Introduction The development of bare-metal stents (BMS) and optimal post-procedure antiplatelet therapy in the mid 1990s resulted in marked reductions in acute and subacute stent thrombosis, bleeding complications, negative remodeling, and acute vessel recoil, and a consequent dramatic rise in procedural success. Yet, intermediate-term outcomes remained hampered by restenosis from aggressive neointimal hyperplasia. The advent of drug-coated stents, which impede the growth of neointima while maintaining the strengths of BMS, heralded their widespread utilization to approximately 90% of procedures. Recently, however, very late stent thrombosis rates in excess of those seen with BMS were noted, prompting a reduction in the penetration of drug-coated stents to approximately 65% of all procedures. To date, there are no official guidelines on choice of bare-metal versus drug-coated stents in clinical practice. This article, therefore, summarizes current opinion regarding situations where BMS might be preferable to drug-coated stents, and where one drug-coated stent might be preferable to another, in hopes of guiding the clinician on optimal stent selection. History of Percutaneous Coronary Intervention Although a major breakthrough in technology and patient care, balloon angioplasty was fraught with several significant limitations, including uncontrollable plaque disruption and acute elastic recoil, leading to coronary occlusion and acute procedural myocardial infarction (MI). In addition, a 20–40% incidence of restenosis within the first 6–9 months was seen, due to a combination of negative vessel remodeling and neointimal hyperplasia.1 In addition to the development of recurrent ischemic symptoms, restenosis in this setting portended a poor prognosis.2 Balloon-expandable stents were developed to address these limitations. By scaffolding the vessel at the site of lesion disruption, they mitigated acute occlusion and elastic recoil, while virtually eliminating subsequent negative remodeling. However, there was an apparent increase in neointimal hyperplasia as a reaction to vessel injury from and inflammation to the foreign body, resulting in restenosis, albeit at lower rates than seen in the balloon angioplasty era.3,4 Two landmark trials solidified these findings and the benefits of BMS. The North American STRESS trial showed lower angiographic restenosis (31.6% vs. 42.1%) and lower target vessel revascularization (TVR, 10.2% vs. 15.4%) in patients who received a BMS versus balloon angioplasty, while the European BENESTENT trial confirmed these results (restenosis, 22% vs. 32% and TVR, 13.1% vs. 22.9%).5,6 Despite the marked improvement in procedural complications and intermediate-term clinical outcome with widespread utilization of BMS, two problems remained: subacute stent thrombosis and in-stent restenosis. Subacute stent thrombosis, defined as acute vessel occlusion within 30 days of the index procedure, was eventually reduced from 15% to the current rate of The Rise of Drug-Coated Stents Despite the significant relative reduction in restenosis provided by BMS, observed restenosis rates remained in the 20–30% range in randomized controlled trials, and as high as 50–60% in real-world experience, particularly in the high-risk subgroups of diabetics, bifurcation lesions, small caliber vessels, chronic total occlusions, long lesions, and overlapping stents.1,5,6,11 As mentioned previously, neointimal hyperplasia was the dominant contributor to in-stent restenosis in the BMS era. Indeed, due to the higher degree of vessel wall injury with stent placement compared to balloon angioplasty, a more exuberant neontimal response was evidenced.12 Endothelial and medial injury as a result of stent deployment causes inflammation, leading to recruitment of macrophages and lymphocytes to the target lesion. The subsequent release of cytokines and growth factors activate dormant vascular smooth muscle cells, which proliferate, enter the intima, and produce abundant extracellular matrix, a process that continues for 6 to 9 months.13 After multiple failed attempts to control neointima formation with pharmacotherapy and brachytherapy,1,14 polymer-coated stents that elute anti-proliferative drugs in a controlled-release fashion to coincide with the restenosis cascade emerged as the dominant treatment modality. The sirolimus-coated Cypher stent (Cordis Corporation, Johnson & Johnson, Miami Lakes, Florida) was approved by the FDA in April 2003, followed by the paclitaxel-coated Taxus stent (Boston Scientific, Natick, Massachusetts) the following year. Initial trials studying the effects of drug-coated stents were performed on low-risk patients and lesions, involving primarily single de novo lesions less than 30 mm in length and in vessels of moderate caliber (2.5–3.5 mm). Seven randomized controlled trials involving 2487 patients examined the efficacy and safety of