‘Each scientist must stand up, at all costs, for the truth’

It was possible when I was a graduate student in the mid 1980s to have read every article in my field of study, bacterial chemotaxis. I did so, and I was known among my colleagues at the University of California, Berkeley for having achieved a feat that would be impossible now.

At a Gordon conference I attended on sensory transduction in microorganisms, some graduate students, postdoctoral researchers and one or two professors were discussing how the signal created when a bacterium responds to the concentration of a chemical in its environment is conveyed. Despite being rather a junior scientist, I was not shy and, in response to a statement made by one of the others, I interjected: “But what about this paper (I named the authors)?” The answer I received was striking: “Everyone knows that paper is wrong.”

I was stunned. Having confirmed with more senior researchers that this was indeed the universal view, and having learned the reasons, I enquired whether the publication had been retracted or corrected, or whether any subsequent article had stated that the data were incorrect. The answer was no.

Let us acknowledge that I was a naive graduate student. My question was: “How is it possible for such a state of affairs to exist?” The scientific literature absolutely needed to be as accurate a reflection of current knowledge as possible.

It was asserted to me that corrections to the literature were rare, and that scientists were often reluctant to indicate the errors of others in print. However, questions remained. Who was everyone in “everyone knows that paper is wrong?” I didn’t know prior to attending the meeting, and I was in the field. What about those who were not and who would be misled or would incorrectly cite that erroneous article? Of the published and the word-of-mouth scientific literatures, which was the real corpus and which was the shadow? Was retraction reserved for articles in which there was misconduct, or was it also appropriate if results in the article were wrong?

Allow me to begin my story again, further in the past. When I was a student at a high school in Riverdale, New York City, I had an English teacher who was surrounded by a cult of personality. He was insanely egomaniacal, egregiously demeaning to students, and an exhibitor of shameless favouritism towards particular minions. I complained to the head of the upper school about his insulting and bigoted pronouncements and the fact that he evidently graded assignments without having read them, but the administrator excused his behaviour on both counts.

I was eventually able to transfer out of the class, but it has subsequently emerged that I was but one member of a large exodus over the course of the teacher’s career – in a school that saw multiple accusations of sexual abuse by other teachers, too, including of some of my classmates. Again, I admit to being a naive student; I did not know about the abuse at the time and could not have imagined it.

But I learned from this incident two things. First, that administrators and institutions often fail to take action when misconduct is occurring and thereby bear responsibility for its perpetuation. And, second, that individuals engaged in one form of unacceptable behaviour frequently display others.

Experience continued its sequence of lessons. As a visiting scientist at the University of California, San Francisco, I was part of a group working on what eventually became two highly cited review articles for Nature. I tried to repeat my earlier bibliographic feat by reading the entire corpus of articles written about a group of important molecules now called GTPases (a term my coauthors and I popularised). It was a daunting task, but one that I came close to accomplishing; it was probably the last moment that the attempt could even be made because the number of articles on the topic subsequently exploded.

My close reading led to the discovery that a prolific author in the field appeared to be writing substantially the same article over and over again, with just the protein names or protein tissue source being altered. Textual overlap is permitted between the methods sections of different research articles, but the overlap in these articles extended throughout. Two highly similar articles would appear in two different journals – almost always the same two – with only a short interim between the publication dates.

One time, I actually received a manuscript from the prolific author to review and I recognised the writing and some of the data as having appeared in an article in the other journal. My colleagues and I communicated our concerns to the editor of that journal and the editor contacted the author. However, the author proclaimed innocence and victimisation and the editor let him off with a warning.

During my time as a postdoctoral researcher at the Whitehead Institute for Biomedical Research (affiliated with the Massachusetts Institute of Technology) there was periodic discussion of the misconduct allegations – ultimately dismissed – against Thereza Imanishi-Kari regarding a 1986 paper she co-authored with David Baltimore; the 1975 Nobel prizewinner ran the laboratory adjacent to the one where I worked. Then there was the MacArthur (“genius grant”) fellow who had erroneously claimed to have discovered the “cancer metastasis gene” when he worked in the laboratory of Robert Weinberg, also on the same floor of the Whitehead as ours. The careerist tendencies of some of my fellow scientists, as well as the old boy network that too hastily elevates some while unjustly holding back others, were on dramatically clear display there.

Richard Mulligan, my postdoctoral adviser, was a pioneer in the use of viruses for the introduction of genetic material into cells for gene therapy. My initial project was to design engineered retroviruses capable of targeting particular cells. Eventually, I realised that our understanding of the normal process of retroviral entry into cells was insufficient, so I focused on reducing that deficiency.

