Altered corpus callosum development and corpus callosum-cerebellar spatial relationships in preterm infants at term-equivalent age Authors: Xiao-Ying Qi, Hou-Qing Pang, Ling Wang, Xi-Yue Zhang, Yi-Fei Tan, Hong Luo, Yan Luo Published: 2026 ABSTRACT AbstractBackgroundPreterm (PT) birth disrupts critical third-trimester brain maturation, particularly affecting the corpus callosum (CC) and cerebellum. As interconnected components of cerebro-cerebellar networks, their developmental trajectories and spatial relationships remain incompletely understood. This study used serial cranial ultrasonography to characterize these alterations from birth to term-equivalent age (TEA).MethodsThis prospective observational study included 117 neonates, comprising 60 term infants and 57 PT infants. A total of 719 cranial ultrasound examinations were analyzed. PT infants underwent serial weekly ultrasonography from birth to 40 weeks’ corrected gestational age (CGA), whereas term infants were examined within the same TEA window. CC morphometry, including CC length (CCL), CC curve length (CCCL), height, area, and thickness, CC-fastigium length (CCFL), CC-fastigium angle (CCFA), and vermis morphometry were measured. Developmental trajectories were assessed using nonlinear regression, segmented linear regression, and linear mixed-effects models. Multivariable regression was used to identify factors independently associated with morphometric outcomes.ResultsAt TEA, the PT infants had shorter CCL (39.8±2.3 vs. 42.5±3.8 mm, adjusted P=0.015), shorter CCCL (53.7±6.0 vs. 68.5±17.4 mm, adjusted P=0.004), smaller vermis area (400.3±94.6 vs. 455.9±81.1 mm2, adjusted P=0.041), shorter CCFL (42.1±10.3 vs. 48.6±3.0 mm, adjusted P<0.001), and larger CCFA (43.1±6.8° vs. 33.6±4.5°, adjusted P<0.001) than the term controls. In the PT infants, segmented regression showed rapid CCL growth before 35 weeks’ CGA (1.04 mm/week, P<0.001), followed by no significant linear growth thereafter (P=0.059). CCFL continued to increase after 35 weeks but at a slower rate (0.55 mm/week, P<0.001), whereas CCFA showed an exploratory nonlinear trajectory. In the multivariable analysis, PT birth was independently associated with shorter CCL [β=−2.30, 95% confidence interval (CI): −4.15 to −0.46, P=0.015] and larger CCFA (β=7.04, 95% CI: 3.60–10.48, P<0.001).ConclusionsPT birth is associated with quantitatively measurable alterations in CC development, cerebellar vermis size, and CC-cerebellar spatial configuration at TEA. Serial cranial ultrasonography captured divergent postnatal growth trajectories, including a late-PT inflection in CC growth and persistent enlargement of CCFA. These findings support ultrasound-based longitudinal morphometric monitoring of postnatal brain development in PT infants.Keywords: Preterm infant (PT infant), corpus callosum (CC), cerebellar vermis, cranial ultrasonography, corpus callosum-cerebellar spatial relationships (CC-cerebellar spatial relationships) Introduction Despite improved survival rates, preterm (PT) birth interrupts the critical third trimester, a period characterized by rapid organ development, accelerated white matter maturation, and extensive synaptic organization (1,2). This disruption renders the PT brain particularly vulnerable to white matter injury (WMI), a major pathological consequence of prematurity and an important factor affecting subsequent brain maturation. As the largest commissural white matter tract in the brain, the corpus callosum (CC) is especially susceptible to prematurity-related injury. A substantial body of evidence indicates that alterations in CC morphology in PT infants primarily reflect underlying WMI rather than isolated structural abnormalities (3,4). This vulnerability is largely attributable to disrupted pre-oligodendrocyte maturation, a critical process for subsequent myelination, as well as diffuse or cystic WMI (5). Consequently, reduced CC size and altered morphometric features are increasingly regarded as imaging biomarkers of white matter volume and microstructural integrity (3,6). These morphometric alterations should be distinguished from major congenital anomalies such as CC agenesis, a distinct developmental condition associated with heterogeneous neurodevelopmental outcomes, including cases with preserved cognitive function (7,8). The present study focused on subtle variations in CC biometry in structurally intact brains. These quantitative changes are more likely to reflect disturbances in white matter maturation associated with prematurity rather than primary malformations. In parallel with supratentorial white matter development, the cerebellum—particularly the vermis—undergoes rapid growth during the third trimester and is similarly vulnerable to prematurity-related insults (9). Supratentorial lesions may also affect cerebellar vermis growth (10). Further, the CC and cerebellum are functionally linked through cerebro-cerebellar networks that support motor coordination, cognition, and higher-order neurodevelopmental functions (11). However, the CC and cerebellar vermis differ fundamentally in their embryological origins and mechanisms of injury: the CC is a telencephalic commissural structure, whereas the cerebellum originates from the rhombencephalon. Accordingly, alterations in the development of these structures should not be interpreted as manifestations of a shared pathological process, but rather as distinct, yet potentially interacting, components of brain maturation. Existing