Introduction

Germinal matrix hemorrhage (GMH) represents the most important adverse neurologic event for preterm and very low birth weight (VLBW) newborns during neonatal period [1, 2]. Pathogenesis is mainly attributed to germinal matrix (GM) inherent vascular fragility and fluctuations in cerebral blood flow (CBF) [3]. Among GMH, intraventricular hemorrhage (IVH), defined as blood leakage into the ventricular space, is the main form of presentation [1] with subsequent development of posthemorrhagic hydrocephalus (PHH) in 35% of cases. Approximately 20% of preterm and VLBW newborns develop IVH. Higher-grade hemorrhages are more common as age and weight decrease [4]. More than 50% of hemorrhages will develop on the first 24 h postpartum, and 90% of them within the first week after partum [5]. Diagnosis is made by cranial ultrasound performed at bedside using the Papile grading system severity scale [6]. Grade I IVH minimally involves the ventricle (< 10%) or is restricted to subependymal parenchyma, grade II IVH extends into the ventricle but does not expand or occupy more than 50% of it, and grade III involves more than 50% of the ventricle and distends it [7]. Periventricular hemorrhagic infarction (PVHI), previously classified as grade IV IVH, refers to an extensive IVH with parenchymal involvement, likely due to terminal venous occlusion with venous infarction and subsequent secondary hemorrhage (Table 1) [8].

Table 1 Comparison of grading systems; GM, germinal matrix; PVHI, periventricular hemorrhagic infarction [6, 8]
Full size table

Risk factors

Incidence of IVH in preterm and VLWB infants had a relevant decline from the 1980s to the late 1990s, mainly associated with routine implementation of antenatal steroids [10, 11]. Unfortunately, occurrence in the past two decades has remained stationary [12, 13]. Several environmental and medical risk factors have been implicated in IVH pathogenesis and progression, most of them associated with alterations in cerebral blood flow and central nervous system blood pressure [13]. In this article, we present the verified risk factors and discard those that remain controversial or lack data to sustain them (Table 2) [1, 10, 12,13,14].

Table 2 Classification of risk factors [1, 10, 12,13,14]; GM, germinal matrix; IL, interleukin; TNF, tumor necrosis factor
Full size table

Pathogenesis of GMH-IVH

Two major factors are thought to contribute to IVH in the preterm infant: the inherent fragility of the GM vasculature and disturbances of the cerebral blood flow (CBF) [13].

The high risk of hemorrhage of the GM vasculature has been attributed to a high vascularity and a fragility of the vessels caused by alterations in the components and stability of the blood-brain barrier (BBB). Because of its high metabolic activity and high oxygen demand, this area is in a relative hypoxic state [13]. Hypoxia upregulates the production of vascular endothelial growth factor (VEFG), which, together with angiopoietin-2 (ANGPT-2), is responsible for the rapid angiogenesis in the GM. High vascularity enhances the probability of hemorrhage [16,17,18,19]. Additionally, differences in the composition of the vasculature of the GM increase its predisposition to bleed. A reduction of the glial fibrillary acidic protein (GFAP) in the astrocyte end feet affects the mechanical strength of the vessels [18]; a decrease of the fibronectin expression in the basal lamina alters its structural stability [20]; and a reduced signaling of TGF-β1 decreases the density of pericytes in the GM vessels [17]. It has been observed that antenatal glucocorticoids help prevent GMH by increasing GFAP and fibronectin [18, 20].

Disturbances of CBF due to the immaturity of the cardiovascular system, together with an impaired cerebrovascular reactivity, increase the risk of GMH [1, 21]. Cerebral autoregulation is a physiological mechanism that maintains constant blood flow to the brain despite variations in cerebral perfusion pressure [22]. In preterm infants, cerebral autoregulation may be impaired, so their CBF becomes BP-passive [23]; this means that their CBF will change in concordance with BP changes. It is speculated that cerebral pressure-passivity is a risk factor for cerebrovascular injury [21, 24]. Hypercapnia might play a role in the impairment of cerebral autoregulation by producing vasodilation of the cerebral resistance arterioles. In the setting of an increased BP and hypercapnia, hypercapnic hyperemia will overcome the autoregulatory mechanism resulting in an increased CBF [22].