Defective viruses that could enter a cell and convey genetic material but not replicate were the main experimental tool. During my studies, it became evident that there was a problem in multiple laboratories where cells that purportedly were incapable of generating retroviruses that could reproduce were, in fact, producing such potentially hazardous viruses. This finding, I can assure you, did not make certain people very happy, and efforts were made to keep it under wraps.

Retrovirus-related materials are an important component of any mammalian genome. Yet there are different retroviruses in different orders of mammals. The defective viruses that were being used were mouse viruses and were being produced in mouse cells. One potential source of problems was the combining of the defective mouse virus with material already in the mouse cell that could result in a replicating retrovirus. A proposed solution was to engineer a human cell to produce the defective mouse virus instead; an individual in another laboratory, with extensive assistance from us, designed one and published it.

Many of us recognised the utility of these cells and, after some delay, our laboratory received them. Soon, a colleague and I recognised that the cell line had been made incorrectly (one of the nucleic acids introduced was not the one described in the materials and methods section of the article) and that transmissible mouse viral material had been introduced. We notified the authors and suggested that they correct the publication. Instead, they decided to issue a private letter to people who had requested the cell line.

At a Cold Spring Harbor retrovirus meeting, a graduate student gave a presentation on his work studying the transmission of a different mouse retrovirus (we’ll call it Virus A) using the human cell line. This mechanism was perfect, it seemed, for the investigations because it lacked any other potentially interfering mouse virus genetic material (such as from Virus B). Astonishingly, however, the student reported that his analysis of genetic material being transferred by the cell had detected material from Virus B.

This, of course, was the virus material introduced into the cell line by the publication authors. But the student had no way of knowing that, and he appealed to his audience for suggestions. I approached him privately after the talk and explained to him the origin of his findings. He thanked me profusely; he insisted that I had saved him from additional years of fruitless work having already wasted substantial time on the project.

But there may well have been others who were not so lucky. The erroneous article, without any correction ever having been made, has been cited nearly 3,000 times.

Being in the field of gene therapy was not conducive to continued confidence concerning scientific integrity. It was a running source of amusement in the Mulligan laboratory that the diffuse staining of an arterial wall depicted in another lab’s “classic” article that was supposed to be the result of gene transfer was nothing more than background staining. And besides the tragic, well-publicised death of Jesse Gelsinger in 1999 in an early gene-therapy experiment, there were other premature human investigations, as well as animal data that were ignored.

In my studies of the proteins that were responsible for the entry of the modified retroviruses into cells, I encountered some erroneous published data. I would contact the authors, including an explanation of how the mistakes almost certainly arose, but could virtually never get them to admit the existence of a problem, never mind correct the literature.

In one example, I contacted a laboratory director whom I knew personally soon after he published an article that was certainly erroneous and whose conclusion had direct impact on my own research. I shared my analysis and asked him to send me the materials that he claimed had been generated, so that I could test them directly myself.

First, he told me that those materials had been lost in a recent move; he would have someone in the laboratory recreate them. A while later, he said that an error had been introduced into the materials when they were making them again. Finally, he sent me nothing. It may be worth remarking that the scientist was later convicted of sexual abuse of a child under the age of 14 (see the lesson learned from my high-school experience).

In another troubling case, our laboratory and that of a collaborator together submitted a manuscript on our research on the structure of an enzyme to a major scientific journal (Journal A). After some delay, we received a rejection notice. Peculiarly, on the same day, an article on the structure of a closely related protein was published as an “accelerated” publication (meaning there was a shorter than usual period between its submission and publication) in Journal B. I noticed the coincidence at the time but was merely bemused by it.

After having been altered in response to reviewers’ comments, our article was accepted at another respected journal (Journal C). Subsequently, the group associated with the accelerated publication produced another one (in Journal D). This cited a hypothesis – with a reference to our article in Journal C – that had been present in the manuscript submitted to Journal A but not in the one submitted to Journal C (nor, therefore, in the published article).

A reasonable conclusion was that the group had had access to the manuscript submitted to Journal A, rushed their article into Journal B, and had relied on their memory of our manuscript – rather than on a reading of our published article in Journal C – when they wrote their Journal D article. That meant there had been a violation of the confidentiality of peer review, which is a form of research misconduct.

I contacted the editor of Journal D, and he agreed that there was something very suspicious about the events. When we contacted an editor of Journal A, he conducted a very cursory investigation and relied on the statements of the authors that they had done nothing wrong. They had found a meeting abstract that had referred to our discarded hypothesis, they explained, and relied on it to justify their repeated citation of an article that never mentioned that hypothesis.

Source: 

Getty montage

My watershed moment did not come until a few years later, however.

While travelling to my office at Purdue University in late 2010, I heard a story on National Public Radio about a bacterium that was reported to use arsenate instead of phosphate as a nutrient and in whose cellular DNA the arsenic had replaced some of the phosphorus. Having studied enzymatic phosphoryl transfer as a graduate student and faculty member, I knew that the claimed results were impossible. Biology may teach us about novel chemistry, but biology doesn’t violate the laws of chemistry.

My analysis of the data in the paper and, especially, the supplementary material indicated that the authors should have been aware of the invalidity of their claims, and that their results were the result of chemical contamination. The article – published in Science – was nevertheless covered by the international media and was the subject of a press conference by Nasa (which had partly funded the research). The agency’s invitation promised “an astrobiology finding that will impact the search for evidence of extraterrestrial life”.

Other scientists also rapidly found and publicised numerous flaws in the article, and Science published eight critical technical comments on the article and two refutations of its conclusions. However, it has never been retracted and is still cited, sometimes as if it possessed some worthwhile content. Neither Nasa, the authors nor the reporters who touted the research or downplayed its failings have ever made a public acknowledgement of their errors.

This was despite my best efforts. I was interviewed by the media, contributed commentary on blogs and exchanged electronic messages with the authors, the editor-in-chief of Science and the reporters. Because of my prominence in the case, I was invited to give seminars about it around the world. This incident was a paradigm of the forces that were distorting the scientific enterprise, and I would both describe the trouble and propose solutions.

These lectures consistently generated the most extended discussion of any that I have delivered. Most interesting was the frequency with which people would approach me afterwards and share their experience: “If you think that’s bad, you should take a look at this!” They would ask me to advise them or to investigate other violations of scientific norms. I also received tips electronically. It was the beginning of a whole new research programme for me.

I was not the first and nor am I the most prolific among investigators of research improprieties. A number of the fellowship are more highly skilled than I am, and one positive development is that their number has grown in the past few years; some have actually uncovered whole networks of rogues. Nevertheless, there are comparatively few who both conduct primary inquiries into the scientific literature and are willing to do so using their own names. Also, not many of the detectives uncover both data misrepresentation and plagiarism, as I do.

Each time I received some intelligence, I reviewed the article(s) in question and decided whether the matter should be pursued further. If there were serious deviations from scientific standards, I would proceed to scrutinise other articles by the same authors. My experience, which I believe is shared by other members of the investigatory fellowship, is that people who are less than scrupulous in one publication are frequently found to be devious in others. Also, although many individuals repeatedly commit a particular infraction, it is not uncommon for them to cut multiple corners.

Generally, I do not know the authors personally, so if there are major issues, I notify the journals in question. Initially, however, I encountered extreme reluctance by editors to take any action. It was comparatively rare for authors to acknowledge even obviously problematic data or text, and editors appeared happy to take their word for it; a co-author and I have previously itemised the illogical excuses proffered by authors and accepted by editors. In some cases, action was only taken when members of the media expressed interest.

With time, however, some journals recognised that there are real questions of integrity and credibility at stake, so they changed their policies to address the concerns. I have personally presented analyses to journals that have led to dozens of corrections or retractions. Persistence succeeds.

Most journals, however, continue in the tradition of ignoring and delaying appropriate remedies. Moreover, amendment of the scientific literature is virtually the only reward for the many hours that my investigations take up. There have certainly been no benefits to the advancement of my career. Moreover, some academics resent my activities and have engaged in tactics directed at undermining them, including via the courts.

The retribution coming from those who flout standards of behaviour – and their allies – could be more readily avoided if I reported my findings of research transgressions anonymously. I certainly understand why some tipsters want to keep their identities concealed. However, the more people that commit to critiquing the scientific literature’s failures openly, the less likely it is that any individual can be successfully targeted. In addition, anonymity contravenes the principles of the modern scientific endeavour.

Most scientists I know are dedicated to honesty in conducting and reporting research. Science continues to be the best basis for addressing the challenges that face society. But science faces assaults from those who oppose it for ideological reasons, and some colleagues are afraid that exposing wrongdoing by researchers is strengthening the enemies of science. Those who do not adhere to scientific norms, however, are the ones who undermine the trust of the public. They arrogate resources that are thereby denied to upright researchers. They also set a terrible example for young scientists.

The research community must not add to the public’s grounds for mistrust by rewarding those who cheat. Nor should anyone turn a blind eye to misconduct. Perhaps I am still naive, but it seems to me that even in a post-truth society, it is an essential duty of each scientist to stand up, at all costs, for the truth.

Science must clean its own house. And those who help it do so should be commended, not condemned.

David A. Sanders is associate professor in the department of biological sciences at Purdue University.