Readers of this blog are aware that many of the retractions we’ve covered involve the misadventures of post-docs. That makes some superficial sense: post-docs, after all, are trainees, and therefore might be more likely to make mistakes. They’re also hungry to break into their chosen specialty, and how better to do that than by producing spectacular results? (None of this is to say that post-docs are by nature incompetent or venal — only that the raw ingredients exist for typecast villainy.)
But one figure about whom we haven’t written (to the best of our knowledge) is the career lab tech — until today.
Matthew Connors was working as a tech at the University of Chicago (Chief Research Technologist, on his online resume) when he became a co-author on a 2007 study, published in the journal Small, purporting to show a technique for encapsulating pancreatic islet cells in a coating that’s opaque to the immune system. As the researchers explained:
Immunoisolation through microencapsulation offers the opportunity for immunosuppression-free cellular transplantation and would therefore substantially increase the supply of islets from additional donor sources. The method we have demonstrated for encapsulating pancreatic islets fulfils many of the requirements of an effective immunoisolation technique. The thin coatings that surround the islets permit rapid diffusion of essential small molecules, such as oxygen, glucose, and insulin. Large molecules, including those involved in the immune response, are prevented from reaching the protected islets. A successful microencapsulation scheme that immunologically isolates donor tissue from the host immune system raises the possibility of using plentiful nonhuman (for example, porcine) islets as replacement therapy for Type I diabetes. These in vitro results thus warrant further in vivo study.
In other words, the new approach could in theory lead to a virtually inexhaustible supply of insulin-producing cells available for transplanting into diabetics.
The paper has been cited 14 times, according to Thomson Scientific’s Web of Knowledge. But there was a problem. According to a retraction notice this month in Small:
After the publication of this article, the corresponding authors discovered that the results of Figure 6 could not be reproduced. Upon bringing this matter to the attention of the University of Chicago Academic Fraud Committee, the Committee came to the conclusion that Matthew Connors fabricated these data. The Committee determined that there was no evidence that the rest of the authors engaged in unethical behavior or were aware of the data being fabricated.
Although the bulk of the paper describes the encapsulation of the cells, Figure 6 seems to be the rubber-meets-the-road moment when the researchers attempt to demonstrate “that the encapsulated islets functioned normally, as measured by their release of insulin in response to glucose.”
We used a standard assay to determine the concentration of insulin in the fluid and, from these data, we constructed the curves in Figure 6 that relate insulin release to glucose stimulation of the islets.
“Constructed” indeed.
However, the retraction notice concludes:
No questions were raised with respect to the results describing the design of the apparatus for the selective withdrawal method and the photochemical cross-linking of a polymeric shell around the islets. The corresponding authors and the rest of the authors (other than Matthew Connors) have requested and agreed to this retraction.
We reached Marc Garfinkel, a former University of Chicago surgeon and senior author of the paper, by e-mail, but he refused to elaborate on the retraction notice. We’ve attempted to contact several other co-authors of the paper — who include researchers from across the university and the Howard Hughes Medical Institute — and will update this post when we hear from them. We also have tried to reach the fraud committee for comment and are waiting for a response.
Connors left Chicago in February 2007, having spent four years there, and moved to the University of Miami, where according to his Linkedin page he performed “diabetes research using small animal models, and work on synthetic models.” He left Miami in September 2008 and seems to have been unable to find a job as a lab tech since then.
Connors’ online resume lists 11 publications, including several in high-powered titles such as Arthritis and Rheumatism, Diabetes, the Journal of Immunology. The 2007 Small paper is not among them.
So far, no other articles on which Connors was a co-author have been retracted.
“[…] post-docs, after all, are trainees, and therefore might be more likely to make mistakes.”
Really? Sure you ain’t talking about PhD students?
“Small” is a brilliant journal name! I hope they print all their articles in font size 4.
It seems quite unbelievable that a lab tech fakes results. Why should he do that and what about the responsibility of the PI? There is really no incentive for a tech since they do not get any benefits out of it, e.g. degree or grants.
It would really be interesting if some more information could be obtained here — how did the lab members decide to initiate a fraud investigation after having difficulty reproducing the data?
Data irreproducibility is, in my experience, not as uncommon as one might expect. Often the reason is unclear — there must have been something obviously suspect for the lab members to call up the fraud committee. Therefore, I wonder why the co-authors didn’t catch it before submission? People love to have their names listed on manuscripts, but seem to retreat into “self-preservation” mode whenever suspicions are raised. It bothers me how co-authors seem to have their cake and eat it too with respect to responsibility for what they ascribe their names to.
I’m not sure about the opening statements here regarding post-docs. I was once a post-doc and I do agree that some of them are a little bit “dodgy” in the lab, but from what I can see the biggest problem is with the PI. Let us not forget that ultimately the buck stops with them. In most cases they are the ones putting pressure, overt or otherwise, on their staff and are all too happy to see perfect results. They don’t ask the difficult questions when they see a good result, they just take it at face value. Many of them are blatently out of touch with the latest technical developments and do not understand the pitfalls of modern protocols. Several of my colleagues are very poor bench scientists (MDs) but still have an army of workers generating lab data. They are just not qualified to deal with the type and amount of data they receive. The papers that result from this kind of envrionment lack experimental rigour. These days papers are more about how many figures you have, rather than how well the experiment was performed. I fear that this website will become more and more relevant because this situation is goionly going to get worse.
Great analysis Dave, you really got the point. There is surely still too little attention on the PI side, way too little.