WMI, the most common form of brain injury in preterm infants, occurs most often among infants born between 23 and 32 weeks’ gestation with a peak incidence at 28 weeks’ gestational age. The prevalence of WMI varies according to the gestational age of the cohort, as well as the timing and modality of imaging. A recent systematic review has described that up to 40% of preterm infants are born before 28 weeks’ gestational age.¹ The spectrum of preterm WMI ranges from discrete lesions to diffuse volume loss and includes three pathologic forms: cystic necrosis, microscopic necrosis, and nonnecrotic lesions. The most severe injury consists of destructive lesions of all cellular elements with areas of cystic necrosis greater than 2 mm in diameter, known as cystic PVL. The incidence of these lesions has decreased substantially in the last few decades and occurs in current cohorts of very low birth weight infants at a rate of less than 5%.^2,3 More commonly, microscopic foci less than 1 mm in diameter occur and evolve into areas of gliosis, commonly referred to as noncystic PVL. Interestingly, the incidence of noncystic WMI is decreasing over time in some cohorts of preterm neonates, without a clear explanation.⁴ In contemporary cohorts of preterm newborns, the most common form of injury is diffuse WMI. Unlike necrotic injury, diffuse WMI is caused by selective degeneration of premyelinating oligodendrocytes (pre-OLs) that fail to mature into oligodendrocytes, leading to myelination failure and secondary axonal disturbances.⁵

Notably, the patterns of WMI associated with very preterm infants, born before 32 weeks, are increasingly being recognized in other populations. Recent studies have demonstrated WMI in a substantial number of moderate to late preterm infants, though severe lesions are less common than at lower gestational ages.^6–8 In a prospective study of neonates born between 32 and 36 weeks, WMI was the most frequent form of brain injury encountered, with MRI at TEA revealing diffuse injury in 23% and punctate lesions in 16%.⁸ Term-born infants with severe congenital heart disease demonstrate similar lesions,^9,10 postulated to be a result of destructive and dysmaturational disturbances that overlap with the mechanisms in preterm infants.¹¹ Punctate lesions have also been reported among near-term and term-born infants associated with perinatal asphyxia and genetic diagnoses.¹² Nevertheless, as the hallmark brain lesion among very preterm infants, the imaging characteristics, pathogenesis, and outcomes of WMI are currently best understood in the context of prematurity as illustrated below.

The evolution of preterm WMI from the first few days of life until TEA has been extensively characterized by serial cranial US imaging.¹³ This method has been used to diagnose cystic WMI since the 1980s due to its availability, safety, and reliability. Typically, the initial presentation of WMI on US is enhanced periventricular echogenicity. This may persist or resolve within several days; after resolution, it may then evolve into gliotic or cystic changes or, alternatively, leave no abnormality.¹³ Cysts may only appear after 2 to 6 weeks and may subsequently disappear due to resorption of fluid within gliotic brain tissue and can therefore be missed without sequential imaging. At TEA, US may only demonstrate ex vacuo ventriculomegaly and enlarged extra-axial spaces. Thus the number and timing of US studies are critical to consider when evaluating the diagnostic performance of this imaging modality.

When cystic WMI persists, it is associated with significant developmental abnormalities that are primarily motor impairments, classically, spastic diplegic CP. A widely utilized grading classification for the severity of cystic PVL was developed by de Vries et al.¹⁴ and increasing severity of PVL according to this classification is associated with increased incidence and severity of CP.¹⁵ Prolonged duration of periventricular hyperechogenicities, even in the absence of cystic evolution, has also been associated with adverse motor outcomes.^16,17

MRI is the preferred modality to diagnose diffuse WMI. At TEA and onward, diffuse WMI may appear on MRI as ventriculomegaly, irregularly shaped ventricles, white matter loss (enlarged cerebrospinal fluid spaces surrounding the sulci), white matter tract thinning (e.g., thin corpus callosum), diffuse signal intensity changes, and myelination disturbances.¹⁸ Scoring systems for these white matter abnormalities have been developed that are inversely associated with neurodevelopmental outcomes at preschool age.^19–21

Many centers also perform early MRIs, at approximately 32 weeks postmenstrual age, or as soon as infants are sufficiently stable to undergo the exam. Increased use of early MRI has demonstrated more subtle focal lesions that may not be appreciated on US imaging or follow-up TEA imaging.²² Visualized soon after the insult, WMI lesions may appear as clusters of punctate hyperintense T1 signal abnormalities often accompanied by diffusion restriction.^18,23,24 Linear punctate lesions with a hemorrhagic component, indicated by signal loss on susceptibility-weighted imaging, may also be seen.^23,25 Clinical factors associated with punctate lesions include higher gestational age and birth weight among very preterm neonates and the presence of intraventricular hemorrhage (IVH).^26,27 These multifocal WMI lesions are most readily visualized on early-life MRI. In a significant number of infants, these early lesions fade or become challenging to detect on MRI at TEA, though they are still associated with neurodevelopmental outcomes, even if harder to detect on these later age MRI scans. The change in imaging characteristics of these lesions over time supports the destructive/necrotic nature of these early lesions and not that they are transient imaging phenomena.^22,28 Importantly for clinicians, both the volume and location of lesions on early-life MRI are predictive of outcomes, with anterior lesions being most concerning for motor and cognitive deficits.^29,30 A summary of US and MRI findings can be found in Table 4.1.

More sophisticated imaging modalities reveal microstructural disturbances and aberrant connectivity patterns among preterm infants with WMI. Diffusion tensor imaging (DTI) shows reduced fractional anisotropy in white matter tracts, indicating delayed maturation among preterm neonates with focal lesions,^31,32 including distal to the original site of injury³³ and even in the absence of overt injury on conventional MRI.³⁴ Diffuse white matter abnormalities are also associated with abnormal radial and axial diffusivities on DTI.^35,36 Functional resting state MRI has also demonstrated altered connectivity among neonates with WMI.^37–39

WMI does not typically have a clinical correlate in the immediate neonatal period and the diagnosis is usually made by routine imaging. Guidelines from the American Academy of Pediatrics and Canadian Paediatric Society recommend performing serial cranial US in very preterm infants in the first week of life, at 4 to 6 weeks, and again prior to discharge, with minor differences between the guidelines regarding the gestational age cutoff and timing of the first US.^40,41 The first US is primarily meant for the detection of IVH and the latter for WMI, although, as discussed earlier, US may miss the more prevalent diffuse WMI, better demonstrated on MRI. Some controversy exists surrounding the routine use of MRI as a screening tool at TEA given its limited overall predictive value for long-term outcomes,^42–44 and in current guidelines, MRI is suggested for infants with abnormal imaging findings on US. Other experts offer evidence-based indications for using routine MRI in high-risk infants in conversation with the family concerning the performance of this imaging modality to predict long-term neurodevelopment.⁴⁵ It is important to note that the absence of WMI carries a particularly high negative predictive value for significant adverse outcomes,⁴⁶ allowing clinicians to be reassuring when WMI is not demonstrated. And, as noted above, multifocal WMI lesions are more readily detected on early-life MRI rather than at TEA.

The central mechanisms of WMI in preterm neonates are hypoxia-ischemia and infection-inflammation, leading to oxidative stress, glutamate-mediated excitotoxicity, and the production of proinflammatory cytokines and reactive oxygen species (ROS) that are toxic to the developing white matter. The primary cellular target of injury is oligodendrocyte precursors, called premyelinating oligodendrocytes (pre-OLs). Extensive human and experimental studies have demonstrated the vulnerability of pre-OLs, the predominant white matter cells in the preterm brain during the peak window of WMI.⁴⁷ Oligodendrocytes, primarily derived from radial glial cells, mature in four principal stages that have distinct morphology, immune markers, and functional characteristics.⁴⁸ The earliest form of oligodendrocyte progenitors can migrate and proliferate, whereas pre-OLs are proliferative but nonmigratory, and the more mature oligodendrocytes have myelinating properties. Pre-OLs are uniquely vulnerable to oxidative stress due to maturation-dependent features, including an immature antioxidant system that renders them vulnerable to ROS.^49,50 This intrinsic susceptibility is not seen in earlier and later oligodendrocyte lineage stages, or in cortical neurons,⁵¹ and it is potentiated by both hypoxia-ischemia and inflammation, which in turn, potentiate each other.⁵² The role of hypoxia-ischemia is postulated to be a result of disturbances in cerebral autoregulation, heart-rate-dependent cardiac output, and the immature vasculature precipitated by episodes of hypoxia, bradycardia, and hypotension.⁵³ Infection-inflammation is mediated by microglial activation via Toll-like receptors (TLRs), as a result of common inflammatory disturbances in the preterm neonate, including sepsis and necrotizing enterocolitis (NEC).⁵⁴ These mechanisms trigger a cascade of events, leading to the degeneration of pre-OLs and a subsequent increase in dysmature oligodendrocyte progenitors that are unable to fully differentiate into myelin-producing cells, causing diffuse injury characterized by impaired myelination and volume loss.⁵⁵

