Systemic Hyperoxia and Retinal Vasomotor Responses

Abstract

purpose. Several studies have investigated the changes in retinal vessel diameter during physiological stress or pathologic conditions. These studies were principally based on individual fundus photographs and as such did not allow the evaluation of vessel dynamics over time. The research objective was to detail the time course and amplitude changes in the diameter of arteries and veins across all retinal quadrants, during and after hyperoxic vascular stress.

methods. The dynamics of changes in retinal vessel diameter were quantified with a retinal vessel analyzer, which digitizes fundus images in real time and simultaneously quantifies vessel diameter. The arterial and venous diameters within one disc diameter of the optic nerve head in each quadrant were studied. Twenty young adults participated in this study in which the vessel diameters were measured during successive phases of breathing either room air or pure oxygen. The oxygen saturation level (SaO₂), end-tidal carbon dioxide (EtCO₂), pulse rate (PR), respiratory rate (RR), and blood pressure (BP) were also monitored throughout testing.

results. Breathing 100% O₂ caused an increase in SaO₂ and a decrease in the EtCO₂. All other systemic parameters measured (PR, RR, BP, and ocular perfusion pressure [OPP]) remained unchanged. However, the retinal veins and arteries constricted by ∼14% and ∼9% respectively, in all retinal quadrants. After experimental hyperoxia, inhalation of room air was associated with a progressive increase in the caliber of vessels toward their pretest size. The amplitude and overall profile of vessel reactivity to and recovery from hyperoxia was the same across retinal quadrants.

conclusions. These data indicate that, during systemic hyperoxic stress, the retinal vessels change in caliber uniformly across retinal quadrants in healthy young adults. This type of physiological vascular provocation could be used to investigate the quality of vascular regulation during aging and in vascular diseases of the eye.

The retinal vasculature does not have sympathetic innervation ¹ ; rather, it maintains optimal nutrition and oxygenation of the retina through vascular autoregulation. Autoregulation is the ability to keep blood flow constant despite altered perfusion pressure, or to adjust blood flow to the metabolic requirements of a given tissue. ² These mechanisms are mediated, for example, by factors such as the pH level, various endothelial vasoactive agents (endothelin, nitric oxide), or oxygen and carbon dioxide tension. ³  

Ocular hemodynamic responses and retinal vessel reactivity have been studied by using various noninvasive procedures such as aerobic exercise, ^{4 5} altered body position, ⁶ cold pressor stimulation, ⁷ or blood gas perturbations. ^{8 9}  

Provocation through breathing pure oxygen has been used to demonstrate retinal vascular autoregulation, ^{10 11 12 13} seen as a vasoconstriction of retinal vessels ^{10 14 15 16} or a decrease in retinal blood flow. ^{15 16} Pure oxygen breathing has also been used to reveal abnormal autoregulation in ocular diseases such as diabetic retinopathy ¹⁷ or glaucoma. ¹⁸ Other studies have used pure oxygen breathing and reported either regional retinal differences in vascular reactivity, ^{13 19} or no regional differences. ²⁰  

The investigation of retinal vascular reactivity in the various regions of the retina is particularly important in view of reports indicating that some vascular ocular diseases may affect a portion of the retina and/or visual field to a larger degree. In glaucoma, it has been reported that the superior visual field was affected more often than the inferior visual field ²¹ and that retinal vessel narrowing with progression of disease was more pronounced in the inferior retina. ²² In diabetes, vascular anomalies have been reported to be more frequent in the superior retina, ^{23 24} although it has also been shown that decreased visual field sensitivity tends to be localized in the superior quadrants. ²⁵  

To our knowledge, in only one study has the human retinal vessel reactivity to systemic hyperoxia in all retinal quadrants been investigated. ¹³ The authors used a retinal vessel analyzer (RVA) ^{26 27} to study the effect of prolonged hyperoxia on the time course of retinal vasoconstriction. They reported a greater vessel reactivity to oxygen in the temporal than in the nasal arteries and veins. Unfortunately, no data were presented for the individual quadrants or the superior versus inferior retina. Furthermore, no attempt was made to evaluate the time course for recovery from the vascular stress. 

Our specific objective was to detail the time course and magnitude of changes in the diameter of arteries and veins across all retinal quadrants, before, during, and after hyperoxic vascular stress. 

Methods

Subjects

Twenty (9 men, 11 women) healthy, nonsmoking subjects (mean age, 24.1 ± 2.9 years) participated in the present study. All subjects had 20/20 (6/6) visual acuity and good systemic and ocular health and were not taking any vasoactive medication. Subjects were instructed to abstain from alcohol and beverages containing caffeine for 24 hours before the study day. Each subject signed a written consent form after the nature of the study and experimental protocol were explained in detail. All experimental procedures conformed to the tenets of the Declaration of Helsinki and the University of Montreal’s ethical guidelines for human research. 

Subject Preparation

The intraocular pressure (IOP) in the test eye was measured the day before vessel measurements to avoid optical degradation of fundus images. The right pupil of each participant was dilated with 1 drop of tropicamide 1% and 1 drop of phenylephrine 2.5%. Each subject was asked to sit and relax for at least 10 minutes before testing to stabilize the systemic blood pressure (BP) and pulse rate (PR). 

Continuous Blood Pressure Measurement

A continuously recording, noninvasive BP system (NIBP 7000; Colin, San Antonio, TX) was used to monitor systemic BP throughout the experiment. A piezoelectric sensor placed over the radial artery was calibrated to yield BPs identical with those of the brachial artery. The NIBP output was processed on computer (Acknowledge software; BioPac, Santa-Barbara, CA) which provided real-time (25 Hz) measurements of the systolic and diastolic BP and the quantification of the BP_mean and ocular perfusion pressure (OPP). The OPP was derived online from the BP_mean data and the IOP that had been entered into the analysis program for each subject. The following formulas were used for deriving the BP_mean and the OPP: BP_mean = BP_diast + (BP_syst − BP_diast)/3 and OPP = \({2}/{3}\) BP_mean − IOP. 

Gas Breathing System and Physiological Variables

A soft rubber mask (7930 series; Hans-Rudolph, Kansas City, MO) was placed over the subject’s mouth and nose, with the nares clipped closed, to ensure that breathing would be through the mouth only. The subject was then positioned at the fundus camera component of the RVA system. Two one-way valves allowed the test gas to be inhaled on one side and exhaled on the other side of the mask. The exhaled air was fed into a capnograph/oximeter system (model 7100 CO₂SMO, Novametrix; Trudell Medical, Montréal, Québec, Canada) for quantification of the end-tidal carbon dioxide (EtCO₂) and respiratory rate (RR). An infrared sensor attached to the CO₂SMO system was clipped to the middle finger of the left hand for oxygen saturation (SaO₂) and PR measurements. The EtCO₂, RR, SaO₂, and PR were monitored continuously throughout the experiment. 

Retinal Vessel Analyzer

The RVA system is composed of a fundus camera (Model FF 450; Carl Zeiss Meditec, Jena, Germany) that visualizes the fundus vessels and an image digitization system for real-time (∼25 Hz) and off-line analysis of the vessel caliber. 

The subject’s fixation in the fundus camera was adjusted to position the optic nerve head (ONH) in the center of a fundus monitor. The fundus camera and the green background illumination were adjusted to provide uniform lighting and maximum focus of the retinal vessels displayed on the screen monitor. Then, the experimenter positioned a square box over the retinal region of interest (ROI) containing the vessel to be analyzed (Fig. 1) . This ROI contained reference landmarks used to track eye movements and optimize recordings of the vessel diameter. A rectangular cursor was then placed over the vessel of interest to identify the vessel length, approximately 0.5 mm, to be analyzed for changes in caliber throughout experimentation. The RVA program then initiated analysis of vessel diameter over the entire length of vessel within the rectangular cursor. Simultaneously, an event marker was put on the NIBP and CO₂SMO systems to indicate the start of the experiment. Throughout testing, the subject was encouraged to blink and maintain steady fixation. Sections of recordings contaminated by eye movements or blinks were automatically eliminated from analysis. Fundus images were stored on an s-VHS videotape recorder for off-line measurement of other vessels within the captured field of view. All vessels analyzed were located within 1 disc diameter (DD) from the ONH border. For each subject, one artery and vein were analyzed in the superior-temporal (ST), superior-nasal (SN), inferior-temporal (IT), and inferior-nasal (IN) quadrants of the retina. 

