Assessment of causal associations between uric acid and 25-hydroxyvitamin D levels

Abstract

Background: Previous observational studies have revealed the association between serum uric acid and 25-hydroxyvitamin D. However, the causality and the direction of the associations remain unknown. Thus, we performed a two-sample bidirectional Mendelian Randomization (MR) analysis to investigate the causal association between uric acid and 25-hydroxyvitamin D and to determine the direction of the association.

Method: Based on the summary-level GWAS data from large genome-wide association studies, several steps were taken in our analysis to select eligible single-nucleotide polymorphisms (SNPs), which were strongly related to exposure as the instrumental variables. We used different analytical methods, such as inverse-variance weighting (IVW) method, weighted median, MR-Egger regression, and weighted mode method, to make our result more robust and reliable. The IVW method was used as the primary analysis. The Cochran's Q test, MR-Egger intercept test, MR-PRESSO method, and "leave-one-out" sensitivity analysis was performed to evaluate the heterogeneities, horizontal pleiotropy, and robustness of the results. MR analyses were also conducted using genetic risk scores (GRS) as instrumental variables in both directions by using the same summary-level GWAS data.

Results: Our two-sample MR analysis suggested a causal association of genetically predicted uric acid on 25-hydroxyvitamin D [IVW method: β(SE), -0.0352(0.0149); p = 0.0178], which suggested that a per mg/dl increase in uric acid was associated with a decrease of 0.74 nmol/L of 25-hydroxyvitamin D, and the above results remained stable in the sensitivity analysis. By contrast, four MR methods suggested no causal relationship of 25-hydroxyvitamin D on serum uric acid [IVW β(SE), 0.0139 (0.0635); p = 0.826; MR-Egger β(SE), 0.0671 (0.108); p = 0.537; weighted median β(SE), 0.0933 (0.0495); p = 0.0598; weighted mode β(SE), 0.0562 (0.0463); p = 0.228, respectively]. After excluding the SNPs, which were associated with confounding factors and outlier SNPs, the IVW method suggested that there was still no causal association of 25-hydroxyvitamin D on serum uric acid. The GRS approach showed similar results.

Conclusions: Serum uric acid may causally affect the 25-hydroxyvitamin D levels, whereas the causal role of 25-hydroxyvitamin D on uric acid was not supported in our MR analysis. Our findings suggest that increased levels of uric acid should prompt investigation for vitamin D deficiency.

Keywords: 25-hydroxyvitamin D; causality; inverse-variance weighting method; mendelian randomization; uric acid.

Figure 1

Two-sample bidirectional Mendelian Randomization study of the association of serum uric acid and 25-hydroxyvitamin D. (A) Data sources for investigating the causal association of serum uric acid on 25-hydroxyvitamin D. (B) Data sources for investigating the causal association of 25-hydroxyvitamin D on serum uric acid.

Figure 2

Flow chart about the analytical methods and how the MR analysis was performed step-by-step. (A) serum uric acid on 25-hydroxyvitamin D; (B) 25-hydroxyvitamin D on serum uric acid.

Figure 3

Scatter plot showing the associations of the SNP effects between serum uric acid and 25-hydroxyvitamin D. Circles indicate genetic associations between serum uric acid and 25-hydroxyvitamin D. Error bars indicate 95% confidence intervals. (A) Serum uric acid on 25-hydroxyvitamin D; (B) 25-hydroxyvitamin D on serum uric acid. MR, Mendelian Randomization; SNP, single-nucleotide polymorphism.

Figure 4

(A) Forest plot of causal association for uric acid on 25-hydroxyvitamin D. (B) Forest plot of causal association for 25-hydroxyvitamin D on uric acid.

Figure 5

(A) Leave-one-out analysis of the effect of the uric acid on 25-hydroxyvitamin D. (B) Leave-one-out analysis of the effect of the 25-hydroxyvitamin D on uric acid.

Figure 6

(A) Genetic risk score GRS_SUA for 25-hydroxyvitamin D. (B) Genetic risk score GRS_25(OH)D for serum uric acid. The estimate of causal association is shown by a red solid line with gradient, and 95%CIs are denoted by red dashed lines.