The Host Tissue Repair Vulnerability Index (HTRVI): A Composite Biomarker for Predicting Osteoradionecrosis in Locally Advanced Nasopharyngeal Carcinoma
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe manuscript presents an interesting and clinically relevant study evaluating the Host Tissue Repair Vulnerability Index (HTRVI) as a novel composite biomarker for predicting osteoradionecrosis (ORNJ) in patients with locally advanced nasopharyngeal carcinoma treated with definitive chemoradiotherapy. The topic is timely, the rationale is biologically plausible, and the statistical analyses are generally appropriate. The manuscript is well written and the results are clearly presented. Nevertheless, several issues should be addressed before the manuscript can be considered for publication.
Major Comments
1. Novelty and positioning within the existing literature
The authors propose HTRVI as a novel host-related biomarker integrating inflammation, immune status, nutritional reserve, and tissue oxygenation. While this concept is interesting, the Introduction and Discussion focus almost exclusively on blood-derived biomarkers. Contemporary personalized management of nasopharyngeal carcinoma increasingly relies on the integration of circulating biomarkers with advanced functional imaging, particularly [18F]FDG PET/CT, which has demonstrated predictive value for treatment response, prognosis, and risk stratification.
The discussion would therefore benefit from a broader perspective on multimodal biomarkers. In particular, the authors should discuss the complementary role of metabolic imaging in the identification of high-risk patients and cite the following recent study:
Quartuccio N, Sireci F, Pulizzi S, Nicolosi S, D'Oppido D, Ialuna S. Predictive Value of [18F]FDG PET/CT for Neoadjuvant Chemoradiotherapy Response in Nasopharyngeal Carcinoma. J Clin Med. 2025;14(18):6508.
Although this study focuses on treatment response rather than osteoradionecrosis, it provides important evidence supporting the role of PET-derived biomarkers in individualized management of nasopharyngeal carcinoma. Discussing these findings would strengthen the manuscript by placing HTRVI within the broader framework of multimodal predictive models that combine biological and imaging biomarkers.
2. Risk of overfitting
Only 24 ORNJ events were observed. Despite the authors' efforts to limit the number of covariates and perform bootstrap validation, the reported AUC values approaching 1.0 are unusually high and raise concerns regarding model optimism. The Discussion should acknowledge this issue more explicitly and emphasize that external validation is essential before clinical implementation.
3. Clinical applicability of HTRVI
Although HTRVI demonstrates promising predictive performance, the manuscript does not clearly explain how the index could influence clinical decision-making. The authors should provide practical examples describing how patients classified as high risk would be managed differently, including modifications in dental surveillance, preventive interventions, follow-up schedules, or radiotherapy planning.
4. Comparison with existing predictive models
The manuscript compares HTRVI only with Hb and GINI. It would be valuable to discuss how this index compares conceptually with other established inflammatory or nutritional scores (e.g., NLR, SII, PNI, GPS, CAR) and to explain why HTRVI may offer incremental value beyond these existing biomarkers.
Minor Comments
5. Clarification of biomarker construction
The biological rationale for the mathematical formulation of HTRVI should be expanded. Although each component is individually justified, the manuscript does not sufficiently explain why these variables are combined multiplicatively rather than through alternative approaches.
6. Statistical reporting
Please report calibration metrics in greater detail. Although bootstrap validation is appreciated, inclusion of a calibration plot or calibration slope would further support the robustness of the prediction model.
7. Figures
Figures 2–4 are informative, but the legends could be more self-explanatory. The ROC figure would benefit from indicating confidence intervals directly within the plot or legend.
8. Limitations
The limitations section should explicitly acknowledge that laboratory parameters were measured only at baseline. Longitudinal changes during and after chemoradiotherapy may also influence tissue repair capacity and ORNJ development.
9. Language
The manuscript is generally well written. Nevertheless, several paragraphs in the Introduction and Discussion are excessively long and repetitive. Moderate editing would improve readability without affecting scientific content.
