Data_Sheet_2_Causal relationship between obesity and iron deficiency anemia: a two-sample Mendelian randomization study.xlsx

BackgroundObservational studies have suggested an association between obesity and iron deficiency anemia, but such studies are susceptible to reverse causation and residual confounding. Here we used Mendelian randomization to assess whether the association might be causal.
MethodsData on single-nucleotide polymorphisms that might be associated with various anthropometric indicators of obesity were extracted as instrumental variables from genome-wide association studies in the UK Biobank. Data on genetic variants in iron deficiency anemia were extracted from a genome-wide association study dataset within the Biobank. Heterogeneity in the data was assessed using inverse variance-weighted regression, Mendelian randomization Egger regression, and Cochran's Q statistic. Potential causality was assessed using inverse variance-weighted, Mendelian randomization Egger, weighted median, maximum likelihood and penalized weighted median methods. Outlier SNPs were identified using Mendelian randomization PRESSO analysis and “leave-one-out” analysis.
ResultsInverse variance-weighted regression associated iron deficiency anemia with body mass index, waist circumference, trunk fat mass, body fat mass, trunk fat percentage, and body fat percentage (all odds ratios 1.003–1.004, P ≤ 0.001). Heterogeneity was minimal and no evidence of horizontal pleiotropy was found.
ConclusionOur Mendelian randomization analysis suggests that obesity can cause iron deficiency anemia.

背景：观察性研究已提示肥胖与缺铁性贫血存在关联，但此类研究易受反向因果关系及残余混杂因素影响。本研究采用孟德尔随机化（Mendelian randomization）方法，以评估该关联是否存在因果关系。
方法：我们从英国生物银行（UK Biobank）的全基因组关联研究中，提取了可能与多种肥胖人体测量学指标相关的单核苷酸多态性（single-nucleotide polymorphisms, SNPs）作为工具变量；同时从该生物银行的全基因组关联研究数据集内，提取缺铁性贫血相关遗传变异数据。采用逆方差加权回归、孟德尔随机化埃格回归（Mendelian randomization Egger regression）及科克伦Q统计量对数据异质性进行评估；通过逆方差加权法、孟德尔随机化埃格法、加权中位数法、最大似然法及惩罚加权中位数法评估潜在因果关系；借助孟德尔随机化PRESSO分析与留一法（leave-one-out）分析识别异常SNPs。
结果：逆方差加权回归分析显示，缺铁性贫血与体重指数、腰围、躯干脂肪量、体脂总量、躯干脂肪率及体脂率均存在显著关联（所有比值比为1.003~1.004，P≤0.001）。数据异质性极低，未发现水平多效性证据。
结论：本项孟德尔随机化分析表明，肥胖可导致缺铁性贫血。