Comparison of Different Hyperoxic Paradigms to Induce Vasoconstriction: Implications for the Investigation of Retinal Vascular Reactivity

Abstract

purpose. To compare the impact of three different techniques used to induce hyperoxia on end-tidal CO₂ (P_ETCO₂). The relationship between change in P_ETCO₂ and retinal hemodynamics was also assessed to determine the clinical research relevance of this parameter.

methods. The sample comprised 10 normal subjects (mean age, 25 years; range, 21–49 years). Each subject attended for three sessions. At each session, subjects initially breathed air followed by O₂ only; O₂ plus CO₂, using a nonrebreathing circuit (with CO₂ flow continually adjusted to negate drift of P_ETCO₂); or air followed by O₂, using a sequential rebreathing circuit. In addition, using a separate sample of eight normal subjects (mean age, 26.5 years; range, 24–36 years), a methodology that initially raised P_ETCO₂ and then returned to homeostatic levels was used to determine the impact, if any, of perturbation of P_ETCO₂ on retinal hemodynamics.

results. The difference in group mean P_ETCO₂ between baseline and elevated O₂ breathing was significantly different (t-test, P = 0.0038) for O₂-only administration with a nonrebreathing system. The sequential rebreathing technique resulted in a significantly lower difference (i.e., before and during hyperoxia) of individual P_ETCO₂ (t-test, P = 0.0317). The P_ETCO₂ perturbation resulted in a significant (P < 0.005) change of retinal arteriolar diameter, blood velocity, and blood flow.

conclusions. The sequential rebreathing technique resulted in a reduced variability of P_ETCO₂. A relatively modest change in P_ETCO₂ resulted in a significant change in retinal hemodynamics. Rigorous control of P_ETCO₂ is necesssary to attain standardized, reproducible hyperoxic stimuli for the assessment of retinal vascular reactivity.

Administration of oxygen (O₂) has been used as a stimulus to provoke retinal vascular reactivity. Vasoconstriction of retinal vessels ^{1 2} and the resulting reduction of retinal blood flow ^{3 4 5 6 7 8 9 10 11 12 13 14 15 16} has been demonstrated with a variety of measurement techniques, including laser Doppler and blue-field entoptic phenomena. Retinal blood flow varies inversely with the partial pressure of arterial oxygen (Po ₂) to maintain retinal oxygenation at a relatively constant level ¹⁶ ; however, retinal blood flow also varies directly with the partial pressure of arterial carbon dioxide (Pco ₂). ¹² The change of end-tidal CO₂ concentration (P_ETCO₂; the maximum concentration of CO₂ during each expiration) reflects the change in arterial Pco ₂. ¹⁷ Indeed, CO₂ is thought to represent a more potent vasoactive agent than O₂. ⁷ Change of retinal perfusion, measured using laser Doppler blood flow techniques, induced by perturbation of O₂ or CO₂, can be used to provide a measure of the magnitude of retinal vascular reactivity. 

All previously published retinal vascular reactivity studies have used gas delivery systems that comprise a reservoir bag and one-way valves to essentially negate the mixing of inspired and expired gases (i.e., a nonrebreathing system). Hyperoxia typically stimulates hyperventilation (faster or deeper respiration); however, this results in an uncontrolled and variable reduction of Pco ₂. ^{18 19} Any reduction of Pco ₂ produces an exaggerated vasoconstrictive effect on retinal vasculature. The vasoconstrictive effect previously attributed to O₂ when administered by nonrebreathing circuits probably represents the combined effect of elevated Po ₂ and reduced Pco ₂. Harris et al., ^{16 20 21} Roff et al., ¹⁴ and Chung et al. ¹⁵ have recognized this potential artifact and have attempted to correct for the reduction in Pco ₂ during hyperoxia by adding CO₂ to the inspired gases of the nonrebreathing system. ^{14 15 16 20 21} The maintenance of homeostatic Pco ₂ is termed isocapnia. Another method to prevent reduction of Pco ₂ involves the use of a sequential rebreathing circuit that provides a feedback loop to compensate for any hyperventilation induced reduction in Pco ₂. ²² This system has the advantage that it passively adjusts the inspired CO₂ to the minute ventilation to stabilize Pco ₂. 

