The Journal of Clinical Investigation has retracted two papers from the lab of one of Stanford University’s most prominent cancer researchers over concerns about the integrity of the data.
The articles, published in 2012 and 2014, described work on ways of priming the immune system to enhance the activity of drugs to fight cancer.
The first author on the two articles was Holbrook “Brook” Kohrt, a superstar young faculty member who died in 2016 of complications of hemophilia. Kohrt was the subject of this 2013 profile in the New York Times, which also wrote an obituary of him.
According to a person familiar with the articles, who did not want to be identified, Kohrt was responsible for the data — the senior author was Ronald Levy, a major figure in blood cancers — but Stanford learned of problems with the images after his death. When the institution could not find Korht’s relevant lab notebooks — the data were kept there, rather than electronically, according to the person familiar with the research — it decided it had no other option but to retract the work.
The 2012 article was titled “Stimulation of natural killer cells with a CD137-specific antibody enhances trastuzumab efficacy in xenotransplant models of breast cancer.” According to the JCI, the journal recently heard from Stanford:
of concerns regarding Figure 4, A and C, and indicated that the original source data for these figures could not be located. In accordance with the institutional recommendation, the JCI is retracting this article.
Same for the second paper, “Targeting CD137 enhances the efficacy of cetuximab:”
Stanford University School of Medicine recently notified the JCI of concerns regarding Figure 4, B, C, and E, Figure 5, B, D, and F, and Figure 6B and indicated that the original source data for these figures could not be located. In accordance with the institutional recommendation, the JCI is retracting this article.
Both papers have been flagged on PubPeer, here and here. The 2012 article has received 142 citations, while the 2014 paper has garnered 106, according to Clarivate Analytics’ Web of Science.
According to the source, Blood soon will be retracting another paper by Kohrt, but that should be the last article to fall. However, Mark Paglia, the editorial manager for Blood, would not comment on the status of the paper. [See update on this post.]
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up for an email every time there’s a new post (look for the “follow” button at the lower right part of your screen), or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
I don’t see any PubPeer discussion; the links given are just reposts of the retraction notices. Does anyone have a link to an actual discussion of the issues?
The first paper shows no RETRACTION watermark as yet.
https://www.jci.org/articles/view/61226
It is unlikely that there would have been any discussion at PubPeer previously.
The survival curves look odd, in that
Paper 1: Figs 4 B, D
Paper 2: Figs 4 C, F; 5 C, E, G;
show survival curves with zero deaths among 10 mice in 100 days for one or two favoured treatment groups when all the mice in the other groups are dead before 100 days. It looks too good to be true, but it’s difficult to raise any serious discussion beyond speculation at such fantastic results. This is the conundrum when “superstar” scientists publish such papers. The rest of us look at our mouse data, with some animals growing runaway tumours and dying within 2 or 3 weeks, and we look at the Stanford data, where all 10 animals in all 5 groups are alive at day 50, and we reflect on our own inadequacies relative to the superstars at Stanford. But it turns out it’s easier to cheat when you are at Stanford, or Duke, or other hyped universities, because people just expect to see “superstar” results there.
The tumour size graphs show only averages, not the tumour size trajectories of each animal, and they are truncated at day 50 or 60 whereas the survival curves go on out to day 100 or 120. So tumour growth information is hidden, though in this case who knows if there was any real data to begin with. The survival curves look odd, with all animals surviving two months, then all dropping dead or being sacrificed within a few days.
It’s hard to make a case about this kind of odd looking tumour growth and survival curve data. Most non-statistician readers don’t have a good feel for what such data typically look like when graphed. It’s not like spotting duplicated lanes or bands in Western blot alterations. So lack of any prior discussion on PubPeer is not surprising here.
“But it turns out it’s easier to cheat when you are at Stanford, or Duke, or other hyped universities, because people just expect to see ‘superstar’ results there.” What a statement! Casteism is rife among scientists, and this statement captures it very well.
Let us not forget that the entire foundation of this work was very shaky to start: a study of immune cell changes due to human cell line xenografts onto severely immunocompromised mouse strains. If I submitted such a paper to JCI (or the like) in 2011 I would expect (and deserve) to receive a snarky “do you even know what you’re doing?” rejection letter.
But this paper:
“Received: September 30, 2011; Accepted: January 4, 2012”
That’s the effect of of unchallenged scientific “star power”.
As for the survival and growth data, yep need to call that out more. I received a paper just this week with the same kind of “magic”.
The second paper is available here:
https://www.jci.org/articles/view/73014
https://www.jci.org/articles/view/73014/pdf
Targeting CD137 enhances the efficacy of cetuximab
The PDF shows no “Retracted” watermark, though the web page is marked.
Odd and unfortunate that the PDF is still available with no RETRACTED marking.
Before restricted those papers, did anyone repeat those experiments?