Gaithersburg, MD - On the second and final day of the FDA's drug-eluting stent (DES) safety hearing, members of the FDA's Circulatory Systems Devices Advisory Panel left open the possibility that death and MI related to late stent thrombosis might indeed be increased in patients who receive the devices off-label, a link they rejected yesterday when reviewing the data for devices used according to labeling indications.

"Although there's no question that DES are a great advance, they also created a new entity—some would say a monster—which is this very small but statistically significant increase, even in the ideal patient population, of late stent thrombosis," panel member Dr Eric Topol (Scripps Translational Science Institute, La Jolla, CA) stated. "It would be inappropriate, I believe, to extrapolate from the fact that late thrombosis does not track with death and MI in the randomized trials. I'm not prepared to make that big leap in off-label use."

I hope the message that goes forward Monday morning is that any application of stent technology that is off-label should give someone pause, because we don't have enough clinical experience.

Likewise, Dr Clyde W Yancy (Baylor University Medical Center Dallas, Dallas, TX) concluded, "I hope the message that goes forward Monday morning is that any application of stent technology that is off-label should give someone pause, because we don't have enough clinical experience. If we were here talking about a relatively recent medical intervention in which 60% or more of the use was off-label, the flavor of these comments would be vastly different. Trying to tease out who is the best group or the worst group for DES, with what we have available—I can't support that. Off-label is off-label."

Some panel members believed the FDA should go even further. Dr Norman S Kato (Cardiac Care Medical Group, Encino, CA), referring to several registry studies, pointed out that the cumulative risk with DES increases steadily over time, whereas the curves for bare-metal stents, while higher to start with, do not climb as sharply. "I think that these outcome lines are really diverging, they are not coming together. Mortality is increasing in a linear fashion and may be accelerating, and it's making me very nervous about proceeding with DES in general," he said. "It's fairly safe to say that on-label use is safe and effective, but I don't think we have enough data at all to say that putting in a DES for other reasons is safe and effective, and therefore not only can we say that this is off-label use, it needs to be said more forcefully, even contraindicated. In the regulatory world that would be equivalent to a black-box warning."

Panel chair Dr William Maisel (Beth Israel Deaconess Medical Center, Boston, MA) rejected such a move as overkill. "I think a black-box warning is far too strong for a large number of thoughtful physicians who choose to use this device off-label," he said.

The panel's predicament was unique. By definition, "off-label" use refers to practices for which a device or drug is not recommended by regulators, yet the FDA was asking the panel to make recommendations about such use.

"It is not our usual structure to extensively regulate the practice of medicine," Dr Bram Zuckerman (FDA Office of Device Evaluation) acknowledged. "The relationship between the patient and physician and their choice of device used is something that we don't want to tamper with. But when there is a significant health problem, the FDA and Health and Human Services cannot ignore it. In our two days of deliberations, we've heard that there are signals that need to be very seriously looked at."

The panel spent most of the day digesting what amounted to an outpouring of data—25 presentations in total, described by one panel member as ranging in quality from "good" to "terrible." There were other, often-impassioned statements by surgeons, professional societies, and even a member of the public who had undergone DES implantation.

The surgeons' presentations about the mortality benefits of CABG in multivessel disease, in particular, struck a chord with many panel members. The surgeons' primary contention was that DES should not be compared with bare-metal stents in multivessel disease, since the standard of care in this subset is CABG.

Dr Peter K Smith, chief of thoracic surgery at Duke University, Durham, NC, presented an analysis arguing that DES is associated with increased mortality as compared with bare-metal stents; thus, for the most important clinical outcome—mortality—DES is at best "noninferior" to bare-metal stenting. From this perspective, it is relevant to compare mortality outcomes between bare-metal stenting and CABG for multivessel disease, he said. In an analysis that combined the Northern New England Database (1994-2001), the New York State Database (1997-2000), and a Duke observational study (1986-2000), Smith showed that the absolute survival advantage for CABG over bare-metal stents over seven years would be 6.3%, appearing to increase with each year of follow-up from one to seven years. The mortality impact on a per-patient basis would be roughly 6500 worldwide and almost 3700 in the US alone.

"So if the patient's question is, 'Am I more likely to be dead if treated with a DES,' the answers is 'yes, if you have three-vessel disease and are enabled to have DES or bare-metal stent implantation rather than CABG,'" Smith concluded.

Dr Robert A Guyton,chief of cardiothoracic surgery at Emory University, Atlanta, GA, was even blunter: "DES is inferior to CABG in multivessel disease: do we not need a DES labeling change to reflect this fact?" he asked panel members.

At the very least, he argued, patient information needs to be less biased in favor of an interventional approach and should involve a multidisciplinary team of health professionals to provide a more rounded view.

All of us should be concerned about the underuse of surgery and the overuse of angioplasty. That said, there's substantially more equipoise in the community over the question of whether surgery really is superior.

Ultimately the panel declined to specifically single out multivessel disease in the label—a proposal put forward by the Society of Thoracic Surgeons, but several panel members were vocal in their support of the need for a CABG arm in all future DES trials.

Panel member Dr Sharon-Lise Normand (Harvard School of Public Health, Boston, MA), who repeatedly pointed to the lack of comparative data in the bulk of the off-label registry studies, called for future DES registries to also track a contemporary CABG-treated population.

Others, however, reminded their fellow panel members that studies specifically examining surgery vs DES were ongoing, and it would be premature to change the label when important information is still pending.

"All of us should be concerned about the underuse of surgery and the overuse of angioplasty," Dr Michael J Domanski (National Institutes of Health, Bethesda, MD) acknowledged. "That said, there's substantially more equipoise in the community over the question of whether surgery really is superior, and that finds expression in the fact we have two major trials ongoing, one in three-vessel disease and left main disease, and one in diabetics and left main disease."

And Dr George Vetrovec (Medical College of Virginia, Richmond, VA) added, "I think we have to be careful in all these discussions of surgery vs stents not to forget that some of the people who we are pushed into a corner to treat have been turned down for surgery and are still having symptoms."

Perhaps most galvanizing of the off-label data was the Swedish registry data, presented Thursday by Dr Lars Wallentin (Uppsala Clinical Research Centre, Sweden). These data made major headlines in Sweden following the Barcelona World Congress of Cardiology meeting after local press got hold of data from countrywide PCI records, as reported previously by heartwire. Now, presenting the data for the first time after rigorous analysis, Wallentin showed that rates of death, MI, and death/MI after six months were significantly higher among DES-treated patients than among bare-metal-stent-treated patients over 2.5 years of follow-up and after adjustment for background characteristics. According to Wallentin, the relative risk in death/MI in patients treated with DES as compared with bare-metal stents was 20%, for mortality 32%, and for MI 12%. Restenosis rates, however, were halved in the DES-treated patients.

"I think what we see is an early clinical benefit, [a decrease in] restenosis, followed by a loss of this benefit over time due to the increase in death/MI," Wallentin told heartwire. "I think the message should be that drug-eluting stents should be used only in patients with the highest risk of restenosis, which is what we are now doing in Sweden." DES usage rates in Sweden have dropped dramatically since the data were first made public.

Asked during a press conference whether he could name an appropriate level of off-label DES use, panel chair Maisel said it would be impossible to name such a figure.

"I don't know that there is a number that can be assigned to appropriate off-label use," he said. "I think after today, the panel would like to see the off-label use go down in the higher-risk subsets that we've identified."

As with the on-label discussions on Thursday, the panel seemed flummoxed by the profound lack of randomized clinical-trial data for more complex lesions and patients, despite the fact that these off-label situations make up the bulk of DES usage, ranging from 60% to 75%, according to presentations today. As one panel member pointed out, the panel was reluctant to make pronouncements for or against all off-label usage, since other evidence hinted that DES use should be restricted to patients at the highest risk for restenosis, a category that would embrace many patients and disease conditions not currently covered in labeling.

But unlike day 1 of the meeting, where it quickly became clear that the panel was reluctant to propose any meaningful changes in terms of on-label use, day 2's discussions signaled change early on, with several members making it clear they weren't content to leave the labeling untouched altogether.

The existing package insert was partly responsible for "how we got to this point," Kato argued. "To leave it alone means a tacit acknowledgement that we're not going to change anything and that this meeting is a waste of time."

Ultimately, the panel more or less agreed that DES labeling should be updated to specifically state that the risk of stent thrombosis, MI, and death is higher when DES are used off-label as compared with on-label usage, a caution that also holds true for bare-metal stents, several panel members noted. They also singled out specific situations in which off-label DES might be more likely to increase stent thrombosis risk: namely, bifurcations, overlapping stents, and vessels with thrombus. Several members of the panel acknowledged that what they'd seen presented during the hearing would have at least a minimal impact on their practices.

Asked whether there were different safety concerns with off-label use of the Cypher as compared with the Taxus, the panel found no evidence to suggest there was any difference between the two stents. They also cited a lack of evidence in making any special statements on the off-label use of DES in diabetic patients.

As with day 1 of the meeting, the panel spent the good part of an hour haggling over appropriate dual antiplatelet recommendations in off-label use, ultimately deciding, if not unanimously, to specify 12 months as appropriate, given the sense that stent-thrombosis risk may be increased in these patients. But several panel members expressed their dissatisfaction with the decision due to the lack of any truly applicable evidence to support this apparently arbitrary cutoff, as well as the fact that several studies suggested that clopidogrel may have limited impact on stent-thrombosis risk beyond six months.

"There is a gut feeling that higher-risk patients may benefit from prolonged therapy, but the fact is the majority of thrombotic events occurred [when patients were still taking antiplatelet therapy]," Dr Jeffrey A Brinker (Johns Hopkins Hospital, Baltimore, MD) said. "I'm happy for the time being with the decision as it stands—that is, up to one year if tolerated—but I would encourage further study."

The panel also agreed that, going forward, new DES studies should have to be considerably longer, enroll much greater numbers of patients, and include stent thrombosis as a study end point. Furthermore, a large, randomized study looking specifically at appropriate duration of dual antiplatelet therapy is also urgently needed, they said. Postapproval registries should also be several years in duration, the panel specified—a recommendation prompted by the fact that, to the shock and disbelief of panel members, Cordis/Johnson & Johnson's postapproval registry was only one year in duration.

I find it extremely concerning that we don't have registry data beyond one year.

"In trying to reconcile why that data aren't in front of us, I can think of only two explanations: one, you've chosen not to show us the data, or two, you haven't done your due diligence to present to the panel," Maisel blasted Cordis VP for worldwide regulatory and clinical affairs, Dr Dennis Donohoe. "This is critical data for us to have our discussion and make recommendations, and I find it extremely concerning that we don't have registry data beyond one year."

In the company's defense, Donohoe pointed out that the panel hadn't originally asked for longer registry follow-up. Confirming this, the FDA's Zuckerman pointed out that at the time the Cypher stent was approved by the FDA, stent thrombosis as a phenomenon was not expected to occur after the first few months postimplantation. " In retrospect the agency and Cordis would like to have done it differently," Zuckerman said.

Speaking with heartwire after the hearing, panel member Dr Steven Nissen (Cleveland Clinic, OH) said he was "very pleased" with the panel's decisions.

"I think we said very clearly that we shouldn't allow a device to get on the market and to get into use in something like 90% of patients without having adequate data to know that that was the right thing to do. There's a terribly important message here, and that's the law of unintended consequences: if you've never studied a patient population or a risk group, you shouldn't assume that they will have the same benefits as other groups. And we should also realize that short-term clinical outcomes don't always define long-term clinical benefit."

Drug trials are frequently five years in duration, yet the DES were approved on the basis of nine-month data, he noted.

Nissen acknowledged that the panel was more dynamic on the second day of the hearing and less attached to the status quo.

"I was a little frustrated and concerned yesterday," Nissen conceded. "But I think it's important to understand that we were asked yesterday to consider only the narrow issue of whether these devices were safe and effective in the population for which they'd been studied in randomized trials. And people were okay with that. Today we got into the more extended use, and I think the panel got it, and they got it right. Some of us obviously felt a bit stronger about these issues, but by and large, the majority was in a solid, centrist position, which was to go where the evidence was taking us and avoid this stampede toward using DES as a substitute for well-proven therapies, including bypass therapy."

Asked during the press conference what steps the FDA would be taking, Dr Daniel G Schultz, director of FDA's Center for Devices and Radiological Health, said the FDA would be convening its own internal group of experts to review the panel's recommendations and may or may not update the label. Schultz agreed that the FDA needs to put out some kind of "public message," although what form that will take is not yet known.

As for changes that might be required of upcoming studies of next-generation DES, Schultz said the panel's recommendations would likely apply primarily to studies not yet in the pipeline. While changes could still be made to studies already under way—of Medtronic's Endeavor, Abbott's Xience-V, and the CoStar stent made by Conor (recently purchased by Cordis/Johnson & Johnson)—these trial protocols have already been reviewed by the FDA and are likely to continue as originally designed. Some may already meet the proposed refinements: Medtronic's PROTECT study, for example, comparing the Endeavor with the Cypher, is enrolling 8000 patients, with definite or probable late stent thrombosis at three years as one of its primary end points.

Indeed, during the hearing, the panel seemed impressed by the combined Endeavor trial data presented by Dr Rick Kuntz, senior VP for Medtronic, which showed very low rates of stent thrombosis—ranging from 0.3% to 0.5% percent using the per-protocol or Academic Research Consortium probable/definite definitions, respectively, out to three years. Strikingly, unlike with the accumulating data for the approved DES, the stent thrombosis rate, as well as Q-wave MI and all-cause mortality rates, among DES-treated patients in the Endeavor trials are consistently numerically less than those of the bare-metal-stent-treated patients.