Natural History of Central Sparing in Geographic Atrophy Secondary to Non-exudative Age-related Macular Degeneration

Abstract

Background:

The macular central 1-mm-diameter zone is crucial to patients’ visual acuity, but the long-term natural history of central sparing in eyes with geographic atrophy (GA) is unknown.

Methods:

We manually segmented GA in 210 eyes with GA involving central 1-mm-diameter zone (mean follow-up = 3.8 years) in the Age-Related Eye Disease Study. We measured the residual area in central 1-mm-diameter zone and calculated central residual effective radius (CRER) as square root of (residual area/π). A linear mixed-effects model was used to model residual size over time. We added a horizontal translation factor to each dataset to account for different durations of GA involving the central zone.

Results:

The decline rate of central residual area was associated with baseline residual area (P = 0.008), but a transformation from central residual area to CRER eliminated this relationship (P = 0.51). After the introduction of horizontal translation factors to each dataset, CRER declined linearly over approximately 13 years (r² = 0.80). The growth rate of total GA effective radius was 0.14 mm/year (95% CI = 0.12–0.15), 3.7-fold higher than the decline rate of CRER (0.038 mm/year, 95% CI = 0.034–0.042). The decline rate of CRER was 53.3% higher in eyes with than without advanced age-related macular degeneration in the fellow eyes at any visit (P = 0.007).

Conclusions:

CRER in eyes with GA declined linearly over approximately 13 years and may serve as an anatomic endpoint in future clinical trials aiming to preserve the central zone.

INTRODUCTION

Geographic atrophy (GA) is the late stage of nonexudative age-related macular degeneration (AMD), affecting over 5 million people worldwide.¹ GA is characterized by the presence of atrophic lesions in the macula, with progressive degeneration of photoreceptors, retinal pigment epithelium, Bruch’s membrane, and choriocapillaris.² GA lesions usually develop in the parafovea with central sparing so most eyes with GA maintain good visual acuity until the atrophy involves the fovea.^3–5 Although there are currently no approved therapies for slowing GA progression, many clinical trials are ongoing⁶ and a recent phase 2/3 trial demonstrated that a complement C5 inhibitor significantly reduced GA growth rate.⁷

Total GA area is the most common primary endpoint in clinical trials in GA,⁶ but it is weakly associated with patients’ visual function.^{4, 8–10} Recent studies have demonstrated that the degree of GA involvement in the central zone significantly correlates with visual acuity.^{4, 9–13} We recently showed that although total GA area was poorly associated with visual acuity (r² = 0.07), GA area in central 1-mm- diameter zone had the highest correlation with visual acuity (r² = 0.45) as we varied the diameter of the central zone from 0 to 10 mm.⁴ Full GA coverage of central 1-mm-diameter zone was associated with a loss of 34.8 ETDRS letters.⁴ Therefore, central residual size may be an alternative endpoint for treatment trials.

Although central sparing has gained increased attention, only a few studies have assessed the longitudinal change of central sparing in eyes with GA.^{3, 11, 14} In these studies, the definition of the central zone varies widely and the reported mean decline rate of the residual area of the central zone ranges from 0.04 to 0.25 mm²/year.^{3, 11, 14} Moreover, the pattern of GA progression in the central zone is unknown and may be influenced by the unique anatomic microstructure (e.g., high density of cones or macular pigment). A better understanding of the natural history of central sparing is essential to patient counseling and the design of treatment trials aiming to preserve the central zone. Therefore, we investigated the natural history of the residual size in central 1-mm-diameter zone in eyes with GA using data from the age-related eye disease study (AREDS).