sirolimus-coated stents.15–21 Compared to BMS, there was significantly lower restenosis culminating in fewer target lesion revascularizations (TLR) without an increase in death or MI at 8 months. Similarly, five randomized controlled trials involving 3513 patients assessed the efficacy and safety of the paclitaxel-coated stent, showing significantly reduced restenosis and TLR at 9 months without an increase in death or MI.22–26 Together, randomized controlled trials demonstrated 70% or greater reductions in the rate of TLR compared to BMS, an effect that seemed consistent across all patient and lesion subgroups, with TLR rates in the single digits.27 Such dramatic results brought about the rapid and widespread adoption of drug-coated stents, rapidly reaching 80% of revascularization procedures by 2005 and 90% by 2006.27,28 Intrinsic to this deep market penetration, drug-coated stents became the mainstay therapy for both “on-label” and “off-label” situations, the latter comprising roughly 60% of clinical practice. With their marked efficacy and no obvious safety issues, such penetration seemed entirely reasonable and was, in fact, confirmed by various “real-world” registries and randomized trials in various patient subsets out to one-year follow up.21,29–38 “Off-label” in this context included complex lesions, long lesions requiring overlapping stents, in-stent restenosis, saphenous vein grafts (SVG), chronic total occlusions, and primary angioplasty for acute MI.28 Safety Concerns: Mortality and Stent Thrombosis Shortly after their market release, reports surfaced of late stent thrombosis; that is, stent thrombosis after 30 days post-procedure.39–41 Of concern was the fact that late and very late (> 1 year) stent thrombosis had been rarely reported with BMS, and that such out-of-hospital thrombosis resulted in MI and death in upwards of 70–80% of patients.42 The question of whether drug-coated stents increase mortality was first raised in 2006 when a meta-analysis reported a statistically significant increase in mortality at 2–3 years in patients who received drug-coated stents.43,44 An analysis of the Swedish Coronary Angiography and Angioplasty Registry (SCAAR) also showed increased mortality in patients receiving drug-coated stents.45 However, in reviewing this latter registry in more detail, it was apparent that drug-coated stents were preferentially utilized on higher-risk lesions and patient populations, thereby resulting in higher event rates compared to BMS. Indeed, an updated and larger report by the SCAAR investigators showed a 50% reduction in restenosis, a significant reduction of death and MI within the first 6 months, and similar long-term mortality compared to BMS out to 4 years.46 Registries from Denmark and Canada subsequently supported this conclusion regarding overall safety.47,48 Further, in contrast to the initial meta-analysis presented in 2006, subsequent evidence from randomized controlled trials and several meta-analyses have now confirmed that drug-coated stents do not result in excess mortality at 4–5 years.49 In fact, most data now support reduced mortality with the use of drug-coated stents, particularly when the sirolimus-coated stent is utilized,49,50 probably related to its marked reduction in restenosis-related events. Indeed, it is now recognized that restenosis is not benign, presenting as MI in up to 20% of patients.51–53 While drug-coated stents do not appear to increase overall mortality at 4-5 years, available evidence does point to an increase in very late (> 1 year) stent thrombosis. Utilizing patient-level data from the major randomized controlled trials out to 4 years, the frequency of protocol-defined stent thrombosis did not differ at one year (late stent thrombosis), but was significantly higher in drug-coated stents from 1–4 years out (very late stent thrombosis).54 A rate of very late stent thrombosis as high as 0.4–0.6% per year has been reported, compared to 0.2% per year for BMS.49,55,56 Given that stent thrombosis results in death or MI in upwards of 70–80% of cases, one would expect long-term survival to be reduced with the use of drug-coated stents. However, as noted previously, this has not been shown at 4–5 year follow up. On the contrary, several reports suggest improved survival with the use of drug-coated stents. This has led many to believe that any increase in death and MI attributed to the low-frequency event of very late stent thrombosis is likely counterbalanced by a reduction in death and MI from the marked improvement in restenosis.27,57 Indeed, a recent meta-analysis comparing drug-coated stents and BMS before and after the one-year time point post-procedure found no difference in mortality at all time points, despite a reduction in MI and TLR during the first year post-procedure, and an increase in stent thrombosis after the one-year time point.57 Etiology and Risk Factors