Skip to main content
Download PDF

Epigenetic scores derived in saliva are associated with gestational age at birth

Clinical Epigenetics volume 16, Article number: 84 (2024) Cite this article

Abstract

Background

Epigenetic scores (EpiScores), reflecting DNA methylation (DNAm)-based surrogates for complex traits, have been developed for multiple circulating proteins. EpiScores for pro-inflammatory proteins, such as C-reactive protein (DNAm CRP), are associated with brain health and cognition in adults and with inflammatory comorbidities of preterm birth in neonates. Social disadvantage can become embedded in child development through inflammation, and deprivation is overrepresented in preterm infants. We tested the hypotheses that preterm birth and socioeconomic status (SES) are associated with alterations in a set of EpiScores enriched for inflammation-associated proteins.

Results

In total, 104 protein EpiScores were derived from saliva samples of 332 neonates born at gestational age (GA) 22.14 to 42.14 weeks. Saliva sampling was between 36.57 and 47.14 weeks. Forty-three (41%) EpiScores were associated with low GA at birth (standardised estimates |0.14 to 0.88|, Bonferroni-adjusted p-value < 8.3 × 10−3). These included EpiScores for chemokines, growth factors, proteins involved in neurogenesis and vascular development, cell membrane proteins and receptors, and other immune proteins. Three EpiScores were associated with SES, or the interaction between birth GA and SES: afamin, intercellular adhesion molecule 5, and hepatocyte growth factor-like protein (standardised estimates |0.06 to 0.13|, Bonferroni-adjusted p-value < 8.3 × 10−3). In a preterm subgroup (n = 217, median [range] GA 29.29 weeks [22.14 to 33.0 weeks]), SES–EpiScore associations did not remain statistically significant after adjustment for sepsis, bronchopulmonary dysplasia, necrotising enterocolitis, and histological chorioamnionitis.

Conclusions

Low birth GA is substantially associated with a set of EpiScores. The set was enriched for inflammatory proteins, providing new insights into immune dysregulation in preterm infants. SES had fewer associations with EpiScores; these tended to have small effect sizes and were not statistically significant after adjusting for inflammatory comorbidities. This suggests that inflammation is unlikely to be the primary axis through which SES becomes embedded in the development of preterm infants in the neonatal period.

Graphical abstract

Background

Preterm birth (delivery <37 weeks' gestation) affects around 10% of births worldwide and is closely associated with increased likelihood of cerebral palsy, neurocognitive impairment, behavioural, social and communication difficulties, and mental and cardiometabolic health diagnoses across the life course [1,2,3,4,5]. These adverse outcomes can be explained, in part, by deleterious effects of early exposure to extrauterine life on brain and cardiac development, and they are often accompanied by changes in blood proteins, including those reflecting the perinatal innate and adaptive immune response [6,7,8].

Socioeconomic status (SES) is also associated with the adverse neurodevelopmental and health outcomes listed above [9,10,11,12], and social deprivation is consistently overrepresented among preterm children and their families [13, 14]. In a meta-analysis of 43 studies (n = 111,156 individuals), low SES associated with increased inflammatory markers of disease risk (C-reactive protein [CRP] and interleukin-6 [IL6]), which suggests that pro-inflammatory pathways may be important mechanisms for translating social inequalities into health disparities [15]. However, only four studies included participants under 10 years of age, leaving uncertainty about SES-inflammation correlations in early life [16,17,18,19].

Although protein levels are commonly used as biomarkers of exposure and disease risk, their use is  limited because they are often phasic in the systemic circulation, rely on venepuncture, and may not capture baseline status or chronicity. For example, inflammation is often measured using acute-phase inflammatory proteins such as CRP [20, 21], but it is not always reliable [22], particularly in neonates, and a single-time-point measure may not reflect baseline inflammation or capture chronic inflammation [23]. These challenges have been addressed by the development of DNA methylation (DNAm) markers of protein expression (EpiScores), which are derived from a linear weighted sum of DNAm sites that are correlated with protein levels. Several EpiScores are associated with magnetic resonance imaging (MRI) measures of brain health, cognition, child mental health, stroke, ischaemic heart disease, Alzheimer’s disease, and lung cancer [24,25,26,27,28,29,30,31]. In neonates, DNAm CRP is associated with birth gestational age (GA), perinatal inflammatory processes, and MRI features of encephalopathy of prematurity [32]. Childhood SES is associated with differential DNAm in inflammation-related genes [33, 34] and at CpG sites that correlate with an inflammation index [35]. Adult SES and social mobility are associated with variations in DNAm in inflammation-related genes [33, 36]. Importantly, SES-related DNAm variations are associated with differences in gene expression so may have functional consequences [33, 36].

Several maternal factors are associated with DNAm in term infants sampled soon after birth, including maternal smoking [37], diabetes [38, 39], obesity [40, 41], and mode of delivery [42, 43]. It is unknown whether these associations apply in preterm infants, who have a curtailed in utero exposure, and are sampled after prolonged exposure to neonatal intensive care, which is known to have widespread effects on the methylome [44].

We investigated relationships between preterm birth, SES, and 104 EpiScores enriched for inflammation-related proteins [26,27,28, 31, 45]. We tested the following hypotheses: First, low GA is associated with differences in EpiScores; and second, SES is correlated with EpiScores, and interacts with birth GA, but the relationship is attenuated by inflammatory disease burden in preterm infants.

Methods

Participants

Participants were preterm infants (born \(\le\) 33-weeks' gestation) and term-born infants born at the Royal Infirmary of Edinburgh, UK. These infants were recruited to a longitudinal cohort study designed to investigate the effect of preterm birth on brain development and outcomes with multimodal data collection [46]. Infants were recruited between February 2012 and December 2021.

Exclusion criteria were congenital malformation, chromosomal abnormality, congenital infection, cystic periventricular leukomalacia, haemorrhagic parenchymal infarction, and post-haemorrhagic ventricular dilatation. These criteria mean the cohort is representative of the majority of survivors of modern intensive care practices [46].

Final participants included were 217 preterm infants (born \(\le\) 33-weeks' gestation) and 115 term-born infants, with median birth GA of 29.29 and 39.71 weeks, respectively. Their demographic characteristics are shown in Table 1. The three SES measures (Scottish Index of Multiple Deprivation (SIMD 2016) [47], maternal education, and maternal occupation) differed between the preterm and term groups (Cohen’s d effect sizes 0.52–0.68). Ethnicity did not differ between groups and is representative of the Edinburgh area [48].

Table 1 Participant characteristics
Full size table

DNA methylation

Saliva samples for DNAm were collected at term equivalent age using Oragene OG-575 Assisted Collection kits (DNA Genotek, ON, Canada), and DNA was extracted using prepIT.L2P reagent (DNA Genotek, ON, Canada). Saliva sampling was used due to accessibility and the non-invasiveness of the method; DNAm patterns measured via saliva samples correlate with brain and other tissue DNAm patterns [49, 50]. We chose to sample at the term equivalent gestation time point to include the allostatic load of both prenatal and early postnatal exposures.

DNA was bisulphite converted and methylation levels were measured using Illumina HumanMethylationEPIC BeadChip (Illumina, San Diego, CA, USA) at the Edinburgh Clinical Research Facility (Edinburgh, UK). The arrays were imaged on the Illumina iScan or HiScan platform, and genotypes were called automatically using GenomeStudio Analysis software version 2011.1 (Illumina). DNAm was processed in four batches.

