Appendix C: Additional Information Regarding 1,4-Dioxane Cancer Toxicity Values

Section 1: Introduction
Different toxicity values protective of cancer risk have been derived for 1,4-dioxane. These differences are due to the selection of key studies and endpoints, as well as assumptions about the shape of tumor dose-response curves in the low-dose region. The following table summarizes the linear cancer slope factors (CSFs) derived by the U.S. Environmental Protection Agency’s (USEPA) Integrated Risk Information System (IRIS) program, USEPA’s Office of Chemical Safety and Pollution Prevention (OCSPP), and the threshold-based Tolerable Daily Intake derived by Health Canada (Table C-1). Please note that as of the writing of this section, USEPA (2019) and Health Canada (2018) assessments are drafts; therefore, content and conclusions may be subject to change.

Table C-1. Example of cancer toxicity values

Source of Cancer Toxicity Value	Value	Basis

USEPA IRIS (USEPA 2013a)*	Oral slope factor: 1 x 10^-1 per mg/kg/day Drinking water unit risk: 2.9 x 10^-6 per µg/L Inhalation unit risk: 5 x 10^-6 per (µg/m³) 0.35 µg/L for a 1 in a million risk	The IRIS assessment concluded there was insufficient evidence to support hypothesized mode of actions (MOAs) or to deviate from using a linear low-dose extrapolation approach to evaluate cancer risks for the oral, inhalation, or dermal exposure routes. Therefore, USEPA IRIS applied a linearized model to evaluate 1,4-dioxane’s cancer risk. “The Kano et al. (2009) drinking water study was chosen as the principal study for derivation of an oral cancer slope factor (CSF) for 1,4-dioxane” (USEPA 2013a, 127). Benchmark Dose Software (BMDS) modeling of the female mouse liver tumor response (Lowest Observed Effect Level [LOAEL] of 66 mg/kg/day with no available No Observed Effect Level [NOAEL] since the lowest dose generated 35 combined adenomas and carcinomas out of 50 female mice. There were 31 adenomas and 6 carcinomas counted). A Benchmark Dose Level (BMDL₅₀)of 32.93 mg/kg/day was estimated and converted to a BMDL_50HED of 4.95 mg/kg/day. Human Equivalent Dose (HED) was calculated with a body weight scaling factor. “A chronic bioassay of 1,4-dioxane by the inhalation route reported by Kasai et al. (2009) provides data adequate for dose-response modeling and was subsequently chosen as the study for the derivation of the IUR [inhalation unit risk] for 1,4-dioxane.” (USEPA 2013a, 129) BMDS modeling of combined tumor types reported by Kasai et al. (2009) was employed to derive the IUR. A Benchmark Concentration Level (BMCL₁₀) of 30.3 parts per million (ppm) was estimated and converted to a BMCL_10HEC of 19.5 mg/m³. The Human Equivalent Concentration (HEC) was calculated assuming a continuous exposure for 24 hours per day for 7 days per week.

USEPA OCSPP** Draft (USEPA 2019e)	Oral slope factor: 2.1 x 10^-2 per mg/kg/day Inhalation unit risk (worker): 1 x 10^-6 (µg/m³)^-1	USEPA concluded that inadequate information exists to support a threshold-based MOA for liver cancer; no MOA information for other cancers (nasal, peritoneal mesotheliomas, subcutis fibroma, kidney, breast) supporting the default, linear modeling of the cancer risk due to unresolved uncertainties in the possible MOAs. BMD modeling of the cancer data from Kano et al. (2009) and Kasai et al. (2009).^# The oral CSF was derived with MS_Combo (including Liver) from the Kano et al. (2009) drinking water study tumors in male F344/DuCrj rats. IUR was calculated from combined cancer modeling of tumors reported in Kasai et al. (2009) and adjusted for application to an occupational exposure scenario.

Health Canada Draft (Health Canada 2018)	Tolerable daily intake: 0.0054 mg/kg/day 0.050 mg/L (50 µg/L) maximum acceptable concentration (MAC)	Canada applied a threshold approach to modeling 1,4-dioxane’s cancer risk: “Since 1,4-dioxane acts through a non-genotoxic MOA and is known to operate via non-linear kinetics, a non-linear (threshold) risk assessment is considered appropriate.” (39). BMDL modeling of a 5% benchmark response generated a 5.4 mg/kg/day point of departure for male and female rats combined from the Kociba et al. (1974) noncancer liver findings. A 1,000-fold uncertainty factor adjustment was applied (10x for interspecies, 10x for intraspecies variability, and 10x for database deficiencies such as poor characterization of reproductive and developmental toxicity). A relative source contribution of 20% (0.2) was included to generate a proposed MAC of 0.050 mg/L.

#OCSPP calculated a dermal CSF (in mg/kg/day^-1), but is not relevant for environmental exposures via ingestion of contaminated drinking water.
*USEPA’s Integrated Risk Information System.
**USEPA’s Office of Chemical Safety and Pollution Prevention (OCSPP).

With respect to the selection of toxicity values for use in risk assessments within the United States, toxicity information to assess a chemical’s potential human health risks should be based on the most recent, credible, and relevant data and methods. Beginning with USEPA Risk Assessment Guidance for Superfund, Volume I, Human Health Evaluation Manual (Part A), agency guidance recommends selecting toxicity criteria based on the most recent data (USEPA 1989, 7–15). This recommendation has since been implemented in numerous USEPA Office of Solid Waste and Emergency Response (OSWER) directives (USEPA 1993, 2003) that further establish a hierarchy and process for selecting toxicity criteria.

Several groups rely on USEPA’s assessments based on their use of standardized methods and peer review; however, IRIS toxicity values are not always recent and up to date, nor are they available for all chemicals of potential concern at a site. Therefore, consistent with USEPA OSWER directives (USEPA 1993, 2003), alternative sources’ toxicity values—dubbed “Tier 2 and Tier 3” toxicity values—are recognized and accepted by USEPA and several other groups (e.g., USEPA’s RSLs program, the North Carolina Department of Environmental Quality, the Agency for Toxic Substance and Disease Registry [2005]). Importantly, comparison of available safe doses to ensure validity of the approach is now a routine part of regulatory toxicology. As stated in the original OSWER 1993 directive, “in some cases more recent, credible and relevant data may come to the Agency’s attention. …[T]he Agency should evaluate risk based upon its best scientific judgment and consider all credible and relevant information available to it” (USEPA 1993, 2). Also consistent with USEPA directives and guidance documents, priority is given to toxicity information that meets criteria, including transparency of the information and methods, level of external and independent peer review, and use of established methodology consistent with best scientific information and practices used by USEPA. These guiding criteria are presented in the USEPA white paper on selecting Tier 3 toxicity values (USEPA 2013b, 8 and 11).

Toxicity values derived by OCSPP under the Frank R. Lautenberg Chemical Safety for the 21st Century Act (LCSA) present a unique and new challenge. When the LCSA was signed into law, it required a robust weight of evidence methodology for examining both cancer and noncancer risk. These more stringent legal requirements for weight-of-evidence were not required when IRIS presented its final risk assessment in 2013a. USEPA’s 2003 memorandum still specifies IRIS values as the first choice, with no guidance on how to treat toxicity values derived by LCSA Section 6 risk evaluations.

Evaluation of multiple sources of toxicity information ensures that the information used is current, peer reviewed, transparent, and the best available information. Understanding the inherent uncertainties in toxicity values is important, especially for supporting various risk management options. This flexibility recognizes that new chemical-specific information may become available and that risk assessment practices are continually evolving; therefore, selection of a toxicity value should be based on the most recent, credible, and relevant data and risk assessment methods available while accounting for uncertainty. Online and updated databases exist to make such comparisons easier, such as the International Toxicity Estimates for Risk (ITER) and USEPA Chemistry Dashboard.

Therefore, selecting which value (e.g., IRIS, OCSPP, Health Canada, or other toxicity values) to apply in a 1,4-dioxane risk assessment requires knowledge of the underlying toxicology data, the assumptions applied in dose-response modeling, and, most importantly, the selection between a linear (which holds when no MOA exists) or a threshold low-dose extrapolation procedure (which holds that doses below those required to trigger an early key event in the cancer MOA can preclude tumor development). It is important to recognize that significant controversy and argument exists over USEPA’s use of threshold approaches to cancer risk assessment, especially when a chemical is not acting through a mutagenic MOA, such as 1,4-dioxane. Historically, scientific fraud and misconduct have been identified in how radiation-induced cancer risks were designated to be a linear, dose-proportional concept (Calabrese 2019). USEPA adopted the Linear No Threshold approach (LNT) emerging from the Biological Effects of Ionizing Radiation’s (BIER I) genetics panel for addressing chemically induced cancer risk. Since USEPA adopted BIER I’s LNT model in the mid-1970s, significant research and analysis on chemical carcinogenesis has been published supporting the use of thresholds as the basis for many types of chemically induced cancers (Lutz et al. 2002; Wolf et al. 2019; Doe et al. 2019; Tubiana et al. 2009; Bogen 2016; Bevan and Harrison 2017; Felter et al. 2018). This is evident in Health Canada’s draft (2018) selection of a “key event” of liver injury preceding the development of liver cancer as the threshold basis for its proposed MAC value for 1,4-dioxane.

The following review addresses the underlying information and approach used by IRIS, OCSPP (draft), and Health Canada (draft) that impact the confidence (uncertainty) in each approach, useful when selecting 1,4-dioxane’s cancer toxicity values in risk assessment and risk management decision making.

Section 2: Rodent Cancer Data for 1,4-Dioxane
At least seven published cancer bioassays have reported various tumors in rats and mice. Many of these studies support dose-response modeling for establishing cancer toxicity values. These data are graphically depicted below as the tumor evidence is described. The reader can review the extensive summaries on cancer provided in USEPA’s 2013 IRIS risk assessment, USEPA’s OCSPP 2019 draft risk evaluation, and Health Canada’s 2018 draft risk evaluation for additional information beyond what is summarized here. There are many other publications relevant to questions around mutagenicity, genotoxicity, and key events relevant to how 1,4-dioxane’s cancer risk can be understood for either risk assessment or risk management applications.

Figure 1 shows the liver tumors (combined adenomas and hepatocellular carcinomas, as well as cholangiocarcinomas) reported in male and female rats and mice. All of the data shown for the Kociba et al. (1974) study were obtained from the unpublished 1971 report that breaks out the tumor data by male and female rats and provides additional noncancer histopathology information. USEPA (2013a) reports that “precise incidences cannot be calculated since the number of rats per group was reported as 28–32” for Hoch-Ligeti, Argus, and Arcos (1970) and Argus et al. (1973). For graphing purposes, the total number of animals per treatment group for the Hoch-Ligeti Argus, and Arcos (1970) data set was assumed to be 30 in Figure 1.

Figure 1. Male and female rat liver tumor response to 1,4-dioxane exposure.
Source: ITRC 1,4-Dioxane Team, 2020.

