Sholto David brings you a new true crime story. This time featuring two twisted orthopaedics professors from the University of Miami! Will they end up in Sholto’s science jail for their fake cancer research?

The perpetrators, Francis J. Hornicek and Zhenfeng Duan, already received a deserved yet somewhat botched retraction at SAGE Publishers, where the retraction notice was reused from an unrelated paper (as I reported in Friday Shorts). Back then, Duan posted his 9-point rebuttal on PubPeer, with points 1 and 9 both demanding that “SAGE’s retraction was based upon erroneous information and should be reversed.”

Still, even if SAGE is occasionally incompetent, they tried to do right, so nothing was reversed, rather a second paper of Duan’s got retracted. This crooked Miami orthopaedist has almost 30 papers on PubPeer, most are co-authored by Hornicek. Who has a determined, but very unimaginative method of defending these fabrications.

---

Miami Vice: Francis Hornicek & Zhenfeng Duan

By Sholto David

Francis Hornicek is an orthopaedic surgeon and academic, he is currently chair of the department of orthopaedics at the University of Miami. Hornicek previously held senior positions at UCLA, and Harvard. His long-time collaborator is the University of Miami professor of orthopaedics Zhenfeng Duan, with the two having published over 100 papers together since 2008. It seems quite plausible that Hornicek is a skilled surgeon, but the research he has published with Zhenfeng Duan has been seriously corrupted by image duplications. For their impressive feats of research malfeasance Duan and Hornicek have already earned themselves a coveted spot in Schneider Shorts, this blog will explore in further detail the “expert” science of Hornicek and Duan.

Problems in Hornicek’s papers were first noted in 2018, with PubPeer contributor Indigofera tanganyikensis identifying duplicated images of ovarian cancer cells in the journal Oncotarget.

Yan Gao, Rosemary Foster, Xiaoqian Yang, Yong Feng, Jacson K. Shen, Henry J. Mankin, Francis J. Hornicek, Mansoor M. Amiji, Zhenfeng Duan Up-regulation of CD44 in the development of metastasis, recurrence and drug resistance of ovarian cancer Oncotarget (2015) doi: 10.18632/oncotarget.3220

Indigofera tanganyikensis also identified a duplicated colony formation assay in a Scientific Reports paper, a correction was subsequently issued by the journal.

Yunfei Liao, Slim Sassi, Stefan Halvorsen, Yong Feng, Jacson Shen, Yan Gao, Gregory Cote, Edwin Choy, David Harmon, Henry Mankin, Francis Hornicek, Zhenfeng Duan Androgen receptor is a potential novel prognostic marker and oncogenic target in osteosarcoma with dependence on CDK11 Scientific Reports (2017) doi: 10.1038/srep43941

Towards the end of 2022 I started a review of Hornicek and Duan’s publications, including an effort to cross-reference images between different papers. This has led to the identification of around 20 more papers with troubling duplications, as well as additional problems in the previously mentioned papers.

Particularly prevalent are republished or overlapping images of histology images with conflicting descriptions. For example, in two similarly titled papers:

• “T-LAK cell-originated protein kinase (TOPK): an emerging prognostic biomarker and therapeutic target in osteosarcoma“
• “T-LAK cell-originated protein kinase (TOPK) is a Novel Prognostic and Therapeutic Target in Chordoma“

The same slide has been used to represent different types of cancer (green squares below). In the chordoma paper there is an additional overlap within Figure 1 (red rectangles).

Pichaya Thanindratarn, Dylan C Dean, Scott D Nelson, Francis J Hornicek, Zhenfeng Duan T-LAK cell-originated protein kinase (TOPK) is a Novel Prognostic and Therapeutic Target in Chordoma Cell Proliferation (2020) doi: 10.1111/cpr.12901

The authors were able to source new images to write an Erratum for one of the above papers, however when asked to provide raw data for cell growth experiments in the same paper Zhenfeng Duan replied:

“the raw data was gone due to a hard drive problem without backup during the Covid-19 era“…

Unfortunate!

