Added indignity: NIH labels Fisher’s papers as “SCIENTIFIC MISCONDUCT”

The midterm elections of 1994 brought on political change that cost Dingell chairmanship of Energy & Commerce and Oversight and Investigations. He lost most of his staff, and there was no reason to continue oversight hearings of NSABP.

Also, at the end of the year, NCI Director Samuel Broder announced that he would go off and join a company making a taxane drug.

Though the players changed, the NSABP scandal continued to chug along, now driven by a volatile mixture of legal procedure, institutional politics, futile investigations, bureaucratic inertia—and, for the lack of a neutral word—idiocy.

Thus, in February 1995, I reported a shocking story about annotations that were placed on Fisher’s and NSABP’s publications. This story is being reprinted in full:

Under ultimatum, NIH databases remove “misconduct” tags from papers by Fisher

Vol. 21 No. 8 | Feb. 24, 1995

Users of medical literature databases run by NCI and the National Library of Medicine have been finding an announcement tagged to papers that list Bernard Fisher as an author:

Another tag proclaimed:

Scientific misconduct? By whom?

The HHS Office of Research Integrity has not announced the findings of its investigation of Fisher. The only misconduct related to his case was committed by Roger Poisson, a Montreal surgeon.

Does the Poisson case warrant inserting the “misconduct” tag on at least 88 papers that list Fisher as an author?

Among those papers are primary studies that clearly include Poisson’s fraudulent data. However, also tagged were publications in which Fisher expresses his opinions, a review of literature by The Cancer Letter confirmed. The review also found that publications are flagged inconsistently in Medline and Cancerlit, two databases operated by NIH. Also, at least one paper was flagged by Cancerlit even though the data was obtained prior to the first documented incident of fraud at Poisson’s hospital.

Last week, an attorney for Fisher gave NIH an ultimatum: Remove the tags within 48 hours or face legal action. In response, officials at the HHS Office of Research Integrity ordered that the words “scientific misconduct” be struck from the tags.

The question that remains is what is to be done with the remainder of the tag: DATA TO BE REANALYZED. Should it continue to adorn Fisher’s “The Evolution of Paradigms for the Management of Breast Cancer: A Personal Perspective”?

“It’s very simple,” said Robert Charrow, Fisher’s attorney. “They ought to pull the flags off all the papers until they can figure out what to do.

“They likely owe Dr. Fisher some damages,”

Charrow, of the Washington firm Crowell & Moring, said to The Cancer Letter.

Charrow contends that NIH officials went beyond warning cancer researchers about fraudulent data from Montreal.

“It appears that NLM and NCI designated articles without regard to the data in those articles, and instead, based their decision solely or largely on whether Dr. Fisher was an author,” Charrow wrote in his Feb. 15 letter to NIH Legal Advisor Robert Lanman.

In an interview with The Cancer Letter, Lyle Bivens, ORI director, said the wording of the tag was an oversight on his part.

Bivens said the tags were written by the National Library of Medicine. ORI provided the library with a list of papers that were believed to include data from Poisson’s institution, St. Luc Hospital, he said.

“We didn’t ask for a ‘scientific misconduct’ flag on it,” Bivens said “NLM is used to that when they get a request from my office, because usually it is as a result of a misconduct finding.

“I was not explicit enough in what the statement should have been,” he said.

Bivens said that earlier this week he directed that the words “scientific misconduct” be removed. “Today I sent a memo to NLM asking that they take ‘scientific misconduct’ label off,” he said in an interview Feb. 21.

An NLM official contradicted Bivens’s statement, saying that the tag was written by ORI.

In an interview with The Cancer Letter, Lois Ann Colaianni, NLM associate director for library operations, said ORI had specifically asked the library to use the words “scientific misconduct.”

Colaianni said NLM did not select the papers for tagging.

“We had nothing to do with identifying the papers,” she said. “It was through a special request that we labelled these.

