Skip directly to: content | left navigation | search
  1. ATSDR ›
  2. NER ›
  3. Benzene

Benzene-A Review of the Literature from a Health Effects Perspective

    AUTHORS: Ginger L. Gist, Ph.D.
    JeAnne R. Burg, Ph.D.

    AFFILIATIONS: Exposure and Disease Registry Branch, Division of Health Studies, Agency for Toxic Substances and Disease Registry, Public Health Service, U.S. Department of Health and Human Services

    CORRESPONDENCE: Ginger L. Gist, Ph.D., Exposure and Disease Registry Branch, Agency for Toxic Substances and Disease Registry, 1600 Clifton Road, Mailstop E-31, Atlanta, GA 30333 (404)639-6202

    ABBREVIATIONS: ACTH, adrenocorticotrophic hormone
    ATSDR, Agency for Toxic Substances and Disease Registry
    CI, confidence interval
    HPA, hypothalamic-pituitary-adrenocortical
    IARC, International Agency for Research on Cancer
    IDDM, insulin-dependent diabetes mellitus
    NCHS, National Center for Health Statistics
    NCI, National Cancer Institute
    NER, National Exposure Registry
    NHIS, National Health Interview Survey
    NPL, National Priorities List
    NIDDM, non-insulin-dependent diabetes mellitus
    O/E, observed-to-expected ratios
    PNH, paroxysmal nocturnal hemoglobinuria
    PWS, Prader-Willi Syndrome
    SEER, Surveillance, Epidemiology and End Results Program
    USEPA, United States Environmental Protection Agency
    ppb, parts per billion
    ppt, parts per trillion

    KEY WORDS: National Exposure Registry, environmental exposures, benzene

    RUNNING TITLE: Benzene review

    ABSTRACT

    A literature review of the impact on human health of exposure to benzene was conducted. Special emphasis in this report is given to the health effects reported in excess of national norms by participants in the Benzene Subregistry of the National Exposure Registry—people having documented exposure to benzene through the use of benzene-contaminated water for domestic purposes. The health effects reported in excess (p≤.01) by some or all of the sex and age groups studied were diabetes, kidney disease, respiratory allergies, skin rashes, and urinary tract disorders; anemia was also increased for females, but not significantly so.

    BACKGROUND

    Benzene is a volatile, colorless, highly flammable liquid that was first discovered in 1825 by Faraday, who isolated it from a liquid condensed by compressing oil gas (ATSDR, 1996). Today, most (98%) benzene is commercially derived from petrochemical and petroleum refining industries. Benzene is a by-product of various combustion processes, such as forest fires and the burning of wood, garbage, organic wastes, and cigarettes (IARC, 1982a; Fishbein, 1984; Wester et al., 1986; Hattemer-Frey et al., 1990); it is also released to the air from crude oil seeps and volatilizes from plants (Brief et al., 1980).

    Currently, worldwide production of benzene is estimated at approximately 15 million tons (13.6 ×106 metric tons) (Fishbein, 1992). Production in the United States alone is increasing at a rate of 3% annually (ATSDR, 1996), approaching 6 million tons (5.4 × 106 metric tons) of benzene produced in the United States in 1990 and 14.0 billion pounds in 1993 (Fishbein, 1992).

    Benzene is one of the world's major commodity chemicals. Its primary use (95% of production) is as an intermediate in the production of other chemicals, predominantly ethyl benzene (for styrofoam and other plastics), cumene (for various resins), and cyclohexane (for nylon and other synthetic fibers) (ATSDR, 1993). Benzene is an important raw material for the manufacture of synthetic rubbers, gums, lubricants, dyes, and pharmaceutical and agricultural chemicals; it is also found in consumer products such as glues, paints, and marking pens.

    Benzene is also a natural component of crude and refined petroleum. The mandatory decrease of lead alkyls in gasoline has led to an increase in the aromatic hydrocarbon content of gasoline to maintain high octane levels and antiknock properties. In the United States, gasoline typically contains less than 2% benzene by volume (USEPA, 1985), but in other countries benzene concentrations can be as high as 5% (Wolff, 1992).

    Occupational exposures predominantly account for human exposure to benzene. In 1987, approximately 238,000 workers employed by the rubber industry, oil refineries, chemical plants, the shoe manufacturing industry, gasoline storage facilities, and service stations were exposed to benzene (OSHA, 1987; Mehlman, 1991), with an additional 2 to 3 million US workers potentially exposed to benzene (Fishbein, 1992).

    Benzene is ubiquitous in the environment, having been measured in air, water, and human biological samples (Antoine et al., 1986; USEPA, 1986; Wakeham et al., 1986; Wallace, 1986; Wallace et al., 1987; Hattemer-Frey et al., 1990). The major environmental sources include automobile exhaust, automobile refueling, hazardous waste sites, underground storage tanks that leak, waste water from industries that use benzene, chemical spills, chemical manufacturing sites, and petrochemical and petroleum industries (Fishbein, 1992; Edgerton and Shah, 1992).

    As can be seen from the environmental sources listed, inhalation accounts for up to 99% of the total daily intake of benzene (Hattemer-Frey et al., 1990; ATSDR, 1996). Anthropogenic emissions to the air are approximately 34,000 metric tons per year (USEPA, 1989), primarily because of industry-related releases to the environment (TRI 92, 1994). Levels of benzene found in various places are listed in Table 1. Smoking, however, is the largest anthropogenic source of benzene exposure for the general public. The estimates of daily intake of benzene from a single cigarette vary: from 5.9 to 73.0 micrograms (μg) (Brunnemann et al., 1990), from 10 to 31 μg (Thomas, 1986), 30 μg (Powell and Tucker, 1986; Fishbein, 1992), 40 μg (Travis et al., 1990), 57 μg (Higgins et al., 1983; Wallace et al., 1987), and 90 μg (Gilbert et al., 1982). Passive or "secondhand" smoke is also a source of exposure. According to Hattemer-Frey et al. (1990), nonsmokers who live with a smoker have about 30% to 50% higher benzene levels in their breath than do nonsmokers who do not live with a smoker.

    Benzene is also found in other environmental media. It is released to water through discharge of industrial waste water, leachate from landfills, and gasoline leaks from underground storage tanks (CDC, 1994). According to the US Environmental Protection Agency (EPA) STOrage and RETrieval (STORET) database, benzene can be detected in 15% of surface water in the United States, at a median concentration of 5 parts per billion (ppb) (Staples et al., 1985); benzene is also found in unpolluted coastal water (5 to 15 parts per trillion [ppt] ), polluted ocean water (5 to 40 ppt), and polluted coastal water (50 to 175 ppt) (Sauer, 1981). Benzene is found in soil following releases from industrial discharges, land disposal of benzene wastes, gasoline leaks from underground storage tanks, and underground injection of wastes containing benzene (ATSDR, 1996).

    TABLE 1. Levels of benzene in air. (Source: Powell and Tucker, 1986; USEPA, 1987; Shah and Singh, 1988; ATSDR, 1996).

    Location of Air Sample Level (ppb*)
    Urban

    4.0-160

    Remote/Rural

    0.16-3.5

    Suburban residential-remote from traffic

    1.8-4.5

    Indoor

    1.8

    Workplace

    2.1

    Near chemical plant

    14.0

    Near refineries

    9.0

    Gas stations

    <1.0-32

    *ppb = parts per billion

    Ingestion of contaminated food items (Table 2) has been suggested as a potentially important pathway of human exposure to benzene (Hattemer-Frey et al., 1990; NRC, 1980; Gilbert et al., 1982). Benzene, however, has only a low-to-moderate bioconcentration potential in aquatic organisms (Miller et al., 1985; Ogata et al., 1984) and some plants (Geyer et al., 1984); therefore, ingestion of contaminated food items probably accounts for less than 1% of the average daily intake of benzene by the general population of the United States (Gilbert et al., 1982; Hattemer-Frey et al., 1990).

    TABLE 2. Foods containing benzene. (Source: Powell and Tucker, 1986; Lee et al., 1983; Marcus, 1987; Thomas, 1986).

    Foods Containing Benzene (level in μg/kg* where available)

    Vegetables
    Dry red beans
    Leek
    Mushroom
    Onion, roasted
    Parsley
    Potato, cooked peel
    Soybean milk
    Trassi, cooked
    Beverages
    Cocoa
    Coffee
    Jamaican rum (120)
    Tea
    Whiskey

    Fruits
    Apple
    Citrus fruit
    Cranberry and bilberry
    Black currants
    Guava
    Cayenne pineapple
    Strawberry (trace)
    Tomato, hothouse
    Dairy products
    Butter (0.5)
    Blue cheese
    Cheddar cheese
    Other cheese

    		 Meat, Fish, and Poultry
    Cooked beef (2-19)
    Irradiated beef (19)
    Cooked chicken (<10)
    Egg, hard-boiled (500-1900)
    Egg, uncooked (2100)
    Haddock fillet (100 to 200)
    Lamb, heated (<10)
    Mutton, heated (<10)
    Veal, heated (<10)
    Codfish
    Nuts
    Filbert, roasted
    Peanut, roasted
    Macadamia nut

    *μg/kg = micrograms per kilogram

    Toxicokinetics

    Absorption of benzene varies with route of exposure. In humans, respiratory uptake has been determined to vary from approximately 47% (Nomiyama and Nomiyama, 1974) to 80% (Srbova et al., 1950), although dermal absorption can range from 0.05% to 0.2% (Franz, 1984). Absorption data for oral exposure in humans is not available; however, in animals, absorption rates following oral exposure to benzene were found to be from 90% to almost 100% (Parke and Williams, 1953; Sabourin et al., 1987) and is vehicle-dependent.

    Benzene must be metabolized to exert its toxic effects (Smith et al., 1989; Sabourin et al., 1989a; Huff et al., 1989; Subrahmanyam et al., 1991; ATSDR, 1993; Snyder, 1987; Seaton et al., 1995; Medinsky et al., 1995; Irons, 1985; Snyder and Hedli, 1996); however, this process is complex, consisting of multiple pathways (Figure 1). Benzene is metabolized by cytochrome P-450-dependent multifunction oxidase enzymes (Medinsky et al., 1994). According to Birnbaum (1994), benzene metabolism leads either to more toxic products, via pathways involving those leading to ring breakage and benzoquinone, with muconic acid and hydroquinone, respectively, as conjugates, with prephenyl and phenyl mercapturic acids and phenyl conjugates as the markers, respectively. A more in-depth discussion can be found in ATSDR (1996).

    Phenol, hydroquinone, and catechol are the major metabolites of benzene in mammals (Medinsky et al., 1989; Smith et al., 1990; Ciranni et al., 1991; Ong et al., 1996); however, these metabolites can affect each other's rate of metabolism because they are substrates for the same cytochrome P-450 enzymes (Cox, 1991; Medinsky et al., 1994; Medinsky et al., 1995). According to Cox (1991), these interactions might help to explain the nonlinear relation between administered benzene concentrations and internal doses of metabolites. Indeed, it has been reported that a higher proportion of metabolites are produced at lower exposure concentrations (Sabourin et al., 1989a,b, 1990; Henderson et al., 1989).

    Finally, sex differences are believed to play a role in the toxicity of benzene; however, the literature has been mixed. A number of studies have reported the female to be more susceptible in both humans (Mallory et al., 1939; Ito, 1962a; Sato et al., 1975) and animals (Hirokawa, 1955; FIGURE 1. Scheme for metabolism of benzene. (Adapted from: Snyder and Hedli, 1996; Henderson et al., 1992).

    Figure 6

    Ito, 1962b; Ikeda, 1964; Leong, 1977; Meyne and Legator, 1980; Siou et al., 1981; Ciranni et al., 1991). In contrast, male mice (Gad-El-Karim et al., 1984, 1986; Tice et al., 1989) are more susceptible to benzene-induced chromosomal damage than are female mice. Still other studies (Li et al., 1994; Yin et al., 1994, 1996) have found male and female humans to be equally susceptible to the effects of benzene exposure. It is important to note that the end points of these studies are variable.

    The health effects of benzene exposure also depend both on the species exposed (Snyder, 1987; Neun et al., 1992; Orzechowski et al., 1995; Zhu et al., 1995) and the route of exposure (Cox, 1991). All things being equal, humans tend to form lower internal doses of reactive metabolites than do animals (Reitz et al., 1989); route of exposure has little or no effect on subsequent metabolism of benzene in humans (Sabourin et al., 1987, 1988, 1989a,b). It should be noted, however, that Goldstein (1977) postulated that humans might have a genetic predisposition to benzene toxicity. Finally, age can also play a role in metabolism of benzene, with younger animals displaying a higher rate of metabolism, and a greater susceptibility to toxic effects, than do older animals (McMurry et al., 1994a,b; Snyder and Kocsis, 1975).

    Following inhalation exposure, most benzene is excreted unchanged in exhaled air. Human excretion of absorbed benzene involves a biphasic urinary excretion (Sherwood, 1988) of conjugated derivatives (sulfates and glucuronides). Animal data show a similar pattern; that is, unmetabolized benzene is excreted primarily through exhalation, but metabolized benzene is excreted mainly in the urine (ATSDR, 1996).

    Ethanol has been shown to alter benzene metabolism. For example, ethanol has been shown to induce CYP2E1, a cytochrome P-450 enzyme responsible for benzene metabolism (Johansson and Ingleman-Sundberg, 1988; Koop et al., 1989; Guenrich et al., 1991; Gut et al., 1993; Medinsky et al., 1994). Nakajima et al. (1985) found that pre-exposure of male Wistar rats to ethanol not only increased the rate of metabolism of benzene by hepatic microsomes sixfold, but also significantly increased the rate of clearance of benzene from the blood. Ethanol has also been shown to enhance the toxicity of benzene in humans (Whitehead et al., 1978; Wynagaarden and Smith, 1985; Kristensson-Aas et al., 1986; Midzenski et al., 1992), as well as in animals ( Snyder et al., 1981a; Baarson et al., 1982; Rosenthal and Snyder, 1984; Nakajima et al., 1985; Seidel et al., 1990; Baarson and Snyder, 1991).

    Other substances may also affect the metabolism of benzene. Workers exposed to a combination of benzene and toluene produced significantly lower urinary phenol (a biomarker for benzene exposure) than those exposed to either benzene or toluene alone (Inoue et al., 1988); toluene has also been shown to lower the toxicity of benzene in animals markers, or to detoxification, via pathways leading to mercapturic acid products and phenyl (Hsieh et al., 1990a). It should be noted that Aroclor 1254 and phenobarbital are also known to alter the toxicity and metabolism of benzene (ATSDR, 1993; Medinsky et al., 1994).

    Physiologically based pharmacokinetic (PBPK) models are used to allow interdose, interspecies and interroute pharmacokinetic extrapolation as well as prediction of target tissue exposure (Spear et al., 1991). According to Cox and Ricci (1992), PBPK models are useful tools to use to correct risk assessments for nonproportional relations between administered and internal doses in test species; differences between routes of administration in terms of internal doses formed from a given amount of administered benzene; and interspecies metabolic differences in the production of external doses from administered doses. Several models are currently available (Medinsky et al., 1989ab; Woodruff et al., 1989; Travis et al., 1990; Bois et al., 1991a; Bois and Paxman, 1992); each varies in structure, the parameter values assigned, the data from which the values were derived, and metabolic constants. For discussions of these various models, the reader is referred to Bois et al., (1991b), Spear et al. (1991), Cox and Ricci (1992), and ATSDR (1996).

    THE BENZENE SUBREGISTRY

    In 1989, benzene was selected as a primary contaminant for establishment of a subregistry of the National Exposure Registry (ATSDR, 1995). At the time of its selection, benzene was one of the most prevalent contaminants found at National Priorities List (NPL) sites, having been identified at 746 (78%) of NPL sites. Benzene-contaminated groundwater was used as a source of drinking water at 56% (n = 263) of the NPL sites with benzene. Of these sites, 29% (n = 76) had private well systems, 7% (n = 19) had municipal systems, and 60% (n = 158) used both private and municipal systems to provide residents with drinking water (ATSDR, 1988).

    A key purpose of the Benzene Subregistry is to determine if there is excess reporting of adverse health conditions for registrants-that is, people with documented exposure to benzene-compared with a national sample. To date, this objective has been pursued by comparing Benzene Subregistry data about health conditions with National Health Interview Survey (NHIS) data (NCHS, 1990a,b) and comparing Benzene Subregistry cancer outcomes with cancer incidence data from the National Cancer Institute Surveillance, Epidemiology and End Results (SEER) program (NCI, 1989). Health, demographic, occupational, and environmental information was collected for 1,143 benzene-exposed persons (1,127 living, 16 deceased) who met the eligibility criterion for National Exposure Registry participation; that is, they resided for more than 30 consecutive days during the period of exposure at a site address. Registrants were exposed to a maximum of 66 ppb benzene in the water they used for domestic purposes. The remaining site information and data analyses are discussed in detail elsewhere (ATSDR, 1995).

    Comparisons of reporting rates for 25 health outcomes for Benzene Subregistry and NHIS data were made using age and sex as covariates. The results (Table 3) revealed increased, but not statistically significant (p≤.01), reporting rates by all but two female age groups and one male subgroup for anemia or other blood disorders. (Note: This level was intentionally conservative in order to control the likelihood of false negative or false positive results.) Several statistically significant increases were found for some age and sex subgroups for other outcomes (diabetes, kidney disease, respiratory allergies, skin rashes, and urinary tract disorders). Registrants reported some health outcomes (arthritis, rheumatism, or other joint disorders; asthma, emphysema, or chronic bronchitis; hearing impairment; and speech impairment) less frequently (statistically significant) than participants in the NHIS. These outcomes are often self-diagnosed; therefore, the decreased reporting by registrants was an expected result because the Registry data had the qualifier of physician confirmation and the NHIS data did not.

    TABLE 3. Summary results of Benzene Subregistry-National Health Interview Survey (NHIS) comparison.

    Disease Category Age Groups (years)
    0-9 10-17 18-24 25-34 35-44 45-54 55-64 ≥65 All All
    M F M F M F M F M F M F M F M F M F

    Anemia or other
    blood disorders

    		  

    I

    		  

    I

    		  

    I

    		        

    I

    I

    I

    		  

    I

    		        
    Arthritis                                    

    R

    Asthma,
    emphysema

    		                                    

    R
    Diabetes  

    -

    X

    -

    		                              

    Kidney disease

    		                        

    X

    X

    		          

    Hearing
    impairment

    		                                    

    R

    Respiratory
    allergies

    X

    X

    		                                  
    Skin rashes

    X

    X

    		                        

    X

    X

    		      

    Speech
    impairment

    		                                    

    R

    Urinary tract
    disorders

    		                                  

    X

    		  

    I = Observed/expected ratio greater than 3.4, but not statistically significantly different (p ≤ .01).

    X = Statistically significant differences (p ≤ .01), Benzene Subregistry rate higher.

    R = Statistically significant differences (p ≤ .01), NHIS rate higher.

    ─ = Insufficient data.

    HEALTH CONDITIONS

    Anemia or Other Blood Disorders

    The outcome anemia or other blood disorders was reported in greater numbers for the Benzene Subregistry population at baseline data collection; however, the increase was not statistically significant. It is notable that all of the observed-to-expected (O/E) ratios for females exceeded 1.8 (most exceeded 3.4); however, they were not statistically different from 1.0. For this outcome, the time frame of reporting was “within the last 12 months.”

