- Email: [email protected]
Early vascular response to overlapped paclitaxel-eluting stents in swine coronary arteries
Cardiovascular Revascularization Medicine 8 (2007) 251 – 258
Original Article
Early vascular response to overlapped paclitaxel-eluting stents in swine coronary arteries☆ Paul S. Seifert, Barbara A. Huibregtse⁎, Jason Polovick, Bradley Poff Boston Scientific Corporation, Natick, MA 01752, USA Received 8 May 2007; received in revised form 13 August 2007; accepted 13 August 2007
Abstract
Background: The early response to the TAXUS Express2 paclitaxel-eluting stent (PES) system was compared to the response to the Express2 bare metal stent (BMS) system in porcine arteries. Methods: Swine coronary arteries were implanted with overlapping PES or BMS and examined at 1, 2, 4, 10, and 20 days postimplantation using scanning electron microscopy or light microscopy. Results: Vascular healing in terms of strut coverage, reendothelialization, degree of inflammation, and absence of thrombus was equivalent in both groups from 1 to 20 days. Interstrut member spaces were unaffected by stent deployment and remained covered with endothelium from Day 1. In both groups at 2 days, small patches of endothelial cells covered approximately 5–10% of the stent surface. At 4 days, endothelial cell coverage progressed to nearly 50% in both groups. After 10 days, endothelial cell strut coverage was nearly complete (N90%), with regions of incomplete coverage located primarily in strut overlap regions in both groups. BMS exhibited a fibrocellular neointima and no parastrut fibrin, whereas PES exhibited a developing but immature fibrocellular neointima and prominent parastrut fibrin. By Day 20, an endothelialized neointima was present in both groups, with comparable coverage of proximal and distal stented regions. The neointima of PES was more fibrocellular and parastrut fibrin was still comparable to that at 10 days. Conclusion: Early vascular response was comparable for both PES and BMS, with similar rates of reendothelialization, limited inflammatory response, and absence of thrombus, but differed parastrut fibrin clearance and neointimal maturation rate. © 2007 Elsevier Inc. All rights reserved.
Keywords:
Drug-eluting stent; Paclitaxel; Coronary artery disease; Swine
1. Introduction Paclitaxel-eluting stents (PES; TAXUS Paclitaxel-Eluting Express2) have been clinically demonstrated to be effective in reducing restenosis in humans [1–3]. The swine coronary artery model is an accepted and commonly used preclinical test system for evaluating coronary artery stenting, with data ☆
This study was sponsored and supported by Boston Scientific Corporation. ⁎ Corresponding author. Boston Scientific Corporation, One Boston Scientific Place, Natick, MA 01760, USA. Tel.: +1 508 652 5535; fax: +1 508 647 2241. E-mail address: [email protected] (B.A. Huibregtse). 1553-8389/07/$ – see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.carrev.2007.08.002
being generally reported at 28 and 90 days [4–10]. Early vascular response to the polymer-based TAXUS Express2 PES (Boston Scientific Corporation, Natick, MA) in the noninjured swine coronary artery model has not yet been reported. The current study of the early postdeployment period assessed the short-term vascular healing response in PES versus bare metal stents (BMS). By comparing the vascular response of PES to that of Express2 BMS, we sought to determine whether primary differences occur in the first 20 days of tissue response. The normal process of stent healing initially involves the deposition of plasma protein and/or a thrombotic coating of fibrin adherent to or associated with struts and containing variable amounts of red blood cells (RBCs), platelets, and
252
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
leukocytes [11]. This coating material is common to all intravascular biomaterials and presents a substrate that promotes endothelial cell coverage. In this study, we demonstrate that strut-associated fibrin/thrombus persists longer in PES than in BMS and does not appear to be prothrombogenic or to represent an embolic risk. In addition, neointimal hyperplasia is suppressed in PES relative to BMS at 10 and 20 days. There were no other prominent differences in vascular response between these two stent types. Whether the slower fibrin resolution of PES is responsible for or connected to the reduction in restenosis rates observed clinically remains to be established. The study was carried out under Good Laboratory Practice (GLP) regulations for submission to health-regulatory bodies for device approval. 2. Materials and methods 2.1. Device description The polymer-based paclitaxel-eluting TAXUS Express2 and bare metal Express2 coronary stent delivery systems were used in the present study. The TAXUS Express2 stent system consisted of a slow-release formulation of paclitaxel (dose density, 1 μg/mm2) and a Translute-coated 316L stainless-steel balloon-expandable stent mounted on a lowprofile monorail delivery system. The Express2 stent system consisted of a 316L stainless-steel balloon-expandable stent mounted on an over-the-wire delivery system. Both PES and BMS were 2.75, 3.0, or 3.5 mm in diameter and 8.0 mm long. 2.2. Animals Forty nulliparous nonpregnant female domestic pigs (Sus scrofa) weighing 35 to 50 kg were assigned to five study groups. The animals were treated with aspirin (325 mg po) and clopidogrel (75 mg po) for 3 days prior to stent implantation, and with Procardia XL (30 mg po) 1 day prior to stent implantation. On the day of stent implantation (Day 0), each animal was appropriately anesthetized and implanted with one pair of the same type of overlapping stents (one pair each of PES or BMS) in the right coronary artery, left circumflex coronary artery, and left anterior descending coronary artery, for a total of three overlapping stent pairs per animal at an expansion ratio of 1.1:1. The animals received aspirin (81 mg po) and clopidogrel (75 mg po) daily after stent implantation until the end of the study. Six of 40 animals received two stent pairs because vascular anatomy prevented placement in all three vessels. Vessels in which stents were not overlapped were excluded from statistical analysis. To confirm the consistent delivery expansion of each stent and to evaluate stenosis by angiography, quantitative coronary angiography was performed at the time points listed in Table 1.
Table 1 Study design Number of BMS
Number of PES
Time point
Histology
SEM
Histology
SEM
20–28 h (1 day) 48±6 h (2 days) 96±6 h (4 days) 10±1 day (10 days) 20±1 day (20 days) Total stents
0 20 20 20 a 21 b 81
6 6 6 6 6 30
0 20 22 22 22 86
6 6 6 6 6 30
a b
Two nonoverlapping BMS were excluded from the final analysis. One single stent was implanted and excluded from the final analysis.
Peripheral blood was collected for analyses of standard hematology and serum chemistry parameters at two time points: prior to implantation and prior to termination. All animal surgical procedures, euthanasia, necropsy, and stent and tissue collection were performed by Charles River Laboratories (Osceola, WI). The test facility was accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International. The study was conducted in accordance with 21 CFR Part 58 (Food and Drug Administration), ENV/MC/CHEM (98) 17 (Organization for Economic Cooperation and Development [OECD]), or CLIA88. 2.3. Tissue collection Animals were euthanized at study termination. Comprehensive necropsy was performed on each animal. Any abnormalities were appropriately recorded. After euthanasia, hearts were harvested and pressure perfusion fixed with 10% buffered formalin. After fixation, the hearts were radiographed to map stent positions prior to excision. Stented vessels were dissected from the myocardium, allowing adequate proximal and distal reference vessel lengths in addition to the stented segment, and radiographed using high-resolution Faxitron to assess for appropriate overlap and stent strut fractures. The dissected vessels, along with the remaining heart tissues, routine tissue samples from the liver and kidneys, and any abnormal-appearing tissues in the chest or abdomen were assessed by the study pathologist. 2.4. Histopathology Stented vessels were prepared for light microscopy (LM) and scanning electron microscopy (SEM) analysis. Sections were obtained from the proximal unstented vessel, proximal nonoverlapped portion of the stented vessel, midoverlapped portion of the stented vessel, distal nonoverlapped portion of the stented vessel, and distal unstented vessel. Myocardial tissues adjacent and distal to the stent and any lesions were examined as well using standard histology techniques. All histological
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
253
Fig. 1. Scanning electron micrographs of PES (left) and BMS (right) specimens from 1 to 10 days. (A and B) Original magnification ×25; Day 1 examples demonstrate minimal thrombotic material associated with the struts of both stent types. (C and D) Original magnification ×25 and ×300; Day 2 examples demonstrate small patches of endothelial cells on the strut surfaces, as well as adherent leukocytes. Note that thrombotic material is not greater than Day 1 specimens and that overlapped stent regions (original magnification ×25; C) are also not more thrombogenic. (E and F) Original magnification ×25; Day 4 examples of endothelium cover significant areas of the strut surface, as well as adherent multinucleated giant cells. (G and H) Original magnification ×25; Day 10 examples demonstrate that most of the struts are covered with endothelialized neointima.
254
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
Fig. 1 (continued).
specimens were prepared for LM and SEM using standard procedures.
3. Results 3.1. Mortality, gross and clinical pathology, and stent integrity Forty animals were implanted with a total of 227 stents (108 overlapping BMS, 116 overlapping PES, and 3 individual or nonoverlapping BMS, which were excluded from the final analysis) in this study. There were no unscheduled mortalities, no abnormalities in hematology and serum chemistry values, and no abnormalities at necropsy. No strut fractures were detected using highresolution radiography. 3.2. One-day results No gross thrombi were observed on any of the three PES or three BMS pairs examined by SEM (Fig. 1A and B). There was a thin coating (average, b50 μm) of thrombus-like
material (fibrin and platelets) covering variable proportions of the stent struts. Leukocytes were adherent to the arterial endothelium and to the stent struts of overlapping PES and BMS, and appeared to favor the lateral strut surfaces as opposed to the upper flow surface. Most of the luminal arterial surface was covered with endothelium (i.e., areas in the open cells of the stent). The endothelium was not observed on stent struts. There were no differences between PES and BMS at the Day 1 time point, indicating that neither stent surface was thrombogenic nor proinflammatory. The deposition of thrombotic material occurred in both stent types primarily at bends in the strut configuration. In addition, the overlapped stent regions did not exhibit noticeably more thrombotic material than the nonoverlapped regions. No specimens were processed for histology in the 1-day cohort. 3.3. Two-day results No gross luminal thrombi were observed on any of the three PES or three BMS pairs examined by SEM (Fig. 1C and D) or LM. In addition, the thin thrombotic coating covering variable proportions of the struts was not greater than that observed on Day 1 and was not markedly different
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
255
Fig. 2. PES (left) and BMS (right) specimens from 2 to 20 days. (A and B) Day 2: a thin thrombotic material equivalently covers the struts of both PES and BMS at 2 days. (C and D) Day 4: endothelium usually covers struts associated with thrombotic material. (E and F) Day 10: neointimal formation in BMS specimens (F) and residual amounts of parastrut fibrin are observed. There appeared to be more neointimal parastrut fibrin in the 10-day PES specimens compared to the 4-day PES specimens. (G and H) Day 20: BMS groups exhibited a more mature and well-developed fibrocellular neointima than PES; parastrut fibrin was not commonly observed with BMS specimens, whereas PES specimens continued to exhibit parastrut fibrin and RBCs in the neointima.
256
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
Fig. 2 (continued).
between overlapping PES and BMS (Fig. 2A and B). The overlapped stent regions did not exhibit noticeably more thrombotic material than the nonoverlapped regions. Leukocytes were adherent to both struts and endothelium in PES and BMS, and were not at substantially different levels compared to Day 1. Endothelial cells were observed mostly as multicellular patches that were randomly distributed on the stent struts of both PES and BMS and were estimated to cover approximately 5–10% of the stent strut area (Fig. 1C and D). Occasional multinucleated giant cells (fused macrophages) were observed on the surface of struts in both stent groups (Table 2). 3.4. Four-day results No gross luminal thrombus was observed on any of the three PES or three BMS pairs examined by SEM and LM (Fig. 1E and F). In addition, the overlapped stent regions did not exhibit noticeably more thrombotic material than the nonoverlapped regions. Endothelium usually covered struts when they had thrombotic material associated with them and was less common when thrombus was lacking on struts (Fig. 2C and D). Leukocytes were present in the thrombotic material and were adherent to the endothelium.
Multinucleated giant cells (MNGCs) were observed adhering to the struts of both PES and BMS in areas devoid of endothelium. The peak time point for strutadherent MNGC was 4 days (Table 2). In many instances, they accounted for almost half of the strut coverage at this time point and had the appearance of covering strut areas devoid of the endothelium. As endothelial cell coverage progressed over time, there was diminution in MNGC presence (Table 2). 3.5. Ten-day results No luminal thrombus was observed on any of the three PES or three BMS pairs examined by SEM or LM (Fig. 1G and H). Both PES and BMS groups exhibited between 90% and 100% of the luminal surface covered by endothelium (Table 2). In areas where endothelial cell coverage was not yet complete, MNGCs were present, covering the struts. By 10 days, the surface of both stent types was largely covered by the endothelium; given the degree of endothelial coverage, observations of MNGCs were substantially reduced in numbers compared to the 4-day time point. Between 4 and 10 days, there was neointimal formation in BMS specimens and residual amounts of parastrut fibrin (Fig. 2E and F). The neointima in the overlapped stent regions was mildly less
Table 2 MNGC count on strut surfaces and endothelial coverage as a function of time BMS
PES
Day
MNGC count [mean±S.D. (n)]
% Samples with N90% endothelialization
MNGC count [mean±S.D. (n)]
% Samples with N90% endothelialization
2 4 10 20
0.7±1.4 (30) 5.5±4.4 (30) 2.5±3.1 (36) 0.1±0.5 (33)
0 52 97 100
0.2±0.5 (30) 5.8±5.5 (33) 2.8±4.9 (30) 0.9±2.2 (33)
0 47 82 100
n=sample size; S.D.=standard deviation.
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
developed than the nonoverlapped regions. PES specimens exhibited a neointima that was primarily composed of parastrut fibrin, RBCs, and mild amounts of leukocytes with a thin fibrocellular cap at the luminal aspect. The interstrut regions exhibited more developed fibrocellular tissues. Neovessels were sometimes observed in the neointima of both PES and BMS specimens. There appeared to be more neointimal parastrut fibrin in the 10-day PES specimens compared to the 4-day PES specimens. Subendothelial RBCs were observed in PES specimens but were not common in BMS specimens. The majority of struts were covered with a neointima in specimens from both groups. Leukocytes were present in the thrombotic material and were adherent to the endothelium but did not reach extensive levels in either group. As a marker of inflammation, leukocytes were not observed in large numbers, indicating that BMS does not provoke an inflammatory response beyond that expected following implantation of a device. Eosinophils were not observed in either group, suggesting the absence of a hypersensitivity reaction. 3.6. Twenty-day results Both PES and BMS groups exhibited a fibrocellular neointima with an endothelial covering (Fig. 2G and H). Parastrut fibrin was not commonly observed with BMS specimens, whereas PES specimens continued to exhibit parastrut fibrin and RBCs in the neointima, suggesting delayed clearance of the material. The neointima of PES specimens was more fibrocellular than the 10-day samples. The overlapped stent region exhibited more parastrut fibrin and less fibrocellular tissue than the proximal or distal nonoverlapped regions. No luminal thrombi were observed in either stent group. Occasional macrophages, often vacuolated, were present at the periphery of the parastrut fibrin in PES specimens but were not observed in BMS specimens. 4. Discussion This study sought to determine what histological differences, if any, exist between PES and BMS in the first 20 days of stent residence in swine coronary arteries. From previous work, it has been known that parastrut fibrin is present in PES but is uncommon in BMS at 30 days [12] and 90 days (unpublished data). We report here that both PES and BMS initially have equivalent amounts of fibrin or thrombotic material adherent to the stents at 1, 2, and 4 days. The deposition of thrombotic material appeared to be the result of turbulent flow following stent implantation, as it occurred in both stent types primarily at bends in the strut configuration. In addition, stents from both groups became covered with endothelium at approximately the same rate, and neither stent type provoked overt inflammatory cell presence or generated gross luminal thrombi. The first noticeable difference between the two stent types occurred at
257
10 days, when it was observed that BMS had low amounts of parastrut fibrin in a fibrocellular neointima, whereas PES exhibited persistent parastrut fibrin enveloped within a mostly immature fibrocellular neointima. This implies that fibrin resolution was more active between 4 and 10 days with BMS and probably occurred via a fibrinolytic pathway since a specific leukocyte infiltrate was not observed. The persistence of fibrin in PES and the appearance of foamy macrophages confined to the vicinity of fibrin suggest that paclitaxel-coated stents may inhibit such fibrinolysis and that fibrin is cleared by cell-mediated phagocytosis. Thus, it was demonstrated that the parastrut fibrin observed in PES at 30 days and beyond is present initially for both PES and BMS, but in the case of PES is removed more slowly through a different mechanism (phagocytosis) compared to quicker removal through a fibrinolytic pathway in the case of BMS in swine. The results demonstrate that PES do not inhibit endothelial cell proliferation or migration. Both stent types became covered with endothelium at roughly similar rates, and there were no overt differences in cell morphology. In addition and most importantly, in considering endothelial cell functionality, no luminal thrombi were observed at any time point in either stent group, nor was there evidence of embolic myocardial ischemia or infarction. As this observation held true for the period before the stents became endothelialized, it further demonstrates that coated stents are not more thrombogenic than BMS in this model. These results suggest that the low levels of late stent thrombosis observed in clinical studies may not be due to the stent itself and may reflect idiosyncratic reactions to drugs or polymers in a small percentage of patients with underlying atherosclerotic disease, rather than systematic effects on the presence or condition of the endothelium in general. Neutrophils and monocytes were associated with both stent types in the first few days, as would be expected as a response to acute injury. As a marker of inflammation, they were minimal in number and were not increased in PES compared to BMS, indicating that PES do not provoke an undesirable inflammatory response. Eosinophils were not observed, thus suggesting the absence of a hypersensitivity reaction. Macrophages were observed at the 20-day time point in PES specimens in association with parastrut fibrin and were usually foamy in appearance, indicating that they actively participated in the clearance of thrombotic materials, platelets, and RBCs. The early evaluation points in this study preclude an exact comparison to previously reported evaluations of drug-eluting stents (DES) in animal models [4–10,13] because the last evaluation in this study (20 days) occurs prior to the first evaluation in other studies (30 days). However, some similarities in healing were demonstrated between models when procedural differences were taken into consideration. The average stent-to-artery ratio in this study (1.03–1.25) was less than that reported for
