Epigenomic signatures of accelerated epigenetic aging are associated with congenital heart disease in newborns

Abstract

Background

Congenital heart disease (CHD) lesions are the most common birth defects and despite advances in care, are associated with short- and long-term co-morbidities. The exact mechanisms that may influence outcomes in survivors with CHD remain unclear and are likely multi-factorial; exploring the epigenome in these cases may provide novel insights into predictive biomarkers contributing to outcomes. The present study characterizes the impact of CHD on the newborn epigenome through assessments of epigenetic age.

Methods

This is a prospective, single-site case–control pilot study conducted at Children’s National Hospital with subjects enrolled from the Washington DC Metropolitan area. Genomic samples were collected between 2018 and 2024 and analyzed using the Illumina MethylationEPIC BeadChip array platform. PedBE was used to assess infant biological age acceleration.

Results

Using a methylation array approach, we analyzed the epigenetic age of 33 newborns with complex CHD requiring neonatal cardiac surgery and 26 healthy controls. There was a significant accelerated epigenetic age in newborns with CHD (+ 72.9 days) compared to newborns from uncomplicated pregnancies (+ 13.9 days, p < 0.001, unadjusted). Further subgroup analysis within the CHD cohort revealed that both single- (+ 56.6 days) and two-ventricle CHD (+ 64.8 days) displayed significant accelerated epigenetic age, with transposition of the great arteries (TGA) cases having the greatest accelerated age (107.2 days, p = 0.0001). Stepwise analysis of clinical measures in the CHD group revealed significant age acceleration was associated with low blood oxygen saturation.

Conclusions

This pilot study reveals accelerated epigenetic aging in newborns with critical CHD compared to healthy controls. Though the mechanisms behind these findings are not well-defined, the association between postnatal measures of oxygen saturation and epigenetic age suggests hypoxia may play a significant role and should be explored in future studies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12920-025-02189-2.

Introduction

With a reported prevalence of 6–13 per 1,000 liveborn infants, congenital heart disease (CHD) is the most common category of birth defect [1–7]. Though CHD encompasses a wide range of anatomical cardiac malformations [1–8], critical CHD constitutes 15–25% of all CHD cases and is defined as severe, life-threatening CHD that requires surgical intervention in the newborn period [9, 10]. Examples of critical CHDs include hypoplastic left heart syndrome (HLHS), transposition of the great arteries (TGA), tetralogy of Fallot (TOF), tricuspid atresia, and truncus arteriosus. Survivors with CHD are at increased risk for lifelong health adversities, particularly in neurobehavioral function [1–7]. The mechanisms of altered neurodevelopment in critical CHD are likely multifactorial and now are increasingly recognized to have origins in the prenatal period [11–16]. While the molecular underpinnings of impaired neurodevelopment associated with CHD are largely understudied, it is of great interest to understand what role physiological stressors may have on the genome and epigenome, which may contribute to short- and long-term outcomes.

External stressful stimuli, such as chronic hypoxemia, especially in early-life are closely linked to epigenetic changes, including DNA methylation, which could serve as biomarkers for long-term neurodevelopmental impairment [17–19]. It has been previously shown that hypoxia results in aberrant DNA methylation in the context of fetal growth restriction and may involve differential expression of hypoxia-inducible factors (HIFs), which are critical to brain development [19–23]. Other stressful perinatal environments have been linked not only to differential DNA methylation, but also to impaired long-term neurodevelopmental outcomes in children [13, 24–26]. DNA methylation after stressful exposures, such as premature birth and perinatal stress, has also been linked to premature epigenetic aging [24–28]. Across the lifespan, epigenetic age acceleration is associated with increased risk of mortality, cardiovascular, metabolic and neurodegenerative diseases [29, 30]. Though less is known regarding epigenetic aging in newborns, early measures of epigenetic age may provide novel insights into early development, as well as potential biomarkers of adverse early prenatal and environmental exposures [31–33]. Moreover, accelerated epigenetic aging in newborns has been associated with impaired short- and long-term developmental outcomes, including reduced brain volume and increased incidence of internalizing psychological disorders [27, 28]. Taken together, epigenetic aging may identify potential biomarkers of adverse exposures for neonates with CHD, as well as identify children at-risk for impaired neurodevelopmental outcomes. Given that very little is known regarding epigenetic age in newborns with CHD, the primary objective of our study was to determine if there are differences in epigenetic age between newborns with critical CHD and healthy term newborns born from uncomplicated pregnancies.

