Introduction

Urate is the final product of purine catabolism in humans. Elevated serum urate (sUA) can result from excessive dietary intake of purine-rich foods, underexcretion of urate [1,2], or urate overproduction due to enhanced production of xanthine oxidoreductase, an enzyme that catalyzes urate generation [3]. High fructose intake may induce urate generation [4]. Many epidemiological studies have associated elevated sUA concentration with higher risk of cardiovascular disease (CVD) and its risk factors [2]. Meta-analyses [5–7] found marginal to moderate associations of sUA with coronary heart disease (CHD) events, stroke incidence and mortality, and incident heart failure, independent of traditional CVD risk factors. Most existing evidence was in older adults already at higher CVD risk (i.e., greater disorders in classic CVD risk factors, such as body mass index (BMI)) [8],[9]. Elevated sUA concentration appeared to be a circulating marker for classic CVD risk factors in some studies [2], while others inferred a possible causal association between sUA and CVD events [10,11].

In limited evidence concerning younger and healthy populations, a cross-sectional report in 30–45 year olds found that sUA and early atherosclerosis became non-significantly related with adjustment for BMI [12]. We recently reported that repeatedly measured sUA predicted future subclinical atherosclerosis in a cohort of generally healthy people enrolled at ages 18–30 years in the Coronary Artery Risk Development in Young Adults (CARDIA) study [13].We found that sUA through middle age was associated with metabolic factors such as BMI and, that its association with subclinical atherosclerosis when sUA was measured during young adulthood was attenuated by adjustment for BMI. However, the same association with subclinical atherosclerosis when sUA was measured in early middle-age had strengthened, independent of BMI.

In this further longitudinal analysis in CARDIA, we therefore hypothesized that higher sUA concentrations predict incident clinical CVD events. In parallel with our findings for subclinical atherosclerosis [13], we expected that sUA measured in early middle-age (i.e. 4^th to 5^th decade) in CARDIA would predict CVD events independently of other metabolic factors.

Methods

Statistical Analysis

All analyses were conducted using SAS version 9.3 (SAS Institute, Inc, Cary, NC). All P-values were two-sided and statistical significance was set at P<0.05. sUA concentration was assessed in 5049, 3870, 3605 and 3147 of the 5115 participants at years-0, 10, 15 and 20, respectively. However, the current study excluded year-20 sUA concentration as an exposure, considering the limited number of CVD events occurring from year 20 to 27. Serial analyses were conducted to predict CVD outcomes using sUA at year-0, year-10, and year-15 in separate models as study exposures (not adjusting for sUA at other exam years simultaneously in the model).

We excluded subjects who had a low glomerular filtration rate (≤60 mL/min per 1.73 m²) at any of years-0, 10 or 15 (n = 45), leaving a maximum of 5070 participants for any of the following analyses. Compared to the participants who were included, the 45 who were excluded were slightly older and drank more alcohol. The metabolic syndrome prevalence at baseline was also higher among those excluded. For each part of our analysis, anyone with prevalent CVD or missing covariate data at the year of sUA measurement was further excluded from the 5070 participants. Therefore, the sample size varied across sUA exposures and covariates at different exam years, as reported for each specific analysis in the tables.

The concentration of sUA was modeled in three ways: (1) as a continuous variable; (2) as gender-specific tertiles that were pooled to incorporate the potential gender difference in sUA concentrations [25]; (3) as hyperuricaemia (i.e. sUA ≥6.8 mg/dL [26,27]), the concentration above which sUA becomes insoluble and starts depositing as monosodium urate in tissues, which is the precursor of gout. Participant characteristics (i.e. in means or proportions) were described by gender-specific sUA tertiles.

Multiple Cox proportional hazards regression models were used to examine the longitudinal association between sUA and CVD. For each regression model, covariates including sex, race, age, clinic, education, total energy, alcohol, and protein intakes at the year of sUA measurement were adjusted. Additionally, we adjusted for lifestyle factors (e.g. smoking status), BMI, metabolic syndrome (yes vs. no), insulin resistance, the ratio of total cholesterol and HDL-cholesterol, blood pressure, and waist circumference as the classical CVD risk factors, plus the usage of anti-hypertension medication, the usage of diuretics, and glomerular filtration rate (as a continuous variable).

No significant sex- or race-interaction was found. Because both sUA concentrations [25] and CVD risk [28] may differ between men and women, we also report a sensitivity analysis stratified by gender.

