Ketamine Preserves and Propofol Potentiates Hypoxic Pulmonary Vasoconstriction Compared with the Conscious State in Chronically Instrumented Dogs 

All surgical procedures and experimental protocols were approved by an institutional animal care and use committee.

Twenty conditioned, male mongrel dogs (27 ± 2 kg) were premedicated with morphine sulfate (10 mg intramuscularly) and anesthetized with intravenous pentobarbital sodium (20 mg/kg) and fentanyl citrate (15 μg/kg). After tracheal intubation, the lungs were mechanically ventilated. Anesthesia was maintained with halothane (about 1.2% end-tidal). Using sterile surgical technique, a left lateral thoracotomy was performed via  the fifth intercostal space. The pericardium was incised ventral to the phrenic nerve. Heparin-filled Tygon catheters (1.02 mm internal diameter; Norton, Akron, OH) were inserted into the descending thoracic aorta, left and right atria, and main pulmonary artery and were secured with purse-string sutures. After careful dissection and isolation, a hydraulic occluder (18 mm internal diameter; In Vivo Metric, Healdsburg, CA) was positioned loosely around the right main pulmonary artery, and an electromagnetic flow probe (10 mm internal diameter; Zepeda, Seattle, WA) was placed around the left main pulmonary artery. After loose apposition of the pericardial edges, the free ends of the catheters, occluder, and flow probe were threaded through the chest wall and were tunneled subcutaneously to a final position between the scapulae. A chest tube placed in the left thorax before closure was removed 1 day after surgery. Morphine sulfate (10 mg) was administered intramuscularly after surgery for pain as required. Ampicillin (1 g), cefazolin (1 g), and gentamicin (80 mg) were administered intravenously during surgery and on a daily basis for 10 days after surgery. The dogs were allowed to recover for at least 2 weeks before experimentation.

Vascular pressures were measured by attaching the fluid-filled catheters to strain-gauge manometers (Isotec; Quest Medical, Allen, TX) and were referenced to atmospheric pressure with the transducers positioned at midchest at the level of the spine with the dogs in the right lateral position. Heart rate was calculated from the phasic systemic arterial pressure trace. Left pulmonary blood flow (LQ) was measured by connecting the flow probe to an electromagnetic flowmeter (SWF-5RD; Zepeda). The flow probe was calibrated in vivo  on a weekly basis via  the thermal dilution technique. This was achieved by acutely inserting a 7-French balloon-tipped thermal dilution catheter into the pulmonary artery through a percutaneous jugular puncture after topical lidocaine anesthesia. The catheter was positioned 2–3 cm beyond the pulmonic valve. The implanted perivascular hydraulic occluder was then inflated to occlude the right main pulmonary artery completely, which directed total pulmonary blood flow through the left pulmonary artery (and flow probe). LQ was then measured by thermal dilution (HEMOPRO²; Spectramed, Oxnard, CA) with multiple 10-ml sterile injectates of 5% dextrose in water. Values for LQ were referenced to body weight (ml · min⁻¹· kg⁻¹). The aortic and pulmonary artery catheters were used to obtain blood samples to measure systemic arterial and mixed venous blood gases, respectively. Systemic arterial and mixed venous p  H, carbon dioxide tension (P^CO2), and oxygen tension (P^O2) were measured with an ABL-600 (Radiometer, Copenhagen, Denmark). Oxyhemoglobin saturation (S^O2) was measured with a Hemoximeter OSM-3 (Radiometer).

All experiments were performed with each healthy, chronically instrumented dog lying on its right side in a quiet laboratory environment. Conscious dogs were not sedated, and anesthetized dogs were not paralyzed. LPQ plots were used to assess the effects of the various physiologic and pharmacologic interventions on the pulmonary circulation. LPQ plots were constructed by continuously measuring the pulmonary vascular pressure gradient (pulmonary arterial pressure [PAP]− left atrial pressure [LAP]) and LQ during gradual (approximately 1 min) inflation of the hydraulic occluder implanted around the right main pulmonary artery. This technique to measure the LPQ relationship is highly reproducible and has little or no effect on systemic hemodynamics, blood gases, or the zonal condition of the lung. 9 

