{"paper_id":"cb514533-bcb0-4401-af87-0ee498e0c1dc","body_text":"Title: Neonatal brain volumes and birth characteristics predict \nbehavioural outcomes in toddlerhood  \nShort title: Neonatal predictors of toddlerhood outcomes \nYumnah T. Khan1*, Alex Tsompanidis1, Carrie Allison1, Richard A.I. Bethlehem2, and Simon \nBaron-Cohen1,2 \n1. Autism Research Centre, Department of Psychiatry, University of Cambridge, Cambridge, UK, CB2 8AH.  \n2. Department of Psychology, University of Cambridge, Cambridge, CB2 3EB, UK. \n*Corresponding author: Y umnah T. Khan ( yk415@cam.ac.uk). 18b Trumpington Road, Cambridge, \nCambridgeshire, UK CB2 8AH \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\n \nABSTRACT  \nBackground: Early brain structure and birth factors (e.g., sex, birth weight, gestational age at \nbirth) are understood as critical to shaping lifelong developmental and psychopathological \noutcomes. Methods: Using data from the Developing Human Connectome Project, we \nexamined whether neonatal brain volumes and birth factors predict developmental and \nsocioemotional outcomes in toddlerhood. Structural MRI scans were acquired from 391 \ninfants at birth (193 females, 198 males; mean age = 8 days), with follow-up behavioural \nassessments conducted in toddlerhood (mean age = 18 months). Results: Results \ndemonstrated that larger neonatal brain volumes were associated with lower autistic traits and \nhigher cognitive, language and motor outcomes. A higher gestational age and weight at birth \nwere associated with higher scores on various of these outcomes - an effect that was partially \nmediated by larger brain volumes at birth. Females showed higher language scores compared \nto males, though this effect was suppressed rather than mediated by neonatal brain volumes. \nConclusions: These findings demonstrate how birth factors interactively shape early \ndevelopmental and neuropsychiatric outcomes.  \nKeywords: brain structure, birth factors, early behaviour, sex differences, autism  \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nINTRODUCTION \nThe prenatal and early postnatal periods are critical phases in human brain development. \nDuring these periods, the brain undergoes structural growth with remarkable speed and \ncomplexity, laying the foundation for subsequent brain development and function. Evidence \nis accumulating for the significance of this early period in predicting lifelong cognitive, \nbehavioural, developmental, and health outcomes (1–4). However, while brain-behaviour \nassociations have been predominantly researched during later stages of development, these \nlinks remain comparatively less characterised during the earliest stages of life – a period \nduring which both brain structure and behaviour undergo rapid and dynamic changes (5–7).  \nExisting research, predominantly in infants born preterm, has shown that larger brain \nvolumes at birth are associated with higher scores on a range of childhood behavioural and \ncognitive outcomes. For instance, larger neonatal cerebellar volumes have been linked to \nhigher scores on psychomotor functioning (8) and autistic traits (9); larger hippocampal \nvolumes with reduced hyperactivity, emotional, behavioural and peer problems (10); and \nlarger thalamic and basal ganglion volumes with higher scores on motor, intelligence, and \nacademic measures (11). However, the majority of existing studies have focused \npredominantly on preterm or clinical populations. Studying broader, population-level \nvariability is critical to building a more comprehensive and generalisable picture of how early \nlife biology shapes the course of development. \nBeyond brain structure, broader birth factors, such as gestational age and weight at birth, are \nalso robust predictors of future developmental outcomes. While the effects of preterm birth \nare well-documented, studies suggest that even within the term range, variations in \ngestational age can impact future outcomes (12). For instance, being born at a later \ngestational age has been linked to higher scores on motor, cognitive, language, academic, and \nmental health measures (13,14). Similarly, birth weight has been positively associated with \ncognitive, neurodevelopmental, and psychopathological outcomes, even within the typical \nbirth weight range (Cortese et al., 2021; Gonçalves et al., 2024.; Pettersson et al., 2019). It is \nimportant to note, however, that both gestational age at birth and birth weight are strongly \nassociated with neonatal brain volumes (18). As such, it is important to distinguish whether \nthese factors are directly linked to developmental outcomes, or whether their effects are \nmediated indirectly via brain structure (i.e., a smaller gestational age/weight at birth impacts \nbrain structure, which, in turn, impacts future outcomes).  \nSex at birth is another important biological factor that affects both brain development as well \nas a range of outcomes across the lifespan. Studies have shown that sex differences in brain \nvolumes are present from birth (19), with males and females showing diverging growth \ntrajectories from prenatal development (7). These neuroanatomical differences are paralleled \nby sex differences in behaviour and cognition, which have also been observed as early as \nbirth (20). During infancy, differences also begin to emerge in motor, language, and cognitive \noutcomes, with females generally showing faster maturation compared to males (21–24). For \ninstance, males, on average, are more likely to take longer to develop language skills (25) and \naspects of social cognition, such as joint attention (26). They are also more likely to have \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nhigher autistic traits compared to females, as measured in parent-report questionnaires (27). \nWhile neonatal brain volume has recently been shown to associate negatively with autistic \ntraits (28), the role of baseline sex differences in this effect remains unclear. Additionally, \nwhile sex differences in both the brain and behaviour are fairly well-established, the link \nbetween the two is less understood in early life. Since differences are concurrently present \nacross both domains, it is possible that sex differences in behavioural outcomes can be \npartially explained by early emerging differences in brain structure.   \nTo understand how perinatal growth predicts future cognitive development and behaviour, the \npresent study aimed to: (a) investigate whether neonatal brain structure predicts a range of \nbehavioural and cognitive outcomes in toddlerhood (e.g., motor, language, and cognitive \ndevelopment, internalising and externalising traits, and autistic traits), (b) examine whether \nneonatal brain volumes mediate the associations between birth factors (e.g., gestational age, \nbirth weight) and these outcomes, and (c) explore whether sex differences in early brain \nstructure mediate sex differences in future outcomes. To address these objectives, we \nleveraged data from the developing Human Connectome Project (dHCP) (29), where infants \nunderwent structural MRI shortly after birth followed by behavioural assessments in \ntoddlerhood. \nMETHODS AND MATERIALS \nParticipants \nParticipants were recruited as part of the developing Human Connectome Project (dHCP) \n(29). Structural MRI data were acquired neonatally (mean age = 8.55 days), and follow-up \nbehavioural assessments were conducted in toddlerhood (mean age = 18.98 months). The \nexclusion criteria employed in this study included preterm births (< 37 weeks gestational \nage), multiple births, the presence of brain anomalies in the MRI scan with likely analytical \nand clinical significance, pregnancy or neonatal clinical complications, and a postnatal age > \n28 days at the time of the MRI scan to capture the neonatal period. Participants with any \nmissing behavioural measures at follow-up were also excluded. The final sample consisted of \n391 infants (193 females, 198 males), and further sample characteristics are reported in Table \n1. As assessed by two-sample t-tests, no sex differences were observed in any of the sample \ncharacteristics.  \nOutcome Measures \nQ-CHAT \nThe Quantitative Checklist for Autism in Toddlers (Q-CHA T) is an 18-item parent-report \nquestionnaire designed to quantitively measure autistic traits in toddlers through assessing \nvarious social-communication, repetitive, stereotyped, and sensory behaviours (27). While \nnot a diagnostic instrument, it aims to capture variations in autistic traits, which are \nunderstood to exist along a continuum within the general population. \nThe Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III) \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nBayley-III is a standardised assessment, administered by a trained professional, that evaluates \ninfant development through structured tasks and observations (30). The assessment provides \nage-normed scores across cognitive, language (receptive and expressive), and motor (gross \nand fine) domains. \nChild Behavioural Checklist (CBCL) for Ages 1.5–5 \nThe CBCL is a 100-item questionnaire measuring the frequency of behavioural and \nemotional problems in young children (31). The measure contains problem behaviour \nsyndrome subscales which can be grouped into two higher-order factors: internalising \n(measuring emotionally reactive, anxious/depressed, somatic, and withdrawn traits) and \nexternalising (measuring attention problems and aggressive behaviour).  \nMRI Data Acquisition \nAnatomical data acquisition parameters are specified in the dHCP protocol (29). Data were \ncollected using a 3-Tesla Philips Achieva system (Philips Medical Systems) with the dHCP \nneonatal brain imaging system, which included a neonatal 32 channel phased array head coil \nand a customised patient handling system (Rapid Biomedical GmbH, Rimpar, Germany). \nInfants were scanned without sedation after being fed and swaddled. Earplugs (President \nPutty, Coltene Whaledent, Mahwah, NJ, USA) and neonatal earmuffs (MiniMuffs, Natus \nMedical Inc., San Carlos, CA, USA) were used for auditory protection. Heart rate, oxygen \nsaturation, and temperature were monitored throughout the scans.  \nThe imaging protocol was designed to maximise contrast-to-noise ratio by using a Cramer \nRao Lower bound approach (32). Both T2-weighted and T1-weighted inversion recovery Fast \nSpin Echo (FSE) images were obtained in sagittal and axial planes with relaxation times set \nat T1/T2: 1800/150ms for gray matter and at T1/T2: 2500/250 ms for white matter (33).  The \nin-plane resolution was 0.8 × 0.8 mm², with a slice thickness of 1.6 mm and an inter-slice \noverlap of 0.8 mm; for T1-weighted sagittal images, the overlap was 0.74 mm. Specific \nsequence parameters included: for T2-weighted images, a repetition time (TR) of 12000 ms, \necho time (TE) of 156 ms, and SENSE acceleration factors of 2.11 (axial) and 2.60 (sagittal); \nfor T1-weighted images, TR/TI/TE values were 4795/1740/8.7 ms with SENSE factors of \n2.27 (axial) and 2.66 (sagittal). In addition, 3D MPRAGE images were collected at an \nisotropic resolution of 0.8 mm, with TR/TI/TE of 11/1400/4.6 ms and a SENSE factor of 1.2 \nin the right-left direction. To enable high-resolution volumetric analysis, images were \nprocessed with motion correction algorithms, and axial and sagittal scans were combined into \na single 3D image for high resolution and accurate segmentation (34). \nMRI Data Pre-Processing \nStructural MRI data were pre-processed using the dHCP pipeline (35). T2-weighted images \nunderwent motion correction, bias field correction, and brain extraction using the Brain \nExtraction Tool (Smith, 2002). A probabilistic tissue atlas was then aligned to the bias-\ncorrected T2-weighted scans, and initial classification of tissue types (cerebrospinal fluid, \nwhite matter, cortical grey matter, and subcortical grey matter) was carried out using the \nDraw-EM segmentation algorithm (37). Labelled brain atlases (38) were registered to each \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nsubject’s images using multi-channel registration that incorporated both T2-weighted image \nintensities and gray matter probability maps derived from the initial segmentation. This \nprocess generated a detailed segmentation encompassing 87 grey and white matter regions \n(37–39).  \nStatistical Analysis \nFor each behavioural measure, associations between gray matter volumes at birth and \noutcomes in toddlerhood were assessed using linear regression models. Covariates included \nsex, postconceptional age at the neonatal scan, and infant age at follow-up (with the \nexception of Bayley-III scores, which were already standardised based on the toddler’s age). \nThese models were applied across all brain volumes and outcomes of interest. To isolate \nregional effects independent of overall brain size, additional models were ran using total \nbrain volume as a covariate.  \nSex differences in behavioural outcomes were assessed using two-sample t-tests (for Bayley-\nIII scores) or ANOV As (for all remaining measures) while controlling for age at follow-up. \nAssociations with birth weight and gestational age at birth were examined using linear \nregressions. Mediation analyses (bootstrapping with 5,000 resamples and bias-corrected 95% \nconfidence intervals) were then conducted to test whether neonatal brain volumes accounted \nfor any observed associations. In these models, sex, gestational age, or birth weight served as \npredictors, behavioural measures as outcomes, and pre-selected neonatal brain volumes \n(which showed significant associations with both the predictor and outcome) as mediators \n(see Figure 1). Gestational age at scan was included as a covariate in all models, while total \nbrain volume was included as a covariate in models where regional effects, independent of \ntotal brain size, were identified. False discovery rate (FDR) corrections (40) were applied at a \nthreshold of 0.05 within the following sets of analyses: (a) models assessing brain-behaviour \nassociations, separately for each outcome, (b) all models assessing associations between \ngestational age at birth and outcomes, (c) all models assessing associations between birth \nweight and outcomes, (d) all models assessing sex differences in outcomes, and (e) mediation \nmodels, separately for each outcome.  \nFigure 1. Mediation model \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\n \n \nRESULTS \nCorrelations between behavioural outcomes \nTable 2 presents pairwise correlations between all outcome measures at 18 months. Motor, \ncognitive, and language scores all showed moderate associations with one another. Higher Q-\nCHAT scores were associated with higher externalising and internalising traits, and with \nlower cognitive, language, and motor scores. Internalising showed negative associations with \ncognitive, language, and motor outcomes, while externalising showed no significant \nassociations with these outcomes. \nAutistic traits \nHigher scores on the Q-CHAT were associated with lower total brain, total subcortical grey \nmatter, and total white matter volumes. Regionally, higher scores were associated with lower \ngrey matter volumes in the bilateral thalamus, right lentiform nucleus, bilateral posterior \nparahippocampal gyri, brainstem and left anterior and bilateral posterior medial and inferior \ntemporal gyri (Table 3, Supplementary Table 1). No significant regional associations were \nidentified after controlling for total brain volume (Supplementary Table 2). \nBayley-III \nHigher scores on the cognitive subscale were associated with increased cortical grey matter \nvolumes (Table 4). Regionally, higher scores were associated with increased grey matter \nvolumes in the left thalamus, right occipital lobe, left posterior lateral occipitotemporal gyrus \n(fusiformis gyrus), and right medial and inferior temporal gyrus (Table 4, Supplementary \nTable 3). No significant regional associations emerged after controlling for total brain volume \n(Supplementary Table 4).  \nHigher language scores were associated with increased total brain, total white matter, and \ntotal cortical and subcortical grey matter volumes (Table 5). Regionally, higher scores were \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nassociated with increased grey matter volumes in various cortical and subcortical structures \n(Table 5, Supplementary Table 5). No significant regional associations were identified after \ncontrolling for total brain volume (Supplementary Table 6).  \nHigher motor scores were associated with decreased ventricular volumes (B= -0.001, SE = \n0.000, pFDR = 0.023) (Supplementary Table 7). After controlling for total brain volume, higher \nmotor scores were associated with decreased volumes in the brainstem, ventricles, and left \ncerebellum (Table 6, Supplementary Table 8). \nCBCL \nInternalising traits were negatively associated with absolute cerebellar volumes, and, after \ncontrolling for total brain volume, positively associated with frontal lobe volumes – though \nthese associations did not survive multiple comparison corrections (Supplementary Tables 9 \n& 10). There were no significant associations between externalising traits and any global or \nregional brain volumes both before and after controlling for total brain volume \n(Supplementary Tables 11 & 12).  \nMediations  \nSex Differences  \nSex differences were observed in language outcomes (t = 3.44, pFDR = 0.004, Cohen’s d = \n0.35), with higher scores in females (M = 101.84, SD = 14.74) compared to males (M = \n96.54, SD = 15.70). All regions that showed significant brain-language associations \nsignificantly suppressed, rather than mediated, the relationship between sex and language \noutcomes (Supplementary Table 13). Males showed significantly higher externalising (F = \n3.99, p = 0.046, η ²p = 0.01) and autistic traits (F = 4.36, p = 0.037, η ²p = 0.01), though these \nassociations did not survive multiple comparison corrections (autistic traits: pFDR = 0.093, \nexternalising: pFDR = 0.093). No sex differences were observed in internalising (F = 0.00, \npFDR = 0.975, η ²p = <0.001), motor (t= -0.14, pFDR = 0.975, Cohen’s d = -0.01), or cognitive (t \n= 1.62, pFDR = 0.161, Cohen’s d = 0.16) outcomes. \nGestational age at birth  \nGestational age at birth showed significant positive associations with cognitive (\nβ  = 1.305, \npFDR = 0.021, R2 = 0.019), motor (β  = 1.064, pFDR = 0.021, R2 = 0.018), and language \noutcomes (β  = 1.821, pFDR = 0.021, R2 = 0.018). The associations with cognition were \nmediated by volumes in the right occipital lobe (prop. mediated = 0.575, 95% CI [0.155, \n1.170], pFDR = 0.024) and right parahippocampal gyrus (anterior part) (prop. mediated = \n0.245, 95% CI [0.040, 0.550], pFDR = 0.016). There were no significant mediations with \nlanguage and motor outcomes (Supplementary Tables 14 & 15). No significant associations \nwere identified with internalising (\nβ  = -0.193, pFDR = 0.446, R2 = 0.002), externalising (β  = \n0.026, pFDR = 0.933, R2 = 0.000), or autistic traits (β  = -0.402, pFDR = 0.422, R2 = 0.003).  \nBirth weight \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nBirth weight showed significant associations with cognitive (β  = 2.527, pFDR = 0.029, R2 = \n0.012) and language outcomes (β  = 4.019, pFDR = 0.015, R2 = 0.015), while no significant \nassociations were identified with motor (β  = 1.697, pFDR = 0.095, R2 = 0.007), internalising (β  \n= -0.736, pFDR = 0.163, R2 = 0.005), externalising (β  = 0.784, pFDR = 0.294, R2 = 0.003), or \nautistic traits (β  = -1.571, pFDR = 0.085, R2 = 0.007). The relationship between birth weight \nand cognition was significantly mediated by volumes in various of the structures that showed \nsignificant brain-cognition associations (Supplementary Table 16). Similarly, the relationship \nbetween birth weight and language was significantly mediated by volumes in various of the \nstructures that showed significant brain-language associations (Supplementary Table 17).   \nDISCUSSION \nIn the present study, we investigated the role that neonatal brain structure, sex, and perinatal \nfactors play in predicting early cognitive and behavioural outcomes. We found that while \nneonatal brain structure is a reliable predictor of autistic traits, language, cognitive, and motor \nskills, it offers limited predictive value for psychopathological traits (i.e., internalising and \nexternalising traits) in toddlerhood. Additionally, in some cases, neonatal brain volumes \npartially explain the relationship between toddlerhood outcomes and factors such as sex, \nweight, and gestational age at birth.   \nAutistic traits  \nAlthough associations between neonatal brain structure and autistic traits have been \npreviously been reported in the dHCP dataset (28), the present study reanalysed these data in \norder to (a) account for key sample differences to assess generalisability to a healthy, term-\nborn sample (i.e., the prior study included pre-term births, which were excluded from the \npresent research), and (b) investigate the potential role that sex and birth factors play in \nshaping this relationship. Consistent with the prior study, we observed that reduced brain \nvolumes were generally associated with higher autistic traits, confirming that this association \ngeneralises to term-born samples. While autism has traditionally been associated with \nincreased brain volumes, various explanations have been proposed to account for this \nobservation (28). For instance, mixed findings in the literature may reflect the considerable \nheterogeneity within the autism spectrum, where different subtypes may follow distinct \ndevelopmental trajectories. Additionally, studies have shown that autistic infants initially \npresent with smaller head circumferences at birth followed by accelerated brain growth later \nin infancy, indicating that “catch-up” effects may also play a role (41). \nMotor, language, and cognitive outcomes  \nAt the global level, larger cortical grey matter volumes were associated with higher scores on \ncognitive measures, replicating the well-established association between grey matter and \ncognition observed across various stages of development (42–44). Increased gray matter may \nreflect greater neuronal density, synaptic complexity, and dendritic arborisation – factors \nwhich may collectively impact cognitive processing capacities (45,46). At the regional level, \nconsistent with prior literature, positive associations were identified with the thalamus, insula, \ntemporal, and occipital lobes (47). These regions are broadly implicated in sensory \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nprocessing and integration, and their structural maturation may facilitate early cognitive \ndevelopment. Language outcomes showed widespread associations with total grey and white \nmatter, as well as with various temporal, parietal, frontal, occipital, and subcortical structures. \nAfter controlling for total brain volume, no significant associations emerged for either \ncognitive or language outcomes, indicating that the observed findings are primarily driven by \nglobal brain characteristics than by region-specific effects. \nWhile motor outcomes showed no significant global or regional associations with absolute \nbrain volumes, they were associated with reduced volumes in the brainstem and left \ncerebellum after controlling for total brain volume. These findings are unsurprising as both \nthe cerebellum and brainstem play well-established roles in motor control, movement, \nbalance, and coordination (48–50). As such, relatively increased volumes in these structures \nat birth may support motor development during infancy. These findings also indicate that \nindividual differences in motor development are driven by localised regional variations rather \nthan broad, global brain differences.  \nPsychopathological outcomes \nInternalising traits were negatively associated with absolute cerebellar volumes, a region \nfrequently implicated in psychopathology (51), and positively associated with frontal lobe \nvolumes after controlling for total brain volume. These associations did not, however, survive \na correction for multiple comparisons. There were no significant associations between \nexternalising traits and any global or regional brain volumes. It therefore appears that \nneonatal brain structure is a stronger predictor of developmental than psychopathological \noutcomes in toddlerhood. It is possible that meaningful associations may become more \napparent during later stages of development (i.e., adolescence), when mental health traits \nbecome more pronounced and begin to show greater variability (52). Additionally, since \nmental health outcomes are heavily shaped by brain-environment interactions (53,54), \nassociations with brain structure may be minimal prior to extensive environmental influence. \nGestational age at birth and birth weight \nAs expected, our findings showed that a greater gestational age at birth was linked to \nimproved language, cognitive, and motor outcomes. Importantly, brain volumes partially \nmediated these associations for cognitive outcomes, suggesting that gestational age at birth \ninfluences later development, in part, through its effects on early brain structure. Prior \nresearch has shown that, at term-equivalent age, infants born earlier within the term period \nhave reduced brain volumes compared to those born later (18). This is because the last \ntrimester of gestation is characterised by rapid cortical development (e.g., accelerated cortical \ngrowth, synaptogenesis, proliferation, myelination, and gyrification), marking it as a critical \nperiod for establishing the structural and functional foundations of the brain (5,6). Birth \noccurring even a few weeks earlier may interrupt these processes, leading to long-term \nalterations in brain structure and associated outcomes (13).   \nWe additionally identified that birth weight was associated with cognitive and language \noutcomes, and these relationships were also mediated by neonatal brain volumes. Both birth \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nweight and gestational age at birth are shown to have long-term developmental significance, \npredicting brain volumes and behavioural outcomes even later in life (Gonçalves et al., 2024.; \nMa et al., 2022; Silva et al., 2021). Our findings extend prior literature by demonstrating their \neffects are not independent but are rather partially mediated via neonatal brain volumes. It is \nalso worth noting that while birth factors predicted early language and cognitive outcomes, \nthey were not significantly associated with early internalising, externalising, or autistic traits.  \nSex differences  \nWe next examined whether sex plays a role in shaping early developmental outcomes via its \neffects on neonatal brain structure. Males showed higher externalising and autistic traits in \ntoddlerhood compared to females, though these associations did not survive a correction for \nmultiple comparisons. Females showed improved language outcomes in toddlerhood \ncompared to males – a finding that is consistent with a well-established developmental trend \nin which females exhibit an on-average advantage in language, social cognition, and \nempathising (23,57,58). Interestingly, mediation analyses revealed that brain volumes at birth \nsuppressed rather than mediated this relationship. While larger neonatal brain volumes were \ngenerally associated with improved language outcomes, females – despite on average having \nsmaller neonatal brain volumes (19) – showed higher language scores compared to males. \nAlthough seemingly counterintuitive, this pattern can be explained by compensatory \nmechanisms. It has been proposed that sex differences in the brain may serve not only to \ncreate behavioural differences, but also to prevent or mitigate them by balancing hormonal or \nphysiological factors that might otherwise amplify them further (59). According to this \nnotion, larger brain volumes in males may act protectively to reduce a language performance \ngap that might otherwise be even more pronounced. It is also possible that these sex \ndifferences are better explained by metrics beyond brain volumes (e.g., functional \nconnectivity or white matter microstructure) and vary depending on the developmental stage \nunder investigation.  \nStrengths and limitations \nTo the best of our knowledge, this study is among the few to examine associations between \nneonatal brain structure and a broad spectrum of outcomes in a large, representative, term-\nborn sample. This prospective, longitudinal design (i.e., brain imaging at birth and outcomes \nassessed at 18 months) enabled us to investigate how early neuroanatomy shapes future \nbehavioural, developmental, and socioemotional outcomes. While causality cannot be \nassumed, the design offers some insight into the directionality and temporal sequence of these \nassociations. The large sample size also provides adequate statistical power to detect \nmeaningful effects across multiple domains, while the focus on a non-clinical, term-born \ncohort provides insights into general, population-level variability in development.  \nSeveral limitations should also be considered. First, brain structure was assessed at a single \ntimepoint, which fails to capture the rapid and dynamic changes that occur across the first \nfew years of life. It is possible that infants with initially reduced brain volumes may \nexperience relative overgrowth, while those with initially increased volumes may experience \nundergrowth. These developmental trajectories, rather than brain structure at a single point in \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\ntime, may serve as more informative predictors of future behavioural outcomes (V anes et al., \n2023). Additionally, it is important to note that brain–behaviour associations are likely stage-\nspecific and may vary, or become stronger, when both measurements are more temporally \naligned (61). Secondly, the follow-up was limited to one timepoint in toddlerhood, and many \nof the studied traits continue to evolve beyond 18 months. Longitudinal studies incorporating \nrepeated brain and behaviour measurements may provide a richer developmental account and \ndemonstrate how changes in brain morphology over time relate to emerging behaviour. \nThirdly, while we focused primarily on volumetric measures, other imaging modalities (e.g., \nwhite matter microstructure, functional connectivity) may serve as more sensitive predictors \nof behavioural outcomes. Finally, while perinatal factors such as gestational age and birth \nweight were considered, a range of additional environmental and psychosocial factors (e.g., \nparenting quality, maternal mental health, and socioeconomic status) were not assessed and \nmay serve as important mediators or moderators. Future research may consider incorporating \na wider array of imaging techniques and environmental variables to more comprehensively \nmap early brain-behaviour associations.   \nConclusion \nCollectively, these findings demonstrate a broad pattern in which larger neonatal brain \nvolumes are associated with higher scores on cognitive, behavioural, psychopathological, and \nneurodevelopmental measures in toddlerhood. Additionally, neonatal brain structure appears \nto hold domain-specific predictive value, showing stronger associations with certain \noutcomes (e.g., language, cognition, motor and autistic traits) compared to others (e.g., \ninternalising and externalising traits). Associations with these latter traits may become more \npronounced later in development, as individual variability increases and brain–environment \ninteractions become more influential. Our findings further demonstrate that birth factors, such \nas gestational age weight, influence future outcomes in part through their impact on early \nbrain structure, reinforcing the developmental significance of late gestational development \nand fetal growth. Finally, sex differences may influence brain–behaviour associations through \na multifaceted interplay of mediatory and compensatory mechanisms which may not be fully \ncaptured by brain volumes alone. Collectively, these findings provide further evidence that \nneonatal brain structure and birth factors are important markers for shaping future \nbehavioural development. \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\n \nReferences \n1. Alex AM, Buss C, Davis EP, Campos G de L, Donald KA, Fair DA, et al. (2023): Genetic \ninfluences on the developing young brain and risk for neuropsychiatric disorders. Biol \nPsychiatry 93(10):905–920. \n2. de Mendonça Filho EJ, Barth B, Bandeira DR, de Lima RMS, Arcego DM, Dalmaz C, et \nal. (2021): Cognitive development and brain gray matter susceptibility to prenatal adversities: \nmoderation by the prefrontal cortex brain-derived neurotrophic factor gene co-expression \nnetwork. Front Neurosci 15:744743. Available from: \nhttps://www.frontiersin.org/articles/10.3389/fnins.2021.744743/full \n3. Gilmore JH, Knickmeyer RC, Gao W. (2018): Imaging structural and functional brain \ndevelopment in early childhood. Nat Rev Neurosci 19(3):123–137. \n4. Raznahan A, Greenstein D, Lee NR, Clasen LS, Giedd JN. (2012): Prenatal growth in \nhumans and postnatal brain maturation into late adolescence. Proc Natl Acad Sci U S A \n109(28):11366–11371. \n5. Ackerman S. (1992): The development and shaping of the brain. Discov Brain 308:221–\n225. \n6. Estrin GL, Bhavnani S. (2020): Brain development: structure. [Book section; publisher not \nspecified.] \n7. Khan YT, Tsompanidis A, Radecki MA, Allison C, Lai MC, Bethlehem RAI, et al. (2025): \nMapping brain growth and sex differences across prenatal to postnatal development. bioRxiv \n2025.03.17.643640. Available from: \nhttps://www.biorxiv.org/content/10.1101/2025.03.17.643640v1\n \n8. Vanes L, Fenn-Moltu S, Hadaya L, Fitzgibbon S, Cordero-Grande L, Price A, et al. (2023): \nLongitudinal neonatal brain development and socio-demographic correlates of infant \noutcomes following preterm birth. Dev Cogn Neurosci 61:101250. \n9. Aydin E, Tsompanidis A, Chaplin D, Hawkes R, Allison C, Hackett G, et al. (2024): Fetal \nbrain growth and infant autistic traits. Mol Autism 15(1):11. \n10. Rogers CE, Anderson PJ, Thompson DK, Kidokoro H, Wallendorf M, Treyvaud K, et al. \n(2012): Regional cerebral development at term relates to school-age social-emotional \ndevelopment in very preterm children. J Am Acad Child Adolesc Psychiatry 51(2):181–191. \n11. Loh WY , Anderson PJ, Cheong JL Y , Spittle AJ, Chen J, Lee KJ, et al. (2017): Neonatal \nbasal ganglia and thalamic volumes: very preterm birth and 7-year neurodevelopmental \noutcomes. Pediatr Res 82(6):970–978. \n12. Boyle EM, Poulsen G, Field DJ, Kurinczuk JJ, Wolke D, Alfirevic Z, et al. (2012): Effects \nof gestational age at birth on health outcomes at 3 and 5 years of age: population based cohort \nstudy. BMJ 344:e896. Available from: https://www.bmj.com/content/344/bmj.e896.short\n \n13. Gale-Grant O, Fenn-Moltu S, França L, Dimitrova R, Christaens D, Cordero-Grande L, et \nal. (2021): Effects of gestational age at birth on perinatal structural brain development in \nhealthy term-born babies. bioRxiv 2021.05.02.442327. Available from: \nhttps://www.biorxiv.org/content/10.1101/2021.05.02.442327v2\n \n14. Sun Z, Zhang L, Ge Q, Xiao C, Gale-Grant O, Falconer S, et al. (2025): Gestational age \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nat birth and cognitive outcomes in term-born children: evidence from Chinese and British \ncohorts. Early Hum Dev 204:106237. \n15. Cortese M, Moster D, Wilcox AJ. (2021): Term birth weight and neurodevelopmental \noutcomes. Epidemiology 32(4):583–590. \n16. Gonçalves R, Gaillard R, Ferguson KK, Sammallahti S, Hillegers MH, Steegers EAP, et \nal. (2024): Associations of fetal and infant growth patterns with behavior and cognitive \noutcomes in early adolescence. JCPP Adv e12278. \n17. Pettersson E, Larsson H, D’Onofrio B, Almqvist C, Lichtenstein P . (2019): Association of \nfetal growth with general and specific mental health conditions. JAMA Psychiatry 76(5):536–\n543. \n18. Knickmeyer RC, Xia K, Lu Z, Ahn M, Jha SC, Zou F, et al. (2017): Impact of \ndemographic and obstetric factors on infant brain volumes: a population neuroscience study. \nCereb Cortex 27(12):5616–5625. \n19. Khan YT, Tsompanidis A, Radecki MA, Dorfschmidt L, Adhya D, Ayeung B, et al. \n(2024): Sex differences in human brain structure at birth. Biol Sex Differ 15(1):81. \n20. Khan YT, Tsompanidis A, Radecki MA, Allison C, Lai MC, Bethlehem RAI, et al. \n(2025): Mapping brain growth and sex differences across prenatal to postnatal development. \nbioRxiv 2025.03.17.643640. Available from: \nhttps://www.biorxiv.org/content/10.1101/2025.03.17.643640v1 \n21. Connellan J, Baron-Cohen S, Wheelwright S, Batki A, Ahluwalia J. (2000): Sex \ndifferences in human neonatal social perception. Infant Behav Dev 23(1):113–118. \n22. Adani S, Cepanec M. (2019): Sex differences in early communication development: \nbehavioral and neurobiological indicators of more vulnerable communication system \ndevelopment in boys. Croat Med J 60(2):141–149. \n23. Dinkel D, Snyder K. (2020): Exploring gender differences in infant motor development \nrelated to parents’ promotion of play. Infant Behav Dev 59:101440. \n24. Etchell A, Adhikari A, Weinberg LS, Choo AL, Garnett EO, Chow HM, et al. (2018): A \nsystematic literature review of sex differences in childhood language and brain development. \nNeuropsychologia 114:19–31. \n25. Krogh MT, Væver MS. (2019): Does gender affect Bayley-III scores and test-taking \nbehavior? Infant Behav Dev 57:101352. \n26. Adani S, Cepanec M. (2019): Sex differences in early communication development: \nbehavioral and neurobiological indicators of more vulnerable communication system \ndevelopment in boys. Croat Med J 60(2):141–149. \n27. Mundy P, Block J, Delgado C, Pomares Y , Vaughan Van Hecke A, Parlade MV . (2007): \nIndividual differences and the development of joint attention in infancy. Child Dev \n78(3):938–954. \n28. Allison C, Baron-Cohen S, Wheelwright S, Charman T, Richler J, Pasco G, et al. (2008): \nThe Q-CHAT (Quantitative Checklist for Autism in Toddlers): a normally distributed \nquantitative measure of autistic traits at 18–24 months of age: preliminary report. J Autism \nDev Disord 38(8):1414–1425. \n29. Tsompanidis A, Chang K, Khan Y , Radecki M, Dorfschmidt L, Hampton S, et al. (2025): \nPerinatal brain growth and autistic traits in toddlers. Research Square. Available from: \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nhttps://www.researchsquare.com/article/rs-5462151/v1 \n30. Edwards AD, Rueckert D, Smith SM, Abo Seada S, Alansary A, Almalbis J, et al. (2022): \nThe Developing Human Connectome Project neonatal data release. Front Neurosci \n16:886772. Available from: \nhttps://www.frontiersin.org/articles/10.3389/fnins.2022.886772/full \n31. Dubois J, Dehaene-Lambertz G, Kulikova S, Poupon C, Hüppi PS, Hertz-Pannier L. \n(2014): The early development of brain white matter: a review of imaging studies in fetuses, \nnewborns and infants. Neuroscience 276:48–71. \n32. Gilmore JH, Shi F, Woolson SL, Knickmeyer RC, Short SJ, Lin W, et al. (2012): \nLongitudinal development of cortical and subcortical gray matter from birth to 2 years. Cereb \nCortex 22(11):2478–2485. \n33. Holland D, Chang L, Ernst TM, Curran M, Buchthal SD, Alicata D, et al. (2014): \nStructural growth trajectories and rates of change in the first 3 months of infant brain \ndevelopment. JAMA Neurol 71(10):1266–1274. \n34. Knickmeyer RC, Gouttard S, Kang C, Evans D, Wilber K, Smith JK, et al. (2008): A \nstructural MRI study of human brain development from birth to 2 years. J Neurosci \n28(47):12176–12182. \n35. Ouyang M, Kang H, Detre JA, Roberts TPL, Huang H. (2017): Short-range connections \nin the developmental connectome during typical and atypical brain maturation. Neurosci \nBiobehav Rev 83:109–122. \n36. Ball G, Aljabar P, Arichi T, Tusor N, Cox D, Merchant N, et al. (2016): Rich-club \norganization of the newborn human brain. Proc Natl Acad Sci U S A 111(20):7456–7461. \n37. Alexander-Bloch A, Raznahan A, Bullmore E, Giedd J. (2013): The convergence of \nmaturational change and structural covariance in human cortical networks. J Neurosci \n33(7):2889–2899. \n38. Fitzgibbon SP, Harrison SJ, Jenkinson M, Baxter L, Robinson EC, Bastiani M, et al. \n(2020): The Developing Human Connectome Project (dHCP) automated resting-state \nfunctional processing framework for newborn infants. NeuroImage 223:117303. \n39. Eyre M, Fitzgibbon SP, Ciarrusta