Author + information
- Received November 20, 2009
- Revision received February 1, 2010
- Accepted February 22, 2010
- Published online May 1, 2010.
- Giulio Guagliumi, MD⁎,⁎ (, )
- Giuseppe Musumeci, MD⁎,
- Vasile Sirbu, MD⁎,
- Hiram G. Bezerra, MD, PhD†,
- Nobuaki Suzuki, MD†,
- Luigi Fiocca, MD⁎,
- Aleksandre Matiashvili, MD⁎,
- Nikoloz Lortkipanidze, MD⁎,
- Antonio Trivisonno, MD⁎,
- Orazio Valsecchi, MD⁎,
- Giuseppe Biondi-Zoccai, MD‡,
- Marco A. Costa, MD, PhD†,
- on behalf of the ODESSA Trial Investigators
- ↵⁎Reprint requests and correspondence:
Dr. Giulio Guagliumi, Division of Cardiology, Ospedali Riuniti di Bergamo, Largo Barozzi 1, 24121 Bergamo, Italy
Objectives We designed a randomized trial exploiting optical coherence tomography (OCT) to assess coverage and apposition of overlapping bare-metal stents (BMS) and drug-eluting stents (DES) in human coronary arteries.
Background Overlapping DES impair healing in animals. Optical coherence tomography allows accurate in vivo assessment of stent strut coverage and apposition.
Methods Seventy-seven patients with long coronary stenoses were randomized to overlapping sirolimus-eluting stents (SES), paclitaxel-eluting stents (PES), zotarolimus-eluting stents (ZES), or BMS. The primary goal of the study was to determine the rate of uncovered/malapposed struts in overlap versus nonoverlap segments, according to stent type, at 6-month follow-up with OCT.
Results A total of 53,047 struts were analyzed. The rate of uncovered/malapposed struts was 1.5 ± 3.4% and 0.6 ± 2.7% in overlap versus nonoverlap BMS (p = NS), respectively, and 4.3 ± 11% and 3.6 ± 8% in overlap versus nonoverlap DES (p = NS), respectively. There were no differences in the rates of uncovered/malapposed struts between overlapping BMS and DES, likely due to low frequency of uncovered/malapposed struts in ZES (0.1 ± 0.4%), which offset the higher rates observed in SES (6.7 ± 9.6%) and PES (6.7 ± 16.5%, p < 0.05). Overlap segments showed greater neointimal volume obstruction versus nonoverlap segments in all DES (p < 0.05 for all DES types). Strut-level neointimal thickness at overlap and nonoverlap segments were lowest in SES (0.16 ± 0.1 mm and 0.12 ± 0.1 mm, respectively) compared with PES (0.27 ± 0.1 mm and 0.20 ± 0.1 mm, respectively), ZES (0.40 ± 0.16 mm and 0.33 ± 0.13 mm, respectively), and BMS (0.55 ± 0.31 mm and 0.53 ± 0.25 mm, respectively, p < 0.05).
Conclusions As assessed by OCT the impact of DES on vascular healing was similar at overlapping and nonoverlapping sites. However, strut malapposition, coverage pattern, and neointimal hyperplasia differ significantly according to DES type. (Optical Coherence Tomography for Drug Eluting Stent Safety [ODESSA]; NCT00693030)
- coronary artery disease
- optical coherence tomography
- percutaneous transluminal coronary angioplasty
Drug-eluting stents (DES) are used with increasing frequency in complex lesions and patients. However, blocking cellular proliferation is not without consequences, and higher incidence of late stent thrombosis (LST) has been associated with DES compared with bare-metal stents (BMS) (1). Postmortem studies have identified uncovered struts as the most powerful pathologic risk factor for LST (2). Moreover, DES have been associated with higher incidence of malapposition or incomplete apposition, also predictors of LST (3,4).
