Original Article Endothelial progenitor cells in patients on

17. Wittamer V, Franssen JD, Vulcano M et al. Specific recruitmentof antigen-presenting cells by chemerin, a novel processed ligandfrom human inflammatory fluids. J Exp Med 2003; 198:977–985

18. Zabel BA, Allen SJ, Kulig P et al. Chemerin activation by serine pro-teases of the coagulation, fibrinolytic, and inflammatory cascades. JBiol Chem 2005; 280: 34661–34666

19. Cash JL, Hart R, Russ A et al. Synthetic chemerin-derived peptidessuppress inflammation through ChemR23. J Exp Med 2008; 205:767–775

20. Takahashi M, Takahashi Y, Takahashi K et al. Chemerin enhancesinsulin signaling and potentiates insulin-stimulated glucose uptakein 3T3-L1 adipocytes. FEBS Lett 2008; 582: 573–578

21. Wang LY, Wei L, Yu HY et al. Relationship of serum chemerin to obe-sity and type 2 diabetes mellitus. Zhonghua Yi Xue Za Zhi 2009; 89:235–238

22. Sell H, Laurencikiene J, Taube A et al. Chemerin is a novel adipocyte-derived factor inducing insulin resistance in primary human skeletalmuscle cells. Diabetes 2009; 58: 2731–2740

23. Weigert J, Neumeier M, Wanninger J et al. Systemic chemerin isrelated to inflammation rather than obesity in type 2 diabetes. ClinEndocrinol (Oxf) 2009; 72: 342–348

24. Lehrke M, Becker A, Greif M et al. Chemerin is associated with mar-kers of inflammation and components of the metabolic syndrome butdoes not predict coronary atherosclerosis. Eur J Endocrinol 2009;161: 339–344

25. Kaur J, Adya R, Tan BK et al. Identification of chemerin receptor(ChemR23) in human endothelial cells: chemerin-induced endothelialangiogenesis. Biochem Biophys Res Commun 2009; 391: 1762–1768

26. Suliman ME, Stenvinkel P, Qureshi AR et al. Hyperhomocysteinemiain relation to plasma free amino acids, biomarkers of inflammationand mortality in patients with chronic kidney disease starting dialysistherapy. Am J Kidney Dis 2004; 44: 455–465

27. Valli A, Carrero JJ, Qureshi AR et al. Elevated serum levels of S-adenosylhomocysteine, but not homocysteine, are associated with car-diovascular disease in stage 5 chronic kidney disease patients. ClinChim Acta 2008; 395: 106–110

28. Giral P, Jacob N, Dourmap C et al. Elevated gamma-glutamyltransferaseactivity and perturbed thiol profile are associated with features of meta-bolic syndrome. Arterioscler Thromb Vasc Biol 2008; 28: 587–593

29. Stejskal D, Karpisek M, Hanulova Z et al. Chemerin is an independentmarker of the metabolic syndrome in a Caucasian population—a pilotstudy. Biomed Pap Med Fac Univ Palacky Olomouc Czech Repub2008; 152: 217–221

30. Bozaoglu K, Segal D, Shields KA et al. Chemerin is associated withmetabolic syndrome phenotypes in a Mexican–American population.J Clin Endocrinol Metab 2009; 94: 3085–3088

31. Ceccarini G, Flavell RR, Butelman ER et al. PET imaging of leptinbiodistribution and metabolism in rodents and primates. Cell Metab2009; 10: 148–159

Received for publication: 30.3.10; Accepted in revised form: 21.5.10

Nephrol Dial Transplant (2010) 25: 4023–4031doi: 10.1093/ndt/gfq552Advance Access publication 27 October 2010

Endothelial progenitor cells in patients on extracorporeal maintenancedialysis therapy

Detlef H. Krieter1, Regina Fischer1, Karin Merget2, Horst-Dieter Lemke2,3, Andreas Morgenroth4,Bernard Canaud5 and Christoph Wanner1

1Division of Nephrology, Department of Medicine, University of Würzburg, Würzburg, Germany, 2EXcorLab GmbH, Obernburg,Germany, 3Research and Development, Membrana GmbH, Wuppertal, Germany, 4Dialysis Center, Elsenfeld, Germany and5Department of Nephrology, University of Montpellier, Hôpital Lapeyronie, Montpellier, France

Correspondence and offprint requests to: Detlef H. Krieter; E-mail: [email protected]

AbstractBackground. Chronic renal failure patients have a highcardiovascular disease burden, low numbers and impairedfunction of endothelial progenitor cells (EPCs). We hy-pothesized that enhanced uraemic toxin removal restoresEPCs in haemodialysis patients.Methods. In a prospective, randomized, cross-over trial,18 patients were subjected to 4 weeks of each low-fluxhaemodialysis, high-flux haemodialysis and haemodiafil-tration differing in uraemic toxin removal. EPCs were de-termined at baseline and at the end of each 4-week period.A cohort of 16 healthy volunteers served as control. EPCswere studied after culture on fibronectin (CFU-Hill) andcollagen-1 (ECFC).

Results. Dialysis patients had a lower number of ECFCsthan in healthy controls (P < 0.001) and a reduced frac-tion of vital ECFCs (P < 0.05), whereas the formation ofendothelial cell colonies (ECCs) was increased (P < 0.05).Different middle molecular uraemic toxin removal had noeffects on EPC numbers. The number of prototypicalEPCs (CD34+/VEGFR2-KDR+/CD45− ECFCs) wassimilar between patients and controls. Correlations ofplasma C-reactive protein with ECC count, CFU-Hillcolony count and CD34+/VEGFR2-KDR+/CD45− sub-population of both ECFC and CFU-Hill cells wereobserved.Conclusions. Different middle molecule removal has noeffect on EPCs. Reduced vitality and enhanced ECC for-

EPCs in maintenance dialysis patients 4023

© The Author 2010. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.For Permissions, please e-mail: [email protected]

Downloaded from https://academic.oup.com/ndt/article-abstract/25/12/4023/1867162by gueston 12 April 2018

mation suggest growth induction of impaired EPCs inchronic renal failure and are associated with inflammation.

