Models of Idiosyncratic Drug-Induced Liver Injury

Drug-induced liver injury (DILI) is a leading cause of attrition during the early and late stages of drug development and after a drug is marketed. DILI is generally classified as either intrinsic or idiosyncratic. Intrinsic DILI is dose dependent and predictable (e.g., acetaminophen toxicity). However, predicting the occurrence of idiosyncratic DILI, which has a very low incidence and is associated with severe liver damage, is difficult because of its complex nature and the poor understanding of its mechanism. Considering drug metabolism and pharmacokinetics, we established experimental animal models of DILI for 14 clinical drugs that cause idiosyncratic DILI in humans, which is characterized by the formation of reactive metabolites and the involvement of both innate and adaptive immunity. On the basis of the biomarker data obtained from the animal models, we developed a cell-based assay system that predicts the potential risks of drugs for inducing DILI. These findings increase our understanding of the mechanisms of DILI and may help predict and prevent idiosyncratic DILI due to certain drugs.

Cancer Hazard Identification Integrating Human Variability: The Case of Coumarin

Coumarin is a naturally occurring sweet-smelling benzopyrone that may be extracted from plants or synthesized for commercial uses. Its uses include as a flavoring agent, fragrance enhancer, and odor-masking additive. We reviewed and evaluated the scientific evidence on the carcinogenicity of coumarin, integrating information from carcinogenicity studies in animals with mechanistic and other relevant data, including data from toxicogenomic, genotoxicity, and metabolism studies, and studies of human variability of a key enzyme, CYP2A6. Increases in tumors were observed in multiple studies in rats and mice in multiple tissues. Our functional pathway analysis identified several common cancer-related biological processes/pathways affected by coumarin in rat liver following in vivo exposure and in human primary hepatocytes exposed in vitro. When coumarin 7-hydroxylation by CYP2A6 is compromised, this can lead to a shift in metabolism to the 3,4-epoxidation pathway and increased generation of electrophilic metabolites. Mechanistic data align with 3 key characteristics of carcinogens, namely formation of electrophilic metabolites, genotoxicity, and induction of oxidative stress. Considerations of metabolism, human variability in CYP2A6 activity, and coumarin hepatotoxicity in susceptible individuals provide additional support for carcinogenicity concern. Our analysis illustrates the importance of integrating information on human variability in the cancer hazard identification process.

Isoflurane is a suitable alternative to ether for anesthetizing rats prior to euthanasia for gene expression analysis

Diethyl ether (ether) had been widely used in Japan for anesthesia, despite its explosive properties and toxicity to both humans and animals. We also had used ether as an anesthetic for euthanizing rats for research in the Toxicogenomics Project (TGP). Because the use of ether for these purposes will likely cease, it is required to select an alternative anesthetic which is validated for consistency with existing TGP data acquired under ether anesthesia. We therefore compared two alternative anesthetic candidates, isoflurane and pentobarbital, with ether in terms of hematological findings, serum biochemical parameters, and gene expressions. As a result, few differences among the three agents were observed. In hematological and serum biochemistry analysis, no significant changes were found. In gene expression analysis, four known genes were extracted as differentially expressed genes in the liver of rats anesthetized with ether, isoflurane, or pentobarbital. However, no significant relationships were detected using gene ontology, pathway, or gene enrichment analyses by DAVID and TargetMine. Surprisingly, although it was expected that the lung would be affected by administration via inhalation, only one differentially expressed gene was extracted in the lung. Taken together, our data indicate that there are no significant differences among ether, isoflurane, and pentobarbital with respect to effects on hematological parameters, serum biochemistry parameters, and gene expression. Based on its smallest affect to existing data and its safety profile for humans and animals, we suggest isoflurane as a suitable alternative anesthetic for use in rat euthanasia in toxicogenomics analysis.

Data-intensive drug development in the information age: applications of Systems Biology/Pharmacology/Toxicology

Pharmaceutical companies continuously face challenges to deliver new drugs with true medical value. R&D productivity of drug development projects depends on 1) the value of the drug concept and 2) data and in-depth knowledge that are used rationally to evaluate the drug concept's validity. A model-based data-intensive drug development approach is a key competitive factor used by innovative pharmaceutical companies to reduce information bias and rationally demonstrate the value of drug concepts. Owing to the accumulation of publicly available biomedical information, our understanding of the pathophysiological mechanisms of diseases has developed considerably; it is the basis for identifying the right drug target and creating a drug concept with true medical value. Our understanding of the pathophysiological mechanisms of disease animal models can also be improved; it can thus support rational extrapolation of animal experiment results to clinical settings. The Systems Biology approach, which leverages publicly available transcriptome data, is useful for these purposes. Furthermore, applying Systems Pharmacology enables dynamic simulation of drug responses, from which key research questions to be addressed in the subsequent studies can be adequately informed. Application of Systems Biology/Pharmacology to toxicology research, namely Systems Toxicology, should considerably improve the predictability of drug-induced toxicities in clinical situations that are difficult to predict from conventional preclinical toxicology studies. Systems Biology/Pharmacology/Toxicology models can be continuously improved using iterative learn-confirm processes throughout preclinical and clinical drug discovery and development processes. Successful implementation of data-intensive drug development approaches requires cultivation of an adequate R&D culture to appreciate this approach.

Evaluation of DNA microarray results in the Toxicogenomics Project (TGP) consortium in Japan

An important technology used in toxicogenomic drug discovery research is the microarray, which enables researchers to simultaneously analyze the expression of a large number of genes. To build a database and data analysis system for use in assessing the safety of drugs and drug candidates, in 2002 we conducted a 5-year collaborative study in the Toxicogenomics Project (TGP1) in Japan. Experimental data generated by such studies must be validated by different laboratories for robust and accurate analysis. For this purpose, we conducted intra- and inter-laboratory validation studies with participating companies in the second collaborative study in the Toxicogenomics Project (TGP2). Gene expression in the liver of rats treated with acetaminophen (APAP) was independently examined by the participating companies using Affymetrix GeneChip microarrays. The intra- and inter-laboratory reproducibility of the data was evaluated using hierarchical clustering analysis. The toxicogenomics results were highly reproducible, indicating that the gene expression data generated in our TGP1 project is reliable and compatible with the data generated by the participating laboratories.

Toxicogenomics discrimination of potential hepatocarcinogenicity of non-genotoxic compounds in rat liver

Long-term carcinogenicity testing of a compound is exceedingly time-consuming and costly, and requires many test animals, whereas the Ames test, which is based on the assumption that any substance that is mutagenic may also exert carcinogenic potential, is useful as a short-term screening assay but has major drawbacks. Although, in fact, 90% of compounds that give a positive Ames test cause cancer in laboratory animals, a good proportion of compounds that give a negative Ames test are also carcinogens; that is, there is no good correlation between carcinogenicity and negative Ames test results. As an alternative to these two approaches, we have tried applying toxicogenomics to predict the carcinogenicity of a compound from the gene expression profile induced in vivo. To establish our model, male Sprague-Dawley rats were orally administered test compounds (12 hepatocarcinogens and 26 non-hepatocarcinogens) for 28 days. Analysis of liver gene expression data by Support Vector Machines (SVM) dividing compounds into 'for training' and 'for test' (20 cases assigned randomly) allowed a set of marker genes to be tested for prediction of hepatocarcinogenicity. The developed prediction model was then validated with reference to the concordance rate with training data and test data, and a good performance was obtained. We will have new gene expression data and continue the validation of our model.

Predicting drug-induced hepatotoxicity using QSAR and toxicogenomics approaches

Quantitative structure-activity relationship (QSAR) modeling and toxicogenomics are typically used independently as predictive tools in toxicology. In this study, we evaluated the power of several statistical models for predicting drug hepatotoxicity in rats using different descriptors of drug molecules, namely, their chemical descriptors and toxicogenomics profiles. The records were taken from the Toxicogenomics Project rat liver microarray database containing information on 127 drugs ( http://toxico.nibio.go.jp/datalist.html ). The model end point was hepatotoxicity in the rat following 28 days of continuous exposure, established by liver histopathology and serum chemistry. First, we developed multiple conventional QSAR classification models using a comprehensive set of chemical descriptors and several classification methods (k nearest neighbor, support vector machines, random forests, and distance weighted discrimination). With chemical descriptors alone, external predictivity (correct classification rate, CCR) from 5-fold external cross-validation was 61%. Next, the same classification methods were employed to build models using only toxicogenomics data (24 h after a single exposure) treated as biological descriptors. The optimized models used only 85 selected toxicogenomics descriptors and had CCR as high as 76%. Finally, hybrid models combining both chemical descriptors and transcripts were developed; their CCRs were between 68 and 77%. Although the accuracy of hybrid models did not exceed that of the models based on toxicogenomics data alone, the use of both chemical and biological descriptors enriched the interpretation of the models. In addition to finding 85 transcripts that were predictive and highly relevant to the mechanisms of drug-induced liver injury, chemical structural alerts for hepatotoxicity were identified. These results suggest that concurrent exploration of the chemical features and acute treatment-induced changes in transcript levels will both enrich the mechanistic understanding of subchronic liver injury and afford models capable of accurate prediction of hepatotoxicity from chemical structure and short-term assay results.

Practical application of toxicogenomics for profiling toxicant-induced biological perturbations

A systems-level understanding of molecular perturbations is crucial for evaluating chemical-induced toxicity risks appropriately, and for this purpose comprehensive gene expression analysis or toxicogenomics investigation is highly advantageous. The recent accumulation of toxicity-associated gene sets (toxicogenomic biomarkers), enrichment in public or commercial large-scale microarray database and availability of open-source software resources facilitate our utilization of the toxicogenomic data. However, toxicologists, who are usually not experts in computational sciences, tend to be overwhelmed by the gigantic amount of data. In this paper we present practical applications of toxicogenomics by utilizing biomarker gene sets and a simple scoring method by which overall gene set-level expression changes can be evaluated efficiently. Results from the gene set-level analysis are not only an easy interpretation of toxicological significance compared with individual gene-level profiling, but also are thought to be suitable for cross-platform or cross-institutional toxicogenomics data analysis. Enrichment in toxicogenomics databases, refinements of biomarker gene sets and scoring algorithms and the development of user-friendly integrative software will lead to better evaluation of toxicant-elicited biological perturbations.

The Japanese toxicogenomics project: application of toxicogenomics

Biotechnology advances have provided novel methods for the risk assessment of chemicals. The application of microarray technologies to toxicology, known as toxicogenomics, is becoming an accepted approach for identifying chemicals with potential safety problems. Gene expression profiling is expected to identify the mechanisms that underlie the potential toxicity of chemicals. This technology has also been applied to identify biomarkers of toxicity to predict potential hazardous chemicals. Ultimately, toxicogenomics is expected to aid in risk assessment. The following discussion explores potential applications and features of the Japanese Toxicogenomics Project.

Nimesulide aggravates redox imbalance and calcium dependent mitochondrial permeability transition leading to dysfunction in vitro

Nimesulide (selective cyclooxygenase-2 inhibitor) is a nonsteroidal anti-inflammatory drug for the symptomatic treatment of painful conditions like osteoarthritis, spondilitis and primary dysmenorrhoea. Nimesulide induced liver damage is a serious side effect of this otherwise popular drug. The mechanism involved in nimesulide induced hepatotoxicity is still not fully elucidated. However, both mitochondrial dysfunction and oxidative stress have been implicated in contributing to liver injury in susceptible patients. Mitochondria besides being the primary source of energy, act as a hub of signals responsible for initiating cell death, irrespective of the pathway, i.e. apoptosis or necrosis. The present study was aimed to explore the role of compounding stress, i.e. Ca(2+) overload and GSH depletion in nimesulide induced mitochondrial toxicity and dysfunction. Our study showed that, nimesulide (100 microM) treatment resulted into rapid depletion of GSH (60%) in isolated rat liver mitochondria and significant Ca(2+) dependent MPT changed. Enhanced ROS generation (DCF fluorescence) was also observed in mitochondria treated with nimesulide. An important finding was that the concentration at which nimesulide oxidized reduced pyridine nucleotides (autofluorescence of NAD(P)H), it affected mitochondrial electron flow (MTT activity decreased by 75%) and enhanced mitochondrial depolarization significantly as assessed by Rhodamine 123 fluorescent probe. Therefore, nimesulide was found to aggravate redox imbalance and affect Ca(2+) dependent mitochondrial membrane permeability transition leading to dysfunction and ultimately cell death.

