Spatial Distribution of Hyperreflective Choroidal Foci in the Macula of Normal Eyes

Abstract

Purpose: To investigate variations in the spatial distribution of hyperreflective foci in the choroid of the macula in normal eyes.

Methods: We included eyes with a normal fundus from patients who had undergone optical coherence tomography angiography, covering a 6-mm × 6-mm area centered on the fovea. The macular area was divided into nine sectors according to the modified Early Treatment of Diabetic Retinopathy Study grid. Hyperreflective choroidal foci (HCF) distribution, choriocapillaris vascular density, and choroidal stromal density were determined on en face images of the choroid in each sector.

Results: We included 35 eyes from 35 participants, with a mean age of 52.7 ± 16.8 years. The mean number and area fraction of HCF at the 5-mm macular area were 35.6 ± 7.8 foci/mm² and 3.0% ± 0.7%, respectively. The number of HCF in the central circle (50.7 ± 20.9 foci/mm²) was greater than that in the inner (35.1 ± 13.0 foci/mm²) or outer rings (35.6 ± 6.5 foci/mm²) (P < 0.001, P < 0.001, respectively). The area fraction of HCF in the central circle (4.84% ± 3.36%) was greater than that in the inner (2.62% ± 1.17%; P < 0.001) or outer rings (3.12% ± 0.67%; P = 0.004). The HCF distribution did not significantly correlate with the choriocapillaris vascular density or choroidal stromal density in each sector.

Conclusions: HCF were more densely distributed in the macular center than in the pericentral or peripheral macular areas.

Translational Relevance: HCF measurement and spatial distribution could provide additional information for evaluating choroidal stromal characteristics.

Introduction

Since its introduction as an ophthalmic imaging diagnostic technique, optical coherence tomography (OCT) has been widely used to investigate variations in the anatomic structure of the retina in normal and diseased eyes. Hyperreflective foci are discrete round or oval-shaped spots that exhibit hyperreflectivity with well-circumscribed margins on OCT images.^1–3 Hyperreflective foci have been observed in the eyes of patients with various chorioretinal diseases. Furthermore, they are used as a measure to predict the severity and progression of retinal diseases.¹ Recent advances in swept-source OCT (SS-OCT) technology and enhanced depth imaging techniques allowed us to obtain detailed choroidal images.⁴ Hyperreflective foci have also been found in the choroids of patients with diabetic macular edema, Stargardt disease, and choroideremia.^5–7 

The choroid is a highly vascularized tissue composed of blood vessels and stromal tissue. Although choroidal vascular changes have been observed in eyes with chorioretinal diseases, choroidal stromal changes have not been widely studied. Variations in the distribution of hyperreflective choroidal foci (HCF) may depend on changes in the choroid.^8,9 Diseases that occur in specific areas of the retina may be accompanied by changes in the corresponding area of the choroid, which induces changes in HCF.^10,11 The distribution of HCF in the macular area of healthy eyes could be an important indicator for identifying changes caused by these diseases. 

Studies are currently under way to explore the occurrence of HCF in specific diseases and changes in their distribution over the course of the illness, as it can manifest in different macular regions.^5–7 However, no reports are available on the distribution of hyperreflective spots observed on OCT in the macular area of normal individuals. Understanding potential differences depending on the macular area in normal eyes can provide essential baseline data for interpreting disease presentation across different macular regions. 

In this study, we investigated variations in the spatial distribution of HCF according to macular location in normal eyes using en face OCT images. 

Methods

This study was reviewed and approved by the Institutional Review Board of the Korea University Anam Hospital and complied with the Declaration of Helsinki. In this retrospective study, we consecutively included eyes without chorioretinal disease in patients who underwent OCT angiography (OCTA) with a nominal scanning area of 6 × 6 mm centered on the fovea. OCTA imaging was conducted for patients with cataracts or epiretinal membrane between October 2022 and April 2023. We selected one eye with better OCTA image quality and no segmentation errors from participants with cataracts or included normal fellow eyes of patients with a unilateral epiretinal membrane. The 6-mm × 6-mm macular OCTA scan was performed using an SS-OCTA device (DRI OCT Triton; Topcon Corp., Tokyo, Japan) with the manufacturer's default settings using a real-time eye-tracking system after pupil dilation. All patients underwent a comprehensive ophthalmic examination, including Snellen visual acuity testing, anterior segment slit-lamp microscopy, intraocular pressure measurement, fundus examination and imaging, OCT, and OCTA. Patients with retinal or choroidal disease history, high myopia (refractive error <−6.0 diopters or axial length ≥27 mm), severe cataract with low image quality or segmentation errors, moderate to severe glaucoma, history of retinal laser treatment, or ocular surgery other than cataract surgery were excluded. In addition, patients with low-quality OCT images (image score <60) or segmentation errors were excluded. 

