Ethylene Oxide (EtO) Risk Assessment and Cancer Risk

RoseyFields conducts Ethylene oxide (EtO) risk assessments using the following methodology, which inte grates the U.S. Environmental Protection Agency’s (EPA’s) hazard and dose-response conclusions with the underlying peer-reviewed toxicological and epidemiological literature on which those conclusions rest. Be cause EtO risk assessment draws on several distinct scientific literatures — occupational epidemiology, ro dent bioassay toxicology, physiologically based pharmacokinetics, and molecular genotoxicology — Rosey Fields’ approach is organized to make explicit which primary studies drive each quantitative judgment, so that the resulting toxicity values can be defended, cross-examined, and applied to case-specific exposure scenarios with full transparency.

1. Chemical Identity, Regulatory Background, and Exposure Sources

Ethylene oxide (CASRN 75-21-8) is a small, highly reactive, flammable epoxide that is a gas at room tem perature. It is produced industrially by the catalytic oxidation of ethylene and is used predominantly as a chemical intermediate in the manufacture of ethylene glycol, glycol ethers, and nonionic surfactants. Its second major use — and the one of greatest relevance to occupational and community exposure litigation — is as a low-temperature sterilizing and fumigating agent. Because EtO penetrates packaging and diffuses uniformly through complex geometries without heat or moisture, it remains the sterilant of choice for heat and moisture-sensitive medical devices (catheters, sutures, implants, single-use surgical kits) and is also used to fumigate spices and other dried agricultural commodities.

EtO is designated a hazardous air pollutant under the 1990 Clean Air Act Amendments and is regulated by the EPA’s Office of Air and Radiation for ambient and facility-level emissions, and by the EPA’s Office of Pes ticide Programs under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) for its sterilant and fumigant uses. EPA’s current hazard and dose-response conclusions for inhalation carcinogenicity are set out in its 2016 IRIS carcinogenicity assessment, which classifies EtO as carcinogenic to humans by the inhala tion route and derives an inhalation unit risk estimate; the International Agency for Research on Cancer has likewise classified EtO as carcinogenic to humans (Group 1). Because the current federal cancer assessment addresses inhalation carcinogenicity only, RoseyFields’ noncancer inhalation toxicity value is derived inde pendently, following the same point-of-departure/uncertainty-factor framework applied by other health agencies to the same primary animal toxicology database (see Sections 3–6, below).

Exposure pathways of regulatory and litigation relevance fall into three broad categories. First, occupational exposure occurs among workers in EtO production and glycol-manufacturing plants, and — of greatest relevance to the bulk of the toxic tort and workers’-compensation litigation on this chemical — among sterilization workers at commercial sterilization facilities and hospital central sterile processing departments, where EtO gas is used in chamber sterilization cycles for medical devices (Ambroise et al., 2005; Kardos et al., 2003). Second, community or “fenceline” exposure occurs among residents living near EtO production, glycol manufacturing, or commercial sterilization facilities, arising from stack and fugitive emissions and chamber off-gassing; typical ambient background concentrations away from point sources are extremely low (sub-ppb), rising measurably within the vicinity of an active sterilization or production facility. Third, low-level residual exposure can occur from off-gassing of sterilized medical devices themselves, though this pathway is generally minor relative to occupational and fenceline exposure once aeration protocols are followed. RoseyFields’ exposure/dose reconstruction work (Section 8) typically begins by characterizing which of these pathways is operative in a given matter and by identifying whether individual- or area-level exposure monitoring data, process/engineering records, or exposure-modeling approaches analogous to those developed for the underlying occupational cohort studies (Hornung et al., 1994; Greife et al., 1988) are available to reconstruct historical exposure.

2. Hazard Characterization and Mode of Action

EtO’s carcinogenic hazard characterization rests on three converging lines of evidence: rodent bioassay data, human epidemiological data, and genotoxicity/mechanistic data supporting a mutagenic mode of action.

