Author + information
- Received February 21, 2010
- Revision received June 29, 2010
- Accepted August 20, 2010
- Published online November 1, 2010.
- Michail I. Papafaklis, MD, PhD⁎,
- Christos V. Bourantas, MD, PhD§,
- Panagiotis E. Theodorakis, PhD∥,
- Christos S. Katsouras, MD⁎,†,
- Katerina K. Naka, MD, PhD⁎,†,
- Dimitrios I. Fotiadis, PhD⁎,‡ and
- Lampros K. Michalis, MD⁎,†,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Lampros K. Michalis. Department of Cardiology, Medical School, University of Ioannina, University Avenue, Ioannina GR 45110, Greece
Objectives We aimed to explore the relationship of neointimal thickness (NT) to shear stress (SS) after implantation of sirolimus-eluting stents (SES) and paclitaxel-eluting stents (PES) compared with bare-metal stents (BMS). We then tested the hypothesis that drug elution attenuates the SS effect.
Background Neointimal thickness after BMS implantation has been associated with SS; pertinent data for drug-eluting stents (DES) are limited.
Methods Three-dimensional coronary artery and stent reconstruction was performed in 30 patients at 6-month follow-up after SES (n = 10), PES (n = 10), or BMS (n = 10) implantation. Baseline SS at the stent surface was calculated using computational fluid dynamics and NT at follow-up was computed in 3-dimensional space.
Results Neointimal thickness was lower in DES versus BMS (0.03 ± 0.07 mm vs. 0.16 ± 0.08 mm, p < 0.001) and maximum NT was reduced in SES versus PES (0.33 ± 0.13 mm vs. 0.46 ± 0.13 mm, p = 0.025). In the total population, both SS (slope: −0.05 mm/Pa, p < 0.001) and DES (coefficient for DES vs. BMS: −0.17 mm, p = 0.003) were independent predictors of NT. Subgroup analysis demonstrated a significant negative relationship of NT to SS in PES (slope: −0.05 mm/Pa, p = 0.016) and BMS (slope: −0.05 mm/Pa, p = 0.001). Sirolimus elution significantly attenuated the effect of SS on NT (interaction coefficient for SES vs. BMS: 0.04 mm/Pa, p = 0.023), whereas the SS effect remained unchanged in PES (interaction coefficient for PES vs. BMS: 0.01 mm/Pa, p = 0.71).
Conclusions Neointimal thickness is significantly correlated (inversely) to SS in PES as in BMS. Sirolimus elution abrogates the SS effect on the neointimal response following stent implantation, whereas the SS effect is unchanged in PES.
Coronary artery stenting successfully reduces atherosclerotic stenoses providing a patent artery with preserved blood flow following intervention. In-stent restenosis due to neointimal growth constitutes a rather frequent sequela (incidence: 20% to 40%) after implantation of a bare-metal stent (BMS) and often necessitates repeat revascularization (1). Among standard clinical and procedural characteristics influencing restenosis occurrence, local hemodynamic factors, known to play a role in atherosclerosis (2), have also been implicated in the pathobiology of neointimal hyperplasia. Blood flow–derived shear stress (SS) on the in-stent endothelial surface has been demonstrated to influence neointimal thickness (NT), which is increased in stent regions with low SS (3).
Drug-eluting stents (DES) have currently dominated offering very low restenosis rates owing to their potent antiproliferative effect (4). The first-generation DES (i.e., sirolimus-eluting stents [SES] and paclitaxel-eluting stents [PES]) have been proven to reduce neointimal growth of smooth muscle cells (SMC) and endothelial cells leading to decreased neointimal hyperplasia or even tissue regression (5–7). The possible role of SS in predicting NT after DES implantation has been investigated by only 2 studies in small patient series with SES (8,9), and they reported contradictory results: 1 found an overall significant negative correlation of NT to SS in 6 patients, and the other reported no association between neointimal hyperplasia volume and average SS in 11 diabetics with SES. No such data are available for a patient group with PES.
In this report, we assess NT at follow-up after SES and PES implantation, and investigate its relationship with baseline in-stent SS in comparison with a BMS patient group to test the hypothesis that drug elution attenuates the local effect of SS. We used in vivo 3-dimensional coronary artery reconstruction and blood flow simulation for NT and SS computation.
Thirty patients, who underwent implantation of SES (Cypher, Cordis/Johnson & Johnson, Bridgewater, New Jersey) (n = 10), PES (Taxus, Boston Scientific, Natick, Massachusetts) (n = 10), or BMS (n = 10), were studied by angiography and intravascular ultrasound (IVUS) at 6-month follow-up when referred for cardiac catheterization at the University Hospital of Ioannina because of positive routine follow-up noninvasive stress testing. All patients had an otherwise uneventful follow-up period. Patients were eligible if they had undergone angiographically successful stent implantation (stent length ≥12 mm; located at least 5 mm away from major bifurcations) for de novo lesions in native coronary arteries and consented to follow-up catheterization using IVUS. The exclusion criteria were primary stenting in acute myocardial infarction (≤12 h after symptom onset), chronic total occlusion lesions, ejection fraction <50%, and serum creatinine >1.5 mg/dl. None of the studied patients had angiographically evident in-stent restenosis (>50% in-stent diameter stenosis) at follow-up.
All patients had received post-intervention dual antiplatelet therapy: aspirin 100 mg/day indefinitely and clopidogrel 75 mg/day for 1 month after BMS implantation and for 12 months after DES implantation. The study was approved by the local institutional ethical committee, and each patient provided informed consent.
3D reconstruction of coronary arteries and blood flow simulation
Three-dimensional (3D) reconstruction of the coronary arteries was performed using a validated method based on the fusion of biplane angiographic and IVUS data (details in the Online Appendix) (10,11). Briefly, the end-diastolic biplane angiographic images were used to reconstruct the catheter path in 3D space, whereas the lumen and stent (determined by the leading edge of the stent struts) borders were detected in the digitized end-diastolic IVUS images (Online Fig. 1S). The detected borders were placed onto the 3D catheter path to reconstruct the lumen at follow-up and the stent surface (Figs. 1A to 1C); the 3D stent reconstruction represented the approximate lumen at baseline just after stent implantation (12).
Blood flow simulation was performed using computational fluid dynamics techniques for generating the finite volume (hexahedral) mesh of the 3D reconstruction (details in Papafaklis et al. ) and solving the 3D Navier-Stokes equations (ICEM CFD 5 and CFX 10, ANSYS, Inc., Canonsburg, Pennsylvania). We modeled blood as a homogeneous and Newtonian fluid with a dynamic viscosity of 0.0035 Pa × s and a density of 1,050 kg/m3, while the arterial wall was considered rigid. Laminar flow was assumed and a patient-specific steady flow with a fully developed profile was specified at the inlet. Coronary blood flow for each patient was calculated by the number of the post-procedural angiogram frames required for the opacified material to pass from the inlet to the outlet of the arterial section being studied, the corresponding 3D lumen, and the cine frame rate (12.5 frames/s) (14). Zero pressure conditions were imposed at the outlet and the no-slip condition was applied at the wall. The detailed characteristics of the intracoronary flow field were obtained (Fig. 1D), and baseline SS at the stent surface (Fig. 1E) was computed as the product of the gradient of blood velocity at the arterial wall and viscosity.
The analysis was performed only for the stented arterial segments. Area and volume measurements were derived from the 3D reconstructed models; volume measurements derived from 3D models are more accurate than IVUS-based ones, which involve linear stacking of IVUS images and neglect curvature effects (15). Neointimal thickness at follow-up was calculated as the distance between the stent (baseline) and the lumen (follow-up) reconstructed surfaces in 3D space using an in-house developed algorithm implemented in Visual Fortran (Compaq Computer Corporation, Houston, Texas). Positive NT values corresponded to areas of neointimal hyperplasia, whereas negative NT denoted that the lumen border was outside the reconstructed stent surface and thus, tissue regression with lumen enlargement was present (Fig. 1F and Online Fig. 1S). Neointimal thickness was determined along the axial direction per 0.5 mm and around the vessel circumference per 10° (at 36 locations) for each cross section; SS was computed at the same locations. Cross sections corresponding to in-stent side branches (identified by IVUS) and their adjacent segments with a length equal to the diameter of the side branch, as well as a part (2.5 mm in length) of the segment at the entrance and exit of the stent, were excluded from the analysis to minimize their influence on the flow field and SS computations.