Overall Recommendation
This manuscript addresses an important clinical problem and proposes a promising biomarker for individualized ORNJ risk assessment. After revision, particularly by strengthening the discussion of the current literature—including recent evidence on multimodal predictive biomarkers in nasopharyngeal carcinoma—and by better addressing the issues of model generalizability and clinical applicability, the manuscript would represent a valuable contribution to the field.
Author Response
Response to Reviewer 1.
We sincerely thank the reviewer for the careful evaluation of our manuscript and for the thoughtful, constructive comments. We appreciate the recognition of the study's potential contribution as well as the valuable suggestions to improve its clarity, interpretation, and clinical context. We have carefully considered each comment and revised the manuscript accordingly. Specifically, we have strengthened the Discussion by better contextualizing our findings within precision oncology, further clarified the potential clinical implications and limitations of HTRVI, refined figure legends and language where appropriate, and made additional editorial revisions to improve conciseness and readability. We believe these revisions have substantially strengthened the manuscript, and we are grateful for the reviewer's insightful feedback.
Major Comments
Comment 1. Novelty and positioning within the existing literature
The authors propose HTRVI as a novel host-related biomarker integrating inflammation, immune status, nutritional reserve, and tissue oxygenation. While this concept is interesting, the Introduction and Discussion focus almost exclusively on blood-derived biomarkers. Contemporary personalized management of nasopharyngeal carcinoma increasingly relies on the integration of circulating biomarkers with advanced functional imaging, particularly [18F]FDG PET/CT, which has demonstrated predictive value for treatment response, prognosis, and risk stratification.
Response 1. We thank the reviewer for this insightful comment. We agree that the original manuscript focused predominantly on circulating biomarkers and did not sufficiently position HTRVI within the broader context of contemporary multimodal prediction in nasopharyngeal carcinoma. Accordingly, we expanded the Discussion to acknowledge the complementary role of functional imaging biomarkers derived from ^18F-FDG PET/CT, which primarily characterize tumor-related metabolic activity, burden, and biological heterogeneity. We now clarify that HTRVI addresses a distinct but complementary biological dimension by estimating host tissue repair vulnerability through the integration of inflammatory, immune, nutritional, and oxygenation-related factors. We further emphasize that future prediction models combining imaging-derived tumor features with host-related biological and treatment-related variables may provide a more comprehensive framework for individualized risk assessment. However, we also note that such models require dedicated development and validation for normal-tissue toxicity endpoints, including ORNJ, rather than extrapolation from models developed for tumor response or survival. The corresponding discussion has been added, with relevant references.
Comment 2. The discussion would therefore benefit from a broader perspective on multimodal biomarkers. In particular, the authors should discuss the complementary role of metabolic imaging in the identification of high-risk patients and cite the following recent study: Quartuccio N, Sireci F, Pulizzi S, Nicolosi S, D'Oppido D, Ialuna S. Predictive Value of [18F]FDG PET/CT for Neoadjuvant Chemoradiotherapy Response in Nasopharyngeal Carcinoma. J Clin Med. 2025;14(18):6508.
Response 2. We thank the reviewer for this valuable suggestion. In accordance with the reviewer's recommendation, we expanded the Discussion to address the complementary role of functional 18F-FDG PET/CT-derived imaging biomarkers within multimodal prediction strategies for nasopharyngeal carcinoma. Specifically, we incorporated and cited the recommended study by Quartuccio et al. (2025) and discuss how imaging-derived biomarkers primarily characterize tumor-related biological features, whereas HTRVI was specifically developed to estimate host tissue repair vulnerability. We further emphasize that these approaches should be regarded as complementary rather than competing, and that future multimodal models integrating tumor-related imaging features with host-related biological biomarkers may provide a more comprehensive framework for individualized risk assessment.
Comment 3. Although this study focuses on treatment response rather than osteoradionecrosis, it provides important evidence supporting the role of PET-derived biomarkers in individualized management of nasopharyngeal carcinoma. Discussing these findings would strengthen the manuscript by placing HTRVI within the broader framework of multimodal predictive models that combine biological and imaging biomarkers.
