Drug Trials With a Dose of Doubt

BETHESDA, Md. —  On Jan. 10, 2001, pharmaceutical giant Merck & Co. gathered its forces in a hotel conference room here with a clear-cut mission: Win a favorable vote for a new antifungal drug from a federal advisory committee -- a victory that would position the product for swift government approval and for hundreds of millions of dollars in sales.

But after hours of speeches and slides, the committee members, appointed by the U.S. Food and Drug Administration, had yet to vote. The members were focused on the quality of Merck’s case for the new drug, which rested on the treatment of only 69 patients.

Merck summoned to the microphone one of its announced consultants, a man whose government job was nearby, at the National Institutes of Health. Dr. Thomas J. Walsh assured the committee that Merck’s data describing the patients was “extremely robust and very, very rigorous.” He said his government staff had assisted in vetting the company’s data. About 30% of the patients were helped by the drug, he said.

The advisory committee voted unanimously to endorse the drug, called Cancidas. Sixteen days later, the FDA approved it. Doctors would later prescribe it for patients whose immune systems had been ravaged by chemotherapy and who were presumed to have a potentially deadly, invasive fungal infection. In its first five years on the U.S. market, Cancidas would generate $859 million in sales for Merck.

U.S. law generally prohibits a federal employee from representing an outside party before a government agency.

In building a career as an influential government scientist, Walsh has served as both a paid and unpaid advisor to pharmaceutical companies and has helped lead clinical trials that tested the effectiveness of their products. With his help, the companies have brought new antifungal drugs to market, but controversy has flared over whether results from two of the studies were misleading and whether some of the participating patients received adequate treatment.

In written comments for this article, Walsh said his advice to industry did not conflict with his position at the NIH’s National Cancer Institute, or affect his scientific judgment.

“I am not and have never been a representative of, or advocate for, any pharmaceutical company,” Walsh said.

Two drug makers involved with his federal research, Merck and Pfizer Inc., said they have paid fees to Walsh. Merck and another company, Fujisawa USA Inc., have made financial or other donations to support Walsh’s federal research with the approval of his NIH superiors, interviews and government records show.

From 1997 to 2003, Walsh appeared at meetings with FDA committees or staff alongside representatives of Pfizer, Fujisawa and Merck, according to videotapes, transcripts and other government records. He also helped design, oversee and interpret the results of major clinical studies of four antifungal drugs made by those companies. The studies helped win FDA approvals for three of the drugs.

In separate letters to a leading medical journal, other researchers criticized two of those studies. They questioned whether the studies artificially boosted the new products by comparing them to drugs that were given at doses that were too low.

More patients died who took the “comparator” drugs than those who got the new products.

Walsh, in journal articles and in remarks to medical leaders, noted the disparities in deaths while describing the advantages of the newer drugs. In published responses to the scientific critics, he said the doses of the comparator drugs reflected the general standard of care at the participating hospitals.

What led to the higher death rates of the control-group patients in the two major studies may never be known: A limited number of autopsies were performed, and factors other than fungal infections, such as the patients’ cancer, could have caused the deaths.

No published study has established that a higher dose of an antifungal drug is more effective in treating suspected infection, and some studies have suggested that lower dosing may provide similar benefits. But the possibility that patients did not receive adequate doses, combined with Walsh’s advisory role with the drug companies, adds a new dimension to the furor over NIH scientists’ ties to industry.

Earlier revelations of the agency scientists’ outside arrangements called into question their impartiality and the independence of the NIH, the nation’s largest agency for experimental medical research, prompting congressional hearings, policy reforms and ethics investigations.

However, even as the NIH moved recently to ban some of the activities with industry, the agency’s director said the arrangements had apparently not jeopardized patients in clinical studies.

“Thus far, we have not identified any situations where patients were harmed as the result of financial arrangements NIH employees had with outside parties,” NIH Director Elias A. Zerhouni told a Senate subcommittee in 2004. “I will, however, reserve final judgment until all internal and external reviews are completed.”

In response to questions from The Times regarding Walsh, Zerhouni responded generally in a prepared statement. “We revamped our rules last year, and continue to carry out a vigorous program of education, oversight and enforcement,” he said, adding, “Violations of the ethics rules are unacceptable, and I remain determined to pursue any information brought to my attention.”

Walsh, 54, heads a medical research and treatment unit within the pediatric branch of the National Cancer Institute, where he arrived in 1986. He said that collaborating with companies has been fundamental to his government work.

