| Rheumatoid arthritis（RA）is a chronic inflammatory autoimmune disease,mainly manifested as cartilage damage.The possibility of disability and the need for lifelong treatment pose a substantial disease burden.While previous epidemiological studies have emphasized the importance of individual modifiable lifestyle factors such as smoking,body mass index,diet,and physical activity in RA risk management,it remains to determine the association between comprehensive lifestyle patterns and RA risk due to the inter-relatedness of these individual lifestyle factors.Furthermore,the collection of lifestyle factor data typically relies on self-reported questionnaires,which can introduce measurement and recall biases in the association between lifestyle and RA.Metabolomics can provide a more accurate reflection of the overall perturbations of biological systems corresponding to different lifestyles by measuring small molecule metabolites.Recent studies have utilized specific sets of metabolites to construct metabolomic profiles of healthy lifestyles,which well explain the protective effects of healthy lifestyles against various chronic diseases.However,it remains to be explored whether metabolomic signatures of healthy lifestyles can explain their beneficial effects on RA.Moreover,with the development of genome-wide association studies（GWAS）,Mendelian randomization（MR）analysis has been widely applied to evaluate the causal associations between interested exposure factors and diseases.MR analysis can combine the advantages of genetics and metabolomics to determine whether the observed associations between metabolomic signatures of healthy lifestyles and RA are indeed causal.In addition,previous studies have mainly established risk prediction models for RA based on traditional epidemiological factors and genetic susceptibility,and further research is needed to investigate whether incorporating metabolomics analysis provides incremental value in RA risk prediction.Against this backdrop,this study aims to investigate the association between metabolic signatures of healthy lifestyles and RA risk in the UK Biobank（UKB）population.Firstly,we will explore the metabolic signatures that reflect a healthy lifestyle and their prospective association with RA;Next,we will use GWAS to identify the genetic loci associated with the metabolic signatures and conduct MR analysis to determine the causal association between the metabolic signature and RA;Finally,we will integrate this metabolic signature and genetic predisposition to improve the risk prediction for RA,with the goal of providing a scientific basis for precision prevention of RA.Part I Association between metabolic signature of healthy lifestyle and the risk of RA:A prospective cohort studyObjective:To identify the metabolic signature that reflect a healthy lifestyle,investigate its prospective association with the risk of RA,and assess whether it mediate the association between a healthy lifestyle and RA.Methods:This study excluded individuals with missing lifestyle questionnaire data and metabolomics data measured by nuclear magnetic resonance（NMR）spectroscopy from the UK Biobank,as well as those who were lost to follow-up or had self-reported or diagnosed RA at baseline,resulting in a total of 87,258 individuals included in the analysis.Healthy lifestyle was based on a five-point scale,with one point awarded for each of the following:healthy diet,regular exercise,non-smoking,moderate alcohol consumption,and normal body mass index（BMI）.NMR metabolomics data measured249 metabolites from non-fasting baseline plasma samples,covering multiple metabolic pathways,including 14 lipoprotein subclasses,fatty acids and their constituents,and various low molecular weight metabolites such as amino acids,ketone bodies,and glycolytic metabolites.Elastic net regression was used to evaluate the correlation coefficients between 249 metabolites and healthy lifestyle scores,and a metabolite signatures reflecting a healthy lifestyle was constructed using the weighted concentration of non-zero coefficients.Cox regression models were used to evaluate the hazard ratios（HRs）and 95%confidence intervals（CIs）of the associations between healthy lifestyle and metabolic signatures with the risk of RA.Causal mediation analysis was also applied to examine the magnitude of the mediating effect of metabolic signatures on the association between healthy lifestyle and RA.To test robustness and potential differences in different subgroups,the associations between healthy lifestyle,metabolic signatures,and RA risk were also examined in age（<50 years,50-60years,>60 years）and sex（female,male）subgroups.In addition,metabolic signatures were divided into three groups based on the 20th and 80th percentiles,i.e.,unfavorable（0-20th）,moderate（20-80th）,and favorable（80-100th）,with unfavorable metabolic signatures serving as the reference level,to assess whether moderate/favorable metabolic signatures had a protective effect on RA.Results:During a median follow-up time of 8.1 years,a total of 557 incident cases of RA were identified among 87,258 participants.Metabolic signatures reflecting healthy lifestyle were characterized by larger mean diameter of high-density lipoprotein（HDL）,increased concentrations of total choline,citrate,polyunsaturated fatty acids（PUFAs）,and omega-3 fatty acids,and decreased concentrations of phosphatidylcholine and acetylglycoprotein.The metabolic signatures were significantly associated with the healthy lifestyle score(Spearman r=0.45,P=4.2×10-15).In prospective association analyses,the metabolic signatures were inversely associated with the risk of RA（HR0.76 per SD,95%CI:0.70-0.83）.The metabolic signatures largely mediated the beneficial effect of healthy lifestyle on RA risk（mediation proportion:64%,95%CI:50-83%）.Compared with participants with unfavorable metabolic signatures,those with favorable metabolic signatures had a 49%lower risk of developing RA（95%CI:33-62%）.The subgroups with the relatively highest predicted absolute risks were those aged>60 years,women,and those with unfavorable metabolic signatures.Conclusion:The metabolic signatures of a healthy lifestyle are associated with adequate energy supply,a favorable lipoprotein spectrum,increased beneficial fatty acids,and reduced inflammation levels.The metabolic signature plays an important mediating role in the reduced risk of RA incidence associated with a healthy lifestyle.Improving lifestyle to maintain favorable metabolic signatures may contribute to RA prevention,especially with greater potential benefits for older women.Part II Association between metabolic signature of healthy lifestyle and the risk of RA:A Mendel randomization studyObjective:To identify the single nucleotide polymorphisms（SNPs）loci associated with the metabolic signatures of healthy lifestyle,and use Mendel randomization（MR）analyses to detect the causality between the metabolic signatures and RA.Methods:Based on a prospective study population,86,675 participants were included in the genetic analysis after further exclusion of genetic mismatches,heterozygosity outliers,missing genotypes,high relatedness,and non-Caucasian individuals.Quality control of imputed genotypes was performed using QCTOOL software,and SNPs with a minor allele frequency（MAF）<0.5%,an imputation information score（INFO）<0.3,or failing Hardy-Weinberg equilibrium test(P<1×10-6)were excluded.Afterward,we conducted a genome-wide association study（GWAS）of the metabolic signature on 22autosomal chromosomes assuming an additive genetic model based on genotype dosages with BOLT-LMM software,which employs a linear mixed model algorithm,and adjusting for age,sex,healthy lifestyle scores,the first 10 genetic principal components,genotyping arrays,and assessment centers.Utilizing the above GWAS summary statistics,the polygenic risk scores（PRS）of the metabolic signature were constructed by weighted allele effect,and the association between the PRS and RA risk was checked in the Cox model.In addition,single-sample and two-sample MR analyses were conducted.For single-sample MR analysis,two-stage least-squares（2SLS）regression was performed,where in the first stage,linear regression was used to estimate the predicted values of metabolic signatures for the genetic instrumental variable（the PRS for metabolic signatures）,and in the second stage,Cox regression was used to regress the predicted values of metabolic signatures on RA with adjustment of age,sex,fasting time,household income,education level,history of cardiovascular disease,diabetes,and cancer.For two-sample MR analysis,independent SNPs associated with metabolic signatures were selected as genetic instrumental variables from the metabolic signature GWAS results using linkage disequilibrium（LD）（r2=0.1,±500kb）and suggestive significance thresholds(P<1×10-5).The association effects between the selected SNPs and RA were extracted from two published RA GWAS meta-analyses:the Finn Gen biobank round 5 analysis（6,236 RA cases and 147,221controls）and the GWAS meta-analysis by Ha E et al.（14,361 RA cases and 43,923controls）.A total of 183 and 190 SNPs from these non-overlapping sample RA GWAS summary statistics were separately included in the two-sample MR analyses The inverse-variance weighted method（IVW）is mainly used in two-sample Mendelian randomization（MR）analysis to combine the effect estimates of multiple SNPs.The two-sample MR analysis results obtained from the two different RA GWAS datasets were then combined into a comprehensive result through an inverse-variance weighted fixed-effect meta-analysis.Results:The GWAS results showed that the heritability of metabolic signatures based on SNPs was 0.1528（SD=0.0132）.The genomic control inflation factorλwas 1.15,which may be due to the large sample size and polygenicity.2708 SNPs reached genome-wide significance(P<5×10-8),representing 42 independent loci and 7important gene clusters,including GCKR,CTB-40H15.4,MLXIPL,RP11-136O12.2,ALDH1A2,SLC13A5 and PLTP.The PRS of metabolic signatures was significantly inversely associated with RA（HR 0.76 per SD,95%CI:0.70-0.83）.Single-sample and two-sample MR analyses also showed inversely causal relationships between genetically predicted metabolic signatures and RA risk（ORs,95%CI:0.84,0.75-0.94and 0.84,0.73-0.97）.Conclusion:The GWAS results suggest the polygenicity of metabolic signatures associated with healthy lifestyle involving biological pathways such as lipid and glucose metabolism.The MR results indicate a significant causality between the metabolic signatures and a reduced risk of RA.Part III Improve RA risk prediction by combining the metabolic signature and genetic predispositionObjective:To investigate whether combining metabolic signature and genetic predisposition can improve RA risk prediction.Methods:The study population was the same as in the previous study.Three independent RA risk prediction models were developed:the basic prediction model,the metabolic feature prediction model,and the genetic predisposition prediction model.The basic prediction model included traditional RA risk factors（age,sex,family income,education level,BMI,smoking,alcohol consumption,diet,and physical activity）.The metabolic signature prediction model was consistent with the prospective cohort study analysis using a Cox model that included metabolic signatures.The genetic predisposition of RA was measured by calculating the polygenic risk score（PRS）,which included SNPs in both the major histocompatibility complex（MHC）region and the non-MHC region.The genetic predisposition prediction model not only included the PRS but also included established gene-environment interaction terms for the human leukocyte antigen-shared epitope（HLA-SE）effect alleles（0/1/2）and smoking pack-years（never smoked,≤10 pack-years,10-20 pack-years,≥20 pack-years）.Furthermore,a more comprehensive combined model was constructed by nesting these three prediction models.The area under the receiver operating characteristic curve（AUC）was used to measure the predictive accuracy of the models.The calibration of the models was evaluated by comparing the actual absolute risk with the predicted absolute risk deciles.The continuous net reclassification improvement（c NRI）and integrated discrimination improvement（IDI）were used to compare the improvement in predictive performance of the combined model relative to the three models,with positive c NRI and IDI indicating improved predictive performance and negative c NRI and IDI indicating decreased predictive performance.Results:The basic prediction model,metabolic signature prediction model,and genetic predisposition prediction model showed moderate accuracy in predicting the risk of RA,with corresponding AUCs（95%CI）of 0.674（0.652-0.696）,0.658（0.634-0.682）,and0.673（0.651-0.695）,respectively.The combined model had the highest relative accuracy（AUC 0.702,95%CI 0.680-0.724）.All models had calibration slopes between0.97 and 1.04,indicating a relative consistency between predicted and actual risks thus reaching good calibration.In addition,when comparing the combined model with the basic prediction model,metabolic signature prediction model,and genetic predisposition prediction model separately,the predictive ability of all models improved to a certain extent:c NRIs>0.3（all P<0.001）and IDIs≥0.07（all P<0.001）.Conclusion:The combination of metabolic signatures and genetic predisposition can moderately improve the prediction of RA risk.The generalizability of the conclusions needs to be further validated in external validation cohorts. |
PDF Full Text Request