Abnormal Magnetic Resonance Signal in the Internal Capsule Predicts Poor Neurodevelopmental Outcome in Infants With Hypoxic-Ischemic Encephalopathy
Objective. The aim of this study was to establish whether abnormal signal intensity in the posterior limb of the internal capsule (PLIC) on magnetic resonance imaging is an accurate predictor of neurodevelopmental outcome at 1 year of age in infants with hypoxic-ischemic encephalopathy (HIE).
Methods. We have examined 73 term neonates with HIE between 1 and 17 days after birth with cranial magnetic resonance imaging and related the magnetic resonance imaging findings to neurodevelopmental outcome at 1 year of age.
Results. All infants with an abnormal signal intensity in the PLIC developed neurodevelopmental impairment although in 4 infants with very early scans the abnormal signal was not apparent until up to 4 days after birth. A normal signal intensity was associated with a normal outcome in all but 4 cases; 3 of these infants had minor impairments and all had persistent imaging changes within the white matter. The 4th infant with a normal signal intensity on day 2 died before a further image could be obtained. The absence of normal signal predicted abnormal outcome in term infants with HIE with a sensitivity of 0.90, a specificity of 1.0, a positive predictive value of 1.0, and a negative predictive value of 0.87. The test correctly predicted outcome in 93% of infants with grade II HIE, according to the Sarnat system. Applying a Bayesian approach, the predictive probability of the test (the probability that the test would predict an outcome correctly) was distributed with a mean of 0.94 and 95% confidence limits of 0.89 to 1.0.
Conclusion. Abnormal signal intensity in the PLIC is an accurate predictor of neurodevelopmental outcome in term infants suffering HIE.
- MRI =
- magnetic resonance imaging •
- HIE =
- hypoxic-ischemic encephalopathy •
- PLIC =
- posterior limb of the internal capsule •
- T1WSE =
- T1-weighted spin echo •
- T2WSE =
- T2-weighted spin echo •
- IR =
- inversion recovery •
- DQ =
- developmental quotient
Magnetic resonance imaging (MRI) in infants with hypoxic-ischemic encephalopathy (HIE) is able to detect early patterns of injury that are associated with permanent structural change to the brain1-7 and abnormal neurodevelopmental outcome.2 Documented patterns include absence of the normal signal intensity in the posterior limb of the internal capsule (PLIC), bilateral abnormalities within the basal ganglia and thalami, loss of gray/white matter differentiation in the hemispheres, and highlighting of the cerebral cortex. Although absence of the normal signal intensity from myelin within the PLIC may sometimes be seen on the first day after birth in infants with HIE, other abnormalities are often most easily detected at the end of the first week after delivery.1 A normal term infant will have high signal in the posterior half of the posterior limb on an inversion recovery (IR) sequence image (Fig 1A). The aim of this prospective study was to establish whether absence of the normal signal intensity in the PLIC during the first 2 weeks after suspected birth asphyxia is an accurate predictor of neurodevelopmental impairment at 1 year of age.
Ethical permission for these studies was obtained from the Royal Postgraduate Medical School Research Ethics Committee.
The infants were either born at the Hammersmith or Queen Charlottes Hospitals or were referred from other hospitals. All infants included in the study were born at term (37 to 42 weeks, as assessed by early ultrasound and/or clinical gestational age assessment), were suspected of having birth asphyxia, and were diagnosed as having HIE in the first few days of life. The diagnosis of HIE was made in the presence of all of the following signs: fetal distress; Apgar score of <5 at 1 minute; a need for resuscitation; and one or more of the following neurologic abnormalities in the first 2 hours of life: abnormal tone, poor feeding, convulsions, or decreased conscious level. Fetal distress was defined as an abnormal cardiotocograph with type II dips, decreased heart rate variability, or bradycardia (<100 beats/minute) with or without meconium-stained liquor. HIE was staged according to Sarnat and Sarnat:8 infants with stage I showing transient abnormalities of tone; infants with stage II having convulsions, more persistent abnormalities of tone, and feeding difficulties; and infants with stage III being unresponsive, with or without convulsions. For the purposes of this study the infants were staged according to their clinical course during the first week of life. Infants with congenital abnormalities, congenital infection, or metabolic disorders were excluded from this study.