METHODS

Study population and image grading

We obtained data from the AREDS via the database of Genotypes and Phenotypes (dbGaP Study Accession: phs000001.v3.p1) after approval.¹⁵ The AREDS was a prospective, multicenter, randomized controlled trial to evaluate the effects of oral supplements on the progression of AMD and cataracts.^16–18 Eleven retinal specialty clinics recruited 4757 participants aged 55 to 80 years from 1992 to 1998, and randomly assigned them into 4 groups (Table 1).¹⁸ The dietary nutrient intake was estimated based on a self-administered, 90-item, semiquantitative food frequency questionnaire at enrollment.¹⁷ Color fundus photographs (CFPs) were taken annually.¹⁶ The AREDS obtained informed consent from all participants before enrollment. Our study was exempted from the need for approval by the Yale University Institutional Review Board and adhered to the tenets of the Declaration of Helsinki.

Table 1.

Impact of oral supplements on the decline rate of central residual effective radius after adjustment for age, sex, and baseline central residual effective radius

	Placebo	Antioxidants*	Zinc†	Antioxidants plus Zinc
Number of patients	39	33	48	38
Age, years, mean (SD)	70.3 (5.8)	68.5 (5.2)	71.8 (4.9)	71.0 (5.4)
Sex, female, n (%)	22 (56.4)	12 (36.4)	31 (64.6)	25 (65.8)
Duration of follow-up, years, mean (SD)	4.4 (2.7)	4.5 (2.6)	3.5 (2.0)	3.6 (2.3)
Advanced AMD in the fellow eye at any visit, n (%)	26 (66.7)	23 (69.7)	32 (66.7)	24 (63.2)
Baseline central residual effective radius, mm, mean (SD)	0.40 (0.09)	0.39 (0.11)	0.38 (0.11)	0.38 (0.12)
Decline rate of central residual effective radius, mm/yr, mean (95% CI)	0.041 (0.033–0.051)	0.039 (0.029–0.049)	0.039 (0.030–0.048)	0.036 (0.037–0.046)
P compared with the decline rate in placebo‡	NA	0.85	0.70	0.67

AMD, age-related macular degeneration; CI, confidence interval; FE-Advanced AMD, the fellow eye had advanced AMD; FE-None, the fellow eye did not have advanced AMD; GA, geographic atrophy.

^*Antioxidants = 500 mg of ascorbic acid; 400 IU of dl-alpha-tocopherol acetate; and 15 mg of beta carotene

^†Zinc = 80 mg as zinc oxide and copper; and 2 mg as cupric oxide.

^‡P value of interaction between treatment group and year after adjusting for age, sex, and baseline central residual effective radius. Overall P = 0.97.

In our analysis, we included eyes with GA involving central 1-mm-diameter zone for ≥ 2 follow-up visits. We included only those visits of an eye between the first visit when GA involved central 1-mm-diameter zone (defined as baseline visit) and the last visit before GA covered the entire central zone. We included 1 eye per patient for analyzing the effect of oral supplements and dietary nutrients.¹³ For patients with bilateral GA, the eye with the highest number of eligible visits was included. If the numbers of eligible visits was the same, the right eye was used.¹⁹ We included all eligible eyes in the analysis of the natural history of central sparing. We excluded CFPs with poor image quality, eyes with neovascular AMD at any visit, and eyes with GA covering the entire central zone at the first available visit.

The University of Wisconsin fundus photograph reading center previously graded all CFPs from the AREDS for the presence of GA, the status of AMD, and other AMD-related fundus abnormalities.²⁰ However, segmentations of GA lesions were not available in the AREDS data files. We manually segmented GA on CFPs of each eye using ImageJ software (version 1.52p; National Institutes of Health, Bethesda, Maryland, USA)²¹. We described the methods of image segmentation in a separate paper⁴ and in online supplementary methods. Based on these gradings, we measured GA area in the central zone (1 mm in diameter; Figure 1A–C).

Figure 1.

