Registration Dossier

Administrative data

Description of key information

Schnatter et al 2020 derived an occupational exposure limit for benzene using quality assessed data. The epidemiological carcinogenicity data was assessed in the initial scoping phase of this work, however, the MOA described by North et al., 2020a indicated genetic toxicity and/or haematotoxicity were key events preceding carcinogenic outcomes, therefore the carcinogenicity endpoint was not used to inform on a point of departure in the OEL derivation. 

 

Seventy-seven genotoxicity and thirty six haematotoxicity studies in workers were scored for study quality with an adapted tool based on that of Vlaanderen et al., 2008 (Environ Health. Perspect. 116 1700−5). Lowest and No- Adverse Effect Concentrations (LOAECs and NOAECs) were derived from the highest quality studies (i.e. those ranked in the top tertile or top half) and further assessed as being “more certain” or “less certain”. Several sensitivity analyses were conducted to assess whether alternative “high quality” constructs affected conclusions. The lowest haematotoxicity LOAECs showed effects near 2 ppm (8 h TWA), and no effects at 0.59 ppm. For genotoxicity, studies also showed effects near 2 ppm and showed no effects at about 0.69 ppm. Several sensitivity analyses supported these observations. These data define a benzene LOAEC of 2 ppm (8 h TWA) and a NOAEC of 0.5 ppm (8 h TWA). Allowing for possible subclinical effects in bone marrow not apparent in studies of peripheral blood endpoints, an OEL of 0.25 ppm (8 h TWA) is proposed. 

 

In a paper on risk models for benzene leukaemia by North et al., 2020b, a number of publications related to characterisation of benzene leukaemogenic effects in humans have been outlined. Occupational exposure to benzene at levels of ≥10 ppm has been associated with increased risk of acute myeloid leukaemia (AML). In conclusion, an OEL of 0.25 ppm, would prevent the initiating events – haematotoxicity and genotoxicity leading to adverse outcomes: myelodysplastic syndrome (MDS) and AML.  

 

Additional information

Benzene exposure can result in haematotoxic (including immunotoxic), genotoxic and carcinogenic (i.e. leukaemogenic) effects. Early studies (e.g. Aksoy et al., 1971; Goldwater, 1941; Greenburg et al., 1939) found that high benzene exposure had a suppressive effect on peripheral blood cell counts. Lower exposures (e.g. < 10 ppm) produce only mild reduction in blood cell counts through changes in marrow function. This reduction in circulating blood cells is likely one of the earliest clinical effects of benzene exposure (DECOS, 2014; North et al., 2020a).  

 

The occurrence of cytogenetic damage in individuals exposed to high levels of benzene has been recognised from the 1960s. Available experimental data suggests that benzene genotoxicity is related to chromosomal damage rather than gene mutation. 

 

An extensive literature review of epidemiological studies of benzene exposure reveals that most early studies (before 1980), focused on higher exposures, with benzene concentrations often exceeding 50 ppm in air. More recent studies have examined effects of lower exposures, and a significant number of these studies have also reported a depression of peripheral blood counts, or changes in chromosomal damage, and micronucleus formation, often in the absence of aplastic anaemia or pancytopenia. Especially for genotoxicity, study groups exposed to lower concentrations often come from environments with co-exposure to chemicals, making it more difficult to attribute the effects to benzene or other chemical exposure such as polycyclic aromatic hydrocarbons (PAHs).  

  

Repeated Dose – Haematotoxicity 

 

Studies used for weight of evidence to derive an OEL (Schnatter et al 2020) 

  

Qu et al., 2003 saw significant trends for decreases in white blood cells (WBC), red blood cells (RBC) and neutrophils in 130 shoe, glue and sporting goods Chinese workers versus 51 age-, sex- and smoking-matched controls in Tianjin, China. Workers were exposed to 4-week average concentrations of 0.08 to 54.5 ppm, and lifetime cumulative exposures of 6.1 to 632 ppm-years. The most sensitive parameter from this study was reduction of the neutrophil count. For blood count reductions, the 4-week average exposure is the most relevant metric, keeping with the preference for the quantification of recent exposure. A significant dose-response for neutrophils (as well as other blood elements) was found for exposure groups 0-5, 5-15, and 15+ ppm. The 0-5 ppm group showed a 12% reduction in neutrophil count that was of greater magnitude than other blood elements. The mean concentration of benzene over four weeks for this low dose group was 2.26 ppm (1.35 SD). Thus 2.26 ppm can be regarded as a LOAEC from this group. Other analyses that were conducted in Qu et al., 2002, 2003 concerning a 0 – 0.5 ppm group, were not used because of an inappropriate control group. Specifically, a group of 16 female workers exposed to concentrations < 0.5 ppm (i.e. average 4-week concentration = 0.14 ppm) had significant 15% decrements in WBC and RBC parameters and a 19% decrement in neutrophils which remained after control for confounders. However, this analysis is flawed since female subjects are being compared to males and females (females have a priori lower blood counts), and none of these workers smoked, while the comparison group did. Since smoking raises neutrophil counts, the comparison (for this small subgroup only) is inappropriate and confounding by gender and smoking was not (and indeed, cannot be) accounted for, despite the attempt to control for it. Thus, the LOAEC for this study is 2.26 ppm which is the mean exposure of the 0-5 ppm group, which contains sufficient numbers and enough heterogeneity by gender and smoking to allow for control of these confounders. The report does not allow for the calculation of a NOAEC. However, the Qu neutrophil data was modelled using the PROAST software which calculates the benchmark dose lower limit. Using a 22% response factor which is biologically justified a BMDL of 1.7 ppm was derived and can serve as a NOAEC. (Cembrowski G et al 2016, North et al 2019)

[Cembrowski G, Topping K , Versluys et al 2016 The use of serial outpatient complete blood count (CBC) results to derive biologic variation: A new tool to gauge the acceptability of hematology testing. Int J Lab Hematol 38 (2) 111-8 North, C.M., Rooseboom, M., Dalzell, A., 2019. Benchmark dose modeling for hematologic effects of occupational benzene exposure, in: Abstracts of the 55th Congress of the European Societies of Toxicology (EUROTOX 2019). Presented at the EUROTOX 2019, Elsevier, p. S263. Qu, Q., Shore, R., Li, G., Jin, X., Chen, L.C., Cohen, B., Melikian, A.A., Eastmond, D., Rappaport, S., Li, H., Rupa, D., Waidyanatha, S., Yin, S., Yan, H., Meng, M., Winnik, W., Kwok, E.S.C., Li, Y., Mu, R., Xu, B., Zhang, X., Li, K.,. Validation and evaluation of biomarkers in workers exposed to benzene in China. Res Rep Health Eff Inst 1–72; discussion 2051 73-87.]  

  

 

Swaen et al. 2010 analysed haematological data from an exposed group of 701 petrochemical workers (8,532 blood samples) and a non-exposed group of 1,059 production workers (12,173 blood samples). A job exposure matrix was used to assign employees an exposure value at the time that blood was drawn. Smoking and age were controlled in regression models. A stratification of the exposed population into three subgroups (<0.5 ppm, 0.5-1 ppm and >1 ppm) resulted in no declines in nearly all assessed blood indices (i.e. haemoglobin, haematocrit, WBC, lymphocytes, neutrophils, basophils and monocytes). Only a reduction in eosinophils was observed for the 0.5 – 1 ppm and > 1 ppm groups. Based on this single finding among several blood indices, the LOAEC lies within the 0.5 – 1 ppm (midpoint = 0.75 ppm) group.     

 

While effects on eosinophils could be an early marker for benzene-induced haematotoxicity, Swaen indicates these changes as small and not clinically significant. Also, many blood indices were studied and a few would be expected to be significant by chance. Thus, a strong interpretation of this finding is not justified, and it is flagged as a less certain LOAEC.     

 

The 0 – 0.5 ppm (midpoint = 0.25 ppm) group did not show effects for any blood cell type and can be regarded as the NOAEC.  This agrees closely with the mean exposure for all blood tests, which was 0.22 ppm (range 0.01 – 1.85 ppm) and could be justified as a NOAEC if the eosinophil finding is disregarded as a finding due to multiple comparisons. Thus, the NOAEC in this study is derived two different ways and the two values are very close, making this finding the more certain derivation compared to the LOAEC.     

[Swaen, G.M.H., van Amelsvoort, L., Twisk, J.J., Verstraeten, E., Slootweg, R., Collins, J.J., Burns, C.J., 2010. Low level occupational benzene exposure and hematological parameters. Chem. Biol. Interact. 184, 94–100. https://doi.org/10.1016/j.cbi.2010.01.007]  

 

Schnatter et al., 2010 examined several blood indices among a large population of 928 shoe and rubber workers in Shanghai China, and 73 internal controls not exposed to benzene. Exposure ranged from 0.02 ppm to 273 ppm, with mean exposures of 1.8 ppm among males and 3.4 ppm among females. These were derived from over 2900 individual monitoring readings. Several confounders (age, gender, BMI, smoking, alcohol use and polymorphisms of key genes) were controlled in regression models. For continuous blood parameters, Schnatter et al. did not group employees into categories, but instead used change point regression techniques to delineate the lowest concentration in which the slope for benzene’s effect differed from 0, a technique that lends itself more readily toward identifying lowest effect levels. Of 12 parameters studied by Schnatter et al. 2010, 10 parameters showed an effect, mostly for benzene exposures greater than 10 ppm. The lowest threshold for a clear signal was 7.77 ppm for reductions in neutrophil counts, which can be regarded as the LOAEC. Schnatter et al., also examined out-of-range readings for some blood parameters. The strongest effect was seen for mean corpuscular volume (MCV), with exposures above 10 ppm having an effect. For anemia, Schnatter et al., indicates significantly elevated risks in the <1 ppm and >10 ppm groups, which is difficult to interpret due to non-monotonicity.  This result was present in males but not females.  Due to this internal inconsistency and the lack of a clear dose-response, 7.8 ppm is regarded as the LOAEC in this study. The lower confidence interval for this neutrophil changepoint (2.9 ppm) can be regarded as an approximate NOAEC.    

