●

hamymewRaac“qlatrd©

I

Utspdgpaeoetr

fscdeptutl

A

J-Shaped Mortality Relationship for Uric Acid in CKD

Mohamed E. Suliman, MD, PhD, Richard J. Johnson, MD, PhD, Elvia García-López, MD,A. Rashid Qureshi, MD, PhD, Hadi Molinaei, MD, Juan Jesús Carrero, PhD,

Olof Heimbürger, MD, PhD, Peter Bárány, MD, PhD, Jonas Axelsson, MD, PhD,Bengt Lindholm, MD, PhD, and Peter Stenvinkel, MD, PhD

Background: Hyperuricemia is a common feature in patients with chronic kidney disease (CKD). Hyperuricemiaas been associated with increased cardiovascular mortality in the general population, but less is known about thisssociation in patients with CKD. Methods: To explore possible associations of serum uric acid with all-causeortality and comorbidity in patients with CKD, we studied 294 incident patients with CKD stage 5 (185 men; age, 53 � 12

ears) starting renal replacement therapy with a median glomerular filtration rate of 6.4 mL/min/1.73 m2 (0.11L/s/1.73 m2; range, 0.8 to 14.3 mL/min/1.73 m2 [0.01 to 0.24 mL/s/1.73 m2]). Survival was determined from the day of

xamination and during a mean follow-up period of 27 months (range, 3 to 72 months); 94 patients died. Patientsere divided into 3 groups based on serum uric acid levels (low quintile, 3 middle quintiles, and high quintile).esults: In a nonadjusted analysis, patients in the high quintile, followed by patients in the low quintile, had greaterll-cause mortality compared with patients in the 3 middle quintiles (log-rank test chi-square, 6.8; P � 0.03). Afterdjusting for age, sex, glomerular filtration rate, cholesterol level, phosphate level, C-reactive protein level,ardiovascular disease, diabetes mellitus, diuretics, and allopurinol treatment, the association showed aJ-shaped” association with hazard ratios of 1.96 (confidence interval, 1.10 to 3.48; P � 0.02) for the highuintile and 1.42 (confidence interval, 0.76 to 2.66; P � not significant) for the low quintile. Moreover, uric acid

evels correlated positively with levels of triglycerides, phosphate, C-reactive protein, and intracellulardhesion molecule 1 and negatively with levels of calcium, high-density lipoprotein cholesterol, and apolipopro-ein A. Conclusion: Serum uric acid levels showed a J-shaped association with all-cause mortality, with the lowestisk in the 3 middle quintiles. Moreover, uric acid level was associated with calcium/phosphate metabolism,yslipidemia, and inflammation. Am J Kidney Dis 48:761-771.2006 by the National Kidney Foundation, Inc.

NDEX WORDS: Uric acid; chronic kidney disease (CKD); mortality; phosphate; dyslipidemia; inflammation.

thica

DoHHi

1

o

esaI

Rdk

RIC ACID IS GENERATED by xanthinedehydrogenase or xanthine oxidase during

he degradation of purines.1 Because of progres-ive loss of glomerular filtration rate (GFR),atients with chronic kidney disease (CKD) haveecreased renal clearance of uric acid and thusreater serum uric acid levels than the generalopulation. Approximately two thirds of uriccid is excreted by the kidney, and the rest isxcreted by the gastrointestinal tract; this routef elimination may be the major mechanism forxcretion in the setting of renal failure.2 In addi-ion, some uric acid is degraded in the body aftereaction with oxidants or peroxynitrite.3

The role of uric acid as an independent riskactor for cardiovascular risk has been controver-ial.4,5 Whereas several studies showed an asso-iation between uric acid and cardiovascularisease (CVD), as well as mortality, in the gen-ral population,6-10 this association was not inde-endent in some, but rather related to concomi-ant risk factors.6-8 Studies examining the role ofric acid on clinical outcome are complicated byhe frequent association of uric acid with estab-

ished cardiovascular risk factors, such as hyper-

merican Journal of Kidney Diseases, Vol 48, No 5 (November), 2

ension, obesity, dyslipidemia, hyperglycemia,yperinsulinemia, and decreased insulin sensitiv-ty.11,12 Moreover, other pathophysiologicalauses of atherosclerosis, such as inflammationnd oxidative stress,13,14 also are associated with

From the Divisions of Renal Medicine and Baxter Novum,epartment of Clinical Science, Intervention and Technol-gy, Karolinska Institutet, Karolinska University Hospitaluddinge, Stockholm, Sweden; and Division of Nephrology,ypertension, and Renal Transplantation, University of Flor-

da, Gainesville, FL.Received May 24, 2006; accepted in revised form August

5, 2006.Originally published online as doi:10.1053/j.ajkd.2006.08.019

n October 2, 2006.Support: None. Potential conflicts of interest: B.L. is

mployed by Baxter Healthcare, Chicago, IL. R.J.J. wasupported by NIH grants DK-52121 and HL-68607. R.J.J. islso a consultant for TAP pharmaceuticals and Nephromicsnc.

Address reprint requests to Peter Stenvinkel, MD, PhD,enal Medicine, K56, Karolinska University Hospital Hud-inge, 141 86 Stockholm, Sweden. E-mail: peter.stenvinkel@i.se

© 2006 by the National Kidney Foundation, Inc.0272-6386/06/4805-0008$32.00/0

doi:10.1053/j.ajkd.2006.08.019

006: pp 761-771 761

ibwoblf

aiaaaC

tSpp

2ymotabsapcytsc

ppn(pclpcaoicioaprid

tatobctwwt

B

ap2LsiMiihmermttgmCsi

N

otioemnMmpgw(d0cpel2sf

SULIMAN ET AL762

ncreased uric acid levels. Recently, a relationshipetween C-reactive protein (CRP) and uric acidas reported in patients with CKD stage 5.15 More-ver, because recent studies showed that phosphateinders had a lowering effect on serum uric acidevels in hemodialysis patients,16,17 this mayurther confound interpretations.

To date, only one previous study evaluated thessociation between uric acid level and mortalityn dialysis patients.18 In the present study, weim to investigate possible links of uric acid withll-cause mortality, lipid level, and phosphatend calcium metabolism in incident patients withKD stage 5 starting renal replacement therapy.

