Imaging selective vulnerability in the developing nervous system

Abstract

Why do cells in the central nervous system respond differently to different stressors and why is this response so age-dependent? In the immature brain, there are regions of selective vulnerability that are predictable and depend on the age when the insult occurs and the severity of the insult. This damage is both region and cell population specific. Vulnerable cell populations include the subplate neurons and oligodendrocyte precursors early in development and the neurons closer to the end of human gestation. Mechanisms of injury include excitotoxicity, oxidative stress and inflammation as well as accelerated apoptosis. Advanced imaging techniques have shown us particular patterns of injury according to age at insult. These changes seen in the newborn at the time of injury on magnetic resonance imaging correlate well with the neurodevelopmental outcome. New questions about how the injury evolves and how the newborn brain adapts and repairs itself have emerged as we now know that injury in the newborn brain can evolve over days and weeks, rather than hours. The ability to follow these processes has allowed us to investigate the role of repair in attenuating the injury. Neurogenesis and angiogenesis exist in response to ischemic injury and can be enhanced by processes that are known to protect the brain. The injury response in the developing brain is a complex process that evolves over time and is amenable to repair.

Introduction

Advanced imaging techniques have allowed a window into the delineation of patterns of injury that occur in the newborn human brain. From imaging we have learned that vulnerability rests with different populations of cells critically sensitive at specific developmental timepoints. Animal studies have reinforced the findings in human newborns by showing that there are characteristically vulnerable cell populations that exhibit sensitivity to hypoxia–ischemia (HI) and excitotoxic insults (McQuillen & Ferriero, 2004). In the preterm brain, the subplate neurons and oligodendrocyte precursors are most vulnerable, and in the term brain, projection neurons especially in the deep gray nuclei are at greatest risk during ischemic insults. These vulnerabilities rest with the phenotypic characteristics of these cells at these developmental timepoints, where the expression of glutamatergic receptor subtypes that favor calcium entry and excitability, and the endogenous antioxidant mechanisms are deficient. In addition, the newborn neuron is programmed for cell death to allow for proper pruning and development but this propensity increases the vulnerability of certain cells in the early developmental window. Imaging has allowed us to capture the metabolic changes of oxidative stress and cell death, as well as to map the patterns of cell loss and degeneration. From these findings we know that the injury is just the tip of the iceberg as brain development seems to be affected by focal injuries (Miller & Ferriero, 2009). In addition, functional outcomes are affected from disturbance of these pathways. It is commonly believed that the newborn brain is plastic and resilient but followup human studies have provided mixed results (Ballantyne et al. 2008; Ment et al. 2009b). However, it has been learned from animal studies that the newborn brain can repair itself through enhanced neurogenesis (Gonzalez et al. 2007). The future will bring these therapies to the clinic and plasticity may become a reality.

Early vulnerable cell populations

Subplate neurons are among the earliest born cells of the neocortex that play a critical role in cortical development, especially in the formation of thalamocortical connections (Kostovic et al. 2002). Subplate neurons form a transient cell population that undergoes programmed cell death in the first postnatal week in mice. In humans, the subplate zone peaks at the onset of the developmental window of vulnerability to periventricular white matter injury (gestational week 24), undergoes dissolution during the third trimester and is largely absent after 6 months of postnatal age. At its peak of development in human, the subplate zone is four times the width of the cortical plate (Kostovic & Rakic, 1990).

The subplate is an area that has been shown by magnetic resonance imaging (MRI) in humans to be affected by HI in the preterm newborn human (Fig. 1), thereby making these cells regionally vulnerable. In a rodent model of perinatal HI, the subplate neurons appear to be selectively vulnerable and their loss corresponds to the severity of the insult (McQuillen et al. 2003). Recent in-vitro studies substantiate the relative sensitivity of these cells when compared with cortical neurons (Nguyen & McQuillen, 2010) to glutamate toxicity, suggesting that the glutamate receptor is playing a role in the pathogenesis. Loss of some of these cells may result in abnormal thalamocortical connectivity and explain the visual and somatosensory impairment seen in prematurely born humans who survive perinatal HI (Hoon et al. 2009).

