White matter injury in infants with intraventricular haemorrhage: mechanisms and therapies

Abstract

Intraventricular haemorrhage (IVH) continues to be a major complication of prematurity that can result in cerebral palsy and cognitive impairments in survivors. No approved therapy exists to prevent IVH or to treat its consequences. IVH varies in severity and can present as a bleed confined to the germinal matrix, small-to-large IVH or periventricular haemorrhagic infarction. Moderate-to-severe haemorrhage dilates the ventricle and also damages the periventricular white matter. This white matter injury (WMI) results from a constellation of blood-induced pathological reactions, including oxidative stress, glutamate excitotoxicity, inflammation, perturbed signaling pathways and remodeling of the extracellular matrix. Potential therapies for IVH are currently undergoing investigation in preclinical models, and evidence from clinical trials suggests that stem cell treatment and/or endoscopic removal of clots from the cerebral ventricles could transform the outcome of infants with IVH. This Review presents an integrated view of new insights into the mechanisms underlying WMI in premature infants with IVH and highlights the importance of early detection of disability and immediate intervention in optimizing the outcomes of IVH survivors.

Introduction

Germinal matrix haemorrhage (GMH) and intraventricular hemorrhage (IVH) are the most common neurological disorders of preterm infants, occurring in about 12,000 infants every year in the USA.^1,2 The major neurodevelopmental sequelae of IVH include cerebral palsy, cognitive deficits and hydrocephalus. With technological and scientific advances in perinatal and neonatal care, the survival of extremely premature infants (<1,000 g birth weight) has improved remarkably and neonatal morbidity profiles in IVH survivors have shifted towards milder neurocognitive impairments.^3,4 The rate of severe cerebral palsy has declined, whereas the rate of mild cerebral palsy has increased.⁵

IVH typically originates in the germinal matrix (also known as the ganglionic eminence) of premature newborn babies at 23–32 weeks gestational age (Fig. 1). As a germinal matrix bleed enlarges, the underlying ependyma breaks, filling up the ventricle with blood. The diagnosis of IVH is made on screening cranial ultrasonography, which is performed in extremely premature infants (≤32 weeks gestation) in neonatal intensive care units (NICU) during the first week of life. Most of these infants are asymptomatic during their stay in the NICU, although some might manifest subtle abnormalities in consciousness, movement, tone, respiration or eye movements.

Figure 1. Germinal matrix and intraventricular haemorrhage in human preterm infants.

a | Grade I germinal matrix haemorrhage. Microscopic image of hematoxylin-eosin stained section showing haemorrhage (arrowheads) confined to the germinal matrix and not extending to the ventricle in a 25-week preterm infant at postnatal day 3. Arrows indicate the ependymal layer. b | Grade II intraventricular haemorrhage (IVH). Coronal section of the forebrain of a 25-week infant showing blood (block arrows) in the germinal matrix and lateral ventricle. c | Grade II IVH. Coronal section of the forebrain of a 24 week infant with IVH and ventricular dilatation (red arrows). d | Grade IV periventricular haemorrhagic infarction (PHVI). Coronal section of the forebrain of a 28 week preterm infant revealing IVH with ventricular dilatation of ventricle (red arrow) as well as an infarct dorsal to the ventricle (PVHI, block arrow).

The severity of IVH is variable and is classified according to the ultrasound scan findings as haemorrhage confined to the germinal matrix (grade I), IVH without ventricular dilatation (grade II) or IVH with acute ventricular dilatation (grade III).^6,7 The presence of an intraparenchymal echodense region is often referred to as grade IV IVH; however, such lesions usually represent periventricular haemorrhagic infarction (PVHI), which is attributed to compression of the medullary veins draining the white matter rather than the consequence of IVH extension.^6,8,9 Accumulation of blood in the cerebral ventricles of infants with IVH also damages the adjacent white matter. Autopsy studies of brain samples have revealed that 50–78% of premature infants with IVH show evidence of white matter injury (WMI).^10–12 About 45–50% of infants with moderate-to-severe IVH survive, whereas more than 80% of infants with grade I–II IVH survive.^13,14

In the USA, after discharge from the NICU, infants with IVH are followed in the neonatal follow-up clinic and in publicly funded Early Intervention Programmes for disabled children. Survivors of IVH often develop neurodevelopmental impairments, including motor, cognitive, speech, hearing and visual deficits. About 15% of all survivors of IVH develop cerebral palsy and 27% have moderate-to-severe neurosensory impairments at 18–24 months of age.¹³ Morbidity is increased/high in infants with moderate-to-severe IVH and about one-half to two-thirds of them develop features of hydrocephalus, cerebral palsy, cognitive deficits and/or intellectual disability.^15–17 About one-quarter of non-disabled survivors of IVH develop behavioural disorders and problems with executive function in infancy and childhood.^15–17 As these infants grow into childhood and adolescence, they often show symptoms of neuropsychiatric disorders such as attention deficit hyperactivity disorder (ADHD), major depressive disorder episodes, obsessive-compulsive disorders and epilepsy.^16–18 Studies of the psychiatric sequelae of IVH in adolescents who developed IVH following preterm birth show that these children have an increased risk of ADHD and tic disorders at 6 years of age, as well as an increased risk of major depressive disorder and obsessive-compulsive disorder at 16 years of age.^18,19 Parenchymal lesions and ventricular enlargement exacerbates the risk of these neurobehavioural deficits.¹⁸ Thus, IVH adversely affects brain maturation and the trajectory of growth and development. Despite the high incidence of this disorder and its high societal and financial burden, the treatment of disabled survivors of IVH is currently limited to physical, occupational and speech therapy aimed at optimizing their motor, cognitive and communication skills.

In this Review, we discuss novel mechanisms of WMI in IVH as well as the role of neuroimaging in diagnosis of IVH and prediction of IVH outcome. Up-to-date neurodevelopmental data are included to illustrate the need for early diagnosis of IVH and timely intervention. Mechanism-based strategies successfully employed in preclinical models and ongoing clinical trials to improve the neurological outcome of these infants are also discussed.

Patterns of IVH-induced injury

The major neuropathological findings associated with IVH include germinal matrix damage, acute and post-haemorrhagic dilatation of the lateral ventricles and PVHI.^6,20,21 On gross examination of the brain at autopsy, blood clot(s) can be seen in the ventricle, germinal matrix or brain parenchyma of infants who lived only for a short period after the onset of IVH. In infants who survive, resorption of the clot occurs over several weeks. A small clot in the germinal matrix is often replaced by a germinolytic cyst, whereas PVHI evolves into either a single cavity communicating with the lateral ventricle or multiple cysts that remain separate from the lateral ventricle. Dilatation of the lateral ventricles indicates post-haemorrhagic ventricular dilatation (PHVD).

Typical microscopic findings during the early stage of IVH evolution are apoptosis, necrosis and cerebral oedema along with infiltration of leukocytes and microglia. Subsequently, scarring and astrogliosis develop, and reactive astrocytes, haemosiderin-laden macrophages and calcification are seen in the periventricular white matter and germinal matrix.^11,20 Immunostaining reveals an increased number of non-myelinating pre-oligodendrocytes, a diminished number of mature oligodendrocytes, and hypomyelination of the white matter at near-term age (35–42 weeks’ gestation).²² Haemosiderin and iron deposition can be observed using specialized stains.

PVHI is characteristically observed in infants with large IVHs and has been reported in 15% of all infants with IVH (Fig. 2).²⁰ This haemorrhagic necrosis of periventricular white matter is typically unilateral and asymmetrical. PVHI can involve the ventricular and/or subventricular zone and adjacent white matter of mostly the frontoparietal lobe.²³ These infarcts subsequently evolve into a single porencephalic cyst or multiple periventricular cysts. The pathogenesis of PVHI is attributed to compression of the terminal vein and its tributaries, which drains the cerebral white matter and germinal matrix.^6,20 In one postmortem case series of infants who died with IVH, 75% had associated WMI.¹⁰ WMI is often haemorrhagic in infants with IVH.^10,24 Ischaemic WMI and pontine neuronal necrosis frequently accompany IVH, but are not necessarily caused by it.²⁵

Figure 2. Sequential neuroimaging studies of a preterm infant with bilateral intraventricular haemorrhage and periventricular haemorrhagic infarction.

a | Cranial ultrasound (coronal view) shows a bilateral germinal matrix haemorrhage (arrowheads) and periventricular haemorrhagic infarction (PVHI), seen as a large echogenic lesion in the white matter (arrows) in a preterm infant of 29 weeks gestational age. b | A T2-weighted MRI sequence performed on day 10 confirmed the PVHI (arrows) and also shows some abnormalities (arrowhead) in the white matter on the contralesional side. c | A subsequent T1-weighted MRI scan performed at term-equivalent age shows an area of cavitation (arrows) and asymmetry of myelination of the posterior limb of the internal capsule (PLIC). d | A direction-encoded collar map confirms asymmetry of the PLIC (seen as blue, rostral-caudal direction), with lower fractional anisotropy suggestive of reduced myelination on the affected side.

The presence of IVH causes axonal degeneration and damages oligodendroglial progenitor cells (OPCs) resulting in reduced myelination of the white matter²⁶. Accordingly, MRI studies have shown evidence of reduced myelination in premature infants with IVH compared to infants without IVH.^27,28 Diffuse axonal injuries have been reported in the white matter adjacent to large necrotic lesions in human infants.²⁹ Consistent with these human studies, studies in rabbit and rat models of IVH have demonstrated reduced myelination and the presence of axonal degeneration in the white matter of preterm animals with IVH relative to controls without IVH (Fig. 3).^30–32

Figure 3. Degenerating axons and reduced numbers of myelinated axons in an animal model of IVH.

a | Representative brain coronal section immunolabeled with amyloid-β precursor protein (APP)-specific antibody showing swelling along contiguous axons (varicosity, arrowhead) or terminal bulbs suggesting axonal damage in periventricular corona radiata of premature rabbit kits (embryonic day (ED) 29; term 32 days) with intraventricular haemorrhage (IVH) at postnatal day 1 and day 3, but not in day 3 controls without IVH. b | Fluoro-Jade-C-labelled coronal section showing neurofilaments and varicosities (arrowheads) in the periventricular corona radiata of prematurely born rabbit kits with IVH at postnatal day 7 and day 14, but not in day 14 controls without IVH. c | Electron micrograph showing fewer myelinated axons (yellow arrows) in the corona radiata of a kit with IVH compared to a control kit without IVH, both at postnatal day 14. Features of axonal degeneration can also be seen in the corona radiata of the kit with IVH, including an intra-axonal vacuole (orange arrowhead), granular disintegration to total loss of microtubules and rupture of the axolemma (black arrows). Scale bar 1μm.

Pathogenesis

Inducers of white matter injury

Collection of blood into the lateral ventricle destroys the periventricular germinal matrix, damages the corpus callosum and corona radiata of the white matter and causes mass effects owing to compression of periventricular structures. IVH increases intracranial pressure, reduces cerebral blood flow and cerebral oxygen metabolism, and also seems to disrupt the blood–brain barrier and permeability of the blood–CSF barrier.^33,34 The haematoma formed in the cerebral ventricles and brain parenchyma releases thrombin, complement, haem and iron, which are toxic to the brain (Fig. 4). Early brain injury is produced by plasma components of the blood, including thrombin, complement, immunoglobulins and other bioactive molecules. Subsequently, the cerebral injury is exacerbated by lysis of red blood cells, which releases cytotoxic haemoglobin and iron.³⁵ The toxicity of these blood components has been demonstrated in both cell culture experiments and in adult animal models of intracranial haemorrhage.^35,36

Figure 4: Mechanisms and treatment options for white matter injury.

Intraventricular haemorrhage (IVH) results in the collection of blood in cerebral ventricles, clot lysis and the subsequent release of haemoglobin, iron, thrombin and complement. These blood products induce oxidative stress, glutamate excitotoxicity and inflammation, which are the main pathological reactions that result in damage to oligodendrocyte progenitor cells (OPCs), reduced myelination. In addition, IVH causes mass effects that compress the brain region around the ventricle and damage the blood–brain barrier. Therapeutic strategies to improve myelination have been classified into three categories: removal of blood products; inhibition of inflammation, glutamate excitotoxicity and oxidative stress; and promotion of proliferation and maturation of OPCs. BMP, bone morphogenetic protein; COX2, cyclo-oxygenase 2; TNF, tumour necrosis factor.

Among the products of clot lysis, thrombin and iron have been widely studied and seem to be the key inducers of brain injury in neonatal and adult animal models of IVH.^37–40 Thrombin is an enzyme principally known for its role in blood coagulation. However, thrombin also has chemotactic properties that induce inflammation and mitogenic effects that stimulate cell growth. For example, thrombin stimulates both inflammation and brain injury by activating proteinase-activated receptors (PARs) 1, 2, and 4. PAR1 is expressed on OPCs and inhibition of PAR1 both reduces inflammation and promotes myelination in the spinal cord of a neonatal mouse model of IVH.⁴¹ Thrombin injection into the cerebral ventricle of neonatal rats on postnatal day 2 (P2) induces apoptotic neural cell death in the periventricular regions and dilatation of the cerebral ventricles.^36,42 The induction of ventricular enlargement by thrombin has been attributed to a loss of cadherin in the choroid plexus and increased CSF production owing to activation of PAR1 receptors.⁴²

Lysis of red blood cells releases haemoglobin into the extracellular space, which is partly scavenged by haptoglobin and endocytosed by macrophages. Haemoglobin is toxic to the brain, and treatment with haemoglobin scavengers reduces tissue injury and periventricular inflammation in preterm rabbit kits with IVH.^43,44 Haemoglobin induces cerebral oedema and injury in adult rats, via activation of haeme oxygenase 1 and by release of its degradation products.⁴⁵ Non-scavenged haemoglobin is metabolized to haeme and then to iron. Both haeme and iron trigger the Fenton reaction, which is generation of highly reactive hydroxyl radicals from hydrogen peroxide. This leads to oxidative damage of neurons and glia.^46,47 Indeed, iron chelation therapy can reduce the toxic effects of reactive oxygen species (ROS) on cultured oligodendrocytes and has offered neuroprotection in both neonatal and adult rat models of IVH.^47,48,49

