Rheumatology 2009;48:1197–1201 Advance Access publication 9 June 2009
doi:10.1093/rheumatology/kep130
Reduced number of endothelial progenitor cells is predictive of early relapse in anti-neutrophil cytoplasmic antibody-associated vasculitis Jakub Za´vada1,2, Lindy Kideryova´3, Robert Pytlı´ k3, Zdenˇka Hrusˇ kova´1 and Vladimı´ r Tesarˇ 1 Objectives. ANCA-associated vasculitis (AAV) is an inflammatory disorder of small- to medium-sized vessels with relapsing/remitting progression. Microvascular endothelial injury is a major feature of AAV. Circulating endothelial progenitor cells (EPCs) may provide an endogenous repair mechanism to counteract ongoing endothelial damage. We hypothesized that decreased capacity for endothelial regeneration paralleled by low EPC numbers could increase the risk of relapse in patients with AAV. Methods. The number of circulating EPCs was determined by a colony-forming assay in a cohort of 41 patients with AAV. The patients were stratified into three subgroups according to the initial EPC count (low, medium and high) and prospectively followed. The primary goal was to determine the association between baseline EPC level and the time to the first relapse of the disease. Results. A total of 19 (46%) patients relapsed during the study period. The cumulative relapse-free survival increased stepwise across three increasing baseline levels of EPCs (P ¼ 0.013). EPC levels were not predictive of progression of renal disease, number of organs involved or death from any cause. Conclusion. Low numbers of EPCs are associated with increased propensity for early relapse of AAV. KEY WORDS: Vasculitis, Anti-neutrophil cytoplasmic antibody, Relapse, Outcome, Endothelial progenitor cell, Endothelium, Pauci-immune glomerulonephritis, Wegener’s granulomatosis.
Introduction
Patients and methods
ANCA-associated vasculitis (AAV) is a group of inflammatory disorders, characterized by inflammation and necrosis of blood vessels and relapsing remitting progression. Relapses can result in end-organ damage and are associated with considerable morbidity and mortality. Even though a substantial number of patients may never undergo a disease relapse, most patients with AAV are treated by a prolonged maintenance immunosuppressive therapy to prevent reactivation of AAV. Therefore, the identification of subgroups of patients with AAV at increased risk of relapse and the subsequent adjustment of (potentially harmful) immunosuppressive treatment (IST) are of major importance. Since resident endothelial cells infrequently proliferate [1], it has been postulated that there are other sources of vascular replenishment in response to continuous damage [2]. Recent studies have identified a population of presumably bone marrow-derived cells called endothelial progenitor cells (EPCs) that circulate in peripheral blood, express a variety of endothelial surface markers, incorporate into sites of neovascularization and home to sites of endothelial denudation [3–11]. EPC counts proved to be a useful biomarker of vascular function and injury [12, 13], and to be predictive of clinical outcome in different types of disease [14–19]. Decreased pool or impaired mobilization of EPCs could hamper the mechanism of endothelial repair and may contribute to repeated relapses in AAV. We hypothesized that low EPC numbers could predict the propensity to early relapse in patients with AAV.
Characterization of patients and controls A cohort of 41 patients with a diagnosis of AAV according to the international consensus definitions [20] were recruited from the Department of Nephrology, Charles University Hospital, Prague in 2004–05. The EPCs were measured in these patients by a colony-forming unit assay cross-sectionally at that time. The cohort was then prospectively followed until September 2008, when the current analysis was performed. At least once in their history, all patients were ANCA positive and 95% of them had histologically proven pauci-immune glomerulonephritis. WG was diagnosed according to the criteria of the ACR [21] and the Chapel Hill Consensus Conference definition [20], microscopic polyangiitis (MPA) was diagnosed according to the Chapel Hill Consensus Conference definition. Baseline characteristics of the patients are shown in Table 1. Patients were sampled during the diagnostic work up or follow-up visits. Nineteen patients were sampled at diagnosis and before initiating IST, 12 after 4–6 weeks of induction treatment and 10 after reaching remission. To check for possible variability of EPCs due to the disease activity and/or effect of IST, 10 AAV patients were sampled twice during the course of their disease, either with active disease both before and after institution of IST, or before and after reaching the remission of the disease. All patients with active disease were treated by an induction regimen consisting of pulse cyclophosphamide (10–15 mg/kg according to the level of renal function) and oral corticosteroids. Patients in remission of the disease were mostly treated by AZA, mycophenolate mophetil and/or low-dose prednisone. The whole study was approved by the ethical committee of the General Teaching Hospital, Prague. The investigation conformed to the principles outlined in the Declaration of Helsinki and a written consent was obtained from each subject.
