Semaglutide Use and Risk of Diabetic Retinopathy and Non-Arteritic Anterior Ischemic Optic Neuropathy in Type 2 Diabetes: A Single-Center Retrospective Cohort Study
- Design
- Retrospective cohort · 100 participants · Hard outcome
- Match to healthy normal-weight adults aged 55–75
- Different populationINDIRECT: every participant had type 2 diabetes. Otherwise the group was close to the target profile (mean BMI 21.51, 17 of 100 with obesity, mean age 59.74), but diabetes is a metabolic disease the target group does not have, and no results are reported for anyone without it.
- Could weight loss explain it?
- PossiblyNo weight or HbA1c change on treatment and no mediation analysis are reported; HbA1c was recorded at baseline only. The authors suggest the retinal findings may reflect rapid glucose lowering rather than the drug itself, which they could not test without serial HbA1c.
- Study tier
- Study tier 4VERY_LOW: a single-centre retrospective cohort of 100 semaglutide users with no comparison group, no NAION events and 13 composite eye complications; uncontrolled and too small for a rare outcome, the single-centre, few-event case the model rates VERY_LOW.
- Assessment
- Version 4 · Two independent AI reviews agreed · Nov 11, 2026
Study facts come from the paper. Population match, weight-loss explanation and study tier are our judgments, made against the reference group of healthy normal-weight adults aged 55–75.
This study reviewed the records of 100 people with type 2 diabetes treated with semaglutide at one hospital in Thailand; their average BMI was 21.51 and 17 of 100 had obesity, and there was no comparison group. No cases of NAION, a sudden loss of blood supply to the optic nerve, were seen over 299 person-years, but the authors say the group was too small to draw conclusions about NAION risk. Advanced (proliferative) diabetic retinopathy developed in 10.00% and diabetic macular swelling in 3.00%; without a comparison group, these figures cannot show whether semaglutide raised or lowered either risk.
01Findings
What the study reported
- Drugs
- Semaglutide
- Comparator
- none (single group of semaglutide users; the only comparisons are between users with and without ophthalmic complications)
- Dose
- subcutaneous 0.25–1.0 mg weekly or oral 7–14 mg daily
- Route
- various
- Treatment duration
- median exposure 30 months (Table 1: mean 29.96 ± 14.38 months); at least 12 months of GLP-1RA therapy required for entry
- Primary outcome
- New-onset NAION (secondary: new proliferative diabetic retinopathy or diabetic macular edema)
- Effect
- No NAION cases over 299 person-years (crude incidence 0 per 1,000 person-years); proliferative diabetic retinopathy in 10.00% and diabetic macular edema in 3.00%; no comparison group
- 95% confidence interval
- 95% CI: 0-6.17 (crude NAION incidence per 1,000 person-years)
- Follow-up
- 299 person-years in total; mean follow-up 35.86 ± 14.82 months (Table 1); Kaplan-Meier analysis to 60 months
- Adverse events
- No NAION events; new proliferative diabetic retinopathy in 10.00% and diabetic macular edema in 3.00% of semaglutide users, with no comparison group
- Limitations
- Small single-centre retrospective cohort of 100 patients with no comparison group and few events; the authors say the sample size precludes definitive conclusions about NAION risk and note intermediate follow-up, possible surveillance bias from uneven examination intervals, no serial HbA1c, a composite endpoint and an exploratory Cox model.
Who was studied
- Mean age
- 59.74 ± 21.13 years (Table 1, total)
- Age min
- 18
- Sex distribution
- 64% women (64 of 100)
- Bmi mean
- 21.51
- Obesity status
- mixed (obesity not required; 17 of 100 had obesity; mean BMI 21.51)
- Diabetes status
- type 2 diabetes required
- Baseline condition
- type 2 diabetes treated with semaglutide (GLP-1RA therapy for at least 12 months), without diabetic retinopathy or macular edema at baseline
- Sample size
- 100
- Inclusion exclusion
- Adults 18 or older with T2DM on a GLP-1RA for at least 12 months; excluded prior NAION or other optic neuropathy, baseline diabetic retinopathy or macular edema, follow-up under 12 months, and laser or anti-VEGF treatment in the prior 12 months
- Normal weight included
- true
Study quality details
- Study design
- Retrospective cohort
- Sample size
- 100
- Randomization
- no
- Blinding
- n/a
- Comparator
- none (single group of semaglutide users; the only comparisons are between users with and without ophthalmic complications)
- Follow up duration
- 299 person-years in total; mean follow-up 35.86 ± 14.82 months (Table 1); Kaplan-Meier analysis to 60 months
- Outcome type
- hard
- Replication
- not assessed
- Consistency with other evidence
- not assessed
- Population applicability
- INDIRECT
- Statistical precision
- CI reported: 95% CI: 0-6.17 (crude NAION incidence per 1,000 person-years)
- Risk of bias
- not assessed
- Peer review status
- yes
02Funding
Funding and conflicts
- Funding
- No funding was received
- Industry funded
- No
- Manufacturer
- not stated in source
- Sponsor role
- not reported
- Author conflicts
- Authors declare no conflicts
- Independent replication
- unknown
Funding is shown on every study and never used to score it.
