Sleep Deficit and Inflammatory Markers as Combined Risk Factors for Glaucoma Progression: A Prospective Longitudinal Observational Cohort Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

RE:  Manuscript review  International Journal of Molecular Science 26-4091927

 

            The manuscript entitled “Sleep deficit and inflammation markers as combined risk factors for glaucoma progression: a prospective cohort study” deals with the prospective assessment of the  influence of sleep deficit and pro-inflammatory cytokine levels on retinal nerve fiber layer loss.

 

The study topic is quite interesting and it is of clinical relevance considering that glaucoma pathogenesis is considered to be multifactorial and, although intraocular pressure (IOP) increase represents the most important risk factor for glaucoma onset and progression, several other IOP-independent variables have be identified or are actually under investigation.  Sleep disorders have been associated with glaucoma prevalence whereas pro-inflammatory cytokine levels have been correlated with glaucomatous morphological damage.

 

The study is well designed, described and discussed.

Anyway, some specific issues need to be addressed, which include:

 

1. Introduction, page 1, line 43: Reference n.1 is old and seems to be not appropriate. Please consider to change Reference;
2. In order to help the study results comprehension, the Materials and Methods section should follow Introduction and precede Results;
3. Results section: please do not replicate the numerical data shown in tables and figures into the text. The text of the Results section should report only a comment about the presence or not of statistically significant difference amongst groups;
4. Discussion section: glaucomatous treated group included into the present study received Latanoprost eyedrops. As known prostaglandins may have pro-inflammatory effects. The possibility to found different results with different anti-glaucoma medications should be considered and briefly discussed;
5. Materials and Methods, page 9, lines 318 nd 319: in order to increase the study comprehension, the Pittsburgh Sleep Quality Index and the Epworth Sleepiness Scale should be reported in a Table or as appendix.

Author Response

Thank you for your valuable insights and for taking the time to review our manuscript!

Comment1: Introduction, page 1, line 43: Reference n.1 is old and seems to be not appropriate. Please consider to change Reference;

Response 1: The reference cited contains some information regarding the global burden determined by glaucoma, however you are correct, it is not a leading reference in the field and is more than 10 years old. We have updated the reference list with two new entries to better sustain the text in the article (citations 1 and 2)

New_1: Stuart, K.V.; Pasquale, L.R.; Kang, J.H.; Foster, P.J.; Khawaja, A.P. Towards Modifying the Genetic Predisposition for Glaucoma: An Overview of the Contribution and Interaction of Genetic and Environmental Factors. Mol. Asp. Med. 2023, 93, 101203. https://doi.org/10.1016/j.mam.2023.101203. PubMed PMID: 37423164

New_2: Stein, J.D.; Khawaja, A.P.; Weizer, J.S. Glaucoma in Adults—Screening, Diagnosis, and Management: A Review. JAMA 2021, 325, 164–174. https://doi.org/10.1001/jama.2020.21899. PubMed PMID: 33433580.

Comment 2: In order to help the study results comprehension, the Materials and Methods section should follow Introduction and precede Results;

Response 2: Thank you for your observation, but we have followed precisely the official instructions for authors (available here: https://www.mdpi.com/journal/ijms/instructions) and the official Word template, provided by the International Journal of Molecular Sciences (available here: https://mdpi-res.com/files/word-templates/ijms-template.dot). IJMS uses a different layout than the standard IMRAD order.

Comment 3: Results section: please do not replicate the numerical data shown in tables and figures into the text. The text of the Results section should report only a comment about the presence or not of statistically significant difference amongst groups;

Response 3: We have corrected all duplications regarding numbers present both in text as well as in their respective tables and figures.