Complement activation leads to formation of the membrane attack complex (C5b–9), which can damage neurons and glia. Complement components C3a and C5a are potent chemoattractants that trigger cerebral inflammation in adult models of IVH.⁵⁰ Although the effects of C3a have not been studied in neonatal models of IVH, experiments in a hypoxic-ischaemic brain injury model suggest that the effect of C3a is complex and dependent on context, timing and developmental stage.⁵¹ The presence of C3a induces neutrophil infiltration, inflammatory brain injury and expansion of infarct size.⁵² Conversely, C3a signalling promotes neuroplasticity by enhancing neurogenesis, axonal regeneration and synaptogenesis in animal models of hypoxic-ischaemic brain injury.^53–55 Lipid lysophosphatidic acid (LPA) is another bioactive molecule found in the blood that works though the G-protein-coupled receptors LPAR1–LPAR6.⁵⁶ Studies in a mouse model of IVH have shown that the presence of LPA in blood contributes to the development of post-IVH hydrocephalus via activation of LPAR1.^57,58 Moreover, LPA inhibits the survival and maturation of OPCs and might exacerbate WMI.⁵⁹

Together, a number of reports have highlighted the roles of thrombin and iron in inducing brain injury in both adult and neonatal animals with IVH. However, no studies have addressed the effects of neonatal IVH on complement activation, blood–brain barrier damage and brain–CSF barrier disruption.

Effects on OPCs

At 23–28 weeks gestational age, the human cerebral cortex is populated mostly by early and late OPCs, which are pre-myelinating cells.⁴⁷ At around 30–32 weeks, a wave of OPC differentiation is observed, which is characterized by a three-fold increase in myelinating OPCs and a marked increase in myelinated axons.^60,61 Infants born prematurely during the developmental window of OPC maturation (23–28 weeks) are at the highest risk of IVH and are vulnerable to IVH-triggered OPC injury. As both IVH and OPCs originate in the lateral and medial ganglionic eminence,⁶² IVH exposes early and late oligodendrocyte progenitors (pre-oligodendrocytes) to blood-induced injury. Studies in preterm rabbit kits have shown that IVH induces apoptosis of pre-oligodendrocytes (Fig. 5) and reduces their proliferation^26,63 In addition, preterm animals with IVH have an abundance of pre-oligodendrocytes (pre-myelinating) and paucity of immature and mature oligodendrocytes (myelinating), which suggests that maturation of OPC is arrested by IVH at the pre-oligodendrocyte stage.²⁶ A similar arrest of OPC maturation resulting in hypomyelination is observed in neonatal models of hypoxic-ischaemic brain injury and in extremely preterm human infants.^22,64 By contrast, enhanced OPC proliferation is seen in rodents with hypoxic-ischaemic brain injury.²² In conclusion, reduced myelination in IVH is attributed to increased apoptosis, reduced proliferation and arrested maturation of OPCs, which causes a dearth of myelinating oligodendrocytes and leads to myelination failure of the white matter (Fig. 5).

Figure 5: IVH induces apoptosis of oligodendrocyte progenitor cells and reduces myelination of the white matter.

a | Coronal sections of the periventricular white matter of the frontal lobe of the forebrain of a preterm (born at ED29; term 32 days) rabbit kit with intraventricular haemorrhage (IVH) on postnatal day 2 and a premature human infant (born at gestational age 27 weeks) on postnatal day 3 with IVH. Both sections were immunolabelled with O4 antibody (red) and TUNEL (terminal deoxynucleotidyl transferase dUTP nick-end labeling, which detects DNA breaks) staining (green). Apoptotic O4-positive and TUNEL-positive cells are shown (arrowheads). Scale bar 20 μm. b | Representative coronal section of the forebrain of an ED29 rabbit kit with IVH on postnatal day 14 immmunolabelled with an antibody targeting myelin basic protein versus a control kit without IVH, also on day 14. Reduced myelination is evident in the corona radiata of the kit with IVH. ED, embryonic day; V, ventricle. Scale bar 100 μm.

Oligodendrogenesis is regulated by several different families of transcription factors.⁶⁵ Basic helix-loop-helix (bHLH) transcription factors can be inhibitory (ID2 and ID4) or excitatory (OLIG1, OLIG2, and MASH1). The most important homeodomain transcription factors are Nkx2.2 and Nkx6.1, and the high mobility group transcription factors include SOX9, SOX10, and SOX17. The presence of IVH downregulates the excitatory transcription factors OLIG1, OLIG2 and SOX10 and upregulates inhibitory transcription factors, including ID2 and ID4, in preterm rabbits,²⁶ thereby inhibiting OPC differentiation and contributing to myelination failure.

Underlying mechanisms

The development of IVH stimulates a constellation of signaling cascades that trigger inflammation, generate ROS and cause glutamate excitotoxicity in the periventricular germinal matrix and adjacent white matter.^63,66,67 Oxidative stress, excitotoxicity and inflammation are the interlinked consequences of glial cell reactions to the presence of haemorrhage.⁶⁸ Activated microglial cells and invading macrophages release various pro-inflammatory cytokines, chemokines, ROS and reactive nitrogen species, which collectively induce inflammatory and oxidative injury.^63,67 For example, IVH leads to a release of glutamate into the extrasynaptic space. Owing to the reduced capacity of glial transporters to clear this glutamate, OPCs experience glutamate excitotoxicity and calcium-mediated injury.⁶⁹ Similar mechanisms of perinatal brain injury have been described in models of hypoxic-ischaemic brain injury and neuroinflammation in neonatal animals.^70,71

Studies in preterm rabbit kits and mouse pups have shown that IVH results in apoptotic cell death, infiltration of neutrophils and microglia, increased protein expression of cyclooxygenase 2 (COX2) and prostaglandin E2, and increased levels of pro-inflammatory cytokines in the periventricular germinal matrix and white matter.^30,72,38,41 Consistent with these findings, autopsy samples from human preterm infants with IVH have demonstrated a similar escalation of apoptosis and immune cell infiltration in the periventricular white matter.³⁰ Importantly, inflammation-suppressing treatment with COX2 or TNF inhibitors has restored myelination in rabbit kits with IVH, which suggests that IVH-induced inflammation inhibits myelination.

