Uric acid is not an independent predictor of cardiovascular mortality in type 2 diabetes: A population-based study

Atherosclerosis 221 (2012) 183–188

Contents lists available at SciVerse ScienceDirect

Atherosclerosis journal homepage: www.elsevier.com/locate/atherosclerosis

Uric acid is not an independent predictor of cardiovascular mortality in type 2 diabetes: A population-based study F. Panero a , G. Gruden a , M. Perotto a , P. Fornengo a , F. Barutta a , E. Greco a , C. Runzo a , G. Ghezzo b , P. Cavallo-Perin a , Graziella Bruno a,∗ a b

Department of Internal Medicine, University of Torino, Italy Santo Spirito Hospital, Casale Monferrato, Alessandria, Italy

a r t i c l e

i n f o

Article history: Received 30 March 2011 Received in revised form 25 November 2011 Accepted 30 November 2011 Available online 26 December 2011 Keywords: Uric acid Mortality Survey Diabetes Cardiovascular diseases

a b s t r a c t Objective: Although some studies have suggested that uric acid is a risk factor for mortality, this relationship is still uncertain in people with type 2 diabetes. Methods: The study base was the population-based cohort of 1540 diabetic subjects (median age 68.9 years) of the Casale Monferrato Study. The role of serum uric acid on 15-years all-cause, cardiovascular and non-cardiovascular mortality was assessed by multivariate Cox proportional hazards modeling. Results: Baseline levels of serum uric acid were negatively correlated with HbA1c, were higher in men and in the elderly and were independently associated with components of the metabolic syndrome. Out of 14,179 person-years, 1000 deaths (514 due to cardiovascular diseases) were observed. Compared to the lower quartile of uric acid, HRs (95% CI) in the upper quartile were 1.47 (1.22–1.76) for all-cause mortality; 1.40 (1.09–1.80) for cardiovascular mortality and 1.50 (1.15–1.96) for non-cardiovascular mortality. In multiple adjusted models, however, HRs were 1.30 (1.06–1.60) for all-cause mortality, 1.13 (0.85–1.50) for cardiovascular mortality and 1.50 (1.11–2.02) for non-cardiovascular mortality (men 1.87, 1.19–2.95; women 1.20, 0.80–1.80); the latter appeared to be due to neoplastic diseases (HR in all combined quartiles vs. lower quartile: both sexes 1.59, 1.05–2.40; men 1.54, 0.83–2.84, women 1.68, 0.95–2.92). Conclusions: In diabetic people, uric acid is associated with components of the metabolic syndrome but it may not be accounted as an independent risk factor for cardiovascular mortality. The increased all-cause mortality risk with higher levels of uric acid might be due to increased neoplastic mortality and deserves future studies. © 2012 Published by Elsevier Ireland Ltd.

Traditional risk factors account only for a part of individual cardiovascular risk, whereas the residual risk still represents an unknown and fundamental burden hanging over [1,2]. Many studies have indicated that serum uric acid – the end product or purine catabolism – is associated with hypertension, obesity, hyperinsulinemia and dyslipidemia, suggesting that it could be part of the cluster of factors of the metabolic syndrome [3]. High levels of serum uric acid are associated with lowering glomerular filtration rates, peripheral vascular disease, stroke and vascular dementia [3,4]. Moreover, hyperuricemia is usually found in patients at high cardiovascular risk [3,5]. Though the association between hyperuricemia and cardiovascular disease has been speculative for

∗ Corresponding author at: Department of Internal Medicine, University of Turin, Corso Dogliotti 14, I-10126 Torino, Italy. Tel.: +39 11 6336 709; fax: +39 11 6634 751. E-mail address: [email protected] (G. Bruno). 0021-9150/$ – see front matter © 2012 Published by Elsevier Ireland Ltd. doi:10.1016/j.atherosclerosis.2011.11.042

many years, several studies have tried to elucidate the independent role of serum uric acid as cardiovascular risk factor in the general population [6–9]. As uric acid is strictly linked to many cardiovascular risk factors, it may be simply a marker of increased cardiovascular and renal risk. Whether uric acid plays either a causative role or behaves as a marker of atherosclerotic disease is clinically relevant, as several therapies to treat hyperuricemia might be employed. The identification of new risk factors or biomarkers that improve prediction of cardiovascular events might be even more important in diabetic people, in whom the residual risk is still high, in spite of adequate treatment of known risk factors. However, studies examining the role of serum uric acid on all-cause and cardiovascular mortality of diabetic people are limited [10–13]. HbA1c has been shown to be inversely related to serum uric acid, and inconsistent results of studies might be due to inadequate control for the confounding effect of glycemic control over the study period; other possible explanations include the retrospective study design [10],

184

F. Panero et al. / Atherosclerosis 221 (2012) 183–188

the limited number of recruited people [11,12], the short length of follow-up [13] and the pattern of confounders assessed in multivariate analyses [10–12]. In this 15-years follow up of the population-based Casale Monferrato cohort of type 2 diabetes we aimed to examine variables associated with serum uric acid and its relationship with all-cause, cardiovascular and non-cardiovasculr mortality, independently of known and novel risk factors and cumulative HbA1c during the study period.

