Hepatic Toxicity and Cancer
This chapter reviews information presented in the Environmental Protection Agency (EPA) draft Integrated Risk Information System (IRIS) assessment of the toxic and carcinogenic effects of tetrachloroethylene on the liver. The metabolism of tetrachloroethylene by the liver is critical for its toxicity and carcinogenicity in that organ. The major metabolites of tetrachloroethylene responsible for hepatic effects are formed by the oxidative metabolic pathway (see Chapter 2 for an overview of toxicokinetics). The following sections address hepatotoxicity and hepatocarcinogenicity separately, but they are not necessarily independent end points. This information is considered in the context of the other evidence on carcinogenicity in Chapter 11, where EPA’s assessment of carcinogenic risks posed by tetrachloroethylene is evaluated.
The draft IRIS document on tetrachloroethylene points out that hepatotoxicity associated with tetrachloroethylene has been shown in rodents in several studies. A number of studies have been conducted with acute administration, but the draft correctly focuses on subchronic and chronic exposures, particularly those involving inhalation as a route of administration. Most of the toxicologic findings focus on increased liver weight, hypertrophy, and histologic lesions, including necrosis.
Damage to the liver by all or most of the chlorinated hydrocarbons has been demonstrated. Tetrachloroethylene is a weaker hepatotoxic agent than, for example, carbon tetrachloride and chloroform; this was shown by studies conducted in the middle 1960s (Klaassen and Plaa 1966, 1967).
The IRIS document overemphasizes a few studies. One is that by Kjellstrand et al. (1984), which is also mentioned in Chapter 3, on neurotoxicity. According to that study, exposure to tetrachloroethylene at 9 ppm for 30 days
caused a significant increase in liver weight (not corrected for body weight) in mice. The study also reported an increase in plasma butyrylcholinesterase (BuChE) in mice exposed to tetrachloroethylene at over 9 ppm for 30 days. Although the importance of the change in BuChE is not clear, the exposure in the study was so much lower than those in the other studies cited by EPA that it is important in considering the noncarcinogenic liver end points. EPA does not note that the increase in BuChE at 9 ppm was not significant (it was significant only at 37 ppm and above). It would be valuable for EPA to discuss this study critically in comparison with others in which much higher lowest observed-adverse-effect levels were found. In particular, it should be mentioned that increased BuChE in the Kjellstrand et al. study occurred at 37 ppm only when the exposure was continuous for the entire period, not when exposure at this concentration was intermittent, whereas other studies have involved intermittent exposure (usually 3-6 hours/day). Therefore, the total dose per mouse in the Kjellstrand et al. study must have been several times higher than that in other studies, and the information given in the draft (p. 4-12 and Table 4-2 on p. 4-14) is misleading. It would also be useful for EPA to discuss the quality of studies (for example, deficiencies in reporting by Kjellstrand et al.) and the toxicologic meaning, if any, of the reported effects. Furthermore, the increase in BuChE as a toxic effect does not appear to have been considered important by other investigators, on the basis of citations of the Kjellstrand et al. paper, nor does the effect seem to have been reported by others. Thus, a more critical analysis of the study is necessary to determine the significance of its findings in comparison with other reports of hepatotoxicity that required higher exposure concentrations.
The National Toxicology Program (NTP 1986) and Japan Industrial Safety Association (JISA 1993) studies lend some support to the possibility of hepatotoxicity associated with exposure to tetrachloroethylene. In the NTP 13-week study in rats, hepatotoxicity was evidenced as congestion in the liver. In the 13-week study in mice, there was leukocytic infiltration, centrolobular necrosis, and bile stasis in animals exposed to tetrachloroethylene at 400, 800, or 1,600 ppm. Liver degeneration was observed to occur in a dose-dependent fashion in the 2-year study in mice. In the JISA study, there was an increase in spongiosis hepatitis in Crj:BDF1 mice, but it is a common finding in these mice and is likely to be unrelated to chemical exposure. Hyperplasia was not statistically significantly increased; there were increases in angiectasis and central degeneration.
In updating and revising the draft IRIS assessment, EPA should include a new 30-day gavage study in Swiss Webster mice given tetrachloroethylene at 150, 500, and 1,000 mg/kg/day (Philip et al. 2007). The metabolism of tetrachloroethylene and its toxicity were examined. That is one of the few studies that were conducted with oral administration and repeated dosing. The investigators found that hepatic injury peaked at 7 days but then was repaired. That suggests that single-dose studies demonstrating hepatic damage on the basis of measurements made after short periods might not mimic the effects of repeated dosing.
EPA also discusses hepatotoxicity in humans. Most of the studies cited in the IRIS draft involved dry-cleaners and found no evidence of an association. However, the EPA document gives undue weight to a couple of studies. One (Brodkin et al. 1995) used sonographic analysis of scattering of fat in liver. This was the only study to report such effects in tetrachloroethylene exposed populations and the importance of the fat changes as an indicator of toxic response is unclear. Furthermore, serum transaminases were not increased in the exposed population. Thus, interpretation of the result is difficult. EPA also considers the study of Gennari et al. (1992). They reported an increase in gamma-glutamyltransferase-2 in tetrachloroethylene-exposed dry-cleaners. The relevance of that finding as an indicator of hepatotoxicity is unclear. The investigators did not find any other indicators of hepatotoxicity despite an extensive serum-enzyme profile. It is likely that the concentrations of tetrachloroethylene that humans were exposed to in those studies were too low to induce frank hepatotoxicity. Further studies are needed.
The NTP (1986) and JISA (1993) studies showed, as is the case with many of the halogenated solvents, that there is a dose-dependent increase in hepatic tumors after exposure to tetrachloroethylene in both sexes of mice but not in rats. The draft IRIS assessment’s section on hepatic carcinogenicity is written reasonably well in a descriptive sense, with regard to the style of the presentation of the cancer-relevant results of long-term studies with tetrachloroethylene. However, the presentation would benefit if the table on page 5-37, which now gives cumulative tumor incidence, were expanded to include information on species; strain; dose; duration; incidence and multiplicity of adenomas, carcinomas, and other hepatic tumors (such as hemangiosarcomas); and the literature cited.
Tetrachloroethylene induces hepatocellular carcinomas and adenomas in mice. The yield of tetrachloroethylene-induced hepatocellular carcinomas is statistically significant in both male and female B6C3F1 mice after either oral or inhalation exposure. Both male and female Crj:DBF1 mice also have an increased incidence of hepatocellular carcinomas after inhalation exposure to tetrachloroethylene. The earlier studies of the National Cancer Institute (NCI 1977) were repeated, and the findings were confirmed by Nagano et al. (1998). As discussed in more detail below in the section on mode of action, some metabolites of tetrachloroethylene—including trichloroacetic acid (TCA), dichloroacetic acid (DCA), and chloral hydrate (if it is formed)—cause hepatic cancer in mice, and DCA causes hepatic cancer in rats. In the study by Nagano et al., both
males and females incurred dose-related increases in incidences of hepatic carcinoma and combined hepatic adenoma and carcinoma.
A difficulty in interpreting the significance of the mouse hepatic tumors is that they have a high spontaneous background incidence in mice. Such tumors have been commonly encountered after exposure to other halogenated solvents, such as dichloromethane, trichloroethylene, tetrachloroethane, carbon tetrachloride, and 1,1,2-trichloroethane.
The curious observation of hepatic and splenic hemangiosarcomas reported in male mice in one of the tetrachloroethylene mouse bioassays (JISA 1993) is mentioned several times in the EPA draft as a potentially important finding; however, there is little discussion of these tumors, the potential mode of action, or the relevance to human risk. Reference to the tumors is presented in Figure 5-14, Table 5-5, and Table 5-9. The analysis is complicated by the fact that the JISA report does not describe the tumors as hemangiosarcomas, but rather as hemangioendothelioma; this term is usually associated with benign tumors, but JISA lists it as a malignant hepatic tumor in male mice. The term is also used for both benign and malignant tumors of the spleen. Furthermore, because of the cell types involved, the hepatocellular carcinomas being of hepatocellular origin and the hemangiosarcomas being of endothelial-cell origin, it is scientifically inappropriate to lump these tumors in with carcinomas, as is done by EPA (Figure 6.4 and Table 6.4).
