Uric acid (UA) is the end-product of the metabolic breakdown of purine nucleotides and is excreted in urine. Increased levels of UA because of decreased renal excretion are the primary cause of arthritis urica. Arthritis urica patients, including individuals with asymptomatic hyperuricemia, have an increased risk of coronary artery disease (CAD); however, the mechanisms behind this association are not yet fully elucidated.
UA levels have been found to be associated with both adverse events and mortality in patients with acute myocardial infarction (MI) 1,2 as well as with the severity and mortality of acute CAD 3,4. Moreover, elevated levels of UA have been associated with an increased risk of subsequent clinical events 5 and mortality in patients with stable CAD 6. In contrast to these observations, both Cheong et al. 7 and Nossent et al.8 did not find any relationship between either all-cause or cardiovascular mortality and UA levels in the general population.
Also, the role of UA as a risk factor for coronary artery calcification (CAC) remains controversial as some studies have reported a significant association between UA and CAC 9–12, whereas others did not find any relation between UA and CAC 13.
With the purpose of further illuminating the role of UA in relation to stable and acute clinical manifestations of CAD, we aimed to investigate (a) whether UA levels correlate with the presence and severity of CAC in asymptomatic middle-aged individuals from the general population and (b) whether UA levels are generally elevated or differ between subgroups of chest pain patients with suspected or definite MI.
Patients and methods
DanRisk population (the Danish risk score study)
In 2009, a random sample of 1825 asymptomatic individuals from the general population, 50 or 60 years old, was invited to participate in the DanRisk screening study 14. A total of 1257 (69%) patients accepted the invitation. Exclusion criteria were as follows: known CAD (n=32), atrial fibrillation (n=9), heart valve disease (n=1), and diabetes (n=59). Furthermore, patients in whom noncontrast computed tomography (CT) scan was not performed (n=4) or in whom biobank material was insufficient (n=136) were excluded. Thus, a total of 1016 asymptomatic individuals were eligible for the final analysis. To assess the presence and severity of CAC, all participants underwent a noncontrast cardiac CT scan. An Agatston score was calculated by summing up the scores from each of the coronary artery foci 15. Experienced cardiologists measured the Agatston score, and the participants were categorized into five groups: score 0 U, score 1–9 U, score 10–99 U, score 100–399 U, and score >400 U. On the day of the CT scan, nonfasting blood samples were collected, centrifuged, and serum was stored at −80°C for future analysis.
DEF-AMI population (consequences of the universal 2007 definition of acute myocardial infarction studied prospectively in a Danish consecutive hospital population)
From January 2010 to January 2011, a total of 3762 consecutive patients with a suspected acute MI, who had serial cardiac troponin I values (TnI) measured during admission to Odense University Hospital, were included within 24 h after the first TnI sampling; all patients were subjected to an assessment of supplementary history by dedicated personnel, with a special focus on symptoms at admission, clinical characteristics, and co-morbidity. From 2390 patients who provided informed consent, a supplementary nonfasting blood sample was collected and stored at −80°C in the hospital’s biobank. Depending on the results of serial TnI measurements, patients were categorized as having elevated TnI concentrations (>0.03 µg/l) or nonelevated TnI concentrations (≤0.03 µg/l). Subsequently, patients with elevated TnI concentrations were classified as having an acute MI or as not fulfilling the MI diagnosis according to the criteria of the universal definition 16. Patients with acute MI were subgrouped as having a type 1 or a type 2 MI. Type 1 MI results from spontaneous myocardial ischemia caused by coronary plaque erosion and/or rupture, fissuring, or dissection, whereas type 2 MI occurs secondary to myocardial ischemia caused by increased oxygen demand and/or decreased oxygen supply 16. Patients classified as type 1 or type 2 MI all had at least one TnI value above the 99th percentile of