Foveal Sparing of Reticular Drusen in Eyes With Early and Intermediate Age-Related Macular Degeneration

Abstract

Purpose: To analyze the central distribution of reticular drusen (RDR) in eyes with early and intermediate AMD without soft drusen or pigmentary changes within the central subfield using confocal scanning laser ophthalmoscopy (cSLO) and spectral-domain optical coherence tomography (SD-OCT).

Methods: Fifty-two eyes of 46 subjects (median age: 76.3 years, interquartile range [IQR], 71–80) were examined by simultaneous combined near-infrared cSLO and raster SD-OCT imaging. The appearance and the topographic distribution of RDR were analyzed within the macula area using the Early Treatment Diabetic Retinopathy Study grid. In addition, longitudinal examinations during an observation period of at least 6 months were included (median observation time: 1.5 years, IQR, 0.9–2.8).

Results: The RDR involvement within the central subfield (46%) was less compared with the surrounding subfields (62%–100%), slices (67%–100%), and zones (94%–100%) (P < 0.001). RDR were typically distributed as one continuous zone around the fovea in an incomplete or complete ring-shaped pattern, whereas the fovea itself was either spared or only a few lesions were present. Over time, the RDR density increased and new RDR lesions occurred at the border of the RDR zone toward a closure of the ring-shaped pattern. Within the fovea, development of RDR was observed in 8 of 28 eyes.

Conclusions: The fovea appears to be less vulnerable to RDR development as compared with peripheral macula areas. Factors for initial sparing of the foveal retina are yet unknown but may relate to topographic differences of choroidal blood flow and/or photoreceptor distribution.

Drusen represent a typical phenotypic hallmark in patients with AMD. Initially in 1990, reticular drusen (RDR) have been distinguished from soft, hard, calcified and cuticular drusen.¹ Various terms, including “reticular pseudodrusen,”² “reticular pattern,”³ “reticular macular disease,”⁴ and “subretinal deposits,”⁵ have been used to describe these subtle lesions. Due to the introduction of high-resolution imaging technologies, particularly confocal scanning laser ophthalmoscopy (cSLO) and spectral-domain optical coherence tomography (SD-OCT), their detection has markedly improved.^5,6 Longitudinal analysis revealed an increase in the involved retinal area quantified in fundus autofluorescence images and a possible growth of single lesions identified by SD-OCT images over time.^5,7–10 In addition, several studies indicate that RDR may represent a risk factor for AMD progression.^2,11–13 

Differences in the topographic distribution of RDR have already been noted by initial reports.^2,5,6,11 It has subsequently been confirmed by several reports that RDR are most prevalent in the superior macula.^6,10,14–19 Sarks et al.²⁰ observed in an AMD cohort of different AMD stages a relative sparing of RDR in the fovea in contrast to soft drusen that have a predilection for the fovea. These findings were supported by histologic findings that demonstrated a paucity of subretinal drusenoid deposits (SDD; i.e., reticular drusen) in the fovea and an abundance of SDD in the perifovea in AMD patients.²¹ Furthermore, it has been shown that RDR may also occur nasal to the disc (i.e., outside the macula).¹⁵ The systematic study by Joachim et al.²² demonstrated an increased risk for development to late-stage AMD if RDR were initially located outside the Early Treatment Diabetic Retinopathy Study (ETDRS) grid. Of note, and as discussed previously, the analysis of foveal involvement was challenging in previous reports that mainly included subjects with late AMD. At that stage, the possibility to detect RDR may be particularly challenging in the center of the macular due to the presence of other AMD-related lesions, such as atrophy, exudation, or subretinal fibrosis. 

The morphological substrate and the underlying mechanisms for the development of RDR remain a matter of debate, with controversial histopathological and in vivo findings.^2,20,23,24 An increasing knowledge on RDR manifestation and their variation over time may help to understand the underlying yet unknown pathophysiological process. The aim of the study was to analyze the appearance, topographic distribution, and longitudinal changes of RDR. To specifically evaluate the RDR occurrence with the fovea, this analysis was performed in eyes with early and intermediate AMD in absence of any other AMD-related phenotypic alterations in the central subfield of the ETDRS grid. 

