Risk factors for diabetic retinopathy, diabetic macular edema, and sight-threatening diabetic retinopathy.

OBJECTIVE: To identify the risk factors for the development of diabetic retinopathy (DR), diabetic macular edema (DME), and sight-threatening DR (STDR) based on a city-wide diabetes screening program. RESEARCH DESIGN AND METHODS: Diabetic patients were prospectively recruited between June 2016 and December 2022. All patients underwent dilated fundus photography centered on the disc and macula or macular spectral domain optical coherence tomography (SD-OCT) scan. Complete medical history was documented. Systematic examination, blood analysis, and urinalysis were performed. Multivariate logistic regression analysis adjusting for age and sex was conducted. RESULTS: Out of 7274 diabetic patients, 6840 had gradable images, among which 3054 (42.0%) were graded as DR, 1153 (15.9%) as DME, and 1500 (20.6%) as STDR. The factors associated with DR, DME, and STDR included younger age (odds ratio [OR]: 0.96, 0.97, and 0.96 respectively), lower BMI (OR: 0.97, 0.95, and 0.95 respectively), longer duration of diabetes (OR: 1.07, 1.03, and 1.05 respectively) and positive of urinary albumin (OR: 2.22, 2.56, and 2.88 respectively). Other associated factors included elevated blood urea nitrogen (OR: 1.22, 1.28, and 1.27 respectively), higher LDL-cholesterol, lower blood hemoglobin (OR: 0.98, 0.98, and 0.98), insulin intake, presence of diabetic foot pathologies and diabetic peripheral neuropathy. We also identified novel risk factors, including high serum potassium (OR: 1.37, 1.46, and 1.55 respectively), high-serum sodium (OR: 1.02, 1.02, and 1.04 respectively). Better family income was a protective factor for DR, DME, and STDR. Alcohol consumption once a week was also identified as a protective factor for DR. CONCLUSIONS: Similar risk factors for DR, DME, and STDR were found in this study. Our data also indicates high serum sodium, high serum potassium, low blood hemoglobin, and level of family income as novel associated factors for DR, DME, and STDR, which can help with DR monitoring and management.

Clinical features of COVID-19-related optic neuritis: a retrospective study.

Objective: This retrospective study aimed to investigate the clinical features of optic neuritis associated with COVID-19 (COVID-19 ON), comparing them with neuromyelitis optica-associated optic neuritis (NMO-ON), myelin oligodendrocyte glycoprotein-associated optic neuritis (MOG-ON), and antibody-negative optic neuritis (antibody-negative ON). Methods: Data from 117 patients (145 eyes) with optic neuritis at the Shantou International Eye Center (March 2020-June 2023) were categorized into four groups based on etiology: Group 1 (neuromyelitis optica-related optic neuritis, NMO-ON), Group 2 (myelin oligodendrocyte glycoprotein optic neuritis, MOG-ON), Group 3 (antibody-negative optic neuritis, antibody-negative ON), and Group 4 (optic neuritis associated with COVID-19, COVID-19 ON). Characteristics of T2 and enhancement in orbital magnetic resonance imaging (MRI) were assessed. Best-corrected visual acuity (BCVA) was compared before treatment, at a short-term follow-up (14 days), and at the last follow-up after treatment. Results: The COVID-19-associated optic neuritis (COVID-19 ON) group exhibited 100% bilateral involvement, significantly surpassing other groups (P < 0.001). Optic disk edema was observed in 100% of COVID-19 ON cases, markedly differing from neuromyelitis optica-related optic neuritis (NMO-ON) (P = 0.023). Orbital magnetic resonance imaging (MRI) revealed distinctive long-segment lesions without intracranial involvement in T1-enhanced sequences for the COVID-19 ON group compared to the other three groups (P < 0.001). Discrepancies in optic nerve sheath involvement were noted between the COVID-19 ON group and both NMO-ON and antibody-negative optic neuritis (antibody-negative ON) groups (P = 0.028). Before treatment, no significant difference in best-corrected visual acuity (BCVA) existed between the COVID-19 ON group and other groups. At the 14-day follow-up, BCVA in the COVID-19 ON group outperformed the NMO-ON (P < 0.001) and antibody-negative ON (P = 0.028) groups, with no significant difference observed compared to the myelin oligodendrocyte glycoprotein optic neuritis (MOG-ON) group. At the last follow-up after treatment, BCVA in the COVID-19 ON group significantly differed from the NMO-ON group (P < 0.001). Conclusion: Optic neuritis associated with COVID-19 (COVID-19 ON) predominantly presents with bilateral onset and optic disk edema. Orbital magnetic resonance imaging (MRI) demonstrates that COVID-19 ON presents as long-segment enhancement without the involvement of the intracranial segment of the optic nerve in T1-enhanced images. Glucocorticoid therapy showed positive outcomes.

