Friday, October 19, 2007

Memory Hole (19 October): Harvard paper in "error"

12 years ago today: Harvard paper in "error"

On 19 October 1995 the journal Nature reported on the case of three Harvard Medical School researchers:

R. Alan B. Ezekowitz (now senior vice president at Merck),
John B. Mulliken and
Judah Folkman.

This is an instructive case and I have therefore provided it with an exclusive Memory Hole.

One wonders how scientific scientific misconduct or fraud might be defined given this piece. Below is the text reproduced from the database of Professor A.C. Higgins (highlighting added). Perhaps you can avoid anything simply by being important (see BMJ, 2007;335:582-583). This is Ouija board science.

The linguistics of the news report are Familiar to me relating to the first part of a publication trilogy involving myself.
"Inquiry at Harvard prompts research paper corrections,"
Dalton, Rex. Nature 377 (19 October 1995), p. 569.

Three prominent physicians from Harvard Medical School working at the Children's Hospital in Boston have been forced to publish an extensive correction in connection with federally funded research on a treatment for rare, life-threatening birthmarks in infants. The correction has been published in the New England Journal of Medicine (NEJM) after a lengthy investigation by Harvard into the conduct of the physicians, R. Alan B. Ezekowitz, John B. Mulliken and Judah Folkman. The latter is internationally renowned for his discoveries on angiogenesis, the process by which blood vessels grow.

It follows a briefer correction, published unilaterally last year by the three when the investigation was still under way, which included corrected data about some of the patients involved in the study, but without any accompanying comment - or even identification of the names of the authors.

The investigation concerned an article published in the NEJM (326, 1456 63; 1992) on the effectiveness of interferon alfa-2a in reducing large hemangiomas, growths that can occasionally block an infant's airway, affect its vision or compromise a vital organ.

The article, based on observations of 20 cases, claimed that the therapy had had a substantial effect. But Harvard officials say that after it was published, other physicians who had referred the children to the team complained about various inaccuracies.

The committee found that there was no misconduct or fraud, says Nathan. But it did find many errors needing correction. These included the fact that some hemangiomas were reported to have regressed more than they did; that the lesions were not measured, as the article suggested; that some patients were not treated for the period reported; that prior therapies used on some patients were not listed correctly; and that monthly blood tests were not taken on all patients as reported, with the result that no conclusion could be reached on drug toxicity.

The results of Harvard's investigation were sent to the National Institutes of Health (NIH)'s Office of Research Integrity (ORI), which, according to Nathan, accepted the university's findings. But it appears that neither the National Cancer Institute (NCI), nor the National Center for Research Resources -the NIH agencies whose grants were cited as funding the research- was notified of the Harvard investigation or of any corrective actions. Collette Freeman, the NCI grant officer, described the situation as "extraordinary" and "bizarre" when told of the correction.

Harvard officials refuse to comment on the case, or to say whether any sanctions were issued against the researchers. Nor were any of the three physicians prepared to comment. But in their correction, published on 31 August, they wrote: "We are embarrassed by our errors and regret them. We apologize to the readers for any misunderstanding and thank our colleagues for their assistance in correcting our report." Nathan called the interferon research "a very important paper" by "distinguished investigators." But, he claims, the researchers "got very busy" and "just slapped this paper together." He accepts that almost all the errors in the eight-page article tend to improve the apparent success of the therapy. "That is how the original complaint was made," says Nathan. "It is curious that it turned out that way."

The lengthy investigation appears to have been marked by contention, defiance by the researchers and anger. Nathan says that when the Harvard review started, the three "became very irritable," submitting material for correction to the journal against his advice. "The first correction wasn't certified by Harvard," said Nathan. "I felt it was wiser to be patient, collect all the errors and deal with them in one fell swoop," he says, adding that the medical school feels strongly that corrections have to be certified by the institution.

Jerome P. Kassirer, who took over as editor-in-chief at the NEJM after the first correction was published, says that the Harvard researchers submitted "material that was very confusing" for this correction. Earlier this year, he says, the three physicians submitted further material, approved by Harvard, but this was little easier to interpret. "The editors had a great deal of trouble figuring out what they were driving at."

Despite the errors, Nathan said the conclusions of the study remain valid, and that interferon remains the preferred therapy for such hemangioma cases. But some physicians feel a random, prospective study is still required to back up this conclusion, as the lesions are known to regress on their own as children age.

"I have to express reservations about any non-randomized study, especially one in which it now appears that the data was brushed up," says Ajay J. Vora, a paediatric haematologist at Children's Hospital in Sheffield, in the United Kingdom, who originally questioned the interferon study in a letter to the NEJM in October 1992.
Earlier|Later|Main Page


Anonymous said...

This appears to be the same kind of "science" that garnered FDA approval for rDNA insulin. I have carefully read the application submitted by Lilly to FDA. The fact that treatment of millions of diabetics (today) was premised on such suspect "science" is incredible. From the human insulin application, it appears that whenever unexplained events occurred, they were diminished, trivialized or ignored. Or, strangely, whenever the label of "identical" (to pork insulin) was beneficial to Lilly's application, human insulin was "identical" to pork insulin (in bG lowering effects) but was "different" when a differences was 'needed.'

This first-ever genetically-engineered, marketable drug SHOULD HAVE BEEN scrutinized with every tool available, and held to higher standards since the science needed to evaluate it was still emerging. Instead, it was treated as merely another small-molecule drug. Because he was so enamored of the "epochal event" that WAS rDNA insulin, Henry Miller was willing to quash unexplained (and perhaps significant) observations, protect them with the "identical" label that became quite convenient, and rush this product through the process.

Anonymous said...

The John B. Mulliken link is wrong.

Anonymous said...

Folkman has just died. His eulogy in Scientific American did not mention the fraud