# Development of a predictive model based on aqueous cytokines data for response to anti-VEGF therapy in diabetic macular edema

## TL;DR

Researchers developed a model using cytokine levels in eye fluid to predict which diabetic macular edema patients will respond to anti-VEGF therapy.

## Contribution

A predictive model based on baseline aqueous cytokines for anti-VEGF response in diabetic macular edema was developed and validated.

## Key findings

- A model using five cytokines (VEGF, IL-6, Ang-2, MCP-1, ICAM-1) predicted anti-VEGF response with an AUC of 0.85 in discovery and 0.84 in validation.
- Non-responders had persistently elevated Ang-2, IL-6, IL-8, MCP-1, and PIGF levels even after anti-VEGF treatment.
- Cytokines beyond VEGF are implicated in DME pathogenesis, suggesting new therapeutic targets.

## Abstract

Intravitreal anti-vascular endothelial growth factor (VEGF) agents improve visual acuity in diabetic macular edema (DME). However, resistance, non-response, or recurrence occurs in many patients. Predictive biomarkers for anti-VEGF response are lacking. We aim to identify cytokine markers predictive of anti-VEGF response and elucidate cytokines involved in poor-response DME pathogenesis.

A luminex assay was carried out to measure the concentration of cytokines in the aqueous humor. A predictive model based on baseline cytokines was constructed in a discovery set that comprised 46 responders and 20 non-responders. In addition, an analysis of baseline cytokines of 15 responders and 12 non-responders was conducted as a validation set. The performance of the nomogram was determined using the area under the receiver operating characteristic curve (AUC) and Hosmer–Lemeshow goodness of fit test.

Baseline concentrations of angiogenesis-related cytokines VEGF (P < 0.0001), placenta growth factor (PIGF) (P < 0.0001), angiopoietin-2 (Ang-2) (P < 0.0001), inflammatory factor interleukin-6 (IL-6) (P < 0.0001), IL-8 (P < 0.0001), chemokine monocyte chemoattractant protein-1 (MCP-1) (P < 0.0001), and adhesion factor intercellular adhesion molecule-1 (ICAM-1) (P < 0.001) were significantly increased compared to controls. The prediction nomogram model included five baseline cytokines: VEGF, IL-6, Ang-2, MCP-1, and ICAM-1 were constructed. The AUC for the discovery set was 0.85 (95% CI: 0.74–0.96) and for the internal validation was 0.84, indicating that the prediction model has good predictive accuracy. The Hosmer–Lemeshow goodness of fit test showed good model calibration (P = 0.295). The levels of Ang-2 (P = 0.0042), IL-6 (P < 0.0001), IL-8 (P < 0.0001), MCP-1 (P < 0.0001), and PIGF (P < 0.0001) were still significantly increased at the 6-month timepoint after multiple injections of anti-VEGF drugs for non-response group patients.

The baseline cytokine-based model helped to assess the individual probability of response to anti-VEGF therapy. There are cytokines beyond VEGF that are involved in the pathogenesis of DME, therapeutic regimens targeting these cytokines may improve the visual acuity and reduce macular edema.

The online version contains supplementary material available at 10.1186/s40662-026-00479-z.