No immunity for Dendreon’s cancer vaccine

(UPDATED: See below.) For almost two months, it has seemed that the FDA might be ready to approve an entirely new attack on cancer — a “cancer vaccine” that immunizes patients with tumor fragments in order to activate the body’s immune defenses against cancerous cells.

That dream suffered a major setback earlier today, when the FDA threw up a fresh roadblock to the approval of Dendreon‘s Provenge, the first such cancer vaccine to go before the agency for regulatory approval. In March, an FDA advisory panel recommended approval of Provenge as a treatment for prostate cancer, despite some reservations about its efficacy in clinical trials. Instead, the FDA issued what is known as an “approvable” letter — a misnomer that actually means the agency won’t approve a drug’s application without additional data. Providing that sometimes requires additional lengthy and expensive clinical trials.

Although the near-term impact on Dendreon’s stock was severe — its shares plunged by 60 percent on the news — chances are good that Provenge or at least one of many other cancer vaccines now in clinical trials will eventually work. To explain why, I’ll set the stage in this post with a walk through the troubled history of cancer vaccines in general and Provenge in particular. In subsequent posts, I’ll take a look at other vaccines currently in late-stage trials and the efforts of venture companies in this area.

First, though, a quick note on terminology: This class of therapeutic cancer vaccines shouldn’t be confused with more traditional infectious-disease vaccines like Merck‘s Gardasil. The Merck product protects against the human papilloma virus linked to cervical cancer, but isn’t really a “cancer vaccine” except in the loosest possible sense of the term.

One of the things that makes tumors dangerous is the way they evade the immune-system surveillance that guards against infection by bacteria and viruses. The goal of most cancer vaccines is to somehow stimulate the body’s defenders to identify tumor cells as enemy “others,” effectively calling in the immune system’s beat cops and strike teams to swarm and overwhelm them. (For a detailed description of how this should work in practice, click here.)

The problem has long been figuring out exactly how to trigger an immune response strong enough to attack the tumor but specific enough to leave other healthy parts of the body alone. There are two basic challenges: Identifying proteins specific to tumors that can be used to inoculate cancer patients, and finding the right way to expose those “antigens” to the guardians of the immune system.

Researchers have tried a variety of approaches, most of which have failed. In recent years, for instance, Antigenics has seen a cancer vaccine directed against “heat shock” proteins extracted from patients’ own tumors fail twice in large trials — once against kidney cancer, and again in skin cancer. A vaccine against pancreatic cancer made by Therion Biologics didn’t improve patient survival, effectively imploding the closely held company. Similarly, a startup called CancerVax pulled the plug on a melanoma vaccine in mid-2005 and later effectively vanished in a reverse merger with the European biotech Micromet.

Provenge, known generically as sipuleucel-T, took a different yet still relatively straightforward approach. Prostate-cancer patients underwent pheresis to filter out certain immune-system cells known as “antigen-presenting cells.” In the lab, researchers then cultured the cells together with a tumor protein called prostatic acid phosphatase in order to “prime” them for attack once they were reintroduced to the patient.

But Provenge’s history has been anything but smooth. The company’s first large-scale trial of the vaccine failed to demonstrate that the vaccine could hold prostate cancer in check, although Dendreon tried to argue that it appeared to work better in men with less aggressive prostate cancer — an after-the-fact analysis that amounted to moving the goalposts during the game. (For a technical summary of clinical-trial statistics and why Dendreon’s strategy didn’t hold water, click here.) The company rejiggered an existing trial to study Provenge’s effect in this particular subgroup of men, only to discover that a followup analysis of the original patients showed that men who received Provenge appeared to live longer than their counterparts. That analysis, however, involved just 127 men, a group so small that no one could be sure if the survival data was merely a statistical fluke.

Although the second trial also suggested that Provenge improved survival, it was also quite small and failed tests of statistical significance. What’s more, neither trial showed that the vaccine could actually slow the progress of the disease — a paradox that concerned experts assembled for an FDA advisory panel in March, although it nevertheless ended up recommending approval of Provenge by a 13-4 vote.

Dendreon’s shares shot up almost fivefold following that recommendation, although naysayers continued to warn that the vaccine hadn’t proven itself. Two no-voting members of the advisory panel wrote letters to FDA — which quickly leaked to the newsletter Cancer Letter — urging rejection of Provenge. In the end, the FDA apparently agreed, although it isn’t yet clear what sort of additional data the agency is looking for. Under one scenario, Dendreon may have to complete a large clinical trial that’s already underway, but which may not report data until 2010.

UPDATE: I’ve revised the description of Provenge’s clinical-trial history for accuracy.

Next: The outlook for other late-stage cancer vaccines.