the upper reference limit (URL) and showed a rise and/or fall pattern as required according to the universal definition 16. Patients with elevated cardiac TnI concentrations but not fulfilling the MI criteria 16 were categorized as having myocardial injury as described previously 17. Case ascertainment and individual evaluation of all patients was carried out by an experienced adjudication committee. Considering patients with myocardial injury, the following classification was used: first, patients with clinical conditions known to cause myocardial ischemia (e.g. arrhythmias, heart failure, aortic valve stenosis, chronic CAD, hypertension) who did not fulfill the biomarker criteria of the MI diagnosis were grouped. Indeed, these patients did have one or more TnI values above the 99th percentile of the URL, but the values were stable and did not show any rise and/or fall pattern as required according to the universal definition 16. In the present study, these patients are classified as having ischemic myocardial injury. Second, patients with myocardial injury related to nonischemic cardiac conditions (e.g. major cardiac surgery, myocarditis, radiofrequency ablation) were classified as having Nonischemic myocardial injury. Third, patients with myocardial injury associated with systemic clinical conditions (e.g. infections, chronic kidney disease, pulmonary embolism, cerebrovascular events) were classified as having systemic myocardial injury. Finally, for comparison, patients presenting with nonelevated TnI values were included (Fig. 1). A detailed description of the DEF-AMI study, including classification of patient subgroups, has been published previously 17–19.
In this substudy, we included a total of 772 patients from the biobank. The patients were all type 1 MI patients with blood samples available (n=177), all type 2 MI patients (n=36), and all patients with myocardial injury (n=196) (68 with ischemic myocardial injury, 51 with nonischemic myocardial injury, and 77 with systemic myocardial injury). Finally, from the group with nonelevated TnI values, a random selection of patients (n=363), with a distribution of men versus women of ~2 : 1 as in the other subgroups, was included. To eliminate kidney disease as a cause of elevated UA values, only patients with an enzymatic creatinine value within the reference interval (≤90 µmol/l for women and ≤105 µmol/l for men) were included. As a consequence, 191 patients were excluded before the statistical analyses because of creatinine results above the URL and these patients are not included in the above-mentioned numbers.
Both the DanRisk and the DEF-AMI studies were approved by the Regional Scientific Ethical Committee for Southern Denmark (S-20080140 and S-20090082, respectively) and were carried out in accordance with the Declaration of Helsinki. Written informed consent was obtained from each participant.
Before biochemical analysis, serum samples were thawed, mixed, and centrifuged for 4 min at 2400g. Blood samples were frozen and thawed maximally three times before analysis of UA and creatinine in this substudy. The effect of three freeze/thaw cycles was investigated and on average UA and creatinine showed deviations of −0.6 to −0.9% after the three cycles, which is considered acceptable compared with the analytical coefficients of variation (data not shown).
UA and enzymatic creatinine were analyzed using a DxC 800 (Beckman Coulter Inc., Brea, California, USA) and the DxC 800 assay kits (Beckman Coulter Inc.). The analytical coefficients of variation for UA and enzymatic creatinine were 1.6–5.8 and 1.4–5.8%, respectively.
TnI values were analyzed at the time of hospitalization on an Architect c16000 (Abbott Diagnostics, Abbott Park, Illinois, USA). TnI has an analytical coefficient of variation less than 20% at 0.02 µg/l and less than 10% at 0.028 µg/l, with an upper reference limit of the 99th percentile of 0.028 µg/l. A concentration of more than 0.03 µg/l is used as the decision limit for the diagnosis of MI. C-reactive protein (CRP) was analyzed on an Architect c16000 (Abbott Diagnostics), with analytical coefficients of variation of 2.8–5.5%.
The aim of this study was to investigate the possible association between UA levels and (i) varying degrees of CAC in asymptomatic individuals and (ii) different groups of patients hospitalized with chest pain.
