- Email: [email protected]
Serum uric acid and risk of ischemic stroke: The ARIC Study
Atherosclerosis 187 (2006) 401–407
Serum uric acid and risk of ischemic stroke: The ARIC Study Atsushi Hozawa a,b , Aaron R. Folsom a,∗ , Hassan Ibrahim c , F. Javier Nieto d , Wayne D. Rosamond e , Eyal Shahar a a
b
Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, 1300 South 2nd Street, Suite 300, Minneapolis, MN 55454-1015, USA Department of Public Health and Forensic Medicine, Division of Epidemiology, Tohoku University Graduate School of Medicine, Sendai, Japan c Department of Medicine, University of Minnesota, Minneapolis, MN, USA d Department of Population Health Sciences, University of Wisconsin, Madison, WI, USA e Department of Epidemiology, University of North Carolina, Chapel Hill, NC, USA Received 12 July 2005; received in revised form 26 August 2005; accepted 20 September 2005 Available online 18 October 2005
Abstract Aims: Since serum uric acid (UA) is strongly associated with cardiovascular risk factors, it has been debated whether serum UA is a stroke risk factor or whether UA may be simply “marking” subjects with other, causal risk factors. We therefore investigated the relation between UA and ischemic stroke in the Atherosclerosis Risk in Communities (ARIC) Study. Methods and results: Of 15,792 ARIC participants, 13,413 who were free of recognized stroke or coronary heart disease (CHD) at baseline and had a baseline UA measurement were included in the analysis. We followed the participants for ischemic stroke incidence (N = 381) over 12.6 years. Although serum UA was independently and positively related to ischemic stroke incidence when we adjusted for age, sex, race, and education, the positive relation was weakened when additionally adjusted for possible confounding variables. The positive multivariateadjusted association between serum UA and ischemic stroke was observed among subjects not using diuretics (adjusted relative hazard in the highest quartile versus the lowest: relative hazard (RH) = 1.49; 95% confidence interval (CI): 1.00–2.23) (P for trend: 0.02), but not among diuretic users (P for interaction: 0.08). Conclusion: Our findings suggest that UA is an independent predictor of ischemic stroke among subjects not using diuretics, but that elevated UA itself may not cause ischemic stroke. © 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Uric acid; Stroke; Diuretics; Prospective studies
1. Introduction Elevated serum uric acid (UA) has been associated with an increased risk of cardiovascular disease in studies spanning five decades [1–11]. However, since UA is frequently elevated in subjects with other cardiovascular disease risk factors [1,2], it has been debated whether UA is an independent cardiovascular disease risk factor through an ability to induce inflammatory and vascular mechanisms, or whether UA may be simply “marking” subjects with other, causal risk ∗
Corresponding author. Tel.: +1 612 626 8862; fax: +1 612 624 0315. E-mail address: [email protected] (A.R. Folsom).
0021-9150/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.atherosclerosis.2005.09.020
factors [1,2,12–14]. Alternatively, it has been hypothesized that hyperuricemia may be in part a compensatory mechanism to counteract oxidative damage related to atherosclerosis and aging [15–17]. We previously investigated the relation between UA and coronary heart disease (CHD) in the Atherosclerosis Risk in Communities (ARIC) Study and found that UA was not an independent risk factor for CHD [12]. Prior studies of the relation between UA and stroke incidence have been inconsistent [4–10,18,19]. Some studies reported a positive independent relation between UA and stroke [4–8]; others demonstrated that UA did not relate significantly to stroke occurrence [9,10,18,19].
402
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407
Diuretics are known to increase serum UA. If UA itself is harmful, the relation between UA and stroke should be similar between diuretic users and non-users. However, if elevated UA is simply “marking” subjects with other, causal risk factors or part of a compensatory mechanism to oxidative damage, the relation between UA and stroke might differ between diuretic users and non-users. Reyes has proposed a compensatory mechanism hypothesis that elevation in serum UA caused by drugs that reduce its renal excretion, such as diuretics, may actually be beneficial to cardiovascular prognosis [16,17]. Therefore, we believed that analyzing the relation between UA and ischemic stroke separately for diuretic users and non-users may help address whether UA itself is harmful to ischemic stroke. However, there are no prior studies of the effect of serum UA on ischemic stroke in diuretic users and non-users. Therefore, in this study, we investigated the relation between UA and ischemic stroke incidence in the middleaged general population. We also investigated whether the association varied by use of diuretics to help clarify whether UA itself increases ischemic stroke risk.
2. Methods 2.1. Study design and subjects The Atherosclerosis Risk in Communities Study is a multi-center prospective cohort study investigating the natural history of atherosclerotic disease in four US communities: Forsyth County, NC; Jackson, MS; Washington County, MD; the northwest suburbs of Minneapolis, MN [20]. The cohort comprised 15,792 men and women aged 45–64 years who were selected by probability list or area sampling. Only African Americans were recruited in the Jackson study center. From 1987 to 1989, a baseline home interview assessed participants’ sociodemographic characteristics, smoking and alcohol-drinking habits, medication use, reproductive history, and personal history of diseases. A clinical examination included measurement of various risk factors and cardiovascular conditions, B-mode ultrasound examinations of the carotid and popliteal arteries, and a 12-lead electrocardiogram (ECG). 2.2. Baseline examination After an overnight fast, blood was drawn from the antecubital vein of seated participants into vacuum tubes containing sodium citrate for hemostatic factors, EDTA for lipids, and serum separator gel for chemistries including UA. The tubes were centrifuged at 3000 × g for 10 min at 4 ◦ C. Plasma and serum aliquots were frozen at −70 ◦ C until analysis a few weeks later. UA was measured by the Uricase method. The reliability coefficient of UA, assessed by repeated measurements taken at least one week apart in 40 subjects (10
per ARIC center), was 0.91, and within-person variability was 7.2% [21]. High-density lipoprotein (HDL) cholesterol [22] was measured after dextran–magnesium precipitation of non-HDL lipoproteins. Details also have been reported for centralized measurement of von Willebrand factor (vWF) and serum creatinine and albumin [23–26]. Estimates of intraindividual variability in blood measurements have been reported [27,28]. Smoking status (current, former, or never smoking) and usual ethanol intake were estimated from interview [29]. Subjects were asked whether they currently drank alcoholic beverages and, if not, whether they had done so in the past. Current drinkers were asked how often they usually drank wine, beer, or hard liquor. In calculating the amount of alcohol consumed (in grams per week), it was assumed that 4 oz of wine contains 10.8 g, 12 oz of beer contains 13.2 g, and 1.5 oz of liquor contains 15.1 g of ethanol [30]. Sitting blood pressures (BP) were measured three times using a randomzero sphygmomanometer after 5 min rest [31]. The mean of the last two measurements was used for analysis. Use of anti-hypertensive medications within the past two weeks of baseline interview was self-reported [29]. Prevalent diabetes mellitus was defined as a fasting glucose of >126 mg/dL, or a reported history of physician-diagnosed diabetes, or current use of diabetes medication. For anthropometric measurements participants wore only light clothing. Body weight was measured to the nearest pound and height was measured to the nearest centimeter. Body mass index (BMI) was computed as weight (kg) divided by height squared (m2 ). Waist-to-hip ratio (WHR), an indicator of fat distribution, was computed as the circumference of the waist (umbilical level) in centimeter divided by the maximum circumference of the hips in centimeter. A 12-lead ECG [32] was used to define left ventricular hypertrophy (LVH), using the Cornell score [33]. Preexisting CHD at baseline was defined by self-reported prior physician diagnosis of myocardial infarction (MI) or coronary revascularization, or prevalent MI by ARIC ECG [32]. Pre-existing stroke was defined by any self-reported prior physician diagnosis of stroke. The ankle/brachial index (ABI) was defined as the ratio of ankle to brachial systolic BP, and peripheral arterial disease (PAD) was defined as ABI ≤ 0.90 for men and ≤0.85 for women. ARIC’s ultrasound measurements were based on the technique validated by Pignoli et al. [34], using a scanning protocol common to the four field centers [35,36], and standardized central reading of scans [37]. Analyses were based on mean intima-media thickness (IMT) of the far wall for 1-cm lengths of the carotid bifurcation and the internal and common carotid, right and left, adjusted for site-specific reader differences and downward measurement drifts in IMT over the baseline visit, with the means at missing sites imputed by maximal likelihood methods [38]. The means at the six sites were combined in an unweighted average to produce an overall mean IMT. In some analyses, extreme IMT was defined as 1.0 mm.
