Traditional and nontraditional risk factors as predictors of cerebrovascular events in patients with end stage renal disease

Page 1

C

2468 Original article

Traditional and nontraditional

risk factors as predictors ofcerebrovascular events in patients with end stagerenal diseaseGiovanni Tripepia, Francesco Mattace-Rasob, Francesco Rapisardac,Benedetta Stancanellid, Lorenzo Malatinod, Jacqueline Wittemanb,Carmine Zoccalia and Francesca Mallamacia

Objectives and methods End stage renal disease (ESRD)

patients exhibit a higher risk of cerebrovascular events as

compared with the general population. In 283 ESRD

patients followed up for 10 years, we investigated the

long-term predictive value for stroke and transient ischemic

attacks of traditional and nontraditional risk factors. Data

analysis was performed by a modified Cox regression

analysis for repeated events and by a competing risks

analysis.

Results During the follow-up, 61 cerebrovascular events

occurred in 47 patients. On univariate Cox analysis, the risk

of cerebrovascular outcomes was directly related to age,

smoking, diabetes, BMI, systolic and pulse pressures,

triglycerides, hemoglobin, history of stroke/transient

ischemic attacks, arrhythmia and left ventricular mass

index. Nontraditional risk factors in ESRD such as

norepinephrine, homocysteine, interleukin-6 and

asymmetric dimethylarginine failed to predict these events.

In a multivariate Cox model for repeated events only

smoking [hazard ratio: 2.45, 95% confidence interval (CI):

1.29–4.65], age (hazard ratio: 1.05, 95% CI: 1.01–1.08),

hemoglobin (hazard ratio: 1.28, 95% CI 1.06–1.54),

triglycerides (hazard ratio: 1.04, 95% CI 1.01–1.08), pulse

pressure (hazard ratio: 1.53, 95% CI 1.01–2.23) and left

ventricular mass index (hazard ratio: 1.02, 95% CI 1.01–

1.04) maintained an independent relationship with

cerebrovascular events. The direct link between

hemoglobin and cerebrovascular events was significantly

stronger (P < 0.05) than that of the same variable and death.

opyright © Lippincott Williams & Wilkins. Unautho

0263-6352 � 2010 Wolters Kluwer Health | Lippincott Williams & Wilkins

Conclusion The risk of stroke in ESRD depends mainly on

traditional risk factors, high hemoglobin and left ventricular

hypertrophy. Multiple interventions aimed to reduce arterial

stiffness, left ventricular mass and smoking as well as to

maintain hemoglobin within the recommended therapeutic

range may have beneficial effects on the risk of

cerebrovascular events in ESRD patients. J Hypertens

28:2468–2474 Q 2010 Wolters Kluwer Health | Lippincott

Williams & Wilkins.

Journal of Hypertension 2010, 28:2468–2474

Keywords: asymmetric dimethylarginine, cerebrovascular events,homocysteine, interleukin 6, norepinephrine

Abbreviations: ADMA, asymmetric dimethyl arginine; BMI, body massindex; BP, blood pressure; CAPD, chronic ambulatory peritoneal dialysis;CREED, Cardiovascular Risk Extended Evaluation in Dialysis patients; ESRD,end stage renal disease; HPLC, high performance liquid chromatography;HR, hazard ratio; IL-6, interleukin 6; LVH, left ventricular hypertrophy; LVMI,left ventricular mass index; TIA, transient ischemic attack

aCNR-IBIM, Reggio Calabria, Italy, bDepartment of Internal Medicine, Departmentof Epidemiology, Erasmus MC of Rotterdam, The Netherlands, cInstitute ofInternal Medicine ‘L. Condorelli’ and dClinica Medica, Universita degli studi diCatania, Ospedale Cannizzaro, Catania, Italy

Correspondence to Professor Carmine Zoccali, CNR-IBIM c/o EUROLINE, diAscrizzi Vincenzo, Via Vallone Petrara 55-57, Reggio Calabria 89132, ItalyTel: +39 0965 397 002; fax: +39 0965 26879; e-mail: [email protected]

This study was supported by grants of the National Research Council (CNR),Regione Calabria, Department of Health and Fondazione per le Malattie Renali el’Ipertensione of Reggio Calabria, Italy.

Received 6 May 2010 Revised 16 July 2010Accepted 26 July 2010

IntroductionEnd stage renal disease (ESRD) patients exhibit a higher

risk of cerebrovascular events as compared with the

general population [1] and stroke is recognized as a major

cause of death and disability in the dialysis population.

Observational and experimental research on stroke in

ESRD is still limited. The issue is of particular relevance

because recent clinical trials of statins [2,3] and of dar-

bopoietin [4] have exposed a disturbing increase in the

risk for stroke by these interventions, thus making

research on cerebrovascular disease a clinical and public

health priority in this population. In this regard, it is

important noting that risk factors for cerebrovascular

complications identified so far do not fully account for

the high incidence rate of these events in these patients

[1,5–10]. In the last decade, nontraditional risk factors

such as anemia [11], inflammation [12], hyperhomocys-

teinemia [13], high sympathetic activity [14,15] and

endothelial dysfunction as assessed by asymmetric

dimethylarginine (ADMA) [16,17] have emerged as risk

factors of paramount importance in ESRD, but their

predictive value for strokes and transient ischemic attacks

(TIA) in this population is still unknown. Furthermore,

left ventricular hypertrophy (LVH) is now considered as

the strongest predictor of death in ESRD [18,19].