METHODS

The study protocol was approved by the Ethics Commit-ee of Karolinska University Hospital Huddinge, Stockholm,weden, and informed consent was obtained from eachatient. Patients were investigated as part of an ongoingrospective study.19

In the present study, post hoc analyses were performed in94 patients with CKD stage 5 (185 men; age, 53 � 12ears; range, 22 to 70 years) at a time when renal replace-ent therapy was planned to be initiated with a median GFR

f 6.4 mL/min (0.11 mL/s; range, 0.8 to 14.3 mL/min [0.01o 0.24 mL/s]), estimated by using the mean of creatininend urea clearances from a 24-hour urine collection. Fastinglood samples were obtained from 205 patients beforetarting dialysis treatment (median predialysis time, 21 days)nd 89 patients just after starting dialysis treatment (medianostdialysis time, 9 days; see Results). Study exclusionriteria were age younger than 18 years or older than 70ears, clinical signs of acute infection or liver disease at theime of evaluation, or unwillingness to participate in thetudy. Moreover, no patient with “acute” or “acute-on-hronic” renal failure was included.

Causes of CKD were chronic glomerulonephritis in 79atients (27%), diabetic nephropathy in 96 patients (32%),olycystic kidney disease in 32 patients (11%), interstitialephritis in 5 patients (2%), nephrosclerosis in 6 patients2%), and other or unknown in 76 patients (26%). Theresence of clinical CVD was defined by medical history,linical symptoms, and/or findings of cardiac, cerebrovascu-ar (stroke), and/or peripheral vascular disease. One hundredatients (34%) had a clinical history or signs of CVD,erebrovascular disease, and/or peripheral vascular diseaset the start of the study. Of these 100 patients, 59 patients hadne or more myocardial infarctions or clinical signs ofschemic heart disease (angina pectoris) or had undergoneoronary artery bypass surgery, 22 patients had peripheralschemic vascular disease, 19 patients had a history of stroker cerebral bleeding, and 5 patients had a history of an aorticneurysm. In this study, 161 patients (55%) were started oneritoneal dialysis treatment as an intention to treat, and theest were started on hemodialysis treatment. However, dur-ng follow-up, the majority of patients changed the initial

ialysis modality, and the majority underwent hemodialysis S

reatment. Two hundred fifty-two patients were administeredntihypertensive medications. Most were on double or tripleherapy, including angiotensin-converting enzyme inhibitorsr angiotensin-receptor blockers (n � 148), calcium channellockers (n � 118), and �-blockers (n � 165), with variousombinations of these drugs. Two hundred thirty-one pa-ients were administered phosphate binders, 191 patientsere administered vitamin D supplements, 253 patientsere administered diuretics, and 42 patients were adminis-

ered allopurinol.

iochemical MethodsSerum uric acid level was determined by means of a fully

utomated routine Trinder (AOX) method using uricase/eroxidase from Instrumentation Laboratory on a Konelab0XT centrifugal analyzer (Instrumentation Laboratory Inc,exington, MA). Absorbance was measured at 540 nm. High-ensitive CRP was measured by means of an enzyme-linkedmmunosorbent assay kit (R&D System Inc, Minneapolis,

N). Serum levels of tumor necrosis factor � (TNF-�) andnterleukin 6 (IL-6) were measured by using enzyme-linkedmmunosorbent assay kits obtained from Boehringer Mann-eim (Mannheim, Germany). Soluble intracellular adhesionolecule was determined by using a commercially available

nzyme-linked immunosorbent assay kit (R&D System Eu-ope Ltd, Abingdon, UK). Serum albumin (bromcresol purpleethod), creatinine, urea, calcium, phosphate, total choles-

erol, triglyceride, high-density lipoprotein (HDL) choles-erol, apolipoprotein A (apo A), apo B, hemoglobin, andlucose and urinary creatinine and urea were determined byeans of routine procedures at the Department of Clinicalhemistry, Karolinska University Hospital Huddinge. A

pecific radioimmunoassay kit was used to analyze plasmansulin (Pharmacia, Uppsala, Sweden).

utritional StatusSubjective global assessment (SGA) was used to evaluate

verall protein-energy nutritional status. SGAincluded 6 subjec-ive assessments; 3 based on the patient’s history of weight loss,ncidence of anorexia, and incidence of vomiting and 3 basedn the physician’s grading of muscle wasting, presence ofdema, and loss of subcutaneous fat. Based on these assess-ents, each patient was given a score of 1 to 4, indicating

ormal, mild, moderate, or severe malnutrition, respectively.alnutrition was defined as SGA score of 2 to 4, and patientseeting these criteria are grouped together as malnourished

atients. Body mass index was calculated as weight in kilo-rams/(height in meters)2. Fat body mass and lean body massere evaluated by using dual-energy X-ray absorptiometry

Lunar Corp, Madison, WI). The coefficient of variation forual-energy X-ray absorptiometry measurement is 2.0% �.7%. Total fat mass index and lean body mass index werealculated according to Kyle et al20 and expressed as kilogramser square meter. Protein intake was estimated from the proteinquivalent of nitrogen appearance (PNA), which was calcu-ated based on urea kinetic modeling using urea excretion in a4-hour urine collection. PNA calculation was normalized totandard body weight based on patient height, sex, age, andrame size using National Health and Nutrition Examination

urvey tables.21

O

u(tpf(

S

repCoWaaadacapppiCcwvrtptpaMtts

PA

tdut�tEdsm

padwrT

T

SABGCWCDHSSCITS

WUHPCCP

TTHAAHGISD

(cacmmmttmm

URIC ACID AND MORTALITY IN CKD 763

utcome AscertainmentSurvival was determined from the day of examination

ntil March 5, 2006, with a mean follow-up of 27 monthsrange, 3 to 72 months). Patients were censored at death,ransplantation, or when completing the 6-year follow-uperiod, with no loss of follow-up of any patient. Within theollow-up period, 94 patients (32%) died and 125 patients43%) underwent transplantation.

tatistical AnalysesAll values are expressed as mean � SD or median and

ange unless otherwise indicated. P less than 0.05 is consid-red statistically significant. Kolmogorov-Smirnov test waserformed to evaluate the normal distribution of variables.omparisons between 2 groups were assessed for continu-us variables by using Student unpaired t-test, Mann-hitney test, or chi-square test, as appropriate. Differences

mong 3 groups were analyzed by using analysis of vari-nce, 1-way analysis of variance, or Kruskal-Wallis test, asppropriate. Spearman rank correlation (�) was used toetermine correlations of uric acid levels with other vari-bles. For correlation of uric acid levels with phosphate,alcium, and calcium-phosphate (Ca � P) product values,nalysis of covariance was used to adjust for the use ofhosphate binder and vitamin D. Survival analyses wereerformed with the Kaplan-Meier survival curve or Coxroportional hazard model. Hazard ratios (HRs) for mortal-ty were determined by using univariate and multivariateox regression analysis and presented as HR and 95%onfidence interval (CI). Cox proportional hazards modelas used to examine effects of baseline and follow-upariables on outcome variables. Plots of log (�log [survivalate]) against log (survival time) were performed to establishhe validity of the proportionality assumption. Because theroportional hazards model assumes a constant HR overime, the assumption for each baseline covariate of the Coxroportional hazards model was tested before the finalnalysis by evaluating a time-covariate interaction term.aximum-likelihood test at P less than 0.05 was considered

o accept that the HR of the particular variable depended onime. Statistical analysis was performed using statisticaloftware SAS, version 9.1.4 (SAS Institute, Cary, NC).