1. Materials and methods The study-base included 1540 patients with known type 2 diabetes who were residents in the town of Casale Monferrato in 1991, North-West of Italy (93,477 inhabitants), were invited to a baseline examination in 1991–92 to assess the prevalence of diabetic nephropathy and cardiovascular risk factors and were then followed up to 31/12/2006 [14–16]. They were identified in the first prevalence survey of the Casale Monferrato Study, using multiple independent data sources (diabetes clinics, general practitioners, hospital discharges, administrative registries for prescription, strips and syringes), obtaining a high estimated completeness of ascertainment (80%) [14–16]. Surveys conducted within Italy showed they were representative of Italian diabetic patients as regard age, sex, duration of diabetes, body mass index (BMI), and type of anti-diabetic treatment. As described in detail elsewhere, at baseline all patients were interviewed and examined by trained investigators [14–16], provided they had given informed consent. The study was approved by the local ethic committee, in accordance with the Declaration of Helsinki. For all patients enrolled the date of diagnosis was retrieved and recorded. Venous blood samples were collected in the fasting state for determinations of triglycerides, total cholesterol, HDL-cholesterol (enzymatic–colorimetric method, after precipitation with Mn2+ ), apolipoprotein (apo) A1 and B (turbidimetric method, BM/Hitachi 717; BBR, Tokyo, Japan), hemoglobin A1c (HbA1c) (HPLC, Daiichi, Menarini, Japan, laboratory reference range 3.8–5.5%) and fibrinogen (Clauss method). LDL-cholesterol was calculated with the Friedewald’s formula for all subjects in the cohort whose triglycerides values were <4.48 mmol/l. Albumin excretion rate (AER) was calculated on the basis of urinary albumin concentration measured on a single timed overnight urine collection by nephelometric method (Behring Nephelometer Analyzer, Behring Institute, Marburg, Germany), after having excluded urinary tract infection, congestive heart failure or other known causes of non-diabetic renal diseases. Micro and macroalbuminuria were defined as AER between 20 and 200 ␮g/min and AER > 200 ␮g/min, respectively. All laboratory determinations were centralized. Hypertension was defined as systolic blood pressure > 140 mmHg and/or diastolic blood pressure > 90 mmHg or treatment with antihypertensive drugs. Smoking habit was classified into three categories: never smoker, ex-smoker (if patient stopped smoking at least one month before the visit) and smoker; daily alcohol intake was also estimated in g/die from self-referral habits. Coronary heart disease was defined on the basis of abnormalities on a 12-leads electrocardiography according to the Minnesota code (as probable for major Q and QS items (codes 1.1 and 1.2) or as possible for minor Q and QS items, S-T/T items (codes 1.3, 4.1–4.4 and 5.1–5.3)). We estimated the glomerular filtration rate (eGFR) by the four-component abbreviated equation from the Modification of Diet in Renal Disease Study (all patients were Euripids): eGFR = 186 × (serum creatinine [mg/dl])−1.154 × age − 0.203 (×0.742 if female). During the period 1991–2006, patients were regularly examined, 3–4 times a year either at the diabetes clinic or by general practitioners. Cumulative individual average values of HbA1c,

updated at each follow-up examination during the study period, were calculated.

2. Statistical analysis distributed variables are presented as Normally means ± standard deviation (SD), whereas variables with skewed distribution were analyzed after logarithmic transformation (triglycerides, AER, creatinine) and results presented as geometric means and 95% CIs. Differences in clinical characteristics of patients were assessed by the ANOVA test for continuous variables and 2 test for categorical ones. Pearson’s correlation coefficients for serum uric acid and continuous variables were also calculated. Variables assessed were age, sex, BMI, triglycerides, total cholesterol, HDL-cholesterol, apoB/apoA1, systolic and diastolic blood pressure, hypertension, CHD, smoke, fibrinogen, AER, eGFR, alcohol consumption and treatment with diuretics, hypoglycemic, hypolipemic and hypo-uricemic drugs. We assessed with ordinal logistic regression analyses variables independently associated with odds of shifting through quartiles of serum uric acid (first quartile as reference), after adjustment for age and sex. All continuous variables were categorized into quartiles of their distribution, except age (continuous variable), BMI (<26, 26–29, >29 kg/m2 ), HbA1c (quintiles), AER (<20 ␮g/min or ≥20 ␮g/min) and eGFR (<60, ≥60 ml/min). We performed ordinal logistic regression, comparing nested models through both the backward and the forward strategy. Two models are nested if both contain the same predictors and one has at least an additional predictor. In the final analysis, we included variables that were significantly associated with the independent variable on a likelihood ratio test basis or that modified odds ratios (ORs) for other variables included. Information on deaths up to December 31, 2006 was obtained from the demographic registries of the town of residence, hospital discharge and autopsy records. The underlying causes of death were ascertained and coded according to the ICD-9 classification by two authors. In twelve individuals (two in the first quartile, four in the second and six in the forth quartile of serum uric acid) the cause of death could not be ascertained; therefore, they were included in all-cause mortality analyses only. Mortality rates were calculated by dividing the number of deaths occurring during the study period by the number of person-years of observation. The role of uric acid as a predictor of mortality, independently of conventional and new risk factors and confounders, was assessed by multivariate Cox proportional hazards modeling. We included covariates in models on the basis of their biological plausibility, previous mortality findings in this cohort [15,16] and statistically significant univariate associations with uric acid. Models were constructed with variables as continuous measures to provide maximum power for detecting association between uric acid and mortality. Variables included in models fulfilled the proportional hazard assumptions, since the results of tests based on Schoenfeld residuals were not significant. The right form of covariates entering each model was checked by Martingale residuals; goodness of fit of the fully adjusted model was ascertained by the Cox–Snell residuals. We tested for possible interaction between serum uric acid and sex, age, BMI and eGFR by adding a single interaction term one at time in the fully adjusted model. The likelihood ratio (LR) test was used to assess the statistical significance of variables in nested models. We tested for linear trend across categorical variables by entering a single ordinal term in the Cox regression model. The relevant time scale for the analysis was time since diabetes diagnosis to death or to 31 December 2006; given that, all models were adjusted also for known duration of diabetes. The p values were two-sided and values lower than 0.05 were considered statistically significant. For

