Interventional Cardiology Unit
San Raffaele Scientific Institute
Division of Cardiology
Department of Medicine
University of Cape Town, South Africa
Interventional Cardiology Unit
San Raffaele Scientific Institute
Interventional Cardiology Unit
EMO Centro Cuore Columbus
The introduction of stents that elute drugs to inhibit restenosis was met with considerable enthusiasm by cardiologists, physicians and patients. Although drug-eluting stents (DES) have not completely eliminated restenosis, they have resulted in substantial reductions in the degree of neointimal hyperplasia and the rates of angiographic restenosis and repeat revascularisation. Recently, the question of whether DES are safe has become an area of considerable interest and controversy with the publication of a relatively small randomised trial (BASKET-LATE),(1) an observational study(2) and a meta-analysis,(3) reporting increased rates of stent thrombosis (ST), myocardial infarction (MI) and death with DES compared with bare-metal stents (BMS).
The “eye of the storm” was the presentation of a meta-analysis at the European Society of Cardiology meeting in Barcelona in September 2006, which suggested an increase in the risk of death and MI following implantation of sirolimus-eluting stents.(4) These studies made for sensational headlines in the lay press and even had large-scale impact on DES sales around the world.
However, these trials were limited by small sample sizes, by limited duration of follow-ups, by lack of access to source data or by using landmark analyses that excluded events in the first six months. In the past year there has been a flood of new studies and information published. A Medline search with the keywords “drug-eluting stent thrombosis” found 462 papers published since 2002, with 60% (276) of them since September last year. In the face of this avalanche of new information, we have selected the most appropriate studies that provide a current appraisal of DES safety and have highlighted the issues important in interpreting the available data.
The strongest research evidence that we have in medicine on which to base our decisions is large, randomised, controlled trials. However, the randomised trials comparing DES with BMS were not designed to look at rare endpoints such as ST and are thus underpowered to make these comparisons. In these situations, our next best source of evidence comes from systematic reviews (including meta-analyses) of multiple randomised trials. However, not all meta-analyses are created equal, and performing the analysis using only study-level (published) data rather than patient-level data can result in unreliable results with artificial significance.
Pooled meta-analysis of trials
Stone and colleagues have performed a unique, patient-level, pooled meta-analysis of four pivotal trials that randomised 1,748 patients to either sirolimus-eluting Cypher stents (SES, Cordis, Miami Lakes, Florida, USA) or BMS; and the five pivotal trials that randomised 3,513 patients to either paclitaxel-eluting Taxus stents (PES, Boston Scientific, Natick, Massachusetts, USA) or BMS.(5) The analysis performed by Stone et al demonstrated that the overall incidence of ST at four years (as defined in the study protocols) does not differ significantly between DES & BMS (1.2% with the SES vs 0.6% with BMS, p = 0.20; 1.3% with PES vs 0.9% with BMS, p = 0.30). However, the time distribution of these thrombotic events appears to differ, and both SES and PES are associated with a small but significant increase in the incidence of late ST between one and four years after implantation (0.6% with SES vs 0% with BMS, p = 0.025; 0.7% with PES vs 0.2% with BMS, p = 0.028).
The rates of death or MI did not differ significantly between the groups with DES and those with BMS. It is uncertain why these increased rates of ST, seen more than one year after DES implantation, did not translate into higher rates of death or MI in the randomised studies. It may be that the sample size in these studies was not large enough to detect small differences between the treatment groups in a relatively rare event such as ST. An additional and somewhat overlooked alternative explanation may be that the increased rates of death or MI due to DES thrombosis may have been offset by a reduction in adverse events associated with in-stent restenosis and repeat revascularisation.(6)
There is increasing evidence that in-stent restenosis is not as benign a process as previously thought, and may present as an acute MI in up to 10% of cases.(7,8) It is also important to note that the protocol definition of ST in most trials censored thrombotic events that occurred after target-lesion revascularisation (TLR). Since patients with BMS are more likely to require reintervention for restenosis, thromboses occurring in these patients are censored more frequently, thus lowering the thrombosis rate after BMS implantation and introducing a bias against DES.
