Neointimal tissue characterization after implantation of drug-eluting stents by optical coherence tomography: quantitative analysis of optical density

Abstract

Normalized optical density (NOD) measured by optical coherence tomography represents neointimal maturity after coronary stent implantation and is correlated with morphologic information provided by both light and electron microscopy. We aimed to test the hypothesis that even second generation drug-eluting stents (DESs) are problematic in terms of neointimal maturity. We implanted bare-metal stents (BMS: n = 14), everolimus-eluting stents (EESs: n = 15) or zotarolimus-eluting stents (ZESs: n = 12) at 41 sites in 32 patients with stable coronary artery disease. OCT was performed at up to 12 months of follow-up, and the average optical density of neointima covering struts was evaluated. NOD was calculated as the optical density of stent-strut covering tissue divided by the optical density of the struts. We also measured circulating CD34+/CD133+/CD45low cells, and serum levels of stromal cell-derived factor (SDF)-1, interleukin (IL)-8 and matrix metalloproteinase (MMP)-9 at baseline and follow-up. NOD was lower in the EES (0.70 ± 0.06) group than in the BMS (0.76 ± 0.07, P < 0.05) and ZES (0.76 ± 0.06, P < 0.05) groups. The mean neointimal area (R = 0.33, P < 0.05) and mean neointimal thickness (R = 0.37, P < 0.05) were correlated with NOD. Although NOD was not correlated with percent changes in circulating endothelial progenitor cells, and the levels of SDF-1 and IL-8, it was negatively correlated with the change in MMP-9 level (R = −0.51, P < 0.01). Neointimal maturity might be lower at EES sites than BMS or ZES sites. This might lead to impaired neointimal tissue growth and matrix degradation. These results suggest a specific pathophysiology after DES implantation.

Introduction

Drug eluting stents (DESs) have substantially reduced angiographic and clinical restenosis rates across broad lesion and patient subsets compared with bare metal stents (BMSs). However, current DESs do not necessarily benefit all patients with coronary artery disease (CAD). A unique issue beyond restenosis after DES implantation is the potential for impairment of re-endothelialization, which leads to lack of neointima formation and neointimal coverage over the stent struts. Exposed stent struts potentially cause stent thrombosis, which could happen even very late after stent implantation [1-5]. In addition, impaired re-endothelialization delays wound healing and causes a relapse of inflammation at the stent-injured sites when the anti-inflammatory effects of the drugs coated on DESs disappear [6], potentially inhibiting the formation of neointima. In the process of wound healing after stent implantation, CD34-positive cells are mobilized into the circulation and are positively correlated with an increased risk of restenosis [7].

Optical coherence tomography (OCT) assessment of stent-vessel interaction has been used in vivo as a surrogate endpoint to evaluate the healing process after DES implantation. The pixel intensity (optical density) of the tissue covering the stent struts based on OCT imaging, normalized for optical density of the stent struts, provides information about neointimal maturity [8-11].

First-generation DESs, such as sirolimus-eluting stents (SES) and paclitaxel-eluting stents (PESs), were followed by the development of second-generation DESs, such as zotarolimus-eluting stents (ZESs) and everolimus-eluting stent (EESs), and these newer DESs are now mainly used in clinical practice. Although these new-generation DESs have been suggested to be better in terms of local wound healing in the stent-injured vessel wall sites compared with first-generation DESs, this has not yet been completely established.

This study was designed to compare neointimal maturity after BMS, ZES and EES placement, and to investigate the pathophysiological mechanism of neointimal maturity and its role in vascular healing after stent implantation.

Methods

Study design

The subjects included 32 consecutive patients (22 men and 10 women, aged 70 ± 9 years) with stable coronary artery disease who underwent elective coronary stent implantation for organic lesions in a single coronary artery using a BMS (n = 13), ZES (n = 7) or EES (n = 12). When multiple stenting was required in a single patient, the same type of stents were implanted. Stent choice among BMSs, ZESs or EESs was based upon each operator’s decision. All of the patients received dual anti-platelet therapy with 81 mg of aspirin and 75 mg clopidogrel at least until the follow-up coronary angiography. Follow-up coronary angiography was recommended at 12 months after stent implantation, but was performed earlier if necessary based on clinical indications. At the time of follow-up coronary angiography, we assessed re-endothelialization and neointimal growth at each of the stented sites using OCT imaging. Also, peripheral blood samples were collected from all patients at baseline before stenting and at the follow-up study. The blood samples were immediately collected into tubes containing ethylene diaminetetraacetate (EDTA) and plain tubes. We measured the number of circulating CD34+/CD13+/CD45low progenitor cells using the blood in the EDTA tubes, and serum level of biomarkers such as stromal cell-derived factor (SDF)-1α, interleukin (IL)-8, and matrix metalloproteinase (MMP)-9 using the blood in the plain tubes. All patients provided written informed consent. The study protocol was approved by the local institutional review board in Dokkyo Medical University (Mibu, Tochigi, Japan) and complied with the Declaration of Helsinki.

