Systematic evaluation of the evidence base on ethyl tert-butyl ether and tert-butyl alcohol for carcinogenic potential in humans; lack of concern based on animal cancer studies and mechanistic data

Highlights

• •A systematic evaluation of carcinogenic hazard of ETBE and TBA in humans was conducted.
• •Animal cancer studies show a low incidence of liver tumors in ETBE-exposed rats.
• •Animal cancer studies show a low incidence of male rat kidney and female mouse thyroid tumors with exposure to TBA.
• •ETBE and TBA lack genotoxicity with overall limited strength in mechanistic data across KCCs.
• •Collectively, evidence supports that ETBE and TBA are unlikely to be a carcinogenic hazard in humans.

Abstract

Ethyl tert-butyl ether (ETBE), a fuel additive, and tert-butyl alcohol (TBA), a solvent and metabolite of ETBE and methyl tert-butyl ether (MTBE), may be encountered via inhalation, oral, or dermal exposure. This assessment evaluated the human carcinogenic hazard of ETBE and TBA by systematically reviewing available human, animal, and mechanistic data. No epidemiological studies were identified, and two standard cancer bioassays were available for each compound. Tumor responses were limited to low incidences at high exposure levels: liver tumors in male F344 rats, kidney tumors in male F344 rats, and thyroid tumors in female B6C3F1 mice exposed to ETBE or TBA, respectively. Mechanistic evidence was organized within the framework of the key characteristics of carcinogens (KCC) and established rodent non-genotoxic modes of action (MoAs) for the overall evaluation. Aside from supportive evidence for KCC2 (is genotoxic) and KCC10 (alters cell proliferation, death, or nutrient supply), mechanistic data across KCCs were sparse and inconsistent. Both substances lacked genotoxic activity with available data supporting non-genotoxic MoAs that are not relevant to humans. Overall, the evidence indicates little concern for a carcinogenic hazard of ETBE or TBA in humans.

Introduction

Ethyl tert-butyl ether (ETBE) (CASRN 637-92-3) is used almost exclusively as a fuel oxygenate to allow fuel to burn more completely, leading to a reduction of exhaust emissions. Benefits to using ETBE include the maintenance of high-octane numbers in fuel, along with the reduction of content of aromatics, including benzene, in vapor emissions (Scholz et al., 1990). ETBE is used in Europe and Asia as a fuel component to increase octane ratings in gasoline. Tert-Butyl alcohol (TBA) (CASRN 75-65-0) can also be used as a fuel oxygenate and in the manufacture of methyl methacrylate plastics and flotation devices. It is also used in paint removers, nail enamels and polishes, and in the manufacture of food flavorings (EPA, 2021b). Exposure to ETBE or TBA may occur through inhalation, ingestion of contaminated water or food, and dermal contact, particularly in occupational settings. Human exposure to either ETBE or methyl tert-butyl ether (MTBE) results in detectable levels of TBA in blood since TBA is a primary metabolite of both ETBE and MTBE observed in both human and rat studies as reviewed by McGregor, 2006, McGregor, 2007.

The U.S. Environmental Protection Agency’s (EPA) Integrated Risk Information System (IRIS) program recently published risk assessments for ETBE and TBA (EPA., 2021a, EPA., 2021b). Both ETBE and TBA were categorized as having suggestive evidence of carcinogenic potential, indicating limited evidence of carcinogenic effects in animals but insufficient support for a stronger classification. For ETBE, this conclusion was based on one inhalation bioassay in rats in which liver tumors were observed in male rats (Saito et al., 2013), and a negative drinking water cancer bioassay in rats (Suzuki et al., 2012), a study considered by EPA to be inadequate for a definitive cancer risk assessment due to the lack of tumor response (EPA, 2021a). For TBA, the category of “suggestive evidence” was based on a drinking water study resulting in a low incidence of male rat kidney tumors and mouse thyroid tumors at high exposure concentrations (Cirvello et al., 1995, NTP, 1995). Both ETBE and TBA demonstrated a lack of genotoxic activity, and the observed tumor responses occurred at sites and under conditions not considered of concern for human carcinogenicity. The American Conference of Governmental Industrial Hygienists (2001) similarly concluded that TBA is not classifiable as a human carcinogen, identifying narcotic effects at high concentrations as the primary health concern.

In 2025, the International Agency for Research on Cancer (IARC) evaluated ETBE and TBA for the first time and published summary classifications (Turner et al., 2025). ETBE was classified as possibly carcinogenic to humans (Group 2B), based on sufficient evidence in experimental animals and strong mechanistic evidence, whereas TBA was not classifiable as to its carcinogenicity to humans (Group 3). IARC’s evaluations are hazard-based and consider evidence from three streams—human, animal, and mechanistic data—without accounting for exposure levels. Mechanistic evidence is organized according to the key characteristics of carcinogens (KCC), along with the evaluation of modes (or mechanisms) of action for animal tumor responses that may not be relevant to humans (IARC, 2019).

The KCC framework, originally proposed by Smith et al. (2016), describes ten biological properties commonly exhibited by established human carcinogens. While the KCC approach facilitates systematic organization of mechanistic evidence, concerns have been raised regarding its lack of specificity and potential to oversimplify causality (Becker et al., 2025). Nevertheless, others have emphasized its utility for identifying and mapping mechanistic data to key events within mode of action (MoA) or adverse outcome pathway (AOP) constructs to evaluate biological plausibility (Meek et al., 2014, Meek and Wikoff, 2023). This framework has been adopted by several regulatory programs, including the U.S. EPA IRIS and the National Toxicology Program’s Report on Carcinogens (EPA., 2022, NTP, 2018c).

The objective of this assessment is to systematically review the available evidence for ETBE and TBA to evaluate their carcinogenic potential in humans. This review applies systematic review methodology and critical appraisal tools to evaluate data reliability across three evidence streams: human cancer studies, animal cancer studies, and mechanistic evidence. Mechanistic endpoints were evaluated within the KCC framework and aligned with key events of known rodent tumor MoAs. The approach follows the principles outlined in the IARC Preamble (2019) for organizing and integrating evidence to develop conclusions regarding carcinogenic potential.

Methods

Overall approach

The approach used for assessing the potential carcinogenic hazard of ETBE and TBA is outlined in Fig. 1. This systematic evaluation, as described for MTBE (Borghoff et al., 2025a) was built on general principals and procedures described within the IARC Preamble (2019) for conducting cancer hazard evaluations. All available evidence identified was integrated: epidemiological; experimental animal; mechanistic data including absorption, distribution, metabolism and excretion (ADME); mechanistic data organized within KCC (Smith et al., 2016), along with consideration of the evidence base and evaluations of MoA of tumor outcomes in experimental animals; and, if available, information on any emerging carcinogenic mechanisms. Methods used for assessing and integrating mechanistic data were based on biological pathway-based concepts described by Wikoff et al. (2019) for tumor formation outcomes observed in animal and human studies and considered in the assessment as described by the World Health Organization (WHO) International Program on Chemical Safety (IPCS) (Meek et al., 2014, Meek and Wikoff, 2023).

Fig. 1.

Approach for systematic assessment of carcinogenic potential of ETBE and TBA. A simple representation of the volume and characterization of evidence for each stream identified for this assessment with additional details on the workflow in the methods and results.

Evidence identification – systematic literature search and screening

A protocol for this systematic evaluation for literature searching and screening is available in Supplementary Data A1. Using a broad syntax, consisting of the chemical name, major synonyms, and CAS number, searches of peer-reviewed literature for both ETBE and TBA (literature search syntax provided in Supplementary Data A2) were conducted in two citation databases, PubMed and Embase, on February 14, 2024. Citations were deduplicated using a reference manager (EndNote) and imported to Sciome Workbench for Interactive Computer-Facilitated Text-Mining (SWIFT) for prioritization by evidence stream (human, human-relevant animal models, in vitro or quantitative epidemiological) and health effects related to carcinogenesis. Citations meeting SWIFT prioritization criteria (i.e., evidence streams and any health outcome or any KCC) were prioritized for Title and Abstract (TiAb) review. An updated literature search was conducted in January 2025 with date filters from February 15, 2024 – Present (January 13, 2025) using the same search syntax and deduplication process (Supplementary Data A2). TiAbs of identified articles from the updated literature search were manually reviewed for relevancy.

TiAbs prioritized in SWIFT were evaluated for relevance in DistillerSR by reviewers based on categorization of the reported test article(s), evidence stream(s) (i.e., ADME, human cancer, animal cancer, mechanistic data (KCC), tumor MoA reviews). Articles meeting criteria at TiAb were then further reviewed for relevance based on the full-text article. TiAb and full-text review were conducted by a single reviewer per article. If foreign language papers were identified as relevant, these papers were translated to English for full-text review. Relevant full-text articles were then extracted to capture study design parameters and results of the outcome assessments. Quality control for extraction was performed on approximately 20% of the articles providing mechanistic data and animal cancer studies.

Additional handsearching was performed via Google Scholar and general internet searching to identify secondary literature (e.g., authoritative documents, literature reviews). These were reviewed for relevant data (e.g., citations of publicly available study reports) and used to cross-check the primary literature from the citation database searches (EPA., 2021a, EPA., 2021b).

Inclusion and exclusion criteria were established in the protocol (Supplementary Data A1) based on a Population, Exposure, Comparator, and Outcome (PECO) statement. These criteria, listed below, were used to identify relevant articles, i.e., human, animal and mechanistic data for assessing the carcinogenic potential of ETBE and TBA. Inclusion/Exclusion criteria for prioritization of TiAb were as follows:

• •
  Include:

  • -
    Human epidemiological studies investigating cancer outcomes or relevant biomarkers and ETBE or TBA exposure
  • -
    Animal cancer studies and ETBE or TBA exposure
  • -
    Studies that evaluated the relationship between exposure to ETBE or TBA and mechanistic endpoints relevant to one or more KCC
  • -
    Studies that evaluated the relationship between exposure to ETBE or TBA and emerging carcinogen mechanisms as well as MoA for tumor outcomes
  • -
    Studies and review articles related to adsorption, metabolism, distribution or excretion (ADME) of ETBE or TBA
• •
  Exclude:

  • -
    Studies not obtainable or studies that did not report original data (e.g., reviews of cancer or mechanistic data and/or commentaries)
  • -
    Studies in a foreign language that could not be translated
  • -
    Clinical human studies conducted for the dissolution of gall stones unless there were measures of ETBE or TBA in blood or endpoints measured that reflected mechanistic data relevant to cancer outcomes.

Toxicological databases such as EPA CompTox Chemicals Dashboard and IARC KC-Hits software were reviewed for high-throughput (HT) assay data in February 2024 (Reisfeld et al., 2022, Williams et al., 2017). This review included ToxCast quantitative structure–activity relationship (QSAR) models for predicting Estrogen Receptor (ER) and Androgen Receptor (AR) binding and activity using the CERAPP (Mansouri et al., 2016) and COMPARA (Mansouri et al., 2020) consensus models, respectively, as reported within the EPA CompTox Chemicals Dashboard. Data was identified for ETBE, but not for TBA (EPA., 2024a, EPA., 2024b).

Extraction and appraisal for quality and informativeness

Human, animal, and mechanistic data were extracted from studies confirmed as relevant following full-text review (Fig. 1). Extracted studies were appraised for quality and reliability utilizing appropriate tools as described below.

Cancer studies in humans

Human epidemiology studies identified in the literature that evaluate the association between ETBE or TBA exposure and cancer outcomes were evaluated for quality and risk of bias (RoB) using considerations described by NTP Office of Health Assessment and Translation (OHAT) (NTP, 2015) and the IARC (Berrington de González, 2024) bias-assessment guidance.

Cancer studies in experimental animals

Standard two-year cancer bioassays were evaluated for tumor outcomes in experimental animals. Non-standard studies such as initiation-promotion studies or studies on chronically exposed animal models (∼2 years) allowed to live until their natural death were evaluated and used, if necessary as supportive information only. Details of the study designs, which included species, strain, sex, duration, age, number of animals per exposure group, chemical administration and characterization (i.e., route, purity, vehicle), and results (i.e., tumor incidence, statistical significance, degree of mortality), were tabulated. Conclusions of tumor outcomes from authoritative bodies and other existing reviews were considered to provide contextual information in the final assessment of these studies (EPA., 2021a, EPA., 2021b, McGregor, 2007, McGregor, 2010).

Identified animal cancer studies were appraised for study quality and reliability, both objectively and consistently, according to general systematic review principles. Science in Risk Assessment and Policy (SciRAP) (Beronius et al., 2018), a critical appraisal tool (CAT), was used and the output interpreted for reliability categorization by subject matter experts (SME) (authors IAL and SJB). Use of a CAT facilitated a transparent and granular appraisal of study methods (appropriateness of the study design, study conduct) and reporting (completeness in the reporting of study design, conduct, and results) guided by 30 reporting quality criteria and 18 methodological quality criteria. As described in a similar evaluation of MTBE (Borghoff et al., 2025a), reviewers assigned fulfillment categories for each of the methodological criteria based on the following:

• •Fulfilled: Direct and/or specific information was considered that could be understood in the context of the study. Study design and methods aligned with Organization for Economic Co-operation and Development (OECD) recommendations and conduct or similar level of validity in the case where there is not a guideline equivalent.
• •Partially Fulfilled: Study reported partial information on criteria to assess methodological component or did not fully align with guideline methods.
• •Not Fulfilled: Critical aspects of the methodology were not conducted.
• •Not Reported: Information reported was insufficient to assess the methodological criteria.

Reliability categorization was qualitatively derived based on method conduct consistent with the criteria reported in the IARC Preamble (IARC, 2019), such as test substance characterization, dose monitoring, dose levels, duration and frequency, appropriate animal species/strains, number of animals per dose group, randomization, and adequate histopathological review. Consistent with approaches taken by authoritative bodies (Lahr et al., 2023), SME evaluation was integral to assessing study informativeness and took into consideration factors not explicitly specified in the CAT, including evaluating animal health status and considering the historical control data and human relevance. Studies were categorized based on the overall assessment using SciRAP and SME. Reliability categories included:

• •Reliable: Study was conducted using a method generally in alignment with standardized test guidelines, e.g., OECD TG 451 (carcinogenicity studies).
• •Partially Reliable: Study was conducted using a well-defined and widely accepted method without significant deviation but did not follow standardized test guidelines. Design or reporting of study resulted in some limitations for data interpretation, e.g., use of single dose levels precluding assessment of a dose–response relationship for any observed effects.
• •Not Reliable: Study was not conducted using a well-defined and widely accepted method and did not follow standardized test guidelines. Design or reporting of study resulted in significant limitations for data interpretation, e.g., unknown pathogen status for animals used in the study.

