Natalie Filmann and Eva Herrmann
Goethe-University Frankfurt
Objectives. We have developed and evaluated virus kinetics models for the HBsAg-kinetics and the anti-HBs PK after liver transplantation based on existing models for chronic hepatitis B and the general Target Mediated Drug Disposition Model (TMDD-Model) [1, 2, 3, 4]. Model parameters were estimated by non-linear fitting of individual patient data. Individual fitting was used because we have focused on model building for a small, but detailed data set consisting of 18 patients who exhibit heterogeneous complex dynamics.
Aim of this study is to compare the individual and the population approach in this context.
Methods. Model parameters were estimated by non-linear fitting with MATLAB (R 2010a) using a maximum likelihood approach for individual fitting and using the population approach using nonlinear mixed-effect models in Monolix (version 4.3.3) by the SAEM algorithm, respectively.
The more detailed dataset 1 (n=18) was originally collected in the context of a prospective study [5] and was used for model building due to its richness. Datasets 2 (n=25) and 3 (n=26) were collected retrospectively and were used for model validation.
Results. The population approach leads to more stable estimates.
The individual approach allows fitting details in the observed kinetics that were exhibited by some patients and assumed to be important for model building. Fitting these details seems difficult with the population approach. A limitation of the population approach is that assumptions concerning the distributions of the residuals and random errors apply, which might be difficult to test for small datasets.
Conclusion. The individual approach seems to be more adequate for model building, whereas the population approach leads to more stable estimates and provides a valuable tool for post-hoc analyses and further validation of the model.
References:
[1] M. A. Nowak, S. Bonhoeffer, A. M. Hill, R. Boehme, H. C. Thomas and H. McDade. Viral dynamics in hepatitis B virus infection. Proceedings of the National Academy of Sciences, 93:4398–402, 1996.
[2] D. E. Mager and W. J. Jusko. General pharmacokinetic model for drugs exhibiting target mediated drug disposition. Journal of Pharmacokinetics and Pharmacodynamics, 28: 507–32, 2001.
[3] N. Filmann, J. Rosenau, M. P. Manns, H. Wedemeyer, E. Herrmann. Modeling the PK/PD of hepatitis B immunoglobulin after hepatitis B induced liver transplantation by an extended Target Mediated Drug Disposition Model. Presented at oral presentation at the 24th meeting of the Population Approach Group in Europe ; 2015 Jun 2-5; Hersonnisos, Crete, Greece. Available from www.page-meeting.org/?abstract=3454
[4] N. Filmann, J. Rosenau, M. P. Manns, H. Wedemeyer, E. Herrmann. Modeling the PK/PD of Hepatitis B immune globulin after Hepatitis B-induced liver transplantation by an extended Target Mediated Drug Disposition Model. [submitted]
[5] J. Rosenau, T. Kreutz, M. Kujawa, M. Bahr, K. Rifai, N. Hooman, A. Finger, G. Michel, B. Nashan, E. R. Kuse, J. Klempnauer, H. L. Tillmann and M. P. Manns. HBsAg level at time of liver transplantation determines HBsAg decrease and anti-HBs increase and affects HBV DNA decrease during early immunoglobulin administration. Journal of Hepatology, 46:635–44, 2007.
Reference: PAGE 25 (2016) Abstr 5970 [www.page-meeting.org/?abstract=5970]
Poster: Drug/Disease modeling - Infection