In this age of globally accessible information, patients are more careful consumers than ever. So I am bracing myself this month for questions as my patients consider a new drug for diabetes care that was approved by the United States Food and Drug Administration in December 2017. Semaglutide, a subcutaneously injected, long-acting glucagonlike peptide-1 analogue, is indicated as an adjunct to diet and exercise to improve glycemic control among adults with type 2 diabetes. The clinical trial results have been impressive. Significant improvements in reducing hemoglobin A1c and greater weight loss have been demonstrated in randomized controlled trials that compared semaglutide head-to-head with other commonly used antiglycemic medications for type 2 diabetes, including insulin glargine and a same-class glucagonlike peptide-1 receptor agonist. In addition to these health benefits, the fact that this long-acting drug is dosed only once weekly will make it an attractive alternative to many.
Why the concern about possible ophthalmic effects? Under “Warnings and Precautions,” the drug label includes the following language: “Diabetic Retinopathy Complications: Has been reported in a clinical trial. Patients with a history of diabetic retinopathy should be monitored.” Further detail provided in the package insert notes that whereas “diabetic retinopathy complications” occurred at a rate of 3% in patients treated with semaglutide, these events occurred in only 1.8% of placebo-treated patients in a 2 year trial that enrolled type 2 diabetic patients with high cardiovascular risk. The insert also states that patients with vs without a history of diabetic retinopathy experienced a larger risk of complications. Based on these findings, I’ve already been queried by some diabetologist colleagues as to whether they should restrict the use of this medication in their diabetic patients.
But does semaglutide therapy truly worsen diabetic retinopathy outcomes? A deeper dive into the data suggests that the answer is not clear. The methods used to evaluate diabetic eye disease in this cardiovascular-focused trial were sloppy at best. These did not include standardized reporting or photographic assessment of all study participants, but instead relied on patient self-reported or investigator-reported events based on medical records or any form of eye examination (dilated or undilated), which could have been performed by nonophthalmic personnel. With all due respect to our medical colleagues, I place minimal value on an undilated eye examination performed in a primary care setting. Diabetic eye complications were defined as any reported event of vitreous hemorrhage, treatment with retinal photocoagulation or intravitreous injection, or study-defined “diabetes-related blindness” consisting of a single episode of vision loss to 20/200 or worse. None of these events needed routine confirmation by an eye care specialist. Patients were allowed to enroll in the study in the midst of planned ocular treatments, but would then be counted as having an eye complication if they received the next in an ongoing series of intravitreous injections once they entered the study. As a clinical trialist who specializes in research in diabetic retinopathy, I consider the way in which “diabetic retinopathy complications” were defined in this study as making the outcome potentially meaningless.
The data are the data, however, and although I think it is highly possible there is no actual increased retinopathy risk with semaglutide, we must consider the possibility that this finding was not due to chance alone. Some have suggested that perhaps the higher rates of events are due to rapid reductions in hyperglycemia over the first 16 weeks of the trial. A substantially higher risk of diabetic eye disease worsening in patients who rapidly experienced greater decreases in hemoglobin A1c was clearly documented in the landmark Diabetes Control and Complications Trial. Does the early worsening effect apply to patients who improved their glycemic control with semaglutide? A post hoc, exploratory analysis of the study data reveals that patients in both the placebo and semaglutide arms who had greater reductions in hemoglobin A1c had higher rates of retinopathy events. It is worthwhile to note that the group of patients who experienced early worsening in the Diabetes Control and Complications Trial still had better ocular outcomes in the long run than those who did not institute strict glycemic control. Optimistically, one might think that the longer-term benefits of improving glycemic control with semaglutide might actually improve diabetic eye disease outcomes in the long run. However, the available data don’t address this question.
The Food and Drug Administration approved semaglutide without requiring a retinopathy postmarketing study. Nevertheless, a well-designed, prospective study carefully focused on retinopathy outcomes is needed to fully address the possible risk of worsening diabetic retinopathy with semaglutide treatment. For the sake of both better scientific understanding and greater clarity in patient management, I hope that one is performed soon. In the meantime, I will advise my patients and colleagues not to restrict the use of this drug in patients with retinopathy, but to ensure, as always, that patients are being routinely followed by an eye care provider with dilated eye examinations to identify and treat diabetic eye disease in a timely and appropriate manner.