Study design
This study used 2 blood pressure phenotypes and 5 gout phenotypes for Mendelian randomization analysis, the blood pressure phenotypes were obtained from the IEU database (https://gwas.mrcieu.ac.uk/), and the gout phenotype was obtained from the FinnGen database (FinnGen http://www.finngen.fi/en). Statistical analyses were performed by the “TwoSampleMR” package (version 0.5.6) of the R program (version 4.2.2). Only summary-level statistics were used in our study and therefore ethical approval was not required.
GWAS data for hypertension
The GWAS summary data for blood pressure were obtained from the International Consortium of Blood Pressure, Participants used sex, age, age-squared, BMI, and chip as covariates and Meta-analysis of cohort/cross-sectional studies as study design. There were 757,601 European populations of systolic blood pressure Sample size and 7,088,083 SNPs were tested. The diastolic blood pressure phenotypescame from 757,601 European descents and 7160619 SNPs were tested.
GWAS data for gout, idiopathic gout, unspecified gout, strictly defined gout, and drug-induced gout
Summary statistics for gout, idiopathic gout, unspecified gout, strictly defined gout, and drug-induced gout were obtained from the FINNGEN Consortium. They all refer to ICD criteria as diagnostic criteria.
There were 7461 Finnish populations of gout cases and 221323SNPs, 429209 people referred to the ICD criteria for gout as a diagnostic criterion and 9568 people were finally screened., with a mean age at first event 67.05 years for females and 64.43 years for males.
The Idiopathic gout phenotype, which includes 1851 cases and 335038 controls. 429209 people referred to the ICD criteria for idiopathic gout as a diagnostic criterion, and 2433 people were finally screened, with a mean age at first event (years) of -68.44 years for females and 66.23 years for males.
Drug-induced gout: There are 110 cases and 342389 controls. 429209 people referred to the ICD criteria for drug-induced gout as a diagnostic criterion, and 129 people were finally screened, with a mean age at the first event (years) of -70.89 years for females and 67.77 years for males.
Unspecified gout: There are 4607 cases and 335038 controls. 429209 people referred to the ICD criteria for unspecified gout as a diagnostic criterion, and 5945 people were finally screened, with a mean age at the first event (years) of -69.10 years for females and 67.64 years for males.
Strictly defined gout: There are 3768 cases and 335038 controls. 429209 people refer to the ICD criteria for unspecified gout as a diagnostic criterion, and 5162 people were finally screened, with a mean age at the first event (years) of -65.70 years for females and 61.58 years for males.
Mendelian randomization analysis
MR analysis must satisfy the three main assumptions that SNPs are strongly correlated with exposure (correlation assumption), that SNPs are not correlated with outcome, and that SNPs are not correlated with confounders. To satisfy the correlation assumption, we used P < 5 × 10− 8 and linkage disequilibrium (LD) < 0.001 as SNPs correlated with exposure (hypertension in this study). SNPs were strongly correlated with exposure when F > 10. To satisfy the exclusivity assumption, we manually excluded SNPs that were associated with outcomes at P < 5 × 10− 8. In addition, we also check the Phenoscanner website (http://www.phenoscanner.medschl.cam.ac.uk/) to see if SNPs are correlated with underlying factors that might potentially affect the outcome.
In detail, we obtained 461 SNPs associated with systolic blood pressure strength and 460 SNPs with diastolic blood pressure strength (P < 5 × 10− 8, LD r2 < 0.001). We then extracted the exposure SNPs in the GWAS data of five gout phenotypes and excluded SNPs associated with them (P < 5 × 10− 8). For SNPs missing from the outcome, proxies were identified in high LD (r2 > 0.8) according to the European reference panel of the 1000 Genomes Project. For those SNPs for which no suitable proxies were found, we discarded them. Alleles of the exposure and outcome SNPs were then aligned to make sure that the effect alleles were concordant, and SNPs with intermediate effect allele frequencies (EAF > 0.42) and incompatible alleles were discarded for MR analysis.
We applied the inverse variance weighted (IVW) estimation in the main analysis, which combines the Wald ratio of each SNP to the outcome to obtain a pooled causal estimate. The IVW method has significantly higher statistical power than other MR methods and is thus typically used as the primary approach[20]. Besides, we considered a significant association based on an IVW estimation with P < 0.005 (Bonferroni correction: 0.05/2 exposures/5 outcomes). Two other MR models, including the weighted median (WM) and MR-Egger method, were used as complementary methods. The WM method assumes that at least half of the instruments were valid, and the statistical power is mildly weaker than the IVW method[21]. MR-Egger regression provides consistent estimates accounting for pleiotropy when all the instruments are invalid[22].
Sensitivity analysis
To test the reliability of our MR analysis, a sensitivity analysis was required. We used Cochran's Q test, leave one out (LOO), and MR-Egger intercept analysis for sensitivity analysis[22]. Heterogeneity exists when the P-value of Cochran's Q test is less than 0.05[23]. We assessed horizontal pleiotropy results based on the intercept term derived from the MR-Egger intercept analysis[22]. When P < 0.05 for MR-Egger intercept analysis, there was pleiotropy[24]. To determine whether causal estimates were driven by any individual SNPs, we performed a LOO analysis, discarding each exposure-related SNP in turn by LOO, to replicate the IVW analysis.
Risk factors analysis
To explore the potential mechanisms underlying the phenotypic genetic association of hypertension and gout, we further conducted MR analyses to investigate the causal effect of hypertension on several potential mediators, including body mass index (BMI), and smoking. The genetic effects of BMI came from the Genetics of Anthropometric Traits Research (GIANT) consortium[25]. Smoking phenotype GWAS summary data were obtained from the FinnGen database. The IVW method was used as the major analysis, with P < 0.05 considered as a significant association.