Abstract
PURPOSE: Intermediate age-related macular degeneration (iAMD) is a risk factor for progression to advanced stages, but rates of progression vary between individuals. Predicting individual risk is advantageous for programing timely and effective treatment and for patient stratification into future clinical trials. METHODS: We conducted a prospective and noninterventional study following patients with iAMD for 24 months. Optical coherence tomography parameters related with drusen, hyper-reflective foci (HRF), presence of incomplete retinal pigment epithelial and outer retinal atrophy (iRORA) and ellipsoid zone (EZ) status were explored at the baseline. Patients were reclassified at the end of the follow-up period and divided according to their progression. A risk prediction model for progression to late AMD was developed. RESULTS: A total of 135 patients were enrolled in the study and 30.4% developed late disease. A multivariate logistic regression model was created using those optical coherence tomography parameters, further optimized by backward feature elimination. Parameters offering the best fit in prediction progression were presence of iRORA, EZ status, drusen area and presence of HRF. iRORA is the feature that provides a higher probability of developing late AMD (odds ratio, 12.91; P = 0.000), followed by EZ disruption status (odds ratio, 3.54; P = 0.0018). The area under the receiver operating characteristic curve calculated for the testing set was 0.77 (95% confidence interval, 0.56-0.98). CONCLUSIONS: The combination of iRORA and EZ disruption constitute a high risk of progression to complete RORA within 2 years. TRANSLATIONAL RELEVANCE: We propose a practical and useful model to help clinicians in their daily practice in predicting individual progression to advanced AMD.