We planned a prospective study in January 2015 recruiting patients affected by macular atrophy who were consecutively registered in our database until January 2018. All patients affected by EMAP, DTGA or non-diffuse-trickling geographic atrophy (nDTGA) were enrolled in the study, with one planned examination visit every 12 months. All recruited patients underwent a complete ophthalmologic examination measuring BCVA using Early Treatment Diabetic Retinopathy Study (ETDRS) charts; slit-lamp biomicroscopy with fundus examination; IOP measurement with a Goldman tonometer; and FAF and optical coherence tomography (OCT) imaging, using Spectralis HRA + OCT (Heidelberg Engineering, Germany). We also investigated the presence of cardiovascular (CV) risk factors (including hypertension, hypercholesterolemia, stroke, heart failure, heart attack, and cardiac rhythm diseases) and history of glaucoma.
The research followed the principles of the Declaration of Helsinki and was approved by the Ethics Committee of the IRCCS Scientific Institute San Rafael Hospital. Informed consent was obtained from all recruited subjects.
The study included patients with macular atrophy secondary to EMAP, DTGA, or nDTGA. The diagnosis of EMAP was defined when the following criteria were met: (1) the presence of macular atrophy with symmetrical, polycyclic borders, with a prominent vertical axis; (2) pseudodrucen-like deposits around atrophic lesions; (3) peripheral floor-stone collapse; (4) When visual symptoms appear (either nyctalopia, low vision or metamorphopsia) under the age of 55 years.
Patients diagnosed with macular atrophy due to AMD were subclassified based on their appearance on FAF imaging as described by Holz et al.4 Briefly, patients were assigned to the DTGA group if the atrophic area met the following criteria: (1) at least 55 years of age at symptom onset; (2) increased AF signal extending beyond the edge of diffuse atrophy; (3) Coalescent, gray, lobular aspect.
AMD patients with macular atrophy side were assigned to criterion (2) or (3) nDTGA group (Figure 1).
A senior grader (MBP) confirmed the diagnosis of cases meeting the criteria for one of the three groups.
Exclusion criteria were: minimum follow-up of less than 12 months, high media opacities, insufficient fixation for high-quality imaging, systemic or other ocular conditions known to cause chorioretinal atrophy, history of neovascular complications, or administration of intravitreal anti-VEGF agents. . When both eyes of the same patient were eligible for inclusion in the study, only one eye was randomly selected.
Acquisition protocol, imaging analysis and outcome variables
Structural OCT images were obtained using a Spectralis HRA + OCT (Heidelberg Engineering, Heidelberg, Germany). The standard acquisition protocol consisted of a radial pattern of at least six B-scans with a 30° angle and a high number of frames (ART > 25) and one raster pattern of 19° angle B-scans (ART > 9). distance of 234 µm. An enhanced depth imaging (EDI) modality was selected to provide a better view of the choroid.
The area of atrophy at baseline and at the last available follow-up visit was measured on FAF images with blue light (30° × 30° field of view) focused on the fovea using a semi-automated area finder tool provided by Heidelberg HEYEX software.12,13. OCT scans and infrared images supported the identification and confirmed the presence of foveal involvement and expansion of atrophic areas as suggested by the classification of atrophy reports.14,15. The following parameters were provided by the software: total atrophic area, change from baseline examination, rate of change in mm2 per year. Quantitative analysis of circularity index was then performed using FAF images and corresponding masks of atrophic areas, exported as .bmp files from Area Finder and imported into ImageJ software (National Institutes of Health, Bethesda, MD). Image scale was set using HEYEX software to match dimensions in millimeters and pixels for each FAF image. A contour mask of the atrophic area exported by Region Finder was selected automatically using a tracing tool on the image (instead of delineating it manually) and saved as a “region of interest” (ROI). The ROI was then superimposed on the original FAF image and area, perimeter and circularity were calculated using the “measure” command (this method showed excellent inter-rater reliability in a previous study).16 We thus obtained the following variables at both baseline and final follow-up examinations for each eye included in the study:
Atrophy area (mm2)
Progression rate (mm2/y), calculated by subtracting the baseline area from the follow-up area and dividing by the overall length of observation in years.17.
Progression rate after square-root transformation of area measurements (mm/y) to reduce dependence on baseline size of atrophic lesion18.
Circularity, whose formula is 4π*area/perimeter2Identifies random shapes with values close to 019.
