SPIRIT III: Comparison of the Xience V Everolimus-Eluting Coronary Stent System with the Taxus Paclitaxel-Eluting Coronary Stent

The purpose of the Xience V stent was to try and put together a best-in-class drug, enhanced durable polymer and optimized metallic stent backbone. The drug that was used was everolimus, very similar to sirolimus, which is known to have very good properties in terms of low late loss and a very wide toxic-therapeutic ratio, so it's relatively easy to work with. The polymer layer, consisting of a unique durable flouropolymer, has had widespread use in other medical applications, is a very thin, and is also non-adhesive, or non-tacky, so it avoids many of the bonding and webbing problems other polymers have had. As a thin polymer layer, it would theoretically not add much in the way of profile to the stent, and as such it would promote strut endothelialization. It would also tend not to cover side branches and thus lead to few periprocedural myocardial infarctions from side branch compromise, which some thicker polymers have issues with. Finally, there is the stent backbone itself, the Vision Stent, which is a .0032 thick cobalt chromium stent. It is very flexible, low-profile and deliverable, all desirable attributes when treating complex coronary artery disease. From preclinical studies performed by Dr. Renu Virmani in a 14-day rabbit iliac model, it appears as if endothelialization was faster with the Xience V stent compared to either sirolimus-eluting, paclitaxel-eluting or zotarolimus-eluting stents. The endothelium also appeared more functional, based on a series of tests measuring platelet cell adhesion molecule (PCAM, an endothelial cell marker) expression and inhibition of vascular endothelial growth factor (VEGF, a substance made by cells that stimulates new blood vessel formation). Thus, in the SPIRIT III trial, we hoped that Xience V would prove to be a very effective and potentially safer stent, though the relevance of rabbit studies to the human condition is, of course, unknown. Preliminary human studies that were performed in Europe, the SPIRIT I and SPIRIT II trials, seem to support that contention. Of course, Xience V had to be tested in a large-scale randomized trial in the United States for regulatory approval and to confirm these favorable preliminary observations. What was the design of the SPIRIT III trial? SPIRIT III enrolled 1,002 patients with mostly stable coronary artery disease, randomized in a 2 to 1 ratio to the Xience V everolimus-eluting stent or the paclitaxel-eluting Taxus stent (Boston Scientific, Natick, MA). The lesion criteria were very similar to those that are approved for the Taxus stent; that is, lesions up to 28mm in length, and 2.5 to 3.75mm in reference vessel diameter. However, we did allow up to two lesions per patient to be enrolled in this study, with one lesion in two separate vessels. Patients were followed at: 30 days; 8 months (at which time, protocol-driven angiographic follow-up was planned in a cohort of 564 patients); 9 months for the primary clinical endpoint; 12 months; then yearly for 5 years. What did the 9-month data show? There were two major endpoints. The primary endpoint was in-segment late loss. The previous study, SPIRIT II, had shown a statistically significant reduction in its primary endpoint of in-stent late loss. There is a potentially important difference between in-stent late loss and in-segment late loss, since in-segment late loss also includes the edges. The edges are always important to consider, because clinical restenosis can occur either at the edges or within the stent. SPIRIT III was designed for sequential non-inferiority and superiority testing as regards in-segment late loss. We found the Xience V stent was not only non-inferior to the TAXUS stent, but actually was superior in terms of reducing in-segment late loss. Xience reduced late loss by about 50%, from a mean of 0.28 mm with Taxus to 0.14 mm with Xience. It also markedly reduced in-stent late loss, and even though it was not powered for binary restenosis, it nearly showed a reduction in in-segment binary restenosis, from 8.9% to 4.7%, p = 0.07. The intravascular ultrasound (IVUS) substudy verified the angiographic results, demonstrating a 40-50% reduction in volumetric tissue growth within the stent. Late acquired malapposition was infrequent with both stent types. The second major endpoint was termed the major secondary endpoint, though indeed it was really another primary endpoint, because both of the endpoints had to be met for the trial to be considered successful. This was target vessel failure (TVF): death, myocardial infarction (MI) or target vessel revascularization (TVR). In this regard, SPIRIT III was designed for non-inferiority; we did not think we had enrolled enough patients to show superiority. There was a trend towards a reduction in TVF, with 9% (Taxus) versus 7.2% (Xience), about a 21% reduction. Clearly not inferior the numbers were lower with Xience than with Taxus, but still, it did not reach statistical significance for superiority. However, we were pleased to see that when reviewing lesion-specific indicators of clinical risk, in particular, MACE (major adverse cardiac events: cardiac death, MI or target lesion revascularization [TLR]), that there was a statistically significant 44% reduction in MACE at 9 months. This positive finding occurred as a result of a borderline reduction in TLR itself (p = 0.06), almost exactly paralleling the reduction in binary restenosis, which is what one would expect. There were no differences in the rates of death, MI, and stent thrombosis between the two stents, though the composite of cardiac death and MI tended to be slightly lower with Xience, contributing to the lower rate of MACE. Stent thrombosis occurred in 3 patients in the Xience arm, and in no patient who received Taxus. Thus, in summary, we found the Xience V stent, compared to the paclitaxel-eluting Taxus stent, was: Superior in terms of late loss, with a strong trend towards a reduction in binary angiographic restenosis; Superior in terms of reducing volumetric obstruction within the stent, as assessed by IVUS; Non-inferior in TVF, but reduced MACE primarily because of a strong trend toward a reduction in ischemic TLR. Also, when we looked at overall TLR, which is ischemic plus non-ischemic, there was a marked reduction from 9% to about 3.8% with Xience as compared to Taxus. It is important to emphasize that the Taxus stent performed very well in SPIRIT III Xience V was just somewhat more effective. These trial findings are particularly important because they actually mirror the similar findings from the SPIRIT II trial. SPIRIT II was a smaller trial performed in Europe with only 300 patients that compared Xience V to Taxus in very similar lesions and patients. It also showed a significant reduction in MACE at one year. The fact that two consecutive trials essentially demonstrated the same results makes it more likely that the findings are robust. How are drug-eluting stent clinical trial endpoints utilized in trials today? DES endpoints have for many years been standardized, but now are somewhat evolving. The classic time for primary endpoint measurement had been 9 months, with long-term follow-up to 5 years, for the final assessment the safety and efficacy of a drug-eluting stent. The time course for the primary endpoint is now being pushed out to 12 months by FDA, with a significant amount of data also being requested to 2 years before U.S. approval. We have also learned that DES efficacy and safety may or may not overlap. TLR is the purest measure of DES efficacy. Drug-eluting stents were designed to reduce angiographic restenosis, which translates into a reduction in ischemic TLR. We want to be able to accomplish that primary goal safely; that is, with similar rates of death (especially cardiac death), MI and stent thrombosis. Ideally, we would see an improvement in the safety profile with an advanced DES, but at the very least we want to make sure that these potent anti-proliferative drugs aren't causing any harm. The causes of death and MI can be very different than prevention of restenosis, although occasionally restenosis or the procedures required to treat restenosis results in death or MI. The problem is that stent thrombosis is very difficult to accurately define and measure, and death and MI are sufficiently rare that in order to accurately assess these endpoints, a study of 8,000“10,000 patients would be required. Second, long-term follow-up is required. Considering stent thrombosis, there are three periods of interest: up to 30 days, 30 days to 1 year, and greater than one year. The reason to consider these three periods separately is that the mechanisms of stent thrombosis may be different in each. Thrombotic episodes occurring within the first 30 days seem to principally relate to mechanical causes such as suboptimal stent implantation or lack of expansion, and possibly clopidogrel resistance. We're still learning about what the causes actually are. In the 30-day to one-year period, so-called late stent thrombosis episodes may relate to different factors, such as ongoing polymer inflammation with lack of healing, strut fracture and antiplatelet resistance. The same factors may or may not be important beyond one year in producing very late stent thrombosis, a period in which most restenosis has already occurred. These adverse events are infrequent, and large studies are required to accurately assess their incidence. Unfortunately it can be hard to differentiate late sudden or cardiac death due to underlying heart disease or comorbidities from very late stent thrombosis. How are patients chosen for angiographic follow-up? It's usually done randomly in a consecutive series of patients. In SPIRIT III, routine angiographic follow-up was pre-specified for the first 564 patients that were enrolled in the study. How did the angiographic and clinical results compare in SPIRIT III? They were actually quite consistent. Follow-up angiography is performed to provide precise quantitative measures to explain the observed clinical outcomes. The primary angiographic finding of interest is the degree of renarrowing with the stent, variably measured as late loss, follow-up diameter stenosis, and binary restenosis. It is then important to examine effects within the stent versus at the edges outside the stent, and determine whether the pattern of restenosis is focal or diffuse. This translates clinically into ischemic TLR. We also look angiographically for adverse effect, such as aneurysm formation, vessel obstruction and thrombus formation. These low-frequency events would translate more into clinical safety endpoints, such as death, MI or stent thrombosis. What can we say about Xience V safety? To really have confidence on safety, given how rare such adverse events are, we really need data on 5,000“10,000 patients, with long-term follow-up to 2 to 5 years. It will be some time before we have this level of information on Xience. What we can say now is that from approximately 1300 randomized patients in SPIRT II and SPIRIT III followed to 9-12 months, the Xience V stent results in similar rates of death, MI and stent thrombosis compared to the Taxus stent. Importantly, with both stents, the rates of stent thrombosis were low out to 9 months. Overall rates of major adverse cardiac events are approximately 50% less with Xience compared to Taxus, though this difference was primarily driven by fewer revascularization events, with weak, nonsignificant trends for less cardiac death and non Q-wave MI with Xience. Buttressing these clinical results to date, Xience V has also demonstrated safety in preclinical experimental models. In porcine implants ranging from follow-up duration from 12 month to several years, the stent is well-healed, with low levels of inflammation, and complete endothelialization. Renu Virmani has reported that Xience V appeared to be the safest of 4 stents studied in a rabbit iliac model. All that is favorable, but how the results from a 14-day rabbit iliac artery model translate to a person is unknown. Until we really have data in thousands of patients followed for years, we can't be certain as to the ultimate safety profile, but at this point there are no signals to suggest there are any concerns. What data were presented at EuroPCR in May? At EuroPCR as a late-breaking trial we presented a combined patient-level pooled meta-analysis of SPIRIT II and III, and the 3-year data from the small SPIRIT First trial. The most important lesson from SPIRIT First was that no stent thromboses occurred at 3 years with the Xience V stent in 26 patients. In the SPIRIT II and III meta-analysis, which examined the 9 month outcomes of Xience V compared to Taxus in 1,302 randomized patients, Xience V resulted in: A 53% reduction in TLR (5.1% with Taxus vs. 2.4% with Xience V, p = 0.01); Similar rates of cardiac death (0.76% vs. 0.35% respectively, p = 0.32); Similar rates of MI (2.7% vs. 1.7% respectively, p = 0.22); Similar rates of stent thrombosis (0.25% vs. 0.45%, p = 0.29); Reduced rates of MACE (8.0% vs. 4.1% respectively, p = 0.004). These data confirm the clinical benefit of the Xience V stent in a large patient population at 9 months. Follow-up to 5 years is ongoing to assess the durability of these results. What lies ahead for SPIRIT III data presentations? At TCT in October, we'll present one-year data from SPIRIT III, the stent thrombosis definitions re-adjudicated by the Academic Research Consortium (the ARC definitions), and data for different subgroups. Those will be the major rollouts of these data later this year. Dr. Stone can be contacted at gstone @ crf.org

NULL