Study design and population
The UK Biobank is comprised of data from a population-based cohort study that recruited more than 500,000 participants (aged 40–79 years) who attended one of 22 assessment centers across the UK between 2006 and 2010 and were followed up to 2021 [26]. The present analyses were restricted to individuals who were of white British ancestry because they had available genetic information. Among 502,507 total participants, 84,993 were excluded in this cohort study, including 3124 with prevalent gout, 29,640 of non-British descent, 13,416 with missing data on genetic risk, 38,113 with missing data on any of the lifestyle factors (i.e., alcohol consumption, smoking status, physical activity, and diet) at baseline, and 1033 who were lost to follow-up examinations. This left 416,481 participants remaining for the current study (Additional file 1: Fig. S1).
UK Biobank has ethics approval from the North West Multi-Centre Research Ethics Committee (11/NW/0382). Appropriate informed consent was obtained from participants and ethical approval was covered by the UK Biobank. This research has been conducted using the UK Biobank Resource under the project number of 45676.
Data collection
Information on sex, age, education, socioeconomic status, and employment status was collected through a touchscreen questionnaire and interview, and BMI was obtained from physical measurement. Education was defined as college or university, upper secondary, lower secondary, vocational, or other. Socioeconomic status was defined based on the Townsend deprivation index [27] (encompassing information on social class, employment, car availability, housing) and categorized as low (highest quintile), middle (quintiles 2 to 4), or high (lowest quintile) [28]. Employment status was categorized as working, unemployed, retired, or other. BMI was calculated as weight (kg) divided by height squared (m2) and was categorized as <25 kg/m2 and ≥25 kg/m2. Information on urate, C-reactive protein, serum creatinine, cholesterol, and triglyceride levels was obtained from blood samples collected at study recruitment.
Information on disease history was derived from medical examinations, self-reported medical conditions, and hospital inpatient records, including data on admissions and diagnoses from the Hospital Episode Statistics in England (dating back to 1997), the Scottish Morbidity Record (dating back to 1981), and the Patient Episode Database in Wales (dating back to 1998). Information on death was obtained through linkage to national death registries from May 2006 to February 2021, and the main cause of death for each participant was identified based on International Classification of Diseases 10 (ICD-10) codes.
In the UK Biobank, genotyping was performed by Affymetrix on two arrays, namely the UK BiLEVE Axiom array and UK Biobank Axiom array. Genotypes were imputed using computationally efficient methods combined with the Haplotype Reference Consortium and UK10K haplotype resource. Genetic analysis in this study was conducted using version 3 of the UK Biobank imputed data, which was released in March 2018. Our genotyping data were restricted to 13.7 million single nucleotide polymorphisms (SNPs) following the Neal-lab-performed variant quality control filters (https://github.com/Nealelab/UK_Biobank_GWAS).
Assessment of gout
Gout was ascertained based on information from self-report (Data-Field 20002, code: 1466), medical records (ICD-10 codes: M10.0, M10.2, M10.3, M10.4, M10.9), and death records (ICD-10 codes: M10.0, M10.2, M10.3, M10.4, M10.9) from the Hospital Episode Statistics (England), the Scottish Morbidity Record (Scotland), and the Patient Episode Database (Wales).
Assessment of lifestyle factors
Information on alcohol consumption, smoking status, and physical activity was obtained from the touchscreen questionnaire; diet was derived from the Food Frequency Questionnaire. Alcohol consumption was calculated based on self-reported intake of red wine, white wine, beer, spirits, and fortified wine. No/moderate alcohol consumption was defined as 0 to 14 g/day alcohol for women and 0 to 28 g/day alcohol for men [29]. Smoking status was dichotomized as smoking vs. non-smoking. Physical activity was measured as minutes per week spent walking or engaged in moderate or vigorous activity according to the International Physical Activity Questionnaire (IPAQ). Regular physical activity was defined as engaging in moderate activity ≥150 min per week, vigorous activity ≥75 min per week, or moderate and vigorous activity ≥150 min/week [30]. A healthy diet score was generated based on the seven commonly eaten food groups following a more recent definition of ideal intake of dietary components for cardiometabolic health [31]. A healthy diet was based on intake of at least four of these seven commonly eaten food groups [31]. Additional file 1: Table S1 provides additional details regarding the assessment of healthy lifestyle factors.
In the current study, we confirmed four healthy lifestyle factors including no/moderate alcohol consumption, not smoking, regular physical activity, and a healthy diet. Participants were categorized into three groups according to the number of healthy lifestyle factors: (1) unfavorable (0 or 1 healthy lifestyle factors), (2) intermediate (2 factors), and (3) favorable (3 or 4 factors).
Assessment of polygenic risk score
A weighted polygenetic risk score (GRS) for gout was calculated to assess the cumulative effect of genetic risk on gout. Using an LD clumping cut-off of r2 < 0.01 and conditional analyses, we selected 33 independent SNPs that have been previously associated with gout in a Global Urate Genetics Consortium study of individuals of European descent [18]. The GRS for gout disease is based on genome-wide association studies of individuals of European descent [18]. Therefore, the present study was restricted to individuals whose self-reported ethnic background was white. Details regarding the selected SNPs are provided in Additional file 1: Table S2. Briefly, we summed the number of risk alleles (0, 1, or 2) for each SNP weighted by the effect size (β coefficient) between that SNP and gout from previous GWAS studies (Additional file 1: Methods). An individual-level genetic risk score (GRS) was then derived from the sum of the number of risk alleles present at each SNP weighted by the effect sizes from all SNPs included in the UK Biobank, which was produced using the PLINK “–score” command. Participants were divided into low (lowest tertile), middle (tertile 2), and high (highest tertile) genetic risk categories according to their Z-standardized GRS.
Assessment of cardiometabolic diseases
Cardiometabolic diseases (CMDs) were defined as cardiovascular disease, hypertension, and/or type 2 diabetes. Information on cardiovascular disease status (i.e., presence of coronary heart disease, heart failure, atrial fibrillation, or stroke) was derived from medical records (ICD-10 codes I20-I25, I42, I48, I50, I60-I64, G45, G46). Diabetes was ascertained on the basis of medical records (ICD-10 codes E111), glycated hemoglobin ≥6.5%, and use of anti-diabetic drugs. Hypertension was defined as systolic blood pressure (SBP) ≥140 mmHg or diastolic blood pressure (DBP) ≥90 mmHg, use of anti-hypertension agents, or medical records (ICD-10 codes I10-I13, I15).
Statistical analyses
Baseline characteristics of the study population were compared by incident gout status using t tests for continuous variables and chi-squared tests for categorical variables. If continuous variables did not follow a normal distribution, the Mann–Whitney U test was applied. Incidence rates (IRs) and 95% confidence intervals (95% CIs) per 1000 person-years were calculated for each genetic predisposition and lifestyle profile category.
Cox proportional hazards regression models were applied to estimate the hazard ratios (HRs) and 95% CIs of gout in relation to genetic risk and lifestyle factors. Follow-up year was used as the time scale in the model. Follow-up time was calculated as the time from baseline assessment until the first event of gout, death, or February 31, 2021, whichever occurred first. The proportional hazards assumptions for the Cox model were tested using the Schoenfeld residuals method, and no violations of the assumption were observed. Models were adjusted for sex, age, socioeconomic status, education level, cardiovascular disease, diabetes, hypertension, and concentrations of C-reactive protein, serum creatinine, cholesterol, triglycerides, and BMI. If data was missing for a covariate, we used multiple imputations based on five replications and utilized a chained-equation method to account for the missing data [32]. The combined effect of the lifestyle profile and genetic predisposition on gout disease risk was assessed by creating dummy variables based on the joint exposures to both factors. The presence of an additive interaction was examined by estimating the relative excess risk due to interaction (RERI), the attributable proportion (AP), and the synergy index (SI). Additionally, we examined the multiplicative interaction between lifestyle and genetic predisposition by incorporating the two variables and their cross-product term in the same model. Furthermore, we conducted stratification analysis by CMD status (with vs without) to investigate whether the associations between gout and the joint exposures of lifestyle profile and genetic predisposition varied by CMD status.
Several additional analyses were performed to assess the robustness of our study results. First, we conducted a weighted healthy lifestyle score based on the β coefficients of each lifestyle factor in the Cox proportional hazards regression model adjusted for sex, age, socioeconomic status, education level, C-reactive protein, serum creatinine, cholesterol, triglyceride, cardiovascular disease, diabetes, hypertension, and BMI. Weighted lifestyle score = (β1*factor 1+β2*factor 2+β3*factor 3+β4* factor 4) [4/(β1+ β2 + β3 + β4 )] [33]. This weighted score ranges from 0 to 4 points. Lifestyle was recorded as “favorable” (3 or 4 points), “intermediate” (2 points), and “unfavorable” (0 or 1 points). Next, we used stratification analysis to examine whether the association between gout and the joint exposures of lifestyle profile and genetic predisposition varied by age (<60 vs. ≥60 years) or by sex. Additionally, to address the role of potential reverse causality, we repeated the main analyses in a sample excluding participants who developed incident gout in the first 3-year follow-up period and participants who died within 3 years from baseline. Furthermore, we excluded participants with diuretic antihypertensive drugs at baseline to assess the robustness of our study results. Finally, we assessed the competing risk of non-gout death on the association between gout and the joint exposures of lifestyle and genetic predisposition using the subdistribution method proposed by Fine and Grey [34].
All analyses were performed using STATA 15 statistical software (Stata Corp, College Station, TX, USA) and R (version 3.6.1, R Foundation for Statistical Computing). All P values were two-sided, and statistical significance was set at 0.05.