Investigating altered brain development in infants with congenital heart disease using tensor-based morphometry

Isabel Ng, Alexandra F. Bonthrone, Christopher Kelly, Lucilio Cordero‐Grande, Emer Hughes, Anthony N. Price, Jana Hutter, Suresh Victor, Andreas Schuh, Daniel Rueckert, Joseph V. Hajnal, John Simpson, A. David Edwards, Mary Rutherford, Dafnis Batallé, Serena J. CounsellScientific 2020 · DOI

Abstract

Magnetic resonance (MR) imaging studies have demonstrated reduced global and regional brain volumes in infants with congenital heart disease (CHD). This study aimed to provide a more detailed evaluation of altered structural brain development in newborn infants with CHD compared to healthy controls using tensor-based morphometry (TBM). We compared brain development in 64 infants with CHD to 192 age- and sex-matched healthy controls. T2-weighted MR images obtained prior to surgery were analysed to compare voxel-wise differences in structure across the whole brain between groups. Cerebral oxygen delivery ($CDO_2$) was measured in infants with CHD (n = 49) using phase contrast MR imaging and the relationship between $CDO_2$ and voxel-wise brain structure was assessed using TBM. After correcting for global scaling differences, clusters of significant volume reduction in infants with CHD were demonstrated bilaterally within the basal ganglia, thalami, corpus callosum, occipital, temporal, parietal and frontal lobes, and right hippocampus ($p < 0.025$ after family-wise error correction). Clusters of significant volume expansion in infants with CHD were identified in cerebrospinal fluid spaces ($p < 0.025$). After correcting for global brain size, there was no significant association between voxel-wise brain structure and $CDO_2$. This study localizes abnormal brain development in infants with CHD, identifying areas of particular vulnerability.

Introduction

Congenital heart disease (CHD) is the most prevalent birth defect globally. While advancements in surgical techniques have significantly improved survival rates, many survivors face long-term neurodevelopmental challenges, including impairments in motor skills, executive function, and overall cognitive ability.

Previous neuroimaging research has established that infants with CHD often possess smaller total brain volumes. However, it remained unclear whether this reduction was a uniform global effect or if specific brain regions were disproportionately affected. Traditional volumetric analysis often relies on pre-defined regions of interest, which may overlook subtle, localized variations. This study addresses this gap by employing tensor-based morphometry (TBM) to map structural abnormalities at a voxel-wise level, identifying specific "vulnerability zones" before surgical intervention.

Methods and Datasets

The study cohort included 64 newborn infants with various forms of CHD, such as Transposition of the Great Arteries and Coarctation of the Aorta, recruited from St Thomas’ Hospital, London. These were compared against a control group of 192 healthy, age- and sex-matched infants (3:1 ratio) from the Developing Human Connectome Project (dHCP).

Table 1: Demographic and clinical characteristics of the study cohort, including specific cardiac diagnoses.

Researchers utilized Tensor-Based Morphometry (TBM) to analyze T2-weighted MR images. TBM calculates the mathematical "warping" or deformation required to map an individual infant's brain onto a standard age-matched template. By analyzing the "Jacobian determinant" of these deformations, the team could determine if specific voxels were expanded or contracted relative to the template. Crucially, global scaling was factored out to isolate structural differences that persist even after accounting for the overall smaller brain size in the CHD group. Additionally, phase-contrast MR imaging was used to measure cerebral blood flow and calculate cerebral oxygen delivery ($CDO_2$) in a subset of 49 infants.

Key Results

The TBM analysis revealed that volume loss in CHD is not uniform. Significant localized volume reductions ($p < 0.025$) were identified bilaterally in the thalami, basal ganglia (caudate nuclei and globus pallidi), corpus callosum, and the right hippocampus, as well as regions within the frontal, parietal, and occipital lobes.

Figure 1: Voxel-wise TBM results showing (a) regions of significant volume reduction (blue) and (b) regions of significant volume expansion (orange/yellow) in infants with CHD compared to controls.

Conversely, significant volume expansion was localized in the cerebrospinal fluid (CSF) spaces, particularly the third and fourth ventricles and the extra-axial spaces around the sylvian fissures. This expansion often serves as a marker for "missing" brain tissue (ex-vacuo expansion).

While lower $CDO_2$ was significantly associated with smaller total brain volume and cortical/deep grey matter volumes, it did not predict specific localized structural variations once global brain size was corrected. This suggests that while hypoxia drives overall growth restriction, other factors—such as genetic influences or the timing of cardiac development—may dictate the failure of specific regions to thrive.

Discussion and Future Directions

The localization of volume reductions in the deep grey matter and hippocampus is particularly noteworthy, as these areas are critical hubs for sensory processing, motor control, and memory. These findings likely explain the specific neurodevelopmental profiles observed in CHD survivors. The vulnerability of the thalami, which have high metabolic demands, suggests that "brain sparing" mechanisms in CHD are insufficient to protect these high-traffic regions from chronic substrate deficiencies.

A limitation of this study is its cross-sectional nature, capturing a snapshot prior to surgery. Future longitudinal research is needed to link these voxel-wise TBM maps to long-term neurodevelopmental outcomes at school age. Such models could eventually facilitate early identification and intervention for infants at the highest risk for cognitive and motor delays.

Summary updated

Abstract

Magnetic resonance (MR) imaging studies have demonstrated reduced global and regional brain volumes in infants with congenital heart disease (CHD). This study aimed to provide a more detailed evaluation of altered structural brain development in newborn infants with CHD compared to healthy controls using tensor-based morphometry (TBM). We compared brain development in 64 infants with CHD to 192 age- and sex-matched healthy controls. T2-weighted MR images obtained prior to surgery were analysed to compare voxel-wise differences in structure across the whole brain between groups. Cerebral oxygen delivery (CDO<sub>2</sub>) was measured in infants with CHD (n = 49) using phase contrast MR imaging and the relationship between CDO<sub>2</sub> and voxel-wise brain structure was assessed using TBM. After correcting for global scaling differences, clusters of significant volume reduction in infants with CHD were demonstrated bilaterally within the basal ganglia, thalami, corpus callosum, occipital, temporal, parietal and frontal lobes, and right hippocampus (p < 0.025 after family-wise error correction). Clusters of significant volume expansion in infants with CHD were identified in cerebrospinal fluid spaces (p < 0.025). After correcting for global brain size, there was no significant association between voxel-wise brain structure and CDO<sub>2</sub>. This study localizes abnormal brain development in infants with CHD, identifying areas of particular vulnerability.
BibTeX
@inproceedings{ng2020investigating,
  author = {Isabel Ng and Alexandra F. Bonthrone and Christopher Kelly and Lucilio Cordero‐Grande and Emer Hughes and Anthony N. Price and Jana Hutter and Suresh Victor and Andreas Schuh and Daniel Rueckert and Joseph V. Hajnal and John Simpson and A. David Edwards and Mary Rutherford and Dafnis Batallé and Serena J. Counsell},
  title = {{Investigating altered brain development in infants with congenital heart disease using tensor-based morphometry}},
  booktitle = {Scientific},
  year = {2020},
  doi = {10.1038/s41598-020-72009-3},
}
Figure