Issues in Assessing the Health Risks of n-Butanol

Abstract

A literature review and health effects evaluation were conducted for n-butanol, a chemical that occurs naturally in some foods, is an intermediate in the production of butyl esters, and can be used as a gasoline additive or blend. Studies evaluating n-butyl acetate were included in the review as n-butyl acetate is rapidly converted to n-butanol following multiple routes of exposure. The primary n-butanol health effects identified were developmental and nervous system endpoints. In conducting the literature review and evaluating study findings, the following observations were made: (1) developmental findings (e.g., embryo/fetal loss, altered growth, or non-neurological structural alterations) were consistently identified; (2) neurodevelopmental findings (e.g., dilation of brain ventricles and subarachnoid space) were inconsistent; (3) evidence for nervous system effects was weak, and there were study quality issues; (4) comparing internal doses from oral and inhalation exposures using physiologically based pharmacokinetic (PBPK) models involves uncertainties due to the lack of oral pharmacokinetic data and a limited understanding of the toxicological mechanisms; and (5) a lack of mechanistic information for n-butanol resulted in the reliance on mechanistic data for ethanol, which may or may not be applicable to n-butanol. This paper presents the literature review findings on the health effects of n-butanol and proposes additional research that may help to reduce the uncertainty that exists due to database limitations.

Graphical Abstract

INTRODUCTION

A literature review was conducted to evaluate the potential human health hazards associated with n-butanol following oral or inhalation exposure. n-Butanol is one of the biobutanol isomers that can be used as a gasoline additive or blend included in the mandate of the Energy Independence and Security Act (EISA Act, 2007; U.S. DOE, 2014). Studies examining the effects of n-butyl acetate were also considered in this paper because n-butyl acetate has been demonstrated to rapidly convert to n-butanol in a 1:1 ratio following exposure via multiple routes (Teeguarden et al., 2005). Thus, test animals exposed to either n-butyl acetate or n-butanol are expected to receive nearly equivalent internal doses of n-butanol, and n-butyl acetate studies could contribute to the understanding of potential human health effects associated with n-butanol exposure. Following the literature review, key toxicity endpoints were identified and evaluated. An EPA-modified version of the physiologically-based pharmacokinetic (PBPK) model (Teeguarden et al., 2005) for n-butanol was used to compare internal doses from selected studies. This paper summarizes the results of the literature review and several key issues that became apparent in the evaluation of study findings on the human health hazard of n-butanol exposure.

n-Butanol is a high-production-volume chemical, and in 2012, had a domestic production of 2,028 million lb/year (U.S. EPA, 2012). Also referred to as n-butyl alcohol, butyl alcohol, 1-butyl alcohol, 1-butanol, and butyric alcohol, n-butanol is a four-carbon, straight chain alcohol that is a highly flammable clear liquid with a strong odor. Besides commercial production, n-butanol occurs naturally and is a volatile component of apple and pear aroma and grape essence, certain cheeses, roasted filberts, fried bacon, peppermint oil, tea, and human milk. For concentrations of both n-butanol and n-butyl acetate in common foods and beverages, see Table 1. Other natural sources include animal wastes, microbes, and insects (National Library of Medicine, 2014; U.S. EPA, 1994a).

Table. 1.

Concentrations of N-Butanol and N-Butyl Acetate in Foods and Beverages

Foods	Concentration	Reference
n-butanol
Apples (volatile emissions)	8.6–35.1 pg/kg-houra	NLM 2015
Baked goods (as flavoring agent)	32 mg/kg (maximum)	WHO 1987
Beef (raw), Frankfurter sausages	Identified/not quantified	NLM 2015
Beverages (as flavoring agent)	12 mg/L (maximum)	WHO 1987
Candy (as flavoring agent)	34 mg/kg (maximum)	WHO 1987
Cordials (as flavoring agent)	1 mg/L (maximum)	WHO 1987
Cream (as flavoring agent)	4 mg/kg (maximum)	WHO 1987
Ice cream and ices (as flavoring agent)	7 mg/kg (maximum)	WHO 1987
Mussel (Mytilus edulis-Japan)	0.27 ppm	NLM 2015
Nuts [roasted filberts; almonds (volatile emissions)]	Identified/not quantified	NLM 2015
Soybeans	199.9–1,586.2 μg/kg	NLM 2015
n-butyl acetate
Apples	29.5 mg/kg (maximum)	WHO 2005
Apples (Golden Delicious)	81 μg/6 L head space (14 days after harvest)	NLM 2012
Apple juice	2.2 mg/kg (maximum)	WHO 2005
Apricots, plums, nectarines	Identified/not quantified	WHO 2005
Beer	0.2 mg/kg (maximum)	WHO 2005
Brandy	0.4 mg/kg (maximum)	WHO 2005
Cider	1.3 mg/kg (maximum)	WHO 2005
Coffee, cocoa, black tea	Identified/not quantified	WHO 2005
Crabs (legs, body, carapace)	0.9, 0.8, 1,5 μg/kg, respectively	NLM 2012
Grapes. mangoes, melons, strawberries	0.1 mg/kg (maximum)	WHO 2005
Honey	Identified/not quantified	WHO 2005
Melon (Queen Anne’s Pocket)	124 μg/kg (skin); 111 (pulp) μg/kg	NLM 2012
Milk, cheese	Identified/not quantified	WHO 2005
Nuts (roasted)	Identified/not quantified	WHO 2005
Potatoes (baked; volatile flavor component)	Identified/not quantified	NLM 2012
Vinegar	166 mg/kg (maximum)	WHO 2005
Wine	Identified/not quantified	WHO 2005

The major use of n-butanol is as an industrial intermediate in the production of butyl esters. It is also used to manufacture dibutyl phthalate, various polymers and other butyl compounds (OECD, 2001), and as a solvent or diluent (National Library of Medicine, 2014). It is used as an intermediate in pharmaceuticals and as an extractant in manufacturing antibiotics, vitamins, and hormones. Therefore, workers are exposed to n-butanol in industrial settings, and the general population is exposed through diet and other consumer products. In addition, as mentioned earlier, n-butanol can be used as an additive or a gasoline blend, so in this context, people can be exposed through fumes. Although exposures to n-butanol may occur through various pathways, the focus of this paper is on hazard assessment. Thus, exposures and associated risk have not been characterized here.

