Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Real-World Results of Curative Open Colorectal Cancer Surgery in Octogenarians: Meaningful Long-Term Survival despite High Frailty Burden

Humm, not sure ‘meaningful’ should be included in the title given the very small sample size on non-frail group.  It would be a bit more accurate without overstating the results.

Small point in the methods section:  define function dependence. 

It would be nice to have a table that is the 3 frailty score by characteristics. (expanding table 1)

I’m a little concerned with table 2 to run statistical analysis with very small numbers.  Not sure that is helpful since it is underpowdered to show an effect, I believe.

I didn’t get any sense if complications were mostly from a few people or many (suspect these were not unique but rather more than one complication could be counted as unique). 

Similarly, I’m concerned about the ROC given your sample size for ROC reporting. For 30‑day mortality, the event count is extremely low (n=5). While age (and possibly a continuous frailty index),  may still offer some discriminative signal, the other covariates will not. The ROC curve will produce very wide and uninformative confidence intervals, and the resulting AUC will not be meaningful.

It’s not clear whether the mFI is being treated as a continuous score or a dichotomous frailty indicator. If continuous, ROC analysis is acceptable, though confidence intervals will still be wide and should be shown. If dichotomous, it has the same limitations as the other binary comorbidities;  in my opinion, ROC adds almost nothing beyond sensitivity and specificity.

The comorbidities and a dichotomous mFI (if coded that way) do not have adequate sample size to support meaningful ROC curves. A binary predictor yields only one non‑trivial ROC point, so the AUC collapses to a simple function of sensitivity and specificity. In my opinion, this produces a blunt and unstable measure, especially with small cell counts, and the AUC CI will be extremely wide. In these situations, ROC adds little beyond simply reporting sensitivity and specificity with exact Cis.  Similarity because 30‑day mortality is rare, traditional ROC analysis becomes even less informative. In highly imbalanced settings, precision–recall (PR) curves are often more appropriate because they reflect performance in terms of positive predictive value vs sensitivity, which is more aligned with decision‑making under class imbalance.

 

For Kaplan-Meier analysis, it isn’t super clear what was the procedure category since in the table the procedures are broken into many more than 2 categories.   Need to have ‘procedures’ more clearly described for  the KM earlier.  I see it is later listed in table 5 but alittle confusing with Table 1 has a subheading called procedures.  Probably think of combining.    

For the KM, might want to add shaded (95% CI bands and number at risk table after the KM.  From these suggestions, the discussion needs to be balanced with the findings that the authors are able to provide results with more confidence based on small sample size.

 

 

 

Author Response

Manuscript ID: medsci-4130941

Title: Real-world results of curative open colorectal cancer surgery in octogenarians: long-term survival despite high frailty burden

Response to reviewers

Dear Editor,

            Thank you for giving us the opportunity to submit a revised manuscript in your esteemed journal. We appreciate the time and effort that you and the reviewers dedicated to our study, and we are grateful for the interesting comments and valuable suggestions you have made to the paper.

            In this revised manuscript we have incorporated most of the comments made by the reviewers and highlighted the changes to the text in red. Please find below our point-by-point replies and changes to the reviewers’ suggestions.

            We carefully reviewed the iThenticate similarity report and revised the manuscript accordingly. In the revised version, we have rephrased and structurally modified the sections with the highest similarity, in the Methods section, which represented consistent phrasing used across our previously published work by our research group, representing acceptable self-referencing and a uniform manuscript structure rather than inappropriate text reuse.

            We note that a proportion of the remaining similarity is attributable to the use of standard methodological terminology and widely accepted descriptions of statistical methods and frailty indices, which cannot be meaningfully altered without compromising reproducibility.

            We would like to emphasize that all overlapping content is appropriately cited, no data or interpretations are duplicated, and the current study represents original research with distinct objectives, analyses, and results. We therefore believe that the revised manuscript adequately addresses the similarity report and complies with the journal’s standards for originality and ethical publication.

 

REVIEWER 1

Comment 1

Humm, not sure ‘meaningful’ should be included in the title given the very small sample size on non-frail group. It would be a bit more accurate without overstating the results.