“Generally we steer away from labeling papers. We have retractions, and errata, and comments. At the time these went in as comments. However, there was concern that it wasn’t enough to cause people to read the reanalyzed data,” she said.

From documents and interviews The Cancer Letter has learned that the ORI staff selected the papers that were ultimately tagged.

Removing the words “scientific misconduct” does not wipe out the damage to Fisher and to scientific literature, Charrow said.

“Had this language been contained in a news report and had Medline and Cancerlit been a private publication, their actions would have constituted defamation,” Charrow said.

“While the government may believe it is immune from defamation claims, private publishers are not,” Charrow said. “We are advising the scientific community that we will take action against anyone who republishes the defamatory statements contained in Medline.”

The NIH tags are beginning to filter into medical 1 iterature. Two flags can be found on p. 318 of the Feb. 15 issue of the Journal of the National Cancer Institute.

Fisher case different from the start

ORI’s actions in the Fisher case were unprecedented, Bivens said.

“This is the first case we’ve had where we put out a notification prior to any scientific misconduct finding,” he said in an interview.

“We made a commitment to let the clinical community know even prior to an investigation if there is a problem with publications that may inform treatment decisions,” he said.

Asked why ORI decided to go beyond flagging publications that involved Poisson data, Bivens said, “As the situation developed, it became a possibility that either bad data or suspect data might be contained in other publications.

“That is the primary question we are asking.” A review of publications was an essential part of the Fisher case, Bivens said. ORI had to compile its own list because Fisher’s cooperative group, the National Surgical Adjuvant Breast & Bowel Project, failed to provide a complete list of publications to the investigators, he said.

“We never have gotten a complete list of publications from NSABP that we feel we need,” Bivens said. “We needed to find out what publications had been submitted from NSABP when it was known that St. Luc data was shown to be falsified.”

The anatomy of a tagging

From documents and interviews, The Cancer Letter was able to reconstruct the process that resulted in the flagging of Fisher’s publications.

On April 25, 1994, NCI Director Samuel Broder wrote a memorandum to Donald Lindberg, NLM director, in which he asked the library to help the Institute ensure that the databases denote serious error, fraud or scientific misconduct in research supported by NCI.

The memorandum, which did not mention Poisson, Fisher or NSABP, sought to set up a mechanism for NCI and NLM to work together.

In a May 4 memo, Lindberg informed Broder that Colaianni was designated to work with NCI on denoting in the literature episodes of misconduct and fraud.

Broder designated Susan Hubbard, head of the NCI International Cancer Information Center, to work on the project. ICIC runs the Cancerlit database.

Initially, ORI officials requested that NCI compile

a list of papers that could have been affected by fraud.

However, documents indicate that NCI officials did not perform the selection of papers that were subject to flagging. Instead, NCI provided to ORI a list of NSABP publications, leaving it to the investigators to decide which publications ought to be flagged.

A memorandum from Bivens dated May 20 and addressed to Colaianni confirms that ORI investigators were working with a list of NSABP publications, crossing out the publications that in their judgment were unrelated to the clinical trials in question.

Attached to the memorandum was a list on which some entries were marked with an “X.” The library was instructed to avoid flagging the articles so marked.

However, in the memorandum, Bivens said the list was compiled with the help of NCI. In an interview with The Cancer Letter, he repeated that statement.

“NCI gave us a list,” Bivens said. “NCI identified the articles as containing data that might need to be reanalyzed.”

Sources said that the same scientific papers were to be flagged in both NLM’s Medline and NCI’s Cancer lit.

However, Charrow’s letter includes 88 “unique identifier” numbers for publications authored by Fisher and contained in Cancerlit and 19 identifier numbers for publications cited in Med line. A review of the databases by The Cancer Letter indicates that many of the articles flagged in Cancer! it were not flagged in Medline.

“The action that was taken was to try to flag articles that had data from St. Luc,” Hubbard said to The Cancer Letter. “The purpose was not to make the cancer community feel that Dr. Fisher was responsible for scientific misconduct. The language used did not make that clear.

