When scientists at the California-based pharmaceutical company, Immune Response, proposed changes to James Kahn's article on the failure of the company's experimental AIDS treatment, Kahn refused. The University of California at San Francisco (UCSF) researcher perceived the company's post-hoc analysis to be “spin.” The company described it as good science. When Kahn submitted the study to the Journal of the American Medical Association, they immediately agreed to publish it, accompanied by separate articles on the perils of industry-sponsored research. The story quickly became front-page news.

Even before the journal was printed, Immune Response sought arbitration and was asking for $7 to $10 million in damages. UCSF responded with a counter claim, asking for release of all data collected as part of the $30 million study. And the already hot issue of industry influence on campus moved even higher on the research policy agenda. The 1 November article concluded that patients taking IRC's therapeutic AIDS vaccine, Remmune, didn't fare any better than patients on standard therapy.

“We went forward because we felt that our patients and colleagues had the right to know,” says Kahn. The company did not try to block the study's publication, as reported, but simply wanted to make sure it included all the data, insists IRC vice president Ronald Moss. Missing from the article was a subgroup analysis conducted by the company concluding that the drug has some effect on immune response, he said. “The data is there,” says Moss. “Why not let people see it and come to their own conclusion?” Kahn says he had no problem with the data, but felt the company's analysis was flawed.

Disagreements between industry sponsors and academic researchers are not uncommon, but they rarely reach this stage, says John Bartlett, the director of the clinical research at the Duke University Center for AIDS Research in Durham, North Carolina. His program has at least six industry sponsored AIDS studies running at any given time. The best way to avoid court proceedings is to spell out all the details from the start, including who gets to approve publications, says Bartlett.