Endothelial progenitor cells and plaque burden in stented coronary artery segments: an optical coherence tomography study six months after elective PCI

Background Endothelial progenitor cells (EPC) are involved in neovascularization and endothelial integrity. They might be protective in atherosclerosis. Optical coherence tomography (OCT) is a precise intracoronary imaging modality that allows assessment of subintimal plaque development. We evaluated the influence of EPC on coronary plaque burden in stable disease and implemented a novel computational plaque analysis algorithm using OCT. Methods Forty-three patients (69.8% males, 69.6 ± 7.7 years) were investigated by OCT during re-angiography 6 months after elective stent implantation. Different subpopulations of EPCs were identified by flow cytometry according to their co-expression of antigens (CD34+, CD133+, kinase domain receptor, KDR+). An algorithm was applied to calculate the underlying total plaque burden of the stented segments from OCT images. Plaque morphology was assessed according to international consensus in OCT imaging. Results A cumulative sub-strut plaque volume of 10.87 ± 12.7 mm3 and a sub-stent plaque area of 16.23 ± 17.0 mm2 were found within the stented vessel segments with no significant differences between different stent types. All EPC subpopulations (mean of EPC levels: CD34+/CD133+: 2.66 ± 2.0%; CD34+/KDR+: 7.50 ± 5.0%; CD34+/CD133+/KDR+: 1.12 ± 1.0%) inversely correlated with the identified underlying total plaque volume and plaque area (p ≤ 0.012). Conclusions This novel analysis algorithm allows for the first time comprehensive quantification of coronary plaque burden by OCT and illustration as spread out vessel charts. Increased EPC levels are associated with less sub-stent coronary plaque burden which adds to previous findings of their protective role in atherosclerosis.


Background
Atherosclerosis is a systemic disease. Our understanding of its underlying pathomechanism has been extended beyond typically contributing cardiovascular risk factors like hyperlipidemia, hypertension, diabetes mellitus or nicotine abuse. Lately, individual biological factors such as endothelial progenitor cells (EPC) are in the focus of translational research. EPC are pluripotent, mononuclear cells deriving from bone marrow. They are found to be involved in new vessel formation and vascular regeneration after arterial injury [1][2][3][4]. There is growing evidence that some EPC subpopulations, characterized by surface markers like CD34, CD133 or KDR, might exert protective properties in atherosclerosis [5][6][7][8]. Common cardiovascular risk factors like smoking, obesity, hypercholesterolemia, diabetes, inactivity or chronic inflammation are associated with reduced numbers and impaired function of EPC [9][10][11][12]. Whereas statins, ACE inhibitors, omega-3 fatty acids, estrogens and highdensity lipoprotein among others have been correlated with increased EPC levels [13,14].
Optical coherence tomography (OCT) has emerged as the new gold-standard of intracoronary imaging due to its high resolution, fast acquisition and ease of handling in the catheterization laboratory [15,16]. Consensus in interpreting OCT images regarding its various aspects (e.g. vessel wall, stent struts, scaffold boxes, thrombi, etc.) has been established [17][18][19]. However, standardized algorithm to analyse OCT pullbacks and to enhance comparability of results are still lacking [20]. Coronary plaques, their relation to disease progression and to acute coronary syndromes ("vulnerable plaques") as well as to individual biomarkers are of increasing interest. Therefore and beyond a thorough analysis of atherosclerotic plaques, quantification of coronary plaque burden is needed.

Objectives
We (1) assessed the influence of EPC on coronary plaque burden using OCT in stented vessel segments in stable coronary artery disease (CAD). (2) A novel parameter of plaque burden "luminal surface area" was introduced and a computational OCT algorithm was implemented that allows quantification of coronary plaques in a segment of interest and subsequent visualization as comprehensive spread-out vessel charts.

Study design and patient population
The study was conducted according to the principles of the Declaration of Helsinki and approved by the local ethics committee of the University Hospital of Jena. The main study (OCTOPUS trial) is registered at clinicaltrials.gov, NCT:01056744. All study participants gave written informed consent. The data presented in this manuscript are collected in a sub-study from the OCTOPUS trial. The design and results of the main study have already been published [21,22]. Briefly, the trial was intended to randomly compare two interventional devices (Xience V™ drug-eluting stents vs. bare metal stents post-dilated with a paclitaxel-coated balloon) in patients with stable CAD.

