Washington, DC - Two new meta-analyses of the Cypher and Taxus trial program results—this time conducted by investigators who participated in the trials themselves—are causing heads to spin among clinicians trying to sort through conflicting study results and learn how best to limit the impact of late stent thrombosis in day-to-day practice. Additional registry data presented at the TCT 2006 meeting pointing to a 0.4% increased risk of stent thrombosis with drug-eluting stents (DES) in real-world patients is only adding to widespread agreement that the phenomenon exists; however, its clinical implications and what can be done to prevent them are being hotly disputed.

To confuse matters further, new standardized definitions for stent thrombosis were also unveiled at the TCT meeting and applied to the Cypher and Endeavor stent data sets to show that late thrombosis risk might not, in fact, be higher in DES than in bare-metal stents. To see heartwire's coverage of the new Academic Research Consortium's definition of stent thrombosis, click here.

Dr Gregg Stone

During a special session devoted to the "ESC Firestorm," Drs Gregg Stone and Martin Leon (Columbia University, New York, NY), the two course directors of the TCT, showed results from "independent physician-led patient-level" meta-analyses of both stent programs that pointed to small but significant increases in late stent thrombosis for both drug-eluting stents, compared with bare-metal-stent controls. The studies were conducted through the Cardiovascular Research Foundation, which also sponsors the TCT meeting.

For the Taxus randomized clinical trials, that increase was 0.5% between one and four years after stent implantation, a rate of approximately 0.15% per year. For the Cypher trials, the late-stent-thrombosis rate was 0.6% between one and four years, or roughly 0.2% per year.

In neither group, however, did this increased risk of late stent thrombosis appear to translate into increased death or MI. Their hypothesis, explained by Stone earlier in the meeting, is that a counterbalancing effect of reduced restenosis in improving hard outcomes puts mortality and MI rates with DES on par with those of bare-metal stents.

Cypher meta-analysis: Clinical outcomes out to four years

End point	DES (%)	Bare-metal stent (%)	p
Stent thrombosis	1.2	0.6	0.200
Late stent thrombosis (1 year to 4 years)	0.6	0	0.025
MI	6.4	6.2	0.860
Death	6.7	5.2	0.190
Cardiac death	3.5	2.6	0.320
Death or MI	11.6	10.3	0.390

Taxus meta-analysis: Clinical outcomes out to four years

End point	DES (%)	Bare-metal stent (%)	p
Stent thrombosis	1.3	0.9	0.290
Late stent thrombosis (1 year to 4 years)	0.7	0.2	0.033
MI	7.0	6.3	0.640
Death	6.1	6.6	0.700
Cardiac death	2.4	3.0	0.520
Death or MI	12.4	11.8	0.770

New results contrast with WCC data

Dr Edoardo Camenzind

But Stone and Leon's results, at least in terms of death and MI, contrast starkly with data that first stoked the controversy from a spark to a blaze at the World Congress of Cardiology (WCC) 2006. Two meta-analyses, also based on the Cypher and Taxus clinical-trial data and presented by Drs Edoardo Camenzind (University Hospital Geneva, Switzerland) and Alain J Nordmann (University Hospital Basel, Switzerland), as reported by heartwire, first raised the specter of increased death and MI, attributed to a higher rate of late stent thrombosis, in DES-treated patients. These two studies were presented again here at the TCT meeting, with the presenters participating in a panel discussion following the sessions. Dr Peter Wenaweser (Thorax Center, Rotterdam, the Netherlands), whose Bern-Rotterdam registry data presented at the WCC 2006 meeting also pointed to an increased and persistent stent-thrombosis risk, presented his study again for the TCT audience.

The different findings for mortality and MI between the WCC and TCT presentations stem from the fact that Stone and Leon obtained raw data from the manufacturers and sponsors: from Cordis for the Cypher trials and Boston Scientific for the Taxus trials. Camenzind and Nordmann obtained their data from published and presented studies or in a restricted fashion from the companies themselves. At the WCC meeting, Nordmann described to the press the difficulty in obtaining raw data from the trial sponsors, and one of the points made by Stone at a press conference was the need for more transparent, reasonable, and practical methods for obtaining data of this sort from industry.

But what about real-world patients?

Although experts attending or participating in the packed session here seemed somewhat mollified by Stone and Leon's presentations, questions persist, not the least of which stem from the age-old problem of extrapolating from clinical trials to the real world. Dr Matthias E Pfisterer (University Hospital Basel, Switzerland), whose BASKET-LATE trial was one of the first studies to raise alarm bells about increased mortality in DES-treated patients, told heartwire he was not surprised to see no increase in death or MI in the analyses presented by Stone and Leon, because they relied on trials that enrolled stable, relatively low-risk patients.

"If you look at the data from the pivotal studies, which Dr Stone and Dr Leon presented, they are in very selected patients with very clear lesions," Pfisterer commented. "But if you look at the BASKET study and the Bern-Rotterdam registry, they all show that ACS is the only or most important predictor of late stent thrombosis, and this is off-label use. But ACS patients were not included in the pivotal trials."

At least 40% of people receiving DES have ACS, and this is the most important predictor of adverse outcomes following stenting, Pfisterer emphasized. "If ACS is the most important predictor of events, then that is why they didn't see it. They cannot say whether death, MI, or stent thrombosis is or is not true, or that it's not a problem in the real world, because they didn't look at these patients."