Optical coherence tomography imaging and analysis

At the time of follow-up coronary angiography, OCT was performed using a frequency-domain system (C7-XR FD-OCT Intravascular Imaging System; St. Jude Medical, St. Paul MN) after administration of intracoronary nitroglycerin. Two independent readers from the Cardiovascular Imaging Core Laboratory (Cardiovascular Imaging Core Laboratory, Harrington Heart & Vascular Institute, University Hospitals, Cleveland Medical Center, OH), blinded to patient information, performed quantitative and qualitative OCT analyses using dedicated software (Off-line Review Software, version E.0.2, St. Jude Medical). All cross-sectional images were initially screened for quality assessment and excluded if any frame had a stent that was out of the screen or if the image had poor quality caused by residual blood, artifact or reverberation [12]. For morphometric analysis, standard definitions of cross-sectional area and volume measurements were applied, as previously reported [12]. Strut-level analysis was performed using all analyzable frames along the stented segment. Stent strut coverage and malapposition were defined and measured as described previously [13, 14].

Optical density of stent strut covering tissue

Based on OCT imaging properties, we evaluated the pixel intensity (optical density) of stent strut covering tissue localized on the inner side of the struts, and normalized the value for the optical density of the stent struts (Fig. 1). The values of optical density were obtained automatically using proprietary computer-assisted analysis software (St. Jude Medical). A previous study showed that normalized optical density (NOD), defined as the optical density of stent strut covering tissue divided by the optical density of the stent struts, was significantly correlated with morphologic information provided by both light and electron microscopy [8]. NOD was measured on all struts in the cross-section every 0.6 mm in the same frames used to perform quantitative analysis. A region of interest was manually drawn by two experienced OCT analysts, and the results were reviewed by a third analyst. The average of all the measured struts was calculated as a marker of neointimal maturity [9-11].

Fig. 1.

Calculation of normalized optical density (NOD)

Biomarker measurements

We measured circulating CD34+/CD133+/CD45low cells, which included endothelial progenitor cells, using flow cytometry based on a previously described method [7, 15, 16] with minor modifications. The reagent mixture included a peridinine chlorophil protein (PerCP)-conjugated anti-CD45 antibody (Becton Dickinson), a fluorescein isothiocyanate (FITC)-conjugated anti-CD34 antibody (Becton Dickinson) and a phycoerythrin (PE)-conjugated anti-CD133 antibody (Miltennyi Biotec). Isotype controls were used as negative controls based on the species and immunoglobulin (Ig) G control antibodies (IgG1 isotype control; Becton Dickinson). Flow cytometric analysis was performed using the FACS Calibur laser flow cytometer (Becton Dickinson) according to the manufacturer’s instructions. The absolute number of CD34+/CD133+/CD45low cells per milliliter was calculated based on the cells-to-the white blood cell count ratio. To minimize any methodological variations, each sample was analyzed with two independent experiments, and the mean value was calculated.

In this study, we also measured serum levels of stromal cell-derived factor (SDF)-1α, interleukin (IL)-8, and matrix metalloproteinase (MMP)-9, using an enzyme-linked immunosorbent assay (ELISA). We used commercially available ELISA kits (Quantikine ELISA kits; R&D Systems) to measure SDF-1α (Human CXCL12/SDF-1α Immunoassay), IL-8 (Human CXCL8/IL-8 Immunoassay), and MMP-9 (Human MMP-9 Immunoassay). The procedure was performed according to the manufacturer’s instructions. Each sample was assayed in duplicate for standards and controls; and high, medium and low standards were included in each run. The colorimetric reactions were read on an automatic microplate reader set to 450 nm. The concentration of each biomarker in a sample was determined by interpolation from a standard curve.

