Bruch's Membrane Opening-Minimum Rim Width Assessment With Spectral-Domain Optical Coherence Tomography Performs Better Than Confocal Scanning Laser Ophthalmoscopy in Discriminating Early Glaucoma Patients From Control Subjects.

PURPOSE: To compare the diagnostic performance and evaluate diagnostic agreement for early glaucoma detection between a confocal scanning laser ophthalmoscope (CSLO) and a spectral-domain optical coherence tomography (SD-OCT). PATIENTS AND METHODS: Fifty-five eyes of 55 open-angle glaucoma patients and 42 eyes of 42 healthy control subjects were enrolled in this observational, cross-sectional study. All participants underwent comprehensive ophthalmic examination, visual field testing, and optic nerve head and retinal nerve fiber layer imaging by CSLO (HRT3) and SD-OCT (Spectralis OCT). The agreements of categorical classifications were evaluated (κ statistics). Area under receiver operating characteristic curves (AUROC) and sensitivity at 95% fixed specificity were computed. RESULTS: The agreements of HRT3 and Spectralis OCT categorical classifications were fair to moderate (κ ranged between 0.33 and 0.54), except for Moorfields regression analysis of the HRT3 and the OCT global Bruch's membrane opening-minimum rim width (BMO-MRW) (criterion 1 κ=0.63, criterion 2 κ=0.67). The AUROC of OCT global BMO-MRW (0.956) was greater than those of HRT3 cup-to-disc area ratio (0.877, P=0.0063), vertical cup-to-disc ratio (0.872, P=0.0072), and cup area (0.845, P=0.0005). At 95% specificity, Spectralis OCT global BMO-MRW attained a higher sensitivity than HRT3 cup-to-disc area ratio (P<0.001). CONCLUSIONS: The BMO-MRW assessment with SD-OCT performed well in discriminating early glaucoma patients from control subjects and had a better performance than CSLO. The diagnostic classifications of HRT3 and Spectralis OCT may reach good agreement.