Topographic Progression of Geographic Atrophy and Visual Acuity in Nonexudative Age-Related Macular Degeneration
Article information
Abstract
Purpose
To investigate long-term topographic progression of the geographic atrophy (GA) area based on location and analyze its correlation with visual acuity in patient with GA secondary to non-neovascular age-related macular degeneration in South Korean patient cohort.
Methods
Medical records and imaging data of 58 eyes from 34 patients with GA were retrospectively reviewed using fundus autofluorescence (FAF). Regions of interest were defined as concentric ring-shaped zones with diameters of 1, 2, 3, and 4 mm centered on the fovea (zones 0–3), each subdivided into superior, temporal, inferior, and nasal sectors. The foveal center was determined using optical coherence tomography, and sectoral GA areas were measured on FAF with the semiautomated software (RegionFinder ver. 2.6.2.0). Correlations among GA area enlargement, GA growth rate, and changes in best-corrected visual acuity (BCVA) were evaluated.
Results
The mean GA area enlarged from 4.10 to 16.57 mm2 and mean BCVA decreased from 0.34 to 1.06 logMAR at 5 years of follow-up. The mean overall GA area growth rate was 1.96 mm2/yr. During yearly follow-up from baseline to 5 years, there were significant differences in GA growth rate of zone 3 inferior (p < 0.005). The GA area changes of zone 3 inferior was significantly correlated with BCVA. In the subgroup with BCVA decreased under 1.0 logMAR during follow-up, there was a higher growth rate in zone 2 nasal sector, 2 to 3 years before.
Conclusions
The overall GA growth rate was 1.96 mm2/yr in the 5-year follow-up of our cohort. Changes in the GA growth rate in the nasal perifoveal region (zone 2) may be associated with subsequent clinically meaningful visual decline.
Geographic atrophy (GA) is advanced form of age-related macular degeneration (AMD), and it is characterized by a degeneration of photoreceptors, retinal pigment epithelium (RPE) cells as well as the inner most layer of the choroid [1–3]. It causes decreases central vision over time, and have been demonstrated that a dramatic drop in visual acuity (VA) occurs once the fovea becomes involved [4–7].
Color fundus photography (CFP), fundus autofluorescence (FAF), and optical coherence tomography (OCT) imaging can be used to evaluate GA lesions. In 2001, Holz et al. [8] reported GA growth in hyperreflective regions using FAF images. Numerous recent clinical trials of GA therapies have also used FAF images to measure changes in area [9–12]. FAF has been predominant modality for assessing GA lesion and progression that allows a sharp discrimination of a lesion’s boundary.
FAF is capable of assessing GA area or growth rate to gauge disease progression. However, the resultant values do not accurately reflect VA. While the alteration in the overall GA lesion area is presently employed as a primary endpoint in clinical trials [9,10], the growth rate of the total GA area does not reliably predict the residual VA, especially in cases with foveal sparing. In various studies investigating the correlation between changes in VA and GA area, a moderate association has been observed between alterations in the total GA lesion area and VA [13]. Notably, a significant visual decline is evident when the GA area encroaches upon the fovea.
A previous study demonstrates that GA progression into the perifoveal area, especially 1 to 2 mm from fovea, had significant correlation with decreases of VA [14]. Understanding the nature of GA progression over time and the associated visual impairment is critical for patient’s care. In South Korean studies, You et al. [14] reported a mean GA growth rate of 2.44 mm2/yr, Jeong et al. [15] documented 1.14 mm2/yr, and Lee et al. [16] reported 1.47 mm2/yr. But data in longitudinal variation in VA and GA cohorts are rather limited, especially in Asian population. In this study, we evaluated long-term topographic progression of GA based on location and analyzed its correlation with VA for modeling topographic and temporal progression of FAF in Korean patients’ cohorts.
Materials and Methods
Ethics statement
This study was approved by the Institutional Review Board of Kyung Hee Medical Center (No. 2025-07-095). Informed consent was waived due to the use of deidentified data and the retrospective nature of the study. The study was conducted in accordance with the Declaration of Helsinki.