Measurement of the central residual area on colour fundus photographs of the right eye of a representative patient. The solid white lines represent the borders of geographic atrophy (GA) lesions and the broken black circle represents central 1 mm diameter zone. We defined central residual area as the nonatrophic area in central 1 mm diameter zone. We calculated the central residual effective radius as square root of (central residual area/π). For this eye, central residual area was (A) 0.61 mm2 at year 2, (B) 0.25 mm2 at year 5 and (C) 0.22 mm2 at year 8 after enrollment into the study.

Statistical analysis

The statistical analysis was performed in MATLAB (The MathWorks, Inc, Natick, MA) and R 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria). We calculated central residual area (CRA) as the maximum area in central 1-mm-diameter zone (0.785 mm²) – GA area in the central zone. We defined central residual effective radius (CRER) as$\sqrt{\frac{CRA}{\pi }}$. We also calculated total GA area and GA effective radius (i.e.,$\sqrt{\frac{TotalGAarea}{\pi }}$.^{22, 23} We determined the decline rate of CRA and CRER of the entire cohort via a linear mixed-effects model using the eye as the unit of analysis (“lme4” package²⁴ in R).

Oxidative stress to the retina has been suggested to play a role in the pathogenesis of AMD.⁶ Several studies demonstrated that oral supplementation with antioxidants (e.g., vitamin C, vitamin E, and carotenoids) and minerals (e.g., zinc) significantly reduced the risk of progression from intermediate AMD to advanced AMD.^{17, 25, 26} But the impact of these compounds on GA progression in the central zone is unknown. Therefore, we performed linear mixed-effects models to model the change in central residual size (area or effective radius) over the years of follow-up for the effects of AREDS supplements and dietary nutrients¹⁷ while adjusting for confounding factors (listed in tables). We calculated a nutrient density (nutrient intake/total energy intake) for each nutrient.¹⁷

A separate analysis was performed to explore the relationship between the decline rate of central residual size and fundus characteristics. We estimated a yearly decline rate of CRA and CRER in each eye using a linear regression. We then performed a univariable linear mixed-effects model to model the decline rate as a function of baseline central residual size, baseline GA focality, and the fellow eye status. Next, we performed a multivariable linear mixed model, including the 4 characteristics as well as age and sex. We classified the status of the fellow eye based on the presence of GA and the presence of advanced AMD (central GA or neovascular AMD; termed as FE-Advanced AMD and FE-None). We defined central GA as GA involving the foveal center point.¹³ We used the original AREDS gradings to determine the AMD status for the fellow eyes that we were unable to segment GA.

We next investigated the hypothesis that CRER declined linearly over time. We first plotted CRER as a function of follow-up time. To account for different baseline durations of GA in the central zone, we added a horizontal translation factor (in years) to each dataset,^{5, 22, 27–32} which converted the horizontal axis from “follow-up time” to “inferred duration of GA in the central zone”. We estimated the translation factors by adjusting 1 translation factor by 1 month at a time iteratively until the r² was maximized for the cumulative trendline.^{5, 22, 27–32} We predefined the trendline as a linear regression with a y-intercept of 0.5 mm and a slope of CRER decline rate estimated from a linear mixed-effects model. We performed the entry time realignment using a custom MATLAB program.^{30, 31}

RESULTS

Intergrader reproducibility and patient characteristics

We found good intergrader reproducibility of CRA between the 2 teams (intraclass correlation coefficient = 0.89), with a mean difference of 0.02 mm² (Supplementary Figure 1). We included 865 eligible visits of 210 eyes in 158 patients (57% were females). The mean ± standard deviation of follow-up duration was 3.8 ± 2.3 years for central sparing. At baseline, the age was 70.5 ± 5.4 years, total GA area was 4.21 ± 6.00 mm², and CRA was 0.51 ± 0.23 mm². The growth rate of total GA area was 1.34 mm²/year (95% confidence interval (CI) = 1.11–1.57), and the growth rate of total GA effective radius was 0.14 mm/year (95% CI = 0.12–0.15). In comparison, the decline rate of CRA was 0.067 mm²/year (95% CI = 0.060–0.074). The decline rate of CRER was 0.038 mm/year (95% CI = 0.034–0.042).