[Schnatter AR Kerzic PJ, Zhou Y, Chen M, Nicolich MJ, Lavelle K, Armstrong TW, Bird MG, Lin L, Fu H, Irons RD. 2010 Peripheral blood effects in benzene-exposed workers. Chem Biol Interact. 2010 Mar 19;184(1-2):174-81. doi: 10.1016/j.cbi.2009.12.020. Epub 2009 Dec 23]  

  

Ward et al (1996) used a case control design, selecting 78 cases of leukopenia who showed a blood count reading below the first percentile of pre-placement blood test readings. 5637 matched controls, having a normal blood test within six months of the case’s blood test date. Exposure over the preceding 30, 60, and 180 days was calculated for each case and control. Age, gender, year of employment were controlled. Results were analysed via conditional logistic regression A monotonic increase in risk of low white counts was seen with benzene exposure over the preceding 180 days. The best fitting model was for exposure summed over the previous 180 days. The authors superimposed confidence intervals along the plot of the OR for select values of exposures over the range of exposures studied. These confidence intervals indicated that the interval at 2.2 ppm (400 ppm-days) included unity (i.e. there was no significant risk), while the interval at 7.2 ppm (1300 ppm-days) excluded unity, indicating a significant risk of leukopenia in 180 days prior to blood test date. These two values are taken as the NOAEC and LOAEC, respectively.  

[Ward, E., Hornung, R., Morris, J., Rinsky, R., Wild, D., Halperin, W., Guthrie, W., 1996. Risk of low red or white blood cell count related to estimated benzene exposure in a rubberworker cohort (1940- 3 1975). Am. J. Ind. Med. 29, 247–257. https://doi.org/10.1002/(SICI)1097-2132 0274(199603)29:3]  

  

Collins et al., (1991) studied several haematological parameters among 468 Monsanto chemical plant workers (200 exposed) between 1982-89. A total of 1112 TWA-8-hour monitoring results were used to estimate exposures; however, no detail is given on the monitoring method. Geometric mean benzene concentrations were between 0.01 and 1.4 ppm (95% were between 0.01 and 0.1 ppm, Table 2) for different jobs. A job exposure matrix was used to assign exposures at the time of phlebotomy. Arithmetic mean exposures were not provided but are likely higher assuming the exposure distribution shows usual right-skewness. As such, results are likely conservative based on geometric means. Age, sex, race, smoking, alcohol use and exercise were controlled in the analyses. Most blood elements showed no effects when examining several metrics of benzene exposure. An isolated effect on MCV was noted. While this could indicate MCV as the most sensitive indicator of benzene exposure in the study, the MCV changes were within clinically normal ranges, MCV can be a non-specific effect of general stress, and several other parameters showed no effect (RBCs, WBCs, HgB, platelets, and WBC differentials). Collins reports only one job assignment above 0.19 ppm (Table 2), which is taken as the maximum reliable exposure that showed no effects, or the NOAEC.  

[Collins, J.J., Conner, P., Friedlander, B.R., Easterday, P.A., Nair, R.S., Braun, J., 1991. A study of the hematologic effects of chronic low-level exposure to benzene. J Occup Med 33, 619–626.]  

Rothman et al (1996) studied 44 workers heavily exposed to benzene. Workers were selected from rubber paint and tape factories, while 44 age and gender matched controls were selected from a factory in the same district that manufactured sewing machines and an administrative facility. Workers wore passive monitoring badges for 1-2 weeks before phlebotomy. Urinary metabolites were collected via a spot urine sample.  Age sex smoking, alcohol use and the Quetelet index was controlled in all analyses. Subjects were initially analysed according to whether they fell above or below the median measurement of 31 ppm. A subgroup (n=11) never exposed above 31 ppm (median exposure = 7.6 ppm) was also studied. Nearly all blood elements were reduced for those exposed above and below the median value of 31 ppm. However, for the subgroup of workers never exposed above 31 ppm, only the absolute lymphocyte count was reduced. Thus, the LOAEC for this study is 7.6 ppm.   

[Rothman, N., Li, G.L., Dosemeci, M., Bechtold, W.E., Marti, G.E., Wang, Y.Z., Linet, M., Xi, L.Q., Lu, W., Smith, M.T., Titenko-Holland, N., Zhang, L.P., Blot, W., Yin, S.N., Hayes, R.B., 1996. Hematotoxicity among Chinese workers heavily exposed to benzene. Am. J. Ind. Med. 29, 236–246. https://doi.org/10.1002/(SICI)1097-0274(199603)29:3]  

 

The Lan et al., 2004 study compared 250 subjects from two shoe factories with 140 unexposed controls. Subjects were categorized into four groups by mean benzene levels measured during the month before phlebotomy (control; <1 ppm; 1 to <10 ppm; and ≥10 ppm). Controls were from a clothing manufacturing factory, but were frequency matched on demographic variables. Models  were adjusted for age, sex, smoking, alcohol use, BMI and recent infections. The authors reported significant depressions in WBC, granulocytes, lymphocytes, CD4 T cells, B cells, monocytes, and platelets at benzene concentrations less than 1 ppm based on a group of 109 subjects categorised as exposed to < 1 ppm. The mean exposure in this group was 0.57 ppm. While a level of 0.57 ppm has been mentioned as showing an effect (i.e. a LOAEC) on blood cell elements in Lan et al. 2004, this finding comes with several caveats and is too much uncertain to be regarded as a clear LOAEC.  For example, over half of the workers categorized as exposed to < 1 ppm had a past cumulative exposure of over 40 ppm-years, although the average length of employment was only 6.2 years, and many workers were also exposed to a range of other chemicals, including toluene, pentane, ethyl benzene, xylene, trichloroethane and heptane. In addition, this study shows no difference between the < 1 ppm and 1-10 ppm groups, bringing into question whether the control group is appropriate. The non-monotonic risk is difficult to interpret. In addition, if the urinary benzene concentrations found in the <1ppm group are taken as a more precise indication of exposure, DFG values can be used to calculate a potential airborne LOAEC. DFG air/urine values suggest that the Lan et al., 2004 <1ppm group is really exposed to an approximate mean of 2.2 ppm rather than the reported air exposure of 0.57 ppm. It is not unrealistic for the working conditions applied that dermal exposure contributed to the overall exposure. It is proposed that this value of 2.2 ppm, which could account for previous criticisms of higher exposures among the group that were unaccounted for, be used as a preferred LOAECfor this study.   

[Lan, Q., Zhang, L., Li, G., Vermeulen, R., Weinberg, R.S., Dosemeci, M., Rappaport, S.M., Shen, M., Alter, B.P., Wu, Y., Kopp, W., Waidyanatha, S., Rabkin, C., Guo, W., Chanock, S., Hayes, R.B., Linet, M., Kim, S., Yin, S., Rothman, N., Smith, M.T., 2004. Hematotoxicity in Workers Exposed to Low Levels of Benzene. Science 306, 1774–1776. 1970   

  

Koh et al., (2015) studied 10,702 Korean benzene workers in various industries and categorized exposures as 0.01 – 0.1 ppm (with a group mean of 0.04 ppm), 0.1 – 0.5 ppm (group mean = 0.21 ppm), and >0.5 ppm (group mean = 2.6 ppm). The group mean levels have been obtained from personal communication with Dr. Koh. A linear mixed-effects model was used, treating worker as a random variable with repeated inter-correlated blood readings, while controlling for age and sex. In general, Koh et al. used a tiered approach for information quality and showed that most blood parameters are unaffected by benzene, even when using the highest standard for information quality. Koh et al., (2015) saw no consistent trends in white cell counts nor platelets. However, Koh does report a relationship between clinically defined anemia (reduced red cells) in the highest exposed group - above 0.5 ppm (group mean = 2.6 ppm) for the highest quality exposure information. This result is specific to males. The reason for positive RBC-based findings only among males is uncertain, though males could potentially be more sensitive than females, and RBCs could be an early indicator of benzene-induced haematotoxicity. Koh et al. mention study limitations. For example, in low-level exposure environments, benzene inhaled from cigarette smoking may be important and the study did not account for any potential impact of this. Furthermore, the authors indicate that workers could have been co-exposed to other chemicals with haematological effects like formaldehyde and the blood tests were conducted at more than 150 hospitals using different counting machines and devices, which may lead to some systematic errors in cell counts. Because of multiple potential interpretations of the Koh et al. (2015) results, it is flagged as a less definitive LOAEC.    

Koh et al. Interpret their results as showing potential effects as low as 0.5 ppm. In contrast, none of the lower dose groups (0.01 - .1 ppm and 0.1 - 0.5 ppm) showed consistent effects for red blood cells, nor white blood cells and subtypes. Thus, the LOAEC of 2.6 ppm and NOAEC of 0.21 ppm can be derived form this study.     

[Koh, D.-H., Jeon, H.-K., Lee, S.-G., Ryu, H.-W., 2015. The relationship between low-level benzene exposure and blood cell counts in Korean workers. Occup Environ Med 72, 421–427.https://doi.org/10.1136/oemed-2014-102227 ]   

 

Pesatori et al., (2009) examined 153 Bulgarian petrochemical workers exposed to benzene from 0.01 to 23.9 ppm versus 84 unexposed subjects. Exposure groups were defined as controls, <1 ppm (mean = 0.3 ppm) and >1 ppm (mean = 4.9 ppm). Measurements were taken the day before phlebotomy. Age, gender, smoking, and toluene exposure were controlled in statistical analyses.  Most blood parameters were not reduced when comparing these two groups versus the controls. However, a decline in eosinophils was noted, although the decline was only observed for smokers. Since this is an unexplained finding and not consistent with other blood cell types, it was not selected as a LOAEC; a NOAEC is appropriate. A weighted average of mean exposures in this study from Table 1, (106 * 0.3 + 47 * 4.9) is 1.7 ppm, which can serve as a NOAEC.    

[Pesatori, A.C., Garte, S., Popov, T., Georgieva, T., Panev, T., Bonzini, M., Consonni, D., Carugno, M., Goldstein, B.D., Taioli, E., Fontana, V., Stagi, E., Bertazzi, P.A., Merlo, D.F., 2009. Early effects of low benzene exposure on blood cell counts in Bulgarian petrochemical workers. Med Lav 100, 83–90.]  