The magnitude of change and the variability of P_ETCO₂ should be quantified and compared across the various techniques used to induce hyperoxia. Three different techniques were compared: administration of O₂ only with a nonrebreathing system; O₂ with added CO₂ with a nonrebreathing system (with CO₂ flow continually adjusted to negate “drift” of P_ETCO₂); and O₂ using a sequential rebreathing system (with O₂ flow set equal to the subjects’ minute ventilation). In addition, the relationship between change in P_ETCO₂ and retinal blood flow was assessed to determine the clinical research relevance of this parameter. 

Materials and Methods

Sample

The study received approval by the University of Waterloo Office of Research Ethics. Informed consent was obtained form each subject after explanation of the nature and possible consequences of the study, according to the tenets of the Declaration of Helsinki. The sample comprised four men and six women of average age 25 years (range, 21–49 years). To determine the relationship between P_ETCO₂ and retinal blood flow, a second sample of six men and two women of average age 26.5 years (range, 24–36 years) was subsequently recruited. Subjects with any cardiovascular or respiratory disorders were excluded from the study. 

Procedures

Each subject attended three sessions of approximately 30 minutes each. The group mean number of days between each of the three sessions was 11 days. At each session, subjects initially breathed air followed by O₂ only, or O₂ plus CO₂ using a nonrebreathing system (Fig. 1) , or compressed air followed by O₂ using a sequential rebreathing system (Fig. 2) . Each gas condition (air or O₂) was administered for 15 minutes. The nonrebreathing system comprised a silicone mouthpiece and two low-resistance one-way valves connected to the gas supply by a reservoir bag. The sequential rebreathing system comprised fresh gas and rebreathed gas reservoirs that were interconnected by two one-way valves and a single peep valve. It was assembled by adding a gas reservoir to the expiratory port of a commercial three-valve oxygen-delivery system (Hi-Ox^SR; ViasysHealthcare, Yorba Linda, CA). For the purpose of this study, a silicone mouthpiece was attached to the sequential rebreathing system, which in turn was connected to the gas supply. For both systems, flow from the gas tanks was controlled using standard rotometers as flowmeters. O₂ and CO₂ were mixed in a baffled container before being administered to the subject through the nonrebreathing circuit. 

Each subject was seated for 5 minutes before commencing the study. In every situation, an initial air-breathing period was used to allow stabilization of baseline breathing parameters (e.g., respiration rate). For the O₂ plus CO₂ with a nonrebreathing system (Fig. 1) , CO₂ flow was continually adjusted to negate drift of P_ETCO₂. For the O₂ using a sequential rebreathing system (Fig. 2) , O₂ flow was set equal to the subjects’ minute ventilation (determined while breathing air). 

To determine the impact, if any, of perturbation of P_ETCO₂ on retinal blood flow, the sequential rebreathing system was used to manipulate P_ETCO₂ while retinal blood flow was quantified with a laser blood flowmeter (CLBF 100; Canon, Tokyo, Japan). A steady state perturbation of P_ETCO₂ was produced because the time between P_ETCO₂ fluctuation and its impact on retinal hemodynamics is unknown and because the CLBF does not provide a continuous measurement of retinal blood flow. A methodology that initially raised P_ETCO₂ and then returned to homeostatic levels was used. After stabilization of cardiovascular and respiratory parameters, air flow delivered to the subject through the sequential rebreathing system was reduced to elevate P_ETCO₂ by approximately 5 mmHg (i.e., the volunteers were compelled to rebreath). At this point, a minimum of 10 CLBF readings was acquired. Air flow was subsequently returned to baseline levels, and a further six CLBF readings were acquired. 