Descriptive statistics and bivariate analyses
Descriptive statistics were assessed according to data type and appropriateness of the normality assumption (continuous data: mean and SD vs. median and range; categorical data: frequencies and respective percentages). Bivariate correlation between UA (mmol/l) and CAC as measured by the Agatston score was visualized by a scatter plot, and the distributions of UA (mmol/l) across calcification categories (Agatston score=0 vs. 1–9 vs. 10–99 vs. 100–399 vs. >400) as well as heart disease groups (type 1 MI vs. type 2 MI vs. ischemic myocardial injury vs. nonischemic myocardial injury vs. systemic myocardial injury vs. nonelevated TnI) were shown by means of box plots. The potential effect of different calcification categories and DEF-AMI patient subgroups on UA (mmol/l) was explored by one-way analysis-of-variance (ANOVA) including Bonferroni-adjusted multiple comparison tests for the analysis of intergroup differences 20.
Interplay of uric acid levels with the presence and severity of coronary artery calcification in asymptomatic middle-aged individuals
The influence of UA (mmol/l) in addition to the usual risk factors for CAC as measured by the Agatston score was investigated in two ways: first, the five above-mentioned categories of the Agatston score were used in ordinal logistic regression using the known risk factors age, sex, hypertension (no vs. newly diagnosed vs. known), hyperlipidemia, and smoking (never vs. active vs. former) as further covariates. The results were presented as generalized odds ratios and respective 95% confidence intervals (CI). Second, the quantitative measurement of the Agatston score was used as a dependent variable, considering its basic nature of being a count process, and zero-inflated negative binomial regression 21 was applied, again using the known risk factors as covariates. The results were shown as incidence rate ratios and respective 95% CIs. Zero-inflated negative binomial regression accounts for the excessive number of zero values of the Agatston score and makes use of as much information of calcification as possible, whereas the categorization of quantitative measurements into few categories is debatable 22–24. Model comparison with and without UA (mmol/l) was performed using a Likelihood ratio test. The applied zero-inflated negative binomial model was also contrasted with the zero-inflated Poisson model as well as the negative binomial model (i.e. without zero-inflation) using Likelihood ratio tests and Vuong’s test 25, respectively.
Role of uric acid levels in subgroups of chest pain patients with suspected or definite myocardial infarction
A multinomial logistic model was fitted for group affiliation using nonelevated TnI as a reference category, in which both known risk factors and clinical history were used as explanatory variables as follows: creatinine, CRP, age, sex, smoking (never vs. current vs. former vs. unknown), family history (none vs. yes vs. unknown), diabetes mellitus, hypertension, hyperlipidemia, previous MI, previous coronary artery bypass graft surgery, previous percutaneous coronary intervention, heart failure, previous stroke, peripheral artery disease, known kidney disease, and chronic obstructive pulmonary disease (COPD). The results were shown as relative risk ratios with respective 95% CIs. Moreover, binary logistic regression was applied on the group affiliations type 1 MI and type 2 MI as well as type 1 MI and ischemic myocardial injury using the former in both cases as the reference category and the above-mentioned covariates as explanatory variables. The results were presented as odds ratios and respective 95% CIs. For both the multinomial and the binary logistic regression model, the original UA (mmol/l) concentrations were multiplied by 10 to facilitate interpretation of 0.1 U changes instead of the usual 1 U changes.
The level of significance was 5%. All analyses were carried out using Stata/MP 15.0 (StataCorp. LP, College Station, Texas, USA).
DanRisk participants had a median age of 59.4 (49.4–61.1) years, and 54% were women. The demographics and clinical variables of the DanRisk population are presented in Supplementary Table (Supplemental digital content 1, http://links.lww.com/MCA/A177, which contains the demographics and clinical variables of the Danrisk population). The correlation between UA and CAC as measured by the Agatston score is visualized by a scatter plot (Supplementary Figure, Supplemental digital content 2, http://links.lww.com/MCA/A177, which shows the scatter plot of UA vs. CAC) and the distribution of UA across CAC categories (Agatston score) is shown by a box plot (Fig. 2). Using one-way ANOVA, UA was shown to be associated significantly with increasing Agatston score (Supplementary Table, Supplemental digital content 3, http://links.lww.com/MCA/A177, which shows the results of one-way ANOVA). However, this effect was abolished when adjusting for other risk factors. Ordinal logistic regression showed limited additive value of UA in addition to a model with usual risk factors (age, sex, smoking, hypertension, and hypercholesterolemia), P=0.08 (Supplementary Table, Supplemental digital content 4, http://links.lww.com/MCA/A177, which shows the results of ordinal logistic regression), as did zero-inflated negative binomial regression, P=0.95 (Supplementary Table, Supplemental digital content 5, http://links.lww.com/MCA/A177, which shows the results of zero-inflated negative binomial regression).
