Cadmium Risk Assessment and Kidney and Lung Cancer Risk

RoseyFields conducts Cadmium risk assessments using the following methodology. This document summarizes the chemical-specific hazard identification, dose-response, and toxicity value derivation approach applied by RoseyFields when cadmium is a chemical of concern at a site or in an exposure reconstruction, litigation support, or regulatory compliance matter. Cadmium presents an unusual and important methodological wrinkle: the governing EPA IRIS noncancer and cancer values date to 1987–1989 and have never been updated, while ATSDR’s 2012 Toxicological Review derived substantially more current, biomarker and benchmark-dose-based Minimal Risk Levels (MRLs) using a modernized version of the same underlying toxicokinetic model. RoseyFields tracks both value sets and explains, for each matter, which is being relied upon and why.

1. Chemical Identity, Regulatory Background, and Exposure Sources/Pathways

Cadmium (Cd; CASRN 7440-43-9) is a naturally occurring heavy metal that exists in the environment exclusively in the +2 oxidation state and does not undergo oxidation-reduction reactions. It occurs naturally in the earth’s crust at concentrations of 0.1–0.5 ppm and is commonly associated with zinc, lead, and copper ores. Cadmium is not degraded in the environment; once released, it persists indefinitely, redistributing among soil, water, and sediment compartments depending on pH and organic matter content, and it bioaccumulates substantially in both plants and the human body, with a biological half-life in humans estimated at 10–30 years.

Regulatory background. EPA’s IRIS assessment for cadmium was developed in two parts: an oral reference dose (RfD) finalized in 1989 and a carcinogenicity assessment finalized in 1987, both based on the 1985 EPA Drinking Water Criteria Document and its 1985 mutagenicity/carcinogenicity addendum. EPA has never evaluated an inhalation reference concentration (RfC) for cadmium and has not updated the oral RfD or the cancer assessment since the 1980s. ATSDR’s 2012 Toxicological Review for Cadmium substantially updated the noncancer methodology, deriving chronic oral, chronic inhalation, and acute inhalation Minimal Risk Levels (MRLs) using benchmark dose (BMD) modeling and a modernized multicompartment toxicokinetic model, applied to a large modern database of environmental biomonitoring studies not available to EPA in the 1980s. RoseyFields treats the ATSDR MRLs as the more scientifically current noncancer values, while continuing to apply the EPA IRIS RfD and cancer values where a matter specifically requires reliance on EPA IRIS (for example, in federal Superfund risk assessments, where IRIS values are the presumptive default).

Exposure sources and pathways. For the general nonsmoking population, diet is the dominant source of cadmium exposure; estimated daily intakes for U.S. nonsmoking adults are approximately 0.35 µg/kg-day (males) and 0.30 µg/kg-day (females), with leafy vegetables, grains, potatoes, peanuts, soybeans, and sunflower seeds contributing disproportionately, and shellfish and organ meats (liver, kidney) representing additional concentrated sources. Cigarette smoking is a major independent and additive source: each cigarette contains approximately 1.7 µg of cadmium, of which roughly 10% is inhaled when smoked, and geometric mean blood cadmium levels in heavy smokers (as high as 1.58 µg/L) run several-fold higher than the U.S. adult population mean (approximately 0.38 µg/L). Ambient air and drinking water are minor contributors for most of the general population, though inhalation exposure becomes a dominant pathway near cadmium-emitting industrial sources (non-ferrous metal mining/refining, phosphate fertilizer manufacturing, fossil fuel combustion, waste incineration) and in occupational settings such as nickel-cadmium battery manufacturing and cadmium smelting, which are also the settings underlying most of the epidemiological dose-response data described below.

2. Point-of-Departure (POD) Selection: Critical Studies and Endpoints

RoseyFields tracks two independently derived, non-interchangeable POD frameworks for cadmium, reflecting the different eras and methodologies of the EPA IRIS and ATSDR assessments.

EPA IRIS oral RfD POD (1989)

The IRIS oral RfD is anchored to a critical renal cadmium concentration of 200 µg Cd/g wet human renal cortex, identified as the highest renal cadmium level not associated with significant proteinuria (U.S. EPA, 1985). Rather than relying on a single epidemiological study, EPA applied an early toxicokinetic model — assuming that 0.01% of the cadmium body burden is eliminated per day, with gastrointestinal absorption of 2.5% from food and 5% from water — to back-calculate the chronic daily oral exposure (NOAEL) that would result in this critical renal concentration: 0.005 mg/kg-day from drinking water and 0.01 mg/kg-day from food. EPA’s IRIS summary describes high confidence in this NOAEL specifically because it is not drawn from any single study but instead synthesizes toxicokinetic parameters (absorption, distribution, elimination) derived from the broad cadmium literature available at the time.

ATSDR chronic oral MRL POD (2012)

ATSDR’s substantially more current approach uses urinary cadmium concentration (µg/g creatinine) — rather than renal tissue concentration — as the internal dose metric, reflecting its status as a well-validated, minimally invasive biomarker that correlates closely with cadmium body burden and kidney cadmium concentration in the dose range below overt nephrotoxicity. ATSDR evaluated three independent approaches to POD selection: (1) a single-study NOAEL/LOAEL approach using Järup et al. (2000), a study of 1,021 individuals living near or working at a Swedish nickel-cadmium battery factory, which identified a urinary cadmium level of 1 µg/g creatinine (recalculated to 0.5 µg/g creatinine per European Chemicals Bureau, 2007) associated with a doubling of abnormal pHC (a low-molecular-weight protein biomarker) prevalence; (2) selection from among eight published benchmark dose analyses, the lowest of which was a BMDL10 of 0.7 µg/g creatinine from Uno et al. (2005), a study of 828 Japanese adults, corroborated by a similar BMDL10 of 0.81 µg/g creatinine from Suwazono et al. (2006); and (3) a meta-analysis pooling dose-response functions across seven environmental exposure studies (spanning European, Japanese, and Chinese populations) to estimate the urinary cadmium level associated with a 10% excess risk of low-molecular-weight proteinuria (UCD10). ATSDR selected the third approach — the meta-analytic UCD10 — because it draws on the full dose response curves from multiple independent studies rather than a single dataset. The resulting POD was the 95% lower confidence limit on the pooled UCD10 from the European studies (the lowest of the three regional estimates), UCDL10 = 0.5 µg/g creatinine — numerically convergent with, but methodologically far more rigorous than, the single-study approach.

Cancer POD

The cancer dose-response assessment relies entirely on human occupational data: a retrospective cohort mortality study of 602 white male cadmium production workers employed for a minimum of six months between 1940 and 1969, followed through 1978, with urinary cadmium data on a subset confirming high historical exposure (Thun et al., 1985). This is the only cadmium cancer dataset EPA considered adequately quantitative, notwithstanding evidence of carcinogenicity in animal inhalation bioassays (Section 6).

3. Pharmacokinetic Modeling and Dose Conversion

Cadmium risk assessment is unusually dependent on multicompartment toxicokinetic modeling because the relevant internal dose metrics (renal cortex concentration, urinary cadmium) cannot be measured directly in living exposed populations at the population scale, and because cadmium’s very long biological half-life means steady-state body burden, not instantaneous exposure, drives toxicity. Both the EPA IRIS and ATSDR frameworks rely on variants of the same foundational model.

The Nordberg-Kjellström model (Kjellström and Nordberg, 1978) is a multicompartment pharmacokinetic model simulating cadmium absorption, distribution among blood/liver/kidney/other-tissue compartments, and urinary excretion over a lifetime of constant exposure. EPA’s 1985/1989 application of this modeling approach (assuming 0.01%/day whole-body elimination and 2.5%/5% oral absorption from food/water) is comparatively simple. ATSDR’s 2012 application substantially modernized the model: age- and sex-specific growth algorithms and organ weights were incorporated (Choudhury et al., 2001), lean-body-mass-based creatinine excretion rates were derived from pooled human data, and sex-specific gastrointestinal absorption fractions of 5% (males) and 10% (females) were applied — reflecting the well-established finding that women absorb a substantially greater fraction of ingested cadmium than men, plausibly related to lower iron stores. The modernized model simulates cadmium accumulation to age 55, the age at which renal cortex cadmium concentration is predicted to peak under constant lifetime intake, and interconverts between three dose metrics: administered oral intake (µg/kg-day), urinary cadmium excretion (µg/g creatinine), and renal cortex tissue concentration (µg/g wet tissue).

