Abstract
Purpose :
To quantitatively compare sector-specific glaucoma structure-function models using RNFLT circle sectors around the optic nerve head (ONH) with models using thickness measured directly at the ONH border (NRRT).
Methods :
From a large clinical glaucoma practice, circumpapillary optical coherence tomography (OCT) RNFLT scans around ONH and accompanying SITA Standard 24--2 Humphrey VFs (Cirrus HD OCT/Humphrey Field Analyzer, Carl Zeiss Meditec, Jena, Germany) were selected if the following reliability criteria were fulfilled: signal strength >= 6, no missing RFNLT data within the Cirrus standard circle (1.73 mm radius around ONH), no eye movement artifacts detected by a trained observer (OCT); false positive/negative rates <=20%, fixation loss rate <=33% (VF). One reliable pair of OCT/VF measurements of one randomly chosen eye per patient was selected. RNFLT on standard circle and NRRT, measured between Bruch's membrane opening and a machine-defined landmark on internal limiting membrane (Fig. 1a), were transferred from the Cirrus machine. VF and RNFLT/NRRT were partitioned into sectors according to the Garway-Heath map [1] (Fig. 1b). For each sector, linear regression models of VF mean total deviation (TD) were compared with RNFLT, NRRT, or both combined as regressors. A sector-specific Bayesian model comparison with Bayes Factor (BF) as selection criterion was calculated.
Results :
445 eyes of 445 patients met the selection criteria (age [mean±std. dev.]: 58.9±13.0 years, MD: -3.0±4.3 dB). RNFLT and NRRT were significantly (p<0.01) correlated to TD for all sectors (Fig. 1c). In the sector-specific RNFLT/NRRT model comparison (Fig. 2a), NRRT outperformed RNFLT for the central VF sector (BF=442) and RNFLT outperformed NRRT in all other sectors (BF between 8.9 and 1.3e+19). For all sectors, the combination of RNFLT and NRRT outperformed the respective best single parameter with positive evidence for the inner/temporal and very strong evidence for the nasal/peripheral VF sectors (Fig. 2b; BF between 4.8 and 886).
Conclusions :
A combination of RNFLT and NRRT improves glaucoma structure-function modeling, particularly in nasal/peripheral VF sectors. NRRT, which is available on the Cirrus printout, should be taken into consideration as well for the interpretation of VFs.
[1] Garway-Heath et al. Ophthalmology 107:1809-15, 2000
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.