Vinyl Chloride Risk Assessment and Liver Cancer

Prepared for Regulatory and Litigation Support Use RoseyFields — Toxicology & Risk Assessment Practice Group

RoseyFields conducts vinyl chloride risk assessments using the following methodology, which integrates the peer-reviewed toxicological, epidemiological, and pharmacokinetic literature on vinyl chloride (VC) into a defensible framework for quantifying noncancer hazard and cancer risk. This methodology is applied consistently across regulatory support, remediation decision-making, and toxic tort and environmental litigation matters involving VC exposure.

1. Chemical Identity, Regulatory Background, and Exposure

Sources/Pathways

1.1 Chemical Identity

Vinyl chloride (chloroethene; CAS No. 75-01-4) is a two-carbon haloalkene (CH =CHCl, molecular weight 62.5) that is a colorless gas with a mild, sweet odor at room temperature and pressure. It is structurally related to other chlorinated solvents and degradation products, including trichloroethylene and 1,1-dichloroethylene, and is characterized by high vapor pressure and relatively low water solubility, properties that govern its volatilization behavior in environmental media and its physicochemical fate at contaminated sites (Cowfer and Magistro, 1983).

VC is manufactured almost exclusively as a chemical intermediate for the production of polyvinyl chloride (PVC) resin, typically via thermal cracking of 1,2-dichloroethane. Because it is produced, transported, and polymerized as a high-volume industrial monomer, the principal historical hazard setting for VC has been occupational exposure in VC monomer (VCM) production and PVC polymerization facilities (Cowfer and Magistro, 1983).

1.2 Regulatory Background

VC’s regulatory history was shaped decisively by the clinical recognition, in the early 1970s, of clusters of hepatic angiosarcoma — an exceptionally rare tumor — among PVC polymerization workers, a finding that established one of the clearest causal chemical-carcinogen associations in occupational medicine (Creech and Johnson, 1974). That discovery precipitated an extensive body of cohort mortality studies of VC/PVC workers over the following two decades and led federal agencies to regulate VC as an occupational carcinogen and to develop quantitative noncancer and cancer toxicity values for use in ambient, drinking-water, and site-specific risk assessment. RoseyFields’ assessors treat VC as a Group A/“known human carcinogen” compound (see Section 7) and apply the toxicity values summarized in Sections 5–7 to compound-specific hazard and risk calculations.

1.3 Exposure Sources and Pathways RoseyFields’ case-specific exposure assessments typically consider the following sources and pathways, which reflect the environmental and occupational contexts in which VC exposure most commonly arises:

• Occupational exposure in VC monomer production, PVC polymerization, and PVC fabrication facilities, historically via inhalation of VC vapors and, to a lesser extent, dermal contact; this is the exposure context for essentially all of the human epidemiological cohorts underlying the cancer assessment (Fox and Collier, 1977; Jones et al., 1988; Simonato et al., 1991; Wong et al., 1991).

• Groundwater and soil-gas/vapor-intrusion exposure at hazardous waste sites and landfills, where VC is frequently encountered not as a directly disposed contaminant but as a reductive-dechlorination breakdown product of chlorinated solvents such as tetrachloroethylene (PCE) and trichloroethylene (TCE); this pathway is a recurring feature of Superfund and RCRA corrective-action matters.

• Drinking-water ingestion where VC-contaminated groundwater serves as a potable water source, historically a basis for EPA’s oral toxicity assessment (Section 2 sample calculations use a 1 mg/L drinking-water exposure scenario).

• Ambient/fenceline air exposure in communities near VC or PVC manufacturing facilities.

• Indoor air/vapor intrusion into overlying structures from contaminated subsurface sources, an increasingly common exposure pathway in current vapor-intrusion assessments.

RoseyFields’ exposure and dose reconstruction work (Section 8) is tailored to the specific pathway(s) relevant to a given site or claim, drawing on the dose-metric and PBPK tools described in Sections 3–4 to translate site-specific exposure concentrations into internal dose estimates comparable to those underlying the toxicity values themselves.