Mechanistic evidence

The mechanistic evidence identified and reviewed within this cancer assessment includes a summary review of ADME data collected in humans and animal models, and extraction of mechanistic data organized within the ten KCC (Smith et al., 2016). Additional mechanistic data identified as potential emerging issues in carcinogenesis were also extracted to assess their possible contribution to ETBE’s or TBA’s carcinogenic potential.

Absorption, Distribution, Metabolism, and Excretion (ADME)

Studies in humans and in animal models were identified that provided information on the ADME of ETBE and TBA. The focus for this assessment was on the most relevant review articles, recent peer-reviewed publications not covered in previous reviews, and summaries in regulatory assessments. These sources were used to summarize the metabolism and toxicokinetic properties of ETBE and TBA across different exposure routes, with emphasis on the similarities and/or differences between humans and animal models. Additionally, key metabolites of ETBE and/or TBA were identified to consider their potential carcinogenesis role.

Mechanistic data organized by KCC

Mechanistic data were organized into KCC following the IARC Preamble (IARC, 2019) and the framework described by Wikoff et al. (Wikoff et al., 2019). Guidance for mapping of endpoints to each KCC was established using publications with assays/endpoints grouped into KCC categories (Guyton et al., 2018, Madia et al., 2021, Smith et al., 2020). Data extraction of mechanistic evidence involved a systematic evaluation of study details and categorization of the model (e.g., in vitro: human, in vivo: mammalian, etc.) as well as assessing model strength, activity, and study/endpoint reliability.

As recently described by Borghoff et al. (2025a), study details including model, strain (e.g., Fischer), sex, age at start of exposure, route of exposure, dose level or concentration of ETBE, duration of exposure, sample analyzed (i.e., tissue), and evaluations of cytotoxicity in vitro or systemic toxicity in vivo were extracted. Endpoint details included KCC category (e.g., genotoxicity, oxidative stress), specific endpoint measured (e.g., chromosomal aberrations), activity status (active:1 vs. inactive:0), and endpoint summary (e.g., no significant increase in micronucleus formation). Activity status was reported based on effects at doses/concentrations determined to be statistically significant from control, as stated by the study authors, excluding effects occurring at cytotoxic doses in vitro or causing systemic toxicity in vivo which were considered unreliable data for inclusion in the overall evaluation. For in vitro studies,whether the study controlled for ETBE volatility was considered when integrating data. As ETBE would likely volatize out of the test system if not sealed, lack of control for volatility could potentially lead to false-negative findings. Study quality for all KCC data was evaluated using Klimisch criteria with reliability scores from 1 to 3: 1 = reliable without restriction, 2 = reliable with restriction, and 3 = not reliable. Reliability scores were informed by the ToxRTool red criteria (Schneider et al., 2009). Critical review of oxidative stress endpoints related to oxidative damage and reactive oxygen species (ROS) was evaluated for reliability using critical reviews by Halliwell and Whiteman (2004) and Murphy et al. (2022). Additionally, given the importance of genotoxicity data in cancer evaluations, these data were reviewed by SME with a focus on alignment with OECD standards widely accepted for genotoxicity. These guidelines were also used to guide the assessment of the assays’ overall reliability.

In this assessment, model strength was assigned based on a weighting structure from Wikoff et al. (2019). Data from exposed humans were rated highest with a score of 8, followed by in vitro human primary cells and in vivo mammalian study data with a score of 4. In vitro human cell lines, non-human mammalian primary cells and cell lines, and Ames assays (due to their high ability to predict mutagenicity) received a score of 2, while all other in vitro and in vivo non-mammalian models scored 1.

Consideration of Strength of Mechanistic Data Activity per KCC

Within each KCC, mechanistic data extracted were reviewed for consistency across study types and endpoints. Studies considered reliable (Klimisch score of 1 or 2) were included in data integration. Mechanistic data from studies considered not reliable (Klimisch score of 3) may be discussed in the evaluation within each KCC in instances where limited data were available, hence discussed for transparency, mainly to highlight why the data would not be used for consideration of strength of activity. All studies reviewed and extracted, along with reliability justification are provided in the Supplementary Data (Table S2-S3) which provide a weight-of-evidence approach to evaluate mechanistic activity within each KCC.

Data were also evaluated within specific model types, such as exposed humans, in vitro human primary cells and tissues, in vivo animal studies, in vitro human cell lines or non-human mammalian primary cells or cell lines. The in vitro non-mammalian Ames assays were categorized with in vitro non-human mammalian cells based on their predictive ability for mutagenic response. The strength of biological activity was assessed for each model type by looking at endpoints in reliable studies. The activity was categorized as:

• •Consistent activity
• •Inconsistent activity
• •Minimal or weak activity
• •Overall negative activity
• •No data or only data from unreliable studies

This evaluation was performed considering endpoint activity across KCC and to identify relevant biological activities (key events) that support specific MoAs for animal tumor formation.

Evaluation of mechanistic activity across KCC

Strength of the mechanistic data was classified with consideration of the IARC framework (IARC, 2019), with slight modifications as described by Borghoff et al., (2025a):

• Strong evidence: Assigned when the overall evidence base (across each KCC) was coherent with consistent activity in reliable studies/endpoints. This included: a) strong evidence in exposed humans; b) strong evidence in human primary cells or tissues (human receptors and enzymes, etc.); or c) strong evidence in experimental in vivo animal models supported by evidence in several studies in human primary cells or tissues.

• Limited evidence: Assigned when the overall evidence base (across each KCC) was suggestive of activity, with studies covering a narrow range of experiments, endpoints, and species. Also assigned because of inconsistencies in similar studies, and/or incoherence across studies of different endpoints or different experimental systems.

• Inadequate evidence: Assigned when there were few or no data available, or when there were unresolved questions about the adequacy of the study design, conduct, or interpretation, or the results were negative.

Overall, strength of mechanistic data

The overall strength of the mechanistic activity was evaluated by examining the robustness of the data within each KCC, as well as identifying individual KCC considered to have strong findings, or the combination of two of more KCC that together supported a strong indication of carcinogenic activity associated with exposure to ETBE or TBA. ETBE or TBA metabolism in humans and animals was also considered in reliability of model type where activity occurred for overall strength of data. Classification of overall strength of the mechanistic data was considered Strong, Limited, or Inadequate, similar to the description in IARC’s Preamble (IARC, 2019).

Integration across evidence streams

Assessment of the overall carcinogenic potential of ETBE and TBA in humans was conducted using a systematic approach, grounded in weight-of-evidence considerations and established scientific principles within and across these evidence streams (i.e., human cancer studies, animal cancer studies, and mechanistic data including specific rodent tumor MoA evidence). Tumor responses measured in reliable cancer bioassays in rodents and mechanistic data were integrated to inform the biological plausibility of carcinogenic response in humans. Metabolism of ETBE or TBA in animal models was compared to humans to inform transferability of data to human exposures. Mechanistic data, organized by KCC, were integrated into biological pathways (i.e. AOPs) for consideration of carcinogenic potential when overall activity (across model types) was strong. Strong (i.e consistent) KCC activity in experimental animal models was integrated into rodent-specific tumor MoAs within frameworks that identify criteria to support KE activity both for male rat kidney tumors and rodent thyroid tumors (EPA., 1991, EPA, 1998a, IARC., 1999). Mechanistic data considered inadequate were not integrated. In accordance with the IARC preamble and KCC guidance, due to the role of oxidative stress in numerous processes (not unique to carcinogenesis) oxidative stress would only be considered in the context of other strong mechanistic data within selected KCC and was not considered by itself for consideration of carcinogenic activity (IARC, 2019, IARC., 2025).

For the overall evaluation of carcinogenic potential of ETBE and TBA in exposed humans, relevance and reliability of each evidence stream were considered. Tumors identified in animal studies that form through a MoA considered not to operate in humans were not used as supporting evidence for carcinogenic potential in humans in the overall evaluation for human cancer hazard of ETBE and TBA.

Results

A total of 164 citations for ETBE and 1446 citations for TBA were prioritized via SWIFT and were screened for relevancy via TiAb from the initial literature search in February 2024. An additional 9 and 8 papers for ETBE and TBA, respectively were identified from hand-searching, giving a total of 173 publications for ETBE and 1454 publications for TBA. Of these, 40 publications for ETBE and 43 publications for TBA were retrieved and reviewed for relevance at the full-text level. No relevant human cancer studies were identified for either ETBE or TBA. Six studies in experimental animal models evaluated cancer outcomes associated with ETBE and two studies for TBA exposure. Twenty articles contained mechanistic data for ETBE, and 35 articles for TBA (Fig. 1).

Cancer studies in humans

No human cancer studies were identified in the peer-reviewed literature for either ETBE or TBA.

Cancer studies in experimental animal models

Two standard 2-year cancer bioassays were identified for ETBE (Saito et al., 2013, Suzuki et al., 2012), with one non-standard cancer bioassay (Maltoni et al., 1999) and three initiation-promotion studies considered non-standard cancer bioassays as well (Hagiwara et al., 2011, Hagiwara et al., 2015, Hagiwara et al., 2013). Two standard 2-year cancer bioassays were identified for TBA that were conducted in rats and mice (Cirvello et al., 1995, NTP, 1995).

ETBE animal cancer studies

Standard and non-standard cancer studies in animal models conducted for ETBE, along with study quality summaries are reported in Table 1. Study details and study quality summaries for initiation-promotion studies are provided in Supplemental Data Table S1 and criteria used for study reliability provided in Fig. S1-S2. In one reliable 2-year standard cancer bioassay, male and female F344 rats were chronically exposed (104 weeks) via inhalation to 0, 500, 1500 and 5000 ppm ETBE (high dose: 1350–1435 mg/kg-bw/day) with an increase in the incidence of hepatocellular adenomas observed in male, but not female rats (at the highest exposure concentration (p < 0.01) (Table 1) (Saito et al., 2013). Body weights in the high dose groups (male and female) were consistently lower than controls; at study termination body weights of exposed rats were 22 % (female) or 25 % (male) lower than controls. No significant increase in any tumor type was reported in a reliable standard 2-year cancer bioassay in which F344 rats were exposed to ETBE via drinking water (104 weeks) at water concentrations up to 10,000 ppm (∼550 mg/kg-bw/day) (Suzuki et al., 2012). Unlike exposure to MTBE (Bird et al., 1997), no kidney tumors were observed in male F344 rats following chronic exposure to ETBE via inhalation or drinking water (Saito et al., 2013, Suzuki et al., 2012).

Table 1.

Incidence of significant tumors reported in reliable ETBE and TBA standard cancer bioassays with study reliability scoring and rationale.^a

Test Compound Reference	Study design Species, strain (sex) Age at Start Duration	Route Purity Vehicle, Dose(s) No. of animals at start % of animals surviving in each dose group	Tumor Incidenceb (# tumors/# examined)	Overall Reliability Criteria used by SME to assign overall reliability category
ETBE Saito et al., 2013	Standard chronic bioassay Rat, F344/DuCrlCrlj (M&F) Age: 6 weeks Duration: 6 h/day, 5 days/week for 104 weeks	Inhalation Purity: >99 % Air 0, 500, 1,500, 5,000 ppm Calculated high dose: 1,350–1,435 mg/kg/dayc 50 M/ 50F for each dose level M: 88 %, 76 %, 79.6 %, 60 % F: 76 %, 78 %, 60 %, 60 %	No significant increase in tumor incidence attributed to ETBE exposure in female rats. Full necropsy and histopathological examination performed according to OECD TG 451 Hepatocellular adenoma M: 0/50d, 2/50, 1/49, 9/50d F: 1/50, 0/50, 1/50, 1/50 Hepatocellular carcinoma M 0/50, 0/50, 0/49, 1/50 F: 0/50, 0/50, 0/50, 0/50	Reliable Strengths: Generally, in compliance with OECD TG 451 and GLP requirements. Concentrations of ETBE vapor in inhalation chambers monitored by gas chromatography every 15 min. 50 rats/sex/dose group sufficient for statistical analysis. Limitations: Reduced survival in high dose males (attributed to increased CPN). Final mean body weight was > 20 % lower than controls in high dose males and females. No blinding or peer review process described for histopathology findings. Organ weight and clinical pathology measures taken only at terminal sacrifice.
ETBE Suzuki et al., 2012	Standard chronic bioassay Rat, F344/DuCrlCrlj (M&F) Age: 6 weeks Duration: 104 weeks	Oral (drinking water) Purity: >99 % Deionized and filtered tap water 0, 625, 2.500, or 10,000 ppm Calculated average dose: M: 0, 28, 121, 542 mg/kg-bw/week F: 0, 46, 171, 560 mg/kg-bw/day 50 M/ 50F for each dose level M: 76 %, 74 %, 68 %, 68 % F: 72 %, 74 %, 76 %, 76 %	No significant increase in tumor incidence attributed to ETBE exposure in male or female rats. Full necropsy and histopathological examination performed according to OECD TG 451.	Reliable Strengths: Generally, in compliance with OECD TG 451 and GLP requirements. 50 rats/sex/dose group sufficient for statistical analysis. ETBE stability in drinking water assessed following 8 days at room temperature and determined to be stable. Histopathology findings verified using a peer-review process. Limitations: Terminal mean body weight in high-dose females was > 15 % lower than controls. Terminal body weight in males < 10 % lower than controls. Organ weight and clinical pathology measures taken only at terminal sacrifice.
ETBE Maltoni et al., 1999	Non-standard carcinogenicity assay Rat, Sprague-Dawley (M&F) Age: 8 weeks Duration: 4 days/week for 104 weeks with animals observed until natural death	Oral (gavage) Purity: >94 % Olive oil 0, 250, 1,000 mg/kg-bw, 4-days/week 60 M/ 60F for each dose level [M: 25 %, 22 %, 11 %] [F: 30 %, 28 %, 35 %]	Full necropsy and histopathological examination performed. Total mouth (oral cavity, tongue and lips) epithelial tumors (group comprised acanthomas, squamous cell dysplasia, borderline dysplasia with in situ carcinoma and carcinomas) M: 6/60, 14/60, 15/60e Total uterine tumors (group comprised carcinomas and sarcomas) F: 2/69, 10/60e, 2/60; (Reevaluation by NTP confirmed no dose related trend). Note: Non-dose related increase in total number of forestomach lesions in males (acanthomas, squamous cell dysplasia, and squamous cell carcinoma). Increased incidence of lymphoma and leukemias (combined).	Not Reliable Strengths: Total number of animals in exposures groups were comparable to OECD TG 451 Limitations: Noted that the study was conducted in conformity to the principles of GLP, but not “according to GLP guidance”. Reporting is not presented in GLP manner with no evidence of auditing the full study. Reporting is limiting based on cancer outcomes report for multiple chemicals. Pathogen status of animals not provided; health monitoring surveillance not provided. Use of animals bred ‘in-house’ noted to have high incidence of infection with of M. pulmonis; influences tumor outcome. Presentation of tumor data did not follow standard practices in combining proliferative lesions. Lack of specific historical control data (same timeframe, dose route strain) provided. Statistical analyses failed to account for differences in survival age.
TBA Cirvello et al., 1995, NTP, 1995	Standard chronic bioassay Rat, F344/N (M&F) Age: 7 weeks Duration: 103 weeks	Oral (drinking water) TBA purity: >99 % Deionized water M: 0, 1.25, 2.5, or 5 mg/mL F: 0, 2.5, 5, or 10 mg/mL Calculated average dose: M: 90, 200, or 420 mg/kg-bw/day F: 180, 330, or 650 mg/kg-bw/day 60 M/ 60F for each dose level M: 33 %, 10 %, 7 %, 2 % F: 47 %, 40 %, 37 %, 20 %	No significant tumor responses attributed to TBA exposure in female rats. Full necropsy and histopathological examination performed. Renal adenoma, single (standard and extended evaluation) M: 7/50, 7/50, 10/50, 10/50 Renal adenoma, multiple (standard and extended evaluation) M: 1/50, 4/50, 9/50f, 3/50 Renal carcinoma (standard and extended evaluation) M: 0/50, 2/50, 1/50, 1/50 Renal adenoma or carcinoma combined (standard and extended evaluation) M: 8/50, 13/50, 19/50f, 13/50 Note: Step section evaluation of male rat kidneys performed in an extended analysis to evaluate the observed proliferative lesions	Reliable Strengths: Generally, in compliance with OECD TG 451 and GLP requirements. Dose formulations analyzed every ∼ 8 weeks. 60 rats/sex/dose group sufficient for statistical analysis. Histopathology findings verified using a peer-review process. Limitations: Survival of mid- and high-dose females was significantly lower than controls. More than 50 % of animals in each group survived to week 85 and was considered adequate for detection of possible effects. Final mean body weights in male and female high = dose groups were > 20 % lower than controls.
TBA Cirvello et al., 1995, NTP, 1995	Standard chronic bioassay Mouse, B6C3F1(M&F) Age: 7 weeks Duration: 103 weeks	Oral (drinking water) TBA purity: >99 % Deionized water 0, 5, 10, or 20 mg/mL Calculated average dose: M: 540, 1,040, or 2,070 mg/kg-bw/day F: 510, 1,020, or 2,110 mg/kg-bw/day 60 M/ 60F for each dose level M: 45 %, 60 %, 57 %, 28 % F: 60 %, 58 %, 68 %, 70 %	Full necropsy and histopathological examination performed. Thyroid gland follicular cell adenoma M: 1/60, 0/59, 4/59, 1/57 F: 2/58, 3/60, 2/59, 9/59 g,h Thyroid gland follicular cell adenoma and/or carcinoma combined M: 1/60, 0/59, 4/59, 2/58	Reliable Strengths: Generally, in compliance with OECD TG 451 and GLP requirements. Dose formulations analyzed every ∼ 8 weeks. 60 mice/sex/dose group sufficient for statistical analysis. Histopathology findings verified using a peer-review process. Limitations: Reduced survival in high-dose males, but at least 50 % survived to study termination and was sufficient for data analysis. Final mean body weight in high-dose females was > 10 % lower than controls. Cause of increase unknown, but TBA concentration was considered excessive for male mice.

