Review of the SPIRIT IV Trial
Everolimus-Eluting versus Paclitaxel-Eluting Stents in Coronary Artery Disease
N Engl J Med 2010;362:1663-74.
Background: The RAVEL and TAXUS-IV trials compared the sirolimus- and paclitaxel-eluting 2nd generation stents to 1st generation bare metal stents. Both trials reported improvements in surrogate endpoints - “in-stent luminal loss” was the primary endpoint of RAVEL and “ischemia-driven target-vessel revascularization” was the primary endpoint of TAXUS-IV. Neither trial showed differences in hard endpoints like death or MI but were not powered for such events.
The observation that restenosis still occurred with 2nd generation stents drove interest in developing newer stent technology with improved bioavailability and drug delivery. The 3rd generation everolimus-eluting stent was felt to represent such a development but like its predecessors had only been tested in experiments using surrogate endpoints that were not driven by clinical symptoms.
Thus, the SPIRIT IV trial sought to test the hypothesis that 3rd generation everolimus-eluting stents would reduce patient-driven clinical outcomes compared 2nd generation paclitaxel-eluting stents. Furthermore, it was designed to be large enough to provide data on important subgroups, especially patients with diabetes.
Patients: Limited details are provided about inclusion and exclusion criteria in the main manuscript and readers are directed to a previous publication and supplemental appendix. Lesion characteristics had to be less than 28 mm in length with a reference-vessel diameter between 2.5 to 3.75 mm. Patients were excluded if they had features making them complex from either a clinical or angiographic standpoint.
*Note to learners: Be especially skeptical of trials that do not include at least an abridged version of important inclusion and exclusion criteria in the main publication manuscript. This often indicates that the criteria are complex and that patients are highly selected, which limits the generalizability of the findings to routine practice.
Baseline characteristics: The average age of patients was 63 years and 68% were men. Approximately 32% of patients had diabetes with about one quarter being insulin-dependent. Over 20%of patients smoked and a similar percentage had a previous heart attack.
Three quarters of patients had 1 target lesion, 22% had 2 target lesions and 3% had 3 and 11% of patients had 1 or more complex lesions. The average lesion length was 15 mm, reference-vessel diameter was 2.75 mm, minimal luminal diameter was 0.75 mm, and average % stenosis was 72%.
Procedures: Patients were randomized in a 2:1 ratio to receive an everolimus- or paclitaxel-eluting stent. They were stratified based on having diabetes or not, whether they had a single or complex lesion, and study site. Operators were not blind to the stent being used.
At least 300 mg of aspirin was administered before catheterization and at least 300 mg of clopidogrel was recommended before the procedure and was required within 1 hour after stent implantation. Patients took at least 80 mg of aspirin daily for an indefinite period and 75 mg of clopidogrel for at least 12 months. Clinical follow-up visits were scheduled at 30, 180, 270, and 365 days and yearly through 5 years.
Endpoints: The primary end point was ischemia-driven target lesion failure at 1 year defined by the composite of cardiac death, target-vessel myocardial infarction, or ischemia-driven target-lesion revascularization. As was the case in the TAXUS-IV trial, “ischemia-driven” did not necessarily mean “symptom-driven”. Two major secondary endpoints were also prespecified which included ischemia-driven target-lesion revascularization and the composite of death or target-vessel MI.
The trial was powered for sequential testing of noninferiority and superiority for both the primary and 2 major secondary endpoints. The criteria for noninferiority would be met if the upper limit of the 97.5% confidence interval was not more than 3.1%. This was based on an assumed 1 year target-lesion failure rate of 8.2% for both groups. The trial had 90% power to show non-inferiority. Superiority testing was prespecified if the criterion for noninferiority was met. It was estimated that 3690 patients would have 90% power to detect a 2.9% absolute reduction in the primary end point, at a two-sided alpha of 0.05.
The trial also had 90% power to test noninferiority for ischemia-driven target-lesion revascularization and the composite of cardiac death or target-vessel MI at a 2.1% margin. It had 90% and 91% power to test for superiority of these endpoints if noninferiority was met.
*Note to learners: The statistical analysis plan for this trial demonstrates 2 important concepts in hypothesis testing. First, trials can be powered in a prespecified manner for non-inferiority and superiority testing. Second, trials can be powered for prespecified hypothesis testing of more than just a single endpoint.
Results: Patients were enrolled over a 2 year period from 66 U.S. sites. There were a total of 3,687 patients included in the final analysis with 2,458 in the everolimus-eluting stent group and 1,229 in the paclitaxel-eluting stent group. There were some significant differences for patients receiving everolimus-eluting stents that included the number of stents per lesion, total stent length per lesion, the ratio of stent length to lesion length and the maximum pressure used.
At 1 year, everolimus-eluting stents met non-inferiority for the primary and major secondary endpoints and met superiority for 2 of 3. Everolimus-eluting stents reduced the primary endpoint of target-lesion failure (4.2% vs 6.8%; RR 0.62; 95% CI 0.46 to 0.82) and the major secondary endpoint of ischemia-driven target lesion revascularization (2.5 vs 4.6%; RR 0.55; 95% CI 0.38 to 0.78) but not the other major secondary endpoint of cardiac death or target-vessel MI (2.2% vs 3.2%; RR 0.69; 95% CI 0.46 to 1.04).
Differences in target-lesion failure were driven by statistically significant reductions in target-lesion revascularization (2.5% vs 4.6%) as well as MI (1.9% vs 3.1%) but not all-cause (1.0% vs 1.3%) or cardiac death (0.4% vs 0.4%). Stent thrombosis was also significantly reduced but rates were very low in both groups and the trial was not powered for this endpoint.
Interestingly, subgroup analysis of the primary endpoint revealed a statistically significant interaction for treatment efficacy in patients with diabetes such that diabetics did not appear to benefit from everolimus-eluting stents (6.4% vs 6.9%) compared to non-diabetics (3.3% vs 6.7%; p for interaction = 0.02).
Conclusions: In patients with stable CAD who underwent generally non-complex PCI procedures, 3rd generation everolimus-eluting stents compared to 2nd generation paclitaxel-eluting stents reduced a composite endpoint of ischemia-driven target-lesion failure by 38% with a number needed to treat of approximately 40 patients. This was associated with statistically significant reductions in nonfatal MI with a NNT of approximately 100 patients and ischemia-driven target lesion revascularization with a NNT of approximately 50 patients. Everolimus-eluting stents did not reduce death.
There was an interaction noted for diabetic patients who did not appear to significantly benefit from everolimus-eluting stents. Notably, diabetics exhibited more severe angiographic disease with a higher prevalence of multivessel disease, diffuse plaque burden, and a greater likelihood of left main coronary artery involvement. This subgroup finding along with the highly selected nature of the study cohort reduces our confidence that the 3rd generation everolimus-eluting stent confers significant advantages over 2nd generation stents for many patients who receive them in clinical practice.
One final consideration is that the trial was single blinded and operators were aware of stent type which could have biased their performance and the study results.