 

Zhang et al., 2016 studied peripheral blood counts and micronuclei frequency in 317 shoe factory workers and 102 controls recruited from local bank and school office personnel. No matching was described in the study. Air benzene exposure was assessed three times per day for 15 minutes by area monitoring in the breathing zone of shoe factory workers, while no description of air benzene measures for controls was reported. Median air benzene levels were reported as 1.57 ppm (sewing), 2.60 ppm (moulding), and 1.79 ppm (packing). Ambient concentrations ranged from 0.80 – 12.09 ppm, with a median value of 1.60 ppm. However, a weighted median based on the number of subjects can be calculated to be 2.11 ppm. Glue brushing (manually dipping brushes into open containers of glue) was suspected as the primary source of benzene inhalation, though this practice is also likely to lead to dermal exposure. Benzene from smoking was also estimated separately, thus some controls had non-zero benzene exposure. 6 Average white blood cell counts were significantly decreased for the aggregate exposed group versus controls, and for cumulative exposures of 10.72 ppm-years or more. Analyses were not performed by concentration of recent exposure. The fraction of subjects with abnormal white blood cell count (defined by the authors as < 4.3 x 109 /L, the 5th percentile of the control distribution) was significantly reduced for all exposed workers versus controls but was not significantly increased by cumulative exposure category. Dose response assessment by benchmark response modelling was performed by cumulative benzene exposure to both micronuclei frequency and white blood cell counts. No confounders were controlled, though age was modelled as an effect modifier. The authors report dose response modelling for categorized exposure groups for dichotomous outcomes (using BMDS), substituting the median value for the average exposure, and do not address what measure was used for variability in the model. BMD modelling of summary data depend on use of an average, variation statistic (standard deviation or standard error), and count of subjects. The authors decision makes the modelling of dichotomous outcomes unreliable as a dose response modelling exercise, as medians and means can be substantially different. As such, reported values should be rejected as unreliable. However, the authors also performed benchmark dose response modelling using continuous measures of cumulative benzene exposure and micronuclei or white blood cell count changes (using PROAST with each individual’s estimated exposure). BMD estimates for reduced white blood cell count were generally higher than the values for micronuclei. However, the authors did not examine WBC subsets, such as neutrophils. The study can be interpreted as reporting a LOAEC of >2.11 ppm for changes in white blood cell count and micronuclei frequency. The use of median rather than mean exposure, area monitoring to estimate worker exposure, and lack of a measure for total benzene exposure (i.e., biomonitoring information) in the presence of work practices expected to result in dermal benzene exposure, are anticipated to underestimate individual exposure.  

[Zhang GH, Ji BQ, Li Y, Zheng GQ, Ye LL, Hao YH, Ren JC, Zhou LF, Xu XW, Zhu Y, Xia ZL (2016) Benchmark Doses Based on Abnormality of WBC or Micronucleus Frequency in Benzene-Exposed Chinese Workers. J Occup Environ Med 58: e39-44.]  

  

Collins et al. (1997) used periodically collected exposure and blood count readings among 387 workers employed at a chemical plant in the U.S. Exposures averaged 0.55 ppm. Individual readings covered a range from 0.01 to 87.69 ppm, although average annual readings showed a range between 0.07 and 1.85 ppm from 1980 through 1993. 12 of the 387 workers had an initial benzene exposure reading and blood test before working in a benzene-exposed job, and thus were selected to be examined in a longitudinal analysis that examined exposure and blood counts specific to at least six points in time. Other cross-sectional analyses were conducted for the larger group of 387 workers. The authors examined several blood cell types but, since results were similar, focused on lymphocytopenia, which they thought was the most sensitive outcome based on previous literature. Lymphopenia was defined as 1.05 and 1.40 thousands per cubic millimeter for non-smokers and smokers, respectively. No lymphopenia was found for the individual analyses, and slight positive slopes were found over time versus a negative slope that was expected for benzene exposure. No relationships with benzene exposure level were reported for the cross-sectional analyses. Effects of age, sex, and smoking were controlled. This study suggests no effects on blood counts for an average exposure of 0.55 ppm, which can serve as the NOAEC.  

[Collins, J.J., Ireland, B.K., Easterday, P.A., Nair, R.S., Braun, J., 1997. Evaluation of lymphopenia among workers with low-level benzene exposure and the utility of routine data collection. J. Occup. Environ. Med. 39, 232–237.]  

  

Tsai et al., (2004) studied 1200 employees in a benzene monitoring program and compared their 7 blood values to 3227 employees not exposed to benzene. Who had at least two blood tests during the period 1977 to 2003. Six haematologic measures were analysed, with all performed by a single laboratory. The data were analysed with respect to readings below a certain cut-off; the cut-off was specified by the laboratory (for WBCs a cut-off of <4000/mm3 was used) and a comparison of mean values between exposed and unexposed workers. Age, sex, race, and smoking were adjusted for. There were no significant differences for the proportion of tests below cut-offs between the two groups. There was only one significant difference in means (MCV), but it was in a direction opposite to that expected if benzene had an effect. Also, the MCV difference was not significant after adjusting for multiple comparisons. The benzene surveillance group had a mean benzene exposure of 0.6 ppm 8 hr TWA between 1977 and 1987 and 0.14 ppm 8 hr TWA from 1988 through September 2003. A weighted average of these readings is 0.33 ppm which can serve as the NOAEC for this study.  

[Tsai, S.P., Fox, E.E., Ransdell, J.D., Wendt, J.K., Waddell, L.C., Donnelly, R.P., 2004. A hematology surveillance study of petrochemical workers exposed to benzene. Regul. Toxicol. Pharmacol. 40, 67– 73. https://doi.org/10.1016/j.yrtph.2004.05.010]  

  

Khuder et al. (1999) examined 105 workers at a Texas refinery. All workers worked in sections of the refinery with benzene exposure and had at least two blood tests and two personal monitoring measurements applicable to them over the period 1977 – 1994. Average annual exposures ranged from 0.14 – 2.08 for the study period. Results were analysed both in aggregate (mean blood counts and benzene readings by year) and individually (pre-mid-late interval trends in blood counts and benzene exposures for each worker). No relationships with benzene concentration (nor time x Benzene concentration) were found in either analysis for white blood cells, red blood cells, MCV, hemoglobin, nor platelets. The average benzene level thoughout the period of the study was 0.81 ppm, which can be taken as the NOAEC.  

[Khuder SA, YoungdaleMC, Bisesi MS and Schaub 1999 Assessment of complete blood count variations among workers exposed to low levels of benzene. J Occup Environ Med 41 (9) 821-6] 

   

Repeated Dose – Genotoxicity 

 

Studies used for weight of evidence to derive an OEL (Schnatter et al 2020)  

  

Qu et al. 2003 Qu et al., 2003 studied 130 benzene exposed workers from shoe factories in Tianjin, China as well as 51 other production workers not exposed to benzene. They examined several indices via both interphase FISH techniques (Chromosome 1 and 7) and conventional cytogenetic assays. Benzene exposure was examined via both a 4-week average exposure as well as lifetime cumulative exposure. FISH analyses did not identify effects on G0 lymphocytes and granulocytes using probes for Chromosome 1 and for chromosome 7 nor were clear effects related to recent exposure identified using chromosome 1 tandem probes in 48 h lymphocyte cultures nor with chromosome 7 single probe investigation of 72 hour lymphocyte cultures. Accordingly, the authors focused data analysis on materials related to conventional metaphase cytogenic methods. For continuous four week exposure average, there was a moderate association for chromatid gaps, chromosomal breaks and hypodiploidy, but not any of the other cytogenetic indices when appropriate control for confounders was taken into account (in Table 28). The exposure categories are shown in Table 27 while this data is without control for confounders. When the three endpoints (viz. chromatid gaps, chromosomal breaks and hypodiploidy) that show a significant dose-response trend with confounder control (Table 28) are examined by exposure categories, hypodiploidy and chromatid gaps do not show a monotonic dose response pattern, but chromosome breaks does. Moreover, for chromatid gaps analysis according to Table 1 of Appendix A there is no good basis to pool the data from phase 1 and phase 2 as Qu has done, also chromatid gaps are of doubtful biological significance. For the chromosome aberrations, the lowest dose that shows an effect is the 15-30 ppm group. The four week mean exposure in this group is 19.9 ppm, which could be considered as a LOAEL. However, this group was based on only 8 individuals and subsequently, a formal test of means indicated that the p-value was 0.141 (not significant) for the >15-30 group. Thus, this analysis should not be used to establish a genotoxicity LOAEL for the Qu et al. study. For other cytogenetic endpoints (e.g. total aberrations) that did not show a dose-response relationship when confounders were controlled, exposure category analysis (without confounder adjustment) showed that the lowest group of workers (0-5 ppm) had a higher rate of most cytogenetic parameters versus controls. However, lower rates were often seen in the next two exposure categories (i.e. 5-15 and 15-30 ppm), with only the highest exposure category (>30 ppm) showing a higher rate than the 0-5 ppm group. In addition, when the 0-5 group was analysed further, only a subgroup of workers exposed from 0-0.5 ppm showed higher cytogenetic indices vs. controls. The authors note that the median exposure (prior to the four weeks) for this group was 2.7 ppm, which may have played a role in the finding. In addition, since confounders were not controlled, other factors could have affected the trends seen. The authors also analysed cumulative exposure to benzene, and stronger more consistent dose-response patterns emerged for most cytogenetic indices in exposure categorical analyses. However, when cumulative exposure was broken down into its two components (exposure duration and exposure intensity) different cytogenetic indices were related to the two different components. The authors state that it is not clear whether these associations are due to chance fluctuations. The data from Qu et al. allows the assessment of not only whether a linear trend was significant (which was reported by the authors) but also whether a specific categorical exposure group was significantly different than the controls. When looking at this combination (i.e. specific categorical exposure group is significantly different than controls (particularly when adjusted for confounders) and a significance for the linear trend test) only total chromosomal aberrations (except gaps), hypodiploidy (45 chromosomes) and aneuploidy (45 or 47 chromosomes) showed statistical significant effects with benzene exposure (Table 27, Table 28, Table 15 in Appendix A, Table 16 in Appendix A). Of these total chromosomal aberrations (except gaps) is seen at the lowest exposure group (p = 0.022 ) and also this response is monotonic since the rate for controls and the four dose groups are: 1.78, 2.74, 3.23, 3.03, and 3.09 (Table 15 in Appendix A). The last three numbers are very similar and essentially equal, suggesting an increase and then a plateau of response at 5.89 ppm (mean value of 5-<15 ppm group). The recent (4-week) exposures reported for the 0-<5ppm average concentration group has a mean value of 3.07 ppm and is considered the LOAEL from the Qu et al. Study. 

