Article Text

Download PDFPDF

Choroidal structure in eyes with drusen and reticular pseudodrusen determined by binarisation of optical coherence tomographic images
  1. Federico Corvi1,2,
  2. Eric H Souied2,
  3. Vittorio Capuano2,
  4. Eliana Costanzo2,
  5. Lucia Benatti1,
  6. Lea Querques1,
  7. Francesco Bandello1,
  8. Giuseppe Querques1,2
  1. 1Department of Ophthalmology, Istituto di Ricovero e Cura a Carattere Scientifico (IRCCS) San Raffaele Scientific Institute, University Vita-Salute San Raffaele, Milan, Italy
  2. 2Department of Ophthalmology, University Paris Est Creteil, Centre Hospitalier Intercommunal de Creteil, Creteil, France
  1. Correspondence to Dr Giuseppe Querques, Department of Ophthalmology, Istituto di Ricovero e Cura a Carattere Scientifico (IRCCS) Ospedale San Raffaele, University Vita Salute San Raffaele, Via Olgettina 60, Milan 20132, Italy; giuseppe.querques{at}hotmail.it

Abstract

Purpose To compare luminal and stromal area of the choroid in eyes with drusen and reticular pseudodrusen (RPD) and to investigate their changes over 24 months.

Methods In eyes with drusen and RPD and control subjects, total choroidal, luminal and stromal area were measured on optical coherence tomography B-scans converted to binary images, at baseline and after 24 months.

Results Eighteen eyes of 18 subjects for each group were included. In drusen and RPD, we found reduction of mean total choroidal (p=0.0005 and p<0.0001, respectively), luminal (p=0.003 and p<0.0001, respectively) and stromal area (p=0.007 and p=0.0002, respectively) from baseline to month 24; no change of ratio between luminal–stromal and the choroidal area was recorded. Mean luminal, stromal and total choroidal areas were reduced in RPD, as compared with drusen and controls at both baseline and month 24 (p<0.05 for all). In RPD, the stromal area was more represented, as we found lower mean ratio of luminal and total choroidal area compared with drusen and control at both baseline and month 24 (p<0.05 for all).

Conclusions Mean total choroidal, luminal and stromal area decreased over 24 months similarly in eyes with drusen and RPD. Mean total choroidal, luminal and stromal area were more reduced in eyes with RPD, as compared with eyes with drusen and controls; however, stromal area was more represented in eyes with RPD suggesting a possible role of choroidal vascular depletion and fibrotic replacement in the pathogenesis and disease progression.

  • Imaging
  • Macula
  • Retina
  • Choroid

Statistics from Altmetric.com

Introduction

Age-related macular degeneration (AMD) is a progressive retinal disease with genetic, environmental and constitutional factors.1 ,2 Early stages of AMD are usually asymptomatic and have been characterised, across various classification systems, by the presence of drusen and pigmentary alterations within 2 disc diameters of the fovea.3 Drusen are composed of focal deposits of extracellular matrix and inflammatory components located between the basal lamina of the retinal pigment epithelium (RPE) and the inner collagenous layer of Bruch's membrane. Their formation is thought to be due to the continued outer segment constituents phagocytosis and deposition with an unbalance of production and clearance associated with lipid-rich material deposits.4 ,5 On the other side, reticular pseudodrusen (RPD) were identified as an additional lesion strongly associated with AMD. Based on integrated imaging6–9 and histopathological study,10 it has been proposed that atrophy and fibrosis of the choroid could lead to the derangement of the RPE and secondary accumulation of photoreceptor outer segments above the RPE as RPD deposits (discrete collections of hyper-reflective material). Histological analysis has shown that in eyes with AMD, changes of the choroidal interstitial stroma can occur including oedema, fibrosis and inflammation with cellular infiltration.11 However, the histological processing induces artefacts that makes difficult evaluating the changes in the choroid caused by the disease processes, especially the vascular tone and structure.12 ,13 Enhanced depth imaging spectral domain-optical coherence tomography (EDI SD-OCT) is a non-invasive imaging method to visualise with reliable images the full thickness of the choroid.14 Recently, it has been reported on a new method with high repeatability to differentiate and quantify the choroidal luminal area from the stromal area using an open access software named ImageJ.15

The purpose of this study was to compare the luminal and stromal area of the choroid in eyes with drusen and RPD and to investigate their change over 24 months.

Methods

Study participants

This is a retrospective analysis of patients from two institutions (Department of Ophthalmology of the University Paris Est Creteil, and Department of Ophthalmology of the University San Raffaele in Milan) who first presented between January 2012 and December 2013 with a diagnosis of early AMD, and for whom a minimum follow-up of 24 (±2) months was available. To perform this retrospective observational study, an informed consent was obtained from all subjects in agreement with the Declaration of Helsinki for research involving human subjects. This study was institutional review board (IRB) approved at both sites and was carried out in compliance with local and national IRB guidelines.

Criteria for inclusion were (1) age ≥55 years; (2) presence of five or more medium-large drusen (63–124 μm) within the macula due to early AMD in at least one eye without RPD; (3) RPD (defined by the peculiar yellowish reticular pattern at the macula, whose visibility was enhanced by IR reflectance)10 not accompanied by soft drusen; (4) customised high-resolution EDI SD-OCT (Spectralis Heidelberg Engineering, Heidelberg, Germany) for a minimum follow-up of 24 (±2) months; (5) axial length between 23.5 and 26.5 mm. The exclusion criteria were (1) presence of neovascular AMD in the study eye; (2) presence of atrophy in the study eye; (3) any previous treatment in the study eye including intravitreal injection of antivascular endothelial growth factor agents, photodynamic therapy or laser photocoagulation; (4) history of ocular inflammation in the study eye; (5) significant media opacities; (6) any other retinal disease (such as retinal vein occlusion, diabetic retinopathy or macular dystrophy) in the study eye. Age-matched and sex-matched control subjects with no ocular diseases and axial length between 23.5 and 26.5 mm were also included in the current analysis. In all subjects, a review of paper and electronic medical records was performed. Demographic data and findings from the clinical examination including EDI SD-OCT were collected at baseline and at 24 months.

High-resolution EDI SD-OCT assessment

The method to obtain EDI OCT images has been previously described in detail.14 A 19-horizontal-line protocol (6×6 mm area), each consisting of 1.024 A-scans per line, was performed. Line scans were saved for analysis after up to 100 frames were averaged, using the automatic averaging and eye tracking features of the proprietary device. Given that medium-large drusen are most prevalent in the central macula, while RPD superiorly and inferiorly to the macula,16 in all eyes we analysed the central (the 10th line of this protocol, in fovea), the upper (the first line of this protocol) and the lower line (the last line of this protocol) scans in order to have a more exhaustive overview on macular choroidal changes. Follow-up mode was used to compare the baseline examination at month 24.

Evaluation of total choroidal area, luminal area and stromal area by binarisation technique

The examined area was determined for a large 3000 μm-wide area because preliminary study showed that the sampling of a small area tended to have a large diversity of the ratio of luminal/stromal area.17 EDI SD-OCT images were displayed on a computer screen and evaluated by three masked graders independently (GQ, EHS and FB). When two or more graders determined that the choroidal images at baseline and at month 24 of follow-up were clearly distinguishable, images were deemed acceptable and used for the following analysis by two readers (FC and VC). The upper margin of the region of interest (ROI) was the RPE line and the lower margin was the chorioscleral border of the EDI SD-OCT images. The binarisation of the choroidal area of the OCT image was done by a modified Niblack method as previously reported.15 Briefly, the OCT image was analysed by ImageJ V.1.47 (National Institutes of Health, Bethesda, Maryland, USA; available at http://imagej.nih.gov/ij/). An ROI was selected and set by the ROI manager of the OCT image. The average reflectivity of three choroidal vessels with lumens larger than 100 μm was settled as average brightness to the minimum value to minimise the noise of the OCT image. Then, the image was converted to 8 bits and adjusted by the auto local threshold of Niblack. The binarised image was reconverted to an RGB image, and the luminal area was determined using the threshold tool. After adding the data for the distance of each pixel, the total choroidal, luminal and stromal areas were automatically calculated. The inter-rater agreement was examined for each case by two readers. For each participant, examinations were performed at baseline and at month 24 and each parameter was estimated as the mean of the measurements performed on the three lines of the EDI OCT protocol.

Statistical analyses

Statistical calculations were performed with GraphPad Prism V.5.0 (GraphPad Software, San Diego, California, USA). All variables were tested for normal distributions, according to the Kolmogorov–Smirnov test. The data were summarised with the mean±SD. The significance of differences between patients with drusen and RPD and control subjects was determined by Mann–Whitney rank sum test for continuous data and the χ2 test for categorical data. The differences in patients and controls during follow-up were analysed using Wilcoxon signed-rank test. Univariate linear regression models and Pearson correlation were used to study the relationship among variables (each choroidal structure and the age, sex). All tests were two-sided and a p value <0.05 was considered significant.

Results

A total of 18 eyes from 18 patients with drusen (12 female (66.6%), mean age 76.3±7.2 years) and 18 eyes from 18 patients with RPD (13 female (72.2%), mean age 76.6±10.9 years) met inclusion criteria and were included in the study. A total of 18 eyes from 18 control subjects (12 female (66.6%), mean age 75.4±6.1 years) were also included in the study. The mean follow-up was 24.7±0.7 months in eyes with drusen, 24.8±0.8 months in eyes with RPD and 25±0.8 months in controls.

