Hexavalent Chromium (Cr-VI) Toxicology and Lung Cancer

RoseyFields evaluates the toxicology of hexavalent chromium using the following evidence base, drawn from the primary peer-reviewed and government-sponsored animal toxicology literature underlying EPA’s 2024 IRIS toxicological review, together with the occupational epidemiology and mechanistic literature relevant to interpreting that animal database. Cr(VI) presents an unusually well-characterized, organ-system-specific toxicological picture, with quantitative reference values derived separately for gastrointestinal, hepatic, developmental, and hematological effects (oral route) and for upper and lower respiratory effects (inhalation route), and with carcinogenicity assessed separately — and by different anchoring evidence — for the oral and inhalation routes of exposure.

1. Chemical Identity and Physicochemical Properties Relevant to Toxicity

Hexavalent chromium, Cr(VI) (CASRN 18540-29-9), refers to a group of inorganic compounds containing chromium in the +6 oxidation state, most commonly as the chromate (CrO4 2-) or dichromate (Cr2O7 2-) anion. Cr(VI) compounds are widely used industrially in pigment manufacturing, corrosion inhibition, and metal finishing. Many Cr(VI) compounds are highly water-soluble, giving them high environmental mobility in soil and groundwater and creating potential for drinking water contamination; Cr(VI) may also be released to air by industrial sources and by fossil fuel combustion.

A defining toxicological property of Cr(VI) is its structural resemblance to naturally occurring divalent anions: the chromate anion is similar in size and charge distribution to sulfate and phosphate, allowing it to enter cells efficiently through nonspecific anion transport channels that normally handle these essential ions. This is in marked contrast to the trivalent form, Cr(III), which exists primarily as a cation, does not readily cross cell membranes through these channels, and is comparatively inert biologically (and, at low levels, is considered an essential trace nutrient). This differential membrane permeability between oxidation states — rather than persistence or bioaccumulation in the classical sense — underlies most of Cr(VI)’s distinctive toxicological behavior: Cr(VI) is not itself a stable, bioaccumulative species in tissue, but is progressively reduced to Cr(III) intracellularly and extracellularly, and it is this reduction step, and the reactive intermediates it generates, that drives both the noncancer toxicity and the mutagenic carcinogenicity discussed below (Zhitkovich, 2011).

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

Absorption

Following oral ingestion, Cr(VI) absorption and systemic availability depend heavily on dose. At oral ad libitum doses below approximately 1 mg/kg-day, Cr(VI) absorbed by the gastrointestinal tract is largely reduced to Cr(III) by the liver (and, to a lesser extent, by plasma and red blood cells in the portal circulation) before reaching most systemic tissues, though the gastrointestinal tract and liver themselves remain directly exposed to Cr(VI). At doses at or above approximately 1 mg/kg-day, Cr(VI) absorption exceeds the reducing (‘detoxifying’) capacity of the liver and is distributed more widely to systemic tissues, including kidney, lung, and brain. Notably, this dose-dependence does not imply a true threshold below which no systemic absorption occurs: human studies using radiolabeled chromium demonstrated that even very low concentrations of ingested Cr(VI) (on the order of 1.3 × 10-5 mg/L) can be absorbed intact by the small intestine and distributed systemically (Donaldson and Barreras, 1966), a finding with direct bearing on the low-dose linear extrapolation approach used for Cr(VI) oral carcinogenic risk.

Gavage and injection routes of administration bypass much of this first-pass detoxification: oral gavage compresses the exposure into a short time window that can exceed the stomach and liver’s reducing capacity even at doses that would be adequately detoxified if delivered continuously via drinking water or diet, and intravenous or intraperitoneal injection bypasses first-pass gastric and hepatic reduction entirely, exposing systemic tissues to substantially higher effective Cr(VI) doses than an equivalent oral gavage dose.

Distribution and Biomonitoring

Because only total chromium (not oxidation state) can be reliably measured in tissue, distinguishing systemic Cr(VI) absorption from background Cr(III) uptake requires indirect biomarkers. Red blood cell (RBC) chromium is a particularly useful indicator: RBCs rapidly take up and reduce circulating Cr(VI), and the resulting Cr(III) becomes bound to hemoglobin and released back to circulation only slowly, so elevated RBC chromium (and an elevated RBC to-plasma chromium ratio) persists longer than plasma chromium following systemic Cr(VI) absorption and serves as a relatively specific marker that unreduced Cr(VI) reached the bloodstream rather than having been fully detoxified in the gut or liver.

Metabolism (Reduction)

Cr(VI) is not metabolized in the conventional sense but is progressively reduced to Cr(III) through both extracellular and intracellular pathways: extracellular reduction occurs in gastric juice, plasma, and other body fluids; intracellular reduction (following cellular uptake through anion channels) is mediated primarily by ascorbate and, to a lesser extent, glutathione and other cellular reductants, generating short-lived reactive intermediate oxidation states (Cr(V), Cr(IV)) and reactive oxygen species en route to the stable Cr(III) end product (Zhitkovich, 2011). This intracellular reduction step, rather than Cr(VI) itself, is the proximate source of the DNA damage and oxidative stress discussed in Section 3.

Excretion

Chromium is excreted primarily in urine, and urinary chromium has been used extensively as a biomarker of recent occupational exposure in the human epidemiological literature discussed in Section 6, though it does not distinguish Cr(VI) from Cr(III) or reflect the internal dose delivered to specific target tissues.

Species and Sex Differences

Species differences in gastrointestinal detoxification capacity are central to interpreting the animal carcinogenicity bioassay: the chronic drinking-water bioassay in rats and mice produced tumors at different gastrointestinal sites in the two species — squamous cell tumors of the oral cavity/tongue in rats, reflecting direct mucosal contact with Cr(VI) prior to swallowing and gastric reduction, versus adenomas and carcinomas of the small intestine in mice, reflecting Cr(VI) that survived gastric transit and reached the small intestine before being fully reduced. This site-specificity required two different human-equivalent-dose extrapolation approaches: physiologically based pharmacokinetic (PBPK) modeling of gastric Cr(VI) reduction capacity for the mouse intestinal tumor data, and simple body-weight (BW3/4) allometric scaling for the rat oral cavity tumor data, since oral cavity exposure precedes gastric reduction entirely. Sex differences are also evident: the oral cancer slope factor is anchored on tumors in female rats specifically (squamous cell carcinoma/papilloma of the oral mucosa or tongue), and chronic hepatic inflammation — the basis for the hepatic noncancer reference value — was likewise identified in female rats rather than males, while the hematological reference value was based on decreased hemoglobin in male rats specifically, indicating that sex specific sensitivity varies by organ system/endpoint rather than following a single consistent pattern across all Cr(VI) effects.

3. Mechanism(s) of Action / Mode of Action

Cr(VI) carcinogenicity operates through a mutagenic mode of action that EPA has determined is sufficiently supported in laboratory animals and relevant to humans, triggering linear low-dose extrapolation for both the oral and inhalation cancer risk estimates. The mechanism begins with cellular uptake of the chromate anion through nonspecific sulfate/phosphate anion channels (a consequence of its structural mimicry of these essential anions, as described in Section 1), followed by intracellular reduction via ascorbate and glutathione to unstable pentavalent and tetravalent intermediates and, ultimately, stable trivalent chromium. This reduction cascade generates reactive oxygen species and reactive chromium intermediates capable of forming chromium-DNA adducts, DNA-protein crosslinks, and oxidative DNA damage, providing a coherent mechanistic basis for Cr(VI) mutagenicity and, ultimately, carcinogenesis (Zhitkovich, 2011; Costa and Klein, 2006).

