Please visit the new BioModels platform to access the latest content. This website is no longer updated and will be retired on 31 May 2019.
BioModels Database logo

BioModels Database


Bucher et al., 2011. A systems biology approach to dynamic modeling and inter-subject variability of statin pharmacokinetics in human hepatocytes

February 2012, model of the month by Christine Seeliger
Original model: BIOMD0000000328

Developing and testing new pharmaceutical drugs is an expensive and time consuming process, since many drugs fail in quite late stages of the development process. Frequently, toxicity and poor pharmacokinetics are the reason for the failure. Pharmacokinetics - the interaction of the drug with the human organs, is an important property of a drug. In many cases, the already available in vitro data generated during the development of a new drug are integrated to prevent failure of drugs in later stages of the development process. Systems biology and especially modelling, can provide the necessary tools and methods to gather the already present knowledge at certain steps of the development process [1]. This can help to understand the underlying mechanisms of drug interactions with the organism to guide later stages of the development process.

The model presented by Bucher et al. (2011) [2, BIOMD0000000328], aims to develop a deterministic model of the interactions of the drug Atorvastatin with liver cells. It includes transport processes as well as detoxification steps and is used to study interindividual variability, which is another important aspect during the drug designing process. Atorvastatin (AS) is a synthetic drug belonging to the family of Statins. Statins are structurally similar to 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) that is reduced to mevalonate. This is the rate limiting step in cholesterol biosynthesis, catalyzed by the enzyme 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMG-CoA reductase) in hepatocytes. The result of statin treatment is mainly to decrease cholesterol levels in the blood. In addition, it increases the uptake of low-density lipoproteins (LDL) bound to cholesterol by hepatocytes. High levels of these cholesterol-bound LDL in the bloodstream are associated with many health problems and cardiovascular diseases, hence commonly called "bad cholesterol". Today, Atorvastatin is one of the major drugs used to treat cholesterol related dyslipidemias like hypercholesterolemia and the prevention of cardiovascular diseases, e.g. stroke prophylaxis in diabetes type II patients [3]. Despite these positive effects, Atorvastatin can exhibit negative side effects such as hepatotoxicity, myopathy or rhabdomyolosis. Drug-drug interactions should also be taken into account.

Figure 1

Figure 1: Schema of the modelled Atorvastatin metabolism. The model accounts for import (via OATP1B1, OATP2B1), export (via MDR1, MRP2) as well as passive diffusion (dashed) as well as para- and ortho-hydroxy-metabolites (ASpOH, ASoOH, ASLpOH, ASLoOH). Figure taken from [2].

Figure 2

Figure 2: CYP3A4 and UGT1A3 concentration in human liver microsomes of 150 individuals. Figure taken from [2].

Atorvastatin exists either as hydrophobic lactone (ASL) or hydrophilic acid (AS) in the cells. The model proposed by Bucher et al. (2011), takes both forms into account along with their interconversion and their ortho- and parahydroxy metabolites (ASoOH, ASpOH, ASLoOH, ASLpOH). However, lactonization of AS to ASL is negligible at physiological pH. The main enzymes catalyzing these reactions are UGT1A3 (lactonization) and CYP3A4 (hydroxylation). AS is imported into the cells via organic anion transport polypeptides (OATPs) and is considered to occur via an active mechanism rather than facilitated diffusion. Transport of AS, ASoOH, and ASpOH out of cells is considered to be ATP-dependent via MDR and MRP transporters. Due to the more hydrophobic properties of the lactones, passive diffusion might play a more important role for them, than for the acidic forms. The last effect considered in the model is unspecific binding, especially of the lipophilic forms to the intracellular and extracellular proteins, and to the cell surfaces. A scheme of the modelled Atorvastatin metabolism model is given in figure 1.

Model parameters were either fixed to values found in literature or estimated by minimizing the difference between experiments and model results. This optimization as well as the initial model validation was based on data obtained from hepatocyte cultures of a single individual.

Analysis of parameter sensitivity and identifiability, made it necessary to reduce the initial model by combining active transport and passive diffusion mechanisms, for export and import processes into apparent rates. This reduced model was used for simultaneous validation with data from two additional individuals. Whereas, the model was well in agreement with data from individual 2, it was not possible to find a good agreement between the model and individual 3, especially with regard to intracellular ASL and ASoOH and extracellular ASpOH. The hypothesis that individual 3's discrepancies stemmed from his/her type II diabetes mellitus and the influence of that on beta oxidation could not be verified, pointing to other differences in individual 3's metabolism and other effects that might not have been discovered yet.

The model with an optimized parameter set was linked to protein expression data obtained from 150 individuals to address interindividual variability. Figure 2 shows the measured variability with regard to the laconizing and hydroxylizing enzymes UGT1A3 and CYP3A4. This study addresses, how the observed interindividual variability influences the intracellular AS metabolites. Figure 3 shows the resulting probability density functions for AUC, cmax and t(cmax) for AS alone (solid) or the sum of AS, ASpOH and ASoOH (dashed). In all cases, the summed results peak later than AS alone due to the fact that ASpOH and ASoOH have to be synthesized from AS. Most of the fitted distributions have a narrow shape with a relative standard deviation of less then 50%.

The results undermine the possibility of an active uptake of AS or the presence of so far unknown transport mechanisms. Some discrepancies between clinical studies and the proposed model are pointed out by the authors. They might stem from the fact that the model is based on data obtained from isolated hepatocytes in cultures. Experiments in cultures are usually done on much shorter timescales than clinical studies as well. To overcome problems in measuring minimal concentrations, a high initial AS concentration was used that could also result in differences. The paper also addresses identifiable issues that resulted in a reduction of the model, pointing out the balance between model complexity, available data and experimental limits. Parameter errors in model development can be judged whether or not they are small enough to suit the requirements of the pharmacological application of the drug. The results of the dynamic analysis of the model based on CYP3A4 and UGT1A3 expression levels show a high individual variability that is important to be considered in drug design and application. This variability study has to be extended to other processes, e.g. OATP-C, one of the essential transporter proteins.

The model presented here, helps to understand the structure and function of the underlying metabolic network. It provides a clearer and structured view of the pathways and transport processes that are influenced by the drug AS. Studying the model enables researchers to understand the detailed mechanistics and the influence of individual variations that might influence the effect of the drug, hence facilitating and guiding drug testing and development. Including studies like this in drug development, can help to determine suitability of a drug much earlier in the development process and save money on expensive following studies. The more detailed mechanistic understanding that the model provides, could also help to find the problems with a specific drug and improving it, and speed up the development process.

Figure 3

Figure 3: Dynamic analysis of interindividual variability in AS metabolism. Fitted probability density functions for AS (dotted) and the the sum of AS, ASoOH and ASpOH (dashed). Figure taken from [2].

Bibliographic References

  1. Huisinga W, Telgmann R, Wulkow M. The virtual laboratory approach to pharmacokinetics: design principles and concepts. Drug Discov Today. Sep;11(17-18):800-5, 2006. [CiteXplore]
  2. Bucher J, Riedmaier S, Schnabel A, Marcus K, Vacun G, Weiss TS, Thasler WE, Nüssler AK, Zanger UM, Reuss M. A systems biology approach to dynamic modeling and inter-subject variability of statin pharmacokinetics in human hepatocytes. BMC Syst Biol. May 6;5:66, 2011. [CiteXplore]
  3. Law MR, Wald NJ, Rudnicka AR. Quantifying effect of statins on low density lipoprotein cholesterol, ischaemic heart disease, and stroke: systematic review and meta-analysis. BMJ. Jun 28;326(7404):1423, 2003. [CiteXplore]