Structural brain network in relation to language in school-aged extremely preterm children: A diffusion tensor imaging study

Between 22 and 45 % of children born preterm experience difficulties with expressive and receptive language when they reach school age. Little is currently known about the neural mechanisms behind their linguistic performance. This study investigates the brain areas and white matter connections that form the structural language network in extremely preterm-born children who have reached school age. Structural brain connectivity was quantified using diffusion-weighted imaging (DWI) and tractography in n = 58 (62 % female) extremely preterm-born children aged 8-12 years. Language outcomes were assessed using the CELF-4-NL Recalling Sentences subtest. Language scores were below average in n = 13 (22 %) children. Language outcomes related significantly to a subnetwork of 16 brain regions (p = 0.012). The network comprised brain regions from the left hemisphere including the pars orbitalis, middle and superior frontal gyrus, frontal pole, pre- and postcentral gyrus, superior temporal gyrus, insula, caudate nucleus, thalamus, and putamen. In the right hemisphere, the anterior cingulate was part of the network. These findings suggest that extremely preterm children rely mostly on their left hemisphere during language processing, which is similar to typically developing children. However, they seem to use compensatory neural pathways that include brain areas right next to the areas typically involved in language processing. These areas include the pars orbitalis (adjacent to Broca's area) and the putamen and caudate nucleus (adjacent to the limbic system). It is important to note that language difficulties were not necessarily related to brain injury around birth.