Reviewing Proveng’s chances of FDA approval from the point of view of investors, Matthew Herper gives an overview useful for patients as well. Herper writes:

The FDA has a legal deadline of May 15 to make a decision–although it is possible to break or extend it. But buying the stock in front of this deadline is probably highly risky. There is a reasonable chance the FDA will decide to go against the panel vote and ask for more data. And the downside to the stock if the FDA does delay approval by asking Dendreon for more data is probably much bigger than any potential gain.

Why? For one thing, an approval might be priced in. Two years ago, Geoffrey Porges, an analyst at Sanford C. Bernstein, analyzed the likely performance of biotech stocks following FDA decisions.

Some 70% of the time, Porges found, drugs were approved, and shares in their makers gradually rose. But when a drug failed, stocks cratered 40%. The net result is that betting that drugs are going to get approved is a good way to lose money. In contrast, if you shorted stocks before FDA decisions, you’d make money.

Another problem is that the FDA is much less likely to follow the whims of its advisory committees than many investors seem to think. The agency does follow the advice of these committees most of the time, but there are plenty of reasons to believe Dendreon might be an exception.

Dendreon’s application to the FDA contains two late-stage studies with a combined total of 225 patients. That’s tiny; many drugs are tested in studies of thousands of patients. Both studies missed their main goal, but the more recent one–a three-year, 124-patient study–shows that patients on the drug lived, on average, an extra four months.

But statisticians, and the FDA, tend to toss out studies that don’t make their main goal. To cancer patients, this can sound insensitive. But there’s a real reason for it. When a study gives a result that scientists didn’t predict ahead of time, it’s simply more likely that the result happened by chance.

The panel did weigh those risks, and it still recommended Provenge by a 13-4 vote. One study shows that when an FDA panel recommends approval, the FDA will approve the drug 97% of the time. Both a concerned investor and Dendreon’s outside public relations firm point to this research, conducted by the National Research Center for Women & Families.

That statistic is misleading, though, because the study counts approvals that happened months or years after the panel voted. Nobody expects the FDA to reject Provenge outright. Instead, the FDA would probably grant an “approvable letter,” basically saying it will approve the drug if certain conditions are met. Dendreon is already conducting a 500-patient study that will give a much clearer answer as to whether Provenge works. If the drug is approved, it may make it difficult to conduct that placebo-controlled trial.

Also, the NRCWF report basically argues that the FDA approves drugs too often. It is exactly the kind of criticism that could lead to the agency becoming more reticent. And it picked committees at random, missing the advisory committee for cancer drugs, where a lot of controversies happen. Also, it doesn’t include two cases where the FDA did go against a committee.

An advisory committee recommended that Merck’s (nyse: MRK – news – people ) Vioxx be put back on the market; the FDA found a way to not bring it back without actually contradicting the committee’s recommendation. And a committee recommended approval for Pargluva, a diabetes drug from Bristol-Myers Squibb (nyse: BMY – news – people ). After safety issues emerged, the FDA asked for a large study, basically killing the drug’s chances.

If the agency wants to ask Dendreon for more data, it certainly has some outs. The FDA changed the wording of a question during the panel, so instead of just saying Provenge should be approved, it said there was substantial evidence for its efficacy. Also, the panel that looked at Provenge was the one for cellular, tissue and gene therapies, not the panel that normally deals with cancer drugs.

Richard Pazdur, head of the Office of Oncology Drug Products, has set a high bar for judging the effectiveness of cancer drugs. His office is completely removed from the one that convened the Provenge panel.

Many of the panelists on the cellular, tissue and gene therapies panel were researchers, not doctors who treat patients. One of the “no” votes was Howard Scher, of the Memorial Sloan-Kettering Cancer Center, who has made a career of designing trials to test prostate cancer drugs. The Cancer Letter, an industry newsletter, obtained a confidential letter Scher sent to the FDA arguing that Provenge should not be approved. Memorial Sloan-Kettering says that Scher had nothing to do with making the letter public.

In the letter, Scher voices four concerns: The recommendation to approve fell short of regulatory requirements; the statistics are “inadequate” to determine whether patients were helped; the data on safety are incomplete; and the advisory committee seemed to be using a different set of criteria compared to those for other cancer drugs.

The panel vote does make an approval more likely. But the FDA can go against its panels, and it has in the past. There are plenty of reasons to delay approval until a big safety study emerges. And the pain of a rejection could be worse than the pleasure of approval. Until Dendreon gets a fax from the FDA, what happens next is anybody’s guess. (Forbes, The Danger In Dendreon, April 20)