Early Prognostic Indicators of Outcome in Infants With Neonatal Cerebral Infarction: A Clinical, Electroencephalogram, and Magnetic Resonance Imaging Study
Objective. The aim of this study was to identify prognostic factors in newborns with cerebral infarction.
Design. Antenatal and perinatal factors and early clinical, electroencephalogram (EEG), and magnetic resonance imaging (MRI) findings were compared with neurodevelopmental outcome in 24 children with evidence of cerebral infarction on neonatal MRI.
Results. Out of 24 infants, 19 had an infarction in the territory of a major cerebral vessel and 5 in the borderzone between cerebral arteries. Neuromotor outcome was normal in 17 and abnormal in 7 infants. Of these 7 infants, 5 infants showed a definite hemiplegia, whereas the other 2 showed some asymmetry of tone or function but no definite hemiplegia.
None of the adverse antenatal or perinatal factors was significantly associated with abnormal outcome. Neonatal clinical examination was also not always predictive of the outcome. The extent of the lesion on MRI was a better predictor. In particular, it was the concomitant involvement of hemisphere, internal capsule and basal ganglia that was always associated with an abnormal outcome whereas the involvement of only one or two of the three tended to be associated with a normal outcome.
EEG was also very helpful. Abnormal background activity either unilateral or bilateral was found in 6 infants and 5 out of 6 developed hemiplegia. In contrast, the presence of seizure activity in presence of a normal background was not related to abnormal outcome.
Conclusions. Early MRI and EEG can help to identify the infants with cerebral infarction who are likely to develop hemiplegia.
- cerebral infarction
- neurological examination
- magnetic resonance imaging
- MCA =
- middle cerebral artery •
- EEG =
- electroencephalogram •
- MRI =
- magnetic resonance imaging •
- CTG =
- cardiotocography •
- CT =
- computed tomography
Until relatively recently, neonatal cerebral infarction was thought to be an uncommon condition, usually associated with perinatal complications and with a very unfavorable outcome. The reported incidence was mainly based on findings from postmortem studies.1 ,2 In the last two decades, as a result of wider availability of brain imaging in neonates, cerebral infarction has been shown to be relatively common.3–7 The majority of the infarcts occur in full-term infants who do not show signs of birth asphyxia, but present with seizures in the first hours or days of life. The new techniques of neuroimaging also provide more detailed information on the extent of the infarcts, showing that they most frequently involve one or more branches of the middle cerebral artery (MCA), with the left hemisphere more frequently affected than the right.
The majority of previous studies on neonatal infarcts described infants with lesions in the territory of the MCA. In most of these studies the incidence of hemiplegia was quite high, in fact similar to that observed in adults suffering a stroke.9–14 Only a few studies have reported that infarction is not always associated with abnormal outcome.3 ,8 ,15 ,16 As a result, parents of newborns with perinatal infarcts have often been counseled about the high risk of their children developing hemiplegia. There is little information on whether there are perinatal or neonatal factors that may differentiate the children who have a normal outcome from the ones who develop hemiplegia.
The aim of this study was to evaluate whether there are antenatal/perinatal factors, early clinical signs, electroencephalograms (EEG), or magnetic resonance imaging (MRI) findings in the neonatal period that might help in predicting later neurodevelopmental outcome.
SUBJECTS AND METHODS
The study is part of an ongoing longitudinal prospective project aimed at documenting the evolution of neonatal cerebral lesions. As part of this study all the infants who present with birth asphyxia and/or neonatal convulsions undergo neonatal brain MRI. The study has been approved by the Research Ethical Committee of the Imperial College School of Medicine. Twenty-four full-term infants born at or referred within 48 hours of birth to the Hammersmith Hospital and who showed evidence of cerebral infarction on neonatal MRI were enrolled. In all the lesions were also evident on the cranial ultrasound scan performed at the end of the first week of life. All the infants were investigated after the onset of convulsions between day 1 and day 3. All were treated with phenobarbitone that was always discontinued by the end of the first week of life.
A detailed history of antenatal and perinatal events was obtained from the parents and from obstetric and pediatric notes and recorded on a proforma. The antenatal factors investigated were evidence of maternal systemic diseases, intrauterine growth retardation, elevated blood pressure, bleeding, infections, reduced fetal movements, trauma, and pain. Perinatal factors were meconium-stained liquor and abnormal cardiotocography (CTG), cord pH, type of delivery, and Apgar scores.
Details of the neurological status of the infants at birth and before the onset of seizures were obtained from the obstetric and pediatric notes. Neurological examination was performed and recorded on a standardized proforma17 soon after admission.
