolving hydrocephalus.

4. Appraise interventions to reduce intracranial pressure based on current evidence and recognize practices that are not beneficial.

Hydrocephalus is a clinical diagnosis of cerebrospinal fluid (CSF) accumulation in the ventricles and brain spaces accompanied by an increase in intracranial pressure (ICP). CSF is a clear, colorless liquid that nourishes the neural tissue and removes toxic metabolites. The choroid plexus (of the lateral ventricles) and the tela choroidea (of the third and fourth ventricles) produce 70% of the body’s CSF; the remainder is secreted by the ependymal epithelium and blood-brain barrier capillaries. (1) CSF production begins by the sixth week of gestation and approximately 300 to 500 mL of CSF is produced daily in the neonate. The CSF travels from the lateral ventricles to the third and fourth ventricles via the foramen of Monro and the cerebral aqueduct, respectively (Fig 1). It then enters the subarachnoid space surrounding the cerebral hemispheres and spinal cord. The superior sagittal sinus of the dural venous system has arachnoid granulations (or villi) to absorb CSF, which then further circulates to the internal jugular veins and inferior vena cava. (1) Arachnoid villi become visible between 6 and 18 months of age and increase in number and size over the first few years of age. CSF is continuously produced as it is absorbed, and there is a delicate balance between rates of CSF production and absorption.

The mechanism by which the accumulation of CSF in hydrocephalus leads to brain injury is not fully understood. When alterations in CSF flow dynamics (most commonly due to obstruction in the CSF circulatory pathway) gives rise to ventriculomegaly, the developing brain may be affected in several ways. Increased intraventricular pressure distends and compresses periventricular tissue, and this may result in ischemia, hypoxia, and cerebral edema. Hydrocephalus can also bring about changes in brain cytoarchitecture, structure, and metabolism. These changes may progress to altered blood-brain barrier transport as well as reduced amyloid clearance, dendritic and synaptic changes, and neuronal cell death. While the effect of apoptosis in the cerebral cortex is mitigated by the sheer number of cortical neurons, oligodendrocyte injury can result in demyelination, axonal degeneration, and gliosis with adverse effects on motor, behavioral, cognitive, and neurodevelopmental outcomes. (2)

Neonatal hydrocephalus is broadly categorized as congenital or acquired (Fig 2). Congenital hydrocephalus is further categorized as syndromic or isolated/nonsyndromic based on whether or not the hydrocephalus is part of an underlying genetic syndrome.

Genetic forms of hydrocephalus are rare, but the most common form is X-linked recessive hydrocephalus (Fig 3). This is a family of related entities, collectively known as “L1 disease,” that result from mutations in the Xq28 gene encoding L1CAM (L1 protein). These conditions include HSAS (hydrocephalus due to congenital stenosis of the aqueduct of Sylvius), MASA syndrome (mental retardation, aphasia, shuffling gait, adducted thumbs), spastic paraparesis type 1, and X-linked agenesis of the corpus callosum. The L1 protein is part of a superfamily of neuronal cell adhesion molecules expressed on the axons of developing neurons. (3) Together, the estimated incidence of these syndromes is 1 in 25,000 to 60,000 males, and L1 disease is thought to play a role in up to 25% of isolated congenital hydrocephalus in males. (4) Unfortunately, infants with L1 disease often have severe developmental delay, even with early CSF diversion. Hydrocephalus may also be present in conditions affecting chromosomes 8, 9, 13, 15, 18, and 21.