Results

Participant characteristics

Mean sUA concentrations increased over time in both genders. The mean (SD) sUA concentrations at years-0, 10 and 15 in men were 6.16 (1.15), 6.35 (1.24), and 6.42 (1.27) mg/dL, respectively; and in women were 4.46 (0.97), 4.63 (1.00), and 4.75 (1.07) mg/dL, respectively. Metabolic profiles substantially deteriorated over time (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Unadjusted participant characteristics (mean±SD or percent) of participants across exam years.

https://doi.org/10.1371/journal.pone.0138067.t001

Association of sUA concentration with incident CVD

Over the 27-year follow-up, 164, 79, 40, and 40 participants developed any fatal or non-fatal CVD, CHD, CHF or stroke, respectively. Adjusting for baseline age, sex, race, clinic, education, physical activity, smoking and intakes of alcohol, protein and total energy (Table 2), baseline sUA concentration was positively associated with the incident CVD by year 27 (hazard ratio (HR) per mg/dL of sUA = 1.21; 95% confidence interval (CI) = 1.05, 1.39). Compared to participants with sUA concentration in the lowest tertile at year-0, those in the highest tertile had a greater risk (HR = 1.58; 95%CI = 1.08, 2.31) for any fatal or non-fatal CVD by year 27. This positive association was substantially attenuated with further adjustment for baseline BMI, systolic and diastolic blood pressure, anti-hypertension medication use (excluding those taking diuretics), diuretics use, and glomerular filtration rate (HR = 1.09, P = 0.24). Hyperuricemia at baseline related positively to incident CVD by year 27, with findings attenuated by adjustment.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Longitudinal association between sUA and the incidence of any fatal or nonfatal CVD endpoints by year 27.

https://doi.org/10.1371/journal.pone.0138067.t002

sUA at both year-10 and year-15 had larger estimated magnitude of association with incident CVD by year 27 (both HR = 1.27, P≤0.001) than was found for sUA at year-0. Prediction from sUA at year-10 (based on 133 incident CVD events) was attenuated in the full multivariable model, but remained significant with additional adjustment for individual year-10 BMI, metabolic syndrome, total cholesterol/HDL-C, HOMA-IR, or blood pressure (data not shown). The association between sUA at year-15 and incident CVD, adjusting for demographic and lifestyle factors, was of a comparable magnitude to that at year-10; and remained significant but moderate in the full model (HR = 1.19; 1.02, 1.40; P = 0.03). Being hyperuricemic at year-15 significantly predicted CVD development, even after adjustment for metabolic factors.

In sensitivity analysis, the estimated association between sUA and CVD was nominally stronger in men than in women (S1 Table); p<0.05 for sex-interaction with sUA only at year-15. Like associations with incident CVD, associations with incident CHD, HF, and stroke with sUA concentrations (each examined separately) were generally positive and estimates strengthened when predicting from at years-10 and 15 vs. prediction from year-0 (S2, S3 and S4 Tables). However, associations for separate outcome were not statistically significant.

Discussion

In this young adult cohort that was generally healthy at baseline, a higher sUA concentration was associated with a moderate and significant risk of developing any fatal or non-fatal CVD over a 27-year follow-up. The prediction of CVD based on sUA concentration in young adults was largely attributable to the correlation of sUA with classical CVD risk factors, but appeared to strengthen for sUA measured in early middle age. The sUA concentrations were highly correlated across examinations and those with elevated sUA concentrations had worse metabolic profiles, as previously reported [13]. Coupled with the deterioration in multiple classic CVD risk factors with aging, sUA appeared to be involved as a part of a complex web of changing metabolic variables over early adult life in its prediction of CVD endpoints.

Our present findings were generally consistent with our report on the positive association between sUA concentration and the risk of subclinical atherosclerosis (comprising coronary artery calcified plaque and maximum common carotid intima-media thickness) [13]. In both studies, we found that the associations of sUA with subclinical atherosclerosis and CVD were more substantial in men than in women, although the sex-interaction was generally not statistically significant in the present analyses. The prediction of both subclinical atherosclerosis and CVD endpoints strengthened using sUA concentration in early middle age compared to sUA in young adulthood. Of note, some previous studies reported that the association between sUA and CVD outcomes (or intermediate outcomes) was stronger in women than in men [5], while others showed the opposite [29–32]. While this potential gender interaction is interesting, it is not clearly significant in our study and deserves further study.