“If we were to do it again, we would all do it differently.”

Hubbard said that once NSABP’s reanalysis of the B-06 study is published, the “data to be reanalyzed” flag will be removed from both Medline and Cancerlit. Instead, a comment will refer readers to the publication of the reanalysis.

Can all data from St. Luc be excluded?

The controversy over what is to be flagged is likely to rekindle the question of whether the government has the legal authority to exclude all St. Luc data from the trials, regardless of whether the data were submitted by Poisson or other researchers.

So far, the government has been guided by the ORI recommendations contained in the 1993 report on the investigation of Poisson. “The reliability of the entire data set from St. Luc Hospital remains questionable,” Bivens wrote in a memorandum that accompanied that report.

“It would not be unreasonable to exclude all data on patients from this institution from any future analyses,” Bivens wrote.

The ORI report on the Poisson investigation listed only five publications affected by scientific misconduct.

The ORI report did not claim to offer the authoritative list of affected publications. However, another list, compiled by NSABP interim leadership following Fisher’s firing, claimed to be complete.

That list, contained in an appendix to an action plan for restructuring the cooperative group, listed

18 papers submitted between 1986 and 1994.

In an interview, Fisher’s attorney Charrow said he plans to challenge the exclusion of all St. Luc data regardless of whether it was submitted by Poisson.

“The only person they had any proof against was Dr. Poisson,” Charrow said to The Cancer Letter. “There is no proof against any other physician at St. Luc who was enrolling patients in NSABP clinical trials.”

Dangerous data?

A review of literature by The Cancer Letter found that at least five papers labelled “scientific misconduct” were expressions of opinion by Fisher.

Two of those publications listed Fisher as the only author:

• “The Evolution of Paradigms for the Management of Breast Cancer: A Personal Perspective,” Cancer Research, 52(9):23 71-23 83, 1992. The article appeared under the heading “Perspectives in Cancer Research.”

• Fisher’s 1992 Steiner Award lecture, published in International Journal of Cancer, 55(2): 179-180, 1993.

Also labelled were:

• “On the Underutilization of Breast-Conserving Surgery for the Treatment of Breast Cancer,” an editorial by Fisher and Leora Ore, published in the Annals of Oncology, 4:96-98, 1993.

• “New Perspectives on Cancer of the Contralateral Breast: A Marker for Assessing Tamoxifen as a Preventive Agent,” an editorial by Fisher and NSABP biostatistician Carol Redmond, JNCI, Sept. 18, 1991.

• “Adjuvant Therapy in Node-Negative Breast Cancer. A Panel Discussion.” The discussion between Fisher, William McGuire, Martin Abeloff, John Glick, I. Craig Henderson and C. Kent Osborne was published in Breast Cancer Research and Treatment, 13(2):97-115, 1989.

The only possible explanation for flagging editorials and panel discussions is their reliance on NSABP studies, Charrow said. However, if that criterion is to be applied to Fisher, it should be applied to other authors who cite NSABP studies, he said.

“Followed to its illogical conclusion, a warning flag should be placed on the conclusions of the final document of the 1990 NIH Consensus Development Conference on the Treatment of Early-Stage Breast Cancer,” Charrow said.

Drawing heavily on NSABP data that included patients from St. Luc, that conference concluded that breast preservation is the preferable treatment for women with stage I and II disease.

Differences between databases

In at least one case, a tag was placed on a paper from a study that clearly fell outside the time frame of the Poisson investigation, which, according to ORI documents, found that falsified records at St. Luc existed since 1976.

The study traced long-term mortality among patients who received radiation treatment prior to 1975:

“Cause-Specific Mortality in Long-term Survivors of Breast Cancer Who Participated in Trials of Radiotherapy,” John Cuzick, et al., Journal of Clinical Oncology, 12(3):447-453, March 1994. The paper, which lists Fisher among the authors, was flagged both in Medline and Cancerlit.

A review of Medline and Cancerlit shows that publications were not flagged in a coordinated fashion.

Medline did not flag at least three papers that listed Poisson among authors and included fraudulent data. Two of those papers were cited in the ORI report on Poisson.

The papers are:

• Fisher et al., “Eight-Year Results of a Randomized Clinical Trial Comparing Total Mastectomy and Lumpectomy With or Without Irradiation in the Treatment of Breast Cancer,” New England Journal of Medicine, 320( 13): 822-828, March 30, 1989.

• Fisher et al., “A Randomized Clinical Trial Evaluating Tamoxifen in the Treatment of Patients with Node-Negative Breast Cancer who Have Estrogen-Receptor-Positive Tumors,” NEJM, 320(8):479-484, Feb. 23, 1989.

Another paper that escaped the flag in Medline despite the fact that it listed Poisson as an author and contained St. Luc data was:

• Fisher et al., “Two Months of Doxorubicin­ Cyclophosphomide With and Without Interval Reintroduction Therapy Compared With Six Months of Cyclophosphamide, Methotrexate, and Fluorouracil in Positive-node Breast Cancer Patients With Tamoxifen-Nonresponsive Tumors: Results from the NSABP Project B-15,” Journal of Clinical Oncology , 8(9):1483-1496, 1990.

All three papers were flagged in Cancerlit.

In fact, the parameters of Cancerlit appeared to have been altered to allow for tagging of an expanded number of Fisher’s articles, Charrow said.

Typically, Cancerlit citations are arranged in reverse chronological order, and at this time, the database runs back from 1994 to 1988. However, in the case of Bernard Fisher, the bottom boundary drops back to 1979, Charrow said.

For those extra nine years, a Cancerlit user sees nothing but flagged papers by Fisher. A literature check for entries on Poisson found three papers written before 1988.

An unpopular action

The flagging of Fisher’s papers has met with sharp criticism from clinical cancer researchers nationwide.

• “This is the computer equivalent of the air brush that removes people from the reviewing stand at the May Day parade,” said O. Ross McIntyre, former chairman of the Cancer and Leukemia Group B, invoking imagery from the Moscow Trials of the 1930’s.

• “I think this is the most unfortunate approach and a disservice to clinical trials,” said Norman Wolmark, chairman of NSABP. “I hope NCI will remedy this transgression.”

• ”It seems unbelievably extreme to me,” said Charles Coltman, chairman of the Southwest Oncology Group, who said he was surprised by NCI’ s use of the tag line beyond papers that reported primary clinical trials.

“I don’t know where you stop when you begin doing that,” Coltman said. “I am not surprised that Bernie and his attorneys are outraged by this approach.”

• “It’s gratuitous, vindictive, inaccurate, and it contributes nothing to our understanding of cancer,” Emil J Freireich, professor of oncology and hematology at M.D. Anderson Cancer Center said.

Freireich said he was stunned to find the tag on Fisher’s “Evolution of Paradigms” paper.

“That paper is a brilliant, innovative, original formulation of the modern paradigm of breast cancer, and it has been confirmed over and over again,” he said. “There is no controversy about the science here. This is personal.”

• “This is the blunderbuss approach to government,” said James Holland, professor at the Mt. Sinai School of Medicine.

“Bernard Fisher is one of the towering figures in medical science of the last half of the 20th century. To paste him as if he were a villain is a complete disregard of the scientific process.

“I react to this with dismay that the NLM has been dragged in to the fiasco that I believe represents the conduct of NCI in this attempt to sort out the problems that faced NSABP.”

• ”First of all, I think the data from all NSABP trials has been reanalyzed, both with the fraudulent data included and excluded, and Dr. Fisher eloquently presented this analysis at the May 1994 meeting of the American Society of Clinical Oncology,” said John Glick, director of the Univ. of Pennsylvania Cancer Center and ASCO president-elect.

“While those results have not been published, the oral presentation showed that none of the NSABP scientific conclusions were altered by the removal of fraudulent data. Therefore, the NSABP contribution to scientific advancement of breast cancer research remains valid.

“Obviously, we are awaiting the republication of NSABP papers in peer reviewed journals, and I think those papers will convince the public that none of the NSABP results have significantly changed,” Glick said.

Asked whether he believes that the publication in which he appears as a coauthor warranted a “scientific misconduct” tag, Glick said:

“That was a panel discussion, and there is nothing in that paper that would warrant any warning about scientific misconduct whatsoever.”

Letter to the Editor

Two weeks later, Lyle Bivens, then director of the HHS Office of Research Integrity, explained that he didn’t mean to say that the “misconduct” annotation and his comments were not intended as presumption of guilt on the part of Fisher.

I still fail to follow the curvature of his clarification, but here it is:

Bivens: No intent to link flags with NSABP inquiry

To the Editor:

In your Feb. 24 issue, I was quoted as saying that, “This is the first case we’ve had where we put out a notification prior to any scientific misconduct finding.” However, I did not mean to connect the issuance of the Medline and Cancerlit notifications to an ongoing ORI investigation of NSABP.

As stated in my original request to the National Library of Medicine, I asked that a flag be placed in Medline “to indicate that a reanalysis of the [NSABP) study may be needed based on a [June 1993] finding of scientific misconduct on the part of one of the contributors [Dr. Poisson].” The Medline and Cancerlit notifications were predicated solely on the confirmed findings of scientific misconduct by Dr. Poisson, a contributor to the data base underlying anumber of NSABP publications, and were not related to the current NSABP investigation.

Although the current NSABP investigation will examine the data underlying a number of NSABP publications and may therefore generate a fact-base to indicate what publications need reanalysis due to the Poisson misconduct, it is only in this sense that the Poisson and the ongoing NSABP case are related. My comments were not intended to, and should not be construed as, a determination that any authors of NSABP publications, other than Dr. Poisson, have engaged in scientific misconduct.

I did not intend to link the NSABP investigation to the notifications, and to the extent that message was conveyed, it was in error.

Lyle Bivens

Director, Office of Research Integrity

In September 1997, after years of litigation against NCI, the University of Pittsburgh, and the law firm of Hogan & Hartson, Fisher received an apology and a check for $2.75 million.

Here is The Cancer Letter’s story about the settlement:

Bernard Fisher settles suit for $2.75M, retains title; University apologizes

Vol. 23 No. 34 | Sept. 5, 1997

Cancer researcher Bernard Fisher last week agreed to drop his suit against NCI, the University of Pittsburgh and the Washington law firm of Hogan & Hartson.

Under an agreement that settles the suit, Fisher will receive $2.75 million and retain his title of Distinguished Service Professor, but would collect no salary or employee benefits, legal documents state.

Pursuant to the agreement, the university issued an apology for “any harm and public embarrassment that Dr. Fisher sustained which was in any manner related to the activities of the University of Pittsburgh and/or its employees.”

NCI, also a defendant in the civil action brought by Fisher, issued a statement that enumerated Fisher’s contributions to the understanding and treatment of breast cancer.

“Through his role as the scientific leader of the National Surgical Adjuvant Breast and Bowel Project, [Fisher] has not only changed the way breast cancer is treated, but enlightened medical science to view breast cancer as not just a tumor confined to the breast, but as a systemic disease requiring more than surgical intervention,” NCI said in a statement.

The Institute contributed $300,000 to the overall settlement, to cover a portion of Fisher’s legal expenses, sources said.

The settlement was reached Aug. 27, six days before the case was scheduled to go to trial at the US District Court for the Western District of Pennsylvania.

Though all parties agreed not to discuss the terms of the settlement agreement, The Cancer Letter obtained a copy of the document under the Freedom of Information Act.

Tangled controversy concluded

“I am glad to be alive to see this vindication,” Fisher said to The Cancer Letter. “I feel that I am still in a position to continue to make contributions, and I want to go forward in the best way I can: to write, and to complete data that needs to be put out.”

Fisher declined to describe his plans for the future. “My plans are in the process of being formulated,” said Fisher, who is 78. “At this time, I am scientific director of the NSABP, and I would like to continue in that mode, and to make whatever contributions that I can to the organization as it now exists.”

The scientific director’s post can have a significant impact on generating interest in clinical trials and development of protocols, sources said.

The statements by NCI, the University of Pittsburgh, and Fisher appear on pages 3 and 4.

The settlement is likely to conclude the tangled controversy that began on March 13, 1994, when an article in the Chicago Tribune disclosed that Montreal surgeon Roger Poisson had contributed falsified data to NSABP clinical trials. Soon after the publication of the story, Fisher was removed from leadership of the cooperative group.

The oversight and investigations subcommittee of the House Committee on Commerce conducted hearings on the matter, and the HHS Office of Research Integrity was brought in to investigate possible misconduct by Fisher and two other officials at the cooperative group.

Ultimately, the subcommittee, then headed by Rep. John Dingell (D-MI), accepted the mea culpa from NCI and Pitt and bowed out. Earlier this year, the ORI completed its investigation, finding no misconduct by Fisher and other NSABP officials (The Cancer Letter, March 7).

In the just-settled suit, Fisher claimed that NCI officials had “unlawfully terminated” him as principal investigator of NSABP and “crafted multiple false accusations” against him.

“In an effort to keep millions of federal research dollars flowing to the University,” Pitt officials assisted NCI in Fisher’s firing, Fisher’s attorneys stated in the most recent version of the complaint, filed in December 1995.

The suit also named Martin Michaelson, an attorney with Hogan and Hartson, who was hired by the university to handle the matter in its initial stages.

“Defendants Michaelson and Hogan & Hartson obtained Dr. Fisher’s confidences by representing that a privileged attorney-client relationship existed between them,” but ultimately shared these privileged and confidential communications with the university, NCI, ORI, and the staff of the Subcommittee on Oversight and Investigations of the House Committee on Commerce, the complaint states.

Under the settlement agreement, Fisher is to receive a single check for $2.75 million, documents say. The check would be issued by the University of Pittsburgh “on behalf of all defense interests” within 45 days of the signing of the agreement

The issue of integrity of the research is now completely put to rest. There wasn’t anything that would have altered data or would have changed the results.

Bernard Fisher

While the federal government will contribute$300,000 toward the settlement, it is unspecified how much of the remaining $2.450 million would come from Pitt and how much (if anything) would come from Hogan & Hartson.

Though Fisher will retain his title, he would collect neither a salary nor employee benefits. If he chooses to stay at the university, Fisher would be provided with office space, but no support staff. If he accepts a position elsewhere, he would receive no office space.

Regardless of whether he stays or goes, Fisher would continue to have unrestricted access to NSABP data kept at the university, the agreement states.

In the document, federal defendants stated that all investigations surrounding Fisher have been concluded and that all previously imposed sanctions have been lifted. The text of that section of the memorandum follows:

• “Dr. Fisher is not required to submit his manuscripts to the NCI prior to publication… [Any] requirement to that effect was rescinded April 10, 1995.

• “Dr. Fisher is not precluded from participation in any federally funded cancer research project;

• “A grant applicant, including any applicant for the Operations Center [based at Allegheny General Hospital in Pittsburgh] or Biostatistical Center [based at the University of Pittsburgh] NSABP grants, may list Dr. Fisher as a participant on an application, and any such application will be reviewed through the normal peer review process;

• “NCI will consider Dr. Fisher for a position on a top advisory committee at the NCI, taking into account his achievements and reputation;

• “After a thorough investigation, the Office of Research Integrity did not make a finding of scientific misconduct.”

Pitt apologizes to Fisher, expresses pride in his work

The text of the joint statement by Fisher and the University of Pittsburgh:

NCI: Fisher a dominant force in breast cancer for 40 years

The text of the NCI statement:

Fisher: “So many people don’t understand clinical trials”

Bernard Fisher’s statement to The Cancer Letter: