In Vivo Quantification of Retinal Changes Associated With Drusen in Age-Related Macular Degeneration

Abstract

Purpose.: Drusen alters retinal architecture in early age-related macular degeneration (AMD). However, abnormalities also may exist in drusen-free areas of the AMD retina. This study examines retinal thickness above drusen relative to drusen-free areas in the same patient and a normal population.

Methods.: Patients with early to intermediate AMD (n = 122) or no disease (n = 30) were examined at the Center for Eye Health. Spectral domain optical coherence tomography (SD-OCT) scans through single, isolated druse (n = 125) or confluent drusen (n = 54) were obtained. The thickness of individual retinal layers was measured above the druse and in a drusen-free area, 150 μm from the drusen edge.

Results.: Intraeye comparisons found total retinal thickness above drusen was 16 ± 0.6% less than drusen-free areas. Thinning was mostly in the retinal pigment epithelium/photoreceptor layer (32 ± 1% reduction) and the outer nuclear layer (22 ± 1% reduction). Confluent drusen showed similar thinning of the outer retina as well as inner retina loss (5%). Thinning was strongly correlated with drusen height, but only modestly correlated with drusen width. When compared to the normal population, retinal thickness above drusen and drusen-free areas were significantly reduced.

Conclusions.: We confirm outer retina thinning above drusen in early/intermediate AMD compared to drusen-free areas in the same retina or a normal population. Interestingly, drusen-free areas in AMD patients were not the same as control patients suggesting “normal” areas of the AMD retina are abnormal. The strong correlation between retinal thinning and drusen height, rather than width, suggests current grading systems for AMD may need refinement.

Introduction

Age-related macular degeneration (AMD) is the leading cause of irreversible blindness and low vision in developed countries, affecting 6.5% of the United States population aged 40 years and older.^1–7 In the late stages of the disease, patients may experience permanent vision loss due to geographic atrophy or choroidal neovascularization.^8–11 Although dietary supplements can reduce the likelihood of AMD progression in some patients, no treatments currently are available to completely prevent progression to late stage AMD.^12,13 Understanding the retinal changes that occur during early AMD may be important for developing strategies that modify progression and to provide a better understanding of visual functional changes. 

A hallmark finding of early AMD is the presence of drusen within the macula region. Drusen are extracellular deposits of lipids, proteins, and other cellular material that build up between the RPE and the inner collagenous layer of Bruch's membrane.^14–16 Drusen characteristics on fundus photography, such as size and location, have important clinical implications, including predicting risk of AMD progression and effectiveness of dietary supplementation.^17–19 More recently, optical coherence tomography (OCT) has been used to describe cross-sectional characteristics of drusen and overlying retina, which may be associated with AMD progression.^20–24 Specifically, photoreceptor layer thinning and disruption of the inner segment ellipsoid (ISe) line above drusen has been linked to localized loss in visual function.^25–35 These changes, however, mostly have been quantified for large drusen (>125 μm) and, thus, may not be an ubiquitous feature of all drusen. 

Beyond drusen, there also is evidence of changes in the “normal” areas of the macula and peripheral retina of early AMD patients. Photoreceptor layer thickness has been reported to be abnormal in retinal areas immediately adjacent to drusen.²⁵ Retinal thickness across the AMD macula also is altered compared to the normal population, although this finding is not consistent.^23,36–38 Functional studies, such as multifocal electroretinography, indicate dysfunction of the peripheral retina in AMD patients.³⁹ Greater dysfunction of rod photoreceptors than cone photoreceptors, based on dark and light adaptation thresholds and histopathology, further suggest involvement of the peripheral retina in AMD.^40–42 Thus, assessing drusen-free areas of the central and peripheral retina of early AMD patients may be just as important as assessing drusen affected areas to characterize and understand the pathophysiology of AMD. 