Patients with long coronary lesions often require multiple DES in overlap and might be at higher risk of stent thrombosis, because double layers of DES might expose the vessel wall to higher and potentially toxic doses of drug and polymer and alter flow (5). The impairment of arterial healing observed at overlapping DES in nonatherosclerotic preclinical animal models supports this hypothesis (6), but little is known about the local in vivo vascular effects of overlapping DES in humans.
Optical coherence tomography (OCT) can detect small degrees of in-stent neointima more accurately than intravascular ultrasound (IVUS) (7,8) and enables detailed strut-level analysis of tissue coverage and apposition in humans (9). We planned a randomized clinical trial with OCT to investigate stent coverage and apposition 6 months after deployment of overlapping DES or BMS.
Design, patients, and procedures
The ODESSA (Optical Coherence Tomography for Drug-Eluting Stent Safety) trial was a single-center randomized open-label trial with independent core laboratory imaging analyses (University Hospitals Cleveland Core Laboratory). The study protocol was approved by the Ethics Committee, and all patients provided written informed consent.
Eligible subjects were older than 18 years of age, with stable or unstable coronary syndromes, long stenosis (>20 mm) in native coronary arteries, 2.5 to 3.5 mm in diameter, requiring percutaneous coronary intervention (PCI) with overlapping stents. Exclusion criteria were left main disease, ongoing/recent myocardial infarction, previous target vessel stenting, ejection fraction ≤30%, creatinine >2.5 mg/dl, no suitable anatomy for OCT (ostial lesions and extreme tortuosity), and inability to comply with dual antiplatelet therapy and follow-up requirements.
Consecutive patients who signed informed consent were 2:2:2:1 randomized to sirolimus-eluting stents (SES) (Cypher, Cordis, Miami Lakes, Florida), paclitaxel-eluting stents (PES) (Taxus Libertè, Boston Scientific, Natick, Massachusetts), zotarolimus-eluting stents (ZES) (Endeavor, Medtronic, Santa Rosa, California) or BMS (Libertè, Boston Scientific). Only a single stent type was allowed in each patient.
Coronary angioplasty was performed according to standard techniques. Direct stenting was allowed, and the recommended overlap length was 2 to 4 mm by visual estimation. Glycoprotein IIb/IIIa inhibitors were at the operator's discretion. All patients were pretreated with aspirin 100 mg and clopidogrel 300 mg, followed by daily administration of clopidogrel 75 mg for at least 6 months after discharge plus aspirin. Patients were followed-up at 1 month and 6 and 12 months after discharge by office visits and readmitted for planned follow-up angiography, IVUS, and OCT imaging at 6 months.
Quantitative coronary angiography and IVUS
Digital angiograms were analyzed offline with CAAS II (PIE Medical, Maastricht, the Netherlands) (10). The stented segment plus 5-mm distal and proximal edges were selected for analysis. Reference vessel diameter, minimum luminal diameter, percent diameter stenosis, and lesion length were obtained. Procedure success was defined as <10% residual diameter stenosis with normal distal flow (Thrombolysis In Myocardial Infarction flow grade 3).
The IVUS was performed after injection of 200 µg nitroglycerin after the procedure and at 6-month follow-up with Atlantis SR Pro 40 MHz and iLab (Boston Scientific). Imaging included stents and the 5-mm stent edges with a 0.5 mm/s pullback. All IVUS data were digitally stored and subsequently analyzed with Curad 4.32 (Curad, WijkbijDuurstede, the Netherlands) (11). Lumen and stent cross-sectional areas were automatically measured, and neointimal hyperplasia (NIH) was calculated as the difference between stent and lumen area (12).
