2005 Annual Scientific Meeting - CLICK FOR PROGRAMME

Pharmacokinetic variability

Frédérique Servin, MD - Hôpital Bichat – Paris

Variability amongst individuals is an everyday observation. Age, sex, weight, body mass index, pathological states (hepatic or renal impairment, cardiac failure, anxiety …) create huge differences in the way bodies will deal with administered drugs, and thereby in the doses required to generate a given effect, even in the absence of any changes in the dose / response relationship.

An obvious variability is body weight, and it seems logical to give bigger doses to heavier individuals, so much so that most manual bolus / infusion schemes are given per kilo body weight, even if it can lead to gross overdosing (ex: remifentanil). Nevertheless, this normalisation is not sufficient and it is common knowledge for example that many drugs dosages have to be modified in the elderly, even if no pharmacodynamic changes can be demonstrated [1].

So long as the only way to administer intravenous drugs was by giving bolus doses and manually controlled infusions, the anaesthetists coped with this variability by reducing or increasing the usual dose according to their feeling and experience. The launching of TCI in clinical practice has dramatically changed the problem. TCI is more precise than manual control and reduces per se the variability to the drug response [2].

But the anaesthetist does not decide anymore which physical amount of drug is given. He/she targets an effect through a predicted concentration hopefully creating this desired effect. The true amount of drug is calculated and administered by the device. Therefore, if the pharmacokinetic model implemented in the device is not adapted to the patient, the amount of drug given may be wrong; leading to inaccurate predicted concentrations when compared to measured ones, and thereby to unwanted signs of either over or under dosage.

The adequacy of a pharmacokinetic model may be estimated by statistical parameters such as bias (mean prediction error) or imprecision (mean absolute prediction error). Whatever the effort to reduce those parameters, some degree of variability still remains and usually imprecision cannot be lowered fewer than 20%, a “good” pharmacokinetic model yielding no bias and an imprecision around 30% [3], which is much less than pharmacodynamic variability.

A constant bias over time is not a big issue, since by titrating the target concentration to effect the anaesthetist will be able to ensure the same quality of clinical anaesthesia despite the difference between predicted and measured concentrations [4]. More troublesome are either an important imprecision[5] or an inadequacy of the model at certain phases of anaesthesia (ex induction) and not at other (maintenance). This can be observed for example when models designed for young patients are applied to elderly patients in whom the initial distribution of the drug is altered.

In the following figure, the Marsh model implemented into the Diprifusor™ has been used to administer anaesthesia to patients over 80 years old. The left side of the figure displays the discrepancy between predicted and measured concentrations at induction of anaesthesia whereas the right hand side displays the agreement between predicted and measured concentrations during recovery. This discrepancy is due to the fact that changes with ageing in propofol pharmacokinetics are mainly limited to initial distribution[6] as demonstrated in the following figure where the same manual dosing regimen has been simulated in a 25 yr and a 85 yr individuals (PK model of Schnider including age as a covariate).

It is therefore desirable to use a model which has been built taking into account the population you wanted to treat. It is not always possible, and current research improves the models, for example by considering cardiac output in physiological models, more quickly than marketed devices may follow. Does it mean that TCI has to be limited to patients strictly following the population described in the models? Certainly not in most patients (children are a different problem). TCI means easy titration, and this way even imperfect models may lead to better anaesthesia than manual control [7]. Therefore, so far TCI has always demonstrated better clinical results for intravenous anaesthesia than manual control, and even if it is good practice to choose the available model better suited to the treated population, a good knowledge of the devices associated with titration will allow adequate handling of intravenous drugs in everyday practice.

References:

1.      Rupp SM, Castagnoli KP, Fischer DM and Miller RD. Pancuronium and vecuronium pharmacokinetics and pharmacodynamics in younger and elderly subjects. Anesthesiology 1987; 67: 45 - 49.

2.       Hu C, Horstman DJ and Shafer SL. Variability of target-controlled infusion is less than the variability after bolus injection. Anesthesiology 2005; 102: 639-645.

3.       Short TG, Tam YH, Tan P and Oh TE. Pharmacokinetic model-controlled infusion of midazolam. A prospective evaluation during general anaesthesia. Anaesthesia 1993; 48: 187-191.

4.       Ausems ME, Stanski DR and Hug CC. An evaluation of the accuracy of pharmacokinetic data for the computer assisted infusion of alfentanil. Br J Anaesth 1985; 57: 1217-1225.

5.       Oei-Lim VL, Kalkman CJ, Makkes PC, Ooms WG and Hoogstraten J. Computer controlled infusion of propofol for conscious sedation in dental treatment. Br Dent J 1997; 183: 204-208.

6.       Schnider TW, Minto CF, Gambus PL, et al. The influence of method of administration and covariates on the pharmacokinetics of propofol in adult volunteers. Anesthesiology 1998; 88: 1170-1182.

7.       Passot S, Servin F, Pascal J, Charret F, Auboyer C and Molliex S. A comparison of target- and manually controlled infusion propofol and etomidate/desflurane anesthesia in elderly patients undergoing hip fracture surgery. Anesth Analg 2005; 100: 1338-1342, table of contents.