This study determined the changes in retinal layer thickness that occur above drusen and at nearby “normal” drusen-free areas in early AMD. We used a population comparison (comparing retinal thickness of drusen and drusen-free areas in AMD patients to matching locations in a normal population) and an intraeye comparison (comparing retinal thickness of drusen and drusen-free areas to each other within the same AMD eye) to determine how early to intermediate AMD affects the retina at and away from the obvious sites of disease. We went beyond the scope of previous studies to encompass a variety of drusen sizes, including druplets (<63 μm), medium drusen (63–125 μm), and large drusen (>125 μm),¹⁸ and different drusen types (i.e., single isolated druse and confluent drusen) to determine if changes in retinal thickness above drusen and at adjacent, drusen-free areas, are linked to specific drusen characteristics. 

Methods

Study Population

The medical records of 122 patients with nonexudative AMD (aged 50 years and over) and 30 age-matched patients with no history of AMD, drusen, or any other retinal disease were reviewed. All patients were examined between 2010 and 2013 at the Center for Eye Health (CFEH) Sydney, Australia, a referral-only clinic that provides advanced diagnostic testing services and consultations by specially trained optometrists and ophthalmologists.⁴³ Patient written consent was obtained in accordance with the Declaration of Helsinki and approved by the Biomedical Human Research Ethics Advisory Panel of the University of New South Wales. 

Patients were included in the study if their records included current demographic information (specifically age, sex, best corrected visual acuity), a color fundus photograph for each eye with the corresponding Age-Related Eye Disease Study (AREDS) simplified scale score,¹⁷ a macular cube OCT scan (comprised of 50–61 horizontal line scans through the macula) of the right eye obtained by the Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany) and a macula 512 × 128 cube scan (128 B-scans, each comprised of 512 A-scans) obtained by the Cirrus HD-OCT (Carl Zeiss Meditec, Jena, Germany). All patient records and images had been reviewed independently by at least two experienced clinicians in concordance with CFEH's diagnostic protocols.⁴³ The characteristics of the final patient cohort are summarized in Table 1. 

Table 1

 

Characteristics of Normal and AMD Cohorts

Characteristic	Normal	AMD, Single Druse	AMD, Confluent Druse
Demographics
No of subjects	30	47	23
Male	16	31	11
Female	14	16	12
Mean age ± SD, y	65 ± 12	69 ± 9.3	73 ± 8.7
Age range, y	44–92	44–88	59–88
Median BCVA	20/20	20/20	20/24+1
BCVA range	20/16−20/40	20/12.5–20/63	20/20–20/40
Ethnicity, n (%)
Caucasian	24 (80)	45 (96)	24 (100)
Asian	6 (20)	2 (4)
AREDS score, n (%)
1		8 (17)	2 (10)
2		26 (55)	10 (43)
3		3 (6)	1 (4)
4		10 (21)	10 (43)
Drusen characteristics
Number of lesions measured		125	49
Druplets (<63 μm)		5	0
Medium drusen (63–125 μm)		49	0
Large drusen (>125 μm)		71	49

Definitions of Drusen Type

Drusen were classified as either isolated druse (singular dome-shaped drusen deposit surrounded by an area of retina that was free of distortions from adjacent lesions) or confluent drusen (areas of multiple drusen peaks whose borders overlap and merge into one another). Reticular pseudodrusen, defined as drusenoid lesions appearing above the RPE, were excluded. 

Image Selection

Individual horizontal line scans from patient right eye macular cube OCT scans were reviewed and 125 isolated druse in 47 eyes and 49 confluent drusen from 26 eyes were suitable for analysis. A total of 14 eyes was measured for both drusen types so that 45 of 125 isolated druse and 30 of 49 confluent druse were from eyes containing both drusen types. All data were pooled as there was no significant difference between the range of drusen width or height between the eyes with only one druse type and those with both drusen types. Drusen were selected only if the individual horizontal scans were of sufficient image quality to allow the specific retinal layers measured to be clearly distinguishable from one another without interference from features, such as pigment migration or hyper-reflective areas.²⁷ For drusen that spanned multiple horizontal line scans, the scan showing the largest horizontal diameter of the lesion was used. For accuracy, drusen were excluded if they were <20 μm in height or <500 μm from the foveal center. To avoid subject bias, no more than eight isolated druse were measured in a single eye (limit of two per quadrant) and no more than four confluent drusen were measured in a single eye (one per quadrant). For eyes with multiple acceptable drusen, the most clearly delineated lesions were selected. Drusen were identified by an experienced clinician (JR) and a subset were verified by a second experienced clinician with a subspecialty in medical retina ophthalmology (NA). 

