Medical Device Daily Washington Editor

GAITHERSBURG, Maryland – There are times when changing a clinical study protocol midstream creates major headaches for device makers, but Biosense Webster (Diamond Bar, California) managed to sail through an advisory committee hearing last week with a unanimous vote for approvability despite the switch.

Webster's radio frequency NaviStar catheter with the Thermocool feature, a saline rinse used to reduce unwanted damage to heart tissues, was originally approved for treatment of atrial flutter, but will be the first ablation devices approved for atrial fibrillation (AF), assuming FDA follows the panel's unanimous recommendation for approval.

However, the panel saved its recommendation for only the NaviStar and declined to recommend that FDA authorize the same indication for the company's other Thermocool catheters. As for post-approval studies, Webster also managed to dodge a recommendation that the post-approval study (PAS) include a control arm, which would have substantially upped the cost of a study.

Leading off the company's presentation at the hearing, Marcia Yaross, PhD, its VP for clinical, quality and regulatory affairs, hinted at an unmet need in reminding the panelists that "no ablation catheter has been approved ... in the U.S." for paroxysmal AF. She also said that the Navistar has marketed in 39 nations since its introduction in 1998.

According to the data presented by Webster, the probability of chronic success (at nine months) for the patients on ablation was 62.7%, which was deemed significantly better than the 17.2% probability for controls, who were on medical therapy for their arrhythmias.

The data also indicated that there were no device-related serious adverse events (SAEs) such as death, heart attack, or stroke within seven days of the procedure, and no clinically significant pulmonary vein stenosis in patients receiving ablation. Overall SAEs in the study at 90 days were about half that in the control population, or 18.4% vs. 35.1%.

Regarding the study conducted to bolster the new indication, Yaross said that the study population "all had failed at least one anti-arrhythmic drug" and that the company had found that "enrollment ... was extremely challenging."

This was the reason Webster had requested FDA's stamp of approval on the change from a traditional statistical analysis plan – the so-called frequentist model – to a request for a Bayesian analysis plan, which FDA approved in September 2007. She said that an interim analysis led to a conclusion of success for the device, which led to suspension of further enrollment.

Yaross said the final enrollment was of 167 subjects at 19 centers and that no centers accounted for more than 30% of enrollees.

Donald Berry, PhD, reminded the panel that the agency agreed to the use of Bayesian statistical approaches and issued a guidance in 2006 on the topic. He said FDA had agreed to the interim analysis in 2007, and that the analysis of 148 patients led to a predictive probability of success was greater than the 90% threshold set by the agency and the sponsor for ceasing patient accrual.

Laura Thompson, PhD, reviewed the agency's statistical summary. She reminded panelists that Bayesian statistical analysis is used primarily to analyze data as they come in, but remarked that "FDA believed the change was potentially problematic." Still, FDA applied a penalty to the analysis to account for the intra-study switch to Bayesian analysis.

"Normally, FDA does not recommend a change" in the middle of a trial," she said, but acknowledged that "it is likely that the original frequentist analysis would have shown statistical significance," in the difference in outcomes.

Thompson said the first problem is that "the trial was unblinded," to the interventionists who participated. She also cited as a problem that the determination of AF "was not entirely objective" inasmuch as patient reporting was part of that determination. Consequently, "it is unclear to what extent the results are due to a placebo effect."

"Finally, the largest site performed better" than the smaller study sites, she said, adding that "the primary effectiveness endpoint was met" despite that a penalty was paid for the switch from frequentist to Bayesian analysis, Thompson acknowledged.

Panelist David Slotwiner, MD, of the Long Island Jewish Medical Center (New York), gave an analysis of the application, noting that each of the procedures in the trial were performed with computerized tomography or magnetic resonance imaging data, giving the electrophysiologist "an enormous amount of information" on the patient's heart. However, he said, "we may need to consider" whether it is appropriate to allow use without three-dimensional imaging, "especially for less experienced operators."

Slotwiner also asked, "Should we give approval to all five of these catheters when only one was studied?" He also observed that the European sites taken together "look better than all the U.S. sites," which suggests that physician experience might be a factor.

"Overall I want to commend the sponsors for what is clearly one of the most important fibrillation trials to date," Slotwiner said.

One of the questions FDA put to the panelists was whether they were put off by the sponsor's decision to exclude the use of amiodarone as a treatment in the control arm. Slotwiner said, "for these younger patients ... with paroxysmal atrial fibrillation and no special heart disease," it would not be used in any case, "so I think that was an appropriate and reasonable design." Another panelist seconded the age-related comment.

Bram Zuckerman, FDA's director of cardiovascular devices, replied that the device might be used in older populations, but panelist John Somberg, MD, of Rush University Medical Center (Chicago), countered that "you couldn't design a trial for amiodarone" because its side effects are "horrible" and the adverse event rate seen in the drug would still leave the device with better outcomes

As to whether safety had been established, panelists were comfortable with the data for the NaviStar, but some were less comfortable with expanding the indication to other Thermocool devices. Efficacy was also seen as sufficiently demonstrated, although some panelists expressed reservations about its use on older patients.