Molecular Differences between Squamous Cell Carcinoma and Adenocarcinoma Cervical Cancer Subtypes: Potential Prognostic Biomarkers

Round 1

Reviewer 1 Report

The authors present potentially very interesting data. The attempt to distinguish between squamous cell carcinoma (CC) and adenocarcinoma of the cervix on the basis of gene expression profile should be welcomed. Three different, unequal databases were used to obtain this data. The results obtained provide insights into the biological signaling pathways that are activated in both subtypes. However, the study has several serious flaws: misleading statements, incomplete presentation of results and poorly structured discussion.   1) Already in the introduction the readers are confronted with a curious statement, "according to the National Comprehensive Cancer Network (NCCN) guidelines, the treatment for both is based on concurrent chemoradiotherapy, and the survival outcomes remain uncertain [8]". The treatment cornerstone of CC is surgery, in the (most frequently diagnosed) early stages without any adjuvant treatment. The randomly chosen reference (which are not the NCCN guidelines) is misleadingly cited. 2) Similarly, the role of the HPV in the pathogenesis of CC is mentioned superficially and reduced to the two main HPV types. 3) Accordingly, the HPV infection is included into the analysis in a very simplified way, without any differentiation between subtypes. Most important, some critical information is missing, e.g. if the author mean the "HPV infection" as the current infection or HPV infection diagnosed in the past; or if the authors mean any HPV infection or (current or previous) infection with high risk HPV. 4) The quality of survival data is limited, as frankly stated by the authors at the end of the study. 5) Although data regarding disease stage at diagnosis are available, information is lacking as to whether distinct molecular profiles were associated with early versus advanced stage at first diagnosis. 6) The discussion contains a lot of cited studies, however this part of the manuscript would benefit from more structure and clarity. Ideally either summarizing of the main findings as a figure, a table or at least subheadings.   Some minor comments: a) e.g. abbreviations introduced without full name, b) unusual citation format within the text (provided ere as "Name" + "Initial"  + "and colleagues" instead of "Name et al.", c) exact data including p values in the introduction.

Author Response

The authors present potentially very interesting data. The attempt to distinguish between squamous cell carcinoma (CC) and adenocarcinoma of the cervix on the basis of gene expression profile should be welcomed. Three different, unequal databases were used to obtain this data. The results obtained provide insights into the biological signaling pathways that are activated in both subtypes. However, the study has several serious flaws: misleading statements, incomplete presentation of results and poorly structured discussion.  

1) Already in the introduction the readers are confronted with a curious statement, "according to the National Comprehensive Cancer Network (NCCN) guidelines, the treatment for both is based on concurrent chemoradiotherapy, and the survival outcomes remain uncertain [8]". The treatment cornerstone of CC is surgery, in the (most frequently diagnosed) early stages without any adjuvant treatment. The randomly chosen reference (which are not the NCCN guidelines) is misleadingly cited.

Thanks for your comment, we modified the main text and we updated the reference.

“These histologic types are relevant in terms of patient prognosis; according to National Comprehensive Cancer Network (NCCN) guidelines, both are usually treated similarly, i.e., treatment is based on surgery for early disease and chemoradiotherapy for advanced disease, with survival outcomes for both histologic types being uncertain (8)”.

 

2) Similarly, the role of the HPV in the pathogenesis of CC is mentioned superficially and reduced to the two main HPV types.

Thanks for your observation. We add more high-risk HPV subtypes and add new reference.

 

3) Accordingly, the HPV infection is included into the analysis in a very simplified way, without any differentiation between subtypes. Most important, some critical information is missing, e.g. if the author mean the "HPV infection" as the current infection or HPV infection diagnosed in the past; or if the authors mean any HPV infection or (current or previous) infection with high risk HPV.

The HPV data from TCGA patients do not specify whether they are infections at diagnosis or a current infection. But data from the Mexican-mestizo cohort and the validation cohort, HPV infection was detected at diagnosis. This was clarified in supplementary table S2.

 

 

4) The quality of survival data is limited, as frankly stated by the authors at the end of the study.

Thank you for your comment. In this regard, we clarify in the materials and methods section that "OS was defined as the time from diagnosis to date of death or last contact".  Also, it is important to clarify that for the OS analyses, we have complete 5-year follow-up information for the Mexican-mestizo cohort and the Mexican independent cohort. However, for the TCGA patients, we used only the data that are available, since follow-up information is not available for all patients.

 

5) Although data regarding disease stage at diagnosis are available, information is lacking as to whether distinct molecular profiles were associated with early versus advanced stage at first diagnosis.

Thank you for your observation, it is very interesting but the objective of the study was to find a molecular profile that distinguishes between histologic subtypes (ADC vs. SCC) regardless of clinical stage. Also, as shown in table 1, early vs advanced stages are more frequent.

 

6) The discussion contains a lot of cited studies, however this part of the manuscript would benefit from more structure and clarity. Ideally either summarizing of the main findings as a figure, a table or at least subheadings.  

Thank you for your comment, in this regard we have improved the discussion by including references from the introduction.

 

Some minor comments:

  a)e.g. abbreviations introduced without full name,

No problem, we have corrected it

1. b) unusual citation format within the text (provided ere as "Name" + "Initial"  + "and colleagues" instead of "Name et al."

No problem, we have corrected it

1. c) exact data including p values in the introduction.

The exact p data are relevant and we consider it important to show them in the introduction to make clear the statistical significance between subtypes.

Reviewer 2 Report

Please read the attached file.

Comments for author File: Comments.pdf

Author Response

Thank you for the opportunity to read and evaluate this work. Researchers present an interesting and important issue, which aims to assess the molecular differentiation of the types of cervical cancer. I believe that with a few corrections, this article will have scientific value for many readers.

 

1. Please write the citations in the recommended format [1,2,3] instead of [1], [2], [3],.

Thank you, the reference format was corrected

 

2. Please include more current literature in references. Apart from three works from 2020, the rest are older

Excellent, we have added some new references  

 

3. Please improve the English language

The manuscript was edited by a native English speaker, please find attached a copy of the article.

 

4. Please correctly translate the abbreviations, e.g. HR-HPV is high-risk human papillomavirus and not high-risk papillomavirus

An apology, the translation has been corrected

 

5. Please write p (p-value) correctly and consistently throughout the work

We have corrected this error.

 

6. In the introduction, the works of researchers who have achieved statistical significance and have not been described for too long are also repeated in the discussion. It reduces engagement and is difficult to read

Thank you for your comment, in this regard we have improved the discussion by including references from the introduction.

­

 

7. In materials and methods "Stage 1-IV", please write numbers consistently

We have corrected this error

 

8. Please translate all abbreviations before using them, e.g. LACC

Ready, thanks for the observation

 

9. In Materials and methods is a description that the material was obtained by "surgical excision" - please specify what the procedure is

Thank you for your important observation. We have corrected the main text, since it was not surgery, it was a biopsy obtained by punch.

 

10. P adj <0.05 please explain

Thanks for your comment. It is important to clarify that we use two types of P-value, the adjusted P-value and the P-value. Specifically, the adjusted P-value is generally used for microarray analysis, where a correction is made to the P-value in order to eliminate false positives this information was mentioned in materials and methods section.

 

11. Patients from the TCGA database have a follow-up description, while patients from the observation group have no follow-up description or no

Thanks for your observation. It is important to mention that for the Mexican-mestizo cohort and the validation of the independent Mexican cohort we have complete 5-year follow-up information, but for the OS analysis of the TCGA patients we used the available data, since this database lacks of follow-up information for some patients. This was add to materials and methods section.

 

12. Please write down what significance your results have for the clinic or the extension of diagnostics. Do they have mainly scientific value?

You are right in your comment, these results are very relevant and significant for the clinic. This was mentioned in the discussion and conclusion section. “These findings are very relevant since they show a high concordance of molecular differences for ADC versus SCC between independent cohorts; independently of HPV type, thus opening a window of opportunity to identify new prognostic biomarkers by histological type, nonetheless, further studies are required to define and implement them”.

13. 13. Please work on the clarity of the message in the text

Important comment. In this regard, we have made changes to the abstract, introduction, discussion and conclusion.

 

Reviewer 3 Report

The authors report 70 differentially expressed genes between two subtypes of cervical cancer, namely squamous cell carcinoma and adenocarcinoma, obtained by analyzing and comparing two publicly available expression datasets. Subsequently, they validated by means of qRT-PCR three of these genes (GABRB2, TMEM40 and TSPAN8) with a potential impact on prognosis in a cohort of 31 cervical cancer patients and found an association with prognosis.

The paper needs several minor improvements:

* Abstract
In the abstract, page 1 line 29, the authors report that 70 DEGs are
found in the TCGA cohort and then validated in the Mexican cohort,
whereas in the paper the authors report 1678 DEGs found in the TCGA
cohort. Whereas 70 arises from the intersection between the DEGs of
both cohorts. The abstract is misleading as it suggests that all of
initial DEGs were subsequently validated in an independent cohort.

* Materials and methods
** Analysis of differential expression
 Clustering methods must be described

** CC transcriptome and pathway analyses.
The webgestalt website is mentioned, however the specific method used among over representation analysis, gene set enrichment analysis or network topology analysis is not specified.

* Results
pag 5 line 178. The sentence: "70 were also differentially expressed
in the TCGA 178 cohort and even presented the same log2FoldChange
values" should not use the word "same" but "concordant" or "similar"
instead.

* Figure 1 and 2
The gene and sample lists shown aside the plot are too small and not particularly useful and should be removed. Furthermore, a legend showing the colors associated with ADC or SCC needs to be added to the plot.

