Drug metabolism in animal models and humans: Translational aspects and chances for individual therapy

Abstract

Basic research in the biomedical field works with models from the subcellular up to the organism level with demands to translate them into the human situation. Starting from subcellular systems and ending with animal models, scientists are confronted with increasing complexity hampering the interpretation of their results.1 Therefore, it is important to recognize the differences between the experimental models and the situation in humans to design and evaluate the test system accordingly. For testing of drugs, drug candidates as well as toxic substances, the animal model is still an indispensable and frequently used intermediate step.2-4 Its relevance has been shown in simulations of cardiovascular diseases,5, 6 diabetes7, 8 or kidney diseases.9, 10 However, limitations are evident for pathophysiological conditions that are hard to mimic in animal models, for example Alzheimer's disease. Differences in physiology between rodents and humans are obvious and well studied. However, the pharmacokinetic aspect is often not given sufficient attention. Especially in basic physiological research, little attention is paid to metabolism and data on it is rarely collected. The pharmacokinetics of tested substances can differ greatly between different species. Thus, knowledge of metabolism is a key factor in finding a suitable test system in advance and, if necessary, in critically questioning the results obtained. To tackle that issue first, let us look at the situation in humans, as metabolism already varies greatly between individuals. Age, gender, ethnics and pre-existing illnesses have a major impact.11 The range and concentration of enzymes in the liver change significantly, especially in the first few years of life, and there are also gender-specific differences in the expression of enzymes.12 Above all, polymorphisms of drug-metabolizing enzymes lead to increased side effects or the absence of an effect depending on the genotype. This can be impressively demonstrated by means of rare serious side effects. In the therapy of tuberculosis, the WHO recommends the standard therapy consisting of the four drugs as follows: isoniazid, rifampicin, pyrazinamide and ethambutol. Serious side effects that occur in some cases during the therapy are related to the enzyme N-acetyltransferase 2. Patients with two low-active alleles of N-acetyltransferase 2 are significantly more likely to suffer hepatic toxicity, neurotoxic effects and systemic lupus erythematosus because of reduced isoniazid degradation.13 The genetic interindividual differences can be visualized with the help of a Hardy–Weinberg equilibrium. This is a mathematical simulation, which reflects the distribution of alleles in an ideal population without the influence of evolution. The Hardy–Weinberg equilibrium makes it possible to calculate ideal drug dosages depending on the number and the different variants of the metabolizing genes. When looking at the activity of enzymes as a dependent function of the genotype and thus the alleles, a special phenomenon becomes apparent. Alleles show a co-dominance for the enzymatic metabolism of xenobiotic substances. In the recessive inheritance of disease, carriers of a homozygous variant with two healthy alleles do not differ in phenotype from carriers of a heterozygous variant with only one healthy allele. In the case of xenobiotic-metabolizing enzymes, phenotypes from poor metabolizer over intermediate metabolizer, extensive metabolizer to ultra-rapid metabolizer can be observed depending on the allele combination (Figure 1).14 The vast majority of people in a given population are extensive metabolizers. Loss of function alleles (poor metabolizers) and gene duplications (ultra-rapid metabolizers) are less frequent than the other phenotypes but have the greatest clinical impact. Since polymorphism can occur in many different metabolizing enzymes, extreme variants in an enzyme critical for xenobiotic metabolism are not all that uncommon. The complexity of a Hardy–Weinberg equilibrium increases with the number of enzymes involved. For each enzyme encoding gene, which plays a role in the respective metabolic pathway, the allele variants with their phenotype and thus with their clinical outcome are included in the calculation. The result is a distribution with the respective probabilities for the occurrence of certain combinations of the respective enzyme variants (Figure 2).15 The genotype distribution can then be correlated with a theoretical optimal dosage of a drug. However, since other parameters such as co-medication, pre-existing diseases, polymorphisms in transporters, etc must also be taken into account, the correlation of dosage and metabolism is only a small piece of the puzzle for individual therapy. Several projects have been created to compile these pharmacogenetic correlations in online databases. Progress in digitization allows access to large numbers of drugs with their metabolism as well as the influence of different gene variants (eg, see Pharmacogenomics Knowledgebase [pharmgkb.org]).16 The aim of these databases is to list a constantly expanding library of drugs with their respective known metabolic pathways. If the genotype is known, it is also possible to enter the respective data of the specific enzyme variants into an online tool to receive a therapy and dosage recommendation for a drug based on the findings. Considering the various combinations of metabolizing enzymes with different genotypes, it quickly becomes clear that targeted therapies are a great challenge. However, the relevance of these theoretical considerations only becomes apparent when a serious side effect occurs, because of an undetected genotype that is unfavourable for the chosen treatment. Genotyping in medicine is not a dream of the future, as it is already being used in therapy. Abacavir, a reverse transcriptase inhibitor, is a drug used for HIV treatment. Unfortunately, in some cases, a life-threatening hypersensitivity reaction occurs after approximately 4 weeks. Carriers of the HLA-B*5701 polymorphism experience such an event with a 23-fold higher probability than patients without polymorphism. Since 5% of the European population are carriers of this variant, genotyping before starting the therapy is mandatory to minimize the risk.17 This is one of the many current examples of why genotyping has entered the world of medicine and can certainly be a valuable tool for better drug selection and dosing in the future.18 Drugs are usually affected by metabolism to chemically modify the substance so that it can be excreted by the body. In this process, activation, inactivation or toxification of the pharmacologically active substance occurs depending on the drug. Prodrugs represent an inactive form that is only converted into the active form by metabolism. In this case, high metabolic activity ensures high bioavailability. In addition, there