Methyl-tert-butyl ether (MTBE): integration of rat and mouse carcinogenicity data with mode of action and human and rodent bioassay dosimetry and toxicokinetics indicates MTBE is not a plausible human carcinogen

ABSTRACT

Methyl-tert-butyl ether (MTBE) is a fuel oxygenate used in non-United States geographies. Multiple health reviews conclude that MTBE is not a human-relevant carcinogen, and this review provides updated mode of action (MOA), exposure, dosimetry and risk perspectives supporting those conclusions. MTBE is non-genotoxic and has large margins of exposure between blood concentrations at the overall rat 400 ppm inhalation NOAEL and blood concentrations in typical workplace or general population exposures. Non-cancer and threshold cancer hazard quotients range from a high of 0.046 for fuel-pump gasoline station attendants and are 100–1,000-fold lower for general population exposures. Cancer risks conservatively assuming genotoxicity for these same scenarios are all less than 1 × 10⁻⁶. The onset of MTBE nonlinear toxicokinetics (TK) in rats at inhalation exposures less than 3,000 ppm, a dose that is also not practically achievable in fuel-use scenarios, indicates that high-dose specific male rat kidney and testes (3,000 and 8,000 ppm) and female mouse liver tumors (8000 ppm) are not quantitatively relevant to humans. Mode of action analyses also indicate MTBE male rat kidney tumors, and lesser so female mouse liver tumors, are not qualitatively relevant to humans. Thus, an integrated analysis of the toxicology, exposure/dosimetry, TK, and MOA data indicates that MTBE presents minimal human cancer and non-cancer risks.

KEYWORDS:

• Methyl-tert-butyl ether
• MTBE
• mode of action
• carcinogenicity
• exposure
• toxicokinetics
• cancer risk

Introduction

Methyl tert-butyl ether (MTBE, CAS # 1634–04-4) was introduced into commerce in the United States in 1979 as a response to the United States Environmental Protection Agency (EPA) restriction on use of lead in gasolines. In 1990, amendments to the United States Clean Air Act mandated the use of oxygenates in motor gasolines, and MTBE was one of the series of fuel oxygenate compounds widely used to enhance combustion of motor vehicle fuels and reduce emissions of carbon monoxide and nitrogen oxides. Oxygenated fuels containing up to 15% MTBE were marketed primarily in the winter months in the United States beginning in the 1990ʹs to abate concerns of high carbon monoxide levels, and year-round at a 12% concentration to address ozone pollution in geographies experiencing excessive volatile organic compound emissions. MTBE use in reformulated gasolines began to decline in 1999 due to several state bans and was largely eliminated from use in the United States by 2006 due to passage of the United States Energy Policy Act of 2005 which eliminated the oxygenate requirement (NTP (National Toxicology Program) 1985; McGregor 2006; Cruzan et al. 2007; Roberts et al., 2015; Bogen and Heilman 2015). However, MTBE is still used in Europe and the Asia-Pacific region, and the 2019 global demand for MTBE was estimated at 25.2 million metric tons (IHS and Argus 2019).

Due to potential concerns for MTBE occupational, consumer, and general population exposures, MTBE has been extensively evaluated for potential human health hazards and risks. Many of the toxicology studies supporting its hazard and/or risk characterization have been evaluated as integrated reviews in the peer-reviewed published literature and in governmental or non-governmental institutional assessments. Importantly, human exposures have been well characterized and provide the dose context necessary to assess the health relevance of toxicity findings observed in experimental doses that often substantially exceed real-world human exposures. Finally, mode of action (MOA) studies, including metabolism and toxicokinetic (TK) studies, have further informed quantitative extrapolation of animal toxicity observations to potential adverse human health consequences.

The objectives of this review were to: 1) succinctly summarize key findings of past MTBE hazard and risk assessment evaluations (Table 1) and associated key rodent tumorigenicity findings (Table 2) underpinning those assessments. This information is intended to highlight the key cancer endpoints of concern and current risk assessment positions/perspectives on the health relevance of those endpoints; 2) introduce additional and/or more recent literature findings that have not been previously addressed in the summarized integrated reviews, and which further inform MTBE health assessments; 3) provide further MOA and TK perspectives assessing the plausibility of non-cancer and cancer outcomes associated with documented real-world exposures to MTBE; and 4) for typical human occupational, consumer and general population exposures, provide additional MTBE human health risk perspectives by comparisons of systemic blood margins of exposure associated with typical human exposure scenarios and low (400 ppm) and high (8000 ppm) rat C_max concentrations; and threshold cancer and non-cancer hazard quotients (HQ) and non-threshold cancer potency risk analyses.

Table 1. Summary conclusions of governmental, institutional, and other literature reviews addressing the toxicity and modes of action of MTBE and its primary oxidative metabolite TBA

Review/Assessment	Conclusions	Review/Assessment Detailed Bases for Conclusions
ATSDR (Agency for Toxic Substances and Disease Registry) (1996)	Minimal Risk Levels (MRLs): Inhalation Acute (≤ 14 days) = 2 ppm Intermediate (15–364 days) = 0.7 ppm Chronic (≥ 365 days) = 0.7 ppm Oral Acute (≤ 14 days) = 0.4 mg/kg/day Intermediate (15–365 days) = 0.3 mg/kg/day Chronic (≥ 365 days) = ND	MRL basis, Inhalation: Acute: NOAEL = 800 ppm (rat, neurological) Intermediate: NOAEL = 400 ppm (rat, neurological) Chronic: NOAEL = 400 ppm (female rat, kidney ↑ CPN) MRL basis, Oral: Acute: NOAEL = 40 mg/kg/day (rat, neurological) Intermediate: minimal LOAEL = 100 mg/kg/day (rat, mild liver) Chronic: ND (female rat, ↑ mortality at lowest tested dose, 250 mg/kg/day); Not genotoxic
EPA (United States Environmental Protection Agency) (1993)	RfC = 0.8 ppm (3 mg/m^−1)	Chronic: NOAEL = 400 ppm (rat, ↑ liver/kidney weight, ↑ female CPN)
IARC (International Agency for Research on Cancer) (1999)	IARC Cancer Group 3: “not classifiable as to its carcinogenicity to humans”	Human data: “inadequate evidence” (no human studies available) Animal data: “limited evidence” (↑ male rat kidney tumors due to α2u-globulin nephropathy); ↑ liver tumors in female mice at 8000 ppm attributed to ↑ mitogenicity secondary to altered estrogen regulation; ↑ rat testes tumors de-weighted due to low control incidence (inhalation) and nonstandard lifetime protocol (oral); Not genotoxic.
NTP (National Toxicology Program) (2000)	Cancer: Not listed after evaluation for possible “reasonably anticipated to be a human carcinogen”	Rat kidney and female mouse liver tumors may be due modes of action not relevant to humans.
EU (European Union) (2002)	Cancer: not classified Mutagenicity: not classified Reproduction: not classified Worker, EU conclusion (iii): risk reduction measures to control for long-term effects to skin in maintenance/automotive repair; Consumers and humans exposed via environment, EU conclusion (ii): no need for further information, testing or risk reduction beyond those currently applied	Cancer: Mode of action evaluation indicates human relevance of rat kidney and testes, and female mouse liver tumors “probably not very significant” Non-cancer: Not regarded as eye or skin irritant, sensitizer, or developmental/reproductive hazard in absence of maternal toxicity
ACGIH (2002)	Occupational: 8 hr TWA TLV = 50 ppm; Cancer classification A3	Cancer: A3 = “Confirmed animal carcinogen with unknown relevance to humans”; tumors and modes of action not regarded as human relevant; “predominantly negative” genotoxicity Non-cancer: No effects in volunteer study up to 50 ppm; NOAEL = 400 ppm, repeated exposures, rat
SCOEL (Scientific Committee on Occupational Exposure Limits) (2006)	Occupational: 8 hr TWA = 50 ppm, STEL (15 min) = 100 ppm; Cancer: A threshold for cancer potential is assumed	Cancer: Male rat kidney and testes (Leydig cell), female mouse liver induced by modes of action, not regarded as human relevant, e.g., α2u-globulin nephropathy; Not genotoxic
McGregor (2006)	Cancer: “Unconvincing” evidence for MTBE and its metabolites TBA and methanol Non-cancer: No specific developmental or reproductive toxicity; neurobehavioral or functional toxicity not likely under conditions of human exposure	Cancer: Male rat kidney tumors due to combination of α2u-globulin nephropathy and exacerbation of CPN modes of action not human relevant; Rat testes (Leydig cell) tumors confounded by inter-group age differences and lower than expected control incidences, and likely lack of human mode of action relevance; Female mouse liver tumors “extremely weak” evidence of human significance; Rat lung hemolymphoreticular tumors confounded by chronic inflammatory changes in lung; Mouse thyroid benign tumors after oral TBA were not observed with inhaled MTBE; MTBE and TBA are not genotoxic
Cruzan et al. (2007)	Cancer: Weak MTBE male rat kidney and testes, female mouse liver, and TBA female mouse thyroid tumor responses unlikely to present cancer hazard/risk at real-world human exposures	Cancer: Male rat kidney tumors attributed to α2u-globulin and exacerbation of CPN, modes of actions not regarded as human relevant; rat testes tumors equivocal due to high background incidence and evidence of a threshold mode of action; female mouse liver tumors due to disruption of estrogen function controlling mouse liver tumor development that is not relevant to human liver tumors; mouse TBA thyroid tumors secondary to altered thyroid gland function for which humans are far less susceptible; MTBE and TBA are not genotoxic.
McGregor (2010; TBA only)	Cancer: Male rat kidney tumors not human relevant; female mouse thyroid tumors not due to direct toxicity to thyroid Non-cancer: No specific development or reproductive toxicity	Cancer: Male rat kidney tumor mode of action due to α2u-globulin nephropathy exacerbated by CPN; Mouse thyroid tumor mode of action data limited but suggestive of increased T4/T3 clearance due to induction of Phase II metabolism; Not genotoxic
de Peyster and Mihaich (2014); endocrine only	Totality of weight-of-evidence analyses indicates no direct effects on the endocrine system based on the tested hypotheses of potential endocrine toxicity identified in mammalian and fish models and in in vitro screening assays	Weight-of-evidence protocol used semi-qualitative relevance weighting of data from each endpoint to test overall endocrine hypotheses of interactions with estrogen, androgen, and thyroid pathways and steroidogenesis; Isolated and inconsistent reports of high dose-specific effects in rodent and fish models not regarded as evidence of endocrine effects
Bogen and Heilman (2015)	Cancer Potency: Plausible upper-bound estimate for lifetime exposure, assuming genotoxic mode of action, slope factor of 0.003 per mg/kg/day (2 X 10^−5 per mg/kg/day). MTBE is among the weakest 10% of chemical carcinogens evaluated by US EPA; Cancer Potency, non-genotoxic mode of action: Assuming non-genotoxic (threshold) mode of action, RfD estimated as 0.27 mg/kg/day, and drinking water RfC of 9.6 ppm as protective of occurrence of any plausible carcinogenic precursor event in the general population. For individuals exposed for a small fraction of a lifetime, the drinking water RfC is approximately 10-fold greater (96 ppm); cancer potency is expected to be zero below the drinking water RfC.	Mode of action analysis of MTBE male rat kidney and testes and mouse liver, and TBA male rat kidney and mouse thyroid tumors attributed to non-genotoxic mode of action and endpoints not likely or of questionable mode of action relevance to humans; however, Cancer potency estimated from arithmetic mean of rodent cancer slope factors from rat oral testes tumors and mouse inhalation liver adenomas and liver adenomas + carcinomas (male rat renal tumors excluded due to lack of human relevance) and assumption of a 5% likelihood that a genotoxic mode of action is true (versus 0% chance if non-genotoxicity mode of action is true); Non-cancer RfD based on male and female rat ↑ kidney weights in 2-year drinking water study (Dodd et al. 2013) and a combined adjustment factor of 3 for interspecies uncertainty (due to PBPK interspecies dosimetry), 10 for inter-human variability and a LOAEL-to-NOAEL factor of 1 (↑ kidney weights not regarded as frankly adverse effect). Total combined Adjustment Factor = 30 applied to kidney weight increase, LOAEL = 8.2 mg/kg/day = 0.27 mg/kg/day.