Subjects
Patients over 55 years of age with GA secondary to AMD were included in this study. GA was defined as a region of atrophy occupies an area with a diameter of over 250 μm [17]. Those who were followed up for more 3 years and had best-corrected visual acuity (BCVA) and FAF testing were included. Subject underwent BCVA testing, color fundus photo, spectral-domain OCT (SD-OCT), and FAF (FAF, 30° × 30° and 768 × 768 pixels; Heidelberg Engineering). Patients with a history of retinal surgery, choroidal neovascularization, and RPE atrophy secondary to other causes (e.g., macular dystrophy, high myopia) were excluded. Patients with cataracts or media opacities limiting image analysis were excluded to provide good images quality and image analysis. Demographic data, such as age, sex, diabetes mellitus, hypertension, phakic states, and diagnosis of fellow eye were collected.
Imaging analysis
Atrophic area in each image were measured using the RegionFinder ver. 2.6.2.0 (Heidelberg Engineering). All GA area of FAF image were independently measured and verified by two retinal specialists. The 5-year growth rate of GA and the annualized alteration in BCVA were calculated. The location of the foveal center was determined using SD-OCT. Area were manually drawn in concentric circles starting from the foveal center determined from OCT. Zone 1, 2, and 3 were divided into concentric circles with diameters of 2, 3, and 4 mm form the fovea center. Each zone was subdivided into four quadrants, and a total of 12 sectors were used for analysis (Fig. 1). GA area was defined as partial or complete disruption of RPE on the FAF images and it confirmed with OCT in border area. GA was segmented with RegionFinder’s seed-and-grow procedure and edge refinement. Foveal diameter of 1 mm (zone 0) was excluded from analysis. In a previous study, GA area of 1 to 2 mm from fovea had significant correlation with decreases of VA [14]. To reduce bias, zone 0 was excluded. We calculated the growth rate for each respective area and also obtained the growth rate for the total atrophy area.
A fundus autofluorescence images of a patient with geographic atrophy. The area of geographic atrophy was measured based on the fundus autofluorescence images. The images were divided into 12 concentric circles with diameters of 2, 3, and 4 mm form the fovea center. S = superior; T = temporal; I = inferior; N = nasal.
VA decreased subjects
To explore in greater detail the dynamics and timing of VA decline in patients with specific characteristics, a subgroup analysis was conducted focusing on those who experienced a decline to a VA of 1.0 logMAR. Subjects with a reduction to 1.0 logMAR were identified, and their data were included from the onset of vision loss up to 3 years prior. In follow-up period, there were 17 eyes which VA decrease below 1.0 logMAR. For these 17 cases, the GA growth rate 3-year before the BCVA decreased to 1.0 log-MAR was analyzed.
Statistical analysis
All statistical analyses were performed using PASW SPSS ver. 18.0 (IBM Corp.). The growth rate of GA was computed for each specific area and further subdivided on an annual basis. The Spearman correlation analysis method was employed to investigate the relationship between alterations in the GA area for each region and changes in VA. The Kruskal-Wallis test method was used to analyze the difference in growth rate for each year. Pairwise post hoc comparisons were performed with Dunn test using multiplicity-adjusted p-values. A p-value of <0.05 was considered statistically significant for all analyses.
Results
A total 58 eyes of 34 patients with GA were included in this study (Table 1). Mean age of patients was 77.95 ± 7.16 years and mean follow-up period was 4.24 ± 1.91 years. The baseline BCVA (58 eyes) was 0.34 ± 0.45 logMAR and mean size of GA area was 4.10 ± 4.60 mm2 at baseline (the area of 1-mm diameter at the center of the fovea was excluded). Forty eyes (69.0%) were multifocal GA and 34 eyes (58.6%) were bilateral GA. The mean GA growth rate for 5 follow-up years was 1.95 ± 1.22 mm2/yr. At baseline, the GA area demonstrated a progressive increase, with measured as 5.70 mm2 at 1 year, 7.56 at 2 years, 9.18 at 3 years, 13.60 at 4 years, and 16.57 at 5 years. Concurrently, BCVA values were observed to be 0.38 logMAR at 1 year, followed by 0.45 at 2 years, 0.64 at 3 years, 0.87 at 4 years, and 1.06 at 5 years, respectively (Fig. 2A, 2B).