Magnetic Resonance Imaging
Infants were studied as soon after birth as practicable. They were usually examined during natural sleep although chloral hydrate (50 mg/kg) was given for sedation if necessary. The infants were monitored with an electrocardiogram and pulse oximetry and a pediatrician was present throughout the examination.
The infants were examined with a 1 Tesla Picker HPQ system (Picker International, Cleveland, OH). Images were obtained in the transverse plane with T1- and T2-weighted spin echo (T1WSE and T2WSE) and age-related IR sequences. The sequence parameters are given in Table1.
Images were assessed by two experienced observers (J.M.P., S.J.C.). Each observer assessed the images separately and neither observer was aware of the clinical outcome in the infants. The signal intensity within the posterior limb was assessed on T1WSE, T2WSE, and IR images and classified as normal, equivocal, or abnormal for each infant. The normal appearance of the PLIC is high signal on T1WSE and IR in at least one third of the limb (Fig 1A) and a much smaller area of low signal intensity within the high on T2WSE (Fig 1B). Abnormal appearances were defined as ranging from partial or complete absence of the normal high signal within the PLIC on T1WSE and IR sequence (Fig2A) and loss of the normal low signal intensity on T2WSE. The term equivocal was used when there was only an asymmetry of the signal intensity within the PLIC. The images were also assessed for abnormal signal intensities elsewhere in the brain.
All infants were followed up with repeat imaging and clinical, neurologic, and developmental assessments. Neurodevelopmental outcome was assessed at 1 year of age, using Griffiths developmental scales9 and a detailed neurologic examination. Neurologic signs were quantified using an optimality score,2 5 based on the neurologic examination. The score includes items for tone, posture, passive and elicited motility and tone, interaction, reflexes, and vision and hearing responses. The maximum possible optimality score was 21. (The score sheet is available on request from the corresponding author.)
Neurodevelopmental outcome was categorized as normal or abnormal. Infants were classified as normal if they had no abnormal neurologic signs and achieved a Griffith's developmental quotient (DQ) of ≥85. All other infants were regarded as having an abnormal neurodevelopmental outcome and this was classified as mild, moderate, or severe. Infants with minor neurologic signs, such as hypotonia or asymmetry of tone, but normal development (DQ ≥85) were classified as mild. Infants with DQ of 75 to 84 were classified as mild, with a DQ of 50 to 74 as moderate, and with a DQ of <50 as severe. A further group of infants included those with no discernible development (estimated DQ <20.) All infants who died did so as a consequence of their brain injury, and for the purposes of analysis, infants who died were considered together with those with severe neurologic impairment.
In patients who had more than one scan, the most abnormal scan was used for analysis. Both T1- and T2-weighted images were used. Interobserver variability in assessment of the magnetic resonance signal intensity in the internal capsule was analyzed by calculating the kappa (κ) statistic. The following ranges for agreement were used: 0.00, poor; 0.00 to 0.20, slight; 0.21 to 0.40, fair; 0.41 to 0.6, moderate; 0.61 to 0.8, substantial; and 0.81 to 1.0, almost perfect.10 The images were initially reviewed independently to measure interobserver variability but a consensus was then reached for further analysis of the results. The predictive value of the test was assessed by calculation of sensitivity and specificity as well as positive and negative predictive values. A Bayesian approach was used to analyze the value of the test further. β density curves were calculated according to the method of Berry11 and the probability that a single examination would yield a correct prediction (the “predictive probability”11) determined, together with 95% confidence limits for that value.
