In March 1999, Walsh appeared as the lead author of an article in the New England Journal of Medicine that reported detailed results from the study that had compared AmBisome with amphotericin. The article said the drug dosages were "deliberated upon and adopted by consensus of the investigators" who conducted the study.
Physician-researchers from Germany questioned the design of the study in a letter to the journal seven months later.
"We think that the design of this randomized trial was not adequate because the dose of conventional amphotericin B (0.6 mg per kilogram of body weight per day) that was used does not reflect widely used standards of care," wrote Drs. Thomas Fischer, Gudula Heussel and Christoph Huber of Johannes Gutenberg University. "Most institutions in Europe and the United States would agree that treatment of this patient population requires a dose of at least 1 mg."
The physicians said it seemed "very likely" that if Walsh and his collaborators had used a "normal," higher dose of conventional amphotericin, fewer patients who took that drug would have had fungal infections emerge or progress. (Fischer declined to be interviewed for this article; he said by e-mail that AmBisome had proved to be a useful drug.)
Skepticism about the dose of conventional amphotericin used by Walsh also was reflected in a 2001 medical reference book issued by the British Society for Haematology and other groups. The authors said that conventional amphotericin had been given to similar patients in Europe at doses up to twice as high as in the study that Walsh helped lead.
The lower dose, the authors wrote, "may bias the results in the favour of AmBisome" and "could entirely explain the differences observed."
Fujisawa had agreed to allow doctors conducting the study to double the dose of either drug, depending on patients' conditions. But doctors ultimately increased the dose for 17% of the patients who took amphotericin -- while doses were increased for 34% of the patients who took AmBisome. The New England Journal of Medicine report cowritten by Walsh described the study as "blinded," so that neither the doctors nor their patients were supposed to know which of the two drugs was being administered.
Walsh and two colleagues, in a reply published by the journal, said the dose of amphotericin reflected "the standards of care" at the participating research centers. Walsh also suggested that the toxicity of amphotericin had prevented the administration of "appropriate doses" to some patients.
The three officials who wrote to The Times on Walsh's behalf, including Robert H. Wiltrout, a research director at the National Cancer Institute, defended the dose of conventional amphotericin.
"There is no rational motivation for an investigator or sponsoring company to design a trial with a control arm that is not standard of care," wrote Wiltrout, along with Drs. Lee J. Helman and Frank Balis of the National Cancer Institute.
As Walsh defended his study in the New England Journal of Medicine, he was helping write new medical-practice guidelines suggesting far higher doses for some patients.
In a paper submitted for publication in October 1999, Walsh and other authors said that, following prompt and aggressive evaluations of the patients, doctors should consider "maximum tolerated doses" of conventional amphotericin if aspergillus infection, specifically, was suspected. They defined those doses as 1 to 1.5 milligrams per kilogram of body weight, daily.
Standing With Merck
By 1999, Walsh was collaborating with Merck & Co., on its new antifungal drug, Cancidas. One Wall Street firm predicted that Cancidas could generate annual sales of $330 million. But first Merck needed FDA approval.
AmBisome, the same drug that Walsh had just helped guide to FDA approval, was picked as the comparator.
Merck paid for the study. Walsh designed it in collaboration with Merck and one other researcher, who received fees from Merck. The initial dose selected for AmBisome -- 3 milligrams per kilogram of body weight, daily -- was the same as in the earlier study.
In a statement delivered to The Times last week, Walsh said "there was and is no evidence" that higher dosages of AmBisome would offer better effectiveness.
Previously, Walsh supported higher doses of AmBisome for patients with aspergillus.
At the 1997 conference in Toronto, Walsh said that a lower dose might suffice for a yeast-type infection. But for aspergillus or for other mold infections that resist treatment, he said, "I would submit that we probably should be using more.... There are good experimental data to show that more is better."
When choosing a dose, Walsh added, "I think it depends on what disease one is treating.''
At a September 1999 conference in San Francisco, Walsh, along with Fujisawa's medical director and several other scientists, described having used AmBisome doses from 7.5 to 15 milligrams per kilogram of body weight per day in patients with possible, probable and proven infections.