Fig. 1.

Imaging the subplate and vulnerable white matter regions in the preterm brain. (A) Diffusion-weighted image of a 4-day-old infant born at 28 weeks gestation. The high signal in the periventricular white matter is consistent with periventricular white matter injury that is diffuse. (B) An apparent diffusivity coefficient map of this newborn shows the subplate region in red.

Oligodendrocytes that have yet to mature are also vulnerable, located in regions susceptible to perinatal HI in the preterm brain. Studies have shown that the late oligodendrocyte progenitors appear to be the most vulnerable in the lineage and fail to mature, having arrested development (Back et al. 2002; Segovia et al. 2008). This arrest in maturation results in myelin abnormalities and subsequent axonal disruption (Billiards et al. 2008). A number of explanations have been proposed for the sensitivity, including vulnerability to oxidative stress (Haynes et al. 2003) and glutamate receptor immaturity (Deng et al. 2003) but it has also been speculated that the cells cannot differentiate because of the environment that is rich in hyaluronic acid. The hypothesis has been put forth that hylauronic acid, rich in the glial scar, allows for a microenvironment that inhibits oligodendrocyte differentiation (Back et al. 2005).

Cell populations vulnerable at term

Vulnerability to specific neuronal populations in the term brain is due to multiple factors but, like white matter vulnerability in the preterm brain, excitotoxicity and oxidative stress play major roles in term injury. There is an over-expression of certain glutamate receptors in selective regions like the basal ganglia. The N-methyl-D-aspartate (NMDA) glutamate receptor subtype is the predominant mediator of this type of injury because of its coupling to neuronal nitric oxide synthase-containing neurons in the postsynaptic density complex. The NMDA receptor subunit composition changes with development with the NR2B subunit predominating early, followed by increasing expression of NR2A. NMDA receptors with NR2B have slower deactivation and higher conductance. Following HI there are differential effects on NMDA receptor subunit composition and these effects differ by age. This interaction ultimately results in the generation of both nitrogen and oxygen free radicals that in turn injure nearby cells (reviewed in McQuillen & Ferriero, 2004)). The vulnerability to neurons in the basal nuclei also appears to be related to its environment. The basal nuclei are rich in neuronal nitric oxide synthase-containing neurons that themselves are relatively resistant to severe HI (Ferriero et al. 1988). There is also an overabundance of NMDA receptors in this region at term (Black et al. 1995) allowing for the robust glutamatergic synapses necessary for long-term potentiation and connectivity but also allowing the neuron to be more vulnerable to glutamate attack. Studies have shown that eliminating these neurons either pharmacologically or by gene knockout will render the region less vulnerable and prevent neuronal loss (Ferriero et al. 1995, 1996). The non-neuronal nitric oxide synthase-containing neurons are also more sensitive to oxidative stress.

The newborn brain is rich in free iron and lacking in antioxidant defenses (such as glutathione peroxidase and superoxide dismutase), thereby providing a fertile environment for oxidative damage. There is an imbalance of antioxidant enzymes at this stage so that excessive H₂O₂ is produced but overexpression of the enzyme glutathione peroxidase can compensate for this excessive accumulation (Sheldon et al. 2004; LaFemina et al. 2006). Low molecular weight iron is also high at this period, allowing for the generation of toxic hydroxyl radicals, but this toxicity can be partially ameliorated with desferoxamine, an iron-chelating agent (Sarco et al. 2000).

In addition to oxidative stress and excitotoxicity, inflammation plays a key role in cell vulnerability. The cytokine production that accompanies infection participates in cell damage and loss. Local microglia are activated early and produce pro-inflammatory cytokines such as tumor necrosis factor-α, interleukin-1β and interleukin-6, as well as glutamate, free radicals and nitric oxide. In animal models, even systemic administration of these cytokines increases the lesions caused by excitotoxicity (Dommergues et al. 2000) and blocking microglial activation and cytokine release can protect the brain (Dommergues et al. 2003).