The OCT was conducted at 6-month follow-up, after 200 µg intracoronary nitroglycerine. A time domain OCT system (M2CV OCT Imaging System, LightLab Imaging, Westford, Massachusetts) was used. The occlusive technique was adopted to completely remove blood from the artery (9). Briefly, the occlusion balloon (Helios Goodman, Advantec Vascular, Sunnyvale, California) compatible with 6-F guiding catheters (0.071-inch inner diameter) was advanced distal to the stented segment over a conventional angioplasty guidewire. The guidewire was then replaced by the 0.019-inch OCT ImageWire (LightLab Imaging). The occlusion balloon was repositioned proximal to the stented segment and inflated at low pressure (0.4 to 0.7 atm for a maximum of 60 s; average 52.8 ± 7.3 s), with simultaneous infusion of Ringer's solution at 37°C with a flow rate of 0.5 to 1 ml/s through the distal tip of the catheter. Images were acquired with an automated pullback at a rate of 1.0 mm/s, generating 15 frames/s (Fig. 1). Images were digitally stored and submitted to the core laboratory for offline evaluation and subsequent analysis.
Measurements of OCT cross-sectional images were performed with a dedicated, automated contour-detection system (OCT system software B.0.1, LightLab Imaging) developed in collaboration with the University Hospitals imaging core laboratory. All cross-sectional images (frames) were initially screened for quality assessment and excluded from analysis if any portion of the image was out of the screen; a side branch occupied >45° of the cross-section; or the image had poor quality caused by residual blood, sew-up artifact, or reverberation. Frames including the ostium of side branches were excluded from analysis. A strut was considered suitable for analysis only if it had: 1) well-defined, bright “blooming” appearance; and 2) characteristic shadow perpendicular to the light source. Qualitative image assessment was performed in every frame (i.e., every 0.06 mm), whereas quantitative measurements were performed every 5 frames (i.e., every 0.33 mm) along the entire stented segment. In stented segments with multiple overlapping sites, overlap segments were labeled in numerical order from distal to proximal. The stented segment between 2 overlapping stents was labeled as middle nonoverlapping segment. In addition, OCT data were computed for every 2.5-mm consecutive segments, considering all the frames in this interval.
Strut-level qualitative OCT analysis was performed in each individual strut along the entire target segment. Struts were qualified in 4 categories: struts covered by tissue and not interfering with the lumen contour were defined as “non-protruding covered struts”; those interfering with the lumen contour were defined as “protruding covered struts”; uncovered struts were defined as “protruding uncovered” or “malapposed struts” depending on quantitative measures as described in the following text (Fig. 2).
The sharp contrast between lumen and vessel wall in OCT images allows fully automated delineation of the lumen contour. The inner and outer contours of each strut reflection (blooming) were delineated semiautomatically. The center of the luminal surface of the strut blooming was determined for each strut, and its distance to the lumen contour was calculated automatically to determine strut-level intimal thickness (SIT). Struts covered by tissue had positive SIT values, whereas protruding uncovered struts or malapposed struts had negative SIT. Distance between inner and outer strut reflection boundaries was measured in 2,250 struts to determine strut blooming thickness, which is essential to properly determine malapposition as described in the following text. Strut malapposition was determined when the negative value of SIT was higher than the sum of strut thickness plus polymer thickness, according to the specifications of each stent, corrected by strut blooming thickness, which was 37 ± 8 µm. To determine reproducibility of OCT measurements, quantitative analyses in 333 struts were repeated by 2 independent analysts and repeated 3 months after the initial analysis. The difference in stent area measurements between 2 analysts was 0.01 ± 0.04%, whereas absolute difference in SIT was 0.01 ± 0.02 µm (R = 0.997).
Study end points
The primary end point was the proportion of uncovered or malapposed struts in overlap versus nonoverlap segments in DES and BMS, as defined by OCT analysis. Secondary imaging end points included percentage of uncovered or malapposed struts in SES, PES, and ZES; percentage of NIH volume by IVUS; and late lumen loss by angiography at overlapping versus nonoverlapping sites in DES versus BMS. Secondary clinical end points included 12-month major adverse cardiac event rates, including death, myocardial infarction, and target vessel revascularization, as well as target lesion revascularization and Academic Research Consortium-defined stent thrombosis (13). All clinical events were adjudicated by an external independent clinical event committee.
