Deep Learning to Detect OCT-derived Diabetic Macular Edema from Color Retinal Photographs: A Multicenter Validation Study.

PURPOSE: To validate the generalizability of a deep learning system (DLS) that detects diabetic macular edema (DME) from 2-dimensional color fundus photographs (CFP), for which the reference standard for retinal thickness and fluid presence is derived from 3-dimensional OCT. DESIGN: Retrospective validation of a DLS across international datasets. PARTICIPANTS: Paired CFP and OCT of patients from diabetic retinopathy (DR) screening programs or retina clinics. The DLS was developed using data sets from Thailand, the United Kingdom, and the United States and validated using 3060 unique eyes from 1582 patients across screening populations in Australia, India, and Thailand. The DLS was separately validated in 698 eyes from 537 screened patients in the United Kingdom with mild DR and suspicion of DME based on CFP. METHODS: The DLS was trained using DME labels from OCT. The presence of DME was based on retinal thickening or intraretinal fluid. The DLS's performance was compared with expert grades of maculopathy and to a previous proof-of-concept version of the DLS. We further simulated the integration of the current DLS into an algorithm trained to detect DR from CFP. MAIN OUTCOME MEASURES: The superiority of specificity and noninferiority of sensitivity of the DLS for the detection of center-involving DME, using device-specific thresholds, compared with experts. RESULTS: The primary analysis in a combined data set spanning Australia, India, and Thailand showed the DLS had 80% specificity and 81% sensitivity, compared with expert graders, who had 59% specificity and 70% sensitivity. Relative to human experts, the DLS had significantly higher specificity (P = 0.008) and noninferior sensitivity (P < 0.001). In the data set from the United Kingdom, the DLS had a specificity of 80% (P < 0.001 for specificity of >50%) and a sensitivity of 100% (P = 0.02 for sensitivity of > 90%). CONCLUSIONS: The DLS can generalize to multiple international populations with an accuracy exceeding that of experts. The clinical value of this DLS to reduce false-positive referrals, thus decreasing the burden on specialist eye care, warrants a prospective evaluation.

Remote Tool-Based Adjudication for Grading Diabetic Retinopathy.

PURPOSE: To present and evaluate a remote, tool-based system and structured grading rubric for adjudicating image-based diabetic retinopathy (DR) grades. METHODS: We compared three different procedures for adjudicating DR severity assessments among retina specialist panels, including (1) in-person adjudication based on a previously described procedure (Baseline), (2) remote, tool-based adjudication for assessing DR severity alone (TA), and (3) remote, tool-based adjudication using a feature-based rubric (TA-F). We developed a system allowing graders to review images remotely and asynchronously. For both TA and TA-F approaches, images with disagreement were reviewed by all graders in a round-robin fashion until disagreements were resolved. Five panels of three retina specialists each adjudicated a set of 499 retinal fundus images (1 panel using Baseline, 2 using TA, and 2 using TA-F adjudication). Reliability was measured as grade agreement among the panels using Cohen's quadratically weighted kappa. Efficiency was measured as the number of rounds needed to reach a consensus for tool-based adjudication. RESULTS: The grades from remote, tool-based adjudication showed high agreement with the Baseline procedure, with Cohen's kappa scores of 0.948 and 0.943 for the two TA panels, and 0.921 and 0.963 for the two TA-F panels. Cases adjudicated using TA-F were resolved in fewer rounds compared with TA (P < 0.001; standard permutation test). CONCLUSIONS: Remote, tool-based adjudication presents a flexible and reliable alternative to in-person adjudication for DR diagnosis. Feature-based rubrics can help accelerate consensus for tool-based adjudication of DR without compromising label quality. TRANSLATIONAL RELEVANCE: This approach can generate reference standards to validate automated methods, and resolve ambiguous diagnoses by integrating into existing telemedical workflows.

A Review of Antimicrobial Therapy for Infectious Uveitis of the Posterior Segment.

Treatment of infectious posterior uveitis represents a therapeutic challenge for ophthalmologists. The eye is a privileged site, maintained by blood ocular barriers, which limits penetration of systemic antimicrobials into the posterior segment. In addition, topical and subconjunctival therapies are incapable of producing sufficient drug concentrations, intraocularly. Posterior infectious uveitis can be caused by bacteria, virus, fungi, or protozoa. Mode of treatment varies greatly based on the infectious etiology. Certain drugs have advantages over others in the treatment of infectious uveitis. Topical and systemic therapies are often employed in the treatment of ocular infection, yet the route of treatment can have limitations based on penetration, concentration, and duration. The introduction of intravitreal antimicrobial therapy has advanced the management of intraocular infections. Being able to bypass blood-ocular barriers allows high drug concentrations to be delivered directly to the posterior segment with minimal systemic absorption. However, because the difference between the therapeutic and the toxic doses of some antimicrobial drugs falls within a narrow concentration range, intravitreal therapy could be associated with ocular toxicity risks. In many cases of infectious uveitis, combination of intravitreal and systemic therapies are necessary. In this comprehensive review, the authors aimed at reviewing clinically relevant data regarding intraocular and systemic antimicrobial therapy for posterior segment infectious uveitis. The review also discussed the evolving trends in intraocular treatment, and elaborated on antibiotic pharmacokinetics and pharmacodynamics, efficacy, and adverse effects.

Use of Autologous Serum Tears for the Treatment of Ocular Surface Disease From Patients With Systemic Autoimmune Diseases.

PURPOSE: To describe the safety and efficacy of autologous serum tears (AST) in managing ocular surface disease resistant to conventional therapy in patients with systemic autoimmune disease(s). DESIGN: Retrospective, interventional case series. METHODS: Records of patients from 2009 to 2015 with systemic autoimmune disease treated with AST (20%-50%) for chronic surface disease were analyzed. Standardized measures of subjective dry eye symptoms, objective dry eye staining of the cornea, and slit-lamp findings including punctate epithelial erosion (PEE), filamentary keratopathy (FK), and corneal epithelial defects (KED) were compared during first and last visit. We attempted to standardize outcomes by creating a scale from 1 to 4 for subjective and objective components: worsening (1), no improvement (2), partial improvement (3), and complete resolution (4). RESULTS: Fifty-one patients (101 eyes) were included. The mean age was 59.8 ± 13.2 years (72.5% female). Average use of AST was 14.3 ± 11.7 months. Complete objective improvement of initial slit-lamp findings was achieved in 30% and partial improvement in 55% of eyes. Presence of PEE, FK, and KED decreased from 92.1% to 52.5% (P < .001), from 22.8% to 9.9% (P = .02), and from 5% to 2% (P = .44) of the eyes, respectively. Full subjective improvement of symptoms was achieved in 34.6%, partial in 50.5%, and none in 14.9% of patients. No adverse side effects were noted during follow-up. CONCLUSIONS: AST are a safe and effective adjunct therapy in improving both objective signs and subjective symptoms of ocular surface disorders associated with systemic autoimmune disease(s).

Impact of Total Pars Plana Vitrectomy on Postoperative Complications in Aphakic, Snap-On, Type 1 Boston Keratoprosthesis.

PURPOSE: To determine the impact of total pars plana vitrectomies (PPVs) with peripheral shaving of the vitreous base on the rates of postoperative complications in patients with aphakic, snap-on type I Boston keratoprostheses (KPros). DESIGN: Retrospective, consecutive case series. PARTICIPANTS: A total of 48 eyes in 46 patients with implantation of aphakic, snap-on type 1 Boston KPros performed at a tertiary care facility between January 1, 2007, and December 31, 2013, were included. METHODS: The cumulative incidences of postoperative complications were compared between patients who underwent total PPVs with shaving of the vitreous base (n = 22) and those who had partial PPVs or anterior vitrectomies (AVs) at the time of KPro implantation (n = 26). MAIN OUTCOME MEASURES: Rates of complications between patients who underwent total PPVs and partial PPVs or AVs. RESULTS: The rate of total postoperative complications was lower in the total PPV group (P = 0.018, log-rank test). In particular, eyes that underwent total PPVs had lower rates of retroprosthetic membranes (RPMs) requiring intervention (P = 0.049) and less vision loss due to glaucoma progression (P = 0.046). There was also a trend for fewer corneal melts (P = 0.060) and less sight-threatening complications (P = 0.051) in the total vitrectomy group. There was no difference in the rates of KPro extrusion (P = 0.41), endophthalmitis or vitritis (P = 0.15), retinal detachments (P = 0.76), cystoid macular edema (P = 0.83), or timing of complications between the 2 groups. The mean preoperative and postoperative visual acuities were similar between the 2 groups (P = 0.97). The mean follow-up was 49±22 months. CONCLUSIONS: Eyes that underwent total PPVs during implantation of aphakic, snap-on, type I Boston KPros had less postoperative complications than eyes with partial PPVs or AVs during the average 4 years of follow-up.