1
Department of Nephrology, Charles University in Prague, First Faculty of Medicine and General Teaching Hospital, 2Institute of Rheumatology and 3First Department of Internal Medicine and Haematooncology, Charles University in Prague, First Faculty of Medicine and General Teaching Hospital, Prague, Czech Republic.
Definition of relapses, follow-up Disease activity was scored with the use of the validated Birmingham Vasculitis Scoring Index (BVAS). Remission was indicated by a complete absence of clinical activity using the
Submitted 12 January 2009; revised version accepted 21 April 2009. Correspondence to: Jakub Za´vada, Institute of Rheumatology, Na Slupi 4, 128 50, Praha 2, Czech Republic. E-mail:
[email protected]
1197 ß The Author 2009. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email:
[email protected]
Jakub Za´vada et al.
1198 TABLE 1. Baseline patient characteristics Characteristics
Group 1 (low EPC level)
Group 2 (medium EPC level)
Group 3 (high EPC level)
Number of patients Age, yearsa Gender, n (%) Male Female Type of disease, n (%) WG MPA Type of ANCA, n (%) Anti-PR3 Anti-MPO Organ involvement, n (%) Kidney Lung Ear–nose–throat area Serum creatinine, mmol/la Patients on dialysis, n (%)a EPC measurement, n (%) In active disease In remission Medicationa, n (%) ACEi/ARB Statins Erythropoietin
14 63 (50–67)
14 52 (42–57)
13 60 (57–65)
8 (57) 6 (43)
7 (50) 7 (50)
5 (38) 8 (62)
8 (57) 6 (43)
10 (71) 4 (29)
4 (31) 9 (69)
7 (50) 7 (50)
11 (79) 3 (21)
3 (23) 10 (77)
P-value 0.1 0.63 0.11 0.02
14 10 5 373 3
(100) (71) (36) (166–500) (21)
14 10 9 251 3
(100) (71) (64) (117–500) (21)
12 9 4 290 3
(92) (69) (31) (119–500) (23)
10 (71) 4 (29)
10 (71) 4 (29)
11 (85) 2 (15)
5 (36) 5 (36) 4 (29)
6 (43) 5 (36) 3 (21)
6 (46) 4 (31) 1 (8)
0.33 0.99 0.16 0.45 0.56 0.66
0.85 0.95 0.39
Categorical data are presented as number (percentage of total within groups), and continuous data are expressed as medians (interquartile range). P-values < 0.05 are marked in bold. aAt the time of the EPC measurement.
BVAS item list. Relapse could only occur in patients who reached remission lasting at least 1 month. Relapse was defined as vasculitic signs and symptoms in any organ system with BVAS > 1. Major relapse required recurrence or new appearance of one of the following major organ involvements attributable to active vasculitis: an increase in serum creatinine of >30% within a period of 3 months or clinical, radiological or bronchoscopic evidence of pulmonary haemorrhage or granulomata. Minor relapse was defined as recurrence of disease activity of lesser severity, such as epistaxis, crusting, myalgia, arthralgia, arthritis, mouth ulcers, rash, episcleritis, scleritis and pulmonary symptoms without or with minor radiological changes.
Quantification of circulating endothelial cells Circulating EPCs were quantified by the colony-forming unit assay as described by Hill et al. [13]. Briefly, blood was taken by venepuncture, and after discarding the first portion, 20 ml of blood anti-coagulated with heparin was centrifuged on Ficoll-Hypaque (Amersham, Upsala, Sweden) to obtain the mononuclear fraction. Peripheral blood mononuclear cells were washed in phosphate-buffered saline and counted on haemocytometer. In 4 ml of the Endocult medium, 107 recovered cells were re-suspended (StemCell Technologies, Vancouver, Canada) and divided into two equal portions. Each of the portions, containing 5 106 cells in 2 ml of Endocult was seeded in one well of the BD BioCoat Gelatin six-well plate (BD Biosciences, Franklin Lakes, NJ, USA) and cultivated for 48 h at 378C in an atmosphere with 5% CO2. Non-adherent cells were then removed together with the rest of the medium and counted, and 2 106 cells from each portion were seeded in two wells of the BD BioCoat fibronectin 24-well plate (i.e. from each sample four wells containing 106 non-adherent mononuclear cells were seeded). After another 72 h, endothelial colony-forming units (EPC-CFU) were scored under inverse microscope. Only colonies with at least 20 cells, containing rounded cells in the middle and elongated cells at the periphery, were considered as endothelial colonies. The phenotype of EPC colonies was determined in five subjects. EPC colonies were grown on fibronectin coated glass chamber slides. CD31 (clone HC1/6, Chemicon, Temecula, CA, USA), vascular endothelial growth factor receptor (VEGFR-1) (clone FLT-19, Sigma Saint Louis, MO, USA), VEGFR-2 (clone 89106, R&D Systems, Minneapolis, MN, USA) and von
Wildebrand factor (clone 2F2-A9, BD Biosciences) positivity were determined by immunofluorescent fluorescein isothiocyanate (FITC)-labelled antibodies. Similarly, Ulex Europeus Agglutinin (UEA)-1 lectin binding was evaluated with FITC-labelled UEA-1 lectin. Cell nuclei were visualized with 40 -6-diamidino-2phenylindole dihydrochloride (DAPI) (Sigma-Aldrich, Munich, Germany) staining, which enabled to discriminate endothelial colonies from scattered adherent cells. As this procedure is quite laborious, after confirming that EPC colonies are positive for the above mentioned markers, we did not repeat these studies in other subjects.
