Monday, March 8, 2021

They give birth astride a grave, the light gleams an instant, then it's night once more

This post was earlier cross-posted at Leonid Schneider's site, hence the unfrivolous tone. The version there is improved by Leonid's editing and frame story, and cartoon, which I am stealing here.

Please admire these single-use burner Yahoo identities, used once each for submitting a manuscript to the Journal of the Balkan Union of Oncology (J. BUOn), then consigned to the realms of darkness and silence as soon as it was accepted. The list evokes a sense of mortality and the tragic brevity of existence. Their creators have confidence in their capacity to span an endless succession of delightful personae, and "DanyelSsche" is the only one I have seen recycled to submit a second manuscript.

One might expect the editors to wonder why Yi Xiao (for instance) opted to communicate with them through the mouthpiece of "" -- especially when the same author had used a more convincingly institutional e-address for two other papers, immediately before and after in the journal's pagination. On the other hand, more situationally-aware editors would have noticed the same Figures turning up again and again in manuscripts purportedly contributed by otherwise-unrelated research teams (to say nothing of the journal's peer-review corps who were equally oblivious to the repetitions). On the prehensile tail, nor were editorial suspicions aroused by the tsunami of authors asking Santa Claus for "Urgent changes to their publications" for Xmas (replacing one or more bogus Figures with something less obviously fake).

The editors spent the festive season dashing around behind the scenes like one-armed puppeteers, tasked with making Stealth Revisions. For a series of J. BUOn papers had inspired Indigofera Tanganyikensis and Elisabeth Bik and Tabebuia Rosea to comment upon their problematic Figures, at the PubPeer platform for post-publication peer-review. It looks like people read PubPeer (which will be gratifying for the site's founders)... Ninja Corrigenda followed, and so much has been swept under the carpet in the J. BUOn editorial office that it must now resemble a scale model of the Himalayas.

There is nothing in the journal's Tables of Contents to announce that the versions of the papers currently hosted there differ from the versions originally published (and indexed in PubMed), and the only clue is the recency of the Created / Modified field in the PDF metadata. Most of these reconstruction projects had been archived at ResearchGate on behalf of the putative authors, allowing a comparison between iterations and documenting the closeness of the collaboration between the journal editors and their imaginative contributors. As it happens, the editors are already called back behind the scenes for a second round of evidence-erasure, as falsifications come to light that the first round overlooked, or because the hastily-prepared replacement Figures are if anything more bogus than the fakes they replaced. I'll come back to that.

Right now, welcome to the first papermill of 2021! Yes, there is an on-line spreadsheet. I propose to approach the topic indirectly, circuitously (as is the custom of my people). A second introductory theme is 'hipster ethnopharmacology'. We have already met the first related theme: Yahoo e-addresses and their tragic ephemerality. These provide a useful clue to provenance, by the way: a paper might be papermill-sourced even if the corresponding e-address is plausibly related to the nominal author's name, but if the address is (say) "", then the real author is seeking financial compensation rather than recognition.

The academic speciality of ethopharmacology is flourishing, if the multitude of journals serving it are any guide. If a plant has any place in any folk pharmacopoeia, sooner or later it will take its turn under the microscope (or more accurately, inside the chromatograph), with eager researchers isolating and characterising its active ingredients. Or failing that, isolating any secondary metabolites distinguishing it from other plants. For evolution, and the advantage that comes in the evolutionary race from not being eaten, has taught plants to deter potential consumers by packing their leaves and fruit and roots with molecules that are sand in the metabolic machinery of herbivorous mammals (and birds, insects, molluscs). The corollary follows (as corollaries are wont to do) that these secondary metabolites must have some therapeutic applications. It remain, then, to test each molecule in vitro or in animal models of all possible medical conditions (since folk-medicine indicative categories like "excess black bile" or "imbalance of qi energy" are not as helpful as one might think).

Oncological candidates are especially helpful, for it goes without saying that phytochemicals designed to deter animal consumption will also cripple the previously-unchecked proliferation of cancer cells. The industry has not yet exhausted all combinations of "phytochemical X" and "vat-grown cancer cell-line Y". Perhaps the papermills can help!