Data Acquisition and Analysis

Tidal gas concentrations were continuously sampled from the mouthpiece using a rapid-response critical care gas analyzer (Cardiocap 5; Datex-Ohmeda, Louisville, CO). In addition, hemoglobin oxygen saturation through pulse oximetry and respiratory and pulse rate were also continuously recorded. All data outputs were downloaded to an electronic data acquisition system (S5 Collect; Datex-Ohmeda). Data were analyzed using box plots that depicted the median, top 25th and bottom 75th percentiles (SD) and outliers of inspired- and end-tidal gas concentrations. Data points lying outside the top 25th or lower 75th percentiles were excluded from the analysis, because all these values were found to be erroneous—that is, these points resulted from inappropriate interpretation of tidal waveforms by the gas monitor. 

Group mean inspired and expired O₂ (FiO₂ and P_ETO₂, respectively), inspired and expired CO₂ (FiCO₂ and P_ETCO₂, respectively), and respiration rates (RR) as a function of delivery system are shown in Table 1 . The group mean FiO₂ was greater than 90% for all techniques. 

Results

The group mean difference of P_ETCO₂ (and SD) between baseline and elevated O₂ breathing for each of the three techniques is shown in Table 2 . Figure 3 shows P_ETCO₂ for each individual at baseline (i.e., air) and during hyperoxia for each of the three techniques. 

For O₂ breathing only, with a nonrebreathing system, group mean P_ETCO₂ reduced from 5.32% ± 0.18% at baseline to 5.06% ± 0.18% during hyperoxia (Fig. 3A) . For O₂ with added CO₂ using a nonrebreathing system, group mean P_ETCO₂ was 5.34% ± 0.16% at baseline and 5.29% ± 0.17% during hyperoxia (Fig. 3B) . For a sequential rebreathing system, group mean P_ETCO₂ was 5.13% ± 0.15% at baseline and 5.07% ± 0.13% during hyperoxia (Fig. 3C) . The difference in group mean P_ETCO₂ between baseline and hyperoxia was significantly different for O₂ only and a nonrebreathing system (t-test, P = 0.0038) but not for the other two techniques. 

The sequential rebreathing technique resulted in a significantly lower difference (i.e., a smaller difference between baseline and during hyperoxia) of individual P_ETCO₂ (as reflected in the standard deviations, Table 2 ) than either of the other two techniques (t-test, P = 0.0008 and P = 0.0317 for O₂ only and O₂ with added CO₂, respectively). 

The group mean difference between the elevated (group mean 5.69% ± 0.44%) and homeostatic (group mean 5.03%± 0.59%) P_ETCO₂ conditions was 0.66% ± 0.21%—that is, a 5.00 ± 1.58 mm Hg change. This perturbation of P_ETCO₂ resulted in a significant reduction (i.e., in response to a lowering of P_ETCO₂) in retinal arteriolar diameter, blood velocity and blood flow of 6.18 μm (P < 0.0050), 6.68 mm/sec (P = 0.0005), and 3.04 uL/min (P < 0.0005), respectively (Fig. 4) . 

Discussion

Elevating Po ₂ by simply raising the FiO₂ without taking any measures to control Pco ₂ resulted in a significant reduction from baseline in mean P_ETCO₂. This group mean reduction was ameliorated with coadministration of O₂ and CO₂ or the sequential rebreathing technique. Of the latter two methods, the sequential rebreathing technique had a significantly smaller variability in individual P_ETCO₂ measurements. 

Nonrebreathing techniques involve the administration of gas using a reservoir bag through a one-way “demand” valve—that is, a valve that opens at the onset of each inspiration. Expired gas leaves the system through a second one-way valve. Riva et al. ⁴ were the first to describe retinal vascular effects using 100% O₂ and laser Doppler velocimetry. In terms of vision-science–based studies, Harris et al. ^{16 20 21} were the first to consider the potential confounding factor of change in P_ETCO₂ during hyperoxia by coadministering O₂ and CO₂ with a nonrebreathing system. Roff et al. ¹⁴ and Chung et al. ¹⁵ also used this technique in their ocular blood flow studies. These studies did not report the magnitude of individual variability of P_ETCO₂. This study demonstrates that the maintenance of homeostatic P_ETCO₂ levels using nonrebreathing techniques apply to groups as a whole, but are less reliable for individual subjects. 

