First approvals for geographic atrophy (GA), puzzling side effects of the new GA drugs, and hits and misses with artificial intelligence (AI) applied to retinal disease were among the top stories in age-related macular degeneration (AMD) during 2023.
First Approved Therapy for GA
The year had barely begun when the FDA approved pegcetacoplan (Syfovre) for GA secondary to AMD. Despite an estimated prevalence of at least 1 million adults in the U.S., ophthalmologists had only off-label treatment options, which were largely ineffective.
GA arises from inflammation-driven cumulative damage to the retina, involving multiple regulatory pathways including the complement system. Pegcetacoplan targets the C3 protein, a key driver of the complement cascade.
Support for the approval came from two randomized, sham-controlled phase III trials involving a total of 1,258 patients with confirmed GA. Treatment with pegcetacoplan decreased GA progression by an average of 22% and by as much as 30% as compared with sham injections. Statistically significant differences occurred at 12 and 24 months.
"Until now, there have been no approved therapies to offer people living with GA as their vision relentlessly declined," said Eleonora Lad, MD, PhD, of Duke University Medical Center in Durham, North Carolina, in a from Apellis Pharmaceuticals. "With Syfovre, we finally have a safe and effective GA treatment for this devastating disease, with increasing effects over time."
Second Approved Option for GA
A condition that began the year with no approved therapies had two approved options before the summer ended, as the FDA greenlighted avacincaptad pegol (Izervay).
Also working through the complement system, avacincaptad pegol targets the C5 protein. In two phase III sham-controlled trials, treatment with the C5 inhibitor slowed GA progression by as much as 35% after 12 months.
"As a C5 inhibitor, Izervay has shown to slow GA progression by targeting the source of retinal cell death and may preserve the upstream benefits of the complement system," said Arshad Khanani, MD, of Sierra Eye Associates in Reno, Nevada, in a from Astellas Pharma and Iveric Bio. "The FDA approval of Izervay is great news for the retina community and our patients suffering from GA."
The two phase III trials investigated the effects of avacincaptad pegol on GA after 6 and 12 months of treatment. Each of the trials showed that the C5 inhibitor slowed GA progression at both assessments, reaching a maximum difference of 35% from sham control at 12 months (P<0.01).
Puzzling Effects of GA Drug
Reports of puzzling ocular effects associated with pegcetacoplan added a bump in the road for the otherwise smooth news ride for the two new GA therapies.
A report at the American Society of Retina Specialists (ASRS) meeting summarized 21 cases (at that time) of unexplained intraocular inflammation in patients treated with pegcetacoplan. The case classifications (some of which had more than one) included 17 cases of panuveitis, seven of retinal vasculitis, and seven of occlusive vasculitis. All the cases arose 7 to 13 days after the initial intravitreal injections, the etiology remained unclear, and patient outcomes continued to evolve.
"In summary, we don't know what this is," said Andre Witkin, MD, of the New England Eye Center of Tufts Medical Center in Boston, and chair-elect of the ASRS Research and Safety in Therapeutics Committee. "We don't know what the etiology is. We will keep you all informed of any new details that we may be able to provide."
In a press release that coincided with Witkin's report, drug manufacturer Apellis Pharma stated that there is "no indication that drug product or manufacturing issues contributed to rare events of retinal vasculitis...Zero events were reported in clinical trials, following more than 23,000 clinical trial injections to date."
About 2 months later, an article in JAMA Ophthalmology provided details about a small number of cases of symptomatic eye floaters associated with pegcetacoplan injections. A retrospective review of 55 patients treated with the C3 inhibitor showed that 16 developed floaters 2 to 4 weeks after the initial injection, including 14 cases of new symptomatic floaters.
Unlike the report on ocular inflammation of unidentified etiology, the floaters were presumed to be associated with silicone oil droplets in the vitreous cavity. Similar cases have been reported in patients who received intravitreal injections of other medications. No patients in the series had signs of ocular inflammation, infection, or decreased visual acuity.
"It's not debilitating, but it's an annoyance," said Amr Dessouki, MD, of Retinal Diagnostic Center in Campbell, California, and first author of the report. "Imagine seeing bubbles floating around in your eye."
