Derivation of a reference dose and drinking water equivalent level for 1,2,3-trichloropropane

Food and Chemical Toxicology 48 (2010) 1488–1510

Contents lists available at ScienceDirect

Food and Chemical Toxicology journal homepage: www.elsevier.com/locate/foodchemtox

Derivation of a reference dose and drinking water equivalent level for 1,2,3-trichloropropane Robert G. Tardiff *, M. Leigh Carson The Sapphire Group, Inc., 3 Bethesda Metro Center, Suite 830, Bethesda, MD 20814, USA

a r t i c l e

i n f o

Article history: Received 16 September 2009 Accepted 15 March 2010

Keywords: Trichloropropane DWEL Reference dose Cancer Benchmark dose Non-linear dose–response

a b s t r a c t In some US potable water supplies, 1,2,3-trichloropropane (TCP) has been present at ranges of non-detect to less than 100 ppb, resulting from past uses. In subchronic oral studies, TCP produced toxicity in kidneys, liver, and other tissues. TCP administered by corn oil gavage in chronic studies produced tumors at multiple sites in rats and mice; however, interpretation of these studies was impeded by substantial premature mortality. Drinking water equivalent levels (DWELs) were estimated for a lifetime of consumption by applying biologically-based safety/risk assessment approaches, including Monte Carlo techniques, and with consideration of kinetics and modes of action, to possibly replace default assumptions. Internationally recognized Frameworks for human relevance of animal data were employed to interpret the findings. Calculated were a reference dose (=39 lg/kg d) for non-cancer and Cancer Values (CV) (=10– 14 lg/kg d) based on non-linear dose–response relationships for mutagenicity as a precursor of cancer. Lifetime Average Daily Intakes (LADI) are 3130 and 790–1120 lg/person-d for non-cancer and cancer, respectively. DWELs, estimated by applying a relative source contribution (RSC) of 50% to the LADIs, are 780 and 200–280 lg/L for non-cancer and cancer, respectively. These DWELs may inform establishment of formal/informal guidelines and standards to protect public health. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Abbreviations: AA, acrylamide; AIC, Akaike’s Information Criterion; ALT, Alanine Aminotransferase; AST, Aspartate Aminotransferase; ATSDR, Agency for Toxic Substances and Disease Registry; BMD, benchmark dose; BMD10, benchmark dose corresponding to a 10% increase in extra risk; BMDL, benchmark dose lower bound (corresponding to 95% lower confidence limit); BMDS, benchmark dose software; bw, body weight; BUN, blood urea nitrogen; CHO, Chinese hamster ovary; CI, confidence interval; CV, cancer value (RfD-equivalent); DBCP, 1,2-dibromo-3chloropropane; DCA, 1,3-dichloroacetone; DHS, Department of Health Services; DWEL, drinking water equivalent level; EMS, ethylmethane sulfonate; ENU, ethylnitrosoura; GA, glycidamide; IPCS, International Programme on Chemical Safety; LADI, Lifetime Average Daily Intake; LB, lower bound from 90th percentile confidence interval; LI, BrdU Labeling Index; LOAEL, lowest-observed adverse effect level; MCL, Maximum Contaminant Level; MCLG, Maximum Contaminant Level Goal; MMS, methylmethane sulfonate; MNU, methylnitrosourea; MTD, maximum tolerated dose; MN, micronucleus; MoA, mode of action; N7-GA-Gua, glycidamide (GA)-derived DNA adduct, N7-(2-carbomyl-2-hydroxyethyl)guanine; NOAEL, noobserved adverse effect level; NTP, National Toxicology Program; OEHHA, Office of Environmental Health Hazard Assessment; PHG, public health goal; PoD, point of departure; RfD, reference dose; RSC, relative source contribution; SDWA, US Safe Drinking Water Act; TCP, 1,2,3-trichloropropane; TDI, tolerable daily intake; UF, uncertainty factor; US EPA, United States Environmental Protection Agency; WHO, World Health Organization. * Corresponding author. Tel.: +1 301 657 8008x202; fax: +1 301 657 8558. E-mail addresses: [email protected] (R.G. Tardiff), [email protected] (M.L. Carson). 0278-6915/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.fct.2010.03.016

1,2,3-Trichloropropane (TCP), a chlorinated alkane, is manufactured for use as an intermediate in chemical processing, and has been used as a solvent, paint remover, and degreasing agent. It may also be produced as a by-product during the processing of other chlorinated compounds, such as epichlorohydrin (Bikales, 1969; WHO, 2003). Historically, TCP has been reported to have been present in certain soil fumigant pesticides (WHO, 2003). TCP is structurally similar to compounds of recognized toxicity such as 1,2-dibromo-3-chloropropane (DBCP). Primary exposure of the general population to TCP appears to be via either water or air; however, it might be incorporated in foods that come into contact with TCP-containing irrigation water or tap water used in food preparation. Presently, while few data are available on environmental levels of TCP, concentrations ranging from 0.1 to 74 lg/L (ppb) have been detected historically in drinking water in Hawaii and California (ATSDR, 1992; WHO, 2003; HSDB, 2009; CDPH, 2007). In Hawaii, however, the levels in finished drinking water have been below the State standard (0.6 ppb) for many years, since all groundwater containing TCP is treated with granular activated carbon to remove this compound (Kawata, 1992). ATSDR (1992) reported from the US Environmental Protection Agency (US EPA) STORET database that in the mid-1980s

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

approximately 39% of 941 samples of groundwater (use for human consumption not specified) were found to have a median concentration of 0.7 ppb TCP (range in the 39% of samples 6 level of detection to 2.5 ppb). At 2 out of 10 sites in an agricultural community in New York, TCP was measured in groundwater (use unspecified) at 6 and 10 ppb (Lykins and Baier, 1985). Low levels (0.2–2 ppb) of TCP have also been detected in soils in California and Hawaii (ATSDR, 1992). The purpose of the work presented herein is the development of a drinking water equivalent level (DWEL) for human daily consumption of TCP over a lifetime. Toward that end, a reference dose (RfD, as defined generically by US EPA (2002)) is used to establish safe levels of exposure to a chemical for exposure from all sources; for carcinogens, a RfD may also include probabilistic estimation of cancer risk when the compound of interest by default is assumed to have a linear dose–response relationship to a zero intersect based on the assumption that the compound’s carcinogenicity is based on a no-threshold mode of action (MoA). A DWEL, as applied under regulations issued under the US Safe Drinking Water Act (US Congress, 1996), represents that fraction of the RfD reserved for tap water consumed directly as well as indirectly from use in food preparation. 2. Approach The approach to estimate safe levels of exposure to TCP in drinking water includes: a critical analysis of its toxicological properties [no epidemiologic data on TCP were found in our search of the scientific literature]; an examination of data on metabolism, kinetics, and possible MoAs; an evaluation of the dose–response information to estimate toxic potency, and a description of the method used to estimate a DWEL and a RfD. The approach used herein relies on the broad construct of risk assessment embodied by the current flexible guidelines presently evinced by US EPA (2002, 2005) and the World Health Organization (WHO, 2006). The approach applies internationally recognized Frameworks for evaluating the relevance to humans of toxicological data in laboratory animals. 2.1. Hazard evaluation The toxicity of TCP has been investigated in rats and mice using subchronic and chronic (i.e., near lifetime) durations of exposure. Three subchronic studies have been conducted in rats and one in mice; and a chronic study has been conducted in each of these two species. While findings from these studies are described below, detailed reviews of TCP’s toxicology data have been prepared by ATSDR (1992), ACGIH (1996), DECOS (1998), WHO (1995, 2003), and US EPA (2009a). Relevant descriptive, kinetic, and MoA studies were critically examined, and the major findings were evaluated with a focus on determining the relevance to humans of descriptive and mechanistic investigations. 2.1.1. Non-cancer toxicity of TCP Of the three subchronic toxicity studies of TCP in rats, one study administered TCP via drinking water, and the remainder did so via corn oil gavage (Table 1). Male and female Sprague–Dawley rats were exposed via drinking water to TCP at 1, 10, 100, and 1000 mg/L for 13 weeks (Villeneuve et al., 1985). Drinking water intake was significantly decreased in females exposed to 100 and 1000 mg/L and in males exposed to the highest dose, likely due to reduced palatability; therefore, these animals had a reduced intake of TCP (modified doses: 17.6 at 100 mg/L and 149 mg/kg d for females and 113 mg/kg d for males at the highest concentration level). Body

1489

weight gain was significantly reduced (32% in males, 27% in females) in the highest dose for both sexes, indicating exceedance of the maximum tolerated dose (MTD). Relative liver- and kidney-to-body weight ratios were increased (6–17% liver-to-body weight, 11–31% kidney-to-body weight) significantly in females at doses 100 and 1000 mg/L. In males, liver- and kidney-to-body weight ratios were significantly increased (22–27%) in the high dose. The authors state that the increased kidney-to-body weight ratio was likely unrelated to TCP administration, because the wet weight was unaffected and the affect on the kidney:body weight ratio was assumed to be due to the decreased body weight. In the highest dose group, histopathologic changes (defined as ‘‘minor” in severity) were reported only in the liver (asinokaryosis, accentuated zonation, fatty vacuolation, and biliary hyperplasia), kidney (eosinophilic inclusions, pyknosis, nuclear displacement, fine glomerular adhesions, interstitial reactions, and proteinura), and thyroid (reduction in colloid density, follicular angular collapse, and increased epithelial height); no pancreatic acinar cell hyperplasia was reported. Increased (24–51%) clinical chemistry measures and liver enzymes (serum cholesterol in females and aminopyrine demethylase and aniline hydroxylase in males) were reported at 1000 mg/L. The study authors observed that the lowest-observed adverse effect level (LOAEL) for TCP was 149 mg/ kg d (1000 mg/L) in females and 113 mg/kg d (1000 mg/L) in males. The study authors also determined a TCP dose of 15– 20 mg/kg d (100 mg/L) to be a no-observed adverse effect level (NOAEL) for males and females. In a subchronic corn oil gavage study, male and female Sprague–Dawley rats were exposed to TCP at doses of 0, 1.5, 7.5, 15, and 59 mg/kg d (0, 0.01, 0.5, 0.1, and 0.4 mmol/kg d) for 90 days (Merrick et al., 1991). Body weights were significantly reduced (14–19%) in both sexes in the 59 mg/kg d group. In male rats, organ weights (brain, kidney, and testes) relative to body weight were significantly increased (15–25%) in the high dose group, while the liver-to-body weight ratio was significantly increased (13– 25%) in the 15 and 59 mg/kg d groups. In females, kidney and liver weights relative to body weight were significantly increased (8–60%) in the 15 and 59 mg/kg d groups. Increased (80–90%) concentrations of liver enzymes (Alanine Aminotransferase [ALT], Aspartate Aminotransferase [AST]) in females, but not males, in the high dose group corroborate a hepatotoxic response. While moderate changes were also seen with creatinine and blood urea nitrogen (BUN), these effects were determined by the investigators to be unrelated to dose and/or were pathologically not significant. Mild to moderate liver necrosis was reported, more frequently in male rats with increasing incidence from low to high dose, but the relevance of these effects was uncertain due to the presence of mild hepatic lesions in four control animals. Significantly increased (31–89%) serum cholesterol was also reported in female rats at doses 15 and 59 mg/kg d. Serum chemistry indicated no renal toxicity. Hyperplasia was also observed in the bile duct of both sexes in the high dose groups; based on toxicokinetic data, this effect may be due to the presence of excreted and reabsorbed stable reactive metabolites of TCP. Additionally, in both sexes at the high dose, a diffuse inflammation-associated necrosis of the cardiac myocardium was reported, indicating cardiotoxicity. We concluded a dose of 15 mg/kg d to be LOAEL for both sexes, and a dose of 7.5 mg/kg d to be a NOAEL. In a second gavage study in 1993, the National Toxicology Program (NTP) exposed F344 rats and B6C3F1 mice to TCP via oral gavage with corn oil (doses: 0, 8, 16, 32, 63, 125, and 250 mg/kg d) for 5 days/week for 17 weeks (interim sacrifice at 8 weeks) (NTP, 1993; Irwin et al., 1995), which served as pilot investigation for subsequent carcinogenicity studies [described below]. All female rats in the high dose group died by the second week from renal or hepatic toxicity, while males in the high dose all died

Species/sex

Duration

Drinking water exposure Sprague–Dawley 90 days rats/male and (3 months) female

1490

Table 1 Noncancer studies of 1,2,3-trichloropropane. Results

LOAEL

NOAEL

Reference

0, 1, 10, 100, or 1000 mg/L [100 mg/L = 17.6 mg/kg d in females; 1000 mg/ L = 113 mg/kg d in males; 149 mg/kg d in females]

1000 mg/L: (113 [M] and 149 [F] mg/kg d)=  Significantly decreased body weight gain [M: 32%; F: 27%] and water intake [M/F]  Significantly increased liver-to-body weight ratio [M: 22%; F: 17%]  Liver changes (anisokaryosis, accentuated zonation, occasional fatty vacuolation) [M & F]  Kidney changes (eosinopilic inclusions, pyknosis, nuclear displacement, fine glomerular adhesions, interstitial reactions) [M & F]  Thyroid changes (angular follicular collapse, reaction in colloid density, increased epithelial height); increased hepatic aminopyrene demethylase [M & F]  Significantly increased biliary hyperplasia and serum cholesterol [F: 24–51%]  Significantly increased aniline hydroxylase [M: 38%]

113 mg/kg d [M] 149 mg/kg d [F]

15–20 mg/kg d [M & F]

Villeneuve et al. (1985)

15 mg/kg d [M & F]

7.5 mg/kg d [M & F]

Merrick et al. (1991)

8 mg/kg d [M] 16 mg/kg d [F]

No NOAEL for males 8 mg/kg d [F]

NTP (1993)

100 mg/L: (17.6 mg/kg d [F]) = significantly increased liver-to-body weight ratio [F: 6%] Corn oil gavage exposure Sprague–Dawley 90 days rats/male and (3 months) female

0, 1.5, 7.5, 15, or 59 mg/kg d

59 mg/kg d=  Decreased body weight gain (M: 25% [10-days], 19% [90-days]; F: 23% [10-days], 14% [90-days]) [M & F]  Increased organ weight ratios [kidney = M: 15%, F: 18%, liver = M: 25%, F: 60%, brain = M: 20%, testis = M: 26%]  Effects on cardiac myocardium (inflammation, degeneration, necrosis) [M & F]  Bile duct hyperplasia [M & F]  Increased ALT, AST [F: 80–90%], and serum cholesterol [F: 89%]  Mild to moderate liver necrosis [M]: relevance uncertain 15 mg/kg d=  Increased organ weight ratios [liver = M: 13%, F: 14%, kidney = F: 8%]  Increased serum cholesterol [F: 31%]  Mild to moderate liver necrosis [M]: relevance uncertain 7.5 mg/kg d = mild to moderate liver necrosis [M]: relevance uncertain 1.5 mg/kg d = mild to moderate liver necrosis [M]: relevance uncertain

F344/N rats/male and female

17 weeks (4 months)

0, 8, 16, 32, 63, 125, or 250 mg/kg d

250 mg/kg d=  All rats died [M & F]  Liver necrosis [M & F]  Kidney necrosis and hyperplasia/karyomegaly [M & F]  Nasal turbinate necrosis and attenuation/imflammation [M & F] 125 mg/kg d=  Decreased body weight [M: 21%; F: 24%]  Increased absolute & relative liver weight [M: 36% & 78%; F: 61% & 105%]  Increased absolute and relative kidney weight [M: 19% & 54%, F = 11% & 43%]  Decreased erythrocyte mass [M & F]  Increased ALT [F: 248%]  Decreased AST [M: 32%]  Decreased urea nitrogen [M: 21%, F: 22%]  Liver necrosis [F]  Multifocal hemorrhage and bile duct hyperplasia [F]  Kidney regenerative hyperplasia with karyomegaly [M & F]  Nasal turbinate necrosis [M]/attenuation [M & F]/inflammation [M] 63 mg/kg d=  Decreased body weight [M: 11%]  Increased absolute & relative liver weight [M: 47% & 23%, F: 32% & 37%]  Increased absolute and relative kidney weight [M: 5% & 26%, F = 25% & 32%]

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

Doses

   

Decreased erythrocyte mass [M & F] Decreased AST [M: 39%] Decreased urea nitrogen [F: 21%] Kidney regenerative hyperplasia [M] with karyomegaly

32 mg/kg d=  Increased absolute and relative liver weight [M: 24% & 26%, F: 17% & 18%]  Increased absolute and relative kidney weight [M: 12% & 15%]  Decreased erythrocyte mass [M & F]  Decreased AST [M: 27%]  Decreased urea nitrogen [F: 21%] 16 mg/kg d=  Increased absolute and relative liver weight [M: 10% (absolute weight only), F: 12% & 18%]  Decreased erythrocyte mass [M & F] 8 mg/kg d = increased absolute liver weight [M: 11%] 2 years (24 months)

0, 3, 10, or 30 mg/kg d

30 mg/kg d=  Decreased survival (probability of survival = 0%); all survivors killed at 60 weeks [F] and 77 weeks [M]  Decreased body weight [M: 13%; F: 9%]  Forestomach hyperplasia [M & F]  Pancreatic hyperplasia [M & F]  Kidney hyperplasia [M & F]

3 mg/kg d [M & F]

None

NTP (1993), Irwin et al. (1995)

125 mg/kg d [M] 63 mg/kg d [F]

32 mg/kg d

NTP (1993)