Categorical variables were compared between groups using the chi-square test or Fisher exact test. The unpaired t test or 1-way analysis of variance were used for comparing continuous variables (presented as the mean ± SD) between 2 or 3 groups, respectively. The correlation between NT at follow-up and baseline SS at the stent surface for each patient was investigated using linear regression analysis. A linear mixed-effects model (random intercept and slope) controlling for patient-to-patient variation was used to estimate the overall NT-SS relationship in the patient groups and the influence of drug (sirolimus and paclitaxel) elution on this relationship (using BMS as the reference group). The parameters entered in this model were SS, stent type, and the interaction between stent type and SS (stent type × SS). An adjusted (Bonferroni correction) alpha level of 0.025 (0.05 ÷ 2) was used to determine statistical significance for the 2 contrasts (SES × SS vs. BMS × SS and PES × SS vs. BMS × SS) testing the interaction, whereas the conventional alpha level of 0.05 was used for all other analyses; all tests were 2-tailed. The SPSS statistical software package (version 16.0 for Windows, SPSS, Inc., Chicago, Illinois) was used for the analysis.
The baseline clinical and procedural characteristics of the SES, PES, and BMS patient groups are presented in Table 1. Except for a difference regarding patient sex between SES and BMS patients, the groups had similar characteristics.
Neointimal thickness in SES, PES, and BMS
Quantitative measurements in the 3D models (Table 2) demonstrated significantly lower neointimal hyperplasia and higher tissue regression in the DES compared with the BMS group. The SES group, compared with the PES group, demonstrated even lower neointimal hyperplasia and higher tissue regression burden. Furthermore, mean NT was significantly lower in DES patients than in BMS patients (0.03 ± 0.07 mm vs. 0.16 ± 0.08 mm, p < 0.001). The difference in mean NT between SES (0.007 ± 0.04 mm) and PES (0.054 ± 0.08 mm) patients did not reach statistical significance (p = 0.123), but the average of maximum NT was significantly lower in SES (0.33 ± 0.13 mm vs. 0.46 ± 0.13 mm, p = 0.025). Low NT in DES also reflected the more frequently observed tissue regression (negative NT) in this group (44% of all measured in-stent locations, minimum NT: −1.09 mm; 49% for SES and 38% for PES) compared with the BMS control group (11% of all measured in-stent locations; minimum NT: −0.15 mm).
Relationship of NT at follow-up with baseline SS at the stent surface
Average baseline SS at the stent surface was similar between DES and BMS (2.02 ± 0.81 Pa vs. 1.74 ± 0.49 Pa, respectively, p = 0.33); the respective averages of minimum SS (0.51 ± 0.49 Pa vs. 0.45 ± 0.37 Pa, p = 0.75) and maximum SS (4.67 ± 2.04 Pa vs. 4.27 ± 2.19 Pa, p = 0.63) were also similar. Furthermore, there was no significant difference between SES and PES in average SS (1.98 ± 0.58 Pa vs. 2.05 ± 1.01 Pa, respectively, p = 0.85) and the averages of minimum SS (0.61 ± 0.57 Pa vs. 0.41 ± 0.43 Pa, p = 0.37) and maximum SS (4.17 ± 1.38 Pa vs. 5.17 ± 2.51 Pa, p = 0.29).
Linear regression analysis for each patient in the DES group (Figs. 2A and 2C) demonstrated a significant negative correlation of NT at follow-up to baseline in-stent SS in 12 patients (range of the correlation coefficient [r]: −0.67 to −0.07, p < 0.02). The respective relationship in the BMS group was significant in 8 patients (r range: −0.61 to −0.11, p < 0.001) (Figs. 2B and 2D). Group analysis (Table 3) resulted in a significant inverse correlation between NT and SS in both DES (slope: −0.03 mm/Pa, p = 0.004) and BMS patients (slope: −0.05 mm/Pa, p = 0.001). After pooling all data from both DES and BMS patients, DES were significantly associated with lower NT (DES vs. BMS: −0.17 mm, p = 0.003) and baseline in-stent SS was a significant predictor of NT at follow-up (slope: −0.05 mm/Pa, p < 0.001), whereas the relationship of NT to SS was not found to be significantly different between DES and BMS (Table 3).
