The Norwegian Mother, Father and Child Cohort Study (MoBa) is a population-based pregnancy cohort study. Pregnant women and their partners were recruited from all over Norway between 1999 and 2008. The women consented to participation in 41% of the pregnancies, and fathers were invited from 2001 onwards. Since women could participate with more than one pregnancy, the cohort includes approximately 114,500 children, 95,000 mothers, and 75,000 fathers. Using a unique personal identification number assigned to residents in Norway, data from questionnaires were linked to information from the Medical Birth Registry of Norway and the National Patient Registry. Blood was drawn from both parents at recruitment, at around gestational week 18. This study included a subset of 3742 mothers participating in MoBa with DNA methylation measurements available from blood samples taken at recruitment and who therefore had epigenetic age acceleration estimates available ( Figure 1 ). A total of 922 of the 3742 mothers were sampled for measurement because they had used assisted reproductive technologies (ART) to conceive, while the others constituted a random sample of the cohort in general.

Definition of study population

Establishment and initial data collection in MoBa was based on a license from the Norwegian Data Protection Agency and approval from The Regional Committees for Medical and Health Research Ethics. The MoBa cohort is currently regulated by the Norwegian Health Registry Act. The present study was approved by the Regional Committees for Medical and Health Research Ethics of South/East Norway (reference number: 277291). All participants provided a written informed consent.

We applied 7 different epigenetic age clocks in this study. This included the Hannum, Horvath’s pan-tissue, Horvath’s skin and blood, Levine, and GrimAge clocks, which estimate epigenetic age at sampling time, while the Lu DNAmTL clock estimates telomere length (in kb) at sampling time and the DunedinPACE clock reflects the pace of aging over the past 10 to 15 years at sampling time. Despite these variations, the epigenetic age estimates were treated the same way in our analyses.

The epigenetic biomarkers of aging were derived by calculating weighted averages over the DNA methylation levels at selected sites where cytosine and guanine are separated by phosphate (CpG) sites. The weights, that is, the coefficient estimates from penalized regressions, were sourced from the publications of the epigenetic age clocks. To obtain a measure of epigenetic age acceleration, we performed a linear regression of each epigenetic age estimate against chronological age at blood sampling. We then extracted the residuals from this model as the deviation between the epigenetic and chronological age. These residuals were subsequently standardized into z-scores and used as the outcomes of interest.

The measurement of peripheral blood-derived DNA methylation in the current dataset is described in detail elsewhere. Briefly, DNA methylation among the MoBa mothers was measured using the Illumina Infinium MethylationEPIC V1 Array (Illumina, San Diego, CA), and after quality control, 770,586 autosomal probes remained. Some CpGs were excluded because they did not pass the quality control procedures described previously. The number of excluded CpGs was as follows: 8 CpGs for Hannum (71–>63), 19 CpGs for Horvath’s pan-tissue (353–>334), 4 CpGs for Horvath’s skin and blood (391–>387), none for Levine (513–>513), 228 CpGs for PCGrimAge (78,464–>78,237), 20 CpGs for DNAmTL (140–>120), and 14 CpGs for the DunedinPACE (173–>159) clock. PCGrimAge was calculated using the R code available on Higgins-Chen et al’s GitHub repository: https://github.com/MorganLevineLab/PC-Clocks . For DunedinPACE, we used Belsky et al’s GitHub repository: https://github.com/danbelsky/DunedinPACE .

Information on diagnoses of endometriosis/adenomyosis was obtained by self-report at the time of recruitment into MoBa (when the blood sample was taken for DNA methylation measurements) and consultations in specialist healthcare services as recorded in the national patient registry from 2008 onwards. At the time of recruitment, women were asked if they had ever been diagnosed with endometriosis or adenomyosis. No additional information was available on the age or method of diagnosis. Self-report of endometriosis has been shown to have a high validity when compared to clinical and surgical records. The patient registry includes all inpatient and outpatient consultations except private practitioners without any agreements with the government, of which there are few. Linkage to the patient registry contained follow-up information on all MoBa participants due to the mandatory registration of all contact with specialist healthcare services in this registry, resulting in no loss to follow-up. Diagnoses in the patient registry are coded according to the International Classification of Diseases version 10, and we used registrations of the International Classification of Diseases version 10 codes N80.0 to N80.9 to indicate endometriosis/adenomyosis. During the study period, given the diagnostic traditions in Norway at the time, we assume that the majority of the diagnoses recorded in the registry were based on surgical verification for endometriosis and vaginal ultrasound or posthysterectomy diagnoses for adenomyosis. A minority may have been diagnosed with endometriosis using specialized ultrasound examinations or magnetic resonance imaging, as these modes of diagnosis have gained foothold in recent years. Due to the large overlap between the 2 conditions, we did not attempt to evaluate differences in epigenetic age acceleration between women with endometriosis and women with adenomyosis.