ESRD patients frequently suffer from recurrent cerebro-

vascular events [20], but studies performed so far in the

rized reproduction of this article is prohibited.

DOI:10.1097/HJH.0b013e32833eaf49

Page 2

C

Cerebrovascular events in ESRD patients Tripepi et al. 2469

dialysis population focused only on the time to the first

event analysis [7–10]. Because population-based obser-

vations show that about 10% of individuals who experi-

ence an acute TIA or a minor stroke have a subsequent

cerebrovascular event within 3 months [21] and because

only 1/3 of stroke episodes are fatal in ESRD patients [9],

survival data analysis restricted to the first cerebro-

vascular event may generate a significant loss of

prognostic information.

We modeled the long-term predictive value for cerebro-

vascular outcomes of traditional (Framingham) and non-

traditional risk factors by a modified Cox regression

analysis for repeated outcomes that allows investigation

of multiple failure-time data of ordered events of the

same type [22]. Furthermore, because death from

causes other than stroke precludes the occurrence of

cerebrovascular outcomes, the independent predictors

of cerebrovascular events in ESRD patients were also

investigated by a competing risks model accounting for

nonstroke death [23].

MethodsProtocolThe protocol was in conformity to the ethical guidelines

of our institutions and informed consent was obtained

from each participant.

Study cohortAn incident-prevalent cohort of 283 ESRD patients [231

on haemodialysis and 52 on chronic ambulatory perito-

neal dialysis (CAPD)] who had been on regular dialysis

treatment for at least 6 months with left ventricular

ejection fraction more than 35%, without cardiac circu-

latory congestion, major infections or intercurrent ill-

nesses requiring hospitalization was considered eligible

for the Cardiovascular Risk Extended Evaluation in

Dialysis patients (CREED) study. All patients were

virtually anuric (diuresis <200 ml/day). Enrolled patients

represented about 70% of the whole dialysis population

treated (between January 1997 and February 1998) in two

urban areas (Reggio Calabria and Catania) in southern

Italy. The remaining 30% of patients were excluded

because of the presence of circulatory congestion or major

infections or because they were hospitalized for inter-

current illnesses (20%) or for logistic reasons/unwilling-

ness to participate in the study (10%).

Follow-up studyFor the purpose of this study, we considered fatal and

nonfatal cerebrovascular events (stroke or TIA) and death

occurring from January 1997 to August 2007. Detailed

analyses of etiological and prognostic factors explaining

all-cause death and cardiovascular events in the CREED

cohort are reported elsewhere [6,11–14,16]. The diag-

nosis of stroke was made on the basis of computed

tomography, MRI and/or clinical and neurological exam-

opyright © Lippincott Williams & Wilkins. Unauth

ination including patient’s history and evaluation of

symptoms surrounding the episode. Each event was

reviewed and assigned an underlying cause by a panel

of five physicians. TIA was defined according to the

revision of the National Institutes of Health classification

document [24], as a sudden, focal neurological deficit

lasting less than 24 h and confined to an area of the brain

or eye perfused by a specific artery. Neurological deficits

include hemiparesis, hemiparesthesia, dysarthria, dys-

phasia, diplopia, perioral numbness, imbalance, and

monocular blindness.

Laboratory measurementsLipids, albumin, calcium, phosphate and Hb were

measured by using standard methods. Serum levels of

interleukin-6 (IL-6) were measured by enzyme-linked

immunosorbent assay with the use of Quantikine High

Sensitivity kits (R&D Systems Inc, Minneapolis,

USA) (normal range <3.12 pg/ml). The plasma concen-

tration of ADMA was determined by high performance

liquid chromatography (HPLC) (normal range<2.2 mol/l).

The plasma concentration of norepinephrine was

measured by a commercially available RIA kit (Ami-

cyl-test TM, Immunological Laboratories, Hamburg,

Germany) (normal range <3.54 nmol/l). Plasma total

homocysteine was determined by an HPLC method

based on ammonium-7-fluorobenzo-2-oxa-1,3-diazole-

4-sulphonate fluorescence derivatization (normal range

<15 mmol/l).

EchocardiographyThese studies were performed on a nondialysis day for

haemodialysis patients and at empty abdomen for CAPD

patients within 2 h after blood sampling. Left ventricular

mass (LVM) was calculated according to the Devereux

formula and indexed to height2.7 [left ventricular mass

index (LVMI)] [25]. LVH was defined by a LVMI more

than 47 g/m2.7 in women or more than 50 g/m2.7 in men.

The height-based indexing of LVM minimizes any

potential distortion attributable to extracellular volume

expansion (BSA indexing being weight-sensitive) and

provides more accurate prognostic information than that

by BSA in the dialysis population [18].

Blood pressure measurementsIn haemodialysis patients, predialysis and postdialysis

blood pressures (BPs) were calculated as the average

value of all recordings [12 measurements (i.e. three/

week)] taken during the month preceding the study.

