TCE was widely used as a solvent and metal degreasing chemical in the United States from the early 1900s to 1980s (Bakke et al., 2007). Workers were initially subjected routinely to quite high exposures. NIOSH (1989) estimated that 401,000 persons at 23,225 plants in the United States were potentially exposed to TCE. TCE has been identified at over one half of the nearly 1300 hazardous waste sites that make up the EPA’s National Priorities List (ATSDR, 2006a; Pohl et al., 2008). It is released into the atmosphere from vapor degreasing operations; however, direct discharges to surface waters and groundwater from disposal operations have been frequent occurrences. As a result, TCE can be released to indoor air by vapor intrusion through underground walls and floors and by volatilization from the water supply. TCE has recently received a great deal of attention from the scientific and regulatory communities, due in large part to the EPA’s effort to update the chemical’s two-decade-old risk assessment (EPA, 2009).
Moderate to high doses of TCE, as with other halocarbons, are associated with a number of noncancer toxicities (ATSDR, 1997d; Barton and Das, 1996). TCE has been implicated in the development of autoimmune disorders and immune system dysfunction (Blossom and Gilbert, 2006; Cooper et al., 2009; Lan et al., 2010), and has been investigated for its potential as a male reproductive toxicant (Forkert et al., 2003; Xu et al., 2004). The effect of gestational exposure to TCE or its oxidative metabolites on cardiac development is a subject that has invoked considerable debate, as conflicting results have been published (Fisher et al., 2001; Johnson et al., 2003; Watson et al., 2006). The issue of TCE’s potential effect on ocular development has also recently reemerged (Warren et al., 2006). Nevertheless, cancer remains the dominant issue for TCE. TCE and its potential health risks have recently been reviewed by the NAS (2006, 2009), Jollow et al. (2009), and the EPA (2009).
The adverse effects of TCE, other than CNS depression, are generally believed to be associated with TCE’s metabolites. CNS depression is due to the parent compound and a major metabolite, TCOH. Knowledge of TCE’s metabolism in different species and under different exposure conditions is thus a prerequisite to understanding its mechanisms of action and assessing human health risks. A great deal of experimentation at many biological levels has been conducted over the last five decades to these ends.
TCE is rapidly and extensively absorbed into the systemic circulation via the oral and inhalation routes. Dermal absorption is considerably slower and less extensive. The majority of TCE is oxidized in the liver, while a small amount is conjugated in the liver with GSH by GSTs. The oxidative pathway is shown on the left of Fig. 24-5, while the GSH pathway is shown on the right. This diagram is a simplification of a more complex metabolic scheme described in detail by Lash et al. (2000a).
General scheme of metabolism for TCE. Abbreviations: β-Lyase, cysteine conjugate β-lyase; DCA, dichloroacetate; DCVC, S-(1,2-dichlorovinyl)-l-cysteine; DCVSH, 1,2-dichlorovinylthiol; DCVCSO, DCVC sulfoxide; DCVG, S-(1,2-dichlorovinyl)glutathione; DP, dipeptidase; FMO, flavin-containing monooxygenase; GGT, γ-glutamyltransferase; GST, GSH S-transferase; P450, cytochrome P450; TCA, trichloroacetate; TCOH, trichloroethanol; TCOG, TCOH glucuronide. (Reproduced from Lash et al., 2005, with permission from Elsevier.)
The initial step in the oxidative pathway is catalyzed by microsomal CYPs. CYP2E1, as described previously in the subsection “Metabolism” in the section “Toxicokinetics,” is the primary isozyme responsible for oxidation of low concentrations of TCE (Lipscomb et al., 1997; Ramdhan et al., 2008; Kim and Ghanayem, 2006). CYP-catalyzed oxidation of TCE in rodents and humans, in decreasing order of magnitude, is as follows: mice >> rats > humans (Lash et al., 2000a). Whether or not TCE is initially converted to TCE oxide is controversial. Cai and Guengerich (2001) were able to detect formation of trace amounts of the epoxide by phenobarbital-induced rat liver CYPs, but not by human liver CYPs. The majority of TCE is apparently converted to an oxygenated TCE–CYP intermediate, which rearranges to form chloral, a major metabolic intermediate. Chloral is oxidized to chloral hydrate (CH), a sedative and hypnotic still widely used in medical and dental procedures for infants and children (Buck, 2005; Heistein et al., 2006). CH is both oxidized to TCA and reduced to TCOH. Much TCOH is conjugated with glucuronic acid (GLU) and excreted in the urine. TCOH–GLU that is excreted in the bile is extensively hydrolyzed in the gut, reabsorbed, and oxidized in part to TCA (Stenner et al., 1997). Chiu et al. (2007) observed that concentrations of TCA were significantly lower than TCOH and TCOH–GLU concentrations in the blood of humans who inhaled TCE at 1 ppm for six hours. Modest amounts of DCA apparently are produced from TCA and TCOH in mice, but relatively little DCA is formed in rats or humans. Trace amounts of DCA were detected in one study of TCE-exposed humans (Fisher et al., 1998) but not in other studies (Lash et al., 2000b; Bloemen et al., 2001). Very high doses of both TCA and DCA have been shown to be hepatic carcinogens in mice (Bull, 2000). It is generally accepted that TCA is a nongenotoxic liver carcinogen in B6C3F1 mice, although its ability to cause liver cancer in humans has been discounted by findings in numerous laboratory investigations (Bull, 2000). The possible role of DCA in human liver cancer is even more controversial (Walgren et al., 2005; Caldwell and Keshava, 2006; Klaunig et al., 2007).
The GSH conjugation pathway is quite similar qualitatively, but not quantitatively, in rats and humans. The initial step in this second, minor pathway involves conjugation of TCE with GSH to form S-(1,2-dichlorovinyl)GSH (DCVG). DCVG formation occurs primarily in the liver at a rate about 10 times greater in rats than in humans (Green et al., 1997). Much of the DCVG is excreted via the bile into the intestines and converted to S-(1,2-dichlorovinyl)-l-cysteine (DCVC). That metabolite is reabsorbed and taken up by the liver, where a portion is detoxified by N-acetylation. Bernauer et al. (1996) exposed rats and humans to TCE vapor at up to 160 ppm for six hours. The rats excreted eight times more N-acetyl-DCVC in their urine than did the human volunteers at each exposure level. Some DCVC is taken up by the kidneys and further metabolized by the enzyme β-lyase to S-(1,2-dichlorovinyl)thiol (DCVSH). DCVSH is then converted to highly reactive products, including DCVC sulfoxide (DCVCS), chlorothioketene, and thionoacylchloride (Lash et al., 2000a). Metabolic activation of DCVC to chlorothioketene was shown to occur 11 times more rapidly in rats than in humans (Green et al., 1997). Lash et al. (2001) also demonstrated that cultured rat renal cells were more sensitive to DCVC than human renal cells. Chlorothioketene and similarly unstable congeners are capable of covalently binding to renal cellular proteins and DNA. This results in genotoxicity and cytotoxicity, with ensuing regenerative hyperplasia and potentially renal cell cancer (RCC).
Modes of Carcinogenic Action in Target Tissues
Both metabolic pathways are implicated in the carcinogenicity of TCE: reactive metabolite(s) of the GSH pathway in kidney tumors in rats and oxidative metabolites in liver and lung tumors in mice. That tumor formation in many cases is species-, strain-, sex-, and route of exposure–dependent has provided clues as to TCE’s modes of carcinogenic action. Whereas substantial progress has been made on the mechanistic front, the reader should not infer from the text that follows that all modes of action are known with absolute certainty.
It is well established that TCE, when given chronically in very high doses by gavage, can produce an increased incidence of hepatocellular carcinoma in B6C3F1 mice, but not in other strains of mice or in rats. This differential susceptibility can be explained in part by the greater capacity of the mouse to bioactivate relatively large quantities of TCE via the oxidative pathway. The B6C3F1 mouse produces a substantially larger amount of TCA after TCE exposures than do unresponsive strains of mouse, rats, or humans. The susceptibility of the B6C3F1 mouse is also likely related to the high (42.2%) incidence of liver adenoma/carcinoma in male controls (Haseman et al., 1998). This phenomenon may be due to an abnormally high population of spontaneously initiated cells in these animals’ liver. Mice express very low levels of epoxide hydrolase (Lorenz et al., 1984), the enzyme that catalyzes the hydrolytic degradation/detoxification of reactive epoxide metabolites of TCE and other VOCs.
CH induces hepatic tumors in male B6C3F1 mice, but not in F344 rats (Leaky et al., 2003). Female B6C3F1 mice gavaged with up to 100 mg CH/kg per day for 104 weeks showed no increase in liver tumors, but the male mice did exhibit increased incidences of hepatoma and/or hepatocellular carcinoma. CH is rapidly converted to TCA and TCOH in rodents and humans. Merdink et al. (2008) detected only trace amounts of DCA in blood and urine of male human subjects dosed with 500 or 1500 mg of CH. An epidemiology study of the possible association of short-term clinical administration of CH as a sedative–hypnotic and cancer risk in 2290 patients was conducted by Haselkorn et al. (2006). The authors concluded there was no persuasive evidence of a causal relationship between CH and cancer in humans.
TCA and DCA are the most likely candidates for proximate hepatocarcinogens produced by TCE’s oxidative pathway. TCA is a species-specific carcinogen that induces peroxisome proliferation and hepatocellular carcinoma in male and female B6C3F1 mice when administered in very high doses in drinking water or by gavage. It does not produce liver tumors in any strain of rats tested under these conditions (NAS, 2006, 2009). Very high doses of DCA, however, produce hepatic tumors in both B6C3F1 mice and F344 rats. Large, repeated doses of DCA and TCA initially stimulate, and then depress the growth of normal hepatocytes (Bull, 2000). This may confer a growth advantage to initiated cells, and is referred to as negative selection. At high tumorigenic doses, DCA (but not TCA) is thought to stimulate cell replication within liver tumors. If indeed both DCA and TCA contribute to tumorigenesis, findings by Bull et al. (2002) indicate they do so by distinct mechanisms. DCA-promoted liver tumors differed phenotypically from those promoted by TCA. DCA- and TCA-induced tumors also differ as to whether their depression of cytosine methylation is reversible on cessation of treatment (Tao et al., 1998). This is particularly important as DNA hypomethylation, including that of the proto-oncogenes c-jun and c-myc, may be an epigenetic mechanism for the tumorigenicity of DCA and TCA (Tao et al., 2004). It appears that hypomethylation due to TCA and DCA induces DNA replication and prevents the methylation of newly synthesized strands of DNA (Ge et al., 2001). Tao et al. (2000) have reported that DCA and TCA do so by virtue of their depletion of S-adenosylmethionine, which normally supplies the methyl group for the methylation process. It should be recognized that these are not genotoxic mechanisms. The effects disappear as soon as these metabolites are eliminated.
A primary mode of action of TCA and to a smaller extent DCA is activation of PPARα. As peroxisomes contain a variety of oxidative enzymes, PPARα activation produces oxidative stress, which can be manifest as lipid peroxidation, oxidative DNA damage, and transcription factor activation (O’Brien et al., 2005). Stimulation of PPARα can enhance DNA replication, resulting in expansion of some clones of hepatocytes and suppression of apoptosis, so initiated and precancerous cells will be spared. Male wild-type mice dosed orally with TCE exhibit hepatocyte proliferation and changes in expression of genes involved in cell growth (Laughter et al., 2004). PPARα-null mice are refractory to those effects, which are associated with carcinogenesis. Mice expressing human PPARα fail to show increases in markers of cell proliferation and are resistant to liver cancer if treated with PPARα agonists (Morimura et al., 2006; Yang et al., 2008). The concentration of PPARα in human cells is about 10% of that in the livers of rodents (Palmer et al., 1998; Klaunig et al., 2003). Many toxicologists have judged that the mode of action for hepatic carcinogenesis observed in mice after administration of peroxisome proliferation–inducing drugs and other chemicals, such as TCA, makes it unlikely that such chemicals pose a hepatic cancer risk in humans (Cattley et al., 1998; Clewell and Andersen, 2004; Klaunig et al., 2007; Gonzalez and Shah, 2008). It was concluded by an expert review panel that the PPARα mode of action for liver cancer in mice is not relevant to humans (NAS, 2006). However, others have raised questions about the interpretation of PPARα actions and whether it is the only relevant mode of action for such chemicals (Keshava and Caldwell, 2006). This continues to be a subject of active debate (NAS, 2008).
