The relationship among drug dosing, drug concentration, and drug response in anesthesia is complex and differs considerably from the steady-state equilibrium pharmacokinetics that we were taught in medical school. With inhalation anesthetics, end-tidal monitoring can give us an estimate of drug concentrations in the body, but we do not have any real-time monitor of drug concentrations for IV agents. We learn largely empirically the drug doses and concentrations that produce appropriate responses (or lack thereof) in a wide variety of patients; this is part of the art of anesthesia.
The relationships between dose or concentration and response can be illustrated for teaching and in clinical practice using software packages such as StanPump, RUGLOOP, IVA-SIM, and TIVATrainer, which have been reviewed recently.1 These applications have been available online for some years and incorporate a wide range of pharmacokinetic and pharmacodynamic models. However, they remain essentially tools for teaching, research, or the enthusiast. It has been the incorporation of this technology into commercial clinical devices that has introduced and demonstrated these concepts to a wide range of users.
This process began in 1996 in many parts of the world with the introduction of the Diprifusor (Astra Zeneca, United Kingdom) for administering a target-controlled infusion (TCI) of propofol. TCI devices allow the user to select a drug concentration that the pump tries to attain using a built-in pharmacokinetic model of the drug. The user can observe how the infusion rate changes to meet the set target. Many TCI pumps incorporate estimates of effect-site concentrations and some graphically display both the history and predictions of future concentrations (at the current infusion setting). More recently, “open TCI” pumps have expanded the concept to a wider range of drugs. Some of these devices allow the user to choose between various pharmacokinetic models for a drug.
Different models of the same drug deliver different amounts of drug to attain the same (modeled) concentration. This is confusing for users,2 and poses the question of which model is the most accurate.3 In this issue, Masui et al.4 have explored this question by examining the performance of 4 different propofol models under a range of conditions. They looked at 3 conventional compartmental models and 1 physiologically based recirculation model. They also reviewed the history and derivation of these models, providing useful background information and context for the various models. Masui et al. concluded that “the Schnider model… has the fewest shortcomings.”4 This guarded recommendation suggests that we still have a way to go to find ideal models.
Despite these limitations, TCI has been shown to be a useful approach to drug delivery. When a TCI pump is not available for a particular drug, or for volatile anesthetics for which target control of volatile concentration is relatively new, an alternative “advisory” approach utilizes embedded models that show drug concentrations in various compartments along with their history and predictions in the near future. The user adjusts drug delivery to achieve the desired predicted drug concentration. We have developed this approach for use with inhalation anesthetics. In our system, measured fresh gas flow rates and vaporizer settings are used as inputs to a validated uptake model, which is used to predict both end-tidal and effect-site concentrations over the following 10 minutes. Because the model is updated every 10 seconds, determining the inputs that produce a desired pattern, such as adjusting the vaporizer after reducing fresh gas flows or anticipating various stages of surgery or anesthesia, is straightforward. We have shown that this approach enables the user to make changes more rapidly5 and may facilitate the reduction of flow rates6 with a consequent significant cost reduction.
Although TCI frees the user from the complexities of dosing, especially during induction, most systems attempt to achieve a given target as rapidly as possible. With an advisory system the user can directly control the rate of change, which may be an advantage at some stages of anesthesia.
All of the above goes a long way to revealing the mysteries of pharmacokinetics in a clinically useful “real-time” setting for individual drugs. As has been discussed in the preceding editorials,7,8 and is illustrated by articles in this issue9,10 and others,11 drug interactions have a significant influence on various anesthetic end points. New devices incorporating these interactions, SmartPilot View (Drager Medical, Lubeck, Germany) and Navigator Suite (GE Healthcare, Helsinki, Finland), are the latest step in bringing the lessons and realities of pharmacological modeling to a wide range of users. These devices display both the kinetics of individual drugs and the combined interaction effects. These effects can be characterized as the probabilities of consciousness and of response to noxious stimulation (either sympathetic or motor).12 Both are advisory rather than control devices and present what is essentially the same information in quite different ways. Examples of these displays are shown by way of introduction rather than as a comprehensive review (Figs. 1, 2).
Our early experience with these devices is that they work as advertised to illustrate the drug interactions. They provide forward estimates of drug concentrations on the basis of current settings, and they have the potential to influence drug dosing at all stages of anesthesia. Interaction data are presented with reference to population norms, typically EC50 and EC95 concentrations for various responses.
As with many new monitors, a useful starting point is to follow normal patterns of drug administration and observe the predicted concentrations and effects in relation to these population norms. These devices illustrate how the common rapid bolus induction technique produces effects well beyond the EC95 for both lack of recall and lack of response to noxious stimulation. It is easy to see how these drug concentrations can be associated with very low values for Bispectral Index and arterial blood pressure and to see how alternative combinations of drugs may produce more hemodynamic stability during induction, but still obtund the stress response to tracheal intubation. The data accumulating on the potential detrimental side effects of overgenerous delivery of anesthetics13–15 suggest a need for this more precise titration.
During the maintenance phase, we can see the relative adjustments in dosage of different drugs to achieve the same effect. These displays suggest that the same end point can be achieved quickly and reliably with a variety of combinations of drugs, and we can be more confident when moving beyond our standard recipes without the need for trial and error. We can choose the set of drugs best indicated for the given situation, refining what has previously worked satisfactorily for us.
During emergence, the displays can be used to guide the timing of return of consciousness and to assist with titration of analgesia. Both systems can be used to titrate fentanyl, which is not included in most single drug TCI systems. This is useful for “loading” the patient with fentanyl before awakening to achieve adequate analgesia at the point of return of consciousness and to ensure that the level of analgesia does not decrease too rapidly. Similarly converting from predominately remifentanil-based intraoperative use to fentanyl-based postoperative analgesia becomes very straightforward.
Although these devices show a lot of promise, there is still considerable work to be done evaluating the underlying models and concepts and also the data presentation. Are all combinations, both of a pair of drugs and of the whole library of drugs that theoretically produce the same effect, really clinically equivalent? Many drugs and techniques we use routinely are not incorporated into the interaction models. For instance, morphine is not modeled, and local anesthetics and adjuvants such as clonidine may affect the response to noxious stimulus, but will not be reflected in the predictions. With inhaled agents in most anesthesia delivery systems, there is a considerable lag between a change in vaporizer setting and the change in predictions of effect that are based on measured end-tidal concentrations. With one system at a fresh gas flow of 1L/min, it took over 2 minutes for any change in prediction to occur and further time for the predictions to stabilize. Using population norms as a reference for the various effects, although similar to minimum alveolar concentration, is conceptually different from targeting a drug concentration. How should the data best be displayed and how do the display formats fit with our perceptions of our goals during anesthesia?
The accuracy of the models may not be as important as is the ability of the interface to convey information to the user. We have argued elsewhere that models perhaps only need to reach some level of “good enough” so that they can be used to adjust drug concentrations to either maintain a constant state or make discrete step changes with some consistency.16 The appropriate “state” for the patient at each stage of anesthesia can be determined by reference to various monitors of actual patient physiology, such as heart rate and arterial blood pressure, electroencephalogram-based hypnosis monitors, and new and emerging indices of the degree of noxious stimulation,12,17 in combination with knowledge of the pattern of drug concentrations.
Tools such as Navigator and SmartPilot have the potential to significantly alter the way we administer anesthesia, especially when combined with a range of monitors of effect. They should allow us to more closely titrate anesthesia delivery to patient need at different stages of surgery, and to demonstrate what is actually happening with drug concentrations. These devices may help us learn the “art” (trial, error, and experience) of anesthesia more rapidly, and encourage us to use a wider palette of drugs. Although lacking the esthetic appeal of fine art, this technology may produce a tangible improvement in outcome.
1. Struys MM, De Smet T, Mortier EP. Simulated drug administration: an emerging tool for teaching clinical pharmacology during anesthesiology training. Clin Pharmacol Ther 2008; 84:170–4
2. Enlund M. TCI: Target controlled infusion, or totally confused infusion? Call for an optimised population based pharmacokinetic model for propofol. Ups J Med Sci 2008;113:161–70
3. Absalom AR, Mani V, De Smet T, Struys MM. Pharmacokinetic models for propofol—defining and illuminating the devil in the detail. Br J Anaesth 2009;103:26–37
4. Masui K, Upton RN, Doufas AG, Coetzee JF, Kazama T, Mortier EP, Struys MM. The performance of compartmental and physiologically based recirculatory pharmacokinetic models for propofol: a comparison using bolus, continuous, and target-controlled infusion data. Anesth Analg 2010;111:368–79
5. Kennedy RR, French RA, Giles S. The effect of a model-based predictive display on the control of end-tidal sevoflurane during low flow anesthesia. Anesth Analg 2004;99:1159–63
6. Kennedy RR, French RA. Changing patterns in anesthetic fresh gas flow rates over 5 years in a teaching hospital. Anesth Analg 2008;106:1487–90
7. Gin T. Clinical pharmacology on display. Anesth Analg 2010;111:256–8
8. Short TG. Using response surfaces to expand the utility of MAC. Anesth Analg 2010;111:249–50
9. Syroid ND, Johnson KB, Pace NL, Westenskow DR, Tyler D, Bruhschwein F, Albert RW, Roalstad S, Costy-Bennett S, Egan TD. Response surface model predictions of emergence and response to pain in the recovery room: an evaluation of patients emerging from an isoflurane and fentanyl anesthetic. Anesth Analg 2010;111:380–6
10. Johnson KB, Syroid ND, Gupta DK, Manyam SC, Pace NL, LaPierre CD, Egan TD, White JL, Tyler D, Westenskow DR. An evaluation of remifentanil–sevoflurane response surface models in patients emerging from anesthesia: model improvement using effect-site sevoflurane concentrations. Anesth Analg 2010;111:387–94
11. Schumacher PM, Dossche J, Mortier EP, Luginbuehl M, Bouillon TW, Struys MM. Response surface modeling of the interaction between propofol and sevoflurane. Anesthesiology 2009;111:790–804
12. Luginbühl M, Schumacher PM, Vuilleumier P, Vereecke H, Heyse B, Bouillon TW, Struys MM. Noxious stimulation response index: a novel anesthetic state index based on hypnotic–opioid interaction. Anesthesiology 2010;112:872–80
13. Kheterpal S, O'Reilly M, Englesbe MJ, Rosenberg AL, Shanks AM, Zhang L, Rothman ED, Campbell DA, Tremper KK. Preoperative and intraoperative predictors of cardiac adverse events after general, vascular, and urological surgery. Anesthesiology 2009;110:58–66
14. Monk TG, Saini V, Weldon BC, Sigl JC. Anesthetic management and one-year mortality after noncardiac surgery. Anesth Analg 2005;100:4–10
15. Leslie K, Myles PS, Forbes A, Chan MT. The effect of bispectral index monitoring on long-term survival in the B-Aware trial. Anesth Analg 2010;110:816–22
16. Kennedy RR. Individualising target-controlled anaesthesia. Better models or better targets? Anaesth Intensive Care 2010;38:421–3
© 2010 International Anesthesia Research Society
17. Chen X, Thee C, Gruenewald M, Wnent J, Illies C, Hoecker J, Hanss R, Steinfath M, Bein B. Comparison of surgical stress index-guided analgesia with standard clinical practice during routine general anesthesia: a pilot study. Anesthesiology 2010;112:1175–83