A pressing unmet need in the field of glaucoma diagnostics is to find methods for objective detection of disease worsening or prediction of visual field (VF) progression in eyes with advanced disease. Eyes with advanced glaucoma are at high risk of losing the remaining vision and blindness. Retinal nerve fiber layer (RNFL) and optic nerve head measures reach their measurement floor as glaucoma progresses beyond the early stages. Hence, functional assessment of the central VF is currently the main tool for monitoring advanced glaucoma. Our central hypothesis is that assessment of the macular retinal ganglion cell (RGC)/axonal complex can lead to improved detection or prediction of disease progression since the last RGCs to disappear in glaucoma reside in the central retina (the macula). We will test this hypothesis in a cohort of glaucoma subjects just reaching 5 years of follow-up and validate our methods in separate cohorts of glaucoma and normal subjects. Aim 1. Are macular thickness measures able to detect change earlier and with a stronger signal compared to RNFL measures in advanced glaucoma? We will measure progression rates for global and local macular and RNFL measures within a Bayesian hierarchical framework. We will compare progression rates and the proportion of progressing eyes/regions/sectors for macular and RNFL measures to normal eyes and account for differing scales, age-related decay, and treatment. Aim 2A. Can macular OCT thickness changes confirm and predict changes in central VFs for advanced glaucoma? We will estimate longitudinal/temporal structure-function relationships with Bayesian joint hierarchical longitudinal modeling of macular OCT and central 10° VF measures. These models will determine whether there is a contemporaneous or lagged deterioration of OCT and VF. We will assess the influence of baseline disease severity, treatment and other covariates on these joint longitudinal models. We will also compare the joint macular/central VF models to joint models of RNFL and 24° VFs and develop functional prediction models from 1 to 4 years ahead. Aim 2B. To validate the performance of prediction models, we will initiate a second prospectively enrolled cohort of patients meeting similar inclusion criteria and matched to the original cohort by age, gender, ethnicity and baseline glaucoma severity. We will compare VF point predictions (e.g., one- or two-visit step ahead) to the observed VF data. Aim 3. Develop software for combining macular structural and functional data in real time as a clinical tool for detection or prediction of progression. It will provide clinicians with structural/functional rates of change and structural ...