Article Text

Download PDFPDF

Drug-eluting stents: a critique
  1. N Melikian1,2,
  2. W Wijns1
  1. 1
    Cardiovascular Centre Aalst, OLV Ziekenhuis, Aalst, Belgium
  2. 2
    Cardiology Department, King’s College Hospital, London, UK
  1. Dr W Wijns, Cardiovascular Centre Aalst, OLV Ziekenhuis, Moorselbaan 164, 9300 Aalst, Belgium; william.wijns{at}


Despite advances in the design of balloons and stents, restenosis remains a major drawback of coronary angioplasty. Multiple randomised trials have demonstrated that drug-eluting stents (DES) can significantly reduce rates of restenosis by 60–75% across both lesion and patient subsets. In recent years there has been an exponential increase in the worldwide use of DES. This expansion has occurred as a result of an enthusiastic extrapolation of results from randomised trials leading to “off-label” use of DES in anatomical or clinical high-risk scenarios, or both. However, emerging medium- to long-term follow-up data have raised concerns about the safety of DES. A number of analyses have recently shown increased rates of late stent thrombosis in patients with DES. The exact mechanisms leading to stent thrombosis remain unclear. This article critically reviews the available efficacy and safety data on DES and discusses the factors influencing our current practice and perception of the net value of DES.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

“Sans la liberté de blâmer, il n’est point d’éloge flatteur”

Pierre-Augustin Caron de Beaumarchais, 1732–1799

Since the first balloon angioplasty in the late 1970s, percutaneous coronary intervention (PCI) has undergone rapid transformation to become an essential component of modern management for coronary artery disease.1 PCI is now widely accepted as being an important mode of coronary revascularisation, with annually over half a million procedures being performed in the United States alone.

PCI as practised by the modern-day interventional cardiologist is testament to two decades of significant technopharmacological advances, dealing with three key obstacles—namely, overcoming mechanical problems with remote delivery of balloons/stents, prevention of platelet activation/aggregation leading to thrombus formation, and reduction in rates of restenosis. New designs and materials now allow delivery and effective expansion/deployment of balloons and stents in remote parts of the coronary circulation, in often rigid, tortuous and severely stenotic vessels. Dual antiplatelet therapy with a combination of aspirin and a thienopyridine (such as clopidogrel) reduces devastating PCI-associated thrombosis. And the widespread adoption of drug-eluting stents (DES), the most recent of these advances, has in turn reduced the rates of restenosis by 60–75% across both lesion and patient subsets.13

DES have been hailed as “breakthrough” technology (fig 1), thus satisfying the criteria for fast track approval by regulatory authorities.4 Therefore, unlike the first two of the aforementioned challenges in the development of PCI, “real-world” medium- to long-term data on the efficacy and safety of DES are only now becoming widely available to the interventional community. This information has re-ignited the debate as to whether DES have finally overcome restenosis, the “Achilles heel” of PCI, and if so at what cost.

Figure 1 A model for “breakthrough” technology using DES. CABG, coronary artery bypass grafting.4

In this article we critically review the available efficacy and safety data on DES and discuss the issues influencing our current practice and perception of the net value of this technology.


Despite major advances in balloon and stent design, significant restenosis (20–40%) remains a major drawback of PCI without the use of DES.2 3 The ability to demonstrate a reduction in neointimal proliferation in animal models of PCI using systemic rapamycin (an antibiotic with potent immunosuppressant properties) paved the way for the design of specific polymer-coated stents, with the ability to elute locally measured quantities of antiproliferative drugs to inhibit neointimal hyperplasia.5

Although over 20 DES have been granted CE mark approval so far, our current understanding of the efficacy of DES is primarily derived from robust clinical data on three commercially available stents: the sirolimus-eluting stent, Cypher (SES; Cordis/Johnson & Johnson, Miami Lakes, Florida, USA), the paclitaxel-eluting stent, Taxus (PES; Boston scientific Corp, Natick, Massachusetts, USA) and, more recently, the zotarolimus-eluting stent, ENDEAVOR (ZES; Medtronic, Santa Rosa, California, USA).