Other biochemical and haematological tests and ANCA detection Biochemical and haematological tests were performed in Central Laboratories of General Teaching Hospital, Prague, according to the institutional guidelines. ANCA antibodies were determined by ELISA (Orgentec, Mainz, Germany) and by IIF microscopy.
Statistical analysis The association between baseline levels of EPCs (measured in 2004–05) and the occurrence of relapse was evaluated in September 2008. The levels of EPCs were analysed as categorical variables—patients were divided into three pre-specified, numerically equal groups corresponding to patients’ EPC counts (low, medium and high) at the time of enrolment. Survival (time to the first relapse) was determined with the use of the Kaplan–Meier method and the Cox regression analysis. The log-rank test was used to determine statistical differences in terms of survival. Patients were censored if they had not relapsed at their last follow-up visit or if they died of other cause than AAV activity. Because of a relatively small number of patients we were not able to perform a robust multivariate Cox proportional analysis, but we performed univariate analyses and a multivariate analysis with three putative predictors of relapse included: type of ANCA (anti-MPO vs anti-PR3), type of disease (WG vs MPA) and EPC group (low, medium and high EPC counts). The hazard ratio for the EPC group represents the predicted change in the hazard for a unit increase in the predictor (e.g. an increase from low to medium or from medium to high in the number of EPCs). Data were tested for normality by Kolmogorov–Smirnov and
EPC number predicts relapses in AAV
Results Cohort description Thirty-one patients with active disease and 10 patients in remission were enrolled to initial EPC sampling. Ten patients were sampled twice during the course of their disease, either with active disease both before and after institution of IST, or before and after reaching remission of the disease. We have neither found any significant difference nor a trend in EPC counts between these paired measurements. Similarly, we did not find a significant difference in EPC numbers between the three subgroups of AAV patients (active before IST, active after IST, in remission—more detailed description of the cohort was presented elsewhere [22]). The average time from EPC measurement to censorship or relapse was 21 months. Cohort characteristics with respect to baseline EPC counts are shown in Table 1. There were no significant differences in the baseline characteristics between the EPC groups, except for the type of ANCA antibodies, as the anti-MPO-positive patients were underrepresented in the medium EPC level group and overrepresented in the high EPC level group. The number of EPC colonies (EPC-CFU) per millilitre of blood ranged from 0 to 39 with a mean of 4.2 CFU/ml. Group 1 represents patients with virtually no growth of EPC colonies (EPC-CFU 3 EPCCFU/ml).
Impact of baseline EPC counts on relapse A total of 19 (46%) patients relapsed during the study period: 11 in the low EPC group, 5 in the medium group and 3 in the high EPC group, respectively (P ¼ 0.01). This trend was most striking with respect to renal relapse, when seven (50%) patients in the low EPC group but none in the medium and high EPC groups suffered from renal relapse (P < 0.001). There was no significant difference in the relapse rate in other organs between the EPC groups. Cumulative relapse-free survival increased in a stepwise fashion across increasing levels of EPC-CFUs (Fig. 1). In univariate analyses, both the EPC group (P ¼ 0.008) and type of disease (WG vs MPA, P ¼ 0.02) were found to be associated with a propensity to earlier relapse. There also was a trend towards early relapse in patients with anti-PR3 antibodies as compared with anti-MPO antibodies (P ¼ 0.09). When these three suggested risk factors were combined in the risk analysis (type of ANCA, type of disease and EPC group), just the EPC group was independently associated with the time to relapse [hazard ratio for relapse of 0.42 (CI 0.21, 0.85) with an increase from low to medium or medium to high EPC group (P ¼ 0.02)]. The frequency of outcomes and results of subsequent analyses are shown in Tables 2 and 3. Similar results were obtained when analysing solely the 31 patients sampled in active disease: nine (90%), four (40%) and three (27%) of these patients relapsed (P ¼ 0.01), and seven (70%), zero (0%) and zero (0%) patients suffered from renal relapse (P < 0.001) in the low, medium and high EPC groups, respectively. The survival analysis for time to any relapse in this smaller subgroup was not significant (P ¼ 0.1), but was significant for time to renal relapse (P < 0.001).
