Science Integrity Digest

Concerns about Marseille’s IHUMI/AMU papers – Part 1

Last March, I shared my concerns about a paper from the IHU-Méditerranée Infection (IHUMI)/ Aix-Marseille University (AMU) claiming that Hydroxychloroquine in combination with Azithromycin could reduce coronavirus viral loads faster than no treatment.

Other papers by this group of researchers led by Professor Didier Raoult and/or his right-hand man Professor Eric Chabrière, also appear to contain problems, ranging from potentially duplicated images to ethical concerns.

In this blog post I have gathered the papers by the Raoult and Chabrière group that have image concerns. This post is not an accusation of misconduct, but a compilation of the potential problems found in 22 different papers by this group. I welcome the authors removing any concerns by providing the original figures.

Concerns about duplicated images

A set of at least 22 papers from the AMU/IHUMI group appear to have problematic images, including unexpected similarities across or even within figure panels. All of these have been posted on Pubpeer, including nineteen that I found and posted there. As of today the authors have not replied to most of these PubPeer posts, nor have they corrected the errors.

Category I duplications

Some papers, such as DOI: 10.1093/ajcp/101.3.318 [PubPeer] or 10.1016/j.ijid.2006.10.005 [PubPeer] appear to contain ‘simple’ duplications in figure panels. As explained in a previous post, I call these Category I duplications, where the exact same photo is used to represent two different experiments. These kinds of simple duplications can often be honest errors, where something went wrong during the paper submission or manuscript handling. A cheerful “Oops, yes you are right, we made an error” would have been enough, and these duplicated panels could be easily addressed with a Corrigendum. Unfortunately, the authors have not replied to most of the PubPeer posts.

Source: DOI: 10.1093/ajcp/101.3.318 [PubPeer]
Figures 1 and 2 appear to show the same photo
Source: 10.1016/j.ijid.2006.10.005 [PubPeer]
Panels b and c look more similar to each other than expected
Source: DOI 10.1371/journal.pntd.0001540 [PubPeer]
Figure 2 appears to be identical too Figure 1 in https://doi.org/10.4269/ajtmh.12-0212 [PubPeer]
as noted by François-Xavier Coudert

In DOI 10.1002/jcb.22135 [PubPeer], the two 32h flow cytometry panels look unexpectedly similar, while the gated percentages are different. One of the authors initially replied on Pubpeer and Twitter that the two images did not look identical, but later admitted they looked ‘intriguingly similar’.

Source: DOI 10.1002/jcb.22135 [PubPeer],
where two panels look very similar but have different gated percentages

The same photo appears to have been used in two papers, DOI: 10.1016/j.nmni.2017.12.006 [PubPeer] where it represents Bacillus salis bacteria, and DOI: 10.1002/mbo3.638 [PubPeer] where it represents Gracilibacillus timonensis bacteria.

Source: DOI 10.1016/j.nmni.2017.12.006 [PubPeer] and DOI: 10.1002/mbo3.638 [PubPeer].

Category II duplications

Two papers in this set contain Category II duplications, where images overlap or might have been rotated, mirrored, etc.

DOI 10.4269/ajtmh.15-0436 [PubPeer] has been corrected after posting it on PubPeer. In this paper, three of the four panels shown in Figure 1 were showing the same specimen. The aspect ratio of these overlapping areas, marked below with blue and magenta boxes, was not always the same. It appears the images might have been stretched differently, which would be unexpected if the errors were unintentional. Yet, the journal accepted a new set of panels in the July 2020 correction. The authors wrote: “These resulted from errors by a researcher that were missed by other authors. Specifically, a single image was inappropriately inserted to represent three different experiments.

Source: DOI 10.4269/ajtmh.15-0436 [PubPeer]
Three panels appear to overlap with each other. This concern was addressed with a correction.

Paper DOI 10.1371/journal.pone.0010041 [PubPeer] contains a figure in which two panels might be showing the same leaf.

Source: DOI 10.1371/journal.pone.0010041 [PubPeer]
Leaf A looks very similar to leaf B

In DOI 10.1128/JCM.01714-06 [PubPeer], published by the American Society for Microbiology (ASM), Figure 3A and 4A appear to show the same blot, albeit cropped differently. Of note, the name of senior author Didier Raoult was removed some time between the acceptance of the paper [JCM Accepts version; archived] and the publication of the PDF on the journal’s website. In 2006, the year this paper was published, Raoult was banned from publishing in ASM journals for 1 year, which might have something to do with this author name removal (source: https://science.sciencemag.org/content/335/6072/1033).

Source: DOI 10.1128/JCM.01714-06 [PubPeer]
Panel A in Figure 3 looks very similar to panel A in Figure 4

In DOI 10.1089/vbz.2012.1083 [PubPeer], two Western blot strips look unexpectedly similar, albeit at different exposures and cropping. Found by PubPeer user Trichoderma Viridescens.

Source: DOI 10.1089/vbz.2012.1083 [PubPeer]
Two Western blot strips representing different sera look more similar than expected

In DOI: 10.1016/j.cimid.2018.06.004 [PubPeer], two blot panels incubated with different sera look remarkably similar, albeit shown at different exposures and with different labels. The authors did not reply on PubPeer, but admitted to the journal that two panels were indeed duplicated. The paper was corrected a couple of months later.

Source: DOI: 10.1016/j.cimid.2018.06.004 [PubPeer] – now corrected
two blot panels look remarkably similar

Category III duplications

At least 10 AMU/IHUM papers with image problems contain Category III duplications, where figures appear to have been altered or to contain duplicated elements. These types of duplications are the most likely to be the result of an intention-to-mislead.

Category III duplications were found in these papers:

Source: DOI 10.1128/jcm.35.7.1715-1721.1997 [PubPeer]
Two pairs of lanes in Western blots look more similar than expected
Source: DOI 10.1128/iai.68.10.5673-5678.2000 [PubPeer]
Some Western blot lanes appear to look more similar than expected
Source: DOI 10.1128/IAI.69.4.2520–2526.2001 [PubPeer]
Cells in these microscopy panels appear to be surrounded by sharp horizontal and vertical background transitions.
Source: DOI 10.1128/JCM.39.2.430–437.2001 [PubPeer]
Boxes of the same color show areas (some including bands) look more similar to each other than expected in these DNA gels
Source: DOI 10.1086/379080 [PubPeer]
Certain areas in this Southern blot look unexpectedly similar
Source: DOI 10.1128/JCM.43.2.945–947.2005 [PubPeer]
Some parts of these Western blots look more similar than expected, as marked with boxes of the same color
Source: DOI 10.1128/AEM.03075-05 [PubPeer]
Some areas in this Transmission Electron Microscopy photo look unexpectedly similar to each other
Source: DOI 10.3389/fmicb.2010.00151 [PubPeer]
The bottom photo appears to have been created with parts of the top photo, which in turn was taken from Wikipedia without attribution. The original by Philippinjl can be found here:  https://commons.m.wikimedia.org/wiki/File:Api20e.jpg
Source: DOI 10.1371/journal.ppat.1003827 (now retracted) [PubPeer]
Three lanes in this DNA gel appear to look very similar to each other
Source: DOI 10.1016/B978-0-323-55512-8.00069-7 [PubPeer]
Two areas in Western blots A and C appear to look very similar to each other, while the marker lane looks different

Part 2 of this series — still to be written — will explore papers by the IHUMI/AMU group with potential ethical concerns.