lable at ScienceDirect

Atherosclerosis 242 (2015) 37e44

Contents lists avai

Atherosclerosis

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

Diabetic nephropathy is an independent factor associated to severesubclinical atheromatous disease

Clara Barrios a, b, Julio Pascual a, b, *, Sol Otero c, Maria J. Soler a, b, Eva Rodríguez a, b,Silvia Collado a, b, Anna Faura a, b, Sergi Mojal b, Juan F. Navarro-Gonz�alez d, e,Angels Betriu f, Elvira Fernandez f, 1, Jose M. Valdivielso f, 1, on behalf of the investigators ofthe NEFRONA studya Department of Nephrology, Hospital del Mar, Barcelona, Spainb Institute Mar for Medical Research, Barcelona, Spainc Nephrology Service, Consorci Sanitari del Garraf, Barcelona, Spaind University Hospital Nuestra Se~nora de Candelaria, Santa Cruz de Tenerife, Spaine Research Unit, University Hospital Nuestra Se~nora de Candelaria, Santa Cruz de Tenerife, Spainf Department of Nephrology, Hospital Arnau de Vilanova, Lleida, Spain

a r t i c l e i n f o

Article history:Received 14 March 2015Received in revised form5 June 2015Accepted 25 June 2015Available online 30 June 2015

keywords:Atheromatous diseaseDiabetic nephropathyCarotid ultrasound

* Corresponding author. Department of NephrologySpain.

E-mail address: julpascual@gmail.com (J. Pascual)1 EF and JMV share senior authorship.

http://dx.doi.org/10.1016/j.atherosclerosis.2015.06.0480021-9150/© 2015 Elsevier Ireland Ltd. All rights rese

a b s t r a c t

Background: Atheromatous disease (AD) is a risk factor for death in renal patients. Traditional CV riskfactors do not predict the presence of AD in this population. The aim of this study is to analyze whetherthe etiology of the primary renal disease influences in the risk of having silent AD.Study design: Observational cross-sectional study in chronic kidney disease patients without previouscardiovascular events.Settings and Participants: 2436 CKD subjects without any previous CV event included in the prospectiveSpanish multicenter NEFRONA study. Patients were classified according to primary renal disease: dia-betic nephropathy (n ¼ 347), vascular nephropathy (n ¼ 476), systemic/glomerular disease (n ¼ 447),tubulointerstitial and drug toxicity nephropathy (n ¼ 320), polycystic kidney disease (n ¼ 238), non-filiated nephropathy (n ¼ 406) and other causes (n ¼ 202).Predictors: B-mode and Doppler ultrasonography analysis of the carotid arteries were performed tomeasure intima media thickness (IMT) and the presence of plaques. Clinical and laboratory parametersrelated to CV risk were also determined.Outcomes: AD was scored according with the ultrasonography findings and the ankle-brachial index intotwo large groups: absence or incipient AD and severe AD.Results: In multivariate regression analysis, older age (OR 1.09/year [1.088e1.108]), smoking habit (OR2.10 [1.61e2.74]), male gender (OR 1.33 [1.09e1.64]), grade-5D of CKD (OR 2.19 [1.74e2.74]), and diabeticnephropathy (OR 2.59 [1.93e3.48]) are independent risk factors for severe AD. The prevalence of silentAD was highest for diabetic nephropathy with grade-5D of CKD (82.2%) and lowest with stages 2e3 CKDsystemic/glomerular disease (36.6%).Limitations: Observational study with the potential for confounding.Conclusion: In CKD patients without any CV event in the background clinical history, diabetic ne-phropathy as primary renal disease is the most significant factor associated to severe silent AD.Furthermore, this difference was independent of other conventional risk factors for atherosclerosis andCV events.

© 2015 Elsevier Ireland Ltd. All rights reserved.

, Hospital del Mar, Barcelona,

.

rved.

1. Introduction

Cardiovascular (CV) events represent the leading cause ofmortality for patients with renal disease as well as for the general

C. Barrios et al. / Atherosclerosis 242 (2015) 37e4438

population [1]. Furthermore, chronic kidney disease (CKD) is awell-known risk factor for CV disease. CV mortality in end-stagerenal disease (ESRD) patients is around 10e15 times higher thanin the general population [2,3]. The high CV mortality in CKD pa-tients highlights the lack of efficacy of the prophylactic strategiestaken to date. For instance, different studies fail to showa beneficialeffect of using statins or ACE inhibitors in reducing CV risk in ESRDpatients [4e6]. Traditional CV risk factors underestimate theatheromatous disease in this population [7]. In this regard, effortsshould be addressed to detect patients at higher risk. Since carotidatheromatous disease (AD) is related to coronary disease and CVrisk [7,8] the use of imaging techniques such as carotid ultrasoundmay add valuable information in the study of the population withrenal disease. Thus, we have previously shown in the NEFRONAStudy that the presence of carotid plaques is significantly increasedeven in early stages of renal disease [9,10] Furthermore, the infor-mation provided by the analysis of arterial stiffness in addition toimaging techniques improves overall knowledge of the AD of thepatients [11,12].

Recent guidelines recommend classifying CKD based on etiol-ogy, estimated glomerular filtrate (eGFR) and urinary albumin/creatinine ratio [13]. Previous observational studies have beenlimited to assess the relevance of the etiology of kidney disease tothe kidney disease progression [14,15]. Recently, a subanalysis ofthe Study of Heart and Renal Protection (SHARP) has suggested thatthe causes of CKD have substantial prognosis implications in therisk of ESRD. Thus, patients with cystic kidney disease are at muchhigher risk of ESRD and much lower risk of death. By contrast pa-tients with diabetic nephropathy (DN) are at high risk of deathbefore reaching ESRD [15]. These findings point out that etiology ofrenal disease influences the prognosis of renal outcome but differ inthe prognosis of death.

