The conclusions of 14 published breast and colon cancer studies are “monotonously similar” with or without fraudulent data submitted by a Montreal surgeon, Bernard Fisher said to clinical cancer researchers in Dallas last week.
Women enrolled in studies conducted by the National Surgical Adjuvant Breast & Bowel Project or treated as a result of the cooperative group’s published data have received appropriate treatment, said Fisher, the ousted NSABP principal investigator.
Standing ovations punctuated the beginning and the end of Fisher’s presentation at the plenary session of the American Society of Clinical Oncology annual meeting.
“Eight weeks ago…my life and that of my associates and that of the entire NSABP precipitously began to unravel,” Fisher said. “We and our families are completely devastated as a result of the recent events.”
Last March, The Chicago Tribune revealed that an NIH investigation concluded a year ago that Roger Poisson, of St. Luc Hospital in Montreal, had falsified enrollment data for women he entered onto NSABP studies.
Later in the month, an NCI investigation at NSABP headquarters at Univ. of Pittsburgh found the cooperative group had stopped auditing its member institutions in violation of NCI guidelines for cooperative group funding. NCI placed NSABP on probation, halted enrollment to its trials, and ordered the Univ. of Pittsburgh to remove Fisher as NSABP principal investigator.
Fisher, 75, citing ill health, declined to testify last month at a hearing of the House Oversight and Investigations Subcommittee, chaired by Rep. John Dingell (D-MI).
At the ASCO presentation Fisher looked robust.
Part of the way through the presentation, Fisher introduced Carol Redmond, former director of the NSABP Biostatistical Center, who said it was reasonable for NSABP to leave the St. Luc data in its publications even after learning about the fraud. “There are scientific and ethical justifications for not excluding these real patients who were randomized, treated, and followed up in NSABP studies,” Redmond said.
Fisher and Redmond are under investigation by the NIH Office of Research Integrity over inclusion of the data in question in studies submitted for publication.
ASCO officials estimated that approximately 6,500 meeting participants heard the May 16 presentation at the Dallas Convention Center.
Leaving the stage, Fisher raised his fist in a defiant salute. He refused reporters’ requests for interviews as a phalanx of ASCO officials and security guards escorted him out of the convention center.
Encore For An ASCO President
Setting a conversational tone for his remarks, Fisher began by referring to his term as 1992-93 ASCO president.
“I guess I’m the only past president who has ever been invited to give an encore,” he said.
“Last year I served as your president. That honor came to me after 35 years of total commitment to laboratory and clinical research aimed at understanding the biology of cancer and applying that information to improve the lives of women with breast cancer.
“Consequently, in my presidential address, I stressed the importance of laboratory and clinical research. I emphasized that clinicians could make important contributions to science by participating in large clinical trials, a mechanism to which I dedicated so much of my life. I indicated that such trials offer the best opportunity to obtain more credible, and definitive information that other mechanisms of data collection by individuals or a few physicians.
“Eight weeks ago, when information was beginning to come from recently completed NSABP protocols, and when new protocols that promised to result in major advances were either being conducted or initiated, my life and that of my associates and that of the entire NSABP precipitously began to unravel.
“This devastating circumstance had its origin in 1991. On Feb. 6, 1991, the NSABP headquarters unequivocally concluded that a physician at St. Luc Hospital in Montreal, one of several thousand NSABP investigators over the years, had falsified data. Accrual to that institution was immediately terminated. The NCI project officer was immediately notified. The Office of Scientific Integrity [now Office of Research Integrity] was notified by the NCI. OSI investigation was begun.
“We were instructed by the OSI that the matter was not to be discussed during the investigation. On April 26, 1993, two years later, we were notified of the OSI’s final action, and during the OSI investigation our data were reanalyzed by NSABP statisticians. Their analyses confirmed that women on our studies and those not on our studies but treated as a result of our published reports, had received appropriate therapy.
“Most importantly, this information indicated that no public health problem had occurred. In March of this year, a manuscript was submitted to the New England Journal of Medicine containing reanalyses of three major NSABP protocols previously published in that journal. One the NSABP B06 protocol evaluating the worth of lumpectomy; another, B13 which evaluated the worth of adjuvant chemotherapy in patients with negative nodes and negative estrogen receptors ; and B14, the protocol for patients with negative nodes, positives receptors who received tamoxifen.
“Many issues have been raised during the past eight weeks,” Fisher said. “In seems most appropriate in this forum, however, that we address the findings obtained following the reanalysis of data from all previously published NSABP clinical trials that contained patients from the Montreal institution.”