Table 2. MTBE rat and mouse chronic/carcinogenicity bioassays

Study	Administered Dose (descriptor)	Delivered Dose Male (mg/kg/day)	Delivered Dose Female (mg/kg/day)	Tumor/Toxicity
Belpoggi, Soffritti, and Maltoni (1995), SD rat, lifetime gavage	250 1000 (mg/kg/day)	250 1000	250 1000	Testes (Leydig cell) adenomas ↑ at 1000 mg/kg/day beyond typical 2 yr study termination; ↑ Female leukemias/lymphomas at 1000 mg/kg/day
Bird et al. (1997), F344 rat, 104 wk inhalation	400 3000 8000 (ppm)	130^a 940 2,700	130 940 2,700	Male rat kidney and testes (Leydig cell adenoma) tumors ↑ at ≥ 3000 ppm; ↑ Substantial mortality ≥ 3000 ppm males due ↑ CPN; NOAEL toxicity and cancer = 400 ppm
Dodd et al. (2013), Wistar rat, 104 wk drinking water	0.5 3 7.5 15 (mg/ml)	25 140 330 NC	49 232 NC 1042	Marginal ↑ brain astrocytomas in male high dose not attributed to treatment; No effects on survival; ↑ CPN severity males/females in high dose
Benson et al., (2014), F344 rat, 104 wk inhalation, GMVC	112 560 1120 (MTBE ppm)	36^b 175 378	36 175 378	↑ Male rat (F344) renal tumors similar between GVC and GMVC – MTBE did not ↑ renal tumors beyond GVC alone; kidney weights not increased; other endpoints (testes, thyroid tumors) also similar; ↑ Testes tumors in GMVC due to low incidence in controls; incidence similar to GVC at all doses.
Bird et al. (1997), CD1 mouse 18 mo inhalation	400 3000 8000 (ppm)	372^c 2,794 7,452	414 3,110 8,295	↑ Female liver adenomas at 8000 ppm; ↑ Hepatocellular hypertrophy both sexes ≥ 3000 ppm; NOAEL toxicity = 400 ppm; NOAEL cancer = 3000 ppm
Cirvello et al. (1995), F344 rat, 104 wk drinking water, TBA	(M) (F) 1.25 2.5 2.5 5 5 10 (mg/ml)	85 195 420	175 330 650	↑ Male rat renal tumors, mid-dose only; ↑ High-dose male/female mortality ↑ Kidney transitional epithelial cell hyperplasia, mid/high dose males and high-dose females
Cirvello et al. (1995), B6C3F1 mouse, 104 wk drinking water, TBA	5 10 20 (mg/ml)	535 1035 2065	510 1015 2105	↑ Female mouse high-dose thyroid follicular cell adenomas

^aDelivered doses from Dourson and Felter (1997). McGregor (2006) estimated delivered doses of 40, 310, and 830 mg/kg/day for males and 60, 450, and 1,190 for females based on 6 L/hr inhaled air and MTBE 40% retention. However the 6 L/hr inhaled volume is low compared to estimates for F344 rats, e.g., Mauderly (1986) estimated approximately 14 L/hr for 350 g rat at approximately 12 mo of age. Using Mauderly (1986) minute volume of 0.67 ml/g bw for an approximate 12 mo old male F344 rat weighing 380 g, 40% MTBE retention, and 8,000 ppm MTBE = 28.56 mg/L, MTBE dose for 8,000 ppm exposure = 2,749 mg/kg/day. Given close agreement to Dourson and Felter (1997), doses from that publication were used.

^bDelivered doses interpolated from Dourson and Felter (1997) estimates of Bird et al. (1997) F344 inhalation doses.

^cDelivered doses estimated from Fairchild (1972): CD1 mouse minute volume (MV) = 1.81 ml/g bw for 72–76 day old male mice weighing 37.4 g; MV/hr = 4.062 L; inhaled volume for 6 hr inhalation exposure = 24.4 L; assume 40% MTBE retention; female mice body assumed as 10% less than males = 33.6 g. Dose estimates are higher than CD1 mouse estimates in McGregor (2006) of 210, 1,559, 4,150 (males) and 250, 1,850, 4950 (females) which assumed MV/hr = 1.8 L. Fairchild (1972) data were used in that MVs were estimated from plethysmography.

Generally, any commentary/analyses developed in this review relying on references dated after approximately 2010 (except for Bogen and Heilman 2015; de Peyster and Mihaich 2014); Table 1; (Dodd et al. 2013) in part represent newly presented commentary/analyses.

Governmental, Institutional, and Other Literature Reviews

The toxicity of MTBE and its potential human health implications associated with occupational, consumer, and general population exposures have been described in a series of evaluations published by governmental and non-governmental institutions, and the primary conclusions of these reviews are summarized in Table 1 EPA (United States Environmental Protection Agency) 1993; ATSDR (Agency for Toxic Substances and Disease Registry) 1996; IARC (International Agency for Research on Cancer) 1999; NTP (National Toxicology Program) 2000; EU (European Union) 2002; ACGIH 2002; SCOEL (Scientific Committee on Occupational Exposure Limits) 2006. In addition to the governmental/institutional reviews, several other reviews evaluating the cancer and non-cancer risks of MTBE and its primary metabolite tert-butanol (TBA) have been published in the peer-reviewed literature (Table 1; McGregor 2006; Cruzan et al. 2007; McGregor 2010; de Peyster and Mihaich 2014; Bogen and Heilman 2015). All of the reviews concluded that MTBE presents very low, if any, human non-cancer and cancer hazards or risks. Depending on animal cancer outcomes, mode of action assessments indicate that tumor responses are either qualitatively and/or quantitatively not relevant to humans. None of the reviews view MTBE as a potential human carcinogen under conditions of occupational, consumer, or general population exposures to MTBE (Table 1).

The focus of these past reviews was directed to animal bioassay findings available at the time of the assessments: an oral gavage lifetime study in rats (Belpoggi, Soffritti, and Maltoni 1995); 2-year and 18-month inhalation bioassays in rats and mice, respectively (Bird et al. 1997); and rat and mouse drinking water studies with tert-butanol (Cirvello et al. 1995; NTP (National Toxicology Program) 2000) Tables 2. The rat chronic drinking water study (Dodd et al. 2013) has been reviewed only by Bogen and Heilman (2015); the rat inhalation bioassay in which MTBE was co-administered as a gasoline vapor condensate mixture reflecting its real-world use as a gasoline fuel additive (Benson et al. 2011; Table 3) has not been examined in any of the cited reviews. The findings of these latter two bioassays, however, are consistent with the conclusion that MTBE is not a cancer hazard or risk in that neither of these studies resulted in MTBE-increased cancer responses (Tables 2, 3).

Table 3. Summary of MTBE toxicity conducted as a co-mixture with gasoline vapor condensates (GMVC) compared to gasoline vapor condensate only (GVC)

Study	Conclusions	Bases for conclusions
Benson et al. (2014): F344 Rat chronic carcinogenicity, inhalation, GVC and GMVC, 0, 2, 10 and 20 g/m^3 hydrocarbon (112, 560, 1,120 ppm or 36, 175, 378 mg/kg/day MTBE in males and females)^a	↑ Male rat (F344) renal tumors similar between GVC and GMVC – MTBE did not ↑ renal tumors beyond GVC alone; kidney weights not increased; other endpoints (testes, thyroid tumors) also similar	↑ Testes tumors in GMVC due to low incidence in controls; incidence similar to GVC at all doses.
Schreiner et al. (2014): S-D Rat 4 week inhalation MN and SCE evaluation, GVC and GMVC, 0, 2, 10, 20 g/m^3 hydrocarbon	GMVC did not increase cytogenicity (MN and SCE) compared to GVC.	MN: no statistically significant ↑ in MN at any dose for GVC and GMVC; SCE: males/females significantly ↑ SCEs at all doses of GVC and GMVC (paired or trend analysis)
Clark et al. (2014) S-D rat subchronic (13 wk) inhalation, GVC and GMVC, 0, 2, 10 and 20 g/m^3 hydrocarbon	GVC and GMVC NOAEL < 2 g/m^3, male rat nephropathy, recovery 4 wk post-exposure; GMVC NOAEL < 2 g/m^3, increased male/female kidney weight, female kidney weight reversed 4 week recovery; GVC NOAEL (other endpoints) = 10 g/m^3	Kidney weight changes not observed in F344 rats chronically exposed to GVC or GMVC.
Gray et al. (2014) S-D rat 2-generation reproduction, GVC and GMVC, 0. 2, 10 and 20 g/m^3 hydrocarbon	GVC and GMVC NOAEL for reproductive and offspring parameters = 20 g/m^3	No neuropathological effects, including brain glial fibrillary acidic protein content; ↑ male kidney weight GVC and GMVC P0 and F1 rats, ↑ female kidney weight GVC only, no histological correlate
Roberts et al. (2014a) S-D rat developmental toxicity, GVC and GMVC, 0, 2, 10 and 20 g/m^3 hydrocarbon	GVC and GMVC NOAEL for developmental toxicity = 20 g/m^3	GMVC maternal body gain and food consumption ↓ for gestation day 8–11 interval
Roberts, Gray, and Schreiner (2014b) CD-1 mouse developmental toxicity, GVC and GMVC, 0, 2, 10 and 20 mg/m^3 hydrocarbon; replicate GMVC study at same doses/time + 30 g/m^3 exposure gestation days 5–10	GVC maternal/developmental NOAEL = 10/20 g/m^3 (slight ↓ maternal body weight gain and ↓ fetal body weight; GMVC maternal/developmental NOAEL = 20/20 (study 1) and 10/20 g/m^3 (study 2)	GMVC (study 1) no maternal toxicity or statistically significant fetal effects, but rare malformation (ectopia cordis) in 1 fetus in 2 g/m^3 and 2 fetuses in 10 g/m^3; not considered treatment related (no high dose effect) and not confirmed in study 2; maternal toxicity GMVC study 2 NOAEL = 10 g/m^3 (labored breathing)
O’Callaghan et al. (2014) S-D rat 13 wk neurotoxicity, GVC and GMVC, 0, 2, 10 and 20 mg/m^3	GVC and GMVC NOAEL = 20 g/m^3 for function observational battery (FOB), motor activity, neuropathology (H&E stain) and regional glial fibrillary acidic protein analyses	FOB: assessments of home cage, handling, open field, reflex, grip strength, landing foot splay, hind limb extensor strength, and air righting ability; Motor activity measured by photobeam procedure; CNS and peripheral neuropathology
White et al. (2014) Female S-D rat 4 wk immunotoxicity, GVC and GMVC, 0, 2,10 and 20 mg/m^3	GVC and GMVC immunotoxicity NOAEL = 20 g/m^3	No significant changes in humoral components of immune system as measured by IgM antibody forming cell (AFC) response to T-dependent antigen, sheep erythrocytes (AFC/10^6 spleen cells or total spleen activity, AFC/spleen).