In the present study we hypothesized that the type of under-lying kidney disease in CKD patients may be associated with adifferent risk of subclinical AD, and could help us to better detectand stratify CV risk in the CKD population. Notably, the relationshipbetween the etiologies of the renal disease and CV risk, especially inpatients with subclinical AD, has been scarcely analyzed. Our aim isto study whether the etiology of the renal disease affects the risk ofhaving silent AD in a large population with different stages of CKDand complete absence of any CV event in their background.

2. Methods

2.1. The NEFRONA study

Study samples were obtained from the NEFRONA study. This is amulticenter, prospective observational study aimed to assess thepredictive value of imaging techniques and biomarkers for CVdisease in a large group of patients from outpatient clinics innephrology departments with different grades of CKD and renal-healthy subjects, as previously described [16,17]. Briefly, 3004subjects aged 18e75 years old were enrolled from 81 Spanishcenters between October 2010 and June 2012. The exclusion criteriawere: previous CV events, active infections (HIV, tuberculosis),pregnancy, having received any organ transplantation or having alife expectancy of less than 1 year. Each physician team recordedetiologies of kidney disease according to the evidences of patient'sclinical history. The eGFR was calculated using the Modification ofDiet in Renal Disease Study (MDRD4) equation. Each local ethicscommittee approved the study, and subjects were included afterproviding informed consent. Subjects were stratified accordingwith the KDIGO-CKD guidelines in different grades of CKD. For ouranalysis we only included patients with eGFR <60 ml/min/1.73 m2

or between 60 and 89 ml/min/1.73 m2 if they had a urine albumin

to creatinine ratio higher than 300mg/g (Grade2 A3). Subjects weresubsequently categorized into 7 groups according to the etiologiesof CKD: diabetic nephropathy (DN), vascular nephropathy (VN),glomerular and systemic disease (Systemic/Glom), tubulointer-stitial and drug toxicity nephropathy (TIN), polycystic kidney dis-ease (Cystic), non-filiated nephropathy and other causes (“Others”)inwhichwe include genetic, tumor, renal hypoplasia among others.

2.2. Clinical and biochemical data

At recruitment, patients were asked to complete a questionnaireincluding family history on premature CV disease, clinical history(diabetes mellitus, hypertension and dyslipidemia), CV risk factors(such as smoking habit), and medication use. Biochemical param-eters were obtained from a routine blood test performed within a3-month period of the vascular exploration. Blood pressure wasmeasured with a validated semi-automatic oscillometer (OmronHEM-705CP).

All subjects underwent ultrasound exploration to evaluate thecarotid and femoral arteries. For this study we analyzed data ob-tained from carotid exploration. The ultrasound was performedaccording to a standardized protocol by three itinerant teamsuniformly trained (a nurse and a radiology technician) belonging tothe UDETMA (Unit for Detection and Treatment of Athero-thrombotic Diseases, Hospital Universitari Arnau de Vilanova,Lleida, Spain). The itinerant teams also collected the anthropo-metric parameters as well as blood samples, which were storedwithin 24 h at the centralized biobank of the Spanish Network forNephrological Research (REDinRen) at the University of Alcal�a(Madrid, Spain).

2.3. Carotid ultrasound

Analysis of the carotid arteries were performed to measure in-tima media thickness (IMT) and the presence of plaques. B-modeultrasound of the carotid arteries was performed using the VividBT09 apparatus (General Electric instrument) equipped with a6e13 MHz broadband linear array probe. A unique reader in ablinded fashion performed the analysis by using the semi-automatic software EchoPAC Dimension (General Electric Health-care). To assess the quality of the reading and the intra-observerreliability, a sample of 20 individuals was measured 3e5 times ondifferent days. A kappa coefficient of 1 was obtained, indicatingexcellent intra-observer reliability.

2.4. Ankle-brachial index

Vascular Doppler MD2 Hungleigth was used with an 8 MHztransducer and a sleeve for making manual blood pressure. Thedetermination of blood pressure was performed in the brachialartery in both arms and in both feet. To calculate the ankle-brachialindex (ABI) the higher brachial blood pressure was used or theclosest in time to the malleolar measure.

2.5. Severity of atheromatous disease

Atheromatous disease (AD) was initially scored in 4 groups ac-cording with the ultrasonography findings and the ABI measure-ment [18,19]: Stage 0; subjects with ABI > 0.9 and IMT < 80%according reference range (RR). Stage 1; ABI between 0.7 and 0.9and/or carotid IMT � 80% according (RR): Stage 2; carotid plaquewithout stenosis (peak systolic velocity < 125 cm/s), Stage 3;ABI < 0.7 and/or carotid plaque with stenosis >50% (peak systolicvelocity � 125 cm/s). Finally, we separated our population into twolarge groups: Absent/incipient AD (Stage 0e1) and severe AD (Stage

C. Barrios et al. / Atherosclerosis 242 (2015) 37e44 39

2e3).

2.6. Statistical analysis

Descriptive analysis included the absolute and relative fre-quencies for qualitative variables as well as the mean and standarddeviation (SD) or median and interquartile range (IQR) for quanti-tative variables. The Pearson chi-square test was used to comparethe distribution of qualitative variables among the groups. TheANOVA or the non-parametric ManneWhitney and KruskaleWallistests were used to compare the distribution of variables betweenthe groups. The estimate of the factors related to the presence ofsevere AD was achieved by fitting a logistic multivariate regressionmodel, including all variables in bivariate analysis with significanceof p < 0.1. Subsequently the step backward method was used toremove variables that did not provide a statistically significantimprovement in the model. The statistical level of significance wasfixed at 0.05. The SPSS V 18.0 (IBM corp.) and STATA V12 programswere used for all statistical analyses.

