Alterations of the Ganglion Cell Complex in Age-Related Macular Degeneration: an AMISH Eye Study Analysis

Abstract

Purpose:

To compare the Ganglion Cell Complex (GCC) thickness in eyes with age-related macular degeneration (AMD) versus healthy controls in an elderly Amish population

Design:

Cross-sectional study

Methods:

This is a post hoc analysis of the family-based prospective study of Amish subjects. Study subjects were imaged by the Cirrus HD-OCT (Carl Zeiss Meditec Inc, Dublin, CA) using a macular cube protocol of 512 × 128 scans (128 horizontal B-scans, each compromising 512 A-scans) over a 6 mm × 6 mm region centered on the fovea. The ganglion cell analysis algorithm calculated the GCC thickness by segmenting the outer boundaries of the retinal nerve fiber layer (RNFL) and inner plexiform layer (IPL) in all B-scans of the volume, with the region between these boundaries representing the combined thickness of the GCL and the IPL layer. A number of parameters were used to evaluate the GCC thickness: the average GCC thickness, minimum (lowest GCC thickness at a single meridian crossing the elliptical annulus), and sectoral (within each of six sectoral areas: superior, superotemporal, superonasal, inferior, inferonasal, and inferotemporal). The stage of AMD was graded on color fundus photographs in accordance with the Beckman Initiative for Macular Research classification system.

Results:

Of 1339 subjects enrolled in the Amish eye study, a total of 1294 eyes of twelve hundred and ninety-four subjects had all required imaging studies of sufficient quality and were included in the final analysis, and of these 798 (62%) were female. Following age adjustment, the average GCC thickness was significantly (p<0.001) thinner in AMD subjects (73.71 ± SD; 13.77 μm) compared to normals (77.97 ± 10.42 μm). Independent t test showed that, early (75.03 ± 12.45 μm), and late AMD (61.64 ± 21.18 μm) groups (among which GA eyes had the lowest thickness of 58.10 ± 20.27 microns) had a statistically significant lower GCC thickness compared to eyes without AMD. There was no significant difference in average GCC thickness between early AMD and intermediate AMD (76.36 ± 9.25 μm) eyes.

Conclusions:

The GCC thickness in AMD eyes is reduced compared to normal eyes, but the relationship is complex with the greatest reduction in late AMD eyes (particularly GA eyes) but no difference between early and intermediate AMD eyes.

Table of Contents Statement:

The ganglion cell complex thickness is reduced in eyes with late age-related macular degeneration, and in particular geographic atrophy, which may have implications for therapeutic approaches aiming to restore vision for these patients.

Introduction:

Age-related macular degeneration (AMD) is a leading cause of irreversible blindness in individuals over 65 years of age, particularly in developed nations.¹

AMD is a degenerative process that primarily involves the outer retinal layers, including the photoreceptors, retinal pigment epithelium, Bruch’s membrane, and choriocapillaris.^2–5 However, several studies have demonstrated that the inner retinal layers, including the ganglion cell layer (GCL), inner plexiform layer (IPL), and retinal nerve fiber layer (RNFL) may be secondarily impacted in both dry and wet AMD.^6–9

Lee and colleagues reported significant thinning of the GCL in the fellow eyes of patients with macular neovascularization (MNV) compared to age-matched normal subjects.¹⁰ Additionally, eyes with early AMD and subretinal drusenoid deposits demonstrated significant thinning of the macular ganglion cell-inner plexiform layer compared to eyes with drusen alone.¹¹ Borrelli et al¹² observed a strong correlation between ganglion cell complex thickness and photoreceptors normalized reflectivity in eyes with intermediate AMD. Finally, histopathological studies have shown that postmortem eyes with MNV have 47% fewer GCL neurons than healthy eyes.¹³

One limitation of previous studies is that ganglion cell complex (GCC) thickness is a highly heritable trait. Studies have suggested that additive genetic factors are responsible for 81% of the variance in age-adjusted thickness, while environmental factors account for the remaining 19% of the variance.¹⁴ The use of a relatively isolated population can mitigate some of these concerns. Thus, in this present study, we analyzed GCC thickness in a cohort of elderly subjects with and without AMD recruited as part of the Amish Eye Study.

Methods:

Of the 1339 subjects (2668 eyes) who were initially enrolled in the Amish Eye Study, 45 eyes of forty-five subjects were excluded from this study due to missing either color fundus images or Cirrus OCT data (24 eyes) and poor image quality (21 eyes). Only the right eye of each subject was included in this post hoc analysis, which thus consisted of 1294 eyes of 1294 subjects in this cross-sectional study. Subjects were recruited between February 2013 and July 2016 from three Amish communities in Ohio, Indiana, and Pennsylvania with centralized analysis of images at the Doheny Image Reading and Research lab (DIRRL). The study was approved by the Institutional Review Boards of the University of Pennsylvania, University of Miami, Case Western Reserve University, and University of California – Los Angeles. The study was performed in accordance with the Health Insurance Portability and Accountability Act and adhered to the principles of the Declaration of Helsinki. Written informed consent was obtained from all the subjects before enrollment in the study. Subjects self-identified as Amish were included in the study if they were ≥50 years of age and members in a sibship with at least 1 person reported to have AMD. Subjects were excluded from the study if they had glaucoma or were deemed to be a glaucoma suspect, poor image quality (defined below), or presence of any eye disorder that might skew the interpretation of AMD features on imaging.

Study Procedures

The clinical examination consisted of a complete ophthalmological evaluation and multimodal imaging, including spectral domain optical coherence tomography (SD-OCT). Clinical exam data was recorded by clinical research coordinators, board-certified ophthalmologists, and ophthalmic technicians at designated examination sites. Collected demographic parameters included age, date of birth, sex, family history of AMD, and presence of other ophthalmological conditions. All clinical data was transmitted and entered into the database at the Amish Eye Study Coordinating Center at the University of Miami.

After pharmacological dilatation, study subjects were imaged by the Cirrus HD-OCT (Carl Zeiss Meditec Inc, Dublin, CA) using a macular cube protocol of 512 × 128 scans (128 horizontal B-scans, each compromising 512 A-scans) over a 6 mm × 6 mm region centered on the fovea. Raw OCT volume data (.dcm file format) were exported and transmitted to DIRRL for analysis.

Two independent DIRRL graders reviewed all images and excluded cases from the analysis if the signal strength was less than eight, or if motion artifacts, misalignment, blinking artifacts, or algorithm segmentation failure were evident.

The ganglion cell analysis (GCA) algorithm (Cirrus OCT software, version 11.5.2) was utilized to automatically compute the GCC thickness within a double elliptical ring region centered around the fovea; the vertical inner and outer radii were 0.5 and 2.0 mm, respectively, while the horizontal inner and outer radii were 0.6 and 2.4 mm, respectively. The computed values from the inner ring exclude the foveal 1mm diameter region from the analysis as the GCC is too thin to be accurately detected in this location. The outer ring includes the area in which the GCC is thickest in normal eyes.¹⁵ The annulus is horizontally stretched by a factor of 20% in order to match the normal distribution of the thickest area of the GCC. The dimensions and the shape of the annulus were determined based on a preliminary analysis that was executed using GCC maps of 47 healthy eyes.¹⁶

The GCA algorithm calculates the GCC thickness by segmenting the outer boundaries of both retinal nerve fiber layer RNFL and IPL from all B-scans in the volume. The distance between these two boundaries represents the combined thickness of the GCL and the IPL layer.^12,17 The following parameters are generated by the GCA algorithm: average GCC thickness across the entire volume, minimum (lowest GCC thickness in a single meridian crossing the elliptical region), sectoral GCC thickness (from six sectoral areas: 1 - superior, 2 - superonasal, 3 - inferonasal, 4 - inferior, 5 - inferotemporal, and 6 - superotemporal). The instrument software compares the GCC thickness values against the built-in age-matched normative database to determine whether the GCC thickness is within the normal range (P= 5% - 95%), borderline (P= 1% - 5%), or reduced (P <1% of normal distribution percentiles). The retinal map analysis was used to calculate central subfield retinal thickness, mean macular thickness, and mean macular volume.

To compare GCC thickness among different grades of AMD, color fundus photographs were graded according to the Beckman Initiative for Macular Research classification, which includes the following stages: 0. no apparent aging changes (no drusen, no pigment changes), 1. normal aging changes (only small drusen ≤ 63 μm), 2. early AMD (medium size drusen > 63 μm and ≤ 125 μm and no pigmentary changes), 3. intermediate AMD (iAMD; large drusen > 125 μm and/or pigmentary changes), and 4. Late AMD (neovascular AMD (MNV) and/or geographic atrophy (GA)).¹⁸ No apparent aging changes and normal aging changes were considered normal (no AMD) for this analysis. Companion OCT images were also assessed to confirm the accuracy of the color photo staging and to evaluate for features of exudation suggestive of MNV.

Grading of the AMD stage was also performed by two independent DIRRL graders. If there was a disagreement, a third senior grader was consulted.

Statistical Analysis

Descriptive statistics were reported as means ± standard deviations. The Shapiro-Wilk test was performed for all quantitative variables to detect departures from normality distribution. One-way repeated measures analysis of covariance (ANCOVA) was conducted to compare GCC thicknesses between different groups. Tukey post hoc tests were applied to all multiple pairwise comparisons. An independent t-test was performed to test the difference between the two groups. All p values presented in the tables and results adjusted for age and after Bonferroni correction. Other potential confounding factors such as refractive error, presence of glaucoma, and body mass index (BMI) were also adjusted for in the analysis among the different AMD groups. Statistical analyses were performed using SPSS Statistics version 20 (IBM, Armonk, NY). A P value < 0.05 was considered statistically significant.

Results:

Twelve hundred and ninety-four eyes of 1294 subjects were included in the final analysis, and of these, 798 (62%) were female. Among these, 959 eyes were graded as normal or normal aging, 149 eyes had early AMD, 138 eyes had intermediate AMD, and 48 eyes had late AMD. The mean age of the normal group was 64.1 ± 9.25 years, and that was significantly lower than the AMD group, which was 71.29 ± 10.62 years; p <0.001. Consequently, all subsequent analyses were adjusted for age to account for this significant difference. There was no significant (p>0.05) difference in GCC thickness between male and females for any AMD group (supplementary table 2).

Central subfield retinal thickness was significantly (p=0.009) lower in AMD subjects when compared to the normal group, 266.76 ± 32.47 vs. 259.86 ± 42.92 respectively, whereas mean macular thickness, and mean macular volume were not significantly different between normal and AMD subjects (Table 1).

Table 1:

Mean GCC Thickness Metrics for Normal and AMD Eyes

	Normal and Normal Aging	Early, Intermediate, and Late AMD	p
N (eyes)	959	335
Age (Years)	64.1 ± 9.25 (50–91)	71.29 ± 10.62 (50–99)	<0.001
Central Subfield retinal Thickness	266.76 ± 32.47 (64–517)	259.86 ± 42.92 (74–498)	0.003
Macular Volume	10.08 ± 0.6 (6.9–14.2)	9.99 ± 0.81 (6.6–17.7)	0.02
Macular Thickness	279.99 ± 17.54 (170–395)	277.35 ± 22.49 (182–492)	0.03
Average GCC Thickness	77.97 ± 10.42 (11–139)	73.71 ± 13.77 (14–120)	<0.001
Minimum GCC Thickness	73.82 ± 14.97 (0–102)	65.35 ± 20.92 (0–93)	<0.001
GCC Thickness Superior	77.98 ± 11.53 (16–125)	72.88 ± 16.09 (12–135)	<0.001
GCC Thickness Superonasal	79.38 ± 10.84 (8–120)	74.4 ± 15.04 (7–109)	<0.001
GCC Thickness Inferonasal	79.01 ± 23.35 (7–704)	74.85 ± 14.11 (13–126)	0.003
GCC Thickness Inferior	77.43 ± 19.78 (18–572)	72.54 ± 15.32 (12–134)	<0.001
GCC Thickness Inferotemporal	78.94 ± 11.23 (19–157)	76.95 ± 38.7 (6–709)	0.16
GCC Thickness Superotemporal	77.11 ± 11.15 (15–144)	73.74 ± 14.2 (15–127)	<0.001

AMD: Age-related macular degeneration, GCC: Ganglion cell complex. Normal or AMD Stage defined by Beckman Staging

One-way ANOVA demonstrated that the average GCC thickness was significantly differed in all AMD groups (p<0.001 supplementary table 1). Average GCC was significantly (p=0.002) thinner in AMD subjects (75.68 ± 11 μm) compared to those without AMD (normal/normal aging) (77.97 ± 10.42 μm). Minimum and sectoral GCC thickness were also lower in subjects with AMD than normals (p<0.05) (Table 1). The relationship persisted when excluding eyes with late AMD with early/intermediate AMD eyes showing thinner GCC compared with eyes without AMD (Table 2). Among the 48 eyes with late AMD, 36 eyes had GA and 12 eyes had MNV. Average GCC was numerically thinner in GA eyes (58.10 ± 20.27 microns) compared with MNV (71.75 ± 21.75 microns) eyes, though this was not statistically significantly (p=0.06).

Table 2:

Mean GCC Thickness in Normal vs Early/Intermediate AMD Eyes

	Normal and Normal Aging	Early and Intermediate AMD	p
N (eyes)	959	287
Age (Years)	64.1 ± 9.25 (50–91)	69.61 ± 10.02 (50–99)	<0.001
Central Subfield retinal Thickness	266.76 ± 32.47 (64–517)	264.6 ± 35.96 (120–498)	0.40
Macular Volume	10.08 ± 0.6 (6.9–14.2)	10.04 ± 0.6 (6.6–13.2)	0.0.99
Macular Thickness	279.99 ± 17.54 (170–395)	278.72 ± 16.55 (182–366)	0.98
Average GCC Thickness	77.97 ± 10.42 (11–139)	75.68 ± 11 (27–96)	0.002
Minimum GCC Thickness	73.82 ± 14.97 (0–102)	69.36 ± 17.57 (0–93)	<0.001
GCC Thickness Superior	77.98 ± 11.53 (16–125)	75.11 ± 13.13 (12–97)	0.001
GCC Thickness Superonasal	79.38 ± 10.84 (8–120)	76.22 ± 12.93 (7–98)	<0.001
GCC Thickness Inferonasal	79.01 ± 23.35 (7–704)	76.69 ± 10.85 (28–115)	0.11
GCC Thickness Inferior	77.43 ± 19.78 (18–572)	74.63 ± 11.63 (28–94)	0.03
GCC Thickness Inferotemporal	78.94 ± 11.23 (19–157)	76.79 ± 11.45 (21–97)	0.006
GCC Thickness Superotemporal	77.11 ± 11.15 (15–144)	75.25 ± 12.23 (15–98)	0.02

AMD: Age-related macular degeneration, GCC: Ganglion cell complex. Normal or AMD Stage defined by Beckman Staging

Differences in retinal and GCC thickness between normal subjects and individual AMD stages were also assessed (Table 3). Central subfield retinal thickness was significantly thinner in subjects with late AMD when compared to normal subjects (p<0.001), but not significantly different between normals vs. early AMD (p=0.79) and intermediate AMD (p=0.09). Average GCC thickness was significantly thinner in early (p=0.003) and late AMD (p<0.001) when compared to normal subjects. Interestingly average GCC thickness was not significantly different between the normal subjects and iAMD subjects (p=0.09) (Table 3).

Table 3:

Mean GCC Thickness Among Beckman AMD Stages

	Normal and Normal Aging	Early AMD	p *	Intermediate AMD	p **	Late AMD	p***
N	959	149		138		48
Age	64.1 ± 9.25 (50–91)	68.14 ± 10.44 (50–99)	<0.001	71.17 ± 9.35 (51–93)	<0.001	81.13 ± 8.55 (59–95)	<0.001
Thickness Central Subfield	266.76 ± 32.47 (64–517)	267.56 ± 32.91 (203–450)	0.99	261.53 ± 38.76 (120–498)	0.37	231.29 ± 65.41 (74–474)	<0.001
Macular Volume	10.08 ± 0.6 (6.9–14.2)	10.04 ± 0.54 (8.2–11.3)	0.89	10.03 ± 0.65 (6.6–13.2)	0.82	9.68 ± 1.56 (7.4–17.7)	<0.001
Macular Thickness	279.99 ± 17.54 (170–395)	278.79 ± 14.99 (228–315)	0.89	278.64 ± 18.07 (182–366)	0.87	269.14 ± 43.12 (205–492)	0.001
Average GCC Thickness	77.97 ± 10.42 (11–139)	75.03 ± 12.45 (27–96)	0.02	76.36 ± 9.25 (36–94)	0.40	61.64 ± 21.18 (14–120)	<0.001
Minimum GCC Thickness	73.82 ± 14.97 (0–102)	69.03 ± 19.38 (7–93)	0.005	69.71 ± 15.52 (0–92)	0.03	40.75 ± 23.11 (0–85)	<0.001
GCC Thickness Superior	77.98 ± 11.53 (16–125)	74 ± 15.39 (12–96)	0.003	76.28 ± 10.19 (35–97)	0.46	58.84 ± 24.17 (18–135)	<0.001
GCC Thickness Superonasal	79.38 ± 10.84 (8–120)	76.22 ± 12.99 (27–98)	0.02	76.22 ± 12.91 (7–96)	0.02	63.23 ± 21.27 (21–109)	<0.001
GCC Thickness Inferonasal	79.01 ± 23.35 (7–704)	75.85 ± 11.48 (28–98)	0.36	77.58 ± 10.11 (46–115)	0.89	63.6 ± 23.65 (13–126)	<0.001
GCC Thickness Inferior	77.43 ± 19.78 (18–572)	73.9 ± 13.07 (28–94)	0.16	75.39 ± 9.87 (37–93)	0.64	59.78 ± 25.8 (12–134)	<0.001
GCC Thickness Inferotemporal	78.94 ± 11.23 (19–157)	76.18 ± 13.69 (21–97)	0.51	77.44 ± 8.49 (51–95)	0.88	77.91 ± 100.24 (6–709)	0.99
GCC Thickness Superotemporal	77.11 ± 11.15 (15–144)	73.89 ± 14.41 (15–98)	0.02	76.67 ± 9.28 (32–98)	0.98	64.03 ± 20.89 (17–127)	<0.001

AMD- Age-related Macular Degeneration, GCC - ganglion cell complex

^*significance levels between No AMD and Early AMD

^**significance levels between No AMD and Intermediate AMD

^***significance levels between No AMD and late AMD

GCC thickness in different AMD groups after adjusting for different covariates:

The mean spherical equivalent refractive error was statistically significantly (p=0.04) different between normal subjects (0.37 ± 1.85 Diopters) and subjects with AMD 0.61 ± 1.54 Diopters), though the actual numerical difference was small. The body mass index (BMI) was not significantly (p=0.87) different between normal subjects (33.62 ± 7.01) and subjects with AMD (33.69 ± 8.35). After adjusting for refractive error, the GCC was significantly (p<0.001) thinner in subjects with AMD (73.97 ± 13.84 microns) when compared to normal subjects 77.94 ± 10.74 microns).

Of 1294 eyes, thirty-three eyes also carried a diagnosis of glaucoma; average GCC thickness remained significantly different between the AMD groups even after adjusting for eyes with presence of glaucoma (p<0.001).

Discussion:

In this study, we evaluated the thickness of the ganglion cell complex among different stages of AMD and normal eyes in an elderly Amish population after adjusting for other potentially important covariates including age, BMI and eyes with glaucoma. The Amish population in the United States is a founder population that originated from two immigration waves from Switzerland into the United States in the 18^th and 19^th centuries and is genetically and culturally isolated. Consequently, the Amish have less genetic and environmental variability than the general population.¹⁹ With regards to environmental factors, Amish individuals do not smoke, and this is relevant as a detrimental effect of smoking on the GCC has also been reported. Chronic smokers showed a decrease in GCC thickness as well as diminished amplitude and delayed latency in the pattern electroretinogram and photopic negative response.²⁰ In-vitro studies have also demonstrated a direct injury to the retinal ganglion cells of rats by cigarette smoke extract.²¹ By mitigating the potential confounding effects of genetic and environmental diversity by using a population with reduced background heterogeneity, we were able to confirm that the GCC is thinner in different AMD grades compared to normal eyes. Before the advent of OCT, previous histologic studies reported an attenuation of ganglion cells in AMD eyes.^13,22 Clarke demonstrated considerable loss of ganglion cells in the macular region of AMD patients.²² With the broad availability of OCT, multiple studies have provided insight into potential alterations of the inner retinal layers in various stages of AMD, including the GCL and IPL.^23–25 These inner retinal alterations were also strongly associated with poor photoreceptors integrity and retinal vascular changes.^12,26 Photoreceptor impairment and subsequent photoreceptor loss during the evolution of AMD can be associated with significant remodeling and alteration of the inner retina. In addition to trans-synaptic degeneration of connected inner retinal neurons, significant migration, rearrangement, and rewiring of the inner retinal neurons can occur as a result of denervation from loss of input from rods and cones. Although these alterations can more substantially impact the bipolar and horizontal cells in the adjacent inner nuclear layer, they can extend to the ganglion cell layer as well.²⁷

Microvascular alterations and associated chronic ischemia may also result in damage to cells in the ganglion cell layer and inner plexiform layer in eyes with AMD.²⁸ Ganglion cell alterations and loss in AMD may occur from other mechanisms as well; Maynard and coworkers reported dysfunction of melanopsin expressing intrinsically photosensitive retinal ganglion cells that resulted in reduced sleep efficiency in patients with advanced AMD.²⁹

Overall, to summarize, several mechanisms have been proposed to be the trigger of ganglion cell injury and loss in AMD. First, as noted above, photoreceptor damage might lead to a chronic reduction of input to inner retinal layers, which may cause post-receptor trans-synaptic degeneration.^30–32 This hypothesis has been reinforced by the strong correlation between GCC thickness and ellipsoid zone normalized reflectivity in AMD patients and the absence of this correlation in normal eyes, suggesting pathological dysregulation in AMD eyes.¹² Additionally, there have been reports indicating the presence of dystrophic ganglion cells and alterations in glial cell immunoreactivity of muller cells in eyes with AMD.³⁰ Second, it has been postulated that inner retinal ischemia or alteration of the morphology of inner retinal vasculature may be the cause of GCC thinning in AMD eyes.^26,33 Third, mechanical pressure from drusen has been hypothesized to cause severe morphological changes in multiple retinal layers and subsequent retinal ganglion cell damage; this hypothesis was supported by the negative correlation between drusen area within the 5-mm circle and macular GCC thickness.⁸ However, this has been questioned by Borrelli et al, who did not find any relationship between the GCC complex thickness and drusen area or volume.¹² Recently, Abdolrahimzadeh and colleagues reported reduced thickness of GCC in early AMD and more thinning in eyes with subretinal drusenoid deposits (SDD) compared to eyes with drusen alone.¹¹ Based on the greater distribution of SDD within the choroidal watershed zones and the growing body of evidence of the association between SDD and choroidal vascular insufficiency, as well as the correlation between GCC thickness and choroidal thickness, they proposed that GCC thinning in AMD could be a component of the generalized spectrum of vascular abnormalities of the choroid and retina.^34–37

Ganglion cell death has been reported after overexposure to reactive oxygen species such as hydrogen peroxide, superoxide anions, and hydroxyl radicals.³⁸ However, they exhibit resistance to oxidative stress due to their endogenous peroxidases, which might be protective under conditions of axonal damage.³⁹ Several in vitro and in vivo studies have shown a protective role for nuclear transcription factor-kB, polysaccharides, transforming growth factor-beta, nerve growth factor, brain-derived neurotrophic factor, glutathione, superoxide dismutase, and catalase.⁴⁰ Approaches to preserve the inner retina in AMD eyes may be of particular relevance with regenerative medicine therapeutic strategies, such as optogenetics.⁴¹

Finally, while our study found that the GCC was thinner in eyes with overall AMD, the relationship appears to be more complex. In particular, eyes with intermediate AMD did not show a statistically significant reduction in GCC thickness for many metrics compared to normals, however the GCC was numerically thinner for all metrics and statistically significantly thinner for minimum GCC thickness. GCC thickness was generally not thinner in intermediate AMD eyes, compared to early AMD eyes. Notably, the GCC thickness was markedly thinner in late AMD eyes compared to early/intermediate AMD eyes. Taken together, these findings suggest that there may subtle but significant alterations in the inner retina in the transition from normal aging to early AMD, but substantial changes do not occur until the disease progresses to the late stage. We did observe a numerically thinner GCC in eyes with geographic atrophy compared to eyes with MNV. While this difference was not statistically significant, we suspect this is because of the relatively small number of MNV eyes, and thus our sample was underpowered for this assessment. We would also note that the GCC thickness in MNV eyes was numerically closer to early and intermediate eyes than to GA eyes, highlighting that GA eyes were primarily driving the lower GCC thickness among late AMD eyes. This observation is consistent with our overall finding, as we would consider GA to be a more advanced or later stage even compared to MNV, as MNV is now considered as a “last-ditch” response to try to prevent the development of atrophy. It should be noted, however, that thickness is only one metric of disease-related alterations. It is possible that there may have been a compensatory proliferation of other cells (e.g., glial cells) during the remodeling process that may have mitigated the observed magnitude of thinning in intermediate AMD eyes.

Our study has some limitations which need to taken into account when assessing our findings. First, although the Amish Eye Study was a prospective population study, the present analysis is post hoc and thus may still be subject to ascertainment bias. Another limitation is that the normal subjects in our population were younger than the AMD subjects. This is perhaps not surprising since AMD is an age-associated disease, but age itself can have an impact on GCC thickness. On the other hand, we did adjust for age in our statistical analysis. We also did not study the impact of other AMD biomarkers such as subretinal drusenoid deposits, intraretinal hyperreflective foci, drusen volume and choroidal thickness on GCC thickness measurements. Finally, our analysis was cross-sectional, and thus while we observed differences in GCC thickness based on the presence and stage of AMD, we could not specifically demonstrate a decline in GCC thickness as AMD developed and progressed in individual subjects.

In summary, we observed that the ganglion cell complex was significantly thinner in AMD eyes compared to healthy eyes in this presumably genetically and environmentally uniform Amish population. Future therapeutic strategies for AMD may need to consider these inner retinal alterations and their potential impact on response to therapy.