A baseline LPQ plot was obtained during normoxia in each conscious dog (n = 7). A conical face mask was then placed over the dog's snout. Room air was administered via  the mask using a semiclosed circulation system, and after 15 min a normoxia LPQ plot with face mask was obtained. The delivered room air was then blended with gases from sources consisting of 100% nitrogen, oxygen, or carbon dioxide. The gas flows were titrated to the fraction of inspired oxygen tension (FI^O2= 13.5 ± 0.2) that resulted in a gradual decrease in systemic arterial P^O2to about 60 mmHg. After reaching a new steady state (about 15 min), a hypoxic LPQ plot was generated. The gas flows were then titrated to decrease systemic arterial P^O2to ∼50 mmHg (FI^O2= 11.2 ± 0.2), and a second hypoxic LPQ plot was generated.

To determine the conscious response to hypoxia, LPQ plots were generated in each dog (n = 6) as described in protocol 1. On a different day, this protocol was repeated in the same six dogs during ketamine anesthesia. A baseline LPQ plot was generated during normoxia in the conscious state. Anesthesia was then induced and maintained with the intravenous administration of ketamine (5 mg/kg bolus and 1 mg · kg⁻¹· min⁻¹infusion). The trachea was intubated and ventilation was controlled with a respirator with zero end-expiratory pressure. Tidal volume was fixed at 15 ml/kg. Baseline systemic arterial blood gases were matched to values measured in the conscious state by adjusting the respiratory rate to between 10 and 20 breaths per minute and by supplying supplemental O². Inspiratory and end-tidal O²and CO²tension were monitored continuously at the adapter end of the endotracheal tube (Solar 7000; Marquette Electronics, Milwaukee, WI). Body temperature was monitored and maintained between 38 and 39°C with a warming blanket during the experiment. Twenty minutes after induction of ketamine anesthesia, an LPQ plot was generated. The hypoxic gas mixture was then administered as described in protocol 1, and LPQ plots were generated with systemic arterial P^O2about 60 mmHg and about 50 mmHg.

To determine the conscious response to hypoxia, LPQ plots were obtained in each dog (n = 7) as described in protocol 1. On a different day, a baseline LPQ plot was generated during normoxia in the conscious state. After administration of propofol (5 mg/kg, intravenously), the trachea was intubated and ventilation was controlled as described in protocol 2. Intravenous propofol was infused continuously at the rate of 0.5 mg · kg⁻¹· min⁻¹immediately after intubation. Twenty minutes after induction, an LPQ plot was obtained during propofol anesthesia. The hypoxic gas mixture was then administered via  the endotracheal tube. Once a steady state was obtained with systemic arterial P^O2∼60 mmHg and ∼50 mmHg, hypoxic LPQ plots were generated. The infusion rate for propofol was then increased to 1 mg · kg⁻¹· min⁻¹, and after 15 min an LPQ plot was obtained with systemic arterial P^O2about 50 mmHg. On a different day, we tested the effect of 10% intralipid vehicle (Cleveland Clinic Pharmacy, Cleveland, OH) on the HPV response in three of the dogs used in protocol 3. After obtaining a baseline LPQ plot, intralipid was administrated intravenously at the dose corresponding to that used in the propofol protocol (5 mg/kg bolus and 0.5 mg · kg⁻¹· min⁻¹infusion). After 20 min, LPQ plots were obtained while the conscious dogs were breathing room air, room air via  mask, and during hypoxia (systemic arterial P^O2about 50 mmHg).

