New research released on Wednesday suggests that Ozempic and similar medications may have unexpected side effects. Doctors in Massachusetts have found a link between the drug’s active ingredient semaglutide and a higher risk of a rare condition that can cause blindness called nonarteritic anterior ischemic optic neuropathy (NAION). The findings do not prove a causal connection between the two, however, and more research is needed to know for certain, the researchers say.

The study was conducted by ophthalmologists at Harvard Medical School and Mass Eye and Ear, a specialty hospital focused on eye issues affiliated with Harvard. The team was inspired to look into the potential link after they had come across three patients who began to lose their vision from NAION within a week’s time. This would already be strange, since NAION isn’t very common (estimates suggest that about 6,000 new cases in the U.S. occur annually), but all three patients were also taking semaglutide.

The researchers decided to do a deep dive into their existing data. They analyzed the medical records of about 17,000 Mass Eye and Ear patients dating back to 2017, when the first semaglutide-based medication Ozempic was approved for use as a treatment for type 2 diabetes. Nowadays, Ozempic is also used as an off-label medication for weight loss along with Wegovy, a higher-dose version of semaglutide approved for obesity. Semaglutide and similar drugs have proven to be much more effective at helping people lose weight than diet and exercise alone.

They focused on nearly 1,700 patients who were diagnosed with diabetes or with being overweight/obese but who had no prior history of NAION. Then they tracked NAION diagnoses between similarly matched groups of patients who were prescribed semaglutide or who were prescribed other diabetes or obesity medications.

The patients were tracked for about three years. During that time, 46 NAION cases were diagnosed among the matched groups of patients. But those prescribed semaglutide were significantly more likely to be diagnosed with NAION, the researchers found. This pattern was seen across both groups of patients, though the risk appeared higher for those prescribed the drug for weight loss. Compared to people taking other obesity medications, the risk of NAION was over seven times higher for those prescribed semaglutide.

“This study’s findings suggest an association between semaglutide and NAION,” the doctors wrote in their paper, published Wednesday in the journal JAMA Ophthalmology.

As far as the researchers know, this is the first piece of evidence ever to suggest that semaglutide could raise the risk of NAION. But they’re quick to point out that their work alone cannot show a cause-and-effect relationship between semaglutide and NAION, especially since it’s based on retrospective, observational data. And there are other caveats to the findings.

The data available only tracks whether people were prescribed these medications, for instance, not whether they took them or continued to take them throughout the entire study period. Given its focus, Mass Eye and Ear also sees many more rare eye cases than would be expected in the general population. At the same time, the small number of NAION cases seen in the study makes it difficult to quanify the actual added risk potentially posed by semaglutide (even a few cases in one direction or the other would greatly change the risk ratio). Scientists also don’t fully understand why NAION happens, much less how these drugs could be raising people’s risk of it. That said, semaglutide mimics the hormone GLP-1, which binds to receptors found in the cells of the optic nerve, potentially explaining how complications could arise, the researchers explained.

While NAION is rare overall, it’s the second most common cause of blindness due to optic nerve damage, according to the scientists. And given the surging popularity of these GLP-1 drugs and much higher relative risk of NAION seen in this study, more extensive research is needed to know whether this potential complication is real or not. Larger retrospective studies with data collected from multiple locations, randomized clinical trials that follow patients over time, or analyses tracking reported adverse events of all GLP-1 drugs are examples of studies that would help confirm or refute their findings, the authors say.

“As this was an observational study, future study is required to assess causality,” they wrote.