Although not genetic in nature, the association of hydrocephalus with open neural tube defects such as myelomeningocele (MMC) is well known and is a common cause of congenital syndromic hydrocephalus. Approximately 90% of infants with MMC develop hydrocephalus from Chiari malformation type II and require CSF diversion. (5) The “unified theory” proposes that an in utero open neural tube causes leakage of CSF that prevents appropriate accumulation of CSF in the embryonic ventricles. This, in turn, leads to abnormal development of the posterior fossa and subsequent descent of the cerebellum, brainstem, and fourth ventricle through the foramen magnum. (6) The descent interrupts CSF flow with resultant hydrocephalus. The Management of Myelomeningocele Study trial compared outcomes of prenatal and postnatal MMC repair; it was noted that infants who underwent fetal repair were half as likely to need a ventricular shunt (40% vs 82%), were less likely to develop Chiari malformation, and had better neuromotor outcomes with higher rates of ambulation at 30 months of age. (7) Based on these results, prenatal repair of MMC is considered to be the standard of care in centers with fetal surgery expertise. Vigilant screening of mothers with prenatally diagnosed MMC can help identify infants who may benefit from this complex and challenging intervention.

Hydranencephaly is the archetypical cause of nonsyndromic congenital hydrocephalus, occurring in 1 in 10,000 births. Although several causes have been proposed, the most common is intrauterine bilateral carotid artery occlusion with subsequent infarction of all cerebral tissue supplied by the anterior circulation and obliteration of the cerebral hemispheres. (8) On neuroimaging, the meninges and skull are intact and the empty cranial cavity is filled with CSF. The vertebrobasilar circulation is spared, preserving the thalamus, brainstem, and cerebellum (Fig 4). In newborns with seizures who have not been prenatally diagnosed, the absence of electrical activity on postnatal electroencephalography with the exception of occipital spikes is suggestive of hydranencephaly. (9) Prognosis is extremely poor and most infants die within the first year after birth. For this reason, treatment is largely symptomatic and supportive. Distinguishing hydrancephaly from maximal hydrocephalus is important for prognostication. In the latter, the cortical mantle is severely stretched and distorted but is preserved and has the ability to develop and regain function after appropriate CSF diversion.

Acquired hydrocephalus occurs after the development of the brain and ventricles and is the result of secondary mechanisms such as infection (toxoplasmosis, other agents, rubella cytomegalovirus, and herpes simplex [TORCH], tuberculosis, or bacterial meningitis), intraventricular hemorrhage (IVH), trauma, and structural abnormalities (tumors or cysts). These secondary causes impair CSF absorption and/or lead to obstruction, particularly at the aqueduct. Infections, specifically cytomegalovirus or toxoplasmosis may result in cerebral atrophy and inflammation of arachnoid granulations, impairing CSF absorption. IVH can lead to both scarring at the aqueduct and impaired absorption at arachnoid granulations.

By far, the most common cause of hydrocephalus in the neonatal population is IVH resulting in posthemorrhagic hydrocephalus (PHH). Neonatal PHH occurs in 15% to 35% of preterm infants with IVH. The vascular supply to the germinal matrix of preterm infants, particularly those born at less than 34 weeks’ gestation, is not invested with supporting stroma. This confers vulnerability of blood vessels in this area to rupture with increased pressure passive cerebral blood flow, which may occur during periods of cardiovascular instability and during resuscitation. Intraventricular blood induces an inflammatory cascade that leads to permanent remodeling and scarring of ventricular walls, with subsequent impairment in CSF flow and absorption. Although the risk factors for IVH are relatively well established— prematurity, low birthweight, and cardiovascular instability—the factors that influence which infants develop PHH are not as well understood. (10)(11) As with other types of hydrocephalus, PHH results in increased cerebrovascular resistance, decreased cerebral perfusion, and stretching and compression of brain parenchyma.

Additional causes of acquired hydrocephalus result from any abnormality that involves obstruction at the outlet of the fourth ventricle, the foramen magnum, or the tentorial notch. Posterior fossa tumors or cysts (arachnoid, colloid) can physically obstruct normal CSF flow. These may be congenital or acquired. Tectal gliomas that obstruct the aqueduct of Sylvius are not usually present in the neonate. Choroid plexus papillomas can present in the neonatal period and lead to hydrocephalus because of both obstruction and overproduction of CSF in the choroid plexus. Dandy-Walker malformation and its variants account for 2% to 10% of children with hydrocephalus. It is defined as hypoplasia of the cerebellar vermis with cystic dilation of the fourth ventricle. Hydrocephalus occurs because the posterior fossa cyst seen in this syndrome does not always communicate with the fourth ventricle outlet. Similarly, congenital aqueductal stenosis or membranous obstruction of the aqueduct can also result in hydrocephalus. It accounts for 20% of cases of hydrocephalus detected in fetuses. Blockage at the aqueduct is suspected when the lateral and third ventricles are enlarged, but the fourth ventricle is small. Identification of the anatomic site and cause of CSF obstruction is critical in determining the appropriate treatment. Some neonates may have a temporary absorptive problem and only require temporary CSF diversion, while others may need long-term shunting.

“Benign” external hydrocephalus (BEH) is a unique entity with less potential for adverse neurodevelopmental outcomes. It is defined by enlarging subarachnoid spaces, often over the frontal lobes, with normal or moderately enlarged ventricles and macrocephaly. On neuroimaging, there may be a widened interhemispheric fissure as well as an enlarged third ventricle and basal cisterns. It is usually diagnosed after 6 months of age and is rarely observed in newborns. An important distinction from true hydrocephalus is that BEH is not associated with increased ICP or associated clinical features. The macrocephaly associated with BEH is usually self-limited and typically normalizes after 1 year of age. It is therefore often managed expectantly and only rarely requires shunting or medical therapy. Some affected children have hypotonia and neurodevelopmental delays, particularly in gross motor functioning, though most of these children later catch up to their peers. Although BEH is idiopathic in most cases, it has been reported in the context of prematurity, meningitis, high venous pressure from cardiac dysfunction, and trauma. (12) It has been theorized that BEH may be related to immature arachnoid villi that are transiently unable to adequately absorb CSF. Approximately 40% of children with BEH have at least 1 relative with a head circumference in the 95th to 98th percentile. (13) Familial BEH is thought to have an autosomal dominant or multifactorial mode of transmission. (14) BEH is also associated with several other conditions including craniosynostosis, achondrodysplasia, Sotos syndrome, and glutaric aciduria type 1.

Clinical signs of hydrocephalus in the preterm or full-term infant are secondary to increased ICP and include a bulging fontanelle, splayed sutures, prominent scalp veins, irritability, lethargy, poor feeding, recurrent vomiting, high-pitched cry, seizures, and “sunsetting” eyes. The sunsetting appearance of the eyes, with upgaze paresis and downward displacement, results from excessive CSF pressure on the suprapineal recess of the midbrain and periaqueductal structures. Papilledema may be seen on ophthalmologic examination. In neonates, increasing CSF pressure affects the flexible skull and prevents appropriate fusion of the skull bones along suture lines, resulting in macrocephaly. A head circumference that enlarges rapidly in the neonatal period or is later observed to cross percentiles on a growth chart warrants a thorough evaluation with neuroimaging.

Prenatal diagnosis of hydrocephalus can be accomplished by fetal ultrasonography or magnetic resonance imaging (MRI). However, MRI is superior to ultrasonography in delineating neuronal myelination and any associated cerebral malformations. Specific MRI sequences to compensate for fetal movement are now available and allow for more precise evaluation of brain development and ventricular size. When measured with MRI, ventricle sizes are often larger than those observed on ultrasonography by 1 mm. (15) On fetal imaging, the transverse atrial width, often referred to as the atrial diameter, is the most useful measurement. An atrial diameter of less than 10 mm is considered normal. A width of 10 to 15 mm is diagnosed as mild to moderate ventriculomegaly, of which 14% progress to hydrocephalus, 57% remain stable, and 29% regress spontaneously. (16) A width of more than 15 mm is considered severe ventriculomegaly. (17) Ventriculomegaly seen on fetal imaging does not always result in a postnatal diagnosis of hydrocephalus as it may result from in utero destruction or maldevelopment of periventricular structures rather than hydrocephalus per se. Therefore, it is important to discern the etiology of fetal ventriculomegaly to best predict outcome and guide antenatal treatment.