Response 3. We thank the reviewer for this insightful observation. We agree that the study by Quartuccio et al., although focused on treatment response rather than osteoradionecrosis, provides an excellent example of how imaging-derived biomarkers can contribute to individualized management of nasopharyngeal carcinoma. Accordingly, the Discussion has been expanded to acknowledge this distinction and to position HTRVI within the broader framework of multimodal prediction. We now emphasize that ^18F-FDG PET/CT-derived biomarkers primarily characterize tumor-related biological features, whereas HTRVI was specifically developed to estimate host tissue repair vulnerability. These biomarkers therefore interrogate complementary biological domains, and future prediction models integrating imaging-derived tumor characteristics with host-related biological biomarkers may provide a more comprehensive approach to individualized risk assessment.
Comment 4. Risk of overfitting: Only 24 ORNJ events were observed. Despite the authors' efforts to limit the number of covariates and perform bootstrap validation, the reported AUC values approaching 1.0 are unusually high and raise concerns regarding model optimism. The Discussion should acknowledge this issue more explicitly and emphasize that external validation is essential before clinical implementation.
Response 4. We thank the reviewer for this important comment. We fully agree that the limited number of ORNJ events warrants cautious interpretation of the model's apparent discrimination despite the use of a parsimonious prespecified modeling strategy and bootstrap internal validation. Accordingly, we have strengthened the Discussion and Limitations sections to explicitly acknowledge the potential for model optimism and coefficient instability associated with event-limited datasets. We now emphasize that bootstrap validation cannot substitute for independent external validation and that the reported predictive performance should be regarded as preliminary. The revised manuscript further clarifies that HTRVI should currently be considered an investigational biomarker requiring confirmation in larger independent cohorts before routine clinical implementation.
Comment 5. Clinical applicability of HTRVI: Although HTRVI demonstrates promising predictive performance, the manuscript does not clearly explain how the index could influence clinical decision-making. The authors should provide practical examples describing how patients classified as high risk would be managed differently, including modifications in dental surveillance, preventive interventions, follow-up schedules, or radiotherapy planning.
Response 5. We sincerely thank the reviewer for this insightful and constructive suggestion. We agree that the potential clinical implications of HTRVI warranted further clarification. Accordingly, we expanded the Discussion to provide practical examples illustrating how HTRVI could potentially contribute to future risk-adapted supportive-care strategies, provided its predictive performance is confirmed through independent external and prospective validation. Specifically, we now discuss that patients identified as having greater host tissue repair vulnerability might undergo more intensive pre- and post-radiotherapy dental assessment, shorter surveillance intervals, earlier management of periodontal or mucosal disease, reinforcement of oral-hygiene measures and nutritional support, and closer monitoring following dental extraction or other mandibular trauma. We further note that HTRVI could be evaluated alongside established clinical and dosimetric variables when considering preventive dental procedures, extraction timing, and mandibular dose-reduction strategies during radiotherapy planning, provided that tumor coverage and oncologic efficacy are not compromised. To avoid overstating the current evidence, we explicitly emphasize that these potential applications remain hypothetical, that no validated HTRVI risk thresholds or interventional studies currently exist, and that further prospective validation will be required before HTRVI can be considered for routine clinical implementation.
Comment 6. Comparison with existing predictive models: The manuscript compares HTRVI only with Hb and GINI. It would be valuable to discuss how this index compares conceptually with other established inflammatory or nutritional scores (e.g., NLR, SII, PNI, GPS, CAR) and to explain why HTRVI may offer incremental value beyond these existing biomarkers.
