# AI-based fluid quantification in neovascular age-related macular degeneration and diabetic macular edema treated with faricimab – a real-life study

## TL;DR

This study shows that AI-based fluid measurement is more effective than traditional methods in tracking treatment response in eye diseases treated with faricimab.

## Contribution

AI-based fluid volumetry is shown to be more sensitive than central subfield thickness in measuring retinal fluid reduction in nAMD and DME patients treated with faricimab.

## Key findings

- AI fluid volumetry showed 37% and 56% fluid reduction in nAMD and DME, respectively, after faricimab treatment.
- Fluid volume changes were 2-4 times larger than central subfield thickness changes, indicating higher sensitivity.
- Treatment-naïve nAMD patients had 49% fluid reduction, higher than 32% in switch patients.

## Abstract

To assess faricimab treatment in neovascular age-related macular degeneration (nAMD) and diabetic macular edema (DME) with artificial intelligence-based fluid quantification on spectral-domain optical coherence tomography (SD-OCT).

Retrospective patients with nAMD and DME; treatment-naïve, or on another intravitreal medication (switchers), loaded with 4 monthly intravitreal faricimab. SD-OCT (Heidelberg Engineering) at baseline and 16 weeks, then processed using the Fluid Monitor® (RetInSight). Sum of fluid volumes in the central 1 mm (pigment epithelial detachment, subretinal fluid, and intraretinal fluid) was computed (SF) and correlated with central subfield thickness (CST).

Thirty-four nAMD (25 switchers) and 21 DME (20 switchers) eyes were included. SF (nL) was 126.68 ± 17.24 and 37.84 ± 8.31 at baseline reduced to 80.78 ± 15.56 (p < 0.0001) and 15.28 ± 4.94 (p < 0.0001) for nAMD and DME, respectively. CST (µm) reduced from 405.12 ± 24.95 and 354.97 ± 15.89 to 320.33 ± 19.80 (p = 0.0001) and 302.41 ± 11.55 (p < 0.0001) in nAMD and DME, respectively. Mean intraindividual change between baseline and 16 weeks was larger using SF than with CST for nAMD (36.5% and 17.6%, respectively) and for DME (56.2% and 13.1%, respectively). A similar pattern was observed for each retinal compartment.

When loaded with faricimab, total fluid decreased by 37% in nAMD and 56% in DME. Fluid volumetry appears more sensitive for retinal fluid.

When AI-fluid volumetry is applied real-world nAMD and DME cases, then fluid volume offers a more sensitive and quantifiable measure of disease activity than CST.

What is known

• Retinal fluid exudation is a key biomarker for disease activity in neovascular age-related macular degeneration (nAMD) and diabetic macular edema (DME). Central subfield thickness (CST) is widely used as a proxy for the latter but has limitations in sensitivity and reproducibility.

• Faricimab has demonstrated efficacy in reducing retinal fluid and CST, showing potential superiority in fluid resolution compared to existing treatments..

What is new

• This study showed AI-based fluid volumetry (Fluid Monitor®) as a more sensitive and quantifiable measure of retinal fluid compared to CST. Fluid volume reductions were notably 2 and 4 times larger than CST reductions for nAMD and DME respectively, thus offering improved sensitivity for assessing treatment efficacy.

• The presented real-world data revealed a significant fluid reduction after faricimab upload, with a 36.5% reduction in nAMD and 56.2% in DME.

• Treatment-naïve nAMD patients showed greater reductions in total fluid volume (49%) compared to switch patients (32%), suggesting a more pronounced anatomical response in previously untreated eyes.