Serum Uric Acid and Blood Pressure in Children atCardiovascular Risk Q:1; 2; 3

WHAT’S KNOWN ON THIS SUBJECT: Uric acid (UA) is associatedwith hypertension in children, even after body weight adjustment.Whether the whole spectrum of variables, such as visceraladiposity, insulin resistance, puberty, and renal function,influence the relationship between UA and blood pressure isunknown.

WHAT THIS STUDY ADDS: In a cohort of children at relatively highcardiovascular risk, the association between UA and bloodpressure levels is independent of several well-known factorsimplicated in the development of hypertension, such as insulinresistance, pubertal status, and renal function Q:6.

abstractOBJECTIVES: Hyperuricemia Q:4has been shown to be a strong correlateof hypertension in children. However, the complex interaction betweenserum uric acid (UA), systemic blood pressure (BP), and possiblyconfounding factors has been elucidated only in part.

METHODS: We evaluated office BP as well as clinical and biohumoralparameters in a cross-sectional cohort of 501 children (280 boys and221 girls) aged between 6 and 18 years (mean = 10.8 years)consecutively referred for cardiovascular risk assessment.

RESULTS: Overall, 156 (31.1%) were normotensive, 122 (24.4%) showedtransient hypertension, 87 (17.4%) had prehypertension, and 136(27.1%) had hypertension. Altogether 33.3% and 40.5% of the studygroup were overweight or obese, respectively. There was a trend to-ward greater weight and waist circumference and higher BMI, Homeo-stasis Model Assessment index, and UA levels as the BP categoriesrose. Moreover, the prevalence of pubertal children, obesity, andwaist-to-height ratio above 0.50 progressively increased from lowerto upper BP categories. After adjusting for puberty, gender, BMI (z-score), Homeostasis Model Assessment index, and renal function, UAwas found to be directly related to systolic and diastolic BP values (P= .03). Using normotensive children for comparison, the risk ofshowing prehypertension or hypertension increased by at least50% for each 1 mg/dL UA increase (P , .01), whereas it doubledfor children in the top gender-specific UA quartile (P , .03).

CONCLUSIONS: Increased UA levels showed an independent predictivepower for the presence of higher BP levels among a cohort of childrenat relatively high cardiovascular risk Q:5. Pediatrics 2013;132:1–8

AUTHORS: Francesca Viazzi, MD,a Laura Antolini, PhD,b,c

Marco Giussani, MD,d Paolo Brambilla, MD,d Sara Galbiati,MD,b Silvana Mastriani, MD,e Andrea Stella, MD,b RobertoPontremoli, PhD,a Maria Grazia Valsecchi, PhD,b,c andSimonetta Genovesi, MDb,e

aUniversity of Genoa and IRCCS Azienda Ospedaliera UniversitariaSan Martino-IST Istituto Nazionale per la ricerca sul Cancro,Genova, Italy; bDepartment of Health Science, and cCenter ofBiostatistics for Clinical Epidemiology, University of Milano-Bicocca, Monza, Italy; dFamily Pediatrician, Milano, Italy; andeNephrology Unit, San Gerardo Hospital, Monza, Italy

KEY WORDSuric acid, children, hypertension, pre-hypertension, HOMA index,obesity

ABBREVIATIONSBP—blood pressureCI—confidence intervalCV—cardiovascularDBP—diastolic BPeGFR—estimated glomerular filtration rateHOMA—Homeostasis Model AssessmentHT—hypertensiveNT—normotensiveNW—normal weightOB—obeseOR—odds ratioOW—overweightPH—prehypertensiveSBP—systolic BPTH—transiently elevated blood pressureUA—uric acidWtHr—waist-to-height ratio

Drs Viazzi and Genovesi conceptualized and designed the study,carried out the initial analyses, and drafted the initialmanuscript; Drs Antolini, Galbiati, and Valsecchi contributed toanalysis and interpretation of data, and reviewed themanuscript; Dr Giussani designed the data collectioninstruments, coordinated and supervised data collection, andcritically reviewed the manuscript; Dr Mastriani gavea substantial contribution to acquisition of data and revised thearticle critically; Drs Brambilla, Stella, and Pontremolicontributed to interpretation of data and critically reviewed themanuscript; and all authors approved the final manuscript assubmitted.