Optical coherence tomography and identification of coronary plaques
Time-domain OCT imaging (M2 CV system, LightLab Imaging Inc., Westford, MA, USA) of the stented vessel segment (region of interest, ROI) was performed six months after the index percutaneous coronary intervention (PCI) according to the study protocol. Due to the old occlusive technique OCT imaging was limited to the stented vessel segment. OCT images were analyzed frame by frame.
A novel algorithm was applied to calculate total plaque burden in terms of underlying total plaque volume and total luminal plaque surface area of a stented vessel segment: Each cross section was analyzed regarding the existence of atherosclerotic plaques situated abluminal from stent struts. Plaque morphology was assessed according to international consensus and each plaque was classified as fibrotic, fibrocalcific, fibroatheroma or mixed [17][18][19]. Subintimal/sub-strut plaques in the treated vessel segment were manually traced in each cross section and entered into a computational algorithm (Fig. 1).
Quantitative plaque assessment consisted of: -Plaque count of each subtype -Volumes of each plaque were computed through the integral of cross-sectional plaque measurements over plaque length ( Fig. 1) according to the formula: (Vol Plaque : plaque volume, x p : proximal plaque location, x d : distal plaque location, CSA Plaque : cross-sectional plaque area on each frame) -Luminal surface area of each plaque, as displayed in the spread-out vessel chart ( Fig. 2 -top), was calculated using the Gauss's area formula, also known as Shoelace formula. Fig. 3 describes this mathematical method to determine areas as simple polygons "projected" at the luminal surface for coherent visualization (Fig. 3) [23,24].
-Minimal cap thickness, manually measured at its narrowest site of magnified OCT cross-sections, using the mean value of two measurements.
The changing position of the imaging catheter might have been misleading if only polar coordinates using its position had been used. Therefore, the centroid of the cross-section and not the imaging catheter was used as a reference for determination of coordinates.
Finally, total volume and total luminal surface area were computed for each morphologic plaque subtype (fibrotic, fibroatheroma, fibrocalcific and mixed) for the analysed vessel segment (Fig. 2 -bottom). In 7 patients measurements of plaque surface area and volume have been performed separately by two independent observers (K.H., K.N.) to assess variability of observations.

Statistical Analysis
All parameters were archived in a custom-made database (Microsoft Access, Microsoft Inc.). All calculations were done with SPSS (version 21.0, IBM SPSS statistics). Continuous variables are expressed as mean ± SD and categorical variables are presented as percent. Bivariate analysis with the Pearson correlation coefficient was used to determine the strength of relationship between EPC levels and plaque burden. Multiple linear regression analysis was done to test for potential confounding variables. The variability of plaque volume and surface area measurements has been assessed using 2-tailed paired Student's t test and Bland and Altman agreement analysis between two independent observers blinded to each other's results. Statistical significance was assumed for p-values <.05 (two-tailed).

Study population and procedural characteristics
In this sub-study of the OCTOPUS trial 44 lesions in 43 patients were investigated with OCT during reangiography 6 months after elective coronary stenting. Table 1 shows the clinical characteristics of the study population. Most patients were male (69.8%) and a high percentage of coronary risk factors were observed (Table 1). All patients received statin medication. Procedural characteristics of the study population are described in Table 2. Most stented lesions were rather complex (only AHA/ ACC type B or C lesions; 20.5% chronic total occlusions, Table 2). All lesions were treated with a drug-    [26]. No differences in EPC counts were found between the groups of DES vs. BMS/DCB.

Coronary Plaque Burden and Association with EPC
EPC levels were measured in all patients at baseline, and their mean values are displayed in Table 1. Leukocyte and lymphocyte numbers, which could bias EPC counts, ranged within standard normal values in all patients. A total of 110 coronary plaques were identified in 44 stented vessel segments (ROI) by OCT at 6-month invasive f/u. More precisely, coronary plaques were found in all, but 2 vessel segments. Table 3 shows the distribution of different plaque types per ROI. There was a mean of 2.5 ± 1.5 plaques per ROI (Table 3). Intracoronary thrombi within the treated vessel segment were not observed. Since we investigated stented vessel segments thin-cap fibroatheroma (TCFA) were not observed by definition. Focal tissue prolapse into the lumen between stent struts was found in five stents with a range of 1 to 5 different intra-stent localizations. Microvessels were only observed in one ROI. Overall, a total plaque volume of 7.5 ± 8.5 mm 3 and a luminal plaque surface area of 12.4 ± 13.7 mm 2 were found per ROI with no significant differences between different stent types (Table 3). Of note, we did not find any neoatherosclerotic plaques within neointima tissue formation.
Variability of observations has been assessed on two independent sets of measurements of total plaque volume and total plaque surface area within the stented segment. The mean absolute differences and their standard deviations were: 0.27 ± 0.51 mm 3 and 0.55 ± 1.12 mm 2 , respectively. Corresponding Bland and Altman plots are given in Fig. 4. No significant differences were found between independent measurements of each analyzed plaque with respect to surface area and volume.
All three EPC subpopulations inversely correlated with the identified sub-stent plaque volume and luminal plaque surface area (Fig. 5, Table 4) showing a significance of p ≤ 0.012. No associations were found between coronary plaque burden and various proliferation parameters (angiography and OCT) as well as stent apposition and stent strut coverage. Multivariable analysis could exclude parameters like LDL-cholesterol, age and creatinine as confounders of the correlation between EPC and plaque burden (Table 5). Using a univariate analysis in a general linear model including also the treatment arm no further significant associations were found.

Visualization of OCT findings as spread out vessel charts
To visualize OCT findings, spread out vessel charts of each stented vessel segment (ROI) were generated using the computational algorithm. Vessel segments are displayed longitudinally, unfolded and bisected in these spread outs, and give an en-face view of atherosclerotic plaques and their spatial relation to adjacent structures like overlaying stent struts or side branches (Fig. 2). Color-coding was added to identify lumen and stent  perimeter, plaque type and stent coverage (neointimal thickness in grey color bar) at a glance (Fig. 2).

Discussion
In this study, we investigated atherosclerotic plaque burden of treated vessel segments by OCT sixmonths after elective PCI. To the best of our knowledge, this is the first study analyzing OCT-derived coronary plaque measurements and their association to individual EPC levels.
We found a mean of 2.5 ± 1.5 plaques with a volume of 7.5 ± 8.5 mm 3 per stented vessel segment. Most plaques were fibroatheroma or mixed, lipid-enriched plaques. Neoatherosclerotic plaque formation was not observed six months after stent implantation.
EPC are seen as a surrogate biomarker of cardiovascular health and vascular functional status [7]. A decreased EPC pool has been linked to adverse cardiovascular events, in-stent neointimal hyperplasia and endothelial dysfunction [5][6][7]25]. Current evidence for the protective role of EPC mainly derives from animal and ex vivo studies, while most clinical studies rely on surrogate parameters for atherosclerotic disease severity [27][28][29][30]. So far, human EPC studies measuring atherosclerotic plaques directly have not been reported except for studies investigating carotid-intima media thickness [31][32][33]. Our data show a strong inverse association between the numbers of circulating EPC and coronary plaque burden measured as plaque volume and plaque surface area by OCT in a clinical study of patients with stable CAD and previous PCI.
Evidence regarding the role of EPC in atherogenesis has been somewhat conflicting. Their protective effects were doubted, and even pro-atherogenetic properties have been postulated in the past [34][35][36]. A few fundamental questions have to be raised in this controversy: (1) Which phenotype of EPC was investigated? (2) How is their functional status? (3) How do they execute their effects and which mediators and mechanism trigger their action? Unfortunately, there is no exclusive EPC marker. Despite a large number of in vitro and in vivo studies, it is still methodologically challenging and remains unclear which discrete population of EPC really triggers adult vasculogenesis and exerts anti-atherosclerotic effects. Currently, the combined expression of CD34+, CD133+ and KDR+, are considered as the definition for EPC [9,37]. Our study, investigating this predefined EPC phenotype shows that a high EPC count is associated with only Recently, EPC have been incorporated as a key therapeutic target in a novel stent design for interventional treatment of coronary stenoses: The Combo™ stent (OrbusNeich Medical) is a drug-eluting stent consisting of sirolimus coating on the abluminal stent side and EPC capturing antibodies on the luminal stent side [38,39]. IVUS imaging revealed a relevant difference regarding formation of neoatherosclerosis. EPC capturing led to a more homogenous fibrotic in-stent neointima without signs of a confluent necrotic core, which adds to previous EPC studies on plaque stabilization by modifying plaque composition, and thus reducing plaque vulnerability. [28,38] However, this could not translate into a measurable clinical benefit for the treated patients. Our study discovered the link between EPC count and plaque burden for the first time. In our study we only provide plaque burden of stented vessel segments at a predefined time point. Our data do not allow conclusion on plaque stabilization since we did not perform OCT at baseline. Further studies are needed to longitudinally evaluate possible alterations in plaque burden and composition with respect to variations of EPC counts and systemic or local therapeutic interventions (e.g. drugeluting devices, statins, anti-diabetic medication). Due to the relatively small sample size we are not able to further evaluate possible interactions of the used antiproliferative drug (everolimus vs. paclitaxel), duration of drug action (polymer coating vs. single shot application by a balloon) and individual biological factors like EPCs. Unfortuantely previous interventional studies investigating EPC counts and outcomes did not state the used types of stents [8,40]. Also, we cannot anticipate the strength of correlation between the plaque content of a stented vessel segment and an entire native coronary vessel in stable CAD.
Currently, the PREMIER trial is investigating a strategy of intense lipid elimination (statin therapy + LDL-apheresis) on atheroma volume reduction and EPC mobilization compared to standard statin therapy [41]. A bias regarding co-medication in our study is unlikely since all patients received statin therapy.