For the inhalation pathway specifically, ATSDR paired the Nordberg-Kjellström model with the ICRP Human Respiratory Tract Model (ICRP, 1994), which simulates deposition, mucociliary clearance, and absorption of inhaled cadmium particles as a function of particle size (activity median aerodynamic diameter) and cadmium compound solubility class (oxides/hydroxides classified as slow-dissolving; sulfides/halides/nitrates as moderate; chloride salts as fast-dissolving). This two-model combination allowed ATSDR to predict, for a given inhaled air concentration and particle characteristics, the resulting steady-state urinary cadmium level — critically, after also accounting for the substantial background dietary cadmium contribution, since inhalation exposure in the real world is never the sole cadmium source. RoseyFields uses this same combined-pathway modeling logic when reconstructing exposure at sites with both air and dietary/soil cadmium sources, since treating the pathways independently would misstate the internal dose driving renal risk.

4. Uncertainty Factor Derivation

Because the EPA IRIS and ATSDR frameworks were developed decades apart using different methodologies, their uncertainty factor derivations differ substantially in both magnitude and rationale.

EPA IRIS oral RfD

EPA applied a single composite uncertainty factor of 10, described as accounting solely for intrahuman variability in the toxicity of cadmium in the absence of chemical-specific data on sensitive individuals. No interspecies factor was applied (the POD is human-derived), no factor was applied for LOAEL-to-NOAEL extrapolation (a NOAEL was used), and no explicit database uncertainty factor was applied, consistent with IRIS’s own statement that cadmium’s toxicological database was unusually large and well-characterized even in 1989.

ATSDR chronic oral MRL

ATSDR applied a composite uncertainty factor of 3 for human variability to the POD (0.33 µg/kg-day, the female-specific dietary intake corresponding to the UCDL10). This factor is notable for its specific, named rationale: concern that individuals with diabetes may be especially sensitive to cadmium-induced renal toxicity, combined with the fact that several of the underlying environmental exposure studies specifically excluded diabetics from their analysis populations, meaning the pooled dose-response data may not adequately characterize this subgroup’s sensitivity.

ATSDR chronic inhalation MRL

ATSDR applied a composite factor of 9: an uncertainty factor of 3 for human variability (again reflecting diabetic sensitivity) and a modifying factor of 3 to account for the lack of adequate human data comparing the relative sensitivity of the respiratory tract versus the kidney as target organs for inhaled cadmium — a gap specific to the inhalation route, since the respiratory tract is a target unique to that exposure pathway.

ATSDR acute inhalation MRL

ATSDR applied a substantially larger composite factor of 300 to the LOAELHEC of 0.01 mg Cd/m³ derived from a 2-week rat inhalation study (NTP, 1995): a factor of 10 for use of a LOAEL rather than a NOAEL (no NOAEL was identified in the study), a factor of 3 for animal-to-human extrapolation (already partially addressed through dosimetric adjustment using EPA’s regional deposited dose ratio methodology), and a factor of 10 for human variability.

5. Derivation of the Toxicity Values

The final noncancer values are summarized in the table above. EPA’s oral RfD is bifurcated by exposure medium — 5 × 10⁻⁴ mg/kg-day for water-based exposure and 1 × 10⁻³ mg/kg-day for food-based exposure — reflecting the differential gastrointestinal absorption assumed for the two media (5% from water versus 2.5% from food) even though the underlying critical renal concentration and elimination rate are identical. RoseyFields applies the water-based (more protective, lower) RfD by default for drinking-water pathway assessments and the food-based RfD for dietary/soil-ingestion pathway assessments, consistent with EPA’s own bifurcated presentation.

ATSDR’s chronic oral MRL of 0.1 µg/kg-day (1 × 10⁻⁴ mg/kg-day) is numerically lower than — and thus more health-protective than — EPA’s food-based oral RfD by an order of magnitude, and modestly more protective than the water-based RfD as well. This divergence is attributable primarily to the difference in POD derivation (biomarker-based BMD meta-analysis of modern low-dose environmental data, versus a 1980s renal-tissue concentration toxicokinetic back-calculation) rather than to any difference in the underlying toxicological understanding of cadmium’s mechanism or target organ. RoseyFields’ default practice, absent a jurisdiction specific requirement to use EPA IRIS exclusively, is to apply the ATSDR chronic oral MRL as the operative noncancer oral toxicity value, given its more current and more rigorously documented derivation, while disclosing the EPA IRIS RfD as the applicable federal Superfund default where relevant.

For inhalation, EPA IRIS has never evaluated an RfC; ATSDR’s chronic inhalation MRL of 0.01 µg Cd/m³ is therefore the only available chronic noncancer inhalation toxicity value from either federal source, and RoseyFields applies it as the default for inhalation pathway noncancer risk characterization. ATSDR’s acute inhalation MRL of 0.03 µg Cd/m³ additionally supports short-duration exposure scenarios (e.g., acute release events) not addressed by either the chronic MRL or the EPA IRIS framework.

