Scientific deep-dive
GLP-1 Eye Risks: Diabetic Retinopathy and NAION Are Not the Same Thing
Two distinct eye questions get merged into one headline. Diabetic retinopathy complications are on the semaglutide label and trace to blood sugar falling fast; NAION is a separate optic-nerve signal that Europe's drug-safety committee recommended adding to product information in June 2025, and US labeling still does not mention.
Part of Safety and side effects, separated from the headlines
Headlines about GLP-1 drugs and eyesight are usually welding two unrelated problems together. One is diabetic retinopathy getting worse, which is on the semaglutide label, came out of a cardiovascular trial, and is mostly discussed as a consequence of blood sugar falling fast rather than the drug harming the retina.[1] The other is NAION, a sudden loss of blood supply to the optic nerve, which is a newer and much rarer signal that European regulators decided in June 2025 belongs in product information and that US labeling still does not mention as of October 2026.[9] They involve different tissue, different people and different conversations.
Issue one: diabetic retinopathy complications
This one is old, well documented, and only concerns people who already have diabetes. It traces to SUSTAIN-6, a two-year cardiovascular outcomes trial of semaglutide in 3,297 adults who had type 2 diabetes and a high cardiovascular risk profile. The trial met its cardiovascular goal. It also turned up a retinopathy imbalance: complications defined as vitreous hemorrhage, blindness, or a condition needing an intravitreal injection or laser photocoagulation ran higher on semaglutide, at a hazard ratio of 1.76 with a confidence interval of 1.11 to 2.78.[6]
The Ozempic label carries those numbers in section 5.3, and the split by baseline status is the important part. Complications reached 3.0% on semaglutide against 1.8% on placebo overall. Among people who already had diabetic retinopathy at the start, the figures were 8.2% and 5.2%. Among people with no known retinopathy at the start, they were 0.7% and 0.4%.[1] Almost all of the absolute difference sits with people who arrived with the disease already present.
Why the speed of improvement is the suspect
A post hoc analysis pooled the retinopathy data across the SUSTAIN program and looked for what distinguished the affected patients. There was no imbalance in retinopathy adverse events across SUSTAIN 1 through 5 or the Japanese trials. Within SUSTAIN-6, the analysis attributed most of the effect to the size and speed of the drop in glycated hemoglobin over the first sixteen weeks, concentrated in patients who had pre-existing retinopathy, poor glycemic control at baseline, and insulin treatment.[7]
That is a recognized phenomenon with a long history. Bringing badly controlled blood sugar down quickly can transiently worsen retinopathy, and it has been documented with insulin for decades. The paper's conclusion was that the SUSTAIN-6 finding looks like the same thing, and that the guidance already given for insulin may be appropriate here.[7] The mechanism under discussion, in other words, is the correction rather than the drug.
We would rather not oversell that. It is the leading explanation and it is the one the labels reflect, but the labels themselves stop short of closing the question. Both semaglutide labels state that the effect of long-term glycemic control with semaglutide on diabetic retinopathy complications has not been studied.[1][2] The trial designed to answer it, FOCUS, was still listed as active and not recruiting when we checked, with primary completion estimated for November 7, 2027.[10] So a labeled concern from 2016 will not have a definitive trial answer for at least another year.
The tirzepatide labels say something different, and less
Precision matters here, because search results routinely transplant the semaglutide numbers onto tirzepatide. Mounjaro and Zepbound carry a retinopathy warning with no trial imbalance behind it. Lilly revised both sections in August 2026. They now say two things: temporary worsening of diabetic retinopathy has been reported when glucose control improves quickly, and anyone with a retinopathy history should be monitored for progression.[4][5] The earlier version also said tirzepatide had never been studied in people with proliferative retinopathy, macular edema, or non-proliferative disease needing acute treatment. That sentence is gone, and no rate replaced it.
That is a warning built on a general mechanism and on reports, not on a counted excess against placebo. It is weaker evidence than semaglutide's and it is also a narrower claim. Neither reading favors one molecule over the other, and nobody has run the trial that would.
