Animal experiments indicate that the inner retina keeps its oxygen extraction constant despite systemic hypoxia. For the human retina no such data exist. In the present study we hypothesized that systemic hypoxia does not alter inner retinal oxygen extraction. To test this hypothesis we included 30 healthy male and female subjects aged between 18 and 35 years. All subjects were studied at baseline and during breathing 12% O2 in 88% N2 as well as breathing 15% O2 in 85% N2. Oxygen saturation in a retinal artery (SO2art) and an adjacent retinal vein (SO2vein) were measured using spectroscopic fundus reflectometry. Measurements of retinal venous blood velocity using bidirectional laser Doppler velocimetry and retinal venous diameters using a Retinal Vessel Analyzer (RVA) were combined to calculate retinal blood flow. Oxygen and carbon dioxide partial pressure were measured from earlobe arterialized capillary blood. Retinal blood flow was increased by 43.0 ± 23.2% (P < 0.001) and 30.0 ± 20.9% (P < 0.001) during 12% and 15% O2 breathing, respectively. SO2art as well as SO2vein decreased during both 12% O2 breathing (SO2art: −11.2 ± 4.3%, P < 0.001; SO2vein: −3.9 ± 8.5%, P = 0.012) and 15% O2 breathing (SO2art: −7.9 ± 3.6%, P < 0.001; SO2vein: −4.0 ± 7.0%, P = 0.010). The arteriovenous oxygen difference decreased during both breathing periods (12% O2: −28.9 ± 18.7%; 15% O2: −19.1 ± 16.7%, P < 0.001 each). Calculated oxygen extraction did, however, not change during our experiments (12% O2: −2.8 ± 18.9%, P = 0.65; 15% O2: 2.4 ± 15.8%, P = 0.26). Our results indicate that in healthy humans, oxygen extraction of the inner retina remains constant during systemic hypoxia.
- retinal oxygen metabolism
- retinal blood flow
- oxygen extraction
hypoxia is a major trigger of retinal disease (3, 6, 45). The human eye has a complex system of oxygen delivery. The posterior pole of the eye is nourished by two vascular beds with different circulatory characteristics (31, 43, 44, 51). The retinal circulation delivers oxygen to the inner retina including the retinal ganglion cells. It is characterized by a high arteriovenous oxygen difference, high vascular resistance, and, accordingly, by a low blood flow rate. Retinal vessels after the level of the lamina cribrosa lack neural innervation. Hence, blood flow in the retina is primarily regulated by local metabolic and myogenic factors. The choroidal circulation delivers oxygen to the outer retina including the photoreceptors. The choroid has a very high perfusion rate and is accordingly characterized by a low arteriovenous oxygen difference (27). The choroidal vessels are richly innervated, and vascular tone is under neural control (44).
Most of our knowledge regarding oxygen metabolism in the retina stems from animal experiments using invasive techniques to get insight into the local oxygen tension. Microelectrodes have been introduced more than 50 years ago to measure oxygen tension and have since then been improved and reduced in size, thereby preventing a potential effect of the electrode itself on the local O2 environment (26, 35, 57, 62). In humans microelectrodes have been employed intra-operatively to study retinal oxygen tension in the vitreous (54).
A method that is applicable in humans is fundus reflectometry. The technique is based on Lambert Beer's law of absorption of light. Such technology has been used for decades to measure systemic oxygen saturation using finger plethysmography in transmission, but is more difficult to be used in reflection, as it is required for the retina (9, 11). The development of high-speed charge coupled device cameras has improved reliability and reproducibility of the measurements, and two techniques recently became commercially available to assess oxygen saturation in retinal arteries and veins in humans. As such values for retinal arterial and venous oxygen saturation have been published in patients with glaucoma (40, 41), age-related macular degeneration (17), diabetes (20, 25), and retinal vascular occlusion (23, 24). It is, however, difficult to fully accomplish the meaning of these data, because none of the studies has measured retinal blood flow and could therefore draw conclusions on retinal oxygen extraction.
In the present study we measured oxygen saturation in retinal arteries and veins during normoxia and at two different levels of hypoxia in healthy humans. In addition, we measured retinal blood flow and calculated retinal oxygen extraction from these values. Based on data from previous animal experiments (7, 60), we hypothesized that retinal oxygen extraction would stay relatively constant during systemic hypoxia.