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407
We calculated 10-year probabilities of ischemic stroke using a previously reported prediction model [39]. To calculate the probabilities, we used sex-specific models involving age, race, smoking status, history of diabetes, anti-hypertensive medication, ECG LVH, systolic BP, BMI, albumin, von Willebrand factor, WHR, ethanol intake, low level of HDL cholesterol, carotid IMT, and PAD [39]. 2.3. Ascertainment of incident events Stroke incidence was ascertained by contacting participants annually, identifying hospitalizations during the previous year, and by surveying discharge lists from local hospitals and death certificates from state vital statistics offices for potential cerebrovascular events [29,40,41]. Potential strokes were found by selecting records with ICD-9 CM discharge codes 430–438 or with mention of stroke or neuroimaging in the discharge summary. For these, hospital records were copied and abstracted by a trained nurse. Each eligible case was classified by computer algorithm and by a physician reviewer, according to criteria adapted from the National Survey of Stroke [42]. Differences in diagnosis were adjudicated by another reviewer. Details on quality assurance for ascertainment and classification of events are presented elsewhere [41]. The criteria for classification were based on combinations of symptom type, duration, and severity, results of neuroimaging and other diagnostic procedures, and autopsy evidence when available [40,41]. A stroke was classified as ischemic, which we have taken as the event of interest, if a brain CT or MR revealed acute infarction or showed no evidence of hemorrhage [43]. 2.4. Data analysis A total of 15,792 ARIC subjects participated in the baseline ARIC exam. Of these, 13,413 were free of recognized stroke or CHD at baseline and had a baseline UA measurement. Of the original sample, we excluded 329 participants who had history of stroke, 1028 participants who had a history of CHD, 119 subjects without UA data, 170 for taking allopurinol or other uricosuric agents, and 39 for creatinine values ≥2.0 mg/dL (these four conditions were not mutually exclusive). We further excluded, due to small numbers, non-White and non-African American subjects in Minneapolis and Washington County (N = 42). Subjects who failed to fast for blood drawing and subjects who did not have complete information on confounding factors (N = 652) were also excluded. The excluded participants were 0.6 years younger, 31.9% more likely to be Black, and 24.6% more likely to have diabetes compared with who had complete information on confounding variables (N = 13,413). The association of baseline UA with other risk factors was assessed using analysis of variance or χ2 -test, as appropriate. The association between UA and ischemic stroke incidence was estimated from Cox proportional hazard models. Relative hazards (RH) for stroke were computed across
403
approximate quartiles of UA using the bottom quartile as the reference category. In Model 1, we adjusted for age, race (African American, White), and education (less than high school graduate, high school graduate, greater than high school graduate). In Model 2, we adjusted for major risk factors for ischemic stroke in our previous study, i.e., systolic BP (continuous variable), diabetes mellitus (yes, no), anti-hypertensive medication (yes, no), cigarette smoking (current, former, never), ethanol intake (continuous variable), serum albumin, vWF, BMI, WHR, and low HDL cholesterol (cutpoints for “low” being 60 mg/dL in women and 50 mg/dL in men) [39]. In a supplemental analysis, we further adjusted for ECG LVH, carotid IMT, and PAD. Although these variables are predictors of ischemic stroke [39], we did not include these variables in the main model (Model 2), as we were concerned to do so would be over adjustment, i.e., these variables might be on the causal pathway or might be surrogates of variables on the causal pathway. In supplemental analyses, we used Model 2 for adjustment. P-values for tests of linear trends were calculated by treating the categories for quartile of UA as an ordinal variable. Differences with a two-tailed P-value < 0.05 were considered statistically significant. All statistical analyses were performed using SAS software, Version 8.2 (SAS Institute, Cary, NC).
3. Results 3.1. Uric acid and other cardiovascular risk factors The mean value ± S.D. of UA was 5.97 ± 1.52 mg/dL (range 0.5–15.9 mg/dL). Table 1 shows the association between UA and other potential risk factors for stroke. Mean values of age, systolic BP, ethanol intake, serum albumin, vWF, WHR, and BMI were higher in higher UA categories. Similarly, the proportions of subjects who were men, diabetic, or ever smoked; who had less than a high school education, used anti-hypertensive medications, or had low HDL cholesterol; or who had ECG LVH, increased IMT, or PAD were increased across UA quartiles. Consequently, the predicted 10-year ischemic stroke probability was higher in the higher UA categories. 3.2. Uric acid and ischemic stroke The mean duration of follow-up was 12.6 years (maximum 15.1 years), during which 381 incident ischemic strokes occurred. Table 2 shows the RH for ischemic stroke according to approximate quartiles of baseline UA. Serum UA was independently and positively related to ischemic stroke incidence when we adjusted for age, sex, race, and education (RH in the highest quartile: 1.74; 95% confidence interval (CI): 1.27–2.38), compared with the first quartile. However, the positive relation was weakened (RH in the highest quartile: 1.25; 95% CI: 0.91–1.73) when additionally adjusted
404
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407
Table 1 Prevalences or mean values of baseline participant characteristics according to approximate quartiles of serum uric acid: the Atherosclerosis Risk in Communities (ARIC) Study, 1987–1989 Characteristic
P-valuea
Serum uric acid quartile (mg/dL) ≤4.8
4.9–5.8
5.9–6.8
≥6.9
3534 53.0 86.6 20.6 116.4 7.0
2852 53.9 66.3 23.8 119.5 8.5
3792 54.4 44.7 24.8 121.5 10.0
3235 54.6 31.6 30.6 125.2 12.6
Smoking Current (%) Former (%) Never (%)
27.2 22.4 50.3
27.4 28.5 44.1
24.6 34.8 40.5
22.4 41.1 36.4
<0.001
EducationHigh school (%)
18.6 36.4 45.0
20.3 34.9 44.7
22.3 31.3 46.4
26.2 29.6 44.0
<0.001
16.1 50.9 24.8 3.85 111.0
21.3 60.8 32.3 3.86 114.7
26.7 69.8 46.8 3.88 116.7
43.2 77.0 63.6 3.91 123.7
<0.001 <0.001 <0.001 <0.001 <0.001
Waist-to-hip ratio Body mass index (kg/m2 )
0.87 25.2
0.91 27.0
0.94 28.3
0.96 29.7
<0.001 <0.001
LVH (%) PAD (%)c Elevated carotid IMT (%)d
0.8 1.8 4.5
1.5 2.1 5.2
1.8 1.8 6.8
3.2 3.1 9.5
<0.001 <0.001 <0.001
10-Year stroke probability (%)
2.0
2.7
3.4
4.6
<0.001
N Age Sex (women, %) Race (African American, %) Systolic blood pressure (mmHg) Diabetes (yes, %)
Use of anti-hypertensive medication (%) Low HDL cholesterol (%)b Ethanol intake (g/week) Serum albumin (g/dL) von Willebrand Factor (%)
<0.001 <0.001 <0.001 <0.001 <0.001
N: number of subjects; LVH: left ventricular hypertrophy on electrocardiogram; PAD: peripheral artery disease; IMT: intima-media thickness. a P-value from a test of equality of all means (analysis of variance) or a contingency table (χ2 with 1 d.f. or 2 d.f.). b Cutpoints for “low” were 60 mg/dL in women and 50 mg/dL in men. c Ankle brachial pressure index ≤ 0.90 for men and ≤0.85 for women. d IMT ≥ 1.0 mm.
Table 2 Relative hazards (95% confidence interval) for ischemic stroke incidence according to quartile of serum uric acid at baseline, ARIC, 1987–2001 Serum uric acid quartile (mg/dL)
Total N Events Model 1 Model 2 Diuretic non-user N Events Model 1 Model 2 Diuretic user N Events Model 1 Model 2
P for linear trends
≤4.8
4.9–5.8
5.9–6.8
≥6.9
3534
2852
3792
3235
59 0.96 (0.68–1.37) 0.86 (0.60–1.23)
114 1.30 (0.95–1.77) 1.09 (0.79–1.49)
141 1.74 (1.27–2.38) 1.25 (0.91–1.73)
2534
3183
2303
45 0.99 (0.66–1.49) 0.92 (0.62–1.39)
85 1.41 (0.98–2.04) 1.23 (0.84–1.78)
86 1.89 (1.29–2.76) 1.49 (1.00–2.23)
318
609
932
14 0.72 (0.35–1.48) 0.60 (0.29–1.25)
29 0.69 (0.37–1.28) 0.71 (0.38–1.33)
55 0.72 (0.40–1.30) 0.73 (0.40–1.34)
67 1.00 1.00 3243 51 1.00 1.00 291 16 1.00 1.00
<0.01 0.06
<0.01 0.02
0.46 0.62
Model 1: adjusted for age, sex, race, and education; Model 2: Model 1 + systolic blood pressure, diabetes mellitus, anti-hypertensive medication, cigarette smoking status, ethanol intake, serum albumin, von Willebrand factor, BMI, WHR, and low HDL cholesterol; N: number of subjects; BMI: body mass index; WHR: waist-to-hip ratio.