Sample size was estimated from unpublished data, assuming a difference of 1.0% in the segment prevalence of uncovered/malapposed struts in overlap versus nonoverlap segments in each stent group, an SD of 1.0%, and aiming for 1% alpha and 10% beta when comparing 2 groups at a time (yielding 15 patients/group). After accrual of an additional 16 patients to take into account potential deviations from normality and losses to 6-month follow-up, 76 patients were necessary.
Continuous variables are expressed as mean ± SD or median (5th to 95th percentile), and categorical variables are expressed as n (%). For per-patient and per-lesion analyses, continuous variables were compared with analysis of variance and Gosset t tests, and categorical variables were compared with chi-square or Fisher's exact test. For per-segment or per-strut analyses, continuous variables were compared with a complex samples general linear model (CSGLM), depending on cluster features. Given the highly skewed distributions, all analyses for primary and secondary end points were confirmed with nonparametric tests. Computations were performed with SPSS version 16.0 (SPSS, Chicago, Illinois), with statistical significance set at the 0.05 level, and p values unadjusted for multiplicity are reported throughout.
Baseline and procedural data
A total of 77 patients were included in the study, with similar baseline and procedural characteristics (Table 1). Group allocations were as follows: SES (n = 22, 2.5 ± 0.5 stent/patient ratio), PES (n = 22, 2.4 ± 0.5 stent/patient ratio), ZES (n = 22, 2.5 ± 0.7 stent/patient ratio), and BMS (n = 11, 2.4 ± 0.5 stent/patient ratio), with ≥3 stents deployed in 30 lesions (38.9%).
The OCT imaging at follow-up was performed successfully in 75 eligible patients, because 1 patient died within 30-days after procedure, and 1 patient was excluded from analysis because of protocol violation (a covered stent was deployed in the target lesion due to a perforation during the index procedure). There were no adverse events associated with the OCT imaging procedures.
A total of 53,047 struts in 6,968 cross-sections of 250 stented segments were analyzed (with 10 segments unsuitable for analysis, and 40 segments of suboptimal quality due to: vessel contour out-of-screen [n = 16], side branches [n = 16], and artifacts [n = 8]). The primary end point (i.e., the rate of uncovered or malapposed struts) was similar in overlap versus nonoverlap segments, irrespective of stent type (p > 0.05 for any stent type) (Tables 2 and 3).⇓⇓ Nevertheless, higher overall rates of uncovered or malapposed struts were observed in SES (8.1 ± 11.2%) and PES (4.05 ± 10.3%) stented segments in comparison with ZES (0.06 ± 0.24%) and BMS (0.86 ± 2.9%, p < 0.001), which was driven mainly by differences in nonoverlap segments. Accordingly, the rate of uncovered/malapposed struts in nonoverlapping segments was 3.7 ± 8.0% for the whole DES group versus 0.6 ± 2.7% for BMS (p = 0.002).
Arterial response was heterogeneous among different DES platforms (Tables 2 and 3, Fig. 3). The SES had homogenously high rates of uncovered and malapposed struts in overlap and nonoverlap segments. The PES showed a trend of higher rates of uncovered and malapposed struts in the overlap segment. Notably, the ZES exhibited homogenous and almost complete strut coverage. The longitudinal distribution of uncovered and malapposed struts along the entire stented vessel is shown in Figure 3.
Percentage of volume obstruction was conversely higher in overlap compared with nonoverlap segments in all DES (Tables 2 and 3, Fig. 4). Strut level intimal thickness was significantly different across the different stent types.
Quantitative coronary angiography, IVUS, and clinical data
Quantitative coronary angiography and IVUS data are reported in Table 4. Angiographic and IVUS effectiveness parameters (i.e., late lumen loss, binary angiographic restenosis, NIH volume, and % NIH obstruction) were concordant with OCT findings, showing SES to have the lowest proliferative response compared with other stent types. Stent malapposition was noted by IVUS after procedure in 7 SES, 9 PES, 6 ZES, and 3 BMS only at nonoverlap segments (p = NS). At follow-up only SES and PES had newly acquired stent malapposition in nonoverlap (SES 11.54% and PES 1.92%) and overlap segments (SES 12.9%).
