Innovative AI technologies expected to enhance AMD space
Key takeaways:
- There is much potential for artificial intelligence in the AMD space.
- AI will help tailor treatments to the right patients.
The potential of artificial intelligence for the diagnosis and management of age-related macular degeneration is vast.
Specifically, experts in the field are learning more about staging of the disease with the use of OCT, which allows a 3D higher-resolution rendition of disease and characterization of stage before progression to geographic atrophy.
“AI can be used to interpret images and make predictions, including conversion of more advanced stages of the disease, such as wet AMD or geographic atrophy,” Eleonora M. Lad, MD, PhD, associate professor of ophthalmology and vitreoretinal diseases and director of the ophthalmology clinical research unit at Duke University Medical Center, told Healio. “AI will improve the standard of care in AMD in a very positive way, and it will not take retina specialists out of a job. There will be plenty of avenues that only we can handle as specialists, but AI can do a lot of the heavy lifting for us and help tailor our skill set to the right patients in our very busy clinics.”
AI at home
One way AI is being studied is to monitor patients with AMD at home.
“In the very near future, we are looking forward to having Home OCT (Notal Vision), which will be used in patients’ homes to monitor their fluid and prevent occurrence of new fluid that requires treatment,” Lad said. “We do not want to tolerate too much subretinal fluid over time due to concern for hemorrhage that will decrease vision and lead to fibrosis if not adequately treated. We don’t want to undertreat either, but at the same time, we want to be able to use an AI algorithm to provide fluid quantification and provide ways to maximize the benefit-to-risk ratio for each patient. We want to use these algorithms built into the Home OCT device to extend the patient for as long as possible to minimize the injection burden yet maximize their visual outcomes based on what we know right now about the types of fluid, how they respond to treatment and how much fluid each patient’s eye can tolerate.”
This is especially important following the recent FDA approval of Syfovre (pegcetacoplan injection, Apellis Pharmaceuticals), the first therapeutic option for geographic atrophy, Lad said.
“There is a dose-dependent increase in exudative events due to Syfovre, so it will be very useful to be able to monitor these patients with Home OCT to pick up any conversion to wet AMD as early as possible for optimal vision outcomes,” Lad said. “The sooner we pick up a conversion to wet AMD, the better outcomes have been demonstrated. If a patient converts to neovascular AMD, the goal is early diagnosis and treatment in the office with anti-VEGF injections.”
The ForeseeHome device (Notal Vision) is another option that has shown promise in the AMD space.
“A study that compared the ForeseeHome device vs. the Amsler grid was stopped early due to superiority of ForeseeHome to identify cases of early conversion to wet AMD,” Lad said. “The ForeseeHome device, which is covered by insurance, is used by patients multiple times per week to monitor and ensure there is no conversion from intermediate or high-risk AMD to wet AMD. If the device generates an alert, the patient is notified right away in addition to the physician’s office so that an appointment can be made in a timely manner.”
AI could be the solution to monitoring these patients and diseases, Lad said.
“We don’t want to conduct a 5 to 10 years study of intermediate AMD that would allow this stage to reach the outcome of conversion to geographic atrophy, for example,” she said. “If we gain a standard of care approved medicine, such as the recent Syfovre approval, AI can help us identify the intermediate AMD or pre-geographic atrophy patients at high risk for progression to geographic atrophy so we can follow them closely and offer therapy sooner. We may also be able to use AI to identify patients at high risk for conversion to exudative AMD, and we can monitor and follow them closely and ensure the best visual outcomes with early treatment.”
Looking ahead
In the future, Lad said that AI will assist physicians in selecting the best patient cases for specific pathology, selecting subjects for clinical trials and identifying patients who may respond best to anti-VEGF treatments.
“There’s quite a bit of heterogeneity in AMD, so we can learn more about outcomes in terms of fluid type, visual acuity with anti-VEGF treatments, as well as different types of pathologies for AMD,” she said. “It is also important for physicians to be cautious about fibrosis. Fibrosis is still a clinical problem for which we currently do not have solutions for in patients with AMD, so if we could figure out a treatment paradigm that is more individualized for each patient to avoid fibrosis, this would be extremely helpful from a clinical standpoint.”
Another area in which there may be a synergy of intelligence between AI and clinicians is in the general increase of insights into clinical and imaging data for greater efficiency of monitoring of treatment regimens, which could lead to better patient outcomes and lower costs of care, Lad said.
“If we can minimize treatment and injection burden, as well as treatment costs, then that would be a game changer for the health care system and would help patients as well,” she said. “Moreover, being able to scale down our efforts in clinic by leveraging AI initiatives so that we can preselect patients that get referred to our clinics and minimize the number of people we screen would be beneficial. The standard of care in AMD will be changing within the next 5 years to treatment of earlier-stage disease. We will need help from AI algorithms trained using large datasets to help us understand what type of patients to treat, how frequently to treat and the population that is at highest risk for progression.”
This type of automation could decrease fatigue and help populate clinics with the right patients, Lad said.
“Overall, the more we know, the more prognostication we can provide for our patients, which will help us improve the doctor-patient relationship,” she said. “We can then make better-educated guesses about what the disease stage means for each patient within the next 5, 10, 20 years of their lives to help them plan. Unfortunately, patients with AMD often ask when their visual function will decrease to a level that will impact their activities of daily living. Currently, I can use imaging to make an educated guess on what I think is going to happen in the next 5 to 10 years. Instead, it would be ideal if I had the evidence based on large data sets and robust AI algorithms.”
References:
- FDA approves Syfovre (pegcetacoplan injection) as the first and only treatment for geographic atrophy (GA), a leading cause of blindness. https://investors.apellis.com/news-releases/news-release-details/fda-approves-syfovretm-pegcetacoplan-injection-first-and-only. Published Feb. 17, 2023. Accessed April 28, 2023.
For more information:
Eleonora M. Lad, MD, PhD, of Duke Health, can be reached at nora.lad@duke.edu.
Collapse