Because the reproducibility of the mean value of pre-

dialysis and postdialysis BPs approaches that of 24 h

ambulatory blood pressure monitoring better than that

of predialysis and postdialysis BPs [26], we considered

this average value for global statistical assessment. In

CAPD patients, BP values were obtained by averaging

home blood pressure measurements (10–20 measure-

ments/month).

orized reproduction of this article is prohibited.

Page 3

C

2470 Journal of Hypertension 2010, Vol 28 No 12

Statistical analysisData are reported as mean � SD, median and interquar-

tile range or as percentage frequency, as appropriate.

Because more than one episode of cerebrovascular events

can occur in a single patient and because the occurrence

of death from causes other than stroke precludes the

occurrence of subsequent cerebrovascular events, the

independent prognostic factors for these outcomes was

assessed by a modified Cox regression analysis using a

conditional risk set method (time from entry) that allows

investigation of multiple failure-time data of ordered

events of the same type [22] and by a competing risks

model accounting for nonstroke death [23]. To assess the

prognostic performance (precision) of the two models for

predicting cerebrovascular events [conditional risk set

approach (repeated events analysis) versus competing

risks method (a kind of first event analysis accounting

for competing risks)] we compared the widths of confi-

dence intervals (CIs) of the hazard ratios of each covariate

provided by the two approaches.

In a first step, we performed unadjusted Cox regression

analyses for cerebrovascular outcomes by considering all

risk factors listed in Table 1. Multiple Cox regression

models were constructed by including all covariates

opyright © Lippincott Williams & Wilkins. Unautho

Table 1 Characteristics of patients and unadjusted Cox regression ana

Cox regreDescriptive data Units of in

Age (years) 61.0�15.4 1 yearDuration of dialysis (months) 41 (18–93) 10 monthTreatment modality (CAPD/HD) 52/231 0¼HD; 1Male sex [n (%)] 158 (55.8) 0¼ femaleSmokers [n (%)] 117 (41.3) 0¼ no; 1¼Diabetics [n (%)] 43 (15) 0¼ no; 1¼BMI (kg/m2) 24.9�4.4 1 kg/m2

Systolic pressure (mmHg) 132.0�22.7 10 mmHgDiastolic pressure (mmHg) 74.3�12.2 10 mmHgPulse pressure (mmHg) 57.7�16.5 10 mmHgHeart rate (beats/min) 80.8�10.7 5 beats/mOn antihypertensive treatment [n (%)] 117 (41.3%) 0¼ untreaOn treatment with erythropoietin [n (%)] 149 (52.7%) 0¼ no; 1¼EPO dose (U/week) 6000 (0–18000) 3000 U/wCholesterol (mg/dl) 207.6�56.2 10 mg/dlTriglycerides (mg/dl) 176.4�89.9 10 mg/dlHDL cholesterol (mg/dl) 41.1�11.7 5 mg/dlLDL cholesterol (mg/dl) 128.0�52.0 10 mg/dlHemoglobin (g/dl) 10.7�2.0 1 g/dlCalciumMphosphate (mol2/l2) 4.43�1.13 1 mmol2/l2

Albumin (g/dl) 4.0�0.5 1 g/dlPrevious cerebrovascular complicationsMyocardial infarction [n (%)] 35 (12.4) 0¼ no; 1¼Stroke/TIA [n (%)] 42 (15) 0¼ no; 1¼Angina pectoris [n (%)] 82 (29.0) 0¼ no; 1¼Arrhythmia [n (%)] 22 (7.8) 0¼ no; 1¼Peripheral vascular disease [n (%)] 42 (15) 0¼ no; 1¼LVMI (g/m2.7) 64�20 1 g/m2.7

LVEF (%) 58�10 1%Norepinephrine (pg/ml) 561 (314–993) 100 pg/mHomocysteine (mol/l) 27.3 (19.9–41.2) 1 mol/lIL-6 (pg/ml) 5.9 (2.9–9.8) 3 pg/mlADMA (mol/l) 3.03 (1.73–4.08) 1 mol/l

The relationship between each risk factor and cerebrovascular events is expressed as hCAPD, chronic ambulatory peritoneal dialysis; CI, confidence interval; HD, hemodialysis;left ventricular ejection fraction; LVMI, left ventricular mass index.

which were related (P< 0.05) to cerebrovascular events

at univariate analysis. Unadjusted (crude) and adjusted

point estimates of the probability of cerebrovascular

events associated to hemoglobin (Hb) levels were calcu-

lated by using the equations derived from Cox regression

analyses. The adjusted estimates were calculated by a

(multiple) Cox equation in which all terms (see Table 2)

but Hb were set to the corresponding average value. CIs

of estimated probabilities were calculated by the standard

formula [27]. Correction for optimism in overfitted sur-

vival models was performed by the shrinkage method

[28]. Data are expressed as hazard ratio, 95% CI and Pvalue. All calculations were done using a standard stat-

istical package (STATA 9, StataCorp LP, Texas, USA).

