Sunday, September 26, 2010

FT.com / US regulator sets October GSK deadline

GlaxoSmithKline must by next month provide US regulators with details of how it will investigate concerns of bias in a clinical trial it ran for its diabetes drug Avandia.

By late October, the UK pharmaceutical group must submit a timetable to the Food & Drug Administration for a review by independent experts of the long-running study, called Record, that GSK claimed showed no increased cardiovascular risk to diabetics taking its medicine.The action is one of a number of concerns about the conduct of clinical trials raised by Avandia which will trigger reflection by international drug regulators following a near-blanket ban on the drug imposed last week.

The Record trial has highlighted a rift between regulators. The European Medicines Agency appeared tougher than its US counterpart when it imposed the study a decade ago, designed to identify any safety concerns, as a condition of first approving Avandia.

Yet a detailed analysis by the FDA issued this summer by Thomas Marciniack, its cardiovascular medical team leader, described Record as “a complex, flawed design and a flawed execution”. He said the FDA would never have approved the study and he highlighted “completely unacceptable” issues in the way some patient data were handled, including breaches of “firewalls” designed to ensure independence between GSK and Quintiles, the clinical research consultancy that carried out the detailed data collection.

Whereas the EMA largely relies on summary results of clinical trials prepared by pharmaceutical companies to make its assessments, the FDA has more capacity to study the underlying raw data.

That greater scrutiny allowed Mr Marciniack to identify anomalies, but also required large amounts of time: he was only able to sift through a small quantity of the underlying records.

GSK maintains that the anomalies Mr Marciniack unearthed did not change the overall conclusions of Record which, it argues, remain “robust and reliable.”

Nevertheless, the outcome of its new independent review may spark transatlantic reflection on how exhaustively to analyse the results of trials submitted by drug companies.

A second trial concerning Avandia that has sparked concern was called 172, which was not submitted to the FDA. Some reports suggested GSK had tested its own drug against the rival Actos, found the alternative to be safer and suppressed the results.

In fact, 172 did not directly test Avandia at all. It was a small study designed to see whether a comparative trial would likely find in Avandia’s favour, by studying cholesterol levels in patients treated with Actos and comparing these results with separate studies it had already conducted using Avandia.

That raises an ethical issue of whether companies should provide regulators with all findings of tests they conduct, even when they do not directly relate to their own drug.

A third issue that emerged during the Avandia scrutiny was how far regulators can rely on imperfect evidence. Some research identified at least as early as 2007 a higher risk of heart attacks in patients taking Avandia than Actos.

Such “meta-analyses” – which attempt to tease out findings from other clinical studies never designed for that purpose or directly comparable – are imperfect. Yet the more comprehensive studies concluded in recent months that triggered the regulators to pull Avandia came up with very similar results. That suggests some meta-analysis may need to be taken more seriously.

The option taken by regulators was instead to seek more robust, conclusive evidence by running a first formal trial called Tide, comparing Actos with Avandia to identify any differences in heart attacks. But with the imperfect data already suggesting that Avandia carried a greater risk, scientific outcry led the regulators to suspend the trial.

Waiting for more detailed and reliable analysis before approving Avandia would have required much more time, and would have been so costly that the drug developers might never have agreed to proceed at all.

In the case of Avandia, that might have been a better solution. The downside more generally would be that drugs with potentially great benefits to patients may never make it to the market.

via ft.com

Posted via email from Jack's posterous

No comments: