Article Text
PDF
Abstract
Purpose To investigate the intraretinal structural and vascular alterations in patients featuring reticular pseudodrusen (RPD), RPD with outer retinal atrophy (ORA), and drusen.
Design Observational cross-sectional study.
Methods Clinical practice study including 68 eyes of 57 patients (22 eyes of 17 patients with RPD; 24 eyes of 21 patients with RPD+ORA; 22 eyes of 19 patients with drusen). Each patient underwent spectral-domain optical coherence tomography (OCT) and OCT angiography (OCT-A). Measurement of retinal layers’ thickness was obtained by the automated segmentation protocol of the Spectralis OCT (Heidelberg Eye Explorer V.1.9.10.0). The superficial capillary plexus (SCP) and the deep capillary plexus (DCP) vessel density, as well as the size of the foveal avascular zone were calculated on 3×3 OCT-A. Main outcome was to compare vessel density at the SCP and DCP among the groups and controls.
Results At the SCP, the vessel density was lower in RPD and RPD+ORA patients with respect to controls (P=0.02 and P=0.003, respectively). At the DCP, meaningful disparity was found between the study groups and the healthy subjects in the vessel density (P<0.001, P=0.04 and P=0.001 for RPD, RDP+ORA and drusen, respectively). The ganglion cell layer (GCL) was thinner in all patients affected either by RPD, RPD+ORA or drusen compared with healthy subjects (P=0.02, P=0.03 and P=0.004, respectively).
Conclusion Significant retinal vascular loss is a common feature of patients with non-exudative age-related macular degeneration, more pronounced in those featuring RPD and RPD+ORA. It is associated with retinal thinning, localised particularly at the GCL, compared with controls.
- retina
- imaging
Statistics from Altmetric.com
Introduction
Reticular pseudodrusen (RPD) has been first described in 1990 as peculiar yellowish lesions in the fundus with enhanced visibility when viewed using blue light.1 Despite this phenotype has been identified in different retinal conditions, such as pseudoxanthoma elasticum or Sorsby fundus dystrophy, the most characteristic association is with age-related macular degeneration (AMD).2–5
Several histopathological studies have found significant overlap between RPD and subretinal drusenoid deposits (SDD): both contain unesterified cholesterol, vitronectin, complement factor H and apolipoprotein E. However, RPD lack immunoreactivity for photoreceptors, Müller cells and retinal pigment epithelium (RPE) marker proteins, and have lower concentrations of esterified cholesterol and other neutral lipids with respect to SDD.6 7
Pathophysiologically, this material should represent an epiphenomenon of a functionally impaired RPE and Bruch’s Membrane Complex, leading to misdirection of transport of photoreceptor remnants into the subretinal space.8 9
An important association between RPD and late manifestations of AMD, including both neovascular and atrophic forms, has been described.9 10 In detail, the natural progression of RPD is the development of choroidal neovascularisation (CNV) in about 43% of patients, extensive geographic atrophy (GA) in nearly 20% or regression of the SDD in the other cases. Eyes with regression of RPD are characterised by progressive thinning of the outer retinal layers, visualised as loss of the external bands and diffuse backscattering of the signal on optical coherence tomography (OCT). This condition has been called outer retinal atrophy (ORA) and has been recognised as a different manifestation of advanced AMD as compared with the conventional entities of CNV and GA.11 According to the latest theories, RPD and ORA can be interpreted as two different stages in the spectrum of SDD disease.
A considerable amount of data is available on the abnormalities involving the outer retinal layers and the choroidal structures in patients with RPD.12–14 On the other hand, there is scant information regarding the condition of the inner retina in the different stages of the disease. Some authors have hypothesised that AMD might also affect the three innermost retinal layers: the inner plexiform layer (IPL), the ganglion cell layer (GCL) and the retinal nerve fibre layer (RNFL). Thanks to the new generation spectral-domain OCT (SD-OCT) software, it is now possible to calculate high-resolution quantitative maps of these retinal layers.15
OCT angiography (OCT-A) detects blood flow at both big vessels and capillary level by measuring the decorrelation rate in consecutive B-scans taken at the same location.16 Since its clinical introduction, OCT-A has been used for en face visualisation of the blood flow at different anatomic retinal layers, and the vascular anatomy at the posterior pole has been further subdivided into a superficial capillary plexus (SCP) and a deep capillary plexus (DCP)—at the retinal level—and into choriocapillaris (CC) and choroid— below the RPE. Pathological changes at the CC and the choroidal level on OCT-A in patients affected by RPD have been already described by our group, and are beyond the purpose of our study.14
The aim of the present study was to investigate the retinal alterations in patients featuring RPD with or without ORA, and to compare these data to patients featuring drusen, to understand if vascular changes occur in different stages of the same disease.
Methods
Cross-sectional clinical practice study carried out between October 2015 and May 2017 at the Medical Retina and Imaging Unit of San Raffaele Hospital in Milan. A consecutive series of patients with non-exudative AMD were enrolled. All patients signed a general written consent. All the procedures were performed in agreement with the principles of the Declaration of Helsinki.
Inclusion criteria were: age ≥50 years and diagnosis of AMD; clear dioptric media and stable fixation were also required to allow OCT-A examination. Patients featuring any other ocular disorder, including glaucoma or advanced AMD (GA or CNV), or systemic vascular conditions, as uncontrolled blood pressure, diabetes mellitus or peripheral vasculopathy, were excluded from the analysis. Subjects with a myopic refractive error greater than −6 D or axial length greater than 26 mm were also excluded.
A group of healthy age-matched subjects (one eye for each subject was included in the analysis) was enrolled as controls.
Each patient underwent best-corrected visual acuity (BCVA), biomicroscopy, applanation tonometry, multicolour and infrared (IR) imaging, short-wavelength (488 nm) fundus autofluorescence (SW-FAF) (Spectralis, HRA, Heidelberg, Heidelberg, Germany), OCT-A and SD-OCT. Based on ophthalmoscopy, multicolour, IR reflectance, SW-FAF and SD-OCT, patients with AMD were divided into three groups: those with RPD not accompanied by ORA (group 1), those presenting both RPD and ORA (group 2) and those with drusen (group 3). The diagnosis of RPD was defined by the presence of a reticular pattern measuring at least two disc diameters at the posterior pole, whose visibility was enhanced by IR reflectance; the presence of RPD was confirmed by SDD on SD-OCT imaging. The diagnosis of ORA was made on the presence of regressed SDD, attenuated ellipsoid zone (EZ) and hyper-transmission under the RPE in at least one scan on SD-OCT.11 The diagnosis of drusen was based on presence of five or more medium to large (63–124 µm) yellow lesion within the macula, confirmed by the presence of hyper-reflective deposits under the RPE on SD-OCT.
Nineteen horizontal SD-OCT B scans (1024 pixels in length and 496 pixels in depth) were acquired in each eye, covering a 6×5 mm retinal area, centred on the fovea. Measurement of retinal layers’ thickness was obtained through an automated segmentation protocol (Spectralis OCT, Heidelberg Eye Explorer V.1.9.10.0); layer segmentation was revised and manually adjusted when needed before image processing. The layers included in the analysis were: (1) RPE, determined from the hyper-reflective Bruch’s membrane to the hyper-reflective EZ (the former IS/OS boundary); (2) outer nuclear layer (ONL) from the external limiting membrane to the outer border of the hyper-reflective band of the outer plexiform layer (OPL); (3) OPL, from the inner boundary of the ONL to the outer rim of the hyporeflective inner nuclear layer (INL); (4) INL from the OPL to the hyper-reflective IPL; (5) IPL; (6) GCL and (7) RNFL, from the hyper-reflective internal limiting membrane to the GCL.17 18 Retinal layers thickness were measured in the nine sectors of the ETDRS macular grid at 3 mm, and then averaged. In the same subjects, an average central retinal thickness (CRT) in the 3 mm diameter circular area was obtained from the Spectralis SD-OCT software (figure 1). The mean thickness maps generated in the control group were used as a normative database. Thickness measurements among sectors between the different study groups were compared with the normative database using analysis of variance (ANOVA) statistics.