Response 6. We sincerely thank the reviewer for this thoughtful suggestion. We agree that numerous inflammatory, immune, and nutritional biomarkers, including NLR, SII, PNI, GPS, CAR, and many other composite indices, have been investigated in oncology. However, the primary objective of the present study was not to perform a comprehensive head-to-head comparison across the broad spectrum of existing biomarkers. Rather, this study was specifically designed to develop a biologically constructed composite biomarker and to determine whether integration of hemoglobin with GINI provides incremental predictive information beyond its constituent components. Accordingly, hemoglobin and GINI represented the most appropriate biological and mathematical comparators for evaluating the added value of HTRVI. We believe that expanding the Discussion to compare HTRVI conceptually with numerous additional biomarker indices, none of which were evaluated in the present study, would shift the focus beyond the predefined objectives of the manuscript and could inadvertently imply comparative conclusions that are not directly supported by our data. We agree that comprehensive comparative evaluations of HTRVI against other established biomarker indices would be scientifically valuable; however, such investigations require dedicated study designs and fall beyond the scope of the present work.
Minor Comments
Comment 1. Clarification of biomarker construction: The biological rationale for the mathematical formulation of HTRVI should be expanded. Although each component is individually justified, the manuscript does not sufficiently explain why these variables are combined multiplicatively rather than through alternative approaches.
Response 1. We thank the reviewer for this helpful comment. We agree that the biological rationale underlying the mathematical construction of HTRVI required clearer explanation. Accordingly, during revision we substantially expanded the Methods section to clarify the conceptual framework used in developing HTRVI. Specifically, we now explain that the formulation was biologically prespecified rather than empirically optimized, that each component was selected to represent a distinct biological domain involved in tissue repair (systemic inflammation, immune competence, nutritional reserve, and oxygen-carrying capacity), and that the multiplicative structure was intentionally chosen to reflect the interdependent nature of these biological processes rather than independent additive effects. We also emphasize that HTRVI was designed as a biologically motivated composite biomarker rather than a statistically derived prediction score. We believe these revisions address the reviewer's concern.
Comment 2. Statistical reporting: Please report calibration metrics in greater detail. Although bootstrap validation is appreciated, inclusion of a calibration plot or calibration slope would further support the robustness of the prediction model.
Response 2. We sincerely thank the reviewer for this valuable suggestion. We agree that detailed assessment of model calibration is an important component of prediction model evaluation. Accordingly, we performed bootstrap-based internal validation and reported optimism-corrected performance measures to assess the stability of the model. During revision, the statistical reporting was carefully balanced in response to the reviewers' recommendations to avoid excessive methodological detail while retaining the information most relevant to the interpretation of model performance. Given the relatively limited number of ORNJ events, we believe that additional calibration analyses, such as calibration plots or calibration slope estimates, would be susceptible to instability and potentially overinterpretation. We therefore elected to retain the current validation strategy, while explicitly acknowledging that independent external validation remains essential before clinical application.
Comment 3. Figures: Figures 2–4 are informative, but the legends could be more self-explanatory. The ROC figure would benefit from indicating confidence intervals directly within the plot or legend.
Response 3. We thank the reviewer for this helpful suggestion. The figure legends were carefully reviewed to improve their independent interpretability. The legend of Figure 2 has been expanded to clarify the meaning of the violin distributions, embedded boxplots, and individual observations. The legend of Figure 3 already reports the AUC and corresponding 95% confidence interval for HTRVI, low hemoglobin, and GINI, together with the results of the pairwise DeLong comparisons; these details have therefore been retained. Figure 4 already identifies the fitted probability curve, 95% confidence interval, distributional tick marks, and the overall and nonlinear association tests. We therefore limited the revision to Figure 2 to avoid unnecessary repetition and preserve figure clarity.
Comment 4. Limitations: The limitations section should explicitly acknowledge that laboratory parameters were measured only at baseline. Longitudinal changes during and after chemoradiotherapy may also influence tissue repair capacity and ORNJ development.