Patients with suspected or definite myocardial infarction
The demographics and clinical variables of the DEF-AMI population according to the different subgroups are presented in Table 1. Further, in Fig. 3, the distribution of UA levels in the different patient subgroups is illustrated using a box plot. One-way ANOVA showed multiple significant differences in UA levels between the different DEF-AMI subgroups (Supplementary Table, Supplemental digital content 6, http://links.lww.com/MCA/A177, which shows the results of the one-way ANOVA). Altogether, patients with type 2 MI and patients with ischemic myocardial injury showed significantly elevated UA levels, compared with type 1 MI, nonischemic myocardial injury, systemic myocardial injury, and nonelevated TnI patient subgroups. Using multinomial logistic regression, patients with nonelevated TnI were shown to have significantly lower UA levels compared with type 2 MI patients (P=0.02) and those with ischemic myocardial injury (P=0.01) (Table 2). Binary logistic regression showed significant differences between type 1 MI and type 2 MI, P=0.002, as well as between type 1 MI and ischemic myocardial injury, P=0.001 (Table 3), with the type 2 MI and the ischemic myocardial injury subgroups having higher levels of UA than the type 1 MI subgroup.
Earlier studies have suggested an association between UA and CAC severity in asymptomatic individuals 9,10,26,27, as well as between UA and both acute and stable CAD 1–6. In the present study, both asymptomatic individuals without known CAD and patients with suspected or definite MI were included. Using this design, the levels of UA could be measured in a broad spectrum of individuals ranging from the clinically stable to the clinically unstable situation, including patients with different subtypes of MI, and patients in whom the MI diagnosis could not be confirmed, despite elevated TnI values.
We did not find any association between the levels of UA and the presence or severity of CAC in asymptomatic middle-aged individuals. In hospitalized patients with chest pains, we found significant differences in UA levels between type 1 and type 2 MI patients. Thus, patients with type 2 MI had higher UA levels than those with type 1 MI. Furthermore, we showed that also in patients with TnI elevations, who did not fulfill the diagnostic MI criteria, the UA levels varied significantly depending on the associated clinical conditions.
Uric acid in asymptomatic individuals
CAC is an established predictor of cardiac events 28 and several studies have suggested an association between UA and CAC severity in asymptomatic individuals 9,10,26,27. However, these findings might have been caused by insufficient statistics. Thus, in some of these earlier studies, Agatston scores were merely divided into the presence or absence of CAC 9,10,27, as opposed to clinically relevant groups, or the highest tertile or quartile of UA was only compared with the lowest in terms of the Agatston score 10,26,27, which is purely data driven and not clinically relevant. In contrast, Neogi et al. found no association between UA and CAC when testing for a trend over quartiles of UA 13. In the present study, we also found no association between UA and CAC severity after adjusting for common risk factors, and in addition, there was only limited value when adding UA from our asymptomatic middle-aged individuals to a risk model predicting atherosclerosis. This observation is consistent with the findings obtained by Kavousi et al.29, who also did not find any improved risk prediction for CAD in a middle-aged population, when adding UA to the Framingham Cardiovascular Risk Score, which has otherwise been proposed to improve the prediction of CAD in the elderly 30. As the predictive value of the Framingham Cardiovascular Risk Score may decrease markedly with older age 31,32, this may explain why UA appears to improve the prediction in the older but not in the middle-aged individuals, consistent with our observation that adding UA to an atherosclerotic risk model only results in very limited value.