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407
for systolic BP, diabetes mellitus, anti-hypertensive medication, cigarette smoking, ethanol intake, albumin, vWF, BMI, WHR, and low HDL cholesterol (Model 2). Because diuretics affect serum UA level, we analyzed the relation separately for diuretic users and non-users. In Model 2, there remained a positive statistically significant relation between UA and ischemic stroke in subjects not using diuretics (RH = 1.49 in fourth quartile; 95% CI: 1.00–2.23) (P for linear trend: 0.02). The RH per 1 mg/dL greater UA level was 1.10 (95% CI: 1.01–1.21). There was no significant relation between UA and ischemic stroke in diuretic users (RH = 0.73 in fourth quartile; 95% CI: 0.40–1.34) (P for linear trend: 0.62). The interaction between UA and diuretic use for stroke incidence was marginally significant (P for interaction: 0.08, using Model 2). These results were unchanged if we also adjusted for ECG LVH, carotid IMT, and PAD. Removal of serum albumin and vWF from Model 2 also had no appreciable impact. 3.3. Supplemental analyses As shown in Table 3, we conducted a subgroup analysis among subjects not taking diuretics, stratifying on age (age < 55 years and ≥55 years at baseline), sex (men and women), race (African American and White), 10-year predicted ischemic stroke risk (<5% and ≥5%), and sub-clinical conditions (LVH, PAD, or extreme IMT). There were no significant interactions of UA with the stratifying variables (all P > 0.10). However, the magnitude of the RH of ischemic stroke in the highest UA quartile was greater in men than women, and in subjects with versus without sub-clinical conditions.
405
4. Discussion In this prospective cohort study, we demonstrated an independent positive moderate association between UA and ischemic stroke incidence restricted to those not using diuretics. A motivation of this study was to clarify whether UA itself is a risk factor for ischemic stroke by analyzing the relation between UA and ischemic stroke separately for diuretic users and non-users. Some mechanisms have been proposed to explain a potential relation between UA and atherosclerosis. First, the relation between UA and atherosclerosis may reflect risk factor clustering such as metabolic syndrome, often due to insulin resistance, or a pathogenic role of UA in hypertension [1,2,11,44]. However, when we adjusted for other risk factors, even though the RH was somewhat attenuated, we still found a positive relation between UA and ischemic stroke among subjects not taking diuretics. Thus, risk factor clustering did not fully explain the positive relation between UA and ischemic stroke, and UA was an independent predictor of ischemic stroke among subjects not taking diuretics. Second, UA may have a direct ability to induce inflammatory and vascular mechanisms that may contribute to the development of cardiovascular disease [1]. However, if UA itself affects atherosclerosis, one might hypothesize that UA would be associated with ischemic stroke risk even in diuretic users. Yet, it is possible that the duration of exposure of arteries to high UA is less among diuretic users than those with physiologically induced hyperuricemia. Third, since UA has strong antioxidant properties, and atherosclerosis has been linked to increased oxidative stress, UA could be a compensatory
Table 3 Stratum-specific relative hazards (95% confidence interval) for ischemic stroke incidence in subjects not using diuretics (N = 11,263) according to quartile of serum uric acid at baseline, ARIC, 1987-2001 N
Events
Serum uric acid quartile (mg/dL) ≤4.8
4.9–5.8
5.9–6.8
≥6.9
P for interaction
Older (age 55 years) Younger (age < 55 years)
4970 6293
179 88
1.00 1.00
0.79 (0.48–1.29) 1.23 (0.59–2.60)
1.05 (0.67–1.66) 1.68 (0.84–3.38)
1.64 (1.03–2.63) 1.22 (0.56–2.66)
0.64
Men Women
5125 6138
149 118
1.00 1.00
1.01 (0.48–2.13) 0.85 (0.51–1.41)
1.30 (0.67–2.53) 1.22 (0.75–1.99)
1.63 (0.83–3.19) 1.27 (0.70–2.30)
0.51
White African American
8733 2530
164 103
1.00 1.00
0.81 (0.48–1.35) 1.14 (0.58–2.27)
1.03 (0.64–1.66) 1.53 (0.82–2.84)
1.27 (0.76–2.12) 1.79 (0.93–3.45)
0.88
10-Year ischemic stroke prediction 5%a 10-Year ischemic stroke prediction < 5%a
1301 8794
111 124
1.00 1.00
0.95 (0.47–1.93) 0.78 (0.45–1.36)
1.64 (0.90–3.01) 0.85 (0.49–1.47)
1.29 (0.67–2.48) 1.49 (0.85–2.65)
0.12
With sub-clinical conditionb Without sub-clinical conditionb
943 9178
61 175
1.00 1.00
0.76 (0.29–2.01) 0.84 (0.51–1.36)
1.32 (0.57–3.06) 1.05 (0.67–1.64)
2.46 (1.08–5.62) 1.04 (0.63–1.71)
0.16
Adjusted for age, sex, race, education, systolic blood pressure, diabetes mellitus, anti-hypertensive medication, cigarette smoking status, ethanol intake, serum albumin, von Willebrand factor, BMI, WHR, and low HDL cholesterol. N: number of subjects; BMI: body mass index; WHR: waist-to-hip ratio; sub-clinical condition: having increased intima-media thickness, electrocardiogram left ventricular hypertrophy, or peripheral artery disease. a Excludes 1168 subjects who did not have information on variables for calculating 10-year ischemic stroke probability. Among 1168 subjects, 7, 673, 222, 20, and 246 did not have information on smoking status, intima-media thickness, peripheral artery disease, HDL cholesterol, and electrocardiographic left ventricular hypertrophy, respectively. b Excludes 1142 subjects who had incomplete information on sub-clinical conditions. Among 1142 subjects, 674, 249, and 219 did not have information on intima-media thickness, electrocardiographic left ventricular hypertrophy, and peripheral artery disease, respectively.
406
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407
mechanism to protect the body from pro-oxidants [15,16]. This hypothesis could explain differences between diuretic users and subjects without diuretics: i.e., increased UA due to oxidative stress could be harmful, whereas diuretic-induced hyperuricemia does not cause ischemic stroke. This hypothesis is also consistent with the finding from randomized controlled trials that diuretics and other anti-hypertensive drugs are equivalent in reducing stroke occurrence [45,46]. If the diuretic-induced elevation of UA played an important role in stroke incidence, diuretics would not be expected to reduce stroke risk equivalently to other anti-hypertensives. Of course, there is a possibility that diuretics have a greater efficacy for stroke prevention than other anti-hypertensive drugs, but this is offset by diuretic-induced high UA; however, we consider this possibility unlikely. There is also the possibility that the measured UA level of many diuretic users does not reflect lifetime values, and hence was misclassified; such misclassification would have tended to bias the association we observed toward the null hypothesis. Because we cannot establish these mechanisms in an observational study, additional clinical trials might be needed to clarify whether UA itself plays a causal role in ischemic stroke incidence. Previous prospective studies focused on the relation between UA and stroke, but yielded inconsistent findings [4–10,18,19]. Most positive reports, but not all [10,19], involved high-risk subjects, i.e., hospital-based subjects [7], subjects with diabetes [8], an older general population [5], or older subjects with isolated systolic hypertension (ISH) [6]. Studies finding no association examined lower risk subjects, e.g., a Japanese middle-aged general population [18]. Accordingly, we analyzed higher risk and lower risk subjects, separately. Although we found that the relation between UA and ischemic stroke was somewhat stronger in subjects with sub-clinical conditions and higher predicted stroke probability, no statistically significant interactions were observed. Therefore, our data did not support the hypothesis that the relation between UA and stroke is stronger in high-risk subjects than in lower risk subjects, but power was limited to detect even quite substantial differences. Further experimental or epidemiological data might be needed to clarify whether there is an interaction between UA and background risk of stroke. Strengths of our study included its large, biracial population-based cohort, its careful assessment of stroke incidence over an extended follow-up period, and its ability to control for many stroke risk factors. Our study had some limitations. First, we had only a single measurement of UA. Second, since we did not measure indicators of oxidative stress, we could not evaluate whether oxidative stress-induced UA increase was harmful.
