Medical Device Daily Washington Editor

GAITHERSBURG, Maryland — One of the issues that cropped up at Wednesday's meeting of the cardiovascular devices advisory committee was the sponsor's use of Bayesian statistical analysis to analyze data from the two trials used to support the application.

Thanks to that analysis, panelists repeatedly noted that the purportedly small data set offered by TherOx (Irvine, California) for its DownStream aqueous oxygen system left them uncertain as to the device's risk-benefit profile. The Bayesian dilemma for panelists combined with a higher observed number of fatalities in the study arm of the primary clinical trial compared to controls combined to sink the firm's application for DownStream.

A low volume of data did not seem to trouble the panel in last November's hearing for the Thermocool system (Medical Device Daily, Nov. 24, 2008) made by Biosense Webster (Diamond Bar, California), but that application was to expand the indications for its Thermocool device from reducing ablation-related damage during treatment of atrial flutter during ablation to the same use during ablation for atrial fibrillation. The Downstream, on the other hand, would be indicated for reduction of infarct size during percutaneous coronary intervention (PCI), and thus, would be the first such device on the U.S. market.

TherOx did not lack for heavyweights on its side. Gregg Stone, MD, the director of cardiovascular research at New York Presbyterian Hospital (New York), was the principal investigator for the pivotal study and spoke on the company's behalf. Stone's prestige is abetted by his directorship of the annual Transcatheter Cardiovascular Therapeutics symposium on behalf of the Cardiovascular Research Foundation (New York). Stone told the panel, "we're in a national epidemic of acute myocardial infarction," which, when left untreated as it is now, "translates to a death approximately every minute." He also said that TherOx's primary trial showed "a strong linear relationship between infarction ... and mortality." He said that a study published in the New England Journal of Medicine indicated that a reduction in infarct size correlates "strongly with lower rates of mortality and morbidity" outcomes.

"This is a major healthcare issue for all of us," Stone asserted.

Jeff Creech, PhD, the program manager for TherOx, told the panel that the company, which employs 25, filed the investigational device exemption in 1999 and completed the PMA about a year ago. The idea of the DownStream, he said, "is to bring the benefits of the hyperbaric oxygen chamber into the cardiac cath lab," but since it is impractical to operate a cath lab in such a chamber, the device would provide "localized hyperbaric oxygen," with the intent of providing "the equivalent of an additional 25-30%" more oxygen than the tissue would ordinarily receive.

Creech also reminded the panel that four other devices have won approval with the help of Bayesian-derived statistics, including the Thermocool. "The advantage of the Bayesian model is that ... it allows you to get the therapy to the public sooner, and if it's ineffective, it tells you that sooner."

The two trials that formed the basis for the Bayesian analysis were the AMIHOT I (Acute Myocardial Infarction Hyperoxemic Therapy) and its Bayesian-by-design successor, AMIHOT II. Stone explained that AMIHOT I enrolled 135 controls and 134 study subjects, with study subjects experiencing nine major adverse cardiac events (MACE) vs. the seven seen in the control arm. Among these MACE events were a total of six fatalities, two in the control arm and four in the study article arm. While the expected reduction in infarct was not met for all the patients in this study, who were all treated within 24 hours of the onset of symptoms, patients who made it into the cath lab within six hours showed promising results, and TherOx opted to build a second study onto AMIHOT I based on that timeline.

AMIHOT II enrolled 304 patients in a 2.8:1 ratio (222 to 79) of study subjects to controls to examine the study hypothesis that study subjects would have at least 5% less infarct than controls at 14 days. When the data from the six-hour patients from AMIHOT I were pooled with the data from AMIHOT II, study subjects exhibited a reduction in infarct size on average of 6.9%. However, there were no fatalities in AMIHOT II controls whereas four of the study subjects passed away during the course of follow-up.

Panelists questioned a number of the ideas behind the study, including the possibility that the oxygen-saturated saline therapy would create ionizing radicals in heart tissue and what some saw as a tentative connection between a 5% reduction in infarct and long-term outcomes for patients. However, the question that dogged the firm's application most consistently was whether an expanded data set might reverse the differences in fatalities, a consideration that was behind a number of comments about the "small" data set.

TherOx did not lose its bid due to bad relations with FDA. Julie Swain, one of the reviewers, said "the company was extremely responsive" and gave TherOx high scores for cooperating with the agency.

In summarizing a panel discussion, panel chairman Jeffrey Borer, MD, of Weill Cornell Medical College (New York), said there is "a sense of concern even though the numbers were too small to allow us to draw conclusions, which perhaps I shouldn't say because of the Bayesian analysis." He also noted concerns about the impact of oxygen on heart tissue.

When it came time to vote up or down, the panel appeared momentarily to give the product a nod for approvability. Discussions of what sort of conditions the panel would recommend seemed to give the application new life. This surge was probably prompted in part by closing comments offered by Stone, who made the argument that "it appears that the MACE rates were quite low" for study subjects and fell "within statistical comparability" with controls. Stone also argued that "we're creating a nation of heart failure patients," with the current lack of therapy for acute infarction, and said "for those patients ... with large infarctions," some treatment is needed.

All the same, the vote for approvability died with only three votes registering as ayes, while the following vote for disapprovability came in at 9-5.