Randomised trials

The first study in man of a DES evaluated the ability of the SES to reduce restenosis: 30 patients who received a SES were shown to have 0% restenosis at 4 months’ follow-up, paving the way for the initial two well-designed prospective, randomised, controlled trials (RCTs) of SES.6 The RAVEL (120 patients in each group) and SIRIUS (533 patients in the SES group and 525 patients in the bare-metal stent (BMS) group) trials compared the SES with its bare-metal counterpart.7 8 Both trials demonstrated hitherto unseen reductions in rates of restenosis (percentage restenosis—RAVEL 0% vs 28.8%, SIRIUS 3.2% vs 35.4%) at 6-month angiographic follow-up. Similarly, the 1-year incidence of major adverse cardiac events (MACE) was also significantly lower for patients receiving SES in both trials; driven essentially by a reduction in repeat revascularisation procedures.7 8 On the basis of data on the 1296 patients enrolled in these studies, the US Food and Drug Administration (FDA) in 2003 approved the use of SES for patients with previously untreated coronary stenosis of <30 mm in length and a reference vessel diameter of 2.50–3.75 mm. Similar results with the PES followed rapidly, leading to its approval by the FDA in 2004. Although, the ZES stent is now available in Europe, approval by the FDA for commercial use in the United States is still awaited.

Currently data are available from over 50 RCTs, non-randomised trials and registries investigating the efficacy and safety of DES.2 3 The overall message emanating from both individual studies and meta-analyses of the pooled data indicates that in the short- to medium-term DES are highly efficacious in preventing restenosis and their use is as safe as using BMS.2 3 This is well illustrated in a recent meta-analysis of RCTs of SES and PES (presenting data on a total of 8987 patients; 4574 with DES and 4413 with BMS) by Roiren et al.2 Analysis demonstrated significantly lower MACE and restenosis rates with DES (fig 2). The overall occurrence of MACE was reduced from 19.9% to 10.1% (odds ratio (OR) = 0.46, 95% confidence interval (CI) 0.41 to 0.52, p<0.001) and adjusted rates of angiographic restenosis (all but one study had angiographic follow-up data on >85% of patients) from 31.7% to 10.5% (OR = 0.25, 95% CI 0.22 to 0.29, p<0.001) with DES. Furthermore, there was no difference in overall rates of mortality (DES: 0.9%, BMS: 1.2%, OR = 1.02, 95% CI 0.64 to 1.64; p = 0.92), or stent thrombosis (DES: 1.0%, BMS: 0.9%, OR = 0.82, 95% CI 0.41 to 1.25; p = 0.44) between DES and BMS groups (fig 2). Rates of both Q and non-Q wave myocardial infarction were also similar between the two groups.2

Figure 2 Odds ratio and p value for rates of major adverse cardiac events (MACE) (A), restenosis (B), mortality (C) and subacute thrombosis (D) for sirolimus-eluting stents (SES) and paclitaxel-eluting stents (PES) and their bare-metal stent (BMS) counterparts. DES, drug-eluting stents. Adapted from Roiren C, et al.2

Other earlier meta-analyses also yield comparable results.3 For example, Babapulle et al in their meta-analysis of 11 RCTs (2641 with DES and 2449 with BMS) demonstrated similar significant reductions in angiographic restenosis (DES: 8.9%, BMS: 29.3%, OR = 0.18, 95% CI 0.06 to 0.40) and MACE rates (DES: 7.8%, BMS: 16.4%, OR = 0.42, 95% CI 0.32 to 0.53) with DES.3

Comparison between DES

Although, all three commercially available DES have significant benefits over their bare-metal counterpart, there may be differences in the magnitude of benefit between each type of DES.2 9 10 For example, in a randomised head-to-head comparison of SES and PES, Windecker et al demonstrated that at 6-month follow-up patients receiving PES had higher MACE rates (SES: 6.2%, PES: 10.8%, OR = 0.56, 95% CI 0.36 to 0.86; p = 0.009), driven primarily by lower rates of revascularisation in the SES group (SES: 4.8%, PES: 8.3%, OR = 0.56, 95% CI 0.34 to 0.93; p = 0.03).9 Direct comparisons with ZES have also been made.10 The ENDEAVOR III trial was a direct comparison of SES and ZES. Data from this study demonstrated that patients who received a ZES had greater angiographic late lumen loss and binary restenosis. However, despite the weaker antiproliferative properties of ZES, clinical outcome at follow-up was similar between the two groups (clinically driven target lesion revascularisation (TLR), SES: 3.5%, ZES: 6.3%, p = 0.34).10 The ENDEAVOR IV trial, which will be reported by the end of 2007, is a direct comparison of PES and ZES.

