MD Anderson postdoc faked results of Novartis anti-cancer compound study

jun fu
Raymond Sawaya, director of MD Anderson’s brain tumor program, presents Jun Fu with the 2014 Caroline Ross Endowment Fellowship.

A former postdoc at MD Anderson Cancer Center faked the results of a mouse study of a Novartis compound designed to fight brain tumors, according to the Office of Research Integrity (ORI).

Jun Fu “admitted to knowingly and intentionally falsifying Figure 8a” in “Novel HSP90 Inhibitor NVP-HSP990 Targets Cell-Cycle Regulators to Ablate Olig2-Positive Glioma Tumor–Initiating Cells,” a paper published in Cancer Research on May 15, 2013:

Specifically, the Respondent falsified survival times of mice to show that NVP-HSP990 prolonged survival rates in glioblastoma tumor bearing mice when experimental data were incomplete and unusable.

Fu agreed to have his research supervised for two years, and that he would not serve on any NIH peer review committees for the same period of time.

It’s unclear what will happen to the paper in question, which has been cited twice, according to Thomson Scientific’s Web of Knowledge. MD Anderson, according to the ORI,

recommended that the senior author of this paper take any appropriate steps with the journal to correct the scientific literature.

Here’s more information on NVP-HSP990, which has been tested in early-stage clinical trials. One phase I study was presented at the American Society of Clinical Oncology meeting last year.

The case seems to have moved quickly. Just three months ago, MD Anderson announced that Fu was one of the winners of the 2014 Caroline Ross Endowment Fellowship.

Update, 12:15 pm Eastern, 8/1/14: Al Yung, the corresponding author of the Cancer Research paper, confirmed for us that Fu is no longer employed by MD Anderson, and that the team plans to retract the paper.

37 thoughts on “MD Anderson postdoc faked results of Novartis anti-cancer compound study”

  1. Did the gentleman have to give back his prestigious “Caroline Ross Endowment Fellowship” back, or is he still fully supported by it, while being supervised for the mere two years?

    1. This whole idea of ridiculously soft punishments makes one wonder: is it a method to stop delinquents from spilling the beans on who else was involved? Is it then a surprise that a PI is usually an innocent victim of a devious postdoc?

      1. Who else was involved, is a pertinent question.Inadequate and/or flawed supervision was found to have played a role in a significant number of cases where a post doc was found guilty of misconduct.(1) This would be less likely to occur if the NIH teaching branch established reasonable student to mentor ratios and “flawed” supervisors/training programs were identified and, where indicated, individual mentors penalized.
        NIH should expect that the institution’s next training grant application would demonstrate that it has taken appropriate action to assure that this will not happen again.

        Don

        1. Kornfeld, DS , Acad. Med.2012,87:877-882

        1. Donald, I can’t trace that paper of yours, but it is important, I feel, if there is a way of quantifying PD vs PI “flaws”. Can you please provide the title and the web-link. Thanks (hoping there is no paywall).

          1. Thanks.
            ‘Kornfeld,DS, Misconduct in Research: The Search for a Remedy., Academic Medicine, 2012,87(7) 877-882.

            Re PD v PI “flaws”, It’s been awhile since I’ve worked with the data. I would estimate 10% really flawed ;
            mentors,25% lazy ones.

            Don 1.

  2. Has someone checked the other papers of this author? Fraudulent practices tend to become the habit of those who engage in them.

    1. Paper cited twice so far. Citation does not mean ‘reproduced’. It could be in the introduction (background research), methodology (“carried out as described in ..”) or discussion. IMO treating citation of a paper as a measure of its reproducibility is an entirely incorrect notion

      1. Indeed. Don’t judge automatically those poor slobs who thought this would be just perfect to buff the introduction section.

  3. Another ‘supervised’ researcher being punished to work under ‘supervision’!! When will this mockery stop?

    ORI also left the fate of the paper in the hands of senior co-auhtor who is very likely to come out with “figures were falsified, data was manipulated but none of these affect the conclusions of the paper”

    1. Yeah, amazing. I am just finishing reviews for a federal health care agency. 8 grants, about 32 hours for the reviews, I get $400 for the reviews and $400 for the in-person meetings. It’s above minimum wage, but not going to fund my retirement on it.

    2. The federal “administrative actions” as imposed by ORI, are said NOT to be “punishment” or “sanctions” – instead, their intent is to protect the Federal Government and taxpayers’ funds. ORI always imposed this non-service action, to prevent the dishonest researcher from reviewing or influencing the reviewer of other research by other (honest) persons.