Baseline characteristic of patients with GA (n = 58)
Changes in best-corrected visual acuity (BCVA) and geographic atrophy (GA) area 5-year follow-up in 58 eyes from 34 patients: (A) GA area (mm2) and (B) GA area (square root) (mm).
A comparative analysis of growth rates within different sectors revealed that zone 1 exhibited a uniform growth rate across all four quadrants. In contrast, in zone 2 exhibited a notably higher growth rate in the nasal quadrant. In zone 3, it was observed that the growth rate in the inferior quadrant was comparatively lower than in the other three quadrants (Fig. 3A–3C). Spearman correlation analysis revealed a relation of GA area and BCVA changes. Among the 12 sectors, inferior sector of zone 3 showed significant correlation with changes in BCVA (p = 0.010) (Table 2).
Zone-specific (superior [S], temporal [T], inferior [I], nasal [N]), geographic atrophy area growth rates (mm2/yr) and the total zone growth rate per eye over a 5-year follow-up in 58 eyes from 34 patients are shown. (A) Zone 1. (B) Zone 2. (C) Zone 3.
Correlation between geographic atrophy progression of each sectors and best-corrected visual acuity changes over time
VA decreased subjects
At the time of VA drop detection, the mean GA area was 13.29 mm2 and 10.92, 8.64, and 6.25 mm2 at 1, 2, and 3 years before the VA drop, respectively. During the same period, the mean BCVA was 1.17 logMAR at the time of VA drop detection, and 0.45, 0.31, and 0.29 logMAR at 1, 2, and 3 years before the VA drop, respectively (Fig. 4A–4I). Total GA growth rate in VA decreased subjects was 2.28 ± 1.22 mm2/yr. The zone 2 of nasal quadrants showed a significantly different annual growth rate compared to other quadrants or areas (p = 0.013) (Table 3). Dunn post hoc comparisons showed that the 1 year before growth rate differed significantly from both 2 years before (adjusted p = 0.023) and 3 years before (adjusted p = 0.034), whereas the comparison between 2 and 3 years before was not significant (adjusted p = 0.812). Approximately 2 to 3 years before VA decreased to 1.0 logMAR, the zone 2 of nasal quadrants showed a higher growth rate than other quadrants (Fig. 4).
Geographic atrophy area growth rates in the 3 years preceding the visit, at which best-corrected visual acuity first reached 1.0 log- MAR, in 17 eyes. Bars show mean growth rate (mm2/yr) by retinal quadrant within zones 1–3. One year before the 1.0 logMAR visit: (A) zone 1, (B) zone 2, and (C) zone 3. Two years before the 1.0 logMAR visit: (D) zone 1, (E) zone 2, and (F) zone 3. Three years before the 1.0 logMAR visit: (G) zone 1, (H) zone 2, and (I) zone 3. S = superior; T = temporal; I = inferior; N = nasal.
Changes in geographic atrophy area growth rate over 3 years prior to BCVA decreased to 1.0 logMAR according to retinal location
Discussion
In the South Korean cohort study, longitudinal analysis revealed an increase in GA area over a 5-year period, escalating from a baseline of 4.10 to 16.57 mm2. This growth pattern was characterized by a yearly growth rate augmenting from 1.60 to 2.60 mm2/yr, with an average growth rate of 1.96 mm2/yr across the span. Concurrently, VA deteriorated from a baseline of 0.34 to 1.06 logMAR after 5 years. Additionally, a topographic analysis of GA progression was conducted, revealing a significant correlation between the inferior zone 3 and VA, as determined by Spearman correlation analysis. A subgroup analysis of subjects who experienced a drop in VA to 1.0 logMAR indicated that at baseline, their VA was 1.17 logMAR and the GA area was 13.29 mm2. Retrospectively examining the period 3 years prior to the onset of vision drop, the VA was found to be 0.29 logMAR, and the GA area was 6.25 mm2. Furthermore, in subjects with decreasing vision, a topographic analysis of GA progression exhibited significant yearly differences, particularly in the nasal zone 2, 2 to 3 years before the observed drop in VA.