In rats, saturation of 1,4-dioxane metabolism to its primary metabolite, 2-hydroxyethoxyacetic acid (HEAA; Figures 2 and 3), has been reported to occur between 30 and 100 mg/kg/day (Health Canada 2018). Complete recovery of radio-labeled [¹⁴C] 1,4-dioxane and metabolites from the urine (primary elimination route with recoveries >75%), feces, and expired air (as nonmetabolized 1,4-dioxane or ¹⁴CO₂) has been reported following single or 17 sequential daily doses in rats (Young et al. 1978). This observation—combined with information showing that the oxidation products of 1,4-dioxane do not induce preneoplastic lesions in hepatic tissue of rats dosed 3 days per week for 12 weeks (Koissi et al. 2012)—shows that 1,4-dioxane is not metabolized to a reactive intermediate that covalently binds to subcellular molecules to any appreciable extent (or at all). 1,4-Dioxane’s saturation kinetics is linked to the dosages causing liver tumors and is an important key event consideration. Dose-dependent increases in the percentage of 1,4-dioxane dose expired unchanged as 1,4-dioxane supports the conclusion that metabolic saturation occurs and yields increased internal doses of the parent compound (Young et al. 1978; reviewed in Dourson et al. 2017 and USEPA 2013a).

Figure 2. Metabolic pathways for 1,4-dioxane.
Source: ITRC 1,4-Dioxane Team, 2020. Adapted from Woo, Argus, and Arcos 1977.

Figure 3. 1,4-Dioxane blood concentrations as a function of dose.
Source: ITRC 1,4-Dioxane Team, 2020. Adapted from Young, Braun, and Gehring 1978.

1,4-Dioxane administration, either via drinking water or inhalation, has fairly consistently shown an increase in squamous cell carcinomas of the anterior nose in male rats, but not in mice. These data are shown in Figure 4. For the NCI (1978) study, the data are reported as the incidence for combined squamous cell carcinomas and adenomas for both male and female rats.

Figure 4. Male and female rat nasal tumor response to 1,4-dioxane exposure.
Source: ITRC 1,4-Dioxane Team, 2020.

Various other tumors are reported as treatment-related or -elevated in the published rat cancer bioassays. Peritoneal mesotheliomas, a unique background tumor observed in aging rats, is fairly consistently elevated due to treatment (Figure 5). Next are the somewhat unique subcutis fibromas observed in rats following 1,4-dioxane treatment. Much less consistently, and only at higher or the highest dosages, sometimes exceeding the maximally tolerated dose, tumors of the breast and kidney, and the Zymbal gland have been reported. Figure 5depicts all except for Zymbal gland tumors. For Kociba et al. (1971), no male mammary tumors were reported in the study, while the female’s mammary tumors from the study represent combined fibroadenomas (majority), cystoadenoma, and adenocarcinoma. Regarding subcutis fibromas observed in rats, in humans, collagenous fibroma (desmoplastic fibroblastoma) “is an extremely rare benign soft tissue tumor of fibroblastic origin,” where the “majority of reported cases have been located in the deep subcutis, fascia, aponeurosis, or skeletal muscle of the extremities, limb girdles, or head and neck regions” (Huang et al. 2002).

Figure 5. Other (not liver or nasal) tumor response to 1,4-dioxane exposure.
Source: ITRC 1,4-Dioxane Team, 2020.

In general, lifetime 1,4-dioxane dosages exceeding 10 mg/kg/day are required to produce excesses in these specific tumor types in rats. This point of departure aligns with the current position that 1,4-dioxane to HEAA metabolic saturation occurs somewhere between 30 and 100 mg/kg/day, pointing to the buildup of 1,4-dioxane blood concentrations as a necessary key event, and potentially a molecular initiating event using the Adverse Outcome Pathway construction of a mode-of-action.

There are two cancer studies in mice , both using a drinking water exposure design (NCI 1978; Kano et al. 2009). Unlike rats, only treatment-related liver tumors have been observed in the mouse studies. Figure 6 shows the liver tumor response as a function of dose in male and female mice.

Figure 6: Male and female mouse liver tumor response to 1,4-dioxane exposure.
Source: ITRC 1,4-Dioxane Team, 2020.

In summary, the most consistent tumor responses in mice and rats are liver, nasal, male peritoneal mesotheliomas, and subcutis fibroma. The female mouse liver tumors reflect the most sensitive tumor response, but this tumor response has been challenged for various reasons as of limited use for risk assessment purposes. The other rat/mouse liver tumors and nasal tumors offer a reasonable treatment-related cancer response for deriving toxicity factors. The male peritoneal mesotheliomas, although treatment related, are a unique tumor response, as are the subcutis fibromas (see below). Neither the mammary (adenoma or fibroadenoma) nor kidney (carcinoma) tumors present a robust and sensitive response. Critical evaluation of the candidate oral CSF values presented in Table 5-10 of USEPA 2013a clearly shows that the hepatic tumor responses (regardless of species or sex) are more sensitive than tumor responses in other tissues across species. This information is reflected in the choice of endpoints selected for derivation of cancer toxicity values by USEPA and Health Canada.

Section 3: Regulatory Interpretation of 1,4-Dioxane Cancer Bioassay Results
The USEPA IRIS program (2013a), USEPA’s OCSPP draft risk evaluation (2019), and Health Canada’s draft risk evaluation (2018) use different data and approaches to derive cancer toxicity values for 1,4-dioxane (see Table 1). Additional critical information is included here for each of these derivations, to provide supporting information for the data already included in Table 1 and the preceding figures. Each agency’s original draft risk assessment reports should be studied for additional details of the risk evaluation exercise.

USEPA

USEPA’s Integrated Risk Information System
The IRIS CSFs for 1,4-dioxane were finalized in 2013. IRIS selected the Kano et al. (2009) drinking water study to derive the oral CSF for 1,4-dioxane based on liver adenomas and carcinomas occurring in female Crj:BDF₁ mice. USEPA (2013a) selected the mouse liver tumor response because it was the most sensitive response (Figure 7).

Figure 7. Female mouse liver tumors reported in Kano et al. (2009).
Source: ITRC 1,4-Dioxane Team, 2020. Adapted from Kano et al. 2009.

USEPA IRIS (USEPA 2010) found that the multistage model did not adequately fit the female mouse liver cancer data and, as a result, chose to rely on an alternative model (log-logistic model) and a notably higher than typical benchmark response (e.g., 30% and 50% modeled independently) for dose-response modeling. Notably, USEPA’s (2019) OCSPP draft risk evaluation elected to reject the Kano et al. (2009) female mouse liver tumors as a basis for establishing an oral CSF. USEPA (2013a, Table 5-10) IRIS also selected a linear oral CSF of 1.0 x 10^-1 mg/kg/day from a number of tumor endpoints; see Table С-2.

Table С-2. Benchmark dose modeling and CSF results from USEPA (2013a)

Study	Gender/strain/species	Tumor type	BMD_HED^a (mg/kg-day)	BMDL_HED^a (mg/kg-day)	Oral CSF (mg/kg-day)^-1
Kano et al. (2009)	Male F344/DuCrj rats^b	Hepatocellular adenoma or carcinoma	17.43	14.33	7.0 × 10^-3
	Female F344/DuCrj rats^c		19.84	14.43	6.9 × 10^-3
	Male Crj:BDF1 mice^d		5.63	2.68	3.7 × 10^-2
	Female Crj:BDF1 mice^d		0.83	0.55	1.8 × 10^-1
	Female Crj:BDF1 mice^{d, e}		3.22^e	2.12^e	1.4 × 10^-1
	Female Crj:BDF1 mice^{d, f, h}		7.51^f	4.95^f	1.0 × 10^-1
	Female F344/DuCrj rats^g	Nasal squamous cell carcinoma	94.84	70.23	1.4 × 10^-3
	Male F344/DuCrj rats^g	Nasal squamous cell carcinoma	91.97	68.85	1.5 × 10^-3
	Male F344/DuCrj rats^b	Peritoneal mesothelioma	26.09	21.39	4.7 × 10^-3
	Female F344/DuCrj rats^d	Mammary gland adenoma	40.01	20.35	4.9 × 10^-3
Kociba et al. (1974)	Male and female (combined)/Sherman rats^g	Nasal squamous cell carcinomas	448.24	340.99	2.9 × 10^-4
Kociba et al. (1974)	Male and female (combined)/Sherman rats^b	Hepatocellular carcinomas	290.78	240.31	4.2 × 10^-4
NCI (1978)	Male Osborne Mendel rats^d	Nasal squamous cell carcinomas	16.10	10.66	9.4 × 10^-3
	Female Osborne Mendel rats^d	Nasal squamous cell carcinomas	40.07	25.82	3.9 × 10^-3
	Female Osborne Mendel rats^d	Hepatocellular adenoma	28.75	18.68	5.4 × 10^-3
	Female B6C3F₁ mice^c	Hepatocellular adenoma	23.12	9.75	1.0 × 10^-2
	Male B6C3F₁ miceⁱ	Hepatocellular adenoma or carcinoma	87.98	35.67	2.8 × 10^-3

^aValues associated with a BMR of 10% unless otherwise noted.
^bProbit model, slope parameter not restricted.
^cMultistage model, degree of polynomial = 2.
^dLog-logistic model, slope restricted ≥1.
^eValues associated with a BMR of 30%.
^fValues associated with a BMR of 50%.
^gMultistage model, degree of polynomial = 3.
^hSee BMDS model output in USEPA (2013a), Figure D-12.
ⁱGamma model.

These various CSFs depict a range of approximately two orders of magnitude, with most of them being three or more times less conservative than the 1 x 10^-1 mg/kg/day CSF.

The combined tumor incidence reported from the male rat inhalation bioassay, published in Kasai et al. (2009), was used to derive the IUR (USEPA 2013a) (see Figure 8 and Figure 4 for the rat tumor data reported in Kasai et al. [2009]).

Figure 8. Combined tumor incidence data reported in Kasai et al. (2009).
Source: ITRC 1,4-Dioxane Team, 2020. Adapted from Kasai et al. 2009.

Table 5-11 provided in the IRIS (2013a) risk evaluation lists individual tumor BMCL₁₀ estimates and the composite BMCL₁₀ value derived from the Kasai et al. (2009) inhalation study in male F344 rats (see Table C-3 below).

Table C-3. Dose-response modeling summary results for male rat tumors associated with inhalation exposure to 1,4-dioxane for 2 years (USEPA 2013a)

Tumor Type^a	Multistage model degree^b	Point of departure^c				IUR estimate (µg/m³)^-1
		Bioassay exposure concentration (ppm)		HEC (mg/m³)^d
		BMC₁₀	BMCL₁₀	BMC₁₀	BMCL₁₀
Nasal cavity squamous cell carcinoma	1	1,107	629.9	712.3	405.3	2.5 × 10 ^-7
Hepatocellular adenoma or carcinoma	1	252.8	182.3	162.7	117.3	8.5 × 10 ^-7
Renal cell carcinoma	3	1,355	1,016	872	653.7	1.5 × 10 ^-7
Peritoneal mesothelioma	1	82.21	64.38	52.89	41.42	2.4× 10 ^-6
Mammary gland fibroadenoma	1	1,635	703.0	1,052	452.4	2.2× 10 ^-7
Zymbal gland adenoma	3	1,355	1,016	872	653.7	1.5 × 10 ^-7
Subcutis fibroma	1	141.8	81.91	91.21	52.70	1.9× 10 ^-6

BMDS MS_Combo total tumor analysis^f		40.4	30.3	26.0	19.5	5.0 × 10 ^-6

^aTumor incidence data from Kasai et al. (2009).
^bBest-fitting multistage model degree (p>0.0, lowest AIC). See Appendix G of USEPA (2013a) for modeling details.
^cBMC = Concentration at specified extra risk (benchmark dose); BMCL = 95% lower bound on concentration at specified extra risk.
^dHuman continuous equivalent estimated by multiplying exposure by [(6 hours)/(24 hours) × (5 days)/(7days) × (molecular weight of 1,4-dioxane)/24.45].
^eThe IUR (µg/m³)^-1was derived from the BMCL₁₀, the 95% lower bound on the concentration associated with a 10% extra cancer risk—specifically, by driving the BMR (0.10) by the BMCL₁₀—thus representing an upper bound, continuous lifetime exposure estimate of cancer potency.
^fResults in the table are from the BMDS MS_Combo model. Additionally, Bayesian analysis using WinBUGS was performed and yielded similar results.

Notably, the addition of peritoneal mesotheliomas (and possibly subcutis fibromas) significantly increased the unit risk per mg/m³of exposure. Both of these tumors are of questionable human relevance.

IRIS elected to apply a linear model to the cancer data, concluding that the Japanese cancer bioassays (Kano et al. 2009; JBRC 1998) demonstrated female mouse tumors (liver) occurring absent evidence of cytotoxic lesions. IRIS noted it did not have noncancer histopathological data from the Kociba et al. (1974) drinking water study and did not seek to obtain the data from the unpublished 1971 toxicology study report. USEPA also concluded that evidence was lacking for demonstrating a relationship between increasing 1,4-dioxane blood and tissue levels to give importance to the metabolism saturation kinetics of 1,4-dioxane to 2-hydroxyethoxyacetic acid (HEAA). It appeared that the IRIS (2013a) MOA conclusions centered on the requirement that cytotoxicity be evident to support increased cell proliferation (regenerative repair). IRIS (2013a) further stated that other tumors (nasal, peritoneal mesotheliomas, and mammary gland) had no MOA evidence.

In the case of 1,4-dioxane, there is insufficient biological support to identify key events and to have reasonable confidence in the sequence of events and how they relate to the development of tumors following exposure to 1,4-dioxane; thus, the data are not strong enough to ascertain the mode of action applying the Agency’s mode of action framework. (USEPA 2013a, 136)

Importantly, USEPA IRIS (USEPA 2013a) did not conclude that a mutagenic MOA explains how 1,4-dioxane causes rodent tumors. Regardless, IRIS’ conclusion that insufficient information existed on any MOA led it to adopt a linear modeling approach in accordance with USEPA’s Guidelines for Carcinogen Risk Assessment (2005). Readers are directed to the 2013a IRIS risk assessment for a complete explanation for USEPA’s justification for rejecting a threshold modeling approach for the rat and mouse tumor response.

USEPA’s Office of Chemical Safety and Pollution Prevention Draft Risk Assessment
In 2019, OCSPP published its draft risk evaluation of 1,4-dioxane according to the newly implemented LCSA or Toxic Substances Control Act (TSCA) reform. OCSPP selected the rat cancer bioassay data from Kano et al. (2009) to derive the draft oral CSF and data from Kasai et al. (2009) to derive the draft inhalation unit risk (IUR) value. Similar to the IRIS (2013a) risk assessment, OCSPP concluded a linear modeling of the data was warranted, claiming insufficient MOA evidence for liver tumors and no MOA for the other tumor types. Like IRIS, OCSPP concluded that a mutagenic MOA was not operative, although there is some evidence of high-dose genotoxicity. In its discussion of MOA, the OCSPP draft report stated that liver tumors occurred in female and male mice absent cytotoxicity: “However, available data do not rule out some role for either genotoxicity or cytotoxicity (or both) in a possible mode of action” (USEPA 2019e [MG41] , 99). USEPA also speculated that “alternative metabolic pathways (e.g., not CYP450) may be present,” arguing against the proposed buildup of 1,4-dioxane in blood after the metabolism saturation dose of 1,4-dioxane is achieved (around 30 to 100 mg/kg/day in rats and approximately 200 mg/kg/day in mice). Notably, no evidence of an alternative metabolite to HEAA has ever been reported, so the opinion pointing to some unknown metabolite as the explanation for 1,4-dioxane carcinogenicity is without scientific support at this time. USEPA also concluded that dosages causing cytotoxicity and increased cell proliferation were greater than dosages causing tumors. This holds true for the female mouse liver tumor response reported in Kano et al. (2009), but readers should examine this for the other tumors, especially rat liver and nasal tumors, along with the relationship between dosages causing metabolic saturation as well as increased tumors. OCSPP did not access the Kociba et al. (1971) unpublished noncancer histopathology data to determine how these key events relate to the apical outcome.

For the draft IUR, a number of BMCL₁₀ values were derived for portal of entry (nasal, Zymbal gland) and systemic (liver adenomas/carcinomas, renal cell carcinomas, peritoneal mesotheliomas, mammary gland fibroadenomas, and subcutis fibromas) from the Kasai et al. (2009) inhalation study in male F344 rats (USEPA 2013a, 116–117, Table 4-9) plotted in Figure 4 above and shown below in Table C-4. This is identical to the IRIS approach for generating the IUR, although the IRIS value of 5 x 10^-6 (µg/m³)^-1 resulted in a fivefold greater theoretical risk compared to the draft 1 x 10^-6 (µg/m³)^-1 IUR from OCSPP, due solely to OCSPP’s use of the MS_Combo program to model combined tumor incidences.

Table C-4. Dose-response modeling summary results for male rat tumors associated with inhalation exposure to 1,4-dioxane for 2 years (USEPA 2019e, Table 4-9)

Portal of Entry Effects
Tumor Type^a	Multistage model degree^b	BMC₁₀ (ppm)^c	BMCL₁₀ (ppm)^c	BMCL_ADJ (worker ppm)^d	BMCL_HEC(worker mg/m³)^e,f	IUR estimate^g (µg/m³)^-1
Nasal cavity squamous cell carcinoma	1	1,107	630	473	221	4.52E-07
Zymbal gland adenoma	1	1,975	958	719	337	2.97E-07
MS_Combo portal of entry		709	449	337	158	6.34E-07
Systemic Effects
Tumor Type^a	Multistage model degree^b	BMC₁₀ (ppm)^c	BMCL₁₀ (ppm)^c	BMCL_ADJ (worker ppm)^d	BMCL_HEC(worker mg/m³)^e,f	IUR estimate^g (µg/m³)^-1
Hepatocellular adenoma or carcinoma	1	253	182	137	492	2.03 E-07
Renal cell carcinoma	1	1,975	958	719	2,589	3.86E-08
Peritoneal mesothelioma	1	82.2	64.4	48	174	5.75E-07
Mammary gland fibroadenoma	1	1,635	703	527	1,900	5.26E-08
Subcutis fibroma	1	142	81.9	61.4	221	4.52E-07
MS_Combo systemic (including liver)		41.2	32.8	24.6	88.6	1.13E-06
MS_Combo systemic (omitting liver)		49.2	37.9	28.4	102	9.76E-07
Portal of Entry Effects and Systemic Effects
Tumor Type^a	Multistage model degree^b	BMC₁₀ (ppm)^c	BMCL₁₀ (ppm)^c	BMCL_ADJ (worker ppm)^d	BMCL_HEC(worker mg/m³)^e,f	IUR estimate^g (µg/m³)^-1
MS_Combo portal of entry+ systemic (including liver)		38.9	31.3	23.5	84.6	1.18E-06
MS_Combo portal of entry+ systemic (omitting liver)		46.0	35.9	26.9	97.0	1.03E-06

^aTumor incidence data from Kasai et al. (2009). Data quality evaluations for all endpoints are high.
^bBest-fitting multistage model degree following current BMDS guidance (USEPA 2014, 2012). Model selections for renal cell carcinoma and Zymbal gland adenoma differ from USEPA (2013a) IRIS assessment.
^cBMC₁₀= Concentration at specified extra risk (benchmark dose); BMCL = 95% lower bound on concentration at specified extra risk.
^dPOD_ADJ(ppm) = BMCL₁₀ × 6 hours ÷ 8 hours.
^ePOD_ADJ(ppm) values were converted to mg/m³ values based on the following: BMCL_ADJ(ppm) × molecular weight of 1,4-dioxane (88.1 g/mole) ÷ gas constant at 760 mm Hg and at 25°C.
^fPOD_HEC = BMCL_ADJ × DAF (i.e., RGDR_ET).
^eThe IUR (µg/m³)^-1was derived from the BMCL₁₀, the 95% lower bound on the concentration associated with a 10% extra cancer risk—specifically, by driving the BMR (0.10) by the BMCL₁₀—thus, representing an upper bound, continuous lifetime exposure estimate of cancer potency.
^hPOD_HEC(mg/m³) = BMCL_ADJ× DAF (i.e., (Hb/g) A ÷ (Hb/g) H)ⁱPOD_HEC(mg/m³) for the MS_Combo, including both portal of entry and systemic effect used the DAF of (Hb/g) A ÷ (Hb/g) H.

The OCSPP draft (USEPA 2019e) calculated a “dermal” CSF from the inhalation data of Kasai et al. (2009) and the oral drinking water studies of Kano et al. (2009) and Kociba et al. (1974)—that is, converting the CSFs from the inhalation and oral routes to a dermal exposure pathway. These values were summarized in Table 4-12, 126–127, of the 2019 draft report, as shown in Table C-5.