Linear regression analysis for each patient also revealed differences between the SES and PES group (Figs. 2A and 2C). For only 3 of 10 SES patients, a significant negative correlation of NT to SS was found (r range: −0.31 to −0.15, p < 0.001), whereas 9 PES patients showed a significant negative correlation (r range: −0.67 to −0.07; p < 0.02). Subgroup analysis (Table 3) demonstrated a significant correlation of NT to SS only in the PES group (slope: −0.05 mm/Pa, p = 0.016).
Overall group analysis (Table 3) for prediction of NT at follow-up, including 3 groups and using BMS as the reference group, demonstrated an independent and significant effect of sirolimus elution on the relationship of NT at follow-up with baseline in-stent SS (interaction coefficient for SES vs. BMS: 0.04 mm/Pa, p = 0.023). Paclitaxel elution, when compared with BMS, was not found to significantly alter the relationship between NT and SS.
The present study explores the effect of baseline in-stent SS on NT following DES or BMS implantation in humans by use of realistic 3D coronary artery reconstruction and blood flow computational analysis. The main findings of this study are: 1) baseline SS at the stent surface is a significant predictor of NT at follow-up (inverse correlation) in PES and BMS; and 2) sirolimus elution significantly attenuates the effect of SS on NT, whereas the SS effect remains unchanged under paclitaxel elution. This is the first study that investigates the impact of in-stent SS on neointimal response following stent implantation in both SES and PES compared with BMS based on the largest up-to-date patient series.
NT: drug elution and the effect of SS
In-stent neointimal hyperplasia depends on complex mechanisms within the arterial wall involving mainly SMC, matrix formation, and endothelial cells. The interplay of growth factors released during vessel injury controls SMC migration and proliferation through intracellular positive (cyclin-dependent kinases) and negative (cyclin kinase inhibitors) regulators of cell cycle events (16). Smooth muscle cells migrating from the media leave their quiescent state and undergo phenotypic modulation switching from a contractile to a synthetic phenotype. Low SS, known to induce phenotypic changes in endothelial cells and create a proatherogenic environment (17), can also modulate SMC proliferation and migration either via endothelial cells or directly in cases of a damaged endothelium, such as those immediately after stenting (18), and thus, may amplify the neointimal response to vessel injury.
NT following BMS implantation in humans has been reported to be inversely correlated to SS (12,19), indicating in vivo that there is a higher probability for neointimal hyperplasia to occur and be more profound in low SS regions (13). Although, in our population, average SS after stenting was within the physiologic range (approximately 2 Pa), there was heterogeneity in the SS distribution at the stent surface due to the local geometry, and, thus, low SS areas were present. An overall significant correlation between NT at follow-up and baseline in-stent SS was found (p < 0.001), and SS was demonstrated as a significant predictor of NT independently of the lower NT in DES compared with BMS. Of note, DES subgroup analysis showed: 1) the NT-SS relationship was significant only in PES; and 2) sirolimus elution, but not paclitaxel elution, significantly attenuated the effect of SS on NT compared with BMS controls. Two previous studies have investigated the effect of SS on NT in SES. Gijsen et al. (8) reported that NT 6 months after SES implantation was significantly related to SS in 4 of the 6 patients studied, and tissue regression was attributed to high SS. The other available study investigated the correlation between neointimal hyperplasia volume and SS in a diabetic population 9 months after SES and found no association between the 2 variables (9). The present study demonstrated a significant inverse NT to SS relation in only 3 of the 10 SES patients with an overall nonsignificant correlation in the SES group, whereas this correlation was significant in 9 of the 10 PES patients. Although some differences in data analysis do not allow direct comparison of this study to the previous ones, it seems that the role of SS after SES implantation is limited due to an attenuation of the SS effect by sirolimus elution, as supported by our results.
