Changes In System
REHIRED: Following her exoneration, Thereza Imanishi-Kari was named an associate professor at Tufts University.
The conclusion of the decade-long scientific misconduct case against Thereza Imanishi-Kari-she was exonerated in a June 21 decision of an appeals panel of the Department of Health and Human Services (HHS)-would appear to be a clear-cut victory for the Tufts University immunologist. Certainly it has been greeted that way by Imanishi-Kari and the other coauthors of a disputed Cell paper (D. Weaver et al., Cell, 45:247-59, 1986), including Nobelist David Baltimore of the Massachusetts Institute of Technology (MIT).
Imanishi-Kari tells The Scientist that she was frightened during the 10 years of numerous investigations. "But honest people can't come out with any other verdict," she says. Two days after the verdict, she celebrated with dinner at Baltimore's home.
Victory or not, many scientists and policy analysts use similar language to describe the case's resolution, terming it "a huge mess" or "a tragedy."
The case left observers and participants alike calling for changes in the government's mechanism for overseeing the integrity of biomedical research.
In the words of Suzanne Hadley-who, as an investigator for the Office of Scientific Integrity and then for a congressional committee, helped develop the evidence against Imanishi-Kari-the case was a "lose-lose" situation. Even leaving this case aside, Hadley says, "the system for dealing with alleged misconduct is a mess. The system is a shambles."
Baltimore hopes the case "will lead to a complete overhaul" of the Office of Research Integrity, the successor of OSI. "That organization is unable to distinguish between fraudulent activity and sloppiness," he states.
For the primary participants in the case, there are no winners. Imanishi-Kari had her career interrupted for 10 years. In 1986, she left her assistant professorship at MIT, where the disputed research was conducted, for what was to be an associate professorship at nearby Tufts University. She was instead hired as an assistant professor and was demoted to a contract researcher in 1994 when ORI issued a finding of 19 counts of scientific misconduct against her. Although continuing her research at Tufts with small private grants, she had no federal support after 1990. Earlier this month she was officially rehired as an assistant professor in the pathology department of Tufts' School of Medicine.
CHANGES URGED: Cell paper coauthor David Baltimore calls for "a complete overhaul" of ORI
When a scientific panel assembled by the National Institutes of Health found errors in the paper but no misconduct in 1989, then-NIH Director James Wyngaarden wrote Baltimore that if he and the other coauthors had "met to consider seriously the allegations or to reexamine the data," a full investigation might have been unnecessary. Baltimore does not agree. He says, "We paid attention to O'Toole at the start. The fact that we didn't agree [with her] doesn't mean we didn't pay attention."
The issue, however, shrouded his 18-month presidency of Rockefeller University. In a letter of resignation dated Dec. 2, 1991, Baltimore wrote, "The reason I have decided to take this step is that the Cell paper controversy created a climate of unhappiness among some in the university that could not be dispelled. . . . Trying to govern the university under these conditions has taken a personal toll on me and my family which I can no longer tolerate" (S. Hall, Science, 254:1447, 1991).
DISPUTE: Margot O'Toole contends that the issue was not merely a matter of interpretation.
She did not feel that any of the scientists she would normally ask for a recommendation, such as Imanishi-Kari herself or Henry Wortis, her thesis adviser and head of the informal Tufts inquiry that dismissed her complaints, would provide a good one, so she never asked, she contends.
Did the episode harm O'Toole's chances of finding work in immunology? Although "O'Toole recognized some real problems with the biology" of the Cell paper, comments Stanford University immunologist Lenore Herzenberg, who briefly considered hiring O'Toole herself, the entire affair "did leave a sour taste, which alone would have damaged her chances of getting work in the field."
Before Imanishi-Kari's appeal to the HHS Departmental Appeals Board in late 1994, no fewer than five investigations had been made of the Cell paper. At Tufts and MIT, friends of the coauthors conducted informal inquiries. Meanwhile, Walter Stewart and Ned Feder, a pair who have made a second career of investigating and exposing scientific misconduct, started examining the issues raised by O'Toole. Their continual questioning ultimately led Baltimore to call for an NIH inquiry.
In a March 17, 1987, letter to J. Edward Rall, deputy director for intramural research at NIH, Baltimore suggested that Imanishi-Kari would send materials from informal reviews at MIT and Tufts.
PARTICIPANTS: NIH's Walter Stewart, left, and Ned Feder aided Rep. John Dingell's subcommittee investigation.
