Endothelial progenitor cells in chronic obstructive pulmonary disease and emphysema

Endothelial injury is implicated in the pathogenesis of COPD and emphysema; however the role of endothelial progenitor cells (EPCs), a marker of endothelial cell repair, and circulating endothelial cells (CECs), a marker of endothelial cell injury, in COPD and its subphenotypes is unresolved. We hypothesized that endothelial progenitor cell populations would be decreased in COPD and emphysema and that circulating endothelial cells would be increased. Associations with other subphenotypes were examined. The Multi-Ethnic Study of Atherosclerosis COPD Study recruited smokers with COPD and controls age 50–79 years without clinical cardiovascular disease. Endothelial progenitor cell populations (CD34+KDR+ and CD34+KDR+CD133+ cells) and circulating endothelial cells (CD45dimCD31+CD146+CD133-) were measured by flow cytometry. COPD was defined by standard spirometric criteria. Emphysema was assessed qualitatively and quantitatively on CT. Full pulmonary function testing and expiratory CTs were measured in a subset. Among 257 participants, both endothelial progenitor cell populations, and particularly CD34+KDR+ endothelial progenitor cells, were reduced in COPD. The CD34+KDR+CD133+ endothelial progenitor cells were associated inversely with emphysema extent. Both endothelial progenitor cell populations were associated inversely with extent of panlobular emphysema and positively with diffusing capacity. Circulating endothelial cells were not significantly altered in COPD but were inversely associated with pulmonary microvascular blood flow on MRI. There was no consistent association of endothelial progenitor cells or circulating endothelial cells with measures of gas trapping. These data provide evidence that endothelial repair is impaired in COPD and suggest that this pathological process is specific to emphysema.


Study sample
The Multi-Ethnic Study of Atherosclerosis (MESA) COPD Study is a multicenter case-control study nested within two prospective cohort studies, MESA [40] and the Emphysema and Cancer Action Project (EMCAP) [4], in addition to a small number from the community. We selected all eligible participants in the MESA Lung Study [41] and oversampled participants with COPD or emphysema from the remainder of MESA and EMCAP, collected from May 2009 through July 2011. Participants were age 50-79 years with greater than 10 pack-year smoking histories and free of clinical cardiovascular disease, stage IIIb-V kidney disease, asthma prior to age 45, other lung disease, prior lung resection and cancer.

Ethics statement
Protocols were approved by the institutional review boards of the participating institutions (Columbia University Medical Center, University of Vermont, Johns Hopkins University, Northwestern Medical Center, University of California at Los Angeles, University of Washington, University of Iowa and Hannover University) and the National Heart, Lung and Blood Institute (NHLBI). Written informed consent was obtained from all participants.

Flow cytometry methods: EPCs and CECs
Freshly drawn whole blood was drawn into heparinized tubes and shipped overnight in temperature controlled packages, as detailed in S1 Supporting Information. Peripheral blood mononuclear cells were prepared and labeled following standardized protocols (see online supplement). We defined EPCs as cells expressing CD34, a stem cell marker, and KDR (also known as VEGFR2), an endothelial cell marker; we further defined less differentiated EPCs (a subset of EPCs) by the additional presence of CD133, a bone marrow derived stem cell marker [42,43]. For CECs, we chose CD45dim to rule out leukocytes, CD146 and CD31 as endothelial cell markers and CD133 to rule out progenitor cells. Samples were fixed in 1% paraformaldehyde and kept refrigerated in the dark until analyzed by flow cytometry (BD LSR II). Endothelial populations were expressed as percent of PBMCs as gated using forward and side scatter and a gating strategy is shown in the online supplement (S3 Fig, S4 Fig). Biovariability and delayed processing impacts were analyzed (S1 Supporting Information, S1 Fig, S2 Fig) and are acceptable for epidemiology studies.

Spirometry
Spirometry was conducted on a dry-rolling-sealed spirometer (Occupational Marketing, Inc., Houston, TX) in accordance with American Thoracic Society/European Respiratory Society (ATS/ERS) guidelines [44] following the MESA Lung protocol [45].
The extents of emphysema and emphysema subtypes were assessed qualitatively by experienced chest radiologists following a highly standardized protocol without access to other study information, as previously described [47]. The inter-rater intraclass correlation coefficient of extent of emphysema and emphysema subtypes were 0.77 and 0.42-0.93, respectively [47].
The percent of emphysema-like lung (percent emphysema) was defined as the percent of lung voxels below -950 HU at total lung capacity and gas trapping on CT was defined as voxels below -856 HU on expiratory scans in a subset, both assessed using Apollo 1.2 software (Vida Diagnostics) [46].

