Heterogeneous, temporally consistent, and plastic brain development after preterm birth

Abstract

The current view of neurodevelopment after preterm birth presents a strong paradox: diverse neurocognitive outcomes suggest heterogeneous neurodevelopment, yet numerous brain imaging studies focusing on average dysmaturation imply largely uniform aberrations across individuals. Here we show both, spatially heterogeneous individual brain abnormality patterns but with consistent underlying biological mechanisms of injury and plasticity. Using cross-sectional structural magnetic resonance imaging data from preterm neonates and longitudinal data from preterm children and adults in a normative reference framework, we demonstrate that brain development after preterm birth is highly heterogeneous in both severity and patterns of deviations. Individual brain abnormality patterns are also consistent for their extent and location along the life course, associated with glial cell underpinnings, and plastic for influences of the early social environment. Our findings extend conventional views of preterm neurodevelopment, revealing a nuanced landscape of individual variation, with consistent commonalities between subjects. This integrated perspective implies more targeted theranostic intervention strategies, specifically integrating brain charts and imaging at birth, as well as social interventions during early development.

Fig. 1. Study overview. Schematic of the study workflow, from hypotheses (left) to data sources (middle) to analysis steps (right).

a Cross-sectional and longitudinal cortical thickness (CTh) and surface area (SA) data were obtained from three developmental cohorts, including preterm (PT) and full-term (FT) participants. Data were parcellated into 34 bilateral cortical regions. Population CTh and SA life course trajectories were extracted from a normative model. For each participant, individual regional deviations from population life courses were classified as infranormal (i.e., <5th percentile) or supranormal (i.e., >95th percentile) for each cortical region. The regional deviation profile of a certain modality is defined as the measure of the individual brain abnormality pattern (IBAP) for a given participant. IBAP heterogeneity was assessed in two ways: by quantifying the number of subjects with extranormal deviations in each region, and by measuring spatial similarity of regional deviation profiles across subjects. For the latter, binarized regional deviation profiles were cross-correlated to determine the average correlation between each subject’s IBAP with all others. b Consistency in initial brain injury following preterm birth was examined in terms of extent, location, and cellular underpinnings. Extent consistency was examined by relating a subject’s number of extra-normal regional deviations to their gestational age (GA). c Location consistency was exami ned by within-subject longitudinal comparison of IBAPs along childhood and adulthood. d Cellular underpinning consistency was examined by linking gestational age with the spatial correlation of eight brain cell type distributions with adult IBAPs, respectively. Investigated cell types were astrocytes (Astro), endothelial cells (Endo), microglia (Micro), excitatory neurons (Neuro-Ex), inhibitory neurons (Neuro-In), oligodendrocytes (Oligo), oligodendrocyte precursors (OPC), and pericytes (Per). e To investigate the developmental plasticity of IBAPs in children and adults, we captured spatial variability of IBAPs across 34 cortical regions using Principal Component Analysis (PCA) as the main axis of deviation across regions (i.e., PC1) and linked it with social environmental factors during early life.

Fig. 2. Brain development after preterm birth is individually heterogeneous.

a Cortical thickness (CTh) average dysmaturation outcome for 26-year-old adults after preterm birth estimated by linear regression models correcting for age and sex of between-group differences (p_FDR <0.05, see Supplementary Table S2j for exact p values) suggests abnormalities shared between preterm subjects. b Individual regional deviations were estimated from a previously published normative framework and classified as infranormal (i.e., <5th percentile) or supranormal (i.e., >95th percentile) for a given cortical region. The lifespan trajectory of the rostral anterior cingulate CTh in males is shown as an example region, with the rostral anterior cingulate CTh of male preterm (PT, red) and full-term (FT, black) subjects plotted on top. c Individual deviation score profiles of example subjects with similar gestational age (GA) and birth weight (BW). Left column: Deviation scores of all 34 cortical regions, right column: only infra- (i.e., <5th percentile, dark blue) and supranormal (i.e., >95th percentile, dark red) deviations. Adult 1: GA = 30 weeks, BW = 1450 g; adult 2: GA = 30 weeks, BW = 1460 g; adult 3: GA = 30 weeks, BW = 1465 g. d The percentage of subjects sharing an extranormal deviation in any given region is less than 30%, illustrating individual heterogeneity of CTh after preterm birth. e Binarized representation of extranormal deviations for each subject. f Spearman correlation matrix of binarized extranormal deviations across subjects. g Distribution of averaged correlation coefficients for each subject with all others. h Further example individual deviation score profiles. Left and right columns as in panel (c). See Supplementary Table S1 for region abbreviations. Source data are provided as a Source Data file.

