The peak bispectral index time cannot predict early phase propofol pharmacodynamics with effect site-controlled infusion algorithm.

OBJECTIVES: The plasma-effect site equilibration rate constant (ke0) of propofol was determined with peak bispectral index (BIS) time (T(PEAK)) in our previous study. The present study has been conducted to evaluate the ke0's performance with effect site-controlled infusion algorithm.
MATERIALS AND METHODS: Forty unpremedicated patients were randomized to group TE1 (Schnider's pharmacokinetic model with ke0 adapted to T(PEAK) = 74s) and TE2 (T(PEAK) = 96s). In stage 1, all patients received propofol with effect-site concentration (Ce) controlled infusion. Once the pump had injected the mass of propofol necessary to achieve pre-set Ce and while the infusion was stopped, target was reset at 0 μg/ml. When BIS returned to 80 or above, then, in stage 2, the patients received plasma concentration controlled infusion for 10 min. The time of loss of responsiveness (LOR) and BIS were recorded. The differences of Ce at the time of LOR, lowest BIS between stages 1 and 2, hysteresis loop were used to evaluate the performance of ke0.
RESULTS: In both groups, the calculated propofol Ce at the time of LOR in stages 1 and 2 differed significantly (P<0.01); the mean lowest BIS in stage 1 were significantly higher than those in stage 2 (P < 0.05).The relations of propofol Ce versus BIS revealed the apparent hysteresis loop.
CONCLUSIONS: The study cannot clinically validate the accuracy of application of ke0 derived from the T(PEAK) = 74 s of BIS with Schnider propofol pharmacokinetic model.

RCT Entities:   Population Interventions Outcomes

Introduction

As the plasma is not the site of drug effect, hysteresis exits between plasma concentration and clinical effect. To eliminate the lag, the extension of pharmacokinetic model with a pharmacodynamic model requires one additional parameter, e.g. plasma-effect site equilibration rate constant, ke0, when target-controlled effect site concentration infusion was used.[1] Targeting the effect site may offer advantages compared with plasma, particularly under the non-steady-state conditions that frequently occur in clinical anesthesia.[2] Theoretically, effect-site concentration calculated with appropriate ke0 should be able to accurately predict the clinical effect of drugs. Because the definite pharmacokinetic study may not incorporate a pharmacodynamic component, Minto et al.,[3] proposed an approach for combining pharmacokinetics with pharmacodynamics based on the peak effect time, T. Subsequently, Muñoz et al.,[4] reported the ke0 values of propofol for children and adults with T approach using auditory evoked potential monitor. They later found that the implementation of the T measured with the bispectral index (BIS) in the Paedfusor's and Kartaria's models of propofol to target the effect site resulted in a clinically acceptable performance, which was evaluated on the basis of the difference between the predicted time of peak concentration at the effect site and the measured time of peak effect in the BIS (terror), during induction of anesthesia in children.[2] Our previous work[5] also calculated the ke0 of propofol according to the T approach. The study found that the calculated ke0 value based on T of BIS was different from other reports.[46-8] It is necessary for the ke0 to be validated before being used in clinic. In the current study, the authors want to test and verify the accuracy of ke0 derived from T of BIS using propofol pharmacokinetic model reported by Schnider et al.[6]

Materials and Methods

Patients’ recruitment and preparation

The protocol was approved by the Ethics Committee of Renji Hospital of Shanghai, where the study was conducted. Written informed consent was obtained from each patient. Forty unpremedicated patients with ASA physical status I~II, aged 25-86 years, scheduled to undergo elective surgery with general anesthesia were recruited. The patients with known or suspected cardiac, liver, renal, neurologic disorder, long- or short-term intake of any sedative, analgesic and psychoactive drug and any known adverse effect to the study drug were excluded. After overnight fasting, the patients were brought to a quiet operating room where a cannula was inserted into an antecubital vein for the injection of propofol and for fluid replacement, and another cannula was inserted into contralateral radial artery for the measurement of invasive blood pressure. Routine monitoring included electrocardiogram, heart rate, end-tidal partial pressure of carbon dioxide and oxyhemoglobin saturation throughout the study. On the day of surgery, patients were allocated randomly to one of the two groups. All randomizations were performed by drawing lots.