^aInitiation-promotion studies for ETBE are available in Table S1.

^bNeoplastic lesion with increased statistical significance shown in bold.

^cCalculated ETBE uptake, based on a minute volume for male adult rats of 561 mL/min/kg body weight, by inhalation exposure (5,000 ppm) was approximately 4,222 mg/kg/day. Because the estimated respiratory uptake of ETBE was ∼ 32 %–34 %, ETBE intake by inhalation exposure (5,000 ppm) was equivalent to approximately 1,350–1,435 mg/kg/day (Hagiwara et al., 2015).

^dP < 0.01 by Fisher’s exact test and Peto’s test for dose response relationship.

^eP ≤ 0.05 by Chi squared test. Histopathology was re-evaluated by EPA and NTP and showed no dose related trends in oral or uterine tumors (Malarkey and Bucher, 2011).

^fP < 0.01.

^gP < 0.05 significantly different from the control group by the logistic regression test.

^hHistorical incidence of follicular cell adenomas was 8/238 (3.4 % ± 2.2 %) in female mice, with a reported range of 0–5 %. (NTP, 1997).

No additional tumors were identified in these standard chronic bioassays. However, in the non-standard bioassay (designated ‘not reliable;’ Supplementary Data Table S1; Figs. S1-2), oral and uterus/vagina tumors as well as lymphoma and leukemia were reported (Maltoni et al., 1999) (Table 1). In this study, Sprague-Dawley rats were administered 250 or 1000 mg/kg-bw/day ETBE by oral gavage, 4 days/week for 104 weeks, following which the animals lived until their natural death. When combined, the incidence of mouth epithelial (oral cavity, tongue and lips) lesions including acanthomas, squamous cell dysplasia, borderline dysplasia with carcinoma in situ, and carcinomas were increased in male rats administered 1000 mg/kg-bw/day group. This study was deemed ‘Not Reliable’, and findings were not integrated into the assessment of ETBE’s carcinogenic potential. Both the study design and the reporting had significant flaws which included the use of rats bred in an in-house facility not verified as pathogen free, thus leading to the potential presence of Mycoplasma pulmonis, known to cause chronic inflammatory reactions (Schoeb et al., 2009).

Three partially reliable initiation–promotion studies in male Wistar and F344 rats evaluated the tumor-promoting potential of ETBE at oral gavage doses up to 1000 mg/kg-bw/day following initiation with known potent mutagens such as DMBDD (which included diethylnitrosamine, methyl-nitrosourea, butyl-hydroxybutylnitrosamine (BBN), dihydroxypropylnitrosamine (DHPN), and dimethylhydrazine), BBN alone, or N-ethyl-N-(2-hydroxyethyl)nitrosa alone (Hagiwara et al., 2011, Hagiwara et al., 2015, Hagiwara et al., 2013). The details of these study designs are outlined in Supplementary Data Table S1. Reported findings from these studies included increased incidences of thyroid follicular cell adenomas and carcinomas, hepatocellular adenomas, colon adenocarcinomas, urinary bladder neoplasms, and renal tubule adenomas with initiation of differing mutagens and promotion of high ETBE dose levels. Given that only liver tumors were observed in one standard reliable 2-year cancer bioassay, these initiation–promotion studies did not provide independent evidence of carcinogenic potential of ETBE exposure only. As such these studies were not integrated for consideration of ETBE as a carcinogen in experimental animal models.

TBA animal cancer studies

Two standard chronic cancer bioassays were conducted in male and female B6C3F1 mice and F344 rats exposed to TBA via drinking water; these study findings and a summary of study quality assessment are provided in Table 1 and criteria used for study reliability provided in Supplementary Data Figs. S1-2. These studies were both considered reliable standard cancer bioasays. In exposed female mice, there was a significant increase in the incidence of thyroid gland follicular cell adenomas at the high concentration of 20 mg/mL (∼2110 mg/kg-bw/day) TBA compared to controls (9/59 vs 1/58, respectively; Table 1). The combined incidence of follicular cell adenomas or carcinomas was marginally, but not statistically, increased at 10 mg/mL TBA as reported. A single carcinoma was reported in one male at 20 mg/mL, but no carcinomas were reported in female mice (Cirvello et al., 1995, NTP, 1995). A significantly increased incidence of follicular cell hyperplasia (compared to controls) was observed in all exposed males and in females at 10 and 20 mg/mL TBA (Table 1). There was a significant dose-related trend in survival in male mice (p < 0.001) and in pairwise comparisons, and significantly fewer males in the high dose group survived to study termination compared to control mice (27 vs 17 mice respectively). No difference was observed in the low- and mid-dose group survival or body weight in male or female mice (Cirvello et al., 1995, NTP, 1995).

In male rats (0, 1.25, 2.5, or 5 mg/mL), proliferative lesions were observed in kidneys during the initial evaluation which led to an extended analysis and Pathology Working Group (PWG) review using kidney step sections (Hard et al., 2011). Results of these combined analyses showed a dose-related increase in renal tubule adenomas with the incidence in the 2.5 mg/mL group significantly greater than controls (Table 1). Renal tubule carcinomas were identified in two rats, one in the 2.5 mg/mL and 5 mg/mL groups. When combined, the incidence of renal tubule adenomas or carcinomas in the 2.5 mg/mL group was significantly greater than controls (19/50 vs 5/50, respectively) and there was a statistically significant dose related trend in tumor incidence. These rats also showed a statistically significant increased incidence of hyperplasia in the 5 mg/mL group (14/50 vs. 25/50) with advanced chronic progressive nephropathy (CPN) in male rats. A dose-related decreased survival that was attributed to the severity of CPN was reported in both male and female rats. From week 20, the body weight gain of high dose males was lower than controls, leading to a 24 % decrease for high dose males compared to controls. In F344 female rats, no tumors were reported in this 2-year drinking water study (0, 2.5, 5, or 10 mg/mL TBA).

Mechanistic evidence

ADME: Metabolism and toxicokinetics of ETBE in humans and animal models

Biotransformation of ETBE is qualitatively and quantitatively similar in humans and rats following inhalation exposure (Dekant et al., 2001). Metabolism in both species is mediated through cytochrome P450 oxidation to tertiary-butyl alcohol (TBA) and acetaldehyde as reviewed in McGregor, 2007 with a more recent summary presented in the EPA IRIS ETBE assessment (EPA., 2021a, McGregor, 2007) (Fig. 2). Cytochrome P450 metabolic oxidation is commonly associated with high-dose metabolic saturation. Several lines of evidence support ETBE being metabolized by liver P450 enzymes in humans (CYP2A6) (Hong et al., 1999, Le Gal et al., 2001) and rats (CYP2B1) (Turini et al., 1998). Although generation of acetaldehyde, an endogenous metabolite, may be of concern associated with disease outcomes, it is rapidly detoxified to acetic acid unless metabolism via aldehyde dehydrogenase is saturated with the potential to form DNA adducts (McGregor, 2007). The two main metabolic products derived from the metabolism of ETBE’s major metabolite TBA, are 2-methyl-1,2-propanediol and α-hydroxyisobutyric acid (Bernauer et al., 1998), both of which have been detected in the urine of rats and humans exposed to ETBE (Amberg et al., 1999, Amberg et al., 2000). Kinetics of excretion of ETBE and its metabolites is slower in humans compared to rats, however, ETBE is still rapidly exhaled and metabolized in humans with metabolites rapidly eliminated in urine, hence not anticipated to accumulate in human tissues (Amberg et al., 2000, Dekant et al., 2001).

Fig. 2.

Proposed metabolic pathway for ETBE and TBA in humans and rats. (Sources include: Amberg et al., 2000, Bernauer et al., 1998; McGregor, 2010).

ADME: Metabolism and toxicokinetics of TBA in humans and animal models

TBA metabolism is also similar in humans and rats as noted in Fig. 2. TBA is metabolized to 2-methyl-1,2-propanediol and α-hydroxyisobutyrate (Bernauer et al., 1998, McGregor, 2007, McGregor, 2010), with both metabolites detected in human and rat urine following exposure to ETBE, or TBA alone (McGregor, 2007, McGregor, 2010). With intravenous doses of TBA (37–300 mg/kg) to rats, the distribution half-life of TBA is approximately 3 min and the elimination half-life approximately 3.8 h in male and female rats for doses of less than 300 mg/kg-bw. At a dose of 300 mg/kg-bw, the half-life increased to 4–5 h (Poet & Borghoff, 1997).

ETBE and TBA mechanistic data organized and evaluated by KCC

Mechanistic evidence for ETBE and TBA were organized using the KCC framework, as described in the methods, to assess the biological plausibility of a response in humans, and/or consider plausibility of proposed MoAs for the tumors identified in animal models that do not, or are not likely to operate in humans. Both EPA., 1991, EPA, 1998a) and IARC (1999) have established criteria in which rodent tumor MoAs (i.e., kidney, and thyroid) are supported. These criteria frameworks were used to identify data to support KEs in these MoAs. Summaries of endpoint and assay data identified and mapped within the KCC framework or support KEs in MoAs (described as mechanistic data) are available in Supplementary Data, Table S2 or S3 for ETBE or TBA, respectively. Consideration of weight-of evidence of mechanistic activity within each KCC was made for overall data and by model category (Supplementary Data Fig. S3 and S4). KCC 2 “Is genotoxic,” KCC 8 “Modulates receptor-mediated effects”, and KCC 10 “Alters cell proliferation, cell death and nutrient supply” were considered to have strong mechanistic data in experimental animals models (Table 3a, Table 3b, Table 3c, Table 3d; Table 4; Supplementary Data Table S-S3), and were used to support the evaluation of tumor specific MoAs for ETBE and TBA. Since other KCCs did not have strong activity overall across model types, information was not available to consider other possible biological pathways (i.e., AOPs) for consideration of carcinogenic potential in humans. KCC with minimal data were not discussed. A high-level summary of available findings is presented in Table 2, and additional information on these KCC data can be found in Supplementary Data Table S2-S3.

Table 3a.

In vitro non-mammalian mutagenicity assays for ETBE and TBA. (All Data Provided in Supplementary Data; Table S2-S3).

Chemical Name	Assay Endpoint	Test System (species, strain)	Concentration of Test Substancea	Results		Comments/ Quality Assessmentc	Reference
				Without Metabolic Activation	With Metabolic Activation
ETBE	Ames Assay	S. typhimurium TA1535, TA97, TA100, TA98	0, 100, 333, 1000, 3333, 10000 μg/plate	−	−	10 % and 30 % Rat and Hamster S9 were utilized	Zeiger et al., 1992
TBA	Ames Assay	S. typhimurium TA98, TA100, TA1535, and TA1537	0, 100, 333, 1000, 3333, 10,000 μg/plate	−	−	10 % hamster S9, 10 % rat S9 was utilized	Zeiger et al., 1987; Mahler, 1997
TBA	Ames Assay	S. typhimurium TA102	0, 100, 200, 500, 1000, 2500, 5000 μg/plate	−	−	S9 was utilized; in accordance with OECD TG 471	McGregor et al., 2005
TBA	Ames Assay	S. typhimurium TA102	0, 0.75, 1.5, 2, 3, 3.75 mg/plateb [750–3750 µg/plate]	NT	+	S9 was utilized	Williams-Hill et al., 1999

Abbreviations: NT – not tested; “-“ negative activity; “+” positive activity; “+/-“ equivocal results

^aSignificant concentrations/doses are bolded.

