Retinal Vascular Oxygen Saturation in a Sample of Chinese Myopic Adults

Abstract

Purpose: The aim of the study was to investigate the normal retinal vascular oxygen saturation and to elucidate the factors affecting it or associated with it in a Chinese population.

Methods: This was a population-based observational cross-sectional study. Chinese myopic adults aged 18 to 45 years old were enrolled. Spherical equivalent (SE) and ocular biometry, including average keratometry (K), central corneal thickness, intraocular pressure, and axial length, were measured. The retinal arteriolar (SaO₂) and venous (SvO₂) oxygen saturation were measured after pupil dilatation.

Results: A total of 1373 participants were enrolled, and the mean age, SE, SaO₂, and SvO₂, were 26.31 ± 6.93 years, −6.40 ± 3.12 D, 93.47% ± 1.58%, and 63.08% ± 4.46%, respectively. In the multivariate analysis, SaO₂, SvO₂, and retinal arteriovenous oxygen saturation difference (AVD) were significantly correlated with age (β = 0.077, P = 0.006 for SaO₂ and β = 0.080, P = 0.006 for AVD) and SE (β = 0.220, P < 0.001 for SaO₂; β = 0.131, P < 0.001 for SvO₂; and β = −0.050, P = 0.048 for AVD, respectively). Higher SaO₂ was associated with larger K values (β = 0.094, P = 0.001).

Conclusions: SaO₂ increased with older age, lower myopia, and larger K values, older age was associated with higher SaO₂ and AVD, and SvO₂ decreased with the deepening of myopia. Our present study provided normative values for healthy Chinese myopic adults with different myopic refractive errors.

Retinal oximetry is used to detect retinal oxygen saturation of the central retinal arteries and veins and their branches. The results are influenced by ocular diseases and systemic physiologic status, such as diabetic retinopathy,¹ exudative age-related macular degeneration,² glaucoma,³ refractive status,⁴ retinitis pigmentosa,⁵ and factors affecting systemic physiologic status, such as age and gender.⁶ Therefore, it is necessary to establish normative values for retinal oxygen saturation in healthy populations. 

Holden et al.⁷ predicted that by 2030, myopia would account for 39.9% of the world's population, and ametropia has been an unavoidable problem in the analysis of retinal oxygen saturation. The effect of myopia on retinal oxygen saturation has been inconsistently concluded by different studies,^8,9 and current studies on the effects of ametropia on retinal oxygen saturation in healthy Chinese adults generally have small sample sizes.⁸ On the other hand, in studies on the effect of age on retinal oxygen saturation in adults, there are also inconsistent findings in different studies.^6,10,11 In particularly, there are no studies on the effect of age on retinal oxygen saturation in adults in a healthy Chinese population. 

In this study, a large sample of healthy Chinese adults with myopia was included to investigate the normal retinal vascular oxygen saturation and to elucidate the factors of age, gender, spherical equivalent (SE), axial length (AL), intraocular pressure (IOP), central corneal thickness (CCT), and average keratometry (K) affecting it or associated with it in a Chinese population. 

Methods

Patients who visited Ineye Hospital of Chengdu University of TCM for the proposed myopia correction surgery and preoperative examination from January to September 2022 were included in this study. All examination procedures were performed in accordance with the tenets of the Declaration of Helsinki and approved by the Medical Ethics Committee of the Ineye Hospital of Chengdu University of TCM (2021yh-022). All participants signed an informed consent form and were aware of all possible risks. 

This was a population-based observational cross-sectional study of Chinese myopic adults aged 18 to 45 years. The included criteria were adult (18 ≤ age ≤ 45 years old), SE ≤ −0.50 D, corrected distance visual acuity (CDVA) logMAR ≤0.10, and myopia increase of less than 0.5 D within 2 years. The excluded criteria were concomitant fundus lesions, glaucoma, or systemic diseases that could contribute to ocular pathology (Fig. 1). Biswas et al.¹² reported the high prevalence of glaucoma among those seeking myopia correction surgery; therefore, participants included in the study were examined for intraocular pressure and cup-to-disc ratio to exclude glaucoma. All participants had no history of ocular surgery or therapy. 

Figure 1.

 

The flowchart of the patient recruitment.

