Enhanced late-outgrowth circulating endothelial progenitor cell levels in rheumatoid arthritis and correlation with disease activity

Rheumatoid arthritis (RA) is a chronic and destructive inflammatory disease affecting the joints. RA is now well known to be associated with striking neovascularization developed in inflammatory joints [1]. Indeed, angiogenesis, leading to an increased number of synovial vessels through local endothelial cells, is a cornerstone of synovial hyperplasia occurring in RA. Disturbances in endothelial cell turnover and apoptosis as well as in angiogenic factors such as vascular endothelial growth factor (VEGF) have been reported in RA synovium [2, 3]. However, despite the abundant synovial vasculature, there are areas of synovial hypoxia contributing to synovial and cartilage damage [4, 5]. Hypoxia is highly suggested to activate the angiogenic cascade, thereby contributing to the perpetuation of RA synovitis [6].

In addition to angiogenesis issued from resident cells, cells derived from bone marrow and named circulating endothelial progenitor cells (EPCs) are able to promote new blood vessel formation (vasculogenesis) and may therefore contribute to RA synovitis [7].

EPCs were originally identified by (a) the expression of markers shared with hematopoietic stem cells such as CD34 and CD133, (b) specific endothelial cell markers such as KDR (vascular endothelial growth factor receptor-2 [VEGFR-2] or kinase-insert domain receptor), and (c) their capacity to differentiate into functional endothelial cells [8‐10]. However, there is no consensus on the precise definition of EPCs [11]. Evidence showed that there is more than one endothelial progeny, monocytic versus hemangioblastic, within the circulating blood, and two distinct cell types of EPCs are currently recognized according to their growth characteristics and morphological appearance: early-outgrowth EPCs and late-outgrowth EPCs [7, 12].

In the currently available human studies, variations in the level of circulating EPCs were reported in different diseases affecting the vascular system and were suggested to be a biomarker for vascular function and tumor progression [13, 14]. In the field of RA, contradictory results have been reported. Indeed, some studies suggested a lower circulating EPC number in RA patients compared with controls [9, 10], but conversely, some others reported higher values [15], and finally some other reports did not find any difference [8, 16]. Several studies have shown an increase of EPCs within the RA synovial tissue [17, 18].

Altogether, these observations underline the difficulty of accurately quantifying EPC populations. The major issue is the identification of the different types of circulating endothelial cells (CECs) issued respectively from the vessel wall or from bone marrow progenitors. The use of accurate methods allowing the detection of rare events by flow cytometry is thus critical. Within this context, our group contributed to recommendations aiming at the improvement of EPC detection and characterization [19]. In line with these latter recommendations and our background in systemic sclerosis [20], we focused on late-outgrowth EPCs that represent the progenitors with the more genuine endothelial properties. In parallel to EPCs, CECs detached from vessel walls (CECs) may also be a relevant biomarker of vascular disease. We hypothesized that coupled raised levels of these two populations may reflect the vascular status of the disease and thus represent innovative biomarkers. Therefore, our aims were (a) to enumerate EPCs and late-outgrowth endothelial colony formation in RA patients and controls, (b) to assess correlations between EPC counts, CEC counts, and RA activity, and (c) to correlate EPC and CEC counts with levels of serum markers of synovitis or endothelial injury.

Our results obtained by using a well-characterized definition of late-outgrowth EPCs in a relatively large number of patients show enhanced levels of this cell population and relationships with RA disease activity. Available data have reported conflicting results about the EPC counts in this inflammatory condition. Several methodological issues could explain such discrepancies. We herein followed recent recommendations and used a previously validated method for late-outgrowth EPC enumeration [20].

The quantification of EPCs by flow cytometry first requires enrichment techniques to select a correct number of this scarce population and a specific marker combination to select the subpopulation of hemangioblastic EPCs. In culture, these 'true' angioblast-like EPCs are represented by cells that enable late outgrowth with higher proliferative potential, while endothelial cell colonies that appear early might more preferentially originate from monocytes or CECs [11, 24]. In the study herein, we excluded the monocytic EPC subpopulation from the quantification and thus selected hemangioblastic EPCs by the lineage-positive cell depletion including CD14⁺ cells. We also extended the circulating EPC definition with an additional marker of viability to select non-apoptotic cells. One may suggest that the several steps required by our technique may induce procedural loss of progenitors which are prone to undergo apoptosis. However, the controlled design of our study limits this potential bias but this will need additional work. In parallel, EPC counts were also determined by the selection in culture of the late-outgrowth EPCs according to the delays before their appearance.

None of the previous studies that reported EPC levels in RA patients was based on these methods. The previous data quantified, conversely to our study, circulating EPCs in whole blood with only three surface markers (CD34/CD133/VEGFR-2) [8‐10, 16]. These authors also assessed EPC-CFU numbers, focusing on the 'early outgrowth' subpopulation and finding or not finding results consistent with those of flow cytometry quantification. These methodological differences may account for the discrepancy with regard to our results. Indeed, differences between the various EPC studies may not relate to the characteristics of the RA population enrolled. Our population of RA patients, issued from consecutive inclusions, did not display differences with other studies based on criteria known to modulate EPC levels -- such as age (mean of 53 to 59 years) and frequency of use of methotrexate or low doses of corticosteroids -- or on the choice to exclude patients with previous cardiovascular events [25‐27]. In addition, it is noteworthy that EPC counts did not differ according to the use of biologics, although the cross-sectional design limits the analysis of the influence of such therapies in our RA patients. The only specificity of our RA patients may be the relatively long disease duration in comparison with other works. Nevertheless, disease duration was never reported to be associated with EPC levels and thus may not account for our findings. We excluded RA patients with cardiovascular risk factors in order to rule out the bias of the specific effects of atheroma on EPC counts and thus to focus on relationships between disease activity and EPC counts as this has been done in many previous studies [9, 10]. One may suggest that this may have introduced a selection bias and the use of this exclusion criterion may have obscured a negative influence of cardiovascular risk factors on EPC counts.

We herein provide the demonstration of the identification of the late-outgrowth subset by the association between circulating cell counts and culture isolations. Indeed, two different subpopulations of EPCs, namely early and late EPCs, can be derived from peripheral blood depending on the different culture methods and times [24, 28]. Although both EPCs express endothelial markers, they have different morphologies, patterns of growth, and angiogenic properties and thus might have different roles in neovasculogenesis [24, 29‐31]. Late-outgrowth EPCs that represent the progenitors with the more genuine endothelial properties such as tube-forming activity in vitro and in vivo need to be better characterized in the context of inflammatory conditions. Previously, the late EPC subpopulation has been studied in systemic sclerosis by our group and their endothelial properties confirmed by angiogenic tests [32]. As reported in systemic sclerosis, we observed that the RA patients, having high Lin^-7AAD^-CD34⁺CD133⁺VEGFR-2⁺, displayed a higher number of EPC-CFUs. However, the size of the sample reduced by the non-systematic achievement of EPC-CFUs could explain the limited increase of EPC-CFU numbers in RA patients as well as the lack of association with disease activity and will need larger studies.

We thus assume that our EPC definition allows a well-characterized quantification of late-outgrowth EPCs. In RA, our data support the contribution of late-outgrowth EPCs to synovitis according to our finding of a strong link with disease activity. The direct correlation of EPC counts with DAS-28 levels is strengthened by the fact that RA patients in remission displayed EPC levels comparable to those of controls. Together with evidence of CD133/VEGFR-2⁺ cells in RA synovial tissue [17], our results emphasize a key role for vasculogenesis and EPC mobilization in RA.

Preliminary data on CEC outcome in vascular diseases have suggested a relationship between the detachment of mature CECs and vascular hurting [33]. We concomitantly determined the value of CECs as compared with EPCs. We found a correlation in RA between these two circulating cell levels but CEC levels in RA did not differ for the ones in controls. Using a CD146 immunoselection in whole blood, one study found enhanced CEC levels but failed to identify a specific association with blood inflammatory markers [34]. The best combination of surface markers including exclusion of dead cells by viability marker seems to be required for CEC quantification.

One of the pitfalls of EPC quantification in RA is the potential involvement of atherosclerosis in EPC changes [35, 36]. However, we excluded from our study individuals with classical cardiovascular risk factors and also those with previous clinical events. Furthermore, we did not find an increase of CEC levels in our RA population which reflects endothelial injury. Indeed, we presume that EPC increase relates to RA disease activity and synovial inflammation, although measurement of infra-clinical atheroma would be necessary to definitely rule out endothelial dysfunction.

We then attempted to correlate EPC counts with blood markers reflecting inflammation (ESR), endothelium injury (sVCAM), or synovial involvement (COMP and YKL-40). While we failed in the identification of any link, sVCAM, COMP, and YKL-40 were found to be increased in RA patients, thus confirming the activity of the disease in our sample of RA patients. We assume that, despite the lack of a link with serum markers, the identification of a relationship between EPCs and DAS-28 is highly relevant. EPC count is probably influenced by several factors, including inflammation, vascular injury, and potentially the immune response with bone marrow changes. The multifactorial regulation probably precludes the identification of a correlation with one single serum marker. Therefore, EPCs could represent a new 'integrative' biomarker. Its predictive value on disease outcome, including both articular and cardiovascular issues, will have to be evaluated in upcoming prospective studies.

We demonstrate enhanced levels of late-outgrowth EPCs in RA and a relationship with disease activity. This supports the implication of vasculogenesis in the perpetuation of the synovitis that occurs in RA. Late-outgrowth EPC isolation also offers a unique opportunity to determine an RA endothelial signature.