The introduction of rapidly acting and metabolized IV anesthetic drugs has led to a marked increase in the use of continuous IV techniques. The complexity of pharmacokinetics and pharmacodynamics during continuous administration makes it difficult to predict easily the duration of drug effect after the infusion is discontinued. This issue is certainly clinically relevant because, to some extent, anesthesiologists base the selection of a drug on their perception of how long its effect will persist.
The terminal elimination half-life is often discussed as though it predicts recovery. However, Shafer and Varvel  have shown that this use of the terminal half-life can be misleading. Using computer simulations, they demonstrated that the rate of decreasing either plasma or effect site concentration is a sensitive function of the duration of that infusion, rather than a result of elimination half-lives. Thus, the time necessary for a 50% decline in plasma concentration after a continuous infusion is terminated has been called the context-sensitive half-time. A term introduced by Hughes and co-workers , it refers to the infusion time as context. These computer simulations, based on pharmacokinetic modeling, disclosed that sufentanil has a shorter decrement time compared with alfentanil-at least after infusions up to eight hours' duration . However, the 50% decrease in drug concentration could not be specifically identified with clinical recovery end points. Furthermore, drug interactions with simultaneously administered anesthetics may influence recovery.
The aim of this study, therefore, was to compare the clinical recovery characteristics of the two analgesics, sufentanil and alfentanil, with the estimates of the model-based approach and to determine relative decrement values associated with recovery. Additionally, we tried to determine the validity of currently available equipotency values for the two analgesics.
After we obtained institutional ethics committee approval and written, informed consent, we studied 40 consecutive patients of both sexes, ASA physical status I or II, undergoing major abdominal surgery (Table 1). Patients with impaired renal and hepatic function or a known history of chronic drug or alcohol abuse were excluded, as well as those who were obese, as expressed as a body mass index >30 kg/m2.
All patients were given 20 mg of clorazepate dipotassium as premedication in the early morning on the day of surgery. The patients were allocated randomly to receive either sufentanil and propofol (Group 1) or alfentanil and propofol (Group 2) for the induction and maintenance of anesthesia. The opioids were administered by means of target-controlled infusions (TCI) [4-6], whereas propofol was titrated manually using Graseby 3400 (Graseby Medical Ltd, Watford, UK) syringe pumps. The pharmacokinetic variable sets used were derived from Hudson et al.  for sufentanil and Scott and Stanski  for alfentanil (Table 2).
After the induction of anesthesia with propofol 2 mg/kg, both sufentanil (0.45 ng/mL) and alfentanil (225 ng/mL) were targeted to equipotent effect site concentrations (CeT; nomenclature according to ), assuming a relative potency of 500:1 (sufentanil to alfentanil), according to available data from the literature [1,10,11]. The target concentrations were kept constant during the whole anesthetic, to achieve a pseudo steady-state condition. Vecuronium 0.1 mg/kg was administered to facilitate tracheal intubation, with subsequent incremental doses according to the requirements of the surgical procedure.
Continuous recording of the electroencephalogram (EEG) was started in the awake patient by bilateral fronto-occipital silver-silver chloride electrode montages with an Fpz reference. The processed EEG parameter 90% spectral edge frequency was derived using the Drager pEEG monitor (Dragerwerk, Lubeck, Germany). Electrode impedance was always <2k Omega. The propofol infusion was titrated to achieve a 90% spectral edge frequency value between 10 and 12 Hertz in all patients to adjust a comparable level of anesthesia .
After tracheal intubation, ventilation was controlled with 65% nitrous oxide in oxygen using a fresh gas flow of 1 L/min. Respiratory and cardiovascular monitoring was performed according to the institution's standard practice and consisted of continuous ECG, invasive arterial pressure using a radial artery 20-gauge cannula, central venous pressure, pulse oxymetry, end-tidal CO2, and core temperature, which was kept above 35.5[degree sign]C using warming devices. Mean arterial pressure was maintained above 70 mm Hg, and the end-tidal CO2 was kept between 35 and 45 mm Hg. Neuromuscular block was monitored by train-of-four counts and reversed after closure of the peritoneum.
At the end of the surgical procedure, the propofol infusion was stopped, and the CeT for sufentanil and alfentanil were set to 0 while nitrous oxide was discontinued to the breathing circuit. Assisted ventilation was provided with 40% oxygen in air until the patients began to breathe spontaneously. Patients were tracheally extubated when respiration rate was >or=to6 breaths/min and end-tidal CO2 was <or=to50 mm Hg, and they were transferred to the postanesthesia care unit (PACU). After the infusions, patient recovery was documented and scored every 5 min by an anesthesiologist blinded to the opioid received by the individual patient (Table 3). Discharge from PACU to the surgical intermediate care unit was provided when all of the conditions (Table 3) were fulfilled at two consecutive observation points within a 10-min interval.
