Angiogenesis 2026: Highlights

This year’s Angiogenesis, Exudation, and Degeneration conference, held virtually on February 8, was packed with data and discussion. Led by Course Director Philip J. Rosenfeld, MD, experts from all over the world covered everything from imaging (including AI applications) and new therapeutic approaches in the clinic, to myriad therapies under investigation for AMD, diabetes, inflammatory disease, and rare and inherited retinal disease. Here, we highlight a few key talks from the day. 

NOVEL APPROACHES TO DRY AMD

Several experts discussed the role of ophthalmic artery stenosis in the development of AMD and a new treatment approach under investigation, ophthalmic artery angioplasty.

Omer Trivizki, MD, MBA, kicked off the discussion by outlining the age-related changes that happen in the eye, such as decreasing choroidal thickness and choriocapillaris perfusion, that play into the development of AMD. He referenced the INFLATE study, which used swept-source OCT angiography to show that patients with significant carotid artery stenosis treated with carotid endarterectomy experienced improved choroidal perfusion.¹ Next, he discussed the Tel-Aviv Correlation Study, which showed that eyes with dry AMD are more likely to have severe stenosis and established imaging protocols using magnetic resonance angiography. In conclusion, Dr. Trivizki emphasized that decreased choriocapillaris perfusion doesn’t cause AMD—it exacerbates the condition in genetically at-risk patients.

In the next talk, Mario J. Saravia, MD, PhD, shared an update on a study first presented at Angiogenesis 2024 in which patients with geographic atrophy (GA) were treated with a novel ophthalmic artery angioplasty procedure.² At 6 months, patients (n = 11) experienced improved vision of about 7 lines, as well as increased choroidal thickness. He explained that cardiovascular outcomes tend to deteriorate after approximately 2 years after angioplasty, and his team was interested to see the long-term effects of treatment on smaller vessels, such as the ophthalmic artery. The 24-month data suggest patients experienced some visual decline, according to Dr. Saravia, but still have improved vision above baseline (by about 5 lines). The same was true for choroidal thickness, he added.

Other novel therapeutic approaches for GA discussed during the meeting include fas inhibition with xelafaslatide (ONL Therapeutics) by David N. Zacks, MD; VOY-101 (Perceive Biotherapeutics) by Anne Fung, MD; kamuvudine K8 (Inflammasome Therapeutics) by Jayakrishna Ambati, MD; once-daily zervimesine (Cognition Therapeutics) by Roger A. Goldberg, MD; and the PRIMA System (Science Corporation) by Frank G. Holz, MD.

WET AMD THERAPIES IN THE PIPELINE

During the wet AMD therapy session, Mark R. Barakat, MD, shared long-term outcomes from the OPTIC trial of ixoberogene soroparvovec (ixo-vec, Adverum). Treatment with this intravitreal gene therapy led to sustained aflibercept protein expression and a cumulative 86% reduction in mean annualized anti-VEGF injections through 5 years. The phase 2 LUNA trial, presented by Charles C. Wykoff, MD, MBA, was designed to determine the optimal dose of ixoberogene soroparvovec for the phase 3 trials, with safety and efficacy data supporting the 6E10 vg/eye dose.

Dr. Wykoff also presented first-time results of the phase 1b JADE trial investigating faricimab (Vabysmo, Genentech/Roche) versus OLN324 (Ollin Biosciences), a bispecific VEGF and angiopoietin-2 inhibitor for the treatment of wet AMD and diabetic macular edema (DME). At 12 weeks, approximately 50% more patients achieved an absence of edema in the DME arm treated with OLN324 compared with the faricimab treated patients, with equivalent drying between all groups in the wet AMD cohorts. The company plans to initiate phase 3 trials this year, according to Dr. Wykoff.

Tong Li, MD, PhD, presented the 1-year results from the phase 2 VENUS trial of the subretinal LX102 (Innostellar Biotherapeutics) gene therapy to treat wet AMD. At 1.5 years, 100% of treated patients in both the medium- and high-dose cohorts (n = 38) remained injection free with stable BCVA and central subfield thickness (CST). The phase 3 trial began earlier this year, Dr. Li said.

Arshad M. Khanani, MD, MA, FASRS, presented encouraging long-term safety and efficacy data from the phase 1/2 PRISM study evaluating 4D-150 (4DMT), an investigational gene therapy for the treatment of wet AMD. The results showed that visual acuity and CST were maintained with 4D-150 compared with aflibercept (Eylea, Regeneron) up to 2 years. In addition, consistent reductions in treatment burden were observed across multiple groups: an 85% reduction across a broad range of patients; 79% in those with severe, recalcitrant disease; and 92% in recently diagnosed patients. Regarding safety, the findings revealed no serious ocular adverse events associated with 4D-150, including no cases of hypotony, endophthalmitis, occlusive/nonocclusive retinal vasculitis, or choroidal effusions.