Cyclooxygenase inhibition was achieved via  the intravenous administration of indomethacin 5 mg/kg (Sigma Chemical Company, St. Louis, MO). This dose inhibits prostaglandin synthesis and abolishes the pulmonary pressor response to arachidonic acid. 4Indomethacin was dissolved in sterile water with 114 g sodium carbonate. To determine the intact (no drug) response to hypoxia (systemic arterial P^O2about 50 mmHg) in the conscious state, LPQ plots were generated in each dog (n = 7) as described in protocol 1. On a different day, LPQ plots were generated in the same conscious dogs at baseline, 45 min after administration of indomethacin, while breathing room air via  mask, and during hypoxia with systemic arterial P^O2approximately 50 mmHg. This protocol was repeated on different days in the same 7 dogs during propofol anesthesia with or without pretreatment with indomethacin. Anesthesia with propofol (5 mg/kg plus 0.5 mg · kg⁻¹· min⁻¹, intravenously) was induced and maintained as described in protocol 3. LPQ plots were generated at conscious baseline, 45 min after administration of indomethacin, 20 min after induction of propofol anesthesia, and during hypoxia with systemic arterial P^O2at about 50 mmHg.

K^+ATPchannel inhibition was achieved via  the intravenous administration of glibenclamide (3 mg/kg, Sigma). This dose inhibits the pulmonary vasodilator response to a selective activator of K^+ATPchannels, lemakalim, in dogs. 8Glibenclamide reduced blood glucose by 20 mg/dl 30 to 45 min after administration, but the dogs recovered promptly from the experiments and showed no adverse effects from K^+ATPchannel inhibition. Glibenclamide was dissolved in 0.1 N NaOH and diluted in 5% dextrose. To determine the intact (no drug) response to hypoxia (systemic arterial P^O2approximately 50 mmHg) in the conscious state, LPQ plots were generated in each dog (n = 6) as described in protocol 1. On a different day, LPQ plots were generated in the same conscious dogs at baseline, while breathing room air via  mask, 15 min after administration of glibenclamide, and during hypoxia with systemic arterial P^O2about 50 mmHg. This protocol was repeated on different days in the same six dogs during propofol anesthesia with or without pretreatment with glibenclamide. Anesthesia with propofol (5 mg/kg plus 0.5 mg · kg⁻¹· min⁻¹, intravenously) was induced and maintained as described in protocol 3. LPQ plots were generated at conscious baseline, 20 min after induction of propofol anesthesia, 15 min after administration of glibenclamide, and during hypoxia with systemic arterial P^O2approximately 50 mmHg.

Phasic and mean vascular pressures and LQ were displayed continuously on an eight-channel strip-chart recorder (model 2800; Gould, Eastlake, OH). Mean values for LQ and vascular pressures were obtained using passive electronic filters with a 2-s time constant and were measured at end-expiration. All vascular pressures were referenced to atmospheric pressure before and after each LPQ plot. The analog pressure and LQ signals were digitally converted and multiplexed (PCM-8; Medical Systems, Greenvale, NY) and stored on videotape (videocassette recorder AG-1260; Panasonic, Secaucus, NJ) for later playback and analysis using software by Labview (Austin, TX). The LPQ relationship was measured continuously over the empirically measured range of LQ in each individual experiment. In all protocols, the LPQ relationship was linear by inspection over the empirically measured range of LQ. Therefore, linear regression analysis was used to calculate the slope and intercept for PAP − LAP (or PAP − 0 if LAP was ≤ 0 mmHg) as a function of LQ in each individual experiment. The correlation coefficient for the LPQ relationship in each protocol averaged ≥ 0.98. The composite LPQ plots summarized in the figures were generated by using the regression parameters from each individual continuously measured LPQ plot to calculate PAP − LAP at 10 ml · min⁻¹· kg⁻¹intervals of LQ over the empirically measured range of LQ. The minimum and maximum values of LQ in each composite LPQ plot represent the average minimum and maximum values of LQ for the dogs studied in that protocol. Multivariate analysis of variance (ANOVA) in the form of Hotelling's T²was used to assess the effects of the various interventions on the LPQ relationship. One-way ANOVA followed by Student t  test for paired comparisons was used to assess the pulmonary vascular effects of hypoxia within each group. Two-way ANOVA followed by Student t  test for paired comparisons was used to assess the effects of the anesthetics on the magnitude of HPV. All values are presented as means ± SEM.

Figure 1depicts mean values of PAP, LAP, and LQ, and the associated LPQ plots, in an individual conscious dog during normoxia and hypoxia. The LPQ plots were generated by gradual occlusion of the right pulmonary artery, which directs pulmonary blood flow through the left pulmonary artery and flow probe. In conscious dogs, breathing room air via  face mask had no effect on the baseline normoxic LPQ relationship (fig. 2A). During hypoxia (P^O2approximately 60 mmHg) there was a leftward shift (P < 0.01) in the LPQ relationship, indicating HPV (fig. 2A). Decreasing P^O2to 50 mmHg caused a further leftward shift (P < 0.01) in the LPQ relationship (fig. 2A). The magnitude of the HPV response (i.e. , the increase in PAP − LAP from normoxia to hypoxia at each common value of LQ) was stimulus-dependent (P < 0.01) and increased (P < 0.01) as LQ increased at both levels of P^O2(fig. 2B). Steady-state hemodynamics and blood gases are summarized in tables 1 and 2, respectively. Hypoxia increased PAP and heart rate, and decreased systemic arterial and mixed venous P^O2and S^O2. Hypoxia also caused a mild respiratory alkalosis in the conscious state.

Ketamine anesthesia had no effect on the baseline LPQ relationship during normoxia compared with the conscious state (fig. 3A). During ketamine anesthesia, hypoxia caused pulmonary vasoconstriction (P < 0.01), which increased (P < 0.01) as P^O2decreased (fig. 3A). The HPV response during ketamine anesthesia was not significantly different from that measured in the conscious state (fig. 3B). Ketamine increased heart rate during both normoxia and hypoxia (table 1). Hypoxia increased PAP, but did not further increase heart rate, during ketamine anesthesia. Systemic arterial and mixed venous blood gases were similar during normoxia and hypoxia in the conscious and ketamine-anesthetized states (table 2).

Propofol anesthesia had no effect on the baseline LPQ relationship during normoxia compared to the conscious state (fig. 4A). During propofol anesthesia, hypoxia caused pulmonary vasoconstriction (P < 0.01), which increased (P < 0.01) as P^O2decreased (fig. 4A). Increasing the dose of propofol did not further increase the HPV response (fig. 4A). The magnitude of the HPV response was potentiated (P < 0.01) during propofol anesthesia compared to the conscious state (fig. 4B). Intralipid alone had no effect on the LPQ relationship during normoxia or hypoxia. Propofol decreased SAP and LQ during normoxia. Hypoxia increased PAP and heart rate to the same extent in conscious and propofol-anesthetized dogs. Systemic arterial and mixed venous blood gases were similar during normoxia and hypoxia in the conscious and propofol-anesthetized states.

The effects of cyclooxygenase inhibition on the HPV response were assessed in the same chronically instrumented dogs studied in protocols 1 and 3. In the conscious state, cyclooxygenase inhibition with indomethacin had no effect on the LPQ relationship during normoxia (fig. 5A), whereas the magnitude of HPV was enhanced (P < 0.01) after cyclooxygenase inhibition (fig. 5B). Similar effects of indomethacin were observed during propofol anesthesia (fig. 6A and 6B). Indomethacin had no effect on baseline steady-state hemodynamics or blood gases. Changes in steady-state hemodynamics and blood gases in response to hypoxia were similar to those observed in protocols 1 and 3.

The effects of K^+ATPchannel inhibition on the HPV response were assessed in the same dogs studied in protocols 1 and 4. In the conscious state, K^+ATPchannel inhibition with glibenclamide had no effect on the LPQ relationship during normoxia (fig. 7A). The magnitude of HPV was enhanced (P < 0.01) in the conscious state after K^+ATPchannel inhibition (fig. 7B). Glibenclamide had no effect on the LPQ relationship during normoxia in propofol-anesthetized dogs (fig. 8A). Moreover, glibenclamide had no effect on the HPV response during propofol anesthesia (fig. 8B). Glibenclamide had no effect on steady-state hemodynamics in the conscious state, but decreased (P < 0.05) LQ in propofol-anesthetized dogs during both normoxia (65 ± 2 to 43 ± 2 ml · min⁻¹· kg⁻¹) and hypoxia (63 ± 8 to 47 ± 2 ml · min⁻¹· kg⁻¹). Changes in blood gases in response to hypoxia during protocol 5 were similar to those observed in protocols 1 and 3.

