Background/aims To analyse appearance, development over 2 years and characteristic patterns of reticular drusen (RDR) in eyes with high-risk characteristics for progression to late-stage age-related macular degeneration (AMD) (age-related eye disease study stages 3 and 4).
Methods 98 eyes of 98 patients (median age 73.4 years, IQR [69–78]) participating in the Molecular Diagnostic of Age-related Macular Degeneration study were included. Simultaneous combined confocal scanning laser ophthalmoscopy (cSLO) and spectral-domain optical coherence tomography (SD-OCT) imaging as well as colour-fundus imaging was performed at baseline and at 24 months. Two independent graders determined the presence of different RDR phenotypes (cSLO modalities: ‘dot’, ‘target’, ‘ribbon’; SD-OCT: ‘spike’ and ‘wave’) at both visits.
Results At baseline, RDR were detected in 44% (κ 0.96). They were always visible in near-infrared reflectance images. Detection rate was 42% using fundus autofluorescence (FAF), 39% on SD-OCT (waves: 100%; spikes: 90%) and 26% on blue reflectance (BR). ‘Dots’ were more frequently detected in all imaging compared with ‘targets’. The ‘ribbon’ pattern was most frequently observed in colour images, BR images and FAF images. In 8 of the 48 eyes with no signs of RDR in any imaging modality at baseline, the development of RDR lesions was observed at 24 months (16.6%, κ 0.42).
Conclusions Careful and meticulous analysis using three-dimensional in vivo imaging reveals distinct characteristic RDR patterns underlying detectable dynamic changes over a period of 2 years. RDR in eyes with early or intermediate AMD are a common observation but appear to be overall less common compared with eyes with geographic atrophy.
Statistics from Altmetric.com
Mimoun and coworkers first described reticular pseudodrusen in AMD eyes as “retinal lesions with a variable diameter of about 100 microns that did not appear hyperfluorescent on fluorescein angiography”.3 Subsequently, various terms have been introduced including ‘reticular drusen’ (RDR), ‘reticular macular disease’ or ‘subretinal drusenoid deposits’ (for a review, see Alten et al4). With the development of high-resolution imaging technologies including confocal scanning laser ophthalmoscopy (cSLO) and spectral-domain optical coherence tomography (SD-OCT), the detection of RDR has become easier and the morphological substrate is now assumed to be located anterior to the retinal pigment epithelium (RPE) cell monolayer in contrast to the sub-RPE location of ‘regular’ drusen. Several lines of evidence indicate that RDR confer a high-risk factor for the development of late-stage AMD.5
The prevalence of RDR in population-based studies has been reported to be 0.7% with incidence rates of 2% over 5 years and 3–4% over 15 years.6 ,7 In AMD cohorts, a marked variation of RDR prevalence has been published, ranging from 8.7% to up to 62%.8–12 These differences may be partially due to the inclusion of different AMD stages, whereas most studies did not include early or intermediate AMD cases. Overall, it is a consistent finding that combined cSLO and SD-OCT imaging is superior in the detection of the subtle RDR lesions compared with standard fundus camera photography.9 ,12–14 Furthermore, standardised imaging protocols and careful image acquisition appear to be an essential perquisite for a meaningful RDR analysis.15
Along with high-resolution retinal imaging, morphological variations of RDR have been increasingly recognised.16–18 Using en-face imaging, these variations have been described as ‘dot’, ‘target’ and ‘ribbon’ appearance, but not yet systematically assessed and compared by different imaging modalities. In addition, little data are available about dynamic changes of RDR over time, particularly in eyes with early or intermediate AMD showing high-risk characteristics for the development of late disease stages.14 ,15 ,19 ,20
The purpose of this study was to determine the presence of characteristic RDR patterns at baseline and the development over 2 years in AMD eyes with high-risk characteristics for progression to late disease stages. This analysis was based on retinal imaging data consisting of different modalities and that had been obtained within a natural history study using standardised image acquisition procedures.
Patients were enrolled in the Molecular Diagnostic of Age-related Macular Degeneration (MODIAMD) study (http://www.modiamd.de) between November 2010 and September 2011. The MODIAMD study is a prospective, non-interventional, observational, longitudinal natural history study over three years with patients at high risk for developing late-stage AMD in the study eye (see below). The primary endpoint of this study is not the scope of this manuscript and will be reported elsewhere.
For inclusion, patients had to be >50 years of age and manifesting retinal changes classified as age-related eye disease study (AREDS) stages 3 or 4, that is, having at least one eye without advanced AMD that would be considered to be at high risk for developing late stages of the disease. The AREDS classification illustrates a well-established AMD severity score.21 Briefly, AREDS stage 3 is characterised by patients with intermediate drusen (>63 µm but <125 µm), large drusen (>125 µm), drusen area ≥360 µm or RPE changes in one eye. Geographic atrophy can occur but must not be within the centre macula. The fellow eye has to show at the most the same stage of severity. AREDS stage 4 is characterised by advanced AMD (either central geographic atrophy (GA) or choroidal neovascularisation (CNV)) in one eye, while the other eye has changes of early or intermediate AMD.
Exclusion criteria comprised any ocular disease that may confound the assessment of the retina other than AMD and previous or concomitant therapy to treat AMD in study eye.
In AREDS category 3, the eye with more pronounced changes (larger drusen area and more pigmentary changes) was selected as study eye. In AREDS category 4, the eye without advanced AMD was selected as study eye.
At each visit, all subjects underwent a complete ophthalmic examination including assessment of best-corrected visual acuity using Early Treatment Diabetic Retinopathy Study charts and dilated fundus examination; retinal images were collected using combined and simultaneous cSLO+SD-OCT imaging and fundus camera photography. Follow-up visits were performed annually, that is, 12 and 24 months after the baseline visit.
Retinal imaging was performed according to standardised operating procedures. After dilation of the pupils with 1.0% tropicamide, colour fundus photography was performed using a standard fundus camera (Visucam 500, Carl Zeiss Meditec AG, Jena, Germany). For the first image, the field of view was set at 30°×30° and centred on the macula (field 2). Furthermore, two additional fields were obtained, one temporal to the macula (field 3M), the other one nasal to the macula with the optic disc in the centre (field 1M).
High-speed combined and simultaneous cSLO+SD-OCT imaging (768×768 pixel) was performed with the Spectralis HRA+OCT (Heidelberg Engineering, Heidelberg, Germany) device and included acquisition of central 30°×30° near-infrared reflectance (IR, λ=820 nm, automatic real time (ART) at least 15 frames), central blue reflectance (BR, λ=488 nm, ART at least 15 frames) and three-field fundus autofluorescence (FAF, exc λ=488 nm, em λ=500–800 nm, at least 15 frames) images. In addition to single horizontal and vertical combined cSLO+SD-OCT scans through the fovea, a volume scan (λ=870 nm, 20°×15°, ART at least four frames, 19 B-scans, distance 240 µm) was recorded. At baseline, fluorescein angiography (FAG) was performed in both eyes.
Definitions and nomenclature
In the cSLO modalities, RDR were characterised as a group of hyporeflective dots (=dot), hyporeflective rings with a hyperreflective centre (=target) or irregular wriggled pattern (=ribbon) as described before.16 ,22 ,23 For FAG, late-phase images (after 5 min) were analysed. On colour fundus images, RDR were detected as greyish/yellow changes of 50–200 µm in diameter. In SD-OCT images, RDR were seen as ≥5 hyperreflective mounds above the RPE (=waves) or as ‘spikes’ (lesions breaking through the external limiting membrane) in ≥1 scan.8 ,13
For the current analysis, all study eyes were included. Two independent graders (JSS and APG) analysed all images at baseline and at month 24 for the presence and characteristic patterns of RDR. In a second analysis, both graders discussed and arbitrated each individual case with initial discrepant finding(s).
Data were compiled with a standard spreadsheet program (Microsoft Excel) and analysed using commercially available statistical software (IBM SPSS, Armonk, New York, USA). Kappa statistics, Wilcoxon–Mann–Whitney and Pearson χ2 test were used for further statistical analysis.
