To evaluate contrast sensitivity function (CSF) in convalescent Vogt-Koyanagi-Harada (VKH) disease and investigate the relationship between CSF and chorioretinal thickness in VKH patients with and without sunset glow fundus (SGF). This is a cross-sectional study. Seventy-six eyes of VKH patients and 56 eyes of normal controls were evaluated. Patients were divided into SGF and non-SGF groups. The best-corrected visual acuity (BCVA) of all the participants was ≤0.0 logMAR. Their CSF and macular chorioretinal thickness were measured with quantitative CSF (qCSF) and Optical Coherence Tomography (OCT) and compared using repeated measures analysis of variance at the group level. Relationships between CSF and macular chorioretinal thickness were evaluated using generalized estimating equations. The CSF was significantly impaired in the SGF group compared to that in the control group (p = 0.001), especially at medium and high spatial frequencies. No significant CSF difference was found between the non-SGF group and control group, nor between the SGF and non-SGF groups. Compared to the controls, outer retinal thickness (ORT) in both VKH subgroups was significantly reduced (P < 0.001 or 0.005, respectively), although their outer nuclear layer thickness (ONLT) and choroidal thickness (CT) were not significantly different (both P = 1.000, P = 0.829 or 0.112, respectively). We found no significant correlation between CSF metrics and outer retinal thickness. Despite good recovery of visual acuity, reduced CSF and outer retina thickness were found in convalescent VKH patients. CSF may be an important and sensitive metric to evaluate functional vision in VKH disease.

Dear Editor,Submacular hemorrhage（SMH） is the accumulation of blood in the macular region caused by changes in retinal or choroidal circulation [1] . SMH may be caused by age-related macular degeneration（AMD）, pathological myopia, polypoid choroidal angiopathy（PCV） or retinal macroaneurysm [2] .

Abstract Purpose To study the relationship between macular thickness in Graves' ophthalmopathy (GO) patients and the occurrence and progression of the disease. Method 32 eyes of GO patients and 32 eyes of normal controls were included. All patients and control group participants underwent eye examinations, which included optical coherence tomography (OCT) analysis, best corrected visual acuity (BCVA) measurement, intraocular pressure (IOP) analysis, slit lamp biomicroscopy, OCT examination and visual field examination. We evaluated the retinal structure of GO patients by measuring the thickness of the macular area and the retinal nerve fiber layer (RNFL), studied the relationship between the thickness of the macular area and IOP, and compared the macular thickness of GO patients with and without visual field defects and visual impairment. Results Here, we evaluated structural retinal abnormalities, including the thickness of the macular area and the RNFL, in patients with GO. Part of the RNFL thickness of GO patients were significantly thinner than those in normal controls.In particular, after 7 months of follow-up, the macular thickness of GO patients was reduced. Macular thickness thinning was related with visual field defects and visual loss, and had a negative correlation with IOP. Conclusion Loss of macular and RNFL thickness associated with GO can be detected by OCT. The macular retina becomes thinner as the disease progresses. Peripapillary, macula thickness parameters may be used as an indicator for the diagnosis, prediction of disease progression, and assessment of visual function loss in GO patients.

AIM: To determine the clinical characteristics of intraretinal microvascular abnormality(IRMA)and associated neovascularization. METHODS:This was a prospective, observational study. We recruited treatment-naive patients with diabetic retinopathy(DR)with IRMAs or retinal neovascularization, confirmed using fundus fluorescein angiography(FFA)between October 7, 2016 and December 10, 2017. Under the guidance of FFA, IRMAs and neovascularizations were scanned using optical coherence tomography angiography(OCTA). Origins, initial layers, morphologic features, retinal nonperfusion areas(NPAs), location with capillary nonperfusion(CNP)and leakages demonstrated by FFA of IRMAs and associated neovascularization were documented and compared. Retinal nonperfusion areas were measured using Image J software. RESULTS: Thirty-nine eyes of 36 patients were enrolled in this prospective study. High quality images of twenty IRMAs and 22 IRMA-associated neovascularizations were identified using OCTA. All IRMAs originated from and drained into veins in pruned-tree-like shapes. IRMAs originated from the major retinal vessels at the margin of the CNP, extended into retina and were always confined within a single original nonperfused area. All IRMA-associated neovascularizations originated from IRMAs with a sea-fan-like appearance. The IRMA associated neovascularizations crossed retinal venous and extended to both sides. The main part of these structures was intraretinal, except some advancing tips that breached the internal limiting membrane(ILM)to form neovascularization, and were adhered firmly to the retina; 91%(20/22)of IRMA-associated neovascularizations were located in the CNPs, and 9%(2/22)were located at the margin of CNPs. The affiliated NPAs of IRMA-associated neovascularizations were 26.1mm2±4.2mm2, significantly larger than the IRMAs(12.9mm2±4.7mm2, P&lt;0.05). The initial layers showed no statistic difference between the groups(P&gt;0.05).CONCLUSION: OCTA is an effective method for detecting both IRMA and neovascularization in DR. IRMA and associated neovascularization had significantly different clinical characteristics that can be differentiated by OCTA, and therefore may be useful to better understand pathophysiological mechanisms and to guide efficient therapeutic strategies for DR patients.

PURPOSE:To explore the pathological features and clinical significance of three types of neovascularization elsewhere (NVE) in proliferative diabetic retinopathy.METHODS:Neovascularization elsewhere was classified based on the origins and morphologic features using fluorescein angiography and angiographic and structural optical coherence tomography. The topographical distribution, vitreoretinal interface, and responsiveness to panretinal photocoagulation were compared among three types of NVE.RESULTS:One hundred and twenty-seven NVEs were classified into three types. Type 1 NVE was concentrated along or adjacent to vascular arcades; Type 2 was distributed more peripherally than were Types 1 and 3 NVE. The arch bridge-like vitreoretinal interface accounted for 79% of Type 1 NVE. The flat and flat-forward vitreoretinal interface accounted for 95% and 100% in Type 2 and Type 3 NVE, respectively. At 3 months after panretinal photocoagulation, the regression rates for Types 1, 2, and 3 NVE were 82%, 100%, and 80%, respectively. Type 2 NVE showed best regression rate after panretinal photocoagulation (both P < 0.01).CONCLUSION:Three types of NVE determine the distinctly topographical distributions, vitreoretinal interface features, and differential responsiveness to panretinal photocoagulation treatment. This new concept may have important clinical implications in assessing the treatment and prognosis of proliferative diabetic retinopathy.

PURPOSE: To classify retinal neovascularization in untreated early stages of proliferative diabetic retinopathy (PDR) based on optical coherence tomography angiography (OCTA). DESIGN: A cross-sectional study. METHODS: Thirty-five eyes were included. They underwent color fundus photography, fluorescein angiography (FA), and OCTA examinations. Neovascularizations elsewhere (NVEs), neovascularizations at the disc (NVDs), and intraretinal microvascular abnormalities (IRMAs) were scanned by OCTA. The origin and morphology of NVE/NVD/IRMA on OCTA were evaluated. Retinal nonperfusion areas (NPAs) were measured using ImageJ software. RESULTS: In 35 eyes successfully imaged, 75 NVEs, 35 NVDs, and 12 IRMAs were captured. Three proposed subtypes of NVE were identified based on the origins and morphologic features. Type 1 (32 of 75, 42.67%) originated from the venous side, in a treelike shape. Type 2 (30 of 75, 40.00%) originated from capillary networks, with an octopus-like appearance. Type 3 (13 of 75, 17.33%) originated from the IRMAs, having a sea fan shape. NVD originated from the retinal artery, from the retinal vein, or from the choroid, and arose from the bending vessels near the rim of the optic disc. IRMA originated from and drained into retinal venules, extending into the retina. The initial layer and affiliated NPA were significantly different in the 3 subtypes of NVEs (all P<.01). CONCLUSIONS: OCTA allowed identification of the origins and morphologic patterns of neovascularization in PDR. The new classification of retinal neovascularization may be useful to better understand pathophysiological mechanisms and to guide efficient therapeutic strategies. (C) 2018 Elsevier Inc. All rights reserved.

Purpose To investigate the retinal vascular network alterations in highly myopic eyes. Methods Thirty-three highly myopic eyes from 21 subjects and 47 mildly myopic or emmetropic eyes from 24 healthy control subjects were enrolled. Optical coherence tomography angiography (OCTA) was used to image the superficial, deep, and whole retinal vascular plexuses at the macular region. Highly myopic eye images were analyzed after adjusting the ocular magnification using Bennett's formula. Fractal analysis (box counting method, Dbox) representing vessel density was performed in different annular and quadrantile zones of both large vessels and microvessels. Correlations between the vascular density, axial length, and spherical equivalent refractive error were analyzed. Results The average density (Dbox) of the superficial retinal annular zone (0.6-2.5 mm) microvessels was 1.741 ± 0.018 in highly myopic eyes and was shown to be significantly lower than that of the controls (1.773 ± 0.010, P < 0.001). Individual annular zone (bandwidth of 0.16 mm) analysis of highly myopic eyes revealed a significant level of microvessel alteration in all zones compared with the same zones in control eyes (P < 0.001). Furthermore, in the highly myopic group, the microvessel density was significantly correlated with axial length elongation in all three layers (r = -0.38 to -0.48; P < 0.05). Conclusions This study reveals retinal microvascular network alterations in highly myopic eyes, which correlates with axial length elongation. Fractal analysis of the microvasculature by OCTA images may help to characterize the underlying pathophysiological mechanisms involved in high myopia.