Comment 4: Discussion section: glaucomatous treated group included into the present study received Latanoprost eyedrops. As known prostaglandins may have pro-inflammatory effects. The possibility to found different results with different anti-glaucoma medications should be considered and briefly discussed;

Response 4: Thank you for your observation. We agree, Prostaglandin analogues (PGAs) such as latanoprost can be associated with ocular surface hyperemia and local inflammatory changes, and these effects can be more pronounced with preservatives (especially BAK). However, in your study we measured serum IL-6/TNF-α, not tear/aqueous cytokines. PGA-induced inflammation is most consistently documented locally (ocular surface/tears) and do not translate directly to systemic cytokine levels. Moreso, the actual product used (MONOPOST – LINE 290) was chosen specifically because it does not contain any preservatives, so that we can lower even more the possibility to induce inflammation. Having said that, we agree that inflammation due to ocular surface hyperemia can induce a potential influence on our study, therefore we have included this situation as a confounder and highlighted it in the discussions section. The following text was added:

“Because the treated group received Latanoprost, a prostaglandin analogue, medication-related inflammatory effects should be considered. PGAs are well known to cause conjunctival hyperemia and have been associated with changes in ocular surface/tear inflammatory markers, particularly with preserved formulations. Although our inflammatory profiling was performed in serum (IL-6, TNF-α) and thus may be less sensitive to local ocular surface effects, and we especially used a market product that delivers Lantanoprost in a formula without preservatives (Monopost) it still remains possible that different IOP-lowering drug classes could yield different inflammation–progression patterns.”

Comment 5: Materials and Methods, page 9, lines 318 nd 319: in order to increase the study comprehension, the Pittsburgh Sleep Quality Index and the Epworth Sleepiness Scale should be reported in a Table or as appendix.

Response 5: Thank you for your observations. The two mentioned scales do not require permissions to be used in a clinical setting, that is why we were able to use them in our research, however, since they are copyrighted, we do need permission to reproduce them in a journal, especially in a fully open access, like IJMS is, either as a table or as an appendix (these scales are not provided in an editable format). Please clarify if this is needed so we can arrange for official permission to reproduce the scales from their respective copyright owners.

Reviewer 2 Report

Comments and Suggestions for Authors

This study investigates the associations among sleep deficit, inflammatory markers, and RNFL loss. The topic holds notable clinical and academic significance, and the manuscript is overall well-structured. However, the following aspects need to be clarified or revised to enhance the robustness of the research.

 

1. Rigorously speaking, this work is not a prospective study but a retrospective analysis utilizing data collected from an established prospective cohort. A rigorous prospective study design should involve enrolling glaucoma patients stratified by the presence or absence of sleep deprivation at baseline, followed by longitudinal observation of RNFL changes. However, the current study appears to have enrolled a cohort of glaucoma patients without stratifying them by sleep status at baseline; it then performed follow-up and data collection, and subsequently designated sleep status as an influencing factor for RNFL loss during the data analysis phase.

 

2. The Methods section should be reordered to follow the Introduction and precede the Results, which conforms to the conventional logical structure of academic papers.

 

3. It’s recommended to provide baseline data on visual field, visual acuity, and IOP for all enrolled participants.

 

4. As a well-recognized critical risk factor for RNFL loss, IOP has not been adequately addressed in the manuscript. Furthermore, comprehensive information on the status of IOP control during the follow-up period is required to supplement in the manuscript.

 

5. The P-values cited at Line 145 (P=0.349) and Line 149 (P=0.27) both fail to reach the predefined significance threshold of P < 0.05 specified in the Methods section. Does this have an impact on the conclusions presented in the manuscript? Please provide explicit explanations of the statistical implications of these non-significant results, as well as their potential impact on the study’s overall conclusions.

 

6. The manuscript employs a uniform threshold of 8 hours of sleep to define adequate sleep status. Please provide relevant literature to substantiate the rationale for selecting this cutoff value. In addition, it is important to acknowledge that sleep duration requirements typically decrease with age, which renders the 8-hour criterion potentially unsuitable for the elderly subset of the study population.

 

7. Please clarify the detailed calculation method of the composite risk scores and cite supporting references

 

8. The supplementary materials referenced in the manuscript were unavailable for retrieval during the review process. Please verify whether these materials were inadvertently omitted. If they exist, submit them as supplementary files; if not, remove all references to supplementary materials throughout the manuscript.

 

9. Sample size is smaller after stratification. Please clarify whether the post-stratification sample size possesses adequate statistical power, or provide results from a formal statistical power analysis to substantiate the robustness of the subsequent analyses.

