Retinal Blood Vessel Positional Shifts and Glaucoma Progression

Abstract

To determine the characteristics and significance of retinal blood vessel (RBV) positional shifts over time in a cohort of patients with progressive glaucoma. Retrospective cohort study. Baseline and serial stereophotographs from 1 eye of 125 patients with open-angle glaucoma with ≥8 reliable Swedish interactive threshold algorithm standard visual fields (VFs) were included. On the basis of global rates of threshold sensitivity change, patients with glaucoma were divided into groups of minimal (<-0.02 decibels [dB]/year), moderate (-0.02 to-0.65 dB/year), or fast (≥-0.65 dB/year) progression. To determine whether graders' assessments of RBV positional shifts were false-positives, a control group consisting of 33 patients with glaucoma with 2 sets of photographs taken on the same day was included. Masked graders reviewed serial photographs aligned with automated alternation flicker (EyeIC, Narbeth, PA) and assessed them for the presence of any discrete RBV positional shifts (2 graders) and for traditional measures of structural progression (2 graders), including neuroretinal rim loss, parapapillary atrophy progression, and disc hemorrhage (DH). Presence or absence of RBV positional shifts, rates of VF progression, and presence or absence of traditional measures of structural progression. A total of 158 image sets (125 longitudinal and 33 same-day controls) from patients with glaucoma were included. Retinal blood vessel shifts were noted in 33 of 125 (26.4%) longitudinally followed glaucomatous eyes and 2 of 33 (6%) same-day control patients (P= 0.01). Agreement between graders I and II was 90.4% (kappa=0.77; P< 0.001). Eyes with RBV positional change progressed more rapidly than those without (-0.55 vs.-0.29 dB/year; 95% confidence interval [CI], 0.03-0.48); P= 0.03). Retinal blood vessel shift was present in 12.1% of minimal progressors versus 31.5% of moderate and fast progressors (P= 0.04). Rate of VF progression was statistically associated with RBV shift (odds ratio [OR], 2.2; 95% CI, 1.1-4.5; P= 0.03). Other variables significantly associated with RBV shift included neuroretinal rim loss (OR, 21.9; 95% CI, 5.7-83.6; P< 0.001) and DH (OR, 4.6; 95% CI, 1.5-15.5; P< 0.01). A multivariable model revealed that rim loss and DH, but not rate of functional change, were significantly associated with RBV shift. Retinal blood vessel positional shifts occurred in eyes with functionally progressive glaucoma, neuroretinal rim loss, and DH. This is a novel clinical finding that could help identify glaucoma progression or individuals at higher risk for future progression.