The median age at baseline was 73.4 years (IQR [69–78], range 51–90 years). There were 30 men (30%) and 68 women (70%). There were 23 patients with AREDS stage 3 and 75 patients with AREDS stage 4. Over 2 years, 19 eyes converted into late-stage AMD (19.4%), whereby 10 eyes developed CNV and 9 eyes GA (p=0.289).
Presence of RDR in different imaging modalities
At baseline, RDR were detectable in 43.9% of the subjects in at least one imaging modality (figure 1 and table 1). If present, RDR were always visible in IR images (43.9%) as dot pattern (100%) whereas ‘targets’ were found in 81.4%. No ‘ribbon’ pattern was observed. In FAF images, RDR were visible in 41.8% (‘dot’ pattern=97.6%, ‘target’=48.8%, ‘ribbon’=26.9%). In SD-OCT images, RDR were detectable in 38.8% (waves= 100%, spikes=89.5%). The appearance of RDR in FAG, BR and colour fundus images was lower. No target lesions were noticed in FAG or colour fundus images. The ‘ribbon’ pattern was most pronounced in colour fundus (41.7%) and in BR (32.0%) images. The presence of RDR in at least one modality was similar in AREDS 3 (43.5%) and AREDS 4 (44.0%) patients. Further details of the analysis of the presence of RDR are shown in table 1.
Development of RDR during the observational period
In 8 of the 48 eyes with no signs of RDR at baseline, the development of RDR lesions in at least one imaging modality was recorded within 24 months (16.6%, figure 2 and table 2). New RDR were always detectable in IR (eight eyes) and SD-OCT (eight eyes) images. Disappearance of RDR was not seen in any of the 43 eyes with RDR at baseline. Within the observational period of 24 months, no marked changes of the ribbon pattern were noticed. As described previously, the density of individual dots and targets increased, while some dots changed into the target appearance.20 The development of RDR was higher in patients with AREDS 4 (seven eyes, 18.4%) than in patients with AREDS 3 (one eye, 10%), but not statistically significant (p=0.445). Further detailed information is shown in table 2.
Association with additional features
The analysis of demographic factors at baseline revealed a statistically significant association for the presence of RDR with age (p=0.018) and with female gender (p=0.006). No statistically significant association was found for the AREDS stage (p=0.965) or the conversion to late AMD stages (p=0.601). Further detailed information is shown in table 3.
In this study, a systematic analysis was conducted of the presence of different RDR patterns at baseline and their development over 2 years in subjects with high-risk characteristics for late AMD using multimodal high-resolution imaging techniques that had been obtained using standardised operational procedures.
High-resolution imaging is an important prerequisite for the detection of small, subtle lesions such as RDR.8 ,10 ,13 Despite improvements in imaging resolution, the prevalence rates may yet be underestimated due to lack of clear-cut imaging quality threshold9 and other phenotypic alterations such as hard or soft drusen that may interfere with RDR detection. The IR and the FAF modes of the cSLO and corresponding SD-OCT scans have been previously shown to reveal the highest sensitivity in detecting RDR.6 ,8 ,9 ,13 ,24 Using colour fundus images only, the detection rate is much lower. Studies based on colour fundus photography only would certainly underestimate the prevalence of RDR.9
In this study, the presence of RDR in 98 patients with either AREDS 3 or 4 was 44% in at least one imaging modality. The analysis of a multicentre study of fellow eyes of patients with neovascular AMD by Hogg et al24 disclosed a similar prevalence of 41% in fellow eyes of patient with unilateral CNV. Studies with high-resolution imaging in patients with more advanced AMD stages23 including GA (62%)9 or pigment epithelial detachment (54%)10 reported a higher prevalence of RDR. This underscores the relevance of RDR as a phenotypic feature associated with various manifestations of AMD. Furthermore, the lower rate in subjects with other high-risk features may reflect RDR as part of the progressive disease process in AMD. Both size of individual lesion and extension of the involved RDR area show an expansion over time.20
Until the present study, no systematic analysis of the different RDR pattern had been performed. In any modality, the ‘dot’ pattern is seen most frequently followed by the ‘target’ variant. ‘Targets’ are the biggest single RDR lesions that, even if they are present, are usually less numerous than single dots. Interestingly, ‘target’ lesions are not found in FAG and colour fundus images. In FAG images, this might be explained by the interference with other retinal alterations such as soft drusen. In colour fundus photographs, the difference in light intensity within a target between the ring (=hyporeflective in IR) and the centre (hyperreflective in IR) is usually invisible. If present, the ‘ribbon’ pattern is hardly detectable in IR images but well seen on BR and colour images. On corresponding IR images, no ribbon pattern is noticed but rather coalescent lesions (figure 2). Interestingly, if present, the ‘ribbon’ pattern may occur together with dots and targets in the same eye. Overall, we found a good intragrader variability in IR, FAF and SD-OCT modalities (κ 0.96, 0.88 and 0.89, respectively).
