Immune and body composition signatures of neurodevelopmental impairments in preterm preschoolers.

Immune and body composition signatures of neurodevelopmental impairments in preterm preschoolers

Abstract

Background

Preterm birth is a primary risk factor for long-term neurodevelopmental deficits. While acute neonatal inflammation is a known disruptor of brain development, the persistence of immune dysregulation and its relationship with body composition in the preschool years remains poorly characterized. This study aimed to identify immunological and anthropometric signatures associated with neurodevelopmental outcomes in preschoolers born preterm.

Methods

We conducted a prospective cohort study of preschoolers born at ≤ 34 weeks gestation. We performed comprehensive anthropometry, high-dimensional flow cytometry, and quantification of serum neuroinflammatory biomarkers, on children with Normal, Delay or Risk in neurodevelopment.

Results

Neurodevelopmental impairments predominantly affected motor, language and cognitive domains in our cohort. Children in the Delay and Risk groups exhibited significantly reduced subcutaneous adiposity and lower fat mass compared to the Normal group, despite comparable linear growth. Immunologically, the Risk group was characterized by decreased neutrophils, platelets, and IP-10 levels. The B-cell compartment was the strongest predictor of clinical status; children with neurodevelopmental deficits exhibited a persistence of immature transitional B cells (CD38+CD24+IgD+), whereas those with normal development were enriched for mature class-switched memory B cells (CD27+IgD-).

Conclusion

Preterm preschoolers with neurodevelopmental impairments showed a systemic profile of biological immaturity (reduced subcutaneous adiposity, lower neutrophil and platelet counts and a persistence of transitional B cells). Our identification of a distinct B-cell signature and altered body composition profiles in preschoolers born preterm offers promising new avenues for early detection and intervention.

Keywords: neurodevelopment, neutrophils, platelets, preschoolers, preterm, subcutaneous adiposity, transitional B cells

1.Introduction

Preterm birth remains a significant global health issue that accounts for approximately 16% of all births, totaling 13.4 million worldwide (1–4). Preterm neonates are a vulnerable population exposed to adverse intrauterine and postnatal events that interrupt organ development during critical growth stages (5). Consequently, preterm survivors face increased risk of long-term comorbidities, including immunological disorders, delayed growth, altered body composition, and neurodevelopmental impairments (3, 4, 6).

A significant challenge in these children is encephalopathy of prematurity, a condition characterized by brain injury, often triggered by hypoxia-ischemia, inflammation, and infection during fetal and postnatal stages, that can result in neurodevelopmental disorders later in life (7). While the central nervous system (CNS) is traditionally considered an immunologically privileged site, it can be severely affected by systemic infections occurring during pregnancy or the postnatal period (8). Furthermore, the immaturity of the preterm brain renders their CNS particularly susceptible to the hostile environment of the Neonatal Intensive Care Unit (NICU), where persistent stressors and inflammatory events can lead to permanent neurodevelopmental alterations (9–12).

Tissue dysfunction, rather than just acute damage, can trigger persistent inflammatory processes, and even after the primary stimuli ceases, a cycle of inflammation-dysfunction may establish low-grade chronic inflammation with long-term systemic repercussions (13). Within this inflammatory context, immune cells such as T cells, B cells, and antigen-presenting cells are of particular interest; while they do not normally reside in the brain, they may modulate neural function by crossing the blood-brain barrier (BBB) or releasing soluble factors, particularly in preterm infants with altered BBB permeability (14–17).

Neurodevelopmental disorders may clearly manifest in children between 3 and 5 years, as this is a sensitive period for the consolidation of cognitive, motor, and language functions (18). Prematurity has been linked to a significantly higher risk of motor, language, cognitive, and psychiatric disorders. However, most preterm studies focus on children under three years, school-age children, and adolescents (19, 20), despite evidence showing that functional difficulties frequently manifest at preschool age (21–23). Biological immaturity and anthropometric alterations associated with prematurity can disturb the mechanisms of postnatal neurobiological development, resulting in poor performance during neurodevelopmental assessments, even in the absence of early neurological complications (24).

The assessment of neuroimmune inflammation biomarkers (25), together with the evaluation of growth rates and body composition (26, 27), hold promise for early prediction of adverse neurodevelopmental outcomes in preterm infants (28).

The aim of this study was to characterize the body composition and immunological profile of preschool children born preterm and their association with neurodevelopmental outcomes.

2.Materials and methods

2.1.Research design and study population

Observational prospective longitudinal study with a cohort design with eligibility criteria of children born preterm at the National Institute of Perinatology “Isidro Espinosa de los Reyes” (INPer) between March 2018 and May 2020. NEOPREM (ImmuNe system-intEstinal micrObiota-brain axis and neurodeveloPment of infants born pREterM from birth to age five) cohort recruited children 42 to 48 months old between November 2021 and November 2023 and followed up until 60 months. Inclusion criteria were female and male children, born at ≤ 34 weeks gestational age by vaginal birth and cesarean section from mothers > 18 years. Exclusion criteria were children with major congenital defects, autoimmune and genetic diseases, cerebral palsy, non-residents of Mexico City and Metropolitan Area. Elimination criteria were children diagnosed with intellectual disability during neurodevelopmental assessments, and children without blood samples. For the analysis, children were classified in 3 study groups according to their results in the Child Development Evaluation Test Second Edition (EDI-II): Normal (n = 17), Delay (n = 27), and Risk (n = 21).

2.2.Data collection

Infant and maternal data were retrieved from electronic medical records. Infant variables included: mode of delivery (programmed C-section, emergency C-section, eutocic birth, dystocic birth), gestational age (GA), sex, parity (singleton, twins, triplets), birth weight, Apgar at 1 and 5 minutes, invasive mechanical ventilation (IMV), length of hospitalization, antibiotic (ABT) administration during hospitalization. Infant clinical diagnoses during hospitalization included: early-onset sepsis (EOS), late-onset sepsis (LOS), necrotizing enterocolitis (NEC) II and III, atypical pneumonia, moderate and severe bronchopulmonary dysplasia (BDP), apnea, retinopathy or prematurity (ROP), multifactorial hyperbilirubinemia (MFHB), brain hemorrhage (intraventricular, intracerebral, subependymal, intraparenchymal, periventricular). Maternal variables included maternal age, years of education, marital status (single, married, consensual union), socioeconomic status (low, medium), pregestational nutritional status according to body mass index (BMI) (healthy weight, overweight, obesity), preeclampsia.

2.3.Neurodevelopmental assessment

Neurodevelopmental assessments were conducted in distraction-free environments by certified psychologists from the Department of Developmental Neurobiology at INPer that were blinded to the patients’ medical history (see Supplementary Methods for details).

Screening assessment: All preschoolers (n = 65) were evaluated with the Child Development Evaluation Test Second Edition (EDI-II, by its acronym in Spanish) between 42- and 48-months, and the average assessment duration was 15 minutes. The EDI-II test was designed and validated for Mexican children aged 1–60 months, assesses five domains (gross motor, fine motor, language, social, and cognitive skills), explores the general neurological status, identifies biological risk factors related to prenatal care and neonatal complications, and addresses warning signs. Results are classified by a traffic-light system: Green (Normal development), Yellow (Developmental Lag, hereafter referred as Delay), and Red (Risk of developmental delay, hereafter referred as Risk) (29, 30).

Diagnostic Assessment: Children’s neurodevelopment was assessed between 42- and 48-months (n = 61), and at 60 months (n = 55) using the McCarthy Scales of Children’s Abilities (MSCA), and the average assessment duration was 45 to 60 minutes. The MSCA test evaluates six domains in children aged 2.5-8.5 years: verbal, perceptual-performance, quantitative, memory, motor, and general cognitive index (GCI, a composite score derived from the verbal, perceptual-performance and quantitative scales). Domain scores have a mean of 50 ± 10, except for the GCI, which has a mean of 100 ± 15. Raw values were standardized by child age. Higher scores indicate better developmental outcomes (Supplementary Figure 8) (22, 31–33).

2.4.Nutritional and anthropometric assessment

Anthropometric measures were performed in duplicate by experienced and trained nutritionists following Lohman’s methodology (34), as previously described for infants (35). Measurements included weight (Digital scale SECA 876), height (Stadiometer SECA 264), and circumferences (head, arm, abdomen, thigh, calf) using a W606PM Lufkin tape (Apex Tool Group). Skinfold thicknesses (biceps, triceps, subscapular, abdominal, thigh, calf) were measured using a Lange caliper (Beta Technology).

Calculated Indices: Body Mass Index (BMI): Weight (kg)/Height (m2); Waist-to-Height Ratio (WHR): Abdominal Circumference (m)/Height (m) (36, 37); Arm Muscle Area (AMA): [MUAC - (π x TSF)]2/(4 x π)], where MUAC is the mid-upper arm circumference and TSF is triceps skinfold thickness; Central-to-Total Skinfold Ratio: SSF+ASF/ΣSF, where SSF is subscapular and ASF is abdominal skinfold thickness, and ΣSF is the sum of biceps, triceps, subscapular, abdominal, thigh, calf skinfolds thickness.

