Effects of Imposed Acid–Base Derangement on the Cardiovascular Effects and Pharmacokinetics of Bupivacaine and Thiopental

The study protocol was jointly approved by the Animal Care and Ethics Committee of the Royal North Shore Hospital, St. Leonards, Australia, and the University of Technology, Sydney, Australia. The subjects consisted of 13 nonpregnant Merino-cross ewes (age, 1–2 yr; weight [mean ± SD], 46 ± 5 kg); 11 subjects contributed data on bupivacaine, 9 contributed data on thiopental, and 7 contributed data on both drugs. Six additional subjects were used in pilot experiments. All subjects were vaccinated against clostridial diseases and caseous lymphadenitis and were drenched with the anthelmintic ivermectin before delivery to the animal house. They were allowed to acclimatize to their new environment for at least 1 week before surgical preparation for studies.

Surgical preparation consisted of two stages, a left thoracotomy and cannulation of blood vessels performed during general anesthesia, at least 7 days apart. The thoracotomy involved placement of (1) active redirection transit time probes (Triton Technology, San Diego, CA) around the pulmonary trunk (21 mm) and left main coronary artery (6 mm) for respective measurement of cardiac output and left coronary artery blood flow; (2) a pressure transducer catheter (3-French, 100 cm, model SPR-524; Millar Instruments, Houston, TX) transmurally into the left ventricle to provide left ventricular pressure, from which the maximum value of its positive first derivative (LV dP/dt^max) was determined as an index of myocardial contractility; (3) a polyurethane catheter (45-cm, 16- or 18-gauge Cavafix Certo 358 or 255; B Braun Melsungen AG, Melsungen, Germany) in the coronary sinus for blood sampling; and (4) electrodes made from fishing wire (20-pound nylon-coated stainless steel) around the dorsal and ventral extremities of the fifth rib to acquire electrocardiographic signals. The left hemiazygos vein, which drains into the coronary sinus in the sheep, was ligated to prevent contamination of coronary sinus blood samples with systemic venous blood.

At surgery for cannulation, two polyurethane catheters (16- or 18-gauge, 45-cm, Cavafix Certo 358 or 255) for measurement of arterial blood pressure and sampling of arterial blood were placed in the left carotid artery to the brachiocephalic trunk. A double-lumen central venous catheter (70-cm Cavafix Duo, B Braun Melsungen AG) was placed via  the jugular vein for administration of drugs. After recovery, cannulated sheep were kept in metabolic crates so that patency of catheters could be maintained by continuous flushing with heparinized (5 U/ml) saline (0.9%) at approximately 3 ml/h per catheter by a high-pressure infusion system.

Flucloxacillin (1 g intravenous) was given before incision and hourly during surgery as prophylaxis against infection. Postoperative analgesia was provided with 0.5% bupivacaine by blockade, during chest closure, of the intercostal nerves cranial and caudal to the thoracotomy incision, along with carprofen (75 mg intravenous) and buprenorphine (0.3 mg intravenous) after anesthetic induction, during anesthetic recovery, and also two or three times daily for approximately 4 days thereafter.

A period of 4–7 days was left between the second surgery and the first study in accord with the recovery of the animals. Studies were performed with the animals conscious and supported by a sling in a metabolic crate to prevent alteration of their position relative to the monitoring system and to support their weight if they wanted to relax.

Five treatments were studied: test drug infusion with normal arterial pH, with acidemia from intravenous infusion of lactic acid, and with alkalemia from intravenous infusion of sodium bicarbonate; in addition, acidemia and alkalemia control studies were performed in which saline (0.9%) was substituted for the test drug. The test drugs consisted of rac -bupivacaine HCl (37.5 mg) or rac -thiopental sodium (250 mg) infused intravenously over 3 min at 10 ml/min. Based on our previous experiments, the dosage regimen of bupivacaine was expected to be sub-convulsive and that of thiopental was expected to be subanesthetic; both doses, however, were expected to cause significant myocardial depression. Bupivacaine studies preceded thiopental studies.

