Article A Custom-Made Semiautomatic Analysis of Retinal Nonperfusion Areas After Dexamethasone for Diabetic Macular Edema Lisa Toto1,*, Rossella D’Aloisio1,*, Antonio Maria Chiarelli2, Luca Di Antonio1, Federica Evangelista1, Giada D’Onofrio1, Arcangelo Merla2, Mariacristina Parravano3, Guido Di Marzio1, and Rodolfo Mastropasqua4 1 Ophthalmology Clinic, Department of Medicine and Science of Ageing, University G. D’Annunzio Chieti-Pescara, Chieti, via dei Vestini 31, 66100, Italy 2 Department of Neuroscience, Imaging, and Clinical Sciences, University G. D’Annunzio Chieti-Pescara, Chieti, via dei Vestini 31, 66100, Italy 3 IRCCS – Fondazione Bietti, Rome, via Livenza 3, 00198, Italy 4 Institute of Ophthalmology, University of Modena and Reggio Emilia, Modena, Via del Pozzo 71, 41125, Italy Correspondence: Rossella D’Aloisio, Purpose: To evaluate the changes of retinal capillary nonperfusion areas and retinal via dei Vestini, 66100, Chieti, Italy. capillary vessel density of the superficial capillary plexus (SCP) and deep capillary plexus e-mail: [email protected] in patients with diabetes with diabetic macular edema treated with an intravitreal Received: January 30, 2020 dexamethasone implant (IDI). Accepted: March 29, 2020 Methods: We enrolled 28 patients with diabetic retinopathy and diabetic macular Published: June 11, 2020 edema candidates to IDI. All patients underwent widefield optical coherence tomog- Keywords: retinal nonperfusion raphy angiography with PLEX Elite 9000 device with 15 × 9 mm scans centered on the areas; intravitreal dexamethasone foveal center at baseline, 1 month, 2 months, and 4 months after IDI. In all the patients, implant; diabetic macular edema; the variation of the retinal capillary nonperfusion areas and of the retinal vessel density widefield OCT angiography of the SCP and deep capillary plexus were calculated using an automatic software written in Matlab (MathWorks, Natick, MA). Citation: Toto L, D’Aloisio R, Chiarelli AM, Di Antonio L, Evangelista F, Results: During follow-up, SCP showed a statistically significant reduction of ischemic D’Onofrio G, Merla A, Parravano M, areas at 1 month after IDI (P = 0.04) and slightly increased not significantly thereafter Di Marzio G, Mastropasqua R. A (P = 0.15). The percentage of nonperfusion areas changed from 11.4% at baseline, to custom-made semiautomatic 6.3% at 1 month, 8.1%, at 2 months, and 10.2% at 4 months. The whole vessel density of analysis of retinal nonperfusion SCP slightly increased (not significantly) from 35.30% at baseline to 38.00% at 1 month, areas after dexamethasone for and then decreased to 37.85% at 2 months and 36.04% at 4 months (P = 0.29). Retinal diabetic macular edema. Trans Vis capillary nonperfusion areas and retinal vessel density at the deep capillary plexus did Sci Tech. 2020;9(7):13, not change significantly (P = 0.31 and P = 0.73, respectively). https://doi.org/10.1167/tvst.9.7.13 Conclusions: Widefield optical coherence tomography angiography showed a decrease in retinal capillary nonperfusion areas after dexamethasone implant suggesting a possi- ble drug-related reperfusion of retinal capillaries particularly evident in the early period. Translational Relevance: A custom-made automatic analysis of retinal nonperfusion areas may allow a better and precise evaluation of ischemic changes after intravitreal therapy. and proliferative DR (PDR) of 6.96% in the diabetic Introduction population. Diabetic macular edema (DME) affects central vision at any stage of DR occurring in 6.81% Diabetic retinopathy (DR) is a leading cause of patients with diabetes.1 of visual loss in working-age populations with an Several studies demonstrated the importance of estimated prevalence of any form of DR of 34.6% retinal ischemia detection in DR as a predictor of Copyright 2020 The Authors tvst.arvojournals.org | ISSN: 2164-2591 1 This work is licensed under a Creative Commons Attribution 4.0 International License. Retinal Capillary Nonperfusion Areas Changes TVST | June 2020 | Vol. 9 | No. 7 | Article 13 | 2 progression of the disease. Increased risk of DR fied according to the simplified version of the Early progression was associated with greater involvement of Treatment Diabetic Retinopathy Study classification the peripheral retina in terms of retinal ischemia and of the American Academy of Ophthalmology Guide- neovascularization.2,3 lines Committee, complicated by center-involved DME Fluorescein angiography (FA) is an essential and candidates to IDI were recruited in the period diagnostic tool in DR staging and can identify both between December 2018 and May 2019 at our retina primary vascular lesions and particularly retinal center of the Ophthalmology Clinic of University G. nonperfusion related to neovascularization.4 Recently, D’Annunzio, Chieti-Pescara, Italy. The diagnosis of the introduction of ultra-widefield FA allowed to DR was established by means of retinal fundus exami- better investigate peripheral areas of nonperfusion nation, FA, and spectral domain OCT.22 and neovascularization that were missed with previous The study adhered to the tenets of the Declara- standard field imaging techniques and to establish tion of Helsinki and was approved by our Institutional correlation between peripheral and central ischemia Review Board. Written informed consent was obtained and between retinal ischemia and DME.5,6 from all participants of the study. In addition, changes of retinal ischemia after intrav- Criteria for inclusion were (1) age greater than 18 itreal treatment either were assessed using anti-vascular years old, (2) diagnosis of diabetes mellitus with DR endothelial factor (VEGF) or steroids was assessed.7–11 complicated with DME, and (3) a central macular Optical coherence tomography angiography thickness of greater than 300 micron. (OCTA) offers major advantages compared with The exclusion criteria were (1) previous intravitreal FA, including faster acquisition times, depth-resolved injections of anti-VEGF or dexamethasone implant retinal layer information, and the lack of invasive dye in the study eye, (2) diagnosis of glaucoma or ocular administration.12 hypertension (intraocular pressure of ≥21 mm Hg) at The first studies using conventional OCTA have baseline (T0) evaluation, (4) significant media opacities focused their attention on retinal perfusion parame- (cataract, vitreous hemorrhage), and (5) inflammatory ters of the macular area. They reported a decrease in diseases such as uveitis or retinal vascular occlusion, vascular density of the superficial and the deep capil- which may cause macular edema. lary vascular network in patients with diabetes with or without DR compared with healthy patients with decreasing values at increasing DR severity.13–17 Study Protocol However, a significant limitation of OCTA has All patients underwent a complete ophthalmologic always been a small field of view not allowing explo- examination, including best-corrected visual acuity, ration of retinal periphery. intraocular pressure reading, slit-lamp biomicroscopic For this reason, the introduction of widefield OCTA evaluation, and dilated fundoscopic examination. (WFOCTA) is an important step forward especially for All examinations were performed at T0 and at 1 the study of retinal vascular disease. month (T1), 2 months (T2), and 4 months (T3) after Some studies have investigated peripheral retina the dexamethasone implantation. in diabetic eyes to asses retinal capillary nonperfu- Moreover, in all enrolled patients OCTA was 17–20 sion and perfusion density using WFOCTA. In performed using widefield PLEX Elite 9000 device addition, reversal of peripheral ischemia after intrav- (Carl Zeiss Meditec Inc., Dublin, CA) at all time points. itreal anti-VEGF treatment was investigated in DR using WFOCTA.21 The aim of this study is to investigate the changes of Workflow Protocol retinal capillary nonperfusion areas and whole retinal vessel density of superficial capillary plexus (SCP) Imaging Protocol: Graders Review Quality, Segmenta- and deep capillary plexus (DCP) in patients with DR tion, and Artifacts Correction complicated by DME after intravitreal dexamethasone All eyes were scanned with the PLEX Elite 9000 implant (IDI) using WFOCTA. device (Carl Zeiss Meditec Inc.) that is able to acquire 100,000 A-scans per second with an axial resolution of about 5 μm in tissue, and with a lateral resolution at the retinal surface at about 14 μm. Methods This device uses a swept laser source which as a central wavelength of 1050 nm (1000–1100 nm full 23 Study Participants bandwidth). For each eye, WFOCTA volumes covering a 15x9 In this prospective observational study, 28 eyes of mm retinal area and centered at the fovea were acquired 28 patients with diabetes suffering from DR, classi- at each time point. Retinal Capillary Nonperfusion Areas Changes TVST | June 2020 | Vol. 9 | No. 7 | Article 13 | 3 Figure 1. (a) Raw OCTA image and the corresponded OCT B-scan. (b) Processed OCTA images. During image acquisition, a FastTrack motion infer regions of ischemia. The algorithm proceeded in correction software was used. five steps (Fig. 1). Poor quality images showing a signal strength index In the first step, each raw WFOCTA image was lower than 8 and with relevant motion or tilt artifacts low-pass filtered based on a two-dimensional gaussian were excluded from the analysis and repeated. smoothing kernel with standard deviation of 5 mm All participants’ eyes were imaged three times each, (350 pixels).24 and the best quality image was