Body Composition: Percentage of Fat Mass (% FM) was estimated using Slaughter’s equations for Girls: 1.33 x (TSF+SSF) - 0.013 x (TSF+SSF)2 - 2.5 and Boys: 1.21 x (TSF+SSF) - 0.008 x (TSF+SSF)2 - 1.7 (38); Absolute Fat Mass (kg): [(% FM/100) x Weight (kg)]; Fat-Free Mass (FFM): Weight (kg) - FM (kg).

Dietary Intake: A food frequency questionnaire was administered to the primary caregiver to assess daily meal frequency and consumption of specific food groups and beverages. Standardized measuring cups and spoons were used to estimate portions.

2.5.Blood sampling, hematic biometry and blood chemistry

Peripheral blood was collected in K2EDTA (500 µL) and SST Amber tubes (600 µL) (BD Microtainer) at 42–48 months old when EDI-II test was administered. K2EDTA blood samples were used for hematic biometry (Beckman Coulter DxH 500) and immunophenotyping by flow cytometry. Serum samples were used for blood chemistry (glucose, triglycerides, total cholesterol, HDL-cholesterol, creatinine, uric acid and urea) with a dry chemistry analyzer (Respons®910 DiaSys), and quantification of neuroimmune inflammation biomarkers by Multiplex Immunoassays.

2.6.Immunophenotyping

Innate (NK cells) and adaptive (T cells and B cells) immune populations were analyzed using 70 µL of peripheral blood per panel. Samples were washed with PBS, stained with Zombie Aqua fixable viability dye (BioLegend), blocked with human TrueStain FcX (BioLegend), incubated with fluorochrome-conjugated monoclonal antibodies against CD14 (M5E2), CD16 (3G8), CD56 (HCD56), CD8α (RPA-T8), CCR7 (G043H7), CD45RA (HI100), CD4 (RPA-T4), CD3 (UCHT1), IgD (IA6-2), CD24 (ML5), CD20 (2H7), CD27 (M-T271), CD38 (HIT2), CD19 (HIB19), HLA-DR (L243) for 30 min at room temperature (all antibodies from BioLegend), and red blood cells were lysed with BD FACS Lysing Solution (BD Biosciences). Data were acquired on a BD FACSAria III flow cytometer and analyzed using FlowJo v10. Gating strategies are detailed in Supplementary Figures 5–7.

2.7.Computational flow cytometry analysis

Briefly, prior to clustering, raw .fcs files were evaluated using FlowJo software to exclude debris, doublets, and dead cells via a Fixable Viability Dye (FVD). Only single, viable, intact mononuclear cells were exported as cleaned .fcs files and loaded into the CATALYST environment for subsequent high-dimensional analysis.

Preprocessing: Raw .fcs files were processed in R (v4.5.2) using flowCore and CATALYST. Channel names were standardized, and fluorescence intensities were arcsinh-transformed (cofactor 150).

Dimensionality Reduction & Clustering: UMAP was performed on downsampled data (1,000 cells/sample) for visualization. Unsupervised clustering was conducted using FlowSOM (10×10 grid) followed by ConsensusClusterPlus metaclustering to define 15 metaclusters per panel.

Annotation & Differential Abundance: Metaclusters were auto-annotated based on median marker expression thresholds (≥ 2.0 considered positive). Differential abundance between clinical groups (Normal vs. Risk/Delay) was assessed using generalized linear models (GLM) via edgeR with Benjamini-Hochberg FDR correction.

Correlation & Predictive Modeling (RAPID): Spearman rank correlation identified cell populations tracking with clinical severity (Normal=1, Delay=2, Risk=3). To identify a minimal immune signature predictive of clinical outcome, we applied a machine learning approach using LASSO (Least Absolute Shrinkage and Selection Operator) regularized logistic regression, implemented via the glmnet package in R (39). The LASSO model performs feature selection by applying a penalty that shrinks the regression coefficients of non-informative features to exactly zero, thereby selecting only the most robust predictors and preventing overfitting. The optimal regularization parameter (λ) was determined through k-fold cross-validation (with k adapted to sample size), which minimizes the model’s prediction error (Binomial Deviance). Clusters with non-zero coefficients at this optimal λ were identified as the final predictive immune signature. To evaluate the performance of the final model, we calculated standard classification metrics on the training set itself, as sample size limitations precluded the use of an independent hold-out test set. The model’s predictions on the training data were used to compute Accuracy, Sensitivity (the proportion of true Delay/Risk cases correctly identified), and Specificity (the proportion of true Normal cases correctly identified) (see Supplementary Methods for details).

2.8.Neuroimmune inflammation biomarkers assessment

Serum biomarkers were quantified using two bead-based immunoassays (LEGENDplex, BioLegend) according to the manufacturer’s instructions: 1) Essential Immune Response Panel: IL-1β, IL-2, IL-4, IL-8, IL-10, IL-12p70, IL-17A, IFN-γ, IP-10 (n=60); 2) Neuroinflammation Panel: BDNF, CCL2 (MCP-1), CX3CL1, IL-6, IL-18, sRAGE, sTREM-1, sTREM2, TNF-α, TGF-β1, VILIP-1, VEGF, β-NGF (n=57). Sample sizes varied due to serum availability. Data was acquired on a FACSAria III flow cytometer and analyzed using QOGNIT software.

2.9.Statistical analysis

Statistical analysis of demographic, clinical, nutritional, environmental, and maternal variables from studied infants were performed using GraphPad Prism version 10.4.1. Descriptive statistics are reported as mean ± standard deviation (SD) for normal continuous variables, median [25th percentile-75th percentile] for non-normal continuous variables, or frequencies (%) for categorical variables. Normality was assessed using the Shapiro–Wilk test. Group comparisons were performed using one-way ANOVA with Šídák’s post hoc test for normally distributed variables or Kruskal–Wallis test with Dunn’s post hoc test for non-normally distributed variables. Fisher’s exact test and chi-squared test were used to compare the frequencies of categorical variables as appropriate. A p-value < 0.05 was considered statistically significant. Correlations between MSCA scores, anthropometry, and biomarkers were evaluated using Spearman’s rank correlation.

3.Results

3.1.Preschoolers born preterm exhibit persistent impairments in motor, language, and cognitive domains

Sixty-five preschoolers born preterm from the NEOPREM cohort were included in the analysis, and their neurodevelopment was evaluated with a screening test (Child Development Evaluation Test Second Edition, EDI-II), and a diagnostic test (McCarthy Scales of Children’s Abilities, MSCA) between 42 to 60 months (Figure 1).

The global result of EDI-II test was used to evaluate the overall performance of children at 42–48 months in five neurodevelopmental domains (gross motor, fine motor, language, social, and cognitive), and to classify them in three groups: Normal (n = 17, 26.2%), Delay (n = 27, 41.5%), and Risk (n = 21, 32.3%) (Figure 2A). Within the Normal group, 76.5% were females, whereas in the Delay and Risk groups the proportion of females and males was similar (female 55.6% vs male 44.4% in the Delay group; female 42.9% vs male 57.1 in the Risk group); no statistical differences between groups were found (p = 0.113) (Table 1). EDI-II results analysis by neurodevelopmental domain revealed that children presented Delay/Risk performance in gross motor (46.1%), cognitive (33.9%), language (33.8%), fine motor (26.1%) and social skills (21.5%) (Figure 2A, Supplementary Table 1).

Table 1Demographic, clinical, and maternal data from studied preschoolers.

Neonatal variable Normal (n = 17) Delay (n = 27) Risk (n = 21) p-value

GA (weeks) 32.3 [29-33] 30.3 [28-32.3] 32 [28.3-33.5] 0.232

Female (%) 13 (76.5) 15 (55.6) 9 (42.9) 0.113

Male (%) 4 (23.5) 12 (44.4) 12 (57.1) 0.113

Birth weight (g) 1325 ± 294.4 1288 ± 494.7 1305 ± 422.6 0.961

IMV (%) 3 (17.6) 11 (40.7) 6 (28.5) 0.261

Antibiotic days 3 [0-15.5] 5 [0-22] 7 [0-22] 0.392

Hospitalization days 46.5 ± 20.1 61.9 ± 33.9 51.6 ± 32.4 0.235

EOS (%) 5 (29.4) 13 (48.1) 9 (42.8) 0.465

LOS (%) 4 (23.5) 8 (29.6) 6 (28.6) 0.902

Atypical pneumonia (%) 2 (11.8) 9 (33.3) 6 (28.6) 0.271

NEC (%) 3 (17.6) 2 (7.4) 4 (19) 0.434

BPD (%) 6 (35.3) 11 (40.7) 8 (38.1) 0.936

Apnea (%) 8 (47) 9 (33.3) 8 (38.1) 0.660

MFHB (%) 14 (82.3) 25 (92.6) 17 (80.9) 0.434

Brain hemorrhage (%) 4 (23.5) 5 (18.5) 6 (28.6) 0.713

ROP (%) 3 (17.6) 10 (37) 5 (23.8) 0.334

Maternal variable

Maternal age (years) 32 [26-35] 28 [24-34] 26 [22-35.5] 0.534

PG healthy weight (%) 9 (52.9) 12 (44.4) 10 (47.6) 0.859

PG overweight (%) 6 (35.3) 11 (40.7) 6 (28.6) 0.682

PG obesity (%) 2 (11.8) 4 (14.8) 5 (23.8) 0.643

Preeclampsia (%) 9 (52.9) 5 (18.5) 10 (47.6) 0.033

GA, gestational age; IMV, invasive mechanical ventilation; EOS, early-onset sepsis; LOS, late-onset sepsis; BPD, bronchopulmonary dysplasia; MFHB, multifactorial hyperbilirubinemia; PG, pregestational; ROP, retinopathy of prematurity. Data are presented as mean ± standard deviation or median [interquartile range]. One-way ANOVA or Kruskal-Wallis tests were used for continuous variables. Fisher's exact test or Chi-Square test were used for categorical variables. Statistically significant p-values (p < 0.05) are shown in bold.