Contact-Lens-Associated Purpureocillium Keratitis: Risk Factors, Microbiologic Characteristics, Clinical Course, and Outcomes.

PURPOSE: To study the risk factors, microbiologic characteristics, clinical course, and outcomes of patients with Purpureocillium keratitis at a tertiary eye care center in south Florida. MATERIALS AND METHODS: All medical records during a seven-year period starting January 1, 2007, were reviewed. Twenty-eight culture-proven Purpureocillium keratitis cases with complete medical records presenting to our institution were included in this retrospective, observational case series. Data collected included predisposing factors, therapeutic interventions, treatment duration, and visual outcomes. RESULTS: Twenty patients (71.4%) had a history of soft contact lens use, with only two for therapeutic use. Other identified risk factors were trauma and immunosuppression. Fifteen patients (53.6%) received topical corticosteroid treatment prior to the diagnosis of fungal keratitis. Thirteen patients (46.4%) were on Natamycin treatment prior to Purpureocillium identification. As a group, the average best-corrected visual acuity (BCVA) at presentation was 1.1 logMAR; upon the final evaluation, it was 1.0 logMAR. The BCVA on last evaluation for the eight patients presenting to our institution within two weeks of onset of symptoms was 0.3 log MAR, and all patients in this group responded to medical management. The final BCVA for 20 patients presenting two weeks after onset of symptoms was 1.2 logMAR. There was a significant difference in the final BCVA between Group 1 and Group 2 (p = 0.004), but no difference in steroid use or previous treatments. Previous steroid use tended to extend time to presentation and was significantly associated with a worse final visual outcome (1.2 versus 0.6 logMAR; p = 0.0474). Previous Natamycin use was significantly associated with a worse final visual outcome (1.4 versus 0.6 logMAR; p = 0.014). CONCLUSION: Purpureocillium keratitis can have devastating consequences to visual function and even lead to enucleation. Physicians should make every effort to arrive at an earlier microbiological diagnosis, as this is associated with better outcomes and less need for surgical intervention. The first line use of voriconazole is recommended, and steroid use should be avoided, as their previous use is associated with worse visual outcomes.

Causes and correction of dissatisfaction after implantation of presbyopia-correcting intraocular lenses.

PURPOSE: The purpose of this study was to assess the causes and possible solutions for patient dissatisfaction after the implantation of presbyopia-correcting intraocular lenses (IOLs). METHODS: This study was a retrospective review of clinical records. All patients who were seen between January 2009 and December 2013 whose primary reason for consultation was dissatisfaction with visual performance after presbyopia-correcting IOL implantation were included in the study. A single treating physician, who determined the most probable cause of dissatisfaction, decided which interventions to pursue following the initial consultation. RESULTS: Data from 74 eyes of 49 patients were analyzed. The most common cause for complaint was blurry or foggy vision both for distance and near (68%). Complaints were most frequently attributed to residual refractive error (57%) and dry eye (35%). The most common interventions pursued were treatment of refractive error with glasses or contact lenses (46%) and treatment for dry eye (24%). Corneal laser vision correction was done in 8% of eyes; 7% required an IOL exchange. After the interventions, 45% of patients had completed resolution of symptoms, 23% of patients were partially satisfied with the results, and 32% remained completely dissatisfied with the final results. CONCLUSION: The most identifiable causes of dissatisfaction after presbyopia-correcting IOL implantation are residual refractive error and dry eye. Most patients can be managed with conservative treatment, though a significant number of patients remained unsatisfied despite multiple measures.

Psychomotor vigilance and visual field test performance.

INTRODUCTION: Visual field test is an invaluable tool to evaluate the detection and progression of glaucoma. On the other hand, as a subjective test, reliable results depend on patients' optimum performance including vigilance during the test. The purpose of this study was to understand patient's attitude and preferences about the visual field test taking, and in addition to assess the relationship between the reliability indices of visual field testing (VFT) and the Walter Reed Psychomotor Vigilance Test (PVT). MATERIAL AND METHODS: In this cross-sectional, non-interventional study, VFT reliability indices were recorded for all 140 patients. In the 46 patients who completed the PVT, average reaction time and minimum reaction times were recorded. All 140 patients completed a survey about their VFT experience. RESULTS: Based on the survey results, most subjects found VFT to be difficult. Subjects who rated their VFT performance excellent/good had similar VFT reliability rates compared to the ones with fair/poor self-assessments. The average reaction time (RT) was 0.6 seconds (0.3-1.9 seconds). Higher average RT was associated with increased age and less formal education (p < 0.001 and 0.03, respectively). There was a marginally significant correlation between average RT and the VFT "reliability" status (p = 0.045). CONCLUSIONS: While VFT is the least favorable part of the work-up for glaucoma patients, their self-assessment about VFT performance did not correlate with current VF reliability indicators. Although reliability of the VFT was not strongly affected by slower reaction times when tested by the PVT, the effect of psychomotor reaction time on other aspects of test outcomes is unknown and warrants further investigation.

Diabetic retinopathy: current management and experimental therapeutic targets.

Diabetic retinopathy, a devastating ocular complication of diabetes mellitus, is the leading cause of blindness among working-age adults in the United States and is a serious public health problem throughout the world. Standard treatment is retinal laser photocoagulation, which is an invasive procedure with considerable limitations and adverse effects. Understanding the biochemical changes and molecular events that occur with diabetes as well as with retinopathy could lead to new and effective treatments. Research indicates an association between oxidative stress and the development of complications from diabetes, including retinopathy. Thus, many sites and sources of oxidative stress that may be involved in the development of diabetic retinopathy have been studied. However, in clinical trials, classic antioxidants have not been beneficial for patients with diabetes. Additional studies are needed to identify treatments that selectively target oxidative stress-mediated protein modification and thus prevent or at least delay diabetic retinopathy.

Poly(ADP-ribose)polymerase inhibition counteracts cataract formation and early retinal changes in streptozotocin-diabetic rats.

PURPOSE: This study evaluated the role for poly(ADP-ribose) polymerase (PARP) in diabetes-induced cataractogenesis and early retinal changes. METHODS: Control and streptozotocin (STZ)-diabetic rats were treated with or without the PARP inhibitors 1,5-isoquinolinediol (ISO; 3 mg kg(-1) d(-1) intraperitoneally) and 10-(4-methyl-piperazin-1-ylmethyl)-2H-7-oxa-1,2-diaza-benzo[de]anthracen-3-1 (GPI-15427, 30 mg kg(-1) d(-1) orally) for 10 weeks after the first 2 weeks without treatment. Lens clarity was evaluated by indirect ophthalmoscopy and slit lamp examination, and retinal changes were evaluated by immunohistochemistry and Western blot analysis. In in vitro studies, cultured human lens epithelial cells and bovine retinal pericytes and endothelial cells were exposed to high glucose or palmitate. RESULTS: PARP is expressed in lens, and poly(ADP-ribosyl)ated proteins are primarily localized in the 38- to 87-kDa range of the protein spectrum, with several minor bands at 17 to 38 kDa. The 38- to 87-kDa and the 17- to 38-kDa poly(ADP-ribosyl)ated protein expression increased by 74% and 275%, respectively, after 4 weeks of diabetes and by approximately 65% early after exposure of lens epithelial cells to 30 mM glucose. Both PARP inhibitors delayed, but did not prevent, the formation of diabetic cataract. The number of TUNEL-positive nuclei in flatmounted retinas increased approximately 4-fold in STZ diabetic rats, and this increase was prevented by ISO and GPI-15427. Both PARP inhibitors reduced diabetes-induced retinal oxidative-nitrosative and endoplasmic reticulum stress and glial activation. GPI-15427 (20 microM) prevented oxidative-nitrosative stress and cell death in palmitate-exposed pericytes and endothelial cells. CONCLUSIONS: PARP activation is implicated in the formation of diabetic cataract and in early retinal changes. These findings provide a rationale for the development of PARP inhibitors for the prevention of diabetic ocular complications.

Aldose reductase inhibitor fidarestat counteracts diabetes-associated cataract formation, retinal oxidative-nitrosative stress, glial activation, and apoptosis.

This study was aimed at evaluating the potent and specific aldose reductase inhibitor fidarestat, on diabetes-associated cataract formation, and retinal oxidative-nitrosative stress, glial activation, and apoptosis. Control and streptozotocin-diabetic rats were treated with or without fidarestat (16 mg kg(-1)d(-1)) for 10 weeks after an initial 2-week period without treatment. Lens changes were evaluated by indirect ophthalmoscopy and portable slit lamp. Nitrotyrosine, poly(ADP-ribose), and glial fibrillary acidic protein expression were assessed by immunohistochemistry. The rate of apoptosis was quantified in flat-mounted retinas by TUNEL assay with immunoperoxidase staining. To dissect the effects of high glucose exposure in retinal microvascular cells, primary bovine retinal pericytes and endothelial cells were cultured in 5 or 30 mM glucose, with or without fidarestat (10 microM) for 3-14 days. Apoptosis was assessed by TUNEL assay, nitrotyrosine and poly(ADP-ribose) by immunocytochemistry, and Bax and Bcl-2 expression by Western blot analyses. Fidarestat treatment prevented diabetic cataract formation and counteracted retinal nitrosative stress, and poly(ADP-ribose) polymerase activation, as well as glial activation. The number of TUNEL-positive nuclei (mean +/- SEM) was increased approximately 4-fold in diabetic rats vs. controls (207+/-33 vs. 49+/-4, p<0.01), and this increase was partially prevented by fidarestat (106+/-34, p<0.05 vs. untreated diabetic group). The apoptotic cell number increased with the prolongation of exposure of both pericytes and endothelial cells to high glucose levels. Fidarestat counteracted nitrotyrosine and poly(ADP-ribose) accumulation and apoptosis in both cell types. Antiapoptotic effect of fidarestat in high glucose-exposed retinal pericytes was not associated with the inhibition of Bax or increase in Bcl-2 expression. In conclusion, the findings, i) support an important role for aldose reductase in diabetes-associated cataract formation, and retinal oxidative-nitrosative stress, glial activation, and apoptosis, and ii) provide a rationale for the development of aldose reductase inhibitors, and, in particular, fidarestat, for the prevention and treatment of diabetic ocular complications.

Peroxynitrite mediates retinal neurodegeneration by inhibiting nerve growth factor survival signaling in experimental and human diabetes.

OBJECTIVE: Recently we have shown that diabetes-induced retinal neurodegeneration positively correlates with oxidative stress and peroxynitrite. Studies also show that peroxynitrite impairs nerve growth factor (NGF) survival signaling in sensory neurons. However, the causal role of peroxynitrite and the impact of tyrosine nitration on diabetes-induced retinal neurodegeneration and NGF survival signaling have not been elucidated. RESEARCH DESIGN AND METHODS: Expression of NGF and its receptors was examined in retinas from human and streptozotocin-induced diabetic rats and retinal ganglion cells (RGCs). Diabetic animals were treated with FeTPPS (15 mg x kg(-1) x day(-1) ip), which catalytically decomposes peroxynitrite to nitrate. After 4 weeks of diabetes, retinal cell death was determined by TUNEL assay. Lipid peroxidation and nitrotyrosine were determined using MDA assay, immunofluorescence, and Slot-Blot analysis. Expression of NGF and its receptors was determined by enzyme-linked immunosorbent assay (ELISA), real-time PCR, immunoprecipitation, and Western blot analyses. RESULTS: Analyses of retinal neuronal death and NGF showed ninefold and twofold increases, respectively, in diabetic retinas compared with controls. Diabetes also induced increases in lipid peroxidation, nitrotyrosine, and the pro-apoptotic p75(NTR) receptor in human and rat retinas. These effects were associated with tyrosine nitration of the pro-survival TrkA receptor, resulting in diminished phosphorylation of TrkA and its downstream target, Akt. Furthermore, peroxynitrite induced neuronal death, TrkA nitration, and activation of p38 mitogen-activated protein kinase (MAPK) in RGCs, even in the presence of exogenous NGF. FeTPPS prevented tyrosine nitration, restored NGF survival signal, and prevented neuronal death in vitro and in vivo. CONCLUSIONS: Together, these data suggest that diabetes-induced peroxynitrite impairs NGF neuronal survival by nitrating TrkA receptor and enhancing p75(NTR) expression.