Raw intensity (.idat) files were read into the R environment using minfi. wateRmelon and minfi were used for preprocessing, quality control, and normalisation [51]. The pfilter function in wateRmelon was used to exclude samples with 1% of sites with a detection p-value > 0.05, sites with beadcount < 3 in 5% of samples, and sites with 1% of samples with detection p-value > 0.05. Cross-hybridising probes, probes targeting single-nucleotide polymorphisms with overall minor allele frequency ≥ 0.05, and control probes were also removed. Samples were removed if there was a mismatch between predicted sex (minfi) and recorded sex (n = 3), or if samples did not meet preprocessing quality control criteria (n = 29). Data were danet normalised, which includes background correction and dye bias correction [51]. Saliva contains different cells types, including buccal epithelial cells. Epithelial cell proportions were estimated with epigenetic dissection of intra-sample heterogeneity with the reduced partial correlation method implemented in the R package EpiDISH [52]. Probes located on sex chromosomes were removed before analysis. The cohort includes twins (n = 32); these were randomly removed leaving one participant per twin pair. This left a final sample size of n = 332.

EpiScore calculation

The 104 protein EpiScores included 100 EpiScores from Gadd et al. [26], a study enriched for inflammatory-related proteins, excluding those where the required CpGs were not available, owing to differences in assay platform CpG coverage relative to Gadd et al. [26]. For duplicate proteins, those developed using Olink platform-identified proteins (antibody-based assays) were prioritised over those from SOMAscan platforms (aptamer-based assays), due to specificity and reproducibility, and the variable correlation between the two methods [53,54,55,56] (for details see Supplementary eMethods, Additional File 1). In addition, we included EpiScores for IL6 [28], growth and differentiation factor 15 (GDF15) and N-terminal-pro B-type natriuretic peptide (NTproBNP) [45], and CRP. The CRP EpiScore used was Barker et al.’s seven-CpG variation of Ligthart et al.’s CRP EpiScore [27, 31], as this is known to correlate with birth GA, perinatal pro-inflammatory exposures, and neonatal brain development [32].

For each individual, EpiScores were obtained by multiplying the methylation proportion at a given CpG by the effect size from previous studies. This was performed using the MethylDetectR platform [57] for those inflammatory proteins currently included and using R for those not currently included (CRP, GDF15, IL6, NTproBNP). All CpG sites and coefficients required to calculate the 104 EpiScores are in Supplementary Table 1 (Additional File 2).

Statistics

The predictor variables were SES and birth GA. SES was operationalised in three ways: neighbourhood-level SES using the Scottish Index of Multiple Deprivation (SIMD) [47], and two measures of family-level SES, which were maternal education (highest educational qualification) and maternal occupation (current or most recent occupation). For further details, see Supplementary eMethods, Additional File 1. Birth GA was a continuous variable to maximise statistical power [58, 59]. We adjusted for GA at saliva sampling, DNAm batch, infant sex, and birthweight z-score.

All statistical analyses were performed in R (version 4.3.1) and were preregistered [60].

Principal component analysis (PCA) was used to determine the significance threshold for controlling type 1 error in analyses of multiple EpiScores [61]. We began with a correlation analysis, which showed correlation coefficients between EpiScores of |0.01 to 0.93| (Fig. 1A). To determine the number of statistical “families” among the 104 EpiScores, PCA was performed. This yielded two principal components with eigenvalues > 1, our pre-specified threshold, which explained 59.5 and 17.2% of variance, respectively (Fig. 1B). Standardised component loadings are provided in Supplementary Table 2 (Additional File 1). In all subsequent analyses, we corrected for multiple comparisons across EpiScores and SES measures using a Bonferroni-adjusted p-value threshold of 8.3 × 10−3. This is 0.05/(2 × 3), with two reflecting the two principal components for EpiScores and three reflecting the number of SES measures used.

Fig. 1
figure 1

Determining significance threshold. Principal component analysis was used to determine the adjusted statistical significance threshold, given multiple statistical comparisons. A A correlation matrix of 104 EpiScores, showing correlation coefficient as red for positive and blue negative associations when significant (p < 0.05). B A scree plot of principal components, with the eigenvalues for each component. Standardised component loadings for principal components one and two are provided in Supplementary Table 3 (Additional File 1)

Full size image

We constructed general linear regression models for each EpiScore as outcome measure to assess associations between GA, each of the three SES measures (separate models for each of SIMD, maternal education, and maternal occupation), and the product interaction term SES*birth GA (removing the term if not significant), and adjusting for GA at sampling, sex, and batch.

For the preterm subgroup, we additionally adjusted for perinatal inflammatory exposures known to be associated with the CRP EpiScore as, to our knowledge, this is the only DNAm proxy of an inflammatory protein that has been studied in this context [32]. These were histological chorioamnionitis (HCA), sepsis, bronchopulmonary dysplasia (BPD), and necrotising enterocolitis (NEC). Maternal smoking and preeclampsia were not associated with DNAm CRP, so are not included as covariates [32, 44]. For definitions see Supplementary eMethods (Additional File 1) and for frequencies see Supplementary Table 3 (Additional File 1).

For EpiScores with significant associations with GA or SES, we performed a post hoc sensitivity analysis, adjusting for maternal factors that have been associated with neonatal methylome in term infants in prior research: maternal smoking, diabetes, obesity, and mode of delivery. A change of standardised \(\beta\) by \(\ge\) 20% or change of p-value to \(\ge\) 0.05 was considered significant, and we report adjusted R2 values of each model. For definitions, see Supplementary eMethods (Additional File 1), and for frequencies, see Table 1.

Results

Associations between gestational age and EpiScores

Gestational age associated with 43 of the 104 EpiScores after adjustment for SIMD, maternal education, or maternal occupation (Fig. 2A–C). The proteins represented by the 43 EpiScores are listed in Table 2 categorised by functional annotation adapted from the STRING database [62], and their broader roles in immune processes and inflammation, and the pathogenesis of neonatal diseases, where known, are described in Supplementary Table 4 (Additional File 1).

Fig. 2
figure 2

EpiScores associated with gestational age in regression models adjusted for socioeconomic status. EpiScores associated with gestational age in regression models with A Scottish Index of Multiple Deprivation, B maternal education, and C maternal occupation. A (n = 331) shows 39 associations, B (n = 323) shows 35 associations, and C shows (n = 328) shows 39 associations. Points and bars represent standardised beta and 95% confidence intervals, with red indicating positive and blue negative associations. Covariates included in all models: age at sample, birthweight z-score, sex, and methylation processing batch. Bonferroni-adjusted p-value < 8.3 × 10−3 CCL11 C-C chemokine 11, CCL18 C-C chemokine 18, CCL21 C-C chemokine 21, CCL22 C-C chemokine 22, CCL25 C-C chemokine 25, CD5L CD5 antigen-like protein, CD6 T-cell differentiation antigen, CD163 scavenger receptor cysteine-rich type 1 protein M130, CI confidence interval, CRP C-reactive protein, CRTAM cytotoxic and regulatory T-cell molecule, CXCL9 C-X-C motif chemokine 9, CXCL10 C-X-C motif chemokine 10, FAP fibroblast activation protein alpha, FCGR3B low-affinity immunoglobulin gamma Fc region receptor III-B, FcRL2 Fc receptor-like protein 2, FGF21 Fibroblast growth factor 21, GDF15 growth/differentiation factor 15, GHR growth hormone receptor, HCII heparin cofactor II, HGF hepatocyte growth factor alpha chain, IGFBP4 insulin-like growth factor-binding protein 4, MMP9 matrix metalloproteinase-9, NCAM1 neural cell adhesion molecule 1, PIGR polymeric immunoglobulin receptor, SCGF stem cell growth factor, SIMD Scottish Index of Multiple Deprivation, SKR3 serine/threonine protein kinase receptor R3, SLITRK5 SLIT and NTRK-like protein 5, VCAM1 vascular cell adhesion protein 1, VEGFA vascular endothelial growth factor A, WFIKKN2 WAP Kazal immunoglobulin Kunitz and NTR domain-containing protein 2

Full size image
Table 2 Protein EpiScores associated with birth gestational age
Full size table

Twenty-nine (67%) EpiScores negatively associated with birth GA (standardised estimates |0.14–0.76|, adjusted p-value < 8.3 × 10−3), and 14 (33%) EpiScores positively associated with birth GA (standardised estimates 0.14–0.88, adjusted p-value < 8.3 × 10−3).