The final common pathway for these mechanisms of injury is the degeneration of pre-OLs and subsequent myelination failure. In severe focal lesions, myelination failure results from pan-cellular necrosis causing degeneration of glial cells and axons within the foci of injury, as initially described by Banker and Larroche,¹²⁹ as well as axonal injury in surrounding gliotic areas beyond the regions of necrosis.¹³⁰ The immature unmyelinated axons demonstrate a vulnerability to ischemia that is similar to pre-OLs¹³¹ and the severity of axonopathy seems to be related to the severity of WMI necrosis.¹³²

In the more prevalent chronic form of diffuse nonnecrotic WMI, myelination failure is not mediated by severe axonal injury, but rather by dysmaturation. This milder form of injury is characterized by the acute degeneration of pre-OLs that is accompanied by inflammatory astrocytosis and microgliosis. Interestingly, the loss of pre-OLs also stimulates a reactive proliferation of oligodendrocyte progenitor cells, resulting in a paradoxical increase in the pre-OL population.¹³³ Human and animal studies have shown a marked increase in progenitor cells primarily within WMI lesions, with a small increase also in the subventricular zone, the transient fetal structure where they are naturally generated.^133–135 These progenitors give rise to a large pool of regenerated pre-OLs in the injured areas; however, these pre-OLs are unable to differentiate into myelinating cells.⁵ The mechanisms behind their disrupted maturation are still being elucidated but seem to be related to the surrounding astrogliosis and microgliosis in the damaged extracellular matrix. Reactive astrocytes in the extracellular matrix secrete hyaluronic acid and the products of hyaluronic acid degradation by specific hyaluronidases inhibit the maturation of pre-OLs.^136–138 Increased levels of hyaluronic acid have been found in preterm sheep models of WMI and reactive astrocytes with the hyaluronic acid receptor CD44 have been found in human preterm WMI.^55,56 The accumulation of selective bioactive hyaluronic acid fragments resulting from this degradation induce signal transduction via TLRs to attenuate the transcription of myelin basic protein, thereby impeding myelination. Other proposed mechanisms for myelination failure include interactions between CD44 and hyaluronic acid that involve recognizing and producing these fragments, independent of signaling.¹³⁹

Data from studies on adult demyelinating diseases have revealed additional potential mechanisms for dysmaturation involving signaling pathways such as Notch and Wnt-beta catenin that disrupt oligodendrocyte maturation,¹⁴⁰ as well as epidermal growth factor receptor (EGFR) signaling that promotes maturation.¹⁴¹ The Wnt signaling pathway was subsequently found to be involved in neonatal WMI mediated by hypoxia-inducible factor alpha (HIFα), critical for angiogenesis and myelination.^142,143 More recently, HIFα was also found to regulate myelination, independent of this signaling pathway.¹⁴⁴ Elucidating the exact mechanisms is crucial for developing targeted treatments. In contrast to irreversible necrotic injury with resultant death of all cellular components, dysmaturation seems to be a form of regeneration and plasticity that may be potentially modifiable and reversible. Novel treatments that have been investigated include hyaluronidase blockers that have been shown to promote pre-OL maturation¹⁴⁵ and a selective EGFR ligand administered intranasally to mice with diffuse WMI that promotes recovery and maturation of oligodendrocytes.¹⁴⁶

Numerous neonatal complications, including comorbid conditions and NICU procedures and therapies, are associated with neonatal WMI. Hemorrhagic infarcts of deep white matter that accompany IVH and evolve into porencephalic cysts are discussed in the chapter on IVH. As noted above, the prevalence of WMI is higher in neonates with IVH for a given gestational age at birth. White matter dysmaturation can also occur following IVH, via the release of blood products such as iron and thrombin that induce oxidative stress, inflammation, and glutamate excitotoxicity.^147,148

Neonatal infections are highly associated with WMI in multiple studies.^98,149–152 This association has been observed in culture-positive infections from the cerebrospinal fluid,¹⁵² as well as from sites remote from the brain (blood and tracheal cultures),¹⁵¹ and in culture-negative clinical sepsis.¹⁵³ Regarding the type of WMI, cystic PVL has been linked with neonatal sepsis,¹⁵⁴ while noncystic progressive WMI has been linked to recurrent infections in the neonatal period.¹⁴⁹ Multiple infections (≥3) have also been associated with indices of dysmaturation in white matter tracts on DTI in the absence of overt injury on conventional MRI.⁹⁸ The pathogenesis is related to proinflammatory cytokines and activated microglia via expression of TLRs as discussed above.^54,155

NEC has also been associated with an increased risk of WMI, particularly when it requires surgical management.^156–158 The mechanism of NEC-related brain injury, long suspected to be mediated by the innate immune response and TLRs,^54,97 has only recently begun to be elucidated in mouse models, shedding new light on the gut-brain signaling axis. Nino et al.¹⁵⁹ discovered the release of proteins functioning as TLR ligands in the intestine that promote microglial activation and accumulation of ROS in the brain. This group of researchers subsequently demonstrated a key role for intestinal CD4+ T-lymphocytes in inducing brain injury via release of interferon-gamma and microglial activation.¹⁶⁰ Importantly, inhibition of these lymphocytes and neutralization of interferon-gamma with targeted antibodies were able to prevent WMI, revealing a potential therapeutic avenue to avert brain injury in neonates with NEC.