3. Results

3.1. Characteristics of the study population

Demographic and clinical data according to the etiology of theCKD are shown in Table 1. A total of 2436 CKD subjects were

Table 1General, epidemiological, laboratory and imaging parameters according to the etiology o

DN VN System/Glom

N 347 476 447Age (years) 59.13 ± 12.40 63.42 ± 9.24 52.48 ± 13.7Gender (% male) 63.98 66.18 63.09Smoking (%) 20.46 17.86 20.58Body mass index (kg/m2) 29.69 ± 5.93 29.72 ± 5.02 27.38 ± 5.12Diabetes mellitus (%) 100 18.49 9.62Hypertension (%) 96.54 100 90.83Dyslipidemia (%) 77.81 68.70 65.10Family history early CV death (%) 7.20 9.03 8.50ACEIs/ARB (%) 78.96 78.78 74.05Statin treatment (%) 70.32 61.13 59.96Antiplatelet treatment (%) 59.65 25.84 16.78Total cholesterol (mg/dL) 171.08 ± 39.48 183.60 ± 39.18 179.80 ± 42.LDL cholesterol (mg/dL) 95.52 ± 34.78 107.32 ± 34.15 102.01 ± 34.Triglyceride (mg/dL) 134 [92e185] 125 [95e171] 127.2 [93e17Glycade haemoglobin (%) 7.39 ± 1.39 5.84 ± 0.90 5.66 ± 0.86Calcium (mg/dL) 9.26 ± 0.58 9.41 ± 0.55 9.24 ± 0.61Phosphorus (mg/dL) 4.22 ± 1.03 3.72 ± 0.97 4.30 ± 1.29Calcium*phosphorus product 38.93 ± 9.07 34.99 ± 9.25 39.58 ± 11.5I-PTH (pg/mL) 135 [77.7

e225]96 [59e176] 135.1 [69e22

Uric acid (mg/dL) 6.67 ± 1.67 6.74 ± 1.60 6.86 ± 1.60Haemoglobin (gr/dL) 12.49 ± 1.64 13.38 ± 1.76 12.59 ± 1.65hsCRP (mg/L) 2.36 [1.1e5.16] 2.07 [1.1e3.98] 1.56 [0.8e3.9

Ferritin (ng/mL) 142 [65e269] 157 [73.5e289]

190.1 [88e34

Urine albumin/Cr (mg/g)(a) 194 [24.2e762]

67 [8.08e300.8]

235.4 [52e915.5]

eGFR (mL/min) 27.9 [20.3e40.5]

39 [26.7e48.2] 26.6 [19.7e39.3]

Intima emedia thickness (mm,mean ± SD)

0.758 ± 0.149 0.770 ± 0.138y 0.683 ± 0.13

Severe atheroumatous disease(% of patients)

74.99 68.08jj 45.86§

DN; Diabetic Nephropathy, VN; Vascular Nephropathy, TIN; Tubulointerstitial and drugconverting enzyme inhibitors/angiotensin type II receptor blockers, Ca x P: Calciumglomerular filtration rate (MDRD4 equation). Values for categorical variables are given asrange]. *p: Indicates differences between the groups. (a)data of urine Albumin/Cr ratio wevs diabetic nephropathy. jjP < 0.05 vs diabetic nephropathy, §p < 0.001 vs diabetic neph

analyzed; 924 in grade 2e3, 619 in grade 4, 216 in grade 5 (non-dialysis) and 677 on dialysis treatment, grade 5D. Patients with VNand non-filiated nephropathy exhibited the highest average agefollowed by those in diabetic nephropathy group.

There were no differences among diabetic nephropathy patientsand the Non-diabetic etiologies of CKD related, smoking habit andfamily history of early CV disease. TIN and Cystic etiologies showedthe lowest percentage of males without differences regardinggender among the others groups.

As expected, diabetic nephropathy patients had higher fre-quency of diabetes mellitus comparing with the other CKD etiol-ogies. Hypertension and dyslipidemia were also more frequent indiabetic nephropathy and vascular nephropathy patients. Of notethe groups of TIN, Cystic and “Others” etiology showed the lowestpercentage of patients regarding this two prevalent diseases. Thepercentages of patients under ACE/ARB inhibitors, statin or anti-platelet treatment were higher in the diabetic nephropathy group.

Laboratory parameters according to the etiology of the CKD areshown in Table 1. The features in the lipid profile showed thatdiabetic nephropathy patients displayed the lowest levels in totaland LDL cholesterol but the highest levels of triglycerides. Patientsin Cystic, Systemic/glom and DN groups showed the poorest controlof Ca*P product and phosphorus levels. Although the differencesdid not reach statistical significance, hemoglobin and ferritin werelower in patients with diabetic nephropathy. Patients in DN groupshowed the highest levels of high sensitive C-reactive protein

f the chronic kidney disease.

TIN Cystic Non-filiated Others p*

320 238 406 2023 57.05 ± 13.06 54.07 ± 10.79 58.82 ± 12.33 55.22 ± 14.40 <0.05

48.75 53.36 66.50 61.39 <0.00120.62 21.01 16.50 21.78 0.427.81 ± 5.19 27.17 ± 4.51 28.04 ± 4.73 26.86 ± 5.02 <0.00110.94 5.88 18.72 13.37 <0.00177.50 89.50 85.22 78.22 <0.00157.50 56.72 60.59 55.45 <0.0015.94 11.76 7.14 3.47 <0.0560.31 61.34 60.10 58.91 <0.00155.94 51.26 53.94 48.51 <0.00113.12 11.34 22.17 17.82 <0.001

52 181.26 ± 38.35 177.35 ± 40.09 175.45 ± 36.61 178.81 ± 39.2 <0.00148 104.14 ± 32.54 101.65 ± 34.25 99.46 ± 30.05 101.95 ± 33.53 <0.0018] 118 [88.57e163] 122 [87e165] 126 [94e185] 122 [91e176] 0.16

5.57 ± 0.74 5.35 ± 0.681 5.81 ± 1.05 5.81 ± 1.29 <0.0019.31 ± 0.61 9.22 ± 0.63 9.31 ± 0.65 9.36 ± 0.61 <0.0013.92 ± 1.05 4.36 ± 1.10 3.99 ± 1.05 4.05 ± 1.11 <0.05

0 36.50 ± 9.80 40.16 ± 10.01 37.10 ± 9.86 37.95 ± 10.62 <0.055] 140.8 [81.13

e251.1]157.9 [91.5e254]

116 [68.87e244]

131 [68.6e232]

<0.05

6.40 ± 1.58 6.61 ± 1.37 6.69 ± 1.49 6.60 ± 1.66 0.1612.85 ± 1.66 12.48 ± 1.50 12.78 ± 1.77 12.85 ± 1.84 <0.001

6] 2.23 [1.12e4.85] 1.91 [0.9e3.97] 2.04 [0.94e5.04]

2 [0.88e5.08] <0.05

2] 147 [69.8e280] 167.5 [78.3e330]

150.3 [76.8e304]

168 [74e339.7]

0.056

50.9 [13.4e269.3] 77.6 [16.5e204] 66.2 [5.7e413] 46.4 [4e245.7] <0.001

30.2 [21e43.7] 26.3 [14.5e37.5]

35.1 [22e49] 31.5 [20.9e44.8]

<0.001

8‡ 0.711 ± 0.142‡ 0.685 ± 0.122‡ 0.746 ± 0.143‡ 0.711 ± 0.158‡ <0.001

53.46§ 44.44§ 58.87§ 52.00§ <0.001

toxicity Nephropathy, Cystic; Polycystic kidney disease, ACEIs/ARB; angiotensin-ePhosphorus Product. HsCRP: high-sensitive reactive C protein. eGFR; estimatepercentage; values for continuous variable as mean (±SD) or median [Interquartilere available in 45% of the participants. yP ¼ 0.12 vs diabetic nephropathy. ‡P < 0.001ropathy.

C. Barrios et al. / Atherosclerosis 242 (2015) 37e4440

(hsCRP) 2.36 [1.14e5.16] (mg/L). Urinary data were available only in45% of the participants. As expected, DN and Systemic/Glom pa-tients groups had the highest levels of urinary Albumin/Creatinineratio 194 [24.2e762] and 235 [52e915.5] (mg/g) respectively,without statistically significant differences between groups(p ¼ 0.51).

The average of eGFR showed statistical differences between thegroups. In a post hoc analysis, the eGFR of DN group (27.9[20.3e40.5] mL/min/1.73 m2) was significantly lower than in thevascular nephropathy group.

3.2. Atheromatous disease in the different CKD aetiologies andstages

Using the findings of the ultrasonography and the Ankle-Brachial index, we compared the carotid intima-media thicknessand the AD score among the etiologies groups (Table 1). There wasno difference between the groups with diabetic nephropathy vsvascular nephropathy; however, both showed thicker carotid in-tima media than CKD patients due to other any causes. The groupwith diabetic nephropathy had higher percentage of patients withsevere AD compared to vascular CKD patients (75.0% vs 68.1%respectively p < 0.05); moreover this difference was more evidentwhen compared to any other causes of CKD.

To assess whether the increased prevalence of patients withsevere AD was equally distributed by grades of CKD, we subse-quently performed the comparison by aetiologies and grades ofkidney disease (Fig. 1). Patients with diabetic nephropathy in grade5D of CKD displayed the highest percentage of patients with severeAD. The figure shows comparisons in the bivariate model with thereference group being those having diabetic nephropathy. While in

Fig. 1. Severe carotid atheromatous disease in different chronic kidney disease grades. Cyssystemic renal diseases, TIN; Tubulointerstitial and drug toxicity Nephropathy, VN; Vascula

grade 5D of CKD the percentage of patients with severe AD washigher in the reference group than in vascular nephropathy (82.2%vs 77.4%, respectively, p < 0.05), this difference was not significantin earlier grade of CKD. The results point out that any etiology ofCKD show lower percentage of patients with severe AD and in anygrade of kidney function comparing with the DN group (p < 0.001).

We conducted an initial analysis among all the other studyvariables (Table 1) and their relationship to carotid AD (Table 2).Family history of early CV disease, treatment with ACE/ARB in-hibitors, total cholesterol, corrected Ca, P and Ca*P product, I-PTH,hemoglobin and uric acid were not statistically associated with ADin the bivariate model.

To assess the potential association of urinary Albumin/Creati-nine ratio with severe silent AD we analyzed 1015 patients withurine data available. The urine Albumin/Creatinine ratio was notsignificantly related to the presence of severe AD (p ¼ 0.53). Whenwe restricted the analysis to a subset of patients with macro-albuminuria (>300 mg/dL, n ¼ 322), the relationship remainedabsent (p¼ 0.6). Other variables such as age, gender, smoking habit,body mass index, diabetes mellitus, hypertension, dyslipidemia,triglyceridemia, uric acid, hsCRP and ferritin showed a nominallysignificant relationship with severe AD. Patients with severe ADreceived statin or antiplatelet treatment in a higher percentagethan patients with carotid absence/incipient AD.