Uric acid in acute myocardial infarction
The results of the present study indicate that the levels of UA were significantly higher in patients with type 2 MI compared with type 1 MI patients. This observation may, according to the available literature, be surprising as a type 1 MI per definition results from acute coronary plaque rupture, plaque erosion, or fissuring, and thus to some extent must be associated with CAD 16. A patient with type 2 MI, however, may or may not have CAD, and a type 2 MI per definition arises from an imbalance in myocardial oxygen supply/demand – without the involvement of an acute coronary plaque rupture, erosion, or fissuring 16. Thus, from a theoretical point of view, if a positive association between UA levels and the burden of CAD including plaque rupture exists, one would expect patients with type 1 MI to have higher UA concentrations than patients with a type 2 MI. However, the lack of stratification of patients with CAD in earlier studies into specific subgroups on the basis of the type of MI or associated clinical conditions 1–6 might have masked the fact that UA is elevated in some subgroups, but not in others, such as type 1 MI.
Indeed, the observation that UA is not elevated in type 1 MI is supported by our results in asymptomatic individuals where we did not find any association between UA and the presence or severity of CAC. These individuals may be defined as having clinically stable CAD and thus our combined results indicate that UA is not a specific marker for either stable or acute clinical CAD.
A causal role for UA in CAD has been proposed in several studies using Mendelian randomization 33–35. Causality was suggested in a follow-up of patients hospitalized for coronary angiography 33 and for patients at high cardiovascular risk 34. However, it cannot be ruled out that the causal association is affected by pleiotropy 35. Furthermore, both data by Testa and colleagues and White and colleagues were comprised of a number of studies and therefore have utilized a very ‘broad’ definition of CAD, making it impossible to determine whether the association between UA and CAD is actually a general finding or is restricted to specific subgroups of CAD patients such as the ones that we examine in the present study.
Uric acid in patients with myocardial injury
On the basis of the results of this study, it may be hypothesized that UA is involved in the process leading to myocardial ischemia because of an imbalance between myocardial oxygen demand and myocardial oxygen supply. Indeed, not only in type 2 MI patients but also in patients with ischemic myocardial injury, we observed significantly higher UA concentrations compared with the concentrations in patients with type 1 MI. The relevant clinical conditions in these patients were atrial fibrillation, chronic heart failure, aortic valvular stenosis, hypertension, and chronic CAD 17. Previously, an association has been reported between UA and atrial fibrillation in the general population 36,37 as well as in patients with ventricular tachycardia 38. It is noteworthy that it has been shown that treatment of chronic heart failure patients with allopurinol improved prognosis 39. Earlier data suggest that UA may be a marker of ischemia as UA has been shown to reduce the release of nitric oxide, a vasodilator, from human umbilical vein endothelial cells 40. Reduced vasodilation might be part of the mechanism behind ischemia in CAD with elevated UA, which fits with the beneficial effect of allopurinol treatment 39. In addition, in patients with chronic heart failure, UA has been found to be associated inversely with maximum oxygen uptake, suggesting that increased UA may reflect an impairment in oxidative metabolism in these patients 41. In concordance with this observation, we found that patients in the ischemic myocardial injury group had a higher incidence of heart failure compared with both nonelevated TnI and type 1 MI. Furthermore, 25% of the type 2 MI patients in this study had been diagnosed with COPD, a condition with chronic hypoxia and with which elevated levels of UA have been associated 42. COPD in the binary logistic regression did not reach the level of statistical significance (Table 3; P=0.11); however, this may be because of the relatively low number of patients in the type 2 MI subgroup.
