Vinyl Chloride Toxicology and Liver Cancer

CAS No. 75-01-4 RoseyFields — Chemical-Specific Toxicology Page

RoseyFields evaluates the toxicology of vinyl chloride using the following evidence base, which draws on the peer-reviewed animal toxicological, mechanistic, and human epidemiological literature to characterize the hazard profile of vinyl chloride (VC) across organ systems, exposure routes, and life stages.

1. Chemical Identity and Physicochemical Properties Relevant to Toxicity

Vinyl chloride (chloroethene, chloroethylene, monochloroethene; CAS No. 75-01-4) is a two-carbon haloalkene, CH =CHCl, with a molecular weight of 62.5. At room temperature and pressure it is a colorless gas with a mild, sweet odor; it has a high vapor pressure (approximately 2,660 mm Hg at 25°C) and is only sparingly water soluble, with reported solubility values of roughly 1,100–2,760 mg/L (Sax and Lewis, 1989; Cowfer and Magistro, 1983).

These properties are directly relevant to VC’s toxicological behavior. High volatility and low water solubility mean that VC partitions rapidly from water and soil into air, favoring inhalation as the dominant environmental exposure route and contributing to its behavior as a vapor-intrusion contaminant at sites where it arises from reductive dechlorination of other chlorinated solvents. Its modest octanol-water partition coefficient (log K of 1.36; NIOSH, 1986) is not indicative of a ow bioaccumulative substance, and this is corroborated toxicokinetically: VC is not stored in adipose or other tissue to an appreciable degree and instead is rapidly metabolized and eliminated (Section 2), so bioaccumulation and biomagnification are not significant concerns for VC as they can be for more lipophilic halogenated compounds.

VC itself is chemically stable enough to be transported and stored as a compressed gas or refrigerated liquid, but it is far more reactive metabolically than it is abiotically: its toxicological and carcinogenic hazard derives almost entirely from oxidative bioactivation to a reactive epoxide (Section 3–4), not from intrinsic reactivity of the parent molecule with biological macromolecules. This combination — environmentally mobile and persistent as a parent compound, but a metabolically activated genotoxicant once absorbed — is central to how RoseyFields frames VC’s hazard profile: environmental persistence and human health persistence (bioaccumulation) are not coupled for this chemical, but metabolic activation drives essentially all of its toxicologically significant endpoints.

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

2.1 Absorption VC is rapidly and efficiently absorbed by inhalation in both humans and animals. In five young adult male volunteers exposed to 7.5–60 mg/m³ VC by inhalation, approximately 42% of inhaled VC was retained on average, with maximum retention reached within 15 minutes and retention independent of the inspired concentration, although individual variation was substantial (30%–71% retention after 6 hours at 30 mg/m³) (Krajewski et al., 1980). Following cessation of exposure, expired-air VC concentrations fell rapidly, to roughly 4% of the inhaled concentration within 30 minutes (Krajewski et al., 1980). Animal inhalation studies confirm rapid pulmonary uptake: male Wistar rats exposed to 1,000–7,000 ppm VC for 5 hours reached equilibrium blood levels within 30 minutes, with blood levels falling to near-undetectable within 2 hours after exposure ended (Withey, 1976).

No human data on oral absorption of VC were located, but animal data indicate rapid, essentially complete oral absorption. Peak blood VC levels were reached within 10 minutes after aqueous gavage dosing in rats (up to 92 mg/kg), while gavage dosing in oil produced more delayed absorption, with peak blood levels at about 40 minutes (Withey, 1976). At high oral doses (100 mg/kg in oil), unmetabolized VC was still detectable in exhaled air 72 hours post-dosing, indicating saturation of metabolic capacity at that dose (Watanabe and Gehring, 1976; Watanabe et al., 1976a). In a dietary study in which VC was administered as residual monomer in PVC powder, only 8%–17% of the ingested VC was recovered unabsorbed in feces (as VC still enclosed within PVC granules) across doses of 2.3–21.2 mg/kg-day, supporting the assumption that free VC monomer is nearly completely absorbed from the gastrointestinal tract (Feron et al., 1981).