Conveniently sourcing replacement images, but not sharing the raw data is a habit. In a figure showing primary and recurrent tumour tissues, the authors inadvertently showed overlapping sections of the same sample (purple rectangles):

Pichaya Thanindratarn, Xiaoyang Li, Dylan C Dean, Scott D Nelson, Francis J Hornicek, Zhenfeng Duan Establishment and Characterization of a Recurrent Osteosarcoma Cell Line: OSA 1777 Journal of Orthopaedic Research (2020) doi: 10.1002/jor.24528

Zhenfeng Duan was able to dig out a new photograph to replace one of these images, but was unable to share the original image in its entirety.

Hornicek’s lab appears to have been a relatively early adopter of the CRISPR-Cas9 technology. In 2015 he published a CRISPR-Cas9 paper in the Journal of Orthopadic research which included a wound healing assay… One of the images was later republished in Oncotarget labelled as a different experimental condition (yellow rectangles). The republished figure included a second overlap (green rectangles):

Yong Feng, Slim Sassi, Jacson K Shen, Xiaoqian Yang, Yan Gao, Eiji Osaka, Jianming Zhang, Shuhua Yang, Cao Yang, Henry J Mankin, Francis J Hornicek, Zhenfeng Duan Targeting CDK11 in osteosarcoma cells using the CRISPR-Cas9 system Journal of Orthopaedic Research (2015)
doi: 10.1002/jor.22745

In 2016 another CRISPR-Cas9 paper was published in Oncotarget, and the images from this paper were later repurposed in Cellular Physiology and Biochemistry. Once again, the paper with recycled images included an additional duplicated element, some of the images below were rotated.

Tang Liu, Zhihong Li, Qing Zhang, Karen De Amorim De Amorim Bernstein, Santiago Lozano-Calderon, Edwin Choy, Francis J. Hornicek, Zhenfeng Duan Targeting ABCB1 (MDR1) in multi-drug resistant osteosarcoma cells using the CRISPR-Cas9 system to reverse drug resistance Oncotarget (2016) doi: 10.18632/oncotarget.13148

The response to PubPeer comments from Zhenfeng Duan and Francis J Hornicek has varied… Zhenfeng Duan has frequently replied and initially seemed willing to correct duplications. Hornicek on the other hand has more recently begun leaving quite belligerent and unhelpful responses on PubPeer.

In reply to the below duplicated panels (identified by imagetwin.ai) Hornicek wrote a series of short comments on PubPeer, including the well-tested classic:

“Conclusions of this peer reviewed manuscript are not altered by the above.“

When pressed on whether the images were duplicated or not, Hornicek seemed unwilling to confirm the obvious, only saying that the data was generated in Japan, and the files were very large. Upon further enquiry Hornicek finally made an insightful comment, perhaps by accident:

“Sometimes images with similarities would require looking at or evaluating original images/data for verification.“

I’m sure most would heartily agree with this statement! But despite having access to the images, and identifying the importance of the raw data for verification, the images have not materialised.

Another unproductive discussion was held regarding an overlap identified in a 2010 paper in Cancer Letters.

Zhenfeng Duan, Diana Ji, Edward J. Weinstein, Xianzhe Liu, Michiro Susa, Edwin Choy, Cao Yang, Henry Mankin, Francis J. Hornicek Lentiviral shRNA screen of human kinases identifies PLK1 as a potential therapeutic target for osteosarcoma Cancer Letters (2010) doi: 10.1016/j.canlet.2010.01.014

Hornicek responds: “Fluorescent photomicrographs – A3 and A4 – are taken from adjacent wells from a 96 well plate.” – Surely he must realize that a) those are tiny cells, he was using a fluorescence microscope and not a magic space-bending device, and b) even then, the adjacent wells are separated by a considerable amount of plastic which would be visible in the picture?

After being challenged on this, Hornicek responded,

“I do not have access to the raw data and I have no way of evaluating what you are saying.”

and then…

“The conclusions of this manuscript have been supported by other peer reviewed publications.”

After further prompting on the integrity of his data Hornicek decided to paste conclusions from seven papers written by other groups that he claims support his work into lengthy unformatted comments on PubPeer. In an amusing (and somewhat predictable) twist, at least two of his chosen references include image problems of their own. In any case, other authors supporting his conclusions obviously doesn’t negate the problems in his own work.

