If gefitinib (Iressa) becomes the first oncology drug to be withdrawn after receiving accelerated approval from the FDA, the impact might not just be felt by its manufacturer AstraZeneca.

With the FDA facing pressure from other quarters claiming that it prizes speed before safety, researchers are worried that this could affect all cancer drugs going through the accelerated approval process.

The potential benefit of novel treatments for patients with fatal diseases means that the FDA is willing to accept treatments on the basis of positive surrogate endpoint data rather than clinical effect.

Iressa was approved in 2003 with data from single-arm Phase II studies showing that it reduced tumour proliferation in patients with advanced non-small-cell lung cancer who had not responded to other treatments.

As there is always a chance that surrogate endpoints might not correlate with clinical benefit, the so-called subpart H regulatory guidelines of the accelerated approval process stipulate that the company continues with clinical trials after the drug is marketed.

AstraZeneca was asked to run a confirmatory study to show that Iressa's ability to shrink lung tumours would lead to improved survival, and this failed.

We need to consider whether it is better for patients to delay approval of a potentially valuable drug, or whether to approve and monitor it. , Len Lichtenfeld, American Cancer Society

The FDA hasn't completed its evaluation of Iressa, and so couldn't comment on whether it will ask AstraZeneca to withdraw Iressa from the market, or whether it will reassess the accelerated approval process.

But it is entirely rational to decide to trade off the risk of a false positive for the general good of earlier approvals, says Stephen George, Professor of Biostatistics at Duke University Medical Center. “Despite all the hype about potential early markers of efficacy in many diseases, there is at present no way to permit such early decisions without increasing the risk of a wrong decision.”

One way of avoiding erroneous conclusions from Phase II trials would be to require Phase III interim data, as is the case for accelerated approvals of HIV drugs.

“This was the case for accelerated approval of oxaliplatin in colorectal cancer,” says George. “There was a built-in analysis of a surrogate endpoint used for accelerated approval with the same study used as a more definitive analysis after further follow-up to confirm or refute the early finding.”

“I hope that the FDA do not overreact and do not abrogate the accelerated approval approach,” says Len Lichtenfeld, Chief Medical Oficer at the American Cancer Society. “We need to consider whether it is better for patients to delay approval of a potentially valuable drug, or whether to approve and monitor it.”

Iressa has shown that the monitoring system can pick up drugs that fail to show a survival benefit, says Lichtenfeld. “It may not be a perfect system, but you can't let the perfect be the enemy of the good.”