Comparative value of retinal structural features versus systemic inflammatory and renal indicators in predicting short-term anti-VEGF response in diabetic macular edema: focus on OCT biomarkers

TL;DR

This study compares retinal scans and blood markers to predict how well diabetic patients will respond to a common eye treatment.

Contribution

The study shows OCT-based retinal features are better predictors of treatment response than systemic inflammatory markers in diabetic macular edema.

Findings

OCT features like disorganization of inner retinal layers and disruption of the ellipsoid zone strongly predict poor treatment response.

Systemic inflammatory indicators like NLR, SII, and PLR showed no significant predictive value.

A predictive model combining OCT features and eGFR achieved an AUC of 0.887, indicating strong discrimination.

Abstract

This study aimed to systemically compare the predictive value of systemic inflammatory indicators, renal function, and optical coherence tomography (OCT)-derived morphological characteristics for the short-term response to anti-vascular endothelial growth factor (anti-VEGF) therapy in patients with diabetic macular edema (DME), and to clarify their relative importance.

A single-center retrospective observational study was conducted involving 81 DME patients who completed three monthly loading doses of anti-VEGF therapy. Baseline data including systemic inflammatory indicators (neutrophil-to-lymphocyte ratio [NLR], systemic immune-inflammatory index [SII], platelet-to-lymphocyte ratio [PLR]), renal function (estimated glomerular filtration rate [eGFR]), and comprehensive OCT morphological features were collected. Patients were categorized into a poor-response group (n = 28) and a response group (n = 53) based on an improvement in best-corrected visual acuity (BCVA) of ≤5 letters at 3 months. Multivariate logistic regression analysis was employed to identify independent predictive factors. A predictive model was constructed and evaluated for its discrimination (using AUC), calibration, and clinical utility.

Univariate analysis identified eGFR, and several OCT features as significant predictors eligible for the multivariate model, whereas none of the systemic inflammatory indicators (NLR, SII, PLR) showed significant predictive value. Multivariate analysis revealed that specific OCT structural features, namely disorganization of inner retinal layers (DIRT) (OR = 0.093, 95% CI [0.021–0.412], p = 0.002) and disruption of the ellipsoid zone/external limiting membrane (EZ/ELM) (OR = 0.142, 95% CI [0.032–0.628], p = 0.010), were strong independent predictors of poor treatment response. A predictive model integrating eGFR and these two OCT features demonstrated excellent discrimination, with an area under the curve (AUC) of 0.869 in the training set and a bootstrap-validated average AUC of 0.887. At the optimal cutoff value, the model achieved a specificity of 0.893, indicating a strong capability to identify high-risk patients.

Our comparative analysis confirms that baseline retinal structural integrity assessed by OCT—specifically disorganization of the inner retinal layers and disruption of the EZ/ELM—shows stronger predictive value for short-term anti-VEGF response in DME compared to systemic inflammatory indicators. The developed prediction model demonstrates good discriminative performance, though its clinical application requires further validation in larger cohorts. These findings support the prioritization of OCT imaging in DME management.

Figures