Redmond Defends Inclusion Of St. Luc Data
Fisher introduced Carol Redmond, chairman of the Dept. of Biostatistics at Univ. of Pittsburgh, to address the issue of reanalysis of the trials.
A comprehensive audit of St. Luc found that the data falsifications “almost exclusively” involved eligibility criterion for patients entering the trials, Redmond said. “There was no indication that the randomized treatment assignments were violated,” she said. “There was only one instance in the audit where follow-up information for study outcomes was apparently misrepresented.”
The issue arose as to how best to analyze multicenter trials where all data from one institution had to be excluded, Redmond said.
“Once the nature and extent of the data falsification that had occurred at St. Luc became known, NSABP statisticians considered that it was not appropriate to exclude all data on St. Luc patients,” Redmond said.
“The basis for leaving St. Luc patients in the data was the following:
“First, there was an ethical concern that excluding all St. Luc patients from follow-up and analyses after they had been treated would result in a failure to report toxicity for long term adverse effects that had already occurred among patients and might result in failure to report such events that might occur in the future.
“Second, statistical literature supports the inclusion of all patients who have been property randomized and followed in end result analyses. Inclusion of all patients as randomized, even those deemed ineligible for the protocol protects against the possibility of bias in assessing treatment effects introduced by selection of patients post randomization. Although it is common practice to present data eliminating ineligible patients, statisticians perform such analyses with great caution when a large proportion of ineligible patients are identified due to concern about post randomization biases. In such situations, statistical power and generalizability of the trial can be appreciably weakened with the exclusion of the ineligible patients.
“Thus, there are scientific and ethical justifications for not excluding these real patients who were randomized, treated, and followed up in NSABP studies.
NSABP clinical trials include “a variety of safeguards” against biases, Redmond said. “In that regard, NSABP data has always been gathered in prospective studies, and patients are randomized within as well as across institutions in a manner that helps to ensure balance among treatment groups.”
The large sample sizes and conservative monitoring rules in reporting data protect the trials when all patients are eliminated from a single institution, Redmond said.
Fisher presented a trial-by-trial analysis of 14 NSABP protocols, showing endpoint data with and without the St. Luc patients. He said the presentation was meant to simulate what was presented in the original publications.
Poisson enrolled 1,511 patients on 22 NSABP protocols, representing 4.5 percent of the total 34,000 patients entered, Fisher said.
The ORI investigation found that Poisson falsified data on records of 99 patients, or 0.3 percent of those entered onto the protocols. Of those, 98 falsifications related to information prior to randomization, Fisher said.
“None related to factors that could have affected the outcome,” Fisher said.
Fisher presented the reanalysis of breast cancer protocols B06, B07, B08, B09, B11, B12, B13, B14, B15, B16, and colon cancer protocols CO1, C02, C03, and R01.
Original findings in all of the studies were confirmed, Fisher said.
Fisher To ASCO: Ensure Academic Freedom
In a defiant closing, Fisher expressed support for clinical investigators and the cancer patients who volunteer for clinical trials, called for the resumption of enrollment to NSABP trials, and called on ASCO to remain “vigilant” in ensuring academic and scientific freedom.
“At its heart, clinical research must rely on the inherent integrity of its investigators,” Fisher said. “Clinical research can go forward only if investigators maintain the highest clinical and ethical standards. We continue to believe, as we have in the past, with only a few exceptions, investigators are honest and ethical.
“We reiterate the safeguards built into large multicenter trials protect the findings from becoming invalidated by events such as occurred in this situation,” he said. “Thus, it is not surprising that the results of our reanalyses confirmed our original findings and conclusions in every trial.
“While we and our families are completely devastated as a result of the recent events, we are even more concerned about the effect of those events on all women, those with and those without breast cancer,” Fisher said. “They must not become the victims of this scenario. Specifically, they must be assured that they have received appropriate therapies and that their participation in NSABP trials over the years has made a major contribution in advancing the treatment of breast cancer.
“To put NSABP clinical trials on hold at this time by suspending accrual to ongoing and planned studies could influence the lives of countless women by prolonging the time for obtaining answers to important questions about their treatment.
“We are grateful to have played a small part in advancing medical knowledge and we hope we have played a role in bringing to women comfort by enhancing the quality and longevity of their lives. We are also grateful to have had the opportunity to have played a seminal role in the implementation and conduct of many important breast and colorectal cancer studies, including the Breast Cancer Prevention Trial, which has already accrued two-thirds of the necessary participants.
“I will always maintain my allegiance to ASCO as long as it remains a strong force in shaping the science and practice of oncology in this country and it remains vigilant in protecting the best interests of patients with cancer by ensuring that intellectual, academic and scientific freedom and integrity are maintained.”