Thirty-three EpiScores associated with low GA irrespective of SES measure used in the model. The results for all 104 EpiScores are provided in Supplementary Figs. 1–9 (Additional File 1).

Associations between SES, EpiScores, and the effect of inflammatory comorbidities of preterm birth.

Three out of 104 EpiScores associated with SES measures or the interaction between SES and GA (Fig. 3). There was a small effect size association between higher afamin EpiScore and higher maternal occupation (standardised \(\beta\) = 0.06, 95% confidence interval (CI) 0.02–0.11, p = 0.0082), and DNAm afamin associated with the birth GA*maternal education interaction term such that afamin positively correlated with birth GA among babies with mothers without university education, and negatively correlated with birth GA among babies with mothers with university education (undergraduate or postgraduate) (standardised \(\beta\) = , 95% CI − 0.20 to − 0.04, p = 0.0041, Supplementary Fig. 10, Additional File 1).

Fig. 3
figure 3

EpiScores associated with socioeconomic status or an interaction between socioeconomic status and birth gestational age. EpiScores associated with socioeconomic status (Scottish Index of Multiple Deprivation, maternal education, or maternal occupation), or with an interaction between socioeconomic status and birth gestational age. Regression models with gestational age at birth, gestational age at sample, birthweight z-score, sex, and methylation processing batch. Sample sizes: for the Scottish Index of Multiple Deprivation n = 331, for maternal education n = 323, and for maternal occupation n = 328. 3/104 EpiScores were significant (Bonferroni-adjusted p-value < 8.3 × 10−3). Points and bars represent standardised beta and 95% confidence intervals, with red indicating positive and blue negative associations. CI Confidence interval, GA Gestational age, HGFI Hepatocyte growth factor-like protein alpha chain, ICAM5 Intercellular adhesion molecule 5, SIMD Scottish Index of Multiple Deprivation

Full size image

Higher intercellular adhesion molecule 5 (ICAM5) EpiScore associated with higher SIMD (standardised \(\beta\) = 0.13, 95% CI 0.03–0.23, p = 0.0079). Hepatocyte growth factor-like protein (HGFI) associated with the birth GA*maternal occupation interaction term (standardised \(\beta\) = −0.10, 95% CI − 0.16– − 0.04, p = 0.0021, Supplementary Fig. 11, Additional File 1), such that HGFI EpiScore positively correlated with birth GA among babies with mothers who were unemployed, homemakers, in full time education, or in unskilled or manual occupations, but negatively correlated with birth GA among babies with mothers in partly skilled, non-manual skilled or professional occupations.

In the planned analysis of preterm-born babies only, when controlling for inflammatory exposures (sepsis, HCA, NEC, and BPD), a greater proportion of R2 was explained (adjusted R2 = 0.031–0.262 in unadjusted models, adjusted R2 = 0.047–0.392 in adjusted models), but the EpiScores no longer met our statistical threshold (p-values > 0.045 with adjusted p-value threshold < 8.3 × 10−3, see Supplementary Table 5, Additional File 1).

Sensitivity analyses

There were few changes to the significant associations between 43 EpiScores and GA, when models were adjusted for maternal smoking, diabetes, obesity, and mode of delivery (see Supplementary Table 6, Additional file 1). There were minor changes in proportion of R2 explained (change in adjusted R2 =|0.0001–0.051|) and change in standardised \(\beta\) was by 0–12%. All adjusted models retained the threshold of p < 0.05, although 4/43 (6.2%) EpiScores no longer met the adjusted p-value threshold < 8.3 × 10−3.

For associations between EpiScores and SES or an interaction between SES and birth GA, there was change in adjusted R2 = |0.002–0.004| and in standardised \(\beta\) by 6–11% (see Supplementary Table 6, Additional file 1). All adjusted models retained p < 0.05, although 3/4 models no longer met the adjusted p-value threshold < 8.3 × 10−3.

Discussion

In this study, we identified several associations between a set of EpiScores enriched for inflammatory proteins and low GA at birth. Few EpiScores associated with SES within the whole sample and these associations were partially attenuated in preterm infants who experienced inflammatory comorbidities. This is the first study to assess the impact of preterm birth and social status using epigenetic signatures designed to reflect the circulating proteome.

Associations between birth GA and EpiScores

43 EpiScores associated with preterm birth when sampled at term equivalent age. The EpiScores reflect chemokines, growth factors, proteins required for neurogenesis and vascular development, cell membrane proteins and receptors, and immune response proteins (Table 2). As well as having specific immunoregulatory roles, in the neonatal setting or relevant animal models, these proteins are associated with several comorbidities and developmental consequences of preterm birth. These include lung development and disease such as BPD [63,64,65,66,67,68,69,70,71,72,73,74,75], in utero and postnatal growth failure [76,77,78,79], HCA [7, 80, 81], patent ductus arteriosus [82,83,84,85], retinopathy of prematurity [86,87,88,89,90], NEC [91,92,93], hyperglycaemia [94], sepsis [93, 95,96,97,98], brain injury [32, 99,100,101,102], and neurodevelopmental outcomes [103,104,105,106,107].

Associations between SES measures and EpiScores

SES appears to play a much smaller role in the patterning of EpiScores compared to birth GA. SES measures, or interactions between birth GA and SES measures, were associated with only three of the 104 EpiScores studied: afamin, ICAM5, and HGFI. Afamin and ICAM5 positively associated with maternal occupation and SIMD, respectively. Afamin and HGFI both associated with the interaction term between SES and birth GA. Afamin is involved in vitamin E transport [108], ICAM5 has a role in microglial regulation [109], and HGFI is a macrophage-stimulating protein [110]. The relationships between these proteins and SES have not previously been investigated, although afamin is associated with the development of metabolic syndrome [111], which varies with SES [112].

Among preterm infants, no SES-EpiScore associations survived adjustment for inflammatory exposures, which suggests that the weak effects of SES on the neonatal proteome that we observed in a small number of EpiScores are at least partially accounted for by inflammatory pathologies in early life. Taken together, the results suggest immune dysregulation, proxied by EpiScores, may not be the primary axis through which SES becomes embedded in the development of preterm infants during neonatal intensive care.

SES has been consistently associated with inflammation in adulthood, including in relation to childhood deprivation [15, 113], but less is known about the relationship between SES and inflammation in the neonatal period. A longitudinal study by Leviton et al. [114], with five sampling time points during the first month of life after preterm birth, showed that maternal eligibility for Medicaid associated with levels of 14 inflammatory proteins (IL6R, TNFR1, TNFR2, IL8, ICAM1, VCAM1, TSH, EPO, bFGF, IGF1, VEGF, PIGF, Ang-1, Ang-2). However, only three were significant at more than one time point during the month after preterm birth (TSH, bFGF, Ang-1), and none was associated at all five measurement time points. These studies, taken together with our results, suggest that the impact of SES on immune regulation is relatively modest and inconsistent in the newborn period but accrues through to adulthood. Further research is required to understand how and when SES becomes embedded in child development and whether early life events such as preterm birth modify that process; EpiScores could be a powerful tool for investigating the temporal dynamics of social determinants of child health.

Sensitivity analyses

Post hoc analyses showed potential small effect of maternal variables on the neonatal methylome in this cohort enriched for preterm birth sampled at term equivalent age. This is consistent with previous findings of no association between smoking or preeclampsia and DNAm within this cohort [32]. This may be due to the reduced in utero exposure to maternal factors for preterm infants, or that the samples were taken at term equivalent age, so neonatal unit exposures may outweigh maternal factors.

Strengths and limitations

Strengths of this study include the large sample of term and preterm neonates; to the best of our knowledge, this is the first examination of multiple DNA methylation-based estimators of circulating proteins in a neonatal sample. We derived EpiScores from minimally invasive sampling (buccal cells from saliva) which overcomes the ethical challenge of venepuncture for research in children. The EpiScores, serving as proxies of inflammatory proteins and sampled at term equivalent age in preterm infants, were selected for their potential to capture chronic, cumulative inflammation associated with preterm birth and neonatal intensive care exposures [23,24,25]. We adjusted for variables associated with DNAm, and additionally for inflammatory exposures to increase the clinical validity of our results.