Measurement of HCF

HCF measurements were performed using en face images derived from structural OCT volume scans, acquired as part of a 6-mm × 6-mm macular OCTA scan protocol. En face structural OCT images were automatically generated using segmentation algorithms provided by the device's built-in software. After confirming the absence of segmentation errors along Bruch's membrane, we modified the default choriocapillaris slab generated using Bruch's membrane–based segmentation. For HCF measurements in the inner choroid at the interface between the choriocapillaris and the medium-sized vessels (Sattler's layer), the modified slab was positioned between 5.2 and 20.8 µm below and parallel to Bruch's membrane,¹⁰ considering the thickness of the hyporeflective choriocapillaris and the device's axial resolution. The HCF were defined as well-circumscribed dots with high reflectivity on en face OCT images. Considering the inability to observe the retinal pigment epithelium (RPE) or nerve fiber layers in en face images, we defined and measured the HCF as spots with reflectivities greater than two standard deviations (SDs) above the mean reflectivity of the slab mentioned above.^8,9,12 HCF were analyzed according to a previously reported method using the ImageJ software (version 1.53m; National Institutes of Health, Bethesda, MD, USA) (Fig. 1).^8,9,12 In brief, en face OCT images were inverted to black and white and converted to 8-bit grayscale images. After conversion, thresholding was performed by setting the threshold below two SDs from the mean reflectance of the entire image per participant. Next, binarization was conducted using the watershed function to separate overlapping points. Subsequently, only choroidal hyperreflective spots with an area ≥314 µm² were calculated for their number and area using the Analyze Particle tool to minimize noise signals measured as HCF. When a discernible large retinal blood vessel was observed in both the fundus photograph and en face OCT image, the area was subtracted from the total area of interest for analysis. 

Figure 1.

 

Measurement of hyperreflective foci in the choroid. The en face slab image of optical coherence tomography covering 6 × 6 mm of the macular area was exported to the ImageJ software (A). It was inverted to black and white and converted to 8-bit grayscale (B). After conversion, thresholding was performed by setting the threshold below the two standard deviations in the mean reflectance, following which binarization was performed using the watershed function to separate overlapping points (C). Subsequently, the density and area fraction of choroidal hyperreflective points were calculated using the Analyze Particle tool (D).

For distribution analysis according to the location in the macula, the macular area was divided into nine sectors, similar to the Early Treatment of Diabetic Retinopathy Study (ETDRS) grid. However, our grid modified the diameter of the outer ring, resulting in a central area with a diameter of 1 mm, four inner pericentral sectors with diameters between 1 and 3 mm, and four outer peripheral sectors with diameters between 3 and 5 mm (Fig. 2). 

Figure 2.

 

Modified ETDRS nine-grid. The central, inner, and outer rings have a diameter of 1, 3, and 5 mm, respectively. Inner and outer rings were further subdivided into four sectors: superior, nasal, inferior, and temporal.

Measurement of Subfoveal Choroidal Thickness, Choriocapillaris Vascular Density, and Choroidal Stromal Density

To evaluate the influence of choroidal structural variations around the HCF on their spatial distribution, we measured the subfoveal choroidal thickness, choriocapillaris vascular density, and choroidal stromal density in the inner choroid.^13,14 Subfoveal choroidal thickness was measured on B-scan SS-OCT images. Subfoveal choroidal thickness was determined at the foveal center by measuring the vertical length from Bruch's membrane to the chorioscleral junction using a built-in caliper tool in a horizontal line scan. 

We measured choriocapillaris vascular density and choroidal stromal density in each sector of the macula on en face SS-OCTA images by selecting two different en face SS-OCTA slabs based on Bruch's membrane. We next obtained a slab positioned 0 to 10 µm below Bruch's membrane to measure the vascular density of choriocapillaris. The choriocapillaris vascular density was calculated as the mean gray value of the slab image, using a previously reported method.¹⁵ We measured the choroidal stromal density in the inner choroid by obtaining a slab positioned between 46.8 and 62.4 µm below Bruch's membrane (Fig. 3).^15,16 The stromal density of the choroid was calculated as follows: the image was converted to 8-bit grayscale, processed using the Phansalkar method, and converted into black and white. The Analyze-measure tool was used to calculate the area fraction of the vessels as the vessel density. It was subtracted from 100 to calculate stromal density. To adjust the effect of choroidal stromal density, the density and area fraction of the HCF were divided by the choroidal stromal density. 

Figure 3.

 

Measurement of choroidal stromal density on an en face optical coherence tomography image. A slab between 46.8 and 62.4 µm below Bruch's membrane was obtained and exported to the ImageJ software (A). After converting the image to 8-bit grayscale, it was processed using the Phansalkar method and converted to black and white (B). The area fraction was calculated as the vessel density using the Analyze-Measure tool, which was subtracted from 100 to calculate the stromal density.

Statistical Analysis

All statistical analyses, except multiple correlation tests, were performed using GraphPad Prism for Windows (version 10.1.2; GraphPad Software, San Diego, CA, USA). Continuous variables are expressed as mean ± SD and between groups using the Mann–Whitney U test. Considering that the area of each ETDRS sector is different, the number of HCF per unit area (mm²) of a 6-mm × 6-mm en face image was used to compare the distribution of hyperreflective foci. A repeated-measures one-way analysis of variance with Greenhouse–Geisser correction was used to compare the distribution according to the location of the HCF within the macula, and post hoc pairwise comparisons were performed with Bonferroni's adjustment. We analyzed the correlation between HCF and different parameters using Pearson's correlation coefficient (r). Multiple correlations between sectors in the HCF distribution were analyzed using a step-down Bonferroni correction. Multiple correlation analyses were performed using SAS software for Windows (version 9.4; SAS Institute, Cary, NC, USA). The standard for statistical significance was set at P < 0.05. 

Results

We included 35 eyes of 35 participants (Table 1). The mean age was 52.7 ± 16.8 years. The mean subfoveal choroidal thickness was 274.0 ± 101.7 µm. In the macula (within the outer circle of a modified ETDRS grid with a 5.0-mm diameter), the mean choriocapillaris vascular density was 112.7 ± 2.9 and the stromal density of choroid was 68.7% ± 3.7%. The mean number of HCF per unit area (mm²) was 35.6 ± 7.8 foci and the area fraction was 3.03% ± 0.70%. Subfoveal choroidal thickness and choriocapillaris vascular density negatively correlated with age. However, the number and area fraction of the HCF did not correlate with age (P = 0.180 and P = 0.203, respectively). The mean age, subfoveal choroidal thickness, and all choroidal measurements, such as the number or area fraction of HCF in the macula, did not differ between the female and male participants (all P > 0.05). 

Table 1.

 

General Characteristics and Their Correlations With Age

 

Distribution of HCF in the Macula

The mean number and area fraction of HCF in the central circle were 50.7 ± 20.9 foci/mm² and 4.84% ± 3.36%, respectively (Table 2). They were significantly higher than those in the inner (35.1 ± 13.0 foci/mm² and 2.62% ± 1.17%; P < 0.001 and P < 0.001) and outer rings (35.6 ± 6.5 foci/mm² and 3.12% ± 0.67%; P < 0.001 and P = 0.012) (Fig. 4). This difference in HCF distribution according to the location within the macula was maintained, even when differences in choroidal stromal density were considered (Supplementary Table S1). The distribution of HCF differed among sectors in the inner and outer rings (Fig. 5). In the inner ring, the number of HCF in the superior sector was smaller than that in the temporal sector (P = 0.047), whereas the area fraction did not differ among the sectors. In the outer ring, the inferior sector had a significantly lower number and area fraction than the nasal and temporal sectors (Fig. 5). Furthermore, these differences in the distribution of HCF according to the location within the macula were consistently observed, even when differences in the underlying choroidal stromal density were considered (Supplementary Fig. S1). 

Table 2.

 

Regional Differences in the Number and Area Fraction of Hyperreflective Choroidal Foci Among the Central Circle, Inner Ring, and Outer Ring

 

Figure 4.

 

Comparisons of the number and area fraction of hyperreflective choroidal foci between the central circle, inner ring, and outer ring. Post hoc pairwise comparisons were conducted using Bonferroni's correction.

Figure 5.

 

Distribution and regional differences of HCF among different sectors in the inner and outer rings. The values for the number (A) and area fraction (D) of HCF are reported as the mean ± SD in the nine sectors of the modified ETDRS grid. The number (B, C) and area fraction (E, F) of HCF were compared among different sectors in the inner (B, E) and outer (C, F) rings. P values (B–F) for multiple comparisons were estimated using a repeated-measures one-way analysis of variance with Greenhouse–Geisser correction. Post hoc pairwise comparisons among different sectors were conducted using Bonferroni's correction. *Adjusted P < 0.05. I and INF, inferior; N and NAS, nasal; RM-ANOVA, repeated-measures one-way analysis of variance; S and SUP, superior; T and TEM, temporal.

Factors Associated With HCF Distribution

The number of HCF in the central circle correlated with that in the superior and inferior sectors of the inner ring (Table 3). The number and area fraction of the HCF in the superior and inferior sectors of the inner ring were correlated or tended to correlate with those of the outer ring. The number and area fraction of HCF in the central sector were not significantly correlated with age, subfoveal choroidal thickness, choroidal capillary density, or choroidal stromal density (Supplementary Table S2). However, the number of HCF in the inner superior and inferior sectors was positively correlated with age (r = 0.474, P = 0.004 and r = 0.367, P = 0.030, respectively). In addition, the area fractions of HCF in the inner superior and inferior sectors were positively correlated with age (r = 0.504, P = 0.002 and r = 0.462, P = 0.005, respectively). The HCF area fractions in the superior, nasal, inferior, and temporal sectors of the outer ring were positively correlated with the choriocapillaris vascular density in each sector (r = 0.365, P = 0.031; r = 0.359, P = 0.034; r = 0.421, P = 0.012; and r = 0.445, P = 0.007, respectively). 

Table 3.

 

Correlation of Hyperreflective Choroidal Foci Number and Area Fractions Between Various Sectors

 

Discussion

We used en face OCT images to quantitatively measure the HCF located between the choriocapillaris and Sattler's layer in the choroid. We investigated topographic variations in HCF distribution in the macula of normal eyes. We found that HCF were more densely distributed in the central macular region than in the periphery. Although the origin of hyperreflective foci is unclear, they are observed in various retinal and choroidal diseases and may have different histopathologic origins, depending on the specific pathophysiology of each disease.¹ Establishing normative spatial distribution of HCF in the choroid provides a baseline for future comparisons in diseased eyes. 

Variations in HCF Distribution Within the Macula of Normal Eyes

In addition to identifying the HCF highest density in the macular center, we demonstrated that HCF distribution differed among sectors of the pericentral and peripheral macular areas. It is unclear why the superior sector of the inner ring and the inferior sector of the outer ring have lower HCF densities. Variations in choroidal thickness in the macula and the peripapillary area were previously reported.^17–19 In the current study, subfoveal choroidal thickness did not correlate with the HCF distribution within the central circle. However, choroidal thickness was measured only at the subfovea, preventing a comparison with choroidal thickness and the distribution of hyperreflective spots at different locations across the macula. Regional differences in the developmental process of the macular choroid may have influenced the differences in HCF distribution.¹⁸ However, the differences in HCF distribution depending on the location were maintained despite considering the differences in the underlying choroidal stromal density. Another possibility could be projection artifacts from large retinal vessels, which are more abundant in the superior and inferior pericentral areas than in the nasal and temporal areas. However, we manually erased the areas with projection artifacts from the retinal vessel and reduced their influence on HCF measurements. Further studies with histologic analyses are warranted to elucidate the origin of the differences in HCF distribution in different sectors of the macular region. 

