On 2018 Jan 16, Amanda Capes-Davis commented:

This paper has problems with its authentication testing, resulting in at least three misidentified cell lines (HeLa) being used as models for liver cancer.

The Materials and Methods of the paper state that "All cell lines were regularly authenticated by morphologic observation under microscopy". This is consistent with the policy of the journal, Cancer Research, which strongly encourages authentication testing of cell lines used in its publications.

However, morphologic observation is not a suitable method for authentication testing. Changes in morphology can be subtle and difficult to interpret; cultures can be misidentified before observation begins. To investigate the latter possibility, I examined publicly available datasets of STR genotypes to see if the cell lines listed in the paper are known to be misidentified.

Three of the cell lines used in this paper (Bel-7402, L-02, SMMC-7721) had STR genotypes published by Bian X, 2017 and Huang Y, 2017. All three "liver" cell lines correspond to HeLa and are therefore misidentified.

HeLa and its three misidentified derivatives were used in the majority of figures (Figures 2, 3, 5, and 6). Although the phosphorylation data appear to be unaffected, the conclusions regarding liver cancer metastasis must be re-examined.

What can we learn to improve the validity of our research publications?

For authors and reviewers:

• Morphologic observation is not suitable for authentication testing and, in this case, missed at least three misidentified cell lines.
• STR genotyping is effective for authentication of human cell lines and is the consensus method for comparison of human cell line samples (American Type Culture Collection Standards Development Organization Workgroup ASN-0002., 2010).
• A checklist for cell lines in manuscripts and grant applications is available at http://iclac.org/resources/cell-line-checklist/.

For journal editors and funding bodies:

• Encouragement of authentication testing is a step forward, but is insufficient to stop use of misidentified cell lines.
• Mandatory testing using an accepted method is effective (Fusenig NE, 2017) and would have detected and avoided this problem prior to publication.
• Policy on authentication testing requires oversight and ongoing review in light of such examples. This is important for NIH and other funding bodies who have recently implemented authentication of key resources as part of grant applications.

I am grateful to Rebecca Schweppe, Christopher Korch, Douglas Kniss, and Roland Nardone for their input to this comment and much helpful discussion.

On 2018 Jan 16, Amanda Capes-Davis commented:

This paper has problems with its authentication testing, resulting in at least three misidentified cell lines (HeLa) being used as models for liver cancer.

The Materials and Methods of the paper state that "All cell lines were regularly authenticated by morphologic observation under microscopy". This is consistent with the policy of the journal, Cancer Research, which strongly encourages authentication testing of cell lines used in its publications.

However, morphologic observation is not a suitable method for authentication testing. Changes in morphology can be subtle and difficult to interpret; cultures can be misidentified before observation begins. To investigate the latter possibility, I examined publicly available datasets of STR genotypes to see if the cell lines listed in the paper are known to be misidentified.

Three of the cell lines used in this paper (Bel-7402, L-02, SMMC-7721) had STR genotypes published by Bian X, 2017 and Huang Y, 2017. All three "liver" cell lines correspond to HeLa and are therefore misidentified.

HeLa and its three misidentified derivatives were used in the majority of figures (Figures 2, 3, 5, and 6). Although the phosphorylation data appear to be unaffected, the conclusions regarding liver cancer metastasis must be re-examined.

What can we learn to improve the validity of our research publications?

For authors and reviewers:

• Morphologic observation is not suitable for authentication testing and, in this case, missed at least three misidentified cell lines.
• STR genotyping is effective for authentication of human cell lines and is the consensus method for comparison of human cell line samples (American Type Culture Collection Standards Development Organization Workgroup ASN-0002., 2010).
• A checklist for cell lines in manuscripts and grant applications is available at http://iclac.org/resources/cell-line-checklist/.

For journal editors and funding bodies:

• Encouragement of authentication testing is a step forward, but is insufficient to stop use of misidentified cell lines.
• Mandatory testing using an accepted method is effective (Fusenig NE, 2017) and would have detected and avoided this problem prior to publication.
• Policy on authentication testing requires oversight and ongoing review in light of such examples. This is important for NIH and other funding bodies who have recently implemented authentication of key resources as part of grant applications.

I am grateful to Rebecca Schweppe, Christopher Korch, Douglas Kniss, and Roland Nardone for their input to this comment and much helpful discussion.