2.2 Distribution

VC distributes rapidly and widely throughout the body following absorption by either route, but its storage in tissue is limited by its own rapid metabolism and excretion (Bolt et al., 1977). No evidence indicates that VC or its metabolites accumulate to a meaningful extent in fat or other tissue depots, consistent with its low log K and rapid clearance kinetics. ow

2.3 Metabolism

VC toxicity and carcinogenicity are attributable almost entirely to its oxidative metabolism, not to the parent compound. The primary metabolic step is cytochrome P450 (CYP)-mediated oxidation of VC to chloroethylene oxide (CEO), a highly reactive, short-lived epoxide (Bolt et al., 1977). A substantial fraction of CEO rearranges non-enzymatically to chloroacetaldehyde (CAA), another reactive electrophile, while the remainder is a substrate for epoxide hydrolase (Pessayre et al., 1979). Both CEO and CAA are detoxified principally through conjugation with glutathione (GSH) (Jedrychowski et al., 1985; Tarkowski et al., 1980; Leibman, 1977), reflected in dose-dependent depletion of nonprotein sulfhydryl (largely GSH) levels at higher VC exposure concentrations (Jedrychowski et al., 1985; Tarkowski et al., 1980) and in urinary excretion of GSH-derived conjugates, including N-acetyl-S-(2-hydroxyethyl)cysteine and thiodiglycolic acid, two of the three major urinary metabolites of VC identified in rats (Watanabe et al., 1976b; Hefner et al., 1975; Green and Hathway, 1975).

At low, environmentally and occupationally relevant concentrations, metabolism in both animals and humans is dominated by the high-affinity, low-capacity isozyme CYP2E1 (Guengerich et al., 1991). At higher exposure concentrations, once CYP2E1 becomes saturated, other, lower-affinity P450 isozymes contribute to metabolism; phenobarbital pretreatment (a CYP2B1 inducer) increases VC metabolism substantially (Ivanetich et al., 1977), and P450 induction is necessary to produce acute VC hepatotoxicity in rats, directly implicating metabolic activation rather than the parent compound as the proximate toxicant (Jaeger et al., 1977). At high exposure concentrations, VC metabolism also destroys P450 enzyme itself, an effect potentiated by prior induction (Reynolds et al., 1975; Guengerich and Strickland, 1977), consistent with reactive intermediates damaging the very enzyme responsible for their formation.

This saturable, capacity-limited metabolism is the single most consequential toxicokinetic feature of VC for interpreting dose-response data: because the reactive, DNA- and protein-reactive metabolites (not parent VC) drive toxicity, external exposure concentration is not a linear surrogate for internal target-tissue dose once metabolic saturation is approached, and dose-response relationships derived from high-dose animal bioassays cannot be extrapolated to low-dose human exposure scenarios using simple linear scaling of administered dose. This is the toxicokinetic basis for RoseyFields’ reliance on physiologically based pharmacokinetic (PBPK) modeling, rather than administered-dose comparisons, whenever cross-study or cross-species dose-response comparisons are required.

2.3.1 Species and Sex Differences in Metabolism

The specific complement of P450 isozymes metabolizing VC has not been unequivocally established, but comparative work on the closely related chlorinated solvent trichloroethylene (TCE) is instructive as to the likely pattern for VC: in male Wistar rats and male B6C3F1 mice, CYP2E1 contributes more to low-concentration metabolism in mice than in rats, whereas CYP2C11/6 — a constitutive, noninducible isozyme present only in male rodents — contributes more in rats than in mice, and CYP1A1/2 contributes to uninduced metabolism in mice but not rats (Nakajima et al., 1993). Because CYP2C11/6 is male-rodent-specific, this implies a plausible sex difference in higher-capacity, lower-affinity (“non-2E1”) VC metabolism in rodents that has no clear human correspondent, since some of the relevant low-affinity, high-capacity rodent P450 isozymes lack human orthologs altogether (Guengerich, 1987). This is a key source of species-extrapolation uncertainty at higher, metabolism-saturating exposure concentrations, though it is of less consequence at the lower, CYP2E1-dominated exposure levels most relevant to chronic environmental and occupational risk assessment.

2.4 Excretion

The route of VC elimination is both dose- and route-of-administration-dependent. At low doses via oral or inhalation exposure, GSH-conjugate-derived metabolites are excreted predominantly in urine; once the capacity for metabolism is saturated at higher doses, unmetabolized VC is increasingly eliminated by exhalation (Watanabe et al., 1976b; Watanabe and Gehring, 1976). The route of administration itself also affects elimination: urinary excretion predominates after oral or intraperitoneal dosing, whereas approximately 99% of an intravenously administered dose is exhaled unchanged, likely reflecting the high peak blood concentration achieved with intravenous dosing combined with VC’s relatively low blood:air partition coefficient (Bolt, 1978).