To confirm the EDI-II test results and to perform a deep analysis of children’s neurodevelopmental skills and follow up at 60 months, the MSCA test was administered. At 42–48 months old, the most affected domains with Delay/Risk scores were motor (52.5%), quantitative (47.6%), verbal (39.3%), general cognitive index (31.1%), and memory (27.9%) domains (Figure 2B, Supplementary Table 2). Subsequent evaluation at 60 months old revealed a persistent motor Delay/Risk (45.5%), followed by quantitative impairments (40%) and memory delay (32.7%) (Figure 2B, Supplementary Table 3).

3.2.Neonatal clinical history and maternal sociodemographic profiles

Most preschoolers were delivered by emergency C-section with a median gestational age between 30 to 32 weeks, and we did not observe differences in sex, birth plurality (singleton, twins, triplets), Apgar at 1 and 5 minutes, birth weight, antibiotic days administered during hospitalization, IMV use, and hospitalization length among preschoolers (Table 1, Supplementary Table 4). Next, we analyzed if early-life clinical diagnoses related with encephalopathy of prematurity (infection, inflammation, brain injury) were associated with EDI-II results, and found similar frequencies of EOS, LOS, atypical pneumonia, NEC, BPD, apnea, MFHB, brain hemorrhage and ROP among the 3 study groups (Table 1).

We did not observe differences in maternal age and pregestational nutritional status, and whereas about 50% of mothers in the Normal and Risk groups had preeclampsia during pregnancy there was 18.5% within the Delay group (Table 1).

Most preschoolers belonged to families with low socioeconomic status (SES), but about 30% of children in the Delay group had a medium SES. All mothers had a median schooling length of 12 years, and we did not observe differences in marital status (Supplementary Table 4).

3.3.Neurodevelopmental delay and risk are associated with reduced subcutaneous adiposity despite a healthy nutritional status

To assess the nutritional status of preschoolers we performed a food consumption frequency assessment, hematic biometry, blood chemistry, anthropometric and body composition measures.

We did not observe differences in feeding patterns (Supplementary Table 5); white and red blood cell counts, % hematocrit and hemoglobin concentration (Table 2); blood chemistry analytes (Supplementary Table 6); weight, height, BMI, WHR and AMA among children (Table 3). These results suggested that preschoolers had a healthy nutritional status overall despite the observed neurodevelopmental impairments.

Table 2Hematic biometry from studied preschoolers.

Parameters Normal (n = 17) Delay (n = 27) Risk (n = 21) p-value

WBC (103/μL) 7.5 ± 2 6.1 ± 1.6 6.5 ± 2.4 0.097

RCB (103/μL) 4.9 ± 0.3 4.8 ± 0.3 4.8 ± 0.3 0.320

Hemoglobin (g/dL) 13.7 ± 0.8 13.7 ± 0.9 13.6 ± 0.7 0.917

Hematocrit (%) 41.2 ± 2.5 40.2 ± 2.4 40.2 ± 2.4 0.361

Neutrophils (103/μL) 2.7 [2-3.7] 1.8 [1.3-2.6] 1.7 [0.9-2.7]* 0.035

Lymphocytes (103/μL) 4 ± 1.3 3.5 ± 0.9 3.9 ± 1.2 0.342

Monocytes (103/μL) 0.4 [0.2-0.6] 0.3 [0.3-0.5] 0.3 [0.2-0.4] 0.680

Eosinophils (103/μL) 0.1 [0.0-0.3] 0.1 [0.0-0.1] 0.0 [0.0-0.2] 0.215

Basophils (103/μL) 0.06 [0.03-0.07] 0.04 [0.03-0.07] 0.06 [0.04-0.08] 0.209

Platelets (103/μL) 340 [293.5-400.5] 286 [220-323]* 266 [218.5-337.5]* 0.021

NLR 0.7 ± 0.5 0.6 ± 0.2 0.5 ± 0.4* 0.048

PLR 98.7 ± 42.3 82.6 ± 33.1 73.5 ± 33.2 0.102

WBC, White Blood Cell Count; RBC, Red Blood Cell Count; NLR, Neutrophil-to-Lymphocyte Ratio; PLR, Platelet-to-Lymphocyte Ratio. Data are presented as mean ± standard deviation or median [interquartile range]. One-way ANOVA or Kruskal-Wallis tests were used for continuous variables. Fisher's exact test or Chi-Square test were used for categorical variables. Statistically significant p-values (*p < 0.05) are shown in bold.

Table 3Anthropometric and body composition data from studied preschoolers.

Parameters Normal (n = 16) Delay (n = 24) Risk (n = 20) p-value

Weight (kg) 14.3 ± 2.7 14.0 ± 2.1 13.2 ± 1.6 0.311

Height (cm) 98.4 [93-101.3] 97.3 [94.4-99.5] 96.6 [92.3-98.2] 0.449

BMI (kg/m2) 14.9 [13.6-16] 14.8 [14.3-15.5] 14.4 [13.8-15.2] 0.469

WHR 0.48 ± 0.02 0.48 ± 0.02 0.47 ± 0.02 0.129

AMA (cm2) 12.9 ± 2.1 13.8 ± 2.2 13.1 ± 2.1 0.374

∑SF (mm) 54.8 ± 15.6 45.3 ± 9.8* 43 ± 7.4* 0.009

SSF+ASF/∑SF 0.27 ± 0.03 0.25 ± 0.03 0.24 ± 0.03 0.056

FM (%) 16.5 ± 5.1 13.3 ± 3.6* 12.2 ± 2.6* 0.005

FM (kg) 2.4 ± 1.1 1.9 ± 0.6 1.6 ± 0.4* 0.008

FFM (kg) 11.8 ± 1.8 12.4 ± 1.6 11.5 ± 1.4 0.526

Circumference Normal (n = 17) Delay (n = 25) Risk (n = 21) p-value

Head (cm) 48.8 ± 1.8 48.8 ± 1.5 47.8 ± 1.6 0.115

Arm (cm) 15.2 ± 1.5 15.2 ± 1.3 14.8 ± 1.1 0.501

Abdominal (cm) 47 ± 3.2 46.6 ± 2.7 44.7 ± 1.9* 0.027

Thigh (cm) 27.5 [24.8-29.4] 27.4 [25-28.4] 27.1 [25-28.6] 0.784

Calf (cm) 20.1 [17.6-22.7] 19.7 [18.1-20.9] 19.5 [18.3-20.1] 0.564

Skinfold Normal (n = 16) Delay (n = 24) Risk (n = 20) p-value

Biceps (mm) 6 [4.5-6.3] 5 [4-5.5] 4.5 [4-5] 0.070

Triceps (mm) 8.1 ± 2.4 6.8 ± 1.8 6.5 ± 1.3* 0.035

Subscapular (mm) 6 [5-7.5] 5 [4.5-6] 5 [4.1-5.5]* 0.031

Abdominal (mm) 8.6 ± 2.9 6.3 ± 2.1* 5.7 ± 1.1* 0.0006

Thigh (mm) 13.7 [11.1-16.5] 12.7 [10.1-14.8] 11.2 [8.6-14.3] 0.232

Calf (mm) 10 [8-12.5] 9.5 [8.5-10.7] 9.2 [8-11] 0.722

BMI, Body Mass Index; WHR, Waist-to-height ratio; AMA, Arm muscle area; SSF, Subscapular Skinfold; ASF, Abdominal Skinfold; ∑SF, Total Skinfold; FM, Fat mass; FFM, Fat free mass. Data are presented as mean ± standard deviation or median [interquartile range]. One-way ANOVA or Kruskal-Wallis tests were used for continuous variables. Statistically significant p-values (*p < 0.05) are shown in bold.

Remarkably, body composition assessment revealed that children in the Delay and Risk groups had a reduced sum of skinfolds thickness (subcutaneous adiposity), % fat mass, and abdominal skinfolds thickness compared to children in the Normal group (Table 3). However, the waist-to-height ratio (WHR) showed no significant difference among groups (p = 0.129). This indicates that while children in the Risk group had lower triceps and subscapular skinfolds, and abdominal circumference than those in the Normal group (Table 3), the lack of difference in WHR suggests a nuanced compartmentalization of adipose tissue—a leaner phenotype—rather than a uniform reduction across all central adiposity indices.

3.4.Neurodevelopmental delay and risk are associated with altered myeloid profiles and reduced neutrophil-to-lymphocyte ratio

Detailed analysis of peripheral blood cell counts revealed that preschoolers from the Delay and Risk groups exhibited lower platelet counts than children in the Normal group (Table 2). Furthermore, children in the Risk group had reduced neutrophil counts, and lower neutrophil-to-lymphocyte ratios (NLR) than those with Normal and Delayed development (Table 2). Lymphocytes, monocytes, eosinophils, basophils, and platelet-to-lymphocyte ratios (PLR) were similar among children (Table 2). Despite the observed differences, peripheral blood cell counts remained within normal physiological values.

3.5.B-cell maturation is delayed in children with neurodevelopmental deficits, while T and NK compartments remain unchanged

To further assess the status of innate and adaptive immune cells compartments in the preschoolers, we immunophenotyped peripheral blood NK cells, T cells, and B cells by flow cytometry.

All preschoolers had a similar frequency of NK cell subpopulations (CD56dimCD16+, CD56hiCD16-, CD56-CD16+) (Supplementary Figure 1), as well as CD4+ and CD8+ T cell subpopulations (CD45RA+CCR7+ naïve, CD45RA-CCR7+ central memory, CD45RA-CCR7- effector memory, and CD45RA+CCR7- EMRA) (Supplementary Figure 2).

Interestingly, within the B-cell compartment, children in the Delay group exhibited higher frequency of transitional B cells (CD38hiCD24+) than those in the Normal group, whereas the frequencies of plasmablasts (CD38+CD24-), naive (IgD+CD27-), non-class-switched memory (IgD+CD27+), and class-switched memory (IgD-CD27+) B cells were similar among groups (Figure 3).

3.6.High-dimensional analysis identifies immune clusters predictive of developmental risk

To depict the immune landscape associated with the EDI-II results (Normal, Delay, and Risk), we performed high-dimensional flow cytometry on peripheral blood mononuclear cells. We applied an automated computational pipeline comprising dimensionality reduction (UMAP), unsupervised metaclustering (FlowSOM/ConsensusClusterPlus), and marker enrichment modeling to identify distinct cell populations within the T-cell, B-cell, and NK-cell compartments.

While the T-cell and NK-cell compartments showed a high degree of overlap between clinical groups (Supplementary Figure 3), the B-cell compartment revealed distinct predictive features associated with clinical outcomes (Supplementary Figure 4). To identify specific populations predictive of clinical risk, we applied the RAPID algorithm (LASSO-regularized regression) to train a robust predictive model (Supplementary Figure 9). The relative frequencies (%) of the 15 B-cell metaclusters (Supplementary Figure 4A) were used as input features to predict a binary clinical status (Normal vs. a combined Delay/Risk group) (Figure 4).

The model identified a distinct B-cell signature that stratified the “Normal” status from the “Risk/Delay” status (Figure 4). The strongest predictor of the Risk/Delay outcome was Cluster 13 (Coefficient ~2.36, Figure 4B; indicated by a red bar in Figure 4A), defined as CD38+CD19+CD24+IgD+CD20+HLA-DR+. This phenotype corresponds to transitional B cells, which are immature emigrants from the bone marrow.

Conversely, the model identified Cluster 12 as the strongest predictor of the Normal outcome (Coefficient ~ -4.07, Figure 4B; indicated by a green bar in Figure 4A). This population displayed a CD38+CD19+HLA-DR+CD27+ phenotype, notably lacking IgD and CD24, indicating a Class-Switched Memory B cell profile. Additionally, Cluster 8 (CD27+CD38+ CD19+CD24+IgD+CD20+) and Cluster 5 (CD27+CD38+CD19+CD24+IgD+) were associated with the Normal outcome.

The LASSO regression coefficient’s magnitude indicates the predictive power associated with the outcome. Positive coefficients indicate that a higher frequency of the population is associated with an increased log-odds of being in the Risk/Delay group, while negative coefficients indicate that a higher frequency of the population is associated with an increased log-odds of being in the Normal group (Figure 4B). These results indicate preschoolers with neurodevelopmental impairments are distinguished by a persistence of immature, transitional B-cell phenotypes (Cluster 13), whereas children with normal neurodevelopment exhibit a more mature B-cell profile enriched in class-switched memory cells (Cluster 12). Running this evaluation on our cohort’s training data yielded an Accuracy of 82.26%, a Sensitivity of 97.83%, and a Specificity of 37.50%.

3.7.Reduced IP-10 expression specifically characterizes the at-risk phenotype

The analysis of serum concentrations of cytokines, chemokines, and neuroinflammatory molecules revealed a decrease in IP-10 concentrations in the Risk group compared to the Normal neurodevelopment group, and no other significant differences were observed (Table 4). IL-1β, IL-2 and IFN-γ were undetectable in preschool children’s samples.

Table 4Serum cytokines, chemokines and neuroinflammatory molecules from studied preschoolers.

Analyte (pg/mL) Normal (n = 16) Delay (n = 24) Risk (n = 20) p-value

IL-1β ND ND ND

IL-2 ND ND ND

IL-4 18.1 [11.3-31.9] 17.3 [11.1-24.7] 18.9 [10.9-29.1] 0.908

IL-8 2.3 [0.4-3.3] 2.7 [1-18] 4.14 [1.1-5.3] 0.449

IL-10 3.3 [0.8-4.7] 1.6 [0.6-3.8] 1.3 [0.5-3.2] 0.488

IL-12p70 4.9 [2.7-8.9] 4.3 [2.8-7.3] 4.4 [2.9-9.3] 0.689

IL-17A 6.9 [5.2-10.2] 7.7 [6.1-11] 7.6 [4.9-11] 0.725

IFN-γ ND ND ND

IP-10 80.7 [60.7-103.5] 68.7 [53.8-109.7] 62.8 [44.6-70.4]* 0.037

Analyte Normal (n = 16) Delay (n = 23) Risk (n = 18) p-value

BDNF 11383 [9177-12345] 10881 [9126-11929] 10622 [8546-12115] 0.795

CCL2 121.8 [88.8-167.8] 113.1 [81.4-154.7] 99.7 [82-147.4] 0.523

CX3CL1 3266 [2343-5100] 2678 [2167-3785] 3256 [2917-4886] 0.188

IL-6 65.1 [55.7-73.8] 83.5 [62.2-155.9] 67.5 [51.1-105.2] 0.294

IL-18 248.9 [158.4-379.8] 239.5 [195.7-349] 219.3 [166.3-301.3] 0.864

sRAGE 1261 [836.6-1766] 1121 [852.2-2217] 1350 [1117-1779] 0.587

sTREM-1 503.5 [369.9-948.4] 413.8 [283.4-623.8] 581.4 [350.4-843.9] 0.268

sTREM-2 3467 ± 1174 3372 ± 1265 4182 ± 1349 0.109

TNF-α 29.1 [21.6-55] 41.2 [26.1-71.7] 43.4 [27.9-64.7] 0.536

TGF-β1 78.6 [53.5-122.7] 64.6 [51.3-104.2] 92.5 [61.2-110.9] 0.232

VILIP-1 730.1 [656.3-883.2] 733.2 [584.3-935.1] 852.1 [769.3-974.8] 0.221

VEGF 143.7 [117.4-225.3] 116.9 [99.7-161.8] 117.7 [98.1-165] 0.143

β-NGF 5.2 [4.4-6.2] 5.2 [3.6-7.3] 5.9 [4.5-8.4] 0.511

ND, Not detected. Data are presented as mean ± standard deviation or median [interquartile range]. One-way ANOVA or Kruskal-Wallis tests were used for multiple comparisons. Statistically significant p-values (p < 0.05) are shown in bold.

3.8.Integrative analysis reveals associations between immune signatures, body composition, and neurodevelopment

A correlation analysis was performed between neurodevelopmental domain scores evaluated with the MSCA test (verbal, perceptual-performance, quantitative, general cognitive index, memory and motor), and the anthropometric, body composition and immunological variables. At 48 months, the verbal domain showed a positive correlation with anthropometric indicators (ΣSF, % fat mass, head and abdomen circumferences, subscapular and abdominal skinfolds thickness) and immunological parameters (transitional and NCS memory B cells, IL-10, and VEGF). The perceptual-performance domain was mainly related to adiposity indicators (ΣSF, % fat mass, subscapular skinfold thickness). The quantitative domain showed positive correlations with subscapular skinfold thickness, plasmablasts and NCS memory B cells and a negative correlation with naïve B cells. In the general cognitive index domain positive correlations were observed with markers of adiposity (ΣSF, %vfat mass, abdomen circumference, subscapular and abdominal skinfolds thickness), plasmablasts, NCS memory B cells, and VEGF. The memory domain positively correlated with subscapular skinfold, NCS memory B cells, IL-18 and VEGF. The motor domain showed positive correlations with adiposity indicators (ΣSF, % fat mass, subscapular and abdominal skinfolds thickness) (Figure 5, Supplementary Figure 10).