Apoptosis is a critical component of normal brain development but, as the brain is poised to initiate programmed cell death at these vulnerable timepoints around human birth, it becomes more susceptible to cell death pathway initiation. Although necrosis plays a major role in early neuronal death in both the immature and mature brain following injury, there is a spectrum of cell death that includes apoptosis that occurs within the first 24 h following perinatal HI (reviewed in Northington et al. 2005).

Imaging selective vulnerability in the human brain

Recently, MRI has become the ‘gold standard’ for safe and reliable diagnosis of injury in the newborn brain and has led to insights regarding normal brain maturation (Ment et al. 2009a). In the newborn, acquired brain abnormalities, such as stroke and white matter injury (WMI), are often indicated by discrete (focal) areas of magnetic resonance (MR) signal abnormality. The extent of MRI abnormalities corresponds closely to histopathological changes on postmortem examination and these changes can affect maturation of the brain (Felderhoff-Mueser et al. 1999; Miller et al. 2007).

Advanced MRI techniques, such as MR spectroscopic imaging and diffusion tensor imaging (DTI), now provide a quantifiable assessment of neonatal brain development in vivo. MR spectroscopic imaging measures regional brain biochemistry, including N-acetylaspartate (NAA) and lactate, which are useful in assessing metabolic changes associated with brain development and injury. NAA, an acetylated amino acid found in high concentrations in neurons, increases with advancing cerebral maturity (Kreis et al. 2002). Lactate is elevated with disturbances in the cerebral energy substrate delivery and oxidative metabolism (Kasischke et al. 2004). However, elevated lactate is observed in premature newborns in the absence of overt brain injury. Levels of NAA and lactate can be quantified or are expressed relative to the choline peak. Changes in these metabolites reflect changes associated with normal brain development (e.g. increasing NAA/choline) and are regionally specific. DTI characterizes the 3D spatial distribution of water diffusion in each voxel of the MR image (Mukherjee et al. 2002), providing a sensitive measure of regional brain microstructural development. With increasing maturity, the average diffusivity decreases (Partridge et al. 2004), apparently due to a decrease in water content and to developing neuronal and glial cell membranes restricting proton diffusion (Mukherjee et al. 2002). In the gray matter of the cerebral cortex, fractional anisotropy, a measure of the directionality of proton diffusion, is high early in the third trimester, reflecting the radial organization of the cerebral cortex, and becomes undetectable by term (Deipolyi et al. 2005). As this radial organization is lost, fractional anisotropy decreases, particularly with the maturation of the oligodendrocyte lineage and early events of myelination, providing a sensitive measure of white matter microstructural development (McKinstry et al. 2002; Drobyshevsky et al. 2005).

Brain injury patterns in the preterm

Acquired brain abnormalities, such as WMI or intraventricular hemorrhage, often consist of discrete areas of MR signal abnormality. With the increasing application of MRI to the clinical assessment of brain injury in the premature newborn, ‘focal non-cystic WMI’ is recognized as the most common pattern of brain injury in this population (Miller et al. 2005b). On conventional MRI, focal non-cystic WMI appears as areas of hyperintensity on T1-weighted MR images (Fig. 2). This type of WMI is distinct from cystic periventricular leukomalacia, a more severe abnormality that refers specifically to cystic regions of necrosis in the periventricular white matter that are well detected by brain ultrasound (Volpe, 2009). Focal non-cystic WMI is observed in up to half of premature newborns on MRI (Miller et al. 2003, 2005a) and this type of WMI has been under-recognized with routine clinical ultrasound. These apparently subtle white matter abnormalities on MRI and DTI are very important to recognize as they are commonly associated with abnormalities of visual, motor and cognitive function (Miller et al. 2005a; Woodward et al. 2006).

Fig. 2.

