Nitrous oxide (N2O) and propofol exhibit directionally opposite effects on the cerebral circulation, vasodilation and vasoconstriction, respectively. The authors investigated an interaction between the two anesthetic agents on the dynamic cerebrovascular response to step changes in end-tidal pressure of carbon dioxide (PetCO2) in humans.


Participants with no systemic diseases were allocated into two groups, each of which was anesthetized sequentially with two protocols. Patients in group 1 were anesthetized with 30% O2 + 70% N2O. A continuous intravenous infusion of propofol (7-10 mg x kg(-1) x h(-1)) was then added to the N2O. Patients in group 2 were anesthetized first with continuous infusion of propofol (10 mg x kg(-1) h(-1)), and then 30% O2 + 70% N2O was added to the propofol anesthesia. Using transcranial Doppler ultrasonography, blood flow velocity at the middle cerebral artery (FV(MCA)) was measured during a step increase (on-response) followed by a step decrease (off-response) in PetCO2, with PetCO2 ranging between approximately 28 and 50 mmHg. The dynamic FV(MCA)-PetCO2 relationship was analyzed using a mathematical model that was characterized with a pure time delay, and a time constant and a gain each for the on- or off-response.


The addition of propofol to the N2O anesthesia increased the on-response time constant (P < 0.01), whereas the addition of N2O to the propofol anesthesia increased the time constants for on- (P < 0.01) and off-responses (P < 0.05). However, the addition of either anesthetic did not affect the gains.


Propofol and N2O, when one is added to the other, produce similar dynamic FV(MCA) responses to sudden changes in PetCO2. Addition of each anesthetic slows the dynamic response and produces the response whose magnitude is proportional to the baseline FV(MCA).

PROPOFOL and nitrous oxide (N2O) are known to produce directionally opposite effects on cerebral vasculature: vasoconstriction is induced by propofol, 1,2and vasodilation is induced by N2O. 3–8Arterial carbon dioxide tension (Paco2) is a major determinant of cerebral blood flow (CBF). It has been suggested that the steady-state CBF response to changes in Paco2is preserved during anesthesia with N2O, 9,10propofol, 2,7,11or the combination of the two. 2,12However, to our knowledge, there has been little work on the effects of rapid changes in Paco2on the cerebral circulation. Therefore, in the present study we examined the actions of N2O and propofol, alone and together, on the dynamic CBF response to rapid changes in end-tidal pressure of carbon dioxide (Petco2).

Thirty-five patients (20 male, 15 female; age, 15–41 yr [mean, 29.3 yr]; height, 165.2 ± 8.2 cm [mean ± SD]; weight, 62.0 ± 9.5 kg) undergoing elective surgery took part in the study. Requirements were fully explained to all participants in writing and verbally, and each gave informed consent before participating in the study. The institutional ethics committee approved the research. Participants were not taking any medication, and none had a known history of cardiovascular, cerebrovascular, respiratory, neurologic, or endocrine disease. They fasted preoperatively for 8–10 h, during which intravenous (800–1,000 ml) fluid replacement was given. No premedication was administered.

The experiment was conducted after the induction of anesthesia and before the start of surgery. To minimize experimental time, subjects were divided into two study groups. The first group was studied during anesthesia with N2O, followed by the addition of an infusion of propofol to the N2O anesthesia. The second group was studied during anesthesia with a propofol infusion, followed by the addition of N2O.

A 2-MHz pulsed Doppler ultrasound system (PC-Dop 842, SciMed, Bristol, UK) was used to measure back-scattered Doppler signals from the right or left middle cerebral artery (MCA). The Doppler signals were transformed to the intensity-weighted mean blood flow velocity (FVMCA), which was stored on a computer for off-line analysis. FVMCAwas identified by an insonation pathway through the right or left temporal window using the standard search technique. The transcranial Doppler (TCD) probe was attached securely with a plastic headband at the position where the signal was maximized.

