Behavioural Brain Research, 49 (1992) 7-19 9 1992 Elsevier Science Publishers B.V. All rights reserved. 0166-4328/92/$05.00 BBR01310

Neuroimaging and functional outcome of neonatal leukomalacia G. Cioni a, L. Bartalena b, E. Biagioni", A. Boldrini b and R. Canapicchi c a blstitute of Child Neurology and Psychiatr), Unirersit), of Pisa and Stella Marls Foundation, Pisa (Ital)9, bNeonatal Intensire Care Unit, Unirersity of Pisa, Pisa (Italy) and ~Department of Neuroradiology, St. Chiara Hospital, Pisa (ltaly) (Received 8 August 1991) (Revised version received 30 September 1991) (Accepted 21 November 1991) Key words: Cystic leukomalacia; Prolonged flare; Ultrasound; Electroencephalogram; Magnetic resonance imaging; Neurological outcome

Leukomalacia is a major cause of neurological impairment in the high-risk newborn. It can be identified during the early postnatal period by means of ultrasound (US) imaging of the brain, through the anterior fontanel. Magnetic resonance imaging (MRI) permits an optimal differentiation of brain tissue and of its abnormalities, without resorting to ionizing radiation or intravenous contrast. It is particularly appropriate for following the evolution of leukomalacia, after fontanel closure. Ninety-five fullterm and preterm infants with cystic and non-cystic leukomalacia, documented by US, were clinically followed-up until at least 12 months of corrected age. Thirty-two had a severe neurological outcome (mainly cerebral palsy, sometimes associated with mental retardation and/or cerebral visual impairment). The prognosis was worse in cystic leukomalacia than in prolonged flare. Electroencephalogram (EEG) carried out in the first 2 weeks of life provided valuable indexes of further outcome, especially for U S findings of more uncertain prognosis. M RI was carried out at around 12 months of corrected age, by means of an apparatus operating at 0.5 Tesla. The main categories of abnormalities observed were the following: cystic lesions, enlarged ventricles with irregular outlines, delayed myelination, high intensity areas in the long TR (repetition time) images within the white matter, cortical atrophy. MRI findings correlated well with the results of US imaging and often with motor, cognitive and visual impairments. Nevertheless, clinical features cannot be predicted by neuroimaging alone and a comprehensive approach, including longitudinal functional and electrophysiological testing, is highly recommended.

INTRODUCTION

Perinatal brain lesions, i.e. injuries which affect the central nervous system (CNS) at around the time of delivery, are the major cause of neurological deficits in infancy. Newer imaging techniques, and cranial ultrasound (US) in particular, have enabled the diagnosis of these lesions to be made during the first days of life and their evolution in the first months of life to be followed 'in vivo'. Thanks to this technique, the incidence of the various types of neonatal brain, injuries can now be established in different at risk populations (preterm infants, fullterm asphyxiated newborns, infants small for gestational age...), together with their functional consequences for the CNS. Very low birth weight infants in particular are at high-risk of hypoxic-ischaemic and/or haemorrhagic

Correspondence: G. Cioni, Instituie of Child Neurology and Psychiatry, University of Pisa and Stella Maris Foundation, 56018 Calambrone, Pisa, Italy. Fax:(39)(50)32214.

damage. Recent follow-up studies indicate that, although intraventricular haemorrhage is frequent, only a small percentage of these infants manifest some neurological deficits, whereas ischaemic lesions, although rarer, are more often the cause of a permanent motor, cognitive and sensory impairment 29. Ischaemic lesions tend to occur in different regions of the brain in preterm infants, in comparison with term ones. This is due to the different location in the two populations of the vascular watershed areas between centripetally and centrifugally oriented vessels. Whereas in the preterm infants these watershed regions, sensitive to ischaemic insults, are mainly located in the periventricular white matter, in the fullterm they lie in the boundary zones between the cerebral arteries and in the white matter under the sulci 6. The term 'leukomalacia', literally 'softening of the white matter', is now widely used to indicate this necrotic process. The pathological evolution of these lesions changes in the period after the insult: coagulation necrosis often gives rise to srriall cavities (cysts) which usually disappear, leaving areas of gliosis. US can identify the early phase ofleukomalacia and,

very precisely, its cystic components, highly correlating with the autopsy findings in patients who died 8'32. Leukomalacia may be classified according to US features into cystic and non-cystic leukomalacia (often indicated as prolonged flare), and, on the basis of the location of the lesions, into periventricular leukomalacia and deep white matter (or 'subcortical' as it is generally called) leukomalacia. The advantages and pitfalls of US in the diagnosis of leukomalacia are amply discussed by de Vries et al. 11. Computed tomography (CT) has a limited value in the acute phase ofleukomalacia~4"~6; this is mainly due to the need of transporting patients, the high water content of the immature brain and the low specificity of periventricular hypodensity. Moreover, small cysts are hardly recognizable, because of the thickness of CT slices. On the contrary, magnetic resonance imaging (MRI) shows more promise from the neonatal period on 2"~~176 being able to detect even small cysts in the white matter of regions difficult to explore via US. Moreover, it permits a differentiation of any subacute haemorrhagic components of the lesions. The examination does not resort to ionizing radiation or intravenous contrast, but it requires the patient's sedation and transportation. For these reasons, MRI is particularly more appropriate for following the evolution of leukomalaeia beyond the fontanel closure, when US studies become impossible. Compared to CT scan, in this phase MRI allows the identification of even small areas of gliosis and of other brain abnormalities, especially of the white matter. Moreover, it allows us to follow the progress of myelination, which is often delayed in these patients. In this paper the main MRI abnormalities observable in the early and the late phase of leukomalacia will be described, in connection with US findings. Pathological features shown by imaging techniques will be compared with motor, cognitive and visual outcomes at the follow-up and with the results of neonatal EEG. By reviewing a large set of leukomalacia cases, the diagnostic and prognostic values of US, M R I and EEG will be discussed.

PATIENTS AND M E T H O D S

From the records of all the patients admitted to the Neonatal Intensive Care Unit of the University Hospital of Pisa between January 1, 1987 and April 30, 1990, we selected all the cases fulfilling the following criteria:

(1) serial US scans executed within the neonatal period showing a cystic leukomalacia or a flare lasting more than two weeks, with or without associated intraventricular haemorrhage; (2)no congenital neurological diseases; (3)neurological follow-up till at least 12 months of corrected age. Ninety-five infants (56 M, 39 F) met all the selection criteria. Their mean gestational age was 31.9weeks (range 25-40); 86 were born prematurely, and 9 at term. Mean follow-up duration was 15.3 months of corrected age (range 12-48). Cystic leukomalacia was detected in 25 cases and prolonged flare in 70 infants. Neurological assessment Neurological examination was carried out during the preterm period according to Dubowitz and Dubowitz ~2, at term age by means of Prechtl's method zS, and later on according to Touwen's items 3~. Developmental scales (Griffith test~9; Uzgiris and Hunt scales 33) were applied at 12 months of corrected age and then repeated regularly at 6-12 months intervals. All the infants were ophthalmologically screened in the neonatal intensive care unit; further ophthalmologic examinations and behavioural assessment of visual functions, including visual acuity cards 2z'3~ were carried out in the majority of cases. Ultrasound examination Cranial ultrasound scannings (US) were performed by means of an Echo-real time scanner, equipped with a 5 MHz transducer (and from January 1990 also with a 7.5 M H z head). -Sections were systematically obtained in the coronal and sagittal planes. The first US was carried out in the first 48 h and than repeated at different intervals according to the postmenstrual age of the infant and to the previous US findings, until discharge from the unit. Subsequently, infants with an abnormal scan had at least one further examination before fontanel closure. Leukomalacia was classified according to its distribution in the brain tissue and to the presence of the cysts as cystic leukomalacia (divided into cystic periventricular leukomalacia and the more diffuse leukomalacia) and prolonged flare (for a density lasting more than 2 weeks). Peri-intraventricular haemorrhage was graded according to Volpe34: grade 1 (germin.41 matrix haemorrhage, with no or minimal intraventricular bleeding); grade 2 (intraventricular haemorrhage occupying 10-50~o of the ventricular area); grade 3 (intraventricular haemorrhage occupying more than 50% of the ventricular area, usually distending the ven-

tricles); haemorrhagic intracerebral involvement or other parenchymal lesions. In 22 of the 25 patients with cystic leukomalacia, all preterms, the cysts were confined to the periventricular white matter; in three other cases (all fullterms with a severe asphyxia) large cysts were observed, extending also into the white matter under the sulci (deep white matter of subcortical leukomalacia). Leukomalacia was often associated with peri-intraventricular haemorrhage. In particular, in 12 of the 70 cases with prolonged flare, parenchymal echodensity was complicated by a severe intraventricular haemorrhage (grade 3).

Magnetic resonance hnaging Magnetic resonance imaging (MRI) studies were performed on 33 of the infants with leukomalacia, by means of a system operating at 0.5 Tesla (MR MAX, GE-CGR), with sections of 5 mm in sequence spinecho. T 1, proton density and Tz weighted images were obtained in the axial, coronal and sagittal planes. Infants were examined during sleep, after sedation induced by flunitrazepam. MRI was performed in the neonatal period only in a few patients, when an optimal differentiation of the lesions was needed, and if the infant was suitable for transportation. In the majority of cases, MRI was carried out after fontanel closure, and generally at around 12 months of age (corrected for prematurity). In fact, at this age the progress in the myelin permits a better differentiation of areas of gliosis in the white matter, and of delayed myelination. A final check is foreseen after 3 years, but so far few infants have reached that age. Each infant was examined with the informed consent of a parent. MRI was systematically executed in all the cases of cystic leukomalacia, or when neurological symptoms were still relevant after the first months of life. At the examination of the images, the presence of the following abnormalities, characteristic of leukomalacia according to the literature 2a~ was checked: (a) cystic lesions (presence, size and site); (b)areas of high intensity in the long TR (repetition time) images; (c) reduction in quantity of the white matter in the periventricular regions; (d)signs of atrophy of the corpus callosum and the brainstem; (e)ventricular dilatation, particularly at the trigones, with irregular outlines; (O widening of he subarachnoid spaces of the mantle and deep and prominent sulci; (g) signs oflesions of the basal ganglia; (h) delayed myelination.

Electroencephalogram EEGs, lasting 60-120 min, were performed on the large majority of the infants with leukomalacia. The

examinations were carried out within the first 2 weeks and generally in the first 4-5 days of life. In fact, the literature indicates that the diagnostic and prognostic value of the EEG is considerably higher when recorded in an early phase after a CNS insult. Computer-aided portable devices (BS and BQ-100 by Micromed Company, Italy) were used; these devices, being battery-operated, reduce the number of artefacts due to the electrical apparatuses of the neonatal intensive care unit. Eight active electrodes (Fp~_2, C 3 _ 4 , O1_2, T3_4 of the International 10-20 System) were applied to the scalp. Tracing were transferred to a Central Unit (BQ-240 by Micromed Company) and then examined by an operator who had no information about the clinical and US picture of the newborn. Ictal and background EEG findings and abnormalities were classified according to the literature 21. Particular attention was payed to EEG maturational features. In many cases the EEGs were also submitted to spectral analysis and to brain mapping using a software written at the Institute of Neurophysiopathology of the University of Genoa (CEA-1). After the selection of at least 40-50 artefact free epochs (every epoch lasts 2 s)~ the tracing was submitted to spectral analysis (by means of Fast Fourier transformation). Afterwards, a gross topographic representation ofthepower of selected frequency bands on the scalp (brain mapping) was also obtained. Further details on EEG recording and analysing are given elsewhere4'5.

Neonatal US and Neurological Outcome 70

50 40 30

E c

20 10

c-LM IlK

severe abnorm,

P F + IVH 3 ~

mild abnorm,

PF r---I

normal outcome

Fig. 1. Comparison of neurological outcome at > 12 months of corrected age in 25 infants with cystic leukomalacia (c-LM), 58 with prolonged flare (PF) and 12 with prolonged flare associated with intraventdcular haemorrhage (PF+IVH3), graded according to Volpe (1987).

10

Fig. 2. A: coronal US scan showing hyperdensity in the right thalamus and in the periventricular areas. B: MRI of the same infant 7 days after US; sagittal Tl-weighted image, showing a small cystic lesion in the right thalamus, low intensity of the ~shite matter due to incomplete myelination. C: coronal Ti-weighted image through the trigones; bilateral cystic lesions in the supraventricular white matter. RESULTS

Clinical outcome

The results of developmental follow-up of the 95 infants with leukomalacia are reported in Fig. 1.

According to ,US findings the infants were divided into the three groups of cystic leukomalacia (25 cases), prolonged flare (58 cases) and prolonged fla~e associated with grade 3 intraventricular haemorrhage (12 cases).

11 On the basis of motor and cognitive development, the outcomes were defined as follows: (1) normal outcome: no motor abnormality, developmental quotient (DQ>~85); (2)mild abnormality: isolated mild motor abnormality (no cerebral palsy), and/or mild developmental retardation (DQ = 71-84); (3) severe abnormality: severe motor impairment (cerebral palsy) and/or severe developmental retardation (DQ ~ right) supraventricular cystic lesions; reduction of periventricular white matter with secondary prominence of cortical sulci and fissures, ventricular dilatation with irregular outlines. C: coronal Tl-weighted image, more posterior; cystic lesion in the left supraventricular area is more prominent. D: axial proton density (left) and Tz-weighted (right) images; high intensity of the periventricular white matter; delayed myelination in both parietal areas.

13 one case showed a mild developmental retardation, with a gradual recovery after the first weeks of life, and the last 4 infants showed transient motor abnormalities, but normalized completely afterwards. Infants with a prolonged flare generally had a good evolution. Forty-five of the 58 infants were normal at 12 months, 6 presented a mild developmental retardation (mainly in the motor aspects) and 7 severe neurological abnormalities. They consisted of a cerebral palsy in 5 cases (4 diplegias and one atethoid form), associated in two cases .with a severe developmental retardation. Cerebral visual impairment was also found to be very rare in these patients. The evolution of the patients with prolonged flare associated to grade 3 intraventricular haemorrhage was considerably worse; 5 infants showed a cerebral palsy (one tetraplegia, three diplegias, one hemiplegia). One baby had a mild motor abnormality in one arm, and 6 had a normal development. A cerebral visual impairment was observed in one of the infants.

MRI findings Thirty-three of the 95 infants with a leukomalacia on the US (20 with cystic leukomalacia, 8 with prolonged flare, and 5 with prolonged flare associated to grade 3 intraventricular haemorrhage), had one or two MRI examinations. In the early phase of cystic leukomalacia, MRI permits the identification of even small cysts (Fig. 2), whereas the lack of myelination makes non-cystic lesions in the white matter difficult to detect precisely in the first weeks of life. At the check at around 12 months of corrected age, the full spectrum of MRI characteristics in the leukomalacia may be observed. In the cases of diffuse, deep white matter also called 'subcortical' leukomalacia, mainly occurring in fullterm newborns, cystic areas are still present, after the first months of life, in the watershed areas under the cortex (Fig. 3). Lateral ventricles are usually very dilated, but rarely with the angular shape of the cases with periventricular leukomalacia. Cortical atrophy is generally severe, with widening of the subarachnoid spaces; sulci and fissure are prominent. Areas of hyperintensity are often observable in the periventricular white matter (see Fig. 3C), as a sign of the ischaemic involvement of these areas too, and sometimes of the basal ganglia. In our three cases, with no exception these MRI findings associated with spastic tetraplegia, mental retardation, severe cerebral visual impairment and epilepsy. In the cases of cystic periventricular leukomalacia, which represent the largest group in our series of MRI, we confirmed previous findings of the literature. In the

Fig. 5. A: coronal" US scan through the atria; small bilateral cystic lesions in the periventricular white matter. B: MRI of the same infant at 12 months: coronal proton density (left) and T2-weighted (right) images, high intensity of the periventricular white matter; cystic lesions are no longer observable.

most severe cases (Fig. 4), small cysts may remain observable after the first months; lateral ventricles are enlarged, mainly in the posterior horns, but also anteriorly. Their outlines are irregular; the white matter is generally reduced, with deep sulci; cortical atrophy is indicated by the enlargement of the subarachnoid spaces. In the periventricular regions, areas of high

14 intensity in the long T R images are observed. They correspond to gliotic reaction to the necrosis. The regions most frequently affected coincide with the pathological distribution of the periventricular leukomalacia according to pathological studies 1'6, i.e. in the areas adjacent to the trigones, to the bodies of the lateral ventricles, and to the anterior or posterior horns in some cases. Myelination is always severely delayed in the first year. This constellation of MRI abnormalities is always associated in our series with a severe motor, cognitive and sensory impairment. In the cases of less severe cystic periventricular leukomalacia (Fig. 5) cystic lesions were no longer observable at 12 months. MRI findings on the long T R sequences consisted mainly of the hyperintensity of the white matter in the periventricular regions, particularly at the trigones, and often in the centrum semiovale. Ventricles are mildly to moderately enlarged, at the atria and the posterior horns, which are angular and irregularly outlined. Periventricular white matter is generally reduced and myelination retarded. The corpus callosum is often thinner and atrophic. As reported in the previous section of the results, cystic periventricular leukomalacia is almost invariably associated with a bad neurological outcome, but the degree of motor, cognitive and visual impairment (from mild to severe) may vary quite considerably. In this respect MRI failed to provide predictive indexes for the severity of the functional deficit of these infants. Amplitude and even asymmetries in motor deficits of cerebral palsied patients sometimes did not strictly correlate with the results of MRI, and this held true also for visual and cognitive impairment. This lack of specificity of MRI in the less severe forms of leukomalacia is also indicated by the findings of the prolonged flare group at 12 months. Only a few MRIs were carried out, and their characteristics were very similar (Fig. 6). Morphology and size of the lateral ventricles were generally normal; no other signs of abnormality were observable, besides symmetrical hyperintensity of the signal in the periventricular white matter, generally surrounding the trigones, being sometimes more diffuse. Despite similarities in the MRI performed at 12 months, clinical features may differ quite dramatically, ranging from a mild motor retardation, normalized after the first months, to a spastic diplegia.

Electroencephalogram Our results confirm the high diagnostic and prognostic value of the EEG, When recorded in the early period after a cerebral insult. At least one EEG was recorded in the first 2 weeks

Fig. 6. A" MRI at 12 months of an infant with US findings of prolonged flare; axial proton density (left) and T2-weighted (right) images; high intensity of the periventricular white matter at the trigones; normal ventricular size and outlines. B: MRI at 11 months of another infant Mth prolonged flare; axial proton density (left) and T2-weighted (right) images; high intensity of the periventricular white matter at the trigones and the frontal horns.

of life in 77 of the 95 infants with leukomalacia documented by US. EEG abnormalities were found in about half the cases. They consisted either of background abnormalities or of ictal ones (rarely electroclinical seizures, more often excess of sharp waves). EEG findings, were clearly associated with the severity of the brain lesion and with the prognosis. This is clearly shown in Fig. 7, where EEG findings are compared with clinical outcome in the three groups of patients with leukomalacia. EEG abnormalities 'occur very often in newborns with cystic leukomalacia. We generally found a delay in background EEG maturation ('dysmaturity') in these infants. Maturational features were visually detected by using maturational codes derived from Nolte and Haas 22, and often confirmed by computer analysis 5.

15 N o r m a l EEG

Normal case, PMA 2fl ws

A b n o r m a l EEG

c-LM

02:1" 4

pF4IVIt 3

__~ ,- ~

FPI.('.I .

I1

PF 30

.

.

-

.

-

~

.

~ ~--._

.

~-,-',v./-~r',~.~k..~

.

. ~

..-~"~-'v-:~

~

.-~.,"~...._./~.....,~.---~ ~

i

2o

lo -

o

to

20

3o

n u m b e r Of c a s e s

severe abnorm,

~

mihl

abnorm,

PVL case. PMA 28 wa

[253

normal

outcome C 4-O 2

Fig. 7. Correlation between results of neonatal EEG and of neurological outcome in 77 infants, grouped according to US findings (abbreviations as in Fig. 1).

The association of cystic leukomalacia at the US and dysmature EEG was always indicative of a bad prognosis in our cases (12 out of 17 patients with cystic leukomalacia submitted to an early EEG). On the contrary, Of the 5 infants with cystic leukomalacia and normal EEG, 4 had a normal development and one a mild developmental retardation. In the group of infants with prolonged flare and grade 3 intraventricular haemorrhage, 6 had an abnormal EEG. Their outcome was a severe neurological impairment in 3 cases, a mild motor abnormality in one and normal in the remaining 2 cases. In the same group a normal EEG met with a normal development in 4 cases and with a cerebral palsy in one. An EEG was recorded in 49 patients with prolonged flare. Twenty-seven had a normal tracing, and only one of them presented a mild developmental retardation whereas the others had a normal outcome. Of the 22 cases showing some EEG abnormalities, 4 had an unfavourable evolution, 3 a developmental retardation (mainly motor), and 15 a good outcome. Besides its prognostic value for the final outcome, early EEG recording provided-important clinical cues in the early phase of the leukomalacia, when US findings were still uncertain, as to whether cysts were going to evolve or not. A representative example is given in Fig. 8, where we report the EEG of two infants, both born at 27 weeks and recorded at 28 weeks of postmenstrual age. US scan showed in both cases a periventricular hyperdensity, which had already lasted for 7 days at the time of the EEG recording. In the following days the flare gradually disappeared in the first case, whose developmen-

~

~

~

'

~

k

.

.

~

~

02"T4

01 -T3 T 3 FPl

~

~s~

-

~

~

~ ' - ' - v ' ~ ~

,

/

.

~

~

.TU-L 2oo ~,v I aec

Fig. 8. EEG tracings of two preterm infants recorded at 28 weeks of postmenstrual age. In the EEG of the infant with prolonged flare and normal outcome (top tracing), rhythmic theta activities are observable in the temporal regions, which are absent in the infant who developed cystic periventricular leukomalacia and who had a severe outcome (bottom tracing).

tal outcome was normal, whereas a cystic periventricular leukomalacia and subsequently a spastic diplegia was diagnosed in the second infant. Despite the similar early US findings, the EEG tracing were quite different. The normal for that age theta activity predominant in the temporal regions (temporal sawtooths), evident in the first case (Fig. 8, top tracing), was completely absent in the other infant (Fig. 8, lower tracing). These differences were fully confirmed by computer analysis. By means of the procedure described in the methods, the EEGs can be submitted to Fast Fourier transform; power spectra for the different frequency bands can also be mapped for the different regions of the scalp. The results of this analysis applied to the tracings of the two infants are shown in Fig. 9, where the three columns represent the brain maps for low frequency (0.5-3 Hz) (left), theta (3.5-7.5 Hz) (centre) and rapid

16

F!g. 9. Brain electrical activity mappings of the tracing shown in Fig. 8. Power spectra, obtained by means of Fast Fourier transform, are expressed as perce m values. The infant with prolonged flare is reported in the upper strip; the infant with cystic periventricular leukomalacia in the middle strip and difference maps in the lower strip. In the columns, maps are shown for low frequency (0.5-3 Hz) band (left), theta (3.5-7.5 Hz) band (centre) and rapid activity (8-20 Hz) bands (right). Power values are represented by different colours (see scales).

activity (8-20 Hz) bands (right), in the infant with prolonged flare (upper strip), the one with periventricular leukomalacia (middle), and for the differences between the two cases (lower strip). Different power values are represented by means of different colours: in the first two strips low values are indicated as dark green, high values as red, passing through light green, yellow and light red. On the left side of the map power values are represented: it is important to stress that the lowest value in each frequency band is represented by dark green and the highest tone by red. That holds true whether the range is very wide or quite narrow. In the last stri p (difference maps) dark blue indicates a lower value of the selected frequency band in the second infant (in this case the pathological one), whereas dark red indicates a higher value. In the baby with cystic periventricular leukomalacia a lower incidence of low frequencies (and a higher incidence of rapid activities) is appreciable in frontal regions. But the most important phenomenon is the lower incidence oftheta band in temporal regions, due to the lack of the maturational pattern 'temporal sawtooth' in this infant. Early EEG abnormalities, especially of the background activity, may therefore indicate a high risk of

cystic evolution, with a subsequent bad outcome, in an early phase of still ambiguous periventricular hyperdensity at the US scan.

DISCUSSION

Our results confirm the importance of leukomalacia as a major cause of neurological impairment in the high-risk newborn. This pathological entity was thought to be rare, but recent studies carried out by means of US scan have indicated a much larger incidence than previously supposed. The percentage of at risk-newborns affected varies in the different series (from 5 to 2 6 ~ as reported by de Vries et al.9), depending on the population studied, the type of transducer used, the study protocol applied and the US definition of leukomalacia utilized. Moreover, anatomical reports have demonstrated 2s the presence of pathological features of periventricular leukomalacia in the majority of the infants with very low birth weight. US is the first-choice technique to demonstrate 'invivo' leukomalaeia; it correlates very we!! with MRI-

17 findings (refs. 2, 10, 18 and our data). Correlations with autopsy findings are generally good 8'32, despite some false negatives 3, which can largely be explained by technical reasons (position of the cysts in specific areas, difficult to explore via US, type of transducer...). Serial US allow to distinguish different subgroups of leukomalacia (cystic vs. non-cystic, periventricular and more diffuse, association between leukomalacia and intraventricular haemorrhage...). This grouping is very important for functional prognosis, which is considerably worse for the cystic leukomalacia, according to our data and to the literature, especially if cystic lesions are large and diffuse in the white matter. Infants with prolonged flare often present a transient motor retardation or other mild neurological abnormalities, but they generally normalize after the first months. That occurs more rarely in the cases of prolonged flare associated with grade 3 intraventricular haemorrhage, the combination of both conditions probably causing more severe parenchymal damage. Few infants with cystic periventricular leukomalacia, and the majority of those with prolonged flare, showed a motor, cognitive and visual development apparently normal in the age range of our follow-up (13 years). It will be of great interest to follow the further cognitive and linguistic development of these 'normalized' infants. Besides the crucial role of US in identifying leukomalacia in the first weeks of life, how could we increase the specificity of these techniques for the diagnosis and especially, the prognosis of leukomalacia? More information may be provided by the improvement of the US examination (new apparatus, stricter protocols...). In particular, systematic use of a 7.5 M H z transducer (which was not available in our unit until recently) permits the visualization of even small cysts, not detectable by means of the 5 M H z transducer. In the early phase of leukomalacia, MRI is as yet of little help, due to the danger of patient transportation and sedation and its scarce specificity for white matter noncystic lesions. New approaches to neurological assessment, by observing the spontaneous motility of preterm infants in the incubator, have shown themselves to be of very high prognostic value in infants with an uncertain US as prolonged flare ~5,26. This technique has proved to be reliable, cheap and easy to use in the neonatal intensive care unit. In this paper we have shown the importance of the EEG in the high-risk newborn, when recorded in the first days after birth or after, the brain insult.

Confirming other reports in the literature 7, the presence of a marked E E G abnormality, and Of a background dysmaturity in particular, may indicate those infants whose densities will subsequently evolve into cysts. Moreover, early EEG is a g o o d predictor of neurological outcome, with a limited number of false positives or negatives. Computer analysis of the EEG, via Fast Fourier transformation, allows us to produce power spectra of the different frequency bands, for sleep and wakefulness states, and for each electrode. A gross topographic representation of the electrical activity at the scalp is also obtainable, by computer interpolation. As pointed out in other publications 5"24, computer analysis of the EEG has many advantages. As shown in Fig. 9, it permits patient-to-patient and group-togroup comparison and statistical analysis. However, it has to be underlined that this technique always requires a full knowledge of neonatal EEG. With very few exceptions, it does not detect EEG features non-observable in the raw EEG, but it permits confirmation of the data already provided by the visual analysis of an experienced examiner. The reliability of US examinations, the need for newborn transportation and sedation, and also the limitations of MRI in the study of non-cystic lesions, restrict the use of MRI in leukomalacia to a limited number of neonates. After the first months of life and the closure of the fontanel, MRI, with its lack of biological danger, its sensitivity to the progress in myelination and to white matter abnormalities, becomes the first-choice technique. In our experience, it confirms and completes neonatal US findings. Moreover, it permits retrospective diagnosis of leukomalacia in cases in which US was not available or was of uncertain significance. MRI findings in our series of leukomalacia largely confirmed previous reports in the literature. High intensity areas in Ta-weighted images and the reduction of the white matter, mainly in the periventricular regions, were the most common findings. They represent the MRI correlates of gliosis, demyelination and the loss of white matter in the watershed zones, caused by the ischaemic processes 2. Their distribution in the periventricular regions coincides with the areas of leukomalacia shown by pathological studies. Anatomical lesions indicated by MRI can be compared with the motor, cognitive and visual behaviour of the infants. The most common neurological impairment that w e found consisted of a cerebral palsy (generally a spastic diplegia). Topographic distribution of the white matter

