Article Text

Regional brain volumes, microstructure and neurodevelopment in moderate–late preterm children
  1. Claire E Kelly1,2,
  2. Deanne K Thompson1,2,3,4,
  3. Alicia J Spittle1,5,6,
  4. Jian Chen2,7,
  5. Marc L Seal2,3,
  6. Peter J Anderson1,8,
  7. Lex W Doyle1,3,6,9,
  8. Jeanie LY Cheong1,6,9
  1. 1 Victorian Infant Brain Study (VIBeS), Murdoch Children's Research Institute, Melbourne, Victoria, Australia
  2. 2 Developmental Imaging, Murdoch Children's Research Institute, Melbourne, Victoria, Australia
  3. 3 Department of Paediatrics, The University of Melbourne, Melbourne, Victoria, Australia
  4. 4 Florey Institute of Neuroscience and Mental Health, Melbourne, Victoria, Australia
  5. 5 Department of Physiotherapy, The University of Melbourne, Melbourne, Victoria, Australia
  6. 6 Newborn Research, The Royal Women’s Hospital, Melbourne, Victoria, Australia
  7. 7 Department of Medicine, Monash Medical Centre, Monash University, Melbourne, Victoria, Australia
  8. 8 Turner Institute for Brain and Mental Health, Monash University, Melbourne, Victoria, Australia
  9. 9 Department of Obstetrics and Gynaecology, The University of Melbourne, Melbourne, Victoria, Australia
  1. Correspondence to Claire E Kelly, Victorian Infant Brain Study (VIBeS), Murdoch Children's Research Institute, Parkville, VIC 3052, Australia; claire.kelly{at}mcri.edu.au

Abstract

Objective To explore whether regional brain volume and white matter microstructure at term-equivalent age (TEA) are associated with development at 2 years of age in children born moderate–late preterm (MLPT).

Study design A cohort of MLPT infants had brain MRI at approximately TEA (38–44 weeks’ postmenstrual age) and had a developmental assessment (Bayley Scales of Infant and Toddler Development and Infant Toddler Social Emotional Assessment) at 2 years’ corrected age. Relationships between cortical grey matter and white matter volumes and 2-year developmental outcomes were explored using voxel-based morphometry. Relationships between diffusion tensor measures of white matter microstructure (fractional anisotropy (FA) and axial (AD), radial (RD) and mean (MD) diffusivities) and 2-year developmental outcomes were explored using tract-based spatial statistics.

Results 189 MLPT children had data from at least one MRI modality (volumetric or diffusion) and data for at least one developmental domain. Larger cortical grey and white matter volumes in many brain regions, and higher FA and lower AD, RD and MD in several major white matter regions, were associated with better cognitive and language scores. There was little evidence that cortical grey matter and white matter volumes and white matter microstructure were associated with motor and behavioural outcomes.

Conclusions Regional cortical grey matter and white matter volumes and white matter microstructure are associated with cognitive and language development at 2 years of age in MLPT children. Thus, early alterations to brain volumes and microstructure may contribute to some of the developmental deficits described in MLPT children.

  • imaging
  • neonatology

Statistics from Altmetric.com

What is already known on this topic?

  • Preterm children, including moderate–late preterm (MLPT) children, have poorer neurodevelopmental outcomes than full-term children.

  • Global and regional brain volumes and white matter microstructure are altered in MLPT infants compared with full-term infants.

  • Relationships between brain structure and neurodevelopmental outcomes have been well documented in very preterm children, but less so in MLPT children.

What this study adds?

  • Brain volumes and microstructure at term age are associated with cognitive and language outcomes at 2 years of age in moderate–late preterm (MLPT) children.

  • Regional volumes and microstructure were explored, and specific regions that had stronger associations with different neurodevelopmental domains were identified.

  • Brain volumetric and microstructural alterations may contribute in part to the developmental deficits described in the MLPT population.

Introduction

Moderate–late preterm (MLPT) births (32–36 weeks’ gestational age (GA)), make up the majority of preterm births and are increasing in rate and absolute numbers.1 Most previous research on preterm birth has focused on very preterm (VPT,<32 weeks’ GA) birth. However, recent studies have found that not only VPT children, but also MLPT children, have poorer developmental outcomes than full-term children, including cognitive, language, motor and social-emotional outcomes.2–6 MRI can provide early knowledge of brain alterations that may underlie developmental delays.