The regimens used to produce acidemia and alkalemia were determined empirically in pilot studies. Acidemia was produced by infusing 15%rac -lactic acid at 10 ml/min for 5 min, 5 ml/min for 15 min, then 2.5 ml/min for 25 min; infusion of the test drug or saline began 5 min after the start of the lactic acid infusion. Alkalemia was produced by infusing 8.4% sodium bicarbonate at 13 ml/min for 25 min, then 7.5 ml/min for 25 min; infusion of test drug or saline began 10 min after the start of bicarbonate infusion. For studies with normal blood pH, saline (0.9%) was infused with the same regimen as for the bicarbonate infusion.

Samples of arterial and coronary sinus blood were collected into heparinized syringes just before the start of the infusions to change blood pH and then before and 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 7.5, 10, 15, 20, 30, 40, 45, and 60 min after the start of infusion of test drug or saline. Further arterial samples were taken at 90, 120, 150, and 180 min to better characterize the whole body pharmacokinetics. The samples were stored at −20°C until analysis of drug concentration by chiral high-performance liquid chromatography. 14,15Arterial blood gases and acid–base balance were analyzed in samples drawn before and 1, 3, 5, 10, 20, 30, and 60 min after the start of infusion of drug or saline.

Analog data were acquired and stored digitally by a personal computer interfaced with a physiologic monitoring system (Triton System 6; Triton Technology, San Diego, CA) and data acquisition/analysis system (MP100 with Acqknowledge v3 software; Biopac Inc., Santa Barbara, CA) at a sample rate of 256 Hz. 16Heart rate was obtained from the electrocardiographic trace, and stroke volume was obtained from the quotient of cardiac output and heart rate. The monitoring systems were calibrated, and baseline data were collected for 5 min. Data were acquired for 3 h after the start of infusion of drug or saline. Hemodynamic data were collected as the mean values of 30-s epochs, except for LV dP/dt^max, which was collected as the maximum value in the epoch.

Mean arterial blood pressure, LV dP/dt^max, heart rate, cardiac output, and left coronary artery blood flow showed a baseline drift with alkalemia, so an automatic baseline correction algorithm (see  Appendix) was designed and incorporated into the spreadsheet used for data maintenance (Excel 97; Microsoft, Redmond, WA). This correction was applied to all data sets for consistency; however, its effects were negligible for the normal pH and acidemia data sets.

The electrocardiogram was analyzed for arrhythmias and measurements were made of time between beginning of P wave to beginning of Q wave (or R wave if no Q wave was present) (P to R wave interval), time from onset of Q wave to end of T wave (Q to T wave interval), time between consecutive R waves (R to R wave interval), and width of the electrocardiographic wave complex (QRS width); also, the Q to T wave interval was corrected for heart rate by dividing it by the square root of the R to R wave interval (corrected Q to T wave interval [QT^c]). For these measurements, means were taken of three consecutive complexes that were not part of an arrhythmia in the 10 s immediately preceding the following times: during the control period at 1, 2, 3, 4, and 5 min; at 10 and 15 min while acid–base balance was being changed; and at 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 6, 7.5, 10, 15, 20, 30, 45, and 55 min after the start of drug infusion. Cardiac arrhythmias during the first 10 min of saline or drug infusion were grouped into supraventricular arrhythmias (supraventricular premature beats, supraventricular tachycardia, supraventricular bigeminy) and ventricular arrhythmias (ventricular premature beats, ventricular tachycardia, ventricular bigeminy, ventricular trigeminy).

Hemodynamic and electrocardiographic data were analyzed for the magnitude of greatest effect (E^maxor E^min) and its time of occurrence. In addition, the sum of the effect differences to 10 min for cardiovascular and hemodynamic data and to 5 min for electrocardiographic data were determined to capture any differences between treatments in magnitude and immediate time course attributable to drug effect. 16These data provide useful univariate statistics for time series data that are analogous to areas under the curves.

Statistix for Windows (version 7.1; Analytical Solutions, Tallahassee, FL) was used on a personal computer. Cardiovascular and pharmacokinetic data were analyzed by two-factor analysis of variance with treatment the main effect and subject a repeated factor. The null hypothesis was that there was no difference in effects between treatments. A criterion of P <  0.05 was taken as weak evidence for rejection of the null hypothesis, and P <  0.01 was taken as strong evidence. If there was a significant effect of treatment, differences in mean values between treatment groups were determined by the method of least significant differences.