Quantitative Measurements

Measurements were made using the “measure distance” tool provided within the Heidelberg Eye Explorer Spectralis Viewing Module software in the 1:1 scale viewing format. For each druse, location was determined from the infrared confocal scanning laser ophthalmoscope (IR-cSLO) image of the posterior pole accompanying the macular cube OCT scan (Fig. 1). Eccentricity was determined as the distance from the foveal center to the center of each lesion and lesions were designated according to quadrant (superotemporal [ST], superonasal [SN], inferonasal [IN], inferotemporal [IT]) and anatomical location (fovea = < 500 μm of foveal center, parafovea = 500–1250 μm from foveal center, perifovea = 1250–2750 μm from foveal center, outer macula = more than 2750 μm from foveal center; Fig. 1A). 

Figure 1

 

Quantification method for drusen in AMD eyes. (A) Color fundus image indicating the regions defined in this study in white. This innermost circle is the fovea (<500 μm from foveal center), followed by the parafovea (500–1250 μm from the center of the fovea), perifovea (1250−2750 μm from the center of the fovea), and the outer macula (>2750 μm). The eccentricity of drusen was determined as the shortest diagonal distance from the fovea to the center of the drusen with reference to the relevant quadrant as noted. (B) For OCT measurements of isolated druse, drusen height (h) was determined at the highest point of the drusen and drusen width (w) was determined from either end of the drusen. Comparative measurements in druse-free areas were made 150 μm temporal and nasal to the edge of each drusen deposit. (C) Identical measurements for confluent drusen were made 75 and 225 μm from the edge of the drusen border and the comparative measure was made 150 μm away from the same drusen border. (D) The total thickness of the retina was determined as the distance from the RPE layer to the bottom surface of the NFL. The RPE/PR layer (yellow) was determined as the distance between the bottom of the hyper-reflective RPE band and the top of the external limiting membrane. The ONL (pink), OPL+INL (blue), and IPL+GCL (beige) were measured from the anterior border of the previous layer to the posterior border of the subsequent layer.

For each isolated druse, the horizontal width and vertical height (measured from the center of the base of the druse to the overlying RPE) was determined (Fig. 1B). Individual layer thickness then was quantified for the RPE/photoreceptor (PR) layer as the height from the posterior border of the RPE to the middle of the external limiting membrane (ELM), the outer nuclear layer (ONL) as the height from the middle of the ELM to the posterior border of the outer plexiform layer (OPL), the OPL/inner plexiform layer (INL) as the height from the posterior border of the OPL to the inner plexiform layer (IPL), and the IPL/ganglion cell layer (GCL) as the height from the posterior border of the IPL to the nerve fiber layer (NFL; Fig. 1D). Total retinal thickness was measured from the posterior border of the RPE to the NFL. Corresponding measurements were made 150 ± 2 μm to the horizontal left (temporal) and right (nasal) of the lesion, and averaged to represent “normal” drusen-free areas of retina within AMD patients (Fig. 1C). Confluent drusen were measured in an identical manner, but at 75 ± 2 μm and 225 ± 2 μm inward from the edge of the lesion (Fig. 1C). Measurements of “normal” drusen-free areas near confluent drusen were performed 150 ± 2 μm away from the lesion edge closest to the intradrusen measurements only, since in most instances the confluent drusen spanned beyond the OCT scan making measurements on both sides unobtainable within a single scan. 

Total macular thickness from the RPE to the internal limiting membrane (ILM) also was assessed via an automated analysis of a macula 512 × 128 cube scans obtained with the Cirrus HD-OCT. Patients were considered to have macular thinning or thickening if any Early Treatment of Diabetic Retinopathy Study (ETDRS) subfield fell within the lower or upper 5% of the normative population (Table 2). 