2. Point-of-Departure (POD) Selection

VC’s dose-response assessment is unusual in that both the noncancer and cancer assessments converge on the liver as the critical target organ, and on a common set of chronic rodent bioassays as the source of quantitative dose-response data. RoseyFields’ selection of principal studies and critical effects follows the weight-of-evidence reasoning summarized below.

2.1 Noncancer POD — Principal Study and Critical Effect

Two related chronic dietary rat studies dominate the noncancer database: a higher-dose lifespan study (Feron et al., 1981) and a lower-dose companion study using the same protocol and laboratory (Til et al., 1991). Both studies reported dose-related hepatocellular alterations, including foci of cellular alteration (clear-cell, basophilic, and eosinophilic foci) and liver-cell polymorphism and cyst formation. Because tumor and preneoplastic responses in the Feron et al. (1981) study occurred across essentially the full dose range tested, that study does not resolve a no-observed-adverse-effect level (NOAEL); the lower-dose Til et al. (1991) study, by contrast, identifies a clear NOAEL/LOAEL pair for liver-cell polymorphism and cysts and is therefore the principal study for both the oral reference dose (RfD) and the inhalation reference concentration (RfC). RoseyFields adopts a NOAEL of 0.13 mg/kg-day and a LOAEL of 1.3 mg/kg-day for hepatocellular polymorphism/cyst formation from this study as the noncancer point of departure.

Supporting inhalation studies — including a subchronic study identifying a NOAEL of 50 ppm for combined hepatocellular and testicular histopathology (Sokal et al., 1980), a study reporting minimal liver histopathologic change at low concentrations (Wisniewska-Knypl et al., 1980), and a study of testicular effects at higher exposure levels (Bi et al., 1985) — identify NOAELs at human-equivalent concentrations at or above those derived from the Til et al. (1991) dietary study once dosimetric adjustment is applied (Section 3), and therefore do not displace it as the principal study.

2.2 Cancer POD — Principal Study and Critical Effect

For the cancer assessment, RoseyFields relies on tumor incidence data — combined liver angiosarcoma, hepatocellular carcinoma, and neoplastic nodules — in female rats from the same Feron et al. (1981) dietary bioassay, and, for the inhalation unit risk, on the well-characterized concentration-response inhalation bioassay of Maltoni et al. (1981; 1984), which exposed rats and mice to a wide range of VC concentrations and followed them for tumor development well beyond the exposure period.

RoseyFields does not rely on the corn-oil gavage bioassays of VC (e.g., single-dose gavage components of the Feron protocol) because vehicle-related hepatotoxicity and peroxisome-proliferative effects of vegetable oil vehicles can independently promote hepatic tumor formation, confounding attribution of the tumor response to VC itself; this is a standard methodological screen RoseyFields applies whenever gavage-in-oil bioassays are proposed as a POD source for a genotoxic hepatocarcinogen.

Three occupational cohort mortality studies — Fox and Collier (1977), Jones et al. (1988), and Simonato et al. (1991) — provide human dose-response information and are used by RoseyFields as a check on, rather than the primary basis for, the animal-derived cancer slope factor, because historical occupational exposure reconstruction in these cohorts carries substantial uncertainty. As discussed in Section 7, the human-data-based risk estimates fall within, and are consistent with, the range derived from the animal bioassays, which supports the animal-based POD.

3. Pharmacokinetic Modeling and Dose Conversion

VC toxicity is mechanistically driven by its oxidative metabolism — principally via cytochrome P450 — to the reactive epoxide chloroethylene oxide, which rearranges to chloroacetaldehyde and forms DNA and protein adducts (Bolt et al., 1977). Because the toxic and carcinogenic moieties are reactive metabolites rather than the parent compound, and because VC metabolism is saturable at higher exposure concentrations, external exposure concentration (ppm or mg/kg-day) is not a reliable surrogate for internal target-tissue dose across the wide range of concentrations spanned by the animal bioassays and human exposure scenarios. RoseyFields therefore relies on physiologically based pharmacokinetic (PBPK) modeling, rather than default dosimetric adjustment factors, to convert applied doses to an internal dose metric before conducting dose-response modeling.

