With two sets of conflicting data in front of them, the Cardiovascular and Renal Drugs Advisory Committee's first task for its review of Sanofi-Aventis' Multaq was to determine if the results were compatible. In the end, swayed by the positive findings in a narrower population, the committee supported a limited approval for dronedarone.

Going into the March 18 review, a big part of what the advisory panel had to determine was the causes for the discrepancies between the negative ANDROMEDA trial and the subsequent positive pivotal trial, ATHENA. ANDROMEDA was stopped in 2003 because of the risk of mortality, and led to a "not approvable" letter (¹ Also see "ANDROMEDA Strain: Full Data Published From Multaq Trial" - Pink Sheet, 19 June, 2008.).

The Long Road To Reconciliation

Sanofi offered up three possible explanations for the inconsistencies in data between ANDROMEDA and ATHENA. The first was that the trial results represent a chance finding related to the uncertainty associated with the frequent interim monitoring of small numbers of events in a trial that was terminated early. Essentially, the firm offered up the possibility that the mortality difference in ADNROMEDA was a false positive finding.

While a number of committee members said that the mortality difference could be explained by chance, they were not willing to ignore the results.

"The suggestion that this might in fact be a statistical variant is fair and the data that the sponsor provided suggests that this could be a false positive as much as 20 percent of the time is something we can't discount," said temporary voting member William Calhoun, University of Texas. "However, in that, patient safety is really paramount. I agree that we do have to treat this as a real finding."

Sanofi's second theory, that the ANDROMEDA trial results represent a true finding related to difference in the use of certain background medications following randomization, is something FDA was initially keen on having committee members discuss.

The hypothesis was that increases in serum creatinine related to dronedarone may have led to differential use of RAAS agents. Patients taking ACE inhibitors or ARBs at baseline accounted for 84.2 percent of the placebo group compared to 88.4 percent of the Multaq group, and patients who discontinued ACE inhibitors or ARBs accounted for 5.7 percent in the placebo group compared to 13.2 percent in the Multaq group.

In material released before the meeting, one of FDA's questions for the committee was about the ACE/ARB explanation for the mortality in ANDROMEDA. But the agency appears to have concluded on its own to discount the ACE/ARB theory in the eleventh hour; the question was not part of the final set presented at the meeting.

FDA Division of Cardiovascular and Renal Drug Products medical officer Abraham Karkowsky, giving the agency presentation, noted that such a hypothesis requires that an asymptomatic serum creatinine increase provoked the discontinuation of the ACE inhibitor or ARBs, and argued that the increased mortality in ANDROMEDA cannot be attributed to that reason.

Bad Results Probably Due To Bad Population

Sanofi's third explanation, which the panel unanimously agreed upon, was that the ANDROMEDA results represent a true finding triggered by an inappropriate patient population; because of the broad enrollment, the trial uncovered a "deleterious effect of dronedarone in recently unstable patients who were hospitalized for decompensated heart failure and who did not have non-permanent atrial fibrillation."

Indeed, the crux of ANDROMEDA's problems stemmed from the fact that the study tried to gauge Multaq's effect on major events in high risk patients. ATHENA, on the other hand, gauged only the effects on major events in patients at risk of atrial fibrillation or flutter recurrence.

"There are two options: Either it's due to the instability of the patients or it's spurious," said voting member Michael Lincoff, Cleveland Clinic. "We have to operate under the assumption that it's real. I think it's very reasonable that the patient population was unstable and that's a potentially good explanation for the mortality difference, particularly given the lack of apparent mortality hazard in a lower-risk or less acute population in the ATHENA trial."

Having decided that the patient population was behind the mortality results in ANDROMEDA, the committee concurred that the level of mortality risk in ATHENA in the refined patient population was acceptable and focused on how to craft the indication, as well as how to restrict use to appropriate patients.

Trying (And Failing) For A Mortality Claim

In addition to the proposed indication for patients with either a recent history of or current non-permanent atrial fibrillation or flutter with associated risk factors, Sanofi was hoping to tack on an outcomes claim that the drug has been shown to decrease combined risk of cardiovascular hospitalization or death.

While the committee voted 10-3 to approve the drug for both atrial fibrillation and flutter patients, they were quick to deny the CV mortality claim.

The claim was backed with data from secondary endpoints that both FDA and committee members said were insufficient. The main secondary endpoint of all-cause mortality trended lower in the Multaq treatment arm. Since the main secondary endpoint was non-significant, Sanofi was not supposed to evaluate the additional endpoints of cardiovascular hospitalization alone and cardiovascular death. But the sponsor went ahead with the review anyway, much to the chagrin of the agency.

"It was not a sensible thing to stick all-cause mortality in as first secondary endpoint. There isn't any likelihood that any drug affects all causes of mortality," said Robert Temple, director of the Office of Medical Policy. "We weren't going to give it to them even if they appeared to earn it."

But it was not all bad news for Sanofi, as panelists did come down in favor of a claim that suggested a decrease in cardiovascular hospitalizations.

"The claim should be for reducing cardiovascular hospitalization in low to intermediate risk patients with current or previous history of AFib or flutter," said voting member Sanjay Kaul, Cedar-Sinai Heart Institute, though noting reducing CV hospitalization was "not the most robust of endpoints."

The difference in patient populations in ATHENA and ANDROMEDA may have made it difficult for FDA and committee members to reconcile the studies' diverging outcomes, but it did help in forming the basis for its restricted use. "In effect, [it's] defined a patient who should not receive this drug," said voting member Emil Paganini, Critical Care Nephrology Consulting. "That's a positive thing, not a negative thing. So if you look at it from that point of view ... you've got a very strong set of patients who should not receive this drug."

Committee members were unanimous in restricting Multaq in patients with Class III or higher heart failure and those with LVEF of less than 35 percent. Despite Sanofi's plea that it would be unfair to restrict use in those groups since there was evidence in ATHENA that a subset of such patients benefited, members held their ground. "I would like to make sure the claim is narrow in its scope, that we are very careful about the exclusion of heart failure patients in the labeling," said committee chair Robert Harrington, Duke University.

The committee also recommended that Multaq be contraindicated for initial treatment, and for patients taking potent CYP3A4 inhibitors, Class I and III antiarrhythmic drugs, torsade-de-pointes inducing drugs, and digoxin. The group called for additional post-market trials on dosing, ethnic minorities, long-term data and especially comparisons to amiodarone.

Some Voices Of Dissent

Not everyone was convinced that Multaq should be approved, even if they came to terms with the differences in the study. The three "no" votes shared the same concern than FDA highlighted in its briefing documents - that the safety of Multaq presents a problem that the label alone may not be able to cover.

"I'm not sure based on what we've seen today that the verbiage that would be added to the indication statement would overcome the deleterious effects and differentiation among the patients that would take the drug," said temporary voting member Robert Dubbs, patient representative.

"I agree with that and I add that it doesn't work as well as amiodarone, we have no idea because of a lack of data what these people were hospitalized for, and I don't think it's possible to get this nice, clean clinical trial, but not real-world separation from the ANDROMEDA patients and the ATHENA patients," added temporary voting member Sydney Wolfe, consumer representative.

Multaq's user fee goal date is April 30.

- Lauren Smith (lauren.smith@elsevier.com)

Multaq Panel Resolves Mortality Concerns, Looks At Restricted Claims

More from Archive

More from Pink Sheet