Visual outcome of various dose of glucocorticoids treatment in nonarteritic anterior ischemic optic neuropathy- a retrospective analysis.

BACKGROUND AND PURPOSE: The objective of this investigation was to assess the therapeutic efficacy of distinct glucocorticoid therapy dosages in the management of acute nonarteritic anterior ischemic optic neuropathy (NAION). MATERIALS AND METHODS: This retrospective, unmasked, and non-randomized study included a total of 85 patients. The patients were categorized into four groups: Group 1 (control) consisted of 15 patients who did not receive glucocorticoids, Group 2 included 16 patients administered with oral prednisone at a dosage of 1 mg/kg/d for 14 days, Group 3 comprised 30 patients who received 250 units of methylprednisolone once daily for 3 days, followed by oral prednisone at a dosage of 1 mg/kg/d for 11 days, and Group 4 encompassed 24 patients who received 500 units of methylprednisolone once daily for 3 days, followed by oral prednisone at a dosage of 1 mg/kg/d for 11 days. The best-corrected visual acuity (BCVA) was assessed at baseline and the final follow-up (> 7 days post-treatment). The changes in visual acuity between baseline and the 7-14 day follow-up, as well as between baseline and the concluding appraisal, were employed as metrics for assessing the extent of visual enhancement. RESULTS: No significant differences were noted in the final visual outcomes or in the changes between final visual acuity and baseline across the four groups. In Group 1 (control), the best-corrected visual acuity (BCVA) remained unchanged during final follow-ups compared to baseline. Conversely, the intervention groups exhibited statistically significant enhancements in BCVA during final follow-up (p = 0.012, p = 0.03, and p = 0.009 for Group 2, Group 3, and Group 4, respectively) when compared to baseline. During the 7-14 day follow-up, there was a significant difference in the changes between baseline BCVA and follow-up BCVA across the groups (p = 0.035). Go a step further by Bonferroni correction for multiple comparisons, group 4 showed a greater change in vision compared with group1 (p = 0.045). CONCLUSION: Our study on acute nonarteritic anterior ischemic optic neuropathy (NAION) showed no significant final visual outcome differences. Nevertheless, Groups 2, 3, and 4 demonstrated improved best-corrected visual acuity (BCVA) during the final follow-up. Notably, a 500-unit dose of methylprednisolone resulted in short-term BCVA enhancement. This suggests potential consideration of 500 units of methylprednisolone for short-term NAION vision improvement, despite its limited long-term impact.

The Role of Gut Microbiota in Neuromyelitis Optica Spectrum Disorder.

