Magnetic Resonance and Cranial Ultrasound Characteristics of Periventricular White Matter Abnormalities in Newborn Infants

Clinical Radiology (2001) 56: 647±655 doi:10.1053/crad.2001.0754, available online at http://www.idealibrary.com on

Magnetic Resonance and Cranial Ultrasound Characteristics of Periventricular White Matter Abnormalities in Newborn Infants A N N E - M A R I E C H I L D S *, L U C CO R N E T T E *, L U CA A . R A M E N G H I *, S T E V E N F. TA N N E R{, RO S E MA RY J . A R T H U R{, DE L I A M A R T I N E Z{, M A L CO L M I . L E V E N E * *Division of Paediatrics and Child Health, University of Leeds, U.K., {Department of Medical Physics and Centre of Medical Imaging Research, University of Leeds, supported by the Special Trustees of The General In®rmary at Leeds, U.K. and {Department of Paediatric Radiology, The General In®rmary at Leeds, U.K. Received: 31 November 2000

Revised: 19 February 2001 Accepted: 23 February 2001

OBJECTIVE: To characterize the range of abnormalities within the periventricular white matter (PVWM) in a cohort of newborns using magnetic resonance (MR) brain imaging and to compare the focal MR abnormalities with the cranial ultrasound (CUS) ®ndings. METHODS: Retrospective study of MR brain and CUS ®ndings of infants born in the 18-month period 1998-1999. PVWM abnormalities were identi®ed by MR and focal lesions were characterized by size, number and distribution using a grading scale. Correspondence with CUS ®ndings was assessed. RESULTS: 175 MR examinations corresponding to n ˆ 105 preterm infants, (median GA 28, range 23±36 weeks) and n ˆ 25 term infants (median GA 39, range 37±42 weeks) were analysed for PVWM abnormalities. In the preterm group, MR demonstrated a normal PVWM in n ˆ 76, focal areas of altered signal intensity (SI) in PVWM in n ˆ 26 and venous infarction in n ˆ 3. In the term group, MR demonstrated a normal PVWM in n ˆ 15, focal areas of altered SI in PVWM in n ˆ 4, oedematous PVWM in n ˆ 2 and a middle cerebral artery infarction in n ˆ 4. All infants with normal MR had normal CUS ®ndings. A focal PVWM SI abnormality detectable on MR corresponded with an abnormality on CUS in only n ˆ 10/30. CONCLUSIONS: MR appears considerably more sensitive than CUS in demonstrating the existence and extent of focal PVWM lesions in newborn infants. Satisfactory correspondence between the two imaging investigations is obtained only for cystic PVWM lesions. Childs, A.-M. et al. (2001) Clinical # 2001 The Royal College of Radiologists Radiology 56, 647±655. Key words: brain, ischaemia, neonates, magnetic resonance imaging, cranial ultrasound.

The developing periventricular white matter (PVWM) of the preterm brain is particularly vulnerable to hypoxic± ischaemic insults such as ( perinatal) asphyxia, resulting in a wide spectrum of motor and cognitive impairments in surviving infants [1±4]. However, in the term infant, abnormalities due to ischaemic injury are characteristically observed in the basal ganglia and the cortex, leaving the PVWM intact [5,6]. The observation that di€erent regions of the brain are vulnerable at di€erent gestational ages (GA) has been mainly attributed to di€erences in maturity of the cerebral vasculature [7], and to the duration and severity of the insult [8]. More recently, new insights have been reported on the mechanisms of neonatal brain injury, such as the role of free radicals and glutamate [9], tumour necrosis factor-a [10], transforming growth factor-b 1 [11] Author for correspondence: Dr A.-M. Childs, Department of Paediatrics, D Floor, Clarendon Wing, The General In®rmary at Leeds, Belmont Grove, Leeds LS2 9NS, U.K. Tel: ‡44(0) 113 392 3901; Fax: ‡44(0) 113 392 3902; E-mail: [email protected] Guarantor of study: Professor M. I. Levene 0009-9260/01/080647+09 $35.00/0

and the vulnerability of oligodendroblasts [12]. Hence, any pattern of brain abnormality observed in neonates may be related more to the type of underlying pathological process rather than GA. Although routine neonatal cranial ultrasound (CUS) is able to detect abnormal PVWM, its sensitivity and speci®city in the prediction of neurodisability is less good than originally assumed [13,14]. Firstly, it is dicult to examine the brain completely, especially in term infants with a small anterior fontanelle [15,16]. Secondly, mild abnormalities can be dicult to identify with CUS since they may produce minimal sonographic changes. Thirdly, some discrete lesions may be transient and could be missed. On the other hand, ongoing optimization of magnetic resonance (MR) imaging methods enables a more detailed anatomical de®nition of the normal developing brain [17±19] and currently allows detection of subtle abnormalities. Hence, MR is likely to provide a more objective measure of cerebral pathology in the neonate. However, there has been little research on the correspondence # 2001 The Royal College of Radiologists

648

CLINICAL RADIOLOGY

Table 1 ± MR imaging grading of focal PVWM lesions Grade I (see Fig. 1a,b)

Grade II (see Fig. 1c,d)

Grade III (see Fig. 1e,f)

Grade IV (see Fig. 2a,b)

Number of lesions

1 or 2

53

53

Size of lesion

43 mm diameter

1 lesion 45 mm diameter

Multiple lesions 45 mm diameter

Grade I±III and cystic lesions

Distribution

42 PVWM areas a€ected

42 PVWM areas a€ected

PVWM extensively a€ected

between MR and CUS with respect to discrete focal PVWM abnormalities. The aim of the present study was to characterize the range of PVWM abnormalities observed on brain MR within a cohort of preterm and term infants admitted to a regional neonatal intensive/special care unit, and to compare the focal MR abnormalities with the CUS ®ndings.

METHODS

Subject Selection All infants admitted to the regional neonatal unit over an 18-month study period (October 1996±February 1998) were considered for inclusion. The study population comprised those infants whose parents gave informed consent. The study was approved by the hospital research ethics committee.

MR Acquisition All infants underwent the same imaging protocol on a 1.5 Tesla Gyroscan ACS-NT (Philips Medical Systems, The Netherlands), with a receive-only quadrature head coil. T1weighted images were acquired in the sagittal and axial planes and T2-weighted images in the axial plane. The parameters used for T1-weighted spin-echo images were repetition time (TR) 800 ms, echo time (TE) 13 ms, ®eld of view (FOV) 180 mm, slice thickness 4 mm with a 0.4 mm gap and a scan time of 3 min 52 s. The parameters used for the T2-weighted fast spin-echo images were TR 6000 ms, TE 200 ms, echo train length of 13, FOV 180 mm, slice thickness 3 mm with a 0.3 mm gap and a scan time of 5 min 12 s. Scans were performed after a feed and the majority of the infants were imaged without the use of sedation. If necessary, subjects were sedated with 30± 70 mg/kg chloral hydrate administered orally. Mechanical ventilation, continuous positive airway pressure, administration of intravenous ¯uids and inotropic drugs were continued if necessary. The head was immobilized by a pillow ®lled with polystyrene balls and moulded into shape by vacuum extraction to minimize movement artefact. The infants were continuously monitored with pulse oximetry and ECG recording and a trained paediatrician was present during the transfer of babies between the ward and the scanner, and throughout the MR study.

MR Analysis The MR images were analysed by two paediatricians (AMC and LAR), experienced in the interpretation of neonatal brain MR, who were unaware of the clinical history of the infants and the CUS ®ndings. Images were viewed on a Philips Easyvision stand-alone workstation, using proprietary software. Infants whose PVWM was obscured on MR by movement artefact or by severe post-haemorrhagic ventricular dilatation were excluded from further comparison with CUS. All focal PVWM lesions were graded on a scale of I to IV according to their size, number and distribution, as summarized in Table 1. Areas with an increased signal intensity (SI) on T1-weighted images and a decreased SI on T2-weighted images were de®ned as haemorrhagic lesions [5,20], while cystic lesions were de®ned as areas with SI identical to that of cerebrospinal ¯uid on all pulse sequences. Whenever characteristics of more than one grade were found, the highest grade was used. Other PVWM abnormalities assessed were venous infarction, PVWM oedema and middle cerebral artery infarction. The post mortem reports of infants who died during the study period were also assessed.

CUS Acquisition Serial bedside sonograms were performed according to a standard protocol on day 1, 3, 7, 10 and weekly thereafter until discharge or transfer to another hospital. The ultrasound images were acquired by a senior radiographer using a Toshiba Sonolayer SSH-140A machine, equipped with 5 and 7 Hz probes. All infants had a CUS assessment within 72 h of MR examination. Ultrasound was performed in coronal and parasagittal planes through the anterior fontanelle, obtaining sequential sections according to Levene et al. [21].

CUS Analysis Focal PVWM SI abnormalities observed on MR imaging were assessed for correspondence with CUS densities, ¯ares and cystic lesions. The sonograms temporally related most closely to the acquisition of the MR scan were assessed by one of two consultant radiologists (RJA, DM). If the image quality was felt to be unsatisfactory, the examination was repeated to ensure that an optimal CUS was compared to the MR image. Special attention was paid to the

649

MR AND CUS CHARACTERISTICS OF PERIVENTRICULAR WHITE MATTER ABNORMALITIES IN NEWBORNS

echogenicity of the PVWM, using the choroid plexus as an internal reference, and to focal abnormalities situated within the PVWM.

RESULTS

Study Population Over the 18-month period, 238 infants were admitted to the regional neonatal unit and 151 infants were enrolled in the study. Reasons for not performing MR in 87 infants were: lack of parental consent (n ˆ 50), and discharge (n ˆ 28) or death (n ˆ 9) before MR examination. Followup MR imaging was obtained in 24 of the 151 subjects. Subjects were classi®ed into two di€erent age groups: (1) n ˆ 122 preterm infants (GA 5 37 weeks) and (2) n ˆ 29 term infants (GA 5 37 weeks). The median GA of the preterm infants was 28 weeks, with a range of 23±36 weeks. Median age at the time of MR was 13 days in the preterm group, with a range of 1±42 days. PVWM anatomical details were obscured by movement artefacts in 8 subjects and by underlying abnormalities in 9 subjects. Hence, results are reported for 105 preterm infants (63 male, 42 female). Of the preterm infants, 29 of the 122 were from multiple pregnancies. The median GA of the term infants was 39 weeks, with a range of 37±42 weeks. Median age at the time of MR was 6 days in the term group, with a range of 2±13 days. Results are reported for 25 term infants since PVWM anatomical details were blurred by movement artefacts in 4 subjects.

MR Analysis in Preterm Infants (n ˆ 105) In 58 (55%) infants MR analysis was normal and in a further 18 (17%) infants the PVWM was unremarkable despite abnormalities elsewhere (see Table 2). A unilateral venous infarction was observed in 3 (3%) infants and the remaining 26 (25%) infants all revealed focal SI abnormalities. Of these 26 infants, 10 showed additional PVWM abnormalities, including 5 intraventricular haemorrhage (IVH), 3 germinal matrix haemorrhage (GMH), 1 agenesis of corpus callosum and 1 septo-optic dysplasia (Table 2).

MR Analysis in Term Infants (nˆ 25) Reasons for admission to the neonatal unit and MR ®ndings are summarized in Table 3. In 10 (40%) infants MR analysis was normal and in a further 5 (20%) infants the PVWM was unremarkable despite other brain abnormalities (Table 3). An abnormal PVWM was observed in 10 (40%) infants; oedematous aspect (n ˆ 2), middle cerebral artery infarction (n ˆ 4) and presence of focal abnormalities (n ˆ 4). These focal abnormalities had identical T1and T2 characteristics to those observed in the preterm infant group. Eleven of the 25 infants showed clinical evidence of HIE, but none had focal lesions in the PVWM on MR.

Table 2 ± MR results in preterm infants Normal PVWM (n)

Abnormal PVWM (n)

Additional MR abnormalities (number and nature)

58

0

0

13

0

8-GMH 5-IVH

5

0

2-holoprosencephaly 2-migration disorder 1-agenesis of CC

0

3 venous infarction

3-IVH

0

26 focal SI abnormalities [see Table 4]

3-GMH 5-IVH 1-agenesis of CC 1-SO dysplasia

Total 76

29 105

CC: corpus callosum; GMH: germinal matrix haemorrhage; IVH: intraventricular haemorrhage; PVWM: periventricular white matter; SI: signal intensity; SO dysplasia: septo-optic dysplasia.

Focal MR SI abnormalities in PVWM Focal MR SI abnormalities were observed in 30 of 130 (23%) subjects studied, corresponding to the premature brain in n ˆ 26 (86%) and the term brain in n ˆ 4 (14%). The median GA of this group of infants was 31 weeks, with a range of 28±41 weeks. MR imaging was performed at a median age of 15 days, with a range of 3±35 days. Infants of multiple pregnancies were overrepresented, i.e. 35% compared to a prevalence of 21% in the total study population. In addition, GMH or IVH was observed in 8 of the 30 (26%) infants, (all premature) compared to 13 of the 76 (17%) preterm infants with otherwise normal brain MR (Table 1). Grade I abnormalities were observed in 12 of the 30 infants, all from the preterm population. A unilateral SI change was identi®ed in 10 of those 12 infants, nine of whom had a single lesion con®ned to the optic radiation (Fig. 1a,b). Grade II abnormalities were seen in 7 preterm and 2 term infants (Fig. 1c,d), whereas Grade III abnormalities were noted in 3 preterm and 2 term infants (Fig. 1e,f). Grade II and III lesions were most commonly observed in the two hemispheres, although with an asymmetrical distribution. Only grade III abnormalities in term infants tended to be organized symmetrically. Grade IV abnormalities were identi®ed in 4 preterm infants. In two of these, the cysts were noted in close association with the focal SI abnormalities (Fig. 2a,b).

Comparison with CUS None of the 15 term infants with normal PVWM on MR revealed any abnormality on CUS. In the 76 preterm infants with normal PVWM on MR, none was reported to have any echodensity equal to that of the choroid plexus on US. However, 15 of these infants were reported to have

650

CLINICAL RADIOLOGY

Table 3 ± MR results in term infants Normal PVWM (n)

Abnormal PVWM (n)

Number and nature of additional MR abnormalities

Reason for admission

10

0

0

6 - suspected HIE 1 - cystic ®brosis 1 - hypopituitarism 1 - meningitis 1 - Chiari malformation

5

0

3 - BGSI 2 - CNS malformation

3 - suspected HIE 2 - antenatal diagnosis

0

2 PVWM generally oedematous

2 - BGSI + cortical highlighting

2 - suspected HIE

0

4 MCA infarction

3 - basal ganglia a€ected

3 - ®ts 1 - collapse

0

2 Grade II focal SI Abnormality [see Table 4]

0

1 - collapse ?cause 1 - diaphragmatic hernia

0

2 Grade III focal SI Abnormality [see Table 4]

0

1 - cardiac defect 1 - multifocal ®ts

0

0

0

Total 15

10

1 - TTN 3 - suspected HIE

25 HIE: hypoxic ischaemic encephalopathy; MCA: middle cerebral artery; PVWM: periventricular white matter; SI: signal intensity; TTN: transient tachypnoea of the newborn; BGSI: focal areas of altered signal intensity within the basal ganglia as reported in HIE [6].

some periventricular echogenicity that was greater than expected for their GA, which resolved in all 15 infants within two weeks. The correlation between the MR and CUS ®ndings was less optimal when assessing focal PVWM SI abnormalities on T1 and T2 MR images. CUS was normal in 11 of the 12 infants with a grade I MR abnormality. In the remaining infant CUS revealed a non-correspondent echolucency. In 4 out of 7 preterm infants with grade II and 1 out of 5 preterm infants with grade III MR abnormalities (Fig. 3a,b), CUS failed to detect any consistent PVWM abnormality. In 3 out of 4 infants with cystic change on MR, a cavitation was observed on CUS, although in one case the cysts were not noted until 10 days later despite regular ultrasound examination. In the four term infants, CUS revealed a persistently normal PVWM, while being abnormal on MR.

Neuropathological Correlation Two of the infants with focal SI abnormalities died during the study period and had post-mortem examinations. The brain of a preterm infant with a chromosome defect and grade IV abnormalities on MR showed multicystic periventricular leukomalacia (PVL) corresponding to the cystic change seen on MR and CUS. In addition, focal areas of PVWM gliosis were identi®ed at post mortem and found to correspond to the focal PVWM SI abnormalities noted on MR. The other infant with grade III abnormalities showed areas of focal SI abnormality in the PVWM on MR,

corresponding to focal areas of ischaemic damage, with micro-haemorrhage, at post-mortem examination. The CUS of this infant revealed non-speci®c echodensity and IVH. DISCUSSION

In 39 of 130 infants studied, MR imaging analysis demonstrated a wide spectrum of PVWM abnormalities, ranging from focal SI abnormalities to venous and middle cerebral artery infarctions. Any correspondence between MR and CUS was generally poor for focal abnormalities, except for cystic lesions. It has been suggested that CUS detected abnormalities represent only the `tip of the iceberg' of potential neurodevelopmental problems [22], as up to 50% of preterm babies who develop cerebral palsy have had normal CUS investigations in the neonatal period [13,14]. In comparative studies of PVL, CUS is known to detect fewer PVWM abnormalities than are found at post mortem, suggesting a number of lesions will remain undetected by CUS [23,24]. More recently, Sie et al. [25] investigated acute stage PVL in preterm infants using CUS and MR imaging and concluded that MR allowed for a wider di€erentiation of lesions as compared with CUS alone. The current study con®rms and extends these observations both to term infants and preterm infants with normal CUS. Only 6 of the 22 preterm infants (27%) and none of the term infants with grade I to III changes on MR had any echodensity revealed on CUS. It is noteworthy that when CUS failed to demonstrate cystic

MR AND CUS CHARACTERISTICS OF PERIVENTRICULAR WHITE MATTER ABNORMALITIES IN NEWBORNS

Fig. 1 ± Continued opposite.

651

652

CLINICAL RADIOLOGY

Fig. 1 ± (a) T1-weighted axial image, through Foramen of Monro (FOM), of an infant of 34 weeks GA at 5 days of postnatal age. A single high SI lesion is noted (white arrow) proximal to the posterior horn of the left lateral ventricle, closely related to the optic radiation. (b) T2-weighted axial image, through FOM (same infant as in Fig. 1a). The low SI lesion, noted in relation to the left optic radiation (black arrow), corresponds to the lesion in Fig. 1a. (c) T1-weighted axial image, through the body of the lateral ventricles, of infant of 28 weeks GA at 8 days of postnatal age. Several discrete areas of increased SI are seen bilaterally, adjacent to the lateral ventricles posteriorly (white arrowheads). (d) T2-weighted axial image, through the body of the lateral ventricles (same infant as in Fig. 1c). Discrete areas of low SI (black arrows) are seen posteriorly, corresponding to those noted in Fig. 1c. Resolving subependymal haemorrhage is noted in the wall of the lateral ventricles. (e) T1-weighted axial image, through the body of lateral ventricles, of a term infant at 5 days postnatal age. The infant presented with multifocal ®ts on day 2 and a normal CUS. Extensive high SI lesions are noted throughout the PVWM. The image demonstrates the lesions in frontal (straight arrows) and parieto-occipital regions (curved arrows). ( f) T2-weighted axial image, through the body of lateral ventricles (same infant as in Fig. 1e). Multiple areas of low SI lesions are seen throughout the PVWM. Frontal and parieto-occipital lesions, corresponding to those noted in Fig. 1g, are indicated by black arrows.

lesions observed on MR (n ˆ 1), the diameter of the cysts was less than 5 mm. A number of reasons for a suboptimal agreement between MR and CUS when dealing with focal PVWM lesions can be put forward. Firstly, one could argue that the CUS resolution might not allow the detection of the smallest lesions. However, a 7 MHz sector transducer was used with a lateral resolution of 1 mm. Secondly, it is clear that CUS remains a relatively subjective examination, but our group has had considerable experience with neonatal CUS and it is highly unlikely that the US ®ndings were misinterpreted. Thirdly, one could argue that, despite the CUS being done within 72 h of the MR, focal SI abnormalities were missed on CUS because of their transient nature. This seems unlikely since the infants who underwent repeat MR still had MR evidence of lesions up to three weeks later. The focal MR SI abnormalities, characterized by increased SI on T1-weighted images and reduced SI on

T2-weighted images, are consistent with discrete areas of micro-haemorrhagic change or possible dystrophic calci®cation as blood products are resorbed [5,20]. In one infant, whose brain was examined at post mortem, the grade III MR abnormalities corresponded to focal PVWM areas of ischaemia with microhaemorrhagic change. Similar lesions, usually subcortical and multiple in occurrence, have been noted as an incidental ®nding in MR studies of elderly patients deemed to be at high risk of a cerebrovascular accident [26] and have also been produced in animal models as a result of micro-emboli [27]. Micro-haemorrhagic changes are also known to be associated with PVL [28]. Histological studies have shown that PVL consists of two components, focal PVWM necrosis with loss of all cellular elements and more di€use glial cell damage [29,30]. It is therefore interesting that in one of the infants with focal abnormalities, follow-up MR demonstrated an interruption of the normal appearances of glial cell migration [17]. We therefore suggest that any focal PVWM lesion detected on

MR AND CUS CHARACTERISTICS OF PERIVENTRICULAR WHITE MATTER ABNORMALITIES IN NEWBORNS

653

Fig. 2 ± (a) T1-weighted axial image, through the body of the lateral ventricles, of infant of 29 weeks GA at 5 weeks postnatal age. Bilateral foci of high SI (small white arrows) with associated cystic degeneration on the left (large white arrow) are observed. (b) T2-weighted axial image, through the body of the lateral ventricles (same infant as in Fig. 2a). Bilateral foci of reduced SI (small white arrows) with cystic degeneration on the left (large white arrow) correspond to the abnormalities observed in Fig. 2a.

MR may be the endpoint of a number of di€erent perinatal insults. Discrete focal abnormalities (grade I±III) do not necessarily progress to severe necrosis or cystic degeneration (grade IV), detectable on CUS. Kuban et al. [31] have demonstrated that PVWM abnormalities appear more common in the presence of a GMH. Although our study comprised a slightly higher proportion of preterm infants with GMH and focal PVWM lesions, this was not signi®cant. In only one infant was the site of the PVWM abnormality related to underlying haemorrhage. Although CUS techniques do not demonstrate all MR detected focal PVWM abnormalities in this study or that of Sie et al. [25] and CUS in the neonatal period does not identify all infants at risk of neurodevelopmental impairment [13,14], it does not necessarily follow that brain MR will be more sensitive in identifying at risk infants. The clinical consequence of isolated focal PVWM abnormalities detected on MR is not yet known, but is likely to relate to both the distribution and the extent of the lesions. Infants with extensive PVWM abnormalities in association with hypoxic ischaemic encephalopathy have a poor outcome [32]. The more minor focal abnormalities may have no clinical consequences, although it will not be possible to

detect subtle neurodevelopmental abnormalities, known to be more common in preterm survivors [33±36], until the infants comprising our study population are much older and a more sensitive assessment of their sensorimotor and cognitive abilities can be made. A prospective study addressing this issue is currently in progress [37]. It is also of interest to note the high number of infants from multiple pregnancies found to have focal MR abnormalities in the current study. The increased risk of twins developing cerebral palsy and other neurodevelopmental disabilities has been well reported [38,39] and our future work also aims to compare the type, distribution and outcome of focal PVWM abnormalities in multiple births. It is dicult to obtain precise data on the incidence of minor focal MR abnormalities, since they can be missed for two reasons. Firstly, they may be transient in nature. Secondly, neonatal MR brain imaging cannot, for logistic and economic reasons, replace CUS as a routine investigation in all premature and term infants [4]. Nevertheless, a collaborative e€ort between the growing number of centres performing neonatal MR brain imaging could allow us to assess the clinical impact of focal PVWM lesions.

654

CLINICAL RADIOLOGY

Fig. 3 ± (a) T2-weighted axial image through body of the lateral ventricles of infant of 35 weeks GA at 4 days. Frontal and parieto-occipital foci of reduced SI (Grade III) are noted bilaterally (arrowed). (b) Cranial ultrasound in the coronal plane from same infant as Fig. 3b, acquired on day of MR. Only minimal echodensity is noted in corresponding parieto-occipital region.

Acknowledgements. The authors are much indebted to all sta€ of the neonatal unit and the MR Department for their assistance This work was supported by a grant from the special Trustees of the Leeds General In®rmary.

REFERENCES 1 de Vries LS, Dubowitz LMS, Dubovitz V, et al. Predictive value of cranial ultrasound in the newborn baby: a reappraisal. Lancet 1985;2:137±140. 2 Graziani LJ, Pasto M, Stanley C, et al. Neonatal neurosonographic correlates of cerebral palsy in preterm infants. Pediatrics 1986;78: 88±95. 3 Bennett FC, Silver G, Leung EJ, Mack LA. Periventricular echodensities detected by cranial ultrasonography: usefulness in predicting neurodevelopmental outcome in low birthweight preterm infants. Pediatrics 1990;85:400±404. 4 Van Wezel-Meijler G, van der Knaap MS, Oosting J, Sie LTL, deGrootL,HuismanJ,ValkJ,LafeberHN.?. Neuropediatrics1999;30: 1999;30:231±238. 5 Barkovich AJ, Truwit CL. Brain damage from perinatal asphyxia: correlation of MR ®ndings with gestational age. AJNR Am J Neuroradiol 1990;11:1087±1096. 6 Rutherford MA, Pennock JM, Schwieso JE, Cowan FM, Dubowitz LM. Hypoxic ischaemic encephalopathy: early magnetic resonance imaging ®ndings and their evolution. Neuropediatrics 1995;26: 183±191. 7 Volpe JJ, Herscovitch P, Perlman JM, Raichle ME. Positron emission tomography in the newborn: extensive impairment of regional blood ¯ow with intraventricular hemorrhage and hemorrhagic intracerebral involvement. Pediatrics 1983;72:589±595. 8 Martin E, Barkovich AJ. Magnetic resonance imaging in perinatal asphyxia. Arch Dis Childhood 1995;72:F62±70. 9 Levene MI. Birth asphyxia. In: David TJ, ed. Recent Advances in Paediatrics. Edinburgh: Churchill Livingstone, 1994;13:13±27.

10 Leviton A. Preterm cerebral palsy: is tumour necrosis factor-alpha the missing link?. Dev Med Child Neurol 1993;35:549±558. 11 Meng SZ, Takashima S. Expression of transforming growth factorbeta 1 in periventricular leucomalacia. J Child Neurol 1999;14: 377±381. 12 Jelinski SE, Yager JY, Juurlink BHJ. Preferential injury of oligodendroglioblasts by a short hypoxic±ischaemic insult. Brain Res 1999;815:150±153. 13 Pinto-Martin JA, Riolo S, Cnaan A, Holzman C, Susser MW, Paneth N. Cranial ultrasound prediction of disabling and nondisabling cerebral palsy at age two in a low birth weight population. Pediatrics 1995;95:249±254. 14 Holling EE, Leviton A. Characteristics of cranial ultrasound white matter echolucencies that predict disability:a review. Dev Med Child Neurol 1999;41:136±139. 15 Gray PH, Tudehope DI, Masel JP, et al. Perinatal hypoxic± ischaemic brain injury: prediction of outcome. Dev Med Child Neurol 1993;35:965±973. 16 Siegel MJ, Shackelford GD, Perlman JM, Fulling KH. Hypoxic ischaemic encephalopathy in term infants: Diagnosis and prognosis evaluated by ultrasound. Radiology 1984;152:395±399. 17 Childs AM, Ramenghi LA, Evans DJ, et al. MR features of developing periventricular white matter in preterm infants: evidence of glial cell migration. AJNR Am J Neuroradiol 1998;19:971±976. 18 Battin M, Maalouf EF, Counsell S, et al. Magnetic resonance imaging of the brain in very preterm infants: visualization of germinal matrix, early myelination and cortical folding. Pediatrics 1998;101:957±962. 19 Huppi P, Schuknecht B, Boesch C, et al. Structural and neurobehavioural delay in postnatal brain development of preterm infants. Pediatr Res 1996;39:895±901. 20 Keeney SE, Adcock EW, McArdle CB. Prospective observations of 100 high-risk neonates by high-®eld (1.5 Tesla) magnetic resonance imaging of the central nervous system: I. Intraventricular and extracerebral lesions. Pediatrics 1991;87:421±430.

MR AND CUS CHARACTERISTICS OF PERIVENTRICULAR WHITE MATTER ABNORMALITIES IN NEWBORNS

21 Levene MI, Williams LW, Fawer CL. Ultrasound of the infant brain. Clin Dev Med 1985;92:9±33. 22 Leviton A, Gilles FH. Venticulomegaly, delayed myelination, white matter hypoplasia, and `periventricular' leucomalacia: how are they related?. Pediatr Neurol 1996;15:127±136. 23 Hope PJ, Gould SJ, Howard S, Hamilton PA, Castello AM de L, Reynolds EOR. Precision of ultrasound diagnosis of pathologically veri®ed lesions in the brains of very preterm infants. Dev Med Child Neurol 1988;30:457±471. 24 Paneth N, Rudelli R, Monte W, et al. White matter necrosis in very low birth weight infants: neuropathological and ultrasonographic ®ndings in infants surviving six days or longer. J Pediatr 1990;116: 975±984. 25 Sie LTL, Van der Knaap MS, Van Wezel-Meijler G, Taets van Amerongen AHM, Lafeber HN, Valk J. Early MR features of hypoxic±ischemic brain injury in neonates with periventricular densities on sonograms. AJNR Am J Neuroradiol 2000;21:852±861. 26 Awad IA, Spetzler RF, Hodak JA, Awad CA, Carey R. Incidental subcortical lesions identi®ed on MR imaging in the elderly I. Correlation with age and cerebrovascular risk factors. Stroke 1986;17:1084±1089. 27 Roos MW, Ericsson A. MR imaging of experimental microinfarctions in the rabbit brain. Exp Neurol 1997;146:295±729. 28 Banker BQ, Larroche J-C. Periventricular leucomalacia of infancy: a form of neonatal anoxic encephalopathy. Arch Neurol 1962;7: 386±410. 29 Gilles FH, Leviton A, Dooling EC. The developing human brain: growth and epidemiological neuropathology. Boston: John Wright, 1983, 244±315. 30 Paneth N. Classifying brain damage in preterm infants. J Pediatr 1999;134:527±529.

655

31 Kuban K, Sanocka U, Leviton A, et al. White matter disorders of prematurity: Association with intraventricular hemorrhage and ventriculomegaly. J Pediatr 1999;134:539±546. 32 Rutherford MA, Pennock JM, Schwieso JE, Cowan FM, Dubowitz LM. Hypoxic ischaemic encephalopathy: early and late magnetic resonance imaging ®ndings in relation to outcome. Arch Dis Childhood 1996;75:F145±151. 33 Jongmans M, Mercuri E, de Vries L, Dubowitz L, Henderson SE. Minor neurological signs and perceptual-motor diculties in prematurely born children. Arch Dis Child 1997;76:F9±F14. 34 Levene MI, Dowling S, Graham M, Fogelman K, Galton M, Philips M. Impaired motor function (clumsiness) in ®ve year old children: correlation with neonatal ultrasound scans. Arch Dis Child 1992;67:687±690. 35 Veen S, Ens-Dokkum MH, Schreuder AM, Verloove- VanhorickSP, Brand R, Ruys JH. Impairments, disabilities and handicaps of very preterm and very low-birthweight infants at 5 years of age. Lancet 1991;338:33±36. 36 Maalouf EF, Duggan PJ, Rutherford MA, Edwards AD, et al. Magnetic resonance imaging of the brain in a cohort of extremely preterm infants. J Pediatrics 1999;135:351±357. 37 Cornette L, Toh V, Ramenghi L, et al. Characteristics of punctate lesions in neonatal brain magnetic resonance imaging. (Abstract). Child's Nervous System 2000;16(1):55. 38 Leonard CH, Piecuch RE, Ballard RA, Cooper BAB. Outcome of very low birth weight infants: multiple gestation versus singletons. Pediatrics 1994;93:611±615. 39 Williams K, Hennesy E, Alberman E. Cerebral palsy: e€ects of twinning, birthweight, and gestational age. Arch Dis Child 1996;75: F178±F182.