Longitudinal Studies utilizing Local Neural Retinal Function, measured by Multifocal Electroretinograms, for the Prediction of Diabetic Eye Disease
Diabetes is the number one cause of preventable blindness in working aged Americans. Diabetic macular edema is one of the most common reasons for vision loss in these patients. While treatments are available for macular edema, they are invasive and target the tissue after edema has already developed. They do not restore lost vision, but rather attempt to minimize further vision loss. Earlier and less invasive treatments for edema are needed.
The multifocal electroretinogram (mfERG), a local test of neural function, has been shown to be sensitive to changes in diabetes at all stages. This thesis investigates the relationship between the mfERG and diabetic eye disease (including retinal structure and function changes), ranging from its earliest clinical changes to sight threatening macular edema.
Five related studies are included, each as a separate chapter in this thesis.
The first study examines the reproducibility of the mfERG across instruments in patients with and without diabetes. This experiment was conceived because it was necessary to develop a method to combine data from our laboratory's two mfERG instruments for use in future experiments. Examining the reproducibly of the instruments is the first step in that process.
The second study builds on the past work of the lab. It creates a multivariate model using the mfERG implicit time to predict the onset of diabetic retinopathy in patients with no previous retinopathy. We were able to construct a model, which has 80% sensitivity and 74% specificity for the local prediction of the first signs of clinical retinopathy. This study also revealed strong differences between the mfERGs of type 1 and type 2 diabetics by identifying type of diabetes as a confounder of mfERG in this patient group.
The third study examines the relationship between retinal thickness and other diabetes health measures, such as blood pressure and blood glucose, which could alter thickness measurements. This study looked at factors that may confound our interpretation of retinal edema on an ocular coherence tomography (OCT), as increased retinal thickness accompanies edema in the diabetic retina. This correlation between retinal thickness and edema facilitates OCT to often be used as an outcome measure to identify edema. We found that increased blood pressure is associated with increased retinal thickness, even when blood pressure is in the normal range, in patients with retinopathy. However patients without retinopathy did not display the same trend.
The fourth study presented here is a cross sectional evaluation of patients with diabetic macular edema. We examined how the mfERG correlates with retinal thickness measured by OCT, edema on a fundus photo, and visual acuity, in these patients. We found local correlations between all these factors.
The fifth, and last study, builds on the previous chapters, in a longitudinal study evaluating patients with retinopathy at risk for edema. We used the mfERG to predict the onset of diabetic edema in patients with diabetic retinopathy. We found that mfERG amplitude and implicit time can predict local edema with a 72% sensitivity and specificity. Furthermore, a multivariate model, which includes the mfERG, measures along with systolic blood pressure and sex can predict the onset of edema with 84% sensitivity and 76% specificity.
Overall, we found the mfERG to be predictive of diabetic changes in the retina at many stages of diabetic eye disease. This extends from the onset of retinopathy to vision threatening diabetic edema. Our multivariate models have good sensitivity for making these local predictions. The predictive properties of these measurements in diabetes should be useful both in clinical trials or studies aimed at better treatments for diabetic eye disease at all levels, and for doctors treating patients at risk.