Here "X" comes from a chemical cavalcade including Ambrosin, Baicalein, chrysoeriol, davanone, Eriodictyol, flavokawain-B, Geraniol, helenalin, Imperatorin, Juglone, Kaempferol, Lycopsamine, α-mangostin, Murrayanine, Norwogonin, Oridonin, Pleuromutilin, Royleanone, Sugiol, Tangeretin, Ursolic acid, Voacangine, Wogonin, Xanthohumol and Zingiberene. These are variously classified as chalcones, diterpenes, deterpenoids, flavones, flavonoids, sesquiterpene lactones, terpenoids etc., for it is not enough just to be an 'alkaloid' these days. Coca-coline and Yoyodeine are not on the list, as researchers collectively lacked the commercial nous to sell naming rights for this molecular abecedarium.

If you are like me, you want to know which secondary metabolites are popular and which are passé, and which ones to learn about now so you can boast that you were studying them before they sold out and went mass-market. No-one wants to be in the position of a mystery novelist whose characters are still poisoning with strychnine when all the hip authors have moved on to ricin. It turns out that there is a special place in the cavalcade for Daidzein, with one specific paper accruing 71 citations:

Some of them are systematic reviews, or an inclusion in some Introductory Context section padded out with vaguely relevant stuff where an author had time for a G**gle Search but not enough time to read the hits. But there are so far 50 papers in my spreadsheet citing Hua et al (2018). Sometimes to exemplify concepts and methods which Hua et al never mentioned -- the quantification of RNA with PCR; or ROS and mitochondrial membrane potential levels. They did mention cell-cycle arrest and quantifying apoptosis with flow-cytometry, but hardly as pioneers, so it is remarkable that so many authors independently chose to use them as the reference description. In a standardised way. Often involving covers-lips.

It's conceivable that a computer virus inserted itself into 50 authors' word-processing software and copy-pasted certain paragraphs into whatever manuscripts were half-written at the time. We should also consider the alternative explanation, that a papermill is churning out the product by sticking to templates, changing the phrases just enough to avoid triggering automated plagiarism detectors.

It is hard to choose where to begin with this sprawling continuum, where shared graphic elements link any paper to many other papers. Western-Blot bands change protein ID and lane conditions to turn up in another WB. Flow-cytometry histograms are modular; the peaks can be taken apart and reassembled in new arrangements to demonstrate entirely different claims. Flow-cytometry scatter-plots gain or lose points. Fluorescent-stained panoramas of cell cultures, color-coded by their post-treatment viability, yield enlarged details to illustrate different cell-lines and different treatments.

Aphidicolin, Artemisinin: surprisingly similar effects
It is a kaleidoscopic phantasmagoria, and I need to be reminded that "however entrancing it is to wander unchecked through a garden of bright images, are we not enticing your mind from another subject of almost equal importance?" To remove this paralyzing excess of choices, I will pick the next few cherries from the narrow focus of the 50 papers of the Daidzein corpus.

Carvone, Daidzein: surprisingly similar effects

The Daidzein subgenre also reminds us that these molecule / cancer-type / cell-signaling-pathway MadLibs do get cited (with the usual caveat that citations do not imply a readership, when it is enough for a title and abstract to show up in authors' searches for corroborative detail to lend verisimilitude to their otherwise bald and unconvincing narratives). The 71 citations for Hua et al. are an outlier, riding on the coat-tails of a template, but "Baicalein induces apoptosis in esophageal squamous cell carcinoma cells through modulation of the PI3K/Akt pathway", "Anticancer and apoptotic activities of oleanolic acid are mediated through cell cycle arrest and disruption of mitochondrial membrane potential in HepG2 human hepatocellular carcinoma cells" and "Genistein induced anticancer effects on pancreatic cancer cell lines involves mitochondrial apoptosis, G 0 /G 1 cell cycle arrest and regulation of STAT3 signalling pathway" (respectively accruing 67, 56 and 39 citations) hit the same sweet-spot of clickbait citability.

That said, J.BUOn is hardly a high-impact-factor outlet (Impact Factor = 1.695). If the present papermillers confined themselves to its pages, they would lose their already-limited interest to readers. However, they are naturally ambitious to extend further afield. We find more of the implausible Yahoo identities in Medical Science Monitor (33 papers in the spreadsheet), and in Archives of Medical Science (17 papers). I know nothing of these journals except that they're not dominated by papermill persiflage... it may be that a systematic search would reveal many more Easter Eggs. AMS is from a Polish publisher and has a slow-moving pipeline, in which a paper that was accepted in 2017 might be delayed until there is space for it in a 2019 issue, so its pages are time-capsules of the papermill's stylistic evolution from a few years earlier. Also 15 papers spread across various Elsevier journals (notably Archives of Biochemistry and Biophysics and Biochimie); five in AMB Express, from Springer; and six in the egregious European Review for Medical and Pharmacological Sciences.

