BioCentury's websites will be down for upgrades starting at 9 p.m. PDT on Monday, August 26. We expect the downtime to last no more than 6 hours, and we apologize for any inconvenience.

12:00 AM
Jul 19, 2010
 |  BioCentury  |  Regulation

FDA's Avandia problem

FDA faces Avandia decision without clear guidance from its 2nd advisory panel

FDA officials walked out of an advisory panel meeting on Avandia rosiglitazone last Wednesday evening with the same problem they came in with on Tuesday morning: how to handle a drug that has been accused but not firmly convicted of causing an increase in cardiovascular events in a population that has a very high rate of heart attacks and strokes.

The committee split into three almost equal groups that recommended mild, moderate and extreme regulatory actions: 10 members voted to keep Avandia on the market with existing or updated labeling; 10 advised adding additional limits on its use, such as restricting prescribing to certain physicians or requiring special physician or patient education; and 12 wanted FDA to compel GlaxoSmithKline plc to withdraw the drug from the U.S. market.

While a majority of the agency's advisors recommended keeping Avandia on the market, it is far from certain that FDA will do so.

After perusing over 1,000 pages of briefing documents, sitting through more than 18 hours of testimony and deliberations, and viewing at least 500 slides, members of the Endocrinologic and Metabolic Drugs and Drug Safety and Risk Management advisory committees did little to resolve the acrimonious dispute between FDA's Office of Surveillance and Epidemiology (OSE), which has long called for Avandia's withdrawal, and the Office of New Drugs (OND), which has argued its benefits outweigh its risks.

FDA's internal politics, as well as intense external political pressure, make it virtually impossible for the agency to simply revise the warnings on Avandia.

At the same time, it is not obvious how to keep the drug available to some American patients under highly restricted conditions that theoretically would reduce the risk it would cause ischemic cardiovascular events.

Unfortunately for FDA, the committee gave few clues on how to design a risk minimization strategy, or provide criteria that could be used to determine which patients should be eligible to receive the drug.

For OND staff who continue to believe Avandia's benefits outweigh its risks, the urgent challenge thus is to develop a scientifically defensible risk-minimization plan. Failure to do so could lead the agency's top management to feel it has no alternative other than siding with OSE's adamant calls for withdrawal.

One potential solution would be to restrict use to patients who fail or can't tolerate Actos pioglitazone, another thiazolidinedione from Takeda Pharmaceutical Co. Ltd., and to restrict prescribing to endocrinologists.

In dealing with the mounds of data and conflicting regulatory recommendations the Avandia controversy has generated, FDA and the advisory committee did break new ground in areas that are certain to become recurrent topics in the age of drug safety. These include a consensus that unique safety risks are not acceptable in the absence of unique benefits, and an emerging, if grudging, acceptance of observational and epidemiological safety data as the basis for regulatory decisions.

Broken RECORD?

Although FDA leadership and the political party in power both have changed, much of the data and most of the arguments made at last week's meeting were presented at a similar panel in July 2007 (see BioCentury, Aug. 6, 2007).

The two biggest differences in terms of scientific data were the availability of the completed RECORD study, and completion of a large epidemiological safety study comparing the CV safety of Avandia and Actos.

In 2007, the committee saw interim RECORD data that supported GSK's assertion that the drug's cardiovascular effects are similar to the standard first-line diabetes regimen, a combination of metformin and a sulfonylurea.

The committee also heard preliminary results of an observational study conducted by David Graham, associate director for science and medicine at OSE, based on private insurance claims data. It linked Avandia to increased CV events.

GSK hailed the completed RECORD results when they were published in June 2009, noting the upper bound of the 95% confidence interval for CV risk indicated a maximum of a 16% increase in CV risk, which was considerably lower than the maximum 30% increase permitted in diabetes drug guidance FDA released in 2008 (see BioCentury June 8, 2009 & BioCentury, March 10, 2008).

RECORD also confirmed meta-analyses that found no increase in CV harm for Avandia compared to active therapy.

If FDA or the advisory committee had accepted the RECORD results at face value last week, the meeting would almost certainly have resulted in recommendations far more favorable to Avandia.

The reason the advisory committee decided it could trust neither the RECORD results nor its sponsor can be boiled down to two words: Thomas Marciniak.

Marciniak, a medical team leader in OND's Division of Cardiovascular and Renal Products, prefaced his remarks by telling the committee that when it comes to Avandia, there are three "factions" at FDA.

One faction, Marciniak said, "was involved with the approval of this drug" and had argued in 2007 that it should remain on the market. "It is not surprising they have a bias toward not reversing themselves," he said.

"The other faction," according to Marciniak, "has been saying since 2007 'this is a dangerous drug, it should be removed'" from the market and is equally biased.

Marciniak declared that the third faction consists of himself: "You have me. I have not been involved with rosiglitazone prior to the fall of last year. I have a clean slate. I amount to a tie-breaker as to whether this is a good or bad drug."

Marciniak's presentation did not, however, directly address the "good or bad" issue by assessing overall risk/benefit balance. Rather, he attacked the design and especially the conduct of RECORD.

Marciniak reviewed about 12% of the 4,450 RECORD case report forms and found problems with 13% of them. Problematic forms, he reported, were four times more likely to favor Avandia than the comparator arm; endpoint problems were twice as likely to favor Avandia.

RECORD's "estimates of CV risk must be considered to be lower bounds for risk rather than precise estimates with statistically valid CIs," Marciniak concluded.

Marciniak quoted minutes from a RECORD steering committee indicating "unrestricted availability of unblinded treatment code within Quintiles and GSK," which he said strongly implied GSK had been able to intentionally skew the results.

Philip Home, professor of diabetes medicine at Newcastle University and chair of the RECORD Steering Committee, said the minutes were incorrect and that...

Read the full 5229 word article

User Sign in

Trial Subscription

Get a 4-week free trial subscription to BioCentury

Article Purchase

$150 USD
More Info >