Subjects were placed in a supine position on an operating table with ambient temperature maintained at 25°C. After applying usual anesthetic monitors (pulse oxymetry, a lead II electrocardiograph, noninvasive brachial blood pressure, rectal temperature, and respired gas tensions [O2, CO2, and N2O]), anesthesia was induced with a bolus injection of 2 mg/kg propofol. Tracheal intubation was facilitated by a bolus injection of 0.2 mg/kg vecuronium. Rectal temperature was maintained within the preanesthetic level ± 0.2°C with a warm air blanket and a water blanket. Expired carbon dioxide tension was monitored continuously with an infrared anesthetic gas monitor (Normocap 200 oxy, Datex, Helsinki, Finland). Airflow was measured by a flow meter to define the end-expiratory phase in the off-line analysis. Carbon dioxide tension and airflow signals were stored on a computer with the TCD signals. Brachial blood pressure was measured every 1 min with an oscillometer throughout the study.

Study 1

The experimental protocol is shown schematically in figure 1. After the induction of anesthesia, 19 subjects in group 1 were anesthetized with 30% O2+ 70% N2O and mechanically ventilated at a slight hypocapnic level, Petco2of 27–29 mmHg. Neuromuscular block was obtained with a continuous intravenous infusion of vecuronium at a rate of 12 mg/h. The slight hypocapnia was attained by adjusting a tidal volume with keeping the ventilation rate at 10 breaths/min and maintained for at least 15 min for stabilization of respiratory and circulatory condition.

After a 3-min baseline period, hypercapnia was induced suddenly by adding carbon dioxide to the inspiratory gas mixture. By closely observing Petco2, the CO2flow rate was manually adjusted breath by breath such that Petco2increased to approximately 45–55 mmHg within a few breaths. The target Petco2level was maintained for the subsequent 5 min. The CO2supplementation was then terminated, and the ventilatory rate was immediately increased to 16 breaths/min, keeping tidal volume constant for about 5–7 breaths to induce a rapid decrease in Petco2. The ventilation rate was gradually decreased such that Petco2remained at a constant hypocapnic level for 3 min (recovery period). This resulted in a stepwise Petco2decrease to the baseline Petco2level in several breaths. This measurement session is referred hereafter as the N2O session.

After the N2O session was completed, an intravenous bolus of 1 mg/kg propofol was injected, followed immediately by a continuous intravenous infusion of propofol at a rate of 10 mg · kg−1· h−1. The infusion rate was adjusted such that blood pressure was maintained as close to that during the N2O session as possible. In 5 subjects, the infusion rate was reduced to 7–9 mg · kg−1· h−1. Once the infusion rate was determined, it was kept constant throughout the measurement run. Ventilation tidal volume was also adjusted such that the baseline Petco2was identical to that for the N2O session. After a 20-min accommodation period to the new anesthesia, a series of the carbon dioxide loading with measurements was performed as in the N2O session. This measurement run is referred hereafter as the propofol–N2O session.

We took an extreme care to produce a step change in Petco2that was similar in the N2O and propofol–N2O sessions for each subject. However, the manual adjustment could not achieve such precise control as to attain an identical hypercapnic level across all subjects. Therefore, we had to accept interindividual variances in the hypercapnic level ranging as wide as 45–55 mmHg.

Study 2

After the anesthetic induction and tracheal intubation was performed as mentioned above, 16 subjects in group 2 were anesthetized with a continuous intravenous infusion of propofol, 10 mg · kg−1· h−1. Neuromuscular block was obtained as in study 1. The patients were mechanically ventilated with 30% O2+ 70% N2at the slight hypocapnic level, Petco2of 27–29 mmHg. A series of measurements during stepwise CO2forcing was performed as in study 1 (propofol session). At the completion of the propofol session, the respired gas was switched to 30% O2+ 70% N2O. After a 20-min accommodation period to the new anesthesia, the stepwise carbon dioxide forcing was again performed with measurements (N2O–propofol session).

Data Analysis

Data were analyzed off-line using a mathematical package (Splus 2000, MathSoft, Cambridge, MA). End-expiratory phases were determined from the airflow signal. Petco2was obtained from the readings in the expired carbon dioxide signal at the corresponding end-expiratory phases. Mean values of FVMCAfor every heartbeat and breath-to-breath Petco2were resampled at 1 Hz by a linear interpolation to create a uniform time base. To align interindividual variances in the absolute FVMCAvalues, FVMCAwere expressed as percentages of the baseline mean FVMCAvalue of each anesthesia session for each subject.

