Prognostic Imaging Biomarkers in Diabetic Macular Edema Treated with Anti-VEGF: A Multicenter AI Perspective

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Diabetic macular edema (DME) management requires predictive biomarkers for treatment response.
• Antivascular endothelial growth factor (VEGF) therapy is a standard treatment for DME.
• Optical coherence tomography (OCT) provides detailed retinal imaging for DME assessment.

Purpose of the Study:

• To identify OCT biomarkers predictive of 12-month morpho-functional outcomes in DME patients treated with anti-VEGF therapy.
• To evaluate the utility of an AI-derived platform for analyzing OCT scans and predicting DME treatment response.

Main Methods:

• Retrospective analysis of OCT scans from treatment-naive DME eyes treated with anti-VEGF.
• Utilized an AI platform (Discovery OCT Biomarker Detector) to measure retinal layer thicknesses, fluid volumes (IRF, SRF), and biomarkers.
• Employed a random forest model to assess predictive factors for morphological and functional outcomes at 12 months.

Main Results:

• A higher baseline intraretinal fluid (IRF) volume was a moderate predictor of response.
• A lower outer nuclear layer (ONL) thickness after the loading phase strongly predicted a good morphological response at 12 months.
• AI analysis revealed significant reductions in retinal layers and fluid post-treatment.

Conclusions:

• AI-powered OCT biomarker analysis shows promise for predicting 1-year outcomes in DME.
• Baseline IRF and post-loading phase ONL thickness are significant predictors of structural response in DME management.
• The AI model demonstrated good performance in predicting morphological outcomes.