Intersession Reproducibility of Retinal Nerve Fiber Layer Thickness Measurements by GDx-VCC in Healthy and Glaucomatous Eyes

Abstract

Background: To assess intersession reproducibility of retinal nerve fiber layer (RNFL) thickness measurements on scanning laser polarimetry with variable corneal compensation (GDx-VCC) in a sample of healthy subjects and glaucoma patients. Methods: One eye each from 29 healthy and 29 glaucomatous subjects was selected and underwent RNFL scanning by the same operator at baseline and 1 week later. Glaucoma diagnosis relied on the presence of a reproducible defect on automated perimetry. GDx-VCC parameters considered were those available on page 1 of the printout [TSNIT average and standard deviation (SD), superior and inferior average (SA, IA), Nerve Fiber Indicator]. Reproducibility was assessed by calculating coefficient of variation and intraclass correlation coefficient separately for the two groups and for each parameter. The percentage of eyes with an intersession difference in thickness parameters of more than 5% was also calculated. Results: Coefficient of variation was <6% for TSNIT average, SA and IA in both groups. Corresponding values for TSNIT SD in healthy subjects and in glaucoma patients were 13.7 and 11.4%, respectively, whereas for Nerve Fiber Indicator they were 82.9 and 13.3%. Intraclass correlation coefficient ranged from 0.794 to 0.907 in healthy subjects and from 0.924 to 0.972 in glaucoma patients. In healthy subjects, TSNIT average, SA and IA intersession difference was 5% or less in 55–69% of eyes, whereas the value for TSNIT SD was 34.5%. Corresponding values in glaucomatous eyes ranged from 69 to 79.3% for TSNIT average, SA and IA and was 37.9% for TSNIT SD. Conclusions: Intersession reproducibility of RNFL thickness measurements on GDx-VCC is high, both in healthy and in glaucomatous eyes. In a few cases, however, intersession variation may be larger than 10%. Caution is necessary while interpreting these changes during follow-up, in order to separate physiological variability from real RNFL thickness variations.