Only participants’ right eyes were included, and the left eyes were included without clear images of the right ones. According to SE, participants were divided into four groups: the low myopia group (−3.00 D < SE ≤ −0.50 D), the moderate myopia group (−6.00 D < SE ≤ −3.00 D), the high myopia group (− 9.00 D < SE ≤ −6.00 D), and the extremely high myopia group (SE ≤ −9.00 D). According to AL, participants were divided into four groups: the first group of participants had an axial length of 25 mm or less (AL ≤ 25.00 mm), the second group greater than 25 mm and less than or equal to 26 mm, the third group greater than 26 mm and less than or equal to 27 mm, and the last group greater than 27 mm. SE was expressed as the sphericity diopter (SD) plus half of the cylindrical diopter (CD) (SE = SD + 1/2 CD), and subjective refraction was measured after pupil dilatation using Tropicamide Phenylephrine Eye Drops (Mydrin -P; Santen, Osaka, Japan). All participants underwent tests for visual acuity (logMAR), intraocular pressure (IOP; CT-800, TOPCON Co., Ltd, Tokyo, Japan), K (Topographic Modelling System, TMS-4; Tomey Corporation, Nagoya, Japan), central corneal thickness (CCT), and AL (Haag-Streit Diagnostics, LS 900; Haag-Streit AG, Koeniz, Switzerland). Following the examination of the fundus using an indirect fundoscope to rule out retinal pathology, retinal oximetry examination was performed. 

Oxygen Examination

The multiwavelength structure–function Coupled Retinal Imager (ROSV-M18; Healthsun Vision, Chengdu, China) was used to measure the retinal vascular oxygen saturation, and Healthsun retinal image quantification analysis (HIQA-1, V1) was used to analyze the acquired images.¹³ The ROSV-M18 was composed of a digital fundus camera, an objective imaging lens, a right-angle reflecting prism, two reflectors, two interference filters, and a three-charge-coupled device (CCD). Unlike the two previously systems,^6,14 this system was simpler in structural composition, with dual-wavelength light (570 and 600 nm) imaged at different positions of the same CCD to achieve simultaneous imaging. 

Participants were seated for 20 minutes before the oximetry examination and then sat comfortably with dilated pupils at least 6 mm in front of the oximeter. Images were centered on the optic disc, and then the focal length was adjusted to obtain a clear image. The acquisition view angle was 45°.¹⁴ Clear images obtained the first time were used, and those who could not be captured with a clear image within three times were excluded from the study. 

Two concentric circles were made with the optic disc as the center and 1.5 and 3.0 optic disc diameters as the radius, respectively. Vessels wider than 8 pixels and longer than 50 pixels between the two concentric circles were selected for oxygen saturation analysis¹³ (Fig. 2). 

Figure 2.

 

Disc-centered fundus image for retinal oximetry. Circle A: 3 times optic disc diameters as radius; circle B: 1.5 optic disc diameters as radius. Vessels between the two circles were included in the calculation.

In this study, the weighted average of arterial and venous oxygen saturation was calculated. S represents oxygen saturation, D represents vessel diameter, and the weighted mean was calculated as Equation (1):  

\begin{eqnarray}S = \frac{{{S}_1 \times D_1^4 + {S}_2 \times D_2^4 + {S}_3 \times D_3^4 + {S}_4 \times D_4^4 + {S}_5 \times D_5^4}}{{D_1^4 + D_2^4 + D_3^4 + D_4^4 + D_5^4}}\ \end{eqnarray}

(1)

 

Oxymap Image Analysis Consistency

To evaluate the consistency of the results for retinal vascular oxygen saturation, funds images of 50 eyes were analyzed by two trained technicians back to back. Kendall's W was used to assess the concordance of the analysis of the images. The consistency coefficients were 0.940 and 0.946 for retinal arteriolar (SaO₂) and venous (SvO₂) oxygen saturation, respectively (P < 0.001 for both). The consistency was acceptable. 

Statistical Analysis

The statistical analysis was performed using IBM SPSS software (Version 25.0; IBM Corp., Armonk, NY, USA). One-way ANOVA was used to compare parameters (age, SE, AL, IOP, CCT, K, SaO₂, SvO₂, and retinal arteriovenous oxygen saturation difference [AVD]) among the four groups categorized by refractive error and by AL. Pearson's correlation was conducted to study the associations among the independent variables. Univariate regression analysis was used to assess the relationship between retinal oxygen saturation and the variables of AL and SE. Multivariate regression analysis was performed to investigate the correlations between the independent parameters (age, IOP, SE, AL, CCT, and K) and the dependent parameters (SaO₂, SvO₂, and AVD). Kendall's W was used to evaluate the concordance of the image analysis. Statistical significance was defined as P < 0.05. 