Higher sUA concentration usually clustered with several other metabolic abnormalities such as high blood pressure, hyperinsulinemia, and dyslipidemia [2,33]; sUA also predicted future components of metabolic syndrome [34]. The role of elevated sUA concentration in CVD development has long been debated, particularly whether its predictive capacity added value to the classic CVD risk factors. A few large prospective cohort studies concluded that there was no independent association of sUA with CHD [35,36]; whereas, others found significant association between sUA and CVD with full adjustment for classic CVD risk factors [11,37]. Using Mendelian Randomization, some studies failed to find any evidence showing a causal relationship between sUA and CVD [38,39], while one study in men of average age 70 years and followed for a median of 10 years found that a genetic risk that correlated with sUA was associated with cardiovascular and sudden cardiac death [40]. A few meta-analyses of older adults have reported an attenuated, but significant small to modest association between sUA and CVD outcomes after adjustment for traditional CVD risk factors [5–7]. Older adults generally have metabolic profiles that are deteriorated compared to our CARDIA sample. Therefore, the independent role of sUA found in older adults supports our observation in CARDIA that sUA significantly predicts CVD outcomes independent of traditional CVD risk factor profiles, starting in early middle age. No previous studies have examined how the prediction of CVD from sUA progressed as the sample aged.

Knowles and Reaven recently suggested that sUA per se was a biomarker of insulin resistance in CVD development [41]. Nevertheless, we found that adjusting for insulin resistance or other individual metabolic abnormalities did not completely explain the significant positive association of elevated sUA measured in early middle age with incident clinical CVD risk. Although measurement error in insulin resistance cannot be excluded, as we previously summarized [13], several mechanisms may be involved in the sUA-CVD association, including elevated sUA (1) being a marker of xanthine oxidoreductase activity and oxidative stress related to atherosclerosis; (2) mediating proinflammatory pathways; (3) inducing endothelial dysfunction; (4) activating the renin–angiotensin system; (5) resulting from renal dysfunction leading to reduced sUA clearance (also suggested by Knowles and Reaven [41]). Gaffo et al. found in CARDIA that higher sUA, even for sUA<6.8mg/dL, independently predicted subsequent incident hypertension [26]. Given similar findings elsewhere [42], a 2-stage urate-mediated hypertension mechanism was proposed [43]. It is suspected that increased sUA activates the renin-angiotensin system and reduces the bioavailability of circulating plasma nitrates, resulting in elevated blood pressure. During this first stage, the metabolic deterioration may still be reversible by decreasing sUA concentration. In the second stage, vascular wall thickening and smooth muscle proliferation is thought to occur due to increased oxidative stress; the resulting hypertension would be less responsive to serum urate lowering and lead to adverse cardiovascular outcomes [42,43]. This hypothesis is supported by our current findings because sUA-CVD association in early adulthood was eliminated after adjustment for blood pressure, but it remained significant in middle-age. Thus, we suspect that elevated sUA concentration shares much but not all common pathophysiology with each individual metabolic disorder in CVD development. Although we cannot determine whether reducing sUA concentration would yield clinical benefit, it might be important to track sUA concentration in clinical practice and to conduct further research to better understand sUA-related pathways. In other words, sUA influence in CVD development in early adulthood might still not be evident; whereas as people age, sUA rises, and vascular remodeling is established, the impact of sUA or its precursors may become more evident.

The positive associations of sUA with CHD, stroke, and HF events generally did not achieve statistical significance, consistent with fewer events of these separate outcomes. sUA may have different pathophysiologies between separate CVD outcomes. In a recent systematic review, elevated sUA was related to increased risk of incident HF as well as CVD and all-cause mortality in HF patients; the majority of studies involved middle-aged and elderly adults and the authors cautioned that the culprit for HF development might be unregulated xanthine oxidase rather than sUA itself [7].

Study limitations include that generalizations to older populations and drawing causal inferences should be cautious. Because this cohort has just entered late middle-age and our analyses were based on only 164 incident CVD cases, our statistical power was limited, especially in the analyses using year 15 sUA as an exposure, in examining CVD subtypes (i.e. CHD, stroke and HF), and in studying potential gender-interaction.

Study strengths include that we followed young, healthy adults for 27 years from young adulthood to middle-age, while most of the previous evidence focused on older adults. The current longitudinal design with repeated measurements of both exposures and outcomes provided new evidence to the debate on sUA and future CVD. We carefully tested the potential impacts of classic CVD risk factors, either individually or as a cluster, on the sUA and CVD association.

In conclusion, we observed a significantly positive correlation between sUA concentration and the development of CVD outcomes among young to middle-aged adults during a 27-year follow-up. CVD event prediction based on sUA appeared to strengthen with age, but the magnitude of the prediction was only small to moderate. During young adulthood, sUA may be responsive to individual metabolic abnormalities, while by middle-age adjusting for individual metabolic disorders did not seem to fully diminish the sUA and CVD association. Our study suggested that monitoring sUA concentrations may be of clinical importance to CVD prevention and management. Because the CARDIA study is still on-going and we expect to obtain additional data on this adult cohort as it ages, it would be interesting to extend our analysis with future follow-up to see whether the role of sUA as a predictor continues to evolve as the cohort ages.

Supporting Information