Dessouki told ֱ that he has stopped using pegcetacoplan, at least for the time being, and encouraged ophthalmologists to inform patients about the potential side effect.
Mixed Results with AI for Ophthalmology
A machine-based learning system for grading retinal images outperformed an existing automated grading system and showed potential for reducing the need for manual grading.
The deep-learning autograder (DLAG) achieved significantly higher specificity (P<0.001) for diabetic retinopathy than either consensus manual grading or iGradingM automated grader and had numerically better sensitivity. DLAG produced a 96.58% sensitivity for observable retinal disease and 98.48% sensitivity for identifying retinal disease requiring referral. The retrospective analysis involving data for nearly 180,000 patients in Scotland suggested that the DLAG system could reduce need for manual grading from 50% to 43.8%, as reported in the British Journal of Ophthalmology.
"If this system were used in NHS Scotland Diabetic Eye Service, it might create economic savings by final-grading more patient screening episodes than are final-graded by the current automated system," wrote Joseph Mellor, PhD, of the University of Edinburgh, and co-authors. "The system described in this study may be combined with automated assignment of screening intervals to further increase the efficiency of diabetic eye screening."
AI got a less favorable review for its ability to generate scientific abstracts in ophthalmology. Overall, the abstracts generated by two versions of the ChatGPT chatbot were of "average quality," but an "alarming" 30% of the references were either fake or unverifiable. Additionally, two programs for detecting chatbot-generated text had disparate accuracy scores.
The programs also were tripped up by clinical scenarios involving nuanced decision making, such as the difference in effectiveness between oral and IV corticosteroids for optic neuritis. In response to a question about anti-angiogenic agents for AMD, one chatbot-generated abstract proclaimed without qualification that one agent had demonstrated clear superiority over all others in the class.
"This report calls attention to the pitfalls of using AI chatbots for academic research," concluded Danny A. Mammo, MD, of the Cole Eye Institute at the Cleveland Clinic, and co-authors in JAMA Ophthalmology. "Chatbots may provide a good skeleton or framework for academic endeavors, but generated content must be vetted and verified."
The findings and observations came from a study involving chatbot responses to challenge questions covering seven ophthalmology subspecialties: comprehensive, retina, cornea, glaucoma, pediatrics, neuro-ophthalmology, and oculoplastics.
Early Success for Cancer Drug in AMD
A bioabsorbable ocular implant containing a reformulated cancer drug improved visual acuity and stabilized retinal thickening in a small randomized trial of patients with neovascular AMD.
Patients randomized to the axitinib (Inlyta)-containing implant or bimonthly aflibercept (Eylea) had a 1- to 2-letter improvement in visual acuity at 52 weeks. Patients allocated to the implant had a 20-µm increase in central subfield thickness (CSFT), a nonsignificant difference from the 2-µm decrease with aflibercept. The implant was associated with an 89% reduction in treatment burden over the course of 52 weeks.
"The study showed good safety, no signs of drug-related problems," said Robert Avery, MD, of California Retina Consultants in Santa Barbara, during the ASRS meeting. "The bioresorption of the implant is complete by 12 months, and due to the pharmacokinetics and pharmacodynamics, it seems like this would warrant a re-treatment in 9 to 12 months."
The findings came from a trial of 21 patients with previous exposure to anti-VEGF therapy for AMD. They were randomized 3:1 to the OTX-TKI implant or aflibercept. Patients allocated to the implant received a single intravitreal dose of aflibercept 4 weeks after implantation, followed by sham injections to simulate aflibercept treatment. The primary endpoints were safety, change in visual acuity and CSFT, and need for supplemental anti-VEGF injections.
A separate report on an axitinib suspension (CLS-AX) for intrachoroidal injection provided evidence of safety and a potential for reduced treatment burden. During 6 months of follow-up in 14 evaluable patients, no safety issues arose, and treatment burden decreased by about 80% as compared with the 6 months prior to enrollment in the trial, reported Rahul Khurana, MD, of Northern California Retina Vitreous Associations in Mountain View.
Other AMD-related news for 2023 included:
Demand for Anti-VEGF Injections for Retinal Diseases Expected to Surge
When It Comes to Eye Care, AI Couldn't See Straight
Sticker Shock: Ophthalmic Bevacizumab Biosimilar Could Drive Up Costs