Results in graphic form are reported as mean ± SEM to show the predicted variability of the mean values. Tabulated results are reported as mean with 95% confidence interval. The absolute values for the relevant pretreatment predrug (baseline) cardiovascular parameters, along with maximum after-drug changes (either E^maxor E^min, as appropriate, expressed as percent of baseline values) are also given. Note that after-drug 95% confidence intervals that do not include 100% are significant changes (P <  0.05).17All tests were two tailed. The frequency of subjects having cardiac arrhythmias with each treatment was compared by chi-square analysis.

The effects of acid–base derangement on whole body pharmacokinetics were explored sequentially. First, each drug was treated as a racemate (i.e. , the same as the dosage form) by summing the respective measured blood concentrations of the enantiomers. Each set of arterial blood drug concentration–time data was fitted with a triexponential decay equation by an iterative nonlinear weighted least squares procedure decided on the basis of minimal objective function and of symmetry of residuals. 18From the equation of best fit, relevant pharmacokinetic properties of a three-compartment open model (initial dilution volume [V^c], total apparent volume of distribution [V^ss], mean total body clearance [CL^t], and slow half-life [T1/2γ]) were determined. The effect of acid–base derangement on the pharmacokinetic properties was determined by two-factor (subject × treatment) analysis of variance. Second, the pharmacokinetics of the separate enantiomers were determined as above. 18Using the Student one-sample t  test, enantioselectivity was assessed by comparing, to a value of unity, the ratio of the value for the R -enantiomer to the corresponding value for the S -enantiomer (R:S  ratio) of the respective relevant pharmacokinetic properties. Because the enantiomer concentrations were measured in each blood sample, this ratio reflects the within-subject pharmacokinetic variability. The effect of acid–base derangement on the R:S  ratio of each relevant pharmacokinetic parameter was determined by two-factor (subject × treatment) analysis of variance. Third, regional (myocardial) pharmacokinetics were determined from the concentrations of drug (as racemate) in paired arterial and coronary sinus blood samples and coronary arterial blood flow. Because some sampling catheters and probes failed during the study period, paired data were available for only four subjects with each drug and treatment; hence, calculations of myocardial kinetics by mass balance were based on mean drug concentration–time–blood flow data. 5,6,10 

Comparisons between bupivacaine and thiopental were not contemplated because the principal objectives were to determine whether the pharmacodynamics and pharmacokinetics of each drug were altered by acid–base derangement.

Lactic acid infusions decreased arterial blood pH to approximately 7.25, and sodium bicarbonate infusions increased it to approximately 7.65, with and without concurrent infusions of bupivacaine and thiopental (fig. 1). Concurrently, arterial oxygen tension (PaO²) tended to increase with lactic acid and decrease with bicarbonate infusions; conversely, arterial carbon dioxide tension (PaCO²) tended to increase with bicarbonate and decrease with lactic acid infusion. However, the animals were neither hypoxemic nor hypercapnic during the studies.

Mean within-subject predrug values did not differ between treatments with the one exception of QT^cinterval with thiopental (see discussion of effects of thiopental later in this section). Figure 2shows the representative time course of effects for raw data and for automatic baseline-corrected data (expressed as percent of baseline) for LV with thiopental infusion. With dP/dt^maxnormal arterial pH, both bupivacaine and thiopental always produced maximum drug-induced changes within 5 min with recovery by 10–15 min. The maximum changes (E^maxor E^min) paralleled the changes in sums of effect differences values, which accounted for differences in the time courses of effects; therefore, for brevity, sums of effect differences data are not shown.

The effects of bupivacaine are given in tables 1 and 2. Bupivacaine caused central nervous system (CNS) excitation in three subjects with normal arterial pH and with alkalemia but not with acidemia. When arterial pH was normal, bupivacaine decreased LV dP/dt^max, heart rate, cardiac output, stroke volume, and left coronary artery blood flow and increased mean arterial blood pressure. For all variables except left coronary artery blood flow, there were significant differences between treatment groups. Although some variables showed significant changes with all treatments (e.g. , decreased LV dP/dt^maxand increased mean arterial blood pressure), the changes with bupivacaine infusion were greater. Acidemia attenuated the decrease in cardiac output and stroke volume induced by bupivacaine, but overall, the hemodynamic effects of bupivacaine were not materially modified by changes in arterial pH. Bupivacaine with normal arterial pH also produced small but significant electrocardiographic changes; however, there were no differences related to treatment except for QT^c, for which the differences were small.