Seventy-five infants fulfilled the study criteria but 2 infants were lost to follow-up and thus 73 infants were included in the study. Forty-seven infants were inborn (63%). HIE was classified as stage I in 23 infants, stage II in 37 infants, and stage III in 13 infants. The 2 infants lost to follow-up were classified as stage I.
Signal Intensity Within the PLIC
Scans were performed in the first 10 days after birth in 68 infants and between 10 and 17 days in the remaining 5 infants (Table2). The interobserver variability was very low (κ = 0.83) representing almost perfect agreement.
Normal signal intensity within the PLIC was seen in 32 infants, equivocal signal intensity was seen in 5 infants, and abnormal signal intensity in 36 infants. The signal intensity was easiest to assess on the IR sequence images. The relationship between signal intensity and stage of HIE is documented in Fig 3.
Thirteen infants had two scans within the first 10 days of life (Table2). In 5 infants with an abnormal signal intensity on their second scan, the initial scan at 1 to 4 days of age was either normal or equivocal (Fig 4 A, B).
Twenty-eight infants had a normal neurodevelopmental outcome at 1 year of age.
Forty-five infants had an abnormal outcome, which was mild in 5, moderate in 7, and severe in 33 cases. The DQs and neurologic optimality scores on all the surviving infants are documented in Table3.
The mild outcomes included asymmetries of tone in 3 infants, 1 of whom had a mild developmental delay, speech delay with convulsions in 1 infant, and delay in gross motor function in 1 infant. The optimality scores ranged from 20.1 to 20.8.
Seven infants had a moderate neurodevelopmental impairment at 1 year of age. Their optimality scores ranged from 14.6 to 17.1.
Severe Impairment or Death
Sixteen surviving infants had a severe outcome. Their optimality scores ranged from 7.2 to 12.9. Five of these infants had no discernible development (DQ <20). Their optimality scores ranged from 6.3 to 7.2. Seventeen infants died, 13 during the first week of life and 4 later in the first year. The 13 infants who died early did so as a direct consequence of their brain injury. All 4 infants that died late had severely abnormal tone patterns and required nasogastric or gastrostomy feeding. They died of respiratory complications.
Correlation of Signal Intensity Within the PLIC and Neurodevelopmental Outcome
These results are shown in Fig 5.
Thirty-two infants had a normal signal intensity: 28 of these had a normal outcome and 4 had an abnormal outcome (mild in 3 and severe in 1 infant who died). The 3 infants with a mild outcome all had persistent white matter changes on imaging. The infant who died was only scanned once at 2 days of age.
All 5 infants with an equivocal signal intensity in the PLIC had an abnormal outcome (mild in 2 and moderate in 3). All 5 infants also had changes within the basal ganglia and thalami.
Thirty-six infants had an abnormal signal intensity. All 36 infants had an abnormal outcome, 16 died and 20 survived with a moderate (n = 4) or severe (n = 16) developmental delay at 1 year of age. All infants with an abnormal signal intensity within the PLIC had additional changes within the basal ganglia and thalami. These were most obvious after the first week of life.
An abnormal or equivocal signal intensity within the PLIC was able to predict an abnormal outcome with a sensitivity of 0.90, a specificity of 1.0, and a positive predictive value of 1.0. The negative predictive value was 0.87. The predictive probability that a single test would correctly predict outcome was 0.94 (95% confidence interval, 0.88–1.0) (Fig 6).
Twenty-four infants in this study had stage I HIE, all had a normal signal intensity in the PLIC, and all had a normal neurodevelopmental outcome. Thirty-six infants had stage II HIE. Eight infants had a normal signal intensity, 4 of whom had a normal neurodevelopmental outcome. Twenty-eight infants had an equivocal or abnormal signal intensity and all of these had an abnormal outcome. Thirteen infants with stage III HIE all had an abnormal signal intensity and all had an abnormal outcome.
In infants with stage II HIE the test was able to predict outcome with sensitivity of 0.90, a specificity of 1.00, a positive predictive value of 1.00, and a negative predictive value of 0.65. The predictive probability was 0.91 (0.82–1.0).