Results

Of the 1485 myopic adults included in this study, 112 patients were excluded. Among the excluded participants, 21 had CDVA (logMAR) >0.10, 32 had ocular or systemic disease, and 59 were unable to cooperate in making a qualified image. Hence, 1373 adults (524 [38.2%] men, 849 [61.8%] women) aged 26.31 ± 6.93 years (range, 18–45 years) with myopia were eventually enrolled. The mean SE of all participants was −6.40 ± 3.12 D (range, −16.75 to −0.50 D), mean AL was 26.10 ± 1.39 mm (range, 22.66–30.21 mm), mean IOP was 15.65 ± 2.78 mm Hg (range, 8–27 mm Hg), and mean K was 43.74 ± 1.54 D (range, 38.75–48.82 D). The overall retinal oxygen saturations were 93.47% ± 1.58% (range, 86.20%–97.54%) of retinal arteries and 63.08% ± 4.46% (range, 46.00%–70.00%) of venous arteries. All participants were divided into four groups according to myopic refractive error or axial length. General characteristics are provided in Table 1 and Table 2. 

Table 1.

 

Retinal Vascular Oxygen Saturation by Refractive Error

 

Table 2.

 

Retinal Vascular Oxygen Saturation by Axial Length

 

In the one-way ANOVA analysis, Figure 3 shows the trend of SaO₂ with age. In the simple linear regression, Figure 4 indicates that SaO₂ and AVD increased with age (SaO₂: r² = 0.012, slope = 0.025, P < 0.001; AVD: r² = 0.004, slope = 0.043, P = 0.019). Univariate regression analysis of SvO₂ and age showed no significant statistical significance. Figure 5 shows that both SaO₂ and SvO₂ increased significantly with SE (simple linear regression, r² = 0.046, slope = 0.110 for SaO₂ and r² = 0.017, slope = 0.183 for SvO₂, P < 0.001 for both), while AVD decreased significantly with SE (r² = 0.003, slope = −0.071, P = 0.045). Myopia is represented by negative values (i.e., the more the myopia, the smaller the value of SE), with SaO₂ and SvO₂ decreasing and AVD increasing with the deepening of myopia, respectively. As shown in Figure 6, there were significant associations between SaO₂ and AL, as well as SvO₂ and AL (r² = 0.050, slope = −0.255 and r² = 0.014, slope = −0.381, respectively, P < 0.001 for both). Univariate regression analysis of ADV and AL showed there is no significant statistical significance. 

Figure 3.

 

Graph shows the mean retinal arterial oxygen saturation for each age from 18 to 45 years.

Figure 4.

 

Scatterplot of the relationship between age and SaO₂ (%) (r² = 0.012, slope = 0.025, P < 0.001) and AVD (r² = 0.004, slope = 0.043, P = 0.019).

Figure 5.

 

Scatterplot of the relationship between spherical equivalent and SaO₂ (%) (r² = 0.046, slope = 0.110, P < 0.001), SvO₂ (%) (r² = 0.017, slope = 0.183, P < 0.001) and AVD (%) (r² = 0.003, slope = −0.071, P = 0.045). (SaO₂, retinal arterial oxygen saturation; SvO₂, retinal venous oxygen saturation; AVD, retinal arteriolarvenular oxygen saturation difference).

Figure 6.

 

Scatterplot of the relationship between axial length and SaO₂ (%) (r² = 0.050, slope = −0.255, P < 0.001) and SvO₂ (%) (r² = 0.014, slope = −0.381, P < 0.001).