(Continued on last page)

PEDIATRICS Volume 132, Number 1, July 2013 1

ARTICLE

Serum uric acid (UA) was shown to bea strong correlate of established hy-pertension in children.1–3 Preliminarystudies have also reported an associ-ation between hyperuricemia andnonsustained hypertension,4,5 and, re-cently, a relationship between changesin serum UA over time and bloodpressure (BP) from childhood toadulthood has been reported.6

These findings have sparked growinginterest in the relationship between UAand BP in children that may lead topathophysiological insights andgreater knowledge on the cause-effectrelationship between these factors.7

The complex interaction between se-rum UA, BP changes, and other well-established risk factors has been elu-cidated only in part, and controversyremains as to whether UA is an in-dependent causal factor or merelya marker of the development of hy-pertension.8 It has been suggested thatseveral factors, such as age,9 pu-berty,10 obesity,11,12 insulin re-sistance,13 and impaired renalfunction,14 all of which are related to UAlevels, may play a role in the de-velopment of hypertension. In 2 recenttrials, serum UA reduction by allopuri-nol or probenecid resulted in BP nor-malization in children and adolescentswith increased BP levels, thus sup-porting a causative role for UA in hy-pertension.15,16

The purpose of the current studywas toinvestigate the relationship betweenserum UA and BP, as well as severalpossibly confounding factors, such aspuberty, anthropometric parameters,insulin resistance, and renal function ina large group of children at relativelyhigh cardiovascular (CV) risk.

METHODS

Subjects

We studied a cohort of children agedbetween 6 and 18 years (mean = 10.8

years), consecutively referred by theirprimary care pediatricians to the SanGerardo Hospital, Unit for Cardiovas-cular Risk Assessment in Children, be-cause of evidence of elevated BP valuesand/or because of positive family his-tory of CV disease. The latter was de-fined as the presence in 1 or both of theparents of at least 1 among the fol-lowing: hypertension, type 2 diabetes,dyslipidemia, early ischemic heartdisease, and cerebrovascular disease.

None of these children were affected byimpaired glucose tolerance, diabetes,or renal insufficiency. Specific di-agnostic tests that are needed to ruleout secondary hypertension were car-ried out in all children. Children withsecondary forms of hypertension wereexcluded from the study.

Informed consent was obtained fromthe children’s parents, and the localethical committee approved the studyprotocol.

Definition of Terms and Groups

Height, weight (mobile digital scale[SECA; Vogel & Halke GmbH, Hamburg,Germany]; precision 100 g), and waistcircumference were measured withthe subjects in their underwear. BMIwas calculated as weight (kg)/height(m2). BMI z-scores were calculated byusing the Centers for Disease Controland Prevention charts (available athttp://www.cdc.gov/nchs/). Weight wasapproximated to hectograms, whereasheight precision was 5 mm. Weightclass was defined according to thetables of the International Obesity TaskForce,17 distinguishing among normalweight (NW), overweight (OW), andobese (OB).

Waist circumference was measured tothe nearest 0.1 cm by a nonelasticflexible tape with the children ina standing position. The tape was ap-plied horizontally midway between thelowest rib margin and the iliac crest.Waist-to-height ratio (WtHr) was cal-

culatedbydividingwaist circumferenceby height. A cutoff of 0.5 was chosen todistinguish between low and high WtHrcategory, as suggested by severalauthors.18–21

Pubertal stage was assessed bya medical examination and childrenwere classified into 2 categories: pre-pubertal and pubertal according toTanner staging.22

BP was measured by using an aneroidsphygmomanometer with the appro-priate cuff for the child’s upper armsize. The sphygmomanometer wascalibrated before starting the studyand once a month thereafter witha mercury sphygmomanometer. Sys-tolic BP (SBP) was defined by the firstKorotkoff sound (appearance ofsounds) and diastolic BP (DBP) wasidentified by the fifth Korotkoff sound(disappearance of sounds). BP valueswere approximated to the nearest 2mm Hg. Measurements were per-formedwhile children were sitting withtheir back supported and the cubitalfossa supported at the heart level, af-ter a rest of at least 5 minutes. BP wastaken 3 times (3–5-minute intervals)and SBP and DBP percentiles werecalculated according to the normo-grams recommended by the NationalHigh Blood Pressure Education Pro-gram Working Group on High BloodPressure in Children and Adoles-cents.23 Each child was classifiedaccording to the percentile of the meanof the 3 measurements as being nor-motensive (NT) if both SBP and DBPpercentiles were ,90th; pre-hypertensive (PH) if the SBP and/orDBP percentile was $90th but bothwere,95th; or hypertensive (HT) if theSBP and/or DBP percentile was$95th.Children who were referred by familypediatricians because of evidence ofelevated BP values, but whose SBP andDBP percentiles were both ,90thpercentile when BP was measured atthe Cardiovascular Risk in Children

2 VIAZZI et al

Unit, were classified as having tran-siently elevated BP (TH). Family historyof hypertension was defined as thepresence of at least 1 parent with hy-pertension.

Biochemical Parameters

Blood samples were taken from allsubjects after a 12-hour fasting periodtomeasure plasmaglucose and insulin,and serum UA, total cholesterol, tri-glycerides, andhigh-density lipoproteinconcentrations. Plasma glucose wasmeasured by a glucose oxidasemethodand insulin was evaluated by chemilu-minescent immunometric assay. Ho-meostasis Model Assessment (HOMA)index was calculated by dividing theproduct of plasma insulin (mU/mL) andplasma glucose (mmol/L) by 22.5.24

Glomerular filtration rate was esti-mated (eGFR) by means of the updatedSchwartz formula using serum creati-nine and height measurements and a kconstant of 0.413.25 Serum creatininelevel was measured by using the Jaffemethod referable to the standardizedreference measurement procedure(isotope dilution mass spectrometry),as recommended.

Statistical Methods

The cohortwas categorized into groupsaccording to hypertension category,weight class, and UA gender-specificquartiles. The continuous variableswere described by mean and SD andcompared across groups by Fisher-Snecodor and Student’s t tests. Thecategorical variables were describedby the proportion of subjects fallinginto each category. Proportions werecompared across groups by thex2 test.Box-plots were used to describe thedistribution of UA according to hyper-tension category and weight classbased groups. Correlation betweencontinuous variables was assessed bythe Pearson correlation coefficient.Multiple linear regression was used to

assess the influence of age, pubertalstatus, gender, BMI z-score (or WtHr)and eGFR on UA. The impact of UA wasconsidered on SBP z-score and on DBPz-score by a multiple linear regressionmodel adjusting for age, pubertal sta-tus, gender, BMI z-score (or WtHr), andHOMA index.

Logistic regression models were usedto assess the impact of the aforemen-tioned factors on the odds ratio (OR) ofbeingTH, PH, andHTwith respect toNT in3 separate models on the TH versus NT,PH versus NT, and HT versus NT, re-spectively.

No data were missing for any of thementioned variables. All analyses andgraphics were conducted with thepackage R (available at http://cran.r-project.org/).

RESULTS

Among 648 children consecutively re-ferred toouroutpatientclinic forCVriskassessment between November 2004and March 2012, 147 were excluded (12because they were ,6 years of age,and 135 because data for 1 or more ofthe variables included in the presentanalyses were missing). In the result-ing final cohort of 501 subjects (55.9%boys and 44.1% girls), mean age was10.8 years (SD = 2.4 years), 156 (31.1%)were NT, 122 (24.4%) were TH, 87(17.4%) were PH, and 136 (27.1%) wereHT. Among them, 33.3% and 40.5% wereOW or OB, respectively.

Elevated BP (PH or HT) was found in32.1% among 131 children with NW, in42.5% among 167 OW and 54.2% among203 OB. Table 1 reports the clinicalcharacteristics of study subjects on thebasis of the various BP categories. As

TABLE 1 Clinical Characteristics of Study Subjects According to BP Category (501 Children)

NT, n = 156;31.1%

TH, n = 122;24.4%

PH, n = 87;17.4%

HT, n = 136;27.1%

P Valuea

Age, y 10.5 2.2 10.7 2.3 10.8 2.5 11.1 2.6 .15Gender, boys, n (%) 83 (53.2) 68 (55.7) 50 (57.5) 79 (58.1) .85Puberty, yes, n (%) 56 (35.9) 61 (50.0) 36 (41.4) 69 (50.7) .03Weight, kg 45.8 14.8 49.9 16.5 52.2 16.8 56.7 20.4 ,.001Height, m 1.4 0.1 1.5 0.1 1.5 0.1 1.5 0.1 .19BMI, kg/m2 21.6 4.5 22.8 4.1 24.0 4.9 25.3 6.0 ,.001BMI, z-score 1.04 1.2 1.37 0.8 1.50 0.9 1.57 0.9 ,.001Weight class, n (%) ,.001NW 61 (39.1) 28 (23.0) 16 (18.4) 26 (19.1)OW 42 (26.9) 54 (44.3) 29 (33.3) 42 (30.9)OB 53 (34.0) 40 (32.8) 42 (48.3) 68 (50.0)

WC, cm 71.8 12.0 74.7 11.6 77.0 11.8 79.3 13.8 ,.001WtHr, % 49.8 7.4 51.1 6.2 52.8 6.5 53.6 7.7 ,.001WtHr .50%, n (%) 78 (50.0) 66 (54.1) 60 (69.0) 86 (63.2) .01SBP, mm Hg 108 7.6 113 7.1 119 5.7 130 10.5 ,.001DBP, mm Hg 64 6.4 66 5.8 70 6.0 74 9.5 ,.001SBP, z-score 0.40 0.56 0.73 0.46 1.37 0.26 2.14 0.37 ,.001DBP, z-score 0.20 0.53 0.38 0.50 0.70 0.54 1.03 0.76 ,.001Family history of HT, n (%) 50 (32.1) 46 (37.7) 30 (34.5) 50 (36.8) .92Creatinine, mg/dL 0.52 0.10 0.53 0.12 0.53 0.11 0.55 0.13 .23eGFR, mL/min 116 19.8 118 18.2 116 18.8 114 20.5 .63UA, mg/dL 3.9 0.9 4.2 1.1 4.4 1.0 4.6 1.1 ,.001Glucose, mg/dL 83 7.3 82 6.9 82 7.5 84 6.7 .10Insulin, mM/L 10.8 6.3 10.9 6.8 12.5 7.0 14.0 8.6 ,.001HOMA indexb 2.23 1.39 2.25 1.47 2.55 1.45 2.93 1.87 ,.001Total cholesterol, mg/dL 168 37.4 157 25.8 163 27.4 163 27.8 .03HDL cholesterol, mg/dL 58 14.1 55 13.5 55 13.9 54 12.2 .09Triglycerides, mg/dL 68 35.6 68 35.4 78 41.8 74 39.1 .10

Values are mean and SD unless otherwise indicated. HDL, high-density lipoprotein; WC, waist circumference.a P , .05 was considered statistically significant.b HOMA index = plasma insulin (mU/mL) 3 plasma glucose (mmol/L)/22.5.