Response: We agree and removed the potentially overstating adjective. Revised title: “Real-World Results of Curative Open Colorectal Cancer Surgery in Octogenarians: Long-Term Survival despite High Frailty Burden”. Abstract conclusion wording was also aligned to avoid overstatement.

 

Comment 2

Small point in the methods section: define function dependence.

Response: We agree and added an operational definition of functional dependence consistent with standard mFI-5 use in the Methods section. Lines 112-113

 

Comment 3

It would be nice to have a table that is the 3 frailty score by characteristics. (expanding table 1)

Response: Thank you for your suggestion. We agree and added a new table stratifying baseline characteristics by mFI-5 score (0–3). Lines 171-172

 

Comment 4

I’m a little concerned with table 2 to run statistical analysis with very small numbers. Not sure that is helpful since it is underpowdered to show an effect, I believe.

Response: We agree. For rare complications (very small cell counts), hypothesis testing is underpowered and can be misleading. We therefore removed p-values for individual rare complications and retained inferential testing only for higher-frequency or summary outcomes in Table 2. We have added a note in Table 2 clarifying reasons for not showing p value for several variables. Lines 191-199

 

Comment 5

I didn’t get any sense if complications were mostly from a few people or many (suspect these were not unique but rather more than one complication could be counted as unique).

Response: Thank you. We clarified that complication categories are not mutually exclusive (patients may have multiple complications). The available dataset includes morbidity as a binary indicator and complication types as categories. We have added a footnote in Table 2 stating this.

 

Comment 6

Similarly, I’m concerned about the ROC given your sample size for ROC reporting. For 30‑day mortality, the event count is extremely low (n=5)… The ROC curve will produce very wide and uninformative confidence intervals, and the resulting AUC will not be meaningful.

Response: Thank you for this important remark. We apologize for the misleading statistics. We fully agree that the low number of 30-day mortality events (n=5) substantially limits the reliability and interpretability of ROC curve analyses for this endpoint. In response, we have removed the ROC curve analyses and corresponding figures and tables for 30-day mortality. Mortality outcomes are now reported descriptively only. ROC analyses were retained only for postoperative morbidity, where event counts were sufficient to support exploratory discrimination analyses.

 

Comment 7

It’s not clear whether the mFI is being treated as a continuous score or a dichotomous frailty indicator…

Response: We clarified that frailty status (non-frail mFI-5=0 vs frail mFI-5≥1) is used for the primary group comparisons, and that the mFI-5 score is additionally retained as an ordinal variable for exploratory prediction/discrimination analyses. The Methods section was updated accordingly. Lines 114-115

 

Comment 8

The comorbidities and a dichotomous mFI … do not have adequate sample size to support meaningful ROC curves… In highly imbalanced settings, precision–recall (PR) curves are often more appropriate…

Response: Thank you for this. Indeed, the small sample size for 30-day mortality prediction is inadequate. We have removed the mortality analysis from the manuscript as even PR curves, with an event sample so low, will also be inappropriate. ROC analyses were retained for postoperative morbidity where event counts were adequate for exploratory discrimination assessment, and all discrimination results are now interpreted cautiously with emphasis on the event-limited nature of the mortality endpoint. The Limitations section in the Discussion was revised to emphasize the above limitations. Lines 360-365.

 

Comment 9

For Kaplan-Meier analysis, it isn’t super clear what was the procedure category since in the table the procedures are broken into many more than 2 categories…

Response: Thank you for this remark. In the revised manuscript we have clarified the procedure grouping used in Kaplan–Meier analyses. A new paragraph was added in the Methods section and in the footnotes of tables. Lines 135-137, Lines 222-224

 

Comment 10

For the KM, might want to add shaded (95% CI bands and number at risk table after the KM.

Response: Thank you for your suggestion. In the revised version we updated the figures including the shaded 95CI and numbers at risk table.

 

Comment 11

From these suggestions, the discussion needs to be balanced with the findings that the authors are able to provide results with more confidence based on small sample size.

Response: We agree and revised the Discussion to more explicitly acknowledge limitations from sample size and low event counts, and to avoid overinterpretation of subgroup and discrimination analyses. Lines 360-365

Reviewer 2 Report

Comments and Suggestions for Authors

It is a single center, retrospective, study on 112 octogenarians subjected to curative open colorectal surgery. Authors, have studied the long-term survival of this group in relation to their frailty status and other co-morbidities.

Despite the limitations [aknowledged by authors: [[i] single center; [ii] open surgery only; [iii] retrospective] the study is well designed; a plenty of patient exclusion make the cohort totally homogenate, the results to be without bias.

I have 3 comments:

[I] do  colon rectal surgery patients had been subjected to Neo-adjouvant therapy; how many of the total 112

[ii] do the authors perform preventing stomas? they are referred to a small number of anastomotic dehiscence

[iii] I would like to read in discussion a comparison with their results in a younger population, if they have such data

Author Response

Manuscript ID: medsci-4130941

Title: Real-world results of curative open colorectal cancer surgery in octogenarians: long-term survival despite high frailty burden

Response to reviewers

Dear Editor,

            Thank you for giving us the opportunity to submit a revised manuscript in your esteemed journal. We appreciate the time and effort that you and the reviewers dedicated to our study, and we are grateful for the interesting comments and valuable suggestions you have made to the paper.

            In this revised manuscript we have incorporated most of the comments made by the reviewers and highlighted the changes to the text in red. Please find below our point-by-point replies and changes to the reviewers’ suggestions.

            We carefully reviewed the iThenticate similarity report and revised the manuscript accordingly. In the revised version, we have rephrased and structurally modified the sections with the highest similarity, in the Methods section, which represented consistent phrasing used across our previously published work by our research group, representing acceptable self-referencing and a uniform manuscript structure rather than inappropriate text reuse.

            We note that a proportion of the remaining similarity is attributable to the use of standard methodological terminology and widely accepted descriptions of statistical methods and frailty indices, which cannot be meaningfully altered without compromising reproducibility.

            We would like to emphasize that all overlapping content is appropriately cited, no data or interpretations are duplicated, and the current study represents original research with distinct objectives, analyses, and results. We therefore believe that the revised manuscript adequately addresses the similarity report and complies with the journal’s standards for originality and ethical publication.

REVIEWER 2

Comment 1: do colon rectal surgery patients had been subjected to Neo-adjouvant therapy; how many of the total 112

Response: Thank you for this remark. 40 patients (35.7%) underwent long course neoadjuvant CRT. This has been added in Table 1.

 

Comment 2: do the authors perform preventing stomas? they are referred to a small number of anastomotic dehiscence

Response: Thank you. Yes, it is our standard practice to perform diverting loop ileostomy in all low colorectal anastomoses which underwent neoadjuvant CRT or if there are other risk factors for anastomotic leak, such as advanced age or significant comorbidities. In the revised version, we have added the rate of diverting stomas in our cohort in Table 1.

 

Comment 3: I would like to read in discussion a comparison with their results in a younger population, if they have such data

Response: Thank you for this suggestion. Unfortunately, for this study we do not have a comparative younger group of patients, although we are currently extending the database to 60-80 years old patients. In an incoming study we will compare outcomes between octogenarians and younger patients, in a propensity case matched fashion.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I'm not sure why ROC was presented based on previous concerns of lack of clear findings.  CI was not included.  If the authors insist on using ROC for outcomes that have limited findings given the sample size, CI needs to be added; which suggests separate ROC for each one of the outcomes.  Appreciated adding the large CI for KM.  Appreciated the addition of   material to clarify the study design.  It doesn't look like the discussion was updated besides saying small number for 30 day mortality.  Rather than a reviewer going through this to mention where it needs to be updated, the authors should more carefully review the findings based on previous feedback and make adjustments.  