Origins of Hyperreflective Foci

Hyperreflective foci were first described in the neurosensory retina.^20,21 These foci were observed in all retinal layers and the subretinal space on OCT images. They are speculated to originate from cells containing hyperreflective pigment granules such as melanosomes, lipofuscin, melanolipofuscin, or granule aggregates.^22–27 Candidate cells suspected of containing such hyperreflective substances include ectopic RPE cells^{21,23,27–31} and mononuclear phagocytes,^{24,25,32–38} including resident microglia and recruited inflammatory macrophages that ingest and retain RPE organelles. 

Hyperreflective foci have also been found in the choroid of eyes with diabetic macular edema,⁵ choroideremia,⁷ angioid streaks,³⁹ Stargardt disease,^6,40 retinitis pigmentosa,^41,42 Vogt–Koyanagi–Harada disease,⁹ geographic atrophy,³ and the eyes of normal individuals.^8,43 However, due to the limited histopathologic studies available, the exact origins of choroidal HCF remain unknown and are less studied compared to intraretinal hyperreflective foci. Several hypotheses have been proposed to clarify hyperreflective foci origin in multimodal imaging studies. In eyes with diabetic macular edema, HCF are assumed to result from intraretinal hyperreflective foci that migrate to the choroid after disruption of the external limiting membrane.^5,44 In choroideremia, Romano et al.⁷ speculated that HCF may originate from recruited macrophages or migration of progressively degenerated RPE cells. Among eyes with angioid streaks, HCF was more frequently found in eyes with choroidal neovascularization than in those without, indicating a potential relationship between macrophages and choroidal neovascularization development.³⁹ HCF has also been observed in retinal degenerative diseases caused by toxic phototransduction by-products leading to photoreceptor and RPE degeneration.⁴⁵ In Stargardt disease, the number of choroidal hyperreflective spots correlated positively with disease duration⁶ and was higher at the pathological border,⁴⁰ suggesting that they may represent lipofuscin deposits migrating from the outer retina.⁴¹ In retinitis pigmentosa, Huang et al.⁴¹ reported the frequent presence of hyperreflective foci in the retina and choroid, suggesting these foci to be the result of either migrated diseased RPE cells containing lipofuscin granules or the unmasking of choroidal melanocytes in atrophic lesions. In addition to an unclear origin of HCF in disease-affected eyes, a significant number of hyperreflective foci have been observed in the choroid of normal subjects.^8,12,43 In a previous study,⁸ a greater number of hyperreflective foci were observed with an increase in the stromal area. The choroidal stroma comprises nerve processes, collagen and elastic fibers, large melanocytes, fibroblasts, nonvascular smooth muscle cells, and several immune cells such as macrophages, mast cells, and lymphocytes.^46,47 Of these components, melanocytes, containing numerous melanosomes with a large size between 20 and 50 µm,^8,46,48 are proposed as a possible explanation for HCF presence in normal eyes without RPE degeneration or inflammation.^8,12,43 In a recent study,⁹ it was reported that HCF almost disappeared in the choroid of patients with Vogt–Koyanagi–Harada disease who manifested a sunset glow fundus due to the loss of choroidal melanocytes, implying that melanocytes could be related to the origin of choroidal hyperreflective foci.^8,9 

Our study revealed variations in HCF distribution within the macula of normal eyes, with higher density and area fraction observed at the macular center despite variability in their numbers. It is unclear why the choroid in the macular center has a denser HCF distribution. However, if HCF are associated with melanocyte distribution, as suggested in previous studies,^8,9,41,43 it is ascribed to variations in the distribution of choroidal melanocytes. The choroidal distribution of melanin, which is produced and stored in the melanosomes of choroidal melanocytes, was determined histologically in previous studies.^49–51 Miura et al.⁵¹ reported that the central region of the macula contains more choroidal melanin content than the peripheral region. These results support the hypothesis that HCF are associated with melanocytes.⁸ Melanocytes within the choroid absorb light, protect cells functioning as optical fibers, manage reactive oxygen species, and contribute to immune modulation.⁵² Melanocytes, which are most abundantly distributed in the macular center, aid in maintaining the visual function of the macular center, which receives the highest intensity and amount of incident light. However, additional comparative studies of hyperreflective foci on OCT images and melanocytes on histologic sections are required to confirm this hypothesis. 

Factors Related to HCF Distribution

Our study found no difference in the density of HCF in the macular center between males and females and showed no correlation with age, consistent with previous studies,^8,12 in which the HCF were evaluated only at the macular center. In this study, we investigated HCF distribution in the pericentral area of the macula. We found that the density of HCF in the superior and inferior pericentral areas was positively correlated with age. It is unclear why the distribution of pericentral HCF, but not macular central HCF, increases with age. Thus, several hypotheses can be considered. One possibility is that aging influences HCF changes in the pericentral area, whereas the macular center, which has a high density of HCF and macular pigments, is more resistant to age-related changes in HCF distribution. Aging can also influence various choroidal changes such as choroidal thickness, choriocapillaris vascular density, and choroidal stromal changes.^53–55 In this study, choriocapillaris vascular density was correlated with age. Moreover, the area fractions of HCF in the pericentral area, but not in the macular center, were positively correlated with the choriocapillaris vascular density in each sector. These findings may support our hypotheses. However, further studies are needed to elucidate the age-related changes in HCF. 

In this study, en face images analysis rather than B-scan analysis was performed to better understand the topographical distribution of each parameter. We found that the distribution of choroidal stroma in the layer immediately below the layer measured for HCF did not affect HCF distribution. These findings suggest that, although there is a positive correlation between the amount of full-thickness choroidal stroma and HCF, HCF measurements were not directly affected by the distribution of choroidal vessels and stroma around the layer selected for HCF measurement. These findings indicate that variations in the regional distribution of surrounding blood vessels do not result in a significant measurement error in HCF and support the possibility of HCF usage as an independent imaging parameter. 

Limitations

This retrospective study, conducted on a small sample of participants across a wide age range, demonstrated broad variability in measurements, even with high-quality images of normal eyes. The relationships between variables and HCF distribution were unclear, and the lack of reference values limited result validation, requiring careful interpretation of our results. Further studies with a larger number of participants are required to establish a normalized HCF database and a deeper understanding of the relationships between different variables. Herein, choroidal stromal density did not affect HCF distribution. We measured the stromal density at a single fixed depth in the middle of Sattler's layer, although variations in choroid thickness might affect the relative position of the measurement layer. However, we believe that these random effects did not significantly affect our overall results, which requires caution while interpreting our results. 

Another limitation is the lack of exploration into the relationship between imaging findings and histologic studies, making it difficult to distinguish whether the quantified HCF is physiological or pathological. Further studies are required to elucidate HCF origin and implications. In addition, we did not compare en face images at all depths or analyze HCF of all layers. Images of the hyperreflective spots may become enlarged and distorted with an increase in depth owing to the blooming effect.⁵⁶ Therefore, only images from the inner choroidal layer could be compared. Nonetheless, we believe that our findings are valuable, as this is the first study to analyze the spatial distribution of HCF in the macula using en face OCT images. 

Conclusions

This study demonstrated that HCF are located between the choriocapillaris and Sattler's layer in the choroid of eyes with a normal fundus. HCF distribution in en face structural OCT images differed depending on their locations within the macula. The macular central 1-mm area exhibited the highest HCF density and area fraction. HCF measurement and spatial distribution could provide additional information for evaluating choroidal stromal characteristics. Future studies on patients with various diseases will offer greater insights into the evaluation of changes in the HCF of the choroid. 

Acknowledgments

Supported by the National Research Foundation of Korea grant funded by the Korean government (MSIT) (No. 2022R1A2C2092118, NTIS number: 1711179625). A part of this study is based on the unpublished master's thesis submitted by Myung-Sun Song for a master's degree at Korea University, Seoul, Korea, 2024. It was presented as a poster (number: PO-313) at the 39th Asia-Pacific Academy of Ophthalmology Congress, February 22–25, 2024, in Bali, Indonesia. 

Disclosure: M-S. Song, None; Y.H. Kim, (P); J. Oh, (P)