578001 research-article2015
PRF0010.1177/0267659115578001PerfusionMathew et al.
Original Paper
Performance of target-controlled infusion of propofol using two different pharmacokinetic models in open heart surgery - a randomised controlled study
Perfusion 1–9 © The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0267659115578001 prf.sagepub.com
PJ Mathew,1 S Sailam,1 R Sivasailam,1 SKS Thingnum2 and GD Puri1
Abstract We compared the performance of a propofol target-controlled infusion (TCI) using Marsh versus PGIMER models in patients undergoing open heart surgery, in terms of measured plasma levels of propofol and objective pharmacodynamic effect. Methods: Twenty-three, ASA II/III adult patients aged 18-65 years and scheduled for elective open heart surgery received Marsh or PGIMER (Postgraduate Institute of Medical Education and Research) pharmacokinetic models of TCI for the induction and maintenance of anaesthesia with propofol in a randomized, active-controlled, non-inferiority trial. The plasma levels of propofol were measured at specified time points before, during and after bypass. Results: The performances of both the models were similar, as determined by the error (%) in maintaining the target plasma concentrations: MDPE of -5.0 (-12.0, 5.0) in the PGIMER group vs -6.4 (-7.7 to 0.5) in the Marsh group and MDAPE of 9.1 (5, 15) in the PGIMER group vs 8 (6.7, 10.1) in the Marsh group. These values indicate that both models over-predicted the plasma propofol concentration. Conclusions: The new pharmacokinetic model based on data from Indian patients is comparable in performance to the commercially available Marsh pharmacokinetic model. Keywords propofol; open heart surgery; pharmacokinetics; target-controlled infusion
Introduction The technique of target-controlled infusion (TCI) has become increasingly popular since its introduction in the 1990s. TCI is, essentially, a pharmacokinetic modeldriven, computer-controlled device that allows intravenous anaesthetic drugs to be administered to a theoretical target plasma concentration, calculated mathematically by the delivery system algorithm.1 The choice of pharmacokinetic model and infusion control algorithm are major determinants of the performance of a TCI system. By using a pharmacokinetic model derived from its own population and including co-variates such as age, weight and sex, the performance of the TCI can be improved to minimize bias and improve precision. When delivering propofol by TCI, good predictive performance in healthy adult patients and easy titration of depth of anaesthesia have been demonstrated.1 The main purpose of this study was to compare the performance of a target-controlled infusion of propofol in
patients undergoing open heart surgery under cardiopulmonary bypass, using an indigenously developed pharmacokinetic model, hereafter referred to as the PGIMER model and the commercially available Marsh pharmacokinetic model.2 The PGIMER pharmacokinetic model used in this study was derived from the 1Department
of Anaesthesia and Intensive Care, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh, India 2Department of Cardiothoracic Surgery, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh, India Corresponding author: Goverdhan D Puri Department of Anaesthesia and Intensive Care Postgraduate Institute of Medical Education and Research (PGIMER) Chandigarh, PIN-160012 India. Email: [email protected]
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
2
Perfusion
Indian population and pharmacokinetic parameters were calculated after a single bolus dose of propofol in twenty-five healthy adult patients.3
Methods After approval from the Institutional Ethics Committee and written informed consent, patients aged 18-65 years, belonging to American Society of Anesthesiologists (ASA) physical status II or III and scheduled for elective open heart surgery under general anaesthesia were recruited in a randomised, active-controlled, non-inferiority study design. Patients with a weight >30% of ideal bodyweight, neurological disorder, severe pulmonary artery hypertension, severe stenotic valvular lesions with ventricular dysfunction, cyanotic heart disease, New York Heart Association (NYHA) class IV or severe valvular dysfunction were excluded. Premedication consisted of oral alprazolam 0.25-0.5 mg the night before and on the morning of surgery. Routine physiological monitoring was done (electrocardiography, pulse oximetry, non-invasive blood pressure monitoring), using the S/5 Avance anaesthesia monitor (Datex Ohmeda Inc., Madison, WI). The bispectral index (BIS) was obtained with disposable sensors attached to the forehead. Morphine sulphate 0.15 mg/kg i.v was administered before induction, during which an in-dwelling arterial cannula, a central venous line and pulmonary artery catheters were inserted. Patients were randomly allocated to one of the two groups in a stratified manner - Marsh or PGIMER pharmacokinetic model group - for the induction and maintenance of anaesthesia, using computer-generated