 

10. In the Discussion, the study compares its reported RNFL loss rate with values from cited references. But Reference 23 only documents visual field progression rates and contains no relevant RNFL loss data; please verify the suitability of this citation. Additionally, was the RNFL loss rate calculated using the Guided Progression Analysis (GPA), Linear Mixed Effects Model, or other approach? Please supplement this detail in the Methods section. It’s suggested to cite studies adopting the same methods when comparing.

Author Response

Thank you for taking the time to review our manuscript and provide us with such invaluable support!

Comment 1: Rigorously speaking, this work is not a prospective study but a retrospective analysis utilizing data collected from an established prospective cohort. A rigorous prospective study design should involve enrolling glaucoma patients stratified by the presence or absence of sleep deprivation at baseline, followed by longitudinal observation of RNFL changes. However, the current study appears to have enrolled a cohort of glaucoma patients without stratifying them by sleep status at baseline; it then performed follow-up and data collection, and subsequently designated sleep status as an influencing factor for RNFL loss during the data analysis phase.

Response 1: We appreciate the reviewer’s careful distinction. We have revised the wording throughout to describe this work as a “prospective longitudinal observational cohort study” with analysis-stage (post hoc) exposure classification. Participants were enrolled and followed prospectively with repeated OCT and biomarker assessments; however, sleep status was not used to stratify recruitment at baseline. We now explicitly state this in the Methods section and throughout the entire paper. We also clarify that our analyses identify associations between self-reported sleep deficit/inflammatory trajectory and RNFL loss, and we propose future studies with baseline sleep-stratified enrollment (and objective sleep monitoring) to strengthen causal inference.

Comment 2: The Methods section should be reordered to follow the Introduction and precede the Results, which conforms to the conventional logical structure of academic papers.

Response 2: Thank you for your observation, but we have followed precisely the official instructions for authors (available here: https://www.mdpi.com/journal/ijms/instructions) and the official Word template, provided by the International Journal of Molecular Sciences (available here: https://mdpi-res.com/files/word-templates/ijms-template.dot). IJMS uses a different layout than the standard IMRAD order.

Comment 3: It’s recommended to provide baseline data on visual field, visual acuity, and IOP for all enrolled participants.

Response 3: Indeed, having baselines for these parameters is crucial for the overall comprehensiveness of the study. We do have such data, available in the main DB (Excel file). Please advise how should we provide this, as a mini-table with the following details (IOP in mmHG, Visual field: MD (dB) ± SD (and PSD), plus VF test type and Visual acuity: BCVA (logMAR) ± SD (or Snellen converted, upon preference) or as a larger, non-intepreted appendix?

Comment 4: As a well-recognized critical risk factor for RNFL loss, IOP has not been adequately addressed in the manuscript. Furthermore, comprehensive information on the status of IOP control during the follow-up period is required to supplement in the manuscript.

Response 4: We thank the reviewer for emphasizing the importance of IOP as a key determinant of RNFL loss. In response, we have strengthened the manuscript by (1) explicitly describing the IOP measurement schedule at each visit (baseline, 6, 12, 18, and 24 months), and (2) adding a concise summary of baseline IOP and follow-up IOP control by group (Table 2). These additions clarify that IOP remained stable within groups during follow-up while remaining substantially different between groups, which is expected given treatment status and supports interpretation of structural progression beyond pressure control alone.

LINE 99 - added text: “Baseline and follow-up IOP control are summarized in Table 2. Mean follow-up IOP remained stable within groups, while between-group differences persisted in line with treatment status”.

LINE 99 – added a new Table, Table 2. Table 2 now becomes Table 3.

LINE 198 – added text: “Given that IOP is a major determinant of glaucomatous progression, reporting longitudinal IOP control is essential for interpreting structural change; in our cohort, IOP was stable within groups over follow-up, yet RNFL loss differed, supporting the contribution of non-IOP factors examined here”.

LINE 317 – added text: “Intraocular pressure (IOP) was recorded at each study visit (baseline, 6, 12, 18, and 24 months) and summarized both at baseline and across follow-up (mean IOP over 6–24 months) to describe IOP control during the observation period”.

