The Real Story Behind FDA’s Delayed Approval Of Eliquis

With what seemed to be stellar results from the ARISTOTLE trial, showing the first superiority over warfarin on bleeding and mortality for a novel oral anticoagulant, Bristol-Myers Squibb/Pfizer’s Eliquis (apixaban) was expected to have a clean trip through FDA.

So why was agency approval of the third novel oral anticoagulant to come down the regulatory pathway in recent years delayed by nine months?

The answer, which was largely hidden from investors and competitors, boils down to study conduct and oversight – things that should have been a piece of cake for experienced sponsors.

It turns out that the much-ballyhooed, 18,000-patient ARISTOTLE trial had a few problems, according to FDA review documents that are dissected in the June issue of Elsevier Business Intelligence’s Pharmaceutical Approvals Monthly, part of a regular series of drug review profiles (see the lead story, free for the next 30 days, here).

What were the problems? Well, for one thing there was documented evidence of fraud by employees of BMS and its contract research organization, PPD, at a Chinese study site. It seems these individuals altered source records ahead of an FDA inspection to cover up good clinical practice violations.

FDA’s need to further investigate this issue, as well as the data integrity for other Chinese sites and the impact on the overall ARISTOTLE results, led to a three-month extension in the original PDUFA date.

Publicly, BMS/Pfizer said only that the review extension resulted from its submission of a “major amendment to the application.” An accurate statement? Absolutely. But the fact that this “major amendment” comprised a more detailed accounting of the fraud was a juicy, and likely market-moving, tidbit not shared with the public at-large.

To its credit, BMS discovered the fraud and reported it to the FDA, and the alleged perpetrators were terminated. In contrast, the agency had to root out on its own answers to the second major problem that delayed apixaban’s approval – dispensing errors in ARISTOTLE.

Buried on page 88 of the clinical study report was a statement that 7.3% of subjects in the apixaban group and 1.2% of subjects in the warfarin arm received “a container of the wrong type” of medicine at some point during the double-blind, double-dummy study. This overall high rate of dispensing errors, and the disparity between treatment arms, troubled FDA, in part because these figures were based only on the sponsor’s analysis of one incomplete source of data.

FDA believed further investigation into the true rate of dispensing errors was warranted. Furthermore, agency reviewers seemed incredulous that the unusual number of medication errors failed to prompt a “serious inquiry” by the sponsor prior to NDA submission and that such errors occurred throughout the course of the trial without meaningful corrective measures, suggesting shortfalls in trial oversight.

So annoyed were agency reviewers by the whole situation, including BMS/Pfizer’s partial and evolving responses to FDA’s questions, that the team leader on the application said the NDA would have received a “refuse-to-file” letter had agency staff known about the dispensing errors issue at the time of submission.

Ultimately, FDA issued a “complete response” letter specifically directing the sponsor to get to the bottom of the problem – not that you would have known this from the sponsor’s public statements.

In a press release, BMS/Pfizer said only that the letter requested “additional information on data management and verification from the ARISTOTLE trial.” Again, not a falsehood, but also not exactly the type of information that would have been helpful to assessing what was really going on with apixaban’s prospects for a near-term approval.

Ultimately, the companies submitted data that convinced FDA reviewers that even under a worst-case scenario, the dispensing errors would not have disturbed the key efficacy and safety findings in ARISTOTLE.

So, all's well that ends well for BMS and Pfizer, right?

Well, not exactly. Eliquis failed to gain a coveted mortality benefit claim in the Indications statement, which would have set it apart from its two competitors who beat it to market, Boehringer Ingelheim GMBH’s Pradaxa (dabigatran) and Bayer AG/Johnson & Johnson’s Xarelto (rivaroxaban) (see PAM's analysis of how FDA reviewers picked apart the statistical significance here [$]).

Eliquis generated just $22 mil. in its first full quarter on the market, according to Bristol's first-quarter earnings report.

-- Sue Sutter (s.sutter@elsevier.com)