The PBPK model RoseyFields applies for VC compartmentalizes the animal and human body into blood, liver, richly and poorly perfused tissue, and fat compartments, and describes VC metabolism as occurring via two saturable oxidative pathways — a low-capacity, high-affinity pathway and a higher-capacity, lower-affinity pathway — both yielding chloroethylene oxide as an intermediate (Clewell et al., 1995c). This two-pathway structure was needed because single-pathway models could not simultaneously reproduce the observed continued rise in total VC metabolism and glutathione depletion at higher exposure concentrations. The model has been calibrated against gas-uptake and whole-body radiolabeled metabolism data spanning mice, rats, hamsters, monkeys, and controlled human inhalation exposures, providing cross-species validation of the metabolic description (Bolt et al., 1977; Clewell et al., 1995c).

Two internal dose metrics are used, depending on the endpoint being modeled:

• “RISK” metric — total amount of VC metabolized per unit volume of liver — used for liver-specific noncancer and cancer endpoints, on the premise that hepatic tumor and toxicity risk tracks local metabolic production of the reactive metabolite within the liver.

• “AMET” metric — total amount of VC metabolized per unit body weight — used for endpoints without an organ-specific dose metric.

For the noncancer assessment, the model is used to convert the NOAEL/LOAEL from the oral Til et al. (1991) study to a human-equivalent internal dose, and then to back-calculate the continuous human oral or inhalation exposure level that would produce an equivalent internal liver dose metric in a human — yielding NOAEL(HEC) and LOAEL(HEC) values expressed in conventional exposure units (mg/kg-day or mg/m³) that are directly comparable to the animal POD but reflect human physiology and metabolic capacity (Clewell et al., 1995c).

For the cancer assessment, dose-response modeling (Section 7) is conducted directly on the internal dose metric rather than on converted human-equivalent exposure concentrations, consistent with the principle that tissues experiencing equivalent average concentrations of a carcinogenic moiety over a lifetime should be presumed to carry equivalent lifetime cancer risk. The resulting risk-per-unit-internal-dose relationship is then converted to a risk-per-unit-external-exposure value (the slope factor/unit risk) by running the model forward for a specified continuous human exposure scenario (e.g., 1 mg/L in drinking water, or a specified continuous air concentration) and determining the corresponding internal dose metric, which is linear with external exposure in humans across the range of regulatory interest (Clewell et al., 1995c; Reitz et al., 1996).

This PBPK-based approach is generally preferred by RoseyFields over default body-weight or body-surface-area scaling because it directly addresses interspecies differences in absorption, distribution, and — critically for a metabolically activated carcinogen like VC — metabolic capacity, while leaving the pharmacodynamic (tissue-sensitivity) component of interspecies extrapolation to be addressed separately through uncertainty factors (Section 4) (Barton et al., 1998).

4. Uncertainty Factor Derivation

RoseyFields applies the standard EPA-style uncertainty factor (UF) framework to the PBPK-adjusted noncancer POD, selecting factor values based on the specific evidentiary gaps in the VC database rather than applying defaults mechanically.

Uncertainty/Modifying Value Basis Factor Applied

Interspecies (UFA) Reduced from the full default of 10 because PBPK modeling addresses the pharmacokinetic component of interspecies extrapolation; retained at 3 (rather than 1) because the toxicodynamic (tissue-sensitivity) component remains uncertain, and limited evidence suggests possible human susceptibility to VC-related hepatic effects (Ho et al., 1991).

Intraspecies (UFH) Full default retained to protect susceptible human subpopulations, given the absence of data adequate to characterize the range of human variability in VC metabolism and hepatic sensitivity.

Subchronic-to-chronic (UFS) Not applied — principal study (Til et al., 1991) is a lifetime chronic bioassay.

LOAEL-to-NOAEL (UFL) Not applied — a defined NOAEL (rather than only a LOAEL) is available from the principal study.