The 12-month rates of death, myocardial infarction, or target vessel revascularization were not significantly different across the groups (p = 0.097) (Table 5). There was 1 case adjudicated as probable subacute stent thrombosis in the SES group, and no late stent thromboses.
The present study establishes the feasibility and safety of using OCT to evaluate stent strut coverage and malapposition in a prospective, randomized, and controlled clinical trial. The OCT showed increased NIH at the overlap compared with nonoverlap stent segments, suggesting reduced efficacy at the site of overlapping DES. Conversely, similar rates of uncovered and malapposed struts were observed between overlap and nonoverlap DES, suggesting the safety of overlapping DES in humans. The study also confirmed the efficacy of DES (with SES being superior to PES, and PES being superior to ZES) in suppressing NIH in long coronary lesions at the expense of an overall higher incidence of uncovered or malapposed struts in SES and PES compared with ZES and BMS.
The success of DES in preventing restenosis likely contributed to the recent expansion in PCI indications to include complex coronary artery disease. Despite limited scientific evidence, the treatment of long coronary stenoses with multiple overlapping DES has become routine practice worldwide (14).
Strut coverage/apposition in overlapping stents
Histopathologic evaluations of DES have revealed a high incidence of uncovered struts (4,15), and suggested the ratio of uncovered to total number of stent struts to be the best morphometric predictor of LST (2). Previous IVUS studies have also suggested a role for stent malapposition in the pathogenesis of DES thrombosis (3), but limited image resolution hampered clinical confirmation. The present study used OCT to evaluate rate and distribution of in vivo uncovered and malapposed struts.
Overlapping SES and PES were associated with poor endothelial coverage in some animal models (15). In the present study, vascular responses among DES platforms were highly heterogeneous. There was an overall higher incidence of uncovered or malapposed struts in SES and PES, but this was not influenced by overlap. Uncovered or malapposed struts were equally distributed between overlap and nonoverlap in the SES, ZES, and BMS platforms. The relevance of the nonhomogeneous distribution of malapposed PES struts with numerically yet not statistically significant higher incidence of malapposed struts at the site of PES overlap compared with nonoverlap is unclear and warrants further investigation.
The rates of uncovered or malapposed struts observed in the present study are lower than previously suggested by histopathology data, even after restructuring the data in 2.5-mm segments (Fig. 3) to parallel histomorphometric analyses (2). The maximum rate of segments containing >30% uncovered struts was 10%, which was found in nonoverlapping SES segments. Differences in definitions and methodology might explain discrepancy between in vivo and postmortem studies (4,6). Of note, the previous histopathology study included a small number of highly selected postmortem samples, and the mean length of overlap in the experimental study was 9.8 mm, which does not correspond to common clinical practice as performed in the present study.
The increased rates of stent malapposition in the SES group demonstrated by IVUS are consistent with the OCT findings. Pooled data and retrospective analyses from previous studies have shown the early (<1 year) safety of overlapping PES and SES. Reconciliation of the present OCT findings with clinical data would require larger studies with serial OCT assessment and longer follow-up evaluations, because the vascular responses depicted by OCT might have a delayed clinical appearance. At this early stage of OCT clinical trials, no cause-effect relationship can be established between OCT findings and clinical outcomes. Nevertheless, the low frequency of uncovered or malapposed struts in DES platforms, even in the SES, matches the reported low rates of very late clinical stent thrombosis.
Neointimal thickness in overlapping DES
The present study used OCT, in addition to IVUS and angiography, to provide a sensitive measure of neointimal thickness at the strut level. Our findings suggested decreased efficacy at the site of overlap compared with nonoverlap DES segments. This vascular response was noted even in the SES group. The ZES and PES had an overall lower SIT compared with BMS, although NIH was similar in the overlap sites of these DES platforms compared with BMS. Previous pre-clinical experiments have suggested increased neointimal thickness in overlapping ZES compared with SES. Whether excessive vascular injury caused during deployment or by a permanent dual layer of struts—which are more rigid—are associated with these findings remains to be determined.
The findings of higher degrees of NIH and apparent higher degrees of strut malapposition at the site of overlap PES is somewhat novel and suggest that neointimal proliferation does not always translate into strut coverage or apposition. The OCT confirmed in vivo the heterogeneous character of the arterial wall response to coronary stents, with the co-existence of completely embedded struts with exuberant NIH and malapposed uncovered struts within a single cross-sectional image. The variable coverage of stent struts in DES might be related to the distinctive arterial reaction produced by each specific stent type and heterogeneous characteristics of the atherosclerotic plaque.
Drawbacks of this study include the single-center design, selected population, and reliance on surrogates. In addition, OCT cannot detect <20 µm tissue coverage or differentiate between very small amounts of thrombus or fibrin deposition or even inflammatory cellular response from the underlying NIH. New OCT imaging systems based on frequency-domain or Fourier-domain technology will enable coronary imaging without need for balloon occlusion. It is also possible that cross-sectional area measurements would be different between occlusive versus non-occlusive techniques, but it is unlikely that assessment of stent coverage and neointimal thickness would be affected.
The 6-month OCT revealed a similar impact of DES on stent coverage at overlapping and nonoverlapping sites but reduced efficacy at overlap. In addition, OCT showed a heterogeneous vascular response according to DES type, with higher rates of uncovered or malapposed struts in PES and SES compared with low rates observed in BMS and ZES.
The authors thank Giuseppe Sangiorgi, MD, Marco Valgimigli, MD, PhD, and Dominick J. Angiolillo, MD, PhD, for their helpful comments and suggestions.
This work was supported by Ospedali Riuniti di Bergamo and the University Hospitals and Case Western Reserve University, with unrestricted grant support from Medtronic Vascular and Boston Scientific Corporation. Dr. Guagliumi reports receiving consulting fees from Boston Scientific and Volcano and receiving grant support from LightLab, Medtronic Vascular, Boston Scientific, and Abbott Vascular. Dr. Biondi-Zoccai reports receiving consulting fees from Abbott Vascular, Biotronik, Boston Scientific, Cordis, Invatec, and Medtronic. Dr. Costa reports receiving consulting fees from LightLab, Medtronic, Scitech, Cordis, Boston Scientific, and Abbott Vascular.
- Abbreviations and Acronyms
- bare-metal stent(s)
- drug-eluting stent(s)
- intravascular ultrasound
- late stent thrombosis
- neointimal hyperplasia
- optical coherence tomography
- paclitaxel-eluting stent(s)
- sirolimus-eluting stent(s)
- strut-level intimal thickness
- zotarolimus-eluting stent(s)
- Received November 20, 2009.
- Revision received February 1, 2010.
- Accepted February 22, 2010.
- American College of Cardiology Foundation
- Finn A.V.,
- Joner M.,
- Nakazawa G.,
- et al.
- Cook S.,
- Wenaweser P.,
- Togni M.,
- et al.
- Joner M.,
- Finn A.V.,
- Farb A.,
- et al.
- Balakrishnan B.,
- Tzafriri A.R.,
- Seifert P.,
- et al.
- Finn A.V.,
- Kolodgie F.D.,
- Harnek J.,
- et al.
- Suzuki Y.,
- Ikeno F.,
- Koizumi T.,
- et al.
- Matsumoto D.,
- Shite J.,
- Shinke T.,
- et al.
- Mintz G.
- Cutlip D.E.,
- Windecher S.,
- Mehran R.,
- et al.
- Tsagalou E.,
- Chieffo A.,
- Iakovou J.,
- et al.
- Finn A.V.,
- Nakazawa G.,
- Joner M.,
- et al.