Available epidemiologic evidence does not support an association between tetrachloroethylene and hepatic cancer. Two cohort mortality studies of drycleaner union members (Ruder et al. 2001; Blair et al. 2003) and a large (N = 77,965) cohort mortality study of aerospace workers (Boice et al. 1999) report no association with hepatic-cancer mortality. A sizable subcohort (N = 2,631) of the aerospace workers routinely exposed to tetrachloroethylene had a standardized mortality ratio of 2.05 (95% confidence interval [CI], 0.83-4.23) on the basis of seven observed deaths. However, an analysis that used an internal cohort referent population to reduce confounding yielded no overall association and no exposure-response relationship. Because hepatic cancer is fatal, assessments of mortality represent the burden of the disease in the population. Essentially null associations are reported in studies of incident cancers in laundry workers residing in Nordic countries. In the one study cited (Lynge et al. 1995) that reported an increased standardized incidence ratio (SIR) for hepatic cancer in women (2.7; 95% CI, 1.5-4.5; 14 observed cases, all cases were in laundry workers, and no cases were observed in dry-cleaning workers, whose exposure to tetrachloroethylene is more likely. (The EPA document does not cite this correctly in Table 4B-1a; the reference should be to Lynge et al. 1995, which is an update of Lynge and Thygesen 1990.) Those studies identified laundry and drycleaning workers on the basis of the census in 1970 and 1980, so the extent of
exposure is unknown. Several population-based case-control studies of hepatic cancer and exposure to solvents (determined by occupation) have been conducted over the last 30 years. Overall, they have not reported an association between tetrachloroethylene and hepatic cancer. Some evidence is suggestive of an association between solvent exposure and laundry work and hepatic cancer in women, but the exposure models for these studies are crude, and methods of control selection raise questions about the validity of the results.
The draft IRIS assessment does not use that limited evidence of an association between tetrachloroethylene and hepatic cancer as supportive of classifying tetrachloroethylene as a carcinogen. The argument that human epidemiologic evidence supports classification as “likely to be a carcinogen” is limited to other cancers, specifically esophageal and lymphoid cancers. The exclusion of hepatic cancer as supporting evidence is appropriate.
Mode of Action
The draft assessment describes the mode of action (MOA) of tetrachloroethylene’s hepatic toxicity and carcinogenicity in several places. The most comprehensive description of the available body of information and identification of potential key events in the MOA are included in Section 4.4.4. The MOA summary is provided in Section 4.10.3, including Table 4-13; Appendix 4A details the EPA-conducted analysis of the consistency between carcinogenicity of tetrachloroethylene and that of one of its major oxidative metabolites, TCA; and Section 6.1.5 includes a short summary of the liver MOA with regard to the human hazard potential of tetrachloroethylene.
EPA concludes that “the MOA for tetrachloroethylene-induced mouse liver cancer is not well understood, and it is highly likely that more than one MOA is operative” (EPA 2008, p. 4-16). In support of that conclusion, EPA describes pathways that could lead to hepatic tumors but does not clearly describe the weight-of-evidence approach for determining the key elements in the tumorigenicity of tetrachloroethylene for the possible MOAs presented. The difficulty in characterizing the MOA is not surprising given the complexity of the metabolic pathways for tetrachloroethylene, the closely related chlorinated solvent trichloroethylene, and their common primary oxidative metabolites, TCA and DCA. The following major events are put forth as plausible components of the MOA of hepatic carcinogenicity of tetrachloroethylene (in no particular order with regard to a temporal sequence):
Metabolism of tetrachloroethylene to TCA and DCA, which are both considered ultimate hepatotoxic metabolites.
Activation of peroxisome proliferator-activated receptor-α (PPARα) and the downstream cascade of the molecular events that include induction of peroxisomes, increase in cell proliferation, and decrease in rates of apoptosis.
Other nongenotoxic events, such as promotion of growth of previously initiated foci, changes in epigenetic status, cytotoxicity, and oxidative stress.
Genotoxic events, such as DNA damage by tetrachloroethylene metabolites or chromosomal aberrations.
Although the discussion of the PPARα-mediated events and their possible roles in species differences with regard to the hepatocarcinogenic potency of tetrachloroethylene is extensive, other important potential MOAs or key events are largely overlooked. For example, the possible role of epigenetic changes caused by TCA and DCA is mentioned, but there is little discussion of the studies that have been conducted on this subject. Similarly, cytotoxicity and secondary oxidative stress that may result from microsomal enzyme induction are insufficiently considered. Adding such discussions would strengthen EPA’s MOA analysis and conclusions.