Materials & methods

Study design, setting, and participants

This case–control pilot study prospectively recruited pregnant individuals in the Washington DC Metropolitan area at Children’s National Hospital and was approved by the Children’s National Institutional Review Board (Protocol 00003898).

In this longitudinal observational study with planned follow-up through two years of age, data collected through the immediate newborn period was included in this report. Individuals in the case arm of the study were recruited after undergoing fetal echocardiogram (echo) before 28 weeks of gestation with a confirmed diagnosis of critical CHD. Critical CHD was categorized into four classes [34]: Class 1 refers to two ventricle defects (2 V) with no outlet obstruction, including complete atrioventricular canal defect (CAVC), double outlet right ventricle (DORV), TGA, TOF, and truncus arteriosus. Class 2 also refers to 2 V plus aortic obstruction (2 V-AO) including coarctation of the Aorta (CoA) and interrupted aortic arch ventricular septal defect (IAA-VSD) and Shone’s complex. Class 3 and 4 represent SV defects without and with aortic obstruction (SV-AO), respectively. Class 3 includes hypoplastic right heart syndrome (HRHS), pulmonary atresia (PA), and tricuspid atresia (TA), while class 4 includes HLHS and HRHS (Supplemental Table 1). For subgroup analyses based on cardiac diagnosis, we further categorized cases into three groups: SV (Class 3 and 4), 2 V (Class 1 and 2, minus TGA cases), and 2 V TGA. Of note, because TGA is a 2 V lesion that represents a severe form of critical CHD associated with systemic and cerebral hypoxia, sub-group analyses with TGA-only lesions were performed and not included in the aforementioned 2 V analysis.

Into the control arm we recruited healthy women before 28 weeks of gestation who had normal prenatal screens and otherwise uncomplicated pregnancies. Exclusion criteria for both arms included: chromosomal and single gene anomalies diagnosed by amniocentesis that are causative of CHD (e.g., CHARGE syndrome), clinical phenotyping that included structural non-cardiac anomalies or dysmorphic features by ultrasound, pregnancies later than 28 weeks, multiple gestations, or evidence of congenital infections. Written consent to participate in the study was obtained from all participants prior to sample collection.

Maternal-infant data

Maternal-infant demographic data, including gestational age (GA) at birth, GA at buccal swab collection, infant length, weight, head circumference, sex, mode of delivery, Apgar 1 min, and 5 min, and maternal race, ethnicity, and age at delivery, were systematically collected and recorded in a centralized, secure clinical data repository, REDCap [35, 36]. Racial and ethnic background were self-reported by parents. Clinical genetic testing is routinely offered to patients with CHD at our center. Additional clinical data from the Cardiac Intensive Care Unit (CICU), including postnatal oxygen saturation, blood gas, use of ventilation, SNAPPE score, and infant demise, were extracted from electronic health records. Lowest oxygen saturation, serum pH, pO2, base deficit and highest pCO2 and lactate were recorded from birth prior to buccal swab collection. A complete breakdown of demographic and clinical information can be found in Table 1 and Supplemental Table 1.

Table 1.

Demographic and clinical information associated with CHD and control (CTL) newborns from uncomplicated pregnancies

		Control (CTL)		Case, all (CHD)		p
Maternal Data		total (n=26)		total (n=33)
Maternal age at delivery, in years		36.5	5.0	31	6.3	<0.01
Gravida (range)		2	(1,4)	2	(1,4)	0.78
Maternal Education*						0.05
	High-School	1	1%	5	9%
	Partial College	0	0%	5	9%
	College Graduate	7	13%	10	18%
	Graduate Degree	16	29%	12	21%
Delivery mode
	Vaginal	12	46.2	18	54.5	0.18
	Cesarean, scheduled	9	34.6	13	39.4
	Cesarean, emergency	5	19.2	1	3.0
Infant Data
Self-reported race, infant
	Asian	0	0.0	1	3.0	0.65
	Black	3	11.5	5	15.2
	Hispanic	1	3.8	5	15.2
	White	15	57.7	15	45.5
	Multiple	4	15.4	4	12.1
	Other/Unknown	3	11.5	4	12.1
Self-reported ethnicity, infant
	Hispanic	1	3.8	7	21.2	0.11
	Non-Hispanic	25	96.2	31	93.9
	Preferred not to answer	0	0.0	2	6.1
Infant sex, female		9	34.6	12	36.4	0.43
Infant GA at birth, in weeks		39.16	1.49	38.63	0.81	0.08
Infant GA at collection, in weeks		43.37	2.24	40.13	2.53	<0.001
Infant weight at birth, in grams		3301.0	445	3104.0	396	0.08
Infant head circumference at birth, in cm**		34.5	1.3	33.3	1.7	0.01
Infant length at birth, in cm*		50.4	2.5	48.6	2.4	0.01
Infant Apgar 1		7.9	1.1	7.7	1.4	0.57
Infant Apgar 5		8.8	0.7	8.4	0.8	0.05
Ventilatory support at birth, yes				23	69.7
CHD type
	Single ventricle (SV)			12	36.4
	Two ventricle, without TGA (2V)			13	39.4
	Two ventricle, TGA only (2V TGA)			9	27.3
Infant blood gas***
	lowest pH (pH), mmHg			7.3	0.1
	lowest oxygen (pO2), mmHg			34.9	8.4
	highest CO2 (pCO2), mmHg			49.7	11.4
	low oxygen saturation (low O2 saturation), %			60.5	16.7
	high lactate, mmol/L			4.3	3.4
SNAPPE score				12.4	9.9
Infant demise before 18 month follow-up, yes				6	18.2

Continuous Measures: Mean (SD); Categorical Measures: N (%)

P-values: t-test for Continuous Measures; Chi-Square/Fisher Exact Test for Categorical Measures

*Data available for 24/26 controls, 32/33

**Data available for 21/26 controls, 31/33 cases

***CICU clinical data was recorded from birth, prior to buccal swab-DNA collection and cardiac leison repair

Neonatal buccal DNA collection and extraction

Between 2018 and 2024, a total of 34 buccal swabs were collected in the neonatal period (approximately 2–4 weeks postnatal age) prior to cardiac surgery in newborns with CHD and 26 were collected from control (CTL) full-term newborns during an outpatient study visit. For all subjects, ORAcollect pediatric buccal swabs (OG-175, DNA Genotek) were collected just prior to feeds and stored at 4˚C for batch processing. DNA was extracted from swabs with the PrepIT-L2P kit (DNA Genotek) and quantified using the Qubit Broad Range dsDNA assay kit and Qubit 4 fluorometer (Thermofisher Scientific). DNA was stored at −20˚C until ready for downstream applications.

Genome-wide DNA methylation array

The Infinium MethylationEPIC BeadChip array platform (both version 1 and version 2) was used to assess global differences in DNA methylation throughout the genome at over 850,000 CpG sites, an input of 300 ng of DNA was bisulfite-converted using the DNA Methylation-Lightning kit (Zymo Research). After whole-genome amplification and enzymatic fragmentation, samples were hybridized to BeadChip arrays using the Infinium MethylationEPIC BeadChip kit according to the manufacturer’s protocol (Illumina). Intensity values at the over 850,000 methylation sites on the BeadChips were measured across the genome at single-nucleotide resolution using iScan or Nextseq 550 platform with BeadChip adapter (Illumina). Of note, a portion of the control arm for this study was previously published and the raw data has been made available in the Gene Expression Omnibus (GEO) under accession number GSE229463 [37].

Differential methylation analysis

Quality filtration, functional normalization, and differential methylation analysis were performed on all arrays in the R programming language using R Studio (version 4.4.1) [38, 39]. Functional normalization of the data was executed using the minfi package with ssnoob background and dye correction [40]. Additional filtration using minfi excluded all probes that had a signal not significantly above background control probes (p-value > 0.05), all probes representing single nucleotide variants present in the general population with a minor allelic frequency > 5% (dropLociWithSnps), and cross-reactive probes (maxprobes) [41, 42]. Normalized intensity values were filtered using wateRmelon, which removes probes if bead count is less than 3 in ≥ 5% of the samples [43]. Cell-type estimation was determined using meffil on resultant beta values [44].

DNA methylation aging via PedBE (methylclock)

Filtered and normalized beta values were input into the methylclock package in R/R Studio downloaded from Bioconductor [25, 45]. The “checkClocks” function was used to determine if there were any missing CpGs of the 96-age informative sites PedBE leverages for analyses and all analyses met and exceeded the minimum of 80% coverage of these CpG sites. The “DNAmAge” command was used to generate predicted epigenetic age in years and subsequently converted into days.

Statistical analysis

For epigenetic aging analyses, chronological days of life were subtracted from the estimated PedBE DNA methylation age to obtain epigenetic age acceleration (EAA). Significant EAA differences between cases and controls were determined through generalized linear regression modeling with adjustments for covariates. Clinical associations with epigenetic age acceleration were assessed both through generalized linear regression modeling and stepwise regression analysis.

Results

Description of critical CHD cohort

Among the 60 newborns that were recruited in this study, 34 CHD cases and 26 controls were analyzed for epigenetic age differences. The CHDs represented in the present cohort (n = 34) include SV, 2 V (without TGA), and TGA (2 V TGA) (Supplemental Table 1). The details of each CHD lesion are included in Supplemental Table 1. Initial cell-type composition analysis was performed and identified one sample in the CHD cohort after informatic processing and was removed from subsequent analysis due to high granulocyte (57.7%) contamination and low buccal cell composition (35.0%) (Supplemental Fig. 1, Supplemental Table 2). The final cohort-level analysis includes 33 CHD cases and 26 controls.

Subgrouping of critical CHD cases revealed 13 with SV disease, 11 with 2 V, and 9 with TGA. While the single Shone’s complex case was included in the larger CHD cohort analysis, it was subsequently removed from the subgroup analyses, given the complex phenotype, as it obstructs blood flow at multiple sites across the left side of the heart and therefore stood apart from the other CHD lesions in this cohort [46].

All CHD cases received postnatal genetics consultation post-delivery, with no additional concerns for syndromic CHD based on clinical evaluation. Though clinical genetic testing is offered universally to all patients with critical CHD, 47% opted to undergo standard clinical genetic testing, none of which reported positive for pathogenic results. Method of clinical genetic testing varied by participant but included one or more of the following options: amniocentesis with fluorescence in situ hybridization (FISH), chromosomal microarray (CMA), or karyotype, noninvasive prenatal testing (NIPT) with microarray or sequencing panel, postnatal microarray, and/or disease-specific customized postnatal sequencing panel (i.e., Noonan Spectrum Disorders panel) (Supplemental Table 1). Two cases returned microarrays with variants of unknown significance, both heterozygous duplications, one at 10q21.3 and the other at 1p36.31p36.23; however, neither was deemed to be causative of CHD pathology upon clinical genetics review (Supplemental Table 1). While we do not have reason to believe these genetic findings would influence epigenetic aging, due to the ever-evolving body of research in this field, we did want to ensure that these clinical genetics findings were noted.

The clinical and demographic metrics between the overall CHD and CTL cohorts reflect the 59 subjects included in the subsequent analysis. Maternal age at delivery was significantly greater in controls, compared to cases (36.57 vs. 31.04 years, respectively, p = 0.001). Though there was no significant differences in GA at birth or infant birth weight, both infant birth length and head circumference were significantly smaller in CHD compared to controls (Table 1). In addition, there was a significant difference in the GA at buccal swab collection (Table 1).

Accelerated aging is associated with critical CHD in newborns and varies by diagnosis

Using PedBE we analyzed the CHD and CTL cohorts to determine if either group was associated with accelerated epigenetic aging [25]. Overall, critical CHD cases had significantly accelerated aging (p < 0.027) compared to controls (Fig. 1, Table 2), accounting for maternal age at delivery, delivery mode, infant GA at birth and birth length, as well as Apgar scores at 1 and 5 min. Among the additional metrics analyzed, only Apgar score at 5 min were significantly correlated with EAA.

Fig. 1.

Accelerated epigenetic age between CHD and Controls. Legend. PedBE reveals significant accelerated epigenetic aging in newborns with CHD. Epigenetic age acceleration was determined using chronological days of life (DOL) between critical CHD cases (n = 33, red), and CTLs (n = 26, green). Generalized linear regression modeling was used to compare epigenetic age difference between cohorts and p-value of comparative analyses can be found in Table 2

Table 2.

Association of Accelerated Aging and CHD

	Estimate	Standard Error	t value	Pr(>|t|)
Cohort (CHD vs. CTL)	70.27	3.03	23.2	0.027
Maternal Age at Delivery	-2.55	0.26	-9.88	0.064
Delivery Mode (c-section, scheduled vs. vaginal)	-1.8	2.94	-0.61	0.651
Delivery Mode (c-section, emergency vs. vaginal)	-44.71	3.71	-12.05	0.053
Infant GA at Birth	6.66	1.94	3.44	0.180
Infant Length at Birth	1.08	0.36	2.99	0.205
Apgar Score 1	-9.87	1.93	-5.12	0.123
Apgar Score 5	43.65	1.57	27.81	0.023

Generalized Linear Regression Model (GLM)

Relative to controls, in keeping with the cohort level analysis, SV cases had significant average age acceleration of 67.7 days (p = 0.001), 2 V cases had a significant average age acceleration of 64.8 days (p < 0.001), and 2 V TGA cases had a significant average age acceleration of 107.2 days (p = 0.0001) (Fig. 2, Table 3). Inter-individual variability in the clinical presentation for the two TGA outliers relative to the other TGA cases did not reveal any significant differences between groups (Supplemental Table 4). A re-analysis excluding the two TGA outliers confirmed a significant age acceleration of 45.7 days (p = 0.01). Taken together, epigenetic age is increased for all CHD subtypes in our study, with the greatest acceleration noted in the overall 2 V-TGA subgroup (Table 3, Supplemental Table 3).

Fig. 2.

Accelerated aging among CHD diagnostic subgroups. Legend. Critical CHD subgroup epigenetic age acceleration difference analysis between 2 V (blue), 2 V TGA (gray), and SV (red) subgroups in reference to controls (CTL, green). Mean and standard deviation of PedBE and chronological age, age difference, and p-value of comparative analyses for each group compared to CTLs can be found in Table 3

Table 3.

Accelerated epigenetic aging in newborns with CHD and by specific diagnostic subgroup

	N	PedBE age days, mean	sd	Chronological DOL, mean	sd	EAA days, mean	sd	p
CTL	26	43.3	32.2	29.3	11.0	13.9	29.8	-
CHD, all	33	83.5	76.6	10.6	17.0	72.9	76.7	0.0005
SV	14	67.6	48.6	10.9	13.5	56.6	48.4	0.0013
2V	10	83.3	49.1	18.5	42.9	64.8	42.9	0.0003
2V TGA	9	108.5	125.5	1.2	0.4	107.2	125.6	0.0010

P-values: t-test between all CHD cases and subgroups with controls

Accelerated epigenetic age observed in CHD may be influenced by postnatal hypoxemia

In addition to exploring differences in epigenetic age within the CHD cohort, we further evaluated the association between epigenetic age and potentially relevant clinical factors. Linear regression analysis with all variables in the model returned no significant associations (Table 4). However, in stepwise regression analysis of the listed variables, low blood oxygen saturation levels were significantly correlated with accelerated epigenetic age (p = 0.022, Table 5, Fig. 3).

Table 4.

Regression Analysis Association of Accelerated Aging (CHD)

	Estimate	Standard Error	t value	Pr(>|t|)
Ventilation, yes	52.12	106.26	0.49	0.63
Delivery Mode (c-section, scheduled vs. vaginal)	68.89	51.62	1.33	0.21
Delivery Mode (c-section, emergency vs. vaginal)	4.55	116.38	0.04	0.97
Apgar Score 1	-17.39	32.14	-0.54	0.60
Apgar Score 5	42.22	46.19	0.91	0.38
pH	412	401	1.03	0.33
pO2	3.12	6.82	0.46	0.66
pCO2	-4.11	3.69	-1.11	0.29
O2 saturation	-4.46	3.29	-1.35	0.21
SNAPPE score	1.41	3.08	0.46	0.66
TGA vs 2V	45.18	71.59	0.63	0.54
SV vs. 2V	26.25	60.94	0.43	0.68

Linear Regression Model based on AIC

Table 5.

Stepwise Analysis Association of Accelerated Aging (CHD)

	Estimate	Standard Error	t value	Pr(>|t|)
Intercept	-20.49	16.75	-1.22	0.24
Oxygen saturation	-2.840	1.150	-2.470	0.022

Stepwise Linear Regression Model based on AIC

Fig. 3.

Accelerated epigenetic aging in CHD cases is associated with low blood oxygen saturation. Legend. In CHD, epigenetic age is significantly associated with low blood oxygen saturation levels via stepwise regression analysis (Table 5)

Discussion

CHD represents a variety of structural changes in the heart that may disrupt perfusion and oxygenation throughout the developing body and brain [47]. Though neurodevelopmental outcomes are multifactorial, influenced in part by severity and type of CHD as well as co-morbid conditions and pathogenic gene mutations [48], intrauterine and post-natal environments, including medical, surgical and psychosocial interventions, are also known to contribute to short- and long-term outcomes. These studies highlight the need to better characterize the collective exposures and risk factors that may influence outcomes for children with CHD. As such, leveraging epigenomic approaches to explore additional molecular mechanisms that may contribute to outcomes is of considerable interest.

In this hypothesis-generating study, we report the findings of accelerated epigenetic aging in newborns is associated with critical CHD diagnoses. While the genomic contributions that result in CHD are not fully elucidated, recent studies are increasingly identifying genetic causes of CHD [49, 50]. Concurrent with these studies is the recognized overlap in molecular pathways of disrupted cardiac development as well as neurodevelopment [51–54]. Despite these advances, there remain additional environmental risk factors that also are known to influence outcomes in children with CHD, including prenatal and neonatal exposures, such as hypoxia, hypoperfusion, medications and surgical interventions [48]. These studies suggest that in addition to identifying underlying genetic differences that result in CHD, it is critical to assess subsequent genomic changes that may result from CHD, such as those relating to epigenetic aging. Hypoxia, including chronic hypoxia that can occur with unrepaired CHD, is one such pathway known to result in epigenetic dysregulation, including accelerated aging, and contribute to adverse neurodevelopment [55–58].

While this field is still emerging and requires further investigation, epigenetic age acceleration may prove to have clinical utility in predicting adverse outcomes associated with certain diseases, specifically in newborns [24, 25, 59]. Based on previous studies that observed accelerated epigenetic aging after premature birth and its relationship with neurodevelopment, we sought to explore the role of epigenetic age in our cohort of critical CHD cases. Though we observed significant epigenetic age acceleration in the overall critical CHD case cohort, we also observed an average accelerated epigenetic age of 13.9 days in the control arm. This may reflect how the PedBE clock, while able to predict age acceleration for a broad pediatric population, may not fully capture the rapid, exponential changes that occur early in development in the neonatal period [60]. Despite this limitation, we report a significant increase in age acceleration compared to controls in each of the CHD subgroup analyses, with TGA having the greatest age acceleration, followed by SV, and 2V disease [27, 61]. Notably, we also observed that low blood oxygen saturation was associated with greater epigenetic age acceleration among the CHD group. This finding is particularly salient, given that many critical CHD cases are cyanotic in presentation and hypoxia, both acute and chronic, is known to cause aberrant DNA methylation [22, 56, 62]. This finding requires further study but may imply that molecular or cellular oxidative stress, due to low oxygen environments, causes an increase in epigenetic age. Interestingly, a recent report also noted that children and adults with Fontan circulation, the final stage of surgical repair for those with SV CHD, demonstrated epigenetic age acceleration, with greater acceleration in adults relative to children, which may suggest progressive changes throughout the lifespan [63].

Unexpectedly, we also observed that higher Apgar 5 scores are significantly associated with greater accelerated epigenetic age acceleration in the overall cohort of cases and controls. While we report this finding, it is also important to highlight that the Apgar scores at 5 min reflect a relatively normal perinatal transition. As such, the clinical relevance of this finding is unclear. Future studies should further explore epigenetic age associations with perinatal transition that build upon Apgar scores, as well as other potentially relevant newborn health assessments.

Together, these findings suggest an important role for the “exposome,” or collection of exposures for newborns with CHD [64]. Overall, while the exact molecular underpinnings and implications associated with EAA in CHD are not well understood, these significant age-related differences may serve as important biomarkers of stress and disease. This, in turn, could offer improved granularity in understanding the genomic and epigenomic contributions to the clinically complex landscape of CHD.

Limitations

While we present novel insights into epigenetic age acceleration in infants with critical CHD, there are several limitations that deserve mention. First, while our overall sample size may be considered small for a genomic study, this cohort is well-balanced between CHD cases and controls. Nonetheless, the smaller sample size and lack of normative data for EAA in this population underscore the need to validate and expand these findings in a larger population. Moreover, it is important to note that numerous maternal, prenatal, clinical and environmental factors may influence the neonatal epigenome, particularly in CHD, and additional studies will further explore and identify these and other potential confounders. Additionally of note, variability of diagnosis in the CHD cohort required grouping of diagnostic categories, which as a result, may not allow this study to be generalized to all children with CHD. Similarly, the scope of clinical genetic testing performed within this cohort varied considerably (microarrays, custom panels, whole exome sequencing) and thus did not allow for the additional evaluation of specific CHD-associated single-nucleotide polymorphisms in the current analysis. Future endeavors may benefit from accounting for additional genetic contributions, in conjunction with the epigenome and transcriptome, to understand the complex molecular landscape across the spectrum of CHD diagnoses. Large scale longitudinal studies can both leverage the complex relationships between pathogenic genetic changes with the neonatal epigenome and transcriptome and further shed light into the onset and progression of genomic changes across the lifespan. Such work would enhance generalizability to potentially at-risk populations and inform potential mechanisms that may influence outcomes in patients with CHD.

It also is important to note that DNA methylation is known to be highly influenced by age. The advent of user-generated methylation clock packages, like PedBE, was specifically validated for use in subjects 0–20 years old; it is therefore possible that the variation or alterations to the epigenome in the first weeks of life are not accurately captured using this package [24, 25]. This may be reflective of the slight average age acceleration of 13.9 days noted in the control arm of the study. As aforementioned, we included two samples among the TGA cohort that had accelerated ages above 300 days and could be considered statistical outliers. Given the pilot nature of this study, we elected to include these in our pilot analysis. As a result, these findings should be interpreted with caution, as they warrant further exploration to discern the potential clinical relevance of these samples. It may be prudent in future studies to focus on TGA-specific clinical variables and outcomes to better understand the unique and significant epigenetic age acceleration noted in certain TGA cases.

Despite these limitations, this report of accelerated epigenetic aging among newborns with CHD and may assist in elucidating the etiological underpinnings of long-term outcomes in this population. While further exploration using additional clinical metrics and large sample size is warranted, this study underscores the utility of methylation clocks to assess epigenetic age across the lifespan and identify periods of risk and resilience in survivors of CHD.

Conclusion

Overall, the present data support the hypothesis that newborns with critical CHD have global differences in DNA methylation when compared to healthy newborns. The significant increase in epigenetic age in newborns with CHD provides insight into potential epigenetic biomarkers of aging in this population, as well as the potential timing and the role of relevant clinical variables, like low blood oxygen saturation, that may contribute to important clinical outcomes, including impaired neurodevelopment. This provides an important framework for future experiments and may allow for further insight into early diagnostic and therapeutic approaches for children with CHD.

Abbreviations

CHD

Congenital Heart Disease

PedBE clock

Pediatric Buccal Epigenetic clock

Echo

Echocardiogram

SV

Single ventricle

2V

Two ventricles

TGA

Transposition of the Great Arteries

HLH/HRH

Hypoplastic Left Heart/Hypoplastic Right Heart

TOF

Tetralogy of Fallot

Authors’ contributions

KK: experiments and investigation, formal analysis, informatics methodology, data visualization, writing of original draft and editing of manuscript; JN: statistical methodology and analysis, data visualization, writing of original draft and review and editing of manuscript; SB: informatics methodology and analysis, review and editing of manuscript; MD: interpretation of cardiology diagnosis and clinical variables, review and editing of manuscript; CL: funding acquisition, resources, supervision, validation, editing and review of manuscript; NA: conceptualization of project, data curation, clinical recruitment supervision, validation of data analysis and visualization, writing of original draft and editing of manuscript. All authors read and approved the final manuscript.

Funding

This study was funded by the National Heart, Lung, and Blood Institute (NHLBI R01 HL116585-01) and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (5T32HD098066). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Data availability

The datasets generated and analyzed in the current study will be made publicly available in the NCBI-Gene Expression Omnibus (GEO) upon publication. Data for previously published samples can be found under GEO accession number: GSE229463. New data associated with the present study are currently private but can be viewed at GEO accession number: GSE293799 using the token: kfwruocelnuzpcb until the time of publication.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board at Children’s National Hospital in Washington DC, USA under Protocol 00003898 and 00011077. All methods were carried out in accordance with institutional guidelines and regulations. Written informed consent to participate in the study was obtained from all participants and from the parents/legal guardians for minors.

Consent for publication

N/A.

Competing interests

The authors declare no competing interests.