Changes in Stromal and Luminal Areas of the Choroid in Pachychoroid Diseases: Insights Into the Pathophysiology of Pachychoroid Diseases

Abstract

Purpose: To evaluate and compare changes in choroidal vascular and stromal areas in patients with three major pachychoroid diseases for better insight into the pathophysiology of these diseases.

Methods: Eighty-six eyes of 86 patients (50 men and 36 women; mean age, 49.1 years) were evaluated, including 21 patients with chronic central serous chorioretinopathy (CSC), 14 with pachychoroid pigment epitheliopathy (PPE), 19 with pachychoroid neovasculopathy (PNV), 14 with myopic choroidal neovascularization (mCNV), and 18 controls. Multimodal retinal imaging, including enhanced-depth imaging optical coherence tomography (EDI-OCT), was performed. Each EDI-OCT image was binarized with ImageJ software, and luminal (dark pixels) and stromal (light pixels) areas were calculated (3000 μm wide in the subfoveal choroid centered on the fovea).

Results: The subfoveal choroidal thickness (SFCT) was greater in the three pachychoroid groups than in the control group (430.01 vs. 282.61 μm, P < 0.001). There was no significant difference in SFCT among the three pachychoroid groups. The luminal-to-total choroidal ratio (L/C) was highest (ANOVA, P = 0.001) and the stromal-to-total choroidal ratio (S/C) lowest in the CSC group (ANOVA, P = 0.001). Interestingly, stromal area changes were not correlated with SFCT in the CSC and PNV groups, in contrast to the good correlation between luminal area changes and SFCT in these groups.

Conclusions: The eyes of CSC patients had significantly smaller choroidal stromal areas than those of controls or of PPE, PNV, or mCNV patients. The differences in choroidal stromal area, L/C, and S/C in different pachychoroid diseases may reflect different predominant pathogenic processes.

The recent advancements in optical coherence tomography (OCT) imaging, such as enhanced-depth imaging OCT (EDI-OCT), swept-source OCT (SS-OCT),¹ and en face OCT,² have provided good visualization of the choroid. Several methods, including binarization of OCT images,^1–4 have allowed quantitative assessment of the choroid. Changes in the total choroidal area and choroidal vascular (luminal) and stromal areas measured after binarization of OCT images have been reported in patients with several diseases or conditions with increased or decreased choroidal thickness^5–7 or other specific choroidal abnormalities.⁸ 

Pachychoroid spectrum diseases, including pachychoroid pigment epitheliopathy (PPE), pachychoroid neovasculopathy (PNV), and central serous chorioretinopathy (CSC), have been described. The phenotypes of these entities are characterized by diffuse or focal areas of increased choroidal thickness, dilated choroidal vessels (pachyvessels), and structural OCT changes, including thinning or absence of choriocapillaris and Sattler's layer overlying the pachyvessels, corresponding to areas where indocyanine green angiography (ICGA) typically shows “choroidal vascular hyperpermeability.”^9–13 Although a number of studies have described various manifestations of these diseases, their pathophysiologies are still debatable. Recently, an increased choroidal vascular area or a dilated choroidal luminal area has been suggested to be a major component of increased choroidal thickness in pachychoroid diseases, and changes in the choroidal vascular area after anti–vascular endothelial growth factor (anti-VEGF) therapy or photodynamic therapy (PDT) in CSC have been demonstrated.^14,15 

However, choroidal stromal changes in these diseases have not been well described, likely because the major manifestation of a thickened choroid is easily identified by dilated large choroidal vessels with hyporeflective vascular structures on EDI-OCT or SS-OCT. In addition, studies of the choroidal stroma have been limited because variable choroidal stromal changes, such as active inflammatory reactions or scarring, fibrosis, and atrophy cannot be well differentiated with current OCT imaging. The choroidal stroma is the connective tissue surrounding blood vessels and is composed of neural tissues, melanocytes, fibroblasts, resident macrophages, dendritic cells, collagen, elastin, and other extracellular components.^16,17 Several studies have reported choroidal stromal changes in the reticular pseudodrusen (RPD) either by histology^18,19 or by en face OCT^20,21 and reported that fibrosis of the choroidal stroma and loss of vascularity were responsible for the development of RPD; RPD has been proposed as a marker of choroidal ischemia on histology. More recently, using en face OCT, diffuse choroidal thinning, loss of the choriocapillaris, and increased choroidal stroma in the inner choroid were demonstrated in eyes of patients with RPD,^20,21 and choroidal fibrosis and ischemia may lead to the formation of RPD.²¹ 

