A lifespan perspective on brain-behavioural heterogeneity following very preterm birth

o Study #1: to identify differences in neonatal structural brain volumes in subgroups of VPT born children screening negatively and positively for autism spectrum conditions and to explore the role of developmental delay in mediating and moderating childhood autism traits.

o Study #2: to parse heterogeneity in neonatal clinical and social risk and childhood behavioural outcomes using data-driven integrative consensus clustering and to explore differences in neonatal brain volumes and structural and functional connectivity between the resultant subgroups.

o Study #3: to compare resting state functional connectivity and structural volumes between groups of children stratified both in terms of their clinical characteristics (i.e., VPT and fullterm (FT) birth) and their behavioural profiles identified using data-driven consensus clustering regardless of their gestational age at birth. Post-hoc analyses aimed to elucidate whether clinical and environmental factors differ within VPT or FT children belonging to distinct subgroups.

o Study #4: to use the same approach employed in Study #3 to investigate resting state functional connectivity in a sample of adults born VPT and FT.

Methodology: Distinct psychometric screening criteria (Study #1) and advanced data-driven clustering approaches (Studies #2-4) were used to parse behavioural heterogeneity at different time points throughout development: toddlerhood (Study #1), early childhood (Study #2), middle childhood (Study #3), and adulthood (Study #4). Magnetic resonance (MR) imaging data were collected and analysed using advanced whole-brain approaches such as Tensor Based Morphometry, Tract Based Spatial Statistics, graph theory metrics, and Network Based Statistic to quantify brain volumes and structural and functional connectivity patterns characterising behavioural heterogeneity.

Study participants: Participants were drawn from two cohort studies: (i) the Evaluation of Preterm Imaging Study (ePrime), which evaluated brain development using multi-modal MR imaging (at term equivalent age and at 7-12 years) and behavioural assessments (at 2, 4-7, and 7-12 years); and (ii) the University College Hospital London (UCHL) study, which studied adults (median age 30 years) using multi-modal MR imaging and behavioural assessments.

o Study #1: Smaller neonatal brainstem volumes and high levels of developmental delay were seen in one of two subgroups of VPT toddlers screening positively for autism according to different psychometric criteria, relative to those screening negatively. Developmental delay in this positively screening subgroup was seen to be mediating and moderating the onset of autism traits in childhood. Smaller neonatal cerebellar volumes differentiated between the two distinct subgroups of VPT children screening positively for autism, despite exhibiting similar extent of autism traits in early childhood. Together, results suggest the presence of distinct aetiological trajectories associated with autism traits.

o Study #2: In early childhood, three data-driven behavioural subgroups of VPT children were delineated: (i) a ‘Resilient’ subgroup with favourable behavioural outcomes and a more cognitively stimulating home environment, (ii) an ‘At-risk’ subgroup with behavioural difficulties and high neonatal clinical risk, and (iii) an ‘Intermediate’ subgroup with intermediate behavioural outcomes, low neonatal clinical risk and a less cognitively stimulating home environment. Relative to the ‘Intermediate’ subgroup, the ‘Resilient’ subgroup displayed larger neonatal fronto-limbic regional volumes and functional connectivity and the ‘At-risk’ subgroup showed widespread fronto-temporo-limbic white matter alterations. These findings highlight the value of studying neonatal patterns of functional and structural brain development as potential biomarkers of childhood outcomes, as well as the importance of a supportive home environment to foster child development following VPT birth.

o Study #3: Results show evidence of widespread volumetric alterations and increased functional connectivity in VPT children relative to controls. In middle childhood, stratifying the sample into two data-driven behavioural subgroups, independently of birth status identified: (i) a ‘General Difficulties’ subgroup displaying widespread decreases in functional connectivity and greater behavioural difficulties relative to a (ii) ‘General Resilience’ subgroup. A three-subgroup solution was also explored, identifying: a (I) ‘Neurodevelopmental Difficulties’ subgroup with socio-emotional and higher-order cognitive difficulties and reduced rostro-lateral prefrontal, brainstem, occipital, and cerebellar volumes, and a (II) ‘Psychiatric Difficulties’ subgroup exhibiting psychiatric and executive function difficulties with reduced dorsolateral prefrontal and cerebellar volumes, relative to a (III) ‘Typical Development’ subgroup. All brain differences, apart from cerebellar and occipital volumetric alterations, significantly differentiated between distinct data-driven behavioural subgroups after adjusting for preterm birth, highlighting the presence of VPT-specific neural alterations as well as unique neural patterns underlying behavioural difficulties in the general population, independently of birth status. Furthermore, VPT (but not FT) children belonging to Neurodevelopmental Difficulties or Psychiatric Difficulties subgroups displayed greater social risk relative to those in the Typical Development subgroup.

o Study #4: Complex widespread patterns of both increased and decreased functional connectivity were found in VPT compared to FT born adults in default mode, visual, and ventral attention networks. In adulthood, when VPT and FT born adults were stratified in terms of their behavioural profiles (irrespective of preterm birth), two data-driven subgroups were identified: (i) an ‘At-risk’ subgroup with more behavioural difficulties and reduced functional connectivity in frontal opercular and insular areas relative to a (ii) ‘Resilient’ subgroup with more favourable behavioural outcomes. These results indicate that functional connectivity between the default mode, ventral attention, and visual networks characterise clinically defined groups (VPT and FT) and are different from the connectivity patterns that characterise adults subdivided in terms of their behavioural profiles (irrespective of birth group), which are anchored in insular and frontal opercular regions. Moreover, social risk was found to be greater within adults born VPT belonging to At-risk relative to Resilient subgroups, while this relationship was not identified in those born FT.

Conclusions: Collectively, these findings indicate a long-lasting presence of neurodevelopmental heterogeneity within VPT and FT samples which seems to persist throughout the lifespan. Specific structural and functional alterations differentiating between distinct behavioural profiles across both VPT and FT samples are also identified; whereby alterations localised to fronto-temporolimbic regions seem to be characteristic of behavioural difficulties in VPT and FT samples regardless of their birth status, while alterations to brain regions including visual and cerebellar areas may be characterising biomarkers of outcomes specifically in VPT samples. Implications of these findings suggest a potential benefit of using MRI to detect neurobiological markers of behavioural outcomes, which can in turn guide the implementation of personalised interventions for those at-risk of developing specific behavioural difficulties. Results presented here also highlight the importance of fostering an enriching environment to probe resilience against developing behavioural difficulties, particularly for those born VPT.