Hard Exudates in Diabetic Macular Edema: Automated OCT Quantification and Longitudinal Response to Faricimab

To develop and validate an automated OCT-based pipeline for hard exudates (HEs) quantification and to characterize their longitudinal evolution in eyes with diabetic macular edema (DME) treated with faricimab.

Retrospective, single-center observational study.

Treatment-naïve and previously treated DME eyes initiating intravitreal faricimab with ≥2 visits and a minimum follow-up of 3 months.

All patients underwent best-recorded visual acuity (VA) assessment and serial OCT imaging (Spectralis HRA system, Heidelberg Engineering). A custom ImageJ-based pipeline was used to quantify HE across volumetric datasets. Extracted metrics included total lesion count, square-root-transformed area, lesion density, and spatial distribution. Faricimab was administered using a treat-and-extend regimen following a 3-monthly injection loading phase.

Agreement with manual grading assessed using intraclass correlation coefficients and Bland-Altman analysis; longitudinal changes in VA, central subfield thickness (CST), and HE metrics modeled using linear mixed-effects models with evaluation of nonlinear trajectories via natural cubic splines and interaction effects with clinical and treatment-related predictors.

Eighty-three eyes from 65 patients (mean age 64 ± 10 years) were included, of which 72 (87%) presented with HE at baseline. Automated quantification showed good to excellent agreement with manual grading (intraclass correlation coefficient 0.85-0.92). Greater baseline HE area was associated with treatment-naïve status (β = 56.5, P = 0.02), higher CST (β = 24.3 per 100 μm, P = 0.04), and the presence of subretinal fluid (β = 71.8, P = 0.02). Over a median follow-up of 13.5 months, all HE metrics decreased significantly (P < 0.001), following nonlinear trajectories characterized by an initial plateau and subsequent accelerated decline. Higher faricimab injection frequency during the first year was associated with faster HE resolution (β = -1.06 per injection per month, P = 0.02). Hard exudate burden remained positively associated with CST over time (β = 0.34 μm per unit √HE area, P < 0.001) and with worse VA at any given time point (β = 2.34 × 10^-4 logarithm of the minimum angle of resolution per unit √HE area, P = 0.048), without significantly influencing the rate of functional recovery.

Automated OCT-based quantification enables reproducible assessment of HE burden and its longitudinal dynamics in DME. Faricimab treatment was associated with a significant, nonlinear reduction in HE, accompanied by improvements in CST and VA over follow-up.

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.