A follower on Twitter asked me to look at two identical papers. I agreed that they looked very similar, did some searches, and found six more. All eight papers presented the same survival curves, table values, and similar line graphs. But they were published in different journals by different authors, at different institutes, on different patients, and different cancers.
In this blog post, I present to you the Mysterious Case of the Octopaper.
My investigation started with a tweet by Nirmalya Saha, who asked me to look at two identical papers, PMID: 30305431 (Liu et al., Open Biology, 2018) and PMID: 30117667 (Cai et al., J Cell Mol Med, 2018)
Of course, I was intrigued, so I compared the two papers to each other. They were from different authors at different hospitals, published in different journals, and although both groups studied miR-125a-5p expression, they did so in different types of cancer (gastric vs lung cancer).
Two papers with very similar text…
The text of the two papers, in particular the abstracts, Methods, Results, and Discussion sections were very similar to each other, as the SimTexter analysis below showed. Blocks of text in the same color show verbatim textual similarities. Only parts of the introduction and patient description were significantly different, which is not surprising because these papers described very different types of cancers.
… and nearly identical figures
The figure similarities were also quite striking. Here are both Figures 1 compared to each other, with Cai et al. shown on the left, and Liu et al on the right. Even though two very different patient groups with very different types of cancer were studied, they showed a remarkably similar survival (panels B) and microRNA expression and methylation levels (panels D).
Here are both Figures 2. The layout is somewhat different, and the Cai paper has two extra photos, but what a remarkable similar set of results!
Figures 3 were also very similar, with some shuffling of the Western blots.
Figures 4: spot the differences.
Figures 5: again, we see the same line graphs, although now the Liu paper has been decorated with some images of mice with tumors.
Same table values, too
Tables 2 and 3 from both papers are remarkably similar as well. The risk ratios for survival of either lung or gastric cancer are …. identical – at least if we believe these tables.
But wait – there is more
If two papers are nearly identical, there might be more. I first thought that one paper might have copied/pasted data and graphs from the other paper, but the papers were submitted pretty close to each other in time. Because of our previous work on the Tadpole Paper mill, I was starting to suspect that both papers might be the products of the same paper mill, a company that sells fabricated papers in English to doctors in China who need a paper for their promotion.
So I sailed out my data detective boat onto the big internet sea to find more papers. Based on my experience, it is hard to do successful reversed image Google searches using panels from scientific papers, so I used a different strategy. I looked for the values listed in tables 2 and 3 in Google Scholar.
This approach worked. I found six more papers that had the values that the Cai and Liu papers had in common in Tables 2 and 3. A full list of the combined eight papers is given below. All papers were published between 2017 and 2020, in five different journals, by non-overlapping authors from seven different hospitals in different cities in China, about different sets of patients with different types of cancers.
Although the figure and textual similarities were not always as striking as with the first two papers, all eight papers had very similar or identical Kaplan-Meier survival curves:
Similarly, all eight papers had identical or very similar data in tables showing the univariate and multivariate “hazards models” (risk factors for long term cancer survival).
Could this have been caught during peer review?
Although the first two papers were textually pretty similar, this would have been very hard to have been caught with a plagiarism checker. One paper was in peer review but not yet accepted or published when the second paper was submitted for peer review. The other papers are textually less similar. Most importantly, each of these papers by themselves looks legit. It took a Google Scholar search for table values to detect that all eight papers are connected.
My current suspicion is that all eight papers were generated by the same paper mill company. The company might have taken data from a real paper and adapted it to a different patient set, microRNA, or cancer type. For example, the data values all appear similar to that of an older paper, from 2015 (Zhang et al. Journal of Experimental & Clinical Cancer Research (2015), DOI 10.1186/s13046-015-0176-z), which might be the original.
In most cases, they mixed in some different figures or variations, but in the Cai and Liu papers, the company might have sold a two nearly identical copies by accident to two different group of authors.
