Earlier this month, DynaMed, an online database used by physicians, added to its Provenge entry a note on the JNCI paper, but calls the concern “not substantiated.” ACS’s Lichtenfeld says the analysis “might inhibit some patients and doctors from going ahead with a very expensive drug.”

Huber says she plans to approach European regulators as they consider Dendreon’s application to approve Provenge.

Investor message boards have lit up in response to the new paper. In February, an anonymous commentator on InvestorVillage.com warned that Huber’s work was about to be published “a few days before our earnings. Her agenda is obvious.”

IFFY STATISTICS

Critics of the new analysis argue the number of cells removed is too small to suppress the immune system. Charles Drake, an oncologist and immunologist at Johns Hopkins Sidney Kimmel Comprehensive Cancer Center, said there is no evidence the placebo men in IMPACT suffered more infections or other effects of a depleted immune system than the Provenge men.

The scientist who led IMPACT, oncologist Philip Kantoff of Dana Farber Cancer Center, said colleagues in immunology “dismissed as nonsense the idea that leukapheresis could hurt individuals.”

He takes issue, too, with the statistics. Dividing the men by whether they are older or younger than 65, he said, is “arbitrary” and to pick apart data retrospectively is a statistical no-no.

Dendreon’s Frohlich also criticizes the statistics: “If you do enough of these (post-hoc analyses) then by chance alone you’d expect to get one positive finding.”

In other words, it is almost always possible to find a subset of patients who do better than others.

When Dendreon divided the men by whether they were older or younger than about 71, he added, they found no red flags.

The FDA agrees that such post-hoc statistical analyses “are exploratory” and their results “must be interpreted with caution, as acknowledged by the authors.” Yet a paid consultant to Dendreon before the IMPACT trial agreed with many of Huber’s concerns.

“The control vaccine used in IMPACT and in the predecessor trial had never been used anywhere for anything and may well have been detrimental to patients,” said Donald Berry of MD Anderson Cancer Center, a leading biostatistician. “Here’s a great way to get your drug approved: Kill the control patients.”

Despite the heated rhetoric, Provenge may go out with a whimper more than a bang. Promising new agents for advanced prostate cancer include an oral drug from Medivation Inc called enzalutamide, in the final phase of clinical trials, and Zytiga from Johnson & Johnson, which won FDA approval in 2011. A vaccine that targets PSA, from Bavarian Nordic Immunotherapy, is in late-stage trials.

Dendreon does not disclose how many patients have been prescribed Provenge. CEO Johnson said that about 70 percent of its Provenge revenue comes from sales to community hospitals and doctors and 30 percent from academic medical centers. Some of the latter decline to use Provenge, deterred by lingering concerns over whether it provides a meaningful benefit.

Three such facilities in the Midwest, contacted at random by Reuters, confirmed they do not recommend Provenge. All asked not to be named for fear of receiving threats.

“It is my policy not to make public comments about this drug,” said one oncologist. Patients who ask for it “are referred to another facility.”

###

By Sharon Begley