Evidence-based causation in toxicology: A 10-year retrospective.

We introduced Evidence-based Toxicology (EBT) in 2005 to address the disparities that exist between the various Weight-of-Evidence (WOE) methods typically applied in the regulatory hazard decision-making arena and urged toxicologists to adopt the evidence-based guidelines long-utilized in medicine (i.e., Evidence-Based Medicine or EBM). This review of the activities leading to the adoption of evidence-based methods and EBT during the last decade demonstrates how fundamental concepts that form EBT, such as the use of systematic reviews to capture and consider all available information, are improving toxicological evaluations performed by various groups and agencies. We reiterate how the EBT framework, a process that provides a method for performing human chemical causation analyses in an objective, transparent and reproducible manner, differs significantly from past and current regulatory WOE approaches. We also discuss why the uncertainties associated with regulatory WOE schemes lead to a definition of the term "risk" that contains unquantifiable uncertainties not present in this term as it is used in epidemiology and medicine. We believe this distinctly different meaning of "risk" should be clearly conveyed to those not familiar with this difference (e.g., the lay public), when theoretical/nomologic risks associated with chemical-induced toxicities are presented outside of regulatory and related scientific parlance.

Reevaluating cancer risk estimates for short-term exposure scenarios.

Estimates of cancer risk from short-term exposure to carcinogens generally rely on cancer potency values derived from chronic, lifetime-exposure studies and assume that exposures of limited duration are associated with a proportional reduction in cancer risk. The validity of this approach was tested empirically using data from both chronic lifetime and stop-exposure studies of carcinogens conducted by the National Toxicology Program. Eleven compounds were identified as having data sufficient for comparison of relative cancer potencies from short-term versus lifetime exposure. The data were modeled using the chronic data alone, and also using the chronic and the stop-exposure data combined, where stop-exposure doses were adjusted to average lifetime exposure. Maximum likelihood estimates of the dose corresponding to a 1% added cancer risk (ED(01)) were calculated along with their associated 95% upper and lower confidence bounds. Statistical methods were used to evaluate the degree to which adjusted stop-exposures produced risks equal to those estimated from the chronic exposures. For most chemical/cancer endpoint combinations, inclusion of stop-exposure data reduced the ED(01), indicating that the chemical had greater apparent potency under stop-exposure conditions. For most chemicals and endpoints, consistency in potency between continuous and stop-exposure studies was achieved when the stop-exposure doses were averaged over periods of less than a lifetime-in some cases as short as the exposure duration itself. While the typical linear adjustments for less-than-lifetime exposure in cancer risk assessment can theoretically result in under- or overestimation of risks, empirical observations in this analysis suggest that an underestimation of cancer risk from short-term exposures is more likely.

The acetaminophen regioisomer 3'-hydroxyacetanilide inhibits and covalently binds to cytochrome P450 2E1.

3'-Hydroxyacetanilide has been previously studied as a nontoxic regioisomer of the analgesic acetaminophen (4'-hydroxyacetanilide). The radiolabeled derivative has been shown to covalently bind to liver proteins at levels similar to that observed with hepatotoxic doses of radiolabeled acetaminophen with no evidence of hepatic damage. Using an anti-arylacetamide antiserum the primary protein adduct detected following administration of 3'-hydroxyacetanilide (300 and 600 mg/kg) to mice was a 50 kDa microsomal protein that co-migrated with cytochrome P450 2E1. Cytochrome P450 2E1 enzyme activity (p-nitrophenol hydroxylase) was decreased by 79% in the mice treated with 3'-hydroxyacetanilide (600 mg/kg). Incubation of 3'-hydroxyacetanilide with hepatic microsomes resulted in a time dependent 47% decrease in cytochrome P450 2E1 activity. Pre-incubation of acetaminophen with microsomes did not result in covalent binding to the cytochrome P450 nor was there a decrease in p-nitrophenol hydroxylase activity. These data suggest that 3'-hydroxyacetanilide covalently binds to cytochrome P450 2E1 with preferential loss of activity.

Detection of trichloroethylene-protein adducts in rat liver and plasma.

Trichloroethylene is an industrial chemical with widespread occupational exposure and is a major environmental contaminant. In a Western blot using antiserum that recognizes trichloroethylene covalently bound to protein, a single 50 kDa microsomal adduct was detected in the livers of trichloroethylene-treated Sprague-Dawley rats. To determine if trichloroethylene-protein adducts could be detected in blood, plasma proteins were immunoaffinity purified using an antidichloroacetyl column. A single 50 kDa protein was detected in the affinity-purified fraction in a Western blot using dichloroacetyl antiserum. This protein was also immunochemically reactive with anti-cytochrome P450 2E1 antibodies. The 50 kDa trichloroethylene-protein adduct may be formed in the liver and released into the blood following exposure to trichloroethylene. The significance of adduct formation with respect to trichloroethylene toxicity remains to be established; however, the data suggest that this approach may be useful in the investigation of trichloroethylene-protein adducts and adverse effects following exposure.

Covalent binding of xenobiotics to specific proteins in the liver.

Chemicals that cause toxicity though a direct mechanism, such as acetaminophen, covalently bind to a select group of proteins prior to the development of toxicity, and these proteins may be important in the initiation of the events that lead to the hepatotoxicity. Disruption of the cell is measured by release of intracellular proteins such as alanine aminotransferase and occurs late in the time course following a hepatotoxic dose of a direct toxin. Prior to this disruption, there appears to be a large number of proteins covalently modified by a reactive metabolite. There are at least two possible mechanisms that may cause the toxicity. First, some critical protein is a target of the reactive metabolite. Disruption of the enzymatic function (or a critical pathway for a regulatory protein) may lead directly to cell death. With the direct hepatotoxin acetaminophen, there is a decrease in the activity of several of the early target proteins, but how this disruption of critical proteins leads to the toxicity is still unclear. The early targets appear to be proteins with accessible nucleophilic sulfhydryl groups, and usually the target has a high concentration of the protein within the cell. It is possible that the binding to some of these proteins represents a detoxification protecting more critical targets within the cell. A second mechanism for the direct toxicity is that more and more proteins become targets in the time course following administration of a direct toxin, and eventually the cells machinery is overwhelmed. The cell can then no longer function, or there is a disruption the redox balance within the cell due to the decreased function of numerous proteins. In contrast to the direct-acting toxins, the chemical-protein conjugates that initiate toxicity through an activation of the immune system appear to have a limited number of target proteins and are localized within one subcellular fraction. Halothane produces adducts almost exclusively in the microsomal fraction, and these adducts appear to be limited to selective proteins with high concentrations in this fraction. The substitution level is an important factor in the development of an immune response. Halothane hepatitis patients' antibodies primarily recognize proteins with a high substitution level. For halothane and diclofenac, the proteins are accessible to the immune system through exposure on the plasma membrane. Trichloroethylene binds primarily to a 50-kDa microsomal protein, and preliminary evidence has been presented which indicates that a trichloroethylene-protein conjugate is released into the blood following exposure, where contact with the immune system can occur. In order to elicit an immune response the immune system requires multiple exposure to the chemical-protein conjugates. With halothane hepatitis and with diclofenac hepatitis, as well as occupational and environmental exposure to trichloroethylene, there are multiple exposures leading to repeat presentation of the protein adducts to the immune system; this situation is not generally found with acetaminophen overdose patients. In summary, direct toxicants such as acetaminophen covalently bind to selected targets which may be critical to the development of hepatotoxicity, and they later form adducts with numerous proteins which may overwhelm the cell's capacity to maintain homeostasis, leading to loss of vital function and cell death (Fig.3). In contrast, indirect toxicants that elicit an immune-mediated toxicity such as halothane, and possibly diclofenac and trichloroethylene, appear to have a limited number of protein targets with a high substitution level, and the immune system is exposed repeatedly to the modified proteins.

Covalent binding and inhibition of cytochrome P4502E1 by trichloroethylene.

1. To investigate the effects of trichloroethylene on cytochrome P4502E1 (CYP2E1), an isozyme responsible for its metabolic activation, mice were treated with trichloroethylene and Western blot staining with both anti-dichloroacetyl and anti-CYP2E1 antisera detected a comigrating 50 kDa protein band. There was a dose-dependent increase in the intensity of the 50 kDa protein adduct stained immunochemically with anti-dichloroacetyl. 2. CYP2E1 enzyme activity was decreased from control levels in a dose-dependent manner in mice treated with 250-500 mg/kg TRI. 3. Microsomal incubations with trichloroethylene resulted in covalent binding to several proteins including a 50 kDa adduct, which is in contrast with the selective binding to the 50 kDa protein observed in vivo. 4. CYP2E1 enzyme activity levels were significantly decreased following microsomal incubation with NADPH and trichloroethylene, and additionally there was a time- and NADPH-dependent decrease in enzyme activity indicating that trichloroethylene is a mechanism-based inhibitor of CYP2E1.

Covalent binding of acetaminophen to N-10-formyltetrahydrofolate dehydrogenase in mice.

The analgesic acetaminophen is frequently used as a model chemical to study hepatotoxicity; however, the critical mechanisms by which it produces toxicity within the cell are unknown. It has been postulated that covalent binding of a toxic metabolite to crucial proteins may inhibit vital cellular functions and may be responsible for, or contribute to, the hepatotoxicity. To further understand the importance of covalent binding in the toxicity, a major cytosolic acetaminophen-protein adduct of 100 kDa has been purified by a combination of anion exchange chromatography and preparative electrophoresis. N-Terminal and internal amino acid sequences of peptides from the purified 100-kDa acetaminophen-protein adduct were found to be homologous with the deduced amino amino acid sequence from the cDNA of N-10-formyltetrahydrofolate dehydrogenase. Antiserum specific for N-10-formyltetrahydrofolate dehydrogenase and acetaminophen react in a Western blot with the purified 100-kDa acetaminophen-protein adduct. Administration of a toxic dose of acetaminophen (400 mg/kg) to mice resulted in a 25% decrease in cytosolic N-10-formyltetrahydrofolate dehydrogenase activity at 2 hr. The covalent binding of acetaminophen to proteins such as N-10-formyltetrahydrofolate dehydrogenase and the subsequent decreases in their enzyme activity may play a role in acetaminophen hepatotoxicity.

Protein targets of xenobiotic reactive intermediates.

Many xenobiotics are metabolically activated to electrophilic intermediates that form covalent adducts with proteins; the mechanism of toxicity is either intrinsic or idiosyncratic in nature. Many intrinsic toxins covalently modify cellular proteins and somehow initiate a sequence of events that leads to toxicity. Major protein adducts of several intrinsic toxins have been identified and demonstrate significant decreases in enzymatic activity. The reactivity of intermediates and subcellular localization of major targets may be important in the toxicity. Idiosyncratic toxicities are mediated through either a metabolic or immune-mediated mechanism. Xenobiotics that cause hypersensitivity/autoimmunity appear to have a limited number of protein targets, which are localized within the subcellular fraction where the electrophile is produced, are highly substituted, and are accessible to the immune system. Metabolic idiosyncratic toxins appear to have limited targets and are localized within a specific subcellular fraction. Identification of protein targets has given us insights into mechanisms of xenobiotic toxicity.

Immunochemical detection of protein adducts in mice treated with trichloroethylene.

Trichloroethylene has been shown to produce tumors in rodents and is a suspect human carcinogen. In addition, a number of case reports raise the possibility that trichloroethylene can induce an autoimmune disorder known as systemic sclerosis. To investigate whether covalent binding of reactive trichloroethylene metabolites may be involved in the mechanisms underlying these toxic responses, we have developed a polyclonal antibody that can recognize trichloroethylene--protein adducts in tissues. The antibody was prepared by immunizing a rabbit with dichloroacetic anhydride-modified keyhole limpet hemocyanin. Enzyme-linked immunosorbent assay data indicated that the serum antibody recognized dichloroacetic anhydride-modified rabbit serum albumin, but not unmodified protein. In addition, N epsilon-dichloroacetyl-L-lysine was the most potent inhibitor of antibody binding to dichloroacetic anhydride-modified rabbit serum albumin, indicating that the antibody recognizes primarily dichloroacetylated lysine residues. Immunoblots revealed the presence of two major trichloroethylene adducts at 50 and 100 kDa in liver microsomal fractions from male B6C3/F1 mice treated with trichloroethylene. The formation of trichloroethylene adducts was both dose and time dependent. Furthermore, the 50-kDa adduct was found to comigrate on a polyacrylamide gel with cytochrome P450 2E1. These data show that reactive metabolites of trichloroethylene are formed in vivo and bind covalently to discrete proteins in mouse liver. The data also suggest that one of the protein targets is cytochrome P450 2E1. Further studies will be necessary to elucidate the relationship between covalent binding of trichloroethylene and trichloroethylene toxicity.

Glutamate dehydrogenase covalently binds to a reactive metabolite of acetaminophen.

The mechanism of the hepatotoxicity of the analgesic acetaminophen is believed to be mediated by covalent binding to protein; however, critical targets which effect the toxicity are unknown. It has been shown that mitochondrial respiration in vivo is inhibited in mice as early as 1 h following a hepatotoxic dose of acetaminophen, and it is postulated that covalent binding to critical mitochondrial proteins may be important. A time course of mitochondrial proteins stained with anti-acetaminophen in an immunoblot detected two major adducts of 50 and 67 kDa as early as 30 min after a hepatotoxic dose of acetaminophen in mice. To further understand the role of covalent binding to mitochondrial proteins and acetaminophen hepatotoxicity, we have purified and identified a 50 kDa mitochondrial protein which becomes covalently bound to a reactive metabolite of acetaminophen. An N-terminal sequence of the 50 kDa adduct was 100% homologous with the deduced amino acid sequence of glutamate dehydrogenase. In addition, the purified protein was immunochemically reactive with rat liver anti-glutamate dehydrogenase. Enzyme activity of glutamate dehydrogenase was significantly decreased in mice 1 h following hepatotoxic treatment with acetaminophen. These data suggest that acetaminophen hepatotoxicity may in part be mediated by covalent binding to glutamate dehydrogenase.

Immunochemical detection of oxidized proteins.

An immunochemical assay was developed to detect carbonyl moieties that result from oxidative damage to proteins. Bovine serum albumin was reacted with hydroxyl radicals generated via a Fenton-like mechanism or by a radiolysis mechanism. The resulting albumin-derived carbonyls were reacted with 2,4-dinitrophenylhydrazine, giving the corresponding hydrazones, which were detected by Western blot using anti-dinitrophenyl antisera. The immunoblot demonstrated a concentration-dependent increase in carbonyl formation, as well as fragmentation of the albumin into two distinct bands with molecular masses of 51 and 45 kDa when oxidized with the Fenton-like mechanism, and 62 and 46 kDa when oxidized by radiolysis. Analysis of the immunoblot using laser densitometry indicated a linear relationship between carbonyl groups and increasing treatment from radiolysis. This immunochemical assay was approximately 3 orders of magnitude more sensitive than the spectrophotometric method and was able to determine the molecular mass of carbonyl-modified polypeptides in the detection of oxidative damage.