Morphologic aging changes in the choroid have been investigated in both mice and humans. In aged mice, melanocyte abnormalities and fibrotic changes were often observed in Haller's layer.¹⁶ In addition, complement activation and increased leukocyte extravasation in the choroidal tissues of aged mice have been reported.²² In humans, age-related changes in the choroid have also been reported.^23,24 Ruiz-Medrano et al.²⁴ showed that the choroidal area, luminal area, and ratio of the vascular-to-total choroidal area decreased with increasing age, but interestingly, the stromal area remained stable. Kinoshita et al.²⁵ investigated diurnal variations in choroidal thickness and demonstrated that thickness of the luminal area, not the stromal area, mainly affected these diurnal variations. There was no significant variation in the stromal area. 

The aim of the present study was to evaluate and compare changes in both the choroidal vascular and stromal areas in patients with three major pachychoroid diseases, namely, PPE, PNV, and CSC, to gain better insights into the pathophysiology of these diseases. In our previous study,¹³ we suggested that the pathophysiology of chronic venous disease might parallel that of pachychoroid phenotypes. Local hemodynamic disturbances exerted by pachyvessels can cause chronic venous congestion and increased vascular hyperpermeability, and venous capillary pressure can produce extravasation of proinflammatory and prothrombotic proteins and mediators, including fibrinogen (fibrin) and matrix metalloproteinase-2, into the choroidal stroma.^26,27 Then, metalloproteinase activity degrades the extracellular matrix, resulting in focal tissue degeneration or atrophy, mainly in the choroidal stroma. Therefore, we hypothesized that there might be significant changes not only in the luminal area but also in the stromal area in patients with pachychoroid diseases compared to those in controls without pachychoroid phenotypes. Patients with myopic choroidal neovascularization (mCNV) were also included to compare data from patients with pachychoroid diseases. Significant differences in the stromal area between different pachychoroid diseases were demonstrated in the present study. 

Methods

Patients

Eighty-nine eyes from 89 patients diagnosed with PPE, PNV, or chronic CSC at Konkuk University Medical Center, Seoul, Korea, from January 1, 2013, to December 31, 2017, were included. For this retrospective study, out of these 89 patients, we reviewed the medical records for 54 eyes of 54 patients (14 PPE, 19 PNV, and 21 chronic CSC). Six eyes were excluded because the patients had uncontrolled systemic diseases, such as ischemic heart disease, chronic kidney disease, or systemic inflammatory disease. Twenty-four eyes were excluded for the following reasons: vitreous or retinal hemorrhage, diabetic retinopathy, epiretinal membrane, glaucoma, previous intraocular surgery (except for cataract and refractive surgeries), a history of uveitis, past treatments such as laser photocoagulation or intravitreal injection, PNV with obvious polypoidal lesions, and PNV with evidence of CSC. We also excluded five eyes due to poor EDI-OCT image quality with motion artifacts and projection artifacts that made binarization of the choroid structure difficult. We also reviewed 14 mCNV eyes and 18 normal eyes from age- and sex-matched control patients who visited our clinic, and had symptoms of vitreous floaters and had no other retinal diseases, as control groups to compare with patients in the pachychoroid disease groups. Finally, 86 eyes from 86 patients and controls (14 PPE, 19 PNV, 21 chronic CSC, 14 mCNV, and 18 normal eyes) were included in the present study. All patients underwent a standard ophthalmologic examination at every follow-up, which included measurement of intraocular pressure, best-corrected visual acuity (BCVA), slit-lamp examination, fundus examination, fundus photography with the Topcon TRC-50IX (Topcon Medical Systems, Inc., Paramus, NJ, USA) or the Optos P200Tx (Optos, Inc., Dunfermline, Scotland, UK) and EDI-OCT (Spectralis HRA + OCT2 device; Spectralis, Heidelberg Engineering, Heidelberg, Germany). All other tests (fluorescein angiography [FA], ICGA, and OCT angiography [OCTA]) were reviewed to determine the patients' status and decide the subgroups. All patients who were diagnosed with pachychoroid diseases were divided into three groups (PPE, PNV, and CSC) by three retinal specialists. In cases of disagreement in dividing patients into groups, these specialists held face-to-face meetings to review the patients' charts and images together to reach a consensus. We used the definitions of pachychoroid diseases based on previous publications.^9–13 Briefly, eyes with PPE exhibited pachyvessels without neovascular or polypoidal lesions, no history of CSC, and retinal pigment epithelium (RPE) changes, such as sub-RPE drusen-like deposits, and small serous pigment epithelial detachments. PNV was diagnosed based on the presence of exudative macular disease in patients with no previous history of CSC. For the CSC group, we selected patients who had persistent symptoms for more than 6 months, as described by Kinoshita et al.¹⁵ Since type 1 CNV is a well-recognized complication of chronic CSC, we excluded PNV patients with any evidence of antecedent neurosensory detachment attributable to CSC. Patients with definitive polypoidal lesions were also excluded. All participants completed at least 12 months of follow-up. The study was performed in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board/Ethics Committee at Konkuk University Medical Center. 

EDI-OCT Image Binarization

First, we supposed that the choroid fundamentally consisted of stromal and luminal areas. We wanted to determine the ratio between the stromal and luminal areas of each group and analyze the results. A 7-horizontal line scan (30° × 5°) centered on the fovea was performed. The distance between each B-scan was 235 μm. We collected only images at the center of the fovea for each subject for binarization. The EDI-OCT images were analyzed with ImageJ software (ImageJ, V.1.47; National Institutes of Health, Bethesda, MD, USA). The examined area was 3000 μm wide in the subfoveal choroid centered on the fovea and extended vertically from the outer border of the RPE to the choroid–scleral border. The examined area was manually selected from the RPE to the choroid–scleral border. The three largest choroidal vessels in each cropped image were manually selected with the Oval Selection Tool on the ImageJ toolbar. The average brightness of the three areas was set as the minimum value to minimize noise in the EDI-OCT image. Then, the image was converted to an 8-bit image and adjusted with the Niblack auto local threshold. The binarized image was converted to a Red-green-blue (RGB) image again, and the luminal area was determined using the Threshold Tool. The light area was defined as the stromal area, and the dark area was defined as the luminal area. After adjusting each image by adding the data of the distance of each pixel (scaling for X was 11.18 μm/pixel, and scaling for Z was 3.87 μm/pixel), the luminal and stromal areas were automatically calculated. All these processes were performed according to the methods described by Sonoda et al.²⁸ (one example of our image-processing procedure is depicted as Supplementary Material S1). The subfoveal choroidal thickness (SFCT) and central outer plexiform layer plus outer nuclear layer thickness (OPL/ONL thickness) were manually measured using the caliper function on the ImageJ toolbar. The OCTA images in the choriocapillaris and choroidal layers were binarized in a representative case from each group. Each layer slab was based on default slabs of OCTA software preset using an automated segmentation algorithm. The upper boundary for the choriocapillaris layer was set to Bruch's membrane, and the lower boundary was less than 20 μm from Bruch's membrane. The upper boundary for the choroidal layer was set to less than 20 μm from Bruch's membrane, and the lower boundary was less than 100 μm from Bruch's membrane. All binarization processes and thickness measurements were primarily performed by a trained ophthalmologist (ML) and confirmed by a retinal specialist (HC). All the patient data and binarization EDI-OCT measurements were gathered at the initial visit. If patients with PNV, CSC, or mCNV had intravitreal injections of anti-VEGF, additional binarization of EDI-OCT images was performed at the last visit to compare with the baseline. 