studies have characterized CC development across both the fetal and postnatal periods. Early magnetic resonance imaging (MRI)-based investigations have demonstrated altered CC size and growth trajectories in PT infants (12-14), while ultrasound-based studies have provided accessible bedside assessments of neonatal brain development (15,16). Recent studies have further extended cranial ultrasonography beyond simple linear measurements to enable estimation of global brain volume and longitudinal growth trajectories in PT infants, thereby supporting its utility for serial assessment of postnatal brain development (17,18). Despite these advances, most studies have focused on isolated structural measurements, with limited attention to spatial relationships between different brain regions. In this context, the spatial relationship between the CC and cerebellum may provide additional insights into coordinated brain development beyond single-structure metrics. Quantitative distance- and angle-based measures describing the relative positioning of supratentorial and infratentorial structures may provide complementary information beyond conventional morphometric assessments. While CC biometry primarily reflects white matter integrity, such geometric measures may capture aspects of coordinated brain maturation that remain poorly characterized in PT infants. Therefore, this study aimed to quantitatively evaluate the CC and its spatial relationship with the cerebellar vermis using serial cranial ultrasonography. By characterizing developmental trajectories from birth to term-equivalent age (TEA), this study sought to identify morphometric features associated with altered postnatal brain development in PT infants. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0473/rc). Methods Study participants The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of West China Second University Hospital (No. 2024-338). Written informed consent was obtained from the parents or legal guardians of all participants. This prospective cohort study included singleton neonates born at West China Second University Hospital, a tertiary referral neonatal center with Level III neonatal intensive care capabilities, between November 2024 and September 2025. The center provides comprehensive care for PT and critically ill neonates, including invasive and non-invasive respiratory support. Infants were included if they were appropriate for gestational age (GA) at birth, defined as having a birth weight between the 10th and 90th percentiles (19), and had no major structural abnormalities. GA and corrected gestational age (CGA) were calculated based on crown-rump length measurements obtained during the first trimester. Infants were categorized into term (≥37 weeks) and PT (<37 weeks) groups based on GA at birth. PT infants were further subclassified as extremely PT (<28 weeks), very PT (28+0 to 31+6 weeks), moderate PT (32+0 to 33+6 weeks), and late PT (34+0 to 36+6 weeks). The exclusion criteria were as follows: (I) major congenital defects, particularly central nervous system or other significant malformations; (II) neonatal intracranial infections or severe neonatal complications, including severe intraventricular hemorrhage (IVH; grade III/IV), periventricular leukomalacia (PVL), necrotizing enterocolitis (NEC), culture-proven neonatal sepsis, or moderate-to-severe bronchopulmonary dysplasia; (III) confirmed chromosomal abnormalities; and/or (IV) incomplete maternal or neonatal clinical data or suboptimal image quality for neurosonographic analysis. The sample size was estimated based on previously published studies reporting differences in CC morphometry between PT and term infants (3,6). Assuming a moderate effect size (Cohen’s d=0.5), a two-sided alpha level of 0.05, and a statistical power of 80%, approximately 50 participants were required per group. Considering potential attrition and incomplete follow-up, we aimed to recruit approximately 60 neonates per group. PT infants received standardized nutritional and clinical management according to institutional neonatal intensive care protocols. Clinical data, including primary admission indications, respiratory support, and major postnatal morbidities, were systematically collected from electronic records to characterize the study cohorts. Cranial ultrasound examination TEA was defined as 40 weeks’ CGA, with an allowable imaging window of ±1 day. PT infants underwent serial cranial ultrasonography approximately weekly from birth until TEA. If PT infants were discharged before 40 weeks’ CGA, weekly outpatient follow-up examinations were continued until the TEA scan was completed. Term control infants underwent 1–3 cranial ultrasonography examinations after birth, and the examination performed within the TEA window was used for cross-sectional comparison. Morphometric comparisons between PT and term infants were based exclusively on measurements obtained during the TEA examination. Neonates were positioned supine, and scanning was performed via the anterior fontanel acoustic window using a Resona M10 system (Mindray, Shenzhen, China) equipped with a C11–3S neonatal brain probe (frequency range: 3–11 MHz). Comprehensive multiplanar imaging, including standardized midsagittal and coronal views, was obtained to visualize the CC, cerebellar vermis, and other key intracranial structures. Cranial ultrasonography was performed by two experienced sonographers, each with more than 5 years of experience