To determine the decline in drug concentrations of sufentanil and alfentanil, arterial samples from the radial artery cannula were taken for analysis at the following time points: 5 min before stopping the infusion (T1); at the time of stopping (T2); 5, 10, 15, 20, 30, 45, 60, 90, and 120 min (T3-T11) thereafter; and 3, 6, 12, and 24 h (T12-T15) thereafter. The samples were collected in lithium-heparinized vials, immediately centrifuged, and stored at -20[degree sign]C.
The samples were analyzed using commercial sufentanil (R 33 800) and alfentanil (R 39 209) radioimmunoassay kits (Janssen Biotech, Olen, Belgium). Because of the low-concentrated samples compared with alfentanil, the determination of sufentanil required a sorbent extraction procedure beforehand using Extrelut 3 solid-phase extraction columns (Merck, Darmstadt, Germany). The assays were performed according the method of Woestenborghs et al. . Activity was measured in a LKB Wallace (Turku, Finland) 1219 Rackbeta liquid scintillation counter, and the results were calculated by using spline approximation. The limits of quantitation for sufentanil and alfentanil were 18 pg/mL and 35 pg/mL, respectively. The limits of detection, however, were somewhat lower. Precision analysis disclosed a within-day variation coefficient of 5.1%-8.5% and a between-day variation coefficient of 8.5%-10% for both assays.
Descriptive data are presented as mean values +/- SEM. After testing normal distribution, the between-group comparisons were performed by using Student's t-test, based on the assumption that a minimal difference in end point criteria of 30 min, which we defined as clinically relevant, could be assessed with a power of 0.9. Probability values <0.05 were considered significant. The relationship of end point criteria to the duration of infusion was determined by using linear regression analysis.
The prediction error (PE) was calculated to compare the predicted (calculated) and measured sufentanil and alfentanil concentrations as Equation 1
Data are expressed as bias and precision. Bias (mean PE) provides an estimate of a systematic over- or underprediction, whereas precision (mean absolute PE) is a measure of the scatter of the data around the line of perfect prediction. All calculations were performed by using the StatView statistical software package (Abacus Concepts, Berkeley, CA).
Forty patients were enrolled in the study. All demographic data, including age, weight, and gender, were comparable across the two treatment groups, as was the duration of surgery. However, the alfentanil infusion was a mean 23 min shorter compared with the sufentanil infusion. This was not statistically significant, but it could have influenced recovery. There was a considerable difference in the total amount of propofol required per hour (Table 4).
Three patients could not be included in the analysis because of the following protocol violations: one patient in the sufentanil group developed a central anticholinergic syndrome and had to be treated with physostigmine. In the alfentanil group, one patient continued to receive mechanical ventilation because of severe hypothermia, and the serum samples for one patient could not be assayed reliably.
The declines of sufentanil and alfentanil Cpm are displayed in Figure 1 and Figure 2. The pharmacokinetic parameters for sufentanil and alfentanil used for TCI resulted in negative PE values (bias) of -17.1% and -16.8% and median absolute PE values (precision) of 20.1% and 24.5%, respectively. These values represent data derived from samples taken at a time of zero infusion rate (T3-T15). The negative bias indicates that, on average, the measured plasma concentration was lower than the targeted concentration, reflecting an overprediction of the TCI system. Regression plots comparing Cpm and Cpcalc are displayed in Figure 3. It shows the largest variability in the highest concentration ranges for both sufentanil and alfentanil. This represents sampling time points T1 and T2, at which times the high infusion rates still contributed to relevant drug intake. There were only minor differences between the groups regarding absolute end point criteria. Extubation time was almost identical in both groups (Group 2 46.4 +/- 4.2 min versus Group 1 48.7 +/- 3.6 min), whereas discharge criteria were fulfilled significantly (P = 0.039) earlier in the alfentanil group (Group 2 99.5 +/- 11.0 min versus Group 1 131.3 +/- 15.3 min).
These results should be considerably dependent of the duration (context) of infusion. However, we found no significant relationship or dependency between extubation time and discharge from the PACU to the duration of infusion for either group.
To better assess the characteristics of the recovery of sufentanil and alfentanil, we calculated the relative decline in plasma concentrations necessary for extubation and discharge (Table 5).
The results show a significant difference in these values, which indicates that alfentanil requires a smaller relative decline to recovery end points compared with sufentanil. In other words, a 48% decrease in alfentanil mean drug concentration matched with extubation criteria, whereas a 62% decline was necessary for sufentanil to meet the same criteria. A similar difference can be found when comparing time to discharge from the PACU. Comparing the time to extubation and discharge with simulated decrement times of a similar relative decrease, the mean values are more likely described by the model. Thus, tracheal extubation after sufentanil is expected in 60 min, compared with after 55 min after alfentanil, and discharge is expected after 150 min, compared with 110 min.
