Interpreting ERAP2 as an Ageing‐Derived Biomarker for Premature Ovarian Insufficiency

We write regarding the article by Meng et al. [1] reporting ERAP2 as a biomarker associated with ovarian ageing and spontaneous ovulation recovery in women with premature ovarian insufficiency (POI). While we appreciate the focus on identifying markers related to ovarian ageing, we would like to raise several methodological considerations regarding the interpretation of ERAP2 as a predictive biomarker. Candidate biomarkers were identified using ovarian transcriptomic data from the GTEx v8 database by comparing younger (20–30 years) and older (50–60 years) healthy female groups. This strategy raises questions about the extent to which molecular characteristics associated with physiological ovarian ageing can be extrapolated to the prediction of pathological premature ovarian failure (POI). Unlike normative ovarian ageing, POI represents a heterogeneous condition that may involve immune-mediated, genetic, or idiopathic mechanisms, which may not be fully reflected in age-stratified transcriptomic differences observed in healthy populations [2]. In addition, it remains unclear whether changes in ERAP2 expression precede the clinical onset of POI or instead represent molecular correlates of established ovarian dysfunction. Predictive biomarkers are most clinically valuable when they identify individuals at increased risk before overt symptoms or biochemical abnormalities emerge. Without longitudinal evidence demonstrating altered ERAP2 levels prior to the development of amenorrhoea or gonadotrophin elevation, it is difficult to determine whether ERAP2 enables early identification of healthy women at risk or those with subclinical or occult POI, as opposed to serving primarily as a marker associated with manifest disease. We acknowledge that the cohort design of the current analyses may limit temporal inference, but this distinction is important for clinical translation [3]. This study is of great value in exploring more clinically significant biomarkers of ovarian ageing. We hope the authors address these limitations in future work, and we welcome further discussion. X.F. contributed to conceptualisation and wrote the original draft. Y.J. contributed to reviewing and editing. R.W. contributed to conceptualisation and reviewing. The authors have nothing to report. The authors declare no conflicts of interest. The data from this study are available on reasonable request from the corresponding author.