Abstract
Purpose :
To quantify the intervisit test-retest (TRT) reproducibility for measurements of superior and inferior arcuate nerve fiber layer (SaNFL and IaNFL, respectively) in glaucoma patients with a range of visual field severities.
Methods :
Analyses were conducted with the same OCT scans, same eyes, and same visits as those previously reported by Mwanza 2010. Repeat scans on different days (n=5) were obtained during an interval too short for physiologic progression to occur. Thickness thresholds were applied to each thickness map to separate the rNFL into a deep and superficial portion. The superficial portion, or thickest rNFL region, was defined as the arcuate rNFL, and was divided into SaNFL and IaNFL based on location around the optic nerve. 61 different thresholds ranging from 20µm to 140µm were tested producing SaNFL and IaNFL volume, area, and thickness measurements for each eye at each threshold. TRTSD were obtained for each using MinQUE variance component analyses in which subject and threshold height were included as random effects. Because TRT variances were correlated with the means of the measurements, Box-Cox analyses were used to find optimal transformations for each measurement. TRTSD were calculated at each threshold on Box-Cox transformed measurements and the reproducibility was summarized with the intraclass correlation coefficients (ICC) of reproducibility and coefficient of variation (COV).
Results :
Box-Cox analyses of SaNFL and IaNFL volume, area, and thickness measurements from 50 eyes of 50 patients identified different transformations to uncouple variance from measurement means. Volumes were raised to the 0.67 power. Areas were raised to the 0.5 power, but this relationship broke down for areas measured at thresholds <50 µm, so these were excluded from further analysis. Thickness measurements were limited to a threshold of 45 µm and were not transformed. ICCs for all six types of measurements were >0.95 over a broad range of thresholds. For these thresholds, COVs were in the range 7-9% for volume, 7-10% for area, and <3% for thickness.
Conclusions :
Excellent ICCs of reproducibility and COVs between for SaNFL and IaNFL measurements of volume, area, and thickness were comparable to those previously published for standard RNFL thickness. This suggests possible efficacy for their use in identifying glaucoma and detecting progression.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.