ARVO Retina Research Highlights Progress in Geographic Atrophy, DME Therapies

Credit: Scott Camazine/Medical Images.
Research presented at the ARVO 2023 meeting shows a growing armamentarium of retinal treatments and monitoring technologies.

Ophthalmologists know the Association for Research in Vision and Ophthalmology (ARVO) meeting is a key conference, where a truly astounding volume of research is presented. It would be impossible to consume all this information during the conference’s few short days, and just as unlikely that anyone in eye care could go a year without having their practice affected by something presented at ARVO. 

Although the meeting is comprehensive in nature, featuring findings from across all subspecialities, this year’s ARVO retina research offerings represented significant advancements in care. Early in 2023, pegcetacoplan became the first approved treatment for geographic atrophy (GA) —  an advanced form of dry age-related macular degeneration (AMD) that commonly leads to​​ significant bilateral central vision loss. The approval of the intravitreal injection came on the heels of a plethora of research — most significantly 2 large phase 3 trials.  

This first approval of a GA treatment may feel reminiscent of early research into anti-vascular endothelial growth factor (VEGF) injections for the treatment of neovascular diseases of the retina. If that’s the case, researchers may be finding new applications for it even 20 years from now. That’s what’s happening today with anti-VEGF drugs, in fact. Presentations at ARVO are showing that aflibercept 8 mg injections delivered at 12- and 16-week intervals (after 3 monthly doses) are as effective for most patients with diabetic macular edema (DME) as aflibercept 2 mg at the traditional 8-week interval (after 5 monthly doses).   

And advances in eye care are not limited to pharmaceuticals. Researchers are constantly uncovering new potential applications for the rapidly developing field of artificial intelligence in ophthalmic research and practice. Investigators are finding that, with the use of advanced imaging techniques, such as optical coherence tomography angiography (OCT-A), these digitized assistants can predict most cases of AMD, high myopia, and glaucoma, in some cases, before any clinically detectable changes occur. 

When you look at the primary analysis, you will see that in the last 6 of the 24 months, there was an increasing treatment effect.

Geographic Atrophy Treatment Options Available 

When pegcetacoplan was approved for the treatment of geographic atrophy (GA) earlier this year, it fulfilled a long-promised prophecy in eye care — that a pharmaceutical option could preserve vision for patients with this condition. Prior to its launch, no treatment options for geographic atrophy existed. Although several experimental drugs have been through clinical trials, none have previously made it to market. 

A number of ARVO retina research presentations evaluated the application of pegcetacoplan and other treatment options, as well as the use of imaging technologies for grading and evaluation of geographic atrophy.

One of the prime presentations in this arena reviewed the findings of the OAKS and DERBY phase 3 trials ( Identifier: NCT03525600).

The investigation demonstrated the efficacy of pegcetacoplan across patient subgroups and with both monthly and every-other-month (EOM) dosing schedules. 

After 24 months, GA lesions in sham arms of the study grew at rates consistent with natural history studies. However, those treated with pegcetacoplan (monthly or EOM) had slower rates of GA progression, and those effects increased over time.1

“I think there are some keys we can take from the data. When you look at the primary analysis, you will see that in the last 6 of the 24 months, there was an increasing treatment effect. Overall, patients in the OAKS and DERBY had approximately 20% reduction in GA lesion growth if they were treated monthly, and about 17% with treatment every other month,” explained Allen Chiang, MD, of the Wills Eye Hospital in Philadelphia, who led the presentation. “But if you look at that last 6 months, there’s starting to be a greater divergence compared to sham where it’s 30% for monthly and 24% for every other month. This underscores the importance of staying on treatment. Some patients may have transportation or other issues, but if they can somehow manage even just every other month it can be significant. It can help slow their disease progression and potentially deliver some functional benefits that we’re finding in post-hoc analyses, and hopefully that we’ll see more of in GALE.”

GALE ( Identifier: NCT04770545) is a phase 3, multicenter, open label, extension study evaluating long-term safety and efficacy of pegcetacoplan. It will evaluate long-term incidence and severity of ocular and systemic treatment emergent adverse events as well as change in the total area of GA lesions.

In other GA research, a phase 2 study speculated about the use of minocycline as a treatment. Researchers with the National Eye Institute and other facilities hypothesized that since geographic atrophy development and progression may be triggered by phagocytic innate immune cells, particularly activated outer retinal microglia, the anti-inflammatory agent minocycline could potentially slow the disease’s progression. 

Although cell culture and animal models supported minocycline’s neuroprotective role in retinal degeneration, the phase 2 study did not deliver such promising results.

That research enrolled 37 participants (mean age, 74.3 year; 57% women). The primary and select secondary outcome evaluations showed no significant difference in measures between the phases (Table). For the primary outcome, the mean difference between enlargement rates during the treatment versus run-in phases was -0.03 mm/year (standard error 0.03; P=0.45). Of the 123 treatment-emergent adverse events reported to date, 46 (37%) were considered related to minocycline (no severe or ocular events), including raised thyroid stimulating hormone (11 participants), skin hyperpigmentation (6 participants), and gastrointestinal symptoms (7 participants).

“No anatomic or visual acuity effects were observed in the trial. Oral minocycline may not be beneficial in slowing GA progression. This might relate to suboptimal inhibition of microglia or absence of their causal involvement in GA progression,” the presenters reported.

Anti-VEGF Advancements

Geographic atrophy therapy might be grabbing the most headlines in retinal care right now, but it is hardly the only ARVO retina research topic featured at this year’s meeting. 

A study out of the UK presented at the conference shows that patients with aggressive neovascular AMD responded well when switched to faricimab. That research evaluated the effect on both function and structure of 267 eyes (mean age, ​​74.1±7.6 years) that were switched to faricimab from either aflibercept (89.9%) ranibizumab (10.1%).3

At 4.27±0.62 weeks after undergoing 2 faricimab injections, patients’ mean central subfield thicknesses reduction was 20.19±7.01 μm (95%CI=[3.091, 37.296]; P =.015), and 60% of maculae were dry, a statistically significant increase, the investigators reported. Mean best-corrected visual acuity was maintained at +0.52±0.69 early treatment diabetic retinopathy study (ETDRS) letters, however this difference did not reach statistical significance (P =.19), suggesting maintained visual acuities and improved anatomical outcomes after switching.3

Faricimab is likely to relieve some of the treatment burden for patients with AMD as dosing can be spaced further apart than current anti-VEGF options. In a similar vein, research presented at ARVO 2023 shows that a current anti-VEGF therapeutic, at a higher dose, can safely manage diabetic macular edema (DME) at 12- and 16-week intervals.4,5

Researcher Diane V. Do, MD, a professor of ophthalmology at Byers Eye Institute, Stanford University, presented findings derived from the PHOTON trial ( Identifier: NCT04429503) showing that aflibercept 8 mg is noninferior to aflibercept 2 mg for patients with DME in terms of best-corrected visual acuity outcomes and has no new safety signals through 48 weeks. Most patients given the higher dose were able to have their treatment schedules extended to 12-week dosing or more.

“Overall, aflibercept 8 mg provides greater therapeutic benefit, an expanded injection interval, and equivalent safety vs aflibercept 2 mg,” according to the presentation.4

Further data extrapolated from the trial specifies that patients with DME who could not maintain a randomized dosing interval had more severe disease at baseline.5

Benefits of aflibercept 8 mg over 2 mg is not limited to DME, patients with AMD are also demonstrating improved visual and anatomical outcomes, according to researcher Jordana Goren Fein, MD, who reviewed findings from the CANDELA trial ( Identifier: NCT04126317).

That trial evaluated treatment-naïve patients with neovascular AMD and compared a group given 3 monthly doses of aflibercept 2 mg (n=53) with those given aflibercept 8 mg (n=53), followed by additional doses at 20 and 32 weeks. By week 16, the group that received the 8 mg dose compared with the 2 mg patients had more patients with no intraretinal fluid, subretinal fluid, or sub-retinal pigment epithelium fluid in the central subfield (38% vs 19%; nominal, P=.031). By week 44, those numbers were  25% vs 11% (nominal P =.076).

The 8 mg group also had a greater proportion of eyes (27%) with a more than 239 µm reduction in central subfield thickness than the 2 mg group (20%). In terms of visual acuity, the 8 mg group had a greater proportion of eyes that gained 15-letters or more letters by week 44 (33%) than the 2 mg group (14%), as well as more who achieved BCVA of 20/40 or better (61% vs 49%). More patients in the 8 mg group compared with the 2mg group also achieved BCVA of 20/20 or better (10% vs 2%). 

The study also notes that a smaller proportion of eyes in the 8 mg (20%) had vision loss or no BCVA change at week 44 (27%). For patients who had baseline BCVA worse than 20/40, a greater proportion in the 8 mg group gained 10 letters or more compared with the 2 mg group (65% vs 47%), and a similar difference was seen among those who gained 15 letters or more (43% vs 18%).6

Artificial Intelligence and the Retina

Any conference attendee could see from this year’s ARVO retina research that treatments are expanding, opening up new avenues of disease management. But a key element of any modern approach is going to require robust monitoring and early diagnosis. Artificial intelligence is primed to assist clinicians in this role.

One study was able to show that, when it comes to grading fluid volume in neovascular AMD, DME, or branch retinal vein occlusion (BRVO), an automated approach is comparable to human experts. The researchers used a validated automated deep learning algorithm to evaluate 6898 OCT scans for central, peri- and parafoveal intraretinal and subretinal fluid in patients with AMD, DME, or BRVO. It found that “In addition to information about presence of fluid, automated quantification of retinal fluid provides volumetric information, and its fluid detection is comparable to human detection. Thus, automated fluid quantification is a feasible tool for objective treatment decision support and disease monitoring in clinical practice.”7

Another presentation shows the possibility that artificial intelligence can uncover hidden biomarkers that can predict retinal disease long before any clinically noticeable changes occur. The study used deep learning to automatically propose clusters of OCT images that share similar features. “Without any prior knowledge of AMD, our method rediscovered the set of biomarkers used in clinical grading systems and others that have previously been linked to early AMD,” the researchers explained. 

The automated system identified several clusters of like images and was able to associate them with disease. The findings it noted as biomarkers included large drusen, Complete retinal pigment epithelium (RPE) and outer retinal atrophy (cRORA), serrated drusen, double-layer sign, and basal laminar deposits. These are all established biomarkers, but the fact that artificial intelligence was able to develop it implies that it could identify new biomarkers as well, the researchers suggest. In fact, at least one cluster of images could qualify as “a potentially novel biomarker depicting cases of hypertransmission where the choroid remains thick and visible,” according to the study.8

These findings were based on images gathered for the ongoing PINNACLE trial ( Identifier: NCT04269304).

Those are just a couple of the dozens of studies presented on the potential applications of artificial intelligence in retina research and clinical practice. 

The retina research presented every year at ARVO helps to set the tone for investigators and practitioners, not just for the year ahead, but likely for many years to come.


  1. Chiang A, Bliss C, Ribeiro R. Assessment of geographic atrophy (GA) lesion progression in the phase 3 OAKS and DERBY trials. Presented at: Association for Research in Vision (ARVO) 2023; April 23-27. Abstract 986
  2. Keenan TDL, Bailey CC, Bellur S, et al. A phase ii trial evaluating oral minocycline in the treatment of geographic atrophy in age-related macular degeneration. Presented at: Association for Research in Vision (ARVO) 2023; April 23-27. Abstract 5058
  3. Sim SY, Chalkiadaki E, Koutsocheras G, et al. Anatomical and visual outcomes of patients with active neovascular age related macular degeneration(nAMD) requiring monthly anti-VEGF treatment following switch to faricimab. Presented at: Association for Research in Vision (ARVO) 2023; April 23-27. Abstract 2206 – C0159
  4. Brown D. Baseline disease characteristics of patients who maintained 12- and 16-week aflibercept 8 mg dosing versus patients with shortened treatment intervals through week 48 in the phase 2/3 PHOTON trial. Presented at: Association for Research in Vision (ARVO) 2023; April 23-27. Abstract 2813.
  5. Do D. Aflibercept 8 mg for diabetic macular edema: 48-week results from the phase 2/3 PHOTON trial. Presented at: Association for Research in Vision (ARVO) 2023; April 23-27. Abstract 2814.
  6. Fein JG. Additional visual and anatomic outcomes of intravitreal aflibercept injection 8 mg versus 2 mg: a post hoc analysis of the Phase 2 CANDELA study. Presented at: Association for Research in Vision (ARVO) 2023; April 23-27. Abstract 2180 – C0133.
  7. Goldbach F, Gerendas BS, Leingang O, et al. Human expert grading versus automated quantification of fluid volumes in nAMD, DME and BRVO. Presented at: Association for Research in Vision (ARVO) 2023; April 23-27. Abstract 1286.
  8. Holland R, Leingang O, Hagag AM, et al. Deep-learning-based clustering of OCT images for automated biomarker discovery in age-related macular degeneration. Presented at: Association for Research in Vision (ARVO) 2023; April 23-27. Abstract 1283.