Presented at the Neonatal Society 2014 Autumn Meeting.
Rogers PS1, Salvan P1, Abaei M1, Edwards AD1,2,3, Counsell SJ1, Aljabar P1, Arichi T1,2,3
1 Centre for the Developing Brain, King’s College London, St Thomas’ Hospital, London, UK
2 Neonatal Intensive Care, St Thomas’ Hospital, Guys and St Thomas’ NHS Trust, London, UK
3 Department of Bioengineering, Imperial College London, South Kensington Campus, UK
Background: At rest, distinct and highly reproducible spatial patterns of correlated brain activity (known as “resting state networks”) can be readily identified with functional magnetic resonance imaging (fMRI). In the neonatal brain, resting state networks emerge during the third trimester in accordance with a period of rapid neural growth, such that a facsimile of adult network architecture can be identified at full term (Doria et al., 2010). Here we investigate the consistency of these networks across the cohort of infants studied at term equivalent age, and study the effects of specific clinical and demographic variables on their formation.
Methods: The study population consisted of 62 prematurely born infants who were studied at term equivalent age (38 to 42 weeks PMA, 29 male), all of whom were born prematurely and recruited as part of the NIHR-funded ePRIME study (REC ref: 09/H0707/98). fMRI data was acquired over 6.5 minutes on a 3-T Philips MRI system located on the neonatal unit at the Queen Charlotte and Chelsea Hospital, London. Infants with any evidence of overt brain abnormalities and/or focal brain pathology were excluded from the study group. The majority of infants (50/62) were sedated with chloral hydrate (30- 50mg/kg) prior to the MRI scan.
fMRI data analysis was done using FSL (www.fmrib.ox.ac.uk/fsl). A statistical data-driven analysis was done on the fMRI data from all 62 infants using probabilistic independent component analysis (P-ICA) to identify a full repertoire of 14 resting state networks from the group. The consistency of networks was then studied using the dual-regression approach by comparing the effects of key variables (whether they received sedation, sex, age at birth, and age at scan) on the spatial distribution of the empirically derived networks using a general linear model and permutation methods (Filippini et al. 2009). Correction for multiple comparisons was made on a voxel-wise level using the family-wise-error rate, and for overall significance using the Bonferroni correction.
Results: As described in the literature, group ICA yielded a total of 14 resting state networks, corresponding to the somatosensory, motor, visual, auditory and cognitive systems. The spatial representation of all networks was not found to be significantly different between any of the study subgroups. In particular, sedation with chloral hydrate did not have a significant effect (example shown in figure is the sensori-motor network in sedated (left) and un-sedated (right) infants).
Conclusion: At term equivalent age, infants have consistent and robust resting state networks, which are independent of the effects of sedation, sex, and age at birth. The results suggest that as in adults, resting state networks are reproducible and consistent in neonates at term-equivalent age, and are therefore amenable for systematic studies examining the effects of specific brain pathologies and therapies.
Corresponding author: email@example.com
Doria et al., 2010. Emergence of resting state networks in the preterm human brain. PNAS; 16: 20015-20.
Filippini et al. 2009. Distinct patterns of brain activity in young carriers of the APOE4 allele. PNAS; 106(17): 7209-14.