A second patient-level pooled analysis by Mauri et al,(9)using the Academic Research Consortium (ARC) definition of ST,(10) included the thrombotic events occurring after the treatment of restenosis, and found no differences in the overall incidence of ST between DES and BMS during four years of follow-up.
Kastrati et al(11) performed a different analysis of individual data on 4,958 patients enrolled in 14 randomised trials comparing SES and BMS. In this study, the risk of death (HR, 1.03; 95% CI, 0.80 to 1.30) or the combined risk of death or MI (HR, 0.97; 95% CI, 0.81–1.16) was not significantly different for SES vs BMS. SES was associated with a sustained reduction in the need for reintervention up to five years after implantation. The overall risk of ST was not different between SES and BMS (HR, 1.09; 95% CI, 0.64–1.86), although there was a slight increase in ST with SES (0.6% vs 0.05%; p = 0.02) after the first year. However, five cases of ST that occurred in the BMS group after TLR were censored as per the study definitions of eight of the original trials and not included in this analysis.
The largest meta-analysis of DES vs BMS and SES vs PES was recently published in The Lancet. Stettler et al(12) performed a network analysis with a mixed-treatment comparison of 38 randomised trials (18,023 patients) with a follow-up of up to four years. The safety endpoints assessed in the study included mortality, MI and definite ST, while the effectiveness outcome was TLR. Importantly, the triallists included secondary ST occurring after a patient had undergone a TLR. This analysis again confirmed that DES and BMS were associated with similar rates of overall and cardiac mortality. SES were associated with the lowest risk of MI (HR, 0.81; 95% CI, 0.66–0.97, p = 0.03 vs BMS and HR, 0.83; 95% CI, 0.71–1.0, p = 0.045 vs PES). Thus, about 100 patients will have to be treated with a SES instead of a BMS or PES to prevent one MI over four years. DES were not associated with an increased risk of definite ST from 0 days to four years. PES, however, were associated with an increased incidence of late ST (after 30 days) compared with BMS (HR, 2.11; 95% CI, 1.19–4.23, p = 0.02) and SES (HR, 1.85; 95% CI, 1.02–3.85, p = 0.04).
The study also confirmed the efficacy of DES in markedly reducing TLR, which is maintained for up to four years. In essence, about seven patients will have to receive SES or eight patients PES, instead of BMS, to prevent one TLR at four years.
Although these meta-analyses further confirm the long-term safety and effectiveness of DES, it must be remembered that they cannot replace multicentre randomised trials. They are limited by the choice and quality of the data and have been proven wrong one-third of the time by subsequent large randomised controlled trials.(13) Also, the complex network analysis performed by Stettler and colleagues may be more prone to error than a routine meta-analysis as different treatments are compared not directly but by statistical inference. Some critics would argue that in “real-world” practice most DES are implanted into high-risk lesions and/or patients not included in the RCTs (also known as off-label use), and that in these situations the risk of ST may be higher. In these situations, data from large observational registries can be helpful.
Registry studies in Europe
In Northern Europe, two very large registry studies have been performed and analysed to compare DES with BMS.(14,15) In the Western Denmark Heart Registry, 12,395 consecutive patients treated with DES (n = 3,548) and BMS (n = 8,847) were followed up for 15 months after stent implantation.(14) All-cause and cardiac mortality were similar between the DES and BMS group. There was also no difference in cumulative MI (DES: 3.2% vs BMS: 3.0%, p = 0.65) or the risk of definite ST between the two groups (DES: 0.65% vs BMS: 0.61%, p = 0.78). A difference in MI or ST between the two groups was only found when an analysis was performed of events occurring between 12 and 15 months. Once again, DES resulted in a marked (43%) reduction in TLR (4.6% vs 7.1%; p < 0.001). An important caveat of this study is that, in the majority of patients, dual antiplatelet therapy was recommended for a year. The Swedish Coronary Angiography and Angioplasty Registry (SCAAR) study group recently presented an analysis at the European Society of Cardiology meeting in Vienna of a much larger patient cohort (n = 35,262) than they previously published (n = 19,771) and with an additional year of follow-up.(15) Contrary to their previous findings, the study showed no difference in mortality up to four years between the 13,785 patients receiving at least one DES and 21,477 patients who received only BMS. These large registry studies are reassuring of the safety of DES and confirm the data from randomised trials. However, it must be remembered that these studies are observational and at best are hypothesis-generating.
