Catch and kill: What it’s like to try to get a NEJM paper corrected

Last month,  the New England Journal of Medicine (NEJM) published a letter to the editor and a response reflecting a quite modest correction.  Essentially, the three letters “miR” will be removed from throughout a manuscript as the data, to date, do not support there being a human novel microRNA blood-based biomarker for myocarditis, as the original manuscript claimed.  

At the time of this posting, however, that change – which itself is well over a year in the making – has not yet occurred. And we really don’t understand why. This is our story of the arduous journey to improve the medical and scientific literature.

In May of 2021 the NEJM published “A novel circulating microRNA for the detection of acute myocarditis.” One of us (Marc Halushka), a practicing cardiovascular pathologist and microRNA researcher, recognized this paper was squarely in his wheelhouse.  The concept of a novel microRNA blood-based biomarker was exciting, but also curious. 

After reading the paper, Arun Patil, a postdoctoral fellow in one of our (MH’s) labs, looked for this novel microRNA (hsa-miR-Chr8:96) in vast human sequencing data he was researching ). He didn’t find it. We then brought in Bastian Fromm, a microRNA expert and developer of MirGeneDB, who brought in his collaborators Kevin J. Peterson and Michael Hackenberg for their thoughts.  Within a few days and between our work on expression, structure, and evolutionary aspects, multiple lines of evidence determined there was no hsa-miR-Chr8:96 and no clear link between the author’s mouse and human work.

We were concerned about the damage that could happen if people began studying hsa-miR-Chr8:96 as if it were a microRNA. They might report functionalities that are simultaneously physiologically impossible, but experimentally discoverable due to secondary effects, bad experiments, wishful thinking, and the like. Entire areas of science have been built upon very questionable foundations when the data sound plausible and the implications are desirable.  We all know of manuscripts in our scientific areas that we wish never saw the light of day and pushed our fields into the wrong direction.

Once a quorum of papers exist that build upon each other, it is unlikely that area of dubious scientific inquiry can be silenced.

We thought about our options.  We discussed writing to the author. However, this rarely goes well, and would be unlikely to make the concerns public. We discussed Twitter. A quick way to respond, but ephemeral and too small a scientific audience. It is difficult to tell a cogent story in 280 characters. We discussed PubPeer. A possibility, but it too has a limited audience and comments there would likely have little impact on the persuasive power of a publication in the NEJM. 

We settled on writing a letter to the editor of the NEJM. A letter would forever be linked to the manuscript and could warn researchers of reasons to not pursue science on this non-microRNA. If the journal published such a letter in their typical 3-month window, this response would be the strongest message we could offer. 

We were confident a journal as serious as the NEJM would, upon seeing our concerns, take action. We submitted a letter to the editor on June 4, 2021, 8 days after the original publication date,  providing several lines of evidence about why there is no hsa-miR-Chr8:96. Although the editors we worked with were initially pleasant, progress was slow, and explanations of their deliberations were cryptic.  

However, by October of 2021, the editorial staff of the NEJM decided our letter had merit and was accepted for publication. They had us sign their standard copyright forms for its publication. We expected our letter to be published in November of 2021 –a bit of a delay, but within a reasonable time frame for such a concerning discovery that we had made.

But that didn’t happen.

There is a term in the tabloid media, “catch and kill,” in which a tabloid buys the rights to a story and then chooses not to publish the material to protect the accused. This experience has been quite similar. As with any submission, prior publication elsewhere of the same information would negate publishing in NEJM. Upon signing the NEJM copyright forms, we transferred ownership rights to any text, figures, tables, illustrations, and even our reply to correspondence concerning our contribution. 

That means once a letter has been written to the NEJM (and likely other similar journals have equivalent policies), the critique is essentially muzzled until the journal choses to publish the letter. If they choose not to publish a letter, one is free to do whatever one likes with the information. Instead, we have been silenced for 11 months awaiting the opportunity for the correction to appear in the place where it would have the greatest impact. A lawyer we consulted said that going public before that time carried legal risk, as the NEJM could have enforced the copyright agreement.

