H

Mpinmot

Hac

bj

1

ypertension and Acid-Base/Electrolyte Disorders

ADMA, C-Reactive Protein, and Albuminuria in Untreated EssentialHypertension: A Cross-sectional Study

Costas Tsioufis, MD, PhD,1 Kyriakos Dimitriadis, MD,1 Eirini Andrikou, MD,1

Costas Thomopoulos, MD,1 Dimitris Tsiachris, MD,1 Elli Stefanadi, MD,1 Costas Mihas, MD,1

Antigoni Miliou, PhD,1 Vassilios Papademetriou, MD, PhD,2 andChristodoulos Stefanadis, MD, PhD1

Background: Asymmetric dimethylarginine (ADMA) and subclinical inflammation are associated withatherosclerosis progression, whereas microalbuminuria is an established index of hypertensive organdamage.

Study Design: Cross-sectional.Setting & Participants: In an outpatient hypertensive unit, 296 nondiabetic and untreated partici-

pants with hypertension were studied. Participants with atherosclerotic cardiovascular disease, severevalvulopathy, congestive heart failure, presence of neoplastic or other concurrent systemic disease,atrial fibrillation, serum creatinine level � 1.5 mg/dL in men and � 1.4 mg/dL in women, and urinaryalbumin excretion � 300 mg/24 h were excluded.

Predictors: ADMA and high-sensitivity C-reactive protein (hs-CRP) levels.Outcome Variable : Albuminuria assessed using albumin-creatinine ratio (ACR).Measurements: Participants underwent ambulatory blood pressure monitoring, echocardiography,

routine assessment of metabolic profile, ADMA, and hs-CRP, whereas ACR was determined as themean of 3 values in nonconsecutive morning spot urine samples.

Results: 64 participants had an ACR of 30-300 mg/g. Stratification based on ADMA level showed thatparticipants with hypertension in quartile [Q] 4 compared with those in Q3, Q2, and Q1 showed the highestACRs (53.2 vs 31.2 vs 30.4 vs 16.7 mg/g; P � 0.008 for all). Moreover, stratification based on hs-CRP levelshowed that participants with hypertension in Q4 (69.8% had microalbuminuria) showed the highest ACRs(72.2 vs 25.6, 16.2, and 19.2 mg/g for Q3, Q2, and Q1, respectively; P � 0.008 for all). Stepwise regressionanalysis showed that age, 24-hour systolic blood pressure, hs-CRP level, ADMA level, and the interaction ofhs-CRP with ADMA were independent predictors of ACR (R2 � 0.674; P � 0.001).

Limitations: Cross-sectional study.Conclusions: In patients with untreated essential hypertension, increased hs-CRP and ADMA levels

are associated with microalbuminuria, suggesting the involvement of inflammation and endothelialdysfunction in vascular and kidney damage.Am J Kidney Dis 55:1050-1059. © 2010 by the National Kidney Foundation, Inc. Published by ElsevierInc. All rights reserved.

INDEX WORDS: Endothelium; inflammation; kidney; atherosclerosis; hypertension.

flsau

dio

icroalbuminuria is an integrated markerof endothelial health and an independent

redictor of adverse cardiovascular events inndividuals with hypertension1-3 that is con-ected to diverse pathways of low-grade inflam-atory activation.3-6 Moreover, increased levels

f high-sensitivity C-reactive protein (hs-CRP),he most established downstream marker of in-

From the 1First Cardiology Clinic, University of Athens,ippokration Hospital, Athens, Greece; and 2Hypertensionnd Cardiovascular Research Clinic, Veterans Affairs Medi-al Center, Washington, DC.

Received June 29, 2009. Accepted in revised form Novem-er 16, 2009. Originally published online as doi:10.1053/

.ajkd.2009.11.024 on March 2, 2010.

American Journal of K050

ammation, characterize patients with hyperten-ion,7-11 contribute to target-organ damage,4-6,12

nd unfavorably modify the relation of albumin-ria to blood pressure (BP).13

Focusing on the endothelium, compelling evi-ence suggests that asymmetric dimethylargin-ne (ADMA), an endogenous inhibitor of nitricxide synthase, is an emerging risk factor14-16

Address correspondence to Costas Tsioufis, MD, PhD,ESC, 43, Agias Marinas Str, 15127 Melissia, Athens,reece. E-mail: ktsioufis@hippocratio.gr© 2010 by the National Kidney Foundation, Inc. Published by

lsevier Inc. All rights reserved.0272-6386/10/5506-0012$36.00/0

FG

E

doi:10.1053/j.ajkd.2009.11.024

idney Diseases, Vol 55, No 6 (June), 2010: pp 1050-1059

tmstotattrmatii

ah(alps

S

amaA

o2spl

P

hhpwthscsincc(

eo

tdhctoc

cilpama1luswpiw

Albuminuria, ADMA, and C-Reactive Protein 1051

hat has considerable cardiovascular effects,ainly by modulating arterial resistance and

tructure in the renal and systemic circula-ion.14,17,18 Thus, increased ADMA levels arebserved in patients with endothelial dysfunc-ion states, including chronic kidney disease,19

nd essential hypertension.20-22 There is no doubthat the vascular effects of ADMA are connectedo subclinical atherosclerosis,14-16,23 cardiacemodeling/hypertrophy,24 and renal impair-ent19,25,26 in a way that contributes to an unfavor-

ble prognosis. However, in the setting of essen-ial hypertension, no data are available aboutnterrelationships among albuminuria, chronicnflammation, and ADMA.

On this basis, the aim of the present study is tossess the possible associations of ADMA ands-CRP levels with albumin-creatinine ratioACR) and investigate whether there is an inter-ction of high ADMA and increased hs-CRPevels with albuminuria in nondiabetic partici-ants in the early stages of essential hyperten-ion.