random numbers in sealed, opaque envelopes. PJM generated the random allocation sequence while GDP and SS enrolled and assigned the participants, who were blinded to group allocation to the study groups. The anaesthetic technique consisted of a fentanyl bolus of 3µg/kg over a period of three minutes before induction. The patients were induced and maintained on propofol TCI to achieve a pre-specified target plasma propofol concentration ranging from 1.8 to 2.2 µg/ml. After the loss of consciousness, neuromuscular blockade was achieved with vecuronium 0.1mg/kg to facilitate endotracheal intubation. In the PGIMER group, a laptop computer was used: i) to implement the pharmacokinetic model using a control algorithm, ii) to provide a user interface and iii) to control communication through serial ports (RS 232) with the infusion system (Pilot C, Fresenius, Paris, France) and a vital signs monitor. The control algorithm calculates the loading dose (LD) of propofol, using the formula: LD = Target concentration × Volume of central compartment. Subsequently, the propofol concentration was maintained between 1.8 to 2.2 µg/ml, using the equation: R = LD × (K12 e-K21t + K13 e-K31t),
Table 1. Pharmacokinetic parameters used in PGIMER and Marsh models. Parameter
PGIMER model
Marsh Model
Volume of central compartment (L/kg) K12 (min-1) K21 (min-1) K13 (min-1) K31 (min-1)
0.214
0.228
0.13 0.109 0.056 0.014
0.114 0.055 0.0419 0.0033
where R is the rate of infusion, LD is the loading dose, K12= the distribution rate constant for transfer from central to tissue compartment, K21= the elimination rate constant for transfer from tissue to central compartment, K13=the distribution rate constant from central to deep tissue compartment and K31= the elimination rate constant from deep tissue to central compartment. The pharmacokinetic parameters used for these calculations are given in Table 1. In the Marsh group, Diprifusor TCI (software version 2.0; AstraZeneca Pharmaceuticals, Cambridge, UK) was used for induction and maintenance, which incorporates pharmacokinetic parameters described by Marsh et al.2 All patients received a continuous fentanyl infusion at a rate of 1µg/kg/hr and a continuous infusion of vecuronium at a rate of 50µg/kg/hr. In both groups, arterial blood samples were obtained at the following time points: before endotracheal intubation, before skin incision, before sternotomy, at 10 minute intervals during the period between sternotomy and aortic cannulation, before aortic cannulation, at 5, 10 and 15 mins after starting cardiopulmonary bypass (CPB), at the start of rewarming, at 30 minutes of rewarming, post CPB and at the end of surgery for estimating the plasma propofol concentration, using high-performance liquid chromatography (HPLC) as described by Pavan et al.4 wherein the detection limit of propofol in human serum was 0.1 µg/ml. The samples were analysed in the laboratory, blinded to group allocation. At the end of surgery, propofol, fentanyl and vecuronium infusions were stopped and the patient shifted to the post-anaesthesia care unit without antagonizing muscle relaxants for elective mechanical ventilation. The patients were extubated on meeting standard criteria for extubation. Post-operatively, all patients were monitored continuously for haemodynamic stability and patients were subjected to a structured interview for intra-operative awareness as modified by Brice and colleagues.5 Statistical analysis The primary outcome measure was the performance of the system as assessed by calculating the differences in predicted/target plasma concentration and measured plasma concentrations of propofol using the methods
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
3
Mathew et al.
described by Varvel et al:6 median performance error (MDPE) and median absolute performance error (MDAPE). MDPE and MDAPE are measures of bias and precision of the system, respectively. The performance error was calculated by the formula: Cp ( measured ) − Performance error ( % ) = × Cp ( predicted ) 100 / Cp ( predicted ) Median performance error (MDPE) -It is a measure of the systematic tendency of the system (bias) to underestimate or overestimate the measured concentration of blood propofol. If the bias has a positive value, it indicates that the measured value is, on average, greater than the system prediction and vice versa. , MDPEi = median {PEij j = 1, , Ni} Median absolute performance error (MDAPE): indicates the measure of inaccuracy in the ith subject. , MDAPEi = median { PE i j j = 1,., Ni} Where Ni is the number of |PE| values obtained for the ith subject. The secondary outcome measures were the differences in bispectral index, heart rate and mean arterial pressure at various time points during the procedure. A sample size of 26 was calculated with an alpha error of 0.05 and beta error of 0.2 in order to establish a noninferiority margin of 0.2 for the PGIMER model which had an MDAPE of 8.5% compared to the Marsh model which had an MDAPE of 4.5% during the pilot study. The normality of