Mechanistic Considerations in 1,4-Dioxane Cancer Risk Assessment

Abstract

The risk assessment of many carcinogens involves extrapolation across large exposure differences between the dose levels used in animal studies and the much lower human exposures. This is true for 1,4-dioxane which has a consistent liver carcinogenic effect in both genders of rats and mice. These data have been applied to risk assessment assuming a linear low dose extrapolation in some cases but non-linear or threshold models have been used in others. This choice hinges on our understanding of the 1,4-dioxane cancer mechanism. While 1,4-dioxane is not genotoxic in standard test batteries and has non-linear toxicokinetics, the mechanism for its carcinogenic effect remains unknown and is an active area of research. This review summarizes the possible modes of action for this chemical, data gaps and application to risk assessment. We find that the cytotoxicity/hyperplasia and metabolic saturation hypotheses do not explain the carcinogenic response and do not take into account 1,4-dioxane’s induction of its own metabolism, leading to less likelihood for saturation during chronic exposure. There is evidence for other mechanisms, especially oxidative stress associated with the induction of CYP2E1 and in vivo genotoxicity that is not seen in vitro. The dose response for these effects needs further exploration compared to the time course and dose response for 1,4-dioxane-induced carcinogenesis. An additional consideration is the manner in which these 1,4-dioxane effects may augment naturally occurring and disease-related processes that contribute to the increasing rate of human liver cancer. These factors add to the rationale for using a non-threshold linear approach for extrapolating to low dose for this carcinogen, which is consistent with the default for carcinogens which do not have a clearly defined mode of action.

Introduction

The highly persistent drinking water contaminant, 1,4-dioxane, has come under increasing public health concern due to its environmental properties of long-term residence in groundwater, difficulty in removal from water resources, and widespread occurrence in drinking water supplies [1, 2]. This is compounded by evidence for a reproducible carcinogenic effect in both genders of rats and mice [3]. While unregulated by the United States Environmental Protection Agency (USEPA), several US states and other international bodies have established water quality targets (Table 1). These values span a large range, from 0.3 to 50 μg/L (micrograms per liter, or part per billion), with a key difference being the method for extrapolation of high dose studies in animals to the much lower human doses possible from ingesting contaminated drinking water. This extrapolation is associated with considerable uncertainty because of the magnitude of the high dose to low dose extrapolation and the lack of epidemiological studies. This uncertainty is typically addressed by mechanistic information which can inform how to extrapolate across species and orders of magnitude of exposure. Chemicals acting via a genotoxic/mutagenic mode of action receive a linear, non-threshold extrapolation to low dose. Non-mutagens which can be demonstrated to act via a threshold mechanism may receive a margin of exposure approach whereby uncertainties are addressed via uncertainty factors, and an acceptable level of exposure is calculated in a manner akin to non-cancer risk assessment [4]. The linear extrapolation approach typically targets a 1 in a million (di minimus) level of cancer risk when setting drinking water criteria for carcinogens, and this turns out to be well below (more stringent) than criteria based upon threshold approaches as suggested in recent reviews for 1,4-dioxane [5, 6].Thresholds may involve toxicokinetic mechanisms, in which metabolic clearance of the compound is saturated above a certain dose level leading to a buildup of chemical that wouldn’t occur at lower doses. Alternatively, thresholds may involve toxicological mechanisms whereby host defense and repair capabilities are overwhelmed at high doses but would be adequate at lower doses. Invoking a threshold mechanism requires knowing how the chemical causes cancer in the first place, why this carcinogenic effect can only occur above some metabolic or toxicologic threshold, and sufficient dose-response information to show where this threshold occurs, preferably both in humans and in the animals studied in cancer studies [4, 7].

Table 1.