^bApproximate concentrations determined from graph.

^c“–” no comments on study or quality assessment worth noting.

Table 3b.

In vitro mammalian mutagenicity and genotoxicity assays for ETBE and TBA. (All Data Provided in Supplementary Data; Table S2-S3.

Chemical Name	Assay, Endpoint	Species, Tissue/Cell Line	Concentrationa	Results	Comments/Quality Assessmentb	Reference
ETBE	Chromosomal Aberration	Chinese Hamster Ovary (CHO)	0, 0.1, 0.3, 1, 3, 5 mg/mL	−	Conducted in the absence and presence of S9	Vergnes, 1995
TBA	Mutant Colonies	Mouse, L5178Y	0, 1000, 2000, 3000, 4000, 5000 µg/mL; 0, 625, 1250, 5000 µg/mL	−	Conducted in the absence and presence of S9	McGregor, 1988, Mahler, 1997
TBA	Sister Chromatid Exchange	Chinese Hamster Ovary (CHO)	0, 160, 500, 1600, 5000; 2000, 3000, 4000, 5000 µg/mL	−	No longer an accepted method for measuring genotoxicity (OECD, 2014). Non-reproducible positive result in second trial for studies conducted in the absence of S9	Galloway et al., 1987, Mahler, 1997
TBA	Chromosomal Aberration	Chinese Hamster Ovary (CHO)	0, 160, 500, 1600, 5000; 2000, 3000, 4000, 5000 µg/mL	−	Cytotoxicity observed at concentration at which effects were observed (5000 µg/mL)	Galloway et al., 1987, Mahler, 1997
TBA	Comet Assay, DNA strand breaks	Rat-1, fibroblast	0, 0.44 mM	+	–	Sgambato et al., 2009

Abbreviations: NT – not tested; “-“negative activity; “+” positive activity; “+/-“ equivocal results.

^aSignificant concentrations are bolded.

^aTang et al. (1997) data for TBA was not incorporated due to inability to translate publication.

^b“–” no comments on study or quality assessment worth noting.

Table 3c.

In vivo mammalian mutagenicity and genotoxicity assays for ETBE. (All Data Provided in Supplementary Data; Table S2).

Assay, Endpoint	Sex, Species, Tissue	Dosea	Dose Regimen (route); Study Duration	Results	Comments/Quality Assessmentb	Reference
cll mutation frequency	Male, Fischer 344 Big Blue homozygous transgenic rat, Bone marrow	0, 500, 1500, 5000 ppm	Inhalation, 6 h/day, 28 days	−	Performed in accordance with OECD TG 488.	Gollapudi and Rushton, 2023
cll mutation frequency	Male, Fischer 344 Big Blue homozygous transgenic rat, Liver	0, 500, 1500, 5000 ppm	Inhalation, 6 h/day, 28 days	−	Performed in accordance with OECD TG 488.	Gollapudi and Rushton, 2023
Micronucleus	Male & Female, C57BL/6 Mouse, Blood	0, 500, 1750, 5000 ppm	Inhalation, 6 h/day 5 day/wk for 13 wks	+	Effects only observed in males.	Weng et al., 2013
Micronucleus	Male & Female, F344/DuCrlCrlj Rat, Bone Marrow	0, 1600, 4000, 10,000 ppm	Drinking Water, 13 weeks	−	Performed in accordance with OECD TG 408, 413 and 474 and OECD Principles of GLP.	Noguchi et al., 2013
Micronucleus	Male & Female, F344/DuCrlCrlj Rat, Bone Marrow	0, 50, 150, 500, 1500, 5000 ppm	Inhalation, 6 hr/day 5 day/wk for 13 wks	−	Performed in accordance with OECD TG 408, 413 and 474 and OECD Principles of GLP.	Noguchi et al., 2013
Micronucleus	Male & Female, F344/DuCrlCrlj Rat, Bone Marrow	0, 250, 500, 1000, 2000 mg/kg/day	IP injection, 13 weeks	−	Performed in accordance with OECD TG 408, 413 and 474 and OECD Principles of GLP.	Noguchi et al., 2013
Micronucleus	Male & Female, F344/DuCrlCrlj Rat, Bone Marrow	0, 500, 1500, 5000 ppm	Inhalation, 6 h/day, 5 days/week, 13 weeks	−	–	JBRC, 2007
Micronucleus	Male & Female, CD-1 Mouse, Bone Marrow	0, 400, 2000, 5000 ppm	6 h/day, 5 days	−	–	Vergnes and Kubena, 1995
Micronucleus	Male, B6C3F1 Mouse, NR	0, 1300, 1700 mg/kg	IP injection, 72 h	−	–	NTP, 2018a
Micronucleus	Male, F344 Rat, NR	0, 625, 1250, 2500 mg/kg	IP injection, 72 h	−	–	NTP, 2018b
Comet Assay, DNA damage	Male & Female, C57BL/6 Mouse, Liver, Leukocytes	0 500, 1750, 5000 ppm	Inhalation, 6 h/day, 5 days/week, 13 weeks	+	Increase in males at highest dose, no effect in females. Authors did not measure effects on bodyweight. Mice were sacrificed 20–24 h post-exposure OECD TG 489 recommends 2–6 h post-exposure.	Weng et al., 2011, Weng et al., 2012
Comet Assay, DNA damage	Male, C57BL/6 Mouse, Sperm	0, 500, 1750, 5000 ppm	Inhalation, 6 h/day 5 day/wk for 13 wks	+	Mice were sacrificed 20 h post-exposure, OECD TG 489 recommends 2–6 h post-exposure. Changes in body weight were not reported.	Weng et al., 2014
Comet Assay, DNA damage	Male, C57BL/6 Mouse, Sperm	0, 50, 200, 500 ppm	Inhalation, 6 h/day 5 day/wk for 9 wks	−	Mice were sacrificed 20–24 h post-exposure OECD TG 489 recommends 2–6 h post-exposure. Changes in body weight were not reported.	Weng et al., 2014
Comet Assay, DNA damage	Male, C57BL/6 Mouse, Blood	0, 50, 200, 500 ppm	Inhalation, 6 h/day 5 day/wk for 9 wks	−	Time mice were sacrificed were not specified. OECD TG 489 recommends 2–6 h post-exposure. Changes in body weight were not reported.	Weng et al., 2019
hOOG1 modified Comet assay, Oxidative DNA damage	Male & Female, C57BL/6 Mouse, Liver	0 500, 1750, 5000 ppm	Inhalation, 6 h/day 5 day/wk for 13 wks	+	Increase in males at highest dose, no effect in females. Mice were sacrificed 20–24 h post-exposure. OECD TG 489 recommends 2–6 h post-exposure. Changes in body weight were not reported.	Weng et al., 2012
hOOG1 modified Comet assay, Oxidative DNA damage	Male, C57BL/6 Mouse, Sperm	0, 500, 1750, 5000 ppm	Inhalation, 6 h/day 5 day/wk for 13 wks	+	Mice were sacrificed 20 h post-exposure, OECD TG 489 recommends 2–6 h post-exposure. Changes in body weight were not reported.	Weng et al., 2014
hOOG1 modified Comet assay, Oxidative DNA damage	Male, C57BL/6 Mouse, Sperm, Liver	0, 50, 200, 500 ppm	Inhalation, 6 h/day 5 day/wk for 9 wks	−	Mice were sacrificed 20 h post-exposure, OECD TG 489 recommends 2–6 h post-exposure. Changes in body weight were not reported.	Weng et al., 2014, Weng et al., 2019
hOOG1 modified Comet assay, Oxidative DNA damage	Male & Female, C57BL/6 Mouse, Blood	0, 500, 1750, 5000 ppm	Inhalation, 6 h/day 5 day/wk for 13 wks	−	OECD TG 489 recommends 2–6 h post-exposure. Changes in body weight were not reported.	Weng et al., 2013

Abbreviations: NT – not tested; “-“negative activity; “+” positive activity; “+/-“ equivocal results.

^aSignificant dose levels are bolded.

^b“–” no comments on study or quality assessment worth noting.

Table 3d.

In vivo mammalian mutagenicity and genotoxicity assays for TBA. (All Data Provided in Supplementary Data; Table S3).

Assay, Endpoint	Sex, Species, Tissue	Dosea	Dose Regimen (route); Study Duration	Results	Comments/Quality Assessmentb	Reference
Comet Assay, DNA strand breaks	Male, CD-1 Mouse, Thyroid	0, 125, 250, 500, 1000, 2000 mg/kg/day	Gavage, 2 days	−	Informed by OECD TG 489	Thompson et al., 2024
Comet Assay, DNA strand breaks	Male, C57BL/6 Mouse, Leukocytes	0, 5, 20 mg/mL	Drinking Water, 6 weeks	+	–	Lin et al., 2020
Micronucleus Assay, Micronucleated PCE	Male, F344 Rat, Bone Marrow	0, 39, 78.125, 156.25, 312.5, 625, 1250	IP injection, 3x for 24 h	−	Mortality of all animals reported at highest dose evaluated; units of doses not provided	Mahler, 1997
Micronucleus Assay, Micronucleated NCE	Male & Female, B6C3F1 Mouse, Bone Marrow	0, 3000, 5000, 10000, 20000, 40000 ppm	Drinking Water, 13 weeks	−	–	Mahler, 1997

Abbreviations: NT – not tested; “-“negative activity; “+” positive activity; “+/-“equivocal results.

^aSignificant dose levels are bolded.

^b“–” no comments on study or quality assessment worth noting.

Table 4.

In vitro and in vivo mammalian cell proliferation, cell death and nutrient supply assays for ETBE and TBA (All Data Provided in Supplementary Data; Table S2-S3)*.

Chemical Name	Endpoint, Assay	Sex, Species, Tissue	Dose/Concentrationa	Dose Regimen (route); Study Duration	Results	Comments/Quality Assessment	Reference
In vitro mammalian
TBA	Cell Cycle Arrest, Protein levels (retinoblastoma protein (pRb), cyclin-dependent kinase inhibitor (p27) and cyclin D1)	Rat-1, fibroblast	0, 0.44 mM	48 h	↑	–	Sgambato et al., 2009
TBA	Cell Cycle Arrest, G0/G1 phase	Rat-1, fibroblast	0, 0.44 mM	48 h	↑	–	Sgambato et al., 2009
TBA	Cell Proliferation; Thymidine Incorporation,	Rat, primary astrocytes	0, 0.1, 0.3 % (v/v)	NR	↓	Decrease cell proliferation in phorbol ester stimulated cells. Klimisch = 3; No cytotoxicity measured.	Kötter et al., 2000
TBA	Cell Proliferation Thymidine Incorporation,	Rat, primary astrocytes	0, 0.3 %	NR	↓	Decrease in cell proliferation in PDGF-BB stimulated cells. Klimisch = 3; No cytotoxicity measured.	Kötter and Klein, 1999
In vivo mammalian
ETBE	Cell Proliferation BrDU labeling,	Male, F344/DuCrlCrl Rat, Liver	0, 1000 mg/kg,	Oral Gavage, 2x/day for 3, 10, 17, 28 days	↑	Increased cell proliferation at 3 and 28 days	Kakehashi et al., 2015
ETBE	Cell Proliferation, BrDU labeling,	Male & Female, CD-1 Mouse, Liver	0, 500, 1750, 5000 ppm	Inhalation, 1, 4, 13 weeks	↑	Males: increased labeling indices (LI) at1750 and 5000 ppm (1, 4 weeks). No change at week 13. Females: increase in LI at 1750 and 5000 ppm week 1 and week 13.	Medinsky et al., 1999
ETBE	Cell Proliferation, BrDU labeling	Male & Female, F344 Rat, Kidney	0, 500, 1750, 5000 ppm	Inhalation, 1, 4, 13 weeks	↑	Males: Increase in LI at all exposure concentration at week 1, and at 5000 ppm at week 4. No change at week 13. Females: LI at 5000 ppm (all timepoints). At 13-week increased LI at 500 and 1750 ppm, but not at 5000 ppm.	Medinsky et al., 1999
ETBE	Cell Proliferation, PCNA labeling	Male, F344/ DuCrlCrlj Rat, Liver	0, 150, 1000 mg/kg bw/day	Oral Gavage, 2x/day for 2 weeks	↑↓	Increased PCNA positive cells at high dose at week 1 with decrease in PCNA positive cells at high dose at week 2	Kakehashi et al., 2013
ETBE	Apoptotic Cells, ssDNA staining,	Male, F344/ DuCrlCrlj Rat, Liver	0, 150, 1000 mg/kg bw/day	Oral Gavage, 2x/day for 2 weeks	↑	Increase in apoptotic cells at high dose at week 2 but not week 1	Kakehashi et al., 2013
TBA	Cell Proliferation, BrdU labeling	Male & Female, F344 Rat, Kidney	0, 250, 450, 1750 ppm	Inhalation; 6 h/day for 8 or 10 days	↑	Significant increase in cell proliferation in males, but not females.	Borghoff et al., 2001
TBA	Cell Proliferation, PCNA labeling	Male, F344 Rat, Kidney	0, 0.25, 0.5, 1, 2, and 4 % (W/V)	Drinking Water; 90 days	↑	Increased cell proliferation at 2 % and 4 %. Significant mortality reported in at 4 % at 5 and 12 weeks.	Takahashi et al., 1993, Lindamood et al., 1992
TBA	Histopathology, Transitional epithelium hyperplasia	Male & Female, F344 Rat, Urinary Bladder	0, 0.25, 0.5, 1, 2, and 4 % (W/V)	Drinking Water; 94–95 days	↑	Males: reported at 4 % (7/10) Females: reported at 4 % (3/10) Significant mortality reported in males and females at 4 %.	Cirvello et al., 1995, Lindamood et al., 1992
TBA	Histopathology, Transitional epithelium hyperplasia	Male & Female, B6C3F1 Mouse, Urinary Bladder	0, 0.25, 0.5, 1, 2, and 4 % (W/V)	Drinking Water; 94–95 days	↑	Males: reported at 2 % (6/10) and 4 % (10/10). Females: reported at 4 % (3/9) Significant mortality in males at 4 % and significant decrease in bodyweight for both males and females at 4 %.	Cirvello et al., 1995, Lindamood et al., 1992
TBA	Cell Proliferation, Ki-67 immunohistochemical staining	Female, B6C3F1 Mouse, Thyroid	0, 5, 10, 20 mg/mL	Drinking Water; 14- or 28-days	−	No significant change in cell proliferation.	Borghoff et al., 2025b

Abbreviations: NT – not tested; “↑” increase activity; “↓” decreased activity; “↑↓” equivocal results, “-” negative results.