data was checked by the Kolmogorov Smirnov test of normality. Continuous variables were expressed as mean and standard deviation. Intergroup comparisons were made using one-way analysis of variance (ANOVA). Categorical data was analysed by the Chisquared (Fisher’s exact) test. The intra-operative heart rate and mean arterial blood pressure at different time intervals were compared between the groups by ANOVA, with appropriate post-hoc testing with the Bonferroni correction. All tests were evaluated for 95% confidence limits. A p-value of 0.05) (304.1±66.72 min in the PGIMER group vs 290±69.48 min in the Marsh group). The mean cardiopulmonary bypass time was also similar in both the groups (127.90±44.14 min in the PGIMER group vs 115.80±46.64 min in the Marsh group). When a plasma concentration of 1.8 µg/ml was targeted, the measured plasma propofol concentration (Figure 2) was less than the predicted value in both the PGIMER group and the Marsh group during pre-CPB and post-CPB. In both groups, the plasma concentration increased steadily over time and reached equilibrium during the pre-CPB period. During CPB, the plasma concentration fell by 20-30% during the initial 15 minutes and reached a steady state over time, as seen during the start of re-warming. In the post-CPB period, the concentration increased to pre-CPB levels. A similar pattern was observed in patients where the target concentrations were 2 µg/ml and 2.2 µg/ml in the Marsh and PGIMER groups (Figures 3 and 4). Performance Error: Performance errors during the procedure in both groups are depicted in Figures 5a and 5b. The MDPE during the entire procedure was -5.0 (-12.0, 5.0) in the PGIMER group vs -5.5 (-11.1 to 5.2) in the Marsh group (p=0.70), which reflects a negative bias and indicates that the pharmacokinetic parameters used in both groups over-predicted the plasma propofol concentration similarly. A similar bias also was observed during the pre-CPB and post-CPB periods (Tables 4a and 4b). Overall, the MDAPE in the PGIMER group of 9.09 (5, 15) was comparable to the Marsh group’s 7.78 (5.5, 12.43) (p=0.27). Bispectral Index: The mean BIS values at various time points during the surgery were lower in the Marsh group compared to the PGIMER group in all the three subgroups of target concentrations of 1.8, 2 and 2.2 µg/ml (Figure 6). This difference was statistically significant during cardiopulmonary bypass: CPB 10 min, CPB 15 min and start of re-warming. None of the patients had intra-operative awareness as assessed by the modified Brice questionnaire. In both groups, the heart rate remained stable during the entire surgery. There was a statistically significant difference in the mean arterial pressure at time points loss of consciousness (LOC), before intubation and at the end of surgery (Table 5).
Discussion In the current study, a constant concentration-targeted infusion was used throughout the surgery. During the induction and initial part of maintenance, plasma propofol concentrations remained below the target, as
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
4
Perfusion
Figure 1. CONSORT Flow Diagram. Table 2. Distribution of patients who received the specific target concentration with TCI pumps. Target concentration
Marsh group (n= 10)
PGIMER group (n= 13)
1.8 µg/mL 2.0 µg/mL 2.2 µg/mL
4 5 1
4 7 2
reflected by the plasma concentration before endotracheal intubation. However, subsequent concentrations remained near the target in both groups. During CPB, the plasma concentration decreased during hypothermia and, subsequently, it steadily increased towards the target concentration and overshot the target after CPB. The increase in plasma concentration after the start of re-warming could be attributed to accumulated propofol during CPB due to the effect of hypothermia on the metabolism of propofol. It could also be due to a decrease in the volume of distribution as a result of excretion of excess plasma volume as the patient is prepared to separate from CPB. This is similar to a previous
Table 3. Patient characteristics.
Age (years) Sex (M:F) BMI Height (cm) Weight (kg) Type of surgery (CABG/Valve replacement/CHD).
PGIMER group Marsh group (n=l 3) (n=10)
p-value
48.9 ± 13.46 8:5 22.82 ± 2.56 160 ± 5.9 60 ±9.1 5/7/1
0.19 0.65 0.57 0.77 0.38 0.71
43.7 ± 15.12 6:4 23.61 ± 3.54 164 ± 10.5 62 ± 19.7 4/5/1
All values except gender distribution and type of surgery are mean ± SD. M:F : male: female; BMI: body mass index; CABG: coronary artery bypass grafting; CHD: coronary heart disease.
study by Coetzee et al.1 who studied adult patients undergoing non-cardiac surgery, using a propofol TCI incorporating the Marsh pharmacokinetic model. In an another study by Barvais et al.,7 the measured plasma concentration was higher than the predicted in adult patients undergoing coronary artery surgery with a target-controlled infusion of propofol using the Marsh
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
5
Mathew et al.
Figure 2. Measured plasma concentrations when the target concentration was set at 1.8 µg/ml.
Marsh model, n=4; PGIMER model, n=4. The values are shown as mean. The standard deviation is shown on the positive side of the Marsh group and on the negative side of the PGIMER group.
Figure 3. Measured plasma concentrations when the target concentration was set at 2 µg/ml.
Marsh model, n=5; PGIMER model, n=7. The values are shown as mean. The standard deviation is shown on the positive side of the Marsh group and on the negative side of the PGIMER group.
pharmacokinetic model wherein the MDPE was +21.2% during the pre-CPB period and +9.6% during CPB. We found that both the TCI systems overestimated the plasma propofol concentrations though the magnitude
of bias was less in both the groups: (-2.6% (off CPB) and -10.5% (on CPB) in the Marsh group and -4.2% (off CPB) and -8.3% (on CPB) in the PGIMER group) when compared to the bias reported by Barvais et al.7 One
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
6
Perfusion
Figure 4. Measured plasma concentrations when the target concentration was set at 2.2 µg/ml.
Marsh model, n=2; PGIMER model, n=1. The values are shown as mean. The standard deviation is shown on the positive side of the Marsh group and on the negative side of the PGIMER group.
Figure 5a. Performance Errors during the procedure in Marsh group.
Each box depicts the interquartile range with the central line indicating median performance error. The line outside the box indicates 5th and 95th percentiles of performance error.
possible explanation for this obvious difference could be a variable rate infusion and the higher concentration targeted in their study as higher and variable rated infusion tend to affect the performance, as reported by Swinehoe et al.8 It may be worth noting that both Marsh and PGIMER models were derived from studies on healthy adults. Despite this fact, we found the performance of these two pharmacokinetic models acceptable
during complex open heart surgery in physiologically challenging patients during propofol anaesthesia. Bias and predictive accuracy of both models were within the clinically acceptable range, even much better than described by Varvel et al.6 We observed a sudden fall in the measured propofol concentrations during CPB, even as early as five minutes after the initiation of CPB. This decrease in plasma
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
7
Mathew et al.
Figure 5b. Performance Errors during the procedure in PGIMER group.
Each box depicts the interquartile range with the central line indicating median performance error. The line outside the box indicates 5th and 95th percentiles of performance error.
Table 4A. MDPE and MDAPE during pre CPB and post CPB.
MDPE MDAPE
PGIMER group(n=13)
Marsh group(n=10)
p-value
−4.22 (–10.00, 6.17) 8.95 (5.00, 12.74)
−2.61 (–7.50, 6.81) 7.22 (4.53, 11.11)
0.42 0.15
MDPE: median performance error; MDAPE: median absolute performance error.
Table 4B. MDPE and MDAPE during CPB.
MDPE MDAPE
PGIMER group
Marsh group
p-value
−8.33 (–17.00, –0.91) 10.00 (3.89, 19.44)
−10.50 (–13.83, –4.58) 10.81(6.21, 13.83)
0.92 0.754
MDPE: median performance error; MDAPE: median absolute performance error.
concentration is similar to the findings of Barbosa et al.9 Hiraoka et al.10 observed an initial reduction in total drug concentration as a result of haemodilution and an increase in the volume of distribution at the beginning of the CPB procedure. Dawson et al.11 also reported that CPB caused a reduction in total propofol plasma concentrations, while unbound concentrations remained stable.
The present study demonstrated a good relationship between BIS, which is a pharmacodynamic parameter, and plasma propofol concentration, which is a pharmacokinetic parameter. BIS values fell after induction and remained stable in the pre-CPB period and post-CPB period. During CPB, BIS values decreased significantly, which could be attributed to the effect of the hypother-
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
8
Perfusion
Figure 6. Bispectral Index (mean) during the procedure.
The p-values were not significant (p>0.05) except for the time-points CPB 10 minutes, CPB 15 minutes, start of re-warming, where the p-values are marked.
Table 5. Mean arterial pressure at specific time points.
Baseline LOC Before intubation Before incision Before sternotomy Sternotomy 10 min Sternotomy 20 min Before cannulation Post CPB End of surgery
PGIMER group (n=13)
Marsh group (n=13)
p-value
86±13.13 71±10.3 72±5.39 75±12.30 83±13.24 82±13.17 82±8.98 77±5.15 67±8.52 67±7.17
98±14.92 92±16.01 90±15.40 86±13.47 89±17.20 84±16.06 86±10.30 78±11.97 73±15.01 77±12.37
0.057 0.003* 0.003* 0.065 0.419 0.677 0.329 0.690 0.237 0.045*
(Values in mean±SD). LOC: loss of consciousness; CPB: cardiopulmonary bypass. p
PRF0010.1177/0267659115578001PerfusionMathew et al.