There is increasing evidence that MLPT infants have brain alterations compared with full-term infants at approximately term-equivalent age (TEA, 38–44 weeks’ postmenstrual age).7 Our group and others have found that MLPT infants have smaller brain volumes, including smaller corpus callosum volume, larger cerebrospinal fluid volume, and altered diffusion tensor MRI values (lower fractional anisotropy (FA) and higher axial (AD), radial (RD) and mean (MD) diffusivities) in many white matter regions compared with full-term infants at TEA.8–10 These differences are similar, but smaller in magnitude, to the differences in brain volumes and microstructure seen between VPT and full-term infants at TEA.9

Very few studies have investigated whether the brain alterations documented in MLPT infants contribute to developmental problems. We previously reported that global brain tissue volumes at TEA were associated with 2-year neurodevelopmental outcomes in MLPT children; larger total brain tissue, white matter and cerebellar volumes were associated with better cognitive, language and motor outcomes.11 However, regional volumes within the cortical grey matter and white matter, and also white matter microstructure, may provide further insight into specific brain structure-function relationships in MLPT children. In VPT children, regional brain volumes and microstructure at TEA have been associated with developmental outcomes at 2 years of age.12–19 It is not known whether these brain structure-function relationships are present in MLPT infants, and they cannot be assumed to be so, because brain development is rapid during the perinatal period,20 and VPT and MLPT infants are exposed to different stressors during different stages of brain development.21 Therefore, the aim of the current study was to explore whether regional cortical grey matter and white matter volumes and white matter microstructure at TEA are associated with developmental outcomes at 2 years of age in MLPT children.

Methods

Participants

MLPT infants (n=201) were recruited between 2009 and 2012 from the Royal Women’s Hospital, Melbourne, into a prospective longitudinal cohort study.7 Infants with congenital abnormalities or genetic syndromes were excluded. Written informed consent was obtained from parents.

MRI

Infants had MRI at TEA at the Royal Children’s Hospital, Melbourne on a 3T Scanner (Siemens Magnetom Trio, Tim System). Infants were fed, were tightly swaddled, had earplugs inserted, had their ears covered with noise attenuators (MiniMuffs; Natus, San Carlos, California, USA), were placed in a MedVac (CFI Medical Solutions, Fenton, Michigan, USA) and were scanned during natural sleep without sedation. Infants were monitored with an apnoea monitor and oxygen saturation probe, and if required, oral sucrose was administered with parental consent. T2 -weighted images were acquired with turbo spin echo sequences (repetition time (TR) 8910 ms; echo time (TE) 152 ms; flip angle 120°; field of view (FOV) 192×192 mm; matrix size 192×192; 1 mm3 isotropic voxels; acquisition time ~6 min 30 s). Diffusion-weighted images were acquired with a multi b-value echo planar imaging (EPI) sequence (TR 20 400 ms; TE 120 ms; FOV 173×173 mm; matrix 144×144; 1.2 mm3 isotropic voxels; 45 non-collinear gradient directions; b-values ranging from 100 s/mm2 to 1200 s/mm2 in increments of 50 (23 different b-values in total, with 1–3 gradient directions per b-value); three b=0 s/mm2 volumes). All infants were scanned with the same diffusion sequence, including the same b-values. The total diffusion sequence was acquired as three separate acquisitions (17 directions requiring ~6 min, 16 directions requiring ~5 min 40 s, and 15 directions requiring ~5 min 30 s), which were concatenated prior to further processing. If the acquisitions had unacceptable levels of motion artefact, the scans were repeated whenever possible until acceptable images were acquired.

Structural image processing

T 2 images were bias corrected22 to improve skull stripping using the Brain Extraction Tool23 from the functional MRI of the brain (FMRIB) Software Library (FSL). Skull stripping was visually examined and manual edits were performed as required. Skull stripped images were segmented into tissue types using morphologically adaptive neonatal tissue segmentation (MANTiS).24 MANTiS generates segmentations in native space and in the standard space of a neonatal template.25 Standard space cortical grey matter and white matter images were analysed using voxel-based morphometry (VBM). VBM preprocessing involved combining all participants’ individual standard-space cortical grey and white matter images into 4D files, and smoothing them (full width at half maximum (FWHM)=2 mm).