The presence of haemorrhage in the ventricle induces glutamate excitotoxicity by activation of AMPA receptors and an increased influx of calcium into OPCs, causing their degeneration and death.⁶⁶ Moreover, treatment with AMPA receptor inhibitors reduced inflammation, apoptosis of OPCs, myelination failure and neurological dysfunction in preterm rabbits with IVH, which suggests that glutamate excitotoxicity is involved in OPC injury.⁶⁶ Glutamate excitotoxicity causing OPC injury and restoration of myelination with AMPA receptor inhibition have also been demonstrated in neonatal models of hypoxia-ischaemia injury.^73,74

Low antioxidant levels in premature infants render them more vulnerable than term infants to the deleterious effects of free radicals. The reduced levels of several antioxidant molecules, including superoxide dismutase, glutathione peroxidase, glutathione, melatonin, ceruloplasmin and vitamin E, are attributed to low endogenous production as well as interrupted maternal transfer of these products owing to premature delivery.^70,75,76,77 Indeed, excessive production of superoxide, hydrogen peroxide and other free radicals damages white matter by inducing apoptotic cell death and inflammation in an animal model of IVH.⁶⁷ Free-radical generation could be mediated by a number of oxidative pathways involving NAD(P)H oxidase, cyclooxygenases, xanthine oxidase or nitric oxide synthase and mitochondria.⁶⁷ NAPH is expressed primarily in microglia and neutrophils in response to injury, and contributes to inflammatory brain injury.⁷⁸ In preterm animals with IVH, NADPH oxidase is the major source of free radicals and treatment with NAPDH inhibitors offers neuroprotection.⁶⁷ These findings are consistent with observations made in adult models of hypoxic-ischaemic brain injury and neurodegeneration.^78,79 In a postnatal day 9 (P9) mouse model of hypoxic-ischaemic injury, gene expression of Cybb (which encodes cytochrome b-245 heavy chain, also known as NADPH oxidase 2, is elevated at 24 h and 72 h after the insult, but genetic or pharmacological inhibition of NADPH oxidase 2 did not reduce brain injury in these animals.⁸⁰ The lack of any neuroprotective effect of NADPH oxidase 2 inhibition in these full-term mice can be attributed to either the timing of the experiment or to specific changes in NAPDH and glutamatergic signaling in the maturing brain of the neonatal mouse. Since activation of NADPH oxidase initiates redox signaling, inflammation, and injury to periventricular oligodendrocytes in a rabbit model of IVH,⁶⁷ we speculate that increased NADPH oxidase 2 activity contributes to OPC injury in premature infants with IVH.⁸¹ Together, inflammation, oxidative stress, and glutamate toxicity contribute to both WMI and neurological deficits in survivors of IVH.

Dysregulated OPC maturation

Key signaling pathways.

Bone morphogenetic protein (BMP), Notch, Wnt and sonic hedgehog signaling pathways are the key determinants of oligodendrogenesis during the perinatal period (Fig. 6).^26,82,83 In addition, epidermal growth factor (EGF), insulin-like growth factor 1 (IGF-I, also known as somatomedin-C) and thyroid hormone signaling in the OPC, as well as hyaluronan levels in the extracellular matrix, have important roles in OPC generation and maturation.^84–86

Figure 6: Regulation of oligodendrogenesis.

Key signalling pathways that regulate the production and maturation of oligodendrocyte progenitor cells (OPCs) are likely to be affected by intraventricular haemorrhage (IVH)-induced injury to OPCs. (1) Bone morphogenetic proteins (BMPs) activate BMP receptors 1 and 2 and phosphorylate downstream SMAD proteins (SMAD1, SMAD5, SMAD8) to induce transcription of the DNA-binding protein inhibitors ID2 and ID4, which suppresses both the specification and maturation of OPCs. (2) Sonic hedgehog protein (SHH) binds to protein patched homolog 1 (PTC1) resulting in activation of smoothened homolog (SMO). Stimulation of SMO increases the levels of zinc finger proteins GLI1 and GLI2. This upregulation of GLI1 and GLI2 and degradation of transcriptional activator GLI3 contribute to the upregulation of oligodendrocyte transcription factor 2 (OLIGO2), which promotes oligodendrogenesis. (3) Notch activation releases the notch protein intracellular domain (NICD), which then translocates to the nucleus and induces the transcription of Notch-targeted genes. Notch activation inhibits the differentiation of OPCs. (4) Wnt activation results in dissociation of the β-catenin destruction complex. The resultant rise in cytoplasmic levels of β-catenin, and its translocation to the nucleus, induces transcription of Wnt target genes. IVH downregulates this Wnt signaling. APC, adenomatous polyposis coli protein; CK1, casein kinase 1 isoform-α; CSL, a family of transcription factors; GSK3B, glycogen synthase kinase-3β; LRP, lipoprotein receptor-related protein; PKA, protein kinase A; STK36, serine/threonine-protein kinase 36, also known as fused homolog; SUFUH, supressor of fused homolog; TCF7L2, transcription factor 7-like 2.

BMPs act as negative regulators of OPC differentiation⁸⁷ and inhibit oligodendrocyte lineage commitment by diverting oligodendroglial precursors towards an astrocytic lineage.⁸⁸ BMP4 suppresses both specification and maturation of OPCs by inducing ID2 and ID4 transcription factors.⁸⁹ In both preterm rabbit kits with IVH and premature human infants with IVH, levels of BMP2 and BMP4 and levels of their downstream mediators phosphorylated SMAD1, SMAD5 and SMAD8 are elevated. Furthermore, BMP inhibition by human recombinant noggin can restore OPC differentiation and myelination in preterm rabbits with IVH.²⁶ Consistent with these studies, noggin overexpression in a transgenic mouse model of hypoxic-ischaemic brain injury reverses the maturational arrest of OPCs.⁹⁰ BMP and transforming growth factor-β (TGFβ) belong to the same superfamily. However, TGFβ signaling drives OPC differentiation and promotes myelination by modulating the expression of MYC and p21 genes.⁹¹ TGFβ1 and TGFβ2 levels are elevated in a rat model of IVH,⁹² but TGFβ inhibition by decorin and colchicine does not affect motor impairment or hydrocephalus in these animals with IVH.^92,93 Both BMP and TGFβ levels are raised in animal models of IVH, and BMP elevation inhibits myelination in preterm animals with IVH. Wnt signaling has an important role in oligodendrocyte production and maturation and affects oligodendrogenesis in a context-dependent manner. Treatment with Wnt antagonists⁹⁴ and inactivation of Wnt signaling in mice increases OPC production,^95,96 although later studies showed that Wnt signaling promoted oligodendrogenesis in both in vivo and in vitro experiments.^97,98 This suggests that Wnt signaling has distinct roles in OPC proliferation, differentiation, and myelination in a context dependent manner. In preterm rabbit kits with IVH, activation of β-catenin and expression of TCF4 and AXIN2 transcription factors are all reduced versus their counterparts in controls without IVH, suggesting that downregulation of Wnt signaling occurs in the context of IVH.⁸² Importantly, activation of Wnt signaling by glycogen synthase kinase-3β (GSK3β) inhibition, using the small molecule AR-A014418, accelerates maturation of OPCs, myelination and neurological recovery in preterm rabbits with IVH.⁸² In these experiments, GSK3β inhibition suppressed both inflammation and Notch signaling, the absence of which would contribute to myelination. During development, Notch signaling inhibits OPC differentiation and myelination both in vitro and in vivo.^99,100 In agreement with these studies, treatment with a peptide inhibitor of GSK3β reduces pro-inflammatory microglial activation, WMI and neurobehavioural deficits in a postnatal day 1 (P1) mouse model of neuroinflammation.¹⁰¹ The sonic hedgehog signaling pathway also promotes the formation of ventrally derived OPCs in the forebrain (and thus fosters myelination) by inducing production of the OLIG2 transcription factor.^102,103

Hormones and growth factors.

