Chloroform Toxicology and Liver and Kidney Toxicity

RoseyFields evaluates the toxicology of chloroform using the following evidence base drawn from the primary peer-reviewed toxicological literature.

1. Chemical Identity and Physicochemical Properties Relevant to Toxicity

Chloroform (trichloromethane, CHCl3; CAS No. 67-66-3) is a small, dense, volatile chlorinated one-carbon compound with a molecular weight of 119.37 g/mol, a boiling point of 61.12°C, and a vapor pressure of approximately 159–197 mm Hg at 20–25°C (Boublik et al. 1984). Its water solubility (7,950 mg/L at 25°C) is high relative to its lipophilicity (log Kow 1.97), which together with a moderate Henry’s law constant (3.00–3.67×10⁻³ atm-m³/mol) governs its environmental partitioning and, ultimately, the routes by which humans and animals are exposed (Gossett 1987; Nicholson et al. 1984). Chloroform is not flammable under ordinary conditions (Deshon 1979).

These properties are directly relevant to systemic toxicity. Because chloroform is small, uncharged, and lipophilic, it diffuses readily across biological membranes, distributes into lipid-rich tissue, and volatilizes rapidly from surface water and moist tissue surfaces alike; the same volatilization tendency that limits its environmental persistence in surface water (calculated half-life on the order of hours) also underlies its rapid pulmonary uptake and excretion in mammals (Dilling 1977; Dilling et al. 1975). In the atmosphere, chloroform reacts slowly with hydroxyl radicals, giving an atmospheric half-life on the order of months, sufficient for long-range transport but not for substantial bioaccumulation (Class and Ballschmidter 1986; Hubrich and Stuhl 1980). Hydrolysis in water is negligible at environmentally relevant pH (half-lives of decades to millennia) (Jeffers et al. 1989). Chloroform has low soil sorption (log Koc 1.5–2.4), so it leaches readily to groundwater and is relatively persistent there under oxygenated conditions (Sabljic 1984). It does not meaningfully bioconcentrate in aquatic organisms — measured bioconcentration factors of 6–13 in fish (Barrows et al. 1980; Veith et al. 1980) and 690 in green algae (Mailhot 1987) — and no evidence of biomagnification has been reported. Under anaerobic conditions typical of some wastewater and subsurface environments, chloroform is biodegradable, with methanogenic and other anaerobic bacteria capable of dehalogenating it to dichloromethane and ultimately carbon dioxide, although chloroform itself is toxic to these same microbial populations at concentrations above roughly 1–100 mg/L (Rhee and Speece 1992). None of these fate properties indicate substantial environmental persistence or bioaccumulation, but the volatility and lipophilicity of the parent compound are central to understanding both its rapid absorption/elimination kinetics and its distribution to lipid-rich, well-perfused organs (brain, liver, kidney, adipose tissue) described below.

2. Toxicokinetics: Absorption, Distribution, Metabolism, and Excretion

Absorption. Chloroform is well absorbed by all three exposure routes. Following inhalation, uptake is governed by the blood/air partition coefficient, which is lower in humans (7–11) than in rodents (15–21), predicting proportionally greater pulmonary retention in rodents at a given air concentration (Corley et al. 1990). In human volunteers exposed to chloroform released from shower water, a one-compartment pharmacokinetic model estimated roughly 71% (range 40–80%) systemic absorption of the inhaled dose (Xu and Weisel 2005); studies of swimmers and pool attendants confirm that blood chloroform concentrations rise in proportion to airborne and waterborne exposure during swimming (Aggazzotti et al. 1993, 1995; Cammann and Hübner 1995; Lévesque et al. 1994). Oral absorption is essentially complete (close to 100%) in both humans and animals, with peak blood levels reached within about one hour (Fry et al. 1972; Taylor et al. 1974). Notably, the extent and rate of oral absorption depends heavily on vehicle: aqueous administration produces faster, more complete gastrointestinal uptake than administration in an oily vehicle, because a lipophilic solute delivered from an aqueous phase partitions into gut mucosal lipid more readily than one already dissolved in oil (Withey et al. 1983; Pereira 1994). This vehicle effect proves critical for interpreting the animal toxicology literature, since bolus gavage dosing (particularly in oil) produces a markedly different internal dose-time profile, and greater toxicity, than the same total dose delivered gradually in drinking water. Dermal absorption also occurs and is temperature- and vehicle-dependent; in human forearm studies, roughly 1.6–7.8% of an applied dermal dose was systemically absorbed within 8 hours depending on whether the vehicle was ethanol or water (Dick et al. 1995), and studies in hairless rats confirm rapid dermal uptake with evaporative loss from the skin surface having a half-time of only 2–3 minutes (Islam et al. 1996, 1999a, 1999b).