Table 2

 

Classification of Macular Thickness by Automated OCT Analysis

Drusen Type	Macular Thickness Grading, n(%)
	Within Normal Limits	Thinner Than Normal	Thicker Than Normal	Thinner and Thicker Than Normal
Isolated drusen	28 (60)	10 (21)	7 (15)	2 (4)
Confluent drusen	10 (47)	5 (24)	5 (24)	1 (5)

 For each patient, automated macular thickness analysis was performed using Zeiss Cirrus OCT cube scans that were classified as normal if the ILM to RPE thickness for all ETDRS subfields was within the limits of the normative population. Patients were classified as having a thinner or thicker than normal macula if any ETDRS subfield was outside the normative range.

Normative data of total retinal layer thickness from control subjects was obtained by measuring the retinal thickness of seven horizontal OCT line scans covering this area of the macula cube scan at 500 to 5000 μm radial distance from the fovea. Quantitative measurements were obtained by an experienced clinician (JR) and two nonclinicians (LNS, CZ) unaware of the purpose of the study. The measurements were then cross-checked by each observer. 

Statistical Analysis

Statistical analysis between paired samples was performed by a Student's two-tailed t-test in Graphpad Prism (GraphPad Software, Inc., La Jolla, CA, USA). Correlations were determined using a two-tailed Spearman's correlation with 95% confidence interval. Significance was assumed for a P < 0.05. 

Results

Drusen Selection Criteria

The criteria used to identify drusen showed no bias in retinal location, with 32% of drusen located in the parafoveal region, 53% in the perifoveal region and 15% in the outer macula (Fig. 2A). The criteria also yielded a large range of drusen in terms of height (22–90 μm) and width (50–460 μm) across all locations (Figs. 2B, 2C). Similar results were seen for confluent drusen samples (data not shown). 

Figure 2

 

Characteristics and distribution of isolated druse. (A) The retinal location of all isolated druse measured in this study showing a random sampling of drusen locations. Across all retinal eccentricities, there was no relationship in (B) drusen width or (C) height.

Retinal Thickness Changes Associated With Isolated Druse in AMD Eyes

Total retinal thickness directly above drusen (Fig. 3, black squares) was significantly less than the total retinal thickness of age-matched control eyes (Fig. 3, shaded area) at matching eccentricities (Student's t-test, P < 0.05 or greater for all quadrants). This effect was greater in the perifoveal and outer macula than within the parafovea. Total retinal thickness of normal, drusen-free areas (Fig. 3, gray dots) also was significantly less than normal with a similar trend of increased retinal thinning with increased distance from the fovea (Student's t-test, P < 0.05 or greater in all quadrants). Interestingly, the total macular thickness of AMD patients as determined by the automated macular thickness analysis on Cirrus OCT found the majority of patients had macular thickness within normal limits for all ETDRS subfields (Table 2). Thinning or thickening of any macular subfields did not appear to be clearly linked with drusen distribution within the patient's macula (personal observation, data not shown). 

Figure 3

 

Total retinal thickness above isolated druse and adjacent, drusen-free areas compared to normative population. The total retinal thickness of drusen (gray circles) and drusen-free regions (150 μm nasal or temporal to the drusen edge. black squares) were compared to the retinal thickness of normal age-matched subjects at the matching eccentricities. Normative data are presented as the mean (white squares) ± SD (gray shading). Dotted lines delineate the parafovea, perifoveal, and outer macula. Asterisks in the individual foveal regions indicate a significant difference between the total retinal thickness above drusen (D) or drusen-free (DF) areas to the normal population. *P < 0.05, **P < 0.01, ***P < 0.001. ns, not significant.

When intraeye comparisons were made, the mean total retinal thickness above drusen was 16 ± 0.6% less than the total retinal thickness 150 μm away from the lesion border (Figs. 4A, 4F). When individual retinal layers were measured, this thinning was attributed mostly to the outer retina. Specifically, the RPE/PR layer directly above drusen displayed a 32 ± 1% reduction in thickness and the ONL displayed a 22 ± 1% reduction in thickness compared to the drusen-free area (Figs. 4B, 4C, 4F). The OPL/INL layer directly above drusen was only slightly thinner than in adjacent drusen-free areas (5% ± 1% reduction) and there was no significant change in IPL/GCL layer thickness (1 ± 1% increase; Figs. 4D–F). The quadrant location or eccentricity of drusen had no significant effect on retinal thickness ratio in any retinal layer (1-way ANOVA, P > 0.05 for all quadrant comparisons). Additionally, there were no significant differences in retinal thickness ratio of any retinal layer for drusen measured in patients with isolated druse only, or isolated and confluent drusen (data not shown). 