Optical coherence tomography (OCT) and OCT angiography (OCT-A) image processing for quantitative analysis of retinal layers’ thickness, vessel density and foveal avascular zone (FAZ) in a 62-year-old patient featuring reticular pseudodrusen (RPD). (A–B) Infrared imaging and OCT, showing multiple subretinal hyper-reflective RPD with preservation of the ellipsoid zone and of the retinal pigment epithelium. Measurement of retinal layers’ thickness was obtained by the automated segmentation protocol of the Spectralis OCT in the nine sectors of the 3 mm ETDRS macular grid. (C–D) 3×3 OCT-A segmentation at the superficial capillary plexus (SCP, left) and at the deep capillary plexus (DCP, right) of the same subject. (E–F) Binarisation of the OCT-A scans. The adjust threshold tool with dark-background option selection on ImageJ was applied. A region of interest centred on the fovea, corresponding to a circle of 3 mm diameter was selected; the area outside the circle was coloured as blue. The FAZ area was outlined at SCP and DCP level and coloured as blue (not shown). White pixels were measured as vessel, black pixels as background, and blue pixel were eliminated from the analysis.
A 3×3 OCT-A (Zeiss AngioPlex, CIRRUS HD-OCT 5000, Carl Zeiss Meditech, Dublin, Ohio, USA) scan, relying on optical microangiography algorithm, was recorded for each patient.19 20
Images with Signal Strength Index (SSI) below 6 were excluded from the analysis. To validate the reliability of the examination, average SSI from each group has been compared. Automated segmentation into the SCP and DCP inner retinal vascular plexuses, outer avascular retina and CC were performed.21 Two trained retina specialists (MVC and AR) independently performed the qualitative analysis of the images for segmentation. Segmentation artefacts were manually adjusted in cases of automatic software errors.
All 3×3 OCT-A images were digitally analysed as previously described (figure 1).14
Variables included in the analysis were: age, sex, BCVA, foveal avascular zone (FAZ) area at the SCP and DCP; vessel density in SCP, DCP; CRT; retinal thickness for each segmented layer. The BCVA was expressed as logarithm of the minimum angle of resolution (LogMAR) for statistical purposes. Differences between patients and controls were calculated by means of Student t-test or ANOVA with Bonferroni post hoc correction for continuous variables. The relationship between variables was explored using Pearson (parametric) correlation. All data were expressed as mean±SD, and all tests were two sided. A P value less than 0.05, after adjustment when required, was considered significant. Prism V.6.0 software (GraphPad Software, San Diego, California, USA) was used to analyse the data.
The main outcome of the study was to compare FAZ area at the SCP and DCP and vessel density in the SCP and DCP between patients featuring RPD and ORA versus patients featuring RPD only, drusen and controls. The secondary aim of the study was to compare each measured retinal layer thickness among the different study categories. Finally, CRT and retinal layers’ thickness were correlated to vessel density, in order to find any evidence of vascular damage in patients suffering from RPD.
Results
A total of 68 eyes of 57 patients were enrolled in the study (22 eyes of 17 patients of group 1; 24 eyes of 21 patients of group 2; 22 eyes of 19 patients of group 3), mean age was 75.8±6.3 years. The groups showed no difference in age and gender (P=0.2 and P=0.6, respectively). Of 22 healthy age-matched and sex-matched controls (eight males, 36.4%), 22 eyes were included. Mean BCVA was 0.18±0.14 LogMAR in the study population and 0.01±0.03 LogMAR in the control group; mean refractive error was 0±2.0 D spherical equivalent. Demographics and main clinical characteristics of the patients are listed in table 1.
