Comparing three different modes of electroretinography in experimental glaucoma: diagnostic performance and correlation to structure

To compare diagnostic performance and structure-function correlations of multifocal electroretinogram (mfERG), full-field flash ERG (ff-ERG) photopic negative response (PhNR) and transient pattern-reversal ERG (PERG) in a non-human primate (NHP) model of experimental glaucoma (EG). At baseline and after induction of chronic unilateral IOP elevation, 43 NHP had alternating weekly recordings of retinal nerve fiber layer thickness (RNFLT) by spectral domain OCT (Spectralis) and retinal function by mfERG (7F slow-sequence stimulus, VERIS), ff-ERG (red 0.42 log cd-s/m(2) flashes on blue 30 scotopic cd/m(2) background, LKC UTAS-E3000), and PERG (0.8A degrees checks, 99% contrast, 100 cd/m(2) mean, 5 reversals/s, VERIS). All NHP were followed at least until HRT-confirmed optic nerve head posterior deformation, most to later stages. mfERG responses were filtered into low- and high-frequency components (LFC, HFC, > 75 Hz). Peak-to-trough amplitudes of LFC features (N1, P1, N2) and HFC RMS amplitudes were measured and ratios calculated for HFC:P1 and N2:P1. ff-ERG parameters included A-wave (at 10 ms), B-wave (trough-to-peak) and PhNR (baseline-to-trough) amplitudes as well as PhNR:B-wave ratio. PERG parameters included P50 and N95 amplitudes as well as N95:P50 ratio and N95 slope. Diagnostic performance of retinal function parameters was compared using the area under the receiver operating characteristic curve (A-ROC) to discriminate between EG and control eyes. Correlations to RNFLT were compared using Steiger's test. Study duration was 15 +/- 8 months. At final follow-up, structural damage in EG eyes measured by RNFLT ranged from 9% above baseline (BL) to 58% below BL; 29/43 EG eyes (67%) and 0/43 of the fellow control eyes exhibited significant (> 7%) loss of RNFLT from BL. Using raw parameter values, the largest A-ROC findings for mfERG were: HFC (0.82) and HFC:P1 (0.90); for ff-ERG: PhNR (0.90) and PhNR:B-wave (0.88) and for PERG: P50 (0.64) and N95 (0.61). A-ROC increased when data were expressed as % change from BL, but the pattern of results persisted. At 95% specificity, the diagnostic sensitivity of mfERG HFC:P1 ratio was best, followed by PhNR and PERG. The correlation to RNFLT was stronger for mfERG HFC (R = 0.65) than for PhNR (R = 0.59) or PERG N95 (R = 0.36), (p = 0.20, p = 0.0006, respectively). The PhNR flagged a few EG eyes at the final time point that had not been flagged by mfERG HFC or PERG. Diagnostic performance and structure-function correlation were strongest for mfERG HFC as compared with ff-ERG PhNR or PERG in NHP EG.