Later in the session, Allen C. Ho, MD, shared 1-year results from the phase 2 study of surabgene lomparvovec (sura-vec/ABBV-RGX-314, Regenxbio/Abbvie), an investigational gene therapy for wet AMD. The study evaluated a high (1.3 x 1011 GC/eye, n = 30) and low (6.4 x 1010 GC/eye, n = 30) dose of sura-vec delivered via subretinal injection. Each cohort demonstrated stable to improved visual acuity (high dose, +2.7 ETDRS letters; low dose, +1.1 letters) and mean CST (high dose, +4.7 µm; low dose, -16.0 µm). In addition, patients experienced a significant mean reduction in anti-VEGF injection burden at 1 year (high dose, 76%; low dose, 77%). No serious adverse events occurred that were associated with the study drug.

For the first time with a gene therapy, the study also evaluated fellow eye treatment (n = 11) and found that bilateral administration of sura-vec led to good anatomic and visual acuity outcomes, as well as a 92% reduction in anti-VEGF injections at 1 year in fellow eyes.

AI-POWERED IMAGING ANALYSIS

A proof-of-concept study presented by David Eichenbaum, MD, used an AI-driven workflow to identify qualified candidates with GA for clinical trial enrollment. The research team conducted a retrospective analysis to compare an AI-generated list of potential candidates for enrollment with a historical screening reference from the GATHER1/2 trials, in which 26 candidates were screened, six of whom were enrolled (23.1%). Of the 20 remaining screen failures, 90% were a result of not meeting imaging criteria. The team used an AI model to first filter the GATHER1/2 participant pool (n = 10,421) based on imaging analysis, which led to a group of 270 who met GA lesion criteria. This group was then further filtered via electronic health record (EHR) reconciliation, leading to 24 qualified candidates left after completing both imaging analysis and EHR review. Assuming 100% of these candidates would ultimately be enrolled, this AI-driven outcome may be approximately 4x more efficient than the historical reference. The next step for evaluating the benefits of this AI-drive workflow, shared Dr. Eichenbaum, would be to integrate it into a current, prospective trial for GA.

Hasenin Al-khersan, MD, shared findings from an ongoing pilot analysis of patients with GA from Retina Consultants of Texas using AI-powered analysis of real-world data. Dr. Al-khersan describes the challenges his team faced, including limited data availability and visual acuity variability compared with clinical trials, as well as issues with imaging quality and OCT scan density variability. One potential solution to these barriers to high-quality real-world data analysis may involve increasing data scale; to this end, he discussed RCA Insights, a data platform dedicating to centralizing retinal imaging data nationally to be used for such projects. He also pointed out that the use of autorefractors, standardized OCT acquisition protocols, and real-time flags for poor OCT image quality may lead to improved data quality, although reimaging may affect workflow efficiency.

Ursula Schmidt-Erfurth, MD, presented a process to enhance the precision of AI-powered OCT segmentation in GA analysis. One issue with AI algorithms is that clinical trial data does not encompass the full range of presentations of GA. To counter this issue, Dr. Schmidt-Erfurth described the enrichment of data by analyzing more than 200 cases of GA, 50% of which were real-world cases typically labeled “challenging” due to small or large lesion size, presence of fluid, or other unique features. Next, she discussed the need for refined annotation protocols for ellipsoid zone layers and loss of retinal pigment epithelium in intermediate AMD and GA and the importance of extended image augmentations (ie, contrast changes, noise/noise patches, rotation/shifting/bending, vessel shadows, flipping) to prevent overfitting and expand the range of scans that can be processed using AI. When her team evaluated the AI model, they found a high correlation between human and algorithm measurements and observed that the model could catch subtle ellipsoid zone thinning, an important biomarker for predicting functional vision outcomes.

ROBUST EDUCATION, EVERY YEAR

The meeting was a triumph of education, with more than 12 hours of presentations from experts around the globe. Attendees walked away with an appreciation for the robust research that continues to push the field of retina forward. We look forward to next year’s meeting!

1. Zhang Y, Zhou SW, Noam N, et al. Influence of carotid endarterectomy on choroidal perfusion: the INFLATE study. Ophthalmol Retina. 2024;8(1):62-71.

2. Lylyk P, Lylyk I, Lylyk PN, et al. Ophthalmic artery angioplasty in a cohort of patients with geographic atrophy secondary to non-exudative age-related macular degeneration. J Neurointerv Surg. 2026;18(5):1364-1371.