To simulate 24-2 visual field (VF) using optical coherence tomographic angiography (OCTA) for glaucoma evaluation.

Cross-sectional study.

One eye each of 39 glaucoma and 31 age-matched normal participants was scanned using 4.5-mm OCTA scans centered on the disc. The peripapillary retinal nerve fiber layer plexus capillary density (NFLP-CD, %area) was measured. The NFLP-CD and 24-2 VF maps were divided into 8 corresponding sectors using an extension of Garway-Heath scheme.

Sector NFLP-CD was transformed to a logarithmic dB scale and converted to sector simulated VF deviation maps. Comparing simulated and actual 24-2 VF maps, the worst sector was in the same or adjacent location in the same hemisphere 97% of the time. VF mean deviation (VF-MD) was simulated by NFLP mean deviation (NFLP-MD). The differences between NFLP-MD and VF-MD in early, moderate, and severe glaucoma stages were −0.9 ± 2.0, 0.9 ± 2.9, and 5.8 ± 3.2 dB. NFLP-MD had better (P = .015) between-visit reproducibility (0.63 dB pooled standard deviation) than VF-MD (1.03 dB). NFLP-MD had a significantly higher sensitivity than VF-MD (P < .001) and overall NFL thickness (P = .031).

OCTA-based simulated VF agreed well with actual 24-2 VF in terms of both the location and severity of glaucoma damage, with the exception of severe glaucoma in which the simulation tended to underestimate severity. The NFLP-MD had better reproducibility than actual VF-MD and holds promise for improving glaucoma monitoring. The NFLP-MD had better diagnostic accuracy than both VF-MD and overall NFL thickness and may be useful for early glaucoma diagnosis.