Most other comments on his research have not eliceted such a bungled defence. In reply to overlaps published only last year in Orthopaedic Surgery, Hornicek simply said “The first and last author will check.”

Juncheng Cui, Dylan Dean, Francis J Hornicek, Guoliang Yi, Zhenfeng Duan Expression and Clinical Significance of High-Mobility Group AT-hook 2 (HMGA2) in Osteosarcoma Orthopaedic Surgery (2022) doi: 10.1111/os.13167

Hornicek’s response to the below muddle including three different figures with multiple overlaps was to say: “The last author contacted the first author and both are investigating the above.”

Wenlong Feng, Dylan C. Dean, Francis J. Hornicek, Jinglu Wang , Yanyan Jia, Zhenfeng Duan, Huirong Shi ATR and p-ATR are emerging prognostic biomarkers and DNA damage response targets in ovarian cancer Therapeutic Advances in Medical Oncology (2020) doi: 10.1177/1758835920982853 

But can we trust the authors checking their own work? In December 2022 I identified one overlap in a paper published in Bioscience Reports. The authors checked the images and confirmed the mix up. They then provided a replacement image. However, it would seem that their “check” wasn’t very effective, as imagetwin.ai was able to find two further overlaps in the same figure which I had originally missed…

Tao Guo, Ran Wei, Dylan C. Dean, Francis J. Hornicek, Zhenfeng Duan SMARCB1 expression is a novel diagnostic and prognostic biomarker for osteosarcoma Bioscience Reports (2022) doi: 10.1042/bsr20212446

So how effective was the check by the authors? Surely it should be easier for them since they have access to all of the pictures, file names, metadata, and the folder structure… And yet they failed to spot the additional problems in their own figure even when they apparently looked. Like most responders on PubPeer they react to the problems that are spoon-fed to them, rather than conducting a credible review, never finding further errors on their own.

One publisher deserves credit in this story, Sage Publishing has not allowed Hornicek’s team to publish obvious errors and then “check” their own work.

After comments were left on PubPeer regarding duplicated images of osteosarcoma cells in a 2021 paper, the authors contacted the journal requesting corrections to one image only.

Hangzhan Ma, Dylan C. Dean , Ran Wei , Francis J. Hornicek, Zhenfeng Duan Cyclin-dependent kinase 7 (CDK7) is an emerging prognostic biomarker and therapeutic target in osteosarcoma Therapeutic Advances in Musculoskeletal Disease (2021) doi: 10.1177/1759720×21995069

Sage Publishing conducted an internal review of the paper, and chose to retract it instead, it is worth reposting the retraction notice, as the review was impressively thorough:

“Sage Publishing was contacted by the authors, requesting a replacement of Figure 5C’s U-20S 0 µM treatment control panel. The authors explained that they accidentally inserted the KHOS 0 µM treatment control panel twice in the creation of this figure.

During an internal review of the remaining figures, further concerns were raised, specifically:

• In Figure 2A, lane 2 of the Tubulin panel appears different compared to the background of the other Western blot lanes.
• In Figure 4C, lanes 3,4, and 5 of the CDk7 panel, appear to have different backgrounds compared to the other Western blot lanes.
• In Figure 4C, lanes 4 and 5 of the RNAPII ser5 panel, appear to have an image overlaid.
• In Figure 4D, lane 1 of the CDK7 panel appears to have a background that differs compared to the other Western blot lanes.
• In Figure 5D, lane 5 of the CDK7 panel appears to have an adjusted background compared to the other Western blot lanes.
• In Figure 5D, there is an apparent splice line between lanes 4 and 5 of the Mcl-1 panel
• In Figure 5F, lanes 2 and 3 of the RNAPII panel appear to have a different background compared to the other Western blot lanes.
• In Figure 5F, there appears to be an apparent splice line between lanes 1 and 2 of the Mcl-1 panel.

The authors provided the images underlying these figures for assessment, but Sage Publishing is unable to validate the integrity of these images as they do not meet our criteria of raw, unedited images of the experiments conducted.
This article has been retracted due to uncertainty around the integrity of the original images that call into question the validity of the findings.

The authors did not respond for comment when notified of this retraction.”

Sage Publishing wrote that the authors did not respond to this retraction, although on PubPeer Hornicek said of the western blots in this paper

“There are no problems with the Western blots in Figure 5”

Perhaps he could let the editor know?

Sage Publishing has also retracted another paper for similar reasons, after the authors tried to “correct” the problems in the figure below, the journal found further issues in the western blots and retracted the paper, the authors disagreed with the retraction.

Wenlong Feng, Dylan C. Dean, Francis J. Hornicek, Dimitrios Spentzos, Robert M. Hoffman, Huirong Shi, Zhenfeng Duan Myc is a prognostic biomarker and potential therapeutic target in osteosarcoma Therapeutic Advances in Medical Oncology (2020) doi: 10.1177/1758835920922055

In December 2022 I contacted the university of Miami with my concerns regarding Hornicek and Duan’s research. Jerry Engel a research integrity investigator at the university of Miami has responded, and we have had a reasonably productive exchange of emails. As expected though, the investigation has proceeded at the speed of a university administration, and my impression is that they will only focus on research conducted at Miami, rather than considering the researcher’s behaviour holistically.

Hornicek and Duan did not respond to my email request for a comment to put in this blog, however Hornicek did respond to Leonid Schneider:

Dear Dr. Schneider,

I appreciate your interest.  As you likely know, such matters are managed within institutions in confidential processes.  Accordingly, I have no comment and will allow the appropriate processes to address any concerns raised about my scientific research.  I will only confirm that every conclusion offered in papers on which I am one of the authors is sound and can be relied upon.

Again, thank you for your interest.

Sincerely,

FJH

I added the bold to emphasise Hornicek’s preoccupation with the conclusions of his papers. It’s astonishing that he thinks this is good response, and he has repeated this argument several times on PubPeer. Surely he must understand that conclusions are supported by data. Continuing to insist on the reliablity of his conclusions in the face of serious questions about the data isn’t just unscientific, it’s obviously stupid.

In summary, Francis Hornicek is another high flying academic who lends his name as an author to seriously flawed laboratory research, no doubt it benefits his career substantially, but from his responses on PubPeer it appears as though he may not understand much about the research he “authors” and has a somewhat dubious grasp of the scientific process.

Hornicek frequently published images derived from human patients; rudely dismissing concerns about how poorly this data has been handled is a bad look. Presumably when people consent to their specimens being used for research they expect a high standard of scientific integrity to be applied.

It is worth mentioning that I have not even read probably the majority of papers which Hornicek and Duan have published, I focused only on those which clearly contained lab research based on the title. There may well be more skeletons in the closet!

PS: There is an amusing sidenote to this story. Hornicek and Duan appear to have had their own data stolen by a team of Malaysian researchers.

Two images of cells published by Hornicek and Duan in BMC Cancer in 2009 were subsequently republished in the Hindawi journal BioMed Research International in 2013. There were no common authors on the latter paper, and the labelling of the images conflicted with the original experimental conditions.

I think this sidenote goes some way to showing how common fraud is in biomedical research, even cheaters are having their homework stolen…

Michiro Susa, Arun K Iyer, Keinosuke Ryu, Francis J Hornicek, Henry Mankin, Mansoor M Amiji, Zhenfeng Duan Doxorubicin loaded Polymeric Nanoparticulate Delivery System to overcome drug resistance in osteosarcom BMC Cancer (2009) doi: 10.1186/1471-2407-9-399

Abdullahi Shafiu Kamba , Maznah Ismail , Tengku Azmi Tengku Ibrahim , Zuki Abu Bakar Zakaria A pH-sensitive, biobased calcium carbonate aragonite nanocrystal as a novel anticancer delivery system BioMed Research International (2013) doi: 10.1155/2013/587451

With thanks to other contributors on PubPeer (Indigofera tanganyikensis, Illex illecebrosus, Hoya camphorifolia, Actinopolyspora biskrensis) and imagetwin.ai which spotted several issues that would have been very hard to find otherwise.

I comment on PubPeer as Mycosphaerella Arachidis. I tweet here and make YouTube videos here.

---