Female reproductive longevity, controlled by the timing of menarche and menopause, can vary greatly depending on genetics, lifestyle, and environmental exposures. While variations in age at menarche (AAM) and age at natural menopause (ANM) have a complex multi-factorial aetiology, the biological mechanisms underlying these variations are still not fully understood. Identifying biomarkers allows for a better understanding of the pathophysiology underpinning variations in AAM and ANM and their interconnection, while the same molecules could represent potential targets to pharmacologically modify timing of these events. High-throughput metabolomics studies have led to the discovery of a number of candidate biomarkers for a variety of traits. However, simultaneous measurement of hundreds of circulating metabolites in case-control studies is cost prohibitive and subject to confounding and reverse causation. Large genome-wide association studies (GWAS) have become available for both metabolites and AAM and ANM, advancing our knowledge on the genetic determinants of these traits. Mendelian randomisation (MR) is an established method in genetic epidemiology that explores whether a modifiable exposure is causally linked to an outcome by using genetic variants for this exposure as instrumental variables. The MR design can avoid potential bias from confounding that are typical in conventional observational studies by taking advantage of the fact that inherent genetic variants are not susceptible to environmental risk factors and reverse causation. In a setting known as two-sample MR, GWAS data for an exposure and an outcome measured in independent populations can be used to test causality of risk factors on complex health outcomes.
By employing two-sample MR, we screened hundreds of previously measured circulating metabolites for causal association with AAM and ANM. Genetic variants associated with each metabolite were extracted from four large metabolomic GWAS and effects of these variants on AAM and ANM were retrieved from the largest GWAS conducted for AAM (N = 329,345), and ANM (N = 200,000). We discovered 12 blood metabolites with evidence of a causal relationship with the AAM and 114 metabolites associated with ANM. Using multivariable MR, we found that the majority of these metabolites affect the timing of AAM or ANM regardless of body mass index (BMI). These molecules cluster in specific pathways, such as that of amino acid synthesis and glycerophosphocholine synthesis. We identified two of the candidate metabolites for AAM as significantly associated with ANM in ALSPAC participants (data requested as part of a separate project [B3667], aiming to predict AAM using a combination of genetic and non-genetic risk factors). Specifically we found an association between higher phosphatidylcholine levels and a later onset of AAM, and a link between lower isoleucine levels and an earlier onset of AAM, and the magnitude of the effects were comparable with those of the MR study. In this proposal, we request data on metabolites levels and AAM of ALSPAC mothers, in order to seek validation for a portion of the 114 candidate metabolites for ANM prioritized by our MR study, which have been measured in ALSPAC. This validation will provide further support to our MR findings, suggesting a causal role of the above metabolites in ANM.