are several pharmacologically active substances whose metabolites continue to be pharmacologically active or have a different mode of action. A clinical example of this is the conversion of codeine to morphine via CYP2D6. Codeine has an antitussive effect as a weak opioid but can be converted via metabolism into the much more potent morphine. A polymorphism in the gene for the CYP2D6 enzyme ensures that people of different ethnic backgrounds experience widely differing effects when using codeine. While about 8.9% of the British population are poor metabolizers and accordingly have a low conversion to morphine, about 20% of the people in Saudi Arabia are ultra-rapid metabolizers, some of whom suffer from morphine intoxication.19, 20 Another metabolic way is the toxification of drugs. Probably the most prominent example is the toxification of paracetamol. In vivo, paracetamol is converted into reactive quinoneimine, which is detoxified with the help of glutathione. Once the glutathione reserves are exhausted, the quinoneimine reacts with various compounds in the liver. This can lead to liver failure and consequently to death. Misuses and overdosage of paracetamol is the leading cause for acute liver failure in the United States and Europe.21 If a drug, after being absorbed through the intestine, is transported from the portal vein to the liver, where it is predominantly converted into inactive metabolites, this drug is subject to the first-pass effect. These considerations play a role in animal experiments too. Rapid inactivation via the first-pass effect is already generally considered to adapt the mode of application accordingly in the animal model. Here, intravenous22 and intramuscular injection offers some advantages in safe dosing and bioavailability in animal models and is therefore frequently used. The question that arises from this high variability in drug metabolism is How can these processes be best represented in a model? The first step should be extensive research of the substances to be used in the experiments. The focus should be on pharmacokinetic data on bioavailability, mode of application, solubility and, of course, the type and extent of metabolism, including metabolites. Especially, when it comes to already approved drugs, the data on bioavailability and metabolism are usually available. In the context of toxic substances and non-approved pharmacologically active substances, this is not always the case. Therefore, collecting own data via experiments can be useful and sometimes necessary. One way to circumvent these problems with the individual- and species-related variants in drug metabolism and to enable the translation of result to the human is to use humanized transgenic mice. Here, individual genes relevant for the metabolism (typically the main variants of the cytochrome P450 isoenzymes) are transferred to the mouse to obtain a more exact picture of human biotransformation.23 This method avoids many problems of the animal model. However, the whole procedure is time-consuming. Therefore, humanized mice are mainly used in investigations of drug candidates in pharmaceutical development. For basic physiological research, their application is not practically feasible. Thus, a fundamental understanding of biotransformation in the animal model is always necessary. In the context of animal experiments, CYP2D6 is noteworthy because it is the only known human CYP2D enzyme. In contrast, the mouse expresses more than six different representatives of this family, namely CYP2D9, CYP2D10, CYP2D11, CYP2D12, CYP2D13 and CYP2D22.24 If we look at the overall distribution of cytochrome P450 families, we find that mice have a total of 34 different families and genes that code for 103 different isoenzymes, while humans only have 18 families and 57 isoenzymes. This alone strongly illustrates the dilemma. The cytochrome P450 enzymes are responsible for about 70%-80% of the enzymatic reactions in xenobiotic metabolism and for a variety of endogenous enzymatic reactions,25 as they are responsible for the majority of the oxygenations.26 Comparing the mice and rats to humans, it becomes apparent that not only do the essential enzymes of the cytochrome P450 family differ but also the relative metabolism rate does. The extrapolation of pharmacokinetic data is not possible without further ado, since rodents have a larger liver in correlation with body size and body weight and therefore have a higher proportion of cytochrome P450 enzymes. Isoforms in humans as well as in mice or rats have highly conserved regions but also small differences in the amino acid sequence.26 This, however, can already have a significant impact on substrate specificity.27 The positive aspect is that 35 out of 57 isoenzymes are of clinical relevance in humans. When breaking it down to the central families, we can limit ourselves to the CYP1, CYP2 and CYP3 families in most cases. Among these three enzyme families, the isoenzyme CYP3A4 is certainly the most important representative. This enzyme in particular shows great similarities between male rats, mice and humans so that substrates that are metabolized exclusively by CYP3A4 can be simulated in the animal model.28 However, CYP3A4 is also strongly inhibitable and inducible, so that the application of several substances can be critical. In addition to animal models, isolated mouse and rat cytochrome P450 isoenzymes are commercially available, making it possible to test critical substances in advance in vitro. It must also be pointed out that the metabolism neither in humans nor in animals is fully elucidated. Several enzymes that are involved in human xenobiotic metabolism are often not given sufficient attention, because of the strong focus on cytochrome P450 enzymes.29, 30 Furthermore, in the case of many substances, the occurrence of metabolites has been described without knowing the exact correlation in the biotransformation. The diversity in enzyme expression and activity in animal models requires a careful investigation of the substance/drug applied in the experiments. A basic understanding of the metabolism in combination with analyses of metabolites in urine, faeces and plasma31 are necessary for serious interpretation of effects or side effects. Further, such a procedure will reduce the variability of results because of polymorphism at a critical point in metabolism or altered kidney, liver or intestinal function. If these findings from the animal model are combined with the results from cell culture32 and recombinant enzymes, basic research can certainly better represent the processes in humans. Individual therapy is an exciting medical milestone of the 21st century and will become more and more important. Understanding the underlying mechanisms is already a valuable tool in many scientific disciplines. Further progress in individualized therapy may be the combination of an interdisciplinary understanding of physiological principles, statistics and clinical implications. None.

Topics & Concepts