It is important to recognize that stimulation or inhibition of cell growth through PPARα activation ceases when the metabolites are eliminated (Miller et al., 2000). Thus, such alteration of cell signaling is not a genotoxic mechanism of action. Very high concentrations of DCA and CH have a weak genotoxic action in vitro. Bull (2000) and Moore and Harrington-Brock (2000), however, conclude that it is unlikely that those metabolites would cause tumors in any organ through genotoxicity or mutagenicity at exposure concentrations relevant to humans.
As discussed in the review paper by Green (2000), inhaled TCE is carcinogenic to the mouse lung but not to that of the rat. Oral TCE is not carcinogenic to the lung, probably due to first-pass hepatic metabolism that limits the amount of TCE reaching the lungs. The primary target of TCE in the mouse lung is the nonciliated Clara cell. Cytotoxicity is characterized by vacuolization and increased replication of these cells in the bronchiolar epithelium. A dose-dependent reduction in the CYP activity in Clara cells is observed as well. This loss of metabolic activation capacity can be thought of as an adaptive response. Clara cells recover morphologically during repeated daily inhalation exposures to TCE.
Chloral is the putative toxicant responsible for pulmonary tumor formation. Clara cells of the mouse efficiently metabolize TCE to chloral. Chloral accumulates, due to its efficient production and low activity of ADH, the enzyme responsible for its reduction to TCOH. The Clara cells’ lack of glucuronosyltransferase, the enzyme that normally catalyzes the formation of TCOH–GLU, has also been implicated in chloral accumulation. Species differences in susceptibility of the lung to TCE are due in part to mouse lung Clara cells having a much higher level of CYP2E1 than those of the rat, and thus a much higher capacity to metabolize TCE to chloral. Also, Clara cells in mice are much more numerous than in rats. Clara cells are rare in human lungs. A critical role for chloral is supported by the findings that its administration to mice, but not TCA or TCOH, causes Clara cell toxicity identical to that of TCE. Chloral does appear to have some genotoxic potential, especially in regard to inducing aneuploidy. However, the fact that tumors are not seen in species where cytotoxicity does not occur strongly implicates cytotoxicity and reparative proliferation in tumor formation. In an effort to test the hypothesis that bronchiolar damage by TCE is associated with bioactivation within Clara cells, Forkert et al. (2006) administered TCE i.p. to CD-1 mice. The result was dose-dependent production of dichloroacetyl lysine adducts in Clara cells (used as an in vivo marker of TCE metabolism) that correlated with bronchiolar damage. The work also suggested that CYP2F2 may play a more important role than CYP2E1 in TCE metabolism and cytotoxicity within the mouse lung (Forkert et al., 2005).
TCE was given in corn oil to F344/N rats and B6C3F1 mice of both sexes by gavage five times weekly at doses up to 1000 mg/kg in rats and 6000 mg/kg in mice in a 13-week study, as well as up to 1000 mg/kg to both species and sexes in a 103-week study (NTP, 1990). A low, but statistically significant increase in renal tumor incidence was observed only in the male rats given TCE at 1000 mg/kg for two years. Two-year gavage studies of TCE, in four additional rat strains, were also conducted (NTP, 1988). In all strains of rats tested, cytomegaly and karyomegaly of tubular cells in the renal corticomedullary region were seen. Frank toxic nephropathy was observed with higher frequency in male rats beginning at 52 weeks of exposure. Renal adenomas or adenocarcinomas were occasionally seen in male rats of different strains after two years of the repetitive, high-dose oral exposure regimen.
Adverse effects of TCE on the kidneys are due largely to metabolites formed via the GSH conjugation pathway (Lash et al., 2000b). As described previously, conjugation of TCE with GSH to form DCVG occurs primarily in the liver. DCVG is secreted into bile and blood. That in the bile is converted in the gut to DCVC, which is reabsorbed into the bloodstream. Humans have a lower capacity than rats to metabolize TCE by the GSH pathway. Lash et al. (1999) were able to detect DCVG in the blood of humans who inhaled TCE at 50 or 100 ppm for four hours, but Bloeman et al. (2001) could not find DCVG or DCVC in the urine of similarly exposed subjects. DCVG in the blood is taken up by the kidneys and metabolized to DCVC by γ-glutamyltransferase and a dipeptidase. Lash et al. (2001) observed the following decreasing order of toxic potency in freshly isolated rat cortical cells: DCVC > DCVG >> TCE. DCVC can be detoxified by acetylation or activated further by two pathways: (1) cleavage by renal cytosolic and mitochondrial β-lyases to dichlorothioketene, which in turn can lose a chloride ion to yield chlorothioketene or tautomerize to form chlorothionacyl chloride (the latter two moieties are very reactive and acylate proteins and DNA); and (2) oxidation by renal CYPs or flavin-containing monooxygenases to DCVCS, a reactive epoxide. Lash et al. (1994) reported that DCVCS was a more potent nephrotoxicant than DCVC in vitro and in vivo in rats. Apoptosis was observed after as little as one hour of incubation of cultured human renal proximal tubular (RPT) cells with DCVC and DCVCS (Lash et al., 2003, 2005). Cellular proliferation, accompanied by increased expression of proteins associated with cellular growth, differentiation, stress and apoptosis, was also an early response to low doses. Necrosis, however, was a late, high-dose phenomenon. Exposure of human RPT cells to DCVC at lower concentrations for 10 days also resulted in expression of genes associated with cell proliferation, apoptosis, and stress (Lash et al., 2005), as well as repair and DCVC metabolism (Lash et al., 2006).
Animal studies typically provide insight into mode of action, but in the case of TCE-induced RCC, human studies have been of significant value. Bruning et al. (1997a) analyzed tumor tissues from 23 RCC patients with occupational histories of long-term, high-level TCE exposure. Tumor cell DNA was isolated and analyzed for somatic mutations of the von Hippel–Lindau (VHL) tumor suppressor gene. Compared with VHL gene mutation rates of 33% to 55% in unexposed RCC patients, all 23 TCE-exposed RCC patients exhibited aberrations of the VHL gene. In a follow-up study, Brauch et al. (1999) sought to determine whether TCE produced a specific mutation of the VHL gene. These investigators analyzed VHL gene sequences in DNA isolated from RCC tissues from patients exposed to high levels of TCE in metal-processing factories. Renal cell tumors of TCE-exposed patients showed somatic VHL mutations in 33 of 44 cases (75%). Of the 33 cases with VHL mutations, a specific mutational hot spot at VHL nucleotide 454 was observed in 13 cases. The nucleotide 454 mutation was not found in any of the 107 RCC patients without TCE exposure or among 97 healthy subjects, 47 of whom had a history of TCE exposure. These data suggest that the VHL gene may be a specific and susceptible target of reactive GSH pathway metabolites, a concept strengthened by a more recent study reporting VHL nucleotide 454 mutation among TCE-exposed but not in nonexposed RCC patients (Brauch et al., 2004). More recently, however, Charbotel et al. (2007) found no associations between the number and type of VHL mutations in TCE-exposed and unexposed RCC patients. The subject awaits further clarification (Chow and Devesa, 2008).
The aforementioned reactive products of DCVC are both cytotoxic and mutagenic. Adverse effects on proximal tubule cells include alkylation of cytosolic and mitochondrial structural and enzymatic proteins, oxidative stress, marked ATP depletion, and perturbations of calcium homeostasis. Tubular necrosis ensues, with subsequent proliferation that can alter gene expression, which may modify cell growth and differentiation. Genes associated with stress, apoptosis, repair, and proliferation were upregulated almost twofold in cultured human renal tubular cells exposed to subtoxic doses of DCVC for 10 days (Lock et al., 2006). Mechanisms of noncarcinogenic and carcinogenic action are discussed in detail by Lash et al. (2000b) and Lock and Reed (2006).
Bruning and Bolt (2000) opine that reactive metabolite(s) of the GSH pathway may have a genotoxic effect on the proximal tubule of the human kidney, but that full development of a malignant tumor requires a promotional effect such as cell proliferation in response to tubular damage. If this is true, RCC secondary to TCE exposure would be a threshold response. The question of whether chronic tubular damage is a prerequisite to renal tumor formation is quite important. Evidence has come from the use of electrophoresis to examine protein excretion patterns in the urine of RCC patients with and without a history of chronic, high-level TCE exposure (Bruning et al., 1996). Protein excretion patterns indicative of tubular damage were identified in all their 17 TCE-exposed cases, but in only about one half of 35 controls. Bruning et al. (1999) subsequently published the results of a larger study supportive of this concept. Approximately 93% of 41 RCC patients with high TCE exposure exhibited elevated urinary α1-microglobulin excretion versus 46% of 50 RCC patients without a history of TCE exposure. Similar findings were reported in an updated study by Bolt et al. (2004), and in an investigation by Green et al. (2004) of 70 electronics workers who inhaled an average of 32 ppm TCE for four years.
Results of an investigation by Mally et al. (2006) provide additional insight into the TCE renal carcinogenesis threshold premise. A strain of rats (Eker) uniquely susceptible to renal carcinogens was administered 100, 250, 500, and 1000 mg TCE/kg by gavage five days per week for 13 weeks. The Eker rat is a unique animal model for RCC, carrying a germ-line alteration of the Tsc-2 tumor suppressor gene. Results showed a significant increase in cell proliferation in renal tubular cells but no increase in preneoplastic renal lesions or tumor incidence. In vitro studies were conducted on primary Eker rat renal epithelial cells by exposing them to the TCE metabolite DCVC dissolved in water at 10 to 50 μM for eight, 24, and 72 hours. Concentrations of DCVC that reduced rat renal cell survival to 50% also resulted in cell transformation. No carcinogen-specific mutations were identified in the VHL or Tsc-2 tumor suppressor genes in the transformed cells. RCCs in the Eker rat have substantial similarities to human RCC. It is not entirely clear that this or any contemporary experimental animal model adequately mirrors humans with regard to the effects of TCE-induced mutations in the VHL gene, but the authors firmly suggest that TCE-mediated renal carcinogenicity may occur only secondarily to nephrotoxicity and sustained regenerative cell proliferation.
Metabolism of TCE by the GSH pathway is similar qualitatively, but not quantitatively in rats and humans. N-Acetyl-DCVC, the major detoxification product of DCVC, was found in the urine of humans and rats after six hours of inhalation of up to 160 ppm TCE (Bernauer et al., 1996). Cumulative excretion of the N-acetyl derivative was seven- to eight-fold higher in the rats. Bruning et al. (1997b) originally reported an increased likelihood of RCC patients with high TCE exposure having a functional GST isozyme GSTT1 or GSTM1 genotype. This isozyme GSTT1 is thought to be primarily involved in TCE metabolism, while GSTM1 is thought to detoxify epoxides. A reassessment by the investigators of a larger population, however, revealed no such relationships (Wiesenhutter et al., 2007). Moore et al. (2010), recently completed an assessment of an even larger population. There was an increased risk of RCC in workers with at least one intact GSTT1 allele, but not in persons with two deleted alleles (ie, null genotype).