RESEARCH DESIGN AND METHODS
The protocol of the present study was approved by the Ethics Committee of the Medical University of Vienna and followed the guidelines set forth in the Declaration of Helsinki. A total of 30 healthy male and female subjects between 18 and 35 years were included in the present study. All subjects gave their written informed consent and passed a screening examination including physical examination, 12-lead electrocardiogram, assessment of visual acuity, slit lamp biomicroscopy, funduscopy, and measurement of intraocular pressure. Exclusion criteria were ametropia ≥ 3 diopters, anisometropia ≥ 1 diopter, current or former smoking, any sign of ocular disease, and any clinically relevant illness as judged by the investigators to potentially interfere with the aims of the present study. In addition, blood donation or intake of any medication in the 3 wk preceding the study were exclusion criteria. Twelve hours before the study day, participants had to abstain from beverages containing alcohol or caffeine.
The study was performed in a randomized two-way cross-over design. Subjects received 12% oxygen plus 88% nitrogen and 15% oxygen plus 85% nitrogen (Messer Group, Vienna, Austria) as inhalant during the study day. The order of the two breathing periods was masked and randomized.
To achieve dilatation of the pupil one drop of tropicamide (Mydriatikum AGEPHA, Vienna, Austria) was instilled into the study eye. Thereafter, a resting period of at least 20 min was scheduled. When blood pressure had stabilized, retinal blood flow in a major retinal vein was measured by combining measurements of retinal vessel diameters using the RVA and blood velocity using bidirectional laser-Doppler velocimetry (LDV). RVA was used for measuring retinal arterial diameter. The fundus images obtained by the RVA were used to measure retinal oxygen saturation in retinal arterioles and venules. In addition to these variables, baseline measurements of blood pressure, pulse rate, systemic arterial oxygen tension, and blood gases from arterialized blood samples out of the ear lobe were taken. Thereafter, one of the breathing periods to induce hypoxia was scheduled. This breathing period was scheduled for 30 min, and the outcome parameters were assessed after 15 min of inhalation. After the first breathing period a break of 2 h was scheduled. Afterward, baseline values were assessed again and the second breathing period was started thereafter. All periods followed exactly the same time scheduled, and outcome parameters were measured in a predetermined order.
Measurement of retinal vessel diameters and oxygen saturation.
The diameters of one major temporal retinal artery (Dart) and vein (Dvein) within 1 to 2 disc diameters from the center of the optic disc were measured using the RVA (IMEDOS, Jena, Germany) (15). The same device was used to measure arterial and venous retinal oxygen saturation levels (SO2art, SO2vein) using an oxygen module (IMEDOS). This system consists of a fundus camera (FF 450; Carl Zeiss Meditec AG, Jena, Germany), a high-resolution digital video camera, and a personal computer with analyzing software. The measurement of oxygen saturation in retinal vessels is based on fundus reflectometry using spectral analysis of reflected light at selected wavelengths (21, 22). In the present system two fundus pictures at wavelengths of 610 and 545 nm are taken simultaneously. Measurement of retinal oxygen saturation is based on the fact that oxygenated hemoglobin has different light absorption characteristics as compared with nonoxygenated hemoglobin. The isobestic wavelength of 548 nm is the point within the light spectrum where oxygenated and nonoxygenated hemoglobin show identical absorption. At a wavelength of 610 nm, oxygenated hemoglobin is nearly transparent and absorption is dominated by nonoxygenated hemoglobin. Measuring the contrast at these wavelengths enables determination of the relation between oxygenated and total hemoglobin and thus calculation of oxygen saturation.
The measuring procedure and the subsequent image analysis were performed as published previously (21, 22). For this purpose fundus images centered at the optic nerve head were taken. Measurement of SO2vein and SO2art was performed at the same position where diameter assessment was done. The arteriovenous oxygen difference was calculated as the difference between SO2art and SO2vein.
For measurement of retinal venous blood velocity a fundus camera-based bidirectional laser Doppler velocimetry system was used (LDV-5000; Oculix, Arbaz, Switzerland). Measurements were performed in retinal veins at the same locations at which diameter and oxygen saturation were assessed. For this purpose the fundus image automatically taken with the RVA was used and the site of measurement was marked during the assessment of vessel diameter. This image was then used to find the exact same position for oxygen saturation as well as blood flow velocity measurements. The principle of LDV is based on the optical Doppler effect. Laser light of high coherence length with a wavelength of 670 nm is scattered by moving red blood cells. The frequency shift is proportional to the blood flow velocity in the retinal vessel and also depends on the angle between the incident light and the blood flow in the vessel. The maximum Doppler shift corresponds to the centerline erythrocyte velocity at the center of the vessel - Vmax (46). To overcome the angle dependence the Doppler shift power spectra are recorded simultaneously for two directions of the scattered light in the image plane of the fundus camera. With the use of this approach, the absolute blood velocity can be calculated independent of the angle of incidence (47, 48). The scattering plane can be rotated and adjusted in alignment with the direction of Vmax. After calculation of the absolute blood velocity, the angle of incidence can be calculated based on the data of both detection channels (61), enabling a quality control for the measurements. Only if the angle of incidence as calculated from the two channels is equal, the measurements can be considered accurate. In the present study we considered the measurements adequate if the agreement was within 0.5 rad. If this criterion was not reached, data were not accepted as accurate. From Vmax mean blood velocity in retinal vessels can be calculated as follows assuming a parabolic velocity profile: vel = Vmax/2, where vel is mean blood velocity and Vmax is the centerline erythrocyte velocity.
Blood flow in the retinal vein under study was calculated as follows: flow = vel·Dvein2·(π/4).
Measurement of systemic hemodynamic parameters.
Systolic blood pressure, diastolic blood pressure, and mean arterial blood pressure were measured on the upper arm by an automated oscillometric device (Infinity Delta, Dräger, Vienna, Austria). The same device automatically recorded pulse rate and systemic oxygen saturation (SaO2) by a finger pulse oximeter.
Blood gas analysis and calculation of oxygen content.
Arterialized capillary blood from the earlobe was collected from a lancet incision into a thin glass capillary tube after a paste was applied to induce capillary vasodilatation (Finalgon; Thomae, Biberach, Germany). Arterial pH, Pco2, and Po2 were determined with an automatic blood gas analysis system. Hemoglobin concentration was measured using standard methods. Oxygen content in the retinal arteries (cO2art) and veins (cO2vein) was estimated using Henry's law: cO2 = hemoglobin concentration·1.34·SO2 + Po2·0.003.
Arterial Po2 was taken from the data obtained with the blood gas analysis system. Venous Po2 was estimated from the oxygen-binding curve at a Pco2 of 37 mmHg and a temperature of 37°C. The arteriovenous difference in oxygen content was calculated as: cO2diff = cO2art − cO2vein.
Finally, retinal oxygen extraction was calculated as: retinal oxygen extraction = flow·cO2diff.
Data are presented as means ± SD. A repeated measures ANOVA model was used to analyze data. Post hoc analysis was done using planned comparisons. Linear correlation analysis was performed to study the association between the change in SO2art and SaO2 as well as the change in arteriovenous oxygen difference and retinal blood flow. This was done for both breathing periods separately to include only independent data in the analysis. A P value <0.05 was considered the level of significance. CSS Statistica for Windows (Version 6.0; Statsoft, Tulsa, CA) was used of statistical calculations.
In two subjects fundus images as obtained with the RVA were not sufficient to measure retinal oxygen saturation. In another subject the LDV data did not fulfill the angle criterion mentioned above. Hence, data presented in this article arise from 27 healthy subjects with a mean age of 25.2 ± 3.9 years (13 male, 14 female). The mean hemoglobin concentration in this population was 14.1 ± 1.4 g/dl. The systemic hemodynamic outcome variables are summarized in Table 1. Breathing of 12% oxygen caused a small decrease in systolic and mean arterial blood pressures, but not in diastolic blood pressure. By contrast, 15% oxygen breathing did not cause any change in systemic blood pressure. Both gas mixtures increased pulse rate with a more pronounced increase during 12% oxygen breathing. As expected, systemic hypoxia was reflected in a dose-dependent decrease in Po2. This decrease in Po2 was accompanied by an increase in pH. The effects on Pco2 were small and only significant during 15% oxygen breathing.
Breathing of gases with reduced oxygen content induced vasodilation in both retinal arteries and veins (Fig. 1). These effects were more pronounced during 12% oxygen breathing than during 15% oxygen breathing (P = 0.02). The increase in Dart was 6.6 ± 4.5% during 12% oxygen breathing (P < 0.001 vs. baseline) and 4.5 ± 4.7% during 15% oxygen breathing (P < 0.001 vs. baseline). The increase in Dvein was in the same order (12% oxygen: 8.0 ± 3.9%, P < 0.001 vs. baseline; 15% oxygen: 4.0 ± 3.3%, P < 0.001 vs. baseline) and also significantly more pronounced during 12% oxygen breathing than during 15% oxygen breathing (P = 0.003). Hypoxia also caused an increase in blood flow velocity. This increase was slightly more pronounced during 12% oxygen breathing (22.5 ± 19.1%, P < 0.001 vs. baseline) than during 15% oxygen breathing (20.0 ± 17.3%, P < 0.001 vs. baseline), but the difference did not reach the level of significance (P = 0.43). Hence flow was also increased by 43.0 ± 23.2% (P < 0.001) and 30.0 ± 20.9% (P < 0.001) during 12% oxygen breathing and 15% oxygen breathing, respectively. The increase in flow was higher during 12% oxygen breathing than during 15% oxygen breathing (P = 0.008).
Figure 2 summarizes the effects of the gas mixtures on SaO2 and on oxygen saturation in the retinal vessels. As expected a decrease in SaO2 was seen during breathing 12% oxygen (−15.8 ± 6.9%, P < 0.001) as well as during breathing 15% oxygen (−6.2 ± 3.7%, P < 0.001). These changes were also reflected in retinal arteries where SO2art decreased by −11.2 ± 4.3% (P < 0.001) and −7.9 ± 3.6% (P < 0.001) during 12% oxygen breathing and 15% oxygen breathing, respectively. On average SO2vein also decreased. This reduction was significant during 12% oxygen breathing (−3.9 ± 8.5%, P = 0.012) and during 15% oxygen breathing (−4.0 ± 7.0%, P = 0.010). In some individuals, however, we observed an increase in SO2vein during both breathing periods. The arteriovenous oxygen difference decreased accordingly during both breathing periods (12% oxygen: −28.9 ± 18.7%, 15% oxygen: −19.1 ± 16.7%, P < 0.001 each). This decrease was more pronounced during 12% oxygen breathing than during 15% oxygen breathing (P = 0.002).
Figure 3 shows the correlation between the decrease in SaO2 and SO2art for both breathing periods separately. Although the correlation was significant for both regimens a higher correlation coefficient was found during 12% oxygen breathing. Correlation analysis was also done between the change in flow and the change in the arteriovenous oxygen difference (Fig. 4). For both gas mixtures, a significant negative association was observed.
The results for oxygen content and oxygen extraction are presented in Fig. 5. Arteriovenous oxygen content difference decreased during 12% oxygen breathing (−30.1 ± 17.7%, P < 0.001) as well as during 15% oxygen breathing (−19.9 ± 14.9%, P < 0.001). Oxygen extraction did, however, not change during either of the gas mixtures (12% oxygen: −2.8 ± 18.9%, P = 0.65; 15% oxygen: 2.4 ± 15.8%, P = 0.26).
In keeping with animal experiments our data indicate that inner retinal oxygen extraction stays almost constant during graded hypoxia in healthy humans. Whereas both breathing of 12% oxygen and 15% oxygen mixtures caused a significant decrease in arteriovenous oxygen difference, retinal blood flow increased, thereby leaving retinal oxygen extraction constant. In addition, we observed a significant negative association between the change in arteriovenous oxygen difference and the change in retinal blood flow during both breathing periods (Fig. 4). This indicates that the increase in blood flow as observed in the present study is closely related to the degree of hypoxia.
Animal experiments in several species have shown that in the inner retina tissue Po2 is regulated when SaO2 is decreased. In the cat preretinal tissue Po2 decreases by 0.14 to 0.18 mmHg/mmHg systemic Po2 when the systemic Po2 is above 35 mmHg (2, 12). When the systemic Po2 decreases below 35 mmHg the value for Po2 in the inner retina decreases more steeply. The regulation of inner retinal tissue Po2 during systemic hypoxia is due to the increase in retinal blood flow. This was also shown in recent rat experiments in which phosphorescence lifetime imaging was used to measure retinal arterial and venous Po2 and red-free and fluorescent microsphere imaging to measure retinal blood flow (58). Whereas, in keeping with our findings, the oxygen extraction in the inner retina was well regulated during breathing 15% O2 + 85% N2, it declined when the hypoxia was more severe (10% O2 + 90% N2). This was related to the increase in retinal blood flow, which was seen during 15% oxygen breathing, but not further enhanced during 10% oxygen breathing (58). To the best of our knowledge no previous data on oxygen extraction are available in humans during systemic hypoxia. It was, however, shown that breathing 12% O2 + 88% N2 in healthy humans for 5 min causes a slight increase in implicit times of the major negative (N95) components of the pattern electroretinogram, but did not affect the major positive component (P50) (28).
