THINK THERE’S TOO MUCH MONEY IN POLITICS?
I do! Click here to help fix it.
From the Sacramento Bee, by Bruce Mehlman and Larry Irving (thanks, Peter):
The impending exaflood of data is cause for excitement. It took two centuries to fill the shelves of the Library of Congress with more than 57 million manuscripts, 29 million books and periodicals, 12 million photographs, and more. Now, the world generates an equivalent amount of digital information nearly 100 times each day. The explosion of digital information and proliferation of applications promises great things for our economy and our nation, as long as we are prepared.
So I got a lot of blowback from yesterday’s post, as you might expect.
It hit three themes (plus one email saying the drug isn’t needed anyway – just take aspirin and eat curried cauliflower):
1. Two of the doctors recommending against approval of Provenge have big conflicts of interest.
2. The question was changed, yes, but to CORRECT it for what FDA standards say it should be.
3. The second trial people lived shorter than the first trial placebo people, but the second trial people were sicker. Apples to apples, Provenge beat placebo both times.
I guess there are two issues here:
a) In an ideal world, should the FDA approve it?
b) In the real world of current law, WILL the FDA approve it? A much narrower question, and maybe heartless, but certainly relevent to the stock price.
Here’s a sampling of the emails, only one of which, in red, was positive . . . followed by my friend’s response.
Kerry: ‘Provenge works and the FDA only denied it do to politics and the Conflict of Interest Doctors, Scher and Hussain.’
Richard Lake: ”Substantial Evidence’ IS the FDA standard and if you don’t know that you have no business writing about it and if you did know that you are no better than a paid basher on the DNDN message boards. I personally believe you are ‘short’ ethics.’
Richard Berman: ‘You stated that the FDA changed the efficacy question in the Advisory Committee panel on Provenge. That is correct, however, they changed the question to fit the statutory and regulatory language and to be consistent with the language used for virtually all other cancer drugs. The original language, ‘establish efficacy’ appears no where in the statute or regulations. It is a very high standard, and to scientists is even higher. Remember, these are people to whom gravity is still a theory. The changed language ‘provide substantial evidence of efficacy’ is the statutory and regulatory language and is the language used in other cancer advisory committees. The question was corrected, not changed.’
Richard: ‘I agree with your conclusion regarding DNDN but not with your reasons (or, I guess, your friend’s reasons). For example, he mocks the company for not knowing how its drug works. In fact we don’t know how many drugs work. Phrases like ‘the exact mechanism of action is unclear’ appear with frightening frequency in drug applications. The head of Genetech describes oncology research as a series of ‘semi-rational leaps of faith.’ I try to remember that the FDA is not a drug testing organization but a statistics analyzing one. Provenge’s box score is not too good. FDA Standard: Two trials meet predefined targets with 95% level of significance each, resulting in a composite level of 99.5%. Provenge results: Two trials, neither of which met predefined target. So, after special pleading: Two trials, one of which met post hoc defined target with 95% level of significance but the other still didn’t so, after special pleading: One trial and a meta-trial which combined the two trials into a single data set (sort of like copying from the person sitting next to you) meet the post hoc defined targets with 95% level of significance each resulting in a composite level of who knows?’
Thomas H. Jones: ‘Please go read Dendreon’s 10K per 2007 for the REAL facts about Provenge (access at dendreon.com 10K from SEC filings on website). And please suggest that ALL your readers do the same so they have the facts for themselves. ALL the trial results your ‘friend’ referenced are there for the viewing and they are all POSITIVE . . . The average Overall Survival was 4 1/2 months, with some patients living for years. Eduardo Garcia, who spoke at the FDA advisory council meeting, is a 6+ year survivor.’
Trond Hildahl: ‘You are probably going to get a lot of hate mail – there are some fanatical Dendreon fans. Not all of us are foaming at the mouth, please believe me – but I do want to respond to a couple of your ‘shorting’ points. The FDA’s own requirements state the proper wording should be ‘substantial efficacy.’ Yet the question was worded ‘establish efficacy.’ . . . Maha Hussein, who you mention, actually fell asleep during the patient testimony session! This is not knowable from reading the transcript. Also, Hussein and Dr. Scher, 2 of the 4 NO votes, both have enormous conflicts of interest. Information that has come out since show that both never should have been allowed to serve on the advisory committee. These Drs. BOTH set a new low also by writing letters to the Cancer Letter and then ‘leaking’ them to the press. I could go on and argue a number of your other points, ESPECIALLY regarding the second trial and the number of deaths in the Provenge arm versus the placebo – I think you are flat out wrong in your conclusions . . . but I am not a statistician and this is already too long. Keep in mind the FDA recently advised that they will accept interim trial data from 9902B which will be available in mid 2008. Don’t stay short too long!’
Kevin M. Ward: ‘Maha Hussain – deep conflicts of interest, both from funding for her investigative work AND personal investments. Also, an expert in CHEMO, but NOT in immunotherapies. Should not have been on the panel. . . . Good luck with your short position. As of right now, it’s moved 23 cents against you. The day all this gets resolved and Provenge gets its approval, any short will be toast.’