“My efforts are in service of the public interest in sound, reliable science concerning potentially effective agents for the treatment of life-threatening infections in children and adults with cancer,” Walsh said in a statement to The Times. “This mission frequently includes collaboration with companies that research and develop new compounds in this area -- for example, utilizing my [staff’s] expertise to ensure that clinical trials relating to these compounds are designed and implemented in a manner that elicits reliable and useful results.”

He said he has appeared before the FDA only “as a government scientist providing information and/or evaluation” regarding clinical trials. Referring to studies he helped lead, Walsh said, “There is no conflict of interest, and the trials were well and appropriately designed.”

The full extent of Walsh’s ties with industry is not open to view by outsiders. His yearly financial-activity reports at the NIH are exempt from release under the Freedom of Information Act, as are the reports for most senior researchers at the agency.

None of Walsh’s outside arrangements were listed in records that the NIH turned over to a congressional committee that had sought details of connections between agency scientists and the drug industry.

Although Walsh declined to answer a number of questions about his financial arrangements with the drug companies for this article, he said in a telephone conversation on May 18: “On the personal issues, I’ve made mistakes.”

Walsh also said he preferred to let colleagues address questions regarding the dosages selected for the two major studies he helped design. Two private lawyers representing him, H. Bradford Glassman and Jeffrey D. Robinson, noted in a letter to The Times that the dosages were chosen with the assent of other researchers, and not by Walsh, individually.

Walsh is well-known in his field, having written or cowritten more than 230 medical journal articles over the last decade. A medical graduate of Johns Hopkins University in Baltimore, he has won honors within the NIH as a mentor, receiving the agency’s Distinguished Clinical Teacher Award. In 1996, he received an Outstanding Service Medal from the U.S. Public Health Service for “sustained and outstanding advances in the treatment, prevention and diagnosis of invasive fungal infection in children with cancer and HIV infection.”

Three of Walsh’s superiors at the National Cancer Institute contacted The Times by e-mail and defended as scientifically sound the two major studies that he helped lead. The officials noted that the designs of both studies were reviewed and approved by the FDA and by boards at the medical sites where patients were treated. Dosages for one of the studies, they wrote, were selected based on a consensus of participating researchers.

Eight doctors, including seven who participated in one or the other major study with Walsh and who are not employed by the NIH, also contacted the newspaper and said they stood behind the validity of the research. The study designs, they said, “were both scientifically and medically sound, reflect the state of the art in the field, and have advanced supportive care, improving the management of patients worldwide and saving lives.”

Other researchers have said that doses of comparator drugs that are inadequate may endanger patients or make a new drug look more effective than it is.

“I can see why the companies are eager to get an easy comparison, a drug they can beat,” said Dr. Curt D. Furberg, who formerly headed clinical research at the NIH’s National Heart, Lung, and Blood Institute. “But for [scientific] investigators to go along with that, it’s just a bad practice.”

Picking the Patients

From the late 1950s to today, the drug of choice for many doctors treating potentially lethal fungal infections has been a powerful compound called amphotericin B.

Nurses and doctors have long dubbed the drug, derived from spores found in Venezuela’s Orinoco River region, “ampho-terrible.” Some patients tremble violently as the solution, infused intravenously, courses through their bodies. Fever and vomiting also can result. In some cases, the drug can cause fatal kidney damage.

Approved by the FDA in 1958, amphotericin gained greater acceptance in the United States in the 1980s after research conducted in Europe and at the National Cancer Institute suggested that the drug decreased patients’ vulnerability to an internal fungal infection.

For decades, amphotericin has been available worldwide in relatively cheap, generic formulations. By the early 1990s, several firms were aiming to modify it into their own brand-name products -- agents that they hoped would be better and that could fetch far higher prices. The new products would deliver the amphotericin in fatty mixtures, changing the characteristics of the drug to reduce the risk of kidney damage.

The modified amphotericin products would cost as much as $800 a day, compared with about $16 per day for the older drug.

In order to get their reformulated drugs approved by the FDA, the companies had to conduct human studies. The FDA held two public meetings, in 1994 and 1995, to hear experts’ opinions regarding design standards for the studies.

The FDA was under pressure to cooperate more closely with the pharmaceutical industry. Amid complaints that existing standards had stymied the development of new drugs, the agency had been directed by Congress and the White House to streamline its medical reviews.

For makers of the new antifungal drugs, less burdensome clinical-study standards could make it easier to get the products approved. For instance, some companies wanted to enroll cancer patients with suspected -- but unproven -- fungal infections. These would be patients who had abnormally low levels of infection-fighting white blood cells and fevers lasting at least four days, despite treatment with a standard antibiotic.

Walsh has stressed the need for treating suspected infections quickly, noting that persistent fever may be the only sign and that delaying treatment could lead to increased deaths. As envisioned by Walsh and others developing the new products, the drugs would be assessed on several factors, including whether the patients’ fevers abated.

Some cancer and infectious-disease specialists questioned that approach. Every enrolled patient would have a fever, but would its disappearance mean that the drug had defeated a fungal infection?

Noting that fever can have many causes, the specialists stressed the importance of studying patients with proven, as opposed to suspected, infections. Eliminating fever in the patients with proven infections, they said, would provide better evidence of effectiveness.

But at the 1995 FDA meeting, Walsh said enrolling and treating more of the patients with proven fungal infections would pose “financial and logistical limitations,” meaning the major studies would take longer and cost more. He estimated that it would take years to identify and enroll a sufficient number of patients with proven infections.

“I think it is appropriate to have a relatively low frequency of the proven infections,” Walsh said.

Walsh also told the FDA committee it was essential to launch separate studies that would more directly examine a drug’s effect on specific fungal infections.

The FDA accepted the approach of designing the major studies to enroll and treat patients with persistent fevers who had suspected but unproven infections. An FDA medical officer, Dr. Teresa Wu, said the approach “largely was motivated by” the ease of enrolling such patients.

Roughly half the patients would get new drugs, made by companies that helped pay for the research. The remaining “control” patients would get dosages of an older, comparator drug.

The choice of dose might determine patients’ survival: If the dose of the comparator drug in the first study, amphotericin, was insufficient, patients could be left more vulnerable to an infection invading the lungs or other organs.

One of the fungal infections, aspergillus, typically kills 50% or more of the patients who develop it. And it is notoriously difficult to diagnose: Because the patients are so sick, doctors often are reluctant to collect a sample of lung tissue, which might confirm an underlying infection. Aspergillus often cannot be confirmed before autopsy.

Yet if the dose were set too high, patients, including those who turned out not to have a fungal infection, would be put at greater risk of kidney damage.

Walsh did not commit to an exact dosage of amphotericin at the 1995 FDA meeting. He did, however, say that a drug used in the new studies would need to be powerful enough to treat aspergillus or other devastating mold-type fungal infections, not just yeast-type fungal infections, such as candida, which are lethal less often and are commonly treated with lower doses. Since the early 1990s, experts in the U.S. and Europe had reported increases in the frequency of aspergillus.

“If we are really trying to protect the high-risk patients,” he said, “we have to appreciate that there are more than just yeasts that we are trying to prevent or to impact upon.”

Disputed Results

The first major study that Walsh helped lead compared one of the new, modified drugs, AmBisome, with conventional amphotericin.

The study was paid for by the developer of the new drug, Fujisawa USA Inc., and by a grant from the NIH. Walsh had conferred about the study design with Fujisawa and with a national network of other physicians who would carry out the project.

The dosage for patients who would be given amphotericin was 0.6 milligram per kilogram of body weight, daily.

One expert invited by the FDA advisory committee, Dr. John H. Rex of the University of Texas Medical School at Houston, said in April 1995 that the dosage in such a study “probably, actually, should be higher.” Asked if he favored the higher dosage even for the yeast-based fungal infections, Rex added: “The general feeling is ... somewhere between 0.6 and 1 is the correct dose.”

Walsh had foreshadowed concern about using low-dose amphotericin for suspected aspergillus. In a 1990 article published by Seminars in Oncology, Walsh and a colleague wrote: “When Aspergillus pneumonia is suspected or proven, higher doses of amphotericin B (1 to 1.5 mg/kg/d as opposed to the standard 0.5 mg/kg/d used in other infections) are indicated to optimize successful outcome.”

A dose of 0.6 milligram per kilogram of body weight, daily, for patients with suspected but unproven infections “obviously is not sufficient against” aspergillus, Walsh and his co-authors wrote in a 1991 article published by Reviews of Infectious Diseases.

In the AmBisome study, Walsh supported using the 0.6-milligram dosage. Walsh had told the FDA committee in 1995, without referring to the ongoing study with Fujisawa, that he preferred “flexibility in dosage.” This would allow increases if a patient faltered. He said that “some experimental data” suggested that higher doses of conventional amphotericin might be more effective.

The study treated 687 patients: 343 with AmBisome, at 3 milligrams per kilogram of body weight, daily; and 344 with conventional amphotericin, at 0.6 milligram.

The patients, treated at sites throughout the United States and ranging in age from 2 to 80, were enrolled within 16 months, at what Walsh later called “a remarkably rapid rate.” He also would describe the patients as a “very high-risk population,” vulnerable to fungal infections.