Figure 4

 

Change in retinal layers above drusen and drusen-free areas. The percentage difference between the (A) total retinal thickness, (B) RPE/PR layer thickness, (C) ONL thickness, (D) OPL+INL thickness, and (E) IPL + GCL thickness above drusen was compared to the thickness of matching retinal layers in drusen-free areas based on drusen eccentricity. Thickness of the drusen-free areas was averaged from measurements 150 μm temporal and nasal of the drusen. The line represents a linear regression fitted to the data and the values to the right of the regression line indicate the Spearman's correlation coefficient (r) and relative P value (P). As there was no clear relationship between retinal thickness change and eccentricity, data were pooled for each layer for all eccentricities to give the (F) mean thickness change ± SE mean for each retinal layer.

Thinning of the outer retinal layers above drusen was modestly correlated with drusen width (Spearman's correlation, RPE/PR, r = −0.29, P < 0.001; ONL, r = −0.24, P < 0.01; Figs. 5A–C). Drusen height, on the other hand, displayed a strong linear correlation with thinning of the RPE/PR layer (Spearman's correlation, r = −0.71, P < 0.001) and thinning of the ONL (Spearman's correlation, r = −0.68, P < 0.001; Figs. 5F, 5G) and a modest correlation with changes in the inner retina (OPL/INL, r = −0.36, P < 0.001; IPL/GCL, r = −0.24, P < 0.01; Figs. 5H–J). 

Figure 5

 

Changes in retinal thickness with drusen width and height. The percentage difference between (A, F) total retinal thickness, (B, G) RPE/PR layer thickness, (C, H) ONL thickness, (D, I) OPL+INL thickness, and (E, J) IPL + GCL thickness above drusen compared to the adjacent drusen-free area relative to (A–E) drusen width and (F–J) drusen height. The line represents a linear regression fitted to the data, and the Spearman's correlation coefficient (r) and relative P value (P) are noted in the top right of each graph.

Retinal Thickness Changes Associated With Confluent Drusen in AMD Eyes

Analysis of retinal thickness above confluent drusen yielded results similar to isolated druse (Supplementary Fig. S1). Total retinal thickness above both measurement locations (75 and 225 μm) within the confluent drusen was significantly decreased compared to the retinal thickness of age-matched controls within the perifoveal region (Student's t-test; P < 0.05 or less for all quadrants). A small decrease in retinal thickness was seen in the drusen-free areas, but significance was seen only in the ST quadrant (Student's t-test, P < 0.05). 

When retinal thickness at drusen and drusen-free areas was compared, total retinal thickness was significantly reduced above confluent drusen (Fig. 6). Analysis of individual retinal layers showed significant thinning of the RPE/PR at both drusen locations (75 μm, 19 ± 2 μm reduction; 225 μm, 15 ± 3 μm) and to a lesser degree, thinning of the ONL layer (75 μm, 14 ± 3 μm reduction; 225 μm, 11 ± 3 μm). The OPL/INL layer thickness between drusen and the adjacent, drusen-free, area was not significantly different. Interestingly, the IPL/GCL layer thickness was significantly reduced over confluent drusen compared to drusen-free areas (75 μm, 4 ± 1 μm reduction; 225 μm, 5 ± 2 μm). When the results for confluent drusen (Fig. 6, columns) were compared to the same measurements for isolated druse (dotted gray line), there was significantly less thinning of the RPE/PR, ONL, and total retinal thickness over confluent drusen than isolated druse (Student's t-test, P < 0.001), but significantly more thinning of the IPL/GCL layer over confluent drusen compared to isolated druse (Student's t-test, P < 0.001; Fig. 6) possibly indicating inner retinal remodelling in large confluent drusen. 

Figure 6

 

Change in retinal thickness between confluent drusen and adjacent, drusen-free area. The mean percentage difference ± SEM in retinal layer thickness at 75 and 225 μm from the edge of the confluent drusen compared to the adjacent, drusen-free area. The dotted lines indicate the mean change in thickness of the same layer for isolated druse from Figure 3F.