It is also worthy of note that male rats and humans are apparently at greater risk of TCE-induced kidney cancer than their female counterparts. Blood DCVG concentrations were 3.4-fold higher in male than in female volunteers inhaling 50 or 100 ppm TCE (Lash et al., 1999). Male rats display higher GSH conjugation, γ-glutamyl transpeptidase activity, and cysteine conjugate β-lyase activity than female rats. Taken together, results of the cited studies indicate that both male humans and male rats possess GSH conjugation capacity and can produce the critical TCE metabolite DCVC. Renal carcinoma has been observed in male rats and male workers when both have been exposed to very high TCE concentrations for prolonged periods of time. These observations show data congruence, indicating that the conjugation pathway plays a central role in induction of renal carcinoma in males of both species. As described previously, rats have a significantly greater capacity to metabolically activate TCE to DCVC by this pathway than do humans. Rat renal cortical cells, in turn, are more susceptible to injury by DCVC than their human counterparts.
Cancer Epidemiology Studies
There have been many published studies of cancer incidence and mortality in TCE-exposed populations. Most of the epidemiology studies in the United States prior to 2000 involved workers in the aircraft maintenance and manufacturing industries. There were also investigations of Swedish, Finnish, German, and Danish worker cohorts. Results of these assessments have been mixed, ranging from no association to limited evidence. The major studies with some exposure data constituted what the Wartenberg et al. (2000) meta analysis referred to as Tier I studies, which received a greater weighting when making causal inferences than lower tier cohort, case–control, and community-based investigations. Among the Tier I studies, evidence for an excess incidence of cancer was strongest for kidney (rate ratios [RR] = 1.7, 95% confidence interval [CI] = 1.1–2.7), liver (RR = 1.9, CI = 1.0–3.4), and non-Hodgkin lymphoma (NHL) (RR = 1.5, CI = 0.9–2.3) (Wartenberg et al., 2000).
The report of Henschler et al. (1995) was the first in a series of German studies that have provided some of the strongest evidence to date for an association between TCE and RCC. These authors described a cohort of male cardboard factory workers who were exposed to moderate to extremely high concentrations of TCE vapor. By the closing date of the study, five of the 169 exposed workers had been diagnosed with kidney cancer versus none of the 190 controls. This resulted in standardized incidence ratios (SIRs) of 7.97 (CI = 2.59–18.59) and 9.66 (CI = 3.14–22.55), using Danish and German Cancer Registry data for comparison, respectively. German researchers also conducted a hospital-based case–control study with 58 RCC patients and 84 patients from accident wards who served as controls. Of the 58 RCC patients, 19 had histories of occupational TCE exposure of at least two years, compared with only 5 of the controls. After adjustment for potential confounders, an association between RCC and long-term exposure to TCE was reported (odds ratio [OR] = 10.80, CI = 3.36–34.75) (Vamvakas et al., 1998). In an expanded German case–control study of 134 RCC cases and 401 controls, excess risks for those working longest with TCE (OR = 1.8, CI = 1.01–3.20) and those experiencing narcotic symptoms (OR = 3.71, CI = 1.80–7.54) were reported (Bruning et al., 2003). Narcosis was thought to be associated with peak exposures. These workers reported frequent dizziness, requiring them to seek fresh air several times daily. More recent investigations have generally involved surveys of populations with lower exposures. The results of these investigations have been suggestive of an association between TCE and RCC, although some findings were negative or did not reach statistical significance (Alexander et al., 2006; Boice et al., 2006; Chang et al., 2005; Mandel et al., 2006; Raaschou-Nielsen et al., 2003; Wong, 2004). A number of these assessments involved a limited number of subjects. Charbotel et al. (2006) studied 86 RCC patients and 316 matched controls from an area in France with high TCE exposures in local industries. There was a 64% increase in RCC risk with TCE exposure, with the risk doubling in persons with a high cumulative dose, increasing more when peak exposures were also taken into account (OR = 2.74, CI = 1.06–7.07). The OR was still high, but not statistically significant after adjusting for exposure to cutting fluids, because of the study’s lack of power. Chow and Devesa (2008) reviewed recent epidemiological evidence of a rising incidence of RCC in the United States. Cohort studies showed associations with smoking, obesity, diminished physical activity, hypertension, and certain chemical exposures.
Epidemiological evidence of NHL or cancer of the liver, lung, or other organs has been weaker than that for the kidney. There appeared to be a dose-dependent increase in the SIR for NHL with increased duration of TCE exposure in Danish workers (Raaschou-Nielsen et al., 2003). These researchers also reported a SIR of 1.7 for esophageal adenocarcinoma. Seidler et al. (2007) described the association between TCE exposure of >35 ppm per year and malignant lymphoma (OR = 2.1, CI = 1.0–4.8) as being of borderline statistical significance. Lan et al. (2010) recently observed a dose-dependent decline in major types of lymphocytes in TCE-exposed workers in China. Little or no association with NHL incidence was seen by Boice et al. (2006) or Zhao et al. (2005) in aerospace employees. Evidence of lung cancer in persons occupationally exposed to TCE is limited to nonexistent (Boice et al., 2006; Hansen et al., 2001; Zhao et al., 2005; Raaschou-Nielsen et al., 2003). The latter group of scientists’ estimates of SIRs for lung cancer, for example, were 1.4 for men and 1.9 for women. Most estimates of liver cancer risk in TCE-exposed workers have also been low. SIRs of 2.6 (Hansen et al., 2001) and 2.8 (Raaschou-Nielsen et al., 2003), resulting from a small number of cases, were among the highest reported.
Attempts to understand the mechanistic underpinnings of TCE’s carcinogenicity in rodent models and their relevance to humans have resulted in a massive body of published data. TCE provides a stellar example of how experimental data from laboratory and epidemiological studies may or may not impact cancer risk assessment in a regulatory context. The EPA began in the 1990s to revise its guidelines for cancer risk assessment. The final EPA Guidelines for Carcinogen Risk Assessment were released in March 2005. TCE was utilized as a pilot chemical for evaluation and implementation of this guidance document. The guidelines emphasize a scientific “weight of evidence” approach that includes characterization of dose–response relationships, modes of action, and metabolic/TK processes. Where adequate data are available to support reversible binding of the carcinogenic moiety to biological molecules as the initiating event, a nonlinear (ie, threshold) risk assessment/approach is to be used. Otherwise, the default assumption of a linear (ie, no-threshold) model/approach is to be used to estimate cancer risk. As described above, there is considerable evidence that TCE’s oxidative metabolites act via nongenotoxic modes of action. PPARα activation and induction of mouse liver tumors were judged to be irrelevant to humans (NAS, 2006). The EPA, in its 2001a draft TCE Risk Assessment, concluded it was difficult to establish with sufficient certainty the important TCE metabolites, the key events they cause, and their relevance to humans. Despite a substantial increase in information in the last decade, EPA scientists (Caldwell and Kesheva, 2006) contend that knowledge of mechanisms and human relevance is still insufficient to depart from the default assumption. This is exemplified by the critical response of EPA scientists (Caldwell et al., 2006) to the attempt of Clewell and Andersen (2004, 2006) to apply a margin of exposure approach (nonlinear dose–response extrapolation) in their TCE risk assessment. In contrast to liver cancer, kidney cancer is widely accepted to be qualitatively similar in rats and humans, although rats form greater quantities of reactive metabolites via the GSH pathway. Despite genotoxic events, kidney tumor formation in humans is generally believed to require promotion resulting from frank cytotoxicity. Caldwell and Kesheva (2006), however, opine that there may be other modes of action of multiple metabolites operative at low doses. This logic has been retained in the EPA’s most recent IRIS Toxicological Review of TCE (2009).
The EPA (2001a) draft TCE Risk Assessment elicited considerable debate, which prompted the EPA and other federal agencies to request a scientific review of the document by a NAS expert panel. Their report was released in 2006. It concluded there was concordance between rat and human studies of renal carcinogenicity. The preponderance of evidence indicated that humans would be much less susceptible than mice to liver and lung carcinogenesis. It was recommended, among other things, that a new meta-analysis of epidemiological data be conducted. Kelsh and his coworkers recently completed meta-analyses of occupational study data. Studies were classified as group I or II, depending on the quality of their design and exposure assessment. The summary relative risk estimate (SRRE) across all such studies was 1.42 (CI = 1.17–1.77) for kidney cancer (Kelsh et al., 2010). The same research group performed meta-analyses for occupational TCE exposures and liver cancer (Alexander et al., 2007) and NHL (Mandel et al., 2006). The highest SSREs calculated for liver cancer and NHL in any study groupings were 1.30 and 1.59, respectively. The authors noted the results of many studies were inconsistent, information on TCE exposure was often quite limited, and recognized confounding factors were not always taken into account. NAS (2009) recently evaluated published occupational epidemiology studies, and concluded there was limited/suggestive evidence of associations between chronic TCE exposure and kidney cancer. Evidence was deemed inadequate/insufficient to determine whether there was an association with NHL or cancer of the liver, lung, or any other organ. In contrast, Scott and Chiu (2006) opined that modest RR elevations (1.5–2.0) typically reported in positive epidemiology studies provided support for the kidney, liver, and lymphatic systems as target organs. The EPA (2009), in its IRIS document released in September 2011, concluded there was convincing evidence of a causal association between TCE exposure and kidney cancer. Epidemiological evidence was said to be less compelling for NHL, and even more limited for liver cancer. Nevertheless, TCE was characterized by the EPA as “carcinogenic in humans by all routes of exposure.” This is a departure from previous national and international classifications.
Tetrachloroethylene (perchloroethylene, PERC) is commonly used as a dry cleaner, fabric finisher, degreaser, rug and upholstery cleaner, paint and stain remover, solvent, and chemical intermediate. The highest exposures usually occur in occupational settings via inhalation. Much attention is being focused on adverse health effects that may be experienced by dry cleaners and other persons living in the proximity of such facilities (Garetano and Gochfeld, 2000). Echeverria et al. (1995), for example, reported adverse effects on visuospatial functions in dry cleaners. PERC is frequently detected in the low ppt range in the breath and blood of the general populace (Ashley et al., 1994; Churchill et al., 2001; Blount et al., 2006). Although releases are primarily to the atmosphere, PERC enters surface and groundwaters by accidental and intentional discharges (ATSDR, 1997c). Levels in the ppb range were reported in municipal water in areas of New England, where PERC was used in a process to treat plastic water pipe (Paulu et al., 1999). Pohl et al. (2008) reported that PERC was the third most frequently found chemical contaminant in groundwater at hazardous waste sites in the United States. A dry cleaner adjacent to Camp Lejeune, in North Carolina, was the source of PERC contamination of some wells that supplied drinking water to the marine base. This contamination prompted a recent study of potential health effects in employees and persons stationed at the base (NAS, 2009).
The systemic disposition and metabolism of PERC and TCE are similar in many respects, although PERC is much less extensively metabolized (ATSDR, 1997c; Chiu et al., 2007). Both chemicals are well absorbed from the lungs and GI tract, distributed to tissues according to their blood flow and lipid content, partially exhaled unchanged, and metabolized. PERC, like TCE, is metabolized by CYP-catalyzed oxidation and GSH conjugation. CYP2E1, however, is not thought to play a major role, in that PERC is considered to be oxidized primarily by the CYP2B family in the rat (Hanioka et al., 1995). In humans, CYP2B6 is the primary isoform responsible for PERC metabolism, and there are minor contributions by CYP1A1 and CYP2C8 (White et al., 2001). The initial metabolite is the epoxide PERC oxide. This metabolic intermediate can be biotransformed to several products (Lash and Parker, 2001). The primary one is trichloroacetyl chloride, which reacts with water to form TCA, the predominant PERC metabolite found in the urine of rodents and humans (Birner et al., 1996; Volkel et al., 1998). Some TCA is converted to DCA.
A small proportion of absorbed PERC undergoes conjugation with GSH to form S-(1,2,2-trichlorovinyl)GSH (TCVG). That initial metabolic step is catalyzed by GSTs and occurs primarily in the liver. TCVG is converted to S-(1,2,2-trichlorovinyl)-l-cysteine (TCVC). TCVC, like the DCVC formed from TCE, is both detoxified in the liver by N-acetylation and metabolically activated by β-lyases in the kidneys. 2,2-Dichlorothioketene can decompose to DCA. Hence, DCA is derived from both GSH- and CYP-dependent biotransformation of PERC. PERC is conjugated with GSH more extensively by rats (1%–2%) (Dekant et al., 1986) than is TCE (<0.005%) (Green et al., 1997). The extent of GSH conjugation of PERC increases when the oxidative pathway begins to become saturated at high exposure levels. Metabolic products of GSH conjugates of TCE and PERC are primary contributors to these halocarbons’ nephrotoxicity (Lash et al., 2007).
Modes of Cytotoxicity/Carcinogenicity
PERC-induced hepatic injury is believed to be a consequence of its oxidative metabolism (Lash and Parker, 2001). Although PERC is more poorly metabolized by CYPs than TCE, two additional intermediate metabolites of PERC contribute to its hepatocytotoxicity: the initial oxidation product, PERC oxide, and one of its convertants, trichloroacetyl chloride. TCA is primarily responsible for activation of the nuclear receptor PPARα, which stimulates peroxisomal enzymes and selected CYPs involved in lipid metabolism. This results in peroxisome proliferation, which generates reactive oxygen moieties that can cause lipid peroxidation, cellular injury, and altered expression of cell-signaling proteins (Bull, 2000). Lash et al. (2007) more recently demonstrated that CYP inhibition resulted in reduced injury of hepatocytes isolated from male F344 rats and exposed to PERC. GSH depletion increased cellular injury, apparently because of a shift from the GSH to the oxidative pathway.
The metabolism and mode of nephrotoxicity of PERC and TCE appear to be quite similar, although PERC and its metabolites are somewhat more potent. Renal effects of both halocarbons are due primarily to metabolites formed via the GSH pathway (Lash and Parker, 2001). The sites, enzymes, and products associated with PERC biotransformation are almost identical to those associated with TCE. The primary difference is that TCVG and TCVC are produced from PERC, and DCVG and DCVC are formed from TCE. TCVC can be detoxified by acetylation or cleaved by renal cytosolic and mitochondrial β-lyases to trichlorothioketene, which loses a chloride ion to form dichlorothioketene. The latter is a very reactive moiety that binds to cellular proteins and DNA. TCVC, as noted above, can be enzymatically oxidized to form the very reactive TCVC sulfoxide (TCVCS) (Krause et al., 2003). TCVCS was shown to be more nephrotoxic than TCVC in male rats after i.p. injection (Elfarra and Krause, 2007). TCVC caused more pronounced necrosis of RPT cells in male rats than did DCVC after i.v. injection (Birner et al., 1997). Lash et al. (2002) similarly found that PERC and TCVG were more toxic than TCE and DCVG in vivo to renal cortical cells of F344 rats. Cells from male rats were more sensitive than cells from females to PERC- and TCVG-induced mitochondrial state 3 respiratory inhibition and cytotoxicity. Isolated rat hepatocytes and their mitochondria, however, were unaffected by PERC and TCVC. Elevated GSH levels enhanced TCE-induced and PERC-induced cytotoxicity in suspensions of rat renal cortical cells but not hepatocytes (Lash et al., 2007). Thus, PERC’s GSH metabolites are both sex- and organ-specific.
PERC, like TCE, has a limited ability to adversely affect the liver or kidneys of rodents. Near-lethal i.p. doses of PERC were required to cause acute liver injury in mice (Klaassen and Plaa, 1966). Buben and O’Flaherty (1985) saw manifestations of modest hepatocellular damage in male mice given 500 to 2000 mg PERC/kg daily for six weeks by corn oil gavage. A lack of dose dependence reflected saturation of metabolic activation in this dosage range. Philip et al. (2007) reported that male mice given 150 to 1000 mg PERC/kg by aqueous gavage initially exhibited a dose-dependent increase in serum alanine aminotransferase (ALT) activity, but levels of the enzyme regressed substantially over a 30-day dosing period. Hayes et al. (1986) found no consistent dose-related changes in any clinical chemistry parameter in male or female rats that consumed up to 1440 mg PERC/kg in their drinking water for 90 days. Liver injury was not seen in B6C3F1 mice or Osborne–Mendel rats gavaged with up to ~1000 mg/kg daily for 78 weeks (NCI, 1977a,b). Dose-dependent karyomegaly was observed in the renal proximal tubule (RPT) epithelium of mice and rats inhaling PERC for 103 weeks (NTP, 1986). This change was most prominent in the male rats. Tinston (1995) reported mild, progressive glomerulonehropathy and increased pleomorphism of renal tubular nuclei in male but not female rats that inhaled 1000 ppm PERC for up to 19 weeks. Green et al. (1990) found increases in protein droplets and cell proliferation in the proximal tubular epithelium of male F344 rats gavaged with 1500 mg PERC/kg daily for up to 42 days. These alterations were accompanied by an increase in α2u-globulin, a male rat-specific protein (Fig. 24-6). Goldsworthy et al. (1988) reported similar findings in male but not female F344 rats gavaged for 10 days with 1000 mg PERC/kg per day. The accumulation of α2u-globulin is cytotoxic, causing cellular necrosis and compensatory cellular proliferation in the P2 segment of renal proximal tubules (Borghoff et al., 1990).
Proposed mechanism of solvent-induced kidney cancer in male rats involving α2u-globulin. (Used with permission from Borghoff et al., 1996a.)
Humans should be less susceptible to hepatorenal injury by PERC than rodents, due to lower target organ doses of the parent compound and its bioactive metabolites. Rats achieve a substantially higher internal dose of VOCs than do humans on inhalation exposures. Volkel et al. (1998) subjected rats and people to identical PERC inhalation regimens. Blood TCA concentrations were 3 to 10 times higher in the rats. DCA was not detectable in human urine, but substantial amounts were found in rat urine. A study of the urinary excretion of total trichloro-metabolites by PERC-exposed workers led Ohtsuki et al. (1983) to conclude that the capacity of humans to metabolize PERC was rather low. Lash and Parker (2001) noted that saturation of PERC metabolism occurred at lower doses in humans than in rodents. This indicates that humans have lower capacity to form biologically active metabolites from moderate to high PERC doses. The difference is reflected in the finding of much lower concentrations of protein adducts in the blood of humans than in the blood of rats subjected to equivalent PERC inhalation exposures (Pahler et al., 1999).
PERC has limited ability to damage the liver of humans. Stewart et al. (1977) found no evidence of hepatotoxicity in six male and six female volunteers exposed randomly to PERC at 0, 25, or 100 ppm 5.5 hours per day, five days per week for 11 weeks. Serum ALT activity was not increased in 22 dry cleaners examined by Lauwerys et al. (1983). A research group in Italy studied 141 employees exposed to PCE in small laundries and dry-cleaning shops (Gennari et al., 1992). No worker exhibited clinical signs of hepatic dysfunction or abnormal serum enzyme concentrations, although there did appear to be an increase in one isozyme of γ-glutamyltransferse, which was said to be associated with hepatobiliary impairment. Another investigation of dry cleaners failed to reveal increases in serum enzymes, but did show mild to moderate changes in hepatic parenchyma revealed by ultrasonography (Brodkin et al., 1995). Thus, considerable experience in occupational settings demonstrates that humans may develop mild but reversible liver injury on chronic exposure to high concentrations of PERC (ATSDR, 1997c).
Occupational exposures to PERC vapor have led to several reports of mild renal tubular damage (ATSDR, 1997c). Employees of dry-cleaning shops have been the subjects of a number of studies of potential kidney effects. Increased concentrations of urinary lysosomal or β-glucuronidase activity were described in dry cleaners exposed to 10 ppm (Franchini et al., 1983) and 23 ppm PERC (Vyskocil et al., 1990) for 9 to 14 years. In a more comprehensive study, a number of urinary indices indicative of early glomerular and tubular changes were increased in 50 dry cleaners who inhaled ~15 ppm PERC for 10 years (Mutti et al., 1992). Verplanke et al. (1999) monitored several indices of tubular and glomerular function in Dutch dry-cleaning workers, but found an increase only in retinol-binding protein in their urine. Other investigators have failed to find evidence of renal effects in such populations. A laboratory study of 10 male and 10 female adults, who inhaled PERC at up to 150 ppm for as long as 7.5 hours per day for five days, did not show changes from preexposure baseline BUN levels (Stewart et al., 1981). Hake and Stewart (1977) described a dry cleaner who was found unconscious in a pool of PERC, where he had been for an estimated 12 hours. Laboratory tests revealed hematuria and proteinuria that lasted for 10 and 20 days, respectively. Mild hepatic damage was revealed by transient increases in serum enzymes. On the basis of the foregoing occupational experiences, PERC has shown limited ability to cause diffuse changes along the nephron, although extremely high exposures can lead to pronounced changes in humans.
Cancer Bioassays in Rodents
High, chronic doses of PERC have been demonstrated to produce species-specific and in certain cases strain- and gender-specific tumors in some organs of mice and rats. Male and female B6C3F1 mice, gavaged with very high doses of PERC (stabilized with epichlorohydrin) for up to 78 weeks, exhibited an increase in hepatocellular carcinoma (NCI, 1977a). No increase in tumor incidence was seen in Osborne–Mendel rats. A significant increase in hepatocellular neoplasms (adenomas and carcinomas combined) was seen in B6C3F1 mice inhaling 100 or 200 ppm PERC for two years (NTP, 1986). There was also a low incidence of renal adenomas and carcinomas in male but not female F344 rats exposed for two years to 200 or 400 ppm PERC vapor. The incidence of renal neoplasms in the males was 1 of 49 controls, 3 of 49 exposed at 200 ppm, and 4 of 49 exposed at 400 ppm (NTP, 1986). Even though the changes were not statistically significant, it was noted that these particular tumors are rare in F344/N male rats, so they were believed to have been caused by PERC exposure. There was also a significant increase over controls in mononuclear-cell leukemia (MNL) in the male and female rats, but it was not clearly dose-dependent. The incidence of this form of leukemia can exceed 70% in F344 controls (Caldwell, 1999; Ishmael and Dugard, 2006). MNL in this rat strain apparently arises from large granular lymphocytes. This lukemic origin is very uncommon in humans (Caldwell, 1999). Gliomas were found in two female and four male F344/N rats exposed to PERC at 400 ppm and in one control male (NTP, 1986). The increase in incidence of this tumor was not statistically significant. Thus, the brain tumors were not considered by NTP (1986) to have been induced by exposure to PERC. The overall incidence of Leydig cell (testicular) tumors was 70%, 80%, and 82% in the 0-, 200-, and 400-ppm groups, respectively. Haseman et al. (1998) reported that NTP control F344 rats have an extremely high spontaneous incidence (89.1%) of Leydig cell tumors. Therefore, these tumors are believed to be irrelevant to men (Ishmael and Dugard, 2006). No increases in lung proliferative lesions seen in B6C3F1 mice of either sex after inhalation of PERC at 100 or 200 ppm for 103 weeks, nor were lung neoplasms seen in male or female F344/N rats exposed at 200 or 400 ppm for 103 weeks (NTP, 1986).
Cancer Epidemiology Studies
There have been many epidemiology studies of cancer incidence and mortality in groups of dry cleaners and other persons occupationally exposed to PERC (ATSDR, 1997c; NAS, 2009; EPA, 2008). Some researchers have reported findings of excess incidences of different cancers, while others have not. Frequently, there was not a determination of the degree or duration of PERC exposure, or consideration of major confounding factors (eg, exposure to other solvents/chemicals, smoking, alcohol consumption, socioeconomic status). Nevertheless, there was sufficient information for Weiss (1995) to conclude that cigarette smoking and alcohol consumption could only partially account for an increased rate of esophageal cancer in dry cleaners. In this instance kidney cancer incidence did not appear to be elevated. Blair et al. (2003) conducted and updated a mortality assessment of a cohort of dry cleaners and found an increased risk of death from esophageal cancer (standardized mortality ratio [SMR] = 2.2, 95% CI = 1.5-3.3). No exposure–response pattern was seen. Chang et al. (2005) did not find any cases of esophageal cancer in a Taiwanese cohort of electronics workers, while Lynge et al. (2006) observed a decreased incidence in dry cleaners in Nordic countries. Nevertheless, NAS (2009) concluded there was limited/suggestive evidence of an association between chronic PERC exposure and esophageal cancer.
