Lebanon and Hanover, NH - For the second time in as many months, researchers have published a Medicare analysis comparing outcomes with drug-eluting stents (DES) vs bare-metal stents, showing that DES reduce the risk of repeat procedures but do not increase the risk of death or MI [1]. Just last month in the Journal of the American College of Cardiology, Dr Peter Groeneveld (Philadelphia Veterans Affairs Medical Center, PA) and colleagues published an analysis of more than 75 000 Medicare patients treated with a Cypher DES within the first nine months of the device's FDA approval, showing that DES not only reduce the risk of revascularization procedures but also appeared to reduce the risk of death and MI [2].

Now, writing in the June 25, 2008 issue of the Journal of the American Medical Association, a second group, this time from Dartmouth Medical Center and led by Dr David Malenka (Dartmouth-Hitchcock Medical Center, Lebanon, NH), report that the need for repeat revascularization procedures was indeed reduced over a two-year period in Cypher-treated patients as compared with bare-metal stent-treated patients. Malenka et al, however, saw no difference in unadjusted mortality risks at two years and a small but statistically significant decrease in ST-elevation MI.

In an interview with heartwire, Malenka acknowledged that the publication of the two studies, ostensibly asking the same question in the same patients—yet reaching different conclusions about the MI/mortality effects of DES—may seem confusing. The key difference between the two studies, he explained, was the time period analyzed. In the Groeneveld paper, the authors included all Medicare patients who received a DES between April 2003, when the devices were approved, and December 2003. In Malenka et al's paper, authors waited for five months post-DES approval, including patients who received a DES between September and December 2003.

"When we first started to do this analysis we did it the same way they did, but we worried and worried about selection bias that we might be missing," Malenka explained. When they looked at survival rates, patients treated with bare-metal stents showed a spike in mortality rates in the first few months after DES were approved—something he and his colleagues believe to be clear evidence of selection bias. "When new technology comes out, people tend to use it cautiously at first, they tend to use it in people in whom they are likely to get a good result," Malenka said. "So what we think was happening was an important selection bias, with a sicker patient population getting the bare-metal stent instead of the DES. So all of a sudden, the mortality for bare-metal-stent-treated patients seems to go up. . . . The only explanation that I can come up with for that mortality pattern is that somehow the patients who were getting DES must have been healthier or more likely to survive than patients who got a bare-metal stent."

To avoid this selection bias, Malenka et al opted to include DES-treated Medicare patients after a short interval, while operators became more accustomed to and comfortable with using the devices.

In all, their analysis included 38 917 Medicare patients who received bare-metal stents between October 2002 and March 2003 and 28 086 patients who received a DES between September and December 2003. Over two years of follow-up, repeat revascularization procedures (PCI or CABG) were lower in the DES-treated patients than in the bare-metal-stent-treated patients; unadjusted rates of ST-elevation MI were also slightly lower, while unadjusted death rates were no different. But after adjustment for age, sex, race, history of MI, CHF, vascular disease, etc, there were no differences between rates of death or MI between the two groups. A landmark analysis, looking only at event rates after three months—marking the recommended cutoff point for dual antiplatelet therapy when DES were first approved—also showed no differences in adjusted mortality or MI rates between the two stent groups.

"I think this is reassuring," Malenka told heartwire. "I think it clearly shows that, whatever risk DES pose associated with late stent thrombosis, by decreasing the risk of restenosis and the associated adverse events related to restenosis, DES have a net benefit. We probably should be using them more, but with the appropriate caveats: that people need to be on long-term antiplatelet therapy, and we need to make sure we are identifying the patients who will be compliant and can afford it."

Malenka acknowledged that the reduction in restenosis with DES (3.8%) was lower than that seen in clinical trials but consistent with other population-based studies where angiographic follow-up is not mandated. "You never see the magnitude of effect in the real world that you see in a randomized trial. . . . This is still a clinically meaningful difference," he said.

I think this provides tremendous reassurance and solid evidence that on average, DES are effective.

Malenka also emphasized that the key point of the JAMA paper is more or less the same as that of Groeneveld et al—namely, that in real-world patients DES reduce the risk of restenosis, without increasing the risk of death or MI. Even Groeneveld et al acknowledged the possibility of selection bias boosting the apparent mortality benefit in their study: in both papers, the very large patient numbers lend statistical significance to small differences that would seem less relevant in smaller studies. As Groeneveld told heartwire when his paper was published, the most important observation from this Medicare data set is that there is no signal of harm.

Malenka agrees.

"I think it's very hard for some clinicians and certainly many patients to wrap their minds around competing risks and benefits," he said. "This study very clearly demonstrates that the benefit of DES, in terms of decreasing repeat revascularization and the associated morbidity and mortality, clearly outweighs the risk of late stent thrombosis."

He continued: "That doesn't mean we shouldn't strive to create a new generation of DES that endothelialize more reliably, and it doesn't mean we should forget about the need for long-term Plavix therapy with DES, which will be an issue for some patients. But I think this provides tremendous reassurance and solid evidence that on average, DES are effective."

Contacted by heartwire for his thoughts on the discrepant results, Groeneveld said, "It's not immediately clear why the Malenka group's results were different, other than they were comparing all patients who received stents from October 2002 to March 2003 with all patients who received stents from October to December 2003." While the first group received only bare-metal stent patients, the second group was actually just 60% DES.

"The survival of these two groups was the essentially same, so they concluded that DES must not improve survival," Groeneveld explained. "I have some problems with this, since it presumes that exactly the same kinds of patients were getting PCI before and after DES entered the market. In fact, we know based, on the significant decline in CABG rates after DES, that numerous patients were getting PCI after April 2003 who would have received CABG before DES came along. It's possible that the October-December 2003 cohort was 'sicker' than the October 2002-March 2003 cohort in ways that Malenka and colleagues could not observe, and thus the beneficial effect of DES in this cohort was blunted."

Groenveld added that he and his team are "trying to reproduce the Malenka analysis and delve into these issues," with the intention of putting their findings in a letter to the editor of JAMA.