    Benzene is a known hematopoietic poison and bone marrow depressant (Goldstein, 1977, 1983, 1985, 1989a; Marcus, 1987; Savitz and Pearce, 1987; Ning et al., 1991; Zhu et al., 1995), having first been reported to produce anemia and leukopenia as long ago as 1897 (Santesson, 1897). According to ATSDR (1993), benzene was used at levels ranging from 43 to 71 milligrams per kilogram per day (mg/kg/day) as a treatment for leukemia before 1913. Case reports show the patients had decreased erythrocyte and leukocyte levels (Selling, 1916); however, it is uncertain whether these observations were a result of the disease or the treatment. Since that time, a variety of hematopoietic disorders, such as pancytopenia, aplastic anemia, preleukemia, leukopenia, acute myeloblastic anemia, thrombocytopenia, leukopenia associated with thrombocytopenia, nucleated red blood cells, and eosinophilia (NIOSH, 1974, 1988; Aksoy, 1980, 1989a; USEPA, 1980; Goldstein and Snyder, 1982; IARC, 1982a,b; Aksoy et al., 1987; Rinsky et al., 1987; Goldstein, 1988; Paci et al., 1989; Rinsky, 1989; Ciranni et al., 1991; Cox, 1991; Ruiz et al., 1991; Midzenski et al., 1992; ATSDR, 1993; Medinsky et al., 1994; Travis et al., 1994; Niculescu et al., 1995) have been associated with exposure to benzene. Myelofibrosis, a myeloproliferative disease in which the bone marrow is replaced with fibrous tissue, has also been associated with benzene exposure (Zoloth et al., 1986; Schottenfeld and Fraumeni, 1982; Wintrobe et al., 1981).

    Worker studies have provided the basis for the description of hematologic outcomes in humans. Fishbeck et al. (1978) reported that 282 workers exposed at a chemical factory to 25 ppm benzene for an average of 9 years and 7 months had an increased mean corpuscular level at the end of their high exposure period, but that levels had returned to normal 11 years later after exposure had ceased. Townsend et al. (1978) reported slight decreases in red blood cell counts in this same population, but these decreases were not correlated with levels or durations of exposure (ATSDR, 1993). In a study of 459 workers in the rubber industry exposed to 15 to 75 ppm benzene from 1940 through 1975, decreased blood cell counts that did not persist when the exposure was decreased were reported; significant decreases in leukocyte and erythrocyte counts and hemoglobin were reported during the years of highest exposure (Kipen et al., 1989). Doskin (1971) reported lymphocytosis in workers exposed to 3 to 13 ppm benzene for 1 to 3 years. Finally, Midzenski and coworkers (1992) reported that, of 15 workers acutely exposed over several days to high concentrations of benzene (>60 ppm), 9 had at least one hematologic abnormality consistent with benzene exposure. One year later, six workers (40%) had persistent abnormalities; six (40%) had numerous large granular lymphocytes on peripheral blood smears.

    Aksoy and coworkers (1971) followed the health of 217 male Turkish workers exposed to a maximum of 210 ppm benzene for a period of 4 months to 17 years and found leukopenia, thrombocytopenia, leukopenia associated with thrombocytopenia, pancytopenia, and eosinophilia. In a second study, Aksoy et al. (1974) reported preleukemia and acute leukemia in 28,500 workers exposed to 210 to 650 ppm benzene for 1 to 15 years. It should be noted that the clinical features of preleukemia include one or more of the following: pancytopenia, bone marrow hyperplasia, pseudo-Pelger-Heut anomaly, and splenomegaly. Finally, of 44 pancytopenic patients followed postexposure, 23 went into complete remission; however, the others died from complications of the disease or from a successive blood disorder, such as myeloid dysplasia or leukemia (Aksoy and Erdem, 1978).

    Yin et al. (1987a, 1989) followed 528,729 Chinese workers, mainly from paint, shoe, rubber, leather, or adhesives factories, who were exposed to a range of benzene levels from <13 ppm (the majority) to 264 ppm. These workers reported leukopenia, aplastic anemia, and leukemia. The authors reported that these findings were similar to those of an additional 2,740 shoe factory workers they were also following.

    A few cases of paroxysmal nocturnal hemoglobinuria (PNH) have been reported following benzene exposure (Aksoy et al., 1975; Davies and Levine, 1986; Kwong and Chan, 1993). PNH is a chronic, acquired blood cell dysplasia in which there is a mutation of hematopoietic stem cells resulting in the production of erythrocytes, granulocytes, and platelets that are abnormally sensitive to complement lysis because of a lack of complement-inactivating proteins. PNH can occur concomitantly or sequentially in patients with aplastic anemia, or vice versa (Young, 1992; Kwong and Chan, 1993), and can also be a paraneoplastic disorder (Goldstein, 1977).

    Several studies were located in which no adverse hematologic effects were reported following benzene exposure. Tsai and coworkers (1983) reported no adverse hematologic effects in refinery workers exposed to low levels of benzene (less than 1 ppm) for 1 to 21 years. Wong (1987a) found substantially lower death rates for diseases of the blood in refinery workers in California when compared with workers with no occupational exposure to benzene. This result was confirmed in a followup of this cohort (Dagg et al., 1992); however, neither study included women in most of the analyses and a strong “healthy worker” effect was believed to have played a significant role in the outcomes. In a study of 200 workers exposed to 0.01 to 1.40 ppm benzene per 8-hour time-weighted average for a 10-year period, Collins et al. (1991) found no difference between the exposed workers and 268 nonexposed workers in the same plant.

    Guiguet et al. (1995) conducted a case-control study of new cases of aplastic anemia enrolled in a French national register of the disease. No differences were found between cases and controls for occupation, based on a 15-year occupational history, or for exposure to solvents. It should be noted, however, that a positive relationship between exposure to glues and paints-both of which could contain benzene-was found. The authors indicated that further investigation was needed for these relationships.

    Animal studies have shown that chronic benzene exposure has a hematotoxic effect on hematopoietic stem cells, erythropoietic cells, and granulopoietic progenitors, as well as on erythrocytes and leukocytes (Wolf et al., 1956; Snyder et al., 1980; Cronkite et al., 1982, 1985; Ward et al., 1985; NTP, 1986; Thomas, 1986; Hsieh et al., 1988b; Luke et al., 1988a,b; Huff et al., 1989; Seidel et al., 1989a,b; ATSDR, 1993). Using a mathematical model of murine hematopoiesis, Scheding et al. (1992) found (1) erythropoietic cells were the most sensitive, (2) granulopoietic cells were about half as sensitive as erythropoietic, and (3) hematopoietic stem cells exhibited a sensitivity that ranged between that of erythropoietic and granulopoietic cells. It should be noted, however, that some reports on differentiated granulopoietic marrow and blood cells are ambiguous (Snyder et al., 1978; Snyder et al., 1980, 1981b, 1982; Baarson et al., 1982; Rosenthal and Snyder, 1984; Cronkite et al., 1985; Scheding et al., 1992). Fetuses and offspring of pregnant mice exposed to benzene have also shown long-term functional changes in hematopoiesis (ATSDR, 1993; Keller and Snyder, 1986, 1988; Schardein and Keller, 1989).

    Vacha et al. (1990) reported that ferrokinetic indicators of benzene-exposed mice showed only a slightly enhanced production of heme and erythrocytes in the spleen; however, the life spans of late erythroblasts and circulating erythrocytes were reduced. Low-molecular-weight bleomycin- detectable iron was also observed in the bone marrow following administration of benzene to female rats (Pandya et al., 1990). The authors suggested that the accumulation might lead to the formation of tissue-damaging species, such as lipid peroxide radicals or superoxide anions.

    In other animal studies, Aoyama (1986) exposed mice to 200 ppm benzene for 6 hours per day for 7 days or 50 ppm for 14 days. He reported that the mice had decreased levels of white blood cell counts in both the blood and spleen. Similar results (decreasing circulating leukocytes) were seen for rabbits, rats, and guinea pigs (NTP, 1986; Huff et al., 1989; Wells and Nerland, 1991; Li et al., 1992) following various exposure scenarios. Rats (Hoechst, 1978) and mice (Hoechst, 1978; Cronkite et al., 1982, 1985, 1989; Hsieh et al., 1988b, 1990b) were also found to have leukopenia, but in conjunction with anemia, following various exposure scenarios. Finally, mice exposed to 300 ppm benzene for 6 hours per day, 5 days per week, for 16 weeks were found to have granulocytic hyperplasia in the bone marrow (Farris et al., 1993).

    It has long been recognized that animal responses to benzene are varied and can depend on factors such as species, strain, duration of exposure, and route of exposure (continuous or intermittent) (Deichmann et al., 1963; Longacre et al., 1980; Snyder et al., 1980; Gaido and Wierda, 1985; Snyder, 1987; Neun et al., 1992; Corti and Snyder, 1996). Effects reported include anemia (Cronkite et al., 1982; Rozen et al., 1984; Ward et al., 1985) and bone marrow abnormalities (Toft et al., 1982; Ward et al., 1985), as well as the previously mentioned abnormalities in the numbers of specific cell types (Wolf et al., 1956; Cronkite et al., 1982, 1985; Ward et al., 1985; Luke et al., 1988a,b; Seidel et al., 1989b). Depression of lymphocytes and nucleated red cells formed in the spleen have also been noted in mice (Baarson et al., 1984). Snyder et al. (1980) reported that mice chronically exposed to benzene intermittently over their lifetime developed bone marrow hypoplasia and pancytopenia (ATSDR, 1993).

    As stated previously, the hematotoxic effects of benzene are increased by ethanol. This has been demonstrated for blood cell counts and bone marrow cellularity in mice and rats (Snyder et al., 1981a; Baarson et al., 1982; Nakajima et al., 1985) and for the hemopoietic progenitors (Seidel et al., 1990). It can be seen that the combination of ethanol and benzene leads to the same effects as an increase in the benzene dose or a prolongation of the exposure period. Of particular interest is the reported transient appearance of nucleated red cells (normoblasts) in the circulating blood of animals exposed to benzene and ethanol. It has been shown that the bone marrow is the source of these normoblasts, because the cells also appear in the circulating blood of exposed, splenectomized animals (Rosenthal and Snyder, 1984; Baarson and Snyder, 1991).

    Several authors have reported recovery of hematopoiesis following termination of chronic benzene exposure (Snyder et al., 1980; Cronkite et al., 1982, 1985; Seidel et al., 1989a,b; Dempster and Snyder, 1990; Baarson and Snyder, 1991). These studies have shown that hematopoietic stem cell recovery is rapid and complete for short exposure periods (less than 8 weeks); however, longer exposure periods result in a considerable delay in stem cell recovery, indicating a benzene-induced residual hematopoietic injury (Cronkite et al., 1985; Scheding et al., 1992).