F. Panero et al. / Atherosclerosis 221 (2012) 183–188

non-dichotomous categorical variables, p values for trend were estimated according to Mantel–Haentzel. All analyses were performed with STATA (Stata Release 11.0, Stata Corporation, College Station, TX, 2009). 3. Results Measurements of serum uric acid were available for 1509 of 1540 diabetic people of the Casale Monferrato cohort, with mean value 321.79 ± 99.33 ␮mol/l. Sex differences in serum uric acid were evident, with higher values in men than in women (p < 0.001), even after adjustment for age, BMI, blood pressure and eGFR (330.11 ␮mol/l, 95% CI 322.98–337.85 vs. 314.65 ␮mol/l, 95% CI 308.11–321.19). Frequency of clinically relevant hyperuricemia (values higher than 416.36 ␮mol/l) was found in 13.2% of the cohort, 14.2% of men and 12.5% of women (p = 0.003). The baseline characteristics of type 2 diabetic people by serum uric acid quartile are shown in Table 1; people with values in the highest quartile were more likely to be male, older, overweight or obese, hypertensive, to have a worse kidney function and lipid profile, to be insulin-treated and to have suffered from previous CHD; moreover, they had lower duration of diabetes and serum glucose and HbA1c values. Indeed, uric acid levels were negatively correlated with HbA1c (r = − 0.09, p = 0.0005). In ordinal logistic regression analysis, variables independently associated with serum uric acid, after adjustment for age and sex, were eGFR, BMI, triglycerides, HDL-cholesterol, hypertension, cumulative HbA1c and fibrinogen (online Table). Both HbA1c and HDL were negatively associated with uric acid. Further adjustment for either CHD, smoke or treatment with diuretics, hypoglycemic drugs and allopurinol did not modify observed associations. During the 15-year follow-up period (median 13.7 years, interquartile range 7.5–15.1), 1000 people with type 2 diabetes died over 14,179 person-years of observations, which represented an all-cause mortality rate per 1000 person-years of 70.5 (95% CI 66.3–75.0). Fig. 1 shows survival curves of the cohort by quartiles of uric acid (log-rank test p = 0.0015). As shown in Table 2, subjects in the highest quartile of uric acid experienced a significantly higher hazard of all-cause mortality than subjects in the lowest quartile (unadjusted HR = 1.47, 95% CI 1.22–1.76, p for trend <0.0001), which was reduced after adjustment for age, sex and diabetes duration (model 1, HR = 1.30, 95% CI 1.08–1.56, p for trend = 0.002). HR was unmodified and still statistical significant after further adjustment for other risk factors (model 2, HR = 1.30, 95% CI 1.06–1.60, p for

0.00

0.25

0.50

0.75

1.00

all-cause mortality by quartile of uric acid

0

5

analysis time

uric_acid = <255.76 uric_acid = 303.36-374.72

10

15

uric_acid = 255.76-303.35 uric_acid = >374.72

Fig. 1. Survival curves of the Casale Monferrato survey, by quartiles of serum uric acid. Log-rank test p = 0.0015.

