No Increased Death, MI With DES in Randomized Trials and Registries, "Mega"-Meta-Analysis Suggests

Shelley Wood

March 29, 2008

March 29, 2008 (Chicago, IL) - In what's being dubbed a "mega-meta-analysis," researchers who combined all known randomized trials and registry studies of Cypher and Taxus drug-eluting stents (DES) vs bare-metal stents say the newer devices are just as safe as bare-metal stents, even when used off label. Indeed, in the combined registry analysis, which included more than 174 000 patients, mortality was actually reduced, by 20%, in DES-treated subjects.

In an interview with heartwire , Dr Ajay Kirtane (Columbia University, New York, NY), one of the lead investigators on the new analysis, emphasized that while the mortality reduction suggested in the registry data is likely not so striking outside of registry studies, where patient selection bias may skew results, it is at least clear that mortality with DES is not increased, as earlier studies hinted. Indeed, wildly divergent registry results presented or published over the past 18 months have been largely responsible for stoking the controversy over DES safety.

"It's pretty clear that for each of the three end points we looked at--mortality, MI, and target vessel revascularizations [TVR]--in both the randomized controlled trials [RCTs] and the registry studies, and in looking at a lot of different subsets, there is no obvious signal of adverse safety outcomes associated with DES," Kirtane commented.

Others who reviewed the data for heartwire had contradictory views as to the significance and weight of the results.

Largest-ever DES meta-analysis

Kirtane presented the results during a satellite symposium held on the eve of the American College of Cardiology (ACC) 57th Annual Scientific Session; the study was not a part of the official ACC program, which undergoes a process of peer review by the ACC planning committee, but rather an industry-sponsored CME session. Co–principal investigator Dr Gregg Stone (Columbia University) will be presenting the same data at other satellite sessions during the ACC meeting. But both Kirtane and Stone emphasized that they opted to present these data at a sponsored symposium because their data were finalized after the ACC abstract deadline had passed.

"This was literally a matter of timing," Kirtane said. "We felt this is such a hot item, we wanted to get these data out there as quickly as possible, rather than wait until the next major US meeting."

In all, the study includes information on more than 180 000 patients from 52 studies, with clinical follow-up extending out three to four years.

The aim of the analysis was in part to explore whether DES are as safe in higher-risk, "real-world" patients as they are in the regulated randomized clinical-trial environment and in part to see whether reductions in TVR seen with DES in the trial setting are as "robust" in the real world, where routine angiographic follow-up is not the norm.

Kirtane and Stone combined all studies with at least 100 patients for which mortality data were reported and that had at least one year of clinical follow-up. A fixed-effects model was used as the primary model unless significant heterogeneity between studies was present, in which case a random-effects model was employed.

They report that in 22 randomized controlled trials, including almost 10 000 patients, mortality and MI rates trended lower among DES-treated patients and TVR was significantly reduced by 55%. In the 30 registry studies, in which more than 174 000 patients received either a DES or bare-metal stent, mortality was significantly reduced by 20%, MI was reduced by 11%, and TVR was reduced by 47% in DES-treated patients.

"The favorable results of DES from the randomized controlled trial and registry analysis populations were robust and consistent for both on-label and off-label use and for clinical follow-up extending to three to four years," Kirtane concluded. "These findings strongly suggest that DES are safe for both on-label and off-label use and have comparable efficacy in both RCTs and in the 'real-world.' "

All-cause mortality
Study type Patients, n Trials, n Relative risk p
RCT: all 8867 21 0.97 0.72a
RCT: on-label 4818 10 1.05 0.69a
RCT: off-label 4049 12 0.84 0.24a
Registries 161 232 28 0.80 <0.001b
a. Fixed-effects model
b. Random-effects model
Study type Patients, n Trials, n Relative risk p
RCT: all


20 0.94 0.54a
RCT: on-label 4318 9 1.03 0.82a
RCT: off-label 4532 12 0.77 0.19b
Registries 129 955 24 0.89 0.023b
a. Fixed-effects model
b. Random-effects model Target-vessel revascularization
Study type Patients, n Trials, n Relative risk p*
RCT: all 7291 16 0.45 <0.001
RCT: on-label 4618 9 0.53 <0.001
RCT: off-label 2673 8 0.38 <0.001
Registries 73 819 17 0.53 <0.001
*Random-effects model

Confidence or obfuscation?

Commenting on the analysis for heartwire , Dr Bonnie Weiner, president of the Society for Cardiovascular Angiography and Interventions, pointed out that the analysis reaffirms the benefit of DES in reducing TVR, something established in previous research. "Now it is also clear that this does not come at the price of increased death or MI, and in fact there may be an unexpected benefit in these outcome measures as well," she said. "This adds to both operator and patient confidence that when the choice is a DES to reduce restenosis in patients who can and will be compliant with dual antiplatelet agents, they will do well, have low rates of restenosis, and will not be at increased risk of complications related to the stent."

Weiner cautioned that there are always limitations to meta-analyses, given the variability in populations studied in the different registries. "But in this case that can also be seen as a strength, since the results are fairly consistentacross them all and across multiple types of patient subsets."

Weiner's sole question, however, was over the lack of inclusion of trials that so far have been presented only in abstract form: the onslaught of new registry data over the past year has been difficult to keep up with, but some of these studies have produced ambiguous results.

Dr Sanjay Kaul (Cedars Sinai, Los Angeles, CA), who also reviewed the data for heartwire , congratulated the researchers for their "meticulous culling of the literature" but also expressed some concerns about how different registry studies were used. "It appears that studies that do not favor DES are selectively missing from the analysis. For example, I could not find inclusion of the GRACE trial that showed a five- to sixfold increase in adjusted mortality at days 180 to 730. Similarly, data from BASKET-LATE trial, which showed an increased risk from six months to 18 months, are not included. . . . Furthermore, data from SCAAR and the Emilia Romagna Italian and Massachusetts studies are missing from the TVR registry data--not surprisingly, the absolute risk reduction in TVR rates in these studies was less than 4%."

Kaul points out that, as yet, there is no plausible mechanism to explain the 18% to 20% mortality reduction with DES, making confounding a much more likely explanation.

"While there remains substantial uncertainty regarding the impact of DES on mortality or MI, it is clear that there is a significant benefit with regard to restenosis and TVR. However, the magnitude of this benefit is rather modest," Kaul said. "Coupled with the increased cost of DES and the amortized risk associated with late stent thrombosis--0.2% to 0.6% per year--and the monetary and nonmonetary risks (such as bleeding, noncompliance, rebound, inconvenience) associated with antiplatelet therapy, it becomes abundantly clear why a profligate policy on DES is neither reasonable nor supportable, despite what this meta-analysis tells us."

Kaul continued: "This meta-analysis, with all its limitations, is useful in the sense that it helps frame the question. Now it is our collective job to think about ways to answer it in a manner that truly informs, not obfuscates, clinical practice."

Safety in numbers

Kirtane, in response, clarified that he and his colleagues have conducted many more analyses of the data then they were able to present at the Friday night symposium and are continuing to examine their data from different angles, with the aim of submitting the analysis for publication within the next few weeks. Responding to the charge that some registries may be missing from specific analyses, for example, the Massachusetts registry included all repeat revascularizations, not just TVR, and therefore could not be included; SCAAR did not report TVR but instead looked at all repeat PCIs and "restenosis," which more closely matches target lesion revascularization, not target vessel.

The primary objective of the study was to bring all the data together in one place, Kirtane emphasized, recognizing the inherent problems in doing so. "As a clinician, you want to try to put things together in a way that will help you decide how to treat your patients better," he told heartwire . "Registry data are definitely confounded. The 47%, 11%, and 20% reductions [in TVR, MI, and mortality] that we report may be exaggerated by the limitations of the original registries, but what is clear is that there do not appear to be adverse safety signals associated with DES, and the reduction in TVR remains significant."

Kirtane disclosed receiving a past honorarium from Boston Scientific and serving as a consultant/speaker for Medtronic and Abbott. Stone disclosed receiving research grants from Boston Scientific and Abbott. Weiner and Kaul report no conflicts of interest.

The complete contents of Heartwire
, a professional news service of WebMD, can be found at
, a Web site for cardiovascular healthcare professionals.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.