ARTICLE

PEDIATRICS Volume 132, Number 1, July 2013 3

expected, there was a trend towardhigher BMI z-score and higher percen-tages of OW or OB, increasing WtHr, andHOMA index as BP categories rose.Moreover, the prevalence of pubertal,OB, children whose WtHr was above0.50 progressively increased from thelower to the upper BP categories. De-spite similarities in age, family historyof hypertension, renal function, fastingserum glucose, and lipid profile, thechildren showed a significant rise inserum UA levels along with increasingBP levels.

The distribution of UA levels based onweight classes is shown in Fig 1. Therewas a significant trend in increasing UAlevels from NW to OB.

Univariate analysis showed that amongthewholestudygroup,UAwasrelated toage (r = 0.38, P, .001), pubertal stage(t = –7.68, P , .001), BMI z-score (r =0.29, P , .001), WtHr (r = 0.27, P ,.001), eGFR (r = –0.22, P , .001), high-density lipoprotein cholesterol (r = –

0.26, P, .001), triglycerides (r = 0.23, P, .001), and HOMA index (r = 0.29, P,.001).

The relationship between UA and thesecorrelates was further investigated bymeans of multiple linear regressionanalysis. A significant and independentassociation with UA values was shownfor age (b= 0.11, P, .001), gender (b=0.11, P = .02), pubertal status (b = 0.35,P = .001), BMI z-score (b = 0.34, P ,.001), and eGFR (b = –0.006, P, .001).Similar results were found when WtHrwas used in the model as a measure ofweight excess instead of BMI z-score.HOMA index ceased to be significantlycorrelated to UA when included in thismodel (data not shown).

Univariate analysis showed that amongthe whole study group, UA was signifi-cantly related to both systolic BP (SBP)(z-score, r = 0.21, P, .001) and DBP (z-score, r = 0.11, P = .02).

A multiple regression model showedthat the only variables significantlyrelated to SBP (z-score) were BMI (z-score) (P, .001), HOMA index (P = .02),and serum UA (P = .03), whereas UAand pubertal status were related toDBP (z-score) (P = .03 and P = .02, re-spectively).

WhenWtHrwas included in themultiplelinear regression model as a measureof weight excess instead of BMI (z-score), the only variables significantlyrelated to SBP(z-score) were WtHr (P,.001) and UA (P = .03), whereas WtHrwas related to DBP (z-score) (P = .01).Including eGFR in the statistical modelyielded superimposable results.

Mean UA level was 4.38 (SD = 1.14) inboys and 4.15 (SD = 0.95) in girls (t teststatistic = –2.5, P = .01). Table 2 showsthe clinical characteristics of studysubjects on the basis of gender-specific serum UA quartiles. It can benoticed that, as UA quartiles increased,children were more likely to be older,pubertal, and OB. Moreover, there wasa trend toward higher BMI z-score,WtHr, BP parameters, serum creati-nine, and more unfavorable HOMA in-dex and lipid profile.

After adjusting for pubertal status,gender, BMI z-score, and HOMA index,the OR for the presence of HT amongthe whole population was 1.40 (95%confidence interval [CI] 1.12–1.75, P ,.01) for each 1 mg/dL increase in UAand 1.67 (95% CI 1.00–2.82, P = .05) forsubjects in the IV UA quartile (.4.7mg/dL for girls;.5.0 mg/dL for boys). Theimpact of UA was not found to be sig-nificant when we compared childrenwith TH versus those with NT. Using NTas comparison, the risk of being PH andof being HT increased by at least 50%(OR = 1.60 and P, .01, OR = 1.54 and P, .01, respectively, Table 3) for each 1mg/dL increase in UA, whereas it dou-bled (OR = 2.25, 95% CI 1.15–4.38, P =.01 and OR = 2.04, 95% CI 1.12–3.73, P =.02, respectively) for children in the IVUA quartile. The relationship betweenthe distribution of BP categories andUA quartiles is emphasized in Fig 2.When WtHr was used instead of BMI (z-score) as a measure of weight excess,the OR of being PH and of being HTwas2.17, 95% CI = 1.10–4.28 (P = .03) and1.95, 95% CI = 1.06–3.60 (P = .03),

FIGURE 1Boxplot distribution of UA (mg/dL) in groups defined according to weight class in 501 subjects. Boxplotexplanation: upper horizontal line of box = 75th percentile; lower horizontal line of box = 25th percentile;horizontal bar within box = median; square within box = mean; vertical lines out of the box = minimumand maximum. The P value displayed is referred to the overall comparison.

4 VIAZZI et al

respectively. Including eGFR in the sta-tistical model yielded almost superim-posable results.

DISCUSSION

In this large group of children at rela-tively high CV risk, serum UA was in-dependently associated with BP levels

across different BP categories, fromnormotension to transient and thenprehypertension, up to establishedhypertension.

Although the association between mildhyperuricemia and hypertension haspreviously been described in severalstudiesboth inadults and in children,1,3,

4,6,26 a relationship between UA andprehypertension has mainly beenreported in adults,27 with only a few,relatively small studies carried out inchildren with white coat and border-line hypertension.4,5,16 Although theclinical setting of our study is crosssectional, it provides us with an op-portunity to investigate the complexassociation between UA and othervariables (such as insulin resistance,obesity, puberty, and BP) in a group ofsubjects in whom pathophysiologicalmechanisms and changes leading tothe development of hypertension arestill at play. This is even more in-teresting in light of the recentlyreported finding that UA levels may bea predictor of the future developmentof hypertension in adults.6

The relationship between UA and BPthat we observed becomes clinicallyrelevant for higher UA values, sincechildren in the top UA quartile (ie,$5.0mg/dL for boys and $4.7 mg/dL forgirls) showed a twofold higher risk ofbeing PH and HT (Table 3, Fig 2). On theother hand, although a trend towardhigher UA values was present acrossall BP categories, UA levels could nothelp identify children with non-sustained hypertension.

The prevalence of OW and OB was rel-atively high in our study sample, asexpected on the basis of recruitmentcriteria. Nonetheless, the associationbetween UA and BP persisted afteradjusting for BMI z-score and for WtHr,a finding that clearly supports a path-ogenetic role for serum UA in the earlyphases of hypertension development.

With the aim of better elucidating therole of UA in the development of es-sential hypertension, in the currentstudy we excluded patients with renalinsufficiency and/or diabetes. However,we analyzed the relationship betweenUA and those parameters, such as eGFRand HOMA index, which, even in therange of normality, could be implicated

TABLE 2 Clinical Characteristics of Study Subjects on the Basis of UA Gender-Specific Quartiles(501 Children)

I, n = 145;28.9%

II, n = 118;23.6%

III, n = 114;22.7%

IV, n = 124;24.8%

P Valuea

Age, y 9.7 2.3 10.6 2.0 11.1 2.2 11.9 2.5 ,.001Gender, boys, n (%) 78 (53.8) 71 (60.2) 61 (53.5) 70 (56.5) .70Puberty, yes, n (%) 34 (23.4) 52 (44.1) 52 (45.6) 84 (67.7) ,.001Weight, kg 39.3 13.3 48.6 12.8 53.7 14.6 63.9 19.2 ,.001Height, m 1.4 0.1 1.5 0.1 1.5 0.1 1.5 0.1 ,.001BMI, kg/m2 20.3 4.2 22.8 4.3 24.1 4.2 26.5 5.5 ,.001BMI, z-score 0.90 1.1 1.35 1.0 1.55 0.7 1.67 0.8 ,.001Weight class, n (%)NW 64 (44.1) 33 (28.0) 17 (14.9) 17 (13.7) ,.001OW 43 (29.7) 39 (33.1) 50 (43.9) 35 (28.2) ,.001OB 38 (26.2) 46 (39.0) 47 (41.2) 72 (58.1) ,.001

WC, cm 67.3 10.9 74.8 10.5 77.2 10.1 83.9 12.7 ,.001WtHr, % 48.9 6.8 51.7 6.9 52.2 6.3 54.6 7.7 ,.001WtHr .50%, n (%) 59 (40.7) 73 (61.9) 69 (60.5) 89 (71.8) ,.001SBP, mm Hg 112 10.8 116 10.4 118 8.7 123 14.0 ,.001DBP, mm Hg 66 7.6 67 7.5 69 6.8 71 10.0 ,.001SBP, z-score 0.93 0.87 1.06 0.79 1.20 0.70 1.32 0.88 ,.001DBP, z-score 0.49 0.69 0.50 0.64 0.56 0.60 0.69 0.76 .08Family history of HTN, n (%) 50 (34.5) 39 (33.1) 43 (37.7) 44 (35.5) .68Creatinine, mg/dL 0.48 0.10 0.53 0.09 0.54 0.11 0.59 0.13 ,.001eGFR, mL/min 121 20.4 115 18.0 116 19.1 111 18.6 ,.001UA, mg/dL 3.1 0.4 4.0 0.2 4.6 0.2 5.7 0.7 ,.001Glucose, mg/dL 83 6.1 82 7.6 84 7.4 83 7.6 .33Insulin, mM/L 9.1 5.5 11.3 6.6 13.0 7.9 15.0 8.0 ,.001HOMA indexb 1.88 1.14 2.33 1.44 2.73 1.76 3.10 1.73 ,.001Total cholesterol, mg/dL 168 36.4 161 27.1 163 27.1 160 30.0 .14HDL cholesterol, mg/dL 60 13.2 57 14.1 53 11.6 50 12.6 ,.001Triglycerides, mg/dL 60 26.5 64 36.5 77 35.9 85 46.1 ,.001

