Teflon toxicity paper fails to stick

toxicological sciencesAn advanced online paper on prenatal toxicity of perfluorooctanoic acid (PFOA), an industrial chemical used to make waterproof coatings and Teflon, is being retracted due to “some minor errors.”

High blood levels of PFOA have been tied to kidney disease in humans, as well as several cancers in animal models. The majority of the U.S. Environmental Protection Agency’s scientific advisory board deemed PFOA “likely to be carcinogenic in humans” in 2006, though a decade later the EPA has yet to make a decision on regulations. The retracted paper found that exposing pregnant mice to PFOA altered hormone pathways in mammary glands.

According to the notice in Toxicological Sciences, there was a duplicated image in one of the figures, as well as “some minor errors.” Here’s figure 5B:

Screen Shot 2015-04-08 at 5.31.26 PM

The notice states that the duplication, which seems to be between PPARβδ and PPARγ, is due to “mistakes made in conversion of raw RT-qPCR data and in file naming.”

The editor of Toxicological Sciences, Gary Miller, told us that raw data provided to him by the authors — based at the National Institute of Environmental Health Sciences and the EPA — have indicated “a series of careless errors by an individual in the laboratory” that would require two figures to be corrected.

The authors wanted to submit an updated version for republication, but Miller declined.

Here’s his full statement to us:

Shortly after this paper was published on Advance Access, Oxford University Press’ online-before-print format, a reader contacted the editorial office.  This reader noticed that one of the images was duplicated within a particular figure. I confirmed this error and we immediately initiated an investigation following the COPE guidelines.  We sent an inquiry to the senior author and requested the raw data, image files, and data files. The senior author complied and said that the research team would investigate what happened.  Several days later the senior author contacted the editorial office and explained that they had identified where errors occurred in the generation of the figures and labeling of files.  The senior author provided a very detailed explanation and it appeared that the problem could be traced to what I perceived as a series of careless errors by an individual in the laboratory.  During their review of the data they realized that two figures would need to be corrected. The revised figures were sufficiently different such that they would need to undergo peer review.  Given that the manuscript was already in production I felt that it would be inappropriate to bypass the peer review system and correct the manuscript. At that point I suggested that it would be best to retract the manuscript and the senior author agreed to do so.

The retraction notice was published on Advance Access, like the original paper, as soon as possible. Our publishers agreed that the journal should not publish the print version of the paper with errors.

The authors requested to submit a new and revised version, but I am not allowing them to do so.  Except for not allowing a resubmission, the journal is not taking any punitive action against the authors and is willing to entertain future submissions from this group.

Miller also addressed our concerns that the duplicated bands had identical shapes but different smudging along the sides of the bands. Although this could possibly suggest manipulation, rather than a simple file mixup, he didn’t see any evidence of that:

I saw all of the data and images.  I did notice exactly what you describe and that was my immediate thought.  I even ran some of the software comparing the various images, but after considerable analysis I could not see any advantage in the way the figures were constructed. The densitometry was performed on the original blots and was consistent with the original blot, but the student creating the figures was following poor practices in cutting and pasting lanes and tried to make the figure a couple of different times (weeks apart).  The generated images were mislabeled…I agreed with the senior author that there was no manipulation to gain advantage, just sloppy work.  The senior author and a more senior co-author reran everything from scratch and came up with the same results and conclusions, which is why I did not refer this to their institutional official.

I will say that I approached this with a very skeptical eye. I also consulted with one of my Associate Editors and she agreed with my assessment, as did my Managing Editor.

A spokesperson for the National Institute of Environmental Health Sciences gave us a statement via email:

Sue Fenton, Ph.D., one of the lead authors on the paper, voluntarily requested the retraction of the paper. When the duplication of a Western blot inset was reported to her in one of the figures, she requested a hold on publication so that the figure could be fixed. In reviewing supporting Excel data sheets, some minor errors made in converting raw data, which did not impact the overall conclusions, but were not as precise as the authors would have liked. Thus, rather than issue an erratum, she requested the retraction. The paper will be submitted for review in the coming weeks.

Here is the notice for “Prenatal PFOA exposure alters gene expression pathways in murine mammary gland”, first published online December 8, 2014:

Following the Advance Access publication of the paper “Prenatal PFOA exposure alters gene expression pathways in murine mammary gland” by Macon et al. in Toxicological Sciences, a reader noticed a duplicated image within Figure 5B. Upon notification by the editorial office, the senior investigator initiated a full review of the data presented in the manuscript, after which the results of the investigation and much of the raw data were provided to the editorial office. During the evaluation of the raw data and data analysis by the senior author and two others, some minor errors were identified. It was determined that the errors were not the result of data or image manipulation, but rather mistakes made in conversion of raw RT-qPCR data and in file naming. Since the problem involved more than the original image identified by the reader, the authors have requested that the article be withdrawn from publication so that these mistakes can be corrected.

Published by Oxford University Press on behalf of the Society of Toxicology 2015. This work is written by US Government employees and is in the public domain in the US.

We’ve contacted all three authors, and will update if we hear back.

Like Retraction Watch? Consider supporting our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, and sign up on our homepage for an email every time there’s a new post.

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.