High-risk patients and complex lesions

Although, there is clear evidence that DES are highly efficacious in reducing restenosis in de novo lesions of moderate length and diameter (see FDA recommendation above for “on label use”), the quality of the data on the efficacy of DES in anatomically complex coronary lesions, such as treatment of long and bifurcation stenosis, left main coronary disease, chronically occluded vessels (CTO) and saphenous venous grafts, and in patients at higher risk such as those with diabetes and acute coronary syndromes, is limited. Relatively few specifically designed RCTs are available in these important groups of patients, with the bulk of evidence being based on results from either registries or subgroup analysis of RCTs which were originally set up to investigate an alternative primary end point.

Currently available evidence provides some support for the use of DES in long coronary stenoses.1113 Subgroup analysis of 100 patients from the C-SIRIUS trial (lesion length between 15 and 32 mm; rates of angiographic restenosis—SES: 2.3%, BMS: 52.3%, and TLR—SES: 4%, BMS: 18%]11 and 96 patients from the RESEARCH registry (mean lesion length 61.2 mm; rates of angiographic restenosis: 11.9%, TLR: 4.2%) demonstrated the clinical efficacy of DES in long coronary stenosis.12 More recently, the TAXUS IV trial (448 patients with mean lesion length of 20.6 mm) has also confirmed significantly lower TLR rates (PES: 6.8%, BMS: 18.9%; p<0.001) with the PES at 9-month follow-up.13 No differences in MACE rates (PES: 16.4%, BMS: 22.5%; p = 0.12) were found between the two groups. However, pooled analysis of restenosis as a function of lesion length demonstrates that lesion length continues to remain an independent predictor of restenosis even with DES (lesion length:angiographic restenosis rates; <33 mm: 4%; 33–51 mm: 7.1%; >51 mm: 15.4%).14

In contrast, evidence for the use of DES in bifurcation stenosis is more limited and importantly there are no randomised head-to-head comparisons of DES and BMS. Most published studies to date have been designed to investigate the optimal technique for treatment of coronary bifurcation stenosis—namely, comparison of the use of two DES in both main vessel and side branch (MV + SB) or stenting of the main vessel and optimal treatment of the side branch (MV).1517 The Nordic bifurcation study is the largest such study, with 413 patients randomised to receive SES in either MV + SB or MV. No significant differences in MACE and angiographic restenosis rates were identified between the two groups.16 Overall results from a number of studies have been promising with TLR rates of between 5 and 18%.1517 Outcome in bifurcation stenosis is dependent on the ability to achieve complete and appropriate mechanical scaffolding of the target lesion. Using current techniques and stent designs this is not always possible. To deal with this important factor a number of dedicated drug-eluting bifurcation stents, including self-expanding devices, are currently under investigation.

CTOs constitute a sizeable proportion of patients undergoing PCI (up to 5% of all case in the USA).1820 Currently, no large randomised studies are investigating the role of DES for treatment of CTOs. However, results from a number of small studies (up to 100 patients) comparing restenosis and TLR rates in CTO patients treated with a DES, with historical matched BMS controls, have been encouraging (restenosis rates—DES: 2–19%, BMS: 32–51.1%).1820

Another important area is the management of patients with unprotected left main stem (ULMS) coronary disease, a traditional bastion of coronary bypass surgery. Several small studies have reported lower rates of restenosis, MACE and TLR with DES, in comparison with historical BMS controls.2123 Results from the SYNTAX trial, which randomly compares DES with coronary artery bypass graft (CABG) surgery in complex and ULMS stenosis, will further clarify the role of DES in management of ULMS coronary disease; an area where coronary surgery is highly effective with excellent long-term outcome.