Matt Kim, MD: ‘The second trial was indeed supportive of the first, the skewed data on first read was shown to be caused by the Provenge group having sicker patients than the control untreated group, on later analysis which the FDA accepted as proper and requires the same analysis for the ongoing trial the second study again like the first showed clear evidence of longer survival . . . I obviously don’t know your Harvard friend, but he is talking out his rear and that is verifiable if you have the integrity to check. I realize this statement is harsh but you put absolute lies in print on your website. Please don’t let your name be associated with lies.’
☞ Get the gist? Well, here’s the thing. It’s possible these folks are right in buying the DNDN line, and that my friend is wrong. But he wasn’t born a DNDN short. He goes long drug stocks from time to time, also, and could have gone long this one. Before he goes long or short, he does a lot more analysis than most investors . . . and he has a great deal of experience understanding FDA procedures.
Confronted with the emails above, he offers the following:
(1) There were two doctors who were on the panel and who voted no (among the 4) who subsequently published letters to the FDA in a trade journal known as the Cancer Letter. One was Maha Hussain, the chair person of the Oncology Drugs Advisory Committee.
The second is Howard Scher, head of prostate cancer at Memorial Sloan Kettering. That speaks for itself.
The third doctor who published a letter opposing the approval was a statistician who was asked by the Biologics division to be on the Provenge panel, but was unable to attend the meeting.
All three of the letters explain in detail why the data for show no evidence of efficacy, citing lots of data.
(2) The question “Establish Efficacy” was worded by the FDA. It is published on the website. The original testimony from the division director, Celia Witten, is that this is the question the FDA wanted. What actually happened is that three doctors voted “no,” so then the chairman asked if the question could be changed. Dr. Witten said that, well, this was the question they wanted answered, though the committee could also answer a different question. They then went back and asked the fourth doctor, who voted “no.” Then Dr. Witten said that the law uses the words “substantial evidence of efficacy,” so she and the chairman changed the question to “substantial evidence of efficacy.” All of this can be found in the last 20-30 pages of the transcript. The FDA has a technical definition of “substantial evidence of efficacy.” There is a difference in what this means from a regulatory standpoint and what a guy on the street (or an orthopedist–look at the background of the people on the panel, found at the beginning of the transcript) might think it means.
(3) The entry criteria for the two trials were the same. Thus, the outcomes should be the same. It also turns out that the patients who received Provenge in the first trial were much healthier than those who received the control, and this difference could easily explain the apparent difference in survival in the first trial.
(4) In order for a trial to make a claim, it must demonstrate statistical significance on the primary endpoint. Dor Pharma [a different company with a different drug recently rejected] did not do this. IDM [ditto] did not do this. Dendreon did not do this. IF you do not do this, then, as noted for Dor (and all other companies), any additional data that might be generated from the trial is useful only for generating hypotheses and planning additional trials. It is “post-hoc” analysis. Richard Feynman in The Meaning of It All has an excellent discussion in the second lecture about why post-hoc data analyses are never valid in statistics. The FDA is following Feynman (the Nobel prize winner) in this regard. In reviewing IDM, the FDA concluded that since it did not meet its primary endpoint, it did not show “substantial evidence of efficacy.” Ditto for Dor. Ditto for Dendreon. Ditto for Pharmacyclics (PCYC), which received a rejection letter from the cancer division of the drug division of the FDA earlier this year for a product to treat brain mets. Ditto for Neoprobe (NEOL), which got a rejection letter for brain cancer for the same reason.
The FDA has not approved Provenge, will not approve Provenge, and should not approve Provenge. If Provenge would like an approval, it should demonstrate statistically valid evidence of efficacy, like all other approved drugs.
☞ He believes DNDN has nothing, sadly, and will gradually fade away. I, obviously, have no clue. I just know he could be wrong – but generally is right.
Certainly, very little ‘smart money’ expected the advisory committee to recommend approval, or the stock would not have been at $4 for so long just before it did (and jumped to $24). And, at least so far, that smart money has been right: the FDA – headed by a prostate cancer expert – has not approved the drug.
Meanwhile, from PeterK: ‘Provenge’s ineffectiveness against prostate cancer is no loss except to DNDN and its investors. Aspirin and turmeric are already proven effective in the prevention and treatment of prostate cancer. Aspirin and other NSAIDS help prevent cancers, heart disease and possibly Alzheimers by reducing inflammation, which researchers believe causes or promotes the spread of these and other diseases. Click here. (‘Among daily NSAID users, there was a 12% reduced risk of prostate cancer in those aged 50 to 59, a 60% lower risk in those 60 to 69, and an 83% drop in risk for those 70 and over.’) And here. (‘Rutgers researchers have found that the curry spice turmeric holds real potential for the treatment and prevention of prostate cancer, particularly when combined with certain vegetables.’) Said vegetables include cauliflower, which when fried is delicious with turmeric. Be careful though; turmeric stains things yellow!’