Statistical analyses

The normality of the distributions of variables was assessed using the Kolmogorov–Smirnov test with Lilliefors’ correction. If the distributions were not parametric, the data were transformed into logarithmic values. Data are presented as means ± standard deviations (SD). Differences among the three groups were assessed with a chi-square test for categorical variables and with an analysis of variance (ANOVA) followed by Scheffe’s post hoc test for continuous variables. Intra-group comparisons between baseline and follow-up were performed using a paired t-test for normally distributed data and a Wilcoxon Rank Sum test for data with a skewed distribution. Spearman’s correlation analysis was used to assess the relationship between two parameters. P < 0.05 was considered significant.

Results

Baseline characteristics were compared among each patient group that had BMS, ZES or EES implantation (Table 1). There were no significant differences in age, gender, coronary risk factors or medications among the three groups. The follow-up times were also similar in the BMS, ZES and EES groups (8.3 ± 3.0, 9.7 ± 2.1 and 9.6 ± 2.1 months, respectively).

Table 1.

Baseline characteristics

	BMS (n = 13)	ZES (n = 7)	EES (n = 12)
Age: years	69 ± 11	72 ± 6	70 ± 8
Male gender [n (%)]	10 (77)	5 (71)	6 (46)
Coronary risk factors [n (%)]
Hypertension	11 (85)	6 (87)	12 (92)
Diabetes	5 (39)	2 (29)	7 (54)
Dyslipidemia	9 (69)	6 (86)	11 (85)
Current smoking	7 (54)	4 (57)	7 (54)
Medications [n (%)]
Statins	9 (69)	5 (71)	12 (92)
ACEIs/ARBs	10 (77)	5 (71)	12 (92)
Antidiabetic agents	3 (23)	2 (29)	5 (39)
Follow-up interval (months)	8.3 ± 3.0	9.7 ± 2.1	9.6 ± 2.1

BMS bare metal stent, ZES zotarolimus-eluting stent, EES everolimus-eluting stent, ACEI angiotensin converting enzyme inhibitors, ARB angiotensin receptor blocker

In the 32 study patients, a total of 41 stents (BMS: n = 14, ZES: n = 12, EES: n = 15) were implanted. The conventional OCT image analysis data for each stent site are shown in Table 2. The mean neointimal area (BMS vs. EES, P < 0.001; ZES vs. EES, P < 0.001), neointimal volume (BMS vs. EES, P < 0.001; ZES vs. EES, P < 0.05) and mean neointimal thickness (BMS vs. EES, P < 0.001; ZES vs. EES, P < 0.01) were lower in the EES group than the BMS and ZES groups. The percentage of uncovered struts was higher in the EES group than the BMS (P < 0.05) and ZES (P < 0.01) groups. The percentage of malapposed struts was also higher in the EES group than the BMS (P < 0.05) and ZES (P < 0.05) groups. Furthermore, the NOD was lower in the EES group than the BMS (P < 0.05) and ZES (P < 0.05) groups (Table 2).

Table 2.

Conventional OCT image analysis data

	BMS (n = 14)	ZES (n = 12)	EES (n = 15)
Mean reference diameter (mm)	2.83 ± 0.38	0.59 ± 0.26	0.41 ± 0.46
Mean reference area (mm2)	6.45 ± 1.75	5.41 ± 1.10	4.79 ± 1.83
Mean stent area (mm2)	8.80 ± 2.91	8.22 ± 1.57	6.53 ± 2.06
Mean luminal area (mm2)	2.49 ± 0.53	3.88 ± 1.19	4.20 ± 1.99
Mean neointimal area (mm2)	3.61 ± 1.55	2.46 ± 0.86	0.93 ± 0.59**,†††
Mean neointimal thickness (μm)	401 ± 151	265 ± 75	119 ± 93**
Uncovered struts (%)	0.91 ± 1.56	0.32 ± 0.57	11.0 ± 15.3*,††
Malapposed struts (%)	0.05 ± 0.19	0.06 ± 0.20	1.34 ± 2.85*,†
NOD	0.76 ± 0.07	0.76 ± 0.06	0.70 ± 0.06*,†

BMS bare metal stent, ZES zotarolimus-eluting stent, EES everolimus-eluting stent, OCT optical coherence tomography, NOD normalized optical density

^*P < 0.05

^**P < 0.001 versus BMS

^†P < 0.05

^††P < 0.01

^†††P < 0.001 versus ZES

In 28 of the 32 study patients, circulating CD34+/CD133+/CD45low cells and levels of SDF-1α, IL-8 and MMP-9 were measured both at baseline before stenting and at the time of follow-up. There were no significant differences in the values of these biomarkers among the three groups at either baseline or follow-up. The percent change in the value ([follow-up value – baseline value] – 100/baseline value) of CD34+/CD133+/CD45low cells was less in the EES group than the BMS group (P < 0.05). However, there were no significant differences in the percent changes in the levels of SDF-1α, IL-8 and MMP-9 among the three groups (Table 3).

Table 3.

Comparison of circulating biomarkers among the BMS, ZES and EES groups

	BMS (n = 13)	ZES (n = 6)	EES (n = 9)
CD34+/CD133+/CD45low cells
Baseline	262 ± 120	326 ± 240	342 ± 88
Follow-up	432 ± 251	323 ± 226	336 ± 108
% Change	69 ± 67	16 ± 77	−1 ± 25*
SDF-1α
Baseline	2109 ± 670	1859 ± 713	1878 ± 749
Follow-up	2220 ± 799	1677 ± 387	2098 ± 939
% Change	27 ± 82	0 ± 34	16 ± 52
IL-8
Baseline	203 ± 72	173 ± 29	191 ± 25
Follow-up	196 ± 77	183 ± 47	213 ± 62
% Change	1 ± 30	7 ± 30	13 ± 35
MMP-9
Baseline	3220 ± 1797	2655 ± 748	2481 ± 906
Follow-up	2341 ± 760	2331 ± 729	2600 ± 1547
% Change	−17 ± 28	−11 ± 24	11 ± 53

SDF-1α stromal cell-derived factor-1^α, IL-8 interleukin-8, MMP-9 matrix metalloproteinase-9

^*P < 0.05 versus BMS

The mean neointimal area (R = 0.33, P < 0.05) and mean neointimal thickness (R = 0.37, P < 0.05) were correlated with the NOD value (Fig. 2). The NOD value was not correlated with each of the percent changes in CD34+/CD133+/CD45low cells (R = 0.16), SDF-1α level (R = 0.10) and IL-8 level (R = 0.32), but it was negatively correlated with the percent change in MMP-9 level (R = −0.51, P < 0.01) (Fig. 3).

Fig. 2.

Relationship between NOD and neointimal growth. The mean neointimal area and mean neointimal thickness were correlated with the NOD. NOD normalized optical density

Fig. 3.

Relationship between NOD and percent change in biomarkers ([value at follow-up – baseline value] × 100/baseline value). NOD was not correlated with changes in CD34+ CD133+ CD45low cells or SDF-1α and IL-8 levels, but was negatively correlated with the change in MMP-9 level. SDF stromal cell-derived factor, IL interleukin, MMP matrix metalloproteinase

Discussion

In the present study, we demonstrated that the OCT-based NOD as well as neointimal growth and neointimal strut coverage were lower at EES sites than at BMS and ZES sites, and that the mean neointimal area and mean neointimal thickness were correlated with the NOD over the entire length of the stented segment. Representative OCT cross-sectional images with NOD analysis for each stent are shown in Fig. 4. The results suggest that neointima might be immature at EES sites, compared with BMS and ZES sites, and that neointimal maturity might lead to neointimal tissue growth.

Fig. 4.

Representative OCT cross-sectional images with NOD analysis for BMS (a and b), ZES (c and d) and EES (e and f) are shown. The methodology for calculation of NOD is described in Fig. 1

Generational advances in DES technology have resulted in reduced rates of target lesion revascularization across broad patient and lesion subsets with improved safety with respect to stent thrombosis. However, concerns over incomplete stent healing even with second-generation DESs persist because the annual rate of target lesion failure still remains at 2–4%, which is similar to the rates observed after implantation of BMSs or first-generation DESs [17]. From a vascular biology perspective, there is consensus that late-lumen loss and neointimal thickening (i.e., restenosis) are the biological response to vascular injury characterized by a sequence of endothelial denudation, platelet deposition and inflammatory cell recruitment, smooth muscle cell migration and proliferation, and extracellular matrix deposition. Re-endothelialization and complete neointimal coverage of the stent struts are commonly viewed as markers of favorable vascular healing [2, 18]. Recently, characterization of tissue maturation after DES implantation has been recognized as an important step forward in the assessment of vascular healing. Histological analysis of regions of mature neointimal tissue showed significantly higher smooth muscle cells and, in turn, proteoglycan/collagen, fibrin and inflammatory cells. In contrast, immature neointimal tissue was predominantly composed of macrophages with minimally interspersed smooth muscle cells [11]. Thus, neointimal immaturity is potentially associated with development of neoatherosclerosis, the atherosclerotic change within neointima [19] that has been regarded as a primary mechanism of late-stent failure after DES implantation [20]. A similar incidence of neoatherosclerosis has been observed for both second-generation and first-generation DESs [21]. Tsujita et al. showed that intravascular ultrasound (IVUS)-derived necrotic and less-fibrotic neointimal formation were frequently observed in the site of DESs implantation using dedicated software [IVUS-based tissue characterization software: iMap (Boston Scientific, Natick, MA)]. The result of our study that NOD was lower at EES sites than at BMS and ZES sites suggests a potential risk for neoatherosclerosis at EES sites. These findings are consistent with previous reports.

We have been investigating the role of circulating bone marrow-derived progenitor cells on the biological response after stent deployment using translational research techniques that include both flow cytometry and OCT imaging. First, we demonstrated that mobilization of CD34+ increased from baseline during the acute phase (7–14 days) after BMS deployment, and was associated with late-lumen loss and restenosis at follow-up coronary angiography. The first-generation SES suppressed late-lumen loss and CD34+ cell mobilization, raising the question of whether neointimal suppression is inexorably linked with impaired re-endothelialization [2, 7]. Next, we observed that even second generation DESs, including ZESs and EESs, suppressed the mobilization of CD34+/CD133+/CD45low cells during the acute phase after stent deployment (on day 7) and inhibited subsequent neointimal formation. Furthermore, EES deployment was associated with an increased number of uncovered stent struts, compared with BMS deployment [22]. Finally, we also performed a study in vitro to investigate the pharmacological action of the drugs coated on the surface of DESs on the differentiation of progenitor cells into vascular cells. The results showed that everolimus, zotarolimus and sirolimus all inhibited differentiation of progenitors into endothelial cell as well as smooth muscle cell-like lineages, indicating a drug class effect on progenitor cell differentiation [23]. These results strongly suggest that neointimal suppression utilizing DESs with -limus derivatives are biologically linked with impaired wound healing responses (i.e., re-endothelialization and incomplete stent strut coverage) as a consequence of reduced mobilization of stem cell progenitors, especially in the case of EESs. In the present study, we could not measure CD34+/CD133+/CD45low cells during the acute phase but did so at the time of follow-up coronary angiography. The percent change in the number of CD34+/CD133+/CD45low cells were less in the EES group than the BMS group. These results suggest the possibility that lower progenitor cell mobilization might be associated with neointimal immaturity in the EES lesions. We also measured serum levels of SDF-1α, IL-8 and MMP-9, all of which are considered to be related to mobilization of progenitor cells. Consequently, the NOD value was negatively correlated with the percent change in MMP-9 level. Since MMP-9 is a proteolytic enzyme that promotes matrix degradation and plays a pathophysiologic role in atherosclerotic plaque vulnerability and rupture, our results strongly suggest that MMP-9 contributes to neointimal immaturity in the vascular healing process after stent injury. Although DESs strongly inhibit the inflammatory reaction during the acute phase after stent implantation, there is impairment of the subsequent wound-healing response initiated by impaired re-endothelialization. This causes a relapse of inflammation, and inflammatory cells such as lymphocytes as well as macrophages play an essential role [6, 24-26]. The MMP-9 released from these cells possibly inhibits neointimal maturity, leading to impairment of vascular healing at the stented vessel sites.

Limitations and clinical implications

In the present study, we assessed NOD as a marker of neointimal maturity. The NOD was calculated based on the average neointima for all stent struts in the analyzed frames. We assessed the average neointimal maturity because we tried to perform quantitative assessment. However, tissue characteristics of neointima are heterogeneous, and our analysis did not include any histological validation. For example, collagen fiber-rich neointima exhibits stronger optical scattering and therefore a stronger OCT signal, whereas tissues made up of lipids, such as the necrotic core, appear as low-intensity areas with diffuse borders [27]. Therefore, detailed qualitative assessment for each type of neointima is needed.

Despite the suitability of NOD assessment for vascular healing after stent implantation, this model remains hampered by its inability to reflect important patient comorbidities and other dependent parameters, such as the presence of risk factors and the influence of medications that may affect stent healing. Since the number of samples was limited, further studies are required to confirm the significance and application of our findings in a larger cohort. Despite a number of limitations, our results suggest that it is important to assess neointimal quality such as tissue characteristics in addition to neointimal growth in terms of physiological vessel wound healing after stent deployment.

Conclusions

Neointimal maturity might be lower at EES sites, compared with BMS and ZES sites. This might lead to impaired neointimal tissue growth, and might be associated with matrix degradation. These results suggest a specific pathophysiology after DES implantation.