Database completeness Not applied — the database includes chronic bioassays by two (UFD) routes, adequate reproductive/developmental studies, and mechanistic/genotoxicity data sufficient to characterize the noncancer hazard without an additional database gap factor.

Modifying factor (MF) Not applied — principal study is of high quality, and effects at organs other than the liver occur only at substantially higher exposure levels.

Composite UF 3 (UFA) × 10 (UFH) × 1 × 1 × 1 (MF) = 30, applied identically to both the RfD and the RfC.

This UF structure reflects a deliberate asymmetry that RoseyFields applies consistently across noncancer and cancer assessments for metabolically activated genotoxic carcinogens: because VC’s noncancer hepatic mode of action is less well characterized mechanistically than its genotoxic carcinogenic mode of action, the toxicodynamic uncertainty factor is retained for the RfD/RfC but is not layered onto the PBPK-based cancer slope factor, where human epidemiological risk estimates independently corroborate that humans are not more sensitive than rodents to VC carcinogenicity (Section 7) (Chen and Blancato, 1989; Storm and Rozman, 1997).

5. Derivation of the Oral Reference Dose (RfD) and Reference

Concentration (RfC)

5.1 Oral RfD

Using the PBPK model, RoseyFields converts the animal NOAEL of 0.13 mg/kg-day (Til et al., 1991) to a human-equivalent NOAEL(HEC) of 0.09 mg/kg-day. Dividing by the composite uncertainty factor of 30:

RfD = 0.09 mg/kg-day ÷ 30 = 0.003 mg/kg-day (3 × 10⁻³ mg/kg-day)

5.2 Inhalation RfC

The same principal study and dose metric are used for the RfC, since the Til et al. (1991) oral study — once converted through the PBPK model — yields a lower and more protective human-equivalent concentration than the available inhalation bioassays (e.g., the Sokal et al. (1980) NOAEL(HEC) of 93 mg/m³ for hepatocellular and testicular effects). The oral NOAEL(HEC) converts to an inhalation-equivalent NOAEL(HEC) of 2.5 mg/m³. Dividing by the same composite uncertainty factor of 30:

RfC = 2.5 mg/m³ ÷ 30 = 0.083 mg/m³ ≈ 1 × 10⁻¹ mg/m³

RoseyFields applies these values to characterize noncancer hazard from chronic VC exposure by the oral and inhalation routes, respectively (Section 8), while noting the Agency’s own recognition that, because VC’s carcinogenic effects dominate at exposure levels well below those producing noncancer hepatic effects in most exposure settings, the RfD/RfC are typically of secondary practical significance relative to the cancer risk values in real-world site assessments.

6. Carcinogenicity Classification and Cancer Slope Factor / Unit Risk

Derivation

6.1 Weight-of-Evidence Classification

RoseyFields classifies VC as a known human carcinogen by the oral and inhalation routes, and as highly likely to be carcinogenic by the dermal route, based on the convergence of four independent lines of evidence:

• Consistent epidemiologic evidence of a causal association between occupational VC exposure and hepatic angiosarcoma — an extremely rare tumor with an overwhelming attributable fraction to VC exposure — as well as excess hepatocellular carcinoma, across multiple independent cohorts (Fox and Collier, 1977; Jones et al., 1988; Simonato et al., 1991; Wong et al., 1991).

• Consistent carcinogenicity in rats, mice, and hamsters by both the oral and inhalation routes, with the liver as the shared critical target organ in animals and humans (Feron et al., 1981; Maltoni et al., 1981, 1984).
• A well-characterized genotoxic mode of action: oxidative metabolism to chloroethylene oxide/chloroacetaldehyde, formation of persistent etheno-DNA adducts, and downstream mutation of ras and p53 in both rodent and human VC-associated liver tumors (Swenberg et al., 1992; DeVivo et al., 1994; Marion et al., 1996).

• Efficient absorption of VC by all tested exposure routes and rapid systemic distribution (Krajewski et al., 1980; Bolt et al., 1977).

Because the mode of action is genotoxic and DNA-reactive, RoseyFields applies low-dose linear extrapolation, consistent with standard practice for genotoxic carcinogens, rather than a nonlinear/threshold approach.

