Abstract
Purpose::
To evaluate the diagnostic accuracy of GDx-VCC parameters in identifying glaucoma patients from a group of healthy eyes having atypical retardance pattern (ARP).
Methods::
The retinal nerve fiber layer estimates (RNFL) were obtained using GDx-VCC (Carl Zeiss Meditec). The RNFL was classified as having typical retardance pattern (TRP) or ARP on the basis the software-provided typical scan score. The typical scan score ranges between 0 (maximally atypical pattern) and 100 (completely typical pattern). A score of < 80 was considered ARP. RNFL data were collected for 176 eyes (115 healthy and 61 glaucoma) with ARP in the Glaucoma Service of the Ophthalmologica Department of the Semmelweis University. All eyes with glaucoma group had a repeatable perimetric defect. The sensitivity, specificity and ROC area were obtained and difference in ROC area was tested for statistical significance using Hanley and McNeil method (Radiology 1983).
Results::
The mean age and standard deviation (SD) of healthy and glaucoma group were 62.64 (10.01) and 62.87 (10.66) years respectively which was not significantly different (p=0.89). Overall the age of study participants was skewed towards older age and did not follow normal distribution and age (Shapiro-Wilk p<0.0001). The mean and SD values of the various GDx software-derived parameters for TSNIT (temporal, superior, nasal, inferior, temporal) Average, Superior Average, Inferior Average, and Nerve Fiber Indicator (NFI), were 59.30 (5.37), 69.19 (8.57), 67.74 (8.11) and 17.9 (9.01) for the healthy subjects, and 55.35 (9.99), 60.37 (11.63), 60.72 (13.52) and 33.09 (17.49) for the glaucoma group. The area under the ROC curve for the software-derived parameters (TSNIT Average, Superior Average, Inferior Average, and NFI) was 0.68, 0.74, 0.71 and 0.79, respectively. At a fixed specificity of 80 % the sensitivity for TSNIT average, Superior Average, Inferior Average and NFI was 54.1, 57.4%, 50.8% and 55.7%, respectively. The between-group difference in ROC area of both NFI and Superior Average were significantly greater than TSNIT average (p<0.05).
Conclusions::
None of the GDx-VCC parameters investigated appears optimal for identification glaucoma on eyes with ARP as the sensitivity was low at 80% specificity.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • imaging/image analysis: clinical • visual fields