No Effect of GLP-1s on Anti-VEGF Injection Frequency for Diabetic Macular Edema

NEW ORLEANS — GLP-1 receptor agonists did not significantly affect anti-VEGF treatment frequency in patients with diabetic macular edema (DME), a large retrospective cohort study showed.

Over 24 months of follow-up, “no sustained or progressive change” in anti-VEGF treatment occurred after patients started a GLP-1 agonist versus those who did not. Moreover, visual acuity and retinal thickness, two key outcomes of DME treatment, did not change appreciably.

The “clinically neutral” findings added to prior studies suggesting that GLP-1 drugs may influence the risk for DME, both increasing and decreasing the risk, reported Preeti Nagalamadaka, a medical student at Warren Alpert Medical School of Brown University in Providence, Rhode Island, at the American Academy of Ophthalmology (AAO) meeting.

“It adds because we’re looking at patients who already have DME,” she said. “[On the basis of these results] DME management should continue to be guided by ocular disease activity and anti-VEGF response. This study provides reassuring data amid ongoing uncertainty about GLP-1 RAs’ [receptor agonists’] ophthalmic safety. Future studies should continue to clarify ophthalmic associations with GLP-1 RAs.”

In response to a question about potential interaction with DME severity, she said that variable was addressed in propensity matching, but planned future studies will examine the issue in more detail.

“We tried to rigorously match all of the factors, demographics and DME risk factors, but there could also be some unmeasured differences,” said Nagalamadaka. “Our primary research question was whether there was a change after GLP-1 RA initiation. So, it doesn’t matter necessarily that there’s a difference before, just the trend after.”

Funneling the Data

Noting that the study relied on the AAO’s IRIS Registry, Jennifer Sun, MD, MPH, of Harvard Medical School in Boston, said that studies based on large datasets are frequently challenged by “a huge funnel as you clean the dataset and you’re getting down to very selected cohorts that may or may not be generalizable to larger groups. Do you have a sense of what your funnel was? You ended up with 10,000 matched pairs. What did you start with?”

“There are millions of patients in the IRIS database,” said Nagalamadaka. “I think we started with hundreds of thousands with DME. They had to have anti-VEGF within a certain time frame and they had to keep taking anti-VEGF for a certain number of months … and making sure there’s enough follow-up. That reduced the cohort. That’s a great point, that it was funneled down for strict inclusion/exclusion criteria.”

Another member of the audience asked if investigators established the pretreatment follow-up period for patients who did not take GLP-1 receptor agonists. Nagalamadaka said patients who initiated GLP-1 agonist treatment were matched with control patients who had DME at the same time.

Three other studies reported at the meeting reflected the ongoing interest in GLP-1 drugs’ potential impact on the eye, as well as the mixed messages that have emerged. One study initially suggested an increased risk of age-related macular degeneration (AMD) in patients on GLP-1 agonists, but after matching and adjustment, the data suggested the drugs may lower AMD risk. Two studies of anterior ischemic optic neuropathy (AION) showed the condition occurs at a younger age among GLP-1 receptor agonist users, and among people who have sequential AION, the second-eye event occurred significantly sooner when the patient was a GLP-1 agonist user.

Previous research on GLP-1 receptor agonists and DME primarily focused on incident disease, Nagalamadaka said. Limited data exist on the association between GLP-1 drugs and treatment burden in DME. For the current study, investigators retrospectively analyzed data for two propensity-matched groups of 9,896 patients with DME. Monthly anti-VEGF injection probability was estimated for the 12 months before the GLP-1 agonist group started the medication and the 24 months afterward.

The primary objective was to determine whether GLP-1 agonist treatment influenced anti-VEGF injection frequency, particularly whether treatment burden decreased.

The data showed no significant variation in the injection trajectory for either group over the 36-month period (interaction coefficients -0.18 to 1.59). Similarly, resulting graphs for visual acuity and central subfield thickness showed no significant differences between the matched groups (-0.05 to 0.04 logMAR and -14.12 to 33.58 µm, respectively).

GLP-1 Agonist Use and AMD, AION

Several recent studies have examined associations between GLP-1 receptor agonist use and AMD, yielding inconsistent results. Controlling for potential confounders — such as diabetes status, AMD subtype, and other medications — might improve data analysis, said Franklin Ye Ruan, a medical student at Carle Illinois College of Medicine in Urbana.

Using the Epic Cosmos electronic records database, investigators found 15 million GLP-1 receptor agonist users, 1.4 million patients with AMD, and 115,056 GLP-1 agonist users who had AMD. An unadjusted analysis produced an AMD odds ratio of 1.68 among GLP-1 drug users, including 1.53 for neovascular (n)AMD and 1.89 for dry AMD.

After adjustment for multiple variables, the odds ratio reversed in some cases. Adjustment for type 2 diabetes produced odds ratios of 0.69 and 0.62 for any AMD and nAMD, respectively. Adjustment for hypertension yielded odds ratios of 0.65 and 0.59, and adjustment for hyperlipidemia produced odds ratios of 0.63 and 0.57. On the other hand, adjustment for body mass index resulted in an OR of 1.17 for nAMD and age adjustment produced an OR of 1.94 for nAMD.

“Our hope is that these findings encourage more rigorous studies of GLP-1 medications and macular degeneration, taking other underlying health conditions into account,” said Ruan.

Two other studies by Carle investigators added more information to the ongoing discussion of GLP-1 receptor agonists’ potential association with AION. Using Epic Cosmos data, they searched for cases of AION during 2016-2025 and identified 2,755 patients who were GLP-1 drug users and 26,828 who were not.

Overall, the age at AION onset did not change during the study period, reported medical student Ashley Tin. However, AION occurred at an average age of 62.98 among GLP-1 agonist users versus 66.45 among non-users (P<0.0001). An analysis of age-specific rates showed that GLP-1 drug users had more than a twofold greater risk of AION for age groups 30-39, 40-44, and 45-49, and a relative risk of 1.80 for age ≥50 (P=0.021 for association by age group). Tin pointed out that the vast majority of the GLP-1 agonist users had traditional risk factors (diabetes and obesity) as compared with the non-users.

The second study examined patients with both eyes affected by AION, which happens within 5 years in about 15% of patients who have first AION, said Tin. A query of the Epic database returned about 22,000 patients with both eyes affected by AION, including 1,440 GLP-1 receptor agonist users. The analysis showed that 37.5% of second-eye events occurred within 3 months among GLP-1 agonist users versus 22.5% of non-users, and 56.8% versus 40.6% for second-eye events within 6 months.

The study suggests a need for vigilance among clinicians caring for patients who have one AION event and also a need to consider the findings within the context of the benefits GLP-1 receptor agonists offer to patients.

“GLP-1 use should be an important part of the medication history in patients presenting with AION, particularly when counseling patients after a first-eye event,” said Tin. “These drugs have important benefits for diabetes, obesity, and cardiovascular health, and our study identifies an association rather than proving that the medications cause AION.”