The study has some limitations. The EpiScores used were developed in adult cohorts [26,27,28, 45] and have not been validated with neonatal protein levels. A validation study would be challenging because the phasic nature of circulating proteins and maturational variation in protein expression would require serial venepuncture, which presents ethical and practical barriers in preterm infants. Of note, we have previously established that neonatal DNAm CRP scores correlate with cumulative clinical inflammatory exposures, which is corroborative evidence that the score developed in adults is relevant in neonates [32]. The 104 EpiScores we tested explain 1–58% of variance of protein levels [26,27,28, 45]. However, even those that capture a relatively low proportion of the variance associate with incident diseases such as cardiovascular disease, type 2 diabetes, cognitive function and brain health [26, 115,116,117]. This magnitude of variance explained is also comparable to that achieved with polygenic risk scores, which have proved useful in risk stratification [118,119,120]. The EpiScores were also trained using blood samples [26,27,28, 45], whereas we have projected these scores into saliva samples. However, previous studies have successfully used similar cross-tissue techniques [32, 121, 122], and in neonates saliva provides a noninvasive and accessible sample method. Not all inflammatory-related proteins are represented, as we were limited by available EpiScores, so we may have underestimated the full complexity of the relationship between birth GA, SES, and inflammation. Longitudinal investigations are imperative for elucidating whether the DNA methylation signatures associated with gestational age identified in this study exert a causal influence on the inflammation-associated mechanisms in preterm birth. It remains crucial to discern whether these signatures represent a direct downstream consequence of GA itself or are induced by specific factors correlated with GA, yet not necessarily driven by chronic inflammation. Mendelian randomisation studies, integrating genomic and epigenomic determinants, are a promising methodological approach to disentangle the directionality of these intricate relationships. The sensitivity analyses suggested a potential weak effect of maternal exposures on associations between GA and EpiScores at term equivalent age. However, the prevalence of some of the maternal factors was low, for example 18/332 (5.4%) mothers had diabetes. Therefore, larger sample sizes enriched for the variable of interest, with longitudinal sampling from birth, would be required to investigate the relative contributions of antenatal/intrapartum versus postnatal events on the associations we observed.

The study population is comparable to other neonatal populations in high-income, majority white settings, but these results may not generalise to settings with different socioeconomic or ethnicity profiles. We studied several measures of SES but were not able to include all that could be relevant, such as household income.

Conclusion

We identified 43 EpiScores enriched for inflammatory proteins that associated with low birth GA. These 43 proteins offer novel insights into the physiological response to preterm birth and warrant further study to explore their role in the relationship between preterm birth, inflammation, and longer-term outcomes. We found only three EpiScores associated with SES in the neonatal period, none of which survived adjustment for perinatal pro-inflammatory exposures, suggesting that inflammation is unlikely to be the primary axis through which SES becomes embedded in the development of preterm infants in the neonatal period.

Availability of data and materials

DNA methylation data are available to researchers subject to the terms of the Data Access Policy: https://www.ed.ac.uk/centre-reproductive-health/tebc/about-tebc/for-researchers/data-access-collaboration.

References

  1. Ohuma EO, Moller A-B, Bradley E, Chakwera S, Hussain-Alkhateeb L, Lewin A, et al. National, regional, and global estimates of preterm birth in 2020, with trends from 2010: a systematic analysis. Lancet. 2023;402:1261–71.

    Article  PubMed  Google Scholar 

  2. Johnson S, Marlow N. Early and long-term outcome of infants born extremely preterm. Arch Dis Child. 2017;102:97–102.

    Article  PubMed  Google Scholar 

  3. Agrawal S, Rao SC, Bulsara MK, Patole SK. Prevalence of autism spectrum disorder in preterm infants: a meta-analysis. Pediatrics. 2018;142:1–14.

    Article  Google Scholar 

  4. Twilhaar ES, Wade RM, deKieviet JF, vanGoudoever JB, vanElburg RM, Oosterlaan J. Cognitive outcomes of children born extremely or very preterm since the 1990s and associated risk factors: a meta-analysis and meta-regression. Jama Pediatr. 2018;172:361.

    Article  PubMed  PubMed Central  Google Scholar 

  5. Crump C. An overview of adult health outcomes after preterm birth. Early Hum Dev. 2020;150: 105187.

    Article  PubMed  PubMed Central  Google Scholar 

  6. 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.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. Sullivan G, Galdi P, Blesa-Cábez M, Borbye-Lorenzen N, Stoye DQ, Lamb GJ, et al. Interleukin-8 dysregulation is implicated in brain dysmaturation following preterm birth. Brain Behav Immun. 2020;90:311–8.

    Article  CAS  PubMed  Google Scholar 

  8. Raisi-Estabragh Z, Cooper J, Bethell MS, McCracken C, Lewandowski AJ, Leeson P, et al. Lower birth weight is linked to poorer cardiovascular health in middle-aged population-based adults. Heart. 2023;109:535–41.

    CAS  PubMed  Google Scholar 

  9. Farah MJ. The neuroscience of socioeconomic status: correlates, causes, and consequences. Neuron. 2017;96:56–71.

    Article  CAS  PubMed  Google Scholar 

  10. Kivimäki M, Batty GD, Pentti J, Shipley MJ, Sipilä PN, Nyberg ST, et al. Association between socioeconomic status and the development of mental and physical health conditions in adulthood: a multi-cohort study. Lancet Public Heal. 2020;5:e140–9.

    Article  Google Scholar 

  11. Duncan GJ, Ziol-Guest KM, Kalil A. Early-childhood poverty and adult attainment, behavior, and health. Child Dev. 2010;81:306–25.

    Article  PubMed  Google Scholar 

  12. Pillas D, Marmot M, Naicker K, Goldblatt P, Morrison J, Pikhart H. Social inequalities in early childhood health and development: a European-wide systematic review. Pediatr Res. 2014;76:418–24.

    Article  PubMed  Google Scholar 

  13. Thomson K, Moffat M, Arisa O, Jesurasa A, Richmond C, Odeniyi A, et al. Socioeconomic inequalities and adverse pregnancy outcomes in the UK and Republic of Ireland: a systematic review and meta-analysis. BMJ Open. 2021;11: e042753.

    Article  PubMed  PubMed Central  Google Scholar 

  14. Ruiz M, Goldblatt P, Morrison J, Kukla L, Švancara J, Riitta-Järvelin M, et al. Mother’s education and the risk of preterm and small for gestational age birth: a DRIVERS meta-analysis of 12 European cohorts. J Epidemiol Commun H. 2015;69:826–33.

    Article  Google Scholar 

  15. Muscatell KA, Brosso SN, Humphreys KL. Socioeconomic status and inflammation: a meta-analysis. Mol Psychiatr. 2020;25:2189–99.

    Article  Google Scholar 

  16. Broyles ST, Staiano AE, Drazba KT, Gupta AK, Sothern M, Katzmarzyk PT. Elevated C-reactive protein in children from risky neighborhoods: evidence for a stress pathway linking neighborhoods and inflammation in children. PLoS ONE. 2012;7: e45419.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Dowd JB, Zajacova A, Aiello AE. Predictors of inflammation in U.S. children Aged 3–16 Years. Am J Prev Med. 2010;39:314–20.

    Article  PubMed  PubMed Central  Google Scholar 

  18. Schmeer KK, Yoon AJ. Home sweet home? Home physical environment and inflammation in children. Soc Sci Res. 2016;60:236–48.

    Article  PubMed  PubMed Central  Google Scholar 

  19. Schmeer KK, Yoon A. Socioeconomic status inequalities in low-grade inflammation during childhood. Arch Dis Child. 2016;101:1043.

    Article  PubMed  Google Scholar 

  20. Chiesa C, Natale F, Pascone R, Osborn JF, Pacifico L, Bonci E, et al. C reactive protein and procalcitonin: reference intervals for preterm and term newborns during the early neonatal period. Clin Chim Acta. 2011;412:1053–9.

    Article  CAS  PubMed  Google Scholar 

  21. Brown JVE, Meader N, Cleminson J, McGuire W. C-reactive protein for diagnosing late-onset infection in newborn infants. Cochrane Database Syst Rev. 2019;10:16–78.

    Google Scholar 

  22. Bower JK, Lazo M, Juraschek SP, Selvin E. Within-Person variability in high-sensitivity C-reactive protein. Arch Intern Med. 2012;172:1519–21.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. Dammann O, Leviton A. Intermittent or sustained systemic inflammation and the preterm brain. Pediatr Res. 2014;75:376–80.

    Article  PubMed  Google Scholar 

  24. Stevenson AJ, McCartney DL, Harris SE, Taylor AM, Redmond P, Starr JM, et al. Trajectories of inflammatory biomarkers over the eighth decade and their associations with immune cell profiles and epigenetic ageing. Clin Epigenet. 2018;10:159.

    Article  CAS  Google Scholar 

  25. Stevenson AJ, McCartney DL, Hillary RF, Campbell A, Morris SW, Bermingham ML, et al. Characterisation of an inflammation-related epigenetic score and its association with cognitive ability. Clin Epigenetics. 2020;12:113.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Gadd DA, Hillary RF, McCartney DL, Zaghlool SB, Stevenson AJ, Cheng Y, et al. Epigenetic scores for the circulating proteome as tools for disease prediction. Elife. 2022;11: e71802.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Ligthart S, Marzi C, Aslibekyan S, Mendelson MM, Conneely KN, Tanaka T, et al. DNA methylation signatures of chronic low-grade inflammation are associated with complex diseases. Genome Biol. 2016;17:255.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Stevenson AJ, Gadd DA, Hillary RF, McCartney DL, Campbell A, Walker RM, et al. Creating and validating a DNA methylation-based proxy for interleukin-6. J Gerontol Ser Biol Sci Med Sci. 2021;76:2284–92.

    Article  CAS  Google Scholar 

  29. Conole ELS, Stevenson AJ, Maniega SM, Harris SE, Green C, del Hernández M, CV, et al. DNA methylation and protein markers of chronic inflammation and their associations with brain and cognitive aging. Neurology. 2021;97:e2340–52.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. Green C, Shen X, Stevenson AJ, Conole ELS, Harris MA, Barbu MC, et al. Structural brain correlates of serum and epigenetic markers of inflammation in major depressive disorder. Brain Behav Immun. 2021;92:39–48.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. Barker ED, Cecil CAM, Walton E, Houtepen LC, O’Connor TG, Danese A, et al. Inflammation-related epigenetic risk and child and adolescent mental health: a prospective study from pregnancy to middle adolescence. Dev Psychopathol. 2018;30:1145–56.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Conole ELS, Vaher K, Blesa-Cábez M, Sullivan G, Stevenson AJ, Hall J, et al. Immuno-epigenetic signature derived in saliva associates with the encephalopathy of prematurity and perinatal inflammatory disorders. Brain Behav Immun. 2023;110:322–38.

    Article  CAS  PubMed  Google Scholar 

  33. Needham BL, Smith JA, Zhao W, Wang X, Mukherjee B, Kardia SLR, et al. Life course socioeconomic status and DNA methylation in genes related to stress reactivity and inflammation: the multi-ethnic study of atherosclerosis. Epigenetics. 2015;10:958–69.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Stringhini S, Polidoro S, Sacerdote C, Kelly RS, van Veldhoven K, Agnoli C, et al. Life-course socioeconomic status and DNA methylation of genes regulating inflammation. Int J Epidemiol. 2015;44:1320–30.

    Article  PubMed  Google Scholar 

  35. McDade TW, Ryan C, Jones MJ, MacIsaac JL, Morin AM, Meyer JM, et al. Social and physical environments early in development predict DNA methylation of inflammatory genes in young adulthood. Proc Nat Acad Sci. 2017;114:7611–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. Smith JA, Zhao W, Wang X, Ratliff SM, Mukherjee B, Kardia SLR, et al. Neighborhood characteristics influence DNA methylation of genes involved in stress response and inflammation: the multi-ethnic study of atherosclerosis. Epigenetics. 2017;12:662–73.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Joubert BR, Felix JF, Yousefi P, Bakulski KM, Just AC, Breton C, et al. DNA methylation in newborns and maternal smoking in pregnancy: genome-wide consortium meta-analysis. Am J Hum Genet. 2016;98:680–96.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  38. Howe CG, Cox B, Fore R, Jungius J, Kvist T, Lent S, et al. Maternal gestational diabetes mellitus and newborn DNA methylation: findings from the pregnancy and childhood epigenetics consortium. Diabetes Care. 2019;43:98–105.

    Article  PubMed  PubMed Central  Google Scholar 

  39. Tobi EW, Juvinao-Quintero DL, Ronkainen J, Ott R, Alfano R, Canouil M, et al. Maternal glycemic dysregulation during pregnancy and neonatal blood DNA methylation: meta-analyses of epigenome-wide association studies. Diabetes Care. 2022;45:614–23.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Sasaki A, Murphy KE, Briollais L, McGowan PO, Matthews SG. DNA methylation profiles in the blood of newborn term infants born to mothers with obesity. PLoS ONE. 2022;17: e0267946.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Godfrey KM, Reynolds RM, Prescott SL, Nyirenda M, Jaddoe VWV, Eriksson JG, et al. Influence of maternal obesity on the long-term health of offspring. Lancet Diabetes Endocrinol. 2017;5:53–64.

    Article  PubMed  Google Scholar 

  42. Krumpolec P, Kodada D, Hadžega D, Petrovič O, Babišová K, Dosedla E, et al. Changes in DNA methylation associated with a specific mode of delivery: a pilot study. Front Med. 2024;11:1291429.

    Article  Google Scholar 

  43. Chen Q, Ming Y, Gan Y, Huang L, Zhao Y, Wang X, et al. The impact of cesarean delivery on infant DNA methylation. BMC Pregnancy Childbirth. 2021;21:265.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. Wheater ENW, Galdi P, McCartney DL, Blesa-Cábez M, Sullivan G, Stoye DQ, et al. DNA methylation in relation to gestational age and brain dysmaturation in preterm infants. Brain Commun. 2022;4(2):fcac056.

    Article  PubMed  PubMed Central  Google Scholar 

  45. Gadd DA, Smith HM, Mullin D, Chybowska O, Hillary RF, Kimenai DM, et al. DNAm scores for serum GDF15 and NT-proBNP levels associate with a range of traits affecting the body and brain. Medrxiv. 2023;43:1715.

    Google Scholar 

  46. Boardman JP, Hall J, Thrippleton MJ, Reynolds RM, Bogaert D, Davidson DJ, et al. Impact of preterm birth on brain development and long-term outcome: protocol for a cohort study in Scotland. BMJ Open. 2020;10: e035854.

    Article  PubMed  PubMed Central  Google Scholar 

  47. Scottish National Statistics. SIMD—Scottish index of multiple deprivation: SIMD16 technical notes. Edinburgh: Scottish National Statistics; 2016 pp 1–69

  48. National Records of Scotland. Scotland’s Census [Internet]. 2020 [cited 2023 Nov 23]. Available from: https://www.scotlandscensus.gov.uk/search-the-census/

  49. Lowe R, Gemma C, Beyan H, Hawa MI, Bazeos A, Leslie RD, et al. Buccals are likely to be a more informative surrogate tissue than blood for epigenome-wide association studies. Epigenetics. 2013;8:445–54.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  50. Braun PR, Han S, Hing B, Nagahama Y, Gaul LN, Heinzman JT, et al. Genome-wide DNA methylation comparison between live human brain and peripheral tissues within individuals. Transl Psychiat. 2019;9:47.

    Article  Google Scholar 

  51. Pidsley R, Wong CCY, Volta M, Lunnon K, Mill J, Schalkwyk LC. A data-driven approach to preprocessing Illumina 450K methylation array data. BMC Genom. 2013;14:293.

    Article  CAS  Google Scholar 

  52. Zheng SC, Webster AP, Dong D, Feber A, Graham DG, Sullivan R, et al. A novel cell-type deconvolution algorithm reveals substantial contamination by immune cells in saliva, buccal and cervix. Epigenomics. 2018;10:925–40.

    Article  CAS  PubMed  Google Scholar 

  53. Katz DH, Robbins JM, Deng S, Tahir UA, Bick AG, Pampana A, et al. Proteomic profiling platforms head to head: leveraging genetics and clinical traits to compare aptamer- and antibody-based methods. Sci Adv. 2022;8:eabm5164.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. Haslam DE, Li J, Dillon ST, Gu X, Cao Y, Zeleznik OA, et al. Stability and reproducibility of proteomic profiles in epidemiological studies: comparing the Olink and SOMAscan platforms. Proteomics. 2022;22: e2100170.

    Article  PubMed  PubMed Central  Google Scholar 

  55. Raffield LM, Dang H, Pratte KA, Jacobson S, Gillenwater LA, Ampleford E, et al. Comparison of proteomic assessment methods in multiple cohort studies. Proteomics. 2020;20:1900278.

    Article  CAS  Google Scholar 

  56. Joshi A, Mayr M. In aptamers they trust: the caveats of the SOMAscan biomarker discovery platform from SomaLogic. Circulation. 2018;138:2482–5.

    Article  PubMed  PubMed Central  Google Scholar 

  57. Hillary RF, Marioni RE. MethylDetectR: a software for methylation-based health profiling. Wellcome Open Res. 2021;5:283.

    Article  PubMed  PubMed Central  Google Scholar 

  58. Preacher KJ. Extreme groups designs. Encycl Clin Psychol. 2015;2015:1–4.

    Google Scholar 

  59. Fisher JE, Guha A, Heller W, Miller GA. Extreme-groups designs in studies of dimensional phenomena: advantages, caveats, and recommendations. J Abnorm Psychol. 2020;129:14–20.

    Article  PubMed  Google Scholar 

  60. Mckinnon K, Conole ELS, Vaher K, Binkowska J, Sullivan G, Hillary R, et al. The relationship between socioeconomic status, preterm birth and systemic inflammation using DNA methylation proxies. 2023. OSF. https://doi.org/10.17605/OSF.IO/4JNFV.

  61. Gao X, Becker LC, Becker DM, Starmer JD, Province MA. Avoiding the high Bonferroni penalty in genome-wide association studies. Genet Epidemiol. 2010;34:100–5.

    Article  PubMed  PubMed Central  Google Scholar 

  62. Szklarczyk D, Kirsch R, Koutrouli M, Nastou K, Mehryary F, Hachilif R, et al. The STRING database in 2023: protein–protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic Acids Res. 2022;51:D638–46.

    Article  PubMed Central  Google Scholar 

  63. Kandasamy J, Roane C, Szalai A, Ambalavanan N. Serum eotaxin-1 is increased in extremely-low-birth-weight infants with bronchopulmonary dysplasia or death. Pediatr Res. 2015;78:498–504.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  64. Almudares F, Hagan J, Chen X, Devaraj S, Moorthy B, Lingappan K. Growth and differentiation factor 15 (GDF15) levels predict adverse respiratory outcomes in premature neonates. Pediatr Pulmonol. 2023;58:271–8.

    Article  PubMed  Google Scholar 

  65. Al-Mudares F, Reddick S, Ren J, Venkatesh A, Zhao C, Lingappan K. Role of growth differentiation factor 15 in lung disease and senescence: potential role across the lifespan. Front Med. 2020;7: 594137.

    Article  Google Scholar 

  66. Wan Y, Fu J. GDF15 as a key disease target and biomarker: linking chronic lung diseases and ageing. Mol Cell Biochem. 2023;479:453–66.

    Article  PubMed  PubMed Central  Google Scholar 

  67. Zhang Y, Jiang W, Wang L, Lingappan K. Sex-specific differences in the modulation of growth differentiation factor 15 (GDF15) by hyperoxia in vivo and in vitro: role of Hif-1α. Toxicol Appl Pharm. 2017;332:8–14.

    Article  CAS  Google Scholar 

  68. Wang G, Wen B, Deng Z, Zhang Y, Kolesnichenko OA, Ustiyan V, et al. Endothelial progenitor cells stimulate neonatal lung angiogenesis through FOXF1-mediated activation of BMP9/ACVRL1 signaling. Nat Commun. 2022;13:2080.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  69. Collaco JM, McGrath-Morrow SA, Griffiths M, Chavez-Valdez R, Parkinson C, Zhu J, et al. Perinatal inflammatory biomarkers and respiratory disease in preterm infants. J Pediatrics. 2022;246:34-39.e3.

    Article  CAS  Google Scholar 

  70. Oak P, Hilgendorff A. The BPD trio? Interaction of dysregulated PDGF, VEGF, and TGF signaling in neonatal chronic lung disease. Mol Cell Pediatrics. 2017;4:11.

    Article  Google Scholar 

  71. Course CW, Lewis PA, Kotecha SJ, Cousins M, Hart K, Watkins WJ, et al. Characterizing the urinary proteome of prematurity-associated lung disease in school-aged children. Respir Res. 2023;24:191.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  72. Bose CL, Dammann CEL, Laughon MM. Bronchopulmonary dysplasia and inflammatory biomarkers in the premature neonate. Arch Dis Child Fetal Neonatal Ed. 2008;93:F455.

    Article  CAS  PubMed  Google Scholar 

  73. Zasada M, Suski M, Bokiniec R, Szwarc-Duma M, Borszewska-Kornacka MK, Madej J, et al. Comparative two time-point proteome analysis of the plasma from preterm infants with and without bronchopulmonary dysplasia. Ital J Pediatr. 2019;45:112.

    Article  PubMed  PubMed Central  Google Scholar 

  74. Lassus P, Heikkilä P, Andersson LC, von Boguslawski K, Andersson S. Lower concentration of pulmonary hepatocyte growth factor is associated with more severe lung disease in preterm infants. J Pediatr. 2003;143:199–202.

    Article  CAS  PubMed  Google Scholar 

  75. Wagner BD, Babinec AE, Carpenter C, Gonzalez S, O’Brien G, Rollock K, et al. Proteomic profiles associated with early echocardiogram evidence of pulmonary vascular disease in preterm infants. Am J Respir Crit Care Med. 2018;197:394–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  76. Guasti L, Silvennoinen S, Bulstrode NW, Ferretti P, Sankilampi U, Dunkel L. Elevated FGF21 leads to attenuated postnatal linear growth in preterm infants through GH resistance in chondrocytes. J Clin Endocrinol Metab. 2014;99:E2198–206.

    Article  CAS  PubMed  Google Scholar 

  77. Spencer R, Maksym K, Hecher K, Maršál K, Figueras F, Ambler G, et al. Ultrasound and biochemical predictors of pregnancy outcome at diagnosis of early-onset fetal growth restriction. Medrxiv. 2023;2023:298.

    Google Scholar 

  78. Qiu Q, Bell M, Lu X, Yan X, Rodger M, Walker M, et al. Significance of IGFBP-4 in the development of fetal growth restriction. J Clin Endocrinol Metab. 2012;97:E1429–39.

    Article  CAS  PubMed  Google Scholar 

  79. Voller SB, Chock S, Ernst LM, Su E, Liu X, Farrow KN, et al. Cord blood biomarkers of vascular endothelial growth (VEGF and sFlt-1) and postnatal growth: a preterm birth cohort study. Early Hum Dev. 2014;90:195–200.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  80. Kim MJ, Romero R, Kim CJ, Tarca AL, Chhauy S, LaJeunesse C, et al. Villitis of unknown etiology is associated with a distinct pattern of chemokine up-regulation in the feto-maternal and placental compartments: implications for conjoint maternal allograft rejection and maternal anti-fetal graft-versus-host disease. J Immunol. 2009;182:3919–27.

    Article  CAS  PubMed  Google Scholar 

  81. Romero R, Chaemsaithong P, Chaiyasit N, Docheva N, Dong Z, Kim CJ, et al. CXCL10 and IL-6: markers of two different forms of intra-amniotic inflammation in preterm labor. Am J Reprod Immunol. 2017;78: e12685.

    Article  PubMed  PubMed Central  Google Scholar 

  82. Sellmer A, Henriksen TB, Palmfeldt J, Bech BH, Astono J, Bennike TB, et al. The patent ductus arteriosus in extremely preterm neonates is more than a hemodynamic challenge: new molecular insights. Biomol. 2022;12:1179.

    CAS  Google Scholar 

  83. Olsson KW, Larsson A, Jonzon A, Sindelar R. Exploration of potential biochemical markers for persistence of patent ductus arteriosus in preterm infants at 22–27 weeks’ gestation. Pediatr Res. 2019;86:333–8.

    Article  CAS  PubMed  Google Scholar 

  84. Waleh N, Seidner S, McCurnin D, Giavedoni L, Hodara V, Goelz S, et al. Anatomic closure of the premature patent ductus arteriosus: the role of CD14+/CD163+ mononuclear cells and VEGF in neointimal mound formation. Pediatr Res. 2011;70:332–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  85. Xu C, Su X, Chen Y, Xu Y, Wang Z, Mo X. Proteomics analysis of plasma protein changes in patent ductus arteriosus patients. Ital J Pediatr. 2020;46:64.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  86. Cheng Y, Zhu X, Linghu D, Xu Y, Liang J. Serum levels of cytokines in infants treated with conbercept for retinopathy of prematurity. Sci Rep-uk. 2020;10:12695.

    Article  CAS  Google Scholar 

  87. Sato T, Kusaka S, Shimojo H, Fujikado T. Simultaneous analyses of vitreous levels of 27 cytokines in eyes with retinopathy of prematurity. Ophthalmology. 2009;116:2165–9.

    Article  PubMed  Google Scholar 

  88. Rathi S, Jalali S, Patnaik S, Shahulhameed S, Musada GR, Balakrishnan D, et al. Abnormal complement activation and inflammation in the pathogenesis of retinopathy of prematurity. Front Immunol. 2017;8:1868.

    Article  PubMed  PubMed Central  Google Scholar 

  89. The Royal College of Ophthalmologists. Treating Retinopathy of Prematurity in the UK. The Royal College of Ophthalmologists; 2022

  90. Rivera JC, Holm M, Austeng D, Morken TS, Zhou Ellen T, Beaudry-Richard A, et al. Retinopathy of prematurity: inflammation, choroidal degeneration, and novel promising therapeutic strategies. J Neuroinflamm. 2017;14:165.

    Article  Google Scholar 

  91. Klerk DH, Plösch T, Verkaik-Schakel RN, Hulscher JBF, Kooi EMW, Bos AF. DNA methylation of TLR4, VEGFA, and DEFA5 is associated with necrotizing enterocolitis in preterm infants. Front Pediat. 2021;9: 630817.

    Article  Google Scholar 

  92. Olaloye OO, Liu P, Toothaker JM, McCourt BT, McCourt CC, Xiao J, et al. CD16+CD163+ monocytes traffic to sites of inflammation during necrotizing enterocolitis in premature infants. J Exp Med. 2021;218: e20200344.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  93. Pammi M, Suresh G. Enteral lactoferrin supplementation for prevention of sepsis and necrotizing enterocolitis in preterm infants. Cochrane Db Syst Rev. 2020;3:CD007137.

    Google Scholar 

  94. Satrom KM, Ennis K, Sweis BM, Matveeva TM, Chen J, Hanson L, et al. Neonatal hyperglycemia induces CXCL10/CXCR3 signaling and microglial activation and impairs long-term synaptogenesis in the hippocampus and alters behavior in rats. J Neuroinflamm. 2018;15:82.

    Article  Google Scholar 

  95. Groselj-Grenc M, Ihan A, Derganc M. Neutrophil and monocyte CD64 and CD163 expression in critically Ill neonates and children with sepsis: comparison of fluorescence intensities and calculated indexes. Mediat Inflamm. 2008;2008: 202646.

    Article  Google Scholar 

  96. Kingsmore SF, Kennedy N, Halliday HL, Velkinburgh JCV, Zhong S, Gabriel V, et al. Identification of diagnostic biomarkers for infection in premature neonates. Mol Cell Proteomics. 2008;7:1863–75.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  97. Zilow EP, Hauck W, Linderkamp O, Zilow G. Alternative pathway activation of the complement system in preterm infants with early onset infection. Pediatr Res. 1997;41:334–9.

    Article  CAS  PubMed  Google Scholar 

  98. Mohammed A, Okwor I, Shan L, Onyilagha C, Uzonna JE, Gounni AS. Semaphorin 3E regulates the response of macrophages to lipopolysaccharide-induced systemic inflammation. J Immunol. 2020;204:128–36.

    Article  CAS  PubMed  Google Scholar 

  99. Leviton A, Allred EN, Fichorova RN, O’Shea TM, Fordham LA, Kuban KKC, et al. Circulating biomarkers in extremely preterm infants associated with ultrasound indicators of brain damage. Eur J Paediatr Neuro. 2018;22:440–50.

    Article  Google Scholar 

  100. Schultz SJ, Aly H, Hasanen BM, Khashaba MT, Lear SC, Bendon RW, et al. Complement component 9 activation, consumption, and neuronal deposition in the post-hypoxic–ischemic central nervous system of human newborn infants. Neurosci Lett. 2005;378:1–6.

    Article  CAS  PubMed  Google Scholar 

  101. Morales DM, Townsend RR, Malone JP, Ewersmann CA, Macy EM, Inder TE, et al. Alterations in protein regulators of neurodevelopment in the cerebrospinal fluid of infants with posthemorrhagic hydrocephalus of prematurity. Mol Cell Proteom. 2012;11:M1110.11973.

    Article  Google Scholar 

  102. Ochoa TJ, Sizonenko SV. Lactoferrin and prematurity: a promising milk protein? Biochem Cell Biol. 2017;95:22–30.

    Article  CAS  PubMed  Google Scholar 

  103. Che X, Hornig M, Bresnahan M, Stoltenberg C, Magnus P, Surén P, et al. Maternal mid-gestational and child cord blood immune signatures are strongly associated with offspring risk of ASD. Mol Psychiatr. 2022;27:1527–41.

    Article  Google Scholar 

  104. Leviton A, Joseph RM, Fichorova RN, Allred EN, Taylor HG, O’Shea TM, et al. Executive dysfunction early postnatal biomarkers among children born extremely preterm. J Neuroimmune Pharm. 2019;14:188–99.

    Article  Google Scholar 

  105. Allred EN, Dammann O, Fichorova RN, Hooper SR, Hunter SJ, Joseph RM, et al. Systemic inflammation during the first postnatal month and the risk of attention deficit hyperactivity disorder characteristics among 10 year-old children born extremely preterm. J Neuroimmune Pharm. 2017;12:531–43.

    Article  Google Scholar 

  106. Aly H, Khashaba M, Nada A, Hasanen B, McCarter R, Schultz S, et al. The role of complement in neurodevelopmental impairment following neonatal hypoxic-ischemic encephalopathy. Am J Perinatol. 2009;26:659–65.

    Article  PubMed  Google Scholar 

  107. Limbrick DD, Morales DM, Shannon CN, Wellons JC, Kulkarni AV, Alvey JS, et al. Cerebrospinal fluid NCAM-1 concentration is associated with neurodevelopmental outcome in post-hemorrhagic hydrocephalus of prematurity. PLoS ONE. 2021;16: e0247749.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  108. Voegele AF, Jerković L, Wellenzohn B, Eller P, Kronenberg F, Liedl KR, et al. Characterization of the vitazmin E-binding properties of human plasma afamin. Biochemistry. 2002;41:14532–8.

    Article  CAS  PubMed  Google Scholar 

  109. Paetau S, Rolova T, Ning L, Gahmberg CG. Neuronal ICAM-5 inhibits microglia adhesion and phagocytosis and promotes an anti-inflammatory response in LPS stimulated microglia. Front Mol Neurosci. 2017;10:431.

    Article  PubMed  PubMed Central  Google Scholar 

  110. Skeel A, Yoshimura T, Showalter SD, Tanaka S, Appella E, Leonard EJ. Macrophage stimulating protein: purification, partial amino acid sequence, and cellular activity. J Exp Med. 1991;173:1227–34.

    Article  CAS  PubMed  Google Scholar 

  111. Kronenberg F, Kollerits B, Kiechl S, Lamina C, Kedenko L, Meisinger C, et al. Plasma concentrations of afamin are associated with the prevalence and development of metabolic syndrome. Circ Cardiovasc Genet. 2018;7:822–9.

    Article  Google Scholar 

  112. Montez JK, Bromberger JT, Harlow SD, Kravitz HM, Matthews KA. Life-course socioeconomic status and metabolic syndrome among midlife women. J Gerontol: Ser B. 2016;71:1097–107.

    Article  Google Scholar 

  113. Castagné R, Delpierre C, Kelly-Irving M, Campanella G, Guida F, Krogh V, et al. A life course approach to explore the biological embedding of socioeconomic position and social mobility through circulating inflammatory markers. Sci Rep. 2016;6:25170.

    Article  PubMed  PubMed Central  Google Scholar 

  114. Leviton A, Allred EN, Dammann O, Joseph RM, Fichorova RN, O’Shea TM, et al. Socioeconomic status and early blood concentrations of inflammation-related and neurotrophic proteins among extremely preterm newborns. PLoS ONE. 2019;14: e0214154.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  115. Chybowska AD, Gadd DA, Cheng Y, Bernabeu E, Campbell A, Walker RM, et al. Epigenetic contributions to clinical risk prediction of cardiovascular disease. Circ Genom Precis Med. 2024;17:004265.

    Article  Google Scholar 

  116. Cheng Y, Gadd DA, Gieger C, Monterrubio-Gómez K, Zhang Y, Berta I, et al. Development and validation of DNA methylation scores in two European cohorts augment 10-year risk prediction of type 2 diabetes. Nat Aging. 2023;3:450–8.

    Article  CAS  PubMed  Google Scholar 

  117. Smith HM, Moodie JE, Monterrubio-Gómez K, Gadd DA, Hillary RF, Chybowska AD, et al. Epigenetic scores of blood-based proteins as biomarkers of general cognitive function and brain health. Clin Epigenet. 2024;16:46.

    Article  CAS  Google Scholar 

  118. Cross B, Turner R, Pirmohamed M. Polygenic risk scores: an overview from bench to bedside for personalised medicine. Front Genet. 2022;13:1000667.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  119. Lewis CM, Vassos E. Polygenic risk scores: from research tools to clinical instruments. Genome Med. 2020;12:44.

    Article  PubMed  PubMed Central  Google Scholar 

  120. Yousefi PD, Suderman M, Langdon R, Whitehurst O, Smith GD, Relton CL. DNA methylation-based predictors of health: applications and statistical considerations. Nat Rev Genet. 2022;23:369–83.

    Article  CAS  PubMed  Google Scholar 

  121. Suarez A, Lahti J, Lahti-Pulkkinen M, Girchenko P, Czamara D, Arloth J, et al. A polyepigenetic glucocorticoid exposure score at birth and childhood mental and behavioral disorders. Neurobiol Stress. 2020;13: 100275.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  122. Blostein FA, Fisher J, Dou J, Schneper L, Ware EB, Notterman DA, et al. Polymethylation scores for prenatal maternal smoke exposure persist until age 15 and are detected in saliva in the Fragile Families and Child Wellbeing cohort. Epigenetics. 2022;17:2223–40.

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

The authors are grateful to the families who consented to take part in the study.

Funding

This work was supported by Theirworld (www.theirworld.org) and a UKRI MRC Programme Grant MR/X003434/1. KM receives salary from NHS Scotland. GS is funded by an MRC Clinician Scientist Fellowship MR/X019535/1. SRC is supported by a Sir Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society (221890/Z/20/Z).

Author information

Authors and Affiliations

  1. Centre for Reproductive Health, Institute for Regeneration and Repair, University of Edinburgh, 4-5 Little France Drive, Edinburgh, EH16 4UU, UK

    Katie Mckinnon, Kadi Vaher, Justyna Binkowska, Amy Corrigan & James P. Boardman

  2. Lothian Birth Cohorts, Department of Psychology, University of Edinburgh, Edinburgh, UK

    Eleanor L. S. Conole & Simon R. Cox

  3. Centre for Genomic and Experimental Medicine, Institute of Genetics and Cancer, University of Edinburgh, Edinburgh, UK

    Robert F. Hillary, Danni A. Gadd, Anna J. Stevenson & Riccardo E. Marioni

  4. Centre for Clinical Brain Sciences, University of Edinburgh, Edinburgh, UK

    Kadi Vaher, Gemma Sullivan, Heather C. Whalley & James P. Boardman

  5. Edinburgh Clinical Research Facility, University of Edinburgh, Edinburgh, UK

    Lee Murphy

  6. School of Philosophy, Psychology, and Language Sciences, University of Edinburgh, Edinburgh, UK

    Hilary Richardson

Authors
  1. Katie Mckinnon
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Eleanor L. S. Conole
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Kadi Vaher
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Robert F. Hillary
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Danni A. Gadd
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Justyna Binkowska
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Gemma Sullivan
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Anna J. Stevenson
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Amy Corrigan
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Lee Murphy
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Heather C. Whalley
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Hilary Richardson
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Riccardo E. Marioni
    View author publications

    You can also search for this author in PubMed Google Scholar

  14. Simon R. Cox
    View author publications

    You can also search for this author in PubMed Google Scholar

  15. James P. Boardman
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

KM was responsible for conceptualisation, methodology, formal analysis, writing (original draft), writing (review/editing), and visualization. ELSC and JB took part in formal analysis and writing (review/editing). KV participated in formal analysis, writing (review/editing), and visualization. RH and DG took part in methodology, formal analysis, and writing (review/editing). GS contributed to investigation and writing (review/editing). AJS was involved in methodology and writing (review/editing). AC participated in investigation, resources, data curation, and writing (review/editing). LM helped with investigation, resources, and writing (review/editing). HCW assisted with methodology and writing (review/editing). HR took part in conceptualisation, writing (original draft), writing (review/editing), and supervision. REM and SRC participated in conceptualisation, methodology, writing (original draft), writing (review/editing), and supervision. JPB was responsible for conceptualisation, methodology, writing (original draft), writing (review/editing), supervision, project administration, and funding acquisition.

Corresponding author

Correspondence to James P. Boardman.

Ethics declarations

Ethics approval and consent to participate

Ethical approval for the study was obtained from the National Research Ethics Service, South-East Scotland Research Ethics Committee (REC 11/55/0061, 13/SS/0143, 16/SS/0154), and NHS Lothian Research and Development (2016/0255). Written informed consent was obtained from parents.

Consent for publication

Not applicable.

Competing interests

LM has received speaker and consultancy fees from Illumina. REM is a scientific advisor to the Epigenetic Clock Development Foundation and to Optima Partners. All other authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Mckinnon, K., Conole, E.L.S., Vaher, K. et al. Epigenetic scores derived in saliva are associated with gestational age at birth. Clin Epigenet 16, 84 (2024). https://doi.org/10.1186/s13148-024-01701-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13148-024-01701-2

Keywords

Clinical Epigenetics

ISSN: 1868-7083

Contact us