I must not overlook the 18 productions which found their way into journals from Spandidos Publications (Oncology Reports, Oncology Letters and Molecular Medicine Reports), or the nine into journals from e-Century (International Journal of Clinical and Experimental Pathology and American Journal of Translational Research). There are undoubtedly more. In fact the dates suggest that the scriveners began their depredations there, with J.BUOn as only a secondary host. 'E-Century' is located several sub-barrels below the bottom of the barrel, but is still subsidised by the US government which archives their contents and lists them in PubMed. IJCEP and AJTR are 'International' and 'American' in the sense that the proprietor does indeed reside in Wisconsin, thereby fulfilling the requirement levied on Chinese authors that they publish in a non-China journal.

As for the Spandidos press, a fall-from-grace narrative is waiting to be told there. The eponymous founder Demetrios Spandidos was widely seen as a contender for a Nobel Prize until someone asked to see the receipts for all the laboratory rodents on which he conducted his purported experiments... everything went tits-up and he turned to predatory publishing instead. Episodes of that tale are scattered across footnotes to other people's stories and this is not the place to bring them all together.

In those early manifestations, the Photoshopping was clumsy and the contempt for the audience was poorly-concealed. The results were homages... to Monet waterlilies (h/t Elisabeth Bik), or to the classic Space Invaders arcade game, with serried ranks of cell-cluster sprites.

I hasten to add that appearance in J.BUOn is not definitive proof that a faked paper came from our current papermill. The attribution of some entries in my spreadsheet could be false. It is not as if our subjects held territorial rights over J.BUOn to stop other fantasists from using its services as well (one entry seems to be from the 'Contractor' papermill). Nor is there anything, other than stringent peer-review, to stop authors of genuine papers from making innocent mistakes like using the same Figure twice (not overlapping with other papers with different authorship). Even with that caveat, though, there are at least 250 of these mock-science fantasies (and the end is nowhere in sight), collecting citations and competing with real cancer-research findings.

The Yahoo email instantiations are dispositive, but not universally present. Their use did not become standard until 2018, so the attribution of earlier papers is more speculative. There is stylistic continuity and recycled images although the Space-Invader compositions are more skillful in later examples.

It is anyone's guess whether the papermillers have a research day-job, and came by their own Figures, or just found them elsewhere in the literature, insufficiently fastened down. In one case they appropriated Western Blots that are clearly fakes from the 'tadpole' mill, which aroused Indigofera's suspicions and set off this whole line of inquiry.

Our anonymous scriveners are most comfortable composing fairy-tales about apoptosis pathways to cancer-cell destruction -- to be more accurate, about autophagy pathways. Of course this does not narrow the scope enough to dox them. Autophageosomes -- one of the Four Horsemen of the Apoptocalpyse -- occupy a central place in their enhanced and recycled Transmission Electron Microphotographs.

The Apoptocalypse Horsemen include fluorescent stains, lighting up cells blue or green or yellow or red according to their viability or stage of self-destruction.

To be fair, the scriveners are willing to step outside their comfort zones, and on reasonable request they turn their talent for bricolage to the micro-RNA genre of fiction instead.

Now I am not saying that J.BUOn is a knowingly predatory journal, in Jeffrey Beall's sense. I am confident that when oncologists in the Balkans decided to unionise and to found a scholarly organ, they had benign ambitions and no long-term intention of creating a conduit by which papermills could disguise their MadLibs-composed cancer-cure CV-stuffers behind a veneer of academic approval. This is probably something they drifted into, as the journal was hollowed out by the Invisible Hand of the Market wielding an equally-invisible melon scoop. Along the way they collected a corps of reliably-complaisant peer reviewers, all members of the Totem of the Rubber Stamp, though they may or may not be more sock-puppets. And now J.BUOn rings the changes on the phytochemical novelties, pairing each one with different cell-lines of domesticated carcinoma (always with promising results to report, warranting further study).

The PubPeer back-end automagically sends authors invitations to join the discussions, and some have replied. Some ascribe the recycling or manipulation of images to poorly-mentored students, and promise retractions to come.

