Group issues model retraction over antibody error

The authors of a 2013 paper on antibody production in patients with rheumatoid arthritis have retracted the work in what looks to us like a case study in how to handle operator error.

The paper, “Monoclonal IgG antibodies generated from joint-derived B cells of RA patients have a strong bias toward citrullinated autoantigen recognition,” was published in the Journal of Experimental Medicine by a group from the Karolinska Institutet (KI) in Sweden and elsewhere, and has been cited 128 times, according to Clarivate Analytics’ Web of Science. The last author, Vivianne Malmström, is a specialist in cellular immunology at the KI.

Here’s an excerpt from the lengthy notice:

The editors of the Journal of Experimental Medicine have been notified by Dr. Vivianne Malmström of the Karolinska University Hospital, Karolinska Institutet, Stockholm, Sweden, that she and the other authors of the above article wish to retract the paper.

The authors state:

In our efforts to understand the contribution of autoreactive B cells and autoantibodies in rheumatoid arthritis (RA), we have continued to drive projects where we study B cells and plasma cells from different anatomical compartments and express patient-derived recombinant mAbs for downstream applications. Some of these efforts have recently been published (e.g., Steen, J., et al. 2018. Arthritis Rheumatol. https://doi.org/10.1002/art.40699). However, when using the mAbs first described in the above article as comparators to the newly generated mAbs in affinity measurements, we observed that the Kd values from our study could not be reproduced. After reviewing the original SPR data, we learned that the discrepancy was due to an incorrect instrument setting, and we can now conclude that the mAbs in our paper have no measurable affinity for the tested citrullinated peptides, while a number of our new mAbs do.

The notice goes on to explain how the authors refined and improved their methods over time, and how readers should put the earlier work into proper context:

Moreover, although we initially used widely spread protocols for both the purification of recombinantly expressed mAbs and the subsequent testing of their reactivity in ELISA, we grew aware that these protocols were not fully optimized for our research setting. Hence, in the time between the above article and the follow-up studies, we have continuously improved our methods and significantly refined our protocols based on state-of-the-art recombinant antibody methodology. We believe that this is critical for studies of antibodies from patients with pronounced autoreactivity. Caveats include storage of antibodies, aggregation tests, and excluding unspecific so-called polyreactivity (generally driven by charge or hydrophobic interaction). So, although the antibodies presented in the above article were positive based on published methodologies and the cut-offs used at the time, today we no longer regard them as citrulline-specific autoantibodies based on subsequent findings in our follow-up studies. The precise adjustments of the protocols to allow the distinction of sticky/polyreactive versus antigen-specific autoantibodies will be presented elsewhere.

Notably, mAbs from the above article still represent Ig sequences generated from the RA joint with interesting biochemical characteristics, which may provide insights into joint pathology. Indeed, some of them demonstrate the capacity to functionally contribute to disease by mechanisms that we continue to investigate.

The authors wish to communicate their apologies for their mistake concerning the SPR measurements and the misinterpretation of the ELISA data reported in the paper.

Malmström told us in an email that:

We became fully aware of the problems with the SPR assays when we re-tested two of our ‘old’ monoclonal antibodies described in the 2013 JEM paper in conjunction with our new antibodies described in the recently published paper by Steen at al (A&R August 27, 2018). At this time we collaborated with experienced SPR users, while in our first study we used a core facility instrument in a way that proved to be erroneous after scrutiny and comparison with the new SPR study.

In parallel to the comparative SPR experiment, we had already seen that the first generation of antibodies (stemming from the 2013 JEM paper) did not perform as well as the new generation. In fact, these new antibodies (e.g. Steen et al) helped us to substantially refine our assays, which was difficult originally without a good positive control monoclonal. Hence, we had a dilemma with the interpretation of the data in our 2013 paper not reflecting our current understanding, while at the same time the ELISA data published in the JEM 2013 paper describe exactly the data that were obtained using the protocols described in that publication. Moreover, the reported sequences and mutation rates of the different monoclonals derived from single B cells from RA joints are still valid as published  and this information is, in our opinion, still useful for the scientific community.

We communicated our fault in the SPR assays to JEM editor in September 2018, and we initially suggested to retract only the data related to the SPR assay and not the entire paper. However, as described also in our retraction letter, after a discussion with the JEM editor they suggested to retract the entire paper based on the fact that currently used and more stringent Elisa protocols failed to demonstrate specific reactivity to citrullinated peptides of the monoclonal antibodies described in the 2013 paper.

She also said that other articles from her group are “indirectly affected” by the discovery:

as the monoclonal antibodies from the JEM paper have been used in parallel to polyclonal affinity-purified human antibodies from ACPA-positive RA patients. The polyclonal ACPA have provided the conceptual and major evidence for effects of ACPAs on various cell types.  Since then, we have used the recently produced and published monoclonal ACPAs to verify the functional effects of the polyclonal antibodies (Steen et al, Titcombe et al).

In order to clarify the concepts of effects of polyclonal and well as monoclonal ACPAs and also to describe that the functional effects of the originally described monoclonals from the 2013 JEM paper may not be due to citrulline reactivity, we have notified the editor of the ARD ([Annals of the Rheumatic Diseases] where two of the previous papers were published) with a letter where  the original conceptual findings of functions of ACPAs are further reproduced using the new generation of monoclonal antibodies and where we also comment on and discuss the functional properties of the 2013 JEM monoclonals. This letter is presently under consideration for publication in ARD.

Like Retraction Watch? You can make a tax-deductible contribution to support our growth, 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 team@retractionwatch.com.

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.