Background: To evaluate the ability of scanning laser polarimetry with variable corneal compensation (GDx-VCC) in separating healthy from glaucomatous patients with early visual field (VF) loss.
Methods: Sixty-two healthy and 48 glaucomatous age-matched patients with early glaucoma [mean deviation (MD): -1.74dB +/-1.69] underwent complete ophthalmological evaluation, automated achromatic perimetry (AAP) and retinal nerve fiber layer (RNFL) measurement with GDx-VCC. One randomly selected eye from each subject was considered. Glaucomatous VF defects had either Glaucoma Hemifield Test (GHT) outside normal limits or pattern standard deviation (PSD) outside 95% confidence limits. Mean (+/-SD) MD, PSD and GDx-VCC parameters in the two groups were compared by t-test. For each GDx-VCC parameter, area under receiver operating characteristics (AUROC) curve and sensitivity at predetermined specificity > or =80% and > or =95% were calculated. Moreover, the parameter with largest AUROC was evaluated by likelihood ratios (LRs).
Results: Mean values for MD, PSD and ten of 14 GDx-VCC parameters were significantly different between the two groups (P<0.001). The three parameters with largest AUROCs were the nerve fiber indicator (NFI) (0.870), superior average (0.817) and normalized superior area (0.816) (P=0.08 for differences between AUROCs). NFI displayed sensitivity values of 80.2% and 60.4% for specificity > or =80% and > or =95%, respectively. At NFI cutoff value of 30, positive LR was 34.9 (95% CI: 4.9-247.6) and negative LR was 0.45 (95% CI: 0.32-0.61). Interval LRs showed large effect on post-test probability for NFI values < or =18 or > or =31.
Conclusions: In our sample of eyes with early VF loss, GDx-VCC showed moderate-to-good discriminating ability. Among the best performing parameters, NFI had the largest AUROC, but several glaucomatous eyes (21, 43.8%) had NFI <30. This suggests that algorithm for NFI calculation requires some refinement when eyes with early VF loss are evaluated.