Comment 5: The P-values cited at Line 145 (P=0.349) and Line 149 (P=0.27) both fail to reach the predefined significance threshold of P < 0.05 specified in the Methods section. Does this have an impact on the conclusions presented in the manuscript? Please provide explicit explanations of the statistical implications of these non-significant results, as well as their potential impact on the study’s overall conclusions.

Response 5: Thank you for this important clarification request. The non-significant p-values in our mediation analysis actually support our conclusion that inflammation doesn’t significantly mediate sleep-progression relationship. It is congruent with the interpretation that sleep deficit and inflammation are independet processes, rather than a single mediated pathway. 

We have redesigned Figure 3 as a forest plot that clearly displays the confidence intervals crossing zero and updated the figure caption “Forest plot showing mediation analysis results. Error bars represent 95% confidence intervals. Neither indirect pathway through IL-6 nor TNF-α reached statistical significance, indicating that sleep deficit and inflammation function as parallel rather than sequential risk factors”

Revised Figure 3 is provided in the revised version of the manuscript uploaded into the platoform. 

Comment 6: The manuscript employs a uniform threshold of 8 hours of sleep to define adequate sleep status. Please provide relevant literature to substantiate the rationale for selecting this cutoff value. In addition, it is important to acknowledge that sleep duration requirements typically decrease with age, which renders the 8-hour criterion potentially unsuitable for the elderly subset of the study population.

Response 6: We thank the reviewer for this important point. We have revised the manuscript to clarify that the 8-hour threshold was used as a pragmatic benchmark for defining “sleep deficit/adequate sleep” in secondary stratified analyses, and we now support this choice with consensus sleep-duration recommendations indicating that healthy adults typically require 7–9 h/night, while older adults (≥65 years) are generally recommended 7–8 h/night. We also acknowledge that sleep need may decrease with age; therefore, we added a limitation and state that future studies (and sensitivity analyses) should apply age-specific thresholds and/or analyze sleep duration as a continuous exposure, which avoids arbitrary cutoffs.

LINE 319 – added text (Methods / Laboratory Methods): “Consistent with consensus recommendations that adults typically require 7–9 h/night (and 7–8 h/night in older adults), we used 8 h as a pragmatic benchmark for defining ‘adequate’ sleep in stratified analyses; primary models additionally treated sleep duration as a continuous variable (hours) to avoid reliance on a single threshold [New citation 25, New citation 26]”.

LINE 268 – added text (Discussions / Limitations): “Sleep duration requirements vary across the lifespan; in particular, recommended sleep duration is generally lower in older adults (7–8 h/night). Therefore, a uniform 8-h criterion may misclassify adequacy in elderly participants; future work should apply age-specific thresholds and incorporate objective sleep measures to refine exposure classification [New citation 27]”.

New citation 25: Hirshkowitz, M.; Whiton, K.; Albert, S.M.; Alessi, C.; Bruni, O.; DonCarlos, L.; Hazen, N.; Herman, J.; Adams Hillard, P.J.; Katz, E.S.; et al. National Sleep Foundation’s sleep time duration recommendations: Methodology and results summary. Sleep Health 2015, 1, 40–43. PubMed PMID: 29073412.

New citation 26: Watson, N.F.; Badr, M.S.; Belenky, G.; Bliwise, D.L.; Buxton, O.M.; Buysse, D.; Dinges, D.F.; Gangwisch, J.; Grandner, M.A.; Kushida, C.; et al. Recommended amount of sleep for a healthy adult: A joint consensus statement of the American Academy of Sleep Medicine and Sleep Research Society. Sleep 2015, 38, 843–844. https://doi.org/10.5665/sleep.4716. PubMed PMID: 26039963.

New citation 27: Centers for Disease Control and Prevention (CDC). FastStats: Sleep in Adults (recommended ≥7 h/night). Updated 15 May 2024. (Accessed on 15.01.2026).