There has been little evidence of an association between occupational PERC exposure and liver cancer. Increased incidences or mortality were not seen in cohort studies (Blair et al., 2003; Chang et al., 2005) or in a case–control study (Lynge et al., 2006). A RR of just 0.76 (CI = 0.38–1.52) was estimated in the latter investigation of Nordic dry-cleaning workers. A mortality odds ratio (MOR) of 2.57 (CI = 1.21–5.46) was reported by Lee et al. (2003) in another case–control study, but the exposure assessment was weak. The NAS (2009) recently concluded there was insufficient/inadequate evidence to determine whether an association exists between chronic PERC exposure and hepatobiliary cancer.
Epidemiology studies of populations exposed to PERC generally have shown no increase in risk of lung cancer. Blair et al. (2003), however, in an updated mortality analysis of dry cleaners, reported a SMR of 1.5 (CI = 1.2–1.9) for workers with medium to high exposures. This study adequately characterized exposures and included a relatively large number of subjects. Thus, the NAS (2009) concluded there is limited/suggestive evidence of an association between chronic PERC exposure and lung cancer.
Investigations of kidney cancer risk in PERC-exposed populations have yielded mixed results. A number of studies have not revealed an increased risk (Blair et al., 2003; Boice et al., 1999; Lynge et al., 2006; Mundt et al., 2003; Ruder et al., 2001). However, some of the assessments lacked power due to a limited number of study subjects. Although Blair et al. (2003) reported a very low SMR for kidney cancer deaths in dry cleaners with little or no PERC exposures, the SMR was 1.5 for those with medium or high exposure. Increased incidences have also been reported by Mandel et al. (1995), Pesch et al. (2000), and McCredie and Stewart (1993). The NAS (2009) concluded there is limited/suggestive evidence of an association between chronic PERC exposure and kidney cancer in humans.
A great deal of research has been conducted to characterize PERC’s dose–response relationships, modes of action, and metabolism/TK in rodents and humans. It is apparent from the previous discussion there is a considerable body of information available from human experience/exposures to PERC, as well as extensive data sets from studies in different species of laboratory animals. Both types of data are important in assessing risks to human health. Knowledge of the relevance of the animal data to humans is essential in order to meaningfully extrapolate from one species to another.
Humans appear to be less susceptible than rodents to the toxic or carcinogenic actions of PERC, as they are to those of TCE. Humans absorb less inhaled PERC and TCE, attain lower target organ doses of the parent compounds, have lower oxidative and GSH conjugation capacity, and inactivate epoxide intermediates more efficiently. On equivalent inhalation exposures of Wistar rats and humans to PERC, the rats excreted substantially larger amounts of TCA, DCA, and acetylated TCVC (Volkel et al., 1998). These observations were consistent with in vitro findings of greater conversion of PERC to TCVC by rat liver cytosol and of 10-fold higher β-lyase-mediated metabolism of TCVC by rat than by human kidney (Green et al., 1990). These data indicate that the human kidney has limited capacity to generate reactive metabolites from PERC by the GSH and β-lyase pathways. Biotransformation of PERC by the GSH conjugation pathway appears to be quite similar qualitatively, but not quantitatively in male rats and humans.
PERC-induced accumulation of α2μ-globulin in the P2 segment of renal proximal tubules and the accompanying cytotoxic and regenerative changes are unique to the male rat (Borghoff et al., 1990). α2μ-Globulin is synthesized under androgenic control in the liver of male rats, reabsorbed by the P2 segment, and undergoes hydrolytic digestion. Accumulation of α2μ-globulin in cytoplasmic droplets elicits tubular cell necrosis and compensatory cell proliferation (Goldsworthy et al., 1988). Sustained cell proliferation can promote clonal expansion of spontaneously or chemically initiated cells in proximal tubules to form preneoplastic and neoplastic lesions (Swenberg, 1993). The increase in cell proliferation is reversible, as is the binding of halocarbon metabolites to α2μ-globulin. Thus, this mode of action is not genotoxic and is considered by International Agency for Research on Cancer (IARC) and the EPA to be irrelevant to humans. Melnick and Kohn (1999), however, argue that some data are inconsistent with this conclusion, and that alternative mechanisms may exist. Doi et al. (2007) find some contribution of α2μ-globulin nephropathy to renal tumors produced by several structurally diverse chemicals in rats, but conclude that other critical processes are probably involved.
Several comprehensive reviews of PERC’s toxicity and carcinogenicity have been published. Lash and Parker (2001) compiled an excellent review of the metabolism, toxicity, and modes of toxicity of the chemical in laboratory animals and humans. The NAS (2009) recently published an evaluation of risks of use of TCE- and PERC-contaminated drinking water. It contained an extensive review of the metabolism, toxicity, and carcinogenicity of the two VOCs. The EPA (2008) released a draft document in support of its IRIS risk assessment, entitled Toxicological Review of Tetrachloroethylene (Perchloroethylene). Quite recently, the NAS (2010) published a review of the EPA document. Questions were raised that called into question the soundness and reliability of EPA’s proposed toxicity and cancer risk estimates. Stated weaknesses included a lack of critical analysis of data, preference given to studies (eg, genotoxicity, epidemiological) showing positive results, and lack of an integrated consideration of the weight of evidence. It is anticipated this review will result in EPA’s publication of a revised PERC risk assessment document.
TRI (methyl chloroform) is a widely used organic solvent. Its popularity as a metal degreaser, general purpose solvent, spot cleaner, and component of aerosols and a variety of household products increased substantially with the decline in manufacture of other halocarbons found to be high-dose rodent carcinogens and potential human carcinogens (ATSDR, 2006b). Utilization of TRI diminished during the 1990s, however, due to its ozone-depleting properties (Doherty, 2000). The VOC was to be phased out under the Montreal Protocol by 2002, but is still manufactured in the United States and utilized as a precursor for hydrofluorocarbons. TRI is also still present in some cleaning products. It was found in groundwater at 21% of hazardous waste sites surveyed in the United States (Pohl et al., 2008). TRI and other VOCs in groundwater can remain trapped for years and serve as a source of low-level exposure. The highest exposures to TRI occur via inhalation in occupational settings, but many persons encounter the VOC at home by use of commercial products and tap water containing it. Ashley et al. (1994) reported TRI and other VOCs in the blood of 75% of >600 non-occupationally exposed individuals. TRI and TCE were also frequently detected in a subset of 982 adults examined in the NHANES III survey (Churchill et al., 2001), and in 951 members of the general population (Blount et al., 2006).
Toxicokinetics and Metabolism
The TK of TRI has been characterized in rodents and in humans. TRI is rapidly and extensively absorbed from the lungs (Schumann et al., 1982) and GI tract (White et al., 2013). Systemic uptake and blood levels are elevated in humans inhaling TRI and other VOCs when they exercise (Astrand et al., 1973). Dallas et al. (1989) observed that uptake of inhaled TRI by rats diminished from >80% to 50% to 60% during two hours of exposure, due to its systemic accumulation as a result of slow metabolism. Monster et al. (1979) similarly noted in humans that the capacity of the body to absorb TRI vapor was less than for TCE, due to TRI’s relatively poor metabolism. White et al. (2013) reported the bioavailability of comparable oral doses of TRI and TCE in rats to be ~85% and 45%, respectively. TRI, like other VOCs, is distributed throughout the body, with fat achieving the highest concentrations (Schumann et al., 1982). As TRI biotransformation is quite limited, the VOC is cleared primarily by exhalation by rodents (Reitz et al., 1988; Schumann et al., 1982) and humans (Nolan et al., 1984).
A number of PBTK models for TRI in rodents and in humans have been published. Lu et al. (2008) recently evaluated the suitability of each of these for use in the EPA IRIS database. The model of Reitz et al. (1988) proved the most satisfactory, in that it accurately predicted the time-course of TRI in mouse, rat, and human blood for different inhalation scenarios. No empirical data were available for assessing the accuracy of simulations of brain TRI levels. Warren et al. (1998), however, demonstrated a high degree of correlation between brain and blood levels in TRI-exposed mice and rats, as well as reasonable correlation between brain levels and CNS effects of the VOC. TK data from human and rodent studies were used by Lu et al. (2008) to simulate internal dose metrics for appropriate human exposure scenarios for derivation of IRIS reference values.
The primary pharmacological manifestation of acute or chronic inhalation of TRI is CNS depression, ranging in severity from slight headache or dizziness to anesthesia and death (ATSDR, 2006b). The current TLV of 350 ppm was established to prevent decrements in workers’ mental and physical functions. Volunteers who inhaled 450 ppm TRI for four hours showed no effects on psychophysiological functions (Salvini et al., 1971). Two of 11 other subjects inhaling 500 ppm TRI 6.5 to 7.0 hours daily for five days exhibited difficulty balancing on one foot (Stewart et al., 1969). Muttray et al. (2000) found no psychophysiological effects in volunteers breathing 200 ppm TRI for four hours, but did report EEG changes and slight tiredness in their subjects. Warren et al. (2000) observed rapid, parallel increases in blood and brain TRI concentrations in mice inhaling the VOC. Mice breathing high vapor levels initially exhibited increased locomotor activity, followed by decreased activity. Mattsson et al. (1993) saw large evoked potential and EEG changes in F344 rats during inhalation of 2000 ppm TRI, but no evidence of neurotoxicity (eg, residual neurologic functional or morphological changes) after 13 weeks of exposure of ≤2000 ppm six hours daily, five days per week. Most studies of long-term occupational exposures have not revealed residual neurologic effects, although one assessment of workers subject to near-anesthetic vapor levels did reveal deficits in memory and balance (ATSDR, 2006b). It is worthy of note that very high inhaled concentrations of TRI, particularly when accompanied by hypoxia and stress, can sensitize the myocardium to catecholamines, producing cardiac arrhythmias (Reinhardt et al., 1973).
TRI has a very limited cytotoxic potential, ostensibly due to its limited biotransformation to relatively nontoxic metabolites. A near-lethal acute i.p. dose (3350 mg TRI/kg) was required to significantly enhance serum ALT activity in mice (Klaassen and Plaa, 1966). Male rats given a single oral dose of ~2500 mg/kg exhibited a transient increase in serum aspartate aminotransferase activity, but no ALT increase (Tyson et al., 1983). These enzymes are released from damaged or necrotic hepatocytes into the bloodstream. Male rats gavaged five times weekly for as long as 12 days with up to 5 g TRI/kg died from effects of repeated, protracted CNS depression, but exhibited only slight hepatotoxicity (Bruckner et al., 2001). Quast et al. (1988) saw only minimal histological changes in the liver of F344 rats inhaling 1500 ppm TRI six hours daily, five times per week for up to two years. Rats gavaged daily for 21 days with a high dose of TRI showed increased urinary N-acetyl-β-d-glucosaminidase (NAG) activity, but no microscopic evidence of injury indicative of renal toxicity (NTP, 1996). Liver and/or kidney injury are usually absent or quite modest in occupationally exposed populations, even in fatal cases (ATSDR, 2006b). A report by Hodgson et al. (1989) is an exception. They described four TRI-exposed workers who exhibited elevated serum ALT activity and fatty vacuolation of hepatocytes. In addition, Brogen et al. (1986) reported that 10% of a group of metal workers exposed to TRI, TCE, and Freon 113 had elevated NAG activity in their urine.