New registry data

For the time being, these kinds of data will likely come from registry studies, one of which was presented during the same "Firestorm" session by Dr Jose M de la Torre Hernandez (University Hospital Marques de Valdecilla, Santander, Spain). The ESTROFA study, conducted at 17 hospitals across Spain between 2002 and 2006, tracked stent thromboses among the patients treated over this time period. As de la Torre Hernandez showed, preliminary results from 15 centers where 13 500 patients underwent DES implantation show an overall rate of stent thrombosis of 1.2%, with 0.48% occurring after the six-month mark. Although the incidence of stent thrombosis in the first six months was similar to historic rates for bare-metal stents, the incidence after six months is important new information, but consistent with findings from other studies, he said.

"There appears to be a small (0.4%) increase in stent thrombosis after six months with both Cypher and Taxus DES, compared with bare-metal stents," de la Torre Hernandez concluded. "In that case, the beneficial decrease in in-stent restenosis with drug-eluting stents . . . outweighs any associated risk."

A need for transparency

Meanwhile, Camenzind, whose Cypher and Taxus meta-analyses were positioned as incorrect, or at least ill-informed, alongside Stone and Leon's, was cautious in his comments during the panel discussion. Everyone agrees, he said, that there is a need for "transparency and clarification."

Asked by heartwire if he was comfortable with the results from the new Taxus and Cypher meta-analyses that differed so much from his own, Camenzind said carefully: "If we have the same data, we should come to the same conclusions. My data are drawn from the literature and from official presentations, so what we have to see is the whole, real, official, correct data and then there should be no problem. We should get to the same conclusions."

He also pointed to the pitfalls of comparing trials that use different end points or definitions of events. Concerning the rate of stent thrombosis seen in Stone and Leon's analyses, Camenzind noted that it was similar to that reported by Wenaweser et al's Bern/Rotterdam registry. "The rate of stent thrombosis is . . . about 0.6% per year and continuing for three to four years. If it goes on like that and doesn't come to a plateau, that's a problem. Maybe it does, but that's what we need to find out."

Mulling over the mountain and the mole hill

In the panel discussion, as well as in the press conference that followed the presentations, experts tried to put the information in some kind of context. Dr Stuart Pocock (London University, UK) estimated that the risk of stent thrombosis is approximately one event per 500 patient-years of follow-up. "So the absolute risk is small," he reasoned. "There is an excess risk from the best data available, it is relatively low, and it needs to be seen in the context of the overall benefits."

But Dr Bram Zuckerman, director of the FDA's division of cardiovascular devices, although acknowledging that Pocock's comments were "a propos," reiterated that the existing patient-level meta-analyses only speak to low-risk patients of the type covered in the indications for use with the FDA-approved DES, not the large proportion of higher-risk patients who actually receive the stents.

Absent from much of the discussion here was speculation about the relative safety of the Taxus vs the Cypher stent—a question that has dogged both stents at different times in the recent past. Most of that, experts hinted, is "industry noise." At the press conference, Stone asked the panel of six experts whether any of them believed there was a clinically important difference in terms of safety between the Cypher and Taxus stents. The panel was silent.

"The answer is, no one is sure," said Stone. "Certainly the totality of data suggests that there is not a difference in safety. One issue is stent thrombosis, but that is only one of the many ways a device can harm you. There are hypersensitivity reactions, there are strut fractures, there's malapposition and aneurysm, and all sorts of other complications. So you have to look at overall patient outcomes, and the event rates are sufficiently low that you'd need a huge head-to-head randomized trial. I don't think anyone can convincingly say there is a difference, but I don't think anyone can say with scientific certainly that there isn't."

Clopidogrel decisions left to physician discretion

Instead, the most pressing question for interventional cardiologists concerned about late stent thrombosis remains clopidogrel safety—one of the focus points for the FDA panel discussion scheduled for early December. In addition, Cordis announced Tuesday that its 30 000-patient registry already under way in Europe and Asia/Pacific will be extended to the US and include a prospective randomized subset of patients to better understand stent-thrombosis risk and the use of antiplatelet therapy.

In the meantime, experts here at the TCT offered insight into how they are dealing with the dual-antiplatelet uncertainty, despite, as Leon put it, "flying by the seat of our pants."

Dr Spencer King (Piedmont Hospital, Atlanta, GA) told the panel that he is putting patients on dual antiplatelet therapy indefinitely but "hoping to find an exit strategy somewhere along the way."

Dr Sigmund Silber (Dr Müller Hospital, Munich, Germany), discussing the role of clopidogrel earlier at the meeting, said that he puts most patients on six months of clopidogrel but extends it out to two years in more complex patients. One hint, he suggests, is making better use of tests for clopidogrel resistance when deciding what type of stent a patient should receive.

Silber's opponent in the debate, Dr David Faxon (Brigham and Women's Hospital, Boston, MA), takes another tack: in lower-risk patients, he tries to convince surgeons to operate on patients without taking them off dual antiplatelet therapy. "Somebody told me to tell the surgeons that if they stop clopidogrel, the patient has a one in 10 chance of having a heart attack during the operation, and they say, okay, we'll do it on the drugs."

Stone had an even cheekier suggestion: "I tell the surgeons, if they don't operate on the drugs, there's a nine in 10 chance I'm going to find another surgeon. And that works even better."