Real-World Data on the Effectiveness of Immunotherapy on Advanced NSCLC: A Retrospective Cohort Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article entitled “Real-World Data on the Effectiveness of Immunotherapy on Advanced NSCLC: A Retrospective Cohort Study". The article is academically sound and clinical, and it offers useful practical information on the efficacy of immunotherapy in patients with progressive NSCLC. A major revision will enable the manuscript to be accepted for publication in the Journal.

1. Temporal bias is a potential issue that should be addressed, as the study involves patients over a long period (1999-2024). There has been a change in treatment and access to immunotherapy. It is suggested that social stratification or adjustment across time periods be used.
2. The study is retrospective, which could introduce selection bias and confounding, particularly when comparing the immunotherapy and chemotherapy groups. Could you explain how the confounders were addressed during the analysis?
3. Molecular profiling and PD-L1 testing were performed in only a subset of patients, which could have influenced treatment decisions. The abstract and the discussion should address this restriction more prominently.
4. The conclusion that immunotherapy monotherapy was more effective in terms of survival than combination therapy should be taken with a grain of salt, as it may be influenced by patient selection (e.g., PD-L1 expression levels). Do revise the wording, please.
5. Adverse event or toxicity data are not included in the study, and this is significant when studying in the real world. This limitation is better discussed.
6. TTNT should be used as a substitute for progression-free survival, and stronger arguments, with relevant references and a discussion of its shortcomings, should be provided.
7. Please elaborate further on the selection criteria used in treatment, including the decision-making process for selecting immunotherapy and chemotherapy, and how decisions were made regarding combination regimens.

Minor comments/suggestions

8. It would be better to note the study's limitations in the abstract to enhance transparency. The figures could be made clearer, more resolved, and labeled better to make them easier to read.
9. Figure legends could also be shorter but more informative, without any repetition.
10. Standardize the reporting of statistical findings, especially hazard ratios and confidence intervals. 1
11. It is possible to consider introducing a graphical abstract or schematic summary to emphasize the main findings and enhance interaction with the reader.

Author Response

We sincerely thank the reviewer for the positive evaluation of our manuscript and for the constructive comments. We have carefully addressed all comments and revised the manuscript accordingly.

Comments 1: “Temporal bias is a potential issue that should be addressed, as the study involves patients over a long period (1999-2024). There has been a change in treatment and access to immunotherapy. It is suggested that social stratification or adjustment across time periods be used.”

Response 1: We thank the reviewer for highlighting the potential for temporal bias due to the long study period. In response, we have included calendar period as a categorical variable (1999–2018, 2019–2024) in the multivariable analysis to account for changes in treatment availability and clinical practice over time. This has been added to the Methods section and further discussed as a limitation in the Discussion.

“Specifically, calendar period was included as a categorical variable with two major time groups (1999-2018, 2019-2024). (lines 130-131, page 4).

“First, the design and the data collection introduces inherent selection bias and temporal confounding, especially given the long study period (1999–2024). Differences in treatment availability over time, incomplete molecular profiling, especially during the early years and the absence of systematic adverse event data further limit generalizability of the findings” (lines 348-352, page 12).

Comments 2: “The study is retrospective, which could introduce selection bias and confounding, particularly when comparing the immunotherapy and chemotherapy groups. Could you explain how the confounders were addressed during the analysis?”

Response 2: We agree that confounding is an inherent limitation of retrospective analyses. To address this, we expanded our multivariable models to include clinically relevant covariates selected a priori, including age, sex, histology, performance status, and calendar period. Despite these adjustments, residual confounding cannot be excluded, and this has now been explicitly acknowledged in the Discussion.

“Multivariable models included age, sex, histology, performance status, smoking status, stage, and calendar period to account for temporal bias. Specifically, calendar period was included as a categorical variable with two major time groups (1999-2018, 2019-2024).” (lines 128-131, page 4).

“Differences in treatment availability over time, incomplete molecular profiling, especially during the early years, and the absence of systematic adverse event data further limit generalizability of the findings. Another important confounding factor was the small subset of patients undergoing PD-L1 testing, which in combination with the diverse molecular testing may have influenced treatment selection and outcome interpretation” (lines 349-354, page 12).

Comments 3: “Molecular profiling and PD-L1 testing were performed in only a subset of patients, which could have influenced treatment decisions. The abstract and the discussion should address this restriction more prominently.”