Comment 7: Please clarify the detailed calculation method of the composite risk scores and cite supporting references

Response 7: Sure, the composite risk scores were calculated using the formula: Risk_Score = 0.40 × (Sleep_Deficit/4) × 100 + 0.30 × (IL-6/10) × 100 + 0.30 × (TNF-α/5) × 100, where the weights (40%, 30%, 30%) were derived from each factor’s standardized regression coefficient in our multivariable progression models. Division by reference values (4 hours, 10 pg/mL, 5 pg/mL respectively) normalizes each component to a 0-100 scale based on clinically meaningful ranges observed in our cohort. This weighting approach is analogous to established cardiovascular risk scores (e.g., Framingham, SCORE) that combine multiple factors with empirically-derived weights.

The proper citation to sustain this is the following:

Gilles R. Dagenais, Annie St-Pierre, Patrick Gilbert, Benoît Lamarche, Jean-Pierre Després, Paul-Marie Bernard and Peter Bogaty. Comparison of prognosis for men with type 2 diabetes mellitus and men with cardiovascular disease. CMAJ January 06, 2009 180 (1) 40-47; DOI: https://doi.org/10.1503/cmaj.071027

Comment 8: The supplementary materials referenced in the manuscript were unavailable for retrieval during the review process. Please verify whether these materials were inadvertently omitted. If they exist, submit them as supplementary files; if not, remove all references to supplementary materials throughout the manuscript.

Response 8: Indeed, there is a reference in the submitted manuscript for a supplementary file. We apologize for the confusion, an initial draft of the paper included also a supplement material, however in the later versions, and, in the submitted one we have included all the relevant results in the main draft. To clarify this we have deleted the entry in the text pointing to a supplementary material.

Comment 9: Sample size is smaller after stratification. Please clarify whether the post-stratification sample size possesses adequate statistical power, or provide results from a formal statistical power analysis to substantiate the robustness of the subsequent analyses.

Response 9: Yes, we do agree that this imposes to declare the stratified investigation as exploratory and not confirmatory analysis. We have added explicit acknowledgment of reduced power in post-stratified analyses, such as:

LINE 286 – added text (Materials and Methods): “However, post-stratification analyses (risk tertiles: n=12-13 per group; phenotype subgroups: n=1-8 per cell) have reduced power and should be considered exploratory/hypothesis-generating.”

Results (existing text that highlights the caution) - LINE 173: “These subgroup analyses should be interpreted with caution due to small sample sizes in some phenotype strata.”

Comment 10: In the Discussion, the study compares its reported RNFL loss rate with values from cited references. But Reference 23 only documents visual field progression rates and contains no relevant RNFL loss data; please verify the suitability of this citation. Additionally, was the RNFL loss rate calculated using the Guided Progression Analysis (GPA), Linear Mixed Effects Model, or other approach? Please supplement this detail in the Methods section. It’s suggested to cite studies adopting the same methods when comparing.

Response 10:

In the Discussion, the study compares its reported RNFL loss rate with values from cited references. But Reference 23 only documents visual field progression rates and contains no relevant RNFL loss data; please verify the suitability of this citation.

Thank you for this important observation. We verified that De Moraes et al. (Arch. Ophthalmol. 2011;129:562–568) evaluates visual field (VF) progression only and does not report OCT-derived RNFL thinning rates. Accordingly, we have removed this citation from the paragraph where we compare RNFL loss rates across studies.

To maintain an appropriate structural comparison, we have replaced it with an OCT-based longitudinal study that explicitly reports RNFL thinning in both controls and glaucoma: Wessel et al. (IOVS 2013), which reports an estimated annual RNFL decrease of 0.60 μm/year in healthy eyes and 2.12 μm/year in progressing glaucoma (vs 1.18 μm/year in non-progressing glaucoma). We also ensured that De Moraes et al. is cited only in contexts where VF progression/risk factors are discussed, consistent with its scope.

New citation (new reference 23): Wessel, J.M.; Horn, F.K.; Tornow, R.P.; Schmid, M.; Mardin, C.Y.; Kruse, F.E.; Juenemann, A.G.; Laemmer, R. Longitudinal Analysis of Progression in Glaucoma Using Spectral-Domain Optical Coherence Tomography. Invest. Ophthalmol. Vis. Sci. 2013, 54, 3613–3620. https://doi.org/10.1167/iovs.12-9786. PubMed PMID: 23633657.

