Causal Relationships Between Birth Weight, Childhood Obesity and Age at Menarche

A Two-Sample Mendelian Randomization Analysis

Lianke Wang; Fei Xu; Qiang Zhang; Jiajun Chen; Qianyu Zhou; Changqing Sun


Clin Endocrinol. 2023;98(2):212-220. 

In This Article


In the current study, we detected that lower BW and higher CBMI were associated with later AAM by performing comprehensive MR analyses using GWAS summary statistics. To efficiently avoid possible violation of the MR model assumptions, we carefully selected the genetic variants associated with BW and CBMI to serve as valid IVs and performed extensive sensitivity analyses to ensure the validity of MR analysis. In addition, power calculation results in the MR analyses also revealed a significant association. These findings support the hypothesis that low BW and high CBMI may play causal roles in the pathway of decreasing the time of AAM.

To the best of our knowledge, this is the first two-sample MR analysis using two large GWAS summary statistics to investigate the causal association between BW, CBMI and the AAM outcomes. The relationships between BW, CBMI and AAM were consistent with the majority of previous observational studies. A meta-analysis including 17 longitudinal studies concluded that lower BW and higher body weight in infancy and childhood may increase the risk of early menarche.[22] Later, a cohort study also indicated that small size at birth and rapid infant growth were significantly related to early pubertal age in girls.[19] Another cohort and sibling-matched study demonstrated that overweight and obese girls attained earlier menarche onset compared with normal-weight children, and point estimates ranged −18.0 to −3.1 months per 5 kg/m2 higher CBMI.[20] However, findings regarding the influence of prepubertal adiposity were inconsistent and some failed to support such association.[18,23,44] The divergent results may be possible due to different analytical methods applied in the studies, incomplete measurements of obesity during childhood, the lack of uniform measures and adequate control for potential confounding factors that occur during the individual life. Therefore, we attempted to maximize statistical power by leveraging the summary statistics from the large available GWAS studies to avoid the spurious conclusion obtained from the previous observational studies. Indeed, by using the two-sample MR approach to control for confounders, it showed that lower BW and higher CBMI were risk factors for earlier menarche.

The mechanisms underlying the causal association between BW, childhood obesity and AAM can be understood by several plausible potential explanations. First, previous studies reported low BW children had higher adrenal androgen secretion before puberty and showed no sex difference, this precocious adrenarche could contribute to earlier menarche in girls.[45] Second, another possible mechanism is that a child born with low BW will gain weight rapidly in compensation with a rich postnatal environment, contributing to further acceleration of puberty maturation. The levels of leptin may be the critical link between adiposity and earlier menarche, leptin is a kind of adipocyte-derived protein hormone, and found higher in children with high BMI.[46] Several studies demonstrated that leptin-deficient mice treated with leptin enter puberty earlier than control mice.[47,48] Additionally, the appropriate leptin level is indispensable for the maturation of the hypothalamic–pituitary–gonadal axis. Previous studies revealed that leptin may accelerate the secretion of gonadotropin-releasing hormone, then stimulate the production of luteinizing hormone and follicle-stimulating hormone from the pituitary, resulting in the production of oestrogen in girls.[49] Third, childhood obesity was associated with both insulin resistance and hyperinsulinemia, which may initiate the early onset of menarche either by regulating the hypothalamic–pituitary–gonadal axis. Moreover, insulin can act on some organs including adrenals, ovary, liver and fat cells to increase the bioavailability of sex hormones.[47] Fourth, another potential mechanism for the association is through the increased aromatase activity. The elevated adiposity in obese prepubertal children could lead to increased aromatase activity, contributing to the conversion of androgens to oestrogens and then activating early puberty in girls.[50] The above may be relevant mechanisms, but further studies are needed to examine exact biological mechanisms and find which factor to be targeted for the interventions.

The present study has several strengths. First, this MR study may minimize confounding factors and reverse causal effects which usually appeared in observational studies by using large-sample GWAS summary statistics data. Moreover, the statistical power is large enough to get the causal conclusion. Second, no pleiotropic effects were found in this study and the sensitivity analyses were consistently confirmed by the IVW method, the results were robust. Third, prior observational studies mostly measured BMI at some time points and the association inference may be underestimated by the use of such scattered longitudinal data. In this MR analysis, CBMI was set as a continuous variable in the GWAS, thus we can capture the long-term change of CBMI and obtain an overall effect estimation trend. Nonetheless, there are still several limitations to the study. First, we selected the GWAS data sets used in our study to maximize the sample sizes yet avoid overlapping samples, there was still a very small part between exposures and outcome, thus might introduce bias to the causal estimates. Nevertheless, given the sample size employed, this influence would likely be small. Second, we assumed that the relationship between genetically determined CBMI with AAM was linear, and linear analysis was used to quantify the average causal effect of exposure. We could not explore the nonlinear effects of CBMI because of our use of GWAS summary-level statistics. Further investigation will be needed to explore nonlinear relationships by using individual-level data. Third, our study was limited to individuals of European descent, so our results may not be generalizable to other ancestral groups. Future MR studies in different populations are needed to extend our findings. Fourth, the exposure data sets used in the MR study included both females and males, and there were no sex-specific summary statistics for us to choose from. However, we thought the data sets used in the present study weakened the relationship between BW, CBMI and AAM, thus making our results conservative. For the BW, almost all substudies showed that the BW was a little higher in males than in females, so the genetic association between SNPs and BW was considerably stronger in males than in females.[31] In addition, the causal effect value obtained in the MR study was based on the regression model, so the causal effect value would be smaller for males in comparison with females. Therefore, future MR analysis may be warranted to confirm our findings in female-only samples.

In conclusion, our study suggests that lower BW and higher childhood BMI are causally associated with an increased risk of early menarche. The interventions to improve foetal weight and control childhood obesity should be implemented to prevent early menarche development.