We could demonstrate the appearance of RDR after a time period of 24 months, that is, in eyes that did not show RDR at baseline. The detection of new RDR lesions over the observation period of 2 years was the highest in IR, FAF and SD-OCT modalities (16%). Population-based studies revealed 15- year incidence of 3–4%6 ,7 ,25 using colour fundus images. Interestingly, if any RDR developed after 2 years, no ‘ribbon’ pattern occurred while ‘dot’ and ‘target’ lesions were typically seen. It may, thus, be hypothesised that the ‘ribbon’ pattern takes more time to develop whereas small single ‘dots’ or ‘targets’ are already visible at early stages of RDR evolution. Zweifel et al19 recently reported the potential growth in height of single RDR in SD-OCT images so that it can be assumed that dots progress to targets over time. It is not clear whether or not the ‘ribbon’ pattern develops from single dots and targets or whether it represents a separate RDR phenotype. Further longitudinal studies are needed to address this in detail.
The prevalence of RDR correlates significantly with increasing age and female gender.6 ,8 ,10 ,13 Presuming that RDR represent a high-risk factor for the progression to late AMD, the conversion rate in eyes with high-risk features was assumed to be higher than in eyes without RDR. Interestingly, no correlation was found between the conversion rate into late AMD. This might be explained by the limited observation time (24 months) and the sample size of the study, which encompassed 98 patients with high-risk characteristics at baseline.
In patients classified as AREDS 3, bilateral RDR were detected in 90% (p<0.005). This high interindividual symmetry is in accordance with the findings of previous studies.8 ,9 ,11 The prevalence of RDR at baseline in the fellow eyes of the AREDS 3 study eye was 43.5%, which was similar to the overall prevalence in all study eyes. The fact that no RDR developed over time in this group of eyes may be related to the limited number of patients. However, it indicates that patients with intermediate AMD in both eyes (AREDS 3) might not have a higher incidence rate than patients with AREDS 4.
In summary, high-resolution imaging based on combined cSLO and SD-OCT imaging and acquired under standardised operational procedures allows for a detailed analysis of different RDR patterns. The rate of 44%, visible as the ‘dot’ and ‘target’ pattern by cSLO near-infrared imaging, suggests that RDR are a common observation also in eyes at high risk for progression to late AMD stages. Compared to the previously reported rate of 63% in subjects with GA or 54% in subjects with pigment epithelium detachments (also based on cSLO imaging), these findings also underscore the dynamic nature of RDR that were visible within a time period of 2 years in this study itself.
Contributors The manuscript was written by JSS and SS-V and reviewed by all coauthors. The original data were collected by APG, JSS, MF, SS-V and FGH. The data analysis was primarily performed by JSS, APG and SS-V. Additional support was provided by MF and FGH.
Funding Gertrud Kusen foundation, German Ministry of Education and Research (BMBF), FKZ 13N10349 conducted at the Department of Ophthalmology, University of Bonn, Germany.
Competing interests None.
Patient consent Obtained.
Ethics approval The study follows the tenets of the Declaration of Helsinki and was approved by the local ethics committee (Ethik-Kommission der Universität Bonn Lfd-Nr: 175/10).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.