Additionally, was the RNFL loss rate calculated using the Guided Progression Analysis (GPA), Linear Mixed Effects Model, or other approach? Please supplement this detail in the Methods section. It’s suggested to cite studies adopting the same methods when comparing.

We thank the reviewer for highlighting this explicitly. We used ordinary least squares (OLS) regression for each patient individually, not GPA (Guided Progression Analysis).

LINE 340 - added text (Methods / Statistical Analysis section): “Individual RNFL progression rates (slopes, μm/year) were calculated for each patient using ordinary least squares (OLS) regression of RNFL thickness against time (in years). This approach has been validated against automated trend-based progression algorithms and provides unbiased estimates for short-to-medium follow-up periods”.

Reviewer 3 Report

Comments and Suggestions for Authors

This study confirmed the association between sleep deficit and glaucoma progression. It was a prospective, 24-month cohort study with an appropriate experimental design (19 normal/glaucoma/RNFL samples each, totaling 57 samples). The authors must have put a great deal of effort into the experimental design and cohort management, resulting in optimal results. I have also experienced patient recruitment and healthy subjects in clinical practice, and I understand the challenges of IRB approval, patient persuasion, patient enrollment, and sample collection. While the results are interesting, publication in IJMS will be difficult unless several critical points are addressed.

1. IRB approval number: Lines 401-403p provided the IRB approval and approval number, which was approved in 2025-09. However, Lines 287-294p show patient recruitment from 2022-01 to 2022-03. Was this a post-approval? I suggest mentioning this. And Line 277: You must cite a paper related to the STROBE guidelines.

2. Table 1: A p-value must be obtained through a three-group ANOVA.

3. Line 258: The expression "our relatively young cohort" is scientifically inaccurate. Please provide a reference for comparison with an existing cohort.

4. Objective sleep indicators: Were polysomnography or actigraphy performed? Since the PSQI is a questionnaire-based survey (well-used but lacks objectivity), has it been compared to objective indicators?

5. Lines 266~ "Fifth, serum cytokines may not fully reflect local ocular inflammation." Cytokines exhibit diverse patterns in various body fluids. In the Discussion section, we could suggest simultaneous measurement of aqueous humor IL-6/TNF-α in future studies. If measurement data are available in this study, it would be helpful to provide them.

6. The authors propose IL-6 and TNF-alpha in tear films as biomarkers of glaucoma progression. As far as I know, most glaucoma patients use eye drops for long periods of time. Preservatives can cause surface inflammation and dry eye, and cytokine concentrations can vary depending on various factors, such as individual lifestyle and atmospheric conditions. Lines 254-268 present limitations, but adding the above information would be helpful.

7. Overall figures: The resolution is too low.
Figure 1: This appears to have been drawn using R's ggplot, but the y-axis seems to start at 40, not 0. While this is fine, it would be better to indicate this in the legend, and to display the x- and y-axes in solid black. Individual observations are presented as points, but the three factors are not clearly distinguishable. Could you please make the display a bit bolder?
Figure 2: This is a very good visualization strategy. The RNFL slope is presented as a heatmap scale, with the x-axis representing sleep deficit and the y-axis representing TNF-alpha. The lower the slope, the closer it is to red, effectively representing the risk status. However, can you draw the same figure for cataracts, normal RNFL, and IL-6 in addition to TNF-alpha? This would make Figure 2 richer.
Figure 3: The error bars are too large. What do they mean? Are they CIs? This is a visualization of actual data, but I don't trust it. I suggest organizing this in a different visualization or table format.
Figure 4: The error bars in the first category of the x-axis are strange. The legend should include an explanation of the error bars.

8. If possible, could you provide a schematic diagram that clearly explains the conclusion? This would be helpful for understanding the research and could be widely cited.

Author Response

Thank you for taking the time to review and analyse our manuscript and for providing such useful insights!

Comment 1: IRB approval number: Lines 401-403p provided the IRB approval and approval number, which was approved in 2025-09. However, Lines 287-294p show patient recruitment from 2022-01 to 2022-03. Was this a post-approval? I suggest mentioning this. And Line 277: You must cite a paper related to the STROBE guidelines.