In eyes with drusen we observed, after 24 months, a significant reduction of mean total choroidal (from 2065908±652018 to 1943579±616875; p=0.0005), luminal (from 1438420±547387 to 1351658±474675; p=0.003) and stromal area (from 627488±141255 to 591920±172466; p=0.007); no change of ratio between luminal–stromal and total choroidal area was recorded (p=0.1 and p=0.1, respectively) (table 1) (figure 1). Similarly, in eyes with RPD we found a significant reduction in mean total choroidal (from 1512251±399661 to 1385979±430932; p<0.0001), luminal (from 1007640±290166 to 924802±335829; p<0.0001) and stromal area (from 504611±121907 to 461176±113626; p=0.0002) at month 24 compared with baseline; no change of ratio between luminal–stromal and total choroidal area was recorded (p=0.1 and p=0.1, respectively) (table 1) (figure 2). In eyes of control subjects, from baseline to month 24, we did not record any significant change of mean total choroidal (from 2113163±620464 to 2033480±529013; p=0.08), luminal (from 1475130±507254 to 1410308±439604; p=0.1) and stromal area (from 637959±155297 to 623171±125901; p=0.4); in a similar fashion we did not find any change of ratio between luminal–stromal and total choroidal area (p=0.9 and p=0.9, respectively) (table 1) (figure 3).

Table 1

Results of choroidal binarisation method analyses in patients with drusen and RPD and in control subjects at baseline and at 24 months

Figure 1

Enhanced depth imaging-optical coherence tomography (EDI-OCT) of the choroid converted to binary images in a patient with drusen at baseline (left panels) and month 24 (right panels). EDI-OCT of the fovea (the 10th line of the protocol) shows drusen, as collections of hyper-reflective material located under the retinal pigment epithelium. The area surrounded by yellow lines defines the region of interest of the choroid (first row). The EDI-OCT was converted to a binary image using ImageJ software, which can allow the visualisation of the luminal area (black) and the stromal area (white) (second row). In the region of interest the luminal area is excluded by the yellow colourisation (third row). Overlay of binary images to EDI-OCT showing the colourised (yellow) stromal area and the non-colourised luminal area (fourth row).

Figure 2

Enhanced depth imaging-optical coherence tomography (EDI-OCT) of the choroid converted to binary images in a patient with reticular pseudodrusen (RPD) at baseline (left panels) and month 24 (right panels). EDI-OCT of the fovea (the 10th line of the protocol) shows RPD, as collections of hyper-reflective material located above the retinal pigment epithelium. The area surrounded by yellow lines defines the region of interest of the choroid (first row). The EDI-OCT was converted to a binary image using ImageJ software, which can allow the visualisation of the luminal area (black) and the stromal area (white) (second row). In the region of interest the luminal area is excluded by the yellow colourisation (third row). Overlay of binary images to EDI-OCT showing the colourised (yellow) stromal area and the non-colourised luminal area (fourth row).

Figure 3

Enhanced depth imaging optical coherence tomography (EDI-OCT) of the choroid converted to binary images in a control subject at baseline (left panels) and month 24 (right panels). EDI-OCT of the fovea (the 10th line of the protocol) shows any retinal alteration. The area surrounded by yellow lines defines the region of interest of the choroid (first row). The EDI-OCT was converted to a binary image using ImageJ software, which can allow the visualisation of the luminal area (black) and the stromal area (white) (second row). In the region of interest the luminal area is excluded by the yellow colourisation (third row). Overlay of binary images to EDI-OCT showing the colourised (yellow) stromal area and the non-colourised luminal area (fourth row).

Interestingly, mean luminal, stromal and total choroidal areas were significantly reduced in eyes with RPD, as compared with eyes with drusen and controls at both baseline and month 24 (table 1). Moreover, in eyes with RPD the stromal area was more represented, as we found lower mean ratio of luminal and total choroidal area compared with eyes with drusen and controls at both baseline and month 24 (table 1). No significant differences were found between eyes with drusen and control subjects at both baseline and month 24 (table 1). Correlation analysis did not reveal any significant relationship at baseline and at month 24 between the three groups (data not shown).

Intrarater agreement was high with an intraclass correlation coefficient (ICC) of 0.990 (CI 0.983 to 0.994) for the total choroidal, 0.991 (CI 0.984 to 0.995) for the luminal and 0.990 (CI 0.982 to 0.994) for the stromal area. The intersession agreement was also high with an ICC of 0.962 (CI 0.954 to 0.968) for the total choroidal, 0.958 (CI 0.95 to 0.961) for the luminal and 0.953 (CI 0.941 to 0.966) for the stromal area.

Discussion

In this study, using the image binarisation method we compared the luminal and stromal area of the choroid in eyes with drusen and RPD and investigated their change over 24 months. In eyes with drusen and RPD we found significant reduction of the mean total choroidal, luminal and stromal area from baseline to month 24; however, no change of ratio between luminal–stromal and the total choroidal area was recorded.

The metabolic support for the RPE and outer retina is provided by the choroid.18 In particular the choriocapillaris has unique features including high flow rate and low oxygen extraction that are mandatory to sustain normal photoreceptor metabolism.19 Any reduction in blood flow of choroidal circulation would have meaningful clinical effects as it is the sole supply of nutrients to the outer retina. One of the negative consequences may be the development of drusen and RPD. A recent study showed that in patients with dry AMD, there is an association between increased drusen extent and decreased choroidal blood velocity and flow.18 Another recent study found that eyes with more drusen had the lowest choriocapillaris density.20 In patients with RPD an overall thinned choroid along with choroidal atrophy and fibrosis underlying RPD was demonstrated, favouring the hypothesis that the choroid may be involved in RPD pathogenesis.7 In the current study, we found progressive thinning of the total choroidal area in patients with drusen and RPD, involving in a similar manner both luminal and stromal area whether we did not find change of the mean ratio between luminal–stromal and the total choroidal area. On the other hand, in control subjects, we did not find any significant change from baseline to month 24. In agreement with previous studies7 the choroid was significantly thinner in eyes with RPD claiming the possible role of choroidal vascular depletion in their development. At both baseline and 24 months, mean luminal, stromal and total choroidal area were reduced in eyes with RPD, as compared with eyes with drusen and controls. Interestingly, in eyes with RPD, we found lower mean ratio of luminal and total choroidal area, and higher mean ratio of stromal and total choroidal area, compared with patients with drusen and control subjects. These findings suggest that the luminal area (vascular component) was less represented in patients with RPD (choroidal vascular depletion). Arnold et al10 in their histopathological report showed a loss of the small choroidal vessels and increased spacing between the large choroidal veins, and proposed that a loss of vascularity and fibrotic replacement of the choroidal stroma may be responsible for RPD. Querques et al7 proposed that in RPD development and progression there may be first a diffuse loss of small choroidal vessels (and thus a diffuse choroidal thinning), and later a fibrotic replacement (and thus a slight thickening) mainly in the area of higher concentration of RPD. The authors also proposed that the derangement of the RPE occurs because of underlying atrophy and fibrosis of the choroid could lead to the accumulation of photoreceptor outer segments above the RPE (subretinal deposits). Our current results support role of choroidal vascular depletion and possible fibrotic replacement in the pathogenesis of RPD.

In RPD eyes, the lower mean ratio of luminal area (not large though significant) is in agreement with the study of Ueda-Arakawa et al21 in which thinned vessels were found in the choroid of eyes with RPD studied by en face images of high-penetration swept source OCT(SS-OCT) suggesting a choroidal involvement in the pathogenesis of RPD. Besides the reduced choroidal thickness at both baseline and month 24, in our study we demonstrated the mean ratio of stromal and total choroidal area to be more represented in eyes with RPD compared with eyes with drusen and controls. These differences in the choroid, with less blood vessels and more stroma in RPD, may be responsible for RPD development, and may help understanding the increased risk for disease progression conferred to eyes with RPD.10

The main limitations of the present study are the small sample size and its retrospective nature. However, despite the limited number of eyes, we obtained significant results. Furthermore, we had strict inclusion criteria for drusen, RPD and control group, with all patients being followed up for 24±2 months.

In conclusion, by means of binarisation of OCT images we demonstrated in vivo that mean total choroidal, luminal and stromal area decreased over 24 months in a similar manner in eyes with drusen and RPD (no changes in mean luminal–stromal ratio). Mean total choroidal, luminal and stromal area were more reduced in eyes with RPD, as compared with eyes with drusen and controls at both baseline and month 24. The stromal area was more represented in eyes with RPD suggesting a possible role of choroidal vascular depletion in the pathogenesis and disease progression.

References

View Abstract

Footnotes

  • Contributors All the authors meet the ICMJE recommendations for authorship credit (substantial contributions to the conception or design of the work, or the acquisition, analysis or interpretation of data; drafting the work or revising it critically for important intellectual content; final approval of the version published; agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved).

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval This study was institutional review board (IRB) approved at both sites and was carried out in compliance with local and national IRB guidelines.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

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.

Linked Articles

  • At a glance
    Keith Barton James Chodosh Jost Jonas