Comparative Survival Outcomes of Hyperthermic Intraperitoneal Chemotherapy, Intraperitoneal Chemotherapy and Intravenous Chemotherapy for Primary Advanced Ovarian Cancer: A Network Meta-Analysis

Round 1

Reviewer 1 Report

The manuscript sent for review concerns important and interesting issues of choosing the best route of administration of cytostatic agents in patients with ovarian cancer. The advantage of the study is the holistic assessment of the problem of various routes of drug administration. However, the manuscript does not add anything new on this subject. It does not discuss side effects and quality of life.

 

Lines 169-171 are incomprehensible to me

Figure 1 - probably an error - the same number of screened and excluded records?

All the Tables are too complex and therefore illegible, incomprehensible.

 

The manuscript in its present form is not suitable for publication.

Needs deep corrections and explanations.

Author Response

Dear reviewer:

Thanks very much for taking your time to review our manuscript. We really appreciate all your comments and suggestions. According to your suggestions, we have supplemented several data here and corrected some mistakes in our previous manuscript. Based on your comments, we also attached a point-by-point letter to you. We have made extensive revisions and the detailed point-by-point responses are listed below.

 

The manuscript sent for review concerns important and interesting issues of choosing the best route of administration of cytostatic agents in patients with ovarian cancer. The advantage of the study is the holistic assessment of the problem of various routes of drug administration. However, the manuscript does not add anything new on this subject. It does not discuss side effects and quality of life.

A: Some articles reported all grades of AEs, while others only reported AEs ≥ grade 3, and some articles did not provide AEs. Considering that AE ≥grade 3 has a greater impact on the patient's body, and the AEs available for analysis is limited. We only analyzed the difference in AE (≥ grade 3) between patients in the HIPEC group and those non-HIPEC group. We added the results in our manuscript (line 187-194).       Due to the limited articles available for statistical analysis, we did not analyze the quality of life in our network meta-analysis. However, we discussed the impact of HIPEC on patients' quality of life (line 263-267).

Lines 169-171 are incomprehensible to me

A: Sorry for our unclear expression, we have revised this sentence (line 164-165).

Figure 1 - probably an error - the same number of screened and excluded records?

A: We are so sorry about this mistake in our manuscript. We added the search database according to the comments of another reviewer, and we carefully checked and modified Figure 1.

All the Tables are too complex and therefore illegible, incomprehensible.

A: According to the suggestions of reviewers, we have made corresponding modifications. We simplified and integrated the pictures and tables in the manuscript.

Author Response File: Author Response.docx

Reviewer 2 Report

The paper investigates the efficacy of various regimens and ways of administration for chemotherapy in ovarian cancer setting by performing a systematic review and meta-analysis. This is a very relevant subject while falling within the aim and scope of the journal. However, it raises concerns over the lack of novelty due to existing literature on this subject. Nevertheless, the paper adds value and potential novelty by including multiple comparisons between treatments. Therefore, I make the following comments and suggestions:

Text and Methodology:

                The manuscript is somewhat comprehensive and follows the PRISMA guidelines. However, I see no record of a pre-registered protocol. Please note that all systematic reviews and meta-analyses should be pre-registered in the PROSPERO database (https://www.crd.york.ac.uk/prospero/) to avoid duplication. This is not mandatory by any means and comes more as a recommendation. Perhaps the authors failed to provide the registry number. If the authors did not pre-register the study, they might consider doing that before publication. There are several alternatives to register your work, and the editors should give further guidance. Otherwise, the study is well constructed and touches every point of the PRISMA guidelines.

                The text is, however hard to follow, and there are some suggestions I make:

-line 39: Please insert citations regarding statistics of ovarian cancer. Do you refer to its incidence or prevalence? Feel free to use and cite Globocan statistics, which are very accurate and exact. https://gco.iarc.fr/today/home

-lines 42 and 43 – please insert citations to argue the best treatment course

-lines 58 and 59 – the sentence ends abruptly and makes no sense. Please revise and also insert citations

-line 74 – the authors give an example of the search strategy. Please provide the exact search strategy used in every database. Perhaps the authors could provide this information in a Supplementary Material

-Please provide PICO-formulated inclusion criteria; this will group all the parameters you are filtering for

-CENTRAL database was not searched. This is one of the most comprehensive databases providing access to published and undergoing trials. Please revise the search and CENTRAL. Should the authors consider, they might expand the search to other databases as well, such as SCOPUS or Springer

-The authors should be aware of the latest Cochrane recommendations regarding the time-to-event analysis when performing such a complex task. All recommendations point towards using the log-transformed values of HR. This will render confidence intervals to appear symmetrical and is the preferred method for analyzing HR. Standard Error is also used, and data can be computed into the Inverse Variance method for the meta-analysis. You can always check and cite the handbook provided by Cochrane that gives detailed explanations regarding the best methods for such analyses. https://training.cochrane.org/handbook/current/chapter-10#section-10-9