Diffusion image processing

Diffusion images were corrected for motion and eddy current-induced distortions using FSL’s ‘eddy_correct’, including b-vector reorientation. We did not exclude individual diffusion directions, as this could introduce bias due to inhomogeneous directions between participants. Rather, we excluded participants if their diffusion acquisitions had severe motion artefact (n=24). We calculated motion parameters from the motion correction (absolute and relative displacement, rotation and translation), and these did not correlate with our outcome, neurodevelopmental scores at age 2 years. Susceptibility-induced distortion correction was performed based on the average gradient echo field map of n=10 participants’ field maps, because field maps were not acquired for all participants. The average field map was non-linearly registered to the b=0 s/mm2 image of each participant. Distortions were corrected using the aligned field map and FMRIB's utility for geometrically unwarping echo planar images (FUGUE). We visually compared images before and after susceptibility-induced distortion correction, and found correction reduced the distortions. The corrected images were brain extracted.23 The diffusion tensor model was fitted with the weighted linear least squares method using FSL, generating FA, AD, RD and MD images. Diffusion tensor images were analysed using tract-based spatial statistics (TBSS).26 Each participant’s FA image was aligned to every other participant’s FA image using FSL’s non-linear registration tool, each warp field was summarised by its mean displacement, and the most representative participant’s image was chosen as the image with the minimum mean distance to all other participants’ images. All FA images were aligned to the most representative image. The non-linear registrations were applied to the AD, RD and MD images. A mean FA image was created and thinned to generate a mean FA skeleton (threshold=0.2, chosen based on visual inspection of the skeleton, which showed it included most major white matter regions and excluded non-white matter regions). Every infant’s aligned FA, AD, RD and MD images were projected onto the mean FA skeleton.

Developmental assessments

Cognitive, language and motor development were assessed at 2 years of age (corrected for prematurity) using the Bayley Scales of Infant and Toddler Development (Bayley-III).27 The Bayley-III has age-standardised norms (mean=100; SD=15); higher scores indicate better outcome. Social-emotional and behavioural problems were assessed using a parent-report questionnaire developed for children aged 12–36 months, the Infant Toddler Social Emotional Assessment.28 Age-specific and gender-specific T-scores (mean=50, SD=10) were calculated for externalising behaviour problems, internalising behaviour problems, dysregulation and social-emotional competence. Higher scores for externalising behaviour, internalising behaviour and dysregulation indicate more behavioural problems, while lower scores for social-emotional competence indicate more problems.

Statistical analysis

Voxel-wise statistical analysis of brain volumes and diffusion tensor measures was performed using FSL’s Randomise, a non-parametric permutation-based method, which allows statistical modelling of neuroimaging data using the general linear model.29 Associations between MRI measures and 2-year developmental outcomes were investigated, adjusted for postmenstrual age at MRI and sex. Volume analyses were performed with and without adjusting for intracranial volume, to examine whether any specific regions were associated, over and above the overall head size. Results are reported at p<0.05 following 5000 permutations, threshold-free cluster enhancement and familywise error rate correction. Statistically significant regions were anatomically localised using a neonatal atlas.30

Results

Participants

Of the 201 MLPT infants recruited, 198 MLPT infants had MRI, of which we included 189 MLPT infants who had usable data for at least one MRI analysis (VBM or TBSS) and for at least one developmental domain (online supplementary figure 1). Of these 189 infants, 163 infants had both usable structural images for VBM and developmental outcome data, while 170 infants had usable diffusion images for TBSS and developmental outcome data. Exclusions were due to incomplete image acquisitions or image artefact (n=32 for VBM; n=24 for TBSS), or not attending or completing the developmental assessments (n=3 for VBM; n=4 for TBSS).

Table 1 shows characteristics of the included MLPT participants. We previously reported that the MLPT children had poorer cognitive, language, motor and social-emotional competence scores compared with full-term children at 2 years of age.2 Characteristics were similar between MLPT participants and MLPT non-participants (table 2).

Table 1

Characteristics of the moderate–late preterm participants who contributed data to the current study from at least one MRI analysis (voxel-based morphometry or tract-based spatial statistics) and at least one neurodevelopmental assessment (Bayley Scales of Infant and Toddler Development (Bayley-III) or Infant Toddler Social Emotional Assessment (ITSEA))

Table 2

Characteristics of the moderate–late preterm participants and non-participants

MRI and cognitive outcomes

Smaller volumes in the cortical grey and white matter regions shown in figure 1 and listed in table 3 (6% of the total cortical grey matter and 1% of the total white matter respectively) were significantly associated with poorer cognitive scores. When adjusting for intracranial volume, most of these regions were no longer significantly associated with cognitive scores, except regions within the right temporal grey matter and left frontal white matter (0.001% of the cortical grey matter and 0.02% of the white matter respectively; figure 1, table 3).

