Retraction, tell-all style, for breast cancer radiology paper

Here at Retraction Watch, we don’t believe in the expression “TMI.” But this case features a level of detail we’re not sure we’ve seen before.

Acta Radiologica has pulled a 2012 article on breast cancer imaging for being a duplicate publication — a sin the retraction notice takes great pains to point out.

The notice, written by journal editor Arnulf Skjennald, has the blow-by-blow feel of a police report:

The authors of article “Contrast-enhanced US quantitatively detects changes of tumor perfusion in a murine breast cancer model during adriamycin chemotherapy” submitted this article for publication in Acta Radiologica on 17 October 2012. After going through peer review, the article was accepted for publication on 24 March 2013 and published online on 12 May 2013 with DOI: 10.1177/0284185113488019, and thereafter published in print (1). The following mail was received from the corresponding author on 6 November 2013: “Soon after online publication of our paper entitled ‘Contrast-enhanced US quantitatively detects changes of tumor perfusion in a murine breast cancer model during adriamycin chemotherapy’ in Acta Radiol (2013 Oct 1; 54(8):882–888), we became aware that part of the data (tumor volume and pathologic results) has been published in J Ultrasound Med (Wang JW, Zheng W, Chen Y, et al. Quantitative assessment of tumor blood flow changes in a murine breast cancer model after adriamycin chemotherapy using contrast-enhanced destruction-replenishment sonography. 2013 Apr;32(4):683–690). I don’t know whether this would be partial duplicate publication”. This manuscript was submitted to J Ultrasound Med on 22 May 2012 and accepted on 13 September 2013 (2).

After an evaluation of the two articles we came to the conclusion that the manuscript submitted to Acta Radiologica had to be regarded as a duplicate publication. The authors of the two articles belong to the same research group. We then sent this conclusion to the corresponding author and received the following response:

Currently, there are two different means to measure tissue perfusion based on contrast-enhanced ultrasound: one is bolus kinetics (used in the paper published in Acta Radiol), and the other one is destruction-replenishment kinetics (used in the paper published in J Ultrasound Med). The bolus kinetics is simple and easy to manipulate, and extensively used in assessing tumor response to treatment, however only semi-quantitative parameters (such as peak enhancement, area under the curve, and so on) could be obtained. On the other hand, the destruction-replenishment kinetics could be used to calculate quantitative parameters (blood flow) and mostly used in assessing brain and myocardium perfusion. Theoretically, destruction-replenishment kinetics would be more accurately in calculating tissue perfusion than bolus kinetics as quantitative parameters could be acquired. However, the setting of destruction-replenishment kinetics is more complicate, and not easy to manipulate in clinical studies. The purpose of our study was to determine these two different ultrasound means in assessing tumor response to adriamycin, respectively, however same mice samples were used because we don’t want to waste more mice life. The experiment of destruction-replenishment kinetics was tested firstly after administration of contrast agent, and about 1 hour later the experiment of bolus kinetics was tested using different ultrasound machine settings and another dose of contrast agent was administrated. During the second study, one mouse died during the experiment. Therefore, the ultrasound data were totally different in these two papers due to different theory and method used in the two studies although the supporting data (tumor volume and pathologic results) were similar. Furthermore, the supporting data were just used to prove that the treatment of adriamycin was effective. Very regretfully, there is no reference to the article published in J Ultrasound Med about the using of supporting data (tumor volume and pathologic results) in the paper published in Acta Radiol due to the short publishing interval between the two papers. (When the paper in Acta Radiol was accepted for publication on 24 March 2013, the article in J Ultrasound Med had not been published online. So we didn’t know how to cite the article of J Ultrasound Med at that time).

After this statement the section editor of our journal made a new evaluation of the two articles and came up with the following conclusion:

These two papers are based on the exactly same study material and the presented results are from the exactly same experimental study.

The main aim of the study was the same for both papers:

• J Ultrasound Med (first paragraph of Discussion): “This study was conducted to assess quantification of tumor blood flow changes with contrast-enhanced destruction-replenishment sonography in the evaluation of conventional chemotherapy in a murine breast cancer model and to compare tumor volume, tumor cell density, and microvascular density.”

• Acta Radiol (first paragraph of Discussion): “The present study was conducted to assess quantifying tumor perfusion changes with contrast-enhanced ultrasound in the evaluation of conventional chemotherapy in a murine breast cancer model by using tumor volume and tumor cell density as gold standards.”

The main results from the two publications (as presented in the first paragraph of Discussion):

• J Ultrasound Med: “Our results showed that treatment with Adriamycin significantly inhibited the growth of the MCF7 breast cancer cells compared with control tumors.”

• Acta Radiol: “Our results showed that treatment with Adriamycin significantly inhibited the growth of the MCF7 breast cancer as compared with control tumors.”

The Introduction, Material and Methods, Results, and Discussion are very similar in the two publications.

Furthermore:

• The graph used in Fig. 3 in both papers is the same (the columns for “Control” and “Treated” just changed positions as first and second, respectively).

• The graph used in Fig. 4 is the same in both publications (again, the authors have only changed the position for “Control” and “Treated”, respectively).

• The immunohistochemical Fig. 4c is the same in both papers.

In my view, the “minor” difference of the ultrasound parameters applied in the two papers for presenting the results cannot justify the submission of two papers. There are obviously differences between “bolus kinetics” and “destruction-replenishment kinetics”, but if this was an important issue we should have expected that the presented figures and Discussion would have been very different in the two papers. The different parameters and results (values) presented in Table 1 of the two publication do not prevent the impression of a “duplicate publication”.

It must have been quite obvious to the authors at submission of the manuscript to Acta Radiologica that the study design, the Material and Methods, the Discussion, and the aim and main results of the two papers would be regarded as duplicate publication. It is remarkable that the authors not at all did mention the submission of their manuscript to J Ultrasound Med when they submitted the paper to Acta Radiologica.

After that conclusion the following message was received from the authors:

Soon after online publication of our paper in Acta Radiol (2013 Oct 1; 54(8):882–888), we became aware that part of the supporting data (tumor volume and cell density) have been published in J Ultrasound Med 2013 (Apr;32(4):683–690). In light of this, we are requesting to retract the article published in Acta Radiol although the purpose, theory, methods and ultrasound data behind the two papers were totally different, and all of us have never realized the possibility of partial data duplicated publication in beforehand. We deeply apologize for any inconvenience that we may have caused.

The article is hereby officially retracted.

The paper has yet to be cited, according to Thomson Scientific’s Web of Knowledge.