And who knows, there might be more sets of these. Let me know below in the comments if you find another one!
List of the eight papers
|Mingzhi Cai, Qiuxian Chen, Juntao Shen, Chenbing Lv, Lisheng Cai||Epigenetic silenced miR-125a-5p could be self-activated through targeting Suv39H1 in gastric cancer||J Cell Mol Med. 2018;22:4721–4731||10.1111/jcmm.13716||Zhangzhou Affiliated Hospital of Fujian Medical University, Zhangzhou City, China||https://pubpeer.com/publications/39A1613F4546DA16064BA441B29A0F|
|Hongxu Liu, Yegang Ma, Changhao Liu, Pengfei Li, Tao Yu||Reduced miR-125a-5p level in non-small- cell lung cancer is associated with tumour progression||Open Biol. (2018) 8: 180118||10.1098/rsob.180118||Cancer Hospital of China Medical University, Shenyang, Liaoning Province, China||https://pubpeer.com/publications/5C4CE1170930C517CB660480C85DA9|
|Xiuli Wang, Zenghui Li, Beihua Kong, Chen Song, Jianglin Cong, Jianqing Hou, Shaoguang Wang||Reduced m6A mRNA methylation is correlated with the progression of human cervical cancer||Oncotarget, 2017, Vol. 8, (No. 58), pp: 98918-98930||10.18632/oncotarget.22041||Qilu Hospital, Shandong University, Jinan, Shandong, China||https://pubpeer.com/publications/71BD811C298523B3DA169B33990572|
|Xiaofei Yan, Jian Zhao, Rui Zhang||Interleukin-37 mediates the antitumor activity in colon cancer through β-catenin suppression||Oncotarget, 2017, Vol. 8, (No. 30), pp: 49064-49075||10.18632/oncotarget.17093||Liaoning Cancer Hospital & Institute, Cancer Hospital of China Medical University, Shenyang, Liaoning Province||https://pubpeer.com/publications/BCC3ABDF2CDE9AE0DA9EED65E7E848|
|Feng Wang, Weihua Zhang, Tianfeng Wu, Heying Chu||Reduced interleukin-38 in non-small cell lung cancer is associated with tumour progression||Open Biol. (2018) 8: 180132||10.1098/rsob.180132||The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China||https://pubpeer.com/publications/EF2A21C6D3071CF820BDF72AEA23D6|
|Jian Zhang, Tao Mao, Shuyun Wang, Dongsheng Wang, Zhaojian Niu, Zhenqing Sun, Jianli Zhang||Interleukin-35 expression is associated with colon cancer progression||Oncotarget, 2017, Vol. 8, (No. 42), pp: 71563-71573||10.18632/oncotarget.17751||The Affiliated Hospital of Qingdao University, Qingdao, China||https://pubpeer.com/publications/440CE72BA5492F9C5D713FDB585913|
|Kuangkuang Zhu, Dong Sun, Xiaoqin Zou, Ruixia Liu, Zhen Wan||Interleukin-36 receptor antagonist is associated with the progression of renal cell carcinoma||International Immunopharmacology 84 (2020) 106474||10.1016/j.intimp.2020.106474||Affiliated Suzhou Hospital of Nanjing Medical University, Suzhou, Jiangsu, China||https://pubpeer.com/publications/BAF4E0274D18F7F8DA25734CBF9053|
|Mingfei Sun, Xianjie Zheng, Qingjiang Meng, Yanjun Dong, Guoyu Zhang, Dexin Rao, Xiaokang An, Zhongxin Yang, Lihong Pan, Shuanglin Zhang||Interleukin-35 Expression in Non-Small Cell Lung Cancer is Associated with Tumor Progression||Cell Physiol Biochem 2018;51:1839-1851||10.1159/000495706||The First Affiliated Hospital of Henan University, Kaifeng City, Henan Province, China||https://pubpeer.com/publications/46F59FBF8E26449CA9F87868812E08|