Associations Of Serum Homocysteine Levels And Its Related Genetic Variants With Risk Of All-Cause,Cardiovascular,and Cancer Mortality

One-carbon metabolism（OCM）consists of various biochemical reactions that transfer one-carbon units between different sites,and involves multifarious physiological processes including amino acid homeostasis,epigenetic maintenance,nucleic acid synthesis,and redox regulation.Homocysteine（Hcy）is the downstream product of OCM,its elevated levels reflect disorders in OCM.Previous studies have proposed that elevated Hcy levels are associated with higher risk of all-cause mortality,however,several limitations remained among these studies including limited sample size and insufficient adjustment for confounding factors.Furthermore,the associations between Hcy levels and risk of mortality from CVD and cancer,the two major cause-specific mortality risks,were inconclusive,and whether the risk of mortality from coronary heart disease（CHD）and stroke,two major subtypes of CVD,were associated with Hcy levels remained elusive.Previous studies have investigated the associations between elevated Hcy levels and the risk of all-cause,cardiovascular,and cancer mortality,however,the causality of these associations remain elusive.During recent years,elevation of Hcy levels has been a general trend in the globe.In China,the prevalence of hyperhomocysteinemia（HHcy;Hcy levels >15 μmol/L）has also been increased rapidly.Therefore,it is of great significance in the field of public health to estimate the associations between Hcy levels and the risk of all-cause,cardiovascular,and cancer mortality,as well as to assess the causality of these associations in prospective cohort study with large sample size.Within human bodies,Hcy is metabolized after food intakes,and its levels may be affected by various environmental factors such as dietary intakes,smoking,and exercise.Besides environmental factors,Hcy levels are also affected by genetic factors.The latest and largest meta-analysis of Hcy-related genome-wide association study（GWAS）has identified 18 Hcy-related single nucleotide polymorphisms（SNPs）,explaining more than 5.9% of the Hcy variance among 44 147 Europeans.However,differences in genotype frequency among different races and areas are common.GWAS with large sample size conducted in different populations may discover novel Hcy-related SNPs,as well as verify the reported related SNPs.Gene-environment interaction study can serve as a useful tool in discovering interaction effect of environmental exposure with genetic variant on adverse health outcome.Studies have found that Hcy levels had interaction effects with its related genetic variants on CVD and cancer.However,whether Hcy levels had interaction effects with its related genetic variants on risk of all-cause,cardiovascular,and cancer mortality remained to be elucidated.This kind of investigation may provide novel insights for the precise prevention of the increased mortality risk related to elevated Hcy levels.On the base of the aforementioned background,the current study was established using a large sample of Chinese population from the prospective Dongfeng-Tongji cohort（DFTJ cohort）,with the primary objectives as follow:（1）To comprehensively explore the environmental and genetic risk factors of serum Hcy levels by performing cross-sectional study and GWAS;（2）To investigate the associations between serum Hcy levels and risk of mortality from all-cause,CVD,CHD,stroke,and cancer by prospective study,and further assess the causality of these associations using Mendelian randomization（MR）study;（3）To establish Hcy-related genetic risk score（GRS）based on the SNPs found in our GWAS,and assess the interaction effects between serum Hcy levels and its genetic variants on risk of mortality from all-cause,CVD,CHD,stroke,and cancer by gene-Hcy interaction study.The present dissertation consists of the following three parts:Part Ⅰ Factors associated with serum homocysteine levelsObjective: To explore factors on major lifestyles,occupations and other aspects that are associated with serum Hcy levels by cross-sectional study,and to discover genetic variants that affect serum Hcy levels through GWAS,with the aim to provide novel insights for prevention and intervention of elevated Hcy levels.Methods: This part of the study was based on the first follow-up of DFTJ cohort in 2013 when Hcy was first detected in the cohort.Participants with severe adverse health conditions including CHD,stroke,cancer,and severely abnormal electrocardiogram,or with missing data on Hcy levels were excluded.After the exclusion,19 826 participants were included in the cross-sectional study.On the basis of 31 155 participants who underwent genome-wide genotyping（Illumina Infinium Omni