Young women with poor ovarian response exhibit epigenetic age acceleration based on evaluation of white blood cells using a DNA methylation-derived age prediction model

Study question: Is poor ovarian response associated with a change in predicted age based on a DNA methylation-derived age prediction model (the Horvath algorithm) in white blood cells (WBCs) or cumulus cells (CCs)?

Summary answer: In young women, poor ovarian response is associated with epigenetic age acceleration within WBC samples but is not associated with age-related changes in CC.

What is known already: The majority of human tissues follow predictable patterns of methylation which can be assessed throughout a person's lifetime. DNA methylation patterns may serve as informative biomarkers of aging within various tissues. Horvath's 'epigenetic clock', which is a DNA methylation-derived age prediction model, accurately predicts a subject's true chronologic age when applied to WBC but not to CC.

Study design, size, duration: A prospective cohort study was carried out involving 175 women undergoing ovarian stimulation between February 2017 and December 2018. Women were grouped according to a poor (≤5 oocytes retrieved) or good (>5 oocytes) response to ovarian stimulation. Those with polycystic ovary syndrome (PCOS) (n = 35) were placed in the good responder group.

Participants/materials, setting, methods: DNA methylation patterns from WBC and CC were assessed for infertile patients undergoing ovarian stimulation at a university-affiliated private practice. DNA was isolated from peripheral blood samples and CC. Bisulfite conversion was then performed and a DNA methylation array was utilized to measure DNA methylation levels throughout the genome. Likelihood ratio tests were utilized to assess the relationship between predicted age, chronologic age and ovarian response.

Main results and the role of chance: The Horvath-predicted age for WBC samples was consistent with patients' chronologic age. However, predicted age from analysis of CC was younger than chronologic age. In subgroup analysis of women less than 38 years of age, poor ovarian response was associated with an accelerated predicted age in WBC (P = 0.017). Poor ovarian response did not affect the Horvath-predicted age based on CC samples (P = 0.502). No alternative methylation-based calculation was identified to be predictive of age for CC.

Limitations, reasons for caution: To date, analyses of CC have failed to identify epigenetic changes that are predictive of the aging process within the ovary. Despite the poor predictive nature of both the Horvath model and the novel methylation-based age prediction model described here, it is possible that our efforts failed to identify appropriate sites which would result in a successful age-prediction model derived from the CC epigenome. Additionally, lower DNA input for CC samples compared to WBC samples was a methodological limitation. We acknowledge that a universally accepted definition of poor ovarian response is lacking. Furthermore, women with PCOS were included and therefore the group of good responders in the current study may not represent a population with entirely normal methylation profiles.

Wider implications of the findings: The process of ovarian and CC aging continues to be poorly understood. Women who demonstrate poor ovarian response to stimulation represent a common clinical challenge, so clarifying the exact biological changes that occur within the ovary over time is a worthwhile endeavor. The data from CC support a view that hormonally responsive tissues may possess distinct epigenetic aging patterns when compared with other tissue types. Future studies may be able to determine whether alternative DNA methylation sites can accurately predict chronologic age or ovarian response to stimulation from CC samples. Going forward, associations between epigenetic age acceleration and reproductive and general health consequences must also be clearly defined.

Study funding/competing interest(s): No external funding was obtained for the study and there are no conflicts of interest.

Trial registration number: N/A.