Downloaded from fn.bmj.com on July 10, 2014 - Published by group.bmj.com
Original article
The relationship between ventricular size at 1 month and outcome at 2 years in infants less than 30 weeks’ gestation Lisa M Fox,1,2 Pauline Choo,1,3 Sheryle R Rogerson,1,2 Alicia J Spittle,1,4 Peter J Anderson,4,5 Lex Doyle,1,2,4 Jeanie L Y Cheong1,2,4 1
Neonatal Services, Royal Women’s Hospital, Melbourne, Australia 2 Department of Obstetrics and Gynaecology, University of Melbourne, Melbourne, Australia 3 Department of Paediatrics, Hospital Tuanku Ja’afar Seremban, Negeri Sembilan, Malaysia 4 Murdoch Childrens Research Institute, Melbourne, Australia 5 Department of Paediatrics, University of Melbourne, Melbourne, Australia Correspondence to Dr Lisa M Fox, The Royal Women’s Hospital, Locked Bag 300, Parkville, VIC 3052, Australia; [email protected] Received 2 May 2013 Revised 29 November 2013 Accepted 7 December 2013 Published Online First 9 January 2014
ABSTRACT Background Cranial ultrasound cerebral biometric measurements have been used in preterm neonates, particularly in cases of ventriculomegaly. While cerebral biometric measures using MRI have been found to correlate with long-term outcome, the relationship between cranial ultrasound biometric measures and neurodevelopmental outcome has not been established. Objective To assess the relationship between ventricular size at 1 month of age using cranial ultrasound and neurodevelopmental outcome at 2 years in very preterm infants. Method Digital cranial ultrasound images taken between 25 and 35 days of age of 44 infants born at less than 30 weeks’ gestation were analysed independently by two observers. Infants with significant ultrasound abnormalities were excluded. A range of ultrasound linear measures were correlated with Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III) motor, language and cognitive composite scores at 2 years using linear regression. Results Larger lateral ventricular sizes (anterior horn width, ventricular height, midbody ventricular height) and larger ventricular-brain biparietal ratios were related to poorer motor composite score at 2 years. A ventricular-brain ratio of less than 0.3 was reassuring with regard to motor outcome. Poorer language composite scores at 2 years were associated with larger midbody ventricular heights. There was little evidence of a relationship with the cognitive composite score. Conclusions Larger lateral ventricles in the parietal region at a month of age were related to poorer motor development at 2 years. Larger ventricular measurements were also related to slower early language development. The role of cranial ultrasound biometric measures as biomarkers of later outcome in very preterm infants warrants further investigation.
INTRODUCTION
To cite: Fox LM, Choo P, Rogerson SR, et al. Arch Dis Child Fetal Neonatal Ed 2014;99:F209–F214.
Cranial ultrasound scanning (CUSS) is a useful bedside tool for imaging the neonatal brain. Despite limitations with field of view, grey-white matter differentiation and subtle parenchymal abnormalities, CUSS has an established role in detection of important brain pathology that relates to outcome in preterm infants such as intraventricular haemorrhages (IVH), periventricular haemorrhagic infarctions and cystic periventricular leukomalacia (PVL).1–3 It is well recognised that these abnormalities detected on CUSS do not reflect all forms of preterm brain injury and that a proportion of very
What is already known ▸ Diffuse brain injury that results in white matter loss and adversely affects neurodevelopmental outcome is difficult to detect on cranial ultrasound. ▸ Linear cerebral biometric measures of brain structures and volumes using MRI at term corrected age correlate with neurodevelopmental outcome in very preterm infants. ▸ Cranial ultrasound linear measures of the lateral ventricles at term corrected age correlate with MRI abnormalities.
What this study adds ▸ Ventricular size measured early in the postnatal course using linear ultrasound biometric measures correlate with 2-year neurodevelopmental outcome in very preterm infants. ▸ Larger lateral ventricles in the parietal region at 1 month of age are associated with poorer motor outcome at 2 years. ▸ These biometric measures have very good interobserver reliability.