Information on the role of DES for the treatment of saphenous vein grafts remains conflicting and once again limited to non-randomised studies and registries.2427 A number of small non-randomised studies have shown improved short- and medium-term outcome in patients receiving a DES (restenosis rate: 10%, TLR rate 3.3%) in comparison with historical matched patients treated with BMS (restenosis rate: 26.7%, TLR rate 19.8%).2426 However, not all studies have been encouraging. For instance, the Prospective Multicenter German Cypher Stent Registry, designed to investigate real-world outcome from SES use, has shown revascularisation rates of around 21% with SES in vein grafts; a figure similar to that achieved with BMS.27 Critics of DES argue that the prothrombotic milieu of the venous conduits, combined with the delayed healing associated with DES, are important arguments against routine use of DES for vein graft PCI. Furthermore, DES do not deal with the fundamental question of distal embolisation, which is an important risk of PCI in venous grafts. Final results from the RRISC trial, which randomly compares SES with BMS in venous grafts, do not support the use of SES for this indication.24

Debate continues about the use of DES for high-risk patients such as those with diabetes and acute myocardial infarction (AMI). Patients with diabetes are known to have worse clinical outcome after coronary revascularisation, regardless of the method used.28 29 Initial results from subgroup analysis of diabetic patients from a number of RCTs (including the RAVEL, SIRIUS and TAXUS IV) have been encouraging, demonstrating significant reductions in rates of restenosis, TLR and/or MACE in patients receiving a DES.28 29 In the ARTS II study, which compared real-life patients receiving SES with matched controls of the CABG surgery arm of the previously conducted ARTS I trial, 1-year MACE rates for diabetic patients were similar between the two groups (SES: 15.7%, CABG: 14.6%).30 However, TLR remained significantly higher in the SES group (SES: 12.6%, CABG: 4.2%).30 At a preliminary glance these results are highly encouraging but key criticisms remain: the small number of diabetic patients enrolled in these studies as well as the post hoc analysis design. Furthermore, not all patients, especially in the TAXUS IV trial, underwent angiographic follow-up.31 This is important considering diabetic patients have higher rates of silent ischaemia than non-diabetic patients.

To date only two small, randomised studies of exclusively diabetic patients have been published.32 33 The DIABETES trial compared outcome in 80 diabetic patients randomly assigned to receive a SES or BMS. The SES group had significantly lower TLR (7.3%) and MACE (11.3%) at 9-month follow-up.32 The ISAR-DIABETES trial compared SES and PES in diabetic patients.33 Although the SES arm had lower angiographic late loss, the study was not powered to demonstrate clinically important differences between the two groups.33 In addition, there is little information on the optimal use of adjuvant pharmacological therapies in diabetic patients undergoing PCI with DES.28 29 Results from current studies (such as FREEDOM and CARDIA), designed specifically to examine the challenging problems of diabetic patients, are required before DES or a specific device can become the recommended choice in diabetic patients.

The role of DES in primary PCI has also received significant attention. To date seven RCTs of DES in AMI have been published.34 Four RCTs demonstrated significant reductions in MACE rates and the remaining three demonstrated a non-significant trend towards lower event rates with DES.34 In a recent meta-analysis of these studies, MACE rates (death, repeat MI and revascularisation) were significantly lower with DES (DES: 9.3%, BMS: 17.6%, OR = 0.53, 95% CI 0.43 to 0.66), once again driven by significantly lower TLR rates in the DES group (DES: 4.8%, BMS: 12.0%, OR = 0.40, 95% CI 0.30 to 0.54).34 Of importance, the incidence of death, repeat MI and stent thrombosis were similar between DES and BMS groups in both individual studies, as well as in the meta-analysis. Despite these encouraging short- to medium-term results, there are currently no data on long-term (>1 year) stent thrombosis or survival with DES in the context of primary PCI.

However, despite the absence of conclusive evidence for the use of DES in a number of important areas, there has been a quantum leap in the use of DES world wide. This expansion is brought about by enthusiastic extrapolation of results from RCTs, leading to “off-label” use of DES in anatomical and clinically high-risk situations. Critics argue that in extreme cases as little as 20% of DES deployed may conform to the evidence-based recommendations. As a consequence, there is close scrutiny of the emerging medium- and long-term safety data of DES to investigate the implications of such widespread and often off-label use DES (table 2).


The introduction of any new medical device/treatment is followed by an intense period of safety monitoring.35 36 DES have not been exempt from this scrutiny. Since their approval by the FDA, multiple studies and analyses have investigated mortality, stent thrombosis and infarction rates in patients receiving DES. However, this process has proved to be much more complex than expected. Differing use of definitions, in particular for mortality (all-cause mortality, versus cardiac mortality, versus non-cardiac mortality) and stent thrombosis (defined using clinical or angiographic criteria, as well as the time of suspected thrombosis: early, late and very late), in conjunction with differences in trial design and data collection, have revealed major inconsistencies in the data and introduced uncertainty within the cardiac community.