Injury to the preterm newborn brain. (A) Minimal white matter injury in a premature newborn born at 31 weeks gestational age and scanned at 6 days of life. The spoiled gradient echo volumetric scan shows small foci of T1 hyperintensity (arrow) without cavitation in the periventricular white matter. This newborn also had intraventricular hemorrhage. (B) Moderate white matter injury in a premature newborn born at 29 weeks gestational age and scanned at 2 weeks of life. The spoiled gradient echo volumetric scan shows more numerous foci of T1 hyperintensity (involving less than 5% of the hemisphere involved) without cavitation (arrow) in the posterior periventricular white matter. (C) Severe white matter injuries in a premature newborn born at 29 weeks gestational age and studied at 4 weeks of age. The spoiled gradient echo volumetric scan demonstrates confluent areas of T1 hyperintensity (arrows) without cavitation throughout the periventricular white matter of both cerebral hemispheres.

Brain injury patterns in the term

Magnetic resonance imaging illustrates the timing and heterogeneity of brain injury associated with neonatal hypoxic–ischemic encephalopathy in the term newborn. Previous retrospective studies suggested that neonatal encephalopathy is primarily related to antenatal risk factors and events that occurred well before birth (Badawi et al. 1998). However, two large prospective cohort studies of term newborns with encephalopathy evaluated with MRI demonstrate that brain injury actually occurs at or near birth (Cowan et al. 2003; Miller et al. 2005b). As the insult occurs near birth, it may be amenable to postnatal interventions, such as hypothermia, in the first days of life. Hypothermia is rapidly becoming standard therapy for full-term neonates with moderate to severe hypoxic–ischemic encephalopathy (Azzopardi et al. 2009). Recent clinical trials in neonates have demonstrated that induced moderate hypothermia reduces the combined outcome of mortality and long-term neurodevelopmental disability at 12–24 months of age (Gluckman et al. 2005; Shankaran et al. 2005; Azzopardi et al. 2009).

A remarkable regional vulnerability is observed in the brain of term newborns following HI, resulting in two major patterns of injury detectable by MRI: (i) a watershed predominant pattern involving the white matter, particularly in the vascular watershed, extending to cortical gray matter when severe and (ii) a basal ganglia predominant pattern involving the deep gray nuclei and perirolandic cortex, extending to the total cortex when severe (Fig. 3) (Barkovich et al. 1998; Sie et al. 2000). These patterns in the human newborn are associated with different antenatal risk factors and neurodevelopmental outcomes (Miller et al. 2005b; Roland et al. 1998). In fact, the pattern of brain injury on MRI is even more predictive of neurodevelopmental outcome than the severity of the lesions (Cowan et al. 2003; Miller et al. 2005b).

Fig. 3.

The predominant patterns of injury in term newborn with hypoxic–ischemic encephalopathy. (A) Watershed injury pattern: axial T2-weighted image above the body of the lateral ventricles, demonstrating the characteristic T2 hyperintensity of the cortex and white matter in the posterior watershed regions. (B) Basal ganglia injury pattern: axial T1-weighted image demonstrating marked hyperintensity of the caudate, putamen and thalamus.

Injury affects brain development

Newer quantitative MRI techniques allow the detection of acquired abnormalities of brain development in the newborn period. For example, using serial DTI, focal WMI prior to term age is followed by diffuse abnormalities of white matter development as the premature newborns develop to term-equivalent age (Chau et al. 2009). The volume and pattern of thalamocortical connections may also be disrupted in premature newborns with WMI on MRI, resulting in visual dysfunction (Kostovic et al. 2002; Counsell et al. 2007). Importantly, microstructural abnormalities in particular white matter regions correspond with subsequent developmental impairments. Furthermore, regional tissue losses, such as smaller hippocampal volumes, are linked to abnormalities in cognitive development, such as working memory deficits (Beauchamp et al. 2008). The recent ability to image the subplate zone (Maas et al. 2004) and cortical microstructure (McKinstry et al. 2002; Deipolyi et al. 2005) reveals new ways to determine the impact of focal injuries on subsequent brain development.