Methods

Subjects

The database of the AMD outpatient clinic at the Department of Ophthalmology, University of Bonn, was retrospectively screened for eyes with RDR and early or intermediate AMD from March 1, 2014, to December 31, 2014. Eligible eyes had to show high-quality retinal imaging with a clearly distinguishable RDR area of at least two disc areas in size. In addition, follow-up examinations of a time period of at least 6 months had to be available. Exclusion criteria in the study eye comprised late AMD (geographic atrophy [GA] or choroidal neovascularization [CNV]) or any other confounding ocular disease (e.g., diabetic retinopathy, glaucoma, retinal vessel occlusion, retinal dystrophies, uveitis, dense cataract), as well as history of retinal surgery. If the inclusion and exclusion criteria were met, both eyes were included. 

Clinical data were collected from the medical records. The data archived included age, best-corrected visual acuity, and lens status. Using stereo biomicroscopy, the AMD stage of both eyes had been initially classified according to the AMD classification system by Ferris et al.²⁵ To confirm the AMD stage, the cSLO plus SD-OCT imaging data (see below) of every eye included in the current study was reviewed. 

The study followed the tenets of the Declaration of Helsinki. According to the Medical Association's professional code of conduct (§15 Berufsordnung für die nordrheinischen Ärztinnen und Ärzte, Ärztekammer Nordrhein), no ethical approval was required for this retrospective data analysis. 

Imaging Protocol

After dilation of the pupils with 1.0% tropicamide and 2.5% phenylephrine, all subjects received a minimum retinal imaging protocol. Simultaneous near-infrared (NIR) cSLO and SD-OCT imaging was performed using the Spectralis HRA+OCT (Heidelberg Engineering, Heidelberg, Germany) at all visits. Subjects underwent a central cSLO NIR-reflectance image (λ = 820 nm, 30° × 30°, high-speed mode: 768 × 768 pixels, automatic real time (ART; at least nine frames) combined with an SD-OCT raster scan (λ = 870 nm, 20° × 20°, centered on the fovea), at least 19 B-scans (at the most 240-μm distance between each B-scan). This raster scan was repeated at every visit using the AutoRescan tool, which automatically places follow-up scans in the same location as the baseline scan. 

Definitions and Nomenclature

In the NIR cSLO images, RDR were characterized as a group of hyporeflective dots (= dot) or hyporeflective rings with a hyperreflective center (= target) as described before.^4–6,26 In SD-OCT scans, RDR were seen as hyperreflective mounds above the RPE (= “waves”) or as “spikes” (lesions breaking through the external limiting membrane).^6,16 

Analysis

To analyze the presence and topographic distribution of RDR within the macula and particularly the fovea, an ETDRS grid was centered on the fovea (Fig. 1). As the visualization of RDR is challenging, or even RDR might be no longer detectable when other AMD-related pathological alterations such as regular drusen are present, all eyes with regular soft drusen or pigmentary changes within the central subfield of the ETDRS grid were excluded from the further analysis. In the remaining eyes, the distribution of RDR within the fovea was classified in three different categories: 1 = no RDR, 2 = few RDR (five or fewer lesions) and 3 = many RDR (more than five lesions). Foveal sparing was defined by the absence of any RDR within the central subfield (category 1). Furthermore, the presence of RDR was assessed in each of the other eight subfields of the ETDRS grid. The central subfield was defined as the central zone, the four subfields between the inner and middle circles as the middle zone, and the four subfields between the middle and outer circles as the outer zone, respectively. Furthermore, the data of each of the two superior, inferior, nasal, and temporal subfields were summarized in four “slices.” The analysis was performed at baseline and at the last performed follow-up examination. 

Figure 1

 

A representative example of a combined cSLO NIR reflectance and SD-OCT image. The ETDRS grid with the inner (diameter 1600 μm), the middle (3200 μm), and the outer (7200 μm) circles is centered on the fovea.