[Qu, Q., Shore, R., Li, G., Jin, X., Chen, L.C., Cohen, B., Melikian, A.A., Eastmond, D., Rappaport, S., Li, H., Rupa, D., Waidyanatha, S., Yin, S., Yan, H., Meng, M., Winnik, W., Kwok, E.S.C., Li, Y., Mu, R., Xu, B., Zhang, X., Li, K., 2003. Validation and evaluation of biomarkers in workers exposed to benzene in China. Res Rep Health Eff Inst 1–72; discussion 73-87.] 

  

The studies of Xing et al 2010, Ji et al 2012 and Marchetti et al 2012 examined aneuploidy and chromosomal aberrations in sperm from men exposed to benzene within 3 Chinese factories making shoes, paper bags and sandpaper and compared it with that from benzene unexposed workers from 2 factories making ice-cream and meat products. These three studies are related because they are all from one population and one laboratory and therefore do not provide independent replication of the reported findings. Xing et al 2010 report examination of sperm using a X-Y-21-sperm FISH assay whilst Ji et al 2012 uses the same data and applied a parallel investigation to peripheral blood lymphocytes. Marchetti et al 2012 investigated chromosome 1 by using sperm FISH to investigate partial chromosomal duplications or deletions of 1cen or 1p36.3 or breaks within 1cen-1q12 and numerical aberrations. Subjects were frequency matched for age and smoking-habits. Other important confounders for genotoxicity were assessed and controlled for in a multivariate statistical model as covariates, although X-ray exposure has not been considered as confounder in these studies. In Xing et al and Ji et al exposure by urinary tt-MA levels at the median (6.7 mg/L) was used to categorise exposed workers into low (median tt-MA 1.9 mg/L, median air 1ppm, geometric mean air 1.0 ppm, median benzene in urine 4.3 µg/L) and high (median tt-MA 14.4 mg/L, median air 7.7 ppm, geometric mean in air 7.6 ppm, median benzene in urine 52.5 µg/L) exposure groups. Marchetti et al 2012 reassigned benzene-exposed workers into 3 groups based on urinary benzene levels into low (median benzene urine 2.9 µg/L, median air 1.2 ppm, geometric mean air 1.0 ppm, median tt-MA 1.7 mg/L) , moderate (median benzene urine 11.0 µg/L, median air 3.7ppm, geometric mean air 3.0 ppm, median tt-MA 7.9 mg/L) and high (median benzene urine 110.6 µg/L, median air concentration 8.4ppm, geometric mean air 7.6 ppm, median tt-MA 13.4 mg/L). In this study the results were similar when benzene exposure was categorized according to tertiles of urinary tt-MA. The authors also analysed sperm aneuploidy in the context of urinary benzene exposure by continuous data analysis (Xing et al, Ji et al). The measures of exposure (i.e. benzene in breathing zone and urinary benzene and tt-MA) are reported to be highly correlated among exposed men and although it is noted that tt-MA is not normalized to creatinine, the air exposures seem to be reasonably aligned with urinary biomarkers of exposure when using existing correlations (DFG) between air and urinary benzene. The authors did not analyse their results by air exposure (personal communication). Xing et al reported statistically significant aneuploidy increases in low- and high-exposed groups for disomy X (incidence-rate ratio (IRR) 2.0 and 2.8, respectively) and overall hyperhaploidy (IRR 1.6 and 2.3, respectively) and disomy Y and total hyper- and hypoploidy only at high exposure group (IRR 2.6 and IRR 1.7, respectively). Despite statistical significance in the study all these aneuploidy effects for unexposed and exposed are within the normal control range seen in healthy men as reported in various studies (Williams et al., 1993, Griffin et al 1995, Robbins et al., 1995, Robbins et al., 1997, Martin et al., 1996, Baumgartner et al., 1999, Frias et al., 2003). Ji et al 2012 reported (in addition to sperm aneuploidy results reported previously by Xing et al) that in lymphocytes trisomy 21 was not statistically significantly different from the unexposed and only weakly positively associated with urinary benzene levels (only high-exposed permuted < 0.1) and there was no association of benzene with X and Y aneuploidy in lymphocytes. The lymphocyte chromosomal pattern is in contrast with the sperm pattern. Marchetti et al 2012 reported benzene-related effects on structural aberrations of chromosome 1 (at all exposure groups 1p36.3 duplications and deletions, but not the 1cen region, 1q12 breaks only at high exposure group) with incidence-rate ratios (IRR) of 1.42, 1.44 and 1.75 for total structural aberrations and 4.31, 6.02 and 7.88 for 1p36.3 deletions in low-, moderate- and high-exposed groups, respectively. There were no effects on numerical aberrations of chromosome 1, although the authors caution that their technique might be insensitive to detecting aneuploidy. Again despite statistical significance in the study itself for structural aberrations in chromosome 1, the values are within the normal control range seen in healthy men as reported in various studies (Baumgartner et al., 1999, McInnes et al., 1998, Van Hummeln et al., 1996, Sloter et al., 2000, Sloter et al., 2007). Whilst germ cell aneuploidy is expected to lead to infertility and loss of pregnancy, such developmental and reproductive toxicity effects are not a known phenomenon for benzene (DECOS 2014, IARC 2018). Moreover, Marchetti et al report that there was no difference between exposed and non-exposed groups in the values of 4 WHO sperm quality criteria (sperm concentration, sperm count, semen volume and percent sperm motility). There was no imbalance between the groups of individuals with low scores for these criteria indicating that there was no effect of benzene on these sperm quality factors. The authors indicate in the noted limitation in design and analysis that the 1 month sampling period did not cover the whole meiotic period and that some of the damage measured by the assay may have occurred at an earlier time point than that monitored by the exposure assessment. Thus, it remains possible that the exposure may have been misclassified. In addition, the workers were categorized in exposure categories based on exposure data on two working days only. It is not clear how representative these two days are for the exposure period of interest and what could have been the likelihood for workers to be classified in the wrong exposure category. There are further concerns about the suitability of the control group. Critically no reference is made to the duration of the work history to benzene exposure (i.e. worked at the factory for at least one year was used as inclusion criteria) nor to prior exposure to benzene and other chemicals. Xing et al states that levels of toluene and xylene were also assessed in their study but do not report any of the effect of these exposures on sperm parameters, nor do they discuss potential effects of other potential co-exposures in shoe manufacturing such as heavy metals. It is noted that whilst the authors criticise other studies as being based on small cohorts (~15 individuals) the low and high exposure groups of Xing et al and Ji et al include only 17 and 16 men, respectively. In the study of Marchetti et al exposed groups contained 10 and the control group 11 individuals. Sperm hyperhaploidy data presented by Xing et al (figure 1) showed that the top two highest frequencies were in low exposed individuals suggesting significant heterogeneity in the data set. It is clear that data from Xing et al, Ji et al and Marchetti et al relate to a novel situation of aneuploidy events in sperm, an observation that has not been replicated in other studies at such exposure levels but has been shown at much higher exposure levels (13-27 ppm range) in other studies (Li et al., 2001, Liu et al., 2000, Liu et al., 2003, Zhao et al., 2004). Ji et al also looked at lymphocytes - a more studied cell type - but these somatic cells did not show effects in the low-exposure group and a borderline increase in trisomy 21 at the high-exposure group. It is noted that the sperm aneuploidy data have been produced using non-standardised methods and different DNA probes, FISH procedures and scoring criteria impact observed aneuploidy levels (Ji et al) and thus the data may contain more variability than data from more usual methods. It has also been published that there are technical challenges of using FISH techniques for detecting aneuploidy in sperm (Downie et al., 1997, Shi et al., 2000). Despite statistical significance and dose-response seen on aneuploidy and chromosomal aberrations in sperm these effects are within the normal control range seen in healthy men as reported in various studies. For all of these reasons, therefore, data from these studies should be interpreted with much caution. Finally, in each of the three studies, a geometric, rather than arithmetic mean is reported. Arithmetic means are preferred for summarising health outcome data. Assuming a log normal distribution, arithmetic means can be calculated as 1.6 ppm for the low-exposure group. It is noted that Xing et al. applied a GM calculation twice (i.e. first calculates GM for an individual based on two results and then used these GM values to calculate the GM for the group) so there is likely an underestimation of the true arithmetic exposure due to the calculations that were made. Hence >1.6 ppm could serve as a conservative LOAEL from the studies for aberration and aneuploidy effects on sperm. The peripheral blood lymphocytes indicate a NOAEL of 7.6 ppm for aneuploidy based on the geometric mean concentration of the high exposed group in Ji et al (2012). 

  

References for Xing et al 2010, Ji et al 2012 and Marchetti et al 2012  

Baumgartner A, Van Hummelen P, Lowe XR, Adler ID, Wyrobek AJ. 1999. Numerical and structural chromosomal abnormalities detected in human sperm with a combination of multicolor FISH assays. Environ Mol Mutagen 33(1):49-58. 

DECOS health-based recommended occupational exposure limits, Health Council of the Netherlands, 2014, GSW/1560 space 98 – 33 Downie, SE, Flaherty SP and Mathews 1997 Detection of chromosomes and estimation of aneuploidy in human spermatozoa using fluorescence in situ hybridisation. Molecular Human Reproduction 3 (7) 585-598. 

Frias S, Van Hummelen P, Meistrich ML, Lowe XR, Hagemeister FB, Shelby MD, Bishop JB, Wyrobek AJ. 2003 NOVP chemotherapy for Hodgkin’s disease transiently induces sperm aneuploidies associated with the major clinical aneuploidy syndromes involving chromosomes X, Y, 18, and 21. Cancer Res 63(1):44-51. 

Griffin DK, Abruzzo MA, Millie EA, Sheean LA, Feingold E, Sherman SL, Hassold TJ. 1995. Nondisjunction in human sperm: Evidence for an effect of increasing paternal age. Hum Mol Genet 4(12):2227–2232. 

IARC Monographs on the evaluation of carcinogenic risks to humans. Benzene monograph volume 120, 2018 

Ji, Z., Weldon, R.H., Marchetti, F., Chen, H., Li, G., Xing, C., Kurtovich, E., Young, S., Schmid, T.E., Waidyanatha, S., Rappaport, S., Zhang, L., Eskenazi, B., 2012. Comparison of aneuploidies of chromosomes 21, X, and Y in the blood lymphocytes and sperm of workers exposed to benzene.  

