1,4-Dioxane Risk Assessment and Cancer Toxicity

1. Methodology Overview

RoseyFields conducts 1,4-dioxane risk assessments using the following methodology: we identify the exposure pathways and populations relevant to the scenario under evaluation — occupational inhalation and dermal exposure, general population drinking water exposure, general population ambient air exposure, and, where relevant, aggregate exposure from multiple contributing sources; we reconstruct or adopt exposure estimates (central tendency and high-end) appropriate to each pathway; we calculate non-cancer margins of exposure (MOEs) against EPA’s benchmark MOEs and excess lifetime cancer risk against EPA’s inhalation unit risk (IUR) and cancer slope factor (CSF); and we interpret the resulting risk estimates in light of the confidence and uncertainty EPA has documented for each exposure pathway and condition of use (COU). The summary below documents EPA’s exposure assessment approach, risk characterization methodology, and scenario-specific risk results for 1,4-dioxane as reflected in its 2024 supplemental risk evaluation, and explains how RoseyFields applies this same framework in practice.

2. Populations, Exposure Scenarios, and Toxicological Endpoints Evaluated

EPA’s risk characterization for 1,4-dioxane is organized around three populations of interest, each evaluated for both acute (short-term) and chronic (long-term) exposure durations:

• Workers: adolescent (16 years and older) and adult workers, evaluated for a single 8-hour acute exposure and for chronic exposure across an 8-hour workday, 260 days per year, for 40 working years.
• General population — drinking water: adults, children, and formula-fed infants, evaluated for a 24-hour acute exposure and for chronic exposure over 33 or 78 years.
• General population — ambient air: people exposed to 1,4-dioxane in ambient air, evaluated for a 24-hour acute exposure and for continuous chronic exposure over 33 or 78 years.

Each of these populations is evaluated against the same underlying hazard values: an acute/short-term non-cancer benchmark MOE of 300 (based on systemic liver toxicity), a chronic non-cancer benchmark MOE of 30 (based on effects on the olfactory epithelium and liver toxicity), and cancer risk values (IUR and CSF) applicable to lifetime, duration-adjusted exposure. Because these hazard values do not vary by exposure pathway, the risk assessment for each scenario reduces to (1) reconstructing or estimating human exposure for that pathway and population, and (2) applying the standard MOE or cancer-risk equations described in Section 3.

3. Risk Characterization Methodology

Non-Cancer Risk: Margin of Exposure (MOE)

Non-cancer risk is expressed as a margin of exposure, calculated separately for acute and chronic exposure durations:

MOE (acute or chronic) = Non-Cancer Hazard Value (POD) ÷ Human Exposure

The resulting MOE is interpreted relative to a benchmark MOE, which is typically the total composite uncertainty factor used to derive that POD (300 for the acute liver-toxicity endpoint; 30 for the chronic olfactory-epithelium/liver-toxicity endpoint). An MOE below the benchmark MOE indicates a human health risk; an MOE at or above the benchmark MOE indicates risk is not indicated. As with the other risk values in this framework, EPA treats these MOE estimates as informative rather than as bright-line indicators of unreasonable risk, retaining discretion to weigh other risk-related factors.

Cancer Risk: Extra Lifetime Cancer Risk

Cancer risk is expressed as the extra (incremental) probability of an individual developing cancer over a lifetime as a result of repeated exposure, calculated using the applicable IUR or CSF:

Inhalation Cancer Risk = Human Exposure × IUR

Dermal/Oral Cancer Risk = Human Exposure × CSF

Human exposure for cancer risk calculations is expressed as a lifetime average daily concentration or dose (LADC/LADD, in ppm or mg/kg-day), consistent with the lifetime-averaged basis on which the IUR and CSF were derived. EPA’s 2024 supplement specifically expanded its cancer risk analysis to consider a full 78-year lifetime exposure duration, in addition to the 33-year exposure duration used in the original 2020 evaluation, noting that lifetime (78-year) cancer risk estimates are 2.36 times greater than the 33-year estimates presented for the same exposure concentration.

Aggregate and Cross-Route Considerations

Because multiple facilities or sources can contribute to 1,4-dioxane concentrations at a single drinking water intake or within a single geographic area, EPA’s 2024 supplement introduces probabilistic aggregate exposure modeling for both surface water (accounting for down-the-drain releases, indirect releases, and other upstream sources) and air (accounting for clusters of facilities within 10 km of one another). EPA also discusses, though does not fully quantify, the extent to which aggregating risk across exposure routes (e.g., combined inhalation, dermal, and oral exposure for the same individual) would alter its risk conclusions, retaining single-route risk estimates as the primary basis for risk characterization while noting this as an area of ongoing refinement.

4. Occupational Exposure and Risk

EPA estimated occupational cancer and non-cancer risk for a range of conditions of use (COUs) in which 1,4-dioxane is present as an unintentional byproduct — including uses such as textile dyes, antifreeze, surface cleaners, dish and laundry detergents, and paints and lacquers. Because cancer risk is the primary risk driver across most of these scenarios, EPA’s summary focuses on cancer risk estimates, while also flagging non-cancer risk (relative to the corresponding benchmark MOE) as a concern for many of the same COUs.

Across all evaluated occupational COUs, cancer risk estimates for inhalation exposure ranged from 4.8 × 10⁻¹¹ to 1.9 × 10⁻⁴ for central tendency exposures and from 4.8 × 10⁻¹⁰ to 7.4 × 10⁻³ for high-end exposures; cancer risk estimates for dermal exposure ranged from 8.1 × 10⁻⁷ to 7.3 × 10⁻³ for central tendency exposures and from 5.0 × 10⁻⁶ to 2.8 × 10⁻² for high-end exposures. For 7 of the 10 occupational COU subcategories evaluated, high-end cancer risk estimates exceeded 1 in 10,000 (1 × 10⁻⁴). Risks were highest for PET manufacturing, hydraulic fracturing operations, ethoxylation processes, and textile dyes, each of which exceeded 1 in 10,000 for both central tendency and high-end exposures.

Representative Occupational Risk Estimates by Condition of Use

Overall confidence in these occupational risk estimates ranges from low to high depending on the specific condition of use, reflecting differences in the age, sample size, and representativeness of the underlying exposure monitoring or modeling data — for example, EPA specifically flags uncertainty about whether monitoring data collected in the 1990s (textile dyes) or 1987 (military vehicle painting) remain representative of current workplace practices and engineering controls. Confidence in the underlying inhalation unit risk itself, by contrast, is rated high across all occupational scenarios.

5. Drinking Water Exposure and Risk

EPA evaluated general population drinking water risk resulting from four categories of release: direct facility releases to surface water, down-the-drain (DTD) releases from consumer and commercial product use, releases in hydraulic fracturing wastewater, and 1,4-dioxane reaching groundwater through landfill leachate.

Individual Facility and Down-the-Drain Releases

Risk from individual facility releases to surface water varies substantially within and across COUs, with cancer risk estimates ranging from 5.4 × 10⁻¹³ to as high as 0.025 (2.5 × 10⁻²) depending on the facility, release volume, and receiving water body characteristics. Cancer risk estimates from modeled down-the-drain releases are highest in locations where large populations contribute to these releases and where they are ultimately discharged to streams with low flow, since low-flow receiving waters provide less dilution of the same mass loading.

Hydraulic Fracturing Wastewater

Cancer risk estimates for drinking water exposure resulting from hydraulic fracturing wastewater released to surface water are 3.9 × 10⁻⁸ for median modeled releases and 1.5 × 10⁻⁶ for 95th-percentile modeled releases. For hydraulic fracturing wastewater reaching groundwater/drinking water via land disposal, cancer risk estimates range from 4.0 × 10⁻⁷ for median modeled releases to 8.6 × 10⁻⁶ for 95th-percentile modeled releases.

Landfill Leachate

Cancer risk estimates from 1,4-dioxane reaching groundwater via landfill leachate are highest under disposal scenarios that result in higher 1,4-dioxane concentrations in the leachate itself and higher landfill loading rates — meaning the risk is sensitive to both what is disposed of and how much of it is disposed of at a given site, rather than to landfill characteristics alone.

Aggregate Drinking Water Risk from Multiple Sources

Because multiple upstream sources can contribute to 1,4-dioxane concentrations at a single drinking water intake, EPA used probabilistic modeling to estimate aggregate surface water concentrations accounting for down-the-drain releases, indirect releases, and other upstream contributions, assuming no removal through water treatment and no further downstream dilution beyond the point of release. This analysis showed wide variation in both cancer and non-cancer risk within and across COUs; notably, high-end aggregate cancer risk estimates were very similar to high-end risk estimates for individual facility releases alone, indicating that high-end aggregate risk is driven primarily by the single highest-releasing facility rather than by the cumulative effect of many smaller sources. Overall confidence in this aggregate analysis is medium to high for COUs where release estimates rely on direct monitoring-based reporting (Toxics Release Inventory Form R or Discharge Monitoring Reports), medium where releases are based on TRI Form A reporting, and low to medium where release estimates rely on surrogate or modeled data rather than direct facility reporting.