Biomarkers for drug-induced liver injury

Of the estimated 10,000 documented human drugs, more than 1000 have been associated with drug-induced liver injury (DILI), although causality has not always been established clearly. Numerous biomarkers for DILI have been explored, but less than ten are adopted or qualified as valid by the US FDA. The biomarkers for DILI are individual or a panel of proteins, nucleic acids or metabolites from various sources, such as the liver, blood and urine. While most DILI biomarkers are drug independent, some possibly 'drug-specific' DILIs have been explored, but specificity and sensitivity of both types need to be improved for the diagnosis of DILI during drug development and in clinical practice. Novel approaches for DILI biomarkers have been actively investigated recently, but produced mainly animal-based biomarkers, which are possibly useful for drug development, but are not suitable or have not been validated for clinical applications. This review summarizes the current practice and future perspectives for DILI biomarkers.

Gene expression profiling in rat liver treated with compounds inducing elevation of bilirubin

We have constructed a large-scale transcriptome database of rat liver treated with various drugs. In an effort to identify a biomarker for the diagnosis of elevated total bilirubin (TBIL) and direct bilirubin (DBIL), we extracted 59 probe sets of rat hepatic genes from the data for seven typical drugs, gemfibrozil, phalloidin, colchicine, bendazac, rifampicin, cyclosporine A, and chlorpromazine, which induced this phenotype from 3 to 28 days of repeated administration in the present study. Principal component analysis (PCA) using these probes clearly separated dose- and time-dependent clusters in the treated groups from their controls. Eighteen more drugs in the database, reported to elevate TBIL and DBIL, were estimated by PCA using these probe sets. Of these, 12 drugs, that is methapyrilene, thioacetamide, ticlopidine, ethinyl estradiol, alpha-naphthylisothiocyanate, indomethacin, methyltestosterone, penicillamine, allyl alcohol, aspirin, iproniazid, and isoniazid were also separated from the control clusters, as were the seven typical drugs causing elevation of TBIL and DBIL. The principal component 1 (PC1) value showed high correlation with TBIL and DBIL. In the cases of colchicine, bendazac, chlorpromazine, gemfibrozil, and phalloidin, the possible elevation of TBIL and DBIL could be predicted by expression of these genes 24 h after single administration. We conclude that these identified 59 probe sets could be useful to diagnose the cause of elevation of TBIL and DBIL, and that toxicogenomics would be a promising approach for prediction of this type of toxicity.

Gene expression profiling in rat liver treated with various hepatotoxic-compounds inducing coagulopathy

A large-scale transcriptome database of rat liver (TG-GATEs) has been established by the Toxicogenomics Project in Japan. In the present study, we focused on 8 hepatotoxic compounds within TG-GATEs, i.e., clofibrate, omeprazole, ethionine, thioacetamide, benzbromarone, propylthiouracil, Wy-14,643 and amiodarone, which induced coagulation abnormalities. Aspirin was selected as a reference compound that directly causes coagulation abnormality, but not through liver toxicity. In blood chemical examinations, for all the coagulopathic compounds there was little elevation of aspartate aminotransferase (AST) and/or alanine aminotransferase (ALT), suggesting no severe cell death by treatment with the compounds. We extracted 344 probe sets from the data for these 8 typical drugs, which induced this phenotype at any time from 3 to 28 days of repeated administration. Principal component analysis using these probe sets clearly separated dose- and time-dependent clusters of the treated groups from their controls, except aspirin and propylthiouracil, both of which were considered to cause coagulopathy not due to their hepatotoxicity but due to their direct effects on the blood coagulation system. Reviewing the extracted genes, changes in lipid metabolism were found to be dominant. Genes related to blood coagulation were generally down-regulated by these drugs except that vitamin K epoxide reductase complex subunit 1 (Vkorc1) like 1, a paralogous gene of Vkorc1, was up-regulated. As expected, expression changes of these genes were least prominent in aspirin or propylthiouracil-treated liver. We concluded that these probe sets could be a good starting point in developing mechanism-based biomarkers for diagnosis or prognosis of hepatotoxicity-related coagulation abnormalities in the early stage of drug development.

Toxicogenomic biomarkers for liver toxicity

Toxicogenomics (TGx) is a widely used technique in the preclinical stage of drug development to investigate the molecular mechanisms of toxicity. A number of candidate TGx biomarkers have now been identified and are utilized for both assessing and predicting toxicities. Further accumulation of novel TGx biomarkers will lead to more efficient, appropriate and cost effective drug risk assessment, reinforcing the paradigm of the conventional toxicology system with a more profound understanding of the molecular mechanisms of drug-induced toxicity. In this paper, we overview some practical strategies as well as obstacles for identifying and utilizing TGx biomarkers based on microarray analysis. Since clinical hepatotoxicity is one of the major causes of drug development attrition, the liver has been the best documented target organ for TGx studies to date, and we therefore focused on information from liver TGx studies. In this review, we summarize the current resources in the literature in regard to TGx studies of the liver, from which toxicologists could extract potential TGx biomarker gene sets for better hepatotoxicity risk assessment.

Scoring multiple toxicological endpoints using a toxicogenomic database

As information regarding microarray data sets and toxicogenomic biomarkers grows rapidly, the process of analyzing data and interpreting the results is increasingly complicated. To facilitate data analysis, a simple expression ratio-based scoring method called the TGP1 score was previously proposed [Kiyosawa, N., Shiwaku, K., Hirode, M., Omura, K., Uehara, T., Shimizu, T., Mizukawa, Y., Miyagishima, T., Ono, A., Nagao, T., Urushidani, T., 2006. Utilization of a one-dimensional score for surveying chemical-induced changes in expression levels of multiple biomarker gene sets using a large-scale toxicogenomics database. J. Toxicol. Sci. 31, 433-448]. Although the TGP1 score has demonstrated its efficacy for rapid comprehension of large-scale toxicogenomic data sets, inclusion of low quality gene expression data in the biomarker gene set produced flaws in the calculated score. To overcome this shortcoming, we tested a new scoring method called the differentially expressed gene score (D-score), where Detection Call as well as signal log ratios generated by MAS5 algorithm on Affymetrix GeneChip data were considered for the calculation. Four prototypical toxicants, namely acetaminophen, phenobarbital, clofibrate and acetamidofluorene, were used for detailed analysis. A toxicogenomics database (TG-GATEs) was utilized as a reference data set. The D-score successfully alleviated the effects of low quality data on the score calculation, and captured the overall direction of expression changes as well as the magnitude of expression change level of a set of genes, highlighting the affected toxicological endpoints elicited by chemical treatment. The D-score will be useful for high-throughput toxicity screening using a toxicogenomic database and biomarkers.

Species-specific differences in coumarin-induced hepatotoxicity as an example toxicogenomics-based approach to assessing risk of toxicity to humans

One expected result from toxicogenomics technology is to overcome the barrier because of species-specific differences in prediction of clinical toxicity using animals. The present study serves as a model case to test if the well-known species-specific difference in the toxicity of coumarin could be elucidated using comprehensive gene expression data from rat in-vivo, rat in-vitro, and human in-vitro systems. Coumarin 150 mg/kg produced obvious pathological changes in the liver of rats after repeated administration for 7 days or more. Moreover, 24 h after a single dose, we observed minor and transient morphological changes, suggesting that some early events leading to hepatic injury occur soon after coumarin is administered to rats. Comprehensive gene expression changes were analyzed using an Affymetrix GeneChip approach, and differentially expressed probe sets were statistically extracted. The changes in expression of the selected probe sets were further examined in primary cultured rat hepatocytes exposed to coumarin, and differentially expressed probe sets common to the in-vivo and in-vitro datasets were selected for further study. These contained many genes related to glutathione metabolism and the oxidative stress response. To incorporate human data, human hepatocyte cultured cells were exposed to coumarin and changes in expression of the bridging gene set were examined. In total, we identified 14 up-regulated and 11 down-regulated probe sets representing rat-human bridging genes. The overall responsiveness of these genes to coumarin was much higher in rats than humans, consistent with the reported species difference in coumarin toxicity. Next, we examined changes in expression of the rat-human bridging genes in cultured rat and human hepatocytes treated with another hepatotoxicant, diclofenac sodium, for which hepatotoxicity does not differ between the species. Both rat and human hepatocytes responded to the marker genes to the same extent when the same concentrations of diclofenac sodium were exposed. We conclude that toxicogenomics-based approaches show promise for overcoming species-specific differences that create a bottleneck in analysis of the toxicity of potential therapeutic treatments.