It was observed that infants who required mechanical ventilation experienced more fluctuations in BP and CBF velocity secondary to asynchrony with the respiratory effort and the ventilator. The use of pancuronium in the first 24 h removed infant’s contribution to the ventilation, thereby reducing CBF velocity fluctuating pattern along with the incidence and severity of IVH [25]. Nonetheless, this drug was found to have multiple cardiovascular side effects and to cause an increase in the need for higher ventilator pressures to maintain adequate oxygenation [26]. Some ventilator modalities, such as synchronized intermittent mandatory ventilation and assist control, can reduce this asynchrony between the patient and the machine [13, 26].

Hypotension is the most common BP alteration found among preterm infants [27]. Numerous studies have found a direct association between hypotension and an increased risk of IVH [21, 23, 28]. Hypotension produces vasodilation of the brain vasculature [21], thereby increasing the risk of vessel rupture and hemorrhage. Others have found an association between GMH and a hypoperfusion-reperfusion pattern [3, 28, 29].

Pathogenesis of posthemorrhagic hydrocephalus

The mechanism by which hydrocephalus develops after GMH-IVH is still uncertain. A most common held view is that initially blood clots obstruct the cerebrospinal fluid (CSF) flow by blocking the cerebral aqueduct or the fourth ventricle outlets, thus producing an obstructive hydrocephalus [30]. Later on, a delayed communicating hydrocephalus develops, and it is speculated that it is due to an impairment of CSF resorption caused by an increased production of extracellular matrix (ECM) proteins throughout the cerebroventricular system [1, 31]. Increased TGF-β1 in CSF might play a role in the pathogenesis PHH by upregulating the genes encoding ECM proteins, such as fibronectin and collagen [1, 32, 33]. Others have proposed that thrombin may play a role as well [34, 35]. Some evidence supports the role of iron in the pathogenesis of PHH. By generating hydroxyl radicals and inducing oxidative damage, iron may cause ventricular dilation and neuronal death [2, 9]. The use of iron chelators was found to attenuate ventricular dilation by reducing iron overload after GMH and iron-induced brain injury [2, 36]. Ventricle dilation can lead to hypoxia, ischemia, decreased cerebral perfusion, increased free radicals, and ultimately white matter destruction [30].

Diagnosis

The clinical presentation of IVH typically follows three patterns: (a) catastrophic deterioration which occurs over minutes or hours and mimics the rapid neurological deterioration of an older patient with large intracranial hemorrhages (associated with a very poor prognosis); (b) the saltatory course which evolves over hours to days and includes decreased alertness and activity, hypotonia, abnormality tight popliteal angle, abnormal eye movements, and respiratory difficulties; (c) and a clinically silent course, which supports the use of surveillance cranial ultrasonography [37].

An excessive head enlargement is defined as a persistent increase of 2 mm per day. When measuring the head circumference, it is not easy to detect a difference day to day but rather with the cumulative head growth from several days, or when there is additional evidence of raised intracranial pressure like increased splaying of the cranial sutures, a full tense fontanel and worsening of the apnea and bradycardia episodes, lethargy, and feeding intolerance [7]. These clinical findings have a limited reliability. However, the most reliable clinical indication of increased pressure is progressive splaying of the sagittal suture [1].

Visualization of the cerebral ventricles is essential in investigation of the child with suspected hydrocephalus. Real-time ultrasound (RT-US) has been widely accepted as a reliable method to diagnose intraventricular hemorrhage and hydrocephalus in infants, the latter defined as a ventricular enlargement with index > 97th percentile. The limitation was that there was no data on the size of the ventricular system in preterm infants; for this reason, Levene performed a study where he produced reference ranges for ventricular index according to the gestational age (from week 26 to week 42), using RT-US [38, 39]. The ventricular index is measured from the falx to the lateral wall of the body of the lateral ventricle. An index of 4 mm over the 97th centile (p97+4) has been widely accepted as the “action line” to start the interventions (Fig. 1) [7].

Fig. 1
figure 1

Reference values of ventricular width. Taken from Whitelaw A. and Aquilina K. [7]

Full size image

Sometimes, ventricles do not expand laterally but become rounded or expand occipitally. Therefore, different reference ranges for anterior horn width and third ventricular width, measured in coronal plane, as well as the thalamo-occipital dimension, measured in the sagittal plane are now used as an alternative definition of posthemorrhagic ventricular dilation (PHVD): anterior horn width > 4 mm (1 mm over 97th centile), thalamo-occipital dimension > 26 mm (1 mm over the 97th centile), and third ventricular width > 3 mm (> 1 mm over 97th centile) [7].