10 mg/kg d=  Decreased survival (probability of survival = M: 30%; F: 16%)  Forestomach hyperplasia [M & F]  Pancreatic hyperplasia [M & F]  Kidney hyperplasia [M] 3 mg/kg d = forestomach hyperplasia [M & F]; pancreatic hyperplasia [M & F] B6C3F1 mice/male and female

17 weeks (4 months)

0, 8, 16, 32, 63, 125, or 250 mg/kg d

250 mg/kg d=  Decreased survival [M: 80%]  Increased absolute and relative liver weight [M: 25% & 30%; F: 22% & 24%]  Decreased absolute and relative kidney weight [F: 13% & 14%]  Liver necrosis [M & F] and karyomegaly [M]  Bronchial epithelium regeneration [M & F]  Forestomach hyperkeratosis and acanthosis [F]  Decreased blood urea nitrogen [M: 53%]  Decreased creatinine [M: 30%]  Increased ALT [F: 75%] 125 mg/kg d=  Increased absolute and relative liver weight [M: 22% & 10%, F: 12% & 25%]  Increased relative kidney weight [F: 17%]  Forestomach hyperkeratosis [M & F] and acanthosis [F]  Bronchial epithelium regeneration [M & F]  Decreased blood urea nitrogen [M: 50%]  Decreased creatinine [M: 44%]

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

F344/N rats/male and female

63 mg/kg d=  Increased absolute liver weight [M: 4%]  Decreased relative kidney weight [F: 11%]  Bronchial epithelium regeneration [F]  Forestomach hyperkeratosis and acanthosis [F]  Decreased blood urea nitrogen [M: 30%]  Decreased creatinine [M: 30%] 1491

(continued on next page)

1492

Table 1 (continued) Species/sex

Duration

Doses

Results

LOAEL

NOAEL

Reference

6 mg/kg d

None

NTP (1993), Irwin et al. (1995)

32 mg/kg d=  Increased absolute liver weight [M: 14%]  Decreased relative kidney weight [F: 13%]  Decreased blood urea nitrogen [M: 48%]  Decreased creatinine [M: 23%] 16 mg/kg d=  Decreased relative kidney weight [F: 17%]  Decreased blood urea nitrogen [M: 46%]  Decreased creatinine [M: 21%] 8 mg/kg d =  Decreased blood urea nitrogen [M: 46%];  Decreased creatinine [M: 16%] 2 years

0, 6, 20, or 60 mg/kg d

60 mg/kg d=  Decreased survival (probability of survival = 0%) [M & F]  Decreased body weight [M: 16%, F: 18%]  Forestomach hyperplasia [M & F]  Uterine polyp [F] 20 mg/kg d=  Decreased survival (probability of survival = 0%) [M & F]  Forestomach hyperplasia [M & F] 6 mg/kg d=  Decreased survival [probability of survival = M: 36%; F: 26%]  Forestomach hyperplasia [M & F]

M = male and F = female.

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

B6C3F1 mice/male and female

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

by the fifth week. Body weight gain was significantly decreased (21–24%) in the 125 mg/kg d group for both male and female rats and in the 63 mg/kg d dose group (11%) for male rats. Relative liver weights were significantly increased (24–78%) in male rats at doses P32 mg/kg d, while absolute liver weights were significantly increased (10–36%) at all doses. In female rats, relative and absolute liver weights were significantly increased at P16 mg/ kg d (12–105%), while absolute and relative kidney weights were significantly increased (5–54%) (accompanied by regenerative hyperplasia) at P32 and P63 mg/kg d in males and females, respectively. Changes in liver and kidney weights were consistent with clinical chemistry and histopathology findings. Liver enzymes (ALT) were significantly elevated (248%) in females at 125 mg/kg d, while the liver enzyme, AST, was decreased (27–38%) in males at P32 mg/kg d [no consistent dose–response]. Urea nitrogen levels were significantly reduced (21%) in males receiving 125 mg/ kg d and in female rats (21–22%) receiving P32 mg/kg d. Multifocal and/or centrilobular hepatocellular necrosis and karyomegaly of hepatocytes were significantly increased (45– 100%) at the 8-week interim evaluation in liver in males receiving 250 mg/kg d and in females (25–100%) receiving 125 and 250 mg/ kg d at the 8-week interim and 17-week evaluation, with hepatocellular necrosis more predominant in female rats. In the kidney, multifocal necrosis was reported in males and females receiving 250 mg/kg d, while regenerative hyperplasia was reported in males and females receiving 63, 125, and 250 mg/kg d, and karyomegaly was reported in males receiving 125 and 250 mg/kg d and in females receiving 125 mg/kg d. In the nasal turbinates, epithelium attenuation, necrosis of olfactory and respiratory epithelium, and multifocal necrosis were reported in both sexes at 125 and 250 mg/kg d, while inflammation was reported in males receiving 125 and 250 mg/kg d and in females receiving 250 mg/kg d. No dose–response was noted for the nasal lesions in the male rats, while a minor dose–response was noted for liver necrosis and nasal turbinate necrosis and attenuation at the 8-week interim evaluation. We concluded that 8 mg/ kg d bw in male rats and 16 mg/kg d in female rats are LOAELs; no NOAEL was identified. In mice from the same subchronic corn oil gavage study (NTP, 1993; Irwin et al., 1995), 16 out of 20 males in the high dose died by the fourth week, and 7 out of 20 females died by the second week in the high dose group. Body weights (8%) were significantly affected only in the male high dose group, and not in any of the female dose groups. Relative liver weights were significantly increased (10–30%) in males receiving P125 mg/kg d, while absolute liver weight was significantly increased (4–25%) at P32 mg/kg d. In females receiving 125 and 250 mg/kg d, relative and absolute liver weights were significantly increased (12–25%). Relative kidney weights were significantly reduced (13–17%) in females at P16 mg/kg d, while absolute kidney weight was only significantly decreased at the high dose. The effect on liver and kidney weight was consistent with histopathologic lesions and clinical chemistry measures. In males, BUN (30–53%) and creatinine (16–44%) levels were significantly decreased at all doses [no dose response]. In females, the liver enzyme, ALT, was significantly increased (75%) at the high dose [no dose response]. Lesions were significantly increased at the 8-week interim and 17-week evaluations in the liver (focal hepatocellular necrosis and hepatocellular degeneration) (58–75%) in males receiving 125 and 250 mg/kg d and females (35–66%) receiving 250 mg/kg d, lungs (epithelial regeneration1) in males at 125 and

1 This effect in the lungs was not reported in other studies, and is likely to have resulted from improper gavage administration.

1493

250 mg/kg d (75%) [at the 17-week evaluation only] and females receiving 63, 125, and 250 mg/kg d (50–83%), and forestomach (hyperplasia and hyperkeratosis) in males receiving 125 mg/kg d (58–75%) and in females receiving 63, 125, and 250 mg/kg d (50– 100%). No histopathologic lesions were reported for the kidneys. A dose–response was observed for lung epithelial regeneration in male mice, but not in the females. We estimated that 125 mg/kg d for males and 63 mg/kg d for females were LOAELs. We determined the dose of 32 mg/kg d to be a NOAEL for both sexes of mice. A lifetime gavage study in rats (F344/N) and another in mice (B6C3F1) were conducted to determine whether the TCP is capable of causing cancer following chronic exposures (NTP, 1993; Irwin et al., 1995). Assorted non-cancer responses were reported and are noted here [tumor responses are described later in this monograph]. In these chronic bioassays, TCP was administered via corn oil gavage to F344 rats (doses: 0, 3, 10, and 30 mg/kg d) and B6C3F1 mice (doses: 0, 6, 20, and 60 mg/kg d) for 5 days/week for 24 months with an interim sacrifice after 15 months of treatment. Male and female rats in the high dose group died or were killed moribund before the end of the study (probability of survival = 0%); in the mid-dose groups of rats, the probability of survival was reported to be 30% for male rats and 16% for females. In the control animals, the probability of survival was 70% and 62% for the male and female rats, respectively, suggesting problems in the animal colony unrelated to the test compound. At the 15-month interim sacrifice, pancreatic acinar hyperplasia was reported in male rats at 10 and 30 mg/kg d (70–100%) and in female rats at 30 mg/kg d (25%). In male rats, renal tubule hyperplasia was reported at 10 and 30 mg/kg d (20–75%), while it was reported only at the high dose (25%) for female rats. A dose–response was noted for renal tubule hyperplasia for male rats. In addition, the liver enzyme, ALT, was significantly reduced (31%) in male rats in the high dose. No effect on liver enzymes was reported in female rats at the 15month interim sacrifice (NTP, 1993). In rats at all doses at the end of the study, pancreatic acinar hyperplasia was significantly increased, and kidney hyperplasia (focal hyperplasia of the renal tubule epithelium) was significantly increased (43–56%) at 10 and 30 mg/kg d for males and only at the high dose for females (20%). A significant increase (8–44%) in forestomach basal cell and squamous hyperplasia was reported at all doses in females, while males had significantly increased basal cell hyperplasia (12–20%) at all doses and squamous hyperplasia (22–47%) at 3 and 10 mg/kg d. A dose–response was noted for hyperplasia of the renal tubule cells of both sexes and for pancreatic acinar cells of male rats (NTP, 1993; Irwin et al., 1995). In the same NTP (1993) chronic study, male and female mice in the high and mid-dose group died or were killed moribund by the end of the study (probability of survival = 0%); in the low dose groups of mice, the probability of survival was reported to be 36% and 26% for male and female mice, respectively. In the control animals, the probability of survival was 81% and 82% for males and females, respectively, by the end of the study, suggesting a problem in the animal colony unrelated to the test substance. At the 15-month interim sacrifice, male rats reported significantly increased creatinine kinase levels (235%) at the high dose (suggesting cardiovascular tissue injury), while only inconsistent effects were reported for the female mouse. Liver enzymes were unchanged from control values. At the end of the study, forestomach squamous cell hyperplasia was increased (38–64%) above that in controls in males and females at all doses (no dose–response). We estimated that 3 and 6 mg/kg d were LOAELs for rats and mice, respectively; a NOAEL was not identified for either species. The foregoing indicates that in laboratory rodents the target organs for repeated and prolonged oral exposures to TCP are the kidneys, liver, and pancreas (corn oil studies only) with the caveat that

1494

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

interpretation is clouded by high premature mortality at the highest dose levels. Much of the non-cancer effects were changes in body weight and in organ:body weight ratios, and some hyperplasia which may have been in response to cell damage. The magnitude of those test doses mitigate the possibility of reliably extending these findings to humans exposed to doses and at dose rates 5–6 orders of magnitude below the test doses. Nonetheless, in the absence of human data, it is assumed that the same organs would likely be affected in humans at sufficiently high doses of TCP. These data also suggest that TCP is less toxic, by a factor of 5–10, when administered in drinking water than when administered by corn oil gavage. The available data suggest that the LOAELs for the administration via drinking water was 10 times greater than corresponding LOAELs for administration in corn oil; comparing NOAELs shows that the drinking water administered TCP was approximately five times less toxic than TCP administered in corn oil. These findings are consistent with results reported by La and Swenberg (1997a) and La et al. (1996). 2.1.2. Carcinogenicity of TCP Lifetime carcinogenicity bioassays in rats and mice NTP (1993) were conducted to determine whether TCP is capable of causing cancer following repeated and prolonged exposures. Study details were described above. The tumor data generated from these studies are presented in Table 2. TCP has been shown to be a multi-site carcinogen in both rats and mice after repeated and prolonged exposure. At the 15-month interim sacrifice of the NTP (1993) study, renal tubule adenomas and preputial carcinomas in male rats were reported at the high dose (63% and 13%, respectively). Pancreatic adenomas, but not adenocarcinomas, were reported in male rats at the high dose (25%). In female rats, clitoral gland adenomas and mammary gland adenomas and adenocarcinomas were reported in the high dose (25% and 13%, respectively). Forestomach papillomas of male and female rats receiving 10 and 30 mg/kg d (30–100%) and oral mucosa squamous cell papillomas at 30 mg/ kg d (38%) were reported. Forestomach carcinomas were reported in male and female rats at the high dose (13–25%); a dose–response was noted for forestomach papillomas in both sexes. Oral mucosa squamous cell carcinomas were found only in high dose females (25%). At the end of the 2-year study (NTP, 1993; Irwin et al., 1995), among the tumors observed in rats, a statistically significant increase (23–40%) in mammary gland adenocarcinomas at 10 and 30 mg/kg d was reported in female rats, with an observed dose–response. Benign neoplasms were also significantly increased; they included pancreatic acinar adenomas in male rats (all doses with a dose–response), renal tubule adenomas in male rats (10 and 30 mg/kg d with a dose–response), preputial adenomas in male rats (high dose and a week dose–response was noted), clitoral gland adenomas in females (10 and 30 mg/kg d with a weak dose–response). Male and female rats also had a significant increase, with an observed dose–response, in oral cavity squamous cell papillomas and carcinomas. For intestinal lesions, NTP concluded that the number of rats affected in any particular dose group was ‘‘low and not significantly greater than the number of affected controls”. Forestomach squamous cell papillomas were significantly increased in male and female rats at all doses (with a dose–response), while squamous cell carcinomas were increased significantly, with a dose–response, at all doses in male rats and in 10 and 30 mg/kg d in female rats (NTP, 1993; Irwin et al., 1995). As noted previously, a low probability of survival to the end of the study was calculated for the mid- and high dose groups, suggesting that either these doses exceeded the MTDs or some unre-

lated problem was present in the animal colony. The LOAEL for both male and female rats was estimated to be 3 mg/kg d for tumors; a NOAEL was not identified. For mice in the same study (NTP, 1993), at the 15-month interim sacrifice, no increase above controls was observed for squamous cell papillomas of the oral mucosa, and for adenomas of the liver, Harderian gland, and uterine endometrium at all doses of both sexes. Hepatocellular adenomas, but not carcinomas, were reported in male mice at the high dose (50%) and females (11–100%) receiving 20 and 60 mg/kg d. Papillomas (50–100%) and carcinomas (67–100%) of the forestomach were observed in males and females receiving the mid and high dose of TCP. A dose–response was noted for the forestomach carcinomas in male mice and liver adenomas in female mice. Male mice had no significant increase in oral cavity neoplasms, whereas female mice had a significant increase in oral cavity squamous cell carcinomas at the high dose only. At the end of the 2-year mouse study (NTP, 1993; Irwin et al., 1995), hepatocellular carcinomas were statistically significantly increased in male mice only at the lowest dose (6 mg/kg d) [not statistically significant when combined with adenomas], but were considered to not be compound-related because no hepatocellular carcinomas were not elicited at higher doses. In addition, no corresponding statistically significant increase in hepatocellular carcinomas was found in females. Benign hepatocellular adenomas were also significantly increased, with an observed dose–response, in male mice at 20 and 60 mg/kg d and female mice at the high dose. Harderian gland adenomas were increased in male mice at 20 and 60 mg/kg d and in females at the high dose. Female mice had a significant increase in uterine carcinomas at all doses, with a questionable dose–response. The LOAEL for tumors in male and female mice was determined to be 6 mg/kg d; no NOAEL was found. As previously stated, a low probability of survival to the end of the study calculated for all the dose groups suggested that these doses are likely to have exceeded the MTD. The foregoing indicates that in laboratory rodents the target organs for tumors from repeated and prolonged exposures to TCP are the kidneys, preputial gland, clitoris, mammary gland, and oral cavity in rats, and liver, uterus, and oral cavity in mice. In the absence of human data, it is assumed that all tissues having a homologous counterpart in humans might be target organs for carcinogenicity in humans exposed to sufficiently high doses of TCP. However, the high dose rates tested exceeded the MTD, and were sufficiently large as to have caused substantial premature mortality and excessive loss of body weight among test subjects; the magnitude of those test doses complicate extending these findings to humans exposed to doses and at dose rates many orders of magnitude below the test doses. 2.2. Data influencing dose–response of TCP Several types of toxicological information are valuable in distinguishing between safe and dangerous levels of exposure to a chemical such as TCP: (1) the influence of the body on the disposition (toxicokinetics: absorption, distribution, metabolism, and excretion) of the chemical and (2) the behavior of the chemical on tissues, cells, and subcellular components of the body (toxicodynamics related to MoAs). 2.2.1. Toxicokinetics of TCP The disposition of TCP and its kinetics have been reported by Volp et al. (1984), Mahmood et al. (1991), and Weber and Sipes (1992), and summarized in part by WHO (1995).