      1. Alan,

        I understand that ORI’s administrative actions are limited,and your explanation of the restriction on participation in review committees clarifies the purpose of the imposed sanctions.

        ORI is also permitted to forward more egregious cases for prosecution by the Justice Dept.
        What is unclear is why such prosecutions are so rare.?
        We are all aware of individuals who have been found by ORI to have used significant federal funds to produce fraudulent research and have not been prosecuted. Apparently,,as demonstrated in a recent case,at Iowa State University, it takes the intervention of a senator legislator,such as Senator Grassley, for such action to be taken. Why is that so ?

        Don

        ,

        1. Yes, Don.

          ORI always discussed its cases with the HHS OIG, which could refer them to the DoJ – and sometimes DoJ expressed interest in a case – but in my experience, because of the high cost to DoJ to pursue a criminal-law investigation and prosecution, the AUSAs wanted to be sure that DoJ could recover a million dollars. Prosecuting, for example, a low level foreign-born researcher working in this country — whose work may well be only a small of the research grant and the paper in question — is not going to lead to such a recovery.

          Thus, the ORI-imposed administrative-law actions (usually debarment from receiving federal funds for 3 years or more) is left. However, the institution or state may seek recover of funds directly from the perpetrator, or pursue other civil or criminal actions, at the discretion of the local or state prosecutor.

          1. Alan,

            I do understand that rationale for the scenario you describe. However,t I have the impression that US citizen-researchers who have received multimillion dollar grants have also not been prosecuted.
            In the history of ORI (25+ years ) perhaps only 3 such prosecutions.

            Don

          2. Yes, Don, but just because a PI has multi-million dollar grants, the recovery of funds sought by NIH and DoJ is based on the fraction of the award that was impacted by the falsification or fabrication or other fraud – it is not the whole grant. [It is noteworthy that in qui tam (False Claims Act) suit by relators, that may be taken over by DoJ, there is the possibility to recover triple-damages, so DoJ pursues them under threat of getting three times the amount of the grant back for the Government.]

            Yes, there have been DoJ prosections based on ORI misconduct cases:

            In 2005, Dr. Eric Poehlman, Professor at the University of Vermont, formerly at the University of Maryland at Baltimore, falsified massive human subject physiology testing. DoJ imposed a 1 year prison term plus 2 years probation for lying in an NIH grant application, with a $180,000 civil fine — and ORI debarred him for life:
            http://www.nytimes.com/2006/10/22/magazine/22sciencefraud.html?_r=1&pagewanted=6
            http://edocket.access.gpo.gov/2005/pdf/05-5876.pdf

            In 2006, Mr. Paul Kornak, Clinical Research Coordinator at the Stratton New York Veterans Administration Medical Center, falsely claimed to be an M.D. and falsified clinical trial data, including that to enroll an ineligible veteran patient, who died from the treatment? DoJ found negligent homocide with a 6 year prison term — and ORI debarred him for life.
            http://web.archive.org/web/20060203101242/http://www.usdoj.gov/usao/nyn/NewsReleases/2005/2005-02/200502031553.htm
            http://www.gpo.gov/fdsys/pkg/FR-2006-02-24/html/E6-2667.htm

            IN 2010, Dr. Elizabeth Goodwin, professor at the University of Wisconsin, falsified data in multiple publications, She was prosecuted by DoJ, and the Federal District Count in Wisconsin required her to pay a $500 fine plus $50,000 in restitution to the University scholarship fund and to the Federal Government.
            http://www.justice.gov/usao/wiw/Press/June%2025,%202010%20%20Elizabeth%20Goodwin%20Pleads%20Guilty.pdf
            http://www.gpo.gov/fdsys/pkg/FR-2010-08-25/pdf/2010-21048.pdf

            – – – –

            There have also been local or state prosecutions of respondents in ORI cases:

            In 2010, Dr. Vipul Bhrigu, postdoctoral fellow at the University of Michigan Medical School, who had been found to have tampered with research materials of others, causing false data to be generated in the laboratory, was locally prosecuted in Michigan District Court and pleaded guilty to misdemeanor charges for malicious destruction of property. He received 6 months probation and nearly $10,000 of fines and costs — and in 2011 ORI debarred him from federal funding for for three years.
            http://classic.the-scientist.com/news/display/58157/

            In 2005, Mr. Randall Luce, technician at the University of Buffalo, fabricated subject interviews. He was locally prosecuted in New York and pleaded guilty to grand larceny, admitting to misappropriation of funds for the research interviews, for which he was fined — and ORI debarred him for three years.
            http://www.gpo.gov/fdsys/pkg/FR-2005-08-10/html/05-15777.htm