Table C-5. Dose-response modeling summary results for oral CSFs and route-to-route extrapolated dermal CSFs (adapted from USEPA 2019e, Table 4-12)

Study (Data Quality)^a	Gender/strain/ species	Endpoint	BMR	MS*	BMD (mg/kg-d)	BMDL (mg/kg-d)	BW_A(g)	POD^b (mg/kg-d)	Oral CFS (mg/kg-d)^-1	Dermal CSF^c (mg/kg-d)^-1
Kano et al. (2009) (high)	Male F344/DuCrj rats	Nasal squamous cell carcinoma	10%	2	365	242	432	65.6	1.5E-3	4.9E-05
		Peritoneal mesothelioma	10%	2	77.7	35.4		9.60	1.0E-02	3.3E-04
		Hepatocellular adenoma or carcinoma	10%	2	61.7	28.3		7.67	1.3E-02	4.2E-04
		Subcutis fibroma	10%	1	154	85.0		23.0	4.3E-03	1.4E-04
		MS_Combo (excluding liver)	10%	N/A	55.2	28.1		7.62	1.3E-02	4.2E-04
		MS_Combo (including liver)	10%	N/A	35.1	17.8		4.83	2.1E-02	6.7E-04
	Female F344/DuCrj rats	Nasal squamous cell carcinoma	10%	1	376	214	267	51.4	1.9E-03	6.2E-05
		Mammary gland adenoma	10%	1	177	99.1		23.8	4.2E-03	1.3E-04
		Hepatocellular adenoma or carcinoma	10%	2	79.8	58.1		14.0	7.1E-03	2.3E-04
		MS_Combo (excluding liver)	10%	N/A	120	76.5		18.4	5.4E-03	1.7E-04
		MS_Combo (including liver)	10%	N/A	57.6	41.6		10.0	1.0E-02	3.2E-04
	Male Crj:BDF1 mice	Hepatocellular adenoma or carcinoma	10%	1	71.0	44.0	47.9	6.88	1.5E-02	4.7E-04
Kociba et al. (1974) (high)	Sherman rats (M+F)	Nasal squamous cell carcinomas	10%	2	1,981	1,314	325	332	3.0E-04	9.6E-06
Kociba et al. (1974) (high)	Sherman rats (M+F)	Hepatocellular carcinoma	10%	1	940	584	325	147	6.8E-04	2.2E-05
NCI (1978) (low)	Female OM rats	Nasal squamous cell carcinomas	10%	1	176	122	310	30.4	3.3E-03	1.1E-04
	Female OM rats	Hepatocellular adenoma	10%	1	132	94.1	310	23.5	4.3E-03	1.4E-04
	Male B6C3F₁ mice	Hepatocellular adenoma or carcinoma	10%	1	164	117	32	16.5	6.1E-03	1.9E-04
	Female B6C3F₁ mice	Hepatocellular adenoma or carcinoma	10%	1	49.1	38.8	30	5.40	1.9E-02	5.9E-04

^aApplies to all of the endpoints listed in this table for each study.
^bPOD = dose × (BW_A/BW_H)^0.25. BW_H = 80 kg. BW_A values are study-specific (obtained from Table 5-9 of the 1,4- dioxane IRIS assessment).
^cDermal CSF (mg/kg-d) ^-1 = Oral CSF (mg/kg-d)^-1 × 3.2% (dermal absorption) ÷ 100% (oral absorption).

However, as the introduction of this review mentioned, the most likely route of environmental exposure is from drinking water, with dermal contact being insignificant.

Health Canada’s Draft Health-Based Value for 1,4-Dioxane in Drinking Water
Health Canada’s draft (2018) selected the Kociba et al. (1974) rat liver cancer study to derive its draft cancer toxicity value. However, unlike USEPA, Health Canada concluded that the MOA justified a threshold approach. Furthermore, it selected a key event (a noncancer liver response) rather than using the dose-tumor response to develop a toxicity value protective against developing cancer (i.e., a tolerable daily intake [TDI]). The logic for regulating on a cancer MOA key event evokes the concepts of MOA and adverse outcome pathways (AOPs), whereby preventing a critical step (a key event and its key event relationships) in the process of tumor promotion yields safe levels of exposure that will prevent the development of cancer. Canada derived a draft Canadian Maximum Allowable Concentration (MAC) value of 50 µg/L (or 50 parts per billion). It is important to note that Canada further reduced the cancer toxicity value by one-fifth using a default allocation factor for drinking water, allowing the theoretical cancer risk from drinking water to account for 1,4-dioxane exposures coming from nondrinking water sources. Thus, if no other exposure to 1,4-dioxane outside of drinking water occurs, the MAC could be five times higher—250 parts per billion—while still not exceeding the TDI.

In evaluating the MOA, Health Canada (2018) reached the following conclusion: “Studies in both rats and mice found liver toxicity to be the most sensitive endpoints of concern; thus, this section (Section 9.3 of the 2018 draft report) focuses on the mode of action (MOA) evidence for liver tumors” (34). With respect to the saturation of metabolism as a key event, Health Canada found that the parent compound is “the toxic moiety and not a metabolite” (36). At 1,4-dioxane blood levels below 30 µg/ml, up to an oral dose of 10 mg/kg/day, no adverse effects in rats occur. To place this 30 µg/ml NOAEL blood concentration into perspective, the National Health and Nutrition Examination Study (NHANES) recently reported no detectable 1,4-dioxane blood concentrations in over 3,000 U.S. citizens (2017). The lower limit of detection and the imputed 1,4-dioxane blood levels were 0.5 ng/ml and 0.35 ng/ml, respectively. Applying the imputed 0.35 ng/ml (0.00035 µg/ml) against the 30 µg/ml NOAEL blood concentration generates a margin of exposure of 85,700.

The draft Health Canada (2018) assessment described other key events, including increased DNA synthesis (direct mitogenic stimulation) and regenerative repair triggering secondary mitogenic stimulation. Overall, it reached the following conclusion:

Analysis of the weight of evidence indicates that 1,4-dioxane is not a mutagen and does not cause DNA repair or initiation; however, 1,4-dioxane appears to promote tumors, stimulate DNA synthesis, and evoke cancer following saturation of 1,4-dioxane metabolism or elimination. Analysis supports a non-genotoxic MOA involving cytotoxicity followed by regenerative hyperplasia and stimulation of endogenously formed mutations. Since 1,4-dioxane acts through a non-genotoxic MOA and is known to operate via non-linear kinetics, a non-linear (threshold) risk assessment is considered appropriate. (39)

Hepatocellular necrosis for both male and female rats reported in the Kociba et al. (1971, 1974) drinking water study were modeled to yield a BMDL₅ of 5.4 mg/kg. For context, Figure 9 shows the combined male and femaleliver tumor response to 1,4-dioxane, along with a number of key events.

Figure 9. Combined male and female liver tumors and noncancer liver toxicity findings from Kociba et al. (1974) and the unpublished Kociba et al. (1971) report.
Source: ITRC 1,4-Dioxane Team, 2020. Adapted from Kociba et al. 1971, 1974.

This graph illustrates the various key events relative to the combined liver tumors (apical outcome). The hepatocellular cytoplasmic vacuolar degeneration and hepatic necrosis observations of rats with no, mild, moderate, and severe findings reported in Kociba et al. (1971) were weighted and combined to tally a severity-number score. Because these histopathology observations were obtained at study termination, interim information ascertaining an earlier emergence of hepatotoxicity to better understand the key event–apical outcome relationship is not available. It can be seen that histopathology associated with degenerative repair (e.g., increased hepatocellular cytoplasmic vacuolar degeneration and hepatocellular degeneration) and increased proliferation (increased nodules and hyperplastic nodules as well as bile duct hyperplasia) either exceed the tumor response or parallel liver tumor development. In addition to the 2-year cancer bioassays, several other publications reveal important data that identify key events by which liver, nasal, and kidney cancer is promoted by noncancer toxicity. The key events thus noted represent a key event relationship that supports a regenerative repair proliferative tumor promotion MOA, as noted by Health Canada (2018)’s draft assessment.

For additional context, the noncancer hepatic histopathology for male and female rats data are shown in Tables C-6 and C-7, including the last table, where the foci data are weighted for severity and combined to calculate a proliferative response index.

Table C-6. Combined male and female rat noncancer and cancer liver histopathology findings

Combined Male and Female Rat Liver H&E (Kociba et al. 1971, 1974)
Dose	Discrete pale foci	Increased nodules	Hepatocelluar cytoplasmic vacuolar degeneration			Hepatocellular necrosis			Anisonucleosis
mg/kg/day			Min	Mod	Severe	Min	Mod	Severe
0	5	1	8	1	0	1	1	1	1
9.6	5	1	5	0	5	7	1	0	1
94	14	4	21	5	2	14	7	2	2
1,015	31	19	16	16	0	19	24	12	34

Combined Male and Female Rat Liver H&E (Kociba et al. 1971, 1974)
Dose mg/kg/day	Bile duct hyperplasia	Hyperplastic nodules	Hepatocellular carcinoma	Cholangiolar carcinoma	Cholangioma
0	10	1	1	1	0
9.6	1	1	0	0	0
94	9	3	1	0	0
1,015	23	9	10	0	2

Table C-7. Combined male and female rat noncancer and cancer liver histopathology findings adjusted for severity

Combined Male and Female Rat Liver H&E (Kociba et al. 1971, 1974) Degeneration and Necrosis Adjusted for Severity and Summed
Dose mg/kg/day	Hepatocellular cytoplasmic vacuolar degeneration*				Hepatocellular necrosis*				Combined liver tumors***
	Adj. Min	Adj. Mod	Adj. Severe	Total**	Adj. Min	Adj. Mod	Adj. Severe	Severe Total**
0	8	2	0	10	1	2	15	18	2
9.6	5	0	0	5	7	2	18	27	0
94	21	10	6	37	14	14	21	49	1
1,015	16	21	15	52	19	48	24	91	12

* Adjusted (Adj.): min x 1; mod x 2; severe x 3.
** Total summed across the adjusted scores.
*** Combined liver tumors were hepatocellular carcinomas, cholangiocarcinomas, and cholangiomas.

To the 5.4 mg/kg/day BMDL₅ estimate, Health Canada’s (2018) draft assessment applied a combined uncertainty factor of 1,000 (10x for interspecies variability, 10x for intraspecies variability, and 10x for database deficiencies, including a lack of sufficient reproductive-developmental toxicity information). In addition, a default allocation factor (similar to a relative source contribution term) of 0.2 was applied to yield a TDI of 5.4 μg/kg/day and associated MAC value of 50 µg/L for drinking water.

Section 4: Confidence/Uncertainty in 1,4-Dioxane Cancer Toxicity Values
There is significant debate in the scientific and regulatory communities as to the scientific support for a threshold approach, or the decision to use a precautionary-based linear approach, for cancer risk modeling of 1,4-dioxane exposures. Different interpretations of the carcinogenicity and toxicology information have generated a greater than 100-fold range in 1,4-dioxane’s cancer-based drinking water criteria, depending on whether a linear or threshold modeling approach is used to evaluate the dose-cancer incidence relationship (USEPA 2013a, 2019; Health Canada 2018). This difference generates confusion over which approach to use, which resulting drinking water criteria is sufficiently health protective, and which criteria is more scientifically correct. USEPA’s linear cancer toxicity value, known as the CSF, estimates a population lifetime extra cancer risk (i.e., above background) at any level of exposure (e.g., 1 in 10,000 people to 1 in 1,000,000 people when exposed to 1,4-dioxane drinking water concentrations from 3.5 to 0.035 µg/L). Health Canada’s draft evaluation (2018), on the other hand, applies a threshold approach. Its draft toxicity value determines that no population or individual cancer risk from 1,4-dioxane exposure exists if the drinking water concentration is below 50 parts per billion. The confidence and uncertainty in either the linear or threshold approach depend on how one interprets the biological basis (known as the MOA) by which 1,4-dioxane induces different tumor types in rats and mice. At the same time, a range of risk approaches provides risk managers with options, depending on the health-protective nature of the risk assessment and exposure circumstances.

The information that follows addresses aspects of the linear modeling of the cancer risk adopted by USEPA and the draft threshold approach used by Health Canada. The decision to use a linear versus a threshold approach requires an understanding of the toxicology data, interpretations, agency-specific regulatory guidance, and assumptions, all of which go toward the question of confidence and uncertainty in 1,4-dioxane risk assessment. Information provided in the following sections reviews and summarizes:

The fundamental concepts of MOA that underpin an agency’s decision to apply either a linear or threshold approach to cancer risk assessment
Why the USEPA and Health Canada cancer toxicity values are different
The scientific justification for a threshold approach over a linear modeling of the cancer risk

Publications such as Dourson et al. (2014, 2017) and risk evaluations published by regulatory agencies other than USEPA and Health Canada provide perspective on the threshold versus linearity question.

4.1 Fundamental Concepts of MOA

How a chemical causes cancer, also known as its cancer MOA, plays a critical role in the modeling method a government agency will employ in assessing the cancer risk and assigning toxicity values used in setting safe levels of exposure. Cancer MOA(s) can be simply divided into:

Carcinogens that are mutagens, that alter the genetic code, and bring about the expression of oncogenes (or cancer promoting genes). These mutations are in addition to the many mutations that spontaneously occur every day in each cell due to natural causes (e.g., oxidative stress, mutagens in our diet). It is held by some, including USEPA, that mutagenic carcinogens should have their cancer risk treated as a linear function of exposure.
Carcinogens that are not mutagens but stimulate the rate at which cells divide and replicate, thereby facilitating the buildup of naturally occurring mutations. Nonmutagenic carcinogens are equated with tumor promotion, whereby the chemical promotes (speeds up) the background biology of carcinogenesis and is reversible provided the tumor promotion stimulus is discontinued at an early stage of tumor promotion.

A number of publications have laid out the framework for assembling an MOA-human relevance evaluation of the underlying toxicology data (Meek et al. 2014). By applying an examination of the “Evolved Bradford Hill Considerations” for examining the cancer MOA information for 1,4-dioxane described in Meek et al. (2014), a MOA weight of evidence assessment can be assembled (see the earlier discussion of MOA in Section 3).

In addition to the MOA-human relevance approach, the Organisation for Economic Co-operation and Development (OECD) developed an AOP framework (2019) that has further refined the MOA methodology in a systematic manner. The framework begins with the molecular initiating event and proceeds through higher levels of biology to include information on biochemical, cellular, tissue, organ, and whole organism response to a chemical leading to an apical outcome like cancer. The AOP is another useful tool that is built on and compliments the MOA-human relevance framework methodology. Readers can consult the OECD AOP (2019) website for AOPs that 1,4-dioxane’s tumor promotion MOA may align with (https://www.oecd.org/chemicalsafety/testing/adverse-outcome-pathways-molecular-screening-and-toxicogenomics.htm).

4.2 USEPA’s Linear Handling of 1,4-Dioxane Cancer Risk-Exposure Relationship
USEPA concluded that 1,4-dioxane is not mutagenic at environmentally relevant doses (USEPA 2010, 2013a). USEPA recognized that 1,4-dioxane will not add to the mutation burden that spontaneously occurs naturally in the cells of the body. However, it concluded that not enough was known about 1,4-dioxane’s nonmutagenic cancer MOA in producing liver cancer to justify a threshold approach. For the other cancer types caused by 1,4-dioxane (tumors of the nasal mucosa, peritoneal mesotheliomas in male rats, kidney tumors, subcutis fibromas of the skin, Zymbal gland, and mammary gland tumors), USEPA claimed that no MOA information was available. For example, following a review of the published evidence, USEPA’s OCSPP (2019) stated the following about the MOA in its draft risk assessment:

Kociba et al. (1974) reported hepatic degeneration and regenerative hyperplasia at or below dose levels that produced liver tumors, but incidence for these effects was not reported. Therefore, a dose-response relationship could not be evaluated, and the events cell proliferation and hyperplasia are not supported by available data. Finally, the doses in hepatotoxicity studies where cytotoxicity and cell proliferation were observed were greater than cancer bioassay dose levels. Integrating data across studies, dose-response relationships between cytotoxicity and tumor formation are not well established in the rat and mouse data and are inconsistent among bioassays and across exposure duration.

However, an unpublished report by Kociba et al. (1971) is available upon request and provides the incidence data for key events (these data were evaluated by Health Canada [2018] and Dourson et al. [2014, 2017] and are discussed in more detail earlier in Section 3).

One policy/guidance reason why USEPA adopted a threshold approach is found on page 3-22 of the 2005 Cancer Guidelines (USEPA 2005a):

When the weight of evidence evaluation of all available data are insufficient to establish the mode of action for a tumor site and when scientifically plausible based on the available data, linear extrapolation is used as a default approach, because linear extrapolation generally is considered to be a health-protective approach. Nonlinear approaches generally should not be used in cases where the mode of action has not been ascertained. Where alternative approaches with significant biological support are available for the same tumor response and no scientific consensus favors a single approach, an assessment may present results based on more than one approach.

USEPA concluded that insufficient evidence exists to support a threshold MOA for the liver, even though it has clearly ruled out a mutagenic MOA for 1,4-dioxane. As a result, it relies on the policies described in the 2005 Cancer Guidelines to default to a linear low-dose extrapolation approach. However, the other option for USEPA is to derive both a linear and nonlinear toxicity value, as specified in the 2005 Cancer Guidelines:

A nonlinear approach should be selected when there are sufficient data to ascertain the mode of action and conclude that it is not linear at low doses and the agent does not demonstrate mutagenic or other activity consistent with linearity at low doses. Special attention is important when the data support a nonlinear mode of action but there is also a suggestion of mutagenicity. Depending on the strength of the suggestion of mutagenicity, the assessment may justify a conclusion that mutagenicity is not operative at low doses and focus on a nonlinear approach, or alternatively, the assessment may use both linear and nonlinear approaches.

Both linear and nonlinear approaches may be used when there are multiple modes of action. If there are multiple tumor sites, one with a linear and another with a nonlinear mode of action, then the corresponding approach is used at each site. If there are multiple modes of action at a single tumor site, one linear and another nonlinear, then both approaches are used to decouple and consider the respective contributions of each mode of action in different dose ranges. For example, an agent can act predominantly through cytotoxicity at high doses and through mutagenicity at lower doses where cytotoxicity does not occur. Modeling to a low response level can be useful for estimating the response at doses where the high-dose mode of action would be less important.

USEPA (2013a) did not indicate why it did not perform the evaluation using linear and nonlinear approaches.

4.3 Health Canada’s (Draft) Adoption of a Threshold Approach
During a toxicological evaluation, Health Canada (2010) requires that a weight of evidence analysis be performed to classify a substance as a potential human carcinogen and to determine the most appropriate shape of the cancer dose-response curve in the low-dose region, given the available evidence. In so doing, Health Canada (2010) determines if a substance should be treated as a nonthreshold of threshold carcinogen based on consideration of the following questions:

Has the chemical or its metabolites been associated with relevant malignant tumors in animals or people? (Did it increase the number or type of tumors? Did it decrease the development time for tumors to occur?)
Has the chemical or its metabolites been found to interact with DNA? (Does the weight of evidence from a battery of genotoxicity tests indicate direct mutagenic action?)
Is the chemical structure of the substance very similar to any other chemical that is considered to be a genotoxic carcinogen?
Do supporting or corroborative occupational or other studies on human subjects exist?

Health Canada’s guidance goes on: “If the toxicological data indicate a ‘yes’ response to questions (1) and either (2) or (3), it is probably most appropriate (and conservative) to assume that the chemical is a non-threshold-response chemical, particularly in the absence of clear evidence to the contrary.” As a result, Health Canada’s (2010) approach recommends that a nonthreshold approach be used when the weight of evidence indicates that a substance induces cancer via direct mutagenic action and a substance, or its metabolites, have been associated with relevant tumors in animals.

Applying these questions to evaluation of 1,4-dioxane, Health Canada (2018) concluded in its draft assessment that the available data indicate that “the pattern of genotoxicity is inconsistent with a MOA where genotoxicity is an early and influential key event in the carcinogenic MOA.” USEPA (2013a) actually reached the same conclusion. Simultaneously, Health Canada (2018) concluded in its draft assessment that sufficient information demonstrated that 1,4-dioxane increased cell proliferation facilitating the accumulation of background mutations, thereby promoting and enhancing cancers that naturally develop in rats and mice, thereby supporting a threshold approach. For liver and nasal tumors, a number of histopathological observations have been published that identify toxicity, creating the precancerous lesions that ultimately drive tumor promotion, as determined by Health Canada’s draft (2018). Much of this was previously summarized in Section 3 of Appendix C.

4.4 The Scientific Justification for a Threshold Approach for Cancer Risk Assessment

4.4.1 Concluding Information on MOA for Justifying a Threshold Approach Versus a Linear No Threshold Approach

USEPA has relied on its 2005 Cancer Risk Assessment guidelines to select the linear modeling of 1,4-dioxane’s animal carcinogenicity data. The 2005 guidelines contain much rational guidance, but for its one significant scientific limitation of promoting use of the LNT approach as a default:

This wording demands that even if the agent is not mutagenic there must be “sufficient data to ascertain the mode of action and conclude that it not linear at low doses” (emphasis added). It can almost always be argued that the available toxicology information for a chemical does not fully support a hypothetical MOA. But for 1,4-dioxane, the general consensus is that 1,4-dioxane is not mutagenic since neither the parent compound nor its metabolite, 2-hydroxyethoxyacetic acid, is chemically reactive or chemically capable of alkylating DNA. Since USEPA’s IRIS and OCSPP are forced to rely on guidance (the 2005 U.S. USEPA Cancer Guidelines) that has set the bar high for determining MOA, linearity will more often than not be the default. As to what constitutes adequate information for USEPA to conclude “Sufficient Data” is up to expert judgment. However, USEPA’s 2005 guidance also allows that “alternatively, the assessment may use both linear and non-linear approaches” (3-22).

As summarized by both the USEPA’s IRIS (2013a) and OCSPP draft (2019) risk assessments on 1,4-dioxane, the available MOA data supports a number of threshold key events for modeling the carcinogenic risk in a threshold manner. Dourson et al. (2014, 2017) laid out the following MOA components for liver cancer that can be constructed into an adverse outcome pathway:

Metabolism Saturation of 1,4-Dioxane to HEAA Is a Critical Key Event—the Molecular Initiating Event

Saturation of 1,4-dioxane’s oxidative conversion to HEAA occurs between 30 and 100 mg/kg/day in rats and over 200 mg/kg/day in mice. It is at this saturation point that the key events behind tumor promotion emerge and where tumors begin to occur. For example, Figure 10 shows the relationship between 1,4-dioxane blood levels, measured after 12 weeks of exposure in the 13-week Kasai et al. (2008) study and the liver and nasal tumor response observed after 2 years in the Kasai et al. (2009) inhalation cancer bioassay. The parallel relationship between blood 1,4-dioxane levels and tumor development supports a causal relationship linkage between 1,4-dioxane’s saturation kinetics, with a steep rise in 1,4-dioxane blood levels beyond the 30–100 mg/kg/day saturation dose, and tumor development that likewise begins after this metabolic saturation dose range.

Figure 10. Dose-response of nasal and liver tumors in rodents exposed to 1,4-dioxane for 13 weeks to 2 years via inhalation (Kasai et al. 2008, 2009).
Source: ITRC 1,4-Dioxane Team, 2020. Adapted from Kasai et al. 2008, 2009.

Despite these data, USEPA’s OCSPP (2019) concluded in its draft risk assessment that there is no evidence of a correlation between 1,4-dioxane blood levels and tumor development (98).

Hepatoxicity and Liver Tumor Promotion

Kociba et al. (1974) and the unpublished 1971 report, presented in Section 3 and Figure 9, show how hepatotoxicity (e.g., hepatocellular necrosis) occurs before or at a greater frequency at dosages as do the combined liver tumors (refer to the discussion under Figure 9 for more description).

Much of this evidence supporting the key events is explained in Dourson et al. (2014, 2017), which can be consulted for these additional lines of key event data.

Kasai et al. (2009) reported a number of hepatic lesions that are precursor key events in a regenerative proliferation-based MOA. As shown in Figure 11, there was a greater occurrence of various types of foci indicative of a proliferative response elicited at 1,250 ppm (903 mg/kg/day) compared to a lower incidence of hepatocellular adenoma (HA) and hepatocellular carcinoma (HC) in the male F344 rats (female rats or mice were not studied). Spongiosis hepatis, a unique rat proliferative response due to activation of stellate cells, was also reported, suggesting the rat liver tumor response may have MOA components (e.g., stellate cell activation caused by 1,4-dioxane) that are not relevant to human cancer risk (Karbe and Kerlin 2002).

Figure 11. Focal lesions and liver tumors reported following 2 years of 1,4-dioxane inhalation in rats.
Source: ITRC 1,4-Dioxane Team, 2020. Adapted from Kasai et al. 2009.

By contrast, Kano et al. (2009) reported a higher incidence of combined HA and carcinoma in male and female rats relative to foci number following chronic exposure via drinking water. The Kano et al. drinking water study did not mention other noncancer histopathology (e.g., spongiosis hepatis or centrilobular necrosis). The apparent discrepancy in liver histopathology relevant to tumor promotion key events reported between the two studies is unresolved. Except for the Kano et al. drinking water study, in the absence of hepatotoxicity, liver tumor promotion can be argued as unlikely. In a challenge to this statement, USEPA points out that the liver tumors reported in the Japanese cancer bioassays (Kano et al.’s [2009] mouse liver tumors in particular) occur at a lower dosage than do noncancer hepatoxic responses, thereby questioning dose-concordance between key events and liver tumors. Again, this is an uncertainty that gives rise to less confidence in asserting a threshold MOA for cancer risk modeling of 1,4-dioxane exposures.

3. Nasal Mucosa Toxicity and Nasal Tumor Promotion
Nasal tumors, which occur consistently in rat bioassays, also have a reasonable key event database to examine MOA. In Section 6, Risk Evaluation, of the July 2019 draft, USEPA’s OCSPP listed a number of nasal mucosa histopathological (H&E) observations reported in the chronic inhalation study by Kasai et al. 2009 (Table 4-7A of USEPA’s 2019 draft report).

The Kasai et al. (2009; 2008) studies also showed atrophy of the nasal epithelium in rats and the JRBC (1998) study also observed atrophy of the nasal epithelium. (98)

The histopathological dose-response relationships for nasal tumor key events are shown graphically in Figure 12. OCSPP derived benchmark concentrations for each of these plausible key event occurrences, along with the nasal tumor response, shown in Figure 12b. These data demonstrate that nasal mucosa injury followed by hyperplasia (proliferative response), and its adaptive response (olfactory epithelium [OE] and respiratory epithelium [RE] metaplasia), occurs at dosages lower than those causing nasal mucosa tumors in a manner that is consistent with a key event relationship to the apical outcome.

Figure 12. Histopathological dose-response relationships for nasal tumor key events. (a) Rat noncancer key event histopathology and nasal tumors. (b) BMCL estimates following 2 years of inhalation exposure to 1,4-dioxane.
Source: ITRC 1,4-Dioxane Team, 2020. Adapted from Kasai et al. 2009.

As further evidence for a threshold, Torkelson et al. (1970) observed no tumors in rats exposed over a lifetime to 111 ppm. Despite this evidence, OCSPP (USEPA 2019) claimed no MOA information existed for the rat nasal tumors:

However, neither study reported evidence of cytotoxicity in the nasal cavity therefore, cytotoxicity, as a key event is not supported. (96)

Kano et al. (2009) Female Mouse Adenoma Outlier

USEPA argues that the absence of liver injury precursors reported by the Japanese investigators, as potential key events driving tumor promotion, supports its conclusion that inadequate information exists to derive a cancer MOA. In particular, USEPA points to the lack of an NOAEL in the increased female mouse liver tumor incidence reported at 66 mg/kg/day (Kano et al. 2009), absent evidence of hepatotoxicity to drive a regenerative repair MOA. Furthermore, this tumor response at 66 mg/kg/day occurred at dosages below the metabolic saturation point of 1,4-dioxane to HEAA (Dourson et al. 2017). In this case, the lowest dosage applied (66 mg/kg/day) caused a 70% incidence of liver tumors versus 5% in the controls, 82% in the next highest dose (278 mg/kg/day), and 92% in the highest dose (964 mg/kg/day). Figure 13 shows the Kano et al. (2009) female mice tumor results, along with the 1978 NCI mice liver tumor results. The NCI (1978) 380 mg/kg/day dose group developed a 21% incidence of liver tumors versus no liver tumors in the control group and a 94.6% response at the highest dosage of 860 mg/kg/day. The 66 mg/kg/day 70% response rate, when contrasted to the NCI (1978) response appears to be an outlier, keeping in mind the potential for differences in strain responsivity (Kano et al. [2009] used the BDF1 mouse and NCI [1978] used the B6C3F1 mouse).

Figure 13. Comparing the female mouse liver tumor response reported in the NCI (1978) study versus Kano et al. (2009) following lifetime drinking water ingestion.
Source: ITRC 1,4-Dioxane Team, 2020.

Several important considerations have been raised about this 70% response at the 66 mg/kg/day 1,4-dioxane dose from Kano et al. (2009) that reduce confidence in this tumor response:

The Kano et al. (2009) slides are not available for a review by a pathology working group to confirm and corroborate their findings of liver tumors occurring at low dosages (66 mg/kg) and, as noted above, which seem inconsistent with the higher dosages needed in the NCI (1978) cancer bioassay in B6C3F1 mice. The BDF1 female mouse has been reported to express an approximate 10% incidence of foci and HA in a small study of 50 female mice allowed to live out their lifespan (Yamate et al. 1990). When 499 female control BDF1 mice were examined, a range of specific foci (e.g., 0%–6% for basophilic foci) was reported (Katagiri et al. 1998). Up to 8% and 4% of female control BDF1 mice were found to have hepatocellular adenomas and carcinomas, respectively (Katagiri et al. 1998). Without reporting noncancer observations of hepatotoxicity relative to liver tumors, the Kano et al. (2009) study leaves many unanswered questions to evaluate its reported liver tumor response to 1,4-dioxane.
The Kano et al. (2009) study in Crj:BDF1 female mouse demonstrated a near maximum liver tumor response at the lowest dosage tested (66 mg/kg/day). In contrast, the 1978 NCI study in B6C3F1 female mice demonstrated a more gradual increase in treatment-related liver tumors requiring hundreds of mg/kg/day to occur.
The 13-week drinking water study (Kano et al. 2008) reported non-neoplastic liver pathology that somehow was not seen in the 2-year study. Based on this apparent discrepancy, Health Canada (2018) explained its decision to set aside the Kano et al. (2009) female mouse liver tumors as unreliable:

The absence of non-cancer histopathological changes and the concomitant increase in liver enzymes in the JRBC studies despite the presence of both endpoints in the subchronic studies from the same group (described in sections 9.2.2.1 and 9.2.2.5) lend credence to the uncertainty surrounding the development of tumors at this low dose. (Health Canada 2018 30)

This issue was also commented on by Dourson et al. (2017):

The lack of liver noncancer histopathologically in JBRC (1990s) is unexpected, especially since an increase in liver enzymes associated with cell damage is found in this same study. Also, the JBRC (1990b) 13-week study showed extensive liver noncancer histopathology at suitable adjusted-to-chronic doses. Unfortunately, this internal inconsistency is not resolvable because slides or pictures from a sufficient number of experimental animals are not available for the current reanalysis. (49)

It is uncertain whether the evolving pathological assessment of the Japanese mouse drinking water findings inadvertently counted altered hepatic foci as tumors, thereby inflating the number of liver tumors reported across the dose groups in Kano et al. (2009).

Diagnostic improvement from the hepatic hyperplasia in our preliminary report (Yamazaki et al., 1994) to the altered hepatocellular foci and hepatocellular adenomas in the present studies according to the current criteria (Mohr, 1997; Deschl et al., 2001) was found to increase the incidences of hepatocellular adenomas in the 1,4-dioxane-dosed groups, resulting in the definite dose-hepatocarcinogenic response relationships as compared with those in our preliminary report. (2783)

Notably, no altered hepatocellular foci (e.g., basophilic) were reported to evaluate proliferative precancerous lesions. No histological pictures of the before-and-after assessments were provided to understand how these investigators changed their previous histopathology evaluation. Furthermore, changing the histological diagnoses would require deviations or amendments to the protocol if this were a GLP study. The fact that the diagnoses were altered, without much explanation of evidence supporting this action, creates uncertainty around the updated diagnostic decisions generating a positive liver tumor response at 66 mg/kg/day.