Study powerThe study power was calculated by using standard soft-

ware (NCSS-PASS, Utah, USA). In previous studies in

the CREED cohort over a shorter follow-up, we found

that IL-6 [9], ADMA [29] and LVM [18] provided 7.0, 3.3

and 14.0% additional explanatory power, respectively, for

cardiovascular events when added to Cox models includ-

ing traditional risk factors. With this background in mind

and by considering an expected cerebrovascular event

rate of 4.8 events/100 person-years [1], we calculated that

rized reproduction of this article is prohibited.

lysis for cerebrovascular events (conditional risk set approach) [22]

ssion analysis (unadjusted)crease Cerebrovascular events (HR, 95% CI and P)

1.05 (1.03–1.08), P<0.001s 1.00 (0.96–1.04), P¼0.93¼CAPD 1.43 (0.67–3.06), P¼0.36; 1¼male 0.86 (0.48–1.53), P¼0.60

yes 2.51 (1.42–4.44), P¼0.002yes 2.13 (1.08–4.18), P¼0.03

1.05 (1.00–1.11), P¼0.051.14 (1.00–1.30), P¼0.050.89 (0.68–1.16), P¼0.381.35 (1.16–1.58), P<0.001

in 0.92 (0.67–1.27), P¼0.60ted; 1¼ treated 0.86 (0.45–1.65), P¼0.65

yes 0.65 (0.36–1.16), P¼0.15eek 0.96 (0.89–1.04), P¼0.36

0.98 (0.93–1.04), P¼0.511.03 (1.01–1.06), P¼0.0030.98 (0.88–1.08), P¼0.680.96 (0.90–1.02), P¼0.171.16 (1.01–1.32), P¼0.041.08 (0.89–1.31), P¼0.440.76 (0.46–1.25), P¼0.27

yes 1.53 (0.70–3.34), P¼0.28yes 2.72 (1.40–5.28), P¼0.003yes 1.10 (0.62–1.97), P¼0.74yes 3.09 (1.27–7.50), P¼0.01yes 1.86 (0.87–3.97), P¼0.11

1.02 (1.01–1.04), P¼0.0050.99 (0.96–1.02), P¼0.62

l 1.00 (0.95–1.06), P¼0.911.08 (0.94–1.10), P¼0.661.00 (0.96–1.03), P¼0.841.09 (0.96–1.25), P¼0.18

azard ratio, 95% confidence interval and P. ADMA, asymmetric dimethyl arginine;HDL, high-density lipoprotein; HR, hazard ratio; LDL, low-density lipoprotein; LVEF,

Page 4

C

Cerebrovascular events in ESRD patients Tripepi et al. 2471

Table 2 Multivariate Cox regression model for cerebrovascular events (conditional risk set approach)[22]

Units of increase HR and 95% CI P HR and 95% CI (shrinkage corrected) [28]

Smoking 0¼ no; 1¼ yes 2.45 (1.29–4.65) 0.006 2.07 (1.09–3.92)Age 1 year 1.05 (1.01–1.08) 0.009 1.04 (1.01–1.07)Hemoglobin 1 g/dl 1.28 (1.06–1.54) 0.01 1.22 (1.01–1.47)Triglycerides 10 mg/dl 1.04 (1.01–1.08) 0.03 1.03 (1.01–1.07)Pulse pressure 10 mmHg 1.53 (1.01–2.33) 0.04 1.41 (1.00–2.15)LVMI 1 g/m2.7 1.02 (1.01–1.04) 0.04 1.02 (1.00–1.04)History of arrhythmia 0¼ no; 1¼ yes 2.42 (0.82–7.10) 0.11 2.05 (0.70–6.01)History of stroke/TIA 0¼ no; 1¼ yes 1.88 (0.82–4.29) 0.14 1.67 (0.73–3.81)Systolic arterial pressurea 10 mmHg 0.80 (0.57–1.14) 0.21 0.84 (0.59–1.18)BMI 1 kg/m2 1.01 (0.93–1.10) 0.73 1.01 (0.93–1.10)Diabetes 0¼ no; 1¼ yes 0.96 (0.40–2.31) 0.92 0.96 (0.40–2.32)

Data are expressed as HR, 95% CI and P value. CI, confidence interval; HR, hazard ratio; LVMI, left ventricular mass index; TIA, transient ischemic attack. a Systolic pressurefailed to significantly predict cerebrovascular events (P¼0.61) also in a model excluding pulse pressure.

Fig. 1

Unadjusted (upper panel) and adjusted (bottom panel) 10-yearprobability of cerebrovascular events associated to hemoglobin levels.Data are point estimates and 95% confidence intervals (see statisticalanalysis for more details).

a sample including at least 246 ESRD patients achieved

80% power to detect as significant (a-error¼0.01) the

associations between IL-6, homocysteine, norepi-

nephrine, ADMA and LVM and the incidence rate of

cerebrovascular outcomes.

