Semaglutide and retinal vascular events

A 2026 scoping review counted 15 retinal vascular events among 17,478 people assigned to semaglutide and four among 17,334 assigned to placebo across 13 randomized trials^[1]. The numerical imbalance is a safety signal worth investigating. It is not proof that semaglutide caused the events, and the very small counts make the size of any risk uncertain.

What the review examined

The authors searched PubMed, Embase, ClinicalTrials.gov and Google Scholar in April 2025 and again in March 2026. They looked for clinical trials of semaglutide in type 2 diabetes or obesity that reported retinal vascular events. These include retinal artery occlusion, where arterial blood flow to the retina is blocked, and retinal vein occlusion, where venous drainage is blocked^[1]. Both can threaten sight and require urgent assessment.

They included 13 randomized trials, most double-blind and placebo-controlled. The review also tracked ischemic optic neuropathy as a reference outcome to see whether its method could detect another rare vascular event affecting the optic nerve. A scoping review maps available evidence and identifies patterns. Unlike a formal meta-analysis, it does not necessarily produce one pooled relative-risk estimate.

The large evidence base came partly from long outcome trials. SELECT enrolled adults with overweight or obesity and established cardiovascular disease but without diabetes and followed cardiovascular outcomes^[2]. FLOW enrolled people with type 2 diabetes and chronic kidney disease and focused on major kidney and cardiovascular outcomes^[3]. Neither trial was primarily designed around retinal vascular events, which affects how completely rare eye outcomes are captured and adjudicated.

The retinal event counts

Across placebo-controlled trials, 15 retinal vascular events were reported with semaglutide and four with placebo^[1]. Those totals are small relative to more than 34,000 participants. They show a numerical pattern, but sparse data are unstable: adding or reclassifying a few events could materially alter the apparent difference.

In SELECT, SOUL and FLOW, the review calculated rates from zero to 0.33 events per 1,000 person-years in semaglutide groups and zero to 0.05 in placebo groups^[1]. Person-years combine the number of participants with their follow-up time. Even the upper reported rate corresponds to roughly one event per 3,000 person-years, so a much larger dataset is needed for a stable comparison.

Trials with active comparators showed rates from zero to 5.99 per 1,000 person-years with semaglutide and zero to 3.52 with comparator medicines^[1]. These ranges should not be compared as if they came from one uniform trial. Study populations, follow-up duration, event definitions and comparator treatments differed.

The optic nerve reference outcome

For ischemic optic neuropathy, the review found six events among 11,757 semaglutide-treated participants and one among 11,105 placebo participants. Active-comparator trials reported two versus one^[1]. Ischemic optic neuropathy is not a retinal vascular occlusion, but it also involves impaired blood supply within the visual system.

Including it helped the authors ask whether searching trial reports could recover a rare ophthalmic signal. It does not make the two diagnoses interchangeable. A retinal artery occlusion, retinal vein occlusion and ischemic optic neuropathy have different anatomy, diagnostic criteria and possible mechanisms.

Why causation remains unresolved

Randomization is a strength because it balances many baseline differences, yet these eye events were not the primary endpoint of the included trials, reports can use inconsistent terms and a trial may record only events meeting a serious-adverse-event threshold. That matters here. The review also selected trials that mentioned usable outcomes, so a missing report does not necessarily mean no event occurred.

Participants with diabetes, kidney disease, obesity and cardiovascular disease already have vascular risk factors. Randomization helps with confounding inside each trial, but combining sparse counts across different populations still requires care. The scoping review did not establish a biological pathway or a dose-response relationship^[1].

Rapid improvement in glucose can also affect diabetic retinopathy in some settings, but that is a separate outcome from retinal vascular occlusion. The new review should not be summarized as evidence that all semaglutide-related eye problems share one cause. Readers can see the broader semaglutide safety overview for context.

What patients should take from it

The review supports further registry studies with standardized diagnoses and enough people to evaluate rare events. It does not support stopping a prescribed medicine without speaking to the prescriber. Sudden painless vision loss, a dark curtain, new field loss or a major change in vision requires urgent medical assessment regardless of which medicine a person uses.

What we don't yet know

We do not yet have a precise relative or absolute risk estimate for each retinal vascular diagnosis. It is unclear whether any association differs by diabetes status, existing retinopathy, kidney disease, dose, speed of metabolic change or treatment duration. Large linked health registries could address those questions, but they will need careful outcome validation.

The trial pattern is therefore neither nothing nor a verdict. It is a low-count imbalance that should be tested with datasets built for rare eye outcomes. Until then, the most accurate message is that a possible signal has been identified and its clinical meaning remains uncertain.

Medical disclaimer: This article is for educational and journalistic purposes only and does not constitute medical advice. Peptides discussed may be classified as prescription medicines or research chemicals depending on your jurisdiction. Always consult a qualified healthcare professional before using any peptide product. PeptideMethods.com does not sell, distribute, or facilitate the sale of any peptide product.