Chronic inhalation bioassays in both rats and mice demonstrate that EtO is a multi-site carcinogen. In a two year inhalation bioassay in Fischer 344 rats, Snellings, Weil, and Maronpot (1984) reported dose-related increases in mononuclear cell leukemia, peritoneal mesotheliomas, and brain tumors; the brain tumor findings from this same bioassay were characterized in greater detail by Garman, Snellings, and Maronpot (1985, 1986), who described the tumors’ cell type, distribution, and dose-response pattern. A parallel inhalation bioassay in F344 rats and B6C3F1 mice conducted for the National Institute for Occupational Safety and Health likewise identified carcinogenic and other toxicologic effects of inhaled EtO (Lynch et al., 1984a). The National Toxicology Program’s chronic inhalation bioassay in B6C3F1 mice (NTP, 1987) found significantly increased incidences of lymphocytic neoplasms and, in females, mammary gland carcinomas. Molecular characterization of the mouse mammary and lung tumors from this and related bioassays has since shown a distinctive pattern of K-ras and p53/H-ras mutations consistent with EtO’s direct mutagenic activity at these tumor sites (Houle et al., 2006; Hong et al., 2007).

Human evidence comes primarily from occupational cohort studies of EtO production and sterilization workers. The pivotal cohort is the NIOSH cohort of 18,235 workers employed at 13 U.S. sterilization fa cilities, for whom individual retrospective exposure estimates were developed from a validated statistical exposure-reconstruction model incorporating sterilizer size, chamber cycle characteristics, and industrial hygiene monitoring data (Hornung et al., 1994; Greife et al., 1988). Early mortality follow-up of this cohort found excess lymphohematopoietic cancer mortality (Steenland et al., 1991); an exposure-response anal ysis of the same cohort (Stayner et al., 1993) and extended mortality follow-up through 1998 (Steenland, Stayner, and Deddens, 2004) confirmed a positive exposure-response trend for lymphoid cancer, concen trated among males. A companion incidence study of the female sub-cohort with individual interview data on reproductive and other breast-cancer risk factors found a positive exposure-response trend for breast cancer incidence (Steenland et al., 2003). Independent, smaller cohorts of EtO-exposed workers in the United Kingdom (Coggon et al., 2004), Sweden (Mikoczy et al., 2011), and U.S. chemical manufacturing (Swaen et al., 2009) provide broadly supportive, though individually less statistically powerful, evidence.

The genotoxicity database is unusually extensive for EtO and supports a mutagenic mode of action operat ing through direct DNA alkylation. EtO forms N7-(2-hydroxyethyl)guanine and other alkylated DNA adducts in a dose-related manner (Marsden et al., 2007; Marsden et al., 2009), induces base excision repair gene responses and apurinic/apyrimidinic sites in rodent tissue following inhalation exposure (Rusyn et al., 2005), and produces chromosomal aberrations, sister chromatid exchange, and hemoglobin adducts in both ex perimentally exposed animals and occupationally exposed human populations (Tates et al., 1991). Reviews of this mechanistic database by Preston et al. (1995), Natarajan et al. (1995), and Preston (1999) conclude that the weight of evidence supports a mutagenic mode of action for EtO carcinogenesis — a conclusion with direct quantitative consequences, because it is this mode-of-action determination that justifies both (a) linear low-dose extrapolation without a threshold in the cancer risk assessment (Section 7) and (b) the application of age-dependent adjustment factors for early-life exposure.

3. Point-of-Departure Selection

RoseyFields’ EtO risk assessments require two distinct points of departure (PODs) because the chemical carries both a cancer risk value and a noncancer (systemic/reproductive/neurotoxic) risk value, each derived from a different segment of the toxicological literature.