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

3.1 Carcinogenic Mode of Action

VC’s carcinogenic mode of action is genotoxic and reasonably well characterized mechanistically. CEO is generally regarded as the principal carcinogenic metabolite: it is directly mutagenic in bacterial reverse-mutation assays without exogenous metabolic activation (Bartsch et al., 1975; Rannug et al., 1976), is roughly 400-fold more potent as a bacterial mutagen than CAA (Perrard, 1985), and is tumor-initiating in mouse skin initiation-promotion assays, while CAA is negative in the same assay (Zajdela et al., 1980). A structurally informative comparator, 2,2′-dichlorodiethylether — a precursor of CAA but not CEO — fails to produce either preneoplastic hepatocellular foci or hepatic DNA adducts in rats under conditions where VC itself produces both, further supporting CEO as the operative carcinogenic species (Gwinner et al., 1983; Bolt, 1986).

CEO and CAA react with DNA to form a family of etheno-bridged adducts. Although 7-(2-oxoethyl)guanine (OEG) accounts for roughly 98% of total VC-derived DNA adducts, it is rapidly repaired and does not appear to cause replication miscoding, and is therefore not considered central to carcinogenesis (Swenberg et al., 1992). VC carcinogenicity is instead attributed to four minor but far more persistent etheno adducts (Fedtke et al., 1990; Dosanjh et al., 1994) that can cause defective transcription (Singer et al., 1987) and, for ethenoguanine specifically, produce G→A transition mutations in bacterial systems (Cheng et al., 1991). Persistence of these adducts reflects the balance of formation and repair rates (Singer, 1985); in humans all four appear to be repaired by a common DNA glycosylase but at markedly different rates, with ethenoguanine repair notably slower (Dosanjh et al., 1994).

More sensitive adduct measurement showed that apparent adduct “persistence” partly reflects ongoing endogenous adduct formation independent of VC exposure: humans and rats both carry measurable endogenous N²,3-ethenoguanine, at somewhat higher levels in humans, and dose-related increases in this adduct in VC-exposed rats show a steep dose-response at lower concentrations with relative flattening at higher concentrations, supporting low-dose linear extrapolation of cancer risk (Swenberg et al., 1999; Morinello et al., 1999). Hepatic sinusoidal cells — the cell of origin for angiosarcoma — accumulate roughly three times the adduct level of hepatocytes despite having little CYP2E1 of their own, an effect explained largely by markedly lower expression of the relevant DNA-repair glycosylase in sinusoidal cells, suggesting limited repair capacity, more than local metabolic activation, drives their selective vulnerability (Swenberg et al., 1999).

At the molecular level, VC-associated liver tumors show two at least partly distinct mutational routes. Hepatic angiosarcoma is associated with a p53-dependent pathway: p53-null mice spontaneously develop angiosarcoma and lymphoma but not hepatocellular carcinoma (Donehower et al., 1992), and VC-exposed workers show p53 mutation with a clear positive dose-response relationship to cumulative exposure (Smith et al., 1998; Luo et al., 1999), with VC-associated human angiosarcomas carrying specific A→T transversions at p53 codons 179, 249, and 255 not seen in angiosarcomas unrelated to VC exposure (Chen et al., 1999; Soini et al., 1995). Hepatocellular carcinoma, by contrast, is more closely linked to myc/ras oncogene overexpression (Sandgren et al., 1989) and is not consistently associated with p53 loss (Greenblatt et al., 1994), and ras mutation is itself frequently observed in VC-associated human angiosarcoma, with combined mutant-ras/mutant-p53 status carrying a reported predictive value of roughly two-thirds for liver tumors in VC-exposed workers (DeVivo et al., 1994; Marion et al., 1996; Trivers et al., 1995).

