Recent studies have reported negative associations between maternal obesity during pregnancy and cognitive/neurodevelopmental outcome of children. It is speculated that neuro-programming differs in offspring of obese and normal weight women. Our recent study has revealed lower white matter integrity in widespread regions in infants born to obese but otherwise healthy mothers. In this study, we evaluate and compare the resting-state functional connectivity in 2-week-old infants born to normal weight or obese mothers. Thirty-four full-term healthy infants from uncomplicated pregnancies were included in this IRB approved study. Inclusion criteria for the pregnant women were: pre-pregnancy self-reported BMI 18.5-24.9 (normal weight) or 30-35 (obese); second parity, singleton pregnancy; ≥ 21 years of age; conceived without assisted fertility treatments. Exclusion criteria were: pre-existing medical conditions; medical complications during pregnancy; medications during pregnancy known to influence fetal growth; smoking or alcohol drinking. All enrolled pregnant women had their body composition, BMI and maternal IQ measured within the first 10 weeks of gestation, and gestational weight gain measured at 36 weeks. Only infants born full term (≥ 37 weeks of gestation), with size at birth appropriate for gestational age (AGA), and without medical conditions known to influence growth and development were included for the MRI study (N=18 born to normal weight mothers and N=16 born to obese mothers). At age 2 weeks, the infants underwent an MRI examination of the brain without sedation on a 1.5 Tesla Philips Achieva MRI scanner. A single-shot gradient echo T2*-weighted EPI sequence with TR/TE 2400ms/50ms, acquisition voxel size 2X2X4 mm3 and 150 dynamics was used to acquire the resting-state fMRI data. 3D T1 weighted images were also acquired for structural references. Raw data were exported to a workstation with FMRIB Software Library V5.0 (FSL) installed on a VMware Linux virtual machine for independent component analyses (ICA) using the MELODIC toolbox and associated functions. The computed functional connectivity components were visually inspected to label meaningful networks. In addition to the group ICA analysis for all infants, group ICA for all infants born to normal weight mothers and for all infants born to obese mothers were run separately. Dual regression and Randomise tools in FSL were used for voxel-wise comparisons of functional connectivity between groups for each component. The threshold-Free Cluster Enhancement (TFCE) option in FSL with 5000 permutations was used. To control for potential confounders, a number of variables including maternal IQ and gestational weight gain and infant birth anthropometrics parameters were added into the randomization of FSL as covariates. p<0.05 after multiple comparison correction across voxels was regarded as significant.

The group ICA analysis showed meaningful components representing functional connectivity networks at resting-state consistent with literature findings on term newborns. These components include: primary motor; primary visual; visual association; auditory-left and right; basal ganglia; cerebellum; somatosensory; prefrontal; and default mode network. Dual regression analysis revealed two networks with significant differences in functional connectivity between infants born to normal weight or obese mothers. One is the somatosensory/posterior insula network, in which infants born to obese mothers have lower connectivity (p<0.05, corrected) for recruitment of medial prefrontal cortex (mPFC) compared to the infants born to normal weight mothers (Figure 1 left); the other is the prefrontal cortex network, in which infants born to obese mothers have lower connectivity (p<0.05, corrected) for recruitment of dorsal anterior cingulate cortex (dACC) compared to those born to normal weight mothers (Figure 1 right). No other voxels/clusters in any other components/networks showed significant differences between the two groups.

Region of interest analyses showed that for the two regions with clusters showing significant differences in functional connectivity, the parameter estimate (PE, reflection of connectivity) for the general linear model in the dual regression negatively correlated with maternal fat mass percentage. (Figure 2 left: Partial Pearson correlation r=-0.41, p=0.027 for the mPFC in the somatosensory/posterior insula network; Figure 2 right: Partial Pearson Correlation r=-0.52, p=0.004 for the dACC in the prefrontal cortex network).

Our results suggest significant differences in resting-state functional connectivity associated with maternal obesity in newborn infants.