Now, our letter and their response letter have appeared. Our brief letter  provides three lines of evidence – a lack of synteny between mouse and human “microRNA” chromosomal segments, an incorrect structure for microRNA processing, and a lack of any small RNA sequencing data that contains the purported biomarker.

The response letter continues to suggest they have discovered a segment of RNA that is “functionally analogous” to a mouse microRNA, yet it does not even share the same critically-important seed sequence of the microRNA. No new data was provided by the original authors in the public correction to refute the significant concerns raised, despite having over a year to bolster their arguments. The NEJM has treated the problem as an error of semantics, rather than the acknowledgment of a failure of peer review. However, the lack of detection of a microRNA in small RNA sequencing reads, as we reported, is unprecedented and, among the other raised concerns, evidence for a serious problem with the manuscript. 

Addressing that would have meant acknowledging that the authors misinterpreted or misunderstood aspects of their work, the reviewers may not have been well-versed in microRNA biology, and the NEJM editorial staff may have been too enamored by the exciting concept of the findings without a good understanding of the data. 

What could NEJM have done differently? From the moment the NEJM editorial staff read our letter, or at least after they vetted our concerns, we believe the proper approach would have been to put an expression of concern on the manuscript right away. An expression of concern would have allowed a public and frank conversation over the merits of the manuscript. This would have alerted others in the medical/scientific community to perhaps take a second look at the data or begin a scientific discussion on the facts in play. We had made strong arguments there was no equivalent human microRNA (or small RNA), but a scientific debate would have been healthy. If we were in error, the expression could be removed with no harm done. 

Instead, without the expression of concern, we believe great harm has occurred to the community, which will be difficult to quantify and correct. The paper has been cited nearly three dozen times, according to Clarivate’s Web of Science, including by society working groups and position papers. How many unfruitful research programs were started to understand the non-microRNA’s biology? How many funding decisions for myocarditis grants were affected by this manuscript in the past year? The story of a microRNA biomarker for myocarditis has been picked up by the medical press and has reached audiences that are unlikely to read the NEJM.  How does this get unwound?

Indeed, just this week, one of the co-authors of the original paper published an important myocarditis review article in NEJM continuing to call the finding a microRNA. If the journal was interested in fixing the record, they missed their first opportunity to refer to it as the corrected hsa-Chr8:96.

The slow movement to correct the scientific literature by journals has a corollary in the legal maxim “justice delayed is justice denied.” Many articles are found to be fraudulent or plagiarized years after their publication, when justice is a stronger motivation for correction. Particularly for established fraud, we have the efforts of Elisabeth Bik, Retraction Watch and others who tirelessly bring attention to scientific misconduct. 

However, for brand new works, whether fraud or just a factual mistake, a lack of swift response opens a Pandora’s Box of problems that cannot be easily remedied after the inaccuracies are allowed to gain traction . It is hard not to find the NEJM complicit in any negative impact this paper has on the diagnosis and treatment of myocarditis, as they were made aware of the paper’s shortcomings a mere 8 days after publication, and continue to support the work despite grave concerns about plausibility.  Journals risk losing credibility when they choose not to act in a timely fashion or be responsive to legitimate concerns as they are only as credible as the papers they publish.

At the end of a year-long saga, we are left wondering if there would have been a better way to provide a faster, more enduring response to the flawed manuscript. We wonder what policies journals can agree on to move with more speed to take appropriate corrective action and alert the scientific community to legitimate concerns raised on their publications.

Marc Halushka is a professor of pathology at Johns Hopkins University and president of the Society for Cardiovascular Pathology. Bastian Fromm is a group leader at UiT, The Arctic University of Norway. Arun Patil is a postdoctoral trainee at Johns Hopkins University.

Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, 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 [email protected].