While MRI is more sensitive in identifying subtle white matter damage associated with prematurity [40], rapid ultrasound provides a method for measuring ventricular size and allows diagnosis of dilated ventricles given gestational age, and regular assessment of size permits early detection of deviation in ventricular growth [38]; it is also a better choice for preterm and unstable infants during the first week of life [40].

The diagnostic method of choice is still RT-US because of its sensitivity and specificity in the diagnosis of this pathology, and because of the benefit of being a bedside study. A disadvantage of this method is the fact that it is operator dependent and that it does not confer continuous monitoring. New methods have been proposed for the diagnosis and monitoring of PVHD, such as Doppler ultrasound, which provides additional information about intracranial pressure and parenchymal perfusion [7, 41]; flash visual evoked potentials and amplitude-integrated electroencephalography that can aid in the monitoring of increased intracranial pressure even before clinical manifestations appear [42], and 3D ultrasound which can give us a more accurate characterization of ventricle dilation measures [43].

Treatment

Currently, the only intervention that seems to be effective in the prevention of IVH is the use of antenatal steroid therapy. Many medical interventions, such as postnatal phenobarbital and prophylactic indomethacin, have been proposed for the control and management of IVH but have not proven to be effective. For the moment, the optimal management of IVH is based on the early diagnosis with serial RT-US, control of ventricular dilation by different invasive procedures to drain the excessive CSF out of the ventricular system, and prevention of future complications which are directly related to the degree of ventricular dilation and subsequent parenchymal damage. Definitive treatment of progressive PHH is made with the placement of a VP (ventriculoperitoneal) shunt; however, this procedure is rarely used as a first intervention for several reasons: the risk of skin ulceration in VLBW infants, the high incidence of shunt malfunction, and frequent need of revision, among others. In many cases, the ventricular dilation will be controlled with the use of temporizing neurosurgical procedures (TNPs): lumbar punctures (LP), external ventricular drainage (EVD), insertion of a ventricular access device (VAD), or a ventriculosubgaleal (VSG) shunt. In cases where PHH is not controlled with these procedures, the placement of a VP shunt is indicated [1, 7]. The lack of conclusive recommendations for the treatment of this pathology makes it very difficult to have a standardized protocol; most centers use institutional guidelines, medical expertise, and available resources to guide their treatment (Table 3).

Table 3 Summary of the treatments of IVH; LP, lumbar puncture; VSG, ventriculosubgaleal; EVD, external ventricular drainage; VAD, ventricular access device; VP, ventriculoperitoneal; DRIFT, drainage, irrigation, and fibrinolytic therapy; PHH, posthemorrhagic hydrocephalus [43,44,45,46,47,48,49,50,51,52,53,54,55]
Full size table

Pharmacological treatment

Postnatal phenobarbital

Considering the role CBF fluctuations play in the pathogenesis of GMH, it was hypothesized that the administration of phenobarbital would reduce the incidence of IVH by stabilizing the blood pressure. However, Cochrane review meta-analysis failed to demonstrate a reduction in the risk of IVH and showed an increase on the need for mechanical ventilation [44].

Prophylactic indomethacin

Indomethacin has shown to reduce the risk of severe IVH; however, its effectiveness is limited to short-term outcome as it has not shown to reduce mortality or to improve neurodevelopmental outcomes [45].

Diuretic therapy

Acetazolamide and furosemide decrease the production of CSF, and the use of these drugs was hoped to decrease the need of surgery and to increase disability-free survival [56, 57]. A trial conducted by the International PHVD Drug Trial Group included 177 infants that underwent treatment with acetazolamide (100 mg/kg/day) and furosemide (1 mg/kg/day) once their ventricular width was 4 mm above 97th percentile. The results of this trial were disappointing with an increased risk of death, neurological morbidity, and shunt placement in comparison to the control group [57]. The use of acetazolamide and furosemide in posthemorrhagic hydrocephalus is not recommended [58].

Intraventricular fibrinolytic agents

It was hypothesized that the early use of a fibrinolytic agent could prevent the progression of permanent hydrocephalus, thus reducing the need of a permanent shunt [59]. Studies have been conducted using various intraventricular agents, recombinant tissue plasminogen activator (rTPA), streptokinase, and urokinase, and results have failed to prove effectiveness [58]. Additionally, studies with streptokinase were found to increase the risk of meningitis and secondary IVH [59]. The use of intraventricular fibrinolytic therapy is not recommended [58].

Temporizing neurosurgical procedures for CSF removal

Lumbar and ventricular punctures

It was hypothesized that the repeated CSF tapping would reduce intracranial pressure and remove proteins and blood, thereby allowing the flow and absorption of CSF. Controlled trials have failed to demonstrate any benefit in the reduction for VP shunt placement [46, 58]. Some of the advantages that this procedure confers are an immediate CSF drainage with the possibility of sampling [58], without the need of moving the patient to an operating room, with no need of a neurosurgeon to perform the procedure. However, every tap can only evacuate 10 ml/kg of CSF per tap, which might not be enough to decrease the ventricular dilation [47]. Recent recommendations suggest performing a maximum of 3 LPs, and if ventricular dilation progresses, another intervention should be considered [48]. Repeated taps are associated with an increased risk of CSF infection (meningitis or ventriculitis) and increased risk of shunt infection [46, 56]. Although repeated LPs are not recommended as a definitive treatment, they may be useful for drawing of CSF as an immediate treatment of increased intracranial pressure (ICP) [58]. An exception can be made when the ventricles are not connected with the lumbar subarachnoid space [7], like in the case of aqueductal stenosis.

Ventricular punctures are not a procedure recommended in the control of ventricular dilation because of the high risk of porencephalic cyst development and higher CSF infection rates [46]. This procedure should only be considered in cases of extreme urgency, when the patient’s life is at risk and there is no other method available [47].

External ventricular drains

EVDs consist in the insertion of a catheter in the dilated anterior horn of the right lateral ventricle that externally connects to an adjustable draining system. The main advantage offered by this technique is its effectiveness in evacuating sufficient CSF volume in a continuous manner to reduce intracranial pressure. Several authors have reported different rates of infection, ranging from very low rates to unacceptable levels (5.4 to 40%) [49, 56, 60]. Controversy exists regarding the relationship between the duration of catheterization and risk of infection, and most studies consider extended catheter duration, usually exceeding 5 days, to be an important risk factor for subsequent infection [50, 51]. Other problems, such as over drainage and development of subdural hygromas, have also been reported [49, 56, 60].

Ventriculosubgaleal shunt

VSG shunt procedure consists on the insertion of a ventricular catheter into the dilated lateral ventricle. The catheter is then fixed to the dura and placed in a subgaleal pouch that is made by the surgeon during the procedure (Fig. 2) [49]. This low pressure and continuous and more physiological drainage of CSF helps to decrease the intracranial pressure and potentially decrease shunt dependence by continuously draining hemorrhagic debris, thus reducing further progression of hydrocephalus [56, 61]. Some of the benefits that this procedure confers over VADs are the decrease in need of daily taping, maintenance of a closed system in which no fluids and electrolytes are lost, and a potential earlier discharge [52, 61]. An additional advantage seen in this procedure is its low cost. Reported complications of VSG shunt are CSF leakage from the incision site, meningitis, malfunction, migration of the catheter from the ventricle or slippage into the ventricle, and intraparenchymal hemorrhage. The rate of leakage varies from 4.7 up to 32%, infectious complications occur in approximately 8% of the cases, while the other complications are rarely seen. Permanent VP shunt is necessary in 60–85% of cases [49].

Fig. 2
figure 2

Ventriculosubgaleal shunt

Full size image

Ventricular access devices/reservoirs

The use of ventricular reservoirs allows controlled, intermittent transcutaneous CSF withdrawal to be performed on as-needed basis [52, 62]. Intermittent CSF removal from the reservoir might promote the return of normal CSF absorption, thereby reducing shunt dependence (Fig. 3) [52]. However, the conversion rates to VP shunt reported in the literature range from 75 to 88% [63]. Potential benefits include the facility to obtain CSF to examine it and the possibility to simultaneously infuse medication and avoid the BBB [62]. Some complications reported from this procedure include increased risk of CSF infection, which occurs in approximately 9.5% of the cases and might be due to continuous tapping that exposes the intraventricular environment to the exterior [49, 62, 64], CSF leakage, skin necrosis, and loss of electrolytes and fluids. The appearance of hyponatremia after repeated CSF tapping has been reported [65] and in cases of repeated LPs and ventricular taps [66]. It is essential to identify the electrolyte alteration and replace sodium loss promptly in cases of frequent CSF tapping [65, 66]. Despite possible complications, evidence has demonstrated that the use of VAD decreases morbidity and mortality compared with EVDs [58].

Fig. 3
figure 3

Ventricular access device

Full size image

DRIFT

The DRIFT (drainage, irrigation, and fibrinolytic therapy) randomized trial involved 77 infants with PHH, in which irrigation and drainage were performed through the placement of two catheters, one in the right frontal and the other in the left occipital, and fibrinolysis was carried out with rTPA. The rationale behind this protocol was to eliminate the blood clots obstructing the absorption pathways of CSF and to decompress the distended ventricles by draining the CSF containing proinflammatory cytokines, iron, and free radicals, thus decreasing secondary injury to the brain. The aim was to decrease the mortality and the severe cognitive, motor, and sensory disabilities [67]. The trial was suspended because of the increased risk of secondary bleeding. However, developmental outcome at 2 years showed a reduction of death and severe cognitive disability in survivors in comparison to the control group [68]. DRIFT is not recommended for the treatment of PHH [56].

Endoscopic ventricular lavage

Endoscopic ventricular lavage has been proposed as treatment of PHH. The hypothesis of Schulz et al. was based on the 2-year outcome results of DRIFT trial, and the removal of the hematoma would eliminate the proinflammatory cytokines, iron, and free radicals and reduce the risk of progression of the hydrocephalous. Several studies have shown a decreased conversion to VP shunt in those who undergo an early endoscopic ventricular lavage; however, these studies have potential limitations, including the design of the studies and the small patient sample. This procedure is technically feasible but there is insufficient evidence to recommend its use [69, 70] .

Endoscopic third ventriculostomy

A third ventriculostomy is generally indicated for obstructive hydrocephalus, as seen in the initial phase of PHH. However, controversy exists with the use of this procedure in children under age of 2 years; failure rates are higher among infants under age of 6 months [71]. There is insufficient evidence to recommend the use of this procedure as a treatment for PHH [58].

Ventriculoperitoneal shunt

Despite the fact that VP shunt is the best definite treatment for progressive PHH [56], many complications of the procedure may be seen when used as the primary therapy in VLWB infants [7]. There is a high risk of shunt obstruction due to the high concentrations of proteins and blood in the CSF, which leads to a higher rate of shunt revisions. Because of the fragile and thin skin of the VLWB infants, there is a high incidence of ulceration of the skin above the valve [56]. There are no standardized criteria to define when to convert to a permanent shunt [58]. Nonetheless, the following must be present for a VP shunt to work: a mature immune system, an adequate absorption capacity of the abdomen, the effective elimination of blood products from CSF flow, and the sufficient thickness of subcutaneous tissue [58]. The most important parameter that can be monitored and used as guidance to decide when conversion is feasible is the body weight. Some authors recommend waiting until the baby reaches 2.5 kg [7], others suggest the surgery once the infant weighs 2 kg [72, 73], and there are reports of good outcomes if surgery is performed in patients weighing over 1.5 kg [53, 56, 74, 75]. Because of the high incidence of shunt obstruction, it has been proposed that protein concentration should be low to avoid these malfunctions. There are different recommendations as to what is the ideal CSF protein content. A retrospective study of Fulkerson et al. showed no statistical significance with an increased risk of shunt failure and the CSF content of protein [76]. Some studies report better results when waiting for protein concentration to decrease, some recommend to wait until proteins are below 1.5 g/l [7], others suggest to wait until they are less than 1000 mg/dl [49], while others prefer to wait until protein concentration is below 200 mg/dl due to a decrease risk of shunt malfunction and the need of short interval revisions [56, 74, 77]. The authors of this article use as parameters of conversion a weight above 1.5 kg and proteins below 200 mg/dl in a study, with results pending publishing, and we suggest for future studies to register these parameters to make comparisons. Additionally, it should not be considered as a first-line treatment since not all patients will require a VP shunt conversion as the TNP will suffice to control the hydrocephalus; a correlation can be made with the grade of IVH at the time of the diagnosis and the need for VP shunt. Different studies have reported that only 1% of their patients with IVH grades 1 and 2 required shunt placement, while grade 3 and grade 4 ranged between 18 and 29% [78, 79].