Statistical Analysis

Data were compiled with a standard spreadsheet program (Microsoft Excel; Microsoft Corporation, Redmond, WA, USA) and analyzed using commercially available statistical software (IBM SPSS, Armonk, NY, USA). Friedman statistics were used for further statistical analysis. 

Results

A total of 105 eyes of 89 subjects were retrieved from the AMD database with RDR and early or intermediate AMD. Fifty-three eyes from 43 patients were excluded because of soft drusen or pigmentary changes within the central ETDRS subfield. The remaining 52 eyes of 46 subjects were included. The median age at baseline was 76.3 years (interquartile range [IQR], 71–80, range, 65–88 years). There were 17 men and 29 women. The mean visual acuity was 20/25 (range, 20/100–20/20). According to the AMD classification system by Ferris et al.,²⁵ 16 eyes were classified with early AMD and 36 eyes with intermediate AMD. In 6 patients, both eyes were included and in 40 patients only 1 eye was included. Of the nonincluded fellow-eyes, there were 31 eyes with CNV, 1 eye with GA, and 8 eyes with intermediate AMD. There were 20 pseudophakic and 32 phakic eyes, respectively. The median follow-up period between baseline and follow-up examinations was 1.5 years (IQR, 0.9–2.8, range, 0.5–6.1). 

Figure 2 (upper row) demonstrates the overall RDR involvement of different subfields, slices, and zones for the baseline examination. Reticular drusen were particularly found superior (100%) and nasal (81%) to the fovea. Reticular drusen involvement within the fovea (46%) was clearly less common compared with all other subfields (62%–100%, P < 0.001), slices (67%–100%, P < 0.001), and zones (94%–100%, P < 0.001). 

Figure 2

 

The involvement of RDR in different subfields (left column), slices (middle column) and zones (right column) for baseline (upper row) and follow-up (lower row) examinations.

Reticular drusen involvement was higher at follow-up compared with baseline in all subfields, slices, and zones (provided it had not been 100% at baseline) (Fig. 2, lower row). The largest increase was detected in the inferior and temporal subfields that had revealed lower rates of involvement at baseline. At the follow-up examination, both the middle and the outer zone did always show RDR involvement (100%), whereas foveal involvement was 62%. 

If RDR were present in more than one subfield, the RDR area always involved neighboring subfields (i.e., two or more separated RDR areas were not visible in any eye). Typically, RDR areas were distributed around the fovea in an incomplete or complete ring-shaped pattern, whereas the fovea itself was spared or only a few RDR lesions (five or fewer RDR) were found. Figure 3 shows representative examples demonstrating different degrees of ring-shaped patterns around the fovea. 

Figure 3

 

Representative examples demonstrating different degrees of ring-shaped patterns around the fovea (first row: one-quarter ring, second row: one-half ring, third row: three-quarters ring, fourth row: complete ring). In each row, the native and the processed image with the area of RDR involvement is highlighted in blue.

To further describe this typical topographic distribution of RDR involvement, individual eyes were systematically grouped according to the extent of ring-shaped involvement using four different categories, and the degree of RDR involvement in the fovea itself (Table). Although most eyes showed a three-quarters ring (n = 15) or a complete ring involvement (n = 26), most eyes also showed a complete foveal sparing (n = 28) or only a few RDR lesions (n = 13) in the central subfield. Representative examples for complete foveal sparing, and presence of few RDR and many RDR are shown in Figure 4. 

Figure 4

 

Six examples of foveal RDR involvement (left: no RDR, middle: a few RDR, right: many RDR). For each example the whole grid (first and fourth rows), the enlarged fovea with the surrounding subfields (second and fifth rows) and the schematic RDR distribution (third and sixth rows) are illustrated.