in neonatal neurosonography, according to standardized protocols. To ensure measurement accuracy, only high-quality images that clearly and completely displayed the CC, cerebellar vermis, cavum septi pellucidum, thalamus, and pons in the standardized midsagittal and coronal planes were included for morphometric analysis. Two examiners blinded to the clinical data independently performed offline measurements using ImageJ software (version 1.53, National Institutes of Health, USA) on images stored in Digital Imaging and Communications in Medicine format. Inter- and intra-observer reproducibility were assessed using intraclass correlation coefficients (ICCs) based on repeated measurements obtained from all enrolled neonates. Multiple CC parameters were obtained in the midsagittal plane, including CC length (CCL), CC curve length (CCCL), CC height, segmental CC thicknesses, total CC area, CC-fastigium length (CCFL), and CC-fastigium angle (CCFA). The midsagittal CC area was further subdivided into six anatomical regions [genu, rostral body (RB), anterior mid-body (AMB), posterior mid-body (PMB), isthmus, and splenium] according to a previously published neonatal MRI-based geometric parcellation scheme described by Thompson et al. (20). Briefly, five perpendicular lines were placed at intervals of 1/6, 1/3, 1/2, 2/3, and 4/5 along the anterior-posterior axis of the CC, generating six subregions. CCL was defined as the linear distance between the most anterior margin of the genu and the most posterior margin of the splenium. CC height was defined as the perpendicular distance from the midpoint of the CCL to the superior border of the CC. CCFL was defined as the linear distance between the most anterior margin of the CC genu and the fastigium of the fourth ventricle (21). CCFA was defined as the angle formed at the intersection of the CCL and CCFL (22). Cerebellar vermian height and area were also evaluated in the midsagittal plane. In addition, the vertical diameter of the CC genu was measured in the coronal plane through the anterior fontanelle. Representative views and measurements are provided in Figure 1 and Figure S1. Schematic representation of the CC and cerebellar measurements obtained from midsagittal cranial ultrasonography. (A) Measurements of the CCL, CCFL, and CCFA. (B) Measurement of the CCCL. (C) Measurements of the thickness of the CC genu, body, and splenium in the midsagittal plane. (D) Measurements of the midsagittal CC area and cerebellar vermis area. CC, corpus callosum; CCCL, corpus callosum curve length; CCFA, corpus callosum-fastigium angle; CCFL, corpus callosum-fastigium length; CCL, corpus callosum length. Statistical analysis Continuous maternal and neonatal variables were compared using Student’s t-test or one-way analysis of variance, as appropriate, while categorical data were analyzed using Chi-squared tests or Fisher’s exact tests, as appropriate. Segmented linear regression and nonlinear regression models were employed to characterize the developmental trajectories of CC and cerebellar parameters across CGA. The inflection point for segmented regression was empirically determined based on apparent plateauing and changes in slope observed in the nonlinear regression curves; therefore, the segmented regression analyses were considered exploratory. To account for repeated within-subject measurements, linear mixed-effects (LME) models were additionally performed with infant identity included as a random intercept and CGA as a fixed effect. For trajectory analyses, the optimal model fit was determined based on the Akaike Information Criterion and coefficient of determination (R2). Univariate and multivariable regression models were used to evaluate associations between PT birth and neuroimaging indicators. To minimize bias, multivariable regression analyses were performed with adjustment for potential confounders. The covariates for the multivariable analyses were selected based on their clinical relevance, previous literature, and associations observed in univariate analyses, while accounting for sample size limitations and the potential risk of model overfitting. All statistical analyses were performed using R software (version 4.4.2; R Foundation for Statistical Computing, Vienna, Austria). To account for multiple comparisons, the Benjamini-Hochberg procedure was applied to control the false discovery rate (FDR), and adjusted P values were reported where appropriate. A two-tailed P<0.05 was considered statistically significant. Missing data were minimal (<5% for all variables); therefore, a complete-case analysis was performed. Results A total of 234 neonates were screened during the study period. After excluding 87 infants before formal eligibility assessment, 147 neonates underwent eligibility evaluation. Following additional exclusions in the term (n=3) and PT (n=27) cohorts based on predefined clinical criteria, missing data, or incomplete follow-up, 117 neonates (60 term infants and 57 PT infants) were included in the final analysis. The participant recruitment and exclusion process is summarized in Figure 2. Flowchart of participant recruitment, eligibility assessment, exclusion, and final cohort selection. BPD, bronchopulmonary dysplasia; FGR, fetal growth restriction; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis; PVL, periventricular leukomalacia. A total of 719 cranial ultrasound examinations were included in the trajectory analyses. The PT infants underwent a median