Table 3 shows the variables associated to an elevated risk ofcarotid severe AD, after adjustment for baseline characteristics(age, gender, smoking status, grade and etiology of CKD) (Model A).Old age, male gender, being smoker or former smoker and grade 5Dof CKD were associated with a higher risk of having severe AD. Themultivariate model estimates an odds ratio of 2.59 of having severeAD when diabetic nephropathy was the etiology of the CKD

tic; Polycystic kidney disease, DN; Diabetic Nephropathy, Glom/System: glomerular orr Nephropathy, *p < 0.05 **p < 0.001.

Table 2Demographic, clinical and analytical variables and their relationship to carotid atheromatous disease (AD).

Absent or incipient AD Severe AD p

Age (years) 51.06 ± 13.88 62.37 ± 9.34 <0.001Gender (% males) 55.57 65.93 <0.001Smoking habit (%) 47.95 59.35 <0.001Family history early CV (%) 7.81 7.79 0.9Body mass index 27.41 ± 5.13 28.88 ± 5.16 <0.001Diabetes mellitus (%) 17.09 32.14 <0.001Arterial hypertension (%) 86.23 91.21 <0.001Dyslipidemia (%) 55.59 68.71 <0.001ACEIs/ARB (%) 70.12 68.43 0.37Statin treatment (%) 53.7 61.86 <0.001Antiplatelet treatment (%) 17.29 30 <0.001Total cholesterol (mg/dL) 178.6 ± 39.33 178.46 ± 40.3 0.3LDL cholesterol (mg/dL) 102.17 ± 33.64 101.95 ± 33.6 0.07Triglyceride (mg/dL) 121 [89e166] 128 [94e177] <0.05Glycade haemoglobin (%) 5.94 ± 1.3 6.21 ± 1.27 <0.001Phosphorus (mg/dL) 4.08 ± 1.49 4.04 ± 1.09 0.4CaxP product 37.96 ± 10.43 37.51 ± 9.98 0.8I-PTH (pg/mL) 129 [69e226] 123 [71e226] 0.7Uric acid (mg/dL) 6.72 ± 1.49 6.64 ± 1.64 0.9Haemoglobin (gr/dL) 13.0 ± 1.7 12.82 ± 1.73 0.6hsCRP (mg/L) 1.78 [0.82e3.88] 2.18 [1.06e5.08] <0.001Ferritin (ng/mL) 148 [71e278] 167 [78.6e326] <0.001Urine albumin/Cr (mg/g)a 119 [14e442.9] 88.2 [9.8e453.8] 0.53

ACEIs/ARB, angiotensin-converting enzyme inhibitors/angiotensin type II receptor blockers, CaxP: CalciumePhosphorus Product, hsCRP; high sensitive C-reactiveprotein. Values for categorical variables are given as percentage; values for continuous variable as mean (±SD) or median [Interquartile range]).

a Data of urine Albumin/Cr ratio were available in 45% of the participants.

Table 3Multivariate logistic regression model for the presence of severe carotid atheromatous disease.

Odd ratio CI 95% p

Model AAge (risk/year) 1.098 [1.088e1.108] <0.001Gender (Male) 1.33 [1.09e1.64] 0.005Smoker 2.10 [1.61e2.74] <0.001Former-smoker 1.56 [1.24e1.96] <0.001Grade 5D of CKDa 2.19 [1.74e2.74] <0.001Other grades of CKDb Grade 5D (Reference) Grade 5D (Reference)Grade 2-3 0.40 [0.31e0.51] <0.001Grade 4 0.42 [0.32e0.56] <0.001Grade 5 0.63 [0.44e0.91] 0.014Etiology of CKD (Diabetic Nephropathy)c 2.59 [1.93e3.48] <0.001Others etiologies of CKDd Diabetic Nephropathy (reference) Diabetic Nephropathy (reference) .Vascular nephropathy 0.47 [0.33e0.67] <0.001Systemic/Glomerular 0.35 [0.24e0.50] <0.001TIN 0.38 [0.26e0.55] <0.001Cystic 0.28 [0.18e0.42] <0.001Non-Filiated 0.41 [0.28e0.58] <0.001Others 0.37 [0.24e0.57] <0.001Model BAge (risk/year) 1.094 [1.083e1.105] <0.001Gender (Male) 1.34 [1.06e1.69] 0.012Smoker 2.15 [1.59e2.90] <0.001Former-smoker 1.73 [1.34e2.90] <0.001Grade 5D of CKD 2.03 [1.57e2.63] <0.001Etiology of CKD (Diabetic Nephropathy) 2.73 [1.95e3.82] <0.001Body mass index 1.019 [0.997e1.042] 0.081Dyslipidemia 1.25 [0.99e1.57] 0.054Triglyceride (mg/dL) 1.078 [0.861e1.349] 0.66hsCRP (mg/L) 1.033 [0.940e1.134] 0.49Ferritin (ng/mL) 1.151 [1.029e1.286] 0.013

Model A: Adjusting by age, gender, smoking habit, grade of CKD and etiology of CKD.Model B: Including clinical and laboratory variables significantly associated with severe AD in the bivariate model, except DM and Hypertension.Abbreviations: CI; confidence interval, hsCRP; high sensitive C-reactive protein.

a Grade 5D (Dialysis) compared to any other grades of CKD.b Grade 5D of CKD as reference group.c DN compared to any other causes of CKD all together.d DN as reference group.

C. Barrios et al. / Atherosclerosis 242 (2015) 37e44 41

C. Barrios et al. / Atherosclerosis 242 (2015) 37e4442

compared to any other etiologies of CKD. Considering those havingdiabetic nephropathy as the reference group, the fully adjusted ORswere significantly lower for patients with all the other CKD etiol-ogies (Table 3). Further adjustment for other variables nominallyassociated with carotid severe AD did not materially change theseestimates (Model B). After including these variables, with theexception of diabetes and hypertension, the effects of age, gender,smoker status, grade and etiology of CKD were still present in themodel.

Other factors such as hypertriglyceridemia, dyslipidemia or BMIdid not reach a significant association with a higher risk for severeAD after adjusting for baseline characteristic in this model. Afteradjusting, inflammatory parameters as ferritin remained significantassociated with severe AD, but this was not the case for hsCRP.

Diabetes mellitus and arterial hypertension are well known CVrisk factors. To ensure that etiology of CKD is an independent factorof severe AD further than diabetes or hypertension, we restrictedour analysis to a populationwith only diabetes or only hypertensivepatients (Table 4, panels A and B). As displayed and despite asubstantial decrease in the sample size, the results remained un-changed, which reinforce our findings in the overall population.

The estimated prevalence of carotid severe AD based on themultivariate logistic regression model including gender, etiologiesof CKD and age effect is showed in Fig. 2. As illustrated, age andgender showed an important effect in the presence of severe AD.The presence of severe AD was higher with age, although this as-sociation was weaker for women than for men (p < 0.001). Thepresence of severe AD was clearly higher among patients withdiabetic nephropathy than among patients with others any causesof CKD (p < 0.001).

4. Discussion

To our knowledge, this is the first study comparing the presenceof silent carotid AD in CKD patients with different etiologies. Theanalyses showed that the absolute and adjusted prevalence rates ofsevere carotid AD were more than two fold higher in patients withdiabetic nephropathy compared to those with any other causes ofkidney disease. Moreover, the prevalence of silent AD increaseswith the severity of CKD, confirming previous findings in the wholecohort [9]. This study also highlights the potential role of imagingtechniques as early diagnostic tools for endothelial damage andcardiovascular risk in CKD population.

The uses of ACEIs/ARB, antiplatelet or statins are established onreducing CV risk [20,21]. For example, in non-dialyzed CKD pa-tients, statins have demonstrated 20% reduction of stroke and

Table 4Adjusted model including only subjects with diabetes n ¼ 626 (Panel A) and onlysubjects with Hypertension n ¼ 2169 (Panel B).

Panel A (diabetic patients) Odd Ratio CI 95% p

Age (risk/year) 1.088 [1.068e1.109] <0.001Gender (male) 1.63 [1.04e2.56] 0.003Smoker 1.32 [0.81e2.14] 0.25Former-smoker 1.32 [0.74e2.34] 0.33Grade 5D of CKDa 2.20 [1.28e3.79] 0.004Etiology of CKD (Diabetic Nephropathy)b 2.01 [1.33e3.03] 0.001Panel B (Hypertensive patients) Odd ratio CI 95% pAge (risk/year) 1.096 [1.085e1.107] <0.001Gender (male) 1.34 [1.07e1.66] 0.008Smoker 2.24 [1.68e2.99] <0.001Former-smoker 1.55 [1.22e1.97] <0.001Grade 5D of CKDa 2.19 [1.74e2.74] <0.001Etiology of CKD (Diabetic Nephropathy)b 2.24 [1.76e2.86] <0.001

a Grade 5D (Dialysis) compared to any other grades of CKD.b DN compared to any other causes of CKD all together.

myocardial infarction [22]. Although the aim of our study was notto assess the use and effectiveness of these drugs in this population,we noticed that patients with diabetic nephropathy were signifi-cantly more treated compared to CKD patients for any other cause(Table 1). However, diabetic nephropathy patients showed thehighest rate of vascular damage, which highlights that primaryprevention measures currently applied in this group may not beenough. Diabetic nephropathy is a microvascular complication ofdiabetes mellitus characterized by initially albuminuria and aprogressive decline of the renal function. It is well known that both,loss of the renal function and albuminuria are established riskfactors for CV events [23]. However, our findings remark that thisrelationship is strongly affected by the causes leading to the renaldamage. For example the group of Systemic/Glom and diabeticnephropathy had the highest values of urine album/creatinine ratiowithout differences between these two groups. Nevertheless, pa-tients with renal damage due to systemic or glomerular diseasesdisplayed the lowest incidences of carotid damage in all the gradesof CKD.

It is well known that elevated calciumephosphorus product aswell as hyperparathyroidism are important risk factors for AD[24,25]. However, those parameters were equally distributedamong patients with absence/incipient AD and severe AD. In thesameway, lipid profile is awell-established CV risk factor but in ourpopulation total cholesterol and LDL-cholesterol displayed similarlevels between the different scores of AD. Due to the cross-sectionalnature of our observation, we cannot rule out the casual associationand discriminate the effect of the treatments received. Otherwise,triglyceride levels were higher among patients with severe AD butthe significance was lost in the adjusted model. These results are inagreement with previous reports, showing that the associationbetween the lipid profile and risk of myocardial infarction is weakerin patients with lower baseline eGFR [26,27]. We also analyzed theinfluence of other non-traditional risk factors such as inflammatoryparameters; ferritin or hsCRP. Our findings agree with previousresults showing an association of ferritin and C-reactive proteinwith coronary atherosclerosis, metabolic syndrome and insulinresistance [28,29]. Increasing ferritin level could indicate chronicsubclinical inflammation, which agrees with the idea that AD is anactive inflammatory process [30].

The etiology of CKD was one of the factors showing a higherweight in the association with severe AD. This factor remainedsignificant even in a more restricted diabetes population. Thisshould reinforce the recommendation of classifying CKD based onetiology [13,27]. Considering those having diabetic nephropathy asthe reference group, the second etiology with higher risk wasvascular nephropathy. Arterial hypertension is the second leadingcause of CKD in the westernworld and an established CV risk factor[1,3,31]. In our findings, other well known risk factors such as age,former and current tobacco use, body mass index or advancedgrade of CKD were associated with an increase in severe AD prev-alence, independently of other factors. Notably, the well-knownprotective effect of female sex was still present in CKD patients.

Our study has some limitations. The cross-sectional design doesnot allow us to make causal associations. Due to the nature of thepopulation studied, the distribution of the prevalence of CKD is notrepresentative of the general population with only one third ofpatients in grade 2e3 of CKD. Studies with a larger sample of thosesubjects would help to correlate our findings in patients with earlyrenal function decline. Moreover, patients with previous CV eventswere excluded from the NEFRONA study. It is important to recog-nize that while this remarks the utility of our approach for an earlydetection of patients at risk, we cannot ensure similar results in apopulation with CV disease clinically established and furtherstudies should be addressed to uncover that hypothesis. Finally,

Fig. 2. Estimated prevalence of carotid severe atheromatous disease by the multivariate logistic regression model including gender, etiologies of chronic kidney disease and the ageeffect. Cystic; Polycystic kidney disease, DN; Diabetic Nephropathy, Glom/System: glomerular or systemic renal diseases, TIN; Tubulointerstitial and drug toxicity Nephropathy, VN;Vascular.

C. Barrios et al. / Atherosclerosis 242 (2015) 37e44 43

each physician team recorded the causes of kidney disease ac-cording with the evidences of patient's laboratory and clinicalhistory, but we do not have histological confirmation, which mayintroduce errors in the classification of patients. Since other reliableand non-invasive markers have been established, we cannot over-come this limitation.

In conclusion, patients with diabetes-induced renal disease aremore likely to have endothelial damage beyond the kidneys. Pa-tients with diabetic nephropathy at any grade of CKD are atparticularly high risk of subclinical severe AD compared to anyother causes of CKD. Thereby, it should be considered an increase inprophylactic measures and a closer monitoring of these patients.The use of carotid ultrasound can help in the early detection of AD,particularly in CKD population, in whom the classical risk scoringhas been proven inefficacious.

Author's contributions

Research idea and study design: C.B., J.P., E.F., J.M.V.; dataacquisition: A.F., A.B., E.F., J.M.V.; data analysis/interpretation:C.B., J.P., S.O., M.J.S., E.R., S.C., S.M., A.B., J.F.N.G., E.F., J.M.V.; sta-tistical analysis: C.B., S.M.; supervision or mentorship: J.P., E.F.,J.M.V. J.P., E.V., and J.M.V. are the guarantors of this work and, assuch, had full access to all the data in the study and takes re-sponsibility for the integrity of the data and the accuracy of thedata analysis.

Financial disclosure

The NEFRONA study is funded by a research grant from AbbVieand the Spanish government RETIC (RD12/0021) and FIS PI13/01565. The authors declare that they do not have other relevantfinancial interests.

Acknowledgments

The authors would like to thank the NEFRONA support team andthe Biobank of RedInRen for their invaluable support. The NEFRONAstudy group is composed by a number of investigators depicted in aSupplementary List.

Support: The NEFRONA Study is sponsored by the Spanish So-ciety of Neohrology. Julio Pascual is supported by FIS ISCIII-FEDERPI13/00598, RedinRen RD12/0021/0024 and Programa deIntensificaci�on de la Actividad Investigadora ISCIII. María J Soler issupported by FIS ISCIII-FEDER PI11/01549. Research activity by DrNavarro is supported by Instituto de Salud Carlos III (Programa deIntensificaci�on de la Actividad Investigadora e Convenio ISCIII/Comunidad Aut�onoma Canarias).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atherosclerosis.2015.06.048.

References

[1] Chronic Kidney Disease Prognosis, C, K. Matsushita, M. van der Velde, et al.,Association of estimated glomerular filtration rate and albuminuria with all-cause and cardiovascular mortality in general population cohorts: a collabo-rative meta-analysis, Lancet 375 (2010) 2073e2081.

[2] C.A. Herzog, R.W. Asinger, A.K. Berger, et al., Cardiovascular disease in chronickidney disease. A clinical update from kidney disease: improving global out-comes (KDIGO), Kidney Int. 80 (2011) 572e586.

[3] A.S. Go, G.M. Chertow, D. Fan, et al., Chronic kidney disease and the risks ofdeath, cardiovascular events, and hospitalization, N. Engl. J. Med. 351 (2004)1296e1305.

[4] C. Baigent, M.J. Landray, C. Reith, et al., The effects of lowering LDL cholesterolwith simvastatin plus ezetimibe in patients with chronic kidney disease(study of heart and renal protection): a randomised placebo-controlled trial,Lancet 377 (2011) 2181e2192.

[5] P.M. Ridker, J.S. Group, Rosuvastatin in the primary prevention of cardiovas-cular disease among patients with low levels of low-density lipoproteincholesterol and elevated high-sensitivity C-reactive protein: rationale and

C. Barrios et al. / Atherosclerosis 242 (2015) 37e4444

design of the JUPITER trial, Circulation 108 (2003) 2292e2297.[6] J.F. Mann, H.C. Gerstein, J. Pogue, et al., Renal insufficiency as a predictor of

cardiovascular outcomes and the impact of ramipril: the HOPE randomizedtrial, Ann. Intern Med. 134 (2001) 629e636.

