Effectiveness of Drug-Eluting Stents in Real-World Patients

Drug-eluting stents are a major advance in percutaneous coronary intervention (PCI) because they reduce the need for repeat target lesion revascularization several-fold compared with bare-metal stents. Early randomized clinical trials separately testing sirolimus-eluting stents^{1 - 2} and paclitaxel-eluting stents^{3 - 4} against their bare-metal counterparts demonstrated that rates of target lesion revascularization were reduced from approximately 10% to 15% to approximately 4% to 5%, whereas the rates of death, myocardial infarction, and stent thrombosis were similar between bare-metal and drug-eluting stents. Because these studies were performed for device regulatory approval, they enrolled patients and lesions with relatively low risk for restenosis and adverse events. Subsequent head-to-head trials of sirolimus- and paclitaxel-eluting stents in cohorts with a broader risk profile have provided inconsistent results; some trials have suggested equivalent rates of target lesion revascularization for these stents,^{5 - 6} whereas others have reported superiority of sirolimus-eluting stents.^{7 - 9}

Initial clinical trials in any area of medicine are primarily designed to test efficacy (ie, is the drug or device beneficial under ideal conditions?) rather than to test effectiveness or relevance (ie, is the drug or device beneficial in everyday practice?).¹⁰ Phase 4 (postapproval) studies are expected to be conducted to help bridge this gap while monitoring safety. The sirolimus-eluting stent was the first to become clinically available, and early postapproval data from the Rapamycin-Eluting Stent Evaluated At Rotterdam Cardiology Hospital (RESEARCH) registry showed that the unrestricted use of sirolimus-eluting stents in the real world was safe and effective in reducing target lesion revascularization as well as major adverse cardiac events at 1 year compared with bare-metal stents—findings similar to the regulatory approval studies.¹¹ Subsequently, observational data from the Taxus-Stent Evaluated At Rotterdam Cardiology Hospital (T-SEARCH) registry suggested that the unrestricted use of paclitaxel-eluting stents was safe and associated with a similar adjusted outcome when indirectly compared with sirolimus-eluting stents.^{12 - 14} There was an “inferior trend” in crude outcome observed with paclitaxel-eluting stents—although this finding was attributed to higher-risk characteristics among patients receiving such stents,¹² since practice had become comfortable with a wider application of drug-eluting stents over time.

Thereafter, a number of small, direct-comparison trials were reported, with most individually observing a similar outcome for patients receiving sirolimus- vs paclitaxel-eluting stents. On the other hand, a meta-analysis of 16 such randomized head-to-head trials determined sirolimus-eluting stents to be superior to paclitaxel-eluting stents with regards to target lesion revascularization and stent thrombosis.¹⁵ Likewise, a collaborative network meta-analysis of 38 trials including 18 023 patients and having a follow-up of up to 4 years also suggested that sirolimus-eluting stents provided a clinical advantage over paclitaxel-eluting stents.¹⁶

The discordance in findings among some trials and subsets of patients (eg, those with diabetes) has been confusing and challenging to interpret. In this issue of JAMA, Galløe and colleagues¹⁷ aim to add clarity and relevance to the issue by comparing outcome among patients randomized to receive sirolimus- vs paclitaxel-eluting stents in a study design reflecting everyday clinical practice by including unselected patients and by using symptom-driven clinical end points. They report in this large-scale practical randomized trial that there were no significant differences in clinical outcomes between patients receiving sirolimus- and paclitaxel-eluting stents.¹⁷ Their preliminary data were included in the recent meta-analyses^{15 - 16} and can now be further examined. A noteworthy feature of the study by Galløe et al is that it is the largest randomized direct comparison of these drug-eluting stents; yet with low event rates and the challenges presented by enrollment of all-comers, it has several limitations that need to be considered.

First, the study randomized less than a third (27.7%) of potentially eligible patients, which suggests that the cohort may not be as unselected as the authors intended. This raises some concern of selection bias and generalizability of the results and is compounded by the observation that clinical outcomes were worse among patients who were eligible for enrollment but not included in the study—again, suggesting a potential selection bias and possibly reducing the power to observe a meaningful difference between stents. Nevertheless, since there were no stringent inclusion and exclusion criteria for this study, the results should be more reflective of real-world patient outcome than those from clinical trials.

Second, the authors' observation that there were no significant differences between sirolimus-eluting and paclitaxel-eluting stents merits closer inspection. As Galløe et al point out, the study was underpowered due to the small sample size coupled with relatively low event rates; there was less than a 30% chance to detect a meaningful difference in major adverse cardiac events between the 2 drug-eluting stents if a difference existed. As a secondary end point, the authors did observe a statistically nonsignificant 28% lower rate of target lesion revascularization with sirolimus-eluting stents (hazard ratio [HR], 0.72; 95% confidence interval [CI], 0.48-1.07; P = .10). The lack of statistical significance is more likely a type II error than a lack of true difference in rates of target lesion revascularization between the stents, because the authors' findings are consistent with data reported in the recent meta-analyses by Schömig et al¹⁵ (8695 patients; HR, 0.74; 95% CI, 0.63-0.87; P < .001) and by Stettler et al¹⁶ (18 023 patients; HR, 0.70; 95% CI, 0.56-0.84; P = .002).

Third, Galløe et al report no difference in rates of stent thrombosis between sirolimus-eluting stents (2.5%) and paclitaxel-eluting stents (2.9%) (HR, 0.87; 95% CI, 0.52-1.46; P = .60). These absolute rates, however, seem higher than would be expected with bare-metal stents (approximately 1.6%)¹⁸ or drug-eluting stents in a large registry¹⁴ and reinforce the need to explore ways to reduce this often catastrophic event, including meaningful preprocedural patient-physician discussion regarding the feasibility of administering dual antiplatelet therapy for at least 1 year. In short, while this study alone cannot clarify the superiority of one drug-eluting stent over another in everyday practice, the overall data presented add support to the safety and effectiveness of such stents. The higher-than-anticipated stent thrombosis rate did not appear to translate into a higher-than-expected mortality rate.

In 2008, clinicians will have additional choices of drug-eluting stents with the availability of second-generation devices—namely, the everolimus-eluting stent,¹⁹ which yielded similar or fewer major adverse cardiac events among patients as compared with the paclitaxel-eluting stent, and the zotarolimus-eluting stent,²⁰ which was shown to be noninferior to the paclitaxel-eluting stent. The ongoing choice of a drug-eluting stent will likely depend on multiple factors that will include safety, effectiveness, deliverability, and—given recent cuts in reimbursement—cost of the device. The current literature for drug-eluting stents can be challenging to interpret because of differing criteria for study enrollment, definitions for acute stent thrombosis and other clinical end points, and varied intervals of dual antiplatelet therapy and follow-up after stent implantation.²¹ Similarly, current real-world registries are usually limited by lack of valid control groups and often use historical controls.²¹ A large longitudinal database for patients receiving these various drug-eluting stents with open entry to fully capture all procedures may help determine the safest and most effective revascularization practice possible and should help guide future recommendations.²¹