Choroidal variables and outcomes
Subfoveal and mean choroidal thickness (CT), Haller layer thickness (HLT) and Sattler layer-choriocapillaris complex thickness (SLCCT), and choroidal vascularity index (CVI) were measured on OCT B- at baseline and at the end of follow-up. Scans focused on the fovea according to current literature11,20,21. Large choroidal vessels defined the Haller layer, while SLCCT was defined as the distance between the lower border of the retinal pigment epithelium-Bruch complex and the upper border of the Haller layer, as previously reported.11.
All choroidal measurements were taken by two trained graders (LB and AS). Specifically, subfoveal CT was manually measured between Bruch’s membrane and the sclerochoroidal interface at the subfoveal location, and at 0.5-mm intervals, from 1.0 mm nasal to 1.0 mm temporal fovea; While mean CT, HLT and SLCCT were calculated from the mean values of five samples (subfoveal, 750 mm [right- left] and 1500 mm [right-left]).
Patients with EMAP, DTGA and nDTGA belong to different age groups by definition, and choroidal thickness decreases with age.22. We also calculated the ratio of SLCC to mean CT (i.e. Sattler/Choroid ratio [SCR]) as an additional parameter, to standardize our results.
CVI was calculated as previously described in the literature21. Specifically, after importing the scans into ImageJ software, we calculated the CVI by selecting the entire choroid as the ROI, since the CVI is not affected by the considered region.23. Bruch’s membrane and the sclerochoroidal junction represent the upper and lower limits of the area. Areas of hyper- and hypo-transmission were also included in the ROI. The entire image was then converted to 8-bit form, binarized with an autolocal threshold of black and then converted back to RGB to isolate the choroidal region. ROIs with hyper- and hypo-transmission areas were selected and deleted from the choroidal area to be analyzed, to avoid under- and over-estimation of CVI (see Figure 2). After binarization, CVI was obtained through an in-house plug-in that automatically calculates the ratio of the luminal choroidal area (black pixels) to the total choroidal area (black pixels + white pixels).
Finally, IR + OCT scans were reviewed for the presence of choroidal lipid globule cavities, defined on structural OCT as non-reflective spheres in polyhedral structures with a posterior hypertransmission tail, often hyperreflective on IR images in cases of RPE loss. has been found to be associated with RPE atrophy in AMD (Figure 3).24,25,26. These cavities were previously identified as hyperreflective dots in IR images24and later confirmed on the corresponding OCT scan.
Retinal variables and outcomes
A diffuse Bruch-RPE separation has been described in both EMAP and DTGA3,5, so we designed a surrogate quantitative parameter for this phenomenon, and called it “EZ barrier distance”. In the horizontal OCT scan, we examined the presence of Bruch-RPE separation at the edge of the atrophic lesion (Figure 4). Then, on the temporal side of the horizontal IR + OCT 30° scan of patients showing this feature, we identified a point where the ellipsoid zone (EZ) faded at the point where the Bruch-RPE complex splits. In the corresponding IR image, we measured the distance between that point and the nearest edge of the atrophic lesion (Figure 5). The temporal side was chosen because of the presence of optic disc near the nose and frequent extension of atrophy beyond 30° along the vertical axis. All measurements were taken by two graders (LB and AS) and an intraclass correlation coefficient was calculated to assess intergrader agreement.
When the Bruch-RPE separation was absent at baseline or extended beyond the total length of the horizontal OCT scan, it was not considered possible to measure the EZ barrier distance in that eye, while if this feature disappeared during follow-up, the barrier distance progressed to 0 µm.
All descriptive data were expressed as mean ± standard deviation (SD) for continuous normally distributed variables, as median (interquartile range, IQR) for non-normal, and as frequencies and percentages for categorical ones. Tests for normality of continuous variables were assessed using the Kolmogorov-Smirnov test. Comparison of frequencies between groups was performed using the chi-square test. Differences between baseline and follow-up measurements of continuous variables were assessed using a paired samples t-test. We adopted Pearson’s correlation test for baseline SCR and CVI and Spearman’s correlation test for circularity to assess the relationship with the rate of atrophy progression and BCVA loss.
For the purpose of statistical analysis, we merged all GA patterns except DTGA, meaning we compared 3 subgroups: EMAP, DTGA and non-diffuse-trickling geographic atrophy (nDTGA). Continuously distributed variables were compared between EMAP and other groups using the Tukey HSD post hoc test for ANOVA; Non-normal variables were compared using the Kruskal-Wallis test.
All tests were two-sided and set at the level of significance p< 0.05. Analyzes were performed using SPSS Statistics 23 (IBM; Armonk, NY).