6. General Population Air Exposure and Risk

Industrial Air Releases (TRI-Reported Facilities)

Cancer risk estimates for people living near facilities reporting 1,4-dioxane air releases to the Toxics Release Inventory were generally highest within 1,000 meters of the facility and declined at greater distances, consistent with standard atmospheric dispersion behavior. Based on land use patterns around modeled facilities, fenceline community exposure has the potential to occur at roughly half of the facilities evaluated (11 of 22 GIS-mapped facilities) where modeled fenceline air concentrations correspond to a cancer risk greater than 1 × 10⁻⁶.

Aggregate Risk from Neighboring Facilities

EPA separately screened for aggregate air risk where multiple facilities reporting 1,4-dioxane releases are located within 10 km of one another, identifying five such facility groups (each with two to four facilities). Aggregated risk in these groups was generally dominated by whichever single facility in the group had the greatest individual risk, and this aggregate screening analysis did not identify any location where combined cancer risk exceeded 1 × 10⁻⁶ that was not already above that threshold based on an individual facility alone.

Hydraulic Fracturing Operations

Lifetime cancer risk estimates for general population exposure within 1,000 meters of hydraulic fracturing operations range from 3.9 × 10⁻⁷ to 7.1 × 10⁻⁵ for high-end (95th-percentile) release estimates and from 2.2 × 10⁻⁸ to 4.1 × 10⁻⁶ for central tendency (50th-percentile) release estimates, across a range of topographical and meteorological modeling scenarios. Acute and chronic non-cancer risk estimates for this pathway did not indicate risk relative to the applicable benchmark MOEs at any modeled exposure concentration. Overall confidence in this pathway is medium: while the underlying air dispersion modeling methodology is robust and the cancer inhalation unit risk itself is rated high confidence, no air monitoring data were available to confirm actual 1,4-dioxane concentrations near hydraulic fracturing sites, release estimates are based on probabilistic modeling rather than site-specific data, and there is additional uncertainty around how many years a nearby resident would actually be exposed given the typically limited operational lifetime of an individual well.

Industrial and Institutional Laundries

Cancer risk estimates within 1,000 meters of industrial and institutional laundries range from 1.5 × 10⁻¹¹ to 3.8 × 10⁻⁸ across the range of high-end and central tendency air concentrations modeled for maximum release scenarios — several orders of magnitude lower than the corresponding risk range for hydraulic fracturing operations, reflecting the comparatively smaller and more localized nature of laundry facility emissions.

7. Potentially Exposed or Susceptible Subpopulations (PESS) in Risk Characterization

EPA incorporated PESS considerations into the risk characterization itself, not only into the underlying hazard assessment, in two distinct ways: through exposure-side adjustments (using population- and lifestage-specific exposure factors) and through hazard-side adjustments (applying uncertainty factors to account for biological susceptibility that cannot be separately quantified).

• Lifestage: general population drinking water exposure scenarios incorporate lifestage-specific exposure factors for adults, children, and formula-fed infants, reflecting differences in body weight and water intake rate by age group. Because inhalation exposure is evaluated based on ambient air concentration (rather than an inhaled-dose metric that varies by lifestage), inhalation exposure estimates are held consistent across lifestages, with the standard 10-fold human-variability uncertainty factor serving as the primary safeguard for any lifestage-specific differences in internal dose that cannot currently be quantified for 1,4-dioxane.
• Pre-existing disease: EPA did not identify specific health conditions expected to increase exposure itself, but qualitatively describes the potential for pre-existing disease (e.g., liver disease) to alter metabolism, behavior, or treatment in ways that could increase biological susceptibility; this is addressed through the same 10-fold human-variability uncertainty factor rather than a disease-specific quantitative adjustment.
• Lactational exposure: EPA did not quantitatively evaluate the breast-milk exposure pathway for 1,4-dioxane, based on its physical-chemical properties and the absence of studies evaluating its potential for accumulation in milk; EPA identifies this as a remaining source of uncertainty rather than a pathway it has ruled out.

RoseyFields treats these PESS considerations as directly relevant to risk characterization: where a specific exposure scenario involves children, formula-fed infants, or populations with elevated background exposure (e.g., communities served by drinking water systems already affected by other upstream 1,4-dioxane sources), RoseyFields applies the lifestage-specific exposure factors described above rather than defaulting to adult-only exposure assumptions, and flags the qualitative, non-quantified nature of the pre-existing-disease and lactational-exposure considerations as residual uncertainty in the resulting risk estimate.

