Optic Disc Photography vs. Optical Coherence Tomography (OCT) in Detection of Glaucoma Progression

Abstract

Purpose : To compare glaucoma specialists’ evaluation of progression of glaucoma using optic disc photos and OCT, according to the intra- and inter-rater reproducibility of each technique, agreement between techniques, and ability of each technique to detect visual field (VF) progression.

Methods : We collected monoscopic disc photos, OCT reports (minimum rim width and retinal nerve fiber layer scans) and 24-2 VF tests from glaucoma subjects and healthy controls followed in longitudinal observational cohorts. Subjects with a baseline disc photo and OCT (taken on the same day) and a follow-up disc photo and OCT with at least 3 years of follow-up were included. Five glaucoma specialists evaluated the disc photos and OCT reports, ranking the likelihood of glaucoma progression as follows: definitely no progression, likely no progression, likely progression, definitely progression, corresponding to scores 1-4, respectively (Fig 1). Eight subjects were duplicated to determine intra-rater reproducibility and another 8 subjects were duplicated and presented in reverse order to estimate false positive (FP) progression detection. The controls were used to provide another estimate of FP and excluded from other analyses. VF glaucoma progression was defined as a classification of ‘possible’ or ‘likely’ on Guided Progression Analysis software, using VFs between the baseline and follow-up examinations. Reproducibility analyses were conducted with intraclass correlation (ICC). Average scores given by examiners were compared with Wilcoxon test. Accuracy to detect VF progression was explored using area under receiver operating characteristics curve (AUC).

Results : We included 72 glaucoma and 7 control subjects. Intra-rater reproducibility was higher for OCT than disc photos (mean ICC= 0.82 vs. 0.55). Intra-rater agreement between disc photos and OCT was poor (mean ICC= 0.23). Inter-rater agreement between the 5 examiners was moderate for OCT and poor for disc photos (ICC= 0.62 vs. 0.37). The mean score for inverted order images was higher in photos than OCT (2.1 vs. 1.6, p= 0.07), which could suggest greater FP in photos than OCT. However, the mean score in controls was the same (1.6) with both techniques. The techniques had a comparable AUC (Fig 2).

Conclusions : OCT evaluations had higher intra and inter-rater agreement than evaluations of optic disc photos. Both techniques had similarly low detection of VF progression.

This is a 2021 ARVO Annual Meeting abstract.