Furthermore, the link between change in P_ETCO₂ and retinal hemodynamics (namely retinal arteriolar diameter, blood velocity, and blood flow) has been demonstrated. The group mean change of P_ETCO₂ of 0.66% produced a group mean 27% change in retinal blood flow (i.e., a 2-mmHg drift in P_ETCO₂) that invariably occurs with nonrebreathing techniques, resulted in a 10% to 12% artifactual change in blood flow. This emphasizes the importance of using breathing circuits that facilitate control of P_ETCO₂ measurements during administration of elevated O₂. 

An alternative method of preventing reduction of P_ETCO₂ with increases in ventilation (as induced by exposing subjects to O₂) is to increase the dead space of the circuit so that rebreathing occurs. Adding circuit dead space may not limit the reduction of P_ETCO₂ with hyperventilation, as spontaneously breathing subjects (as opposed to those being mechanically ventilated) will overcome the effects of rebreathing by increasing respiratory volume. The sequential rebreathing method developed by Sommer et al. ²² and Banzett et al. ²³ used in this study, passively matches the inhaled CO₂ to increases in minute ventilation thereby preventing the expected reduction in Pco ₂. The sequential rebreathing technique is effective irrespective of the pattern of breathing. Compared to a nonrebreathing system and adding CO₂ to inspired gas, this system has the advantage of avoiding the risk of raising P_ETCO₂ and consequently eliciting subject discomfort. The flow of fresh gas (air or O₂, not containing CO₂) is set to just match the patient’s minute ventilation during resting conditions while breathing air. This flow is identified by observing that the fresh gas reservoir just collapses at the end of each breath. The gas exhaled by the subject is “stored” in a second reservoir bag and is available for rebreathing on the next inspiration. Because the flow of the fresh gas is fixed, any increases in ventilation will proportionally increase the volume of previously exhaled gas that is rebreathed. Only the fresh gas (O₂ in this case) contributes to the elimination of CO₂. As the flow of fresh gas is equal between air and O₂, the rate of elimination of CO₂ is constant across the two conditions. The constant P_ETCO₂ also maintains the subject’s breathing comfort. Figure 5 shows a typical example of the change in retinal blood flow (i.e., the magnitude of retinal vascular reactivity) as measured by the laser blood flowmeter (Canon), induced by O₂ delivery through the sequential rebreathing circuit. The flowmeter has been described in detail elsewhere. ²⁴  

Impaired vascular reactivity has been implicated in the pathogenesis of diabetic retinopathy and glaucoma. ^{6 16 20 21 25 26 27 28} Hyperoxia has been used frequently to provoke vascular reactivity. A recent study has shown that endothelin-1 plays a major role in hyperoxia-induced vasoconstriction in humans. ²⁹ Other factors that may be responsible for regulating vascular tone in the retina include the endothelins, ^{30 31 32 33 34 35 36} nitric oxide, ³⁷ prostacyclins, ³⁸ and angiotensin. ³⁹ Regulation of blood flow is necessary to maintain structure and function of tissue. This is achieved using systemic controls (nervous influences) and/or local factors (metabolic or myogenic). The vascular response to metabolic factors, sometimes loosely described as autoregulation, stabilizes local blood flow by making the necessary adjustments to ensure a steady supply of metabolites to tissues by altering blood flow as appropriate. 

In this study, the group mean reduction of P_ETCO₂ was ameliorated by the coadministration of O₂ and a small amount of CO₂ using the nonrebreathing system. CO₂ is a potent vasoactive agent. Retinal blood flow varies directly with the arterial Pco ₂, as reflected in the P_ETCO₂. However, with this method, there were significant intrasubject variations of P_ETCO₂. In addition, individuals responded in different ways, and thus it would be more difficult to standardize hyperoxia using the coadministration of O₂ and CO₂. In contrast, the sequential rebreathing technique was shown to allow the administration of elevated O₂ levels without a reduction in P_ETCO₂ and, importantly, reduced variability of P_ETCO₂ measurements when compared with the nonrebreathing techniques. Steady state manipulation of P_ETCO₂ unequivocally demonstrated that relatively modest change in P_ETCO₂ resulted in significant change of retinal hemodynamics. 

Conclusions

Published retinal vascular reactivity studies have used a nonstandardized hyperoxic stimulus without control of P_ETCO₂, making the results difficult to interpret. Compounded vasoconstrictive effects occur when no control for the reduction of systemic Pco ₂ levels is made. In addition, blood flow measurements taken under these conditions may exhibit exaggerated variability due to continuous alterations in Pco ₂. Rigorous control of P_ETCO₂ using a modified commercially available sequential rebreathing circuit will allow the establishment of a standardized, reproducible hyperoxic stimulus for the investigation of vascular reactivity of the human retina. 

 Supported by an Operating Grant and New Investigator Award (CH) from the Canadian Institutes of Health Research, and by a Premier’s Research Excellence Award (CH).

 Submitted for publication November 10, 2003; revised April 7, 2004; accepted May 3, 2004.

 Disclosure: E.D. Gilmore, None; C. Hudson, None; S.T. Venkataraman, None; D. Preiss, None; J. Fisher, 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.

 

Schematic diagram showing the components of the nonrebreathing system.

Figure 2.

 

Schematic diagram showing the components of the sequential rebreathing system.

Table 1.

 

Group Mean and SD of Inspired O₂, Expired O₂, Inspired CO₂, End-tidal CO₂, and Heart Rate as a Function of Technique

	Inspired O2 (%)			Expired O2 (%)			Inspired CO2 (%)			End-tidal CO2 (%)			Heart Rate (beats/min)
		Air	O2	Air	O2	Air	O2	Air	O2	Air	O2
Pure O2	20.33	91.51	14.73	81.92	0.07	0.07	5.42	5.17	72.96	66.96
SD	0.05	2.27	0.56	7.06	0.07	0.08	0.19	0.19	4.00	4.55
O2+CO2	20.28	92.06	14.82	82.19	0.08	0.97	5.34	5.29	77.03	71.39
SD	0.14	1.01	0.68	3.16	0.12	0.28	0.16	0.17	5.77	7.03
O2 (SRB)	19.91	93.50	15.30	87.21	0.42	0.74	5.13	5.07	76.91	70.54
SD	0.46	2.48	0.63	5.82	0.33	0.39	0.16	0.13	4.51	5.58

 SRB, sequential rebreathing.

Table 2.

 

Group Mean Difference in P_ET CO₂ (%) between Baseline and Oxygen Breathing Using three Different Techniques

	Pure O2	O2+CO2	SRB
Group mean difference (O2-air)	−0.21	−0.06	−0.06
Group mean SD	0.24	0.17	0.08

 SRB, sequential rebreathing.

Figure 3.

 

Change in end-tidal CO₂ concentration for each individual using (A) pure O₂ delivered by a nonrebreathing system, (B) O₂ with added CO₂ delivered through a nonrebreathing system, and (C) O₂ delivered through a sequential rebreathing (SRB) system.

Figure 4.

 

Change in retinal (A) arteriolar diameter, (B) blood velocity, and (C) blood flow induced by a change in end-tidal CO₂ concentration. The group mean difference between the elevated (group mean 5.69% ± 0.44%) and homeostatic (group mean 5.03% ± 0.59%) P_ETCO₂ conditions was 0.66% ± 0.21%. Error bars, SD.

Figure 5.

 

Change in retinal blood flow (as measured by a laser blood flowmeter) induced by O₂ delivered using the sequential rebreathing circuit. Oxygen was administered at 5 minutes. The data have been fit with a sigmoid type function; equation y = {[(v1 − v2)/[1 + euler∧[x − (v3/v4)]]] + v2} where v1 and v2 are the upper and lower asymptotes, v3 and v4 localize the midpoint of the descending part of the function on the x-axis (R = 0.85). P_ETCO₂ air = 5.00%; P_ETCO₂ O₂ = 4.88%. FiO₂ air = 20.31%; FiO₂ O₂ = 93.88%.