Severe ROP is associated with altered microstructural integrity and delayed maturation of white matter tracts, predominantly in the posterior white matter.^161–163 These changes are predictive of lower motor and cognitive scores at 18 months old.¹⁶² The pathogenesis is postulated to be linked to levels of insulin growth factor one,¹⁶⁴ an anabolic hormone which exerts effects on the retina and the brain, but the exact mechanisms have yet to be clarified. Other hypotheses include a role of systemic inflammation and hyperoxia, as recently reviewed by Morken et al.¹⁶⁵

BPD or prolonged mechanical ventilation is associated with a spectrum of WMI and delayed maturation of white matter at TEA.^166–170 Several underlying factors seem to be related to adverse white matter development in BPD including duration of mechanical ventilation,^171,172 and precursors to BPD, such as cumulative supplemental oxygen and mean airway pressure in the first few weeks of life, which are independently associated with adverse neurodevelopmental outcomes.¹⁷³ Respiratory disturbances that are often associated with mechanical ventilation such as hypocarbia and hyperoxia are also linked to WMI.^174–176

There is conflicting evidence to support a role for chorioamnionitis as a risk factor for WMI. Intrauterine inflammation leading to the release of proinflammatory cytokines and fetal immune response syndrome is related to neonatal morbidities, such as postnatal infection, IVH, and adverse outcomes, that are also associated with WMI,^177–179 but several studies have been unable to show an independent association between chorioamnionitis and WMI.^{171,180–182}

Given the function of myelinated white matter in the rapid transmittal of information across neural networks, it is not surprising that WMI can have a significant and broad impact on motor, cognitive, and social development. Motor-wise, white matter enables efficient and accurate communication between brain regions that include the motor cortex, basal ganglia, and cerebellum; intact myelination between these areas is vital for motor skills requiring fluency, coordination, and automatization. Accordingly, WMI is associated with a spectrum of motor deficits of varying severity, ranging from developmental coordination disorder (DCD) to CP, with more severe imaging abnormalities predicting more severe impairment.^{18,46,183,184}

Although the incidence of the severe lesion of cystic WMI has decreased, when it occurs, it remains highly predictive of CP. In a recent population-based study over two decades in Canada, approximately 80% of infants with this injury were diagnosed with CP,³ consistent with other studies.^185–187 In the absence of cystic injury, myelination disturbances in the corticospinal tract are also associated with CP.^188,189 Clinically, motor abnormalities predictive of CP may be detected early in infancy; in the first few weeks of life, abnormal general movements, including the absence of normal fidgety movements, are highly sensitive and specific for the subsequent diagnosis of CP.¹⁹⁰ In preterm infants, these aberrant patterns of general movements are associated with WMI on TEA-MRI, white matter microstructural abnormalities, lower regional white matter volumes, and adverse motor and cognitive outcomes.^191,192 Diffuse white matter microstructural abnormalities, in the absence of gross abnormalities on conventional MRI, are also associated with childhood DCD. Clinically, DCD manifests as difficulty learning and executing coordinated motor movements and reduced fluency in motor skills and the neuroanatomical correlates of these deficits last into childhood.^193,194

Several studies have demonstrated increased rates of long-term cognitive impairments at school age among preterm-born children with overt WMI and, as with motor outcomes, the severity of injury is related to the severity of impairment.^21,195–199 Microstructural disturbances on DTI, even in the absence of lesions on conventional imaging, are also associated with cognitive outcomes.²⁰⁰ Specifically, impairments in complex skills such as executive functions, responsible for cognitive flexibility, self-control, and working memory are associated with white matter microstructural abnormalities among children born preterm who suffered inflammatory neonatal complications, such as BPD, NEC, and sepsis.²⁰¹ The link between executive dysfunction and inflammatory complications is bolstered by the association between elevated neonatal inflammatory markers in the first month of life and executive function deficits at age 10 years.²⁰² These impairments may explain the increased rates of attention deficit hyperactivity disorder seen in the preterm population.²⁰³