Figure 1

Regions where MRI measures (cortical grey matter and white matter volumes and white matter diffusion tensor values) were associated with cognitive scores. The number of voxels that were associated with cognitive scores and their percentage of the total cortical grey matter, white matter or mean fractional anisotropy skeleton are also shown above the images. Regions are overlaid on the same 10 axial brain slices. Red-yellow: positive associations; blue-light blue: negative associations. FWE, familywise error rate; ICV, intracranial volume; TFCE, threshold-free cluster enhancement.

Table 3

List of the brain regions at term-equivalent age that were associated with cognitive and language outcomes at 2 years of age in the moderate–late preterm children (at p<0.05, familywise error rate-corrected)

Lower FA and higher MD and RD in the white matter regions shown in figure 1 and listed in table 3 (4% of the skeleton for FA; 0.6% of the skeleton for MD; 4% of the skeleton for RD) were significantly associated with poorer cognitive scores. There were no significant associations between AD and cognitive scores.

MRI and language outcomes

Smaller volumes in the cortical grey and white matter regions shown in figure 2 and listed in table 3 (30% of the cortical grey matter and 13% of the white matter respectively) were significantly associated with poorer language scores. When adjusting for intracranial volume, many of these associations were no longer significant, although some remained (0.5% of the cortical grey matter and 0.6% of the white matter respectively; figure 2, table 3).

Figure 2

Regions where MRI measures (cortical grey matter and white matter volumes and white matter diffusion tensor values) were associated with language scores. The number of voxels that were associated with language scores and their percentage of the total cortical grey matter, white matter or mean fractional anisotropy skeleton are also shown above the images. Regions are overlaid on the same 10 axial brain slices. Red-yellow: positive associations; blue-light blue: negative associations. FWE, familywise error rate; ICV, intracranial volume; TFCE, threshold-free cluster enhancement.

Lower FA and higher MD, AD and RD in the white matter regions shown in figure 2 and listed in table 3 (27%, 12%, 4% and 22% of the skeleton, respectively) were associated with poorer language scores.

MRI and motor and social-emotional outcomes

There were no significant associations between any MRI measures and motor composite scores, or internalising behaviour, externalising behaviour, dysregulation or social-emotional competence scores.

Discussion

Brain volumes and microstructure were associated with cognitive and language outcomes, but not motor or behavioural outcomes, at age 2 years in MLPT children. These associations were in the direction that would generally be expected if size or microstructure of the brain was important to developmental outcomes, that is, lower cognitive and language scores were associated with lower volume and FA, and higher diffusivities,9 with no voxels where associations were in the opposite direction. This builds on our previous work11 in that we explored regional differences within brain tissues and identified specific regions that had stronger associations with different neurodevelopmental domains. This is one of the first studies of brain structure-function relationships in the MLPT population, and the results are important, suggesting that regional volume and microstructure alterations previously reported at TEA in MLPT infants compared with full-term infants8–10 may contribute to some of the cognitive and language delays described in MLPT children.

Our findings indicate that global brain volume is associated with cognitive and language outcomes in MLPT children. This is in line with our previous study,11 in which smaller total brain volumes at TEA were associated with poorer cognitive and language outcomes at age 2 years in MLPT children. Our current results after adjustment for intracranial volume could suggest that temporal grey matter and frontal white matter regions are particularly associated with cognitive outcome, and occipital grey matter and frontal white matter regions are particularly associated with language outcome, though the adjusted results were generally small in spatial extent. The association between occipital grey matter and language is difficult to explain, but may partly relate to the nature of many of the language items, in which the activities involved visual stimuli.27 In a previous study, late preterm children aged 6–13 years had smaller total brain tissue volume and larger cerebrospinal fluid volume, along with poorer cognitive outcomes compared with full-term children. This suggested smaller total brain volumes may underlie cognitive problems in older late preterm children,31 which is in line with our early childhood findings. In the VPT population, volume across much of the brain at TEA has been related to cognitive outcomes at age 2 years,12 32 which is similar to our findings in MLPT children.

