CardioExchange Archive

The FDA has released a 538-page briefing document for an advisory panel meeting on Wednesday and Thursday that will reassess a key clinical trial and reconsider the fate of the now-tarnished former blockbuster diabetes drug rosiglitazone (Avandia, GlaxoSmithKline). (Click here for the FDA documents.) As reported last week, the re-adjudication of the RECORD safety trial performed by the Duke Clinical Research Institute (DCRI) confirmed the initial finding of the trial that rosiglitazone was not associated with an increased risk for cardiovascular events.

For the most part, the FDA documents released today express strong support for the DCRI re-adjudication. But one FDA official, Thomas Marciniak, remains highly critical of RECORD and says the trial data and, therefore, the analysis of the data from GSK and DCRI, are completely unreliable. All parties agree, however, that the fundamental underlying design flaws of RECORD —  in particular, its open-label design — mean that data from the trial will never provide definitive assurance about the safety of rosiglitazone.

The roster for the panel may also provide some evidence to support assertions by Steve Nissen, the most prominent Avandia critic, that the makeup of the advisory panel was designed to “whitewash” the FDA’s role in the Avandia controversy. By an initial rough count, most of the 16 of the 33 panel members who sat on the 2010 panel and who will not be participating in this week’s panel originally voted either to restrict or withdraw the drug. Nissen has also said that he was denied a request to make a formal presentation to the panel.

One FDA reviewer said the DCRI review of the mortality results in RECORD was “well-conceived and comprehensive” and “no stone was left unturned.” But the same reviewer states:

There is no amount of analytical rigor that can compensate for a weak trial design that is exacerbated by elements of poor execution, both of which afflicted RECORD. Its open-label non-inferiority design was simply problematic, especially for ascertainment of non-mortality MACE during trial execution… Thus, while we agree with the analytical findings of the DCRI mortality re-analysis, we would emphasize that RECORD’s design irreparably hampers its ability to characterize definitively the CV risk of rosiglitazone.

The panel may well accept the findings of the re-adjudication and the FDA analysis. In that case, the terrifying specter looming over the FDA and the rest of the medical establishment — that not just rosiglitazone but the entire drug development and approval process is fundamentally flawed and unreliable — will be put to rest, at least for now.

But acceptance of the re-adjudication does not necessarily mean that the FDA or the panel will want to remove the current strict restrictions on rosiglitazone or that the drug will once again be commercially viable. Certainly the chances for a rosiglitazone rebirth have now improved, but given the remaining serious concerns about RECORD, the panel may well prefer to err on the side of caution and retain most of the restrictions on rosiglitazone.

After two days of discussion the committee will be asked to vote on only one question: whether to remove, continue, or modify the current restrictions on rosiglitazone use, or whether the drug should be withdrawn from the market.

An updated meta-analysis from the FDA of randomized trials continues to suggest an increased risk associated with Avandia, though the results are far from conclusive. FDA reviewers also found evidence hinting that rosiglitazone was less safe than pioglitazone (Actos, Takeda). The FDA reported that rosiglitazone use declined from 5.1 million prescriptions in 2008 to about 12,600 prescriptions last year.

A Bitter Feud

Buried in the massive document is a bitter feud between an FDA rebel, Thomas Marciniak, and his bosses and other senior officials in the FDA’s drug division. (Marciniak was the subject of a recent New York Times article that publicly revealed another dispute with some of the same officials over the safety of angiotensin receptor blockers.)

Marciniak unleashes a barrage of rhetoric and data in an attempt to discredit RECORD and the DCRI. He argues that the DCRI is inherently unreliable because it relies on industry funding. He also points to DCRI’s involvement in the ARISTOTLE and PLATO trials, which have come into question. Finally, he invokes the name of a disgraced Duke researcher (but who had no connection with DCRI):

I would have thought that the two words ‘Anil Potti’ are sufficient for convincing anyone that Duke University is a poor choice for a contractor whose task it is to confirm the integrity of scientific research.

The panel will undoubtedly spend a great deal of time assessing Marciniak’s criticism of the data. The FDA goes to great lengths to respond to Marciniak. In one extraordinary memo, top FDA officials Norman Stockbridge, Ellis Unger, and Robert Temple write:

Dr. Marciniak’s memoranda were neither assigned by his supervisors in the Division of Cardiovascular and Renal Products nor solicited by the Division of Metabolic and Endocrine Product… He did not seek supervisory review or concurrence prior to filing his memoranda… Thus, although Dr. Marciniak’s documents are part of the official FDA record… his views do not necessarily reflect official FDA positions.

…

At CDER, we encourage, even cherish, scientific debate. Although Dr. Marciniak did not request supervisory concurrence for his reviews, we support and respect his right to express his scientific views. Regrettably, however, some of the views expressed in his memoranda go beyond scientific and regulatory principles and issues. They include disparaging remarks about FDA staff and review units, as well as serious allegations regarding the competence and behavior of DCRI and GlaxoSmithKline, and, indeed, all investigations and organizations that are paid by a drug company for any clinical trial conduct or analysis activity. The Division of Cardiovascular and Renal Products and the Office of Drug Evaluation-I reject these personal remarks…

…

In essence, Dr. Marciniak alleges that parties involved in the sponsorship, the re-adjudication, the statistical review, and the inspection of RECORD were all unreliable because of contractual arrangements, flaws in some unrelated matter, or both. We disagree with his assessment.

…

Apart from the content and basis for Dr. Marciniak’s critique, we find much of the language used in his memoranda regrettable. A number of passages insult our own review staff (Dr. Preston Dunnmon), our Biostatistical staff, DCRI, and the applicant. We find this language unprofessional, inappropriate, and not commensurate with our standards. As noted, we welcome and thrive on honest debate, but it is unprofessional to insult someone because he or she disagrees with you.

…

After re-adjudication, we do not find a cardiovascular ‘signal’ of concern in RECORD. The mortality data are, in fact, reassuring.

The FDA reviewers also point out that although Marciniak has been extremely critical of other analyses of RECORD, his own evaluation suffered from serious flaws, even, notably, a less than rigorous blinding process. According to one reviewer:

Regrettably, the case reviews in his original consult were not blinded; although he states that some type of blinding occurred, it appears that he himself did the redactions, then followed with his own review. This is not an acceptable review procedure if one wants to put forth one’s review work as blinded.