185

trend = 0.004). No investigated interactions were statistically significant; in particular, the interaction between sex and uric acid was not significant, indicating that the role of uric acid on survival was not different between men and women. With regards to cardiovascular mortality (ICD 390–459), rates were based on 514 (51.4%) deaths, giving a mortality rate per 1000 person-years of 36.2 (95% CI 33.2–39.5). Subjects in the highest quartile of uric acid experienced a significantly higher hazard than subjects in the lowest quartile (unadjusted HR = 1.40, 95% CI 1.09–1.80, p for trend = 0.02). HR, however, was reduced to a not significant level after adjustment for age, sex, diabetes duration (model 1, HR = 1.25, 95% CI 0.97–1.60, p for trend = 0.08) and other risk factors (model 2, HR = 1.13, 95% CI 0.85–1.50, p for trend = 0.27). We then examined whether the independent association between serum uric acid and all-cause mortality could be mainly attributed to non-cardiovascular mortality (474/1000 deaths). As shown in Table 2, people with serum uric acid in the upper quartile had significantly higher unadjusted hazard of dying for non-cardiovascular diseases (HR = 1.50, 95% CI 1.15–1.96, p trend = 0.001). Differently from cardiovascular mortality, HRs were almost unmodified after adjustment for other risk factors (model 2, HR = 1.50, 1.11–2.02, p for trend = 0.003). Table 3 shows results of analyses performed by sex; the direction of the associations were similar to that found in the whole cohort in both men and women for all-cause and cardiovascular mortality, whereas as regards noncardiovascular mortality a significant increasing trend was evident in men only. Finally, we analyzed separately neoplastic mortality (ICD-9 codes 140.0–239.9), which was based on 183 deaths only (92 deaths in men and 91 in women), showing that with respect to the lowest quartile, point estimates of HRs were similarly high in all upper quartiles: HRs = 1.59 (1.01–2.53), 1.60 (0.99–2.59) and 1.56 (0.94–2.57). Compared to the lowest quartile, HRs in all combined upper quartiles were 1.59 (1.05–2.40, p = 0.028) in both sexes, 1.54 (0.83–2.84) in men and 1.68 (0.95–2.92) in women.

4. Discussion Our study aimed to investigate the role of serum uric acid on all-cause and cardiovascular mortality in a large population-based cohort of people with type 2 diabetes, during a 15-years followup period. The cross-sectional analysis of our study confirms the association between serum uric acid and variables included in the cluster of the metabolic syndrome (obesity, hypertension, HDLcholesterol and triglycerides). In the prospective analysis, however, uric acid was not significantly associated with cardiovascular mortality. Even though in univariate analysis values in the upper quartile conferred a significantly 40% increased risk with respect to values in the lower quartile, this association largely attenuated after adjustment for age, sex and diabetes duration and finally disappeared after adjustment for other cardiovascular risk factors. These findings are consistent with studies showing that hyperuricemia might reflect an underlying insulin-resistance state, which by itself provides higher cardiovascular risk [17]. Our study provides evidence that uric acid has no independent role on cardiovascular mortality in diabetic people, and, therefore it should be considered a marker rather than a determinant of cardiovascular diseases [3]. Finally, we found statistically significant higher hazards of both all-cause and non-cardiovascular mortality. Although the study was not specifically powered to assessed this issue, this excess appeared to be due to neoplastic mortality and deserves further considerations. In the general population, the role of uric acid on cardiovascular mortality has been examined in studies providing inconsistent results [3,4,7,18]. As regards to diabetic people, studies are

186

F. Panero et al. / Atherosclerosis 221 (2012) 183–188

Table 1 Characteristics of the Casale Monferrato cohort of people with type 2 diabetes, by quartiles of serum uric acid. Uric acid (␮mol/l)

n Men (%) Age (years) Duration of diabetes (years) Body mass index (kg/m2 ) Glucose (mmol/l) HbA1c (%) Total cholesterol (mmol/l) LDL cholesterol (mmol/l) HDL cholesterol (mmol/l) Triglycerides (mmol/l) ApoA1 (g/l)a ApoB (g/l) ApoB/apoA1 Fibrinogen (␮mol/l) Creatinine (␮mol/l)a Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Hypertension CHD AER (␮g/min) <20 20–200 >200 Smoking Never Ex Current Alcohol consumption (g/die) Treatment with diuretics Treatment with insulin

<255.76

255.76–303.35

303.36–374.72

>374.72

Total

p

361 134 (36.9%) 67.8 ± 11.4 11.4 ± 7.3 26.2 ± 4.9 9.56 ± 3.54 8.4 ± 2.4 5.73 ± 1.17 3.51 ± 1.00 1.56 ± 0.49 1.30 (1.23–1.37) 1.38 ± 0.36 0.99 ± 0.35 0.73 ± 0.28 10.54 ± 2.93 71.68 (70.16–73.21) 151.3 ± 22.0 86.0 ± 10.5 277 (77.2%) 64 (21.0%)

383 154 (39.9%) 69.3 ± 10.5 11.1 ± 7.3 26.9 ± 4.3 9.16 ± 3.22 8.0 ± 2.2 5.74 ± 1.25 3.58 ± 1.08 1.42 ± 0.39 1.46 (1.39–1.54) 1.34 ± 0.36 1.05 ± 0.39 0.79 ± 0.29 10.41 ± 2.64 74.73 (73.21–76.26) 153.2 ± 20.2 86.6 ± 10.3 323 (83.9%) 79 (24.9%)

