Retinal ultrastructural, electrophysiological, and microvascular morphological outcomes in diabetic macular edema treated with intravitreal bevacizumab

Background Investigation of retinal ultrastructural, electrophysiological, and microvascular morphological changes, as well as correlations between these changes and visual outcome in naive diabetic macular edema (DME) patients after intravitreal bevacizumab therapy (IVBT). Methods This prospective interventional study enrolled 31 DME patients' eyes treated with monthly IVBT for three months. Best-corrected visual acuity (BCVA) and intraocular pressure (IOP) were measured, and fundus fluorescein angiography, optical coherence tomography (OCT), microperimetry, as well as optical coherence tomography angiography (OCTA) were performed before and after IVBT. Patients were grouped based on BCVA improvement after three consecutive IVBT: group 1: > 10 letters, group 2: <= 5 letters, and group 3: between 6 and 10 letters. Results Mean BCVA increased significantly from 34.2 to 39.9 letters (p < 0.001). Central macular thickness decreased significantly from 335.1 to 276.4 mu m (p < 0.001). Fixation stability, retinal sensitivity, and local deficit all improved significantly (p < 0.001 for all). There was no statistically significant change in IOP (p = 0.665). Although OCTA parameters did not change significantly, lower foveal avascular zone (FAZ) area, higher foveal vessel density 300 mu m area around FAZ and deep plexus vascular density were associated with highly improved BCVA, retinal sensitivity, and local deficit. Also, there were no significant intergroup differences in gender, age, baseline BCVA, HbA1c, IOP, phakic/pseudophakic lens ratio, concomitant hypertension, and superficial capillary plexus vascular density. Conclusions IVBT was associated with significantly improved BCVA, retinal ultrastructural integrity, and electrophysiological patterns in naive DME patients. Improvements in retinal electrophysiology correlated with ultrastructural improvements, which could be predicted using OCTA.