These data show that loss of the normal signal from within the PLIC is an accurate early predictor of poor outcome in infants with HIE. In 5 infants with a normal or equivocal signal intensity on the initial scan on days 1 to 4, a repeat scan was needed to identify the abnormal signal intensity. Four of these infants had stage III HIE. All infants with an abnormal signal intensity in the PLIC on one or more scans had a poor outcome with moderate or severe impairment or death. Eighty-eight percent of infants with a normal signal intensity had a normal outcome.
Abnormal signal intensity was easy to identify and we found that the interobserver error was very small (κ = 0.83). Both observers were experienced at assessing neonatal images. To determine the value of the loss of signal intensity from the PLIC for predicting later neurologic impairment, we calculated the sensitivity and specificity, together with the positive and negative predictive values of the test when applied to the population we studied. We used a Bayesian approach to calculate the value of the test for a population. β densities were estimated to calculate the predictive probability that a single examination would give the correct prognosis. This was extremely high and the upper confidence limit approached 100%.
Infants with stage I HIE have a good prognosis with no reports of a severe outcome whereas those with stage III HIE are known to have a probability of between 75% and 100% of suffering neurodevelopmental impairment.8 12 13 Infants with stage II HIE present more difficulties with up to 25% having a poor outcome of severe handicap or death8 12 13 and a more accurate diagnostic test would be particularly valuable. In this study it was possible to predict the outcome in 33 of the 36 (92%) infants with stage II HIE. The positive predictive value was 1.00 and the predictive probability was 0.91 (0.82–1.00). Although there was a high percentage (73%) of stage II infants with a moderately or severely abnormal outcome in this study the use of the Bayesian approach helps to take into account variations in disease prevalence between study populations.
Myelination is associated with a shortening of T1 and T2 on MRI. This is seen as high signal intensity on a T1-weighted sequence and low signal intensity on a T2-weighted sequence. In histologic studies the onset of myelination in the PLIC occurs between 32- and 36-weeks gestational age,14-16 but evidence of myelin is not seen with MRI until 37 weeks gestation.17 Before 37 weeks it may not be possible to identify the PLIC or it may be seen as low signal intensity on T1-weighted images. The absence of a magnetic resonance signal from myelin in the PLIC in an infant of 40-weeks gestation is, however, abnormal. It is therefore important to have an accurate assessment of gestational age before using this sign as a predictor of poor prognosis in infants with signs consistent with HIE. Delayed myelination with absent myelin in the PLIC may also be found in infants with metabolic disorders, some of which may mimic HIE.18
In some infants who develop an abnormal signal intensity, the posterior limb can seem relatively normal in the first 24 to 48 hours. The delay in the loss of signal confirms that there was evidence of myelination before the hypoxic-ischemic insult. It has been postulated that asphyxiated infants show delayed myelination in the internal capsule,19 but these results show that the abnormal signal intensity represents acquired changes in preexisting myelin rather than a lack of myelin.