METHODS

tudyDesign andSetting

A cross-sectional design was chosen to compare hs-CRPnd ADMA levels between patients with hypertension withicroalbuminuria and those with normoalbuminuria, as well

s to compare ACRs between those with high and lowDMA and hs-CRP levels.The study was set in the outpatient hypertensive unit of

ur institution within a period of almost 36 months (October003 to October 2006). Recruitment and selection of thetudy population and clinical and laboratory workup wereerformed within our hypertension unit. All data were col-ected during the mentioned active study period.

articipants

Beginning in October 2003, all patients with essentialypertension who referred or self-referred to the outpatientypertension clinic of our institution were asked to partici-ate in the Hippokration Hellenic Hypertension (3H) Study,hich was approved by the Ethics Committee of our institu-

ion. The 3H Study is an ongoing observational registry ofypertension-related target-organ damage specifically de-igned to evaluate indices of early structural and functionalardiovascular adaptations.27 After written informed con-ent had been obtained, all participants underwent the follow-ng procedures: (1) complete medical history, physical exami-ation, and determination of anthropometric characteristicsarried out by the attending physician; (2) routine biochemi-al analyses of blood and urine; (3) renal ultrasonography;

4) standard 12-lead electrocardiography; and (5) complete

sd

chocardiographic study. Secondary hypertension was ruledut based on the required clinical and laboratory workup.28

Of 2,615 participants enrolled during the first 3 years ofhe registry, we focused on patients with untreated newlyiagnosed (within the last 2 years) stages I-II essentialypertension28 without known diabetes mellitus or cardiovas-ular or other systemic disease (n � 370) for participation inhe present study. The diagnosis of hypertension was basedn 3 outpatient measurements of BP � 140/90 mm Hg andonfirmed by daytime ambulatory BP � 135/85 mm Hg.

In this initial population of 370 patients, after a thoroughlinical and laboratory examination, we excluded those withmpaired fasting glucose levels, atherosclerotic cardiovascu-ar disease, severe valvulopathy, congestive heart failure,resence of neoplastic or other concurrent systemic disease,trial fibrillation, renal insufficiency (serum creatinine � 1.5g/dL in men and � 1.4 mg/dL in women), and overt

lbuminuria (urinary albumin excretion � 300 mg/24 h; Fig). Moreover, no participant had a history or clinical/aboratory evidence of recent infection or inflammation ornderwent medical treatment 1 month before entry into thetudy. There were 296 patients with essential hypertensionho fulfilled the inclusion criteria. The study protocol com-lies with the Declaration of Helsinki, was approved by ournstitutional ethics committee, and all participants gaveritten informed consent.

Figure 1. Flow chart: recruitment and selection of the

tudy population. Abbreviations: CV, cardiovascular; DM,iabetes mellitus; HTN, hypertension.

V

pdAFaefi

P

osasd(bnhdiuA

Tm

3sacgvfiVfftdaCt2ttc

ecEH

5A

S

Aimspss

S

wcawviApepaoesPttmchwfAfmAtbccwAwietmhmcatwa

Tsioufis et al1052

ariables of Interest

The main outcome variable was ACR, whereas potentialredictors of outcome were levels of BP expressed usingiverse ambulatory BP parameters, as well as hs-CRP andDMA levels and the hs-CRP � ADMA interaction term.inally, potential confounders of the predictor variable werege, sex, smoking status, body mass index, waist circumfer-nce, 24-hour heart rate, glucose level, estimated glomerularltration rate, and low-density lipoprotein cholesterol level.

rocedures

OfficeandAmbulatoryBPMeasurement

Office BP measurement was performed at 3 different visits inur outpatient clinic, according to guidelines.28 Patients weretudied supine, having fasted and abstained from smoking,lcohol, and caffeinated beverages in the 12 hours before thetudy. Ambulatory BP was recorded over a working day (Mon-ay through Friday) using the automatic Spacelabs unit 90207Spacelabs, www.spacelabshealthcare.com). The procedure haseen described in detail.29 In brief, the cuff was fixed to theondominant arm and the device was set to obtain automaticeart rate and BP readings at 15-minute intervals during theaytime and 30-minute intervals during the nighttime. In keep-ng with current practice, daytime and nighttime were definedsing short fixed clock time intervals, which ranged from 10:00M to 8:00 PM and from midnight to 6:00 AM, respectively.wenty-four–hour systolic and diastolic BP values were theean of the overall 24-hour recordings after artifact editing.

UrinaryAlbuminExcretionandLaboratoryMeasurements

In all participants, ACR was determined as the average ofvalues obtained from nonconsecutive morning spot urine

pecimens (collected during 2 weeks) using a quantitativessay (DCA 2000; Bayer Diagnostics Europe, www.bayer.om) with a coefficient of variation of 2.8%.4,6 Estimatedlomerular filtration rate was assessed according to the 4ariable isotope dilution mass spectrometry–traceable Modi-cation of Diet in Renal Disease (MDRD) Study equation.30

enous blood samples were obtained from the anticubitalossa between 8:00 AM and 9:00 AM for estimation of lipid,asting glucose, serum creatinine, hs-CRP, and ADMA concen-rations. All blood samples from the study population wereirectly stored at �80°C in our laboratory. hs-CRP level wasssessed using a validated high-sensitivity assay (Dade BehringardioPhase* hsCRP Assay, www.dadebehring.com) with in-

ra- and interassay coefficients of variation of 3.4% and.1%, respectively. ADMA was evaluated using a competi-ive enzyme-linked immunosorbent assay (DLD Diagnos-ika, www.dld-diagnostika.de) with intra- and interassayoefficients of variation of 3.4% and 4.5%, respectively.

