This was a retrospective analysis of glaucoma suspects from the longitudinal cohort in the Diagnostic Innovations in Glaucoma Study (DIGS). ^, All participants underwent optical OCTA imaging (AngioVue; Optovue Inc). Participants from DIGS were eligible if they met the inclusion and exclusion criteria detailed below. Written informed consent was obtained from all participants. The University of California, San Diego Human Subjects Committee approved all study procedures, and the research adhered to the tenets of the Declaration of Helsinki. The time period for data gathering was from May 1995 to August 2023 for fundus photographs, and from March 2015 to October 2022 for OCTA. Of the 1713 glaucoma suspects in the DIGS dataset, 556 patients completed >5 years of follow-ups for fundus photographs and had current OCTA at their last visit. Inclusion criteria for DIGS were: (1) age >18 years, (2) open angles on gonioscopy, (3) best-corrected visual acuity of 20/40 or better, (4) refraction within ±5.0 diopters spherical and within ±3.0-diopters cylinder at study entry. Exclusion criteria were: (1) history of ocular trauma or intraocular surgery (except uncomplicated cataract extraction or glaucoma surgery), (2) coexisting significant retinal disease, (3) uveitis, or (4) nonglaucomatous optic neuropathy. Participants with systemic conditions, including Parkinson’s disease, Alzheimer’s disease, dementia, or prior stroke, were also excluded.

Glaucoma suspect eyes were included with either elevated IOP (≥22 mm Hg) or optic discs with glaucomatous optic neuropathy in the absence of repeatable abnormal visual field (VF) loss. Glaucoma was defined as having at least two consecutive abnormal VF test results, and glaucomatous optic neuropathy was defined as optic disc excavation, focal rim thinning or notching, or localized/diffuse retinal nerve fiber layer (RNFL) atrophy. Abnormal VF tests were defined as a pattern SD outside the 95% CI and/or a Glaucoma Hemifield Test outside normal limits. Eyes were included in this study if optic disc photographs were available at least 5 years before OCTA imaging. All eyes included in this study did not have consecutive abnormal VFs during the follow-ups. During follow-up, structural progression, disc hemorrhage, and β-zone PPA progression were recorded.

Structural progression of the optic disc was evaluated using stereophotographs according to standardized grading procedures by two independent observers. Each photograph pair was first assessed for overall quality, and images were deemed not gradable if inadequate clarity, stereo effect, or RNFL visibility prevented reliable determination of progression. For gradable pairs, the presence of disc hemorrhages was documented if the hemorrhage was located within the optic cup or had a proximal edge situated no more than one-half disc diameter from the disc margin. Although disc hemorrhages were recorded and considered risk factors, their presence alone was not sufficient to classify an eye as demonstrating progression. When progression was observed, it was attributed to thinning of the neuroretinal rim or the appearance of new or enlarged RNFL defects. Rim thinning was defined as narrowing or increased loss of the neuroretinal rim, while RNFL progression referred to the development of new or enlarged defects. These categories were not considered mutually exclusive. Each eye was assigned a grade of either progression or no progression by comparing baseline and follow-up image pairs. See Figure 3 for examples of optic disc stereophotograph assessment.

Optic disc photographs were evaluated masked to MvD status and clinical information. To evaluate β-zone PPA, fundus photographs and superficial slab OCTA images were rotated, resized, and aligned using PowerPoint, then the β-zone PPA area was delineated with the freeform scribble tool. The images were then imported into ImageJ software (vs 1.54), and the topographic parameters of the β-zone PPA ( Figure 1 ) were measured according to previous reports: (1) area, (2) maximal radial extent (width), and (3) angular extent around the disc (circumference). Pixel area and distance were corrected using Littmann’s formula. Progression of β-zone PPA was defined as a 20% or greater increase in area, maximum radial extent, or angular extent around the disc.

The parameters of β-zone peripapillary atrophy: (1) area, (2) maximal radial extent (width), and (3) angular extent around the disc (circumference).

A 4.5 × 4.5 mm² scan centered on the ONH was acquired (304 B-scans × 304 A-scans per B-scan; software version 2018.1.0.43) using AngioVue. For the purpose of this study, an en-face choroidal vessel density map generated from the entire 4.5 × 4.5 mm² scan was used to evaluate MvD. This map encompasses the layers beneath the retinal pigment epithelium, including the choroid and sclera. Image quality was reviewed according to the UC San Diego Imaging Data Evaluation and Analysis Regarding Center protocols. Scans were excluded if they demonstrated: (1) a scan quality index <4, (2) poor clarity, (3) motion artifacts visible as irregular vascular patterns or distorted disc margins, (4) image cropping or local weak signal due to media opacity, or (5) uncorrectable segmentation errors. MvD was defined as a dropout visible in ≥4 consecutive horizontal B-scans, with a diameter >200 µm in at least one scan, and in contact with the optic disc boundary. The optic disc margin was automatically identified by the OptoVue software at the Bruch’s membrane/retinal pigment epithelium complex opening. When automated demarcation was inaccurate, a trained grader, masked to all clinical data, manually adjusted the boundary by locating the Bruch’s membrane opening. Two independent, masked observers assessed each scan for the presence of MvD. In cases of disagreement, a third adjudicator resolved the discrepancy.

Patient and eye characteristics data were presented as mean (95% CI) for continuous variables and count (%) for categorical variables. Categorical variables were compared using Fisher’s exact test. Logistic regression analyses were performed to evaluate the association between the presence of MvD and structural progression. The dependent binary (yes/no) variable was structural progression. To account for within-subject correlation, cluster-robust SEs at the eye level were applied. To determine whether β-zone PPA progression contributed independently to the prediction of structural change, we compared a model that included MvD and covariates with one that also included β-zone PPA. PPA progression was defined as ≥20% increase in β-zone PPA (assessed by area or angle measurements). We also evaluated 10% and 30% as alternative cutoffs for defining progression. This comparison was evaluated using both Wald and likelihood ratio tests. To explore potential effect modification, we also constructed a model including an interaction term between MvD and β-zone PPA progression and covariates. These probabilities were visualized with marginal effect plots for interpretability. For the agreement of structural progression, β-zone PPA progression, and presence of MvD, Cohen’s κ statistic was calculated using 30 independent cases. All analyses were conducted using Stata 17 (StataCorp). Statistical significance was defined as a two-sided P value <.05.