Skeletal Muscle Radiation Attenuation at C3 Predicts Survival in Head and Neck Cancer

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is a timely and well-reported study leveraging C3-level body composition metrics (SMA and AM-RA) extracted with deep-learning pipeline to predict LRC and OS in HNC.

The cohort size (n=904) and use of public datasets are strengths.

Nevertheless, to be much clearer and more reproducible the manuscript would benefit from tightening the prose, harmonizing terminology (CRT/RCT/RTC-, SMA/SMI), resolving inconsistencies in counts and table values, providing additional methodological detail (handling of missing data, PH assumption checks, cut-off justification) and addressing according to the following major comments:

• Table 1 : reconcile the disagree between tummy site counts in table 1 and in the text (larynx vs. nasopharynx; oral cavity %  0.7 vs 0.8)
• Table 3 ("sex" row): please verity HR/CI/p-value , CI 0.79-1.69 does not contain HR 0.46

• Cut-off selection: You used the 25th percentile for low SMA and low SM-RA. Please report the actual cut-off values (cm² and HU), justify the percentile choice (e.g., prior literature vs data-driven ROC/Youden), and provide a sensitivity analysis using alternative thresholds (e.g., tertiles, sex-specific cut-offs).

• Model diagnostics & missing data: State how missing covariates were handled (complete-case vs imputation). Report proportional hazards checks (e.g., Schoenfeld residuals) and model discrimination (e.g., c-index with 95% CI). Clarify whether interactions (e.g., SM-RA×T stage, sex×SM-RA) were explored.

• HPV/p16 confounding: Given the large proportion of unknown HPV (52.1%), please add a sensitivity analysis restricting to cases with known HPV/p16 (oropharynx subset) to test robustness of SM-RA associations.

• L87–91: To enhance the relevance and reproducibility of the study, I recommend updating the TNM classification to the most recent version (8th edition)
• L151–156: Please reconcile site counts between text and Table 1 since text says nasopharynx 66 (7.3%), larynx 34 (3.8%), but table shows larynx 66 (7.3%), nasopharynx 34 (3.8%). 
• L172–174: Choose one of the two oral cavity percentage 0.8% vs 0.7% and be consistent

• L240–251: please clarify SMA vs SMI usage; you cite SMI at L3 but analyze SMA at C3—state explicitly why SMA (not SMI) is appropriate here.

• L252–263: Add one sentence on lack of height/BMI preventing SMI calculation.

• L264–283: The biological plausibility paragraph is strong, but I suggest to consider adding one supporting citation on myosteatosis-related inflammatory milieu in HNC or solid tumors.

• General: Consider to implement with a short clinical implications paragraph (how SM-RA screening could triage nutrition/rehab or treatment intensity).

• L305–325: please add lack of treatment dose/RT plan parameters, scanner/protocol heterogeneity across TCIA datasets, and potential selection bias from cases with missing C3 coverage.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

see 6 page word doc for details and general suggestions. Overall a good mss  but one that can be improved significantly. I look forward to reviewing the revision on this good work.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The authors propose a manuscript about the use of Deep Learning-Assisted Skeletal Muscle Radiation Attenuation using measurements at the level of C3 as prognosticator for patients affected by H&N cancer. The authors analyzed retrospectively a large patient series affected by H&N cancers. Methods are described in details. The conclusions support the use of SMA based on C3 assessment for H&N patients. The manuscript is clear and well written.

 

Two details:

Among limitations the authors should include the retrospective study design

CUP is not reported in the abbreviations

 

Author Response

Comments 1: “Among limitations the authors should include the retrospective study design”

Response 1: We thank the reviewer for pointing this out. We have added the sentence "In addition, our study is retrospective nature, which imposes inherent limitations, particularly the lack of uniform clinical information (e.g., HPV status and anthropometric parameters)." in Limitation section.

 

Comments 2: “CUP is not reported in the abbreviations”

Response 2: We thank the reviewer for this comment. We have now added "CUP, cancer of unknown primary" and "SMD, skeletal muscle density" to the abbriviation list.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Markedly improved.  The 2025 metanalyses likely had not been published at time mss was first being prepared, accounting for the understandable initial omission. They are now nicely incorporated. 

Most of my suggestions are small, but are few are not, such as the simple summary suggestion and I repeat that now Table 1 should contain ONLY SMA and SMRA data, not the clinical correlates (that you can add to the supplementary section if you wish--it is not the topic of our mss, and is distracting from your important point. 

Simllarly, delete sentence, for the same reason
In addition, T stage (T3-T4 166 vs. T0-T2) was associated with a significantly reduced survival in both univariate (HR 2.40, 95% CI 1.82–3.16, p < 0.001) and multivariable analyses (HR 2.15, 95% CI 1.61–2.87, p < 0.001) (Table 1). 

I added the lack of validated cut off points to the Conclusion as you correctly pointed out that as a limitation. Without a cutoff the concept is clear but how to implement is less so.

Do you want to add one clinical sentence that pre-treatment imaging showing a marked concern might warrant consideration of a gastrostomy tube for nutrition in hopes of reducing treatment complications.    You have no data of course, but the implication of the study suggests that is wise, with minimal risk and might be helpful. (There won't be a randomized trial to compare use or non-use as ethically one could argue no one should be in the no G-tube arm.)

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

Responding to the suggestions of different reviewers can be challenging. I think you have integrated your responses thoughtfully and effectively. I agree with the other reviewer regarding cut-off values: glad you added that. I didn’t suggest it because ai didn’t realize you could do so.
I also appreciate your point-by-point responses: makes it easier to re-review…rather than just submitting a revised mss.

I think this mss should be accepted. Great job.