This series of reviews also evaluated the potential of MTBE to induce non-cancer hazards and risks, with the overall conclusions that MTBE is not a threat to reproduction, development, or other endpoints, such as neurotoxicity and immunotoxicity at exposure concentrations experienced occupationally or by the general population (Tables 1, 3). Of particular importance is a series of recent studies that comprehensively examined the non-cancer effects of MTBE as a co-exposure with a gasoline vapor condensate (Table 3). The gasoline-MTBE vapor condensate studies were designed to identify potential worst-case hazards and risks associated with exposure to MTBE resulting from its primary occupational and consumer use as a fuel oxygenate, and overall found no apparent evidence of MTBE-induced toxicity beyond that associated with non-MTBE containing gasoline-only exposures.

The primary tumor responses of concern consist of male rat renal tumors associated with MTBE and TBA treatment, MTBE-induced rat testes (Leydig cells) and female mouse liver tumors, and TBA-only-associated mouse thyroid tumors (Tables 1, 2). All of the published reviews provide analyses of the animal tumor findings that are integrated with MOA evaluations, with overall conclusions that the tumor findings lack qualitative or possibly quantitative human relevance.

The objective of this review is not to reiterate the analyses presented in these comprehensive analyses, but rather only to briefly summarize key findings and provide additional perspectives as appropriate based on new information and/or analyses. Readers are referred to each of the cited reviews for details of the analyses presented therein.

MTBE male rat kidney tumors

In brief, the reviews conclude that the MOA of MTBE and TBA male rat kidney tumors are attributable to a disruption of male rat-specific α2_u-globulin catabolism in the kidney that is further augmented by an exacerbation of rodent-specific chronic progressive nephropathy (CPN). Importantly, chemically induced α2_u-globulin nephropathy is not regarded as qualitatively relevant to humans since α2_u-globulin is only produced in male rats, but not in female rats, either gender of mice, or humans. MTBE and TBA, along with a spectrum of structurally related compounds, have been demonstrated to specifically bind to α2_u-globulin and inhibit its lysosomal catabolism in kidney proximal tubule cells following reabsorption of the glomerular filtered protein. A consequence of the altered catabolism is the formation of microscopically visible crystalline lysosomal hyaline droplets containing α2_u-globulin within the proximal tubule cells that subsequently rupture and initiate tubule cell cytotoxicity and associated onset of reparative tubule cell proliferation. This continuous cell-proliferative stimulus is hypothesized to be a key MOA event responsible for the ultimate progression of tumors (Borghoff, Short, and Swenberg 1990; EPA (United States Environmental Protection Agency) 1991; Hard 2018; Swenberg 1993).

The reviews also importantly discuss whether the overall MTBE and TBA MOA data adequately fulfill all of the criteria required by EPA (United States Environmental Protection Agency) (1991) to attribute the male rat specific renal toxicity of these two agents to an α2_u-globulin nephropathy only response. In particular, McGregor (2006) concluded that although these two agents rank among the weakest potency of chemicals operating with this postulated tumorigenic MOA, the overall data were nonetheless supportive of an α2_u-globulin mediated mode of action. A recent detailed pathology reexamination of the kidney toxicity and tumorigenicity associated with the NTP TBA rat 13-week and 2-year chronic drinking water studies (NTP, 1995), however confirmed the histological presence of two previously unfulfilled α2_u-globulin specific mode of action criteria, tubular granular casts and linear mineralization at the junction of the outer and inner stripes of the outer medulla and in the papilla (Hard et al. 2019). The low severity of these TBA-induced responses was consistent with a weakly active α2_u-globulin MOA. Importantly, the study authors also noted that both of the two additional confirmed renal lesions were considered as pathognomonic for α2_u-globulin nephropathy.

In addition to α2_u-globulin nephropathy, MTBE and TBA have been demonstrated to exacerbate another renal toxicity MOA that is possibly interactive and additive with α2_u-globulin, CPN. The onset of severe CPN in male rats has been identified as the primary cause of renal failure that accounted for excessive mortality and early termination of the 3000 and 8000 ppm treatments in the MTBE inhalation bioassay (Bird et al. 1997; McGregor 2006), and Hard et al. (2019) confirmed that TBA-induced renal toxicity and tumorigenicity was also clearly associated with enhanced CPN. However, because CPN is a small but clear risk factor for rat-specific renal cancer, but does not have any human clinical counterpart, renal cancer outcomes associated with this MOA are not considered as qualitatively human relevant (Hard et al. 2013; Hard, Johnson, and Cohen 2009).

MTBE rat testes toxicity and Leydig cell tumors

The reviews provide detailed analyses of why MTBE-induced rat Leydig cell testes tumors are regarded as lacking quantitative and possibly qualitative human cancer relevance. Interpretation of the testes tumor findings following oral gavage dosing in one of the studies (Belpoggi, Soffritti, and Maltoni 1995) was confounded by the atypical protocol in which animals were retained for treatment until death, resulting in total treatment times exceeding the 104-month period typical of standard rodent cancer bioassays. Since this tumor is associated with advanced age, findings in extended age studies are of unclear significance to human risk.

The high and variable background incidence of Leydig cell tumors in F344 rats used in the inhalation MTBE assay (Bird et al. 1997) is also a significant confounder in the evaluation of the human health relevance of this finding. Limited MOA investigations indicate that high-dose treatments of MTBE induce cytochrome P450s involved in testosterone metabolism, the consequences of which have been postulated as initiating compensatory testosterone-mediated hormonal CNS feedback controls that stimulate testicular interstitial hyperplasia that ultimately progresses to benign Leydig cell tumors in rats. This hormonally mediated MOA was reviewed by McGregor (2006), and it was suggested that this MOA would have a threshold below which no effect would be seen. In addition, Cook et al. (1999) noted that humans are likely substantially less sensitive to Leydig cell tumors than rats, and further concluded that this tumor response operates by a nonlinear (threshold-mediated) MOA. Thus, chemically induced Leydig cell tumors in humans would occur at threshold doses higher than those in rats. In this context, MTBE-induced Leydig cell tumors are observed only after high-dose inhalation exposures ≥3000 ppm MTBE (Table 2; 940 mg/kg/day; Bird et al. 1997). The lack of quantitative human relevance of F344 rat Leydig cell tumors and support for a threshold is further supported by the observation that testis toxicity and tumors were not observed in a chronic drinking water study in Wistar rats in which the 330 mg/kg/day top dose was approximately one-third that of the lowest inhalation tumorigenic dose of 940 mg/kg/day (Table 2; Bird et al. 1997; Dodd et al. 2013). A high-dose threshold for Leydig cell tumors is also consistent with the observation that MTBE induced only mild decreases in rat serum and testicular interstitial fluid testosterone at the top tested oral gavage dose of 1500 mg/kg/day (15 or 28 days of treatment; Williams et al. 2000). Overall, and similar to the hypothesized MOA for mouse liver tumors, rat Leydig cell tumors are likely mediated by parent MTBE in that MTBE’s primary oxidative metabolite, TBA, did not induce testicular tumors in rats (Table 2, Cirvello et al. 1995).

Li, Yin, and Han (2007; 2009) reported that various indicators of increased oxidative stress were observed in rat Sertoli and spermatogenic cells exposed in vitro to high MTBE concentrations, and suggested that “environmental” exposures to MTBE might elicit adverse reproductive toxicity. However, this highly speculative hypothesis is inconsistent with a lack of toxicity to reproduction in both one- and two-generation rat inhalation reproduction studies conducted at MTBE exposures up to 8000 ppm during all critical reproductive periods (reviewed in McGregor 2006). Importantly, no reproductive or offspring effects including neuropathology were observed in a rat 2-generation reproduction study at the top and worst-case human exposure of 20 g/m³ MTBE-containing gasoline vapor condensate (resulting in a 1120 ppm MTBE co-exposure; (Gray et al. 2014); Table 3). These combined findings indicate that MTBE is not an in vivo reproductive toxicity hazard in rats despite its reported ability to induce in vitro oxidative stress in isolated Sertoli and spermatogenic cells at high test concentrations.

MTBE mouse liver tumors

MTBE inhalation exposures significantly increased female CD1 mouse liver tumors only at the highest tested concentration of 8000 ppm (Table 2; Bird et al. 1997). The MOA of the high-dose specific MTBE liver tumors was proposed as being secondary to high dose-specific MTBE parent-compound induction of cytochrome P450 metabolism, a conclusion consistent with the observation that the primary MTBE metabolite TBA did not alter mouse liver tumors following high dose TBA drinking water treatment (Cirvello et al. 1995). The high-dose specific MTBE enhanced cytochrome P450 activity has been postulated to result in enhanced estrogen catabolism with subsequent loss of estrogen-mediated suppression of promotion of spontaneous liver tumors. Importantly, this MOA is not qualitatively relevant to humans in that human liver tumors are not controlled by estrogen (Cruzan et al. 2007).

The high-dose only increase in mouse liver tumors is also quantitatively not relevant to humans since the 8000 ppm tumorigenic exposure likely substantially exceeded saturation of MTBE metabolism and the Kinetically Derived Maximum dose (KMD) for this agent.

MTBE rat leukemia/lymphomas

The oral (gavage) bioassay of MTBE reported by Belpoggi, Soffritti, and Maltoni (1995) described an elevated incidence of lymphoimmunoblastic lymphoma that was primarily localized (85%) in lungs. However, the possible co-occurrence of chronic inflammatory changes in the lungs secondary to respiratory infections has resulted in challenges to these findings as treatment related. Schoeb and McConnell (2011) reported that an analysis of the non-neoplastic lesions from 3 bioassays conducted in the lab of Belpoggi, Soffritti, and Maltoni (1995), including MTBE, found clear evidence of inflammatory lesions in 76% of animals that were associated with infection with mycoplasma pulmonis disease. This high incidence of infection in laboratory test animals was confirmed in a later Pathology Working Group site visit analysis sponsored by the US National Toxicology Program (Gift et al. 2013), with the conclusion that the leukemias/lymphomas identified in studies from this lab were misinterpreted as treatment-related due to the extensive confounding by respiratory infections.

Interestingly, although epidemiology data for MTBE are generally not available, a comprehensive review and meta-analysis of the association of occupational exposures to gasolines containing MTBE and risk of non-Hodgkin’s lymphoma revealed no evidence of a gasoline-associated risk (pooled risk estimate from random-effects meta-analysis = 1.02, 95% confidence interval 0.94–1.12; Kane and Newton, 2014).

TBA female mouse thyroid and male rat kidney tumors

The primary metabolite of MTBE, TBA, induced female mouse thyroid tumors and male rat kidney tumors when evaluated in a drinking water bioassay Tables 1,2; (Cirvello et al. 1995).

MTBE inhalation exposure did not increase mouse thyroid tumors, and MTBE-induced mouse thyroid tumors are postulated as only secondary to high-dose oral administration of TBA (Tables 1, 2). In addition, oral administration of TBA demonstrates that it is a weakly potent thyroid tumorigen in female B6C3F1 mice in that a pair-wise significant rise in tumor incidence (p < .028; Fisher’s exact test) was observed only at the highest tested dose of 2110 mg/kg/day. Tumors were not markedly changed using pair-wise comparison in male mice at the top tested dose of 2070 mg/kg/day; however, an EPA-IRIS (United States Environmental Protection Agency) (2017) post-hoc statistical analysis reported a significant dose–response trend for both female and male mice tumor incidences (mortality adjusted Cochran–Armitage test: p < .028 and p < .041, respectively).

It is important to note, however, that neither of these statistical analyses meets the standards necessary to achieve statistical significance for common tumors observed in 2-year rodent bioassays as defined by Haseman (1983) as those with a background incidence greater than 1% and which was inclusive of the mouse thyroid follicular tumors reported in the NTP TBA drinking water carcinogenicity studies (Cirvello et al. 1995). For pair-wise comparisons, Haseman (1983) concluded that the appropriate statistical comparison for significance should be p < .01 for pair-wise comparisons to avoid generating an inappropriate number of false-positive results. The Haseman (1983) guidance has been adopted by the FDA (US Food and Drug Administration) (2001) for evaluation of studies of pharmaceuticals, with the added caveat that for trend analysis the p value should be less than 0.005. This guidance has also been accepted by the OECD (2012) for statistical analysis of long-term bioassays in rodents. Thus, the TBA thyroid tumor responses do not meet the standard for statistical significance of common tumors.