The abnormal signal intensity within the internal capsule may be caused by edema or infarction. The abnormal low signal intensity does disappear in all infants and a high signal intensity consistent with myelin usually reappears although this may be irregular in appearance and may take several months to appear. In infants with severe atrophy of the basal ganglia and thalami the normal high signal from myelination in the internal capsule may not reappear.2
Studies have suggested that actively myelinating tissue is particularly sensitive to hypoxic-ischemic damage because of its increased metabolic rate.19 This may explain the susceptibility of the corticospinal tracts to damage in infants with HIE. The normal term brain is actively myelinating in the corticospinal tracts as they arise in the postcentral gyrus, traverse the centrum semiovale, and enter the posterior limb of the internal capsule. In the asphyxiated infant changes may be seen on MRI in all these areas, although the PLIC is the only region where originally high signal intensity on T1-weighted sequences is lost.1 In infants with HIE the corticospinal tracts higher in the brain may show an increase in signal intensity with T1-weighted sequences.1The basal ganglia and thalami are also sensitive to hypoxic-ischemic injury probably because of their relatively rapid metabolic rate due in part to the active myelination.20
The exact relationship between abnormalities within the internal capsule and the development of a central motor deficit is not clear. Although an abnormal signal intensity within the PLIC was often the only abnormality seen with an early MRI, later scans always showed the presence of other lesions. Unilateral abnormalities within the PLIC are associated with the development of a hemiplegia in adult stroke patients.21 The presence of additional lesions within the thalamus is associated with a more severe motor outcome in adults.21 In our study, the presence of bilateral abnormalities within the PLIC was always associated with the development of a central motor deficit. The severity of this deficit may be related to the extent of additional basal ganglia and thalamic injury, which will be most obvious during the second week of life.2
Early magnetic resonance imaging of term infants with HIE can be used to predict outcome. Very early scanning of the asphyxiated infant who still requires intensive care may show an abnormal signal intensity within the posterior limb. In our experience, the neurodevelopmental outcome in such infants is always severely abnormal. This information may help make early management decisions, however, an abnormal signal may take up to 4 days to develop and repeat imaging may be necessary in some infants with stage II or III HIE.
A few stage II infants with a persistently normal signal intensity within the posterior limb may have a mildly abnormal outcome. These infants may have either areas of white matter infarction or mild focal lesions within the basal ganglia. White matter infarction can be detected very early with diffusion-weighted imaging.22Therefore the combination of an IR sequence and diffusion-weighted imaging during the first week of life will correctly predict outcome in the majority of infants. The severity of any basal ganglia and thalamic lesions will be apparent on an image taken during the second week of life.2 Conventional imaging during the second week of life will identify the exact pattern of injury throughout the brain and this information can be used to predict both outcome and the severity and nature of an abnormal outcome.2
We thank the Medical Research Council, Picker International, and the Garfield Weston Foundation for their support.
- Received February 15, 1997.
- Accepted February 19, 1998.
Reprint requests to (M.A.R.) Robert Steiner Magnetic Resonance Unit, Hammersmith Hospital, Du Cane Rd, London, W12 OHS, England.
- Keeney SE,
- Adcock EW,
- McCardle CB
- Baenziger OE,
- Martin M,
- Steinlin M,
- et al.
- Keunzle C,
- Baenziger O,
- Martin E,
- Thun-Hohenstein L,
- Steinlin M,
- Good M
- Barkovich AJ,
- Westmark K,
- Partridge C,
- Sola A,
- Ferriero DM
- ↵Griffiths R. The Abilities of Babies. London, England: University of London Press; 1970
- ↵Berry D. Statistics, A Bayesian Approach. London, England: Duxberry Press; 1996:200–215
- ↵Peliiowski A, Finer NN. Birth asphyxia in the term infant. In: Sinclair JC, Bracken MB, eds. Effective Care of the Newborn Infant. Oxford, England: Oxford University Press; 1992:249–279
- ↵Gilles FH, Shankle W, Dooling EC. Myelinated tracts: growth patterns. In: Gilles FH, Lenton A, Dooling EC, eds. The Developing Human Brain, Growth and Epidemiologic Neuropathology. Boston, MA: John Wright; 1983:117–183
- ↵Yakolev PI, Lecours A. The myelogenetic cycles of regional maturation of the brain. In: Minkinski A, ed. Regional Development of the Brain in Early Life. Oxford, England: Blackwell; 1967:3–70
- ↵Larroche JC. The development of the central nervous system during intrauterine life. In: Flakner F, ed. Human Development. Philadelphia, PA: WB Saunders; 1966:257–276
- Ball WS
- Barkovich AJ,
- Truwit LT
- Chugani HT,
- Phelps ME
- Copyright © 1998 American Academy of Pediatrics