            In 2004, Pat Palmer, assistant research scientist at the University of Iowa, fabricated autism family interviews, falsified travel reports, and falsified B.S., M.S. and Ph.D. degrees to get a higher salary. She was prosecuted by the State , and she pleaded guilty to first degree theft and falsification of degrees and paid a $1,000 fine and nearly $19,000 in restitution to the University — and ORI debarred her for three years.
            http://www.news-releases.uiowa.edu/2003/december/hitsdecember03.html
            http://www.gpo.gov/fdsys/pkg/FR-2004-02-17/html/04-3336.htm

            – – – – – –

            Thus, it is up to the prosecutor – whether federal DoJ, state, or local — to pursue criminal cases against persons whom ORI finds to have committed research misconduct, with its findings published in the Federal Register of the United States (as well as the NIH Guide to Grants and Contracts, ORI website, and ORI Annual Reports).

  4. He should lose his Fellowship and the post should be re-advertised so that it can go to a more deserving scientist.

    Further, if the whistleblower (or vigilant reader) responsible for making the errors known, chooses to apply for the the Fellowship, they should automatically get it.

    1. “if the whistleblower (or vigilant reader) responsible for making the errors known, chooses to apply for the the Fellowship, they should automatically get it.”

      I would revisit that concept. Being a whistleblower does not guarantee anything – he/she can be a saint, a competitor, or just another fraudster – we should not make reporting fraud personally profitable, or things will likely to get really nasty!

  5. There are 47 data panels (plus a Table) in this article, including Supplementary, but as per the ORI report the falsification affects only one. Why on earth would you do that? I can’t help but wonder in cases like these whether they are ‘reviewer experiments’.

    The fact is, when you start a project it doesn’t really matter what the results are. If they’re negative you either abandon or you finish it properly accepting it will be a ‘null results article’, but at least with no incentive to make anything up. But when the data looks good, the stakes are higher because you could end up with a high IF publication at the end, and so the incentive to keep finding the ‘right’ results increases. This reaches its absolute maximum once you’ve been offered a Revision in a good journal, if only you can satisfy the damned reviewers. If they’ve recommended experiments, then these experiments HAVE to work – there can only be one outcome, otherwise your paper (and associated prestige) is doomed.

    Combine this perverse pressure with the fact that the original first author may have left the lab by the time of revisions, and you end up with contrived / poorly performed / analysed with bias / occasionally outright manipulated experiments which will always end up ‘proving’ what the reviewer has asked for. They end up undermining what is probably otherwise a good paper with genuine and interesting results.

    Personally I think reviewer-suggested experiments shouldn’t be allowed: if the ‘missing experiment’ is absolutely critical then reject the paper, otherwise review the manuscript as a body of work in its own right. Yes there are more experiments which could be done, there always are – perhaps you could even do them yourself.

    I note, by the way, that it took 8 months to submit the revision…

    1. Well, it seems you’re correct in calling out the faking of one in 47 images as anomalous. The following would suggest the ORI and MDA oversight bodies didn’t pay much attention….

      1) Figure 3C, 50nM treatment with drug, same blot is duplicated in supplemental Figure 2B with 20nM of drug.

      2) Figure 2B, b-actin loading controls in both panels are the same (vertically re-sized) as the loading controls in supplemental Figure 1B. Problem is, the listing of different cell lines along the top of these panels is different between the figures.

      3) Figure 6C, in the p-Rb panel there’s a sharp horizontal cut off immediately below the band of interest, where it seems as if someone has hosed out the background maybe to hide an undesirable extra band?)

      4) b-actin loading contrl in Figure 5B left panel is identical to the one in Figure 7D, for two completely unrelated experimental conditions (treatment with different drugs).

      This may be one of those rare cases where perhaps ORI should have spent more time investigating. By my reckoning, if the faking now extents to 5 instances instead of one, then the sentence should be multiplied 5 times – a 10 year ban and supervised research seems about right!

        1. Better yet, scrap ORI, which wastes US tax-payer’s money, and donate a fraction of that money to PubPeer to fortify its anonimity and feedback by scientists. Scientists that report fraud and errors have one thing that is rapidly starting to be lost in science and science publishing: a conscience.

