Ocular Effects of GLP-1 Receptor Agonists: A Review of Current Evidence and Safety Concerns

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Authors presented an interesting and generally well written review. Some detailed comments below:

Lines 79-82 repetition
It is possible that patients who use GLP-1 receptor agonists (GLP-1RAs) are more health-conscious and aware of disease risks. Such patients may actively seek alternatives to older treatments, leading them to choose GLP-1RAs, and may also monitor diabetes-related complications more closely than the general diabetes population.
If feasible, the authors could calculate the incidence risk per 1,000 or 100,000 patients per year. This would provide readers with a clearer perspective on the magnitude of the issue.
NAION - why only semaglutide?
Authors could improve the MS by including the mechanistic studies performed in murine models. 
Paper lacks a final part that would summarise the topic and provide some suggestions for future studies and for current clinical practice.
References need corrections - some are repeated. Moreover, for a hot-topic like GLP-1RAs, one would expect a bibliography of twice that size.

Author Response

We sincerely thank the Reviewer for the constructive and insightful comments, which have greatly contributed to improving the clarity, scientific accuracy, and overall quality of our manuscript. Below we provide a point-by-point response, highlighting the changes made accordingly.

Comment 1 (Lines 79–82 – Repetition):
The text presents a repetition when listing the main GLP-1RAs (semaglutide, liraglutide, exenatide, dulaglutide).

Response:
We thank the Reviewer for pointing out this redundancy. To improve readability and avoid unnecessary repetition, we have streamlined the text by keeping a single, concise mention of the main GLP-1RAs.

Modification in the manuscript (Lines 86-93): “Agents such as semaglutide, liraglutide, exenatide, ….”

Comment 2:
It is possible that patients who use GLP-1RAs are more health-conscious and aware of disease risks. Such patients may monitor complications more closely than the general diabetes population.

Response:
We have now acknowledged this potential source of bias in the Discussion and Conclusions section, referring to it as the “healthy user effect.” We also clarified that current retrospective designs cannot fully disentangle drug effects from patient behavior.

Modification in the manuscript:
Added sentence in Discussion (lines 361-365):
“Another potential source of bias is patient selection. Individuals prescribed GLP-1RAs may represent a subgroup of patients who are more health-conscious and proactive in disease monitoring. This so-called ‘healthy user effect’ may partly explain differences in ocular outcomes between GLP-1RA users and the general diabetes population.”

Comment 3:
If feasible, the authors could calculate incidence risk per 1,000 or 100,000 patients per year.

Response:
We appreciate the Reviewer’s suggestion. However, since this is a narrative review and not a primary data analysis, absolute incidence rates could not be directly calculated. To address this point, we clarified in the Discussion that most available studies report only relative risks, and that future population-based studies should provide absolute incidence rates.

Modification in the manuscript:
Added in Discussion (lines 365-371):
“Most available studies report relative measures of risk (e.g., hazard ratios, odds ratios) rather than absolute incidence rates, which limits the possibility of expressing results in terms of events per 1,000 or 100,000 patient-years. Future population-based studies are needed to provide absolute incidence estimates.”

Comment 4:
The Reviewer asks why only semaglutide is discussed.

Response:
To clarify, current evidence specifically linking GLP-1RAs to NAION predominantly involves semaglutide, with both observational studies and pharmacovigilance analyses consistently identifying this molecule. We have now emphasized in the NAION section that the focus on semaglutide reflects the available evidence, and not an intentional exclusion of other GLP-1RAs.

Modification in the manuscript:
Sentence added:
“It should be noted that most of the available evidence on NAION currently involves semaglutide, while data on other GLP-1RAs remain limited.”

Comment 5:
The authors could improve the manuscript by including mechanistic studies in murine models.

Response:
To address this point, we have added a specific section entitled “Mechanistic insights from murine models,” where we summarize the main experimental findings from animal studies. This addition allows us to clearly distinguish preclinical evidence from clinical data and provides a stronger mechanistic framework to support the discussion.

