When the goose lays a rotten egg

by William Wells

Silence of the Boots

In November 1994 Betty Dong of the University of California, San Francisco (UCSF) received word that her manuscript had been accepted for publication in the prestigious Journal of the American Medical Association (JAMA). Dong was delighted - an article in JAMA comes along rarely for even the best medical researchers - and publication was set for January 25 1995.

But then on January 13 1995 JAMA received a letter from Dong withdrawing her manuscript. She cited as explanation an "impending legal action by Boots Pharmaceuticals, Inc. against the University of California, San Francisco and the investigators."

Dong's withdrawn article contended that one of Boots' big money-spinners - a thyroid hormone preparation called Synthroid - was biologically equivalent to three competing and cheaper preparations. Over the previous four years Dong had been battling Boots, who funded the study, over the results. She was hamstrung by a contract that she had signed which stated, amongst other things, that any data were "not to be published or otherwise released without written consent" of the company. Such gag orders are forbidden by any sane university, but Dong signed the contract before 1993, the year that administrative review of all such documents became standard practice at UCSF. Now UCSF was unwilling to back up Dong, even as she claimed that Boots was threatening legal action against her personally for any reduction in sales of Synthroid.

Publication was staved off for long enough for the 1995 sale of Boots to Knoll Pharmaceuticals to go ahead successfully. Meanwhile Boots/Knoll came under increasing pressure following a 1996 article in the Wall Street Journal and a series of curt inquiries from the US Food and Drug Administration. The company finally relented and the Dong article, unchanged from the 1994 proofs, was published in April 1997.

But UCSF wasn't in the clear just yet.

Science and money

The Dong case was an extreme one: a bad contract that should never have been signed, and a company trying to hide results with no apparent justification other than worries about profit. Other conflicts like this can and do arise, but the more common concern is not direct conflict, but conflict of interest.

This idea - that money might influence research outcomes - is gaining more attention as profit-making companies pour ever greater amounts of cash into basic and applied research. A total of US$1.9 billion in research funding flowed from industry to academia in 1997, and the 1998 efforts of university technology transfer offices (which shepherd university inventions into the private sector) are estimated to have generated $34 billion of activity, supporting 280,000 American jobs. "This is a significant business for us, as it is for many research universities," says Chris Scott, UCSF's assistant vice-chancellor for research affairs.

It all started 20 years ago, when the Bayh/Dole act declared that universities could patent the results of federally-funded research. Since then the research-results auction has been an ongoing and ever-expanding bazaar, and Scott says that isn't going to stop any time soon.

"The train is out of the station on this," he says. "We have to face the reality that company science and academic science are closely linked. The question is not whether we should be doing the collaboration with industry but how."

Scott was interviewed for an article in the Atlantic Monthly that questioned this financial interdependence, and not so subtly called for it to end. But Scott says that "if we were to take [the magazine's] advice we would stop biotech in its tracks. If we had taken their advice twenty years ago Silicon Valley would still be an orchard."

Not that Scott is hoping for unlimited industry involvement. "For places that are the nucleating areas for new discovery you have to be careful not to migrate away from discovery science," he says. "We don't want to become a scientific contract house for companies."

Meanwhile the university must look out for the dangers that go with greater company involvement. "Are we making some sort of Faustian bargain?" asks Scott. "Is the bargain somehow affecting the objectivity of university scientists?"

For the answer to that question, we have to turn to the researchers who research the researchers.

How does bias arise?

Jordan Cohen, the president of the Association of American Medical Colleges, stated recently that "a conflict of interest exists when the prospect of some personal advantage is strong enough to pose a realistic possibility that [it] might compromise the researcher's primary obligation to adhere to sound scientific procedures in an unbiased search for the truth." In other words, money can make people bend the truth.

Identification of a conflict of interest means only that the possibility of wrongdoing exists, not that wrongdoing has been committed. For most scientists the strongest conflict of interest is personal. Basic science brings few financial advantages, so success (and the associated notoriety) is a far stronger motivating force. It has prompted more than a few researchers to commit research fraud.

But monetary conflicts of interest are more easily quantified, and are therefore the ones subject to regulation and scrutiny. These conflicts range from company funding of an academic's research, to companies paying the academic for speaking engagements, consulting sessions, and involvement on an advisory board. Increasingly, academics also own equity in companies that are exploiting their basic research.

Most people, and perhaps particularly scientists given their commitment to the truth, think they will not be influenced by such temptations. "Everybody has in their own mind that they understand the situation better than everyone else," says Scott. "But the reality is there is subtle bias in everything we do."

That bias showed up in studies of the cost-effectiveness of six anti-cancer drugs. Charles Bennett of Northwestern University in Chicago, Illinois, analyzed the results of these studies, and reported in the October 1999 JAMA that company-sponsored trials were one-eighth as likely to reach a negative conclusion about the cost-effectiveness of these drugs when compared to trials sponsored by non-profits.