Magnetic resonance imaging
Pulmonary microvascular blood flow was assessed on MRI at one site using a modified version of the cardiac MRI protocol of the fifth examination of MESA on a 1.5 Tesla whole-body MR system (Signa LX, GE Healthcare) with phased-array coil for signal reception using a coronal Time Resolved Imaging of Contrast Kinetics (TRICKS) sequence and a contrast bolus of 0.1 mmol/kg bodyweight gadolinium diethylenetriaminepentaacetic acid (Magnevist, Berlex, Wayne, NJ) at an injection rate of 5 mL/s. Mean values of pulmonary microvascular blood flow and volume were assessed on a coronal slice at the level of the trachea in the peripheral 2 cm of the lung, as previously described [48].

DLco and plethysmography
DLco and plethysmography were performed in participants at the same site. Single-breath DLco was measured with a Sensormedics Autobox 220 Series instrument (Viasys Healthcare, Yorba Linda, CA) following ATS/ERS guidelines [49]. Body plethysmography was performed using a V6200 Series Autobox (Sensormedics, Yorba Linda, CA) following ATS/ERS recommendations [50].

Covariates
Age, gender, race/ethnicity, educational attainment, smoking status, pack-years, and medical history were self-reported. Medication use was assessed by medication inventory.
[51] Height, weight, blood pressure, oxygen saturation, glucose, total cholesterol, high-density lipoprotein (HDL) levels and complete blood counts were measured using standardized approaches. Smoking status was confirmed by cotinine.

Statistical analysis
EPCs and CECs displayed skewed distributions and were therefore log-transformed. The association between each of the cellular measures and COPD status and severity was tested in linear regression models in which categories of COPD status and severity were treated as independent variables and cellular populations were treated as the dependent variables. A test of trend across categories of COPD severity was performed, in addition to pairwise comparisons.
Linear regression models were then used to adjust for potential confounders, which were selected based on biologic plausibility and correlations with covariates. The base model was adjusted for age, gender, race/ethnicity and cohort of selection. We then additionally adjusted for smoking status and pack-years, then additional potential confounders of educational attainment, diabetes, hypertension, oxygen saturation, sleep apnea, height, BMI, as they are known confounders for COPD, and statin use, HDL, and white blood cell (WBC) count, as statins have been implicated in endothelial progenitor cell mobilization [52-54], HDL may affect endothelial health and is related to percent emphysema [55], and WBCs to account for unrelated inflammatory white cell interference. Models for the qualitatively measured percent emphysema were additionally adjusted for mA. Models for pulmonary perfusion were additionally adjusted for cardiac output.
The associations of cellular measures with emphysema extent and other continuous outcomes were assessed using linear regression. Because the study recruited based upon COPD status, the assumption underlying linear regression of independent observations may not hold. Analyses of continuous outcomes were therefore weighted according to cohort-specific probabilities of selection and enrollment, with cases recruited from the community assigned the same weights as those from EMCAP.
The p-values were two-tailed with statistical significance defined as p<0.05. Analyses were performed in SAS 9.2 (Cary, NC) and R version 2.14.1 (Vienna, Austria).

Results
The flow diagram for the participants with spirometry, CT and endothelial cell populations is shown in Fig 1. The 257 participants had a mean age of 68.3 +/-6.9, 28.0% were current smokers, and the race/ethnic distribution was 51.3% white, 36.9% African-American, 14.8% Hispanic and 7.0% Chinese. Table 1 shows the clinical characteristics of the study sample stratified by COPD severity. Participants with severe COPD were more likely to be male, white and have greater packyears of smoking.

EPCs, CECs and COPD severity
The CD34+KDR+ EPC population showed a significant decrease with increasing COPD severity in minimally adjusted (p = 0.03 for trend), smoking adjusted (p = 0.01 for trend) and fully adjusted (p = 0.02 for trend) models (Table 2). Compared to controls, CD34+KDR+ cells were significantly lower in moderate and severe COPD. For CD34+KDR+CD133+ cells, only severe cases were significantly different from controls. There was no evidence that CECs were increased across categories of COPD severity (Table 2).

EPCs, CECs and Emphysema
Whereas there was little-to-no evidence that CD34+KDR+ EPCs were reduced with greater extent of emphysema, CD34+KDR+CD133+ EPCs were significantly lower with total emphysema (Table 3). Evaluation by emphysema subtypes demonstrated that CD34+KDR+ and particularly CD34+ KDR+CD133+ EPCs were significantly decreased with increasing extent of panlobular emphysema (Table 3). CD34+KDR+CD133+ EPCs only were (weakly) associated with centrilobular emphysema and there was no evidence for associations with paraseptal emphysema. Fig 2 illustrates the association of CD34+KDR+ and CD34+KDR+CD133+ populations with major emphysema subtypes. CEC cells showed no association with any qualitative emphysema measure (Table 3). There was no evidence for an association of either EPC population or CECs with percent emphysema when assessed qualitatively (adjusted P = 0.24, P = 0.92 and P = 0.39, respectively).

Diffusing capacity and pulmonary microvascular perfusion
Lower levels of CD34+KDR+ and CD34+KDR+CD133+ EPCs were significantly associated with lower diffusion capacity and lower diffusing capacity corrected for alveolar volume in fully adjusted models (S1 Table). CECs were positively associated with diffusion capacity in the fully adjusted model but not in minimally adjusted models or when corrected for alveolar volume.
There were no statistically significant associations between EPCs and pulmonary microvascular blood flow on MRI in fully adjusted models (Online Supplement Table 1). Higher levels of CECs were significantly associated with reduced pulmonary microvascular blood volume in both the minimally and fully adjusted models.

Gas trapping
None of the endothelial cell populations were associated with increased residual volume (RV) or RV/total lung capacity on plethysmography. Gas trapping on CT was associated with a small but significant decrease in the CD34+KDR+ endothelial progenitor cells but otherwise associations were non-significant (S2 Table).

Discussion
EPC populations were reduced in COPD, with CD34+KDR+ EPCs in particular being lower in both moderate and severe COPD compared to controls. CD34+KDR+CD133+ EPCs were preferentially lower with greater emphysema on CT and both EPC populations were lower with greater panlobular emphysema and reduced diffusing capacity. CECs were inversely related to pulmonary microvascular blood flow. In contrast, there was little evidence to support a definitive role for endothelial cell populations and measures of gas trapping. These apparent discrepancies in gas trapping results with respect to cellular measures may be explained, in part, by the fact that not all measures of hyperinflation are created equal [56], where gas trapping was associated with smaller airway lumen diameter, greater dyspnea and chronic bronchitis, and hyper-expansion was associated with percent emphysema, lower BMI and higher hemoglobin levels.
To our knowledge, this is the largest study of endothelial cell populations in COPD and the only one to look at more than one EPC population and COPD subphenotypes. The two mechanisms for reduced circulating cell populations are increased utilization, decreased supply or a combination of both. Since the less mature progenitors (CD34+KDR+CD133+) were significantly reduced only in severe COPD, while the CD34+KDR+ cells were highly significant even after adjustment, these findings support the hypothesis that CD34+KDR+ cells were being preferentially drawn from the circulation during mild and moderate COPD (increased utilization) and it is not until severe COPD that the more bone marrow derived cells (CD34+KDR +CD133+) become depleted (decreased supply). Consistent with this thinking, a study by Peinado et al[57] examining lung tissue from persons with COPD and controls showed an increase in CD34+KDR+ cells in the lung tissue of COPD patients. Additionally, EPCs from COPD patients have been shown to be dysfunctional in proliferation and migration assays in response to specific chemotactic factors (i.e. SDF-1α) [58-61], which may be responsible for a change in circulating EPCs. Together, these data suggest that with increasing severity of COPD, CD34 + KDR + cells are drawn into damaged lung tissue, thus decreasing the number circulating in the peripheral blood. A recent study in mice showed a decrease in circulating CD34+KDR+ cells within 1 day of lps challenge and~5% of the proliferating pulmonary vascular endothelial cells were of BM-derived cells [62]. They also noted that the resident proliferating vascular endothelial cells were expressing CD34 and KDR on their surface during the reparative process.
Interestingly, a different situation may occur in emphysema, where the extent of emphysema was related to less mature progenitors (CD34+KDR+CD133+) and particularly in panlobular emphysema. These findings suggest that while COPD is recruiting from the circulating CD34+KDR+ EPC pool, emphysema repair relates to the less differentiated CD34+KDR+ CD133+ EPCs, the exact mechanism of which remains unclear, emphasizing the differences in this phenotype. The strong association of the EPCs with the panlobular emphysema, which is a more vascular-related emphysema [36,37], is also consistent with recent work showing effects of alpha1 antitrypsin on pulmonary endothelial health [63,64]. This notion is further supported by results for diffusing capacity, which provides an indirect vascular measure, although direct measures of microvascular structure and function on MRI were non-significant for EPCs and only revealed an association for CECs.
In contrast, there was little to no association between endothelial cell populations and measures of gas trapping, which suggests that the biology of endothelial damage and repair may be specific to emphysematous changes in COPD. This data, taken together, illustrates some of the differences in the pathology of COPD and supports further study of sub-phenotypes of COPD and emphysema to better direct future treatment options.
These results suggest that the definition of EPCs and selection of COPD patients and subphenotypes may explain some of the controversy in the literature regarding the role of EPC's in COPD. Many studies [6,27,35,61,[65][66][67] have used CD34+KDR+ or CD34+KDR+CD133 + cells to study COPD, coming to different conclusions based on the characterization of the EPCs, and few have examined COPD subphenotypes. CD133 has been shown to be a marker of early EPC differentiation that is lost as the progenitor differentiates [68]. Thus, studies that utilize CD133 as a marker are looking at less differentiated EPCs, while those that use just CD34+KDR+ are examining both early and late EPC. Our results suggest that the two cell types behave differently, dependent on the phenotype, suggesting increased uptake of EPCs in COPD but loss of bone-marrow derived cells in panlobular emphysema.
One surprising result was the lack of increase in CECs in COPD or emphysema. We hypothesized that damage to pulmonary vasculature would cause an increase in the "sloughing off" of endothelial cells thus increasing the CEC based, in part, on prior findings for EMPs in this study [7] and the reports of elevated CECs in cardiovascular disease [69]. It is possible these negative results may be due, at least in part, to apoptosis of the endothelial cells in the pulmonary vessels, either on the vessel surface, or shortly after the "sloughing off" process. We and others have previously found evidence for pulmonary vascular endothelial cell apoptosis based The MESA COPD study on elevated apoptotic EMPs in mild, moderate and severe COPD [7,9,10], consistent with animal [70,71] and human studies implicating apoptosis in emphysema and COPD [72][73][74][75]. Further, the study by Boos et al [76] showed that CEC levels correlate with damage markers but not with apoptosis. Additionally, a recent study demonstrating that smoking [77] alters the alpha-1 antitrypsin interaction with caspases causing an increase in endothelial apoptosis supports an apoptotic mechanism in emphysema. And lastly, CECs, as we have defined them, may not be a good indicator of what is happening at the local level.
It was also somewhat unusual that the associations for emphysema were specific to radiologist-defined emphysema and non-significant for quantitatively assessed percent emphysema. The latter null result may have been due to the inability of quantitatively assessed emphysema to define subtypes of emphysema.
The strengths of this study are the precisely defined cellular measures, the relative size of the study and the detailed EPC and COPD subphenotyping. Weaknesses include the delayed sample processing, which we attempted to minimize with strict adherence to timing protocols [78][79][80] and which is mitigated by the observed correlation between results from fresh and delayed processing among volunteers. Circulating cellular levels may not be reflective of localized events, yet they provide one of the best methods for evaluating cells in living persons. The choice of cellular markers is, to some degree, subjective but was based on best information available at the inception of this project. Finally selection bias can affect case-control studies but this was minimized by using a nested design with known sampling probabilities within MESA and EMCAP.
In conclusion, CD34+KDR+ cells were significantly decreased with increasing COPD severity. The less differentiated EPC phenotype, CD34+KDR+CD133+ cells, was also lower in severe COPD. These results support the hypothesis that the CD34+KDR+ EPC are being preferentially recruited into the lung for vascular repair and the exhaustion of bone marrow derived cells does not appear to occur until spirometry defined COPD has reached severe levels. In contrast, CD34+KDR+ and CD34+KDR+CD133+ cells were reduced in panlobular emphysema and with reduced diffusing capacity, possibly suggesting a loss of bone-marrowderived EPCs in that form of the disease. This would suggest that the lung vascular injury/ repair mechanism for emphysema subtypes varies greatly and requires further evaluation.