Cancer POD. For the cancer inhalation unit risk, the critical study is the NIOSH sterilization-worker cohort (Steenland et al., 1991, 2003, 2004), selected in preference to the rodent bioassays for four reasons that RoseyFields applies as a general matter when human data are available and adequate: (1) it is the largest oc cupational cohort of EtO-exposed workers with individual, model-based retrospective exposure estimates (Hornung et al., 1994); (2) it includes both males and females, permitting derivation of a female-specific breast cancer risk estimate in addition to the lymphoid cancer estimate; (3) it has long follow-up and an internal-comparison design that minimizes healthy-worker bias and confounding by co-exposure, since ster ilization work involves little exposure to other identified carcinogens; and (4) the underlying exposure as sessment is of unusually high quality for a retrospective occupational study. Rather than a single POD in the traditional NOAEL/LOAEL/benchmark-dose sense, the human-data-based cancer assessment uses an EC01/LEC01 approach: a two-piece linear spline model is fit to the individual-level continuous exposure response data, and the lower 95% confidence bound on the exposure associated with 1% extra risk (the LEC01) is calculated from a life-table analysis of the cohort’s exposure and vital-status data. This LEC01 — on the order of 5 × 10⁻³ ppm for lymphoid cancer mortality and roughly 1.9 × 10⁻³ ppm for lymphoid cancer incidence under the preferred model — functions as the cancer POD from which linear low-dose extrapolation proceeds (Section 7). Three chronic rodent bioassays (Snellings, Weil, and Maronpot, 1984; Lynch et al., 1984a; NTP, 1987) were also modeled as a cross-check; the resulting rodent-based unit risk estimates are roughly two orders of magnitude lower than the human-data-based estimates, a divergence RoseyFields addresses explicitly with retained counsel and experts because it can become a significant point of cross-examination (see Section 9).

Noncancer POD. For systemic, reproductive, and neurotoxic endpoints, no adequate chronic human exposure-response dataset exists with the individual dosimetry needed for POD derivation, so the critical study is an animal bioassay. RoseyFields’ noncancer inhalation toxicity value is anchored to the subchronic inhalation study of EtO in B6C3F1 mice conducted by Snellings, Weil, and Maronpot (1984), which identified impaired neurological function and decreased absolute testicular weight as the most sensitive, consistently observed effects across the tested exposure range, with a free-standing no-observed-adverse-effect level (NOAEL) of 10 ppm (the highest concentration tested without a corresponding LOAEL for these endpoints). This is the same critical study relied upon by other state and federal health agencies that have derived noncancer EtO toxicity values in the absence of an EPA IRIS-derived RfC (for example, ATSDR’s intermediate-duration inhalation minimal risk level and state health department air guidance values are anchored to this same bioassay). RoseyFields selects this endpoint and study because (a) neurological and reproductive/testicular effects are reproducible across independent inhalation studies at comparable exposure levels, (b) the endpoint is biologically plausible given EtO’s direct alkylating mechanism and its documented capacity to produce peripheral neuropathy and testicular toxicity in multiple species, and (c) the study design (subchronic, multiple dose groups, quantitative histopathology) meets standard criteria for POD derivation.

4. Pharmacokinetic Modeling and Dose Conversion

EtO’s dosimetry has been characterized in unusual depth using physiologically based pharmacokinetic (PBPK) modeling, and this literature plays two distinct roles in RoseyFields’ methodology: supporting interspecies extrapolation for the noncancer value, and informing (though ultimately not driving) the cancer assessment, which relies on direct human exposure-response data rather than animal-to-human extrapolation.

Fennell and Brown (2001) constructed PBPK models describing EtO uptake, distribution, and metabolism in mice, rats, and humans, building on kinetic data from Brown, Wong, and Fennell (1996) and Brown, As gharian, Turner, and Fennell (1998). These models simulate a two-limbed elimination process: at low, en vironmentally and occupationally relevant exposure concentrations, EtO is cleared primarily by glutathione conjugation (a first-order process), while at higher concentrations glutathione depletion shifts elimination toward slower, nonenzymatic hydrolysis, producing a disproportionate (greater-than-linear) rise in blood EtO concentration and adduct formation above roughly 200 ppm (Brown et al., 1998). Critically for interspecies extrapolation, Fennell and Brown’s (2001) simulations of 6-hour inhalation exposures show that blood area-under-the-curve values in mice, rats, and humans are all within about 15% of one another for the same inhaled concentration — that is, at a given air concentration, internal (blood) dose is approximately equivalent across species in the range relevant to the chronic bioassays (100 ppm and below). This finding is corroborated by the independent PBPK model of Csanády et al. (2000) for rat, mouse, and human, and by the earlier rat-specific dosimetry model of Krishnan, Gargas, Fennell, and Andersen (1992). Direct measurement of the elimination half-life of EtO in human blood — 42 minutes (Filser et al., 1992) — versus 2.4–3.2 minutes in mice and 11–14 minutes in rats (Brown et al., 1996, 1998) — illustrates the underlying species difference in metabolic clearance rate that the PBPK models reconcile at the level of tissue dose.