The effects of thiopental are given in tables 3 and 4. When arterial pH was normal, thiopental decreased LV dP/dt^maxand stroke volume and increased heart rate and cardiac output. Acidemia enhanced the increased heart rate and also produced a greater decrease in stroke volume than with alkalemia; however, changes in cardiac output and LV dP/dt^maxwere unaffected by acidemia or alkalemia. Thiopental with normal arterial pH also produced small but significant electrocardiographic changes; however, there were no significant differences related to treatment. There were small differences in the pretreatment values of Q T^c(P =  0.04), but there was no material difference in outcome of statistical analyses that ignored this difference or treated predrug value of QT^cas a covariate.

Supraventricular and ventricular arrhythmias were found in some subjects with both drugs and all treatments; however, there was no difference in frequency of any arrhythmia according to treatment with either drug.

Summed enantiomer drug blood concentrations are shown in figure 3. Pharmacokinetic analysis based on rac -bupivacaine found differences between treatments in V^ssand T1/2γ (table 5). Pharmacokinetic analysis as separate enantiomers, followed by determination of R:S  ratios for the respective parameters, did not show pharmacokinetic enantioselectivity with any treatment. In contrast, pharmacokinetic analysis as rac -thiopental showed no differences between treatments (table 6), but there was significant pharmacokinetic enantioselectivity in CL^tand T1/2 with all treatments but no differences in enantioselectivity between treatments.

No trends related to treatment were found when the effects on LV dP/dt^maxwere plotted against either arterial or coronary sinus blood concentration of either drug. Similarly, neither the arterial:coronary sinus blood concentration ratio nor the myocardial net flux differed according to treatment for either drug.

This study found that imposed acid–base derangements did not systematically alter the effects of bupivacaine or thiopental on cardiovascular function in healthy sheep. Although the findings are largely negative, there is a positive message in that acidemia, which is considered by many to amplify the cardiovascular toxicity of bupivacaine, attenuated some of its adverse effects. However, for thiopental, some of the adverse effects were worsened by acidemia. In addition, bupivacaine caused CNS excitation only during alkalemia or when arterial pH was normal, and this is consistent with the observation that transport of lipophilic amines across the blood–brain barrier is inhibited by acidemia. 19 

Effects of directly acting cardioactive drugs, in general, depend on their concentrations in the heart, presumably at particular subsites, and these concentrations depend on both blood and tissue pH, which influence the rate of drug transfer between blood and tissue. Apart from dose and dosing rate, a number of factors can significantly affect the short-term concentration profiles of drugs in the heart. From a mass balance viewpoint, the concentration–time course of drug in this organ is driven by its blood flow and arterial blood-tissue concentration gradient of unbound drug. These factors are, in turn, driven by the physiologic status of microcirculation, oxygenation, and blood-tissue pH difference in the organ, as well as the drug's tissue affinity for and rate of elution from the organ, the latter again being determined by the organ blood flow and tissue-venous blood drug concentration gradient. 20Although much has been learned from studying drugs with isolated organ systems in vitro , 2–4there is still insufficient information relating regional blood flow, blood pH, and pharmacokinetics and pharmacodynamics of drugs in the heart in vivo . Such data can be obtained by using an appropriate large animal model. This chronically instrumented conscious sheep preparation provides a way of obtaining high-fidelity measures of relevant drug effects (e.g. , negative inotropy) in concert with whole body and regional pharmacokinetics. Measurement of drug concentrations in samples of affluent and effluent blood of relevant organs (e.g. , for the heart, arterial and coronary sinus blood), combined with continuous measurement of organ blood flow, allows calculation of rate of drug uptake and elution therein using mass balance principles. Hence, effects of drugs in organs can be related to drug concentrations in those organs. 20 

It is difficult if not impossible to reproduce the clinical conditions of acid–base derangements in normal healthy subjects without foregoing some physiologic fidelity, introducing other perturbations, or both. Acid–base derangement was produced reliably, but the regimens were not free from potential problems. For example, the use of 8.4% sodium bicarbonate to increase pH would inevitably have increased blood tonicity also, resulting in increased blood volume by movement of water from cells, cardiovascular changes, and diuresis. Both cardiovascular changes and high urine output were observed.