Magnetic Resonance Imaging
The infants were imaged on a 1.0-tesla Picker system (Cleveland, OH) using conventional T1 weighted spin echo (SE 860/20), inversion recovery (IR 3800/30/950), and T2 weighted spin echo (SE 3000/120) sequences. Two different classifications were used, according 1) to the arterial distribution of the lesions and 2) to the degree of involvement of cortical and subcortical structures.
Classification According to Arterial Territory
Based on the location, extent, and shape of the lesions, infarcts where characterized as being in the territory of the main arteries or in a borderzone distribution, ie, in the watershed area between the end fields of main arterial territories. These lesions are usually multiple and bilateral involving the posterior convexities and, less frequently, the anterior lobes.
The infarcts in the territory of the main arteries were first classified according to the main artery involved. The infarcts in the territory of the MCA were further subdivided according to a modified version of the criteria suggested by de Vries et al. 5into 1) main branch, 2) cortical branches, and 3) lenticulostriate branches.
In addition to these three categories, we classified separately the infarcts in the territory of the MCA that were associated with contralateral lesions.
Classification Based on the Type of Cerebral Structures Involved
This was based on the site and concomitant involvement of hemispheric and subcortical structures. The lesions were classified according to the possible combination of the involvement of the hemispheres and/or subcortical structures such as internal capsule, basal ganglia, thalamus, and brainstem.
The EEG was performed as soon as possible after the onset of convulsions in the infants with normal Apgar scores. In 14 infants a continuous EEG was recorded using a four-channel Oxford Medilog system recording from F3-P3 and F4-P4 using silver/silver chloride electrodes. From January 1996 a digital Micromed system recording from eight bipolar leads was used. Three infants had a full standard EEG (16 channels) soon after the onset of seizures in the hospital where they were born. EEG was classified according to the continuity and symmetry of frequencies and amplitude of the background activity and to the presence of seizures, into 1) normal (continuous and symmetrical background and no epileptic discharges); 2) normal background but presence of epileptic discharges (uni- or bilateral); 3) abnormal background (discontinuous or asymmetrical) but no epileptic discharges; and 4) abnormal background and presence of epileptic discharges.
Neurological examination was performed using a structured proforma, looking in particular at the symmetry in posture, tone, power, reflexes, hand, and leg function. Griffiths developmental scales18 were used to evaluate global development.
Nonparametric statistical analysis (Fisher's exact test) was used to correlate antenatal, perinatal, and neonatal findings to clinical outcome.
Antenatal and Perinatal Events
Adverse antenatal factors were present in 11 of the 24 infants. Fourteen had an abnormal CTG (continuous decelerations below 90 with slow recovery) and 11 had meconium-stained liquor grade 2 or 3. Two infants had cord pH values below 7.1 and none below 7. Apgar scores below 5 at 1 minute were found in 5 infants. Individual details of antenatal and perinatal findings are shown in Table 1.
Neonatal Neurological Examination
The neurological examination was always performed after the onset and the treatment of convulsions, soon after the admission or on the day when MRI was performed (days 3 to 5). Neurological assessments were rated normal for 6 of the 24 infants; 8 showed mild diffuse hypotonia but no asymmetry; and 6 had some asymmetry in their tone pattern. Individual details are shown in Table 2.
Details of the extent of the lesions on MRI are given in Table 3.
Focal changes in the distribution of one of the main arteries were present in 19 infants; 5 five showed infarcts that were in a borderzone distribution.
Cerebral Artery Infarcts
In all the 19 infants with arterial infarct, the changes were in the territory of the MCA.
Changes in the territory of one of the cortical branches (Fig 1) were present in 12 infants. Two infants showed changes the territory of the lenticulostriate branch (Fig 2), involving the lentiform and caudate in one infant and the lentiform only in the other infant.
Five infants showed an infarct in the territory of the MCA associated with a contralateral ischemic lesions. Three of the 5 infants showed the unilateral involvement of the main branch (Fig 3) associated with contralateral changes in the territory of a cortical branch (n = 2) or in the mesencephalon (n = 1); the other 2 showed unilateral infarction in a cortical branch territory with contralateral white matter or basal ganglia changes.
Five infants showed changes in the watershed areas that were consistent with borderzone lesions (Fig 4).
Hemispheric and Subcortical Involvement
Seven children had hemispheric involvement only with spared internal capsule and basal ganglia. Eight children had hemispheric lesions and involvement of either internal capsule or basal ganglia. Two children had involvement of internal capsule and basal ganglia but normal hemisphere. Involvement of hemispheres, basal ganglia, and internal capsule was seen in 8 children.