-Most times, publications will not provide HR and measures of Variance so appropriate transformations must be undertaken. Tierney's et al paper (https://trialsjournal.biomedcentral.com/articles/10.1186/1745-6215-8-16) accurately describes many ways of transforming data into logHR and estimates of Variance. Feel free to use the calculus spreadsheet provided in the Supplementary Material to aid with data transformation. Detailed explanations can also be consulted in this very useful webinar by Cochrane regarding aggregate time-to-event data meta-analyses: https://training.cochrane.org/resource/meta-analysis-time-event-data

-Please re-check all the included papers to make sure you provided adequate transformations. The R software is very complex and comprehensive, but the authors should perhaps consider using a more user-friendly software that might aid with performing a correct meta-analysis using lnHR and SE values after appropriate transformations

-please consider performing a sensitivity analysis by excluding each study from the statistics and checking if the results remain in the initial CI. I2 was used for heterogeneity assessment. Although this is not an absolute measurement of heterogeneity, I2 is a valuable tool for highlighting the proportion between variances in true effect size or sampling errors. Please consider performing another companion test, such as the Chi2, and provide p-values for heterogeneity.

-Combining RCTs with non-RCTS and other types of studies such cohorts (retrospective and prospective) is problematic and might introduce a lot of bias. Please consider at least  performing a sub-group analysis based on the type of included studies: RCTs compared to other studies. Explore the heteroegenety after and check if the effect carries over.

-Another problem is the pooling of HRs. Please note that a lot of comparative studies that are not RCTs might provide adjusted values for HRs (usually adjusted for age, stage, or other parameters). The studies should always state whether the given values are adjusted or not or the authors should deduct that if the results are from analyses such as a Cox multivariate regression. This is useful since it can lessen the risk of bias pooled from these types of publications. Still, it is problematic since it is not usually compatible with an unadjusted HR from an RCT or an HR deducted from the log-rank test. The authors have no record of this. This is of utmost importance and should be addressed in the manuscript. Please make sure you re-check you included publications.

-Markman M (Refference 45) was categorized as an RCT but had undergone NOS evaluation?

Grammar:

                The paper needs appropriate and extensive English-language revisions. I can give some examples below:

-"literatures published" (line 15) - might want to replace with publications or articles; also reformulate line 15 and 16, short sentences with no sense?

-line 18: sentence begins with And

-consistently using our or we in formal writing

-line 24 sentence begins with and

-line 29: presenting results with the phrasing “there was probably”. Line 34: Advise against conclusions like ”more evidence is needed”

-line 100 "respectively" is not a necessary word

Author Response

Dear reviewer:

Thank you for your decision and constructive comments on our manuscript. We have carefully considered the suggestions of reviewer and make some changes. We have tried our best to improve and made some revised in the manuscript. The red part that has been revised according to your comments. Revision notes, point-to-point, are given as follows:

 

-line 39: Please insert citations regarding statistics of ovarian cancer. Do you refer to its incidence or prevalence? Feel free to use and cite Globocan statistics, which are very accurate and exact. https://gco.iarc.fr/today/home

A: We have inserted the citation in line 39.

-lines 42 and 43 – please insert citations to argue the best treatment course

A: We have inserted the citation in line 42 and 43.

-lines 58 and 59 – the sentence ends abruptly and makes no sense. Please revise and also insert citations

A: We have revised this sentence and inserted citation in line 59.

-line 74 – the authors give an example of the search strategy. Please provide the exact search strategy used in every database. Perhaps the authors could provide this information in a Supplementary Material

A: We provide the exact search strategy used in each database in File A1.

-Please provide PICO-formulated inclusion criteria; this will group all the parameters you are filtering for

A: The PICOS principle has been demonstrated in our inclusion criteria. The details are as follows:

P: Patients newly diagnosed with ovarian cancer, primary peritoneal, or fallopian tube carcinoma

I：Platinum-based chemotherapy, including HIPEC, IV, IP and weekly dose-dense chemotherapy

C：The treatment groups can be controlled by each other. There are no specific restrictions.

O：Survival outcomes and AEs are available

S: There are no specific restrictions.

-CENTRAL database was not searched. This is one of the most comprehensive databases providing access to published and undergoing trials. Please revise the search and CENTRAL. Should the authors consider, they might expand the search to other databases as well, such as SCOPUS or Springer

A: We increased the search of the CENTRAL and SCOPUS databases, and we modified Figure 1.

