Retraction Watch

Tracking retractions as a window into the scientific process

Glaxo asks Nature Medicine to retract paper by fired company scientist

with 18 comments

natmedcoverIn what could be a significant blow to a major pharmaceutical company, Nature Medicine is reportedly set to retract a 2010 article by a group of researchers affiliated with a Chinese arm of the drug giant GlaxoSmithKline.

We’re not the first to report the news — you can read coverage of it on In the Pipeline and Pharmalot, for starters — which includes the revelation that Glaxo has fired Jingwu Zang, a co-author of the suspect paper and former senior vice president and head of research and development at the Shanghai facility: in other words, a big fish. (Big enough to have a profile in, well, Nature Reviews Drug Discovery.)

Pharmalot has quoted a Glaxo spokeswoman:

“We’ve now established that certain data in the paper were, indeed, misrepresented. We’ve shared our conclusion that the paper should be retracted and are in the process of asking all the authors to sign a statement to that effect, which is the procedure the journal requires,” the spokeswoman writes us, adding the drugmaker “is committed to the highest ethical and scientific standards… in this instance, our standards were compromised.”

The site also notes that Zang received a warning from the FDA in 1999

for administering experimental treatments to patients without filing an IND, or Independent New Drug application (here is the letter).

The paper, “Crucial role of interleukin-7 in T helper type 17 survival and expansion in autoimmune disease,”

has implications in the treatment of autoimmune disease…

IL-7 has been area of interest for Glaxo until this week, when, according to the Pink Sheet, it suspended its research program in this target. (Could that mean more studies, perhaps unpublished, are tainted, or is this a precautionary measure?)

The paper has been cited 66 times, according to Thomson Scientific’s Web of Knowledge.

Glaxo launched the Shanghai center in 2005, with Zang at the helm, becoming, according to this Taiwainese report:

the first foreign-invested pharmaceutical company to carry out research and development in traditional Chinese medicine, Guangzhou’s 21st Century Economic Report says. …

Zang said GSK had provided its China center with plenty of resources and the power to decide on the direction of neuroscience research at the company. “China has made great progress in neuroscience research in the past 10 years; GSK hopes to catch up with its progress in this field,” Zang added.

We’re often asked whether drug company research is more or less likely to be retracted than work done in academia. We don’t have any data on that, although some limited studies have looked at this issue. We’ve reported on several retractions of Pfizer studies for serious errors, on a retraction for “unsolved legal reasons” of work by Bristol -Myers Squibb researchers, and on corrections to figures that included a senior Merck scientist as an author.

Update, 7 p.m. Eastern, 6/16/13: Zang told the Wall Street Journal (via FierceBiotech and Andrew Mallon):

I take a certain responsibility. I’m not trying to say I’m free of any responsibility. But what I’m really angry about is, I was dragged into this so-called data fabrication, which I’ve never been involved in.

FierceBiotech notes:

Meantime, one of the other authors of the Nature paper in question has resigned from GSK and taken one for the team in the data fiasco. The researcher, Xuebin Liu, told the business newspaper that he made an honest mistake and never tried to dupe anyone.

Glaxo has also paused a phase I trial of the compound described in the Nature Medicine paper, FierceBiotech reports.

Comments
  • Hubert de Guise June 12, 2013 at 11:15 am

    As a comment to your last paragraph on the origins of likely retractions, please see (on a related but slightly different topic)
    Misconduct accounts for the majority of retracted scientific publications
    Ferric C. Fanga, R. Grant Steenc, and Arturo Casadevall,
    PNAS October 16, 2012 vol. 109 no. 42 17028-17033

  • Andrew June 12, 2013 at 11:29 am

    Kudos to GSK on their professional and unequivocal handling of this situation compared to what we normally see from academic institutions. In drug development, scientific misconduct has no place. There is no reward and severe consequences, GSK losses here will already be very substantial.