Increased levels of UA can be associated with kidney disease 43 and inflammation 44. Although we tried to avoid the established influence of kidney disease as a cause of elevated UA by excluding all patients with a creatinine above the URL from the study, creatinine levels were still higher in the type 2 MI and ischemic myocardial injury subgroups compared with the nonelevated TnI or type 1 MI subgroups. However, these differences did not reach statistical significance, possibly because of the relatively low number of patients in the type 2 MI subgroup. Thus, one of the main reasons for the differences in UA levels among our patient groups might be subclinical kidney dysfunction. In terms of inflammation, CRP levels were not significantly elevated in the type 2 MI and ischemic myocardial injury subgroups compared with the nonelevated TnI or type 1 MI subgroups, and in our opinion, the CPR levels do not seem to be a plausible cause for the elevated UA levels. In addition, CRP was the highest in the nonischemic myocardial injury and systemic myocardial injury subgroups, in which no elevation of UA levels was found.
Strengths and limitations
A limitation in our study is the small number of patients in a number of the subgroups. However, the stratification into specific MI subgroups according to the universal definition may help us to determine and understand the potential involvement of UA in the pathogenic processes underlying different subtypes of MI and myocardial injury. Furthermore, we use two different populations, one healthy and one diseased, the results from which supplement one another.
We did not find any correlation between UA levels and CAC severity in asymptomatic middle-aged individuals from the general population. Furthermore, we found that patients with type 2 MI and patients with ischemic myocardial injury had significantly higher UA levels than patients with type 1 MI. This observation might support the hypothesis that UA is involved in the pathophysiological mechanisms leading to an imbalance in the oxygen supply/demand ratio in type 2 MI and ischemic myocardial injury.
The authors acknowledge the skillful technical assistance of Kirsten Guldborg Rasmussen and Susanne Schmidt Nielsen.
This work was supported by The Danish Heart Foundation (grant number 14-R97-A5094 and 16-R107-A6605-22028), the Odense Patient data Explorative Network (Grant number 2014-024), The Region of Southern Denmark (Grant number 12/21371 and 14/24208), and Odense University Hospital, as well as The Strategic Research Council of Svendborg Hospital (Grant number 13/42602)
Conflicts of interest
There are no conflicts of interest.
1. Kojima S, Sakamoto T, Ishihara M, Kimura K, Miyazaki S, Yamagishi M, et al. Prognostic usefulness of serum uric acid after acute myocardial infarction
(the Japanese Acute Coronary Syndrome Study). Am J Cardiol 2005; 96:489–495.
2. Ranjith N, Myeni NN, Sartorius B, Mayise C. Association between hyperuricemia and major adverse cardiac events in patients with acute myocardial infarction
. Metab Syndr Relat Disord 2017; 15:18–25.
3. Ndrepepa G, Braun S, Haase HU, Schulz S, Ranftl S, Hadamitzky M, et al. Prognostic value of uric acid in patients with acute coronary syndromes. Am J Cardiol 2012; 109:1260–1265.
4. Duran M, Kalay N, Akpek M, Orscelik O, Elcik D, Ocak A, et al. High levels of serum uric acid predict severity of coronary artery disease in patients with acute coronary syndrome. Angiology 2012; 63:448–452.
5. Rothenbacher D, Kleiner A, Koenig W, Primatesta P, Breitling LP, Brenner H. Relationship between inflammatory cytokines and uric acid levels with adverse cardiovascular outcomes in patients with stable coronary heart disease. PLoS One 2012; 7:e45907.
6. Ndrepepa G, Braun S, King L, Hadamitzky M, Haase HU, Birkmeier KA, et al. Association of uric acid with mortality in patients with stable coronary artery disease. Metabolism 2012; 61:1780–1786.
7. Cheong E, Ryu S, Lee JY, Lee SH, Sung JW, Cho DS, et al. Association between serum uric acid and cardiovascular mortality and all-cause mortality: a cohort study. J Hypertens 2017; 35 (Suppl 1):S3–S9.
8. Nossent J, Raymond W, Divitini M, Knuiman M. Asymptomatic hyperuricemia is not an independent risk factor for cardiovascular events or overall mortality in the general population of the Busselton Health Study. BMC Cardiovasc Disord 2016; 16:256.