Test Gas

All testing was performed in a quiet, dimly illuminated laboratory, at normal room temperature and at sea level atmospheric pressure. The test protocol was divided into three consecutive phases in which the subject had to inhale: (1) room air for 2 minutes, (2) 100% oxygen for 12 minutes, and (3) room air for 12 minutes, through the air-tight face mask. These three phases were considered the baseline, pure oxygen-breathing, and recovery phases, respectively. At the end of baseline, the reservoir bag filled with oxygen was connected to the mask to start inhalation of pure oxygen, and this bag was quickly disconnected once the recovery phase began. Event markers were placed on each of the NIBP, CO₂SMO, and RVA systems to indicate the beginning and the end of the three test phases. 

Data Analyses

Each data set for the SaO₂, EtCO₂, PR, RR, BP_mean, BP_syst, BP_diast, and OPP was group averaged across subjects, as a function of time into the experiment, to generate population trends in data over time. To quantify subject responses to each physiological variable, all data for each experimental phase were group averaged. However, for the SaO₂ and EtCO₂, only the most steady segments during breathing of pure oxygen were averaged for that particular experimental phase. 

For each subject and vessel, the data were normalized to a common baseline level by assigning 100% to the mean value averaged over the 2-minute baseline interval. All normalized data were then group averaged across subjects for each artery and vein measured in each retinal quadrant to evaluate changes over time. The nine separate variables measured for vessel responses throughout the experiment are identified schematically in Figure 2 . Data analyses included the following: (1) average of all baseline recordings, (2) time before vasoconstriction response to oxygen, (3) rate of the vasoconstriction, (4) latency before the plateau in vasoconstriction, (5) average of all vessel measurements in the vasoconstriction plateau contained in the pure oxygen phase, (6) time for the onset of recovery from vasoconstriction, (7) rate of the recovery from vasoconstriction, (8) latency before the plateau in recovery, and (9) average of all vessel caliber values in the recovery plateau. These parameters were group averaged across all subjects to reveal the population response to the physiological provocation of breathing pure oxygen. 

Statistical Analyses

All data are expressed as the mean ± SEM. Analyses of variance (ANOVAs) and Student’s t-tests were used for data comparisons between the various parameters and across subjects. Differences across test conditions were considered significant at α = 0.05. 

Results

Figure 1shows the placement of the ROI cursor and the smaller rectangle encompassing the vessel length to be measured throughout the study. The group-averaged values for each test variable and physiological parameter measured throughout each test phase are presented in Table 1 . The group-averaged SaO₂ increased from 97.4% to 98.7% (P < 0.0001) during systemic hyperoxia and returned to baseline during the recovery phase. The group-averaged EtCO₂ decreased by 9.6%, from 39.5 to 35.7 mm Hg (P < 0.0001) during pure oxygen breathing and returned to baseline during the recovery phase. The group-averaged PR, RR, BP_syst, BP_diast, BP_mean, and OPP did not change (P > 0.05) throughout the experiment. Figure 3presents the group-averaged data for SaO₂ (Fig. 3A) , EtCO₂ (B), PR (C), and RR (D), over time during testing. The BP_syst, BP_mean, BP_diast, and OPP did not vary significantly throughout the test phases, as reported in Table 1 . 

Table 2provides vessel data group averaged across retinal quadrants and hemiretina. Group-averaged baseline vessel diameters, combined across temporal and nasal quadrants and superior and inferior quadrants, revealed that the temporal vessels were larger than nasal vessels, but vessels in the superior and inferior retina did not differ. 

Figures 4 and 5present the percentage of change relative to baseline in the group-averaged diameters of the arteries and veins, respectively, in the ST (Figs. 4A 5A) , SN (B), IT (C), and IN (D) retinal quadrants throughout testing. These data indicated that systemic hyperoxia induced a group-averaged arterial constriction of 8.7%, 9.0%, 8.9%, and 10.0% and an averaged venous constriction of 15.5%, 13.9%, 13.5% and 14.9% for the retinal quadrants, respectively (P < 0.0001 for all vessels). The degree of vasoconstriction differed in neither veins nor arteries across the quadrants (P > 0.05). By the end of recording, all vessels were within 1% to 3% of prehyperoxia testing diameter. The data were further analyzed, comparing the vessel reactivity in the temporal versus nasal and the superior versus inferior retina. The results of this analysis indicated that the timing and degree of vasoconstriction of the retinal arteries and veins were uniform in the temporal versus nasal retina (P > 0.05) and superior versus inferior (P > 0.05) retina. 

A further analysis revealed that the degree of vasoconstriction of the veins and the arteries did not differ between male and female subjects in any of the four quadrants (P > 0.05). Furthermore, the degree of vasoconstriction of the veins and the arteries did not differ across quadrants in either the men or the women (P > 0.05). 

An analysis of the rate of vasoconstriction (Fig. 2 , parameter 3), which was derived from each subject’s data for arteries and veins in each quadrant and subsequently group averaged across subjects, revealed that the slope of the line drawn through the data points forming parameter 3 in both the arteries and veins did not differ (P > 0.05) across retinal quadrants. The rate of recovery from retinal vasoconstriction (Fig. 2 , parameter 7) in both the arteries and the veins also did not differ across retinal quadrants (P > 0.05). 

The delay time (Fig. 2 , parameter 2; mean: arteries, 16.4 ± 2.0 seconds; veins, 25.1 ± 2.6 seconds) and latency (Fig. 2 , parameter 4; mean: arteries, 194.2 ± 12.4 seconds; veins, 274.9 ± 15.4 seconds) until vasoconstriction in both the arteries and veins did not differ across retinal quadrants. The delay time (Fig. 2 , parameter 6; mean: arteries, 25.0 ± 1.9 seconds; veins, 35.2 ± 2.3 seconds) and latency (Fig. 2 , parameter 8; mean, arteries, 288.9 ± 10.1 second; veins, 306.4 ± 9.3 seconds) for recovery from vasoconstriction in both the arteries and veins did not differ across retinal quadrants. 

Discussion

Systemic hyperoxia was achieved during breathing of 100% oxygen, as indicated by the increased SaO₂ level. Because our experimental design did not require an isocapnic stimulus, ²⁸ systemic hyperoxia was accompanied by a decrease in EtCO₂, which probably represents the reduced affinity of hemoglobin for CO₂ known to occur when the oxygen partial pressure is increased. ²⁹ Both the SaO₂ and EtCO₂ quickly resumed their baseline values after discontinuation of pure oxygen breathing. The change in EtCO₂ was not accompanied by a change in the RR. Also, there were no alterations in the group-averaged PR during systemic hyperoxia. 

In the present study, the BP measurements were acquired at 25 Hz throughout testing. Lovasik et al. ⁵ were the first to use this NIBP technology to reveal the presence of choroidal blood flow regulation, showing a 43% increase in OPP without a parallel alteration in the choroidal blood flow. To our knowledge, no study has detailed the time course of OPP while breathing pure oxygen. Our present results revealed that the systemic BP and the OPP remained relatively constant throughout transient experimental hyperoxia. The IOP was measured only before testing because an earlier study had shown that pure oxygen breathing reduced the IOP by only 1 mm Hg, a change that would not have modified the OPP significantly. ⁸ Furthermore, it has recently been reported that although the IOP decreases during systemic hyperoxia, it does not alter significantly the level of OPP. ³⁰  

Earlier studies have reported that retinal blood flow is higher in the temporal versus nasal retina but is the same in the superior versus inferior retina. ^{20 31 32} Although blood flow was not measured in the present study, our results indicated that the baseline vessel diameters were larger in the temporal than in the nasal retina, but similar between the superior and inferior retina. This was the case for the comparisons of vessels across retinal quadrants, as well as of the temporal versus nasal or superior versus inferior hemiretinas. Our data therefore are in agreement with earlier blood flow studies, considering that blood flow is a function of the fourth power of the radius of a vessel ³³ (i.e., very small changes in vessel diameter are needed to cause large changes in blood flow). These regional differences in retinal blood flow have been attributed to the fact that the temporal retina is larger than its nasal counterpart and contains the highly metabolic macular area. ^{34 35}  

Our results show that systemic hyperoxia induces a vasoconstriction in the arteries and veins in each of the four retinal quadrants. On average, the veins constricted by 14.5% and the arteries by 9.2%. These results compare well with those in other studies that involved the use of pure oxygen to investigate vessel reactivity. In these studies, investigators recorded a 14.9% decrease in retinal vein diameter, ³⁶ a 12% reduction in artery and vein diameter, ¹⁰ and a 13% to 15% and 11% to 12% reduction in retinal vein and artery diameters, respectively, ¹³ with systemic hyperoxia. 

The degree of hyperoxia-induced vasoconstriction was found not to differ between veins across retinal quadrants, nor did it differ between arteries across quadrants. An additional analysis, looking at the temporal versus nasal retina and superior versus inferior retina also indicated that there were no regional differences in the degree of hyperoxia-induced arterial and venous constriction across retinal sectors. Our results differ from those published recently by Kiss et al., ¹³ who also used the RVA system to investigate the vessel reactivity to an oxygen-induced vascular stress. They reported that the retinal vasoconstriction achieved by the nasal arteries and veins was less pronounced than that of the temporal vessels. Our data are more in line with those presented earlier by Rassam et al. ²⁰ who quantified vessel changes from standard photographs of the ocular fundus. Even though this approach produced limited measurements, their observation that systemic hyperoxia induces a similar degree of constriction in temporal and nasal vessels is in agreement with our present findings. Our data further showed that there were no differences in the level of vasoconstriction across retinal quadrants or between the superior versus inferior retina. Overall, the present study indicated that in young healthy adults, there was no differential reactivity across the retina to a short-lived but potent metabolic vascular stress. 

Because all vessel measurements in this study were made within 1 DD of the ONH border, small-caliber changes in these principal vessels would cause significant blood flow changes upstream in smaller vessels. Additional studies will be undertaken to measure the caliber changes in vessels at different distances from the ONH. Focal differences in vessel reactivity to changes in arterial oxygen saturation may help explain differences in vessel reactivity reported in different studies. 

In this study, we recruited a similar number of men and women as subjects. Although we are not aware of any study showing that the retinal vessel reactivity is different in men and women, some reports have discussed gender-related differences in ocular blood flow. A few studies have reported that women have a higher pulsatile ocular blood flow than do men. ^{37 38 39} On the contrary, it has been reported that the subfoveal choroidal blood flow ⁴⁰ ; the blood flow and vascular resistance in the ophthalmic artery ⁴¹ ; and the blood flow velocity in the ophthalmic artery, central retinal artery, central retinal vein, and short posterior ciliary arteries ⁴² are not different between men and women. Our present study revealed a uniform vasoconstriction across quadrants in both men and women that is consistent with autoregulatory responses to transient hyperoxia. 

Our study showed that the rate of vasoconstriction in both arteries and veins was greater than the rate of recovery from vasoconstriction. This suggests that vasoconstriction is the result of positive innervation, whereas the recovery from vasoconstriction after oxygen inhalation is a more passive phenomenon. This interpretation is consistent with the observation that the time needed to achieve maximum vessel constriction was shorter than the time taken to recover from this level of vasoconstriction. The short delay before the initiation of retinal vasoconstriction and recovery from vasoconstriction closely matched the time course of experimentally induced systemic hyperoxia. The time-response dynamics of the arteries and veins was similar across retinal quadrants. Our measurements indicated that it took about 3 to 4 minutes for retinal vasoconstriction to develop fully in response to systemic hyperoxia, a finding that is in line with other published data. ¹³ Finally, our results also revealed that the recovery from vasoconstriction occurred over a minimum of 4 minutes, but typically followed a longer time course for a return toward baseline. 

This first ever characterization of retinal vessel reactivity and recovery from transient systemic hyperoxia revealed uniform changes in the caliber of both the arteries and veins projecting into the four principal quadrants of the fundus. Such innocuous physiological provocation may be useful clinically for evaluating the functional integrity of retinal vascular autoregulation. Furthermore, the amplitude and timing of vasomotor responses may have diagnostic power for vascular disease of the human retina. 

 Supported by Natural Sciences and Engineering Research Council of Canada OGP0116910 (JVL) and OGP0121750 (HK), Canadian Foundation for Innovation (JVL, HK), Canadian Optometric Education Trust Fund (SJ-L, HK), Fonds de la Recherche en Santé du Québec (HK).

 Submitted for publication October 13, 2004; revised December 14, 2004; accepted January 10, 2005.

 Disclosure: S. Jean-Louis, None; J.V. Lovasik, None; H. Kergoat, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Fundus image as seen on the RVA monitor with alignment cursors. Square white box: selected ROI, where vessels are tracked during eye movements; rectangular white box: the length of vessel measured at 25 Hz for optimal temporal and spatial resolution. Quadrants are delineated by dashed lines.

Figure 2.