for Late Stent Thrombosis The etiology of late stent thrombosis and very late stent thrombosis, in particular, remains unclear, but is the subject of active research. Drug-coated stents have two additional elements compared to BMS, and thus two components that must by definition cause these late events: the anti-proliferative drug and the polymer coating. Both paclitaxel and sirolimus delay endothelialization over and between stent struts, and impair normal endothelial function.58–61 Since endothelial cells are an important regulator of thrombosis and vessel reactivity, such dysfunction may tip the balance towards thrombosis. In addition to its effect on endothelial cells, anti-proliferative drugs also cause persistent fibrin deposition in the vessel well, particularly with the paclitaxel-coated stent. Such persistent fibrin may be a trigger for late and very late stent thrombosis, especially in conjunction with endothelial dysfunction. Polymer coatings have also been linked to late thrombosis. Inflammatory response to polymer coating has been implicated in both late incomplete stent apposition (LISA) and hypersensitivity reaction. LISA, defined as the clear separation of stent struts from the vessel wall when such separation was not present immediately post-procedure, is thought to be due to positive remodeling from chronic inflammation or drug-effect.62 Present in up to 5% of BMS, some reports have found LISA in over 10% of drug-coated stents within 6 months.63–65 At least one study has implicated LISA in late stent thrombosis.66 Although rare, hypersensitivity to the polymer coating has also been reported and implicated in LISA, peri-stent aneurysm formation, and stent thrombosis. In particular, the polymer coating for the sirolimus-coated stent seems prone to chronic inflammation and such hypersensitivity reactions, as its components have been linked to such reactions in other territories.67,68 It has also been hypothesized that late stent thrombosis may be a result of poor interventional technique; that is, poor stent deployment and resultant poor initial stent apposition. While BMS have exuberant neointimal hyperplasia and endothelialization to counteract poor deployment and stent malapposition, the factors discussed above (delayed and dysfunctional endothelialization, LISA, chronic inflammation, and hypersensitivity), together with minimal neointimal formation, may make drug-coated stents more sensitive to poor technique, and hence more prone to stent thrombosis. Many experts currently advocate better attention to stent sizing, along with intravascular ultrasound (IVUS) or post-dilation with high-pressure balloons in order to optimize stent apposition to the vessel wall, and thereby minimize late stent thrombosis.69–73 Multiple studies have now shown that the single most important predictor of stent thrombosis is premature discontinuation of dual antiplatelet therapy.74 Currently, therefore, the FDA has recommended a minimum of one year of therapy for patients who have received a drug-coated stent.75 Other risk factors for late stent thrombosis appear to be renal insufficiency, diabetes, long total stent length, bifurcation stenting, incomplete stent expansion, poor stent apposition, stent strut penetration into a necrotic plaque core, left ventricular dysfunction, stent implantation during an acute coronary syndrome, and treatment of diffuse in-stent restenosis.71,76–84 In addition, one study showed an excess of stent thrombosis and related events in SVG lesions.85 It follows from the above data that patients receiving drug-coated stents should have either post-dilation at higher pressure or IVUS guidance to optimize stent expansion and apposition to the vessel wall. In addition, bifurcation stenting should be avoided whenever possible (in preference to single stent placement and provisional side-branch stenting). Finally, and perhaps most importantly, all patients must receive at least one year of dual antiplatelet therapy. Table 1 lists factors associated with late and very late stent thrombosis according to whether they are related to the patient, the clinical presentation, or target lesion. Choice of Stent: Drug-Coated versus Bare Metal Taken together, available literature supports drug-coated stents in the majority of percutaneous coronary interventions (PCI), including so-called “on-label” and “off-label” situations. This is especially true if care is taken to perform well-deployed stent placement at high pressure, preferentially with IVUS-guidance. Consistent with this, current drug-coated stent penetration is estimated at 75% for the first quarter of 2009, up from approximately 60–65% in 2007.42 Since “on-label” procedures constitute roughly 40% of all intervention, clearly there is a high penetration of drug-coated stents in both “on-label” and “off-label” situations. Table 2 lists situations that might favor use of a BMS