401 195 (48.6%) 68.2 ± 10.5 10.1 ± 6.0 27.8 ± 4.3 8.82 ± 3.13 7.7 ± 2.3 5.78 ± 1.26 3.61 ± 1.13 1.38 ± 0.37 1.56 (1.49–1.63) 1.30 ± 0.39 1.04 ± 0.36 0.84 ± 0.64 10.56 ± 2.47 81.60 (79.31–83.12) 156.4 ± 21.4 88.3 ± 10.1 345 (87.1%) 79 (23.9%)

364 174 (47.8%) 70.0 ± 10.0 10.4 ± 6.8 28.3 ± 4.7 9.10 ± 3.32 7.9 ± 2.1 5.88 ± 1.35 3.69 ± 1.56 1.30 ± 0.39 1.77 (1.68–1.87) 1.24 ± 0.39 1.08 ± 0.39 0.86 ± 0.33 11.08 ± 2.72 89.22 (86.94–91.51) 156.8 ± 23.2 89.0 ± 10.8 320 (88.6%) 92 (29.1%)

1509 657 (43.5%) 68.8 ± 10.6 10.7 ± 6.9 27.3 ± 4.6 9.15 ± 3.31 8.0 ± 2.3 5.78 ± 1.26 3.60 ± 1.10 1.41 ± 0.42 1.52 (1.48–1.55) 1.33 ± 0.38 1.04 ± 0.37 0.81 ± 0.42 10.64 ± 2.70 78.55 (77.79–80.07) 154.5 ± 21.8 87.5 ± 10.5 1265 (83.8%) 314 (20.8%)

0.001 0.02 0.05 <0.001 0.02 <0.001 0.34 0.19 <0.001 <0.001 0.002 0.01 <0.001 <0.004 <0.001 <0.001 <0.001 <0.001 0.13

194 (54.3%) 110 (30.8%) 53 (14.9%)

205 (55.1%) 116 (31.2%) 51 (13.7%)

191 (49.0%) 133 (34.1%) 66 (16.9%)

152 (42.8%) 113 (31.8%) 90 (25.4%)

740 (49.0%) 472 (31.3%) 260 (17.2%)

0.001

238 (67.0%) 63 (17.8%) 54 (15.2%) 5.2 ± 12.8 40 (11.1%) 74 (21.8%)

246 (65.1%) 82 (21.7%) 50 (13.2%) 4.8 ± 13.4 36 (9.4%) 51 (13.9%)

251 (64.2%) 103 (26.3%) 37 (9.5%) 6.2 ± 15.5 39 (9.7%) 48 (12.6%)

227 (64.9%) 85 (24.2%) 38 (10.9%) 4.5 ± 13.0 46 (16.6%) 46 (13.1%)

962 (63.8%) 333 (22.1%) 179 (11.9%) 5.2 ± 13.8 161 (10.7%) 219 (15.3%)

0.05

0.34 0.07 <0.001

p values refer to one-way ANOVA or the 2 test (for categorical variables). a Geometric means (95% confidence interval).

potentially biased by the clinic-based study design [10,12,13] and by the inadequate adjustment for diabetes duration and other risk factors [10–12]. Although hyperinsulinemia as a consequence of insulin resistance causes an increase in serum uric acid by reducing renal uric acid excretion [20,21], both our study and others have shown a negative association between serum uric acid and glycemic control [10,11,19]; lower serum uric acid in people with higher values of plasma glucose might suggest an alteration of uric

acid tubular reabsorption in presence of glycosuria [22], but further studies are needed to clarify this issue. However, higher levels of HbA1c might be associated with both increased cardiovascular risk and lower uric acid levels, causing an attenuation to the null value of the association between uric acid and mortality depending on the level of glycemic control. Differently from other studies, however, our results were independent of cumulative HbA1c over the study period.

Table 2 Mortality rates and results of Cox-regression analyses in the Casale Monferrato Study, by quartiles of serum uric acid. Deaths (n) All-cause mortality Uric acid (␮mol/l) 215 <255.76 255 255.76–303.35 268 303.36–374.72 >374.72 262 p value for trend Cardiovascular mortality Uric acid (␮mol/l) 115 <255.76 134 255.76–303.35 131 303.36–374.72 134 >374.72 p value for trend Non-cardiovascular mortality Uric acid (␮mol/l) 98 <255.76 117 255.76–303.35 137 303.36–374.72 122 >374.72 p value for trend a b