3.2 Noncancer Mode of Action

The mechanistic basis for VC’s noncancer hepatic effects is less well resolved than for its carcinogenicity. Both CEO and CAA can react with cellular nucleophiles, but in vitro work using rat liver microsomes indicates the two metabolites bind preferentially to different macromolecular targets: inhibiting alcohol dehydrogenase (which detoxifies CAA) reduces metabolite binding to protein, while inhibiting epoxide hydrolase (which detoxifies CEO) reduces binding to DNA (Guengerich et al., 1981) — consistent with CEO as the principal DNA-reactive species and CAA as a larger contributor to protein-adduct-mediated, non-genotoxic hepatotoxicity. Because both endpoints depend on the same upstream oxidative metabolism, RoseyFields treats the liver’s noncancer and cancer dose-response relationships as mechanistically linked but not identical, with correspondingly different confidence in mode-of-action characterization for each.

4. Animal Toxicological Evidence by Organ System

4.1 Hepatic

The liver is the dominant noncancer (and cancer) target organ for VC across every well-designed chronic study. In a lifetime dietary study in Wistar rats (bioavailable doses of 0, 1.7, 5.0, or 14.1 mg/kg-day, plus a 300 mg/kg-day gavage-in-oil group), a broad spectrum of dose-related hepatic lesions was observed — clear-cell, basophilic, and eosinophilic foci, neoplastic nodules, necrosis, cysts, and liver-cell polymorphism — with several endpoints significant at the lowest dietary dose tested, and relative liver weight significantly increased at the highest dose (Feron et al., 1981). In a related, lower-dose lifetime dietary study (0, 0.014, 0.13, or 1.3 mg/kg-day), basophilic foci were significantly increased at doses as low as 0.014 mg/kg-day, and liver-cell polymorphism and multiple hepatic cysts were significantly increased at 1.3 mg/kg-day, with a NOAEL of 0.13 mg/kg-day for polymorphism/cysts (Til et al., 1991).

Subchronic and chronic inhalation studies corroborate the hepatic target and establish comparable sensitivity by that route. In male Wistar rats exposed to 50, 500, or 20,000 ppm VC for 10 months, relative liver weight was increased at 500 and 20,000 ppm, and hepatocellular polymorphism and sinusoidal-cell proliferation were significantly increased at those concentrations, with a NOAEL of 50 ppm (Sokal et al., 1980); a closely related study found ultrastructural hepatocellular changes (smooth endoplasmic reticulum proliferation, lipid droplet accumulation) at concentrations as low as 50 ppm, with a minimal LOAEL of 50 ppm for minor liver histopathology (Wisniewska-Knypl et al., 1980). Multi-species subchronic inhalation exposure (rats, guinea pigs, rabbits, dogs; 50–500 ppm) identified rats and rabbits as more sensitive than guinea pigs or dogs, with increased relative liver weight at 100–200 ppm and centrilobular hepatic degeneration/necrosis at 200–500 ppm, and a NOAEL of 50 ppm (Torkelson et al., 1961). Shorter-duration, high-concentration inhalation exposure (15,000 ppm) in male Sprague-Dawley rats produced early biochemical evidence of hepatocellular injury — decreased microsomal glucose-6-phosphatase activity, increased glucose-6-phosphate dehydrogenase activity, and progressive nonprotein sulfhydryl elevation — consistent with mild acute liver injury (Du et al., 1979).

A two-generation inhalation reproduction study (CD rats; 0, 10, 100, or 1,100 ppm) found dose-related increases in parental liver weight and centrilobular hypertrophy, together with altered hepatocellular foci (basophilic, acidophilic, clear-cell) that were more frequent in the second (P2) generation — exposed in utero and through the full postnatal period — than in the first (P1) generation exposed only as adults, with a parental NOAEL of 10 ppm for hepatocellular foci. Because the P2 generation was also exposed longer overall, this study alone cannot cleanly separate a true early-life-susceptibility effect from a simple cumulative-dose effect.

4.2 Reproductive and Developmental

VC developmental toxicity in animals occurs only at maternally toxic exposure concentrations, and there is no consistent evidence of teratogenicity. In an inhalation developmental study across three species (mice, rats, rabbits; 50–2,500 ppm depending on species), mice were most sensitive: 500 ppm produced maternal toxicity and fetotoxicity (increased resorption, reduced fetal weight and litter size, retarded ossification) but no teratogenic effect, while rats showed only maternal body-weight effects at 500 ppm and rabbits showed no toxicity at either dose (John et al., 1977), yielding a mouse NOAEL of 50 ppm and a rabbit NOAEL of 2,500 ppm. A separate rat study (continuous exposure to 1,500 ppm during discrete gestational windows) found VC neither teratogenic nor embryotoxic when administered during the second or final third of gestation, with embryotoxicity and fetal mortality confined to first-trimester exposure (Ungvary et al., 1978).