RESULTS

atient Characteristics and Uriccid Concentrations

Baseline characteristics of the investigated pa-ients are listed in Table 1. Uric acid concentrationsid not follow a normal distribution, and medianric acid concentration was 7.0 mg/dL (range, 2.8o 14.8 mg/dL [416 �mol/L; range, 167 to 880mol/L]). The stated normal range for our labora-

ory is 2.6 to 8.1 mg/dL (155 to 482 �mol/L).ighty-nine patients in this study had been onialysis therapy for a median of 8 days before thetudy. However, serum uric acid, CRP, serum albu-

in, serum creatinine, and plasma folate levels; g

revalences of inflammation, malnutrition, CVD,nd diabetes mellitus (DM); and percentages ofrug users did not differ between the 89 patientsho had already started dialysis therapy and the

est of the patients in the study (data not shown).herefore, all patients were analyzed as one group.Serum uric acid levels were significantly

able 1. Baseline Characteristics of 294 Patients WithCKD Starting Renal Replacement Therapy

ex (men/women) 185/109ge (y) 53 � 12ody mass index (kg/m2) 25 � 4FR (mL/min/1.73 m2) 6.4 (0.8-14.3)VD (%) 34asting (SGA �1) (%) 30RP �10 mg/L (%) 35iabetes mellitus (%) 31ypertension (%) 88erum albumin (g/dL) 3.3 � 0.6erum creatinine (mg/dL) 8.1 � 2.9RP (mg/L) 5.1 (0.2-218)

L-6 (pg/mL) 6.4 (0.75-48)NF-� (pg/mL) 10.0 (3.1-206.0)oluble intracellular adhesionmolecule 1 (ng/mL) 248 (99-655)hite blood cells (109/L) 8.5 � 3.4ric acid (mg/dL) 7.0 (2.8-14.8)emoglobin (g/dL) 10.4 � 1.5hosphate (mg/dL) 5.9 � 1.5alcium (mg/dL) 10 � 1a � P (mg2/dL2) 60 � 18rotein intake (normalized PNA;g/kg body weight/d) 0.71 � 0.16

otal cholesterol (mg/dL) 205 � 58riglycerides (mg/dL) 186 � 106DL cholesterol (mg/dL) 50.27 � 23.2po A (mg/dL) 131 � 33po B (mg/dL) 105 � 34emoglobin A1c (%) 5.3 � 1.5lucose (mg/dL) 97 (54-510)

nsulin (�IU/mL) 16 (2-81)ystolic blood pressure (mm Hg) 151 � 25iastolic blood pressure (mm Hg) 88 � 13

NOTE. Values expressed as mean � SD, medianrange), or number of patients unless noted otherwise. Toonvert GFR in mL/min to mL/s, multiply by 0.01667;lbumin and hemoglobin in g/dL to g/L, multiply by 10;reatinine in mg/dL to �mol/L, multiply by 88.4; uric acid ing/dL to �mol/L, multiply by 59.48; phosphate in mg/dL tomol/L, multiply by 0.3229; calcium in mg/dL to mmol/L,ultiply by 0.2495; total and HDL cholesterol in mg/dL

o mmol/L, multiply by 0.02586; triglycerides in mg/dLo mmol/L, multiply by 0.0113; apo A and apo B in mg/dL tomol/L, multiply by 0.01; glucose in mg/dL to mmol/L,ultiply by 0.05551.

reater in men than women (7.5 � 2.0 versus

6�do�wv�ammm�Unbl

U

t(veaarsn2�

UI

ttP0c

(0a0gv1sunoal00trauCa

s in mg

SULIMAN ET AL764

.7 � 2.3 mg/dL [446 � 119 versus 399 � 137mol/L]; P � 0.001). Uric acid levels did notiffer significantly between patients with or with-ut CVD (7.1 � 2.2 versus 7.2 � 2.1 mg/dL [422

131 versus 428 � 125 �mol/L]) and betweenasted (SGA � 1) or not wasted (7.2 � 2.4ersus 7.1 � 2.1 mg/dL [428 � 143 versus 422

125 �mol/L]) patients. The difference in uriccid levels between patients with (CRP � 10g/L) and without inflammation (CRP � 10g/L; uric acid, 7.4 � 2.2 versus 7.1 � 2.1g/dL [440 � 131 versus 422 � 125 �mol/L]; P

0.09) did not reach statistical significance.ric acid level did not correlate with age, GFR,ormalized PNA, glucose level, insulin level, orody composition parameters (body mass index,ean body mass, and fat body mass).

ric Acid and Blood Pressure

Serum uric acid levels did not differ be-ween patients with and without hypertension7.2 � 2.1 versus 6.8 � 2.5 mg/dL [428 � 125ersus 404 � 149 �mol/L]). However, afterxcluding hypertensive patients administeredngiotensin-converting enzyme inhibitors andngiotensin-receptor blockers (n � 148), theemaining hypertensive patients (n � 104)howed a trend to greater uric acid levels thanormotensive patients (7.4 � 2.1 versus 6.8 �.4 mg/dL [440 � 125 versus 404 � 143

Fig 1. Relationships of serum uric acid levels withtarting renal replacement therapy. To convert uric acid

mol/L]; P � 0.07). n

ric Acid in Relation to Lipids, Phosphate, andnflammation Markers

Serum uric acid levels correlated positively withriglyceride level (� � 0.19; P � 0.01) and nega-ively (Fig 1) with HDL cholesterol (� � �0.23;

� 0.0001) and apo A levels (� � �0.23; P �.001). However, uric acid did not correlate signifi-antly with either total cholesterol or apo B level.

Moreover, uric acid level correlated positivelyFig 2) with plasma phosphate level (� � 0.25; P �.0001) and Ca � P product (� � 0.21; P � 0.001)nd negatively with calcium level (� � �0.13; P �.05). Users of phosphate binders had significantlyreater uric acid levels than nonusers (7.3 � 2.2ersus 6.6 � 2.0 mg/dL [434 � 131 versus 393 �19 �mol/L]; P � 0.05). However, there were noignificant differences in uric acid levels betweensers (7.2 � 2.1 mg/dL [428 � 125 �mol/L]) andonusers (7.2 � 2.0 mg/dL [428 � 119 �mol/L])f vitamin D. After adjusting for phosphate-bindernd vitamin D treatment, correlations of uric acidevels with plasma phosphate (� � 0.24; P �.0001) and Ca � P product values (� � 0.19; P �.001) were still significant; whereas the correla-ion of uric acid level with calcium level did noteach statistical significance (� � �0.11; P � 0.06)fter adjustment. Among inflammation markers,ric acid level was associated significantly withRP (� � 0.15; P � 0.05) and soluble intracellulardhesion molecule (� � 0.14; P � 0.05) levels, but

holesterol and apo A levels in 294 patients with CKD/dL to �mol/L, multiply by 59.48.