The same oxidative-stress mechanism that underlies Cr(VI) mutagenicity is also implicated more broadly in Cr(VI)- induced cytotoxicity across multiple tissues, including the gastrointestinal tract, providing mechanistic support for the noncancer hazards (GI, hepatic) identified from animal histopathology data. A further, more specific mechanistic link has been identified between Cr(VI) exposure and intestinal tumorigenesis: suppression of the cystic fibrosis transmembrane conductance regulator (CFTR) gene has been shown to enhance intestinal tumorigenesis in animal models, and CFTR has been found to be inactivated in mice exposed to Cr(VI), suggesting a mechanistic pathway — beyond direct DNA damage — by which Cr(VI) may promote small intestinal carcinogenesis specifically, and raising the possibility that individuals with genetically impaired CFTR function could experience a compounding reduction in CFTR expression following oral Cr(VI) exposure.

4. Animal Toxicological Evidence by Organ System / Health Outcome

The current Cr(VI) noncancer toxicology database is anchored substantially in a coordinated set of high-confidence National Toxicology Program (NTP) rodent studies, supplemented by other subchronic and continuous-breeding studies, each addressing a specific organ system.

Gastrointestinal Tract

The pivotal evidence for GI toxicity comes from a high-confidence chronic 2-year drinking-water study in F344 rats and B6C3F1 mice of both sexes (NTP, 2008), which reported diffuse epithelial hyperplasia of the duodenum in mice at multiple dose levels. This hyperplasia finding — evaluated using benchmark dose (BMD) modeling in male mice and a LOAEL analysis in female mice, with human equivalent doses derived via PBPK modeling of gastric Cr(VI) detoxification — forms the basis of the overall chronic oral reference dose of 9 × 10-4 mg/kg-day. Multiple high confidence subchronic studies support these chronic findings, and mechanistic evidence implicating oxidative stress in Cr(VI)-induced cytotoxicity across GI tissue further supports the human relevance of this effect.

Hepatic

Hepatic toxicity was assessed using both histological and serum biomarker endpoints across several NTP studies of different durations: elevated serum alanine aminotransferase (ALT) was observed in male rats at 90 days (NTP, 2007) and again at 3 and 12 months (NTP, 2008), and in female rats at 90 days (NTP, 2007), while chronic inflammation was observed in female rat livers and fatty change in female rat livers after 2 years of exposure, and chronic hepatic inflammation was also observed in female mouse liver after 2 years (NTP, 2008). Chronic inflammation in female F344 rats was selected as the basis for the hepatic organ-specific reference dose of 7 × 10-4 mg/kg-day, derived using a LOAEL analysis and the same PBPK-based human equivalent dose approach used for the GI endpoint. Hepatic effects are considered more likely to occur following oral, rather than inhalation, exposure because of the first-pass hepatic exposure that follows gastrointestinal absorption.

Developmental Toxicity

Developmental toxicity was assessed primarily through a continuous-breeding reproductive toxicity study in BALB/c mice (NTP, 1997), which found decreased F1 offspring postnatal growth, forming the basis of a developmental organ specific reference dose of 0.07 mg/kg-day (derived from a NOAEL rather than a LOAEL). Decreased offspring growth was observed across most of the animal studies reviewed, supporting the qualitative conclusion that Cr(VI) is likely to cause developmental toxicity in humans; however, confidence in the specific quantitative reference value is low, because the developmental effect occurred only at doses high enough that other toxicological effects (as reflected in the lower points of departure derived for GI, hepatic, and hematological effects at comparable or lower doses) may have been occurring concurrently and could be contributing to, or confounding, the observed decrease in offspring growth.

Hematological

Hematological toxicity was evaluated using short-term/low-dose data from the NTP (2008) chronic bioassay, which found decreased hemoglobin in male rats after 22 days of exposure. The overall hematological pattern — decreased hemoglobin together with other changes — was judged consistent with a regenerative microcytic hypochromic anemia. Because the effect had its highest magnitude early in exposure and appeared to ameliorate over time, a subchronic-to-chronic uncertainty factor was not applied, and the resulting organ-specific reference dose of 0.01 mg/kg-day was derived using BMD analysis and PBPK modeling. Confidence in this value is medium, diminished specifically by uncertainty over the apparently transient nature of the effect.

Immune

Evidence for immune effects was drawn from a combination of ex vivo white blood cell function assays (in both human and animal studies), antibody responses to a T-cell-dependent antigen (in animal studies), and reduced host resistance to bacterial infection (in animal studies). This combination of findings was judged coherent enough to conclude that Cr(VI) may cause immune toxicity in humans, but confidence in this conclusion is reduced because most of the underlying studies were of low confidence and findings were often inconsistent from study to study.

Male Reproductive

Male reproductive effects on sperm parameters and serum testosterone were reported in both human and animal studies, but most of these studies were considered low confidence, and among the smaller number of high-confidence rodent studies, findings were inconsistent — some showing effects and others not — leaving male reproductive toxicity in the ‘suggestive but inconsistent’ evidence tier rather than a well-established hazard.

Female Reproductive

Evidence for female reproductive toxicity was judged inadequate to support a hazard conclusion. A single low confidence epidemiology study and a series of low-confidence animal toxicology studies suggested an association, but medium- and high-confidence studies did not identify effects, apart from a moderate decrease in maternal body weight in some studies — an effect that, on its own, does not establish specific female reproductive toxicity.

Respiratory (Lower Respiratory Tract)

Lower respiratory toxicity following inhalation exposure was evaluated using six medium-confidence animal inhalation studies examining lung cellular responses and histopathology. Because histopathological and cellular changes co-occurred, and were accompanied by serum biomarkers indicative of an inflammatory response, these findings were judged to represent a coherent, adverse pattern sufficient to conclude that Cr(VI) is likely to cause lower respiratory toxicity in humans, with mechanistic evidence (the oxidative-stress mechanism described in Section 3) supporting the human relevance of the animal findings.

Respiratory (Upper Respiratory Tract / Nasal Cavity)

Nasal cavity effects of Cr(VI) — irritation and ulceration of the nasal mucosa or septum, septal perforation, and nasal bleeding — have been well established in humans since EPA’s 1998 evaluation of the inhalation literature, and hazard identification for this endpoint was not re-performed in the current assessment; instead, the current review focused on identifying data suitable for refining the quantitative dose-response relationship for this endpoint. Unlike most of the organ-system hazards discussed above, the quantitative point of departure for nasal effects — and for the overall inhalation reference concentration of 3 × 10-5 mg/m3 — is derived directly from human occupational data rather than from an animal bioassay, as discussed further in Section 6.

5. Animal Carcinogenicity Bioassay Findings

The pivotal animal carcinogenicity data for Cr(VI) come from a single, high-confidence chronic 2-year drinking-water bioassay conducted in both F344 rats and B6C3F1 mice of both sexes (NTP, 2008), which found ‘clear evidence of carcinogenic activity’ in both species, at two distinct gastrointestinal sites reflecting the species-specific toxicokinetic patterns described in Section 2. In rats, Cr(VI) produced increased incidences of squamous cell carcinomas and squamous cell papillomas of the oral cavity/tongue in both sexes, with the female rat tumor response selected as the principal basis for the human oral cancer slope factor. In mice, Cr(VI) produced increased incidences of adenomas and carcinomas of the small intestine in both sexes. Benchmark dose modeling was applied to both tumor datasets to derive points of departure (BMDL10 for the rat oral cavity tumors and BMDL01 for the mouse intestinal tumors, the lower benchmark response reflecting the more sensitive dose range needed to appropriately characterize the nonlinear low-dose intestinal dosimetry in mice). Human equivalent doses were then derived using PBPK modeling of gastric Cr(VI) reduction for the mouse intestinal tumor data (since these tumors form only after Cr(VI) escapes gastric detoxification) and BW3/4 allometric scaling for the rat oral cavity tumor data (since these tumors form from direct oral mucosal contact prior to any gastric reduction). Small intestinal carcinogenesis is itself a notably rare tumor type in both humans and rodents, underscoring the specificity of this finding to Cr(VI) exposure via the pathway described in Section 3 (CFTR suppression enhancing intestinal tumorigenesis).