Author Response

Dear Editor,

We sincerely thank you and the reviewers for the additional careful evaluation of our revised manuscript and for the constructive comments provided in this second review round. We have carefully reconsidered all remarks and have undertaken further revisions to improve the manuscript. We have refined the discussion and conclusions to better reflect the descriptive nature and statistical limitations of the study, clarified the reasoning behind keeping the ROC curve analysis for morbidity, given the adequate sample size, and added additional subgroup information based on CRT status as suggested. All modifications made in the manuscript are highlighted in red. We hope that the revised version adequately addresses the remaining concerns and meets the journal’s standards for publication.

REVIEWER 1

Comment 1: I'm not sure why ROC was presented based on previous concerns of lack of clear findings.  CI was not included.  If the authors insist on using ROC for outcomes that have limited findings given the sample size, CI needs to be added; which suggests separate ROC for each one of the outcomes.

Response: We thank the reviewer for this comment. We respectfully clarify that ROC analysis was performed exclusively for postoperative morbidity (n=46 events), not for 30-day mortality, which was deleted in the revised version, given the low number of events and we thank the reviewer for pointing that out. The event rate for morbidity in our cohort was approximately 41%, which does not represent a highly imbalanced outcome. We believe ROC curve analysis remains an appropriate and widely accepted method for evaluating discrimination in such settings. Prior methodological studies have demonstrated that ROC curves are robust when outcome prevalence is moderate (The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets - PubMed ; A simulation study of the number of events per variable in logistic regression analysis - PubMed). Furthermore, the number of morbidity events exceeds commonly cited minimum thresholds, i.e. more than 10%. For clarity, we have now explicitly stated in the Methods and Discussion that ROC analyses for morbidity are exploratory and interpreted cautiously (Lines 213-218 and Lines 377-386)

            Regarding the representation of 95% CI, we would like to emphasize that the 95% CI for each variable is presented in Table 5 where each ROC curve is depicted separately.  We kept the figure showing all ROC curves in one to reduce redundancy of multiple figures, but the table shows each ROC curve analysis. In this version we have explicitly referenced the 95% CIs in the figure legend and Results section (Lines 210-213), and we have ensured that AUC values with corresponding 95% CIs are clearly presented. We agree that interpretation must be cautious given sample size, and the Discussions have been revised in this regard.

 

Comment 2: It doesn't look like the discussion was updated besides saying small number for 30 day mortality.  Rather than a reviewer going through this to mention where it needs to be updated, the authors should more carefully review the findings based on previous feedback and make adjustments.  

Response: We are sorry that the Discussion is not suitable. In the first revision we focused more on Methods, presentation of Results and Limitations of the study as we felt this is where reviewers had more remarks and concerns. In this re-revised version, we have changed the first paragraph of the Discussion, where we present the overall findings, and have tempered down the significance of the results considering the low sample. Also, the conclusions have been revised to better acknowledge the limitations of this study (Lines 248-255 and Lines 391-397). Also a new paragraph was added in the Limitations section (Lines 370-387)

 

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors,

although there is a small disadvantage with the small number of participants, without control group, and the 112 patients now should be seperated into with or without neo-adjuvant therapy

I agree your paper to be published as a preliminary report

Author Response

Dear Editor,

We sincerely thank you and the reviewers for the additional careful evaluation of our revised manuscript and for the constructive comments provided in this second review round. We have carefully reconsidered all remarks and have undertaken further revisions to improve the manuscript. We have refined the discussion and conclusions to better reflect the descriptive nature and statistical limitations of the study, clarified the reasoning behind keeping the ROC curve analysis for morbidity, given the adequate sample size, and added additional subgroup information based on CRT status as suggested. All modifications made in the manuscript are highlighted in red. We hope that the revised version adequately addresses the remaining concerns and meets the journal’s standards for publication.

Comment 1: although there is a small disadvantage with the small number of participants, without control group, and the 112 patients now should be seperated into with or without neo-adjuvant therapy

Response: We thank the reviewer for this suggestion. In the revised version we have added a new table (Table 4) describing postoperative outcomes including morbidity, mortality and OS in CRT vs no CRT subgroup analysis. Also, the text in the manuscript was updated accordingly.