HDL c

ot IL-6 or TNF-� levels.

U

hc(vlt�1ll

U

tsn�Huemr

U

adw

ltaqwp�m�

vpam

URIC ACID AND MORTALITY IN CKD 765

ric Acid and Diabetes Mellitus

Of 91 patients (31%) with DM, 48 patientsad type 2. Uric acid levels did not differ signifi-antly between patients with and without DM7.2 � 2.2 versus 7.2 � 2.1 mg/dL [428 � 131ersus 428 � 125 �mol/L]). Moreover, uric acidevels did not differ significantly between pa-ients with type 1 (7.0 � 2.1 mg/dL [416 � 125mol/L]) and type 2 (7.5 � 2.3 mg/dL [446 �37 �mol/L]) DM. Uric acid level did not corre-ate with hemoglobin A1c, glucose, or insulinevels.

ric Acid and Use of Diuretics and Allopurinol

Eighty-six percent of patients were on diureticreatment. As expected, uric acid levels wereignificantly greater in users of diuretics thanonusers (7.3 � 2.2 versus 6.5 � 1.7 mg/dL [434

131 versus 387 � 101 �mol/L]; P � 0.02).owever, there was no significant difference inric acid levels between users (16%) and nonus-rs of allopurinol (7.5 � 2.0 versus 7.1 � 2.2g/dL [446 � 119 versus 422 � 131 �mol/L],

espectively).

ric Acid and Mortality

Figure 3 shows HRs of the 5 quintiles of uriccid levels in relation to mortality. Althoughifferences in mortality among the 5 quintiles

Fig 2. Associations of serum uric acid levels with palues (� � 0.21; P � 0.001) in 294 patients with CKhosphate-binder and vitamin D treatment, correlationsnd Ca � P product (� � 0.19; P � 0.001) were statisultiply by 59.48.

ere not significant, patients in the higher andqt

ower quintiles had greater mortality than pa-ients in the 3 middle quintiles. Therefore, indditional analyses, patients in the 3 middleuintiles were grouped together and all patientsere divided into 3 groups. Group I includesatients in the lower quintile (�5.3 mg/dL [�315mol/L]), group II includes patients in the 3iddle quintiles (5.3 to 8.9 mg/dL [315 to 529mol/L]), and group III includes patients in the

phosphate (� � 0.24; P � 0.001) and Ca � P productting renal replacement therapy. After adjustment forc acid level with phosphate level (� � 0.24; P < 0.0001)significant. To convert uric acid in mg/dL to �mol/L,

Fig 3. HRs for mortality of patients in each of 5

lasmaD star

of uritically

uintiles of serum uric levels. The middle quintile ishe reference (HR � 1).

uuwCidPt

tccBd

vilpa0Cptaaqa

SABLFGCWCDSSCITSHPCCPTTHAAHGSD

imimm

SULIMAN ET AL766

pper quintile (�8.9 mg/dL [�529 �mol/L]) ofric acid levels. Basal characteristics of patientsith CKD in the 3 groups are listed in Table 2. Theox proportional hazards model shows that the

mpact of uric acid on survival was not time depen-ent (proportionality test; Wald chi-square � 9.4;� not significant), suggesting that the propor-

ionality assumption was valid.Within the 6-year follow-up period, 94 pa-

ients (32%) died; 60 patients (64%) died ofardiovascular causes. Table 3 lists causes ofardiovascular and noncardiovascular deaths.ased on cause of death, serum uric acid levels

Table 2. Characteristics of Patients Wi

Gro

ex (men/women)ge (y)ody mass index (kg/m2)ean body mass index (kg/m2) 1at mass index (kg/m2)FR (mL/min/1.73 m2) 5.VD (%)asting (SGA �1) (%)RP �10 mg/L (%)iabetes mellitus (%)erum albumin (g/dL)erum creatinine (mg/dL)RP (mg/L) 4.

L-6 (pg/mL) 5.NF-� (pg/mL) 1oluble intracellular adhesion molecule 1 (ng/mL) 25emoglobin (g/dL) 1hosphate (mg/dL)alcium (mg/dL) 1a � P product (mg2/dL2)rotein intake (normalized PNA; g/kg body weight/d) 0otal cholesterol (mg/dL) 2riglycerides (mg/dL) 1DL cholesterol (mg/dL)po A (mg/dL) 1po B (mg/dL) 1emoglobin A1c (%)lucose (mg/dL) 1ystolic blood pressure (mm Hg) 1iastolic blood pressure (mm Hg)

NOTE. Values expressed as mean � SD, median (rangen mL/min to mL/s, multiply by 0.01667; albumin and hemog

ultiply by 88.4; uric acid in mg/dL to �mol/L, multiply by 5n mg/dL to mmol/L, multiply by 0.2495; total and HDL cho

g/dL to mmol/L, multiply by 0.0113; apo A and apo B inultiply by 0.05551.Abbreviation: NS, not significant.

id not differ significantly between nonsurvi- C

ors. In a nonadjusted analysis (Fig 4A), patientsn the high quintile, followed by patients in theow quintile, had greater all-cause mortality com-ared with patients in the 3 middle quintiles of uriccid levels (log-rank test chi-square � 6.8; P �.03). After adjusting for age, sex, GFR, DM,VD, serum cholesterol level, CRP level, plasmahosphate level, and diuretic and allopurinolreatment, the association between uric acid levelnd all-cause mortality showed a “J-shaped”ssociation (Fig 5). Compared with the 3 middleuintiles (Fig 4B), the higher quintile was associ-ted significantly with high mortality (HR, 1.96;

by Quintile of Serum Uric Acid Level

intile 1)Group II

(middle 3 quintiles) Group III (quintile 5) P

117/56 44/15 �0.00012 52 � 12 53 � 12 NS

25 � 4 25 � 5 NS.2 16.7 � 2.2 16.7 � 2.5 NS.0 7.2 � 3.6 7.9 � 3.8 NS

10.4) 6.7 (1.8-13.7) 6.6 (0.8-14.3) �0.0531 36 NS26 38 NS33 46 NS31 34 NS

.7 3.3 � 0.6 3.5 � 0.6 NS

.2 8.2 � 2.9 7.9 � 3.0 �0.05118) 4.8 (0.2-218) 7.4 (0.2-163) �0.0535) 6.5 (1.1-48) 6.1 (0.7-45) NS) 10 (3-206) 10 (3-28) NS