6.2 Molecular Mode of Action — Angiosarcoma vs. Hepatocellular Carcinoma

RoseyFields’ case narratives distinguish, where mechanistically relevant, between the two principal VC-associated liver tumor types. Angiosarcoma arises from hepatic sinusoidal cells and is associated with a p53-dependent pathway: p53-null mice spontaneously develop angiosarcoma (among other tumors) but not hepatocellular carcinoma (Donehower et al., 1992), and VC-induced human angiosarcomas show frequent ras mutation together with p53 alteration, a combination with a reported predictive value of roughly two-thirds for liver tumors in exposed workers (Marion et al., 1996). Hepatocellular carcinoma, by contrast, is more closely associated with myc/ras overexpression and is less consistently linked to p53 loss (Greenblatt et al., 1994). Because both pathways are represented in human VC-exposed liver tumors, RoseyFields includes both tumor types (together with neoplastic nodules, treated as adenoma-equivalent) in the combined tumor-incidence dataset used for dose-response modeling, consistent with the approach taken in the underlying bioassay analyses.

6.3 Dose-Response Data and Extrapolation

Oral cancer risk is calculated from the combined incidence of liver angiosarcoma, hepatocellular carcinoma, and neoplastic nodules in female rats in the Feron et al. (1981) dietary bioassay; inhalation cancer risk is calculated from the corresponding tumor incidence in female rats in the Maltoni et al. (1981, 1984) inhalation bioassay. Dose-response modeling is conducted on the PBPK-derived internal liver dose metric (Section 3) rather than on administered dose, using two low-dose extrapolation approaches applied in parallel: the linearized multistage (LMS) model, and a linear extrapolation from the lower 95% confidence bound on the dose associated with 10% extra tumor risk (the LED10 approach). The two methods produce closely comparable potency estimates.

6.4 Oral Slope Factor and Inhalation Unit Risk

RoseyFields applies the following recommended values, derived using the LMS model on the internal dose metric and converted to conventional exposure units as described in Section 3: Toxicity Value Recommended Value

Oral cancer slope factor (adult, lifetime continuous 7.2 × 10⁻¹ per (mg/kg)/day exposure)

Inhalation unit risk (adult, lifetime continuous 4.4 × 10⁻⁶ per µg/m³ (most conservative of the exposure) sex/species-specific estimates spanning 1.0–4.4 × 10⁻⁶ per µg/m³)

Inhalation unit risk, exposure beginning at birth 8.8 × 10⁻⁶ per µg/m³ (twofold early-life adjustment; see Section 6.6)

These animal-based values are corroborated by independent risk estimates derived from the three occupational cohort mortality studies. PBPK-based analyses of the Fox and Collier (1977), Jones et al. (1988), and Simonato et al. (1991) cohorts yield 95% upper-bound inhalation risk estimates ranging from roughly 0.3 × 10⁻⁶ to 2.8 × 10⁻⁶ per µg/m³ (Clewell et al., 1995c), consistent with an earlier independent estimate from the Fox and Collier data (Chen and Blancato, 1989) and with a PBPK-based reanalysis of the Maltoni bioassay data (Reitz et al., 1996). RoseyFields treats this convergence — animal bioassay potency and independent human cohort potency landing in the same order of magnitude — as unusually strong support for the recommended unit risk, and as evidence against greater human sensitivity to VC carcinogenicity than observed in rodents (Storm and Rozman, 1997).

6.5 Cross-Species Extrapolation and the Pharmacodynamic Adjustment

Default cross-species dose scaling (e.g., body-weight³⁄⁴ scaling) rests in part on evidence that smaller animals tolerate higher doses per unit body weight owing to more rapid metabolism (Freireich et al., 1966). Because the PBPK model already accounts for interspecies differences in metabolic rate and other dosimetric parameters, RoseyFields does not additionally apply body-weight³⁄⁴ scaling on top of the PBPK conversion; consistent with recommended practice for chemicals with both PBPK data and a validated dose metric, a default pharmacodynamic adjustment of 1.0 is applied for the cancer assessment (Barton et al., 1998). This is supported by the absence of any epidemiologic signal of greater human carcinogenic sensitivity relative to rodents once dose is expressed on a common internal metric (Chen and Blancato, 1989; Storm and Rozman, 1997), a conclusion reinforced by comparable levels of endogenous background etheno-DNA adduct formation in humans and rats (Swenberg et al., 1992; Morinello et al., 1999).

