Purchase this article with an account.
Dolly S. Chang, Karun S. Arora, Harry A. Quigley, Henry D. Jampel, Pradeep Ramulu, Michael V. Boland, Derek Welsbie, David S. Friedman, Wasu Supakontanasan; Detecting Glaucoma with Pupillography. Invest. Ophthalmol. Vis. Sci. 2012;53(14):5621.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To assess pupillography for discriminating glaucoma patients from those without glaucoma, we evaluated 3 approaches: 1) detecting a relative afferent pupillary defect (RAPD) between the two eyes, 2) comparing pupil responses to stimuli aimed at different parts of the retina within the same eye, and 3) comparing overall pupillary responses of eyes with glaucoma to responses of normal eyes.
The RAPiD (Konan Medical, USA), a binocular pupillographic device, is designed to analyze pupil responses at multiple, controlled stimulus intensities [bright (Br) and dim (Di)], while using varied stimulus patterns on the retina [full (F), central (C), peripheral (P), superior (S) and inferior (I)) and color stimuli [white (W), red (R), green (G), blue (B), and yellow (Y)]. This device records precise latency (Lat) and amplitude (Amp) of pupil responses . Glaucoma was defined as having both optic disc abnormalities and visual field (VF) loss. We included a full spectrum of glaucoma severity and did not exclude patients based on treatment and comorbidities. We enrolled controls from the accompanying persons of patients who do not have VF loss (based on testing) and had normal appearing optic nerves. Approximately 2% of the pupillographic data were missing, and these were imputed from the mean of related variables.
Thirty controls (age 62±10 years, 50% female) and 104 glaucoma patients (67±11 years, 51% female, mean deviation in the worse eye = -9.8±8.3 dB) from a single clinic were enrolled. Persons with glaucoma had a significantly larger RAPD compared to controls (0.047 log ratio of the amplitude between the two eyes versus 0.013, p<0.001). By using stepwise multivariate regression, we selected 5 parameters from approach 1 that were associated with glaucoma [RAPD of AmpBrFW, AmpDiCY, AmpBrCG, LatBrPR and LatDiCG]. Parameters from approach 2 and 3 were less associated and not selected. The model using these 5 parameters produced an area under the receiver operating characteristic curve of 0.92 (95% CI: 0.87-0.96). With a specificity of 97%, the sensitivity was 80%. These findings remained robust after bootstrap resampling techniques.
A prototype pupillography device appears able to discriminate glaucoma from normal with relatively high specificity and sensitivity. Screening performance of this device may be lower in community settings and additional study is warranted.
This PDF is available to Subscribers Only