Drinking Water Guidelines and Criteria for 1,4-Dioxane¹

Jurisdiction	Target Concentration (µg/L or ppb)	Type of Target	Date Set [updated year]
Alaska	4.6	Groundwater Cleanup Level	2021
California	3.0 1.0	Public Health Protective Conc Drinking water notification level	2011 2019
Connecticut	3	Drinking water Action Level	2013
Health Canada	50	Drinking Water Guideline	2021
Indiana	4.6	Groundwater Screening Level	2019
Maine	4	Maximal Exposure Guideline	2016
Massachusetts	0.3	Drinking Water Guideline	2014
Michigan	7.2	Drinking Water Criterion	2017
Minnesota	1.0	Drinking Water Guideline	2015
New Jersey	0.4 0.33	Groundwater quality standard Draft Drinking Water MCL2	2015 2021
New York	1.0	Drinking Water MCL	2020
North Carolina	3.0 0.35	Groundwater standard Drinking Water Criterion	2013 2021
Texas	9.1	Protective Concentration Level	2018
USEPA3	0.46	Screening Level for Tap Water	2017
World Health Organization	50	Drinking Water Guideline	2005
Japan	50	Drinking Water Guideline	2015
New Zealand	50	Drinking Water Guideline	2018

¹Source USEPA 2017 [39], ITRC 2021 [40].

²Maximum Contaminant Level

³USEPA has not set an enforceable MCL.

When the cancer mechanism is unknown it is impossible to determine whether thresholds exist in key steps of the carcinogenic process. Where this is the case, the more conservative (health protective) non-threshold linear low dose extrapolation is applied as a default [4]. USEPA in its Integrated Risk Information System (IRIS) toxicological review states why this default is needed for 1,4-dioxane [3] and this approach carries forward in the Agency’s Toxic Substances Control Act (TSCA) risk evaluation of this chemical [8]. While many jurisdictions and agencies have also applied this approach when setting drinking water targets for 1,4-dioxane, some have applied a threshold approach leading to the wide range of criteria in Table 1. This brief review provides an updated summary of the 1,4-dioxane liver cancer evidence relative to potential thresholds and mechanism and concludes that this mechanism is still unknown. Further, there are additional reasons for using a non-threshold assumption when extrapolating to low drinking water exposures.

1,4-Dioxane Metabolic Saturation

There is clear evidence that 1,4-dioxane metabolism is saturable with intravenous, gavage and inhalation studies in rats supporting the concept [9]. It is estimated that a plasma level of 100 mg/L represents a systemic level that can saturate liver metabolism and cause a disproportionate excess of parent compound [10, 11]. The issues with this threshold concept are: 1) there is no mechanistic explanation of how unmetabolized 1,4-dioxane causes cancer so that having this excess of parent compound does not have obvious implications for liver damage or carcinogenesis; 2) there are other excretory pathways for 1,4-dioxane aside from metabolic clearance (e.g. urinary and pulmonary clearance) such that excess non-metabolized chemical does not necessarily build up to great extent in liver [11]; 3) there is evidence that 1,4-dioxane induces its own metabolism such that the metabolic saturation point, if it exists at all under conditions of chronic exposure, would be higher than from single dose exposure [12]. For example, in a detailed repeat dose toxicokinetic study, 13 week inhalation in rats revealed a linearly increasing 1,4-dioxane plasma concentration with increasing dose, with plasma levels of parent compound ranging up to 1054 mg/L without showing any sign of metabolic saturation [13]. Thus, single exposure toxicokinetic evidence of metabolic saturation is likely not representative of the chronic exposure situation, wherein there is evidence of enzyme induction and lack of saturation even at high dose.

1,4-Dioxane Liver Toxicity and Carcinogenesis: Mechanistic Considerations

The toxicology profile of 1,4-dioxane has defied simple classification and still unclear is the potential for genotoxic and mutagenic effects to contribute to cancer causation. While in vitro screens were negative across the typical battery of tests [3, 9], in vivo studies document a genotoxic potential not predicted by the in vitro results. This includes positive gene mutation in the target organ, liver [14]. In that study, 1,4-dioxane administration to rats via drinking water induced gene mutation in liver at the gpt delta locus at 1000 and 5000 ppm exposures but not at lower doses, with a DNA repair gene induced at the highest dose level. Other indications of in vivo genotoxic potential include positive findings in mouse micronucleus [15, 16], and liver genetic damage [17, 18] while other in vivo genotoxicity studies were negative [19, 20]. These results suggest an in vivo genotoxic potential whose mechanism and dose response are still not well understood. As described below, repeat dosing with 1,4-dioxane induces cytochrome P-450 (CYP) 2E1, a CYP enzyme known for its generation of radical species and reactive oxygen [21]. Thus, one plausible mechanism is heightened CYP2E1 activity leading to oxidative stress and genetic damage.

While the contribution of genotoxic effects is uncertain, so is the cytotoxicity/proliferation mechanism. 1,4-Dioxane has produced a range of histopathologic effects in liver including vacuolation, cloudy swelling, spongiosis, inflammation and necrosis. These effects have been found sporadically in some high dose studies but not in others, and have not yielded a consistent pathologic pattern or hallmark of 1,4-dioxane hepatotoxicity. In comparison, the hepatocarcinogenic effect is highly consistent across 1,4-dioxane cancer bioassays [22–25]. Thus, a continuum between hepatotoxicity, proliferation and carcinogenesis has not been a consistent finding. Dourson et al. 2017 report liver necrosis and/or inflammation at dose ranges of 94–219 mg/kg/d in rats and in the 190–200 mg/kg/d range in mice [6]. In some cases, these non-cancer effects were not findings in the original bioassay reports but only in the re-reading of histopathology slides [6, 7]. While plausible that the hepatotoxicity reported in rats in this dose range may have contributed to the carcinogenic effects found in rat bioassays (274 mg/kg/d and above) [3, 24], it is noteworthy that a mutagenic effect in rat liver was found at a drinking water dose of 92 mg/kg/d and this occurred without hepatotoxicity [14]. Since this occurred at a dose below where tumors have been found in rat liver, a mutagenic component to 1,4-dioxane-induced liver carcinogenesis in that species cannot be ruled out. Further, liver tumors have been reported in mice at dose levels below the hepatotoxic range. The cancer effect level in the key study in mice (Kano et al. 2009 [24]) used by USEPA’s IRIS program to calculate 1,4-dioxane’s cancer slope factor was 66 mg/kg/d; this dose was not associated with any liver pathology other than a high incidence of tumors [3, 24]. A recent drinking water study designed to further understand the carcinogenic effect in mice found mild liver pathology that included vacuolation, isolated liver cell apoptosis and a mitogenic response after 13 weeks of exposure [26, 27]. These effects occurred beginning at a dose of 364 mg/kg/d, considerably higher than the cancer effect level in mice [24, 26, 27]. Thus, this study did not provide an explanation of how tumors may be initiated or progressed by the lower doses that were carcinogenic in Kano et al. 2009 [24].

A recent line of evidence points to an oxidative stress response induced in liver by repeated high dose exposures to 1,4-dioxane that involves CYP2E1 induction [28, 29] and increased markers of lipid peroxidation and DNA damage [28, 30, 31]. The fact that these markers were more pronounced in a knockout mouse model that is deficient in glutathione synthesis indicates the importance of cellular anti-oxidative defenses in preventing 1,4-dioxane damage [28]. A plausible key initiating event for this pro-oxidant effect is 1,4-dioxane induction of CYP2E1 given the propensity of this CYP enzyme to produce superoxide and other radical species during its mixed function oxidase catalysis of substrates [21]. 1,4-Dioxane was shown to induce CYP2E1 across multiple tissues in high dose rat and mouse studies [28, 29]. In these studies, 1,4-dioxane induced hepatic CYP2E1 activity and protein levels without inducing CYP2E1 mRNA, suggesting that this induction is via a post-transcriptional mechanism. This explains why studies which only evaluate transcriptomic changes did not detect CYP2E1 induction by 1,4-dioxane [14, 27].

The oxidative stress and genotoxic response have been observed after high doses, with the dose response at lower doses not well explored. Totsuka et al. (2020) evaluated DNA modifications induced by 1,4-dioxane in rat liver after 16 weeks of drinking water exposure spanning a wide range of doses, 20, 200 and 5000 ppm or approximately 1.9, 19 and 440 mg/kg/d [31]. Three different DNA modifications were associated with 1,4-dioxane exposure, all of which were also present in control rat liver, indicating that 1,4-dioxane was able to enhance endogenous processes that produce DNA damage (Figure 1). One of these modifications, 1,4-dioxane_3045, was tentatively identified through detailed mass spectral analysis as 8-hydroxyguanosine which is well known to result from oxidative stress. 1,4-Dioxane increased the level of this modification by 2-fold over control at both the mid and high doses or 19 and 440 mg/kg/d. The low dose (1.9 mg/kg/d) did not increase the level of the tentatively identified 8-hydroxyguanosine modification but did elevate a different DNA modification by 4-fold over control (not significant). This effect became statistically significant at the two higher exposures (Figure 1). While this DNA modification was clearly associated with 1,4-dioxane exposure, its exact identity needs further examination. These results suggest that 1,4-dioxane is capable of enhancing DNA damage processes at relatively low doses (1.9 mg/kg/d) with evidence for oxidative DNA damage at 19 mg/kg/d. While these results are in rats, the more sensitive liver cancer response to 1,4-dioxane appears to be in mice[3, 21, 26].

Figure 1. Dose Response for 1,4-Dioxane-Induced DNA Adducts in Rat Liver.

F334 male rats were dosed with 1,4-dioxane in drinking water for 16 weeks and liver DNA analyzed for adducts by HPLC-MS. Adducts are represented by their peak number and mass to charge ratio (m/z) (reprinted with permission from Reference [31]).

The potential for 1,4-dioxane to induce CYP2E1 at low doses needs further evaluation, especially in light of the ability of another CYP2E1 inducer to be effective at very low dose. Nicotine induction of rat liver CYP2E1 had an ED50 of 10 µg/kg/d with a saturation of this effect at 100 µg/kg/d [32]. Nicotine also induces CYP2E1 via a non-transcriptional mechanism. Thus, for certain agents, demonstration of CYP2E1 induction at high doses may be indicative of an ability to do so at much lower doses as well, and may be a mechanism that can contribute to cancer risk at low doses. Exploration of this dose response for CYP2E1 induction is a key research need for 1,4-dioxane and perhaps other agents.

Potential for Additivity of Cancer Risk with Background Disease Processes

1,4-Dioxane’s ability to induce oxidative stress via CYP2E1 induction is not unique to this chemical and is a common factor in the hepatocarcinogenesis associated with a variety of etiologies [21]. Liver cancer rates have been rising for several decades in the United States with liver damage associated with alcohol ingestion, viral hepatitis, and metabolic disorders contributing to this rise [33, 34]. These predisposing conditions have oxidative stress and CYP2E1 induction as key factors that worsen the liver pathology and promote its progression to cancer [35]. Thus, there would appear to be an intersection between 1,4-dioxane mediated hepatotoxicity and disease processes that are common in the US population. Toxicity that is additive to background aging and disease processes is less likely to be associated with a definable threshold given that the mechanism is already operative above threshold levels in many individuals. In such individuals, host defense mechanisms are already superseded at basal levels of oxidative stress, inflammation, cytotoxicity and DNA damage [36]. Thus, chemicals which contribute to ongoing damage processes have some likelihood of contributing to those processes even at low doses, with that dose response dependent upon a range of chemical-specific and host-specific factors. Knockout models in which host defenses are compromised (e.g., [28]), as well as animal models of human disease [37, 38] are potential ways to explore chemical dose response in vulnerable populations.

Summary

While progress has been made, the 1,4-dioxane carcinogenic mechanism remains unknown and threshold concepts relative to metabolic saturation and cytotoxicity/regenerative hyperplasia remain uncertain. This points towards the continued use of the linear low dose extrapolation as a health conservative default [4]. Both cytotoxic and genotoxic mechanisms are plausible with CYP2E1 induction and related oxidative stress being mechanisms in need of further investigation in terms of dose response and correspondence with tumor outcomes. Oxidative stress mechanisms also appear to contribute to the background rate of liver cancer, highlighting the need for research of 1,4-dioxane dose response in model systems that can evaluate the interaction of chemical-specific and disease-related mechanisms of liver disease and carcinogenesis [36].