Prediction of pulmonary exposure based on plasma pharmacokinetics: A comparison of different model-based approaches.
Anneke Himstedt (1,2), Jens M. Borghardt (2), Sebastian G. Wicha (1)
(1) Dept. of Clinical Pharmacy, Institute of Pharmacy, University of Hamburg, Hamburg, Germany, (2) Drug Discovery Sciences, Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Germany
Introduction: Different model-based approaches were applied to characterize the pharmacokinetics (PK) of orally inhaled drugs, differing prominently in the structural representation of pulmonary absorption processes. While it has been theorized that these empirical models can be used to infer on the pulmonary PK based on plasma data after both intravenous and inhalational administration [1], the predictions and conclusions drawn from these approaches have never been systematically compared.
Objective: To compare five published empirical pulmonary absorption models of varying complexity with regard to the suitability of these models to correctly infer on lung exposure considering different pulmonary PK characteristics. These compartmental models comprised (i) three, (ii) two, (iii) one single (parallel) first-order absorption process(es), (iv) one single absorption process with simultaneous non-absorptive pulmonary loss, and (v) a transit absorption model.
Methods: The five models were built and parameterized based on the respective publications [2-6]. The models were used for a simulation/re-estimation analysis performed with R (Version 3.2.2) [7], employing the deSolve package (Version 1.20) [8]. Since the focus of this work lay on the comparison of pulmonary absorption models, only PK profiles after oral inhalation were considered and the systemic parameters were fixed to the published values during the estimation process. Furthermore, the oral absorption compartments were removed from the models.
The models were deemed exchangeable with regard to the characterization of systemic exposure if the newly predicted plasma concentration-time profiles deviated from the originally simulated data used for fitting by less than five percent. When plasma equivalence was given, the model simulations were compared with regard to lung exposure (measured as area under the lung concentration-time curve, AUC0-inf,Lung) and drug concentration in the lungs after 24 hours (C24h,Lung).
Results: All pulmonary absorption models, except for the one including three parallel absorption processes, were exchangeable regarding the plasma concentration-time profiles. The resulting predictions of lung exposure and lung concentration after 24 hours differed greatly and were not interchangeable even though the plasma profiles indicated so (values for AUC0-inf,Lung and C24h,Lung deviated over 6 and 8 orders of magnitude, respectively).
Conclusions: This simulation/re-estimation study showed that information on plasma PK alone is not sufficient to draw conclusions about the extent and duration of lung exposure using typically applied empirical models. However, it has to be noted that the physiological assumptions implied in the investigated models differ, so that prior knowledge about the pulmonary processes affecting the respective drug after oral inhalation might still allow for correct interpretation of model predictions.
In conclusion, additional quantitative information on the underlying physiological processes is crucial to carefully select an adequate model-based approach to infer on pulmonary PK from systemic exposure.
References:
[1] Borghardt JM et al. AAPS J (2015) 17:853-870.
[2] Borghardt JM et al. Br J Clin Pharmacol (2016) 81:538-552.
[3] Diderichsen PM et al. Clin Pharmacokinet (2013) 52:443-452.
[4] Krishnaswami S et al. Int J Clin Pharmacol Ther (2005) 43:117-122.
[5] Melin J et al. AAPS J (2017) 19:865-874.
[6] Sakagami M. Clin Pharmacokinet (2004) 43:539-552.
[7] https://www.R-project.org/
[8] http://www.jstatsoft.org/v33/i09/