Values are mean and SD unless otherwise indicated.UA gender-specific quartiles (#3.6 mg/dL,.3.6 mg/dL and #4.1,.4.1 mg/dL and #4.7 mg/dL, .4.7 mg/dL for girls, #3.6mg/dL, .3.6 mg/dL and #4.3, .4.3 mg/dL and #5.0 mg/dL, .5.0 mg/dL for boys).HDL, high-density lipoprotein; WC, waist circumference.a P , .05 was considered statistically significant.b HOMA index = plasma insulin (mU/mL) 3 plasma glucose (mmol/L)/22.5.

TABLE 3 Effect of Pubertal Status, Gender, BMI, UA (as continuous measure), and HOMA index onthe risk of TH versus NT, PH versus NT, and HT versus NT

Variable TH versus NT, n = 278 PH versus NT, n = 243 HT versus NT, n = 292

OR (95% CI) Pa OR (95% CI) Pa OR (95% CI) Pa

Puberty, Yes 1.97 (1.13–3.43) .01 1.09 (0.58–2.06) .78 1.47 (0.84–2.57) .17Gender, boys 1.11 (0.67–1.84) .69 1.15 (0.65–2.03) .64 1.29 (0.78–2.15) .32BMI, z-score 1.53 (1.15–2.05) ,.01 1.43 (1.03–1.99) .03 1.44 (1.07–1.93) .02UA, mg/dL 1.17 (0.89–1.54) .26 1.60 (1.15–2.22) ,.01 1.54 (1.17–2.03) ,.01HOMA indexb 0.82 (0.67–1.01) .06 0.96 (0.77–1.19) .70 1.06 (0.89–1.27) .50a P , .05 was considered statistically significant.b HOMA index = plasma insulin (mU/mL) 3 plasma glucose (mmol/L)/22.5.

ARTICLE

PEDIATRICS Volume 132, Number 1, July 2013 5

in the development of hypertension inchildren.13 Even slightly elevated serumUA levels have shown to be strong, in-dependent predictors of the de-velopment of type 2 diabetes in adultcohorts of unselected general pop-ulations,28 as well as of HT patients.29,30

Moreover, recent data strongly supportthe hypothesis that insulin resistancein children might play a role in hyper-tension development regardless ofobesity and fat distribution.13 Othervariables, such as puberty and renalfunction,10,14 have also been shown tobear an influence on the link betweenUA and BP in children.

To date, only a few studies have in-vestigated the relationship of sexualmaturity1,2 and estimated renal func-tion3,4,31 with UA and BP, and none havetaken insulin resistance into consid-eration. In the current study, HOMA in-dex was found to be linearly related toUA levels and increased along with UAquartiles (Table 4Q:7 ), and BP categories(Table 1). However, in multiple linearand logistic regression analysis, the

relationship between this marker ofinsulin resistance and BP levels andcategories almost completely loses itsstrength, whereas serum UA maintainsits independent link to BP. Althoughthese results cannot help to definewhether serum UA is influenced by BPelevation and insulin resistance or viceversa, they certainly point towarda complex relationship among thesefactors.

As forrenal function,we foundthateGFRwas similar in all BP categories, and theaddition of this variable to linear andlogistic regression analyses left therelationship between UA and BP un-changed. Even accounting for bodymassandpuberty, the independent roleof UA variations on BP values did notdecrease.

Our data confirm and expand thosepreviously reported by Loeffler et al3 ina large study conducted on a repre-sentative sample of adolescents in theUnited States. In fact, we were fortu-nate enough to obtain a larger studysample from our records and there-

fore were able to demonstrate an as-sociation between SBP as well as DBPand serum UA levels even in a fullyadjusted multivariate linear re-gression analysis.

Our study has both strengths andweaknesses that deserve to be com-mented on. Among the former, the factthat, unlike other previous large stud-ies,3 the definition of elevated BP andthe allocation of children into differentBP categories was not based on a sin-gle physical examination. BP wasmeasured in at least 2 different occa-sions: previously by the family pedia-trician and then again in our clinic, andin both cases the mean of 3 measure-ments was used for analysis. To fullyaccount for age, gender, and height,according to the tables published in thefourth report on the diagnosis, evalu-ation, and treatment in children andadolescents,23 we used BP z-scoresrather than raw BP values, whichmakes our observations more reliableas compared with those of a variety ofprevious reports.1,2,4,6,16 Furthermore,to the best of our knowledge, for thefirst time several clinical and bio-humoral variables have been concur-rently collected and analyzed ina group of referred children.