Neuromyelitis optica spectrum disorder (NMOSD) is a rare, disabling inflammatory disease of the central nervous system (CNS). Aquaporin-4 (AQP4)-specific T cells play a key role in the pathogenesis of NMOSD. In addition to immune factors, T cells recognizing the AQP4 epitope showed cross-reactivity with homologous peptide sequences in C. perfringens proteins, suggesting that the gut microbiota plays an integral role in the pathogenicity of NMOSD. In this review, we summarize research on the involvement of the gut microbiota in the pathophysiology of NMOSD and its possible pathogenic mechanisms. Among them, Clostridium perfringens and Streptococcus have been confirmed to play a role by multiple studies. Based on this evidence, metabolites produced by gut microbes, such as short-chain fatty acids (SCFAs), tryptophan (Trp), and bile acid (BA) metabolites, have also been found to affect immune cell metabolism. Therefore, the role of the gut microbiota in the pathophysiology of NMOSD is very important. Alterations in the composition of the gut microbiota can lead to pathological changes and alter the formation of microbiota-derived components and metabolites. It can serve as a biomarker for disease onset and progression and as a potential disease-modifying therapy.

Casein kinase-2 inhibition promotes retinal ganglion cell survival after acute intraocular pressure elevation.

Intraocular pressure elevation can induce retinal ganglion cell death and is a clinically reversible risk factor for glaucoma, the leading cause of irreversible blindness. We previously demonstrated that casein kinase-2 inhibition can promote retinal ganglion cell survival and axonal regeneration in rats after optic nerve injury. To investigate the underlying mechanism, in the current study we increased the intraocular pressure of adult rats to 75 mmHg for 2 hours and then administered a casein kinase-2 inhibitor (4,5,6,7-tetrabromo-2-azabenzimidazole or 2-dimethylamino-4,5,6,7-tetrabromo-1H-benzimidazole) by intravitreal injection. We found that intravitreal injection of 4,5,6,7-tetrabromo-2-azabenzimidazole or 2-dimethylamino-4,5,6,7-tetrabromo-1H-benzimidazole promoted retinal ganglion cell survival and reduced the number of infiltrating macrophages. Transcriptomic analysis showed that the mitogen activated protein kinase signaling pathway was involved in the response to intraocular pressure elevation but was not modulated by the casein kinase-2 inhibitors. Furthermore, casein kinase-2 inhibition downregulated the expression of genes (Cck, Htrsa, Nef1, Htrlb, Prph, Chat, Slc18a3, Slc5a7, Scn1b, Crybb2, Tsga10ip, and Vstm21) involved in intraocular pressure elevation. Our data indicate that inhibition of casein kinase-2 can enhance retinal ganglion cell survival in rats after acute intraocular pressure elevation via macrophage inactivation.

Allergy Test Dosage of Fluorescein Detects Diabetic Retinopathy Changes in Fundus Fluorescein Angiography.

BACKGROUND: The aims of this study were to evaluate the feasibility of allergy test dosage of fluorescein sodium (1%) for Diabetic Retinopathy (DR) detection in Fundus Fluorescein Angiography (FFA) examination as compared to the regular dosage (20%). METHODS: Totally 77 eyes from 42 DR patients were included in this prospective study. Capillary non-perfusion area, neovascularization, diabetic macular edema and microaneurysms were measured by FFA and compared at 1, 5 and 15 min after intravenous injection of 1% or 20% fluorescein sodium. RESULTS: There was no statistically significant difference in the proportions of capillary non-perfusion area and diabetic macular edema as well as the amount of neovascularization between the 1% and 20% fluorescein sodium groups. Yet, the 1% group had a significantly a smaller number of microaneurysms than the 20% group at 1 min (p < 0.001) and a smaller number of eyes with diabetic macular edema than the 20% group at 5 (p = 0.032) and 15 min (p = 0.015). The images from patients with clear vitreous had better quality than the images from patients with vitreous opacity (all p < 0.05, except comparison on neovascularization at 5 min: p > 0.999). All examined indexes showed high correlations between the 1% and 20% groups (r > 0.8, p < 0.001). CONCLUSIONS: This study demonstrated that 1% fluorescein sodium could detect the changes of DR comparably to the regular dosage.

Artificial Intelligence-based database for prediction of protein structure and their alterations in ocular diseases.