Both bupivacaine and thiopental are myocardial depressants and, under certain circumstances, can inhibit, initiate, and/or support cardiac arrhythmias. 7,8,21–25To avoid the opposing effects of CNS stimulation on the myocardial effects of bupivacaine, we chose what we thought, on the basis of our previous experiments, would be a subconvulsive dose. 9However, as noted above, this dose was sufficient to cause some CNS excitation in three subjects when arterial pH was normal and during alkalemia; nevertheless, analysis with and without these animals removed from the data pool gave essentially the same results.

The pH changes in this study would have markedly changed the extent of ionization of these two drugs. At normal arterial pH, bupivacaine is predominantly ionized such that the concentration ratio of ionized:nonionized forms is 6.3:1; at pH 7.25, the ratio increases to 8.9:1, and at pH 7.65, it decreases to 3.6:1. Conversely, thiopental is predominantly nonionized at pH 7.40 with a concentration ratio of ionized:nonionized forms of 0.63:1; at pH 7.25, this decreases to 0.45:1, and at pH 7.65, it increases to 1.1:1. These changes have the potential to alter drug distribution, clearance, and effects because the nonionized form is much more lipophilic than the ionized form. 26Although the enantiomers of these two drugs are treated as separate but interacting entities in the chiral environment of the body, the changes in extent of ionization predicted by the Henderson-Hasselbalch equation are the same for both enantiomers. The plasma protein binding of both drugs is affected by pH: Bupivacaine binding is appreciably decreased with decreased plasma pH 27,28and thiopental binding is modestly increased with increased plasma pH. 29However, their binding is also modestly enantio-selective: both R -bupivacaine and R -thiopental undergo slightly greater plasma binding than their respective S -enantiomers in sheep. 13,30Through physicochemical changes, altered pH could thus affect membrane permeability, tissue:plasma distribution, regional uptake and elution, whole body drug disposition, and metabolite activity, as well as receptor-related activity of these drugs.

Despite the potential of altered pH to modify pharmacodynamics, acid–base derangement did not systematically materially alter the overall effects of either drug, except for an amelioration by acidemia of decreased stroke volume and cardiac output induced by bupivacaine. Nor did acid–base derangement systematically alter pharmacokinetics, except for an increased total distribution volume (V^ss) with bupivacaine, which, with an unchanged total body clearance, produced a prolonged slow half-life (T1/2γ). It is reasonable to surmise that this resulted from an increased distribution into tissue of unbound nonionized drug. We did not find that acid–base derangement differentially influenced pharmacokinetics between enantiomers of either drug.

Humans with bupivacaine toxicity may be acidotic because of preexisting abnormality or to consequences of bupivacaine toxicity itself, or alkalotic from effects of resuscitation attempts. Others have found that CNS toxicity of local anesthetic drugs was increased by respiratory acidosis imposed on preexisting metabolic acid–base derangement 31and that respiratory acidosis with moderate hypoxia increased the cardiotoxicity of bupivacaine. 7The importance of combined hypoxia and acidosis was emphasized by another study in which neither hypoxia nor simulated respiratory acidosis enhanced local anesthetic–induced depression of guinea pig atria in vitro , but acidosis plus hypoxia did enhance bupivacaine-induced depression of heart rate and contractile force. 32The aim of our study was to determine the cardiovascular and pharmacokinetic outcome resulting from changes to blood pH. The subjects were not anesthetized and ventilated because of the known confounding effect of general anesthetics on these same variables. 33Our method of imposing acidemia represents the accumulation of plasma lactate during convulsions, with maintenance of airway and coronary blood flow, thus avoiding myocardial hypoxia. At the doses studied, cardiovascular depression was not worsened by acidemia, and this is consistent with the outcome of an in vitro  study with imposed acidosis and lidocaine cardiotoxicity. 34In other experiments with acidemia from electrically induced muscular tetany, combined metabolic and respiratory acidosis were found to decrease the myocardium and brain tissue:plasma distribution coefficients of bupivacaine. 28This was reasoned to have occurred because the difference between intracellular and extracellular pH may have been less than normal (decreased extracellular and unchanged intracellular pH), and therefore, ion trapping in myocardium and brain would have been lessened. The current results show that when myocardial oxygen supply is adequate, acidemia per se  does not alter pharmacokinetics in a way that would enhance bupivacaine toxicity.