258
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
overlapped stents in a rabbit model (1.13–2.45) [13] but was comparable to that reported in a single-stent study of sirolimus-eluting stents (SES) in porcine coronary arteries (1.16–1.29) [6]. In both the rabbit model [13] and this study, there was no significant difference in endothelialization between BMS and PES. However, in contrast to the rabbit model in which only approximately 66–80% of the lumen was endothelialized by 28 days for either BMS or PES [13], endothelialization in the porcine model was more rapid and largely complete in both PES and BMS (including the overlap section) by 20 days. Carter et al. [6] reported equivalent endothelialization between SES and BMS at 30 days in the porcine model, with a similar stent-to-artery ratio as used in this study. The differences in endothelialization rates between the rabbit and pig models are likely due to both species differences and procedural differences in stent-to-artery ratio; however, none of the models demonstrated a significant difference in endothelialization between DES and BMS. While fibrin deposition was similar in PES and BMS in this study at 2 and 4 days, there was greater deposition in PES at 10 and 20 days compared to BMS. Using a different scoring system in a model with greater overstretch, Finn et al. [13] also demonstrated a significantly greater fibrin deposition at 28 days in the rabbit model (as measured by struts with fibrin) in overlapping PES compared to BMS (P=.001 for PES vs. BMS). However, there was no difference in fibrin deposition between PES and BMS in the rabbit model at 90 days [13]. Thus, while both models demonstrate greater fibrin deposition in the overlapping PES compared to BMS at early time points, Finn et al.'s data suggest that this difference does not persist until 90 days. In summary, this study compared the histological response to overlapping PES and BMS in the first 20 days following implantation in a porcine coronary artery model. The absence of atherosclerotic disease and the young age of the animals limit the extrapolation of data from this healthy porcine model to human coronary artery disease patients; however, it serves as a method for the comparison of stent placement in similar conditions. Given these limitations, some observations were consistent throughout the study and were similar to results reported in other animal models. No difference in endothelialization was observed between PES and BMS. Results demonstrate that the only notable difference between PES and BMS is the presence of parastrut fibrin in PES, suggesting that paclitaxel may affect the pathway for fibrin removal. Whether this pathway also contributes to a slowdown of the development of the neointima requires further investigation, but it was observed in this animal model that PES delays neointimal formation, in accordance with the reduction of restenosis in the clinical setting.
Acknowledgments The authors would like to thank Saran Chhabra and Suzanne Wile for help in writing this manuscript and Michael Bravo for study support (all full-time employees of Boston Scientific Corporation). References [1] Grube E, Silber S, Hauptmann KE, Mueller R, Buellesfeld L, Gerckens U, Russell ME. TAXUS I: six- and twelve-month results from a randomized, double-blind trial on a slow-release paclitaxel-eluting stent for de novo coronary lesions. Circulation 2003;107(1):38–42. [2] Stone GW, Ellis SG, Cox DA, Hermiller J, O'Shaughnessy C, Mann JT, Turco M, Caputo R, Bergin P, Greenberg J, Popma JJ, Russell ME. A polymer-based, paclitaxel-eluting stent in patients with coronary artery disease. N Engl J Med 2004;350(3):221–31. [3] Stone GW, Ellis SG, Cannon L, Mann JT, Greenberg JD, Spriggs D, O'Shaughnessy CD, DeMaio S, Hall P, Popma JJ, Koglin J, Russell ME. Comparison of a polymer-based paclitaxel-eluting stent with a bare metal stent in patients with complex coronary artery disease: a randomized controlled trial. JAMA 2005;294(10):1215–23. [4] Lowe HC, Schwartz RS, Mac Neill BD, Jang IK, Hayase M, Rogers C, Oesterle SN. The porcine coronary model of in-stent restenosis: current status in the era of drug-eluting stents. Catheter Cardiovasc Interv 2003;60(4):515–23. [5] Schwartz RS, Edelman ER, Carter A, Chronos N, Rogers C, Robinson KA, Waksman R, Weinberger J, Wilensky RL, Jensen DN, Zuckerman BD, Virmani R. Drug-eluting stents in preclinical studies: recommended evaluation from a consensus group. Circulation 2002;106(14): 1867–73. [6] Carter AJ, Aggarwal M, Kopia GA, Tio F, Tsao PS, Kolata R, Yeung AC, Llanos G, Dooley J, Falotico R. Long-term effects of polymerbased, slow-release, sirolimus-eluting stents in a porcine coronary model. Cardiovasc Res 2004;63(4):617–24. [7] Garcia-Touchard A, Burke SE, Toner JL, Cromack K, Schwartz RS. Zotarolimus-eluting stents reduce experimental coronary artery neointimal hyperplasia after 4 weeks. Eur Heart J 2006;27(8):988–93. [8] Heldman AW, Cheng L, Jenkins GM, Heller PF, Kim DW, Ware M, Nater C, Hruban RH, Rezai B, Abella BS, Bunge KE, Kinsella JL, Sollott SJ, Lakatta EG, Brinker JA, Hunter WL, Froehlich JP. Paclitaxel stent coating inhibits neointimal hyperplasia at 4 weeks in a porcine model of coronary restenosis. Circulation 2001;103(18): 2289–95. [9] van der Giessen WJ, Lincoff AM, Schwartz RS, van Beusekom HM, Serruys PW, Holmes DR, Ellis SG, Topol EJ. Marked inflammatory sequelae to implantation of biodegradable and nonbiodegradable polymers in porcine coronary arteries. Circulation 1996;94(7):1690–7. [10] Schwartz RS, Chronos NA, Virmani R. Preclinical restenosis models and drug-eluting stents: still important, still much to learn. J Am Coll Cardiol 2004;44(7):1373–85. [11] Virmani R, Farb A. Pathology of in-stent restenosis. Curr Opin Lipidol 1999;10(6):499–506. [12] Balakrishnan B, Tzafriri AR, Seifert P, Groothuis A, Rogers C, Edelman ER. Strut position, blood flow, and drug deposition: implications for single and overlapping drug-eluting stents. Circulation 2005;111(22):2958–65. [13] Finn AV, Kolodgie FD, Harnek J, Guerrero LJ, Acampado E, Tefera K, Skorija K, Weber DK, Gold HK, Virmani R. Differential response of delayed healing and persistent inflammation at sites of overlapping sirolimus- or paclitaxel-eluting stents. Circulation 2005;112(2): 270–8.
Original Article
Early vascular response to overlapped paclitaxel-eluting stents in swine coronary arteries☆ Paul S. Seifert, Barbara A. Huibregtse⁎, Jason Polovick, Bradley Poff Boston Scientific Corporation, Natick, MA 01752, USA Received 8 May 2007; received in revised form 13 August 2007; accepted 13 August 2007
Abstract
Background: The early response to the TAXUS Express2 paclitaxel-eluting stent (PES) system was compared to the response to the Express2 bare metal stent (BMS) system in porcine arteries. Methods: Swine coronary arteries were implanted with overlapping PES or BMS and examined at 1, 2, 4, 10, and 20 days postimplantation using scanning electron microscopy or light microscopy. Results: Vascular healing in terms of strut coverage, reendothelialization, degree of inflammation, and absence of thrombus was equivalent in both groups from 1 to 20 days. Interstrut member spaces were unaffected by stent deployment and remained covered with endothelium from Day 1. In both groups at 2 days, small patches of endothelial cells covered approximately 5–10% of the stent surface. At 4 days, endothelial cell coverage progressed to nearly 50% in both groups. After 10 days, endothelial cell strut coverage was nearly complete (N90%), with regions of incomplete coverage located primarily in strut overlap regions in both groups. BMS exhibited a fibrocellular neointima and no parastrut fibrin, whereas PES exhibited a developing but immature fibrocellular neointima and prominent parastrut fibrin. By Day 20, an endothelialized neointima was present in both groups, with comparable coverage of proximal and distal stented regions. The neointima of PES was more fibrocellular and parastrut fibrin was still comparable to that at 10 days. Conclusion: Early vascular response was comparable for both PES and BMS, with similar rates of reendothelialization, limited inflammatory response, and absence of thrombus, but differed parastrut fibrin clearance and neointimal maturation rate. © 2007 Elsevier Inc. All rights reserved.
Keywords:
Drug-eluting stent; Paclitaxel; Coronary artery disease; Swine
1. Introduction Paclitaxel-eluting stents (PES; TAXUS Paclitaxel-Eluting Express2) have been clinically demonstrated to be effective in reducing restenosis in humans [1–3]. The swine coronary artery model is an accepted and commonly used preclinical test system for evaluating coronary artery stenting, with data ☆
This study was sponsored and supported by Boston Scientific Corporation. ⁎ Corresponding author. Boston Scientific Corporation, One Boston Scientific Place, Natick, MA 01760, USA. Tel.: +1 508 652 5535; fax: +1 508 647 2241. E-mail address: [email protected] (B.A. Huibregtse). 1553-8389/07/$ – see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.carrev.2007.08.002
being generally reported at 28 and 90 days [4–10]. Early vascular response to the polymer-based TAXUS Express2 PES (Boston Scientific Corporation, Natick, MA) in the noninjured swine coronary artery model has not yet been reported. The current study of the early postdeployment period assessed the short-term vascular healing response in PES versus bare metal stents (BMS). By comparing the vascular response of PES to that of Express2 BMS, we sought to determine whether primary differences occur in the first 20 days of tissue response. The normal process of stent healing initially involves the deposition of plasma protein and/or a thrombotic coating of fibrin adherent to or associated with struts and containing variable amounts of red blood cells (RBCs), platelets, and
252
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
leukocytes [11]. This coating material is common to all intravascular biomaterials and presents a substrate that promotes endothelial cell coverage. In this study, we demonstrate that strut-associated fibrin/thrombus persists longer in PES than in BMS and does not appear to be prothrombogenic or to represent an embolic risk. In addition, neointimal hyperplasia is suppressed in PES relative to BMS at 10 and 20 days. There were no other prominent differences in vascular response between these two stent types. Whether the slower fibrin resolution of PES is responsible for or connected to the reduction in restenosis rates observed clinically remains to be established. The study was carried out under Good Laboratory Practice (GLP) regulations for submission to health-regulatory bodies for device approval. 2. Materials and methods 2.1. Device description The polymer-based paclitaxel-eluting TAXUS Express2 and bare metal Express2 coronary stent delivery systems were used in the present study. The TAXUS Express2 stent system consisted of a slow-release formulation of paclitaxel (dose density, 1 μg/mm2) and a Translute-coated 316L stainless-steel balloon-expandable stent mounted on a lowprofile monorail delivery system. The Express2 stent system consisted of a 316L stainless-steel balloon-expandable stent mounted on an over-the-wire delivery system. Both PES and BMS were 2.75, 3.0, or 3.5 mm in diameter and 8.0 mm long. 2.2. Animals Forty nulliparous nonpregnant female domestic pigs (Sus scrofa) weighing 35 to 50 kg were assigned to five study groups. The animals were treated with aspirin (325 mg po) and clopidogrel (75 mg po) for 3 days prior to stent implantation, and with Procardia XL (30 mg po) 1 day prior to stent implantation. On the day of stent implantation (Day 0), each animal was appropriately anesthetized and implanted with one pair of the same type of overlapping stents (one pair each of PES or BMS) in the right coronary artery, left circumflex coronary artery, and left anterior descending coronary artery, for a total of three overlapping stent pairs per animal at an expansion ratio of 1.1:1. The animals received aspirin (81 mg po) and clopidogrel (75 mg po) daily after stent implantation until the end of the study. Six of 40 animals received two stent pairs because vascular anatomy prevented placement in all three vessels. Vessels in which stents were not overlapped were excluded from statistical analysis. To confirm the consistent delivery expansion of each stent and to evaluate stenosis by angiography, quantitative coronary angiography was performed at the time points listed in Table 1.
Table 1 Study design Number of BMS
Number of PES
Time point
Histology
SEM
Histology
SEM
20–28 h (1 day) 48±6 h (2 days) 96±6 h (4 days) 10±1 day (10 days) 20±1 day (20 days) Total stents
0 20 20 20 a 21 b 81
6 6 6 6 6 30
0 20 22 22 22 86
6 6 6 6 6 30
a b
Two nonoverlapping BMS were excluded from the final analysis. One single stent was implanted and excluded from the final analysis.
Peripheral blood was collected for analyses of standard hematology and serum chemistry parameters at two time points: prior to implantation and prior to termination. All animal surgical procedures, euthanasia, necropsy, and stent and tissue collection were performed by Charles River Laboratories (Osceola, WI). The test facility was accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International. The study was conducted in accordance with 21 CFR Part 58 (Food and Drug Administration), ENV/MC/CHEM (98) 17 (Organization for Economic Cooperation and Development [OECD]), or CLIA88. 2.3. Tissue collection Animals were euthanized at study termination. Comprehensive necropsy was performed on each animal. Any abnormalities were appropriately recorded. After euthanasia, hearts were harvested and pressure perfusion fixed with 10% buffered formalin. After fixation, the hearts were radiographed to map stent positions prior to excision. Stented vessels were dissected from the myocardium, allowing adequate proximal and distal reference vessel lengths in addition to the stented segment, and radiographed using high-resolution Faxitron to assess for appropriate overlap and stent strut fractures. The dissected vessels, along with the remaining heart tissues, routine tissue samples from the liver and kidneys, and any abnormal-appearing tissues in the chest or abdomen were assessed by the study pathologist. 2.4. Histopathology Stented vessels were prepared for light microscopy (LM) and scanning electron microscopy (SEM) analysis. Sections were obtained from the proximal unstented vessel, proximal nonoverlapped portion of the stented vessel, midoverlapped portion of the stented vessel, distal nonoverlapped portion of the stented vessel, and distal unstented vessel. Myocardial tissues adjacent and distal to the stent and any lesions were examined as well using standard histology techniques. All histological
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
253
Fig. 1. Scanning electron micrographs of PES (left) and BMS (right) specimens from 1 to 10 days. (A and B) Original magnification ×25; Day 1 examples demonstrate minimal thrombotic material associated with the struts of both stent types. (C and D) Original magnification ×25 and ×300; Day 2 examples demonstrate small patches of endothelial cells on the strut surfaces, as well as adherent leukocytes. Note that thrombotic material is not greater than Day 1 specimens and that overlapped stent regions (original magnification ×25; C) are also not more thrombogenic. (E and F) Original magnification ×25; Day 4 examples of endothelium cover significant areas of the strut surface, as well as adherent multinucleated giant cells. (G and H) Original magnification ×25; Day 10 examples demonstrate that most of the struts are covered with endothelialized neointima.
254
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
Fig. 1 (continued).
specimens were prepared for LM and SEM using standard procedures.
3. Results 3.1. Mortality, gross and clinical pathology, and stent integrity Forty animals were implanted with a total of 227 stents (108 overlapping BMS, 116 overlapping PES, and 3 individual or nonoverlapping BMS, which were excluded from the final analysis) in this study. There were no unscheduled mortalities, no abnormalities in hematology and serum chemistry values, and no abnormalities at necropsy. No strut fractures were detected using highresolution radiography. 3.2. One-day results No gross thrombi were observed on any of the three PES or three BMS pairs examined by SEM (Fig. 1A and B). There was a thin coating (average, b50 μm) of thrombus-like
material (fibrin and platelets) covering variable proportions of the stent struts. Leukocytes were adherent to the arterial endothelium and to the stent struts of overlapping PES and BMS, and appeared to favor the lateral strut surfaces as opposed to the upper flow surface. Most of the luminal arterial surface was covered with endothelium (i.e., areas in the open cells of the stent). The endothelium was not observed on stent struts. There were no differences between PES and BMS at the Day 1 time point, indicating that neither stent surface was thrombogenic nor proinflammatory. The deposition of thrombotic material occurred in both stent types primarily at bends in the strut configuration. In addition, the overlapped stent regions did not exhibit noticeably more thrombotic material than the nonoverlapped regions. No specimens were processed for histology in the 1-day cohort. 3.3. Two-day results No gross luminal thrombi were observed on any of the three PES or three BMS pairs examined by SEM (Fig. 1C and D) or LM. In addition, the thin thrombotic coating covering variable proportions of the struts was not greater than that observed on Day 1 and was not markedly different
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
255
Fig. 2. PES (left) and BMS (right) specimens from 2 to 20 days. (A and B) Day 2: a thin thrombotic material equivalently covers the struts of both PES and BMS at 2 days. (C and D) Day 4: endothelium usually covers struts associated with thrombotic material. (E and F) Day 10: neointimal formation in BMS specimens (F) and residual amounts of parastrut fibrin are observed. There appeared to be more neointimal parastrut fibrin in the 10-day PES specimens compared to the 4-day PES specimens. (G and H) Day 20: BMS groups exhibited a more mature and well-developed fibrocellular neointima than PES; parastrut fibrin was not commonly observed with BMS specimens, whereas PES specimens continued to exhibit parastrut fibrin and RBCs in the neointima.
256
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
Fig. 2 (continued).
between overlapping PES and BMS (Fig. 2A and B). The overlapped stent regions did not exhibit noticeably more thrombotic material than the nonoverlapped regions. Leukocytes were adherent to both struts and endothelium in PES and BMS, and were not at substantially different levels compared to Day 1. Endothelial cells were observed mostly as multicellular patches that were randomly distributed on the stent struts of both PES and BMS and were estimated to cover approximately 5–10% of the stent strut area (Fig. 1C and D). Occasional multinucleated giant cells (fused macrophages) were observed on the surface of struts in both stent groups (Table 2). 3.4. Four-day results No gross luminal thrombus was observed on any of the three PES or three BMS pairs examined by SEM and LM (Fig. 1E and F). In addition, the overlapped stent regions did not exhibit noticeably more thrombotic material than the nonoverlapped regions. Endothelium usually covered struts when they had thrombotic material associated with them and was less common when thrombus was lacking on struts (Fig. 2C and D). Leukocytes were present in the thrombotic material and were adherent to the endothelium.
Multinucleated giant cells (MNGCs) were observed adhering to the struts of both PES and BMS in areas devoid of endothelium. The peak time point for strutadherent MNGC was 4 days (Table 2). In many instances, they accounted for almost half of the strut coverage at this time point and had the appearance of covering strut areas devoid of the endothelium. As endothelial cell coverage progressed over time, there was diminution in MNGC presence (Table 2). 3.5. Ten-day results No luminal thrombus was observed on any of the three PES or three BMS pairs examined by SEM or LM (Fig. 1G and H). Both PES and BMS groups exhibited between 90% and 100% of the luminal surface covered by endothelium (Table 2). In areas where endothelial cell coverage was not yet complete, MNGCs were present, covering the struts. By 10 days, the surface of both stent types was largely covered by the endothelium; given the degree of endothelial coverage, observations of MNGCs were substantially reduced in numbers compared to the 4-day time point. Between 4 and 10 days, there was neointimal formation in BMS specimens and residual amounts of parastrut fibrin (Fig. 2E and F). The neointima in the overlapped stent regions was mildly less
Table 2 MNGC count on strut surfaces and endothelial coverage as a function of time BMS
PES
Day
MNGC count [mean±S.D. (n)]
% Samples with N90% endothelialization
MNGC count [mean±S.D. (n)]
% Samples with N90% endothelialization
2 4 10 20
0.7±1.4 (30) 5.5±4.4 (30) 2.5±3.1 (36) 0.1±0.5 (33)
0 52 97 100
0.2±0.5 (30) 5.8±5.5 (33) 2.8±4.9 (30) 0.9±2.2 (33)
0 47 82 100
n=sample size; S.D.=standard deviation.
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
developed than the nonoverlapped regions. PES specimens exhibited a neointima that was primarily composed of parastrut fibrin, RBCs, and mild amounts of leukocytes with a thin fibrocellular cap at the luminal aspect. The interstrut regions exhibited more developed fibrocellular tissues. Neovessels were sometimes observed in the neointima of both PES and BMS specimens. There appeared to be more neointimal parastrut fibrin in the 10-day PES specimens compared to the 4-day PES specimens. Subendothelial RBCs were observed in PES specimens but were not common in BMS specimens. The majority of struts were covered with a neointima in specimens from both groups. Leukocytes were present in the thrombotic material and were adherent to the endothelium but did not reach extensive levels in either group. As a marker of inflammation, leukocytes were not observed in large numbers, indicating that BMS does not provoke an inflammatory response beyond that expected following implantation of a device. Eosinophils were not observed in either group, suggesting the absence of a hypersensitivity reaction. 3.6. Twenty-day results Both PES and BMS groups exhibited a fibrocellular neointima with an endothelial covering (Fig. 2G and H). Parastrut fibrin was not commonly observed with BMS specimens, whereas PES specimens continued to exhibit parastrut fibrin and RBCs in the neointima, suggesting delayed clearance of the material. The neointima of PES specimens was more fibrocellular than the 10-day samples. The overlapped stent region exhibited more parastrut fibrin and less fibrocellular tissue than the proximal or distal nonoverlapped regions. No luminal thrombi were observed in either stent group. Occasional macrophages, often vacuolated, were present at the periphery of the parastrut fibrin in PES specimens but were not observed in BMS specimens. 4. Discussion This study sought to determine what histological differences, if any, exist between PES and BMS in the first 20 days of stent residence in swine coronary arteries. From previous work, it has been known that parastrut fibrin is present in PES but is uncommon in BMS at 30 days [12] and 90 days (unpublished data). We report here that both PES and BMS initially have equivalent amounts of fibrin or thrombotic material adherent to the stents at 1, 2, and 4 days. The deposition of thrombotic material appeared to be the result of turbulent flow following stent implantation, as it occurred in both stent types primarily at bends in the strut configuration. In addition, stents from both groups became covered with endothelium at approximately the same rate, and neither stent type provoked overt inflammatory cell presence or generated gross luminal thrombi. The first noticeable difference between the two stent types occurred at
257
10 days, when it was observed that BMS had low amounts of parastrut fibrin in a fibrocellular neointima, whereas PES exhibited persistent parastrut fibrin enveloped within a mostly immature fibrocellular neointima. This implies that fibrin resolution was more active between 4 and 10 days with BMS and probably occurred via a fibrinolytic pathway since a specific leukocyte infiltrate was not observed. The persistence of fibrin in PES and the appearance of foamy macrophages confined to the vicinity of fibrin suggest that paclitaxel-coated stents may inhibit such fibrinolysis and that fibrin is cleared by cell-mediated phagocytosis. Thus, it was demonstrated that the parastrut fibrin observed in PES at 30 days and beyond is present initially for both PES and BMS, but in the case of PES is removed more slowly through a different mechanism (phagocytosis) compared to quicker removal through a fibrinolytic pathway in the case of BMS in swine. The results demonstrate that PES do not inhibit endothelial cell proliferation or migration. Both stent types became covered with endothelium at roughly similar rates, and there were no overt differences in cell morphology. In addition and most importantly, in considering endothelial cell functionality, no luminal thrombi were observed at any time point in either stent group, nor was there evidence of embolic myocardial ischemia or infarction. As this observation held true for the period before the stents became endothelialized, it further demonstrates that coated stents are not more thrombogenic than BMS in this model. These results suggest that the low levels of late stent thrombosis observed in clinical studies may not be due to the stent itself and may reflect idiosyncratic reactions to drugs or polymers in a small percentage of patients with underlying atherosclerotic disease, rather than systematic effects on the presence or condition of the endothelium in general. Neutrophils and monocytes were associated with both stent types in the first few days, as would be expected as a response to acute injury. As a marker of inflammation, they were minimal in number and were not increased in PES compared to BMS, indicating that PES do not provoke an undesirable inflammatory response. Eosinophils were not observed, thus suggesting the absence of a hypersensitivity reaction. Macrophages were observed at the 20-day time point in PES specimens in association with parastrut fibrin and were usually foamy in appearance, indicating that they actively participated in the clearance of thrombotic materials, platelets, and RBCs. The early evaluation points in this study preclude an exact comparison to previously reported evaluations of drug-eluting stents (DES) in animal models [4–10,13] because the last evaluation in this study (20 days) occurs prior to the first evaluation in other studies (30 days). However, some similarities in healing were demonstrated between models when procedural differences were taken into consideration. The average stent-to-artery ratio in this study (1.03–1.25) was less than that reported for
258
P.S. Seifert et al. / Cardiovascular Revascularization Medicine 8 (2007) 251–258
overlapped stents in a rabbit model (1.13–2.45) [13] but was comparable to that reported in a single-stent study of sirolimus-eluting stents (SES) in porcine coronary arteries (1.16–1.29) [6]. In both the rabbit model [13] and this study, there was no significant difference in endothelialization between BMS and PES. However, in contrast to the rabbit model in which only approximately 66–80% of the lumen was endothelialized by 28 days for either BMS or PES [13], endothelialization in the porcine model was more rapid and largely complete in both PES and BMS (including the overlap section) by 20 days. Carter et al. [6] reported equivalent endothelialization between SES and BMS at 30 days in the porcine model, with a similar stent-to-artery ratio as used in this study. The differences in endothelialization rates between the rabbit and pig models are likely due to both species differences and procedural differences in stent-to-artery ratio; however, none of the models demonstrated a significant difference in endothelialization between DES and BMS. While fibrin deposition was similar in PES and BMS in this study at 2 and 4 days, there was greater deposition in PES at 10 and 20 days compared to BMS. Using a different scoring system in a model with greater overstretch, Finn et al. [13] also demonstrated a significantly greater fibrin deposition at 28 days in the rabbit model (as measured by struts with fibrin) in overlapping PES compared to BMS (P=.001 for PES vs. BMS). However, there was no difference in fibrin deposition between PES and BMS in the rabbit model at 90 days [13]. Thus, while both models demonstrate greater fibrin deposition in the overlapping PES compared to BMS at early time points, Finn et al.'s data suggest that this difference does not persist until 90 days. In summary, this study compared the histological response to overlapping PES and BMS in the first 20 days following implantation in a porcine coronary artery model. The absence of atherosclerotic disease and the young age of the animals limit the extrapolation of data from this healthy porcine model to human coronary artery disease patients; however, it serves as a method for the comparison of stent placement in similar conditions. Given these limitations, some observations were consistent throughout the study and were similar to results reported in other animal models. No difference in endothelialization was observed between PES and BMS. Results demonstrate that the only notable difference between PES and BMS is the presence of parastrut fibrin in PES, suggesting that paclitaxel may affect the pathway for fibrin removal. Whether this pathway also contributes to a slowdown of the development of the neointima requires further investigation, but it was observed in this animal model that PES delays neointimal formation, in accordance with the reduction of restenosis in the clinical setting.
Acknowledgments The authors would like to thank Saran Chhabra and Suzanne Wile for help in writing this manuscript and Michael Bravo for study support (all full-time employees of Boston Scientific Corporation). References [1] Grube E, Silber S, Hauptmann KE, Mueller R, Buellesfeld L, Gerckens U, Russell ME. TAXUS I: six- and twelve-month results from a randomized, double-blind trial on a slow-release paclitaxel-eluting stent for de novo coronary lesions. Circulation 2003;107(1):38–42. [2] Stone GW, Ellis SG, Cox DA, Hermiller J, O'Shaughnessy C, Mann JT, Turco M, Caputo R, Bergin P, Greenberg J, Popma JJ, Russell ME. A polymer-based, paclitaxel-eluting stent in patients with coronary artery disease. N Engl J Med 2004;350(3):221–31. [3] Stone GW, Ellis SG, Cannon L, Mann JT, Greenberg JD, Spriggs D, O'Shaughnessy CD, DeMaio S, Hall P, Popma JJ, Koglin J, Russell ME. Comparison of a polymer-based paclitaxel-eluting stent with a bare metal stent in patients with complex coronary artery disease: a randomized controlled trial. JAMA 2005;294(10):1215–23. [4] Lowe HC, Schwartz RS, Mac Neill BD, Jang IK, Hayase M, Rogers C, Oesterle SN. The porcine coronary model of in-stent restenosis: current status in the era of drug-eluting stents. Catheter Cardiovasc Interv 2003;60(4):515–23. [5] Schwartz RS, Edelman ER, Carter A, Chronos N, Rogers C, Robinson KA, Waksman R, Weinberger J, Wilensky RL, Jensen DN, Zuckerman BD, Virmani R. Drug-eluting stents in preclinical studies: recommended evaluation from a consensus group. Circulation 2002;106(14): 1867–73. [6] Carter AJ, Aggarwal M, Kopia GA, Tio F, Tsao PS, Kolata R, Yeung AC, Llanos G, Dooley J, Falotico R. Long-term effects of polymerbased, slow-release, sirolimus-eluting stents in a porcine coronary model. Cardiovasc Res 2004;63(4):617–24. [7] Garcia-Touchard A, Burke SE, Toner JL, Cromack K, Schwartz RS. Zotarolimus-eluting stents reduce experimental coronary artery neointimal hyperplasia after 4 weeks. Eur Heart J 2006;27(8):988–93. [8] Heldman AW, Cheng L, Jenkins GM, Heller PF, Kim DW, Ware M, Nater C, Hruban RH, Rezai B, Abella BS, Bunge KE, Kinsella JL, Sollott SJ, Lakatta EG, Brinker JA, Hunter WL, Froehlich JP. Paclitaxel stent coating inhibits neointimal hyperplasia at 4 weeks in a porcine model of coronary restenosis. Circulation 2001;103(18): 2289–95. [9] van der Giessen WJ, Lincoff AM, Schwartz RS, van Beusekom HM, Serruys PW, Holmes DR, Ellis SG, Topol EJ. Marked inflammatory sequelae to implantation of biodegradable and nonbiodegradable polymers in porcine coronary arteries. Circulation 1996;94(7):1690–7. [10] Schwartz RS, Chronos NA, Virmani R. Preclinical restenosis models and drug-eluting stents: still important, still much to learn. J Am Coll Cardiol 2004;44(7):1373–85. [11] Virmani R, Farb A. Pathology of in-stent restenosis. Curr Opin Lipidol 1999;10(6):499–506. [12] Balakrishnan B, Tzafriri AR, Seifert P, Groothuis A, Rogers C, Edelman ER. Strut position, blood flow, and drug deposition: implications for single and overlapping drug-eluting stents. Circulation 2005;111(22):2958–65. [13] Finn AV, Kolodgie FD, Harnek J, Guerrero LJ, Acampado E, Tefera K, Skorija K, Weber DK, Gold HK, Virmani R. Differential response of delayed healing and persistent inflammation at sites of overlapping sirolimus- or paclitaxel-eluting stents. Circulation 2005;112(2): 270–8.