While an increasing number of studies on GA are being conducted, there are few studies that have analyzed the longitudinal data between GA growth rate and BCVA, unlike studies on neovascularization AMD. Sunness et al. [4] observed changes in GA area with CFP and reported that in median total atrophic area 2.9 disc area (2.54 mm2), 31% suffered a three-line VA loss from baseline by 2 years, and 53% had a three-line loss by 4 years. In another study, VA modeling in GA was done as a subanalysis of FAM (Fundus Autofluorescence in Age-related Macular Degeneration) study [6,8]. The mean VA was 0.6 ± 0.4 logMAR and the GA area was 6.3 ± 5.0 mm2. An estimated mean VA changes of 0.181 and 0.256 logMAR after 2 and 4 years, respectively. They showed the status of the fovea and the total lesion size were the strongest predictor variables of BCVA. In this study, the baseline GA area was 4.1 mm2, and at 5-year follow-up, GA area increased to 16.57 mm2 and VA changed from 0.34 logMAR at baseline to 1.06 logMAR.
Previous investigations into GA growth rate among the South Korean and Asian population have been limited. In a study using FAF, Jeong et al. [15] reported a mean GA growth rate of 1.14 mm2/yr in patients with a mean follow- up period of 1 year using FAF, while Lee et al. [16] reported a mean GA growth rate of 1.47 mm2/yr in patients with a mean follow-up period of 2 years. Most recently, other studies involving both Asian and non-Asian cohorts reported lower GA growth rates in Asian population, with a mean growth rate of 0.6 and 2.0 mm2/yr for Asian and non-Asian population, respectively [18]. In a previous study, the GA growth rate was 2.44 mm2/yr in South Korean patients with an average follow-up of more than 3 years [14], compared to the 1.98 mm2/yr in the 5-year follow-up group in this study. GA growth rate investigations for the Caucasian population with a minimum follow-up period of 3 years or more yielded results of 1.78 mm2/yr by Lindblad et al. [19], 1.27 mm2/yr by Keenan et al. [20], and 2.60 mm2/yr by Sunness et al. [21]. GA growth rates were found to be higher in long-term follow-up, and similar to the Caucasian population, GA growth rates were measured in the South Korean cohort. While subtle variations exist due to differing baseline characteristics among the various studies, this investigation specifically analyzed GA growth rate in Korean patients with a long-term follow-up period over 3 years.
FAF patterns can affect GA growth rate [22]. Our study had many eyes with a diffuse FAF pattern, which may explain why growth seemed faster than in some reports. To keep enough data in each group, we used a simple split of unifocal versus multifocal GA. While patterns (particularly diffuse-trickling and banded) appear clearly associated with faster average progression, this association is intertwined with factors such as baseline lesion size/circumference, presence of subretinal drusenoid deposits/hyperreflective foci, and lesion location [23]. In the future, larger studies could sort eyes by standard FAF patterns and analyze sectoral growth over time to show pattern-specific growth.
Further analysis showed that GA area expansion in inferior zone 3 among the 12 zones correlated with decreased VA. In a topographic analysis of GA using CFP, Shen et al. [24] reported that GA area was greater in the superior and nasal regions than in other quadrants. They also found that total GA area correlated only weakly with VA, whereas GA area within the central 1-mm foveal zone showed the strongest association with vision. Our previous study reveals an association between VA and perifoveal GA enlargement with early detection before foveal involvement [14]. Similar to previous studies, zone of 1.5 to 2.0 mm from the fovea center (zone 3), and a more specific, the inferior zone, was associated. These results are probably due to the fact that the inferior zone area deteriorates late and slowly and worsens at the time of foveal involvement.
To examine the topographic and temporal progression of VA decline in GA, we selected a subgroup whose VA dropped to 1.0 logMAR and analyzed it retrospectively. In eyes with VA of 1.0 logMAR or less, zone 2 nasal sector showed a significant correlation with visual impairment. No clear causality has been identified in this regard. It has been reported that the U-shaped GA phenotype with foveal sparing is more frequent in the diffuse form [5,25–29]. In the data of this study, there were many diffuse GA with reticular pseudodrusen, so the increase in GA area in the nasal region over the temporal region of the fovea may have contributed to the rapid decrease in VA [30–36]. When looking at the 5-year GA growth rate, although it was not statistically significant, the growth rate in the superior region was initially high, and the growth rate in the temporal region after 3 years and the nasal region after 4 years showed a tendency to be higher than other regions. From this tendency, it can be assumed that the nasal region is the last part to show the highest growth rate before fovea invasion. These observations as hypothesis-generating and recommend replication in larger cohorts. Future work should also incorporate foveal status metrics and OCT- or OCT angiography-based assessments to strengthen structure–function inference.
