Participants

All patients underwent an annual comprehensive ophthalmologic examination, including best-corrected visual acuity, slit-lamp biomicroscopy, Goldmann applanation tonometry, gonioscopy, dilated fundus examination, stereoscopic optic disc photography, ultrasonographic pachymetry, and standard automated perimetry (SAP) in both eyes. Semiannual examinations included intraocular pressure (IOP) measurement, spectral-domain optical coherence tomography (SDOCT) imaging, and SAP testing.

Glaucoma was defined as the presence of glaucomatous optic nerve damage (ie, the presence of focal thinning, notching, or localized or diffuse atrophy of the RNFL) with reliable (fixation losses <33%, false negatives <33%, and false positives <33%) and repeatable abnormal SAP tests using the 24-2 Swedish Interactive Threshold Algorithm with either a Pattern Standard Deviation outside the 95% normal limits or a Glaucoma Hemifield Test result outside the 99% normal limits. Patients with baseline 24-2 mean deviation (MD) >−6.0 dB were included in this study.

Inclusion criteria included (1) age ≥18 years, (2) open angles on gonioscopy, (3) best-corrected visual acuity of 20/40 or better, and (4) ≥3 years of follow-up with a minimum of 4 SDOCT scanning sessions.

Exclusion criteria included (1) history of trauma or intraocular surgery (except for uncomplicated cataract surgery or glaucoma surgery), (2) coexisting retinal disease, (3) uveitis, or (4) nonglaucomatous optic neuropathy. Participants with a diagnosis of Parkinson disease, Alzheimer disease, dementia, or a history of stroke were also excluded. Those with unreliable VFs or poor-quality SDOCT scans were also excluded from this report.

Optic Disc Classification

All color simultaneous stereophotographs were taken using a Nidek Stereo Camera Model 3-DX (Nidek Inc, Palo Alto, California, USA) after maximal pupil dilation. All photograph evaluations were performed using a simultaneous stereoscopic viewer (Asahi Pentax Stereo Viewer II; Pentax, Tokyo, Japan) with a standard fluorescent light bulb. As described by Nicolela and Drance and then in our previous study, the optic discs were classified into 4 types based on appearance: generalized cup enlargement (GE), focal ischemic (FI), myopic glaucomatous (MY), and senile sclerotic (SS). For optic discs with GE, the cup is concentrically enlarged and there are no localized areas of neuroretinal rim loss or pallor. For focal optic disc damage, the neuroretinal rim has a localized notch inferiorly or superiorly, while the remaining tissue is relatively well preserved. For optic discs classified as the high myopia phenotype, there is a tilted appearance, shallow cupping, and a myopic crescent of peripapillary atrophy. Optic discs with sclerotic damage have shallow cupping with marked areas of peripapillary atrophy. The classification of optic disc phenotype was evaluated by 2 experienced graders (E.E. and S.M.) using a stereoscopic viewer. Each grader was masked to the subject’s identity and the other test results. All included photographs were judged to be of adequate quality or better. Discrepancies between the 2 graders were resolved by consensus or adjudication by a third experienced grader. Inability to reach a consensus led to exclusion of the case.

OCT Measurements

The Spectralis SDOCT (software version 5.4.7.0; Heidelberg Engineering, Inc, Heidelberg, Germany) was used to obtain cpRNFL measurements. Details of its operation have been discussed elsewhere. The high-resolution protocol was used, obtaining 1536 A-scans from a 3.45-mm circle centered at the optic disc, providing an axial resolution of 3.9 µm and a lateral resolution of 6 µm. The Spectralis SDOCT included an automatic real time function that gathers multiple frames (B-scans) to increase image quality. The images were then averaged for noise reduction. The quality score ranges from 0 dB (poor) to 40 dB (excellent). To be included, all images were reviewed by experienced graders of the Imaging Data Evaluation and Assessment (IDEA) Center for noncentered scans, accurate segmentation, and a signal strength >15 dB. The cpRNFL measurements evaluated in this study included the mean global index, temporal quadrant (316-45 degrees), superior temporal sector (46-90 degrees), superior nasal sector (91-135 degrees), nasal quadrant (136-225 degrees), inferior nasal sector (226-270 degrees), and inferior temporal sector (271-315 degrees).

Statistical Analysis

Descriptive statistics were calculated as the mean and 95% confidence interval (95% CI). Categorical variables were compared using the χ ² test. The kappa value was used to estimate the extent of agreement between the 2 observers. Mixed effects modeling was used to compare ocular parameters among groups. Models were fitted with ocular measurements as response variable and classification groups as fixed effects. Measurements of bilateral eyes were nested within subject to account for the fact that eyes from the same individual were more likely to have similar measurements. Contributory factors affecting cpRNFL thinning were examined using a univariable and multivariable linear mixed model. Multivariable models were constructed including the following potential confounding factors: age, gender, baseline 24-2 mean deviation (MD), and any other ocular parameters in which the P value was <0.1 in univariable analysis.

The evaluation of the effect of optic disc phenotype parameters on mean rates of change in cpRNFL in each group were performed using a linear mixed model with random intercepts and random slopes. In this model, the average evolution of the outcome variable (cpRNFL measurements) was explored using a linear function of time, and random intercepts and random slopes were introduced with patient- and eye-specific deviations from this average evolution. The model can account for the fact that different eyes can have different rates of cpRNFL thinning over time, while accommodating correlations between both eyes of the same individual. ^, Because cpRNFL decline may depend on disease severity, an unstructured covariance between random effects was assumed, allowing for correlation between intercepts and slopes of change. In addition to putative predictors to examine the effect of variables on the baseline cpRNFL, interaction terms between time and predictors were included in the model to explore whether there is a significant effect of the predictor on changes of the outcome variable over time. Age, race, baseline 24-2 MD, and mean IOP during follow-up were used as predictors in this investigation. In the evaluation of progressors, individual slopes in each phenotype were classified into groups according to a pre-established range for mean rates of global cpRNFL thinning: slow, if change was greater than –1.0 µm/y; moderate, if between –1.0 and –2.0 µm/y; and fast, if change was less than –2.0 µm/y.

All statistical analyses were performed with commercially available software (Stata version 14; StataCorp, College Station, Texas, USA). Statistical significance for tests was set at P ≤.05.