Imaging Biomarkers in DME:
What the 2025 Evidence Says

Central subfield thickness remains the workhorse metric of diabetic macular edema (DME), but it is a blunt instrument. Two eyes with identical thickness can have very different visual potential and very different biology. The maturation of optical coherence tomography — and the 2025 synthesis of the biomarker literature — lets us read structure, not just thickness, and translate it into three practical questions: what visual outcome is realistic, which drug class is most likely to help, and how should this eye be monitored?

The unifying point is efficiency: nearly all of the biomarkers below come from the same SD-OCT (and, where available, OCT-angiography) volume you already acquire. No additional test is required — only a more deliberate read.

Why Imaging Biomarkers Matter

Imaging biomarkers earn their place clinically when they do at least one of three things: predict the visual ceiling, discriminate between treatment strategies, or track response over time. The biomarkers with the strongest evidence do more than one. A 2025 systematic review and meta-analysis of baseline OCT biomarkers and visual acuity in DME (Nanji et al., Ophthalmology, 2026) pooled dozens of studies and quantified these associations, and it anchors much of the discussion here.

Photoreceptor Integrity: EZ/ELM and the COST Line

The integrity of the outer retinal bands is the single most consistent structural predictor of vision in DME. Disruption of the ellipsoid zone (EZ) and external limiting membrane (ELM) carries one of the largest baseline effect sizes on visual acuity, with pooled estimates on the order of several ETDRS letters of difference sustained out to 24 months. The cone outer segment tips (COST) line reflects still deeper photoreceptor compromise and behaves as a strong long-term predictor.

Practically, these bands sit on a severity gradient — an absent COST line, then EZ disruption, then ELM loss mark progressively deeper damage and a lower visual ceiling. Restoration of the EZ/ELM during treatment tracks with visual recovery, which is why photoreceptor status is the first thing to read when counselling a patient on realistic expectations.

DRIL: A Dynamic Correlate of Vision

Disorganisation of the retinal inner layers (DRIL) — the loss of distinct boundaries between the ganglion cell, inner plexiform, inner nuclear and outer plexiform layers — was established as a predictor of visual acuity in centre-involved DME by Sun and colleagues (JAMA Ophthalmology, 2014). Its value is that it tracks vision dynamically: changes in DRIL extent correspond to changes in acuity over follow-up, roughly on the scale of a line of vision per few hundred microns of change. Its main limitation is measurement standardisation, which is improving but not yet uniform across clinics.

Hyperreflective Foci: Reading the Inflammatory Phenotype

Retinal and choroidal hyperreflective foci (HRF) are associated with poorer visual outcomes and, more usefully, they behave as a marker of inflammatory activity. The clinical pearl is longitudinal: HRF that persist after an adequate anti-VEGF course flag an inflammatory phenotype and, in an eye that is responding suboptimally, support considering a switch to or addition of an intravitreal corticosteroid — balanced, as always, against cataract progression and IOP response.

Fluid Compartments: Not All Fluid Is Equal

Where the fluid sits matters more than how much there is.

• Subretinal fluid (SRF) is generally a favourable sign in DME, associated with a better anti-VEGF response.
• Intraretinal cysts require location-aware reading: inner nuclear layer cysts tend to be anti-VEGF responsive, whereas outer nuclear layer and chronic, persistent cysts mark a more recalcitrant, potentially steroid-responsive phenotype.

This SRF-versus-persistent-cyst distinction is one of the higher-yield forks in day-to-day DME decision-making.

The Vitreoretinal Interface

The interface can dominate the picture. Vitreomacular traction (VMT) carries one of the largest negative effects on vision and should prompt consideration of vitrectomy rather than persistent pharmacotherapy. An epiretinal membrane exerts a smaller but real negative effect, while a complete posterior vitreous detachment is favourable — a macula free of tractional forces tends to respond better and more durably.

OCT-Angiography and Widefield Imaging

Where available, OCT-angiography adds perfusion context: an enlarged foveal avascular zone (FAZ) and reduced vessel density — particularly in the deep capillary plexus — correlate with vision and may precede changes in the superficial plexus. These metrics are not yet standardised for routine clinical thresholds, but they enrich the phenotype. On ultra-widefield angiography, peripheral non-perfusion and the pattern of leakage (perivascular versus generalised) help explain recalcitrant macular edema and can direct targeted peripheral laser in selected eyes.

A Biomarker-Guided Decision Framework

The biomarkers cohere into a simple, repeatable workflow — all derivable from one SD-OCT/A acquisition.

Practice Points

• EZ/ELM status sets the realistic visual ceiling — read it first and counsel accordingly.
• SRF versus persistent cysts is a practical fork for anti-VEGF versus steroid selection.
• Persistent HRF after adequate anti-VEGF flags an inflammatory phenotype worth acting on.
• VMT is a surgical signal, not a reason to keep injecting.
• Everything above comes from imaging you already perform — the gain is in the read, not new equipment.