Fig. 3. Individual heterogeneity of cortical thickness after preterm birth is lasting across age groups.

a Cortical thickness (CTh) average dysmaturation outcome after preterm birth estimated by linear regression models correcting for age and sex (p_FDR <0.05, see Supplementary Tables S2 for exact p-values). Results are shown for neonates (dHCP), children (ABCD-10 and ABCD-12), and adults (BLS-38). b Bar plots representing the percentage of preterm (PT) and full-term (FT) subjects sharing an extranormal deviation in any given region. An overlap of less than 20% suggests substantial heterogeneity between subjects. c Distribution of averaged correlation coefficients of binarized extranormal deviation profiles for each subject with all others. See Supplementary Table S1 for region abbreviations and number encoding. Source data are provided as a Source Data file.

Fig. 4. Individual heterogeneity of regional surface area after preterm birth across cohorts.

a Surface area (SA) average dysmaturation outcome after preterm birth estimated by linear regression models correcting for age and sex (p_FDR <0.05). Results are shown for neonates (dHCP), children (ABCD-10), and adults (BLS-26 and BLS-38). b The percentage of preterm (PT) and full-term (FT) subjects sharing an extranormal deviation in any cortical region. Not more than 30% of subjects overlap in any region, demonstrating that spatial heterogeneity between individuals is also evident for regional SA development after preterm birth. As regional percentile SA trajectories for the entorhinal cortex were not available from the normative model, no subjects of any cohort shows extranormal deviations in that region. c Distribution of averaged correlation coefficients of binarized extranormal deviation profiles for each subject with all others. See Supplementary Table S1 for region abbreviations and number encoding. Source data are provided as a Source Data file.

Fig. 5. Consistency of IBAP extent.

The number of supranormal or infranormal CTh deviations per subject was correlated with gestational age for dHCP, ABCD-10, and BLS-26, respectively (dHCP: Spearman rho(129) =  −0.255, p = 0.003, p_FDR = 0.005, CI = [–0.409, –0.087], two-tailed; ABCD-10: Spearman rho(380) = –0.098, p = 0.055, p_FDR = 0.055, CI = [–0.197,0.002], two-tailed; BLS-26: Spearman rho(94) = –0.313, p = 0.002, p_FDR = 0.005, CI = [–0.484, –0.120], two-tailed). Orange lines and shades represent the linear regression line with 95% confidence intervals based on 10,000 bootstrap resamples. Source data are provided as a Source Data file.

Fig. 6. Temporal consistency of IBAP anatomical location.

Deviation scores for regional surface area (SA) were computed in preterm individuals with available longitudinal data using the bilateral Desikan-Killiany parcellation: a for preterm children from the ABCD cohort (n = 296) at ages 10 and 12 years and b for preterm adults from the BLS cohort (n = 46) at ages 26 and 38 years. Five representative regions are shown here; a complete representation can be found in Supplementary Figs. S14–S15. Only individuals who exhibited an extranormal deviation (i.e., infranormal: <5th percentile, supranormal: >95th percentile) at either timepoint are depicted. Within-subject comparisons demonstrate that the anatomical location of infranormal (blue) and supranormal (red) deviations remain largely consistent across time. Intraclass correlation coefficients (ICCs) were used to quantify temporal stability of regional deviations. Regional distributions of ICCs are shown on left hemispheric cortical surfaces for (c) preterm children and d preterm adults. See Supplementary Table S1 for region abbreviations. Source data and exact p-values are provided as a Source Data file (panels a, b) and in Supplementary Tables S4a, b (panels c, d).

Fig. 7. Temporal consistency of IBAP cellular underpinnings.

a Regional Spearman correlation matrix of cortical thickness (CTh) individual brain abnormality patterns (IBAPs) from preterm adults aged 26 years (vertical) with mean expression profile of eight brain cell types (horizontal). CTh IBAPs of three example subjects are plotted for illustration purposes. Investigated cell types were astrocytes (Astro), endothelial cells (Endo), microglia (Micro), excitatory neurons (Neuro-Ex), inhibitory neurons (Neuro-In), oligodendrocytes (Oligo), oligodendrocyte precursors (OPC), and pericytes (Per). b Spearman correlation coefficients for the association between astrocytes, endothelial cells, oligodendrocytes, and OPCs, with individual CTh deviations were significantly related to gestational age in preterm adults (p < 0.05, two-sided). Orange lines and shades represent the linear regression line with 95% confidence intervals based on 10,000 bootstrap resamples. See Supplementary Table S5 and Source Data files for Source Data and exact p values.

Fig. 8. Deviations of children and adults are linked to the social environment during early development.

To capture regional variation of deviation scores across 34 cortical regions, we applied Principal Component Analysis (PCA) to individual deviations (Fig. 1e). Next, we assessed the association between the first principal component (PC1) of regional deviations and early developmental social factors, including socio-economic status (SES) and quality of mother-infant relationship (Parent-Infant Relationship Index, PIRI; BLS-26 only) using Spearman rank correlation. a In preterm children of the ABCD-10 cohort, the association between SES with the first principal component (PC1) of regional cortical thickness (CTh) (left, Spearman rho(380) = –0.002, p = 0.977, p_FDR = 0.977, CI = [–0.102,0.098], two-sided) and of regional surface area (SA, right, Spearman rho(294) = 0.269, p = 0.001, p_FDR = 0.006, CI = [0.160,0.371], two-sided) is shown. b In preterm adults of the BLS-26 cohort, the association between CTh PC1 with SES (left, Spearman rho(94) = –0.217, p = 0.033, p_FDR = 0.088, CI = [–0.400, –0.018], two-sided) and with PIRI (Spearman rho(89) = − 0.211, p = 0.044, p_FDR = 0.088, CI = [–0.400, –0.006], two-sided) is illustrated. Orange lines and shades in panels (a, b) represent the linear regression line with 95% confidence intervals based on 10,000 bootstrap resamples. c In adults, the association between gestational age (GA) and CTh PC1 scores was moderated by SES (interaction term SES x GA in linear regression model: β = 0.104, p = 0.035, SE = 0.057, CI = [–0.0090,0.217], one-sided). Conditional effects of SES on GA are shown on the right. A low (red, effect = 0.246, p = 0.35 × 10 × 10^–4, one-sided) or middle (yellow, effect = 0.143, p = 0.001, one-sided) SES significantly moderates the relationship between SES and GA, whereas a high SES (green, effect = 0.039, p = 0.303, one-sided) does not. Source data are provided as a Source Data file.

Fig. 9. Individual brain deviations after preterm birth are associated with cognitive outcome variability in later life.

Spearman correlations between the first principal component (PC1) of regional cortical thickness (CTh, left) or of regional surface area (SA, right) with measures of cognition in preterm neonates (a), children aged 10 years (b), and adults aged 26 years (c). Orange lines and shades represent the linear regression line with 95% confidence intervals based on 10,000 bootstrap resamples. All statistical tests were two-sided, and uncorrected p values are provided for illustration purposes. Source data are provided as a Source Data file.

Fig. 10. Conceptual transition from average dysmaturation outcome-based view to individual brain abnormality pattern centered model of prematurity.

Schematic representation of the two contrasted concepts. a The prevailing view of brain aberrations based on group-level brain imaging studies after preterm birth suggests injury-induced average dysmaturation outcomes, disregarding individual variability. b The extended concept presented here emphasizes the heterogeneity of individual brain abnormality patterns (IBAPs) among preterm subjects, as a result of individual initial injury patterns (indicated by varying extent and location of abnormalities across three arbitrary subject cases), subsequent dysmaturation (indicated by a brain trajectory for each case), and plastic individual changes due to social environment influences (indicated by horizontal boxes around distinctively developing brain trajectories). Critically, the model of heterogeneous and plastic IBAPs suggests specific theranostic approaches on prematurity (blue shaded vertical boxes): on the one hand, neonatology and chart-based brain imaging approaches to identify at-risk infants, and on the other hand, social psychology approaches to social environment to modify developmental trajectories. Thus, heterogenous IBAPs seem to have the potential to frame and integrate approaches of basic neuroscience, neonatology, brain imaging, and social psychology on human prematurity. Source data are provided as a Source Data file. Icons used from Google’s Material Icons (https://fonts.google.com/icons; licensed under Apache License 2.0).