The two-generation inhalation reproduction study described above (Section 4.1) found no adverse effect of VC exposure (up to 1,100 ppm) on parental fertility, reproductive performance, or F1/F2 pup survival or growth, yielding a reproductive NOAEL of >1,100 ppm — substantially higher than the hepatic NOAEL from the same study. In a dominant lethal assay, male rats exposed to 50–1,000 ppm VC for 11 weeks showed no evidence of pre- or post-implantation loss when mated to untreated females, indicating VC does not produce dominant lethal (germinal) mutations, although reduced male fertility was observed at 250 and 1,000 ppm (Short et al., 1977).

4.3 Testicular/Male Reproductive

Testicular effects have been identified in two independent subchronic inhalation studies, generally at exposure levels above those producing hepatic effects. In Wistar rats exposed to 10, 100, or 3,000 ppm VC for up to 12 months, incidence of seminiferous tubule damage (cellular alteration, degeneration, necrosis) increased in a concentration-related manner, reaching statistical significance at 100 and 3,000 ppm; the NOAEL for testicular degeneration in this study (10 ppm) was also the LOAEL for hepatic weight changes, indicating the liver is the more sensitive target (Bi et al., 1985). In the Sokal et al. (1980) study described above, damage to the spermatogenic epithelium was significantly increased at 500 ppm, with a testicular NOAEL of 50 ppm, again above the hepatic NOAEL in the same study.

4.4 Neurological

Neurological effects have been documented primarily at very high, largely narcotic/anesthetic-range exposure concentrations. Male Wistar rats exposed to 30,000 ppm VC for 4 hours/day, 5 days/week for 10 months showed decreased responsiveness to external stimuli and disturbed equilibrium, with histopathology at 12 months revealing diffuse gray- and white-matter degeneration, atrophied nerve fibers, pronounced cerebellar Purkinje cell degeneration, and fibrous infiltration around peripheral nerve endings (Viola, 1970). These animal findings parallel occupational clinical reports of headache, drowsiness, dizziness, ataxia, and loss of consciousness at presumably high (periodic) VC exposures (Langauer-Lewowicka et al., 1983), and of acute central nervous system symptoms (nausea, dizziness) in human volunteers exposed to ≥ 12,000 ppm for 5 minutes (Lester et al., 1963).

4.5 Renal

Renal findings in animal studies have been inconsistent and generally secondary in severity to hepatic and testicular effects. In the Bi et al. (1985) study described above, relative kidney weight was transiently increased at some time points in the highest exposure group (3,000 ppm) but showed no consistent concentration- or duration-related pattern, and no renal histopathology accompanied these weight changes; RoseyFields does not treat the kidney as a robust independent target organ for VC based on the available animal database.

4.6 Cardiovascular / Hematologic / Other Organ Weight Changes

Several chronic and subchronic studies report isolated, non-dose-consistent changes in relative spleen or heart weight (e.g., Sokal et al., 1980), but these lack corroborating histopathology or a clear concentration-response relationship and are not considered adverse, treatment-related effects on the cardiovascular or hematologic systems. No treatment-related effects on hematology, clinical chemistry, or urinalysis parameters were observed in the principal chronic dietary bioassays (Feron et al., 1981; Til et al., 1991), and no evidence of pulmonary function decrements was found in a cohort of chemical/plastics/rubber workers exposed to VC concentrations up to 250 ppm (Occidental Chemical Corporation data, as discussed in the underlying toxicological literature).

4.7 Endocrine and Immune Systems

The available animal database does not include dedicated immunotoxicology or endocrine-toxicology studies of VC, and no organ-system-specific immune or endocrine findings (beyond the reproductive/testicular endpoints addressed above) have been consistently reported in the chronic bioassays. RoseyFields identifies this as a data gap rather than as evidence of an absence of hazard (see Section 8).

5. Animal Carcinogenicity Bioassay Findings VC is one of the most thoroughly documented rodent hepatocarcinogens in the toxicological literature, with consistent tumor induction across multiple species, strains, sexes, routes of exposure, and independent laboratories.

