Management of diabetic macular oedema

New information gleaned from the DRCR.net Protocol T database has identified optical coherence tomography (OCT)-defined biomarkers that should help improve the management of patients with diabetic macular oedema (DME). The biomarker information was based on a post-hoc analysis of the DRCR.net Protocol T at one year. Analysis included the data of 629 patients with active DME and a best corrected visual acuity (BCVA) of 78-24 ETDRS letters at baseline, who were randomised 1:1:1 to be treated with aflibercept (2mg), bevacizumab (1.25mg) or ranibizumab (0.3mg). “Intraretinal fluid has the greatest predictive value for BCVA during treatment of DME,” said Ursula Schmidt-Erfurth MD, Medical University of Vienna, Austria. In Protocol T, after an initial loading phase, patients were monitored monthly and treated if a change in OCT or BCVA was seen. No treatment was given if the central retinal thickness was <250μm or the BCVA was 20/20 or better. Dr Schmidt-Erfurth’s team analysed the OCTs using automated algorithms that quantified several morphological features, the so-called biomarkers, as seen on spectral domain-OCT. These included subretinal fluid (SRF) volume, intraretinal cystoid fluid (IRF) volume, segmentation of retinal layers and total retinal thickness at baseline and weeks 4, 8, 12 and 24. The goal of Dr Schmidt-Erfurth’s study was to identify and quantify the effects of anti-vascular endothelial growth factor (anti-VEGF) treatment on BCVA, IRF and SRF and total retinal thickness on a population level. “Predictive models using a machine learning approach used 572 potential predictive factors incorporating each morphological feature, its location and its timing. These factors were then correlated with BCVA to devise a ranking of the predictive value of the most important factors,” she explained. Intensive regression of IRF after the first injection proved to be the best predictor of BCVA at week 24. A foveal location of the IRF is, in turn, a more significant predictor than parafoveal IRF. Conversely, SRF is not predictive, as there is a poor correlation between it and BCVA at 24 weeks. Significantly, aflibercept and ranibizumab appear to reduce IRF faster than bevacizumab, consistent with superior visual recovery in all patients with DME. “This analysis path is able to detect relevant features and evaluate their specific impact on the prognosis of DME. Computational biomarker analysis improves patient management, and it might improve the planning of clinical trials in DME,” she said. Ursula Schmidt-Erfurth: ursula.schmidt-erfurth@meduniwien.ac.at