Effects of oral supplements and dietary nutrients on central sparing

The decline rate of CRA was comparable between placebo and treatment groups (0.061–0.074 mm²/year; P = 0.41–0.64; Supplementary Table 1). Similarly, the decline rate of CRER was comparable across all 4 groups (0.036–0.041 mm/year; P = 0.67–0.85; Table 1). We did not find any significant association between dietary nutrient intake (carotenoids and vitamin A, C, and E) and the decline rate of CRA or CRER (Supplementary Table 2–5).

Natural history of central sparing

We investigated the natural history of central sparing based on all 210 eligible eyes. The results from the univariable analysis (Supplementary Table 6) and multivariable analysis (Table 2 and Supplementary Table 7) were consistent. A larger baseline CRA and advanced AMD in the fellow eye at any visit were independently associated with a higher decline rate of CRA (P = 0.008 and 0.002, respectively; Table 2).

Table 2.

Multivariable linear mixed model on the decline rate of central residual area and effective radius after adjustment for age and sex

	Number of Eyes	Decline Rate of Central Residual Area (mm2/yr)		Decline Rate of Central Residual Effective Radius (mm/yr)
		Mean (95% CI)	P *	Mean (95% CI)	P *
Baseline central residual size†			0.008		0.51
≤ 0.39 mm2	63	0.058 (0.040–0.076)		0.050 (0.037–0.064)
> 0.39 mm2	147	0.076 (0.066–0.086)		0.038 (0.032–0.043)
Baseline GA focality			0.87		0.71
Unifocal	117	0.068 (0.057–0.080)		0.043 (0.035–0.051)
Multifocal	93	0.073 (0.060–0.086)		0.040 (0.032–0.047)
Fellow eye status at any visit			0.002		0.007
FE-Advanced AMD‡	148	0.078 (0.068–0.089)		0.046 (0.039–0.053)
FE-None	62	0.052 (0.039–0.065)		0.030 (0.023–0.037)

AMD, age-related macular degeneration; CI, confidence interval; FE-Advanced AMD, the fellow eye had advanced AMD; FE-None, the fellow eye did not have advanced AMD; GA, geographic atrophy.

^*P values were from the multivariable linear mixed model including covariates listed in the table as well as age and sex. Age and sex were not significantly associated with the decline rate of central residual area or effective radius.

^†Baseline central residual size (area or effective radius) was entered as a continuous variable in the multivariable linear mixed model.

^‡Advanced AMD was defined as central GA or neovascular AMD. Among 148 fellow eyes that had advanced AMD at any visits, 130 eyes had central GA, 11 eyes had neovascular AMD, and 7 eye had both central GA and neovascular AMD.

After we converted CRA to CRER, the decline rate of CRER was independent of the baseline CRER (P = 0.51; Table 2). The presence of advanced AMD in the fellow eye at any visit was still independently associated with a higher decline rate of CRER (P = 0.007; Table 2). We found the decline rate of CRER was 53.3% higher in the FE-Advanced AMD group (0.046 mm/year, 95%CI = 0.039–0.053) than in the FE-None group (0.030 mm/year, 95%CI = 0.023–0.037; Table 2). When we defined the fellow eye status using the baseline visit, the difference between the 2 groups was 26.3% (P = 0.13; Supplementary Table 7). The presence of GA in the fellow eye at baseline or any visit was not significantly associated with the decline rate of CRA or CRER (P = 0.26–0.78).

To explore the relationship between the decline rate of central residual size and the baseline residual size further, we classified the cohorts into small (63 eyes) and large (147 eyes) baseline CRA groups using 0.39 mm² as a cut-off (i.e., half of central 1-mm-diameter zone). When CRA was the outcome measure, the decline rate was significantly higher in the large than in the small baseline CRA group (0.083 vs. 0.045 mm²/year, P < 0.001; Figure 2A). However, after we transformed the CRA to CRER, the decline rates became comparable between the 2 groups (0.037 vs. 0.041 mm/year, P = 0.28; Figure 2B).

Figure 2.

Change of the central residual area and effective radius in eyes with small and large baseline residual area over 4 years (N = 210 eyes). We defined the small baseline residual area (orange) as ≤0.39 mm2 and the large baseline residual area (blue) as>0.39 mm2. We chose 0.39mm2 as the cut-off because it is approximately half of the maximum area of central 1 mm diameter zone (0.785mm2). In each figure, the left y axis is for the small baseline residual area group, and the right y axis is for the large baseline residual area group. The error bar represents the 95% CI. (A) The decline rate of the central residual area was significantly higher in the large residual area group (0.083mm2 /year, 95% CI =0.072 to 0.093; blue) than in the small residual area group (0.045mm2 /year, 95% CI=0.036 to 0.054; orange) (p<0.001). (B) After converting the area to effective radius in the y axis, the two lines became closer to each other. The decline rate of the central residual effective radius was similar for the large residual area group (0.041mm/year, 95% CI=0.034 to 0.047; blue) and the small residual area group (0.037mm/year, 95% CI=0.030 to 0.045; orange) (p=0.28).

CRER declined over time after the first observation of GA in the central zone (Figure 3A). The baseline CRER varied widely among individual eyes, suggesting that different eyes might have different durations of atrophy in the central zone at baseline. Also, the correlation between CRER and time was poor (r² = 0.16). After we introduced a horizontal translation factor to each dataset,^{5, 22, 27–32} the datasets followed a linear decline as a function of time in the central zone over approximately 13 years (r² = 0.80; compare Figure 3A with Figure 3B). The decline rate of CRER was consistent across eyes with different baseline CRER (r = −0.08; Supplementary Figure 2).

Figure 3.

Decline of central residual effective radius as a function of time. (A) Decline of the central residual effective radius as a function of time in 210 individual eyes. Each dashed line represents 1 eye, and the solid circles on the dotted line represent the central residual effective radius at baseline and subsequent follow-up(s). At time 0, the central residual sizes varied widely across different eyes, suggesting that these eyes might have different durations of geographic atrophy (GA) in the central zone. Despite the large interindividual variations, the central residual effective radius declined over time (r2 = 0.16). (B) Same data after adding a horizontal translation factors (expressed in years) to each dataset (ie, each dotted line) in (A) to account for different baseline durations of atrophy in the central zone among individual eyes. Horizontal axis now represents the inferred duration of GA in the central zone rather than the follow-up time. After the introduction of the horizontal translation factors, datasets now followed along a straight line with a decline rate of 0.038 mm/year (r2 = 0.80). The decline rate in central residual effective radius appeared to be relatively constant over the elapsed time.

DISCUSSION

To our knowledge, this study is the first to investigate the long-term natural history of central sparing in eyes with GA, and to explore the impact of oral supplements on GA progression in the central zone. We found that the decline rate of CRA was positively correlated with the baseline CRA (P = 0.008); however, after we transformed CRA (in mm²) to CRER (in mm), the decline rate was no longer dependent on the baseline central residual size (P = 0.51; Table 2). Interestingly, baseline CRER varied widely across individual eyes (Figure 3A). A priori, the reasons for the differences are unclear, but at least 2 possibilities exist. First, each eye may represent a unique GA phenotype with a distinct progression pattern. However, a more unifying hypothesis is that different eyes represent the same disease with a similar underlying GA progression pattern in the central zone. If the latter hypothesis is true, the variation of baseline CRER among individual eyes may simply arise from differences in the time at which these eyes entered the study. In other words, on average, eyes with small CRER had longer durations of GA in the central zone compared with eyes with larger CRER at baseline. After we introduced horizontal translation factors to individual datasets to account for different entry times, CRER declined linearly as a function of time over approximately 13 years (Figure 3B). We determined the decline rate of CRER as 0.038 mm/year (95% CI = 0.034–0.042), comparable to our previous estimate (0.041 mm/year) via a meta-analysis⁵. Also, our present study showed that total GA effective radius growth rate (0.14 mm/year) was 3.7-fold higher than the decline rate of CRER (0.038 mm/year), consistent with previous reports that GA progressed at a much faster rate (approximately 3-fold higher) towards the periphery than towards the foveal center.^{3, 5} We found no significant impact of AREDS supplements and dietary nutrients (carotenoids and vitamin A, C, and E) on the decline rate of CRA or CRER.

Our results shed light on the underlying pattern of GA progression. Previous studies showed that the growth rate of total GA area was positively associated with baseline GA size.^{19, 22, 23, 33} One possible explanation is that GA with a larger baseline size represents a more aggressive stage. If this hypothesis is correct, the decline rate of CRA should be higher in eyes with smaller baseline CRA (i.e., more severe GA) than in eyes with larger baseline CRAs (i.e., less severe GA). However, our data showed the opposite results (Figure 2A), and thus does not support this hypothesis. A second hypothesis is that the edge of a GA lesion may expand at a relatively constant rate over time; that is, the radius of a circular GA lesion enlarges linearly as a function of time.^{22, 23, 33} Our finding that CRER declined linearly over time was consistent with this hypothesis. For an intuitive interpretation of CRER, we may consider a circular residual island in central 1-mm-diameter zone (similar to Figure 1C). If the edge of the GA lesion progresses at a constant rate, the radius of the residual island will decrease linearly over time.

The decline rate of CRER was 53.3% higher in eyes with than without advanced AMD in the fellow eyes (0.046 vs. 0.030 mm/year, P = 0.007), suggesting that eyes in the FE-Advanced AMD group may have a more aggressive form of GA. This finding is consistent with previous reports that the growth rate of total GA size was 30–60% higher in eyes with GA or neovascular AMD in the fellow eyes.^{28, 33} Interestingly, when we classified the fellow-eye status based on the baseline visit, we did not find a significant difference in the decline rate of CRER between the FE-Advanced AMD and FE-None groups (0.048 vs. 0.038 mm/year, P = 0.13). A closer examination of our data revealed that among 146 eyes in the FE-None group at baseline, 84 eyes eventually developed advanced AMD in the fellow eyes over the follow-up period. On average, these 84 eyes had a 50% higher GA growth rate in the central zone compared with the remaining 62 eyes that never developed advanced AMD in the fellow eyes (0.045 vs. 0.030 mm/year, P = 0.02). Therefore, the classification of the fellow-eye status based on all follow-up visits instead of the baseline visit may be a better indicator of the aggressiveness of GA, although we recognize that it is less useful as a predictive factor. Future clinical trials in GA should monitor the fellow eye status during follow-up visits and account for the potential confounding effect.

Currently, many ophthalmologists advise patients with GA to take AREDS or AREDS2 supplements, but the benefits of these supplements may not be correctly recognized by many clinicians.³⁴ Based on a survey of 216 ophthalmologists, more than half expected to see a slowing of GA progression in patients taking nutritional supplements.³⁴ However, reports from the AREDS and AREDS2 did not show a statistically significant impact of the oral supplements on the progression from ealy or intermediate AMD to noncentral or central GA,^{18, 26} or on the progression rate of total GA size.^{13, 33} Our present study added that there was no significant difference in the decline rate of central residual size between the placebo and treatment groups in the AREDS (Table 1). Therefore, ophthalmologists should convey this information regarding AREDS and AREDS2 supplements to patients with GA, given the cost of AREDS supplements and potential adverse effects (e.g., increased risk of lung cancer in smokers taking beta-carotene supplementation;³⁵ increased risk of hemorrhagic stroke and all-cause mortality in patients taking Vitamin E supplements^{36, 37}). Ophthalmologists should also highlight the main evidence-based benefit of AREDS and AREDS2 supplements in patients with GA is to reduce the risk of developing neovascular AMD.^{18, 26} Due to the lack of access to AREDS2 data, we do not know if our findings are applicable to AREDS2 suppelements. Future studies are needed to investigate the impact of AREDS2 supplements on GA progression in the central 1-mm-diameter zone.

Most previous clinical trials in GA focused on the efficacy of therapeutic agents in patients with pre-existing GA.⁶ An alternative approach is to treat patients with earlier stages of AMD. The international Classification of Atrophy Meeting group has proposed OCT-anchored multimodal imaging as a method to identify precursor endpoints that allow earlier estimation of tissue loss secondary to AMD.^38–40 For example, the group has incorporated the evidence of overlying photoreceptor degeneration as an OCT-based anatomic feature to define incomplete RPE and outer retinal atrophy, a precursor of GA.⁴⁰ Future longitudinal studies are needed to provide more information on these OCT-based biomarkers and to establish the validity and reproducibility of such structural endpoints for treatment trials.

Our study had several limitations. First, the number of patients in each treatment group was relatively small, and our analysis was retrospectively designed, so we cannot exclude the possibility that the effect from AREDS supplements could be detected in a larger prospective study. Second, the dietary nutrient intake of each patient was estimated based on a questionnaire in the AREDS. The accuracy of the estimates may be affected by the subjective nature of the assessment, patients’ age, GA severity, and comorbidities, and poor recollection by some patients. Third, there are significant measurement errors in determining the foveal center based on CFPs. We tried to improve the accuracy of the foveal center determination by using the best image with a clear foveal center across all visits of an eye, and registered the foveal center marking to all images with GA.⁴ This method allowed us to have a consistent foveal center across different visits of the same eye and to achieve a good intergrader reproducibility on the measurements of GA area in the central zone, which is the main outcome measure in our study. Nonetheless, the median distance between the foveal center markings by the 2 teams was relatively large (199 μm in our study⁴, comparable to the median difference (178 μm) reported by Sunness et al based on CFP⁴¹). This measurement error could contribute to variations in the decline rate of CRER among patients and alter our other results. Future studies may employ optical coherence tomography (OCT) or OCT angiography to identify the foveal center more accurately and repeat our analyses. Fourth, the CFP used in our study was acquired as film images in the AREDS. It is possible that the improvement in photographic quality over recent years may increase the accuracy in measuring the residual area in the central zone and may alter some of our results (e.g., resulting in less variation in the decline rate of CRA and CRER among different individuals). Therefore, we encourage future studies to replicate our analyses using digital, full color fundus photography. Fifth, since our study was based on CFP, we were unable to investigate the associations between the decline rate of CRER and biomarkers identified on fundus autofluorescence or OCT (e.g., fundus autofluorescence signals surrounding GA lesions, subretinal drusenoid deposits, choroidal thickness, and outer retinal tubulation). Future studies are needed to explore these relationships.

In conclusion, we determined the long-term natural history of the residual size in central 1-mm-diameter zone. Although the decline rate of CRA was significantly associated with the baseline residual size, a transformation from CRA to CRER eliminated this significant relationship. CRER declined linearly over approximately 13 years at a rate of 0.038 mm/year (95% CI = 0.034–0.042). CRER may serve as an anatomic endpoint for clinical trials aiming to preserve the central zone. The presence of advanced AMD in the fellow eye is a significant biomarker for the decline rate of CRER.

SYNOPSIS.

The residual effective radius (square root of area/π) of central 1-mm-diameter zone in eyes with geographic atrophy declines linearly over approximately 13 years at 0.038 mm/year (95% confidence interval = 0.034–0.042).