Archived publication

This publication is no longer current or has been superseded.

2.9    Benzene (C6H6)

The following discussion on the toxicity of benzene summarises relevant data from various reviews (WHO, 1993; Baars et al, 2001; NICNAS, 2001; US EPA, 2002; 2003; DEFRA and EA, 2003). Particular attention is given to those studies that have been used in deriving reference health standards. Readers are referred to the original reviews for more details on adverse health effects.

2.9.1   Toxicological status

Benzene is toxic by all routes of administration. The bone marrow is the target organ for the expression of benzene haematotoxicity and immunotoxicity. This is manifested by decreased levels of antibodies and blood cells in exposed people. All blood cells (ie, erythrocytes, leukocytes, and platelets) may be affected to varying degrees. The accelerated destruction or reduction in the number of all three major types of blood cells is termed pancytopenia. Chronic benzene exposure can result in bone marrow depression, expressed as a variety of conditions including aplastic anaemia and leukaemia. Aplastic anaemia can progress to leukaemia, whereas pancytopenia may be reversible. Which effect is observed depends on the dose, length of exposure, and the stage of stem cell development affected (Baars et al, 2001). Neither gastrointestinal effects from oral exposure nor pulmonary effects due to inhalation exposure have been reported (US EPA, 2002).

Benzene is a well-established human carcinogen. Epidemiological studies of benzene-exposed workers have demonstrated a causal relationship between benzene exposure and the production of acute myelogenous leukaemia and also suggest evidence for chronic leukaemias. Other neoplastic conditions that are associated with an increased risk in humans are haematologic neoplasms, blood disorders such as pre-leukaemia and aplastic anaemia, Hodgkin’s lymphoma, and myelodysplastic syndrome (MDS). Benzene is classified as a known human carcinogen (Class 1, Group A) by the IARC (1987) and the US EPA (2000) respectively.

The genotoxicity of benzene has been studied extensively (eg, see Whysner, 2000; US EPA, 2002; 2003; ATSDR, 2007). Benzene does not induce gene mutations in in vitro systems, but several studies have demonstrated induction of both numerical and structural chromosomal aberrations, sister chromatid exchanges and micronuclei in experimental animals and humans after in vivo benzene exposure (WHO, 1993). Benzene does not appear to interact directly with DNA under normal in vivo exposure conditions. It is suggested that the responses elicited are due to interactions of the metabolites of benzene, which most likely damage the DNA of stem cells (WHO, 2000; ATSDR, 2007).

Benzene metabolism occurs mainly in the liver, is mediated primarily through the cytochrome P-450 enzyme system, and involves the formation of a series of unstable reactive metabolites. The main metabolites are phenol, catechol and hydroquinone. Studies suggest that benzene toxicity is the result of the interactive effects of several benzene metabolites formed in both the liver and the bone marrow (WHO, 1993). In rodents the formation of two toxic metabolites, benzoquinone and muconaldehyde, appears to be saturable, ie, reaches a plateau, indicating that a higher proportion of the benzene will be converted to toxic metabolites at low doses than at high doses.

Multiple mechanistic pathways have been proposed that may give rise to cancer and, in particular, to formation of leukaemia from exposure to benzene (Smith, 1996). Debate is currently focused on whether carcinogenic mechanisms observed at high occupational exposures are also occurring at lower environmental exposures (Baars et al, 2001).

2.9.2   New Zealand classification

The Hazardous Substances and New Organisms Act classification of benzene set by ERMA NZ is shown in Table 43. Overall, benzene is of relatively high toxicity (6.1B oral classification), and is a skin and eye irritant (6.3A, 6.4A) and in terms of long-term endpoints benzene is mutagenic (6.6A), a proven human carcinogen (6.7A), developmnetla toxicant (6.8A) and is highly toxic from chronic exposures (6.9A).

Table 43: HSNO classification of benzene

Hazardous property

HSNO classification

Acute toxicity

6.1B

Skin irritation

6.3A

Eye irritation

6.4A

Sensitiser

ND

Mutagenicity

6.6A

Carcinogenicity

6.7A

Reproductive/developmental toxicity

6.8A

Target organ systemic toxicity

6.9A

ND – no classification due to no data/insufficient data/inconclusive data.

2.9.3   Reference health standards

A number of regulatory agencies have developed guideline values for benzene. In contrast to most contaminants, estimates of toxicological intakes are primarily derived from inhalation studies. Some agencies have developed intakes for threshold effects, in addition to non-threshold effects.

Ingestion

Oral intake values for benzene for threshold and non-threshold effects and their bases of derivation are shown in Tables 44 and 45. Most oral values are derived from inhalation studies using route-to-route extrapolation, hence greater detail of the individual studies is provided in that section.

Only US EPA and Dutch agencies have developed threshold values for benzene (Table 44). These values are relatively similar despite different derivation methods.

Oral risk-specific doses (for a risk of 1 in 100,000) range from 0.18 to 0.67 μg/kg bw/day, with most being around 0.3 μg/kg bw/day (Table 45). The exception is the Canadian risk-specific dose, which is an order of magnitude lower. However, insufficient information is available to indicate why this is the case.

Existing New Zealand documents (MfE, 1997; 1999) use a slope factor of 0.029 per mg/kg bw/day for both oral and inhalation reference doses. This nominally arises from the US EPA, but neither of the MfE documents provides the actual reference so it is not possible to critique these values, other than to observe that the more recent US EPA (2000) values will have superseded the original source value.

The WHO drinking water guidelines used a rat gavage study to support guideline values derived from an (unstated) occupational inhalation study (WHO, 2003). This is the only use of animal studies in deriving intake values for benzene.

Table 44: Summary of oral reference health standards for threshold effects of benzene, used by different international agencies

Jurisdiction

Tolerable daily intake (μg/kg bw/day)

Key study1

Critical effect1

Basis of value

Reference

The Netherlands – current

4.3

Not stated

Haematological effects

Vermeire et al (1991). Route-to-route extrapolation from inhalation value (30 μg/m3) – see Table 46

Baars et al (2001)

US ASTDR – chronic duration MRL and US EPA

4.0

Rothman et al (1996)

Decreased lymphocyte count, occupational inhalation exposure

BMDL 1.2 mg/kg/day converted from BMCLadj of = 8.2 mg/m3 (see Table 46 for description of BMCLadj) using route-to-route extrapolation: the BMCLadj is multiplied by the default inhalation rate, multiplied by 0.5 to correct for the higher oral absorption, and divided by the standard default human bodyweight of 70 kg: 8.2 mg/m3 x 20 m3/day x 0.5 / 70 kg = 1.2 mg/kg/day

A UF of 300 was applied, which comprised a factor of 3 for extrapolation from an effect to a no-effect level, a factor of 10 to account for human variability, a factor of 3 for sub-chromic to chronic exposure, and a factor of 3 to account for database deficiencies. The resultant UF (270) is rounded up to 300.

ATSDR (2007)

US EPA (2003)

1 As reported in the reference cited in the reference column.

Table 45: Summary of oral reference health standards for non-threshold effects of benzene used by different international agencies

Inhalation

Benzene has a high vapour pressure (9.95 kPa at 20°C), thus inhalation is an important route of exposure. Most of the knowledge of human effects of benzene arises from occupational inhalation studies, and the majority of toxicological intake values are derived from such studies. The so-called “Pliofilm cohort” is the most thoroughly studied, and forms the basis for most of the values in Tables 46 and 47. According to the US EPA (2000), this study had the least number of confounders and a wide range of exposure to benzene, while other epidemiological studies had some methodological problems, ie, confounding exposures, lack of sufficient power, and other limitations, which precluded their use in quantitative risk estimation. However, exposure estimates for this cohort vary considerably and three different exposure matrices have been used to describe the Pliofilm cohort, ie, those reported by Crump and Allen (1984, cited in US EPA, 2000), by Rinsky et al (1987, cited in US EPA, 2000), and a newer and more extensive one by Paustenbach et al (1993, cited in US EPA, 2000).

The risk estimates provided by Crump (1994) presented 96 unit risk calculation analyses by considering different combinations of the following factors: (1) different disease endpoints, (2) additive or multiplicative models, (3) linear/nonlinear exposure–response relationships, (4) two different sets of exposure measurements (Crump and Allen (1984) vs exposure estimates by Paustenbach et al (1993 cited in US EPA, 2000) and (5) cumulative or weighted exposure measurements. The unit risk estimates range from 8.6 × 10–5 to 2.5 × 10–2 at 1 ppm (3200 µg/m3) of benzene air concentration (Crump, 1994).

US EPA (2000) concluded that at present there is insufficient evidence to reject the concept of linearity at low doses. When a linear model was employed, the choice of cancer unit risk ranged between 7.1 × 10–3 and 2.5 × 10–2 at 1 ppm (2.2 × 10–6 to 7.8 × 10–6 at 1 µg/m3 of benzene in air), depending on which exposure measurements were used, ie, Crump and Allen (1984) or Paustenbach et al (1993)(as both cited in US EPA, 2000). US EPA (2000) does not present a single risk estimate but rather presents this risk range to indicate the scientific uncertainty in the risk estimates.

In the UK, the Expert Panel on Air Quality Standards (DEFRA and EA, 2003 citing DoE, 1994) used the data of Rinsky et al (1981; 1987; both cited in DEFRA and EA, 2003) and Wong et al (1983; 1987; both cited in DEFRA and EA, 2003) to form the basis of their air quality guideline. However, the US EPA (2000) concluded that there were some limitations in the cited Wong (1983; 1987) study that precluded its use in quantitative risk assessment. Specifically, the limitations were imprecise historical industrial hygiene data; unusual distribution of leukaemia cell types, ie, there were no acute cases of myelogenous leukaemia out of seven leukaemia cases; and possible exposure of comparison subjects to potentially carcinogenic solvents other than benzene.

Insufficient information is provided in Environment Canada (2005) and Baars et al (2001) to critically evaluate the derivation of Canadian and Dutch values, which are higher than the majority of values in Table 47.

Table 46: Summary of inhalation reference health standards for threshold effects of benzene used by different international agencies

Jurisdiction

Tolerable daily intake (mg/m3)

Key study1

Critical effect1

Basis of value1

Reference

The Netherlands

0.03

Not stated

Induction of blood toxicity

Vermeire (1993). Derived from LOAEL of 30 mg/m3 for haematological changes in experimental animals using an uncertainty factor of 1000

Baars et al (2001)

US ATSDR – chronic duration MRL and US EPA

0.03

Rothman et al (1996), human occupational inhalation study

Decreased lymphocyte count

BMCLadj of 8.2 mg/m3 converted from BMCL = 7.2 ppm, 8-h TWA. Assuming 25ºC and 760 mm Hg, BMCL (mg/m3) = 7.2 ppm x MW (78.11)/24.45 = 23.0 mg/m3. BMCLadj = 23.0 mg/m3 x 10 m3/20 m3 x 5 days/7days = 8.2 mg/m3. (The BMC was based on a benchmark response of one standard deviation change from the control mean.)

A UF of 300 was applied, which comprised a factor of 3 for extrapolation from an effect to a no-effect level, a factor of 10 to account for human variability, a factor of 3 for sub-chromic to chronic exposure, and a factor of 3 to account for database deficiencies. The resultant UF (270) is rounded up to 300.

ATSDR (2007)
US EPA (2003)

1 As reported in the reference cited in the reference column.

Table 47: Summary of inhalation reference health standards for non-threshold effects of benzene used by different international agencies

Jurisdiction

Acceptable risk level1

Guideline value2 (μg/m3)

Risk-specific dose (μg/kg bw/day

Cancer slope factor (per mg/kg bw/day)

Key study3

Critical effects3

Basis of value2

Reference

New Zealand

10–5

–

0.34

0.029

US EPA (1995)4

Not stated

US EPA (1995)4

MfE (1999)

10–5

–

0.34

0.029

US EPA5

Not stated

US EPA5

MfE (1997)

New Zealand air quality guidelines

Not stated

10

2.9

 

Not stated

Not stated

Stated to be a combination of European Council and UK approaches

An annual average of 3.6 μg/m3 is the stated target for 2010

MfE (2002)

WHO – air quality guidelines

10–5

1.7

0.48

0.021

Crump (1994)

Leukaemia, occupational inhalation exposure “Pliofilm cohort”

Geometric mean of the range of risk estimates reported by Crump (1994) (6 x 10–6 for an exposure of 1 μg/m3)

WHO (2000)

Expert Panel on Air Quality Standards (EPAQS)

10–5

3.2

0.91

0.011

Rinsky et al (1987), Wong (1987a; 1987b)

Leukaemia, occupational inhalation exposure “Pliofilm cohort” plus others

Risk of leukaemia in workers was not detectable with lifetime exposure around 500 ppb (1600 μg/m3), and the risk would be too small to detect in any study. This concentration was divided by 10 to extrapolate occupational exposure to chronological lifetime exposure

An uncertainty factor of 50 (comprising 10 for human variability, and 5 to account for the need to keep exposures of genotoxic compounds as low as practicable) was applied

DEFRA and EA (2003) citing DoE (1994)

UK

10–5

 

0.91

0.011

DoE (1994)

Leukaemia, occupational inhalation exposure “Pliofilm cohort”

DoE (1994)

Estimates from Crump (1994) were said to indicate the risks from inhalation of benzene at 3.2 μg/m3 were “in the order of 10–5”, and therefore supported the EPAQS value

DEFRA and EA (2003)

US

10–6
[10–5]

 

0.037–0.13
[0.37–1.3]

0.0077–0.027

Rinsky et al (1987a; 1987b)

Leukaemia, occupational inhalation exposure “Pliofilm cohort”

Low-dose linearity utilising maximum likelihood estimates (Crump, 1992; 1994).

US EPA (2003)

Canada

10–6
[10–5]

 

0.3
[3]

 

Not stated

Not stated

Health Canada derived a 5% lifetime cancer risk at 15 mg/m3, but no further details were provided for the actual derivation

Environment Canada (2005)

The Netherlands – proposed6

10–4
[10–5]

 

20
[2]

 

EU 1998

Leukaemia, occupational inhalation exposure “Pliofilm cohort”

Stated to be derived by linear extrapolation from the “Pliofilm cohort”

Baars et al (2001)

1 Where the acceptable risk level for a given jurisdiction is not 10–5, the risk-specific dose for a risk of 10–5 is shown in square brackets.
2 Where a guideline value is provided, risk-specific doses have been determined assuming inhalation of 20 m3/day by a 70-kg adult.
3 As reported in the reference cited in the reference column.
4 US EPA (1995) is stated as the source of the slope factor in the text of MfE (1999), but the reference is not provided in the reference list.
5 No specific reference is given, presumably it is US EPA (1995) as per MfE (1999); however, as stated above, the actual reference is not provided in the reference list.
6 Proposed value; this value is yet to be officially adopted.

Dermal absorption

Dermal absorption of volatile organics such as benzene is especially difficult to assess, because most studies have involved occluding (covering) the skin, which may give artificially high skin absorption values, since these compounds would also be expected to volatilise from the skin. Skin absorption in studies on non-occluded benzene is typically considered to be less than 1% (US EPA, 2003) – one in vivo study using human volunteers found that on average 0.023% of the benzene applied to skin was absorbed while the remainder quickly volatilised (Franz, 1984). Skowronski et al (1988) found that soil adsorption decreased the bioavailability of benzene during occluded studies on absorption. A value of 0.05% has been proposed for use by US EPA Region 3 based on these two studies (US EPA, 1995).

Other routes of exposure – background exposure

Inhalation accounts for more than 99% of the exposure by the general population, whereas intake from food and water is minimal. Sources of benzene in ambient air include cigarette smoke, combustion and evaporation of benzene-containing petrol (up to 1% benzene), petrochemical industries, and combustion processes. Smoking has been identified as the single most important source of benzene exposure for the estimated 40 million US smokers (ATSDR, 2007). However, as benzene is a genotoxic carcinogen, all exposure needs to be reduced to as low as reasonably practicable, thus background exposure is irrelevant in developing soil guideline values.

2.9.4   Summary of effects

Benzene is toxic by all routes of administration. Benzene exposure affects the central nervous system (CNS) and haematopoietic system and may affect the immune system. Death due to acute benzene exposure has been attributed to asphyxiation, respiratory arrest, CNS depression, or cardiac dysrhythmia. Acute benzene exposure results in classic symptoms of CNS depression such as dizziness, ataxia, and confusion. These effects are believed to be caused by benzene itself rather than its metabolites, because the onset of CNS effects at extremely high doses is too rapid for metabolism to have occurred. Acute exposure at high doses may also result in skin and eye irritation. The most characteristic systemic effect resulting from intermediate and chronic benzene exposure is arrested development of blood cells.

Benzene can cause serious haematologic toxicity such as anaemia, leukopenia, thrombocytopenia, or pancytopenia after chronic exposure. Haematotoxicity and immunotoxicity have been consistently reported to be the most sensitive indicators of non-cancer toxicity in both humans and experimental animals. Leukocytopenia has been consistently shown to be a more sensitive indicator of benzene toxicity in experimental animal systems than anaemia, and lymphocytopenia has been shown to be an even more sensitive indicator of benzene toxicity than overall leukocytopenia. Chronic low-level exposures have also been associated with peripheral nervous system effects. Benzene-induced aplastic anaemia is generally caused by chronic exposure at relatively high doses (ATSDR, 2007). Aplastic anaemia can progress to leukaemia, whereas depression of blood elements (eg, leukopenia, pancytopenia) may be reversible.

Table 48 summarises effects observed in humans, primarily as a result of occupational exposure to benzene (ATSDR, 2007).

Table 48: Summary of the health effects of benzene in humans

Dose (mg/kg/day)1

Exposure

Effects

>126

Acute

Death

24–34 [150– 210 ppm]

Intermediate occupational exposure
(4 months – 1 year)

Pancytopenia

24 [150 ppm]

Chronic occupational exposure
(4 months – 15 years)

Pancytopenia

9 [60 ppm]

Acute, occupational exposure
(1–21 days, 2.5–8 hours/day)

Skin and mucous tissue irritation, leukopenia (reduced leukocyte count), drowsiness, dizziness, headache – symptoms reversible after removal from exposure

4–4.7 [24–29 ppm]

Chronic occupational exposure
(3.5–19 years)

Aplastic anaemia, leukaemia, pancytopenia

1.8–12

Chronic occupational (0.5–25 years)

Anaemia

0.29

Chronic

Reduced white blood cell count, based on route-to-route extrapolation from occupational inhalation study

0.05

Chronic occupational (18 months)

Lowest reported concentration resulting in cancer (leukaemia) as a result of occupational exposure to benzene

1 Where ppm is given in square brackets, dose has been calculated as follows. Ppm to mg/m3: assuming 25°C and 760 mm Hg, mg/m3) = ppm x MW (78.11)/24.45. Accounting for occupational exposure mg/m3adj = mg/m3 x 10 m3 (volume of air inhaled during workplace exposure)/20 m3 x 5 days/7 days. Concentration to dose = mg/m3adj x 20 m3/70 kg.

2.9.5   Weight of evidence

  • Benzene is considered to be a known human carcinogen by IARC (1987) and the US EPA (2000). Specifically, there is sufficient inhalation data from humans supported by animal evidence, including the oral studies in animals, to demonstrate carcinogenicity in humans. The human cancer predominantly induced by inhalation exposure to benzene is acute non-lymphocytic leukaemia.
  • Other neoplastic conditions associated with an increased risk in humans are haematologic neoplasms, blood disorders such as preleukaemia and aplastic anaemia, Hodgkin’s lymphoma, and myelodysplastic syndrome (WHO, 1993; ATSDR, 2007).
  • Animal studies also support data indicating that exposure to benzene increases the risk of cancer in multiple species at multiple organ sites (WHO, 1993; US EPA, 2002; ATSDR, 2007).
  • Multiple mechanistic pathways leading to cancer and, in particular, to the development of leukaemia, from exposure to benzene have been proposed, with effects observed suggested to be due to interactions of the metabolites of benzene with DNA of stem cells (Whysner, 2000; ATSDR, 2007).
  • Benzene does not induce point mutations in animal bioassays (WHO, 1993; US EPA, 2002; ATSDR, 2007).
  • In vivo and in vitro data from both humans and animals indicates that benzene and/or its metabolites are genotoxic. Chromosomal aberrations in peripheral lymphocytes and bone marrow cells are the predominant effects seen in humans (Whysner, 2000; US EPA, 2002; ATSDR, 2007).
  • Due to the evidence of carcinogenicity and genotoxicity, benzene is considered a non-threshold substance.

2.9.6   Recommendations for toxicological intake values

As benzene is considered to be a genotoxic carcinogen, and therefore is a non-threshold contaminant, risk-specific doses are proposed. As inhalation is the significant exposure pathway, it is discussed first.

US EPA (2000) presents an estimated risk range to indicate the scientific uncertainty in the risk estimates for benzene exposure; however, from a regulatory perspective it is easier to have a single value to ascertain whether exposure results in exceedances or not. Therefore an inhalation risk-specific dose of 0.48 μg/kg bw/day (slope factor of 0.021 per mg/kg bw/day) is recommended (Table 49). This is the WHO (2000) air quality value and is the geometric mean of 94 risk estimates from Crump (1994). This approach appears to provide the most robust value, maximising the use of available data. An alternative would be to take the midpoint of the range of risk estimates of the US EPA, which are also derived from Crump (1994). However, this may not reflect the actual central tendency of the full range of risk estimates.

Due to the lack of oral carcinogenicity data in humans, as well as the lack of a well-demonstrated and reproducible animal model for leukaemia from benzene exposure, oral data has typically been extrapolated from inhalation data. Therefore, the oral risk-specific dose recommended is one extrapolated from the WHO (2000) air quality guideline using the default US EPA assumptions; a 70-kg adult inhales 20 m3/day and 50% of benzene is absorbed via inhalation and 100% via ingestion; this should be revised if different parameters are adopted in the general soil guideline value methodology. This gives rise to an oral risk-specific dose of 0.24 μg/kg bw/day (slope factor 0.042 per mg/kg bw/day). This approach is taken to maintain consistency between the recommended oral and inhalation intake values.

Dermal absorption is less than 1% and therefore would make a negligible contribution to oral risk estimation. A value of 0.05%, as proposed by US EPA region III (US EPA, 1995), could be used to provide a greater refinement for risk estimation. This value is the average of available studies on non-occluded dermal absorption of benzene.

The recommended toxicological criteria for benzene are summarised in Table 49.

Table 49: Recommended toxicological criteria for benzene

Parameter

Value

Basis

Contaminant status

Non-threshold

–

Oral

 

 

Risk-specific dose (μg/kg bw/day)
Slope factor (per mg/kg bw/day)

0.24
0.042

Extrapolated from inhalation dose (see below), assuming inhalation of 20 m3/day by a 70-kg adult, 50% absorption via inhalation and 100% via ingestion

Inhalation

 

 

Risk-specific dose (μg/kg bw/day)
Slope factor (per mg/kg bw/day)

0.48
0.021

Derived from WHO air quality guideline of 1.7 μg/m3 for a 10–5 risk (WHO, 2000), which in turn is derived from the geometric mean 6x10–6 of risk estimates from Crump (1994)

Skin absorption factor

0.0005

Dermal absorption is expected to be <1%

Background exposure (μg/kg bw/day)

NA

Exposure to non-threshold contaminants from all sources should be as low as reasonably practicable