Environ. Mol. Mutagen. 53, 218–226. https://doi.org/10.1002/em.21683 

Li X, Zheng LK, Deng L X , and Zhang Q , 2001 Detection of numerical chromosome aberrations in sperm of workers exposed to benzene series by two-color fluorescence in situ hybridization. Yi Chuan Xue Bao 28 589-594 (cited in Ji et al 2012) 

Liu et al. Zheng L, Deng L, Tang G and Zhang Q, 2000 Detection of numerical chromosomal aberrations in sperm of workers exposed to benzene series by two-color fluorescence in situ hybridization. Zhonghua Yu Fang Yi Xue Za Zhi 34 17-19 (cited in Ji et al 2012) 

Liu XX, Tang GH, Yuan YX, Deng LX, Zhang Q and Zheng LK, 2003 Detection of the frequencies of numerical and structural chromosomal aberrations in sperm of benzene series-exposed workers by multi-color fluorescence in situ hybridization Yi Chuan Xue Bao 30 1177-1182 (cited in Ji et al 2012) Marchetti, F., Eskenazi, B., Weldon, R.H., Li, G., Zhang, L., Rappaport, S.M., Schmid, T.E., Xing, C., Kurtovich, E., Wyrobek, A.J., 2012. Occupational exposure to benzene and chromosomal structural aberrations in the sperm of Chinese men. Environ. Health Perspect. 120, 229–234.https://doi.org/10.1289/ehp.1103921 

Martin RH, Spriggs E, Rademaker AW. 1996. Multicolor fluorescence in situ hybridization analysis of aneuploidy and diploidy frequencies in 225,846 sperm from 10 normal men. Biol Reprod 54:394 – 398. 

McInnes B, Rademaker A, Greene C A, Ko E, Barclay L, Martin R H. 1998. Abnormalities for chromosomes 13 and 21 detected in spermatozoa from infertile men. Human Reproduction, Volume 13, Issue 10, pages 2787–2790. 

Personal Communication: Email : Dr B Eskenazi to Dr AR Schnatter 12 November 2018 Robbins WA, Balch JE, Moore D II, Weier HU, Blakey D, Wyrobek AJ. 1995. Three-probe fluorescence in situ hybridization to assess chromosome X, Y, and 8 aneuploidy in sperm of 14 men from two healthy groups: Evidence for a paternal age effect on sperm aneuploidy. Reprod Fertil Dev 7:799– 809. 

Robbins WA, Meistrich ML, Moore D, Hagemeister FB, Weier H-U, Cassel MJ., Wilson G, Eskenazi B, Wyrobek AJ. 1997. Chemotherapy induces transient sex chromosomal aneuploidy in human sperm. Nat Genet 16:74 –78. 

Shi Q and Martin RH (2000) Aneuploidy in human sperm: a review of the frequency and distribution of aneuploidy, effects of donor age and lifestyle factors. Cytogenet Cell Genet 90 219-226. 

Sloter ED, Lowe X, Moore II DH, Nath J, Wyrobek AJ. 2000. Multicolor FISH analysis of chromosomal breaks, duplications, deletions, and numerical abnormalities in the sperm of healthy men. Am J Hum Genet. 67(4):p862-72. 

Sloter ED,Marchetti F , Eskenazi B, Weldon RH, Nath J, Cabreros D, Wyrobek AJ. 2007. Frequency of human sperm carrying structural aberrations of chromosome 1 increases with advancing age. Fertility and Sterility. Volume 87, Issue 5, Pages 1077-1086. 

Van Hummelen P, Lowe XR, Wyrobek AJ. 1996. Simultaneous detection of structural and numerical chromosome abnormalities in sperm of healthy men by multi-color FISH. Hum Genet 98:608–615. 

Williams BJ, Ballenger CA, Malter HE, Bishop F, Tucker M, Zwingman TA, Hassold TJ. 1993. Nondisjunction in human sperm: Results of fluorescence in situ hybridization studies using two and three probes. Hum Mol Genet 2(11):1929 –1936. 

Xing, C., Marchetti, F., Li, G., Weldon, R.H., Kurtovich, E., Young, S., Schmid, T.E., Zhang, L., Rappaport, S., Waidyanatha, S., Wyrobek, A.J., Eskenazi, B., 2010. Benzene exposure near the U.S. permissible limit is associated with sperm aneuploidy. Environ. Health Perspect. 118, 833–839.https://doi.org/10.1289/ehp.0901531 

Zhao T, Liu XX, He Y, Deng LX and Zheng LK , 2004 Detection of numerical aberrations of chromosomes 7 and 8 in sperms of workers exposed to benzene series by two-color fluorescence in situ hybridization Zhonghua Yi Xue Yi Chuan Xue Za Zhi 21 360-364 (cited in Ji et al 2012) 

Eastmond et al 2001 (Chinese workers) investigated the chromosomal aberrations in chromosome 1 in 44 Chinese workers from three factories utilizing benzene-containing chemicals and 44 frequency-matched on age and sex controls from either a sewing machine factory or administrative facility in the same geographic area. Benzene exposure was monitored using personal passive sampling (3M no.3500; 3M, St Paul, MN) over a full work shift on five separate days during the 1-2 weeks prior to blood collection. The 8-hr time weighted average for the five days was used to calculate a geometric mean exposure. Furthermore, as described by Rothman et al (1996), the benzene exposures were quite high: The median exposure concentration in the exposed group was 31 ppm (8-hour time-weighted average [TWA]) with a measured range from 1.6 ppm to 328.5 ppm. The median benzene exposure in the control subjects was 0.02 ppm, with a range from 0.01 ppm to 0.1 ppm. Clear and significant differences in benzene exposure variables were seen between the two groups. Fluorescence in situ hybridization (FISH) with the tandem labelled probes for chromosome 1 was performed on the 72-hour cultured interphase lymphocytes from the workers and control subjects. The authors report that significant differences were not seen for either endpoint (P>0.05). The median frequency of breakage affecting the 1cen-q12 region in the control subjects was 2‰ with an interquartile range (IQR) of 1‰ to 3‰. An almost identical frequency was seen in the exposed workers with a median of 2‰ (IQR 1‰–4‰). Similar results were seen for hyperdiploidy. The median frequency of hyperdiploidy in the cells from the benzene exposed workers and the control subjects was 2 with an IQR of 1‰ to 4‰. When the exposed group was divided into two exposure categories, one group with exposures below the median of 31 ppm and a second group with exposures above 31 ppm, no difference from the control subjects was seen with either hyperdiploidy or breakage in workers from either of the exposure categories. The median (and IQR) frequencies of breakage were 2‰ (1‰–3‰), 2‰ (1‰–4‰), 2‰ (0‰–4‰) in the control, low exposure, and high-exposure groups, respectively. The median (and IQR) frequencies for hyperdiploidy among these same groups were 2‰ (1‰–4‰), 2‰ (0‰–3‰), and 3‰ (2‰–5‰), suggesting that a slight increase in hyperdiploid cells might be occurring in the workers with the highest exposure (>31 ppm). This pattern of hyperploidy was considered to be similar to that reported by Zhang et al (1996) in which a significant increase in hyperdiploid cells was seen in the highly exposed group when analysed using FISH with a single chromosome 9 probe. In this earlier study, the authors reported that a significant correlation between urinary phenol and hyperdiploidy was seen among the exposed workers. Based on this result, a similar comparison was performed for the chromosome 1 tandem label data. A significant correlation between urinary phenol and hyperdiploidy for chromosome 1 was seen for the exposed workers (P 0.0075; r 0.412) as well as the entire study population (P 0.0146; r 0.322). The authors report that here was a similar association between the frequency of cells hyperdiploid for chromosome 1 and urinary t,t-ma concentration for both the exposed group (P 0.0046; r 0.434) and the entire study group (P 0.0415; r 0.271). No association between urinary phenol or t,t-ma was seen for breakage whether the analysis was based upon the entire study group or only the exposed group. Significant associations also were not observed between (a) the frequency of breakage or hyperdiploidy and (b) age, smoking status, or cigarettes per day. However, there was a weak but statistically significant difference in the breakage frequency was observed between the males and females, however (P 0.038; t test). The mean (± SD) frequency of breakage in the male cells (1.485 ± 0.821) was higher than that seen in cells from the females (1.116 ± 0.789). These results indicate evidence of an increase in aneuploidy of chromosome 1 with exposures >31 ppm benzene in workers which could be considered as the study LOAEL. 

[Eastmond DA, Schulker M, Frantz et al 2001 Characterization and Mechanisms of Chromosomal Alterations Induced by Benzene in Mice and Humans. Res Rep Health Eff Inst 103 1-80 ] Overall therefore the genotoxicity conclusion from these Rothman et al 1996 associated studies (Zhang 2007, Zhang 1998, Smith 1998, Zhang 1996, Eastmond 2001) is that a LOAEC of 13.6 ppm can be determined. 

 

Zhang et al (2012) studied 28 workers exposed to benzene from two shoe manufacturing facilities and 14 unexposed controls from clothing manufacturing facilities in Tianjin, China. These are a subset of workers originally reported on by Lan et al (2004). Controls were frequency matched by sex and age to exposed workers. The authors examined colony forming unit-granulocyte-macrophage (CFU-GM) cells, a myeloid progenitor. The authors were primarily interested in monosomy 7 and trisomy 8, due to their respective roles in MDS and AML, but examined monosomy and trisomy of both chromosomes. They used FISH (fluorescence in situ hybridisation) techniques to measure aneusomy in interphase cells (rather than metaphase cells). Exposure assessment has been reviewed previously in association with the haematological aspects of Lan et al 2004 and outcomes may be influenced by higher previous exposures. In this worker subset, the authors reported two groups exposed to 10 ppm (n = 10) benzene. Exposure was described as a single reading one month before phlebotomy. Having only one reading per person poses a high risk of exposure misclassification, so that the low group may have been contaminated with workers from the high dose group and vice-versa. Means (assumed to be arithmetic means) among the 18 workers exposed to 10 ppm were 2.64 and 24.19 ppm, respectively, while U-Bz levels were 66.39 and 897.70 mg/L, respectively. Using negative binomial models and adjusting for important confounders, the authors reported a significant increase in both monosomy 7 and monosomy 8 for both exposure groups. There was a significant dose-response trend for both outcomes. Trisomy 7 and 8 were not related to benzene exposure. A high rate of monosomy in the control group was hypothesized to be probe overlap inherent to the examination of interphase chromosomes. The background incidence of monosomy for chromosomes 7 and 8 in control population is relatively high with a mean % ± SD of 4.32 ± 2.06 and 5.21 ± 2.17, respectively. The authors attribute this high incidence to the artefact produced by their staining technique. Criticism has been made by Gentry et al 2013 of the use of CFU-GM methods as applied to formaldehyde in the work of Zhang et al 2010. Gentry et al proposed that the scoring of aneuploidy in pooled metaphases rather than of specific aneuploid colonies would not differentiate between aneuploidy that had existedin vivoand that that had developed in the cells during the 14 day culture period. In the case of formaldehyde, Lan et al 2015 recognise the limitation in their study identified by Gentry et al but reason that even if their findings do not definitively identify aneuploidy that existed in vivo it points to cells from exposed workers being more susceptible to developing aneuploidy in vitro. The direct implications of these observations to benzene studies are unclear but point to the need for caution in interpreting the outcome and significance of findings from studies using novel and potentially non-standardised techniques. The urinary benzene levels measured in Zhang et al 2012 (low dose group mean 66.39 µg/L) according to DFG 2017 are not consistent with the reported air exposure (2.64 +/- 2.7 ppm for the 2.64 ppm (standard deviation = 2.70) based on increases in monosomy 7 and monosomy 8 observed at interphase. 

[Gentry et al 2013 Formaldehyde exposure and leukemia : critical review and reevaluation of the results from a study that is the focus for evidence of biological plausibility. Crit Rev Toxicol 43 (8) 661-70 Lan et al 2015 Chromosome=wide aneuploidy study of cultured circulating myeloid progenitor cells from workers occupationally exposed to formaldehyde. Carcinogenesis 36 (1) 160-7 Zhang et al 2010 Occupational exposure to formaldehyde, hematotoxicity and leukemia-specific changes in cultured myeloid progenitor cells. Cancer Epidemiol Biomarkers Prev 19 (1) 80-8 

Zhang, L., Lan, Q., Ji, Z., Li, G., Shen, M., Vermeulen, R., Guo, W., Hubbard, A.E., McHale, C.M., Rappaport, S.M., Hayes, R.B., Linet, M.S., Yin, S., Smith, M.T., Rothman, N., 2012. Leukemia-related chromosomal loss detected in hematopoietic progenitor cells of benzene exposed workers. Leukemia 26, 2494–2498.https://doi.org/10.1038/leu.2012.143] 

 

Carere et al 1995 investigated the possible relationship between occupational exposure to petroleum fuels and cytogenetic damage in peripheral lymphocytes. Twenty-three Italian, non-smoking male station attendants were compared to age-paired, healthy non-smoking blood donors. Although relevant confounders were considered in the study, apart from diet, it is unclear whether the blood donors used are from general population including office workers giving rise to a potential information bias. Peripheral lymphocyte cultures were set up for the analysis of structural chromosome aberrations (CAs), sister chromatid exchanges (SCEs) and micronuclei (MN) in cytokinesis-blocked lymphocytes. Benzene exposure was measured by personal sampling (an average of 6.5 samplings per year and subject) and resulted in mean benzene levels of 1.5 ± 0.7 mg/m3(0.47 ± 0.22 ppm) and a range of 0.1 to 13.1 mg/m3(0.03-4.1 ppm). Benzene exposure of the healthy non-smoking blood donors was not measured but they are reported to have had no occupational exposure to fuels or other chemicals. Mean length of employment of the exposed workers was 22.4 ± 2.3 years. There was a statistically higher level of blood lead detected in exposed versus control subjects. No statistically significant effects (with p < 0.05) were reported on structural chromosome aberrations (CAs), sister chromatid exchanges (SCEs) and micronuclei (MN) in this study between the exposed and the control group. A further analysis was reported of the exposed workers (who had a median benzene exposure of 0.2 mg/m3and a mean benzene exposure of 1.5 mg/m3) by splitting them into two groups around the median. Consequently, there was a low (benzene exposure <0.2 mg/m3) and a high benzene (median benzene exposure ≥ 0.2 mg/m3) group. (Note: for two workers no benzene exposure measurements are available, it is unclear if these were excluded). A significant overall upward trend in percentages of cells with CAs at increasing levels of benzene exposure was demonstrated. Furthermore, the authors claim that filling-station attendants belonging to the high benzene exposure group showed a statistically significant increase of CAs in comparison with controls (p value not given in the publication). For cells with high frequency SCEs no significant overall upward trend in percentages of cells with high frequency SCEs at increasing levels of benzene exposure could be shown. The reported excess of high frequency SCEs could be correlated to the increasing blood lead levels. Thus, with regards to genetic toxicity of benzene a NOAEC of 0.47 ± 0.22 ppm can be derived as no statistically significant effects on CAs, SCEs and MN were reported in this study at this dose. The LOAEC for induction of chromosome aberrations can be taken as the mid-point of the high exposure group which had a range of 0.2-13.1 mg/m3(i.e. 0.06-4.1 ppm) namely 2 ppm. 

[Carere A, Antoccia A, Crebelli R, Degrassi F, Fiore M, Iavarone I, Isacchi G, Lagorio S, Leopardi P, Marcon F, et al 1995 Genetic effects of petroleum fuels: cytogenetic monitoring of gasoline station attendants. Mutat Res 332: 17-26] 

 

Rekhadevi et al 2010 and Rekhadevi et al 2011 It appears that the results of a single study have been provided in two associated papers in which different endpoints were reported separately. These papers concern an investigation into micronuclei (both in PBL and in buccal epithelial), chromosomal aberrations and a comet assay in petrol filling station attendants in India. Two hundred workers who had each worked for at least 5 years in filling stations were recruited together with 200 control subjects from the general population with no history of exposure to fuels, heavy metals for other genotoxic agents. It was noted that the gasoline being handled contained 5-6% benzene compared to < 1% in Europe. Exposure was determined by diffusive sampler in the breathing zone of subjects during the last 8-10 hour shift of the week. Additionally, area monitoring, benzene in blood and various urinary determinations were used as indices of exposure. Summary data indicated an arithmetic mean ± SD for personal benzene exposure of 1500 ± 138 µg/m-3(0.46 ppm) in attendants compared to 175± 15 µg/m-3(0.05 ppm) in controls. Poor working conditions including dermal exposure, as a result of none of the study subjects using hand gloves but frequently getting dirty as they pumped fuel, can be assumed in the study. The authors indicated that benzene in blood was reported as being 5.18 ± 0.98 µg/L in attendants versus 2.12±0.51 in controls. Based on the mean blood benzene level in the filling station attendant category (5.18 ppb), DFG conversions would indicate a total exposure of 1ppm atmospheric air equivalent (see http://onlinelibrary.wiley.com/doi/10.1002/3527600418.bb7143e0003/pdf for end of shift blood benzene correlations to air). Whilst urinary biomarkers of exposure (benzene, ttma, spma and phenol) were monitored, these gave confused results. Urinary benzene concentrations for the low exposure category analysed by the authors (stated atmospheric benzene level 1,100-1,300µ/m3 = 0.34-0.41 ppm) were 8.89 ± 1.41 µg/L (mean ±SD). According to the DFG conversions this indicated that total exposure was the equivalent of 1-2 ppm atmospheric benzene. However, the u-ttma and uspma values were below the validated levels for the DFG conversions (300 µg/g creatinine and 12.0 µg/g creatinine, respectively.) Urinary phenol, apparently reported in the wrong units (µg/L rather than mg/L), appears to have increased with reported category mean air exposure (10.11, 12.04 and 14.14 mg/L respectively as corrected unit values for low, medium and high exposure groups). However, this would be within the normal range for urinary phenol (up to 30 or even 50 mg/L) so could not be used to check on reported air exposures or consider the contribution by dermal exposure. Based on the combined data from both urinary benzene and from blood benzene it can be concluded that total benzene exposure was the equivalent of exposure to ≥ 1 ppm benzene in air 8hTWA. Micronuclei in PBL were present at a higher rate in exposed individuals (11.8 ± 1.4 % compared to 5.83 ± 1.21 % in controls) as were micronuclei in buccal epithelium (4.46 % versus 2.63% - p 0.05). Chromosomal aberrations in PBL were present at 6.2 % compared to 2.39 % in the control group and DNA damage, based on comet tail length, was a mean of 25.1 ± 2.28 compared to 10.27 ± 1.52. As previously described subjects were arbitrarily categorised into low, medium and high benzene exposure according to the service station air measurements. Higher levels of both micronucleus frequency in buccal epithelium and of chromosomal aberrations in PBL were reported in high exposure groups compared to medium exposure and in medium exposure groups compared to low exposure. However, it was not entirely clear if buccal micronuclei and PBL chromosome aberration frequency in the low exposure group (reported area measurement of 1.1-1.3 mg benzene /m3) were statistically different from control. However, if the effects in the low exposure group are taken as being statistically significant from control then the LOAEC for micronuclei in buccal epithelium and chromosome aberrations in peripheral blood are seen at an atmospheric benzene concentration of ≥ 1 ppm as an 8 h TWA based on urinary benzene data. Poor working conditions were documented with oil and grease exposure. This together with co-exposures (toluene, xylene) and air emissions (PAH, particulates etc) gave potentially confounding exposures. Additionally, effects of gender, age, smoking and drinking were not completely matched between groups. The authors attributed statistically significant differences in plasma biochemistry in the attendants compared to controls (elevated glucose, creatinine and cholesterol and reduced protein and albumin) to differences in lifestyle habits, diet and medical history further suggesting that the control and exposed groups differed other than by benzene exposure. Overall, a LOAEC of > 1ppm) can be derived from this study based on micronuclei in buccal epithelium and peripheral blood lymphocytes together with comet DNA damage. The data do not support a NOAEC. However, confidence in these findings is reduced by concerns about co-exposure and confounding. 

[Rekhadevi PV, Rahman MF, Mahboob M, Grover P 2010 Genotoxicity in filling station attendants exposed to petroleum hydrocarbons. Ann Occup Hyg 54: 944-54. 

Rekhadevi PV, Mahboob M, Rahman MF, Grover P 2011 Determination of genetic damage and urinary metabolites in fuel filling station attendants. Environ Mol Mutagen 52: 310 - 318.] 