5.1 Oral Bioassays

In the lifetime dietary bioassay in Wistar rats described in Section 4.1 (0, 1.7, 5.0, or 14.1 mg/kg-day, plus a 300 mg/kg-day gavage-in-oil arm), tumor incidence was increased in all treated groups. Liver angiosarcoma and hepatocellular carcinoma occurred together at the mid and high doses, while only hepatocellular carcinoma was observed at the lowest dose; almost exclusively angiosarcoma was seen in the gavage-in-oil group. Additional, rarer tumor types included pulmonary and extrahepatic abdominal angiosarcoma, Zymbal gland tumors (at ≥5 mg/kg-day), and abdominal mesothelioma (elevated over controls at all doses without a clear dose-response gradient) (Feron et al., 1981). In the lower-dose companion dietary study (0, 0.014, 0.13, or 1.3 mg/kg-day), tumor findings were more limited, with angiosarcoma in a small number of high-dose animals and significant increases limited to hepatocellular carcinoma in males and neoplastic nodules in females at the top dose (Til et al., 1991) — consistent with a dose-response relationship across the two related studies.

5.2 Inhalation Bioassays

The most extensive inhalation carcinogenicity dataset comes from a large multi-species, multi-concentration bioassay in rats (1–30,000 ppm) and mice and hamsters (50–30,000 ppm), with lifetime observation. Liver angiosarcoma incidence and latency were clearly concentration-dependent across species. Additional tumor types included hepatoma, nephroblastoma, and brain neuroblastoma in rats (noted by the study authors as rare in their historical control colony), and mammary, forestomach, and lung tumors in mice, with liver (including angiosarcoma) and forestomach tumors in hamsters (Maltoni et al., 1980, 1981, 1984). Several other bioassays corroborate hepatic carcinogenicity at lower concentrations: rats and mice exposed to 50–1,000 ppm for up to 10–12 months showed significantly increased liver angiosarcoma at ≥250 ppm, with concurrent lung and mammary tumors in mice (Hong et al., 1981; Lee et al., 1978), and rats exposed to as low as 100 ppm for 12 months showed significantly increased angiosarcoma incidence (Bi et al., 1985). At very high concentrations (30,000 ppm), rats developed skin, lung, and bone tumors in addition to liver tumors (Viola et al., 1971), and exposure up to 5,000 ppm produced tumors of the brain, lung, Zymbal gland, and nasal cavity in another study (Feron and Kroes, 1979). A preliminary mouse study (0, 50, 200, or 2,500 ppm) reported concentration-dependent lung adenoma, liver angiosarcoma, and mammary carcinoma (Keplinger et al., 1975), and separate studies focused on VC-induced lung tumor formation found concentration-related increases in bronchioloalveolar adenoma (Suzuki, 1978, 1983; Hehir et al., 1981).

5.3 Age-at-Exposure (Early-Life Susceptibility) Bioassays

Two independent experiments directly addressed whether age at VC exposure modifies carcinogenic response. In four rodent species/strains exposed to 100 ppm VC for 6–24 months beginning at ages of 0, 6, 12, or 18 months, tumor incidence generally increased with exposure duration and decreased with age at first exposure (Drew et al., 1983). In Sprague-Dawley rats exposed to 6,000 or 10,000 ppm VC for a fixed cumulative exposure of 100 hours, delivered either beginning at 13 weeks of age over 52 weeks or beginning at 1 day of age over 5 weeks, angiosarcoma incidence was higher in the newborn-exposed animals, and hepatoma incidence — virtually absent in the adult-exposure group — approached 50% in the newborn-exposure group (Maltoni et al., 1981). Together, these studies provide the animal basis for RoseyFields’ application of an early-life cancer-risk adjustment.

5.4 Sex Differences in Tumor Response

Sex differences in VC carcinogenic sensitivity in animals are modest but consistently favor somewhat greater female sensitivity. Inhalation bioassays found only small sex differences in liver cancer susceptibility, with female rats showing the greatest overall response (Maltoni et al., 1981, 1984), and in the dietary feeding studies, neoplastic nodules and preneoplastic basophilic foci were induced at lower doses in females, who also showed higher incidences of liver cysts and liver tumors overall (Til et al., 1991). No definitive conclusions can be drawn regarding human sex differences, since occupational cohorts have been composed almost entirely of male workers, but the female rat data are used as the more health-protective basis for RoseyFields’ cancer dose-response analyses.