Postnatal diagnosis and surveillance of hydrocephalus can be accomplished using cranial ultrasonography to assess ventricle size and serial head circumference measurement. Again, true hydrocephalus must be distinguished from “hydrocephalus ex vacuo,” which refers to compensatory enlargement of the cerebral ventricles and/or subarachnoid spaces without increased ICP. The latter is usually due to encephalomalacia and/or periventricular volume loss that has occurred in the setting of stroke. The open anterior fontanelle in newborns permits facile measurement of ventricle size using head ultrasonography; serial head ultrasonography is a more reliable surveillance tool than head circumference measurements.

Methods to standardize postnatal ventricle size measurements have been proposed by several groups. Levene defined the ventricular index as the distance from the wall of the lateral ventricle to the falx at the midcoronal level. (18) Growth curves for the progression of this index have served as the basis for timing of interventions in several studies. (19) Davies et al developed additional measures that evaluate ventricle size at multiple locations, which is plotted on standard curves to address the asymmetric ventricular dilation often seen in infants. (20) Finally, others have presented measures of ventricle size that are normalized to head utilization ratios at the frontal and temporal horn levels (fronto-occipital horn ratio and frontotemporal horn ratio). This approach has high inter-rater reliability and validated norms. (21)(22)

The mainstay of management of hydrocephalus is the relief of increased CSF and ICP. A number of medical and surgical strategies have been proposed and studied. We refer the reader to an excellent 10-part systematic review and evidence-based guidelines published for the management of pediatric hydrocephalus in the Journal of Neurosurgery: Pediatrics. This review includes the optimal treatment strategies for the management of PHH in the preterm neonate. (23) Decreasing CSF production with diuretics (either furosemide or acetazolamide) has been ineffective and potentially unsafe. (24)(25)(26) Similarly, use of intraventricular fibrinolysis in the setting of PHH has been explored in small studies with disappointing results and notable safety concerns. (26)(27) As a result, neither diuretics nor thrombolysis are recommended as treatments for PHH in preterm newborns. (23)

The remaining options involve surgical diversion of CSF. Decompression of the ventricles with a temporary device is often undertaken first, especially for small infants or when it is not clear if permanent shunting will be required. If the underlying hydrocephalus does not resolve, the temporary device may be removed and replaced with a fully internalized cerebral shunt. The precise timing of temporary and permanent CSF diversion in preterm neonates is still relatively controversial (23); only recently has there been a movement to formally standardize this intervention using an infant’s weight, ventricle size, and other specific clinical parameters. (28)(29)(30)(31) Moreover, results have been mixed with respect to benefits of early, aggressive intervention. A recent observational study in preterm infants of early versus late intervention (defined by ventricular index) showed that infants receiving early intervention had better outcomes that those with late intervention, even if they later required permanent shunting. The authors concluded that early intervention appears to outweigh potential risks. (31) However, data from a recent randomized trial of early versus late intervention (defined by ventricular index) did not show improvements in the composite outcome of permanent shunt placement or death. Neurodevelopmental outcomes for these patients are currently pending. (30) Additional randomized controlled trials are needed to help inform this important clinical question.

A number of temporary CSF diversion methods are available and used in neonates: ventricular access device (VAD), external ventricular drain (EVD), ventriculosubgaleal shunt, and serial lumbar punctures (LPs) (Fig 5). CSF may be aspirated by way of a VAD, EVD, or LP. An EVD is particularly helpful in diverting CSF in the acute stage of meningitis, and permits serial CSF sampling for culture. A VAD has less morbidity and mortality when compared with an EVD as well as a lower risk of infection when repeated CSF aspiration is undertaken. (32)(33)(34)(35)(36) Ventriculosubgaleal shunts may reduce the need for daily CSF aspiration. (37) A large, prospective study showed that ventriculosubgaleal shunts decrease the need for permanent CSF diversion. (38) Serial LPs are an effective, immediate treatment for elevated ICP but do not obviate the need for permanent CSF diversion; furthermore, they may contribute to infection. (23)(24)(25) Regardless of the specific mode, early CSF diversion is associated with a reduced need for permanent shunting and reduced risk of moderate-to-severe disability. (31)

For patients with PHH, temporary diversion and decompression may relieve ICP but does not address the pathophysiologic cascade following hemorrhage that contributes to chronic hydrocephalus. One approach that has been explored is temporary drainage and continuous irrigation of the ventricles with artificial CSF and a fibrinolytic agent (DRIFT therapy). (27) The original trial did not demonstrate a difference in the primary outcome of death or the need for a shunt and showed increased occurrence of secondary hemorrhage. However, follow-up of the cohort at later ages showed some improvement in developmental outcomes, suggesting that this intervention may warrant further exploration. (39) Current guidelines do not support use of this therapy. (23)

The goal of CSF diversion is to establish a permanent communication between the ventricular CSF and a distal anatomic space. The most common distal destination for CSF diversion is the peritoneum because of the large absorptive surface of the peritoneal lining (ventriculoperitoneal shunt). However, in patients who suffer from peritoneal pathologies (ie, necrotizing enterocolitis, abdominal ascites, peritoneal dialysis), diversion into the atria of the heart (ventriculoatrial shunt) or the pleural cavity (ventriculopleural shunt) may be necessary (Fig 6). Lumboperitoneal shunts are not typically used in neonates.

Shunts typically consist of a few parts: an inflow catheter connected by a valve mechanism to an outflow catheter. The inflow catheter begins at the ventricle or subarachnoid space, exits the skull, and travels under the skin to the chosen diversion site. A valve mechanism controls pressure and CSF flow through the tubing and is usually placed on the top of the head or behind the ear. Shunts may also include reservoirs for CSF sampling and/or medication administration.

Endoscopic third ventriculostomy (ETV), with or without choroid plexus cauterization (CPC), is a safe procedure with varying degrees of success in the treatment of neonatal hydrocephalus. It involves unilateral coronal suture entry to the third ventricle, followed by fenestration of the floor of the third ventricle, to allow CSF to be internally diverted to the basal cisterns and reabsorbed by arachnoid granulations. A rigid or flexible endoscope can be used, and an EVD is inserted for postoperative CSF diversion at the discretion of the operating surgeon. When performed successfully, the ETV stoma allows for an alternative route of CSF flow in obstructive hydrocephalus (Fig 6). Established predictors of ETV and CPC success are older age and greater degree of CPC. (40) It is thought that ETV is more successful in older infants as a result of restoration of CSF absorption at a later age. The ETV Success Score is a validated clinical prediction tool for estimating operative success based on age at surgery, cause of hydrocephalus, and previous history of shunt. (41)

Risks of ETV/CPC include vascular injury, neural structure injury, endocrine or electrolyte dysfunction, CSF leak, and infection. The volume of irrigation during an ETV must be monitored because excessive irrigation can contribute to cerebral herniation and cardiovascular compromise. Intraoperative injury can occur to the fornix, thalamus, hypothalamus, or oculomotor nerve and may be catastrophic. Transient ocular divergence and anisocoria sometimes occur postoperatively because of midbrain stretch. Manipulation or injury to the tuber cinereum during an ETV can result in cerebral endocrine dysfunction with transient or permanent diabetes insipidus. Given this risk, postoperative monitoring of electrolytes is critical. (42) Careful presurgical patient selection is the key to success because the risk of neural injury is higher with ETV than with shunt placement. Unnecessary ETV surgeries must, therefore, be avoided when possible.

Variations in the temporal and spatial progression of hydrocephalus have relevant clinical consequences. A slow progression of ventriculomegaly over months allows the brain to respond with intrinsic repair mechanisms via cellular plasticity. Preferential expansion of the lateral ventricle’s occipital horn can affect optic radiations and result in selective visual deficits without directly affecting motor function. (43)

Untreated neonatal hydrocephalus can be lethal, but with appropriate intervention, patients can survive well into adulthood with minimal complications. However, many shunts will fail over time. Shunt failure may be aseptic and involve obstruction, over- or underdrainage, or occult reasons for malfunction. It can also be related to infection, requiring removal of the entire shunt system and then replacement after appropriate antibiotic treatment. Long-term follow-up has shown shunt infection and obstruction rates as high as 16% and 79%, respectively. (44) When a shunt malfunctions due to obstruction, and revision is required, approximately 40% of patients will have event-free survival at 10 years.

Long-term studies have shown that individuals with shunt placement in childhood require approximately 2 to 4 revisions by the time they reach the third decade of life. (44)(45)(46) Unfortunately, the absence of shunt failure does not always equate to shunt independence because malfunction has been reported to occur up to 20 years after the initial shunt insertion. (46) The durability of the shunt tubing has not been studied well, but when discontinuity or kinking occurs along the shunt tract, shunt failure can result. If a patient ultimately becomes shunt independent, the shunt is not removed given that calcification and scarring along the tubing can increase surgical morbidity with removal. There is also a risk of damaging the highly vascular choroid plexus that may be scarred around or in close proximity to the proximal catheter.

Shunt placement can be complicated by over- or underdrainage. If a programmable valve was used at the time of shunt placement, the valve settings may be changed via noninvasive techniques to either increase or decrease CSF flow through the shunt as needed. However, in the case of nonprogrammable valves, an additional surgery is required to exchange the shunt valves. In neonatal hydrocephalus, overdrainage can result in slit ventricle syndrome and early suture closure (craniosynostosis). Valve regulation has reduced the incidence of over- and underdrainage, but both are important sequelae that must be monitored in the postoperative period. (47)

Shunt infections are often a result of skin flora, and stringent aseptic techniques must be used at the time of surgery to minimize and eliminate postoperative infection risk. Depending on the definition and duration of follow-up, infection has been reported to range from 0.2% to 8% per surgery. (48)(49) Risk factors of shunt infection are young age at the time of initial surgery, prematurity, IVH, and open neural tube defects as etiologic factors in hydrocephalus. (46)

Social integration and overall psychosocial well-being in patients with permanent CSF diversion relies largely on the underlying causes of neonatal hydrocephalus. (46) However, for preterm infants, ventriculomegaly remains one of the strongest predictors of impaired neurodevelopmental outcome. (50)

1. Neonatal, and pediatric hydrocephalus more broadly, is a complex disorder and there is no consensus on its appropriate treatment. A 10-part systematic review of the literature and guidelines was published in 2014 by the Pediatric Hydrocephalus Systematic Review and Evidence-Based Guidelines Task Force.

2. For PHH in the preterm neonate (23):

   • There is level 1 evidence (high clinical certainty) to recommend against the use of thrombolytic agents and diuretics.

   • There is level 2 evidence (moderate clinical certainty) suggesting that ventriculosubgaleal shunts reduce the need for daily CSF aspiration compared with VADs. There is level 2 evidence (moderate clinical certainty) to recommend against the use of serial LP to reduce the need for shunt placement or to avoid the progression of hydrocephalus.

   • There is level 3 evidence (insufficient) to recommend the use of ETV. In addition, there is insufficient evidence to recommend a specific weight or CSF parameters to direct timing of shunt placement.

   • Recently, there has been a movement to formally standardize the timing of temporary and permanent CSF diversion using an infant’s weight, ventricle size, and other specific clinical parameters. Supportive data to help develop these guidelines are being generated.

3. Although fetal hydrocephalus is increasingly detected on routine antenatal surveillance, there are no guidelines to support clinical practice related to serial imaging, parental counseling, or mode and timing (fetal vs neonatal) of surgical intervention. Owing to the complexities involved, treating clinicians may wish to consider consultation with and referral to centers with expertise in fetal diagnosis and surgical intervention.

BEH

benign external hydrocephalus

CPC

choroid plexus cauterization

CSF

cerebrospinal fluid

ETV

endoscopic third ventriculostomy

EVD

external ventricular drain

ICP

intracranial pressure

IVH

intraventricular hemorrhage

LP

lumbar puncture

MMC

myelomeningocele

MRI

magnetic resonance imaging

PHH

posthemorrhagic hydrocephalus

VAD

ventricular access device