Diabetic Retinopathy Diagnosis Based on Convolutional Neural Network in the Russian Population: A Multicenter Prospective Study.

BACKGROUND: Diabetic retinopathy is the most common complication of diabetes mellitus and is one of the leading causes of vision impairment globally, which is also relevant for the Russian Federation. OBJECTIVE: To evaluate the diagnostic efficiency of a convolutional neural network trained for the detection of diabetic retinopathy and estimation of its severity in fundus images of the Russian population. METHODS: In this cross-sectional multicenter study, the training data set was obtained from an open source and relabeled by a group of independent retina specialists; the sample size was 60,000 eyes. The test sample was recruited prospectively, 1186 fundus photographs of 593 patients were collected. The reference standard was the result of independent grading of the diabetic retinopathy stage by ophthalmologists. RESULTS: Sensitivity and specificity were 95.0% (95% CI; 90.8-96.4) and 96.8% (95% CI; 95.5- 99.0), respectively; positive predictive value - 98.8% (95% CI; 97.6-99.2); negative predictive value - 87.1% (95% CI, 83.4-96.5); accuracy - 95.9% (95% CI; 93.3-97.1); Kappa score - 0.887 (95% CI; 0.839-0.946); F1score - 0.909 (95% CI; 0.870-0.957); area under the ROC-curve - 95.9% (95% CI; 93.3-97.1). There was no statistically significant difference in diagnostic accuracy between the group with isolated diabetic retinopathy and those with hypertensive retinopathy as a concomitant diagnosis. CONCLUSION: The method for diagnosing DR presented in this article has shown its high accuracy, which is consistent with the existing world analogues, however, this method should prove its clinical efficiency in large multicenter multinational controlled randomized studies, in which the reference diagnostic method would be unified and less subjective than an ophthalmologist.

Effect of CYP2D6*4, CYP2D6*10 polymorphisms on the safety of treatment with timolol maleate in patients with glaucoma.

OBJECTIVES: Timolol maleate is used for the treatment of glaucoma and metabolized by cytochrome CYP2D6 in the liver. The aim of this study was the evaluation of the influence of CYP2D6*4 and CYP2D6*10 gene polymorphisms on the safety of medications containing 0.5% of timolol maleate as glaucoma treatment in patients with primary open-angle glaucoma (POAG). METHODS: 105 patients with POAG were prescribed glaucoma medications, containing 0.5% timolol maleate. The safety of glaucoma treatment was determined by electrocardiography (ECG) (to assess heart rate (HR) and PQ interval) and blood pressure (BP) measurements. The real-time polymerase chain reaction method was used for the detection of single nucleotide polymorphisms (SNP). RESULTS: The risk of adverse drug reactions was higher in patients with the CYP2D6*4 GA genotype compared with GG: mean HR change at 1 month (2.88 ± 4.68 and 6.44 ± 5.57, p<0.001) and 6 months (5.14 ± 8.93 and 7.88 ± 5.65, p<0.001), mean PQ interval change at 1 (0.01 ± 0.031 and 0.02 ± 0.022, p=0.003) and 6 months (0.01 ± 0.032 and 0.02 ± 0.024, p=0.003). The risk of adverse drug reactions was higher in patients with the CYP2D6*10 CT genotype compared with CC: mean HR change at 1 month (2.94 ± 4.65 and 6.34 ± 5.66, p<0.001) and 6 months (5.20 ± 8.90 and 7.78 ± 5.75, p<0.001), mean PQ interval change at 1 (0.01 ± 0.032 and 0.02 ± 0.021, p=0.014) and 6 months (0.01 ± 0.033 and 0.02 ± 0.022, p=0.014). CONCLUSIONS: CYP2D6*4 and CYP2D6*10 gene polymorphisms may affect a higher risk of timolol-induced bradycardia and increased PQ interval of treatment medications containing 0.5% of timolol maleate in patients with POAG.

Tear Proteome Revealed Association of S100A Family Proteins and Mesothelin with Thrombosis in Elderly Patients with Retinal Vein Occlusion.

Tear samples collected from patients with central retinal vein occlusion (CRVO; n = 28) and healthy volunteers (n = 29) were analyzed using a proteomic label-free absolute quantitative approach. A large proportion (458 proteins with a frequency > 0.6) of tear proteomes was found to be shared between the study groups. Comparative proteomic analysis revealed 29 proteins (p < 0.05) significantly differed between CRVO patients and the control group. Among them, S100A6 (log (2) FC = 1.11, p < 0.001), S100A8 (log (2) FC = 2.45, p < 0.001), S100A9 (log2 (FC) = 2.08, p < 0.001), and mesothelin ((log2 (FC) = 0.82, p < 0.001) were the most abundantly represented upregulated proteins, and ß2-microglobulin was the most downregulated protein (log2 (FC) = −2.13, p < 0.001). The selected up- and downregulated proteins were gathered to customize a map of CRVO-related critical protein interactions with quantitative properties. The customized map (FDR < 0.01) revealed inflammation, impairment of retinal hemostasis, and immune response as the main set of processes associated with CRVO ischemic condition. The semantic analysis displayed the prevalence of core biological processes covering dysregulation of mitochondrial organization and utilization of improperly or topologically incorrect folded proteins as a consequence of oxidative stress, and escalating of the ischemic condition caused by the local retinal hemostasis dysregulation. The most significantly different proteins (S100A6, S100A8, S100A9, MSLN, and ß2-microglobulin) were applied for the ROC analysis, and their AUC varied from 0.772 to 0.952, suggesting probable association with the CRVO.