Chloroform Risk Assessment and Liver and Kidney Toxicity

RoseyFields conducts chloroform risk assessments using the following methodology, which documents how the firm selects points of departure, converts doses across species and routes, derives and applies uncertainty factors, establishes reference values, characterizes carcinogenic potential, and translates each of these into defensible exposure and risk estimates for litigation and regulatory support. The approach is anchored in the toxicological database that supports the current chronic reference dose for chloroform, and every quantitative value cited below is traceable to the underlying peer-reviewed studies from which it was derived.

1. Chemical Identity, Regulatory Background, and Exposure Setting

Chloroform (trichloromethane, CHCl₃, CAS No. 67-66-3) is a colorless, volatile, nonflammable liquid with a distinct sweet odor and a molecular weight of 119.38. It has a density of 1.483 g/mL, a vapor pressure of 197 mm Hg at 25 °C, a water solubility of 7.95 g/L at 25 °C, and a log Kₒw of 1.97. Its Henry’s Law constant of 3 × 10⁻³ atm·m³/mole and high vapor pressure make it readily volatile, so that the appropriate air conversion is 1 ppm = 4.88 mg/m³. Because chloroform partitions so readily out of water and soil into air, exposure is almost never confined to a single medium or route.

The regulatory significance of chloroform arises principally from its status as a trihalomethane (THM) disinfection byproduct. When drinking water and wastewater are chlorinated, free chlorine reacts with natural organic matter to form chloroform along with the other regulated trihalomethanes. As a result, chloroform occurs in most public drinking water supplies, and it is the THM most consistently detected. Historically, chloroform also saw wide use as an industrial solvent, an intermediate in the manufacture of refrigerants, an extraction solvent, and a component of consumer products, all of which contribute to environmental releases from chemical and industrial operations.

For a RoseyFields assessment, the practical consequence is that a person exposed to chloroform in water is typically exposed by three simultaneous pathways: direct ingestion of drinking water and food, inhalation of vapor volatilizing from water into indoor air (showering, bathing, cooking, dishwashing, and laundering all generate substantial airborne chloroform), and dermal contact with contaminated water. Investigations of household water use have shown that inhalation and dermal uptake during a single shower can rival or exceed the internal dose delivered by drinking the same water: breath measurements taken shortly after showering correlate directly with tap-water chloroform concentrations, and inhalation-only exposure produced roughly half the breath burden of a combined shower involving both inhalation and skin contact (Jo et al., 1990). Any credible reconstruction of a claimant’s chloroform dose must therefore account for the volatilization-driven inhalation and dermal pathways, not ingestion alone.

2. Point-of-Departure Selection: Critical Study and Endpoint

The foundation of the noncancer assessment is the identification of the most sensitive, reliable, and human-relevant adverse effect and the dose at which it appears. RoseyFields follows the same reasoning EPA applied in constructing the chloroform database: because no reliable long-term oral studies exist in humans, the point of departure (POD) is drawn from the animal literature, and the candidate list is deliberately broad, spanning chronic liver and kidney toxicity studies, reproductive and developmental studies, and shorter-term studies indexing cell proliferation (labeling index) in liver and kidney.

Across this database, liver toxicity is the most sensitive noncancer endpoint after chronic oral exposure, expressed as increased liver fat and histologic evidence of hepatic cytotoxicity. The critical study is the seven-and-a-half-year beagle dog study of Heywood et al. (1979), in which chloroform administered in a toothpaste base produced a dose related increase in the incidence and severity of moderate-to-marked hepatic fatty cysts, accompanied by a sustained, dose-responsive rise in serum glutamate pyruvate transaminase (SGPT), a marker of low-level hepatocellular injury. This study governs the POD for two reasons: it identifies the lowest lowest-observed-adverse-effect level (LOAEL) in the entire database, 15 mg/kg/day, and it is by a wide margin the longest-duration study available, making it uniquely suited to detecting effects that accumulate over a lifetime. Because 15 mg/kg/day was the lowest dose tested, no no observed-adverse-effect level (NOAEL) could be established.

The choice of hepatic fatty cysts as the critical effect warrants explanation, because it is not a commonly reported endpoint. RoseyFields treats it as authentic and toxicologically relevant for the same reasons EPA did: although fatty cysts appeared in the vehicle-control group as well, both their size and severity were significantly increased in treated dogs, and fat accumulation in the liver is a common, characteristic, and well-replicated response to chloroform across species and studies (Roe et al., 1979; Palmer et al., 1979; Jorgenson et al., 1982; Thompson et al., 1974). The concurrent, dose-related SGPT elevation confirms that the lesion reflects genuine hepatocytotoxicity rather than an incidental finding. Supporting studies bracket the dog LOAEL: increased liver fat in mice and rats exposed via drinking water (Jorgenson et al., 1982), moderate-to-severe fatty liver in mice given chloroform in a toothpaste vehicle (Roe et al., 1979), diffuse hepatic degeneration in mice dosed by corn-oil gavage (Bull et al., 1986), and increased hepatocellular labeling index in female rats (Larson et al., 1995b). Renal endpoints — histologic evidence of tubular cytotoxicity in male Osborne-Mendel rats (Hard et al., 2000) — are also part of the database but are less sensitive than the hepatic effect in the dog.

3. Pharmacokinetic Modeling and Dose Conversion

Chloroform’s toxicity is fundamentally a pharmacokinetic phenomenon: injury occurs in the tissues that most actively metabolize the parent compound, and the internal dose that reaches those tissues depends on route, dose rate, body size, and metabolic capacity. RoseyFields therefore treats dose conversion as a central, rather than incidental, element of the assessment.

Absorption is rapid and extensive by all relevant routes. Following oral dosing in humans, more than 90% of an administered dose is recovered in expired air as either unchanged chloroform or carbon dioxide within eight hours (Fry et al., 1972), and gastrointestinal absorption in animals ranges from 64% to 98%. The dosing vehicle matters: aqueous vehicles yield faster absorption and higher peak blood concentrations than oil vehicles (Withey et al., 1983), a distinction that becomes decisive when comparing corn-oil gavage bioassays to the drinking-water exposures that characterize real human contact. Absorbed chloroform distributes widely, concentrating in body fat, with lower levels in kidney, liver, and brain (McConnell et al., 1975), and is taken up rapidly by the liver and kidney (Pereira, 1994; Gemma et al., 1996).

Metabolism is the key determinant of toxicity. Chloroform is oxidized by cytochrome P450 — chiefly CYP2E1 at low, environmentally relevant concentrations (Constan et al., 1999), with CYP2B1/2 contributing at high concentrations — to trichloromethanol, which spontaneously dehydrochlorinates to the highly reactive electrophile phosgene. Phosgene binds cellular macromolecules, depletes glutathione, and produces the cytotoxicity that drives both the noncancer and cancer responses (Pohl et al., 1977). Because metabolism is enzyme-mediated and saturable, nearly all of a low dose is metabolized, whereas at high dose rates metabolic capacity saturates and increasing fractions are exhaled unchanged. This dose-rate dependence explains why bolus gavage produces more injury than the same daily dose delivered continuously in water.

To move from animal internal dose to human external dose, RoseyFields relies on the physiologically based pharmacokinetic (PBPK) framework first developed for chloroform by Corley et al. (1990), which represents the mouse, rat, and human as systems of perfusion-limited compartments (gastrointestinal tract, lung, liver, kidney, fat, and rapidly and slowly perfused tissues) with saturable Michaelis-Menten metabolism in liver and kidney. The model predicts that metabolized dose per unit tissue is highest in the mouse, intermediate in the rat, and lowest in the human, reflecting the lower mass-specific metabolic, ventilation, and cardiac-output rates of larger species; if equal macromolecular binding is assumed equitoxic, chloroform’s relative potency runs mouse > rat > human. Reitz et al. (1990) extended the model to relate hepatic metabolic rate to cell killing, and Smith et al. (1995) demonstrated that dose-rate-dependent internal metrics (maximal metabolic rate and percentage of hepatocytes killed per day) best reconcile the rodent liver tumor data, while the kidney data are best described by dose scaled to body surface area.

In practice, RoseyFields applies two conversions consistent with the reference-value derivation. For the noncancer POD, the dog LOAEL is adjusted for a discontinuous dosing schedule (six days per week) by multiplying by 6/7. For cross-species cancer dose conversion, animal doses are scaled to human-equivalent doses (HEDs) by body weight to the ¾ power, following EPA’s default: HED = animal dose × (BW_animal / BW_human)^(1/4). Using the study body weight of 0.375 kg for male Osborne-Mendel rats and 70 kg for humans, the administered rat doses of 19, 38, 81, and 160 mg/kg/day correspond to HEDs of 5.1, 10.3, 21.9, and 43.2 mg/kg/day. Where site-specific data support it, RoseyFields uses the Corley PBPK model — including the dermal-and-bathing extension of Blancato and Chiu (1994) — to reconstruct internal dose directly rather than relying on default body-weight scaling.

4. Uncertainty Factor Derivation

RoseyFields derives the oral reference dose (RfD) using the benchmark dose (BMD) approach as the primary method and the traditional NOAEL/LOAEL approach as a cross-check, applying uncertainty factors (UFs) transparently and documenting the basis for each.

Under the NOAEL/LOAEL approach, the composite uncertainty factor is 1,000, built from three tenfold components. A factor of 10 accounts for extrapolation from a LOAEL to a NOAEL, necessary because the critical study identified only a LOAEL (15 mg/kg/day was the lowest dose tested). A second factor of 10 accounts for interspecies extrapolation from the dog to humans. A third factor of 10 accounts for intraspecies variability — the protection of potentially sensitive human subpopulations. Applying the duration-adjusted LOAEL yields:

RfD = [15 mg/kg/day × (6/7)] ÷ 1,000 = 1 × 10⁻² mg/kg/day.

Under the preferred benchmark dose approach, the fatty-cyst incidence data from Heywood et al. (1979) were modeled to a 10% extra-risk benchmark response. The best-fitting model (quantal-linear) produced a benchmark dose (BMD₁₀) of 1.7 mg/kg/day and a lower 95% confidence limit (BMDL₁₀) of 1.2 mg/kg/day. This BMDL — an internally derived surrogate for a NOAEL — is duration-adjusted:

BMDL₁₀ = 1.2 mg/kg/day × (6/7) = 1.0 mg/kg/day.

Because the BMDL already represents a defined low-response level, no LOAEL-to-NOAEL factor is required, so the composite uncertainty factor collapses to 100: a factor of 10 for interspecies extrapolation and a factor of 10 for intraspecies variability. Critically, RoseyFields documents why the remaining conventional factors are set to 1. No subchronic-to-chronic factor is applied because the critical study is a 7.5-year (chronic, indeed near-lifetime) exposure. No database-deficiency factor is applied because the chloroform database is complete: chronic bioassays exist in the dog (Heywood et al., 1979), rat, and mouse (Jorgenson et al., 1982; Jorgenson et al., 1985); oral developmental studies exist in rats and rabbits (Thompson et al., 1974) and by inhalation in rats (Schwetz et al., 1974) and mice (Murray et al., 1979); and a two-generation reproduction study found no effects on fertility at doses producing liver histopathology (NTP, 1988). Finally, because chloroform is rapidly metabolized and excreted and does not bioaccumulate, no adjustment for accumulation is warranted.

5. Derivation of the Oral Reference Dose and Inhalation Reference Concentration

Applying the composite uncertainty factor of 100 to the duration-adjusted benchmark dose lower limit gives the chronic oral RfD:

RfD = 1.0 mg/kg/day ÷ 100 = 1 × 10⁻² mg/kg/day (0.01 mg/kg/day).

The NOAEL/LOAEL and benchmark approaches converge on precisely the same value, 0.01 mg/kg/day — a reassuring, though partly coincidental, consistency that strengthens confidence in the number. RoseyFields adopts the benchmark-based derivation as primary because the BMD method uses the full shape of the dose-response curve, is less sensitive to dose spacing and study power, and is not restricted to the doses actually tested. Overall confidence in the RfD is characterized as medium: the critical effect (hepatic fatty cysts) is uncommon and was observed in a toothpaste-base vehicle whose glycerol and oil constituents could in principle have modified the response, and the critical data set comprised only two doses plus a control, which constrains curve-fitting. These caveats are offset by the study’s exceptional duration, the biological coherence of the hepatic endpoint across the database, and the convergence of the two derivation methods.

For the inhalation reference concentration (RfC), no verified value is available. The reference database did not support derivation of an RfC at the time the assessment was finalized; the endpoint was formally reserved pending further review of the inhalation literature. RoseyFields therefore does not apply a chloroform RfC. Where an inhalation exposure must be evaluated, the firm’s practice is to note the absence of a verified RfC explicitly, to evaluate airborne exposures against the oral RfD only through a defensible route-to-route extrapolation using the Corley et al. (1990) PBPK model when data permit, and otherwise to characterize inhalation risk qualitatively rather than force an unsupported quantitative comparison. The inhalation noncancer database that would inform such work includes controlled animal studies showing olfactory epithelial, hepatic, and renal effects (Templin et al., 1996a; Larson et al., 1996; Nagano et al., 1998) and occupational reports of hepatic and neurological effects at higher air concentrations (Phoon et al., 1983; Li et al., 1993).

6. Carcinogenicity Classification and the Basis for Declining a Slope Factor

Chloroform is characterized as likely to be carcinogenic to humans by all routes of exposure under high-dose conditions that lead to cytotoxicity and regenerative hyperplasia in susceptible tissues, and not likely to be carcinogenic at exposure levels that do not cause cytotoxicity and sustained cell regeneration. This conditional characterization is the single most consequential feature of the chloroform cancer assessment, and it distinguishes chloroform from most environmental carcinogens.

The classification rests on mode-of-action evidence rather than on human epidemiology, which is uninformative for chloroform specifically. Although numerous studies associate chlorinated drinking water with a weak increase in bladder and other cancers, chlorinated water contains many disinfection byproducts, and standard causal criteria do not support attributing any cancer risk to chloroform per se (Cantor et al., 1998). In animals, chloroform increases liver tumors in mice and kidney tumors in male rats and mice, but only at doses that first produce cytotoxicity. The weight of evidence indicates that sustained or repeated cell killing, followed by compensatory regenerative hyperplasia, precedes and is probably causal for the neoplasia. This cytolethality-driven mechanism has been documented directly: cell proliferation and tumor formation track one another across organs and dose in both liver and kidney (Larson et al., 1994b; Templin et al., 1998), and a histologic re-evaluation of the pivotal kidney bioassay established persistent renal tubule injury at exactly the doses that induce renal tumors (Hard et al., 2000). Chloroform is not strongly mutagenic and is not expected to cause cancer through a genotoxic mechanism.

Because cytotoxicity — and therefore carcinogenicity — occurs only above a threshold dose, a nonlinear (margin of-exposure) approach is scientifically appropriate, and RoseyFields consequently does not apply a low-dose linear cancer slope factor (CSF) or inhalation unit risk (IUR) for chloroform. This is a deliberate methodological position, not an omission: applying default linear extrapolation to a threshold carcinogen substantially overstates low dose risk. RoseyFields quantifies cancer potential in two complementary ways consistent with the nonlinear mode of action.

First, in the margin-of-exposure (MOE) analysis, the principal study is Jorgenson et al. (1985), in which male Osborne-Mendel rats received chloroform in drinking water at 0, 200, 400, 900, or 1,800 mg/L (time-weighted average doses of 0, 19, 38, 81, and 160 mg/kg/day) for 104 weeks, producing a dose-related increase in combined kidney tumors. This study is preferred over corn-oil gavage bioassays because exposure was via drinking water — the most human-relevant route, avoiding vehicle and bolus-rate artifacts — and because its five dose groups and large low-dose group provide strong resolution of the low end of the curve (an emphasis independently recommended by NAS, 1987). Modeling the tumor data across dichotomous models yielded consistent (not strongly model-dependent) results; the preferred quantal-linear model produced a central estimate ED₁₀ of 36 mg/kg/day and a point of departure LED₁₀ (lower 95% bound on the dose producing 10% extra tumor risk) of 23 mg/kg/day, expressed as human equivalent doses after body-weight-to-the-¾ scaling. Comparing this LED₁₀ of 23 mg/kg/day to the RfD of 0.01 mg/ kg/day yields a margin of exposure of roughly 2,000, which is regarded as adequately protective of public health for cancer effects given the nonlinear dose-response and the shared cytotoxic pathway underlying both cancer and noncancer effects.

Second, in the RfD approach to cancer, because chloroform’s carcinogenicity is secondary to a threshold toxicity, an RfD-type analysis is scientifically equivalent to the cancer MOE analysis. Since doses at or below the RfD of 0.01 mg/kg/day do not produce cytolethality or regenerative hyperplasia, that RfD is judged protective against increased cancer risk as well as against noncancer effects. RoseyFields’ operating conclusion is therefore that an exposure at or below 0.01 mg/kg/day is not expected to produce either cancer or noncancer effects from chloroform by the oral route. The inhalation cancer assessment, like the RfC, is reserved and no verified inhalation unit risk is adopted. Assessors who encounter older linear slope-factor values for chloroform in legacy databases should recognize that those values reflect a superseded default-linear treatment that the current mode-of-action analysis does not support; RoseyFields does not apply them.

7. Application in Practice: Hazard Quotients, Hazard Index, Cancer Risk, and Dose Reconstruction

RoseyFields’ risk assessors convert these toxicity values into case-specific conclusions through a standard, reproducible sequence.

Exposure and dose reconstruction comes first. For each claimant or exposed population, the assessor reconstructs an average daily dose (ADD, or lifetime average daily dose, LADD, for cancer) from measured or modeled environmental concentrations, contact rates, exposure frequency and duration, and body weight, summing across the ingestion, inhalation, and dermal pathways that chloroform’s volatility makes simultaneously relevant. Because inhalation and dermal uptake during showering and other household water use can dominate the total internal dose (Jo et al., 1990), the firm models volatilization from water into indoor air and, where data support it, uses the Corley et al. (1990) PBPK model — with the bathing/dermal extension of Blancato and Chiu (1994) — to translate external contact into delivered internal dose. Vehicle and dose-rate considerations (Withey et al., 1983; Smith et al., 1995) inform how bolus-type versus continuous exposures are weighted.

Noncancer risk is expressed as a hazard quotient (HQ), the ratio of the reconstructed oral dose to the RfD: HQ = ADD (mg/kg/day) ÷ RfD (0.01 mg/kg/day).

An HQ at or below 1 indicates exposure below the level expected to produce adverse effects; an HQ above 1 signals potential concern and prompts closer evaluation of the exposure estimate and the sensitive endpoint. Where chloroform co-occurs with other trihalomethanes or hepatotoxicants acting on the same target organ, the assessor sums the component HQs into a Hazard Index (HI) for the liver, consistent with dose addition for chemicals sharing a mode of action. Because chloroform lacks a verified RfC, inhalation contributions are handled through route-to-route PBPK extrapolation to the oral RfD rather than through an independent air-based hazard quotient, with the limitation stated explicitly.

Cancer risk is handled through the nonlinear framework rather than by multiplying dose by a slope factor. The assessor computes a case-specific margin of exposure — the ratio of the LED₁₀ point of departure (23 mg/kg/day) to the reconstructed exposure — and evaluates its adequacy in light of the intraspecies and interspecies factors, the 10% response level of the point of departure, and residual pharmacodynamic uncertainty. Equivalently and more simply, because doses at or below the RfD carry no expected cancer risk, the assessor compares the reconstructed dose to the RfD of 0.01 mg/kg/day: an exposure below that value supports a conclusion of no appreciable cancer risk. RoseyFields does not report an “excess lifetime cancer risk” numeric probability (e.g., 1 × 10⁻⁶) derived from a linear slope factor for chloroform, because doing so would misrepresent a threshold carcinogen; when opposing experts present such figures, RoseyFields is prepared to explain on the record why the linear calculation is not scientifically supported for this chemical.

Litigation and regulatory support flow directly from this structure. In toxic-tort and environmental litigation, the methodology supplies a transparent, peer-reviewed-literature-anchored chain from measured concentrations to reconstructed dose to a hazard quotient or margin of exposure, allowing an expert to state clearly whether a claimant’s exposure did or did not exceed health-protective benchmarks, and to withstand Daubert-type scrutiny because each step is documented and defensible. The explicit nonlinear, threshold-based cancer characterization is often pivotal: it lets RoseyFields address causation directly, distinguishing exposures that could plausibly contribute to disease (those high enough to cause sustained cytotoxicity) from the far more common low-level exposures that the mode-of-action science indicates carry no appreciable cancer risk. In regulatory support — permitting, cleanup-level derivation, and comment on drinking-water and remediation standards — the same values feed screening comparisons, help set risk based concentrations, and provide a basis for arguing that action levels should reflect chloroform’s threshold behavior rather than default linearity.

8. Key Uncertainties and Sensitive Populations

RoseyFields characterizes uncertainty explicitly in every chloroform assessment, because the defensibility of a conclusion depends on candor about its limits.

The most important noncancer uncertainties attach to the critical study. Hepatic fatty cysts are an uncommon endpoint, and although fat accumulation in the liver is a well-established chloroform effect, reliance on this specific lesion introduces some uncertainty. The toothpaste-base vehicle contained glycerol and small quantities of natural oils that could in principle have potentiated the response, though this is not judged a likely explanation. The dog appears markedly more sensitive than rodents — the dog-based benchmark dose is far lower than rodent-based values — and the reason is unknown; it may reflect a true species difference or the greater sensitivity of a near-lifetime exposure to detect low-level injury. Finally, the two-dose-plus-control design of the critical data set limits definition of the low dose curve shape. On the cancer side, key uncertainties include the residual pharmacodynamic differences between rat and human that body-weight scaling may not fully capture, the relatively shallow slope of the kidney-tumor dose response, reliance on tumor rather than an earlier precursor endpoint, and — most fundamentally — the strength of the evidence that cytotoxicity is genuinely necessary (not merely associated) for tumorigenesis, on which the entire nonlinear approach depends.

Sensitive populations deserve specific attention. Susceptibility to chloroform is governed largely by CYP2E1- mediated oxidative metabolism, so any factor that alters that enzyme alters risk. Co-exposure to CYP2E1 inducers — ethanol and other alcohols, and various ketones — can increase chloroform metabolism and thereby increase target organ toxicity, while agents that inhibit the enzyme reduce it; individuals with substantial alcohol intake are therefore plausibly more susceptible. Conditions or co-exposures that deplete glutathione remove a key protective buffer against phosgene and may heighten injury. Sex differences are well documented in animals: renal toxicity is generally more severe in males because the responsible cytochrome P450 is induced by testosterone. With respect to children and the developing fetus, the available evidence does not indicate greater susceptibility, and there is a mechanistic reason to expect the fetus may be relatively protected: the enzyme required to generate chloroform’s toxic metabolite, CYP2E1, is low or absent in the fetus and rises only after birth (Hakkola et al., 1998). Reproductive and developmental effects in animals occur only at doses that also produce maternal toxicity (Thompson et al., 1974; Schwetz et al., 1974; Murray et al., 1979), and no fertility effects appear at doses causing liver histopathology (NTP, 1988), so developmental effects appear secondary to maternal toxicity rather than indicative of a uniquely sensitive window. RoseyFields nonetheless evaluates childhood exposure explicitly in every assessment, both because higher intake-to-body-weight ratios can raise a child’s dose per kilogram and because exposure latency over a longer remaining lifespan is relevant to any cumulative endpoint. When case-specific factors — heavy alcohol use, relevant co-exposures, pre-existing liver or kidney disease — are present, the assessor documents how they shift the exposed individual’s position relative to the population-protective reference values.

Bibliography