| Description: |
During the last years, signature-based predictive models have been applied to different pathologies obtaining promising results. Those models take special consideration in situations where there is not clear clinical guidelines to choose between different treatments in a specific case for early application. Moreover, predictive models focused on treatment response play an indispensa-ble role on personalized medicine. Age-related macular degeneration (AMD) is an incurable disease associated with aging that destroys sharp and central vision. Most widespread treatment is intravitreal injection of anti-vascular endothelial grow factor (VEGF) agent to prevent choroidal neovascularization in those pa-tients. However, it is not effective on absolutely all patients since many of them do not show significant response to the treatment after three doses. The develop-ment of a predictive treatment response model to AMD would be very helpful to decide which treatment is suitable for each patient. Here, we present a novel clas-sification model based on a signature composed of 4 mRNAs and 1 miRNA, ob-tained from PBMCs, that predicts the response to ranibizumab with high accu-racy (Area Under the Curve of the Receiver Operating Characteristic curve = 0.968), before treatment. Our classification model represents a robust screening method to identify those patients with poor response to anti VEGF. Furthermore, this model in combination with other information, such as specific baselines char-acteristics, could help to establish a treatment plan on the first visit. |