Hemodynamic and BIS monitoring

Arterial blood pressure, heart rate, oxyhemoglobin saturation and ECG were monitored using the Datex-Ohmeda S/5 monitor (Datex-Ohmeda Division, Instrumentarium Corp., Helsinki, Finland). The drug effect was continuously monitored using the Bispectral Index (BIS) with a plug-in module. The BIS™ Sensor (Aspect Medical Systems, Inc), composed of a self-adhering flexible band holding four electrodes, was applied to the forehead and temple according to the instructions of the manufacturer. Before sensor application, the skin was carefully cleaned with an alcohol swab, allowed to dry, and then cleansed with 3M Red Dot™ Trace prep. (3M Co. Canada) to reduce the impedance of skin and achieve a better trace quality. The bispectral index smoothing rate of the BIS module was set to 15s. Data of blood pressure, heart rate and BIS were recorded automatically with the Datex-Ohmeda software S/5 Collect® onto the computer hard disk every 5 s. The same as our previous study,[5] the BIS lag time of 7s was subtracted for off-line analysis.

Propofol delivery design

Propofol (AstraZenca, Italy) was administered by target-controlled infusion (TCI) system via a Graseby 3500 syringe pump (SIMS Graseby Ltd., Herts, England) using the pharmacokinetic parameter sets reported by Schnider et al.[6] The parameters of model are V1 = 4.27 (l), V2 = 18.9–0.391×(age-53) (l), V3 = 238 (l), Cl1 = 1.89+ ((weight–77)×0.0456) + ((LBM-59)×(-0.0681))+ (height–177)×0.0624 (l/min), Cl2 = 1.29–0.024×(age–53) (l/min), Cl3 = 0.836 (l/min). LBM (lean body mass) was calculated based on the previous method.[9] The pump was controlled by STELPUMP, a program written by Coetzee for Windows 95/NT operating system (Microsoft, USA). The maximum infusion rate of the pump is 1200 ml/h. The delta time of control program was set at 5s, with which the pump was capable of adjusting the infusion rate every 5s. In group TE1 (n=20) and TE2 (n=20), the TCI device targeted and calculated the effect-site concentration with a ke0 value adapted to the T = 74s and 96s as reported by our previous study[5] and original Schnider's study,[6] respectively. The ke0 was calculated corresponding to T using the algorithm described by Minto et al.[3]

In stage 1, the following anesthetic protocol was adhered to after pre-oxygenation administration. The patients in group TE1, TE2 received initially the effect site-controlled propofol concentration at 10 μg/ml and 8 μg/ml, respectively. Once the pump had injected the mass of propofol necessary to achieve this effect site concentration and while the infusion was stopped, the target concentration was reset to 0 μg/ml. Observer's assessment alertness/sedation (OAA/S)[10] were determined every 15 s from the start of infusion. Loss of responsiveness (LOR) was defined as an OAA/S score equal to 0 (i.e., no response to painful trapezius squeeze) by an author (Huang), who did not know the propofol delivery scheme. The time of LOR and BIS was recorded. No attempt was made to rouse the subjects until recovery of responsiveness (OAA/S=5, i.e., response readily to name spoken in normal tone) and/or BIS 80 or above. This was the end of the first anesthetic cycle. Subsequently, in stage 2, the patients in group TE1, TE2 received plasma-controlled propofol concentration at 10 μg/ml and 8 μg/ml for 10 min, respectively. The time of LOR and BIS were monitored and recorded as stage 1. All patients maintained spontaneous breathing via a facemask delivering 100% O2. If, at any time during both stages, spontaneous ventilation was not satisfactory, manual breathing support using a circle system with 100% oxygen was applied.

Pharmacodynamic analysis and evaluation of ke0

We firstly hypothesized that the ke0 values used to control the infusion in the study can accurately reflect the time course of the observed clinical effect. The performance of ke0 was evaluated using the following measures:

Theoretically, the same effect endpoint should occur at the same effect site propofol concentration (Ce). Therefore, given the interindividual (and intraindividual) variability of ke0 (and that pharmacokinetics), the mean target-controlled effect site concentration of propofol at the time of LOR in stage 1 should be comparable with the mean calculated effect site concentration in stage 2 (plasma-controlled infusion); otherwise, it suggests that the ke0 has a poor performance.

With the target-controlled effect site concentration in stage 1, the plasma compartment must be overdosed initially to drive the drug into the effect site so as to produce a peak Ce. The height of the peak Ce is higher transitorily than pre-set “target” concentration. When the plasma concentration (Cp) was targeted in stage 2, Ce approximates slowly to the pre-set Cp. Therefore, using a value of ke0 accurately reflecting the clinical effect, the median lowest BIS in stage 1 should be lower than or comparable with that in stage 2. Otherwise, it suggests that the ke0 has a poor performance.

We also calculated the terror of stage 1, which was defined as the time difference between the predicted peak Ce time and the observed lowest BIS time in both groups.

Visually, plotting BIS versus propofol Ce calculated from an appropriate ke0 will not produce the hysteresis loop. If this is not the case, it suggests that the ke0 has a poor performance.

Statistical Analysis

Data were presented as the means (SD) or the medians and interquartile range (25%, 75%) after checking the distribution of the data with Shapiro-Wilk test. Differences between and within the groups were determined using paired t-test and unpaired t-test with mixed models for repeated measures data, Wilcoxon rank sum test, Chi-square where appropriate. Significance level was set at 0.05. Calculation was performed with SAS 9.1.

Results

Population demographics for two groups are shown in Table 1. Groups were not significantly different for age, gender, weight, height and lean body mass (all P>0.05). Two patients were excluded from related analysis for unable to obtain stable BIS baseline or fail to result in OAA/S=0. Six patients showed transient respiration depress in stage 1, and 10 patients showed respiration depress needing jaw support to maintain normal ventilation in stage 2.

Table 1

Patients and administration of propofol characteristics

As shown in Table 1, patients lost responsiveness more slowly when the TCI device targeted the plasma (75 s and 100 s) than targeted the effect site (60 s and 65 s) in groups TE1 and TE2 (P < 0.05), respectively. In stage1, to obtain a target effect site concentration, the TCI system administered propofol 1.96± 0.30 mg/kg in group TE1 and 2.11± 0.36 mg/kg in group TE2 (NS). The mean (range) injection duration of loading dose was 34 ± 5 (25-45) s in group TE1 and 38 ± 8 (25-50) s (NS) in group TE2.

The measures to evaluate the performance of the ke0s can be found in Table 2. (i) At the time of LOR, the mean target-controlled propofol Ces (8.05 μg/ml, 5.95 μg/ml) in stage 1 were significantly greater than the calculated Ces (6.00 μg/ml, 4.38 μg/ml) in stage 2 in group TE1 and TE2, respectively (P < 0.01). (ii) The median peak BIS (36, 32) of stage 1 were significantly higher than the mean lowest BIS (20, 23) of stage 2 in group TE1 and TE2 (P < 0.01). (iii) In stage 1, the terror -5 s (-20, 15) in group TE1 had no difference compared with 20 s (-18, 46) in group TE2 (NS).

Table 2

Observations and measures during study period

Figure 1 (a, c) depicts the relations between the measured BIS versus propofol plasma concentration for group TE1 and TE2, which revealed the apparent hysteresis in the plasma concentration versus BIS relation. Moreover, as shown in Figure 1 (b, d), the hysteresis loop still occurred with ke0s calculated from T = 74 s and 96 s in both groups, which suggested that the ke0s used had poor performance.

Figure 1

Relations of the measured bispectral index (BIS) and propofol concentration (a and c). Graphs show the relation between the measured BIS versus propofol plasma concentration for both groups, thereby revealing the hysteresis in the relation. Moreover, the hysteresis still occurred with the relation between BIS and effect site concentration (Ce) calculated from ke0 adapted to peak effect time = 74s and 96 s (b, d) for both groups. To visualize the hysteresis loop clearly, we showed only the measured data in stage 1.

Discussion

The ke0, developed to account for the hysteresis in the plasma concentration-effect relation, is useful for targeting the effect site instead of the plasma, for designing and interpreting clinical pharmacologic research.[1-611-13] Our goal was to determine whether the ke0 of propofol (adapted to T of BIS = 74 s) was appropriate for prediction of pharmacodynamics.[5] If so, this must be particularly relevant for target-controlled effect site concentration infusion in our population. For comparison, we also examined the integrated propofol pharmacokinetic-pharmacodynamic (PKPD) model with ke0 adapted to T of BIS = 96 s reported by Schnider et al.[6]

The predicted effect site concentrations should be time independently and unequivocally related to hypnotic effect with an appropriate ke0 value for propofol.[81113] Therefore, any individual should achieve LOR at the same effect site concentration of propofol in different stages of the current study. Given the large interindividual variability (and that of intraindividual) relative to population typical ke0 value (and that pharmacokinetics),[5] the mean effect site concentration should be similar in a population. Furthermore, we used the same concentration, so the mean peak BIS should be comparable with the mean lowest BIS in both steps. In addition, the relations of effect site concentration-BIS should not show hysteresis loop pattern.[14] However, we failed to observe the above-mentioned phenomena, which suggests that the ke0s used in this study had poor performance.

Muñoz et al.,[2] found that the implementation of the T (65 s) of BIS in the Paedfusor's and Kartaria's models of propofol to target the effect site resulted in a clinically acceptable performance in children. They made the conclusion on the basis of the difference between the predicted peak effect site concentration time and the measured peak BIS time (terror), which has also been used in previous studies.[1314] We did not use terror as the primary outcome measurement because it was not a sensitive indicator to detect ke0 difference. Although biased terror of -5 s was found, predicted effect site concentration-BIS hysteresis still occurred as reflected in our results. In view of the variable time delay of BIS and the acute change in BIS with loss of consciousness,[14-16] we also did not use the BIS difference at the time of LOR as primary outcome measurement.

We did not clinically validate the accuracy of ke0 based on the peak BIS time (74 s) reported by our previous study.[5] Interestingly, we also cannot validate the accuracy of ke0 (T= 96 s, ke0 = 0.46 min-1) which was validated by Struys et al.[13] The reasons leading to failure of ke0s might be the followings. First, the ke0 is dependent on the underlying pharmacokinetic model, and if the pharmacokinetic model is biased, the estimate of ke0 will bias by that error.[17] The three-compartment model of Schnider et al.,[6] was selected because previous studies[818] have shown that the model performed well under the condition of constant infusion. An additional advantage of using the Schnider model as a pharmacokinetic basis would have been that we could compare the ke0s derived from different sources with each other. However, as pointed out by Coppens et al.,[17] the central compartment volume of propofol appears to be a determinant of a quicker increase in plasma concentration (and that effect site concentration), especially in the first few seconds after bolus injection.[16] So, the possible differences of pharmacokinetics will inevitably affect the results. In the second place, it is related to the influence of propofol infusion rate on plasma-effect site equilibration. If the pharmacokinetics is correct, this is perhaps the most important factor. Our ke0 was derived (≈7200 ml/h) and validated (≈780 ml/h) in different propofol administrated rate. Struys et al.,[14] found that propofol plasma-site equilibration occurs more rapidly after a bolus than after rapid infusion. The t½ ke0 should be approximately 1.2 min (ke0 = 0.58 min-1) when propofol was administered at bolus rate ≈6300 ml/h. Compared with their rate, we actually used a faster rate in our original study. However, Masui and colleagues[16] reported a ke0 0.414 min-1 with infusion rate 10-160 mg⋅kg-1 ⋅h-1 (much less than the administration rate of our original study). They suggested that the failure of the ke0 performance is an expected consequence of a flawed assumption with conventional mammillary compartmental models of instantaneous mixing in the central compartment.

Several critics might be raised. First, effect compartment is a theoretical compartment; its parameters are often dependent on index of observed signs. The BIS and LOR are different signs. Their ke0s may be different. Therefore, we simultaneously measured BIS and LOR in order to avoid this bias. Second, this study used two targets of 10 μg/ml and 8 μg/ml. As the accuracy of ke0 might be different for different effective concentrations, if so, the inter-group comparison might be biased. One might suspect that 8 μg/ml and 10 μg/ml whether or not are appropriate to titrate for obtaining LOR in 95% of the cases in respective group. However, these concentrations were determined on the basis of simulation with 2×ED50 dose (rounded to integer) for female Chinese patients reported by Short et al.[19] Moreover, the mean propofol dosage measured at LOR was similar between both groups. Third, for consistency purpose, we correct the time delay of the BIS with 7 s as our previous study. However, the exact time delay is unknown[15] although Struys et al.,[14] and Masui and colleagues[16] found that the lag time of 10 s and 19.7-20.3 s resulted in the best fit in their studies. Fourth, we did not measure plasma concentration, so it is difficult to know exactly the source of the difference. However, we wanted to evaluate the clinical performance of the TCI system using the most common pharmacokinetic model, which is available to the clinician. Similar studies have also been performed without measurement of the Cp of propofol.

In conclusion, the present study tried to test the ke0 based on the peak effect time of bispectral index from our previous study to target-controlled infusion system of propofol. Unfortunately, we failed to validate clinically the accuracy of application of ke0 (T = 74 s of BIS) with Schnider propofol pharmacokinetic model.