In one randomized trial it did, and the people it happened to are a narrow group. SUSTAIN 6 was a two-year cardiovascular outcomes trial in type 2 diabetes, and in it semaglutide was associated with a significant increase in diabetic retinopathy complications against placebo [1]. A post hoc analysis attributed most of that to how fast and how far HbA1c fell in the first sixteen weeks, among patients who already had retinopathy, had poor control at baseline and were on insulin. Almost every safety question in this library ends with the evidence not supporting the worry, and this one does not. What follows is an attempt to say exactly who it applies to, because the answer is narrow and the alarm is broad. That gap is the one this site keeps finding between what a trial established and what gets repeated about it.
What the trial found
SUSTAIN 6 was a two-year pre-approval cardiovascular outcomes trial in type 2 diabetes, and in it semaglutide was associated with a significant increase in the risk of diabetic retinopathy complications compared with placebo. That is the finding, stated as the investigators state it, and it is the only randomized retinopathy signal in the program.
Across the rest of the program — SUSTAIN 1 through 5 and the Japanese trials — there was no imbalance in retinopathy adverse events. The signal sat in one trial, which is both reassuring and the reason the trial mattered: it was the longest and it enrolled the sickest. The same is true of the kidney trial covered in semaglutide and kidney disease, where the sickest population is also where the effect showed up.
The mechanism, and what it does not do
A post hoc mediation analysis used the initial change in HbA1c at week 16 as a covariate. Most of the effect could be attributed to the magnitude and rapidity of that reduction, and it concentrated in patients who already had retinopathy, had poor glycemic control at baseline, and were being treated with insulin.
Early worsening of retinopathy after a fast improvement in glucose control is a known phenomenon and is documented with insulin. Guidance exists for it there, and the authors suggest similar guidance may be appropriate for semaglutide.
The other eye signal, and why it is separate
Retinopathy is not the optic nerve question. A separate pooling of eight cohorts put the relative effect for non-arteritic anterior ischemic optic neuropathy at 1.93 against no semaglutide, with a 95% confidence interval of 1.22 to 3.08 [2]. In overweight or obese people specifically it came in at 1.68 (95% CI 0.72 to 3.91). Two different structures, two different literatures, and the second one is read here in the optic nerve evidence by indication. Neither excuses the other, and neither is evidence about the other.
Whether it applies to you
Every participant in these trials had type 2 diabetes. Most people buying these drugs from the sellers tracked here do not, and for them there is no HbA1c to fall rapidly and no pre-existing retinopathy for it to worsen. The finding does not transfer to that population, and nothing in this literature suggests it should.
If you do have diabetes and any degree of retinopathy, this is a conversation to have before starting rather than after — with someone who can look at your retinas. That is the part this market is worst equipped for. Most sellers publish nothing about who reviews a case or how to reach them, which we count in what sellers will not tell you and list in the questions they leave open. A condition that needs monitoring is a poor fit for a service that does not describe any.