Of the patients given conventional amphotericin, 36, or 10.5%, died. Of patients given Fujisawa’s drug, 25, or 7.3%, died.

Those who oversaw the treatments concluded that fungal infection was the primary or contributing cause of death for 11 who received conventional amphotericin and for four treated with the Fujisawa drug. The remaining deaths were attributed to other causes.

On July 16, 1997, Walsh anchored Fujisawa’s presentation of AmBisome to the FDA advisory committee, which met in Silver Spring, Md. The FDA’s agenda listed Walsh as part of the “Fujisawa USA Presentation.”

Fujisawa’s vice president for regulatory affairs, Jerry Johnson, told the FDA committee: “Our presentation will conclude with Dr. Walsh presenting the key results from the U.S. study.”

Walsh narrated a series of slides and told the committee that AmBisome “was more effective in preventing proven invasive fungal infections and fungal-infection-related deaths” than conventional amphotericin.

Within hours, the advisory committee voted unanimously in support of the new drug.

On Aug. 11, 1997, the FDA approved AmBisome for treating presumed fungal infections in children and adults. The dose approved by the FDA -- 3 milligrams per kilogram of body weight, daily -- was the same as that used in the study.

The next month, Walsh told a conference of physicians and research scientists in Toronto that AmBisome was “the first agent shown to be superior to amphotericin B in reducing proven, invasive fungal infections in cancer patients.” AmBisome, he said, was “a new standard” in treatment. Within days, Fujisawa began marketing AmBisome in the U.S.

In the FDA’s final review of the new drug, statistician Thomas Hammerstrom wrote that although AmBisome was similar in effectiveness to amphotericin, there were “inadequate scientific grounds” to judge it superior.

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.

A summary of their research said the high dosages “are safe, well-tolerated, and can provide effective therapy for aspergillosis” and similar infections. (Two years later, their full-length article on the study repeated that conclusion but also said the study did not have enough patients to prove which dosages worked best.)

At a symposium to discuss the treatment of aspergillus infections last October in San Francisco, Walsh was asked by a physician which maximum dose of AmBisome he recommended.

“Certainly, we want to think that more is better,” Walsh replied, adding that while results from clinical trials did not support using more than 5 milligrams per kilogram of body weight, daily, there were data, based on safety and drug concentration in the blood, suggesting a benefit at 7.5 to 10 milligrams.

From January 2000 through August 2002, 116 hospitals and clinics worldwide carried out the study of Cancidas.

The 1,095 patients with suspected fungal infections received either Cancidas or AmBisome. The patients ranged in age from 16 to 83.

At around the same time, Merck persuaded the FDA to conduct a fast-track review of Cancidas for a more narrow use: treating aspergillus in patients who had either not tolerated or failed to improve while taking another antifungal drug.

On Jan. 10, 2001, representatives of Merck -- assisted by Walsh -- presented the company’s case for approval of the drug to the FDA advisory committee in Bethesda.

Walsh, in his statement to The Times, said: “I did not appear as a consultant to Merck.”

But that is how Merck identified him to the FDA committee, both orally and in a slide.

Tamara Goodrow, a Merck regulatory affairs official, said: “Merck has brought several consultants to the meeting today so that they are available to facilitate the advisory committee’s discussion and deliberations.” Goodrow then named the consultants, including “Dr. Thomas Walsh.”

Another Merck official said Walsh served as the head of a company committee of three researchers who assessed how patients with aspergillus infections had responded to treatment with Cancidas in the smaller company study.

Several members of the FDA advisory committee voiced concern about the validity of the small study involving 69 patients. They pointed out that it lacked a “control” group of patients treated at the same time to gauge the comparative effectiveness of Merck’s drug.

They questioned whether the study proved that Cancidas provided patients with a measurable benefit.

At that point, a videotape of the meeting shows, Merck’s senior director of clinical research, Dr. Carole A. Sable, gestured to the audience and said: “Perhaps Dr. Walsh, who is actually the head of our expert panel, would like to make a comment.”

Walsh strode to the podium, took the microphone and assured the FDA committee that Merck’s case-by-case information for the 69 patients was reliable. “I think this was really the largest and most robust set of data that we’ve ever had on individuals,” Walsh said. He said his “whole section” of NIH government scientists had assisted him in reviewing Merck’s data.

A member of the FDA committee, biostatistician William Blackwelder, asked Walsh if he was “confident” that the patients with worsening aspergillus infections had benefited from Merck’s drug. “Yes, indeed,” Walsh concluded.

The committee endorsed the approval of Cancidas for treating patients with aspergillus who had not responded to other drugs. On Jan. 26, 2001, the FDA approved Merck’s application to market it for that narrow use, although doctors were at liberty to prescribe Cancidas as they saw fit.

