Relative ellipsoid zone reflectivity and its association with disease severity in age-related macular degeneration: a MACUSTAR study report

Abstract

Quantification of the relative ellipsoid zone reflectivity (rEZR) might be a structural surrogate parameter for an early disease progression in the context of age-related macular degeneration (AMD). Within the European multicenter, cross-sectional MACUSTAR study, we have devised an automatic approach to determine the mean rEZR [arbitrary units, AU] at two independent visits in SD-OCT volume scans in study participants. Linear mixed-effects models were applied to analyze the association of AMD stage and AMD associated high-risk features including presence of pigmentary abnormalities, reticular pseudodrusen (RPD), volume of the retinal-pigment-epithelial-drusenoid-complex (RPEDC) with the rEZR. Intra-class correlation coefficients (ICC) were determined for rEZR reliability analysis. Within the overall study cohort (301 participants), we could observe decreased rEZR values (coefficient estimate ± standard error) of - 8.05 ± 2.44 AU (p = 0.0011) in the intermediate and of - 22.35 ± 3.28 AU (p < 0.0001) in the late AMD group. RPD presence was significantly associated with the rEZR in iAMD eyes (- 6.49 ± 3.14 AU; p = 0.0403), while there was a good ICC of 0.846 (95% confidence interval: 0.809; 0.876) in the overall study cohort. This study showed an association of rEZR with increasing disease severity and the presence of iAMD high-risk features. Further studies are necessary to evaluate the rEZR's value as a novel biomarker for AMD and disease progression.

Figure 1

Graphical representation of linear mixed-effects model analysis with the rEZR as dependent variable and the age at baseline (color-coded in blue: younger age to older age = dark blue to light blue), as well as the topographic variation (eccentricity [°]) of the rEZR from the fovea within each SD-OCT raster scan differentiated for each study group (controls, early AMD, intermediate AMD and late-stage AMD). Note, linear mixed-effects model showed a decreased rEZR in AMD subjects compared to healthy individuals with overall lowest rEZR in the late AMD group followed by the intermediate AMD group. Further, the rEZR was found to first increase and then decrease with increasing eccentricity respecting highest rEZR values at approximately 5° perifoveally within each subgroup. In addition, the rEZR showed an association with subjects’ age at baseline indicated by lower modelled rEZR values in older AMD subjects and controls. AMD age-related macular degeneration.

Figure 2

SD-OCT image details (pixel intensities displayed: logarithmic transformed (above) and raw (below)) of representative cases for (A) healthy controls, (B) early AMD, (C) intermediate AMD and (D) late AMD and their corresponding reflectivity profiles. The yellow superimposed yellow lines indicate the region of interest (ROI) on the logarithmic (above) and the linear (below) displayed OCT scan. Note, while the logarithmic displayed OCT image is presented for better visualization of outer retinal structure, all actual rEZR analyses in this study were performed in the image raw data (linear displayed).

Figure 3

Bland–Altman plot for the overall test–retest study population (n = 286) of the differences in the mean rEZR values per patient at V1 and V3 (solid line: mean difference; dotted lines: mean ± 1.96 × standard deviation). The rEZR difference between both visits (y axis) is presented against the rEZR average for V1 and V3 (x axis). AMD age-related macular degeneration, V1 screening visit, V3 follow-up visit, rEZR relative ellipsoid zone reflectivity.

Figure 4

Quantification of SD-OCT based outer retina reflectivity in corresponding reflectivity profiles in a representative healthy case. Each hyperreflective outer retinal band (e.g., the external limiting membrane (ELM), the ellipsoid zone (EZ) or retinal pigment epithelial/Bruch’s membrane complex (RPE/BM)) specifically corresponds to a peak in the reflectivity profile (see red overlay on SD-OCT image detail). Peak detection and assessment of the peak height in the reflectivity profile allows for reflectivity quantification.