6.6 Early-Life Susceptibility Adjustment

RoseyFields applies a twofold upward adjustment to cancer risk estimates for exposure scenarios beginning in early life (e.g., residential exposure of children), reflecting evidence of age-dependent sensitivity to VC carcinogenesis from bioassay and mechanistic data (Maltoni et al., 1981) and studies of age-dependent hepatic DNA alkylation and preneoplastic focus induction that show substantially greater responses in young versus adult animals at equivalent VC exposure (this literature is discussed in the source toxicological review at the studies underlying the childhood-susceptibility adjustment). This adjustment is applied as a case-specific modifier to the lifetime unit risk/slope factor rather than as a change to the adult-exposure toxicity values themselves, consistent with current practice for early-life susceptibility adjustments for mutagenic carcinogens.

7. Application in Practice

7.1 Noncancer Hazard Characterization — Hazard Quotient and Hazard Index

For each relevant exposure pathway, RoseyFields calculates a chronic daily intake (CDI) or exposure concentration from site- or case-specific data (environmental sampling results, occupational exposure reconstruction, or biomonitoring data where available) and divides it by the applicable RfD (oral) or RfC (inhalation) to obtain a pathway-specific Hazard Quotient (HQ):

HQ = CDI (or EC) ÷ RfD (or RfC)

An HQ at or below 1.0 indicates that noncancer hazard from that exposure is not expected to be appreciable; an HQ above 1.0 flags a pathway warranting further evaluation, though it is not itself a quantitative risk probability. Where multiple exposure pathways to VC (or to VC together with other liver-toxic co-contaminants) are present, RoseyFields sums pathway-specific HQs into a Hazard Index (HI), consistent with EPA’s chemical-mixtures guidance, subject to an evaluation of whether the endpoints and modes of action being summed are toxicologically comparable (in VC’s case, hepatocellular polymorphism/cyst formation).

7.2 Excess Lifetime Cancer Risk

For cancer risk, RoseyFields calculates pathway-specific excess lifetime cancer risk (ELCR) as the product of the lifetime average daily dose (LADD, oral) or lifetime average exposure concentration (inhalation) and the applicable slope factor or unit risk:

ELCR (oral) = LADD × Oral Slope Factor; ELCR (inhalation) = LAEC × Inhalation Unit Risk

Results are typically expressed against the regulatory risk range of 10⁻⁴ to 10⁻⁶, with the early-life adjustment (Section 6.6) applied where exposure duration includes childhood years, and with pathway-specific ELCRs summed across relevant exposure routes for a cumulative site- or claimant-specific risk estimate.

7.3 Exposure and Dose Reconstruction

In litigation and regulatory matters, the exposure concentration or intake input to the HQ/ELCR calculations above is rarely directly measured for the relevant historical period, particularly in occupational or legacy-contamination settings. RoseyFields’ dose reconstruction work draws on:

• Available environmental monitoring and modeling data (air dispersion modeling, groundwater fate-and-transport modeling, historical industrial hygiene records).
• Job-exposure matrices and historical process/production records, calibrated where possible against the exposure categorization approaches used in the underlying occupational cohort studies (Fox and Collier, 1977; Jones et al., 1988; Simonato et al., 1991).

• The PBPK model (Section 3) to translate reconstructed external exposure estimates into internal dose metrics directly comparable to those underlying the toxicity values, improving the defensibility of dose reconstruction where exposure was intermittent, of variable concentration, or spanned developmental life stages.