Table 2 Summary of dose–response data for non-cancer toxicity and tumors in laboratory rodents exposed to TCP (NTP, 1993; Irwin et al., 1995). Tissue

Rats (doses = 0, 3, 10, 30 mg/kg d)

Mice (doses = 0, 6, 20, 60 mg/kg d)

Magnitude of dose– response relationship

Male

Female

Male

Female

Yes

Liver

–

–

Non-cancer: Eosinophilic foci (2/52, 3/ 51, 8/54, 32/56) LOAEL = NS NOAEL = NS Cancer: Hepatocellular adenoma/carcinoma (13/52, 24/51, 24/54, 31/56) LOAEL = 20 mg/kg d NOAEL = 6 mg/kg d

Non-cancer: Eosinophilic foci (0/50, 6/ 50, 9/51, 34/55) NOAEL = NS LOAEL = NS Cancer: Hepatocellular adenoma/carcinoma (7/50, 11/50, 8/51, 31/55) LOAEL = 60 mg/kg d NOAEL = 20 mg/kg d

Absent in rats Moderate in mice

Yes

Pancreas

Non-cancer: Acinar hyperplasia (28/ 50, 46/50, 46/49, 48/52) LOAEL = 3 mg/kg d NOAEL = NS Cancer: Acinar adenoma (5/50, 21/ 50, 36/49, 29/52) LOAEL = 3 mg/kg d NOAEL = NS

Non-cancer: Acinar hyperplasia (5/50, 14/49, 24/52, 9/52) LOAEL = 3 mg/kg d NOAEL = NS Cancer: Acinar adenoma (0/50, 0/47, 2/52, 0/51) LOAEL = NS NOAEL = 30 mg/kg d

–

–

Moderate in male rats Absent in others

Yes

Kidney

Non-cancer: Renal tubule hyperplasia (0/50, 1/50, 21/49, 29/52) LOAEL = 10 mg/kg d NOAEL = 3 mg/kg d Cancer: Renal tubule adenoma (0/50, 2/50, 20/49, 21/52) LOAEL = 10 mg/kg d NOAEL = 3 mg/kg d

Non-cancer: Renal tubule hyperplasia (0/50, 2/47, 3/52, 10/51) LOAEL = 30 mg/kg d NOAEL = 10 mg/kg d Cancer: Renal tubule adenocarcinoma (0/50, 0/47, 0/52, 1/51) LOAEL = NS NOAEL = 30 mg/kg d

–

–

Moderate in male rats Absent in others

Yes

Reproductive system

Non-cancer: Preputial hyperplasia (0/49, 0/47, 1/49, 1/50) LOAEL = NS NOAEL = 30 mg/kg d Cancer: Preputial adenoma/carcinoma (5/49, 6/47, 8/49, 16/50) LOAEL = 30 mg/kg d NOAEL = 10 mg/kg d

Non-cancer: Clitoral hyperplasia (0/46, 2/46, 3/50, 3/51) LOAEL = NS NOAEL = NS Cancer: Clitoral adenoma/carcinoma (5/46, 10/46, 17/50, 15/51) LOAEL = 10 mg/kg d NOAEL = 3 mg/kg d

–

Non-cancer: Endometrial hyperplasia (43/50, 38/50, 41/51, 52/54) LOAEL = NS NOAEL = NS Cancer: Uterus adenoma/adenocarcinoma (0/50, 5/50, 3/51, 9/54) LOAEL = 6 mg/kg d NOAEL = NS

Weak in both species

Yes

Mammary gland

–

Non-cancer: None identified Cancer: Adenocarcinoma (1/50, 6/49, 12/52, 21/52) LOAEL = 10 mg/kg d NOAEL = 3 mg/kg d

–

–

Moderate in female rats Absent in others

Yes

Oral mucosa

Non-cancer: None identified Cancer: Squamous papilloma/carcinoma (1/50, 4/50, 18/49, 40/52) LOAEL = 10 mg/kg d NOAEL = 3 mg/kg d

Non-cancer: None identified Cancer: Squamous cell papilloma/carcinoma (1/50, 6/49, 28/52, 32/52) LOAEL = 10 mg/kg d NOAEL = 3 mg/kg d

Non-cancer: None identified Cancer: Squamous papilloma (0/52, 0/ 51, 0/54, 2/56) LOAEL = NS NOAEL = 60 mg/kg d

Non-cancer: None identified Cancer: Squamous papilloma/carcinoma (1/50, 0/50, 2/51, 5/55) LOAEL = 60 mg/kg d NOAEL = 20 mg/kg d

Strong in rats Weak in mice

Yes

Intestine

Non-cancer: None identified Cancer: Adenocarcinoma (0/50, 0/50, 2/49, 3/52) LOAEL = NS NOAEL = 30 mg/kg d

Non-cancer: None identified Cancer: Adenocarcinoma (0/50, 0/49, 1/52, 2/52) LOAEL = NS NOAEL = 30 mg/kg d

–

–

Weak in rats Absent in mice

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

Human tissue homologue

(continued on next page) 1495

Moderate in male mice Absent in others Non-cancer: None identified Cancer: Adenoma (2/50, 6/50, 7/51, 10/55) LOAEL = 60 mg/kg d NOAEL = 20 mg/kg d Harderian gland No

NS: not stated.

Non-cancer: None identified Cancer: Adenoma (1/52, 2/51, 10/54, 11/56) LOAEL = 20 mg/kg d NOAEL = 6 mg/kg d

Weak in male rats Absent in others –

Non-cancer: None identified Cancer: Carcinoma (0/50, 1/49, 0/52, 3/ 52) LOAEL = 30 mg/kg d NOAEL = 10 mg/kg d – Zymbal’s gland No

Non-cancer: None identified Cancer: Carcinoma (0/50, 0/50, 0/49, 3/52) LOAEL = NS NOAEL = 30 mg/kg d –

–

Non-cancer: Squamous hyperplasia (10/50, 15/49, 14/51, 31/55) LOAEL = 6 mg/kg d NOAEL = NS Cancer: Squamous papilloma/carcinoma (0/50, 48/50, 50/51, 54/55) LOAEL = 6 mg/kg d NOAEL = NS Non-cancer: Squamous hyperplasia (8/ 52, 29/51, 27/54, 34/56) LOAEL = 6 mg/kg d NOAEL = NS Cancer: Squamous papilloma/carcinoma (3/52, 50/51, 53/54, 55/56) LOAEL = 6 mg/kg d NOAEL = NS Non-cancer: Squamous hyperplasia (1/50, 25/49, 11/51, 15/52) LOAEL = 3 mg/kg d NOAEL = NS Cancer: Squamous papilloma/carcinoma (0/50, 16/49, 37/51, 19/52) LOAEL = 3 mg/kg d NOAEL = NS Forestomach No

Non-cancer: Squamous hyperplasia (3/50, 28/50, 13/49, 6/52) LOAEL = 3 mg/kg d NOAEL = NS Cancer: Squamous papilloma/carcinoma (0/50, 33/50, 42/49, 43/52) LOAEL = 3 mg/kg d NOAEL = NS

Female Mice (doses = 0, 6, 20, 60 mg/kg d)

Male Female

Rats (doses = 0, 3, 10, 30 mg/kg d)

Male

Tissue Human tissue homologue

Table 2 (continued)

Strong in both species

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

Magnitude of dose– response relationship

1496

Orally administered radiolabeled TCP was rapidly absorbed, metabolized, and excreted (Mahmood et al., 1991). In F344 rats and mice, approximately 90% of the compound was absorbed when a single radiolabeled dose of either 30 mg/kg (rats and mice) or 60 mg/kg (mice) was administered by corn oil gavage. Concentrations in tissues peaked at 6 h after administration, and >90% of the administered radiolabeled dose was excreted (50–57% via the urine) approximately 24 h after administration (terminal half-life estimated to be 23 h). Orally administered TCP was found to be distributed (in order of decreasing concentration) to forestomach tissue > liver > kidneys > large and small intestine > glandular stomach tissue > fat. TCP was found to be metabolized to N-acetyl-S-(3-chloro-2hydroxypropyl)-L-cysteine which was the major metabolite in rat urine and a minor metabolite in mouse urine (Mahmood et al., 1991). Also identified was the urinary metabolite 2-(s-glutathionyl)malonic acid). These investigators reported finding three radiolabeled metabolites in bile, one of which was the glutathione conjugate (others not specified). Approximately 2% of the exhaled radiolabeled dose of TCP remained as the parent compound. Formation of the two identified metabolites may have resulted from nucleophilic attack of water on episulfonium ion intermediates. The presence of these metabolites indicates that oxidation and glutathione conjugation play a substantial role in the metabolism of TCP in rodents, and offers suggestions as to similar metabolic pathways for humans. The disposition of TCP (3.6 mg/kg iv) has been evaluated in F344 rats using conventional and physiological toxicokinetics (Volp et al., 1984). TCP was cleared from blood in two phases, the first relatively short (0.3 h) and the second considerably longer (23 h). The clearance time of 14C from blood was 10 times longer in the first phase and approximately seven times longer in the second phase than TCP itself, suggestive of relatively rapid metabolism, excretion, and likely binding. Approximately 5% of the exhaled radiolabeled dose of TCP remained as the parent compound. TCP was excreted in urine (40% dose), bile (30% of dose with only 18% in feces, suggesting reabsorption from the intestines), and expired air (30% dose mostly as CO2 and a small fraction as parent compound), and was dependent upon its metabolism. Based on this excretion profile, TCP appears to behave similarly to DBCP (similar structure). The metabolism of TCP was found to be approximately 95% with approximately 5% being excreted unchanged. Based on evidence from the metabolism of other haloalkanes (i.e., DBCP, 1,2-dichloropropane, 1,2-dichloroethane), the investigators hypothesized that TCP would be metabolized by P-450 oxidation followed by glutathione conjugation. While TCP was distributed to all tissues examined, unchanged TCP remained in adipose tissue longer than in others; and high concentrations of metabolites were observed in the liver and kidneys. TCP’s metabolism has also been investigated in liver homogenates from humans and rats (Weber and Sipes, 1992). Using GC–mass spectrometry and radiolabeled TCP, Weber and Sipes reported that TCP was metabolized to 1,3-dichloroacetone (DCA). Due to its reactivity, DCA is a direct acting mutagen (Merrick et al., 1987) that has been hypothesized to be an intermediate responsible for carcinogenicity in rodents exposed chronically to high doses of TCP. Weber and Sipes (1992) demonstrated that pre-treatment with inducers of P-450 microsomal enzymes increased the rate of DCA production, whereas pre-treatment with inhibitors of those enzymes inhibited the formation of DCA – and TCP’s mutagenicity – suggesting that this pathway was important in the activation and detoxification of TCP. Furthermore, [14C]TCP was found to bind to microsomal proteins (another likely dimension of detoxification and possibly toxicity), particularly in cells pre-treated with P-450 activators. Addition of glutathione (GSH) or N-acetylcysteine to

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

1497

Fig. 1. Possible metabolic pathways of 1,2,3-trichloropropane.

the incubate completely inhibited protein binding, suggesting the importance of GSH in the detoxification of TCP and its reactive metabolites. When alcohol dehydrogenase and NADH were added to the microsomal incubations, two related alcohols (1,3-dichloro2-propanol and 2,3-dichloropropanol) were formed. The proposed metabolic pathways for TCP in rats and humans is presented in Fig. 1. Quantitative differences were observed in the formation of metabolites of TCP. Weber and Sipes (1992) reported that DCA is generated at a rate 10 times faster by rat microsomes than by human microsomes, suggesting that humans may be less susceptible (by virtue of producing less reactive metabolite(s) over time), by perhaps as much as an order of magnitude, to the mutagenic and carcinogenic influence of TCP compared to rats. 2.2.2. Evaluation of MoA(s) of TCP non-cancer toxicity and carcinogenicity Studies on MoAs may provide critical information relevant to conducting mechanistically based safety evaluations and risk assessments. This evaluation of TCP’s possible non-cancer and cancer MoA(s) related to repeated and prolonged exposures ap-

proaches it from two perspectives: (1) non-cancer toxicity for which the biology indicates generically that the dose–response relationship is non-linear (that is, having a threshold, meaning that its dose response reaches background levels of effects) and (2) carcinogenicity which is assumed to have a linear dose–response when a compound produces tumors by eliciting certain mutations that propagate through genetic replication, although it can also be demonstrated that exogenous mutations have a non-linear dose– response by reaching background endogenous mutation levels. Part of this consideration is not only the identification of potentially injurious molecular and cellular pathways but also the role of detoxification, repair, and saturation that can mitigate the manifestation of pathological consequences at the dose range of interest for humans. While it is possible to use default assumptions in lieu of insights into MoAs, particularly for tumor induction, for purposes of distinguishing safe from potentially injurious levels of human exposure to any chemical, it is consistent with regulatory risk assessment guides and practices that recommended that, when possible, scientific knowledge be given serious consideration for replacement of defaults (Sonich-Mullin et al., 2001; Meek et al., 2003; Boobis et al., 2006).

1498

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

2.2.2.1. Examination of TCP data related to non-cancer MoA. This analysis is complicated by high mortality of the test animals and by considerable loss of body weight (>20% of control) at the highest doses administered, suggesting that the high doses exceeded the MTD. It has been recognized in toxicology, including carcinogenicity, that doses exceeding the MTD are of little or questionable value in judging the MoA of a compound, such as TCP, that may be operating in the dose range which humans are likely to experience (namely, several orders of magnitude lower than the MTDs). Therefore, this evaluation focuses predominant on the toxicological events observed at doses at or somewhat greater than the reported LOAELs. In the subchronic (i.e., 3–4 months) rat toxicity studies, TCP’s target organs were consistently the kidneys and liver. At the LOAELs, gross changes were mainly increases in organ-to-body weight ratios. Histologically, little evidence of cellular changes were manifest, except in the drinking water study where substantial histopathological lesions (see Section 2.1.1) were observed at the LOAEL, suggesting the presence of oxidative stress leading to cellular dysfunction and possibly cell death followed by compensatory and reparative hyperplasia. At much higher doses, particularly in the corn oil gavage studies, however, signs of substantial toxicity in these organs were observed histopathologically (e.g., necrosis and hyperplasia) and via clinical chemistry measurements (e.g., increases in bilirubin and liver enzymes, indicating liver damage; and increases in BUN and creatinine denoting renal insufficiency). That kidneys and liver of rats are targets of TCP is consistent with the known distribution of TCP in the body of rodents [first-pass activation in liver (Volp et al., 1984); primary route of excretion of reactive metabolites via kidneys (Mahmood et al., 1991)]. The effects on the liver and kidneys are expected to exhibit a non-linear dose–response consistent with a practical biological threshold region for non-cancer effects. Furthermore, the subchronic liver changes are likely to be reversible, based on similar observations across an array of hepatoxicants [e.g., trichloroethylene (Nagaya et al., 1993), perchloroethylene (Kjellstrand et al., 1984), chloroform (Torkelson et al., 1976), and lindane (Junqueira et al., 1997)]. Renal insufficiency due to high doses of TCP, may be halted by cessation of exposure, but may not always be compensated through natural repair mechanisms. Other organ involvement was also observed in rats at dose levels greater than the LOAELs but they tended to be isolated in single studies, raising the possibility that the effects may have resulted from factors other than the compound. For instance among the studies using corn oil as the vehicle, Merrick et al. (1991) alone reported myocardial inflammation; and NTP in its 4-month study alone reported nasal turbinate necrosis and bronchial epithelium regeneration in TCP-treated rats, which may have been the result of the method of administration (since exhalation was only as CO2 and parent compound). Also the TCP drinking water study alone found changes in cellularity of the thyroid of treated animals at the LOAEL (Villeneuve et al., 1985), suggestive of possible reduced thyroid stimulating hormone production (not measured) resulting from general systemic toxicity. In none of these cases was a dose–response relationship identified, and neither TCP nor its metabolites were reported in these tissues. The inconsistency of these reported findings heightens certainty that these effects are not likely causal, and provide little basis for judging their MoA of TCP relevant for humans. By contrast, the drinking water study found bile duct hyperplasia in male and female rats (Villeneuve et al., 1985); this finding is consistent with the demonstrated biliary excretion of TCP (Volp et al., 1984; Mahmood et al., 1991) and with the hyperplasia observed in the pancreas (acinar cell) in the chronic study (NTP, 1993; Irwin et al., 1995). In the chronic/carcinogenicity study of rats and mice administered TCP, the sentinel non-cancer toxicity was hyperplasia in

the kidneys, pancreas, and forestomach at all doses (NTP, 1993). The pancreatic hyperplasia is consistent with the excretion of TCP and its metabolites in bile (Volp et al., 1984), and may be due to residence time in the pancreas prior to excretion and from secondary contact related to reabsorption from the intestines. Hyperplasia of the forestomach in rats may have arisen plausibly as a result of repeated tissue damage resulting from either irritation due to the bolus dose of TCP in direct contact with the tissue (La and Swenberg, 1997a); or hypothetically, the hyperplasia might have resulted from metabolites of TCP being absorbed and formed in the tissue; Mahmood et al. (1991) reported finding radiolabeled TCP/metabolites in forestomach tissue, however, the precise pathway(s) by which the [14C]-TCP was incorporated in intestinal tissue was not determined. The non-cancer data provide two observations that relate in part to TCP’s MoA in relation to the significance for humans. First, by one measure, duration of exposure appears to have resulted in a qualitative and quantitative change in non-cancer toxicity elicited by TCP. For example, among the corn oil gavage subchronic studies of rats, changing the duration of exposure from 3 to 4 months led to a reduction of the LOAEL by a half (from 15 to 8 mg/kg d); and increasing the duration from 3 or 4 months to 24 months reduced the LOAEL by approximately a third (from 8 to 3 mg/kg d) and also expanded the effects to include hyperplasia of the pancreas, kidneys, and forestomach, which had not been observed in the shorter duration studies. This change may be due to several possibilities including the accumulation of damage with repeated internal exposures that exceed the repair capabilities of the organisms or perhaps because of the variability of the design of the respective studies. Second, a significant difference in toxic potency is apparent between the two vehicles of administration, drinking water and corn oil. In rats, TCP in corn oil was several fold more potent toxicologically than when administered in drinking water. La et al. (1996) published a definitive work demonstrating the difference in DNA adduct formation in target (forestomach and liver), and non-target (glandular stomach and kidney) tissues of male mice exposed to [14C]TCP at a relatively low experimental dose of 6 mg/kg d in either tap water or corn oil for 2 weeks. They reported that, in forestomach, liver, and kidney tissues, that corn oil vehicle group exhibited between 1.4 and 2.4 times more of TCP’s major metabolite (the reactive GSH conjugate) than animals receiving TCP with tap water as the vehicle. They also found in these tissues a 3-fold greater rate of cell proliferation in the corn oil treated mice than those administered TCP in tap water. By contrast, the glandular stomach showed no difference in DNA adducts for each vehicle; however, cell proliferation was greater in the corn oil treated animals than in those administered TCP in tap water. This observation is not unique to TCP since similar observations have been made of other compounds (e.g., chloroform) (La et al., 1996). Further, La and Swenberg (1997a) compared DNA ras-gene-adduct formation in rats treated with TCP (30 mg/kg d for 1 week) by two routes of administration: drinking water and corn oil gavage. Significant increases in endogenously formed DNA adducts were found in liver and forestomach tissue of rats treated by corn oil gavage, whereas such adducts were not found in rats receiving TCP via drinking water, leading the investigators to indicate that observed increases most likely derive from the method of TCP administration: namely high local concentrations of TCP from the bolus exposure may deplete or saturate cellular defense mechanisms. Mice exhibited a different target organ profile than did rats. In mice, the liver effects (cellular necrosis and degeneration) occurred in males exposed to the two highest doses and in females receiving only the highest dose. Unlike in the rats, no histopathology was found in the kidneys of the mice at all doses, suggestive of a more effective clearance of TCP and its metabolite(s) than in the rats.