The EEG was recorded in 20 infants between day 2 and day 4 (mean 46 hours). Of the 20 infants, 8 had a completely normal EEG. One other infant had abnormal background not associated with epileptic features. Six infants showed epileptic discharges in the presence of a normal background whereas other 5 showed epileptic discharges and abnormal background. Individual details of the EEG findings are shown in Table 2.
The duration of the follow-up ranged from 15 months to 5 years and 6 months. Fifteen children were normal on both neurological examination and developmental assessment. Hemiplegia was present in 5 children; 4 of the 5 children showed a hemiplegia associated with normal development in 4 of the 5 and with mild global delay in the other subjects. Two children showed some asymmetry on the neurological examination but no definite hemiplegia; both had normal neurodevelopment. Two other children showed a mild neurodevelopmental delay but normal neurological examination. None of them has, so far, had seizures after the neonatal period. Details are shown in Table 2.
Correlation between Antenatal/Perinatal Factors and Outcome
Antenatal factors, CTG, cord pH, mode of delivery, Apgar scores and resuscitation were all individually assessed and compared with outcome. None of them showed a statistically significant association.
Correlation between Neonatal Neurological Examination and Outcome
Six children had a normal neonatal assessment, 5 of the 6 had a normal outcome and 1 showed a mild delay but no hemiplegia.
Of the 8 children who showed mild generalized hypotonia on the neonatal assessment, 5 had a normal outcome, 1 had some asymmetry but no definite hemiplegia, and 2 showed a definite hemiplegia.
Of the 6 infants who showed some asymmetry on tone patterns on the neonatal assessment, 3 had a normal outcome, 1 showed some asymmetry, and 2 showed hemiplegia.
Correlation between MRI and Outcome: Arterial Distribution
Main Cerebral Artery Infarcts
Ten of the 12 children with unilateral infarction in the territory of a branch of the MCA had a normal outcome. Of the remaining two children, 1 showed some asymmetry in tone on neurological examination and a strong hand preference but no definite hemiplegia, and 1 had hemiplegia.
One of the 2 children with lenticulostriate infarction showed normal outcome and the other a mild global delay with mild hypotonia but no asymmetry.
One of the 5 infants with infarcts of the cerebral artery associated with contralateral lesions had a normal outcome, 1 showed some asymmetry, and 3 hemiplegia.
Four of the 5 infants with borderzone lesions had a normal outcome and one a hemiplegia. Details of this correlation are shown in Fig 5.
Correlation Between MRI and Outcome: Hemispheric and Subcortical Involvement
Figure 6 shows details of the correlation between arterial distribution and outcome.
No significant correlation could be found between the involvement of any of the subcortical structures (internal capsule, midbrain, basal ganglia, and thalami) and the clinical outcome (P> .05).
Of the 7 children had hemispheric involvement only with spared internal capsule and basal ganglia, all the 7 had normal outcome. Of the 8 children with hemispheric lesions and involvement of either internal capsule or basal ganglia, all 8 had normal outcome. Two children had involvement of basal ganglia and internal capsule but normal hemispheres; both were normal. In 8 children in which there was involvement of the hemispheres, basal ganglia, and internal capsule, 1 of the 8 had a normal outcome, 2 showed asymmetry, and 5 showed hemiplegia.
The concomitant involvement of the thalamus did not seem to increase the risk of abnormal outcome.
Details of this correlation are shown in Fig 6.
Correlation Between EEG and Outcome
Of the 8 infants who experienced a normal neonatal EEG with normal background and no epileptic features, 7 of the 8 had a normal outcome and 1 a mild global delay but no hemiplegia. Six children had a normal background but epileptic features, 5 of the 6 had a normal outcome and 1 showed a mild asymmetry but no hemiplegia. Six infants had abnormal background, associated with epileptic features in 5 of the 6, one was normal and the other 5 had hemiplegia. Details of this correlation are shown in Fig 7.
The incidence of hemiplegia observed in our cohort of infants with neonatal cerebral infarction was approximately 20% (5 of the 24 children studied). Another 2 infants showed some asymmetry but no definite hemiplegia. Although the duration of the follow up was only for 15 months in 1 of the 2 infants with asymmetry, we can not exclude that this infant might develop a more severe impairment. Even taking this possibility in account, however the incidence of abnormal motor outcome would still be only 25%. Only 1 of the 5 children with a hemiplegia showed signs of a more global neurodevelopmental delay. However, 2 of the children without motor impairment have a mild global developmental delay.
The aim of this study was to identify whether there are any antenatal, perinatal or neonatal adverse factor that might help to predict hemiplegia in infants with perinatal infarction. None of the antenatal or perinatal adverse factors that we examined were significantly associated with hemiplegia. Abnormal signs on early neonatal neurological examination were also very poor prognostic indicators. Most of these children only showed a mild degree of generalized hypotonia, which could be due to the anticonvulsant, and this showed no significant association with motor outcome. Asymmetries on neurological examinations, when present, were often transient and were also not always associated with late hemiplegia.
The vascular distribution of the lesion on neonatal MRI was not always predictive of the outcome. Both normal and abnormal outcome could be found in infants with focal infarcts or borderzone lesions, although infants with involvement of the main branch associated with contralateral lesions tended to have a less favorable outcome.
In contrast, the overall involvement of hemispheres, basal ganglia, and internal capsule seemed to be more consistently associated with outcome. Although the children with the involvement of all these regions tended to show hemiplegia or asymmetry of tone, the ones who showed involvement of only one a combination of two of the three regions tended to have a normal outcome.
The early neonatal EEG was also a reliable prognostic indicator of motor outcome in our cohort. More specifically, it was the abnormal background that tended to be associated with abnormal outcome whereas other EEG abnormalities such as epileptic discharges in the presence of a normal background, were not associated with abnormal outcome. In most of our cases only two channels recording EEG were available. Nevertheless our findings suggest that an abnormal background even if only recorded on two channels and for a short time is a valuable predictive indicator.
Our results show a much better outcome for perinatal infarction than that reported by others where, with few exceptions3 ,8 ,15 ,16, a high incidence of hemiplegia has been found.5 9–14 The difference can not be due to the inclusion in our study of infants with borderzone lesions because if we exclude them from our cohort the incidence of hemiplegia remains similar. We feel the most likely explanation must lie in the extent of the tissue involvement. Although in most previous studies the lesions are all reported at MCA this does not necessarily imply that the size and sites of the lesions were identical. Most of these studies used ultrasound and computed tomography (CT)7 ,10 ,12 ,15 ,16 19–23 that provide far less anatomic detail than MRI and this makes it difficult to compare size and site of tissue involvement between studies. In the few studies that used MRI, this was mainly performed after the neonatal period.9 ,24 ,25 In the only study in which MRI was performed in the neonatal period5 hemiplegia mainly followed when the lesions resulted from involvement of the main branch of the MCA. The involvement of the main branch of the MCA was observed in only 3 of the 24 infants, always associated with contralateral changes. The motor outcome in all the 3 children was abnormal.
The overall low incidence of hemiplegia in our cohort therefore probably reflects the extent of the lesion. The reason that we have detected many relatively localized lesions may be because of our study design whereby we are easily able to obtain early and follow up magnetic resonance images in all infants presenting to us with seizures. This imaging protocol allows us to detect some lesions not seen on early ultrasound and also identify small lesions that might be less evident on CT or late MRI. As most of these children have a relatively benign neurological course after the first few days these lesions would be missed if not fully investigated with MRI.
With MRI becoming more widely available for investigation of all neonatal seizures, it is likely that the detection of small lesions will be more frequent. Assessment of the sites of involvement on MRI and of the background EEG in the neonatal period can provide early guide to prognosis and we would suggest that these procedures should become standard practice whenever possible. This might help to reduce the anxiety that is inevitable when these lesions are diagnosed in the neonatal period.
We thank David Wertheim and the late Cindy Bradshaw for their help in recording and interpreting the EEG.
- Received March 27, 1998.
- Accepted June 30, 1998.
Reprint requests to (E.M.) Dept of Paediatrics, Hammersmith Hospital, Du Cane Rd, London W12 OHN, United Kingdom.
- ↵Larroche JCL. Occlusion of the cerebral arteries. In: Developmental Pathology of the Neonate. Amsterdam, The Netherlands: Elsevier/North Holland; 1977:135–161
- ↵Dubowitz L, Dubowitz V, Mercuri E. The neurological assessment of the preterm and full-term newborn infant. In: Dubowitz L, Dubowitz V, Mercuri E, eds. Clinics in Developmental Medicine. 2nd ed. Cambridge, UK: Mc Keith Press; 1998
- ↵Griffiths R. The Abilities of Babies: A Study in Mental Measurements. Amersham, The Netherlands: Association of Research in Infant and Child Development; 1976
- Hill AD,
- Martin J,
- Danemann A,
- Fitz CF
- Copyright © 1999 American Academy of Pediatrics