| Product | Retinopathy section | What it rests on | NAION mentioned |
|---|---|---|---|
| Ozempic (semaglutide injection) | §5.3 Diabetic Retinopathy Complications | A two-year cardiovascular trial, with rates given | No |
| Wegovy (semaglutide injection) | §5.8, restricted to patients with type 2 diabetes | The same trial plus a weight-reduction trial in type 2 diabetes | No |
| Rybelsus and Ozempic tablets (oral semaglutide) | §5.3, cross-referencing the injection trial | The same cardiovascular trial | No |
| Mounjaro (tirzepatide) | §5.7, for patients with a retinopathy history; revised August 2026 | Reported worsening and the rapid-correction mechanism | No |
| Zepbound (tirzepatide) | §5.8, restricted to patients with type 2 diabetes; revised August 2026 | Reported worsening and the rapid-correction mechanism | No |
| Foundayo (orforglipron tablets) | §5.7, restricted to patients with type 2 diabetes | The rapid-correction mechanism; not studied in retinopathy or macular edema needing acute treatment | No |
Issue two: NAION, which is a different organ problem
Non-arteritic anterior ischemic optic neuropathy is not a retina disease. It is a loss of blood supply to the head of the optic nerve, and it typically shows up as painless vision loss in one eye that a person notices on waking. It is uncommon in the general population, it does not usually recover, and its known risk factors include a crowded optic disc that most people have no idea they have.
The signal started with an observation rather than a database. Neuro-ophthalmologists at a Boston referral center noticed cases clustering among patients on semaglutide and went back through their registry of 16,827 patients. In the diabetes group they counted 17 NAION events among semaglutide patients against 6 among comparators, a hazard ratio of 4.28. In the overweight or obesity group it was 20 against 3, a hazard ratio of 7.64. The authors said plainly that the design was observational and that causality was not established.[8]
Those ratios traveled a long way in headlines, and they should be read for what they are. A single institution's neuro-ophthalmology registry is where NAION patients end up by referral, which is a structural reason to expect concentrated numbers. The follow-up work is more informative.
What the larger studies found
- US insurance claims, 482,912 matched pairs. People with type 2 diabetes starting a GLP-1 receptor agonist were compared against people starting an SGLT2 inhibitor. Presumed NAION was higher on the GLP-1 arm, hazard ratio 1.85, with a rate difference of 0.29 cases per 1,000 person-years. The authors described the incidence and the absolute increase as small.[11]
- US veterans, 102,361 people. A target-trial emulation in the Veterans Health Administration compared semaglutide starters against SGLT2 inhibitor starters. NAION ran at 123 per 100,000 person-years on semaglutide against 67 on the comparator, a hazard ratio of 2.33.[12]
- Pooled observational evidence. A 2026 systematic review and meta-analysis put the hazard ratio at 2.17 across five studies, and translated it into an absolute risk of 0.014%, which the authors described as roughly one extra case of NAION per 7,000 people treated per year. They graded the certainty of the evidence as low.[13]
- Randomized placebo-controlled trials, pooled. A pooled safety analysis of 96,829 participants across the manufacturer's liraglutide and semaglutide trials found three confirmed NAION cases on drug and five on placebo, roughly three per 100,000 participant-years against roughly six. Its conclusion was that the trial data do not show an increase and do not suggest a relationship.[14]
So the last of those points the opposite way from the first three, and we are not going to average them into a verdict. Two points about the disagreement are more useful. The trial pool was authored partly by employees of the manufacturer whose products it examined, which is disclosed in the paper and belongs in your reading of it. And randomized trials are poor instruments for an event this rare: eight confirmed cases spread across nearly a hundred thousand participants is not enough to detect a doubling of anything. The observational studies have the opposite problem, in that they can see rare events but cannot rule out that people prescribed semaglutide differ from people prescribed something else in ways no adjustment captures.
Where the regulators are, as of October 2026
This is the part that has moved, and the part most articles get wrong by writing from memory. What follows is what we could verify from primary sources on the date of writing.
Europe has acted. On June 6, 2025 the safety committee of the European Medicines Agency, known as the PRAC, decided the condition belonged in EU product information for Ozempic, Rybelsus and Wegovy at the frequency band the European classification calls very rare, which covers events reaching up to one person in 10,000. The stated basis was several large epidemiological studies pointing to roughly a doubling of risk in adults with type 2 diabetes, working out to about one extra case for every 10,000 person-years of treatment. The clinical instruction attached to it was that sudden or rapidly worsening vision loss goes to a doctor without delay, and that semaglutide is stopped where the diagnosis is confirmed.[9]
US labeling has not followed, at least not yet. On October 17, 2026 we pulled the six current US structured product labels from DailyMed and searched each for NAION, for “optic” and for “ischemic optic neuropathy”. The words appear nowhere in any of them: not in Ozempic at version 20,[1] not in Wegovy at version 19,[2] not in the oral semaglutide label at version 14,[3] not in either tirzepatide label at version 40,[4][5] and not in Foundayo at version 10.[16] Both tirzepatide labels changed in August 2026, and the change touched retinopathy, not the optic nerve. The only vision instruction in US labeling sits under the retinopathy heading and is addressed to patients with type 2 diabetes, telling them to contact their physician if their sight changes.[2] Gradual blur in both eyes usually has a more ordinary cause; see blurred vision on a GLP-1. A person taking Wegovy for weight without diabetes is not the audience of that sentence.