Zhong Hua-8 Chips）in the DFTJ cohort,we further excluded participants who failed to pass the quality control of genotyping or had missing data on serum Hcy levels,and included 20 987 participants in the Hcy related GWAS.Serum Hcy levels were measured through chemiluminescence reaction and were natural logarithm transformed（ln-transformed）to approximate normal distribution.HHcy was defined as a Hcy level >15 μmol/L.Generalized linear regression models were used to assess the associations of factors on general demographic characteristics,socioeconomic status,major lifestyles,occupations,and personal health status with serum Hcy levels.Logistic regression models were applied to evaluate the relations between the above factors and the risk of HHcy.GWAS was performed to discover independent SNP loci that affected serum Hcy levels.Results: The cross-sectional study included 19 826 participants（mean age,62.93 years;58.0% female;35.8% HHcy）.The cross-sectional study found that age ≥65 years old,male,body mass index（BMI）≥24 kg/m²,estimated glomerular filtration rate（eGFR）<60 m L/min/1.73 m²,current or former smoker,nighttime sleep duration ≥9 hours,midday napping >90 minutes,duration of past shift work >5 years,and hypertension were associated with higher serum Hcy levels and HHcy risk,and being divorced or widowed was associated with higher serum Hcy levels,while junior high school education or beyond,dietary intakes of animal protein（meats,fishes or seafoods,milk or dairy products）≥5 times/week,vitamin supplementation,and diabetes were associated with lower serum Hcy levels and HHcy risk（all P’s<0.05）.Further analyses showed that BMI,duration of past shift work,systolic blood pressure（SBP）,and diastolic blood pressure（DBP）had positive associations with serum Hcy levels and HHcy risk,frequency of dietary intakes of beans or soy foods had positive association with serum Hcy levels,and age had positive association with HHcy risk;while eGFR,frequency of dietary intakes of animal proteins,total cholesterol（TC）,triglyceride（TG）,fasting glucose（FG）and glycated hemoglobin（Hb A1c）had negative associations with serum Hcy levels and HHcy risk,and metabolic equivalent of task（MET）of exercise had negative association with serum Hcy levels（all P’s<0.05）.The GWAS included 20 987 participants（mean age,64.39 years;55.5% female;38.8% HHcy）.Our GWAS discovered that 20 independent SNPs were significantly associated with serum Hcy levels(r²<0.2,P≤5×10^-8),explaining about 8.57% of the variance of serum Hcy levels.Among which,12 SNPs locating on MTHFR gene,NOX4 gene,and CBS gene were reported to be associated with Hcy levels for the first time,explaining about 2.89% of the variance of serum Hcy levels.Conclusions: Cross-sectional study found that major factors on general demographic characteristics including age,sex,education levels and marital status,lifestyles including smoking status,dietary intakes,vitamin supplementation,exercise,nighttime sleep,and midday napping,occupations including duration of past shift work,and personal health status including FG,Hb A1 c,blood pressure,and lipid profile were significantly associated with serum Hcy levels.GWAS discovered 20 independent SNPs that affected serum Hcy levels.Part Ⅱ Associations of serum homocysteine levels with risk of all-cause,cardiovascular,and cancer mortalityObjective: To investigate the associations of serum Hcy levels with risk of mortality from all-cause,CVD,CHD,stroke,and cancer through prospective cohort study,and explore the causality of these associations by MR study combining the Hcy-related SNPs reported by the GWAS of the Part Ⅰ and the literature.Methods: The same population of 19 826 participants from the cross-sectional study of Part Ⅰ were included in the prospective study,and were followed-up till 31 December 2018.The outcome of the present study was death,and specific causes of death were recorded and identified according to the tenth version of International Classification of Disease（ICD-10）.Serum Hcy levels were detected through chemiluminescence reaction,and were lntransformed.To estimate the hazard ratio（HR）and 95% confidence interval（CI）for risk of mortality from all-cause,CVD,CHD,stroke,and cancer by different serum Hcy levels,we used Cox proportional hazards regression model with stratifications by age at risk（in 5-year intervals）and sex in model 1,and further adjustments for BMI,eGFR,education levels,smoking status,alcohol intake status,dietary intakes of five primary foods（meats,fishes or seafoods,milk or dairy products,beans or soy foods,fruits or vegetables）,vitamin supplementation,regular exercise,hypertension,diabetes,and dyslipidemia in model 2.Restricted cubic spline（RCS）model was used to evaluate the dose-response relations between serum Hcy levels and the risk of above mortality.The MR study included the same population of 20 987 participants from the GWAS of Part Ⅰ.Instrumental variables were selected based on our GWAS and the literature,and the methods of inverse-variance weighted（IVW）,weighted median（WM）and MR-Egger were used to further assess the causality between serum Hcy levels and risk of above mortality.Moreover,global test of MR-pleiotropy residual sum and outlier（MR-PRESSO）method was used to estimate the horizontal pleiotropy of SNPs,and Cochran’s Q statistics was applied to assess the heterogeneity of SNPs.Results: During a median follow-up of 5.69 years,878 deaths were documented,including 293 CVD deaths（consisted of 187 CHD deaths and 106 stroke deaths）and 97 cancer deaths.In model 2,for per unit increase in ln-transformed serum Hcy levels,the risk of mortality from all-cause,CVD,CHD,stroke,and cancer increased by 49%,76%,67%,92%,and 107%,respectively（all P’s<0.05）.The results of RCS models showed that all the relations of serum Hcy levels with the risk of above mortality were of linear positive associations(all P’s_overall<0.001,all P’s_non-linear>0.05).Stratified analyses suggested that the risk of cancer mortality associated with Hcy levels was substantially higher among participants who ever drank alcohol than never-drinkers(P_interaction=0.03).Six reliable SNPs were chosen as instrumental variables in the MR study,and 3 of which were reported by our GWAS for the first time.None of the method of IVW,WM,or MR-Egger found significant causal relationship between serum Hcy levels and the risk of mortality from all-cause,CVD,CHD,stroke,and cancer.Global test of MR-PRESSO showed no significant horizontal pleiotropy of SNPs,and Cochran’s Q statistics detected none heterogeneity among SNPs.Conclusions: Serum Hcy levels had linear positive associations with risk of all-cause,cardiovascular,and cancer mortality.Alcohol assumption exacerbated the risk of cancer mortality regarding Hcy.MR study did not find any causal relationship between serum Hcy levels and risk of all-cause,cardiovascular,and cancer mortality.Owing to the current limited number of Hcy-related SNPs and its small explained variance of Hcy levels,future GWAS with larger sample size is needed to discover more SNPs that are strongly associated with Hcy levels to improve the efficiency of instrumental variables in the MR study in order to verify our findings.Part Ⅲ Interactions between serum homocysteine levels and its related genetic variants on the risk of mortality from all-cause,cardiovascular disease,and cancerObjective: To assess the potential interaction effects between serum Hcy levels and risk of mortality from all-cause,CVD,CHD,stroke,and cancer by establishing the Hcy-related GRS based on the SNPs reported by our GWAS and the literature.Methods: Based on the 19 826 participants included in the prospective cohort study in part Ⅱ,after further exclusion of participants without genetic data or failed to pass the quality control of genotyping,a total of 15 434 participants were included in this study.Weighted GRS20 and unweighted GRS20 were calculated based on 20 Hcy-related SNPs reported by our GWAS,and weighted GRS18 were calculated based on 18 SNPs reported by the largest meta-analysis of Hcy GWAS.The associations between GRSs and serum Hcy levels were evaluated by generalized linear regression models.Cox proportional hazards regression models were used to estimate the interaction effects of serum Hcy level with each GRS and SNP genotype on the risk of mortality from all-cause,CVD,CHD,stroke,and cancer.Results: All GRSs were positively associated with serum Hcy levels（all P’s<0.05）.Both the weighted GRS20 and unweighted GRS20 had significant interactions with serum Hcy levels on CHD mortality（all P’sinteraction<0.05）.No significant interaction effect was found between GRS18 and serum Hcy levels on the risk of all-cause,cardiovascular,and cancer mortality.Analyses regarding SNP genotypes found that the C allele of rs7130284,the G allele of rs957140,and the T allele of rs11018628 on NOX4 gene aggravated the risk of mortality from CVD and CHD regarding Hcy,the A allele of rs154657 on DPEP1 gene aggravated the risk of CHD mortality regarding Hcy,and the T allele of rs150419659 on MTHFR gene aggravated the risk of stroke morality regarding Hcy（all P’sinteraction<0.05）.Conclusions: Higher Hcy-related genetic risk exacerbated the positive association between serum Hcy levels and the risk of CHD mortality.NOX4 gene,DPEP1 gene,and MTHFR gene may participate in inducing higher risk of cardiovascular mortality associated with elevated Hcy levels.The current study provided novel clues and directions for the precise prevention of fatal cardiovascular injury with respect to Hcy.