Additionally, microstructure (lower FA and higher MD, AD and RD) in several white matter regions was also associated with poorer cognitive and language outcomes in our MLPT sample. It is difficult to determine the cellular basis of diffusion tensor measures due to their limited specificity.33 FA increases and AD, RD and MD decrease in the white matter throughout typical neonatal development,20 and MLPT infants have been shown to have lower FA and higher AD, RD and MD in many regions compared with full-term infants.8 9 This suggests lower FA and higher AD, RD and MD reflect atypical white matter microstructural development in MLPT infants, and these atypical patterns are associated with poorer cognition and language at age 2 years. There are many processes occurring during development including increasing myelination, axon density and axon diameter, and decreasing membrane permeability, water content and extracellular space.34 Changes in diffusion tensor measures in MLPT infants could reflect changes in any of these microstructural properties, due to insults to the white matter during the perinatal period (eg, ischaemia or inflammation), or delays in axonal or myelination development.34 35 In VPT children, microstructure of several regions including the corpus callosum, internal capsule, superior longitudinal fasciculus and arcuate fasciculus at TEA has been related to cognitive and/or language outcomes at age 2 years,18 19 36 which agrees with our findings, although we identified some other regions including the posterior thalamic radiation and corona radiata that were also associated with cognition and language in MLPT children.

There was little evidence that regional cortical grey matter and white matter volumes and white matter microstructure at TEA were associated with motor and behavioural outcomes in our MLPT sample. This lines up with our previous study in which smaller volume of only the cerebellum, and not total cortical grey or white matter, at TEA was associated with poorer motor scores at age 2 years in MLPT children.11 The VBM analysis in the current study included cerebral cortical regions only, hence future studies of subregions of the cerebellum would be beneficial. In contrast to our studies in the MLPT population, in VPT cohorts, studies have found that cerebellar volume at TEA is not associated with motor outcome,13 37 while cerebral volume and white matter microstructure at TEA is associated with motor outcome at age 1–2 years.14–19 38 In terms of behavioural outcomes, a previous study found smaller right temporal lobe volume was associated with anxiety in late preterm children during late childhood (age 6–11 years),39 which is in contrast to our findings in early childhood. Differential findings between our study and previous studies likely reflect the heterogeneity of the samples, including birth gestations and ages at follow-up, and different MRI analysis techniques.

Strengths of our study include the large sample, range of early developmental outcomes and early imaging. Limitations include that the sample was from a single tertiary centre, which may not generalise to other cohorts. Head motion is an important issue, but we attempted to minimise its effects by careful infant preparation and MRI acquisition, motion correction and excluding infants with severe motion artefact. Motion parameters were not correlated with developmental scores, suggesting motion was not driving our brain structure-function associations. VBM and TBSS have known limitations; specific to TBSS, the skeletonisation step limits reliability and interpretation of results.40 41

In conclusion, smaller regional brain volumes and altered microstructure at TEA were associated with poorer cognition and language at age 2 years in MLPT children. Hence, brain structural alterations may form the basis of some developmental deficits described in MLPT children. This study documented associations between brain structure and neurodevelopmental outcomes in MLPT children at the group level. The results provide an important basis for future studies into prognosticating the neurodevelopmental outcomes of individual infants born MLPT.

Acknowledgments

The authors thank the members of the Victorian Infant Brain Studies (VIBeS) group, Developmental Imaging group and Melbourne Children’s MRI Centre at the Murdoch Children’s Research Institute for their support. The authors thank the families who participated in the study.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Contributors JLYC, AS, LWD, PA, DT and MS were involved in the conception and design of the study, and the acquisition of the MRI data and clinical data. CEK, JC and DT were involved in the processing of the MRI images. CK performed the statistical analyses and drafted the manuscript. JLYC, AS, LWD, PA, DT, MS, JC and CK revised the manuscript critically for important intellectual content, gave final approval of the version to be published, and agree to be accountable for all aspects of the work.

  • Funding This work was supported in part by the Australian National Health and Medical Research Council (NHMRC) (Project Grant ID 1028822 and 1024516; Centre of Clinical Research Excellence Grant ID 546519; Centre of Research Excellence Grant ID 1060733; Senior Research Fellowship ID 1081288 to PA; Early Career Fellowship ID 1053787 to JLYC, ID 1053767 to AS, ID 1012236 to DT; Career Development Fellowship ID 1141354 to JLYC, ID 1108714 to AS, ID 1085754 to DT), Murdoch Children’s Research Institute Clinical Sciences Theme Grant, the Royal Children’s Hospital, the Department of Paediatrics at the University of Melbourne, the Victorian Government Operational Infrastructure Support Program, and The Royal Children’s Hospital Foundation.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval Human Research Ethics Committee of the Royal Women’s Hospital, Melbourne (approval number Project 09/38).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available from the corresponding author upon reasonable request.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Fantoms
    Ben J Stenson