of 11.0 [interquartile range (IQR): 9.0–14.0] ultrasound examinations per infant compared with 2.0 (IQR: 2.0–3.0) examinations in term infants (P<0.001), with a median follow-up duration of 10.0 (IQR: 8.0–13.0) weeks versus 1.0 (IQR: 1.0–2.0) weeks, respectively (P<0.001). Baseline characteristics The maternal and neonatal characteristics are summarized in Table 1. Among the term infants, the primary indications for admission were pneumonia in 34 cases (56.7%, with or without concomitant jaundice), and isolated pathological jaundice in 25 cases (43.3%). In contrast, all PT infants were admitted to the Neonatal Department for routine clinical management and developmental monitoring following premature birth. Regarding the etiologies of prematurity in our cohort, 13 infants (22.8%) were born following spontaneous PT birth, 13 (22.8%) due to PT premature rupture of membranes, and 31 (54.4%) via iatrogenic (indicated) PT delivery secondary to maternal or fetal complications. Table 1Comparison of maternal and neonatal characteristics between preterm and term infantsCharacteristicsTerm (n=60)Preterm (n=57)PMaternal Age (years)30.9±2.931.8±4.50.346 BMI (kg/m2)21.4±2.121.1±3.40.645 Gravida1.9±1.02.3±1.30.243 Para0.4±0.60.2±0.50.178 Smoking6 (10.0)6 (10.5)0.925 Alcohol8 (13.3)9 (15.8)0.706 Diabetes11 (18.3)21 (36.8)0.025* Hypertension6 (10.0)9 (15.8)0.349 Anemia5 (8.3)3 (5.3)0.718 Cardiovascular disease4 (6.7)3 (5.3)>0.99 IVF5 (8.3)21 (36.8)<0.001*** Emergency cesarean8 (13.3)15 (26.3)0.077 Placental abruption2 (3.3)12 (21.1)0.003** PROM14 (23.3)15 (26.3)0.709 Bile stasis1 (1.7)3 (5.3)0.356Infants Male35 (58.3)27 (47.4)0.235 Birth weight (g)3,208±3491,229±333<0.001*** GA subgroup Extremely preterm (<28 weeks)–14 (24.6)– Median GA, weeks (IQR)–26.5 [25.6–26.8]– Very preterm (28+0 to 31+6 weeks)–29 (50.9)– Median GA, weeks (IQR)–30.0 [29.0–30.6]– Moderate preterm (32+0 to 33+6 weeks)–9 (15.8)– Median GA, weeks (IQR)–33.0 [32.1–33.7]– Late preterm (34+0 to 36+6 weeks)–5 (8.8)– Median GA, weeks (IQR)–35.4 [34.9–36.4]– Weight at TEA (g)3,140 [2,730–3,395]2,510 [2,210–2,670]<0.001*** Extrauterine growth restriction–23 (40.4)– Intracranial hemorrhage grade I/II1 (1.7)18 (31.6)<0.001*** Pathological jaundice44 (73.3)53 (93.0)0.005** Mechanical ventilation9 (15.0)39 (68.4)<0.001*** Patent ductus arteriosus2 (3.3)34 (59.6)<0.001*** Antenatal corticosteroid exposure0 (0.0)42 (73.7)<0.001*** Postnatal corticosteroid exposure0 (0.0)9 (15.8)0.001** Mild bronchopulmonary dysplasia0 (0.0)5 (8.8)0.025* 10-minute Apgar score9.9±0.58.9±0.9<0.001*** Number of ultrasound examinations2.0 [2.0–3.0]11.0 [9.0–14.0]<0.001*** Length of hospital stay (days)7.0 [5.0–14.0]55.0 [39.5–86.0]<0.001*** CGA at discharge (weeks)40.3 [39.6–41.2]38.4 [37.0–40.1]<0.001*** Duration of follow-up (weeks)1.0 [1.0–2.0]10.0 [8.0–13.0]<0.001***Data are presented as mean ± standard deviation, n (%), or median [interquartile range]. *, P<0.05; **, P<0.01; ***, P<0.001. Missing data for all variables were <5% and are not reported. BMI, body mass index; CGA, corrected gestational age; GA, gestational age; IQR, interquartile range; IVF, in vitro fertilization; PROM, premature rupture of membranes; TEA, term-equivalent age. Comparison of maternal and neonatal characteristics between preterm and term infants Data are presented as mean ± standard deviation, n (%), or median [interquartile range]. *, P<0.05; **, P<0.01; ***, P<0.001. Missing data for all variables were <5% and are not reported. BMI, body mass index; CGA, corrected gestational age; GA, gestational age; IQR, interquartile range; IVF, in vitro fertilization; PROM, premature rupture of membranes; TEA, term-equivalent age. Compared with the term group, the PT infants had lower 10-minute Apgar scores (8.9±0.9 vs. 9.9±0.5, P<0.001), a higher prevalence of conception by in vitro fertilization (IVF) (21/57 vs. 5/60, P<0.001), and more frequent placental abruption (12/57 vs. 2/60, P=0.003). Significant differences in postnatal growth were observed at TEA. The PT infants had a significantly lower median weight at TEA compared with the term infants (P<0.001), and 23 of the 57 PT infants (40.4%) met the criteria for extrauterine growth restriction. Further, the PT group showed a higher frequency of clinical morbidities, including mechanical ventilation, patent ductus arteriosus, and intracranial hemorrhage (all P<0.001). No significant differences were observed between the two groups in terms of fetal sex, maternal smoking, alcohol consumption, hypertension, diabetes, cardiovascular disease, or history of labor induction (all P>0.05). Morphological comparisons at TEA At 40 weeks’ CGA, significant morphological differences persisted between the PT and term groups (Table 2). In the sagittal plane, the PT group exhibited significantly shorter CCL (39.8±2.3 vs. 42.5±3.8 mm, adjusted P=0.015) and CCCL (53.7±6.0 vs. 68.5±17.4 mm, adjusted P=0.004). Conversely, CC height was significantly greater in the PT group (adjusted P=0.045). No significant differences were observed in the thickness of the genu (adjusted P=0.449), body (adjusted P=0.449), splenium (adjusted P=0.774), or in the total CC area (adjusted P=0.126). Table 2Comparison of CC and cerebellar vermis morphometric parameters and their spatial relationship between preterm and term infantsParametersTerm (n=60)Preterm (n=57)PAdjusted P (FDR)Coronal plane Vertical diameter of CC genu (mm)2±0.51.7±0.50.019*0.041*Sagittal plane Genu thickness (mm)4.5±0.94.3±1.00.3290.449 