EPC group Low Medium High Low-censored Mediumcensored High-censored
1.0
0.8
Relapse-free survival
Lilliefors tests. Because several of the variables, including EPC numbers, did not follow the normal distribution, non-parametric tests were used for comparisons between subgroups. Statistical significance was assumed when a null hypothesis could be rejected at P < 0.05 (all P-values are two-sided). Statistical analysis was performed with the use of SPSS software, version 11.5, for Windows (Chicago, IL, USA).
1199
0.6
0.4
0.2
P = 0.013 0.0 0.00
10.00
20.00
30.00
40.00
Time (months after EPC sampling)
FIG. 1. Cumulative relapse-free survival of AASV patients in an analysis of relapse of AASV according to baseline EPC count.
Impact of baseline EPC counts on death from any cause and renal outcome There was no significant difference in terms of incidence of death of any cause or measures of renal outcome (such as the mean time averaged change in serum creatinine or dependence on dialysis treatment) with respect to baseline EPC levels (Table 2).
Discussion The present study extends our previous findings [22], and demonstrates that in patients with ANCA-associated vasculitis the reduced number of circulating EPCs is associated with increased risk of early relapse of the disease. The results of our study confirm the previous reports suggesting that circulating EPCs can be used as a predictive biomarker for cardiovascular risk and vascular injury. The levels of circulating EPCs in the peripheral blood correlate with measures of endothelial function [12, 13], and trials aimed to increase the number of EPCs at the site of tissue damage have shown therapeutically meaningful results [23–25]. EPC counts have also been shown to predict clinical outcomes in cardiovascular disease and stroke, as well as in acute lung injury or sepsis [14–19]. In AAV, both low [22, 26] and normal [27] EPC level were reported, as well as inverse [26, 27] or no correlation [22] with markers of activity of the disease. Small number of investigated patients (n ¼ 11 in [27]), differences in the techniques of EPC measurement (CFU assay [22, 26] vs late-outgrowth [27] cultivation method) as well as in the characteristics of studied patient cohorts (in [26] only anti-PR3-positive patients with WG were investigated, active and non-active patients seemed to have different level of renal dysfunction) could have been involved in these discrepancies. In the initial analysis of our cohort [22], we found neither significant changes between paired measurements, nor significant associations between the number of EPC colonies and the activity of the disease scored with the use of the validated BVAS, the number of involved organs or the level of CRP and ANCA antibodies in these patients. The interplay of factors, such as the level of disease activity, CRP, renal dysfunction, types and dose of immunosuppressive drugs and other medications (such as statins, ACE inhibitors or erythropoietin) could have an impact on the level of EPCs [28–32]. Despite the potential of these confounding influences, very low yield of EPC-CFUs seems to signal impaired intrinsic potential to microvascular repair and increased propensity to relapse of the disease.
Jakub Za´vada et al.
1200
TABLE 2. Outcomes of the AAV patients with respect to their baseline EPC levels Characteristics
Group 1 (low EPC level)
Death from any cause Relapse Major relapse Minor relapse Organ(s) involved in relapse Kidney Lung Ear–nose–throat area Others Serum creatinine, mmol/la Overall change in serum creatinine Change in serum creatinine per month of follow-up Patients on dialysis, n (%)a
2 11 8 3 7 4 1 1 206 14 1.1 6
(14) (79) (57) (21)
Group 2 (medium EPC level) 4 5 4 1
(50) (28) (7) (7) (133–500) (160–0) (3.5–0) (43)
0 4 1 0 235 4 0.2 5
(29) (36) (29) (7)
Group 3 (high EPC level) 3 3 2 1
(0) (28) (7) (0) (96–500) (18–22) (0.4–1.6) (36)
0 2 0 1 185 46 2.9 3
(23) (23) (15) (8) (0) (15) (0) (7) (109–371) (132–0) (6.6–0) (23)
P-value 0.65 0.01 0.06 0.43 _0.001 0.65 0.61 0.58 0.56 0.16 0.70 0.55
Categorical data are presented as number (percentage of total within groups) and continuous data are expressed as medians (interquartile range). P-values < 0.05 are marked in bold. aAt the time of the last follow-up visit.
TABLE 3. Cox proportional analysis of putative risk factors for relapse of AAV
Type of disease (WG vs MPA) Type of ANCA (anti-PR3 vs anti-MPO) EPC group (high, medium and low)
Unadjusted hazard ratio (95% CI)
P-value
Adjusteda hazard ratio (95% CI)
P-value
3.10 (1.17, 8.19) 2.27 (0.89, 5.79) 0.43 (0.23, 0.80)
0.023 0.087 0.008
1.73 (0.50, 5.93) 1.35 (0.41, 4.46) 0.42 (0.21, 0.85)
0.39 0.62 0.02
P-values