Two additional important considerations indicate that mouse thyroid tumors lack human relevance. First, the top doses in males and females exceeded by 2-fold the EPA and OECD testing guideline recommended Limit Dose of 1000 mg/kg/day:

The highest dose tested need not exceed 1000 mg/kg bw/day (EPA (United States Environmental Protection Agency) 1998).

A limit of 1000 mg/kg/d may apply except when human exposure indicates the need for a higher dose level to be used (OECD 2009). Table 5 indicates that realistic human occupational, consumer, and general population exposures are well below the dose limit.

Second, the top doses likely exceeded the threshold for the onset of nonlinear TBA TK due to saturated TBA metabolism. As reviewed in McGregor (2010), and likewise summarized in the EPA-IRIS (United States Environmental Protection Agency) (2017) assessment, the primary metabolism of TBA in both rodents and humans is mediated through cytochrome P450 oxidation of the TBA methyl group, resulting in the formation of 2-methyl-1,2-propanediol and 2-hydroxyisobutyrate. Cytochrome P450 metabolic oxidation is commonly associated with vulnerability to high-dose metabolic saturation. TBA is not a substrate for alcohol dehydrogenase, and lesser amounts of TBA metabolism proceed through direct glucuronidation of the parent molecule. Faulkner and Hussain (1989) demonstrated high-dose saturation of TBA metabolism in mice following iv administration of 370, 741, and 1483 mg/kg TBA, as evidenced by an 11.6-fold increase in the high-dose TBA AUC relative to the lowest dose, despite only a 4-fold differential to the actual administered dose (AUC at 370 mg/kg = 28 mmol*hr/L; AUC at 1482 mg/kg = 324 mmol*hr/L). Importantly, the top intraperitoneal (ip) dose of 1482 mg/kg bw used in by Faulkner and Hussain (1989) was significantly lower than the top dose of 2110 mg/kg bw/day in female mice exhibiting thyroid adenomas in the NTP bioassay (Cirvello et al. 1995). Due to the rapid and essentially complete oral absorption of TBA, systemic TK following ip administration are likely to be reasonably parallel to those of oral absorption.

EPA and OECD dose selection guidance for the design of toxicity studies emphasizes that toxicity, including tumorigenicity, observed at doses exceeding metabolic saturation lacks quantitative relevance for meaningful characterization of health hazards and risks (OECD 2009, 2012). Thus, if the mouse NTP TBA drinking water cancer bioassay had been designed in accordance with current EPA and OECD guidance for dose selection, thyroid tumors would likely not have emerged as a substantive cancer concern.

In addition to statistical and dose-selection considerations of the TBA thyroid tumor response, plausible but limited evidence suggests that the low-potency and high-dose specific female mouse thyroid adenomas observed only at a dose of 2110 mg/kg/day (Table 2) were secondary to TBA-increased gene expression of liver Phase II glucuronyl transferase enzymes resulting in corresponding reductions in circulating T4 and T3. Serum TSH was only slightly increased and there was no evidence of histological thyroid changes in the MOA studies. However, the observation that thyroid tumors were not altered in chronic MTBE bioassays (Table 2) also indicates that the weak and high-dose specific TBA response is not strong evidence that MTBE would be a potential thyroid carcinogen (McGregor 2010).

Similar to MTBE, TBA is also a low potency male rat kidney carcinogen. A positive finding of tumorigenicity was not identified in the original NTP bioassay using standard histopathological sectioning of the kidney (NTP, 1995, Cirvello et al. 1995). The combined incidence of adenoma and carcinoma of the kidney was 1/50, 3/50, 4/50, and 3/50 for the control and 1.25, 2.5, or 5 mg/ml drinking water doses, respectively, and significance of the response was only achieved in the mid-dose when subsequent step-sectioning of the kidney was conducted (8/50, 13/50, 19/50, 13/50; McGregor 2006). The procedure of step-sectioning of kidney tissue for obtaining increased renal tumor data for statistical analysis also selects for renal tubule tumors (usually late-occurring adenomas) and precursor foci of tubule hyperplasia associated with advanced CPN (Hard and Khan 2004). In first describing the step-sectioning procedure, Eustis et al. (1994) stated that a notable feature of all of the studies associated with higher renal tumor incidences in male rats after step-sectioning (including the 2-year TBA study) was the presence of chemical-related enhanced severity of CPN.

Advanced and end-stage CPN was reported to be responsible for a low incidence of renal tubule tumors in control rats and is therefore a risk factor for renal cancer development, particularly in male rats (Hard, Betz, and Seely 2012). A number of chemicals exacerbate CPN to advanced stages of severity, and in so doing, marginally elevate the incidence of renal tubule tumors (Hard et al. 2013). A high rate of tubule cell turnover occurs throughout the developmental course of CPN in association with sustained tubule cell injury, and is likely to act as the underlying basis for the spontaneous renal tubule tumor formation associated with advanced to end-stage CPN.

In addition to CPN-associated male rat renal tumors, the dose-dependent rise in kidney transitional epithelial cell hyperplasia reported in TBA treated male and female rats (significantly increased in mid-dose males and high-dose males and females; Cirvello et al. 1995) is not regarded as a human relevant endpoint. This lesion is an accepted component of advanced CPN exacerbated by TBA (Hard et al. 2019). In its advanced stages, CPN represents a wide spectrum of renal parenchyma alterations in both genders, including inflammatory cell clusters and a characteristic form of transitional cell hyperplasia of the renal papilla lining (Frazier et al. 2012; Hard et al. 2011). The association between the papillary lesion and advanced CPN was reported by several investigators (Frazier et al. 2012; Hard et al. 2011; Hard, Johnson, and Cohen 2009; Hard and Khan 2004). The latter publication was a product of the INHAND project, a joint initiative of the Societies of Toxicologic Pathology from Europe, Great Britain, Japan, and North America to develop an internationally accepted set of criteria and nomenclature for proliferative and non-proliferative lesions of the various organs of lab animals. This transitional cell papillary lesion was overlooked in earlier descriptions of CPN (Abrass 2000; Barthold 1979; Gray 1977; Hirokawa 1975; Peter, Burek, and van Zwieten 1986). These reports focused on changes in the glomeruli, tubulo-interstitium, and vasculature of the outer kidney zones in attempts to elucidate the pathogenesis and etiology of this spontaneous disease process, but did not critically evaluate the complexity of changes in advanced or end-stage CPN.

A Pathology Working Group reevaluation of the TBA 13-week and 2-year datasets (Hard et al. 2011) characterized the renal papillary lesion as an edema-like outpouching of the papilla lining followed by collapse of the protruding, single layer of lining upon itself (Hard et al. 2011). This alteration is present to some degree, ranging from trace to moderate in most rats with advanced CPN, and is particularly evident in end-stage CPN. The lesion is well illustrated in a control and TBA-treated male and female rat in Hard et al. (2011); Hard et al. (2019). McGregor (2010) concluded that α2_u-globulin nephropathy and CPN were plausible mutually contributing MOAs accounting for TBA-induced renal tumors. A subsequent PWG review of the TBA renal tumor data (Hard et al. 2011) agreed that both renal tumors in male rats and transitional cell hyperplasia of the renal papilla associated in male and female rats were integral outcomes of advanced CPN stating:

There was unanimous agreement among the members of this independent PWG that both α2_u-globulin nephropathy and CPN exacerbation were the only causative factors in the development of renal tubule tumors observed in male rats exposed to TBA [emphasis added] in drinking water. As neither of these modes of action have human counterparts, the PWG concluded that TBA-related renal changes in rats could not be extrapolated for human health risk assessment, and were unlikely to pose any risk for humans.

Despite the above conclusion, the US EPA-IRIS (United States Environmental Protection Agency) (2017) proposed that the kidney “transitional epithelial hyperplasia” noted in TBA-exposed male and female rats operates by a MOA different from the human irrelevant α2_u-globulin and chronic progressive nephropathy MOA, and used this endpoint to derive a non-cancer TBA reference dose. In response to this postulate, Hard et al. (2019) reevaluated 13-week, 15-month, and 2-year sacrifice histology slides of male and female rat kidneys from the NTP subchronic and 2-year NTP studies (Cirvello et al. 1995) and concluded that the transitional cell hyperplasia lesion reported by NTP was in fact part of advanced chronic progressive nephropathy. This conclusion was supported by the analysis of Souza et al. (2018), who determined that earlier pathological nomenclature classification of the CPN papillary lesion as “urothelial hyperplasia” or “transitional cell hyperplasia” was not accurate and in fact was more appropriately described as “vesicular alteration of the renal papilla lining.” Due to this pathology nomenclature inaccuracy, the US EPA-IRIS (United States Environmental Protection Agency) (2017) determination that the TBA kidney lesion is distinct from CPN is erroneous, and the TBA-induced papillary lesion was affirmed solely as an integral characteristic of advanced CPN in the rat. Importantly, CPN has no qualitative human correlate. Examination of other areas of the renal parenchyma unaffected by CPN identified only lesions consistent with an α2_u-globulin MOA. Thus, the EPA (2017) conclusion that TBA-induced non-cancer and cancer findings in male and female rats are human relevant, and by inference to MTBE (TBA is the primary metabolite of MTBE), is unfounded, and this TBA lesion does not support the existence of a third unexplained MOA accounting for TBA male and female rat kidney toxicity and/or tumorigenicity.

Available evidence indicates that TBA-induced α2_u-globulin nephropathy and exacerbation of advanced CPN are the MOAs accounting for male rat renal tumors and male and female “transitional cell hyperplasia,” and neither are considered relevant to humans. Thus, it is not appropriate to use either of these rat-specific endpoints for hazard or risk characterization of TBA, or inferred to MTBE as the metabolic parent of TBA.

In 2012, Melnick et al. (2012) relied on an analysis of 60 National Toxicology Program (NTP) studies to challenge the conclusion that the exacerbated CPN was a MOA contributing to chemically induced rat kidney toxicity and tumorigenicity. However, a critical shortcoming of that analysis was that it did not incorporate the key importance of the CPN histopathological grading system in examining the relationship of this constellation of lesions to kidney tumor outcomes. In order to reveal an association between exacerbated CPN and renal tubule tumor occurrence, a more sensitive grading system than the conventional grade 0–4 scale (based upon % kidney affected by CPN) used for nephropathy in the NTP chronic studies is required. The negative results of Melnick et al. (2012) substantiate this point. Hard, Betz, and Seely (2012) found that using an extended grading system of 0–8 grades based on lesion progression, and particularly where grade 8 was end-stage advanced CPN, is associated with an elevated incidence of renal tubule adenoma and its precursor, atypical hyperplasia (ATH). The kidneys of male and female control F344 rats (2391 rats in total) from 24 chronic studies in the NTP Archives were re-examined histopathologically and checked for ATH, renal tubule tumors, and severity grade of CPN using the 0–8 scale. In 1236 male control rats, there were 43 ATH and 26 adenomas (no carcinomas) and all but two of these lesions (which were both grade 6) occurred in rats with grade 7 or 8 CPN. The incidence of ATH in male rats with grades 7 and 8 CPN was 7.2%, and with adenoma, 4.3%. Although the incidence in rats with advanced grades of CPN and incidence of these lesions was lower in the 1155 control female rats (as expected due to the known gender difference for CPN occurrence in rats), the distribution of ATH and adenoma showed a similar pattern of distribution according to severity grades of CPN as in the male rats, namely a 2.8% incidence of ATH and 1.4% incidence of adenoma in rats with grades 7 and 8 CPN. There was one carcinoma at grade 5 CPN in the female rats, but this lesion was not associated with CPN-affected tissue and was considered to be a true spontaneous tumor.

In addition to the critical importance of using a robust histopathological CPN grading system to establish a CPN mode of action, Hard et al. (2013) noted a number of significant weaknesses in Melnick et al. (2012). These included: 1) chemicals producing renal tumors by disparate MOAs beyond CPN were included, confounding CPN-only conclusions; 2) diagnoses were varied dependent on whether more sensitive step-sectioning was combined with routine single section studies; 3) as noted above, a CPN grading system with particular focus on the severity of CPN end-stage kidney disease was not used; and 4) the statistical analysis was flawed in that it failed to include all animals on study regardless of CPN grade; if all animals were included, statistical significance was demonstrated between CPN grade and tumor formation.

MTBE adrenal toxicity

Increased adrenal weights have been observed in male and female F344 rats exposed for 13 weeks to 4000 or 8000 ppm MTBE by inhalation (reviewed in McGregor 2006), and blood corticosterone was increased 3- to 5-fold at the latter exposure. In addition, elevated adrenal weights were also noted only at the high-dose 8000 ppm MTBE exposure in a chronic 18-month inhalation exposure in mice (Bird et al. 1997). However, this high-dose specific effect was only observed at exposures exceeding a KMD and thus is not quantitatively relevant to human hazard or risk.

MTBE genotoxicity

McGregor (2006) reviewed the available data on genotoxicity of MTBE and more recently, Bogen and Heilman (2015) examined this issue in detail. The preponderance of evidence supports the conclusion that MTBE is not likely to be of mutagenic concern in vivo, especially at exposure levels that do not saturate the normal metabolic pathways. However, additional perspectives are offered below to contextualize the relevance, or lack thereof, of some of the isolated positive findings reported in the literature in the overall assessment of in vivo mutagenic risk of MTBE.

The in vitro genotoxic profile for MTBE can be described as being 1) negative in the bacterial mutation assays, 2) negative in in vitro cytogenetic tests, and 3) negative for the induction of gene mutations at the hprt locus of Chinese hamster V79 cells. However, positive findings were reported for an in vitro gene mutation assay using mouse lymphoma cells and in a comet assay.

Mackerer et al. (1996) reported that MTBE induced a mutagenic response in the presence of S9 at the TK locus in the in vitro mouse lymphoma forward mutation assay. This positive response is likely due to the formation of a reactive metabolite (e.g., formaldehyde) by microsomes outside of the cell and was not considered reflective of the realistic in vivo intracellular metabolism of MTBE to formaldehyde. Casanova and Heck (1997) demonstrated that formaldehyde generated from MTBE metabolism in primary mouse and rat hepatocytes at concentrations of up to 6.5 mM (650 µg/ml; higher than rat blood C_max = 493 µg/ml following exposure to 8000 ppm MTBE, Table 4) did not result in alterations in DNA-protein cross-links. However, only formaldehyde added exogenously to the hepatocyte cultures markedly increased DNA-protein cross-links relative to MTBE.

Table 4. Margins of Exposure (MOE) of systemic MTBE human blood concentrations biomonitored from various exposure scenarios relative to rat blood C_max blood concentrations measured immediately after 6 -hr inhalation exposure to 400 and 8,000 ppm MTBE

Study	Exposure Scenario	Biomonitored human blood concentrations µg/L (µM)	Margin of Exposure (Rat blood Cmax at 400 or 8,000 ppm/Human biomonitored blood concentration)^a Cmax 400 ppm/human Cmax 8000/human
White et al. (1995)	Gas station pump attendants	15 (0.17)	933	32,867
	Inside car repair garages	1.73 (0.020)	8,092	284,971
	Commuters arriving at work	0.11 (0.0012)	127,272	4,481,818
Moolenaar et al. (1994)	Workers heavily exposed to MTBE-containing gasoline vapors and vehicle exhausts	1.8 (0.020)	7,778	273,889
	Workers heavily exposed to non-MTBE gasoline vapors and vehicle exhausts	0.24 (0.0027)	58,333	2,054,167
CDC (Centers for Disease Control) (2009)	General population, age 20–59 (01–02, 03-04 sample year)	0.17 (0.0019) 0.19 (0.0021)	82,352 73,684	2,900,000 2,594,737

^aRat blood inhalation C_max concentrations from Miller et al. (1997), 400 ppm = 14,000 µg/L (159 µM); 8,000 ppm = 493,000 µg/L (5,593 µM)

In addition to the specific weaknesses of Mackerer et al. (1996), contextual exposures to other common intracellular metabolic xenobiotic sources of formaldehyde (Andersen et al. 2019) substantively diminish the toxicologic plausibility of this primary MTBE metabolite as presenting a realistic genotoxic risk under exposure scenarios producing the highest real-world human MTBE exposures – gasoline fuel oxygenate use. MTBE exposures, calculated as Lifetime Average Daily Doses (LADD) are highest for gasoline station pump attendants, approximately 10 µg/kg/day (Table 5). Conservatively assuming that all of the retained daily doses of MTBE are metabolized into formaldehyde, a 70 kg human would inhale a total daily MTBE dose of 700 µg (8 µmoles), and through metabolism, produce an equivalent total dose of 8 µmoles of formaldehyde. An 8 oz cup of coffee contains approximately 100 mg of caffeine (Mayo Clinic 2018), and because caffeine is essentially completely n-demethylated to formaldehyde by liver oxidation (Kot and Daniel 2008), a 100 mg dose of caffeine conservatively liberates approximately 500 µmoles of formaldehyde (caffeine MW = 194.19). Thus, consumption of a single cup of coffee generates an approximately 62-fold higher dose of metabolically generated formaldehyde than the human LADD associated with gasoline-pump attendant MTBE exposures (500 µmoles divided by 8 µmoles). If a typical pump attendant daily dose is used instead of the LADD, MTBE delivers a formaldehyde dose of approximately 18 µmoles, or 28-fold less than a formaldehyde dose from a single cup of coffee [e.g., Hu et al., 2015, reported mean daily MTBE exposures of 388 µg/m³, translating to a daily retained dose of approximately 1552 µg or 18 µmoles MTBE or formaldehyde (assumes 10 m³ inhaled air per 8 hr workshift and 40% retention of inhaled MTBE)]. Thus, given the contextual dosimetric consideration of formaldehyde metabolically formed from real-world and likely worst case human exposures to MTBE (gasoline station pump attendants) compared to formaldehyde metabolically formed from a common consumer beverage, it is highly implausible that the Mackerer et al. (1966) in vitro findings support a hypothesis that MTBE poses a genotoxic hazard or risk.

Table 5. MTBE HQ and cancer potency risks associated with typical occupational, consumer, and general population exposures

Study (location)	LADD^a µg/kg/day (n)	Hazard Quotient^b (threshold cancer and non-cancer)	Cancer Risk^c (non-threshold)
Hu et al. (2015) Gas station operators (pump attendants) and support staff (Southern China, 8 stations)	Operators(61): Mean = 10.7 Support(36): Mean = 1.9	0.026 0. 0046	2.1 x 10^−7 3.8 x 10^−8
Campo et al. (2016) Gas station pump attendants, cashiers and mixed work, and general population controls (Milan, Italy, 46 stations)	Gas station workers (all, 90): Median = 9.3 Attendants(52): Median = 14.1 Cashiers(12): Median = 0.7 Controls(89): Median = 0.08	0.034 0.052 0.003 0.0003	1.9 x 10^−7 2.8 x 10^−7 1.4 x 10^−8 1.6 x 10^−9
Wu et al. (2012) Residents, suspected hot spot of air pollution (Camden, NJ)	Residents (99): Geometric mean = 0.16 95^th % = 0.64	0.0006 0.0023	3.2 x 10^−9 1.3 x 10^−8
Brown (1997) Multiple datasets inclusive of occupational, consumer, and general population (United States)	Gas station workers (7 datasets): Geometric mean = 12.40 Gas station customer (7 datasets): Geometric mean = 0.19 Gas station or storage neighbor (3 datasets); Geometric mean = 0.15	0.046 0.0007 0.00056	2.4 x 10^−7 3.5 x 10^−9 3.0 x 10^−8
Brown (1997). General population, water affected by leaks and spills, total dose from drinking water, inhalation shower, dermal bath, whole house air (United States)	Geometric mean: 0.01 95^th percentile: 2.0	0.000037 0.007	2.0 x 10^−10 4.0 x 10^−7

^aBreathing zone air concentrations (µg/m³) for Hu et al. (operators, 388; support 71.6); Campo et al. (overall worker median, 408; attendants, 619; cashiers, 32.2; controls, 3.5); Wu et al. (resident geometric mean, 1.44; 95^th percentile, 5.57); Brown et al., air collection method not specified (gas station worker, 2000; gas station customer, 150; gas station or storage neighbor, 30); Brown et al. air collection method not specified (general population, water use, 0.01; 95^th percentile, 2).

^bHQ = LADD/RfD; oral drinking water RfD of 0.27 mg/kg/day from Bogen and Heilman (2015) is less than RfD = 0.42 mg/kg/day estimated from EPA RfC = 3 mg/m³ (Brown 1997).

^cCancer potency = LADD x cancer potency (2 x 10⁻⁵), from Bogen and Heilman (2015).

Tang, Wang, and Zhuang (1997) studied the genotoxicity of MTBE and its metabolites (TBA and α-hydroxyisobutyric acid) in a human cell line (HL-60) at concentrations of 1, 5, 10, or 30 mM using the comet assay. Remarkably, Tang, Wang, and Zhuang (1997) report identical levels of DNA damage for all three chemicals, which have different molecular structures with possible differences in reactive moieties and it would be unlikely that these chemicals would all produce the same comet response. The cell line used in this study is not known to be metabolically competent, and thus it is unlikely that a common proximate metabolite was responsible for the reported responses. Further, the results of this in vitro study are of questionable relevance as the two highest MTBE test concentrations were higher than the rat blood C_max MTBE concentration resulting from an 8000 ppm MTBE exposure (Table 6), and are many orders of magnitude higher than levels of MTBE biomonitored blood concentrations associated with worst-case human exposures (Table 4).

Table 6. Evidence of metabolic saturation after a 6 hr 8,000 ppm MTBE inhalation exposure. Data are from Miller et al. (1997)

MTBE Exposure	MTBE AUC0 → ∞ (µg hr ml^−1)	Blood Cmax MTBE (µg/L)	Exhaled MTBE (% dose recovered)	Urine (% dose recovered)	TBA AUC0 → ∞ (µg hr ml^−1)
400 ppm	84.3	14,000	21.2	64.7	404
8000 ppm (20X↑ dose)	2960 (35X↑)	493,000 (35X↑)	53.6	41.6	6010 (7.4X↑)

A non-GLP in vitro S. typhimurium assay (William-Hills et al. 1999) was conducted with (MTBE/TBA) and resulted in a non-reliable result. The (test substance) barely produced a positive response with the average value of the repeats exceeding a 2-fold increase in mutation incidence, but individual tests failed to meet this threshold in a strain that is known to have a high and variable background incidence of revertants. Thus, caution needs to be taken when evaluating these results, especially in light of the fact that in two independent and GLP-compliant tests in TA102, the William-Hills et al. (1999) findings could not be replicated (McGregor 2006). It is also important to note that the TA102 strain is specifically designed to be highly sensitive to oxidative stress (McGregor 2006, 2010), and thus the negative McGregor (2006) findings also do not support the findings of Sgambato et al. (2009) which showed enhanced comet assay damage and formation of 8-OHdG DNA oxidative damage observed in the presence of high cell toxicity.