18 lesions at the MRI included mainly regions adjacent to the trigones, which are crossed by fibres of the corticospinal tracts and by those for the lower extremities in particular 34. In the case of more diffuse white matter damage, as in the fullterm multicystic leukomalacia, or in the most severe cases of cystic periventricular leukomalacia, the lesions extend to the centrum semiovale and the corona radiata. The clinical form of cerebral palsy more often found in these cases is of a spastic tetraplegia. These patients also show a severe widening of subarachnoid spaces and a ventricular dilation, revealed by the MRI, and a mental retardation. However, besides these cases, recent MRI studies of patients with periventricular leukomalacia 35 and also our personal experience, do not indicate a strict correlation between the severity of mental impairment, measured by IQ test, and MRI findings. Multifactorial genesis of cognitive retardation, which may involve genetic and environmental factors, and lesions of several motor, sensory and association areas, may account for these results. MRI often showed anatomical damages in regions of the brain relevant for visual functions, the visual cortex and the optic radiations in particular, which cross the periventricular white matter near the trigones. In fact, a high incidence of cerebral visual impairment in patients with leukomalacia is reported in the literature z7 and also confirmed by our data. As for the cognitive functions, careful comparison between the resuits of behavioural visual assessment and of specific MRI features failed to reveal significant correlations with the exception of very severe brain lesions, always associated with cortical blindness in our cases. However, cases of dissociation between severe anatomical lesions in the posterior regions of the brain and normal visual functions are also reported t3. In conclusion, the clinical features and outcome of a neonate or an infant with leukomalacia cannot be predicted through the results of the US or of the MRI alone, despite the evident correlations between specific neuroimaging features and some functional deficits. Progress in neuroimaging apparatus and a better knowledge of brain functioning and development may in the future solve some of the present enigmas of dissociation between functional and imaging results. At the present time, a comprehensive approach to an infant with leukomalacia, which includes neuroimaging examinations, but also serial motor, cognitive and perceptive assessment, and electrophysiological testing is highly recommended. This permits a more precise diagnosis and progno-

sis of these patients, while providing valuable data for understanding structure/function correlations in abnormal as well as normal development.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the contribution to this study of Prof. C. Cipolloni and the staff of the Neonatal Intensive Care Unit and of.Prof. P. Planner and the members of the Infant Section of the Institute of Child Neurology and Psychiatry. Dr. J. van Hof-van Duin supervised the visual assessment. We also thank P. Morse for correcting the English of the manuscript and S. Giancola for some of the EEG recordings. This study was supported by grants from the Italian Ministries of Health and of Scientific Research and the University.

REFERENCES 1 Banker, B.Q. and Larroche, J.L., Periventricular leukomalacia in infancy: a form ofneonatal anoxic encephalopathy., Arch. Neurol.~ 7 (1962) 386-410. 2 Barkovich, A.J. and Truwit, C.L., Brain damage from perinatal asphyxia: correlation of MR findings with gestational age, Am. J. Neuroradiol., 11 (1990) 1087-1096. 3 Carson, S.C., Hertzberg, B.S., Bowie, J.D. arid Burger, P.C., Value of sonography in the diagnosis of intraeranial hemorrhage and periventricular leukomalacia, Am. J. Radiol., 155 (1990) 595601. 4 Cioni, G., Biagioni, E., Bartalena, L. and Boldrini, A., Dynamic EEG recordings of the preterm infant: relationship to intracranial haemorrhage, Boll. Lega It. Epil., 66/67 (1989) 165-166. 5 Cioni, G., Biagioni, E. and Cipolloni, C., Brain before cognition: EEG maturation in preterm infants. In I. Kostovie, S. Knezevic, H. Wisniewski and G. Spillieh (Eds.), Neurodevelopmem; Aging and Cognition, Birkhauser, New York, 1992, pp. 75-99. 6 De Reuck, J., Chatta, A.S. and Richardson Jr., E.P., Pathogenesis and evolution ofperiventricular leukomalacia in infancy, Arch. NeuroL, 27 (1972) 229-236. 7 De Vries, L.S., Connell, J.A., Dubowitz, L.M.S., Oozeer, R.C., Dubowitz, V. and Pennock, J.M., Neurological, electrophysiological and MRI abnormalities in infants with extensive cystic leukomalacia, Neuropediatrics, 18 (1987) 61-66. 8 De Vries, L.S., Wigglesworth, J.S. and Dubowtiz, L.M.S., The natural evolution of periventricular leukomalacia during the neonatal period and infancy: correlation of imaging and postmortem findings, Earl)' ttum. Dev., 17 (1988) 205-219. 9 De Vries, L.S., Larroche, J.C. and Levene, M.I., Cerebral ischaemic lesions. In Fetal and Neonatal Neurology and Neurosurgery, M.I. Levene, M.J. Bennett and J. Punt (Eds.), Churchill Livingstone, Edinburgh, 1988, pp. 326-338. 10 De Vries, L.S., Dubowitz, L.M.S., Pennock, J.M. and Bydder, G.M., Extensive cystic leukomalacia: correlation of cranial ultrasound, magnetic resonance imaging and clinical findings in sequential studies, Clin. RadiaL, 40 (1989) 158-166. 11 De Vries, L.S., Eken, P. and Dubowitz, L.M.S.,.The spectrum

19

12

13

14

15

16

17

18

19 20

21

22

of leukomalacia using cranial ultrasound, Behav. Brain Res., 49 (1992) 1-6. Dubowitz, L.M.S. and Dubowitz, V., The neurological assessment of the prcterm and fullterm newborn infant, CDM No. 79, Heinemann, London, 1981. Dubowitz, L.M.S., Mushin, J., De Vries L. and Arden, G.B., Visual function in the newborn infant: is it cortically mediated? Lancet, 1 (1986) 1139-1141. Estrada, M., El Gammal, T. and Dyken, P.R., Periventricular low attenuations: a normal finding in computerized topographic scans of neonates? Arch. NeuroL, 37 (1980) 754-756. Ferrari, F., Cioni, G. and Prechtl, H.F.R., Qualitative changes of general movements in preterm infants with brain lesions, Earl)" Hum. Dev., 23 (1990) 193-231. Flodmark, O., Becker, L.E., ttarwood-Nash, D.C., Fitzhardinge, P.M., Fitz, C.R. and Chuang, S.H., Correlation between computed tomography and autopsy in premature and full-term neonates that have suffered pcrinatal asphyxia, Radiology, 137 (1980) 93-103. Flodmark, O., Roland, E.H., Hill, A. and Whitfield, M.F., Periventricular leukomalacia: radiologic diagnosis, Radiology, 162 (1987) 119-124.. Flodmark, O., Lupton, B., Li, D., Stimac, G.K., Roland, E.H., Hill, A., Whitfield, M.F. and Norman, M.G., MR imaging of periventricular leukomalacia in childhood, Am. J. NeuroradioL, 10 (1989) 111-118. Griffiths, R., The Ability of Babies, University London Press, London, 1954. Keeney, S.E., Adcockr E.W. and McArdle, C.B., Prospective observation of 100 high-risk neonates by high-field (1.5 Tesla) magnetic resonance imaging of the central nervous system. I. Intraventricular and extracerebral lesions, Pediatrics, 87 (1991) 421-430. Lombroso, C.T., Neonatal polygraphy in full-term and premature infants: a review of normal and abnormal findings, J. Clin. NeurophysioL, 2 (1985) 105-155. Mohn, G., Van Hof-van Duin, J., Fetter, W.P.F., de Groot, L. and Hage, M., Acuity assessment of non-verbal infants and children: clinical experience with the acuity card procedure, Dev. Med. Child NeuroL, 30 (1988) 232-244.

23 Nolte, R. and Haas, G., A polygraphic study ofbioelectrical brain maturation in preterm infants, Dev. Med. Child NeuroL, 20 (1978) 167-182. 24 Nuwer, M.R., Quantitative EEG: I. Technique and problems of frequency analysis and topographic mapping, J. Clin. NeurophysioL, 5 (1988) 1-43. 25 Prechtl, H.F.R., The Neurological Examhzation of tire Full-term Newborn brant (2nd revised and enlarged version) CDM No.63, Heinemann, London, 1977. 26 Prechtl, H.F.R., Qualitative changes of spontaneous movements in fetus and preterm infants are a marker of neurological dysfunction, Earl), Hum. Dev., 23 (1990) 151-158. 27 Scher, M.S., Dobson, V., Carpenter, N.A. and Guthrie, R.D., Visual and neurological outcome of infants with periventricular leukomalacia, Dev. Med. Child NeuroL, 31 (1989) 353-365. 28 Shuman, R.M. and Selednik, L.J., Periventricular leukomalacia: a one year autopsy study, Arch. NeuroL, 37 (1980) 231-235. 29 Stewart, A.L., Reynolds, E.O.R., Hope, P.L., Hamilton, P.A., Baudin, J., de L. Costello, A.M., Bradford, B.C and Wyatt, J.S., Probability of neurodevelopmental disorders estimated from ultrasound appearance of brains of very preterm infants, Dev. Med. Child NeuroL, 29 (1987) 3-11. 30 Teller, D.Y., McDonald, M.A., Preston, K., Sebris, S.L. and Dobson, V., Assessment of visual acuity in infants and children: the acuity card procedure, Dev. Med. Child NeuroL, 28 (1986) 779-789. 31 Touwen, B.C.L., Neurological Development hz hfancy, CDM No. 58, Heinemann, London, 1975. 32 Trounce,'J.Q., Fagan, D. and Levene, M.I., Intraventricular haemorrhage and periventricular leukomalacia: ultrasound and autopsy correlation, Arch. Dis. Child, 61 (1986) 1203-1027. 33 Uzgiris, I. and Hunt, J. McV., Assessmem hi hfancy, University of Illinois Press, Urbana, 1975. 34 Volpe, J.J.,~ Neurology of the Newborn, 2nd edn., W.B. Saunders, Philadelphia, 1987. 35 Yokochi, K., Aiba, K., Horie, b.l., Inukai, K., Fujimoto, S., Mariko, K. and Kazuo, K., Magnetic resonance imaging in children with spastic diplegia: correlation with the severity of their motor and mental abnormality, Dev. Med. ChildNeuroL, 33 (1991) 18-25.

Neuroimaging and functional outcome of neonatal leukomalacia.

Leukomalacia is a major cause of neurological impairment in the high-risk newborn. It can be identified during the early postnatal period by means of ...
7MB Sizes 0 Downloads 0 Views