*All assays with Klimisch Scores of 1 or 2 are included. Assays with Klimisch scores of 3 are added if activity is indicated but are not used in the evaluation of KCC strength.

^b“–” no comments on study or quality assessment worth noting.

^aSignificant doses/concentrations are bolded.

Table 2.

Summary of Overall Mechanistic Data Activity by KCC (Supplemental Tables S2-3 and Figs. S3-4).

Key Characteristics	Overall KCC Activity Call and Evidence
	ETBE	TBA
1. Is electrophilic or can be metabolically activated to an electrophile	No data available; ‘inadequate’	Overall evidence was considered ‘inadequate’. Data was available in one experimental animal study. However, data were considered unreliable based on methodology.
2. Is genotoxic	Evidence for genotoxicity in experimental animals, animal primary cells, and human or animal cell lines was overall ‘limited’. Available data was predominantly inactive for endpoints related to DNA damage, oxidative DNA damage, micronucleus formation, chromosomal aberrations along with mutation frequency, mutagenicity.	Available data for genotoxicity in experimental animals, animal primary cells, or human or animal cells lines was ‘limited’. Endpoints related to genotoxicity and mutagenicity were overall inactive in vivo and in vitro. Available genotoxicity and mutagenicity data included endpoints related to DNA damage, micronucleus formation, sister chromatid exchange, and chromosomal aberrations along with mutation frequency and mutagenicity.
3. Alters DNA repair or causes genomic instability	Evidence was overall ‘inadequate’. One in vivo study reported decreased expression of an oxidative DNA repair enzyme.	No data available; ‘inadequate’
4. Induces epigenetic alterations	No data available; ‘inadequate’	Evidence was overall ‘inadequate’. One in vitro study reported an increase in expression and protein levels for a NAD + dependent deacetylase in a mouse hippocampal cell line with a significant change at one concentration where cytotoxicity was not observed.
5. Induces oxidative stress	Evidence was overall ‘limited’ in two in vivo studies measuring levels of 8-OHdg in rats. Activity was reported only at the highest dose in both studies.	Evidence for oxidative stress in experimental animals and in vitro were ‘limited.’ A significant increase in oxidative DNA damage was reported in two in vivo and one in vitro study. Alterations in antioxidant enzymes were reported at a single dose in vitro. Inconsistent alterations in antioxidant enzymes were reported in three in vivo studies.
6. Induces chronic inflammation	No data available; ‘inadequate’	Evidence was overall ‘inadequate’ for chronic inflammation. One reliable in vivo study reported increased inflammation in the urinary bladder.
7. Is immunosuppressive	Evidence was overall ‘limited’ for immunosuppression endpoints in experimental animals. Consistent negative activity for changes in white blood cell levels, inconsistent activity for IgM formation, and decreased T-cell levels were reported.	Evidence was overall ‘inadequate’. One in vivo guideline study reported significant effect on IgM formation.
8. Modulates receptor-mediated effects	Evidence in experimental animals was overall ‘limited’. Alterations in CAR, PXR, PPAR, and liver metabolizing enzymes were reported in experimental animals.	Evidence in vitro and in experimental animals was overall ‘limited’. Endpoints related to steroidogenesis, and estrogen or androgen receptor binding were overall inactive in vitro. In experimental animals in one study T3 and T4 (but not TSH) levels were altered. Alterations in liver metabolizing enzymes were also reported.
9. Causes immortalization	No data available; ‘inadequate’	No data available; ‘inadequate’.
10. Alters cell proliferation, cell death, or nutrient supply	Available data for cell proliferation and apoptosis in experimental animals was ‘limited’. No data was available in vitro. Strong data for cell proliferation in liver of experimental animals that resulted in tumors; however, these effects were observed in other tissues as well where no tumors were observed (e.g., kidney, bladder).	Available data for cell proliferation in experimental animals and in vitro was ‘limited’. Increased cell proliferation and hyperplasia were noted in experimental animals exposed to TBA. These changes were observed in tissues in which tumors were reported (i.e., kidney in male rats) or at high exposure concentrations at or above the maximum tolerated dose (urinary bladder).

ETBE

No data were available for “KCC 1; Is electrophilic or can be metabolically activated to an electrophile,” “KCC 4; Induces epigenetic alterations,” “KCC 6; Induces chronic inflammation,” or “KCC 9; Causes immortalization,” and therefore, the strength of the mechanistic data for these KCC was considered as ‘Inadequate’ (Fig. 3a).

Fig. 3.

A) Evaluation of overall mechanistic activity for ETBE and B) Evaluation of overall mechanistic activity for TBA. ¹Similar to IARC (IARC (International Agency for Research on Cancer), 2019), strength of overall mechanistic activity per KCC was defined using the following criteria, Strong evidence: consistent and coherent activity in reliable studies/endpoints in (a) exposed humans, (b) human primary cells or tissues or (c) experimental in vivo animal models supported by evidence in several studies in human primary cells or tissues. Limited evidence: Inconsistent activity in similar studies and/or incoherent activity across studies with different endpoints or experimental systems. Assigned when the overall evidence base (across each KCC) was suggestive of activity, with studies covering a narrow range of experiments, endpoints, and/or species. Inadequate evidence: Assigned when there were few or no data available, or when there were unresolved questions about the adequacy of the study design, conduct, or interpretation, or the results were negative. ² The overall strength of the mechanistic activity was evaluated by examining the robustness of the data within each KCC, as well as identifying individual KCCs considered to have strong findings, or the combination of two of more KCCs that together supported a strong indication of carcinogenic activity associated with exposure to ETBE or TBA. Classification of overall strength of the mechanistic data was considered Strong, Limited, or Inadequate, similar to the description in IARC’s Preamble (IARC, 2019).

KCC 2 “is genotoxic” (Table 2; Supplementary DataTable S2

ETBE was evaluated for mutagenic activity in the bacterial reverse mutation test (Ames assay) in four of the five bacterial strains recommended by OECD 471 (Salmonella typhimurium TA100, TA1535, TA97, TA98) with and without metabolic activation. No activity was reported at levels of ETBE up to 10,000 µg/plate in the absence or presence of S9 (Table 3a) (Zeiger et al., 1992). No chromosomal aberrations were reported in Chinese hamster ovary (CHO) cells exposed to ETBE at concentrations ranging from 0.1-5 mg/mL with and without S9 (Table 3b) (Vergnes, 1995).

An inhalation study in transgenic male F344 Big Blue rats exposed to 0, 500, 1500, or 5000 ppm ETBE for 28 days (6 h/day) (OECD TG 488) reported no changes in the mutant frequencies at the cll locus of the transgene in liver and bone marrow (Gollapudi & Rushton, 2023). These results confirm the lack of mutagenicity of ETBE in male rats exposed under the same conditions (inhalation exposure up to 5,000 ppm) that resulted in liver tumors in chronically exposed male F344 rats, a rodent model whose metabolism is similar in humans (Table 3c).

Micronucleus formation in blood, bone marrow and liver in mice and rats exposed to ETBE via inhalation, intraperitoneal injection, or drinking water was evaluated. A significant increase in blood micronucleus formation was observed in male, but not female C57BL/6 mice exposed to 5000 ppm ETBE for 6 h/day, 5 days/week for 13 weeks (Weng et al., 2013). In a study conducted in accordance with OECD TG 474, no significant increase in bone marrow micronucleus formation was measured in bone marrow of F344 male or female rats exposed to various concentrations of ETBE via inhalation, drinking water or intraperitoneal administration for 13 weeks (Noguchi et al., 2013). Additionally, no significant change in micronucleus formation in bone marrow was reported in several study reports in mice (B6C3F1 or CD-1) or rats (F344/DuCrlCrlj) exposed to ETBE by inhalation (400–5,000 ppm) or administered by intraperitoneal injection (625–2500 mg/kg) (JBRC., 2007, NTP, 2018a, NTP, 2018b, Vergnes and Kubena, 1995) (Table 3c).

DNA damage was measured in non-guideline in vivo studies using an alkaline comet assay and human 8-oxoguanine DNA N-glycosylase 1 (hOGG1) modified comet assay used for measuring oxidative DNA damage. In a series of studies conducted by Weng et al. (Weng et al., 2014, Weng et al., 2019, Weng et al., 2011, Weng et al., 2012, Weng et al., 2013), male and female C57BL/6 mice were exposed via inhalation to ETBE for different durations, and various concentrations up to 5000 ppm, and with different tissues (blood, sperm, liver and leucocytes) evaluated. Positive responses, such as increased DNA damage, were only noted in liver, sperm and leukocytes following 13 vs 9 weeks of exposure in males (Table 3c). Authors did not measure body weight or other clinical signs of overt toxicity, with the exception of Weng et al. (2019) which reported no significant effects on bodyweight at 50–500 ppm via inhalation exposure which are at concentrations much lower than concentrations in which DNA damage was measured, supporting a high dose effect associated with toxicity.

Data in experimental animal models was inconsistent across endpoints in reliable studies, however, results conducted in accordance with OECD Guideline 488 and 474 were consistently negative (Gollapudi and Rushton, 2023, Noguchi et al., 2013). These findings combined with overall negative results in animal primary cells/cells lines, lead to the conclusion that ETBE is non-genotoxic. The lack of available data in exposed human and primary human cells/tissues lead to an overall strength of evidence for KCC 2 “Is genotoxic”to be ‘limited.’.

KCC 5 “Induces oxidative stress” (Supplementary Data, Table S2)

Two studies measured the effects of ETBE on oxidative stress endpoints in the liver of mice (via inhalation: 500–5000 ppm) and rats (via oral gavage: 150, 1000 mg/kg-bw/day). In mice, increased 8-OHdg levels were reported in male but not female rats at the highest dose (Weng et al., 2012). In a study conducted in rats, significant hydroxyl radical generation and increased 8-OHdg levels were reported at the highest dose (5000 ppm) using two different reliable methods of measurement (Kakehashi et al., 2013).

Overall, the strength of evidence for KCC 5 “Induces oxidative stress” is considered ‘limited’ based on data in two studies in experimental animals measuring levels of 8-OHdg in rats, and lack of data in exposed humans or human primary cells/tissues.

KCC 8 “Modulates receptor-mediated effects” (Table 2; Supplementary Data, Table S2)

No data was publicly available in the peer-reviewed literature for effects on the androgen (AR), estrogen (ER) or thyroid receptor (ThR). However, the EPA’s CompTox Chemicals Dashboard predicted that ETBE would not bind to or activate the estrogen (ER) or androgen receptor (AR), using the COMPARA and CERAPP models for AR and ER, respectively (EPA, 2024a).

Data was available to assess the consideration of constituitive androstane receptor (CAR), pregane X Receptor (PXR) and Peroxisome Proliferator-Activated Receptor (PPAR) activation in animals exposed to ETBE. One study investigated receptor-mediated effects of ETBE in the liver of F344 rats administered ETBE by oral gavage at 150, and 1000 mg/kg bw/day. A significant increase in expression of the PPARα and PPARγ genes was reported in the liver following 1 week of ETBE administration (1000 mg/kg-bw/day). A significant increase in expression of CYP enzymes and P450 content and a significant decrease in UGT1a1 gene expression were also reported following 2 weeks of ETBE administration (1000 mg/kg-bw/day) (Kakehashi et al., 2013). Kakehashi et al. (2013) conducted proteomics and evaluated gene expression changes in rat liver and predicted activation of upstream regulators using ingenuity pathway analysis (IPA). IPA results predicted activation of CAR/PXR nuclear receptors as well as PPARγ, PPARδ, with possible activation of PPARα (see Tumor Specific MoA in Animal Models for details on the human relevance of these receptors in liver tumors). However, IPA did not predict activation of the AhR. In this study peroxisome proliferation, particularly for PPARα, was further investigated using transmission electron microscopy (TEM). TEM findings in hepatocytes showed increased peroxisomes at week 2 of ETBE administration for the low- and high-dose groups (Kakehashi et al., 2013). In other studies, liver hypertrophy was observed in F344 or Sprague-Dawley rats or CD-1 or C57BL/6 mice exposed to ETBE via inhalation (500–5000 ppm) or in drinking water (dose levels estimated to be 5–1000 mg/kg bw/day) (Kakehashi et al., 2013, Kakehashi et al., 2015, Medinsky et al., 1999, Miyata et al., 2014, Weng et al., 2012).

Data in experimental animal models was consistent for activation of CAR/PXR or PPAR and was considered as strong supporting data for integration into the tumor specific MoA for ETBE (see Tumor Specific MoA in Animal Models). However, due to the lack of available data for exposed humans or human primary cells/tissues the overall strength of evidence for KCC 8 “Modulates receptor-mediated effects” is ‘limited.’

KCC 10 “Alters cell proliferation, cell death, and nutrient supply” (Table 2; Supplementary Data, Table S2)

In CD-1 mice, a dose-dependent increase in cell proliferation was reported in hepatocytes following inhalation exposure (500–5000 ppm ETBE). Increased labeling indices were observed at 1 week (male and female mice), 4 weeks (male mice), and 13 weeks (female mice) (Medinsky et al., 1999). In male Fischer (F344) rats, increased cell proliferation was reported in hepatocytes at 3- and 28-days following administration of 1000 mg/kg bw/day 2×/day for 3, 10, 17, or 28 days via oral gavage (Kakehashi et al., 2015). In another study, decreased cell proliferation in male F344 rat hepatocytes was accompanied by increased apoptotic cells with 2 weeks of administration via oral gavage of 1000 mg/kg bw/day ETBE or 150 or 1000 mg/kg bw/day two times per day for 2 weeks (Kakehashi et al., 2013). In the renal proximal tubules cells, increased cell proliferation was observed in male F344 rats at all inhalation concentrations (500, 1750, 5000 ppm ETBE) at 1 week and at 5000 at 4 weeks with no significant change at 13 weeks. In females, a significant increase was observed at 5000 ppm at weeks 1 and 4 and at all concentration at week 13 (Medinsky et al., 1999) (Table 4).

Data in experimental animals was consistent for cell proliferation in hepatocytes and was considered as strong supporting data for integration into the tumor specific MoA for ETBE (see Tumor Specific MoA in Animal Models). However, due to the lack of available data for exposed humans or human primary cells/tissues the overall strength of evidence for KCC 10 “Alters cell proliferation, cell death, or nutrient supply” is ‘limited.’.