Several lines of evidence indicate that the risk of brain injury and abnormal brain development in the premature newborn is altered by systemic illness and by critical care therapies. For example, the severity of chronic lung disease predicts the cognitive outcome at 8 years of age, even after controlling for birthweight and neurological complications (Short et al. 2003). Postnatal infection in preterm newborns is also associated with impaired neurodevelopmental outcomes (Stoll et al. 2004). Recurrent postnatal infection is now recognized as an important risk for progressive WMI (Glass et al. 2008). Recent observations suggest that motor abnormalities seen more commonly in preterm infants with sepsis are mediated by WMI (Shah et al. 2008). However, other observations suggest that both postnatal infection and WMI are independently associated with adverse neurodevelopmental outcomes (Miller et al. 2005a). When a human newborn is exposed to therapeutic corticosteroids for the treatment of chronic lung disease, there may be impaired brain growth, although this effect may be limited to early treatment with dexamethasone (Lodygensky et al. 2005). A dramatic decline in the incidence of cystic periventricular leukomalacia is related somewhat to a decrease in days of mechanical ventilation (Hamrick et al. 2004), possibly by avoiding hypocarbic alkalosis (Fujimoto et al. 1994). In an observational study, less WMI was seen in premature newborns exposed to prolonged indomethacin therapy for patent ductus arteriosus (Miller et al. 2006). These data highlight the importance of recognizing the effect of modifiable factors on brain development in the vulnerable newborn.

Full-term infants with congenital heart disease (CHD) also have a strikingly high incidence of WMI on MRI and at autopsy (McQuillen et al. 2007; Miller et al. 2007). The WMI observed in these term newborns has imaging characteristics that are strikingly similar to those reported in preterms. The pattern of WMI is attributed to developmentally regulated cell populations vulnerable to ischemia and oxidative stress. Although predominant injury to neurons would be the expected response to these insults in term newborns with CHD, WMI nonetheless occurs frequently. Similar to premature newborns, those with CHD are certainly at risk of impaired delivery of energy substrates due to ischemia, inflammation and oxidative stress, particularly with cardiopulmonary bypass. Recent data acquired with MRI, DTI and MR spectroscopic imaging suggest that in-utero brain development is delayed in newborns with two types of CHD: D-transposition of the great arteries and single ventricle physiology, including hypoplastic left heart syndrome (Miller et al. 2007). The pattern of lower NAA/choline, higher average diffusivity and lower white matter fractional anisotropy seen in these newborns is congruous with findings in premature newborns at an earlier age as described above. Neuropathology data in newborns with CHD also indicate that they are more likely to be microcephalic and have an immature cortical mantle (Glauser et al. 1990). Newborns with D-transposition of the great arteries and single ventricle physiology have impaired in-utero brain growth, possibly related to impaired fetal cerebral oxygen delivery (Donofrio et al. 2003; Limperopoulos et al. 2010). Parallel to findings in the premature newborn, where WMI is associated with more widespread impairments in gray and white matter development, brain injury in newborns with CHD prior to surgery also impairs the subsequent development of the corticospinal tracts (Partridge et al. 2006). Together these data highlight the important connection between gray and WMIs in term newborns, and of focal injuries with widespread abnormalities in subsequent brain development.

Conclusions

Recent advances in brain imaging of the human newborn have provided fascinating insights into the age-dependent response of the brain to neurological insult. With this unprecedented view of the brain in critically ill newborns, there is increasing recognition that clinical care practices in the intensive care nursery can have beneficial or detrimental consequences on brain development and injury. A better understanding of the clinical factors that impact brain development and injury will allow us to directly improve the neurodevelopmental outcome of newborns at highest risk of neurodevelopmental impairments, such as those born prematurely, full-term newborns with hypoxic–ischemic encephalopathy and those with congenital heart birth defects. In particular, a clearer view of the connectivity of selectively vulnerable cell populations and circuits in the developing brain will enable the consideration of therapeutic approaches that extend past the first hours of an acute insult. Moreover, the ability to identify and quantify brain injury and abnormal development at a time when intervention is possible will lay the foundation for testing new strategies for preventing or treating brain injury in these populations. As more research is done on human brain development, as outlined in this issue, these strategies can be realized.