Several organizations have developed reference or safety values for exposure to n-butanol and n-butyl acetate. In 1991, EPA posted an IRIS assessment for n-butanol (U.S. EPA, 1991), which derived a reference dose (maximum acceptable oral daily dose if exposed for a lifetime) of 1 × 10⁻¹ mg/kg/day. A reference concentration (maximum acceptable inhalation daily concentration if exposed for a lifetime) was not derived by EPA. However, the Occupational Safety and Health Association (OSHA) has set a workplace Permissible Exposure Level (PEL) of 100 ppm (maximum time-weighted average exposure for 8 hours) for inhalation of n-butanol, and the American Conference of Governmental Industrial Hygienists (ACGIH) has established a Threshold Limit Value (TLV) of 20 ppm (maximum time-weighted average for 8 hours if exposed for a working lifetime) (National Library of Medicine, 2014). For n-butyl acetate, OSHA’s PEL is 150 ppm, and the ACGIH TLV is also 150 ppm (National Library of Medicine, 2015).

METHODS

Literature Search

A literature search was conducted to identify studies characterizing the health effects of n-butanol. The literature search for n-butanol utilized five online scientific databases: PubMed, Toxline, Web of Science, ChemID, and the Toxic Substances Control Act Test Submissions (TSCATS) database through April 2014 and was updated in April 2016 (see supplement A for complete list of search terms). The keywords included in the search were n-butanol, n-butyl acetate (a metabolic precursor to n-butanol), as well as the metabolites of n-butanol, namely n-butyric aldehyde and butyric acid. The overall literature search approach is shown graphically in Figure 1. A total of 5,511 references were identified from the literature searches with 268 added from the review of online regulatory sources bringing the combined dataset total to 5,779. A manual review of titles and abstracts was conducted to screen studies for relevance (see Figure 1).

Figure 1.

Literature search strategy for n-butanol

Screening Strategy

Selection of studies for consideration was based on the information content and relevance, in addition to general study quality considerations. The exclusion criteria shown in Figure 1 were applied, and 5,219 studies were removed from the combined dataset of 5,779. Studies were excluded for the following reasons: 3,145 were not chemical-specific; 791 were cases in which the chemical was used in sample preparation; 196 were cases in which the chemical was used in manufacturing; 35 were cases in which the chemical was used in measurement methods; 9 were cases in which the chemical was used in treatment, disposal or remediation, and 11 were abstracts only. Another 1,032 studies were excluded for “miscellaneous” reasons, the majority of which focused on butyrate supplementation in domestic animals, butanol-producing species, or anti-tumor and other therapeutic applications.

Eighty foreign language studies were also identified. Based on titles and abstracts, 78 studies were not considered further and were among the 1,032 placed in the miscellaneous category. However, two foreign language studies appeared to be relevant so they were translated into English in their entirety; however, both lacked study details and had small sample sizes, ultimately precluding them from inclusion. Among the remaining 560 papers that were not excluded, further criteria, as described in Table 2, were applied to identify the papers that were the primary sources of health effects data (62 publications). The remaining papers (498 studies) represented either supplementary studies or secondary sources of health effects information (see Figure 1).

Table. 2.

Inclusion/Exclusion Criteria for Study Selection

Inclusion Criteria

Exposure is to n-butanol, n-butyl acetate (or n-butanol metabolites: n-butyric aldehyde and butyric acid)

Exposure is measured in environmental/biological media or tissues

Study includes a measure of one or more primary health effect endpoints

Study includes a measure of one or more secondary health effect endpoints (e.g., genotoxicity, oxidative stress, inflammation, etc) evaluating cellular, biochemical, or molecular effects relevant to mode of action

Exclusion Criteria

References not pertinent to evaluating primary sources of potential health effects (e.g., site-specific risk assessments, chemical analytical method studies, review articles, editorials, and environmental fate and transport studies)

References not evaluating n-butanol, n-butyl acetate (or n-butanol metabolites: n-butyric aldehyde and butyric acid)

References using or describing the chemical as part of sample preparation or assays or manufacturing and use

References evaluating measurement methods

References describing chemical treatment, disposal, or remediation

References evaluating butyrate supplementation in domestic animals

References evaluating anti-tumor and other therapeutic applications of butyrate

References evaluating chemical mixtures

References available only as an abstract

Many of the 62 studies that were identified as primary sources of health effects data evaluated developmental effects and nervous system effects. These studies are discussed further in this manuscript. Several of the epidemiologic studies that were identified included exposures to multiple chemicals and thus were not considered useful for assessing the health risk of n-butanol. Other toxicological effects that were evaluated include hematological, hepatic, renal and dermal effects; however, the evidence for these effects is inconclusive, primarily due to conflicting findings or lack of data. Therefore, these effects will not be discussed further in this manuscript.

There are several studies that investigated eye irritation following chronic exposure to n-butanol in the workplace, and these studies consistently have reported adverse effects. For example, two studies (Tabershaw et al., 1944; Cogan and Cogan, 1945) reported increased eye irritation in workers exposed to n-butanol concentrations ranging from 15 to 115 ppm; however, in another occupational study (Sterner et al., 1949), eye irritation was only reported with exposures exceeding 100 ppm. Although recognized as a hazard of n-butanol exposure, eye irritation will not be discussed further in this paper because, to the best of our knowledge, there are no recent non-acute exposure studies on this outcome.

Some of the inhalation exposure studies identified in the literature search reported exposures in parts per million (ppm). These exposures were converted to mg/m³ using a molecular weight of 74.12 g/mol for n-butanol and 116.16 g/mol for n-butyl acetate, assuming a temperature of 25 degrees Celsius and a pressure of 1 atmosphere.

RESULTS

Following the screening of the 62 primary health effects studies identified in the literature search, eleven studies were identified that could be used to assess n-butanol developmental and neurodevelopmental toxicity and nervous system effects. These studies, however, presented some inconsistent findings, which highlighted several limitations in the knowledge base for n-butanol. The overall findings and the issues that became apparent are presented in the following sections.

Developmental Effects

The n-butanol developmental toxicity study findings were fairly consistent. The n-butanol and n-butyl acetate systematic literature search identified several studies designed to assess the primary manifestations of developmental toxicity: survival, growth, structural alterations, and functional outcomes (U.S. EPA, 1991). The relevant studies included prenatal developmental toxicity studies with both n-butanol (Ema et al., 2005; Sitarek et al., 1994; Nelson et al., 1989a,b) and n-butyl acetate (Saillenfait et al., 2007; Hackett et al., 1982). All of the studies were conducted in rats (Wistar or Sprague-Dawley); one of the studies also assessed New Zealand White (NZW) rabbits (Hackett et al., 1982). The studies were designed to administer control or test material to pregnant females from the time of conception or implantation until the end of the gestation period. Near-term gravid uteri were extracted by hysterolaparotomy. The uteri were examined for evidence of embryo or fetal death (e.g., early or late resorptions, empty implantation sites, or dead fetuses). Viable fetuses were weighed, examined for external abnormalities, and processed for the assessment of visceral or skeletal abnormalities using standardized methodologies.

Each study assessed a control and a range of spaced dose levels (except for Hackett et al., 1982, which only used one treatment level) and the oral studies approached or exceeded limit doses (i.e., 1000 mg/kg-d) recommended by EPA and OECD guidelines for developmental toxicity testing (U.S. EPA, 1998a; OECD, 2001). There was a remarkable degree of consistency across the study results. In general, maternal toxicity did not appear to affect the interpretation of fetal findings. Increased fetal mortality was not observed in any of the studies. Evidence of treatment-related effects on fetal growth, e.g., findings such as decreased fetal body weight, decreased crown-rump length, or delays in skeletal ossification, were observed in all of the rat studies (Ema et al., 2005; Sitarek et al., 1994; Nelson et al., 1989a; Saillenfait et al., 2007; Hackett et al., 1982), although not in the rabbit study (Hackett et al., 1982).

Increased incidences of structural alterations (i.e., visceral or skeletal variations or malformations) were noted in nearly all of the developmental toxicity studies, the exception being the rat study with n-butyl acetate (Hackett et al., 1982). Although the exact same findings were not replicated across studies, there is no expectation of complete concordance. Rather, evidence of structural changes in multiple studies can be interpreted to indicate a general propensity for developmental disruption.

Neurodevelopmental Effects

Findings from studies that addressed neurodevelopmental effects were less consistent than those that addressed non-nervous system developmental effects. For example, some studies found structural alterations in the brains of offspring following gestational oral exposures, while others did not. The first study (Sitarek et al., 1994) observed significant dose-related increases in the incidence of dilation of the lateral and/or third cerebral ventricles and subarachnoid space in Wistar Imp:DAK rat fetuses exposed in utero to 300, 1000 and 5000 mg/kg/day n-butanol; however, a later study (Ema et al., 2005) did not find any abnormalities in the fetal brain of Sprague-Dawley rats at similar doses.

Enlarged brain ventricles were observed in Sprague-Dawley rat fetuses whose mothers had been exposed during pregnancy to 24,252 mg/m³ of n-butanol in an inhalation developmental toxicity study (Nelson et al., 1989a). However, this study (Nelson et al., 1989a) reported enlarged brain ventricles as one of several visceral variations that did not reach statistical significance and did not provide incidence data for this specific abnormality. Additionally, in another study (Hackett et al., 1982), brain malformations were observed in two Sprague-Dawley rat fetuses at 7,126 mg/m³: hydrocephaly in one and general brain dysmorphology in the other, although these findings also were not statistically significant.

A possible reason for the discrepancies between the findings of the two primary studies evaluating dilated brain ventricles (Sitarek et al., 1994; Ema et al., 2005) may be due to differences in study design. The latter (Ema et al., 2005) utilized a gestation-only exposure regimen, while the former (Sitarek et al., 1994) also exposed rats prior to and during mating. Yet the exposure regimen probably doesn’t account for the discrepancies as the susceptible period for morphological alterations such as dilation of the brain ventricles in rat fetuses is thought to be during mid and late pregnancy (Kameyama, 1985). More likely, however, there might be differences in strain sensitivity to n-butanol because the earlier study (Sitarek et al., 1994) used Imp:DAK Wistar rats, whereas the later study (Ema et al., 2005) used Sprague-Dawley rats.

Strain sensitivity in rodents to developing dilated brain ventricles has been reported elsewhere in the peer-reviewed literature. For example, following maternal exposure to aspirin during pregnancy, hydrocephalus, which causes dilation of brain ventricles, was reported in fetuses of Wistar rats (Kimmel et al., 1971) and Long-Evans rats (Mankes et al., 1982), but not in fetuses of Sprague-Dawley rats (Gupta et al., 2003).

Structural alterations in the brain are often accompanied by functional changes in behavior; however, the n-butanol and n-butyl acetate studies identified in the literature search did not provide strong evidence for functional effects following developmental exposures. Specifically, an inhalation study (Nelson et al., 1989b) investigated neurobehavioral outcomes in offspring following n-butanol exposures of 9,125 or 18,189 mg/m³ to pregnant mothers or to fathers exposed for 6 weeks prior to mating and identified no treatment-related effects.

It is important to note that the technical methods employed in soft tissue evaluation of fetuses in developmental toxicology studies should ensure an assessment of the inner structure of the brain for approximately half of the fetuses in a rat study, and up to 100% of rabbit fetuses (U.S. EPA, 1998a; OECD, 2001). These procedures were already commonly practiced by many laboratories prior to 1998. If the inner structure of the brain were examined in the fetuses on the n-butanol studies, one might have expected enhanced sensitivity of detection and/or characterization of brain dysmorphology following in utero n-butanol and n-butyl acetate exposures at high doses in laboratory animals. It is possible, however, that apparent lack of consistency in response might have been due to non-apparent procedural differences among laboratories. It might reflect insufficient power of the study to detect rare events. Or perhaps the diversity of adverse effects might be attributed to the incredible complexity of nervous system development, and the possibility that a single toxicological insult might disrupt multiple interconnected adverse outcome pathways.

In summary, the findings from one study (Sitarek et al., 1994) suggest that oral exposure to n-butanol increases the risk for dilation of brain ventricles/subarachnoid space; however, the findings from another study (Ema et al., 2005) suggest it does not. Therefore, additional research studies evaluating the risk for this neurodevelopmental outcome following n-butanol gestational exposure are needed to have more confidence in making a hazard determination. Figures 2 (oral exposure) and 3 (inhalation exposure) present the developmental and neurodevelopmental effects that were identified in the literature and the exposure levels at which they occurred.

Figure 2:

Exposure array for developmental and neurodevelopmental effects following oral exposure to n-butanol.

Figure 3:

Exposure array for developmental and neurodevelopmental effects following inhalation exposure to n-butanol or butyl acetate.

Nervous System Effects in Mature Animals

Studies identified in the literature search suggest that nervous system effects in mature animals may be a hazard of n-butanol exposure. It is difficult, however, to determine the most sensitive adverse nervous system effect and the exposure level (point of departure, POD) at which it occurs because of the low confidence in some of the observed findings resulting from a lack of supporting evidence and study quality issues for some of the studies as discussed below.

The strongest evidence for n-butanol-related nervous system effects was for deficits in motor activity and function. The literature search identified one study that evaluated nervous system effects following oral exposure in adult animals (TRL, 1987), which reported an increased incidence of ataxia and hypoactivity in Sprague-Dawley rats administered 500 mg/kg/day n-butanol by gavage for 13 weeks. These neurobehavioral alterations were transient, disappearing within one hour following cessation of dosing. Although effects that are transient may be of less concern than those that are longer lasting, findings that seem reversible may re-appear later in life or be predictive of later adverse outcomes (U.S. EPA, 1998b). The effects were first observed at week 11 and then were consistently observed through week 13. Similarly, another study (David et al., 1998) identified minor reductions in activity and alertness during 13 weeks of inhalation exposure to 7,126 and 14,252 mg/m³ of n-butyl acetate that ended when the rats were no longer exposed.

Additional evidence for motor activity and functional effects comes from a 13-week n-butanol inhalation study that reported decreased motor coordination at 303 mg/m³ in male rats as demonstrated by performance on the rotarod test (Korsak et al., 1994). A major limitation for these data is the presentation of rotarod performance as a quantal variable as opposed to a continuous maximum latency allowed to prevent confounding by fatigue. Although the quantal percent failures data can provide useful hazard information, these measures require an arbitrary selection of the length of time required for successful performance; however, there is no scientific consensus on an optimal time for this parameter. In addition, when identifying effect levels based on the data presented in this study (Korsak et al., 1994), latencies on the rod of 1 second versus 119 seconds would be treated identically as failures when, in fact, they indicate very different levels of motor coordination.

In addition to the reporting of quantal data, there were several procedural shortcomings, such as the possible confounding by a 6 to 7% weight gain in the n-butanol-treated rats, since weight gain can influence performance on the rotarod. Nonetheless, acute exposure studies support the findings of impaired motor function (Bowen and Balster, 1997; Korsak and Rydzyński, 1994; Korsak et al., 1993; Maickel and Nash, 1985; Wallgren 1960). Confidence in the Korsak et al. (1994) findings is also strengthened because the number of quantal failures increased (from 0–2 to 2–4 of 12 rats/ treatment group) with longer durations of n-butanol exposure, and because none of the 24 control rats were reported to fail at any point during the study.

An additional nervous system effect was identified in a 13-week n-butyl acetate inhalation exposure study (David et al., 2001), which reported decreased mean absolute brain weight of 7% in adult male and female rats at the highest exposure concentration (14,253 mg/m³). No histopathology changes, however, were observed in the brains or other neural tissues from rats exposed for the same length of time and to the same concentration of n-butyl acetate (David et al., 1998). Furthermore, the n-butanol oral exposure study (TRL, 1987) did not find any changes in brain weight. Decreased body weight and food consumption were also found in the study that identified decreased absolute brain weight (David et al., 2001); however, change in absolute brain weight is considered a biologically significant effect regardless of whether changes in body weight occur simultaneously, because brain weight, unlike the weight of most other organs or tissues, is generally conserved even during malnutrition or weight loss (U.S. EPA, 1998a). Although the lack of consistency between the two studies (David et al., 2001; TRL, 1987) might be due to different exposure routes (inhalation vs. oral), one would have more confidence in the decreased brain weight findings (David et al., 2001) if additional studies identified this effect following exposure to either n-butanol or n-butyl acetate.

Consistent with the neurotoxic effects reported in animal studies, neurotoxic effects associated with n-butanol exposure were also reported in humans. An occupational study of workers (n=11) in a cellulose acetate ribbon factory demonstrated hearing impairment associated with exposure to n-butanol (Velazquez et al., 1971). Altered sensory function is considered a neurotoxic effect (U.S. EPA, 1998b). In the occupational study (Velazquez et al., 1971), exposure levels to n-butanol were reported to be 242 mg/m³ and factory noise levels were approximately 72−78 db, but details of the exposure assessment were not described. A comparison group was selected from workers in another factory who were not exposed to n-butanol, but were exposed to higher noise levels (90 to 110 db). The prevalence of hearing loss was higher among the workers exposed to n-butanol compared to the workers exposed to higher-noise levels without the n-butanol exposure.

This study (Velazquez et al., 1971) has severe study-quality limitations in addition to the low number of subjects. First, the comparison group had a different occupational co-exposure than did the n-butanol-exposed workers, and it is not clear if other characteristics were similar between the two groups. Second, the authors assume that n-butanol exposures and noise levels measured at the time of the study were constant for the job tenure of 3–11 years of the study subjects, which may not be accurate as no further information on exposures was provided. Third, confounders such as cigarette smoking and disease status were not considered. In conclusion, confidence in the study findings is low. Additional research is needed to determine whether chronic n-butanol exposure can result in hearing loss.

Because there was only one oral exposure study in laboratory animals (TRL, 1987), which found only one transient effect, increased incidence of hypoactivity and ataxia, additional research is needed that investigates the nervous system effects of oral exposure to n-butanol. Likewise, for inhalation exposures, the two most sensitive endpoints, impaired performance on the rotarod and hearing loss, are both from studies with study quality concerns. Therefore, new higher quality studies are needed to assess these endpoints.

Figures 4 (oral exposure) and 5 (inhalation exposure) present the neurotoxic effects that were identified in the literature and the exposure levels at which they occurred.

Figure 4:

Exposure array for nervous system effects following oral exposure to n-butanol.

Figure 5:

Exposure array for nervous system effects following inhalation exposure to n-butanol or butyl acetate.

Comparison of Oral and Inhalation Dose-Response Data and Potential for Route-to-Route Extrapolation

As described above, after exclusion of studies with quality issues, the n-butanol database is limited for oral ingestion, but may be stronger for inhalation exposure if data for n-butyl acetate are considered. However, when considering only developmental neurotoxicity, which is the greatest concern of the structurally similar ethanol, very little inhalation exposure data exists. Because of these limitations in the toxicological database, a means of cross-route comparison of the dose-response data and the potential for extrapolating between oral and inhalation exposures should be considered. For example, could the oral exposure study (Sitarek et al., 1994) be extrapolated to an inhalation exposure value or, alternatively, could an inhalation exposure value be extrapolated to an oral exposure value? How can the existing studies be compared?

PBPK models are tools used to simulate absorption, distribution, metabolism, and excretion (ADME) in animals and humans, allowing for the prediction of an internal concentration or measure of exposure (i.e., an internal dose metric) by which data for different routes of exposure can be compared. Commonly used internal dose metrics are the peak concentration or area-under-the-concentration-curve (AUC) of a chemical in the blood. If the dose metric is predictive of toxicity (e.g., the peak blood concentration of an alcohol should be predictive of acute motor impairment), then it can be used to compare the dose-response for endpoints related to that metric, observed from different routes of exposure. Using an appropriate internal dose metric, PBPK models can also be used to extrapolate toxicological dose-response data between species (e.g., laboratory animal-to-human), and to estimate equivalent exposures by different routes (e.g., the inhalation concentration that would yield the same internal dose as one associated with an oral toxicity level). However, extrapolation can also add uncertainty depending on the available data, mechanistic information, and PBPK models (McLanahan et al., 2012).

A PBPK model for n-butyl acetate and its butyl metabolites (including n-butanol) that can be used for a route-to-route comparison (Barton et al., 2000) was later refined (Teeguarden et al., 2005) and describes blood kinetics for n-butyl acetate, n-butanol and n-butyric acid, in rats and humans. For the current purposes, the oral route of exposure for n-butanol was incorporated into the refined model (Teeguarden et al., 2005) by assuming that the absorption rates for n-butanol are the same as those of ethanol (based on studies by Sultatos et al., 2004; Pastino and Conolly, 2000).

Available data are not sufficient to clearly identify the appropriate internal dose metric for the nervous system and developmental effects of n-butanol. Both the duration and magnitude of elevated concentrations at affected sites have been correlated with effects from ethanol exposure (Riley and McGee, 2005; Driscoll et al., 1990). The PBPK model for n-butanol, n-butyl acetate, and metabolites simulates concentrations in blood, liver, and a lumped compartment representing other tissues (Teeguarden et al., 2005), but it does not explicitly simulate concentrations in the brain or in the developing fetus. In the absence of dose metric-specific data, parent compound concentration in the blood is generally selected as a surrogate because it represents the circulating matrix for distribution in the body. To account for both magnitude and duration of exposure, the average area under the curve (AUC) for n-butanol concentration in arterial blood was used as the dose metric for the neurotoxicity and developmental toxicity observations in animals, and for dosimetry extrapolation to humans. As mentioned earlier, there are uncertainties when utilizing route-to-route extrapolations. For nervous system and developmental toxicity effects of n-butanol, particularly when extrapolating from the oral route of exposure to the inhalation route of exposure, the following uncertainties exist:

• The proportion of parent chemical and metabolites may differ depending on the route of exposure, and the metabolite n-butyric aldehyde may contribute to the developmental and other effects of exposure to n-butanol. Hence, the effects may be a function of the combined concentration of n-butanol and the metabolite, or a weighted sum of those concentrations. But in the absence of empirical data to inform the relative potency of each, it is unclear how one would factor in the proportions of the parent and metabolite when evaluating dose-response, or how the effects might differ by route of exposure.

• There are limited oral toxicokinetic data for n-butanol and n-butyl acetate for adequate validation of the PBPK model for oral exposures. The model was calibrated using inhalation data only, and the true impact of the first-pass effect in vivo is unknown. This increases uncertainty around the internal dose metric following oral exposure. The actual n-butanol concentration, as well as the ratio of parent to metabolite, are highly uncertain for oral ingestion.

• For oral exposures, the peak concentration is highly dependent on the oral ingestion pattern; i.e., if the individual tends to consume water in large (less frequent) boluses or drinks more gradually throughout the day. Hence, uncertainty in the ingestion pattern for an animal drinking water study and variability in human drinking patterns leads to significant uncertainty in extrapolation when peak concentration is the dose metric. The AUC is much less sensitive to the exposure pattern.

• For a different but related chemical (ethanol), a study (Oshiro et al., 2014) measured neurodevelopmental effects of inhalation exposure to ethanol at high concentrations. This study found that when comparing effects of ingested or inhaled ethanol, the oral route of exposure to ethanol resulted in greater neurodevelopmental toxicity in Long-Evans rats than did inhalation, despite both routes of exposure yielding the same blood AUC of the parent compound ethanol. This suggests that it is plausible that, at equal n-butanol blood levels, neurodevelopmental effects might not be equivalent following oral and inhalation exposures.

Because of these issues, the PBPK model is considered too uncertain for use in route-to-route (oral to inhalation) extrapolation. Nevertheless, it can be informative to perform these calculations to understand what the model suggests for the two routes of exposure. Tables 3 and 4 outline the administered exposures, internal doses, and external human equivalent exposures (including route-to-route extrapolations) that were calculated using the modified PBPK model. Table 3 contains values for the inhalation bioassays for n-butanol and n-butyl acetate, while Table 4 contains values for the drinking water bioassays for n-butanol. For additional information on PBPK model parameters and code, see Supplement B and the EPA Health and Environmental Online Database (https://hero.epa.gov/hero/index.cfm/reference/details/reference_id/4353650).

Table 3.

Internal doses and human equivalent exposures for n-butanol and n-butyl acetate inhalation studies

Study/description	Chemical	Exposure profile	Inhaled conc.a (ppm)	Internal doseb (AUC24, mmol-h/L)	HECc (ppm n-butanol)	HEDd (mg/kg-d n- butanol)
Rat (Sprague-Dawley), M+F (David et al., 2001)	n-Butyl acetate	6h/d, 5d/w	500 (N)	0.177	265	64.8
			1,500 (L)	0.536	799	154
			3,000	1.10	1630	271
Rat (Sprague-Dawley), F (dams) (Saillenfait et al., 2007)	n-Butyl acetate	6h/d, 7d/w	501 (N)	0.248	371	84.7
			1,010 (L)	0.502	749	146
			1,997	1.01	1500	253
			2,966	1.52	2240	352
Rat (Wistar), M (Korsak et al., 1994)	n-Butanol	6h/d, 5d/w	50 (N)	1.27e-02	19.0	6.03
			100 (L)	2.54e-02	38.1	11.7
Rat (Sprague-Dawley), M+F (Nelson et al., 1989a)	n-Butanol	7h/d, 7d/w	3,500 (L)	1.47	2170	343
			6,000	2.55	3720	542
			8,000	3.45	4990	704
New Zealand White rabbit, M+F (Hackett et al., 1982)	n-Butyl acetate	7h/d, 7d/w	1,500 (L)	N/A	437e	N/A

^a(N) and (L) indicate corresponding NOAEL and LOAEL across all endpoints evaluated in the study (see Figures 3 and 4).

^bPBPK modeling using the rodent model assumed exposure time profiles defined by the study to estimate the internal n-butanol dose (AUC₂₄) at the rodent inhaled concentration. AUC₂₄ is the area under the curve of n-butanol blood concentration (mmol-h/L). The corresponding steady-state blood concentration, assuming continuous exposure, can be obtained by dividing the AUC₂₄ by 24 h, which is also the average blood concentration (C_avg), given in Supplemental Table 1.

^cThe human equivalent concentration (HEC) is the inhaled n-butanol concentration (ppm) needed to achieve an AUC₂₄, using the human PBPK model assuming continuous 24-hour/day exposure.

^dThe human equivalent dose (HED) is the ingested n-butanol dose (mg/kg-d) needed to achieve an AUC₂₄, using the human PBPK model assuming a drinking water exposure profile. It was assumed that humans would consume n-butanol in drinking water as bolus events 6 times/day at varying fractions (sum = 100% daily dose over 1 day) of the total daily dose.

^eTo account for intermittent exposure (7 hours/day, 7 days/week), the reported concentration of 1,500 ppm was multiplied by (7/24) to obtain a time-adjusted value. No PBPK model was available for rabbits. As recommended (U.S. EPA 1994b), for extra-respiratory effects, the time-adjusted value was extrapolated to the HEC by multiplying it by the rabbit-to-human ratio of blood:gas partition coefficients. Because the partition coefficients are unknown, the default value of one was used for the ratio.

Table 4.

Internal doses and human equivalent exposures for n-butanol drinking water studies

Study/description	Exposure profile	Oral dosea (mg/kg-d)	Internal doseb (AUC24, mmol-h/L)	HEDc (mg/kg-d)	HECd (ppm)
Rat (Sprague-Dawley), F (dams) (Ema et al., 2005)	Drinking water (ad libitum)	316	9.88e-2	40.1	148
		1,454 (N)	0.734	197	1090
		5,654 (L)	9.02	1670	12400
Rat (Imp:DAK Wistar), F (dams) (Sitarek et al., 1994)	Drinking water (ad libitum)	300 (L)	9.33e-2	38.2e	140
		1000	0.409	125	610
		5000	7.38	1390	10300

^a(N) and (L) indicate corresponding NOAEL and LOAEL across all endpoints evaluated in the study (see Figure 2).

^bFor oral drinking water modeling in the rat, the drinking water exposure model was based on data from the study (Spiteri, 1982). The corresponding steady-state blood concentration, assuming continuous exposure, can be obtained by dividing the AUC24¬ by 24 h, which is also the average blood concentration (C_avg), given in Supplemental Table 2.

^cThe human equivalent dose (HED) is the ingested n-butanol dose (mg/kg-d) needed to achieve an AUC₂₄, using the human PBPK model assuming a drinking water exposure profile. AUC₂₄ is the 24-hour average area under the curve of n-butanol blood concentration (mmol-h/L). It was assumed that humans would consume n-butanol in drinking water as bolus events 6 times/day at varying fractions (sum = 100% daily dose over 1 day) of the total daily dose.

^dThe human equivalent concentration (HEC) is the inhaled n-butanol concentration (ppm) needed to achieve an AUC₂₄, using the human PBPK model assuming continuous 24-hour/day exposure.

^eThis is the lowest HED at which adverse health effects (in this case the neurodevelopmental outcome of brain ventricle dilation) were identified.

Mechanistic Data

The primary toxicities identified in laboratory animal and human studies following exposure to n-butanol are associated with the central nervous system (neurotoxicity) and development (developmental and neurodevelopmental toxicity). While identifying toxicities associated with a chemical is needed for hazard characterization, it is also important to understand the underlying mechanisms leading to the resultant toxic effect. Knowledge of the mechanisms leading to the observed toxic effect could decrease uncertainty associated with potential toxicodynamic differences between animals and humans. With respect to the observed toxicities following exposure to n-butanol, the corresponding mechanistic data do not provide a clear understanding of how the effects could be caused and their relevance to humans.

Most of the available mechanistic studies for n-butanol only provide information on initial events that may lead to the observed toxic effect. Many of the hypothesized intermediate and late-stage mechanisms for n-butanol are speculative because they are based on mechanistic data for the structurally similar ethanol. The analysis leading to the hypothesized mechanisms for the observed neurotoxicity and developmental toxicity effects are described below.

Hypothesized mechanisms leading to developmental toxicity effects

Developmental toxicity effects, such as skeletal variations, have not been studied with n-butanol even though these effects were consistently observed with n-butanol and its metabolic precursor, n-butyl acetate. With ethanol, it is known that heavy alcohol consumption in adults leads to loss in bone density through mechanisms of osteocyte apoptosis and oxidative stress (Maurel et al., 2012). Additionally, when mouse embryos were treated with ethanol, decreased expression was observed of a gene (msx2) known to regulate and promote osteoblast proliferation (Rifas et al., 1997). A human homolog of msx2 has been identified as being critical to proper skull development, so this mechanism plausibly is relevant to humans (Rifas et al., 1997). Again, it is unclear if these mechanisms are operative in the skeletal observations following gestational exposure to n-butanol; however, the similarities between the two alcohols warrant additional investigation.

Hypothesized mechanisms leading to neurodevelopmental effects

The hypothesized mechanisms for neurodevelopmental toxicities associated with n-butanol exposure were previously discussed and analyzed in another publication (Bale and Lee, 2016). Briefly, they include the interaction of n-butanol with the (1) L1 cellular adhesion molecule (CAM) (Tang et al., 2011) and (2) phospholipase D (PLD) and perturbation of the PLD-mediated cell signaling pathway (Kötter and Klein, 1999). It is known that these initial interactions between n-butanol and L1 CAM and/or PLD leads to inhibition of neurite outgrowth (Watanabe et al., 2004) and decreased astrocyte proliferation (Kötter and Klein, 1999; Kotter et al., 2000). Beyond these initial changes, it is unknown how n-butanol-mediated inhibition of neurite outgrowth and decreased astrocyte proliferation could lead to the observed neurodevelopmental toxicities. These highlighted mechanistic changes are consistent with ethanol, a known neurodevelopmental toxicant. There are, however, several hypothesized mechanisms contributing to ethanol-related neurodevelopmental toxicity that may or may not be applicable to n-butanol. For example, ethanol induces microglial activation and oxidative stress, leading to increased neuronal death (Guizzetti et al., 2014).

Hypothesized mechanisms leading to neurotoxicity effects

Several studies demonstrated that acute and subchronic exposure to n-butanol (or its precursor, n-butyl acetate) results in neurotoxic effects including observations of decreased motor activity and impaired motor coordination (David et al., 1998; Korsak et al., 1994; TRL, 1987; Bowen and Balster, 1997; Korsak and Rydzyński, 1994; Korsak et al., 1993; Maickel and Nash, 1985; Wallgren, 1960). A review of the mechanistic data for n-butanol suggests that interactions with the excitatory N-methyl-D-aspartate (NMDA)-glutamate receptor and the inhibitory gamma-aminobutyric acid (GABA) receptor which are found ubiquitously throughout the brain and are involved in motor function may be initial events leading to neurotoxic outcomes. In vitro, n-butanol inhibits NMDA-glutamate receptor function (Peoples and Weight, 1999; Dildy-Mayfield et al., 1996; Lovinger et al., 1989) and potentiates GABA receptors (Mascia et al., 2000; Peoples and Weight, 1999; Ye et al., 1998; Dildy-Mayfield et al., 1996; Nakahiro et al., 1991). There are no mechanistic studies that evaluated intermediate steps leading to the observed nervous system effects reported for n-butanol following this initial interaction with n-butanol and perturbation of NMDA-glutamate and GABA receptor function.

As described earlier, where there were mechanistic data gaps for n-butanol, data from ethanol were used to develop hypothetical mechanisms for n-butanol. Mechanistic studies with ethanol reported that following increased activation of GABA receptors, mitotic spindles in neural progenitor cells were disrupted (Tochitani et al., 2010) resulting in a decreased number of neural progenitor cells in adult Sprague-Dawley rats following a 5- or 28-day treatment with ethanol (Nixon and Crews, 2002). Similarly, following inhibition of NMDA-glutamate receptor function by ethanol, the release of neuronal nitric oxide synthase (nNOS) from postsynaptic density protein 95 (PSD 95) is inhibited; nNOS is an enzyme that interacts with the glucose transporter (Ferreira et al., 2011; Fattoretti et al., 2003). Decreased nNOS levels following ethanol exposure has been demonstrated to decrease glucose uptake and glucose levels causing an increase in apoptotic markers leading to increased neuronal death (Fattoretti et al., 2003). The combination of increased neuronal death and decreased number of neurons in the brain would be expected to lead to neurotoxicity that is observed for n-butanol. A schematic of the hypothesized mechanisms leading to neurotoxicity based on both n-butanol and ethanol mechanistic data is presented in Figure 6.