7.4 Regulatory and Litigation Support

RoseyFields’ risk assessors use the outputs above to support: (i) baseline and site-specific human health risk assessments under CERCLA/RCRA, including establishment of preliminary remediation goals for soil, groundwater, and indoor air; (ii) drinking-water and ambient-air standard-setting and compliance evaluations; (iii) causation and specific-causation analysis in toxic tort and personal injury litigation, including Daubert/Frye-compliant expert reports that connect reconstructed exposure to the quantitative dose-response relationships summarized in Sections 5–6; and (iv) rebuttal analysis of opposing experts’ toxicity value selection, dose reconstruction assumptions, and uncertainty factor application.

8. Key Uncertainties and Sensitive Populations

8.1 Key Uncertainties

• Oral dose-metric derivation. The oral-to-internal-dose conversion for the principal noncancer study relies on estimates of VC volatilization loss and fecal recovery from the dietary bioassays, introducing dosimetric uncertainty distinct from, and layered on top of, ordinary interspecies extrapolation uncertainty (Feron et al., 1981; Til et al., 1991).

• Mode-of-action asymmetry between noncancer and cancer effects. Because the mechanistic basis for VC’s noncancer hepatic effects is less thoroughly characterized than its genotoxic carcinogenic mode of action, RoseyFields retains a toxicodynamic component in the noncancer interspecies uncertainty factor while applying no comparable adjustment to the cancer slope factor/unit risk (Section 4) — an asymmetry that should be disclosed and defended explicitly in any expert report relying on both values.

• Human epidemiological exposure characterization. All three occupational cohort studies used to corroborate the cancer potency estimate suffer from imperfect historical exposure reconstruction and, in the largest cohort, from potential exposure misclassification across pooled worksites (Simonato et al., 1991); RoseyFields therefore treats the human data as corroborative rather than as the primary quantitative basis for the recommended slope factor/unit risk.

• Oral vs. inhalation potency divergence. The oral slope factor and the inhalation unit risk, when cross-converted using the PBPK model, do not perfectly agree (extrapolating the oral estimate to an inhalation-equivalent unit risk yields a substantially higher value than the recommended inhalation unit risk), a discrepancy attributed largely to the higher incidence of neoplastic nodules in the oral bioassay relative to the inhalation bioassays, potentially reflecting differences in rat strain or dosing route; RoseyFields discloses this divergence rather than reconciling it artificially.

• Preneoplastic and adduct-based dose-response data. Etheno-DNA adduct dose-response data (Swenberg et al., 1992; Morinello et al., 1999) support the plausibility of low-dose linear extrapolation but are not themselves used to quantitatively extend the tumor dose-response curve, because the relationship between adduct burden and tumor risk is not yet quantitatively established.

8.2 Sensitive Populations

• Early-life-exposed individuals. Bioassay and mechanistic evidence of enhanced sensitivity to VC carcinogenesis during early-life exposure (Maltoni et al., 1981) is the basis for the twofold early-life risk adjustment (Section 6.6); this is a first-order consideration in any claim involving childhood residential or in-utero exposure.

• Individuals with pre-existing hepatic disease or hepatotoxicant co-exposure. Because both the noncancer and cancer critical effects are hepatic, individuals with independent hepatic compromise, or concurrent exposure to other hepatotoxic or hepatocarcinogenic agents, may experience an altered exposure-response relationship not captured by the general-population toxicity values; some limited evidence of persistent liver dysfunction in VC-exposed workers (Ho et al., 1991) informed the retention of the toxicodynamic interspecies uncertainty factor (Section 4).

• Workers with high cumulative historical exposure. The occupational cohorts underlying the corroborative human risk estimates, and much of the qualitative hazard characterization for acroosteolysis and other high-exposure clinical findings, reflect legacy exposure conditions no longer representative of current industrial hygiene practice; RoseyFields flags this when the toxicity values are applied to modern occupational exposure scenarios, where actual exposures are typically far lower.

• Populations exposed via multiple concurrent pathways. Communities near legacy contamination sites where groundwater, vapor intrusion, and historical ambient air pathways may operate simultaneously require the cumulative, multi-pathway HQ/HI and ELCR aggregation described in Section 7, rather than single-pathway analysis alone.

References