Fudged figures sink breast cancer paper

oncogeneA prestigious cancer journal has pulled an article over “concerns” regarding some of the figures, which PubPeer commenters had tagged as suspect.

A few weeks after the paper was published on June 9, comments on PubPeer began accumulating. Commenters called out both potentially manipulated and repeated images. The exact timeline is not clear, because Oncogene does not list a date on the retraction notice, but by August 29 the paper had been retracted.

Here’s the notice for “IL-6 secreted by cancer-associated fibroblasts induces tamoxifen resistance in luminal breast cancer,” by researchers at Shanghai Jiaotong University School of Medicine and Ruijin Hospital, both in Shanghai, China, and the University of Michigan:

The authors wish to retract this article due to concerns raised regarding some of the data presented in Figures 2, 4, 5, 6 and Supplementary Figures 6 and 7.

According to the data presented in Figures 1, 3 and 7, the major conclusion drawn from this article is still valid: IL-6 secreted by CAFs is causal factor for tamoxifen resistance in luminal breast cancer.

The authors are deeply regretful and apologize to the readers.

A spokesperson for the journal declined to comment, and suggested we speak with the authors. We’ll let you know if we hear back.

Hat tip: Leonid Schneider

22 thoughts on “Fudged figures sink breast cancer paper”

  1. There seem to be a specific problem with the deluge of scientific publications originating from China/Chinese scientists. We’ve used applied onomastics (the science of proper names) to analyse the growth of such publications in PubMed/Medline between 2002 and 2012 and how much ‘digested’ they are by the international research community. We found interesting cultural biases: scientists with Chinese names have a positive bias towards citing other scientists with Chinese names, but not more than many other countries/cultures. More interestingly, we found that the negative bias of the international scientific community (from the British to the Italian) has grown wider and wider: the more publications by Chinese scientists, the less cited they are. This trend -along with cases of retractions such as the one mentioned in RetractionWatch- could seriously undermine the reputation of China as a growing scientific and industrial power. Read further
    http://namesorts.com/2014/08/20/is-china-really-becoming-a-science-and-technology-superpower/

    1. Leonid, I agree 100% with your criticism. But don’t solely blame the authors. My guess is that the majority of non-native English speaking scientists probably can’t understand the legal and ethical subtleties of such notices (maybe not even the jargon in some of them), so I think the approval of such ridiculous and outrageous statements lies squarely on the shoulders of the publishers, in this case, Nature Publishing Group. Please express your outrage each time so we can all give you a thumbs up! At some point in time, a frank and peaceful interview needs to be had with the authors of retracted papers so that the background interaction with the publisher can be assessed.

      1. Jaime, I meant both parties, the one who wrote this statement and the one who thought it appropriate to allow it in a retraction notice. There would be much less of fraudulent papers if journals had any (financial) incentive to combat it. Sadly, they don’t, instead they, especially NPG, combat retractions! This case demonstrates how even a retraction can be hidden and therefore practically undone, by the editors themselves.

  2. Interesting timing for this RW post….

    For was it not over this most recent, late summer, sunny, sleepy weekend that Retraction Watch linked to an article that suggested there was not a problem with Oncogene, relative to Nature itself, because they at Oncogene don’t retract much, if anything at all, so there can’t be a problem? The linked blog article

    http://www.ipscell.com/2014/09/could-natures-2-year-torrent-of-paper-retractions-be-a-good-thing/

    But yet, unless I have gone completely brain dead, then it is a verifiable fact that Oncogene articles with dodgy figures turn up with monotonous regularity at PubPeer. Maybe, however, fingering these articles by PPPR has not lead to significant retraction activity in that particular journal at this moment in time.

    Might it be getting to the stage where PubPeer should allow their database to be searched by journal name? This might help worthy blogs like ipscell.com to stop making assertions that Oncogene does not have a problem publishing fictitious work when “as any fule kno” (google it), it most surely does.

    I for one hold the view that retractions != problem papers in any given journal. And that some journals are still in total denial. Or perhaps, even, that all journals are in denial, but some journals are more in denial than others.

    1. I’ll drink to that! In as far as plant science journals are concerned, there are only specks of recognition here and there that there is a problem. We need topic-by-topic activists and PPPR specialists to take center stage now, use RW to highlight the problems and centralize the themes, and PubPeer and PubMed Commons to list the exact errors. We can then thank Google for allowing these cases to be traced easily if this triangle of information is made for every problematic paper. At some point, entering someone’s name, a keyword or a title will start to link back to one of these three sites. Then, we will start to see the extinction of denial.

      1. Very useful link! Backs up what Scrutineer said at 6:02 ‘ unless I have gone completely brain dead, then it is a verifiable fact that Oncogene articles with dodgy figures turn up with monotonous regularity at PubPeer.’ In fact 100 articles from Oncogene are featured, which is massively above the average. Articles from cancer journals in general seem to be widely featured. Are these high volume journals? If not, this is very worrying for those funding cancer research.