Original Paper
Performance of target-controlled infusion of propofol using two different pharmacokinetic models in open heart surgery - a randomised controlled study
Perfusion 1–9 © The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0267659115578001 prf.sagepub.com
PJ Mathew,1 S Sailam,1 R Sivasailam,1 SKS Thingnum2 and GD Puri1
Abstract We compared the performance of a propofol target-controlled infusion (TCI) using Marsh versus PGIMER models in patients undergoing open heart surgery, in terms of measured plasma levels of propofol and objective pharmacodynamic effect. Methods: Twenty-three, ASA II/III adult patients aged 18-65 years and scheduled for elective open heart surgery received Marsh or PGIMER (Postgraduate Institute of Medical Education and Research) pharmacokinetic models of TCI for the induction and maintenance of anaesthesia with propofol in a randomized, active-controlled, non-inferiority trial. The plasma levels of propofol were measured at specified time points before, during and after bypass. Results: The performances of both the models were similar, as determined by the error (%) in maintaining the target plasma concentrations: MDPE of -5.0 (-12.0, 5.0) in the PGIMER group vs -6.4 (-7.7 to 0.5) in the Marsh group and MDAPE of 9.1 (5, 15) in the PGIMER group vs 8 (6.7, 10.1) in the Marsh group. These values indicate that both models over-predicted the plasma propofol concentration. Conclusions: The new pharmacokinetic model based on data from Indian patients is comparable in performance to the commercially available Marsh pharmacokinetic model. Keywords propofol; open heart surgery; pharmacokinetics; target-controlled infusion
Introduction The technique of target-controlled infusion (TCI) has become increasingly popular since its introduction in the 1990s. TCI is, essentially, a pharmacokinetic modeldriven, computer-controlled device that allows intravenous anaesthetic drugs to be administered to a theoretical target plasma concentration, calculated mathematically by the delivery system algorithm.1 The choice of pharmacokinetic model and infusion control algorithm are major determinants of the performance of a TCI system. By using a pharmacokinetic model derived from its own population and including co-variates such as age, weight and sex, the performance of the TCI can be improved to minimize bias and improve precision. When delivering propofol by TCI, good predictive performance in healthy adult patients and easy titration of depth of anaesthesia have been demonstrated.1 The main purpose of this study was to compare the performance of a target-controlled infusion of propofol in
patients undergoing open heart surgery under cardiopulmonary bypass, using an indigenously developed pharmacokinetic model, hereafter referred to as the PGIMER model and the commercially available Marsh pharmacokinetic model.2 The PGIMER pharmacokinetic model used in this study was derived from the 1Department
of Anaesthesia and Intensive Care, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh, India 2Department of Cardiothoracic Surgery, Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh, India Corresponding author: Goverdhan D Puri Department of Anaesthesia and Intensive Care Postgraduate Institute of Medical Education and Research (PGIMER) Chandigarh, PIN-160012 India. Email: [email protected]
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
2
Perfusion
Indian population and pharmacokinetic parameters were calculated after a single bolus dose of propofol in twenty-five healthy adult patients.3
Methods After approval from the Institutional Ethics Committee and written informed consent, patients aged 18-65 years, belonging to American Society of Anesthesiologists (ASA) physical status II or III and scheduled for elective open heart surgery under general anaesthesia were recruited in a randomised, active-controlled, non-inferiority study design. Patients with a weight >30% of ideal bodyweight, neurological disorder, severe pulmonary artery hypertension, severe stenotic valvular lesions with ventricular dysfunction, cyanotic heart disease, New York Heart Association (NYHA) class IV or severe valvular dysfunction were excluded. Premedication consisted of oral alprazolam 0.25-0.5 mg the night before and on the morning of surgery. Routine physiological monitoring was done (electrocardiography, pulse oximetry, non-invasive blood pressure monitoring), using the S/5 Avance anaesthesia monitor (Datex Ohmeda Inc., Madison, WI). The bispectral index (BIS) was obtained with disposable sensors attached to the forehead. Morphine sulphate 0.15 mg/kg i.v was administered before induction, during which an in-dwelling arterial cannula, a central venous line and pulmonary artery catheters were inserted. Patients were randomly allocated to one of the two groups in a stratified manner - Marsh or PGIMER pharmacokinetic model group - for the induction and maintenance of anaesthesia, using computer-generated random numbers in sealed, opaque envelopes. PJM generated the random allocation sequence while GDP and SS enrolled and assigned the participants, who were blinded to group allocation to the study groups. The anaesthetic technique consisted of a fentanyl bolus of 3µg/kg over a period of three minutes before induction. The patients were induced and maintained on propofol TCI to achieve a pre-specified target plasma propofol concentration ranging from 1.8 to 2.2 µg/ml. After the loss of consciousness, neuromuscular blockade was achieved with vecuronium 0.1mg/kg to facilitate endotracheal intubation. In the PGIMER group, a laptop computer was used: i) to implement the pharmacokinetic model using a control algorithm, ii) to provide a user interface and iii) to control communication through serial ports (RS 232) with the infusion system (Pilot C, Fresenius, Paris, France) and a vital signs monitor. The control algorithm calculates the loading dose (LD) of propofol, using the formula: LD = Target concentration × Volume of central compartment. Subsequently, the propofol concentration was maintained between 1.8 to 2.2 µg/ml, using the equation: R = LD × (K12 e-K21t + K13 e-K31t),
Table 1. Pharmacokinetic parameters used in PGIMER and Marsh models. Parameter
PGIMER model
Marsh Model
Volume of central compartment (L/kg) K12 (min-1) K21 (min-1) K13 (min-1) K31 (min-1)
0.214
0.228
0.13 0.109 0.056 0.014
0.114 0.055 0.0419 0.0033
where R is the rate of infusion, LD is the loading dose, K12= the distribution rate constant for transfer from central to tissue compartment, K21= the elimination rate constant for transfer from tissue to central compartment, K13=the distribution rate constant from central to deep tissue compartment and K31= the elimination rate constant from deep tissue to central compartment. The pharmacokinetic parameters used for these calculations are given in Table 1. In the Marsh group, Diprifusor TCI (software version 2.0; AstraZeneca Pharmaceuticals, Cambridge, UK) was used for induction and maintenance, which incorporates pharmacokinetic parameters described by Marsh et al.2 All patients received a continuous fentanyl infusion at a rate of 1µg/kg/hr and a continuous infusion of vecuronium at a rate of 50µg/kg/hr. In both groups, arterial blood samples were obtained at the following time points: before endotracheal intubation, before skin incision, before sternotomy, at 10 minute intervals during the period between sternotomy and aortic cannulation, before aortic cannulation, at 5, 10 and 15 mins after starting cardiopulmonary bypass (CPB), at the start of rewarming, at 30 minutes of rewarming, post CPB and at the end of surgery for estimating the plasma propofol concentration, using high-performance liquid chromatography (HPLC) as described by Pavan et al.4 wherein the detection limit of propofol in human serum was 0.1 µg/ml. The samples were analysed in the laboratory, blinded to group allocation. At the end of surgery, propofol, fentanyl and vecuronium infusions were stopped and the patient shifted to the post-anaesthesia care unit without antagonizing muscle relaxants for elective mechanical ventilation. The patients were extubated on meeting standard criteria for extubation. Post-operatively, all patients were monitored continuously for haemodynamic stability and patients were subjected to a structured interview for intra-operative awareness as modified by Brice and colleagues.5 Statistical analysis The primary outcome measure was the performance of the system as assessed by calculating the differences in predicted/target plasma concentration and measured plasma concentrations of propofol using the methods
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
3
Mathew et al.
described by Varvel et al:6 median performance error (MDPE) and median absolute performance error (MDAPE). MDPE and MDAPE are measures of bias and precision of the system, respectively. The performance error was calculated by the formula: Cp ( measured ) − Performance error ( % ) = × Cp ( predicted ) 100 / Cp ( predicted ) Median performance error (MDPE) -It is a measure of the systematic tendency of the system (bias) to underestimate or overestimate the measured concentration of blood propofol. If the bias has a positive value, it indicates that the measured value is, on average, greater than the system prediction and vice versa. , MDPEi = median {PEij j = 1, , Ni} Median absolute performance error (MDAPE): indicates the measure of inaccuracy in the ith subject. , MDAPEi = median { PE i j j = 1,., Ni} Where Ni is the number of |PE| values obtained for the ith subject. The secondary outcome measures were the differences in bispectral index, heart rate and mean arterial pressure at various time points during the procedure. A sample size of 26 was calculated with an alpha error of 0.05 and beta error of 0.2 in order to establish a noninferiority margin of 0.2 for the PGIMER model which had an MDAPE of 8.5% compared to the Marsh model which had an MDAPE of 4.5% during the pilot study. The normality of data was checked by the Kolmogorov Smirnov test of normality. Continuous variables were expressed as mean and standard deviation. Intergroup comparisons