The 78-trial review, and what it could not settle
A 2025 systematic review and meta-analysis pooled 78 randomized trials of semaglutide covering 73,640 participants. On the broad measures it was reassuring: semaglutide neither increased nor reduced the risk of eye disorders overall (OR 1.01, 95% CI 0.91 to 1.12) or of diabetic retinopathy (OR 1.04, 95% CI 0.92 to 1.17), with overall risk of bias rated low.[15]
On NAION it found a significant association, an odds ratio of 3.92, and then said something unusual and important about its own limits. The 95% confidence interval runs from 1.02 to 15.02, barely clearing no-effect at one end and reaching a fifteen-fold risk at the other. A trial sequential analysis indicated the pooled sample was large enough to settle the diabetic retinopathy question but not large enough to settle NAION.[15]
Two people, two different conversations
If you already have diabetic retinopathy
You are the person the labeled warning was written for, and the numbers that apply to you are the 8.2% against 5.2% figures, not the whole-trial average.[1] The useful conversation is about having a documented eye examination before you start, agreeing who is going to check your eyes and how often while your blood sugar is coming down, and asking your prescriber whether the pace of that improvement is something they intend to manage. Every one of those is a clinical decision, and none of them is settled by a page like this one.
If you do not have diabetes or retinopathy
The retinopathy warning is largely not about you. US labeling scopes the Wegovy and Zepbound versions of it to patients with type 2 diabetes explicitly.[2][5] NAION is the question that does apply. It is rare, it is unresolved, and European regulators found it likely enough to put in writing while American labeling has not. If you have had NAION before, or an eye specialist has ever mentioned a crowded or small optic disc, say so before you start, because that is exactly the baseline detail nobody thinks to volunteer.
What a compounded vial changes here
Chemically, nothing above changes. The paperwork around it does. No approved label travels with a compounded preparation, which means no version of the retinopathy section has to reach you and no Medication Guide arrives in the box. Reporting runs thin in the same direction: FDA does not require a 503A compounder to send it your adverse event, so the record of what happens to people on compounded product stays sparser than the record for the approved one. That gap is the subject of what a compounded vial contains.
If you want to know which pharmacy is filling the order, our register of licensed dispensers names one for every seller on the site. Figures move, so the compounded semaglutide board and the price tracker stamp a verification date on each. The neighboring warning people fold into this one is the thyroid box, taken apart on its own page at GLP-1 thyroid cancer risk.
Frequently Asked Questions
References
- 1.U.S. National Library of Medicine OZEMPIC (semaglutide) injection, solution [Novo Nordisk Pharmaceutical Industries, LP] — SPL version 20, effective June 1, 2026 DailyMed Structured Product Label. 2026. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=adec4fd2-6858-4c99-91d4-531f5f2a2d79
- 2.U.S. National Library of Medicine WEGOVY (semaglutide) injection, solution [Novo Nordisk Pharmaceutical Industries, LP] — SPL version 19, effective June 18, 2026 DailyMed Structured Product Label. 2026. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b
- 3.U.S. National Library of Medicine OZEMPIC (oral semaglutide) tablet / RYBELSUS (oral semaglutide) tablet [Novo Nordisk Pharmaceutical Industries, LP] — SPL version 14, effective January 30, 2026 DailyMed Structured Product Label. 2026. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=27f15fac-7d98-4114-a2ec-92494a91da98
- 4.U.S. National Library of Medicine MOUNJARO (tirzepatide) injection [Eli Lilly and Company] — SPL version 40, effective August 27, 2026 DailyMed Structured Product Label. 2026. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=d2d7da5d-ad07-4228-955f-cf7e355c8cc0
- 5.U.S. National Library of Medicine ZEPBOUND (tirzepatide) injection [Eli Lilly and Company] — SPL version 40, effective August 28, 2026 DailyMed Structured Product Label. 2026. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=487cd7e7-434c-4925-99fa-aa80b1cc776b
- 6.Marso SP, et al. Semaglutide and Cardiovascular Outcomes in Patients with Type 2 Diabetes N Engl J Med. 2016. PMID: 27633186.
- 7.Vilsbøll T, et al. Semaglutide, reduction in glycated haemoglobin and the risk of diabetic retinopathy Diabetes Obes Metab. 2018. PMID: 29178519.
- 8.Hathaway JT, et al. Risk of Nonarteritic Anterior Ischemic Optic Neuropathy in Patients Prescribed Semaglutide JAMA Ophthalmol. 2024. PMID: 38958939.
- 9.European Medicines Agency PRAC concludes eye condition NAION is a very rare side effect of semaglutide medicines Ozempic, Rybelsus and Wegovy EMA News, 6 June 2025. 2025. https://www.ema.europa.eu/en/news/prac-concludes-eye-condition-naion-very-rare-side-effect-semaglutide-medicines-ozempic-rybelsus-wegovy
- 10.Novo Nordisk A/S Long-term Effects of Semaglutide on Diabetic Retinopathy in Subjects With Type 2 Diabetes (FOCUS), NCT03811561 — active, not recruiting; estimated primary completion November 7, 2027 ClinicalTrials.gov. 2026. https://clinicaltrials.gov/study/NCT03811561
- 11.Tesfaye H, et al. GLP-1RA and the risk of non-arteritic anterior ischaemic optic neuropathy in patients with type 2 diabetes: A population-based study Diabetes Obes Metab. 2026. PMID: 41104517.
- 12.Heberer K, et al. New-Onset Nonarteritic Anterior Ischemic Optic Neuropathy and Initiators of Semaglutide in US Veterans With Type 2 Diabetes JAMA Ophthalmol. 2026. PMID: 41678180.
- 13.Chrzanowski J, et al. Semaglutide-associated risk of nonarteritic anterior ischemic optic neuropathy in patients with type 2 diabetes: A systematic review and meta-analysis of observational studies PLoS Med. 2026. PMID: 42166479.
- 14.Vilsbøll T, et al. Non-arteritic anterior ischaemic optic neuropathy incidence in placebo-controlled clinical trials of liraglutide or semaglutide Br J Ophthalmol. 2026. PMID: 42049287.
- 15.Natividade GR, Spiazzi BF, Baumgarten MW, et al. Ocular Adverse Events With Semaglutide: A Systematic Review and Meta-Analysis JAMA Ophthalmology. 2025. PMID: 40810985.
- 16.U.S. National Library of Medicine FOUNDAYO (orforglipron) tablet, film coated [Eli Lilly and Company] — SPL version 10, effective July 29, 2026 DailyMed Structured Product Label. 2026. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=8ac446c5-feba-474f-a103-23facb9b5c62
- 17.U.S. Food and Drug Administration Drug Safety Communications FDA. 2026. https://www.fda.gov/drugs/drug-safety-and-availability/drug-safety-communications
Compare the sellers
Related research
Is Compounded Tirzepatide Safe? What FDA Has Actually Reported
FDA has never reviewed compounded tirzepatide for safety, effectiveness or quality. That is a fact about regulatory review, not a finding of harm. This covers what the agency has reported, what separates a 503A pharmacy from a 503B facility, and what nobody can tell you.
11 min read
GLP-1 Thyroid Cancer Risk: What the Boxed Warning Actually Says
A boxed warning about thyroid C-cell tumors sits on every semaglutide and tirzepatide label there is. It comes from rodent studies, the labels themselves say the human relevance is undetermined, and the human research published since does not agree with itself.
11 min read
Which GLP-1s Can Be Compounded? The FDA Shortage and Bulks-List Rules
Both federal routes for compounding semaglutide and tirzepatide depended on FDA shortage status. Both shortages are closed, and FDA has proposed shutting the remaining route.
7 min read
How Much Does Compounded Semaglutide Cost? What Drives the Number
Compounded GLP-1 prices are quoted in ways that hide what you will be charged. This is the structure of the bill, and where on this site the dated numbers live.
9 min read
Compounded Semaglutide vs Ozempic and Wegovy: What Is Actually Different
The molecule inside compounded semaglutide can be the one in Ozempic and Wegovy. What differs is everything around it: how it is formulated, at what concentration, on what dose ladder, and with what adverse-event record behind it.
9 min read
How Long Do Semaglutide and Tirzepatide Side Effects Last?
What predicts side effects is where you are on the ladder, not how many weeks you have been taking it. Both labels place the nausea and diarrhea during escalation, and how long any single dose keeps working comes down to half-life: roughly seven days for semaglutide, roughly five for tirzepatide.
11 min read
Where to get tirzepatide online, safely: sellers our editors have checked
These are telehealth sellers our editors have checked. For each one we hold a price, the form the drug comes in, and the states it reaches.
No insurance needed · vetted by our editors
Some of the links on this page earn us money. If you sign up with a provider after following one, that provider may pay GLP Watchdog a commission. Learn more
DudeMeds
From $233/mo
Get started →TrimRx
From $399/mo
Get started →ShedRx
Oral orforglipron alongside the injectables
From $299/mo
Get started →