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

Furthermore, in mouse kidneys, no pathology was reported; whereas in the rat, renal tubular hyperplasia was observed (significant in males receiving 10 and 30 mg/kg d and in females receiving 30 mg/kg d). Hyperplasia of the pancreatic acinar cell was present in rats (significant at all doses in both sexes) but not in mice. In the reproductive system, clitoral hyperplasia (significance not stated) was observed in rats only, and endometrial hyperplasia was reported only in mice (significance not stated). Overall, evidence indicates that at doses sufficiently high to produce general systemic toxicity (even mortality at the highest test doses), damage to target tissues may be the result of overwhelming concentrations of metabolites in these tissues that exceed cellular and tissue defenses. For some studies, NOAELs have been observed, providing some indication of exposure levels at which point defense mechanisms are sufficiently robust to preclude functional impairment; and LOAELs, particularly for relatively mild forms of tissue damage, serve as a benchmark for the types of systemic changes that might serve as sentinels for consequences in humans similarly exposed to TCP. The toxicokinetic data of TCP in rodents provide important evidence that TCP’s metabolites are more important toxicologically than the parent compound; and also that at relatively low doses, the rat can effectively detoxify and eliminate reactive intermediates of TCP. Limited data suggest that, compared to rats, humans may be less capable of activating TCP yet more robust at detoxifying its reactive intermediates. An important observation from the data collectively is that TCP appears to be less potent toxicologically to rats via drinking water than by corn oil gavage. Corroborating data in rats were reported in the work of La et al. (1996). No comparable data exist in mice to determine whether the same quantitative relationship would be observed. Further research is needed to understand TCP’s non-cancer pathology at the molecular level. Using the non-cancer framework of Boobis et al. (2008), the human relevance of the rodent toxicity data was evaluated. To determine human relevance, Boobis et al. (2008) lists four questions to be answered. The first is whether the weight-of-evidence is sufficient to establish a MoA in animals. We conclude that the weight-of-evidence is limited due to the absence of key data at the molecular level (possibly oxidative stress or apoptosis) that indicate the pathway to the observed hyperplasia. As the first question in the evaluation was answered with a negative response, the remaining three questions are not addressed, indicating that the safety or risk assessment should continue. Therefore, human relevance of the non-cancer toxicity cannot be determined without additional data. Nonetheless, as has been demonstrated repeatedly with a wide assortment of chemicals, the non-cancer effects of the ingested TCP are threshold events. 2.2.2.2. Examination of TCP data related to cancer MoA. The tumor data for orally administered TCP are listed in Table 2. Evidence supporting possible and probable MoAs for the rodent carcinogenicity of TCP is evaluated to determine how and to what extent it may permit support for, or deviation from, default approaches and assumptions in distinguishing safe levels of TCP exposure from unacceptable levels of risk to human health. This analysis consists of examining (1) the role of biomarkers of TCP exposure, (2) biomarkers of effects (notably, mutagenicity as a measure of genotoxicity) as a partial or dominant MoA, and (3) possible non-genotoxic biological responses to TCP that might be involved in eliciting tumors in rodents. An internationally recognized framework analysis approach is then used to evaluation the relevance of the findings for humans. 2.2.2.2.1. Biomarkers of exposure to inform TCP’s carcinogenic MoA(s). Biomarkers of exposure are valuable in determining the MoA(s) of a laboratory animal carcinogen, particularly when seeking to determine the significance of animal tumors for humans. For

1499

a substance that may be a genotoxic carcinogen, such biomarkers include formation of adducts to macromolecules, particularly DNA, RNA, and proteins (Swenberg et al., 2008). As noted as early as 1997 by La and Swenberg (1997b), major insights can be obtained as to the MoA of a chemical carcinogen by understanding the intracellular dose and MoA of such adducts across species. Examples of the utility of such information include aflatoxin B1, vinyl chloride, ethylene oxide, propylene oxide (Swenberg et al., 2008), and acrylamide (Zeiger et al., 2009; Tardiff et al., 2010). Weber and Sipes (1990) described the extent of covalent binding of 30 mg/kg (ip) radiolabeled TCP-equivalents to F344 rat hepatic DNA, RNA, and protein. Covalent time course analyses indicated no change over 24 h in [14C]TCP-equivalents bound to hepatic DNA, but binding to hepatic protein was maximal at 4 h post administration and decreased significantly by 48 h post injection. Binding of [14C]-TCP to hepatic protein and DNA was found to be cumulative for two and three doses administered 24 h apart. Pre-treatment with an inhibitor of P-450 enzymes did not change the binding profile; however, depletion of GSH led to substantial increase in protein binding (likely due to reduction of competitive binding sites) and a substantial reduction in binding to DNA (perhaps due to change in metabolite profile). The findings suggest that GSH is involved in both activation of TCP and detoxification of reactive intermediate(s) of TCP [possible pathways of bioactivation were provided by the investigators]. Applying the ‘‘covalent binding index” of Lutz (1979), the investigators, estimated that TCP would likely be a ‘‘weak” to ‘‘moderate” hepatocarcinogen [subsequently, the NTP (1993) studies found TCP-related hepatic cancer in mice but not in rats, even at the highest dose tested]. In comparing the disposition and metabolism of TCP in rats (30 mg/kg) and mice (30 and 60 mg/kg), Mahmood et al. (1991) added support for covalent binding of a reactive metabolite of TCP in rat kidney, liver, and forestomach. They noted that, from 24 to 60 h post dosing, much of the [14C]TCP was not extractable from these tissues, and that which was extracted was metabolites and not parent compound. The bound radioactivity was determined to be released slowly from the tissues. Weber and Sipes (1991) investigated the possibility that TCP might cause DNA strand-breaks, DNA–DNA, or DNA–protein cross links in F344 rats exposed to 30, 100, or 300 mg/kg (ip). TCP (100 mg/kg, a cytotoxic dose) induced strand-breaks 1 h post treatment, and they decreased gradually over the following 48 h, suggesting the presence of some form of DNA repair. The breaks increased in a dose-dependent fashion from 30 to 100 mg/kg; but at 100 and 300 mg/kg, they were identical, suggesting saturation at approximately 100 mg/kg, a dose level that is many orders of magnitude higher than that experienced by humans. Neither DNA–DNA nor DNA–protein cross links were detected. La et al. (1995) examined DNA adduct formation in rats (F344) and mice (B6C3F1) exposed (ip) to [14C]TCP at doses of 3 or 30 mg/ kg (rats) and 6 and 60 mg/kg (mice) [dose identical to those in NTP’s carcinogenicity studies (NTP, 1993; Irwin et al., 1995)]. Tissues were excised for analysis for adducts 6 h post administration. The investigators isolated from target and non-target organs the DNA adduct S-[1(hydroxymethyl)-2-(N7-guanyl)-ethyl]glutathione, an adduct derived also from the structurally related carcinogen 1,2-dibromo-3-chloropane. [A postulated pathway is presented graphically therein.] Radiolabeled adduct levels varied depending upon species, organ, and dose. In rat tissues, [14C]TCP DNA adduct was measured in all but two tissues (preputial gland and palate) examined; whereas the adduct was measured in all mouse tissues (brain, forestomach, glandular stomach, heart, kidneys, liver, lung, spleen, and testes) examined. Adduct levels were as prevalent in target and non-target tissues, and, for the most part, with little difference in adduct concentration in target and non-target tissues. One exception was in the mouse where the adduct

1500

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

concentration at the low dose (6 mg/kg) was 50% higher in glandular stomach (non-target) than in the forestomach (target); and at the high dose (60 mg/kg), it was 5-fold higher in glandular stomach than forestomach. No clear relationship was found between adduct formation and organ-specific tumorigenesis, suggesting that factors other than adduct formation play a role in TCP-induced carcinogenesis and that adducts alone cannot be relied upon to serve as precursors of tumor formation. Furthermore, La et al. (1996) demonstrated that, in forestomach, liver, and kidney tissues, the corn oil vehicle group exhibited between 1.4 and 2.4 times more of TCP’s major metabolite (reactive GSH conjugate) than animals receiving TCP with tap water as the vehicle. These findings further indicate that human exposures to TCP via tap water are likely to be less potent toxicologically than that extrapolated from rodents exposed to TCP in the corn oil vehicle. Collectively, these results indicates that TCP, following metabolic transformation and certainly at high cellular concentrations, is capable of altering genetic material that, depending upon the cell’s ability to handle such change, may or may not be transformed into a mutation that may or may not elicit tumors. The findings also suggest that TCP-induced DNA adducts that might result from the much lower cellular concentrations that humans might experience, may be repaired prior to cell replication, thereby avoiding mutational events. Although assorted DNA adducts, such as S-[1-(hydroxymethyl)2-(N7-guanyl)ethyl]glutathione, have been associated with TCP identified by La et al. (1994, 1995), no evidence was found for the presence of adducts of O6-methyl guanine, which type has generally been associated with tumor initiation. According to Boysen et al. (2009), while N7-guanine adducts are formed frequently endogenously (and increase with age), they do not persist, and are unlikely to be mutagenic. Hence, these types of adducts are considered to have minimal biological relevance in relationship to exposure to mutagenic carcinogens. By contrast, these adducts are considered to be useful markers of internal exposures (Boysen et al., 2009). 2.2.2.2.2. Evidence for a mutagenic MoA for TCP tumor induction. An understanding of the MoA as the genotoxic basis for carcinogenicity is needed to determine consequences of chemical exposure on human populations and to improve health risk assessments. The detailed role of mutagenesis of chemical carcinogenesis has been presented by Swenberg et al. (2008). Salient elements of that evaluation are presented herein as background for understanding the significance of the genotoxicity findings of TCP. Mutations are essential for a substance to have a genotoxic MoA for tumorigenicity. Mutations comprise all irreversible changes in DNA’s primary structure, and may be manifested at the gene or chromosome levels. Faithful DNA replication is needed to maintain the physiological integrity of tissues and organs. Mutations are the result of errors or damage (e.g., DNA adducts) in the DNA that is propagated to daughter cells through replication. While they occur continually (Friedberg, 2003), endogenous mutations are nonetheless considered rare occurrences, and they have background rates that vary by tissue and increase with age. Several chemical processes, including normal metabolism, can chemically alter DNA to produce potentially mutagenic chemicals. Superimposed on this background are mutations from exogenous agents and/or their metabolites. DNA adducts are not mutations per se, and they may or may not be precursors, depending on how the cell processes them. Cells contain a complex array of processes to repair damaged DNA at varying rates and with varying fidelity, as a means of a cell to tolerate continuous damage from endogenous assaults (Friedberg et al., 2004). These protective mechanisms are also effective against DNA damage elicited by exogenous chemicals (Jenkins et al. (2005) found that different types of DNA repair con-

tribute to genotoxic thresholds for alkylating agents). When a cell successfully repairs DNA damage, the cell may continue to function normally, and eventually die to be replaced by its progeny. Consequently, because they are key events, mutations are the critical biomarkers for cancer safety and risk assessments of exogenous chemicals having a mutagenic MoA (Swenberg et al., 2008). TCP has been tested in several mutagenicity protocols to explore the prospect that these effects might play a role in TCP’s MoA in cancer elicitation. The combined results of the studies provide a basis to judge the degree of confidence that TCP has some potential to cause cancer in humans at doses of TCP that they may encounter. In a mutagencity and genotoxicity study of TCP and other members of the haloalkane class (using the in vitro micronucleus [MN] test of human lymphocytes), Tafazoli and Kirsch-Volders (1996) found no reproducible dose–response relationship in MN frequencies when TCP was applied (2–8 mM) with and without metabolic activation. The MN test is believed to detect clastogenic and aneugenic compounds. In comparing mutagenic profiles of treated cells, the investigators suggested a prevalence of deactivating pathways over activating pathways, leading to depletion of metabolites; they further suggested that the test cells might have a high repair capacity or selective elimination of heavily damaged cells. In this same study, TCP was also tested in the comet assay of human lymphocytes, and TCP was found to have strong mutagenic activity (as measured by a dose–response, without a NOAEL) in DNA damage when TCP was applied (2 and 4 mM) without metabolic activation; however, interpretation of these results is limited by high cellular mortality at the three highest test concentrations. TCP [concentrations not stated] has been found to induce mutations in three test strains of Salmonella typhimurium in the presence, but not the absence, of metabolic activation (NIOSH, 1981), consistent with the formation of stable reactive oxidative and GSH conjugate metabolites. In the S. typhimurium mutagenicity assays, TCP was found also to be mutagenic for strains TA100 and TA1535 in the presence of rat hepatic S9 fraction (Ratpan and Plaumann, 1985), and mutagenic in TA100 in the presence, but not absence, of S9 or liver microsomes (Mahmood et al., 1988). NTP (1993) also tested TCP in a battery of in vitro mutagenicity tests, and reported that TCP produced gene mutations in four strains of S. typhimurium with, but not in the absence of, metabolic activation (rat S9); however, no mutations were produced in S. typhimurium strain TA1537 either with or without metabolic activation. A dose–response relationship was observed for gene mutations in each of the four strains and the NOAELs range from 3 to 33 lg/ plate. TCP also produced sister chromatid exchanges and chromosomal aberrations in Chinese hamster ovary (CHO) cells in the presence, but not the absence, of metabolic activation. A NOAEL of 4.7 lg/ml was determined for sister chromatid exchanges, while no NOAEL was found for chromosomal aberrations (LOAEL of 59.5 lg/ml). No dose–response was observed for the results in the CHO cells (NTP, 1993). Furthermore, TCP was positive in the mouse lymphoma assay in the presence, but not the absence, of metabolic activation (NOAEL was 8.4 lg/ml [57 lM]) with a dose response (NTP, 1993, 1999). TCP produced no incidence of dominant lethal mutations in rats (Saito-Suzuki et al., 1982). An important measure of a mutagenic MoA for carcinogenicity is the presence of mutations in the tumors related to the administered compound (Swenberg et al., 2008). Mutation of the ras gene was reported in 10 of 16 TCP-induced tumors of the forestomach of mice. Six tumors had K-ras and H-ras mutations at codon 61, with 5 of 6 having AT to TA transversions in base 2; four tumors had Kras mutations at codon 13, with GC and CG transversions in base 1. The investigators reported finding that TCP administration produced increases in etheno DNA adducts, 1,N6-ethenodeoxyadenosine and 3,N4-ethenodeoxycytodine, which are likely to arise

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

endogenously from lipid peroxidation (Ito et al., 1996), raising the possibility that this may be an indirect MoA for TCP’s carcinogenicity in rodents. According to the investigators, these mutations are inconsistent with miscoding properties of the major DNA adducts, suggesting an alternative MoA. Since TCP administered in corn oil gavage in the dose range that produced forestomach tumors in rodents decreases cellular glutathione levels in the forestomach, the observed mutagenic effect may have been the indirect result of the method of TCP administration (Ito et al., 1996). As a result, genotoxicity appears to be implicated in the high dose carcinogenicity of TCP observed in rodents exposed by oral gavage for a lifetime, most likely via one or more of TCP’s stable reactive metabolites. Historically, such a finding would likely have led some to apply as a default the linear mathematical extrapolation of cancer incidence in the observation range of rodents to the zero intersect of dose and estimated risk of cancer for humans. However, by taking into account the background of endogenous mutations, one should modify that default position to one in which linear extrapolation to low doses of an exogenous mutagen (viz., TCP) reaches the background level(s) of mutations rather than the zero intersect. In essence at such low doses, the dose–response curve for mutations takes on a non-linear shape, consistent with a practical threshold (that is, positive doses below which no increase in mutation frequency exists above that observed in the unexposed control animals and unexposed humans) (Swenberg et al., 2008). It is noteworthy that the dose–response curve for mutations may be non-linear, even if the associated DNA adduct dose–response is linear. However, the available mutagenicity data on TCP provide limited dose–response data from which to extrapolate the compound’s mutagenic potency either toward or into the region where endogenous mutations are prevalent. For TCP to have a non-linear dose–response for mutagenicity in rodent and human cells would not be novel. Several carcinogens with mutagenic MoAs having non-linear dose–response curves for mutagenesis include the following [details provided in Section 3]: acrylamide (Swenberg et al., 2008; Zeiger et al., 2009), dibenzo[a,l]pyrene (Bailey et al., 2009), mitomycin C, Ara-C, colchicine (Asano et al., 2006), ethylene oxide (Swenberg et al., 2008), methylmethane sulfonate (MMS), ethylmethane sulfonate (EMS) (Doak et al., 2007; Johnson et al., 2009), dimethylhydrazine (Parry et al., 2000), ethylnitrosourea (ENU) (Guttenplan, 1990; Favor, 1998; Parry et al., 2000), and other nitroso compounds (Guttenplan,1990). Collectively these studies provide evidence for the existence of a practical biological threshold for non-linear mutagenic carcinogens, suggesting consistency with TCP’s mutagenic MoA. 2.2.2.2.3. Evidence for non-genotoxic MoA for TCP tumor induction. A non-genotoxic MoA may also be acting at tumor sites. TCP, when administered by gavage to male B6C3F1 mice, increases cell proliferation rates (BrdU Labeling Index, LI), while the same daily dose given in drinking water produced no such effect. Mice were given TCP by gavage or drinking water (6 mg/kg d for 10 days over 2 weeks). LI in forestomach of gavage-treated mice were roughly 3fold higher than controls at both 18 and 30 h after the last gavage dose of TCP (La et al., 1996). These findings are consistent with the greater cytotoxicity observed for TCP when administered by gavage (NTP, 1993) rather than by drinking water (Villeneuve et al., 1985). This cytotoxicity can lead eventually to tumor formation, when exposure is repeated and prolonged. Furthermore, the etheno DNA adducts produced by TCP but which are likely to arise from lipid peroxidation also provide support for the indirect tumor effect of TCP as an alternative MoA (Ito et al., 1996). The possibility cannot be excluded that TCP causes cancer in rats and mice by a combination of genotoxic and non-genotoxic MoAs (US EPA, 2005). Whether one MoA might be dominant or whether different MoAs are activated at different dose levels is un-