[7] B. Coll, A. Betriu, M. Martinez-Alonso, et al., Cardiovascular risk factors un-derestimate atherosclerotic burden in chronic kidney disease: usefulness ofnon-invasive tests in cardiovascular assessment, Nephrol. Dial. Transpl. 25(2010) 3017e3025.

[8] R. Komorovsky, A. Desideri, Carotid ultrasound assessment of patients withcoronary artery disease: a useful index for risk stratification, Vasc. Health RiskManag. 1 (2005) 131e136.

[9] A. Betriu, M. Martinez-Alonso, M.V. Arcidiacono, et al., Prevalence of sub-clinical atheromatosis and associated risk factors in chronic kidney disease:the NEFRONA study, Nephrol. Dial. Transpl. 29 (7) (Jul 2014) 1415e1422.

[10] D. Arroyo, A. Betriu, M. Martinez-Alonso, et al., Observational multicenterstudy to evaluate the prevalence and prognosis of subclinical atheromatosis ina Spanish chronic kidney disease cohort: baseline data from the NEFRONAstudy, BMC Nephrol. 15 (2014) 168.

[11] A. Ichihara, N. Yamashita, T. Takemitsu, et al., Cardio-ankle vascular index andankle pulse wave velocity as a marker of arterial fibrosis in kidney failuretreated by hemodialysis, Am. J. Kidney Dis. 52 (2008) 947e955.

[12] J. Blacher, A.P. Guerin, B. Pannier, et al., Arterial calcifications, arterial stiffness,and cardiovascular risk in end-stage renal disease, Hypertension 38 (2001)938e942.

[13] K.M. Andrassy, Comments on 'KDIGO 2012 clinical practice guideline for theevaluation and management of chronic kidney disease', Kidney Int. 84 (2013)622e623.

[14] D.B. Jovanovic, L. Djukanovic, Analysis of factors influencing chronic renalfailure progression, Ren. Fail 21 (1999) 177e187.

[15] R. Haynes, N. Staplin, J. Emberson, et al., Evaluating the contribution of thecause of kidney disease to prognosis in CKD: results from the study of heartand renal protection (SHARP), Am. J. Kidney Dis. 64 (1) (Jul 2014) 40e48.

[16] A. Betriu-Bars, E. Fernandez-Giraldez, Carotid ultrasound for the early diag-nosis of atherosclerosis in chronic kidney disease, Nefrologia 32 (2012) 7e11.

[17] M. Junyent, M. Martinez, M. Borras, et al., Predicting cardiovascular diseasemorbidity and mortality in chronic kidney disease in Spain. The rationale anddesign of NEFRONA: a prospective, multicenter, observational cohort study,BMC Nephrol. 11 (2010) 14.

[18] M. Junyent, R. Gilabert, I. Nunez, et al., Carotid ultrasound in the assessment of

preclinical atherosclerosis. Distribution of intima-media thickness values andplaque frequency in a Spanish community cohort, Med. Clin. Barc. 125 (2005)770e774.

[19] M. Junyent, M. Martinez, M. Borras, et al., Usefulness of imaging techniquesand novel biomarkers in the prediction of cardiovascular risk in patients withchronic kidney disease in Spain: the NEFRONA project, Nefrologia 30 (2010)119e126.

[20] B.J. Hoogwerf, Renin-angiotensin system blockade and cardiovascular andrenal protection, Am. J. Cardiol. 105 (2010) 30Ae35A.

[21] K. Utsunomiya, Treatment strategy for type 2 diabetes from the perspective ofsystemic vascular protection and insulin resistance, Vasc. Health Risk Manag.8 (2012) 429e436.

[22] S.C. Palmer, S.D. Navaneethan, J.C. Craig, et al., HMG CoA reductase inhibitors(statins) for dialysis patients, Cochrane Database Syst. Rev. 9 (2013)CD004289.

[23] B.R. Hemmelgarn, B.J. Manns, A. Lloyd, et al., Relation between kidney func-tion, proteinuria, and adverse outcomes, JAMA 303 (2010) 423e429.

[24] M. Abe, K. Okada, M. Soma, Mineral metabolic abnormalities and mortality indialysis patients, Nutrients 5 (2013) 1002e1023.

[25] C. Huang, S.A. Shapses, X. Wang, Association of plasma parathyroid hormonewith metabolic syndrome and risk for cardiovascular disease, Endocr. Pract.Official J. Am. Coll. Endocrinol. Am. Assoc. Clin. Endocrinol. 19 (2013)712e717.

[26] M. Tonelli, P. Muntner, A. Lloyd, et al., Association between LDL-C and risk ofmyocardial infarction in CKD, J. Am. Soc. Nephrol. JASN 24 (2013) 979e986.

[27] M.K. Elliott, J.A. McCaughan, D.G. Fogarty, Do patients with chronic kidneydisease get optimal cardiovascular risk reduction? Curr. Opin. Nephrol.Hypertens. 23 (2014) 267e274.

[28] A.S. Gonzalez, D.B. Guerrero, M.B. Soto, et al., Metabolic syndrome, insulinresistance and the inflammation markers C-reactive protein and ferritin, Eur.J. Clin. Nutr. 60 (2006) 802e809.

[29] K.C. Sung, S.M. Kang, E.J. Cho, et al., Ferritin is independently associated withthe presence of coronary artery calcium in 12,033 men, Arterioscler. Thromb.Vasc. Biol. 32 (2012) 2525e2530.

[30] P. Libby, I. Tabas, G. Fredman, et al., Inflammation and its resolution as de-terminants of acute coronary syndromes, Circulation Res. 114 (2014)1867e1879.

[31] M. Tonelli, M.C. Riella, Chronic kidney disease: chronic kidney disease and theageing population, Nat. Rev. Nephrol. 10 (2014) 127e128.