That TCA is the major urinary metabolite of tetrachloroethylene and is a mouse hepatocarcinogen suggests that the hepatocarcinogenicity of tetrachloroethylene is due in part to TCA. DCA is another tetrachloroethylene urinary metabolite that is formed both in the oxidative pathway by dechlorination of TCA and, in organs other than the liver, in the glutathione (GSH) pathway. DCA is known to cause hepatic cancer in both rats and mice, so it is possible that DCA contributes to the hepatocarcinogenicity of tetrachloroethylene, although it is not certain to what extent it contributes in that little of it is produced and it is produced primarily in the kidney. Early studies that reported finding DCA as a metabolite may have overstated the amount formed because of problems with analytic methods (Ketcha et al. 1996). Later studies showed very small amounts of DCA, if any, being formed from tetrachloroethylene. Chloral hydrate (if it is formed) is a mutagen and is a hepatocarcinogen in mice and might contribute to the hepatocarcinogenicity of tetrachloroethylene. In addition, metabolites formed from the GSH pathway, such as trichlorovinylglutathione, which is further metabolized by β-lyase in the kidneys, are also genotoxic.
The multiplicity of metabolites formed from tetrachloroethylene that are toxic and carcinogenic—TCA, DCA, tetrachloroethylene oxide, trichloroacetyl chloride, and possibly chloral hydrate—makes it difficult to determine the MOA of hepatocarcinogenicity of tetrachloroethylene. Indeed, there may not be enough data to determine quantitatively the extent to which each metabolite contributes to tetrachloroethylene-induced hepatotoxicity. Perhaps a summary of the available information on hepatocarcinogenicity of TCA, DCA, and chloral hydrate administered alone or in combination with other compounds—for example, from studies of Bull et al. (1990, 2002, 2004) on mixtures and coadministration with gadolinium chloride—should be included in the IRIS assessment and in tabular form (e.g., see table in NRC 2006a, pages 149-156) to better assess the data.
Although the consideration of the metabolic activation of tetrachloroethylene and the comparison with TCA-induced carcinogenesis are useful, the dose-
response information in the draft on tumor formation after TCA administration (Table 4A-2) suggest that very high concentrations of TCA are needed to cause hepatic tumors—far beyond what would be generated after tetrachloroethylene administration.
The peroxisome-proliferator MOA is discussed in great detail. The key events associated with the known links between peroxisome-proliferator chemicals in general and rodent hepatic cancer are identified, and appropriate literature references are included. However, no data or weight of evidence criteria specifically on tetrachloroethylene are provided, and the lack of coherent flow in the document detracts from the intended message. The document might be improved by organizing the information into sections that make clear (1) what parts of this MOA are based on studies with other model peroxisome proliferators, (2) what data are available to support this MOA for tetrachloroethylene, (3) for TCA, (4) the rationale for species differences, and (5) the relevance of this MOA to mouse hepatic tumors induced by tetrachloroethylene or to human risk.
As presented, the draft IRIS assessment seems to be more concerned with critiquing the current dominant view in the field that the peroxisome-proliferator MOA may not be relevant to human hepatocarcinogenesis than with providing evidence of links between tetrachloroethylene and this MOA. The general criticism of the MOA with regard to its relevance to humans is warranted, although it should be expressed in milder terms, and it points correctly to several historical and recent lines of evidence that suggest important inconsistencies that challenge the paradigm of the central role of PPARα in rodent, but not human, hepatocarcinogenesis. However, as pointed out above, the data linking tetrachloroethylene to this MOA are weak to begin with and come largely from studies of trichloroethylene and TCA, not tetrachloroethylene itself. The idea that there are deficiencies in our knowledge of tetrachloroethylene should be made more prominent. Similarly, the discussion of “tetrachloroethylene and PPARα MOA” and the discussion of “relevance of the PPARα MOA to human liver carcinogenesis” should be separated more clearly by EPA.