NIH launched its investigation in May 1987. In 1989, an NIH panel of three scientists reported finding numerous errors but no evidence of fraud; they did not seriously question the authenticity of the data.
Before NIH completed its inquiry, the staff of Dingell's subcommittee began its own investigation. The staff was aided by Stewart and Feder, who were on loan from their permanent jobs as NIH intramural scientists.
At Dingell's behest, the United States Secret Service began analyzing data notebooks of Imanishi-Kari and others. When the Secret Service turned up irregularities in the notebooks, the newly established OSI reopened NIH's investigation. OSI evolved into ORI in 1992 and became part of HHS headquarters in response to dissatisfaction with NIH's handling of misconduct investigations. ORI completed its investigation in 1994, charging Imanishi-Kari with 19 counts of scientific misconduct. She appealed this finding.
The NIH Revitalization Act of 1993, which legislatively mandated the creation of ORI, also gave scientists charged with misconduct by the office an opportunity to appeal to the HHS appeals board. Most ORI findings go unchallenged, but the appeals board has reversed two other high-profile ORI cases in recent years, against Rameshwar Sharma at the Cleveland Clinic Foundation in August 1993 and against former National Cancer Institute researcher Mikulas Popovic in November 1993. ORI then decided not to contest the appeal of Popovic's lab chief, Robert Gallo. Thus, many observers had called the Imanishi-Kari appeal a "must win" for ORI.
The three-person appeals panel heard testimony for 28 days during the summer of 1995, adding 6,500 hearing transcript pages to an already voluminous record. Its 179-page ruling cleared Imanishi-Kari of all 19 charges against her and handed ORI a bitter setback. Observers largely expected the ruling, though perhaps not its sweeping condemnation of the government's case.
The appeals panel-composed of HHS lawyers Cecilia Sparks Ford and Judith Ballard as well as Julius Youngner, Distinguished Service Professor, Emeritus, at the University of Pittsburgh School of Medicine-wrote that the "main problem with ORI's case was not that more evidence was needed but that less equivocal and more independently decisive evidence was required for ORI to carry its burden of proof." Morale at the office is reportedly low, though staff members would not comment on the case at the direction of HHS officials.
He retracted it merely as a formality, he says, when the Office of Scientific Integrity (OSI) concluded in a draft report in 1991 that Imanishi-Kari had falsified and fabricated data. He never doubted the science in the paper. "I was always proud to have it on my C.V.," he says.
Imanishi-Kari, who was recently rehired as an assistant professor in the pathology department of Tufts University School of Medicine in Boston, where she has been since moving from the Massachusetts Institute of Technology in the summer of 1986, has continued to investigate the subject matter of the paper. Since losing National Institutes of Health support in 1990, she has been supported by small, private grants.
The Cell paper explored what happened when an antibody gene from one strain of mice was introduced into another strain of mice. The authors found what they termed an "altered repertoire" of antibodies in the transgenic mice. Several immunologists interviewed for this article say that while it may not be that surprising for an introduced gene to alter the antibodies produced in the recipient, the specific claim of the Cell paper was more startling. Imanishi-Kari and her coauthors reported that many antibody molecules encoded by the endogenous genes expressed a structural characteristic of the introduced antibody gene called the idiotype.
The assertion that an endogenous gene expresses the transgene's idiotype is simply a claim about a fact. It is an important fact, however, because the fact contains within itself the seeds for a mechanism, much as James Watson and Francis Crick's factual claim about the structure of DNA contained within it the seeds for a mechanism for its replication. In this case, Imanishi-Kari and her coauthors suggested that their observation lent support to a particular theory of cellular regulation of antibody expression, known as the "network theory."
Nobel Prize winner Niels Jerne of the Basel Institute of Immunology proposed the network theory in 1974. The theory is based on the idea that antibodies have determinants, called idiotypes, that are themselves recognized by different antibodies ("anti-idiotypes"). Jerne postulated that idiotypes could turn on anti-idiotypes, leading to down- or up-regulation of the original antibodies.
The Cell paper is not principally about the network theory. However, in several cases it asserts that its findings support the theory, which had limited experimental support at the time. According to Alan Stall, an assistant professor of microbiology at Columbia University's College of Physicians and Surgeons, who has confirmed many of the observations of the Cell paper, "there was some evidence, but not in an intact developmental system such as Imanishi-Kari was studying. That was why her findings were so potentially interesting."
