Gregory A. Lodygensky, Caroline Menache Starobinski, Linda S. de Vries, Petra S. Hüppi Despite marked improvements in antenatal and perinatal care, perinatal brain injury remains one of the most important medical complications in the newborn resulting in significant handicap later in life. Experimental advances have helped to understand many of the cellular and vascular mechanisms of perinatal brain damage, showing a correlation between the nature of the injury and the maturation of the brain, but effects on long-term circuit formation are just emerging.1–3 Early identification of brain injury and appropriate prognostication though remain a major challenge to neonatal care. Different forms of imaging are diagnostic tools that have emerged to detect early brain injury and help predict outcome. Magnetic resonance (MR) techniques are one of these relatively new diagnostic tools that allow the assessment of the developing brain in detail thanks to their resolving power and their noninvasiveness. Their capacity to provide detailed structural as well as metabolic and functional information without the use of ionizing radiation is unique. Conventional MR imaging is therefore now widely used for identifying normal and pathologic brain morphology giving objective information about the structure of the neonatal brain during development. Susceptibility-weighted imaging (SWI) with phase postprocessing4 is particularly useful for detecting intravascular venous deoxygenated blood as well as extravascular blood products. Diffusion-weighted imaging (DWI),5,6 spectroscopy7,8 and functional MR imaging (blood-oxygenation-dependent [BOLD] imaging)9 are newer MR techniques that complement conventional MR imaging and can indicate some of the pathophysiologic mechanisms occurring during brain injury in the newborn and the postinjury plasticity. This article will focus on the role of the different MR techniques in the study of perinatal brain injury. The specific patterns of brain injury identified by different imaging techniques will be illustrated by case presentations, followed by the discussion of pathophysiologic and neurodevelopmental outcome associated with the described brain lesion. This approach should allow the reader to make the right choice of imaging method at the right time to decide on intervention, withdrawal of care, and accurate prediction of a range of neurofunctional outcomes. Neonatal brain injury in the term infant is most frequently related to cerebral hypoperfusion and/or hypoxemia followed by reperfusion as the infant is resuscitated, typically shortly after delivery. This is summarized in the term “asphyxia,” progressive hypoxemia, and hypercapnia with significant metabolic acidosis occurring both in the antenatal, intrapartum, and neonatal period.1,10 Perinatal asphyxia may lead to hypoxic-ischemic encephalopathy (HIE) which is the clinically defined condition of disturbed neurologic function in the newborn, characterized by insufficient respiration, depression of tone and reflexes, altered level of consciousness, and often seizures.11 The subsequent neurological deficits of concern are grouped together under the term of cerebral palsy10,12,13 but include different motor deficits, such as spasticity, choreoathetosis, dyskinesia, dystonia, and ataxia. Further cognitive deficits, behavior, memory problems, and seizures might also be the end result of neonatal HIE. The major varieties of neuropathologies that are found in neonatal HIE are listed in Table 15.1. Selective neuronal necrosis is the most common pathology observed in HIE and refers to necrosis of neurons in a characteristic, although often widespread, distribution. The four basic patterns of the topography of the neuronal injury depend on the severity and temporal characteristics of the insult, and on the gestational age. Thus pontosubicular necrosis14 occurs more frequently in premature than in term infants and the basal ganglia neurons of the putamen are more likely to be affected in term infants,15 whereas neurons from the globus pallidus are more frequently affected in premature infants.16 The reason why a term infant with hypoxia-ischemia may develop one of the various patterns of selective neuronal necrosis or primarily parasagittal cerebral injury is not entirely clear but might be related to brain stem and basal ganglia sparing reflex,17 where the blood flow is redirected to basal ganglia, the brainstem, and the cerebellum in case of mild to moderate asphyxia that does not occur in case of severe abrupt hypoperfusion, i.e., in case of