Response 3: We thank the reviewer for this important observation. We have revised both the Abstract and the Discussion to more clearly emphasize the limited availability of PD-L1 expression and molecular profiling data, and their potential impact on treatment selection and interpretation of outcomes.

“While molecular testing rates increased significantly over the study, only a minority of patients had PD-L1 testing, while also broad molecular profiling was incomplete, limiting the interpretation of treatment effects” (lines 51-53, page 2)

“Another important confounding factor was the small subset of patients undergoing PD-L1 testing, which in combination with the diverse molecular testing may have influenced treatment selection and outcome interpretation” (lines 352-354, page 12).

Comments 4: “The conclusion that immunotherapy monotherapy was more effective in terms of survival than combination therapy should be taken with a grain of salt, as it may be influenced by patient selection (e.g., PD-L1 expression levels). Do revise the wording, please.”

Response 4: We agree with the reviewer that this finding should be interpreted with caution. Accordingly, we have revised the wording of the Results and Conclusions to avoid causal interpretations and to reflect the possibility of selection bias, particularly related to PD-L1 expression and patient characteristics.

“Interestingly, immunotherapy monotherapy appeared to have higher OS compared to chemo-immunotherapy combinations. This observation should be interpreted cautiously given the retrospective design and potential confounding by indication and patient selection. Patients receiving combination regimens often had lower PD-L1 expression or more aggressive disease, with higher disease burden, while immunotherapy monotherapy was predominantly administered as first-line treatment in patients with PD-L1 ≥50%, reflecting current clinical practice guidelines and potentially explaining the observed differences in outcome. However, it underscores the importance of optimal patient selection and biomarker-driven treatment decisions” (lines 255-263, page 10).

Comments 5: “Adverse event or toxicity data are not included in the study, and this is significant when studying in the real world. This limitation is better discussed.”

Response 5: We acknowledge that the absence of toxicity data is an important limitation in a real-world study of immunotherapy. This limitation has now been explicitly discussed in the revised Discussion section.

“A further limitation of this study is the lack of available data on adverse events and treatment-related toxicities, which are particularly relevant in real-world settings. The absence of such information precludes a comprehensive assessment of the risk–benefit profile of the evaluated treatments and may limit the interpretation of their overall clinical utility.” (lines 356-360, page 12).

Comments 6: “TTNT should be used as a substitute for progression-free survival, and stronger arguments, with relevant references and a discussion of its shortcomings, should be provided.”

Response 6: We appreciate this helpful comment. We have expanded the Discussion to better justify the use of time-to-next-treatment (TTNT) as a real-world surrogate endpoint, including relevant references and a more detailed discussion of its limitations, such as its dependence on physician decision-making and treatment access.

“In our study, TTNT was used as a pragmatic real-world surrogate for progression-free survival [7], with the understanding that it can be influenced not only by disease progression but also by physician practice, patient preference, treatment access, and toxicity.” (lines 264-266, page 10).

Comments 7: “Please elaborate further on the selection criteria used in treatment, including the decision-making process for selecting immunotherapy and chemotherapy, and how decisions were made regarding combination regimens.”

Response 7: We thank the reviewer for this suggestion. We have further elaborated on the treatment selection process in the Methods section, clarifying that treatment decisions were based on clinical judgment, patient characteristics, and treatment availability during the respective time periods.

“Treatment decisions were based on clinical judgment, patient characteristics, and treatment availability during the respective time periods” (lines 106-108, page 3).

Comments 8: “It would be better to note the study's limitations in the abstract to enhance transparency. The figures could be made clearer, more resolved, and labeled better to make them easier to read.”

Response 8: We agree and have now included key study limitations in the Abstract to improve transparency. Figures have been revised to improve clarity, resolution, and labeling.

“Only a minority of patients had PD-L1 testing, while also broad molecular profiling was incomplete, limiting the interpretation of treatment effects” (lines 52-53, page 2).

Comments 9: “Figure legends could also be shorter but more informative, without any repetition.”

Response 9: Figure legends have been shortened and revised to be more concise and informative, avoiding repetition.

Comments 10: “Standardize the reporting of statistical findings, especially hazard ratios and confidence intervals.”

Response 10: We thank the reviewer for the comment. Statistical reporting has been standardized throughout the manuscript, including consistent presentation of hazard ratios and confidence intervals.

Comments 11: “It is possible to consider introducing a graphical abstract or schematic summary to emphasize the main findings and enhance interaction with the reader.”