Response 1: The provided approval date is for the article sent for publication and not for the study. As mentioned in LINES 408-411 (the Acknowledgements section) the results provided in this article are part of a PhD study that was developed during the timeframe mentioned in the cohort. For this study we have a different Ethics Committee Approval protocol that was provided in the supplementary materials in the revised upload version. We have incorporated these details to better reflect this. Regarding the STOBE guidelines we have provided, in the supplemental files, a completed STOBE checklist for our article and we have inserted the proper link for the STROBE guidelines (https://www.strobe-statement.org/checklists/).

Comment 2: Table 1: A p-value must be obtained through a three-group ANOVA.

Response 2: Surely, thank you for pointing this out. We have redesigned Table 1 to include a p-value column showing one-way ANOVA results for each baseline variable. We also updated the caption: “Baseline characteristics of study participants by group. Values are mean ± SD. P-values from one-way ANOVA”.

Comment 3: Line 258: The expression "our relatively young cohort" is scientifically inaccurate. Please provide a reference for comparison with an existing cohort.

Response 3: Indeed, the phrasing is not suitable for explaining our results. We have changed the whole phrase to better reflect the age-related issue and we also provided 2 references that include cohorts of glaucoma patients, for contrast: “Notably, RNFL loss in our control group was slower (−0.20 ± 0.10 µm/year) than rates reported in typical age-matched cohorts (≈0.5–0.7 µm/year), which may reflect our cohort’s slightly lower mean age compared with landmark glaucoma studies such as the Early Manifest Glaucoma Trial (EMGT) with 68.1 years [NEW_2] and the Diagnostic Innovations in Glaucoma Study cohort (DIGS/STAGE) with 69.0 years [New_3], the relatively short follow-up interval, and/or measurement variability; therefore, absolute between-group comparisons should be interpreted with caution.”

NEW_2: Leske, M.C.; Heijl, A.; Hyman, L.; Bengtsson, B. Early Manifest Glaucoma Trial: design and baseline data. Ophthalmology 1999, 106, 2144–2153. https://doi.org/10.1016/S0161-6420(99)90497-9. PubMed PMID: 10571351.

NEW_3: Proudfoot, J.A.; Zangwill, L.M.; Moghimi, S.; Bowd, C.; Saunders, L.J.; Hou, H.; Belghith, A.; Medeiros, F.A.; Williams-Steppe, E.; Acera, T.; Dirkes, K.; Weinreb, R.N. Estimated Utility of the Short-term Assessment of Glaucoma Progression Model in Clinical Practice. JAMA Ophthalmol. 2021, 139, 839–846. https://doi.org/10.1001/jamaophthalmol.2021.1812. PubMed PMID: 34110362.

Comment 4: Objective sleep indicators: Were polysomnography or actigraphy performed? Since the PSQI is a questionnaire-based survey (well-used but lacks objectivity), has it been compared to objective indicators?

Response 4: We agree that the PSQI is questionnaire-based and does not provide objective sleep physiology. In this study, polysomnography and actigraphy were not performed (logistical/feasibility constraints in a 24-month ophthalmic follow-up cohort). We have clarified this in the Methods and added an explicit Limitations statement. The reason for resorting to PSQI (besides the logistical reasons)  was that PSQI is a well-validated instrument for detecting poor sleep quality (commonly used cut-off >5 with good sensitivity/specificity), and that prior work shows that agreement with objective measures (PSG/actigraphy) is modest, reflecting that subjective sleep quality and objective sleep continuity capture overlapping yet distinct constructs. However, the comment is valid ant, therefore, we now refer in the paper to “sleep exposure” as “self-reported sleep deficit” or “self-reported poor sleep quality” throughout, and we highlight that future studies should include actigraphy and/or PSG to confirm and refine these associations.

Comment 5: Lines 266~ "Fifth, serum cytokines may not fully reflect local ocular inflammation." Cytokines exhibit diverse patterns in various body fluids. In the Discussion section, we could suggest simultaneous measurement of aqueous humor IL-6/TNF-α in future studies. If measurement data are available in this study, it would be helpful to provide them.