Demographics and main clinical characteristics of study population
The OCT-A SSI was comparable among all the groups (ANOVA P=0.2). At the SCP, the vessel density turned out to be lower in RPD and RPD+ORA patients with respect to controls (P=0.02 and P=0.003, respectively), with no significant difference between the two study groups. No difference was found between RPD and RPD+ORA patients compared with patients featuring drusen; on the other hand, patients with drusen did not differ significantly from the controls. The FAZ showed no statistical difference in all the groups analysed (table 2).
Quantitative comparison of the macular vessel density and the foveal avascular zone (FAZ) between patients and controls
At the DCP, meaningful disparity was found between the study groups and the healthy subjects in the vessel density (P<0.001, P=0.04 and P=0.001 for RPD, RDP+ORA and drusen, respectively); no significant difference was found between RPD and RPD+ORA and patients with drusen. Patients featuring RPD and those featuring drusen showed a larger FAZ compared with controls (P=0.03 and P=0.003, respectively) (table 2) (figures 2 and 3).
- Download figure
- Open in new tab
- Download powerpoint
Multimodal imaging of a 78-year-old patient with reticular pseudodrusen (RPD) and outer retinal atrophy. (A–B) Short-wavelength (488 nm) fundus autofluorescence and colour fundus, showing nearly confluent RPD in the macula and some areas of atrophy at the posterior pole. (C) Structural optical coherence tomography (OCT) disclosing poorly defined collections of hyper-reflective material above the retinal pigment epithelium and attenuated but still visible ellipsoid zone (EZ) and external limiting membrane. (D) Magnification of OCT showing backscattering of the light radiation under areas of major RPD reabsorption (red arrows); note how the EZ disappears and the outer nuclear layer becomes thinner (asterisk).
- Download figure
- Open in new tab
- Download powerpoint
Multimodal imaging of an 83-year-old patient with reticular pseudodrusen and outer retinal atrophy. (A–B) Short-wavelength (488 nm) fundus autofluorescence and colour fundus, showing a well-defined reticular pattern at the posterior pole. (C) Structural optical coherence tomography (OCT) disclosing poorly defined collections of subretinal drusenoid deposits (SDD) with an extremely thin choroid and diffuse backscattering. (D) Greater detail of OCT; a good correspondence can be detected between reabsorbing SDD (red arrow) and outer nuclear layer thinning (asterisks).
To verify if this vascular dysfunction was coupled to intraretinal structural changes, measurement of different retinal layers’ thickness was carried out and compared among the study groups (table 3). CRT, RNFL, INL and OPL disclosed no differences among the study groups and between patients and controls. The GCL turned out to be thinner in all patients affected either by RPD, RPD+ORA or drusen compared with healthy subjects (P=0.02, P=0.03 and P=0.004, respectively). The IPL was abnormal only in patients with drusen compared with controls (P=0.009), while the ONL was altered only in patients with RPD+ORA compared with controls (P=0.007) (figure 4).
Retinal layer thickness of study population evaluated on structural spectral-domain optical coherence tomography
Vascular network differences and correlation between vessel density and retinal thickness between patients with reticular pseudodrusen (RPD) (group 1), RPD+outer retinal atrophy (group 2), drusen (group 3), and controls. (Top left) Box plots showing comparison of the superficial capillary plexus (SCP) density and of the foveal avascular zone (FAZ) area between patients and controls. (Top right) Box plots showing comparison of the deep capillary plexus (DCP) density and of the FAZ area between patients and controls. *= statistically significant value. (Centre) Linear regression between SCP vessel density and innermost retinal layers (retinal nerve fibre layer (RNFL), ganglion cell layers (GCL) and inner plexiform layer (IPL)) showing significant correlation in group 1 (P=0.03) and in group 2 (P=0.002), but not in group 3 and controls. (Bottom) Linear regression between SCP vessel density and central retinal thickness (CRT) showing significant correlation in group 1 (P=0.004) and in group 2 (P=0.04), but not in group 3 and controls.