The fundamental role of white matter in these higher-order cognitive skills has been gaining recognition^204,205 and is likely related to connectivity and network efficiency.^206–208 The development of the connectome between 30 and 40 gestational weeks is heavily mediated by white matter microstructure²⁰⁹ and microstructural disturbances among preterm-born children result in impaired connectivity and network efficiency.²¹⁰ These disturbances may also underlie the higher prevalence of autism spectrum disorder (ASD) and anxiety disorders that are seen among children born preterm.^211,212 Long-distance connectivity along several white matter tracts, such as the superior longitudinal fasciculus, is integral to the normal development of social cognition and theory of mind.²⁰⁷ Disruptions to the integrity of these tracts are commonly found in children with ASD^213,214 and aberrant fractional anisotropy values in white matter tracts at 6 months of age can be found among infants who are subsequently diagnosed with ASD.²¹⁵ Among preterm-born children, frank WMI on conventional MRI at TEA has been associated with ASD²¹⁶ and animal models of preterm diffuse WMI provide evidence of inflammatory mechanisms mediating autistic behaviors in mice, mimicking human ASD.²¹⁷

Cognitive outcomes are also likely mediated by grey matter changes that may accompany WMI. MRI studies have shown reduced cortical and thalamic volumes among children with WMI^218–220 and reduced caudate neuronal maturation observed experimentally with hypoxia-ischemia.²²¹ Furthermore, the interaction between early neonatal WMI and diminished childhood thalamic volumes has been associated with cognitive outcomes at school age.²²⁰ Among neonates with PVL, pathology studies have demonstrated gliosis and neuronal loss in the thalamus²²² and DTI studies have demonstrated abnormal microstructure both within the thalamus and in white matter tracts that is associated with diminished thalamic volumes.^219,223,224 Animal models of diffuse WMI have attributed grey matter loss to a reduction in the complexity of the dendritic armor and reduced synaptic activity in cortical neurons and corticothalamic pathways.^{221,225–227}

Cognitive outcomes are also modified by environmental and genetic influences that are not readily accessible on imaging. The substantial impact of socioeconomic status on outcomes²²⁸ suggests potential interventions to improve outcomes, as well as important limitations for predicting outcomes with imaging alone.

There is currently no effective treatment to prevent or to cure WMI and therefore mitigate its long-term consequences. In the neonatal period, avoiding associated risk factors for WMI, such as ventilatory complications, neonatal infection, and NEC, is key. When WMI is diagnosed on imaging, surveillance programs for early intervention are put in place to identify any neurodevelopmental impairments and start physical, occupational, and speech and language therapies. These programs have demonstrated a positive influence on outcomes at preschool age.²²⁹

The more prevalent chronic diffuse WMI that is seen in contemporary preterm populations opens a window for possible therapeutic strategies to mitigate or prevent the maturation and myelination failure that progressively occurs over many weeks. Potential interventions include treatments targeting inflammation and the immune response. Erythropoietin has been investigated over the last decade as a neuroprotective agent due to its antioxidant and anti-inflammatory effects and as a promotor of neurogenesis. In animal models of neonatal WMI, it prevents pre-OL death and promotes pre-OL development.^230,231 Randomized trials subsequently showed improved white matter development and better cognitive outcomes in preterm infants who received erythropoietin.^232–234 A recent randomized, double-blind trial of prophylactic erythropoietin treatment did not show differences in outcome between the placebo and erythropoietin groups.²³⁵ This finding is surprising given the mechanisms of WMI outlined above; as such, erythropoietin as a potential neuroprotection strategy requires further attention (e.g., longer duration of therapy). Other potentially promising interventions in preclinical trials include anti-inflammatory treatments, such as tumor necrosis factor antagonists, associated with reduced gliosis among preterm sheep exposed to inflammation,²³⁶ and anti-interleukin-1 monoclonal antibodies, that reduce brain injury in fetal sheep exposed to ischemia.²³⁷

Supportive measures such as providing adequate nutrition and nutrient intake have also been shown to improve outcomes and diminish brain injury in preclinical and some clinical studies.²³⁸ Specifically, studies on full-term infants suggest that iron may improve myelination; a recent randomized controlled trial showed greater myelin content at 1 year of age among term-born infants who underwent delayed cord clamping associated with higher ferritin levels at 4 months of age.²³⁹ The long-term outcomes of adequate nutritional intake are still being evaluated.²³⁸ Preventing painful procedures is also an important component of improving brain maturation and neurodevelopmental outcomes.²⁴⁰ Painful procedures are associated with impaired white matter development that persists to school age and is linked with cognitive outcomes.^241–243 Optimal strategies to measure pain and provide analgesia in neonates require further investigation.