5. Conclusions In conclusion, UA was independently and positively associated with ischemic stroke incidence among subjects not
using diuretics. However, we found no significant relation between UA and ischemic stroke among diuretic users. This finding suggests that UA is independent predictor of ischemic stroke among subjects not using diuretics, but that elevated UA itself may not cause ischemic stroke. Acknowledgements Dr. Hozawa was supported by a grant (14010301) from the Japan Society for the Promotion of Science Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan and the Banyu Fellowship Program sponsored by Banyu Life Science Foundation International. The Atherosclerosis Risk in Communities Study is carried out as a collaborative study supported by National Heart, Lung, and Blood Institute contracts N01-HC-55015, N01-HC-55016, N01-HC-55018, N01-HC-55019, N01-HC55020, N01-HC-55021, and N01-HC-55022. The authors thank the staff and participants of the ARIC study for their important contributions. References [1] 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. [2] Rich MW. Uric acid: is it a risk factor for cardiovascular disease? Am J Cardiol 2000;85:1018–21. [3] Brand FN, McGee DL, Kannel WB, Stokes 3rd J, Castelli WP. Hyperuricemia as a risk factor of coronary heart disease: the Framingham study. Am J Epidemiol 1985;121:11–8. [4] Tomita M, Mizuno S, Yamanaka H, et al. Does hyperuricemia affect mortality? A prospective cohort study of Japanese male workers. J Epidemiol 2000;10:403–9. [5] Mazza A, Pessina AC, Pavei A, et al. Predictors of stroke mortality in elderly people from the general population. The CArdiovascular STudy in the ELderly. Eur J Epidemiol 2001;17:1097–104. [6] Wang JG, Staessen JA, Fagard RH, et al. Prognostic significance of serum creatinine and uric acid in older Chinese patients with isolated systolic hypertension. Hypertension 2001;37:1069–74. [7] Longo-Mbenza B, Luila EL, Mbete P, Vita EK. Is hyperuricemia a risk factor of stroke and coronary heart disease among Africans? Int J Cardiol 1999;71:17–22. [8] 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. [9] Goldberg RJ, Burchfiel CM, Benfante R, et al. Lifestyle and biologic factors associated with atherosclerotic disease in middle-aged men. 20-Year findings from the Honolulu Heart Program. Arch Intern Med 1995;155:686–94. [10] Franse LV, Pahor M, Di Bari M, et al. Serum uric acid, diuretic treatment and risk of cardiovascular events in the Systolic Hypertension in the Elderly Program (SHEP). J Hypertens 2000;18:1149–54. [11] 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. [12] Moriarity JT, Folsom AR, Iribarren C, Nieto FJ, Rosamond WD. Serum uric acid and risk of coronary heart disease: Atherosclerosis Risk in Communities (ARIC) Study. Ann Epidemiol 2000;10:136–43.
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407 [13] Yusuf S, Bosch J. Urate levels as a predictor of cardiac deaths: causal relation or mere association? Eur Heart J 2002;23:760–1. [14] Culleton BF, Larson MG, Kannel WB, Levy D. Serum uric acid and risk for cardiovascular disease and death: the Framingham Heart Study. Ann Intern Med 1999;131:7–13. [15] Nieto FJ, Iribarren C, Gross MD, Comstock GW, Cutler RG. Uric acid and serum antioxidant capacity: a reaction to atherosclerosis? Atherosclerosis 2000;148:131–9. [16] Reyes AJ. Cardiovascular drugs and serum uric acid. Cardiovasc Drugs Ther 2003;17:397–414. [17] Reyes AJ, Leary WP. The increase in serum uric acid induced by diuretics could be beneficial to cardiovascular prognosis in hypertension: a hypothesis. J Hypertens 2003;21:1775–7. [18] Sakata K, Hashimoto T, Ueshima H, Okayama A, NIPPON DATA 80 Research Group. Absence of an association between serum uric acid and mortality from cardiovascular disease: NIPPON DATA 80, 1980–1994. National Integrated Projects for Prospective Observation of Non-communicable Diseases and its Trend in the Aged. Eur J Epidemiol 2001;17:461–8. [19] De Leeuw PW, Thijs L, Birkenhager WH, et al., Systolic Hypertension in Europe (Syst-Eur) Trial Investigators. Prognostic significance of renal function in elderly patients with isolated systolic hypertension: results from the Syst-Eur trial. J Am Soc Nephrol 2002;13:2213–22. [20] The ARIC Investigators. The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. Am J Epidemiol 1989;129:687–702. [21] Eckfeldt JH, Chambless LE, Shen YL. Short-term, within-person variability in clinical chemistry test results. Experience from the Atherosclerosis Risk in Communities Study. Arch Pathol Lab Med 1994;118:496–500. [22] Warnick GR, Benderson JM, Albers JJ. Quantitation of high-densitylipoprotein subclasses after separation by dextran sulfate and Mg2+ precipitation. Clin Chem 1982;28:1574. [23] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 9: hemostasis determinations, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [24] Wu KK, Papp AC, Patsch W, et al. ARIC hemostasis study. II. Organizational plan and feasibility study. Thromb Haemost 1990;64:521–5. [25] Sharrett AR. The Atherosclerosis Risk in Communities (ARIC) Study: introduction and objectives of the hemostasis component. Ann Epidemiol 1992;2:467–9. [26] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 10: clinical chemistry determinations, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [27] Chambless LE, McMahon RP, Brown SA, et al. Short-term intraindividual variability in lipoprotein measurements: the Atherosclerosis Risk in Communities (ARIC) Study. Am J Epidemiol 1992;136:1069–81. [28] Chambless LE, McMahon R, Wu K, et al. Short-term intraindividual variability in hemostasis factors: the ARIC Study. Ann Epidemiol 1992;2:723–33. [29] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 2: cohort component procedures, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [30] Fuchs FD, Chambless LE, Whelton PK, Nieto FJ, Heiss G. Alcohol consumption and the incidence of hypertension: the Atherosclerosis Risk in Communities Study. Hypertension 2001;37:1242–50.
407
[31] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 11: sitting blood pressure, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [32] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities Study (ARIC). Operations manual no. 5: electrocardiography, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [33] Crow RS, Prineas RJ, Rautaharju P, Hannan P, Liebson PR. Relation between electrocardiography and echocardiography for left ventricular mass in mild systolic hypertension (results from Treatment of Mild Hypertension Study). Am J Cardiol 1995;75:1233–8. [34] Pignoli P, Tremoli E, Poli A, Oreste P, Paoletti R. Intimal plus medial thickness of the arterial wall: a direct measurement with ultrasound imaging. Circulation 1986;74:1399–406. [35] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 6A: ultrasound assessment—part A, ultrasound scanning, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [36] Bond MG, Barnes RW, Riley WA, et al., the ARIC Study Group. High-resolution B-mode ultrasound scanning methods in the Atherosclerosis Risk in Communities Study (ARIC). J Neuroimaging 1991;1:68–73. [37] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 6B: ultrasound assessment—part A, ultrasound reading, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [38] Chambless LE, Heiss G, Folsom AR, et al. Association of coronary heart disease incidence with carotid arterial wall thickness and major risk factors: the Atherosclerosis Risk in Communities (ARIC) Study, 1987–1993. Am J Epidemiol 1997;146:483–894. [39] Chambless LE, Heiss G, Shahar E, Earp MJ, Toole J. Prediction of ischemic stroke risk in the Atherosclerosis Risk in Communities Study. Am J Epidemiol 2004;160:259–69. [40] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 3: surveillance component procedures, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [41] Rosamond WD, Folsom AR, Chambless LE, et al. Stroke incidence and survival among middle-aged adults: 9-year follow-up of the Atherosclerosis Risk in Communities (ARIC) cohort. Stroke 1999;30:736–43. [42] The National Survey of Stroke. National Institute of Neurological and Communicative Disorders and Stroke. Stroke 1981;12(2 Pt 2 Suppl. 1):I1–91. [43] Shahar E, Chambless LE, Rosamond WD, et al. Plasma lipid profile and incident ischemic stroke: the Atherosclerosis Risk in Communities (ARIC) study. Stroke 2003;34:623–33. [44] Klein BE, Klein R, Lee KE. Components of the metabolic syndrome and risk of cardiovascular disease and diabetes in Beaver Dam. Diabetes Care 2002;25:1790–4. [45] Psaty BM, Lumley T, Furberg CD, et al. Health outcomes associated with various antihypertensive therapies used as first-line agents: a network meta-analysis. JAMA 2003;289:2534–44. [46] ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial. Major outcomes in high-risk hypertensive patients randomized to angiotensin-converting enzyme inhibitor or calcium channel blocker vs diuretic: the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). JAMA 2002;288:2981–97.
