Differences in callosal and subcortical volumes and associated neurobehavioural deficits in children with prenatal alcohol exposure

Doctoral Thesis

2019

Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
License
Series
Abstract
Certain high-risk communities in the Western Cape Province of South Africa where heavy maternal prenatal alcohol consumption is perpetuated by historical and societal challenges, have some of the highest prevalence rates of fetal alcohol syndrome (FAS) in the world. FAS has lifelong behavioural and cognitive consequences. Neuroimaging research aims to link deficits in brain structure and function to behavioural outcomes. Manual tracing is considered the gold standard of neuroanatomical volumetric analysis. Combined with neurobehavioural testing it can provide links between structure and function, but is time consuming and labour intensive. Automated segmentation programmes, such as FreeSurfer, are a faster alternative. The challenge is creating automated programmes that can provide results that are comparable to manual tracing, especially in a clinical sample. The aims of this thesis were to investigate (1) the effects of prenatal alcohol exposure (PAE) on the sizes of the caudate nucleus, nucleus accumbens, hippocampus and corpus callosum (CC) and potential relations of regional volumes with IQ and verbal learning, (2) to compare the performance of manual and automated segmentation methods in identifying alcohol-related changes in brain morphometry, and (3) to examine the effects of PAE on inter-hemispheric transfer during adolescence and potential relations of CC size with inter-hemispheric transfer deficits. Participants for this project were recruited from the Cape Town Longitudinal Cohort for whom alcohol exposure data were gathered prospectively from the mothers during pregnancy using the timeline follow-back approach. Participants had been diagnosed previously by two expert dysmorphologists as either control, non-syndromal heavily exposed (HE), partial FAS (PFAS) or FAS. High-resolution T1-weighted images were acquired using a sequence optimized for morphometric neuroanatomical analysis on a Siemens 3T Allegra MRI scanner for 71 right-handed children (9 FAS, 19 PFAS, 24 HE and 19 non-exposed controls) from this cohort at ages 9-11 years. Bilateral caudate nuclei, nucleus accumbens and hippocampi and the CC were manually traced using Multitracer. FreeSurfer was used for automated segmentation. All structures were segmented with both FreeSurfer versions 5.1 and 6.0 to compare progress within development of automated segmentation algorithms. Associations of volumes from manual tracing with IQ and performance on the California Verbal Learning Test-Children’s Version (CVLT-C) were also examined. Inter-hemispheric transfer was assessed using a finger localization task (FLT) administered to 74 participants (12 FAS, 16 PFAS, 14 HE, and 32 controls) from the same cohort at ages 16-17 years. Of these, 34 participants had completed MRI at 9-11 years. Higher levels of PAE were associated with reductions in CC area, as well as bilateral volume reductions in caudate nuclei and hippocampi, effects that remained significant after controlling for alcohol-related reductions in TIV (total intracranial volume). Amongst dysmorphic children (FAS/PFAS), poorer performance on the CVLT-C was related to larger hippocampi and smaller CC. Smaller CC was also associated with lower IQ and partially mediated the effect of PAE on IQ. Manual and automated comparisons showed good agreement in the caudate nuclei, which are simpler to segment, moderate to good agreement in the smaller, more complex nucleus accumbens and hippocampi, and poor agreement in the CC. The latter is not surprising, however, in view of the fact that manual tracing measured the average area of the CC on a mid-sagittal slice, while FreeSurfer measures CC volume over a number of contiguous slices. After controlling for confounders and adjustment for smaller TIV, the latest FreeSurfer version 6.0 provided evidence of alcohol-related volumetric brain reductions comparable to manual segmentation. Only the most severely affected children with FAS demonstrated inter-hemispheric transfer deficits, with the number of transfer-related errors tending to increase with decreasing CC volume among children with PAE. This study confirms and extends evidence of PAE-related decreases in subcortical and CC size and that callosal volume partially mediates alcohol-related impairment in IQ. Although FreeSurfer v 6.0 achieves automated segmentations that are comparable to manual tracing, even in a paediatric clinical sample, performance is more reliable in some structures than others. Improvement and standardization of CC segmentation is especially important given the vulnerability of the CC and its critical role in domains affected by PAE, including verbal learning, IQ and inter-hemispheric transfer of information.
Description
Keywords

Reference:

Collections