In in vivo investigations, MTBE was negative for cytogenetic effects in rodent bone marrow, negative for induction of unscheduled DNA synthesis (UDS) in mouse liver, and negative for induction of gene mutations at the hprt locus of mouse splenic lymphocytes. In a recently completed in vivo gene mutation assay at the cII locus of transgenic Big Blue F344 rats, no mutagenicity was observed in nasal epithelium, liver, bone marrow, and kidneys following 28 consecutive days of exposure (6 hr/day) to MTBE concentrations of up to 3000 ppm (unpublished; summary available in ECHA MTBE dossier; https://echa.europa.eu/registration-dossier/-/registered-dossier/15543/7/7/1).

Overall, there are only a few in vivo studies reporting positive findings. Du et al. (2005) reported the formation of DNA adducts in mouse lungs, liver, and kidneys following the administration of ¹⁴C-labeled MTBE by oral gavage. The dose levels used were relatively small (up to approximately 6 mg/kg body weight) and not expected to exceed metabolic saturation. Du et al. (2005) used an ultrasensitive accelerator mass spectrometry (AMS) technique which has a sensitivity to identify 1 adduct in a trillion nucleotides. In this method, the DNA sample for analysis is converted to elemental carbon or CO₂ for the analysis of ¹⁴C resulting in the loss of adduct’s structural information. Because of this, it is essential that the analyte identified in AMS be chromatographically compared with synthetic standards to ensure that the binding detected is in fact due to DNA adduct formation (Himmelstein et al. 2009). However, Du et al. (2005) made no such comparison with synthetic standard. Because of this deficiency, it is not possible to ascertain whether the radioactivity identified in the DNA samples of Du et al. (2005) study is in fact due to DNA adduct formation and not due to metabolic incorporation of ¹⁴C into DNA through cellular carbon pools fueled with ¹⁴C originating from ¹⁴C-MTBE metabolism (e.g. formation of ¹⁴C-formate). This deficiency is amplified by the observation of Casanova and Heck (1997) who demonstrated substantial metabolic incorporation of radioactivity derived from ¹⁴C-MTBE into DNA and RNA nucleosides relative to potential cross-linking purified in mouse and rat hepatocytes incubated with a wide range of MTBE concentrations.

Schreiner et al. (2014) studied micronucleus (MN) induction in bone marrow and sister chromatid exchanges (SCE) in peripheral blood lymphocyte cultures of rats exposed to gasoline vapor condensate containing MTBE (GMVC). Schreiner et al. (2014) reported increases of up to 1.2-fold in SCE in groups exposed to up to 20 g/m³ of GMVC compared to gasoline vapor condensate (GVC) alone; no change in MN was observed. This is not an especially useful study to assess MTBE in vivo genotoxicity because the test material evaluated was a mixture and not MTBE alone. Further, the biological significance of SCE-induction is not fully understood, and consequently this endpoint is no longer considered as a bona fide genotoxic endpoint as reflected in the recent decision by the OECD to delete the test guidelines for this assay (http://www.oecd-ilibrary.org/docserver/download/9747901e.pdf?expires=1380202705&id=id&accname=guest&checksum=A908EDE6C9CCF59D6A7AA446B4D5E289).

IARC proposed 10 key characteristics of human carcinogens as a means intended to facilitate an improved understanding of key MOAs influencing human and animal cancer outcomes (Smith et al. 2016). Evidence of increased in vitro and/or in vivo oxidative stress is one of the key characteristics of human cancer identified by IARC in part because oxidative damage to DNA has been associated with evidence of mutations. IARC used this proposed MOA assessment tool in recent evaluations of several chemicals, e.g., the determination of “strong evidence” of oxidative stress as a contributing key rationale supporting the classification of glyphosate as a probable (Classification 2A) human carcinogen (IARC (International Agency for Research on Cancer) 2015).

It is important to note that the IARC framework evaluation system has been challenged with respect to its reliability in meaningfully supporting the human relevance of in vitro and in vivo mode of action datasets. Bus (2017a) noted that the data supporting “strong evidence” of oxidative stress for glyphosate suffered substantial experimental limitations, and particularly the use of single doses and/or time points and/or doses/concentrations that were many orders of magnitude greater than any known human exposures. Thus, in the case of glyphosate, essentially all of the studies identifying in vitro oxidative stress occurred at concentrations that would be physically unattainable even in high-dose animal studies, and far lower than in vivo doses and tissue concentrations resulting from real-world human glyphosate exposures.

In a subsequent study examining the same type of high throughput data from the EPA and NTP ToxCast/Tox21 datasets as was used in IARC analyses, Becker et al. (2017) concluded that the IARC MOA framework performed “no better than chance” at identifying human carcinogens as classified by the US EPA. With specific respect to the impact of oxidative stress as a predictor of human carcinogenicity classification, it was determined that a failure to consider cytotoxicity-induced concentrations markedly affected the interpretation of oxidative stress as to cancer classification. Overall, the Becker et al. (2017) analyses indicated that although high throughput data can inform the potential cancer hazards and risks of chemicals to humans, there remains an “urgent need for explicit, transparent, and scientifically robust procedures to evaluate the relevance and reliability of mechanistic datasets and the process of integrating mechanistic results with extant animal toxicity findings and human epidemiology.” The 10-key characteristic classification approach proposed by IARC has not yet established such evaluation procedures.

Such careful considerations are clearly necessary for MTBE. For example, as previously noted oxidative stress was reported in in vitro rat Sertoli and spermatogenic cell studies at concentrations generally ≥0.5 mM (Li et al. 2009; Li, Yin, and Han 2007). However, De Peyster et al. (2008) found no in vivo evidence of oxidative stress, including hallmark 8-hydroxydeoxyguanosine DNA base damage, in the livers of male mice receiving 80–8000 µg/ml MTBE for 28 days in drinking water. The absence of in vivo DNA oxidative damage was consistent with observations that MTBE was also likely not genotoxic in Ames Salmonella lines that are sensitive to oxidative mutagens (summarized in De Peyster et al. 2008). Thus, the carcinogenicity implications of MTBE MOA evaluations based upon consideration of the IARC 10 key characteristics of human carcinogens need to be cautiously interpreted, and particularly for in vitro studies in which the high volatility of MTBE presents experimental dosimetry challenges.

Taken together, and consistent with the analyses reported in the cited reviews, the available database supports the conclusion that MTBE does not pose a mutagenic risk at exposure levels encountered in the environment.

MTBE and endocrine disruption

The potential for MTBE to adversely affect endocrine systems under estrogen, androgen, thyroid, and steroidogenesis pathway control was robustly reviewed in a weight-of-evidence review reported by de Peyster and Mihaich (2014); Table 1. The review indicated that MTBE, when tested in a spectrum of in vitro and in vivo screening assays including rodent and fish models, or apical rodent toxicology tests, exerts no direct or secondary effects of toxicologic concern on the endocrine system. Further conclusions may be found in this publication and are not reproduced here.

The de Peyster and Mihaich (2014) conclusions are also consistent with observations that MTBE rat Leydig cell tumorigenicity and rat and mouse adrenal weight increases, both of which have potential endocrine-mediated MOAs, occur only at high doses that are not quantitatively relevant to humans.

MTBE Studies Conducted as a Co-Mixture of Gasoline Vapor Condensates

In 2014, a series of toxicology studies sponsored by the American Petroleum Institute were reported in several publications that described the comparative inhalation toxicity of gasoline-alone vapor condensate (GVC) and MTBE-containing GVC (GMVC) vapors in rats and mice (Henley et al. 2014); Table 3. The studies were conducted in response to a 1994 final rule issued under the US Clean Air Act by the US EPA that focused on the generation of health effect information on motor fuels and fuel additives. The focus of the US EPA test rule on assessing the potential toxicity of a gasoline-MTBE mixture is entirely consistent with recommendations of the NAS (National Academy of Sciences) (2009) that implementation of toxicity testing programs should incorporate a priori problem formulation in order to improve the utility of the health information for practical use in risk management decisions. Because the vast majority of occupational, consumer, and general population MTBE exposures result from its use in gasoline fuels, it is entirely reasonable to undertake a series of studies in Table 3 to better inform potential toxicity and health risks associated with exposures to volatile GVC and comparatively to MTBE-containing GMVC. Although the current review addresses only the comparative toxicities of GVC and GMVC, the studies also individually evaluated fuels containing the oxygenates ethyl tert-butyl ether (ETBE), t-amyl-methyl ether (TAME), diisopropyl ether (DIPE), tert-butyl alcohol (TBA), and ethanol (EtOH).

The GMVC test material was prepared from the volatile fraction of gasoline containing the maximum amount of 15% MTBE allowed by EPA. The final composition of the vaporized GMVC to which the animals were exposed contained approximately 20% MTBE and other low-molecular weight hydrocarbons including approximately 1.5% benzene (Henley et al. 2014). All of the toxicity studies involved 6 hr/day, 5 days/week exposures to 0. 2, 10, and 20 g/m³ of total GVC or GMVC hydrocarbon for strains of rats and mice appropriate to the endpoints evaluated (Table 3). GMVC test material containing approximately 20% MTBE results in MTBE exposures of approximately 112, 560, and 1120 ppm for the respective total GMVC hydrocarbon doses (Table 3; Benson et al. 2011).

The results of the comparative GVC and GMVC studies characterizing the carcinogenicity, subchronic toxicity, two-generation reproductive toxicity, neurotoxicity and immunotoxicity in rats, and developmental toxicity in rats and mice are summarized in Table 3. These studies were conducted with side-by-side simultaneous exposures to GVC or GMVC to facilitate effective comparisons between the two vapor condensate test materials.

In contrast to the rat cancer bioassay conducted with MTBE alone (Bird et al. 1997), GMVC did not affect the incidence of male rat kidney or testes tumors relative to GVC-alone treatment. Importantly, the lack of MTBE-specific tumors in these studies indicates that the 400 ppm NOAEL identified for these tumor responses in the MTBE-alone study (Bird et al. 1997) may be higher, i.e., at least 1120 ppm, reflecting the highest MTBE exposure in the GMVC study. Although the Bird et al. (1997) MTBE-alone exposure increased the incidence of kidney tumors, the GMVC kidney tumor response was entirely attributable to other non-MTBE components of the vapor condensate mixture, and was consistent with the previous observation of male rat-specific kidney tumors in studies of non-MTBE-containing wholly vaporized unleaded gasoline (MacFarland et al. 1984). As previously noted, even with a higher NOAEL suggested by the GMVC findings, MOA evidence indicates MTBE kidney tumors are not qualitatively relevant to human health risks.

Similar to the lack of GMVC effects on renal tumors at the highest tested dose, the vapor condensate mix also did not alter absolute male or female kidney weights in any of the treatments, while GVC increased these weights in mid- and high-dose female rats only. In addition, the observation that both GVC and GMVC elevated male and female kidney weights at <2 g/m³ in a rat subchronic study indicates that this shorter-term organ weight effect was not differentiated from GVC treatment alone. These findings indicate that the chronic dose threshold for increased kidney weights is >1120 ppm MTBE, compared to the 400 ppm NOAEL identified for this chronic endpoint for MTBE alone. The overall NOAELs for all of the other evaluated endpoints (Table 3) had the highest tested dose of 20 g/m³ GMVC, equivalent to 1120 ppm MTBE. Although it cannot be excluded that an absence of MTBE-specific effects in the GMVC studies may be due to unidentified interactions between MTBE and components in the GVC mixture, problem formulation confirms the mixture as constituting the majority of MTBE exposures encountered by humans and thus highly relevant to risk assessment

Overall, the combined GVC and GMVC datasets indicate that MTBE is not a rat carcinogen when tested under exposure scenarios most relevant to occupational, consumer, and general population exposures, i.e., with co-exposure to a range of volatile hydrocarbons associated with the manufacture, distribution, and use of gasoline.

