Neuroimaging and neurodevelopmental outcome of preterm infants with a periventricular haemorrhagic infarction located in the temporal or frontal lobe

Authors

  • Aneta Soltirovska Salamon,

    1. Department of Neonatology, Wilhelmina Children's Hospital, University Medical Centre Utrecht, Utrecht, the Netherlands
    2. Division of Paediatrics, Department of Neonatology, University Medical Centre Ljubljana, Ljubljana, Slovenia
    Search for more papers by this author
  • Floris Groenendaal,

    1. Department of Neonatology, Wilhelmina Children's Hospital, University Medical Centre Utrecht, Utrecht, the Netherlands
    Search for more papers by this author
  • Ingrid C van Haastert,

    1. Department of Neonatology, Wilhelmina Children's Hospital, University Medical Centre Utrecht, Utrecht, the Netherlands
    Search for more papers by this author
  • Karin J Rademaker,

    1. Department of Neonatology, Wilhelmina Children's Hospital, University Medical Centre Utrecht, Utrecht, the Netherlands
    Search for more papers by this author
  • Manon J N L Benders,

    1. Department of Neonatology, Wilhelmina Children's Hospital, University Medical Centre Utrecht, Utrecht, the Netherlands
    Search for more papers by this author
  • Corine Koopman,

    1. Department of Neonatology, Wilhelmina Children's Hospital, University Medical Centre Utrecht, Utrecht, the Netherlands
    Search for more papers by this author
  • Linda S de Vries

    Corresponding author
    1. Department of Neonatology, Wilhelmina Children's Hospital, University Medical Centre Utrecht, Utrecht, the Netherlands
    • Correspondence to Linda S de Vries, Department of Neonatology, University Medical Centre Utrecht/Wilhelmina Children's Hospital, Lundlaan 6, 3584 EA Utrecht, the Netherlands.

      E-mail: l.s.devries@umcutrecht.nl

    Search for more papers by this author

Abstract

Aim

The aim of the study was to compare clinical and neuroimaging characteristics and neurodevelopmental outcome in preterm infants with a periventricular haemorrhagic infarction (PVHI) located in the temporal or frontal periventricular white matter.

Method

The study was a retrospective hospital-based study of preterm infants with a frontal PVHI (n=21; 11 males, 10 females; mean birthweight 1527g; mean gestational age 30.3wks) or temporal PVHI (n=13; five males, eight females; mean birthweight 1205g; mean gestational age 30.2wks) admitted to the neonatal intensive care unit between 1990 and 2012. The clinical course, results of neuroimaging studies, and neurodevelopmental outcomes of preterm infants with a gestational age less than 34 weeks with a confirmed PVHI on early cranial ultrasonography and/or magnetic resonance imaging were reviewed. For assessment of neurodevelopmental outcome we used the Griffiths Mental Development Scales, the Movement Assessment Battery for Children, the Gross Motor Function Classification System, the Wechsler Preschool and Primary Scale of Intelligence, the Child Behavior Checklist, and ophthalmological assessment. An unfavourable neurodevelopmental outcome was defined as moderately or severely atypical neurological examination during the last visit: presence of cerebral palsy, epilepsy, a hearing or visual impairment, and/or atypical cognitive development (Griffiths Mental Development Scales developmental quotient or Wechsler Preschool and Primary Scale of Intelligence <85).

Results

Unfavourable outcome was observed in 12 out of 13 children with a temporal PVHI compared with six out of 21 children with a frontal PVHI (p=0.002). Only one of the included infants with a PVHI in the temporal white matter developed cerebral palsy, which was due to a parietal PVHI in the contralateral hemisphere. Cognitive impairment was noted in seven infants with a frontal PVHI and five with a temporal PVHI. There were more infants with a temporal PVHI who developed visual impairment (n=5) or behavioural problems (n=7) compared with those with a frontal PVHI (visual impairment (n=2), behavioural problems (n=3).

Interpretation

PVHI located in the temporal or frontal lobe is almost invariably related to a typical motor outcome, but carries a risk of cognitive, behavioural, and visual problems, especially in infants with a PVHI located in the temporal lobe.

Abbreviations
CBCL

Child Behavior Checklist

cPVL

Cystic periventricular leukomalacia

DTI

Diffusion tensor imaging

GMH–IVH

Germinal matrix haemorrhage–intraventricular haemorrhage

GMDS

Griffiths Mental Development Scales

MABC

Movement Assessment Battery for Children

PHVD

Posthaemorrhagic ventricular dilatation

PLIC

Posterior limb of the internal capsule

PVHI

Periventricular haemorrhagic infarction

PWML

Punctate white matter lesions

TEA

Term-equivalent age

USCP

Unilateral spastic cerebral palsy

Periventricular haemorrhagic infarction (PVHI) diagnosed in preterm infants is a complication of a germinal matrix–intraventricular haemorrhage (GMH–IVH). Cystic periventricular leukomalacia (cPVL) and PVHI are the most important brain lesions in the preterm infant. Whereas the incidence of cPVL is decreasing, the incidence of PVHI has remained stable over the last decades.[1] Between 3% and 11% of preterm infants born before 32 weeks' gestation and 5% to 8% of preterm infants with an extremely low birthweight develop a PVHI.[2, 3] PVHI usually develops from 24 to 96 hours after birth and is often seen in the context of cardiorespiratory instability in preterm infants. PVHI is considered to be due to impaired drainage of the medullary veins in the periventricular white matter and may be associated with GMH–IVH and with ischaemia of the periventricular white matter. Maternal, neonatal, and obstetric risk factors have been identified.[4]

Neuroimaging is required to diagnose a PVHI as clinical symptoms may be subtle or may not be recognized. Cranial ultrasound remains the method of choice for diagnosing a PVHI at the bedside, and its accuracy and reliability have been shown in a number of studies.[5, 6] While cranial ultrasound is useful for screening and sequential imaging will allow the evolution of the lesion to be followed, magnetic resonance imaging (MRI) provides more detailed information, is superior in diagnosing additional white matter or cerebellar injury, and allows clinicians to give a more reliable prognosis.[7]

PVHI is often associated with adverse neurological outcomes such as death, cerebral palsy (CP), symptomatic epilepsy, developmental delay, and cognitive impairment. Mortality in preterm infants with a PVHI reported in the literature varies between 30% and 60%.[2, 4] Outcome studies of children with a PVHI, assessed at 2 to 3 years of age, have reported that 50% to 85% of these children develop CP and that cognition is impaired in 20% to 79%.[8, 9] The spatial relationship between the corticospinal tracts and the site of the PVHI is important for predicting motor outcome. Rademaker et al.[6] reported that development of a motor impairment was less common when the site of the lesion on cranial ultrasound was anterior to the trigone, whereas all children with a lesion posterior to the trigone either died or developed a motor impairment, most often unilateral spastic cerebral palsy (USCP).

While a PVHI is most often seen in the parietal white matter, it can sometimes be found in the temporal or frontal periventricular white matter.[10] As a frontal and parietal PVHI are more directly within the ultrasound field of view than a temporal PVHI, a temporal PVHI could be missed when performing only cranial ultrasound. The incidence in the literature varies between 2% and 4%.[10, 11] Outcome data for a PVHI located in the frontal or temporal white matter are still limited.

The aims of this study were to compare clinical and neuroimaging characteristics of preterm infants with a temporal PVHI with those with a frontal PVHI and relate these to their neurological outcome.

Method

Patients and clinical characteristics

In this retrospective observational study, we analysed preterm infants (gestational age <34wks) who had been admitted to the level 3 neonatal intensive care unit of the Wilhelmina Children's Hospital, Utrecht, the Netherlands, between January 1990 and June 2012. We searched the ultrasound and patient database for infants with a diagnosis of PVHI. All cranial ultrasound scans from these infants were reviewed, and infants with a temporal or frontal PVHI on a routine cranial ultrasound, which in most cases was confirmed by MRI, were eligible for the study. Exclusion criteria were ultrasound features of brain malformations, metabolic disorders, and chromosomal abnormalities.

We reviewed the medical records of the infants for obstetric, perinatal, and neonatal risk factors associated with PVHI. These included low birthweight or gestational age, pre-eclampsia, chorioamnionitis, low Apgar score at 1 minute and 5 minutes, low arterial umbilical cord pH, need for emergency Caesarean section, respiratory and circulatory failure, patent ductus arteriosus, seizures, and presence of thrombocytopenia and/or mutation in genes coding for coagulation factors, GMH–IVH grade I, II, or III (according to Papile et al.[12]), posthaemorrhagic ventricular dilatation (PVHD), and need for intervention for PVHD.

