Sylvain Fouliard, Tanguy Jamier, and Marylore Chenel
Department of Clinical Pharmacokinetics and Pharmacometrics, Servier, France
Introduction: The classical mammillary compartmental models in population analyses are becoming more and more complex, integrating physiological or pharmacological features (enterohepatic recycling, multiple absorption sites…). Yet, the interpretation of peripheral compartments parameters is not straightforward. In parallel, minimal PBPK models [1-3] are proposed as a simplification of full PBPK models, allowing the estimation of interpretable parameters.
Objectives: To evaluate a minimal PBPK model in a population analysis framework, and to investigate the relationship between the parameters of an empirical 3-compartment PK model, and that of a minimal PBPK model.
Methods: Blood concentration-time profiles of drug S (an anticancer drug in development) were simulated in 1000 individuals, after a single 1-h infusion intravenous administration using a population mammillary 3-compartment model and an extensive measurement design. The model structure proposed in [1] (composed of a blood compartment and 2 distribution compartments) was implemented in NONMEM, then used to fit the data, with some parameters (cardiac output, blood volume, total body volume…) constrained to physiological values. The model was parameterized in terms of partition coefficients (Kp), tissue volumes, and fractions of blood flow. The constraints on parameters were implemented using logit-normally distributed random effects.
Results: The minimal PBPK model was successful in describing the data, and inter-individual variability was estimated on every parameter but one. As expected, blood clearance value was similar between the two models. The two distribution compartments’ volumes were 20 L and 61 L respectively, and with different Kp (65 and 1 respectively) and perfusion (19% and 28% of cardiac blood flow respectively). This shows a distribution of drug S in both extensively and poorly vascularized tissues, consistently with the large volume of distribution of the drug. A Kp of 65 in poorly vascularized tissues can be explained by a high affinity for adipose tissues (lipophilic compound). Distribution of the drug in extensively vascularized tissues is consistent with drug S anticancer activity (distribution in tumour).
Conclusions: This works illustrates how pharmacometricians can make good use of a minimal PBPK model in population PK analyses of rich data. These should be promoted in a general framework, as it allows better parameter interpretation, hypotheses generation, and assumption testing, than classical mammillary distribution models.
References:
[1] Text Cao Y. et al (2012) “Applications of minimal physiologically-based pharmacokinetic Models”, J Pharmacokinet Pharmacodyn.
[2] Cao Y. et al (2013) “Second-generation minimal physiologically-based pharmacokinetic model for monoclonal antibodies”, J Pharmacokinet Pharmacodyn.
[3] Tsamandouras N. et al (2015) “Combining the ‘bottom up’ and ‘top down’ approaches in pharmacokinetic modelling: fitting PBPK models to observed clinical data”, Br J Clin Pharmacol
Reference: PAGE 24 (2015) Abstr 3358 [www.page-meeting.org/?abstract=3358]
Poster: Drug/Disease modeling - Absorption & PBPK