Serum uric acid and risk of ischemic stroke: The ARIC Study Atsushi Hozawa a,b , Aaron R. Folsom a,∗ , Hassan Ibrahim c , F. Javier Nieto d , Wayne D. Rosamond e , Eyal Shahar a a
b
Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, 1300 South 2nd Street, Suite 300, Minneapolis, MN 55454-1015, USA Department of Public Health and Forensic Medicine, Division of Epidemiology, Tohoku University Graduate School of Medicine, Sendai, Japan c Department of Medicine, University of Minnesota, Minneapolis, MN, USA d Department of Population Health Sciences, University of Wisconsin, Madison, WI, USA e Department of Epidemiology, University of North Carolina, Chapel Hill, NC, USA Received 12 July 2005; received in revised form 26 August 2005; accepted 20 September 2005 Available online 18 October 2005
Abstract Aims: Since serum uric acid (UA) is strongly associated with cardiovascular risk factors, it has been debated whether serum UA is a stroke risk factor or whether UA may be simply “marking” subjects with other, causal risk factors. We therefore investigated the relation between UA and ischemic stroke in the Atherosclerosis Risk in Communities (ARIC) Study. Methods and results: Of 15,792 ARIC participants, 13,413 who were free of recognized stroke or coronary heart disease (CHD) at baseline and had a baseline UA measurement were included in the analysis. We followed the participants for ischemic stroke incidence (N = 381) over 12.6 years. Although serum UA was independently and positively related to ischemic stroke incidence when we adjusted for age, sex, race, and education, the positive relation was weakened when additionally adjusted for possible confounding variables. The positive multivariateadjusted association between serum UA and ischemic stroke was observed among subjects not using diuretics (adjusted relative hazard in the highest quartile versus the lowest: relative hazard (RH) = 1.49; 95% confidence interval (CI): 1.00–2.23) (P for trend: 0.02), but not among diuretic users (P for interaction: 0.08). Conclusion: Our findings suggest that UA is an independent predictor of ischemic stroke among subjects not using diuretics, but that elevated UA itself may not cause ischemic stroke. © 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Uric acid; Stroke; Diuretics; Prospective studies
1. Introduction Elevated serum uric acid (UA) has been associated with an increased risk of cardiovascular disease in studies spanning five decades [1–11]. However, since UA is frequently elevated in subjects with other cardiovascular disease risk factors [1,2], it has been debated whether UA is an independent cardiovascular disease risk factor through an ability to induce inflammatory and vascular mechanisms, or whether UA may be simply “marking” subjects with other, causal risk ∗
Corresponding author. Tel.: +1 612 626 8862; fax: +1 612 624 0315. E-mail address: [email protected] (A.R. Folsom).
0021-9150/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.atherosclerosis.2005.09.020
factors [1,2,12–14]. Alternatively, it has been hypothesized that hyperuricemia may be in part a compensatory mechanism to counteract oxidative damage related to atherosclerosis and aging [15–17]. We previously investigated the relation between UA and coronary heart disease (CHD) in the Atherosclerosis Risk in Communities (ARIC) Study and found that UA was not an independent risk factor for CHD [12]. Prior studies of the relation between UA and stroke incidence have been inconsistent [4–10,18,19]. Some studies reported a positive independent relation between UA and stroke [4–8]; others demonstrated that UA did not relate significantly to stroke occurrence [9,10,18,19].
402
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407
Diuretics are known to increase serum UA. If UA itself is harmful, the relation between UA and stroke should be similar between diuretic users and non-users. However, if elevated UA is simply “marking” subjects with other, causal risk factors or part of a compensatory mechanism to oxidative damage, the relation between UA and stroke might differ between diuretic users and non-users. Reyes has proposed a compensatory mechanism hypothesis that elevation in serum UA caused by drugs that reduce its renal excretion, such as diuretics, may actually be beneficial to cardiovascular prognosis [16,17]. Therefore, we believed that analyzing the relation between UA and ischemic stroke separately for diuretic users and non-users may help address whether UA itself is harmful to ischemic stroke. However, there are no prior studies of the effect of serum UA on ischemic stroke in diuretic users and non-users. Therefore, in this study, we investigated the relation between UA and ischemic stroke incidence in the middleaged general population. We also investigated whether the association varied by use of diuretics to help clarify whether UA itself increases ischemic stroke risk.
2. Methods 2.1. Study design and subjects The Atherosclerosis Risk in Communities Study is a multi-center prospective cohort study investigating the natural history of atherosclerotic disease in four US communities: Forsyth County, NC; Jackson, MS; Washington County, MD; the northwest suburbs of Minneapolis, MN [20]. The cohort comprised 15,792 men and women aged 45–64 years who were selected by probability list or area sampling. Only African Americans were recruited in the Jackson study center. From 1987 to 1989, a baseline home interview assessed participants’ sociodemographic characteristics, smoking and alcohol-drinking habits, medication use, reproductive history, and personal history of diseases. A clinical examination included measurement of various risk factors and cardiovascular conditions, B-mode ultrasound examinations of the carotid and popliteal arteries, and a 12-lead electrocardiogram (ECG). 2.2. Baseline examination After an overnight fast, blood was drawn from the antecubital vein of seated participants into vacuum tubes containing sodium citrate for hemostatic factors, EDTA for lipids, and serum separator gel for chemistries including UA. The tubes were centrifuged at 3000 × g for 10 min at 4 ◦ C. Plasma and serum aliquots were frozen at −70 ◦ C until analysis a few weeks later. UA was measured by the Uricase method. The reliability coefficient of UA, assessed by repeated measurements taken at least one week apart in 40 subjects (10
per ARIC center), was 0.91, and within-person variability was 7.2% [21]. High-density lipoprotein (HDL) cholesterol [22] was measured after dextran–magnesium precipitation of non-HDL lipoproteins. Details also have been reported for centralized measurement of von Willebrand factor (vWF) and serum creatinine and albumin [23–26]. Estimates of intraindividual variability in blood measurements have been reported [27,28]. Smoking status (current, former, or never smoking) and usual ethanol intake were estimated from interview [29]. Subjects were asked whether they currently drank alcoholic beverages and, if not, whether they had done so in the past. Current drinkers were asked how often they usually drank wine, beer, or hard liquor. In calculating the amount of alcohol consumed (in grams per week), it was assumed that 4 oz of wine contains 10.8 g, 12 oz of beer contains 13.2 g, and 1.5 oz of liquor contains 15.1 g of ethanol [30]. Sitting blood pressures (BP) were measured three times using a randomzero sphygmomanometer after 5 min rest [31]. The mean of the last two measurements was used for analysis. Use of anti-hypertensive medications within the past two weeks of baseline interview was self-reported [29]. Prevalent diabetes mellitus was defined as a fasting glucose of >126 mg/dL, or a reported history of physician-diagnosed diabetes, or current use of diabetes medication. For anthropometric measurements participants wore only light clothing. Body weight was measured to the nearest pound and height was measured to the nearest centimeter. Body mass index (BMI) was computed as weight (kg) divided by height squared (m2 ). Waist-to-hip ratio (WHR), an indicator of fat distribution, was computed as the circumference of the waist (umbilical level) in centimeter divided by the maximum circumference of the hips in centimeter. A 12-lead ECG [32] was used to define left ventricular hypertrophy (LVH), using the Cornell score [33]. Preexisting CHD at baseline was defined by self-reported prior physician diagnosis of myocardial infarction (MI) or coronary revascularization, or prevalent MI by ARIC ECG [32]. Pre-existing stroke was defined by any self-reported prior physician diagnosis of stroke. The ankle/brachial index (ABI) was defined as the ratio of ankle to brachial systolic BP, and peripheral arterial disease (PAD) was defined as ABI ≤ 0.90 for men and ≤0.85 for women. ARIC’s ultrasound measurements were based on the technique validated by Pignoli et al. [34], using a scanning protocol common to the four field centers [35,36], and standardized central reading of scans [37]. Analyses were based on mean intima-media thickness (IMT) of the far wall for 1-cm lengths of the carotid bifurcation and the internal and common carotid, right and left, adjusted for site-specific reader differences and downward measurement drifts in IMT over the baseline visit, with the means at missing sites imputed by maximal likelihood methods [38]. The means at the six sites were combined in an unweighted average to produce an overall mean IMT. In some analyses, extreme IMT was defined as 1.0 mm.