Distribution. Because chloroform is lipophilic, it distributes preferentially to adipose tissue, followed by brain, liver, and kidney, in that order (Corley et al. 1990). Tissue/blood and tissue/air partition coefficients are 20- to 40-fold higher in fat than in lean tissue and vary with species, age, and body composition (Corley et al. 1990). Autopsy data from fatal human exposures corroborate this pattern, with the highest tissue concentrations generally found in fat and brain (Nashelsky et al. 1995). In mice, chloroform readily crosses the placenta, with early accumulation in embryonic neural tissue during organogenesis and later accumulation in fetal respiratory epithelium as metabolic capacity develops (Danielsson et al. 1986).

Metabolism. Chloroform metabolism is well characterized and is central to its toxicity (see Mechanisms, below). The dominant pathway is oxidative dehydrochlorination by cytochrome P450 2E1 (CYP2E1) in the liver and kidney, yielding the highly reactive intermediate phosgene; CYP2A6 (in humans) and CYP2B1/2 (in rats) contribute a minor, lower-affinity component of oxidative metabolism (Constan et al. 1999; Testai et al. 1996). Under anaerobic conditions, a minor reductive pathway generates a dichloromethyl radical and, ultimately, carbon monoxide (Pohl et al. 1981). Phosgene formation via this route was directly demonstrated using deuterated chloroform, which was one-half to one-third as cytotoxic as native chloroform and converted to phosgene much more slowly, confirming that metabolism at the carbon-hydrogen bond is required for toxicity (Branchflower et al. 1984). Phosgene reacts with glutathione (GSH) to form a detoxification product, or, when GSH is depleted, covalently binds cellular proteins and lipids, an event mechanistically linked to hepatocellular necrosis (Docks and Krishna 1976). A marked, well-documented sex difference in mice reflects testosterone-dependent CYP2E1 expression: male mice metabolize and accumulate chloroform in the renal cortex to a greater extent than females, correlating with the substantially greater susceptibility of male mice to chloroform-induced renal necrosis and renal tumors, a difference first characterized in strain C3H mice by Deringer et al. (1953) and later linked mechanistically to renal CYP2E1 activity (Taylor et al. 1974; Culliford and Hewitt 1957; Ade et al. 1994). Route of administration also modulates hepatic first-pass metabolism: in rats, roughly 38% of an oral dose undergoes first-pass hepatic metabolism, compared with substantially less following inhalation, so oral (particularly gavage) dosing concentrates metabolic activation — and therefore toxicity — in the liver to a greater degree than an equivalent inhaled dose (Chiou 1975; Wang et al. 1994). There are also substantial species differences: mice convert roughly 80% of an oral dose to CO2, versus about 66% in rats and only 18% in squirrel monkeys, with corresponding differences in the proportion of dose exhaled unchanged (Brown et al. 1974); overall metabolic clearance is calculated to be slower in humans than in rodents (Taylor et al. 1974; Corley et al. 1990). These species and route differences are essential to interpreting cross-species extrapolation of chloroform toxicity data (see below).

Excretion. The major elimination route is pulmonary, either as unmetabolized parent compound or as metabolically generated CO2, with minor contributions from urine and feces (Fry et al. 1972; Mink et al. 1986). In humans, pulmonary elimination half-times of roughly 7.9 hours (breath) and 9–39 minutes (blood, depending on compartmental model) have been reported following controlled exposures (Xu and Weisel 2005).