6. Carcinogenicity Classification and Cancer Slope Factor / Inhalation Unit Risk Derivation

EPA classifies cadmium as Group B1, a probable human carcinogen, based on limited but consistent evidence from occupational epidemiologic studies together with sufficient evidence of carcinogenicity in rats and mice by inhalation and by intramuscular/subcutaneous injection. The human evidence is anchored to the Thun et al. (1985) cohort of 602 cadmium production workers, which found an approximately two-fold excess risk of lung cancer; the Carcinogen Assessment Group’s evaluation concluded that the increased risk was probably not attributable to concurrent arsenic exposure or smoking, though EPA characterized the evidence as “limited” rather than definitive because observed standardized mortality ratios were relatively low and a clear causal relationship specific to cadmium could not be established with certainty. Three additional occupational studies found excess lung cancer risk but were compromised by co-exposure to other known carcinogens (arsenic) or small population size. Four studies also found a statistically significant positive association between cadmium exposure and prostate cancer, though case numbers were small in each and a later update of one underlying cohort (Thun et al., 1985, updating Lemen et al., 1976) did not confirm the excess prostate cancer risk.

The animal carcinogenicity evidence includes a chronic inhalation bioassay in which Wistar rats exposed to cadmium chloride at 12.5, 25, or 50 µg/m³ for 18 months (with a 13-month post-exposure observation period) showed significant, concentration-related increases in lung tumors (Takenaka et al., 1983), together with injection-site and distant-site (including testicular) tumors reported across multiple studies using intramuscular or subcutaneous administration of various cadmium compounds. Notably, seven oral carcinogenicity studies in rats and mice using cadmium salts (acetate, sulfate, chloride) showed no evidence of a carcinogenic response by that route — the basis for EPA’s conclusion that no oral cancer slope factor can be derived for cadmium.

EPA derived the inhalation unit risk (IUR) of 1.8 × 10⁻³ per µg/m³ directly from the human Thun et al. (1985) cohort data, using a two-stage extrapolation model (with only the first stage affected by exposure) applied to observed versus expected lung/trachea/bronchus cancer deaths across three cumulative exposure strata. EPA explicitly considered an alternative IUR derived from the Takenaka et al. (1983) animal bioassay, which yielded a substantially higher (more conservative) unit risk of 9.2 × 10⁻² per µg/m³ — roughly 50-fold greater than the human-data-based value — but elected to rely on the human-data-based IUR because of concerns about species differences in response and differences in the physical/chemical form of cadmium tested (cadmium salt in the animal study versus cadmium fume and cadmium oxide in the occupational human exposure). EPA cautions that the IUR should not be applied above an air concentration of 6 µg/m³, since the dose-response relationship may not remain linear above that level. No oral slope factor is available; RoseyFields does not attempt to derive one via route-to-route extrapolation given EPA’s explicit finding that the oral carcinogenicity database is negative.

7. Practical Application: Hazard Quotients, Excess Lifetime Cancer Risk, Dose Reconstruction, and Litigation Support

Hazard Quotient / Hazard Index

RoseyFields calculates a hazard quotient (HQ) as the estimated chronic daily intake divided by the applicable toxicity value — the ATSDR chronic oral MRL (0.1 µg/kg-day) or chronic inhalation MRL (0.01 µg/m³) by default, or the EPA IRIS oral RfD where required by a governing regulatory framework. Because cadmium’s noncancer critical effect (renal tubular dysfunction) is not unique to cadmium among heavy metals of environmental concern, RoseyFields incorporates cadmium HQs into a renal-effect-specific hazard index when a site involves co-exposure to other nephrotoxic metals.

Excess Lifetime Cancer Risk (ELCR)

For inhalation pathways, RoseyFields calculates ELCR as the estimated chronic daily air concentration multiplied by the inhalation unit risk (1.8 × 10⁻³ per µg/m³), subject to EPA’s caution that this linear unit risk should not be applied above 6 µg/m³. No oral pathway cancer risk calculation is performed for cadmium, consistent with the absence of an oral slope factor; where oral cadmium exposure is a concern at a site, RoseyFields characterizes it exclusively through the noncancer HQ framework.