Last, there are some limitations to ourstudy. First, given the cross-sectionalnature of its design, it would be in-appropriate to derive definitive con-clusions regarding the cause-effectrelationship between UA and BP levels,even though the association we reportpersisted even when several well-known confounding factors weretaken into account. Second, the lack ofa control group may limit the implica-tionsof ourobservations. Nevertheless,within the context of children with rel-atively high CV risk, our data showa clear trend in the BP-UA associationalong with progressively worse BPcategories, from transient to estab-lished hypertension. Higher UA levels

FIGURE 2Distribution of BP categories according toUAquartiles. Stackedbarplot of BP category in groupsdefinedaccording to UA quartiles (I =#3.6 mg/dL, II =.3.6 mg/dL and#4.1, III =.4.1 mg/dL and#4.7 mg/dL,IV =.4.7 mg/dL for girls; I =#3.6 mg/dL, II =.3.6 mg/dL and#4.3, III =.4.3 mg/dL and# 5.0 mg/dL,IV =.5.0 mg/dL for boys) in 501 subjects. Using the NTas comparison, the adjusted risk of being PH (OR= 2.25, 95% CI 1.15–4.38, P = .01) and of being HT (OR = 2.04, 95% CI 1.12–3.73, P = .02) doubled forchildren in IV UA quartile. The P value displayed refers to the overall comparison.

6 VIAZZI et al

show a significant association with theoccurrence of higher BP levels evenamong a cohort of children at relativelyhigh CV risk. Third, ambulatory bloodpressure monitoring was not per-formed in our study and therefore wecould not classify any subject into the“white-coat hypertension” or “maskedhypertension” category. However,there is a general consensus in currentpediatric literature that diagnosis ofhypertension in children and adoles-

cents be based on office BP measuresand not on ambulatory blood pressuremonitoring.

CONCLUSIONS

In children at relatively high CV risk, UAshows a strong relationship with BPvalues across different BP categories,from normal BP to transient, up to pre-and finally to established hypertension.The association between UA and BP

levels is independent of several well-known factors potentially implicatedin the development of hypertension,such as insulin resistance, pubertalstage, and renal function. These datasupport the need for further large,prospective, and interventional studiesto clarify the pathophysiologicalmechanisms underlying the relation-ship between serum UA and BP.

REFERENCES

1. Török E, Gyárfás I, Csukás M. Factors as-sociated with stable high blood pressure inadolescents. J Hypertens Suppl. 1985;3(3):S389–S390

2. Goldstein HS, Manowitz P. Relation betweenserum uric acid and blood pressure inadolescents. Ann Hum Biol. 1993;20(5):423–431

3. Loeffler LF, Navas-Acien A, Brady TM, MillerER III, Fadrowski JJ. Uric acid level and el-evated blood pressure in US adolescents:National Health and Nutrition ExaminationSurvey, 1999-2006. Hypertension. 2012;59(4):811–817

4. Feig DI, Johnson RJ. Hyperuricemia inchildhood primary hypertension. Hyper-tension. 2003;42(3):247–252

5. Rovda II, Kazakova LM, Plaksina EA.Parameters of uric acid metabolism inhealthy children and in patients with ar-terial hypertension [in Russian]. Pediatriia.1990;8(8):19–22

6. Alper AB Jr, Chen W, Yau L, Srinivasan SR,Berenson GS, Hamm LL. Childhood uric acidpredicts adult blood pressure: the Boga-lusa Heart Study. Hypertension. 2005;45(1):34–38

7. Feig DI. Uric acid and hypertension. SeminNephrol. 2011;31(5):441–446

8. Viazzi F, Leoncini G, Ratto E, Pontremoli R.Serum uric acid as a risk factor for car-diovascular and renal disease: an oldcontroversy revived. J Clin Hypertens(Greenwich). 2006;8(7):510–518

9. Galescu O, George M, Basetty S, et al. Bloodpressure over height ratios: simple andQ:8

accurate method of detecting elevatedblood pressure in children. Int J Pediatr.2012;2012: 253497

10. Liang Y, Mi J. Pubertal hypertension isa strong predictor for the risk of adult

hypertension. Biomed Environ Sci. 2011;24(5):459–466

11. Genovesi S, Antolini L, Giussani M, et al.Usefulness of waist circumference for theidentification of childhood hypertension. JHypertens. 2008;26(8):1563–1570

12. Sorof J, Daniels S. Obesity hypertension inchildren: a problem of epidemic pro-portions. Hypertension. 2002;40(4):441–447

13. Genovesi S, Brambilla P, Giussani M, et al.Insulin resistance, prehypertension, hy-pertension and blood pressure values inpaediatric age. J Hypertens. 2012;30(2):327–335

14. VanDeVoorde RG, Mitsnefes MM. Hyperten-sion and CKD. Adv Chronic Kidney Dis. 2011;18(5):355–361

15. Feig DI, Soletsky B, Johnson RJ. Effect ofallopurinol on blood pressure of adoles-cents with newly diagnosed essential hy-pertension: a randomized trial. JAMA. 2008;300(8):924–932

16. Soletsky B, Feig DI. Uric acid reductionrectifies prehypertension in obese adoles-cents. Hypertension. 2012;60(5):1148–1156