At 60 months, the verbal domain was associated positively with head and abdomen circumferences, whereas the perceptual-performance domain showed a positive correlation with plasmablasts and NCS memory B cells and a negative correlation with naive B cells and IL-8. The quantitative domain was related to adiposity indicators (ΣSF and subscapular skinfold thickness) and NCS memory B cells. In the general cognitive index domain, positive correlations included head circumference, plasmablasts, NCS memory B cells, and IL-10, and a negative correlation with naive B cells was observed. In the memory domain, positive associations were identified with head and abdomen circumferences, abdominal skinfold thickness, IL-4, IL-10, IL-12p70, and VEGF. The motor domain showed a positive correlation with plasmablasts, and a negative correlation with naïve B cells (Figure 5, Supplementary Figure 10).

4.Discussion

Premature birth exerts profound and potentially lasting effects on child development, influencing body composition, innate and adaptive immune responses, and critical neurological processes such as neurogenesis, myelination, and synaptogenesis (13). Beyond acute neonatal injuries, prematurity is associated with subtle, persistent alterations in white matter microstructure and functional connectivity, which often manifest as motor, cognitive and language impairments later in life (40, 41). Hence, identifying growth patterns and biological markers associated with these outcomes during the preschool years is crucial. Early identification facilitates timely intervention, potentially mitigating long-term neurodevelopmental limitations.

Consistent with literature indicating that preterm children, even those without overt neurological deficits, score lower on cognitive, attention, and executive function tests (21–23), our cohort exhibited developmental vulnerabilities. Using the EDI-II screening tool, we identified delays in gross motor skills, cognition, and language as the most prevalent impairments in preterm preschoolers at 42–48 months old. These results were corroborated by the diagnostic test MSCA, which identified motor, quantitative, verbal and memory as the most vulnerable domains in preschoolers at 48 and 60 months. Our findings align with reports from preterm infant cohorts evaluated with the Bayley-III scales and the INTERGROWTH-21st Neurodevelopment Assessment at 2-years old (19–21, 42), in which motor, cognitive and language impairments were the most frequent outcomes. The persistence of these deficits in our preterm cohort at the preschool period highlights that natural compensatory mechanisms can fail to overcome developmental gaps, as previously shown in other children’s cohorts (43, 44). Our findings with the EDI-II test, highlight the usefulness of this test to quickly and efficiently detect (15 minutes assessment) neurodevelopmental impairments, which is especially important in field work to assess the neurodevelopment of vulnerable child populations (29, 30).

Our initial screening of the preschooler´s immune system status revealed a decrease in neutrophil counts and the Neutrophil-to-Lymphocyte Ratio (NLR) in the Risk group. This parallels findings in cord blood (45) and 2-year-old preterm infants (46), where reduced neutrophils were associated with lower levels of Insulin-like Growth Factor 1 (IGF-1). Moreover, we found reduced platelet counts in children with neurodevelopmental Delay/Risk, and these cells are emerging as critical players in neurodevelopmental disorders such as autism spectrum disorder (ASD) (47). Platelets share common characteristics in subcellular organization and protein composition with neurons (48, 49), thus the link between myeloid lineage suppression and altered neurodevelopment in children warrants further investigation.

Childhood environmental determinants such as nutrition, cognitive stimulation, and indicators of family well-being like maternal educational and socioeconomic status (SES) can act as neuromodulators that shape brain and behavioral development (50–54). In terms of SES and maternal education, our cohort was very homogeneous, most children belonged to low SES families and a maternal schooling median of 12 years. Low SES and maternal education levels have been linked to reduced brain surface area in regions governing language, reading, executive functions and spatial abilities (55). Therefore, while the biological perturbation caused by preterm birth is central, the lack of cognitively stimulating environments, often associated with poverty, may exacerbate these deficits (56, 57).

We previously reported that preterm infants from our cohort had similar feeding practices during hospitalization and all preterm infants were fed human milk (mothers’ own milk or pasteurized donor human milk) either exclusively or in combination with fortified human milk or preterm formula (58). Concerning post-discharge feeding practices, about 80% of preterm infants started solid foods at 6-months corrected age (CA) vs 20% at 6-months chronological age; and 50% of children were fed human milk either exclusively or in combination with formula and 50% were fed exclusively formula at 6-months CA (Unpublished).

At preschool age, the food consumption frequency assessment did not reveal differences between study groups. However, the anthropometric and body composition assessment provided further insight into the systemic nature of the observed developmental impairments. While growth, dietary intake and general metabolic markers appeared similar, we observed a distinct body composition phenotype in children with neurodevelopmental delay and risk. These children exhibited a “leaner phenotype” characterized by reduced abdominal, triceps, and subscapular skinfolds thickness, and lower FM percentage compared with their normal counterparts.

Altered adipose tissue in preterm infants at term-equivalent age, preschool and school age remains controversial (24, 27, 59–62), our findings align with reports observing no excess adiposity in later childhood. Children with neurodevelopmental impairments frequently exhibit reduced FFM and disproportionate fat mass accumulation. Binder et al. reported that higher FFM Z-scores were significantly and independently positively associated with cognitive, language and motor scores in 3-year-old preschool children born extremely preterm (<28 weeks GA) using Bayley-III scales (63). In our cohort, preschool children were born very preterm to moderately preterm (28–34 weeks GA), and we found reduced FM but not in FFM in children with neurodevelopmental impairments. The leaner phenotype observed in children with neurodevelopmental impairments suggests that energy is prioritized toward vital metabolic functions and basic maintenance over adipose storage, or alternatively, reflects distinct adipose compartmentalization programming (64).

Taking together our results suggest that the combination of the EDI-II test, a routine and low-cost lab test such as hematic biometry, and the subscapular skinfold thickness measure hold promise as useful tools for early identification of preterm children at risk of neurodevelopmental impairments.

Regarding immunological status, our study offers novel insights into the “neuro-immune axis” in preterm survivors. While T-cell and NK-cell compartments appeared homeostatically recovered, the B-cell compartment revealed significant alterations that tracked with neurodevelopmental outcomes. Our high-dimensional computational analysis (RAPID/UMAP) identified Cluster 13, a Transitional B-cell phenotype (CD38+CD24+IgD+), as the strongest predictor of the Delay/Risk outcome. Transitional B cells represent an intermediate stage between immature bone marrow B cells and mature naïve B cells; in healthy development, their frequency typically decreases with age (65). The predictive dominance of this immature population in the Delay/Risk group suggests a state of immunological immaturity that parallels the observed neurodevelopmental delays.

The dominance of this immature Transitional B-cell population (CD38+CD24+IgD+) in the Risk/Delay group raises mechanistical questions regarding the neuro-immune axis. Human transitional B (TrB) cells were first described and characterized by a CD24hiCD38hi phenotype (66). Subsequently, based on differential CD21 expression, two subsets of human TrB cells with distinct functions have been described (67). In mice, TrB cells can be further subdivided into three subsets representing different maturation statuses (68). In humans, the CD24hiCD38hi transitional compartment is highly enriched for regulatory B cells that maintain immune tolerance primarily via IL-10 secretion (69). These studies demonstrate that TrB cells are a heterogeneous population at various developmental stages with diverse functional properties. Because optimal neurodevelopment relies on peripheral immune signals, we hypothesize that this expanded transition pool in delayed children may be functionally stalled, failing to mature and provide adequate neuroprotective cytokine support (such as IL-10) to the developing brain.

Conversely, Cluster 12, characterized as Class-Switched Memory B cells (CD27+IgD-), was strongly predictive of Normal development. The presence of these cells reflects a “trained” adaptive immune system that has successfully developed immunological memory, a trajectory appropriate for preschool-aged children. Collectively, these findings suggest that preterm preschoolers with neurodevelopmental impairments exhibit systemic maturational lag. They have not only failed to fully develop motor and cognitive domains but also retain an immature immune signature characterized by high transitional B cells and a relative paucity of the memory B cells seen in normally developed children.

Ultimately, this striking dichotomy between the arrested B-cell signature (cluster 13) in delayed children and the mature, class-switched memory profile (cluster 12) in normally developing peers underscores that preterm neurodevelopmental deficits are linked to a systemic lag in the parallel maturation of the adaptive immune system and the central nervous system.

Cytokine, chemokine and neuroinflammatory molecules assessment revealed reduced levels of IP-10 (CXCL10) in the Risk group. IP-10 is a potent proinflammatory chemokine whose gene expression is primarily and strongly driven by IFN-γ, a cytokine that was undetectable in our cohort together with IL-1β and IL-2. Our findings are consistent with a recent study in which 39 neuroinflammation biomarkers were evaluated in school children born preterm with neurodevelopmental impairments and reports IL-1β and IFN-γ below the detection range (12). The alteration in IP-10, a chemokine involved in Th1 trafficking, suggests that while overt inflammation may have been resolved, subtle dysregulation in immune signaling pathways persists. A negative association of IP-10 and social behaviors has been reported in 3.8–6 years old children with ASD (70).

In this study we performed a comprehensive neurodevelopmental assessment using both screening (EDI-II) and diagnostic (MSCA) tools, alongside a detailed high-dimensional immunological characterization and meticulous body composition analysis. Our results suggest that preterm birth exerts long-term effects not only on neurodevelopment but also on specific immune and body composition profiles. The association between reduced subcutaneous adiposity, the persistence of immature B-cell phenotypes, and neurodevelopmental delays highlights the systemic nature of prematurity. Integrating these body composition and immunological biomarkers into pediatric follow-up could facilitate the early identification of children at risk, enabling timely interventions to support their long-term cognitive and physical health.

5.Conclusion

Preterm preschoolers with neurodevelopmental impairments showed a systemic profile of biological immaturity (reduced subcutaneous adiposity, lower neutrophil and platelet counts and a persistence of transitional B cells). If confirmed in independent and larger cohorts, body composition and B-cell maturation markers could complement developmental screening for the early identification of children at risk.

6.Limitations

The study has several limitations that should be considered for data interpretation. First, the relatively small sample size precluded sex-stratified analyses, and while our cohort size is comparable to previous immunophenotyping and neurodevelopmental assessment studies in preterm infants, these findings require replication in larger cohorts. Second, the classification of neurodevelopmental groups was based on the EDI-II test, a screening instrument validated for Mexican children (29, 30). The validation of EDI-II test was performed with the Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III), and with The Battelle Developmental Inventory, Second Edition (BDI-2) (71), which have been validated and widely used for neurodevelopmental assessment of children from the United States, Latin America and European countries. Third, data on early-intervention therapies (i.e. physical therapy, language stimulation) received by children between hospital discharge and neurodevelopmental assessment at preschool age, were not systematically collected and analyzed. However, most children were followed up after hospital discharge at the Pediatric Follow-up Department at INPer and were included in Early-Stimulation or Rehabilitation programs according to their neurodevelopmental status. Fourth, due to sample size limitations, the LASSO predictive model’s accuracy, sensitivity, and specificity were calculated on the training set rather than an independent hold-out test set; thus, this specific B-cell immune signature requires future validation in an independent longitudinal cohort. Finally, our initial high-dimensional flow cytometry panel was restricted to phenotypic surface markers. We lacked the expanded phenotypic mapping (e.g., CD21, CD44, CD93, CD23), transcriptional profiling, and functional assays (such as autoreactivity or IL-10 cytokine secretion) needed to fully characterize Cluster 13. Determining the specific cytokine profiles and transcriptional regulation of this distinct preterm B-cell subset are immediate avenues for our future mechanistic research in dedicated, newly recruited cohorts.

Acknowledgments

We thank families of the children that participated in this study; Karla Erandi Alcántara Pérez, Laura Victoria Tenorio Arias, Astrid Quintero Pérez, Jimena Paola García Herrera, Rodrigo Tonalli Camacho Pacheco, Ana Sofía Mendoza Obregón, Maricruz Tolentino Dolores, María Cristina Ramírez González, Miranda Molerés Orduña, María Luisa Pizano Zárate and María de Jesús Nava Benitez, for research assistance. Armida Ivette Bahena-Delgado was supported with the PhD scholarship number 815109 provided by the SECIHTI.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. This research was funded by the National Institute of Perinatology “Isidro Espinosa de los Reyes” (grant numbers 2017-2-61 and 2020-1-25), and the Secretary of Science, Humanities, Technology and Innovation (SECIHTI) (grant number SALUD-2017-02-289734).

Footnotes

Edited by: Henner Morbach, University Hospital Würzburg, Germany

Reviewed by: Yunqiao Li, Soochow University, China

Haokun Tian, Capital Institute of Pediatrics, China

Data availability statement

The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors. The R scripts used for the FlowSOM clustering and LASSO regularized regression are available on GitHub at https://github.com/omedinac/TrB-signatures.

Ethics statement

The studies involving humans were approved by the Research, Ethics in Research and Biosecurity Committees of the National Institute of Perinatology "Isidro Espinosa de los Reyes" (project number 2020-1-25, approval January 1st, 2021). The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants’ legal guardians/next of kin.

Author contributions

AB-D: Validation, Investigation, Formal analysis, Visualization, Data curation, Writing – original draft. CH-C: Validation, Methodology, Writing – review & editing, Investigation. SM-M: Validation, Investigation, Writing – review & editing. GG-M: Writing – review & editing, Investigation, Validation. AC-E: Investigation, Writing – review & editing. IM-H: Investigation, Writing – review & editing. CR-H: Investigation, Writing – review & editing. BR-P: Investigation, Writing – review & editing. AR-C: Methodology, Writing – review & editing, Investigation. OP-P: Methodology, Investigation, Writing – review & editing. JM-G: Writing – review & editing, Resources. GE-G: Resources, Writing – review & editing, Investigation. CI: Investigation, Resources, Writing – review & editing. OM-C: Investigation, Writing – original draft, Methodology, Visualization, Writing – review & editing, Formal analysis. GG-P: Conceptualization, Supervision, Formal analysis, Writing – review & editing, Writing – original draft, Investigation, Methodology, Project administration, Funding acquisition, Visualization, Validation, Data curation.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu.2026.1941190/full#supplementary-material

References

1.
De Costa A, Moller AB, Blencowe H, Johansson EW, Hussain-Alkhateeb L, Ohuma EO, et al.
Study protocol for WHO and UNICEF estimates of global, regional, and national preterm birth rates for 2010 to 2019. PloS One. (2021) 16:e0258751. doi:  10.1371/journal.pone.0258751

2.
Quinn JA, Munoz FM, Gonik B, Frau L, Cutland C, Mallett-Moore T, et al.
Preterm birth: Case definition & guidelines for data collection, analysis, and presentation of immunisation safety data. Vaccine. (2016) 34:6047–56. doi:  10.1016/j.vaccine.2016.03.045

3.
Pinto F, Fernandes E, Virella D, Abrantes A, Neto MT.
Born preterm: A public health issue. Port J Publ Health. (2019) 37:38–49. doi:  10.1159/000497249

4.
Vogel JP, Chawanpaiboon S, Moller AB, Watananirun K, Bonet M, Lumbiganon P.
The global epidemiology of preterm birth. Best Pract Res Clin Obstet Gynaecol. (2018) 52:3–12. doi:  10.1016/j.bpobgyn.2018.04.003

5.
Morniroli D, Tiraferri V, Maiocco G, De Rose DU, Cresi F, Coscia A, et al.
Beyond survival: the lasting effects of premature birth. Front Pediatr. (2023) 11:1213243. doi:  10.3389/fped.2023.1213243

6.
Ismail AQT, Boyle EM, Pillay T, Boyle EM, Modi N, Rivero-Arias O, et al.
Clinical outcomes for babies born between 27 – 31 weeks of gestation: Should they be regarded as a single cohort? J Neonatal Nurs. (2023) 29:27–32. doi:  10.1016/j.jnn.2022.04.00342574925

7.
Fleiss B, Gressens P, Stolp HB.
Cortical gray matter injury in encephalopathy of prematurity: link to neurodevelopmental disorders. Front Neurol. (2020) 11:575. doi:  10.3389/fneur.2020.00575

8.
Hagberg H, Mallard C, Ferriero DM, Vannucci SJ, Levison SW, Vexler ZS, et al.
The role of inflammation in perinatal brain injury. Nat Rev Neurol. (2015) 11:192–208. doi:  10.1038/nrneurol.2015.13

9.
Collins JM, Keane JM, Deady C, Khashan AS, McCarthy FP, O’Keeffe GW, et al.
Prenatal stress impacts foetal neurodevelopment: Temporal windows of gestational vulnerability. Neurosci Biobehav Rev. (2024) 164:105793. doi:  10.1016/j.neubiorev.2024.105793

10.
Damiani F, Cornuti S, Tognini P.
The gut-brain connection: Exploring the influence of the gut microbiota on neuroplasticity and neurodevelopmental disorders. Neuropharmacology. (2023) 231:109491. doi:  10.1016/j.neuropharm.2023.109491

11.
Ryan MA, Murray DM, Dempsey EM, Mathieson SR, Livingstone V, Boylan GB.
Neurodevelopmental outcome of low-risk moderate to late preterm infants at 18 months. Front Pediatr. (2023) 11:1256872. doi:  10.3389/fped.2023.1256872