Statistical Analysis

Statistical analysis was performed with IBM SPSS software version 18 for Windows (IBM Corp., Chicago, IL, USA). The χ² test and Fisher's exact test were performed to compare sex ratios, hypertension, diabetes mellitus (DM), and smoking status. Analysis of variance (ANOVA) was performed to evaluate differences in age, spherical equivalent, luminal-to-choroidal ratio (L/C), stromal-to-choroidal ratio (S/C), SFCT, and OPL/ONL thickness. P < 0.05 was considered to be statistically significant. Levene's test was also performed to determine the homogeneity of variances because the post hoc test is dependent on the homogeneity of variance. If the P value of Levene's test was greater than 0.05, the variance was assumed to be homogenous, and the Bonferroni correction method was adopted for the post hoc test. If the homogeneity of variance was not reached, we used Dunnett's T3 test as the post hoc test. Correlation analysis between SFCT and choroidal components was performed with Pearson's correlation coefficient. Comparisons between the ratios before and after treatment for each group (CSC, PNV, mCNV) were performed with paired t-tests. Graphs were created using Prism 7 software (GraphPad Software, San Diego, CA, USA) and Excel (Microsoft Corp., Seattle, WA, USA). 

Results

Baseline Demographic Data

Eighty-six eyes of 86 subjects, including 50 men and 36 women, were reviewed in the present study. The inclusion criteria and methods used by the three retinal specialists for classification of patients into five groups are described in the Methods section. The demographic characteristics of the subjects are shown in Table 1. The average age of all subjects was 48.07 ± 14.39 years. Patients in the PNV group were significantly older than those in the other groups (ANOVA, P = 0.001). The spherical equivalent was not significantly different among the four groups, except for the mCNV group (ANOVA, P < 0.001). The number of patients who currently smoked in the CSC group was higher than that in other groups (χ² test, P = 0.021). There was no significant difference in the number of patients with DM or hypertension among all groups (χ² test, P = 0.260 and P = 0.126, respectively). 

Table 1

 

Baseline Patient Characteristics

 

Evaluation of Choroidal Area, Luminal Area, and Stromal Area by Binarization

The SFCT was greater in the three pachychoroid disease groups than in the nonpachychoroid disease groups (control and mCNV groups) (P < 0.001); however, there was no significant difference in SFCT among the three pachychoroid groups (ANOVA, P > 0.341). There was a significant difference in OPL/ONL thickness among groups (ANOVA, P < 0.001). The OPL/ONL in the PNV and CSC groups was significantly thinner than that in the control group (P = 0.001, P < 0.001, respectively), whereas the OPL/ONL in the PPE and mCNV groups was similar to that in the control group (P = 0.578, P = 0.074, respectively) (Table 2; Fig. 1). 

Table 2

 

Comparisons of the Ratios of Choroidal Structural Components and the Thickness of the Choroid and Outer Retina for Each Disease on EDI-OCT Binarization

 

Figure 1

 

Comparisons of the thickness of the choroid and outer retina in each disease and the ratio of choroidal structural components on enhanced-depth imaging optical coherence tomography (EDI-OCT) binarization images. (A) The subfoveal choroidal thickness (SFCT) in all three pachychoroid disease groups was higher than that in the control or mCNV groups, although no significant differences were found among the three pachychoroid groups. (B) The outer plexiform layer plus the outer nuclear layer thickness (OPL/ONL thickness) in the control group was significantly higher than that in the PNV and CSC groups. However, there was no significant difference between the control and PPE groups. (C, D) The luminal-to-total choroidal (L/C) and stromal-to-total choroidal (S/C) ratios. In the CSC group, the L/C ratio was the highest whereas the S/C ratio was the lowest among all the groups (n.s., no significant difference; asterisk indicates significantly different [P < 0.05]).

The L/C ratio determined by binarization of the EDI-OCT images was highest in the CSC group among all groups (0.767 ± 0.066 vs. 0.707 ± 0.026, 0.703 ± 0.023, 0.694 ± 0.057, and 0.681 ± 0.076 for the CSC, control, PPE, PNV, and mCNV groups, respectively; ANOVA, P = 0.001). However, the L/C ratio in the PPE and PNV groups was not different from that in the control group. The S/C ratio was lowest in the CSC group among all groups (0.233 ± 0.066 vs. 0.293 ± 0.026, 0.297 ± 0.023, 0.306 ± 0.057, and 0.319 ± 0.076 for the CSC, control, PPE, PNV and mCNV groups, respectively; ANOVA, P = 0.001). Again, the S/C ratio in the PPE and PNV groups was not different from that in the control group. Interestingly, we noticed that changes in the stromal area were not necessarily correlated with SFCT in all groups, contrary to the good correlation between changes in the luminal area and SFCT in all groups (R = 0.559, P = 0.016 in the control group; R = 0.828, P = 0.001 in the PPE group; R = 0.713, P = 0.001 in the PNV group; and R = 0.652, P = 0.001 in the CSC group; Pearson's correlation coefficient) (Fig. 2). In the CSC and PNV groups, the stromal areas were not correlated with SFCT (R = 0.431, P = 0.052 in the CSC group and R = 0.332, P = 0.165 in the PNV group). The stromal area and SFCT in the control and PPE groups were well correlated (R = 0.496, P = 0.036 in the control group and R = 0.743, P = 0.002 in the PPE group). 

Figure 2

 

Relationship between SFCT and choroid components (A) lumen and (B) stroma in each group. Luminal areas were positively correlated with SFCT in all groups. However, a significant positive correlation between stromal areas and SFCT was found only in the control and PPE groups. There was no significant correlation between the stromal area and SFCT in the CSC and PNV groups (R = 0.431, P = 0.052 in the CSC group and R = 0.332, P = 0.165 in the PNV group; Pearson's correlation coefficient).

Of a total of 21 CSC eyes, 12 patients were treated with anti-VEGF intravitreal injections (mean of 3.58 injections with bevacizumab [1.25 mg/0.05 mL]), and 3 patients were treated with focal laser therapy. All 19 patients with PNV were treated with anti-VEGF intravitreal injections (mean of 5.11 injections with bevacizumab [1.25 mg/0.05 mL], ranibizumab [0.5 mg/0.05 mL], and/or aflibercept [2 mg/0.05 mL]). Of the 14 mCNV eyes, 11 patients were treated with anti-VEGF intravitreal injections (mean of 2.07 injections with bevacizumab [1.25 mg/0.05 mL]). One-way ANOVA also showed no significant differences in all choroid components and ratios before and after treatments in all the groups. There was also no significant difference in SFCT in any of the groups between the initial and final visits (Fig. 3). 

Figure 3

 

Comparisons of SFCT (A), OPL/ONL thickness (B), and L/C and S/C ratios (C, D) for each disease at the initial and final visits (before and after treatment) on EDI-OCT binarization images. All the results did not show any significant differences.

Finally, correlations between OCTA and EDI-OCT images were investigated in representative cases in each group. The OCTA images were autosegmented to the choriocapillaris and choroidal layers. The S/C ratio of 6 × 6-mm OCTA images in these two layers was clearly distinguishable in representative cases with CSC compared with those in other groups (Figs. 4155215521552–8). Multimodal images and binarization EDI-OCT measurements for the representative cases in each group are depicted in Figures 4 through 8. 

Figure 4

 

Case 1. A healthy, 42-year-old, female control patient. The best-corrected visual acuity (BCVA) was 20/20 in the right eye. (A) EDI-OCT image. (B) Converted binary image using ImageJ with the area of interest in the choroid demarcated with a yellow line. The choroidal area was measured at approximately 3000 μm wide with the margins of 1500 μm nasal and 1500 μm temporal from the foveal center. The L/C and S/C ratios were 0.685 and 0.315, respectively. (C) Overlays of EDI-OCT images with a demarcated choroid rectangle (yellow line). The dark area is the luminal area, and the bright area is the stromal area. (D) Red-free photograph with the corresponding EDI-OCT scan area shown in (A). (E–G) Autosegmented slab boundary of the choriocapillaris (red line) (E); OCT angiography (OCTA) image of an autosegmented slab of the choriocapillaris (F); and a converted binary OCTA image of the choriocapillaris using ImageJ (G). The S/C ratio at this level on OCTA was 0.613. (H) Fundus photograph. (I–K) Autosegmented slab boundary of choroidal layer (red line) (I); OCTA image of an autosegmented slab of choroidal layer (J); and a converted binary OCTA image of choroidal layer using ImageJ (K). The S/C ratio at this level on OCTA was 0.558.

Figure 5

 

Case 2. A 51-year-old man with PPE who visited our clinic for routine checkups. The BCVA of his right eye was 20/20. (A) EDI-OCT image. (B) Converted binary image. The L/C and S/C ratios were 0.702 and 0.298, respectively. (C) Overlay EDI-OCT images with a demarcated choroid rectangle (yellow line). (D) Red-free photograph with the corresponding EDI-OCT scan area shown in (A). (E–G) Autosegmented slab boundary of the choriocapillaris (red line) (E); OCTA image of an autosegmented slab of the choriocapillaris (F); and a converted binary OCTA image of the choriocapillaris (G). The S/C ratio at this level on OCTA was 0.608. (H) Fundus photograph. (I–K) Autosegmented slab boundary of choroidal layer (red line) (I); OCTA image of an autosegmented slab of choroidal layer (J); and a converted binary OCTA image of choroidal layer (K). The S/C ratio at this level on the OCTA image was 0.526.

Figure 6

 

Case 3. A 54-year-old woman with PNV complained of decreased vision in her left eye for 1 month. The BCVA in her left eye at the initial visit was 20/25. After six intravitreal injections with aflibercept [2 mg/0.05 mL], her vision was improved to 20/20. (A) EDI-OCT image. (B) Binarized EDI-OCT image. The L/C and S/C ratios were 0.704 and 0.296, respectively. (C) Overlay EDI-OCT images with a demarcated choroid rectangle (yellow line). (D) Red-free photograph with the corresponding EDI-OCT scan area shown in (A). (E–G) Autosegmented slab boundary of the choriocapillaris (red line) (E); corresponding OCTA image of the choriocapillaris slab (F); and a binarized OCTA image of the choriocapillaris slab (G). The S/C ratio was 0.666. (H) Fundus photograph. (I–K) Autosegmented slab boundary of choroidal layer (red line) (I); corresponding OCTA image of choroidal layer slab (J); and a binarized OCTA image of choroidal layer slab (K). The S/C ratio was 0.528.

Figure 7

 

Case 4. A 46-year-old man with chronic CSC who experienced blurred and distorted vision in his right eye for 2 years. He had a history of smoking (20 packs/year). (A) EDI-OCT image. (B) Binarized EDI-OCT image. The L/C and S/C ratios were 0.773 and 0.227, respectively. (C) Overlay EDI-OCT images with a demarcated choroid rectangle (yellow line). (D) Red-free photograph with the corresponding EDI-OCT scan area shown in (A). (E–G) Autosegmented slab boundary of the choriocapillaris (red line) (E); corresponding OCTA image of the choriocapillaris slab (F); and binarized OCTA image of the choriocapillaris slab (G). The S/C ratio was 0.440. (H) Fundus photograph. (I–K) Autosegmented slab boundary of choroidal layer (red line) (I); corresponding OCTA image of choroidal layer slab (J); and a binarized OCTA image of choroidal layer slab (K). The S/C ratio was 0.400.

Figure 8

 

Case 5. A 49-year-old man with mCNV who experienced decreased vision in his right eye for 5 years. The BCVA at the initial visit was 20/40. He was treated three times with intravitreal injections of bevacizumab [0.25 mg/0.05 mL]. His vision improved to 20/32. There was no recurrence of macular edema or fluid for 3 years after the last treatment. (A) EDI-OCT image. (B) Binarized EDI-OCT image. The L/C and S/C ratios were 0.673 and 0.327, respectively. (C) Overlay EDI-OCT images with a demarcated choroid rectangle (yellow line). (D) Red-free photograph with the corresponding EDI-OCT scan area shown in (A). (E–G) Autosegmented slab boundary of the choriocapillaris (red line) (E); corresponding OCTA image of the choriocapillaris slab (F), and a binarized OCTA image of the choriocapillaris slab (G). The S/C ratio was 0.600. (H) Fundus photograph. (I–K) Autosegmented slab boundary of choroidal layer (red line) (I); corresponding OCTA image of choroidal layer slab (J); and a binarized OCTA image of choroidal layer slab (K). The S/C ratio was 0.471.

Discussion

To our knowledge, this is the first study to evaluate and compare choroidal luminal and stromal areas among patients with different pachychoroid diseases using binarization of OCT images. This study demonstrated a significantly lower S/C ratio in the CSC group than in the PPE, PNV, and control groups, although there was no significant difference in SFCT among all three pachychoroid groups. This significant reduction in stromal area in the CSC group might be simply attributed to a higher L/C ratio or increased luminal area in this group compared with the L/C ratio or luminal area in patients with other pachychoroid diseases or controls, accompanied by compression of dilated large vessels in the outer choroid. However, interestingly, we noticed that changes in the stromal area were not correlated with SFCT in the CSC and PNV groups, contrary to the good correlation between changes in the luminal area and SFCT in these groups. In eyes of healthy controls and patients with PPE, both the luminal and stromal areas were well correlated with SFCT (Fig. 2). Other studies have demonstrated that choroidal vascular areas at the large choroidal vessel level were larger in eyes of patients with CSC than in age-matched eyes.¹⁴ In addition, the choroidal vascular area (luminal area) was positively correlated with the total choroidal thickness, consistent with our findings. These authors showed that there was no difference in choroidal vasculature or choroidal thickness between active and resolved eyes with CSC, which is also consistent with our results; however, some other authors have reported that the outer choroidal vascular dilatation was reduced after anti-VEGF therapy²⁹ or PDT.¹⁵ Our results showed that both choroidal vasculature and stromal components remained unchanged in patients with CSC or PNV after anti-VEGF treatment during a relatively short-term follow-up period (Fig. 3). CSC is one of the most widely studied diseases both clinically and experimentally among macular diseases; however, the etiology and pathophysiology of CSC have not yet been determined. Previous studies have indicated that the primary pathogenesis of CSC is choroidal vascular disturbances based on ICGA findings.^30,31 More recently, Schubert et al.³² demonstrated that genetic variations in cadherin 5 (CDH5) in CSC patients and corticosteroids may trigger the development of CSC in genetically predetermined patients.³² According to these authors, steroid-induced suppression of CDH5, a major cell–cell adhesion molecule in vascular endothelial cells, may alter the permeability/cell shape/intracellular connections of large choroidal vessels and thus may relate to vessel dilation in CSCs. Our results showed that a reduced stromal area might indicate atrophied stroma in eyes of patients with CSC, suggesting that chronic low-grade inflammation caused by chronic venous congestion and subsequent tissue atrophy might be involved in the pathophysiology underlying this disease. If tissue stress, such as chronic venous congestion or vascular hyperpermeability and subsequent extravasation of proinflammatory and prothrombotic proteins and mediators into the choroidal stroma, including fibrinogen (fibrin) and metalloproteinase-2, occurs for a prolonged period, extracellular matrix destruction results in tissue degeneration and stromal atrophy.^{13,26,27,33,34} Fibrous replacement of the choroidal stroma was also identified in eyes of patients with RPD or hypertension and long-standing DM.¹⁸ 

It is still unknown whether choriocapillaris atrophy or obliteration is the primary event or whether inner choroidal attenuation is secondary to dilated pachyvessels.³⁵ Based on the findings in the present study, we speculate that choriocapillaris obliteration or atrophy might be the primary event in pachychoroid diseases. The fact that the L/C ratio in the PPE and PNV groups was not different from that in the control group suggests that the choroid thickened in proportion to increases in luminal areas; thus, it is difficult to say that choriocapillaris obliteration and ischemia are mainly caused by mechanical compression of dilated pachyvessels. Impaired choroidal vascular autoregulation with steroids, catecholamines, or sympathomimetic medicines can cause CSC,³⁶ and repeated intravenous epinephrine injections in monkeys have been shown to result in damage to the choriocapillaris endothelium and other features of human CSC.³⁷ Saito et al.^38,39 also proposed a possible mechanism of CSC in response to increased sympathetic activity: Sympathetic α-adrenoceptor activation leads to local vasoconstriction of the choroidal arterioles, and this functional blood flow resistance results in disturbed perfusion to the choriocapillaris. They also described that this leads to secondary passive overflow into the surrounding large choroidal veins, leading to changes that are representative of the characteristics of dilated large choroidal vessels in pachychoroid diseases. Histopathologic studies in healthy subjects over 40 years of age or in patients with polypoidal choroidal vasculopathy showed arteriosclerosis in the choroidal vessels, although arteriosclerosis in the retinal vessels was not prominent.^40–42 Disappearance of the choriocapillaris, even in areas of RPE preservation, was identified in patients with polypoidal choroidal vasculopathy.⁴¹ Thus, we assume that choriocapillaris obliteration or ischemia caused by either sympathetic α-adrenoceptor activation in young patients or arteriosclerosis in older patients may cause RPE damage leading to PPE. We also hypothesize that if sympathetic activation is persistent, venous stasis or congestion may be aggravated, causing an increased number of dilated large choroidal vessels accompanied by atrophic stroma derived from chronic low-grade inflammation and the extravasation of tissue-destructive cytokines. These features are consistent with our findings in the CSC group, in which the largest L/C ratio and smallest S/C ratio were identified among all the pachychoroid disease groups. On the other hand, if ischemia at the level of choriocapillaris is a more prominent feature possibly due to older age or atherosclerosis, secretion of VEGF increases, leading to the development of choroidal neovascularization, namely, PNV (Fig. 9). In the PNV group, both the L/C and S/C ratios were not different from those in the control group. Intraocular VEGF levels in PNV have never been reported, whereas one study reported that there was no significant difference in aqueous VEGF levels between patients with CSC and controls,⁴³ suggesting that increased VEGF levels were not the major etiology for the development of CSC. 

Figure 9

 

Proposed mechanism for the development of pachychoroid diseases.

The current study has several limitations. First, the sample size was relatively small, with approximately 20 patients in each pachychoroid disease group. However, we adopted strict criteria to classify and distinguish the three pachychoroid diseases because clear identification of each disease was an important precondition for this study. Second, the current study was carried out with image binarization to distinguish lumen from stroma in choroid tissue using ImageJ software (Niblack method). However, no method can perfectly reflect the actual anatomic structure of the choroid due to the arbitrary nature of the threshold assigned to dark and light areas in the choroid. In addition, unlike a study by Sonada et al.⁶ in which the choroid was divided into inner and outer choroidal areas after binarization, we did not discriminate the inner and outer choroid for our analysis of L/C and S/C ratios because we focused more on the luminal and stromal areas of the outer choroid, and the influence of the inner choroid on the total luminal and stromal areas of choroid was negligible. We performed additional steps for binarization of the choriocapillaris and choroidal layer with OCTA images of representative cases from each group and were able to show that the S/C ratio was substantially lower in the CSC group than in the other groups. However, the number of obtained OCTA images for each group was too small to perform statistical analysis, and the autosegmented layer slabs might not be accurate. More OCTA images must be obtained for each group, and further studies using OCTA images might be needed to identify each component of the choroid with greater accuracy. 

In conclusion, a markedly decreased S/C ratio as measured by binarization of EDI-OCT images in patients with CSC suggests that chronic low-grade inflammation and subsequent tissue atrophy might be involved in the pathophysiology underlying this disease. In addition, these differences in the choroidal stroma as well as changes in the L/C and S/C ratios among patients with different pachychoroid diseases may reflect different predominant pathogenic processes in these diseases. 

Acknowledgments

Supported by the National Research Foundation of Korea (NRF) and funded by the Ministry of Science and ICT (NRF- 2017R1E1A1A01073964). 

Disclosure: M. Lee, None; H. Lee, None; H.C. Kim, None; H. Chung, None