Methodology of coronary plaque assessment
Plaque volume assessment has been mainly conducted with IVUS. OCT studies have been focusing on TCFAs and ruptured plaques in ACS patients. [42][43][44][45][46][47]. Less sophisticated OCT plaque analysis using just the count of observed plaques or plaque circumference measurements (e.g. lipid arc) are insufficient to estimate the entire plaque burden and severity of atherosclerosis. Plaque length was for example only measured in areas with a lipid arc >90°, which also is a somewhat random cut-off. [48] Moreover, lipid arc measurements are prone to significant errors due to the variable eccentric position of the imaging catheter in the vessel. Contrary, our algorithm allows a more detailed quantification of plaque burden in an area or vessel of interest, including all noticeable plaques irrespective of size. We also introduce a novel parameter -luminal surface area of plaques which we see as a more reproducible measure of plaque burden, since it describes and measures for the first time the projection of a plaque on the luminal surface. Given the relevance of TCFAs it might be of interest if this new parameter of plaque dimension is better suited to predict adverse events (e.g. TCFA with  larger luminal surface area) than "classic" plaque size parameters (depth, length, volume). Moreover, we report a computational algorithm that enables in vivo quantification of the atheroma in a region of interest. Our 2D reconstruction of the luminal surface of a vascular segment superimposes information on plaque localization and extension, coverage thickness of struts in addition to strut localization, as well as stent und lumen perimeter (Fig. 2). Nevertheless, further longitudinal studies and especially a consensus about standardized plaque evaluation are needed to help its implementation into clinical practice.

Clinical implications
Taken the current evidence and the results of this study into account, EPC are an interesting specific target for prevention and disease modification of atherosclerosis. Furthermore, a low EPC count seems to be an independent risk factor for coronary atheroma burden. Translational research including in vivo imaging is needed to identify the link between EPCs, chemokines and atherosclerotic plaque development.

Limitations
Measurements of lipid-rich plaques are limited due to axial resolution of OCT. Therefore, precise definition of plaque borders far distally from the lumen can be challenging and partly insufficient. While the volumetric assessment of fibroatheroma might be therefore difficult, the surface area measurements of all plaques should not be hampered by this methodological drawback. Our data are stemming from a subgroup of the OCTOPUS trial with only moderate sample size (n = 44), and therefore the correlation coefficients are relatively low. Due to the occlusive OCT-technique with considerably shorter pullbacks compared to the newer generation OCT, plaque analysis was limited to the stented vessel segment. Since OCT was not performed at baseline, sequential data for coronary plaque burden cannot be provided. Taken these limitations together, our study results are rather descriptive and hypothesis generating, and should be followed by sequential plaque analysis studies in native coronary vessels. Finally, we cannot preclude on potential therapy-related atherosclerotic changes of stented plaques.

Conclusion
Our computational OCT analysis algorithm is feasible, allows quantification of coronary atheroma burden and comprehensible illustration as spread out vessel charts. Increased EPC levels are associated with less coronary plaque burden of a stented vessel segment, which adds to previous findings of their protective role in atherosclerosis. Further research is needed to refine EPC as a possible therapeutic target and thereby address individual biological differences.

Consent for publication
Not applicable.

Ethics approval and consent to participate
The study was conducted according to the principles of the Declaration of Helsinki and approved by the local ethics committee of the University Hospital of Jena. The main study (OCTOPUS trial) is registered at clinicaltrials.gov, NCT:01056744. All study participants gave written informed consent.