Body thickness (mm)2.8±0.62.7±0.40.3270.449 Splenium thickness (mm)4.6±0.74.5±0.70.7740.774 CC length (mm)42.5±3.839.8±2.30.005**0.015* CC height (mm)14.8±1.916.3±3.80.025*0.045* CC curve length (mm)68.5±17.453.7±6.00.001**0.004** Genu area (mm2)50.8±29.947.6±11.10.6560.706 Rostral body area (mm2)23.5±5.320.6±4.60.042*0.063 Anterior mid-body area (mm2)22.3±5.419.0±4.60.020*0.041* Posterior mid-body area (mm2)22.2±6.119.0±3.60.036*0.059 Isthmus area (mm2)26.2±8.220.7±4.80.008**0.021* Splenium area (mm2)52.3±13.249.6±13.50.4500.540 Total CC area (mm²)197.3±48.2176.5±34.00.0910.126 Vermis height (mm)24.2±3.127.0±9.40.0710.106 Vermis area (mm2)455.9±81.1400.3±94.60.018*0.041* CCFL (mm)48.6±3.042.1±10.3<0.001***<0.001*** CCFA (°)33.6±4.543.1±6.8<0.001***<0.001***Data are presented as mean ± standard deviation. *, P<0.05; **, P<0.01; ***, P<0.001. P values were adjusted using the Benjamini-Hochberg procedure to control the FDR. CC, corpus callosum; CCFA, corpus callosum-fastigium angle; CCFL, corpus callosum-fastigium length; FDR, false discovery rate. Comparison of CC and cerebellar vermis morphometric parameters and their spatial relationship between preterm and term infants Data are presented as mean ± standard deviation. *, P<0.05; **, P<0.01; ***, P<0.001. P values were adjusted using the Benjamini-Hochberg procedure to control the FDR. CC, corpus callosum; CCFA, corpus callosum-fastigium angle; CCFL, corpus callosum-fastigium length; FDR, false discovery rate. Regarding the cerebellum, the vermis area was significantly smaller in the PT group (adjusted P=0.041), whereas the vermis height did not differ significantly between groups (adjusted P=0.106). Notably, the PT group demonstrated a significantly shorter CCFL (42.1±10.3 vs. 48.6±3.0 mm, adjusted P<0.001) and a markedly larger CCFA (43.1±6.8° vs. 33.6±4.5°, adjusted P<0.001) compared to the term group. Additionally, coronal imaging demonstrated that the vertical diameter of the CC genu remained smaller in the PT infants (adjusted P=0.041). The ICC analysis demonstrated excellent inter- and intra-observer reproducibility for all parameters (all ICCs >0.90). Developmental trajectories The nonlinear correlation analysis demonstrated that the term neonates consistently exhibited greater CCL than the PT infants throughout the observation period (Figure 3A). In the term group, the CCL showed no marked postnatal changes. In contrast, while the CCL in the PT group increased progressively with advancing CGA, its growth plateaued after approximately 35 weeks. Consequently, the CCL in the PT infants failed to achieve the levels observed in the term infants by TEA. Similarly, the CCCL in the PT infants demonstrated decelerated growth after 35 weeks’ CGA, remaining consistently lower than that in the term infants (Figure 3B). Notably, after 35 weeks’ CGA, the CCFL in the PT infants continued to increase, narrowing the difference between the PT and term groups (Figure 3C). However, the CCFA exhibited a distinct nonlinear developmental pattern and remained larger in the PT infants than in the term infants throughout the later observation period (Figure 3D). The robustness of these growth patterns was further supported by LME modeling. Even after adjusting for individual variance as a random effect, CGA remained a significant independent predictor for the progressive growth of the CCL (β=0.58, P<0.001), CCCL (β=0.38, P=0.002), and CCFL (β=0.88, P<0.001). For the CCFA, the lack of a significant linear association in the LME model (P=0.239) was consistent with its observed nonlinear U-shaped trajectory. Developmental trajectories of CC and spatial relationship parameters across CGA. (A-D) Nonlinear regression curves comparing the developmental trends between preterm (red) and full-term (blue) infants. Shaded areas represent 95% confidence intervals. (E-H) Segmented linear regression analyses for preterm infants using 35 weeks’ CGA as the inflection point. CC, corpus callosum; CCCL, corpus callosum curve length; CCFA, corpus callosum-fastigium angle; CCFL, corpus callosum-fastigium length; CCL, corpus callosum length; CGA, corrected gestational age. Segmented regression analysis of growth rates To quantify these trajectories, segmented linear regression was performed using 35 weeks’ CGA as the inflection point (Figure 3E-3H). Before 35 weeks’ CGA, the CCL demonstrated a relatively rapid increase with GA (1.04 mm/week, R2=0.337, P<0.001), followed by no significant linear growth thereafter (R2=0.006, P=0.059). Similarly, the CCCL showed an increasing trend before 35 weeks’ CGA (1.02 mm/week, R2=0.049, P<0.001) before plateauing (R2=0.003, P=0.216), although considerable interindividual variability was observed. The CCFL exhibited progressive elongation before 35 weeks’ CGA (1.17 mm/week, R2=0.349, P<0.001), followed by a slower but still significant rate of increase thereafter (0.55 mm/week, R2=0.068, P<0.001). In contrast, the CCFA showed a modest decrease before 35 weeks’ CGA (R2=0.017, P=0.002), followed by a gradual increase thereafter (R2=0.039, P<0.001). However, the segmented regression model explained only a small proportion of the observed variability (adjusted R2=0.024); thus, these findings should be considered exploratory. Factors associated with CC morphometric and spatial parameters The factors associated with the CCL and CCFA in the univariate regression analyses are summarized in Tables 3,4, respectively. For the CCL (Table 3), PT birth, advanced maternal age, and lower Apgar scores were associated with shorter CCL measurements. PT birth was also associated with shorter CCCL and CCFL values. In the multivariable model, PT birth remained independently associated with shorter CCL [β=−2.30, 95% confidence interval (CI): −4.15 to −0.46, P=0.015], as did advanced maternal age (β=−0.26, 95% CI: −0.51 to −0.02, P=0.035). For the CCFA (Table 4), the univariate analysis demonstrated that PT birth was associated with larger CCFA values. In the multivariable model, PT birth (β=7.04, 95% CI: 3.60–10.48, P<0.001), maternal hypertension history (β=5.08, 95% CI: 1.39–8.77, P=0.008), and a higher number of pregnancies (β=1.15, 95% CI: 0.03–2.27, P=0.043) were independently associated with larger CCFA values. Table 3Univariate and multivariable regression analyses of the factors associated with corpus callosum lengthVariablesCorpus callosum lengthUnivariateMultivariableβ (95% CI)Pβ (95% CI)PMaternal Preterm birth–2.73 (–4.61 to –0.85)0.005**–2.30 (–4.15 to –0.46)0.015* Age–0.30 (–0.55 to –0.04)0.024*–0.26 (–0.51 to –0.02)0.035* BMI–0.05 (–0.40 to 0.31)0.799 Gravida–0.58 (–1.40 to 0.23)0.159 Para–0.11 (–1.62 to 1.40)0.888 Smoking1.60 (–1.31 to 4.52)0.276 Alcohol0.34 (–2.18 to 2.86)0.789 Diabetes0.18 (–1.93 to 2.29)0.865 Hypertension–2.14 (–4.86 to 0.59)0.122 Anemia1.81 (–1.58 to 5.21)0.291 Cardiovascular disease2.53 (–1.21 to 6.28)0.182 IVF–0.69 (–3.11 to 1.73)0.572 Emergency cesarean–1.99 (–4.29 to 0.30)0.088–1.63 (–3.79 to 0.52)0.136 Placental abruption–1.34 (–4.48 to 1.80)0.396Infants Sex–1.40 (–3.17 to 0.36)0.118 Labor induction0.59 (–3.76 to 4.94)0.788 PROM–1.13 (–3.18 to 0.92)0.274 Bile stasis0.51 (–6.91 to 7.94)0.891 10-minute Apgar 1.65 (0.50 to 2.80)0.006**0.83 (–0.57 to 2.22)0.243*, P<0.05; **, P<0.01. BMI, body mass index; CI, confidence interval; IVF, in vitro fertilization; PROM, premature rupture of membranes. Univariate and multivariable regression analyses of the factors associated with corpus callosum length *, P<0.05; **, P<0.01. BMI, body mass index; CI, confidence interval; IVF, in vitro fertilization; PROM, premature rupture of membranes. Table 4Univariate and multivariable regression analyses of factors associated with corpus callosum-fastigium angleVariablesCorpus callosum-fastigium angleUnivariateMultivariableβ (95% CI)Pβ (95% CI)PMaternal Preterm birth9.54 (6.71 to 12.37)<0.001***7.04 (3.60 to 10.48)<0.001*** Age0.14 (–0.35 to 0.63)0.564 BMI0.34 (–0.31 to 0.99)0.303 Gravida1.77 (0.30 to 3.23)0.019*1.15 (0.03 to 2.27)0.043* Para–0.06 (–2.84 to 2.73)0.967 Smoking–3.59 (–8.94 to 1.76)0.185 Alcohol1.62 (–3.02 to 6.25)0.489 Diabetes1.41 (–2.46 to 5.28)0.468 Hypertension5.73 (0.81 to 10.65)0.023*5.08 (1.39 to 8.77)0.008** Anemia–2.19 (–8.48 to 4.10)0.489 Cardiovascular disease–3.56 (–10.51 to 3.39)0.310 IVF4.64 (0.31 to 8.96)0.036*0.54 (–2.92 to 4.00)0.754 Emergency cesarean3.47 (–0.77 to 7.71)0.107 Placental abruption4.74 (–0.95 to 10.43)0.101Infants Sex–0.78 (–4.10 to 2.53)0.638 Labor induction–4.46 (–12.41 to 3.50)0.267 PROM–0.67 (–4.48 to 3.13)0.724 Bile stasis–3.92 (–17.58 to 9.74)0.568 10-minute Apgar–4.52 (–6.48 to –2.56)<0.001***–1.73 (–3.73 to 0.28)0.090*, P<0.05; **, P<0.01; ***, P<0.001. BMI, body mass index; CI, confidence interval; IVF, in vitro fertilization; PROM, premature rupture of membranes. Univariate and multivariable regression analyses of factors associated with corpus callosum-fastigium angle *, P<0.05; **, P<0.01; ***, P<0.001. BMI, body mass index; CI, confidence interval; IVF, in vitro fertilization; PROM, premature rupture of membranes. Discussion In this prospective cohort study using serial neurosonography, we demonstrated that PT birth is associated with persistent alterations in CC and cerebellar development, as well as in CC-cerebellar spatial relationships, that remain evident at TEA. At 40 weeks’ CGA, PT infants exhibited significantly shorter CC linear dimensions and a reduced cerebellar vermis area compared to term controls, suggesting incomplete structural catch-up of commissural white matter and cerebellar development. Notably, the CCFA—a metric reflecting the spatial orientation between the CC and the cerebellar fastigium—was significantly altered at TEA and exhibited an exploratory nonlinear developmental pattern. These findings suggest that prematurity is associated not only with altered CC growth but also with changes in the spatial configuration of supratentorial and infratentorial structures. In addition to intrinsic neurodevelopmental alterations, postnatal cranial molding may also contribute to the observed trajectories of the CCFA and CCFL. PT infants commonly develop a dolichocephalic or scaphocephalic cranial morphology during early extrauterine life, characterized by a relatively elongated anteroposterior dimension and reduced biparietal width. Recent three-dimensional cranial studies have demonstrated that dolichocephaly is highly prevalent in PT infants and may persist or even progress during the period from birth to TEA rather than normalize during early postnatal development (23,24). Such cranial molding, potentially influenced by prolonged hospitalization, positioning practices, and non-invasive respiratory support, may alter intracranial spatial relationships independently of intrinsic brain growth. Such extracerebral cranial shape changes may alter the relative spatial configuration of supratentorial and infratentorial structures and could partly explain the altered CCFA and CCFL trajectories, as well as the