The apparent absence of noncancer liver pathology in the Kano et al. (2009) mouse study is in contrast to the NCI (1978) mice results, which were available for reevaluation (as described in Dourson et al. (2014, 2017). A reexamination of the NCI (1978) studies provided additional histopathological information supporting key events consistent with a regenerative repair MOA. Figure 14 was created from reread information of the 1978 NCI study as reported in Dourson et al. (2014). Notably, the two dosages of 1,4-dioxane, despite being close to each other, produced an overall increase in the extent and—to a more limited degree—the severity of a number of hepatotoxicity observations associated with tumor promotion. The reread of the 1978 NCI histopathological findings as reported in Dourson et al. (2014) demonstrate a picture of hepatotoxicity combined with proliferative responses establishing a biological plausible MOA of regenerative repair—for example, the centrilobular necrosis occurs at a higher frequency than the tumor response. This centrilobular necrosis provides sufficient proliferation, as noted in the foci and Kupffer cell expansion, to enhance the normal background incidence of liver cancer in B6C3F1 mice.

Figure 14. Noncancer histopathology observations, serving as key events leading to liver tumors in the NCI (1978) mouse drinking water study.
Source: ITRC 1,4-Dioxane Team, 2020. Adapted from NCI 1978.

In summary, the lack of an NOAEL and a low LOAEL of 66 mg/kg/day for the female mouse liver tumors reported in Kano et al. (2009) raises questions and concerns over confidence in applying these findings toward a cancer toxicity value. Agencies like Health Canada (2018) and USEPA OCSPP (2019 have rejected these data for dose-response modeling on various grounds. Another important point is that chemically induced mouse liver tumors have been criticized as irrelevant for evaluating human risk (Lake 2018).

Proposed MOA for Liver and Nasal Tumors

Building on the previous MOA published in the Dourson et al. (2014, 2017) and adding to it the evidence of a direct mitogenic stimulus that occurs independent of late-stage cytotoxicity-driven regenerative repair, the following key events are proposed:

1.      Metabolic saturation of 1,4-dioxane to HEAA (Key Event #1)
               a. Subsequent key events depend on the accumulation of 1,4-dioxane in the blood
2.      Direct mitogenic stimulation (Key Event #2)
3.      Increased single-cell necrosis (apoptosis) (Key Event #3)
4.      Cytotoxicity-driven regenerative repair (Key Event #4
                a. Chronic inflammation
5.      Clonal expansion of preneoplastic foci (Key Event #5)

Using the “comparative weight of evidence analysis” approach set forth by Meek et al. (2014), these proposed key events are discussed in terms of supporting evidence (see Table C-8). This approach uses the “evolved Bradford Hill considerations” of (1) biological concordance, (2) essentiality of key events, (3) concordance of empirical observations among key events, (4) consistency, and (5) analogy. The third—concordance of empirical observations among key events—contains information about dose-response, temporality, and incidence.

Table C-8. Revised Bradford Hill causation considerations applied to the proposed key event and MOA hypotheses

Evolved Bradford Hill Consideration	Supporting data	Inconsistent data	Missing data
Biological concordance	For liver and nasal tumors, the H&E and mechanistic evidence establishes a proliferative response (direct mitogenic and indirect caused by regenerative repair) consistent with tumor promotion. Like many MOAs, there are data gaps (e.g., interim sacrifice data and pathway analyses to understand mechanistic aspects and dose-temporality concordance).	Kano et al. (2008), 13-week mouse liver toxicity data is not reported in the 2009 cancer bioassay.	The liver toxicity information available from the NCI (1978) mouse cancer bioassay contrasts with the missing information from the Kano et al. (2009) mouse study. Key event (e.g., H&E evidence) is not available for the unique male rat peritoneal mesotheliomas, the rat subcutis fibromas, the limited evidence of increase in kidney and breast cancer, and the rat-specific Zymbal gland tumors.
Essentiality of key events	The tumor promotion MOA is highly dependent on saturation kinetics where 1,4-dioxane levels increase as a function of dose. Doses that do not exceed metabolic saturation do not lead to tumors.	The 66 mg/kg/day LOAEL for female mouse liver occurs below the 200 mg/kg/day metabolic saturation proposed for mice.	CYP knockout specific studies to facilitate the buildup of 1,4-dioxane, hypothesized to induce key events and apical outcomes, is a potential data gap. No start-stop bioassays are available to see how termination of dosing eliminates or reduces the tumor incidence.
Concordance of empirical observations	A clear dose-response relationship (doses that do not cause key events or the apical outcomes, including liver, nasal, peritoneal mesotheliomas, subcutis fibroma, kidney, mammary, and Zymbal gland) has been reported in the majority of the cancer bioassays and sub-chronic studies, for example, clear NOAELs and LOAELs. For the most part, the apical outcome (tumors) occurs at higher doses and at a lower incidence than do the key events in the liver and the nasal epithelium. Both key events and apical outcomes depend on saturation kinetics with the buildup of 1,4-dioxane best correlated with the onset of key events and tumors.	The 66 mg/kg/day LOAEL for female mouse liver tumors reported in Kano et al. (2009).	90-day exposures produce minimal key events, suggesting later interim periods are needed to observe the emergence of key events that ultimately drive tumor development. Key event information is not available for male rat-specific peritoneal mesotheliomas, the rat subcutis fibromas, the breast, and Zymbal gland tumors. Interim sacrifice information establishing the temporality onset of key events preceding the tumor response is not available.
Consistency	Yes for liver tumors, in that both rats and mice develop liver cancer. There are differences between mice and rats, with mice largely developing liver cancers and no other tumor site, whereas rats, depending on the study, develop a larger range of tumor types.	Species-specific tumor responses (e.g., male rat peritoneal mesotheliomas and rat nasal tumors not observed in mice) have been reported. Across rat studies, not all report the same increased tumor response, (e.g., subcutis fibroma, peritoneal mesotheliomas).
Analogy	No—1,4-Dioxane does not have any apparent nuclear receptor or transcription factor to link it with a specific class of rodent carcinogens.		Genomics evidence identifying pathways linked to increase mitogenesis.

Figure 15 depicts the key events for both the liver tumor and nasal epithelium tumor response.

Figure 15. Proposed MOA for 1,4-dioxane-induced liver and nasal tumors.
Source: ITRC 1,4-Dioxane Team, 2020.

Risk Assessments Taking into Account Either Linear or Threshold Approach

Dourson et al. (2014) presented an MOA hypothesis from which they derived a 0.05 mg/kg/day reference dose (RfD) designed to be protective of cancer effects. This RfD value protects against liver hyperplasia induced by 1,4-dioxane, thereby preventing an essential key event of 1,4-dioxane-induced tumorigenesis. The 0.05 mg/kg/day RfD, in combination with standard default exposure factors used to derive drinking water criteria, supports a drinking water standard of 350 µg/L.

4.5 Conclusion

USEPA and Health Canada reached divergent conclusions on 1,4-dioxane’s cancer MOA and how it should be modeled. It can be stated with confidence that the USEPA-derived linear CSFs are very conservative relative to the key events that explain 1,4-dioxane’s MOA behind liver and nasal tumors. In addition, the evidence presented above should be reviewed to determine what approach best suits the circumstances, keeping in mind that USEPA’s 2005 Cancer Guidelines allow for the development of both a linear and a threshold estimation of the theoretical cancer risk. If the Kano et al. (2009) mouse liver adenoma results are viewed as unreliable, USEPA has calculated a number of other CSFs that can be used. The reader can refer to the 2013a IRIS report (e.g., Tables 5-10 and 5-11 on pages 138 and 140, respectively) as well as the draft 2019 OCSPP risk evaluation (e.g., Table 4-9 on page 116, Table 4-11 on page 123, and Table 4-12 on pages 126–127), where USEPA calculated numerous slope factors and IUR values, based on different tumors and modeling approaches. USEPA developed these alternative cancer slope values but did not choose them to represent its conclusions about risk. They reflect up to an approximate 1,000-fold range of uncertainty, or more, for predicting a population-level cancer risk based on a linear treatment of the data.

Dourson et al. (2014) modeled the hepatic necrosis key event data from Kociba et al. (1971) to derive a RfD protective of cancer development. A Log-Probit fit of the data yielded a BMDL₁₀ of 20 mg/kg/day for hepatic necrosis that was then used to derive a human equivalent dose (HED) of 5.2 mg/kg/day. In their 2017 publication, Dourson et al. estimated a saturation dose (molecular initiating event) for liver tumors of 9.6–42 mg/kg/day for rats and 57–66 mg/kg/day for mice from which a HED—which would be slightly lower, but in the same approximate range as the 5.2 mg/kg/day BMDL₁₀ (human)—was derived based on hepatic necrosis in rats. Uncertainty factors of 3x (toxicodynamics), 10x (intrahuman variability), and 3x (database uncertainty factor for lack of a two-generation reproduction study) yielded a composite uncertainty factor of 100. An RfD value of 0.05 mg/kg/day was generated, and to this a relative source contribution of 0.2 at 2 L of drinking water/day and a body weight of 70 kg was applied to generate a maximum contaminant level goal (MCLG) of 350 µg/L. This value is 10 times higher than USEPA’s 1 in 10,000 drinking water value using USEPA’s IRIS (2013a) linear CSF. The 750 µg dose, assuming 2 L per day at 350 µg/L ingested per day, is 520 times less than 5.2 mg/kg/day BMDL₁₀ HED.

Uncertainty still exists for the validity, interpretability, and inconsistency of the increased liver tumors in female CrJ:BFD1 mice at 66 mg/kg/day. Dourson et al. (2017) were not able to clarify the apparent absence of key events and the sensitivity of this response based on translated laboratory documents. They did not have access to the original slides to conduct a pathology working group reread to corroborate the original Japanese pathologist interpretations. Even so, this uncertainty is more than balanced and addressed by the reread of the 1978 NCI B6C3F1 mouse key event finding and the relatively consistent key event and dose-response relationship for tumors between the NCI (1978), Kociba et al. (1974), Kano et al. (2009), and Kasai et al. (2009) studies. The data from these studies support a combination of direct and indirect mitogenic stimuli (regenerative repair) in support of threshold modeling of 1,4-dioxane’s cancer risk that must be combined with the saturation kinetics acting as the important molecular initiating event.

References

ATSDR. 2005. Public Health Assessment Guidance Manual (Update).

Argus, M. F., Sohal, R. S., Bryant, G. M., Hoch-Ligeti, C., & Arcos, J. C. (1973). Dose-response and ultrastructural alterations in dioxane carcinogenesis.: Influence of methylcholanthrene on acute toxicity. European Journal of Cancer (1965), 9(4), 237-243.