The aim of the study is to establish an online database for predicting protein structures altered in ocular diseases by Alphafold2 and RoseTTAFold algorithms. Totally, 726 genes of multiple ocular diseases were collected for protein structure prediction. Both Alphafold2 and RoseTTAFold algorithms were built locally using the open-source codebases. A dataset with 48 protein structures from Protein Data Bank (PDB) was adopted for algorithm set-up validation. A website was built to match ocular genes with the corresponding predicted tertiary protein structures for each amino acid sequence. The predicted local distance difference test-Cα (pLDDT) and template modeling (TM) scores of the validation protein structure and the selected ocular genes were evaluated. Molecular dynamics and molecular docking simulations were performed to demonstrate the applications of the predicted structures. For the validation dataset, 70.8% of the predicted protein structures showed pLDDT greater than 90. Compared to the PDB structures, 100% of the AlphaFold2-predicted structures and 97.9% of the RoseTTAFold-predicted structure showed TM score greater than 0.5. Totally, 1329 amino acid sequences of 430 ocular disease-related genes have been predicted, of which 75.9% showed pLDDT greater than 70 for the wildtype sequences and 76.1% for the variant sequences. Small molecule docking and molecular dynamics simulations revealed that the predicted protein structures with higher confidence scores showed similar molecular characteristics with the structures from PDB. We have developed an ocular protein structure database (EyeProdb) for ocular disease, which is released for the public and will facilitate the biological investigations and structure-based drug development for ocular diseases. Database URL:  http://eyeprodb.jsiec.org.

Bilateral retinal nerve fiber layer thickness reduction in a 9-year-old myopic boy suffering from unilateral optic neuritis: A case report.

BACKGROUND: In this paper, we present a 9-year-old boy who demonstrates a complex interplay between myopia progression, axial length (AL) extension, and retinal nerve fiber layer (RNFL) thickness loss in both eyes. Additionally, concurrent optic neuritis has directly impacted RNFL thickness in his right eye, and its potential indirect influence on RNFL and macular ganglion cell layer (mGCL) thickness in his left eye is also noteworthy. CASE SUMMARY: A 9-year-old boy with bilateral myopia presented with diminished vision and pain in his right eye due to optic neuritis, while his left eye showed pseudopapilledema. Steroid therapy improved his vision in the right eye, and 16-mo follow-up revealed recovery without recurrence despite myopia progression. Follow-up optical coherence tomography conducted 16 mo later revealed a notable thinning of the RNFL in both eyes, especially along with a reduction in mGCL thickness in the left eye. This intricate interaction between optic neuritis, myopia, and retinal changes underscores the need for comprehensive management, highlighting potential long-term visual implications in young patients. CONCLUSION: The progression of myopia and AL extension led to the loss of RNFL thickness in both eyes in a 9-year-old boy. Concurrently, optic neuritis directly affected RNFL thickness in his right eye and may indirectly play a role in the thickness of RNFL and mGCL in his left eye.

Optic nerve diseases and regeneration: How far are we from the promised land?

The retinal ganglion cells (RGCs) are the sole output neurons that connect information from the retina to the brain. Optic neuropathies such as glaucoma, trauma, inflammation, ischemia and hereditary optic neuropathy can cause RGC loss and axon damage, and lead to partial or total loss of vision, which is an irreversible process in mammals. The accurate diagnoses of optic neuropathies are crucial for timely treatments to prevent irrevocable RGCs loss. After severe ON damage in optic neuropathies, promoting RGC axon regeneration is vital for restoring vision. Clearance of neuronal debris, decreased intrinsic growth capacity, and the presence of inhibitory factors have been shown to contribute to the failure of post-traumatic CNS regeneration. Here, we review the current understanding of manifestations and treatments of various common optic neuropathies. We also summarise the current known mechanisms of RGC survival and axon regeneration in mammals, including specific intrinsic signalling pathways, key transcription factors, reprogramming genes, inflammation-related regeneration factors, stem cell therapy, and combination therapies. Significant differences in RGC subtypes in survival and regenerative capacity after injury have also been found. Finally, we highlight the developmental states and non-mammalian species that are capable of regenerating RGC axons after injury, and cellular state reprogramming for neural repair.

Retinal transcriptome of neonatal mice after optic nerve injury.

BACKGROUND: The axonal growth capacity of retinal ganglion cells decreases dramatically within the first day of birth, and the axonal regeneration after injury in mature mammals is very limited. Here, this study aimed to delineate the transcriptomic changes associated with altered axonal growth capacity and to identify the key genes associated with axonal regeneration by the RNA sequencing (RNA-Seq) analysis. METHODS: The whole retinas from the mice of embryonic day (E) 20, postnatal day (P) 1 and P3 were collected at 6 hours after optic nerve crush (ONC). Differentially expressed genes (DEGs) for ONC or ages were identified by the RNA-Seq analysis. K-means analysis was conducted for the clustering of DEGs based on expression patterns. Enrichment of functions and signaling pathways analysis were performed based on Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) database, and Gene Set Enrichment analysis (GSEA). Quantitative real time polymerase chain reaction (qRT-PCR) was used to validate the DEGs selected from the RNA-Seq analysis. RESULTS: In total, 5,408 DEGs were identified for ages, and 2,639 DEGs in neonatal mouse retina after ONC. K-means analysis revealed 7 clusters in age-DEGs and 11 clusters in ONC-DEGs. The GO, KEGG and GSEA pathway analyses identified significantly enrichment of DEGs in the visual perception and phototransduction for the age effect, and the break repair, neuron projection guidance, and immune system pathway for the ONC. PPI analysis identified hub genes in the axon-related gene cluster. The expressions of Mlc1, Zfp296, Atoh7, Ecel1, Creb5, Fosb, and Lcn2, thought to be involved in RGC death and axonal growth were validated by qRT-PCR. CONCLUSIONS: This study, for the first time, delineated the gene expression changes following ON injury in embryonic and neonatal mice, providing a new resource of age- and injury-driven data on axonal growth capacity.

Acute phase clinical manifestations of patients with Vogt-Koyanagi-Harada disease in Southern China.

BACKGROUNDS: To characterize the acute phase clinical manifestations and visual outcomes of the patients with Vogt-Koyanagi Harada (VKH) disease in southern China. METHODS: In total, 186 patients with acute-onset VKH disease were recruited. The demographic data, clinical signs, ophthalmic examinations, and visual outcomes were analyzed. RESULTS: Among the 186 VKH patients, 3 were diagnosed as complete VKH, 125 as incomplete VKH, and 58 as probable VKH. All patients visited the hospital within 3 months of onset and complained of decreased vision. For the extraocular manifestations, 121 patients (65%) referred neurological symptoms. Anterior chamber activity was negative in most eyes within an onset of 7 days, which increased slightly with onset beyond 1 week. Exudative retinal detachment (366 eyes, 98%) and optic disc hyperaemia (314 eyes, 84%) were commonly observed at presentation. A typical ancillary examination helped with the diagnosis of VKH. Systemic corticosteroid therapy was prescribed. The logMAR best-corrected visual acuity improved significantly from 0.74 ± 0.54 at baseline to 0.12 ± 0.24 at the 1-year follow-up visit. The recurrence rate was 18% in the follow-up visits. Erythrocyte sedimentation rate and C-reactive protein were significantly correlated to VKH recurrences. CONCLUSION: Posterior uveitis, followed by mild anterior uveitis, is the typical initial manifestation in the acute phase of Chinese VKH patients. Visual outcome improvement is promising in most patients receiving systemic corticosteroid therapy in the acute phase. Detection of the clinical features at the initial onset of VKH could facilitate early treatment and better vision improvement.