Response 5: We agree with the reviewer that serum cytokines may not fully reflect intraocular inflammatory activity, as cytokine profiles can differ across compartments (serum/tears/aqueous humor) and intraocular cytokines may better capture local ocular biology. Therefore, we have revised the Discussion to (i) explicitly acknowledge this limitation, and (ii) recommend that future studies include simultaneous aqueous humor IL-6/TNF-α measurement (e.g., at the time of clinically indicated cataract or glaucoma surgery) to test concordance between systemic and intraocular inflammation and to refine biomarker-based risk stratification. In the present study, aqueous humor samples were not collected, so these data are not available.

Therefore, we modified the Discussion section that now includes:

“Although we measured serum IL-6 and TNF-α as systemic inflammatory markers, serum cytokines may not fully reflect local ocular inflammation, as cytokine concentrations can differ between systemic circulation and ocular fluids. Prior studies have reported altered cytokine levels within the aqueous humor in glaucoma, including TNF-α and other inflammatory mediators, supporting the relevance of intraocular profiling. Future longitudinal studies should consider paired sampling (serum/tears/aqueous humor) and, where ethically feasible, simultaneous aqueous humor IL-6/TNF-α measurement (e.g., during clinically indicated intraocular surgery) to evaluate compartment concordance and determine whether intraocular cytokines improve prediction of structural progression.”

And we have modified the phrase in the Limitations section to better reflect the differences in cytokines concentrations in different compartments: “Fifth, cytokine profiles can be compartment-specific; serum levels may not reflect intraocular inflammation.”

Comment 6: The authors propose IL-6 and TNF-alpha in tear films as biomarkers of glaucoma progression. As far as I know, most glaucoma patients use eye drops for long periods of time. Preservatives can cause surface inflammation and dry eye, and cytokine concentrations can vary depending on various factors, such as individual lifestyle and atmospheric conditions. Lines 254-268 present limitations, but adding the above information would be helpful.

Response 6: We definitely agree with this assessment that preservatives in eye drops products may and can induce long term surface inflammation that can interfere with adhesion or penetration of other ophthalmic products and can alter some determinations. However, all of our patients in the cohort received as topic treatment the following product: MONOPOST (Romanian product) – that contains as active drug LATANOPROST. We specifically chose this product particularly as it does not contain any kind of preservatives, as stated here: https://www.medicines.org.uk/emc/product/2978/smpc or here: https://www.medicines.org.uk/emc/files/pil.2978.pdf.

Comment 7: Overall figures: The resolution is too low.

Figure 1: This appears to have been drawn using R's ggplot, but the y-axis seems to start at 40, not 0. While this is fine, it would be better to indicate this in the legend, and to display the x- and y-axes in solid black. Individual observations are presented as points, but the three factors are not clearly distinguishable. Could you please make the display a bit bolder?
Figure 2: This is a very good visualization strategy. The RNFL slope is presented as a heatmap scale, with the x-axis representing sleep deficit and the y-axis representing TNF-alpha. The lower the slope, the closer it is to red, effectively representing the risk status. However, can you draw the same figure for cataracts, normal RNFL, and IL-6 in addition to TNF-alpha? This would make Figure 2 richer.
Figure 3: The error bars are too large. What do they mean? Are they CIs? This is a visualization of actual data, but I don't trust it. I suggest organizing this in a different visualization or table format.
Figure 4: The error bars in the first category of the x-axis are strange. The legend should include an explanation of the error bars.

Response 7: Following your helpful comment we have provided better captions and higher DPI image quality and also addressed all concerns regarding the well readability of our plots. We are grateful for your effort to highlight improvement to this paper.

Comment 8. If possible, could you provide a schematic diagram that clearly explains the conclusion? This would be helpful for understanding the research and could be widely cited.

Response 8: We have generated a diagram to better highlight the conclusions. Disclaimer: this diagram has been generated with the aid of ChatGPT 5.2 Thinking model.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

This paper, which I reviewed, has been well-revised. Not only I but also two other reviewers suggested appropriate revisions, and the authors have made the necessary revisions. Figure resolution and several other issues have been resolved. I hope this paper will contribute positively to the scientific community.