Context of MTBE Human Exposure and Experimental Toxicology Dose Considerations for Informing MTBE Human Health Risks

The following sections provide additional perspectives on information evaluating the experimental toxicological data and its associated systemic dosimetry in the context of measured or modeled human exposures. Information on margins of exposure for non-cancer and cancer effects may be used as a gauge to determine whether the toxicological data cover a sufficiently broad dose/concentration range to accurately capture potential adverse health effects in humans.

Margins of Exposure (MOE) of rat chronic bioassay blood concentrations to human blood biomonitoring

Table 4 describes MOEs of C_max blood concentrations in rats exposed for 6 hr to 400 and 8000 ppm MTBE (Miller et al. 1997), representing the low- and high- exposure groups in the rat inhalation bioassay (Bird et al. 1997), relative to human blood concentrations associated with typical occupational, consumer, and general population MTBE exposures. Human end-of-shift blood concentrations for typical workplace MTBE exposures associated with gasoline station pump attendants, inside car repair garages (White et al. 1995) and working in area with intense exposures to MTBE-containing gasoline vapors and vehicle exhausts (Moolenaar et al. 1994) are differentiated by approximately 3–5 orders of magnitude from the maximum blood concentrations achieved in rodent bioassays. Moolenaar et al. (1997) also demonstrated that when the same occupational scenario was reexamined approximately 3 months after the MTBE fuel oxygenate program had ended in Fairbanks, Alaska, the median worker blood concentration was reduced by approximately 7.5-fold, expanding the MOE to rat blood concentrations to 4–6 orders of magnitude. MOEs ranged from 4 to 7 orders of magnitude for consumers as represented by commuters (White et al. 1995) or the general population as sampled in the US Centers for Disease Control NHANES nationwide biomonitoring program (CDC (Centers for Disease Control) 2009). The CDC biomonitoring was conducted between 2001 and 2004, when MTBE was used as fuel oxygenate in the United States. For the general population, even the 95^th percentiles of estimated MTBE blood concentrations were almost 100,000-fold less than the maximum (C_max) rat blood concentration observed immediately after a 6 hr exposure to the 400 ppm chronic toxicity NOAEL concentration.

The human blood concentrations depicted in Table 4 also offer an important frame of reference facilitating interpretation of potential health concerns extrapolated from in vitro toxicity tests. The emergence of and focus on high-throughput in vitro toxicity testing as a supplement/replacement to conventional resource-consuming in vivo testing has recently been catalyzed by recommendations of the US National Academy of Sciences (NAS (National Academy of Sciences) 2007; Krewski et al. 2010). However, it has likewise been cautioned that the in vitro test chemical concentrations eliciting biological “hits” of concern might be used to inform priorities for additional higher tier toxicity testing by consideration of the MOEs to known or expected human blood/tissue concentrations (Bus, 2017b; Thomas et al. 2013). The extremely large MOEs of MTBE for typical maximum and minimum real-world exposure scenarios of concern relative to systemic blood concentrations representing the maximum blood concentration resulting from a MTBE NOAEL dose in apical animal bioassays (400 ppm; Tables 1–3) indicate MTBE represents a low concern for adverse human health outcomes. Because the MTBE MOEs are based upon internal systemic concentrations accounting for TK behaviors in animals and humans, MOE’s greater than 10–100 would typically not be regarded as presenting health concerns.

Estimated MTBE human cancer and non-cancer risks associated with occupational, consumer, and general population exposures

Human cancer risks operating with assumed threshold-based modes of action and non-cancer health risks may be estimated from the Hazard Quotient (HQ) ratio of an appropriate reference concentration or reference dose (RfD, RfC) to Lifetime Average Daily Doses (LADD) of MTBE. A cancer potency approach based on estimates of cancer slope factors from animal bioassays is used if cancer outcomes are assumed to be mediated through non-threshold mechanisms (Table 5).

The overall conclusions of the comprehensive reviews and new toxicological data cited in the preceding sections indicate that MTBE is not genotoxic and the observed cancer responses operate via MOAs that lack qualitative and/or quantitative relevance to humans. As such, it is appropriate to apply a HQ analysis for evaluation of MTBE to evaluate potential cancer and non-cancer risks. HQ values of less than 1 are assumed to present a minimal, if any, risk of human cancer. Bogen and Heilman (2015) developed an oral RfD of 0.27 mg/kg/day that was based upon the assumption that increased kidney weights in male and female rats in a rat drinking water chronic bioassay (Dodd et al. 2013), and which the authors described as conservatively representing fairly subtle, even questionable, evidence of an adverse effect. This RfD was used in Table 5 as the basis for the HQ estimates even though the human exposures of concern were by inhalation. This value is more conservative than the RfD of 0.42 mg/kg/day estimated by EPA (United States Environmental Protection Agency) (1993), the inhalation RfC of 3 mg/m³ (Brown 1997) and the ATSDR intermediate-duration Minimal Risk Level (MRL) of 0.3 mg/kg/day (ATSDR (Agency for Toxic Substances and Disease Registry) 1996); Table 1.

To estimate cancer risk assuming limited function of a genotoxic (non-threshold) cancer mode of action, Bogen and Heilman (2015) proposed a plausible upper-bound cancer potency value of 2 × 10⁻⁵ per mg/kg/day MTBE on the assumption MTBE represented a 5% probability of operating with a genotoxic MOA. Such an adjustment is consistent with evidence indicating that MTBE is not genotoxic in a variety of in vitro and in vivo test systems (Tables 1, 3).

The MTBE LADDs for occupational MTBE exposure scenarios in Table 5 were calculated by the following formula: LADD = C_m × IR × EL × ED (BW × LT)⁻¹ × 40%, where C_m = average/median exposure concentration, IR = inhalation rate (m³/hr = 1.4), EL = exposure length (h/day = 8 hr full shift), ED = exposure duration (days at 25 years × 365 days), BW = body weight (70 kg), LT = Lifetime (days: 70 years × 365 days/yr). The LADD was adjusted by 40% to represent retention of inhaled MTBE in the body (McGregor 2006); the other values are standard defaults (EPA (United States Environmental Protection Agency) 2012). For non-occupational scenarios, the LADD was calculated by multiplying the mean, median, or geometric means of reported exposures by 20 m³ of inhaled air per day divided by a 70 kg body weight (EPA (United States Environmental Protection Agency) 2012). Estimates of cancer risk using a cancer potency approach were calculated by multiplying the cancer potency of 2 × 10⁻⁵ per mg/kg/day developed by Bogen and Heilman (2015) using the MTBE LADDs.

The Hu et al. (2016) and Campo et al. (2016) studies were selected for risk evaluations because these represent recent exposure scenarios associated with ongoing MTBE-containing gasoline use in Europe and China. In addition, the controls described in the Campo et al. (2016) study offer a reasonable estimate of the general population-level exposures in that these subjects lived and worked within a 25 km distance of Milan, Italy, where the occupational data were collected. Wu et al. (2012) and Brown (1997) studies represent occupational, consumer, or general population exposures associated with past uses of MTBE gasolines in the US, including potential exposures resulting from MTBE contaminated water. The strength of the Hu et al. (2016), Campo et al. (2016), and Wu et al. (2012) analyses is that exposures were estimated from personal breathing zone air sampling.

The HQ and cancer potency-based risk values calculated for typical MTBE exposures in Table 5 indicate that MTBE presents a low human health concern or risk for both cancer and non-cancer outcomes. The highest risks are associated with occupational exposure and resulted in projected lifetime cancer risks of less than 1 in 1,000,000 assuming the possibility of a weakly active non-threshold cancer MOA. The non-threshold cancer risks were calculated using the LADD. However, even if cancer risks are calculated using mean daily workshift doses, the risks remain low. For example, the cancer risk estimated for the highly exposed gasoline station pump attendants described in Hu et al. (2016) is 5 × 10⁻⁷ (mean exposure = 388 µg/m³; assume 1.4 m³/hr or 11.2 m³/8 hr work shift, 40% retention, 70 kg worker, risk = 5 × 10⁻⁷). All HQ values, which are indicators of risk for toxicities assumed to operate by threshold MOA, were significantly below 1, also supporting very low health risk concerns.

Quantitative relevance of high-dose rodent chronic bioassay doses to potential adverse human health effects

The Society of Toxicology Task Force to Improve the Scientific Basis of Risk Assessment (Conolly, Beck, and Goodman 1999) emphasized that rodent carcinogenicity responses restricted to high test doses that are “many multiples” higher than conceivable real-world human exposures have “dubious” human health hazards and risk relevance. This emphasis certainly applies to MTBE in that the maximum systemic blood concentrations and total daily/lifetime doses resulting from occupational inhalation exposures are multiple orders of magnitude less than inhalation exposures eliciting non-cancer and cancer responses in animal studies (Tables 4, 5). High-dose specific saturation of metabolic processes associated with corresponding onset of non-linear impacts on absorption, distribution, metabolism, and clearance processes (ADME) was identified as a common source of transition to MOAs exhibiting little, if any, quantitative relevance to real-world human exposures (Bus, 2017b; Barton et al. 2006; Carmichael et al. 2006; Doe et al. 2006; Foran 1997; Slikker et al. 2004b, 2004a).

Acknowledging this important toxicologic concept, EPA Cancer Risk Assessment Guidelines (2005) emphasize that changes in TK with increasing dose may result “ … in important differences between high and low dose levels in disposition of the agent or generation of its active forms. These studies play an important role in providing a rationale for dose selection in carcinogenicity studies.” Thus, data indicating the onset of dose-dependent toxicokinetic nonlinearities should be an important consideration in the potential selection of doses for rodent bioassays, and conversely, in the interpretation of tumor findings noted above the threshold of nonlinear pharmacokinetics.

The EPA guidance regarding consideration of TK in the selection of doses of rodent chronic toxicity/carcinogenicity bioassays is also fully consistent with recent OECD guidance (OECD 2012). OECD emphasized that:

Available toxicokinetic data (ADME) should always [emphasis added] be taken into account when selecting dose levels for a chronic toxicity or carcinogenicity study … Many toxicokinetic processes influencing absorption, distribution, elimination and metabolic activation or detoxication may become saturated at higher doses, resulting in systemic exposures to parent compound or metabolites that would not be expected in the real life human exposures for which risk assessments are needed.

Perhaps even more importantly, however, the OECD guidance clearly articulates that toxicity, including carcinogenicity, observed above the threshold for the onset of nonlinear TK has questionable quantitative health relevance to human hazard. Such high-dose specific toxicity is viewed as equivalent to toxicity noted at doses exceeding a Maximum Tolerated Dose (MTD) as determined by conventional body weight and/or other evidence of toxicity (OECD 2012):

Although top dose selection based on identification of inflection points in toxicokinetic nonlinearity may result in study designs that fail to identify traditional target organ or body weight effects, it must be appreciated that metabolic saturation in fact represents an equivalent indicator of biological stress. In this case, the stress is evidenced by appearance of non-linear toxicokinetics rather than appearance of histological damage, adverse changes in clinical chemistry, haematology parameters or decrease in body weight gain.

Furthermore, the dose identified as corresponding to the onset of nonlinear TK has been termed the Kinetically Derived Maximum Dose (KMD) (Saghir et al., 2012) and, as noted above, was recommended as a toxicologically justifiable alternative to the use of conventional MTD approaches for the selection of appropriate top doses in experimental toxicology studies. The KMD approach has been recognized in OECD guidance, e.g. the extended one generation reproduction testing (OECD 2011), as an appropriate means for selection of top doses in animal bioassays:

… care should be taken to avoid high dose levels which clearly exhibit saturation, provided of course, that human exposures are expected to be well below the point of saturation. In such cases, the highest dose level should be at, or just slightly above the inflection point for transition to nonlinear TK behaviour.