Table

 

The Classification of RDR Involvement According to the Extent of Ring-Shaped Manifestation and the Number of Lesions in the Fovea and Distribution in the Study Cohort

 

During follow-up, new RDR lesions typically occurred at the border of the existing RDR area, whereas the individual size and density of RDR lesions within the area observed at baseline increased. Furthermore, if an incomplete ring-shaped pattern of RDR involvement at baseline was visible, the enlargement of the RDR area typically encompassed the fovea spreading toward “a closure” of the ring, similar to a “horseshoe” and up to a “donut”-like configuration. At the same time, a residual “RDR-free” island or an almost RDR-spared central subfield was visible. At least, the density of individual RDR lesions in the central subfield typically appeared to be less compared with the surrounding subfields (Fig. 5). 

Figure 5

 

A demonstration for two examples of the increase of RDR involvement over time. In the left example, the fovea remains spared over time. In the right example, only a few RDR are found within the fovea at baseline. At follow-up examination, an overall marked increase in RDR area and density is seen, also with a change to many RDR within the central subfield.

Discussion

The findings of this study suggest that RDR occur less commonly in the foveal area compared with more eccentric locations in eyes with early or intermediate AMD. The fovea may be spared initially and becomes involved later in the course over time. Furthermore, the development of individual RDR lesions appears to occur to a smaller extent, showing a relatively lower RDR density within the fovea. Eventually, as found in a few eyes in the study, the pattern of foveal involvement may become similar to other macular areas. 

Several previous studies analyzed the topographic distribution of RDR using the ETDRS grid.^15,17 The predilection superior to the fovea, as confirmed in the current study, has been a consistent finding. A reduction in RDR involvement within the fovea also has been observed.^20,21 However, most of these studies did not include subjects with early or intermediate AMD, but rather addressed late AMD stages. Of note, it would be conceivable that the systematic analysis of RDR distribution within the center of the macula would be particularly challenging in the presence of CNV, subretinal fibrosis, or GA. In the current study, we also excluded eyes with soft drusen or pigmentary changes within the central subfield to ensure that no other AMD-related pathological alterations would obscure the detection of RDR lesions in the most crucial subfield: the fovea. In the other subfields, RDR could have existed before the appearance of soft drusen and might not be detectable anymore. The obtained occurrence rates should therefore be seen as minimal rates. Using this strategy, we found that foveal RDR sparing was present in 54% of eyes at baseline, which was significantly less compared with surrounding subfields. The more detailed analysis revealed a typical ring-shaped presence of the RDR area that appeared to encompass the spared or little involved fovea over time (“horseshoe” and “donut”-like pattern). 

Furthermore, we classified the topographic RDR involvement that is based on the extent of ring-shaped manifestation using four different categories according to the number of quadrants involved, and the number of RDR lesions in the fovea itself. We would propose that this classification system presented herein might be used in other studies to identify high-risk characteristics for development of late-stage AMD that may be also helpful to design and monitor future interventional trials for subjects at particular high risk of disease progression and severe visual loss. 

Interestingly, the RDR area and its spread over time followed a “floor”-like growth pattern typically covering a continuous zone with confluent or a high density of single RDR lesions in the center and a more mottled appearance at its borders. This pattern would differ from the occurrence and spread of GA that commonly manifests by multifocal lesions.^27–29 

The preservation of the fovea to avoid loss of central vision is the principal goal in the treatment of any macula diseases. The phenomenon of foveal sparing is also seen in eyes with GA,^30–32 Stargardt disease,^33–35 and mitochondrial retinal dystrophy.³⁶ It would be conceivable that protective factors for a reduced vulnerability of the fovea for RDR involvement would be either similar or different as compared with these other conditions. Previously discussed protective factors for foveal sparing in conditions other than RDR manifestation would be the unique choroidal blood flow of the fovea, protective effects of macular pigment, and the peculiar distribution of photoreceptors.^37–42 Overall, the underlying pathological mechanisms for RDR development and growth over time remain unknown. Particularly, the histopathological substrate has been a matter of debate. Although some studies favor an origin within the choroid, other findings that are mainly based on simultaneous SD-OCT and cSLO imaging rather suggest alterations at the level of the photoreceptors with focal accumulations of hyperreflective material above the RPE and with disintegration of the external limiting membrane.^23,24,43,44 Both hypotheses, alone or in combination, might explain the phenomenon of a lesser vulnerability of the fovea for RDR involvement as reported in the current study. 