The NCI (1977b) conducted a 78-week study in which B6C3F1 mice and F344 rats of both sexes received high doses of TRI daily by gavage. There was no increase in cancers attributable to TRI. In a screening study, Maltoni et al. (1986) observed an apparent increase in leukemias in male and female S-D rats gavaged with 500 mg TRI/kg per day for 104 weeks. Statistical analyses were not presented, and the authors stated that definite conclusions could not be drawn from their work due to limitations in the design and number of animals. A two-year inhalation study in B6C3F1 mice and F344 rats of both sexes revealed no evidence of tumorigenicity due to TRI (Quast et al., 1988). Few epidemiological studies of TRI-exposed populations have been conducted. Infante-Rivard et al. (2005) did report a high risk (OR = 7.55, 95% CI = 0.92–61.97) of childhood leukemia in offspring of women exposed to TRI from two years before pregnancy up to birth. Recently, Gold et al. (2011) reported an OR of 1.8 (1.1–2.9) for multiple myeloma in persons exposed to TRI. It should be recognized that workers are commonly exposed to multiple solvents in the workplace. TRI is currently assigned the classification of D (not classifiable as to carcinogenicity in humans) by the EPA (1998b).
MC (dichloromethane) has enjoyed widespread use as a solvent in industrial processes, manufacture of drugs, degreasing agents, aerosol propellants, agriculture, and food preparation. It was commonly used to decaffeinate coffee and tea. Thus, large numbers of people have been exposed occupationally and in the home. The primary route of exposure to this very volatile solvent is inhalation. The preponderance of MC escaping into the environment does so by volatilization (ATSDR, 2000a). The VOC is also frequently found in wastewater discharges and in air and water at hazardous waste sites (Pohl et al., 2008).
The TK of MC has been well characterized in humans and rodents. MC is rapidly absorbed and distributed throughout the body (Angelo et al., 1986). Inhaled MC reached a near steady state in the blood of human subjects with one to two hours of continuous exposure (DiVincenzo and Kaplan, 1981). Less than 5% of the absorbed dose was exhaled unchanged. Approximately 25% to 34% was exhaled as carbon monoxide (CO), the major end metabolite of MC. Exposure of the volunteers to 50, 100, 150, and 200 ppm for 7.5 hours produced peak blood carboxyhemoglobin saturations of 1.9%, 3.4%, 5.3%, and 6.8%, respectively. MC was very rapidly eliminated from the body and did not accumulate over a five-day exposure regimen. As shown in Fig. 24-7, metabolism of MC in humans and rodents is believed to occur via three pathways (Andersen et al., 1987). One entails CYP2E1-catalyzed oxidation to CO via formyl chloride, a reactive intermediate. The second, a GSH-mediated pathway, involves the theta-class GST, GST-T1. Oxidation is a high-affinity, low-capacity pathway that predominates at the relatively low MC concentrations present in occupational and environmental settings. The GST conjugation is a low-affinity, high-capacity pathway operative at the high exposure levels used in cancer bioassays (Green, 1997). With the third and minor pathway, it is postulated that CO2 is also formed via the oxidative pathway by reaction of formyl chloride with a nucleophile such as GSH (Watanabe and Guengerich, 2006). The abilities of different species to metabolize MC in the liver by the GST pathway are as follows: mouse >> rat > human high conjugators > hamster > human nonconjugators (Reitz et al., 1989; Thier et al., 1998). Interindividual variation in the ability to biotransform MC via GST-T1 is associated with genetic polymorphisms in humans (Haber et al., 2002).
Proposed pathways for methylene chloride (CH2Cl2) metabolism. (1) Mixed-function oxidase pathway; (2) glutathione transferase pathway; and (3) nucleotide pathway. (Modified from Andersen et al., 1987, with permission from Elsevier.)
PBTK models for MC have been developed to assess the relative importance of the oxidative and GST pathways in MC’s toxicity and carcinogenicity. A PBTK model by Andersen et al. (1987) was validated by comparing simulations of blood MC time-course data with data from experiments with mice, rats, and humans. Tumor incidences in mice in chronic bioassays by NTP (Mennear et al., 1988) and Serota et al. (1986a,b) were consistent with model predictions of liver and lung doses of GSH metabolites, but not oxidative metabolites. After extensive review, the EPA adopted this model as a reasonable means of extrapolating NTP bioassay results in mice to humans. This was the first use of PBTK modeling by EPA in a cancer risk assessment. Other investigators such as Reitz et al. (1988, 1989) published deterministic models for MC, which also provided internal dosimetry point estimates of GST pathway metabolites. Forecasts of relatively low tissue doses of GST metabolites in humans resulted in part from the need to saturate the oxidative pathway before appreciable GSH metabolites could be formed.
Recently, probabilistic PBTK models for MC have been developed, which allow for inclusion of intraspecies and interspecies variability in model predictions, as well as quantitative assessment of model uncertainty. Values for model input parameters and TK data from rodent studies from a variety of sources were subjected to MCMC analysis, a Bayesian optimization technique. With this approach, the prior input information can be combined to obtain posterior distributions of key model parameters. El-Masri et al. (1999) and Jonsson and Johanson (2001) used Bayesian analysis solely to evaluate the influence of GST-T1 polymorphism on human cancer risk. Marino et al. (2006) used MCMC analysis to develop a probabilistic PBTK model for MC in mice. The resulting dose metrics (mg MC metabolized by GST/L tissue per day) were three- to four-fold higher than contemporary EPA estimates. David et al. (2006) applied this modeling approach to humans. Inclusion of GSH nonconjugators resulted in a unit cancer risk estimation 500-fold lower than the EPA unit risk at that time. The EPA (2010a) recently concluded in an extensive assessment that this was the best available PBTK model, despite some uncertainties.
Modes of Toxicity/Carcinogenicity
MC has a quite limited cytotoxicity potential. Male and female B6C3F1 mice consuming up to ~2000 mg MC/kg per day in their drinking water for 90 days showed no adverse effects (Kirschmann et al., 1986). Similarly dosed male and female F344 rats exhibited mild to moderate hepatocellular lipid vacuolation and elevated serum enzyme activities at daily dosage levels as low as 166 to 209 mg/kg per day (Kirschmann et al., 1986). Hepatic centrilobular vacuolation and focal necrosis occurred in the liver of rats inhaling 500 to 4000 ppm MC six hours per day, five days per week for two years (Burek et al., 1984; Mennear et al., 1988). Manifestations of kidney damage have been rare in laboratory animals, but have occasionally been reported in persons subjected to high vapor levels (ATSDR, 2000a; EPA, 2010a). There is little information on the identity of MC metabolites that adversely affect the liver or kidney. As described previously, the CO formed by oxidation of MC binds to hemoglobin to produce dose-dependent increases in carboxyhemoglobin. Offspring of pregnant rats inhaling low concentrations of CO have been reported to exhibit permanent learning and memory impairment (De Salvia et al., 1995). It is generally accepted that tissue hypoxia can contribute to CNS depressant effects of MC. There are few reports of residual neurologic dysfunction in MC-exposed workers (Lash et al., 1991; ATSDR, 2000a).
There has been a great deal of research to define mechanisms of MC carcinogenicity, in order to more clearly understand the relevance of the murine tumors to humans (Green, 1997). Liver and lung tumors in mice do not seem to be associated with overt cytotoxicity or increased replicative DNA synthesis (Maronpot et al., 1995). Induction of the tumors in mice is generally believed to be due to a reactive intermediate generated via the GST pathway (Andersen et al., 1987). GST-T1 in liver and lung catalyzes conversion of MC to S-(chloromethyl)glutathione (GSCH2Cl), which apparently breaks down rapidly to GSH and formaldehyde. Both GSCH2Cl and formaldehyde are reactive with DNA. MC is usually mutagenic in bacterial assays containing GSH/GST activity. MC produces DNA single-strand breaks (SSBs) in vitro in mouse hepatocytes and lung Clara cells, in which GST is localized in the nucleus (Mainwaring et al., 1996). DNA SSBs were induced in mouse hepatocytes by a 60-fold lower MC concentration than in rat hepatocytes. Rat liver does not show preferential nuclear localization of GST-T1. Some human hepatocytes apparently exhibit nuclear localization, others cytoplasmic (Sherratt et al., 2002). Negative results have been seen in a variety of genotoxicity assays with rat or hamster cell lines with little or no GST activity (EPA, 2010a). Positive results for sister chromatid exchanges, chromosomal aberrations, and the micronucleus test have been obtained in experiments with human cell lines and isolated cells. Negative results were seen in unscheduled DNA synthesis, DNA SSBs, and DNA–protein cross-links (DPX). There is limited evidence of formation of GSCH2Cl DNA adducts in some hybrid in vitro systems, but the adducts’ instability presents considerable technical challenges to their study. They have yet to be isolated in vivo. Formaldehyde produces both DPX and SSBs, indicative of its prominent role in MC’s carcinogenicity (Graves and Green, 1996).
Cancer Bioassays in Rodents
High, chronic exposures to MC have been found to produce species- and gender-specific tumors in some organs of mice and rats. Serota et al. (1986a,b) administered a series of doses of MC to F344 rats and B6C3F1 mice in their drinking water for two years. The male mice showed a trend for an increase in hepatocellular adenomas and carcinomas, but the modest response was not dose-dependent. There was a statistically significant increase in neoplastic nodules or hepatocellular carcinoma in some groups of female F344 rats. Burek et al. (1984) saw a significant increase in salivary gland sarcomas in male S-D rats that inhaled 3500 ppm six hours per day, five days per week for two years. The number of benign mammary tumors per tumor-bearing female S-D rat increased with increasing concentration of exposure, although the number of tumor-bearing rats was not significantly elevated over controls. Similarly exposed hamsters were unaffected. In a two-year follow-up study (Nitschke et al., 1988), female S-D rats inhaling 500 ppm MC also only exhibited an increased number of benign mammary tumors per tumor-bearing rat. No increase in malignant tumors was manifest in male rats. NTP conducted an inhalation study, in which F344/N rats and B6C3F1 mice inhaled up to 4000 ppm MC six hours daily, five times weekly for two years (Mennear et al., 1988). There were weak trends for neoplastic nodules and hepatocellular carcinoma, as well as benign mammary tumors in female rats. Male and female mice inhaling 2000 or 4000 ppm MC exhibited statistically significant, dose-dependent increases in hepatocellular adenoma and carcinoma, as well as bronchoalveolar adenoma and carcinoma. There were similar findings in a follow-up study of female B6C3F1 mice inhaling 2000 ppm MC for two years (Maronpot et al., 1995). The incidences of these hepatic and lung tumors in control B6C3F1 mice are quite high (Haseman et al., 1998).
Cancer Epidemiology Studies
Despite a substantial number of epidemiology studies of MC-exposed workers, evidence of associations between MC and specific tumors is not strong. There have been four cohort mortality studies of employees at facilities where MC was used as a solvent for cellulose acetate. There were no increased risks of cancer mortality for all tissues or for lung or breast. In just one assessment was there an elevated risk (SMR = 2.98 [95% CI = 0.81–7.63]) of death from liver and biliary tract cancer (Lanes et al., 1993). This was apparently the sole report of increased hepatic cancer mortality in an occupational population. No investigations provide evidence of an association between MC and lung or kidney cancer (ATSDR, 2000a; EPA, 2010a). Cantor et al. (1995) conducted a case–control study of 33,509 occupationally exposed women, but found little association between MC exposure probability and breast cancer mortality. Blair et al. (1998), however, reported a RR of 3.0 (1.0–8.8) for breast cancer in 3605 women employed at Hill Air Force Base. Blair et al. (1998) also estimated RRs of 3.0 (0.9–10.0) and 3.4 (0.9–13.2) for NHL and multiple myeloma, respectively, in MC-exposed aircraft maintenance employees of both sexes. Occasionally, excess risks of other cancers have been found in highly exposed groups. Gibbs et al. (1996), for example, computed a statistically significant SMR of 2.08 for prostate cancer death of cellulose acetate workers with a latency period of at least 20 years since their first exposure to 350 to 700 ppm MC. In a study of association between MC and astrocytic brain cancer, Heineman et al. (1994) calculated an OR of 2.4 for males with a high probability of MC exposure and for intense exposure versus unexposed controls. Lastly, Infante-Rivard et al. (2005) reported an increased risk of childhood leukemia (OR = 3.22 [0.88–11.7]) in offspring of mothers with probable or definite occupational MC exposure. Nevertheless, most investigations have revealed weak or no apparent associations between relatively high MC inhalation exposures in industry and cancers (Dell et al., 1999; Starr et al., 2006).