-593) 238 (99-551) 271 (148-655) �0.05.5 10.4 � 1.5 10.5 � 1.3 NS.5 6.2 � 1.9 6.5 � 1.9 �0.01.9 10.0 � 1.1 9.9 � 1.0 �0.056 61 � 19 63 � 17 �0.05.14 0.71 � 0.17 0.71 � 0.14 NS0 205 � 58 201 � 62 NS2 186 � 106 213 � 142 �0.057 50 � 27 42 � 15 �0.017 131 � 33 122 � 27 �0.011 106 � 35 104 � 33 NS.2 5.4 � 1.6 5.2 � 1.4 NS9 117 � 63 119 � 56 NS2 155 � 24 144 � 26 �0.012 90 � 13 84 � 14 �0.05

mber of patients unless noted otherwise. To convert GFRg/dL to g/L, multiply by 10; creatinine in mg/dL to �mol/L,

hosphate in mg/dL to mmol/L, multiply by 0.3229; calciuml in mg/dL to mmol/L, multiply by 0.02586; triglycerides into mmol/L, multiply by 0.01; glucose in mg/dL to mmol/L,

th CKD

up I (qu

24/3854 � 125 � 55.9 � 27.7 � 45 (1.9-

40343227

3.3 � 07.4 � 20 (0.2-9 (1.1-1 (5-701 (1350.4 � 15.3 � 10.4 � 055 � 1.69 � 001 � 559 � 658 � 242 � 301 � 3

5.1 � 113 � 445 � 286 � 1

), or nulobin in9.48; plesteromg/dL

I, 1.10 to 3.48; P � 0.02). Although the lower

qmcootswf2

sdw5(om9ircDaela

acai

o5gn

mtPot(ti[fi

C

O

mm

URIC ACID AND MORTALITY IN CKD 767

uintile also tended to be associated with highortality, this association did not reach statisti-

al significance (HR, 1.42; CI, 0.76 to 2.66). Thebservation period was calculated from the datef blood sampling. However, when the observa-ion period was calculated from the day of thetart of dialysis treatment, findings were similar,ith an HR of 1.94 (CI, 1.09 to 3.44; P � 0.02)

or high quintile and HR of 1.40 (CI, 0.75 to.60) for the low quintile.

DISCUSSION

In this study, we examine the relationship oferum uric acid level obtained at the initiation ofialysis treatment in patients with stage 5 CKDith subsequent all-cause mortality. In this study,0% of patients were classified as hyperuricemicuric acid � 7.0 mg/dL [�416 �mol/L] in menr �6.5 mg/dL [�387 �mol/L] in women). Theajor finding is that a serum uric acid level of

.0 mg/dL or greater (�535 �mol/L; correspond-ng to quintile 5) had nearly 2.0-fold increasedisk for mortality independent of many potentialonfounders, including age, sex, GFR, CVD,M, diuretic use, cholesterol level, CRP level,

nd phosphate level. However, there also wasvidence that a uric acid level of 5.2 mg/dL oress (�309 �mol/L; corresponding to quintile 1)

Table 3. Causes of Death as a Proportion of AllDeaths and Serum Uric Acid Level in Each Cause

of Death

Cause of DeathNo. of Patients

(%) Uric Acid (mg/dL)

ardiovascular death 60 (64) 6.7 (3.3-12.5)Myocardial infarction 29 (31)Thrombotic stroke 7 (8)Hemorrhagic stroke 8 (9)Sudden death 6 (6)Arrhythmia 1 (1)Mesenteric infarction 1 (1)Heart failure 8 (9)ther causes of deathSepsis/infection 4 (4) 7.1 (5.5-11.3)Cachexia 6 (6) 8.0 (4.6-13.6)Neoplasia 11 (12) 7.0 (3.9-9.8)Respiratory disease 3 (3) 7.0 (6.9-10.8)Others 10 (11) 7.4 (5.8-12.0)

NOTE. N � 94 deaths. Uric acid values expressed asedian (range). To convert uric acid in mg/dL to �mol/L,ultiply by 59.48.

lso conferred a 40% increased risk for mortality,t0

lthough this did not reach statistical signifi-ance. This J-shaped curve relating serum uriccid level to mortality was observed previouslyn both the dialysis18 and general populations.4

The first important finding in the study is thebservation that hyperuricemia was present in0% of patients; this percentage could even bereater because 16% of subjects were on allopuri-ol therapy. To the best of our knowledge, there

Fig 4. (A) Unadjusted and (B) adjusted all-causeortality in relation to serum uric acid levels in pa-

ients with CKD starting renal replacement therapy.atients were divided into 3 groups based on quintilesf uric acid levels. Group I (low) includes patients inhe lower quintile (<5.3 mg/dL [<315 �mol/L]), group IImiddle) includes patients in the 3 middle quintiles (5.3o 8.9 mg/dL [315 to 529 �mol/L]), and group III (high)ncludes patients in the upper quintiles (>8.9 mg/dL>529 �mol/L]) of uric acid. (B) Survival was adjustedor age, sex, GFR, cholesterol level, phosphate level,nflammation, CVD, DM, diuretic use, and allopurinol

reatment (likelihood ratio, chi-square � 53.2; P <.0001).

aa

aplilttssdTwcwpbc

hrltperafw

rdid2ptCtdcadmadrmdioov

ta2fnapmuociooTnCta

uftcau

arttlGC

SULIMAN ET AL768

re no reports of the prevalence of hyperuricemiat the onset of renal replacement therapy.

