"Thus far, we have buried trials 15, 31, 56. The larger issue is how do we face the outside world when they begin to criticize us for suppressing data."
Something similar has happened with the atypicals. A 2006 study in The American Journal of Psychiatry, which looked at 32 head-to-head trials of atypicals, found that 90 percent of them came out positively for whichever company had designed and financed the trial. This startling result was not a matter of selective publication. The companies had simply designed the studies in a way that virtually ensured their own drugs would come out ahead—for instance, by dosing the competing drugs too low to be effective, or so high that they would produce damaging side effects. Much of this manipulation came from biased statistical analyses and rigged trial designs of such complexity that outside reviewers were unable to spot them. As Dr. Richard Smith, the former editor of the British Medical Journal, has pointed out, "The companies seem to get the results they want not by fiddling the results, which would be far too crude and possibly detectable by peer review, but rather by asking the 'right' questions."
Initially, the controversy over atypical antipsychotics was focused largely on Eli Lilly (PDF), the manufacturer of Zyprexa. In early 2009, it settled litigation for a record-breaking $1.4 billion for illegal marketing and allegedly hiding the risks of the drug. More recently, however, the scandal has spread to Seroquel. In April 2010, AstraZeneca agreed to pay $520 million to settle two federal investigations and two whistleblower lawsuits alleging that it had marketed Seroquel illegally and concealed its health risks. The company faces more than 25,000 civil suits.
Documents unsealed in related civil suits suggest an alarming pattern of deception. Sales reps were instructed to tell doctors that Seroquel doesn't cause diabetes, even though the company knew about the link to diabetes as early as 1997. Internal correspondence reveals company officials discussing how to hide or spin potentially damaging studies. "Thus far, we have buried trials 15, 31, 56," wrote a publications manager in 1999. "The larger issue is how do we face the outside world when they begin to criticize us for suppressing data."
One of those potentially damaging studies led back to the University of Minnesota. In the late 1990s, a clinical trial known as Study 15 unexpectedly failed to show that Seroquel was any better than Haldol, a generic antipsychotic that's been on the market since the 1960s. In fact, on the main measures, Seroquel performed worse than Haldol. The study also showed that Seroquel increased the risk of weight gain and diabetes. Internal correspondence repeatedly refers to Study 15 as a "failed study," and company officials discuss possible ways to spin or bury it. "I am not 100% comfortable with this data being made publicly available at the present time," wrote Richard Lawrence, a senior AstraZeneca official, in 1997. "However I understand that we have little choice...Lisa [Arvanitis, a company physician] has done a great 'smoke-and-mirrors' job." Lawrence referred approvingly to a strategy that he said would "put a positive spin (in terms of safety) on this cursed study." Later, apparently hoping to find a way to present Seroquel in a better light, the "commercial support team" performed an analysis of a number of other studies, but even that did not show Seroquel to be better than Haldol. Yet when a summary of the AstraZeneca data was presented at the American Psychiatric Association annual conference in 2000, the author claimed Seroquel was "significantly superior" to Haldol. That author was Dr. Charles Schulz, the University of Minnesota psychiatry department chair—and a well-compensated consultant for AstraZeneca. In a press release claiming Seroquel's superiority over Haldol, Schulz praised it enthusiastically as a "first-choice antipsychotic."
Although the documents unsealed in the Seroquel litigation do not specifically mention the CAFE study in which Dan was enrolled, they do suggest that AstraZeneca planned to establish Seroquel as the "atypical of choice in first-episode schizophrenia," according to a 2000 "Seroquel Strategy Summary." A later document titled "Seroquel PR Plan 2001" discusses the agenda for an advisory panel meeting in Hawaii. Among the potential topics were the marketing of Seroquel to first-episode patients, adolescents, and the elderly. The document refers to these populations as "vulnerable patient groups."
"R&D is no longer responsible for Seroquel research—it is now the responsibility of Sales and Marketing."
Even more alarming are internal documents suggesting that AstraZeneca was designing clinical trials as a covert method of marketing Seroquel. In 1997, when Dr. Andrew Goudie, a psychopharmacologist at the University of Liverpool, asked AstraZeneca to fund a research study he was planning, a company official replied that "R&D is no longer responsible for Seroquel research—it is now the responsibility of Sales and Marketing." The official also noted that funding decisions would depend on whether the study was likely to show a "competitive advantage for Seroquel."
Another set of documents from 2003 describes a glucose metabolism study apparently designed to fend off the charge that Seroquel causes patients to gain weight and become diabetic. One slide describes two purposes for the study: a "regulatory" purpose and a "commercial" purpose. The regulatory purpose was to "produce data that will help us defend the Seroquel label." The commercial purpose was to "produce data that will enable us to generate commercially attractive and competitive messages in relation to diabetes and weight." The document suggests several possible names for the study, including "Flexible Dose Approach Trial for Atypical Responses to Metabolism," which could be usefully shortened to the acronym FATFARM. (When I contacted AstraZeneca, a spokesperson would say only that Seroquel has been found "safe and effective" by the FDA and that it stands behind the CAFE study and the rest of its clinical research.)