We rejected data obtained from subjects who did not present a stable cardiovascular condition throughout the experiment to avoid possible influence of changes in blood pressure on FVMCA. The stable cardiovascular condition was defined as satisfying the following two criteria: (1) variations of the mean blood pressure remained within a range of the value at the beginning of the experiment ±10 mmHg; and (2) heart rate variations remained within a range of the value at the beginning of experiment ±15 beats/min.

Mathematical Modeling

The mathematical model used was described previously. 6,13The model is a simple extension of the steady-state FVMCA–Petco2relationship, in which the magnitude of change of FVMCAwas assumed to be proportional to the magnitude of change in Petco2. In addition, the dynamic model also considers the speed of change in FVMCAin response to change in Petco2. In the model, therefore, the rate of change of FVMCAis proportional to the deviation of FVMCAfrom the value it would obtain in the steady state. Such a dynamic model produces an exponential output (FVMCA) for a step input (Petco2). It is written in the form:


where the function u(t − Td) defines the variation of Petco2as,


where t is time (s), d/dt  denotes a derivative in terms of time, τ (s) is a time constant, G (%/mmHg) is a gain, and Td (s) is a pure time delay. FVMCA* (%) and Petco2* (mmHg) are their respective steady-state values before a step change is undertaken.

To allow for asymmetry between the FVMCAresponse to a step increase (on-response) and to a step decrease (off-response) in Petco2, separate parameter values were estimated for the on- and off-responses. This resulted in five variables for estimation: gains for the on- and off-responses (Gon, Goff), time constants for the on- and off-responses (τon, τoff), and a single time delay (Td).

Model fitting for parameter estimation was performed on the on- and off-responses separately. The on-response model was fitted to the data from duration containing the 3-min baseline period and the first 3-min hypercapnic period with a step Petco2increase in the middle. The off-response model was fitted to the data from duration containing the last 3-min hypercapnic period and the 3-min recovery period with a step Petco2decrease in the middle. The best fit models that minimized the sum of square of residuals between the data and model were computed using a grid search technique. 6,13 

Statistical Analysis

Statistical comparisons of FVMCAand the model parameters between the two sessions in each study were performed using the paired t  test when the normal distribution criterion for each variable or model parameter was fulfilled. Otherwise, the Mann–Whitney U test was used. P < 0.05 was considered significant.

All subjects exhibited normal mean blood pressure (79 ± 6 mmHg [mean ± SD] mmHg) and heart rate (64 ± 4 beats/min [mean ± SD]) at the beginning of the experiment. Four subjects in study 1 and 1 subject in study 2 did not satisfy the stable cardiovascular condition; all of them exhibited increases in blood pressure of greater than 10 mmHg during the hypercapnic period. Therefore, they were discarded from further analysis, and the data obtained from the remaining 15 subjects in each study were analyzed.

Baseline FVMCAvalues are presented in table 1. In study 1, the continuous infusion of propofol added to the N2O anesthesia decreased the baseline FVMCAby 31.6 ± 8.6% (mean ± SD;P < 0.001) compared with that for the N2O session. In study 2, the inhalation of N2O added to the propofol anesthesia did not affect the baseline FVMCA.

Figure 2shows the signals obtained from the subjects and the averages from the subjects. In each anesthesia condition, the manually adjusted carbon dioxide loading produced stepwise changes in Petco2, although the step magnitude ranged from 15 to 25 mmHg among the subjects. Petco2steps tended to be faster and greater in the on-response than in the off-response.

Figure 3shows the responses of FVMCAto step changes in Petco2and the models best fitted to the responses in a representative subject in study 1. Exponential contours in FVMCAresponse curve to step changes in Petco2were observed in the N2O and propofol–N2O sessions. Model fitting performance was good in all subjects, and the model was able to track the dynamic changes of FVMCAfor each anesthesia session.

Tables 2 and 3compare the estimated values of the model parameters between the two anesthesia sessions and between the on- and off-responses. The time delay (Td) after which the FVMCAresponses to step changes in Petco2started was similar in both sessions of each study. In study 1 (table 2), compared with the N2O anesthesia, the propofol–N2O anesthesia increased τon and did not change either Gon or Goff. In study 2 (table 3), compared with the propofol anesthesia, the N2O–propofol anesthesia increased τon and τoff and did not change either Gon or Goff. In both studies, with respect to the comparison between the on- and off-responses, τon was significantly greater than τoff, and Gon was significantly smaller than Goff in any anesthesia session, indicating slower and smaller on-responses than off-responses.

Consideration for Methodologic Limitations

Before interpreting the results of the study, the methods used should be considered. Instead of measuring the true MCA blood flow, this study used FVMCAas an index of MCA blood flow, which has been done in previous studies. 6,14–16The MCA volume blood flow is theoretically the product of FVMCAand the cross-sectional area of MCA. Therefore, FVMCAmay not necessarily represent the volume blood flow if MCA exhibits temporal changes in the cross-sectional area. However, Poulin et al.  13showed that, during CO2-loaded breathing similar to this study, changes in the MCA cross-sectional area were negligible compared with those in the FVMCA.

The manually adjusted carbon dioxide loading could not produce precise step (square-shaped) changes in Petco2with an identical magnitude across all subjects. The nonuniformity in the carbon dioxide loading pattern among subjects may affect the results. In the previous study, we examined effects of nonuniform step Petco2changes on the dynamic FVMCAresponse. 6It indicated that the effects of the nonuniformity were negligible on the results in this study, as long as it remained within the extent produced in this study.

We used only a single dose for each anesthetic because of limited experimental time. Furthermore, in 5 subjects in study 1, we reduced the infusion rate of additional propofol to 7–9 mg · kg−1· h−1instead of 10 mg · kg−1· h−1to align blood pressure as closely as possible between the two anesthesia sessions. Therefore, the results of the present study may not be interpreted as the general relationship between the two anesthetics.

Interpretation of the Results

Effects of the Combination of N2O and Propofol on the Baseline FVMCA.

The continuous infusion of propofol, when added to the N2O anesthesia, decreased the baseline FVMCAby 31.6%. It indicates that propofol at the dose used decreased the baseline MCA blood flow. It was suggested that propofol further decreased CBF as a combination of direct vasoconstriction and decreased metabolism. 1,7Although the present study neither controlled nor measured depth of anesthesia, it was likely that the infusion of propofol deepened the level of anesthesia. Therefore, both mechanisms would be responsible for the decrease in the baseline FVMCAin this study.

By contrast, inhalation of 70% N2O, when added to the propofol anesthesia, did not change the baseline FVMCA. This is similar to the finding reported by Eng et al.  2These results indicate that vasoconstriction induced by propofol is more potent than N2O-induced vasodilation at the doses used, so that effects of propofol on the cerebral vasculature would manifest.

Dynamic Cerebrovascular Responses

Compared with the steady-state condition, the dynamic response not only considers the magnitude but also the speed, the gains, the pure time delay, and the time constants, respectively. Neither propofol nor N2O affected Gon or Goff. This indicates that the magnitude of the dynamic response is proportional to the baseline FVMCAlevel, irrespective of different anesthesia conditions. It would be similar to the steady-state response of CBF to hypocapnia in either inhalational 17–20or intravenous anesthetia. 2,21However, Gon and Goff are approximately twice the magnitude of the steady-state response, which was reported to be 2.1–3.5%/mmHg. This difference may be the result of sustained hypo- or hypercapnia in the steady-state studies, in which gradual adaptation of cerebral vascular regulation toward the baseline level may have occurred. 21–23The difference in the response magnitude between the dynamic and steady-state responses may be reflected on the asymmetry between the on- and off-responses observed in this study. The on-responses were faster (τon < τoff) and smaller (Gon < Goff) than the off-responses in any anesthesia condition. It is comparable with the results of previous studies. 6,13The greater Goff than Gon would be odd because it implies that CBF decreased below the baseline level during the recovery period. It could be attributed to a transient overresponse in the cerebral arteries to a step Petco2decrease, which then fades away in several minutes. 6,13 

As to the speed of the dynamic response, Td represents the time delay after which the FVMCAresponses to step Petco2changes start. It may be attributed mainly to a circulatory delay between the lung and the brain. The Td values obtained in this study, approximately 7 s, are comparable with those reported in previous studies. 13,24The addition of propofol to the N2O anesthesia increased τon, and the addition of N2O to the propofol anesthesia increased τon and τoff. It indicates that the addition of either propofol or N2O slowed the dynamic FVMCAresponse to rapid changes in Petco2. N2O and propofol induce mutually opposing effects on cerebral vessels, vasodilation and vasoconstriction, respectively. Therefore, the mechanisms slowing the dynamic response may differ between the two anesthetic agents. In the case of the addition of propofol, we wonder if the decrease in the baseline CBF produced by propofol would have delayed changes in the vascular and perivascular factors invoked by sudden changes in Paco2, leading to the slowed dynamic response of FVMCA. Furthermore, cerebral metabolism may be reduced by the additional propofol, 7,25which would also work in a direction to slow the response.

However, it is difficult to extrapolate mechanisms responsible for the slowed dynamic response produced by the addition of N2O to the propofol anesthesia. We had anticipated that the addition of N2O, a cerebral vasodilator, would have accelerated the dynamic FVMCAresponse. We could only speculate that the additional N2O might modulate carbon dioxide-induced temporal (phase) sequences arising in the vascular and perivascular factors. 21,26–28 

It is beyond the scope of the present study to specify exact mechanisms enrolled in the interaction between N2O and propofol on the dynamic cerebrovascular response to rapid changes in Petco2. From a clinical viewpoint, however, the combined use of propofol and N2O may work beneficially for the brain protection, irrespective of the order of administration, because it would dampen the immediate changes in CBF at sudden changes in Paco2.

In conclusion, this study examined the effects of propofol and N2O on the cerebrovascular dynamic response to step changes in Petco2when added each other. Despite the directionally opposing effects of the two anesthetics on the cerebral circulation, both induced essentially similar effects. The addition of either anesthetic induced the baseline FVMCA-dependent dynamic response in magnitude and slowed the dynamic response to step changes in Petco2.

The authors thank Rie Kato, M.D., D.Phil. (Department of Anesthesiology, Chiba University Graduate School of Medicine, Chiba, Japan) for constructive criticism in preparing the manuscript.

Ederberg S, Westerlind A, Houltz E, Svensson SE, Elam M, Ricksten SE: The effects of propofol on cerebral blood flow velocity and cerebral oxygen extraction during cardiopulmonary bypass. Anesth Analg 1998; 86: 1201–6
Eng C, Lam AM, Mayberg TS, Lee C, Mathisen T: The influence of propofol with and without nitrous oxide on cerebral blood flow velocity and CO2reactivity in humans. A nesthesiology 1992; 77: 872–9
Field LM, Dorrance DE, Krzeminska EK, Barsoum LZ: Effect of nitrous oxide on cerebral blood flow in normal humans. Br J Anaesth 1993; 70: 154–9
Kirsch JR, Traystman RJ: Anesthetic action on cerebral circulation, Anesthesia: Biologic Foundations. Edited by Yaksh TL, Lynch C, Zapol WM, Biebuyck J, Saidman LJ. Philadelphia, Lippincott-Raven, 1997, pp 1193–203
Hoffman WE, Charbel FT, Edelman G, Albrecht RF, Ausman JI: Nitrous oxide added to isoflurane increases brain artery blood flow and low frequency brain electrical activity. J Neurosurg Anesthesiol 1995; 7: 82–8
Aono M, Sato J, Nishino T: Nitrous oxide increases normocapnic cerebral blood flow velocity but does not affect the dynamic cerebrovascular response to step changes in end-tidal P(CO2) in humans. Anesth Analg 1999; 89: 684–9
Matta BF, Lam AM: Nitrous oxide increases cerebral blood flow velocity during pharmacologically induced EEG silence in humans. J Neurosurg Anesthesiol 1995; 7: 89–93
Matta BF, Lam AM, Strebel S, Mayberg TS: Cerebral pressure autoregulation and carbon dioxide reactivity during propofol-induced EEG suppression. Br J Anaesth 1995; 74: 159–63
Reinstrup P, Ryding E, Algotsson L, Messeter K, Asgeirsson B, Uski T: Distribution of cerebral blood flow during anesthesia with isoflurane or halothane in humans. A nesthesiology 1995; 82: 359–66
Hormann C, Schmidauer C, Haring HP, Schalow S, Seiwald M, Benzer A: Hyperventilation reverses the nitrous oxide-induced increase in cerebral blood flow velocity in human volunteers. Br J Anaesth 1995; 74: 616–8
Cenic A, Craen RA, Howard-Lech VL, Lee TY, Gelb AW: Cerebral blood volume and blood flow at varying arterial carbon dioxide tension levels in rabbits during propofol anesthesia. Anesth Analg 2000; 90: 1376–83
Fox J, Gelb AW, Enns J, Murkin JM, Farrar JK, Manninen PH: The responsiveness of cerebral blood flow to changes in arterial carbon dioxide is maintained during propofol-nitrous oxide anesthesia in humans. A nesthesiology 1992; 77: 453–6
Poulin MJ, Liang PJ, Robbins PA: Dynamics of the cerebral blood flow response to step changes in end-tidal PCO2and PO2in humans. J Appl Physiol 1996; 81: 1084–95
Aaslid R: Cerebral hemodynamics, Transcranial Doppler. Edited by Aaslid R, Newell DW. New York, Raven Press, 1992, pp 49–55
Aaslid R, Lindegaard KF, Sorteberg W, Nornes H: Cerebral autoregulation dynamics in humans. Stroke 1989; 20: 45–52
Dawson SL, Panerai RB, Potter JF: Critical closing pressure explains cerebral hemodynamics during the Valsalva maneuver. J Appl Physiol 1999; 86: 675–80
Scheller MS, Todd MM, Drummond JC: Isoflurane, halothane, and regional cerebral blood flow at various levels of Pa co2in rabbits. A nesthesiology 1986; 64: 598–604
Drummond JC, Todd MM: The response of the feline cerebral circulation to Pa co2during anesthesia with isoflurane and halothane and during sedation with nitrous oxide. A nesthesiology 1985; 62: 268–73
Strebel S, Kaufmann M, Baggi M, Zenklusen U: Cerebrovascular carbon dioxide reactivity during exposure to equipotent isoflurane and isoflurane in nitrous oxide anaesthesia. Br J Anaesth 1993; 71: 272–6
McPherson RW, Briar JE, Traystman RJ: Cerebrovascular responsiveness to carbon dioxide in dogs with 1.4% and 2.8% isoflurane. A nesthesiology 1989; 70: 843–50
Brian JE Jr: Carbon dioxide and the cerebral circulation. A nesthesiology 1998; 88: 1365–86
Ellingsen I, Hauge A, Nicolaysen G, Thoresen M, Walloe L: Changes in human cerebral blood flow due to step changes in PAO2and PACO2. Acta Physiol Scand 1987; 129: 157–63
Poulin MJ, Liang PJ, Robbins PA: Fast and slow components of cerebral blood flow response to step decreases in end-tidal PCO2in humans. J Appl Physiol 1998; 85: 388–97
Poulin MJ, Cunningham DA, Paterson DH: Dynamics of the ventilatory response to step changes in end-tidal PCO2in older humans. Can J Appl Physiol 1997; 22: 368–83
Doyle PW, Matta BF: Burst suppression or isoelectric encephalogram for cerebral protection: Evidence from metabolic suppression studies. Br J Anaesth 1999; 83: 580–4
Faraci FM, Breese KR, Heistad DD: Cerebral vasodilation during hypercapnia. Role of glibenclamide-sensitive potassium channels and nitric oxide. Stroke 1994; 25: 1679–83
Dreier JP, Korner K, Gorner A, Lindauer U, Weih M, Villringer A, Dirnagl U: Nitric oxide modulates the CBF response to increased extracellular potassium. J Cereb Blood Flow Metab 1995; 15: 914–9
Hudetz AG, Smith JJ, Lee JG, Bosnjak ZJ, Kampine JP: Modification of cerebral laser-Doppler flow oscillations by halothane, PCO2, and nitric oxide synthase blockade. Am J Physiol 1995; 269: H114–20