a sentinel event like uterine rupture or cord compression.18 Pre- or postnatal generalized systemic circulatory insufficiency can also generate focal or multifocal ischemic necrosis. The occurrence of a neonatal neurological syndrome requires the search for brain injury either from ante-, intra-, or postpartum events. The neurological symptoms of severe HIE are described in Table 15.2. Important systemic abnormalities (renal, cardiac, hepatic, etc.) related to ischemia accompany the neurological manifestations in most cases but show no relation to outcome.19 Periodic breathing or respiratory failure Intact pupillary and oculomotor response Hypotonia and minimal movements/occasionally hypertonia Persistent but diminished stupor Disturbed sucking, swallowing, gag, and tongue movements Modified from Volpe JJ. Neurology of the Newborn. 6th ed. Philadelphia: Elsevier; 2018. Clinicopathological correlations can be made for the different neuropathological varieties of neonatal HIE in the term infant and are summarized in Table 15.3. The following four cases of term perinatal asphyxia illustrate the clinical course, the neuroimaging characteristics, and the subsequent neurodevelopmental outcome and will illustrate the role and appropriate timing of MR imaging in the evaluation of the term newborn after perinatal asphyxia. Neonatal sonography is still the only bedside technique to image the neonatal brain. In term perinatal asphyxia and HIE, the most typical findings in the acute phase are represented by poor differentiation of cortical sulci and diffuse increase of parenchymal echogenicity and slit-like ventricles. These features can be primarily related to diffuse cerebral edema. Injury in the basal ganglia can lead to hyperechogenic basal ganglia, but these abnormalities take time to develop and are usually first seen 48 to 72 hours after birth. When echogenicity develops in the thalami as well as the basal ganglia, a “four column” pattern can be recognized.20 This pattern has been shown to be predictive of poor outcome21,22 and an echolucent line running between the thalami and basal ganglia corresponding to the posterior limb of the internal capsule has been shown to be predictive of poor outcome. But many early ultrasound scans in neonates with HIE lesions are normal or nonspecific, as illustrated in the cases shown. Ultrasound examinations can be considered as a first-step technique that will exclude significant hemorrhage, periventricular calcifications, overt brain malformation, or signs of well-established injury dating the insult before birth. However, accurate assessment of brain injury should be done with an MRI study. MR has become the technique of choice to evaluate the hypoxic-ischemic brain both in adults and in the newborn. Additional MR imaging techniques, such as the use of DWI, SWI, and MR spectroscopy, have further improved the MR capability to investigate the neonatal brain. Generally, to increase the signal-to-noise ratio, a higher field magnet (1.5–3 T) should be used, allowing for high-resolution imaging and increased sensitivity for spectroscopy.23–25 The basic information in conventional MR imaging is represented by T1- and T2-weighted images. Proton density and FLAIR images help illustrate brain lesions with slightly different contrast than T1- and T2-weighted images. A neonatal MR imaging protocol should provide good quality T1- and T2-weighted images with a maximum field of view of 16 to 18 cm and a slice thickness of 2 mm or less. Due to increased water content of the neonatal brain with longer T1 and T2 relaxation times, the repetition time (TR) should be increased in both T1-weighted and T2-weighted imaging sequences. Typically MR sequence parameters for T1-weighted images should include an increased TR to 800 ms and a TR above 5000 ms with an echo time (TE) of 125 to 150 ms for T2-weighted imaging.23,26 To compensate for a long TR, the echo-train-length can be increased. As shown in the three cases previously with a history of acute perinatal asphyxia, selective involvement of areas with advanced maturation and higher energy demands, i.e., the putamen, ventrolateral thalami, and perirolandic cortex, are particularly vulnerable. Characteristic changes representing selective neuronal necrosis in these areas on T1-weighted images are T1 hyperintensities, which become apparent 3 to 7 days after the insult. These T1 hyperintensities might represent cellular reaction of glial cells and macrophages containing lipid droplets and/or some mineralization of necrotic cells. Some difficulties in identifying these lesions arise from the fact that early myelination shows the same imaging characteristics. T1 hyperintensities in the internal capsule due to beginning myelination need to be differentiated from lateral thalamic lesions and lesions in the putamen. Often the posterior limb appears swollen and has lost its normal T1 hyperintensity/T2 hypointensity, which has a bad prognostic value and is associated with development of cerebral palsy.27 These T1 changes are not apparent till the end of the first week and most infants are now scanned on the day following rewarming, which is often too early to see these changes in the PLIC.26 On T2-weighted images, thalami might appear slightly hyperintense in the acute phase (see Case 1) but these signal changes tend to be very difficult to detect. T2 hyperintensities become more apparent at a later stage, illustrated also by well-defined lesions on proton-density images (see Case 2). Evolution of these lesions is marked by progressive atrophy of the involved area (i.e., putamen, thalami, rolandic cortex) with persistent T2 hyperintensity and possible cavitation.28 Of note, similar lesions in the bilateral thalami, lentiform nucleus, and globus pallidum can be detected also in premature infants with documented severe anoxic insults, most frequently associated with the typical periventricular white matter injuries.16 Isolated parasagittal injury refers to a lesion of the cerebral cortex and the subcortical white matter with a defined distribution, i.e., parasagittal, superomedial aspects of the cortical convexities, usually bilateral but often asymmetric in its extension.29,30 During the acute phase, the cortex might show increased T1-weighted signal intensity or little abnormality on conventional MR imaging (see Fig. 15.10). On the T2-weighted sequence, loss of the cortical ribbon is best seen. Chronic changes involve cortical thinning and atrophy and less often subcortical cysts. Case 5 shows another form of brain injury associated with HIE. Early on (<2 days), conventional MRI is characterized by a diffuse T1 hypointensity and T2 hyperintensity involving both the cortex and the subcortical white matter but sparing the cerebellum and the more basal structures of the medulla (see Fig. 15.9). Late intrauterine generalized prolonged systemic circulatory insufficiency is probably at the origin of these lesions, which evolve into severe cortical atrophy with cavitation and are invariably associated with a severe neurological syndrome. There is still limited data about cerebellar injury on MRI in term neonates with HIE. Conventional T1- and T2- weighted sequences do not often suggest cerebellar injury in neonates with HIE.31 SWI will additionally identify the presence of hemorrhagic lesions (Fig. 15.11). Several studies measured ADC values in cerebellar hemispheres, vermis, and pons. One study looked at 28 infants with HIE and hypothermia, 8 with HIE and normothermia, and 9 controls and found ADC values to be significantly related to death and motor outcome.32 In a recent study, ADC values were found to be significantly reduced in the vermis (p = 0.021) and dentate nucleus (p < 0.001) in infants with HIE compared to controls. ADC values in the vermis were significantly correlated with Purkinje cells injury.33 Another study looked at 59 infants and noted combined pontine and dentate nucleus in eight of them and noted that these abnormalities were always associated with a more severe brain injury pattern as well as being predictive of major disability.34 More advanced MRI studies have identified cerebellar injury in HIE using diffusion tensor imaging (DTI) in the first month after birth.35 DWI measures the self-diffusion of water. The two primary pieces of information available from DWI studies—water apparent diffusion coefficient (ADC) and diffusion anisotropy measures—change dramatically during development, reflecting underlying changes in tissue water content and cytoarchitecture.36 ADC being a quantitative measure (velocity) of overall water diffusion in tissue and anisotropy being a measure of directionality of water diffusion in a given tissue. The developing human brain presents several challenges for the application of DWI. Values for the water diffusion parameters differ markedly between neonatal brain and adult brain and vary with age. As a result, much of the knowledge regarding DWI derived from studies of mature, adult human brain is not directly applicable to the developing brain. In order to perform DWI, the optimum b value required to make