Now that the dust is settling on the controversy surrounding DES, what can we conclude about DES safety and efficacy? First, there is no uncertainty about the efficacy of DES in reducing restenosis and TLR; they have been proven to be superior to BMS in every lesion subset in which they have been tested, and these benefits are sustained for up to four years of follow-up. Secondly, there appears to be no difference in the overall risk of death, MI or ST between DES and BMS up to four years after stent implantation, but the timing of the ST is different. DES thrombosis tends to occur later more commonly than it does with BMS, although BMS thrombosis can also occur late.(16) Landmark analyses that exclude events occurring within the first 6–12 months are the only ones that have shown an increase in MI and death with DES. However, these types of analyses penalise DES and neutralise their beneficial effect on reducing restenosis and thereby MI. The definition of the ARC is an important step toward having a uniform definition for the reporting of ST.(10) However, the probable and possible ST definitions introduce a bias and subjectivity into the adjudication of this contentious event. Although ST may be asymptomatic, when it is symptomatic it results in MI and/or death, and thus it is these two clinical endpoints that are the most important in assessing DES safety. Also, despite the important features of the work performed by the ARC, there is an area of concern when we apply this definition in patients with multivessel coronary artery disease. There is the risk to label any hard endpoint (MI and death) as ST even if it is well known that other causes come into play. Any MI in a patient with stents in the three major coronary vessels will become a “probable ST”.
Durable clinical efficacy
In conclusion, no matter what type of data you look at (registries, meta-analysis, randomised studies), there appears to be no significant difference in MI and/or death between DES (SES or PES) and BMS, up to four years of follow-up. These data confirm the durability of the clinical efficacy for DES with a marked and persistent reduction in TLR compared with BMS that is sustained up to four years after implantation. These encouraging conclusions are welcome but should not shadow problems still present in patients treated with DES. The tight dependence on clopidogrel therapy and the uncertainties regarding the appropriate duration are two major shortcomings that will need to be addressed.
- Pfisterer M, et al. J Am Coll Cardiol 2006;48(12):2584-91.
- Lagerqvist B, et al. NEJM 2007;356(10):1009-19.
- Nordmann AJ, et al. Eur Heart J 2006;27(23):2784–814.
- Camenzind E, et al. European Society of Cardiology Congress 2006, Barcelona, Spain. 2006 Sep. Available from:http://www.escardio.org/knowledge/congresses/CongressReports/hotlinesandctus/707009_Camenzind.htm
- Stone GW, et al. NEJM 2007;356(10):998-1008. 6.
- Farb A, et al. NEJM 2007;356(10):984-7.
- Chen MS, et al. Am Heart J 2006;151(6):1260-4.
- Nayak AK, et al. Circ J 2006;70(8):1026-9.
- Mauri L et al. NEJM 2007;356(10):1020-9.
- Cutlip DE, et al. Circulation 2007;115(17):2344-51.
- Kastrati A, et al. NEJM 2007;356(10):1030-9.
- Stettler C, et al. Lancet 2007;370(9591):937-48.
- LeLorier J, et al. NEJM 1997;337(8):536-42.
- Jensen LO, et al. J Am Coll Cardiol 2007;50:463-70.
- James S. Presented at the European Society of Cardiology Congress 2007, Vienna, Austria. 2007 Sep. Available from: http://www.escardio.org/knowledge/congresses/CongressReports/2007/hl-ctu…
- Rosas Ramos A, et al. Catheter Cardiovasc Interv 2007;70(2):229-32.