It would be conceivable that an impaired choriocapillaris blood flow causing RDR development may less likely and/or later occur in the disease process in the fovea itself due to the peculiar supply by choroidal vessels of the fovea. The assumption would be further underscored by the study of Alten et al.,⁴⁵ who reported of a topographic relationship between localized RDR presence and choroidal watershed zone, suggesting that choroidal hypoxia may have a role in RDR pathogenesis. Furthermore, there is ongoing discussion in the literature if choroidal thickness is reduced along with the presence of RDR.^14,45–47 On the other hand, it may be also speculated that choroidal vascular changes represent an epiphenomenon (i.e., that the choroidal blood flow becomes impaired after RDR development). The second hypothesis postulating that accumulation of material above the RPE might interfere with normal metabolism of the RPE/photoreceptor complex and then resulting in the manifestation of RDR lesions might also explain the lesser vulnerability of the fovea for RDR involvement. It could be speculated that the different photoreceptors might have an important impact on the occurrence and growth patterns of RDR lesions in general; that is, that the cones would be less vulnerable as compared with rods and that the predominance of cones in the fovea would then lead to the observation to a lesser and later RDR involvement over time. This hypothesis would be in accordance with the recent findings by Curcio and coworkers,²¹ who demonstrated a spatial correlation of damage to rods and the presence of subretinal deposits (i.e., reticular drusen). Furthermore, histological findings and structural–functional analysis in AMD subjects have consistently reported of a preferential vulnerability of the rod system and scotopic function, along with a relative preservation of cone photoreceptors and cone function in the early stages of AMD in general.^21,38,39 Finally, we have recently observed a larger extent of impaired scotopic function specifically over RDR areas as compared with only a mild localized photopic dysfunction.⁴⁸ 

High-resolution imaging, along with the three-dimensional mapping of RDR lesions using combined cSLO and raster SD-OCT imaging, is an important prerequisite for the detection of small, subtle lesions, such as RDR.^5,6,26 Limitations of the current study are notably the retrospective design, the small sample size, including subjects with different AMD stages in the fellow eye, and the inclusion of both eyes in some subjects. Although we would not feel that increasing the number of subjects would change the main conclusions, we cannot exclude selection bias of cases. However, the observation of a lesser involvement of the central subfield and the typical ring-shaped pattern was a very consistent finding in our cohort. For the ongoing analysis, we aim to initiate a prospective study with predefined follow-up examinations at regular intervals. Quantitative analysis of individual lesions and the involved area by using automated software algorithm may allow further elucidation of the development and growth pattern of RDR lesions, as well as their impact on the course of AMD in general. Based on the current knowledge, including the findings of this study, we would hypothesize that the use of currently available high-resolution imaging technology would allow detecting marked changes in the RDR distribution and growth pattern within a time period between 1.5 and 2.0 years. 

In summary, we performed an analysis of the macula distribution of RDR in eyes with early or intermediate AMD and without soft drusen or pigmentary changes within the fovea. This study demonstrates a lower vulnerability of the fovea as compared with peripheral macular areas. Factors for initial sparing of the foveal retina are yet unknown but may be related to topographic difference of choroidal blood flow and photoreceptor distribution. 

Acknowledgments

Supported by the Gertrud Kusen Foundation. 

Disclosure: J.S. Steinberg, Optos (F), Heidelberg Engineering (F); M. Fleckenstein, Optos (F), Heidelberg Engineering (F, C, R), Zeiss Meditec (F), P; F.G. Holz, Optos (F, C), Heidelberg Engineering (F, C, R), Zeiss Meditec (F), Novartis (F, C), Bayer Healthcare (F, C), Alcon (F, C), Allergan (F, C), Acucela (C), Boehringer Ingelheim (C); S. Schmitz-Valckenberg, Optos (F, R), Heidelberg Engineering (F, R), Zeiss Meditec (F), Novartis (F, C, R), Bayer Healthcare (F), Allergan (F), Roche (F), Bayer (R)