Phenotype-driven management of diabetic macular edema: multimodal imaging biomarkers and individualized therapy.

Diabetic macular edema (DME) remains a leading cause of vision impairment worldwide. Although intravitreal anti-vascular endothelial growth factor (anti-VEGF) agents are widely regarded as first-line treatment, a substantial proportion of patients demonstrate suboptimal responses, ranging from anatomical non-resolution to functional plateau, reflecting the marked pathophysiological heterogeneity of the disease. Multimodal retinal imaging, particularly optical coherence tomography (OCT), OCT angiography (OCTA), and ultra-widefield fluorescein angiography (UWF-FA)- has enabled detailed characterization of retinal structural and microvascular alterations and facilitated the identification of clinically relevant imaging biomarkers. This review synthesizes established and emerging imaging biomarkers and groups them into two major domains. First, markers of disease activity, represented by intraretinal and subretinal fluid and hyperreflective foci. Second, predictors of visual prognosis, including disorganization of the retinal inner layers, photoreceptor damage, and retinal perfusion deficits. Crucially, based on characteristic biomarker profiles, we propose stratifying DME into five principal clinical phenotypes: the leakage-dominant, inflammatory, tractional, focal-treatment, and poor-prognosis phenotypes. Integrating these phenotypes into proposed, hypothesis-generating decision-support framework-encompassing baseline assessment, phenotype-based stratification, and dynamic optimization, aims to align therapeutic strategies to be more precisely aligned with underlying pathogenic mechanisms. Future developments, including automated biomarker quantification and artificial intelligence-assisted image analysis, may further enhance precision in DME phenotyping and support a definitive shift towards truly individualized disease management.