Effect of Choroidal Vessel Density on the Ellipsoid Zone and Visual Function in Retinitis Pigmentosa Using Optical Coherence Tomography Angiography

Abstract

Purpose: We evaluate the effect of choroidal vessel density on the residual length of the ellipsoid zone (EZ) and visual function in patients with retinitis pigmentosa (RP) using optical coherence tomography angiography (OCTA).

Methods: Fifty-three patients with RP (n = 101 eyes) and 53 normal participants (n = 76 eyes) were enrolled in this study. Patients with RP were assigned to three groups according to their best-corrected visual acuity (BCVA). All patients underwent ophthalmologic examinations, including BCVA, fundus examination performed with a slit-lamp using an indirect 90 diopter (D) lens, OCTA, full-field electroretinogram (ERG), and visual field. The choroidal vessel density in the choriocapillaris-Sattler's layer (D_C-S), Haller's layer (D_H), horizontal length of the ellipsoid (HEL), and vertical length of the ellipsoid (VEL) were assessed using OCTA and Adobe Photoshop CS3 extended software.

Results: A significantly increasing impairment of choroidal vessel density (D_C-S and D_H) was characterized in the RP groups compared to those of the controls (P < 0.05 for all). The magnitude of the reduction in the DC-S and DH was much easier to identify for more severely impaired BCVA in the RP groups (P < 0.05 for all). The D_C-S had the strongest correlation with the HEL, VEL, BCVA, visual field, and b-wave amplitude (r = 0.735, r = 0.753, r = −0.843, r = 0.579, and r = 0.671, respectively).

Conclusions: Using noninvasive OCTA, choroidal microcirculation, especially in the small/middle choroidal vessel layers, was a prominent factor affecting the EZ, visual acuity, visual field, and recordable ERG b-wave amplitude of patients with RP. This may provide new insights into the progress mechanism and treatment of RP.

Retinitis pigmentosa (RP) is a heterogeneous hereditary disease involving the progressive degeneration of photoreceptor cells that eventually causes blindness.¹ The worldwide prevalence of RP is 1 in 2500 to 7000.² The common characteristics of RP are scattered bone spicule-like pigmentation in the retina, attenuated retinal vessels, and waxy optic discs. The choroidal vessels provide blood and oxygen to nourish the choroid and outer retina, which is free of central retinal vascular system blood vessels. Numerous experimental studies have indicated that ocular microcirculation changes in RP. Subfoveal choroidal blood volume and blood velocity measured by confocal laser Doppler flowmetry have been reported to decrease in patients with RP.³ Reduced retinal and choroidal blood flow in patients with RP have been revealed using magnetic resonance imaging.⁴ A thinner choroid also is an apparent morphologic characteristic of patients with RP.⁵ The choriocapillaris has been found to be lost in RP animal models, such as Royal College of Surgeons rats and retinal degeneration mice by histologic analysis.^6,7 Indocyanine green angiography (ICGA) is the gold standard for visualizing choroidal vasculatures in vivo.⁸ However, it is unable to capture any separated layers of choroidal vasculatures.⁹ More importantly, ICGA is not an ideal choice for follow-up studies and is not suitable for all patients as it is an invasive technique. 

The development of optical coherence tomography angiography (OCTA) for use with split-spectrum amplitude-decorrelation angiography (SSADA), an amplitude-based method, can visualize the retinal and choroidal vascular layers without the injection of dye. It works by detecting and calculating the variation in reflected OCT signal amplitude between consecutive cross-sectional scans using a SSADA algorithm.¹⁰ OCTA can separately show superficial and deep retinal capillary networks, which are rarely visible using fundus fluorescein angiography.¹¹ OCTA also can distinguish choroidal choriocapillaris and medium-sized vessel layers from large-sized vessel layers, which are hard to distinguish using only ICGA. Liu et al.¹² conveniently and noninvasively observed choroidal neovascularization using OCTA. 

