Diurnal Variation in Choroidal Parameters Among Healthy Subjects Using Wide-Field Swept-Source Optical Coherence Tomography Angiography

Abstract

Purpose: The purpose of this study was to evaluate the diurnal variation in choroidal parameters in a wide field area among healthy subjects and to identify correlations between choroidal luminal area and stromal area and various systemic factors.

Methods: In this cross-sectional study, 42 eyes from 21 healthy participants (mean age = 32.4 ± 8.8 years) were examined using wide-field swept-source optical coherence tomography angiography (WF SS-OCTA, 24 mm × 20 mm). Measurements of choroidal parameters, including choroidal volume (CV), choroidal thickness (CT), choroidal vessel volume (CVV), and choroidal stromal volume (CSV), were taken at 8:00, 12:00, 18:00, and 22:00. Systemic factors, such as blood pressure and heart rate, were concurrently monitored.

Results: Our study observed significant diurnal variations in the mean total CV, CT, CVV, and CSV, with minimum measurements around 12:00 (P < 0.001) and peak values at 22:00 (P < 0.001). Furthermore, changes in CV in specific regions were more closely associated with fluctuations in CVV than CSV in the same regions. No significant diurnal variations were found in systolic (P = 0.137) or diastolic blood pressure (P = 0.236), whereas significant variations were observed in the heart rate (P = 0.001).

Conclusions: Our study reveals diurnal variations in choroidal parameters and their associations, emphasizing that changes in choroidal volume relate more to the luminal than the stromal area in vessel-rich regions. This enhances our understanding of choroidal-related ocular diseases.

Translational Relevance: Regions with higher choroidal vasculature observed greater choroidal volume changes.

Introduction

The choroid is a vascular layer nestled between the retina and sclera, vital for supplying blood to the external layers of the retina. Beyond this critical role, the choroid undertakes essential physiological functions, such as adjusting the position of the retina through changes in choroidal thickness, moderating retinal temperature, and secreting growth factors.¹ Given its importance, choroidal dysfunction can precipitate an array of ocular diseases such as central serous chorioretinopathy (CSC), uveitis, or neovascular age-related macular degeneration (nAMD). Consequently, quantitative assessment of choroidal vessels can offer valuable insights into the pathophysiology of various chorioretinal disorders. 

Researchers have strived to measure choroidal thickness (CT) using various instruments to better comprehend choroidal abnormalities. In earlier studies, CT was gauged using ultrasonography. However, due to its limited resolution, distinguishing the retina from the choroid was a formidable task.^2,3 The advent of spectral domain optical coherence tomography (SD-OCT) technology, combined with enhanced depth imaging (EDI) techniques and swept source optical coherence tomography (SS-OCT), has revolutionized this process. The higher resolution and longer wavelengths of these technologies provide a detailed exposition of the choroid and enable accurate identification of the choroidal scleral interface (CSI).^4–8 

In recent years, more attention has been paid to the regional distribution of CT. Previous studies revealed significant eccentricity-dependent patterns of change in CT, with decreasing choroidal thinning beyond the parafovea into the periphery and a quadrant thickness distribution with superior CT thickest, followed by the temporal, inferior, and nasal.^9,10 A similar finding was also reported by Ding et al.¹¹ Yet, numerous studies evaluating CT have primarily focused on the choroid's macular region, especially the subfoveal choroid. Given that the choroid's anatomy is distinctly different from the retina and most choroidal diseases are not confined to the macular region, this focus is somewhat narrow. Recently, the advent of widefield (WF) SS-OCT has facilitated the analysis of CT in areas proximal to the vortex vein ampulla, opening avenues to study peripheral choroidal structures in the eyes.^12,13 

It is well established that CT experiences significant diurnal fluctuations—thickening during the night and thinning during the day—in both normal animal and human subjects.^10,14–20 Nevertheless, past studies exploring this diurnal variation have concentrated exclusively on subfoveal choroid thickness (SFCT). As a step toward a more comprehensive understanding, we have examined the diurnal fluctuation in CT beyond the macula in this study. Utilizing WF SS-OCTA, we evaluate the diurnal variation in choroidal parameters in a wide field area among healthy subjects and identify correlations between choroidal luminal area and stromal area and various systemic factors. 

Methods

Subjects and Procedures

This study was performed at the Zhongshan Ophthalmic Center, having received approval from the Institutional Review Board of the Zhongshan Ophthalmic Center at Sun Yat-Sen University in adherence to the Declaration of Helsinki (2023KYPJ116). 

The study's participant group, consisting of hospital employees who live near their workplace, provided a uniform sample, effectively reducing variability linked to environmental factors. Prior to participation, each subject underwent an initial ophthalmic examination to ascertain their ocular health. All participants were instructed to abstain from wearing contact lenses and consuming coffee, alcohol, smoking, and tea on the day of testing. Individuals working night shifts were excluded from the study. 

The subjects’ baseline demographic characteristics, such as age, gender, best-corrected visual acuity (BCVA), spherical equivalent (SE), and ocular axial length (AL), were recorded. Measurements of choroidal parameters, systolic blood pressure (SBP), diastolic blood pressure (DBP), and heart rate (HR) were carried out at 8:00, 12:00, 18:00, and 22:00 in a single day within a dark room. To ensure the consistency and reliability of our data, all assessments were completed within a tightly scheduled 3-day window. Each participant's examination was conducted on the same day, effectively reducing the impact of environmental and other external variables. All examinations were performed by the same experienced investigator (author G.Q.H.). 

Wide-Field Swept-Source Optical Coherence Tomography Angiography

Images of choroidal structure were obtained using WF SS-OCTA (TowardPi BMizar, TowardPi Medical Technology, Beijing, China) centered on the fovea, covering a 24 mm × 20 mm (1280 lines, 120 degrees angular field of fundus view) section. This instrument uses a swept-source vertical-cavity surface-emitting laser (VCSEL) of wavelength 1060 nm, providing a transverse resolution of 10 µm and an in-depth resolution (optical) of 3.8 µm at a scanning rate of 400,000 A-scans per second. Notably, the TowardPi DAQ card, a significant component, operates at 6 GB/second, 12 bits, and a Signal Noise Ratio (SNR) of 56 db. The algorithm applied to detect motion signals is called higher-order moment amplitude decorrelation angiography (HMADA). This innovative approach has been recognized for its efficacy in detailed visualization of both the larger blood vessels and the intricate capillary network within the retinal and choroidal circulations. It achieves this by capturing and analyzing higher-order statistical signals from optical coherence tomography angiography (OCTA) data. 

Leveraging the capabilities of artificial intelligence, this technique allows for precise recognition of each layer within the ocular structure, including Bruch's membrane (BM) and the choroid-sclera interface. The resultant en face OCT and OCTA images enable both automatic and manual segmentation of the retinal and choroidal layers, offering highly detailed visualizations. The accuracy and reliability of this segmentation process, as facilitated by the HMADA algorithm, have been previously validated in studies by Wang et al. and Zhang et al., which we reference for substantiation of our methodology.^21,22 

Imaging Acquisition

Each participant in our study underwent OCTA examinations in a darkened room, a deliberate choice to facilitate natural pupil dilation, thereby ensuring the capture of high-quality images. This method effectively bypasses the potential effects of dilation drugs on choroidal measurements. For WF-OCTA, we conducted high-definition scans covering 24 mm × 20 mm sections centered on the fovea. To bolster the accuracy of our data, we diligently repeated the scanning process for each participant. Furthermore, the reproducibility of measurements, particularly the vessel densities within the choroidal vessel layer (ChV) and thickness measurements, has been previously validated and corroborated in studies by Zhang et al.²² Last, to maintain optimal scan quality and ensure comparability, it was necessary for the WF-OCTA scans to achieve an image quality score of at least 8. 

Furthermore, to ensure the precision of our vascular parameter quantifications using swept-source OCTA (SS-OCTA), each image underwent an initial verification process. This involved two experienced ophthalmologists (authors G.Q.H. and X.Z.Z.), who reviewed the automated segmentation accuracy and manually adjusted it when necessary. For the measurement of choroidal vasculature, we adopted a standardized approach across participants, correcting for AL-related magnification through the modified Littmann formula, following the Bennett procedure. This method ensures consistent and accurate comparisons of the choroidal vascular structures in our study population. 

Choroidal Parameters

The measured choroidal parameters included choroidal volume (CV), CT, choroidal vessel volume (CVV), and choroidal stroma volume (CSV). CT is designated as the vertical distance separating the outer border of the retinal pigment epithelium (RPE) from the chorioscleral interface. The CV is computed as the aggregate of the choroidal vessel volume and the choroidal stroma volume. CVV is defined as the volume of choroidal vessels per unit area (volume/area), expressed in microns. A higher CVV value indicates a greater density of vessels. CSV is similarly defined as the volume of choroidal stroma per unit area (volume/area), also measured in microns. 

For the purpose of analysis, the software partitioned the section into nine grids: superior temporal (ST), superior (S), superior nasal (SN), temporal (T), central (C), nasal (N), inferior temporal (IT), inferior (I), and inferior nasal (IN), as shown in Figure 1. The overall (Tt) and individual parameters from the nine grids were subsequently procured for further analysis (see Fig. 1). 

Figure 1.

 

Illustration of the 24 mm × 20 mm WF SS-OCTA scanning protocol. (A, B, C, D, E, G, H, I) Right eye (OD). (F, H, J) Left eye (OS). (A) Choroidal vasculature images. (B, C, D) Horizontal B-scan centered on the fovea. CT = the vertical distance between the outer border of the RPE and the chorioscleral interface (green line). CVV = choroid vessel volume per unit area (volume/area). (E, F) A grid has been developed for measuring physiological changes in the choroid centered on the macula across nine regions: ST = superior temporal; S = superior; SN = superior nasal; T = temporal; C = central; N = nasal; IT = inferior temporal; I = inferior; and IN = inferior nasal. (G, H) The choroidal thickness map in the same subject. (I, J) Three-dimensional choroidal vessel volume (CVV) maps from the same subject.

Data Analysis

Data were analyzed statistically using SPSS software (version 26.0; SPSS, Inc., Chicago, IL, USA). The Shapiro-Wilk test was used to verify the normal distribution of the study variables. Variations in ocular and systemic parameters were examined using generalized estimating equations (GEEs), accounting for age, sex, AL, and SE parameters. Changes in choroidal thickness during the day were determined as the difference between the thickest and thinnest mean choroidal parameters in a single measurement session. Significant fluctuations, when detected, warranted the use of the Bonferroni test for post hoc analysis. Given the characteristics and distribution of the variables, a multiple logistic regression model was used. A P value of less than 0.05 was considered statistically significant. 

Results

Demographic Characteristics

This research examined 42 eyes from 21 healthy individuals, comprising 4 male subjects and 17 female subjects. The mean age was 32.4 ± 8.8 years (ranging from 24 to 52 years). The mean AL was 24.59 ± 1.52 mm (ranging from 21.41 to 27.88 mm), whereas the mean SE was −3.11 ± 2.49 diopters (ranging from −7.50 to +1.00 diopters). 

Systemic Diurnal Rhythms

No significant diurnal fluctuations were observed in the mean SBP and DBP (P = 0.137 and P = 0.236, respectively). Conversely, significant diurnal variations were recorded in the mean HR (P = 0.001) as shown in Table 1 and Figure 2. 

Table 1.

 

Overview of Systemic and Total Choroidal Parameters at Different Times of the Day

 

Figure 2.

 

Diurnal variations in the systemic and choroidal parameters at four different times of the day. (A) Diurnal variations in the choroidal volume of the total choroid. (B) Diurnal variations in choroidal vessel volume and choroidal stroma volume of the total choroid. (C) Diurnal variations in the choroidal vessel volume across nine regions: ST = superior temporal; S = superior; SN = superior nasal; T = temporal; C = central; N = nasal; IT = inferior temporal; I = inferior; and IN = inferior nasal. (D) Diurnal variations in the systemic parameters.

Ocular Diurnal Rhythms

Regarding total and regional choroidal parameters, significant diurnal variations were noticed in mean total CV (P < 0.001), CT (P < 0.001), CVV (P = 0.010), and CSV (P < 0.001). The mean total CT, CV, CVV, and CSV were at their thinnest at 12:00 (185.69 ± 26.08 µm, 82.56 ± 11.56 mm³, 88.53 ± 12.47 µm, and 126.01 ± 15.46 µm, respectively) and thickest at 22:00 (189.40 ± 28.75 µm, 84.22 ± 12.72 mm³, 89.96 ± 13.17 µm, and 128.40 ± 17.24 µm, respectively; see Table 1, Fig. 2, Fig. 3). 

Figure 3.

 

Illustration of the 24 mm × 20 mm WF SS-OCTA images of CT for 2 healthy subjects at 4 different times of the day. Subject 1: (A, B, C, D, E, F, G, H, I, J, and K). (A, B, C, D, and E) Right eye (OD). (F, G, H, I, and J) Left eye (OS). (A, F) The choroidal layer OCTA en face images of the same subject at 8:00. (B, G) The total choroidal thicknesses maps of the same subject at 8:00. (C, H) The total choroidal thicknesses maps of the same subject at 12:00. (D, I) The total choroidal thicknesses maps of the same subject at 18:00. (E, J) The total choroidal thicknesses maps of the same subject at 22:00. Subject 2: (K, L, M, N, O, P, Q, R, S, T, K, L, M, N, and O) Right eye (OD). (P, Q, R, S, and T) Left eye (OS). (K, P) The choroidal layer OCTA en face images of the same subject at 8:00. (L, Q) The choroidal thicknesses maps across 9 regions of the same subject at 8:00. (M, R) The choroidal thicknesses maps across 9 regions of the same subject at 12:00. (N, S) The choroidal thicknesses maps across 9 regions of the same subject at 18:00. (O, T) The choroidal thicknesses maps across 9 regions of the same subject at 22:00.

There were significant diurnal variations in CV and CT in the ST, S, SN, C, N, IT, I, and IN regions (all P < 0.05), except in the T region (P = 0.079 and P = 0.058, respectively). Significant diurnal variations were also observed in CVV in the ST, S, SN, C, and N regions (all P < 0.05). In contrast, there were no significant diurnal variations in CVV in the T, IT, I, and IN regions (P = 0.487, P = 0.109, P = 0.221, and P = 0.061, respectively). Significant diurnal variations were detected in CSV in all nine regions (all P < 0.05; Table 2). 

Table 2.

 

Overview of Choroidal Parameters Across Measured Regions at Different Times of the Day

 

Correlations of Choroidal Parameters With Systemic Parameters

Multiple linear regression analyses (Table 3) showed that fluctuations in total choroidal volume were significantly correlated with fluctuations in choroidal volume across all nine regions (standardized β = 0.128; standardized β = 0.153; standardized β = 0.13; standardized β = 0.133; standardized β = 0.19; standardized β = 0.167; standardized β = 0.115; standardized β = 0.173; and standardized β = 0.133, respectively, all P < 0.05). Moreover, fluctuations in CVV and CSV across all nine regions were positively correlated with fluctuations in CV within the same region (ST-CVV: standardized β = 0.806, ST-CSV: standardized β = 0.595; S-CVV: standardized β = 0.646, S-CSV: standardized β = 0.663; SN-CVV: standardized β = 0.932, SN-CSV: standardized β = 0.472; T-CVV: standardized β = 0.831, T-CSV: standardized β = 0.501; C-CVV: standardized β = 0.687, C-CSV: standardized β = 0.764; N-CVV: standardized β = 0.935, N-CSV: standardized β = 0.51; IT-CVV: standardized β = 0.786, IT-CSV: standardized β = 0.447; I-CVV: standardized β = 0.516, I-CSV: standardized β = 0.581; IN-CVV: standardized β = 0.934; and IN-CSV: standardized β = 0.445, all P < 0.05; Table 4). Changes in CV in the ST, SN, T, N, IT, and IN regions were more closely associated with fluctuations in CVV than CSV within the same region. 

Table 3.

 

Multiple Regression Analysis of Total Choroidal Volume Changes Between 22:00 and 12:00 in Relation to Variables

 

Table 4.

 

Multiple Regression Analysis of Choroidal Volume Changes (22:00 vs. 12:00) Across Different Regions Relative to Intra-Regional Variables

 

Discussion

The current study explored diurnal variations in choroidal thickness and vascularity parameters, such as luminal and stromal thickness, throughout an extensive field in healthy eyes using WF SS-OCTA. Our findings confirmed the existence of substantial diurnal variations in choroidal volume, exhibiting a trough at 12:00 and a peak at 22:00, a pattern that aligns with earlier investigations.^15,17 We also found that SBP demonstrated the same diurnal trend. The variations in choroidal volume across all nine regions were observed to correspond with total choroidal volume changes. Furthermore, fluctuations in CV in the superior temporal, superior nasal, temporal, nasal, inferior temporal, and inferior nasal regions had stronger associations with changes in CVV than with CSV in the corresponding regions. 

The choroid is one of the most vascularized structures in the human body, providing essential nutrients and oxygen to the outer layers of the retina. As our understanding of the choroidal vasculature grows, we gain invaluable insights into diseases that affect this area. EDI-OCT has been instrumental in this understanding, revealing previously hidden facets of the choroidal structures. Sonoda et al.²³ were the pioneers in analyzing these structures, differentiating the choroid into luminal and stromal components. They introduced a new quantitative parameter, the “choroidal vascular ratio” (LC ratio), which is derived from the vascular luminal area (LA) over the total choroidal area (TCA). The process involved using Niblack binarization conversion on OCT B-scan images, which proved useful in image layout analysis and skew estimation.²⁴ Following this pioneering work, Agrawal et al.²⁵ developed the “Choroidal Vascularity Index” (CVI), which also measures the ratio of LA to TCA, but by converting images to 8-bit grayscale and applying Niblack auto local thresholding before demarcating the CSI for TCA calculation. Although this approach improved the 2D image processing, it had its limitations. Notably, artifacts like retinal shadows remained in the processed images, and the whole process was time-consuming. Our study used in-built software for image analysis, which proved efficient in detecting signal differentiation in tissues of Sattler's layer and Haller's layer. It aided in signal enhancement and artifact elimination, leading to a robust reconstruction of the vessel patterns from these layers. Notably, the algorithm used outperformed existing algorithms in accurately detecting the choroid-sclera interface. With the data being three-dimensional, we were able to perform a quantification analysis of choroidal vessels in 3D, offering an efficient way to obtain reliable data without secondary processing steps. 

Previous studies have presented differing results regarding diurnal variations in CT. Tan et al.¹⁶ demonstrated the thickest CT in the morning, which decreased throughout the day, reaching its thinnest at 5 PM. In contrast, Chakraborty et al. reported CT as thinnest at 12 PM and thickest at 6 PM over 2 consecutive days.¹⁵ Kinoshita et al.¹⁹ proposed CT as thickest at 6 AM, thinnest at 3 PM, and increasing in the evening. Usui et al.¹⁷ also found the choroid to thicken during the night. These findings, strongly supported by animal studies, including ones involving chicks and the common marmoset,¹⁷ underscore the diurnal variation in CT, a phenomenon that our study reaffirms. Additionally, we detected significant diurnal variations in total and central CV, CVV, and CSV. Consistent with earlier research primarily focused on the macular area, our study demonstrated a significant correlation between central CT changes and luminal area changes.¹⁹ For the first time, Kinoshita et al.¹⁹ reported diurnal variations in CT attributed to the luminal area rather than the stromal area. In our study, we found a similar trend in the superior temporal, superior nasal, temporal, nasal, inferior temporal, and inferior nasal regions, which are abundant in vortex veins. We believe this regional variation is related to the distribution of choroidal venous vessels.^26,27 In areas with a higher density of choroidal venous vessels (ST, S, SN, C, and N) significant diurnal rhythms were observed, whereas in areas with fewer choroidal venous vessels, such rhythms were not evident. This emphasizes the crucial role of choroidal vessels in these variations. 

Previous investigations have indicated that BP peaks early in the morning and gradually decreases throughout the day.¹⁹ Other studies, in humans and animals alike, have reported different times for peak blood pressure.^28,29 Our results showed that the SBP was lowest at 12:00 and highest at 22:00, demonstrating that blood pressure is influenced by endogenous circadian rhythm. Moreover, these fluctuations are also associated with various factors, including physical activity and the white-coat effect. Interestingly, variations in vascular smooth muscle contraction and vasoconstriction also contribute to the circadian rhythm of blood pressure.³⁰ Considering that smooth muscle cells constitute a significant component of the vasculature, the alignment of the SBP trend with CT is explicable. Additionally, the duration of the cardiac cycle is inversely related to the HR.³¹ As the HR increases, the cardiac cycle shortens, affecting both the systolic and diastolic phases, with a more significant reduction in the diastolic phase. Consequently, the volume of blood flowing from the aorta to the periphery decreases, leading to an increase in blood volume retained within the aorta, which significantly raises diastolic pressure. However, elevated blood pressure accelerates blood flow, resulting in an increased volume of blood delivered to the periphery during systole, thereby making the increase in systolic pressure relatively smaller. Therefore, changes in HR primarily influence diastolic pressure, in alignment with the consistency observed in this study between HR and diastolic pressure changes, and are also consistent with reports from previous literature.³² The internal “oscillator” regulating the human biological clock is the suprachiasmatic nucleus, which processes external signals, such as environmental light information and inputs from the brain, regulating circadian functions including body temperature, sleep/wake cycles, and the secretion of hormones like melatonin.³³ Circadian clocks are also present in cardiac muscle cells, vascular smooth muscle cells, and endothelial cells. There is a complex interaction between environmental influences and internal mechanisms (i.e. central and peripheral biological clocks). Changes in eating or sleeping patterns, as well as exposure to light at unusual times (such as at night), can lead to a loss of synchrony, explaining the observed asynchrony in rhythm changes between HR and CT in this study. 

Over the past several decades, a wealth of research has established correlations between CT and a variety of factors, including sex, age, AL, intraocular pressure, body mass index, and genetic factors like complement factor H. Notably, studies have found intriguing links between CT and everyday phenomena, such as coffee consumption or fasting.^{14,15,18,34–37} As such, it has been observed that choroidal thickness is significantly higher in male subjects than in female subjects and exhibits a more pronounced diurnal pattern in men.¹⁸ Ding et al.¹¹ demonstrated that SFCT is significantly negatively associated with age in subjects who are 60 years of age or older. In a different context, Nickla et al.³⁸ found that both CT and AL exhibited endogenous circadian rhythms independently of light and dark conditions. This intriguing observation was underscored by the discovery of a substantial negative correlation between variations in AL and CT.¹⁵ In addition, past research has indicated the existence of a diurnal rhythm in the synthesis of proteoglycans in chick sclera, peaking in the afternoon/evening and hitting a low in the early morning.³⁹ A literature review by Debora et al.,⁴⁰ highlighted potential mechanisms that may account for changes in CT. One proposed mechanism was the synthesis of large, osmotically active proteoglycans that absorb water, leading to an increase in CT. Importantly, the current study revealed that the circadian rhythm of CV was not influenced by factors such as AL, age, SD, and sex. This suggests an inherent regulation of CV that is independent of these commonly studied ocular and demographic factors. 

Yet, there are some limitations in our study. We had a relatively small sample size and measured AL only at the baseline. We also measured CT at only four daytime points, neglecting potential variations during the night. Additionally, our research faced constraints due to the demographic composition, with only 4 male and 17 female participants out of a total of 21, limiting a comprehensive comparison of circadian rhythms in choroidal thickness and other parameters across genders. The smaller number of male participants might lead to bias, affecting the analysis based on gender. Moreover, our study included a participant with high myopia who did not exhibit any significant pathological changes. Although the correlation between AL and CT is recognized, our research was primarily aimed at exploring the diurnal variations of the choroid. Given this focus, the inclusion of a high myopia participant without pathological changes is unlikely to have a significant impact on our study outcomes. 

In conclusion, we observed a diurnal rhythm in choroidal parameters across nine regions, peaking at 22:00 and at its lowest at 12:00. Furthermore, regions with more choroidal vasculature exhibited more substantial choroidal volume changes, particularly due to fluctuations in choroidal luminal rather than stromal components. The rapid advancements in scientific and technological applications in medicine contribute significantly to medical developments, aiding our understanding of the choroid and potentially influencing how we approach choroidal diseases. 

Acknowledgments

Disclosure: G. He, None; X. Zhang, None; X. Zhuang, None; Y. Zeng, None; X. Chen, None; Y. Gan, None; Y. Su, None; Y. Zhang, None; F. Wen, None