In the multivariate analysis of retinal vessel oxygen saturation (SaO₂, SvO₂, and AVD) using multiple linear regression, gender, age, IOP, SE, K, CCT, and IOP* CCT were included as independent variables. Retinal vascular oxygen saturations were significantly correlated with age (standardized coefficient: β = 0.077, P = 0.006 for SaO₂; β = −0.055, P = 0.048 for SvO₂; and β = 0.080, P = 0.006 for AVD, respectively), SE (standardized coefficient: β = 0.220, P < 0.001 for SaO₂; β = 0.131, P < 0.001 for SvO₂; and β = −0.050, P = 0.048 for AVD, respectively), and K (standardized coefficient: β = 0.094, P = 0.001 for SaO₂), and SE explained the most variations in SaO₂ (Table 3). In the multivariate analysis of retinal vessel oxygen saturation, with AL included as one of the independent variables, retinal vascular oxygen saturations were significantly correlated with age (standardized coefficient: β = 0.088, P = 0.002 for SaO₂; β = −0.050, P = 0.086 for SvO₂ and β = 0.079, P = 0.007 for AVD), AL (standardized coefficient: β = −0.230, P < 0.001 for SaO₂; β = 0.151, P < 0.001 for SvO₂; and β = −0.065, P = 0.035 for AVD, respectively) (Table 4). In the Pearson correlation analysis between the independent variables of gender, age, IOP, AL, SE, K, and the correlation coefficients were −0.752 and −0.437, respectively (P < 0.001 for both, Fig. 7). The correlation coefficient between AL and SE was too large in absolute values, and therefore, AL could not be included in the multiple linear regression analysis at the same time as SE. 

Table 3.

 

Multivariate Analysis of the Association Between Gender, Age, IOP, SE, K, CCT, and Retinal Vascular Oxygen Saturation

 

Table 4.

 

Multivariate Analysis of the Association Between Gender, Age, IOP, AL, K, CCT, and Retinal Vascular Oxygen Saturation

 

Figure 7.

 

Heatmap of the correlation coefficients between two variables in the seven variables of age, gender, IOP, SE, AL, CCT, and K. *P < 0.05, **P < 0.01, ***P < 0.001.

Discussion

In total, 1485 healthy Chinese adults with myopia were included to explore the associations between retinal oxygen saturation and related factors, such as gender, age, SE, AL, IOP, CCT, and K. This is the first study on retinal vascular oxygen saturation in a large sample of healthy Chinese myopic adults. 

In the healthy myopic population, there was no significant correlation between gender and SaO₂, SvO₂, and AVD. This is consistent with a study¹⁰ of retinal oxygen saturation in adults regarding gender. In a study⁴ of retinal oxygen saturation in children and adolescents, AVD was lower in females than in males, and the article attributed this to a lower basal metabolic rate in females than in males during adolescence. 

Previous studies^10,11 indicated that retinal oxygen saturation at different ages and the trends with age were different. However, the sample size of the study by Jani et al.¹⁰ was only 61, and the age ranged from 19 to 74, which was too wide to reflect in detail the changes in retinal oxygen saturation at different ages. In Waizel et al.,¹¹ the average retinal oxygen saturation was compared between each group at 10 years of age, which could not accurately reflect the change trend with age. In our study, considering the wide range of ages and small sample size in previous studies¹⁰ in adults, 1373 participants were included with an age range of 18 to 45 years, in which the growth and development gradually stabilized. 

In the present study, SaO₂ and AVD increased with age, while SvO₂ decreased with age (i.e., the measured retinal arterial oxygen supply and oxygen consumption by retinal tissue increased with age). Mohan et al.¹⁵ found a negative correlation between the measured saturation values and retinal nerve fiber layer thickness. Older individuals have a thinner retinal nerve fiber layer thickness,¹⁶ and this may be partly responsible for the increase in arterial oxygen saturation with age. Jani et al.¹⁰ compared retinal oxygen saturation in multiethnic adults (19–74 years) and concluded that retinal arterial and venous oxygen saturation decreased with age, which was in contrast to the results in the present study. The reason might be that the age range included in the present study was narrower (18–45 years), whereas the age span in Jani et al.¹⁰ was larger (19–74 years) and the mean age was older (44.1 ± 14.7 years), which was close to the high age value in our study. Therefore, it was proposed that there might be a downward trend for arterial oxygen saturation after age 45 years. In the line graph of SaO₂ in this study (Fig. 3), although SaO₂ in the age range of 18 to 45 years tended to increase, the arterial oxygen saturation decreased at around age 40 years, and thus we postulated that SaO₂ increased before approximately age 40 years and decreased after around 40 years. 

The changes of AVD with age in Waizel et al.¹¹ were consisted with our results. AVD was the difference between arterial and venous oxygen saturation and represented the level of arterial oxygen saturation decreased after being consumed and utilized by retinal tissues. Waizel et al.¹¹ concluded that the metabolism of retinal tissue was similar to that of brain tissue. Compared with children, adults have highly developed aerobic glycolysis and increased energy demand in a mature brain; therefore, they have a greater oxygen demand (i.e., greater AVD levels). In adults, the supply of energy for cellular metabolism is primarily provided by oxidative phosphorylation within the mitochondria. Since mitochondrial function and quality have been associated with normal aging,¹⁷ we assumed that the oxygen utilization rate of the retina decreased with age, and more oxygen was required to provide the needed energy. This was also associated with increased susceptibility of the brain to ischemic and degenerative diseases of older individuals.¹⁸ 

In the multivariate regression analysis, K was significantly positively associated with SaO₂, and to our knowledge, this is the first time that the relationship between corneal curvature and retinal oxygenation has been revealed. In a study on ocular biometry in children and adolescents, Gopinath et al.¹⁹ found that larger K values were significantly correlated with narrow retinal vessel caliber. Previous studies have found a negative relation between retinal oxygen saturation and retinal vessel diameter,²⁰ which may be due to the retinal oxygen saturation and vascular caliber being interrelated. In the multifactor regression analysis (Table 3), we found that the K value was positively correlated with SaO₂, and Table 4 shows that AL was negatively correlated with SaO₂. In patients with myopic refractive error, the higher the average K value, the shorter the AL²¹; therefore, the correlation between K value and SaO₂ may be essentially the AL and SaO₂ being interrelated. However, we prefer to attribute this to the intrinsic pathophysiologic mechanisms that cause individual deviation. 

In studies on the effect of refractive error on retinal oxygenation in adults,^8,22,23 the results were consistent with findings of the present study that the higher the myopia, the lower the measured arterial oxygen saturation. In studies on refractive error and oxygen saturation in children and adolescents,^4,9 the conclusions were opposite to the findings of the present research. Age might be the reason for the different conclusions, and the decline of oxygen demand due to the atrophy of the retina and choroid with the increase of myopia might be the underlying cause. 

In the multivariate regression analyses with AVD as the dependent variable and the simple linear regression analysis of SE and AVD, the AVD increased with increasing myopic refraction, in contrast to the results of Man et al.²⁴ and Zheng et al.²² Longer AL was found to be associated with lower AVD and accompanied by lower retinal function in the study by Man et al.²⁴ Zheng et al.²² found that lower AVD was due to lower oxygen demand by fewer functional neurons. In this study, there was no significant linear relationship between AL and AVD (not exhibited in the results), so we deem that other factors contributed to the refraction power, such as corneal curvature, which affected the measured values of AVD. According to previous studies on the relation between reduced capillary perfusion and AL,²⁵ it was found that the capillary perfusion of the retina was reduced in eyes with longer AL. We assumed that with lower perfusion, in order to obtain adequate oxygen, the retinal tissue needed to extract more oxygen per unit blood flow, which led to an increase in AVD in higher myopia. Thus, we could not conclude that the consumption of oxygen was increased with the deepening of myopia, and this requires further research. 

The patients included in this study were all candidates for corneal refractive surgery, so there might be some selection bias. Particularly, there was an obvious trend of rejuvenation with age, but a wide age range might mask the different trends of retinal oxygen saturation in different age groups. In Figure 3, SaO₂ apparently increased with age before about 35 years, and after 35 years, SaO₂ might slightly decrease with age, but this did not affect the overall trend of SaO₂ with age. In future research, individuals with a larger age range are needed to further investigate the influence of age. Additionally, this study was a population-based cross-sectional study, so the causal relationship between myopia and retinal artery oxygen saturation could not be determined. Therefore, further cohort studies on myopia and SaO₂ are needed to explore the causal relationship. 

Conclusions

In healthy myopic adults, retinal arterial oxygen and venous oxygen saturation decreased with increasing myopia. In the age range of 18 to 45 years, retinal arterial oxygen saturation and retinal arteriovenous oxygen saturation difference increased with age. The association between ocular biometry and retinal oxygen saturation (i.e., average keratometry and retinal arterial oxygen saturation) in the present study indicates that there would be some inherent individual ocular differences that contribute to the different arterial oxygen saturations, but the underlying causes need further investigation. 

Acknowledgments

Disclosure: S. Ge, None; Y. Zhou, None; C. Li, None; M. Zhang, None