We correlated the vessel density at SCP with the sum of the RNFL, GCL, and IPL and the vessel density at DCP with the sum of INL, OPL and ONL in each group. We found a statistically significant relationship between the SCP and the innermost retinal layers (r2=0.23; P=0.03) and the CRT (r2=0.37; P=0.004) in eyes with RPD (figure 3). Such a positive correlation was also found in eyes disclosing RPD and ORA (r2=0.40; P=0.002 for the SCP and r2=0.19; P=0.04 for the CRT) (figure 4). No significant correlation between these variables was found in the drusen group and control subjects. Finally, the SCP correlated with BCVA in patients with RPD (r=−0.47; P=0.04) and in patients disclosing RPD+ORA (r=−0.53; P=0.009) or with intact RPE.
Discussion
AMD is not a static disease, but rather a progressive condition leading to severe late-stage retinal and choroidal damage.6 22 23 While there is strong evidence about the alterations occurring in outer retinal anatomy and choroidal vasculature, there is little published literature about the changes in the inner retina in patients with early and intermediate AMD.13 14 Retinal vascular impairment in these patients has been already described by Toto and colleagues: however, they did not mention the impact of RPD on their analysis.24
The current study was aimed to analyse the innermost retinal vascular state in patients featuring RPD and RPD+ORA, using OCT-A and structural SD-OCT. We also included patients featuring drusen, to analyse all the possible phenotypes of non-exudative AMD.
Both SCP and DCP turned out to be reduced in both RPD and RPD+ORA; the DCP only was impaired in patients with drusen with respect to controls. These data show that vascular impairment is a common feature of dry AMD. Our results are partially discordant with those of Toto and associates, who found significant reduction in the superficial vascular plexus flow density between patients with drusen and healthy controls, and no differences in the deep vascular plexus flow density.25 However, this reduction was significant only in patients with dry AMD featuring distinctive external retina changes predictive of the development of drusen-associated atrophy. We did not include these patients in our analysis.
We investigated the relative thickness of the different neural layer perfused by the two major intraretinal vascular networks.26 As far as it regarded the inner retinal layers, the GCL showed a significative reduction in patients featuring RDP, RPD+ORA and drusen with respect to controls. These data are consistent with previous literature and give support to the hypothesis that AMD could be considered as a neurodegenerative disease, characterised by neuroinflammation and irreversible neuronal loss.27–33 The reduction in macular GCL thickness occurring in RPD phenotypes could be attributed to apoptosis of neurons derived from transneuronal degeneration over time, as the input to the inner retina from functionally impaired photoreceptor is chronically reduced.34 In addition, inner retinal ganglion cells may also die from progressive age-related hypoperfusion.35
Interestingly, macular RNFL was partially spared in the study patients. One possible explanation is that nerve fibres may not be involved yet by retrograde degeneration starting from the outermost retinal structures. Moreover, the chronic deficiency of synapses from photoreceptors to RNFL may induce compensative proliferation of interneuronal cells, such as Müller and horizontal cells, that keeps the thickness of RNFL still within normal limits.36
As far as it regarded the outer retinal layers, the ONL turned out to be involved only in patients with RPD+ORA. This observation is consistent with the finding by Spaide, who described progressive shortening of the photoreceptor length in eyes with SDD regression compared with the eyes that did not show SDD regression. The author already noted diffuse thinning of ONL in these eyes, but did not measure this layer thickness specifically.11 General changes of the entire retina are a common feature of different dry AMD phenotypes; they have been described, along with reduction in ONL, also in patients featuring drusen, especially in areas above the drusen.37
Few reports have correlated OCT-A parameters to anatomical and functional variables. We did not find any significant relationship between vessel density at the superficial and the deep capillary level and relative layer thickness of the retina in control subjects. Further analysis of normative database would either confirm or discard this finding. Interestingly, we found a statistically significant correlation between the SCP and the innermost retinal layers and the CRT in eyes with RPD and only with the CRT in eyes disclosing RPD and ORA.