    Negative results have also been reported for animal studies. Deichmann et al. (1963) found no effects in rats exposed to benzene levels ≤31 ppm for 7 hours per day, 5 days per week, for 88 to 126 days. Jenkins et al. (1970) found no hematologic effects in rats, guinea pigs, or dogs exposed continuously to 17.6 ppm benzene for up to 127 days. Svirbely et al. (1944) exposed rats to 1,000 ppm benzene for 7 hours per day, 5 days per week for 28 weeks and found only a very slight reduction in white cells counts, but no overt signs of toxicity. Other animal studies also reported negative results following benzene exposure ( Frash et al., 1976; Speck et al., 1978; Murray et al., 1979).

    It is believed by many investigators that the myelotoxic (marrow damaging) and hematotoxic (blood-system damaging) effects of benzene are probably mediated by the interactions of two or more benzene metabolites with each other and with target cells in the bone marrow (see Figure 2) (Baarson et al., 1984; Gaido and Wierda, 1984, 1986; Post et al., 1985; Keller and Snyder, 1986; Lewis et al., 1988a,b; Dempster and Snyder, 1989, 1990; Goldstein, 1989b; Smith et al., 1989; Snyder et al., 1989; Thomas et al., 1989; Guy et al., 1990, 1991; Subrahmanyam et al., 1990; Cox, 1991; Eastmond, 1993; Chen and Eastmond, 1995b; Henschler and Glatt, 1995; Niculescu et al., 1995; Rao and Snyder, 1995; Hedli et al., 1996; Irons and Stillman, 1996). The effects seen include delay of bone marrow cell life cycles and change of mitotic indices; killing of stem cells in specific phases of the cell life cycle; inhibition of cellular functions; and damage to the stromal microenvironment that normally regulates stem cell proliferation and differentiation. These cytotoxic effects have been demonstrated experimentally (Cronkite et al., 1989; Cox, 1991;Niculescu et al., 1995; Schoeters et al., 1995; Irons and Stillman, 1996). Niculescu and Kalf (1995) have reported that interleukin-1 alpha (IL-1α), when added concomitantly with benzene, will prevent benzene-induced depression of bone marrow cellularity as a function of dose; however, it has been demonstrated that benzene can interfere with the production of IL-1α (Kalf et al., 1996). Benzene metabolites interact strongly in inhibiting the growth and functioning of erythrocytes (Chen and Eastmond, 1995a). This interaction has been demonstrated in a study by Snyder et al. (1989), in which benzene suppression of erythropoiesis was measured by the uptake of radiolabeled iron into the red blood cell hemoglobin of female mice, and in a study by Dempster and Snyder (1990), in which short-term benzene exposure induced a growth advantage for granulocytic cells in both the bone marrow and spleen of exposed mice.

    FIGURE 2. Schematic of blood cell differentiation and maturation (Source: Thomas, 1986).

    The ability of benzene metabolites to suppress lymphocyte growth and function in vitro correlates both with their oxidation capacity and with their concentration at the target site (Irons et al., 1980; Greenlee et al., 1981; Pfeifer and Irons, 1981, 1982, 1983; Hsieh et al., 1990a). For example, hydroquinone and benzoquinone inhibit proliferation and differentiation in lymphocytes in culture at noncytotoxic concentrations; phenol or catechol suppress lymphocyte growth or function at concentrations that result in cell death (Hsieh et al., 1990a). In addition, hydroquinone and catechol have been shown to reduce the number of spleen and bone marrow progenitor B lymphocytes and to inhibit polyclonal plaque-forming cells (Wierda and Irons, 1982; Hsieh et al., 1990a).

    Benzene affects macrophages as well as lymphocytes (Lewis et al., 1988a; Thomas et al., 1989; Niculescu and Kalf, 1995). According to Cox (1991), experiments with cultured macrophages have shown that benzene metabolites are effective inhibitors of hydrogen peroxide releases from stimulated macrophages. (Note: The release of hydrogen peroxide normally allows macrophages to kill invading bacteria and other threats.) In addition, p-benzoquinone inhibits phagocytosis and cytolysis of tumor cells by macrophages.

    There is considerable evidence supporting the hypothesis that p-benzoquinone is a major source of benzene health effects (Snyder, 1987; Smith et al., 1989, 1990; Cox, 1991). Both hydroquinone and p-benzoquinone (as well as catechol and 1,2,4-benzenetriol at higher concentrations) reduce the ability of the stromal microenvironment to support granulocyte or macrophage stem cell formation in vitro (Schoeters et al., 1995; Zhu et al., 1995). Benzene exposure decreases the subsequent ability of the marrow-adherent layer of stromal cells to support normal stem cell differentiation in vitro (Kalf, 1987; Smith et al., 1989).

    In view of the reported literature, it is possible that chronic exposure to benzene in the environment can result in blood disorders. As mentioned earlier, some researchers have reported that a higher proportion of the toxic metabolites of benzene are produced at lower exposure concentrations (Henderson et al., 1989; Sabourin et al., 1989a, 1990), a phenomenon that might have played a role in the trends reported here. In addition, excess reporting has occurred mainly for females, indicating a greater susceptibility to benzene intoxication, as has been indicated by some researchers ( Mallory et al., 1939; Hirokawa, 1955; Ito, 1962b; Ikeda, 1964; Sato et al., 1975; Meyne and Legator, 1980; Siou et al., 1981; Ciranni et al., 1991).

    Arthritis, Rheumatism, or Other Joint Disorders

    There was a decrease in Benzene Subregistry registrants' reports (46 observed, 101.6 expected, O/E = 0.45, 99% confidence interval [CI] = 0.30, 0.66) of arthritis, rheumatism, or other joint disorders compared with those of the NHIS population. As discussed earlier, arthritis, rheumatism, and other joint disorders are conditions that are likely to be self-diagnosed, and, therefore, increased reporting by the NHIS population would be expected; however, there does not appear to be excess reporting by registrants. The time frame was “within the last 12 months.”

    No studies were located specifically addressing the potential effects of benzene on the development of arthritis, rheumatism, or other joint disorders in either humans or animals. There have been reports, however, of skeletal malformations in the offspring of benzene-exposed animals. Benzene can cross the placental barrier and is found in cord blood at levels equal to or greater than those observed in the mother (Dowty et al., 1976); thus, the fetus might be exposed to relatively high levels of benzene and its metabolites. In one study, Murray et al. (1979) observed offspring of CF-1 mice and those of New Zealand rabbits. Mice exposed to 500 ppm benzene for 7 hours per day on days 6 through 15 of gestation had offspring with mean fetal body weights that were significantly lower (p < 0.05) and minor skeletal variants-delayed ossification-that were significantly greater than those of controls. The variants were considered to be indicative of delayed development, but not of major malformations. The rabbits were exposed in a similar manner but their offspring demonstrated no treatment-related effects.

    In another study, pregnant Sprague-Dawley rats were exposed to 0, 10, 50, or 500 ppm benzene for 7 hours per day on days 6 through 15 of gestation (Kuna and Kapp, 1981). Offspring of the 500 ppm group had significantly reduced mean crown-to-rump length, reduced fetal body weight, and delayed ossification. Four of the offspring from this group had skeletal anomalies- exencephaly, angulated ribs, and out-of-sequence ossification of the forefeet. Offspring of the 50 ppm group also demonstrated reduced fetal body weights and delayed ossification; no other differences were reported for this group or the remaining groups.

    In a similar study, Green et al. (1978) exposed Sprague-Dawley rats to 100, 300, or 2,200 ppm benzene for 6 hours per day on gestation days 6 through 15. A significant decrease in mean fetal weight was reported for the 2,200-ppm group; decreased crown-to-rump length was also significantly reduced in this group. Female offspring in the 300 and 2,200 ppm groups displayed significantly increased delayed ossification of sternebrae; this effect was not seen for the males. In addition, only the females in the 2,200 ppm group had an increase in missing sternebrae. Other studies have also reported skeletal variations following benzene exposure during gestation, including cleft palate, agnathia, and micrognathia (Watanabe and Yoshida, 1970) as well as extra ribs and fused sternebrae (Hudák and Ungváry, 1978).

    Similar studies reporting negative results were also found. Nawrot and Staples (1979) administered 0.3, 0.5, or 1.0 mg/kg/day to pregnant CD-1 mice by means of gavage on gestation days 6 through 15. Even at the highest dose level there were no statistically significant changes in the incidence of malformations. No malformations were seen in offspring of mice exposed to 156.5 or 313.0 ppm benzene for 12 hours per day on gestation days 6 through 15 (Ungváry and Tatrai, 1985). No skeletal variants were reported by other researchers in similar studies with rats (Tatrai et al., 1980; Coate et al., 1984).

    It is interesting to note that Benzene Subregistry males 0 through 9 years of age had an O/E ratio of 3.89; females in the same age group had an O/E ratio of 4.35. Both of these values, however, were based on one case of arthritis, rheumatism, or other joint disorders each; therefore, no conclusions could be drawn.

    Cardiovascular

    No statistically significant results were seen for registrants for either hypertension or the effects of stroke (although 4.6 cases of stroke were expected and 10.0 were reported). It should be noted that, for males 25 through 45 years of age, 4 cases of stroke were found and less than 1 was expected. There are few reports in the literature of cardiovascular problems following benzene exposure, even though it has been suggested that severe, acute benzene poisoning can result in direct myocardial damage (Von Oettinger, 1958). Yin et al. (1994) reported that cardiovascular disorders were the second most common cause of death among benzene workers.Zoloth et al. (1986) found a significantly elevated risk of arteriosclerotic heart disease (PMR = 113) in a working population exposed to benzene and other solvents. Lloyd et al. (1977) also found increased mortality due to heart disease in a similar worker population.

    Case reports of cardiovascular problems following benzene exposure were more commonly found. Avis and Hutton (1993) found ventricular fibrillation to be the cause of death for three people having acute benzene poisoning resulting from an industrial accident aboard a chemical cargo ship. In addition, the brain of each individual appeared grossly normal but showed microscopic evidence of prominent vascular congestion on autopsy. According to Fielder and coworkers (1982), the postmortem findings for a number of fatalities attributed to the inhalation of high concentrations of benzene included extensive petechial hemorrhages in the brain and pericardium. These findings are similar to the autopsy finding of cerebral edema reported by Winek and Collom (1971). Davies and Levine (1986) reported on two series of case studies in which arterial hypertension was cited as an outcome of benzene exposure. These studies included one by Reznik in 1974 in which arterial hypertension was reported in 150 patients with a 15- to 26-year history of chronic poisoning with benzene derivatives. Of the exposed individuals, 52% had either arterial hypertension or arterial pressure within the limits of a “transition zone” (undefined). Davies and Levine (1986) also reported that in 1979 Karmaz found hypercholesterolemia and hyperphosphatidemia in 250 workers employed in coke-benzene production.