8. Interpreting 1,4-Dioxane Risk Estimates

Several cross-cutting points inform how RoseyFields interprets the risk estimates summarized above:

• Cancer risk is the primary driver of concern in most 1,4-dioxane exposure scenarios evaluated in this supplement, with several occupational and drinking-water scenarios producing high-end cancer risk estimates well above the commonly used 1 × 10⁻⁴ (1-in-10,000) benchmark, and some individual facility drinking-water scenarios reaching as high as 2.5 × 10⁻² under high-end assumptions.
• Non-cancer risk is a secondary but non-negligible concern for several of the same occupational COUs where cancer risk is elevated, based on MOE estimates falling below the applicable benchmark MOE (300 acute / 30 chronic); by contrast, non-cancer risk was not indicated for the hydraulic fracturing air pathway at any modeled concentration.
• Central tendency and high-end estimates can differ by several orders of magnitude for the same exposure scenario, reflecting genuine population variability (e.g., across facilities, across individual behavior patterns) as well as uncertainty in the underlying exposure assessment; RoseyFields reports both bounds rather than a single point estimate wherever the underlying EPA data support doing so.
• Confidence in risk estimates varies substantially by pathway and COU, driven primarily by the age, sample size, and representativeness of the underlying exposure monitoring or modeling data, rather than by uncertainty in the hazard values themselves (which EPA rates medium-to-high or high confidence across the board); RoseyFields treats this exposure-side confidence rating as at least as important as the numerical risk estimate when advising on risk management priorities.
• Aggregate exposure from multiple sources can matter, but EPA’s analysis to date indicates that high-end aggregate risk (both in surface water and in air) is typically dominated by the single highest-releasing facility in a given area rather than by the cumulative effect of many smaller contributors — meaning that identifying and addressing the dominant source is likely to be the most effective risk-reduction lever in most aggregate scenarios evaluated so far.

9. Application in RoseyFields Risk Assessments

For a given site- or scenario-specific 1,4-dioxane risk assessment, RoseyFields follows the same sequence EPA uses in this supplement:

• 1. Identify the relevant population(s) and pathway(s) — worker, general population drinking water, or general population air — and the corresponding acute/chronic exposure duration assumptions appropriate to that population (e.g., 8-hour/260-day/40-year for workers; 33- or 78-year lifetime for general population).
• 2. Reconstruct or adopt exposure estimates for the pathway(s) in question, using direct exposure monitoring data where available and validated modeling (Monte Carlo occupational exposure modeling, probabilistic surface water modeling, or air dispersion modeling) where it is not, reporting both central tendency and high-end estimates rather than a single value.
• 3. Calculate non-cancer MOEs against the acute (300) and chronic (30) benchmark MOEs, and excess lifetime cancer risk against the applicable IUR or CSF, using lifetime average daily concentration/dose (LADC/LADD) for cancer calculations and selecting the 33-year or full 78-year lifetime duration assumption appropriate to the population and scenario.
• 4. Evaluate aggregate contributions where multiple sources plausibly affect the same drinking water intake or the same air-exposed community, following EPA’s approach of identifying the dominant contributing source rather than assuming risk from many small sources simply adds without bound.
• 5. Report both the numerical risk estimate and its confidence rating, flagging PESS-relevant considerations (lifestage-specific exposure factors, and the qualitative, non-quantified allowance for pre-existing disease and lactational exposure) wherever the population under evaluation includes children, infants, or other potentially susceptible groups.

This sequence mirrors the risk characterization structure EPA itself uses in the 2024 supplement, and allows RoseyFields to present both the point estimate and the underlying confidence and uncertainty context needed for an informed risk management decision, rather than a bare numerical risk value alone.

10. Conclusion

EPA’s 2024 supplement substantially expands the exposure pathways and populations evaluated for 1,4-dioxane risk beyond the original 2020 Risk Evaluation, while relying on the same underlying hazard values (non-cancer MOEs benchmarked at 300 acute / 30 chronic, and a cancer IUR and CSF derived from chronic rodent bioassay data) across all scenarios. The resulting risk picture is one in which cancer risk, rather than non-cancer risk, is the primary driver of concern across most occupational and drinking-water exposure scenarios, with risk magnitude varying by several orders of magnitude depending on the specific condition of use, release characteristics, and population exposed, and with confidence in individual risk estimates driven largely by the quality of the underlying exposure data rather than by the hazard values themselves. RoseyFields applies this same exposure assessment and risk characterization framework — non-cancer MOE and excess lifetime cancer risk calculations, informed by scenario-specific dose reconstruction, aggregate-source screening, and PESS-relevant exposure adjustments — to characterize 1,4-dioxane risk for the populations and pathways relevant to each site-specific assessment.

Bibliography