The main findings in this study are that (1) ketamine and propofol have no effect on the LPQ relationship during normoxia;(2) the magnitude of HPV is both stimulus- and flow-dependent in the conscious state and during ketamine and propofol anesthesia;(3) the magnitude of HPV is preserved during ketamine anesthesia and potentiated during propofol anesthesia;(4) vasodilator metabolites of the cyclooxygenase pathway and K^+ATPchannel activation modulate the HPV response in the conscious state; and (5) the modulating effect of K^+ATPchannel activation on the HPV response is not observed during propofol anesthesia.

We have previously reported that the magnitude of HPV increases as LQ increases in the conscious and isoflurane-anesthetized states, 4as well as during sevoflurane and desflurane anesthesia. 5The results of the present study extend these observations to the ketamine- and propofol-anesthetized states. The dependence of the HPV response on LQ underscores the importance of comparing the HPV response at the same level of flow when assessing the effects of anesthetics and other interventions on HPV. One advantage of utilizing LPQ plots is that they allow one to assess the effects of an intervention on the pulmonary vascular pressure gradient at a common value of LQ.

We observed a significant and reproducible HPV response when systemic arterial P^O2was decreased to approximately 60 mmHg, and the magnitude of HPV was increased further when P^O2was decreased to approximately 50 mmHg. We have consistently observed a robust HPV response in conscious dogs, 4,5whereas isolated canine lungs have been reported to exhibit a relatively weak HPV response. 10The magnitude of the HPV response is determined by the alveolar P^O2, 11and to a lesser extent by the P^O2of pulmonary arterial (mixed venous) blood. 12There were no differences in systemic arterial or mixed venous P^O2during normoxia or hypoxia among the conscious, ketamine-anesthetized, or propofol-anesthetized states. Thus, the hypoxic stimulus was the same in all three conditions. We did observe a mild respiratory alkalosis during hypoxia in the conscious state. Alkalosis can lead to an attenuation of the HPV response. 13However, the degree of alkalosis was very mild, and systemic arterial P^CO2during hypoxia was essentially identical among the three conditions.

Ketamine had no effect on the baseline LPQ relationship during normoxia in the present study. This result was expected because the pulmonary circulation has little vasomotor tone under baseline conditions in this experimental model. Several clinical studies have suggested that ketamine may induce a small increase in calculated pulmonary vascular resistance, 14,15although it is unclear whether these changes reflect a direct effect of ketamine on pulmonary vasomotor tone. Ketamine caused dose-dependent relaxation in precontracted rat 16and rabbit 17pulmonary arterial rings. Ketamine also induced dose-dependent pulmonary vasodilation in isolated precontracted rat lungs. 18,19Moreover, this effect was inhibited by verapamil, 19which suggests that ketamine may compete with Ca²⁺at Ca²⁺binding sites associated with L-type Ca²⁺channels. We have recently reported that clinically relevant concentrations of ketamine inhibit the amplitude of phenylephrine-induced oscillations in intracellular free Ca²⁺concentration in individual pulmonary artery smooth muscle cells. 20However, neither nifedipine nor verapamil had an effect on the oscillations, 21which suggests that voltage-operated Ca²⁺channels did not mediate the inhibitory effect of ketamine on the oscillations. Ketamine had no effect on the magnitude of HPV in the present study. This result is consistent with previous experimental animal 1,2and human 22studies.

Propofol also had no effect on the baseline LPQ relationship during normoxia. Although propofol is generally recognized as a vasodilator in the systemic circulation, 23,24there are conflicting reports concerning the pulmonary vascular effects of propofol. Propofol has been reported to cause a transient increase in pulmonary vascular resistance in humans, 25whereas pulmonary vasodilation has been observed both in isolated perfused lungs 18,26and in humans with artificial hearts. 27Propofol has been shown to inhibit endothelium-dependent vasodilation, 28which could theoretically result in pulmonary vasoconstriction. However, neither cyclooxygenase inhibition 4nor nitric oxide synthase inhibition 9have an effect on the baseline LPQ relationship in conscious dogs, so we predicted that propofol would not alter the baseline LPQ relationship.

In contrast to ketamine, the magnitude of HPV was increased during propofol anesthesia compared with the conscious state. This result conflicts with previous studies, in which propofol was observed to have no effect on the HPV response. 29–31However, several factors may have influenced the results of those previous studies, including the presence of background anesthetics, acute surgical trauma, the lack of a conscious control group, and concomitant changes in cardiac output.

To investigate the mechanism responsible for the potentiated HPV response during propofol, we tested the hypothesis that propofol attenuated the modulating influence of vasodilator metabolites of the cyclooxygenase pathway. Cyclooxygenase metabolites are released from hypoxic lungs, 32and cyclooxygenase inhibition has been shown to potentiate the HPV response. 33In the present study, we observed that cyclooxygenase pathway inhibition with indomethacin potentiated the HPV response in conscious dogs. Moreover, this effect was still apparent during propofol anesthesia, which indicates that inhibition of this mechanism does not play a primary role in the propofol-induced potentiation of HPV.

We also tested the hypothesis that the potentiated HPV response during propofol was caused by an attenuation in K^+ATPchannel mediated pulmonary vasodilation. K^+ATPchannels have been identified in pulmonary vascular smooth muscle. 34,35We recently reported that inhalational anesthetics can attenuate the pulmonary vasodilator response to an exogenously administered K^+ATPchannel agonist. 8Although K^+ATPchannels have been reported to mediate anoxia-induced pulmonary vasodilation, 36,37their role in the HPV response has not been investigated. In the present study, K^+ATPchannel inhibition with glibenclamide had no effect on the baseline LPQ relationship in either the conscious or propofol-anesthetized states. Glibenclamide potentiated the HPV response in conscious dogs, which indicates that K^+ATPchannel-mediated pulmonary vasodilation modulates the HPV response under these conditions. However, glibenclamide had no effect on the HPV response during propofol anesthesia. We recently demonstrated that propofol attenuates the vasorelaxant response to a K^+ATPchannel agonist in isolated pulmonary arterial rings. 38Taken together, these results suggest that propofol potentiates the magnitude of HPV by inhibiting endogenous K^+ATP-mediated pulmonary vasodilation.

Because propofol was in the same formulation used clinically, we assessed the effects of the vehicle for propofol. The solvent for propofol, intralipid emulsion, is known to cause an increase in PAP resulting from conversion of its precursor fatty acids (arachidonic acid and linoleic acid) to vasoactive metabolites. 39However, in this study, intralipid emulsion had no effect on the LPQ relationship during normoxia or hypoxia in conscious dogs. The diluents for indomethacin and glibenclamide also had no effect on the LPQ relationship. Ketamine contains a preservative (not more than 0.1 mg/ml benzethonium chloride). Although we have not directly assessed the effects of this preservative on the LPQ relationship, ketamine (with preservative) had no effect on the LPQ relationship during normoxia or hypoxia.

The plasma concentration of propofol required to prevent the response to a surgical stimulus is approximately 6 μg/ml in humans and dogs. 40Because of the pharmacokinetic characteristics for propofol in dogs, 40we chose an initial dose of 0.5 mg · kg⁻¹· min⁻¹in order to achieve an appropriate plasma concentration. Doubling the propofol infusion rate did not further potentiate the HPV response, so the initial dose may have been sufficient to induce the maximal effect. The dose of ketamine was chosen to provide reliable anesthesia in all dogs. However, no direct conclusion about the relative anesthetic potencies of these intravenous anesthetics should be inferred. In contrast to propofol, ketamine caused a significant increase in heart rate, presumably because of the central and peripheral sympathomimetic actions of this drug. 41 

In summary, compared with the response measured in the conscious state, the magnitude of the HPV response was preserved during ketamine anesthesia and potentiated during propofol anesthesia. This latter effect does not primarily involve the cyclooxygenase pathway but appears to be mediated by a propofol-induced attenuation of K^+ATP-mediated pulmonary vasodilation.

The authors thank Steve Schomisch, Pantelis Konstantinopoulos, and Mike Trentanelli for technical work, and Ronnie Sanders for secretarial support in preparing the manuscript.