During the development of OPCs, thyroid hormone signaling promotes oligodendrocyte specification and maturation.^104,105 Indeed, hypothyroidism causes myelination failure, whereas thyroxine treatment promotes myelination in adult animal models of demyelination.^106,107 Intracellular availability of triiodothyronine is regulated by activating type 2 deiodinase (D2), and inactivating type 3 deiodinase (D3).^108,109 D2 converts (inactive) thyroxine to (active) triiodothyronine, whereas D3 inactivates both triiodothyronine and thyroxine. In brain samples from humans and rabbits with IVH, D2 levels are reduced and D3 levels are increased versus their levels in controls without IVH.⁸⁵ This situation is likely to result in reduced thyroid hormone signaling owing to diminished levels of T3 in neural cells. Accordingly, thyroxine treatment accelerates the proliferation and maturation of oligodendrocytes, increases transcription of OLIG2 and SOX10 genes, and consequently augments myelination in rabbit kits with IVH.⁸⁵ Therefore, thyroxine treatment restores the maturation of OPCs as well as myelination.

EGF and its receptor EGFR regulate the survival, proliferation, and migration of neural precursor cells and stimulate their differentiation towards an oligodendrocyte lineage.¹¹⁰ Studies by our group show that expression of EGF and EGFR is more abundant in the ganglionic eminence relative to the cortical plate and white matter of human premature infants and that the development of IVH reduces EGF levels, but not EGFR expression, in both rabbits and humans.⁶⁴ Importantly, treatment with recombinant human (rh)EGF promotes proliferation and maturation of OPCs, myelination of white matter, and neurological recovery in rabbits with IVH.⁶⁴ Hence, EGF has an important role in oligodendrogenesis and enhances myelination in newborn infants with IVH.

IGF-I is a neurotropic hormone that plays a crucial part in brain development and maturation. Binding of IGF-I to its receptor IGF1R recruits canonical signaling pathways, including PI3K–Akt and Ras–Raf–MAP kinase signaling pathways, which promote neuroplasticity.⁸⁶ IGF-I affects both neurons and glia. This growth factor enhances the proliferation as well as maturation of OPCs and thus promotes myelination in the neonatal rat model of hypoxic-ischaemic injury and demyelination.^33,111 To our knowledge, IGF-I has not been tested in an animal model of IVH. However, in a phase 2 randomized controlled trial, rhIGF-I complexed with its binding protein rhIGFBP3) was administered to extremely preterm infants to minimize the complications of prematurity.¹¹² This treatment reduced the occurrence of severe bronchopulmonary dysplasia and was associated with a trend towards a decline in the incidence of grade III IVH and PVHI. Long-term follow-up data on these infants are not yet available. As treatment with rhIGF-I can rescue OPCs in the immature white matter and promotes myelination and neurobehavioural function following hypoxic-ischaemic injury,^113,114 we speculate that treatment with rhIGF-I–rhIGFBP3 complex might reduce motor and cognitive deficits in premature infants, and might be particularly effective in infants with IVH.

Neurodevelopmental outcomes

The long-term developmental consequences of IVH in children aged 5–10 years include an increased incidence of cerebral palsy and cognitive impairment after either mild or severe IVH in infancy compared to children without IVH.^115–117 Premature birth or the occurrence of IVH increase the risk of psychiatric disorders in adolescence, including ADHD, poor social skills, anxiety disorder and depression. Among adolescent survivors (72.9%) from the Neonatal Brain Haemorrhage Study (NBHS) cohort of 1,105 infants, the incidence of major depressive disorder (13% versus 5.2%) and obsessive-compulsive neurosis (11.6% vesus 1.4%) was higher in those with preterm birth who developed IVH than in those preterm infants who did not develop IVH.^17,18,19 The presence of ventricular dilatation and parenchymal lesions were associated with an increased risk of developing ADHD.¹⁸ However, isolated GMH-IVH did not increase the risk of autism spectrum disorder.¹¹⁸ The mechanisms leading to this increased risk of psychiatric disorders in IVH survivors are not well understood, but might be related to both WMI and cortical injury in infancy and possibly to an associated cerebellar injury.

Neuroimaging findings

Cranial ultrasonography is routinely performed during the first week of life to detect IVH and early, severe WMI in premature infants of ≤30 weeks of gestational age.¹¹⁹ Cranial ultrasonography is repeated or MRI is performed at near-term age (35–42 weeks of postnatal age), just before the infants are discharged home. Neuro-imaging findings at near-term age predict the risk of neurodevelopmental impairment.^120,121 The advantage of cranial ultrasonography is that it is a cost-effective, bedside technique that can show the evolution of the brain lesion. Following the diagnosis of severe IVH, sequential cranial ultrasonography is performed to identify PHVD and allow timely intervention¹²²

The merits of MRI (especially diffusion-weighted and susceptibility-weighted MRI) include detection of subtle WMI lesions that would be missed by cranial ultrasonography, such as punctate lesions, diffuse white matter abnormalities and small cerebellar haemorrhages.^123,124 In addition, advanced MRI-based imaging techniques, including diffusion tensor imaging (DTI), functional MRI (fMRI) and magnetic resonance spectroscopy (MRS) can provide information on white matter development, brain growth, connectivity, myelination, metabolism and brain function.¹²⁵ DTI has been used to assess maps of white matter tracts and myelination in preterm infants with IVH. DTI reveals extensive white matter involvement with considerably lower factional anisotropy values in infants with both mild and severe IVH compared to matched controls without IVH.^23,126,127 In preterm infants with PVHI, asymmetry of the corticospinal tracts can be assessed using DTI within weeks after its onset (Fig. 2).¹²⁸ In addition, DTI tractography has also demonstrated disruptions in cerebellar white matter in infants with low-grade IVH.¹²⁶ Standard and advanced MRI techniques are important tools that can be employed for diagnosis of IVH, to guide the management of affected patients and in research studies to evaluate the efficacy of interventions.