The resulting oral slope factors for adult exposure ranged from 0.098 to 0.108 (mg/kg-day)-1 across the four sex/species/tumor-site combinations evaluated, a notably tight range given that the estimates derive from two different tumor sites, two species, and two different dosimetric extrapolation methods — providing a measure of internal consistency to the animal carcinogenicity database despite its reliance on a single underlying bioassay. The final adult oral slope factor of 0.16 (mg/kg-day)-1 (and total lifetime exposure oral slope factor of 0.26 (mg/kg-day)-1, incorporating age-dependent adjustment factors for early-life mutagenic susceptibility) is based specifically on the female rat oral cavity tumor response.

6. Comparison of Animal Findings to Human Epidemiological Data

A notable feature of the Cr(VI) evidence base is that the animal-human relationship differs sharply between the oral and inhalation routes of exposure, in effect running in opposite directions for the two routes.

Oral Route: Animal-Anchored, with Slight Human Support

For the oral route, EPA’s determination that Cr(VI) is ‘likely to be carcinogenic to humans’ rests on a combination described explicitly as ‘robust’ animal evidence of GI tract carcinogenicity (the NTP 2008 bioassay discussed in Section 5) together with only ‘slight’ evidence of carcinogenicity from human studies, supported by strong mechanistic evidence for a mutagenic mode of action. In other words, for the oral route, the animal bioassay data — not human epidemiology — provide the primary quantitative basis for the cancer risk estimate, with human data playing a corroborating rather than an anchoring role. This pattern is consistent with the practical difficulty of conducting well powered epidemiological studies of oral Cr(VI) exposure and GI cancer: the relevant human exposure circumstances (contaminated drinking water) are geographically dispersed, exposure levels are typically far lower than the effective doses in the rodent bioassay, and confounding by other dietary and environmental exposures is difficult to fully control.

Inhalation Route: Human-Anchored, with Animal Support

For the inhalation route, by contrast, Cr(VI) has long been established as a known human lung carcinogen based on ‘long-standing evidence of a causal relationship between inhalation of Cr(VI) and increased incidence of lung cancer in humans in occupational settings,’ a determination originally made in 1998 and maintained in the current review specifically because the human epidemiological evidence is considered sufficiently strong and consistent on its own. The quantitative inhalation unit risk is derived directly from a human occupational cohort — chromate production workers at a Baltimore, Maryland facility (Gibb et al., 2000b), with a subsequent analysis of the same cohort examining how relative risk varies by age at exposure (Gibb et al., 2020) — rather than from an animal bioassay. This is the reverse of the oral-route pattern: for inhalation, human epidemiology anchors the quantitative risk estimate, and animal inhalation carcinogenicity data (not detailed further in the reviewed source materials) play a supporting rather than a primary role.

This same human-anchored pattern extends to the noncancer inhalation toxicity value: the overall inhalation reference concentration of 3 × 10-5 mg/m3 is based on human data specifically — longitudinal occupational studies of ulcerated nasal septum (Gibb et al., 2000a) and supporting studies of nasal mucosal pathology (Cohen et al., 1974) and nasal septum ulceration in Swedish chrome-plating workers (Lindberg and Hedenstierna, 1983) — rather than on an animal inhalation bioassow. The three candidate point-of-departure values derived from these three independent human occupational studies were reasonably close in magnitude (LOAELs ranging from 6.6 × 10-4 to 3.4 × 10-3 mg/m3), providing a degree of cross-study convergence analogous to the convergence seen in other human-anchored inhalation toxicity assessments.