Mortality rates/1000 person-years

Unadjusted HR

Model 1 HRa

Model 2 HRb

59.2 69.5 71.4 83.7

1.00 1.21 (1.00–1.45) 1.28 (1.07–1.53) 1.47 (1.22–1.76) <0.0001

1.00 1.09 (0.91–1.30) 1.21 (1.01–1.45) 1.30 (1.08–1.56) 0.002

1.00 1.06 (0.87–1.29) 1.25 (1.02–1.52) 1.30 (1.06–1.60) 0.004

31.7 36.5 34.9 42.8

1.00 1.19 (0.92–1.52) 1.16 (0.90–1.50) 1.40 (1.09–1.80) 0.02

1.00 1.07 (0.83–1.37) 1.10 (0.86–1.42) 1.25 (0.97–1.60) 0.08

1.00 0.99 (0.76–1.30) 1.13 (0.86–1.48) 1.13 (0.85–1.50) 0.27

27.0 31.9 36.5 39.0

1.00 1.22 (0.93–1.60) 1.44 (1.11–1.88) 1.50 (1.15–1.96) 0.001

1.00 1.10 (0.84–1.45) 1.36 (1.04–1.77) 1.32 (1.00–1.73) 0.018

1.00 1.14 (0.86–1.52) 1.44 (1.08–1.92) 1.50 (1.11–2.02) 0.003

Adjusted for age, sex and diabetes duration; Adjusted for the above (a ) and for hypertension, HDL-cholesterol, LDL-cholesterol, triglycerides, BMI, smoking and HbA1c;

F. Panero et al. / Atherosclerosis 221 (2012) 183–188

187

Table 3 Mortality rates and results of Cox-regression analyses in the Casale Monferrato Study, by quartiles of serum uric acid and sex. Deaths (n) Men All-cause mortality Uric acid (␮mol/l) 82 <255.76 108 255.76–303.35 124 303.36–374.72 122 >374.72 p value for trend Cardiovascular mortality Uric acid (␮mol/l) 44 <255.76 57 255.76–303.35 59 303.36–374.72 >374.72 54 p value for trend Non-cardiovascular mortality Uric acid (␮mol/l) <255.76 37 50 255.76–303.35 65 303.36–374.72 >374.72 67 p value for trend Women All-cause mortality Uric acid (␮mol/l) 133 <255.76 147 255.76–303.35 303.36–374.72 144 >374.72 140 p value for trend Cardiovascular mortality Uric acid (␮mol/l) <255.76 71 255.76–303.35 77 303.36–374.72 72 80 >374.72 p value for trend Non-cardiovascular mortality Uric acid (␮mol/l) 61 <255.76 67 255.76–303.35 303.36–374.72 72 >374.72 55 p value for trend a b

Mortality rates/1000 person-years

Unadjusted HR

Model 1 HRa

Model 2 HRb

62.0 76.3 66.5 82.2

1.00 1.26 (0.94–1.68) 1.12 (0.84–1.48) 1.38 (1.04–1.83) 0.07

1.00 1.13 (0.84–1.51) 1.11 (0.84–1.47) 1.25 (0.94–1.66) 0.15

1.00 1.08 (0.79–1.48) 1.18 (0.87–1.60) 1.35 (0.98–1.84) 0.05

33.2 40.3 31.7 36.4

1.00 1.27 (0.86–1.89) 1.02 (0.69–1.51) 1.19 (0.80–1.78) 0.68

1.00 1.15 (0.77–1.71) 0.99 (0.67–1.47) 1.09 (0.73–1.63) 0.89

1.00 1.05 (0.68–1.61) 0.94 (0.62–1.44) 0.96 (0.61–1.51) 0.75

28.0 35.3 34.9 45.2

1.00 1.25 (0.82–1.92) 1.26 (0.84–1.89) 1.60 (1.07–2.40) 0.03

1.00 1.12 (0.73–1.72) 1.29 (0.86–1.93) 1.43 (0.95–2.14) 0.06

1.00 1.15 (0.72–1.83) 1.56 (0.99–2.44) 1.87 (1.19–2.95) 0.002

57.7 65.3 76.2 85.0

1.00 1.18 (0.93–1.49) 1.42 (1.12–1.80) 1.51 (1.19–1.91) <0.0001

1.00 1.07 (0.84–1.35) 1.28 (1.01–1.63) 1.33 (1.05–1.69) 0.007

1.00 1.03 (0.80–1.32) 1.27 (0.98–1.65) 1.24 (0.95–1.62) 0.05

30.8 34.2 38.1 48.6

1.00 1.13 (0.82–1.57) 1.29 (0.93–1.80) 1.60 (1.16–2.20) 0.003

1.00 1.01 (0.73–1.40) 1.15 (0.83–1.60) 1.38 (1.00–1.91) 0.03

1.00 0.95 (0.67–1.35) 1.21 (0.85–1.73) 1.23 (0.85–1.78) 0.14

26.5 29.8 38.1 33.4

1.00 1.20 (0.85–1.70) 1.61 (1.14–2.27) 1.30 (0.90–1.88) 0.05

1.00 1.11 (0.78–1.57) 1.48 (1.05–2.10) 1.18 (0.82–1.70) 0.16

1.00 1.12 (0.77–1.61) 1.38 (0.94–2.03) 1.20 (0.80–1.80) 0.23

Adjusted for age, sex and diabetes duration. Adjusted for the above (a ) and for hypertension, HDL-cholesterol, LDL-cholesterol, triglycerides, BMI, smoking and HbA1c.

