Drug-eluting stents (DES) have largely resolved the problem of restenosis, which is the major limitation of plain balloon angioplasty and bare-metal stenting.1 While several DES platforms have been evaluated in the setting of randomised studies and used in clinical practice, most of the accumulated evidence is related to sirolimus-eluting stents (SES) and paclitaxel-eluting stents (PES).2 These devices are the only DES approved by the US Food and Drug Administration (FDA).3
DES have been linked to a higher risk of late stent thrombosis compared with bare-metal stents (BMS),4,5 a phenomenon that was not identified in the initial trials with short- to mid-term follow-up.6 Furthermore, several studies have suggested that SES and PES may be associated with increased mortality and higher rates of myocardial infarction (MI).4,7–9 Serious concerns have been raised regarding the long-term safety of these DES,10–12 although more comprehensive, patient-based meta-analyses do not justify these concerns.13,14
SES and PES differ importantly with respect to polymer coating and antiproliferative drugs, which may have an impact on the risk of late adverse events associated with these devices. Recently, the results of two meta-analyses suggested that the risk of late thrombosis or death might be different between SES and PES;4,9 however, this difference was inferred from indirect comparisons from trials comparing SES and PES with BMS separately.4,9 Therefore, it remains uncertain whether there are any differences between SES and PES with regard to their long-term safety profile. This uncertainty notwithstanding, a prior meta-analysis including six trials on 3,669 patients followed up for up to one year showed that SES are superior to PES in reducing the risk of restenosis.15 Whether the benefit of SES is maintained beyond this period also remains unknown.
Direct-comparison meta-analysis of randomised trials has the potential to increase the power and improve the precision of treatment effects.16 The availability of individual patient data is the ‘gold standard’ for the analysis of time-to-event or survival data.17 Therefore, we performed a comprehensive meta-analysis with a large emphasis on individual patient data from all clinical trials that have evaluated the long-term outcomes of SES and PES after coronary implantation.
Results
Sixteen randomised trials on a total of 8,695 patients were analysed. The patients were representative of the whole clinical spectrum of coronary artery disease. Individual patient data were available from 11 trials, including 5,562 patients who were followed up for a median of 24.3 months (25th and 75th percentiles: 18.4 and 28.7 months, respectively) in the SES group and 24.3 months (25th and 75th percentiles: 18.3 and 28.5 months, respectively) in the PES group (p=0.51).18–28
Re-intervention, the primary efficacy end-point, was needed in 295 patients in the SES group versus 380 patients in the PES group. Allocation to the SES group was associated with a hazard ratio (HR) for re-intervention of 0.74 (95% confidence interval (CI) 0.63–0.87; p<0.001). There was no significant heterogeneity across trials (p=0.39). The sensitivity analysis yielded HRs that ranged from 0.69 (95% CI 0.59–0.82) to 0.78 (95% CI 0.66–0.92) and were not significantly different from the overall HR (p≥0.58). There was no significant interaction between the treatment effect and inclusion of follow-up angiography in the study protocol (p=0.10). When the analysis was confined to the trials for which individual patient data were available, SES were associated with an HR for re-intervention of 0.72 (95% CI 0.61–0.86; p<0.001). Within the first year, the HR was 0.70 (95% CI 0.57–0.85; p<0.001); after the first year, the HR was 0.79 (95% CI 0.58–1.09; p=0.15). The 30-month probability of re-intervention was 9.5% in the SES group and 12.7% in the PES group.
Stent thrombosis, the primary safety end-point, was observed in 53 patients in the SES group versus 82 patients in the PES group. Allocation to the SES group was associated with an HR for stent thrombosis of 0.66 (95% CI 0.46–0.94; p=0.02). There was no significant heterogeneity across trials (p=0.93). The sensitivity analysis yielded HRs that ranged from 0.55 (95% CI 0.36–0.84) to 0.75 (95% CI 0.51–1.09) and were not significantly different from the overall HR (p≥0.52). When the analysis was confined to the trials for which individual patient data were available, SES were associated with an HR for stent thrombosis of 0.51 (95% CI 0.33–0.80; p=0.003). Within the first year, the HR was 0.64 (95% CI 0.38–1.07; p=0.09); after the first year, the HR was 0.30 (95% CI 0.12–0.72; p=0.004). After the first year, seven patients in the SES group and 21 patients in the PES group incurred stent thrombosis. The 30-month probability of stent thrombosis was 1.2% in the SES group and 2.6% in the PES group. The difference was more evident after the first 12 months.