ResultsPatient characteristicsThe descriptive data of the study cohort are presented in

Table 1. One hundred and seventeen patients were

habitual smokers (23� 16 cigarettes/day) and 15% were

diabetic. Ten patients were treated with statins and 62

with antiplatelet/anticoagulant drugs. ADMA, norepi-

nephrine, IL-6 and homocysteine were above the upper

limit of the corresponding normal range in 99 (35%), 135

(48%), 196 (69%) and 241 (85%) ESRD patients, respect-

ively. LVMI was on average 64� 20 g/m2.7 and the large

majority of ESRD patients (77%) displayed LVH

at echocardiography.

Occurrence of cerebrovascular events in the studycohortDuring the follow-up (range 0.03–131 months), 181

patients died. Forty-seven patients experienced at least

one cerebrovascular event (one episode in 39 patients, two

in six patients, three in one patient and seven in one

patient). Overall, 61 cerebrovascular events occurred

(5.1 events/100 person-years) and among these 31 were

fatal (24 ischemic episodes, six hemorrhagic and one

unclassified) and 30 were nonfatal (26 ischemic episodes,

three hemorrhagic and one unclassified). Cerebrovascular

events were ischemic in 50 episodes (82%), hemorrhagic in

nine episodes (15%) and unclassified in the remaining two

episodes (3%). The proportion of patients with a history of

atrial fibrillation was higher in patients with cerebro-

vascular events than in those without (9 versus 2%).

Predictors of cerebrovascular events: univariateanalysisOn univariate Cox regression analysis (conditional risk

set method), the risk of cerebrovascular outcomes was

directly related to age, smoking, diabetes, BMI, systolic

and pulse pressures, triglycerides, Hb (Fig. 1, upper

opyright © Lippincott Williams & Wilkins. Unauth

panel), history of stroke/TIA, history of arrhythmia

and LVMI (Table 1) whereas norepinephrine, total

homocysteine, IL-6 and ADMA failed to predict these

events. Treatments with antiplatelet/anticoagulant

drugs (P¼ 0.21) and statins (P¼ 0.74) were unrelated

to cerebrovascular events. Univariate data analysis

according to competing risks method produced similar

results (data not shown).

Predictors of cerebrovascular events: multivariateanalysesIn a multiple Cox regression model (conditional risk set

method) including all univariate predictors of cerebro-

vascular events (see Table 1), only smoking, age, Hb

(Fig. 1, bottom panel), triglycerides, pulse pressure and

orized reproduction of this article is prohibited.

Page 5

C

2472 Journal of Hypertension 2010, Vol 28 No 12

Table 3 Multivariate Cox regression model for cerebrovascular events and death (competing risks approach) [23]

Units of increase Cerebrovascular events Death MP for equality of hazard ratios

Smoking 0¼ no; 1¼ yes 2.31 (1.11–4.81), P¼0.025 1.61 (1.05–2.46), P¼0.028 0.43Age 1 year 1.05 (1.01–1.09), P¼0.007 1.03 (1.01–1.05), P¼0.001 0.36Pulse pressure 10 mmHg 1.64 (1.05–2.56), P¼0.03 1.08 (0.83–1.41), P¼0.57 0.15Hemoglobin 1 g/dl 1.27 (1.02–1.58), P¼0.03 1.00 (0.91–1.10), P¼0.98 <0.05LVMI 1 g/m2.7 1.02 (1.01–1.04), P¼0.037 1.03 (1.02–1.04), P<0.001 0.49Systolic pressure 10 mmHg 0.71 (0.35–1.44), P¼0.34 0.72 (0.37–1.40), P¼0.33 0.57Diabetes 0¼ no; 1¼ yes 1.44 (0.60–3.48), P¼0.42 1.57 (0.92–2.71), P¼0.10 0.88History of stroke/TIA 0¼ no; 1¼ yes 1.51 (0.64–3.56), P¼0.34 1.39 (0.85–2.29), P¼0.19 0.87History of arrhythmia 0¼ no; 1¼ yes 1.60 (0.50–5.11), P¼0.43 1.39 (0.69–2.80), P¼0.36 0.84Triglycerides 10 mg/dl 1.03 (0.98–1.08), P¼0.25 1.01 (0.99–1.04), P¼0.30 0.60BMI 1 kg/m2 1.00 (0.92–1.09), P¼0.96 0.96 (0.91–1.00), P¼0.06 0.35

Data are expressed as hazard ratio, 95% confidence interval and P. LVMI, left ventricular mass index; TIA, transient ischemic attack. MP value testing hazard ratios equality ofeach covariate with two study outcomes.

LVMI maintained a direct and independent relationship

with these outcomes (Table 2) and this was also true after

correction for overfitting by the shrinkage method [28]

(Table 2, last column). These relationships, but not that

of triglycerides, were confirmed in a competitive risks

analysis (Table 3) accounting for nonstroke death.

Although the two analyses led to substantially similar

results, the widths of CIs of the hazard ratios for cerebro-

vascular events associated with each covariate were about

20% higher in the competing risks model as compared

with those provided by the repeated events model indi-

cating that the latter provided more precise prognostic

estimates than those obtained by the competing risk

model. Arrhythmia, history of stroke/TIA, systolic arterial

pressure, BMI and diabetes were no longer significant

predictors of cerebrovascular outcomes after multivariate

data adjustment in both conditional and competitive risk

models (Tables 2 and 3). The relationship between Hb

and stroke risk did not change also forcing erythropoietin

use into the models. Of note, the direct link between Hb

and cerebrovascular events was significantly stronger

(P< 0.05) than that of the same variable and all-cause

death (Table 3, last column). Separate analysis of males

and females provided similar results.