    Please have a look at these two links, they give a view upon this issue that is little seen here at RW, from our perspective in the biotech/pharma industry.
    “The unspoken rule is that at least 50% of the studies published even in top tier academic journals – Science, Nature, Cell, PNAS, etc… – can’t be repeated with the same conclusions by an industrial lab”.
    http://lifescivc.com/2011/03/academic-bias-biotech-failures/
    http://lifescivc.com/2012/09/scientific-reproducibility-begleys-six-rules/

    This has fundamentally damaged the progression of therapeutic solutions for patients and continues to do so apace. It is astonishing that this is being tolerated and inadvertently promoted as a pathway to academic success. But… how to fix it?

    • Anonymous crystallographer June 12, 2013 at 5:11 pm

      This is the same company that pled guilty to criminal charges and paid a $3 billion settlement just last year for illegally marketing psychiatric drugs for off-label uses. It’s nice to see them being responsible in retracting this paper, but I’d be a little hesitant before crowing about the superior ethics of the pharmaceutical industry.

      • Pharmapawn June 13, 2013 at 9:53 am

        As a bench scientist in the Pharmaceutical industry, I can assure you that many (probably most) Pharma discovery and development researchers share your disgust at the marketing abuses of our industry. I think it is important, however, to highlight that the basic and even translational research we perform is quite divorced from marketing and sales. In several years at my extremely large company, I have not encountered a single salesperson during the performance of my duties. I don’t think I’ve even received an email from the Sales side of our organization.

        I cannot comment on specifics, but my personal experience and that of close friends at other companies is that in addition to more careful oversight of clinical trial design, execution, and reporting, there are significant changes underway surrounding the oversight of discovery research. For example, I was very recently subjected to a pre-publication audit of my primary data and e-notebooks, and trained in expanded SOP’s for archiving and reporting data for publications. Admittedly, lessons learned from events like those described above are the main drivers, but I applaud this increased oversight.

        Keep an eye on this space, and don’t be surprised when the same level of scrutiny is applied by Pharma in our academic collaborations. Given the overall poor quality of both research and record keeping I’ve encountered in academic collaborations these past few years, I suspect that such increased vigilance will pay rapid dividends for the scientific enterprise as a whole.

  • alahmada June 12, 2013 at 12:37 pm

    Reblogged this on Honest Abe's Blog.

  • Junk Science June 12, 2013 at 2:22 pm

    In figure 2a, the medium and TGF-b plots are identical, so should it be 0.3 or 0.6?

    • Steven McKinney June 12, 2013 at 3:43 pm

      Right you are Junk Science, so it should be “retraction”.

      Authors do state that “Data are representative of three independent experiments” so if duplicate images are shown with differing numbers one can only imagine the kind of experiments that these “data” are representative of.

      With such careful attention to figures in the paper, one again can only marvel and wonder as to the careful attention they exercised in the lab and in the statistical analysis of the findings. Extra points for lack of ethics considering that mice were sacrificed.

      • Junk Science June 12, 2013 at 5:31 pm

        It probably goes way back in time. Looked at some paper while he was at Baylor and in this article (J Immunol. 1999 Dec 15;163(12):6393-402. PMID: 10586029) you will find the old flip-job. Flip the Lck blot of Fig. 3A 180 degrees horizontally and you wlll get the Lck blot of Fig 3B, the same goes for the Lck blots of 3C and 3D.

        • michaelhbriggs June 12, 2013 at 5:50 pm

          Some of the Lck bands in 3C and 3D are the same as those in 3A and 3B, just shifted to the right or left.

          • Anon June 15, 2013 at 5:53 am

            Another one? (Brain, 127(5), 996-1008. doi: 10.1093/brain/awh117). Figure 5 is meant to be from 2 patients but both controls look the same with the possible addition of one spot.

  • Schmuck June 13, 2013 at 11:56 am

    Maybe the Mole and his Ant friend are correct. We need an absolute transparency in science http://jcs.biologists.org/content/126/11/2315.full.html?etoc
    However, in the absence of a scientific repository where scientist should deposit their raw unfiltered data, I may suggest that at least the editors and reviews should have access and should review these raw data before any publication

    • fernando pessoa June 13, 2013 at 2:22 pm

      In reply to Schmuck June 13, 2013 at 11:56 am

      From: http://jcs.biologists.org/content/126/11/2315.full.html?etoc

      “I suggest that the first thing we should do is look forward, not back.”