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407
We calculated 10-year probabilities of ischemic stroke using a previously reported prediction model [39]. To calculate the probabilities, we used sex-specific models involving age, race, smoking status, history of diabetes, anti-hypertensive medication, ECG LVH, systolic BP, BMI, albumin, von Willebrand factor, WHR, ethanol intake, low level of HDL cholesterol, carotid IMT, and PAD [39]. 2.3. Ascertainment of incident events Stroke incidence was ascertained by contacting participants annually, identifying hospitalizations during the previous year, and by surveying discharge lists from local hospitals and death certificates from state vital statistics offices for potential cerebrovascular events [29,40,41]. Potential strokes were found by selecting records with ICD-9 CM discharge codes 430–438 or with mention of stroke or neuroimaging in the discharge summary. For these, hospital records were copied and abstracted by a trained nurse. Each eligible case was classified by computer algorithm and by a physician reviewer, according to criteria adapted from the National Survey of Stroke [42]. Differences in diagnosis were adjudicated by another reviewer. Details on quality assurance for ascertainment and classification of events are presented elsewhere [41]. The criteria for classification were based on combinations of symptom type, duration, and severity, results of neuroimaging and other diagnostic procedures, and autopsy evidence when available [40,41]. A stroke was classified as ischemic, which we have taken as the event of interest, if a brain CT or MR revealed acute infarction or showed no evidence of hemorrhage [43]. 2.4. Data analysis A total of 15,792 ARIC subjects participated in the baseline ARIC exam. Of these, 13,413 were free of recognized stroke or CHD at baseline and had a baseline UA measurement. Of the original sample, we excluded 329 participants who had history of stroke, 1028 participants who had a history of CHD, 119 subjects without UA data, 170 for taking allopurinol or other uricosuric agents, and 39 for creatinine values ≥2.0 mg/dL (these four conditions were not mutually exclusive). We further excluded, due to small numbers, non-White and non-African American subjects in Minneapolis and Washington County (N = 42). Subjects who failed to fast for blood drawing and subjects who did not have complete information on confounding factors (N = 652) were also excluded. The excluded participants were 0.6 years younger, 31.9% more likely to be Black, and 24.6% more likely to have diabetes compared with who had complete information on confounding variables (N = 13,413). The association of baseline UA with other risk factors was assessed using analysis of variance or χ2 -test, as appropriate. The association between UA and ischemic stroke incidence was estimated from Cox proportional hazard models. Relative hazards (RH) for stroke were computed across
403
approximate quartiles of UA using the bottom quartile as the reference category. In Model 1, we adjusted for age, race (African American, White), and education (less than high school graduate, high school graduate, greater than high school graduate). In Model 2, we adjusted for major risk factors for ischemic stroke in our previous study, i.e., systolic BP (continuous variable), diabetes mellitus (yes, no), anti-hypertensive medication (yes, no), cigarette smoking (current, former, never), ethanol intake (continuous variable), serum albumin, vWF, BMI, WHR, and low HDL cholesterol (cutpoints for “low” being 60 mg/dL in women and 50 mg/dL in men) [39]. In a supplemental analysis, we further adjusted for ECG LVH, carotid IMT, and PAD. Although these variables are predictors of ischemic stroke [39], we did not include these variables in the main model (Model 2), as we were concerned to do so would be over adjustment, i.e., these variables might be on the causal pathway or might be surrogates of variables on the causal pathway. In supplemental analyses, we used Model 2 for adjustment. P-values for tests of linear trends were calculated by treating the categories for quartile of UA as an ordinal variable. Differences with a two-tailed P-value < 0.05 were considered statistically significant. All statistical analyses were performed using SAS software, Version 8.2 (SAS Institute, Cary, NC).
3. Results 3.1. Uric acid and other cardiovascular risk factors The mean value ± S.D. of UA was 5.97 ± 1.52 mg/dL (range 0.5–15.9 mg/dL). Table 1 shows the association between UA and other potential risk factors for stroke. Mean values of age, systolic BP, ethanol intake, serum albumin, vWF, WHR, and BMI were higher in higher UA categories. Similarly, the proportions of subjects who were men, diabetic, or ever smoked; who had less than a high school education, used anti-hypertensive medications, or had low HDL cholesterol; or who had ECG LVH, increased IMT, or PAD were increased across UA quartiles. Consequently, the predicted 10-year ischemic stroke probability was higher in the higher UA categories. 3.2. Uric acid and ischemic stroke The mean duration of follow-up was 12.6 years (maximum 15.1 years), during which 381 incident ischemic strokes occurred. Table 2 shows the RH for ischemic stroke according to approximate quartiles of baseline UA. Serum UA was independently and positively related to ischemic stroke incidence when we adjusted for age, sex, race, and education (RH in the highest quartile: 1.74; 95% confidence interval (CI): 1.27–2.38), compared with the first quartile. However, the positive relation was weakened (RH in the highest quartile: 1.25; 95% CI: 0.91–1.73) when additionally adjusted
404
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407
Table 1 Prevalences or mean values of baseline participant characteristics according to approximate quartiles of serum uric acid: the Atherosclerosis Risk in Communities (ARIC) Study, 1987–1989 Characteristic
P-valuea
Serum uric acid quartile (mg/dL) ≤4.8
4.9–5.8
5.9–6.8
≥6.9
3534 53.0 86.6 20.6 116.4 7.0
2852 53.9 66.3 23.8 119.5 8.5
3792 54.4 44.7 24.8 121.5 10.0
3235 54.6 31.6 30.6 125.2 12.6
Smoking Current (%) Former (%) Never (%)
27.2 22.4 50.3
27.4 28.5 44.1
24.6 34.8 40.5
22.4 41.1 36.4
<0.001
Education
18.6 36.4 45.0
20.3 34.9 44.7
22.3 31.3 46.4
26.2 29.6 44.0
<0.001
16.1 50.9 24.8 3.85 111.0
21.3 60.8 32.3 3.86 114.7
26.7 69.8 46.8 3.88 116.7
43.2 77.0 63.6 3.91 123.7
<0.001 <0.001 <0.001 <0.001 <0.001
Waist-to-hip ratio Body mass index (kg/m2 )
0.87 25.2
0.91 27.0
0.94 28.3
0.96 29.7
<0.001 <0.001
LVH (%) PAD (%)c Elevated carotid IMT (%)d
0.8 1.8 4.5
1.5 2.1 5.2
1.8 1.8 6.8
3.2 3.1 9.5
<0.001 <0.001 <0.001
10-Year stroke probability (%)
2.0
2.7
3.4
4.6
<0.001
N Age Sex (women, %) Race (African American, %) Systolic blood pressure (mmHg) Diabetes (yes, %)
Use of anti-hypertensive medication (%) Low HDL cholesterol (%)b Ethanol intake (g/week) Serum albumin (g/dL) von Willebrand Factor (%)
<0.001 <0.001 <0.001 <0.001 <0.001
N: number of subjects; LVH: left ventricular hypertrophy on electrocardiogram; PAD: peripheral artery disease; IMT: intima-media thickness. a P-value from a test of equality of all means (analysis of variance) or a contingency table (χ2 with 1 d.f. or 2 d.f.). b Cutpoints for “low” were 60 mg/dL in women and 50 mg/dL in men. c Ankle brachial pressure index ≤ 0.90 for men and ≤0.85 for women. d IMT ≥ 1.0 mm.
Table 2 Relative hazards (95% confidence interval) for ischemic stroke incidence according to quartile of serum uric acid at baseline, ARIC, 1987–2001 Serum uric acid quartile (mg/dL)
Total N Events Model 1 Model 2 Diuretic non-user N Events Model 1 Model 2 Diuretic user N Events Model 1 Model 2
P for linear trends
≤4.8
4.9–5.8
5.9–6.8
≥6.9
3534
2852
3792
3235
59 0.96 (0.68–1.37) 0.86 (0.60–1.23)
114 1.30 (0.95–1.77) 1.09 (0.79–1.49)
141 1.74 (1.27–2.38) 1.25 (0.91–1.73)
2534
3183
2303
45 0.99 (0.66–1.49) 0.92 (0.62–1.39)
85 1.41 (0.98–2.04) 1.23 (0.84–1.78)
86 1.89 (1.29–2.76) 1.49 (1.00–2.23)
318
609
932
14 0.72 (0.35–1.48) 0.60 (0.29–1.25)
29 0.69 (0.37–1.28) 0.71 (0.38–1.33)
55 0.72 (0.40–1.30) 0.73 (0.40–1.34)
67 1.00 1.00 3243 51 1.00 1.00 291 16 1.00 1.00
<0.01 0.06
<0.01 0.02
0.46 0.62
Model 1: adjusted for age, sex, race, and education; Model 2: Model 1 + systolic blood pressure, diabetes mellitus, anti-hypertensive medication, cigarette smoking status, ethanol intake, serum albumin, von Willebrand factor, BMI, WHR, and low HDL cholesterol; N: number of subjects; BMI: body mass index; WHR: waist-to-hip ratio.
