Abstract and Introduction
Abstract
Objectives: Observational studies suggest birth weight and childhood obesity are closely associated with age at menarche. However, the relationships between them are currently inconsistent and it remains elusive whether such associations are causal. Therefore, the aim of the study was to investigate whether there existed causal relationships between birth weight, childhood obesity and age at menarche.
Design, Patients and Measurements: A two-sample Mendelian randomization (MR) study. The standard inverse variance weighted MR analyses were adopted to evaluate the causal effects of birth weight (n = 143,677), childhood body mass index (BMI) (n = 39,620) on age at menarche (n = 182,416) with summary statistics from large-scale genome-wide association studies (GWASs). Meanwhile, we validated our MR results with some sensitivity analyses including maximum likelihood, weighted-median and MR pleiotropy residual sum and outlier methods.
Results: The present study showed that each one standard deviation (1-SD) lower birth weight was predicted to result in a 0.1479 years earlier of age at menarche (β = .1479, 95% confidence interval [CI] = 0.0422–0.2535; p = 0.0061). We also found that genetically predicted 1-SD increase in childhood BMI was causally associated with early age at menarche (β = −.3966, 95% CI = −0.5294 to −0.2639; p = 4.73E−09).
Conclusions: Our MR study suggests the causal effect of lower birth weight and higher childhood BMI on the increased risk of earlier menarche. It may be the opportune time to carry out weight control intervention in prenatal and early childhood development periods to prevent early menarche onset, thus decreasing the future adverse consequences.
Introduction
Age at menarche (AAM) is an indicator of the timing of puberty in girls, and a significant decline of the average AAM has been observed in many countries.[1–4] Early menarche represents a growing public health problem, considering it is an established risk factor for breast cancer[5] and related to adverse health outcomes in later life such as diabetes,[6] depression,[7] nonalcoholic fatty liver disease,[8] cardiovascular disease[9] and all-cause mortality.[10] AAM is a multifactorial condition determined by genetic, nutritional, socioeconomic and environmental factors, individually or combined.[11,12] Although substantial research has been directed towards the field of AAM, the mechanisms underlying the decline in AAM have not been well elucidated. Both modifiable and less changeable factors urgently need to be detected to better target public health and epidemiological interventions. Heritability studies estimated that above half the phenotypic variation of AAM was due to genetic influences,[13,14] and hundreds of variants associated with AAM have been successfully identified.[15,16] However, the evidence regarding the role of nongenetic factors for AAM was limited and early. One review of epidemiological studies indicated that exposures during the foetal and early pubertal periods had a great influence on AAM.[17]
Growing evidence from observational studies reports birth weight (BW) and childhood obesity are linked with the timing of menarche.[14,18–21] Although some studies demonstrated that lower BW and higher childhood body mass index (CBMI) are associated with earlier AAM,[19,20,22] inconsistent findings still emerged.[14,18,23,24] The causal relationships between BW, childhood obesity and AAM remain uncertain because of the inherent limitations in observational research of unmeasured confounding and reverse causality.[25] Mendelian randomization (MR) analysis can contribute to addressing these issues, which is an instrumental-variable (IV) inference approach for assessing whether the exposure has a causal effect on the outcome.[25,26] This method is analogous to a randomized trial that the individual genotypes are assigned randomly at meiosis and not affected by environmental factors.[27] The MR analysis requires the genetic variants used as IVs must satisfy three important assumptions, which were illustrated in detail in Figure 1. Therefore, we aimed to achieve a definitive conclusion concerning the causality between BW, CBMI and AAM using two-sample MR analyses.
Figure 1.
MR analysis model and three assumptions of MR analysis. MR, Mendelian randomization.
Previous MR studies of such associations utilized inappropriate genetic variants as IVs,[28] and also had small sample sizes as using individual-level data.[29] Now, with the increasing availability of genome-wide association study (GWAS) summary statistics and the establishment of efficient genetic statistical methods, it provides us with an opportunity to assess the causal relationship between phenotypes using two-sample MR analyses. Moreover, a previous study has shown that MR analysis using summary statistics has similar efficiency as using individual-level data.[30] In the present study, we utilized single-nucleotide polymorphisms (SNPs) strongly associated with BW/CBMI as IVs and performed two-sample MR analyses using the effect of IVs on the exposures and outcomes from large GWASs.[15,31,32] The summary-level data were analysed to obtain quantitative estimates of the causal effects of BW and CBMI on the AAM to investigate the potential role of BW and CBMI on AAM. The large sample size used here supports more precise estimates, and the insights from this study will help fill the gaps towards this relationship in the observational studies as well as select the timing for early intervention to avoid early menarche development.
Clin Endocrinol. 2023;98(2):212-220. © 2023 Blackwell Publishing