The correlation between drusen height and change in retinal thickness seen in isolated druse also was observed for confluent drusen (Fig. 7). Confluent drusen height was positively correlated with thinning of the RPE/PR layer overlying drusen at both measurement locations (Spearman's correlation, 75 μm, r = −0.57, P < 0.001; 225 μm, r = −0.37; P < 0.01; Fig. 7A) and this also was the case for the ONL (75 μm, r = −0.36, P < 0.05; 225 μm, r = −0.4, P < 0.05; Fig. 7B). No correlation was present for other layers, likely because there was little change in retinal thickness in these layers (Figs. 7C, 7D). The correlation of drusen height and retinal thickness remained true along the span of the confluent drusen (Figs. 7E–H). Specifically, when the change in drusen height from the 75 to 225 μm location was determined within the same confluent drusen, there was a corresponding change in RPE/PR layer thickness (Fig. 7E). This effect also was evident for the ONL (Fig. 7F), suggesting drusen height has the same effect on the overlying retina independent of its proximity to the edge of the lesion. 

Figure 7

 

Changes in retinal thickness of confluent drusen with drusen height. Change in (A) RPE/PR layer, (B) ONL, (C) OPL+INL, and (D) IPL + GCL thickness at over confluent drusen relative to drusen height at 75 μm (gray circles) and 225 μm (black squares) from the drusen edge. (E–H) Change in retinal thickness from the 75 to 225 μm location within the confluent drusen plotted relative to change in drusen height from the 75 to 225 μm location. Lines represent a linear regression fitted to the data, and the corresponding Spearman's coefficient (r) and P value (P) are in the top right of each graph.

Discussion

Photoreceptor Layer Thinning Is Ubiquitous to Drusen

This study confirmed that outer retinal layers overlying drusen are reduced in AMD. We found a 32% loss in the RPE/PR layer thickness and a 22% loss in ONL thickness over drusen compared to adjacent drusen-free regions in our AMD patient cohort. Thinning of the outer retina occurred regardless of drusen size (i.e., druplets, medium drusen, and large drusen) or drusen type (isolated and confluent), indicating this effect is ubiquitous to drusen in any form in AMD and, thus, possibly part of the underlying disease mechanism. This is supported by functional studies that show decreased photoreceptor layer thickness associated with drusen correlates with reduced rod sensitivity and visual function in AMD patients.^28,32 Alternatively, photoreceptor layer thinning above drusen could be a result of mechanical stress as retinal layers can appear structurally normal on color photography, fundus autofluorescence, and OCT images following drusen regression.^34,44 However, as there is no confirmation that areas of drusen regression retain normal function, photoreceptor layer thinning still may have a pathogenic role in AMD. 

Photoreceptor Layer Thinning Is Correlated With Drusen Height

We found outer retinal thinning above drusen of any size was linearly proportional to drusen height, even across confluent drusen when variation in drusen height was accounted for. Others have indicated previously a correlation between drusen height and retinal thinning, but only for large drusen (>125 μm diameter).^25,27,34 Retinal thinning was not well-correlated with drusen width, which has implications for current methodology, which stages and manages AMD using drusen diameter and number.¹⁷ Drusen height is not part of current AMD grading schemes as it cannot be assessed via color fundus photography. However, drusen quantification of AMD patients by OCT has been shown to correlate closely with current grading schemes,⁴⁵ and multiple studies emphasize the benefits of OCT for monitoring progression and treatment of late AMD.^46,47 Thus, drusen height (as measured by OCT) may be a potentially more informative part of future AMD grading schemes. Drusen volume also has been suggested as a useful indicator of AMD. However, we did not quantify drusen volume in the study as current algorithms only provide drusen volume across the whole retina and not for individual lesions.^21,23 

Thinning of Inner Retinal Layers May Indicate Greater Disease Severity

Significant thinning of the IPL/GCL was observed only for confluent drusen. This thinning may be due to the greater disease severity associated with confluent drusen as all confluent drusen in this study were large drusen, while only 57% of isolated druse were classified as large. Indeed, Schuman et al.²⁷ found no change in inner retinal layer thickness for early AMD patients with at least one drusen size of 125 μm, but global reductions in IPL/GCL thickness in AMD retina have been reported.³⁸ Changes in IPL/GCL thickness may reflect glial cell remodeling and neuronal migration, which has been shown in postmortem analysis of late, nonexudative AMD patients⁴⁸ and in other retinal diseases.^49–51 Thus, monitoring changes in the inner retina may be clinically useful for assessing drusen progression. 