 

Zhang et al 2014 examined the association between benzene exposure and micronucleus (MN) frequency. They additionally examined the effect various polymorphisms have on the expression of micronuclei. 385 benzene exposed workers (190 men and 195 women) were recruited from shoe manufacturers in Wenzhou, Zhejiang province China and compared with 102 indoor workers (internal control) from the same city and 95 teachers from Shanghai (external controls) that were matched by gender and age. Airborne benzene levels were sampled three times a day; they were also collected at three different worksites during the study. Blood samples were collected from each patient (informed consent) during the period May to December 2011. The Cumulative Exposure Dose (CED) for each worker was estimated on the basis of record of work history, location and duration at the factory. The geometric mean of benzene concentration was also estimated for each work site. The intensity of benzene concentration for the exposed group ranged from 2.6 mg/m3to 57.0 mg/m with a median of 6.4 mg/m3(2.0 ppm). The exposed workers were classified into three groups: < 1 ppm (< 3.25 mg/m3), <1.85 ppm (<6.00 mg/m3) or ≥1.85 ppm (≥ 6.00 mg/m3) based on the US (1 ppm) and China standards (6.00 mg/m3). A cytokinesis-block micronucleus assay was conducted on the samples with the results indicating that the MN frequency was significantly increased with all levels of benzene exposure. A dose-response for MN frequency with cumulative exposure (covering the entire exposure duration in the plant) was observed using categorial and continuous data analysis. However, there are limitations with the study, including a lack of dose response relationship for benzene concentration with the MN frequency and that very similar frequency ratios were calculated for each of the exposure levels. The mean MN frequency was quite different between the exposed and controls. In addition, there were only 24 study subjects in the below 1ppm group, while the higher exposure levels had 6 – 8 times as many subjects. There was no control for dermal exposure, which is not unlikely to happen for shoe workers from benzene containing glue, or diet in the study and area measurements were used and not personal air sampling to determine the geometric means for benzene (values not reported). Due to the insufficient exposure assessment and the unclear dose-response for micronuclei, the stratification into the different exposure groups cannot be considered as reliable. However, the mean MN frequency in the exposed group (median of 6.4 mg/m3(2.0 ppm)) was statistically significant from the control frequency (Table 2). Therefore, considering both the limitations and statistical evidence, the LOAEC for this study can be determined to be 2 ppm (range 0.8-18 ppm). 

[Zhang GH, Ye LL, Wang JW, Ren JC, Xu XW, Feng NN, Zhou LF, Ru JG, Hao YH, Tian W, Sun P, Au WW, Christiani DC, Xia ZL 2014 Effect of polymorphic metabolizing genes on micronucleus frequencies among benzene-exposed shoe workers in China. Int J Hyg Environ Health 217: 726-732.] 

  

Pandey et al 2008 investigated DNA damage (comet assay) and micronucleus formation (cytokinesis block micronucleus test) in lymphocytes of Indian petrol-pump workers (PPW). Comet analysis was done in 100 subjects and 100 controls whilst MN was done in 39 PPWs and 18 controls. Air stationary levels for PPW are reported between 100-250 ppb (controls between 5-10 ppb). The mean blood benzene level in the total group of PPW workers is 7.94 ppb (control group 2.82 ppb). DFG correlation (see http://onlinelibrary.wiley.com/doi/10.1002/3527600418.bb7143e0003/pdf for end of shift blood benzene correlations to air) would give a corresponding air level of ~1.5 ppm assuming blood sampling has been done end of shift in Pandey. It is unclear from the paper when the blood samples (used for blood benzene exposure) were taken making the biomarkers of exposure data difficult to translate given the kinetics of benzene in blood. Therefore, the estimation of ~1.5 ppm carries uncertainty. The blood benzene levels in Pandey et al., 2008 (7.94 ppb) are however higher than those seen in Indian filling station attendants in Rekhadevi et al., 2010 (5.18 ppb end of shift) and Rekhadevi blood benzene exposure data corresponds to ~1 ppm in air and reports 1500 ug/m3(~0.5 ppm) in the breathing zone. The study does indicate similarly to Rekhadevi that dermal exposure is likely a contributor to the overall body benzene exposure. Increased tail DNA percentage and tail length were seen in the comet assay in PPW versus control subjects. Therefore, the LOAEL for comet can be set at ~1-1.5 ppm. For MN the data was analysed by low benzene and high benzene PPW groups. There was no MN effect in the low benzene group (Fig 2, inset) compared with control (Table IV) while the high benzene PPW group did show increased MN frequency. The blood benzene levels in low benzene and high benzene PPW groups were ~ 4 ppb and ~15 ppb (Fig 2, inset), respectively. Therefore, the NOAEL for MN is ~4 ppb and the LOAEL is ~15 ppb corresponding to ~0.9 ppm and ~2 ppm benzene in air respectively using the DFG correlation. 

[Pandey, A.K., Bajpayee, M., Parmar, D., Kumar, R., Rastogi, S.K., Mathur, N., Thorning, P., de Matas, M., Shao, Q., Anderson, D., Dhawan, A., 2008. Multipronged evaluation of genotoxicity in Indian petrol-pump workers. Environ. Mol. Mutagen. 49, 695–707. 2033https://doi.org/10.1002/em.20419 

Rekhadevi PV, Rahman MF, Mahboob M, Grover P 2010)Genotoxicity in filling station attendants exposed to petroleum hydrocarbons. Ann Occup Hyg 54: 944-54.] 

 

Eastmond et al 2001 (Estonian workers) investigated the chromosomal aberrations in chromosome 1 and 9 in an Estonian population comprised of shale oil plant benzene factory workers (n=12), exposed to 4.1 ± 8 mg/m3air benzene: 1.29 ppm (blood benzene 85.9 ± 115.3 nmol/L, urinary ttMA 21.7 ± 44.8 µmol/L, urinary SPMA 39.7 ± 64.9 µg/g creatinine, n=9/12 smokers); shale oil plant coke oven workers (n=5) exposed to 1.1 ± 0.5 mg/m3 air benzene : 0.34 ppm (blood benzene 53.6 ± 18.2 nmol/L, urinary ttMA 5.1 ± 3.6 µmol/L, urinary SPMA 12.9 ± 11.0 µg/g creatinine, n=2/5 smokers); controls which were unexposed from a rural village (n=8), however, air benzene was not measured, (blood benzene 22.4 ± 10 nmol/L, urinary ttMA 1.2 ± 1.9 µmol/L, urinary SPMA 2.6 ± 3.1 µg/g creatinine, n=5/8 smokers). The biomarker data is what would be expected at the reported air levels when looking at the DFG correlations. All subjects completed a questionnaire about personal information including smoking and drinking habits, health status, and age. The highest individual values of exposure markers were consistently detected among benzene factory workers, due to the high variability and the small size of the study groups, however, only the difference among blood benzene values in exposed and control subjects attained statistical significance (P≤0.05). The tandem labelling fluorescence in situ hybridization (FISH) analysis was performed on both granulocytes and unstimulated lymphocytes in the blood smears and on the 48-hour cultured lymphocytes. The authors report that breakages in the labelled region of chromosome 1 were generally higher in granulocytes than in unstimulated lymphocytes (P ≤ 0.05). In comparing the benzene factory workers, the coke oven workers, and control group, an analysis of breakage frequencies indicated that the differences among the three groups were not statistically significant (P ≥ 0.05). In addition, there was no significant association observed between the incidence of breakage in either the lymphocytes or granulocytes and age, smoking habits, benzene-exposure markers, or length of employment; the only exception was a significant association observed between breakage in the 1q12 region and smoking (P=0.027), which was by the authors considered to be of doubtful biological significance. For the blood smears the authors report, significant differences in chromosome 1 hyperploidy were not seen between the two cell types or among the study groups. The frequencies of 1q12 breakage in the granulocytes were significantly correlated with those detected in the G0 lymphocytes (P≤0.05), this association, however, was largely due to a single benzene-exposed individual who exhibited high breakage frequencies in both granulocytes and G0 lymphocytes. The tandem labelling analysis of chromosomes 1 and 9 in cultured lymphocytes, harvested 48 hours after stimulation, showed a modest increase of breakage in the cells from benzene production workers compared with both non-exposed control subjects and coke oven workers exposed to lower benzene levels. The median frequencies of breaks in the 1cen-1q12 region were 2‰, 4‰, and 6‰ in control subjects, coke oven workers, and benzene workers, respectively. Significant differences in the frequencies were observed between the benzene-exposed and the control group (P≤0.05). Reported median incidences of breaks affecting the 9cen-9q12 region were 6‰, 7‰, and 10‰ in the control subjects, coke oven workers, and benzene workers, respectively. Although the difference between the chromosome 9 breakage for three groups did not quite attain statistical significance (P=0.053; Kruskall-Wallis test), an excess of breaks was observed in comparing the benzene-exposed workers with the control group using the Mann-Whitney U test (P≤0.05). However, no correlation was seen between the breakage in either 1q12 or 9q12 regions and the exposure biomarkers, age, or smoking status. The frequency of breakage observed in the 9cen-9q12 region was significantly higher than the frequency of breaks in the 1cen-1q12 region when compared across all groups (P≤0.001), which was possibly reflecting a difference in susceptibility of the two targeted regions to breakage (Brogger 1977; Meyne et al 1979). However, an analysis of individual results revealed a strong linear correlation between the results obtained with the two chromosomes (P≤0.001), which demonstrates the overall reliability of results and the reproducibility of the method. In the stimulated lymphocytes, the incidence of hyperploidy for both chromosomes was slightly higher in the benzene exposed workers compared with the other groups but did not attain statistical significance for either chromosome 1 or 9, which was potentially due to the small sample sizes and the overall low frequency of hyperploidy of chromosomes 1 and 9 observed (0.64‰, 16 out of 25,000 scored cells). Furthermore, no association was seen between hyperdiploidy and the exposure biomarkers. However, it should be noted that only 1 cell hyperdiploid for chromosome 1 (out of 10 total observed), and 0 cells hyperdiploid for chromosome 9 (out of 6 total observed), were detected in the 8,000 cells of control subjects. This was suggestive of a possible effect of occupational exposure. The frequencies among the exposed subjects were within the hyperdiploidy ranges seen previously (Eastmond et al 1995). Therefore, the results of this study indicate a LOAEL of 1.29 ppm benzene based on CA breakages. In addition to Eastmond there are 2 previous publications on related study subjects: Marcon et al., 1999 and Surralles et al., 1997. Surralles studied MN in PBL and buccal mucosa in benzene workers at very similar exposures as Eastmond and showed no effects. The Marcon study looked at chromosomal damage and aneuploidy of chromosomes 1 and 9 using FISH and can be seen as being similar to Eastmond. It also reports very similar results as Eastmond. Although the number of subject and number of smokers are identical between Eastmond and Marcon the mean ages and exposures cited differ, while bloods were taken at the same time point (Autumn 1994) and both papers draw on Kivisto et al. 1997 for exposure data. The Eastmond and Marcon data on air benzene and urinary SPMA appear to be different with a higher exposure in Eastmond (Eastmond has 3-fold higher benzene air exposure and 5-fold higher urinary SPMA exposure in the benzene workers), although this is complicated by the use of arithmetic means in Eastmond and geometric means in Marcon.  

