Introduction

DNA methylation profiles change predictably with age, and epigenetic “clock” models leverage methylation at specific CpG sites to estimate biological age. The first-generation clocks by Horvath and Hannum used 71–353 CpGs to achieve high age-prediction accuracy ^[1]. Recently, tissue-specific and simplified clocks have emerged. A new clock introduced a saliva-based epigenetic clock using only 10 CpG sites that were most strongly correlated with age ^[2]. Remarkably, this minimal 10-CpG model still predicts age with high accuracy (r = 0.80, R^2^ = 0.64, mean absolute error ~5.5 years). All ten CpGs in the clock undergo age-related hypermethylation and map to genes involved in developmental, metabolic, and endocrine pathways (e.g., ELOVL2, CHGA, OTUD7A, PRLHR, ZYG11A, GPR158).

Pregnancy is a unique physiological state that might influence biological age. Evolutionary life-history theory posits a trade-off between reproduction and somatic maintenance: energy devoted to pregnancy and childbearing could accelerate ageing ^[3]. Consistent with this, women with higher parity (more pregnancies) have been observed to experience shorter life spans and earlier onset of age-related conditions ^[3].

However, it remains underexplored how an ongoing pregnancy (particularly in later gestation) influences a mother’s epigenetic age in real time. Most prior studies assessed women years after pregnancies or looked at cumulative parity. If pregnancy accelerates biological ageing, we expect pregnant women in their third trimester to exhibit an “older” epigenetic age than expected for their chronological age. Furthermore, maternal epigenetic age might impact pregnancy outcomes; for example, higher epigenetic age acceleration in mothers has been linked to shorter gestational length and risk of preterm birth ^[4]. Identifying epigenetic ageing during pregnancy could therefore provide a biomarker for both maternal health and foetal development outcomes.

In this study, we examine epigenetic age during late pregnancy using a saliva-specific 10-CpG methylation clock. We compare a cohort of pregnant women in the second half of pregnancy to age-matched non-pregnant controls. The focus is on saliva samples for practical applicability, as saliva epigenetic clocks can be deployed easily in clinical or at-home monitoring. By leveraging the concise 10-CpG clock, we aim to detect pregnancy-related epigenetic age acceleration and discuss its implications for maternal health. Additionally, we consider how such epigenetic changes might be mitigated or reversed through prenatal and postnatal interventions, given emerging evidence that biological age is modifiable ^[5].

Materials and Methods

Study Population

This study analysed two groups of female participants: a pregnancy cohort and a control cohort. The pregnancy cohort consisted of 14 women in late pregnancy (gestational age > 20 weeks, corresponding to mid-third trimester for most) with a mean chronological age of 34.9 ± 1.5 years. The control cohort comprised 108 non-pregnant women (not pregnant and with no history of prior pregnancies or children) of comparable age (mean 35.1 ± 0.8 years). Controls were frequency-matched to the pregnant group by age (all were 33–36 years old, similar to the pregnant participants’ age range) to minimise age as a confounding factor. All participants were apparently healthy with no chronic diseases reported. The study was conducted using data gathered via the Muhdo Health data repository.

DNA Methylation Profiling

DNA extraction were performed by Eurofins Genomics (Ebersberg, Denmark) using standard protocols. Genome-wide DNA methylation was quantified with the Illumina MethylationEPIC BeadChip (850k array), which assays >850,000 CpG loci across the genome. The array data for each sample were quality-controlled and processed to obtain β-values (the proportion of DNA methylated at a given CpG site, ranging from 0 to 1) for each CpG.

For the purposes of this study, we focused on the ten CpG sites that constitute the validated saliva epigenetic clock ^[2]. The β-values were normalised (using the same normalisation method as in the 10CpG clock) to ensure comparability to the reference clock data. We then computed the DNA methylation age (DNAm age) for each sample by applying the published linear combination formula of the 10-CpG clock. The resulting DNAm age represents the estimated biological age of the individual based on saliva methylation patterns.

Definition of Epigenetic Age Acceleration

We quantified epigenetic age acceleration (EAA) for each individual as the difference between the DNAm age and the individual’s chronological age. A positive EAA indicates that the person’s epigenetic age is higher than expected (i.e., “older” than their actual age), while a negative value indicates a younger epigenetic age than expected. We calculated each participant’s EAA (in years) as DNAm age minus chronological age. Group-level mean EAA was then compared between pregnant and control groups. We also examined EAA in the context of chronological age matching: since the groups were nearly the same age on average, any systematic difference in DNAm age directly reflects differential biological ageing rather than age differences.

Statistical Analysis

For the primary outcome of DNAm age, we compared pregnant vs. control group means and also performed a one-sample t-test within each group to assess whether DNAm age differed from chronological age (to evaluate overall acceleration within each group). We report group means. A p-value < 0.05 was considered statistically significant.

Results

Participant Characteristics

The pregnant and control groups were closely matched in age. The pregnant women had a mean chronological age of 34.9 years (range 32–37), virtually identical to the control group’s mean age of 35.1 years (range 33–36). The difference in chronological age between groups was not significant (p = 0.81). By design, none of the control women had prior pregnancies, while the pregnant cohort by definition was currently pregnant (for a subset, this was their first pregnancy; a few had one previous child, but all had in common an ongoing pregnancy >20 weeks at sampling). There were no significant differences in other baseline characteristics such as body mass index (BMI ~26 in both groups) or smoking status (0% smokers). Thus, the groups were well suited for isolating the effect of pregnancy on epigenetic age.

Methylation Levels at Clock CpGs in Pregnant vs. Control Groups

At the ten CpG sites comprising the saliva clock, we observed higher DNA methylation levels in the pregnant group compared to controls. The average β-value across the 10 CpGs was 0.256 in pregnant women, versus 0.248 in controls. This represents roughly a 3.2% higher methylation in late pregnancy. Despite the small sample of pregnant individuals, this difference was statistically significant (p = 0.041). In fact, 11 out of 14 pregnant women showed a higher mean clock methylation β-value than the control group’s mean. By contrast, the control individuals’ methylation levels clustered tightly around the lower mean. Each of the 10 CpGs showed a trend toward higher methylation in pregnancy, aligning with the known pattern that these loci undergo age-related gains in methylation. The direction was uniformly positive (hypermethylation in pregnancy at all clock sites). This concordant shift across all markers magnified the composite clock’s output.

Elevated Epigenetic Age in Late Pregnancy

DNA methylation age estimates derived from the 10-CpG clock were notably higher for pregnant women compared to controls. Pregnant individuals had a mean DNAm age of 39.1 ± 5.0 years, whereas controls had a mean DNAm age of 36.0 ± 5.1 years. Despite virtually identical chronological ages, the late-pregnancy group showed on average a >3-year older epigenetic age than the non-pregnant group. This difference was statistically significant (p = 0.032). While control women’s DNAm ages were close to parity with their chronological ages (within ±5 years of 35 for most individuals), the pregnant women’s DNAm ages were consistently elevated above their actual age. In the pregnant cohort, 11 of 14 women had a DNA methylation age older than their chronological age, often by several years. In contrast, the control cohort showed a roughly symmetric distribution of DNAm age around chronological age, with about half the women slightly epigenetically older and half younger than expected (as is common in healthy mid-30s populations).

We directly quantified epigenetic age acceleration (EAA) as DNAm age minus chronological age for each individual. The pregnant group’s mean EAA was +4.2 ± 5.1 years, indicating that on average their epigenetic age was about 4.2 years older than their actual age. In comparison, the control group’s mean EAA was +0.9 ± 5.0 years, i.e., very close to zero (chronological and biological ages aligned). The difference in EAA between groups was about +3.3 years (4.2 vs 0.9) and was statistically significant (p = 0.047). In practical terms, late pregnancy was associated with an epigenetic age that is approximately four years older than one would expect for a 35-year-old woman. Notably, a one-sample test confirmed that the pregnant group’s mean DNAm age was significantly greater than their actual age (p = 0.015), whereas the control group’s DNAm age did not differ significantly from actual age (p = 0.30). This suggests that biological age acceleration is a consistent feature of late pregnancy.

Figure 1. Bar graph showing differences between chronological and biological ages of both groups.

Discussion

In this study, we found that women in late pregnancy exhibit accelerated epigenetic ageing in saliva DNA, as measured by a 10-CpG methylation clock. Pregnant women (>20 weeks gestation) had on average a ~3-year higher epigenetic age than chronologically age-matched non-pregnant women, corresponding to roughly +4 years of biological age acceleration during pregnancy. To our knowledge, this is one of the first reports to directly measure epigenetic age in pregnant women using a saliva-based clock. The findings align with and extend prior research suggesting that pregnancy can impose a short-term biological ageing effect on the mother. Notably, our use of a saliva-specific clock demonstrates that this effect is detectable in an easily accessible tissue, highlighting the potential of saliva epigenetic biomarkers for monitoring maternal health.

Our results are consistent with the evolutionary concept of a reproduction–ageing trade-off ^[3]. The significant increase in epigenetic age during late pregnancy supports the hypothesis that the physiological demands of pregnancy – including endocrine, immunological, and metabolic changes – may accelerate molecular ageing processes. Our study complements current findings by capturing the effect during the pregnancy itself. The magnitude of age acceleration we observed (~3–4 years) is notable given the narrow age range of participants (~35 years) and the relatively short exposure (months of pregnancy). It suggests that pregnancy may induce transient changes to the epigenome that resemble an older biological state. These changes could be driven by factors such as heightened inflammation and oxidative stress, hormonal exposure (e.g., high levels of oestrogen, progesterone, cortisol), metabolic strain (insulin resistance, hyperlipidaemia of pregnancy), and immune system adaptation – all hallmarks of pregnancy that overlap with mechanisms of ageing.

It is important to note that epigenetic age acceleration (EAA) is an established biomarker associated with adverse outcomes. Even modest increases in epigenetic age relative to chronological age have been linked to higher risks of cardiovascular disease, cancer, cognitive decline, and all-cause mortality ^[4]. None of the pregnancies in our cohort resulted in adverse outcomes to our knowledge (no preterm births were recorded), but the sample was small.

An open question is whether the epigenetic ageing observed in pregnancy is transient or persistent. Does a woman’s epigenetic age revert to her baseline (pre-pregnancy) after giving birth and recovering postpartum, or do repeated pregnancies cause a cumulative “ageing” effect that remains? Some longitudinal evidence suggests partial recovery postpartum ^[3]. It is plausible that pregnancy triggers a temporary acceleration in ageing marks that largely normalises after childbirth (perhaps as inflammatory and metabolic stresses subside), but repeated cycles of acceleration could have lasting effects. Unfortunately, our study was cross-sectional and cannot address reversibility directly. We emphasise the need for longitudinal research tracking women from pre-conception through pregnancy and into postpartum, measuring epigenetic age at each stage.

From a translational perspective, the finding of pregnancy-associated epigenetic age acceleration raises the question of interventions. Might supportive interventions slow down this process or assist in recovery, encouragingly, accumulating evidence indicates that epigenetic age is malleable and can be influenced by lifestyle and pharmacological interventions. For instance, a recent pilot randomised trial demonstrated that an 8-week program of diet, exercise, stress reduction, and supplements led to a 3.23-year decrease in DNAm age compared to controls [5]. Notably, that study also used saliva-based methylation measurements and showed the feasibility of reducing epigenetic age through non-pharmacological means. In the context of pregnancy, this opens up exciting possibilities. Optimal prenatal care already encourages healthy diet, physical activity as tolerated, and stress management – interventions that align with those known to slow biological ageing. Our results underscore the importance of these recommendations: pregnant individuals might particularly benefit from ageing-slowing lifestyle measures to counteract the inherent biological strain of pregnancy. It would be worthwhile to test in future studies whether pregnant women who adhere to nutrition and exercise guidelines (or other interventions like mindfulness, adequate sleep, antioxidant supplements, etc.) exhibit less epigenetic age acceleration than those who do not. Likewise, the postpartum period may be an opportunity to “rebound” biological age. Supporting new mothers through diet, exercise, and possibly epigenetic rejuvenation therapies (should they become available) could help reverse any residual age acceleration from pregnancy.

Strengths and Limitations: A strength of our study is the use of an established, tissue-specific epigenetic clock with demonstrated accuracy in saliva. This clock’s validity is supported by strong correlation with chronological age in large datasets, and it focuses on CpGs that are biologically relevant (associated with genes in ageing pathways). By applying this clock to a new context (pregnancy), we extend its utility. Moreover, the control group was tightly age-matched and nulliparous, isolating the effect of current pregnancy. However, the study has limitations. The sample size, particularly of the pregnant group (n = 14), was small, which limits statistical power and the generalisability of the exact magnitude of effects. Our significant findings despite the small n suggest a relatively strong effect size, but replication in larger cohorts is needed. Another limitation is the cross-sectional design – we did not measure these women’s epigenetic ages before pregnancy, so we infer acceleration by comparison to controls rather than direct pre/post changes. It is possible that women who become pregnant at this age differ systematically from those who do not (for example, unmeasured health or lifestyle factors) which could confound epigenetic age differences. We attempted to minimise this by selecting controls of similar age and background; still, a longitudinal within-subject design would be more definitive. Additionally, factors such as parity and pregnancy complications were not deeply examined here. Our pregnant group likely included mostly first pregnancies, but if a few were multiparous, their baseline epigenetic age could have been higher from prior pregnancies. Future studies should stratify by parity and examine whether first pregnancies vs. subsequent pregnancies differ in epigenetic impact. We also did not control for other variables that can affect epigenetic age, such as body mass index, stress levels, or socioeconomic status, due to our small sample. These should be explored in larger samples as moderators of pregnancy-related EAA.

Despite these caveats, our findings provide an initial proof-of-concept that pregnancy is accompanied by a detectable uptick in epigenetic ageing in saliva.

Conclusion

Late-stage pregnancy is associated with significant biological age acceleration, as evidenced by a saliva DNA methylation age that outpaces chronological age. Using a validated saliva-specific 10-CpG epigenetic clock, we found that pregnant women around 35 years old have an epigenetic age approximately 3–4 years higher than their chronological age, in contrast to non-pregnant peers whose biological and chronological ages are nearly aligned. These results support the notion that the substantial physiological demands of pregnancy can accelerate molecular ageing processes in the mother. While pregnancy-induced epigenetic ageing may be largely transient, it could have implications for maternal-foetal health and warrants further investigation.