In the SES group, 169 patients died compared with 173 patients in the PES group. Allocation to the SES group was associated with an HR for death of 0.92 (95% CI 0.74–1.13; p=0.43). There was no significant heterogeneity across trials (p=0.98). The sensitivity analysis yielded HRs that ranged from 0.89 (95% CI 0.72–1.11) to 0.94 (95% CI 0.75–1.16) and were not significantly different from the overall HR (p≥0.85). When the analysis was confined to the trials for which individual patient data were available, SES were associated with an HR for death of 0.92 (95% CI 0.73–1.17; p=0.50). Within the first year, the HR was 1.02 (95% CI 0.73–1.45; p=0.89); after the first year, the HR was 0.84 (95% CI 0.61–1.16; p=0.29). The 30-month probability of death was 6.0% in the SES group and 6.3% in the PES group.
MI occurred in 178 patients in the SES group versus 205 patients in the PES group. Allocation to the SES group was associated with an HR for MI of 0.84 (95% CI 0.69–1.03; p=0.10). There was no significant heterogeneity across trials (p=0.99). The sensitivity analysis yielded HRs that ranged from 0.80 (95% CI 0.64–0.99) to 0.87 (95% CI 0.69–1.10) – not significantly different from the overall HR (p≥0.74). When the analysis was confined to the trials for which individual patient data were available, SES were associated with an HR for MI of 0.81 (95% CI 0.64–1.02; p=0.07). Within the first year, the HR was 0.91 (95% CI 0.71–1.17; p=0.46); after the first year, the HR was 0.45 (95% CI 0.25–0.80; p=0.006). After the first year, 18 patients in the SES group and 36 patients in the PES group incurred an MI. The 30-month probability of MI was 5.3% in the SES group and 7.1% in the PES group. This difference became more evident after the first 12 months. Regarding the composite of death or MI, SES were associated with an HR of 0.86 (95% CI 0.72–1.01; p=0.07).
Discussion
This meta-analysis compared long-term clinical outcomes after implantation of SES versus PES in a large population of patients with various clinical presentations of coronary artery disease. Compared with PES, SES significantly reduced the risk of re-intervention and stent thrombosis with no significant impact on the risk of death or MI.
Three limitations should be acknowledged before commenting on the findings of this study. First, we were able to obtain individual patient data from only two-thirds of the trials. Completeness of patient-level data may increase the accuracy of the analysis. It is, however, reassuring that the treatment effects calculated for the entire population are in accordance with those obtained when individual patient data were analysed.
Second, 10 of the 16 trials included in this meta-analysis had a protocol-mandated follow-up angiography. This may exaggerate the risk of the occulostenotic reflex and lead to an increase in the number of re-interventions, although no significant interaction could be found between this study design feature and treatment effect. In addition, the fact that the difference in the risk of re-intervention between the two DES types persisted even beyond the scheduled time for follow-up angiography (six to nine months) does not support a significant impact of protocol-mandated follow-up angiography on the treatment effect in favour of the SES observed in this meta-analysis.
Third, all trials were open-label due to the impossibility of blinding completely different devices from two different manufacturers. Although all reported events went through a blind adjudication process, these limitations may have had an impact on the evaluation of at least one of the events of interest, namely re-intervention.
SES and PES are the most widely used DES to date. Delayed healing characterised by persistent fibrin deposition, poorer endothelialisation and local hypersensitivity reaction are some of the mechanisms put forward for the explanation of the late occurrence of thrombosis-related events with DES.29 There have been reports that these phenomena are more pronounced with PES than with SES, at least in the presence of overlapping stents.30 We observed that patients treated with SES had a 34% reduction in risk of stent thrombosis relative to patients treated with PES. This finding, coupled with the fact that SES are associated with less late loss than PES,22,23,26,27 does not support the recently reported hypothesis that a greater late loss may have a protective role against stent thrombosis.12 Notably, the risk of both stent thrombosis and MI with PES was most increased after the first year. A different susceptibility to thrombosis after cessation of clopidogrel treatment between the two DES may explain the higher incidence of stent thrombosis with PES. A higher risk of late stent thrombosis with PES versus SES was also recently observed in a registry of a large series of patients.31
Although late stent thrombosis was numerically more frequent with PES, this complication was encountered with both DES types and requires maximum attention to improve long-term safety. These findings may indicate that patients who receive DES require a period of dual anti-platelet therapy longer than that currently recommended;32 this may be particularly important for patients who receive PES.
However, we must acknowledge two factors that may interfere with the results of the analysis of stent thrombosis. First, only recently has there been a strong interest in finding a common definition of stent thrombosis that can be used universally in all DES trials. Although this would constitute an important step forwards for future trials, the value of the retrospective application of new definitions for previously conducted trials has not been proved. Second, although the recommended duration of clopidogrel therapy in each trial was known, we did not have information about the actual length of this therapy and related compliance for individual patients.
Conclusions
SES are superior to PES in terms of a significant reduction in the risk of re-intervention and stent thrombosis. The risk of death was not significantly different between the two DES, but there was a trend towards a higher risk of MI with PES, especially after the first year following the procedure.
This article was originally published in its entirety in the Journal of the American College of Cardiology, 2007;50:1373–80, and is reprinted with permission.