Ventriculoatrial shunts can be considered as an alternative method if a contraindication for VP shunt exists; however, a higher rate of shunt failure has been reported [56, 73].

At the moment, an optimal standardized protocol to ensure a better neurodevelopmental outcome remains to be determined. There are controversial results on which of the TNP is best, taking into account the risk of infection, known to be an independent factor for poor neurodevelopmental outcome [8], and the need for VP shunt conversion. Some studies support that VP shunt placement is directly associated with poor neurodevelopmental outcome, while others relate the adverse outcome to the extent parenchymal involvement [15], with the VP shunt being an indirect factor of poor prognosis associated to the need of its placement due to progressive ventricular dilation and subsequent parenchymal damage. Either way, the conversion to VP shunt is a way to evaluate the effectiveness of TNPs. The most commonly used TNPs are VSG and VAD, and many factors have been analyzed to determine which is superior. Badhiwala et al. compared the risk of infection, obstruction, and surgical revision between these two methods and reported a slightly lower incidence in infection with VSG (9.2% vs. 9.5%), but a lower obstruction rate (7.3% vs. 9.6%) and need of surgical revision (10.8% vs. 12.2%) when using VAD. However, these results lacked statistical significance [64]. No conclusion can be made with these results since there are many variables that may influence the need of permanent shunt, such as the time of intervention and the decision to convert to a permanent shunt [52, 54].

There is still no consensus on when best to intervene once the hydrocephalus has developed. In practice, the decision is made based on the head circumference growth, the fontanel bulging, and the presence of neurological signs of increased ICP. Levene’s ventricular index is also used to make the decision, and usually treatment is started once the ventricles are > p97+4 [56]. Recent trials propose that early treatment (> 97p to 97p+4) could have an impact in the neurodevelopmental outcome and shunt-free survival rate, and early treatment is defined as an intervention done after the ventricular dilation has reached the 97th centile but before crossing the > 4 mm above 97th centile threshold [15, 55, 78]. Between 1992 and 1996, de Vries et al. made a study involving 95 infants, 22 underwent no treatment because ventricular dilation stopped above the 97th centile, 31 were treated early, and the rest started treatment once they crossed > 4 mm above 97th centile (late intervention). Results of this study showed that more infants of the late intervention group required VP shunt placement and had worst neurodevelopmental outcomes at the 24 months of follow-up [55]. An observational study conducted by Brouwer et al. reported similar results, infants in whom treatment was started before crossing > p97+4 had a better developmental quotient at 2 years of follow-up [78].

Developmental outcomes

As mentioned before, IVH increases the risk of cerebral palsy (CP), motor and mental disabilities, and death. However, neurodevelopmental outcome among preterm infants who suffered IVH differs widely. Many trials have been conducted to identify which factors determine the outcome.

A few authors stablished that a poorer outcome was related to the severity grade of IVH, being worse in those with grade IV IVH [80]. Others have stated that the use of VP shunt is an indicator of poor prognosis and that its insertion represents an additional risk factor [79, 81]. More recently, it was identified that the injury of white matter was a determinant for prognosis [82], and based on this assumption, different studies have proposed an early intervention to control ventricular dilation and reduce the risk of white matter injury, having better results in those who were intervened early [39, 55, 78].

With the results of different studies, the authors of this article infer that the developmental outcome is most influenced by the involvement and damage of white matter lesions, and that IVH severity, shunt insertion, and increasing ventricular size are factors that contribute to this damage rather than being determinants for later developmental problems.

Conclusion

IVH is a rare pathology with a complex evolution and usually with a poor outcome. More evidence is required, regarding new technology and results of ongoing clinical trials, to create a standardized protocol that can ensure the best possible outcome for these patients. Special attention must be taken if risk factors are identified and treatment should be decided in interdisciplinary context.