Although the link between elevated serum uric acid (sUA) concentrations and the risk for atherosclerotic cardiovascular and cerebrovascular disease has long been observed, only recently have the pathophysiologic links become clearer 1 . Kanbay and colleagues 2 recently summarized the emerging data suggesting that hyperuricemia may cause not only atherosclerosis in the macrovascular beds such as the coronaries and the carotids but also microvascular damage in the renal vascular bed and may exacerbate vascular disease.

Almost all epidemiological studies performed in populations of higher-than-normal risk have shown a consistent association between sUA and coronary artery disease (CAD) 1 . Studies on the lower-than-normal-risk populations that have relatively few events need to have a very large sample size to be able to measure the magnitude of relative risks observed in the high-risk groups (relative risk of 1.5 to 2.5). In such a context, markers of subclinical atherosclerosis are important outcomes to examine. The detection of coronary artery calcification (CAC) by ultrafast computed tomography (CT) scanning is highly predictive of the presence of histopathologic atherosclerosis 3 , and the extent of calcification correlates well with plaque burden 4 . It is also an accurate (positive predictive value of 84% to 96%) measure of obstructive CAD compared with angiographic evaluation and is a useful tool to study subclinical CAD, especially in population settings 5 . Some argue that, in the setting of observational studies, CAC measurement may even be superior to other measures such as carotid intima-media thickness in predicting cardiovascular outcomes 6 .

The primary objective of this epidemiological study was to understand the relationship between sUA concentration and CAC in relatively young and healthy adults. If the hyperuricemia-CAD link is real, we can expect that the prevalence of CAC among those with higher sUA levels will be greater and that the extent of CAC will be directly proportional to the degree of hyperuricemia - a hypothesis that we tested here.

Of the 5,115 participants at baseline, 3,671 participated in the examination at year 15. Among these, 1,173 were excluded as they met the study exclusion criteria. Overall, there were 2,498 participants (1,211 men and 1,287 women) available for analyses. Table 1 shows the characteristics of these participants. Higher sUA concentrations were associated with greater prevalence of cardiovascular risk factors, metabolic syndrome, and high Framingham risk score (Table 1). Fewer than 20 participants had a self-reported history of probable or definite gout and these could not be independently verified. None of the study population was using urate-lowering medications such as allopurinol, probenecid, sulfinpyrazone, or losartan.

The mean ± standard deviation of sUA concentration was 345 ± 77 μmol/L (5.8 ± 1.3 mg/dL) for men and 238 ± 65 μmol/L (4.0 ± 1.1 mg/dL) for women (P < 0.001). sUA was distributed normally among men and women, whites, and African-Americans. The proportions of participants with sUA of greater than 416 μmol/L (7.0 mg/dL) were 8.4% (n = 211) overall, 16.6% (n = 201) among men, and 0.8% (n = 10) among women. The men had a significantly worse overall cardiovascular risk profile than women (Table 1). sUA concentrations were correlated with male gender (correlation coefficient of 0.6; P < 0.01) and Framingham risk score (0.52; P < 0.01) but were only modestly associated with other known cardiovascular risk factors (all correlation coefficients of less than 0.3; P < 0.05). C-reactive protein levels were not correlated with sUA (P = 0.20).

The majority (90.5%, n = 2,260) of participants had no detectable CAC. Overall, 9.5% (n = 238) of participants had an Agatston score of greater than 0, 6.3% (n = 158) had an Agatston score of greater than 10, and 1.4% (n = 34) had an Agatston score of greater than 100. As expected in a cohort free of clinical CAD, relatively few participants had an Agatston score of greater than 400 (n = 4). Among those with any CAC, mean scores were higher in men than in women (75.0 versus 60.8; Wilcoxon rank sum test P = 0.77) and in whites than in African-Americans (71.6 versus 70.0; P = 0.19), but these findings were not statistically significant.

At each quartile of sUA, men had a greater prevalence of CAC than women (Figure 1). However, the prevalence of CAC increased with increasing sUA in both genders, and the highest quartile had almost two times the prevalence compared with the lowest (odds ratio [OR] 1.87, CI 1.19 to 2.93) (Table 2).

In the bivariate logistic regressions in which presence or absence of CAC was the dependent variable, the highest quartile of sUA concentrations (greater than 393 μmol/L [6.6 mg/dL] for men and greater than 274 μmol/L [4.6 mg/dL] for women) had an OR of greater than 2.0 among both men and women, regardless of the Agatston score cutoff used to define CAC (Table 3).

We developed parallel multivariable logistic regression models that included all of the risk factors of interest (age, gender, race, high- and low-density lipoproteins, triglycerides, smoking, blood pressure class, presence of metabolic syndrome, C-reactive protein, waist circumference, alcohol use, creatinine, and serum albumin concentration) in the model. Data were pooled for men and women, and gender-specific stratification for quartiles of sUA was used after we established that there was no statistically significant interaction between gender and the correlation between sUA concentration and CAC (Table 2). Although these models differed in the definition of subclinical atherosclerosis and in the stratification strategy for sUA, all models showed that the highest quartile of sUA concentration was associated with significantly higher risk for subclinical atherosclerosis. In both multivariable regression models, each unit increase in sUA concentration was associated with an OR of 1.23 (CI 1.09 to 1.39) for subclinical atherosclerosis. The findings were replicated in backward stepwise selection models that eliminated those factors that were not significant individually as well as those in which an alternate stratification strategy for sUA was used (data not shown).

The last set of analyses focused on the association between sUA concentration and the severity of CAC. These analyses included only those subjects who had Agatston score of greater than zero (n = 238). Although the Agatston scores were higher among those with higher sUA concentrations (Figure 2), bivariate correlation analyses showed that the association was not strong (correlation coefficient of 0.13). However, in multivariable OLS regression models, each unit increase in sUA concentration was associated with a significant increase in the log-transformed Agatston score (beta coefficient of 0.288, 95% CI 0.078 to 0.498; P = 0.008; R² = 0.197%). In other words, there was an approximately 22% increase in Agatston score for each unit increase in sUA. When examined separately for each gender, this association persisted for men (beta coefficient of 0.300, CI 0.078 to 0.522) and women (beta coefficient of 0.318, CI -0.502 to 1.138) but was statistically significant only in the former (P = 0.009).

The association between hyperuricemia and the presence of subclinical atherosclerosis has not previously been studied in a cohort of young adults with no risk factors for CAD. Our study found a direct correlation between the prevalence and severity of CAC and sUA concentration in both men and women. This supports the hypothesis that uric acid may be involved in the pathologic process of atherosclerosis independently of conventional risk factors.

Uric acid is a ubiquitous antioxidant in the blood 18 . Abnormally high serum concentrations of uric acid indicate oxidative stress, endothelial dysfunction, and slow coronary artery blood flow 19 20 . Elevated sUA concentration signifies a milieu with high oxidative stress and potentially indicates a vascular pathologic process such as atherosclerosis 21 . An association between hyperuricemia and CAC has been observed in previous studies involving patients with underlying risk factors for CAD such as type 1 diabetes, longstanding hypertension, or metabolic syndrome 22 23 24 25 . Importantly, a cross-sectional analysis of 443 individuals with type 1 diabetes suggested that the chances of progressive CAC were proportional to the magnitude of sUA concentration 22 . Another study involved older age groups compared with our cohort but also found that the correlation between uric acid and CAC was evident among men and women and in similar magnitude 13 . In the INSIGHT (International Nifedipine Study Intervention as Goal for Hypertension Therapy) study, in which CAC was measured in hypertensive patients who were older than 55 years of age and who had at least one more major cardiovascular risk factor, those with a total coronary calcium score (TCS) of greater than zero had a slightly higher sUA concentration compared with those with a TCS of zero (333 ± 89 versus 315 ± 83 μmol/L [5.6 ± 1.5 versus 5.3 ± 1.4 mg/dL]; P = 0.03) 23 . However, other studies, including the GENOA (Genetic Epidemiology Network of Arteriopathy) study on sibships with at least two members with diagnosed hypertension, did not show an association between uric acid concentration and the presence or severity of CAC after adjustment for conventional risk factors 25 26 27 . In addition, the National Heart, Lung and Blood Institute (NHLBI) Family Heart Study did not find a significant relationship between hyperuricemia and CAC in either gender 28 . Among those who underwent coronary angiography for suspected CAD, sUA concentration of greater than 416 μmol/L (7.0 mg/dL) was associated with stable plaques without evidence of remodeling. The authors interpret this as suggesting that uric acid is a marker of atherosclerosis rather than a pathogenic mediator 13 29 .

Coronary atherosclerosis is less likely to be associated with calcification among women compared with coronary atherosclerosis with a similar degree of lumen narrowing in men 30 . In the CARDIA study cohort, the prevalences of CAC were approximately 15% among men and approximately 5% among women overall 14 . This could be because younger women may be 'resistant' to atheroma growth 31 .

Gender might be an important effect modifier in the association between hyperuricemia and CAC because of differences in (a) the distribution of sUA and (b) the prevalence of CAC. Iribarren and colleagues 32 analyzed data from the Atherosclerosis Risk in Communities (ARIC) study and concluded that an association between sUA and cardiovascular risk is evident in men but not women. In contrast, a similar study by Ishizaka and colleagues 33 reported that gender was not a factor. Since the prevalences of hyperuricemia and CAC are both lower among women, a greater sample size would be needed to detect a given effect size of hyperuricemia-CAC association. We defined quartiles separately for men and women prior to pooling. Statistical tests of gender-sUA interaction were not significant in our data. In gender-specific analyses, the direction and magnitude of risk among women were similar to those among men; however, the standard errors were wide because of the lower power for precise estimates.

The major strength of our community-based study is the generalizability of our results to young adults, including men, women, African-Americans, and whites. All of the studies described earlier were performed among patients with greater-than-normal cardiovascular risk, such as those with hypertension, diabetes, metabolic syndrome, psoriasis, or renal disease 13 21 23 26 27 28 29 .

The primary limitation of this study was the cross-sectional nature of data analysis. Survivor bias can affect cross-sectional data analyses in that those with more severe disease die prior to the time point of analysis; however, this is not a major consideration in the CARDIA study as the main cause of mortality in the first 16 years was non-cardiovascular in the vast majority of patients (117/127 deaths out of 5,115 enrollees). Our future studies will examine the rate of progression of CAC over time among patients with hyperuricemia or gout or both. Gouty arthritis has been associated with CAD among middle-aged men 34 , but our study had too few participants with gout (n <20) to allow a formal analysis. A larger number of women would have enabled separate analyses with respect to use of exogenous hormones and menstrual status. Owing to the design of our study, there was relatively little heterogeneity with respect to age (approximately 12 years), precluding an analysis of impact of age on the hyperuricemia-CAC association. However, in our analyses, age was not significantly associated with sUA concentration. sUA concentration is known to vary with time of day and recent dietary intake and possibly with physical exertion, introducing 'random noise' in sUA data. Unless sUA increases due to such variables can be shown to occur preferentially among those with higher CAC scores, this issue cannot explain our findings. As in all other epidemiological studies, unmeasured covariates could have caused residual confounding in our study as well.

We have shown for the first time that sUA is associated with the presence and severity of CAC in young healthy adults, implicating a potential role of uric acid in the pathogenesis of subclinical atherosclerosis. Our data are consistent with the growing body of literature that implicates the vascular injury associated with hyperuricemia - both macrovascular and microvascular 2 22 35 .

CAC: coronary artery calcification; CAD: coronary artery disease; CARDIA: Coronary Artery Risk Development in Young Adults; CI: confidence interval; CT: computed tomography; EBCT: electron beam computerized tomography; OLS: ordinary least square; OR: odds ratio; sUA: serum uric acid; TCS: total coronary calcium score.

EK has consultant/advisor/grant recipient relationships with Takeda Pharmaceuticals International, Inc. (Deerfield, IL, USA). He has been a shareholder of Savient Pharmaceuticals, Inc. (East Brunswick, NJ, USA) and currently holds common stock in that company. He is an investigator for a clinical trial performed by Ardea Biosciences (San Diego, CA, USA). He serves on advisory boards for Takeda Pharmaceuticals International, Inc., URL Pharma (Philadelphia, PA, USA), and UCB (Brussels, Belgium). BJP and OD are employees of Takeda Pharmaceuticals International, Inc. LC declares that she has no competing interests.

EK conceived of the manuscript idea, designed the analysis plan, performed statistical analysis, interpreted the results, and wrote the first draft of the manuscript with assistance from all other authors. He has possession of raw data sets and takes responsibility for the integrity of the data and the accuracy of the data analysis. Takeda Pharmaceuticals International, Inc. did not have access to the raw data, and Takeda Pharmaceuticals International, Inc. authors (BJP and OD) contributed primarily to refinement of study design, interpretation of data, and editing and revising of the initial drafts. All authors read and approved the final manuscript.