There are several potential mechanisms toccount for the increase in uric acid levels inatients with CKD. First, the increase in uric acidevels could be a secondary consequence of renalnsufficiency. Uric acid normally is excretedargely by the kidney, and despite a compensa-ory increase in excretion through the gastrointes-inal tract with progressive renal insufficiency,ome retention is expected to occur.2,22 In ourtudy, the majority of patients were administerediuretics, which also increase uric acid levels.here also is an association of hyperuricemiaith hypertension and metabolic syndrome, both

onditions with increased frequency in patientsith CKD. Finally, hyperuricemia also may com-licate lead-associated renal disease, which iseing recognized increasingly as a significantause of CKD.23

One also must consider the possibility thatyperuricemia itself may be a true risk factor forenal disease, and hence the increased preva-ence observed may be a consequence of the facthat patients with hyperuricemia may be morerone to develop CKD. There now is compellingvidence that hyperuricemia is an independentisk factor for CKD in the general population,2,25

nd in one study, risk for CKD was greater thanor the presence of proteinuria.26 Hyperuricemia

Fig 5. The J-shaped mortality relationship for uriccid levels in patients with CKD stage 5 starting renaleplacement therapy. Compared with the 3 middle quin-iles as a reference, HRs were 1.96 (CI, 1.10 to 3.48) forhe higher quintile and 1.42 (CI, 0.76 to 2.66) for theower quintile. The analysis was adjusted for age, sex,FR, cholesterol level, phosphate level, inflammation,VD, DM, diuretic use, and allopurinol use.

as reported to be an independent risk factor for h

enal progression in patients with glomerulariseases27-29 and also may confer increased riskn subjects with essential hypertension.30 In theays before uric acid–lowering therapies, nearly5% of subjects with gout developed CKD. Ex-erimental hyperuricemia in animals was showno not only cause CKD,31 but also accelerateKD from other causes.32 Additional evidence

hat hyperuricemia may have a role in kidneyisease was provided by a recent prospectivelinical trial in which allopurinol therapy wasssociated with a significant slowing of renalisease progression in subjects with hyperurice-ia and renal insufficiency.33 Additional studies

re necessary to corroborate these findings, butata suggest that uric acid potentially may have aole in the progression of kidney disease. Theechanism by which uric acid may cause renal

isease was shown to be caused by uric acid–nduced endothelial dysfunction,34,35 activationf the renin-angiotensin system,32 and the devel-pment of glomerular hypertension and renalasoconstriction.36

The second important finding in this study ishat subjects with markedly increased serum uriccid levels (�9.0 mg/dL [�535 �mol/L]) had a-fold increased risk for mortality after adjustingor numerous comorbidities. One possible expla-ation is that despite controlling for these vari-bles, uric acid level may correlate with theresence of other risk factors not evaluated inultivariate analysis. For example, increased

ric acid levels were observed in subjects withxidative stress,37 and thus the presence of in-reased uric acid levels may reflect this underly-ng risk factor.4,6 Patients in the highest quintilef uric acid levels were characterized by a clusterf powerful predictors of mortality (Table 2).hus, it is possible that uric acid should notecessarily be viewed as a causative factor forVD, but rather as a valuable biological marker

hat reflects and integrates different risk factorsnd their possible synergistic actions.

An alternative view point is that an elevatedric acid level itself may represent a true riskactor for CVD and mortality. As discussed,here is increasing evidence that uric acid mayause kidney disease, as well as hypertension38

nd metabolic syndrome.39 Experimental hyper-ricemia rapidly results in the development of

ypertension and vascular disease.4 Experimen-

ttheaaavaTai

aLrJhtmpcadbuieecfiifboua

istHtHnsdpwhg

iHstlrceTdudn

wmcpwstiplstut

apcmmsaaastmailphitau

URIC ACID AND MORTALITY IN CKD 769

al hyperuricemia secondary to fructose inges-ion also mediates obesity, insulin resistance,ypertriglyceridemia, and hypertension, and theseffects can be blocked by decreasing serum uriccid levels.39 Uric acid also was shown to medi-te pro-oxidative effects on vascular cells anddipocytes and can stimulate the production of aariety of inflammatory mediators, including CRPnd monocyte chemoattractant protein 1.35,40

hese effects result from the direct effect of uriccid on these cell types, as well as systemicnhibition of endothelial nitric oxide levels.39

The third major finding is the observation thatlow uric acid level (�5.2 mg/dL [�309 �mol/]) tended to be associated with greater mortality

ates compared with the middle 3 quintiles. This-shaped curve of mortality also was reported inemodialysis patients18 and the general popula-ion.4 Hsu et al18 reported that the association of

ortality with low uric acid level may indicateoor protein intake and wasting, as well as defi-iencies in intake of purines or nucleotides, whichre essential nutrients.41 However, our studyoes not show a difference in uric acid levelsetween wasted and not-wasted patients, andric acid level was not associated with proteinntake, estimated from normalized PNA. A differ-nce in dietary purine intake is a conceivablexplanation, but this was not assessed in theurrent study. The most likely explanation for thending that a low uric acid level also confers

ncreased risk for mortality is that uric acid mayunction as an antioxidant and, in particular, canlock peroxynitrite-mediated injury in a numberf experimental systems.42 Thus, a low serumric acid level may reflect inadequate protectiongainst oxidant-mediated stress.4

Several important associations were observedn this study. First, the presence of strong relation-hips between uric acid levels and certain fea-ures of metabolic syndrome were observed.ence, uric acid level was associated with serum

riglyceride level and inversely correlated withDL cholesterol and apo A levels. Conversely,o association was noted between uric acid anderum insulin or glucose levels. Again, this studyoes not distinguish whether this association is arimary or secondary phenomenon. Hence, itas suggested that the common association ofyperuricemia with metabolic syndrome in the

eneral population is caused by effects of hyper- g

nsulinemia to increase uric acid reabsorption.43

owever, this seems less likely in the currenttudy because GFR was seriously decreased andhere was no association of insulin with uric acidevels. An alternative possibility relates to aecent study in which decreasing uric acid levelsompletely corrected triglyceride levels in anxperimental model of metabolic syndrome.39

his suggests a role for uric acid in mediatingyslipidemia. However, in that study, decreasingric acid levels also were associated with aecrease in fasting insulin levels,39 which wasot observed in this analysis.In this study, uric acid levels also correlate

ith the presence of inflammatory mediators,ost notably CRP levels. An independent asso-

iation of inflammatory mediators also was re-orted in subjects with heart failure,44 patientsith uncomplicated and untreated hyperten-

ion,45 and patients with CKD not yet on dialysisherapy.15 Uric acid can stimulate CRP synthesisn human vascular cells, and the effect is moreotent than endotoxin or IL-6.35 Because serumevels of proinflammatory markers are one of thetrongest predictors of outcome in this popula-ion,46 this could provide a mechanism by whichric acid may be associated with increased mor-ality.