Modification in the manuscript:
A new section “Mechanistic insights from murine models” has been added.

Comment 6:
The paper lacks a concluding section summarizing the topic and providing suggestions for future studies and clinical practice.

Response:
We thank the Reviewer for this valuable recommendation. We have now added a comprehensive “Discussion and Conclusions” section, summarizing the dual effects of GLP-1RAs, the methodological challenges, and outlining directions for future research and clinical implications.

Modification in the manuscript:
A new section “Discussion and Conclusions” has been added (page 7).

Comment 7:
References need corrections as some are duplicated (e.g., Ji Q 2017 appears twice). For such a hot topic, the bibliography should be larger.

Response:
We thank the Reviewer for noting this. We corrected the duplication of Ji Q (2017) and carefully reviewed the reference list to remove repeated entries. In addition, we  expanded the bibliography, now including both recent clinical studies and preclinical mechanistic reports, thereby doubling the number of references compared to the original submission.

 

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript incorporates multiple relevant research studies; however, the integration of findings becomes more powerful through a meta-analytical approach when addressing discrepancies between AMD and NAION, rather than presenting them sequentially. This method allows better resolution of inconsistencies across different studies in the literature. The discussion heavily relies on retrospective cohort studies and pharmacovigilance data; however, it does not adequately address statistical limitations, reporting biases, and methods for confounder adjustment.
The framework of methodology remains undefined, which makes it unclear whether this work follows a narrative review format or if a systematic literature search was conducted. The manuscript would benefit from a methodological section that outlines search terms, database sources, and inclusion criteria to enhance transparency and reproducibility even when a systematic review framework is absent. The authors should explicitly define their methodology for distinguishing between mechanistic and clinical studies.
The manuscript requires improvement through the incorporation of visual and tabular summaries. A comprehensive summary table combining study characteristics, such as population demographics, with study design, principal findings, and limitations would make the content more accessible to readers. A mechanistic diagram showing the proposed ocular effects pathways of GLP-1RAs and a schematic risk-benefit profile for clinical use would create a clear visual summary of the evidence for readers.
The section about AMD presents two conflicting studies, yet it fails to investigate why these studies produce different results. This may be due to differences in diabetes status, baseline AMD severity, and duration of GLP-1RA therapy. The discussion on NAION benefits from a population-based incidence rate comparison, along with incidence rates among users of antidiabetic therapy. A comprehensive glaucoma section requires a functional assessment of visual field progression and optic nerve head OCT changes, alongside incidence data.
The document contains minor formatting issues, including unnecessary space before punctuation marks and inconsistent use of semicolons. The initial definitions of NAION and AMD remain unclear when these terms appear in the text without reintroduction, which creates confusion during transitions between disease topics. The reference list includes appropriate and current sources; however, duplicates exist between Ji Q (2017) references #2 and #3, which should be merged into a single entry.

Author Response

We sincerely thank the Reviewer for the thoughtful and constructive feedback, which has helped us strengthen the scientific rigor, transparency, and readability of our manuscript. Below we respond point-by-point and indicate the corresponding revisions.

Comment 1 :
The manuscript incorporates multiple relevant research studies; however, the integration of findings becomes more powerful through a meta-analytical approach when addressing discrepancies between AMD and NAION, rather than presenting them sequentially.
Response:
We thank the Reviewer for this suggestion. As this work was conceived as a narrative review, a formal meta-analysis was not feasible. Nevertheless, we have revised the AMD and NAION sections to highlight and compare differences some differneces in study design ( respective NAION and AMD sections) 

Comment 2:
The discussion heavily relies on retrospective cohort studies and pharmacovigilance data; however, it does not adequately address statistical limitations, reporting biases, and methods for confounder adjustment.

