Wish List and REALITY: Title and subTitle BreakChoice of Stents and End Points for Treatment of De Novo Coronary Artery Lesions

In this issue of JAMA, Morice and colleagues¹ present the results of the much anticipated REALITY trial, a prospective, randomized, nonblinded (operator) comparison of the sirolimus-eluting stent and the paclitaxel-eluting stent in patients with de novo coronary artery disease. Among the 1353 patients enrolled, nearly a third had diabetes mellitus, 40% had a previous infarction, and one quarter presented with unstable angina. The study protocol called for the randomization of 1 or 2 lesions (38% of patients) in (surprisingly) relatively small vessels (mean, 2.4 mm) to paclitaxel-eluting stent or sirolimus-eluting stent with independently conducted angiographic and clinical follow-up at 8 and 12 months, respectively. Of note, more than 90% of the patients returned for angiographic follow-up, an unusually high rate for this kind of trial.

With respect to the primary end point of the trial, there was no statistically significant difference in the incidence of in-lesion restenosis, although for some of the secondary end points, sirolimus-eluting stent appeared to perform better. These findings are not particularly surprising. Indeed, in a meta-analysis of 6 sirolimus-eluting stent vs paclitaxel-eluting stent studies (including REALITY),² there was no heterogeneity in results, and sirolimus-eluting stent reduced in-lesion restenosis from 13.1% to 9.3%. This difference reached statistical significance because of the 3-fold larger number of patients analyzed. Sirolimus-eluting stent inhibits neointimal proliferation better than paclitaxel-eluting stent, but this advantage did not translate into a significant reduction in need for repeat target lesion revascularization (TLR). This discordancy may be because of the nonlinear relation between tissue growth and symptoms or angiographically documented restenosis. Because there was no expectation that either stent would prevent myocardial infarction or cardiovascular death, the lack of such a difference also is not unexpected. Nevertheless, the absence of unexpected findings does not negate the importance of a well-performed study and the additional contributions it makes to current collective knowledge.

So is REALITY the reality? Certain aspects of this trial and its context among other studies must be considered before simply concluding that, as usual, more data are needed. Essentially, there is an apparent disparity between the collective “wish list”—ie, what cardiologists want to find out—and what actually can be determined from REALITY.

First, what questions did the study not answer or address? In the fast-paced world of interventional cardiology, trial design needs to adhere to strict criteria and rules and intelligently “guess” what it is that practitioners really want to know. The REALITY trial investigators opted to select patients with relatively small coronary arteries in an attempt to enrich the primary event pool and enhance the likelihood of detecting the prespecified expected difference in the primary outcome, an unrealistic 43%.

While the expected rate of restenosis for sirolimus-eluting stents was reasonably anchored in data from the SIRIUS trial,³ the 14% expected rate for paclitaxel-eluting stents was not based on similarly strong support. In the TAXUS II study,⁴ available at the time REALITY was designed, paclitaxel-eluting stents were associated with, at most, a 4.3% restenosis rate in patients with vessels measuring a mean of 2.78 mm in diameter and an incidence of diabetes of 17%. Was it likely that different enrollment criteria would have changed the results? Probably not, because although larger vessels generally have a lower rate of restenosis regardless of the stent used, the excess of diabetes in REALITY could not be expected to triple the incidence of restenosis.

The choice of the primary end point needs careful consideration. Because angiographic restenosis and TLR apparently track in nonlinear fashion,⁵ clinically indicated or ischemia-driven TLR might have been a better choice than measurement of diameter stenosis. Indeed, only half of the TLR procedures in REALITY were clinically indicated, while the other half were the direct result of protocol-mandated angiography in largely asymptomatic patients. These data are identical to large US postmarketing registries. For example, the 1-year incidence of TLR was 5.4% among 2458 patients treated with paclitaxel-eluting stent in the ARRIVE 1 Registry⁶ and 3.7% with sirolimus-eluting stent in the T-Search Registry.⁷

Certain subgroups for which the REALITY trial could not generate adequately powered definitive data are of particular concern for practitioners choosing between sirolimus-eluting stents and paclitaxel-eluting stents. In patients with diabetes, particularly those with small vessels, a recent analysis from 2 registries suggested that paclitaxel-eluting stent might be superior.⁷ However, patients with long lesions might be at a disadvantage with paclitaxel-eluting stent: data from the TAXUS V⁸ trial suggested that overlapping paclitaxel-eluting stents are associated with a higher incidence of myocardial infarction. A third subgroup are patients with bifurcation lesions requiring double wiring and potential intervention on both branches; these accounted for 40% of the lesions treated in REALITY.

Another aspect to consider is the issue of acute, subacute, and late stent thrombosis. One could argue that any incremental benefit of one particular drug-eluting stent over another for suppression of neointimal proliferation pales in comparison with the profound reduction in need for TLR with use of bare metal stents. An equally profound concern is that prevention of restenosis carries a cost of delayed endothelialization and late thrombosis, particularly after the discontinuation of dual antiplatelet therapy.^{9 - 11} In the REALITY trial, the incidence of stent thrombosis (defined as cardiac death, Q-wave myocardial infarction, or a documented occluded stent) actually tended to be lower with the sirolimus-eluting stent (0.7% vs 1.9%; P = .06). Most of these events occurred while patients presumably received dual antiplatelet therapy. Furthermore, the incidence of Q-wave myocardial infarction, frequently a consequence of stent thrombosis, was significantly higher in the paclitaxel-eluting stent group (1.2% vs 0.1%; P = .02), but data are not provided on whether all of these patients had stent thrombosis.

In a recent meta-analysis of the paclitaxel-eluting stent vs bare metal stents trials, Bavry et al¹² did not find an increased incidence of stent thrombosis up to 12 months after stenting with paclitaxel-eluting stent. In 6 studies comparing sirolimus-eluting stent with paclitaxel-eluting stent in 3669 patients, the incidence of stent thrombosis was 1.4% vs 1.6%, respectively (P = .62).² Data from registries regarding stent thrombosis are difficult to interpret because of the definition of the event and the nature of the follow-up process.

Is there hope, then, that REALITY will eventually fulfill the clinician's wish list? Further analyses and studies are needed to learn about the effect of stent choice in patients with diabetes and those with long or bifurcating lesions. More importantly, extended follow-up over 24 months may answer 2 critical clinical questions: Will the better inhibition of neointimal proliferation with sirolimus-eluting stents eventually result in a clinically meaningful improvement in outcome? And will the choice of stent affect the incidence of late stent thrombosis after discontinuation of dual antiplatelet therapy? Answers to these 2 questions may well determine the specific way these very helpful devices can be used best.