Geographic atrophy (GA), the advanced form of non-neovascular age-related macular degeneration (AMD), is defined by progressive, irreversible loss of the retinal pigment epithelium, photoreceptors, and choriocapillaris.1
Until 2023, I had nothing to offer patients with this diagnosis except observation. Today, with two FDA-approved complement inhibitors and several years of real-world data behind them, "wait and see" is no longer a neutral holding pattern. It's a choice, and it has a documented cost.
Why surveillance alone is no longer enough
Before 2023, we honestly had no other option but to observe. That became habitual: we were often seeing patients develop GA, unfortunately lose vision, and we didn't have much of an option at that point in time. We could tell a patient the disease would progress, but we couldn't give a definitive timeline, and we had no way to slow it. That rationale made sense when there was nothing to slow it with.
It doesn't anymore, however, it's worth identifying why observation was ever the standard: GA is, by definition, tissue that does not regenerate.1 Once the RPE, photoreceptors, and choriocapillaris are gone in a given area, that retina is gone. Every month spent only watching is a month of lesion growth that cannot be recovered later, treatment or no treatment.
Many retina specialists who have practiced for years learned GA as a slowly growing condition best left alone until it threatened the fovea. That mental model was built in an era with no treatment to offer. It's a habit now, not a clinical necessity, and it's the single biggest barrier I see colleagues struggle to unlearn.
Real-world evidence that delay carries a documented cost
The IRIS Registry data are unambiguous on this point. In a retrospective analysis of more than 69,000 patients with at least 2 years of follow-up, GA eyes with subfoveal involvement had measurably worse visual acuity than nonsubfoveal eyes at every time point, and progression from nonsubfoveal to subfoveal disease was common.2
A separate analysis of nearly 19,000 GA eyes followed for 3 years found an average loss of two to three lines of visual acuity over that window, with the oldest patients and those with moderate baseline impairment losing the most.3 This is what "wait and see" produces in practice, not in theory.
On average, untreated lesions grow at roughly 1.5 to 2mm2 per year, though that rate varies by individual risk factors—bilaterality, multifocality, and the diffuse trickling pattern of hyperautofluorescence we look for on fundus autofluorescence (FAF).
I've had patients progress meaningfully in 3 to 4 months. I had one just yesterday: stable on anti-VEGF therapy for wet AMD in the same eye, no fluid, everything looked great on the scans, and the patient still told me they were losing vision.
We put the FAF images side by side and the GA had clearly moved closer to the fovea in that short window. The wet disease wasn't the problem. The atrophy was, and it hadn't been standing still just because it looked quiet.
That's the conversation I have with patients who feel fine and test 20/20. Visual acuity is a poor early warning system in GA. Reading speed and low-luminance vision decline before Snellen acuity does, and when I ask directly, most patients admit they've noticed something: needing more light to read a menu, reaching for a phone flashlight at a dim restaurant. The lesion is already costing them function they haven't yet named as vision loss.
Clinical trial evidence supporting earlier intervention
Two complement inhibitors now have multi-year, sham-controlled data showing that the benefit of treatment builds the longer patients stay on it. In GATHER2, monthly avacincaptad pegol slowed GA lesion growth relative to sham at 12 months, and that treatment effect more than doubled by 24 months, with the benefit continuing to widen as dosing continued.4,5
Pegcetacoplan shows the same pattern in OAKS and DERBY: a 21% reduction in lesion growth at 12 months in OAKS, growing to 22% at 24 months, with DERBY showing a comparable widening over the same period.6 Two independent complement pathways, two independent trial programs, and the same directional finding: the gap between treated and untreated lesions gets wider with time, not narrower.
That's the argument for continuous treatment, not a course of it. We know that the more these patients receive treatment, the more benefit they get. That compounding effect, the widening of the curve, is the reason I've become more comfortable starting therapy in patients who are asymptomatic and have smaller lesions, rather than waiting for growth toward the fovea to force the conversation.
