COPHy 2026: Weighing in on the evidence on GLP-1 agonists and NAION

At the 17th annual Congress on Controversies in Ophthalmology (COPHy) taking place 20-21 March 2026, in Krakow, Poland, Andrew G. Lee, MD, presents the “no” side of the argument regarding whether GLP-1 agonists cause non-arteritic anterior ischemic optic neuropathy (NAION). NAION is described as “a small vessel ischemia to the optic nerve,” and the discussion centers on whether observed associations between GLP-1 agonist use and NAION reflect causation or are confounded by underlying patient characteristics. Lee is chair of the Blanton Eye Institute, Department of Ophthalmology at Houston Methodist Hospital in Houston, Texas, US, and is professor of ophthalmology, neurology, and neurosurgery at Weill Cornell Medicine.

Lee emphasises that current evidence comes primarily from retrospective cohort studies suggesting an association, with reported risk estimates ranging “somewhere between two and four X for diabetics” and “may be as high as seven times higher for patients who are obese.” However, he cautions that these findings cannot establish causation due to inherent limitations in observational designs. A central challenge is that patients prescribed GLP-1 agonists often already possess vasculopathic risk factors, creating overlap between treatment indication and baseline risk.

Association versus causation and study limitations

A key theme in Lee’s argument is the distinction between association and causation. He says, “the answer is no, we do not have evidence for causation yet,” and reiterates that “association is not causation.” He underscores that retrospective cohort studies can suggest relationships but cannot definitively determine whether GLP-1 agonists directly contribute to NAION.

Lee uses an illustrative analogy to reinforce this concept: increased ice cream consumption and shark attacks both occur in summer, yet “even though both of those things are associated, they’re not causally associated.” This analogy is used to demonstrate how a shared underlying factor—in this case, seasonal conditions—can drive two unrelated outcomes.

He also highlights the limitations of current research methodologies. Because NAION is relatively uncommon, “it’s going to require big data” to study it effectively. Randomised controlled trials, while ideal, are impractical in this context: “it would take millions of patients” to adequately power such a study, especially given the long-term exposure required for vasculopathic changes to manifest. As a result, Lee suggests that alternative designs, such as prospective observational studies, crossover analyses, or case-control approaches using large datasets, are more feasible, though still imperfect.

Mechanistic considerations and clinical implications

From a biological standpoint, Lee notes the absence of a clear mechanism linking GLP-1 agonists to NAION. He states, “there’s no really good biologically plausible mechanism… for why these agents would cause NAION,” given that these drugs are intended to improve glycemic control and promote weight loss. While he acknowledges theoretical parallels with phenomena such as diabetic retinopathy—where rapid glucose control may lead to autoregulatory failure—he emphasises that any similar mechanism in NAION remains “theoretic only” and unproven.

Regarding clinical practice, Lee indicates that there are currently no established screening recommendations or identifiable predictive risk factors that would allow ophthalmologists to stratify patients for NAION risk in the context of GLP-1 agonist use. He notes that NAION “seems to be somewhat random,” and even known anatomical risk factors, such as a small cup-to-disc ratio, would not alter treatment decisions.

In patient counselling, he advises a balanced approach grounded in transparency: “There’s an association. We don’t know if it’s causal association or not.” He emphasises shared decision-making, where patients weigh potential risks against the benefits of GLP-1 agonists, including weight loss and improved glycemic control. Ultimately, his position presented at COPHy is that current evidence does not support a causal relationship, and clinical decisions should reflect individualised risk-benefit considerations rather than presumed causality.