Response 4. We thank the reviewer for this important observation. We agree that longitudinal changes in inflammatory, nutritional, hematologic, and immune parameters during and after chemoradiotherapy may influence tissue repair capacity and subsequent ORNJ development. This limitation is already explicitly acknowledged in the revised Limitations section, where we state that HTRVI was calculated from laboratory parameters measured only at baseline and therefore does not capture dynamic biological changes occurring during treatment or follow-up. We further note that future prospective studies incorporating serial biomarker measurements will be required to evaluate whether temporal changes in HTRVI provide additional predictive information beyond the baseline assessment. Accordingly, no further manuscript modification was considered necessary.
Comment 5. Language: The manuscript is generally well written. Nevertheless, several paragraphs in the Introduction and Discussion are excessively long and repetitive. Moderate editing would improve readability without affecting scientific content.
Response 5. We thank the reviewer for this helpful suggestion. The Introduction and Discussion were carefully re-examined for paragraph length, repetitive phrasing, and overall readability. Minor editorial revisions were made in both sections to improve conciseness and flow without altering the scientific content or interpretation of the findings. In the Introduction, selected sentences were streamlined to reduce unnecessary wording, while the opening paragraph of the Discussion was condensed to minimize repetition of findings discussed in greater detail in the subsequent paragraphs.
Reviewer 2 Report
Comments and Suggestions for AuthorsI would like to thank the authors for submitting this manuscript. The topic is clinically important, as identifying patients at increased risk of osteoradionecrosis of the jaw (ORNJ) remains a major challenge in the management of patients with locally advanced nasopharyngeal carcinoma. The proposed Host Tissue Repair Vulnerability Index (HTRVI) represents an interesting attempt to integrate routinely available laboratory parameters into a clinically applicable prediction model. Overall, I believe the manuscript is well organized and methodologically stronger than many retrospective biomarker studies. Nevertheless, several issues should be addressed before the manuscript can be considered for publication.
In my opinion, the principal limitation of this study is the lack of external validation. Although the authors performed appropriate internal bootstrap validation, the proposed HTRVI is a newly developed composite biomarker derived and tested within a single retrospective cohort. Consequently, its predictive performance should be interpreted cautiously until confirmed in independent patient populations. I encourage the authors to emphasize this limitation more clearly throughout the Discussion and Conclusions and to avoid implying immediate clinical applicability.
Another issue relates to the relatively limited number of ORNJ events. Although the multivariable analyses appear carefully performed and the modeling strategy is appropriate, prediction models developed from relatively few outcome events remain susceptible to model optimism despite internal validation. The remarkably high discrimination achieved by the final model should therefore be interpreted with caution. I believe this point deserves greater emphasis in the discussion.
The statistical methodology is generally appropriate and represents one of the strengths of the manuscript. However, I would encourage the authors to provide a clearer rationale for the development of HTRVI and to explain more explicitly why this composite index should be expected to outperform its individual components from both biological and statistical perspectives. Although this concept is discussed, additional clarification would improve the overall readability of the manuscript.
The Discussion is comprehensive and generally well balanced. However, several sections describing the potential clinical implementation of HTRVI appear somewhat speculative considering the retrospective design and the absence of external validation. In my opinion, these statements should be moderated, and HTRVI should be presented primarily as a promising investigational biomarker that requires prospective validation before routine clinical use.
I also believe that the manuscript could be improved by simplifying several methodological descriptions. Certain statistical explanations are highly detailed and occasionally interrupt the flow of the manuscript. Condensing these sections would improve readability without compromising scientific rigor.
Finally, although the limitations are appropriately acknowledged, I recommend explicitly discussing the retrospective single-center design, the relatively small number of outcome events, the absence of prospective validation, and the possibility that model performance may differ in external cohorts treated under different clinical protocols.
Overall, I believe this manuscript addresses an important clinical problem and presents a well-conducted exploratory study. The statistical methodology is appropriate, the biological rationale is plausible, and the manuscript has the potential to make a meaningful contribution to the field. However, additional revision is necessary to moderate the interpretation of the findings and more clearly distinguish exploratory results from clinically validated evidence.
Author Response
Response to Reviewer 2
We sincerely thank the reviewer for the careful, thoughtful, and constructive evaluation of our manuscript. We greatly appreciate the positive assessment of the study and the insightful comments, which have helped us improve the clarity, balance, and overall quality of the manuscript. In response to these suggestions, we have revised the manuscript accordingly, with all changes described in detail below.
Comment 1. In my opinion, the principal limitation of this study is the lack of external validation. Although the authors performed appropriate internal bootstrap validation, the proposed HTRVI is a newly developed composite biomarker derived and tested within a single retrospective cohort. Consequently, its predictive performance should be interpreted cautiously until confirmed in independent patient populations. I encourage the authors to emphasize this limitation more clearly throughout the Discussion and Conclusions and to avoid implying immediate clinical applicability.
Response 1. We thank the reviewer for this important comment and fully agree that the absence of external validation represents the principal limitation of the present study. HTRVI was developed and evaluated within a single retrospective institutional cohort, and although bootstrap resampling provides an assessment of internal optimism and model stability, it does not establish transportability to independent patient populations.
Accordingly, we revised the manuscript to emphasize this limitation more clearly in the Abstract, Discussion, and Conclusions. We now state explicitly that the reported predictive performance should be interpreted as internally validated rather than externally generalizable, and that HTRVI should presently be considered an investigational biomarker rather than a tool ready for routine clinical application. We also clarified that independent external validation, followed by prospective evaluation in cohorts treated under different clinical and supportive-care protocols, is required to confirm its reproducibility, calibration, generalizability, and clinical utility.
In addition, statements regarding potential clinical implementation were moderated to avoid implying immediate applicability. The revised Conclusions now emphasize that the present findings are hypothesis-generating and that HTRVI should not be incorporated into routine ORNJ risk assessment or treatment decision-making before successful external and prospective validation.
Comment 2. Another issue relates to the relatively limited number of ORNJ events. Although the multivariable analyses appear carefully performed and the modeling strategy is appropriate, prediction models developed from relatively few outcome events remain susceptible to model optimism despite internal validation. The remarkably high discrimination achieved by the final model should therefore be interpreted with caution. I believe this point deserves greater emphasis in the discussion.
Response 2. We thank the reviewer for this important and well-founded comment. We agree that the limited number of ORNJ events represents a major constraint on the interpretation of the final model, particularly given its remarkably high apparent discrimination. Although we deliberately restricted the number of predictors, used a parsimonious prespecified modeling strategy, and performed bootstrap internal validation, these measures cannot fully eliminate coefficient instability or model optimism arising from only 24 outcome events.
We therefore strengthened the Discussion to state explicitly that the high AUC may partly reflect strong separation within the present cohort, particularly because mandibular mean dose and post-CCRT tooth-extraction burden were powerful predictors. We further clarified that bootstrap correction assesses internal optimism but does not establish that the observed discrimination will be reproduced in independent populations. The internal-validation section was also revised to emphasize that the reported performance metrics should be regarded as preliminary and interpreted cautiously until confirmed in larger external cohorts containing an adequate number of ORNJ events.
Comment 3. The statistical methodology is generally appropriate and represents one of the strengths of the manuscript. However, I would encourage the authors to provide a clearer rationale for the development of HTRVI and to explain more explicitly why this composite index should be expected to outperform its individual components from both biological and statistical perspectives. Although this concept is discussed, additional clarification would improve the overall readability of the manuscript.
Response 3. We thank the reviewer for this constructive comment and for recognizing the strength of the statistical methodology. We agree that the rationale underlying the construction of HTRVI and its potential incremental value over its individual components required clearer explanation.
We therefore expanded the Methods section to clarify the biological and mathematical structure of the index. Specifically, we now explain that inflammatory and innate immune-related variables were placed in the numerator, whereas markers of immune competence, nutritional reserve, and oxygen-carrying capacity were placed in the denominator. Accordingly, HTRVI increases when adverse inflammatory activity rises, protective biological reserve declines, or both occur simultaneously. We also clarified that this structure was biologically prespecified and was not derived through data-driven coefficient optimization.
The Discussion was additionally revised to distinguish the biological and statistical rationale for the observed superior discrimination of HTRVI. Biologically, HTRVI captures the balance between multidomain tissue injury and host reparative capacity, whereas Hb primarily reflects oxygen-carrying capacity and GINI reflects inflammatory and immune-nutritional status without directly incorporating hemoglobin. Statistically, integration of complementary but nonidentical biological information may strengthen the aggregate risk signal, reduce reliance on any single potentially noisy marker, and identify patients with moderate abnormalities across several domains. We also emphasized that composite construction does not inherently guarantee superior performance and that the incremental discrimination observed in this cohort requires confirmation through external validation.
Comment 4. The Discussion is comprehensive and generally well balanced. However, several sections describing the potential clinical implementation of HTRVI appear somewhat speculative considering the retrospective design and the absence of external validation. In my opinion, these statements should be moderated, and HTRVI should be presented primarily as a promising investigational biomarker that requires prospective validation before routine clinical use.
Response 4. We thank the reviewer for this important comment and agree that the potential clinical applications of HTRVI were expressed too assertively in parts of the Discussion given the retrospective single-center design and absence of external validation.
We therefore revised the Discussion to present HTRVI explicitly as a promising investigational biomarker rather than a clinically implementable tool. Statements suggesting that HTRVI could currently guide individualized surveillance, dental management, or preventive interventions were moderated. The revised text now clarifies that the present findings do not support the use of HTRVI for directing individual patient management and that any proposed clinical applications remain hypothetical.
We further state that, if independently and prospectively validated, HTRVI may warrant evaluation as a complementary biological component of multivariable ORNJ risk models integrating established clinical, dental, and dosimetric factors. We also emphasize that reproducibility, calibration, incremental clinical value, and treatment utility must be demonstrated in external cohorts before HTRVI can be considered for routine clinical use.
Comment 5. I also believe that the manuscript could be improved by simplifying several methodological descriptions. Certain statistical explanations are highly detailed and occasionally interrupt the flow of the manuscript. Condensing these sections would improve readability without compromising scientific rigor.
Response 5. We thank the reviewer for this helpful suggestion and agree that some methodological explanations were more detailed than necessary and occasionally interrupted the flow of the manuscript.
We therefore condensed the Endpoints and Statistical Analysis section while retaining all information required for methodological transparency and reproducibility. In particular, we shortened the descriptions of the study objectives, descriptive comparisons, ROC analysis, multivariable modeling strategy, restricted cubic spline analysis, and bootstrap validation. Redundant explanations regarding continuous modeling and the separate evaluation of Hb, GINI, and HTRVI were also removed or consolidated.
In addition, the explanatory notes accompanying Tables 3 and 4 were simplified to avoid repetition of information already provided in the Methods. These revisions improve readability while preserving the prespecified modeling strategy, statistical rigor, and reproducibility of the analysis.
Comment 6. Finally, although the limitations are appropriately acknowledged, I recommend explicitly discussing the retrospective single-center design, the relatively small number of outcome events, the absence of prospective validation, and the possibility that model performance may differ in external cohorts treated under different clinical protocols.
Response 6. We thank the reviewer for this valuable recommendation. We agree that the principal limitations of the study should be presented more explicitly and comprehensively. Accordingly, we revised the limitations section of the Discussion to consolidate and expand the discussion of the study's constraints. Specifically, we now explicitly address the retrospective single-center design and its potential for selection bias and residual confounding, the limited number of ORNJ events and the resulting potential for model optimism despite bootstrap internal validation, the absence of prospective and external validation, and the possibility that model performance may differ in independent cohorts managed under different clinical, radiotherapy, dental-management, and supportive-care protocols. We further emphasize that the reported predictive performance should be regarded as preliminary and that HTRVI should be considered an investigational biomarker requiring confirmation through larger external and prospective studies before routine clinical implementation.
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper can be accepted for publication.