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407
for systolic BP, diabetes mellitus, anti-hypertensive medication, cigarette smoking, ethanol intake, albumin, vWF, BMI, WHR, and low HDL cholesterol (Model 2). Because diuretics affect serum UA level, we analyzed the relation separately for diuretic users and non-users. In Model 2, there remained a positive statistically significant relation between UA and ischemic stroke in subjects not using diuretics (RH = 1.49 in fourth quartile; 95% CI: 1.00–2.23) (P for linear trend: 0.02). The RH per 1 mg/dL greater UA level was 1.10 (95% CI: 1.01–1.21). There was no significant relation between UA and ischemic stroke in diuretic users (RH = 0.73 in fourth quartile; 95% CI: 0.40–1.34) (P for linear trend: 0.62). The interaction between UA and diuretic use for stroke incidence was marginally significant (P for interaction: 0.08, using Model 2). These results were unchanged if we also adjusted for ECG LVH, carotid IMT, and PAD. Removal of serum albumin and vWF from Model 2 also had no appreciable impact. 3.3. Supplemental analyses As shown in Table 3, we conducted a subgroup analysis among subjects not taking diuretics, stratifying on age (age < 55 years and ≥55 years at baseline), sex (men and women), race (African American and White), 10-year predicted ischemic stroke risk (<5% and ≥5%), and sub-clinical conditions (LVH, PAD, or extreme IMT). There were no significant interactions of UA with the stratifying variables (all P > 0.10). However, the magnitude of the RH of ischemic stroke in the highest UA quartile was greater in men than women, and in subjects with versus without sub-clinical conditions.
405
4. Discussion In this prospective cohort study, we demonstrated an independent positive moderate association between UA and ischemic stroke incidence restricted to those not using diuretics. A motivation of this study was to clarify whether UA itself is a risk factor for ischemic stroke by analyzing the relation between UA and ischemic stroke separately for diuretic users and non-users. Some mechanisms have been proposed to explain a potential relation between UA and atherosclerosis. First, the relation between UA and atherosclerosis may reflect risk factor clustering such as metabolic syndrome, often due to insulin resistance, or a pathogenic role of UA in hypertension [1,2,11,44]. However, when we adjusted for other risk factors, even though the RH was somewhat attenuated, we still found a positive relation between UA and ischemic stroke among subjects not taking diuretics. Thus, risk factor clustering did not fully explain the positive relation between UA and ischemic stroke, and UA was an independent predictor of ischemic stroke among subjects not taking diuretics. Second, UA may have a direct ability to induce inflammatory and vascular mechanisms that may contribute to the development of cardiovascular disease [1]. However, if UA itself affects atherosclerosis, one might hypothesize that UA would be associated with ischemic stroke risk even in diuretic users. Yet, it is possible that the duration of exposure of arteries to high UA is less among diuretic users than those with physiologically induced hyperuricemia. Third, since UA has strong antioxidant properties, and atherosclerosis has been linked to increased oxidative stress, UA could be a compensatory
Table 3 Stratum-specific relative hazards (95% confidence interval) for ischemic stroke incidence in subjects not using diuretics (N = 11,263) according to quartile of serum uric acid at baseline, ARIC, 1987-2001 N
Events
Serum uric acid quartile (mg/dL) ≤4.8
4.9–5.8
5.9–6.8
≥6.9
P for interaction
Older (age 55 years) Younger (age < 55 years)
4970 6293
179 88
1.00 1.00
0.79 (0.48–1.29) 1.23 (0.59–2.60)
1.05 (0.67–1.66) 1.68 (0.84–3.38)
1.64 (1.03–2.63) 1.22 (0.56–2.66)
0.64
Men Women
5125 6138
149 118
1.00 1.00
1.01 (0.48–2.13) 0.85 (0.51–1.41)
1.30 (0.67–2.53) 1.22 (0.75–1.99)
1.63 (0.83–3.19) 1.27 (0.70–2.30)
0.51
White African American
8733 2530
164 103
1.00 1.00
0.81 (0.48–1.35) 1.14 (0.58–2.27)
1.03 (0.64–1.66) 1.53 (0.82–2.84)
1.27 (0.76–2.12) 1.79 (0.93–3.45)
0.88
10-Year ischemic stroke prediction 5%a 10-Year ischemic stroke prediction < 5%a
1301 8794
111 124
1.00 1.00
0.95 (0.47–1.93) 0.78 (0.45–1.36)
1.64 (0.90–3.01) 0.85 (0.49–1.47)
1.29 (0.67–2.48) 1.49 (0.85–2.65)
0.12
With sub-clinical conditionb Without sub-clinical conditionb
943 9178
61 175
1.00 1.00
0.76 (0.29–2.01) 0.84 (0.51–1.36)
1.32 (0.57–3.06) 1.05 (0.67–1.64)
2.46 (1.08–5.62) 1.04 (0.63–1.71)
0.16
Adjusted for age, sex, race, education, systolic blood pressure, diabetes mellitus, anti-hypertensive medication, cigarette smoking status, ethanol intake, serum albumin, von Willebrand factor, BMI, WHR, and low HDL cholesterol. N: number of subjects; BMI: body mass index; WHR: waist-to-hip ratio; sub-clinical condition: having increased intima-media thickness, electrocardiogram left ventricular hypertrophy, or peripheral artery disease. a Excludes 1168 subjects who did not have information on variables for calculating 10-year ischemic stroke probability. Among 1168 subjects, 7, 673, 222, 20, and 246 did not have information on smoking status, intima-media thickness, peripheral artery disease, HDL cholesterol, and electrocardiographic left ventricular hypertrophy, respectively. b Excludes 1142 subjects who had incomplete information on sub-clinical conditions. Among 1142 subjects, 674, 249, and 219 did not have information on intima-media thickness, electrocardiographic left ventricular hypertrophy, and peripheral artery disease, respectively.
406
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407
mechanism to protect the body from pro-oxidants [15,16]. This hypothesis could explain differences between diuretic users and subjects without diuretics: i.e., increased UA due to oxidative stress could be harmful, whereas diuretic-induced hyperuricemia does not cause ischemic stroke. This hypothesis is also consistent with the finding from randomized controlled trials that diuretics and other anti-hypertensive drugs are equivalent in reducing stroke occurrence [45,46]. If the diuretic-induced elevation of UA played an important role in stroke incidence, diuretics would not be expected to reduce stroke risk equivalently to other anti-hypertensives. Of course, there is a possibility that diuretics have a greater efficacy for stroke prevention than other anti-hypertensive drugs, but this is offset by diuretic-induced high UA; however, we consider this possibility unlikely. There is also the possibility that the measured UA level of many diuretic users does not reflect lifetime values, and hence was misclassified; such misclassification would have tended to bias the association we observed toward the null hypothesis. Because we cannot establish these mechanisms in an observational study, additional clinical trials might be needed to clarify whether UA itself plays a causal role in ischemic stroke incidence. Previous prospective studies focused on the relation between UA and stroke, but yielded inconsistent findings [4–10,18,19]. Most positive reports, but not all [10,19], involved high-risk subjects, i.e., hospital-based subjects [7], subjects with diabetes [8], an older general population [5], or older subjects with isolated systolic hypertension (ISH) [6]. Studies finding no association examined lower risk subjects, e.g., a Japanese middle-aged general population [18]. Accordingly, we analyzed higher risk and lower risk subjects, separately. Although we found that the relation between UA and ischemic stroke was somewhat stronger in subjects with sub-clinical conditions and higher predicted stroke probability, no statistically significant interactions were observed. Therefore, our data did not support the hypothesis that the relation between UA and stroke is stronger in high-risk subjects than in lower risk subjects, but power was limited to detect even quite substantial differences. Further experimental or epidemiological data might be needed to clarify whether there is an interaction between UA and background risk of stroke. Strengths of our study included its large, biracial population-based cohort, its careful assessment of stroke incidence over an extended follow-up period, and its ability to control for many stroke risk factors. Our study had some limitations. First, we had only a single measurement of UA. Second, since we did not measure indicators of oxidative stress, we could not evaluate whether oxidative stress-induced UA increase was harmful.