Stent thrombosis is thought to be the most important underlying factor contributing to both early and late postintervention death and infarction, and thus has received the greatest level of scrutiny. Stent thrombosis can occur early as well as late (fig 3). As for BMS the rigorous use of periprocedural, dual antiplatelet therapy and intraprocedural heparin and/or glycoprotein IIb/IIIa inhibitors has reduced the rates of early stent thrombosis (<30 days) to ⩽1%. However, the emerging data have raised particular concern over the rates of late stent thrombosis with DES, with delayed recognition of the hazard because initial postmarketing surveillance programmes did not extend beyond 1 year after implantation.37

Figure 3 Different stages of stent thrombosis.48

Initial concern over late stent thrombosis was raised by Camenzind et al.37 38 The authors analysed data on the basis of intention-to-treat from the original SES and PES RCTs, as well as from the latest follow-up information on file at Cordis (for SES) and Boston Scientific (for PES). Despite differences in absolute values between SES and PES, as well as between the two sources of data, their findings demonstrated a consistent trend in all analyses towards more frequent deaths and Q-wave MIs with both types of DES.37 38 These results have been replicated in a meta-analysis directed specifically at the issue of late stent thrombosis by Bavry et al.39 Pooled data from 14 RCTs (6675 patients) demonstrated that the incidence of very late stent thrombosis (>1 year after index procedure) was four- to fivefold higher for DES, at 5.0 events per 1000 DES implanted. There were no events in the same time period in patients receiving BMS (risk ratio 5.02, 95% CI 1.29 to 19.52, p = 0.02).39 Of further interest, there were differences between SES and PES, with more events for PES (median follow-up 18 months) than for SES (median follow-up of 15.5 months).

Nordmann et al have also investigated long-term adverse events in patients with DES (meta-analysis of 17 trials with a total of 8221 patients).38 40 The results of this study did not demonstrate differences in late stent thrombosis between DES and BMS groups. However, in their analysis at 2-year follow-up (4631 patients) non-cardiac mortality was higher in patients who had received a DES (OR = 1.72, 95% CI 1.01 to 2.94).38 40 In a sensitivity analysis the SES emerged as being more likely to be associated with increased non-cardiac mortality,40 calling for longer-term follow-up data on a much larger population of patients to substantiate conclusively potential increases in non-cardiac mortality associated with SES.

In response to these observations investigators and sponsors of multiple RCTs have reanalysed all available original data with a specific aim of examining late adverse events and, in particular, late stent thrombosis.4144 The meta-analysis by Stone et al is one of a number of studies demonstrating a pattern of late but rare complications with DES.41 Long-term outcome was analysed in nine of the original RCTs (four trials (1748 patients) comparing SES with BMS and five trials (3513 patients) comparing PES with BMS) of DES deployed in single, previously untreated coronary stenosis. Overall 4-year rates of stent thrombosis were similar between DES and BMS groups.41 However, between years 1 and 4 both DES were associated with significantly higher rates of stent thrombosis (SES 0.6% vs BMS 0%; p = 0.025 and PES 0.7% vs BMS 0.2%; p = 0.028) than their bare-metal counterparts.41 The overall numbers of deaths and MIs were higher in patients who had received a SES; however, the differences did not reach statistical significance.

Although late stent thrombosis and non-cardiac mortality are both rare events (affecting between 0.5 and 1.0% of patients), their importance is only apparent when placed into the perspective of current real-world interventional practice. Considering that around one million DES are deployed annually, an adverse event rate of 1% will potentially account for 10 000 late events a year, a clinical figure that cannot be ignored. Emerging data clearly indicate that the original RCTs for DES and their subsequent meta-analysis have been significantly underpowered to detect rare adverse events. By design, these RCTs were set up primarily to show the greatest procedural benefit for DES (differences in rates of restenosis) and were not suited to detect infrequent adverse events. Furthermore, the coronary anatomy and comorbid conditions of patients recruited to these studies were also significantly less complex than the current real-world practice, hence further decreasing the chance of detecting rare adverse events.

