Extremely Preterm Children Demonstrate Interhemispheric Hyperconnectivity During Verb Generation: a Multimodal Approach

Abstract

Children born extremely preterm (EPT, <28 weeks gestation) are at risk for delays in development, including language. We use fMRI-constrained magnetoencephalography (MEG) during a verb generation task to assess the extent and functional connectivity (phase locking value, or PLV) of language networks in a large cohort of EPT children and their term comparisons (TC). 73 participants, aged 4 to 6 years, were enrolled (42 TC, 31 EPT). There were no significant group differences in age, sex, race, ethnicity, parental education, or family income. There were significant group differences in expressive language scores (p<0.05). Language representation was not significantly different between groups on fMRI, with task-specific activation involving bilateral temporal and left inferior frontal cortex. There were group differences in functional connectivity seen in MEG. To identify a possible subnetwork contributing to focal spectral differences in connectivity, we ran Network Based Statistics analyses. For both beta (20-25 Hz) and gamma (61-70 Hz) bands, we observed a subnetwork showing hyperconnectivity in the EPT group (p<0.05). Network strength was computed for the beta and gamma subnetworks and assessed for correlation with language performance. For the EPT group, exclusively, strength of the subnetwork identified in the gamma frequency band was positively correlated with expressive language scores (r=0.318, p<0.05). Thus, interhemispheric hyperconnectivity is positively related to language for EPT children and might represent a marker for resiliency in this population.

Figure 1:. Functional MRI activation maps and resultant network nodes.

A) Three slices through temporal, parietal, and left frontal language areas showing general linear model (GLM) analysis across all participants (EPT + TC) with typical bilateral activation in response to language stimuli (auditorily presented nouns for verb generation) versus noise condition (p < 0.001, k = 8) and 3D surface rendering of the normalized brain averaged across all participants to show bilateral activation from a sagittal perspective. B) The initial activation map across the entire sample is subjected to a 200-unit parcellation scheme to produce the “nodes” for network analysis, shown in blue.

Figure 2:. Functional connectivity indexed by phase locking value (PLV) for each group

Phase locking value (PLV) throughout the language network identified in Figure 1 for extremely preterm children (EPT, n = 31, shown in pink) and term controls (TC, n = 42, in black) with standard error shaded around the mean (SEM). Areas of group differences on visual inspection include segments of the beta band (20–25 Hz) and the gamma band (61–70 Hz).

Figure 3:. Extremely preterm children show increased beta functional connectivity in language network.

Network “edges” showing significantly increased functional connectivity in EPT versus TC between 20 and 25 Hz during verb generation (observed at various initial thresholds ranging from t =1 to 2, 5000 iterations, p < 0.05, corrected for multiple comparisons).

Figure 4:. Extremely preterm children show increased gamma functional connectivity in language network.

Network “edges” showing significantly increased functional connectivity in EPT versus TC between 61 and 70 Hz during verb generation (observed at various initial thresholds ranging from t =1 to 4, 5000 iterations, p < 0.05, corrected for multiple comparisons).

Figure 5:. Relationship between gamma band-limited connectivity (network strength) in the identified subnetwork supporting language and expressive language scores for the preterm group.

Scores on the Expressive Vocabulary Test (EVT2) versus network strength within the significant subnetwork identified in gamma band-limited connectivity for the EPT group exclusively. Results indicate a positive bivariate correlation (Pearson’s r = 0.32, p < 0.05). Correlations between EVT scores and network strength were not significant across groups or within the term control group.