New data from the Western Denmark registry presented by Dr Anne Kaltoft (Aarhus University Hospital, Skejby, Denmark) during the late-breaking clinical-trial session suggested that between 12 and 24 months after stent implantation, patients treated with the Taxus-eluting stent face a nearly fivefold higher risk of stent thrombosis than bare-metal-stent treated patients and a nearly 50% increased risk of dying.

By contrast, Cypher-treated patients have similar rates of stent thrombosis, MI, and death as patients who underwent bare-metal stenting, but face slightly increased risks of MI between one and two years.

"Our data show that the effects of DES should not be considered a class effect and open the door for improvement of future generations of DES," Kaltoft concluded.

The two-year results from the Western Denmark registry represent the latest update from the researchers, who first observed an uptick in rates of stent thrombosis and MI after 12 months among DES-treated subjects when they presented 15-month data at the 2007 ACC meeting. Their results were later published in the Journal of the American College of Cardiology [1]. At that time, the Danish authors concluded that the overall rate of adverse cardiac events was low and similar between bare-metal stents and DES and that the low risk of stent thrombosis or MI does not outweigh the benefits of DES in reducing revascularization procedures.

In their latest presentation of the Western Denmark data, Kaltoft and colleagues looked specifically at results by stent type. Their registry includes information on all patients who underwent stenting at one of three high-volume centers between January 2002 and June 2005, from among the roughly three million inhabitants of Western Denmark. In all, 8847 patients were treated with bare-metal stents during the period, while 3500 received a drug-eluting stent.

They report that the rate of target lesion revascularization was significantly reduced compared with bare-metal stents, for all DES and for Cypher and Taxus stents analyzed separately. Rates of stent thrombosis were no different for bare-metal stents vs DES for the 24-month period, but when investigators looked specifically at the period between one and two years, the adjusted hazard ratio was 5.25 for DES. When DES were analyzed separately, rates of stent thrombosis over the full 24 months and between 12 and 24 months were no different from bare-metal stents for the Cypher, but during both periods, stent thrombosis rates were significantly higher for the Taxus stent. This difference was the likely driver of the twofold MI risk among all DES-treated patients between 12 and 24 months. In the Taxus-treated patients, both MI rates and mortality were increased between one and two years. Of note, recommended duration of dual antiplatelet therapy was 12 months for most of the study period.

Dr Dariusz Dudek (Instytut Kardiologii, Krakow, Poland), who discussed the registry results following their presentation, urged caution in interpreting the results.

"We should be extremely careful about what can be concluded from a registry," he said. "Registries should not be used for making conclusions related to safety and efficacy between two therapies because we can find many examples of registries and randomized controlled trials that do not support these presented data."

A conspicuous example, he noted, is the Danish SORT-OUT II data, a randomized controlled trial of more than 2000 patients that showed "absolutely no difference" in outcomes between Taxus- and Cypher-treated patients at 18 months. Those results, as reported by heartwire, were published in the Journal of the American Medical Association in January 2008 [2].

"Registries have limited value for clinical decision making, particularly when not powered for comparison of one stent vs another," Dudek stressed. "The highest [level of evidence] is still the adequately powered randomized controlled trial and this should be our goal. . . . We should not use registries and meta-analyses for device selection."

Possibly adding to the sense of general mystification or frustration over registry results, another registry analysis presented during the EuroPCR late-breaking sessions—a two-year follow-up analysis from the STENT registry—showed no difference in outcomes among diabetic patients treated with either the Cypher or Taxus stent.

Commenting on the latest Western Denmark results for heartwire, Dr Carlo Di Mario (Brompton Hospital, London, UK) warned against the dangers of overreacting to registry data. "I agree with Dr Dudek that we somehow need to minimize the role of registries in determining our assessment of events like stent thrombosis," he said. "I understand that the investigators have done a very thorough review of the clinical characteristics, but I'm convinced that you really cannot exclude large differences between patients that led to the use of one stent or the other. You truly cannot correct for these types of differences."

Di Mario added that the diabetic analysis from the STENT registry included roughly the same number of patients as the Danish registry yet showed no differences by stent type, and this in a group of patients where it might be expected that any differences would be amplified. The SCAAR registry, too, has shown no differences in outcomes for the Cypher vs the Taxus stent, he added. "So these results from Denmark are very questionable," he said.