      Even five minutes ago is history. It would let some of the high-ups off the hook.

      A simple example.

      Neoplasia. 2000 Nov-Dec;2(6):505-13.
      Aspirin induces apoptosis through release of cytochrome c from mitochondria .
      Zimmermann KC, Waterhouse NJ, Goldstein JC, Schuler M, Green DR.

      Full article can be found here: http://www.ncbi.nlm.nih.gov/pubmed/?term=11228543

      Source
      Division of Cellular Immunology, La Jolla Institute for Allergy and Immunology, San Diego, CA 92121, USA.

      Please compare the band in the 48 h HeLa lane (right-most lane) of the Cyt c panel of
      figure 6B with the band in the middle lane (+ASA) of figure 5B.

      I think that the treatments are different.

      Figure 5.
      http://www.ncbi.nlm.nih.gov/core/lw/2.0/html/tileshop_pmc/tileshop_pmc_inline.html?title=Click%20on%20image%20to%20zoom&p=PMC3&id=1508093_neo0206_0505_fig005.jpg

      Figure 6.
      http://www.ncbi.nlm.nih.gov/core/lw/2.0/html/tileshop_pmc/tileshop_pmc_inline.html?title=Click%20on%20image%20to%20zoom&p=PMC3&id=1508093_neo0206_0505_fig006.jpg

      • ferniglab June 14, 2013 at 12:59 pm

        I am not over impressed by the argument put forward by Mole and his/her friend. Why? Because what we do tomorrow in science depends on what we currently understand. This in turn depends on the literature. If papers are the product of misconduct and/or fraud and the data do not support the conclusions then we have no evidence.

        I think the confusion is arising, because it is possible to predict the outcome of some experiments with some likelihood of being right. This is NOT the same thing as gathering the evidence. so in some instances the conclusions of the paper may be correct, but the fact remains that there is no hard evidence. The “reward” of publication goes to those who gather such evidence, not to those who take short cuts and make it up.

        A second argument, not addressed by Mole, but which is very important relates to training. If young scientists in training see this sort of practice rewarded by publication, they are likely to become rather cynical and leave science or join the ranks of the fabricators.

        So we do need to clean up the literature.

        Generally in life, politics and so on, acquiescence, complicity and a steadfast lack of attention to history only leads to disaster.

      • Junk Science June 14, 2013 at 3:06 pm

        Just a follow-up to fernando’s post. A Zimmermann and Green favorite:
        http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2173462/figure/fig6/
        6C, UV and CHX both CYTC shrna, identical image just adjusted slightly. Control CHX and CYTC shrna ActD, identical image just adjusted slightly.

        • fernando pessoa June 14, 2013 at 3:44 pm

          In reply to Junk Science June 14, 2013 at 3:06 pm
          If you look at figure 3 in the same paper. J Cell Biol. 2002 March 18; 156(6): 1077–1087.

          Figure 3 http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2173462/figure/fig3/

          the right 1/3rd of the control ARK dsRNA panel and the left 1/3rd of the UV ARK dsRNA panel look very similar. What a coincidence!

          • Erp June 14, 2013 at 5:51 pm

            Er Fernando you might have noticed this as well, but the bottom 1/3rd of the UV ARK dsRNA looks very similar to the top 1/3rd of the control control panel. The right 1/3rd of the control control panel in turn looks like the left 1/3rd of the ActD ARK dsRNA. Apart from that I don’t see any other ‘coincedences’ but it is almost a full set of panels.

  • fernando pessoa June 14, 2013 at 7:21 pm

    In reply to Erp June 14, 2013 at 5:51 pm

    I just had that feeling that there was more to it, but couldn’t spot the overlaps.

    I think that somebody should do a brain study on when you feel that there are more overlaps, but cannot spot them. Something must be going on in our brains.

    All the images reminded me of the Apollo moonshots.

  • Tex October 1, 2013 at 7:14 pm

    Any idea why this paper has not yet been retracted?

  • Post a comment

    Threaded commenting powered by interconnect/it code.