TBA

No data was available for KCC 3; “Alters DNA repair or causes genomic instability,” or KCC 9; “Causes immortalization,” and therefore, the strength of these KCC is designated as ‘Inadequate’ (Fig. 3b).

KCC 1 “Is electrophilic or can be metabolically activated to an electrophile” (Table 2; Supplementary Data, Table S3)

Yuan et al. (2007) measured DNA adduct formation in the lung, liver and kidney of mice exposed to TBA via oral gavage (0, 9.90 × 10^-2 to 9.97 × 10² µg TBA/kg-bw) for one day (Yuan et al., 2007). A dose-dependent increase in DNA-adduct formation was reported in all tissues (Yuan et al., 2007). However, the analytical method used for measuring these adducts was accelerated mass spectroscopy. Bus et al. (2022) identified that without use of a synthetic standard, it is unclear if these results reflect labeled DNA adducts or metabolic incorporation of the ¹⁴C label into DNA through cellular carbon pools generated from the metabolism of ¹⁴C-TBA (e.g., formation of ¹⁴C-formate) (Bus et al., 2022).

One experimental animal study measured DNA adduct formation and was not reliable based on methodology, and lack of data in exposed humans or human primary cells/tissues. No data was available in exposed humans, primary human cells/tissues, human cell lines or animal primary cells/cell lines. Overall, the strength of evidence for KCC 1 “Is electrophilic or can be metabolically activated to an electrophile” is ‘inadequate.’

KCC 2 “Is genotoxic” (Table 2; Supplementary Data, Table S3)

The mutagenic activity of TBA was investigated by Zeiger et al. (1987) as cited by Mahler et al. (1997) in the Ames assay in four of the five OECD 471 TG recommended bacterial strains (Salmonella typhimurium TA100, TA1535, TA1537, TA98) with and without metabolic activation at exposure concentrations ranging from 100 to 10,000 µg/plate (Mahler, 1997). No mutagenic activity was reported in any of the strains evaluated with and without metabolic activation. TBA was negative in S. typhimurium TA102 in the absence or presence of S9 in one assay (McGregor et al., 2005, Williams-Hill et al., 1999), and positive in another assay with metabolic activation (Williams-Hill et al., 1999) with doses ranging from 100–10,000 µg/plate (Table 3a).

Several assays investigating the genotoxicity of TBA were identified as reliable in vitro assay using mammalian cells. McGregor (1988), as cited by Mahler et al. (1997), reported a non-reproducible increase in mutant colonies in the mouse lymphoma cell mutation assay in the presence and absence of S9 at the highest dose tested (5000 μg/mL). The authors concluded that TBA was negative in the assay. Insignificant changes were reported in TBA exposed (160–5,000 µg/mL) Chinese Hamster Ovary cells (CHO) for sister chromatid exchange (SCE) and chromosomal aberrations (CA) (Galloway et al., 1987) as cited by Mahler et al. (1997). The SCE observed effects were non-reproducible, and are no longer accepted as an the OECD Guideline to evaluate genotoxicity (OECD, 2014).The CA observed effects were reported at cytotoxic doses. Therefore, TBA was not considered to induce SCE or CA in vitro. Sgambato et al. (2009) reported an increase in DNA double strand breaks in primary rat fibroblasts (Rat-1) cells exposed to a single high concentration of TBA (0.44 mM) with no significant increase in cytotoxicity (Table 3b).

A significant increase in DNA damage was reported in leukocytes in male C57BL/6 mice exposed to 20 mg/mL TBA via drinking water for 6 weeks (Lin et al., 2020). No significant effect on DNA damage was reported in an OECD TG 489-compliant study in CD-1 mice administered TBA (125–2000 mg/kg-bw/day) via oral gavage for 2 days (Thompson et al., 2024). Additionally, one reliable study reported by NTP (Mahler, 1997) demonstrated no significant increase in micronucleus formation in the bone marrow of F344 rats administered TBA via IP injection 3×/day for 24 h (39–1250) or B6C3F1 mice exposed via drinking water for 13 weeks (3000–40,000 ppm) (Table 3d).

Data in experimental animal models was inconsistent across endpoints in reliable studies, however, results conducted in a comet assay in accordance with OECD Guideline 489 (Thompson et al., 2024) and micronucleus assays by the National Toxicology Program (NTP) were consistently negative (Mahler, 1997, Thompson et al., 2024). These findings combined with overall negative results in animal primary cells/cells lines, including ames assays conducted in accordance with OECD TG 471 (McGregor et al., 2005) lead to the conclusion that TBA is non-genotoxic. The lack of available data in exposed human and primary human cells/tissues lead to an overall strength of evidence for KCC 2 “Is genotoxic” to be ‘limited.’

KCC 5 “Induces oxidative stress” (Table 2; Supplementary Data, Table S3)

Oxidative DNA damage was measured in C57BL/6 mouse liver following exposure to TBA (0, 5, 20 mg/mL) via drinking water for 6 weeks (Lin et al., 2020, Weng et al., 2019) and in vitro in primary rat fibroblasts (Rat-1) with a single exposure concentration of 0.44 mM (Sgambato et al., 2009). A significant increase in oxidative DNA damage was reported in all in vivo and in vitro studies.

Alterations in antioxidant enzymes following TBA exposure were measured in in vitro and in vivo test systems. In a mouse neuronal cell line (HT22), a dose-dependent decrease in superoxide dismutase (SOD) and increased glutathione disulfide (GSSG) and glutathione (GSH) levels were reported at all exposure concentrations (0.1.25–2.0 mM) (Ma et al., 2018). Changes were only considered significant at 0.125 mM as a significant decrease in cell viability was reported starting at 0.25 mM (Ma et al., 2018). In experimental animals, a significant increase in GSH levels in the kidney, but not liver, of male Wistar rats exposed to TBA via drinking water (0.5 % v/v) (Acharya et al., 1995). No significant changes in GSH levels were reported in male Sprague-Dawley rat livers following a single dose of TBA via IP injection (6.44 mmol/kg) (Harris and Anders, 1980). In C57BL/6 mice, increased glutathione oxidation (total GSH, and GSH/GSSG) was reported in serum at the highest TBA drinking water concentration (20 mg/mL) in C57BL/6 (Lin et al., 2020).

Overall, the strength of evidence for KCC 5 “Induces oxidative stress” is ‘limited’ based on data in experimental animals and animal primary cells/cells lines and lack of data in exposed humans or human primary cell/tissues.

KCC 8 “Modulates Receptor-Mediated Effects” (Table 2; Supplementary Data, Table S3)

Changes in steroidogenesis was evaluated in accordance with the EPA test guideline OPPTS 890.1550. No significant changes in estradiol or testosterone production was measured following exposure of a human H295R cell line to (0.0001–100 μM TBA) for 48 h (de Peyster et al., 2014). Study authors also reported that TBA was not an inhibitor of aromatase activity (EPA OPPTS 890.1200) tested up to the highest concentration (0.001–1000 µM) (de Peyster et al., 2014). Moser et al. (1998) reported that TBA did not inhibit 17β-estradiol binding in the human estrogen receptor (ER) at concentrations up to 100 µM (Moser et al., 1998).

In cultured rat Leydig cells exposed to 50 or 100 mM TBA for 3 h, testosterone secretion was significantly decreased at both exposure concentrations compared to controls (de Peyster et al., 2003). However, another study conducted in accordance with EPA test guideline (EPA OPPTS 890.1150) reported that TBA (0.0001–100 μM TBA) did not bind to the androgen receptor (AR) isolated from rat prostate (dePeyster et al., 2014). This data was further supported by EPA’s CompTox Chemicals Dashboard QSAR models which predicted that TBA would not bind to or activate the estrogen (ER) or androgen receptor (AR), using the COMPARA and CERAPP models for AR and ER, respectively (EPA, 2024b).

In B6C3F1 female mice exposed to TBA via drinking water (2 or 20 mg/mL [∼344 and 818 mg/kg-bw/day for the 3 day exposure; ∼418 and 1616 mg/kg-bw/day for the 14 day exposure]) for 3 or 14 days, a significant decrease in triiodothyronine (T3) and thyroxine (T4) levels (but not thyroid stimulating hormone [TSH]) was reported at 20 mg/mL (1,616 mg/kg-bw/day) on day 14 (Blanck et al., 2010). Borghoff et al. (2025b) reported no significant difference in T4 or TSH in B6C3F1 female mice exposed via drinking water at 0, 5, 10, 20 mg/mL for 14 or 28 days, however, a significant decrease in T4 and TSH was reported after 500 or 1000 mg/kg for 5 days via oral gavage (Borghoff et al., 2025b). Blanck et al. (2010) also reported significantly increased hepatic cytochrome P450 content (20 mg/mL, 14 days) and enzyme activity (7-benzoxyresorufin-O-debenzylase (BROD; 2 and 20 mg/mL, 14 days) and 7-pentoxyresorufin-O-dealkylase (PROD, 20 mg/mL, 14 days). No significant changes were observed in 7-ethoxyresorufin-O-deethylase (EROD) activity. Changes in liver enzyme transcripts were also investigated. Authors reported decreased Cyp2b9 (20 mg/mL, 14 days), increased Cypb10 and Sult1a1 (2 and 20 mg/mL at 14 days), and no change in UGT1a1, UGT2b1, or UGT2b5 (Blanck et al., 2010). Increased P450 content, and changes in liver enzyme transcripts was supported by the presence of hypertrophy in the liver of exposed mice (Blanck et al., 2010). Borghoff et al. (2025b) reported upregulation of Phase II transcripts (Sult2a1, Sult2a2, Ugt1a8), with an increase in hepatocellular hypertrophy at 14 and 28-days of exposure, but no change in UGT enzyme activity after 14-days of TBA exposure via drinking water at 0, 5, 10, 20 mg/mL.

Data in experimental animal models were overall minimal/weak, however, data supporting activation of the thyroid receptor, and indicators of CAR/PXR activation determined by induction of phase I and II enzymes were considered to be strong. These data were used for integration into the tumor specific anti-thyroid MoA for TBA (see Tumor Specific MoA in Animal Models). However, due to the lack of available data for exposed humans or human primary cells/tissues the overall strength of evidence for KCC 8 “Modulates receptor-mediated effects” is ‘limited.’

KCC 10 “Alters cell proliferation, cell death or nutrient supply” (Table 2; Supplementary Data, Table S3)

Increased cell proliferation was reported in the proximal tubule of male, but not female, F344 rats at TBA exposure concentrations up to 1750 ppm via inhalation (6 h/day for up to 10 days) (Borghoff et al., 2001). Increased cell proliferation in the proximal tubule of male F344 rats was also reported at concentrations above 1 % (10 mg/mL) TBA in a 90-day drinking water study (0.25, 0.5, 1, 2, 4 % [2.5, 5, 10, 20, 40 mg/mL] TBA) (Lindamood et al., 1992, Takahashi et al., 1993). In a recent publication, there was no increase in cell proliferation in the thyroid of B6C3F1 mice exposed to TBA for 14- and 28- days via drinking water at concentrations up to 40 mg/mL (Borghoff et al., 2025a). Increased incidence of urothelial hyperplasia was reported in the urinary bladder in B6C3F1 mouse and F344 rat exposed to TBA via drinking water (high exposure level in rat: 4% [M or F]; high exposure level in mice: 2 % [M] and 4% [F]) for 13 weeks) (Cirvello et al., 1995, Lindamood et al., 1992, NTP, 1995). In one in vitro study, primary rat fibroblasts (Rat-1) cells were exposed to 0.44 mM for 48 h; there was no increase in cell death, but an increased number of cells were in G0/G1 phase. Cell cycle protein levels (retinoblastoma protein (pRb), cyclin-dependent kinase inhibitor (p27) and cyclin D1were also increased following exposure (Sgambato et al., 2009) (Table 4).

Data in experimental animals was consistent for cell proliferation and was considered as strong supporting data for integration into the tumor specific MoA for ETBE (see Tumor Specific MoA in Animal Models). However, due to the lack of available data for exposed humans or human primary cells/tissues the overall strength of evidence for KCC 10 “Alters cell proliferation, cell death, or nutrient supply” is ‘limited.’

Other mechanistic evidence for ETBE and TBA

High throughput in vitro assay data

High throughput (HT) in vitro assay data for ETBE (up to 100 mM) available through the ToxCast database and the KC-Hits software tool (i.e., EPA ToxCast data in vitrodbv_4.0, iarc-imo / kc-hits · GitLab, v0.6.0) report that within a total of 218 assays tested, all were considered “inactive” (EPA., 2024a, EPA., 2024b). ETBE’s inactivity in these assays is most likely due to its volatility under the conditions of this HT platform (i.e., unsealed wells). As such, these data are not considered to provide any relevant information to this assessment. TBA was not evaluated in HT assays.

Tumor specific MoAs in animal models not relevant to humans

ETBE

The only tumor response in standard chronic rat bioassays was a significant increase (60%) in liver tumors at the highest concentration (5000 ppm) in males only (Saito et al., 2013) (Table 1). The level of statistical significance with which these tumors occurred (p < 0.01) meets the Haseman decision rule (FDA, 2001) for common tumors reducing the likelihood that this finding was a false positive (Haseman, 1983, Haseman, 1984). A proposed non-human relevant MoA for liver tumors has been outlined by Corton et al., 2018, Yamada et al., 2021 and Elcombe et al. (2014), for non-genotoxic chemicals (Table 5) (Corton et al., 2018, Elcombe et al., 2014, Yamada et al., 2021). Mechanistic data organized by KCC were integrated to inform key events including modulates receptor mediated effects (KCC8) and alters cell proliferation, cell death or nutrient supply (KCC10). Induces oxidative stress (KCC5) was integrated as a modulating factor. The key events that are identified for this liver tumor MoA begin with activation of nuclear receptors CAR/PXR or PPARα (KCC8), followed by altered gene expression specific to CAR activation (CAR/PXR) (KCC8), altered cell cycle (PPARα) (KCC8), increased cell proliferation (KCC10), and clonal expansion. Initiation of these events were supported by results presented by Kakehashi et al. (2013) with administration of ETBE 2×/day for 2 weeks via oral gavage (0, 150, 1000 mg/kg bw/day) in F344 rats leading to induced activation of nuclear receptors CAR/PXR, and PPAR (α, γ, δ) followed by gene and protein expression changes in a suite of liver enzymes (Kakehashi et al., 2013). Hepatocellular hypertrophy, an indicator of liver enzyme induction and an associative event by Yamada et al. (2021) (Yamada et al., 2021), was reported in a number of studies following ETBE exposure (Kakehashi et al., 2013, Kakehashi et al., 2015, Medinsky et al., 1999, Miyata et al., 2014, Weng et al., 2012). Lastly, inhibition of cell cycle arrest (Kakehashi et al., 2013) and increased cell proliferation in hepatocytes was reported, supporting mitogenic activity (Kakehashi et al., 2015).