We accomplish that our study has several limitations. First, the correspondence between vascular plexuses and retinal layers may be slightly different from that one presented in the study. Recently, a new classification of OCT-A layers has been proposed, according to which there are up to four retinal vascular networks in the macula. The superficial vascular complex is located primarily in the RNFL and in the GCL, but also includes a part of the IPL. The deep vascular complex includes two vascular networks, above and below the INL, referred to as the ‘intermediate’ (ICP) and DCP, respectively. The fourth network is the radial peripapillary capillary plexus.38 However, the OCT-A machines available in the market do not allow to acquire good quality reconstruction of the deeper vascular plexuses, or to visualise the ICP and the DCP separately. The introduction of a new algorithm of slab segmentation, as the ‘projection-resolved’ OCT-A, would allow visualisation of all the four retinal capillary plexuses known from histology, and quantitative assessment of vessel density at each location.
Furthermore, we accomplish that Henle fibre layer (HFL) may alter the true quantification of ONL in our patients and this should be carefully considered by the readers. In detail, as HFL demonstrates optical properties of birefringence on OCT, they may cause a stronger backscattering when the incidence OCT beam is not perpendicular to the layer orientation.39 The presence of RPD or drusen, distorting the outer retinal architecture, may induce spontaneous visualisation of HFL, mimicking local thinning of the ONL.40
Other limitations include the small number of included eyes, the lack of longitudinal follow-up and the non-inclusion of other AMD phenotypes, as GA or exudative AMD (but it was beyond the scope of this study). Finally, we did not correlate the structural inner retinal changes and alterations in SCP and DCP with clinical features, like RPD stage and number of RPD per area; further investigation will be needed to solve these questions.
In conclusion, patients with non-exudative AMD disclosed relative retinal thinning, particularly at the GCL level, associated with significant retinal vascular loss (more pronounced in those featuring RPD and RPD+ORA). Latest theories have questioned if the RPD spectrum should be considered a clinical entity per se, rather than part of the larger group of AMD.41 Further insights into the vascular and anatomical characteristics of each different phenotype may contribute to a more precise classification of AMD.
Further studies for understanding the primary site affected by RDP would be helpful in tailoring the application of future treatments, such as nutritional supplementation, medical therapy and surgical or parasurgical procedures.
Acknowledgments
FB consultant for: Alcon (Fort Worth,Texas,USA), Alimera Sciences (Alpharetta, Georgia, USA), Allergan Inc (Irvine, California,USA), Farmila-Thea (Clermont- Ferrand, France), Bayer Shering-Pharma (Berlin, Germany), Bausch And Lomb (Rochester, New York, USA), Genentech (San Francisco, California, USA), Hoffmann-La- Roche (Basel, Switzerland), NovagaliPharma (Évry, France), Novartis (Basel, Switzerland), Sanofi-Aventis (Paris, France), Thrombogenics (Heverlee,Belgium), Zeiss (Dublin, USA) Giuseppe Querques consultant for: consultant for: Alimera Sciences (Alpharetta, Georgia, USA), Allergan Inc (Irvine, California,USA), Heidelberg (Germany), Novartis (Basel, Switzerland), Bayer Shering-Pharma (Berlin, Germany), Zeiss (Dublin, USA).
References
Footnotes
Contributors All the authors contributed to the conception or design of the work; the acquisition, analysis and interpretation of data; drafting the work; revising it critically for important intellectual content and gave final approval of the version to be published.
Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Obtained.
Ethics approval San Raffaele Hospital Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- At a glance