肥胖与儿童青春发动时相提前的因果推断：孟德尔随机化研究

The relationship between obesity and early timing of puberty in children has attracted much attention. However, due to the confounding bias and reverse causation in the traditional observational epidemiological studies, the true effect of obesity on early puberty remains to be clear. Mendelian randomization analysis, using genetic variants as instrumental variables, can overcome the deficiency of bias and reverse causation in observational studies, effectively inferring causality between exposure factors and outcomes. In this study, we will conduct an improved Mendelian randomization analysis, based on our ongoing Guangzhou cohort for prevention and control of obesity in school students. Firstly, in 4800 students, we will construct a genetic risk score associated with body fat percentage/obesity to be a robust instrumental variable. Then, we will identify boys and girls with early puberty, through two-year follow up, and will analyze the observational effect size of obesity on early puberty and the expected effect size of instrumental variable on early puberty, applying a 1:2 matched nested case-control study design. Finally, we will compare the difference between the observational and expected effect estimates. We will present the true effect of obesity on early timing of puberty, after the perfect control of confounding bias and reverse causation. This study will provide scientific evidence for screening of children in high risk of early puberty and for making intervention measures of early puberty.

肥胖与儿童青春发动时相提前的关系备受关注。但因传统观察性研究受到混杂偏倚、因果时序混淆干扰，肥胖对青春发动时相提前的真实效应如何尚待明确。孟德尔随机化方法因使用遗传变异为工具变量，可克服混杂、因果时序混淆问题，从而有效推断暴露因素与研究结局的真实关系。本项目拟依托课题组建立的广州市中小学生肥胖防治队列，采用改良的孟德尔随机化研究策略：（1）在4800名学生中构建体脂百分比/肥胖关联的遗传风险评分为稳健工具变量；（2）通过两年随访识别青春发动时相提前的男、女生，应用1：2匹配巢式病例对照研究设计，评价肥胖对青春发动时相提前的观察效应值，并计算工具变量对青春发动提前的期望效应值；（3）比较观察效应值和期望效应值的差异，在排除混杂、因果时序混淆干扰后，揭示肥胖对儿童青春发动时相提前的真实效应，为及早识别青春发动时相提前的高风险儿童提供筛查依据，并为儿童青春发动时相提前的干预措施制定提供参考。

肥胖与儿童青春发动时相提前的关系备受关注。但由于传统观察性研究受到混杂、时序混淆干扰，肥胖与青春发动时相之间的真实关系如何尚待明确。本项目基于纵向研究设计，在2232名广州市儿童中采用双向孟德尔随机化方法（Mendelian Randomization analysis，MR），应用17个肥胖遗传变异以及15个青春期发育遗传变异分别构建遗传风险评分作为工具变量，克服传统观察法的混杂和时序混淆弊端，探讨肥胖与青春发动时相之间是否存在双向因果关系。项目主要发现：（1）体脂率增加或肥胖可引起女童青春发动时相提前，而男童中未见此关联。在观察法分析中，体脂率Z值与女童青春发动时相提前风险呈正关联（OR=1.79，95% CI=1.58-2.03）。肥胖女童的青春发动时相提前风险较非肥胖女童增高约1.4倍（OR=2.41，95% CI=1.82-3.20）。在MR比值法分析中，体脂率每增加1个标准差，女童青春发动时相提前风险增高约2倍（OR=3.01，95% CI=2.21-8.53）。相比于非肥胖女童，肥胖女童的青春发动时相提前风险增高51%（OR=1.51，95% CI=1.27-2.09）。（2）青春发动时相提前可导致男童体脂率增加，但对女童体脂率及肥胖无影响。在观察法中，男童青春发动时相提前与体脂率Z值增加呈负关联（β=-0.12，95%CI=-0.18-0.06）。在MR比值法分析中，与青春发动时相未提前的男童相比，时相提前的男童体脂率Z值增加0.08（β=0.08，95% CI=0.04-0.10）。上述研究发现说明肥胖与儿童青春发动时相的关系存在性别差异。本项目所阐明的肥胖与青春发动时相关系可进一步丰富肥胖及青春发动时相提前的机制理论，可帮助预测儿童肥胖或青春发动时相提前风险，并可为肥胖和青春发动时相提前的预防及干预策略和措施制定提供参考依据。