Abstract
Purpose :
To identify objective criteria from clinical imaging and visual field testing that denote a useful, specific definition of glaucomatous optic neuropathy (GON) for use in comparing outcomes across clinical research studies in glaucoma.
Methods :
Nine glaucoma centers from 5 continents entered clinical data and evaluations, optical coherence tomography results, and visual field tests for 2,850 eyes of 1,531 patients in an online database. Each eye was graded by that patient's glaucoma specialist as either definite, probable, or not glaucomatous optic neuropathy (GON) based on all available information, eliminating eyes whose tests could have non-glaucoma causes. Mean age = 67 years, 53% female, and by derivation: 46% European, 14% Asian, 13% African, and 7% Hispanic. Objective data that matched each of the 3 diagnostic levels were constructed from optical coherence tomography (OCT) or field data, or both combined by eye.
Results :
Overall, 59% of eyes were classified as definite GON, 20% probable, and 22% not. The median field mean deviation (MD) for definite GON eyes was –5.5 dB. OCT data was from Cirrus (77%), Spectralis (14%), Nidek or Topcon (5% each). All fields were Zeiss HFA2 instrument. Potential definitions that used only OCT data (quadrant, clock hour, pointwise) or only field data using statistical abnormality of typical indices had inadequate specificity, identifying 10% each of not GON eyes. Using combined OCT and field statistical abnormality, the best definitional choice was abnormal (red) OCT superior or inferior RNFL quadrant that matched the opposite position of an abnormal glaucoma hemifield test (24-2 field) with 3 points abnormal at p<5% in that corresponding opposite hemifield. This was 73% sensitive, 98% specific for definite GON vs. not. Definite eyes that met this definition had more field loss (mean MD -9 dB) than those that did not (MD = -2 to 3 dB). Data for macular OCT and 10-2 fields were solicited, but available in only 1%. The best criterion achieved on either the first or second OCT/field pair had sensitivity 77% at specificity = 98%.
Conclusions :
Objective criteria to define GON are practical and may be useful for comparisons among clinical research studies to replace clinical assessment alone. Confirmation and improvement in these criteria will be sought in another database with scaled prediction of GON by multiple experts.
This is a 2020 ARVO Annual Meeting abstract.