Increasing numbers of studies have used OCTA to analyze alterations in the retinal and choroidal networks for some ocular diseases.^13–17 It should be noted that these studies mainly focused on choroidal neovascularization, not on the vascular density of different choroidal layers. Moreover, we are unaware of any correlation between choroidal vessels and visual function reported in previous studies. In our study, we investigated the alteration of choroidal vasculature layers in patients with RP using noninvasive OCTA and evaluated the correlation of choroidal vasculature with the ellipsoid zone (EZ) and visual function. 

Materials and Methods

Study Design

This cross-sectional study was conducted in a clinical practice. The study adhered to the tenets of the Declaration of Helsinki and was approved by the institutional review board of Zhongshan Ophthalmic Center, Sun Yat-Sen University, Guangzhou, China. 

Participants and Examinations

Fifty-three patients with primary RP who were diagnosed and followed up at Zhongshan Ophthalmic Center from August 2014 to August 2017 were enrolled in this study. The inclusion criteria for patients with RP included the following: nyctalopia, typical fundus appearance for RP, narrow peripheral visual field, and reduced or even nondetectable electroretinogram (ERG) b-wave amplitude. Exclusion criteria were patients with RP with spherical refraction >3 diopters (D), obscure refracting media, poor fixation ability, macular edema, secondary RP, any history of eye surgery, and other ocular and systemic diseases. The life quality of patients with RP is determined by visual acuity and visual field.¹⁸ The visual acuity has been used as the grouping criteria for patients with RP in several studies.^19–21 Szlyk et al.²² put forward that the borderline of best-corrected visual acuity (BCVA) for patients with RP was 0.5, for it affects their daily task performances. A patient with a BCVA lower than 0.05 is defined as blind.²³ Therefore, our patients were assigned to three groups according to their BCVA: Group A (BCVA ≥ 0.5), Group B (BCVA ≥ 0.05 to < 0.5), and Group C (BCVA < 0.05). Healthy individuals with a BCVA of 1.0, spherical refraction < 3 D, no other ocular diseases, and no history of eye surgery were enrolled as a normal control. All individuals received comprehensive ophthalmic examinations, including measurement of BCVA, IOP examination using noncontact tonometry (Canon TX-20; Canon, Inc., Tokyo, Japan), slit-lamp examination (YZ5S; Suzhou 66 Vision-Tech Co., Ltd., China), fundus examination performed with a slit-lamp using an indirect 90 D lens, and OCTA (RTVue XR100; Optovue, Fremont, CA, USA). Patients with RP also underwent fundus photography (Topcon TRC-50DX; Topcon Corp., Tokyo, Japan), full-field ERG (Roland Consult, Wiesbaden, Germany), and 30–2 threshold visual field testing (Carl Zeiss Meditec, Inc., Dublin, CA, USA). 

OCTA Images

OCTA images were captured through dilated pupils. Cross-sectional scans and three-dimensional OCTA scans were acquired from a 6 × 6 mm macular area for each subject. To enhance the visualization and image quality of the vessel density in the region of interest, the three-dimensional angiogram was projected separately into en face views using an automated algorithm in four layers as follows: the superficial retinal capillary plexus, deep retinal capillary plexus, outer avascular retina, and choriocapillaris layer.¹⁴ The choroidal vascular structure includes the choriocapillaris, Sattler's layer, and Haller's layer, which are considered as small-, medium-, and large-sized vessel layers, respectively. Visualization of the Sattler's and Haller's layers was enabled by the function of the SSADA-based OCT.^24–26 Because there are no definite criteria for the segmentation of the three choroidal layers, we segmented them by their vessel ratios and lumen sizes. Due to projection artifacts, the choriocapillaris layer was not able to be separated accurately and calculated in patients with RP; therefore, we observed the choriocapillaris layer and the Sattler's layer together and treated these two layers as one C-S layer in the study. The scan thickness was set in the range of 20 to 30 μm to locate the scanning image in the middle of the C-S and Haller's layers to obtain the most distinctive images of the choroidal vascular network.⁹ OCTA images with poor quality were excluded according to any of the following objective criteria: artifacts due to the floaters, blinking, or poor fixation, and significant segmentation errors due to the epiretinal membrane.^27,28 