over a drug-coated stent. Since the extent of intimal hyperplasia with bare-metal and drug-coated stents is not dependent on vessel size, it follows that larger vessels should prove least likely to benefit from the use of drug-coated stents. Indeed, randomized-controlled trials routinely excluded vessels > 3.5 mm in diameter. Several studies have since confirmed the lack of any clinically relevant decrease in TLR with use of drug-coated stents in this population, particularly in vessels > 4.0 mm.86,87 In addition, poor stent apposition might be more likely in such large vessels, potentially raising the risk of late and very late stent thrombosis and resultant death or MI.88 Two “off-label” populations that deserve specific mention are the acute ST elevation myocardial infarction (STEMI) patient and the SVG patient. In the former, patient compliance with dual anti-platelet is unclear, due to the emergent nature of the procedure when little time is available to assess barriers to patient compliance. In addition, reports have found a higher incidence of LISA in patients who received drug-coated stents for STEMI, possibly due to the higher frequency of mural thrombus and vessel spasm at the time of procedure, resulting in inadvertently undersized stents. Despite these concerns, a recent randomized controlled trial comparing drug-coated stents to BMS in STEMI patients found no difference in death and MI with improved restenosis at one year.89 However, since very late (> 1 year) thrombosis by definition was not yet reported, drug-coated stent use in this population remains controversial. In the SVG population, at least one randomized trial has noted a higher risk of late mortality in patients who received drug-coated stents, presumably due to stent thrombosis.85 Due to the larger size of these vessels and, therefore, limited relative efficacy of drug-coated stents, their use in SVG disease also remains controversial. Given that premature discontinuation of dual antiplatelet therapy is the strongest predictor of stent thrombosis with drug-coated stents, it follows that a thorough evaluation for factors associated with premature discontinuation must be undertaken before placing these stents. Indeed, premature discontinuation within 6 months of stent placement has been associated with a 25–90 fold increased hazard for stent thrombosis, typically arising within a few days post-discontinuation.77,78 Therefore, patients should be assessed for bleeding risk, and those with bleeding disorders or history of severe gastrointestinal or other bleeding considered for BMS. In addition, patients who require long-term anticoagulation with warfarin, such as those with atrial fibrillation, mechanical heart valve, or hypercoagulable state, already have a high incidence of bleeding, which would be increased by the addition of dual antiplatelet therapy.90 Drug-coated stents are therefore best avoided in these patients as well. A patient who may require elective or urgent surgical procedures over the ensuing year would also be a poor candidate for drug-coated stents, as such procedures typically require discontinuation of anti-platelet therapy. Patients with known malignancies would be in this category, given the unpredictable nature of their disease. BMS, with their requirement for limited duration dual antiplatelet therapy of 2–4 weeks, would be preferable in these scenarios.91 Finally, economic barriers to patient compliance exist, given the cost of dual antiplatelet therapy and the frequency of uninsured patients noted in clinical practice. There is evidence that such barriers do indeed limit drug-coated stent utilization in clinical practice.92 Those without insurance should be screened for their ability to afford dual antiplatelet therapy. In addition, cultural and language barriers similarly exist, and care must be taken to confirm that all patients understand the risks and benefits of drug-coated stents, and the hazard of premature discontinuation of dual antiplatelet therapy. In most instances, this requires discussing such matters in the native tongue or via accepted translation services before a drug-coated stent is implanted. One final category of patients who would be considered poor candidates for drug-coated stents are those with limited life expectancy or quality. In such patients, the risks of stent thrombosis and bleeding due to dual antiplatelet therapy far outweigh any potential efficacy benefit. Patients in this category include the very elderly (> 90 years old) and those with comorbidities that severely limit life expectancy or quality, such as advanced malignancy or severe dementia. Choosing Between Drug-Coated Stents There are currently four distinct drug-coated stent platforms on the market in the United States. These include the first generation sirolimus-coated (Cypher) and paclitaxel-coated (Taxus) stents launched in 2003–2004, and the second generation zotarolimus-coated (Endeavor, Medtronic, Santa Rosa, California) and everolimus-coated (Xience, Abbott, Redwood City, California) stents launched in 2008. A brief description of each is necessary in order to define populations that might theoretically benefit from one over the other, although there remains controversy whether significant differences exist in routine clinical practice. The Cypher sirolimus-coated stent has a stainless steel platform with a 12.6 micron-polymer coating composed of poly(ethylene co-vinyl acetate) and poly(n-butyl methacrylate) that elutes most of the drug within 3-4 weeks, and essentially all of it by 120 days. Sirolimus inhibits the G1 phase of the cell cycle, conferring its antiproliferative properties. Its restenotic efficacy is substantial, with a late loss (amount of neointimal hyperplasia in millimeters) of 0.10–0.15 mm.42 In diabetic patients, however, the late loss likely approximates 0.30 mm.93,94 The polymer has been associated with hypersensitivity reaction and chronic inflammation, possibly linked to late stent thrombosis as noted previously. In addition, the stent struts are thicker (140 microns) than other currently available drug-coated stents, and are therefore believed to be less flexible and deliverable and more prone to stent fracture. Due to its early release and marked clinical efficacy, the Cypher stent has been utilized in more patients worldwide than any other drug-coated stent.42 The Taxus paclitaxel-coated stent also has a stainless steel platform (both Taxus Express and Taxus Liberté, Boston Scientific) but a different 19.6 micron polymer coating composed of poly(styrene-b-isobutylene-b-styrene). Paclitaxel stabilizes microtubules and blocks intracellular signaling, inhibiting smooth muscle cell migration and trophism. In contrast to sirolimus, the elution of paclitaxel is slow and controlled, with a sizable portion of the drug maintained in the polymer after elution.42 Due to a retained drug, there is evidence of continued fibrin activity in the vessel wall, as noted previously, which has been theorized as a cause of late thrombosis.95 However, compared to Cypher, the stent design of the Taxus Express allows for more flexibility and hence better deliverability. The most recent version (Taxus Liberté), in fact, has thinner struts (97 microns) and an even more flexible design, allowing for greater deliverability. The anti-restenotic efficacy is marked by a late loss of 0.30–0.35 mm, somewhat higher and therefore potentially less efficacious than Cypher.42 Due to its in vitro glucose-independent mechanism of action, however, the late loss with paclitaxel appears to be maintained regardless of diabetic status.93,94 The Endeavor zotarolimus-coated stent uses a cobalt-based alloy stent platform, which allows for thinner struts (91 microns) and improved flexibility and deliverability. The drug is an analogue of sirolimus with an identical mechanism of action. The thinnest 4.8 micron polymer coating is phosphorylcholine-based and hydrophilic, and meant to be biocompatible with minimal to no inflammatory response.42 In addition, the drug is incorporated in the polymer in such a way that the drug is eluted entirely within 3–4 weeks, with only 1 micron polymer coating remaining thereafter. Due to the rapid elution that does not cover the entirety of the restenosis cascade, however, late loss is somewhat greater at 0.6–0.7 mm. Consistent with this, recent clinical trials have supported higher restenosis rates with the Endeavor stent when compared to Cypher or Taxus in high-risk patients.94 However, since the polymer is degraded alongside drug elution, the zotarolimus-coated stent may prove safest in terms of stent thrombosis. The Xience everolimus-coated stent also uses a cobalt-chromium alloy with currently the thinnest struts on the market (81 microns), allowing for increased flexibility and deliverability. The drug is also an analogue of sirolimus with an identical mechanism of action. The 7.8 micron polymer is a nonadhesive, durable biocompatible fluoropolymer that elutes 80% of the drug within one month and complete elution over 120 days, similar to the sirolimus-coated Cypher stent.42 Due to its similar drug elution profile, its antirestenotic efficacy is marked and appears similar to Cypher with a late loss of 0.10–0.15 mm. This occurs despite an approximate 50% reduction in the required drug load, which is achievable due to the favorable properties of its polymer.42 Table 3 lists theoretic pros and cons for available drug-coated stents. While many believe the differences are minimal, studies suggest otherwise. A meta-analysis of 16 randomized trials and almost 9000 patients comparing sirolimus-coated and paclitaxel-coated stents found no difference in mortality or MI, but a 26% relative reduction in reintervention and 34% relative reduction in stent thrombosis with use of the sirolimus-coated stent.96 Given the possible improved efficacy of paclitaxel versus sirolimus (and its analogues) in diabetic patients, the ISAR-DIABETES study compared outcomes in this high-risk patient subgroup. Sirolimus-coated stents again were associated with less restenosis and fewer TLR, with a late loss of 0.43 mm versus 0.67 mm.97 While other studies have shown non-inferiority between the two stent platforms, the bulk of evidence favors the sirolimus-coated stent for both safety and efficacy, regardless of diabetic status.94,97 In terms of second generation stents Endeavor and Xience, their thinner struts and polymer coating are associated with improved deliverability, a major advantage in complex lesion subsets. In addition, the thinner polymer coating has been shown to reduce peri-procedural MI, owing to less side-branch occlusion. Such MI are linked to mortality and therefore clinically important.98 Compared to the first generation drug-coated stents, there are fewer data both randomized and “real-world” regarding the second-generation drug-coated stents, owing to their more recent arrival. When the Endeavor stent was compared to the Cypher stent, with late loss as the primary outcome, the Cypher stent proved superior.99 To date, however, there has not been a completed randomized controlled trial comparing these two stents clinically with regard to restenosis, TLR, or stent thrombosis. When the Endeavor stent was compared to the Taxus stent, the former was found to be non-inferior clinically, prompting FDA approval.100,101 In addition, there was some evidence of reduced late stent thrombosis compared to the Taxus stent at long-term follow up, consistent with its possible safer profile.101 Other randomized controlled trials and “real-world” registries have not supported this safety benefit of the Endeavor stent over other drug-coated stents, however, while suggesting higher clinical restenosis.94,102 In “real-world” or randomized clinical trial experience, the Xience stent has not been compared to the Cypher stent to date, but would be expected to have similar efficacy due to its similar drug-elution curve and late loss. Compared to the Taxus stent, however, Xience showed a 45% reduction in major adverse cardiac events at two years.103 The Xience stent was therefore the first to show clinical superiority to another approved drug-coated stent in a randomized controlled trial. Although rates of stent thrombosis at all time points were not statistically different, there was a trend to fewer stent thromboses at two years with use of this second-generation stent.103 Currently, the everolimus-coated stent is the favored platform nationally, with over 50% market share, owing to its late loss similarity to Cypher and clinical superiority when compared directly to another approved drug-coated stent. Its thin cobalt-chromium platform, as with the zotarolimus-eluting stent, allows for greater deliverability and reduced peri-procedural MI compared to first-generation stents. Whether one drug-coated stent is superior to another regarding late and very late stent thrombosis remains unclear. Conclusions Drug-coated stents have proven to be a breakthrough technology, addressing both acute and chronic inadequacies of balloon angioplasty and BMS. However, the phenomenon of very late stent thrombosis is now clearly associated with drug-coated stents, occurring one or more years after placement at a rate of 0.4–0.6% per year. Stent thrombosis appears related to a variety of factors, including chronic inflammation, hypersensitivity, persistent fibrin activity, late incomplete stent apposition and endothelial dysfunction, all of which appear mitigated by prolonged dual antiplatelet therapy and more diligent interventional technique. Newer, so-called second generation stents have been designed to partially address this problem, although long-term data are needed. Until then, drug-coated stents are best used in patients who can comply with a minimum of one-year dual antiplatelet therapy, and in whom the efficacy and safety has been reasonably proven across randomized controlled trials, meta-analyses, and “real-world” registries. In this regard, the approximate 75% market penetration of drug-coated stents in current interventional practice appears appropriate. ___________________________ From the Division of Cardiology, Department of Medicine, Winthrop University Hospital, Mineola, New York. Manuscript submitted June 15, 2009, provisional acceptance given July 16, 2009, accepted July 27, 2009. Disclosure: Dr. Naidu reports that he is a member of the speakers bureau for Abbott Vascular, Cordis Corporation and Medtronic Vascular. Address Correspondence to: Srihari S. Naidu, MD, FACC, FAHA, FSCAI, Director, Cardiac Catheterization Laboratory, Winthrop University Hospital, 120 Mineola Blvd, Suite 500, Mineola, NY 11501. Email: firstname.lastname@example.org.