We obtained information on a range of background characteristics at the time of enrollment from the MoBa questionnaires and the birth registry. This included age at blood sampling, self-reported information on educational level (less than high school, high school/vocational, up to 4 years of higher education, and more than 4 years of higher education), smoking status over the last 3 months before pregnancy (never, former, and current), prepregnancy body mass index (BMI) (underweight [<18.5 kg/m ²], normal weight [18.5–24.9 kg/m ²], overweight [25–29.9 kg/m ²], and obese [≥30 kg/m ²]), parity (primiparous vs multiparous), time to pregnancy (continuous in months), and use of ART (yes vs no). In addition, from the quality-controlled DNA methylation data, we estimated cell type composition using the UniLIFE reference panel implemented in the EpiDISH R package, which is particularly suited for calculating different sets of cell type composition for any age group. Here, for MoBa women, we included Cluster of differentiation4 ⁺ T naive, basophils, Cluster of differentiation4 ⁺ T memory, B memory, B naive, regulatory T cells, Cluster of differentiation8 ⁺ T memory, Cluster of differentiation8 ⁺ T naive, eosinophils, natural killer cells, neutrophils, and monocytes as covariates.

While chronological age and education, smoking, and BMI obtained at the time of the index pregnancy might be considered confounders of the relationship of interest, parity and cell type composition are more appropriately defined as potential mediators.

Differences in background characteristics according to a diagnosis of endometriosis/adenomyosis were evaluated by chi-squared tests or t tests. We used linear regression to estimate mean differences in epigenetic age acceleration measurements between women with and without endometriosis/adenomyosis. Initially, we ran an unadjusted model. Subsequently, in a separate model, we adjusted for chronological age at blood sampling, educational level, smoking status, BMI, and processing batch. In a final model, we explored additional adjustment for cell type composition and parity, as these could act as potential mediators of the relationship. The small proportion of missing data was accounted for by including a separate missing category. Multiple testing was also considered by conducting a Bonferroni correction ( P value threshold.007; 0.05/7 epigenetic age clocks).

To evaluate the robustness of our findings, we performed several additional analyses. First, we stratified the analyses according to whether the woman had used ART or not to conceive the index pregnancy. We did this to account for potential sampling bias as a third of the women were sampled for epigenetic analyses based on having used ART. We have previously shown that there might be a relationship between use of ART and epigenetic age acceleration, and women with endometriosis/adenomyosis have a substantial increased use of ART. We also evaluated the potential influence of undiagnosed disease by comparing the mean difference in epigenetic age acceleration between women with endometriosis/adenomyosis and women without either of these conditions who had conceived the index pregnancy within 3 months of starting to try. This is because there is likely to be a higher proportion of undiagnosed endometriosis or adenomyosis among women who spent longer to conceive. To further evaluate the role of undiagnosed disease, we compared the epigenetic age acceleration among women diagnosed with endometriosis/adenomyosis prior to recruitment into MoBa and those diagnosed after recruitment to those who remained undiagnosed at the end of follow-up. An additional sensitivity analysis excluded women who were only diagnosed with adenomyosis. We also did a stratified analysis with a cutoff of chronological age at 32 years at the time of blood sampling (the median in the population). A formal test for interaction was conducted by including product terms in the multivariable models.

All analyses were conducted using Stata version 17 (Statacorp, Texas) and R version 4.5.1.

Before and while undertaking, we discussed and presented the findings to representatives of the Norwegian Endometriosis Foundation. They were able to comment on the potential interpretation and usefulness of the fundings.