Similarly, the European Chemicals Agency (ECHA (European Chemicals Agency) 2017) has cautioned that:

Often the dose/AUC relationship deviates from linearity above a certain dose.” [A dose level] “ … corresponding to the inflexion point can be regarded as the kinetically derived maximally tolerated dose (MTD) [and] If information in this regard is available, it might be considered setting the highest dose level for repeated doses studies according to the kinetically derived MTD.

MTBE cancer responses in rats (male kidney and testes tumors) are only detected following inhalation exposures of 3000 and 8000 ppm, and in mice at 8000 ppm (liver tumors) (Table 2). Multiple lines of evidence strongly indicate these inhalation doses exceed a KMD. There is clear evidence for the onset of nonlinear TK between rodent inhalation bioassay doses of 400 and 8000 ppm in rats (Miller et al. 1997); Table 6. A 20-fold rise in external dose results in 35-fold elevation in MTBE AUC_{0 →∞} and C_max blood concentration. The disproportionate increases in systemic MTBE dose and maximum blood concentration are also associated with an elevation in the amount of the total systemic dose that is exhaled as MTBE, and with a corresponding decrease in the proportion of doses excreted into urine as MTBE metabolites. Under conditions of linear TK, the amount of total dose recovered in these excreta fractions remains constant over the range of test doses. In addition, consistent with evidence of metabolic saturation, the blood AUC of the primary oxidative metabolite of MTBE, TBA, increases only 7.4-fold between the 400 and 8000 ppm exposures, compared to the expected 20-fold rise if MTBE metabolism was not saturated (i.e., increased MTBE exhalation at saturating exposures leaves a lesser proportion of the total absorbed systemic available for metabolism to TBA).

Physiological-Based Pharmacokinetic (PBPK) modeling of MTBE rat inhalation TK behavior at the inhalation doses used in the rat chronic bioassay (400, 3000, 8000 ppm; Table 2) provides additional evidence that MTBE metabolism is saturated at less than the 3000 ppm intermediate exposure used in the chronic bioassay (Borghoff, Parkinson, and Leavens 2010). The PBPK model predicts that an increasing 80%, 88%, and 93% of the dose following respective 400, 3000, and 8000 ppm MTBE exposures is exhaled in male and female rats after a 91 day, 6 hr/day, 5 day/week treatment protocol. The elevated exhaled MTBE at the 3000 and 8000 ppm doses relative to 400 ppm is matched by a corresponding reduction in % total systemic MTBE dose metabolized to TBA (20%, 12%, and 7% at 400, 3000, and 8000 ppm, respectively). A TK study of MTBE concentrations in male and female rat livers (blood concentrations were not measured) following 6 hr inhalation exposures to 100, 400, 1500, and 3000 ppm MTBE also offers evidence of metabolic saturation at less than 3000 MTBE (Leavens and Borghoff 2009). MTBE liver concentrations immediately after exposure termination were approximately 1320 and 1526 µM (determined from graphically-presented data) for the respective 1500 and 3000 ppm exposures. Since these peak liver concentrations are above the K_m of 1248 µM estimated for MTBE rat metabolism from PBPK modeling (Rao and Ginsberg 1997), it is likely that MTBE is approaching or exceeding saturation at MTBE concentrations as low as 1500 ppm.

Although extensive TK investigations are not available for MTBE in mice, evidence of metabolic saturation is supported by the observation that an increase of 23.2, 37.6, and 69.0% of the respective i.p. MTBE doses of 50, 100, and 500 mg/kg in ddY mice was collected as exhaled parent compound (Yoshikawa et al. 1994). Since MTBE is rapidly absorbed by oral and inhalation routes (McGregor 2006), kinetics following i.p. administration are a reasonable surrogate of oral or inhalation dosing. In addition, since the doses used in this study are well below the total absorbed dose of approximately 8000 mg/kg/day estimated for male and female mice receiving 8000 ppm MTBE (Table 2), it is highly likely that MTBE metabolism is saturated in mice at this exposure concentration. The mouse TK data also illustrate the lack of human health relevance of the high-dose specific elevation in TBA-induced female mouse thyroid tumors to MTBE carcinogenicity (Cirvello et al. 1995); Table 2. Adjusting for molecular weight differences between MTBE and TBA (MTBE MW/TBA MW = 88.15/74.12 = 1.19), approximately 2500 mg/kg/day of inhaled MTBE would be required to produce this high oral dose of TBA, assuming 100% metabolism of absorbed MTBE to TBA. Since the mouse i.p. TK data indicate that a substantial portion of systemic MTBE is exhaled and thus not available for metabolism, and saturation of metabolism is distinctly present at 500 mg/kg, both the dose and MOA required for TBA to produce mouse thyroid tumors do not appear to be quantitatively relevant to MTBE carcinogenicity. As is further expanded on below, the 2500 mg/kg/day MTBE dose necessary to produce the TBA-inducing female mouse thyroid tumors is not realistically attainable under fuel use scenarios in which MTBE is present with other volatile gasoline components.

Consistent with regulatory guidance for dose selection in toxicity studies, and as has been emphasized by Bus (2017b), post-hoc toxicokinetic evidence that tumor responses in animal bioassays are above a KMD need to be interpreted with caution as to their quantitative relevance to human risk. This is particularly so when the inflection point at the onset of nonlinear TK is well separated from human exposures, as is robustly the case for MTBE. If dose selection decisions for implementation of rat and mouse MTBE inhalation bioassays had been considered with a priori knowledge of MTBE TK, the top dose selected would likely have been lower than 3000 ppm for both species. The practical consequence of such a decision is the likelihood that no rodent tumors would be identified as a meaningful risk concern.

The likely lack of MTBE carcinogenicity if tested using a KMD dose selection strategy is pragmatically and substantively reinforced by findings from the rat cancer bioassay conducted using MTBE-containing gasoline vapor condensate (Benson et al. 2011). No MTBE-related tumor findings were detected at the top daily exposure that is likely at or below metabolic saturation, 1120 ppm (378 mg/kg/day) MTBE in the GMVC studies (Benson et al. 2011); Table 2). The GMVC study is particularly important and informative in demonstrating the lack of quantitative human risk relevance of toxicity and tumorigenicity observed above metabolic saturation. As noted previously and in accordance with recommendations by the US National Academy of Sciences (NAS (National Academy of Sciences) 2009), the GMVC studies were designed with specific a priori attention to address the key problem formulation question of what are the potential human health hazards and risks of MTBE encountered under conditions of potential worst-case use conditions, i.e., co-exposures with the volatile fraction of gasoline vapors.

The comparative GVC and GMVC rat bioassays robustly address the central risk problem of concern. GMVC exposures in Benson et al. (2011) represent the maximum possible MTBE exposures humans could encounter from MTBE-containing fuels. The GMVC test material, which was a deliberate concentration of the volatile fraction of gasoline containing MTBE (GMVC = 20% MTBE, volume%) and was tested at a maximum GMVC concentration of 20 g/m³ total hydrocarbon that was 50% of the explosive limit of the GMVC vapor (Henley et al. 2014), did not result in any MTBE-related change in tumor responses. Thus, although the GMVC study findings essentially render moot the need to explore MOA of MTBE tumor responses noted only under metabolic saturation conditions which also are not physically attainable through gasoline exposure scenarios (Bus, 2017b), existing MOA investigations of these responses importantly also indicate that male rat and mouse liver tumors are not qualitatively relevant to humans. These considerations also indicate that the cancer potency estimates derived in Table 5 are extremely conservative. Beyond the likely non-genotoxicity of MTBE, the cancer potency estimates are derived from cancer slope factors modeled from tumor findings inclusive of doses that are also not quantitatively relevant to humans based upon TK and physicochemical considerations.

In addition to TK and practical exposure considerations described above, the top delivered doses delivered to male and female rats and mice were approximately 2700 and 8000 mg/kg/day, respectively, following the 8000 ppm exposure (Table 2). These bioassay doses are markedly above the toxicity testing limit dose of 1000 mg/kg/day recommended in EPA and OECD regulatory dose selection guidance for chronic toxicity/carcinogenicity studies (EPA (United States Environmental Protection Agency) 1998; OECD 2009). EPA guidance recommends that the highest dose in rodent chronic studies “need not exceed 1000 mg/kg/day” (EPA (United States Environmental Protection Agency) 1998), and OECD guidance (OECD 2009) further clarifies use of the 1000 mg/kg/day limit dose by stating that higher doses are only appropriate “ … when human exposure indicates the need for a higher dose level to be used.” The worst-case MTBE human exposures are far lower than even the 400 ppm NOAEL inhalation exposure in rodents (Tables 1,4, 5), and thus the guidance is clear that test doses above 1000 mg/kg/day are not warranted and that any toxicity above this dose should be markedly de-weighted, if not discarded, as to its quantitative risk relevance to adverse human health concerns. The limit dose consideration also applies to TBA-induced female mouse thyroid tumors observed only at an oral daily dose of 2105 mg/kg/day (Cirvello et al. 1995).

Overall, real-world human exposures, including the highest occupational exposures to gasoline station pump attendants, are much lower than non-tumorigenic test doses used in chronic rodent bioassays. Integration of TK data indicating tumor-producing high doses is above metabolic saturation in both rats and mice, and also above what is physically possible to attain in human fuel-associated exposure scenarios, justifies the conclusion that MTBE does not present adverse quantitative human health risks under conditions of past or current occupational or general population exposures.

Summary and Conclusions

MTBE has been well studied in human-relevant inhalation and drinking water animal toxicity studies, particularly because of its extensive global use as a fuel oxygenate additive and the associated potential for occupational, consumer, and general population exposures. Multiple comprehensive reviews of the toxicity findings and associated MOA investigations indicate that the primary cancer responses of concern, male rat renal and testes tumors, and female mouse liver tumors are either qualitatively and/or quantitatively not relevant to humans.

The conclusions of this review indicate that MTBE is a weakly active toxicant relative to potential adverse human health effects in that the maximum blood concentration in rats exposed to a chronic inhalation NOAEL of 400 ppm is multiple orders of magnitude higher than blood concentrations resulting from typical workplace or general population exposures. Consistent with these observations, both HQ (assuming threshold-based non-cancer and non-genotoxic cancer responses) and cancer potency (assuming the possibility of weakly active non-threshold genotoxic MOA) risk estimates do not indicate any meaningful concerns for MTBE-induced adverse human health outcomes. HQ ratios range from a high of 0.046 for fuel-pump gasoline station attendants to values approximately 100–1000-fold lower for general population exposures. Cancer potency risks for these same scenarios are all less than 1 in a million, even assuming operation of a low probability genotoxic MOA. Finally, evidence of the onset of MTBE nonlinear TK at exposures less than 3000 ppm, a concentration that is above that practically achievable for MTBE fuel use scenarios, supports the conclusion that high-dose specific male rat kidneys and testes and female mouse liver tumors are unlikely to be quantitatively relevant to humans using TK and exposure considerations alone. Importantly, other published reviews have also concluded that MOA investigations indicate a high probability that male rat kidney tumors, and lesser so for female mouse liver cancer, are not qualitatively relevant to humans. Overall, an integration of exposure, toxicologic (including genotoxicity), and TK evidence indicates MTBE presents minimal, if any, human cancer and non-cancer health risks.

Note added post-review

It is with great regret that we recognize the passing of co-author Gordon Hard during the final stages of this manuscript revision. Dr. Hard was a world-recognized expertise leader in the field of kidney pathology and disease, and his insightful thinking, which he always freely shared, will be sorely missed.

Additional information

Funding

This work was supported by Sustainable Fuels, www.sustainalblefuels.eu.