Journal of Ophthalmology (Jun 2019)
Predicting the risk of diabetic retinopathy-associated macular edema in patients with type 2 diabetes mellitus
Abstract
Background: Previously, we have reported on the value of prothrombotic platelet phenotype as a factor for the development of diabetic maculopathy (DMP) and diabetic macular edema (DME) in patients with diabetic retinopathy (DR) and type 2 diabetes mellitus (DM2). Purpose: To predict DR-associated DME in patients with DM2 based on platelet dysfunction analysis. Materials and Methods: Ninety patients (92 eyes) with DM2 were included in the study. Of these eyes, 18, 43 and 31 were found to have, respectively, mild non-proliferative DR (NPDR), moderate or severe NPDR, and proliferative DR. Platelet aggregation agonists, adenosine diphosphate (ADP), platelet activation factor (PAF), and collagen (Sigma, St. Louis, MO), were used, and platelet aggregation was assessed with a Chrono-Log aggregometer. Methods for building logistic regression and neural network models were used to identify a set of independent variables associated with the risk for DME. Results: The risk for DME increased with increases in adrenaline- and PAF-induced platelet aggregations (p=0.03 and p=0.02, respectively) and decreased with an increase in collagen-induced platelet aggregation (p=0.046). There was a tendency to increase in the risk for DME with a one per cent increase in angiotensin II (ANG II)-induced platelet aggregation. Neural network analysis revealed non-linear associations of this risk with three independent variables, ANG II-, PAF-, and collagen-induced platelet aggregations. A neural network model with a sensitivity of 77.1% and specificity of 78.1% was created to predict DME based on this set of independent variables.
Keywords