Regulatory and real-world signals reinforcing ongoing management
The regulatory history here matters clinically, not just administratively. Avacincaptad pegol was originally approved with dosing limited to 12 months. In February 2025, the FDA expanded that label, removing the duration limit entirely based on the GATHER2 2-year dataset.7 That update matters because GA is a chronic, progressive disease, not a 2-year problem.
Before that label update, patients would sometimes ask if this was a one-and-done treatment, or something you did for a year or two and stopped. If you don't get the treatment, you lose the benefit. There was never a natural endpoint to this disease, and now there's no artificial one built into the label, either.
The safety data have moved in the same direction as the efficacy data. A pharmacovigilance analysis of FDA Adverse Event Reporting System data found no retinal vasculitis reported for avacincaptad pegol, distinguishing its post-marketing safety profile from the closer monitoring that has been required with the earlier-approved complement C3 inhibitor.8
An open-label extension of GATHER2 followed patients out to 3.5 years and reported no cases of retinal vasculitis or occlusive vasculitis, no new safety signals, and consistently greater protection of retinal tissue in patients who started treatment earlier.9 The vasculitis risk that made a lot of retina specialists cautious in 2023 has not materialized in the same way with this agent, and that's changed how I counsel patients about the trade-off between injections and inaction.
Reframing the patient conversation around "wait and see"
The way I explain this to patients now is with an analogy: we're on a train, and the bridge ahead has collapsed, so the train will crash. We're not preventing that. We're going 100 miles an hour, and we're trying to slow it down to 70. How quickly we get to that end point is the entire question, and treatment is how we control our speed.
I show patients their FAF images, side by side over time when I have them, because a picture of the atrophy is worth more than a description of it. I use the analogy of a carpet with patches worn into it: the retina is the carpet, and the atrophic patches are areas we've already lost. What we're protecting is what's left. Framed that way, "let's keep watching" stops sounding like patience and starts sounding like what it is: an active decision to let the patches spread further before we act.
I don't wait for a lesion to threaten the fovea before I bring up treatment. I mention it as soon as I see GA, and I lean toward starting sooner if I can document growth over 3 to 6 months, because lesions closer to the fovea are harder for patients to accept losing further ground on, and because we get better long-term preservation of tissue and function the earlier we start.
Key takeaways
- Surveillance-only management of GA is a clinical decision with a measurable opportunity cost, not a neutral default, because GA reflects RPE, photoreceptor, and choriocapillaris loss that does not reverse.1
- Real-world registry data confirm that GA causes meaningful, progressive visual acuity loss over 2- to 3-year windows, independent of clinical trial settings.2,3
- Both FDA-approved complement inhibitors show a widening treatment benefit over time in sham-controlled trials, arguing for earlier and sustained initiation rather than a wait-and-treat-later approach.4,5,6
- The 2025 label expansion removing avacincaptad pegol's dosing-duration limit reflects the reality that GA management is chronic and ongoing, with no natural stopping point.7
- Post-approval safety data, including the absence of reported retinal vasculitis with avacincaptad pegol through 3.5 years of follow-up, support greater clinical confidence in starting treatment before a patient becomes symptomatic.8,9
- Patients often have functional symptoms, such as reduced reading speed or low-luminance vision, before they lose visual acuity, so waiting for acuity to decline delays treatment past the point where the most tissue could have been preserved.
- Imaging tools, including serial FAF and emerging AI-based lesion-growth prediction, are shifting GA management toward individualized, risk-stratified decisions rather than a uniform wait-for-fovea approach.
Closing thoughts
The patients I worry about most aren't the ones who accept treatment reluctantly. They're the ones who leave my office having agreed, implicitly, to wait. I understand the impulse: nobody wants an eye injection for a disease that, at that visit, isn't affecting how they see.
But GA doesn't pause while we wait for a clearer signal to act, and the tissue it takes in that interval doesn't come back once we finally do. The earlier retina specialists retire the old model, the more vision we'll have preserved for the patients sitting in front of us right now.
