Serum uric acid and risk of cardiovascular mortality: a systematic review and dose-response meta-analysis of cohort studies of over a million participants

Background Cardiovascular disease (CVD) is the leading cause of death worldwide. Some studies have suggested anassociation between serum uric acid levels and cardiovascular mortality; however, the results have not been summarized in a meta-analysis. Methods A comprehensive search of all related studies until April 2018was performed in MEDLINE/PubMed and Scopus databases DerSimonianand Laird random-effects models were used to combine hazard ratios (HRs) with 95% confidence intervals (CIs). Dose-response analysis was also carried out. Results Thirty-two studies containing forty-four arms with 1,134,073 participants reported association between uric acid and risk of CVD mortality were included in our analysis. Pooled results showed a significant positive association between uric acid levels and risk of CVD mortality (HR 1.45, 95% CI 1.33–1.58, I2 = 79%). Sub-group analysis showed this relationshipwasstronger in women compared to men. Moreover, there was a significant non-linear association between uric acid levels and the risk of CVD mortality (r = 0.0709, p = 0.001). Conclusion Our analysis indicates a positive dose-response association between SUA and CVD mortality risk.


Background
Cardiovascular diseases (CVD) are the first leading cause of death worldwide [1]. This might be due to an increased incidence of chronic diseases such as obesity, diabetes, hypertension, dyslipidemia, and hyperuricemia [2]. Uric acid (UA) is considered the ultimate product of purine metabolism in humans. Evidence suggests that increased levels of serum uric acid (SUA) are associated with the development of hypertension, coronary heart disease (CHD), cardiovascular stroke, cerebrovascular accidents (CVA), and cardiovascular disease [3].SUA concentrations can reflect the amount of purine intake from the diet, inborn purine metabolism, changes in UA secretion (reduced glomerular filtration and tubular secretion, or increased tubular reabsorption), and intestinal degeneration [4]. The relationship between SUA and CVD was first reported more than 50 years ago, and several epidemiological studies were conducted to assess the association between hyperuricemia (HU) and CVD [5]. Although many studies have been conducted assessing the relationship between UA and CVD, there is disagreement about this relationship [6]. These controversies are due to the dual effect of UA in the body [7]. The atherogenic effects of UA include induction of oxidative stress in cells, which reduces the bioavailability of nitric oxide -associated with the activity of platelets and endothelial cells and the differentiation of smooth muscle cells in the vascular system. On the other hand, UA can also have antioxidant properties that can prevent atherosclerosis and improve endothelial function [8].
To gain a greater understanding of the prognostic value of SUA for future clinical decision making, we conducted a meta-analysis of prospective cohort studies with dose-response analysis to determine the relationship between SUA and CVD mortality.

Methods
This meta-analysis conducted by following the Metaanalysis of Observational Studies in Epidemiology study guidelines (MOOSE) [9].

Exclusion and inclusion criteria and data extraction
Non-English articles, reviews papers, editorials, non CVD mortality, non-human studies, in vitro research, case reports, and letters without sufficient data were excluded. Studies that met the following inclusion criteria were included in this meta-analysis: 1) Cohort study design with CVD mortality outcome.
2) Hazard ratio (HR) and the corresponding 95% confidence interval (CI) of CVD mortality were reported based on uric acid levels. The first author, publication year, country, study design, number of participants, mean age, gender, years of follow-up, hazard ratios and 95% CIs of CVD mortality information extracted from included studies.

Statistical analysis
The STATA 14.0 statistical software (Stata Corporation, College Station, Texas, USA) was used for statistical analyses. Combined results of hazard ratio of CVD mortality conducted by Random-effects model [10]. P < 0.10 and I2 < %50 were considered as heterogeneity detection among included studies. In order to find source of heterogeneity, subgroup analysis was conducted based on gender, whereas meta-regression analysis was conducted based on follow-up years, age of participants and HR of CVD mortality. Non-linear association was examined by modeling concentration level using restricted cubic splines [10]. The publication bias among included studies evaluated by Funnel plot, Begg's test, and Egger's regression test.

Literature search
A flow chart of included studies is shown in Fig. 1. In the primary search 611 records were identified, after excluding 74 duplicates studies and 479 irrelevant studies from title and abstract screening, 58 studies remained for full text screening. After reviewing the full text, 26 studies were excluded because they did not meet the inclusion criteria. Finally, 32 studies, containing 44 arms and 1,134, 073 participants, were included in the meta analysis.

Dose-response analysis
The pooled HR from the random-effects dose-response model of included studies showed a significant positive association between SUA and CVD mortality (r = 0.0709, p = 0.001) (Fig. 4). Figure 5 shows funnel plots of CVD mortality. There was publication bias among the studies (the Begg's p = 0.04 and Egger's test). Using the 'Trim and fill'

Discussion
The present meta-analysis of cohort studies revealed that there is a strong relationship between SUA levels and risk of CVD mortality. Moreover, the pooled HR from the random-effects dose-response model indicated that this positive association is stable when SUA is greater than 6 mg/dL. Although the exact mechanism of the relationship between SUA and CVD mortality risk has not yet been elucidated, previous studies have suggested some possible explanations.. An experimental study has reported that using an inhibitor of uricase can elevate blood pressure in rats by activating the renin-angiotensin system and inhibition of nitric oxide synthase [43]. Moreover, it has been shown that patients with hypertension and hyperuricemia had higher carotid intima media thickness in comparison to patients without hyperuricemia [44]. Furthermore, it has been reported that blood atherosclerotic platelets consist of a great amount of UA and higher levels of SUA can simply boost thrombus development which may lead to slow coronary flow (SCF) [45]. In contrast, UA may have some anti-proliferative influence on the endothelium or can damage the process of nitric oxide production [16].
It has been suggested that UA can significantly slow down coronary flow by promoting the calcification of coronary arteries [45,46]. In addition, higher levels of SUA may induce oxidative stress by oxidation of lowdensity lipoprotein cholesterol which may lead to SCF [45]. Another possible mechanism might be explained by the effects of hyperuricemia in induction of crystal shaping on vascular walls that impair endothelial and smooth muscle function leading to atherosclerosis by renin-angiotensin system activation [43,47]. Even, crystals of urate have several noxious effects that activate neutrophilsand macrophage cells to set proteases free and stimulate the coagulation cascade [48].
The main strength of this meta-analysis was the application of cohort studies containing over one million participants, and subsequent dose-response analysis. This study has some possible limitation. The heterogeneity in the study populations is the main limitation of this study. Moreover, the duration of follow-up in our included studies differed. Further, there is an insufficient number of randomized clinical trials to confirm the effects of decreasing SUA levels on CVD deaths; however, our results suggest higher levels of SUA is an independent risk factor associated with CVD mortality. Furthermore, although we searched the literature extensively, we did not explore grey or unpublished literature to limit the possibility of publication bias.

Conclusion
In conclusion, our analysis indicates a positive doseresponse association between SUA and CVD mortality risk; however, further clinical trials are needed to confirm these findings and to determine the possible causeeffect relationship.