To verify that the increasing dependency on complex models produces clinically meaningful results, we compared the recovery characteristics of sufentanil and alfentanil by determining clinical end points. We chose two different end points, which reflect important benchmarks in anesthetic daily practice: the recovery of spontaneous ventilation to tracheally extubate the patient safely, and a condition that allows discharge from direct anesthetic supervision to a lower dependency environment.
Thus, the theoretical assumption of a quicker clinical recovery after sufentanil infusion, based on its shorter context-sensitive half-time, compared with alfentanil, could not completely be confirmed by our results. In fact, patients in the alfentanil group could be discharged significantly earlier, which is contradictory to the pharmacokinetic simulation. This leads us to suggest that alfentanil is the drug of choice when using long-term opioid infusions. Unfortunately, our study population was not identical in terms of drug infusion, and its mean difference of 23 minutes between the groups may have influenced our results.
How does this compare with pharmacokinetic modeling? We used the same kinetic data sets on which the simulations are based for TCI. The prediction error disclosed approximately -17% deviation in terms of bias and 25% for overall precision from targeted values during zero infusion rate. The values of both sufentanil and alfentanil were comparable. In contrast, the bias obtained from the original data set for alfentanil by Scott and Stanski  was -35%, and the inaccuracy (precision) was 36%. Davidson et al.  suggested that the relatively high error in measured blood concentration at times of high infusion rates is presumably the result of inadequate mixing of the drug within the central compartment. This would explain the comparably low PE obtained in our study, which is still an issue of inaccuracy compared with the model simulation.
Based on our data, an approximate 50% decrease in alfentanil drug concentration is appropriate for tracheal extubation, whereas it is not for sufentanil, which requires a 62% decline for the same end point. This has been addressed by Shafer and Varvel  and Youngs and Shafer , who point out that the time course of sufentanil and alfentanil differ significantly when comparing different decrement times. Their simulations show that when a clinical end point requires a larger relative decrease (70%-80% decrement time), alfentanil is superior to sufentanil.
Measures to quantify recovery time may also include pharmacodynamic considerations. The steepness of the dose-response curve, derived by Cp50 or minimum alveolar anesthetic concentration studies, is an important determinant of how narrow the clinical correlates of the decreasing drug concentration will be together. Kapila et al.  compared remifentanil with alfentanil and concluded that the context-sensitive half-time is not always a predictor of recovery time, although they could demonstrate that the context-sensitive half-times of both these opioids and their pharmacodynamic offset, expressed as recovery of minute ventilation, are similar. They found a pharmacodynamic half-time of 54 minutes to adequate spontaneous ventilation for alfentanil, which is comparable to our results for time to tracheal extubation (46 minutes).
Logistic regression analysis, as introduced by Bailey , reveals the probability of drug effect as a function of its concentration (mean effect time). Using this tool, the parameter gamma determines the steepness of the concentration-effect relationship, which can grossly impact the predictive value of the context-sensitive half-times. However, the reported gamma values for alfentanil (10.5) and sufentanil (1.19) are quite different, assuming that even a small decrease in alfentanil drug concentration will result in a substantial decrease in clinical effect, whereas sufentanil requires a significantly larger decrease in concentration to match the same effect. This approach would explain our data better than the context-sensitive half-times do. On the other hand, Stanski and Shafer  analyzed a pharmacodynamic study of propofol/alfentanil interaction and pointed out the difficulties in estimating recovery based on a 50% probability of response to surgical stimulation (C50rec versus C50surg)-the extent of a 90% or higher effect, which we expect is more difficult to quantify. Another possibility, of course, is that the pharmacodynamic end points were not sufficiently sensitive to resolve the differences between the two drugs.
Taking into account that context-sensitive half-times have been calculated for single drugs, one may note that the differences in propofol dosing may have had profound residual effects on the clinical end points studied, because co-administered anesthetics may potentially influence recovery. Surprisingly, even with a significantly higher propofol infusion rate, recovery after alfentanil was still shorter.
Finally, and apparently most important, the clinical appearance of sufentanil and alfentanil recovery may have been biased by the selection of equipotency values. The CpT values for both opioids were chosen as an average of all available equipotency data. These data, however, are highly variable and based on different assessments, such as pronounced delta-wave activity on the EEG , ventilation , surgical stimulation , or minimum alveolar anesthetic concentration reduction of isoflurane . The significantly lower propofol consumption in the sufentanil group could be an indicator of an overprediction in relative potency. Unfortunately, as mentioned above, there is no consensus equipotency value. On the other hand, assuming that propofol consumption based on EEG titration is an indicator of intraoperative opioid requirements, our data suggest a much lower sufentanil versus alfentanil equipotency relationship of approximately 250-300:1. We suppose that, in this case, the recovery time course for sufentanil would be closer to the predicted context-sensitive half-times than our results imply. In contrast, it has been proposed  that sufentanil has intrinsic hypnotic properties that are more pronounced than those of other opioids. This issue must be addressed by further controlled evaluations of equianalgesia.
In summary, taking the limitations of our study into account, the clinical recovery course of sufentanil and alfentanil after continuous infusion is not fully explained by pharmacokinetic simulation. The duration of drug effect is a function of both its pharmacokinetic and pharmacodynamic properties. Computer simulations based on pharmacokinetic models have revealed context-sensitive half-times or decrement times that help to better describe the decay of drug concentrations after continuous infusion. However, we must add more clinical data when using these as estimates for the duration of anesthetic drug effect.
1. Shafer SL, Varvel JR. Pharmacokinetics, pharmacodynamics, and rational opioid selection. Anesthesiology 1991;74:53-63.
2. Hughes MA, Glass PSA, Jacobs JR. Context-sensitive half-time in multicompartment pharmacokinetic models for intravenous anesthetic drugs. Anesthesiology 1992;76:334-41.
3. Gepts E, Shafer SL, Camu F, et al. Linearity of pharmacokinetics and model estimation of sufentanil. Anesthesiology 1995;83:1194-204.
4. Shafer SL, Gregg KM. Algorithms to rapidly achieve and maintain stable drug concentrations at the site of drug effect with a computer controlled infusion pump. J Pharmacokinet Biopharm 1992;20:147-69.
5. Bailey JM, Schwieger IM, Hug CC Jr. Evaluation of sufentanil anesthesia obtained by a computer controlled infusion for cardiac surgery. Anesth Analg 1993;76:247-52.
6. Ausems ME, Vuyk J, Hug CC Jr, Stanski DR. Comparison of a computer-assisted infusion versus intermittent bolus administration of alfentanil as a supplement to nitrous oxide for lower abdominal surgery. Anesthesiology 1988;68:851-61.
7. Hudson RJ, Bergstrom RG, Thomson IR, et al. Pharmacokinetics of sufentanil in patients undergoing abdominal aortic surgery. Anesthesiology 1989;70:426-31.
8. Scott JC, Stanski DR. Decreased fentanyl and alfentanil dose requirements with age: a simultaneous pharmacokinetic and pharmacodynamic evaluation. J Pharmacol Exp Ther 1987;240:159-66.
9. Glass PSA, Glen JB, Kenny GNC, et al. Nomenclature for computer-assisted infusion devices. Anesthesiology 1997;86:1430-1.
10. O'Connor M, Sear JW. Sufentanil to supplement nitrous oxide in oxygen during balanced anaesthesia. Anaesthesia 1988;43:749-52.
11. Lemmens HJM, Burm AGL, Bovill JG, Hennis PJ. Pharmacodynamics of alfentanil as a supplement to nitrous oxide anaesthesia in the elderly patient. Br J Anaesth 1988;61:173-9.
12. Schwender D, Daunderer M, Mulzer S, et al. Spectral edge frequency of the electroencephalogram to monitor "depth" of anaesthesia with isoflurane or propofol. Br J Anaesth 1996;77:179-84.
13. Woestenborghs RJH, Timmermann PMBL, Cornelissen M-LJE, et al. Assay methods for sufentanil in plasma. Anesthesiology 1994;80:666-70.
14. Davidson JA, MacLeod AD, Howie JC, et al. The effective concentration 50 for propofol with and without 67% nitrous oxide. Acta Anaesthesiol Scand 1993;37:458-64.
15. Youngs EJ, Shafer SL. Pharmacokinetic parameters relevant to recovery from opioids. Anesthesiology 1994;81:833-42.
16. Kapila A, Glass PSA, Jacobs JA, et al. Measured context-sensitive half-times of remifentanil and alfentanil. Anesthesiology 1995;83:968-75.
17. Bailey JM. Technique for quantifying the duration of intravenous anesthetic effect. Anesthesiology 1995;83:1095-103.
18. Stanski DR, Shafer SL. Quantifying anesthetic drug interaction: implications for drug dosing. Anesthesiology 1995;83:1-5.
19. Lang E, Kapila A, Shlugman D, et al. Reduction of isoflurane minimal alveolar concentration by remifentanil. Anesthesiology 1996;85:721-8.
20. Bovill JG, Warren PJ, Schuller JL, et al. Comparison of fentanyl, sufentanil and alfentanil anesthesia in patients undergoing valvular heart surgery. Anesth Analg 1984;63:1081-6.