Abstract
Purpose :
To investigate relationships between hyperreflective foci, subretinal hyperreflective material/fibrosis and thickness of retinal layers and fluids on OCT in patients with intermediate and advanced dry and exudative AMD.
Methods :
57 Dry and 20 Exudative AMD (eAMD) OCT volumes (3303 and 1184 B-scans, respectively) from the Medical Retina Department at Jules-Gonin Eye Hospital and 156 Dry and 157 Exudative OCT volumes (4711 and 4758 B-scans) from Noor Eye Hospital (NEH) 2021 public dataset were used. Segmentation was done on the B-scan level with two convolutional neural network models (U-Net architecture, EfficientNet-B5) trained for retinal morphologies segmentation.
Average thickness measurements were calculated per B-scans for: ONL(outer nuclear layer), MZ (Myoid zone), EZ+OPR+IZ (PR - photoreceptors), RPE (retinal pigment epithelium), IRF (intraretinal fluid), SRF (subretinal fluid), PED (pigment epithelial detachments/drusens). To quantify the amount of HRF (hyperreflective foci) and SHRM/FIB (subretinal hyper-reflective material/fibrosis) a B-scan lesion area was calculated.
Relationships between measurements were calculated using Spearman correlation, presented results have P-values < 0.001.
Results :
SHRM/FIB was strongly negatively correlated with RPE, PR, and MZ thickness in both datasets for Exudative and Dry AMD (Exudative: -0.64, -0.68, -0.66: NEH dataset, -0.76, -0.78, -0.78: JG; Dry: -0.64, -0.62, -0.67: JG), positively correlated with SRF (0.41 - JG, 0.53 - NEH), and IRF (0.29 JG, 0.59 - NEH).
HRF are positively correlated (strong and weak correlations) in eAMD with SRF (0.71 - NEH, 0.36 JG), IRF (moderate) (0.46 - NEH) and PED (moderate and strong) (0.57 - JG and 0.65 NEH) and negatively correlated with PR and MZ thickness (-0.53, -0.46 - JG; -0.64, -0.59 - NEH).
For Dry AMD HRF are positively correlated (weak) with PED/Drusen (0.31 - JG, 0.32 - NEH) and negatively correlated (weak) with PR (-0.26 - JG,- 0.2 - NEH).
Conclusions :
The findings of our study highlight the importance of quantification of HRF and SHRM/FIB as biomarkers for AMD progression. The correlation between SRF and HRF in eAMD may point towards a more pronounced inflammatory state of the respective eyes. For dry AMD the increased amount of HRF could be a predictive factor for pronounced GA progression. However, these hypotheses warrant further longitudinal assessment.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.