were made using one-way analysis of variance (ANOVA). Categorical data was analysed by the Chisquared (Fisher’s exact) test. The intra-operative heart rate and mean arterial blood pressure at different time intervals were compared between the groups by ANOVA, with appropriate post-hoc testing with the Bonferroni correction. All tests were evaluated for 95% confidence limits. A p-value of 0.05) (304.1±66.72 min in the PGIMER group vs 290±69.48 min in the Marsh group). The mean cardiopulmonary bypass time was also similar in both the groups (127.90±44.14 min in the PGIMER group vs 115.80±46.64 min in the Marsh group). When a plasma concentration of 1.8 µg/ml was targeted, the measured plasma propofol concentration (Figure 2) was less than the predicted value in both the PGIMER group and the Marsh group during pre-CPB and post-CPB. In both groups, the plasma concentration increased steadily over time and reached equilibrium during the pre-CPB period. During CPB, the plasma concentration fell by 20-30% during the initial 15 minutes and reached a steady state over time, as seen during the start of re-warming. In the post-CPB period, the concentration increased to pre-CPB levels. A similar pattern was observed in patients where the target concentrations were 2 µg/ml and 2.2 µg/ml in the Marsh and PGIMER groups (Figures 3 and 4). Performance Error: Performance errors during the procedure in both groups are depicted in Figures 5a and 5b. The MDPE during the entire procedure was -5.0 (-12.0, 5.0) in the PGIMER group vs -5.5 (-11.1 to 5.2) in the Marsh group (p=0.70), which reflects a negative bias and indicates that the pharmacokinetic parameters used in both groups over-predicted the plasma propofol concentration similarly. A similar bias also was observed during the pre-CPB and post-CPB periods (Tables 4a and 4b). Overall, the MDAPE in the PGIMER group of 9.09 (5, 15) was comparable to the Marsh group’s 7.78 (5.5, 12.43) (p=0.27). Bispectral Index: The mean BIS values at various time points during the surgery were lower in the Marsh group compared to the PGIMER group in all the three subgroups of target concentrations of 1.8, 2 and 2.2 µg/ml (Figure 6). This difference was statistically significant during cardiopulmonary bypass: CPB 10 min, CPB 15 min and start of re-warming. None of the patients had intra-operative awareness as assessed by the modified Brice questionnaire. In both groups, the heart rate remained stable during the entire surgery. There was a statistically significant difference in the mean arterial pressure at time points loss of consciousness (LOC), before intubation and at the end of surgery (Table 5).
Discussion In the current study, a constant concentration-targeted infusion was used throughout the surgery. During the induction and initial part of maintenance, plasma propofol concentrations remained below the target, as
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
4
Perfusion
Figure 1. CONSORT Flow Diagram. Table 2. Distribution of patients who received the specific target concentration with TCI pumps. Target concentration
Marsh group (n= 10)
PGIMER group (n= 13)
1.8 µg/mL 2.0 µg/mL 2.2 µg/mL
4 5 1
4 7 2
reflected by the plasma concentration before endotracheal intubation. However, subsequent concentrations remained near the target in both groups. During CPB, the plasma concentration decreased during hypothermia and, subsequently, it steadily increased towards the target concentration and overshot the target after CPB. The increase in plasma concentration after the start of re-warming could be attributed to accumulated propofol during CPB due to the effect of hypothermia on the metabolism of propofol. It could also be due to a decrease in the volume of distribution as a result of excretion of excess plasma volume as the patient is prepared to separate from CPB. This is similar to a previous
Table 3. Patient characteristics.
Age (years) Sex (M:F) BMI Height (cm) Weight (kg) Type of surgery (CABG/Valve replacement/CHD).
PGIMER group Marsh group (n=l 3) (n=10)
p-value
48.9 ± 13.46 8:5 22.82 ± 2.56 160 ± 5.9 60 ±9.1 5/7/1
0.19 0.65 0.57 0.77 0.38 0.71
43.7 ± 15.12 6:4 23.61 ± 3.54 164 ± 10.5 62 ± 19.7 4/5/1
All values except gender distribution and type of surgery are mean ± SD. M:F : male: female; BMI: body mass index; CABG: coronary artery bypass grafting; CHD: coronary heart disease.
study by Coetzee et al.1 who studied adult patients undergoing non-cardiac surgery, using a propofol TCI incorporating the Marsh pharmacokinetic model. In an another study by Barvais et al.,7 the measured plasma concentration was higher than the predicted in adult patients undergoing coronary artery surgery with a target-controlled infusion of propofol using the Marsh
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
5
Mathew et al.
Figure 2. Measured plasma concentrations when the target concentration was set at 1.8 µg/ml.
Marsh model, n=4; PGIMER model, n=4. The values are shown as mean. The standard deviation is shown on the positive side of the Marsh group and on the negative side of the PGIMER group.