17. Cole TJ, Bellizzi MC, Flegal KM, Dietz WH.Establishing a standard definition for childoverweight and obesity worldwide: in-ternational survey. BMJ. 2000;320(7244):1–6

18. Maffeis C, Banzato C, Talamini G; ObesityStudy Group of the Italian Society of Pedi-atric Endocrinology and Diabetology. Waist-to-height ratio, a useful index to identifyhigh metabolic risk in overweight children.J Pediatr. 2008;152(2):207–213

19. Garnett SP, Baur LA, Cowell CT. Waist-to-height ratio: a simple option for de-termining excess central adiposity in youngpeople. Int J Obes. 2008;32:1028–1030

20. Srinivasan SR, Wang R, Chen W, Wei CY, Xu J,Berenson GS. Utility of waist-to-height ratio

in detecting central obesity and relatedadverse cardiovascular risk profile amongnormal weight younger adults (from theBogalusa Heart Study). Am J Cardiol. 2009;104(5):721–724

21. Mokha JS, Srinivasan SR, Dasmahapatra P,et al. Utility of waist-to-height ratio inassessing the status of central obesity andrelated cardiometabolic risk profile amongnormal weight and overweight/obese chil-dren: the Bogalusa Heart Study. BMCPediatr. 2010;10:73

22. Tanner JM. Growth and maturation duringadolescence. Nutr Rev. 1981;39(2):43–55

23. National High Blood Pressure EducationProgram Working Group on High BloodPressure in Children and Adolescents. Thefourth report on the diagnosis, evaluation,and treatment of high blood pressure inchildren and adolescents. Pediatrics. 2004;114(2 suppl):555–576

24. Matthews DR, Hosker JP, Rudenski AS,Naylor BA, Treacher DF, Turner RC. Homeo-stasis model assessment: insulin re-sistance and beta-cell function from fastingplasma glucose and insulin concentrationsin man. Diabetologia. 1985;28(7):412–419

25. Schwartz GJ, Muñoz A, Schneider MF, et al.New equations to estimate GFR in childrenwith CKD. J Am Soc Nephrol. 2009;20(3):629–637

26. Grayson PC, Kim SY, LaValley M, Choi HK.Hyperuricemia and incident hypertension:a systematic review and meta-analysis.Arthritis Care Res (Hoboken). 2011;63(1):102–110

27. Syamala S, Li J, Shankar A. Associationbetween serum uric acid and pre-hypertension among US adults. J Hyper-tens. 2007;25(8):1583–1589

28. Kodama S, Saito K, Yachi Y, et al. Associa-tion between serum uric acid and

ARTICLE

PEDIATRICS Volume 132, Number 1, July 2013 7

development of type 2 diabetes. DiabetesCare. 2009;32(9):1737–1742

29. Viazzi F, Leoncini G, Vercelli M, Deferrari G,Pontremoli R. Serum uric acid levels pre-dict new-onset type 2 diabetes in hospital-ized patients with primary hypertension:

the MAGIC study. Diabetes Care. 2011;34(1):126–128

30. Wiik BP, Larstorp AC, Høieggen A, et al. Se-rum uric acid is associated with new-onsetdiabetes in hypertensive patients with leftventricular hypertrophy: the LIFE study. AmJ Hypertens. 2010;23(8):845–851

31. Silverstein DM, Srivaths PR, Mattison P,et al. Serum uric acid is associated withhigh blood pressure in pediatric hemodi-alysis patients. Pediatr Nephrol. 2011;26(7):1123–1128

(Continued from first page)

www.pediatrics.org/cgi/doi/10.1542/peds.2013-0047

doi:10.1542/peds.2013-0047

Accepted for publication Apr 5, 2013

Address correspondence to Simonetta Genovesi, MD, Dipartimento di Scienze della Salute Università degli Studi di Milano Bicocca Via Cadore 48, 20900, Monza (MB),Italy. E-mail: simonetta.genovesi@unimib.it

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright © 2013 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

FUNDING: No external funding.

8 VIAZZI et al

1—Titles are limited to 2 lines of printed text; if your title goes beyond this limit, please reduce the length to nomore than 97 characters (including spaces and punctuation).

2—Please verify the corresponding author’s contact information.

3—Please verify all author names, degrees, and affiliations.

4—Please confirm whether funding for this research was provided by the National Institutes of Health. If so,please provide any relevant grant information if it has not already been noted.

5—Per journal style, all abstracts must have fewer than 250 words. If your abstract extends beyond this limit,please edit as necessary (if changes are extensive, include or e-mail an electronic text file with the newabstract).

6—Per journal style, the text for “What’s known on this subject” and “What this study adds” may not exceed40 words each. Please amend as necessary to meet this requirement.

7—No table matches the in-text citation "Table 4". Please supply a table or delete the citation.

8—Ref. 9, please verify page numbers? (i.e., 253-497)? If possible please supply volume and issue numbers.

AUTHOR QUERIES

AUTHOR PLEASE ANSWER ALL QUERIES