(1) An article was published after peer review. Shortly after online publication we received a message from a reader (an academic who works in the same field as the authors) notifying us of a major concern with one of the figures in the article.
“I am writing with regard of manuscript XXX recently published in XXX. These studies raise significant expectations in XXX patients, because the proposed strategy achieved unanticipated therapeutic success in a preclinical model of XXX.
I am writing because I’ve noticed a major problem in Figure X. It appears that two of the panels (X and X), which are supposed to be representative of two different XX specimens from two different experimental groups, belong to the same tissue section (or quasi-identical serial sections)....However, the major problem in this figure is that the two panels are only partially identical. ...These observations raise the concern that the pictures have been intentionally manipulated, and I believe that this concern should be brought to your attention.”
(2) We contacted the authors without revealing the concerned reader's identity and without implying any judgment as follows:
One of our readers has noticed a potential problem with one of the figures in your paper (see below) and I'd like to get your input before deciding how to respond.
“I am writing with regard of manuscript XXX recently published in XXX. These studies raise significant expectations in XXX patients, because the proposed strategy achieved unanticipated therapeutic success in a preclinical model of XXX.
I am writing because I’ve noticed a major problem in Figure X. It appears that two of the panels (X and X), which are supposed to be representative of two different XX specimens from two different experimental groups, belong to the same tissue section (or quasi-identical serial sections)....However, the major problem in this figure is that the two panels are only partially identical.”
(3) The authors responded initially by sending higher resolution images saying:
Please find the attached fig. X. At original magnification it is obvious that the problematic 2 panels are taken from different samples. We are happy to send the original pictures.
(4) We asked one of our editorial board members who has relevant expertise in the area of science and in digital imaging to look at the higher resolution images, in confidence. He responded with the following:
“If the claim is that the three photos of Fig. X represent different samples I would in indeed have a problem with that. Same elements look identical. Difficult to prove though as there are larger parts that differ. Nevertheless, that concerned reader may have a point. I am suspicious, too, that the figures have been manipulated.”
(5) We went back to the authors asking for the original digital images with the following message:
I had forwarded your high resolution version of figure X to the person who originally expressed concerns and received a message in which the person states that his/her concerns remain. I then sent the high resolution image to one of our regular advisers who has experience in neuropathology and digital images. This person agrees that parts of the lower two panels of Figure XB look very similar and that we should ask to see the originals for all three panels in Figure X and obtain detailed information about which software was used and whether the images were digitally manipulated in any way.
As a consequence, I am now asking you to send the original images and any relevant information you have.
(6) They responded by sending three images and the following:
Please find the attached requested original images. Adobe Photoshop was used to reduce brightness, increase contrast and compress the images to JPEG format. The JPEG images were inserted into Powerpoint file where borders of the tumours were marked. No additional manipulations were made to the original images. Please let me know if you need any additional information.
(7) We asked our production people to look at the originals and they felt that there was no better evidence that the images came from different samples. I went back to the authors one more time to give them a chance to tell us more:
“I have now shown the images to several people who know more about digital image generation and processing than I do. All of them find that the striking similarities between groups of cells in the middle and low panels make it hard to believe that the two images come from different tumors (as is implied in the figure legend and your earlier messages).
As it doesn't seem that we are able to resolve the issue easily, we will need to conduct a more formal investigation (most likely with the help of people outside of PLoS). If there is any additional information you can provide, please do so by XXX.
I regret the inconvenience this is causing for all of us, but we feel that we must take the issue seriously.”
(8) To this the authors responded:
“I find this conclusion rather odd and not in accordance to our laboratory protocols. I can assure you that we shall cooperate in any way you see fit and our records on the study are open.”
and, in a separate message a few hours later:
“I am looking again at these pictures. If you carefully study the images you will find that almost all the fields are different and there is no similarity at all. You always can find groups of cells, with a bit of imagination you can make them look similar. I can assure you that all this study is fully controlled and the reproducibility is excellent. We now have even results with the systemic application of the XXX vector to disseminated tumors. I find it hard to believe that I am going to be under investigation questioning my scientific integrity.”
(9) We then enlisted the confidential help of an editor from a different journal who is an expert on manipulation of digital images and has software that can look for specific alterations. We have not heard back from him yet.
What do we do if he (the editor from a different journal) tells us that he has detected manipulation beyond what the authors said they did?
What do we do if he says there is no evidence of manipulation of these images?
What do we do if he says he cannot confirm either way?
More generally, have we handled this process appropriately thus far?
It was agreed that the handling of the matter to date had been exemplary in the editor’s careful and considered approach. With all parties, except for the authors, feeling that some sort of manipulation had taken place with the figure, the only remaining course of action was for the editor to write to the author’s institution and raise the concerns with them.
The author’s institution was contacted and the editor forwarded all relevant correspondence with the author and, in anonymised fashion, correspondence with the scientists who brought this to their attention. The rector of the author's university has assured the editor that they take this seriously, have asked that the journal take no further action while they are investigating and will inform the journal of the outcome.