Table 5.

Non-genotoxic MoAs for liver tumors in rodents; Early key event activity identified for ETBE.

Key Events in MoAs for Liver Tumors (Corton et al., 2018, Elcombe et al., 2014, Yamada et al., 2021)	Data to Support a Rodent Specific Liver Tumor MoA
Activation of nuclear receptors (CAR/PXR, PPAR α) (KCC 8)	Increased peroxisome numbers in hepatocytes (indicator of PPARα activation) of rats exposed to ETBE (150, 1000 mg/kg bw/day) along with predicted activation of CAR/PXR, PPARγ, and PPARδ using Ingenuity Pathway Analysis (IPA) (Kakehashi et al., 2013).
Altered gene expression specific to CAR activation (KCC 8) Associative Eventa: Hepatic CYP2B induction	Changes in gene expression of liver enzymes (increased expression of Cyp3a2, Cyp2b1, Cyp2b2, Cyp1a1, Cyp2e1, Cyp4a1, Cyp4a2, PPARα and PPARγ; decreased gene expression of Ugt1a1, Ugt2b5, Sult1d1, Cyp2c6) along with increased protein levels of P450 isoenzymes in liver (CYP2B1, CYP2B2, CYP3A1, CYP3A2, CYP2C6, CYP2E1, CYP1A1 (Kakehashi et al., 2013). Slight increase in P450 content at 150 mg/kg bw/day, significant increase at 1000 mg/kg bw/day (Kakehashi et al., 2013). Increased hypertrophy (associative event) in liver was reported in CD-1 mouse and Sprague-Dawley rat at 5000 ppm and 400 mg/kg bw/day, respectively (Medinsky et al., 1999, Miyata et al., 2014).
Associative Eventa: Hepatocellular hypertrophy (KCC 8)	Increased hepatocellular hypertrophy (an indicator of liver enzyme induction) following exposure to ETBE in rats via oral or inhalation exposure (Medinsky et al., 1999, Weng et al., 2012, Kakehashi et al., 2013, Miyata et al., 2014, Kakehashi et al., 2015).
Modulating factorsb: Production of oxidative stress – Nrf-2 mediated oxidative stress response (KCC 5); increased oxidative DNA damage	Increased 8-OHdG levels and hydroxyl radical generation in liver of Fischer 344 rats exposed to ETBE (Kakehashi et al., 2013, Weng et al., 2012). In rats exposed to 150 or 1000 mg/kg bw/day ETBE, there was a decrease in gene expression of OGGI (Kakehashi et al., 2013). Increased oxidative DNA damage reported as 8-OHdG levels in Kakehashi et al., 2013 (same data as reported above). No significant increase in DNA damage reported in mouse blood using modified comet assay from inhalation exposure to ETBE at concentrations ranging from 500 to 5000 ppm or 50–500 ppm, respectively (Weng et al., 2013, Weng et al., 2014).
Altered Cell cycle Associative eventa: decreased apoptotic cells (KCC 10)	Decreased measures of cell cycle (PCNA and CD1) at 1000 mg/kg bw/day (Kakehashi et al., 2013) apoptotic cells at 1000 mg/kg bw/day (Kakehashi et al., 2013, Kakehashi et al., 2015). These changes do not support the directional change that would be indicative of this key event
Increased cell proliferation (KCC 10)	Increased cell proliferation (via BrDU labeling) in liver of mice exposed to ETBE via inhalation (500, 1750, 5000 ppm) (Medinsky et al., 1999) and administered ETBE by oral gavage (1000 mg/kg bw/day 2×/day) (Kakehashi et al., 2015) supportive of a mitogenic activity (Medinsky et al., 1999; Kakehashi et al., 2015).
Clonal expansion leading to altered foci	No information identified in the literature

^aassociative event: biological processes that are not necessary key events, but are indicators or markers for key events within the MOA.

^bmodulating factor: biological processes that are not necessary to induce an adverse outcome, but may induce one or more key events or the adverse outcome.

Development of rat liver tumors through PPARα or CAR/PXR activation has also been shown to not be relevant in humans due to differences in biological responses in humans versus rodents in downstream events caused by activation of these receptors (Corton et al., 2018, Yang et al., 2008). The recent EPA IRIS report investigated mechanistic evidence supporting the MoA of liver tumors in rats through activation of (PPAR α, CAR or PXR) and concluded that available data were unable to be used for dose–response assessments or temporal associations, and were therefore inadequate to conclude ETBE induces liver tumors via CAR/PXR mediated MoA. The EPA highlighted data gaps including lack of dose-responsive effects, knock-out studies, data beyond 1–2 week timepoints, and evidence being limited to one species (EPA, 2021a). Although these data would provide additional confidence, to evaluate PPAR or CAR/PXR MoA, they are not necessary for establishing a plausible MoA. Regardless of the data gaps in this MoA, there is only evidence for this tumor response in rats chronically exposed to a high concentration of ETBE in one of two cancer studies, thus provided limited evidence of carcinogenic hazard in humans.

TBA

Chronic exposure to TBA causes a low incidence of kidney tumors in male rats and thyroid tumors in female mice (Table 1). The MoA associated with the low incidence of these tumors were investigated in a number of studies to consider their relevance in assessing human risk. The criteria for male rat specific tumors that do not operate in humans were evaluated using criteria identified by EPA (1991) and IARC (1999). TBA was determined to not be genotoxic through evaluation of evidence organized into the KCC ‘Is genotoxic’ (KCC 2), which is the first criteria for both the male kidney tumors and the female mouse thyroid tumors MoA (Table 6a, Table 6b).

Table 6a.

Non-genotoxic MoA for kidney tumors in male rats; Critical key event activity identified for TBA.

Requried Evidence to Fill Key Event Criteria According to IARC (citation monograph 147), 1999 and EPA, 1991*	Studies that Provide Data to Support a Male Rat Specific Kidney Tumor MoA
*Lack of genotoxic activity (agent and or metabolite) based on an overall evaluation of in vitro and in vivo data (KCC2).	TBA is not considered a genotoxic agent. A recent regulatory assessment cited studies available to assess the genotoxic potential of TBA and reported that, in most studies, TBA was negative in genotoxicity assays (US EPA, 2021). Since this assessment, TBA has been shown to be negative in an in vivo Comet study Table 2(Thompson et al., 2024), confirming its lack of genotoxic activity. ETBE and MTBE, are both metabolized to TBA and are negative in transgenic F344 Big Blue rats exposed via inhalation to support TBA is not a mutagen (Gollapudi and Rushton, 2023).
*Male rat specificity for nephropathy and renal tumorgenicity	TBA causes a low incidence of kidney tumors in male, but not female, rats or mice that was attributed to both α2u-globulin nephropathy (male rat specific) and chronic progressive nephropathy (CPN). There is a higher incidence in CPN in control male vs. female rats that is exacerbated to a greater incidence according to severity grade of the lesion in male vs female rats (Hard et al., 2019). As noted in this publication, all non-neoplastic and neoplastic lesions in the kidneys of male rats exposed to TBA can be attributed to α2u-globulin and CPN; neither MoA is relevant to humans (Hard et al., 2019)
*Induction of the characteristic sequence of histopathological changes in shorter-term studies, of which protein droplet accumulation is obligatory	Several studies, either by inhalation or via drinking water exposure, demonstrate that exposure to TBA results in protein droplet accumulation along with histopathological lesions associated with the progression of α2u-globulin nephropathy such granular casts, karyomegaly and with longer exposure, linear papillary mineralization, foci of tubular hyperplasia (Takahashi et al., 1993, Lindamood et al., 1992, NTP, 1995, Cirvello et al., 1995, Borghoff et al., 2001, Hard et al., 2011; Hard et al., 2019)
*Identification of the protein accumulating in tubule cells as α 2u-globulin	TBA results in the accumulation of protein droplets in male rat kidneys that stained for α2u-globulin (Lindamood et al., 1992; Takahashi et al., 1993; Borghoff et al., 2001).
Reversible binding of the chemical or metabolite to α2u-globulin	TBA has been shown to bind reversibly to α2u-globulin (Williams and Borghoff, 2001).
Induction of sustained increased cell proliferation in the renal cortex (KCC 10)	Increased cell proliferation in male, but not female rats with exposure to TBA via inhalation or drinking water has been reported (Lindamood et al., 1992; Takahashi et al., 1993; Borghoff et al., 2001; Hard et al., 2019).
Similarities in dose–response relationship of the tumor outcome with the histological endpoints (exacerbation of protein droplets, α2u-globulin accumulation, cell proliferation).	The protein droplets in male rats exposed to TBA stained immunohistochemically for α2u-globulin with a statistically significant increase in the concentration of α2u-globulin measured in male rats when compared with unexposed males. A strong positive correlation (r = 0.994) was demonstrated between an increase in labeling index of renal cortical cells (indicating cell proliferation) and mean α2u-globulin concentration (Borghoff et al., 2001).

Table 6b.

Non-genotoxic anti-thyroid MoA for thyroid tumors in mice; Critical key event activity identified for TBA.

Required Evidence to Fill Key Event CriteriaAccording to IARC ( monograph 147), 1999 and (EPA, 1998a, IARC., 1999)*	Studies that Provide Data to Support a Male Mouse Specific Thyroid Tumor MoA
*Lack of genotoxic activity (agent and or metabolite) based on an overall evaluation of in vitro and in vivo data (KCC2).	TBA is not considered a genotoxic agent. A recent regulatory assessment cited studies available to assess the genotoxic potential of TBA and reported that, in most studies TBA was negative in genotoxicity assays (EPA, 2021b). Since this assessment, TBA has been shown to be negative in an in vivo Comet study (Thompson et al., 2024) , confirming its lack of genotoxic activity. ETBE and MTBE, are both metabolized to TBA and are negative in transgenic F344 Big Blue rats exposed via inhalation to support TBA is not a mutagen (Gollapudi and Rushton, 2024; Gollapudi and Rushton, 2023).
Increases in cellular growth; Thyroid weights and histology (KCC 10)	No changes in thyroid weights or histology or rate of cell proliferation with up to 28-days of exposure to TBA to female mice (Borghoff et al., 2025b). Histological observation of increased hyperplasia in the thyroid of male and female mice exposed to TBA for 2 years (NTP, 1995).
Change in thyroid & relevant pituitary hormones (KCC 8)	Blanck et al. (2010) demonstrated a decrease in plasma T3 and T4, but no change in TSH in female B6C3F1 mice exposed to TBA for 14-days. In a recent study reported by Borghoff et al. (2025b), exposure to TBA for 14 or 28 days did not result in any changes in plasma T4 or TSH in exposed female mice. As summarized by McGregor (2010), in female CD-1 mice exposed to 3000 and 8000 ppm MTBE, in which TBA is a primary metabolite, a decrease in total T4 after 31 days was reported (Burleigh-Flayer et al., 1992).
Sites of antithyroid action; Demonstration of induction of phase II enzymes. Measures include liver weight, liver histopathology, microsomal enzyme activity (i.e., UDPGT and SULT).	Liver is a site of antithyroid action beginning with the induction in metabolizing enzymes. TBA did not increase liver weight at 3 or 14 days, but induced centrilobular hepatocellular hypertrophy, an indicator of induction of metabolism noted by Blanck et al., 2010 and more recently by Borghoff et al., 2025b). TBA increased total CYP content after 14 days, with a slight increase in PROD and BROD activity, with up-regulation of transcripts of SULT, but not UGT (Blanck et al., 2010). Borghoff et al. (2025b) reported an increase in SULT and UGT transcripts but no change in UDPGT activity with exposure to TBA. ETBE is metabolized to TBA and has been shown to induce CAR/PXR in mouse liver (Kakehashi et al., 2013).
Dose correlation; A chemical does or does not jointly perturb thyroid-pituitary hormone levels and produce various histological changes in thyroid and thyroid tumors.	No evidence published at this time that TBA demonstrates a dose correlation between perturbation of thyroid-pituitary hormone levels and histological changes in thyroid and thyroid tumors likely due to the mild responses observed at early timepoints.
Reversibility; Thyroid hormone changes, histopathology/cell proliferation recovery to control levels following cessation of exposure.	No treatment-related changes were observed in the thyroid following up to 28 days of TBA exposure; therefore, reversibility could not be assessed. However, liver weight increases and mild hepatocellular hypertrophy observed after up to 28 days of exposure were resolved following a 28-day recovery period without TBA exposure (Borghoff et al., 2025b).

In an analysis by an independent Pathology Working Group, male rat kidney tumors were found to be attributed to both α2u-globulin nephropathy and CPN exacerbation (Hard et al., 2011, 2019), both MoAs identified as contributing to the development of a low incidence of these male rat specific kidney tumors. As outlined in Table 6a, TBA fulfills the key event criteria identified by both IARC and EPA (IARC, 1999; EPA, 1991) for a non-genotoxic agent that produces male rat specific kidney tumor response through its induction of α2u-globulin nephropathy. At the time these key event criteria were published, the role of CPN was noted as a common spontaneous lesion in aging rats that appeared to have a role in contributing to the kidney tumor response in male rats, but since then much effort in characterization of these lesions by Hard and colleagues has been published (Hard et al., 2009, Hard et al., 2011, Hard et al., 2019), demonstrating it as a separate MoA that is not relevant in humans. Across several studies, exposure to TBA via inhalation or drinking water resulted in microscopic kidney lesions characterized by epithelial-cell necrosis, protein droplet accumulation, granular casts, and karyomegaly within the proximal tubules along with linear papillary mineralization and foci of tubular hyperplasia (in studies greater than 90 days) (Takahashi et al., 1993, Lindamood et al., 1992, NTP, 1995, Cirvello et al., 1995, Borghoff et al., 2001, Hard et al., 2011; 2019) (Table 6a).

In a 10-day TBA inhalation study, the protein droplets formed in the kidneys of male rats exposed to TBA were shown to stain immunohistochemically for α2u-globulin with a statistically significant increase in the concentration of this protein compared with unexposed male rats. TBA has also been shown to bind reversibly to α2u-globulin (Williams and Borghoff, 2001), a key event resulting in its increase in male rat kidneys. A strong positive correlation (r = 0.994) was demonstrated between an increase in the mean α2u-globulin concentration and the labeling index of renal cortical cells (indicating tubular cell proliferation; KCC 10) (Borghoff et al., 2001). These findings indicate that TBA, a mild inducer of α2u-globulin nephropathy induces renal cell proliferation in male, but not female, rats (Borghoff et al., 2001, Lindamood et al., 1992, Takahashi et al., 1993). CPN, a common lesion in aging rats with no human relevance which can be exacerbated by chemical exposure, is considered to play a role along with α2u-globulin nephropathy in the development of these tumors, which operates through an MoA not relevant to humans (Hard et al., 2009, 2011, 2019; Souza et al., 2018).

Chronic exposure to TBA also was shown to result in thyroid follicular cell hyperplasia and a low incidence of thyroid adenomas in female mice at the highest concentration tested (Cirvello et al., 1995, NTP, 1995). Based on the absence of genotoxic potential, TBA most likely operates through an anti-thyroid MoA as outlined in Table 6b. The key event criteria for an anti-thyroid MoA, one that does not operate in humans, is based on the criteria identified by EPA, 1998a, EPA, 1998b and IARC (1999). Besides the lack of genotoxicity (KCC 2), mechanistic data identied informed key event criteria within the anti-thyroid MoA including modulates receptor mediated effects (KCC 8) and alters cell proliferation, cell death or nutrient supply (KCC 10).

Changes in thyroid hormones (KCC 8) have been reported in several studies. Blanck et al. (2010) observed decreased T3 and T4 with no change in TSH in female B6C3F1 mice after 14 days of TBA exposure, along with upregulation of SULT but not UGT transcripts, suggesting induction of phase II metabolism (KCC 8). In contrast, Borghoff et al. (2025b) found no changes in T4 or TSH following 14- or 28-day exposure to TBA in drinking water. Although phase II enzyme transcripts (Sult2a1, Sult2a2, Ugt1a8) were upregulated in this study, UGT enzyme activity was not increased after 14 days. Liver changes, including mild hypertrophy and increased weight, were reversible after 28 days without exposure. While the thyroid tumor response in TBA-exposed mice is weak, the combined evidence from Blanck et al. (2010) and Borghoff et al. (2025b) supports the plausibility of a weak anti-thyroid mode of action in female mice.

Evidence integration

ETBE

There were no reliable human data available to evaluate the potential carcinogenic activity of ETBE. Two standard 2-year cancer bioassays in rats (inhalation and drinking water) were considered. In the drinking water study, no increase in tumor incidence was observed. In the inhalation study, an increased incidence of liver tumors was reported in male rats at the highest exposure concentration; however, this response was confounded by markedly reduced survival and a significant reduction in body weight (>20 %) at that dose, raising concern that the effect may have resulted from exceedance of the maximum tolerated dose (MTD). In the drinking water study, no treatment-related tumor responses were observed, although the highest administered concentration was substantially lower than the top dose achieved in the inhalation study (542–560 mg/kg/day vs. ∼1350–1435 mg/kg/day). Taken together, the findings from two reliable cancer bioassays, the absence of genotoxic activity, and a proposed MoA not considered relevant to humans, combined with limited mechanistic activity across most KCC, support the conclusion that ETBE is unlikely to pose a carcinogenic hazard to humans (Fig. 4).

Fig. 4.

Integration of the totality of data that inform the potential carcinogenic hazard of ETBE and TBA in humans.

TBA

There are no reliable human data available to evaluate the carcinogenic potential of TBA. Two reliable, standard 2-year cancer bioassays conducted in rats and mice reported a low incidence of kidney tumors in male rats and thyroid tumors in female mice following exposure to TBA via drinking water. Integration of the findings together with the lack of TBA genotoxic activity, supports the conclusion that the observed tumors arose through rodent-specific MoA. Specifically, the kidney tumors in male rats are consistent with α2u-globulin–mediated nephropathy and CPN, MoAs that do not operate in humans, while the thyroid tumors in female mice are likely associated with an anti-thyroid MoA, also not relevant to humans. Moreover, limited activity across mechanistic endpoints representing seven KCC further supports the conclusion that TBA is unlikely to pose a carcinogenic hazard to humans (Fig. 4).

Discussion

This systematic evaluation of the evidence for the carcinogenic potential of ETBE and TBA in humans included two reliable, standard cancer bioassays for each chemical conducted in experimental animal models, along with mechanistic data mapped across the KCC and evaluated for support of key events within rodent-specific tumor MoAs. No human cancer data were identified for either chemical. Based on the totality of evidence, ETBE and TBA are not considered to pose a carcinogenic hazard to humans (Fig. 4).

Chronic inhalation exposure to ETBE resulted in an increased incidence of hepatocellular adenomas, with a significant increase in combined hepatocellular adenomas and carcinomas observed in male (but not female) F344 rats at the highest exposure concentration (5000 ppm; ∼1350–1435 mg/kg-bw/day) (Saito et al., 2013). A 13-week preliminary study identified this concentration as the maximum tolerated dose (MTD) based on inhibited body weight gain, altered liver and kidney weights, increased proximal renal tubule regeneration, and changes in hematological and biochemical parameters. Similar findings, including a 25% reduction in body weight gain and increased kidney weight, were observed at this concentration in the chronic study, suggesting that liver tumor development was likely driven by systemic toxicity and exceedance of the MTD (Saito et al., 2013). In contrast, lower ETBE doses administered via drinking water (542–560 mg/kg-bw/day) over two years did not increase liver or other tumor incidences in F344 rats of either sex (Suzuki et al., 2012). These differences in tumor response likely reflect route-dependent differences in systemic dose and/or MTD exceedance in the inhalation study (Saito et al., 2013).

A non-standard bioassay by Maltoni et al. (1999) in Sprague-Dawley rats reported multiple neoplastic lesions (oral epithelium, forestomach, uterus, and hemolymphoreticular system). However, this study was deemed Not Reliable based on methodological and reporting deficiencies, including poor documentation, animals held until natural death, lack of pathogen-free verification, and the likely presence of Mycoplasma pulmonis, a known cause of chronic inflammation (Schoeb et al., 2009). Non-standard histopathological classifications and high mortality prompted re-evaluation by the U.S. EPA and NTP (EPA., 2021a, Malarkey and Buchner, 2011), who concluded that respiratory infections confounded interpretation of the reported hemolymphoreticular tumors. The reliability of studies from the Ramazzini Institute, including this one, have been widely questioned (Caldwell et al., 2008; Gift et al., 2013; McGregor, 2007). McGregor (2007) highlighted inconsistencies between neoplastic findings and the absence of associated non-neoplastic lesions, concluding that “…this study does not provide an adequate basis for a thorough evaluation.” More recently, leukemias and lymphomas reported by Maltoni et al. (1999) were excluded from the U.S. EPA IRIS assessment due to diagnostic discrepancies (EPA., 2021a, EPA., 2024c). Malignant schwannomas were observed only at the lowest dose and lacked a dose–response relationship, reinforcing concerns regarding biological significance. Accordingly, the only reliable tumor finding was the increased incidence of liver tumors in male F344 rats exposed to ETBE (Saito et al., 2013).

In evaluating animal cancer studies and mechanistic data, it is important to note that ETBE is metabolized to TBA and acetaldehyde, a presumed genotoxic substance. Both ETBE and TBA are considered non-genotoxic based on consistent in vitro and in vivo evidence. This conclusion is supported by several comprehensive reviews (McGregor, 2007, McGregor, 2010) and the U.S. EPA IRIS assessments (EPA, 2021a,b). More recent studies further dispel concern regarding genotoxic potential, as shown by the negative in vivo comet assay for TBA (Thompson et al., 2024) and the absence of mutagenic activity for ETBE in Big Blue F344 rats exposed by inhalation to 5000 ppm—the concentration associated with liver tumors in rats (Gollapudi and Rushton, 2023).

Mechanistic data for both ETBE and TBA were predominantly derived from animal in vivo studies, with no data available from exposed humans and limited data from human primary cells or tissues. The most robust mechanistic data for both chemicals addressed KCC 2 (genotoxicity), KCC 8 (receptor-mediated effects), and KCC 10 (altered cell proliferation, cell death, or nutrient supply). Organizing mechanistic evidence across the KCC enabled evaluation of evidence strength and integration into proposed MoAs for the tumors identified in ETBE and TBA animal models. Given limited or inadequate data across the remaining KCC, it was not feasible to map these data to key events within human-relevant carcinogenic pathways.

IARC recently evaluated ETBE and concluded that there was sufficient evidence of carcinogenicity in experimental animals, citing liver tumors and malignant schwannomas in rats, along with mechanistic data indicating increased cell proliferation (KCC 10) (Turner et al., 2025). However, based on the lack of reliability of the Maltoni et al. (1999) study, malignant schwannomas were not considered in this evaluation. Furthermore, liver tumors in ETBE-exposed rats occurred only at concentrations exceeding the MTD and were associated with activation of the PPARα and CAR/PXR pathways (Table 5)—key events in a liver tumorigenesis MoA not relevant to humans (Corton et al., 2018, Elcombe et al., 2014, Yamada et al., 2021). Consequently, these findings are unlikely to translate to a carcinogenic hazard for humans.

In alignment with IARC’s classification of TBA (Turner et al., 2025), which identified a low incidence of kidney tumors in male rats, but did not include thyroid tumors in female mice, the overall conclusion of this assessment is consistent—TBA does not pose a carcinogenic hazard to humans. Both the kidney and thyroid tumor responses observed in TBA bioassays were evaluated with respect to their underlying MoAs and human relevance. TBA exposure in male rats has been shown to result in reversible binding to α2u-globulin (Williams and Borghoff, 2001), leading to α2u-globulin nephropathy characterized by the accumulation of α2u-globulin as protein droplets, granular casts, linear papillary mineralization, and focal tubular alterations (Takahashi et al., 1993, Lindamood et al., 1992, Cirvello et al., 1995, NTP, 1995, Borghoff et al., 2001, Hard et al., 2011, 2019). These data fulfill the established key event criteria for a male rat–specific α2u-globulin-mediated MoA for non-genotoxic agents, as identified by both IARC and the U.S. Environmental Protection Agency (EPA, 1991; IARC, 1999). At the time these criteria were developed, chronic progressive nephropathy (CPN) was recognized as a spontaneous lesion in aging rats that could contribute to the kidney tumor response. However, subsequent detailed work by Hard and colleagues (2009; 2011; 2019) demonstrated that CPN represents a distinct MoA separate from α2u-globulin nephropathy, and both are considered not relevant to humans. Overall, as summarized in Table 6a, the criteria for this MoA are fulfilled based on mechanistic data mapped to KCC 2 (genotoxicity) and KCC 10 (altered cell proliferation, cell death, or nutrient supply), as well as other supporting evidence for chemical binding to α2u-globulin and characteristics and involvement of CPN in the development of these kidney tumors (Hard et al., 2019). Consequently, the kidney tumors in male rats exposed to TBA arise through MoAs that do not operate in humans.

Chronic exposure to TBA resulted in thyroid follicular cell hyperplasia and adenomas in female mice at the highest concentration tested (20 mg/mL; ∼2100 mg/kg-bw/day) (Cirvello et al., 1995, NTP, 1995). Although rats are generally more sensitive than mice to thyroid hormone disruption leading to follicular cell tumors (EPA, 1998b; Roques et al., 2013, Qatanani et al., 2005), no thyroid histopathological changes were observed in TBA-exposed rats. The ability of TBA to induce thyroid tumors through an anti-thyroid MoA—which is not considered relevant to humans—was evaluated in this assessment. Key events in this MoA include decreases in circulating thyroid hormones (T3 and T4) and compensatory increases in thyroid-stimulating hormone (TSH), which drive thyroid follicular cell proliferation and tumor development in rodents with chronic exposure (EPA, 1998a, IARC., 1999). This hypothyroid pathway involving elevated TSH is not a risk factor for thyroid follicular tumors in humans. Several studies examined key events in this anti-thyroid MoA in mice exposed to TBA and found similar patterns observed in early key events involving phase II enzyme induction, reflected by changes in liver weight and histopathology (Blanck et al., 2010, Borghoff et al., 2025b). These changes together support mild activation of this pathway with chronic exposure to TBA. Overall, the evidence supports that these thyroid tumors represent a weak response occurring through an anti-thyroid MoA not relevant to humans.

Conclusion

Reliable standard animal cancer bioassays, together with mechanistic endpoints addressing key events relevant to rodent tumor-specific MoAs, were evaluated to assess the carcinogenic potential of ETBE and TBA. A low incidence of liver tumors was observed in one standard rat ETBE study, while kidney tumors in male rats and thyroid tumors in female mice were reported in two standard TBA studies. Both chemicals were identified as non-genotoxic, and the overall strength of mechanistic evidence across the KCC was limited or inadequate, constraining the ability to link observed effects to key events in biological pathways relevant to human cancer. Based on these findings, and the conclusion that the rodent tumor MoAs are not operative in humans, a biologically plausible mechanism by which ETBE or TBA could induce carcinogenic activity in humans is not evident.

Funding/Disclosures

ToxStrategies received consulting fees from LyondellBasell, Sustainable Fuels, Asian Clean Fuels Association, and Associacion de Combustibles Eficientes de Latinoamerica for this systematic review and preparation of the manuscript. Authors are employed by ToxStrategies or Exponent, both consulting firms that provide services to private and public organizations on toxicology and risk assessment issues, along with one author from LyondellBasell who provided oversight on the use and exposure of ETBE and TBA, published literature and discussions concerning any unpublished research efforts supported by LyondellBasell. The work reported in this article was conducted during the normal course of employment, with no authors receiving personal fees. The literature selection, methodological protocol development, analyses, interpretation of research findings, and manuscript writing, formatting, and submission were conducted solely by the authors, and the conclusions and professional judgements were not subject to the funders’ control beyond the authors.

CRediT authorship contribution statement

Brianna N. Rivera: Methodology, Formal analysis, Investigation, Writing – original draft, Writing – review & editing. Isabel A. Lea: Formal analysis, Writing – original draft, Writing – review & editing. Seneca Fitch: Methodology, Resources, Software. Neepa Choksi: Formal analysis, Investigation. Allison Franzen: Formal analysis, Methodology. James Bus: Investigation, Methodology. Erik Rushton: Resources, Supervision. Susan J. Borghoff: Conceptualization, Formal analysis, Investigation, Funding acquisition, Writing – original draft, Writing – review & editing, Supervision.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: ToxStrategies LLC reports financial support and article publishing charges were provided by LyondellBasell, Sustainable Fuels, Asian Clean Fuels Association, and Associacion de Combustibles Eficientes de Latinoamerica. Exponent Inc reports financial support was provided by LyondellBasell, Sustainable Fuels, Asian Clean Fuels Association, and Associacion de Combustibles Eficientes de Latinoamerica. Erik Rushton reports a relationship with LyondellBasell Industries Inc that includes: employment. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article: