Accelerated epigenetic clock aging in maternal peripheral blood and preterm birth

Abstract

BACKGROUND:

Epigenetic clocks use CpG DNA methylation (DNAm) to estimate biologic age; acceleration is associated with cancer, heart disease, and shorter lifespan. Few studies evaluate DNAm age and pregnancy outcomes. AgeAccelGrim is a novel epigenetic clock that combines seven DNAm components.

OBJECTIVES:

We sought to determine if maternal biologic aging (via AgeAccelGrim) is associated with early PTB.

STUDY DESIGN:

A prospective cohort of patients with singleton gestations and at high risk for spontaneous preterm birth (PTB) delivering at a tertiary University hospital were included in this study. Genome-wide CpG methylation was measured using the Illumina^® EPIC BeadChip from maternal blood samples obtained <28 weeks’ gestation. AgeAccelGrim and its seven DNAm components were estimated by the Horvath DNAm age online tool; positive values are associated with accelerated biologic aging whereas negative values are associated with slower biologic aging relative to each subject’s age. The primary outcome was PTB <34 weeks (any indication); secondary outcomes were PTB <37 and <28 weeks. AgeAccelGrim was analyzed as a continuous variable and in quartiles; exploratory analyses evaluated each of the seven DNAm components included in the composite AgeAccelGrim. Data were analyzed by chi-square, t-test, rank-sum, logistic regression (controlling a priori for maternal age, cell counts, low socioeconomic status, and gestational age at time of sample collection), and Kaplan-Meier survival analyses; the log-rank test was used to test the equality of the survival functions.

RESULTS:

163 patients met inclusion criteria; 48% delivered <37, 39% <34, and 21% <28 weeks’ gestation. The median AgeAccelGrim was −0.35 (IQR −2.24, 1.31) years for those delivering at term; those delivering preterm had higher AgeAccelGrim values that were inversely proportional to delivery gestational age [PTB <37 = +0.40 (IQR −1.21, +2.28) years; <34 = +0.51 (IQR −1.05, +2.67 years, and <28 = +1.05 (IQR −0.72, +2.72) years],. Estimated DNAm of the seven epigenetic clock components values were increased among those with PTB <34 weeks’ gestation, although differences were only significant for DNAm of plasminogen activation inhibitor 1. In regression models, AgeAcccelGrim was associated with an elevated risk of PTB with increasing magnitude for increasing severity of PTB. For each one-year increase in AgeAccelGrim value (ie. each one-year increase in biologic age compared to chronologic age), the adjusted odds of PTB were 11% (aOR 1.11, 95% CI 1.00–1.24), 13% (aOR 1.13, 95% CI 1.01–1.26) and 18% (aOR 1.18, 95% CI 1.04–1.35) higher for PTB <37 weeks’, <34 weeks’ and <28 weeks’ respectively. Similarly, individuals with accelerated biologic aging (≥75^th percentile AgeAccelGrim) had over double the odds of PTB <34 weeks’ (aOR 2.36, 95% CI 1.10–5.08) and over triple the odds of PTB <28 weeks’ (aOR 3.89, 95% CI 1.61–9.38). The adjusted odds ratio for PTB <37 weeks’ gestation was 1.73, but spanned the null (aOR 1.73, 95% CI 0.81–3.69). In Kaplan-Meier survival analyses, those in the highest AgeAccelGrim quartile delivered earliest (log-rank p< 0.001).

CONCLUSIONS:

Accelerated biologic aging is associated with PTB among high-risk patients. Future research confirming these findings and elucidating factors that slow biologic aging may improve birth outcomes.

INTRODUCTION:

Preterm birth (PTB) remains the leading cause of neonatal morbidity and mortality among non-anomalous babies delivered in the United States (US).^1–4 In 2021, the US PTB rate rose to 10.49%, the highest reported since 2007.⁵ Unfortunately, the underlying pathophysiology is poorly understood, as is the ability to predict which individuals will deliver preterm.^6,7 Some established PTB risk factors (e.g., experiences of racism and discrimination, history of childhood abuse) are associated with chronic stress and subsequent increased allostatic load (the systematic changes in the endocrine and nervous system in response to heightened and/or frequent stress exposure).,^8–11 High allostatic load may in turn result in accelerated biologic aging. Both chronic stress and allostatic load are associated with PTB, preeclampsia, and other adverse obstetric outcomes; however, whether this pathway functions through accelerated biologic aging remains unclear.^12–17

Biologic aging can be measured utilizing “epigenetic clocks” or “epigenetic age.”¹⁸ Cytosine-guanine nucleotide methylation [CpG DNA methylation (DNAm)] is among the most common epigenetic modifications and has been recognized as a reliable indicator of biologic age.¹⁹ These epigenetic modifications are dynamic and reflect cumulative short- and long-term exposures.²⁰ Though epigenetic age is correlated with chronologic age,¹⁹ accelerations in epigenetic age relative to chronologic age are associated with a variety of adverse health outcomes including cancer,^21,22 heart disease²³, and a shortened lifespan.^24–26 Advanced chronologic age is an established risk factor for several adverse obstetric outcomes including PTB,²⁷ and some studies have appreciated a ‘U’ shaped relationship between maternal age and PTB risk (with those at extremes of age carrying the highest risk).^27,28 However, few studies have investigated whether accelerated biologic age is associated with PTB.

Horvath, et al. developed DNAm GrimAge, an epigenetic clock based on seven DNAm components that include DNAm based surrogates for smoking pack-years (DNAm PACKYRS), adrenomedullin levels (DNAm ADM), beta-2 microglobulin (DNAm B2M), cystatin C (DNAm Cystatin C), growth differentiation factor 15 (DNAm GDF-15), leptin (DNAm Leptin), plasminogen activation inhibitor 1 (DNAm PAI-1), and tissue inhibitor metalloproteinase 1 (DNAm TIMP-1).²⁹ The resulting composite biomarker AgeAccelGrim (DNAm GrimAge adjusted for chronologic age) is a single value, expressed in units of years. AgeAccelGrim has been previously validated and shown to better predict adverse health outcomes including co-morbidity count, time-to-death, and time-to-coronary heart disease relative to other epigenetic clocks.²⁹ Further, lower AgeAccelGrim values are noted with positive lifestyle factors such as healthy diet and education.²⁹ AgeAccelGrim can be obtained from DNAm data and chronologic age using a publicly available, online web tool.²⁹

We sought to determine if biologic aging, assessed in maternal peripheral blood in early pregnancy, could effectively differentiate the subset of high-risk patients ultimately destined to deliver preterm.

MATERIALS AND METHODS:

This was a prospective cohort study of pregnant individuals with singleton, non-anomalous gestations, at high risk for spontaneous PTB (“University of North Carolina Preterm Birth [UNC PTB] Biobank”). Patients were identified and recruited at the University of North Carolina at Chapel Hill (Chapel Hill, NC) inpatient units and outpatient clinics from 2015 to 2018. Patients were considered to be at high risk for spontaneous PTB and eligible for inclusion in the PTB Biobank if they were high risk due to (1) historical factors: e.g., had a history of a delivery between 16 weeks’ and 0 days and 36 weeks’ and 6 days gestation either due to cervical insufficiency (defined as asymptomatic cervical dilation before 24 weeks’ gestation), preterm prelabor rupture of membranes, placental abruption, and/or following the spontaneous onset of premature labor, and/or (2) due to current pregnancy complications: at least one of the following diagnoses in their current pregnancy: preterm prelabor rupture of membranes, asymptomatic cervical shortening <25mm detected by transvaginal ultrasound prior to 24 weeks’ gestation, cervical cerclage in situ, or placental abruption. Patients with a PTB in a pregnancy complicated by fetal aneuploidy or lethal fetal anomalies or a previous PTB due to medical or fetal indications were excluded if they did not have another qualifying delivery or high-risk antenatal course. Those with multiple gestation and individuals unable to provide written, informed consent in English or Spanish were also excluded. Patients with a PTB in a pregnancy complicated by fetal aneuploidy or lethal fetal anomalies or a previous PTB due to medical or fetal indications were excluded if they did not have another qualifying delivery or current high-risk antenatal course. Pregnancy gestational age was calculated using the last menstrual period (if available) and ultrasound using standard ACOG criteria.³⁰ For this current analysis, we included patients from the original PTB Biobank cohort who were carrying a singleton pregnancy and had a blood sample collected before 28 weeks’ gestation. All pregnancy management decisions were made at the discretion of the primary obstetrical provider.

At enrollment, each participant provided a blood sample by standard venipuncture; samples were collected into a PAXgene^® blood DNA tube and stored at −80°C until future analysis. Maternal demographics, medical and obstetric history, antenatal course, and pregnancy outcomes were abstracted from electronic medical records by trained research assistants. All study data were collected and managed using the Research Electronic Data Capture tools, a secure, web-based application designed to support data capture for research studies, hosted at the University of North Carolina at Chapel Hill. This study was approved by the Institutional Review Board of the University of North Carolina at Chapel Hill, and all participants provided written, informed consent.

At the time of analysis, genomic DNA was isolated from frozen samples using the AllPrep DNA/RNA/miRNA Universal Kit (Qiagen, Valencia, CA), in accordance with the manufacturer’s instructions, and then bisulfite-converted using the EZ DNA methylation kit (Zymo Research, Irvine, CA). Bisulfite-converted DNA was hybridized onto the Illumina^® MethylationEPIC BeadChip.^31,32 Samples were labeled with a barcoded study identification number only and were allocated randomly onto arrays by laboratory personnel who were blinded to all clinical parameters, to minimize potential for batch or row variation. Using previously described methods, DNA methylation data were preprocessed using the minfi package within ‘R’.³³ We performed functional normalization with a preliminary step of normal-exponential out-of-band correction method for background subtraction and dye normalization, followed by the typical functional normalization method with the top two principal components of the control matrix.^34–36 Quality control was performed on individual probes by computing a detection p-value and probes with non-significant detection (p>0.01) for 5% or more of the samples were excluded. Lastly, we used the ComBat function from the sva R package to adjust for batch effects from sample plate.^37–39 Data were visualized using density distributions at all processing steps. AgeAccelGrim and its 7 DNAm components (ADM, B2M, cystatin C, GDF, Leptin, PAI-1, and TIMP-1) were estimated by the Horvath epigenetic age tool online.²⁹ Data input into the online tool included DNAm beta values (the numerical representation of DNA methylation levels at each specific genomic loci), female sex, and patient chronologic age at the time of venipuncture.

DNAm GrimAge from participant DNA was regressed on chronologic age of each participant at the time of the blood draw using a linear regression model and AgeAccelGrim was defined as the corresponding raw residual [the difference between the observed value of DNAm GrimAge minus the expected value (chronologic age)], using methodology established by Horvath, et al.²⁹ Online tool outputs included single values for each DNAm component and a composite DNAm GrimAge or AgeAccelGrim result for each individual; for all outputs, negative values indicated biologic aging less than chronologic age, values of zero indicated biologic aging equal to chronologic age, and positive values indicated biologic aging greater than chronologic age.¹⁹

The primary outcome was PTB <34 weeks’ gestation, defined as a dichotomous variable. Early PTB was selected as the primary outcome because of the high-risk nature of the cohort (13% with either a history of a previous classical uterine incision or myomectomy or permanent cerclage suture in situ) and to reflect an outcome associated with the highest risks of morbidity. Secondary outcomes were PTB <37, <28 weeks’ gestation, both defined as dichotomous variables, and delivery gestational age considered as a continuous variable. AgeAccelGrim and each of the seven DNAm individual clock output values were considered continuously and compared between those with and without PTB at each gestational age cutoff using chi-square, Fisher’s exact test and Student’s t-test. In addition, we generated a binary variable whereby we classified each participant as having ‘slower’ biologic aging (AgeAccelGrim value <75^th percentile for this cohort) or ‘accelerated’ biologic aging (AgeAccelGrim value ≥75^th percentile) and evaluated the odds of PTB at each gestational age cutoff utilizing logistic regression models. All regression models evaluating PTB outcomes included several factors a priori known to be associated with PTB or DNAm, including chronologic age, low socioeconomic status (SES), gestational age at blood sampling, and estimated peripheral cell counts (using estimates of natural killer cells, as calculated by the online Horvath tool).^40–42 Finally, a Kaplan-Meier survival model was utilized to evaluate the relationship between AgeAccelGrim value and gestational age at delivery, comparing those with slower biologic aging to those with accelerated biologic aging. The equality of the survivor functions was evaluated using the log-rank test. All analyses were conducted using Stata/MP (version 17.0, College Station, TX).

RESULTS

Of the 271 individuals enrolled in the UNC PTB Biobank, 163 met inclusion criteria and were included in this analysis (Figure 1). Delivery occurred at a median 37 (IQR 28, 38) weeks’ gestation. Rates of PTB were high; 78 (47.9%) delivered <37 weeks, 63 (38.7%) <34 weeks, and 34 (20.9%) <28 weeks’ gestation. Most PTB <37 weeks’ occurred secondary to spontaneous indications; 9 of 78 (11.5%) were medically indicated (4 for hypertensive disorders of pregnancy, 3 for non-reassuring antenatal testing, 1 for intrauterine growth restriction, 1 placenta previa). Participants who delivered <34 weeks’ gestation were less likely to self-identify as White race, less likely to have hypertensive disorders of pregnancy (chronic hypertension, gestational hypertension, or preeclampsia), and had their blood sampled later (median 20.2 weeks vs. 17.3 weeks, p<0.001) compared to those who delivered ≥ 34 weeks’ gestation. Other demographic and baseline characteristics were similar between groups (Table 1). Antenatal course and initial pregnancy outcomes are shown in Table 2. Most participants had at least one intervention for PTB prevention (cerclage and/or progestogen supplementation); though those who delivered ≥ 34 weeks’ were more likely to receive progestogen supplementation, a similar proportion of individuals delivering <34 and ≥ 34 weeks’ gestation had a cerclage placed in pregnancy. Further, the majority (84.6%) had at least one transvaginal cervical length assessment. As expected, those who delivered <34 weeks’ gestation had a shorter mid-trimester transvaginal cervical length and were more likely to have a cervical length <25mm compared to those who delivered later.

Figure 1.

Study enrollment.

Table 1.

Demographic and baseline cohort characteristics. Data are n(%) unless indicated.

Characteristic	Delivery <34 weeks N=63	Delivery ≥ 34 weeks N=100	p-value
Maternal age, mean years ± SD	30.9 ± 6.3	31.9 ± 6.5	0.372
Pre-pregnancy body mass index, mean kg/m2 ± SD	30.8 ± 8.2	30.1 ± 7.9	0.599
White race	29 (29.0)	28 (44.4)	0.044
Black race	24 (38.1)	38 (38.0)	0.990
Hispanic ethnicity	7 (11.1)	23 (23.0)	0.056
Low socioeconomic status *	16 (25.4)	38 (38.0)	0.096
Tobacco use disorder	11 (17.5)	10 (10.0)	0.166
Hypertensive disorders of pregnancy (chronic hypertension or preeclampsia)	12 (19.1)	37 (37.0)	0.015
Pregestational diabetes mellitus	8 (12.7)	11 (11.0)	0.742
Current or previous diagnosis of depression	15 (23.8)	27 (27.0)	0.650
Current or previous diagnosis of anxiety	16 (25.4)	17 (17.0)	0.194
Median number of prior preterm births (IQR)	1 (0, 1)	1 (0, 2)	0.473
Earliest prior delivery, mean weeks ± SD **	26.4 ± 7.5	26.3 ± 7.4	0.941
Gestational age at time of maternal blood sample, mean weeks ± SD	20.2 ± 5.9	17.3 ± 6.0	0.003
Male fetus	35 (56.5)	53 (53.0)	0.668

^*defined as public or no medical insurance ± less than high school education ± annual household income <$25,000.

^**among 114 multiparous patients

Table 2.

Obstetric course and initial pregnancy outcomes. Data are n(%) unless indicated.

Characteristic	Delivery <34 weeks N=63	Delivery ≥ 34 weeks N=100	p-value
Received progestogen supplementation during pregnancy (vaginal progesterone or 17-alpha hydroxyprogesterone caproate)	37 (58.7)	83 (83.0)	0.001
Cervical cerclage	28 (44.4)	40 (40.0)	0.575
Shortest mid-trimester transvaginal cervical length, median mm (IQR) *	14.6 (4.0, 35.0)	33.7 (19.0, 38.0)	<0.001
Short cervical length, <25mm *	30/46 (65.2)	32/93 (34.4)	0.001
Diagnosed with hypertensive disorders of pregnancy	7 (11.1)	24 (24.0)	0.041
Gestational diabetes mellitus	4 (6.4)	15 (15.0)	0.132
Clinical chorioamnionitis diagnosed prior to labor	19 (30.1)	1 (1.0)	<0.001
Placental pathology with chorioamnionitis or funisitis, n(%)	35/53 (66.0)	3/27 (11.1)	<0.001
Birthweight, mean grams ± SD	1173 ± 673	3095 ± 560	<0.001
Infant small for gestational age (birthweight <10% for gestational age and sex)	0 (0.0)	6 (6.0)	0.083
Male fetus	35 (56.5)	53 (53.0)	0.668

^*among 139 individuals with ≥ 1 transvaginal cervical length 16 to 24 weeks’ gestation

The overall median AgeAccelGrim for the entire cohort was −0.044 years (range −8.06 to +10.44, IQR −1.78, 1.49). The median AgeAccelGrim was −0.35 (IQR −2.24, 1.31) years for those delivering at term; those delivering preterm had higher AgeAccelGrim values that were inversely proportional to delivery gestational age [PTB <37 = +0.40 (IQR −1.21, +2.28) years; <34 = +0.51 (IQR −1.05, +2.67 years, and <28 = +1.05 (IQR −0.72, +2.72) years], Figure 2. Estimated DNAm of each individual epigenetic clock component were compared between those with and without PTB <34 weeks’ gestation and are shown in Table 3. Estimated DNAm was higher for all seven individual clock components among those delivering <34 weeks’ gestation compared to those delivering later, but only the estimated DNAm of PAI-1, a protein released in response to inflammation, reached statistical significance for these comparisons.

Figure 2.

Median (IQR) AgeAccelGrim (years) for all subjects and by delivery gestational age.

Table 3.

Age-adjusted epigenetic clock components of GrimAge. Median values (IQR) of each component in early pregnant blood samples are compared between individuals with PTB <34 weeks and those with delivery ≥ 34 weeks. Positive values are associated with accelerated biologic age.

Epigenetic clock component/brief description	Delivery <34 weeks N=63	Delivery ≥ 34 weeks N=100	p-value
DNAmADM Estimated DNAm of adrenomedullin – a vasodilator that increases with HTN, heart failure	1.40 (−7.51, 10.6)	0.60 (−9.24, 7.03)	0.148
DNAmB2M Estimated DNAm of Beta-2 microglobulin, a biomarker associated with CVD, inflammation	12976 (−37903, 60496)	3074 (−46129, 46289)	0.377
DNAmCystatinC Estimated DNAm of Cystatin C, a biomarker of kidney function, CVD	968 (−8691, 13317)	−1128 (−11529, 9430)	0.170
DNAmGDF15 Estimated DNAm of growth differentiation factor 15, in TGF-beta sub-family; implicated in aging, mitochondrial dysfunction	15.9 (−41, 53)	−7.2 (−51, 48)	0.315
DNAmLeptin Estimated DNAm of leptin, a hormone with roles in appetite regulation	90 (−640, 648)	−110 (−1018, 813)	0.535
DNAmPAI1 Estimated DNAm of plasminogen activation inhibitor 1, a protein released in response to inflammation	538 (−950, 1432)	−110 (−1486, 1069)	0.043
DNAmTIMP1 Estimated DNAm of tissue inhibitor of metalloproteinases; TIMP1 naturally inhibits MMPs, and TIMP1 gene transcription is inducible in response to cytokines	97 (−325, 512)	−28 (−339, 279)	0.223

Those with accelerated biologic aging (≥75^th percentile AgeAccelGrim) delivered at a median of 31 weeks’ gestation (IQR 26, 38), significantly earlier than those with slower biologic aging, who delivered at a median of 37 weeks’ (IQR 31, 38). Individuals with accelerated biologic aging were also more likely to deliver <37 (58.5% vs. 44.3%), <34 (53.7% vs. 33.6%), and <28 (36.6% vs. 15.6%) weeks’ gestation (all p<0.001). Similarly, as shown in Figure 1, those who delivered <34 weeks’ gestation were more likely to have AgeAccelGrim values in this top quartile (34.9% vs. 16.0%, p=0.023).

In regression models controlling a priori for maternal age, cell counts, low SES, and gestational age at sample collection, AgeAcccelGrim was associated with an elevated risk of PTB with increasing magnitude for increasing severity of PTB. For each one-year increase in AgeAccelGrim value (ie. Each one-year increase in biologic age compared to chronologic age), the adjusted odds of PTB were 11% (aOR 1.11, 95% CI 1.00–1.24), 13% (aOR 1.13, 95% CI 1.01–1.26) and 18% (aOR 1.18, 95% CI 1.04–1.35) higher for PTB <37 weeks’, <34 weeks’ and <28 weeks’ respectively, Figure 3a. Similarly, individuals with accelerated biologic aging (≥75^th percentile AgeAccelGrim) had over double the odds of PTB <34 weeks’ (aOR 2.36, 95% CI 1.10–5.08) and over triple the odds of PTB <28 weeks’ (aOR 3.89, 95% CI 1.61–9.38). The adjusted odds ratio for PTB <37 weeks’ gestation was 1.73, but spanned the null (aOR 1.73, 95% CI 0.81–3.69), Figure 3b. In Kaplan-Meier survival analyses, individuals with accelerated biologic aging were more likely to deliver earlier across the entire spectrum of gestational ages (log-rank p<0.001), Figure 4.

Figure 3a.

Adjusted odds ratio (95% CI) for preterm birth associated with each 1-year increase in AgeAccelGrim. Regression models controlled a priori for maternal age, cell counts, low SES, and gestational age at sample collection.

Figure 3b.

Adjusted odds ratio (95% CI) for preterm birth associated with accelerated epigenetic aging (top quartile of AgeAccelGrim, ≥1.49 years). Regression models controlled a priori for maternal age, cell counts, low SES, and gestational age at sample collection.

Figure 4.

Kaplan-Meier Survival Curve adjusted for maternal age, cell counts, low SES, and gestational age at sample. Patients with accelerated epigenetic aging (top quartile, AgeAccelGrim ≥1.49 years) are compared to those with slower epigenetic aging (quartiles 1, 2, and 3, AgeAccelGrim <1.49 years). Log-rank p<0.001.

COMMENT

b. Principal findings

In this cohort of individuals at high-risk for PTB, we found that accelerated epigenetic age, as evaluated by AgeAccelGrim, is inversely associated with delivery gestational age. Individuals with the greatest accelerations in epigenetic age in relation to chronologic age had the highest odds of delivering preterm at each gestational age cutoff. In addition, one individual clock component, estimated DNAm of PAI-1, was also associated with PTB and delivery gestational age. Notably, all seven individual epigenetic clock components were higher for those with PTB <34 weeks’ gestation, and these findings were appreciated when AgeAccelGrim was considered continuously and when grouping individuals as having accelerated biologic aging vs. slower biologic aging in relation to others in the cohort.

b. Results in the Context of What is Known

It has been hypothesized that the pathophysiology underlying early PTB may differ compared to late PTB.^43–47 Here, we found the greatest associations between increased epigenetic age assessed in maternal blood and PTB at the earliest gestational ages. Small improvements in rates of PTB in the US between 2007 and 2014 primarily reflected a reduction in late PTB (34–36 weeks), with rates of early PTB <34 weeks remaining constant or rising.⁵ Though only 2.81% of births in 2021 delivered <34 weeks and most premature deliveries occur in the late preterm period, early preterm neonates have the highest risk of mortality and short- and long-term morbidity.⁵ Thus, additional insight into the pathophysiology of prematurity, particularly at these very early gestational ages, is urgently needed.

Most prior studies evaluating the relationship between biomarkers of cellular aging and adverse pregnancy outcomes have measured telomere length, and noted that reduced telomere length and lower maternal peripheral blood telomerase activity are associated with shorter gestational length.^48,49 For example, Page et al⁴⁹ evaluated a subset of 100 self-identified Mexican-origin pregnant women and found that shortened telomere length was significantly associated with PTB <37 weeks’ gestation. In another study, Marrs et al⁴⁸ found that genetic variation in human telomerase reverse transcriptase (a subunit of telomerase) was significantly associated with both preterm labor and preterm prelabor rupture of membranes. While telomere length and DNAm age are both associated with chronologic age, they do not always correlate with each other which suggests that these markers measure different biologic components of aging that are likely both relevant to the pathophysiology of PTB.⁵⁰

Few other studies have examined the relationship between DNAm of maternal blood as a measure of biologic age and PTB. In one prior small study that included 77 individuals (3 preterm births), AgeAccelGrim and DNAm PAI-1 were negatively associated with gestational length.⁵¹ PAI-1 is a protein released in response to inflammatory cytokines and thus may be regarded as a marker for ongoing inflammatory processes.⁵² In a larger study by Lancaster, et al., investigators included 229 participants from 2 cohorts; 89 self-identified as non-Hispanic Black, and 12 (5.2%) delivered preterm. They found that early pregnancy epigenetic age was most strongly associated with delivery gestational age, and reported that the strongest association was in the subset of individuals who self-identified as non-Hispanic Black. ⁵³ Our work supports and expands on these prior findings. In our high-risk cohort enriched for PTB, rates of prematurity were considerably higher, enabling evaluation of prematurity at various gestational age cutoffs and gestational age as a continuous variable, and evaluation of the individual epigenetic clock components in relation to PTB. Previous work has shown that psychosocial stress is related to increased serum levels of inflammatory cytokines during pregnancy.⁵⁴ Racial discrimination and childhood trauma are objective psychosocial stressors that may be contributing to this biological sequelae.⁵⁵

c. Clinical Implications.

Though at the current time, these data are insufficient to support specific clinical or lifestyle recommendations such as vitamin supplementation or behavioral modification (e.g., smoking cessation) specifically for the purposes of epigenetic modification as a way to effect pregnancy outcome changes, they provide a potential mechanism by which a multitude of ‘healthier’ behaviors may be associated with reduced risks of preterm birth and improved pregnancy outcomes. Future mechanistic research, such as that described below, has the potential to target specific epigenetic pathways that have direct clinical implications.

d. Research Implications

Studies outside of pregnancy have evaluated the effect of various interventions on lowering DNAm age and effecting ‘epigenetic age reprogramming’ with promising results.²⁰ In animal models, caloric restriction, cellular reprogramming, and heterochronic parabiosis (the sharing of circulations between a young and older animal) are consistently associated with reversal of epigenetic aging.²⁰ Consistent results are demonstrated in human studies. For example, in one study of 1,036 individuals, Vitamin D supplementation was associated with lower DNAm age acceleration, though only among those individuals with baseline Vitamin D deficiency.⁵⁶ Another study used an 8-week diet and lifestyle intervention to reverse DNAm age.⁵⁷ Ongoing research in this area includes the investigation of novel therapeutic ‘geroprotective’ drugs and antioxidant supplementation aimed at targeting epigenetic pathways, gene pathways known to be implicated in aging, and further refinement of health interventions.⁵⁸ The aforementioned study of epigenetic aging and PTB reported by Lancaster and colleagues included some individuals with longitudinal samples and reported that biologic age was relatively stable over the course of pregnancy.⁵³ However, CpG methylation is dynamic and at some sites varies dramatically by gestational age.^40,41 Thus, future research should also investigate whether, in a cohort of individuals with a history of adverse pregnancy outcomes such as PTB, significant lifestyle modifications during pregnancy or during the interpregnancy interval (e.g., diet, exercise, smoking cessation, other) has the potential to positively impact biologic age and impact pregnancy outcomes.

e. Strengths and Limitations

Our study had several strengths. These data were collected prospectively using a standardized protocol and approach providing a robust, objective assessment of epigenetic age in mid-pregnancy via a simple, non-invasive venipuncture. Rates of PTB in this high-risk cohort were high at each gestational age cutoff, enabling separate evaluation of births at the earliest gestational ages. Biologic age was assessed using a validated, publicly available online tool, producing a single composite biologic marker that is easy to understand and explain to pregnant individuals in the context of chronologic age. Previous work has noted that DNAm age estimates are stable across CpG methylation platforms and different normalization methods, increasing the potential for widespread clinical deployment.⁵⁹

It is important to interpret this study within the context of its limitations. Despite the relatively large number of PTBs, our sample size precluded evaluation of individual PTB subtypes or phenotypes (such as cervical insufficiency or preterm prelabor rupture of membranes). As there are limited data evaluating epigenetic age indices in pregnancy, there are no established normative values, and as such, the cutoff for the 75^th percentile for epigenetic aging as used in this cohort is population specific and may not be generalizable to lower-risk obstetric populations. Though it is biologically plausible that accelerated epigenetic aging is associated with an increased risk of prematurity due to allostatic load, this mechanism is only a hypothesis as the evaluation of factors associated with accelerated biologic aging was beyond the scope of the current investigation. These exploratory findings should be validated in an independent cohort and expanded to include investigations of specific modifiable factors that may be associated with biologic age.

f. Conclusions

These novel findings report an association between accelerated biologic aging and PTB, is a possible mechanism underlying the complex pathophysiology associated with PTB, and highlight the potential use of DNAm as a clinical biomarker for risk of prematurity. Elucidating factors associated with slowing of or reduction in biologic age prior to conception or in early pregnancy may improve birth outcomes. Validation of these findings and subsequent evaluation of DNAm age in intervention-based RCTs pre- and post-treatment may provide additional information regarding PTB pathophysiology and the mechanism(s) of action for possible interventions and provide novel solutions to lower the rate of PTB.

CONDENSATION:

Among high-risk individuals, accelerated epigenetic clock aging in mid-pregnancy peripheral blood is associated with preterm birth <37, <34, and <28 weeks’ gestation.

AJOG AT A GLANCE.

A. Why was this study conducted?

• Biologic aging can be measured utilizing “epigenetic clocks;” accelerated epigenetic aging is associated with multiple adverse health outcomes in non-pregnant individuals

• Advanced chronologic age is an established risk factor for several adverse obstetric outcomes including PTB, but it is unknown if accelerated epigenetic age is associated with PTB

B. What are the key findings?

• Accelerated epigenetic age, evaluated by the online calculator AgeAccelGrim, is inversely associated with delivery gestational age

• Individuals with the greatest accelerations in epigenetic age relative to chronologic age had the highest odds of delivering preterm <37, <34, and <28 weeks’ gestation

C. What does this study add to what is already known?

• Accelerated epigenetic aging is a possible mechanism underlying the complex pathophysiology associated with PTB, and highlights the potential use of DNA methylation as a clinical biomarker for risk of prematurity