Diagnosis and Developmental Care

Increased severity of haemorrhage and earlier gestational ages of premature infants are associated with higher rates of death and neurodevelopmental impairment. In a 2014 study of a large population of preterm babies (n = 1,762 survivors), infants with grade III IVH-PVHI have higher rates of cerebral palsy (30%) and developmental delay (17.5%) than infants with low-grade IVH or no IVH at 2 years of age¹³. In addition, infants with low-grade IVH have substantially worse neurological outcomes than those of infants without IVH in terms of both rates of cerebral palsy (10.4% versus 6.5%) and developmental delay (7.8 versus 3.4%).¹³ These data are consistent with those of other major studies and a meta-analysis evaluating the developmental outcome of infants with IVH.^129–132 However, a few other studies with smaller sample sizes have reported similar outcomes in infants with low-grade IVH and no IVH.¹³³

Very preterm infants with IVH are at an increased risk of motor, cognitive, behavioural and speech impairments in early childhood. Prompt diagnosis of these neurobehavioural impairments is crucial. The initial diagnosis of neurodevelopmental impairment is made by clinical history, physical examination and neuroimaging. A combination of tests, including the Prechtl General Movement Assessment, Hammersmith Infant Neurologic Examination (HINE) and MRI, offers the best predictive validity for detecting cerebral palsy before 5 months of age.^123,134 However, after 5 months of age, the best predictive tools to detect cerebral palsy are MRI, HINE and the developmental assessment of young children (C index).^123,125 In infants with IVH, MRI and cranial ultrasonography findings of cystic WMI, porencephaly, as well as cortical and deep gray matter lesions (representing thalamic and basal ganglia injury) are especially strongly associated with cerebral palsy.^124,134 After the diagnosis of cerebral palsy has been made, the infant should be promptly referred to Early Intervention Program for deficit-specific management and regular medical and neurological monitoring. In addition, these infants should be screened and appropriately treated for other deficits or co-morbidities, including impairment in speech, vision, and hearing, to optimize their outcomes.

Developmental care to improve cognitive and behavioral outcome initiates in the NICUs. This includes promoting skin-to-skin contact with parents (Kangaroo care), enhancing positive auditory stimulation (parents talking to their infants, music therapy), reducing noise and lighting in NICU, initiating oral stimulation (non-nutritive sucking) and supporting breast feeding. These developmental interventions in NICU have positive effects on brain structure as well as on the cognitive and behavioral outcomes.¹³⁵ Following discharge form the NICU, Early Intervention Program in the USA and European Agency for Special Needs and Inclusive Education in the Europe provide developmental care to these infants. These organizations offer multidisciplinary service to children 0–5 years of age and aims to promote child health and wellbeing, accelerate emerging competencies, reduce developmental delays, remediate existing or emerging disabilities, minimize functional deterioration and foster adoptive parenting and family function.

Early intervention could be hospital-based, home-based, or community-based. The focus of developmental intervention should be family-centered, aimed at enhancing parent–infant relationships. Interactions between infants and their parents can benefit from enabling parents to appreciate their infant’s distinctive characteristics, sensitizing parents to the infant’s cues and readiness for interaction and allaying parental anxiety and guilt related to the preterm birth. A number of studies underscore the superiority of early intervention over usual care in this setting.^136,137 Moreover, early intervention services have been effective not only improving the cognitive outcome, but also in family functioning.¹³⁸

Post-haemorrhagic ventricular dilatation

As this Review is primarily focused on WMI, we only briefly describe important advances in the management of post-haemorrhagic hydrocephalus. Increasing evidence supports the importance of early intervention for PHVD. Rather than waiting to initiate treatment after the onset of overt signs and symptoms of hydrocephalus (including bulging fontanelles, splayed sutures, sun-setting ophthalmological sign, apnoeas and a rapid increase in head circumference), intervention should be commenced in response to a rapid increase in measurements of the size of the lateral ventricles on cranial ultrasonography. Several observational retrospective studies have reinforced the notion that infants with PHVD who are managed on the basis of changes in cranial ultrasonography parameters exhibit better outcomes than do those who received treatment after the onset of clinical symptoms.¹³⁹ Intervention is started at specific cut-off values of cranial ultrasonography parameters such as the ventricular index, anterior horn width and thalamic-occipital distance; for example, one such cut-off is 4 mm above the 97th centile in a cross-sectional chart of the ventricular index.¹⁴⁰ The ELVIS (Early Versus Late Ventricular Intervention Study) was conducted to compare the clinical outcomes of infants with PHVD assigned to early intervention (that is, at a ventricular index >97th centile) versus late intervention (at a ventricular index >97th centile + 4 mm). Treatment of these infants included CSF tapping by lumbar puncture followed by reservoir placement.¹⁴¹

Few randomized controlled trials of interventions for PHVD have been conducted. The ELVIS results showed that rates of a composite outcome (consisting of death and cerebral palsy with Bayley Scales of Infant and Toddler Development cognitive and/or motor scores <70) were significantly better in the early intervention than in the late intervention group at 2 years of age, after adjustment for gestational age and severity of IVH.¹²⁴ Moreover, outcome was superior for those infants who did not develop PVHI and did not require CSF diversion surgery. Of note, patient enrollment for ELVIS took ten years to complete. Our group faced a number of difficulties in obtaining parental consent, as the randomized nature of the trial meant that the parents, who did not want to delay intervention, were reluctant to expose their child to a 50% chance of being assigned to the delayed intervention group. Also, infants with a ventricular index exceeding the cut-off for late intervention (97th centile + 4 mm) when the haemorrhage was first diagnosed could not be included in the study. The study procedure was also cumbersome because the enrolled infants required daily cranial ultrasonography to guide treatment decision-making.

The families of infants who develop PHVD experience a substantially increased burden compared to the families of their preterm peers.¹⁴² Associated neurodevelopmental morbidities and low socioeconomic status further adversely affect these families.^143,144 Hence, a family-centred intervention that aims to improve family support systems is necessary. Early intervention delivered according to cranial ultrasonographic parameters in conjunction with family-centred support could offer improved neurocognitive outcome to the survivors of IVH compared to usual care.

Neuroprotective strategies

Myelination is essential for both efficient axonal conduction of electrical impulses and for preserving axonal integrity.^145,146 Myelin sheath formation requires the organization of voltage-dependent channels at nodes and paranodes, which are required for salutatory conduction. The myelin sheath reduces the need for ATP-dependent Na/K exchange in maintaining the resting potential of the axolemma. Delayed or arrested myelination in preterm infants results in continuous conduction through widely distributed sodium channels in the axolemma, thereby increasing ATP consumption. Thus, myelination failure increases energy consumption and axonal damage. Strategies to enhance myelination can be grouped into three categories: removal of blood products; inhibition of pathological processes in the brain, including inflammation, glutamate toxicity, and oxidative stress; and promotion of the proliferation and maturation of OPCs (Fig. 7).

Removal of blood products

The DRIFT (Drainage, Irrigation and Fibrinolytic Therapy) randomized clinical trial was based on the premise that the removal of haematoma and products of blood clot lysis would reduce the incidence of PHVD and neurological impairment in infants with IVH.¹⁴⁷ This UK-based trial included 77 preterm infants with IVH and ventricular dilatation who received fibrinolytic therapy (intracerebroventricular infusions of recombinant tissue plasminogen activator) to lyse the clot, followed by irrigation of the ventricles with artificial CSF to wash out the blood clots. The results showed that DRIFT reduces the incidence of severe cognitive disability (Bayley-II Mental Development Index (MDI) score <55) in survivors at 2 years.¹⁴⁷ The difficulties in performing the DRIFT clinical trial must have been enormous, as this intervention was quite invasive for premature infants.