      1. That is quite a list, maybe it could be hypothetically be the case that they went for the mouse work as wrong doing is easier to prove in a different way. To examine mouse work you would also first go to the records of the mouse unit (where more data is kept independently from the PI or postdoc), if it does not match up with what was claimed there is definitely a serious problem. With gels, even if it looks as questionable as stated then you would still have to go through the hoops of trying to find the original gel pictures etc when they have probably been misplaced, eaten by the lab technician etc.

  6. I wonder how much follow-on research by Novartis was conducted based on this falsified data. While the public sees pharma as having limitless $$$, the reality is they have to prioritize. False data can result in them wasting money on a cancer treatment that actually provides no benefit to patients, while not spending it on other possibly effective treatments.

    According to 4 July Science, a Novartis employee in Japan was actually arrested for falsifying data in an unrelated trial. It sounds like Novartis will need their own internal anti-fraud unit.

    1. But what’s the point of it? (I am truly curious) In academia manipulated/fabricated articles usually cover a very narrow area of a complex research field, tend to be very descriptive and displayed as being well-supported my earlier evidence. Methodological novelties or anything that might facilitate replicate experiments are usually carefully avoided.

      In contrast in drug development sooner or later (but predictably) the ugly truth will surface, the fact that you elongated the survival curve of the lab rats today won’t protect your drug from failing to produce any effect tomorrow in an other lab, or (worst case scenario) in the clinical phase.

      1. @BB – Regarding drug development vis a vis “the ugly truth will surface”, there is of course the quite recent, very well known, and in some quarters highly embarrassing case of Iniparib. This Sanofi-Aventis proto-drug is now a chemical formerly known as a “PARP inhibitor”

        http://en.wikipedia.org/wiki/Iniparib

        Sadly of course, it never, ever, was. Goodness knows what skullduggery went on behind the scenes in that drug company – and it will surely never suffer the oxygen of publicity, no matter how much it deserves it. Does the company even know what went on in its secretive labs? Was anybody fired? The whole PARP inhibitor field was massively fucked up by the disastrous trial results of Iniparib.

        Luckily, our champions in revealing the non-PARP inhibition are common or garden academics happy to publish in a regular cancer journal with their work subject to the much maligned peer review process

        http://www.ncbi.nlm.nih.gov/pubmed/22291137

        Legends!

        Actually, I don’t understand why you are trying to split motivation for fraud between research fields? Opportunity may differ, but people (on average) presumably remain the same. It’s a jungle out there. Nasty things are going on in all research fields. I believe that the current drive to get drug company research results published (given due caveats competitive reasons) will be extremely beneficial. I disagree that the current system leads to the exposure of drug fraudsters: in my view it acts to protect those working in drug companies relative to their colleagues in academia.

  7. BB – Exactly. It makes no sense to falsify data on a compound. It ends up causing the company to waste resources.

    To add to your example, falsified animal or in vitro data would lead to clinical studies that would ultimately fail – but fail at a far higher cost to everyone.

    Patients in initial clinical trials (Phase I) for anticancer compounds are typically those for whom all other treatments have failed. Hence, they would be hurt by being given an ineffective and/or dangerous drug.

    The company is hurt because clinical trials are far more costly than animal & in vitro studies. Also, it keeps them from spending resources on a better compound.

    It hurts the pharma industry in general because it promotes even less trust by the public.

    Hence, as you said “What’s the point?”

    1. Absolutely – I also agree with BB and DTX. Ultimately the drug company’s precious compound goes out into the big wide world where other people will quickly discover whether it works or not. So there is really no benefit (and potentially lots of harm – financially and reputationally) in deluding yourself that it works when it doesn’t.

      But there’s also another significant reason why pharma may be less prone to scientific fraud, and that’s career structure. Scientists in pharma companies are employees: they don’t have to compete for a salary every three years by showing how many dramatic ‘results’ they have found. They are appraised on whether they have performed good experiments with proper processes and documentation, not on the outcome. Similarly on the clinical trials side: you get your bonus if you got all your regulatory paperwork in on time and kept protocol deviations to a minimum, regardless of whether or not the drug worked.

      For all their problems on the commercial side, I actually think we could learn a lot on the science side from pharma!

  8. These are not even glioblastoma cells, his mice are dying in 30-40 days, whereas if you do an intracranial mouse model with authentic primary glioblastoma cells you don’t get many death for months.

  9. In Response to DocMartyn RE survival
    Fu was a 2nd author of another paper (Jan 2014) from the same group
    http://www.ncbi.nlm.nih.gov/pubmed/24038660
    In this they were lookin at another compound ‘RO4929097’ (gamma-secretase inhibitor)
    In the survival using a similar intracranial model control survival was 50-125 days

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.