CardiacUltrasonography

Echocardiographic studies were performed by an experi-nced senior echocardiographer who was blinded to thelinical status of the examined participants, using a Generallectric Medical Systems Vivid 3 ultrasound imager (GE

ealthcare, www.gehealthcare.com) equipped with a 2.25 to p

-MHz transducer according to recommendations of themerican Society of Echocadiography.31

tudy Size

Previous data from our hypertensive unit showed a meanCR of 30.7 � 3.33 (standard deviation) mg/g. Hypothesiz-

ng a mean value � 31.2 mg/g (expected difference of 0.5g/g) at a significance level of P � 0.05, our initial

ample-size target was 399 patients, achieving statisticalower of 85%. Post hoc power analysis of the final sampleize of 296 patients showed that statistical power of thetudy was 75%.

tatistical Analysis

The SPSS statistical package release 12.0 (SPSS Inc.,ww.spss.com) was used for all statistical analyses. All

ontinuous variables with normal distribution are presenteds mean � standard deviation. The Shapiro-Wilk statisticas used to test for any deviation from normality. If aariable had a skewed distribution, descriptive statisticsnclude median and interquartile range. Distributions ofDMA and hs-CRP levels were stratified by quartiles andatients were categorized accordingly. Significant differ-nces between study subgroups were determined using inde-endent-samples t test after checking for equality of vari-nces using the Levene test and analysis of variance. Analysisf covariance was performed to detect significant differ-nces in hs-CRP, ADMA, and ACR values between studyubgroups after adjustment for established confounders.otential associations between continuous variables were

ested using the Pearson correlation statistic. To adjust forhe inflation of type I error caused by the high amount ofultiple comparisons, we used Bonferroni correction. Be-

ause there were 4 study subgroups in the analysis ofs-CRP and ADMA stratification, 6 pairwise comparisonsere possible, and differences were considered significant

or P � 0.008. Because of skewed distributions, hs-CRP,DMA, and ACR values were logarithmically (log10) trans-

ormed before statistical testing. Moreover, before enteringultiple regression models, the independent variables of logDMA and log hs-CRP have been centered (ie, subtracting

he mean score from each data-point) and a new term haseen created that is the interaction between the mentionedentered variables. The mean value used for calculation ofentered log hs-CRP was 0.202 for the total population,hereas the mean value for the calculation of centered logDMA was �0.257 for the total population. Forward step-ise linear multiple regression models were used to examine

ndependent significant predictors of ACR. The candidatexplanatory (independent) variables for entering the mul-iple regression model were age, sex, smoking status, bodyass index, waist circumference, 24-hour systolic BP, 24-

our diastolic BP, 24-hour heart rate, glucose level, esti-ated glomerular filtration rate, low-density lipoprotein

holesterol level, centered log hs-CRP, centered log ADMA,nd the centered log hs-CRP � centered log ADMA interac-ion term. To validate our multiple linear regression models,e tested the normality and heteroscedasticity of the residu-

ls produced. Any potential multicollinearity between ex-

lanatory variables was tested by calculating the variance

ilOvdsl

P

sfuotmtss

wHtun

A

tsmwtTCpc(

AMBWSOOO222LRGeTTHLhAAM

p�a

h

Albuminuria, ADMA, and C-Reactive Protein 1053

nflation factor (VIF) and tolerance (1/VIF). No multicol-inearity was found because tolerance values were � 0.1.ur multiple regression model had the highest adjusted R2

alue of all other models, explaining the variability of ourependent variable in the best way possible. Statisticalignificance was set at P � 0.05 for 2-sided tests duringinear regression modeling.

RESULTS

articipants andBaselineCharacteristics

Of 370 participants with untreated hyperten-ion, 74 did not meet the exclusion criteria: 19or cardiovascular diseases, including left ventric-lar hypertrophy; 33 for disturbed glucose metab-lism; 3 for proteinuria; 7 for significant sys-emic disease; 9 for recent infection; and 3 foredical treatment (apart from antihypertensive

reatment) within 1 month before entry into thetudy. Thus, 296 patients with essential hyperten-ion formed the study population.

Table 1. Characteristics of th

Parameter Q1 (n � 76)

ge (y) 53 � 7en (%) 42ody mass index (kg/m2) 28.3 � 3.9aist (cm) 94 � 10mokers (%) 38ffice systolic BP (mm Hg) 143 � 15ffice diastolic BP (mm Hg) 90 � 9ffice heart rate (beats/min) 75 � 84-h systolic BP (mm Hg) 133 � 74-h diastolic BP (mm Hg) 84 � 84-h heart rate (beats/min) 72 � 8eft ventricular mass index (g/m2) 97.7 � 11.6elative wall thickness 0.41 � 0.06lucose (mg/dL) 97 � 8GFR (mL/min/1.73 m2) 85 � 17otal cholesterol (mg/dL) 225 � 34riglycerides (mg/dL) 120 � 51DL cholesterol (mg/dL) 51 � 12DL cholesterol (mg/dL) 148 � 35s-CRP (mg/L) 2.45 (0.27-9.7)DMA (�mol/L) 0.5 (0.42-0.53)lbumin-creatinine ratio (mg/g) 16.7 (6-45)icroalbuminuria (%) 5

Note: Data are expressed as mean � standard deviaercentage. Conversion factors for units: serum creatinine0.05551; total, LDL, and HDL cholesterol in mg/dL tolbumin-creatinine ratio in mg/g to mg/mmol, �8.84; eGFRAbbreviations: ADMA, asymmetric dimethylarginine; BP,

igh-density lipoprotein; hs-CRP, high-sensitivity C-reactivaP � 0.008 versus Q1.bP � 0.008 versus Q2.

cP � 0.008 versus Q3.

Of 296 participants, 200 were men, mean ageas 51.6 years, mean office BP was 150/96 mmg, and mean 24-hour BP was 141/89 mm Hg. A

otal of 64 (21.6%) patients had microalbumin-ria (ACR, 30-300 mg/g) and 232 patients hadormoalbuminuria (ACR � 30 mg/g).

CRLevels by SerumADMAandhs-CRPQuartiles

Stratification based on ADMA levels showedhat ACRs were higher in patients with hyperten-ion in quartile (Q)3 (31.2 mg/g) and Q2 (30.4g/g) compared with those in Q1 (16.7 mg/g),hereas patients with hypertension in Q4 showed

he highest ACRs (53.2 mg/g; P � 0.008 for all;able 1). Moreover, stratification based on hs-RP level showed that ACRs were higher inatients with hypertension in Q3 (25.6 mg/g)ompared with those in Q2 (16.2 mg/g) and Q119.2 mg/g), whereas those in Q4 showed the

rt by Serum ADMA Quartiles

(n � 73) Q3 (n � 75) Q4 (n � 72)

0 � 8 49 � 7 51 � 865a 79a 69a

9 � 3.4 29.7 � 4.2a 28.8 � 3.90 � 11a 101 � 12a 98 � 11a

51 47 55a

7 � 12 149 � 13a 160 � 12a,b,c

7 � 8a 98 � 8a 100 � 8a,b

7 � 8 79 � 9a 82 � 10a,b

9 � 4a 138 � 3a 152 � 19a,b,c

9 � 6a 89 � 5a 94 � 10a,b,c

5 � 8 78 � 11a 77 � 10a

6 � 12.3 112.2 � 13.1a,b 114.2 � 16.9a,b

2 � 0.06 0.43 � 0.05a 0.43 � 0.05a

6 � 8 94 � 9 97 � 96 � 15 82 � 11 81 � 159 � 34 219 � 31 220 � 334 � 52 126 � 60 128 � 589 � 14 48 � 12 46 � 117 � 31 141 � 26 145 � 31.21-9.5) 2.43 (0.26-9.8) 3.8 (0.93-9.3)a,b,c

.53-0.56)a 0.58 (0.57-0.59)a,b 0.63 (0.6-0.8)a,b,c

-218)a 31.2 (6-184)a 53.2 (5-230)a,b,c

21a 9b 50a,b,c

edian (interquartile range) for skewed distributions, ordL to �mol/L, �88.4; plasma glucose in mg/dL to mmol/L,/L, �0.0259; triglycerides in mg/dL to mmol/L, �0.011;min/1.73 m2 to mL/s/1.73 m2, �0.01667.pressure; eGFR, estimated glomerular filtration rate; HDL,in; LDL, low-density lipoprotein; Q, quartile.

e Coho

Q2

5

28.10

1497

1387

98.0.4

98

21124

142.58 (00.54 (030.4 (6

tion, min mg/mmolin mL/blood

e prote

hQtgtcdu(

F

ocsdl02

a0m0P(lMbcva

Ai2el

AMBWSOOO222LRGeTTHLhAAM

p�a

h

Tsioufis et al1054

ighest ACRs (72.2 mg/g; P � 0.008 for Q4 vs1-3; Table 2). Analysis of covariance showed

hat the mentioned differences among study sub-roups remained statistically significant even af-er adjusting for age, body mass index, waistircumference, smoking status, 24-hour systolic/iastolic BP, 24-hour heart rate, estimated glomer-lar filtration rate, and glucose and lipid levelsP � 0.008 for all).

actorsAssociatedWithUrineACR

In the total study population, the main correlatesf log ACR were age (r � 0.275; P � 0.003), waistircumference (r � 0.192; P � 0.001), 24-hourystolic BP (r � 0.298; P � 0.001), 24-houriastolic BP (r � 0.165; P � 0.005), high-densityipoprotein cholesterol level (r � �0.185; P �.023), log hs-CRP (r � 0.486; P � 0.001; Fig

Table 2. Characteristics of th

Parameter Q1 (n � 75)

ge (y) 51 � 6en (%) 61ody mass index (kg/m2) 27.3 � 3.2aist (cm) 94 � 10mokers (%) 31ffice systolic BP (mm Hg) 150 � 17ffice diastolic BP (mm Hg) 97 � 8ffice heart rate (beats/min) 77 � 94-h systolic BP (mm Hg) 139 � 94-h diastolic BP (mm Hg) 88 � 74-h heart rate (beats/min) 75 � 8eft ventricular mass index (g/m2) 101.1 � 16.2elative wall thickness 0.41 � 0.06lucose (mg/dL) 96 � 8GFR (mL/min/1.73 m2) 87 � 12otal cholesterol (mg/dL) 221 � 37riglycerides (mg/dL) 121 � 60DL cholesterol (mg/dL) 51 � 12DL cholesterol (mg/dL) 147 � 34s-CRP (mg/L) 0.76 (0.21-1.21)DMA (�mol/L) 0.55 (0.48-0.63)lbumin-creatinine ratio (mg/g) 19.2 (6-51)icroalbuminuria (%) 1.3

Note: Data are expressed as mean � standard deviaercentage. Conversion factors for units: serum creatinine0.05551; total, LDL, and HDL cholesterol in mg/dL tolbumin-creatinine ratio in mg/g to mg/mmol, �8.84; eGFRAbbreviations: ADMA, asymmetric dimethylarginine; BP,

igh-density lipoprotein; hs-CRP, high-sensitivity C-reactivaP � 0.008 versus Q1.bP � 0.008 versus Q3.cP � 0.008 versus Q2.

A), log ADMA (r � 0.371; P � 0.001; Fig 2B), C

nd the interaction of hs-CRP with ADMA (r �.214; P � 0.001). Similarly, for log ADMA, theain correlates were 24-hour systolic BP (r �

.315; P � 0.002), log hs-CRP (r � 0.202;� 0.001; Fig 2C), left ventricular mass index

r � 0.282; P � 0.001), and estimated glomeru-ar filtration rate (r � �0.304; P � 0.002).

oreover, the main correlates for hs-CRP wereody mass index (r � 0.280; P � 0.001), waistircumference (r � 0.263; P � 0.001), leftentricular mass index (r � 0.146; P � 0.01),nd 24-hour systolic BP (r � 0.153; P � 0.009).

In a multiple regression analysis model usingCR as the dependent variable and age, sex, smok-

ng status, body mass index, waist circumference,4-hour systolic/diastolic BP, 24-hour heart rate,stimated glomerular filtration rate, glucose level,ow-density lipoprotein cholesterol level, log hs-

ort by Serum CRP Quartiles

(n � 74) Q3 (n � 74) Q4 (n � 73)

0 � 8 51 � 7 52 � 964 66 64.3 � 3.6a 29.1 � 4a 29.8 � 4.4a

9 � 9a 98 � 13a 102 � 11a,b

41 59a,c 61a,c

4 � 13a 152 � 12c 152 � 16c

3 � 9a 98 � 8c 97 � 9c

7 � 10 78 � 7 82 � 9a,b,c

8 � 12 142 � 14 144 � 19a,c

9 � 10 89 � 10 90 � 12a,b,c

4 � 10 74 � 9a 78 � 11.6 � 10.9 110.1 � 17.4a,c 109.9 � 13.7a,c

2 � 0.05 0.42 � 0.06a 0.44 � 0.06a,b

6 � 8 95 � 10 98 � 87 � 16 86 � 12 84 � 143 � 29 215 � 34 225 � 317 � 55 128 � 63 129 � 549 � 12 47 � 13 47 � 121 � 30 141 � 30 149 � 27.22-1.92)a 2.7 (1.94-4.04)a,c 6.2 (4.09-9.8)a,b,c

.42-0.69) 0.57 (0.43-0.67)c 0.58 (0.44-0.8)a,c

-51) 25.6 (6-152)a,c 72.2 (5-234)a,b,c

2.7 13.5a,c 69.8a,b,c

edian (interquartile range) for skewed distributions, ordL to �mol/L, �88.4; plasma glucose in mg/dL to mmol/L,/L, �0.0259; triglycerides in mg/dL to mmol/L, �0.011;min/1.73 m2 to mL/s/1.73 m2, �0.01667.pressure; eGFR, estimated glomerular filtration rate; HDL,in; LDL, low-density lipoprotein; Q, quartile.

e Coh

Q2

5

299

1497

1387

1010.4

98

22124

151.5 (1

0.55 (016.2 (6

tion, min mg/mmolin mL/blood

e prote

RP, log ADMA, and the interaction of hs-CRP

wsap

tlsaAhalphAca3tw

wccnAAlgeavrbofipttA

4

ttth0il

Albuminuria, ADMA, and C-Reactive Protein 1055

ithADMAas independent variables, age, 24-hourystolic BP, log hs-CRP, log ADMA, and the inter-ction of hs-CRP with ADMA were independentredictors of ACR (Table 3).

The interaction of ADMA and hs-CRP implieshat their effects on ACR vary with each other’sevels. Accordingly, based on the multiple regres-ion formula, we constructed 3 prediction vari-bles of ACR according to specific values of logDMA (25th, 50th, and 75th percentiles) whileolding the effect of the other explanatory vari-bles fixed. The same procedure was repeated forog hs-CRP. Three overlaid scatter and linearrediction plots were constructed for the ACR–s-CRP association according to the 3 levels ofDMA (Fig 3A). Another 3 overlaid plots were

onstructed to plot the correlation between ACRnd ADMA level in the 3 levels of hs-CRP (FigB). All interactions between independent predic-ors were tested and no significant associationas found.

DISCUSSION

The salient findings of our study are that patientsith hypertension with microalbuminuria are

haracterized by augmented ADMA and hs-CRPoncentrations and that the combination of pro-ounced inflammatory activation with increasedDMA level shows an adverse interaction withCR. This interaction of ADMA and hs-CRP

evels with albuminuria is independent of demo-raphic, hemodynamic, and metabolic confound-rs. Thus, when increased hs-CRP level is notccompanied by increased ADMA level and viceersa, albuminuria is augmented, but in an infe-ior degree compared with that observed whenoth hs-CRP– and ADMA-associated pathwaysf atherosclerosis progression are activated. Thesendings support the idea that detrimental inter-lay between endothelial status and inflamma-ion is connected to accelerated renal and sys-emic vascular dysfunction (reflected by increasedCR in this setting).

™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™Figure 2. Scatterplots with regression lines of log-

ransformed measurements in 296 individuals with essen-ial hypertension. Positive relationships are shown be-ween (A) log albumin-creatinine ratio (ACR) and logigh-sensitivity C-reactive protein (hs-CRP; R2 linear �.235); (B) log ACR and log asymmetric dimethylargin-

2

ne (ADMA; R linear � 0.138); and (C) log ADMA andog hs-CRP (R2 linear � 0.04).

tra

etvo

A2llh

e

l

Tsioufis et al1056

The cross-sectional nature of our study limitshe ability to infer any temporal and causalelationships of ADMA and hs-CRP levels withlbuminuria. Moreover, ADMA and hs-CRP lev-

Table 3. Analysis of Factors Assoc

Parameters Unstandardized � Coefficie

ge (y) 0.1024-h systolic BP (mm Hg) 2.745

og hs-CRP (mg/L) 0.385og ADMA (�mol/L) 3.233s-CRP � ADMA interaction 0.288

Note: Stepwise linear regression analysis using all 296 pantering the model. Overall P � 0.0001; R2 � 0.674, adjustAbbreviations: ADMA, asymmetric dimethylarginine; B

ipoprotein; hs-CRP, high-sensitivity C-reactive protein.

ls were determined only once, and given thathese markers may show diurnal and intraindi-idual fluctuations in their levels, the accuracy ofur results might be attenuated. Moreover, the

ith Urine Albumin-Creatinine Ratio

tandardized � Coefficient 95% CI P R2

0.092 0.06-0.13 0.04 0.1122.345 2.13-5.68 �0.001 0.1670.312 0.09-0.42 �0.001 0.2292.646 1.99-4.39 �0.001 0.0940.139 0.08-0.45 0.03 0.072

nts. Both log hs-CRP and log ADMA were centered before0.645.

od pressure; CI, confidence interval; HDL, high-density

Figure 3. Scatterplots with lin-ear prediction of (A) log(high-sensi-tivity C-reactive protein [hs-CRP])versus log(albumin-creatinine ratio[ACR]) for fixed values of asymmet-ric dimethylarginine (ADMA); and

iated W

nt S

rticipaed R2 �P, blo

(B) log(ADMA) versus log(ACR) forfixed values of hs-CRP.

eAmocusamriptraactuis

sti9kokioagtmet

lekchwrdsbrrt

iamolw

igtpscphglumaa2ne

otfusCsfiteddAeamcs

Btscge

Albuminuria, ADMA, and C-Reactive Protein 1057

nzyme-linked immunosorbent assay used forDMA determination is not the gold-standardethod, thus attenuating the absolute accuracy

f our results.32 Although we used the universalutoff value for ACR to define microalbumin-ria, an additional analysis (not presented) usingex-specific cutoff values did not significantlylter our results. Because we studied untreated,iddle-aged, and nondiabetic patients who were

eferred to our outpatient hypertension unit, find-ngs may not be directly applicable to otheratient groups or persons undergoing antihyper-ensive treatment in the community. Also, ouresults may not be applicable to different clinicalnd metabolic settings in which there is a lack ofssociation of hs-CRP level with ADMA level, inontrast to the present study.33,34 However, poten-ial strengths of the study are that patients werentreated and ambulatory BP monitoring, whichs more precise than sphygmomanometer mea-urements,35 was performed.

This is the first study to our knowledge thatpecifically showed that ADMA level is relatedo ACR in a selective hypertensive population. Its notable that ADMA level was responsible for.4% of the ACR variance, extending the currentnowledge regarding the pathophysiologic rolef this endogenous inhibitor of nitric oxide onidney-related target-organ damage. In addition,ncreased ADMA level was still a characteristicf patients with microalbuminuria, even afterdjustment for age and BP in conjunction withlomerular filtration rate. Moreover, 50% of pa-ients with the highest ADMA levels showedicroalbuminuria, supporting the link of ADMA

ffects with albuminuria in patients with hyper-ension.

Several mechanisms may explain how ADMAevel is interrelated with ACR. Urinary albuminxcretion levels might reflect prior exposure ofidney structures to ADMA-mediated vascularompromise, even before the development ofypertension.3 This is supported because personsith increased ACR even within the reference

ange are characterized by “early” endothelialysfunction, are more prone to become hyperten-ive, and are exposed to higher risk.1,3 It also haseen shown that increased ADMA levels areelated to impaired vascular compliance and ca-otid atherosclerosis,23,36 reflecting diffuse quali-

ative changes within the arterial wall. Furthermore, t

ncreased formation of collagen cross-linkingnd advanced glycation end products in the renalicrocirculation may be additional pathophysi-

logic mechanisms that may partially explain theink of ADMA-mediated vascular compromiseith pronounced albuminuria.Confirming previous studies, we showed that

nflammatory processes are involved in the patho-enesis of albuminuria in hypertensive pa-ients.4-6 Along these lines, hs-CRP level ex-lained almost 23% of the ACR variance,upporting the notion that “inflammatory mi-roalbuminuria” is of clinical importance in hy-ertension. Of note, almost 70% of patients withypertension in the highest hs-CRP stratificationroup had microalbuminuria. The establishedink of hs-CRP level with ACR could be attrib-ted to the fact that subclinical inflammationay cause direct injury to the renal glomerulus,

ltering its function and leading to the increasedlbuminuria,3,4,6,37 or hs-CRP and ACR may befacets of the same underlying biological mecha-ism associated with diffuse atherosclerotic dis-ase.3,37

Apart from the individual independent effectf increased ADMA and hs-CRP levels on ACR,heir combination is detrimental and responsibleor an additional 7.2% of variation in albumin-ria levels in our setting. Along these lines, aignificant interaction between ADMA and hs-RP on cardiovascular events is shown in the

etting of type 2 diabetes, further supporting ourndings.38 Although additional studies are needed

o assess the underlying mechanisms, a potentialxplanation is that increased hs-CRP levels mayecrease the already impaired endogenous vaso-ilator nitric oxide by augmented circulatingDMA, resulting in a pronounced albuminuric

ffect. Alternatively, one could argue that ADMAnd hs-CRP represent key mediators of distinctechanisms of subclinical vascular damage asso-

iated with increased albuminuria in a hyperten-ive substrate.

The observed associations of 24-hour systolicP with both hs-CRP and ADMA levels indicate

hat the imposition of oscillatory and low sheartress on the vascular beds may activate subclini-al inflammatory and ADMA-associated athero-enetic mechanisms.5,6,20-22,33 Notably, ADMAmerged as a stronger correlate of systolic BP

han hs-CRP, emphasizing the close link of endo-

tpHcptolescicbbdsl

iicuodfllibtitagatap

mhaioai

n

Md2

IaJ

Ktd4

mpC

is

aih

stV

Rh

rs2

i3

l1

ir

PrH

m

ct

ee

Tsioufis et al1058

helial dysfunction with hemodynamic load inatients with untreated essential hypertension.owever, it is not yet clear whether higher BPs

ause mechanical changes in endothelial cells,romoting a cascade of mechanisms of inflamma-ion and proatherogenesis, or whether it is thether way around or even a bidirectional interre-ationship. Alternatively, increased ADMA lev-ls may result from a decreased catabolic rateecondary to dimethylaminohydrolase inhibitionaused by the overwhelming oxidative stress thats a key feature of hypertension.9-11,14,15,20 In ourase, one could suggest that the alterations causedy ADMA and hs-CRP on albuminuria can note explained by merely BP levels because theifferences in ACRs between groups remainedignificant after adjustment for hemodynamicoad.

Our data suggest that in hypertensive patients,nflammation and ADMA are interrelated andnteract with albuminuria. Focusing on the so-alled Janus-faced role of low-grade albumin-ria,3,37 ACR levels could be seen as a barometerf the combined effects of ADMA-mediated en-othelial dysfunction and hs-CRP–associated in-ammatory activation on the hypertensive vascu-

ar tree. These points further emphasize themportant role of albuminuria measurement foretter cardiovascular risk assessment. In addi-ion, based on our findings, to improve prognosisn the early stages of hypertension, an integratedherapeutic approach (ie, renin-angiotensin-ldosterone system blockade) is needed. We sug-est that treatments with antialbuminuric1,3 andnti-inflammatory properties39 that also protecthe endothelium40 should be applied promptlynd in an intensified manner, especially in theresence of microalbuminuria.In conclusion, in patients with hypertension,icroalbuminuria is accompanied by increased

s-CRP and ADMA levels. Moreover, ADMAnd hs-CRP are related and show an adversenteraction on ACR, suggesting the involvementf inflammation and endothelial dysfunction inlbuminuria-related vascular and kidney damagen patients with untreated essential hypertension.

ACKNOWLEDGEMENTSSupport: None.Financial Disclosure: The authors declare that they have

o relevant financial interests. c

REFERENCES1. Pedrinelli R, Dell’ Omo G, Di Bello V, Pontremolli R,ariani M. Microalbuminuria, an integrated marker of car-

iovascular risk in essential hypertension. J Hum Hypertens.002;16:79-89.2. Geirstein H, Mann J, Qilong Yi, et al; HOPE Study

nvestigators. Albuminuria and risk of cardiovascular eventsnd heart failure in diabetic and nondiabetic individuals.AMA. 2001;286:421-426.

3. Tsioufis C, Dimitriadis K, Antoniadis D, Stefanadis C,allikazaros I. Inter-relationships of microalbuminuria with

he other surrogates of the atherosclerotic cardiovascularisease in hypertensive subjects. Am J Hypertens. 2004;17:70-476.4. Tsioufis C, Dimitriadis K, Chatzis D, et al. Relation oficroalbuminuria to adiponectin and augmented C-reactive

rotein levels in men with essential hypertension. Am Jardiol. 2005;96:946-951.5. Pedrinelli R, Dell’Omo G, Di Bello V, et al. Low-grade

nflammation and microalbuminuria in hypertension. Arterio-cler Thromb Vasc Biol. 2004;24:2414-2419.

6. Tsioufis C, Dimitriadis K, Taxiarchou E, et al. Diversessociations of microalbuminuria with C-reactive protein,nterleukin-18 and soluble CD 40 ligand in male essentialypertensive subjects. Am J Hypertens, 2006;19:462-466.

7. Engstrom G, Janson L, Berglund G, et al. Inflammation-ensitive plasma proteins and incidence of myocardial infarc-ion in men with low cardiovascular risk. Arterioscler Thrombasc Biol. 2003;23:2247-2251.8. Sesso HD, Buring JE, Rifai N, Blake GJ, Gaziano JM,

idker PM. C-Reactive protein and the risk of developingypertension. JAMA. 2003;290(22):2945-2951.9. Schillaci G, Pirro M, Gemelli F, et al. Increased C-

eactive protein concentrations in never-treated hyperten-ion: the role of systolic and pulse pressures. J Hypertens.003;21:1841-1846.10. Chae C, Lee R, Rifai N, Ridker P. Blood pressure and

nflammation in apparently healthy men. Hypertension. 2001;8:399-403.11. Dzau VJ. Angiotensin and pathophysiology of vascu-

ar disease: a unifying hypothesis. Hypertension. 2001;37:047-1052.12. Kuwahara F, Kai H, Niiyama H, et al. Roles of

ntercellular adhesion molecule-1 in hypertensive cardiacemodeling. Hypertension. 2003;41:819-823.

13. Stuveling EM, Bakker SJ, Hillege HL, et al;REVEND Study Group. C-Reactive protein modifies theelationship between blood pressure and microalbuminuria.ypertension. 2004;43:791-796.14. Cooke JP. Asymmetrical dimethylargine. The uberarker? Circulation. 2004;109:1813-1819.15. Zoccali C. Asymmetrical dimethylargine (ADMA): a

ardiovascular and renal risk factor on the move. J Hyper-ens. 2006:24: 611-619.

16. Boger RH, Zoccali C. ADMA: a novel risk factor thatxplains excess cardiovascular event rate in patients withnd-stage renal disease. Atheroscler Suppl. 2003;4:23-28.

17. Kielstein JT, Impraim B, Simmel S, et al. Cardiovas-

ular effects of systemic nitric oxide synthase inhibition

w2

Hp2

cis

L

fe2

rl2

de5

oil

ll2

Att2

pc

fh

GtoE2

o

th

tcs

dtEoJ

Hc(A

upe6

bflnl

tr

im

Ctc

tc2

zti

Rsc

Albuminuria, ADMA, and C-Reactive Protein 1059

ith asymmetrical dimethylarginine in humans. Circulation.004;109:172-179.18. Kielstein JT, Bode-Boger SM, Frolich JC, Ritz E,

aller H, Fliser D. Asymmetric dimethylarginine, bloodressure and renal perfusion in elderly subjects. Circulation.003;107:1891-1895.19. Zoccali C, Bode-Boger S, Mallamaci F, et al. Plasma

oncentration of asymmetrical dimethylarginine and mortal-ty in patients with end-stage renal disease: a prospectivetudy. Lancet. 2001;358:2113-2117.

20. Palloshi A, Fragasso G, Piatti P, et al. Effect of oral-arginine on blood pressure and symptoms and endothelialunction in patients with systemic hypertension, positivexercise test and normal coronary arteries. Am J Cardiol.004;93:930-935.21. Takiushi S, Fujii H, Kamide K, et al. Plasma asymmet-

ic dimethylarginine and coronary and peripheral endothe-ial dysfunction in hypertensive patients. Am J Hypertens.004;17:802-808.22. Perticone F, Sciacqua A, Maio R, et al. Asymmetric

imethylarginine, L-arginine and endothelial dysfunction inssential hypertension. J Am Coll Cardiol. 2005;46:518-23.23. Furuki K, Hisashi A, Matsuoka H, et al. Plasma levels

f asymmetric dimethylarginine (ADMA) are related tontima-media thickness of the carotid artery. An epidemio-ogical study. Atherosclerosis. 2007;191:206-210.

24. Zoccali C, Mallamaci F, Maas R, et al. Left ventricu-ar hypertrophy, cardiac remodeling and asymmetric dimethy-arginine (ADMA) in hemodialysis patients. Kidney Int.002;62:339-345.25. Krzyzanowska K, Mittermayer F, Shnawa N, et al.

symmetrical dimethylarginine is related to renal func-ion, chronic inflammation and macroangiopathy in pa-ients with type 2 diabetes and albuminuria. Diabet Med.007;24:81-86.26. Caglar K, Yilmaz MI, Sonmez A, et al. ADMA,

roteinuria, and insulin resistance in non-diabetic stage Ihronic kidney disease. Kidney Int. 2006;70:781-787.

27. Tsioufis C, Tsiachris D, Dimitriadis K, et al. Evidenceor no global effect of metabolic syndrome per se on earlyypertensive sequelae. J Hypertens. 2008;26:773-779.

28. Mancia G, De Backer G, Dominiczak A, et al. 2007uidelines for the Management of Arterial Hypertension:

he Task Force for the Management of Arterial Hypertensionf the European Society of Hypertension (ESH) and of theuropean Society of Cardiology (ESC). J Hypertens. 2007;5(6):1105-1187.29. Tsioufis C, Stefanadis C, Antoniadis D, et al. Absence

f any significant effects of circadian blood pressure varia- t

ions on carotid artery elastic properties in essentialypertensive subjects. J Hum Hypertens. 2000;14:813-818.30. National Kidney Foundation. K/DOQI Clinical Prac-

ice Guidelines for Chronic Kidney Disease: evaluation,lassification and stratification. Am J Kidney Dis. 2002;39(2uppl 1):S1-266.

31. Schiller NB, Shah PM, Crawford M, et al. Recommen-ations for quantitation of the left ventricular dimensions bywo-dimensional echocardiography. American Society ofchocardiography Committee on Standards, Subcommitteen Quantitation of the Two-Dimensional Echocardiograms.Am Soc Echocardiogr. 1989;2:358-367.32. Valtonen P, Karppi J, Nyyssonen K, Valkonen VP,

alonen T, Punnonen K. Comparison of HPLC method andommercial ELISA assay for asymmetric dimethylarginineADMA) determination in human serum. J Chromatogr Bnalyt Technol Biomd Life Sci. 2005;828:97-102.33. Surdai A, Nowicki M, Sandmann J, et al. Reduced

rinary excretion of nitric oxide metabolites and increasedlasma levels of asymmetrical dimethylarginine in men withssential hypertension. J Cardiovasc Pharmacol. 1999;33:52-658.34. Konukoglu D, Firtine S, Serin O. The relationship

etween plasma asymmetrical dimethyl-L-arginine and in-ammation and adhesion molecule levels in subjects withormal, impaired and diabetic glucose tolerance. Metabo-ism. 2008;57(1):110-115.

35. Verdecchia P, Schillaci G, Borgioni C, et al. Ambula-ory pulse pressure: a potent predictor of total cardiovascularisk in hypertension. Hypertension. 1998;32:983-988.

36. Kielstein JT, Donnerstag F, Gasper S, et al. ADMAncreases arterial stiffness and decreases blood flow in hu-ans. Stroke. 2006;37:2024-2029.37. Tsioufis C, Dimitriadis K, Kallikazaros I, Stefanadis

. The emerging role of subclinical inflammation in hyper-ension-associated early renal dysfunction: focus on mi-roalbuminuria. Hellenic J Cardiol. 2006;47(6):361-365.

38. Krzyzanowska K, Mittermayer F, Wolzt M, Schern-haner G. Asymmetric dimethylarginine predicts cardiovas-ular events in patients with type 2 diabetes. Diabetes Care.007;30:1834-1839.39. Di Napoli M, Papa F. Angiotensin-converting en-

yme inhibitor use is associated with reduced plasma concen-ration of C-reactive protein in patients with first everschemic stroke. Stroke. 2003;34:2922-2929.

40. Delles C, Schneider MP, John S, Gekle M, SchmiederE. Angiotensin converting enzyme inhibition and angioten-

in II AT-1 receptor blockade reduce the levels of asymmetri-al N(G), N(G)-dimethylarginine in human essential hyper-

ension. Am J Hypertens. 2002;15:590-593.