Exposure and Dose Reconstruction

Cadmium offers an unusually well-developed basis for biomarker-driven dose reconstruction in litigation and historical exposure matters, because urinary cadmium (µg/g creatinine) is a validated, minimally invasive biomarker that reflects cumulative body burden rather than instantaneous exposure, and because the Nordberg-Kjellström model (as modernized by ATSDR) provides a validated quantitative bridge between measured urinary cadmium, historical daily intake, and target-organ (renal cortex) tissue concentration. Where a claimant has contemporaneous or historical urinary or blood cadmium measurements, RoseyFields uses this modeling framework to reconstruct an estimated historical daily intake and compare it directly to the governing toxicity value, rather than relying solely on environmental sampling and exposure-factor assumptions. Because cadmium accumulates over decades and peak renal concentration is predicted at approximately age 55 under constant exposure, RoseyFields’ dose reconstructions explicitly account for a claimant’s age at the time of measurement relative to this accumulation curve.

Litigation and Regulatory Support

Cadmium matters most commonly arise in occupational settings (battery manufacturing, smelting, electroplating, pigment production) and at sites with historical industrial contamination of soil or groundwater. Because EPA’s governing IRIS values are now more than three decades old while ATSDR’s 2012 MRLs reflect a substantially more rigorous and current dose-response methodology, RoseyFields’ expert reports explicitly document both value sets, the derivation methodology and vintage of each, and RoseyFields’ basis for selecting one over the other in a given matter — a disclosure that is often material to admissibility challenges given the stark methodological gap between the two frameworks for the same chemical.

8. Key Uncertainties and Sensitive Populations

RoseyFields identifies several uncertainties and sensitive-population considerations relevant to applying these values in a specific matter.

• Diabetics are explicitly identified by ATSDR as a plausibly more sensitive subpopulation for cadmium induced renal toxicity, and were excluded from several of the environmental exposure studies underlying the chronic MRL POD — the specific rationale for ATSDR’s human-variability uncertainty factor. Site-specific risk assessments involving a population with an elevated diabetes prevalence should consider this uncertainty explicitly rather than assuming the standard uncertainty factor is fully protective.
• Sex differences in gastrointestinal absorption (approximately 5% in males versus 10% in females, per ATSDR’s modernized model) mean that women are expected to accumulate a higher body burden than men at identical dietary intake, a consideration relevant to sex-stratified exposure assessments and consistent with observed higher blood/urinary cadmium levels in women in NHANES biomonitoring data.
• Children and early-life exposure represent a distinct concern because cadmium is a cumulative, non eliminated toxicant with renal concentration rising roughly linearly from near-zero at birth to a peak between ages 50 and 60; limited but suggestive human and animal evidence indicates that individuals exposed to elevated cadmium beginning in childhood may show greater renal biomarker abnormalities in later adulthood than those first exposed as adults, even at comparable adult urinary cadmium levels, and gestational/lactational exposure has produced measurable renal effects in animal offspring.
• Postmenopausal women and the elderly are a recognized sensitive population for cadmium’s skeletal effects (osteoporosis, decreased bone mineral density, increased fracture risk), historically documented in the Itai-Itai disease outbreaks in cadmium-polluted regions of Japan, though the etiology of Itai-Itai disease also involved multiparity and nutritional deficiencies (calcium, vitamin D, protein) that complicate attribution to cadmium alone.
• Species differences complicate reliance on animal data for cadmium risk assessment: rats are obligate nose-breathers (unlike humans, who breathe through both nose and mouth), producing different lung deposition patterns for inhaled cadmium particles; metallothionein synthesis and inducibility — a key determinant of cadmium sequestration and detoxification — differs by species, strain, sex, and target organ; and the comparatively short lifespan of standard rodent models limits their ability to fully capture the decades-long accumulation dynamics that govern human cadmium toxicity, particularly for bone and renal endpoints that manifest only after prolonged, low-level exposure.
• The roughly 50-fold divergence between the human-data-based inhalation unit risk (1.8 × 10⁻³ per µg/m³) and the alternative animal-bioassay-based unit risk (9.2 × 10⁻² per µg/m³, from Takenaka et al., 1983) is a significant, EPA-acknowledged source of uncertainty in the cancer risk value; RoseyFields discloses this divergence and EPA’s stated rationale for preferring the human-derived value (species differences in response, and differences between the cadmium salt tested in animals versus the cadmium fume/oxide form of actual occupational exposure) in any matter where the IUR materially affects the risk conclusion.

References