preterm infants have diffuse PVL that is not readily detectable using CUSS. This white matter loss (WML) and subsequent decreased brain volume has been demonstrated using MRI at term equivalent age and been shown to be predictive of developmental delay and cerebral palsy.4 Linear measures of cerebral biometrics using MRI have been found to correlate with cognitive and motor development in very preterm children.5 Similar cerebral biometrics can be obtained using CUSS. It has been suggested that ventricular dilatation seen on CUSS can reflect WML related to preterm brain injury.6 Various authors have focused on CUSS assessment of WML, either by directly measuring cerebral structures or using increases in extra axial and ventricular spaces as a sign of reduction in brain size within the cranial vault. These studies have largely involved infants assessed at term corrected age.7–10 Woodward et al4 suggested
Fox LM, et al. Arch Dis Child Fetal Neonatal Ed 2014;99:F209–F214. doi:10.1136/archdischild-2013-304374
F209
Downloaded from fn.bmj.com on July 10, 2014 - Published by group.bmj.com
Original article that CUSS at 6 weeks postnatal age was inferior to MRI at term at predicting outcome; however, the CUSS was interpreted only in regard to major IVH or PVL. There are limited data describing early differences in ventricular size using CUSS biometrics and the relationship to neurodevelopmental outcome. We aimed to establish the interobserver reliability of CUSS biometrics measures at 1 month of age in a group of very preterm infants and to explore the relationship of these measures with corrected gestational age and 2-year neurodevelopmental outcome. We were particularly interested in whether these biometric measures related to outcome in very preterm infants without major IVH and PVL in whom neurodevelopmental outcomes in relation to CUSS findings are less well established.
METHODS Participants Very preterm infants cared for at the Royal Women’s Hospital, Melbourne, who were 0.70 was considered to demonstrate strong agreement between examiners. Linear regression was performed to establish the relationship between CUSS biometrics and corrected gestational age at time of scan. Linear regression was used to establish the relationship between CUSS biometrics and Bayley-III motor, language and cognitive scores,
Neurodevelopmental outcome The children were assessed at 2 years of age (adjusted for prematurity) using the Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III) by examiners blinded to the clinical and sonographic details.16 Ninety-three per cent of the assessments were performed by a single assessor; the F210
Figure 2 height.
Parasagittal view. (A) ventricular height, (B) midbody
Fox LM, et al. Arch Dis Child Fetal Neonatal Ed 2014;99:F209–F214. doi:10.1136/archdischild-2013-304374
Downloaded from fn.bmj.com on July 10, 2014 - Published by group.bmj.com
Original article Twelve infants (27%) were born at 23–25 weeks’ gestation, 19 (43%) at 26–27 weeks and 13 (30%) at 28–29 weeks. Seven infants had minor focal pathology that appeared to have no impact on ventricular size: one had a unilateral cerebellar hemisphere haemorrhage and six had small germinal matrix/ intraventricular haemorrhages (GM/IVH) without ventricular dilatation seen on imaging in the first week that had largely or completely resolved by the time of the study scan.
Interobserver reliability of ultrasound measures The ICCs for interobserver reliability between examiners are shown in table 2. Most measures correlated well between examiners. The ventricular index measures, particularly the right, had lower ICCs compared with the other measures and have therefore been excluded from further analysis. Given the good reliability between the examiners, the measures reported below are the mean of the measurements taken by the two examiners. Figure 3 Study population.
Size of the ventricles and the Bayley-III score first in a univariable analysis, and then adjusted for gestational age at birth, corrected age at scan and sex.
The measurements of the ventricles and the Bayley scores of the study cohort are given in table 3.
Relationship between CUSS measures and corrected gestational age at time of scan RESULTS Of the 120 infants enrolled in the randomised controlled trial, 68 had CUSS that fulfilled our study criteria. The other infants had scans outside of the predefined 10-day study period or had been transferred elsewhere before day 25. Two infants withdrew from the study at their parent’s request and two died before 2 years of age. Six infants with major cranial ultrasound abnormalities (large IVH with ventricular dilatation, parenchymal haemorrhagic infarction or PVL) were excluded. Digital images were not available for 14 infants; thus, 44 infants were included in the study (figure 3). The perinatal characteristics of the study sample are shown in table 1, along with the remainder of the cohort.
The ventricular transverse width and the biparietal diameter increased with increasing age at the time of the scan; the other measures did not (table 4).
Relationship between CUSS biometric measures and 2-year outcome Larger lateral ventricular sizes (anterior horn width, ventricular height, midbody ventricular height) were related to poorer motor development as assessed on the Bayley-III at 2 years (figure 4). There was a reduction in motor scores of a mean (SE) of 3.3 (1.2, right side) and 3.5 (1.3, left side), respectively, for every 1-mm increase in midbody height measurements at 28 days ( p value for both ≤0.01).