Multi-Polymorphism Analysis Reveals Joint Effects in Males With Chronic Central Serous Chorioretinopathy.

Purpose: Central serous chorioretinopathy (CSCR) is a leading cause of central vision impairment in the working-age population with male predilection. Knowledge about the genetic basis of CSCR and its male predilection remained limited. This study aimed to evaluate the association patterns of multiple gene variants in chronic CSCR (cCSCR) in Chinese patients. Methods: This case-control genetic association study included 531 patients with cCSCR and 2383 controls from two independent Chinese cohorts. Nine single-nucleotide polymorphisms (SNPs) of six genes, namely CFH, NR3C2, GATA5, VIPR2, TNFRSF10A, and ARMS2, were genotyped in all subjects. The main outcome measures were the association of individual single-nucleotide polymorphism (SNP) with cCSCR, the sex-stratification effects of individual SNP, and joint effects of different SNPs on cCSCR. Results: Association results in the two cohorts were consistent with low heterogeneities. In the combined analysis, SNPs CFH rs800292 (odds ratio [OR] = 1.25, P = 0.0020), CFH rs1329428 (OR = 1.23, P = 0.0037), and TNFRSF10A rs13278062 (OR = 1.43, P = 0.0014) were significantly associated with cCSCR. In stratification analysis by sex, 3 SNPs in CFH, rs3753394, rs800292, and rs1329428, were associated with cCSCR in male patients, but not in female patients. Joint analysis revealed that subjects homozygous for the risk alleles of CFH rs800292 and TNFRSF10A rs13278062 had over 4-fold of increased risk of cCSCR when compared with subjects homozygous for the non-risk alleles (OR = 4.06, P = 2.30 × 10-5). Conclusions: This study revealed main and joint effects of SNPs in CFH and TNFRSF10A on cCSCR, and suggested CFH as a potential genetic factor underlying the male predilection of cCSCR. Further replication in other study populations is needed.

Automatic Detection of Peripheral Retinal Lesions From Ultrawide-Field Fundus Images Using Deep Learning.

PURPOSE: To establish a multilabel-based deep learning (DL) algorithm for automatic detection and categorization of clinically significant peripheral retinal lesions using ultrawide-field fundus images. METHODS: A total of 5958 ultrawide-field fundus images from 3740 patients were randomly split into a training set, validation set, and test set. A multilabel classifier was developed to detect rhegmatogenous retinal detachment, cystic retinal tuft, lattice degeneration, and retinal breaks. Referral decision was automatically generated based on the results of each disease class. t -distributed stochastic neighbor embedding heatmaps were used to visualize the features extracted by the neural networks. Gradient-weighted class activation mapping and guided backpropagation heatmaps were generated to investigate the image locations for decision-making by the DL models. The performance of the classifier(s) was evaluated by sensitivity, specificity, accuracy, F 1 score, area under receiver operating characteristic curve (AUROC) with 95% CI, and area under the precision-recall curve. RESULTS: In the test set, all categories achieved a sensitivity of 0.836-0.918, a specificity of 0.858-0.989, an accuracy of 0.854-0.977, an F 1 score of 0.400-0.931, an AUROC of 0.9205-0.9882, and an area under the precision-recall curve of 0.6723-0.9745. The referral decisions achieved an AUROC of 0.9758 (95% CI= 0.9648-0.9869). The multilabel classifier had significantly better performance in cystic retinal tuft detection than the binary classifier (AUROC= 0.9781 vs 0.6112, P < 0.001). The model showed comparable performance with human experts. CONCLUSIONS: This new DL model of a multilabel classifier is capable of automatic, accurate, and early detection of clinically significant peripheral retinal lesions with various sample sizes. It can be applied in peripheral retinal screening in clinics.

Evaluation of Myocilin Variant Protein Structures Modeled by AlphaFold2.

Deep neural network-based programs can be applied to protein structure modeling by inputting amino acid sequences. Here, we aimed to evaluate the AlphaFold2-modeled myocilin wild-type and variant protein structures and compare to the experimentally determined protein structures. Molecular dynamic and ligand binding properties of the experimentally determined and AlphaFold2-modeled protein structures were also analyzed. AlphaFold2-modeled myocilin variant protein structures showed high similarities in overall structure to the experimentally determined mutant protein structures, but the orientations and geometries of amino acid side chains were slightly different. The olfactomedin-like domain of the modeled missense variant protein structures showed fewer folding changes than the nonsense variant when compared to the predicted wild-type protein structure. Differences were also observed in molecular dynamics and ligand binding sites between the AlphaFold2-modeled and experimentally determined structures as well as between the wild-type and variant structures. In summary, the folding of the AlphaFold2-modeled MYOC variant protein structures could be similar to that determined by the experiments but with differences in amino acid side chain orientations and geometries. Careful comparisons with experimentally determined structures are needed before the applications of the in silico modeled variant protein structures.

Diversified Treatment Options of Adult Stem Cells for Optic Neuropathies.

Optic neuropathies refer to a group of ocular disorders with abnormalities or dysfunction of the optic nerve, sharing a common pathophysiology of retinal ganglion cell (RGC) death and axonal loss. RGCs, as the retinal neurons in the central nervous system, show limited capacity in regeneration or recovery upon diseases or after injuries. Critically, there is still no effective clinical treatment to cure most types of optic neuropathies. Recently, stem cell therapy was proposed as a potential treatment strategy for optic neuropathies. Adult stem cells, including mesenchymal stem cells and hematopoietic stem cells, have been applied in clinical trials based on their neuroprotective properties. In this article, the applications of adult stem cells on different types of optic neuropathies and the related mechanisms will be reviewed. Research updates on the strategies to enhance the neuroprotective effects of human adult stem cells will be summarized. This review article aims to enlighten the research scientists on the diversified functions of adult stem cells and consideration of adult stem cells as a potential treatment for optic neuropathies in future clinical practices.

Automated multidimensional deep learning platform for referable diabetic retinopathy detection: a multicentre, retrospective study.

OBJECTIVE: To develop and validate a real-world screening, guideline-based deep learning (DL) system for referable diabetic retinopathy (DR) detection. DESIGN: This is a multicentre platform development study based on retrospective, cross-sectional data sets. Images were labelled by two-level certificated graders as the ground truth. According to the UK DR screening guideline, a DL model based on colour retinal images with five-dimensional classifiers, namely image quality, retinopathy, maculopathy gradability, maculopathy and photocoagulation, was developed. Referable decisions were generated by integrating the output of all classifiers and reported at the image, eye and patient level. The performance of the DL was compared with DR experts. SETTING: DR screening programmes from three hospitals and the Lifeline Express Diabetic Retinopathy Screening Program in China. PARTICIPANTS: 83 465 images of 39 836 eyes from 21 716 patients were annotated, of which 53 211 images were used as the development set and 30 254 images were used as the external validation set, split based on centre and period. MAIN OUTCOMES: Accuracy, F1 score, sensitivity, specificity, area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), Cohen's unweighted κ and Gwet's AC1 were calculated to evaluate the performance of the DL algorithm. RESULTS: In the external validation set, the five classifiers achieved an accuracy of 0.915-0.980, F1 score of 0.682-0.966, sensitivity of 0.917-0.978, specificity of 0.907-0.981, AUROC of 0.9639-0.9944 and AUPRC of 0.7504-0.9949. Referable DR at three levels was detected with an accuracy of 0.918-0.967, F1 score of 0.822-0.918, sensitivity of 0.970-0.971, specificity of 0.905-0.967, AUROC of 0.9848-0.9931 and AUPRC of 0.9527-0.9760. With reference to the ground truth, the DL system showed comparable performance (Cohen's κ: 0.86-0.93; Gwet's AC1: 0.89-0.94) with three DR experts (Cohen's κ: 0.89-0.96; Gwet's AC1: 0.91-0.97) in detecting referable lesions. CONCLUSIONS: The automatic DL system for detection of referable DR based on the UK guideline could achieve high accuracy in multidimensional classifications. It is suitable for large-scale, real-world DR screening.

Growth hormone-releasing hormone receptor signaling in experimental ocular inflammation and neuroprotection.

Both inflammation and anti-inflammation are involved in the protection of retinal cells. Antagonists of the hypothalamic growth hormone-releasing hormone receptor (GHRHR) have been shown to possess potent anti-inflammatory properties in experimental disease models of various organs, some with systemic complications. Such effects are also found in ocular inflammatory and neurologic injury studies. In experimental models of mice and rats, both growth hormone-releasing hormone receptor agonists and antagonists may alleviate death of ocular neural cells under certain experimental conditions. This review explores the properties of growth hormone-releasing hormone receptor agonists and antagonists that lead to its protection against inflammatory responses induced by extrinsic agents or neurologic injures in ocular animal models.

Monocyte-derived SDF1 supports optic nerve regeneration and alters retinal ganglion cells' response to Pten deletion.

Although mammalian retinal ganglion cells (RGCs) normally cannot regenerate axons nor survive after optic nerve injury, this failure is partially reversed by inducing sterile inflammation in the eye. Infiltrative myeloid cells express the axogenic protein oncomodulin (Ocm) but additional, as-yet-unidentified, factors are also required. We show here that infiltrative macrophages express stromal cell­derived factor 1 (SDF1, CXCL12), which plays a central role in this regard. Among many growth factors tested in culture, only SDF1 enhances Ocm activity, an effect mediated through intracellular cyclic AMP (cAMP) elevation and phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K) activation. SDF1 deficiency in myeloid cells (CXCL12flx/flxLysM-Cre−/+ mice) or deletion of the SDF1 receptor CXCR4 in RGCs (intraocular AAV2-Cre in CXCR4flx/flx mice) or SDF1 antagonist AMD3100 greatly suppresses inflammation-induced regeneration and decreases RGC survival to baseline levels. Conversely, SDF1 induces optic nerve regeneration and RGC survival, and, when combined with Ocm/cAMP, SDF1 increases axon regeneration to levels similar to those induced by intraocular inflammation. In contrast to deletion of phosphatase and tensin homolog (Pten), which promotes regeneration selectively from αRGCs, SDF1 promotes regeneration from non-αRGCs and enables the latter cells to respond robustly to Pten deletion; however, SDF1 surprisingly diminishes the response of αRGCs to Pten deletion. When combined with inflammation and Pten deletion, SDF1 enables many RGCs to regenerate axons the entire length of the optic nerve. Thus, SDF1 complements the effects of Ocm in mediating inflammation-induced regeneration and enables different RGC subtypes to respond to Pten deletion.

Automatic detection of 39 fundus diseases and conditions in retinal photographs using deep neural networks.

Retinal fundus diseases can lead to irreversible visual impairment without timely diagnoses and appropriate treatments. Single disease-based deep learning algorithms had been developed for the detection of diabetic retinopathy, age-related macular degeneration, and glaucoma. Here, we developed a deep learning platform (DLP) capable of detecting multiple common referable fundus diseases and conditions (39 classes) by using 249,620 fundus images marked with 275,543 labels from heterogenous sources. Our DLP achieved a frequency-weighted average F1 score of 0.923, sensitivity of 0.978, specificity of 0.996 and area under the receiver operating characteristic curve (AUC) of 0.9984 for multi-label classification in the primary test dataset and reached the average level of retina specialists. External multihospital test, public data test and tele-reading application also showed high efficiency for multiple retinal diseases and conditions detection. These results indicate that our DLP can be applied for retinal fundus disease triage, especially in remote areas around the world.