Dosage requirements of thiopental are decreased in the presence of increased blood pH resulting from slight over ventilation. 35This was attributed to a decreased rate of thiopental uptake into fat. 35Previous studies on thiopental disposition in dogs found that hypercapnia-induced acidemia to pH 6.8 reduced plasma thiopental concentrations by 40%, but increasing the pH to 7.7 had no effect. 36Acidemia was believed to have caused an increased rate of drug transfer into tissues, including the CNS. 36However, there were no pharmacokinetic changes found in our study to support any of these observations. Thiopental activity at γ-aminobutyric acid type A receptors is weakly pH dependent with greater activity of both enantiomers found at pH 7.0 than 8.0, 37but pharmacodynamics were not evaluated enantioselectively in the current study. Modestly enantioselective pharmacokinetics were identified for thiopental, consistent with our previous observations, 12,38but were not sensitive to changes in acid–base status.

In conclusion, some effects of bupivacaine and thiopental were modified by changes in acid–base status, e.g. , the detrimental cardiovascular effects of bupivacaine were lessened by acidemia. However, such pharmacodynamic changes could not be ascribed to changes in regional or whole body pharmacokinetics. Acid–base disturbances unaccompanied by hypoxia did not significantly exacerbate the adverse cardiovascular effects of bupivacaine or thiopental at the doses studied.

The authors thank their colleagues Yifei Huang, M.D., Ph.D. (Research Fellow), Ray Kearns (Animal House Manager), Janelle Young (Animal Technician), Sonia Gu, B.Sc. (Research Assistant), Marie Pryor, B.Sc. (Research Assistant), Kylie Wilson, B.Sc. (Technical Officer), Linda Gelgor, Ph.D. (Research Officer), and Michael Iglesias, M.Sc. (Research Assistant), all of the Royal North Shore Hospital, St. Leonards, Australia, for helpful advice and collaboration.

Corrections of baseline slope (baseline linear) drift and baseline first-order curvature (log or first exponential) were effected in two stages. The first corrected for first-order curvature of the baseline, and the second corrected for slope of baseline. Higher-order corrections had to be performed in diminishing order so that the last operation performed was slope correction. We define a first-order curve as one of which the first differential is linear. A second-order curve would be one of which the differential is a first-order curve. Slope is zero-order curvature. A parabolic curve (y = ax  ²+bx = c ) differentiates to a straight line y =  2 ax + b ; thus, a parabola by our working definition is a first-order curve.

Curvature correction was begun by finding the first differential of the data with respect to time and calculating its slope. This slope is the average second differentiation. The data were then differentiated again to find d  ²y /dx  ². Dividing the average of the second differentiation by 2 produced a correction factor that was subtracted from each point during integration of the second differential. A second integration produces the original data set, but corrected for first-order curvature.

The baseline slope was corrected similarly. Finding the slope of the undifferentiated and curve-corrected data and dividing it by two provided the correction factor. On differentiation of the curvature-corrected data, we then corrected for slope by adding the correction factor to each datum at reintegration to provide a fully curvature- and slope-corrected baseline.

Only baseline data (before and after response) were used for finding the two correction factors for data normalization. The corrections were then applied to all data, both baseline and drug response. The abscissa at the time of commencement of drug delivery was set to its unprocessed value. By including the post–drug effect data in the calculation of correction factors, correction was by interpolation, rather than extrapolation, because this is a much safer method for data preservation. When correction factors were calculated from all data, rather than baseline only, the baseline correction was not as effective, although in many cases, the algorithm still improved correction of the baseline drift.