Why the Two Routes Diverge

The most direct explanation for this route-specific divergence in evidence structure lies in toxicokinetics rather than in any difference in intrinsic hazard: the oral route involves substantial, dose-dependent gastric and hepatic first-pass detoxification of ingested Cr(VI) to the comparatively inert Cr(III) form, a process that varies with individual stomach pH and gastric transit time and that is captured quantitatively by the PBPK modeling described in Section 2 — creating a toxicokinetic buffer between external oral dose and internal target-tissue dose that is well suited to characterization in a controlled rodent bioassay with defined dosing. The inhalation route, by contrast, delivers Cr(VI) directly to the respiratory tract mucosa and, via absorption, to the systemic circulation without a comparable detoxification barrier, and occupational cohorts provide direct, decades-long observation of cumulative respiratory tract and systemic effects in real-world exposure circumstances that are difficult to replicate with equivalent statistical power in a rodent inhalation bioassay of feasible size and duration. Population variability in the oral detoxification barrier — for example, in individuals with chronically elevated stomach pH due to hypochlorhydria, acid-reducing medication use, young age (particularly under 30 months), or older age — provides a specific, identified mechanism by which some human subpopulations could receive a higher internal Cr(VI) dose than the animal-derived, average-adult-calibrated PBPK model would predict, a consideration explicitly incorporated into the assessment’s discussion of susceptible populations.

7. Data Gaps and Areas of Ongoing Research

Several limitations in the current Cr(VI) toxicology database were identified as affecting confidence in specific hazard and dose-response conclusions. Most significant is the unresolved status of female reproductive toxicity: the existing evidence is limited to a single low-confidence epidemiology study and a series of low-confidence animal studies, with no effects (other than a moderate decrease in maternal body weight) identified in the medium- or high-confidence studies available; a well-designed, high-confidence female reproductive toxicity study would meaningfully resolve this uncertainty.

Confidence in the developmental toxicity reference value is limited by the confounding possibility that decreased offspring growth reflects concurrent maternal or systemic toxicity at the high doses where the effect was observed, rather than a direct developmental effect; additional studies designed to separate developmental toxicity from general maternal toxicity at a range of doses bracketing the other organ-specific points of departure would help resolve this.

The hematological effects — while reasonably well characterized quantitatively — remain difficult to interpret because of their apparently transient nature; additional chronic-duration hematological monitoring, ideally with serial measurements over the full course of long-term exposure rather than only at 22 days, would clarify whether these effects persist, adapt, or fluctuate over a full chronic exposure period.

Immune and male reproductive toxicity conclusions are both limited by a predominance of low-confidence studies and inconsistent findings even among the available high-confidence rodent studies; additional standardized, high confidence immunotoxicity and reproductive toxicity testing batteries would help resolve whether the suggestive signals in these areas reflect a true, reproducible hazard or reflect the noise inherent in a literature composed mostly of lower-quality studies.

Toxicokinetic uncertainty below the 1 mg/kg-day dose threshold used to interpret the animal database remains a specific, explicitly acknowledged data gap: at these lower, more environmentally and occupationally relevant doses, there is substantial uncertainty as to whether systemic tissues beyond the GI tract and liver receive consistent Cr(VI) exposure across different studies, and inconsistent pharmacokinetic and toxicological results at these lower doses are considered likely simply as a consequence of this exposure variability rather than as evidence against an effect. Refinement of low-dose PBPK modeling, potentially informed by further human tracer studies analogous to Donaldson and Barreras (1966), would help resolve this uncertainty.

Finally, the role of genetic susceptibility factors — polymorphisms affecting DNA repair capacity and carriage of mutated CFTR alleles, both identified as plausible modifiers of Cr(VI)-induced cancer risk — has not yet been incorporated quantitatively into the dose-response assessment; population-based studies characterizing the prevalence and risk-modifying magnitude of these genetic factors would be a valuable addition to future Cr(VI) risk characterization, particularly for the intestinal carcinogenicity endpoint where a specific, biologically plausible CFTR mediated mechanism has already been identified.

Bibliography