Recently reported studies investigating medium- to long-term outcome of DES in real-world practice also indicate a greater incidence of late stent thrombosis with DES.45 46 Daemen et al investigated late adverse events in 8146 patients who had received either a SES or PES.45 During a follow-up period of 3 years, 61 cases of late stent thrombosis occurred at a constant rate of 0.6% a year.45 The number of cases of late thrombosis with PES was higher than with SES (PES: 1.8%, SES: 1.4%; p = 0.03). Results also demonstrated that stent thrombosis could occur at any point during the follow-up period, not necessarily triggered by interruption of dual antiplatelet therapy. In the population studied, diabetes and an acute coronary syndrome at presentation were the only independent determinants of late stent thrombosis.45

Lagerqvist et al conducted a similar analysis using data from the Swedish Coronary Angiography and Angioplasty Registry (SCAAR registry, 6033 patients received DES and 13 738 BMS).46 At 6-months’ follow-up a trend towards lower overall unadjusted event rates was seen in patients receiving DES (13.4 events per 1000 patients per year). However, after 6 months, patients with DES had significantly higher overall adverse event rates (death and myocardial infarction; 12.7 events per 1000 patients per year) than patients receiving BMS (relative risk 1.20, 95% CI 1.05 to 1.37). Both adjusted and unadjusted mortality rates were also higher for the DES group. However, figures for stent thrombosis, although numerically higher, did not reach statistical significance in this analysis.46 As with the study of Daemen et al, a concerning trend was a continuous annual 0.5% increase in the risk of death and 0.5-1.0% increase in the risk of MI in the DES group. This trend did not plateau at the end of the study period.45 46

The exact mechanisms resulting in-stent thrombosis remain unclear. Pathoanatomical studies provide some insight into potential events that follow implantation of a DES.37 Unlike BMS, deployment of a DES results in a pronounced local inflammatory process, leading to delayed healing, re-endothelialisation and, in turn, positive remodelling of the injured artery wall. Clinically, these changes have the potential to result in late-acquired stent malapposition and aneurysm formation. Intravascular ultrasound studies have demonstrated late-acquired stent malapposition in 21% of the RAVEL and 10% of the TAXUS-I studies, and late aneurysm formation in 1.4% of patients in the TAXUS-V study. Alterations in local vascular biology of the stented wall segment, combined with increased blood thrombogenicity often brought about by discontinuation of dual antiplatelet therapy between 6 and 12 months after the index procedure, create the ideal milieu for intracoronary thrombus formation and propagation.37 In line with this hypothesis, a number of reports and studies have demonstrated increased rates of late stent thrombosis with reversion to a single antiplatelet agent after DES implantation.37

Table 2 A summary of continuing randomised controlled trials with drug-eluting stents (DES), which may lead to approval in more complex cases47

The absence of a universally agreed definition for stent thrombosis is a further complexity in the investigation of DES complications. In a recent study Mauri et al have examined this subject by examining rates of stent thrombosis using a hierarchical classification of thrombosis (stent thrombosis as defined in the original study protocols and the recent Academic Research Consortium I (ARC I) series of definitions (table 1)) across pooled results from a number of RCTs (878 patients with SES, 1400 patients with PES and 2267 patients with BMS).44 When the protocol definition of stent thrombosis was used, no significant difference was found in the overall incidence of late thrombosis between either one of the DES and their bare-metal counterpart over a 4-year follow-up period. However, after the initial 1-year follow-up period, numerically there were more events in both DES groups. These data are in line with meta-analyses performed by other investigators. However, as assessed by each of the ARC I categories, the cumulative incidence of stent thrombosis throughout the follow-up period was at the very least twice as low as the observed rates in the per-protocol analysis (fig 4).44 This study highlights the need for uniform definitions to facilitate across the board meaningful comparison of studies.