 

Schematic representation of retinal vessel reactivity throughout the air-breathing, inhalation of pure oxygen, and recovery phases, as a function of time. The first vertical line indicates when inhalation of pure oxygen started (2 minutes) and the second, when it stopped (14 minutes). The numbers beside the schematic curve indicate the various amplitude and timing parameters of vessel dynamics measured: (1) baseline, (2) time before vasoconstriction, (3) rate of vasoconstriction, (4) latency before the plateau in vasoconstriction, (5) plateau of vasoconstriction during the pure oxygen phase, (6) time for the onset of recovery from vasoconstriction, (7) rate of the recovery from vasoconstriction, (8) latency before the plateau in recovery, and (9) plateau of the recovery from vasoconstriction.

Table 1.

 

Group-Averaged Results for the Physiological Variables Obtained throughout the Experiment

Physiological Variables	Baseline (Air-21% O2)	Hyperoxia (100% O2)	Recovery (Air-21% O2)
SaO2 (%)	97.4 ± 0.21	98.7 ± 0.15*	97.6 ± 0.20
EtCO2 (mm Hg)	39.5 ± 0.94	35.7 ± 1.10*	38.5 ± 0.76
PR (bpm)	73.2 ± 2.75	71.5 ± 2.62	74.0 ± 2.62
RR (breaths/min)	14.7 ± 0.90	13.9 ± 0.96	14.3 ± 1.02
BPsyst (mm Hg)	120.0 ± 2.93	120.8 ± 3.52	122.5 ± 3.19
BPdiast (mm Hg)	63.2 ± 2.19	63.6 ± 2.17	62.8 ± 2.64
BPmean (mm Hg)	83.7 ± 1.90	83.9 ± 2.33	84.3 ± 2.40
OPP (mm Hg)	40.6 ± 1.29	40.1 ± 1.58	40.1 ± 1.71

 Data are the mean ± SEM. n = 20.

*  Significant difference: P ≤ 0.05.

Figure 3.

 

Group-averaged (A) SaO₂, (B) EtCO₂, (C) PR, and (D) RR combined across all subjects (n = 20). The experimental phases are identified in (A). Vertical lines: the points at which pure oxygen breathing started (2 minutes) and stopped (14 minutes).

Table 2.

 

Summary Statistics Comparing the Baseline Vessel Diameters

Retinal Region	Vessel Diameter (AU)	P
IN arteries	109.0 ± 3.6	NS*
SN arteries	105.86 ± 3.2
IT arteries	124.72 ± 4.5	NS
ST arteries	120.73 ± 3.3
IN arteries	109.0 ± 3.6	NS
IT arteries	124.72 ± 4.5
SN arteries	105.86 ± 3.2	0.0004
ST arteries	120.73 ± 3.3
IN veins	110.95 ± 2.4	NS
SN veins	115.18 ± 3.8
IT veins	141.7 ± 5.9	NS
ST veins	142.7 ± 4.6
IN veins	110.95 ± 2.4	0.0008
IT veins	141.7 ± 5.9
SN veins	115.18 ± 3.8	0.0001
ST veins	142.7 ± 4.6
I arteries	116.65 ± 3.1	NS
S arteries	113.48 ± 2.5
I veins	126.91 ± 4.0	NS
S veins	129.30 ± 3.7
N arteries	107.33 ± 2.2	<0.0001
T arteries	122.44 ± 2.7
N veins	113.13 ± 2.3	<0.0001
T veins	141.99 ± 3.6

 Data are expressed in arbitrary units ± SEM for the arteries and veins across the various retinal quadrants and hemiretinas and highlight the difference between the nasal and temporal vessel size and the similarity between the superior and inferior vessel size.

*  All NS values are >0.30 except for IN versus IT arteries, where P = 0.0568.

Figure 4.

 

Percentage of change in group-averaged diameter of the arteries combined across all subjects (n = 20) in the (A) ST, (B) SN, (C) IT, and (D) IN retinal quadrants, as a function of time after the beginning of the experiment. Phases and vertical lines are as in Figure 3 .

Figure 5.

 

Percentage of change in group-averaged diameter of the veins combined across all subjects (n = 20) in the (A) ST, (B) SN, (C) IT, and (D) IN retinal quadrants, as a function of time after the beginning of the experiment. Phases and vertical lines are as in Figure 3 .

The authors thank all subjects for their participation in this demanding experiment and Marc Melillo, Normand Lalonde, Micheline Gloin, Denis Latendresse, and André Mathieu for expert computer and technical help.