Thursday, August 26, 2010

Was the CAFE study manipulated by AstraZeneca? Maybe Not.

In Monday's post, I summarized Carl Elliott's recent article in Mother Jones, in which he argues that Dan Markingson's suicide may have been related to the fact that he was enrolled in an AstraZeneca study comparing AZ's Seroquel with two other medications. Like most clinical trials, this study put limits on what medications doctors could give their study patients. These limits are required to discover how effective a drug is for a particular condition. Otherwise, patients in research studies might be taking all kinds of miscellaneous treatments, and you wouldn't know if their improvements were due to the study drug or something else.

The crucial question here is whether this study, called the CAFE trial, was, in fact, designed to answer a valid scientific question, such as: Which of three antipsychotics is most effective? Or was it a rigged study, designed to ensure that Seroquel would look good compared to Zyprexa and Risperdal, purely in order to drive sales? It would obviously be unethical to recruit seriously ill patients into a bogus trial if it was concocted as a marketing strategy.

As Dr. Barney Carroll mentioned in a response to Monday's post, the interesting thing about the CAFE trial is that several of the major investigators were also involved in the CATIE trial. The CATIE trial was of a similar design and involved schizophrenic patients who were randomly assigned to double blind treatment with different antipsychotics. But the CATIE trial was funded by the NIMH rather than by any particular drug company. Therefore, we assume that design decisions were made relatively free of drug company input.

The results of the NIMH-funded CATIE trial were different from the AZ-funded CAFE trial--quite different. The main outcome measure for both studies was the percentage of patients who discontinued the drug. In the CAFE trial, the 12-month discontinuation rates for Zyprexa, Seroquel, and Risperdal were 68.4%, 70.9%, and 71.4%, respectively. In the CATIE trial, the18-month discontinuation rates (the 12-month rates were not reported) were 64%, 74%, and 82% for the same three drugs (two other drugs were also tested in CATIE). In the CATIE trial Zyprexa did much better than Seroquel, but it did not do better in CAFE--which seems odd considering both studies used the same methodology.

What's the explanation here? It could be simply be a statistical variation in outcomes. Perhaps CATIE's results were wrong, and perhaps CAFE's are right. After all, this is exactly why it's important to replicate a research finding. If a finding is replicated, we are much more confident in the truth of the original results.

In this case, however, the divergent results appear more likely due to the fact that in CAFE, Zyprexa and Risperdal were dosed much lower than in CATIE. Look at the table below. Zyprexa was dosed 42% lower in CAFE, Risperdal was dosed 38% lower, while Seroquel was nearly the same in the two--only 7% lower in CAFE. Thus, on the face of it, it seems clear that the CAFE study found Seroquel to be as good as Zyprexa and Risperdal because the design ensured that Seroquel would be dosed much higher than Zyprexa and Risperdal.


AZ-funded CAFE dose

NIMH-funded CATIE dose


11.7 mg

20.1 mg


2.4 mg

3.9 mg


506 mg

543 mg

But why did this happen? Was this a design manipulation ordered by AstraZeneca marketing staff in order to make sure the results were to their liking? Or was it simply a research decision made with pure motives by the scientists who ran the trial? The answer to this question makes a difference. In fact, you could argue that it makes all the difference.

This post would have ended here if I hadn't emailed these questions to the researchers who designed the CAFE study, including
Dr. Joseph McEvoy, Dr. Jeffrey Lieberman, and Dr. Diana Perkins. All three of them were also involved in the CATIE trial. They reminded me of a crucial difference in the two trials, aside from the fact that one was funded by government and the other by Astra Zeneca.

The CAFE study, as should have been obvious to me by the full name ("Comparison of Atypicals in First Episode of Psychosis"), was designed with newly diagnosed patients in mind. Typically, patients with a first episode of psychosis are treated more conservatively, with lower doses of antipsychotics, as opposed to the chronic schizophrenics enrolled in the CATIE trial. So this explains why Risperdal and Zyprexa were dosed lower in CAFE than in the CATIE.

So why, you might ask, wasn't Seroquel also dosed lower? According to Dr. McEvoy, who was the co-principal investigator on both CAFE and CATIE, at that time there were no studies guiding them on Seroquel dosing in first episode patients. "Back then," he wrote me, "many argued that higher doses were needed for optimal efficacy and there was not a side effect ceiling such as risperidone has." Eventually, a study was published showing that the average Seroquel dose needed for first episode psychosis is only 268 mg/day

Dr. Lieberman, who is chair of psychiatry at Columbia University, emailed me that "Although your inferences re dosing are understandable they are entirely incorrect. The CAFE study was wholly investigator initiated and we determined the doses for each drug. There was a clear rationale for the higher quetiapine dose relative to the other drugs which I am happy to explain to you. This study design in no way was influenced adversely by industry influence."

So there you have it. If you can believe these researchers, this may have been an example of an AstraZeneca funded study without undue commercial influence. I think they are believable, and I know that Dr. Lieberman in particular has been quoted regularly in the media recently about various new drugs, and has been appropriately skeptical about their benefits over older agents.

Of course, there are other ways in which CAFE might have been rigged, which are detailed in Carl Elliott's Mother Jones article. I've asked the investigators about these allegations and will let you know if I get a response.

Monday, August 23, 2010

"Making a Killing": New Carl Elliott Article in Mother Jones

There's a fascinating article by Carl Elliott in the current issue of Mother Jones. It's called "Making a Killing," and it shows how clinical trials have become marketing exercises for the pharmaceutical industry, sometimes at the expense of patients' lives. (You can read the article here, but you will need to complete a free registrations process).

The focus is on Astra Zeneca, maker of the antipsychotic Seroquel. Elliott tells the tragic story of Dan Markingson, a young man with schizophrenia who killed himself in a particularly gruesome way in May of 2004. At the time, Markingson was enrolled in an Astra Zeneca-funded study called the CAFE study. The acronym CAFE stands for "Comparison of Atypicals in First Episode of Psychosis." As the name implies, this study randomly assigned patients to three possible antipsychotics: Seroquel, Risperdal, and Zyprexa. The bottom line result, according the abstract in the American Journal of Psychiatry, was that the three drugs were equally effective.

However, Elliott makes the case that the study was rigged in various ways by the Astra Zeneca marketers to ensure that Seroquel would, at the least, not look any worse than its competitors. While this devious tactic may not be much of a news flash, Elliott is
claiming that something more insidious happened--namely, that Dan Markingson was pressured to put his life on the line for a drug company's profits.

Elliott acknowledges that there are risks in any clinical study. When the study is trying to answer a significant scientific questions, these risks may be reasonable. But when the study is designed to boost sales of a drug, these risks are very hard to defend.

As Elliott puts it: "It is one thing to ask people to take risks for science, or the common good, or to help other people. It is another thing entirely to ask them to risk their lives for the marketing goals of AstraZeneca."

I had read the CAFE study before, but as I was preparing this post, I noticed aspects of the design that had not struck me in the past. The study may well have been manipulated in order to make Seroquel look good. In my next post I will delve into the specifics, so stay tuned.

Wednesday, August 4, 2010

Heart Surgeon's Brothel in Dallas

Over at CardioBrief, Larry Husten reviews the industry prospectus offered by an international meeting for cardiac surgeons, occurring in Dallas January 9-11. Apparently, the organizers of the meeting believe they are still in the roaring 90's when the industry gravy train was making stops at every doctor's office, throwing cash into the air and watching as doctors gleefully scooped up as much lucre as they could hold.

At "Dallas Leipzig International Valve 2010," everything is for sale--including the surgeons. Thus, in addition to paying for exhibition space, symposia, wine-tastings, and logos on flash-drives, notebooks, and attendee's foreheads (not really), valve companies can buy an intimate dinner with "four faculty members for a private one-on-one meeting." But act quickly--the program limits each faculty member to only one trick per conference: "Secure your faculty choice early as faculty will be removed from selection list once chosen."