5. Conclusions In conclusion, UA was independently and positively associated with ischemic stroke incidence among subjects not
using diuretics. However, we found no significant relation between UA and ischemic stroke among diuretic users. This finding suggests that UA is independent predictor of ischemic stroke among subjects not using diuretics, but that elevated UA itself may not cause ischemic stroke. Acknowledgements Dr. Hozawa was supported by a grant (14010301) from the Japan Society for the Promotion of Science Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan and the Banyu Fellowship Program sponsored by Banyu Life Science Foundation International. The Atherosclerosis Risk in Communities Study is carried out as a collaborative study supported by National Heart, Lung, and Blood Institute contracts N01-HC-55015, N01-HC-55016, N01-HC-55018, N01-HC-55019, N01-HC55020, N01-HC-55021, and N01-HC-55022. The authors thank the staff and participants of the ARIC study for their important contributions. References [1] 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. [2] Rich MW. Uric acid: is it a risk factor for cardiovascular disease? Am J Cardiol 2000;85:1018–21. [3] Brand FN, McGee DL, Kannel WB, Stokes 3rd J, Castelli WP. Hyperuricemia as a risk factor of coronary heart disease: the Framingham study. Am J Epidemiol 1985;121:11–8. [4] Tomita M, Mizuno S, Yamanaka H, et al. Does hyperuricemia affect mortality? A prospective cohort study of Japanese male workers. J Epidemiol 2000;10:403–9. [5] Mazza A, Pessina AC, Pavei A, et al. Predictors of stroke mortality in elderly people from the general population. The CArdiovascular STudy in the ELderly. Eur J Epidemiol 2001;17:1097–104. [6] Wang JG, Staessen JA, Fagard RH, et al. Prognostic significance of serum creatinine and uric acid in older Chinese patients with isolated systolic hypertension. Hypertension 2001;37:1069–74. [7] Longo-Mbenza B, Luila EL, Mbete P, Vita EK. Is hyperuricemia a risk factor of stroke and coronary heart disease among Africans? Int J Cardiol 1999;71:17–22. [8] 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. [9] Goldberg RJ, Burchfiel CM, Benfante R, et al. Lifestyle and biologic factors associated with atherosclerotic disease in middle-aged men. 20-Year findings from the Honolulu Heart Program. Arch Intern Med 1995;155:686–94. [10] Franse LV, Pahor M, Di Bari M, et al. Serum uric acid, diuretic treatment and risk of cardiovascular events in the Systolic Hypertension in the Elderly Program (SHEP). J Hypertens 2000;18:1149–54. [11] 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. [12] Moriarity JT, Folsom AR, Iribarren C, Nieto FJ, Rosamond WD. Serum uric acid and risk of coronary heart disease: Atherosclerosis Risk in Communities (ARIC) Study. Ann Epidemiol 2000;10:136–43.
A. Hozawa et al. / Atherosclerosis 187 (2006) 401–407 [13] Yusuf S, Bosch J. Urate levels as a predictor of cardiac deaths: causal relation or mere association? Eur Heart J 2002;23:760–1. [14] Culleton BF, Larson MG, Kannel WB, Levy D. Serum uric acid and risk for cardiovascular disease and death: the Framingham Heart Study. Ann Intern Med 1999;131:7–13. [15] Nieto FJ, Iribarren C, Gross MD, Comstock GW, Cutler RG. Uric acid and serum antioxidant capacity: a reaction to atherosclerosis? Atherosclerosis 2000;148:131–9. [16] Reyes AJ. Cardiovascular drugs and serum uric acid. Cardiovasc Drugs Ther 2003;17:397–414. [17] Reyes AJ, Leary WP. The increase in serum uric acid induced by diuretics could be beneficial to cardiovascular prognosis in hypertension: a hypothesis. J Hypertens 2003;21:1775–7. [18] Sakata K, Hashimoto T, Ueshima H, Okayama A, NIPPON DATA 80 Research Group. Absence of an association between serum uric acid and mortality from cardiovascular disease: NIPPON DATA 80, 1980–1994. National Integrated Projects for Prospective Observation of Non-communicable Diseases and its Trend in the Aged. Eur J Epidemiol 2001;17:461–8. [19] De Leeuw PW, Thijs L, Birkenhager WH, et al., Systolic Hypertension in Europe (Syst-Eur) Trial Investigators. Prognostic significance of renal function in elderly patients with isolated systolic hypertension: results from the Syst-Eur trial. J Am Soc Nephrol 2002;13:2213–22. [20] The ARIC Investigators. The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. Am J Epidemiol 1989;129:687–702. [21] Eckfeldt JH, Chambless LE, Shen YL. Short-term, within-person variability in clinical chemistry test results. Experience from the Atherosclerosis Risk in Communities Study. Arch Pathol Lab Med 1994;118:496–500. [22] Warnick GR, Benderson JM, Albers JJ. Quantitation of high-densitylipoprotein subclasses after separation by dextran sulfate and Mg2+ precipitation. Clin Chem 1982;28:1574. [23] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 9: hemostasis determinations, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [24] Wu KK, Papp AC, Patsch W, et al. ARIC hemostasis study. II. Organizational plan and feasibility study. Thromb Haemost 1990;64:521–5. [25] Sharrett AR. The Atherosclerosis Risk in Communities (ARIC) Study: introduction and objectives of the hemostasis component. Ann Epidemiol 1992;2:467–9. [26] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 10: clinical chemistry determinations, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [27] Chambless LE, McMahon RP, Brown SA, et al. Short-term intraindividual variability in lipoprotein measurements: the Atherosclerosis Risk in Communities (ARIC) Study. Am J Epidemiol 1992;136:1069–81. [28] Chambless LE, McMahon R, Wu K, et al. Short-term intraindividual variability in hemostasis factors: the ARIC Study. Ann Epidemiol 1992;2:723–33. [29] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 2: cohort component procedures, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [30] Fuchs FD, Chambless LE, Whelton PK, Nieto FJ, Heiss G. Alcohol consumption and the incidence of hypertension: the Atherosclerosis Risk in Communities Study. Hypertension 2001;37:1242–50.
407
[31] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 11: sitting blood pressure, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [32] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities Study (ARIC). Operations manual no. 5: electrocardiography, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [33] Crow RS, Prineas RJ, Rautaharju P, Hannan P, Liebson PR. Relation between electrocardiography and echocardiography for left ventricular mass in mild systolic hypertension (results from Treatment of Mild Hypertension Study). Am J Cardiol 1995;75:1233–8. [34] Pignoli P, Tremoli E, Poli A, Oreste P, Paoletti R. Intimal plus medial thickness of the arterial wall: a direct measurement with ultrasound imaging. Circulation 1986;74:1399–406. [35] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 6A: ultrasound assessment—part A, ultrasound scanning, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [36] Bond MG, Barnes RW, Riley WA, et al., the ARIC Study Group. High-resolution B-mode ultrasound scanning methods in the Atherosclerosis Risk in Communities Study (ARIC). J Neuroimaging 1991;1:68–73. [37] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 6B: ultrasound assessment—part A, ultrasound reading, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [38] Chambless LE, Heiss G, Folsom AR, et al. Association of coronary heart disease incidence with carotid arterial wall thickness and major risk factors: the Atherosclerosis Risk in Communities (ARIC) Study, 1987–1993. Am J Epidemiol 1997;146:483–894. [39] Chambless LE, Heiss G, Shahar E, Earp MJ, Toole J. Prediction of ischemic stroke risk in the Atherosclerosis Risk in Communities Study. Am J Epidemiol 2004;160:259–69. [40] National Heart, Lung, and Blood Institute. Atherosclerosis Risk in Communities (ARIC) Study. Operations manual no. 3: surveillance component procedures, version 1.0. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina; 1987. [41] Rosamond WD, Folsom AR, Chambless LE, et al. Stroke incidence and survival among middle-aged adults: 9-year follow-up of the Atherosclerosis Risk in Communities (ARIC) cohort. Stroke 1999;30:736–43. [42] The National Survey of Stroke. National Institute of Neurological and Communicative Disorders and Stroke. Stroke 1981;12(2 Pt 2 Suppl. 1):I1–91. [43] Shahar E, Chambless LE, Rosamond WD, et al. Plasma lipid profile and incident ischemic stroke: the Atherosclerosis Risk in Communities (ARIC) study. Stroke 2003;34:623–33. [44] Klein BE, Klein R, Lee KE. Components of the metabolic syndrome and risk of cardiovascular disease and diabetes in Beaver Dam. Diabetes Care 2002;25:1790–4. [45] Psaty BM, Lumley T, Furberg CD, et al. Health outcomes associated with various antihypertensive therapies used as first-line agents: a network meta-analysis. JAMA 2003;289:2534–44. [46] ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial. Major outcomes in high-risk hypertensive patients randomized to angiotensin-converting enzyme inhibitor or calcium channel blocker vs diuretic: the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). JAMA 2002;288:2981–97.