Biotech Values

[AlpineBV_Miller]

Most of what ends up being provocative is overshadowed by significant factual errors...

• The problem with the trial data submitted for the initial panel meeting was size (very small trial and structure). The company blended results to get their final data in a way not valid for FDA statisticians.

Nope. The problem was survival was not the primary endpoint. CBER doesn't require two trials for approval, so DNDN was asking for approval based upon 9901. Pooling and 9902a were used to help understand the strength of the overall data package and as "supporting" data.

The trial was not well constructed, but it was not poorly constructed either.

Because DNDN management is outrageously aggressive and insensitive to the FDA and I am openly skeptical of how they constructed the data. And given what I know of the FDA, that agency will be even more skeptical given management's approach to the approval process. The dunderheads (that is a nicer word than moron) at DNDN are already putting out Provenge sales projections - not a smart move and something the FDA frowns on for drugs not yet approved.

He apparently believed Herper's idiot Forbes piece. The trial has gone through two edits in ten years and the FDA amended the SPA accordingly. Assuming you don't believe Dendreon is lying to their shareholders like SPPI did on satraplatin,'s SPA there is no basis for a charge the trial is "poorly constructed." Prior to the release of the IMPACT data, even bears acknowledged the trial was well designed.

Say what you will about Dendreon management, but they have NEVER been overly aggressive about the data (partnerships and timelines, yes. Data, no). Every time they have characterized the data, with one exception, they have undersold the data. That exception was saying in like 2004 or 05 the patient population in 9902a was similar to the population in 9901. In fact, the 9902a population was much sicker.

Dendreon has not released sales data. They released manufacturing capacity data. There is a BIG difference.

Provenge is not easily extensible to other tumor types. Moving Provenge to another patient set - such as breast cancer - requires new research and development work and wholly new trials and those will take many years.

The ONLY indication Provenge is useful for is prostate cancer. The PLATFORM is extensible, but requires different drugs.

Are there risks? Sure. The author mentions overly optimistic sales projections from Dendreon. I think a bigger risk is Dendreon sandbagging initial sales estimates and freaking people out who still believe it will sell a billion in 2010 and $3 billion in 2011.

The drama queen bears that follow Dendreon are certain to blow up every hiccup in the manufacturing process. Every missed patient appointment, every weather situation around the Jersey plant. They'll obsess over when the J-code is issued. Noise = price volatility, usually to the downside -- at least historically where DNDN is concerned.

Private insurance reimbursement will be an issue, especially if they are charging the price we think. 70% of the patients are likely to be on Medicare, however, which has little or no discretion on reimbursement by Congressional rule. Nothing in the House or Senate version of healthcare reform will allow that to change.

Reimbursement until the J-code is issued could cause a delay in cash flow at the doctor's office. Dendreon will have in place at launch a 3rd party financing mechanism to bridge that cash flow gap.

Provenge was designed with manufacturing in mind. It was designed by people who watched a company they loved be swiped for a song because of an inability to solve manufacturing issues. They have spent the entire 10+ years we have covered the company creating a system to make sure that doesn't happen again. That's worth knowing, I think.