Keywords: atherosclerosis; endothelial progenitor cells; end-stage renaldisease; haemodialysis; uraemic toxins

Introduction

The number of endothelial progenitor cells (EPCs) inblood has been identified as a surrogate biologic markerfor vascular function and cumulative cardiovascular riskin the general population [1]. In patients with coronary ar-tery disease, low numbers of EPCs predict the occurrenceof cardiovascular events and death from cardiovascularcauses [2]. Cardiovascular disease is common in patientson chronic dialysis and represents the principal cause ofdeath in this population. Therefore, the finding that a re-duced number of circulating CD34+ cells, a cell populationthat includes and quantitatively correlates with EPCs [3], issignificantly associated with cardiovascular risks and all-cause mortality in patients on maintenance haemodialysisis not surprising [4].

Uraemic toxins inhibit the in vitro differentiation andfunctional activity of EPCs [3]. Consequently, comparedto the general population, patients with chronic renal fail-ure have reduced numbers and an impaired function of cir-culating EPCs [3,5–7]. These abnormalities aggravate withadvancing chronic kidney disease [7]. In turn, the initiationof dialysis therapy in uraemia ameliorates the number ofEPCs [3]. The quality of the dialysis treatment may signifi-cantly influence EPC quantity and function. In this respect,EPC functions are positively associated with the dialysisdose (Kt/V) and patients on more frequent, i.e. five tosix times per week nocturnal haemodialysis even have re-stored numbers and function of circulating EPCs comparedto conventional haemodialysis patients [5,6]. Furthermore,the levels of uraemic toxins, such as beta2-microglobulin(b2m), inversely correlate with the number of EPCs [8].Consequently, the switch of patients from high-flux haemo-dialysis to online haemodiafiltration is considered to resultin an amelioration of EPC numbers suggesting a beneficialeffect of enhanced middle molecule removal and reducedinflammation triggering effect bymore convective andmorebiocompatible therapy forms [9].

However, most of the present data on EPCs in patientson renal replacement therapy are generated in small, cross-sectional studies with a considerable risk of being biased.Furthermore, there is no uniform definition of EPCs, andseveral studies must be questioned with regard to the ap-plied methods for the identification of EPCs making theinterpretation of their results difficult. To date, there hasbeen no specific unique cell surface marker identified thatpermits the prospective isolation of an EPC [10].

The purpose of the present study was to verify in a pro-spective, randomized setting the effects of extracorporealdialysis procedures differing in middle molecule removalon isolated EPCs as identified by currently acknowledgedmethods.

Materials and methods

Study design

The study design was prospective, randomized and cross-over. Studyapproval was given by the Freiburg Ethics Committee International(registration no. 08/1875).

Patient characteristics

Eighteen stable chronic kidney disease stage 5 patients (mean age 60 ± 14years; 3 female, 15 male) on regular thrice weekly maintenance high-fluxhaemodialysis were enrolled into the study after they had given written in-formed consent. Inclusion criteria were age ≥18 years, maintenance extra-corporeal renal replacement therapy ≥3 months, haematocrit >30%, stableanticoagulation and erythropoietin regimen, no vascular access-relatedproblems and the absence of malignant and ongoing inflammatory orinfectious disease at the time of study enrolment. Exclusion criteria werepregnancy, unstable clinical condition (e.g. cardiac or vascular instability),life expectancy ≤12 months, known coagulation problems and participa-tion in another study. Underlying renal diseases were hypertensive nephro-pathy (n = 5), diabetic nephropathy (n = 3), autosomal dominant polycystickidney disease (n = 3), glomerulonephritis (n = 2), tubulointerstitial neph-ritis (n = 2), nephrectomy (n = 2) and post-renal failure (n = 1). The meanduration on dialysis treatment was 46 ± 29 months (range 11–114 months).The bodymass index averaged at 27.8 ± 4.3 kg/m2. Seventeen patientswerehypertensive, 12 had cardiovascular or peripheral vascular disease, six werediabetics and five patients had hypercholesterolemia. The mean daily urineoutput was 539 ± 719mL including six patients without residual renal func-tion. Seventeen patients had efficient native arteriovenous fistulas, and onepatient had a patent bi-flow dialysis catheter for blood access. The monthbefore and during the trial period, the patient’s concomitant medicationswere continued in an unchanged manner including anticoagulation understudy conditions. All except one patient received erythropoietin subcutane-ously in a weekly average dose of 4500 ± 3400 units (median 4000 units,range 0–15 000 units). Cardiovascular medications included diuretics, beta-blockers, angiotensin-converting enzyme inhibitors, angiotensin receptorantagonists, digitalis, calcium channel blockers and vasodilators.

Study procedures

Starting with a mid-week session, the patients were randomly assigned tosuccessively receive thrice weekly each 4 weeks of low-flux haemodialy-sis (referred as LF-HD; low-flux polysulfone, 1.8 m2, Fresenius MedicalCare AG, Bad Homburg, Germany), 4 weeks of high-flux haemodialysis(referred as HF-HD; high-flux polysulfone, 1.8 m2, Fresenius MedicalCare AG, Bad Homburg, Germany) and 4 weeks of online predilutionhaemodiafiltration (referred as HDF; PUREMA® H high-flux polyether-sulfone, 1.7 m2, Membrana GmbH, Wuppertal, Germany) (Figure 1). HDand HDF were performed using Gambro AK 200 S and AK 200 S Ultramonitors (Gambro AB, Lund, Sweden), which prepare ultrapure dialysate.The dialysate flow (QD) rate was set at 500 mL/min. The infusion flowrate (QI) in HDF was targeted to be 50% of the blood flow rate (QB). Theultrafiltration flow rate (QUF) of each session was set according to theindividual patient’s interdialytic weight gain. Anticoagulation was per-formed by unchanged adoption of the previous routine heparinization.Sixteen patients received standard heparin and two patients fractionatedheparin. For details of the delivered dialysis treatment, refer to Table 1.

Blood sampling and laboratory procedures

Blood samples for the laboratory procedures were drawn from the AVfistula immediately after insertion of the arterial dialysis needle beforethe first session and after completion of each study period. For the meas-urement of urea and b2m, additional samples were obtained at the end ofeach first session from the arterial line of the extracorporeal circuit afterreduction of QB to 50 mL/min for 120 s and the dialysate flow turned off.The respective dialysis doses for urea were quantified by means of theeKt/V [11]. To illustrate differences between the treatment forms with re-spect to middle molecule removal, the reduction ratio (RR) for b2m wascalculated as described elsewhere [12].

Control group

Blood samples were also obtained from a control group consisting of16 healthy volunteers (mean age 54 ± 10 years) of both genders (7 female,

4024 D.H. Krieter et al.


9 male) participating in a blood donation programme after signing in-formed consent. Inclusion criteria were absence of renal impairment,diabetes mellitus and arterial hypertension as well as no history of car-diovascular or peripheral vascular disease. Exclusion criteria weresmoking, malignant and ongoing inflammatory or infectious disease.

Analytical methods

Urea and phosphate (PO4) were determined with a Cobas c 111 analyzer(Roche Instrument Center, Rotkreuz, Switzerland). B2m, albumin andhigh sensitive C-reactive protein (CRP) were measured by laser immuno-nephelometry (BN ProSpec Analyzer, Dade Behring GmbH, Marburg,Germany).

Isolation and culture of EPCs

Peripheral blood mononuclear cells (MNCs) were isolated from hepari-nized whole blood samples by Ficoll (Ficoll-PaqueTM PLUS, StemCellTechnologies, St Katharinen, Germany) density-gradient centrifugation.After washing, isolated cells were subsequently resuspended in growthmedium and plated in two different ways.

In the first approach described earlier by Hill et al. (CFU-Hill) [1],the cells were resuspended in EndoCult® Liquid Medium Kit (StemCellTechnologies, St Katharinen, Germany) and plated on dishes coated withhuman fibronectin (Becton Dickinson, Heidelberg, Germany). Forty-eight hours after an initial preplating step in a fibronectin-coated plate,the non-adherent cells were collected and replated onto fibronectin-

coated plates. After another 72 h, the numbers of colonies were assessedmanually by using a contrasting phase microscope (Axiovert 40 CFL,Zeiss, Göttingen, Germany) under 40-fold magnification. Colonies ofEPCs were identified as multiple thin, flat cells emanating from a cen-tral cluster of rounded cells.

According to the approach proposed by Ingram et al. (endothelialcolony-forming cells, ECFC) [13], isolated MNCs were resuspendedin endothelial growthmedium-2 (EGM-2)medium+ 10%FCS (Promocell,Heidelberg, Germany) and seeded onto tissue culture plates pre-coatedwithtype 1 rat tail collagen (BD 354400 BioCoatTM-6-well plates, Heidelberg,Germany). After 24 h, non-adherent cells and debris were discarded.Adherent cells were washed and then grown with EGM-2 medium, whichwas changed daily until cell harvest after 7 days. Endothelial cell colonies(ECCs) appeared as well-circumscribed monolayers of cobblestone-appearing cells and were visually enumerated with the contrasting phasemicroscope.

Identification of EPCs by fluorescence-activated cell sorting

EPCs were immunophenotyped and enumerated with a FACScan (Bec-ton Dickinson Biosciences, Heidelberg, Germany) flow cytometer andthe CellQuestTMPro software package (Becton Dickinson). After thesupernatant medium and non-adherent cells had been discarded, EPCsweredetached from the respective plates with Accutase (PAA, Colbe, Germany)and a three-colour analysis was performed to characterize the co-expressionof CD34, CD45 and VEGFR2-KDR. A sample of EPC suspension wasincubated with phycoerythrin-Cy5-conjugated CD34 antibodies (PE-

LF-HD HDFHF-HD

LF-HD

LF-HD HF-HD

HF-HD

HDF

HDF

Week 5…8 Week 9...12Week 1…4

Randomisation

Baseli

ne

Blood

sam

ple

Blood

sam

ple

Blood

sam

ple

Blood

sam

ple

Fig. 1. Study design flow chart. Eighteen patients were randomly assigned to consecutively receiving three different treatment modes each lasting for4 weeks. Blood samples were drawn at the beginning and at the end of each treatment period. LF-HD, low-flux haemodialysis; HF-HD, high-fluxhaemodialysis; HDF, predilution haemodiafiltration.

Table 1. Dialysis treatment characteristics and plasma levels of urea, b2m, albumin, PO4 and C-reactive protein

LF-HD HF-HD HDF

Blood flow rate (mL/min) 385 ± 30 381 ± 36 383 ± 28Dialysate flow rate (mL/min) 500 ± 0 500 ± 0 315 ± 63a

Infusion volume (L/session) Not applicable Not applicable 49.5 ± 16.9a

Ultrafiltration volume (L/session) 2.78 ± 1.78 2.62 ± 1.41 2.51 ± 1.20Session duration (min) 268 ± 15 268 ± 15 268 ± 15eKt/V 1.48 ± 0.3b 1.42 ± 0.3 1.38 ± 0.2Urea (day 0/28, mg/dL) 105 ± 23/112 ± 23 109 ± 28/104 ± 24 107 ± 21/106 ± 29b2m reduction ratio (%) −2.8 ± 7.2d 61.0 ± 5.1 68.2 ± 9.3a

b2m (day 0/28, mg/L) 23.0 ± 9.5c/34.2 ± 15.5d 21.8 ± 9.7/24.6 ± 12.9 28.8 ± 14.6a,c/22.7 ± 9.5Albumin (day 0/28, g/L) 39.0 ± 4.9/39.3 ± 2.5 37.5 ± 3.3/39.1 ± 5.0 38.3 ± 3.9/38.0 ± 3.3PO4 (day 0/28, mg/dL) 4.9 ± 1.5/5.1 ± 1.6 4.9 ± 1.5/4.8 ± 1.5 4.5 ± 1.5/4.9 ± 1.5C-reactive protein (day 0/28, mg/L; median/range) 2.5 (0.5–26.3)/2.7 (1.0–26.3) 3.6 (0.9–96.2)/3.5 (0.5–26.3) 1.9 (0.6–118.0) / 3.0 (0.9–96.2)

Data represent mean values ± standard deviation unless otherwise indicated.aP < 0.001 vs LF-HD and HF-HD.bP < 0.05 vs HDF.cP < 0.01 vs day 28.dP < 0.001 vs HF-HD and HDF.



CyTM5-conjugated mouse anti-human IgG1, BD Pharmingen™, Heidel-berg, Germany), fluoresceinisothiocyanat-conjugated CD45 antibodies(BD PharmingenTM, Maus anti-human, IgG1, Heidelberg, Ger-many) and phycoerythrin-conjugated VEGFR2-KDR antibodies [mouseanti-human VEGF R2 (KDR)-PE, R&D Wiesbaden-Nordenstadt, Ger-many]. Blocking experiments with unlabelled antibodies served as con-trols for the three assays and at least 5000 events were counted permeasurement.

In an additional sample, quantitative determination of cells undergo-ing apoptosis was performed using a ‘Cell Viability Solution’ (7-Amino-Aktinomycin D, BD Via-Probe™, Heidelberg, Germany).

Data analysis

Descriptive analysis of the results was performed by calculating meanvalues ± standard deviations (SD) or median and range. Within-subjectbetween-treatment differences were analysed by ANOVA and a Tukeypost hoc test for normally distributed samples and by Kruskal–Wallistest if normal distribution did not apply. Within-subject within-treatmentchanges from baseline were assessed using the two-sided paired t-test.For comparisons of patient samples with the control group, a non-pairedt-test and a Mann–Whitney U-test were used, respectively. Correlationcoefficients were determined for EPCs and pretreatment plasma concentra-

tions of CRP and b2m according to Pearson. A P-value of <0.05 was con-sidered statistically significant. The statistical analysis was performed bymeans of the ‘Minitab 15 Statistical Software’ package (Minitab, Inc., StateCollege, PA, USA).

Results

None of the patients withdrew from the study and no patientdrop-outs were noted.

Treatment data

As a consequence of the online production of infusionfluid (49.5 ± 16.9 L per session), QD in HDF averagedat 315 ± 63 mL/min and was lower (P < 0.001) comparedto both HD modes (500 ± 0 mL/min). Treatment differ-ences were also observed for the dialysis dose eKt/V, theb2m RR and the b2m plasma level. For further details, referto Table 1. In healthy controls, themedian ofCRP (1.7mg/L,

20

15

10

5

0

EC

FC

co

lon

ies/

107

MN

C

Healthycontrols

LF-HDDay 0

LF-HDDay 28

HF-HDDay 0

HF-HDDay 28

HDFDay 0

HDFDay 28

o

Fig. 2. ECFC colony count normalized for 107 MNC. Significantly higher ECC counts of patients versus healthy controls. LF-HD, low-fluxhaemodialysis; HF-HD, high-flux haemodialysis; HDF, predilution haemodiafiltration. °P < 0.05 vs LF-HD, HF-HD and HDF on days 0 and 28;for comparison with the control group, the Mann–Whitney U-test was applied. * represents outlier values.

350000

300000

250000

200000

150000

100000

50000

0

EC

FC

to

tal/1

07 M

NC

Healthycontrols

LF-HDDay 0

LF-HDDay 28

HF-HDDay 0

HF-HDDay 28

HDFDay 0

HDFDay 28

o

Fig. 3. Total ECFC count normalized for 107 MNC. Lower ECFC counts of patients versus healthy controls. LF-HD, low-flux haemodialysis; HF-HD,high-flux haemodialysis; HDF, predilution haemodiafiltration. °P < 0.001 vs LF-HD, HF-HD and HDF on days 0 and 28; for comparison with thecontrol group, the Mann–Whitney U-test was applied. * represents outlier values.



Tab

le2.

Cellcountof

CFU-H

illandECFC

subpop

ulations

norm

alized

for10

7MNC.Medians

andrang

esaregiven

LF-H

DDay

0LF-H

DDay

28HF-H

DDay

0HF-H

DDay

28HDF

Day

0HDF

Day

28Health

ycontrols

CD34

+/VEGFR2-KDR+/CD45

+CFU-H

ill19

471

(547

2–48

146)

2129

6(6285–55

056)

2155

3(529

6–65

275)

2209

7(547

2–48

146)

1600

1(8876–42

460)

2263

9(529

6–59

482)

1377

5(578

0–98

192)

ECFC

4843

(203–1233

1)31

67(73–14

931)

4670

(842–1723

9)68

03(203–1006

5)38

90(73–15

559)

2980

(275–1723

9)95

92a

(201

3–14

913)

CD34

+/VEGFR2-KDR+/CD45

−CFU-H

ill75

7(0–259

0)74

8(0–183

5)66

4(0–4

575)

775

(0–259

0)64

4(0–243

4)94

3(0–457

5)30

4(0–159

1)ECFC

231

(38–13

19)

371

(17–68

7)24

1(25–14

01)

237

(50–13

19)

415

(17–65

1)20

2(13–14

01)

163

(32–56

5)CD34

− /VEGFR2-KDR+/CD45

+CFU-H

ill75

645

(2455

6–33

310

1)11

703

1(2562

5–30

775

7)11

732

2(1107

8–34

498

6)75

645

(2412

3–25

062

0)96

112

(3058

5–29

521

1)11

021

5(2483

3–42

470

5)19

998

2(1372

2–43

214

5)ECFC

1205

8(208–4989

5)14

89(38–94

125)

7715

(499–222

844)

1161

4(151–9734

1)86

71(38–94

125)

6295

(53–22

284

4)56

024b

(970

6–25

473

8)CD34

− /VEGFR2-KDR+/CD45

−CFU-H

ill45

37(216–1433

2)49

40(1590–11

502)

6198

(130

7–18

710)

3671

(136

7–13

702)

5801

(1590–12

866)

5243

(147

0–18

710)

4819

(110

8–10

432)

ECFC

461

(86–94

4)29

6(29–16

34)

612

(26–11

26)

238

(86–91

4)51

6(29–16

34)

213

(27–73

61)

600

(273–316

8)CD34

− /VEGFR2-KDR++/CD45

+CFU-H

ill0 (0–0)

0 (0–0)

0 (0–0

)0 (0–0)

0 (0–0)

0 (0–0)

0 (0–0)

ECFC

0 (0–963

)0 (0–0)

0 (0–2

737)

0 (0–15)

0 (0–385

1)0 (0–658

4)0 (0–0)

a P<0.05.

bP<0.00

1vs

LF-H

D,HF-H

DandHDFon

days

0and28

.



range 0.2–18.0; P = 0.006) and the mean values of b2m(1.7 ± 0.5 mg/L; P < 0.001) and urea (26.0 ± 9.6 mg/dL;P < 0.001) were significantly lower compared to the pa-tients on dialysis.

EPC colony and cell counts

The median of the CFU-Hill colony count in patients ondialysis varied between 5.0 (range 0.0–157.5; HF-HD, day28) and 35.0 (2.5–262.5; HF-HD, day 0) per 107 MNCwithout differences between groups and versus healthycontrols [17.5 (0.0–72.5) per 107 MNC]. In contrast,ECC counts of the dialysis patients [median values between0.0 (0.0–6.5; HDF, day 0) and 0.8 (0.0–17.5; HDF, day 28)per 107 MNC] were significantly (P < 0.05) higher com-pared to the controls [0.0 (0.0 to 0.3) per 107 MNC], inwhich ECFC colony formation was virtually nonexistent(Figure 2).

The medians of the total CFU-Hill cell counts in the treat-ment groups ranged between 801 × 103 (265 × 103 to 1281 ×103; LF-HD, day 28) and 1080 × 103 (110 × 103 to 2079 ×103; HF-HD, day 0) per 107MNC. In healthy controls, it was1620 × 103 (213 × 103 to 2659 × 103) per 107 MNCwithoutbeing significantly different from the dialysis patients (P =0.079). The total ECFC counts were also not different be-tween dialysis treatment groups [from 5.4 × 103 (0.2 × 103

to 101 × 103; LF-HD, day 28) per 107 MNC to 12.7 × 103

(0.9 × 103 60.7 × 103; LF-HD, day 0) per 107 MNC] butwere much lower (P < 0.001) compared to the controls[75.9 × 103 (32.0 × 103 to 318 × 103) per 107 MNC](Figure 3).

EPC subpopulations

Subpopulation counts of CFU-Hill cells and ECFC weresimilar between dialysis modes (Table 2). Significant dif-ferences were only determined versus the healthy controlsfor ECFC subpopulations with the immunophenotypesCD34+/VEGFR2-KDR+/CD45+ and CD34−/VEGFR2-KDR+/CD45+. Particularly for CD34+/VEGFR2-KDR+/CD45− ECFC, which are currently regarded as prototypicalfor EPCs, differences were found neither between treat-ments nor versus controls (Table 2).

The proportion of vital CFU-Hill cells was similar be-tween dialysis patients (93 ± 3%, HDF, day 0, to 96 ± 3%,HDF, day 28) and the control group (97 ± 4%). In contrast,the fraction of vital ECFC was higher (P < 0.05) in controls(86 ± 9%) compared with the dialysis patients which pre-sented a large variation in the degree of apoptosis withoutdifferences between the treatment modes and a cell vitalitybetween 54 ± 25% (LF-HD, day 28) and 68 ± 20% (HF-HD,day 0).

EPCs, inflammation and uraemic toxins

CRP in plasma highly correlated (r = 0.556; P < 0.001) withthe ECFC colony count (Figure 4A) and was also linearlyassociated with the CD34+/VEGFR2-KDR+/CD45− sub-population of CFU-Hill cells (r = 0.372; P < 0.001) andECFC (r = 0.364; P < 0.001) (Figure 4B).

Pretreatment urea and b2m inversely correlated with theCD34+/VEGFR2-KDR+/CD45− subpopulation of ECFCs(r = −0.419; P < 0.001 and r = −0.212; P = 0.034, respect-ively). A negative linear association was also determinedfor the CD34+/VEGFR2-KDR+/CD45+ subpopulation ofECFC (r = −0.381; P < 0.001).

Discussion

In maintenance dialysis patients, cardiovascular disease iscommon and represents the most important cause of death[14,15]. Although the HEMO study was unable to demon-strate an overall beneficial effect of high-flux dialysis onsurvival, patients on this therapy form had a lower relativerisk for cardiac death compared to low-flux dialysis [16].Furthermore, a post hoc analysis of the 4D data indicatedsignificant advantages of increased middle molecule re-moval with respect to survival and to a combined cardio-vascular endpoint [17]. Therefore, the concept that anenhancement of uraemic toxin removal by dialysis treat-ment may restore EPC function and, finally, improve car-diovascular outcome is tempting. In this respect, the resultsof the present study are disappointing. Despite a clear sep-

0

20

40

60

80

100

120

140

0 5 10 15 20

CR

P [

mg

/L]

ECC Count / 107 MNC

r=0.556 p<0.001

0

20

40

60

80

100

120

140

0 200000 400000 600000 800000 1000000 1200000

CR

P [m

g/L

]

r=0.364 p<0.001

CD34+VEGFR2-KDR+CD45- ECFC / 107MNC

A

B

Fig. 4. Pearson correlation between plasma CRP level and EPCs. Cellcount normalized for 107 MNC. (A) CRP vs ECC count. (B) CRP vsCD34+VEGFR2-KDR+CD45− subpopulation of ECFC.



aration in middle molecular uraemic toxin removal verifiedby RRs and the subsequent change of plasma levels ofb2m, the data do not indicate an effect of LF-HD, HF-HDand HDF applied over a 4-week period on isolated andcultivated EPCs.

This finding seems to be in contrast to a previous study,which has demonstrated that HDF is favourably associatedwith the number of EPCs [9]. Although, the rather smallpatient number of the present trial may have contributedto a confounding high variability of EPCs, the discrepancycan be explained by differences and uncertainties in the de-termination of EPCs. There is currently no specific markerto identify and no unique definition for a true EPC [10,18].In the present study, after isolation of MNC and beforeimmunophenotyping and enumeration, EPCs were culti-vated in two different ways, i.e. on fibronectin (CFU-Hill;early outgrowth cells) and collagen (ECFC; late out-growth cells), respectively [1,13]. Both resulting cell popu-lations are involved in neoangiogenesis but only the ECFCprogeny form blood vessels de novo in vivo and particularlyECFC displaying the phenotype CD34+/VEGFR2-KDR+/CD45− are currently regarded as prototypical for a trueEPC [10]. The simple enumeration and immunophenotyp-ing of circulating, not cultivated EPCs as applied inseveral previous studies [8,9,19] is not specific and, thus,inadequate for the identification of true EPCs [10]. Whenexploring the effects of HDF on EPCs, Ramirez et al.defined circulating EPCs as MNC co-expressing CD31,CD14 and VEGFR2-KDR [9]. Cells that co-express CD14among other markers are of hematopoietic progeny but arenot EPCs [10].

Jourde-Chiche et al. identified circulating EPCs based onthe co-expression of CD34, VEGFR2-KDR and CD133 [8],which are highly enriched of hematopoietic progenitor cellsbut fail to specifically isolate ECFC. These cells are in-volved in neoangiogenesis and are predictive biomarkersfor some human diseases, such as coronary artery disease[20]. In their cross-sectional analysis, Jourde-Chiche et al.observed that the number of these cells inversely correlatedwith the level of b2m [8]. Even if we failed to demonstrate abeneficial impact of higher middle molecular toxin removalon EPCs, our data confirm these results. Lower levels ofboth the surrogate marker middle molecule b2m and thesmall solute urea were weakly linearly associated with high-er numbers of several CFU-Hill and ECFC subpopulations,among them, the CD34+/VEGFR2-KDR+/CD45− ECFCprogeny regarded as true EPCs. The lacking effect of en-hanced toxin removal by the more convective dialysisprocedures, i.e. HF-HD and especially HDF, does notcontradict this association because the level of uraemic tox-ins is determined not only by the dialysis treatment but alsoby other parameters, particularly the patients’ residual renalfunction [21], which was preserved in 12 of 18 patients par-ticipating in the study.

The present study only revealed differences between dia-lysis patients and healthy controls but these were restrictedexclusively to ECFCs and did not concern CFU-Hill cells.Consistent with previously published data demonstratingreduced numbers, impaired function and/or augmentedapoptosis of circulating EPCs in maintenance dialysis pa-

tients [3,5–7,19,22,23], the total number of ECFCs as wellas the fraction of vital ECFCs was reduced. In contrast, theECC formation of the dialysis patients was increased com-pared to the controls where it was virtually nonexistent.

The enhanced ECC formation may be interpreted as theresult of EPC induction for vascular repair by stimuli pre-ferentially occurring in dialysis patients. These stimuli maycomprise endothelial lesions in the context of acceleratedatherosclerosis, such as inflammation [24,25], and possiblyalso frequent puncture of the arteriovenous fistula or thedialysis treatment per se. This assumption is endorsed byGeorge et al. who have observed increased colony forma-tion in patients with unstable angina pectoris [26]. In morefavourable conditions outside of the uraemic milieu, suchas in culture with endothelial growth medium on type 1collagen, ECs start forming colonies but due to earlierdamage by uraemic toxins their vitality is impaired leadingto enhanced apoptosis. Since after 24 h non-adherent cellswere always discarded, including particularly those beingapoptotic, the total number of ECFCs after 7 days in cul-ture was lower. Previous studies have shown that in vitroEPC differentiation and functional activity are inhibitedand apoptosis of EPCs is enhanced by uraemic serum[3,8], but our in vitro approach was completely different.However, the assumed state of ineffective ECFC matur-ation requires confirmation. Furthermore, it must be un-derlined that for prototypical EPCs, i.e. ECFCs with theimmunophenotype CD34+/VEGFR2-KDR+/CD45−, dif-ferences were found neither between treatments nor versuscontrols. Significant differences were only determined forthe CD34 +/VEGFR2-KDR+/CD45 + and CD34 − /VEGFR2-KDR+/CD45+ ECFC subpopulations. Cells co-expressing CD45 behave similar to macrophages, whichare well known to express ‘endothelial specific’ proteinsand to participate in regulatory form in neoangiogenesis[10].

Inflammation has been shown to be strongly associatedwith atherosclerosis and cardiovascular outcome in chron-ic renal failure [27,28]. Our data demonstrated a strongcorrelation of the plasma CRP level with the ECC count(r = 0.556; P < 0.001). CRP was also linearly associatedwith the CD34+/VEGFR2-KDR+/CD45− subpopulationof both ECFCs (r = 0.364; P < 0.001) and CFU-Hill cells(r = 0.372; P < 0.001). Disregarding the methodologicaldifferences with respect to previous trials and few con-founding inflammatory incidents occurring over the studyperiod in some patients, such an association has not beenobserved before in patients with chronic renal failure.

Only in the small study by George et al. on subjects withangina pectoris syndromes, a positive correlation of CRP le-vels and the number of circulating EPCs was reported sug-gesting that a systemic inflammatory state could potentiallyfacilitate peripheral endothelial precursor mobilization [26].Although the role of EPCs in coronary artery disease is notfully elucidated [20,29], further evidence for the crucialrole of inflammation in the biological behaviour of blood-derived EPCs comes from in vitro data showing that IL-6receptors are expressed in EPCs and that administration ofIL-6 stimulated EPC proliferation, migration and tube for-mation in a dose-dependent manner [30].



In this regard, other in vitro experiments have producedconflicting results which indicate that EPCs isolated fromblood of healthy volunteers are concentration-dependentlyinhibited in differentiation, survival and function, andpresent increased apoptosis by CRP added to culture[31–33]. However, methodological differences give riseto doubts if these results are directly applicable to ourfindings.

Conclusions

The concept of a favourable effect of enhanced middlemolecule removal by convective extracorporeal renal re-placement therapy on EPCs is not supported by the resultsof the present study. A reduced vitality of EPCs and en-hanced ECC formation in maintenance dialysis patientssuggest growth induction of impaired EPCs in chronicrenal failure. In this respect, inflammation, which is asso-ciated with atherosclerosis, seems to play an important rolein the biological behaviour of blood-derived EPCs besidespotentially other local and systemic stimuli.

Acknowledgements. Part of this study has been published in abstractform and presented at the XXXI Congress of the American Society ofNephrology, 2009, San Diego, USA. The study was supported by a grantfrom Membrana GmbH, Wuppertal, Germany.

Conflict of interest statement. D.H.K. has received lecturer fees fromBaxter Healthcare and Membrana GmbH. H.D.L. is a full-time employeeof Membrana GmbH, Wuppertal, Germany. C.W. serves on the BAXTEREuropean Renal Advisory Board for peritoneal dialysis and received lec-turer fees from Fresenius Medical Care. B.C. has received fees for lec-tures from Fresenius Medical Care, Amgen, Roche and Janssen-Cilag.

References

1. Hill JM, Zalos G, Halcox JPJ et al. Circulating endothelial progenitorcells, vascular function and cardiovascular risk. N Engl J Med 2003;348: 593–600

2. Werner N, Kosiol S, Schiegl T et al. Circulating endothelial progenitorcells and cardiovascular outcomes.NEngl JMed 2005; 353: 999–1007

3. de Groot K, Bahlmann FH, Sowa J et al. Uremia causes endothelialprogenitor cell deficiency. Kidney Int 2004; 66: 641–646

4. Maruyama S, Taguchi A, Iwashima S et al. Low circulating CD34+

cell count is associated with poor prognosis in chronic hemodialysispatients. Kidney Int 2008; 74: 1603–1609

5. Choi JH, Kim KL, Huh W et al. Decreased number and impairedangiogenic function of endothelial progenitor cells in patients withchronic renal failure. Arterioscler Thromb Vasc Biol 2004; 24:1246–1252

6. Chan CT, Li SH, Verma S. Nocturnal hemodialysis is associatedwith restoration of impaired endothelial progenitor cell biology inend-stage renal disease. Am J Physiol Renal Physiol 2005; 289:F679–F684

7. Krenning G, Dankers PY, Drouven JW et al. Endothelial progenitorcell dysfunction in patients with progressive chronic kidney disease.Am J Physiol Renal Physiol 2009; 296: F1314–F1322

8. Jourde-Chiche N, Dou L, Sabatier F et al. Levels of circulating endo-thelial progenitor cells are related to uremic toxins and vascular injuryin hemodialysis patients. J Thromb Haemost 2009; 7: 1576–1584

9. Ramirez R, Carracedo J, Merino A et al. Microinflammation inducesendothelial damage in hemodialysis patients: the role of convectivetransport. Kidney Int 2007; 72: 1008–1013

10. Hirschi KK, Ingram DA, Yoder MC. Assessing identity, phenotype,and fate of endothelial progenitor cells. Arterioscler Thromb VascBiol 2008; 28: 1584–1595

11. Daugirdas JT. Physiologic principles and urea kinetic modeling. In: JTDaugirdas, PG Blake, TS Ing (eds). Part II. Blood-Based Therapies.Handbook of Dialysis IV edn. Wolters Kluwer/Lippincott Williams& Wilkins; pp. 25–58

12. Krieter DH, Morgenroth A, Barasinski A et al. Effects of a polyelec-trolyte additive on the selective dialysis membrane permeability forlow-molecular-weight proteins. Nephrol Dial Transplant 2007; 22:491–499

13. Ingram DA, Mead LE, Tanaka H et al. Identification of a novelhierarchy of endothelial progenitor cells using human peripheraland umbilical cord blood. Blood 2004; 104: 2752–2760

14. Foley RN, Parfrey PS, Sarnak MJ. Epidemiology of cardiovascu-lar disease in chronic renal disease. J Am Soc Nephrol 1998; 9:S16–S23

15. de Jager DJ, Grootendorst DC, Jager KJ et al. Cardiovascular andnoncardiovascular mortality among patients starting dialysis. JAMA2009; 302: 1782–1789

16. Eknoyan G, Beck GJ, Cheung AK et al. Hemodialysis (HEMO) StudyGroup. Effect of dialysis dose and membrane flux in maintenancehemodialysis. N Engl J Med 2002; 347: 2010–2019

17. Krane V, Krieter DH, Olschewski M et al. Dialyzer membrane char-acteristics and outcome of patientswith type 2 diabetes onmaintenancehemodialysis. Am J Kidney Dis 2007; 49: 267–275

18. Yoder MC. Defining human endothelial progenitor cells. J ThrombHaemost 2009; 7: 49–52

19. Ueno H, Koyama H, Fukumoto S et al. Dialysis modality is inde-pendently associated with circulating endothelial progenitor cells inend-stage renal disease patients. Nephrol Dial Transplant 2010; 25:581–586

20. Hristov M, Fach C, Becker C et al. Reduced numbers of circulatingendothelial progenitor cells in patients with coronary artery diseaseassociated with long-term statin treatment. Atherosclerosis 2007;192: 413–420

21. Cheung AK, Rocco MV, Yan G et al. Serum beta-2 microglobulinlevels predict mortality in dialysis patients: results of the HEMOstudy. J Am Soc Nephrol 2006; 17: 546–555

22. Sturiale A, Coppolino G, Loddo S et al. Effects of haemodialysis oncirculating endothelial progenitor cell count. Blood Purif 2007; 25:242–251

23. Schlieper G, Hristov M, Brandenburg Vet al. Predictors of low circu-lating endothelial progenitor cell numbers in haemodialysis patients.Nephrol Dial Transplant 2008; 23: 2611–2618

24. Schiffrin EL, Lipman ML, Mann JF. Chronic kidney disease: effectson the cardiovascular system. Circulation 2007; 116: 85–97

25. Xu Q. Progenitor cells in vascular repair. Curr Opin Lipidol 2007; 18:534–539

26. George J, Goldstein E, Abashidze S et al. Circulating endothelialprogenitor cells in patients with unstable angina: association withsystemic inflammation. Eur Heart J 2004; 25: 1003–1008

27. Stenvinkel P, Heimbürger O, Paultre F et al. Strong association be-tween malnutrition, inflammation, and atherosclerosis in chronicrenal failure. Kidney Int 1999; 55: 1899–1911

28. Wanner C, Drechsler C, Krane V. C-reactive protein and uremia.Semin Dial 2009; 22: 438–441

29. Tousoulis D, Andreou I, Antoniades C et al. Role of inflammationand oxidative stress in endothelial progenitor cell function andmobilization: therapeutic implications for cardiovascular diseases.Atherosclerosis 2008; 201: 236–247

30. Fan Y, Ye J, Shen F et al. Interleukin-6 stimulates circulating blood-derived endothelial progenitor cell angiogenesis in vitro. J CerebBlood Flow Metab 2008; 28: 90–98

31. Verma S, Kuliszewski MA, Li SH et al. C-reactive protein attenuatesendothelial progenitor cell survival, differentiation, and function: fur-ther evidence of a mechanistic link between C-reactive protein andcardiovascular disease. Circulation 2004; 109: 2058–2067

32. Suh W, Kim KL, Choi JH et al. C-reactive protein impairs angio-genic functions and decreases the secretion of arteriogenic chemo-



cytokines in human endothelial progenitor cells. Biochem BiophysRes Commun 2004; 321: 65–71

33. Fujii H, Li SH, Szmitko PE et al. C-reactive protein altersantioxidant defenses and promotes apoptosis in endothelial

progenitor cells. Arterioscler Thromb Vasc Biol 2006; 26:2476–2482

Received for publication: 9.3.10; Accepted in revised form: 17.8.10

Nephrol Dial Transplant (2010) 25: 4031–4041doi: 10.1093/ndt/gfq202Advance Access publication 29 April 2010

Arterial stiffness and functional properties in chronic kidney diseasepatients on different dialysis modalities: an exploratory study

Ada W. Y. Chung1, H. H. Clarice Yang1, Jong Moo Kim1, Mhairi K. Sigrist2, Genevieve Brin2,Elliott Chum2, William A. Gourlay3 and Adeera Levin2

1Department of Cardiovascular Science, Child and Family Research Institute, University of British Columbia, Vancouver, Canada,2Division of Nephrology, University of British Columbia, Vancouver, Canada and 3Division of Urologic Science, University of BritishColumbia, Vancouver, Canada

Correspondence and offprint requests to: Ada W. Y. Chung; E-mail: [email protected]

AbstractBackground. Abnormalities of vascular function and ac-cumulation of oxidative stress have been associated withchronic kidney disease (CKD). Dialysis modalities, peri-toneal dialysis (PD) and haemodialysis (HD) may differen-tially impact on vascular function and oxidative stress.Methods. Patients undergoing living donor transplantationwere studied for vascular stiffness using pulse wave vel-ocity measurements, and inferior epigastric arteries wereharvested to examine in vitro stiffness and functional prop-erties and evidence of oxidative stress. Forty-one patientswere studied representing PD (n = 12), HD (n = 14) andnon-dialysed recipients (n = 15).Results. We demonstrated differences in stiffness from invivo and in vitro measurements such that non-dialysis <HD < PD groups. The stiffness measurements did not cor-relate with duration of CKD nor dialysis duration, but didso with phosphate levels (r = 0.356, P = 0.02). From the invitro isometric force experiments, HD arteries demon-strated decreased contractility and endothelium-dependentrelaxation compared with PD and non-dialysis vessels.Level of oxidative stress (as indicated by the 8-isoprostanelevel) was 30% higher in HD arteries than in PD arteries.Protein expression of inducible nitric oxide synthase,NADPH subunits and xanthine oxidase was upregulatedin HD arteries, while superoxide dismutase was downregu-lated. The compromised vascular function in HD arterieswas improved by pharmacological means that eliminatedoxidative stress.Conclusions. We report associations between vasomotorfunction and oxidative stress in the vasculature of patientsreceiving different dialysis therapies. Oxidative stress,which may be differentially augmented during PD and

HD, may play an important role in the vascular dysfunctionin dialysis populations.

Keywords: chronic kidney disease; haemodialysis; oxidative stress;peritoneal dialysis; vasomotor function

Introduction

Chronic kidney disease (CKD) is awell-recognized risk fac-tor for cardiovascular disease (CVD). Furthermore, theleading cause of death in both CKD and dialysis populationsremains to be CVD [1,2]. There is increasing evidence thatthe cardiac and vascular abnormalities develop early inCKD and become more severe as the disease progressesto end stage. The complex process of vascular calcificationand the consequent arterial stiffening have received substan-tial attention in all CKD populations. However, whether andhow different dialysis modalities might impact on the vascu-lar function has not been the focus of much study.

Vascular dysfunction is characterized by aberrations inendothelial secretion, smooth muscle contractility andmechanical property. Increased arterial stiffness results inelevated left ventricular stress and reduced coronary perfu-sion, and the subsequent hypertrophy and cardiomyopathypredispose to congestive heart failure and sudden death indialysed CKD patients [1–3]. Endothelial dysfunction hasbeen described in the vasculature of patients on both peri-toneal dialysis (PD) and haemodialysis (HD), and is impli-cated as the initial pathological step in the progression ofvascular damage [4,5]. Elevated production of vasocon-strictors, such as norepinephrine, endothelin-1 and angio-

Arterial stiffness in CKD patients on different dialysis modalities 4031

© The Author 2010. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.For Permissions, please e-mail: [email protected]


Documents

Original Article Endothelial progenitor cells in patients on