9. Atar AI, Yilmaz OC, Akin K, Selcoki Y, Er O, Eryonucu B. Serum uric acid level is an independent risk factor for presence of calcium in coronary arteries: an observational case-controlled study. Anadolu Kardiyol Derg 2013; 13:139–145.
10. Grossman C, Shemesh J, Koren-Morag N, Bornstein G, Ben-Zvi I, Grossman E. Serum uric acid is associated with coronary artery calcification
. J Clin Hypertens (Greenwich) 2014; 16:424–428.
11. Wang H, Jacobs DR Jr, Gaffo AL, Gross MD, Goff DC Jr, Carr JJ. Longitudinal association between serum urate and subclinical atherosclerosis
: the Coronary Artery Risk Development in Young Adults (CARDIA) study. J Intern Med 2013; 274:594–609.
12. Kim H, Kim SH, Choi AR, Kim S, Choi HY, Kim HJ, Park HC. Asymptomatic hyperuricemia is independently associated with coronary artery calcification
in the absence of overt coronary artery disease: a single-center cross-sectional study. Medicine (Baltimore) 2017; 96:e6565.
13. Neogi T, Terkeltaub R, Ellison RC, Hunt S, Zhang Y. Serum urate is not associated with coronary artery calcification
: the NHLBI Family Heart Study. J Rheumatol 2011; 38:111–117.
14. Diederichsen AC, Sand NP, Norgaard B, Lambrechtsen J, Jensen JM, Munkholm H, et al. Discrepancy between coronary artery calcium score and HeartScore in middle-aged Danes: the DanRisk study. Eur J Prev Cardiol 2012; 19:558–564.
15. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M Jr, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 1990; 15:827–832.
16. Thygesen K, Alpert JS, White HD. Universal definition of myocardial infarction
. Eur Heart J 2007; 28:2525–2538.
17. Sarkisian L, Saaby L, Poulsen TS, Gerke O, Hosbond S, Jangaard N, et al. Prognostic impact of myocardial injury related to various cardiac and noncardiac conditions. Am J Med 2016; 129:506–514.
18. Saaby L, Poulsen TS, Hosbond S, Larsen TB, Pyndt Diederichsen AC, Hallas J, et al. Classification of myocardial infarction
: frequency and features of type 2 myocardial infarction
. Am J Med 2013; 126:789–797.
19. Saaby L, Poulsen TS, Diederichsen AC, Hosbond S, Larsen TB, Schmidt H, et al. Mortality rate in type 2 myocardial infarction
: observations from an unselected hospital cohort. Am J Med 2014; 127:295–302.
20. Miller R. Simultaneous statistical inference, 2nd ed. New York, NY: Springer; 1981.
21. Hilbe J. Negative binomial regression. Cambridge, UK: Cambridge University Press; 2008.
22. Osborne J. Best practices in logistic regression. Thousand Oaks, CA: SAGE Publications; 2014.
23. Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ 2006; 332:1080.
24. Kuss O. The danger of dichotomizing continuous variables: a visualization. Teach Stat 2013. 78–79.
25. Vuong Q. Likelihood ratio tests for model selection and non-nested hypotheses. Econometrica 1989. 307–333.
26. Berezin AE, Kremzer AA. Serum uric acid as a marker of coronary calcification in patients with asymptomatic coronary artery disease with preserved left ventricular pump function. Cardiol Res Pract 2013; 2013:129369.
27. Krishnan E, Pandya BJ, Chung L, Dabbous O. Hyperuricemia and the risk for subclinical coronary atherosclerosis
– data from a prospective observational cohort study. Arthritis Res Ther 2011; 13:R66.
28. Budoff MJ, Gul KM. Expert review on coronary calcium. Vasc Health Risk Manag 2008; 4:315–324.
29. Kavousi M, Elias-Smale S, Rutten JH, Leening MJ, Vliegenthart R, Verwoert GC, et al. Evaluation of newer risk markers for coronary heart disease risk classification: a cohort study. Ann Intern Med 2012; 156:438–444.
30. Dutta A, Henley W, Pilling LC, Wallace RB, Melzer D. Uric acid measurement improves prediction of cardiovascular mortality in later life. J Am Geriatr Soc 2013; 61:319–326.
31. de RW, Westendorp RG, Assendelft WJ, den Elzen WP, de Craen AJ, le CS, Gussekloo J. Use of Framingham risk score and new biomarkers to predict cardiovascular mortality in older people: population based observational cohort study. BMJ 2009; 338:a3083.
32. Stork S, Feelders RA, van den Beld AW, Steyerberg EW, Savelkoul HF, Lamberts SW, et al. Prediction of mortality risk in the elderly. Am J Med 2006; 119:519–525.
33. Kleber ME, Delgado G, Grammer TB, Silbernagel G, Huang J, Kramer BK, et al. Uric acid and cardiovascular events: a Mendelian randomization study. J Am Soc Nephrol 2015; 26:2831–2838.
34. Testa A, Prudente S, Leonardis D, Spoto B, Sanguedolce MC, Parlongo RM, et al. A genetic marker of hyperuricemia predicts cardiovascular events in a meta-analysis of three cohort studies in high risk patients. Nutr Metab Cardiovasc Dis 2015; 25:1087–1094.
35. White J, Sofat R, Hemani G, Shah T, Engmann J, Dale C, et al. Plasma urate concentration and risk of coronary heart disease: a Mendelian randomisation analysis. Lancet Diabetes Endocrinol 2016; 4:327–336.
36. Nyrnes A, Toft I, Njolstad I, Mathiesen EB, Wilsgaard T, Hansen JB, Lochen ML. Uric acid is associated with future atrial fibrillation: an 11-year follow-up of 6308 men and women – the Tromso Study. Europace 2014; 16:320–326.
37. Kawasoe S, Kubozono T, Yoshifuku S, Ojima S, Oketani N, Miyata M, et al. Uric acid level and prevalence of atrial fibrillation in a japanese general population of 285 882. Circ J 2016; 80:2453–2459.
38. Yamada S, Suzuki H, Kamioka M, Kamiyama Y, Saitoh S, Takeishi Y. Uric acid increases the incidence of ventricular arrhythmia in patients with left ventricular hypertrophy. Fukushima J Med Sci 2012; 58:101–106.
39. Gotsman I, Keren A, Lotan C, Zwas DR. Changes in uric acid levels and allopurinol use in chronic heart failure: association with improved survival. J Card Fail 2012; 18:694–701.
40. Kang DH, Park SK, Lee IK, Johnson RJ. Uric acid-induced C-reactive protein expression: implication on cell proliferation and nitric oxide production of human vascular cells. J Am Soc Nephrol 2005; 16:3553–3562.
41. Leyva F, Anker S, Swan JW, Godsland IF, Wingrove CS, Chua TP, et al. Serum uric acid as an index of impaired oxidative metabolism in chronic heart failure. Eur Heart J 1997; 18:858–865.
42. Durmus KN, Sasak G, Aka AU, Akgun M, Boga S, Sengul A, et al. Serum uric acid levels and uric acid/creatinine ratios in stable chronic obstructive pulmonary disease (COPD) patients: are these parameters efficient predictors of patients at risk for exacerbation and/or severity of disease? Med Sci Monit 2016; 22:4169–4176.
43. Li L, Yang C, Zhao Y, Zeng X, Liu F, Fu P. Is hyperuricemia an independent risk factor for new-onset chronic kidney disease?: a systematic review and meta-analysis based on observational cohort studies. BMC Nephrol 2014; 15:122.
44. Spahic E, Hasic S, Kiseljakovic E, Resic H, Kulic M. Positive correlation between uric acid and C-reactive protein serum level in healthy individuals and patients with acute coronary syndromes. Med Glas (Zenica) 2015; 12:128–132.