Table 1 Clinical characteristics
Gestational age in completed weeks (mean, SD, range) Birth weight in grams (mean, SD, range) Antenatal steroids Chorioamnionitis Female gender Multiple pregnancy Respiratory support* Patent ductus arteriosus Medical treatment Surgical ligation Clinical or culture proven sepsis Necrotising enterocolitis Chronic lung disease† Social risk index‡ (median, range)
USS study sample (n=44)
Remainder of cohort (n=76)
p Value
27 (1.5, 23–29) 877 (223, 432–1424) 37 (84) 8 (18) 27 (61) 12 (27) 43 (98) 28 (64) 23 (52) 0 (0) 21 (48) 0 (0) 16 (36) 1 (0–8)
28 1088 63 19 33 27 76 39 28 1 30 3 19 1
Original article
The relationship between ventricular size at 1 month and outcome at 2 years in infants less than 30 weeks’ gestation Lisa M Fox,1,2 Pauline Choo,1,3 Sheryle R Rogerson,1,2 Alicia J Spittle,1,4 Peter J Anderson,4,5 Lex Doyle,1,2,4 Jeanie L Y Cheong1,2,4 1
Neonatal Services, Royal Women’s Hospital, Melbourne, Australia 2 Department of Obstetrics and Gynaecology, University of Melbourne, Melbourne, Australia 3 Department of Paediatrics, Hospital Tuanku Ja’afar Seremban, Negeri Sembilan, Malaysia 4 Murdoch Childrens Research Institute, Melbourne, Australia 5 Department of Paediatrics, University of Melbourne, Melbourne, Australia Correspondence to Dr Lisa M Fox, The Royal Women’s Hospital, Locked Bag 300, Parkville, VIC 3052, Australia; [email protected] Received 2 May 2013 Revised 29 November 2013 Accepted 7 December 2013 Published Online First 9 January 2014
ABSTRACT Background Cranial ultrasound cerebral biometric measurements have been used in preterm neonates, particularly in cases of ventriculomegaly. While cerebral biometric measures using MRI have been found to correlate with long-term outcome, the relationship between cranial ultrasound biometric measures and neurodevelopmental outcome has not been established. Objective To assess the relationship between ventricular size at 1 month of age using cranial ultrasound and neurodevelopmental outcome at 2 years in very preterm infants. Method Digital cranial ultrasound images taken between 25 and 35 days of age of 44 infants born at less than 30 weeks’ gestation were analysed independently by two observers. Infants with significant ultrasound abnormalities were excluded. A range of ultrasound linear measures were correlated with Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III) motor, language and cognitive composite scores at 2 years using linear regression. Results Larger lateral ventricular sizes (anterior horn width, ventricular height, midbody ventricular height) and larger ventricular-brain biparietal ratios were related to poorer motor composite score at 2 years. A ventricular-brain ratio of less than 0.3 was reassuring with regard to motor outcome. Poorer language composite scores at 2 years were associated with larger midbody ventricular heights. There was little evidence of a relationship with the cognitive composite score. Conclusions Larger lateral ventricles in the parietal region at a month of age were related to poorer motor development at 2 years. Larger ventricular measurements were also related to slower early language development. The role of cranial ultrasound biometric measures as biomarkers of later outcome in very preterm infants warrants further investigation.
INTRODUCTION
To cite: Fox LM, Choo P, Rogerson SR, et al. Arch Dis Child Fetal Neonatal Ed 2014;99:F209–F214.
Cranial ultrasound scanning (CUSS) is a useful bedside tool for imaging the neonatal brain. Despite limitations with field of view, grey-white matter differentiation and subtle parenchymal abnormalities, CUSS has an established role in detection of important brain pathology that relates to outcome in preterm infants such as intraventricular haemorrhages (IVH), periventricular haemorrhagic infarctions and cystic periventricular leukomalacia (PVL).1–3 It is well recognised that these abnormalities detected on CUSS do not reflect all forms of preterm brain injury and that a proportion of very
What is already known ▸ Diffuse brain injury that results in white matter loss and adversely affects neurodevelopmental outcome is difficult to detect on cranial ultrasound. ▸ Linear cerebral biometric measures of brain structures and volumes using MRI at term corrected age correlate with neurodevelopmental outcome in very preterm infants. ▸ Cranial ultrasound linear measures of the lateral ventricles at term corrected age correlate with MRI abnormalities.
What this study adds ▸ Ventricular size measured early in the postnatal course using linear ultrasound biometric measures correlate with 2-year neurodevelopmental outcome in very preterm infants. ▸ Larger lateral ventricles in the parietal region at 1 month of age are associated with poorer motor outcome at 2 years. ▸ These biometric measures have very good interobserver reliability.
preterm infants have diffuse PVL that is not readily detectable using CUSS. This white matter loss (WML) and subsequent decreased brain volume has been demonstrated using MRI at term equivalent age and been shown to be predictive of developmental delay and cerebral palsy.4 Linear measures of cerebral biometrics using MRI have been found to correlate with cognitive and motor development in very preterm children.5 Similar cerebral biometrics can be obtained using CUSS. It has been suggested that ventricular dilatation seen on CUSS can reflect WML related to preterm brain injury.6 Various authors have focused on CUSS assessment of WML, either by directly measuring cerebral structures or using increases in extra axial and ventricular spaces as a sign of reduction in brain size within the cranial vault. These studies have largely involved infants assessed at term corrected age.7–10 Woodward et al4 suggested
Fox LM, et al. Arch Dis Child Fetal Neonatal Ed 2014;99:F209–F214. doi:10.1136/archdischild-2013-304374
F209
Downloaded from fn.bmj.com on July 10, 2014 - Published by group.bmj.com
Original article that CUSS at 6 weeks postnatal age was inferior to MRI at term at predicting outcome; however, the CUSS was interpreted only in regard to major IVH or PVL. There are limited data describing early differences in ventricular size using CUSS biometrics and the relationship to neurodevelopmental outcome. We aimed to establish the interobserver reliability of CUSS biometrics measures at 1 month of age in a group of very preterm infants and to explore the relationship of these measures with corrected gestational age and 2-year neurodevelopmental outcome. We were particularly interested in whether these biometric measures related to outcome in very preterm infants without major IVH and PVL in whom neurodevelopmental outcomes in relation to CUSS findings are less well established.
METHODS Participants Very preterm infants cared for at the Royal Women’s Hospital, Melbourne, who were 0.70 was considered to demonstrate strong agreement between examiners. Linear regression was performed to establish the relationship between CUSS biometrics and corrected gestational age at time of scan. Linear regression was used to establish the relationship between CUSS biometrics and Bayley-III motor, language and cognitive scores,
Neurodevelopmental outcome The children were assessed at 2 years of age (adjusted for prematurity) using the Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III) by examiners blinded to the clinical and sonographic details.16 Ninety-three per cent of the assessments were performed by a single assessor; the F210
Figure 2 height.
Parasagittal view. (A) ventricular height, (B) midbody
Fox LM, et al. Arch Dis Child Fetal Neonatal Ed 2014;99:F209–F214. doi:10.1136/archdischild-2013-304374
Downloaded from fn.bmj.com on July 10, 2014 - Published by group.bmj.com
Original article Twelve infants (27%) were born at 23–25 weeks’ gestation, 19 (43%) at 26–27 weeks and 13 (30%) at 28–29 weeks. Seven infants had minor focal pathology that appeared to have no impact on ventricular size: one had a unilateral cerebellar hemisphere haemorrhage and six had small germinal matrix/ intraventricular haemorrhages (GM/IVH) without ventricular dilatation seen on imaging in the first week that had largely or completely resolved by the time of the study scan.
Interobserver reliability of ultrasound measures The ICCs for interobserver reliability between examiners are shown in table 2. Most measures correlated well between examiners. The ventricular index measures, particularly the right, had lower ICCs compared with the other measures and have therefore been excluded from further analysis. Given the good reliability between the examiners, the measures reported below are the mean of the measurements taken by the two examiners. Figure 3 Study population.
Size of the ventricles and the Bayley-III score first in a univariable analysis, and then adjusted for gestational age at birth, corrected age at scan and sex.
The measurements of the ventricles and the Bayley scores of the study cohort are given in table 3.
Relationship between CUSS measures and corrected gestational age at time of scan RESULTS Of the 120 infants enrolled in the randomised controlled trial, 68 had CUSS that fulfilled our study criteria. The other infants had scans outside of the predefined 10-day study period or had been transferred elsewhere before day 25. Two infants withdrew from the study at their parent’s request and two died before 2 years of age. Six infants with major cranial ultrasound abnormalities (large IVH with ventricular dilatation, parenchymal haemorrhagic infarction or PVL) were excluded. Digital images were not available for 14 infants; thus, 44 infants were included in the study (figure 3). The perinatal characteristics of the study sample are shown in table 1, along with the remainder of the cohort.
The ventricular transverse width and the biparietal diameter increased with increasing age at the time of the scan; the other measures did not (table 4).
Relationship between CUSS biometric measures and 2-year outcome Larger lateral ventricular sizes (anterior horn width, ventricular height, midbody ventricular height) were related to poorer motor development as assessed on the Bayley-III at 2 years (figure 4). There was a reduction in motor scores of a mean (SE) of 3.3 (1.2, right side) and 3.5 (1.3, left side), respectively, for every 1-mm increase in midbody height measurements at 28 days ( p value for both ≤0.01).
Table 1 Clinical characteristics
Gestational age in completed weeks (mean, SD, range) Birth weight in grams (mean, SD, range) Antenatal steroids Chorioamnionitis Female gender Multiple pregnancy Respiratory support* Patent ductus arteriosus Medical treatment Surgical ligation Clinical or culture proven sepsis Necrotising enterocolitis Chronic lung disease† Social risk index‡ (median, range)
USS study sample (n=44)
Remainder of cohort (n=76)
p Value
27 (1.5, 23–29) 877 (223, 432–1424) 37 (84) 8 (18) 27 (61) 12 (27) 43 (98) 28 (64) 23 (52) 0 (0) 21 (48) 0 (0) 16 (36) 1 (0–8)
28 1088 63 19 33 27 76 39 28 1 30 3 19 1