Figure 4 Cumulative incidence of stent thrombosis at 4 years after implantation of FDA-approved drug-eluting stents, according to trial protocol versus Academic Research Consortium I (ARC I; definite or probable categories) definitions. (A and B) Comparison of the incidence of stent thrombosis in patients with sirolimus-eluting stents and paclitaxel-eluting stents, as compared with bare-metal stents according to the definition of stent thrombosis as used in the original randomised controlled trials. (C and D) Data for the same trials with the definition of definite or probable stent thrombosis as recommended by the ARC I consortium. Reproduced with permission from: Mauri L, Hsieh W, Massaro JM, et al. Stent thrombosis in randomized trials of drug-eluting stents. N Engl J Med 2007;356:1020–9.44 Copyright © 2007 Massachusetts Medical Society. All rights reserved.
Table 1 Study protocol and Academic Research Consortium I (ARC I) definitions of stent thrombosis44

Taken together, the overall weight of data clearly provides significant cause for concern and uncertainty about the long-term safety and widespread use of DES. The latest recommendations from the FDA (December 2006) reflect these concerns.47 DES used in the context of their approved indications (on-label use) seem to be safe. In this context the risk of late thrombosis does not outweigh the beneficial effects (reduced restenosis and repeat revascularisation procedures) of DES. In contrast, on the basis of the current evidence, the risk-to-benefit ratio of DES in the context of off-label use is at the best uncertain or in favour of late complications. Future studies are not only required to confirm and quantify this risk, but also to provide clear guidance on the use of DES in real-world practice, granted a larger efficacy gradient in higher-risk subsets.


Medical devices undergo a rigorous period of investigation before being granted a licence. There are often three distinct stages in this period, including preclinical testing (in vitro or animal model studies, or both), detailed and time-consuming clinical trials and postmarketing surveillance, where manufacturers are mandated to monitor both safety and efficacy of the device through real-world registries and robust adverse event reporting systems. The development of DES has been no exception to the rule, with manufacturers strictly adhering to recommendations. However, the unique properties of DES, which combine a medical device with active, dynamic, local pharmacological action, have at every stage proved to be a major challenge to this well-established pathway.35 36

Currently there are no animal correlates of human atherosclerosis where both the mechanical, as well as local, regional and systemic pharmacokinetics of the DES can be tested robustly. Hence the histological characteristics and the injury/healing cycle after DES deployment during preclinical testing in animal models are of limited comparative value to humans.36 Although, the clinical trials investigating the efficacy of DES were conducted rigorously, emerging long-term safety data clearly indicate that these studies were significantly underpowered to detect rare, but vitally important, adverse events. The magnitude of the problem is illustrated by retrospective statistical power calculations for the original DES trials based on the emerging incidence of late adverse events. It is estimated that over 10 000 patients will need to be studied (α = 0.05 (two-sided), power 0.90) in order to detect a difference in late stent thrombosis of 1%.35 The multifaceted postmarketing surveillance of DES, allowing analysis of data from multiple sources including medical device reporting (MDR), claims-based data from a third party, regional and national databases, real-world multicentre registries and device-specific registries, has also failed to identify late adverse events accurately.35 36 As for preclinical testing, none of the aforementioned pathways are suited to monitor simultaneously both the mechanical and the pharmacological properties of DES. The independent collection, storage and fragmented analysis of the postmarketing data, combined with a lack of collaboration between the various sources, have further exacerbated postmarketing problems.

As the significance of rare late adverse events with DES unravels, manufacturers, regulating authorities and independent scientific monitors are evaluating future strategies. Recommendations include the design of larger, adequately powered real-world registries to monitor rare adverse events, integration of the various reporting systems to ensure availability of robust information on low-frequency late events and on denominator values, and finally, distinction between sources of funding for development and monitoring of future devices. It is recommended that funding and organisation of preclinical and clinical studies are best continued by manufacturers. However, the complex postmarket surveillance period should ideally be an independent process, with costs underwritten by government and regulatory agencies.


Despite major concerns over rare late adverse events, DES continue to remain an important part of the interventional cardiologists armamentarium. As outlined above, there is a major concerted drive by scientists, clinicians and manufacturers to move forward an ambitious programme of studies to quantify and investigate the cause of late adverse events in the current generation of DES, and to determine the optimal duration of dual antiplatelet therapy. Data emerging from this international programme should influence the design of a future generation of DES. Manufacturers are already at various stages of investigating the influence of stent strut design, biodegradable polymers and/or stents, new antiproliferative drugs and stents with pro-healing features in animal models or first-in-man studies. Furthermore, it is expected that these new generation drug-device combinations will be subjected to a more stringent and robust monitoring period before patient access will be granted.


NM is British Cardiac Society John Parker Fellow at the Cardiovascular Division, King’s College London School of Medicine at Guy’s, King’s College and St Thomas’ Hospitals, University of London.



  • Competing interests: None declared.