Figure 3. Measured plasma concentrations when the target concentration was set at 2 µg/ml.
Marsh model, n=5; PGIMER model, n=7. The values are shown as mean. The standard deviation is shown on the positive side of the Marsh group and on the negative side of the PGIMER group.
pharmacokinetic model wherein the MDPE was +21.2% during the pre-CPB period and +9.6% during CPB. We found that both the TCI systems overestimated the plasma propofol concentrations though the magnitude
of bias was less in both the groups: (-2.6% (off CPB) and -10.5% (on CPB) in the Marsh group and -4.2% (off CPB) and -8.3% (on CPB) in the PGIMER group) when compared to the bias reported by Barvais et al.7 One
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
6
Perfusion
Figure 4. Measured plasma concentrations when the target concentration was set at 2.2 µg/ml.
Marsh model, n=2; PGIMER model, n=1. The values are shown as mean. The standard deviation is shown on the positive side of the Marsh group and on the negative side of the PGIMER group.
Figure 5a. Performance Errors during the procedure in Marsh group.
Each box depicts the interquartile range with the central line indicating median performance error. The line outside the box indicates 5th and 95th percentiles of performance error.
possible explanation for this obvious difference could be a variable rate infusion and the higher concentration targeted in their study as higher and variable rated infusion tend to affect the performance, as reported by Swinehoe et al.8 It may be worth noting that both Marsh and PGIMER models were derived from studies on healthy adults. Despite this fact, we found the performance of these two pharmacokinetic models acceptable
during complex open heart surgery in physiologically challenging patients during propofol anaesthesia. Bias and predictive accuracy of both models were within the clinically acceptable range, even much better than described by Varvel et al.6 We observed a sudden fall in the measured propofol concentrations during CPB, even as early as five minutes after the initiation of CPB. This decrease in plasma
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
7
Mathew et al.
Figure 5b. Performance Errors during the procedure in PGIMER group.
Each box depicts the interquartile range with the central line indicating median performance error. The line outside the box indicates 5th and 95th percentiles of performance error.
Table 4A. MDPE and MDAPE during pre CPB and post CPB.
MDPE MDAPE
PGIMER group(n=13)
Marsh group(n=10)
p-value
−4.22 (–10.00, 6.17) 8.95 (5.00, 12.74)
−2.61 (–7.50, 6.81) 7.22 (4.53, 11.11)
0.42 0.15
MDPE: median performance error; MDAPE: median absolute performance error.
Table 4B. MDPE and MDAPE during CPB.
MDPE MDAPE
PGIMER group
Marsh group
p-value
−8.33 (–17.00, –0.91) 10.00 (3.89, 19.44)
−10.50 (–13.83, –4.58) 10.81(6.21, 13.83)
0.92 0.754
MDPE: median performance error; MDAPE: median absolute performance error.
concentration is similar to the findings of Barbosa et al.9 Hiraoka et al.10 observed an initial reduction in total drug concentration as a result of haemodilution and an increase in the volume of distribution at the beginning of the CPB procedure. Dawson et al.11 also reported that CPB caused a reduction in total propofol plasma concentrations, while unbound concentrations remained stable.
The present study demonstrated a good relationship between BIS, which is a pharmacodynamic parameter, and plasma propofol concentration, which is a pharmacokinetic parameter. BIS values fell after induction and remained stable in the pre-CPB period and post-CPB period. During CPB, BIS values decreased significantly, which could be attributed to the effect of the hypother-
Downloaded from prf.sagepub.com at Freie Universitaet Berlin on May 4, 2015
8
Perfusion
Figure 6. Bispectral Index (mean) during the procedure.
The p-values were not significant (p>0.05) except for the time-points CPB 10 minutes, CPB 15 minutes, start of re-warming, where the p-values are marked.
Table 5. Mean arterial pressure at specific time points.
Baseline LOC Before intubation Before incision Before sternotomy Sternotomy 10 min Sternotomy 20 min Before cannulation Post CPB End of surgery
PGIMER group (n=13)
Marsh group (n=13)
p-value
86±13.13 71±10.3 72±5.39 75±12.30 83±13.24 82±13.17 82±8.98 77±5.15 67±8.52 67±7.17
98±14.92 92±16.01 90±15.40 86±13.47 89±17.20 84±16.06 86±10.30 78±11.97 73±15.01 77±12.37
0.057 0.003* 0.003* 0.065 0.419 0.677 0.329 0.690 0.237 0.045*
(Values in mean±SD). LOC: loss of consciousness; CPB: cardiopulmonary bypass. p