Risk factors, such as hypertension and microalbuminuria, may lie in the causal path linking hyperuricemia and cardiovascular mortality; therefore, adjustment for such factors would determine an overadjustment bias. However, our negative finding was evident even after adjustment for age, sex and diabetes duration. These findings suggest that the role of uric acid as independent cardiovascular risk factor, if any, is likely to be very limited. We also found a statistically significant 30% increased risk of all-cause mortality in people with uric acid levels in the upper quartile. A similar increased hazard was found in the highest quintile of uric acid of a large prospective study of 22,698 Korean men (14% with diabetes) [11]. That study, however, did not adjust HR for other risk factors, whereas our excess was confirmed even after multiple adjustments. Our data show that the excess of all-cause mortality was almost entirely due to a statistically significant 50% increased non-cardiovascular mortality rate, which was based on 474 deaths (183 deaths due to neoplastic diseases and 291 deaths due to other causes); this association was unmodified by further adjustment for other risk factors, although in subgroup analyses it was evident in men only. Our study was not specifically powered to assess the predictive role of uric acid on neoplastic mortality. It is remarkable, however, that the magnitude of the association between all-cause mortality and uric acid appears to be almost

entirely due to neoplastic diseases, that a statistically increased risk was evident for uric acid values of 255.76 ␮mol/l and over and that the point estimates of risks were similar in men and in women when analyzed separately. Our finding is consistent with another study showing that higher levels of uric acid are associated with incidence of neoplastic diseases [23]. Another large study on 83,683 men followed-up for a median of 13.6 years reported that people with baseline uric acid values > 398.52 ␮mol/l experienced a 40% greater risk for cancer mortality in comparison to those with values < 273.61 ␮mol/l, even when accounting for leadtime bias [24]. Similar findings were found in women also [24]. As the negative predictive role on survival is evident even in studies with long follow-up periods, hyperuricemia might be a very early manifestation of the carcinogenic process [23–27]. Uric acid has antioxidant properties, which may protect against carcinogenesis; however, no evidence of a protective effect has been found neither in gout patients nor in a large cohort of older women [25,28]. Alternately, higher levels of uric acid might be a surrogate marker for lifestyles at increased risk of cancer. The association between uric acid and cancer remains largely unexplored to date and deserves future studies. Strengths of the present study are the prospective design, the large population-based cohort, the centralization of all

188

F. Panero et al. / Atherosclerosis 221 (2012) 183–188

procedures, the representativeness of the Italian diabetic population and the long follow-up period. Moreover, we were able to adjust for classical and novel risk factors as well as for cumulative HbA1c over the study period. The Casale Monferrato Study recruited a population-based cohort of subjects, allowing even subjects cared for exclusively by general practitioners to be included in the analyses. In Italy all citizens are covered by the National Health System; thus, the model of care is not associated with socioeconomic factors. Limitations of our study deserve consideration. We cannot rule out the possibility of residual or undetected confounding. Analyses are based on a baseline serum uric acid measurement and its stability over time cannot be definitively shown. The absence of any association between uric acid and cardiovascular mortality could be due to lower statistical power with respect to all-cause mortality. However, analyses of non-cardiovascular mortality were based on a quite similar number of deaths, but in contrast with cardiovascular mortality, statistical significance was maintained even in the fully adjusted model. In conclusion, in people with type 2 diabetes serum uric acid is associated with components of the metabolic syndrome, but it acts as a marker rather than a determinant of cardiovascular mortality. The statistically significant increased all-cause mortality risk in people with higher levels of serum uric acid might be due to increased neoplastic mortality and deserves future studies. Author contributions PF, researched data and wrote the manuscript, MP, PF, FB, CR and GG researched data, PCP contributed to the discussion and reviewed/edited the manuscript, GB designed the study, researched data and wrote the manuscript. Acknowledgments We thank the patients, the nurses at the diabetes clinic, the diabetologists and general practitioners for long-standing collaboration in this study. The Casale Monferrato Study is supported by grants from the Piedmont Region (Ricerca Sanitaria Finalizzata 2008). The Authors have no conflict of interest with results of the study. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.atherosclerosis.2011.11.042. References [1] Blankenberg S, Zeller T, Saarela O, et al. Contribution of 30 biomarkers to 10year cardiovascular risk estimation in 2 population cohorts: the MONICA, risk, genetics, archiving, and monograph (MORGAM) biomarker project. Circulation 2010;121:2381–3. [2] Melander O, Newton-Cheh C, Almgren P, et al. Novel and conventional biomarkers for prediction of incident cardiovascular events in the community. JAMA 2009;302:49–57. [3] Feig DI, Kang DH, Johnson RJ. Uric acid and cardiovascular risk. N Engl J Med 2008;359:1811–21.

[4] Iseki K, Ikemiya Y, Inoue T, et al. Significance of hyperuricemia as a risk factor for developing ESRD in a screened cohort. Am J Kidney Dis 2004;44:642–50. [5] Johnson RJ, Kang DH, Feig D, et al. Is there a pathogenetic role for uric acid in hypertension and cardiovascular and renal disease. Hypertension 2003;41:1183–90. [6] Niskanen LK, Laaksonen DE, Nyyssonen K, et al. Uric acid level as a risk factor for cardiovascular and all-cause mortality in middle-aged men: a prospective cohort study. Arch Intern Med 2004;164:1546–51. [7] Fang J, Alderman MH. Serum uric acid and cardiovascular mortality the NHANES I epidemiologic follow-up study, 1971–1992. National Health and Nutrition Examination Survey. JAMA 2000;283:2404–10. [8] Strasak AM, Kelleher CC, Brant LJ, et al. Serum uric acid is an independent predictor for all major forms of cardiovascular death in 28,613 elderly women: a prospective 21-year follow-up study. Int J Cardiol 2008;125:232–9. [9] Strasak A, Ruttmann E, Brant L, et al. Serum uric acid and risk of cardiovascular mortality: a prospective long-term study of 83,683 Austrian men. Clin Chem 2008;54:273–84. [10] Ioachimanescu AG, Brennan DM, Hoar BM, Hazen SL, Hoogwerf BJ. Serum uric acid is an indipendent predictor of all-cause mortality in patients at high risk of cardiovascular disease: a preventive cardiology information system (Pre CIS) database cohort study. Arthritis Rheum 2008;58:623–30. [11] Jee SA, Soon YL, Kim MT. Serum uric acid and risk of death from cancer, cardiovascular disease or all cause in men. Eur J Cardiovasc Prev Rehabil 2004;11:185–91. [12] Mazza A, Zamboni S, Rizzato E, et al. Serum uric acid shows a J-shaped trend with coronary mortality in non-insulin dependent diabetic elderly people. The Cardiovascular Study in the Elderly (CASTEL). Acta Diabetol 2007;44:99–105. [13] Zoppini G, Targher G, Negri C, et al. Elevated serum uric acid concentrations independently predict cardiovascular mortality in type 2 diabetic patients. Diabetes Care 2009;32:1716–20. [14] Bruno G, Merletti F, Bargero G, et al. Changes over time in the prevalence and quality of care of type 2 diabetes in Italy: the Casale Monferrato Surveys, 1988 and 2000. Nutr Metab Cardiovasc Dis 2008;18:39–45. [15] Bruno G, Merletti F, Biggeri A, et al. Fibrinogen and AER are major independent predictors of 11-year cardiovascular mortality in type 2 diabetes: the Casale Monferrato Study. Diabetologia 2005;48:427–34. [16] Bruno G, Merletti F, Bargero G, et al. Estimated glomerular filtration rate, albuminuria and mortality in type 2 diabetes: the Casale Monferrato Study. Diabetologia 2007;50:941–8. [17] Bonora E, Kiechl S, Willeit J, et al. Insulin resistance as estimated by homeostasis model assessment predicts incident symptomatic cardiovascular disease in Caucasian subjects from the general population: the Bruneck study. Diabetes Care 2007;30:318–24. [18] Culleton BF, Larson MG, Kannel WB, Levy D. Serum uric acid and risk for cardiovascular disease and death. Ann Intern Med 1999;131:7–13. [19] Lehto S, Niskanen L, Ronnemaa T, Laakso M. Serum uric acid is a strong predictor of stroke in patients with non-insulin-dependent diabetes mellitus. Stroke 1998;29:635–9. [20] Quiniones Galvan A, Natali A, Baldi S, et al. Effect of insulin on uric acid excretion in humans. Am J Physiol 1995;268:E1–5. [21] Muscelli E, Natali A, Bianchi S, et al. Effect of insulin on renal sodium and uric acid handling in essential hypertension. Am J Hypertens 1996;9:746–52. [22] Cook DG, Shaper AG, Thelle DS, Whitehead TP. Serum uric acid, serum glucose and diabetes: relationships in a population study. Postgrad Med J 1986;62:1001–6. [23] Strasak AM, Lang S, Kneib T, et al. Use of penalized splines in extended Coxtype additive hazard regression to flexibly estimate the effect of time-varying serum uric acid on risk of cancer incidence: a prospective, population-based study in 78,850 men. Ann Epidemiol 2009;19:15–24. [24] Strasak AM, Rapp K, Hilbe W, et al. Serum uric acid and risk of cancer mortality in a large prospective male cohort. Cancer Causes Control 2007;18:1021–9. [25] Strasak AM, Rapp K, Hilbe W, et al. The role of serum uric acid as an antioxidant protecting against cancer: prospective study in more than 28 000 older Austrian women. Ann Oncol 2007;18:1893–7. [26] Levine W, Dyer AR, Shekelle RB, Schoenberger JA, Stamler J. Serum uric acid and 11.5-year mortality of middle-aged women: findings of the Chicago Heart Association Detection project in Industry. J Clin Epidemiol 1989;42:257–67. [27] Loh WJ, North BV, Johnston DG, Godsland IF. Insulin resistance-related biomarker clustering and subclinical inflammation as predictors of cancer mortality during 21.5 years of follow-up. Cancer Causes Control 2010;21:709–18. [28] Boffetta P, Nordenvall C, Nyrén O, Ye W. A prospective study of gout and cancer. Eur J Cancer Prev 2009;18:127–32.