Exploring functional connectivity in clinical and data-driven groups of preterm and term adults

Abstract
Adults born very preterm (i.e. at <33 weeks’ gestation) are more susceptible to long-lasting structural and functional brain alterations and cognitive and socio-emotional difficulties, compared with full-term controls. However, behavioural heterogeneity within very preterm and full-term individuals makes it challenging to find biomarkers of specific outcomes. To address these questions, we parsed brain–behaviour heterogeneity in participants subdivided according to their clinical birth status (very preterm versus full term) and/or data-driven behavioural phenotype (regardless of birth status). Participants were followed-up in adulthood (median age 30 years) as part of a wider longitudinal case–control cohort study. The Network Based Statistic approach was used to identify topological components of resting state functional connectivity differentiating between (i) 116 very preterm and 83 full-term adults (43% and 57% female, respectively) and (ii) data-driven behavioural subgroups identified using consensus clustering (n = 156, 46% female). Age, sex, socio-economic status and in-scanner head motion were used as confounders in all analyses. Post hoc two-way group interactions between clinical birth status and behavioural data-driven subgrouping classification labels explored whether functional connectivity differences between very preterm and full-term adults varied according to distinct behavioural outcomes. Very preterm compared with full-term adults had poorer scores in selective measures of cognitive and socio-emotional processing and displayed complex patterns of hyper- and hypo-connectivity in sub-sections of the default mode, visual and ventral attention networks. Stratifying the study participants in terms of their behavioural profiles (irrespective of birth status) identified two data-driven subgroups: an ‘At-Risk’ subgroup, characterized by increased cognitive, mental health and socio-emotional difficulties, displaying hypo-connectivity anchored in frontal opercular and insular regions, relative to a ‘Resilient’ subgroup with more favourable outcomes. No significant interaction was noted between clinical birth status and behavioural data-driven subgrouping classification labels in terms of functional connectivity. Functional connectivity differentiating between very preterm and full-term adults was dissimilar to functional connectivity differentiating between the data-driven behavioural subgroups. We speculate that functional connectivity alterations observed in very preterm relative to full-term adults may confer both risk and resilience to developing behavioural sequelae associated with very preterm birth, while the localized functional connectivity alterations seen in the ‘At-Risk’ subgroup relative to the ‘Resilient’ subgroup may underlie less favourable behavioural outcomes in adulthood, irrespective of birth status.