Response 11: We thank the reviewer for this suggestion. A graphical abstract was already included in the original submission. In the revised version, we have further refined it to improve clarity and better reflect the main findings of the study.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript addresses a clinically relevant question and uses a reasonably large real-world cohort to examine immunotherapy outcomes in advanced NSCLC. The topic is important, and the multicenter design, survival endpoints, and attempt to contextualize molecular profiling trends are all positive aspects. However, the current analysis has major methodological weaknesses, especially substantial risk of immortal-time bias and confounding by indication, limited covariate adjustment, inconsistent cohort definitions, and several numerical and reporting discrepancies. In its present form, the conclusions are stronger than the data can reliably support.

• The most important issue is the high risk of immortal-time bias. Comparing patients who received immunotherapy “at any line” versus those who never received it, while measuring OS from diagnosis of advanced disease, is methodologically problematic because patients must survive long enough to receive later-line immunotherapy. This can substantially exaggerate the apparent survival benefit.

• The primary analysis should be redesigned using a more appropriate framework, such as a time-dependent Cox model, a landmark analysis, or clearly separated line-specific cohorts with aligned time zero.

• There is also strong confounding by indication, especially in the comparison of first-line immunotherapy monotherapy versus chemo-immunotherapy versus chemotherapy. These groups are almost certainly different with respect to PD-L1 status, disease burden, histology, calendar year, performance status, comorbidities, and treatment eligibility.

• The multivariable models are underadjusted. Including only smoking, stage, and performance status is not enough. The analysis should at least consider age, sex, histology, PD-L1, metastatic burden, de novo versus recurrent disease, calendar period, center, and other clinically relevant covariates.

• The claim that first-line immunotherapy monotherapy was superior to chemo-immunotherapy should be interpreted with much greater caution. Without adequate adjustment, this result is very likely driven by patient selection rather than treatment effect.

• Please clarify exactly who was included in the 610-patient outcome cohort and how the exclusions were applied. The current description is not fully consistent across the methods, figure, and supplementary material.

• There is an important inconsistency regarding targeted-therapy exclusions. The main flow diagram indicates exclusion of patients with EGFR/ALK mutations and/or targeted therapy from the outcome analysis, yet the supplementary note states that some patients receiving targeted agents were not excluded. This must be reconciled clearly.
• Please carefully recheck Table 2 percentages, because at least one percentage appears incorrect.

• The manuscript states that TTNT outcomes were consistent across PD-L1 subgroups, but then refers to numerically longer PFS. Since PFS was not reported as an endpoint, this should be corrected.

• Because only a minority of patients had PD-L1 testing, and molecular profiling was incomplete overall, the treatment-effect interpretation is limited. This should be emphasized more strongly in both the discussion and the conclusion.

• Please clarify whether the 24 untreated patients were included in the outcome analysis and, if so, in which comparator group. Their inclusion could materially bias the results.

• The use of TTNT as a surrogate for PFS is acceptable in some real-world settings, but it should be discussed more carefully because TTNT is influenced by physician practice, access, toxicity, and patient preference, not only progression.

• The KM plots should be improved both analytically and visually. Please report the corresponding hazard ratios with 95% confidence intervals and p-values for each comparison, within the figure panels. At present, the survival curves are shown without enough statistical detail to properly interpret the magnitude and significance of the observed differences.

The study has value as a real-world descriptive analysis, but the current analytical approach is not strong enough to support the headline claims without substantial revision.

 

Author Response

We sincerely thank the reviewer for the thorough and insightful evaluation of our manuscript. We also acknowledge the important methodological concerns raised. In response, we have substantially revised the manuscript, including additional analyses, improved statistical modeling, and clearer reporting. All comments have been addressed in detail below.

Comments 1: “The most important issue is the high risk of immortal-time bias. Comparing patients who received immunotherapy “at any line” versus those who never received it, while measuring OS from diagnosis of advanced disease, is methodologically problematic because patients must survive long enough to receive later-line immunotherapy. This can substantially exaggerate the apparent survival benefit.”

Response 1: We fully agree with the reviewer that immortal-time bias is a critical concern in observational studies where treatment is initiated after cohort entry. To address this, we performed a landmark analysis 3 months from the diagnosis of advanced disease, including only patients alive at the landmark time point. Survival was subsequently analyzed from the landmark forward.