the measurement has to be optimized, as it differs between the newborn and adult brain. Generally, a b value corresponding to approximately 1.1/ADC provides the greatest contrast-to-noise ratio for such a measurement.9 In neonatal brain, the high b value is typically on the order of 700 to 1000 mm2/s. EPI rather than SE37-like sequences are generally used, and the recent use of multiband acceleration techniques helps to reduce the acquisition time. Indeed “high angular resolution diffusion imaging” (HARDI) and multicompartmental diffusion techniques require the acquisition of multiple shells (data for several b values in the newborn up to 2600 ms) with multiple diffusion gradient directions to provide an accurate estimation of the diffusion model and the extraction of microstructural features of the developing brain.38 DWI parameters also change in response to brain injury. The decrease in water diffusion associated with injury was initially described for animals39 and adult human stroke,40 and was subsequently confirmed for human infants.41 Case 1 and Case 4 clearly show the marked reduction of ADC in the basal ganglia with only slight hyperintensities on T2-weighted images which can be easily missed. DWI in this case detects the lesion more reliably. There is still debate on the precise mechanism for the decrease in the ADC associated with injury. Changes in ADC following injury are dynamic. ADC values are initially decreased, but subsequently increase so that they are greater than normal and remain so in the chronic phase of injury. During the transition between decreased and increased values, there is a brief period during which values are normal, a process referred to as “pseudo-normalization.” Pseudo-normalization takes place roughly 2 days following stroke in a rat model42 and at approximately 9 days following injury in adult human stroke.43 Preliminary data indicate that the timing of pseudo-normalization in human newborns follows more closely that of adult humans than that of rodents, taking place at roughly 7 days following the injury.44 Interpretation of ADC values to detect acute brain injury in the developing brain needs to be adjusted for the regional differences in ADC values according to age (see Fig. 15.5).45 Case 3 illustrates that without numerical measurement of ADC, acute tissue alteration on diffusion maps can be missed.46 Case 2 illustrates that in the human newborns, very early (<24 hours) DWI might also miss detection of ischemic injury, which has been reported in several studies.45,47,48 From these studies, we can summarize the current role of DWI in the evaluation of the term newborn with HIE: Proton magnetic resonance spectroscopy (1H-MRS) has also entered the clinical arena of MR techniques routinely used for the evaluation of the brain and permits the noninvasive study of metabolic alterations in the brain tissue.49 The physiologist is usually interested in the intracellular concentration of a chemical species in a particular cell type. It must be noted, though, that the in vivo human MR measurement in single voxel MRS is an average (over the sensitive volume) of all tissue types. In the brain, therefore, we generally assess a combination of glial and neuronal cells with different extracellular space depending upon how much white matter, gray matter, or cerebrospinal fluid (CSF) the volume-of-interest contains. When oxidative phosphorylation is impaired, energy metabolism follows the alternative route of anaerobic glycolysis and produces lactic acid. Lactate has a chemical shift of 1.3 ppm and presents as a doublet peak in the in vivo 1H-MRS due to coupling effects. Groenendaal et al. first described markedly elevated lactate levels in five infants with severe perinatal asphyxia.50 The five patients died within the neonatal period. 1H-MRS data has been generated that demonstrates regional differences in lactate elevation after hypoxic-ischemic events in newborns. Single volume 1H-MRS in these patients showed greater increase of the Lac/NAA ratio in the basal ganglia than in the occipito-parietal cerebrum.51 This corresponds to the signal abnormalities observed with early DWI after term hypoxia-ischemia. Case 2 illustrates the typical changes in 1H-MRS after term perinatal hypoxia-ischemia. Early spectroscopy (<18 hours after event) and measurement of high Lac/creatine (Cr) ratio in 1H-MRS correlated well with neurodevelopmental outcome at 1 year.50 This acute phase lactic acidosis is followed by persistently elevated lactate levels not associated with acidosis 1 to 2 weeks after the event to several weeks after the hypoxic-ischemic event.52,53 However, 1H-MRS performed in the first 24 hours after the insult is sensitive to the presence of hypoxic-ischemic brain injury, and seems to be suitable for the detection of brain injury on the first day when conventional MR imaging and DWI might not yet detect the injury. Early MR spectroscopy, within a few hours of insult has been shown to predict outcome more accurately than very early DWI alone.26,54–56 As markers of cell integrity other metabolites visible on 1H-MRS can be used for the assessment of HIE. Ratios of NAA/Cho and NAA/Cr have been used to assess cellular metabolic integrity in neonatal brain injury.57,58 Studies using 1H-MRS at a distance (>1–2 weeks) to the hypoxic-ischemic event showed good correlation between reduced NAA ratios with adverse neurodevelopmental outcome57 whereas in early (acute stage) 1H-MRS Lac/NAA ratios are good predictors of outcome. From these studies, we can summarize the current role of 1H-MRS in the evaluation of the term newborn with HIE: Mild therapeutic hypothermia is now recognized as the only available treatment after hypoxia-ischemia. In such an acute context, thorough assessment of brain integrity is mandatory in order to manage the patient and help the family. Estimating the severity of neonatal resuscitation, the neurological status according to Sarnat classification and assessing the electroencephalographic trace are so far the only available bedside tools. Cerebral ultrasonography performed prior to starting hypothermia can be useful to identify injury of antenatal onset and abnormalities suggestive of HIE mimics. Abnormalities in central gray nuclei can be seen with ultrasound but only after 48 to 72 hours.20 MRI in this setting requiring close collaboration between the neonatologists, nursing staff, and neuroradiologist will give a thorough assessment of brain integrity and map out the extent of injury following asphyxia and might guide neuro-interventions such as hypothermia. To date, relatively few studies have addressed the question of how MRI can be used, when assessing newborns undergoing therapeutic hypothermia or after the completion of therapeutic hypothermia,26,58–61 but generally, the MRI predictability of outcome is not changed by hypothermia when scanned after hypothermia.59 The rational of performing an MRI during hypothermia is to answer whether there is massive brain injury questioning ongoing medical treatment with possible withdrawal of life-sustaining therapies and eventually whether this is significant well-established injury dating the time of insult before birth. It is possible also that it might be used in the future to monitor and tailor the length of hypothermia. The question becomes: What do we know of the validity of neonatal cerebral MRI under hypothermia? What are the effects on proton spectroscopy? What are the effects on DWI? The ADC has a biphasic evolution following HI in newborns in conditions of normothermia.44 Animal data has shown that it corresponds to cytotoxic edema with ongoing cell death and that its measure correlates with caspase-3 activation.62 The measurement of the ADC is directly influenced by temperature, with a calculated 6% reduction of the ADC at 33.5°C. Referring to normative data established by Rutherford et al.45 in conditions of normothermia, the measurement of ADC in normal un-injured tissue might give falsely decreased measures, although the hypoxic-ischemic brain regions express generally a much higher reduction with 35% or more. Moreover, the direct effect of localized cortical cooling in rhesus monkeys has been shown to induce a measurable decrease in ADC only with temperatures as low as 20°C.52 Bednarek et al.60 compared serial ADC values in hypothermia patients compared to a historic control group and showed that indeed the ADC drop after hypoxia-ischemia was prolonged by hypothermia with pseudo-normalization after 10 days rather than between 6 and 8 days as shown without hypothermia. The question remains whether the ADC in this setting is a reliable biomarker of injury. In a small prospective cohort, qualitative analysis of MRI during hypothermia has been performed on day of life 1, 2, and 10.63 Restricted ADC values could be identified in case of brain injury. Similarly, to normothermic condition, significant injury was better defined on ADC map on day 2 rather than within the first 24 hours of life. Phosphorus spectroscopy has been studied during hypothermia in an animal model with a keystone article showing that therapeutic hypothermia reduced the degree of second energy failure after hypoxic-ischemic injury.64 Temperature changes will induce minimal chemical shift of the water peak that can be used to quantify temperature when studied in relation to the temperature-independent shift of the neighboring metabolites65,66 and should not affect the magnitude of metabolic peak of lactate or NAA. Clinical experience and data published by Wintermark and colleagues63 show that spectroscopy remains a reliable tool to assess brain integrity during hypothermia. Interestingly, the repeat exam on day 2 showed greater increase in lactate than the exam done within a few hours of birth. The rational of performing an MRI after the completion of therapeutic hypothermia is to establish the extent of brain injury. The question in this setting is when is the best time to perform the exam? Is the MRI valid? Conventional T1- and T2-weighted after completion of hypothermia (8 days of life: 6–11) has been demonstrated to have the same predictive ability when compared to noncooled newborn infants.61 Unfortunately, the TOBY trial didn’t include early predetermined timing for MRI with the precise analysis of the ADC. Massaro and colleagues found that T2 hyperintensity quantification of the putamen and the thalamus when normalizing to the ocular vitreous signal intensity had superior predictive value compared to T1 or ADC analysis.67 In this study, the MRI was performed during the second week of life, explaining why the ADC had limited value, as injury can produce an ADC value within the pseudo-normalization timeframe. Further studies are needed to define lesion evolution by MRI in neonatal HIE and hypothermia treatment. Recent large retrospective analysis of predictive value of MRI for outcome prognosis confirm the value of MRI equally in cohort with or without hypothermia.59,61 Data from a metaanalysis showed a better predictive value of MRI performed during the first week than the second week of life, most likely due to improved recognition of central gray nuclei injury with DWI.56 As mentioned previously, focal and multifocal ischemic brain necrosis can also occur without HIE and the sole presence of these brain lesions also puts the neonate in the “high-risk infant” category even if they have not suffered asphyxia. However, they do significantly more often have at least one or two intrapartum risk factors compared to controls.68 These lesions occur within the distribution of single or multiple major blood vessels. Almost 90% of the infarcts are unilateral, and of these lesions nearly all involve the middle cerebral artery (MCA), 75% of them involving the distribution of the left MCA for a yet unexplained reason and most often (57%) the posterior branch of the left MCA.69 The major etiologies of these infarcts occurring without significant asphyxia are presented in Table 15.4.
Chapter 15: Magnetic resonance imaging—newer techniques and overall value in diagnosis and predicting long-term outcome
Introduction
Brain injury in the term newborn
Major Neuropathological Varieties of Neonatal HIE in the Term Infant
Characteristics of the Usual Insult According to the Pattern of Injury or Pathogenesis of the Injury
Parasagittal cerebral injury (watershed injury)
Focal and multifocal ischemic necrosis
Birth to 12 hours
12–24 hours
24–72 hours
After 72 hours
Cranial ultrasonography in the evaluation of perinatal asphyxia or HIE
MR techniques in the evaluation of perinatal asphyxia or HIE
Conventional MRI sequences and features in HIE
Selective neuronal necrosis after perinatal asphyxia
Parasagittal cerebral injury
Multicystic encephalomalacia
Cerebellar injury
Diffusion weighted imaging sequences and features in HIE
Magnetic resonance spectroscopy in HIE
MRI in the settings of therapeutic neonatal hypothermia
MRI during hypothermia
MRI performed after hypothermia
Focal and multifocal ischemic brain necrosis without asphyxia
Related posts:
Intraventricular hemorrhage in the premature infant
General supportive management of the term infant with neonatal encephalopathy following intrapartum hypoxia-ischemia
Posthemorrhagic hydrocephalus management strategies
Recent trials for hypoxic-ischemic encephalopathy: Extending hypothermia to infants not previously studied
Cerebral circulation and hypotension in the premature infant: Diagnosis and treatment
Neonatal hypotonia and neuromuscular disorders
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
Magnetic resonance imaging—newer techniques and overall value in diagnosis and predicting long-term outcome