1501

clear presently. Nonetheless, each MoA appears likely operate in ways that the resulting dose–response relationship would be non-linear, providing guidance in extrapolating the data in rodent studies to the considerably lower dose range of TCP which humans may experience. Using the cancer Framework initially developed by Meek et al. (2003) and subsequently updated by Boobis et al. (2006), the human relevance of the rodent carcinogenicity data was evaluated. This Framework is a guideline of the World Health Organization’s (WHO) International Programme of Chemical Safety (IPCS). To determine human relevance, Boobis et al. (2006) list three questions to be answered. The first question seeks to determine whether the weight-ofevidence is sufficient to establish a MoA in test animals; and that is addressed by the framework for determining a cancer MoA adopted by IPCS (Sonich-Mullin et al., 2001). In accordance with the criteria in this framework, a mutagenic MoA for TCP carcinogenicity has been postulated, and key events (metabolism to DNA-reactive and stable metabolites compounds and DNA adduct formation) have been documented, and are supported by the presence of critical parameters (i.e., positive bacterial assays, covalent binding of TCP metabolites to DNA, RNA, and hepatic proteins, and DNA strand-breaks). The weight-of-evidence for a mutagenic MoA is strengthened by the (1) dose-dependent formation of DNA adducts, (2) the temporal relationship between TCP administration and adduct formation and mutagenic events, (3) the consistency of findings across studies (although less so when comparing species), and (4) specificity of the effects due to reactive metabolites. The mutagenic MoA is biologically plausible, although an alternative MoA (cytotoxicity with regenerative and reparative hyperplasia) is also plausible at high doses. Inconsistencies include predominantly the distribution of DNA adducts within rodent tissues that are only indicative of exposure and are not well correlated with tumor formation, and the limited concordance of target tissue sites between species. The hypothesized MoA carries some uncertainty based on the very high doses causing major noncancer toxicity and the attendant high mortality at assorted test doses. Overall, however, the weight-of-evidence appears sufficient to support a mutagenic MoA for TCP-induced carcinogenicity in rodents exposed at high doses daily for a large of their natural lifespan. The second question of the Boobis Framework is whether human relevance of the MoA can be reasonably excluded on the basis of fundamental differences in key events between experimental animals and humans. No evidence convincingly supports the exclusion of the mutagenesis MoA for carcinogenesis in rodents exposed to high doses of TCP, although the hyperplasia which appears to be reparative may play some role in TCP’s elicitation of some tumors. Nonetheless, as a corollary, some types of tumors are judged to not be relevant for humans. Specifically, the pancreatic acinar adenomas are unlikely to be relevant to humans because: (1) the histologic type of tumor (i.e., acinar cell) seen in rodents is distinctly different from tumors of the exocrine pancreas most commonly observed in humans (i.e., ductular) (Longnecker and Millar, 1990); (2) differentiation between hyperplasia and adenoma of acinar cells can lead to misclassification of proliferative cells as adenomas, since the distinction is predicated largely on twodimensional size of the lesion (Eustis and Boorman, 1990; Hansen et al., 1995), possibly leading to overstatement of incidence of adenomas; (3) humans regulate pancreas exocrine secretion via neuronal pathways rather than through direct binding of receptors, such as cholecystokinin, which is expressed to a far lesser extent in human pancreas than in rat pancreas (Ji et al., 2001; Longnecker, 1987; Owyang, 1996); and (4) corn oil alone has been shown to have a significant effect of inducing pancreatic hyperplasia and

1502

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

adenomas by at least 5-fold (Eustis and Boorman, 1985; NTP, 1994), thereby confounding the interpretation of such findings in animals treated with test compounds. Consequently, the weightof-evidence suggests little confidence that this benign tumor type in male rats is relevant for humans. Hence, these pancreatic acinar cell adenomas were excluded from further consideration. Zymbal’s and Harderian gland tumors are considered not applicable to humans, since these tumor sites lack human tissue homologues (Albert et al., 1986; Sheldon, 1994). The Harderian gland is an ocular gland with tubular alveoli present in some animals (rodents, amphibians, reptiles, and birds) but absent in others (dogs, cats, sheep, and goats) (Albert et al., 1986; Sheldon, 1994). In rodents, the gland opens on the outer surface of the third eyelid, where it secretes lipids and porphyrins. Incidence of Harderian gland tumors in rodents has been correlated with porphyrin content (Figge et al., 1942). Forestomach tumors are also considered not relevant to humans. The foremost reason is that it is species-specific. The forestomach is a portion of the gastrointestinal tract of some animals with a squamous cell epithelium located between the esophagus and glandular stomach (Wester and Kroes, 1988; Kroes and Wester, 1986). The forestomach serves as an organ for storage and predigestion of food material. Although the forestomach was one of the few target organs identified in both test species, two reasons indicate that forestomach tumors are of questionable relevance to human health, the first pertaining to biology and the second relating to MoA. With respect to biology, no tissue homologue exists in humans that corresponds directly to the rodent forestomach (Wester and Kroes, 1988; Kroes and Wester, 1986). With respect to MoA, the forestomach typically achieves high concentrations of the chemical following gavage, which occurs over an extended period of exposure time. In a number of cases, these two factors combine to result in forestomach hyperplasia and inflammation (as observed in TCP-exposed rodents), which in turn lead to tumorigenic expression through a tumor promotion mechanism such as reparative hyperplasia. This process is not unique to TCP in as much non-genotoxic chemicals, such as butylated hydroxyanisole and sodium chloride, have been shown to produce forestomach tumors at high concentrations by such a MoA (Ito et al., 1983; Furihata et al., 1996). The third question asks whether human relevance of the MoA can be reasonably excluded on the basis of quantitative differences in kinetics or dynamics between experimental animals and humans. Evidence indicates that the mutagenic MoA for TCP’s carcinogenicity is not likely to operate at human doses 5–6 orders of magnitude lower than those tested in rodents, largely because the dose–response relationship for mutagenesis is expected to (1) reach background rates of mutations (Swenberg et al., 2008), (2) not end at the zero intersect (Swenberg et al., 2008), and (3) unlikely be additive to background because of the presence of effective DNA repair rates. In essence, the mutagenic MoA represents a non-linear dose–response, consistent with a practical biological threshold. In addition, the kinetic behavior of TCP at doses considerably less than the range of doses to which rodents were exposed is expected to favor detoxification over activation. Therefore, the rodent carcinogenicity data are not likely to be relevant to humans at the levels to which they might be exposed via drinking water. The level of confidence is moderate to high for the estimated DWELs and RfDs for TCP consumed by humans in drinking water

for a lifetime. The quality of the chronic toxicity/carcinogenicity studies is adequate in terms of much of the design and execution; however, the dose selection was relatively poor because of the high premature mortality in the high dose groups, suggesting that the MTD had been exceeded. Furthermore, a weakness of these studies is reliance on corn oil gavage rather than tap water as a vehicle. While tap water administration, had it been used, may have increased the variability of dosing, it would have removed several important factors that undermined the ability to estimate toxic potency for purposes of risk assessment and safety evaluation of TCP. 2.3. Estimation of DWEL, RfD, and cancer values (CV; RfD-equivalents) The approach used to estimate a DWEL and RfD for TCP is characteristic, structurally, of that promulgated by the NRC (1983), applied by regulatory and non-governmental organizations worldwide (WHO, 2006; US EPA, 2005; Health Canada, 1994), and used by scientists in academics and industry. For a biologically-based DWEL estimation within this structure, the detailed weight-of-evidence evaluations may lead to replacement of defaults described in assorted guidelines. Eight steps are included in estimating a DWEL for TCP, and are described below. The detailed evaluation in each step is intended to make choices that are supported by the data on TCP and on the general body of knowledge related to the determination of the relevance of laboratory animal carcinogenicity findings to humans, at times displacing default processes and values. Non-cancer and cancer dose–response assessments were conducted in the following steps:  Selection of key events and studies – Major non-cancer endpoints are listed in Tables 2 and 3; from these, renal tubular hyperplasia was selected as the key event relevant for humans because of the consistency with which the kidneys were a target organ in subchronic studies and the understanding that the effect is likely due to the influence of one or more of TCP’s stable reactive metabolites (GSH conjugate and the oxidative product DCA), the likelihood that this hyperplasia (compensatory or otherwise) may be a precursor to renal tumors in rats and mice, and the extent to which DNA adducts (indices of a fraction of tissue dose of TCP) were present. The key study identified is the chronic study (NTP, 1993; Irwin et al., 1995) because of its relatively long duration, extensive histopathological examinations, and consistency with the human duration of exposure envisioned for the estimated non-cancer DWEL – notwithstanding the limitation of extensive premature mortality. Liver pathology was judged to be of somewhat lesser value because the histopathological non-cancer effects occurred in mice only; however, this organ serves a major role in the bioactivation and detoxification of TCP. Pancreatic acinar cell hyperplasia was not selected because the effect was reported only in gavage studies with corn oil, not the drinking water analysis, and corn oil alone has been reported to induce considerable pancreatic acinar hyperplasia in control rats, thus confounding the results and making the relevance to humans uncertain (as previously described; Eustis and Boorman, 1985; NTP, 1994). The preputial and clitoral hyperplasia in male and female rats and endometrial hyperplasia in female mice were not selected because they occurred in only one species and because of their low and not statistically significant incidences from TCP exposure in the chronic studies. No non-cancer pathology was reported for the mammary gland, oral epithelium, and intestines. And the following organs were not selected because they

Table 3 Summary of benchmark dose results for the non-cancer effects of TCP in rats. Endpoint

Data set

Model

AIC

p-Value

BMD10 (mg/kg d)

BMDL (mg/kg d)

Renal tubule hyperplasia

Combined sexes

Log-logistic

282.3

0.135

5.2

3.9

1503

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510 Table 4 Benchmark dose results for the cancer effects of TCP in rats and mice. Tumor site

Species

Sex

Model

AIC

p-Value

BMD10 (mg/kg d)

BMDL (mg/kg d)

Oral cavity Kidney Mammary Clitoral Preputial Liver Uterus Oral cavity

Rat Rat Rat Rat Rat Mouse Mouse Mouse

Combined sexes Male (high dose excluded) Female Female Male Combined sexes Female Combined sexes

Log-logistic Multistage Log-logistic Log-logistic Probit Probit Log-logistic Probit

361.0 85.1 177.4 214.5 178.5 521.4 113.9 89.2

0.085 0.981 0.645 0.091 0.987 0.093 0.060 0.544

2.6 4.5 4.6 11.9 15.0 16.2 36.9 76.3

1.6 2.9 3.1 5.4 10.9 13.2 18.6 58.9

had no tissue homologue in humans: Harderian and Zymbal glands, and forestomach tissue. Eight tumor endpoints in rats and mice (by sex in which tumors were reported) considered relevant for humans are listed in Table 4. Each was selected as a key event because tumors were generated at the LOAEL, a dose which was least likely to have been influenced by premature mortality and excess loss of body weight, the tumor sites have a homologous counterpart in humans, and a dose response was observed for each tumor type. Tumor sites not selected were the pancreas (not considered relevant to humans as previously described), intestine (incidence not significant, and NTP considered findings to be uncertain), Harderian gland (no human homologue), Zymbal gland (no human homologue), and forestomach (no human homologue). The key study identified is the only chronic study (NTP, 1993; Irwin et al., 1995), and was selected because of its relatively long duration, extensive histopathologic examinations, and consistency with the human duration of exposure envisioned for the estimated non-cancer DWEL – notwithstanding the limitation of extensive premature mortality.  Dose metric selection – Because of differences in body weight (as well as metabolic rates and renal clearance) between humans and rodents, adjustment or ‘‘scaling” is considered when applying a dose or dose range to humans from doses administered to rodents. The default approach for interspecies extrapolation involves allometric scaling of dose using body weight raised to the 3/4 power, regardless of the MoA and dose ratio. However, recommendations have been made to consider the MoA when determining the scaling factor, such that an exponent value of 1.0 is appropriate to apply to body weight for chemicals involving a stable reactive metabolite (Clewell et al., 2002), which is likely the case for TCP (N-acetyl-S-(3-chloro-2-hydroxypropyl)-L-cysteine, 2-(s-glutathionyl)malonic acid, and DCA) [noted in Section 2.2.1]. The rationale for the choice is a reported dose-dependency for allometric scaling based on ratios of human to rodent internal doses (Kirman et al., 2003), demonstrating that an exponent of 1.0 as a scaling factor of body weight has an empirical basis, and is more precise in characterizing the shape of the dose–response curve than the power of 3/4 when gavage doses of 10 mg/kg d are used in the test study, as is the case for all but the top dose of TCP for rats in the NTP (1993) study [lowest doses for rats were 3 and 10 mg/kg d for rats; lowest dose for mice was 6 mg/kg d]. Therefore, use of an allometric scaling factor of 1.0 (i.e., unscaled dose) was selected to assess TCP-induced tumors.  Dose–response assessment – The benchmark dose (BMD) approach was selected for this task because it contains the strongest support for characterizing dose–response relationships in the observation range and extending them downward toward the dose range experienced by humans (US EPA, 1995; WHO, 2009). Hence, dose–response modeling was conducted using US EPA’s BMD Software (version 2.1.1), which contains models for dichotomous data (gamma, logistic, log-logistic, multistage, probit, log-probit, quantal-linear, and Weibull). The best-fitting model for each data set was selected based upon visual inspection, Akaike’s Information Criterion (AIC), chi-square goodness of fit p-value, and model

uncertainty. Because the pooled incidence data were converted to extra risk, the background incidence term for the models was fixed at zero. The dose–response models were used to characterize a point of departure (POD, e.g., dose corresponding a 10% increase in extra risk [BMD10] and dose corresponding to the 95% lower confidence limit [BMDL]) for the non-cancer endpoint and the assorted tumor endpoints. To account for differences in the tumor rates in control animals, the incidence data for each dose group were converted to extra risk, where

½Extra Risk ¼ ðIncidence  Control IncidenceÞ =ð1  Control IncidenceÞ: The BMDL (3.9 mg/kg d) obtained from the renal tubule hyperplasia rat data set (combined sexes) was selected as the basis for estimating the RfD (Table 3). This value might be reduced by 50–90% when one considers that the non-cancer potency of TCP consumed in drinking water (if humans were to consume it) is considerably less than when ingested in corn oil. The selected tumor data in rats and mice (Table 4) were combined in a distributional (Monte Carlo) analysis using the BMD10s and BMDLs to assess cancer potency (Table 5; method described below). The resulting mean 50th percentile and lower bound 90th confidence interval (LB CI) were used to estimate a Cancer Value ([CV] RfD-equivalent).  Distributional (Monte Carlo) analysis of tumor data – This procedure adds robustness to the analysis by consolidating multiple data sets and taking into account variability of the data. Although the tumor data sets were assessed separately in a BMD analysis, Monte Carlo methods were applied to characterize a potency distribution for TCP based upon the combined results of potency estimates for each tumor type (Table 4). A beta-PERT distribution was assumed for the potency estimates based upon each tumor type using a minimum of zero, a best estimate based upon 0.1/ BMD10, and a 95th percentile based upon 0.1/BMDL. Sums were calculated assuming independence of tumors (i.e., the probability of rodents having more than one tumor was subtracted from the sum). Combined tumor potency estimates were calculated by summing the individual tumor potencies. Because reproductive tumors are sex-specific, separate calculations were made for males (based upon the sum of liver, kidney, oral cavity, and preputial tumors) and females (based upon the sum of liver, mammary gland, oral cavity and reproductive tissue tumors). The BMD10 value for combined tumors was calculated as 0.1/combined potency estimates. All Monte Carlo simulations were performed using Crystal Ball (Oracle; Version 11.1.1) for Microsoft Excel, and included 10,000 Table 5 Monte Carlo results for cancer potency estimates based on combined tumors. BMD10 (mg/kg d)

Male Female Mean

50th Percentile

90% CI

1.4 1.3 1.4

1.0–2.4 0.9–2.4 1.0–2.4

1504

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

iterations. No correlations were introduced among potency assumptions. The results are reported in Table 5.  PoD selection – For the non-cancer data set, the dose corresponding to the BMDL was used for the PoD (Table 3), namely 3.9 mg/kg d for combined male and female rats. For the cancer data sets, the doses corresponding to BMD10 and the BMDL were used in the Monte Carlo distributional analysis to estimate the BMD10 50th percentile and the 90th LB CI which were used for the PoD (Tables 4 and 5). While the BMDL can serve as the default PoD for quantal data sets (US EPA, 2005), at times, other values for the BMDL have been employed, depending on the type and quality of the incidence data. For the TCP cancer data sets, the BMD10 and distributional analyses provide a more robust approach than the default BMDL alone. The results for combined male and female values indicated that the 50th percentile BMD10 was estimated to be 1.4 mg/kg d, with the 90th CI (lowerand upper) = 1.0 and 2.4 mg/kg d, respectively (Table 5). In addition, a sensitivity analysis indicated that the greatest contribution to the combined values was made by the male rat kidney tumors.  Low dose extrapolation/uncertainty factor (UF) development – This step contains two possible avenues: (1) for effects known or expected to have a biological threshold, the application of UF from the PoD to estimate a human dose expected to be with little or no chance of producing injury and (2) for effects assumed to have no biological threshold (as a default, mutagenic carcinogens), the application of some mathematical model that extrapolates disease incidence from the PoD (generally, over several orders of magnitude) to an estimated risk of disease which is defined as ‘‘acceptable” or ‘‘tolerable” (e.g., the upper bound of one incidence of disease or death in a population of say 100,000 individuals over a lifespan). [Consideration of linear extrapolation for TCP is provided in Section 3] Within the latter avenue, when suitable biological information is available, the safety and risk characterization can displace the default approach (even for a mutagenic carcinogen) with UF, reflecting the view that a substance has, or is more likely to have, a biological threshold for humans, as proposed for mutagenesis resulting from high doses of TCP. When UF are employed, five are included, and they account for (1) interspecies variation (UFa), (2) intraspecies variation (UFh), (3) use of a LOAEL (UFl), (4) subchronic-to-chronic duration adjustment (UFs), and (5) database deficiencies (UFd) (US EPA, 1994, 2002; WHO, 2006). The sum of the UF is described by the notation UFt. These UFs were applied to TCP’s PoDs to obtain a RfD for noncancer toxicty and a CV (RfD-equivalant) for cancer. Interspecies variability in susceptibility (UFa) – This factor addresses interspecies variability in susceptibility between humans and rats or mice that may arise from differences and similarities in metabolism, toxic response(s), and MoA. This UF’s default value of 10, unless adequate data are available to quantify the toxicokinetic and toxicodynamic (‘‘pharmacodynamics”) variability between humans and laboratory animals. This factor is composed of two parts: toxicokinetics and toxicodynamics, which is accounted for by a factor of 4 for toxicokinetics and 2.5 for toxicodynamics in the default UF of 10 (WHO, 2006). For TCP, in vitro data with human cells and comparisons of LOAELs and NOAELs suggests that the potency of TCP in humans may be 5–10 times less than that in rats. The default value of 10 was used for each key endpoint and data set since only suggestive data provided a basis for an alternative value. Intraspecies (human) variability in susceptibility (UFh) – By virtue of their genetic diversity, humans range in susceptibility to exogenous chemicals including pharmaceuticals and chemicals present in the human environment. This factor seeks to address this distribution of sensitivities across a large human population, including those individuals with unique susceptibilities, if any. In general, since many risk factors are believed to confer excess risks of approximately 10 on predisposed persons when compared with

‘‘average” individuals, a default UFh of 10 is considered a reasonable starting place. A 10-fold adjustment is likely to produce a best estimate of the high end of the susceptibility of the distribution for some chemicals (NRC, 1994). The default for this factor is 10 unless adequate data are available to quantify the toxicokinetic and toxicodynamic variability within the human population, in which case a factor of either 3 or 1 may be applied (WHO, 2006; US EPA, 2002). For TCP, no epidemiologic data exist to provide guidance as to the range of human susceptibilities to TCP. No biochemical data (such as genetic polymorphism) indicate one or more subpopulation of humans that is, or might be, more uniquely susceptible to the toxicity from the same effective dose of TCP than the general population. Since TCP stable reactive metabolites are responsible for its non-cancer toxicity and carcinogenicity in rodents, consideration of variations in the microsomal metabolism may offer insight into inter-individual susceptibility to TCP: a study of human intestinal P-450 isoforms indicated that the specific activity was somewhat less than 10-fold (approximately 5–7) (Paine et al., 2006). Consequently, the default value of 10 was selected for each key endpoint and data set. LOAEL to NOAEL conversion (UFl) – When the key data reveal a LOAEL but no NOAEL, this factor is employed to extend downward from the one value to the other. While the default for this factor is 10, considerations such as dose spacing of the key study, slope of the dose–response curve in the experimental range, and use of the entire dose–response curve rather than a single point can be used to modify this default value to either 3 or 1. For TCP, a benchmark dose analysis was conducted incorporating all reliable dose–response data in the observation ranges instead of a single dose, thereby, increasing robustness and reducing uncertainty. It also incorporated a measure of variability based on consideration of the full set of data into the calculations through the application of distributional analysis. Consequently, the BMD values are considered more accurate and reliable than either NOAEL or a LOAEL alone, thereby providing adequate scientific support for the use of a UF of 1. Adjustment for Duration of Exposure from Subchronic to Chronic (UFs) – In cases where a key event is based on subchronic (e.g., 5–10% of the lifespan) duration of exposure but a toxicity value must be estimated for a lifetime of exposure, some adjustment can be considered to bridge this difference. This adjustment would be needed particularly when the body burden of a compound increases in susceptible target tissues with repeated dosing (as is the case for some substances) and/or when the toxic potency increases with increasing duration of exposure. Information from toxicokinetic and toxicodynamic studies may be particularly illuminating about such properties. For this adjustment, the default is 10, but can be decreased to 3 or 1 depending upon the nature and magnitude of data that allow comparison of TCP’s toxic potency for differing durations of exposure. However, when the key study is of a chronic duration (e.g., 24 months for rats), the factor of 1 is justifiable. Since the key studies relied upon for TCP are of chronic duration, an UFs value of 1 is justified, and has been applied. Adjustment for database deficiencies (UFd) – At times, concerns are raised that human exposures to chemicals in general, and persistent ones in particular, may pose a risk of injury, particularly to reproductive performance and to the development of offspring in utero and post partum. Hence, evaluation of the safety and risk posed by a chemical widely distributed in the human environment is expected to include empirical evidence addressing pathological outcomes that could not be fully addressed in subchronic and chronic toxicity studies. The default value for this factor is 10, which can be reduced to 3 or 1 depending upon the quality and quantity of specialized studies of assorted physiological and biochemical processes.

1505

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

The TCP toxicity database includes not only traditional studies (acute, subchronic, chronic), in both rats and mice, but also studies aimed at measuring mutagenicity (in numerous systems), MoA for non-cancer and cancer, and behavior in the body (absorption, distribution, metabolism, excretion). In addition, an oral multi-generation reproductive toxicity has been conducted in mice (NTP, 1990), reductions in fertility has been reported, and NOAELs have been identified, but at higher doses than those for other types of systemic toxicity. Furthermore, the repeated and prolonged dose studies (referenced herein) found no indication of either neurological deficits (except lethargy at near lethal doses) or immune system deficiencies even at doses that produced serious toxicity to other organs and functions (US EPA, 2009a). Consequently, the UF for this consideration is selected as 1, indicating no significant gaps in investigating TCP’s toxicity. This value of UFd is not intended to suggest discouragement of further toxicologic research on TCP but rather to indicate that the available data are sufficient presently for defining safe levels of chronic exposure for humans. UFs were selected and applied to each non-cancer and cancer PoD to estimate RfDs and CVs (RfD-equivalents), respectively, with a UFt = 100 as determined by: (1) UFa = 10; (2) UFh = 10; (3) UFl = 1; (4) UFs = 1; and (5) UFd = 1. An oral non-cancer RfD and a non-linear cancer value (CV; RfDequivalent) were derived for TCP using Eq. (1):

RfD or CV ðmg=kg dÞ ¼ PoD=UFt

ð1Þ

where CV = oral cancer value for a lifetime (mg/kg d); RfD = reference dose for non-cancer toxicity for a lifetime (mg/kg d); PoD = point of departure (human equivalent benchmark dose, mg/kg d); UFt = total uncertainty factors (UFa  UFh  UFl  UFs  Ufd); [100 = (10  10  1  1  1)]. The UF was applied to the BMDLs for non-cancer toxicity to obtain the corresponding RfD (Table 6):

Rat renal tubule hyperplasia ðcombined sexesÞ ¼ 0:039 mg=kg d: The UFt was applied to obtain the corresponding CVs (RfDequivalents):

Combined tumors ðBMD10 50th percentileÞ ¼ 0:014 mg=kg d; Combined tumors ðBMD10 LB 90th percentileÞ ¼ 0:010 mg=kg d:  Relative source contribution (RSC) – Because the RfD and CV apply to total exposure, the relative source contribution (RSC) is used to apportion a specific fraction of the RfD or CV to drinking water including water used for food preparation. Historically, the RSC has ranged from 20% to 80% of the RfD, depending upon contributions to total exposure from non-tap water sources.

Using a default of 20% for tap water exposure would provide the most conservative DWEL for TCP. However, evidence suggests that exposures to TCP in media other than tap water are presently small. As noted previously, primary exposure of the general US population to TCP would be via water, and its presence has been reported in water in six states suggesting limited environmental distribution. When measured in tap water, TCP has usually been found at <10 ppb. Concentrations of TCP in the ambient air have not been reported in the US, but have been reported at a maximum of 0.4 lg/m3 in Germany and 0.21 lg/m3 in Canada. No TCP has been reported in foods. In the US, TCP has been detected in the drinking water of New York, New Jersey, Louisiana, Ohio, California, and Hawaii. After sampling over 900 groundwater samples from the US during the mid-1980s, the median concentration was 0.69 lg/L (ATSDR, 1992). During the 1980s, TCP was detected in drinking water at 0.1–5 lg/L in Hawaii (maximum concentration on Oahu in 1989 was 0.3–2.8 lg/L) and in California (ATSDR, 1992; WHO, 2003; HSDB, 2009). Between 1989 and 1999 in California, TCP was detected at concentrations ranging from 0.02 to 74 lg/L and between 2000 and 2006, concentrations ranged from 0.01 to 57 lg/L (CDPH, 2007). TCP was also detected in the soil of California and Hawaii during the 1980s at 0.2–2 ppb (ATSDR, 1992). TCP is effectively reduced in water through the use of granular activated carbon during the treatment of groundwater to produce water for human consumption; in Hawaii, this process has been applied locally for over a decade (Kawata, 1992). Consequently, a RSC for TCP in tap water of 50% should be sufficient to preclude exposures from exceeding the RfD or CV, and has been selected for estimation of the DWEL for TCP. This RSC was applied to the LADI (mg/person d), and the resulting value is divided by the number of liters of tap water consumed daily, traditionally 2 L per adult, to determine the DWEL.  Estimation of DWEL (unit of amount of TCP per liter of tap water per day) for each data set. The DWEL is obtained by multiplying the Lifetime Average Daily Intake (LADI) by the RSC and dividing the product by volume (L) of water consumed daily by an individual. where,

LADI ¼ RfD ½or CV  body weight ½¼ 80 kg for male and female adults for a lifespan of 75 years for males and 80 for females: Note that the body weight of 80 kg used herein is that which has been recently promulgated by US EPA (2009b) as part of its Exposure Factors Handbook and represents more recently compiled demographic data for humans. Likewise, the lifespan for both sexes of humans has been revised recently from 70 years to 75 and 80 for males and females, respectively (US EPA, 2009b).

Table 6 DWEL calculations for TCP-related to non-cancer and cancer data. Endpoint Non-cancer Renal tubule hyperplasia (combined sexes) Cancer Combined tumorse; 50th percentile (see Table 5) Combined tumorse; lower 90% CI (see Table 5) a b c d e

BMDL or BMD10a (mg/kg d)

UFt

RfD or CVb (mg/ kg d)

LADIc (mg/ person d)

Adjusted by RSC [50%] (mg/ person d)

DWELd (mg/L) for tap water

3.91

100

0.039

3.13

1.56

0.78

1.40

100

0.014

1.12

0.56

0.28

0.99

100

0.010

0.79

0.40

0.20

Non-cancer data used BMDL; combined tumors data used BMD10. RfD or CV = BMDL or BMD10/UF. LADI = RfD or CV mg/kg d  80 kg bw per person to get mg/person d. DWEL = (LADI adjusted by RSC mg/person d)/2 L per d. Combined tumor sites include rat kidney, oral cavity, mammary gland, preputial gland, and clitoral gland and mouse liver, oral cavity, and uterus.

1506

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

The LADI is defined as the oral RfD multiplied by human body weight (80 kg for adults, 10 kg for infants, 20 kg for young children, 30–40 kg for older children; US EPA, 2009b). Daily water consumption from all sources is taken to be 2 L not only for adults but also for an integral of changing body weight from infancy to adulthood; that volume of intake may be adjusted for site-specific circumstances such as climactic conditions. From the RfD and CV, a LADI was calculated by multiplying the specific RfD or CV by the average body weight of an adult, with the presumption of daily exposure for a 75- to 80-year lifespan. For the non-cancer endpoint, the LADI was 3.13 mg/person d based on renal tubule hyperplasia rats. For the combined tumors, the resulting LADIs were 1.12 and 0.79 mg/person d for the 50th percentile and LB 90th percentile, respectively (Table 6). Lifetime DWELs were calculated by multiplying the LADI by the RSC and dividing by 2 L of water consumed daily. The DWEL for TCP considered protective of non-cancer toxicity is 0.78 mg/L (=780 ppb) of drinking water. Likewise, the DWEL for TCP considered protective against tumors is between 0.20 and 0.28 mg/L (=200–280 ppb) of drinking water.

3. Discussion This analysis concludes that tap water can be consumed safely at TCP concentrations of 200 ppb to protect against both non-cancer toxicity and cancer and 780 ppb to protect solely against non-cancer toxicity. Support for these conclusions derives from a critical examination of the available scientific data obtained from laboratory animal studies, a current understanding of mutagenic MoA in experimental carcinogenesis, the application of appropriate stateof-the-art methods for dose–response modeling for humans (viz., BMD and Monte Carlo methods) with provision for the possibility that exposures to TCP may occur via routes other than tap water consumption, and the use of internationally recognized Frameworks to determine the human relevance of data from laboratory animal studies. Given that the cancer DWELs are based on corn oil studies and that corn oil gavage, unlike drinking water exposure, contributes – perhaps extensively – to tumor production, these DWELs may be viewed as conservative. Nonetheless, these findings may prove useful in informing regulatory initiatives at the Federal and State levels such as defining a Maximum Contaminant Level Goal (MCLG) or a chronic Health Advisory; and they may also inform the establishment of private guidelines by purveyors of tap water. The dominant forms of non-cancer toxicity were hyperplasia, not only in tissues where compound-related tumors were later observed but also in others such as bile duct of rats, and necrosis of the liver and kidneys of rats. Among the choices for the key noncancer event in estimating a DWEL for TCP, the kidney hyperplasia stood out as the most relevant for humans (in accordance with the Framework of Boobis et al. (2006)). In particular, TCP’s metabolites were most likely responsible for its toxicity, and the kidneys were the main route of excretion; the cell damage was consistent with a high concentration of the stable reactive metabolites generated in part in the liver and with subsequent regenerative hyperplasia and with adverse changes in kidney weight and kidney:body weigh ratios. Since TCP was known to be metabolized to one or more stable reactive metabolites and some evidence indicated that TCP had to be activated by microsomal enzymes to a proximate toxicant, it is likely that the compound-related toxicity resulted from one or more of TCP’s metabolite(s), particularly in the kidneys and liver where microsomal enzyme activity in rodents is among the highest of the organs and tissues. Support for other organ pathology was less robust. The non-cancer findings in the subchronic studies indicated a significant difference in toxic potency based on the vehicle of

administration. This findings was confirmed by the work of La and Swenberg (1997a) who reported significant increases in endogenously formed DNA adducts were found in liver and forestomach tissue of rats treated by corn oil gavage, whereas such adducts were not found in rats receiving TCP via drinking water. These findings were also suggestive that high local concentrations of TCP from the bolus exposure may deplete or saturate cellular defense mechanisms such as GSH. At the doses tested in rats and mice, TCP produced tumors at multiple sites. The key tumor sites considered relevant for humans were the kidneys, liver, oral cavity, mammary glands, and reproductive tissues (clitoris, preputial glands, and uterus), according to the application of the Framework of Meek et al. (2003) supported by conceptual Framework on MoA for carcinogenesis of Sonich-Mullin et al. (2001). A major determinant of this analysis was the characterization of TCP’s MoA for cancer in rodents as due at least in part to mutagenesis; secondarily, evidence from Swenberg et al. (2008) and others that indicated that (1) DNA adduct formation lead to mutations only when genetic damage was passed on by cell division and (2) the dose–response curves for mutagens had threshold by virtue of reaching background levels of endogenous mutations which were ever present. This information provided the foundation for concluding that TCP carcinogenicity in rodents should be treated as a non-linear event and replace the default linear assumption [the default ‘‘linear dose–response” is assumed to go directly to the zero intersect, rather than to a background region of endogenous mutations]. Supportive findings were reported by Ito et al. (1996) who found in tissues that had TCP-induced tumors, from high dose exposures, the presence of DNA adducts comparable to two metabolites of TCP which were likely generated endogenously from lipid metabolism. Aside from that, DNA adducts were considered indices of exposure rather than effects as described by Swenberg et al. (2008). Note was taken that the TCP-related tumors in rodents were mostly benign and not carcinomas, and that the tumors were found solely at the final 2-year sacrifice and not at the mid-study sacrifice and examination. These observations indicate that the nature of the carcinogenic response is less severe than if the tumors had been all or mostly malignancies found in early to mid-life stages, and that tumor responses were slow in their progression, possibly suggesting a promoting rather than an initiating MoA. It has been postulated that the high mortality observed at the interim sacrifice in the chronic studies was the result of early elicitation (as opposed to end-of-life) of tumors. Data from the Appendices of NTP (1993), however, do not support such a proposition: The number of animals with tumors was relatively small and randomly distributed across sites with an increase in only benign neoplasms of the forestomach, which was most likely caused by the method of administration. The recognition of thresholds and non-linear dose response for mutagenic carcinogens is an important aspect of this analysis, since it replaces a major default assumption. Therefore, determining whether this phenomenon is expressed among other mutagenic carcinogens or is unique to TCP is of value to examine. To that end, data for the following genotoxic carcinogens are presented below:  Findings of non-linearity for genotoxic agents (some of which are potent genotoxicants via alkylation) that are also carcinogens have been reported from numerous studies. Asano et al. (2006) identified three clastogenic compounds (mitomycin C, Ara-C, and colchicine) with a non-linear dose–response curve, providing evidence of a practical mutagenicity threshold for the compounds.  Doak et al. (2007) conducted a study of the significance of low dose exposures employing human lymploblastoid cell which

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510









were treated with alkylating agents (MMS, methylnitrosourea [MNU], EMS, and ENU) that are mutagenic carcinogens. Chromosomal damage and point mutations were quantified using MN test and forward mutation assays. They reported finding that MMS and EMS displayed non-linear curves containing a range of non-mutagenic low doses. The non-linearity could be explained by homeostatic maintenance from faithful DNA repair, which is efficient at low doses of compounds that alkylate predominantly the N7-guanine and rarely attack O atoms. They also observed that MNU and ENU showed linear dose– response relationships; however, the curves ended at a point higher than the zero intersect (i.e., to background levels of mutations, consistent with thresholds). These data support the presence of a practical threshold region for mutagenic carcinogens. These data were corroborated by a later study by Johnson et al. (2009). Favor (1998) examined the mutagenic activity of the genotoxic carcinogen ethylnitrosourea at low doses, and determined that a threshold dose–response properly characterized the resulting data. Additional data on the threshold dose–response of ethylnitrosourea has been reported (Parry et al., 2000; Guttenplan, 1990; Maekawa et al., 1984), along with similar data on other nitroso compounds (Guttenplan, 1990) and dimethylhydrazine (Parry et al., 2000). In an ED001 tumor and biomarker study, rainbow trout were fed low doses of dibenzo[a,l]pyrene for 4 weeks, followed by nine months of exposure to the control diet prior to examination. Using various models, it was shown that the non-linear shape of the DNA adduct and tumor response curves can be very different, demonstrating that DNA adduction cannot be used to accurately predict tumor response for this compound. In addition, the tumor response curve showed increasingly steep slopes with decreasing doses, suggesting the presence of a threshold. The results specific for dibenzo[a,l]pyrene in the ED001 assay suggest the presence of a high degree of conservatism in the US EPA default linear assumption used when assessing genotoxic carcinogens (Bailey et al., 2009). Two recent evaluations have reported thresholds for MN formation in mice orally dosed with acrylamide (AA). Swenberg et al. (2008) evaluated MN data collected from two different laboratories, and demonstrated a threshold for MN formation in the 1–3 mg/kg dose range. This threshold is quite comparable to the one reported by Zeiger et al. (2009) of 1–2 mg/kg in mice orally dosed with AA. Zeiger et al. (2009) found that using internal measures of dose, the glycidamide (GA)-derived DNA adduct, N-7-(2-carbomyl-2-hydroxyethyl)guanine (N7-GAGua) and the hemoglobin adduct from GA, allowed the identification of the threshold. These two studies support the concept that the hemoglobin and DNA adducts produced from AA exposures are biomarkers of exposure and not biomarkers of effect. Furthermore, the clastogenic effects of AA include non-linear, threshold events such as crosslinking chromosomes and/or associated proteins (Carere, 2006). In addition, AA and GA exhibit important differences in the dose–response relationship for mutation frequencies in mouse lymphoma cells, with AA showing an apparent threshold or very small slope (Mei et al., 2008). Although large colony mutants in mouse lymphoma cells (indicative of single-gene mutations) were not significantly increased, the dose–response curve for the small-colony mutants demonstrated a very small slope at low doses, with a point of inflection (i.e., non-linear) above which the dose– response curve becomes increasingly steep (Moore et al., 1987). The repair of DNA damage is a toxicodynamic parameter that may represent the rate-determining step leading to toxic effects (Rozman, 2006). DNA repair would be a likely source of non-linearity in tumor response for AA (Dybing and Sanner, 2003). To

1507

continue the illustration of AA, the predominant DNA adduct identified following exposures to AA is the N7-GA-Gua adduct as described above, and to a much lesser extent, N3-(2-carbomyl-2-hydroxyethyl) adenine, an adduct also derived from GA. Neither of these adducts are promutagenic (Zeiger et al., 2009; Swenberg et al., 2008), but the half-lives of both have been determined in certain rat tissues (Maniere et al., 2005). Once DNA repair rates in humans are available, the impact of these rates could be taken into account in risk assessments of AA, assuming the specific DNA adducts are shown to be critical in the MoA for the effect of interest. If DNA damage is linked to the effects observed for AA, it is expected that humans would repair any DNA damage more efficiently than rodents as humans were reported to repair various types of DNA damage 3–8-fold more efficiently than the rat when species of various lifespans were compared (Cortopassi and Wang, 1996; Wang et al., 1997). Likewise, faithful DNA repair is a likely source of non-linearity in tumor response for TCP. These examples represent a growing body of evidence indicating that genotoxic carcinogens in general can have a non-linear dose–response consistent with thresholds. Therefore, this characteristic dose–response relationship for TCP’s mutagenicity would not be unique, lending support to the proposition that the non-linear (threshold) dose–response analysis for TCP is a justifiable replacement for the default approach. Given the possible interest in comparing the outcome of default linear dose–response assessment for TCP, an illustrative calculation has been performed incorporating the linear default assumption for low dose extrapolation (as identified in US EPA’s Carcinogen Guidelines (US EPA, 2005)). The same data sets as use for the non-linear estimations of DWELs were employed. The resulting estimate of TCP’s cancer potency (CI upper bound average for tumors in males and females) is 0.101 (mg/kg d)1. At an upper bound cancer risk of 1  105, the DWEL is estimated to be 3.96 lg/L (=3.96 ppb) [based on a BW of 80 kg, and water intake of 2 L, with no RSC]. This estimate suggests a considerable overstatement of health risk and an understatement of safety. Limited evidence was found for a non-mutagenic MoA for TCP. It focused largely on the presence of substantial cell damage followed by reparative hyperplasia, as was the case for non-cancer endpoints (described previously). Some cancer sites were judged to not be relevant to humans. They included, based on histopathologic evidence, the pancreas, intestine, and oral mucosa. Based on lack of a human homologue, the following cancer sites were considered not relevant to humans: Harderian and Zymbal glands, and forestomach. Evidence for these are presented in Section 2.3. For the sites that have no human counterpart, evidence from an IARC analysis indicates that for known human carcinogens, no tumors have ever been reported in the animal Harderian gland or forestomach; therefore, their exclusion from the TCP DWEL estimation would not likely understate the risk estimates (Tomatis et al., 1989). In the case of the rodent Zymbal gland, tumors have been reported for 2 of 50 known human carcinogens; however, the Zymbal gland tumors in rodents were unrelated to the actual tumors in humans. Additionally, they would not have been predictive of site-specific tumors in humans because the same rodent studies also reported tumors present at the sites concordant with those in humans (Tomatis et al., 1989). Because the TCP-induced forestomach tumors had among the highest incidence in both species and both sexes, their further consideration has at times been suggested; however, evidence indicates that they are species-specific and not relevant to humans beyond the lack of the same anatomic structure in humans. Indeed, in a review of rodent forestomach tumorigenic MoA with regard to relevance for humans, the recommendation was made that tumors

1508

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

in this tissue ‘‘should not form the basis for carcinogenic classification or quantitative cancer potency estimates for humans” (Proctor et al., 2007). Furthermore, exposures that exceed the MTD for a cancer bioassay should not be considered relevant for human risk assessment (Proctor et al., 2007). The TCP-induced forestomach tumors may likely have arisen as a result of the method of administration (La et al., 1996) producing a long-lasting bolus effect causing tissue damage (it is a storage organ) and reparative hyperplasia leading to tumors from background mutagenesis; the vehicle (corn oil) exacerbated the tissue injury and enhanced tumor formation; and the DNA adducts in forestomach are considerably less than those in the glandular stomach which has a human counterpart (La et al., 1996). In addition, a study of numerous cancer bioassays in which forestomach tumors were observed led to realization that these tumors arise from both mutagenic and non-mutagenic compounds, and that compounds at doses that exceed the maximum tolerated dose are likely to produce tumors by depletion of GSH (Clayson et al., 1990). While it has been postulated that TCP may have caused tumors in the forestomach of rodents, it may have been the result of the stable reactive metabolites produced in the cells in this tissue; however, that appears unlikely since it is low on P-450 activity which would needed to produce TCP’s reactive metabolites. To strengthen the tumor dose–response assessment for TCP, we combined the incidence rates for the remaining tumor types and applied Monte Carlo analysis to obtain estimates of the central tendency (50th percentile) of the BMD10 for male, females, and the combined sexes and of the LB confidence interval (90th percentile) BMD10 for the same data sets. A sensitivity analysis was then performed to identify which tumor data sets contributed most to the distributional values. In taking such steps, the analysis benefitted from the use of a larger amount of tumor data without forcing the selection of a single tumor data set or discarding otherwise useful and defensible information in the estimation of RfD and DWEL. In addition, by generating a distribution of BMD10 values, risk managers are provided a more complete description of the range of possible toxicity values (i.e., lower bound, central tendency, and upper bound), which is consistent with US EPA guidelines (US EPA, 2005). Despite shortcomings in the data, overall, the uncertainty factors are appropriate for non-linear dose–response, and use maximum default values for interspecies and intraspecies extrapolation. Kinetic data and a PBPK model of Volp et al. (1984) were relied extensively to fashion robust interpretations of the descriptive toxicity data; this approach increased appreciably the power of, and confidence in, the conclusions. By contrast, data exploring the nature and extent to which receptor interactions play a role in subchronic and chronic toxicity including carcinogenicity were lacking and might have provided useful insights. Overall, the DWELs described in Table 6 represent a range of sound choices for selecting a MCLG for TCP. These findings are presented in the context of current and proposed regulatory guidelines for TCP. While the US EPA has not established a Maximum Contaminant Level (MCL) for TCP, US EPA drinking water health advisories have been published (10-kg child 1-day and 10-day = 600 ppb, lifetime = 40 ppb; US EPA, 2006). US EPA published a toxicological summary document by the Integrated Risk Information System, and proposed an RfD 0.004 mg/kg d (=4 lg/kg d) based on data from the NTP bioassay and numerous default assumptions (US EPA, 2009a). US EPA’s RfD is approximately one-tenth of the calculated in this monograph. Among the US states, only Hawaii has promulgated a MCL (0.6 ppb) (HDOH, 2005, 2009) representing a theoretical of onein-a-million upper bound risk level in drinking water based on the default linear extrapolation of carcinogenicity data. California

has proposed, using all worst case default assumption, a Public Health Goal (PHG; 0.0007 ppb) (OEHHA, 2009), which is an input into the possible development of a drinking water standard to be promulgated by the California Department of Health. Based on the rodent data and applying its classification definitions, US EPA (2009a) concluded that TCP is a ‘‘likely to be carcinogenic to humans”, while acknowledging the difficulties posed by the high mortality animals treated with TCP at the highest doses. Also, IARC classified TCP as ‘‘probably carcinogenic to humans” Category 2A (WHO, 1995). NTP has concluded that TCP is ‘‘reasonably anticipated to be a human carcinogen” (NTP, 2005). Conflict of Interest The authors declare that there are no conflicts of interest. Acknowledgments The authors gratefully acknowledge the financial support provided for this work by The Dow Chemical Company and Shell Oil Company. The authors thank David Gaylor for his assistance. References Albert, D.M., Frayer, W.C., Black, H.E., Massicotte, S.J., Sang, D.N., Soque, J., 1986. The Harderian gland: its tumors and its relevance to humans. Trans. Am. Ophthalmol. Soc. 84, 321–341. American Conference of Industrial Hygienists (ACGIH), 1996. 1,2,3Trichloropropane. In: Documentation of the Threshold. Limit Values and Biological Exposure Indices, sixth ed. Cincinnati, Ohio. Agency for Toxic Substances and Disease Registry (ATSDR), 1992. Toxicological Profile for 1,2,3-Trichloropropane. US Department of Health and Human Services. Asano, N., Torous, D.K., Tometsko, C.R., Dertinger, S.D., Morita, T., Hayashi, M., 2006. Practical threshold for micronucleated reticulocyte induction observed for low doses of mitomycin C, Ara-C and colchicine. Mutagenesis 21, 15–20. Bailey, G.S., Reddy, A.P., Pereira, C.B., Harttig, U., Baird, W., Spitsbergen, J.M., Hendricks, J.D., Orner, G.A., Williams, D.E., Swenberg, J.A., 2009. Nonlinear cancer response at ultralow dose: a 40800-animal ED001 tumor and biomarker study. Chem. Res. Toxicol. 22, 1264–1276. Bikales, N.M., 1969. Polysulfide polymers. Encyclopedia of Polymer Science and Technology, vol. 11. Interscience, New York, p. 425. Boobis, A.R., Cohen, S.M., Dellarco, V., McGregor, D., Meek, M.E., Vickers, C., Willcocks, D., Farland, W., 2006. IPCS framework for analyzing the relevance of a cancer mode of action for humans. Crit. Rev. Toxicol. 36, 781–792. Boobis, A.R., Doe, J.E., Heinrich-Hirsch, B., Meek, M.E., Munn, S., Ruchirawat, M., Schlatter, J., Seed, J., Vickers, C., 2008. IPCS framework for analyzing the relevance of a noncancer mode of action for humans. Crit. Rev. Toxicol. 38, 87– 96. Boysen, G., Pachkowski, B.F., Nakamura, J., Swenberg, J.A., 2009. The formation and biological significance of N7-guanine adducts. Mutat. Res. 678, 76–94. California Department of Public Health (CDPH), 2007. 1,2,3-Trichloropropane. State of California. . Carere, A., 2006. Genotoxicity and carcinogenicity of acrylamide: a critical review. Ann. Ist. Super. Sanita 42, 144–155. Clayson, D.B., Iverson, F., Nera, E.A., Lok, E., 1990. The significance of induced forestomach tumors. Ann. Rev. Toxicol. 30, 441–463. Clewell, H.J.I.I.I., Andersen, M.E., Barton, H.A., 2002. A consistent approach for the application of pharmacokinetic modeling in cancer and noncancer risk assessment. Environ. Health Perspect. 110, 85–93. Cortopassi, G.A., Wang, E.A., 1996. There is substantial agreement among interspecies estimates of DNA repair activity. Mech. Ageing Dev. 91, 211–218. Doak, S.H., Jenkins, G.J., Johnson, G.E., Quick, E., Parry, E.M., Parry, J.M., 2007. Mechanistic influences for mutation induction curves after exposure to DNAreactive carcinogens. Cancer Res. 67, 3904–3911. Dutch Expert Committee on Occupational Standards (DECOS), 1998. 1,2,3Trichloropropane: Evaluation of the Carcinogenicity and Genotoxicity. Health Council of the Netherlands. Dybing, E., Sanner, T., 2003. Risk assessment of acrylamide in foods. Toxicol. Sci. 75, 7–15. Eustis, S.L., Boorman, G.A., 1985. Proliferative lesions of the exocrine pancreas: relationship to corn oil gavage in the National Toxicology Program. J. Natl. Cancer Inst. 75, 1067–1073. Eustis, S.L., Boorman, G.A., 1990. Exocrine pancreas. In: Boorman, G.A., Eustis, S.L., Elwell, M.R. (Eds.), Pathology of the Fischer Rat: Reference and Atlas. Academic Press, New York, pp. 95–108.

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510 Favor, J., 1998. The mutagenic activity of ethylnitrosourea at low doses in spermatogonia of the mouse as assessed by the specific-locus test. Mutat. Res. 405, 221–226. Figge, F.J.H., Strong, L.C., Shanbrom, A., 1942. Fluorescent porphyrins in the Harderian glands and susceptibility to spontaneous mammary carcinoma in mice. Cancer Res. 2, 335–342. Friedberg, E.C., 2003. DNA damage and repair. Nature 421, 436–440. Friedberg, E.C., McDaniel, L.D., Schultz, R.A., 2004. The role of endogenous and exogenous DNA damage and mutagenesis. Curr. Opin. Genet. Dev. 14, 5–10. Furihata, C., Ohta, H., Katsuyama, T., 1996. Cause and effect between concentrationdependent tissue damage and temporary cell proliferation in rat stomach mucosa by NaCl, a stomach tumor promoter. Carcinogenesis 17, 401–406. Guttenplan, J.B., 1990. Mutagenesis by N-nitroso compounds: relationships to DNA adducts, DNA repair, and mutational efficiencies. Mutat. Res. 233, 177–187. Hansen, J.F., Ross, P.E., Makovec, G.T., Eustis, S.L., Sigler, R.E. 1995. Proliferative and other selected lesions of the exocrine pancreas in rats, GI-6. In: Society of Toxicologic Pathologists (Ed.), Guides for Toxicologic Pathology. STP/ARP/AFIP, Washington, DC. Hawaii Department of Health (HDOH), 2005. Trace Amounts of Organic Chemicals Found in One Maui Water Well. State of Hawaii. Hawaii Department of Health (HDOH), 2009. Contaminants Regulated by the Safe Drinking Water Branch (Effective 3/4/09). State of Hawaii. Hazardous Substance Data Bank (HSDB), 2009. 1,2,3-Trichloropropane. National Library of Medicine (accessed 09.03.09). Health Canada, 1994. Human Health Risk Assessment for Priority Substances. Canadian Environmental Protection Act, Ontario. Irwin, R.D., Haseman, J.K., Eustis, S.L., 1995. 1,2,3-Trichloropropane: a multisite carcinogen in rats and mice. Fundam. Appl. Toxicol. 25, 241–252. Ito, N., Fukushima, S., Imaida, K., Sakata, T., Masui, T., 1983. Induction of papilloma in the forestomach of hamsters by butylated hydroxyanisole. Gann 74, 459–461. Ito, N., La, D.K., Holt, S., Craft, T.F., Sills, R.C., Swenberg, J.A., 1996. Analysis of ras mutations in forestomach tumors from B6C3F1 mice exposed to 1,2,3trichloropropane. Proc. Am. Assoc. Cancer Res. 37, 137. Jenkins, G.J.S., Doak, S.H., Johnson, G.E., Quick, E., Waters, E.M., Parry, J.M., 2005. Do does response thresholds exist for genotoxic alkylating agents? Mutagenesis 20, 389–398. Ji, B., Bi, Y., Simeone, D., Mortensen, R.M., Logsdon, C.D., 2001. Human pancreatic acinar cells lack functional responses to cholecystokinin and gastrin. Gastroenterology 121, 1380–1390. Johnson, G.E., Doak, S.H., Griffiths, S.M., Quick, E.L., Skibinski, D.O., Zair, Z.M., Jenkins, G.J., 2009. Non-linear dose–response of DNA-reactive genotoxins: recommendations for data analysis. Mutat. Res. 678, 95–100. Junqueira, V.B., Koch, O.R., Arisi, A.C., Fuzaro, A.P., Azzalis, L.A., Barros, S.B., Cravero, A., Farre, S., Videla, L.A., 1997. Regression of morphological alterations and oxidative stress-related parameters after acute lindane-induced hepatotoxicity in rats. Toxicology 117, 199–205. Kawata, E., 1992. Personal Communication with Honolulu Board of Water Supply. Kirman, C.R., Sweeney, L.M., Meek, M.E., Gargas, M.L., 2003. Assessing the dosedependency of allometric scaling performance using physiologically based pharmacokinetic modeling. Regul. Toxicol. Pharmacol. 38, 345–367. Kjellstrand, P., Holmquist, B., Kanje, M., Alm, P., Romare, S., Jonsson, I., Mansson, L., Bjerkemo, M., 1984. Perchloroethylene: effect on body and organ weights and plasma butyrylcholinesterase activity in mice. Acta Pharmacol. Toxicol. 54, 414–424. Kroes, R., Wester, P.W., 1986. Forestomach carcinogens: possible mechanisms of action. Food Chem. Toxicol. 24, 1083–1089. La, D.K., Yen, T.Y., Swenberg, J., 1994. Characterization of DNA adducts induced by 1,2,3-trichloropropane. In: Proceedings of the American Association for Cancer Research Annual Meeting 35, 108. 85th Annual Meeting of the American Association for Cancer Research, San Francisco, CA, April 10–13. La, D., Lilly, P., Anderegg, R., Swenberg, J., 1995. DNA adduct formation in B6C3F1 mice and Fischer-344 rats exposed to 1,2,3-trichloropropane. Carcinogenesis (Oxford) 16, 1419–1424. La, D.K., Schoonhoven, R., Ito, N., Swenberg, J.A., 1996. The effects of exposure route on DNA adduct formation and cellular proliferation by 1,2,3-trichloropropane. Toxicol. Appl. Pharmacol. 140, 108–114. La, D.K., Swenberg, J.A., 1997a. The induction of endogenous DNA adducts in tissues of Fischer-344 rats following gavage administration of 1,2,3-trichloropropane. In: Proceedings of the American Association for Cancer Research Annual Meeting 38, 131. 88th Annual Meeting of the American Association for Cancer Research, San Diego, CA, April 12–16. La, D.K., Swenberg, J.A., 1997b. Carcinogenic alkylating agents. In: Sipes, I.G., McQueen, C.A., Gandolfi, A.J. (Eds.), Chemical Carcinogens and Anticarcinogens. Elsevier Science, Oxford, pp. 111–140. Longnecker, D.S., 1987. Interface between adaptive and neoplastic growth in the pancreas. Gut 28, 253–258. Longnecker, D.S., Millar, P.M., 1990. Tumours of the pancreas. In: Turusov, V.S., Mohr, U. (Eds.), Pathology of Tumours in Laboratory Animals, second ed., Tumours of the Rat, vol. I International Agency for Research on Cancer, Lyon, pp. 241–249. Lutz, W.K., 1979. In vivo covalent binding of organic chemicals to DNA as a quantitative indicator in the process of chemical carcinogenesis. Mutat. Res. 65, 289–356. Lykins Jr., B.W., Baier, J.H., 1985. Removal of Agricultural Contaminants from Ground Water, EPA/600/D-85/136. Environmental Protection Agency, Washington, DC.

1509

Maekawa, A., Ogiu, T., Matsuoka, C., Onodera, H., Furuta, K., Kurokawa, Y., Takahashi, M., Kokubo, T., Tanigawa, H., Hayashi, Y., 1984. Carcinogenicity of low doses of N-ethyl-N-nitrosourea in F344 rats; a dose–response study. Gann 75, 117–125. Mahmood, N.A., Burka, L.T., Cunningham, M.L., 1988. Metabolism and mutagenicity of 1,2,3-trichloropropane. Pharmacologist 30, A8. Mahmood, N.A., Overstreet, D., Burka, L.T., 1991. Comparative disposition and metabolism of 1,2,3-trichloropropane in rats and mice. Drug Metab. Dispos. 19, 411–418. Maniere, I., Godard, T., Doerge, D.R., Churchwell, M.I., Guffroy, M., Laurentie, M., Poul, J.M., 2005. DNA damage and DNA adduct formation in rat tissues following oral administration of acrylamide. Mutat. Res. 580, 119–129. Meek, M.E., Bucher, J.R., Cohen, S.M., Dellarco, V., Hill, R.N., Lehman-McKeeman, L.D., Longfellow, D.G., Pastoor, T., Seed, J., Patton, D.E., 2003. A framework for human relevance analysis of information on carcinogenic modes of action. Crit. Rev. Toxicol. 33, 591–653. Mei, N., Hu, J., Churchwell, M.I., Guo, L., Moore, M.M., Doerge, D.R., Chen, T., 2008. Genotoxic effects of acrylamide and glycidamide in mouse lymphoma cells. Food Chem. Toxicol. 46, 628–636. Merrick, B.A., Smallwood, C.L., Meier, J.R., McKean, D.L., Kaylor, W.H., Condie, L.W., 1987. Chemical reactivity, cytotoxicity, and mutagenicity of chloropropanones. Toxicol. Appl. Pharmacol. 91, 46–54. Merrick, B.A., Robinson, M., Condi, L.W., 1991. Cardiopathic effect of 1,2,3trichloropropane after subacute and subchronic exposure in rats. J. Appl. Toxicol. 11, 179–187. Moore, M.M., Amtower, A., Doerr, C., Brock, K.H., Dearfield, K.L., 1987. Mutagenicity and clastogenicity of acrylamide in L5178Y mouse lymphoma cells. Environ. Mutagen. 9, 261–267. Nagaya, T., Ishikawa, N., Hata, H., Otobe, T., 1993. Subclinical and reversible hepatic effects of occupational exposure to trichloroethylene. Int. Arch. Occup. Environ. Health 64, 561–563. National Institute for Occupational Safety and Health (NIOSH), 1981. Trichloropropanes, NTIS PB83-112870. National Research Council (NRC), 1983. Risk Assessment in the Federal Government: Managing the Process. National Academy Press, Washington, DC. National Research Council (NRC), 1994. Science and Judgement in Risk Assessment. National Academy Press, Washington, DC. National Toxicology Program (NTP), 1990. 1,2,3-Trichloropropane Reproduction and Fertility Assessment in Swiss CD-1 Mice when Administered Via Gavage, NTP 90-209. US Department of Health and Human Services, National Toxicology Program, Research Triangle Park, NC. National Toxicology Program (NTP), 1999. NTP Report on Carcinogens background document for 1,2,3-trichloropropane, Final. Research Triangle Park. National Toxicology Program (NTP), 1993. Toxicology and Carcinogenesis Studies of 1,2,3-Trichloropropane (CAS No. 96-18-4) in F344/N Rats and B6C3F1 Mice (Gavage Studies), NTP TR 384. US Department of Health and Human Services, National Toxicology Program, Research Triangle Park, NC. National Toxicology Program (NTP), 1994. Comparative Toxicology Studies of Corn Oil, Safflower Oil, and Tricaprylin (CAS Nos. 8001-30-7, 8001-23-8, and 538-238) in Male F344/N Rats as Vehicles for Gavage, NTP TR 426. National Toxicology Program, Research Triangle Park, NC. National Toxicology Program (NTP), 2005. 1,2,3-Trichloropropane. In: Report on Carcinogens, 11th ed. US Department of Health and Human Services. Office of Environmental and Health Hazard Assessment (OEHHA), 2009. Public Health Goal for 1,2,3-Trichloropropane in Drinking Water. Final. California Environmental Protection Agency. Owyang, C., 1996. Physiological mechanisms of cholecystokinin action on pancreatic secretion. Am. J. Physiol. 271, G1–G7. Paine, M.F., Hart, H.L., Ludington, S.S., Haining, R.L., Rettie, A.E., Zeldin, D.C., 2006. The human intestinal cytochrome P450 ‘‘pie”. Drug Metab. Dispos. 34, 880– 886. Parry, J.M., Jenkins, G.J., Haddad, F., Bourner, R., Parry, E.M., 2000. In vitro and in vivo extrapolations of genotoxin exposures: consideration of factors which influence dose–response thresholds. Mutat. Res. 464, 53–63. Proctor, D.M., Gatto, N.M., Hong, S.J., Allamneni, A.P., 2007. Mode-of-action framework for evaluating the relevance of rodent forestomach tumors in cancer risk assessment. Toxicol. Sci. 98, 313–326. Ratpan, F., Plaumann, H., 1985. Mutagenicity of halogenated three-carbon compounds and their methylated derivatives. Environ. Mutag. 7 (Suppl 3), 15. Rozman, K., 2006. Influence of dynamics, kinetics, and exposure on toxicity in the lung. In: Gardner, D.E. (Ed.), Toxicology of the Lung, fourth ed. Taylor and Francis, Boca Raton, pp. 195–230. Saito-Suzuki, R., Teramoto, S., Shirasu, Y., 1982. Dominant lethal studies in rats with 1,2-dibromo-3-chloropropane and its structurally related compounds. Mutat. Res. 101, 321–327. Sheldon, W., 1994. Tumours of the Harderian Gland, vol. 111. IARC Sci. Publ.. pp. 101–113. Sonich-Mullin, C., Fielder, R., Wiltse, J., Baetcke, K., Dempsey, J., Fenner-Crisp, P., Grant, D., Hartley, M., Knaap, A., Kroese, D., Mangelsdorf, I., Meek, E., Rice, J.M., Younes, M., 2001. IPCS conceptual framework for evaluating a mode of action for chemical carcinogenesis. Regul. Toxicol. Pharmacol. 34, 146–152. Swenberg, J.A., Fryar-Tita, E., Jeong, Y.C., Boysen, G., Starr, T., Walker, V.E., Albertini, R.J., 2008. Biomarkers in toxicology and risk assessment: informing critical dose–response relationships. Chem. Res. Toxicol. 21, 253–265. Tafazoli, M., Kirsch-Volders, M., 1996. In vitro mutagenicity and genotoxicity study of 1,2-dichloroethylene, 1,1,2-trichloroethane, 1,3-dichloropropane, 1,2,3trichloropropane and 1,1,3-trichloropropene, using the micronucleus test and

1510

R.G. Tardiff, M.L. Carson / Food and Chemical Toxicology 48 (2010) 1488–1510

the alkaline single cell gel electrophoresis technique (comet assay) in human lymphocytes. Mutat. Res. 371, 185–202. Tardiff, R.G., Gargas, M.G., Kirman, C.R., Carson, M.L., Sweeney, L.M., 2010. Estimation of safe dietary levels of acrylamide for humans. Food Chem. Toxicol. 48, 658–667. Tomatis, L., Aitio, A., Wilbourn, J., Shuker, L., 1989. Human carcinogens so far identified. Jpn. J. Cancer Res. 80, 795–807. Torkelson, T.R., Oyen, F., Rowe, V.K., 1976. The toxicity of chloroform as determined by single and repeated exposure of laboratory animals. Am. Ind. Hyg. Assoc. J. 37, 697–705. US Environmental Protection Agency (US EPA), 1994. Methods for Derivation of Inhalation Reference Concentrations and Application of Inhalation Dosimetry, EPA/600/8-90/066F. Office of Health and Environmental Assessment, Environmental Criteria and Assessment Office, Research Triangle Park, NC. US Environmental Protection Agency (US EPA), 1995. The Use of Benchmark Dose Approach in Health Risk Assessment, EPA/630/R-94/007. Risk Assessment Forum. United States Environmental Protection Agency (US EPA), 2002. A Review of the Reference Dose and Reference Concentration Process, EPA/630/P-02/002F. Final Report. United States Environmental Protection Agency (US EPA), 2005. Guidelines for Carcinogen Risk Assessment, EPA/630/P-03/001B. United States Environmental Protection Agency (US EPA), 2006. 2006 Edition of the Drinking Water Standards and Health Advisories, EPA-822-R-06-013. United States Environmental Protection Agency (US EPA), 2009a. Toxicological Review of 1,2,3-Trichloropropane: In Support of Summary Information on the Integrated Risk Information System (IRIS), EPA/635/R-08/010F. Washington, DC. United States Environmental Protection Agency (US EPA), 2009b. Exposure Factors Handbook. 2009 Update. External Review Draft, EPA/600/R-09/052a. Washington, DC.

United States 104th Congress, 1996. 1996 Amendments to the Safe Drinking Water Act – Public Law 104–182. Villeneuve, D.C., Chu, I., Secours, V.E., Cote, M.G., Plaa, G.L., Valli, V.E., 1985. Results of a 90-day toxicity study on 1,2,3- and 1,1,2-trichloropropane administered via the drinking water. Sci. Total Environ. 47, 421–426. Volp, R.F., Sipes, I.G., Falcoz, C., Carter, D.E., Gross, J.F., 1984. Disposition of 1,2,3trichloropropane in the Fischer 344 rat: conventional and physiological pharmacokinetics. Toxicol. Appl. Pharmacol. 75, 8–17. Wang, E., Wong, A., Cortopassi, G.A., 1997. The rate of mitochondrial mutagenesis is faster in mice than humans. Mutat. Res. 377, 157–166. Weber, G.L., Sipes, I.G., 1990. Covalent interactions of 1,2,3-trichloropropane with hepatic macromolecules: studies in the male F-344 rat. Toxicol. Appl. Pharmacol. 104, 395–402. Weber, G.L., Sipes, I.G., 1991. Rat hepatic DNA damage induced by 1,2,3trichloropropane. Adv. Exp. Med. Biol. 283, 853–855. Weber, G.L., Sipes, I.G., 1992. In vitro metabolism and bioactivation of 1,2,3trichloropropane. Toxicol. Appl. Pharmacol. 113, 152–158. Wester, P.W., Kroes, R., 1988. Forestomach carcinogens: pathology and relevance to man. Toxicol. Pathol. 16, 165–170. World Health Organization (WHO), 1995. 1,2,3-Trichloropropane. IARC Monogr. Eval. Carcinog. Risk Chem. Hum. 63, 223–244. World Health Organization (WHO), 2003. 1,2,3 Trichloropropane. Concise International Chemical Assessment Document. World Health Organization (WHO), 2006. Guidelines for Drinking Water Quality, third ed. Geneva. World Health Organization (WHO), 2009. Principles for Modelling Dose–Response for the Risk Assessment of Chemicals. Environmental Health Criteria 239. Zeiger, E., Recio, L., Fennell, T., Haseman, J.K., Snyder, R.W., Friedman, M., 2009. Investigation of the low-dose response in the in vivo induction of micronuclei and adducts by acrylamide. Toxicol. Sci. 107, 247–257.