The aforementioned differences in the effect of sirolimus versus paclitaxel elution on the NT-SS relationship reflect the distinct cellular mechanisms of action of the 2 drugs and the association of these mechanisms with the SS signaling cascades. Sirolimus binds to its cellular target (mammalian target of rapamycin) leading to a decrease of the positive (blocks the p70S6 kinase pathway of the cyclin-dependent kinases) and an increase of the negative (cyclin kinase inhibitors p27 and p21) regulators of the cell cycle (20); the final effect is the arrest of the cell cycle at the G0 phase inhibiting cell migration and proliferation. The sirolimus-sensitive p70S6 kinase pathway has also been reported to be a major signaling pathway through which low SS mediates its effects of cell cycle up-regulation (21), deoxyribonucleic acid synthesis and consequently, intimal growth. Further, high SS has been reported to induce cell cycle and growth arrest by up-regulating cyclin kinase inhibitor p21 (22,23). Paclitaxel is a nonspecific microtubule-stabilizing agent resulting in reduced cell division and mobility (16). However, paclitaxel-induced stabilization of the cytoskeleton has been found to sensitize endothelial cells for SS effects (24). In addition, sirolimus and paclitaxel have a dissimilar impact on SMC phenotype in vitro; SMC under sirolimus treatment adopt a differentiated contractile phenotype, whereas paclitaxel treatment induces a dedifferentiated synthetic phenotype (25). Different SMC phenotypes have also been observed in a recent histopathological investigation of in-stent restenotic tissues according to stent type in humans. A synthetic phenotype was more frequent in both PES and BMS, whereas a contractile-intermediate phenotype was mainly found in SES (26). Low SS has also been reported to promote SMC and endothelial cell dedifferentiation (27).
Neointimal growth in SES versus PES
Localized delivery of immunosuppressive and antineoplastic agents (e.g., sirolimus and paclitaxel, respectively) using DES has considerably changed the picture in terms of both in-stent neointimal growth and clinical benefit for patients undergoing percutaneous coronary interventions (28). In our patients, mean NT following DES implantation was significantly and clearly lower compared with BMS reflecting both the reduced neointimal growth and the higher frequency of tissue regression. Differences were also evident between SES and PES; neointimal hyperplasia burden was lower in SES according to all area and volume indexes, and the average of maximum NT was significantly reduced compared with PES (p = 0.025), demonstrating a higher inhibition of neointimal hyperplasia by SES.
The differential degree of neointimal growth inhibition by SES versus PES has been recently demonstrated in large patient registries and randomized trials (29). Neointimal hyperplasia estimated by angiography or IVUS has been reported to be significantly reduced in SES versus PES (30,31). In terms of clinical benefit, significantly lower rates of revascularization have been reported after SES, when compared with PES, implantation both in the total patient population and in high-risk groups (e.g., diabetics) (32,33). Furthermore, DES type (PES vs. SES) has been found to be among the predictive factors of restenosis (34). These differences in clinical studies may be explained in part by the aforementioned distinct biological properties of the eluted drugs and their association with the SS effect on the neointimal response.
Despite the potent antiproliferative effect of DES, neointimal hyperplasia or even in-stent restenosis may still occur especially when DES are used in complex clinical and anatomic settings (35). Lesions in arterial locations with an adverse hemodynamic environment of disturbed flow and low SS (e.g., bifurcations and highly curved segments) are at high risk of restenosis after stenting. Although both SES and PES, when compared with BMS, primarily reduce neointimal growth, sirolimus presumably exerts an additional, second-order, favorable action by abrogating the low SS effect, which could potentially amplify the neointimal response in certain regions. Therefore, the treatment of lesions located in areas with a pro-restenotic hemodynamic milieu could possibly justify the selection of a particular DES type by interventionists for reducing neointimal hyperplasia and improving clinical outcomes. However, large patient studies are needed to confirm our findings and address the efficacy of this strategy.
The number of patients is rather small in each patient group, even if it is the largest to date for such an investigation. The large number of in-stent locations (on average 792 locations per stented segment) axially and circumferentially, where NT and SS were calculated and interrelated, also strengthens the validity of our results.
For the 3D stent reconstructions, only follow-up data were used; this approach increases the accuracy of calculating the difference between the lumen and stent surface in 3D space (i.e., NT), but does not account for any changes in 3D stent geometry over the follow-up period and any suboptimal stent expansion, which may have caused incomplete stent apposition at baseline. However, stent changes over a 6-month follow-up period have been reported to be minor (12), and these limitations are not likely to have significantly influenced our results.
Furthermore, the assumptions (Newtonian viscosity and rigid arterial walls) in the blood flow simulation are acknowledged, but are not considered to have a significant impact on our results, because stented regions are quite rigid and the relative SS distribution is more important rather than its absolute value.
The use of 3D coronary artery reconstruction coupled with computational fluid dynamics for calculating in-stent baseline SS and NT 6 months after DES (SES and PES) and BMS implantation in 30 patients demonstrated an overall inverse correlation of NT to SS. Both DES and SS were found to be independent predictors of NT. Compared with BMS, the relationship between NT and SS is significantly reduced by sirolimus elution but remains unchanged under paclitaxel elution. Common pathways through which both sirolimus and SS modulate cell migration and proliferation explain why sirolimus abrogates the SS effect, whereas the pathobiology of neointimal formation after PES implantation resembles more that of BMS and responds similarly to SS. These differences between the 2 DES types may account for the higher degree of neointimal growth inhibition by SES compared with PES observed in large patient trials.
For more information about the method for 3D coronary artery reconstruction, please see the online version of this article.
The effect of shear stress on neointimal response following sirolimus- and paclitaxel-eluting stent implantation compared to bare metal stents in humans
The authors have reported that they have no relationships to disclose.
- Abbreviations and Acronyms
- bare-metal stent(s)
- drug-eluting stent(s)
- intravascular ultrasound
- neointimal thickness
- paclitaxel-eluting stent(s)
- sirolimus-eluting stent(s)
- smooth muscle cell(s)
- shear stress
- Received February 21, 2010.
- Revision received June 29, 2010.
- Accepted August 20, 2010.
- American College of Cardiology Foundation
- Kimura T.,
- Abe K.,
- Shizuta S.,
- et al.
- Chatzizisis Y.S.,
- Coskun A.U.,
- Jonas M.,
- Edelman E.R.,
- Feldman C.L.,
- Stone P.H.
- Roiron C.,
- Sanchez P.,
- Bouzamondo A.,
- Lechat P.,
- Montalescot G.
- Hong M.K.,
- Mintz G.S.,
- Lee C.W.,
- et al.
- Sousa J.E.,
- Costa M.A.,
- Abizaid A.,
- et al.
- Hong M.K.,
- Mintz G.S.,
- Lee C.W.,
- et al.
- Wentzel J.J.,
- Krams R.,
- Schuurbiers J.C.,
- et al.
- Stone P.H.,
- Coskun A.U.,
- Kinlay S.,
- et al.
- Ueba H.,
- Kawakami M.,
- Yaginuma T.
- Sindermann J.R.,
- Verin V.,
- Hopewell J.W.,
- Rodemann H.P.,
- Hendry J.H.
- Akimoto S.,
- Mitsumata M.,
- Sasaguri T.,
- Yoshida Y.
- Lin K.,
- Hsu P.P.,
- Chen B.P.,
- et al.
- Hastings N.E.,
- Simmers M.B.,
- McDonald O.G.,
- Wamhoff B.R.,
- Blackman B.R.
- Sousa J.E.,
- Serruys P.W.,
- Costa M.A.
- Stefanadis C.,
- Toutouzas K.
- Lemos P.A.,
- Moulin B.,
- Perin M.A.,
- et al.,
- PAINT Trial Investigators
- Petronio A.S.,
- De Carlo M.,
- Branchitta G.,
- et al.
- Lee S.W.,
- Park S.W.,
- Kim Y.H.,
- et al.
- Kastrati A.,
- Dibra A.,
- Mehilli J.,
- et al.
- Alfonso F.