3. Mechanisms of Toxicity

The mechanistic literature converges on a common framework across target organs: chloroform itself is relatively inert, but its cytochrome P450 (principally CYP2E1)-mediated oxidative metabolite, phosgene, is directly cytotoxic. In the liver, phosgene covalently binds macromolecules once glutathione stores are depleted, and hepatotoxicity is abolished in CYP2E1-knockout mice and in liver-specific cytochrome P450 reductase-null mice, confirming that metabolism — not parent chloroform — is required for injury (Constan et al. 1999; Fang et al. 2008); in vitro, pretreatment with the cytochrome P450 inhibitor 1-phenylimidazole similarly prevents chloroform-induced cytotoxicity in isolated rat and mouse hepatocytes (Ammann et al. 1998). In vitro, chloroform exposure produces a biphasic cellular injury: an initial “metabolic phase” of GSH depletion followed by an “oxidative phase” characterized by oxidative stress, mitochondrial permeability transition, and protein nitration (Burke et al. 2007; Hartig et al. 2005; Abbassi et al. 2010). A comparable pathway operates in the kidney, where transgenic mouse studies localize the critical cytochrome P450 activity to the renal proximal tubule itself (Liu et al. 2013), and where sex-hormone-dependent CYP2E1 expression explains the pronounced male-predominant renal susceptibility (Weir et al. 2005; Culliford and Hewitt 1957). Downstream renal mechanisms may include impaired gap-junction intercellular communication (Mally and Chipman 2002) and increased intracellular calcium signaling (Jan et al. 2000). Both the liver and kidney retain a capacity for regenerative repair after sublethal chloroform-induced cytotoxicity, a capacity that can become overwhelmed at high doses (Anand and Mehendale 2004), and repeated low-dose exposure can induce an adaptive, protective tolerance to what would otherwise be a lethal challenge dose, underscoring that cytotoxicity — rather than any direct genotoxic mechanism — drives the dose-response relationship (Anand et al. 2004, 2006; Mehendale 2005; Philip et al. 2006).

This cytotoxicity-driven mechanism is also the accepted mode of action for chloroform carcinogenicity, as reviewed by Boobis (2009) and Borgert et al. (2015): oxidative metabolism to phosgene in metabolically competent target tissue (liver, kidney) produces repeated, sustained cytotoxicity; this in turn drives compensatory regenerative cell proliferation; increased cell division elevates the probability of spontaneous mutation and/or clonal expansion of initiated cells; and hepatic or renal tumors ultimately result. This sequence is considered more consistent with the overall database than a direct DNA-reactive mechanism, since chloroform is, at most, a weak mutagen (see Genotoxicity, below), and is consistent with a non-linear, threshold-like carcinogenic dose-response once cytotoxic exposure levels are exceeded (Meek et al. 2002, 2003).

Outside the liver and kidney, chloroform’s classic acute effect — central nervous system depression, historically exploited in its use as a surgical anesthetic — is generally attributed to the compound acting directly as a lipophilic membrane perturbant, altering the function of membrane ion channels and neurotransmitter receptors rather than requiring bioactivation (Harris and Groh 1985; Jenkins et al. 2001). Chloroform directly binds GABA-A receptors, prolonging synaptic inhibition (Jenkins et al. 2001), and blocks voltage-gated potassium and cardiac ion channels (including hERG), providing a mechanistic basis for both CNS depression and the cardiac arrhythmias observed at high exposure (Doring 1975; Scholz et al. 2006).

4. Animal Toxicological Evidence by Organ System

Hepatic. The liver is the most consistently and heavily documented target of chloroform toxicity in laboratory animals. Rats and mice exposed by inhalation show concentration- and duration-dependent hepatocellular vacuolation, swelling, necrosis, and compensatory proliferation, beginning at levels as low as 10 ppm (acute, mice) and 100 ppm (acute, rats), with mice generally more sensitive than rats (Larson et al. 1994c). Under intermediate-duration inhalation, lowest observed lesion levels were 17 ppm in mice and 25 ppm in rats (Templin et al. 1998). Oral gavage dosing produces hepatic injury at markedly lower doses than equivalent total exposure delivered via drinking water — hepatotoxicity occurred at gavage doses ≥34 mg/kg/day in both rats and mice, versus inconsistent effects at much higher cumulative drinking-water doses — a difference attributed to saturation of GSH-dependent detoxification following bolus dosing (Larson et al. 1994b, 1995a). Chronic gavage studies in B6C3F1 mice and Osborne-Mendel rats produced dose-related hepatocellular necrosis and nodular hyperplasia (Dunnick and Melnick 1993).

Renal. The kidney, particularly the proximal tubule, is the second major target organ, with a striking male-predominant sensitivity in mice attributable to testosterone-driven CYP2E1 expression (Taylor et al. 1974; Weir et al. 2005; Culliford and Hewitt 1957). Renal tubular necrosis and regenerative proliferation occur at inhalation and oral doses overlapping those producing hepatic injury, and, as with the liver, gavage dosing produces greater nephrotoxicity than equivalent drinking-water dosing.

Cardiovascular. No inhalation studies specifically evaluated blood pressure or heart rate in animals, and no exposure-related changes in heart weight or histology were seen at inhalation concentrations up to 300 ppm (rats) or 88 ppm (mice) (Templin et al. 1996b; Larson et al. 1996). By the oral route, a single gavage dose of 149 mg/kg or repeated doses of 37 mg/kg/day for four weeks produced decreased heart rate, increased blood pressure, and prolonged atrioventricular conduction in rats (Müller et al. 1997). Mechanistic in vitro work shows that chloroform damages the cardiac transverse tubular system, produces bradycardia and ventricular fibrillation in isolated hearts, and blocks cardiac potassium channels, including hERG (Doring 1975; Zhou et al. 2011; Scholz et al. 2006).

Immunological. Inhalation exposure to chloroform (10.6 ppm, repeated) increased susceptibility to bacterial (Streptococcus zooepidemicus) lung infection in mice, with impaired alveolar macrophage phagocytic activity and bacterial clearance at 100–1,000 ppm (Aranyi et al. 1986; Selgrade and Gilmour 2010). Low-level inhalation (7 ppm) produced pulmonary inflammatory changes (increased leukocytes and macrophages in bronchoalveolar lavage) in mice (de Oliveira et al. 2015). Oral exposure suppressed humoral immunity (decreased IgM response to sheep red blood cells) in mice at gavage doses ≥50 mg/kg/day, although a comprehensive immunotoxicity battery found no effects at lower drinking-water doses (≤35 mg/kg/day) (Munson et al. 1982; Auttachoat et al. 2009). Proposed mechanisms include chloroform-mediated dissociation of antigen-antibody complexes and upregulation of thymic stromal lymphopoietin in keratinocytes (Lee et al. 2015).

Neurological. CNS depression is the best-established neurotoxic endpoint, occurring at high acute inhalation concentrations in rats, mice, and cats, consistent with chloroform’s historical use as a surgical anesthetic (Gehring 1968; Lehmann and Flury 1943). Below anesthetic concentrations, more subtle effects have been reported: olfactory nerve loss in rats exposed to ≥10.4 ppm for one week (in association with nasal epithelial degeneration) (Mery et al. 1994), impaired operant conditioning in mice at oral doses ≥100 mg/kg/day for 60 days, and conditioned taste aversion at doses as low as 30 mg/kg/day (Balster and Borzelleca 1982; Landauer et al. 1982). Mechanistically, chloroform is thought to act as a lipophilic membrane perturbant affecting ion channel and neurotransmitter receptor function, with direct binding demonstrated at the GABA-A receptor (Harris and Groh 1985; Jenkins et al. 2001).

Reproductive. Acute oral exposure (179 mg/kg/day for 7 days) decreased serum testosterone in male rats (Potter et al. 1996), and inhalation exposure (≥400 ppm) modestly increased the percentage of morphologically abnormal sperm in mice, though the biological significance of this small change was unclear and no other reproductive endpoints were assessed in that study (Land et al. 1981). In pregnant rodents, effects on litter viability (increased resorptions, decreased live fetuses per litter) occurred at inhalation concentrations ≥291 ppm and gavage doses ≥316 mg/kg/day, generally coincident with doses producing overt maternal toxicity (Schwetz et al. 1974; Murray et al. 1979; Thompson et al. 1974).

Developmental. Developmental toxicity in animals is generally, though not universally, confined to maternally toxic dose levels. Schwetz et al. (1974) reported delayed ossification and wavy ribs in rat fetuses at maternal inhalation exposures ≥30 ppm, with more severe malformations (missing ribs, imperforate anus) at ≥95 ppm; Murray et al. (1979) similarly reported delayed ossification and increased cleft palate incidence in mice at 97–99 ppm. However, other inhalation and oral studies in rats and rabbits found no malformations even at maternally toxic doses (Ruddick et al. 1983; Thompson et al. 1974), and fetal weight decrements consistently tracked maternal weight loss rather than occurring independently of maternal toxicity. One drinking-water study reported altered glucose homeostasis and decreased weanling body weight in rat offspring at very low maternal exposure (75 µL/L), though exposure quantification in that study was limited (Lim et al. 2004).

Endocrine. Direct endocrine organ toxicity in animals is limited to one intermediate-duration drinking-water study in which male rats exposed to 175 mg/kg/day for 90 days developed thyroid follicular lesions (reduced follicular size and colloid density); this was not replicated at lower doses or in females, and no histopathological endocrine effects were observed in numerous other inhalation, gavage, or chronic drinking-water studies in rats, mice, or dogs (Heywood et al. 1979).

Genotoxicity. The weight of evidence across dozens of in vitro and in vivo assays indicates chloroform has low genotoxic potential. It is nonmutagenic in the majority of Salmonella reverse-mutation assays, with only occasional, weak positive findings in select strains (Zhang et al. 2021). In mammalian systems, chloroform produced forward mutations in mouse lymphoma cells only with metabolic activation, and in vivo findings are mixed: increased micronucleated kidney cells occurred in rats after a single gavage dose of 478 mg/kg (Robbiano et al. 1998), and sister chromatid exchange increased in bone marrow of gavage-dosed mice (Morimoto and Koizumi 1983), but chromosome aberration assays and other in vivo tests were frequently negative, and no gene mutations were induced in mouse hepatocytes following inhalation exposure up to 90 ppm for as long as 180 days (Butterworth et al. 1998). Epigenetic effects — global DNA hypomethylation and c-myc promoter hypomethylation with increased c-myc expression in liver and kidney — have also been reported following oral exposure in mice, offering a plausible non-genotoxic contributor to the proliferative response described above (Coffin et al. 2000; Tao et al. 2005; Pereira et al. 2001).

5. Animal Carcinogenicity Bioassay Findings

The animal cancer database for chloroform is substantial and forms the primary basis for its carcinogenic hazard classification. In a two-year inhalation bioassay in F344 rats and BDF1 mice, renal adenomas and carcinomas were increased in male mice at concentrations ≥30 ppm; no renal tumors occurred in rats or female mice, and no other target organ showed significant tumor induction (Yamamoto et al. 2002). A follow-up study confirmed that inhalation alone (up to 100 ppm) did not induce renal tumors in male F344 rats, but combined inhalation (100 ppm) plus oral (approximately 45 mg/kg/day) exposure did increase renal adenomas/carcinomas relative to either single route — although the higher combined dose precludes distinguishing additive from synergistic action (Nagano et al. 2006).

By the oral route, a 78-week gavage bioassay in Osborne-Mendel rats and B6C3F1 mice produced renal carcinomas and adenomas in male rats at 180 mg/kg/day and hepatocellular carcinomas in both sexes of mice at ≥138–238 mg/kg/day (Dunnick and Melnick 1993). Strain A mice, which have a low spontaneous hepatoma background, developed liver tumors (“hepatomas”) in essentially all animals given ≥594 mg/kg/day for 30 days (Eschenbrenner and Miller 1945). In a series of drinking-water bioassays, male Osborne-Mendel rats developed renal tubular adenomas/adenocarcinomas at 160 mg/kg/day, while female B6C3F1 mice given up to 263 mg/kg/day did not (Jorgenson et al. 1985); lifetime drinking-water exposure to 200 mg/kg/day did not produce liver tumors in Wistar rats (Tumasonis et al. 1985, 1987), and B6C3F1 mice given drinking-water doses up to 257 mg/kg/day for 52 weeks likewise showed no exposure-related tumors (Klaunig et al. 1986). In a repeated series of experiments using a toothpaste vehicle, male ICI Swiss mice given 60 mg/kg/day for 80 weeks developed both benign renal cortical adenomas and malignant hypernephromas, an effect specific to the ICI strain and to males, and not replicated in three other mouse strains, in Sprague-Dawley rats, or in beagle dogs dosed for 7.5 years (Roe et al. 1979; Heywood et al. 1979; Palmer et al. 1979).

Chloroform has also been evaluated as a tumor promoter/initiator in initiation-promotion protocols with mixed results: it promoted diethylnitrosamine-induced hepatic preneoplastic foci in F344 rats in one study (Deml and Oesterle 1985) but reduced such foci in another (Reddy et al. 1992), and did not promote or initiate liver tumors in several other rat and mouse protocols (Herren-Freund and Pereira 1987). Collectively, this bioassay literature identifies the liver and kidney as the consistent tumor targets, at doses that in nearly every case coincide with the cytotoxic, regenerative-hyperplasia dose range described above, consistent with the accepted mode of action (Boobis 2009; Meek et al. 2002).

6. Comparison of Animal Findings to Human Epidemiological Evidence, and Reasons for Divergence

Human data are considerably less consistent than the animal database, for both hepatic/renal noncancer endpoints and cancer. For hepatotoxicity, case series of surgical patients receiving chloroform anesthesia (exposures up to 20,000–25,000 ppm) documented biochemical evidence of liver injury (elevated bilirubin, LDH, prothrombin time) (Smith et al. 1973; Whitaker and Jones 1965), and occupational cohorts with exposures in the low hundreds of ppm reported enlarged liver, toxic hepatitis, and fatty liver (Bomski et al. 1967; Phoon et al. 1983). However, other occupational cohorts with lower exposures (single-digit to low tens of ppm) found no association with liver function tests (Challen et al. 1958; Li et al. 1993), and case reports of liver regeneration biomarkers after acute poisoning (Rao et al. 1993) show that, as in animals, the human liver retains substantial regenerative capacity following sublethal injury. In broad terms, this pattern — clear injury only at high, often anesthetic-range or occupationally extreme exposures — mirrors the animal cytotoxicity-threshold pattern, though human occupational exposure assessment is far less precise than controlled animal dosing, limiting quantitative comparison.

For cancer, the epidemiological picture is markedly more equivocal than the animal bioassay data would suggest. Numerous case-control and cohort studies of chloroform or trihalomethanes in chlorinated drinking water have examined bladder, colorectal, pancreatic, breast, and other cancers, with inconsistent results: some studies report modest positive associations (e.g., bladder cancer at drinking-water levels ≥38 µg/L (Bove et al. 2007); breast cancer at levels >24.3 µg/L (Font-Ribera et al. 2018); colon cancer and melanoma in the highest exposure quartile of a large prospective cohort (Doyle et al. 1997); rectal cancer in a follow-up of that same cohort (Jones et al. 2019); endometrial cancer (Medgyesi et al. 2022)), while others find null or even inverse associations for the same or related endpoints, including a large multicenter case-control study showing a decreased risk of colorectal cancer with increasing chloroform exposure (Villanueva et al. 2017, 2021) and pooled meta-analyses finding no overall association between chloroform/trihalomethane drinking-water exposure and total cancer risk (Shi et al. 2024; Summerhayes et al. 2021 found only a slight increase in small-for-gestational-age risk, not cancer). None of the occupational case-control studies of chloroform exposure and lymphoma, glioma, kidney, or other cancers found consistent positive associations, with the exception of a single small subgroup finding for pancreatic cancer based on only two exposed cases (Christensen et al. 2013).

Several factors plausibly explain why the strong, reproducible rodent renal/hepatic tumor signal has not translated into a consistent human cancer signal. First, essentially all epidemiological studies estimate chloroform exposure indirectly from ambient drinking-water or air concentrations (typically single-digit to double-digit µg/L or ppb), several orders of magnitude below the cytotoxic, tumor-inducing doses used in the animal bioassays (tens to hundreds of mg/kg/day, or tens to hundreds of ppm by inhalation); if the carcinogenic mode of action truly requires sustained cytotoxicity and regenerative hyperplasia, community-level drinking-water exposures may simply fall below the threshold at which this sequence is initiated (Meek et al. 2002; Boobis 2009). Second, essentially none of the epidemiological studies adequately control for co-exposure to the many other disinfection byproducts (bromodichloromethane, dibromochloromethane, bromoform, haloacetic acids) that co-occur with chloroform in chlorinated water, several of which have independent toxicological and epidemiological signals, confounding attribution to chloroform specifically (Colman et al. 2011; Williams et al. 2018). Third, toxicokinetic differences between species affect internal dose at a given external exposure: humans metabolize chloroform more slowly than rodents, and the blood/air partition coefficient — which governs pulmonary uptake and retention — is lower in humans than in rodents, so a given ambient air or water concentration is predicted to deliver a smaller internal dose of the reactive metabolite phosgene to human target tissue than to rodent target tissue at an equivalent external exposure (Corley et al. 1990; Taylor et al. 1974). Fourth, human epidemiological studies are subject to substantial exposure misclassification (reliance on municipal supply averages rather than individual biomonitoring, limited data on showering/bathing/swimming contributions, and recall or record limitations), which will tend to attenuate true associations toward the null. Taken together, the divergence between a robust, mechanistically well-characterized rodent hepatic/renal carcinogenicity signal and an inconsistent human epidemiological signal is broadly consistent with a nonlinear, cytotoxicity-dependent mode of action operating at exposure levels well above typical community drinking-water or ambient air concentrations, combined with toxicokinetic differences that further reduce the internal dose delivered to human target tissue at those lower exposures.

7. Data Gaps and Areas of Ongoing Research

Despite the size of the chloroform database, several gaps limit confidence in extrapolating current findings to low-level human exposure. Comparative toxicokinetic data in humans remain sparse relative to the extensive rodent database; human tissue distribution data are limited essentially to autopsy case reports, and data on CYP2E1 activity in human olfactory and other extrahepatic tissue are limited and inconsistent, adding uncertainty to route-specific internal dose estimation (Green et al. 2001; Longo and Ingelman-Sundberg 1993). Existing physiologically based pharmacokinetic (PBPK) models for oral exposure are built around bolus gavage administration; models better representing continuous, low-level drinking-water exposure, and nasal-dosimetry PBPK models extended to humans (currently available only for rats), would reduce uncertainty in low-dose extrapolation (Sarangapani et al. 2002). The mechanistic basis for the marked difference in toxicity between gavage and drinking-water administration is understood qualitatively (saturation of GSH-dependent detoxification with bolus dosing) but has not been fully quantified across durations and strains, and additional low-dose drinking-water studies — particularly for reproductive, developmental, and immune endpoints — would help define whether effects seen at bolus gavage doses have any counterpart at more environmentally realistic exposure patterns. Epidemiological research is substantially limited by co-exposure to other disinfection byproducts; well-designed studies with adequate control for the full trihalomethane/haloacetic acid mixture, and with individual-level (rather than community-average) exposure assessment, are needed to clarify whether the associations intermittently reported for bladder, colorectal, and other cancers reflect a true chloroform-specific effect. Children’s and developmental susceptibility also remains incompletely characterized: while available rodent data suggest that fetal effects occur only at maternally toxic doses, this has not been confirmed at low, non-maternally-toxic exposures, and stratified human studies (e.g., by body mass index, given the toxicokinetics of lipophilic compounds in lean versus obese individuals) have not been conducted (La Merrill and Birnbaum 2011). Finally, although the cytotoxicity/regenerative-hyperplasia mode of action for carcinogenicity is well supported, refinement of dose-response modeling near the proposed cytotoxic threshold, and further evaluation of epigenetic contributors (DNA hypomethylation, oncogene expression changes) to tumor promotion, remain active areas of investigation (Coffin et al. 2000; Tao et al. 2005).

Bibliography