-The authors should be aware of the latest Cochrane recommendations regarding the time-to-event analysis when performing such a complex task. All recommendations point towards using the log-transformed values of HR. This will render confidence intervals to appear symmetrical and is the preferred method for analyzing HR. Standard Error is also used, and data can be computed into the Inverse Variance method for the meta-analysis. You can always check and cite the handbook provided by Cochrane that gives detailed explanations regarding the best methods for such analyses. https://training.cochrane.org/handbook/current/chapter-10#section-10-9

A: Thanks for your careful reading of our manuscript and for your good suggestions. Following your suggestion, we carefully read the manual provided by Cochrane and found the most suitable method for us to obtain HR and Cl values.

-Most times, publications will not provide HR and measures of Variance so appropriate transformations must be undertaken. Tierney's et al paper (https://trialsjournal.biomedcentral.com/articles/10.1186/1745-6215-8-16) accurately describes many ways of transforming data into logHR and estimates of Variance. Feel free to use the calculus spreadsheet provided in the Supplementary Material to aid with data transformation. Detailed explanations can also be consulted in this very useful webinar by Cochrane regarding aggregate time-to-event data meta-analyses: https://training.cochrane.org/resource/meta-analysis-time-event-data

​A: Thanks for your suggestions. Some studies we included did not provide HRs and Cl directly, but did provide Kaplan-Meier curves. We used the method presented in Tierney et al. 's paper to extract data from Kaplan-Meier curves and used the calculus spreadsheet they provided to help us the data conversion. We have also stated in lines 107-109 of the manuscript.

-Please re-check all the included papers to make sure you provided adequate transformations. The R software is very complex and comprehensive, but the authors should perhaps consider using a more user-friendly software that might aid with performing a correct meta-analysis using lnHR and SE values after appropriate transformations

A：We converted HR and Cl into lnHR and SE values through Liu et al. 's method and conducted a meta-analysis using R. （https://gb.global.cnki.net/KCMS/detail/detail.aspx?dbcode=CJFD&dbname=CJFDLAST2016&filename=PZXX201607001&uniplatform=OVERSEA&v=GQZdI3orrfQm_FT_JkD8XAQzmE8H088pQaf7psmcOnf4bhZUZ01dXalig9qECG0o）

-please consider performing a sensitivity analysis by excluding each study from the statistics and checking if the results remain in the initial CI. I2 was used for heterogeneity assessment. Although this is not an absolute measurement of heterogeneity, I2 is a valuable tool for highlighting the proportion between variances in true effect size or sampling errors. Please consider performing another companion test, such as the Chi2, and provide p-values for heterogeneity.

A：Thanks very much for the advice. Sensitivity analysis was not performed due to technical limitations and the low to moderate heterogeneity of OS (I2=21%) and PFS (I2=14%) observed in our analysis. TAU2 can also reflect global heterogeneity in our analysis, so we added TAU2.

-Combining RCTs with non-RCTS and other types of studies such cohorts (retrospective and prospective) is problematic and might introduce a lot of bias. Please consider at least  performing a sub-group analysis based on the type of included studies: RCTs compared to other studies. Explore the heteroegenety after and check if the effect carries over.

A: Due to technical limitations, we did not directly conduct subgroup analysis. We analyzed the RCT study separately from other types of studies. This part of the result is added to the manuscript (line 166-172 and line 182-186) and discussed.

-Another problem is the pooling of HRs. Please note that a lot of comparative studies that are not RCTs might provide adjusted values for HRs (usually adjusted for age, stage, or other parameters). The studies should always state whether the given values are adjusted or not or the authors should deduct that if the results are from analyses such as a Cox multivariate regression. This is useful since it can lessen the risk of bias pooled from these types of publications. Still, it is problematic since it is not usually compatible with an unadjusted HR from an RCT or an HR deducted from the log-rank test. The authors have no record of this. This is of utmost importance and should be addressed in the manuscript. Please make sure you re-check you included publications.

A：Many thanks for the valuable comments. We have re-examined carefully the publications we have included. The majority of the non-RCTs studies we included had consistent baseline characteristics. Several articles provide both unadjusted and adjusted HR values. We extracted the unadjusted HR values or univariate analyses for statistical analysis. We have stated in the manuscript according to the comments of the reviewer.

-Markman M (Refference 45) was categorized as an RCT but had undergone NOS evaluation?

A: We are so sorry for this error, and we have carefully checked and corrected it.

Grammar:

                The paper needs appropriate and extensive English-language revisions. I can give some examples below:

-"literatures published" (line 15) - might want to replace with publications or articles; also reformulate line 15 and 16, short sentences with no sense?

-line 18: sentence begins with And

-consistently using our or we in formal writing

-line 24 sentence begins with and

-line 29: presenting results with the phrasing “there was probably”. Line 34: Advise against conclusions like ”more evidence is needed”

-line 100 "respectively" is not a necessary word

A: Thanks for the reviewer's suggestions. We apologize for the poor language of our manuscript. We have sent our manuscript to native English speakers for review and revision. So that some words and sentences were changed but did not change their original meaning, so we did not mark them red in the manuscript.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The manuscript has been improved.