Optical Coherence Tomography Risk Factors for Progression to Late-Stage Age-related Macular Degeneration in the Amish Eye Study

Abstract

Purpose:

To characterize the prevalence of some critical optical coherence tomography (OCT) biomarkers—including cuticular drusen and acquired vitelliform lesions (AVLs)—and to identify their relative risk for progression to late age-related macular degeneration (AMD) over two years in subjects with early or intermediate AMD in the Amish Eye Study.

Design:

Prospective, observational, longitudinal, population-based cohort study.

Participants:

This study included 276 eyes from 171 subjects with early or intermediate AMD at baseline who completed the two-year follow-up.

Methods:

Baseline OCT scans were evaluated for the presence of: cuticular drusen, AVLs, subretinal drusenoid deposits (SDD), high drusen volume (defined as ≥0.2mm³ within the central 5mm), intraretinal hyperreflective foci (IHRF), hyporeflective drusen cores (hDC), thick/thin double-layer sign, and incomplete retinal pigment epithelium and outer retinal atrophy (iRORA). Subfoveal choroidal thickness (SFCT) was also measured.

Main Outcome Measures:

Incidence of late AMD (geographic atrophy or macular neovascularization) at two years as determined by multimodal imaging (OCT, color fundus photography (CFP), and confocal fundus autofluorescence (FAF)).

Results:

By 2-years of follow-up, 26 eyes (10.7%) progressed to late AMD. The most prevalent baseline features in this cohort, in descending order, were cuticular drusen (52.3%), IHRF (17.3%), hDC (16.0%), thin DLS (11.9%), SDD (8.2%), iRORA (7.8%), AVL (7.0%), and high drusen volume (2.5%). The mean SFCT was 243.23 ± 75.45 μm. Univariate analysis demonstrated that the presence of thick DLS, iRORA, AVL, SDD, IHRF, hDC, and SFCT were associated with increased risk of progression. In multivariate regression, however, only the presence of iRORA (OR: 29.60; 95% CI: 6.86–127.84; p < 0.001) and AVL (OR: 15.90; 95% CI: 3.24–78.00; p < 0.001) remained significant, while the presence of IHRF showed borderline significance (OR: 4.71; 95% CI: 1.00–22.16; p = 0.050).

Conclusions:

In this cohort, the presence of iRORA and AVL was independently associated with progression to late AMD over 2 years. Although cuticular drusen were highly prevalent, their presence, as assessed in this study, was not significantly associated with an increased risk of progression to late AMD.

Introduction

Age-related macular degeneration (AMD) is the most common cause of severe vision loss in older adults, particularly in developed countries. As of 2020, approximately 8.7% of people aged 45 and older worldwide have some form of AMD, affecting an estimated 200 million individuals globally. This number is projected to increase to 288–300 million by 2040, driven by the overall aging of populations^{[1, 2]}.

Nonexudative or dry AMD, which can be associated with gradual atrophy of the retinal pigment epithelium (RPE) and photoreceptors leading to irreversible vision loss in its later stages, accounts for 85–90% of all AMD cases^[3]. In the United States, nearly 11 million individuals—around 85% of all AMD cases—are diagnosed with dry AMD, while approximately 1.5 million, or 15%, experience its advanced stages^[4]. Despite its prevalence and significant impact, dry AMD has historically been under-researched, largely due to its perceived slow progression and lack of effective treatments. However, advancements in imaging technologies and growing attention from pharmaceutical pipelines have highlighted the need for deeper investigation. Along with exudative or wet AMD, clarifying its natural course and identifying high-risk subgroups with faster progression are crucial steps toward improving risk stratification, accelerating therapeutic development, and enhancing patient outcomes.

According to the Beckman classification of AMD and its associated prognostic analysis, the estimated five-year risk of progression to late-stage AMD is 3.1% for individuals with early AMD and 47.3% for those with intermediate AMD^[5]. Data from the Age-Related Eye Disease Study (AREDS) and AREDS2 Research Groups further support these findings, reporting that among participants with bilateral medium drusen at baseline, 70.9% progressed to large drusen and 13.8% to advanced AMD over a 10-year period^[6]. A more recent report from the same group indicated that the five-year progression rate to late AMD was 4.5% for individuals with level 1 (early AMD), and ranged from 12.9% to 32.3% for levels 2–3 (intermediate AMD)^[7]. These findings underscore the importance of studying AMD progression across different disease stages to better understand the full disease spectrum.

Previous studies for optical coherence tomography (OCT) biomarkers associated with disease progression have identified intraretinal hyperreflective foci (IHRF)^[8–11], subretinal drusenoid deposits (SDD)^{[8, 12–14]}, high central drusen volume^{[8, 13]}, and hyporeflective drusen cores (hDC)^{[9, 10, 13, 15, 16]} as potential risk factors. More recently, thin double-layer sign (DLS) has been associated with an increased risk of progression to complete retinal pigment epithelium and outer retinal atrophy (cRORA) over two years^[17], while thick DLS has been linked to progression from intermediate AMD to exudative macular neovascularization (MNV)^[18]. However, the relevance of other putative biomarkers such as cuticular drusen and acquired vitelliform lesions (AVLs) in the progression to late AMD remain unclear, particularly in population-based cohorts.

The Amish Eye Study is a prospective, population-based cohort study that included 1,124 eligible participants at baseline. Among this cohort, 46% of eyes were classified as having AMD (defined by the presence of at least 5 medium-sized drusen on color fundus photography), while 49% of eyes showed no signs of AMD^[19]. Over a two-year follow-up period, analysis of the overall cohort (including both AMD and non-AMD eyes) identified drusen volume within the central 3 mm ring (OR: 2.59, p = 0.049) and the presence of IHRF (OR: 57.06, p < 0.001) as significant risk factors for progression to late AMD^[20]. However, the impact of these OCT biomarkers specifically within eyes already diagnosed with early or intermediate AMD remains uncertain and is the question most relevant for risk stratification. Furthermore, newer biomarkers such as thin DLS, AVL, iRORA, and cuticular drusen were not considered in that analysis. This study aims to examine the prevalence of these OCT biomarkers in eyes with early or intermediate AMD within the Amish Eye Study cohort and to evaluate their relative risk of progression to late AMD over a two-year period.

Methods/Materials

The Amish Eye Study is a clinical research initiative supported by the National Eye Institute (NEI), designed to investigate genetic and phenotypic factors associated with AMD and other ocular traits in the genetically isolated Amish population. Participants were recruited from the three largest Amish communities in Ohio, Pennsylvania, and Indiana. This study follows a prospective, observational, longitudinal, population-based design. The inclusion criteria required participants to be ≥50 years old, self-identify as Amish, and belong to a sibship with at least one individual diagnosed with AMD. Individuals with pre-existing or concurrent ophthalmological conditions that could confound the interpretation of AMD-related imaging features were excluded. Data collection and imaging protocols have been detailed in previous Amish Eye Study reports^{[19, 20]}. The study was conducted in compliance with the Declaration of Helsinki, and written informed consent was obtained from all participants. To date, baseline and 2-year follow-up data collection have been completed, and the 7-year follow-up phase is currently ongoing. In total, 1,339 participants (2,668 eyes) were enrolled at baseline.

For the current study, inclusion criteria were: (1) subjects who were from the Amish Eye Study cohort and completed both baseline and 2-year follow-up visits, and (2) subjects with at least one eye diagnosed with early or intermediate-stage AMD at the baseline visit, with AMD staging determined based on the Beckman classification using color fundus photography (CFP)^[5]. Exclusion criteria are eyes with missing imaging data or poor image quality that prevented accurate grading.

Image acquisition

Participants included in the analysis underwent comprehensive multimodal imaging, which included stereo color fundus photography (CFP), confocal fundus autofluorescence (FAF; Spectralis), infrared reflectance (IR; Spectralis), and dense-volume OCT. OCT imaging was performed using both the Cirrus HD-OCT (Carl Zeiss Meditec, Dublin, CA, USA) and the Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany). The Cirrus OCT utilized a macular cube protocol consisting of 128 horizontal B-scans with 512 A-scans per line (6 × 6 mm, centered on the fovea). The Spectralis OCT used a macular cube protocol of 97 horizontal B-scans and 512 A-scans per line, as well as averaging by automatic real-time tracking (ART) of 9 (6 × 6 mm, foveal centered, and EDI mode off). The study included two OCT imaging devices to capitalize on their distinct strengths. The Cirrus was utilized for its capability to perform automated drusen volume quantification, while the Spectralis offered enhanced visualization of outer retinal layers and the choroid through its averaged scans.

Grading procedures

All OCT images from each included case were independently reviewed and graded by two certified graders (Y.C. and M.A., or N.M. and S.V.). In cases where discrepancies arose between the graders, a final decision was made by a senior retina specialist (S.S.). Baseline OCT images were evaluated for the presence of the following features: thin DLS, thick DLS, incomplete RORA (iRORA), cuticular drusen, AVL, IHRF, high drusen volume (defined as drusen volume ≥0.2 mm³ within a 5 mm diameter circle), hDC, and SDD. Subfoveal choroidal thickness (SFCT) was also measured as a continuous variable. Cases with insufficient image quality to allow grading of even one of these features were excluded. Figure 1 illustrates representative OCT B scans of the various evaluated biomarkers.

Figure 1.

Representative images of various OCT biomarkers. A, Cuticular drusen, located beneath retinal pigment epithelium (RPE), are characterized by blunted triangular elevations in a sawtooth configuration. B, Acquired vitelliform lesion (AVL), hyperreflective material between the RPE and photoreceptor layer, often appearing as a dome-shaped elevation with homogeneous reflectivity. C, Incomplete retinal pigment epithelium and outer retinal atrophy (iRORA), characterized by discontinuous hypertransmission band, irregular RPE band and overlying photoreceptor degeneration. D, Intraretinal hyperreflective foci (IHRF), defined as hyperreflective lesions located above RPE band. E, Thick double-layer sign (thick DLS), visible separation of RPE and Bruch’s membrane, occupied by at least two zones of varying reflectivity. F, Thin double-layer sign (thin DLS), single zone of low-to-medium reflectivity. G, Hyporeflective drusen cores (hDC), presence of hyporeflective spaces within drusen, with a minimum height of 40 μm. H, Subretinal drusenoid deposits (SDD), mounds or cone-shaped structures located above RPE.

As described previously, the double-layer sign is characterized by the visible separation of two highly reflective layers: the retinal pigment epithelium (RPE) band and Bruch’s membrane^{[17, 18]}. Thin DLS is identified when a single zone of low-to-medium reflectivity occupies the space between Bruch’s membrane and the RPE. In contrast, thick DLS is defined when multiple layers, consisting of at least two distinct zones with varying reflectivity, are observed between the RPE and Bruch’s membrane. iRORA, as defined by the Classification of Atrophy Meetings (CAM) group^[21], is characterized on OCT by discontinuous signal hypertransmission into the choroid, attenuation or disruption of the RPE, and overlying photoreceptor degeneration. Cuticular drusen are classically described as blunted triangular elevations in a sawtooth configuration^[22]. but more recently their internal hyporeflectivity and associated hypertransmission tail have been emphasized as key features in their recognition. All three morphologic patterns of cuticular drusen were included in this study's definition^[23]. AVLs are characterized on OCT by hyperreflective material between the RPE and photoreceptor layer, often appearing as a dome-shaped elevation with homogeneous hyperreflectivity of the subretinal deposit^[24], though they may appear hyporeflective in some areas, particularly in the superior portions on the lesion. IHRF were defined as hyperreflective lesions located within the neurosensory retinal layers (i.e. above the RPE band), measuring at least three pixels in size, and exhibiting reflectivity comparable to that of the RPE band^[25]. Drusen volume was calculated using Cirrus Advanced RPE analysis software (version 6.0) with a cutoff value of 0.2 mm³ within a 5 mm circle to be considered high. This area was chosen because most drusen are located within the 5 mm circle, whereas the 3 mm circle might underestimate volume^[26]. Additionally, the 5 mm circle provides greater reproducibility as it is less affected by automated foveal center localization. hDC lesions were defined by the presence of hyporeflective spaces within drusen, with a minimum height of 40 μm, as described in prior studies^[27]. SDD were characterized as mounds or cone-shaped structures with medium to high reflectivity, located either at the ellipsoid zone or between the ellipsoid zone and the RPE surface^[28]. Subfoveal choroidal thickness was determined on the central B-scan using the caliper tool in the Heidelberg Spectralis platform to measure the distance between the Bruch’s membrane and the inner scleral surface^[20].

Outcome measurement

The primary outcome of this study was the development of late AMD at the 2-year follow-up. Early/intermediate to late AMD progression was identified based on evaluation of multimodal imaging, including CFP, OCT, or FAF. According to the Beckman classification^[5], late AMD is defined as (1) the presence of MNV, determined by the Consensus on Neovascular AMD Nomenclature (CONAN) criteria^[29], or (2) the occurrence of geographic atrophy (GA) or cRORA as detected on CFP or OCT^[30], or areas of decreased autofluorescence observed on FAF^[31].

Statistical analysis

The statistical analysis was performed using IBM SPSS Statistics (Version 29.0.0). Continuous variables were summarized as means with standard deviations (SD), while categorical variables were presented as counts and percentages. Univariate analysis was conducted to assess the association between baseline OCT features and the development of late AMD, with odds ratios (OR) and 95% confidence intervals (CI) calculated for each variable. Binary logistic regression was then used in the multivariate analysis to assess the independent effects of the selected variables on the risk of late AMD. Variables with a p-value > 0.1 in univariate analysis were excluded from the multivariate regression model. Statistical significance was determined at a two-sided p-value < 0.05.

Results

A total of 1,332 eyes from 666 consecutive subjects completed the two-year follow-up and served as the source population. Among them, 276 eyes from 171 subjects had early to intermediate AMD at baseline, with 133 eyes classified as early AMD and 143 eyes as intermediate AMD. Of these, 243 eyes were included in the final analysis cohort after excluding cases with at least one ungradable variable. At the 2-year follow-up, 26 eyes (10.7%) from 22 participants progressed to late AMD. The flowchart for participants selection in this cohort is shown in Figure 2. Among the 26 eyes that progressed to late AMD, 3 eyes from 3 participants developed MNV, and the remaining 23 eyes from 19 participants developed cRORA.

Figure 2.

Flowchart of Participants Selection and AMD Progression Outcomes. AMD= age-related macular degeneration; CFP=color fundus photography; FAF= fundus autofluorescence; OCT= optical coherence tomography.

Among these 243 eyes, thin DLS was observed in 11.9% (29 eyes), while thick DLS was present in 3.3% (8 eyes). iRORA was identified in 7.8% (19 eyes). Cuticular drusen were the most prevalent feature, detected in 52.3% (127 eyes). AVL were found in 7.0% (17 eyes), and SDD were observed in 8.2% (20 eyes). High drusen volume was noted in 2.5% (6 eyes). IHRF were present in 17.3% (42 eyes), and hDC were identified in 16.0% (39 eyes). The mean SFCT was 243.23 ± 75.45 μm (Table 1).

Table 1.

Prevalence of the Various OCT Features at Baseline

Baseline OCT Variables	Frequency (%) / Mean ± SD
Eyes	243
Thin DLS	29 (11.9)
Thick DLS	8 (3.3)
iRORA	19 (7.8)
Cuticular drusen	127 (52.3)
AVL	17 (7.0)
SDD	20 (8.2)
High Drusen Volume	6 (2.5)
IHRF	42 (17.3)
hDC	39 (16.0)
SFCT	243.23 ± 75.45 μm

AVL = acquired vitelliform lesion; DLS = double-layer sign; hDC = hyporeflective drusen core; High drusen volume = drusen volume ≥0.2mm³ in central 5mm; IHRF = intraretinal hyperreflective foci; iRORA = incomplete retinal pigment epithelium and outer retinal atrophy; SD = standard deviation; SDD = subretinal drusenoid deposit; SFCT=subfoveal choroidal thickness

In the univariate regression analysis (Table 2), several baseline OCT features were significantly associated with an increased risk of developing late AMD at the 2-year follow-up. Thick DLS exhibited an odds ratio (OR) of 9.68 (95% CI: 2.26–41.43, p = 0.002). iRORA demonstrated a strong association, with an OR of 25.71 (95% CI: 8.75–75.55, p < 0.001). AVLs were also significantly associated with progression (OR: 4.07, 95% CI: 1.31–12.67, p = 0.015), as were SDD (OR: 5.78, 95% CI: 2.06–16.23, p < 0.001), IHRF (OR: 6.48, 95% CI: 2.74–15.36, p < 0.001), and hDC (OR: 3.30, 95% CI: 1.35–8.08, p = 0.009). Additionally, lower SFCT was associated with progression (OR: 0.99, 95% CI: 0.98–1.00, p = 0.001). Notably, cuticular drusen were not associated with risk for progression.

Table 2.

Baseline OCT Features as Risk Factors for the Development of Late Age-Related Macular Degeneration in the Univariate and Multivariate Regression Model

	Univariate Analysis				Multivariate Analysis
	Odds Ratio	95% CI	P		Odds Ratio	95% CI	P
Thin DLS	2.53	0.92 – 6.94	0.07		2.15	0.58 – 7.97	0.25
Thick DLS	9.68	2.26 – 41.43	0.002		1.13	0.10 – 12.58	0.92
iRORA	25.71	8.75 – 75.55	<0.001		29.60	6.86 – 127.84	<0.001
Cuticular Drusen	0.90	0.40 – 2.04	0.81		-	-	-
AVL	4.07	1.31 – 12.67	0.015		15.90	3.24 – 78.00	<0.001
SDD	5.78	2.06 – 16.23	<0.001		3.01	0.56 – 16.14	0.20
High Drusen Volume	4.44	0.77 – 25.51	0.095		10.54	0.42 – 267.45	0.15
IHRF	6.48	2.74 – 15.36	<0.001		4.71	1.00 – 22.16	0.050
hDC	3.30	1.35 – 8.08	0.009		1.01	0.25 – 4.05	0.99
SFCT	0.99	0.98 – 1.00	0.001		1.00	0.99 – 1.01	0.62

AVL = acquired vitelliform lesion; CI = confidence interval; DLS = double-layer sign; hDC = hyporeflective drusen core; High drusen volume = drusen volume ≥0.2mm3 in central 5mm; IHRF = intraretinal hyperreflective foci; iRORA = incomplete retinal pigment epithelium and outer retinal atrophy; SDD = subretinal drusenoid deposit; SFCT=subfoveal choroidal thickness.

Bold values denote statistical significance (P < 0.05).

In the multivariate regression model, iRORA remained a strong independent risk factor for progression to late AMD, with an OR of 29.60 (95% CI: 6.86–127.84, p < 0.001). AVLs were also significantly associated with progression (OR: 15.90, 95% CI: 3.24–78.00, p < 0.001). IHRF showed borderline significance (OR: 4.71, 95% CI: 1.00–22.16, p = 0.050). However, thick DLS, SDD, high drusen volume, and hDC did not retain statistical significance in the multivariate analysis. SFCT was not independently associated with late AMD progression (OR: 1.00, 95% CI: 0.99–1.01, p = 0.62).

Since iRORA is a very strong and established risk factor for progression to cRORA^[21], all analyses were repeated after excluding iRORA (Table 3). In the multivariate analysis, AVL (OR: 7.94, 95% CI: 1.88–33.49, p = 0.005) remained statistically significant. IHRF approached significance (OR: 3.88, 95% CI: 0.96–15.76, p = 0.058) but did not reach the threshold for statistical significance. Other factors, such as thick DLS, SDD, hDC, and SFCT, were not significant.

Table 3.

Baseline OCT Features (Excluding Incomplete Retinal Pigment Epithelium and Outer Retinal Atrophy) as Risk Factors for the Development of Late Age-Related Macular Degeneration in the Univariate and Multivariate Regression

	Univariate Analysis				Multivariate Analysis
	Odds Ratio	95% CI	P		Odds Ratio	95% CI	P
Thin DLS	2.53	0.92 – 6.94	0.07		1.67	0.48 – 5.89	0.42
Thick DLS	9.68	2.26 – 41.43	0.002		3.75	0.51 – 27.49	0.19
Cuticular Drusen	25.71	8.75 – 75.55	<0.001		-	-	-
AVL	0.90	0.40 – 2.04	0.81		7.94	1.88 – 33.49	0.005
SDD	4.07	1.31 – 12.67	0.015		2.84	0.60 – 13.44	0.19
High Drusen Volume	5.78	2.06 – 16.23	<0.001		8.19	0.57 – 118.58	0.12
IHRF	4.44	0.77 – 25.51	0.095		3.88	0.96 – 15.76	0.058
hDC	6.48	2.74 – 15.36	<0.001		1.18	0.34 – 4.03	0.80
SFCT	3.30	1.35 – 8.08	0.009		1.00	0.99 – 1.00	0.28

AVL = acquired vitelliform lesion; CI = confidence interval; DLS = double-layer sign; hDC = hyporeflective drusen core; High drusen volume = drusen volume ≥0.2mm3 in central 5mm; IHRF = intraretinal hyperreflective foci; SDD = subretinal drusenoid deposit; SFCT=subfoveal choroidal thickness.

Bold values denote statistical significance (P < 0.05).

Discussion

In this study of Amish eyes with early or intermediate AMD, the presence of iRORA and AVL was identified as strong independent risk factors for progression to late AMD, while the presence of IHRF showed borderline statistical significance. Cuticular drusen were highly prevalent in this cohort, present in more than half of the eyes (52.3%); however, their presence, as graded in this study, was not significantly associated with progression to late AMD.

This finding regarding the prevalence of cuticular drusen in this cohort is notable. In contrast to our results, Goh et al.^[22] examined a cohort of 280 eyes from 140 participants with bilateral large conventional drusen, meeting the definition of intermediate AMD, in which the prevalence of cuticular drusen was 25%. The significantly higher prevalence observed in our study cohort suggests a unique characteristic of the Amish population.

Regarding its association with late AMD, both Goh et al.’s study and our investigation did not demonstrate a statistically significant association between cuticular drusen and progression to late AMD over a relatively short follow-up period of 2–3 years. In our regression analysis, cuticular drusen showed an odds ratio (OR) of 0.90 (95% CI: 0.40–2.04, p = 0.81),indicating no significant prognostic effect.. Similarly, in Goh et al.’s multivariate model, cuticular drusen showed a hazard ratio (HR) of 0.88, with a non-significant p-value of 0.784. Importantly, the lack of statistical significance in these analyses does not imply absence of prognostic relevance, but rather reflects the limited evidence within the constraints of the study design and follow-up duration. On the contrary, other longitudinal studies focusing on patients with cuticular drusen have reported an association with late AMD. For instance, Sakurada et al.^[32] assessed the 5-year risk of geographic atrophy (GA) and macular neovascularization (MNV) in eyes with cuticular drusen, reporting cumulative risks of 28.4% and 8.7%, respectively. Likewise, Yoon et al.^[33] observed a 15.4% incidence of MNV over a mean follow-up of 6.8 ± 0.9 years in AMD eyes exhibiting both cuticular drusen and reticular pseudodrusen (RPD). Taken together, these findings suggest that the prognostic significance of cuticular drusen in general AMD populations remains inconclusive and may depend on cohort characteristics, coexisting phenotypes, and duration of follow-up. Further studies incorporating larger sample sizes, longer longitudinal observation, and refined phenotypic stratification are warranted to better elucidate the role of cuticular drusen in late AMD progression.

In both our study and previous studies, iRORA is strongly associated with progression to late AMD. According to the CAM group consensus^[21], iRORA is recognized as a precursor OCT biomarker and often precedes cRORA. In a cohort study of 135 patients with intermediate AMD, approximately 31% of the variability in AMD progression was attributed to iRORA, with patients exhibiting a 12-fold increased risk of developing late AMD compared to those without iRORA^[34]. In our study, which included 119 eyes with early AMD and 124 eyes with intermediate AMD, iRORA was associated with a 30-fold increased risk of developing late AMD. Figure 3 illustrates that the proportion of progression to late AMD is similar among those with iRORA, regardless of baseline AMD stage.

Figure 3.

Proportions and Numbers of Subjects Progressed to Late AMD Based on the Presence of iRORA at Baseline

Our study also identifies AVL as a significant risk factor for late AMD in this population, marking the first report of this association in the Amish Eye Study. Even after adjusting for iRORA—a well-established and strong predictive variable—AVL remained a significant risk factor in the multivariate regression analysis.

Histopathologic findings reveal that AVL are primarily composed of flocculent material (57–59%), with smaller contributions from RPE organelles (3–22%), outer segments (2–10%), and lipid droplets (0.2–12%). These characteristics share similarities with other AMD-related deposits, such as drusen and SDD, suggesting a dysregulation in lipid transfer pathways involving cone photoreceptors and their supporting cells^[35].

In a retrospective cohort study^[36] of 104 eyes from 63 patients with AVLs in AMD, choroidal neovascularization (CNV) developed in 12.4% of eyes, while 43.8% progressed to foveal atrophy—a significantly higher proportion compared to the other study arm with adult-onset foveo-macular vitelliform dystrophy (AOFVD). Similarly, another study^[35] examining 48 AVLs in 42 AMD eyes found that 50% of AVLs collapsed, leading to cRORA, while 2% developed neovascularization over an average follow-up of 46 months. Additionally, a recent study by Mahmoudi et al.^[37] of 163 eyes with intermediate AMD and AVL reported that over a two-year period, 10% of AVL lesions collapsed to cRORA, and 0.6% developed MNV.

These findings highlight that AVL should be recognized as an independent and important factor in the AMD disease progression. The results from our population-based cohort study further support this hypothesis.

This study has several strengths and limitations. A major strength is its population-based cohort design, which minimizes selection bias and offers a more representative sample of a general population compared to subgroup-specific studies. Our study includes ten OCT biomarkers related to AMD, including emerging biomarkers such as cuticular drusen and AVL, and employs multivariate regression models to compare their significance. Additionally, the image grading protocol is rigorous, with two independent, experienced, and certified graders evaluating all the OCT images under senior retinal specialist supervision to ensure accuracy. However, the study has limitations, including a relatively short two-year follow-up period, which may not fully capture the progression to late AMD. The ongoing seven-year follow-up will provide more comprehensive data on the natural course of the disease. Secondly, while the Amish Eye Study's genetically homogeneous and family-based cohort minimizes genetic background noise and environmental confounding factors, it also lacks ethnic diversity, limiting the generalizability of findings to other populations. Moreover, despite using multimodal imaging for the diagnosis of late AMD, baseline biomarker assessment relied solely on OCT. In addition, most OCT features were graded based on their presence or absence rather than quantitatively. This approach was chosen to provide a clinically intuitive framework for identifying eyes at higher risk, facilitating potential application in daily clinical practice. However, it may reduce sensitivity to subtle differences in feature severity and limit fine-grained risk stratification. Finally, although the extent of iRORA may vary considerably across eyes, and previous studies have suggested that the size of the choroidal hypertransmission defect may confer different risks for progression, the relatively small number of cases with conversion to late AMD and eyes with iRORA at baseline limited the statistical power for reliable quantitative analysis; therefore, qualitative findings were presented in this study.

In summary, in our prospective population-based Amish cohort, we found that cuticular drusen had a high prevalence in early and intermediate AMD. Additionally, iRORA and AVL were identified as risk factors for progression to late AMD over a two-year interval. These findings may facilitate risk prognostication and stratification in clinical practice and early intervention clinical trials. Future research could further explore whether different subtypes or spatial distributions of cuticular drusen have varying impacts on progression in this cohort and incorporate quantitative or volumetric OCT biomarker measurements to improve risk stratification. Additionally, using iRORA as an endpoint to investigate risk factors could provide a deeper understanding of the entire AMD disease spectrum.

Abbreviations and Acronyms

AMD

age-related macular degeneration

AVL

acquired vitelliform lesion

CFP

color fundus photography

cRORA

complete retinal pigment epithelium and outer retinal atrophy

DLS

double-layer sign

FAF

fundus autofluorescence

GA

geographic atrophy

hDC

hyporeflective drusen cores

IHRF

intraretinal hyperreflective foci

iRORA

incomplete retinal pigment epithelium and outer retinal atrophy

MNV

macular neovascularization

OCT

optical coherence tomography

RPE

retinal pigment epithelium

SDD

subretinal drusenoid deposits

SFCT

subfoveal choroidal thickness