Drusen-Free Areas of AMD Eyes Are Abnormal

We found the level of photoreceptor thinning above drusen determined by an intraeye comparison (i.e., thickness above drusen compared to thickness at adjacent drusen-free area) was similar to the values reported by others who compared photoreceptor layer thickness over drusen to matched normal populations.^{25–28,34,35,37} This suggested, intraeye comparisons are appropriate for drusen quantification and may be useful during clinical assessment when individual variations in retinal thickness due to ethnicity, age, patient history, or other ocular pathologies could obscure drusen-associated changes.^52,53 Furthermore, when we compared the total retinal thickness of drusen-free areas to a matched normal population, it was significantly reduced, indicating that the retinal architecture of drusen-free areas in AMD patients also is abnormal. Sadigh et al.²⁵ found thinning in normal areas adjacent to drusen in only 10% of their patient cohort, highlighting that thickening (20%) or no change (70%) in thickness occurred more frequently. This discrepancy may be due to differences in quantification as Sadigh et al.²⁵ measured changes in photoreceptor layer thickness, while this study quantified total retinal thickness. As the overall macular thickness of our patient cohort was mostly within normal limits, thinning seen in drusen-free areas did not cause significant changes throughout the posterior pole. Thus, thinning of drusen-free areas may be due to nearby drusen accumulation rather than generalized retinal dysfunction. Alternatively, since our patient cohort only had early to intermediate AMD, such changes may not yet have been detectable. No consistent trend in total macular thickness has been shown for early AMD, signifying that this measurement may not be ideal for quantifying the initial retinal changes in AMD.^23,36–38 Overall, our data suggested that, although retinal thinning clearly occurs above drusen in comparison with adjacent drusen-free areas of the same eye, these drusen-free areas are themselves significantly thinner than matching retinal areas in the normal population. 

Total retinal thinning of drusen and drusen-free areas compared to the normal population was most profound beyond the parafovea (>1250 μm). Functional and anatomical losses observed in AMD patients also have been reported beyond the parafovea and in the peripheral retina, suggesting rod photoreceptors are mostly lost in early disease.^{39–41,54,55} This is supported by studies using adaptive optics scanning laser ophthalmoscopy, which show normal cone density above drusen and in drusen-free areas of AMD eyes.^56–58 Although RPE/PR thinning may be attributed to rod degeneration, abnormal cone adaptation, sensitivity, and multifocal electroretinograms in early AMD suggest remaining cones still may be dysfunctional. 

Conclusions

This study demonstrated that retinal thickness above drusen in AMD patients is reduced compared to adjacent drusen-free areas. Thinning was most significant in the retinal layers closest to the drusen, specifically the RPE/PR and ONL layers. For confluent drusen, thinning also occurred within the IPL/GCL layer, implying that larger drusen may affect the inner retina as well. Reduced retinal thickness above drusen was linearly proportional to drusen height, but only modestly correlated with drusen width, suggesting current systems used to classify AMD severity based on drusen diameter may need to be refined. Most importantly, when compared to the normal population, retinal thickness was reduced above drusen and at adjacent drusen-free area, indicating that not only is the retinal architecture above drusen abnormal in AMD patients, but these effects extend beyond the borders of drusen into neighboring drusen-free areas of the AMD retina. 

Acknowledgments

We thank Cornelia Zangerl for assistance in acquiring measurements of the OCT images used in this study. 

Supported in part by research grants from the National Health and Medical Research Council of Australia (#1033224) and the University of New South Wales Early Career Research Grant 2014 (#PS35430). The authors alone are responsible for the content and writing of the paper. 

Disclosure: J. Rogala, None; B. Zangerl, None; N. Assaad, None; E.L. Fletcher, None; M. Kalloniatis, None; L. Nivison-Smith, None