DiscussionThis study for the first time shows that traditional risk

factors such as age, smoking and pulse pressure as well as

nontraditional risk factors such as Hb and LVM by echo-

cardiography are independent predictors of cerebro-

vascular events in ESRD patients, whereas biomarkers

of inflammation, sympathetic activity and endothelial

dysfunction are unrelated to these outcomes in this

population.

Cerebrovascular outcomes in end stage renal diseasepatientsPatients with ESRD are at high risk for cerebrovascular

disease [30]. According to previous studies in the dialysis

population [1,9,31,32], we found that 17% of total deaths

were owing to cerebrovascular events and that the inci-

dence rate of these outcomes was 5.1 events/100 person-

years. We also found that 82% of cerebrovascular events

opyright © Lippincott Williams & Wilkins. Unautho

were ischemic in nature, which is again in keeping with

previous findings [1,9]. The fatality rate of cerebro-

vascular events was recently investigated in the Choices

for Healthy Outcomes In Caring for ESRD (CHOICE)

study [9], which reported a 35% case-fatality in a mixed

population of black and white ESRD patients. Our study,

the first investigating the fatality rate of cerebrovascular

events in a large cohort composed exclusively of white

ESRD patients, shows that these events were fatal in

about one half of the cases.

Risk factors for cerebrovascular outcomes in end stagerenal disease patientsEven though in previous studies [12–14,16,29], we docu-

mented that IL-6, homocysteine, ADMA and norepi-

nephrine signal a high cardiovascular risk in the dialysis

population, the present analysis shows that they do not

predict cerebrovascular events. According to studies in

the general population [33], we found that smoking

entailed a 2.5 times excess risk for cerebrovascular events

also in ESRD patients. Such an observation contrasts

with previous studies reporting no relationship between

smoking and cerebrovascular events in ESRD [7–9].

However, the follow-up was shorter than that in our study

(5 versus 10 years) in two previous papers [7,8] whereas in

the third study [9] the multiracial composition of the

cohort may have altered the relationship between smok-

ing and cerebrovascular events [34].

Of note, we found that pulse pressure is a significant

predictor of cerebrovascular outcomes in ESRD patients,

which is consistent with previous data [35]. Hypervole-

mia is common in dialysis patients and it is associated

with pulse pressure, which suggests that lowering systolic

pressure through volume control [36] may reduce stroke

risk in this population. To our knowledge, the relation-

ship between Hb and stroke risk in ESRD patients was

specifically investigated only by Seliger et al. [8], who

reported a link between anemia (Hb <9 g/dl) and stroke

risk. In contrast, we found that 1 g/dl increase in Hb

signals a 27% increase in the risk of cerebrovascular

events, an association that was independent also of ery-

thropoietin use. Our finding is consistent with long-term

rized reproduction of this article is prohibited.

Page 6

C

Cerebrovascular events in ESRD patients Tripepi et al. 2473

prospective studies in the general population [37,38]

where high rather than low Hb is a risk factor for cerebro-

vascular events. Furthermore, a recent study in patients

with chronic kidney diseases (CKD) reported an associ-

ation between high Hb and endothelial dysfunction [39],

an alteration predisposing to cerebrovascular disease while

in the Trial to Reduce cardiovascular Events with Aranesp

Therapy in type-2 diabetics with CKD [4], the higher Hb

target was associated with a higher risk for stroke. LVM

was consistently associated with cerebrovascular events in

the general population [40–42] and our data for the first

time show that the predictive power of LVH for the

outcomes extends to ESRD. In contrast, in the CHOICE

study in the same population [9], the occurrence of these

events was unrelated to LVH. We studied an ethnically

homogeneous cohort of white ESRD patients and defined

LVH on the basis of echocardiography whereas the

CHOICE study [9] included a mixed population of black

and white ESRD patients and defined LVH on the basis of

electrocardiogram (EKG) criteria, a method with sensi-

tivity much inferior to that of echocardiography. Overall,

the risk for stroke in our study mainly depended on

traditional risk factors, Hb and LVH.

Our study has limitations. First, the observational nature of

our findings precludes definitive conclusions about the

nature (causal/noncausal) of the associations we registered.

Second, even though our study was adequately powered to

test independent predictors of cerebrovascular events, our

cohort was relatively small. Therefore, our data await

confirmation in larger studies. Strengths of our study

are: it is the only one considering IL-6, ADMA, norepi-

nephrine and homocysteine as potential predictors of

cerebrovascular events in ESRD patients; the accurate

patient monitoring that allowed a complete case ascertain-

ment in 96% of patients displaying cerebrovascular events;

the long follow-up period (10 years) and the robustness of

statistical methods applied to test the study hypotheses

accounting for both failure order (conditional risk set

method) and failure types (competing risks analysis).

This study generates the hypothesis that multiple inter-

ventions aimed at reducing arterial stiffness, LVMI and

smoking as well as at maintaining Hb within the recom-

mended range may have beneficial effects for the pre-

vention of cerebrovascular events in ESRD patients, a

hypothesis that needs to be confirmed in specifically

designed clinical trials.

AcknowledgmentsThere are no conflicts of interest.

References1 Seliger SL, Gillen DL, Longstreth WT Jr, Kestenbaum B, Stehman-Breen

CO. Elevated risk of stroke among patients with end-stage renal disease.Kidney Int 2003; 64:603–609.

2 Wanner C, Krane V, Marz W, Olschewski M, Mann JF, Ruf G, Ritz E,German Diabetes and Dialysis Study Investigators. Atorvastatin in patientswith type 2 diabetes mellitus undergoing hemodialysis. N Engl J Med 2005;353:238–248.

opyright © Lippincott Williams & Wilkins. Unauth

3 Fellstrom BC, Jardine AG, Schmieder RE, Holdaas H, Bannister K, BeutlerJ, et al. Rosuvastatin and cardiovascular events in patients undergoinghemodialysis. N Engl J Med 2009; 360:1395–1407.

4 Pfeffer MA, Burdmann EA, Chen CY, Cooper ME, de Zeeuw D, Eckardt KU,et al. A trial of darbepoetin alfa in type 2 diabetes and chronic kidneydisease. N Engl J Med 2009; 21:2019–2032.

5 Savazzi GM, Cusmano F, Degasperi T. Cerebral atrophy in patient on long-term regular hemodialysis treatment. Clin Nephrol 1985; 23:89–95.

6 Benedetto FA, Tripepi G, Mallamaci F, Zoccali C. Rate of atheroscleroticplaque formation predicts cardiovascular events in ESRD. J Am SocNephrol 2008; 19:757–763.

7 Iseki K, Fukiyama K. Predictors of stroke in patients receiving chronichemodialysis. Kidney Int 1996; 50:1672–1675.

8 Seliger SL, Gillen DL, Tirschwell D, Wasse H, Kestenbaum BR, Stehman-Breen CO. Risk factors for incident stroke among patients with end-stagerenal disease. J Am Soc Nephrol 2003; 14:2623–2631.

9 Sozio SM, Armstrong PA, Coresh J, Jaar BG, Fink NE, Plantinga LC, et al.Cerebrovascular disease incidence, characteristics, and outcomes inpatients initiating dialysis: the choices for healthy outcomes in caring forESRD (CHOICE) study. Am J Kidney Dis 2009; 54:468–477.

10 Vazquez E, Sanchez-Perales C, Garcia-Garcia F, Castellano P, Garcia-Cortes MJ, Liebana A, Lozano C. Atrial fibrillation in incident dialysispatients. Kidney Int 2009; 76:324–330.

11 Zoccali C, Mallamaci F, Tripepi G. Traditional and emerging cardiovascularrisk factors in end stage renal disease. Kidney Int 2003; 85:S105–S110.

12 Tripepi G, Mallamaci F, Zoccali C. Inflammation markers, adhesionmolecules, and all-cause and cardiovascular mortality in patients withESRD: searching for the best risk marker by multivariate modeling. J AmSoc Nephrol 2005; 16 (Suppl1):S83–S88.

13 Mallamaci F, Zoccali C, Tripepi G, Fermo I, Benedetto FA, Cataliotti A, et al.Hyperhomocysteinemia predicts cardiovascular outcomes in hemodialysispatients. Kidney Int 2002; 61:609–614.

14 Zoccali C, Mallamaci F, Parlongo S, Cutrupi S, Benedetto FA, Tripepi G, etal. Plasma norepinephrine predicts survival and incident cardiovascularevents in patients with end-stage renal disease. Circulation 2002;105:1354–1359.

15 Grassi G, Arenare F, Pieruzzi F, Brambilla G, Mancia G. Sympatheticactivation in cardiovascular and renal disease. J Nephrol 2009; 22:190–195.

16 Zoccali C, Bode-Boger S, Mallamaci F, Benedetto F, Tripepi G, Malatino L,et al. Plasma concentration of asymmetrical dimethylarginine and mortalityin patients with end-stage renal disease: a prospective study. Lancet 2001;358:2113–2117.

17 Coen G, Mantella D, Sardella D, Beraldi MP, Ferrari I, Pierantozzi A, et al.Asymmetric dimethilarginine, vascular calcifications and parathyroidhormone serum levels in hemodialysis patients. J Nephrol 2009; 22:616–622.

18 Zoccali C, Benedetto FA, Mallamaci F, Tripepi G, Giacone G, Cataliotti A,et al. Prognostic impact of the indexation of left ventricular mass in patientsundergoing dialysis. J Am Soc Nephrol 2001; 12:2768–2774.

19 McGill RL, Biederman RW, Getts RT, Hazlett SM, Sharma SB, Duran J,et al. Cardiac magnetic resonance imaging in hemodialysis patients.J Nephrol 2009; 22:367–372.

20 Zoungas S, McGrath BP, Branley P, Kerr PG, Muske C, Wolfe R, et al.Cardiovascular morbidity and mortality in the Atherosclerosis and FolicAcid Supplementation Trial (ASFAST) in chronic renal failure: a multicenter,randomized, controlled trial. J Am Coll Cardiol 2006; 47:1108–1116.

21 Hill MD, Yiannakoulias N, Jeerakathil T, Tu JV, Svenson LW, SchopflocherDP. The high risk of stroke immediately after transient ischemic attack: apopulation-based study. Neurology 2004; 62:2015–2020.

22 Therneau TM, Grambsch PM (2000). Modeling survival data: extending theCox model (statistics for biology and health). New York, USA: Springer-Verlag. pp. 169–229.

23 Lunn M, McNeil D. Applying Cox regression to competing risks. Biometrics1995; 51:524–532.

24 National Institute of Neurological and Communicative Disorders andStroke: a classification and outline of cerebrovascular diseases II. Stroke1975; 6:564–616.

25 de Simone G, Daniels SR, Devereux RB, Meyer RA, Roman MJ, de DivitiisO, et al. Left ventricular mass and body size in normotensive children andadults: assessment of allometric relations and impact of overweight. J AmColl Cardiol 1992; 20:1251–1260.

26 Peixoto AJ, Santos SF, Mendes RB, Crowley ST, Maldonado R, Orias M,et al. Reproducibility of ambulatory blood pressure monitoring inhemodialysis patients. Am J Kidney Dis 2000; 36:983–990.

27 Altman DG. Practical statistics for medical research. London: Chapmanand Hall; 1991; pp. 378.

orized reproduction of this article is prohibited.

Page 7

C

2474 Journal of Hypertension 2010, Vol 28 No 12

28 Steyenberg EW (2009). Clinical prediction models: a practical approach todevelopment, validation, and updating. New York, USA: Springer-Verlag.pp. 232–233.

29 Mallamaci F, Tripepi G, Cutrupi S, Malatino LS, Zoccali C. Prognostic valueof combined use of biomarkers of inflammation, endothelial dysfunction,and myocardiopathy in patients with ESRD. Kidney Int 2005; 67:2330–2337.

30 Schiller A, Covic A. Kidney and brain: a renal perspective of ‘Les LiaisonsDangereuses’. Nephrol Dial Transplant 2010; 25:1370–1373.

31 Kowamura M, Fijimoto S, Hisanaga S, Yamamoto Y, Eto T. Incidence,outcome, and risk factors of cerebro-vascular events in patients undergoingmaintanance hemodialysis. Am J Kidney Dis 1998; 31:991–996.

32 Zoccali C, Mallamaci F, Tripepi G, Cutrupi S, Parlongo S, Malatino LS, et al.Fibrinogen, mortality and incident cardiovascular complications in end-stage renal failure. J Intern Med 2003; 254:132–139.

33 Assmann G, Schulte H, Cullen P, Seedorf U. Assessing risk of myocardialinfarction and stroke: new data from the Prospective CardiovascularMunster (PROCAM) study. Eur J Clin Invest 2007; 37:925–932.

34 Pei YP, Greenwood CM, Chery AL, Wu GG. Racial differences in survivalof patients on dialysis. Kidney Int 2000; 58:1293–1299.

35 Tozawa M, Iseki K, Iseki C, Takishita S. Pulse pressure and risk of totalmortality and cardiovascular event in patients on chronic hemodialysis.Kidney Int 2002; 61:717–726.

opyright © Lippincott Williams & Wilkins. Unautho

36 Agarwal R, Alborzi P, Satyan S, Light RP. Dry-weight reduction inhypertensive hemodialysis patients (DRIP): a randomized, controlled trial.Hypertension 2009; 53:500–507.

37 Kannel WB, Gordon T, Wolf PA, McNamara P. Hemoglobin and the risk ofcerebral infarction: the Framingham study. Stroke 1972; 3:409–420.

38 Kiyohara Y, Ueda K, Hasuo Y, Fujii I, Yanai T, Wada J, et al. Hematocrit as arisk factor of cerebral infarction: long-term prospective population survey ina Japanese rural community. Stroke 1986; 17:687–692.

39 Yilmaz MI, Sonmez A, Saglam M, Gulec M, Kilic S, Eyileten T, et al.Hemoglobin is inversely related to flow-mediated dilatation in chronickidney disease. Kidney Int 2009; 75:1316–1321.

40 Bikkina M, Levy D, Evans JC, Larson MG, Benjamin EJ, Wolf PA, CastelliWP. Left ventricular mass and risk of stroke in an elderly cohort: theFramingham Heart Study. JAMA 1994; 272:33–36.

41 Fox ER, Alnabhan N, Penman AD, Butler KR, Taylor HA Jr, Skelton TN,Mosley TH Jr. Echocardiographic left ventricular mass index predictsincident stroke in African Americans: Atherosclerosis Risk in Communities(ARIC) Study. Stroke 2007; 38:2686–2691.

42 Verdecchia P, Porcellati C, Reboldi G, Gattobigio R, Borgioni C, PearsonTA, Ambrosio G. Left ventricular hypertrophy as an independent predictorof acute cerebrovascular events in essential hypertension. Circulation2001; 104:2039–2044.

rized reproduction of this article is prohibited.