The challenge of Mendelian randomization approach

Abstract

This review deals with methodological problems in current evidence that suggest links between gout and adverse cardiovascular (CV) outcomes and renal disease, and presents a recently adopted approach aimed at overcoming such drawbacks. The review aims to provide more reliable answers regarding the role of uric acid in the pathogenesis and progression of cardiovascular and renal events.

Keywords:

• Mendelian randomization
• hyperuricemia
• cardiovascular risk
• randomized controlled trial

Introduction

Evidence is accumulating suggesting links of gout with adverse cardiovascular (CV) outcomes and renal disease, and studies are being carried out to demonstrate that hyperuricemia can actually be considered an independent major CV risk factor and a causal factor for progression of chronic kidney disease (CKD). A major obstacle to this line of research is the fact that many patients with gout who develop major CV and renal events also possess several known traditional CV risk factors, potentially confounding any true relationship between gout and adverse CV/renal events¹.

This paper deals with methodological problems affecting existent evidence and presents a recently adopted approach aimed at overcoming such drawbacks, and promising to provide more reliable answers about the role of uric acid, in the pathogenesis and progression of cardiovascular and renal events.

Links between uric acid and vascular-renal disease

Hyperuricemia was associated with deleterious effects on endothelial dysfunction, oxidative metabolism, platelet adhesiveness, hemorheology, and aggregation^2,3. A pathogenetic role of hyperuricemia in the progression of renal disease has been proposed based of mechanisms activated by high levels of uric acid, mainly at the vascular level⁴. The activity of xantine oxidase, the enzyme involved in purine catabolism and involved in production of uric acid from hypoxanthine, has been recognized as one of the main sources of oxidative stress within the endothelial cells (Figure 1); this phenomenon is responsible for reduced availability of endothelial nitric oxide, thus indicating the presence of endothelial dysfunction⁶. Chronic hyperuricemia stimulates the renin-angiotensin system, and through this pathway inhibits release of endothelial nitric oxide, contributing to renal vasoconstriction and increasing blood pressure^7,8; at the same time, high levels of uric acid may have a pathogenetic role in interstitial inflammation and progression of renal disease.

Figure 1. Xanthine oxidoreductase reactions. For XDH, xanthine is oxidized to uric acid at the Mo-co and electrons transferred via two Fe/S centers to the FAD where NAD + is reduced to NADH (left). For XO, xanthine is oxidized to uric acid at the Mo-co and electrons are transferred to the FAD where O₂ is reduced to O₂•− and H₂O₂ (right). Conversion from XDH to XO is mediated by post-translational modification. Reprinted from Pharmacol Rep 2015;67:669–74 with permission from Elsevier⁵.

Some clinical trials and experimental studies in animal models suggest a benefit of uric acid-lowering interventions on renal disease progression⁴.

Drawbacks of published evidence

Notwithstanding, several studies suggest the pathophysiological role of uric acid in vascular damage, leading to endothelial dysfunction and increased risk for renal disease.

Definitive evidence of a causal relationship of hyperuricemia with CKD has not been produced, and large-scale clinical trials testing the effect of urate lowering therapy on major clinical end-points of CKD progression are still lacking⁹.

Recently, the efficacy and the risks of urate lowering therapy to improve outcomes in patients with CKD were evaluated by a systematic review of currently published studies, and a meta-analysis of results was carried out¹⁰. Randomized controlled trials in patients with stages 3–5 CKD, greater than 3 months and comparing allopurinol and inactive control, were included. Statistically significant reductions in serum uric acid, systolic, and diastolic blood pressure were found, favoring allopurinol, but there were insufficient data for analysis on adverse events, incidence of end stage renal disease (ESRD), and CV disease. The authors concluded that data are promising, but adequately powered randomized clinical trials are needed to establish whether treatments that lower urate have beneficial renal effects¹⁰.

Observational studies were unable to capture the link between uric acid and CKD, because of methodological inadequacy. Prevailing uric acid levels may not reflect the true uric acid burden in CKD because of variability over time, and the influence of many unstable factors such as the hydration status, the use of diuretics, allopurinol, and other drugs use, acid base status, and glomerular filtration rate (GFR) loss⁹.

Mendelian randomization

Since novel, large scale clinical trials are not yet available, the problem has been recently explored by studies based on the concept of Mendelian randomization. Mendelian randomization is a method that utilizes genetic variants that are robustly associated with such modifiable exposures, to generate reliable evidence regarding interventions that should produce health benefits¹¹.

This approach could overcome the problem of confounding factors present in observational studies, maintain the relevant benefit of randomization, and enable researchers to answer questions by shorter and less expensive studies in comparison with large randomized clinical trials. The environment does not confound genes, by definition: the study of the association of a gene polymorphism with a clinical outcome may give a distinctive assessment. Because genes are transmitted randomly at mating (Mendelian randomization), allele polymorphism may be applied as a surrogate end-point to explore the casual link of risk factors with events in cohort studies⁹. The association between a disease and a polymorphism that mimics the biologic link between a proposed exposure and disease is not generally susceptible to reverse causation or confounding that may distort interpretations of conventional observational studies. One pre-requisite for performing Mendelian randomization is that the investigated genetic loci are exclusively associated with the outcome via their effect on the concentration of the biomarker in question, and not via alternative pathways⁹.

A surrogate marker of hyperuricemia

Uric acid concentrations show a substantial degree of heritability. Most studies report heritability estimates of ∼40%¹². GLUT9 is a major, high-capacity urate transporter in the proximal nephron that exchanges both glucose and fructose for urate. A very large meta-analysis in 28,000 individuals showed that the rs734553 single-nucleotide polymorphism (SNP) of the GLUT9 urate transporter gene was the strongest marker of hyperuricemia described thus far¹³.

The rs734553 polymorphism of the GLUT9 gene was applied as a surrogate of hyperuricemia to explore the link between uric acid level and renal outcomes using the Mendelian randomization approach⁹. A cohort of 755 patients with CKD was enrolled, and the association between the rs734553 SNP and uric acid level was preliminarily confirmed in a series of 211 healthy volunteers from the same geographic area as the patients with CKD. The study end-point was composite: 30% decrease in the GFR, dialysis, or transplantation. In healthy individuals, serum uric acid levels were highest in homozygotes for the T or risk allele, intermediate in heterozygotes for the same allele, and lowest in those without the risk allele (p = .001), but no such relationship was found in patients with CKD. In the CKD cohort, homozygotes (TT) and heterozygotes (GT) for the risk allele had a 2.35-times higher risk (hazard ratio = 2.35; 95% confidence interval = 1.25–4.42; p = .008) of CKD progression in comparison with patients without the risk allele (Figure 2). Adjustment for proteinuria, GFR, and other classical and CKD-peculiar risk factors did not influence the risk for CKD progression by rs734553. The authors concluded that a GLUT9 polymorphism, which is strongly associated with serum uric acid levels in healthy individuals of the general population with normal renal function, holds a strong predictive power for CKD progression⁹.

Figure 2. Cumulative events-free survival (Kaplan–Meyer curves) for renal events in GG and GT + TT patients. The comparison between the two groups was made by the log-rank test. The table under the x-axis indicates the number of patients at risk at relevant time points. Reproduced with permission⁹.

The same surrogate end-point was used later to investigate the link of hypeuricaemia with CV disease¹⁴. A meta-analysis of three cohort studies formed by high CV risk patients (MAURO: 755 CKD patients; GHS: 353 type 2 diabetics and coronary artery disease; and the TVAS: 119 patients with myocardial infarction) was carried out. An increased cardiovascular risk was demonstrated in patients carrying the risk allele. The T allele of the rs734553 polymorphism in the GLUT9 gene predicted a doubling in the risk for incident CV events in patients at high cardiovascular risk (pooled HR = 2.04, 95% CI = 1.11–3.75, p = .02). The results of this study supported the hypothesis of a causal role of hyperuricemia in cardiovascular disease in high-risk conditions¹⁴.

Some conflicting older results of Mendelian randomization studies evaluating different allele polymorphisms associated with uric acid and link with cardiovascular risk should be mentioned. Palmer et al.¹⁵ conducted a Mendelian randomization analysis and found no strong evidence for causal associations between a variant at the SLC2A9 gene (rs7442295), associated with uric acid levels, and ischemic heart disease or blood pressure. Another large meta-analysis, including over 28,000 participants, showed no association between nine different loci associated to serum uric acid levels and the risk of coronary artery disease¹⁶.

Conclusions

Mendelian randomization is a novel approach that helps to study the association of a risk factor with clinical outcomes. This novel approach was applied when large clinical trials were needed, and/or confounders impaired observational studies. Some studies using this method could demonstrate that uric acid level is an independent and causal risk factor for progression of renal disease, and CV events in high-risk subjects.

Although Mendelian randomization studies may appear promising, it cannot be ignored that several scenarios violating the Mendelian randomization assumptions can be limitations, which are sometimes difficult to overcome. These include inadequate phenotype definition; time-varying exposures; the presence of gene-environment interaction; the existence of measurement error; the possibility of reverse causation; and deviations in the expected allelic frequencies due to several causes (population stratification, existence of linkage disequilibrium phenomena, pleiotropic activity of some genes, etc.). The limitations also include the existence of an appropriate instrument (simple or multiple polymorphism, or genetic risk scores) for the study of interest; the lack of compliance with the conditions that should be met by instrumental variables; and problems with sample size.

In conclusion, the results obtained using this novel approach have added new strength to the observation of benefits of urate lowering therapy in patients with CKD, and suggested the usefulness of further clinical trials.

Transparency

Declaration of funding

This review was funded by Fondazione Menarini.

Declaration of financial/other relationships

RP declares sponsorship and speakers bureau from Menarini. Peer reviewers on this manuscript have received an honorarium from CMRO for their review work, but have no other relevant financial relationships to disclose.

Acknowledgments

Editorial assistance for this supplement was provided by Content Ed Net funded by Fondazione Menarini.