National Capital Area SOT, Spring 2004 Symposium
Speaker Abstracts and Slide Presentations
Detecting Liver Injury: Drug-Induced or Not?
John R. Senior, M.D., FDA, Center for Drug Evaluation and Research
Correct attribution of causality as to whether a case of acute liver injury is drug-induced or not is among the more difficult problems in medicine. Drugs or other substances, or their metabolites produced by the liver, may cause liver injuries that mimic all known liver diseases. There is no pathognomonic test or finding, including biopsy, that proves the injury is drug-induced. The diagnosis therefore must be made by ruling out all other possible causes. This is often not done well or thoroughly, so that the data gathered too frequently allow concluding only that the liver injury was “possibly” or “probably” drug-induced.
We chose to address one aspect of this problem: whether we could be reasonably certain of detecting and diagnosing serious liver disease that was NOT drug-induced. The work was done as an innovative research collaboration between the Food and Drug Administration and the Merck Research Laboratories, where the clinical representative served not as a regulator but as a cooperative research partner with the pharmaceutical industry statistician. We examined a large database of 3301 study participants randomized to placebo and observed repeatedly (20 times) over more than 5 years.
We found that serum transaminase elevations were very poorly predictive of serious liver disease, were often transient and unexplained. Only when the transaminase elevations (3-fold) were persistent or progressive and either accompanied or followed by rises in serum bilirubin (2-fold) was the liver disease clinically significant or serious. The combination of a test for hepatocellular injury and a measure of overall liver function was as sensitive as the transaminase alone but far more specific in detecting 6 of 6 cases of serious liver disease incident among 3248 ambulatory participants on placebo for 5 years.
Thus, we could be reasonably sure we could detect serious non-drug-induced liver injury or disease by this combined injury-function test as a biomarker, the same combination that the late Hyman Zimmerman reported for drug-induced liver injury. The methods for exclusion of non-drug causes, and correct and quantitative attribution of causality, will require further work.
Biomarkers of Liver Toxicity and Disease
Yvonne Dragan, Ph.D., FDA, National Center for Toxicological Research
Hepatotoxicity is the number one reason for drug recall and hence it is of concern to the FDA and to the American public. Drug induced toxicities account for nearly half of the cases of acute liver failure in adults over age 50 in the U.S.. There is a need for better predictive tests for agents that have the potential to cause liver toxicity in humans. In addition, early biomarkers of adverse liver effects from exposure to agents are needed and should be developed. Factors that trigger hepatic cell death, permit hepatocyte survival, and trigger hepatocyte proliferation are potential biomarkers. Techniques such as Seldi based proteomics and NMR-based metabonomics provide tools for the development of biomarkers of toxicity and disease. Specific, sensitive, and predictive biomarkers for the induction of liver toxicity and the potential for progression to fulminant liver failure need to be identified and are the long term goal of this proposal.. Animal toxicity data can be used as a surrogate to predict potential human liver toxicity.
Predictive Hepatotoxicity: Correlations in a Structural Database
James H. Kelly, Ph.D., Amphioxus Cell Technologies
We have employed a human liver cell line and a series of high throughput in vitro toxicology assays to examine the structural basis of hepatotoxicity in several families of compounds. The data show that such a system can provide fine discrimination among structural variants. We have examined thiazolidinediones, nonsteroidal anti-inflammatories, fluoroquinolones, and antihistamines with known hepatotoxic properties to show that there is a clear structural basis for the problems encountered with these compounds. With this idea in mind, we are screening a large number of compounds in an attempt to build a predictive toxicology database which, when coupled to the experimental system, can be used to eliminate hepatotoxic liabilities early in the drug development process. The data are gathered directly into software for experiment management, MDL Assay Explorer®. The results are then collated to the associated structure and functional information for each compound using ChemBioAE®. Using cyp1A induction as an example, we are able to show that compounds with similar mechanisms of action, such as ERK inhibition, are common inducers. Likewise, unrelated compounds with similar structural features, such as aristolochic acid and apomorphine, also induce.
In Vitro Assessment of Necrosis and Apoptosis Using Human Hepatocytes
Neal Jensen, Ph.D., In Vitro Technologies
Hepatotoxicity is the most common side effect of new chemical entities (NCE). While hepatotoxicity is generally described as a necrotic event, apoptosis may also play an important role in chemically-induced hepatotoxicity. A number of different models have been used for evaluating the hepatotoxicity of NCEs, including in vivo animal studies and various cell-based assays. Isolated human hepatocytes have been used to study the metabolism and toxicity of drug candidates. Human hepatocytes offer several advantages as an in vitro model. They provide human-specific results, testing can be carried out with small amounts of compounds, and animal use is minimized. Human hepatocytes can be obtained in different formats for use in the evaluation of hepatotoxicity by necrotic or apoptotic pathways. Freshly isolated hepatocytes can be used in suspension or plated as monolayers for short- or long-term toxicity studies. In addition, hepatocytes can be cryopreserved, thawed as needed, and used in suspension for evaluation of drug toxicity. Recently, cryopreserved hepatocytes have become available which, when thawed, can form monolayers to be used in place of fresh hepatocyte monolayers for metabolism or toxicity studies. Different methods for measuring necrosis (MTT, ATP, LDH) and apoptosis (Caspase, DNA fragmentation) using hepatocytes in various formats will be discussed.
Susceptibility Factors in Idiosyncratic Drug-Induced Liver Injury
Steven Yee, Ph.D., NIH, National Heart, Lung and Blood Institute
Drug-induced liver disease (DILD) is a common cause of life-threatening, acute liver failure and is a major reason drugs are removed from clinical development and widespread use. Because of its idiosyncratic nature, accurate prediction of which new drugs will cause DILD and who will be at risk for the development of this disease is difficult. Likewise, the pathogenesis of DILD is complex and appears to involve the formation of reactive drug metabolites that affect critical biochemical functions or elicit an immune response. Recent studies, however, have demonstrated that an alteration in the balance of protoxicants and protectants in the liver can also influence susceptibility to DILD. Environmental influences, such as the inflammagen bacterial lipopolysaccharide, and/or genetic polymorphisms may affect this balance. Indeed, concurrent inflammation can enhance susceptibility to a variety of hepatotoxicants, including drugs, and result in the generation of numerous injurious, protoxicant inflammatory mediators. Moreover, the deficiency of key protective factors in the liver – such as anti-inflammatory cytokines and others – can lead to increased tissue susceptibility, often through unchecked protoxicant activities. Hence, the balance between these susceptibility factors may play a critical role in the overall pathogenesis, with increased levels of protoxicants or decreased levels of protectants leading to this disease. Identification of such factors through the use of toxicogenomics, proteomics and metabonomics will be ideal in better understanding the mechanism and facilitating the prediction of DILD, thereby preventing the removal of otherwise beneficial drugs from the market.
Multi-endpoint Profiling of Suspect Hepatotoxicants in Cultured Hepatocytes
Thomas Flynn, Ph.D., FDA, Center for Food Safety and Applied Nutrition
Hepatotoxicity is the leading cause of post-market withdrawal of drugs and of warning notices for dietary supplements. The present study evaluated the effects of model compounds, some with known hepatotoxicity, on human (HepG2/C3A, WRL-68) and rat (Clone-9) hepatocyte cell lines. Cells were exposed to multiple concentration levels of test agent (up to the limit of aqueous solubility but not more than 1 mg/mL) for up to 48 hr. Specific endpoint assays, which could all be conducted in a plate reading fluorometer or luminometer and which model known mechanisms of hepatotoxicity, included: generation of reactive oxygen species (dichlorofluorescin); depolarization of mitochondria (rhodamine 123); steatosis (nile red); induction or inhibition of cytochrome P450 activities (EROD, BOROD); cell viability (liver enzyme release, Alamar blue, total ATP); and apoptosis (caspase 3). All parameters were normalized for total double stranded DNA content (H33258) per well. Each model compound generated a unique response profile based on which endpoint showed a concentration-related increase, decrease, or no change. Only 3 of 12 compounds tested showed significant cytotoxicity. For some model compounds that are human drugs and have known hepatotoxicity (e.g., valproic acid), some endpoints responded at concentrations comparable to therapeutic blood levels and in ways consistent with the compound’s known mechanism of toxicity. These findings suggest that this test system may serve as a screening assay for hepatotoxicity of food- or dietary supplement-related compounds.
The Use of Microarray Technology to Predict Hepatotoxicity in Humans
William Mattes, Ph.D., Genelogic, Inc.
Because numerous industrial and pharmaceutical chemicals have been found to cause damage to the liver, testing for hepatotoxicity remains a key component in the safety assessment of a new chemical entity. While standard animal models generally predict human hepatotoxicity, there are a number of examples of compounds whose human liver injury was not forseen in rodent models. Gene Logic has taken the premise that gene expression data from livers of compound-treated rats may be contain signals predictive of human hepatotoxicity. To this end studies were conducted with a series of known human and rodent hepatotoxins, as well compounds known to be safe. A database of gene expression results for various treatment times and doses was constructed, and then used to build predictive models composed of hundreds of gene responses. The models are of three types: a model predicting general hepatoxicity, models predicting various pathologies or mechanisms, and models identifying similarity of an unknown compound to one in the reference database. Examples of the results of these models for two paradigm compounds, tacrine and felbamate, will be presented.