In this study, the foveal area, the central 1-mm circle (zone 0), was not included in analysis. As the fovea appears darker than surrounding areas from luteal pigment, evaluating foveal integrity in the presence of GA by FAF alone can be challenging. All total GA areas in this study refer to the area excluding the fovea area. In many other studies, the fovea invasion of the GA area was measured and analyzed in various ways [19,37–39]. However, due to equipment limitations and difficulty in interpretation, some studies have expressed it as probably involved or definite involved [7,9,40,41]. In addition, more consistent data could be obtained by removing the fovea area to reduce program error in calculating the area with a semiautomated program. Nonetheless, OCT-based assessment of foveal integrity would further clarify structure–function relationships.
VA is strongly linked to foveal involvement [13,32,40,41], yet determining macular involvement from FAF images alone is challenging. Assessing the degree of macular atrophy on OCT is therefore critical for interpreting its relationship with vision. In this study, we a priori excluded the central 1-mm foveal zone and focused on perifoveal topography and longitudinal changes outside the foveal center. We speculate that progression of perifoveal atrophy may be informative for determining when to initiate intravitreal therapy, potentially enabling treatment before the macula becomes imminently threatened and helping to identify an earlier window for intervention. Future studies should incorporate foveal status, such as minimum distance to the fovea and time to foveal involvement, and OCT-based metrics of foveal integrity to more directly capture structure–function coupling and strengthen causal inference.
Limitations
This study has some limitations. First, the small number of patients and eyes involved in the analysis. At baseline, 58 eyes were included, but the number of analyzable eyes decreased to 20 eyes at the 5-year follow-up, and only 17 eyes were included in the group with decreased VA. This association should be considered preliminary and hypothesis-generating, and future large-scale studies are needed to confirm the association. Secondly, due to a retrospective study design, the follow-up period was not consistent. Thirdly, the accuracy of quantifying GA area using RegionFinder has not been proven in the literature. Several studies were conducted via RegionFinder software, Biarnes et al. [42] and Abdillahi et al. [43] used RegionFinder to quantify GA and to evaluate the GA growth rate or relationship between GA progression and an increased FAF. Although this study excluded fovea area, measuring near foveal GA area was challenging. Therefore, we used other imaging modalities including OCT, CFP to confirm lesion size. However, there could be the degree of precision in measured atrophic area around the fovea can affect the progression of zone 1 area. Fourth, using Dunn post hoc procedure, only the nasal zone 2 exhibited a selective acceleration pattern: 1 year before the 1.0 logMAR visit differed from both 2 and 3 years before (adjusted p = 0.023 and p = 0.034, respectively), while the 2- versus 3-year comparison was not significant (adjusted p = 0.812). Therefore, the observed patterns should be interpreted as exploratory signals. Furthermore, the modest sample size, particularly at the later time points, likely limited the statistical power to detect sector-specific effects after correction. Thus, these findings warrant confirmation in larger cohorts with pre-specified multiplicity control.
Conclusions
The superior perifoveal area was involved earlier and more extensively, while the inferior perifoveal region was involved later and to a lesser extent. It would be helpful to analyze the regional growth rate of GA, especially the highest growth rate at that follow-up interval to evaluate the treatment effect. In case of small GA lesions, GA area doubling time can be proposed as an additional parameter. A GA progression in the nasal perifoveal region (2–3 mm from the fovea) was associated with subsequent, clinically meaningful visual decline; this signal is exploratory and warrants replication in larger, independent cohorts. Future work should relate total GA area to regional growth dynamics and define practical thresholds to better guide monitoring and timing of treatment.
Notes
Conflicts of Interest:
None.
Acknowledgements:
None.
Funding:
None.