A similar situation arose when the safety of a class of drugs called calcium-channel antagonists was called into question. Multiple articles were written both supporting and opposing the use of these drugs. Of the supportive authors, 96% had ties to a company making calcium channel antagonists, whereas only 37% of the critical authors had such ties.

Clearly there is widespread conflict of interest, and at least some of that conflict leads to bias. So what can we do about it?

Bringing in the rule book

In the November 1 2000 JAMA, Lisa Bero of UCSF published a study of UCSF's response to conflict of interest issues over the last 20 years. She found that by 1999 a full 7.6% of UCSF faculty reported personal financial ties with industry sponsors of their research, and that a supervising committee recommended taking remedial action in 25% of these cases. The actions included sale of company stock, refusing additional payment for talks, resigning from a company management position, or naming a new principal investigator to a university research project.

UCSF's committee tended to be lenient when an academic's research was significantly different from that of the company supplying monetary rewards. But the inconsistency in applying these standards troubled Bero. "You need to look at individual cases, but it would be good if there were more definite rules so consistency was assured," she says. "Right now the whole thing is very ad hoc."

The situation only gets worse at the national level. In the same issue of JAMA Mildred Cho of Stanford University (Stanford, California) reported on the conflict of interest policies of the 100 US institutions with the most funding from the National Institutes of Health in 1998. The regulations varied widely, with only 19% specifying limits on the financial interests that researchers could have in the corporate sponsors of their research.

UCSF's Scott is amongst those who are calling for national guidelines to improve this situation. Such guidelines are probably unlikely to be as strong as those at UCSF, which is one of the few institutions to ban researchers from conducting clinical trials for companies in which they hold any financial stake.

Northwestern's Bennett, meanwhile, is looking at the reasons behind the bias. He has concluded that company-sponsored trials are often, if anything, of a higher quality than other trials. But academics without company ties share one characteristic. "We are more willing to generalize, and the generalizations may not be so optimistic," says Bennett.

Further distortions can arise when drug trials are analyzed in retrospect, because different researchers will analyze the same data with different questions in mind. "The companies ask, how much value did we find," says Bennett. "The academics ask, how bad was it."

Bennett's solution is simple. "The way to fix it is to have prospective, comprehensive registrations of all trials," he says. That way all those trials with negative results cannot be hidden, and the aims of the trial must be declared before the data come in.

UCSF is all in favor of full disclosure before trials begin. Full disclosure might have saved the university from the wrangling that accompanied another study appearing in the November 1 2000 JAMA. The actors had changed since the earlier conflict over Synthroid: now UCSF researcher James Kahn played the part of Betty Dong, and the Immune Response Corp. (IRC; Carlsbad, California) the part of Boots/Knoll. But the subject of the conflict (this one direct, not a conflict of interest) remained the same - the right to publish.

Dong redux, but with less pain

Kahn was a leading investigator on the largest randomized trial among HIV-infected persons conducted during the 1990s. The subject of the study was IRC's drug Remune, which was designed to stimulate an immune reaction to HIV (the AIDS virus) in people already infected. IRC hoped that Remune treatment would decrease the number of HIV-infected people progressing to full-blown AIDS.

Fortunately for the study subjects, but unfortunately for IRC, protease drugs came along during the trial with dramatic effects. As IRC's CEO Dennis Carlo explains, "without enough patients getting AIDS in the control group, it is difficult to show a treatment effect with Remune."

For this and perhaps other reasons, the trial failed to show promise and was halted five months early, in May 1999. Kahn claims that he then requested the final data set (the study was conducted at 77 different centers), at which point the company asked for veto power over the final draft of any manuscript. Kahn refused, and went ahead to publish the results that he did have in JAMA. IRC disagrees with the omission of certain data that it believes are positive, but Kahn says the company "picked and chose the analysis method" to get something that came out positive. "This was something I wouldn't attach my name to because I think it is wrong," he says. IRC has initiated binding arbitration proceedings and is requesting damages.

Unlike the Dong case, here "the protocol gave us an ability to publish our results," says Kahn. But the research contract did not stipulate that all results had to be turned over to the academics. "Based on this I think we'll learn to write better contracts," says Kahn.

Why do companies and academics submit themselves to all this turmoil? Scott believes the two need each other. Increasingly companies have some of the best research tools and leads, while clinical academic researchers have access to patients. Plus "sponsors need to have good honest investigators to do work," says Kahn. "But if they want the ethos of independence that university research brings, they have to abide by the rules of that independence."

Whether by better contracts or clearer regulations, academics and industry alike hope to avoid both direct conflicts and bias from conflicts of interest. Every publicized dispute damages the public's trust in the accuracy of medical research and, as JAMA's editor Catherine DeAngelis stated in the November 1 issue, "without trust, medical research is doomed."

Originally published in the web magazine Access Excellence.