The Comparative Effect of Exposure to Various Risk Factors on the Risk of Hyperuricaemia

Diet Has a Weak Causal Effect

Ruth K. G. Topless; Tanya J. Major; Jose C. Florez; Joel N. Hirschhorn; Murray Cadzow; Nicola Dalbeth; Lisa K. Stamp; Philip L. Wilcox; Richard J. Reynolds; Joanne B. Cole; Tony R. Merriman

Disclosures

Arthritis Res Ther. 2021;23(75) 

In This Article

Abstract and Introduction

Abstract

Background: Prevention of hyperuricaemia (HU) is critical to the prevention of gout. Understanding causal relationships and relative contributions of various risk factors to hyperuricemia is therefore important in the prevention of gout. Here, we use attributable fraction to compare the relative contribution of genetic, dietary, urate-lowering therapy (ULT) and other exposures to HU. We use Mendelian randomisation to test for the causality of diet in urate levels.

Methods: Four European-ancestry sample sets, three from the general population (n = 419,060) and one of people with gout (n = 6781) were derived from the Database of Genotypes and Phenotypes (ARIC, FHS, CARDIA, CHS) and UK Biobank. Dichotomised exposures to diet, genetic risk variants, BMI, alcohol, diuretic treatment, sex and age were used to calculate adjusted population and average attributable fractions (PAF/AAF) for HU (≥0.42 mmol/L [≥7 mg/dL]). Exposure to ULT was also assessed in the gout cohort. Two sample Mendelian randomisation was done in the UK Biobank using dietary pattern-associated genetic variants as exposure and serum urate levels as outcome.

Results: Adherence to dietary recommendations, BMI (< 25 kg/m2), and absence of the SLC2A9 rs12498742 urate-raising allele produced PAFs for HU of 20 to 24%, 59 to 69%, and 57 to 64%, respectively, in the three non-gout cohorts. In the gout cohort, diet, BMI, SLC2A9 rs12498742 and ULT PAFs for HU were 12%, 49%, 48%, and 63%, respectively. Mendelian randomisation demonstrated weak causal effects of four dietary habits on serum urate levels (e.g. preferentially drinking skim milk increased urate, β = 0.047 mmol/L, P = 3.78 × 10−8). These effects were mediated by BMI, and they were not significant (P ≥ 0.06) in multivariable models assessing the BMI-independent effect of diet on urate.

Conclusions: Diet has a relatively minor role in determining serum urate levels and HU. In gout, the use of ULT was the largest attributable fraction tested for HU.

Introduction

Hundreds of genetic variants are associated with serum urate levels,[1–3] and observational studies have associated individual dietary factors (e.g. alcohol, sugar-sweetened beverages, coffee, red-meat consumption[4–8]) and overall eating habits[9,10] with urate levels, along with other environmental (e.g. diuretic use[11,12]) and endogenous factors (e.g. age and sex[13]). Understanding the importance of risk factors and their causal relationship (if any) with HU is critical in developing strategies for the prevention of HU and gout. However, addressing causality is challenging. Observational, longitudinal or migratory studies, and temporal correlations can only be regarded as hypothesis-generating owing to the intractable issue of unmeasured confounding. Any attempts to draw conclusions with respect to causality, including using causal language/inferences, even from an accumulation of studies, mis-represents the evidence.[14] In this context, we note that gout is a multi-stage process beginning with HU, progressing to deposition of monosodium urate crystals and culminating in an innate immune response to crystals.[15] Not all people with HU develop gout,[16] so HU and the progression from HU to gout should not be conflated when considering possible causal risk factors.

The gold standard for testing an exposure for a causal role is the randomised clinical trial (RCT). This approach has demonstrated causality for dissolved sugar (sugar-sweetened beverages) in raising urate levels.[17–19] However, for the majority of suspected causal exposures an RCT is not possible. Mendelian randomisation (MR) exploits the natural randomisation of alleles causal for a particular exposure and is analogous to a RCT. Several MR studies have shown a small causal effect of BMI on urate levels (0.0045 to 0.010 mmol/L [0.075–0.17 mg/dL] increase in serum urate per unit increase in genetically determined BMI[20–22]). Dietary preferences have a heritable component and genetic associations have been reported.[23–25] Mendelian randomisation in the UK Biobank, using genetic variants associated with dietary patterns has demonstrated that a 'healthful' versus 'unhealthful' dietary pattern is not strongly causal for coronary heart disease or type 2 diabetes, despite diet being strongly correlated with these diseases.[26]

Using the widely applied approach of decomposing variance, where the sum of multiple risk factors included in a model is constrained to 100%, overall diet contributed ≤0.3% of variance in urate levels, substantially less than the 23.9% explained by inherited common genetic variants.[10] Why so little variance is explained is unclear, but one possible reason is that overall diet, which comprises some foods associated with increased urate and some foods associated with decreased urate, does not play a strong causal role. This possibility is supported by a RCT that reported a small 0.021 mmol/L (0.35 mg/dL) reduction in urate levels in people following the Dietary Approaches to Stop Hypertension (DASH) diet compared to those on a 'typical' US diet.[27] Another RCT comparing the Mediterranean diet to a 'prudent Westernised diet' reported a small reduction in serum urate levels (0.010 mmol/L [0.17 mg/dL]) over 5 years.[28] Another possible explanation postulated in ref.[9] for the small amount of variance explained is low variability in diet within the US cohorts used in.[10] However, differences in diet between men and women, across age groups, socioeconomic status, ethnicity and BMI strata in the US have been reported[29]—to observe these differences there must be variability in the diet.

Population attributable fraction (PAF) is the proportion of cases for an outcome within a population that can be attributed to a given risk factor, incorporating both the prevalence and the effect size of the exposure.[30] The sum of PAFs for multiple risk factors for a single condition is not constrained to 100%, because an outcome can have multiple risk pathways population-wide. Using the Third National Health and Nutrition Examination Survey PAFs were reported of 44% for being overweight or obese (implying that 44% of HU would be prevented if the entire population had BMI < 25 kg/m2) and 9% for non-adherence to a DASH-style diet.[9] Variances explained were 8.3% and 0.1%, respectively,[9] indicating the two methods agree as to which exposure has a greater impact.

Our first aim was to use attributable fraction to compare the contributions to HU of various genetic, environmental and endogenous risk factors, including lack of use of urate-lowering therapy. The second aim was to use MR to test for a causal role of diet in determining urate levels.

processing....