Figure 6.

Proposed schematic of the hypothesized mechanistic events resulting in neurotoxicity following exposure to n-butanol.

DISCUSSION

Our literature search identified several studies that evaluated developmental, neurodevelopmental, and adult nervous system effects of exposure to n-butanol or n-butyl acetate. Although study results for developmental effects were generally consistent, there were inconsistencies and study quality issues for neurodevelopmental and adult nervous system effects. We identified other issues as well, such as the appropriateness of comparing dose-response data from the oral route to the inhalation route of exposure using PBPK models, and a lack of mechanistic information specific to n-butanol. These issues highlight data gaps in the n-butanol literature and add many uncertainties to the evaluation of the human health risk of n-butanol exposure.

To illustrate how some of these data gaps could be addressed, the focus will be on the discrepancies in neurodevelopmental findings reported in two studies (Sitarek et al., 1994; Ema et al., 2005), which were both considered studies without major study-quality concerns. The first (Sitarek et al., 1994) observed dilation in brain ventricles in Wistar rats following developmental n-butanol exposure to 300 mg/kg-d. Confidence in this value, however, is less than optimal given that the second study (Ema et al., 2005) did not find dilation of brain ventricles at a similar dose in a different rat strain. Although this discrepancy may be due to varying technical procedures or to differences in strain sensitivity, additional studies evaluating neurodevelopmental outcomes would add confidence to a determination of n-butanol’s potential as a developmental neurotoxicant.

Aside from additional neurodevelopmental toxicity testing in rats or mice, which are both costly and time-consuming, new high-throughput screens exist that might be used to evaluate these effects of n-butanol exposure. One option is the zebrafish model, which has already been used successfully to evaluate the developmental neurotoxicity associated with ethanol exposure (Louck and Ahlgren, 2012). Because the zebrafish is a vertebrate species with high physiologic and genetic homology to humans, it is a useful model to study brain disorders in humans. For example, the zebrafish nucleotide sequence shares 70% homology with humans (Kalueff et al., 2014). Another strength of this model is ease of genetic manipulation (Kalueff et al., 2014). Zebrafish assays can also be generated rapidly since zebrafish neurodevelopment occurs within 3 days of fertilization (Lee and Freeman, 2014), and zebrafish produce hundreds of transparent embryos that develop outside of the uterus, which enables analysis of internal structures. Important to the assessment of n-butanol, this model can be used specifically to study brain ventricle development (Lowery and Siva, 2009). Inclusion of zebrafish data in an n-butanol risk assessment could help to clarify the neurodevelopmental potential of this alcohol and could potentially decrease uncertainty with respect to the endpoint of brain ventricle dilation following gestational exposure to n-butanol.

Data from additional in vitro assays could also be used to decrease uncertainty associated with the reported neurodevelopmental effects following gestational n-butanol exposure. One potential target that could be studied further is the neuronal cell adhesion molecule, L1 CAM, which has been hypothesized to be involved in ethanol-induced developmental neurotoxicity (Bearer, 2001). Altered L1 CAM has also been linked in humans to hydrocephalus (Lowery and Siva, 2009), which is characterized by brain ventricle dilation. n-Butanol has been reported to inhibit L1 CAM-mediated cell-cell adhesion in mouse embryo fibroblasts (Ramanathan et al., 1996) and to increase L1 CAM levels in lipid rafts. Following ethanol exposure, increased L1 CAM levels in lipid rafts have been associated with inhibition of neurite outgrowth (Tang et al., 2011). Although it is known that n-butanol interacts with L1 CAM, it is still unclear how this interaction may correlate to observed neurodevelopmental outcomes. Therefore, a useful next step would be to determine if n-butanol-mediated increases of L1 CAM levels in lipid rafts are associated with inhibition of neurite outgrowth. Recently, high-throughput assays have been developed that evaluate neurite outgrowth using human induced pluripotent stem cell-derived neurons (Ryan et al., 2016). Such assays would be useful in elucidating a potential association between n-butanol-mediated increases of L1 CAM levels in lipid rafts and neurite outgrowth inhibition.

Relevance of these data to human risk at environmental or occupational exposure levels could be assessed by using PBPK models to extrapolate from in vitro to in vivo concentrations. The PBPK model can be run assuming various exposure profiles in humans (including aggregate exposure) to estimate a range of inhaled concentrations or ingested doses that may yield internal concentrations that are comparable to those that cause adverse effects in such in vitro experiments. The current PBPK model does not describe pregnancy, and ideally a version capable of predicting concentrations in the fetal brain would be used. It has, however, been shown for methanol that concentrations in the fetus closely paralleled those in the non-pregnant adult for rodents (U.S. EPA, 2013) and, hence, an explicit model of pregnancy was determined not to be necessary. Given that the biochemical properties of n-butanol are similar to methanol, one can expect likewise that blood concentrations of n-butanol in the adult would be a reasonable surrogate for fetal exposure.

In conclusion, n-butanol is not unlike many chemicals that are assessed by the EPA and other health organizations. Several studies evaluate n-butanol health effects, yet data gaps still exist. This paper presents issues and sources of uncertainty that are often encountered when assessing the adverse health effects of chemicals. In addition, it describes how high-throughput screening methodologies might be used to reduce uncertainties that result from data gaps. This ultimately could contribute to higher confidence in risk assessment conclusions.