Interleukins (IL-1β, IL-4, IL-6, and IL-8) in aqueous humor as potential biomarkers of diabetic retinopathy severity.
Aims: To determine the concentrations of interleukins (IL1β, IL4, IL6, IL8) in the aqueous humor and their association with diabetic retinopathy severity.
Materials and Methods: Singlecenter crosssectional study with prospective sampling of 110 adults with type 2 diabetes mellitus spanning five categories of diabetic retinopathy and 25 nondiabetic controls. Aqueous humor (0.1 mL) obtained during phacoemulsification was analyzed by ELISA. Nonparametric tests assessed group differences; optimal thresholds were derived by multiclass onevsall ROC-analysis, followed by clinically relevant binary stratification (mild/ moderate vs severe DR). Spearman rank correlations evaluated relationships with age, diabetes duration, fasting glucose, HbA1c, and Cpeptide.
Results: All interleukins differed between groups (p < 0.001). IL1β and IL6 increased with stage; IL8 peaked at severe nonproliferative disease and declined in proliferative disease; IL4 decreased. Multiclass accuracies were modest: 56%, 58%, 53%, and 44% for IL-6, IL8, IL1β, and IL4. In the binary model, IL6 achieved 72.6% accuracy (95% CI 64.3-79.9), with cutoffs of 34.4-86.2 and >86.2 pg/mL for mild/moderate and severe disease. IL8 was supportive (62.2%; severe >216.9 pg/mL). IL1β and IL4 each yielded about 55.6%. Correlations were weak to moderate: IL6 and IL8 tracked diabetes duration and glycemia, while IL4 inversely tracked Cpeptide.
Conclusions: IL6 is the most informative single local biomarker for binary severity stratification of diabetic retinopathy, with IL8 as an adjunct and IL1β/ IL4 of limited standalone value. IL6 thresholds may guide riskadapted followup and monitoring with optical coherence tomography and optical coherence tomography angiography. External validation, preanalytical standardization, and prospective outcome studies are needed.
Materials and Methods: Singlecenter crosssectional study with prospective sampling of 110 adults with type 2 diabetes mellitus spanning five categories of diabetic retinopathy and 25 nondiabetic controls. Aqueous humor (0.1 mL) obtained during phacoemulsification was analyzed by ELISA. Nonparametric tests assessed group differences; optimal thresholds were derived by multiclass onevsall ROC-analysis, followed by clinically relevant binary stratification (mild/ moderate vs severe DR). Spearman rank correlations evaluated relationships with age, diabetes duration, fasting glucose, HbA1c, and Cpeptide.
Results: All interleukins differed between groups (p < 0.001). IL1β and IL6 increased with stage; IL8 peaked at severe nonproliferative disease and declined in proliferative disease; IL4 decreased. Multiclass accuracies were modest: 56%, 58%, 53%, and 44% for IL-6, IL8, IL1β, and IL4. In the binary model, IL6 achieved 72.6% accuracy (95% CI 64.3-79.9), with cutoffs of 34.4-86.2 and >86.2 pg/mL for mild/moderate and severe disease. IL8 was supportive (62.2%; severe >216.9 pg/mL). IL1β and IL4 each yielded about 55.6%. Correlations were weak to moderate: IL6 and IL8 tracked diabetes duration and glycemia, while IL4 inversely tracked Cpeptide.
Conclusions: IL6 is the most informative single local biomarker for binary severity stratification of diabetic retinopathy, with IL8 as an adjunct and IL1β/ IL4 of limited standalone value. IL6 thresholds may guide riskadapted followup and monitoring with optical coherence tomography and optical coherence tomography angiography. External validation, preanalytical standardization, and prospective outcome studies are needed.
Authors
Petrenko Petrenko, Lytvynenko Lytvynenko, Babenko Babenko, Ziablitsev Ziablitsev
View on Pubmed