Have these findings been confirmed? Imanishi-Kari says they have been by her own lab at Tufts. She cites a pair of 1993 papers (T. Imanishi-Kari et al., Journal of Immunology, 150:3311-26 and 3327-46). Stall does not question Imanishi-Kari's data, but says the specific observation of endogenous genes expressing the transgene's idiotype has not been confirmed by others, for the simple reason that "nobody has really tried to confirm" it.
In part, the requirement for highly specialized anti-idiotypic reagents necessary to study this particular system has discouraged others from repeating the work, he says. "Someone might have [repeated the work more carefully] if this whole stink wasn't raised," comments Stanford University immunologist Lenore Herzenberg, who has worked with her former student Stall on this system.
Moreover, the network theory, never very popular to begin with, has largely fallen out of fashion among immunologists, who now favor other explanations for the diversity and regulation of the immune system. Stall says he still believes idiotype networks may play a functional role in the immune system-but he is in no hurry to write grant applications based on the idea.
As for the many errors in the Cell paper-some of which were the subject of corrections published in Cell after the paper was written (T. Imanishi-Kari, D. Weaver, D. Baltimore, Cell, 57:515-6, 1989)-the authors maintain they are scientifically insignificant. Baltimore wonders whether any other paper "examined with the same microscope" as this one would not have just as many.
Hadley, the former acting director of OSI who is now on loan from NIH to George Washington University to analyze policy regarding scientific accountability, says, "If this is not provable misconduct before the Appeals Board, there never will be."
Imanishi-Kari tells The Scientist that she expected favorable results based on the questions the panel was asking. "But I was scared. The government-OSI, ORI, Dingell-have been so terrible."
The coauthors and their supporters have maintained that O'Toole was merely promoting an alternative interpretation of the data, and that such was "the stuff of science," as Tufts' Wortis put it in a memo that summed up the findings from the Tufts inquiry. O'Toole says the dispute was about experiments and data that were misrepresented, and never about differences in interpretation. In fact, O'Toole feels her reputation is damaged by the notion "that I created a big fuss because I disagreed in matters of interpretation. I was saying at the time that the paper was riddled with errors."
In a June 6, 1986, memo to a senior MIT researcher outlining her position, O'Toole pointed out what she believes to be "serious weaknesses in the data" presented in the paper, including evidence that she contends showed that a reagent did not work as claimed and that a described experiment was not done.
O'Toole remains bitter toward Baltimore, but feels some relief regarding Imanishi-Kari. "I was never happy with [Imanishi-Kari's punishment]," she says. "If she never has to face censorship or punishment, that's okay with me. I never thought the focus should be, 'What do we do with Thereza now?' It should be, 'What are the standards of the profession, and what should we do to uphold the principles?'"
Despite clearing Imanishi-Kari, the appeals panel found the Cell paper "rife with errors of all sorts." Then it went ahead to declare that O'Toole must share "responsibility for the pattern of carelessness in writing and editing of this paper," even though she was not a coauthor and had merely read and commented on a draft. The panel also doubted her version of events, writing, "After hearing Dr. O'Toole and the other witnesses testify and examining all of her statements over the years, we question the accuracy of Dr. O'Toole's memory and her increasing commitment to a partisan stand."
"People came to believe me because they examined the evidence," she says. "The Departmental Appeals Board had to find me not credible when they threw out all the evidence. They reduced this case to my word against all these other people."
The guts of ORI's case did not lie in O'Toole's testimony but in the evidence developed over several years by investigators from ORI, OSI, Dingell's subcommittee, and the Secret Service. As many commentators have pointed out, ORI had no smoking gun. Hadley, who worked on the case at both OSI and for the Dingell subcommittee, says the "coincidence" of statistical, forensic, and scientific evidence made the case against Imanishi-Kari compelling.
The board, however, was not convinced by any of ORI's arguments. Where ORI said that handwritten counts of background radiation in Imanishi-Kari's notebook were fabricated because they deviated from randomness expected with actual counts, her legal team, aided by University of California, Berkeley, statistician Terence Speed, provided an innocent explanation: The counts were obviously rounded, not transcribed precisely from radiation counter tapes. The board agreed, saying, "Dr. Imanishi-Kari was aberrant in her data recording and rounding patterns."
The Secret Service analyzed many of the paper data tapes coming from the radiation counters, which Imanishi-Kari attached to her notebook pages. The service examined the color of the tapes, and the ink and type font of the data recorded on them, in an attempt to determine when a tape was made or to establish that a tape couldn't have been made when claimed. For example, some important tapes containing data supportive of Table 2 in the paper were made, Imanishi-Kari claimed, in June 1985. The Secret Service could find no matches in other researchers' notebooks for these tapes around the date they were said to have been produced. ORI claimed they were made before 1984, and possibly as early as 1981, suggesting that Imanishi-Kari had used data from an old, unrelated experiment to fabricate this so-called June subcloning data.
The appeals panel called ORI's arguments "less than persuasive." The panel also could envision no scientific reason for Imanishi-Kari to create data that are not uniformly supportive of her other results. "We find that the inexplicable nature of the results is, if anything, evidence that it is unlikely that Dr. Imanishi-Kari would fabricate such results," they wrote.
MODIFICATIONS: Attorney Barbara Mishkin notes changes in ORI procedures.
CALL FOR CROSS-EXAMINATION: Imanishi-Kari's lawyer, Joseph Onek, says the system is flawed.
The apparent lack of due process afforded Imanishi-Kari troubled many observers of the case. Donald Kennedy, biologist and former president of Stanford University, wrote in Nature Medicine (2:843-4, 1996) that "Imanishi-Kari had no opportunity to cross-examine witnesses or even to hear specifics about the charges against her or to examine evidence." As Stanford's president, Kennedy himself had a run-in with Dingell's subcommittee in a dispute involving indirect costs (J. Mervis, The Scientist, Aug. 19, 1991, page 3).
Kennedy's assessment is not entirely accurate, according to sources close to the ORI investigation. The bulk of ORI's case, including the statistical analyses, is contained in the draft report that OSI issued in 1991. The forensic evidence was sent to Imanishi-Kari and her attorneys in 1992, and she hired her own expert to examine it.
Even earlier, Imanishi-Kari had a chance to discuss issues that were troubling OSI staff and the five-member scientific panel looking into potential misconduct. At the time, says biochemist William McClure, who was one of the panelists, they did not view the process as a legal one. They were trying to find out what happened. "Five scientists on the panel and four or five OSI staff sat around a table discussing the case," says the Carnegie Mellon University professor of biological sciences. "We tried to find out what was going on. We argued both sides. [The panel realized] we have got to get Imanishi-Kari here, rather than speculate."
Before she answered questions for the panel, she was informed of the questions the scientists had, such as: Were particular pages from her notebook a record of an actual experiment? Were data penciled in from memory or while she looked at the data tape?
Onek maintains that the key element of due process was missing: the chance to cross-examine the experts who developed the evidence. "As soon as my colleagues had an opportunity to question the Secret Service examiners [in front of the appeals panel], all the problems in their case became apparent," he says.
Many observers and participants in the case say that it demonstrates the tension between scientific and legal approaches to a conflict. In the beginning, comments Hadley, "there was a divergence of purposes." The scientists working for OSI and ORI, she asserts, mainly wanted to uncover every detail of what happened. The lawyers' job was to protect their clients. "I appreciate that that is what lawyers are supposed to do," she says, "but scientists are supposed to do something different. The challenge is to see how much we can do both-find out what happened and not destroy someone in the process."
As O'Toole sees it, scientists have a professional obligation to explain their work when questioned, whereas lawyers emphasize the right to remain silent. On the advice of her lawyers, Imanishi-Kari frequently did decline to answer questions or index her notebooks, as requested by investigators. But at the same time, investigators declined to divulge much more than an outline of the developing case against her until the OSI draft report was made public in 1991. "By the time they get to the draft report stage," says Mishkin, "they've already made up their minds."
Ironically, Baltimore had argued strenuously that scientists be judged by their peers, but his colleague owes her exoneration to lawyers. A likely outcome of the case is more, rather than less, involvement of lawyers in scientific misconduct proceedings.
Can scientists not police themselves? Marcel C. LaFollette, a research professor of science and technology policy at George Washington University, is optimistic. "Scientists should police themselves-they know the science best. But they can't do it alone. They don't work their way through the funding process alone, either . . . so it is not unusual for them to seek advice or help from professionals in misconduct cases."
For his part, NIH's Stewart says, "There isn't any alternative" but for scientists to police themselves. "Places like ORI exist because scientists have not taken enough responsibility. Scientists have got to care about making their profession work."
Billy Goodman, a freelance science writer based in Montclair, N.J., is online at email@example.com.