6. Comparison of Animal-Derived Findings to Human Epidemiological

Findings

6.1 Human Epidemiological Findings

VC-induced hepatic angiosarcoma is among the most thoroughly documented examples of chemically induced human carcinogenicity. The first report of an association — three cases of liver angiosarcoma in PVC-plant workers — was published by Creech and Johnson (1974); angiosarcoma is an extremely rare tumor, with only about 20–30 cases per year historically reported in the entire United States. Subsequent independent cohort mortality studies consistently confirmed markedly elevated liver cancer (including angiosarcoma and, in many cohorts, excess hepatocellular carcinoma) risk in VC/PVC workers (Monson et al., 1975; Byren et al., 1976; Waxweiler et al., 1976; Fox and Collier, 1977; Cooper, 1981; Jones et al., 1988; Wong et al., 1991; Simonato et al., 1991), the last being the largest pooled cohort with 24 liver cancer deaths. Weaker, less consistent associations have also been reported between VC exposure and brain/CNS cancer (Monson et al., 1975; Byren et al., 1976; Wong et al., 1991), and, in some cohorts, lymphopoietic/hematopoietic cancer (Simonato et al., 1991) and lung cancer, the latter plausibly attributable to concurrent PVC dust exposure rather than VC itself (Waxweiler et al., 1981). A qualitative review of these cohorts concluded that, given the strong, consistent exposure-response gradient for liver angiosarcoma and the absence of a comparable gradient for other tumor types, VC is unlikely to be strongly associated with cancers other than the liver in humans (Blair and Kazerouni, 1997).

Human noncancer hepatic effects reported in occupational cohorts include portal, subcapsular, and perisinusoidal fibrosis, hepatocellular and sinusoidal-cell hyperplasia, and portal hypertension (Gedigk et al., 1975; Lilis et al., 1975; Marsteller et al., 1975; Popper and Thomas, 1975; Tamburro et al., 1984; Buchancova et al., 1985), with severity of degenerative change correlating with exposure duration, though generally reversible after exposure ceased (Gedigk et al., 1975; Lilis et al., 1975). Studies attempting to link mild, nonspecific liver-function-test abnormalities to low-level VC exposure (e.g., Ho et al., 1991) are confounded by alcohol use and inconsistent exposure characterization, and are considered less reliable than the higher-exposure clinical case series. Acroosteolysis (resorption of the terminal phalanges) and Raynaud’s phenomenon have also been documented in highly exposed workers, particularly tank cleaners (Lilis et al., 1975; Fontana et al., 1995).

6.2 Concordance Between Animal and Human Data

The concordance between animal and human data for VC is unusually strong for a chemical carcinogen. The critical target organ (liver) and even the specific rare tumor type (angiosarcoma, of sinusoidal-cell origin) are identical across humans, rats, mice, and hamsters, by both the oral and inhalation routes (Feron et al., 1981; Maltoni et al., 1981, 1984). The molecular pathway associated with angiosarcoma formation — p53-dependent mutation, with characteristic transversions — is likewise shared between VC-exposed workers and VC-exposed rodents (Chen et al., 1999; Marion et al., 1996; Donehower et al., 1992). Quantitatively, cancer potency estimates derived independently from occupational cohort epidemiology and from PBPK-adjusted animal bioassay data converge to within roughly the same order of magnitude once expressed on a common internal-dose (metabolized-VC) basis (Chen and Blancato, 1989; Clewell et al., 1995c; Reitz et al., 1996) — an unusual degree of cross-species agreement that substantially increases confidence in both the qualitative hazard identification and the quantitative potency estimates for VC.

6.3 Sources of Divergence and Their Toxicokinetic Basis

Where animal and human findings diverge, the divergence is generally attributable to one or more well-characterized toxicokinetic or study-design factors rather than to a qualitative difference in hazard:

• Nonlinear, saturable metabolism across the dose range spanned by animal bioassays. Because VC’s toxic and carcinogenic metabolites are produced by a capacity-limited oxidative pathway, responses observed at the high, near-saturating concentrations used in many animal bioassays do not scale linearly to the lower concentrations typical of human exposure; PBPK-based internal dose metrics, rather than administered dose, are required to make animal and human data quantitatively comparable (Clewell et al., 1995c).

• Imprecise historical human exposure reconstruction. The occupational cohorts most amenable to quantitative modeling (Fox and Collier, 1977; Jones et al., 1988; Simonato et al., 1991) rely on historical industrial-hygiene records and job-category surrogates with substantial uncertainty, and the largest cohort draws on multiple, heterogeneous worksites with potential exposure misclassification; this likely explains much of the roughly order-of-magnitude spread in human-cohort-derived potency estimates.

