2005 Annual Scientific Meeting - CLICK FOR PROGRAMME

Different Models:  The influence of the pharmacokinetic model and ke0 on drug delivery by TCI systems.

 Iain Glen, Glen Pharma Ltd, 35 A Bexton Rd, Knutsford, Cheshire

(jbg@glenpharma.com)

Comparative studies of different pharmacokinetic models for a particular drug are usually done by comparing the accuracy of their predictive performance, ie how well do they achieve the target concentrations they set out to achieve [1,2], using methodology such as that proposed by Varvel et al.[3].  However, this information is often not available and effect site drug concentrations cannot be readily measured to validate TCI systems which aim to control concentrations at the effect site.  This presentation provides information, derived by computer simulation, on the influence of different models on the cumulative amount of drug delivered over time when drugs are delivered by TCI. An understanding of this aspect of TCI may be more useful than information on the predictive performance of different models in providing users with guidance on the likely clinical performance of different models.

The essence of any TCI system is the conversion of a target setting into an infusion rate profile which changes over time as drug input is provided to balance distribution and elimination as predicted by a particular model. The infusion rate profile is influenced by all of the parameters of each model but by far the most important of these are the volume of the central compartment (V₁) and clearance from the central compartment (V₁ x k₁₀). The amount of drug delivered as an initial rapid infusion to achieve the target setting, with a TCI system controlling blood concentration (Cb_T), is directly related to V₁. The greater the value of V₁ in a model the greater the initial ‘bolus infusion’ provided by a TCI system.  On the other hand, clearance is the principal determinant of the infusion rate required to maintain a desired target concentration and the two again are directly related.

Simulations were performed with PK-SIM and data output files copied to Excel to examine drug delivery with a range of models for propofol.

Comparison of Diprifusor (Marsh[4]) with Schnider population PK model[5] (Cb_T)

The output of Diprifusor is not affected by the input of patient age whereas the covariates in the Schnider model influence the values of k₁₂ and k₂₁. This decreases the amount of propofol delivered as age increases from 30 to 80 years by about 20% after 4 min and only 10% after 60 min. These reductions are much smaller than the 47% difference in plasma propofol concentration required to induce unconsciousness in 25 and 75 yr old volunteers [6]. The Diprifusor and Schnider models have similar clearance values and deliver quite similar maintenance infusion rates.  However, V₁ in the Schnider model at 61ml/kg in a 70 kg patient is markedly smaller than the 228 ml/kg in Diprifusor. This means that over the first few minutes of a TCI procedure at similar target settings, the Schnider model delivers about 50% less propofol.  This deficit decreases to about 15% after 30 minutes.

With Diprifusor, as V₁ is scaled to body weight, the amount of drug delivered in terms of mg/kg is independent of body weight. In the Schnider model, as V₁ at 4.27L is independent of body weight, V₁ in terms of ml/kg decreases as body weight increases. This and the influence of height and weight on clearance results in heavier patients receiving less propofol on a mg/kg basis whereas those with a lower lean body mass will initially receive about 30% less than delivered by Diprifusor, but after 30 min, the Schnider model delivers about 15% more.

Influence of mode of administration

With any given model, operation in effect control mode (Ce_T) leads to the administration of a larger initial ‘bolus’.  However, in view of the small V₁ in the Schnider model, when operated in effect control mode with a ke0 of 0.46, the amount of drug delivered in the initial rapid infusion is greater than that provided in blood concentration mode (Cb_T) with this model but no more than that provided at the same target setting with Diprifusor Cb_T.  Thus use of the original Schnider model in Ce_T mode cannot be expected to provide any more rapid onset of effect than use of Diprifusor at the same target setting in Cb_T mode.  Doufas et al [7] have proposed an amended ke0 of 0.17 for the Schnider model and with this value the amount of drug in the initial rapid infusion is increased.

Influence of ke0 on drug delivery with Marsh model for propofol

With the Marsh model, values of ke0 ranging from 0.2 to 1.2 have been proposed [8,9]. With a ke0 of 1.2 the amount of propofol delivered in the initial bolus is increased slightly whereas with a value of 0.2, an excessive initial dose of 3.5 mg/kg is delivered at a CeT of 5μg/ml. Work is ongoing in an attempt to determine the optimum ke0 for use with this model.  Simulations indicate that the influence of ke0 on the size of the initial bolus is non linear, with changes close to 1.2 having a much smaller effect than changes at the other end of the range (Fig 1).

References

1.       Coetzee JF, Glen JB, Wium CA, Boshoff.  Pharmacokinetic model selection for target controlled infusions of propofol.  Anesthesiology 82: 1328-45, 1995

2.       Vuyk J, Engbers FHM, Burm AGL, Vletter AA, Bovill JG.  Performance of computer-controlled infusion of propofol: an evaluation of five pharmacokinetic pharmacokinetic parameter sets.  Anesth. Analg. 81: 1275-82, 1995.

3.       Varvel JR, Donoho DL, Shafer SL.  Measuring the predictive performance of computer-controlled infusion pumps.  J.Pharmacokinetics and Biopharmaceutics 20: 63-93, 1992

4.       Marsh B, White M, Morton N, Kenny GNC. Pharmacokinetic model driven infusion of propofol in children.  Br J Anaesth. 67: 41-8, 1991.

5.       Schnider TW, Minto CF, Gambus PL, Andresen C et al. The influence of method of administration and covariates on the pharmacokinetics of propofol in adult volunteers.  Anesthesiology 88:1170-82, 1998.

6.       Schnider TW, Minto CF, Shafer SL, Gambus PL et al. The influence of age on propofol pharmacodynamics.  Anesthesiology 90: 1502-16, 1999.

7.       Doufas AG, Bakhshandeh M, Bjorksten AR, Shafer SL, Sessler DI.  Induction speed is not a determinant of propofol pharmacodynamics.  Anesthesiology 101: 1112-21, 2004.

8.       White M, Schenkels MJ, Engbers FHM, Vletter A et al.  Effect-site modelling of propofol using auditory evoked potentials.  Br J Anaesth. 82: 333-339, 1999.

9. Struys MRF, De Smet T, Depoorter B, Versichelen LFM et al.  Comparison of plasma compartment versus two methods for effect compartment-controlled target infusion for propofol.  Anesthesiology 92: 399-406, 2000.