For the noncancer POD, this dosimetric equivalence is applied in two steps. First, the animal inhalation exposure concentration is converted to a human equivalent concentration (HEC) using the standard default methodology for a Category 3 (deep-respiratory-tract/systemically acting) gas: a duty-cycle adjustment for the discontinuous experimental exposure schedule (6 hours/day, 5 days/week) and a dosimetric adjust ment factor (DAF) of 1.0, which reflects the PBPK-supported conclusion that blood-level internal dose is approximately dosimetrically equivalent between rodents and humans at a given air concentration. This yields an HEC on the order of 1.8 ppm (roughly 3.2 mg/m³) from the 10 ppm NOAEL. Second, because the PBPK data address the toxicokinetic (dosimetric) component of interspecies variability directly, RoseyFields applies a reduced (rather than full default 10-fold) interspecies uncertainty factor to the HEC, reserving the residual uncertainty factor for toxicodynamic differences alone (Section 5). For the cancer assessment, by contrast, no animal-to-human or internal-dose extrapolation is required for the preferred estimate because the POD is derived directly from human exposure-response data; the PBPK literature is used there primarily as supporting mechanistic evidence (demonstrating that tissue dose scales approximately linearly with air concentration across the relevant exposure range) rather than as a dose-conversion tool.

5. Uncertainty Factor Derivation

Because EtO’s cancer and noncancer toxicity values are derived by fundamentally different statistical frame works, RoseyFields applies uncertainty factors only to the noncancer value; the cancer unit risk estimate is instead adjusted using life-table extrapolation and age-dependent adjustment factors, addressed in Section 7.

For the noncancer inhalation value, RoseyFields applies a composite uncertainty factor built from four components, consistent with EPA’s standard Reference Concentration methodology and with the approach taken by other agencies that have derived EtO noncancer values from the same critical study:

• Interspecies uncertainty factor (UF_A = 3): A full 10-fold default is not applied because the toxicokinetic component of interspecies variability has been directly addressed through PBPK modeling (Fennell and Brown, 2001; Csanády et al., 2000), which demonstrates approximate dosimetric equiv alence between rodents and humans at equivalent air concentrations. The reduced factor of 3 is retained to account for residual toxicodynamic (pharmacodynamic sensitivity) differences between species that the dosimetry models do not address.
• Intraspecies (human variability) uncertainty factor (UF_H = 10): A full default factor is retained be cause the critical study population (inbred mice) does not capture the genetic, physiological, and life stage variability of the human population, including polymorphisms in glutathione S-transferase and microsomal epoxide hydrolase that measurably affect EtO detoxification capacity and hemoglobin adduct burden in occupationally exposed humans (Haufroid et al., 2007).
• Subchronic-to-chronic uncertainty factor (UF_S = 3): Applied because the critical study is a sub chronic (13-week) rather than a full chronic (lifetime) bioassay, and because the endpoints of concern — peripheral neuropathy and testicular toxicity — are not clearly demonstrated to plateau in severity over shorter exposure durations.
• Database uncertainty factor (UF_D = 3): Applied to reflect gaps in the developmental neurotoxi city database and in multigenerational reproductive toxicity data at low exposure concentrations, notwithstanding the substantial overall database on EtO reproductive and neurologic endpoints (e.g., Lynch et al., 1984a, on testicular and reproductive parameters in monkeys and rats).
• LOAEL-to-NOAEL uncertainty factor (UF_L = 1): Not applied, because the critical effect level from Snellings, Weil, and Maronpot (1984) is a true NOAEL (the highest concentration tested, 10 ppm, produced no adverse neurological or testicular effect), not a LOAEL requiring extrapolation.

The individual factors multiply to 90, which RoseyFields rounds to the conventional composite value of 100 for a subchronic-equivalent (intermediate-duration) value; incorporating the additional subchronic-to chronic factor for the chronic value yields a composite uncertainty factor of approximately 300. This UF derivation and its constituent values are documented case-by-case in RoseyFields’ assessment memoranda so that each factor’s toxicological justification (rather than a default reflex application) is available for expert testimony and cross-examination.

6. Derivation of the Chronic Inhalation Reference Concentration Applying the POD, HEC conversion, and composite uncertainty factor above:

RfC = HEC ÷ UF_composite = 3.22 mg/m³ ÷ 300 ≈ 0.011 mg/m³, rounded to 10 µg/m³ (approximately 0.0055 ppm) as RoseyFields’ chronic noncancer inhalation toxicity value for EtO, with impaired neurological func tion and testicular toxicity as the critical, hazard-index-defining effects (based on Snellings, Weil, and Maronpot, 1984).

RoseyFields flags this value’s provenance for every client and every court filing: EPA’s current IRIS assessment for EtO addresses inhalation carcinogenicity only and does not itself derive an RfC or oral RfD for this chemical. The chronic noncancer value above is therefore constructed by RoseyFields using the same POD/UF methodology EPA and other agencies apply generally, anchored to the same primary animal toxicology study that other health agencies (including state environmental health agencies and the Agency for Toxic Substances and Disease Registry) have used to derive their own EtO noncancer inhalation values, several of which cluster in the same order of magnitude (roughly 10–30 µg/m³) despite differing agency specific default assumptions. Where a matter turns on a jurisdiction-specific regulatory value rather than RoseyFields’ independently derived figure, RoseyFields presents both and reconciles the difference in its methodology memorandum. No oral RfD is derived, as inhalation is the exposure route of concern for essentially all EtO matters RoseyFields encounters (EtO’s high volatility and reactivity make stable oral or dermal exposure scenarios rare and generally not supported by the toxicology or exposure-monitoring lit erature).

7. Carcinogenicity Classification and Inhalation Unit Risk Derivation

EtO is characterized as carcinogenic to humans by the inhalation route, a conclusion RoseyFields treats as well-supported by the convergence of rodent bioassay evidence (Snellings, Weil, and Maronpot, 1984; Lynch et al., 1984a; NTP, 1987), human epidemiological evidence (Steenland et al., 1991, 2003, 2004), and mechanistic/genotoxicity evidence supporting a mutagenic mode of action (Preston et al., 1995; Natarajan et al., 1995).

Because a mutagenic mode of action is operative, RoseyFields applies linear low-dose extrapolation with out a threshold, consistent with standard cancer risk assessment practice for genotoxic carcinogens. The quantitative derivation proceeds directly from the human cohort data rather than through a traditional animal-to-human cancer slope factor conversion:

1. A two-piece linear spline model is fit to the individual-level, continuous-exposure data from the NIOSH cohort for each cancer endpoint of interest (lymphoid cancer mortality and incidence, breast cancer incidence in females), using a life-table analysis that accounts for the cohort’s age structure, follow-up time, and competing mortality.

2. The lower 95% confidence bound on the exposure concentration associated with 1% extra risk (the LEC01) is calculated from the lower (low-exposure) segment of the spline, which is the segment most relevant to environmental and typical occupational exposure levels.

3. Linear extrapolation from the LEC01 to the origin yields the inhalation unit risk estimate (unit risk = 0.01 ÷ LEC01) for each endpoint.

4. Endpoint-specific unit risk estimates for lymphoid cancer incidence and female breast cancer incidence are combined (as an upper bound on the sum) to yield a total cancer unit risk estimate. 5. Because the weight of evidence supports a mutagenic mode of action and no EtO-specific early-life susceptibility data exist, increased early-life susceptibility is assumed by default, and age-dependent adjustment factors (a 10-fold potency adjustment for exposure in the first two years of life, 3-fold for ages two through 15, and 1-fold for adulthood) are applied to produce the final, ADAF-adjusted lifetime unit risk estimate used in risk characterization.

The resulting inhalation unit risk RoseyFields applies is approximately 5.0 × 10⁻³ per µg/m³ (total cancer, ADAF-adjusted, based on combined lymphoid and female breast cancer incidence), with endpoint-specific components of approximately 2.6 × 10⁻³ per µg/m³ for lymphoid cancer and 7.0 × 10⁻⁴ per µg/m³ for breast cancer incidence in females. This unit risk is intended for environmental (non-occupational) continuous lifetime exposures up to roughly 40 µg/m³; for occupational-range exposures, RoseyFields calculates ex tra risk directly from the underlying exposure-response models rather than by linear extrapolation of the environmental unit risk, because the occupational cohort data show a distinctly supralinear (steeper-at-low exposure) exposure-response shape that a single linear unit risk value would misrepresent at high cumula tive exposures. RoseyFields’ confidence in the total cancer unit risk estimate is high, driven substantially by the breast cancer incidence component (a large, well-characterized sub-cohort with covariate data on other breast cancer risk factors), with comparatively greater — though still acceptable — uncertainty in the lymphoid cancer component, which rests on a smaller number of events and a more sharply supralinear exposure-response relationship.

8. Practical Application: Hazard Quotients, Excess Lifetime Cancer Risk, and Dose Reconstruction

In practice, RoseyFields’ risk assessors apply the values derived above in three linked steps common to both regulatory consulting and litigation support engagements.

Exposure and dose reconstruction. Because EtO litigation frequently concerns historical occupational or fenceline exposures for which contemporaneous monitoring data are sparse or entirely absent, RoseyFields’ first task is typically exposure reconstruction. Where feasible, this follows the same general approach validated in the NIOSH cohort studies: development of a statistical exposure model incorporating facility- and job-specific determinants of EtO release and ambient concentration (chamber/sterilizer size and cycle pa rameters, ventilation design, aeration protocols, and any available industrial-hygiene monitoring records), calibrated against whatever contemporaneous personal or area monitoring data exist (Hornung et al., 1994; Greife et al., 1988). Where individual monitoring data are available (e.g., from OSHA-mandated personal exposure monitoring at sterilization facilities), RoseyFields uses those data directly, adjusted as needed for exposure duration, work practices, and personal protective equipment use. Biomarker data — hemoglobin adducts, urinary metabolites — are used where available to corroborate or calibrate reconstructed air con centrations, following the biomonitoring methodology developed in the occupational literature (Tates et al., 1991; Boogaard, 2002), with appropriate allowance for genetic polymorphisms (glutathione S-transferase, epoxide hydrolase) that affect the relationship between air concentration and internal adduct burden across individuals (Haufroid et al., 2007).

Hazard Quotient and Hazard Index calculations. For noncancer risk characterization, RoseyFields calcu lates a chemical-specific Hazard Quotient (HQ) as the ratio of the reconstructed or estimated exposure concentration (typically expressed as an appropriately time-weighted, chronic average air concentration) to the chronic inhalation RfC derived in Section 6. An HQ at or below 1.0 indicates that the exposure is not expected to result in appreciable noncancer risk of the neurological or reproductive effects on which the RfC is based; an HQ above 1.0 does not itself establish that harm has occurred, but flags the exposure level as one for which the noncancer margin of safety has been eroded and case-specific medical and exposure evidence takes on correspondingly greater weight. Where EtO co-occurs with other airborne contaminants sharing a common toxicological target (for example, other neurotoxic solvents in a mixed occupational exposure), RoseyFields calculates a target-organ-specific Hazard Index by summing the relevant HQs, consistent with standard EPA mixtures guidance.

Excess lifetime cancer risk (ELCR) calculations. For cancer risk characterization, RoseyFields calculates ELCR as the product of the reconstructed lifetime-average (or, where appropriate, occupational or less than-lifetime, ADAF-adjusted) exposure concentration and the applicable inhalation unit risk derived in Section 7, selecting the lymphoid-cancer-specific, breast-cancer-specific, or combined total-cancer unit risk depending on the tumor type(s) at issue in the matter. For less-than-lifetime or early-life exposure scenarios, age-dependent adjustment factors are applied to each relevant exposure window rather than using the unadjusted adult-based unit risk uniformly across the exposure duration. For occupational-range cu mulative exposures, RoseyFields calculates extra risk directly from the exposure-response models fit to the occupational cohort data (rather than through linear extrapolation of the environmental unit risk), consistent with the supralinear shape of that exposure-response relationship.

Regulatory and litigation support. These calculations support several distinct work products: facility specific risk assessments and permitting support for clients operating under the Clean Air Act’s air toxics program or FIFRA sterilant/fumigant registration requirements; community health risk characterizations in support of or response to regulatory action (e.g., National Emission Standards for Hazardous Air Pollutants residual risk reviews); and, in litigation, general and specific causation analyses that connect a plaintiff’s or cohort’s reconstructed exposure history to the quantitative risk framework above, expert reports addressing both the magnitude of estimated risk and the scientific defensibility of the underlying POD, dosimetry, and uncertainty-factor choices, and rebuttal analysis of opposing experts’ alternative point-of-departure, dose-reconstruction, or extrapolation methodologies.

9. Key Uncertainties and Sensitive Populations

Several uncertainties recur across RoseyFields’ EtO matters and should be addressed explicitly in any case specific application of these values.

Retrospective exposure assessment. The cancer unit risk rests on a retrospective, model-based exposure reconstruction for the underlying occupational cohort (Hornung et al., 1994); while this model was built with unusual care and validated against available monitoring data, individual worker exposure estimates carry non-trivial uncertainty, which propagates into both the exposure-response model fit and, ultimately, into any case-specific application that must, in turn, reconstruct an individual plaintiff’s or facility’s exposure history against a comparably uncertain historical record.

Model selection and the supralinear exposure-response shape. The cohort’s exposure-response relation ship for lymphoid cancer is markedly supralinear (steeper at low cumulative exposure than at high), a pattern that is statistically well-supported but complicates both model selection and the confidence interval on the resulting low-exposure risk estimate; sensitivity analyses across alternative model specifications yield unit risk estimates for lymphoid cancer ranging from roughly half to three times the preferred estimate, a wider band than for the breast cancer component.

Rodent-versus-human unit risk divergence. The roughly two-orders-of-magnitude difference between the human-data-based and rodent-bioassay-based cancer unit risk estimates is a recurring point of dispute in litigation; RoseyFields’ position (consistent with the current weight of scientific opinion) is that the human data-based estimate is preferred when adequate human data exist, because it avoids interspecies extrapolation uncertainty entirely, but the magnitude of the divergence merits case-specific discussion with retained toxicology and biostatistics experts.

Early-life susceptibility. Because no EtO-specific data on early-life susceptibility exist, RoseyFields follows the default, mode-of-action-triggered assumption of increased early-life susceptibility and applies age dependent adjustment factors for any exposure scenario involving in utero, infant, or childhood exposure (e.g., fenceline residential exposure near a sterilization or manufacturing facility, or the children of exposed workers). This is a policy-driven default rather than an EtO-specific empirical finding and should be flagged as such.

Sex-specific risk. The breast cancer component of the cancer unit risk applies only to females; RoseyFields calculates sex-specific ELCR for mixed-sex exposure cohorts rather than applying the combined total-cancer unit risk uniformly, except where a summed, sex-agnostic screening-level estimate is appropriate for a predominantly female (e.g., hospital sterile-processing) or predominantly male workforce.

Genetic and physiological susceptibility. Human polymorphisms in glutathione S-transferase and microsomal epoxide hydrolase — the principal enzymes governing EtO detoxification — measurably affect internal dose (hemoglobin adduct burden) at a given air concentration (Haufroid et al., 2007), meaning that a sub set of any exposed population may carry systematically higher internal dose, and correspondingly higher risk, than the population-average exposure-response relationship implies. Reproductive-age workers and pregnant workers are a further population of concern given the reproductive and developmental toxicity evidence from animal studies (Snellings, Weil, and Maronpot, 1984; Lynch et al., 1984a), and the demon strated heritable-mutation potential of EtO in animal germ-cell assays (Generoso et al., 1990).

Database completeness. Notwithstanding the extensive genotoxicity and occupational epidemiology litera ture on EtO, gaps remain in chronic (as opposed to subchronic) noncancer toxicology, in low-concentration developmental neurotoxicity data, and in exposure-response data for non-sterilization occupational set tings, all of which RoseyFields accounts for through the database uncertainty factor (Section 5) and through case-specific sensitivity analysis rather than treating the derived toxicity values as free of residual uncer tainty.

References