• Route-specific differences in tumor spectrum. Extrapolating the oral-bioassay-derived cancer potency to an inhalation-equivalent value yields a substantially higher estimate than the potency derived directly from inhalation bioassays, a discrepancy attributed largely to a higher incidence of neoplastic nodules in the oral bioassay, potentially reflecting rat strain or route-specific dosing artifacts rather than a true route-dependent potency difference (Feron et al., 1981; Maltoni et al., 1981, 1984).

• Possible interspecies differences in low-affinity/high-capacity metabolism. Some lower-affinity P450 isozymes contributing to VC metabolism at higher, saturating concentrations in rodents may lack clear human orthologs (Guengerich, 1987; Nakajima et al., 1993), a source of potential over- or under-prediction of human risk at higher exposure concentrations, though of limited relevance at the CYP2E1-dominated lower concentrations most pertinent to chronic risk assessment.

• Absence of evidence for greater human sensitivity. No epidemiologic or mechanistic evidence indicates humans are more sensitive to VC carcinogenicity than rodents once dose is expressed on a common internal-dose basis; if anything, humans may be somewhat less sensitive (Storm and Rozman, 1997), and endogenous background levels of the key etheno-DNA adduct are comparable between the species (Swenberg et al., 1999).

7. Data Gaps and Areas of Ongoing Research

• Immunotoxicology and endocrine toxicology. No dedicated chronic immunotoxicology or endocrine-toxicology studies of VC were located; the existing chronic and subchronic study designs were not built to detect immune or endocrine endpoints specifically, leaving these organ systems largely uncharacterized rather than affirmatively cleared.

• Quantitative linkage between DNA adduct burden and tumor risk. Although etheno-DNA adduct dose-response data support the plausibility and shape of low-dose linear cancer risk extrapolation (Swenberg et al., 1999; Morinello et al., 1999), a quantitative model directly relating adduct concentration in a specific tissue to that tissue’s tumor risk has not yet been established, which would improve confidence in low-dose extrapolation if developed.

• Definitive identification of the P450 isozyme(s) responsible for high-concentration (non- CYP2E1) VC metabolism in humans. These isozymes remain incompletely characterized, particularly with respect to human orthologs of the rodent isozymes implicated by analogy to trichloroethylene metabolism (Nakajima et al., 1993; Guengerich, 1987); resolving this would sharpen interspecies extrapolation at higher exposure levels.

• Mechanistic basis for VC noncancer hepatotoxicity. Relative to the well-characterized genotoxic mode of action for carcinogenicity, the mechanism underlying VC’s noncancer hepatic effects is comparatively poorly resolved, including the relative contributions of CEO versus CAA to these endpoints (Guengerich et al., 1981).

• Disentangling early-life susceptibility from cumulative dose. The clearest evidence for early-life carcinogenic susceptibility comes from bioassays with fixed, controlled exposure windows (Drew et al., 1983; Maltoni et al., 1981); by contrast, the two-generation study’s greater focus incidence in the in-utero-exposed generation is confounded by that generation’s longer cumulative exposure, and better-controlled studies isolating age-at-exposure from cumulative dose would strengthen early-life risk adjustments.

• Precision of human exposure reconstruction in occupational cohorts. Further refinement of historical job-exposure matrices for the major occupational cohorts (Fox and Collier, 1977; Jones et al., 1988; Simonato et al., 1991) — for example, through continued mining of contemporaneous industrial-hygiene records — would narrow the current roughly order-of-magnitude spread in human-cohort-derived cancer potency estimates and further sharpen the animal-human concordance discussed in Section 6.2.

• Human sex-difference data. Because occupational cohorts have been composed almost entirely of male workers, no direct human evidence exists to confirm or refute the modest female-predominant sensitivity pattern observed in rodent bioassays (Til et al., 1991; Maltoni et al., 1981, 1984); this remains an open question relevant to whether current toxicity values (based on the more sensitive female rat data) are appropriately protective, over-protective, or under-protective of human females specifically.

• Renal and cardiovascular effects. The limited and inconsistent organ-weight signal in kidney, heart, and spleen across chronic studies has not been followed up with dedicated mechanistic or dose-response studies sufficient to rule these organ systems in or out as VC targets with confidence.

References