Response:
We thank the Reviewer for this important remark. Since this is a narrative review, no formal quantitative synthesis or risk-of-bias assessment was performed, as would be expected in a systematic review or meta-analysis. Nevertheless, to improve transparency and balance, we have expanded the Discussion by explicitly acknowledging the main methodological limitations of the included studies. Specifically, we highlight the potential for confounding by indication, the “healthy user effect,” immortal time bias, outcome misclassification from ICD coding, and surveillance bias in retrospective designs. 

Comment 3 :
The framework of methodology remains undefined, which makes it unclear whether this work follows a narrative review format or if a systematic literature search was conducted. The manuscript would benefit from a methodological section that outlines search terms, database sources, and inclusion criteria.
Response:
We thank the Reviewer for this valuable suggestion. We have now clarified in Section 2 (“Methodology”) that this is a narrative review. We also describe the databases used (PubMed and Scopus), the combination of keywords, and the inclusion of both clinical and mechanistic studies. This addition improves transparency and clarifies how studies were selected.

Comment 4 :
The manuscript requires improvement through the incorporation of visual and tabular summaries. A comprehensive summary table and mechanistic diagrams would improve clarity.
Response:
We fully agree with the Reviewer. We have now added a summary table synthesizing key study designs, populations, findings, and limitations, and we also created a mechanistic figure illustrating the proposed ocular effects pathways of GLP-1RAs. These visual elements aim to provide readers with a clearer overview and facilitate interpretation. Furthemore, a Figure has been added in the discussion 

Comment 5 :

The section about AMD presents two conflicting studies, yet it fails to investigate why these studies produce different results. The discussion on NAION would benefit from population-based incidence rate comparisons.A comprehensive glaucoma section requires a functional assessment of visual field progression and optic nerve head OCT changes, alongside incidence data.

Response:
We thank the Reviewer for this observation. Since this is a narrative review, our purpose was not to perform a direct comparison or quantitative synthesis, but to provide context to the available findings. In the AMD section, we have clarified that both studies are based on large administrative databases using diagnostic and treatment codes (ICD and procedure coding), which may partly explain the discrepancies. We also note possible differences in  baseline AMD severity, comparator groups, and treatment duration (line 264-281). Regarding NAION, instead of attempting direct population-based comparisons, we emphasized that the evidence mainly derives from retrospective cohorts and pharmacovigilance reports, which should be interpreted as hypothesis-generating. We highlighted the methodological heterogeneity and stressed the need for future prospective studies with standardized ophthalmic endpoints, in the discussion. inally, regarding glaucoma, we agree that functional outcomes such as visual field progression and optic nerve head OCT changes are essential endpoints; however, current evidence is mainly retrospective and does not provide such detailed data. We have therefore highlighted this limitation in the manuscript  

Comment 7 

The initial definitions of NAION and AMD remain unclear when these terms appear in the text without reintroduction.
Response:
We agree with the Reviewer. To improve clarity, we have expanded the introductory sentences of both the NAION and AMD sections with concise definitions and clinical context, ensuring that readers unfamiliar with these conditions can follow the discussion without confusion.

Comment 8 :
The reference list includes appropriate and current sources; however, duplicates exist between Ji Q (2017) references #2 and #3, which should be merged.
Response:
We thank the Reviewer for pointing this out. We have carefully revised the reference list, removed the duplicate Ji Q (2017) entries, and verified the consistency of all references.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Authors have sufficiently improved the ms. Two final comments:

1. Figure 1. Lines crossing the middle panel render it harder to read, please move the lines behind the panel.
2. Some things are repeated within the conclusions, please double check and correct

Author Response

e sincerely thank the Reviewers for their constructive and insightful comments, which have significantly contributed to improving our manuscript. As suggested, we have revised the figure on the proposed ocular mechanisms of GLP-1 receptor agonists by adjusting the layout to improve readability. In addition, we carefully revised the Discussion and Conclusions section to remove repetitions and to provide a clearer, more concise synthesis of current evidence, with explicit acknowledgment of methodological limitations and potential biases.