03Claims
Claims this study bears on
- Semaglutide increases the risk of non-arteritic anterior ischaemic optic neuropathy (NAION).Doesn't directly test the claim
Single-centre retrospective cohort of 100 people with type 2 diabetes treated with semaglutide at Rajavithi Hospital, with no comparison group: no NAION cases were observed over 299 person-years (crude incidence rate 0 per 1,000 person-years; 95% CI: 0-6.17), which the authors say is too small for conclusions about NAION risk; proliferative diabetic retinopathy and diabetic macular edema occurred in 10.00% and 3.00%.
04Source
The source, as retrieved
Abstract
[BACKGROUND] Glucagon-like peptide-1 receptor agonists (GLP-1RAs), including semaglutide, are increasingly adopted for type 2 diabetes mellitus (T2DM) due to improvements in glycemic control and weight loss. Although concerns have been raised about a potential association with non-arteritic anterior ischemic optic neuropathy (NAION), this relationship remains uncertain. [OBJECTIVE AND METHOD] In this single-center retrospective cohort study, we reviewed patients with T2DM treated with semaglutide at Rajavithi Hospital. A Kaplan-Meier time-to-event analysis was conducted to estimate the incidence of NAION and proliferative diabetic retinopathy (PDR) progression. [RESULTS] Among 100 patients with T2DM treated with semaglutide, no cases of NAION were observed over 299 person-years of follow-up (crude incidence rate: 0 per 1,000 person-years; 95% CI: 0-6.17). PDR and diabetic macular edema (DME) were the main findings, accounting for 10.00% and 3.00%, respectively. Ophthalmic complications group (13 patients) demonstrated higher body mass index (23.72 ± 1.38 vs 21.18 ± 2.51 kg/m2; p < 0.001), higher fasting plasma glucose (197.77 ± 48.28 vs 148.70 ± 27.84 mg/dL; p < 0.001), and worse baseline best-corrected visual acuity ([BCVA]; 1.71 ± 0.74 vs 0.45 ± 0.39 logarithm of the Minimum Angle of Resolution [logMAR]; p < 0.001). Final BCVA remained significantly worse in the complication group (p < 0.001). Ophthalmic complication remained stable over the 60-month follow-up; longer diabetes duration (>60 months) was modestly associated with better survival (log-rank p = 0.049). By contrast, other demographic, optic disc characteristics, and concomitant systemic vascular diseases were not significantly linked to ophthalmic complication. [CONCLUSIONS] No NAION events were observed over 299 person-years; however, the limited sample size precludes definitive conclusions about NAION risk. A modest trend toward higher proliferative diabetic retinopathy rates suggests that ophthalmic monitoring may be warranted in at-risk patients. Larger prospective studies are needed to confirm these preliminary findings.
Where this record came from
| Source | Retrieved | Identifier |
|---|---|---|
| pubmed | Oct 18, 2026 | 42890826 first ingestion |
| pubmed | Oct 25, 2026 | 42890826 duplicate matched on doi |
| owner-review | Nov 6, 2026 | reviewed_by set to Carrie Schluter (2026-11-05) on the owner's own confirmation. The assessment is unchanged and still the auto-classifier's, so its extracted values stay withheld under the 2026-10-18 decision. |
| reextract | Nov 6, 2026 | dose: '27.84 mg' -> None. Q-043: manufacturer and drug_details.dose re-extracted by auto-classifier-v1.2. manufacturer no longer reads companies named only in the authors' conflicts statement (IN-062) and refuses a multi-company roster; dose is tied to the GLP-1 agent by name, never reads a concentration, and records a multi-dose design whole. Every other value is carried forward unchanged. Owner decision IN-064 option 1, 2026-11-06. |