* Figure 3
Figure 3 may be made more clear by using a scatterplot that shows the relation between fold-changes.

* Figure 4
Panels A and C are the same.

* Supplementary Tables
Supplementary tables need to be more curated. Namely: overall, the  table captions should be more descriptive; ST1 is merely a dump of a database; ST3 caption does not explictly mention the sample cohort and weather the p-value is adjusted or not; the results of differential expression analysis for the TCGA cohort are reported in ST4 only for the 70 genes that are common between cohorts, whereas the whole set of TCGA DEGs deserves to be shown.

* Table 4
Results of multivariate cox analysis are unclear. Namely, the model shown is the result of a variable selection? In that case, the algorithm (forward, backward) should be mentioned and an overall final model p-value should be shown. Why is HPV in the final model?

Author Response

The authors report 70 differentially expressed genes between two subtypes of cervical cancer, namely squamous cell carcinoma and adenocarcinoma, obtained by analyzing and comparing two publicly available expression datasets. Subsequently, they validated by means of qRT-PCR three of these genes (GABRB2, TMEM40 and TSPAN8) with a potential impact on prognosis in a cohort of 31 cervical cancer patients and found an association with prognosis.

The paper needs several minor improvements:

* Abstract. In the abstract, page 1 line 29, the authors report that 70 DEGs are
found in the TCGA cohort and then validated in the Mexican cohort,
whereas in the paper the authors report 1678 DEGs found in the TCGA
cohort. Whereas 70 arises from the intersection between the DEGs of
both cohorts. The abstract is misleading as it suggests that all of
initial DEGs were subsequently validated in an independent cohort.

Thank you for your very accurate comment, we have modified the abstract by adding the relevant data you mention.

* Materials and methods

** Analysis of differential expression Clustering methods must be described

Thanks for your comment. “To construct the heatmaps, a Z-Score transformation was applied to the normalized gene expression values, next we employed a Hierarchical cluster analysis using the Euclidian distance and the complete-linkage clustering algorithm with the hclust R package”. We add this in materials and methods section.

** CC transcriptome and pathway analyses.

The webgestalt website is mentioned, however the specific method used among over representation analysis, gene set enrichment analysis or network topology analysis is not specified.

To CC transcriptome and pathway analysis we using a Gene Set Enrichment Analysis (GSEA) with the Webgestalt platform. This information was added in materials and methods section

 

* Results pag 5 line 178. The sentence: "70 were also differentially expressed
in the TCGA 178 cohort and even presented the same log2FoldChange
values" should not use the word "same" but "concordant" or "similar"
instead.

Thank you for your valuable observation, we have corrected this error

* Figure 1 and 2 The gene and sample lists shown aside the plot are too small and not particularly useful and should be removed. Furthermore, a legend showing the colors associated with ADC or SCC needs to be added to the plot.

Thanks for your observation, in Figure 1 that are too many patients, we cut the list of genes and samples. In Figure 2, we left it with the list of genes and samples because there are fewer patients and it is easier to distinguish. We also attach the legend that represents the SCC and ADC in green and purple, respectively.

*Figure 3 Figure 3 may be made more clear by using a scatterplot that shows the relation between fold-changes.

Thanks for your recommendation, but a scatter plot is more used for correlation analysis and our objective is only to represent the fold change of each cohort.

*Figure 4 Panels A and C are the same.

Thank you for your comment, the figure has been corrected.

*Supplementary Tables.

Supplementary tables need to be more curated. Namely: overall, the  table captions should be more descriptive; ST1 is merely a dump of a database; ST3 caption does not explictly mention the sample cohort and weather the p-value is adjusted or not; the results of differential expression analysis for the TCGA cohort are reported in ST4 only for the 70 genes that are common between cohorts, whereas the whole set of TCGA DEGs deserves to be shown.

ST1. We use the raw data as-is from the TCGA, because readers could use these tables as-is to run analyses in R. Modifying them complicates their analysis. That is why we decided to leave them as they are so that these data can be used for future analysis.

ST3. Thanks for the recommendation, this table has been improved

ST4.  Thank you for your recommendation, we have added an extra table in the supplementary tables, it is supplementary table 3.

* Table 4 Results of multivariate cox analysis are unclear.

Namely, the model shown is the result of a variable selection? In that case, the algorithm (forward, backward) should be mentioned and an overall final model p-value should be shown. Why is HPV in the final model?  

We did not employ a variable selection algorithm for this analysis, clinical variables were arbitrarily selected since in other studies and in clinical practice they have been shown to be important for the diagnosis and prognosis of patients. Global significance of the model (Likelihood ratio, Wald and Logrank tests) was added to the manuscript

 

 

Round 2

Reviewer 1 Report

The manuscript has improved greatly. The authors' efforts are visible throughout the manuscript and are beneficial to the work. Well done!

Reviewer 2 Report

Thank you for corrections.