J, Cordero-Grande L, Price AN, Poppe T, et al. (2021): \nThe Developing Human Connectome Project: typical and disrupted perinatal functional \nconnectivity. Brain 144(7):2199–2213. \n40. Smyser CD, Inder TE, Shimony JS, Hill JE, Degnan AJ, Snyder AZ, et al. (2010): \nLongitudinal analysis of neural network development in preterm infants. Cereb Cortex \n20(12):2852–2862. \n41. Cao M, He Y , Dai Z, Liao X, Jeon T, Ouyang M, et al. (2017): Early development of \nfunctional network segregation revealed by connectomic analysis of the preterm human brain. \nCereb Cortex 27(3):1949–1963. \n42. Turk E, van den Heuvel MI, Benders MJ, de Heus R, Franx A, Manning JH, et al. (2019): \nFunctional connectome of the fetal brain. J Neurosci 39(49):9716–9724. \n43. van den Heuvel MI, Turk E, Manning JH, Hect JL, Hernandez-Andrade E, Hassan SS, et \nal. (2020): Hubs in the human fetal brain network. Dev Cogn Neurosci 42:100737. \n44. Thomason ME, Dassanayake MT, Shen S, Katkuri Y , Alexis M, Anderson AL, et al. \n(2013): Cross-hemispheric functional connectivity in the human fetal brain. Sci Transl Med \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\n5(173):173ra24. \n45. Scheinost D, Spann MN, McDonough L, Peterson BS, Monk C. (2021): Associations \nbetween maternal basal ganglia and fetal brain development. Biol Psychiatry 89(2):122–130. \n46. Clouchoux C, Kudelski D, Gholipour A, Warfield SK, Viseur S, Bouyssi-Kobar M, et al. \n(2012): Quantitative in vivo MRI measurement of cortical development in the fetus. Brain \nStruct Funct 217(1):127–139. \n47. Heringa SM, Jacobusse G, Kan KJ, van den Heuvel MP, Benders MJ, van den Heuvel MI. \n(2024): Functional connectome of the human fetal brain: a systematic review. Cereb Cortex \n34(14):6364–6376. \n48. Thomason ME, Brown JA, Dassanayake MT, Shastri R, Marusak HA, Hernandez-\nAndrade E, et al. (2014): Intrinsic functional brain architecture derived from graph theoretical \nanalysis in the human fetus. PLoS One 9(5):e94423. \n49. van den Heuvel MI, Thomason ME. (2016): Functional connectivity of the human brain \nin utero. Trends Cogn Sci 20(12):931–939. \n50. Scheinost D, Kwon SH, Shen X, Lacadie C, Schneider KC, Dai F, et al. (2016): Preterm \nbirth alters neonatal, functional rich club organization. Brain Struct Funct 221(6):3211–3222. \n51. van den Heuvel MI, Kersbergen KJ, de Reus MA, Keunen K, Kahn RS, Groenendaal F, et \nal. (2015): The neonatal connectome during preterm brain development. Cereb Cortex \n25(9):3000–3013. \n52. Brown CJ, Miller SP, Booth BG, Andrews S, Chau V , Poskitt KJ, et al. (2014): Structural \nnetwork analysis of brain development in young preterm neonates. NeuroImage 101:667–\n680. \n53. Batalle D, Hughes EJ, Zhang H, Tournier JD, Tusor N, Aljabar P, et al. (2017): Early \ndevelopment of structural networks and the impact of prematurity on brain connectivity. \nNeuroImage 149:379–392. \n54. He L, Li H, Holland SK, Y uan W. (2019): Gender differences in structural connectome \ndevelopment across the human lifespan. NeuroImage 203:116033. \n55. Zhao T, Xu Y , He Y . (2022): Graph theoretical modeling of baby brain networks. \nNeuroImage 257:119312. \n56. Thomason ME, Scheinost D, Manning JH, Grove LE, Hect J, Marshall N, et al. (2017): \nWeak functional connectivity in the human fetal brain prior to preterm birth. Sci Rep \n7(1):39286. \n57. Turk E, van den Heuvel MI, Benders MJ, de Heus R, Franx A, Manning JH, et al. (2019): \nFunctional connectome of the fetal brain. J Neurosci 39(49):9716–9724. \n58. Fitzgibbon SP, Harrison SJ, Jenkinson M, Baxter L, Robinson EC, Bastiani M, et al. \n(2020): The Developing Human Connectome Project (dHCP) automated resting-state \nfunctional processing framework for newborn infants. NeuroImage 223:117303. \n59. Ouyang M, Kang H, Detre JA, Roberts TPL, Huang H. (2017): Short-range connections \nin the developmental connectome during typical and atypical brain maturation. Neurosci \nBiobehav Rev 83:109–122. \n60. van den Heuvel MI, Thomason ME. (2016): Functional connectivity of the human brain \nin utero. Trends Cogn Sci 20(12):931–939. \n61. Smyser CD, Inder TE, Shimony JS, Hill JE, Degnan AJ, Snyder AZ, et al. (2010): \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nLongitudinal analysis of neural network development in preterm infants. Cereb Cortex \n20(12):2852–2862. \n62. van den Heuvel MI, Kersbergen KJ, de Reus MA, Keunen K, Kahn RS, Groenendaal F, et \nal. (2015): The neonatal connectome during preterm brain development. Cereb Cortex \n25(9):3000–3013. \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nAcknowledgement \nWe would like to thank Dr Kamen Tsvetanov for providing statistical advice on the mediation \nmodels. We would also like to thank Eden Hymanson, Sehanya Wickramanayake, and Manya \nGupta for assisting with organising the supplementary materials.  \nThese results were obtained using data made available from the Developing Human \nConnectome Project funded by the European Research Council under the European Union’s \nSeventh Framework Programme (FP/2007-2013) / ERC Grant Agreement no. [319456]. Data \nused in the preparation of this manuscript were obtained from the National Institute of Mental \nHealth (NIMH) Data Archive (NDA). NDA is a collaborative informatics system created by \nthe National Institutes of Health to provide a national resource to support and accelerate \nresearch in mental health. Dataset identifier(s): 3995. This manuscript reflects the views of \nthe authors and may not reflect the opinions or views of the NIH or of the Submitters \nsubmitting original data to NDA.  \nY .T.K is supported by the Cambridge Trust and Trinity College, Cambridge.  SBC received \nfunding from the Wellcome Trust 214322\\Z\\18\\Z. For the purpose of Open Access, the author \nhas applied a CC BY public copyright licence to any Author Accepted Manuscript version \narising from this submission. SBC also received funding from the Innovative Medicines \nInitiative 2 Joint Undertaking under grant agreement No 777394 for the project AIMS-2-\nTRIALS. This Joint Undertaking receives support from the European Union's Horizon 2020 \nresearch and innovation programme and EFPIA and AUTISM SPEAKS, Autistica, \nSFARI. SBC also received funding from Autism Action, SFARI, the Templeton World \nCharitable Fund and the MRC.  The funders had no role in the design of the study; in the \ncollection, analyses, or interpretation of data; in the writing of the manuscript, or in the \ndecision to publish the results. Any views expressed are those of the author(s) and not \nnecessarily those of the funders (including IHI-JU2). All research at the Department of \nPsychiatry in the University of Cambridge is supported by the NIHR Cambridge Biomedical \nResearch Centre (NIHR203312) and the NIHR Applied Research Collaboration East of \nEngland. The views expressed are those of the author(s) and not necessarily those of the \nNIHR or the Department of Health and Social Care. \nDisclosures \nR.A.I.B. is a director of and holds equity in Centile Bioscience Ltd. All other authors declare \nthat they have no competing interests. \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nTables \nTable 1. Sample Demographic Characteristics  \nValues represent means and, in brackets, standard deviations.  \na p-values reflect sex differences assessed by independent-samples t-tests. \n Mean (SD) - \nAll  \nMean (SD) - \nMales \nMean (SD) - \nFemales \npa \nGestational age \nat scan (weeks) \n41.39 (1.64) 41.23 (1.61) 41.55 (1.66) \n \n0.270 \nPostnatal age \nat scan (days) \n8.55 (8.10) 7.88 (7.72) 9.25 (8.44) 0.095 \nAge at follow-\nup (months) \n18.98 (1.99) 18.84 (2.02) 19.11 (1.95) 0.162 \nGestational age \nat birth \n(weeks) \n40.17 (1.15) 40.11 (1.15) 40.24 (1.15) 0.270 \nBirth weight \n(kilograms) \n3.44 (0.47) 3.48 (0.45) 3.40 (0.49) 0.089 \nMaternal age \nat birth (years) \n33.84 (4.67) 33.91 (4.40) 33.77 (4.95) 0.756 \nPaternal age at \nbirth (years) \n36.18 (5.94) 36.22 (6.10) 36.14 (5.78) 0.816 \n \nTable 2. Pairwise correlations between behavioural outcomes at age 18 months \nCorrelation matrix displaying associations between behavioural outcomes at 18 months. \nValues represent Pearson correlation coefficients and asterisks indicate significance levels (*p \n< .05, **p < .01, ***p < .001). \n Q-CHAT Externalising Internalising Cognitive Language Motor \nQ-CHAT - 0.21*** 0.39*** -0.32*** -0.49*** -0.22*** \nExternalising 0.21*** - 0.54*** -0.00 -0.06 -0.00 \nInternalising 0.39*** 0.54*** - -0.21*** -0.20*** -0.12* \nCognitive -0.32*** -0.00 -0.21*** - 0.59*** 0.53*** \nLanguage -0.49*** -0.06 -0.20*** 0.59*** - 0.48*** \nMotor -0.22*** -0.00 -0.12* 0.53*** 0.48*** - \n \nTable 3. Regional associations with autistic traits  \nAssociations between brain volumes at birth and autistic traits at 18-month follow-up. Linear \nregression coefficients (β ; both standardised and und), standard errors (SE), and false \ndiscovery rate–corrected p-values (pFDR) are reported for each region. \nRegion Unstandardised \nβ  Coefficient \nStandardised \nβ  Coefficient \nSE pFDR \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nTotal brain volume -0.000 0.000 0.000 0.027 \nTotal subcortical \ngray matter \n-0.001 0.000 0.000 0.031 \nTotal white matter -0.000 0.000 0.003 0.032 \nBrainstem -0.002 -0.001 0.001 0.042 \nRight thalamus -0.003 0.002 0.001 0.049 \nLeft thalamus -0.004 0.002 0.001 0.031 \nRight lentiform \nnucleus \n-0.004 0.001 0.001 0.031 \nRight \nparahippocampal \ngyrus (anterior \npart) \n-0.009 0.000 0.003 0.032 \nLeft \nparahippocampal \ngyrus (anterior \npart) \n-0.008 0.002 0.003 0.049 \nMedial and inferior \ntemporal gyri \n(anterior part) \n-0.003 0.001 0.001 0.036 \nRight medial and \ninferior temporal \ngyri (posterior \npart) \n-0.003 0.001 0.001 0.024 \nLeft medial and \ninferior temporal \ngyri (posterior \npart) \n-0.003 0.001 0.001 0.013   \n \nTable 4. Associations with Bayley-III cognitive subscale \nAssociations between brain volumes at birth and cognitive outcomes at 18-month follow-up. \nLinear regression coefficients (β ; both standardised and unstandardised), standard errors (SE), \nand false discovery rate–corrected p-values (pFDR) are reported for each region. \nRegion Unstandardised \nβ  Coefficient \nSE Standardised \nβ  Coefficient \npFDR \nCortical gray \nmatter \n<0.001 <0.001 0.000 0.050 \nLeft thalamus 0.004 0.002 0.004 0.035 \nRight occipital lobe 0.001 0.000 0.001 0.017 \nLeft lateral \noccipitotemporal \ngyrus  \n0.008 0.003 0.008 \n \n0.050 \nRight medial and \ninferior temporal \ngyrus \n0.004 0.002 0.004 0.034 \nRight 0.009 0.004 0.010 0.042 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\nparahippocampal \ngyrus (anterior \npart) \n \nTable 5. Associations with Bayley-III language subscale \nAssociations between brain volumes at birth and language outcomes at 18-month follow-up. \nLinear regression coefficients (β ; both standardised and unstandardised), standard errors (SE), \nand false discovery rate–corrected p-values (pFDR) are reported for each region. \nRegion Unstandardised \nβ  Coefficient \nSE \n \nStandardised \nβ  Coefficient \npFDR \nTotal brain volume 0.000 0.000 0.000 0.013 \nCortical gray \nmatter \n0.000 0.000 0.000 0.030 \nSubcortical gray \nmatter \n0.001 0.000 0.001 0.020 \nWhite matter 0.000 0.000 0.000 0.010 \nLeft hippocampus 0.021 0.008 0.020 0.030 \nRight cerebellum 0.001 0.000 0.001 0.048 \nLeft amygdala 0.039 0.016 0.041 0.033 \nRight amygdala 0.032 0.013 0.034 0.036 \nRight caudate \nnucleus \n0.01 0.003 0.010 0.011 \nLeft caudate \nnucleus \n0.009 0.003 0.009 0.025 \nRight thalamus 0.005 0.002 0.006 0.049 \nLeft thalamus 0.006 0.002 0.006 0.028 \nLeft lentiform \nnucleus \n0.006 0.002 0.006 0.047 \nRight \nparahippocampal \ngyrus (anterior \npart) \n0.015 0.005 0.015 0.030 \nLeft \nparahippocampal \ngyrus (anterior \npart) \n0.014 0.005 0.014 0.031 \nLeft medial and \ninferior temporal \ngyrus (anterior \npart) \n0.006 0.002 0.006 0.031 \nRight medial and \ninferior temporal \ngyrus (anterior \npart) \n0.006 0.002 0.006 0.025 \nLeft lateral \noccipitotemporal \n0.016 0.007 0.016 0.035 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\ngyrus (anterior \npart) \nRight lateral \noccipitotemporal \ngyrus (anterior \npart) \n0.014 0.006 0.015 0.038 \nRight insula 0.012 0.004 0.011 0.014 \nLeft insula 0.011 0.004 0.011 0.011 \nRight occipital lobe 0.001 0.001 0.001 0.041 \nRight medial and \ninferior temporal \ngyri (posterior part) \n0.004 0.002 0.004 0.022 \nLeft medial and \ninferior temporal \ngyrus (posterior \npart) \n0.004 0.002 0.004 0.035 \nRight posterior \ncingulate gyrus \n0.008 0.003 0.007 0.031 \nLeft posterior \ncingulate gyrus \n0.008 0.003 0.008 0.035 \nLeft frontal lobe 0.001 0.000 0.001 0.038 \nLeft parietal lobe 0.001 0.000 0.001 0.019 \n \nTable 6. Regional associations with Bayley-III motor subscale after controlling for total \nbrain volume.   \nAssociations between brain volumes at birth and language outcomes at 18-month follow-up \nafter controlling for total brain volume. Linear regression coefficients (β ; both standardised \nand unstandardised), standard errors (SE), and false discovery rate–corrected p-values (pFDR) \nare reported for each region. \nRegion Unstandardised \nβ  Coefficient \nSE Standardised \nβ  Coefficient \npFDR \nBrainstem -0.004 0.002 -0.286 0.039 \nLeft Cerebellum  -0.002 0.001 -0.348 0.025 \nVentricles -0.001 0.000 -0.184 0.006 \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint \n\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 2, 2025. ; https://doi.org/10.1101/2025.11.01.686012doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}