An increase in retinal blood flow during systemic hypoxia was seen in cats (1, 38), newborn piglets (39), nonhuman primates (13), and humans (14, 55). Little is, however, known about the mechanisms that mediate this retinal vasodilator response. In the cat it has been shown that nitric oxide contributes to increased retinal blood flow during hypoxia and also contributes retinal hypoxic hyperaemia through a flow-induced mechanism (38). In the newborn pig, however, inhibition of nitric oxide synthase did not alter the retinal vasodilator response to systemic hypoxia (19). A recent in vitro study found hypoxia-induced relaxation of retinal arterioles depended on nitric oxide and prostaglandins in isolated porcine vessels. Furthermore, the presence of perivascular retinal tissue is required to induce vasodilatation in these experiments (30).
In the outer retina systemic hypoxia leads to a pronounced reduction in choroidal Po2 (34), because choroidal blood flow is almost insensitive to changes in oxygen tension (5, 18, 29, 50, 52, 53). Animal experiments indicate a very small arteriovenous oxygen difference in the choroid during normoxia (2). This is most likely also the case in humans although no data are available, because neither choroidal Po2 nor SO2 can be measured noninvasively in humans. As such our results do not provide any information on outer retina oxygen extraction.
The present study has a number of limitations that need to be considered. Baseline SO2art was lower at baseline than SaO2. Whether this is due to a calibration error of the reflectometric measurements is unknown. In the absence of a true gold standard technique this is difficult to answer, because true validation of measurements is impossible. Two previous studies in patients with systemic hypoxia due to either Eisenmenger syndrome (56) or chronic obstructive pulmonary disease (42) have measured retinal oxygen saturation using spectroscopic reflectometry. In both studies SO2art values correlated well with SaO2. This is well compatible with the results of the present study where an association was found between the hypoxia-induced changes in SO2art and SaO2 (Fig. 3).
One also needs to consider that we measured blood flow in only one vein. Moreover, SO2vein was also only measured in the same vein and SO2art in the adjacent artery. This raises the question of whether our results can be generalized to the entire retina. There is, however, no evidence indicating that oxygen extraction of the inner retina may be differentially regulated in different quadrants of the retina. In principle the technique as used in the present study may also be used to measure total retinal blood flow, but for this approach all the vessels entering the optic nerve head need to be measured separately (16, 49). Alternatively, Doppler OCT, a relatively new technique, may be used to measure total retinal blood flow and, combined with spectroscopic reflectometry, allow for calculation of human retinal oxygen extraction (4, 8, 10, 33, 59, 61). In the present study we also assumed that hypoxia induced the same increase in venous and arterial blood flow, which is, however, expected in an end organ. As such it is clear that values for oxygen extraction as presented in Fig. 5 cannot be considered to be valid for the entire retina. During both breathing period Pco2 slightly changed, which may influenced our results. We have, however, previously shown that during 100% oxygen breathing the effects of changes in Pco2 on retinal hemodynamics is relatively small (36).
This does, however, not limit the main conclusion of the present article that during mild hypoxia oxygen extraction remains constant in the human retina in healthy young subjects. Obviously this conclusion only holds true within the limits of detection of our study. Our sample size calculation was based on the previously published reproducibility data of the techniques in our laboratory (32, 37) and a minimum detectable change of 12% in oxygen extraction. Furthermore, the present study was performed in healthy young subjects. To which degree our results can be extrapolated to an elderly population is unclear, and further studies are needed to investigate retinal oxygen extraction in older subjects or patients with vascular disease.
In conclusion, our data indicate that during systemic hypoxia as induced by breathing either 12% O2 + 88% N2 or 15% O2 + 85% N2 does not alter inner retinal oxygen extraction. Although the arteriovenous oxygen difference decreases, an increase in retinal blood flow ensures that the inner retina does not become hypoxic. The data of the present study also indicate that only limited information can be obtained about oxygen metabolism from measurements of SO2art and SO2vein, if blood flow is not measured concomitantly. With the RVA system and other system for the measurements of oxygen saturation some information on the response of retinal blood flow may be obtained when vessel diameter is measured, but this does not allow for quantification of oxygen extraction. Oxygen extraction is a fundamental parameter of retinal metabolism. Measurement of retinal oxygen extraction will lead to a clearer understanding of retinal oxygenation in health and disease. The present technique may be suitable to gain insight into the degree of retinal hypoxia in diseases such as diabetic retinopathy.
Financial support from the Austrian Science Foundation (FWF, project P26157) is acknowledged.
- Copyright © 2014 the American Physiological Society