increased CC height observed in PT infants. Therefore, these geometric parameters should be interpreted as reflecting both brain morphometric development and the influence of postnatal cranial morphology. Potential mechanisms underlying altered CC growth and CCFA trajectories Our trajectory analysis identified a distinct growth inflection during the late-PT period, characterized by an initial rapid increase in CC linear metrics followed by a plateau, whereas the CCFL continued to extend at a decelerated rate. This 35-week threshold identified in our study may reflect a pivotal biological transition from pre-oligodendrocytes to mature myelinating oligodendrocytes, a process that typically begins during this late-PT developmental window (5). This period marks the onset of rapid, organized myelination—a developmental phase that is highly energy-dependent and particularly vulnerable to the systemic inflammation and oxidative stress often encountered in the extrauterine environment. However, the relatively low R2 values observed in several segmented regression models indicate substantial interindividual variability, suggesting that GA alone explained only a limited proportion of morphometric variance. This variability may partly reflect the biological heterogeneity of PT populations across different GA categories. Therefore, the identified trajectory patterns should be interpreted cautiously and primarily as exploratory descriptions of developmental trends rather than definitive predictive models. This pattern supports the hypothesis that the extrauterine environment may fail to fully replicate the optimal intrauterine conditions required for the maturation of commissural pathways during the third trimester. The vulnerability of white matter development in PT infants and the subsequent maturational disruptions—which frequently occur even in the absence of overt parenchymal injury—have been consistently linked to suboptimal neurodevelopmental outcomes (3,25). Further, early postnatal factors may actively modulate these processes. While neonatal brain activity is essential for refining CC microstructural maturation (26), cumulative clinical stressors, including systemic sepsis, NEC, and postnatal corticosteroid exposure, have been shown to adversely affect CC and cerebellar growth trajectories (27). Consequently, the observed alterations in CC-related morphometric and CC-cerebellar spatial parameters likely reflect a multifaceted interplay between intrinsic maturational processes and extrinsic clinical influences. In addition, because some measurements were obtained shortly after birth, early morphometric differences in PT infants may also partly reflect antenatal and intrauterine influences associated with the conditions leading to PT delivery. The divergence between the cessation of CC elongation and the continued increase in CCFL after approximately 35 weeks’ CGA may provide a morphometric basis for the nonlinear evolution of the CCFA. As callosal elongation slows—limiting the further anterior displacement of the genu—the continued increase in the CCFL, reflecting changes in the spatial relationship between the CC and the cerebellar fastigium, may contribute to the progressive enlargement of the CCFA. While this provides a plausible structural framework for the apparent increase in the CCFA observed in PT infants, the specific physiological drivers of this spatial reorganization remain incompletely understood. In addition, postnatal cranial molding and dolichocephalic cranial morphology may also contribute to these altered spatial relationships in PT infants. Regional morphometric alterations in PT brain development The greater CC height observed in PT infants despite shorter overall CC dimensions suggests a relatively rounder CC configuration, a morphometric pattern that has also been reported in previous studies of PT brain development (20,28). Lubián-Gutiérrez et al. also reported greater CC height in PT children with more favorable neurodevelopmental profiles, suggesting that CC height may reflect complex regional developmental patterns rather than simply representing reduced linear growth (29). Such findings may reflect disproportionate regional maturation of the CC in PT infants, including alterations in axonal organization and myelination (4). Relatively reduced anterior-posterior elongation, particularly in posterior callosal regions, may contribute to a more rounded midsagittal appearance and altered CC geometry. Postnatal cranial molding may additionally contribute to this altered geometric appearance. Notably, this study also found that vermis height was numerically greater in PT infants despite a significantly smaller vermis area. Vermis height primarily reflects craniocaudal growth of the medial cerebellar region, whereas vermis area provides a broader two-dimensional estimate of overall vermian development. In a study of very-low-birth-weight (VLBW) PT infants at 40 weeks’ CGA, Liu et al. reported a smaller transverse cerebellar diameter but a relatively preserved vermis height compared with term infants, suggesting that prematurity may affect lateral cerebellar hemisphere development more prominently than medial vermian growth (30). This regional vulnerability may relate to the rapid late-gestational expansion of the cerebellar hemispheres, which undergo substantial growth and granule cell proliferation, potentially rendering them more susceptible to extrauterine injury following PT birth. Postnatal cranial molding may also contribute to the altered geometric appearance of both the CC and cerebellar vermis. Therefore, the increased CC height and vermis height observed in PT infants likely reflect a combination of intrinsic neurodevelopmental alterations and extracerebral cranial morphological influences. CCFA and altered spatial brain configuration in PT infants A key contribution of this study is the application of the CCFA to characterize evolving spatial relationships between the CC and cerebellar structures during postnatal development. The significantly larger CCFA at TEA, together with a reduced vermis area, suggests asynchronous maturation of the commissural and posterior fossa structures in PT infants. This is supported by prior evidence indicating that CC and vermian growth follow distinct trajectories in VLBW infants (10), implying that the spatial coupling between these regions is developmentally vulnerable. Our findings suggest that morphometric alterations in PT infants may differ from those reported in fetal growth restriction (FGR), although direct comparisons were beyond the scope of the present study. In fetuses with late-onset FGR, CC-related lengths are reduced (31-33), while the CCFA often remains stable (22), suggesting relatively preserved geometry despite reduced size. Conversely, our PT cohort exhibited both reduced linear dimensions and altered angular metrics (CCFA) at TEA. This divergence suggests that the extrauterine environment and postnatal exposures may uniquely reshape intracranial spatial relationships, a phenomenon distinct from the intrauterine constraints of FGR. Further, serial neurosonography enabled the characterization of developmental trajectories beyond the limitations of cross-sectional assessments. While recent high-quality fetal studies have established robust reference ranges for CCL (34), these prenatal data cannot capture the postnatal growth dynamics or the spatial reorganization triggered by PT birth. Our longitudinal approach complements existing fetal reference data by identifying growth inflections and morphological divergences following the extrauterine transition. Moreover, key ultrasound-derived parameters of the CC and cerebellum have shown strong agreement with MRI at TEA (35), supporting the clinical feasibility and translational utility of serial cranial ultrasonography for monitoring high-risk PT infants. Clinical implications Our findings underscore the value of serial neurosonography for longitudinal assessment of postnatal brain morphometric development in PT infants. Previous studies have shown that early CC dimensions and growth trajectories may be associated with neurobehavioral performance at TEA (36), and that both CC and cerebellar vermis measurements are linked to neurodevelopmental outcomes (9). Accordingly, these parameters may provide useful morphometric information regarding altered brain development in PT populations, although their long-term clinical significance requires further validation. Notably, even moderate-to-late PT infants exhibit measurable morphological differences from their term peers at TEA, even in the absence of overt brain injury (37). In this context, the CCFA emerges as a quantitative morphometric parameter that exhibited significant between-group differences at 40 weeks’ CGA, and may serve as a quantitative descriptor of altered brain developmental patterns in PT infants. However, further longitudinal studies are required to determine the clinical significance of these morphometric alterations. Limitations This study had several limitations. First, it was a single-center study with relatively small prematurity subgroups, limiting formal GA-stratified analyses and comprehensive adjustment for all neonatal morbidities in the multivariable models. In addition, standardized head circumference measurements were not systematically collected at the time of TEA neurosonography examinations, precluding direct assessment of the relationship between CCFA and global cranial growth. Second, several morphometric parameters demonstrated relatively large variance, which may reflect interindividual anatomical variability, biological heterogeneity, and limited longitudinal sampling. Third, although cranial ultrasonography is a robust bedside tool, it lacks the sensitivity of diffusion MRI for assessing white matter microstructural integrity. Future multimodal studies integrating advanced neuroimaging and long-term neurodevelopmental assessments are needed to clarify the functional significance of these structural alterations. Finally, the observed growth inflection around 35 weeks’ CGA was empirically identified and should therefore be considered exploratory pending further mechanistic validation. Conclusions PT birth is associated with alterations in CC and cerebellar development, as well as CC-cerebellar spatial relationships, that are evident at TEA. Longitudinal assessment revealed distinct developmental trajectories of CC morphometric parameters and CC-cerebellar spatial organization, with CCFA demonstrating an exploratory nonlinear developmental trajectory. These findings support the utility of serial cranial ultrasonography for the longitudinal assessment of postnatal brain morphometric development and highlight the need for further studies to clarify the mechanisms and clinical significance of altered cerebro-cerebellar development and spatial organization in PT infants. Supplementary The article’s supplementary files as