[Brogger, A. 1977 Non-random localization of chromosome damage in human cells and targets for clastogenic action. Chromosome Today 6:297–306. 

Eastmond DA, Schulker M, Frantz et al 2001 Characterization and Mechanisms of Chromosomal Alterations Induced by Benzene in Mice and Humans. Res Rep Health Eff Inst 103 1-80 

Kivisto H, Pkari K, Peltonen K, Svinhugvud J, Veidebaum T, Sorsa M and Aitio A 1997 Biological monitoring of exposure to benzene in the production of benzene and in a cokery. The Science of the Total Environment 199 49-63 

Marcon F, Zijno A, Crebelli R, Carere A, Veidebaum T, PelonenenK, Parks R, Schuler M and Eastmond D 1999 Chromosome damage and aneuploidy detected by interphase multicolour FISH in benzeneexposed shale oil workers Mutat Res 445 155-166 

Meyne, J., L. Lockhart and F. Arrigh 1979 Nonrandom distribution of chromosomal aberrations induced by three chemicals. Mutat Res 63:201–209. 

Surralles J, Autio K, Nylund L, Jarventaus H, Norppa H, Veidebaum T, Sorsa M and Pelonen K . 1997 Molecular cytogenetics analysis of buccal cells and lymphocytes from benzene-exposed workers Carcinogenesis 18 (4) 817-823] 

   

Fracasso et al 2010 examined chromosome aberrations and DNA damage using the alkaline version of the comet assay in the peripheral blood lymphocytes of 33 petroleum industry operators (PIO), 28 service station attendants (SSA), 21 gasoline pump maintenance workers (GPMW). 51 unexposed male workers, working mainly in office jobs, were the controls in this study. The personal benzene exposure was monitored by use of personal samplers (Radiello®) during the work shift; urinary ttMA and SPMA were measured as biomarkers of benzene exposure. Air benzene exposure (median and range) was 40.00 μg/m3 (8.00–260.00), 24.20 μg/m3 (4.60–514.90), 27.80 μg/m3 (1.70–593.50), 5.40 μg/m3 (1.97–16.3) in SSA, GPMW, PIO and controls, respectively. Urinary ttMA and SPMA levels (μg/g creatinine) do not seem to be well correlated with the air benzene levels and the air exposures are too low to use quantitative DFG-based conversions. In controls personal exposure to benzene was influenced by cigarette smoking, but an opposite trend is observed in exposed group. The smoking habits did not substantially modify the metabolites in both groups, the difference being statistically significant only for SPMA in SSA (p = 0.0025). DNA damage was quantified as % of DNA in the tail (TI), tail length (TL), tail moment (TM) and the number of cells with comet. It is to be noted that the authors did not report how long the samples were stored before the comet assay was performed and this could influence the outcome. Significant increases were found for TM, TI and TL in SSA (p = 0.002, p = 0.023, and p = 0.002, respectively) and for TM and TL in PIO workers (p < 0.0001 and p < 0.0001, for TM and TL, respectively) when compared with the control group. Interestingly higher but not statistically significant comet parameters were found in non-smokers almost all groups compared to smokers. Statistically significant positive correlations were found between levels of exposure to benzene and comet parameters (TM, r = 0.509, p = 0.007; TI, r = 0.524, p = 0.005; and number of comets, r = 0.476, p=0.012, in SSA; TM, r = 0.525, p = 0.017 and r = 0.420, p = 0.046, in GPMW and PIO, respectively), whereas negative correlations were found between SPMA and TI (r =−0.465, p = 0.017) or number of comets (r =−0.469, p = 0.016) in the SSA group. The CA analysis was only performed in a subgroup of 19 SSA and controls. No statistically significant difference was determined in CA frequency between exposed workers and control subjects; it was increased (about 60%) in smokers compared to non-smokers in both groups, although the difference was only statistically significant in controls. This study provides no evidence that ppb levels of air benzene exposure affect CA frequency. No personal benzene exposure concentration levels were presented for the SSA subgroup used for the CA analysis, thus the study NOAEC can be considered 12 ppb (2,46-80) based on personal benzene level (med) measured in whole SSA. 

[Fracasso ME, Doria D, Bartolucci GB, Carrieri M, Lovreglio P, Ballini A, Soleo L, Tranfo G, Manno M 2010 Low air levels of benzene: correlation between biomarkers of exposure and genotoxic effects. Toxicol Lett 192: 22-28 ] 

 

Goethel et al 2014 investigated the possible relationship between occupational exposure to benzene and buccal micronucleus frequency and DNA damage using the alkaline version of the comet assay in peripheral blood lymphocytes in 43 gas station attendants (GSA), 34 taxi drivers (TD) and 22 subjects without known occupational exposures (NE group) in Brazil. All participants were males and nonsmokers. Urinary ttMA levels were measured at the end of the work shift, after three consecutive days of exposure as a biomarker of benzene exposure, as neither personal nor ambient benzene levels were measured in all groups. The ttMA levels (mean ± SEM) were found as 117.60 ± 28.0 μg/g creatinine, 439.80 ± 97.30 μg/g creatinine in NE and GSA groups, respectively, but not measured in TD group because this group is not expected to be exposed to benzene according to the authors. Urinary ttMA was positively correlated with time of exposure (r = 0.65; p < 0.001). The ttMA level (assumed to be arithmetic mean ±SEM) was determined to correspond to ~0.6 ± 0,001 ppm air benzene levels using the DFG correlation in GSA, whilst it seems to be less than 95 percentile for general population in NE. DNA damages were visually scored according to tail size into five classes from no tail (0) to maximal (4) tail length, resulting in a single DNA damage score (DNA damage index (DI). Significant increase in the DI and higher DNA damage classes 0 and 1 in both the GSA and TD groups were determined compared to the NE group (p < 0.001). However, in contradiction to the author’s expectation TD has nearly similar Comet results (even higher Class 1 damages) than GSA, which is reported to be the higher exposed group in the study. Thus, increased DNA damage seems to be not related to occupational benzene exposure. In the buccal MN assay, no significant difference was observed among the groups (p > 0.05), with values 0.72 ± 0.17; 2.70 ± 0.67 and 1.30 ± 0.14 MN/1000 cells for the NE, GSA, and TD groups, respectively. Therefore, ttMA level corresponding ~0.6 ± 0,001 ppm air benzene levels in GSA can be considered a NOAEC for buccal MN effects. 

[Göethel G, Brucker N, Moro AM, Charão MF, Fracasso R, Barth A, Bubols G, Durgante J, Nascimento S, Baierle M, Saldiva PH, Garcia SC 2014 Evaluation of genotoxicity in workers exposed to benzene and atmospheric pollutants. Mutat Res Genet Toxicol Environ Mutagen 770: 61-65.] 

 

Pitarque et al 1996 examined the frequency of micronuclei (MN) in cultured blood lymphocytes from a group of 50 male service station workers and 43 male University-based controls in and around Barcelona, Spain. Controls were screened for no previous exposure to suspected genotoxic agents. Attendants wore personal sampling badges near their breathing zone for 16 hours (presumably two eight hour shifts). The benzene metabolite, phenol, was collected from urine at the end of a work shift for both exposed workers and controls. Toluene and xylene air measurements and metabolites were also measured. MN frequencies showed a normal distribution, thus parametric analysis of covariance (ANCOVA) methods were justified and used. Concentrations of benzene were summarised in the 11 service stations where the 50 attendants worked. The mean concentration in the 11 service stations was 0.91 mg/m3, or 0.29 ppm. This concentration is generally too low for phenol to be a reliable biomarker of benzene exposure. However, phenol was significantly increased in service station attendants versus controls. Total MN, as well as binucleated cells with MN, were not significantly different in the station attendants versus controls. ANCOVA indicated no effect from exposure, duration of employment, cigarette smoking nor alcohol consumption. Attendants were older but the age effect was also not significant (p = 0.069). Differential white blood cell counts were also not reduced in exposed workers. In conclusion, this study shows no effect for MN due to occupational benzene exposures, which averaged 0.29 ppm (0.91 mg/m3). Thus, 0.29 ppm is a NOAEL. 

[Pitarque M, Carbonell E, Lapeña N, Marsá M, Torres M, Creus A, Xamena N, Marcos R 1996 No increase in micronuclei frequency in cultured blood lymphocytes from a group of filling station attendants. Mutat Res 367: 161-167.] 

 

    

Carcinogenicity  

 

North et al 2020b (under review) concluded that, there is consistent evidence from epidemiology studies for benzene increasing the risk for AML. MDS is more plausible than any other non-AML potential subtype, but has not been as frequently evaluated. Both AML and MDS are myeloid in nature, and it is plausible both AML and MDS are etiologically related if caused by benzene, with MDS being an earlier or marginally milder dysfunction in the event cascade that includes AML and aplastic anaemia at higher exposures. This myeloid character may initiate with myeloid lineage specific metabolism of hydroquinone or catechol to their respective reactive benzoquinone forms via myeloperoxidase catalysis (Kolachana et al., 1993; Subrahmanyam et al., 1991). These reactive metabolites could act via either indirect genotoxicity from oxidative stress, inhibition of topoisomerase II, or cycles of selective immune-mediated depletion followed by homeostatic proliferation of that myeloid lineage. The likelihood of AML and/or MDS appears strongest amongst all cancer types in relation to benzene (North et al., 2020). It  appears prudent for analysts to consider focusing on benzene AML risk due to potential for inclusion of leukaemia unrelated to benzene to bias potency estimates for benzene-caused leukaemia and the limited number of studies quantitatively considering MDS. 

 

Kolachana, P., Subrahmanyam, V.V., Meyer, K.B., Zhang, L., Smith, M.T., 1993. Benzene and its phenolic metabolites produce oxidative DNA damage in HL60 cells in vitro and in the bone marrow in vivo. Cancer research 53, 1023-1026.

Subrahmanyam, V.V., Kolachana, P., Smith, M.T., 1991. Metabolism of hydroquinone by human myeloperoxidase: mechanisms of stimulation by other phenolic compounds. Archives of biochemistry and biophysics 286, 76-84.