The discussion of the strain and species differences in the peroxisome-proliferation effect of TCA is rather limited. TCA is capable of inducing peroxisome proliferation in the rat, but tetrachloroethylene does not. In addition, the issues of PPARα transactivation by tetrachloroethylene, related chemicals, and their key metabolites and of species differences are important for the discussion of the MOA. Again, a critical look at the quantitative differences in metabolic activation of tetrachloroethylene to TCA between mouse and rat, species that are generally believed to be almost equally sensitive to peroxisome proliferation and differences between mouse and rat in hepatic cancer induced by other compounds of this class should be provided. Specifically, EPA may consider performing additional analyses with the rat data similar to those with the mouse data in Appendix 4A and including a table showing the quantitative differences in affinity between mouse, rat, and human PPARα of tetrachloroethylene and its key metabolites in comparison with the known peroxisome proliferators. Such
analyses and data would greatly facilitate the discussion of quantitative differences between compounds and between species.
The study by Nakajima et al. (2000) is only mentioned in passing on page 4-26 of the draft assessment. It should be discussed in greater detail, especially the data on sex differences and mechanistic considerations. It provides a possible mechanistic explanation for sex differences in susceptibility to carcinogenesis by tetrachloroethylene—information that is important for the discussion of the complexities of and uncertainties in the MOA.
The dose-response relationship in Section 126.96.36.199.6 touches on the important issue. However, the arguments are not supported by adequate literature citations, and the only paper cited is a broad review article, not a primary source of the data. Section 5 contains ample information on dose-response relationships, so appropriate cross-referencing should be included in Section 188.8.131.52.6.
The discussion on nonliver targets in humans that may involve PPARα MOA is interesting, but it is too brief and is not adequately linked to the rest of the chapter to have an appropriate impact. The arguments presented in Section 184.108.40.206.8 may be substantiated by providing a quantitative comparison of PPARα transactivation potential by tetrachloroethylene and its metabolites, as suggested above. Similarly, the discussion of the potential role of PPARβ is inadequate. Specifically, it should be noted that PPARβ may be an important gene for human hepatocellular carcinogenesis.
The committee agrees with EPA that the MOA of tetrachloroethylene-induced hepatic tumors is not clear. Many toxic metabolites are formed from tetrachloroethylene. Hence, it is likely that key events from several pathways operate in tetrachloroethylene-induced hepatocarcinogenesis. It is likely that TCA, DCA, and chloral hydrate (if it is formed)—which are carcinogens in rodents—contribute to tetrachloroethylene-induced hepatocarcinogenesis. It is also likely that mutagenic metabolites of tetrachloroethylene formed via the cytochrome P450 and GSH pathways (tetrachloroethylene-epoxide, TCA, DCA, and TCVG) contribute to hepatocarcinogenesis. And it is possible that activation of PPARα and consequent peroxisomal proliferation; genotoxic events induced by tetrachloroethylene metabolites, including chromosomal aberrations; and other nongenotoxic events—such as promotion of growth of previously initiated foci, changes in epigenetic status, and oxidative stress—may all contribute to the overall MOA through several simultaneous mechanisms. The hypothesis that the mutagenic metabolites of tetrachloroethylene (tetrachloroethylene-epoxide, TCA, DCA, chloral hydrate [if it is formed], and TCVG) initiate carcinogenesis and that tetrachloroethylene-induced promotion of initiated foci, cytotoxicity, and epigenetic events promote carcinogenesis cannot be ruled out.
As with other halogenated solvents, there is evidence in a number of species that tetrachloroethylene can cause liver damage. This was well described by
EPA in the drat IRIS assessment. Two rodent bioassays have demonstrated that high doses of tetrachloroethylene produced liver tumors in mice. While there is clear evidence that this occurs, the basis for their occurrence is not clear and may actually involve more than one MOA. This makes the determination of the relevance to humans more difficult. This is particularly true with respect to the importance of PPARα as the predominant or sole MOA, which led to a split opinion among committee members and a dissenting statement (see Appendix B).
Further studies are needed to define the MOAs for tetrachloroethylene-induced liver tumors, with particular emphasis on the importance of PPARα and whether species difference might exist. In addition, further study is needed to determine the relative roles of metabolites of tetrachloroethylene in tumor development. This may require the development of better analytical methods to detect some metabolites.