Bevan, R.J. and Harrison, P.T.C. (2017). Threshold and non-threshold chemical carcinogens: A survey of the present regulatory landscape. Reg Tox Pharmacol 88: 291-302

Bogen, K.T. (2016). Linear-no-threshold default assumptions for noncancer and nongenotoxic cancer risks: A mathematical and biological critique. Risk Analysis 36: 589-604

Calabrese, E.J. (2019). The linear no-threshold (LNT) dose response model: A comprehensive assessment of its historical and scientific foundations. 2019 Chemico-Biol Interact 301: 6-25

Cohen, S.M., Boobis, A.R., Dellarco, V.L., Doe, J.E., Fenner-Crisp, P.A., Moretto, A., Pastoor, T.P., Schoeny, R.S., Seed, J.G. and Wolf, D.C. (2019). Chemical carcinogenicity revisited 3: Risk assessment of carcinogenic potential based on the current state of knowledge of carcinogenesis in humans. Reg Tox Pharmacol 103: 100-105

Doe, J.E., Boobis, A.R., Dellarco, V., Fenner-Crisp, P.A., Moretto, A., Pastoor, T.P., Schoeny, R.S., Seed, J.G. and Wolf, D.C. (2019). Chemical carcinogenicity revisited 2: Current knowledge of carcinogenesis shows that categorization as a carcinogen or non-carcinogen is not scientifically credible. Reg Tox Pharmacol https://doi.org/10.1016/j.yrtph.2019.01.024

Dourson, M., Reichard, J., Nance, P., Burleigh-Flayer, H., Parker, A., Vincent, M. and McConnell, E.E. (2014). Mode of action analysis for liver tumors from oral 1,4-dioxane exposures and evidence-based dose response assessment. Reg Tox Pharmacol 68: 387-401

Dourson, M.L., Higginbotham, J., Crum, J., Burleigh-Flayer, H., Nance, P., Forsberg, N.D. Lafranconi, M. and Reichard, J. (2017). Update: Mode of action (MOA) for liver tumors induced by oral exposure to 1,4-dioxane. Reg Tox Pharmacol 88: 45-55

Felter, S.P., Foreman, J.E., Boobis, A., Corton, J.C., Doi, A.M., Flowers, L., Goodman, J., Haber, L.T., Jacobs, A., Klaunig, J.E., Lynch, A.M., Moggs, J. and Pandiri, A. (2018). Human relevance of rodent liver tumors: Key insights form a toxicology forum workshop on nongenotoxic modes of action. Reg Toxicol Pharmacol 92: 1-7

Health Canada. (2010). Part V: Guidance on Human Health Detailed Quantitative Risk Assessment for Chemicals (DQRA_Chem). Federal Contaminated Site Risk Assessment in Canada. http://www.healthcanada.gc.ca

Health Canada. (2018). 1,4-Dioxane in drinking water, guideline technical document for public consultation. https://www.canada.ca/en/health-canada/programs/consultation-1-4-dioxane-drinking-water/document.html.

Hoch-Ligeti, C., Argus, M. F., & Arcos, J. C. (1970). Induction of carcinomas in the nasal cavity of rats by dioxane. British Journal of Cancer, 24(1), 164-167.

Huang, H.Y., Sung, M.T., Eng, H.L., Huang, C.C., Huang, W.T. and Chen, W.J. (2002). Superficial collagenous fibroma: immunohistochemical, ultrastructural, and flow cytometric study of three cases, including one pemphigus vulgaris patient with a dermal mass. Apmis, 110(4), pp.283-289.

JBRC. (1998). Two-year studies of 1,4-dioxane in F344 rats and BDF1 mice (drinking water). Japan Bioassay Research Center. Kanagawa, Japan.

Kano, H., Umeda, Y., Kasai, T., Sasaki, T., Matsumoto, M., Yamazaki, K., Nagano, K., Arito, H., Fukushima, S. (2009). Carcinogenicity studies of 1,4-dioxane administered in drinking-water to rats and mice for 2 years. Food Chem. Toxicol. 47 (11), 2776–2784.

Karbe, E. and Kerlin, R.L. (2002). Cystic degeneration/spongiosis hepatis in rats. Toxicol Pathol 30: 216-227

Kasai, T., Saito, M., Senoh, H., Umeda, Y., Aiso, S., Ohbayashi, H., Nishizawa, T., Nagano, K., Fukushima, S., (2008). Thirteen-week inhalation toxicity study of 1,4-dioxane in rats. Inhal. Toxicol. 20, 961–970.

Kasai, T., Kano, H., Umeda, Y., Sasaki, T., Ikawa, N., Nishizawa, T., Nagano, K., Arito, H., Nagashima, H.,

Fukushima, S. (2009). Two-year inhalation study of carcinogenicity and chronic toxicity of 1,4-dioxane in male rats. Inhal. Toxicol. 21 (11), 889–897.

Katagiri, T., Nagano, K., Aiso, S., Senoh, H., Sakura, Y., Takeuchi, T. and Okudaira, M. (1998). A pathological study on spontaneous hepatic neoplasms in BDF1 mice. J. Toxicol Pathol 11: 21-25

Kociba, R.J., Torkelson, T.R., Clashman, A., McCollister, S., Gehring, P.J. (1971). Continuous exposure of Sherman rats to various concentrations of 1,4-dioxane in drinking water for two years. The Dow Chemical Company, Biochemical Research Laboratory. 7901059. March 10.

Kociba, RJ; McCollister, SB; Park, C; Torkelson, TR; Gehring, PJ. (1974). 1,4-dioxane. I. Results of a 2-year ingestion study in rats. Toxicol Appl Pharmacol 30: 275-286. http://dx.doi.org/10.1016/0041-008X(74)90099-4.

Koissi, N; Shah, NH; Ginevan, B; Eck, WS; Roebuck, BD; Fishbein, JC. (2012). Lactone metabolite common to the carcinogens dioxane, diethylene glycol, and N-nitrosomorpholine: aqueous chemistry and failure to mediate liver carcinogenesis in the F344 rat. Chem Res Toxicol 25: 1022-1028.

Lake, B.G., 2018. Human relevance of rodent liver tumour formation by constitutive androstane receptor (CAR) activators. Toxicology research, 7(4), pp.697-717.

Lutz, R.W., Stahel, W.A. and Lutz, W.K. (2002). Statistical procedures to test for linearity and estimate threshold doses for tumor induction with nonlinear dose-response relationships in bioassays for carcinogenicity. Reg Tox Pharmacol 36: 331-337

Meek, M.E., Palermo, C.M., Bachman, A.N., North, C.M. and Lewis, R.J. (2014). Mode of action human relevance (species concordance) framework: Evolution of the Bradford Hill considerations and comparative analysis of weight of evidence. J. Appl. Toxicol., 34(6): 595–606.

NHANES. (2017). National Health and Nutrition Examination Survey 2013-2014 Data Documentation, Codebook and Frequencies. Volatile organic compounds (VOCs) and trihalomethanes/MTBE- blood-special Sample (VOCWBS_H)

NCI (National Cancer Institute). (1978). Bioassay of 1,4-dioxane for possible carcinogenicity. (78-1330 NCICGTR-80). Bethesda, MD. http://ntp.niehs.nih.gov/ntp/htdocs/LT_rpts/tr080.pdf.

Organisation for Economic Co-operation and Development (OECD). (2019). Adverse Outcome Pathways, Molecular Screening and Toxicogenomics. https://www.oecd.org/chemicalsafety/testing/adverse-outcome-pathways-molecular-screening-and-toxicogenomics.htm

Sweeney, L.M., Thrall, K.D., Poet, T.S., Corley, R.A., Weber, T.J., Locey, B.J., Clarkson, J., Sager, S. and Gargas, M.L. (2008). Physiologically based pharmacokinetic modeling of 1,4-dioxane in rats, mice, and humans. Tox Sci 101: 32-50

Torkelson, TR; Leong, BKJ; Kociba, RJ; Richter, WA; Gehring, PJ. (1974). 1,4-Dioxane. II. Results of a 2-year inhalation study in rats. Toxicol Appl Pharmacol 30: 287-298. http://dx.doi.org/10.1016/0041-008X(74)90100-8.

Tubiana, M., Feinendegen, L.E., Yang, C. and Kaminski, J.M. (2009). The linear No-threshold relationship is inconsistent with radiation biologic and experimental data. Radiology 251: 13-22

U.S. Army Public Health Command. (2010). Toxicology report No. 87-XE-08VWR-09 studies on metabolism of 1,4-dioxane March 2010. Preventive Medicine Surveys 40-5f1

U.S. Environmental Protection Agency (USEPA). 1989. Risk assessment guidance for Superfund Volume I, Human Health Evaluation Manual (Part A) [EPA Report]. EPA/540/1-89/002. Washington, DC.

USEPA. 1993. Use of IRIS Values in Superfund Risk Assessment (OSWER Directive 9285.7-16) [EPA Memorandum]. December 21. Washington, DC.

USEPA. 2003. Human Health Toxicity Values in Superfund Risk Assessments (OSWER Directive 9285.7-53) [EPA Memorandum]. December 5. Washington, DC.

USEPA. 2005. Guidelines for carcinogen risk assessment [EPA Report]. (EPA/630/P-03/001F). Washington, DC: Risk Assessment Forum. http://www.epa.gov/cancerguidelines/.

USEPA. 2010. Toxicological review of 1,4-Dioxane (CAS No. 123-91-1) in support of summary information on the Integrated Risk Information System (IRIS) [EPA Report]. (EPA-635/R-09-005-F). Washington, DC. http://www.epa.gov/iris/toxreviews/0326tr.pdf.

USEPA. 2012. Benchmark dose technical guidance. (EPA/100/R-12/001). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.

USEPA. 2013a. Toxicological review of 1,4-Dioxane (with inhalation update) (CAS No. 123-91-1) in support of summary information on the Integrated Risk Information System (IRIS) [EPA Report]. (EPA-635/R-11/003F). Washington, DC.

USEPA. 2013b. Tier 3 Toxicity Value White Paper. Regional 3 Toxicity Value Workgroup. OSWER Human Health Regional Risk Assessors Forum. May 16.

USEPA. 2014. Choosing number of stages of multistage model for cancer modeling: SOP for contractor and IRIS analysts.

USEPA. 2019. Draft risk evaluation of 1,4-dioxane (CASRN: 123-91-1) in support of the Office of Chemical safety and Pollution Prevention (OCSPP) [EPA Report]. (EPA-740-R1-8007). Washington, DC.

Wolf, D.C., Cohen, S.M., Boobis, A.R., Dellarco, V.L., Fenner-Crisp, P.A., Moretto, A., Pastoor, T.P., Schoeny, R.S., Seed, J.G. and Doe, J.E. (2019). Chemical carcinogenicity revisited 1: A unified theory of carcinogenicity based on contemporary knowledge. Reg Tox Pharmacol 103: 86-92

Woo, YT; Argus, MF; Arcos, JC. (1977). Metabolism in vivo of dioxane: Effect of inducers and inhibitors of hepatic mixed-function oxidases. Biochem Pharmacol 26: 1539-1542.

Yamate, J., Tajima, M., Kudow, W, and Sannai, S. (1990). Background pathology in BDF1 mice allowed to live out their life-span. Lab Animals 24: 332-340.

Young, J.D., Braum, L.H., Rampy, L.W., Chenoweth, M.B. and Blau, G.E. (1977). Pharmacokinetics of 1,4-dioxane in humans. J. Toxicol Environ Health 3: 507-520

Young, JD; Braun, WH; Gehring, PJ. (1978). Dose-dependent fate of 1,4-dioxane in rats. J Toxicol Environ Health A 4: 709-726

Print this page/section