Neuroimaging

Cranial ultrasound was part of routine clinical care and was performed using an ATL-UM4 mechanical sector scanner (Philips Healthcare, Best, the Netherlands) and an Aplio XG scanner (Toshiba medical system, Zoetermeer, the Netherlands) during the last 10 years with a multifrequency transducer (5–8MHz) to ensure the best possible resolution. A first cranial ultrasound was always performed as part of the admission procedure; additional examinations were performed two or three times during the first week after birth, then once a week until discharge, and at 40 weeks postmenstrual age; in many of those with PHVD it was repeated several times during the first year. PVHI was diagnosed when a triangular lesion with the apex of the triangle pointing towards the ventricle and the base facing the cortex, or a globular echogenic lesion, often communicating with the lateral ventricle, was seen. Temporal or frontal PVHI was classified according to the Dudink criteria or in relation to the veins likely to be involved.[13] It was noted whether the PVHI was on the left or right hemisphere, and whether the lesion was unilateral or bilateral. Additionally, we characterized associated lesions as GMH–IVH grade I, II, or III,[12] and noted the presence of PHVD, using the criteria of Levene,[14] and focal (punctate) or diffuse white matter lesions. On follow-up cranial ultrasound, the evolution into a porencephalic cyst or multiple cysts, separate from the ventricle, was evaluated.

MRI was performed on a 1.5- or on a 3-Tesla whole-body scanner (Achieva, Philips Healthcare, Best, the Netherlands) using a sense head coil. MRI included sagittal and axial or coronal T1- and T2-weighted images. Diffusion tensor imaging (DTI) was added to the scanning protocol only in 2008. MRI was performed either as soon as the infant was stable enough to be transported to the MRI unit to confirm the PVHI, to look for associated lesions in the ipsilateral as well as the contralateral hemisphere and the cerebellum, and/or at term-equivalent age (TEA) to also assess the symmetry of myelination in the posterior limb of the internal capsule (PLIC) in order to predict development of USCP or bilateral spastic CP. The signal intensity in the white matter was scored on the T2-weighted image as normal or high, and as focal or more diffuse. Punctate white matter lesions (PWML) were seen as low signal intensity on T2-weighted and high signal intensity on T1-weighted images. The size of the lateral ventricles was defined as normal or dilated and the shape as being typical or angular. The size of the extracerebral fluid space was visually assessed as being normal or enlarged. Appearance of the PLIC was assessed on the images obtained at TEA or later, using T1-weighted images. Symmetrical high signal intensity of the PLIC was regarded as normal. In some infants, a repeat MRI was performed at 18 to 24 months of age or later in childhood to assess residual damage.

All cranial ultrasound and magnetic resonance images were reviewed by two authors with extensive experience of neonatal neuroimaging.

Neurodevelopmental outcome

Patients were seen in the follow-up clinic by a developmental paediatrician and a paediatric physiotherapist skilled and trained in neonatal follow-up. All surviving infants were seen at TEA as well as at 6, 15, 24, and 36 months corrected age. A subgroup of infants (n=10) was also assessed between 5 and 7 years of age. Favourable neurodevelopmental outcome was defined as typical or mildly atypical neurological examination during the last visit, absence of CP, epilepsy, or a hearing or visual impairment, and/or typical cognitive development (Griffiths Mental Development Scales [GMDS] developmental quotient [DQ] ≥85).[15] An unfavourable neurodevelopmental outcome was defined as moderately or severely atypical neurological examination during the last visit, presence of CP, epilepsy, or a hearing or visual impairment, and/or atypical cognitive development (GMDS-DQ <85).[15] In those with CP, gross motor functioning was scored using the Gross Motor Function Classification System.[16] From 9 months onwards, the GMDS was used to assess neurodevelopmental outcome. The DQs were calculated as close as possible to the corrected age of 3 years. The Movement Assessment Battery for Children (MABC) test was used for the assessment of motor impairment at 5 years of age, and the Wechsler Preschool and Primary Scale of Intelligence (WPPSI) was used for the assessment of general intellectual functioning, identification of cognitive delay, and learning difficulties, expressed as IQ.[17, 18] Data from a paediatric ophthalmologist assessment were retrospectively collected. In order to obtain information on the children's specific behavioural and emotional problems, the parents of the children completed the Child Behavior Checklist (CBCL), supplemented with teacher's reports if available.[19] Behaviour was assessed with the preschool age checklist (CBCL/1½–5) for children aged 18 months to 5 years and the school age version (CBCL/6–18) for children aged 6 to 18 years. A total score from all questions is derived and interpreted as typical, borderline, or clinical behaviour.

Parental consent to perform MRI was obtained. No permission is required from the medical ethics committee of our hospital for retrospective, anonymous data analysis.

Statistical analysis

Statistical analysis was performed with spss, version 20.0 for Windows (IBM, SPSS Statistics IBM Corp., Armonk, NY, USA) to identify differences in clinical characteristics, neuroimaging, and neurodevelopmental outcome between newborn infants with frontal and temporal PVHI. The Mann–Whitney U test for equality of means for continuous variables and the Fisher's exact test and likelihood ratio G2 test for categorical variables were used. The strength of bivariate association between type of PVHI and outcome was determined based on Fisher's exact test for bivariate analysis of categorical data. In addition, we calculated odds ratios (ORs) and 95% confidence intervals (CIs) to determine the difference between the neurodevelopmental outcome (unfavourable vs favourable) of children with a temporal and frontal PVHI using R-software (version 2.15.1; package epitools; http://cran.r-project.org/). Statistical significance was defined at p<0.05. Finally, test results for GMDS were compared with the theoretical mean of the normative sample using a one-sample t-test.

Results

A total of 5396 preterm infants (gestational age <34wks) were admitted to our neonatal intensive care unit between January 1990 and June 2012, and 213 (4%) were diagnosed as having a PVHI. Using routine cranial ultrasound and/or MRI for determination, it was found that 179 infants (84%) had a parietal PVHI, 13 (6%) had a temporal PVHI, and 21 (10%) had a frontal PVHI. Of the 13 infants with a parietal PVHI, one infant, who died, had a PVHI in both temporal lobes, one also had a smaller PVHI in the contralateral frontal lobe, and one had associated bilateral cPVL in the frontal lobes. These three infants were not included in the statistical analysis when comparing the two groups. Seventy-one (33%) infants died, of whom 70 had a parietal PVHI and one had a temporal PVHI. Clinical details of infants with frontal or temporal PVHI are presented in Table 1.

Table 1. Characteristics of infants with temporal and frontal venous infarction
CharacteristicTemporal periventricular haemorrhagic infarction (n=13)Frontal periventricular haemorrhagic infarction (n=21)
  1. a

    Percentiles according to the new Dutch perinatal growth charts.[28]

Demographic characteristics
Male (n)511
Female (n)810
Birthweight (g), mean (range)1205 (990–2450)1527 (910–2965)
Gestational age (wk), mean (range)30.25 (26.86–33.57)30.35 (28.00–34.43)
Small for gestational age
(<10th centile)a (n)42
Antenatal
Pre-eclampsia (n)12
Chorioamnionitis (n)55
Perinatal
Apgar score at 1min, mean (range)5 (2–10)7 (4–10)
Apgar score at 5min, mean (range)9 (6–10)8 (7–10)
Arterial umbilical cord pH, mean (range)7.17 (7.04–7.32)7.23 (7.04–7.41)
Emergency caesarean section (n)85
Respiratory problems
Infant respiratory distress syndrome (n)814
Pneumothorax (n)11
Mechanical ventilation (n)812
Circulatory failure
Inotropic drugs (n)812
Patent ductus arteriosus (n)44
Sepsis (n)45
Thrombocytopenia (<150 × 109/L)44
Neurological findings
Convulsions (n)74
Periventricular haemorrhagic infarction
Onset <96h post-partum (n)1120
Side of periventricular haemorrhagic infarction
Left (n)37
Right (n)811
Bilateral (n)23
No associated germinal matrix haemorrhage–intraventricular haemorrhage (n)31
Associated germinal matrix haemorrhage–intraventricular haemorrhage
Grade I (n)12
Grade II (n)817
Grade III (n)11
Posthaemorrhagic ventricular dilation
Subcutaneous reservoir (n)42
Ventricular–peritoneal drain (n)22

The mean gestational age of the infants with a frontal or temporal PVHI (30.2wks) was significantly higher than of those with a parietal PVHI (28.4wks; p<0.01).

One hundred and thirty-three survivors were at least 12 months corrected age when last seen. Fifty of the 102 infants with a parietal PVHI (50%) developed CP, compared with one out of 10 infants with a temporal PVHI. The single child who developed USCP suffered a second PVHI in the parietal white matter contralateral to the temporal PVHI and developed USCP contralateral to the parietal PVHI. None of the children with a frontal PVHI developed CP. Two children were blind as a result of retinopathy and both had suffered a parietal PVHI (Fig. 1).

Figure 1.

Flow chart of overall review of children with periventricular haemorrhagic infarction (PVHI) treated in our unit from 1999 to 2012. Only survivors older than 12 months corrected age were included in the analysis of features of neurodevelopmental outcome. Children with temporal PVHI and bilateral frontal cystic periventricular leukomalacia were excluded. CP, cerebral palsy; BSCP, bilateral spastic cerebral palsy; USCP, unilateral spastic cerebral palsy.

Neuroimaging

PVHI in the temporal periventricular white matter

Cranial ultrasound

Temporal PVHI was associated with GMH–IVH in 10 out of 13 infants. One infant had a grade I, eight had a grade II, and one had a grade III GMH–IVH. In eight infants there was a right-sided PVHI, and two infants had a bilateral temporal PVHI. Insertion of a reservoir because of PHVD was required in four of the infants, and two of them subsequently required insertion of a ventriculoperitoneal shunt. Sequential cranial ultrasound showed cystic evolution in 12 infants, being separate (n=9) or in communication with the lateral ventricle (n=3).

Magnetic resonance imaging

MRI was performed in the early neonatal period in 12 infants and at TEA in 11 infants. Three infants underwent MRI in infancy, and in two of these MRI was also performed during the neonatal period. Eight of them had additional lesions such as cerebellar haemorrhage and/or PWML, parietal, small subdural haemorrhage, small frontal lesion, or a cerebrosinovenous thrombosis. One infant had a temporal PVHI and bilateral frontal cPVL and one had an additional small frontal PVHI contralateral to the temporal PVHI. Cavitation at the site of the temporal PVHI was seen in seven children. Gliosis at the site of the PVHI was seen in the six infants who underwent MRI later in infancy or childhood. Asymmetrical myelination of the PLIC was noted in three infants at TEA, in two of them on the side of the PVHI. Only one of these three infants developed a USCP, but the lack of myelination was on the side of the parietal PVHI, contralateral to the temporal PVHI. The other two were assessed with the MABC at 5 years; one child was on and the other below the 5th centile. Five of the infants did undergo DTI, allowing us to assess the integrity of the optic radiation. An abnormal appearance was seen in all, being bilateral in the infant with a bilateral PVHI (Fig. 2).

Figure 2.

Preterm infant, gestational age 26wks 6d, birthweight 1100g. Cranial ultrasound on day 2 showed a bilateral germinal matrix haemorrhage–intraventricular haemorrhage and right-sided temporal periventricular haemorrhagic infarction: (a) coronal and (b) parasagittal view. (c) Follow-up cranial ultrasound showed cystic evolution. (d) Magnetic resonance imaging (MRI) on day 21 confirmed the cranial ultrasound findings without any associated lesions on axial T2-weighted sequence. (e) Follow-up MRI at term-equivalent age showed asymmetry of ventricles, residual hemosiderin, and small cavitation on axial T2-weighted sequence and (f) reduced volume of right optic radiation on diffusion tensor imaging. At the age of 16.5 months the child showed typical gross motor, fine motor, and cognitive development, with a Griffiths Mental Development Scales developmental quotient of 100, and strabismus.

PVHI in the frontal periventricular white matter

Cranial ultrasound

All but one of the 21 infants in the group with a frontal PVHI developed a fan-shaped hyperechogenic white matter lesion on cranial ultrasound during the first 96 hours, located on the left in seven infants, on the right in 11, and bilaterally in three. In one infant there was no associated GMH–IVH, two had a grade I GMH–IVH, 17 had a grade II GMH–IVH, and one had a grade III GMH-IVH. Two infants required surgical intervention for development of PHVD. A subcutaneous reservoir was inserted initially and placement of a ventriculoperitoneal shunt was subsequently required in both infants. Sequential cranial ultrasound showed cystic evolution in 16 infants, being separate (n=10) or in communication with the lateral ventricle (n=6; Fig. 3).

Figure 3.

Preterm infant, gestational age 28wks 2d, birthweight 1330g. Cranial ultrasound on day 2 showed a frontal periventricular haemorrhagic infarction on the right side: (a) coronal and (b) parasagittal view. (c) Follow-up cranial ultrasound showed cystic evolution. (d) Magnetic resonance imaging (MRI) on day 24 confirmed the cranial ultrasound findings and additionally showed punctate white matter lesions and a large subarachnoid cerebrospinal fluid space on the axial T2-weighted sequence. (e) Follow-up MRI at term-equivalent age showed reduced volume of the right hemisphere, a large subarachnoid space, and a large interhemispheric fissure on the axial T2-weighted sequence. (f) Follow-up MRI (inversion recovery sequence) at the age of 33 months still showed the cystic lesion in right frontal lobe, and a large subarachnoid space. At the age of 3 years 9 months the child showed gross and fine motor developmental delay, cognitive developmental delay (Griffiths Mental Development Scales developmental quotient 82), macrocephaly, and had febrile seizures.

Magnetic resonance imaging

In 14 of the 21 infants, MRI was carried out at TEA; 10 of these infants also underwent MRI at around 30 week's postmenstrual age, and three underwent a third MRI in infancy. In two children MRI was performed only in infancy, and four infants had never undergone MRI. Besides confirming the cranial ultrasound diagnosis, additional findings were noted in three infants: a subdural haemorrhage (n=1), blood within the corpus callosum (n=1), and PWML (n=1). Cavitation at the site of the frontal PVHI was seen in nine children. Gliosis at the site of the PVHI was seen in the three infants in whom MRI was performed later in infancy or childhood. Asymmetrical myelination of the PLIC was not seen in any of the children.

Neurodevelopmental outcome

For analysis of features of neurodevelopmental outcome, only survivors who were 12 months corrected age or older were included. Two children who had a PVHI in both the frontal and the temporal white matter were not included in the statistical analysis. One of these two infants was less than 9 months of age at the time of the analysis. One infant with a bilateral temporal PVHI who died was also excluded (Fig. 1). There was evidence of a difference between the neurodevelopmental outcome (unfavourable vs favourable) of children with a temporal and frontal PVHI, p=0.002 (OR 22.5, 95% CI 2.3–218; Table 2). Neurodevelopmental outcome was not significantly associated with the side of the lesion (p=0.57; side right/bilateral: OR 3, 95% CI 0.2–34.5; side left/bilateral: OR 3.7, 95% CI 0.2–51.3), seizures in the neonatal period (p=0.43, OR 2.2, 95% CI 0.4–11.3), PHVD (p=1.0; right vs left, excluding bilateral lesions: OR 0.8, 95% CI 0.2–4.0), or with having additional associated lesions (p=0.11; OR 4.7, 95% CI 0.8–28).

Table 2. Developmental outcome of patients having a frontal or temporal periventricular haemorrhagic infarction in the neonatal period
LocationFavourable (n)Unfavourable (n)Odds ratio95% confidence interval p a
  1. For analysis of features of neurodevelopmental outcome, only survivors older than 12 months corrected age were included. aFisher's exact test.

Frontal lobe (n=21)15622.52.3–2180.002
Temporal lobe (n=10)19

PVHI in the temporal periventricular white matter

One infant with a bilateral PVHI in the temporal lobes died from respiratory and circulatory failure that was unresponsive to treatment. Although one of our patients with a temporal PVHI was still less than 12 months corrected age when last seen and one with a PVHI in the temporal lobe as well as bilateral frontal cPVL was excluded from statistical analysis, only one child in the temporal PVHI group had a favourable outcome based on neurological examination during the last visit and DQ within the normal range as determined by the GMDS. One child with a left-sided temporal PVHI, and a contralateral parietal PVHI, developed a left-sided USCP, global developmental delay, and postneonatal epilepsy. Functional limitation in this child was mild (Gross Motor Function Classification System level I).

Early developmental assessment was performed using the GMDS in 10 children, at a median age of 36 months (18–44mo). The mean DQ was 91, and a DQ less than 85 was observed in four children with a temporal PVHI. Three children were considered to have motor developmental delay as defined by the locomotor subscale of the GMDS.

Later assessment included the MABC test, which was performed in four children, two of whom scored below the 5th centile. The WPPSI was performed in two children, and both showed a decline compared with their early developmental score.

In six children the preschool version of the CBCL was used and in four the school-age version: seven children scored within the clinical range. No child developed hearing impairment, but visual impairment (strabismus or uncoordinated eye movements and/or myopia or hypermetropia, visual field defects) occurred in five of the children with a temporal PVHI. Three of these five infants underwent DTI, which in all cases was abnormal on the affected side. In the child with the left-sided USCP, the ipsilateral optic radiation was involved, but no DTI was performed. Her visual problems could also be due to the contralateral parietal lesion (Table SI, online supporting information).

PVHI in the frontal periventricular white matter

Fifteen out of 21 children with a frontal PVHI experienced a favourable neurodevelopmental outcome based on typical or mildly atypical neurological examination during the last visit and their DQ was within the typical range as determined by the GMDS. None of the children with a frontal PVHI developed CP, but two children exhibited a mild motor developmental delay, as defined by their score on the locomotor subscale of the GMDS.

Early developmental assessment was performed using the GMDS in all children, at a median age of 39 months (26–48mo). The mean DQ was 92, and a DQ less than 85 was observed in five children with a frontal PVHI.

Later assessment included the MABC test, which was performed in six children; four of them scored below the 5th centile. The WPPSI was performed in three children: two of them showed a decline compared with their early developmental score. In fifteen children the preschool version of the CBCL was used and in six the school-age version: three children scored within the clinical range. No child developed hearing impairment, but visual impairments occurred in two of the children with a frontal PVHI (one with strabismus and one with diplopia). No DTI was performed in these two children. Two children developed febrile seizures and no child developed postneonatal epilepsy (Table SII, online supporting information).

Discussion

In this study we were able to show that a PVHI located in the temporal or frontal lobe is almost invariably related to a typical motor outcome but is often associated with visual, cognitive, and behavioural problems. In the group of children with a temporal PVHI, only one child developed a left-sided USCP due to a PVHI located in the contralateral parietal white matter. In a study by Bassan et al.,[4] a more anterior localization was associated with an atypical neurological examination at 1 year of age. These results were not confirmed by Roze et al.,[20] who did not find a relationship between the site of the lesion and functional motor outcome. However, MRI was not used in their study, limiting the possibility of correcting for small associated lesions. In our study, none of the children with a frontal PVHI developed USCP, which is in agreement with the results of our previous study.[6] The spatial relationship between the corticospinal tracts and PVHI is of importance for the prediction of neurological outcome. Involvement of the caudate veins is usually associated with a PVHI anterior to the corticospinal tracts and does not result in USCP. This is not true for a more posterior infarction, when lesions can affect the corticospinal tracts. In one child with a left-sided temporal and a contralateral parietal PVHI, myelination of the PLIC on the side of the parietal PVHI was poor, reflecting interference with the development of the corticospinal tracts followed by USCP. In another two infants with a temporal PVHI, myelination of the PLIC showed mild asymmetry. These infants did not develop USCP, but both showed gross motor developmental delay with a poor score on the MABC test performed at 5 years. In contrast to the positive findings regarding motor outcome, we noted adverse cognitive development in seven (almost one-third) children with frontal and five of those with a temporal PVHI. Previous studies lack specific data on cognitive outcome in children with a frontal and temporal PVHI and report cognitive outcome for all children with a PVHI, irrespective of the site of the lesion.[9] Although development of cognitive dysfunction in both groups could also be due to associated lesions that were noted on neuroimaging studies, we consider this to be less likely, as these were found in only a small number of infants. Furthermore, the associated lesions, such as PWML or punctate cerebellar lesions, were limited in number, and the posterior fossa haemorrhage was small and had resolved by TEA.

In our study we noted that behavioural problems were very common and occurred in nine children (70%) with a temporal PVHI, compared with three (14%) in the frontal PVHI group, a significant difference. This is also higher than the rate of behavioural problems reported in preterm children without parenchymal lesions (13%–20%) and the rates reported in children with all types of PVHI, suggesting that the high rate of behavioural problems in those with a temporal PVHI is likely to be related to the site of the PVHI.[8, 21]

It was also of interest to note that half of the children with a temporal PVHI developed visual impairment, in most of them as a result of visual field abnormalities, which is probably due to involvement of Meyer′s fibres of the optic radiation. The optic radiation carries visual information from the lateral geniculate nucleus through the temporal and parietal lobes to the occipital lobe. The optic radiations form superior, intermediate, and inferior bundles as they emerge from the lateral geniculate nucleus. The superior and intermediate bundles take a direct route to the occipital lobe. The anterior inferior bundle, which carries information from homonymous inferior retinas, courses through the anterior aspect of the temporal lobe and bends around the temporal horn of the lateral ventricle within 4 or 5mm from the tip.[22] Our study is retrospective, with data collected over more than 20 years. We were able to use DTI, to better visualize the optic radiation at TEA, in only five of the infants with a temporal PVHI. This technique allowed us to assess the integrity of the optic radiation and atypical appearance was seen in all, being bilateral in the infant with a bilateral PVHI. We therefore recommend performing DTI in these infants, as this is likely to allow early identification of development of subsequent visual problems. Recent studies have demonstrated the use of DTI fibre tracking to delineate the optic radiation in preterm infants, to assess the microstructure of white matter tracts, and to measure changes within the optic radiation of preterm infants. The authors detected a significant link between the tissue architecture of the optic radiation and visual function in preterm infants.[23, 24]

In this study we focused on preterm infants with a temporal or frontal PVHI. Development of a PVHI in the temporal or frontal white matter was not common, as can be seen from the incidence of only 10% and 6% respectively. In our study, most of the infants developed a PVHI in association with a GMH–IVH with involvement of the caudate, inferior ventricular, and lateral atrial collector veins. PVHI can be recognized on cranial ultrasound by the presence of a typical triangular or globular-shaped lesion. The pathogenesis of a frontal or temporal PVHI is similar to that of a PVHI in the parietal white matter and also tends to be associated with a GMH–IVH. A haemorrhage of the temporal germinal matrix has been well documented.[11] The germinal matrix along the temporal horn traverses the lateral atrial vein on its course in the roof of the posterior part of the temporal horn. Temporal venous infarction above and lateral to the temporal horn is the result of occlusion of the inferior ventricular vein, which drains the superior and lateral borders of the temporal horn towards the basal vein of Rosenthal.[11]

The perinatal course of those with a temporal PVHI was more complicated than in those with a frontal PVHI. More infants exhibited intrauterine growth retardation, developed seizures, and had more associated lesions on their MRI (Table 1). These associated lesions were, however, small and limited in number. The lack of significance of these associated lesions with outcome may also have resulted from lack of power owing to the relatively small sample size. With statistical analyses we found that development of a favourable outcome was about 20 times higher in the group of infants with a frontal PVHI than in those with a temporal PVHI, and it is likely that this can be explained by the site of the lesion.

Brain injury can induce impairment of functional integrity which comprises both connections within the frontal and temporal lobe network as well as connections to distant areas, cerebellum, and basal ganglia, sometimes resulting in maladaptive behaviour, particularly when the injury occurs early in the development. In humans, neuronal misconnections could be involved in a variety of psychiatric disorders. Studies in individuals with epilepsy have concluded that behavioural disorders, including anxiety and mood disorders as well as personality disorders, are less severe in patients with frontal lobe epilepsy than in patients with temporal lobe epilepsy.[25-27] Being aware of these outcome data concerning a temporal PVHI makes the clinician aware of the increased risk of neurodevelopmental impairment in this subgroup of individuals with a PVHI, allowing early intervention.

There are several limitations which need to be addressed. As a frontal or temporal PVHI is less common than a PVHI in the parietal white matter, the number of infants is small, especially regarding those with a temporal PVHI. With regard to cognitive outcome, assessment at 36 months is still too early to explore the full effect of damage, especially in the frontal lobe. Not all children were seen by a paediatric ophthalmologist, only those with clinical symptoms, most often strabismus. As we did not have digitally stored cranial ultrasound images during the first part of the study period, we were not able to measure the size of the lesion, but were able to report only on the size as visually assessed and the site of the lesion, the latter being the most important predictor of outcome. It would be of interest to confirm our findings in a large set of data collected in different centres, and to add new MRI techniques, such as DTI of the optic radiation, to predict visual problems in those with a temporal PVHI.

Conclusion

In contrast to a PVHI in the parietal white matter, a PVHI located in the temporal or frontal lobe is almost invariably related to a typical motor outcome, but carries an increased risk of cognitive, behavioural, and visual problems. Infants with a PVHI in the temporal white matter were most at risk of cognitive and behavioural problems. Because of the common occurrence of visual impairment in infants with a temporal lobe PVHI, early referral of these infants to a paediatric ophthalmologist is recommended.

Ancillary