CDU Washington Editor

GAITHERSBURG, Maryland — Largely at the prodding of industry, FDA has pondered whether it should grant device makers greater leeway in the use of clinical trials other than the gold standard of randomized, controlled trials (RCTs). But for the second time in less than a month, the circulatory systems advisory committee declined to support the idea.

In a late September meeting, the panel decided that RCTs are still necessary for ablation devices designed to treat atrial fibrillation, the panel focusing largely on enrollee recruitment problems.

The agency's circulatory systems panel met again in October and drew similar conclusions about the necessity of RCTs for carotid artery stenting for patients who are not at high risk of an imminent stroke. While recruitment again was discussed, the panelists seemed more concerned about proof of principle issues.

High-risk patients are those who have neurological symptoms — presumably from a transient ischemic attack — and 50% stenosis of a carotid artery, or an asymptomatic patient with 80% stenosis.

Kenneth Cavanaugh, PhD, a biomedical engineer at FDA's Office of Device Evaluation, said that stroke hits more than 700,000 people a year, and he pegged annual U.S. mortality at roughly 160,000. He said that the aggregate cost of stroke is about $57 billion and that "carotid artery disease accounts for about 30% of strokes."

Carotid artery endarterectomy is the surgical gold standard for elimination of carotid stenosis, and FDA has approved or cleared five stents for high-risk patients as an alternative.

Cavanaugh said that the April 2004 meeting for the PMA of the Precise stent, made by Cordis (Miami Lakes, Florida), led to the labeled restriction for high-risk patients. However, he said that the 2004 panel also concluded that there is a need for "a robustly designed, prospective trial to demonstrate proof of concept and support approval for non-high risk indications." Cavanaugh said that for non-randomized trials, "there's a risk that you won't know until the end of the trial that the data are not comparable."

Stan Fink, divisional VP at Abbott Vascular (Abbott Park, Illinois), said the firm "supports the use of randomized trials" to the exclusion of other trial models. He noted that Abbott was able to enroll adequately for the Carotid Revascularization Endarterectomy v. Stent Trial (CREST) study for the Acculink in non-high-risk populations (Abbott picked up the Acculink in its purchase of Guidant's vascular division).

"Randomization removes potential bias" and "produces comparable groups, even with unknown diagnostic factors," Fink said. Added to the low number of physicians available, these factors forced the trial into "a slow ramp-up of sites," Fink said.

He said that non-RCTs would present a further drag on RCTs because "competing non-RCTs would undermine the completion of trials."

Chris White, MD, the director of the Ochsner Heart and Vascular Institute (New Orleans) — telling the panel that he was speaking for the American College of Cardiology (ACC; Washington) and the Society for Cardiovascular Angiography and Intervention (Washington) — said he was "concerned about the population of the panel, and [I] wonder if this question can get a fair hearing" due to the lack of a carotid stenting interventionist on the panel.

"Like Mom and apple pie, RCTs are an excellent source of information, but they are not the only source," White said. He described RCT subjects as "highly selected" and that the results "do not describe the outcomes we see in hospitals today." RCTs, he said, suffer from the fact that they "take a long time, and this is a field that rapidly evolves."

White disagreed with Fink's statement that enrollment in RCTs would be hobbled by concurrent conduct of non-randomized trials. He said he was "utterly confident" that plenty of patients, centers and qualified physicians are available to handle all the trials that might be planned for carotid stents and that any non-RCT trial could be designed to exclude any patients who would be eligible for ongoing RCTs.

A non-RCT trial could use concurrent and cohort controls as well as registries, White said, pointing out that stent grafts for abdominal aortic aneurysms "have all been approved with concurrent, non-randomized control designs."

"Why is it different for carotid stents?" he asked.

During the panel discussion, panel member Eugene Blackstone, MD, the head of clinical research at the Cleveland Clinic, pointed out that a clustered randomized trial, which randomizes entire study sites to one arm of the study, "really hugely inflates your sample size."

However, such a design might get past some of the individual enrollment problems.

"It also has the advantage of including more of the real-life spectrum" due to the large numbers, he said.

Panelist David Milan, MD, a professor of medicine at Massachusetts General Hospital (Boston), said that CREST required eight years to fully enroll and that the benefit of randomization was probably somewhat offset by the fact that the standard of care for controls may have migrated in that time.