“The results remained consistent with the primary analysis also in the landmark analysis using a 3-month cutoff to account for the immortal-time bias” (lines 42-44, page 1).

“In order to mitigate the impact of immortal-time bias we introduced a separate landmark analysis, using as a predefined cutoff at three months” (lines 118-119, page 3).

“In the landmark analysis using a 3-month cutoff, patients who survived beyond the landmark time were included in the analysis. The results remained consistent with the primary analysis, showing that patients receiving immunotherapy at any line showed improved OS compared to chemotherapy regimens (mOS 18.1 vs 10.4 months; HR: 0.55, 95% CI: 0.45-0.68; p < 0.001). No substantial differences in effect estimates were observed, suggesting that immortal-time bias did not significantly influence the findings (Supplementary Figure 2).” (lines 193-199, page 7).

“Patients exposed to immunotherapy demonstrated a doubling of median OS compared with those who never received immunotherapy (17.5 vs. 8.6 months), an effect that remained robust in multivariate analysis after accounting for age, sex, histology, stage, performance status, smoking status and calendar period” (lines 273-277, page 10).

Comments 2: “The primary analysis should be redesigned using a more appropriate framework, such as a time-dependent Cox model, a landmark analysis, or clearly separated line-specific cohorts with aligned time zero.”

Response 2: We thank the reviewer for the comment. To address this, we performed a landmark analysis at 3 months from diagnosis, including only patients alive at the landmark time point. Survival was subsequently analyzed from that time forward.

“In order to mitigate the impact of immortal-time bias we introduced a separate landmark analysis, using as a predefined cutoff at three months” (lines 118-119, page 3).

Comments 3: “There is also strong confounding by indication, especially in the comparison of first-line immunotherapy monotherapy versus chemo-immunotherapy versus chemotherapy. These groups are almost certainly different with respect to PD-L1 status, disease burden, histology, calendar year, performance status, comorbidities, and treatment eligibility.”

Response 3: We agree that confounding by indication is likely present, particularly in comparisons between treatment groups. To mitigate this, we expanded our multivariable Cox regression models to include clinically relevant covariates, including age, sex, histology, performance status, stage, and calendar period. Due to the limited number of patients with available PD-L1 data, we were unable to include it in the main multivariable models without substantially reducing statistical power. However, we performed a subset analysis including only patients with available PD-L1 data, and the results were consistent with the main findings (see Supplementary Table 3). We have now clearly stated this limitation in the Discussion session.

“Multivariable models included age, sex, histology, performance status, smoking status, stage, and calendar period to account for temporal bias. Specifically, calendar period was included as a categorical variable with two major time groups (1999-2018, 2019-2024).” (lines 128-131, page 4).

“A multivariate Cox proportional hazards regression analysis was performed to identify independent predictors of OS. Receipt of immunotherapy at any line of treatment remained independently associated with improved OS (HR: 0.46, 95% CI: 0.36- 0.58; p < 0.001) (Table 3a). Immunotherapy use at any line showed independent improvement in OS when accounting for landmark analysis (HR: 0.53, 95% CI: 0.41–0.69, p < 0.001 (Table 3b). Similarly, receipt of immunotherapy in the first-line treatment was independently associated with improved OS (HR: 0.68, 95% CI: 0.53-0.87; p = 0.002) (Table 3c)” (lines 225-231, page 8).

“Patients exposed to immunotherapy demonstrated a doubling of median OS compared with those who never received immunotherapy (17.5 vs. 8.6 months), an effect that remained robust in multivariate analysis after accounting for age, sex, histology, stage, performance status, smoking status and calendar period” (lines 273-277, page 10).

“Differences in treatment availability over time, incomplete molecular profiling, especially during the early years, and the absence of systematic adverse event data further limit generalizability of the findings. Another important confounding factor was the small subset of patients undergoing PD-L1 testing, which in combination with the diverse molecular testing may have influenced treatment selection and outcome interpretation” (lines 349-354, page 12).

Comments 4: “The multivariable models are underadjusted. Including only smoking, stage, and performance status is not enough. The analysis should at least consider age, sex, histology, PD-L1, metastatic burden, de novo versus recurrent disease, calendar period, center, and other clinically relevant covariates.”

Response 4: We appreciate the reviewer’s assessment. In the revised analysis, we expanded the models to include clinically relevant covariates selected a priori, including age, sex, histology, performance status, stage, and calendar period. Due to incomplete PD-L1 data, this variable was analyzed in a subset of patients with available information (Supplementary Table 3). Additional potential confounders, including metastatic burden, de novo vs recurrent disease, and center were not incorporated in the main multivariable as stage at initial diagnosis was considered to partially capture disease extent. Furthermore, adjustment for center was not performed, given the shared clinical management across participating sites. Nevertheless, these factors are acknowledged as potential sources of residual confounding and are discussed as limitations of the study.

“Multivariable models included age, sex, histology, performance status, smoking status, stage, and calendar period to account for temporal bias. Specifically, calendar period was included as a categorical variable with two major time groups (1999-2018, 2019-2024).” (lines 128-131, page 4).

“Another important confounding factor was the small subset of patients undergoing PD-L1 testing, which in combination with the diverse molecular testing may have influenced treatment selection and outcome interpretation (lines 352-354, page 12).

“Finally, other important factors such as metastatic burden, oncology center and de novo versus recurrent disease could not be assessed, which, may have influenced the outcome analysis.” (lines 360-362, page 12).

Comments 5: “The claim that first-line immunotherapy monotherapy was superior to chemo-immunotherapy should be interpreted with much greater caution. Without adequate adjustment, this result is very likely driven by patient selection rather than treatment effect.”

Response 5: We agree with the reviewer that the observed association between first-line immunotherapy monotherapy and improved survival should be interpreted with caution. In the revised manuscript, all statements implying superiority have been softened to describe associations rather than causal effects. We also explicitly acknowledge the potential influence of residual confounding and patient selection on these findings.

“Interestingly, immunotherapy monotherapy appeared to have higher OS compared to chemo-immunotherapy combinations. This observation should be interpreted cautiously given the retrospective design and potential confounding by indication and patient selection. Patients receiving combination regimens often had lower PD-L1 expression or more aggressive disease, with higher disease burden, while immunotherapy monotherapy was predominantly administered as first-line treatment in patients with PD-L1 ≥50%, reflecting current clinical practice guidelines and potentially explaining the observed differences in outcome” (lines 255-262, page 10).

Comments 6: “Please clarify exactly who was included in the 610-patient outcome cohort and how the exclusions were applied. The current description is not fully consistent across the methods, figure, and supplementary material.”

Response 6: Per the reviewer’s comment we have described the inclusion of patients in detail.  We have clarified the inclusion and exclusion criteria in the Methods section. Specifically, all patients with advanced NSCLC seen at participating centers were initially considered. Patients with EGFR or ALK mutations and/or use of targeted therapy were excluded, resulting in the final 610-patient outcome cohort.

“Tumor molecular profiling data refer to all patients of the cohort with advanced disease, whereas patients harboring EGFR or ALK mutations and/or receiving standard targeted therapy were excluded from outcome analyses, while a small number receiving other targeted agents were retained (Figure 1, Supplementary Table 1). Patients were also excluded from the analysis if they had not received any systemic treatment” (lines 101-106, page 3).

Comments 7: “There is an important inconsistency regarding targeted-therapy exclusions. The main flow diagram indicates exclusion of patients with EGFR/ALK mutations and/or targeted therapy from the outcome analysis, yet the supplementary note states that some patients receiving targeted agents were not excluded. This must be reconciled clearly.”

Response 7: We thank the reviewer for highlighting this important point. In the main flow diagram, patients with EGFR or ALK mutations and/or receiving standard targeted therapies were excluded from the outcome analysis. However, a small number of patients received other targeted agents—specifically lapatinib, cabozantinib, nintedanib, niraparib, and sotorasib—and were not excluded, because in routine clinical practice these therapies are used off-label or in clinical trial settings for NSCLC.

Further information on targeted regimens is shown in Supplementary Table 1.

Supplementary Table 1. Targeted Therapy Regimens

Targeted Therapy Regimen	Number of Patients
Afatinib	34
Osimertinib	22
Erlotinib	11
Gefitinib	6
Crizotinib	3
Sotorasib	2*
Alectinib	1
Lapatinib	1*
Cabozatinib	1*
Nintedanib	1*
Nipararib	1*

*The patients receiving these targeted agents were not excluded from the outcome analysis

 

Comments 8: “Please carefully recheck Table 2 percentages, because at least one percentage appears incorrect.”

Response 8: We thank the reviewer for noting this issue. All percentages in Table 2 have been carefully rechecked and corrected where necessary.

Comments 9: “The manuscript states that TTNT outcomes were consistent across PD-L1 subgroups, but then refers to numerically longer PFS. Since PFS was not reported as an endpoint, this should be corrected.”

Response 9: We thank the reviewer for pointing out this inconsistency. The manuscript has been corrected to ensure that all references to PFS are replaced with TTNT, which was the actual endpoint analyzed.

“TTNT outcomes were consistent across PD-L1 expression subgroups, with numerically longer TTNT observed in patients with PD-L1 ≥50% (lines 223-224, page 8).

Comments 10: “Because only a minority of patients had PD-L1 testing, and molecular profiling was incomplete overall, the treatment-effect interpretation is limited. This should be emphasized more strongly in both the discussion and the conclusion.”

Response 10: We thank the reviewer for highlighting this point. We have now emphasized in both the discussion and the conclusion that only a minority of patients had PD-L1 testing, and molecular profiling was incomplete overall. We explicitly stated:

“Another important confounding factor was the small subset of patients undergoing PD-L1 testing, which in combination with the diverse molecular testing may have influenced treatment selection and outcome interpretation” (lines 352-354, page 12).

“However, these results needs to be interpreted with caution due to the limited availability of PD-L1 testing and molecular profiling, which may have influenced treatment selection” (lines 372-374, page 12).

Comments 11: “Please clarify whether the 24 untreated patients were included in the outcome analysis and, if so, in which comparator group. Their inclusion could materially bias the results.”

Response 11: We thank the reviewer for this comment. The 24 patients who did not receive any systemic therapy were not included in the outcome analysis. We have now clarified this in the Methods section and in the flow diagram to ensure transparency regarding the composition of the outcome cohort.

“Patients were also excluded from the analysis if they had not received any systemic treatment” (lines 105-106, page 3).

“In this analysis patients not receiving any systemic treatment were not included in the analysis” (lines 116-117, page 3).

Comments 12: “The use of TTNT as a surrogate for PFS is acceptable in some real-world settings, but it should be discussed more carefully because TTNT is influenced by physician practice, access, toxicity, and patient preference, not only progression.”

Response 12: We thank the reviewer for highlighting this important point. We have now expanded the Discussion to more carefully describe the limitations of using TTNT as a surrogate for PFS.

“In our study, TTNT was used as a pragmatic real-world surrogate for progression-free survival [7], with the understanding that it can be influenced not only by disease progression but also by physician practice, patient preference, treatment access, and toxicity.” (lines 264-266, page 10).

Comments 13: “The KM plots should be improved both analytically and visually. Please report the corresponding hazard ratios with 95% confidence intervals and p-values for each comparison, within the figure panels. At present, the survival curves are shown without enough statistical detail to properly interpret the magnitude and significance of the observed differences.”

Response 13: We thank the reviewer for this comment. The KM plots have been updated both analytically and visually. Hazard ratios with 95% confidence intervals and p-values for each comparison have been included directly within the figure panels. The figures have also been improved for clarity, including higher resolution, clearer legends, and labels, and the number of patients at risk is now displayed below each curve.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Editor's,

I have carefully reviewed the revised manuscript, which incorporates the suggestions and comments provided during the peer review process. After evaluating it, I am pleased to confirm that the authors have satisfactorily addressed all the major and minor concerns. The revisions have enhanced the work's clarity, rigor, and overall quality, making it suitable for publication in your esteemed journal. The manuscript now meets the standards expected for publication, and I recommend its acceptance without further modifications.

Thank you for the opportunity to contribute to the review process. Please let me know if you need any further input.

Author Response

We appreciate the reviewer 's assessment. 

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have adequately addressed my previous comments, and the revised manuscript is improved. However, it would further strengthen the presentation if the authors added the percentages for each treatment category directly to the stacked bar plot in Figure 2. Including the percentage values for each color-coded therapy segment would improve the clarity and interpretability of Figure 2.

 

Author Response

Comment: The authors have adequately addressed my previous comments, and the revised manuscript is improved. However, it would further strengthen the presentation if the authors added the percentages for each treatment category directly to the stacked bar plot in Figure 2. Including the percentage values for each color-coded therapy segment would improve the clarity and interpretability of Figure 2.

Reply: We thank the reviewer for the comment. We have revised the figure as proposed to include the respective percentages (page 7, line 184).