The Mendelian Randomization Study Of Obesity And Primary Liver Cancer Risk

ObjectiveIn this study,Mendelian randomization method was used in the Chinese population,and genetic variants associated with BMI were used to construct a genetic risk score（GRS）,which was used as an instrumental variable to analyze the causal association between obesity and the risk of primary liver cancer（PLC）.Methods1.A case-control study based on gender and age frequency matching was conducted.1285 newly diagnosed PLC cases were recruited from Shunde Hospital of Southern Medical University during September 2010and October 2018 and 1285 healthy controls were enrolled from hospital and communities in the same period.2.A total of eight genetic variants（rs17782313,rs12597579,rs261967,rs10767664,rs9356744,rs10938397,rs1137101,rs7498665）associated with BMI were selected,which reported by GWAS and verified in Asian populations.Genotyping was detected by Taq Man real-time quantitative PCR.3.BMI was divided into four grades according to quartile method,and multivariate Logistic regression model was used to analyze the observed effect of BMI on the risk of PLC in 1285 cases and 1285controls.4.Unweighted GRS and weighted GRS were constructed using the effect size and risk allele of BMI associated genetic variants on BMI reported in GWAS Meta studies or large-sample studies.Unweighted GRS and weighted GRS were divided into four grades according to the quartile method.The generalized linear regression model was applied to estimate the association effect between GRS and BMI in 1285 healthy controls and weak instrumental variable was tested by calculating F statistics.5.Using multivariate Logistic regression model to estimated the association effect between GRS and PLC in 1285 PLC cases and 1285healthy controls.6.The expected effect of BMI on the risk of PLC was calculated using the formula:OR_IV=exp(Ln(OR_GRS-PLC)/β_GRS-BMI).7.The observed effect and expected effect of BMI on the risk of PLC were examined by the Z test.8.Linear regression model and Logistic regression model were applied to test the pleiotropy of genetic variants and GRS with confounding factors.Results1.The risk of PLC decreased with the increase of BMI level.Among them,individuals with the highest BMI had a 68%lower risk of PLC than those with the lowest BMI（adjusted OR=0.32,95%CI=0.24-0.43）after adjustment by the multivariate model.2.The association of GRS with BMI showed that for each point and level increase of unweighted GRS,BMI increased by 0.04 standard deviation（SD）and 0.05 SD（SD:1.68 kg/m²）,and the unweighted GRS Q4level increased BMI by 0.21 SD;for each increase in weighted GRS by one point and one grade,BMI increases by 0.23 SD and 0.05 SD,respectively.The F statistics of the unweighted GRS and weighted GRS were 12.43 and 11.65,The statistical power of both GRS were 100%.3.Results of association analysis between GRS and the risk of PLC showed that after adjusting for gender,age,cancer family history and BMI,the risk of PLC with the highest risk grade of unweighted GRS increased by 4%（adjusted OR=1.30,95%CI=1.04-1.63）;and the risk of PLC with the highest risk grade of weighted GRS increased by 26%（adjusted OR=1.26,95%CI=1.01-1.58）;In addition,for each grade increase of weighted GRS,the risk of PLC increased by 8%（adjusted OR=1.08,95%CI=1.00-1.16）.4.The expected effect size of the highest BMI grade on PLC was1.45（adjusted OR=1.45,95%CI=1.06-4.91）,which was higher than the observed effect size of 0.32（adjusted OR=0.32,95%CI=0.24-0.43）.Z test showed that the difference between the expected effect size and the observed effect size was statistically significant（P<0.001）.5.The results of the pleiotropic test showed that none of the genetic variants and GRS was statistically associated with confounding factors.ConclusionsThe association between obesity and PLC found in the case-control study may be caused by confounding bias and reverse causation.The result of Mendelian randomization study show that there is a causal association between obesity and PLC.Obesity can increase the risk of PLC.The results of this study still need to be further confirmed by a large sample of Mendelian randomization studies.