A considerable body of scientific information supports the following conclusion: should MC be a carcinogen in humans, it is much less potent than in rodents, notably mice. DPX were detected in hepatocytes that had been isolated from B6C3F1 mice and incubated for two hours with MC. They were not found in hepatocytes of F344 rats, Syrian golden hamsters, or three human subjects (Casanova et al., 1996, 1997). RNA–formaldehyde cross-links, however, were found in hepatocytes of all species. These links were four-, seven-, and 14-fold higher in cells from mice than in cells from rats, humans, and hamsters, respectively. Metabolism of MC via the GSH pathway is an order of magnitude greater in mouse than in rat liver. Metabolic rates in hamster and human liver are even lower (Reitz et al., 1989; Thier et al., 1998; Sherratt et al., 2002). High GST-T1 activity was measured in the nuclei of mouse centrilobular hepatocytes. Mice may be unique in that the extensive metabolic activation of MC to an unstable intermediate occurs in the proximity of the DNA. It would be useful in future PBTK models to include subcompartments for cytosolic and nuclear GST activities (Starr et al., 2006). GST-T1 was also detected in relatively high levels in mouse lung Clara cells and ciliated cells at alveolar/bronchiolar junctions (Mainwaring et al., 1996). Clara cells are present in much lower numbers in rats, and are rare in human lungs. GSCH2Cl apparently causes SSB in vivo and in vitro in DNA of mouse liver and lung (Graves et al., 1995). No DNA breaks were detected in hamster or human hepatocytes in vitro.
The EPA (2010a,b) has recently concluded that MC is likely to be carcinogenic in humans and appears to act via a mutagenic mode of action. The designation of “likely to be carcinogenic in humans” was based largely on the NTP (Mennear et al., 1988) findings of cancer at two sites (liver and lung) in male and female B6C3F1 mice. More limited findings of certain tumors in MC-exposed rats were considered to be supporting evidence. Epidemiological studies were said to provide some evidence of an association between occupational MC exposure and brain and liver cancer. As described in the previous paragraph, bioactivation of MC is qualitatively, but not quantitatively similar in mice and humans. Due to the similarity, the EPA (2010a) reasons that the apparent mode of action (mutagenicity of GST pathway metabolites) is biologically plausible in humans. The currently recommended inhalation risk value is ~47-fold lower than the previous IRIS value. EPA classifies MC as likely to be carcinogenic in humans.
CCl4 previously enjoyed widespread use as a solvent, cleaning agent, fire extinguisher, synthetic intermediate, grain fumigant, and human anthelmintic. Its use has steadily declined since the 1970s, due to its hepatorenal toxicity, carcinogenicity, and contribution to atmospheric ozone depletion (ATSDR, 2005). Nevertheless, CCl4 appears to be ubiquitous in ambient air in the United States, and it is still found in some water wells and waste sites. CCl4 is a classic hepatotoxin, but kidney injury is often more severe in humans.
The timecourse of CCl4-induced acute liver injury has been well characterized (ATSDR, 2005). Early signs of hepatocellular injury in rats include dissociation of polysomes and ribosomes from rough endoplasmic reticulum, disarray of smooth endoplasmic reticulum, inhibition of protein synthesis, and triglyceride accumulation. Hypomethylation of RNA is thought to contribute to inhibition of lipoprotein synthesis, thereby playing a role in steatosis (Clawson et al., 1987). Ingested CCl4 reaches the liver, undergoes metabolic activation, produces lipid peroxidation, covalently binds, and inhibits microsomal ATPase activity within minutes in rats. Single cell necrosis, evident five to six hours postdosing, progresses to maximal centrilobular necrosis within 24 to 48 hours. Most microsomal enzyme activities are significantly depressed (Recknagel et al., 1989). A variety of cytoplasmic enzymes are released from dead and dying hepatocytes into the bloodstream. The activity of these enzymes in serum generally parallels the extent of necrosis in the liver. Cellular regeneration, manifest by increased DNA synthesis and cell cycle progression, is maximal 36 to 48 hours postdosing (Rao et al., 1997).
The metabolism of CCl4 is required for its conversion to a variety of cytotoxic agents. It is widely recognized that CCl4 is bioactivated by cytochromes P450 via reductive dehalogenation to the trichloromethyl radical (CCl3•), which can react in turn with oxygen to form trichloromethyl peroxy free radicals (CCl3OO•). Both unstable radicals bind covalently to a variety of cellular components including enzymatic and structural proteins and polyunsaturated fatty acids in membranes. This results in lipoperoxidation, loss of intracellular and cellular membrane integrity, and leakage of enzymes (Plaa, 2000; Weber et al., 2003). By-products of lipid peroxidation include reactive aldehydes, which can form adducts with proteins and DNA, contributing to cytotoxicity and carcinogenicity, respectively (Manibusan et al., 2007). Liu et al. (1995) have proposed that CCl4 oxidative stress in the liver enhances nuclear factor kappa B activity, which in turn promotes expression of proinflammatory cytotoxic cytokines. Shi et al. (1998) proposed apoptosis as an additional/alternate mechanism of CCl4-induced cell death.
Perturbation of intracellular calcium (Ca2+) homeostasis appears to be an integral part of CCl4 cytotoxicity (Stoyanovsky and Cederbaum, 1996). Increased cytosolic Ca2+ levels may result from influx of extracellular Ca2+ due to plasma membrane damage and from decreased intracellular Ca2+ sequestration. Elevation of intracellular Ca2+ in hepatocytes can cause activation of phospholipase A2 and exacerbation of membrane damage (Glende and Recknagel, 1992). Elevated Ca2+ may also be involved in alterations in calmodulin and phosphorylase activity, as well as changes in nuclear protein kinase C activity (Omura et al., 1999). High intracellular Ca2+ levels activate a number of catabolic enzymes including proteases, endonucleases, and phospholipases, which kill cells via apoptosis or necrosis (Weber et al., 2003). The hydrolytic enzyme calpain mediates progression of acute CCl4-induced liver injury by leaking from dying hepatocytes and attacking neighboring cells (Limaye et al., 2003). Ca2+ may stimulate the release of cytokines and eicosanoids from Kupffer cells. Edwards et al. (1993) demonstrated that destruction of Kupffer cells prior to CCl4 dosing of rats resulted in significant reductions in neutrophil infiltration and hepatocellular injury. Macrophages are known to release a number of inflammatory mediators, such as tumor necrosis factor alpha (TNF-α), that are cytotoxic (Morio et al., 2001).
Development of cellular resistance and tissue repair are important in limiting CCl4 hepatotoxicity, and in recovery (Mehendale, 2005). Alterations in transmembrane carrier proteins have been discovered in hepatocytes of CCl4-treated mice. CCl4 results in reduced expression of genes associated with extraction of bile acids and organic ions from sinusoidal blood, as well as upregulation of certain detoxification genes (Aleksunes et al., 2005). It also produces differential upregulation of multidrug resistance proteins that are involved in export of oxidative stress products and metabolites. Hepatocellular regeneration has been shown to begin within six hours of a small dose of CCl4, just as centrilobular necrosis is becoming evident (Lockard et al., 1983). This early phase regeneration (arrested G2 hepatocytes activated to proceed through mitosis) is followed at ~24 hours by the secondary phase of regeneration (hepatocytes mobilized from Go/G1 to proceed through mitosis) (Bell et al., 1988). CCl4 hepatotoxicity is obviously a complex, multifactorial process that is likely to continue to receive considerable attention.
CCl4 has frequently been used as a model hepatotoxic compound with which to examine the influence of various factors that alter P450s. CYP2E1 is primarily responsible for catalyzing the bioactivation of low doses of CCl4 in humans. CYP3A contributes to the metabolism of higher doses (Zangar et al., 2000). The preeminent role of CYP2E1 in animals is clearly demonstrated by the protection afforded to CCl4-treated rodents by CYP2E1 antibody (Castillo et al., 1992), the CYP2E1 inhibitor 3-amino-1,2,4-triazole (Padron et al., 1996), and the absence of CYP2E1 expression (Wong et al., 1998). As discussed previously in the subsection “P450 Inducers,” a variety of conditions that induce CYP2E1 potentiate CCl4 hepatotoxicity in test animals and humans. Sufficient doses of CYP2E1 inhibitors, including natural constituents of foods (described in the subsection “P450 Inhibitors”), can inhibit CCl4 toxicity (Lieber, 1997). Taieb et al. (2005) discovered protein 8, a transcription factor that regulates the expression of genes that protect cells from stress, rapidly triggering CYP2E1 downregulation in CCl4-dosed mice, thereby minimizing CCl4 bioactivation.
CCl4 has been found to be a hepatocarcinogen in rodents, but there is relatively little experimental evidence on whether it is genotoxic or carcinogenic in humans (ATSDR, 2005; EPA, 2010b). There have been extensive studies of its potential genotoxic and mutagenic effects, but the results are largely negative in bacterial and in mammalian systems. Araki et al. (2004) did report that CCl4 was mutagenic in a strain of E. coli that is particularly sensitive to oxidative damage. As early as the 1940s, the National Cancer Institute conducted a series of chronic bioassays in which very high oral bolus doses of CCl4 were given to mice and rats. Large increases over controls in incidences of liver tumors were found in male and female mice (ATSDR, 2005). Relatively little was seen in rats, but hamsters were susceptible to CCl4-induced liver cancer. Recently, Nagano et al. (2007) published the results of a study in which F344 rats and BDF1 mice of both sexes were exposed six hours daily, five times weekly for 104 weeks to 0, 5, 25, or 125 ppm CCl4 vapor. There was a significant increase in hepatocellular adenomas and carcinomas in the male and female rats at the highest exposure level. Significant increases in these tumors and in adrenal pheochromocytomas were manifest in the 25- and 125-ppm male and female mice. There was a statistically significant, but more modest elevation in hepatocellular adenomas, but not carcinomas, in the 5-ppm female mice. Degenerative and necrotic hepatic changes were seen in livers of all groups of animals with liver tumors except the 5-ppm female mice. There has been limited evidence of associations between occupational CCl4 exposure and certain cancers described in some epidemiology studies, but the data were not conclusive. Exposures to CCl4 were poorly characterized and confounded by the concurrent exposures of most subjects to other chemicals in workplaces. There were no reported associations with liver cancer.
The weight of scientific evidence indicates that CCl4 is more likely an indirect than a direct acting mutagen/carcinogen (EPA, 2010b). Manibusan et al. (2007) concluded that sustained cell death, regeneration, and proliferation enhance the likelihood of unrepaired spontaneous lipid peroxidation and endonuclease-induced mutations that may lead to hepatocarcinogenesis. Jiang et al. (2004) observed changes in expression of genes involved in cell death, proliferation, DNA damage, and fibrogenesis in livers of mice given a high CCl4 dose daily for four weeks. Four weeks after cessation of this treatment, most gene expression profiles returned to control levels, except fibrogenesis. Bioactivated CCl4 can apparently exert modest genotoxic effects, such as DNA breakage and related sequelae only under highly cytotoxic conditions. ACGIH (2012) assigned CCl4 the designation of A2 (suspected human carcinogen), in light of its threshold mode of action and its very weak or absent genotoxicity. Germany placed CCl4 into its category 4, indicating that genotoxicity plays no, or at most a minor, role in its mode of action (MAK, 2011). The EPA (2010b), however, recently categorized the chemical as “likely to be carcinogenic to humans.” Despite acknowledging the correlation between hepatocellular cytotoxicity, regenerative hyperplasia, and induction of liver tumors in rodents, concern about the reactivity of direct and indirect products of CCl4 metabolism and about limited knowledge of key events led the agency to conclude that “the mode of action was unknown.” This led to use of a linear model to estimate human cancer risks from oral and inhalation exposures. PBPK models were used to estimate mouse internal doses and human equivalent doses as part of this risk assessment.
The primary use of CHCl3 (trichloromethane) is in the production of the refrigerant chlorodifluoromethane (Freon 22), but this use is diminishing as chlorine-containing fluorocarbons are phased out under the Montreal Protocol. CHCl3 was among the first inhalation anesthetics, but it was replaced by safer compounds after about 1940. It is a by-product of drinking water chlorination and has been measured in municipal drinking water supplies in concentrations as high as several hundred ppb, although levels are usually <25 ppb (ATSDR, 1997a). CHCl3 has also been found in ppb concentrations in swimming pool water and surrounding air (Aggazzotti et al., 1995). Like many other halocarbons, CHCl3 can invoke CNS symptoms at subanesthetic concentrations similar to those of alcohol intoxication and can sensitize the myocardium to catecholamines, possibly resulting in cardiac arrhythmias.
The reproductive and developmental toxicities of CHCl3 are rather unremarkable. Schwetz et al. (1974) found that inhalation of 100 to 300 ppm CHCl3 by pregnant rats caused a high incidence of fetal resorption, retardation of fetal development, and a low incidence of fetal anomalies. Murray et al. (1979) reported that gestational exposure to 100 ppm CHCl3 resulted in the decreased ability of mice to maintain pregnancy, as well as cleft palate, decreased fetal weight and length, and decreased ossification in pups. Very high CHCl3 concentrations retarded development and induced diffuse cell death in cultured rat embryos (Brown-Woodman et al., 1998). These studies support CHCl3 as a weak teratogen, but negative studies employing maternally toxic doses argue against this characterization (Thompson et al., 1974; NTP, 1997a). The EPA’s 2001 IRIS profile for CHCl3 notes that in reproductive/developmental studies, maternal toxicity and fetal effects occurred at doses higher than those that produced liver toxicity.
Hepatotoxicity serves as the basis for EPA’s benchmark dose–based RfD (EPA, 2001b). Under certain conditions CHCl3 is hepatotoxic and nephrotoxic. These toxicities are potentiated by aliphatic alcohols, ketones, DCA, and TCA (Davis, 1992). Albeit at low doses, numerous disinfection by-products such as the rodent carcinogens TCA and DCA are routinely consumed with CHCl3 in finished drinking water. For this reason, mixture studies are particularly relevant. Consider, for example, the studies of Pereira et al. (2001) in which N-methyl-N-nitrosourea-initiated B6C3F1 mice were exposed to DCA with or without CHCl3. CHCl3 prevented hypomethylation, which would have resulted in increased mRNA expression of the proto-oncogene c-myc and promotion of liver tumors by DCA. Conversely, CHCl3 increased DCA-induced DNA hypomethylation and enhanced the DCA promotion of kidney tumors. Thus, concurrent exposure to two rodent carcinogens, CHCl3 and DCA, resulted in less than additive activity in one organ and synergism in another (Pereira et al., 2001; Tao et al., 2005). This exemplifies the difficulty in assessing the risks posed by solvent mixtures.
Tolerance or adaptation to CHCl3’s hepatorenal toxicity and carcinogenicity has been observed in some mouse strains after repeated exposure. This phenomenon was first investigated by Pereira and Grothaus (1997), who reported pre-exposure of mice to low doses of CHCl3 in drinking water–induced resistance to hepatotoxicity and cell proliferation following a higher gavage dose. This was presumed at the time to result from suicidal inhibition of the CYPs responsible for CHCl3’s activation. More recently, mice exposed daily for 7, 14, and 30 days were found to have a robust regenerative response in target tissues, which prevented the progression of injury (Anand et al., 2005). Blood and tissue levels of CHCl3 after repeated exposure were substantially lower than those following a single exposure, owing to increased elimination of CHCl3 via exhalation (Anand et al., 2006b). These same researchers also reported that priming mice with CHCl3 prior to a lethal dose stimulated compensatory hepatogenic and nephrogenic repair, limiting the progression of injury and resulting in 100% survival. Relative to unprimed mice, there was no difference in hepatic or renal CYP2E1 activity, although GSH and GSH reductase activity were upregulated in the kidney (but not in the liver), with a consequent decrease in renal covalent binding. The area under the blood level versus time curve (AUC) was 40% lower in primed versus unprimed mice, but increased elimination via exhalation was not responsible for the reduction in internal exposure in this particular case (Philip et al., 2006). Taken together, these studies suggest that TK and TD factors contribute to the tolerance observed to CHCl3 toxicity. Mehendele and colleagues have conducted a series of other studies with CHCl3, alone or in combination with other hepatotoxicants, to further discern the role of tissue repair in toxicant-induced injury. These studies have provided valuable insight into the importance of both the timing of the tissue repair response and its magnitude as pivotal determinants of the outcome of toxicant-induced injury, emphasizing the need to consider repair processes in predictive toxicology (Anand et al., 2003, 2005; Mehendale, 2005).
The status of CHCl3 as a rodent carcinogen is indisputable. It causes liver and kidney tumors that are species-, strain-, sex-, and route of exposure–dependent. CHCl3-induced liver tumors in mice and their dependence on ongoing liver necrosis were reported near the end of World War II (Eschenbrenner and Miller, 1945). These same authors observed that male but not female mice suffered kidney necrosis. This observation was supported by the report of Roe et al. (1979) that CHCl3 ingested in a toothpaste base resulted in renal tumors in male but not female mice. This sex difference is thought to be attributable to testosterone-mediated differences in renal CYP activity (Smith et al., 1984). The NCI (1976) cancer bioassay demonstrated renal tumors in male rats and an extremely high incidence of liver tumors in both sexes of B6C3F1 mice gavaged with CHCl3 in corn oil. In 1985, Jorgenson and colleagues reported that daily doses of CHCl3 in drinking water comparable to those in the NCI gavage assay also produced renal tumors in rats, but failed to cause liver tumors in B6C3F1 mice. This finding provided evidence that the dose rate of CHCl3 was a determinant of liver tumor formation, supporting the existence of a threshold mechanism. Hard et al. (2000) have reevaluated the kidneys from the Jorgenson et al. (1985) study and have confirmed the presence of chronic renal tubule injury, indicative of renal tumor formation via an epigenetic mechanism. In what may be a landmark study, Larson et al. (1994) compared cytotoxicity and cell proliferation in female B6C3F1 mice given CHCl3 by gavage in corn oil versus ad libitum ingestion in drinking water. As seen in Fig. 24-8, the hepatocyte labeling index, a measure of the proliferative response, differed between the two exposure regimens at comparable doses. Pereira (1994) reported essentially the same observation. This suggests that ingestion of CHCl3 in small increments, similar to drinking water patterns of humans, fails to produce a sufficient amount of cytotoxic metabolite(s) per unit time to overwhelm detoxification and other protective mechanisms.
Hepatocyte liver LI in female B6C3F1 mice given CHCl3 orally in corn oil (A) or drinking water (B) for four days or three weeks. The LI is defined as the percent of hepatocyte nuclei positive for 5-bromo-2′-deoxyuridine immunohistochemical staining (ie, percent of cells in the S phase, the period of DNA synthesis during the cell cycle). Values represent the mean ± SD for five mice. Asterisks (*) denote a significant difference from similarly treated control mice (<0.05). Note that 1800 ppm CHCl3 in water corresponds to a cumulative uptake of 329 mg/kg per day. (Reproduced with permission from Larson et al., 1994.)
Potentiation of CHCl3’s toxicity by CYP inducers and protection by GSH and P450 inhibitors suggest that a metabolite, presumably phosgene, is responsible for CHCl3’s hepatorenal toxicity. Both target organs metabolize CHCl3 to phosgene. There is evidence that CYP2E1 and CYP2B1/2 metabolically activate CHCl3. The former isoform is thought to catalyze CHCl3 metabolism at a lower substrate concentration than the latter (Nakajima et al., 1995). By using an irreversible CYP2E1 inhibitor and CYP2E1 knockout mice, Constan et al. (1999) have demonstrated that metabolism of CHCl3 by CYP2E1 is required for liver and kidney necrosis and cell proliferation. The electrophilic intermediate generated by CHCl3’s metabolism (ie, phosgene) is initially detoxified by covalently binding to cytosolic GSH. Once GSH is depleted, phosgene is free to covalently bind to hepatic and renal proteins and lipids. Such binding damages membranes and other intracellular structures, leading to necrosis and subsequent reparative cellular proliferation. Sustained proliferation with repeated exposures promotes tumor formation in rodents by irreversibly “fixing” spontaneously altered DNA and clonally expanding initiated cells. The expression of certain genes, including myc and fos, is altered during regenerative cell proliferation in response to CHCl3-induced cytotoxicity (Sprankle et al., 1996; Kegelmeyer et al., 1997). While the identity of phosgene’s intracellular targets is largely unknown, Guastadisegni et al. (1999) have reported that phosgene reacts with phosphatidylethanolamine (PE). The adduct formed appears to consist of two PE moieties cross-linked at the amino head groups by the carbonyl moiety of phosgene. CHCl3-modified PE preferentially accumulates on inner mitochondrial membranes, inducing ultrastructural modifications and inhibiting functions of the organelle. These researchers observed the induction of hepatic apoptosis and necrosis in CHCl3-treated rats and pointed out that apoptosis may be initiated by the release of regulatory factors normally sequestered in mitochondria, in particular Ca2+. Evidence that Ca2+ perturbation plays a role in CHCl3 toxicity comes from a report of Ca2+ mobilization in Madin–Darby canine kidney cells using Fura-2 as a Ca2+ probe. CHCl3, albeit in millimolar concentrations, increased the cytosolic Ca2+ levels by releasing Ca2+ from multiple sites within the cell (Jan et al., 2000).
There is no evidence of covalent binding of CHCl3 metabolites to nucleic acids. There is binding to nuclear histone, which plays a key role in controlling DNA expression and might be a mechanism of CHCl3’s carcinogenicity (Diaz and Castro, 1980; Fabrizi et al., 2003). It has been hypothesized that the induction of oxidative stress and depletion of GSH by CHCl3 may lead to indirect genotoxicity that could contribute to carcinogenicity. This hypothesis is supported by the small dose-dependent increase in M(1)dG adducts (malondialdehyde reacts with DNA to form adducts to deoxyguanosine), DNA strand breakage, and lipid peroxidation in CHCl3-treated rat hepatocytes in the absence of any increase in DNA oxidation (Beddowes et al., 2003). Such a mechanism would still be threshold dependent, given its reliance on the initial depletion of antioxidants.
The EPA (2001b) classifies CHCl3 as a probable human carcinogen (group B2), meaning there is sufficient evidence for carcinogenicity in animals and inadequate or no evidence in humans. Experimental evidence and the prevailing opinion that CHCl3 is nongenotoxic indicates that the relationship between CHCl3 dose and tumor formation is nonlinear. The EPA has, as is called for in its Guidelines for Carcinogen Risk Assessment (EPA, 2005), considered mode of action in the determination of CHCl3’s cancer risk and relied on a nonlinear dose–response approach and the use of margin of exposure analysis. In doing so, the Agency concluded that the RfD for noncancer effects, based on the dog study of Heywood et al. (1979), was adequately protective for cancer by the oral route on the basis of cancer and noncancer effects having a common link through cytotoxicity. The wealth of mechanistic data available continues to inform the risk assessment for CHCl3. For example, Constan et al. (2002) exercised a PBTK dosimetry model to compare hepatic responses in mice and humans with inhaled CHCl3. They concluded that no safety factor was needed to account for interspecies differences in inhalation cancer risk. Additionally, Tan et al. (2003) have published a PBTK/TD model for CHCl3 to describe the plausible mechanism linking the hepatic metabolism of CHCl3 to hepatocellular killing and regenerative proliferation, thereby creating a predictive model that most accurately reflects the current science.