#7 Mingjun Sun 
#4 Ronghuan Wu 
#10 Xiaobin Li 
#8 Ke Zhang 
Alas, although the editors can create and upload a new version of a paper within three days of PubPeer comments about the evidence of papermilling, authors' requests for withdrawal go unheeded for weeks and months. Their correction requests are equally delayed. It's as if the revisions are coming from inside the house!

#2 Ruixin Lin 
Some (notably Fa Hua, Qing-Gang Meng and ShuLi Yang) blamed piracy by rival authors. They have my sympathy; many researchers have neither qualm nor scruple, while many journals -- even the ones with the PubMed imprimatur -- are wretched hives of scum and villainy (we should note that the minutiae of publication dates are such that some of the plagiarism must have occurred even before the manuscripts were submitted).
#2 H He 
Songqing He 
#3 Shuli Yang 
[Tan et al (2019)]

Zhang et al (2018) was pre-emptively pillaged repeatedly. By (at right) Shi et al (2019)...

By Pan & Zhang (2019) (at left) and Luo et al (2019) (at right)...

By Pan & Zhang again, and by Wu et al (2018), and by Chen et al (2019)...

By Zhang, Zhang & Ni (2020), Xiong et al (2019) and Huang et al (2017).

#5 Qing-gang Meng 
We learn elsewhere that Meng fell out of favour, "suspected of serious violations of discipline and law, and is currently undergoing disciplinary review and supervision and investigation". G**gle Translate charmingly renders the Chinese headline as "Meng Qinggang, the former president of Harbin First Hospital, had an accident", but I suspect that "something unfortunate happened to him" might be more accurate [h/t TigerBB8].

Fa Hua is nominally the author of that seminal, citable work on Daidzein.

#2 Fu Hua 
So there are repeat customers, who came back for a second serving... a tribute to the high quality of the papermill's products, or their bargain price. ShuLi Yang is in the list six times. Yi Xiao, thrice. A group from Fudan University Shanghai Cancer Center around Yiqun Du have published six times (thrice as DanyelSsche). One paper has been retracted, for they are not fooling around at Medical Science Monitor, while the fate of two others depends on editorial caprice at Archives of Medical Science and AMB Express, but J.BUOn editors have accepted amendments for the other three.

Coming back to the clandestine corrections, Wei et al (2019) and then Lu et al (2019) needed a double dose, because suspicion fell upon figures left unchanged in the first round. This is what happens when patches are prepared in haste. Other patches had their own dumbprints of fabrication and will need remediation of their own.

In some recent corrigenda, the editors replaced all the original figures, which I take as a tacit admission that the scriveners couldn't remember which ones (if any) were legitimate. If only there were a way that journal editors could signal to readers that the results originally presented in a paper were all fictitious... we could call it (just spit-balling here) "Retraction".

Since the papermiller generated e-addresses for "corresponding author", presumably they also handle the correspondence, including the invitations from PubPeer. It may be that requests for revisions of J.BUOn material generally come from them. In the absence of acknowledgement of the corrections, let alone explanations, we can only speculate.

I am advised by people who contacted the J.BUOn office that the editors present themselves as the real victims here. Nefarious papermillers exploited the journal, taking advantage of loopholes in the system for peer-review, but these have been closed to prevent any recurrence. The credibility of this claim of victim status was undermined by the subsequent stealth amendments, working to improve the plausibility of the impostures rather than retracting them. It is almost as if the editors' real priority is to reward the papermill as a loyal customer.
Dear dr. XXXXXX,
Thank you for your recent informative email. We read very carefully your observations and we agree with them. Quite recently we discovered that we were victims of a number of Chinese authors who submitted their falsified articles to our journal. Other people (pubpeer) have also informed us about this situation. We now use advanced programs to locate and reject such articles.

Best wishes and happier new year,
Nasis Athanasiou
Literature scholars have the useful term 'ekphrasis', meaning 'a detailed, admiring description of an artwork imagined by the author'. The ultimate exercise in ekphrastics is the Hyperotomachia Poliphili, 400 pages of ecstatic raptures over one imagined sculpture or monument after another. Encountering the phenomenon of papermills opened my eyes to the possibility that ekphrasis could also come to dominate science, with a huge literature of descriptions of experiments so beautiful that they should have happened (though they didn't).

I leave you with some final examples of the editorial repairs.

No comments: