Alterations in cortical thickness following very preterm birth have been associated with neuropsychological/behavioural impairments. Here I investigated cortical thickness development in individuals who were born very preterm (gestational age < 33 weeks) between adolescence (15 years old, Time 1) and adulthood (20 years old, Time 2). Using univariate approaches, I compared cortical thickness in preterm-born individuals and controls cross-sectionally, and also assessed cortical thickness changes between Time 1 and Time 2 within each group as well as comparing them between groups. Using a multivariate approach (support vector machine, SVM), I examined spatially distributed between-group differences in cortical thickness at Time 1, and based on this, I predicted group membership of brains scanned at Time 2. Results showed that (1) at Time 1 preterm born adolescents showed greater cortical thickness compared to controls predominantly in occipitotemporal and prefrontal areas as well as temporal poles, differences which decreased by early adulthood; (2) at both time points preterm-born participants showed smaller cortical thickness compared to controls in parahippocampal regions. Longitudinal decrease, but not increase of cortical thickness was observed in all lobes, with the preterm group showing more extensive and widespread changes. SVM identified temporal, occipitotemporal, parietal and prefrontal cortices to be best discriminating between the groups at Time 1 and classified group membership at Time 2 with an accuracy of 86.5%. Longitudinal changes in cortical thickness in left temporal pole, right occipitotemporal gyrus and left superior parietal lobe were significantly associated with executive function scores. To summarise, alterations in cortical thickness development in preterm-born individuals last into early adulthood, with implications for high-order cognitive processing. The proposal, put forward by some previous studies that preterm individuals would developmentally catch up with the control group was consistent with the results from univariate analysis, but not with those from multivariate analysis.
|Date of Award||2015|
|Supervisor||Chiara Nosarti (Supervisor) & Andy Simmons (Supervisor)|