Deformation-based morphometry identifies brain structural damages in 6 month-old infants with neonatal encephalopathy and predicts their developmental outcome
Hosung Kim1, Kevin Shapiro2, Maria Luisa Mandelli2, Hannah Clanley Glass2, Dawn Gano2, ELIZABETH Rogers3, Donna M Ferriero2, Anthony James Barkovich1, and Duan Xu1

1Radiology & Biomedical Imaging, University california San francisco, San Francisco, CA, United States, 2Neurology, University california San francisco, San Francisco, CA, United States, 3Pediatrics, University california San francisco, San Francisco, CA, United States

Synopsis

Neonatal encephalopathy (NE) is a major cause of mortality and permanent neurological disabilities in term infants. Using t1w MRI and DBM, we found that neonatal seizure was related to WM atrophy in multiple locations. Larger birth weight was associated with increased overall GM and WM volumes. A significant association was identified between language ability at 2 years old and increase in GM volume in Wernicke’s area. This DBM approach has the potential for predicting early developmental outcome in infants with NE, as the volume of Wernicke’s area significantly correlated with the scores of language ability evaluated in early childhood.

Background and Purpose

Neonatal encephalopathy (NE) is a major cause of mortality and permanent neurological disabilities in term infants. The prevalence of NE in the US remains at 0.2-0.4% of full-term births. Of affected newborns, up to 40% may die or develop severe and permanent developmental disorders, including those of cognition, visual-motor function or language, hyperactivity, cerebral palsy, and epilepsy 1. The underlying pathophysiology of NE and the related hypoxic-ischemic injury have been investigated using animal models even though differences between these models and human brain limit our understanding. T1-weighted MRI provides high-resolution biometrics related to brain morphological changes. Deformation-based morphometry (DBM) has been one of the best methodological frames for assessing gray matter (GM) and white matter (WM) tissue changes 2 as it does not rely on “error-prone” tissue classification but only on nonlinear registration. We hypothesize that the combined effects of injury and therapy modulate the development of WM myelination and GM maturation. A comprehensive assessment of these multivariate effects may enable a more precise prediction of neurodevelopmental outcome in patients with NE.

Methods

We studied 35 infants with NE (Demographics shown in Table 1). These infants underwent T1-weighted MRI at age 6 months using a 3D IR-SPGR sequence with 1 mm3 isotropic resolution. The images were processed to correct intensity non-uniformity and to normalize the intensity range. Using an established method, we constructed a template representing the whole population in our full-size database (n=60) 3. We then linearly and subsequently nonlinearly registered each infant MRI to this template, producing a deformation field map. For each individual MRI, voxel-wise Jacobian determinants were extracted from the deformation field to measure local volume. We used general linear models to assess changes in relation to a specific clinical variable while considering covariates of other variables. We included gestational age (GA) at birth, sex, and use of therapeutic hypothermia as the covariates while assessing the effect of electrographic seizures or the association with Apgar scores. We also correlated the regional volume with Bayley III scores of language, motor and cognitive function domains measured at 2.5 years old. The multiple comparison correction was performed using the random field theory 4.

Results

We found that neonatal seizures were related to white matter atrophy in multiple locations (Fig 1) as well as enlargement of the 3rd ventricle. Larger birth weight was associated with increased overall GM and WM volumes, primarily in the posterior cerebrum (Fig 2). No significant correlation was found with respect to Apgar scores. A significant association was identified between language ability at 2.5 years old and increase in the GM volume, mainly of supramarginal and posterior superior temporal cortices that comprise the Wernicke’s area (Fig 3). This finding was bilateral and the effects appeared stronger in the left hemisphere. We did not find correlations of brain volumes with scores of motor or cognitive function.

Conclusions

Using an established DBM approach, we found impairment in infant WM development associated with neonatal seizures, suggesting persistent disruption of brain pathways, which may be a precipitating factor in epilepsy or other brain network disorders. As much of the early cortical maturation in infancy occurs in posterior regions of the brain 5, the correlation between birth weight and posterior cortical GM and WM volumes at 6 months old found in the current study may indicate the presence of early posterior cortical damage and/or impedance of structural development related to low birth weight in this cortex. It will be interesting to see whether this DBM approach has the potential for the prediction of early developmental outcome in infants with NE, and to guide early therapeutic interventions, as the volume measurement of the Wernicke’s area significantly correlated with the scores of language ability evaluated in early childhood.

Acknowledgements

No acknowledgement found.

References

1. Gunn AJ et al., Curr Opin Pediatr. 2000; 2. Chung MK et al., Neuroimage 2001; 4. Worsley KJ et al., Neuroimage 2004; 5. Tzarouchi LC et al., Neuroimage 2009

Figures

Fig 1. Volume changes (decrease in blue; increase in red) in relation to neonatal seizure.

Fig 2. Positive correlation between regional volume and birth weight (corrected p<0.05)

Fig 3. Positive correlation between regional volume and score of language ability at 2.5 years old

Table 1. Demographics



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
1408