A Merck spokesman said recently that Walsh was paid a total of $3,000 in fees, in 1999 and 2001, not related to his involvement with the company’s drug. Walsh said in his statement to The Times that Merck had not paid him for any appearance before the FDA.

U.S. conflict of interest law generally prohibits a federal employee from representing anyone before a government agency, regardless of whether outside compensation is paid.

“Outwardly, it looks like it could be a problem,” said John M. Treacy, who formerly directed operations of the FDA’s advisory committees.

Meanwhile, Merck’s hopes for wider use of Cancidas -- in patients with presumed but unproven fungal infections -- rested with the large international study that continued through late 2002.

The results were published in the New England Journal of Medicine on Sept. 30, 2004, with Walsh listed first among the authors. Patients who were given AmBisome as a comparator fared somewhat worse than those who got Merck’s Cancidas: Of the 539 patients given AmBisome, 75, or 13.9%, died. Of 556 patients given Merck’s drug, 61, or 11%, died.

A sharper contrast was reported among 24 patients described as having an aspergillus infection: 11 of the 12 given the older drug died or were otherwise not treated successfully, compared with seven of the 12 patients who got Merck’s drug.

Walsh and his co-authors, who included four Merck employees, noted the differences in their journal article: Among the patients with “baseline” fungal infections, “the rate of death during the study was lower in the [Cancidas] group.” Overall, they said, Cancidas offered “improved survival.”

On Sept. 29, 2004, the FDA approved Merck’s application to market Cancidas for treating patients with presumed fungal infections. On Dec. 8, 2004, a Merck executive told stock-fund managers that the approval could “give us a great opportunity to increase sales in 2005.”

A month later, Dr. Francisco Marty, a specialist from Brigham and Women’s Hospital in Boston who also is an instructor at Harvard Medical School, voiced concern about the Walsh-led study in a letter published in the New England Journal of Medicine.

Patients with early fungal infections who were given AmBisome “may have received suboptimal doses of that drug at a time when frontloading of therapy is critical to gain control of the infection,” Marty and a colleague wrote.

Although the initial dose selected for AmBisome was the same as in the earlier major study, Marty’s letter pointed out a distinction:

In the newer study, the dose was not supposed to be increased until a patient had received treatment for five days on the original dose -- and had continued to deteriorate. (A patient also could be removed from the study and treated differently at the discretion of the physician.)

In an interview at his Boston office, Marty said that the patients whose infections were found early in the Merck study and who were given the lower dose of AmBisome may have been put at a disadvantage.

“You have a bad infection and you don’t get enough drug, you may be dead,” Marty said. He noted that the medical-practice guidelines -- cowritten by Walsh -- suggested a dose of 5 milligrams per kilogram of body weight for aspergillus.

For those patients, Marty said, “you’re not doing a good job with 3 milligrams.”

Other doctors who wrote to the New England Journal of Medicine raised questions. An unusually low percentage of patients in the AmBisome group responded favorably to treatment, wrote Dr. Dimitrios P. Kontoyiannis and a colleague from the University of Texas M.D. Anderson Cancer Center in Houston.

In reply, Walsh wrote in the journal that various groups had advocated both higher and lower dosages of AmBisome. The use of 3 milligrams per kilogram of body weight per day, he and two co-authors wrote in January 2005, was “the most tenable initial dosing strategy.”

Walsh, responding to questions for this article, said that the five-day provision in the second study was intended to standardize the conditions for increasing the dosages. He said the provision was approved by consensus of the participating institutions on the belief that it would not put patients at added risk.

In his statement last week, Walsh pointed to results from a recently completed study, suggesting that a 3-milligram dosage of AmBisome was about as effective against aspergillus as was a 10-milligram dosage.

The points made on his behalf recently by his superiors and by other letter signers, Walsh said, “conclusively refute any possible contention that the two clinical trials violated a standard of care or otherwise called for inappropriate dosages of antifungal medications.”

A 2005 book, “Fungal Infections in the Immunocompromised Patient,” written for doctors caring for patients most at risk, concluded that “much controversy still surrounds the optimal timing, dosage and duration of therapy” for patients with the suspected infections.

Furberg, the former NIH clinical research specialist, said the two major antifungal studies fell short because they left unanswered which drug or dose was best against suspected infections.

“When you set up studies with controversial comparisons, you risk misleading everybody -- regulatory agencies, physicians and patients,” said Furberg, now a professor at Wake Forest University.

Times researcher Janet Lundblad in Los Angeles contributed to this report.