A 10 year follow-up study of 52 children (median age at enrollment 20 days, range 7–28 days) randomly assigned to DRIFT (n=28) or standard care (n=24) showed that the cognitive quotient score was better for DRIFT-treated children than for controls after adjustment for sex, birth weight and IVH grades.¹⁴⁸ The main limitation of the study is the small sample size, as DRIFT was stopped before enrollment was complete as 1/3 of those treated with DRIFT developed a rebleed.. In subsequent studies, a sophisticated neuroendoscopic ventricular irrigation procedure in infants with IVH led to a statistically significant reduction in rates of permanent shunt and post-IVH hydrocephalus.^149–151 In a retrospective study, 56 patients (26 weeks median gestation) underwent neuroendoscopic lavage (NEL) at median postmenstrual age of 31 weeks and followed for 34 months. Of these, 31 patients required permanent ventriculo-peritoneal shunt, suggesting a reduction of shunt placement in 43% cases.¹⁵¹ In a meta-analysis of 700 infants with PHVD performed in 2020, DRIFT was compared to conservative treatment, lumbar puncture, diuretics, as well as fibrinolytic therapy and was found to be superior to other modalities of treatment.¹⁵² Together, there is a need of double blinded randomized controlled trial using DRIFT or NEL, in which neuroendoscopic ventricle irrigation is performed at a correct postnatal age and long term outcome of the infants are assessed. Furthermore, those infants who are undergoing DRIFT can be followed for PHVD using head ultrasound and should be managed according to the recommendations made by de Vries and her co-investigators of the ELVIS trial.^124,153

A number of preclinical studies have been performed in animal models to investigate chemical removal of haemoglobin and iron, or inactivation of downstream mediators of the thrombin signaling pathway. Intracerebroventricular treatment with either haptoglobin (for scavenging haemoglobin) or α1 microglobulin (for eliminating haeme) reduces inflammation and cell injury around the cerebral ventricles of preterm rabbits with IVH.^43,44 However, the effect of these invasive therapies on WMI and neurological outcomes was not assessed, which reduces the translational potential of these studies. Treatment with iron chelators in a neonatal rat model of IVH reduced brain oedema and hydrocephalus.^48,49 Additionally, iron-chelation therapy can alleviate free-radical-induced injury to oligodendrocytes in cell culture experiments.⁴⁷ To counteract thrombin-mediated toxicity to OPCs, dabigatran (a thrombin inhibitor) or SCH79797 (a PAR1 receptor antagonist) have been tested in a neonatal rodent model of IVH.¹⁵⁴ Both treatments resulted in reduced ventricular dilatation, increased myelination and improved neurological outcomes.^41,155,156 Dabigatran is a FDA approved anticoagulant, but has a high risk of inducing hemorrhage in the brain and other organs upon treatment¹⁵⁴. Thus, the safety of using this anticoagulant in premature infants after IVH might limit its use for minimizing WMI in future clinical trials.

Inhibition of inflammatory signaling

Inhibition of inflammation, oxidative stress or AMPA-receptor-induced glutamate excitotoxicity reduces WMI in animal models of IVH.^63,66,67 The COX2 inhibitor celecoxib reduces levels of pro-inflammatory cytokines (TNF and IL1β), free-radical generation and microglial infiltration as well as enhancing myelination and neurological recovery in both rat and rabbit models of IVH.^63,157 Neuroprotective benefits of celecoxib in several other animal models of brain injury reinforce the observations made in models of IVH. Indeed, postnatal celecoxib is a promising treatment for preterm infants with IVH that might enhance their neurological outcomes. Adverse effects of celecoxib include myocardial infarction, stroke, intestinal perforations and gastrointestinal bleeding. However, the concerns regarding myocardial infarction and stroke are relevant only to the geriatric population. Minocycline is a microglia inhibitor that has been tested in a rat model of germinal matrix haemorrhage. Minocycline treatment resulted in reductions in ferritin levels, brain oedema, brain cell death and hydrocephalus in rat pups with IVH relative to their levels in controls without IVH.⁴⁸ Minocycline reduces microgliosis and levels of nitric oxide synthase, COX2 and prostaglandin, thereby exerting both anti-oxidative and anti-inflammatory effects.¹⁵⁸ Moreover, minocycline treatment was effective in improving functional independence and was safe in a clinical trial of adult patients with ischaemic and haemorrhagic stroke¹⁵⁹. However, this clinical trial was not adequately powered to assess its safety and efficacy Although minocycline has adverse effects including pigmentation of skin, teeth and bone, this agent has potential to improve the outcomes of premature infants.

AMPA (or kainate) receptor inhibition has shown neuroprotective potential in a rabbit model of IVH and other models of brain injury.⁶⁶ Two competitive AMPA receptor antagonists, CNQX (6-cyano-7-nitroquinoxaline-2, 3-dione) and NBQX (2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline), have been found to be protective in several models of brain injury. The major limitation of these agents is their renal toxicity and, therefore, they are not in clinical use. However, perampanel, an FDA-approved antiepileptic drug, had similar beneficial effects to NBQX treatment (namely, enhanced myelination and neurological recovery) in a rabbit model of IVH.⁶⁶ Perampanel is a potent and selective AMPA receptor antagonist and a broad-spectrum anticonvulsant.¹⁶⁰ This agent has a favourable adverse-effect profile and lacks psychomimetic properties as it does not interact with NMDA receptors.¹⁶¹ Importantly, perampanel has shown efficacy and safety in phase III clinical trials as a treatment for partial seizures at doses of 8 mg and 12 mg.^162,163 Hence, perampanel is another candidate neuroprotective agent that could be investigated in clinical trials aiming to improve the outcomes of IVH survivors.

Together, downregulation of these three interlinked cascades—inflammation, oxidative stress and glutamate excitotoxicity—offers a unique opportunity to reduce WMI in infants with IVH.

Supporting OPCs and myelination

In animal models, treatment with thyroxine, epidermal growth factor and stem cell treatment promote myelination and neurological recovery. Opportunities exist for translating these discoveries into clinical trials.

Thyroid hormone treatment.

Intramuscular thyroxine treatment increases the proliferation and maturation of OPCs, enhances myelination and promotes neurological recovery in premature rabbits with IVH.⁸⁵ In addition, thyroxine treatment is associated with improved myelination and neurological function in several neonatal and adult models of brain injury. Thyroxine treatment is approved by the FDA for this indication and has been shown to be safe in premature infants in multiple clinical trials.^164–166 A randomized clinical trial of thyroxine treatment in premature infants (delivered at <27 gestational weeks) showed MDI values were 18 points higher at 24 months of age in thyroxine-treated infants than in those who had received placebo.¹⁶⁵ The potential role of thyroxine treatment in minimizing WMI as well as enhancing the neurological outcomes of premature infants with IVH warrants investigation in future clinical trials.

Epidermal growth factor.

EGF treatment can restore myelination and promote clinical recovery in a rabbit model of IVH.^167,168 Intranasal EGF has also been successfully used to enhance myelination in a rodent model of hypoxic–ischaemic demyelination ¹⁶⁸ However, EGF treatment can also induce apoptosis, oxidative cell death, reduced cell maturation and malignant transformation in various cell lines.¹⁶⁹ As EGF treatment has never been tried in humans, evidence of the safety of EGF treatment in infants with IVH is lacking.

Cell-based therapies.

Stem cell therapy has shown promise in several neonatal and adult models of intracranial haemorrhage. Mesenchymal stem cells (MSCs) are the most commonly used stem cells for clinical trials because of their excellent safety profile, ease of isolation and propagation and pleiotropic properties in relieving organ injury.¹⁴⁸ Transplanted MSCs also enhance recovery owing to paracrine effects, including secretion of growth factors, such as brain-derived growth factor and stromal cell-derived factor-1, which induce neural cell proliferation and maturation.¹⁴⁸ MSCs also reduce inflammation and generation of ROS and reactive nitrogen species by affecting microglia, macrophages and lymphocytes. ¹⁷⁰ Moreover, MSCs promote angiogenesis and transfer mitochondria to damaged cells.^170,171

In a rat model of IVH, transplantation of umbilical cord blood-derived MSCs into the cerebral ventricles reduced apoptotic cell death, periventricular inflammation and post-haemorrhagic hydrocephalus. Treated animals also showed enhanced myelination and neurologic recovery.^172,173 MSC therapy has also shown success in neonatal animal models of hypoxia-ischaemia injury.¹⁷⁴ A phase I clinical trial of MSC transplantation in nine premature infants (born at 26.1 ± 0.7 weeks gestation) with grade 4 IVH (PVHI) revealed that this treatment was well-tolerated and not associated with any serious adverse effects.¹⁷⁵ With the success of the phase I trial, a phase II trial has been initiated to determine the therapeutic efficacy of MSC treatment in survivors of IVH (NCT02890953).¹⁷⁵ The main barriers to the success of stem cell therapy in this setting are selection of the right stem cell type, identification of the appropriate dose and postnatal age of the infant, and identification of a suitable mode of cell delivery.

Intranasal delivery can provide a practical and non-invasive method of intracerebral treatment. Therapeutic agents are transported from the nasal cavity to the brain by olfactory and trigeminal nerve routes, bypassing the blood–brain barrier. Indeed, nasal application of neurotrophins and MSCs has offered neuroprotection in experimental models of brain injury.^58,168 In one study, preterm infants with moderate-to-severe IVH received intranasal breast milk (which contains neurotrophins and stem cells) on postnatal days 5–28¹⁷⁶. The incidence of PHVD, shunt surgery and severe porencephalic cyst all showed trends toward a reduction in the 16 infants receiving intranasal breast milk compared to 15 controls.¹⁷⁶ The benefit of intranasal breast milk observed in this small retrospective analysis needs to be confirmed in a randomized controlled trial.

Stem cell therapy might become a promising treatment for preterm infants with moderate or severe IVH. However, in view of the potential adverse effects of this therapy, including tumour formation, graft rejection and inflammation, the risk and benefits of stem cell therapy can be assessed only after completion of randomized clinical trials.

Conclusions and future directions

GMH and IVH continue to be major complications of prematurity that result in WMI and neurological impairments in surviving infants. The past two decades have witnessed a spate of preclinical and clinical work in the field of IVH that has improved our understanding of the mechanisms underlying IVH-induced WMI, which include adverse effects on OPCs and axons triggered by blood components, particularly iron and thrombin

Current therapies for this disorder are limited to rehabilitation measures. As such, an unmet need remains to accelerate the development of new therapies that improve the outcomes of these infants. The most promising strategies investigated to date in infants with IVH include blood clot removal (DRIFT), early intervention (ELVIS) and stem cell transplantation.^124,147 However, a number of barriers have to be overcome before the results of these studies can be translated into clinical practice. The mode of delivery of a therapeutic agent is important; intraventricular injection and surgical intervention are associated with inherent risks and adverse effects. Selection of the optimal window of treatment is also critical for success; in infants with IVH, treatment must be instituted before scarring and irreversible damage sets in. The DRIFT clinical trial was conducted in infants with IVH at a median age of 20 days, which resulted in some benefits, but not as many as anticipated.¹⁴⁷ The median age at intervention in ELVIS was 9 days, which resulted in the lowest rate of ventriculoperitoneal shunts reported so far in infants with IVH and improved outcomes at 2 years of age.¹¹¹ Nevertheless, caution is warranted because very early surgical intervention might worsen IVH and could result in exacerbation of injury.

Several other mechanism-targeted strategies have been investigated in preclinical models. Among the animal studies, oral or intramuscular thyroxine or celecoxib seem the most promising treatments as these agents are already FDA-approved in adults, convenient to deliver, and have a favourable adverse-effect profile. By contrast, the use of iron chelators or anti-thrombin treatment might be poorly tolerated and associated with major adverse effects.

Owing to considerable heterogeneity in the population of patients with IVH, an individualized approach to the selection of treatment is likely to be required. For example, blood clot removal is suitable for infants with grade III IVH, but not for those with isolated PVHI. Likewise, regenerative treatments and agents that counteract inflammation and oxidative stress in the brain (including celecoxib, antioxidants, thyroxine and stem cells) are likely to be most appropriate for infants with moderate-to-severe IVH, whereas the risk of adverse effects associated with these treatments might limit their benefits in infants with grade I IVH. Moreover, tiny premature infants frequently develop other complications, including severe respiratory distress syndrome, low blood pressure, infection, patent ductus arteriosus and low platelet levels that might hinder the timely institution of treatment. Despite these challenges, the establishment of viable treatment options for infants with IVH does not seem too far in the future.

Key points.

• Intraventricular haemorrhage (IVH) results in periventricular white matter injury (WMI) in premature infants of 23–32 weeks gestation; survivors can develop neurodevelopmental sequelae including cerebral palsy, cognitive deficits and hydrocephalus.

• IVH triggers robust inflammation around the cerebral ventricles, damages axons and induces apoptosis and maturational arrest of oligodendrocyte progenitors, leading to reduced myelination of white matter.

• A constellation of blood-induced reactions, including oxidative stress, glutamate excitotoxicity, inflammation, deranged signaling pathways and alteration of the extracellular matrix, contribute to WMI in infants with IVH.

• Early diagnosis of neurobehavioural impairments, timely referral to deficit-specific early intervention and family-centred care is crucial to optimize the neurodevelopmental outcomes of these infants.

• Neuroimaging studies are important for early diagnosis of neurodevelopmental impairments, predicting outcomes and evaluating the efficacy of therapeutic interventions in both individual patients and clinical trials.

• New therapies being tested in preclinical models might reduce cerebral inflammation and promote myelination; ongoing clinical trials are investigating stem cell treatment and endoscopic removal of clots.