Measurement of the Choroidal Vessel Density

The choroidal vascular networks were calculated twice by two independent individuals masked to the groups to avoid experimental bias. The images were binarized by transforming them into black and white pixels using Adobe Photoshop CS3 Extended software (Adobe Systems, Inc., San Jose, CA, USA). The threshold of the conversion could be adjusted from 0 to 255 (0, black; 255, white) to make the vascular status similar to the primary images. The white pixels represented the stromata and the black pixels represented the choroidal vessel lumens. A value of 35 to 50 worked well to outline the choroidal vascular network. The ratio of black pixels to all pixels in the selected zone was calculated as the vessel density (see Fig. 1).^24,29 The vessel densities of the C-S layer and the Haller's layer were abbreviated as D_C-S and D_H, respectively. 

Figure 1

 

OCTA images of the choroidal vessels in the (A–D) C-S layer and the (E–H) choroidal vessels in the Haller's layer are shown for the control group and the RP A, B, and C groups from left to right. (A1–H1) The OCTA images were converted into black and white pixels using Adobe Photoshop. The black pixels represent the lumens of the choroidal vessels and the white pixels represent the stromata. (I) and (J) show a comparison of the Dc-s and the D_H among the control and RP patients. *Significant difference (P < 0.05) in the Dc-s and the D_H between the control and groups. #Significant difference (P < 0.05) in the Dc-s and the D_H between RP Group A and the other two RP groups. §Significant difference (P < 0.05) in the Dc-s and D_H between RP Groups B and C.

Measurement of the Ellipsoid Zone Length

OCT images for measuring the residual length of the EZ were acquired from the crossline model in B scan. The horizontal (HEL) and vertical (VEL) lengths of the ellipsoid were calculated as the distance of the preserved length across the fovea using Adobe Photoshop (see Fig. 2).³⁰ 

Figure 2

 

(A) The HEL and (B) VEL from the crossline scan of the macular fovea were measured as the residual length of the EZ using Adobe Photoshop CS3 Extended software. The length of the white line indicates the residual EZ.

Visual Field Testing

Reliable performance for visual field testing was defined as <20% fixation errors, <15% false positives, and <33% false negatives.³¹ The visual fields of the patients with RP were tested using a Humphrey Field Analyzer II with a central 30–2 threshold test, size III white stimulus, and SITA-fast module. The mean deviation (MD) of the visual field obtained using Humphrey perimetry autosoftware was used to evaluate the extent of visual field loss as reported previously.³² 

Full-Field Electroretinogram Records

Flash ERGs were undertaken based on the International Society for Clinical Electrophysiology of Vision standards.³³ For the ERG examinations, the scotopic ERG test 0.01, maximal scotopic ERG test 3.0, and photopic 3.0 ERG test, which represent rod response, cone and rod response, and cone response, respectively, were performed separately.³⁴ The b-wave amplitudes in the maximal ERG response were acquired to evaluate the function of the cone and rod responses. 

Statistical Analysis

Statistical analysis was performed using SPSS statistical software, Version 22 (IBM Corp., Armonk, NY, USA). All data are presented as means ± SD. BCVA was converted to a logMAR scale. A χ² test was used to analyze the difference in the sex ratios of the normal group and the three RP groups. Nonparametric Kruskal–Wallis tests were used to analyze the difference in the choroid structures and EZs of the normal group and the three RP groups. Spearman's rank correlation coefficients were conducted to analyze the correlation of the D_C-S and the D_H with the HEL, VEL, BCVA, MD of the visual field, and ERG b-wave amplitude. P < 0.05 was considered statistically significant. 

Results

Patient Demographics

A total of 53 patients with primary RP (101 eyes) and 53 normal subjects (76 eyes) were enrolled in this study. The main demographic characteristics of the healthy controls and patients with RP are presented in Table 1. There was no statistically significant difference in age, sex, or spherical refraction between the normal subjects and patients with RP (P = 0.185, P = 0.905, and P = 0.243, respectively). The EZ, BCVA, visual field, and ERG b-wave amplitude all declined in the RP groups compared to the normal subjects (P < 0.05 for all). Seven eyes of four patients (15.6%) in RP Group A, five eyes of three patients (12.2%) in RP Group B, and one eye of one patient (6.7%) in RP Group C had rod-cone responses. Rod and cone responses were nondetectable in the other 45 patients. 

Table 1

 

Demographic Data and Vision Function of Controls and Patients With RP

 

Analysis of Choroidal Vessel Density

The choroidal vessel densities for the normal and three RP groups are presented in Figure 1. Compared to the normal controls, there were overall reductions in the D_C-S and D_H in the RP groups (P < 0.05 for all). The D_C-S and D_H were reduced among the RP groups as visual acuity worsened (P < 0.05 for all). In addition, vessels in the C-S layer in the three RP groups showed a clutter distribution compared to the controls. 

Correlation Analysis

Spearman's tests (Table 2) revealed that the HEL and VEL had a better relationship with the D_C-S than the D_H (r = 0.735 vs. 0.668 and r = 0.753 vs. 0.702, respectively; both P < 0.05). The D_C-S and D_H were significantly correlated with BCVA (r = −0.843 and r = −0.822, respectively; P < 0.05 for all) and MD (r = 0.579 and r = 0.557, respectively; P < 0.05 for all). The D_C-S had a higher coefficient, which suggested that higher D_C-S might be more attributable to the EZ, BCVA, and visual field. Moreover, the D_C-S (r = 0.671) was better correlated with recordable ERG b-wave amplitude. 

Table 2

 

Correlation of Choroid Vessel Structures With the EZ and Visual Function

 

Discussion

The choroid has an important role in nourishing the outer retina, particularly photoreceptor layers. Vascular attenuation in patients with RP has been found to correlate with the degeneration of photoreceptors through the histopathologic study of donor's eyes.³⁵ In addition, various studies have revealed that choroidal blood flow and velocity associated with visual function are reduced in patients with RP using magnetic resonance imaging, Doppler ultrasound, confocal laser Doppler flowmetry, or automated perimeter.^3,4,36–40 Besides vessel caliber, blood flow, and velocity, vessel density also is vital for microcirculation. 

SSADA-based OCT provides a promising new path for clearly distinguishing the choroidal vessel layers and avoiding the risk of fundus angiography. Our study demonstrated the increasing impairment of choroidal vessel density in the small/middle and large choroidal vascular layers in the three RP groups. In addition, the level of impairment was significantly different in the three RP groups: the magnitudes of the reduced D_C-S and D_H were much easier to identify in the more severe stages of RP. In contrast to our study, Battaglia Parodi et al.¹⁴ demonstrated that the density of the choriocapillaris in controls and patients with RP was similar using OCTA. We hypothesized that the differences in the mean age, BCVA, and scanned area of the subjects between these two studies were possible reasons for these different results. Age is considered to have a large impact on choroidal vessel density. The mean ages of the control group and the three RP groups in our study were 40, 38, 43, and 44 years, respectively, and the mean age of the controls and RP patients in the study of Battaglia Parodi et al.¹⁴ study was 53 years. The mean vessel densities of the choriocapillaris-middle-sized vessel layer in the control and three RP groups in our study were approximately 70% and 60% to 70%, respectively. The mean choriocapillaris density of the control and RP groups in the Battaglia Parodi et al.¹⁴ study was approximately 50%. Younger healthy groups have greater vessel density than older groups,⁴¹ which suggests that loss of choroidal vessel density in younger RP groups might be more sensitive to disease progression. RP lesions begin in the peripheral retina and progress to the central retina. Our scan area was 6 × 6 mm, which was larger than the 3 × 3 mm scan area used by Battaglia Parodi et al.¹⁴ The mean BCVA (logMAR) of the RP patients in Groups A to C were 0.16 ± 1.15, 0.732 ± 0.31, and 1.97 ± 0.57, respectively. The mean BCVA (logMAR) of the RP patients in the study of Battaglia Parodi et al.¹⁴ was 0.5 ± 0.3, which is similar to the RP Group B in our study. Different baseline subject characteristics also might be a reason for the different results of these two studies; however, our results were consistent with the following studies. The choroidal luminal areas beneath the degenerated retina decrease in patients with RP using enhanced-depth OCTA.⁴² Toto et al.⁴³ also found that the choriocapillaris density in patients with RP significantly decreased compared to control groups. 

To the best of our knowledge, no work has been undertaken previously to analyze the correlation of choroidal small/middle and large vessel density with retinal photoreceptor cells and visual function in patients with RP. We found that choroidal vessel density, particularly in the choriocapillaris and Sattler's layers, had a close relationship with the EZ and visual function. Higher vessel density in the C-S layer was accompanied by a longer EZ residual length, better BCVA, and less visual field loss. This can be well explained by the critical choroidal function in the photoreceptor layer.⁴⁴ The density of perfused capillaries has a positive correlation with blood flow.⁴⁵ The vessel density of the choroidal layers may affect the blood flow in the subfoveal choroid. Several studies have demonstrated that choroidal blood flow has a central role in visual function.^3,46–48 Since our study showed that the EZ, BCVA, and visual field were all closely correlated with the D_C-S in patients with RP, the D_C-S could be used to monitor the progress of RP and evaluate treatment effects. Egawa et al.⁴⁹ reported that inner choroidal structures, such as the choroidal luminal area ratio, affect the central visual function in patients with RP using enhanced-depth OCTA. Our study indicated that choroidal microcirculation, particularly in the small/middle choroidal vessel layers, might provide new insights into the mechanism and treatment of RP. The decline of choroidal vessel density is the final outcome of different kinds of RP caused by different genes, and choroidal vessel loss will aggravate the dystrophy and dysfunction of retinal photoreceptors regardless of the type of gene mutation. Therefore, the alteration of choroidal circulation also might predict the progress of RP, and more attention should be paid to improving choroidal circulation to protect visual function in patients with RP. Choroidal revascularization has been reported as able to improve visual acuity and the peripheral visual field in atherosclerotic macular dystrophies.⁵⁰ Nilvadipine, as a vasodilator drug, has been reported to be able to delay the progression of visual field defects in patients with RP.⁵¹ Although these treatments have not proved to have very positive effects until now, more effort on the protection and improvement of choroidal microcirculation in patients with RP still may be worthwhile. 

There were some limitations in our study. The number of patients in RP Group C was small. The measurement of choroidal vessels in our study used an approximation algorithm due to the difficulty in distinguishing the choriocapillaris from the choroidal microvasculature. On the other hand, the 6 × 6 mm scan obtained a larger scan area, but a lower resolution compared to the 3 × 3 mm scan.⁵² Another limitation of this study was that the RP patients' genetic information was unable to be detected in our study. It would be better if the characteristics of the choroidal vessels for different genetic types of RP, such as ABCA4 or MAK-associated RP, could be revealed with the noninvasive OCTA assessment.^53,54 This is worth investigating in the near future. 

In conclusion, the alterations in the choroidal vessel layer in patients with RP could be assessed using noninvasive OCTA. Patients with RP with a more preserved choroidal vascular structure had longer EZs, better vision, and lower visual field loss. The choroidal microcirculation, particularly in the small/middle choroidal vessel layers, was a prominent factor affecting the EZ, visual acuity, and visual field of patients with RP. This may provide new insights into the mechanism and treatment of RP. 

Acknowledgment

Supported by grant from the National Natural Science Foundation of China (81770971) (YL). 

Disclosure: R. Liu, None; J. Lu, None; Q. Liu, None; Y. Wang, None; D. Cao, None; J. Wang, None; X. Wang, None; J. Pan, None; L. Ma, None; C. Jin, None; S.V. Sadda, None; Y. Luo, None; L. Lu, None