12.
Van der Zwart S, Knol EF, Gressens P, Koopman C, Benders M, Roze E.
Neuroinflammatory markers at school age in preterm born children with neurodevelopmental impairments. Brain Behav Immun Health. (2024) 38:100791. doi:  10.1016/j.bbih.2024.100791

13.
Madrid M, Brianza-Padilla M, Echeverría JC, Rivera-González R, Bojalil R.
The severity of neurological dysfunction in preschool children, secondary to damage generated during the perinatal period, is associated with a pro-inflammatory pattern of serum molecules. Front Immunol. (2020) 11:595309. doi:  10.3389/fimmu.2020.595309

14.
Jiang Q, Wen J, Ding Y, Cui H.
From vulnerability to resilience: unraveling the role of oxidative stress in preterm brain injury. Ital J Pediatr. (2025) 51:232. doi:  10.1186/s13052-025-02079-4

15.
Herz J, Bendix I, Felderhoff-Müser U.
Peripheral immune cells and perinatal brain injury: a double-edged sword? Pediatr Res. (2022) 91:392–403. doi:  10.1038/s41390-021-01818-7

16.
Nazmi A, Albertsson AM, Rocha-Ferreira E, Zhang X, Vontell R, Zelco A, et al.
Lymphocytes contribute to the pathophysiology of neonatal brain injury. Front Neurol. (2018) 9:159. doi:  10.3389/fneur.2018.00159

17.
Marchetti L, Engelhardt B.
Immune cell trafficking across the blood-brain barrier in the absence and presence of neuroinflammation. Vasc Biol. (2020) 2:H1–18. doi:  10.1530/vb-19-0033

18.
Woodward LJ, Hüppi PS.
Neurodevelopmental follow-up. In: Volpe’s Neurology of the Newborn. Philadelphia, PA:
Elsevier; (2018). p. 255–72. doi:  10.1016/B978-0-323-42876-7.00011-9

19.
Bulbul L, Elitok GK, Ayyıldız E, Kabakcı D, Uslu S, Köse G, et al.
Neuromotor Development Evaluation of Preterm Babies Less than 34 Weeks of Gestation with Bayley III at 18-24 Months. BioMed Res Int. (2020) 2020:5480450. doi:  10.1155/2020/5480450

20.
Do CHT, Kruse AY, Wills B, Sabanathan S, Clapham H, Pedersen FK, et al.
Neurodevelopment at 2 years corrected age among Vietnamese preterm infants. Arch Dis Child. (2020) 105:134–40. doi:  10.1136/archdischild-2019-316967

21.
Battajon N, Bechini C, De Osti F, Galletti A, Frigo AC, Lago P.
Neurodevelopmental outcomes of very low birth weight preterms in preschool childhood: a prospective cohort study. Ital J Pediatr. (2023) 49:56. doi:  10.1186/s13052-023-01467-y

22.
Pereira-Cerro AV, Lanzarote-Fernández MD, Barbancho-Morant MM, Padilla-Muñoz EM.
Evolution of psychomotor development in pre-school children born prematurely. Pediatr (Engl Ed). (2020) 93:228–35. doi:  10.1016/j.anpede.2019.10.010

23.
Sanchez-Joya MDM, Sanchez-Labraca N, Roldan-Tapia MD, Moral Rodríguez T, Ramos Lizana J, Roman P.
Neuropsychological assessment and perinatal risk: A study amongst very premature born 4- and 5-year old children. Res Dev Disabil. (2017) 69:116–23. doi:  10.1016/j.ridd.2017.08.008

24.
Casirati A, Somaschini A, Perrone M, Vandoni G, Sebastiani F, Montagna E, et al.
Preterm birth and metabolic implications on later life: A narrative review focused on body composition. Front Nutr. (2022) 9:978271. doi:  10.3389/fnut.2022.978271

25.
Madrid M, Bojalil R, Brianza-Padilla M, Zapoteco-Nava J, Márquez-Velasco R, Rivera-González R.
The molecular profile of the inflammatory process differs among various neurodevelopmental disorders with or without cognitive component: A hypothesis of persistent systemic dysfunction and hyper-resolution. Front Pediatr. (2023) 11:1132175. doi:  10.3389/fped.2023.1132175

26.
Bortolotto CC, Santos IS, Dos Santos Vaz J, Matijasevich A, Barros AJD, Barros FC, et al.
Prematurity and body composition at 6, 18, and 30 years of age: Pelotas (Brazil) 2004, 1993, and 1982 birth cohorts. BMC Public Health. (2021) 21:321. doi:  10.1186/s12889-021-10368-w

27.
Viswanathan S, Thoene M, Alja’nini Z, Alur P, McNelis K.
Body composition in preterm infants: current insights and emerging perspectives.
MDPI (Multidisciplinary Digital Publishing Institute), Basel, Switzerland, vol. 12. (2025). doi:  10.3390/children12010053.

28.
Pfister KM, Zhang L, Miller NC, Ingolfsland EC, Demerath EW, Ramel SE.
Early body composition changes are associated with neurodevelopmental and metabolic outcomes at 4 years of age in very preterm infants. Pediatr Res. (2018) 84:713–8. doi:  10.1038/s41390-018-0158-x

29.
Serván-Mori E, Fuentes-Rivera E, Quezada AD, Pineda-Antunez C, Del Carmen Hernández-Chávez M, García-Martínez A, et al.
Early neurological development and nutritional status in Mexican socially deprived contexts. PloS One. (2022) 17:e0270085. doi:  10.1371/journal.pone.0270085

30.
Quezada-Sánchez AD, Vesterinen MS, Kinnunen TI, Hernández-Chávez C, García-Martínez A, Madrigal-Ramírez A, et al.
Maternal depressive symptomatology and its association with early child development in socioeconomically deprived Mexican households. Cham:
Springer Nature, vol. 6. (2026). doi:  10.1007/s44192-026-00477-6.

31.
Schrader A, D’Amato RC.
McCarthy scales of children’s abilities. In: Encyclopedia of Clinical Neuropsychology.
Springer New York, New York, NY: (2011). p. 1531–2.

32.
Kupsco A, Estrada-Gutierrez G, Cantoral A, Schnaas L, Pantic I, Amarasiriwardena C, et al.
Modification of the effects of prenatal manganese exposure on child neurodevelopment by maternal anemia and iron deficiency. Pediatr Res. (2020) 88:325–33. doi:  10.1038/s41390-020-0754-4

33.
Williams ME, Corn EA, Martinez Ransanz S, Berl MM, Andringa-Seed R, Mulkey SB.
Neurodevelopmental assessments used to measure preschoolers’ cognitive development in Latin America: a systematic review. J Pediatr Psychol. (2024) 49:321–39. doi:  10.1093/jpepsy/jsad089

34.
Pelletiet D.
Anthropometric standardization reference manual: Abridged edition. In:
Lohman TG, Roche AF, Martorell R, editors. Anthropometric Standardization Reference Manual: Abridged Edition.
Human Kinetics Books, Champaign, IL: (1991). p. vi–90.

35.
Rodríguez-Cano AM, Piña-Ramírez O, Rodríguez-Hernández C, Mier-Cabrera J, Villalobos-Alcazar G, Estrada-Gutierrez G, et al.
Development and validation of anthropometric-based fat-mass prediction equations using air displacement plethysmography in Mexican infants. Eur J Clin Nutr. (2023) 77:748–56. doi:  10.1038/s41430-023-01285-9

36.
Aristizabal JC, Barona-Acevedo J, Estrada-Restrepo A.
Correlation of body mass index and waist to height ratio with cardiovascular risk factors in Colombian preschool and school children. Colomb Med (Cali). (2023) 54:e2014113. doi:  10.25100/cm.v54i1.4113

37.
Valle-Leal J, Abundis-Castro L, Hernández-Escareño J, Flores-Rubio S.
Waist-to-height ratio is an indicator of metabolic risk in children. Rev Chil Pediatr. (2016) 87:180–5. doi:  10.1016/j.rchipe.2015.10.011

38.
Slaughter MH, Lohman TG, Boileau RA, Horswill CA, Stillman RJ, Van Loan MD, et al.
Skinfold equations for estimation of body fatness in children and youth. Hum Biol. (1988) 60:709–23.

39.
Friedman J, Hastie T, Tibshirani R.
Regularization paths for generalized linear models via coordinate descent. J Stat Softw. (2010) 33:1–22. doi:  10.18637/jss.v033.i01

40.
Shaw RJ, Givrad S, Poe C, Loi EC, Hoge MK, Scala M.
Neurodevelopmental, mental health, and parenting issues in preterm infants. MDPI (Multidisciplinary Digital Publishing Institute), Switzerland:
Basel, vol. 10. (2023).doi:  10.3390/children10091565.

41.
Vanes LD, Murray RM, Nosarti C.
Adult outcome of preterm birth: Implications for neurodevelopmental theories of psychosis. Schizophr Res. (2022) 247:41–54. doi:  10.1016/j.schres.2021.04.007

42.
Villar J, Restrepo-Méndez MC, McGready R, Barros FC, Victora CG, Munim S, et al.
Association between preterm-birth phenotypes and differential morbidity, growth, and neurodevelopment at age 2 years: results from the INTERBIO-21st newborn study. JAMA Pediatr. (2021) 175:483–93. doi:  10.1001/jamapediatrics.2020.6087

43.
Shevell MI, Majnemer A, Webster RI, Platt RW, Birnbaum R.
Outcomes at school age of preschool children with developmental language impairment. Pediatr Neurol. (2005) 32:264–9. doi:  10.1016/j.pediatrneurol.2004.12.008

44.
Willoughby MT, Magnus B, Vernon-Feagans L, Blair CB, Family Life Project Investigators .
Developmental delays in executive function from 3 to 5 years of age predict kindergarten academic readiness. J Learn Disabil. (2017) 50:359–72. doi:  10.1177/0022219415619754

45.
Peterson LS, Hedou J, Ganio EA, Stelzer IA, Feyaerts D, Harbert E, et al.
Single-cell analysis of the neonatal immune system across the gestational age continuum. Front Immunol. (2021) 12:714090. doi:  10.3389/fimmu.2021.714090

46.
Ibáñez L, Ong K, Dunger DB, de Zegher F.
Early development of adiposity and insulin resistance after catch-up weight gain in small-for-gestational-age children. J Clin Endocrinol Metab. (2006) 91:2153–8. doi:  10.1210/jc.2005-2778

47.
Padmakumar M, Van Raes E, Van Geet C, Freson K.
Blood platelet research in autism spectrum disorders: In search of biomarkers. Res Pract Thromb Haemost. (2019) 3:566–77. doi:  10.1002/rth2.12239

48.
Burnouf T, Walker TL.
The multifaceted role of platelets in mediating brain function. Blood. (2022) 140:815–27. doi:  10.1182/blood.2022015970

49.
Leiter O, Walker TL.
Platelets: The missing link between the blood and brain? Prog Neurobiol. (2019) 183:101695. doi:  10.1016/j.pneurobio.2019.101695

50.
Tooley UA, Bassett DS, Mackey AP.
Environmental influences on the pace of brain development. Nat Rev Neurosci. (2021) 22:372–84. doi:  10.1038/s41583-021-00457-5

51.
Thomas MSC, Coecke S.
Associations between socioeconomic status, cognition, and brain structure: Evaluating potential causal pathways through mechanistic models of development. Cognit Sci. (2023) 47:e13217. doi:  10.1111/cogs.13217

52.
Rosales FJ, Reznick JS, Zeisel SH.
Understanding the role of nutrition in the brain and behavioral development of toddlers and preschool children: Identifying and addressing methodological barriers. Nutr Neurosci. (2009) 12:190–202. doi:  10.1179/147683009x423454

53.
Bonthrone AF, Miller SP, Boardman JP.
The intersection of neuroscience, socioeconomic status, and preterm birth. Pediatr Res. (2026). doi:  10.1038/s41390-026-05100-6.

54.
Rakesh D, McLaughlin KA, Sheridan M, Humphreys KL, Rosen ML.
Environmental contributions to cognitive development: The role of cognitive stimulation. Dev Rev. (2024) 73:101135. doi:  10.1016/j.dr.2024.101135.

55.
Noble KG, Houston SM, Brito NH, Bartsch H, Kan E, Kuperman JM, et al.
Family income, parental education and brain structure in children and adolescents. Nat Neurosci. (2015) 18:773–8. doi:  10.1038/nn.3983

56.
Ballot DE, Ramdin T, Rakotsoane D, Agaba F, Davies VA, Chirwa T, et al.
Use of the Bayley Scales of Infant and Toddler Development, Third Edition, to assess developmental outcome in infants and young children in an urban setting in South Africa. Int Sch Res Notices. (2017) 2017:1631760. doi:  10.1155/2017/1631760

57.
Villagomez AN, Muñoz FM, Peterson RL, Colbert AM, Gladstone M, MacDonald B, et al.
Neurodevelopmental delay: Case definition & guidelines for data collection, analysis, and presentation of immunization safety data. Vaccine. (2019) 37:7623–41. doi:  10.1016/j.vaccine.2019.05.027

58.
Reyes-García DV, Canul-Euan AA, Rivera-Rueda MA, Cruz-Alvarado CE, Bermejo-Martínez LB, Arreola-Ramírez G, et al.
Neonatal antibiotic treatment can affect stool pattern and oral tolerance in preterm infants. Life (Basel). (2022) 12(7):1043. doi:  10.3390/life12071043.

59.
Uthaya S, Thomas EL, Hamilton G, Doré CJ, Bell J, Modi N.
Altered adiposity after extremely preterm birth. Pediatr Res. (2005) 57:211–5. doi:  10.1203/01.pdr.0000148284.58934.1c

60.
Roggero P, Giannì ML, Forzenigo L, Tondolo T, Taroni F, Liotto N, et al.
No relative increase in intra-abdominal adipose tissue in healthy unstressed preterm infants at term. Neonatology. (2015) 107:14–9. doi:  10.1159/000364855

61.
Giannì ML, Mora S, Roggero P, Amato O, Piemontese P, Orsi A, et al.
Regional fat distribution in children born preterm evaluated at school age. J Pediatr Gastroenterol Nutr. (2008) 46:232–5. doi:  10.1097/MPG.0b013e31814d4df9

62.
Nehab SRG, Villela LD, Soares FVM, de Abranches AD, Junior SCG, Moreira MEL.
Body composition at 4 to 7 years of age in children born <32 weeks gestational age or 1500 g: A cohort study. Early Hum Dev. (2022) 173:105659. doi:  10.1016/j.earlhumdev.2022.105659

63.
Binder C, Calek E, Thajer A, Harreiter K, Longford N, Fuiko R, et al.
Fat-free mass is associated with neurodevelopment outcomes in extremely preterm infants up to 3 years of age. Pediatr Res. (2025) 100(1):249–56.doi:  10.1038/s41390-025-04557-1.

64.
O A, M U, B LF, GC A.
Energy metabolism in childhood neurodevelopmental disorders. EBioMedicine. (2021) 69:103474. doi:  10.1016/j.ebiom.2021.103474

65.
Marie-Cardine A, Divay F, Dutot I, Green A, Perdrix A, Boyer O, et al.
Transitional B cells in humans: Characterization and insight from B lymphocyte reconstitution after hematopoietic stem cell transplantation. Clin Immunol. (2008) 127:14–25. doi:  10.1016/j.clim.2007.11.013

66.
Sims GP, Ettinger R, Shirota Y, Yarboro CH, Illei GG, Lipsky PE.
Identification and characterization of circulating human transitional B cells. Blood. (2005) 105:4390–8. doi:  10.1182/blood-2004-11-4284

67.
Suryani S, Fulcher DA, Santner-Nanan B, Nanan R, Wong M, Shaw PJ, et al.
Differential expression of CD21 identifies developmentally and functionally distinct subsets of human transitional B cells. Blood. (2010) 115:519–29. doi:  10.1182/blood-2009-07-234799

68.
Allman D, Lindsley RC, DeMuth W, Rudd K, Shinton SA, Hardy RR.
Resolution of three nonproliferative immature splenic B cell subsets reveals multiple selection points during peripheral B cell maturation. J Immunol. (2001) 167:6834–40. doi:  10.4049/jimmunol.167.12.6834

69.
Blair PA, Noreña LY, Flores-Borja F, Rawlings DJ, Isenberg DA, Ehrenstein MR, et al.
CD19(+)CD24(hi)CD38(hi) B cells exhibit regulatory capacity in healthy individuals but are functionally impaired in systemic lupus erythematosus patients. Immunity. (2010) 32:129–40. doi:  10.1016/j.immuni.2009.11.009

70.
Shen Y, Ou J, Liu M, Shi L, Li Y, Xiao L, et al.
Altered plasma levels of chemokines in autism and their association with social behaviors. Psychiatry Res. (2016) 244:300–5. doi:  10.1016/j.psychres.2016.07.057

71.
Rizzoli-Córdoba A, Schnaas-Arrieta L, Liendo-Vallejos S, Buenrostro-Márquez G, Romo-Pardo B, Carreón-García J, et al.
Validación de un instrumento para la detección oportuna de problemas de desarrollo en menores de 5 años en México Bol. Med. Hosp. Infant. Mex (2013). Available online at: http://www.scielo.org.mx/scielo.php?script=sci_arttext&pid=S1665-11462013000300003&lng=es (Accessed July 17, 2026).

Associated Data

Supplementary Materials

Data Availability Statement

The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors. The R scripts used for the FlowSOM clustering and LASSO regularized regression are available on GitHub at https://github.com/omedinac/TrB-signatures.