This study also shows that uric acid levels alsore associated with low calcium and elevatedhosphate levels. The explanation for this asso-iation is not evident. However, a link betweenetabolic syndrome (hypertension, dyslipide-ia, hyperglycemia, insulin resistance, and obe-

ity), disturbances in phosphate metabolism,47,48

nd hyperuricemia12,49,50 was proposed. Therelso is evidence in both experimental animalsnd humans that increased uric acid levels canuppress 1,25 dihydroxyvitamin D synthesishrough inhibition of 1-� hydroxylase,51-53 which

ight be expected to exacerbate renal failure–ssociated hypocalcemia. The association withncreased phosphate level is interesting, particu-arly because there is a report that decreasinghosphate levels with sevelamer, a nonabsorbedydrogel phosphate binder, results in a decreasen uric acid concentrations in hemodialysis pa-ients.16,17 Based on this observation, Garg etl16 hypothesized that sevelamer might bind toric acid or guanosine or possibly bind to other

uanidino compounds in the purine metabolic

pos

odwehhesmimfwehG

sachscsmTzptlhttcituukabptfm

a

bimcucfl

d3

t

xfws

pc

n

c

uF

un6

cst

ih

r1

SrT1

d

e

ak

sc

SULIMAN ET AL770

athway. Clearly, identification of an associationf uric acid with elevated phosphate levels de-erves additional studies.

Finally, there also were a few important lacksf associations. As discussed, there are extensiveata correlating an increase in uric acid levelith hypertension in the general population. Nev-

rtheless, in the current study, no association ofyperuricemia with the presence or absence ofypertension was noted. However, one potentialxplanation is that the mechanism of hyperten-ion in a subject with severely impaired GFRay relate to the direct effects of a low GFR to

mpair sodium handling; as a consequence, thisay be the overriding mechanism responsible

or hypertension as GFR decreases. In studies inhich hyperuricemia correlated with the pres-

nce of hypertension, subjects usually did notave longstanding hypertension and had normalFRs.54

Some potential limitations of the present studyhould be considered. First, because we relied on

single determination of uric acid level, weannot take into account any variation that mayave occurred over time. Second, mortality clas-ification was based on death certificates, whichan be biased and inaccurate. Thus, in the presenttudy, we relied on the more robust all-causeortality instead of cardiovascular mortality.hird, the study design does not allow generali-ation of the findings for the entire CKD patientopulations because sampling was performed athe initiation of renal replacement therapy andong-term dialyzed patients are not representedere. Also, because of exclusion of patients olderhan 70 years, study subjects were younger thanypical dialysis patient populations. Finally, be-ause this is a post hoc analysis, there may bencreased risk for type 1 error. However, despitehese limitations, we believe our study contrib-tes to the rather limited knowledge of the role ofric acid in patients with CKD. To the best of ournowledge, this is the first study that showsssociations of uric acid with metabolic distur-ances and calcium/phosphate disturbances inatients with CKD. Moreover, we report a long-erm follow-up period for survival that adjustedor several relevant risk factors, diuretic treat-ent, and potential confounders.In conclusion, the current study shows, during

6-year follow-up period, a J-shaped association d

etween uric acid levels and all-cause mortalityn patients with stage 5 CKD on renal replace-ent therapy. The association is independent of

omorbidity and confounders. In addition, highric acid levels are associated with lipid levels,alcium/phosphate metabolism, and levels of in-ammation markers.24

REFERENCES

1. Harrison R: Structure and function of xanthine oxi-oreductase: Where are we now? Free Radic Biol Med3:774-797, 20022. Sorensen LB: Role of the intestinal tract in the elimina-

ion of uric acid. Arthritis Rheum 8:694-706, 19653. Doehner W, Schoene N, Rauchhaus M, et al: Effects of

anthine oxidase inhibition with allopurinol on endothelialunction and peripheral blood flow in hyperuricemic patientsith chronic heart failure: Results from 2 placebo-controlled

tudies. Circulation 105:2619-2624, 20024. Johnson RJ, Kang DH, Feig D, et al: Is there a

athogenetic role for uric acid in hypertension and cardiovas-ular and renal disease? Hypertension 41:1183-1190, 2003

5. Wannamethee SG: Serum uric acid and risk of coro-ary heart disease. Curr Pharm Des 11:4125-4132, 20056. Alderman M: Uric acid in hypertension and cardiovas-

ular disease. Can J Cardiol 15:S20F-S22F, 1999 (suppl F)7. Culleton BF, Larson MG, Kannel WB, Levy D: Serum

ric acid and risk for cardiovascular disease and death: Theramingham Heart Study. Ann Intern Med 131:7-13, 19998. Lehto S, Niskanen L, Ronnemaa T, Laakso M: Serum

ric acid is a strong predictor of stroke in patients withon-insulin-dependent diabetes mellitus. Stroke 29:635-39, 19989. Fang J, Alderman MH: Serum uric acid and cardiovas-

ular mortality: The NHANES I epidemiologic follow-uptudy, 1971-1992. National Health and Nutrition Examina-ion Survey. JAMA 283:2404-2410, 2000

10. Gertler MM, Garn SM, Levine SA: Serum uric acidn relation to age and physique in health and in coronaryeart disease. Ann Intern Med 34:1421-1431, 195111. Puig JG, Ruilope LM: Uric acid as a cardiovascular

isk factor in arterial hypertension. J Hypertens 17:869-872,99912. Lee J, Sparrow D, Vokonas PS, Landsberg L, Weiss

T: Uric acid and coronary heart disease risk: Evidence for aole of uric acid in the obesity-insulin resistance syndrome.he Normative Aging Study. Am J Epidemiol 142:288-294,99513. Alexander RW: Inflammation and coronary artery

isease. N Engl J Med 331:468-469, 199414. Iuliano L: The oxidant stress hypothesis of atherogen-

sis. Lipids 36:S41-S44, 2001 (suppl)15. Caravaca F, Martin MV, Barroso S, et al: Serum uric

cid and C-reactive protein levels in patients with chronicidney disease. Nefrologia 25:645-654, 200516. Garg JP, Chasan-Taber S, Blair A, et al: Effects of

evelamer and calcium-based phosphate binders on uric acidoncentrations in patients undergoing hemodialysis: A ran-

omized clinical trial. Arthritis Rheum 52:290-295, 2005

eg

li4

ar

ci

cn4

eS

lp

hs

ho

ho

r8

ci

dI

hp

m2

a1

atK

m1

ipc

MgK

af

Rh

fP

lla

taJ

Ian

ih

cH

Ft

T

te

plD

ba

ltn

uK

pN

cp

URIC ACID AND MORTALITY IN CKD 771

17. Castro R, Herman A, Ferreira C, et al: Renagelfficacy in severe secondary hyperparathyroidism. Nefrolo-ia 22:448-455, 200218. Hsu SP, Pai MF, Peng YS, et al: Serum uric acid

evels show a ‘J-shaped’ association with all-cause mortalityn haemodialysis patients. Nephrol Dial Transplant 19:457-62, 200419. Stenvinkel P, Heimbürger O, Paultre F, et al: Strong

ssociation between malnutrition, inflammation, and atheroscle-osis in chronic renal failure. Kidney Int 55:1899-1911, 1999

20. Kyle UG, Schutz Y, Dupertuis YM, Pichard C: Bodyomposition interpretation. Contributions of the fat-free massndex and the body fat mass index. Nutrition 19:597-604, 2003

21. Frisancho AR: New standards of weight and bodyomposition by frame size and height for assessment ofutritional status of adults and the elderly. Am J Clin Nutr0:808-819, 198422. Vaziri ND, Freel RW, Hatch M: Effect of chronic

xperimental renal insufficiency on urate metabolism. J Amoc Nephrol 6:1313-1317, 199523. Lin JL, Lin-Tan DT, Hsu KH, Yu CC: Environmental

ead exposure and progression of chronic renal diseases inatients without diabetes. N Engl J Med 348:277-286, 200324. Iseki K, Ikemiya Y, Inoue T, et al: Significance of

yperuricemia as a risk factor for developing ESRD in acreened cohort. Am J Kidney Dis 44:642-650, 2004

25. Tomita M, Mizuno S, Yamanaka H, et al: Doesyperuricemia affect mortality? A prospective cohort studyf Japanese male workers. J Epidemiol 10:403-409, 200026. Iseki K, Oshiro S, Tozawa M, et al: Significance of

yperuricemia on the early detection of renal failure in a cohortf screened subjects. Hypertens Res 24:691-697, 2001

27. Ohno I, Hosoya T, Gomi H, et al: Serum uric acid andenal prognosis in patients with IgA nephropathy. Nephron7:333-339, 200128. Myllymaki J, Honkanen T, Syrjanen J, et al: Uric acid

orrelates with the severity of histopathological parametersn IgA nephropathy. Nephrol Dial Transplant 20:89-95, 2005

29. Syrjanen J, Mustonen J, Pasternack A: Hypertriglyceri-aemia and hyperuricaemia are risk factors for progression ofgA nephropathy. Nephrol Dial Transplant 15:34-42, 2000

30. Johnson RJ, Segal MS, Srinivas T, et al: Essentialypertension, progressive renal disease, and uric acid: Aathogenetic link? J Am Soc Nephrol 16:1909-1919, 200531. Nakagawa T, Mazzali M, Kang DH, et al: Hyperurice-ia causes glomerular hypertrophy in the rat. Am J Nephrol

3:2-7, 200332. Kang DH, Nakagawa T, Feng L, et al: A role for uric

cid in the progression of renal disease. J Am Soc Nephrol3:2888-2897, 200233. Siu YP, Leung KT, Tong MK, Kwan TH: Use of

llopurinol in slowing the progression of renal diseasehrough its ability to lower serum uric acid level. Am Jidney Dis 47:51-59, 200634. Khosla UM, Zharikov S, Finch JL, et al: Hyperurice-ia induces endothelial dysfunction. Kidney Int 67:1739-

742, 200535. Kang DH, Park SK, Lee IK, Johnson RJ: Uric acid-

nduced C-reactive protein expression: Implication on cellroliferation and nitric oxide production of human vascular

ells. J Am Soc Nephrol 16:3553-3562, 2005 p

36. Sanchez-Lozada LG, Tapia E, Santamaria J, et al:ild hyperuricemia induces vasoconstriction and maintains

lomerular hypertension in normal and remnant kidney rats.idney Int 67:237-247, 200537. Leyva F, Anker S, Swan JW, et al: Serum uric acid as

n index of impaired oxidative metabolism in chronic heartailure. Eur Heart J 18:858-865, 1997

38. Johnson RJ, Feig DI, Herrera-Acosta J, Kang DH:esurrection of uric acid as a causal risk factor in essentialypertension. Hypertension 45:18-20, 200539. Nakagawa T, Hu H, Zharikov S, et al: A causal role

or uric acid in fructose-induced metabolic syndrome. Am Jhysiol Renal Physiol 290:F625-F631, 200640. Kanellis J, Watanabe S, Li JH, et al: Uric acid stimu-

ates monocyte chemoattractant protein-1 production in vascu-ar smooth muscle cells via mitogen-activated protein kinasend cyclooxygenase-2. Hypertension 41:1287-1293, 2003

41. Lopez-Navarro AT, Gil A, Sanchez-Pozo A: Depriva-ion of dietary nucleotides results in a transient decrease incid-soluble nucleotides and RNA concentration in rat liver.Nutr 125:2090-2095, 199542. Kuzkaya N, Weissmann N, Harrison DG, Dikalov S:

nteractions of peroxynitrite with uric acid in the presence ofscorbate and thiols: Implications for uncoupling endothelialitric oxide synthase. Biochem Pharmacol 70:343-354, 2005

43. Muscelli E, Natali A, Bianchi S, et al: Effect ofnsulin on renal sodium and uric acid handling in essentialypertension. Am J Hypertens 9:746-752, 199644. Leyva F, Anker SD, Godsland IF, et al: Uric acid in

hronic heart failure: A marker of chronic inflammation. Eureart J 19:1814-1822, 199845. Zoccali C, Maio R, Mallamaci F, Sesti G, Perticone

: Uric acid and endothelial dysfunction in essential hyper-ension. J Am Soc Nephrol 17:1466-1471, 2006

46. Stenvinkel P: Inflammation in end-stage renal disease:he hidden enemy. Nephrology (Carlton) 11:36-41, 200647. Haglin L: Hypophosphataemia: Cause of the dis-

urbed metabolism in the metabolic syndrome. Med Hypoth-ses 56:657-663, 2001

48. Kalaitzidis R, Tsimihodimos V, Bairaktari E, Siamo-oulos KC, Elisaf M: Disturbances of phosphate metabo-ism: Another feature of metabolic syndrome. Am J Kidneyis 45:851-858, 200549. Yoo TW, Sung KC, Shin HS, et al: Relationship

etween serum uric acid concentration and insulin resistancend metabolic syndrome. Circ J 69:928-933, 2005

50. Denzer C, Muche R, Mayer H, et al: Serum uric acidevels in obese children and adolescents: Linkage to testos-erone levels and pre-metabolic syndrome. J Pediatr Endocri-ol Metab 16:1225-1232, 200351. Hsu CH, Vanholder R, Patel S, et al: Subfractions in

remic plasma ultrafiltrate inhibit calcitriol metabolism.idney Int 40:868-873, 199152. Vanholder R, Patel S, Hsu CH: Effect of uric acid on

lasma levels of 1,25(OH)2D in renal failure. J Am Socephrol 4:1035-1038, 199353. Takahashi S, Yamamoto T, Moriwaki Y, et al: De-

reased serum concentrations of 1,25(OH)2-vitamin D3 inatients with gout. Metabolism 47:336-338, 199854. Feig DI, Johnson RJ: Hyperuricemia in childhood

rimary hypertension. Hypertension 42:247-252, 2003