Many clinical studies place human subjects at risk—at a minimum, the risk of mild discomfort, and at worst, the risk of serious pain and death. Bioethicists and regulators spend a lot of time and energy debating the degree of risk that ought to be permitted in a study, how those risks should be presented to subjects, and the way those risks should be balanced against the potential benefits a subject might receive. What is simply assumed, without much consideration at all, is that the research is being conducted to produce scientific knowledge. This assumption is codified in a number of foundational ethics documents, such as the Nuremberg Code, which was instituted following Nazi experiments on concentration camp victims. The Nuremberg Code stipulates that an "experiment should be such as to yield fruitful results for the good of society," and "the degree of risk to be taken should never exceed that determined by the humanitarian importance of the problem to be solved by the experiment."
But what if a research study is not really aimed at producing genuine scientific knowledge at all? The documents emerging in litigation suggest that pharmaceutical companies are designing, analyzing, and publishing trials primarily as a way of positioning their drugs in the marketplace. This raises a question unconsidered in any current code of research ethics. How much risk to human subjects is justified in a study whose principal aim is to "generate commercially attractive messages"?
In January 2005, the FDA began investigating the circumstances of Dan's suicide. In a report issued that July, before the larger pattern of Seroquel research had begun to emerge, Sharon L. Matson, the FDA investigator, exonerated the university. She wrote, "I did not find any evidence of misconduct, significant violation of the protocol, or regulations governing clinical investigators or IRBs"— the university institutional review board charged with reviewing studies to ensure that they measure up to recognized ethical standards. Matson specifically dismissed the suggestion that Dan was mentally incompetent to consent to the study, writing that "there was nothing different about this subject than others enrolled to indicate that he couldn't provide voluntary, informed consent." (The FDA refused my request to speak with Matson and would not answer questions about the case, citing privacy concerns.) Mary Weiss eventually sued the University of Minnesota, AstraZeneca, Olson, and Schulz, but her case did not even get to trial. District Court Judge John L. Holahan dismissed the suit in 2008 with a partial summary judgment. He ruled that in approving the CAFE study, the university IRB was performing the type of "discretionary function" that is protected from liability under the state's Tort Claims Act. The malpractice suit against Schulz was also dismissed, and the suit against Olson was eventually settled—for $75,000, which Mary says wasn't enough to cover the fees of the expert witnesses her attorneys hired. (Both Schulz and Olson declined to speak about the specifics of the clinical trial or the resulting suit. University spokesman Nick Hanson would say only, "To date, there has been no finding of wrongdoing from any of the investigations or reviews done by the university on this issue.")
The judge also dismissed the case against AstraZeneca. He blasted Mary's lawyers, saying that they had failed to establish that AstraZeneca had a duty to put the interests of research subjects over the interests of the company and the researchers. But he also lamented the lack of case law about clinical trials, saying on this particular point, "Try as it may, this Court's independent research has unearthed not a single case or statute to evidence or support such an alleged duty."
The judge further ruled that Mary's lawyers hadn't shown a causal link between Seroquel and Dan's suicide: An initial drug screening during autopsy had not found any Seroquel in his bloodstream, which suggested that Dan may not have been taking his medication. After the judgment, however, Mary discovered that Seroquel would not be detected in an ordinary drug screening; a special test is required. In the spring of 2008, she called the coroner's office in hopes of getting a special screening for Seroquel. To her surprise, she found that her lawyers and the defendants had already obtained one. The report was dated several days after the summary judgment was issued. It showed 73 nanograms per milliliter of Seroquel in his blood, suggesting that Dan was almost certainly taking the drug, although he may have missed the last scheduled dose before he died.
Although Mary's lawsuit was unsuccessful, it revealed some disturbing financial arrangements at the university. As a patient on public assistance, Dan's treatment would have normally generated little income for the university. Under its arrangement with AstraZeneca, however, the psychiatry department earned $15,648 for each subject who completed the CAFE study. In total, the study generated $327,000 for the department. In fact, during the months before Dan was enrolled, the department was apparently feeling pressure from Quintiles, the CRO that managed the study, to step up recruitment. According to emails written by Jean Kenney, the university's study coordinator, the site had been placed on probation for its recruitment problems, and they were still "struggling to get patients." In November 2002, Olson had managed to recruit only one subject in six months. That began to change in April 2003, when the psychiatry department established a specialized inpatient unit at Fairview hospital called Station 12, in which every patient could be evaluated for research. By December, Olson had recruited 12 more subjects, including Dan, and Olson had been featured in a CAFE study webcast for "turning an underperforming site into a well-performing site."(Quintiles refused to give comment on the case.)
Olson had another financial reason to maintain good relations with AstraZeneca. According to a disclosure statement for a 2006 conference, he was a member of the AstraZeneca "speaker's bureau," giving paid talks for the company. He had similar arrangements with Eli Lilly and Janssen, the makers of the other atypicals being tested in the CAFE study, as well as Bristol-Myers Squibb and Pfizer. In addition, Olson was working as a paid consultant for Lilly, Janssen, Bristol-Myers Squibb, and Pfizer. Although Olson is not required to disclose how much industry money he received, a public database maintained by the Minnesota pharmacy board indicates that Olson received a total of $240,045 from the pharmaceutical industry between 2002 and 2008, with $149,344 coming from AstraZeneca. Dr. Charles Schulz, his co-investigator and department chair, received an even greater sum: more than $571,000 from the industry, with $112,020 coming from AstraZeneca. The database does not reliably distinguish between payment by drug companies for consulting and speaking, which usually goes directly into a physician's pocket, and research grants, which go to the university and are used to help underwrite the salaries of the grant recipients. (Many academic physicians are required by their universities to generate a substantial portion of their salaries by obtaining research grants.)