        1. Cancer research is the biggest money trough there is. The incentive to go into this field and to cheat there is therefore much higher, than, let’s say, in plant sciences. Same applies unfortunately to stem cells, see STAP.

          1. My theory about the plant sciences is that because precisely there is so little funding, the incentive to commit fraud is massive because competition for the few available positions that exist is great (graduates tend to stay in academic because society is almost unable to absorb them into any working positions in the “real world”, so this would also add more pressure to the system). In contrast, in cancer or stem cell research, with a big fat grant, you can also afford to hire a couple of post-docs, so the opportunities appear to be greater than for the plant sciences. As we would say in Portuguese “1000 dogs to one bone”. So, at the end of the day, it seems that wealthy research fields and poverty-stricken ones might both serve as extreme incentives, potentially leading to the same result: misconduct, cuting corners, fabricating data, etc. Most certainly this is an issue worth exploring as more data emerges (first we need a few dozen more retractions from the plant sciences before we can say anything).

          2. Interesting to compare how often articles from different fields feature. Take, for example, Oncogene, Molecular Microbiology and The Plant Journal, which publish a similarish number of articles a year (565, 345 and 341, respectively) and are probably journals of similar reputations in their respective fields. Articles from Oncogene feature 100 times, from Mol Micro 3 times and from The Plant J 1 time. I guess either the plant sciences and microbiology communities aren’t really engaged with PubPeer or Leonid is correct. Just to be on the safe side, I think I’ll be checking my wallet next time I walk past our Cancer Research Institute.

          3. Leonid, don’t get me started. It’s 4:39 am in the land of the rising sun and I don’t want to be forced to write one of my trade-mark long responses…! Needless to say, you probably can imagine on which side of the fence I sit when it comes to Monsanto.

      2. Peer 0 -Thanks for the rapid PPPR of my comment! Any embarrassment at being a bit dim has been well trumped by joy at the commenting activity that has ensued 🙂

        Scanning the list by eye (I might have missed something), here are the journals with >40 entries in PubPeer

        291 JBC
        150 PNAS (129+21 under two spelling variants)
        140 Cancer Res
        102 Nature
        100 Oncogene
        84 Science
        59 Cell
        57 N Engl J Med
        51 Clin Cancer Res
        43 JCI

        In fairness to perceived editorial standards, an obvious correction would be by number of published articles per year. Then the order of the journals would change. Might Oncogene rise up the list?

        A few of those posts might be genuine discussions about the scientific content but, as we all know, most concern visible artefacts in the figures. So we can lay to rest the canard that Nature v. Oncogene retraction count has anything to do with scientific quality standards. Notwithstanding a minority of excellent and ethical cancer researchers, we all know that a certain kind of spiv pours into this field in anticipation of fame, fortune and Nobel prizes, riding roughshod over all that stand in their way. (Oddly enough, not many of these do end up getting their Nobel.)

        Nevertheless, we should be cautious that fields such as plant science have lower levels of fraud. They suffer the same pressures on funding, short term contracts etc. We must be wary of acquisition bias. Figure sleuthers well known to RW readers, such as the pseudonymous Clare Francis, Fernando Pessoa and David Hardman don’t seem to have been heavily scanning articles in this field.

        There are research fields with reduced levels of fabrication such as structural biology. In that field it is because fraudsters have no place to hide. Artefacts in their deposited information give them away sooner rather than later.

        Finally it might be worth noting that there is some PubPeer activity concerning plant research going on as we post. Most of the articles arousing concern were not actually published in the specialised plant journals 😉

        1. Actually taking Scrutineer’s list and doing the suggested correction (dividing by number of articles in a year) changes things completely. The ranking order for the PubPeer Hall of Fame/Shame (take your pick) looks like this.

          1. Cancer Res
          2. Oncogene
          3. N Engl J Med
          4. Cell
          5. Nature
          6. Science
          7. JCI
          8. JBC
          9. Clin Cancer Res
          10. PNAS

          So a clear victory for cancer sciences, followed by the big 4 glamour journals (although frankly who takes anything at face value they publish these days).

          The really shocking thing for me is the complete lack of immunology journals. I demand a recount!

          1. Thank you The Crab for doing the normalisation. As expected, Oncogene shoots up the scale. Still we should be careful about acquisition bias in the upwards as well as the downward direction and not take these rankings too strictly. Though the problem we are facing is of course clear enough.

            Regarding your disappointment with the immunology journals, a recount may not be necessary. I have once again been a tad misleading in my commenting. I blame failing eyesight. For there is this – J. Immunol. (47) – in the list. So JCI would drop out of the top ten. And you would become unschocked “-)

          2. Thanks Scrutineer. The natural order of things is restored. I can sleep soundly once again. However, it is a sobering thought that there may be a field out there dirtier than immunology (if that is indeed what the list means).

  3. Noteworthy for the speed of retraction after publication. Often these matters drag out into months and years of obfuscation. One is tempted to praise the author or authors who took the decision to pull the plug so rapidly, however if they’d taken their time prior to publication to actually properly review all the data and figures, this would never have reached such a damaging conclusion. More haste less speed.

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.