    Negative reports of cardiovascular problems following benzene exposure were also found. For example, Dagg et al. (1992) found significantly lower rates of heart disease and other conditions in a worker population.

    Sudden deaths from cardiovascular problems, predominantly ventricular fibrillation, have also been observed in laboratory animals following acute benzene poisoning (Leong, 1977). For example, Nahum and Hoff (1934) reported extra systoles and ventricular tachycardia of the prefibrillation from electrocardiograms of cats and monkeys, which reflected the effects of acute benzene inhalation exposure on the heart muscle. Moravi et al. (1976) found that rats exposed to benzene invariably exhibited ventricular fibrillation following respiratory arrest. Finally, Vidrio et al. (1986) and Magos et al. (1990) found that benzene increased the arhythmogenic action-specifically ventricular arrhythmias-of epinephrine in rats exposed to 3,526 to 8,224 ppm benzene.

    Diabetes

    A statistically significant increase between the number of cases of diabetes reported by the NHIS and the number reported by Benzene Subregistry respondents was found for the male age group 10 through 17 years (2.0 observed, <0.1 expected, O/E = 40.21, 99% CI = 2.08, 186.45); for females, 0 were expected and 1.0 was reported (statistics could not be calculated because of the zero denominator). It is important to note that the expected value was close to zero (0.05) for the 10 through 17 years of age group, but 3.0 cases were observed . Also notable is that two of the cases were reported by siblings; none of the parents reported having diabetes. Furthermore, given the dates of exposure for the site, these subgroups would have been exposed in utero.

    Few studies mention diabetes in humans as a potential outcome of benzene exposure. In a survey of dry-cleaning workers in Nagoya City, Japan, Takeuchi and coworkers (1981) found one case of diabetes mellitus among workers in one shop who were directly engaged in dry cleaning using a mixture of petroleum solvents and tetrachloroethylene. In a mortality study of refinery workers in California, Wong (1987a,b) found substantially lower death rates from diabetes, a finding that was confirmed in an update of the cohort (Dagg et al., 1992). It should be noted that both of these studies excluded women from most of the analyses and that the authors believed a strong “healthy worker” effect was responsible for much of the outcome.

    Diabetes is a chronic syndrome of impaired carbohydrate, protein, and fat metabolism secondary to insufficient secretion of insulin or target tissue insulin resistance that results in elevated blood glucose levels. There are two major types of diabetes: insulin-dependent diabetes mellitus (IDDM or type I), in which exogenous insulin is necessary, and non-insulin-dependent diabetes mellitus (NIDDM or type II), in which exogenous insulin is not required. Although IDDM can occur at any age, peak onset is 12 years of age. NIDDM typically occurs after 40 years of age, with peak onset occurring from 50 to 60 years of age. Because the excess of diabetes reported for the Benzene Subregistry occurred in the group aged 10 to 17 years, it is likely that IDDM is the type of diabetes most prevalent; therefore, the discussion that follows focuses on this type of diabetes.

    The most generally accepted hypothesis of IDDM pathogenesis follows this course: genetic susceptibility → environmental trigger → immunological activation → progressive beta cell (pancreas) destruction → abnormalities of insulin secretion → early metabolic abnormalities → onset of symptoms (Gale and Bottazzo, 1986). Although there are genetic markers highly associated with the incidence of IDDM (for example, the non-Asp-57 marker), fewer than 3% of the people within a population who have such a genetic marker develop the disorder. Almost all people who develop IDDM have the marker and only a very small percentage having the marker will develop the disease, presumably because of the interaction of host and environmental factors. It might be suspected, therefore, that the susceptible people who develop diabetes have been more exposed than those who do not develop this disease (Tull and LaPorte, 1992).

    Although genetics has been demonstrated to play a major role in the development of IDDM, it cannot explain all cases, which strongly suggests that an environmental influence is superimposed on the heritable component of IDDM (Nathan, 1993). Indeed, at least one study (Gamble, 1980) has suggested that sibling pairs are more likely to have onset of diabetes within a year of one another than would be expected by chance. As noted previously, two of the cases of diabetes reported for the Benzene Subregistry were those of siblings. A similar observation of clustering of time of onset seems to occur in the monozygotic twins of diabetic parents (Olmos et al., 1988). Goldstein (1988) has also suggested that there might be a genetic predisposition to benzene toxicity. This evidence can be interpreted as suggesting a common exposure to an agent that either initiates or precipitates IDDM (Vadheim and Rotter, 1992). In addition, the stability of the annual incidence of IDDM implies a comparable stability in the prevalence of the causal environmental factor, which is uncharacteristic of most viruses and bacteria (Gamble, 1980).

    According to the World Health Organization DIAMOND Project Group on Epidemics, a major difficulty in the area of IDDM research-despite strong epidemiologic evidence that environmental agents are potent causes of IDDM (Diabetes Epidemiology Research International, 1987)-is that the identification of such agents has been elusive. It is noteworthy that several recent epidemiologic studies have reported that the incidence of IDDM is increasing, suggesting that long-term changes in the environment are altering the probability of eventual diabetes.

    According to Palmer and Lernmark (1990), the possible environmental mechanisms involved in the development of IDDM include causing direct toxicity to beta cells, triggering of an immune reaction against beta cells, inducing increased insulin need that cannot be met by damaged beta cells, and altering beta cells to increase susceptibility to damage. Conditions and syndromes associated with IDDM include pancreatic disease from trauma, infections, toxic injury, neoplastic conditions, hormonal disorders (such as hypothalamic lesions), and neurologically active agents (such as epinephrine), as well as post-infancy-acquired pancreatic disease (National Diabetes Data Group, 1985; Harris and Zimmet, 1992). Several epidemiologic studies have reported an excess of pancreatic cancer for individuals following exposure to benzene (Li et al., 1969; Lloyd et al., 1977; Thomas et al., 1982; Zoloth et al., 1986). In animals, Yang et al. (1979) reported that benzene-treated animals altered pancreatic excretion by increasing the flow by more than 900% and by reducing the protein concentration of the pancreatic fluid by more than 50%. The pancreas was not assessed for histopathology; therefore, it was undetermined whether bile duct-pancreatic fluid flow was occurring because of pancreatic damage. Liver damage was ruled out because benzene is not hepatotoxic.

    It should be noted that IDDM is associated with a variety of hematologic changes (such as anemia) and malignancies (such as lymphocytic leukemia, lymphoma, and multiple myeloma) that might be directly related to or simply coincidental with the diabetes (Bern, 1982). From the literature reported in the sections on anemia and cancer, it can be seen that all of these conditions are also associated with exposure to benzene.

    According to Bennett (1990), corticosteroids are associated with the development of diabetes by reducing insulin sensitivity, or possibly by impairing islet function frequently associated with the development of impaired glucose tolerance. The secretion of anti-insulin hormones, such as growth hormones or adrenocorticotrophic hormone (ACTH), are also believed to play an important role in IDDM development (Rodriguez, 1986). Steroid hormones play an important role in determining the severity of beta cell damage in the infected mouse, with androgens and glucocorticoids being particularly critical (Craighead, 1981). For example, diabetes fails to develop in infected castrated males unless androgens are administered. In mice treated with corticosteroids, coagulative necrosis of the islets of Langerhans develops accompanied by severe hyperglycemia (Craighead, 1981). Benzene has been shown to stimulate the hypothalamic-pituitary-adrenocortical (HPA) axis of mice (Hsieh et al., 1991), accompanied by increased ACTH/corticosterone release into blood. The HPA axis is responsible for releasing biogenic amines from the hypothalamus, ACTH from the pituitary, and corticosteroids from the adrenal cortex (Axelrod and Reisine, 1984; Dunn and Kramarcy, 1984).

    It has been demonstrated that most IDDM patients have autoantibodies to the pancreas (Lernmark et al., 1981), as well as to other organs (Whittingham et al., 1971; Nerup and Binder, 1973; MacCuish and Irvine, 1975; Gamble, 1980; Faustman et al., 1991; Nathan, 1993). Autoantibodies have also been reported in animals chronically exposed to small doses of benzene (Alekseeva and Zorina, 1969). Additionally, benzene is known to affect the immune system in other ways (Davies and Levine, 1986; Kalf, 1987; Hsieh et al., 1990a,b; Zeman et al., 1990; Cox, 1991; McMurry et al., 1991, 1994a,b; Reid et al., 1994) which might have an impact on the development of diabetes. In conclusion, although the literature suggests an association between benzene exposure and the development of IDDM, such a determination is premature.

    Immune System Effects

    Although not assessed directly through the Benzene Subregistry, immune system effects do play a role in several of the outcomes that were assessed, such as infections for kidney disease and urinary tract problems and autoimmune components of diabetes and anemia. It is therefore important to review reported effects on the immune system following benzene exposure.

    Benzene is a known immunotoxicant (Dean et al., 1979; Snyder, 1987; McMurry et al., 1991). As discussed in the section on anemia, benzene does cause changes in circulating leukocytes, including lymphocytes (Aksoy et al., 1971, 1987; Yin et al., 1987b, 1982; Kipen et al., 1989; ATSDR, 1993); in addition, benzene-induced aplastic anemia has peripheral lymphocytopenia as an early distinctive feature of the disease ( Irons and Moore, 1980; Goldstein, 1983; Infante and White, 1983; Irons, 1985). It should be noted that there are indications that benzene has a greater depressive effect on T lymphocytes than on B cells (Popp et al., 1992). Zeman et al. (1990) reported decreased CD4:CD8 lymphocyte ratios, abnormal values for lymphocyte populations, and impaired lymphocyte function in petroleum workers exposed to up to 20 milligrams per cubic meter (mg/m3) benzene for 5 years. Changes in humoral immunity have been reported for workers exposed to, but not seriously intoxicated by, benzene (Lange et al., 1973; Smolik et al., 1973); IgA and IgG levels were reduced, but IgM levels were slightly higher. There were, however, concomitant exposures to other solvents. Finally, a reduced mitogenic response of lymphocytes was reported for a population occupationally exposed to low levels of benzene (Yardley-Jones et al., 1988a).

    Benzene also affects both humoral and cellular acquired immunity in animals (Wolf et al., 1956; Cronkite et al., 1982, 1985; Rozen et al., 1984; Rozen and Snyder 1985; Ward et al., 1985; Aoyama, 1986; NTP, 1986; Hsieh et al., 1988b, 1990a; Huff et al., 1989; ATSDR, 1993). As early as 1915 (Simonds and Jones, 1915), studies of rabbits exposed to benzene revealed reduced production of red cell lysins, of agglutinins for killed typhoid bacilli, and of opsonins and the absence of antibacterial antibodies (Camp and Baumgartner, 1915; Hektoen, 1916). Benzene is highly toxic to bone marrow stem cells (Snyder and Kocsis, 1975), as well as mitotic cells and lymphocytes, especially B cells and suppressor T cells (Aoyama, 1986; Snyder, 1987).

    Several studies implicate the immune system as a target for benzene. Inhalation or injection of benzene has been shown to suppress the mitogenic response of B and T lymphocytes (Wierda et al., 1981; Rozen et al., 1984; Rozen and Snyder, 1985; Hsieh et al., 1990a). Other immune system effects seen in animals include development of lymphopenia (Hsieh et al., 1990a), involution of thymic mass and suppressions of both B and T cell mitogeneses (Wierda et al., 1981; Rozen et al., 1984; Rozen and Snyder, 1985; Hsieh et al., 1990a), suppression of mixed lymphocyte culture response to alloantigens, suppression of the tumor lytic ability of cytotoxic T lymphocytes as determined by 51Cr-release assay, and production of antibodies in response to T-dependent antigen (sheep red blood cells) (Wierda et al., 1981; Wierda and Irons, 1982; Rozen and Snyder, 1985; Aoyama, 1986; Hsieh et al., 1990a). IgM levels were also reported to be reduced in benzene-exposed mice (Wierda et al., 1981).

    Animal studies have further defined the effects of benzene exposure on the immune system. Female C57BL/6 mice dosed with benzene at 220, 440, or 880 mg/kg/day for 14 days demonstrated significant decreases in splenocyte proliferation (Reid et al., 1994). A decrease in spleen weight and in the number of circulating leukocytes have been reported in numerous other animal studies (Wolf et al., 1956; Yin et al., 1982; Cronkite et al., 1985, 1989; Ward et al., 1985; NTP, 1986; Huff et al., 1989; Cox, 1991; Wells and Nerland, 1991; Li et al., 1992; Fan, 1992). Inhibition of interleukin-2 (IL-2) production has been demonstrated following exposure to benzene (Post et al., 1985; Hsieh et al., 1990a, 1991; Fan, 1992; Reid et al., 1994). One interesting note is the report of the production of autoantibodies in animals chronically exposed to small doses of benzene (Alekseeva and Zorina, 1969; Tikhacheck and Frash, 1973). Ethanol can enhance the immunosuppressive effects of benzene (Kalf, 1987; Bloch et al., 1990).

    Benzene also affects functional immunity (Snyder et al., 1978; Rosenthal and Snyder, 1987; Cox, 1991; ). Indeed, as early as 1913, studies indicated that benzene-exposed rabbits displayed an increased susceptibility to pneumonia (Winternitz and Hirschfelder, 1913; Hirschfelder and Winternitz, 1913) and tuberculosis (White and Gammon, 1914). Later studies have also indicated decreased responses to pathogenic microorganisms, such as Listeria monocytogenes (Rosenthal and Snyder, 1985) and Klebsiella pneumoniae (Aranyi et al., 1986). Concentrations of 200 or 400 ppm for 4 to 5 weeks (5 days per week) suppressed the primary antibody response to tetanus toxin in mice, but there was no effect at 50 ppm (ATSDR, 1993).

    The metabolites of benzene have also been shown to affect the immune system of animals. These effects include inhibition of IL-2 production (Post et al., 1985); inhibition of maturation and proliferation of B lymphocyte progenitors (Wierda and Irons, 1982; Post et al., 1985); depression of numbers of B and T lymphocytes in the marrow, spleen, and thymus (Irons et al., 1980; Pfeifer and Irons, 1981, 1982, 1983; Post et al., 1985); inhibited macrophage activity (Lewis et al., 1988a); interference with microtubule assembly (Irons et al., 1981; Pfeifer and Irons, 1983); decreased bone marrow cellularity (Wierda and Irons, 1982); inhibition of pre-B cells (IgM-) from maturing to IgM+ cells, as well as reduction of the ability of mitogens to stimulate the proliferation of IgM+ cells to CFU-B colonies (King et al., 1987). It should be noted that benzene and benzene metabolites are not always suppressive (McMurry et al., 1994a, 1991) and at low doses (Pfeifer and Irons, 1981) or on cell-mediated immunity (Aoyama, 1986) can be stimulatory.

    Inhibition of proliferation and production of the T cell lymphokine IL-2 has been seen in mice exposed to p-benzoquinone (Post et al., 1985) as well as to benzene (Hsieh et al., 1990b). As discussed in the section on diabetes, benzene can also have an adverse effect on immune function by means of an activated HPA system (Hsieh et al., 1991), which is responsible for the production of corticosteroids from the adrenal cortex (Axelrod and Reisine, 1984; Dunn and Kramarcy, 1984). Corticosteroids have been reported to inhibit both interleukin-1 (IL-1) and IL-2 production (Snyder and Unanue, 1982; Blecha and Baker, 1986; Goodwin et al., 1986). Benzene can also have a pronounced effect on other brain neurotransmitters (Rea et al., 1984; Paradowski et al., 1985;Hsieh et al., 1988a). In addition, it has recently been demonstrated that various benzene metabolites depress the production of interferon (Cheung et al., 1988; Popp et al., 1992). Acute benzene exposure induced neither immunosuppression nor stimulation in cotton rats (McMurry et al., 1991), results that were believed to be caused by species variation.

    According to Kalf (1987), the immunosuppressive effects of benzene can be altered by other substances. For example, prior administration of a product from Aspergillus ochraceous, which has interferon-inducing properties, has been reported to modulate the effects of benzene exposure (Pandya et al., 1986). Ingestion of ethanol also increases immunosupression by benzene (Nakajima et al., 1985), as well as benzene-induced hematotoxicity (Baarson et al., 1982) in experimental animals.

    Kidney Disease

    Kidney disease was reported in excess for males and females in the 55 through 64 years of age group. For females, 3.0 cases were observed and 0.3 expected; for males, 2.0 were observed and 0.5 were expected. No studies were located that specifically addressed the potential for development of noncarcinogenic kidney disease following exposure to benzene; however, reports of miscellaneous cases of kidney outcomes-Goodpasture syndrome (glomerulonephritis) and antiglomerular basement membrane antibody (Klavis and Drommer, 1970), nephrotoxic anuria (Csata et al., 1971), and kidney congestion (Winek and Collom, 1971)-were found. Both rats (Exxon, 1986) and mice (Shell, 1992) orally exposed to benzene at varying levels did not develop any adverse kidney conditions.

    The excess reports of kidney disease could be due to kidney infections. As indicated in the previous section on immune system effects, in animals metabolites of benzene-including hydroquinone, catechol, and p-benzoquinone-have been reported to damage or suppress the activities and levels of various white blood cells in the immune system, leaving the exposed animal vulnerable to both bacterial pathogens and to transplanted tumor cells (Lewis et al., 1988a; Cox, 1991).

    There were only 5 reported cancers of the urinary organs in the time frames “ever had” or “within the last 12 months” for the Benzene Subregistry. Wong (1987a) reported the standardized mortality ratio (SMR) (140.0) for kidney cancer for continuously exposed workers to be three times that for the occupationally unexposed (48.4). In addition, a study of workers having potential exposure to downstream gasoline at oil distribution centers in the United Kingdom found a significant excess of kidney and suprarenal cancer (Rushton and Alderson, 1983). Negative reports of risk of developing renal cancer following benzene exposure were also found. Enterline and Viren (1985) reviewed the epidemiologic evidence for an association between petrol (gasoline) exposure and kidney cancer and concluded that there was little support for an etiologic link in the 12 cohort, 3 case referent, and 3 ecologic studies included in their review. Similar conclusions had been reached in a workshop on the subject a year earlier (Raabe, 1983) and in a 1987 review of several cohort studies examining the relationship between organic solvents and renal cancer (Harrington, 1987). Three additional studies (Van de Laan, 1980; Divine and Barron, 1987, Harrington et al., 1989) also found no relationship between exposure to solvents and renal cancer.

    Although the literature is too sparse to draw conclusions about any associations between benzene exposure and the development of kidney disease, the literature definitely supports an association between benzene exposure and adverse effects on functional immunity. Further investigation of the causes of the reported cases of kidney disease in the Benzene Subregistry would assist researchers in determining whether immune system dysfunction played a role in the development of kidney disease in this population.

    Liver Disease

    There was no statistically significant difference between the rates of liver disease for the Benzene Subregistry population and the NHIS population. No overall pattern was seen; however, the rates for both registrant and NHIS male age groups 35 years of age or older were of interest; that is, 5 cases were observed and <1 was expected. This question was asked in the “last 12 months” time frame.

    No studies were located that addressed the effect of benzene on the human liver. Several animal studies, however, were identified. Constan et al. (1996) evaluated apoptosis in the livers of Fischer-344 rats following exposure to drinking water containing seven contaminants, one of which was benzene. The authors deduced that apoptosis was directly correlated with changes in cell proliferation that were produced by repeated exposures to the contaminants. Ugurnal et al. (1995) investigated the effect of acute and chronic benzene treatment on the lipid peroxidation and antioxidant system in mouse liver. Malondialdehyde and diene conjugate levels were found to be increased in liver homogenates and microsomes of chronically exposed mice and were unchanged in the acutely exposed; glutathione levels remained unchanged in liver homogenates of all treatment groups. Some parameters, including cytosolic total glutathione peroxidase and glutathione transferase, were increased, but others (selenium-dependent glutathione peroxidase and superoxide dismutase) remained unchanged.

    Pawar and Mungikar (1975) investigated the effect of administering 1,400 mg/kg/day benzene to rats. They found an increase in liver weight in the exposed animals, as well as a decrease in protein in the postmitochondrial supernatant fractions. They also identified changes in hepatic drug metabolism and lipid peroxidation in the exposed animals.

    Negative studies were also located. Tatrai et al. (1980) found slight increases in relative liver weight of CFY rats exposed to 125 ppm benzene for 24 hours per day during gestational days 7 through 13; this outcome was not considered to be adverse, however. No adverse hepatic effects were observed in B6C3F1 mice exposed to 0, 12, 195, or 350 mg/kg benzene in drinking water for 30 days (Shell, 1992).

    Neurotoxicity

    The potential for benzene to have neurotoxic effects was assessed by the Benzene Subregistry through the assessment of three health outcomes-hearing impairments, mental retardation, and speech impairments. These three health conditions were queried in the NHIS questionnaire in the time frame “do you now have?” The time frame for the comparable Benzene Subregistry health conditions was adjusted by counting only registrants who reported that they were “currently receiving treatment” for one of these three conditions. Again, the decreased reporting by the Benzene Subregistry population was expected given that a health care provider was required to confirm the Benzene Subregistry registrants' report of the condition; that is, hearing impairment is a condition that might be reported without health care provider input. If all other factors were equal or the same, increased reporting by the NHIS respondents compared with the Benzene registrants would have been expected. A statistically significant deficit in reporting of hearing impairments was seen in the Benzene Subregistry population compared with the NHIS population. The risk ratio was less than 0.50 for most age and sex groups.

    No summary estimates are available for mental retardation. Two registrants reported being currently treated at baseline for mental retardation (the time frame for reporting was “now being treated”); 7.1 were expected (O/E = 0.28, 99% CI = 0.02, 1.30) based on NHIS rates.

    A total of 6 registrants reported currently being seen by a health care provider for speech impairment. This number represented an overall statistically significant decrease in reporting of this condition by the Benzene Subregistry population compared with the NHIS population. This finding was also consistent with what would have been expected noting the difference in the NHIS and Benzene Subregistry wording of this health condition (the subregistry question was stated as speech impairment, the NHIS question as stammering and stuttering, or other speech impairment) and the Subregistry requirement for confirmation by a health care provider.

    Benzene has been shown to affect both the peripheral (Juntunen, 1982) and central (Leong, 1977; Brief et al., 1980; Goldstein, 1989a) nervous systems. The signs and symptoms reported include vertigo, drowsiness, euphoria, headache, giddiness, narcosis, muscular incoordination, convulsions, paralysis, and unconsciousness (Gerarde, 1960; Drozd and Bockowski, 1967; Brief et al., 1980; Fielder et al., 1982; Davies and Levine, 1986). Chronic industrial exposure has also been reported to cause neurological abnormalities, such as global atrophy of the lower extremities and distal neuropathy of the upper extremities (Baslo and Aksoy, 1982), as well as abnormal electroencephalograms (Schneider and Ursoniu, 1969; Kellerova, 1985).

    Most exposures to benzene occur in the workplace over a relatively short period of time. For example, Midzenski et al. (1992) reported on 15 degassers who were acutely exposed to >60 ppm benzene over several days. Medical surveillance evaluation initially revealed 11 workers (73%) with neurotoxic symptoms. In addition, workers with more than 16 hours of acute exposure were significantly more likely to report dizziness and nausea than those with less than 16 hours of acute exposure. According to Fielder and coworkers (1982), fatalities have also been reported due to the inhalation of high concentrations of benzene in occupational settings, primarily in enclosed spaces such as tanks containing high residual levels of benzene. The findings at autopsy in these cases included extensive petechial hemorrhages in the brain. The authors estimated that inhalation exposure to 20,000 ppm benzene is likely to be rapidly fatal, 7,500 ppm is dangerous after 30 to 60 minutes, and 3,000 ppm is tolerable for only about 30 minutes.

    Only one study was located that could possibly link the health outcomes of speech impairment, hearing impairment, or mental retardation with benzene exposure. Akefeldt et al. (1995) investigated whether parental age and parental preconceptional exposure to various agents, including gasoline or petrol (of which benzene is a major constituent) differentiated children with Prader-Willi syndrome (PWS) from obese children without PWS. (Note: PWS is a congenital disorder characterized by rounded face, almond-shaped eyes, strabismus, low forehead, hypogonadism, hypotonia, insatiable appetite, and mental retardation.) They found that paternal exposure to gasoline or petrol was significantly higher in the PWS group.

    In animals, only one animal study was located that discussed the potential for hearing damage following benzene exposure. Tilson and coworkers (1980) reported that preweaning exposure to benzene had no significant effect on the acoustic startle response in rats. Other studies, however, have indicated that benzene is indeed a neurotoxicant. Acute exposures to high concentrations (>10,000 ppm) of benzene have been shown to be fatal to animals because of the direct narcotic effects (Gerarde, 1960; Jonek et al., 1965; Drew and Fouts, 1974); the threshold for narcotic effects has been estimated to be about 4,000 ppm (Leong, 1977; Fielder et al., 1982).

    Other neurological effects have also been reported. Dempster et al. (1984) reported a 90% decrease in the hind limb grip strength of C57BL/6 mice following a single exposure to 1,000 or 3,000 ppm benzene. Animals in the 3,000-ppm group also had tremors that disappeared after exposure ceased. Evans et al. (1981) reported increased behavioral activity in male CD-1 and C57BL/6 mice 2 weeks after exposure to 300 or 900 ppm benzene for 6 hours per day for 5 days. Narcosis was reported in mice in the 900-ppm group. Piloerection, excitation, and tremors were reported for C57BL/6 mice exposed to 440 to 660 mg/kg benzene (Wierda et al., 1981).

    Hsieh and coworkers (1990b, 1991) have investigated the effect of benzene exposure on the brain of CD-1 mice, specifically on the catecholamines norepinephrine and dopamine; the catecholamine metabolites vanillylmandelic acid, 3,4-dihydroxyphenylacetic acid, and homovanillic acid; and the indoleamine serotonin and its metabolite 5-hydroxy-indoleacetic acid. They found that benzene produced increases of norepinephrine in the hypothalamus, cortex, midbrain, and medulla oblongata; dopamine in the hypothalamus and corpus striatum; and serotonin in all dissected brain regions except the cerebellum. They also reported elevated levels of various monoamine metabolites in these areas. These neurochemicals are known to play important roles in behavior (Poirier and Bedard, 1984; Rogawski and Baker, 1985). Finally, these authors found that benzene stimulated HPA activity, which is critically involved in the regulation of behavioral adaptations (Axelrod and Reisine, 1984; Dunn and Kramarcy, 1984). Li et al. (1992) reported increased levels of acetylcholinesterase in the brains of mice exposed to 3 ppm benzene for 2 hours each day for 30 days. Finally, in another study, de Gandarias et al. (1992) investigated the changes in Lys- and Leu-aminopeptidase activities in several rat brain regions after benzene administration. They found that the activity of both enzymes were significantly decreased in the thalamus, hypothalamus, hippocampus, and amygdala.

    Only one study was located that reported negative results. McMurry et al. (1991) found no gross behavioral changes in cotton rats exposed to benzene.

    Respiratory Conditions

    Asthma, emphysema, and chronic bronchitis are conditions for which decreased reporting by registrants would have been expected based on the NHIS omission of physician confirmation; the results confirmed this. There was a significant decrease in reporting for the Benzene Subregistry respondents compared with the NHIS reporting population (56.0 observed, 94.3 expected, O/E = 0.59, 99% CI = 0.41, 0.83); this finding was consistent across age groups. For the Subregistry, there was a significantly increased reporting for registrants 0 through 9 years of age for other respiratory allergies or problems such as hay fever (37.0 observed, 8.3 expected, O/E = 4.47, 99% CI = 2.80, 6.74).

    According to the National Institute for Occupational Safety and Health (NIOSH) (1988), acute exposure to relatively high levels of benzene in air has been reported to cause respiratory irritation, pulmonary edema, and pneumonia. Chronic exposure to benzene in air can cause labored breathing. In a case report of three fatalities due to acute benzene exposure aboard a chemical cargo ship, Avis and Hutton (1993) reported evidence existed of respiratory injury, including hemorrhagic airless lungs with confluent alveolar hemorrhage, and pulmonary edema. Other studies have reported petechial hemorrhages in the pleura (Fielder et al., 1982), respiratory tract inflammation and lung hemorrhages (Winek and Collom, 1971), irritation of the respiratory tract Drozd and Bockowski, 1967), and intra-alveolar pulmonary (Klavis and Drommer, 1970). Workers exposed to >60 ppm benzene for up to 3 weeks reported mucous membrane irritation and dyspnea (Midzenski et al., 1992).

    In a mortality study of refinery workers in California, Wong (1987a,b) reported substantially lower death rates from nonmalignant diseases of the respiratory system. This finding was confirmed in a followup of the cohort by Dagg and coworkers (1992). In a similar study, Zoloth et al. (1986) reported a significantly reduced death rate from emphysema in commercial pressmen.

    In animals, acute inhalation of benzene by rats, mice, and rabbits caused respiratory paralysis. Also in animals, the lungs, spleen, and lymphatic system showed signs of damage that became more pronounced after exposure to both benzene and ethanol (Bloch et al., 1990). Deichman et al. (1963) reported that the most consistent and significant pathological change in Sprague-Dawley rats exposed to benzene was chronic bronchopneumonia. Driscoll and Snyder (1984) reported that benzene exposure did not affect the respiratory rate or minute volume of mice; however, Sabourin et al. (1990) found a significant decrease in the respiratory rate of mice (but not rats) treated with benzene. Finally, in a study of the ability of the lung to bioactivate or detoxify benzene and the pneumotoxicity of benzene, Chaney and Carlson (1995) found that, although overall metabolism was lower, pulmonary microsomes converted benzene to hydroquinone; however, benzene, injected at a dose of 600 mg/kg body weight in rats, did not cause significant lung cell damage as determined by measurement of gamma-glutamyltransferase and lactate dehydrogenase in bronchoalveolar lavage fluid. Given the types of studies reported in the literature and the information available for the registrants, no obvious conclusions could be reached regarding the association between benzene and this health condition.

    Skin Rashes, Eczema, or Other Skin Allergies

    There was significantly increased reporting of skin rashes for the Benzene Subregistry for the group 0 through 9 years of age-the relative risk was 2.96 (99% CI = 1.77, 4.63) based on 31.0 cases observed compared with 10.5 cases expected. Increased reporting was also found for the group 65 years of age or older, with 13.0 observed and 3.0 expected (O/E = 4.35, 99% CI = 1.90, 8.54).

    In humans, most dermal exposures to benzene have occurred in the occupational setting. Occupational exposures to benzene at levels >60 ppm in air for up to 3 weeks resulted in skin irritation (Midzenski et al., 1992). Other effects seen in other case reports include dry, scaly dermatitis (Fielder et al., 1982); marked irritation due to the defatting action of the solvent (Gerarde, 1963; Fisher, 1975); swelling and edema (Greenburg, 1926); and erythema and vesiculation (Sandmeyer, 1981). In fatal cases, extensive petechial hemorrhages (Fielder et al., 1982) and second-degree chemical burns to the face, trunk, and limbs (Avis and Hutton, 1993) were reported.

    Few animal studies were located in which the dermal effects of benzene were studied. When applied to the skin of rabbits, benzene caused moderate erythema and edema, and moderate necrosis (Wolf et al., 1956). In guinea pigs, benzene caused significant skin changes, and possible irreversible cellular injury, within 15 minutes of topical application (Kronevi et al., 1979). Decreased skin collagen content (Hudák and Ungváry, 1978), edema (Kiernan, 1977), and alopecia (Exxon, 1986) have also been reported for animals exposed to benzene by various routes.

    As can be seen, the literature is too sparse to attempt to draw conclusions about the potential for exposure to benzene in the environment to result in disorders of the skin.

    Ulcers, Gallbladder Trouble, or Stomach or Intestinal Problems

    An overall summary model of the results was obtained for outcomes related to ulcers, gallbladder trouble, and stomach or intestinal problems. No statistically significant differences were found (55.0 were observed, 58.6 were expected, O/E = 0.94, 99% CI = 0.65, 1.32).

    According to the literature, short-term (acute) exposure to benzene in the occupational setting has caused nausea and gastrointestinal irritation (NIOSH, 1988); however, the doses were relatively large. Chronic exposure to benzene in the workplace has been reported to produce anorexia (NIOSH, 1988). One case report was located in which a man swallowed an unspecified amount of benzene and survived, but developed an intense toxic gastritis and later pyloric stenosis (ATSDR, 1993). According to Drozd and Bockowski (1967), early symptoms of benzene toxicity include nausea and vomiting. For these conditions, however, there appears to be no association between reported rates in the literature and the reporting rates of the registrants.

    Urinary Tract Disorders, Including Prostate Trouble

    In interpreting the results for urinary tract disorders, including prostate trouble, the wording of the questions for the NHIS and Subregistry questionnaires differs and might have been a factor in the results obtained (urinary tract disorders, including prostate trouble, were included in the Subregistry questionnaire; disorders of the bladder, other than bladder infections, and diseases of the prostate were included in the NHIS questionnaire). There were differences in reporting, however; rates for females were consistently greater than the rates for male, a finding that supports common knowledge about the occurrence of urinary tract disorders and increased female susceptibility to benzene toxicity. A statistically significant increase was found for all females (26.0 observed, 4.3 expected, O/E = 6.06, 99% CI = 3.44, 9.84); for females 24 years of age or younger, 1.2 were expected and 9.0 were observed.

    Wong (1987b) investigated mortality in refinery workers in California and found significantly lower rates for diseases of the genitourinary system for benzene-exposed workers. A followup of this cohort (Dagg et al., 1992) confirmed these findings; however, women were excluded from most of the analyses because of low numbers. In addition, the authors noted a strong “healthy worker” effect in this population. It should be noted that no studies were located specifically addressing the potential effects of benzene exposure on the development of prostate disorders.

    In 1989, Steineck et al. (1989) observed a small increase in the risk of urinary bladder cancer after exposure to benzene. These results were reported to be similar to those of Wong (1987b), who observed two cases of urinary bladder cancer among workers exposed from 5 to 14 years to benzene and two cases among workers exposed 15 years or more. It should be noted that only five cases of cancer of the urinary organs were reported in the Benzene Subregistry population.

    Urinary tract disorders are frequently related to infection. As discussed in the previous section on immune system effects, benzene is known to adversely affect both humoral and cellular acquired immunity (Dean et al., 1979; Davies and Levine, 1986; Snyder, 1987; McMurry et al., 1991; ATSDR, 1993); indeed, there is considerable evidence for impairment of the human immune system in people chronically exposed to benzene (Lange et al., 1973; Smolik et al., 1973; USEPA, 1984).

    The literature definitely supports the impact of benzene on both humoral and cellular acquired immunity. Whether the increased reporting of urinary tract disorders is related to that effect is unknown.

    Cancer

    There were no statistically significant increases for any age or sex group for cancer in the Benzene Subregistry. The overall risk ratio was 1.74 (99% CI = 0.75, 3.41) based on 13.0 observed and 7.5 expected. It should be noted that one case of leukemia was reported for the “in the last 12 months” time frame. When the time frame was designated as “ever” and the Subregistry rates were compared with the SEER rates, the calculated risk ratios were inconsistent over the years. There appeared to be a correlation of rates between sexes; also, the highest relative risk values occurred for the same years.

    Benzene exposure, and its association with cancer, has been extensively covered in the literature. There is sufficient evidence to have declared benzene a carcinogen (IARC 1987, 1982a,b; Ning et al., 1991). Many authors (Hecht and Hecht, 1987; Kalf, 1987; Smith et al., 1989; Cox, 1991; Bois et al., 1991; Eastmond, 1993) and reviews (Davies and Levine, 1986; Austin et al., 1988) have demonstrated that the metabolites of benzene are primarily responsible for the carcinogenic action of benzene. According to Cox (1991), benzene metabolites are responsible for the progression of a malignant clone of cells from a few (possibly dormant) transformed cells to a clinically detectable neoplasm by means of (1) deregulation of cell differentiation from toxic damage to the bone marrow stromal microenvironment (Thomas et al., 1989); (2) suppression of immune surveillance by toxic metabolites, making it easier for malignant cells to survive and proliferate (Lewis et al., 1988a); or (3) inhibition of proliferation of normal cells by differential cytotoxicity compared with malignantly transformed cells (Cox, 1991).

    Benzene affects macrophages as well as lymphocytes. Experiments with cultured macrophages have shown that hydroquinone, p-benzoquinone, and catechol are all highly effective inhibitors of hydrogen peroxide release from stimulated macrophages (>90% reduction for concentrations of 15 micromoles [μM] for all three metabolites). The release of hydrogen peroxide normally allows macrophages to kill invading bacteria and other threats. P-benzoquinone also inhibits phagocytosis and cytolysis of tumor cells by macrophages at concentrations of ≥10 μM (Lewis et al., 1988a). 1,4-Benzoquinone was also found to be the most potent metabolite in reactive oxygen species formation, which is believed to be a mechanism in the induction of leukemia (Shen et al., 1996).

    Chronic exposure to benzene has been demonstrated to cause leukemia in humans (Mallory et al., 1939; Aksoy et al., 1971; Goldstein, 1977; Aksoy, 1980; Goldstein and Snyder, 1982; IARC, 1982a, b; Aksoy, 1985a,b; NIOSH, 1988; Ning et al., 1991; ATSDR, 1993; Eastmond, 1993; Travis et al., 1994; Swaen and Slangen, 1995). The types of leukemias and lymphoproliferative diseases known to be caused by benzene exposure are acute myelogenous leukemia, acute lymphocytic leukemia, acute erythrocytic leukemia, acute myelomonocytic leukemia, acute promyelocytic leukemia, acute undifferentiated leukemia, hairy cell leukemia, chronic myelogenous leukemia, chronic lymphocytic leukemia, Hodgkin's lymphoma, non-Hodgkin's lymphoma, and multiple myeloma (Infante et al., 1977; Vianna and Polan, 1979; Rinsky et al., 1981; Decouflé et al., 1983; Infante and White, 1985; Mehlman, 1983, 1989, 1990, 1991; Rinsky et al., 1987; Aksoy, 1988, 1989a,b; Goldstein, 1989b, 1990; Dempster and Snyder, 1990; Linos et al., 1991; Wolff, 1992). The most common type of leukemia among individuals chronically exposed to benzene is acute myeloblastic leukemia (Aksoy, 1988), a disease characterized by a proliferation of cells morphologically indistinguishable from myeloblasts.

    Numerous epidemiologic studies and case reports (Aksoy et al., 1971, 1974, 1978, 1987; Infante et al., 1977; Ott et al., 1978; Aksoy 1980, 1989a,b; Rinsky et al., 1981, 1987; Wong 1983, 1987a; Bond et al., 1986; Yin et al., 1987a, 1989; Rinsky, 1989; Cox, 1991; Voytek and Thorslund, 1991; Dagg et al., 1992; Feingold et al., 1992; Hayes, 1992; Paustenbach et al., 1992; Hunting et al., 1995) have demonstrated a causal relationship between benzene exposure and leukemia in occupationally exposed workers. These studies include roughly 28,500 Turkish shoe workers (average daily dose estimated from 210 to 650 ppm) among whom Askoy (1989a) identified a statistically significant number of leukemia cases. Yin et al. (1989) reported 30 leukemia cases observed among 28,460 benzene-exposed Chinese workers whose average daily dose was estimated at over 100 ppm. In preliminary analyses, the leukemia rate in Chinese workers was about twentyfold higher for those having benzene exposure (Hayes, 1992). In a follow-up report, deaths due to all hematopoietic and lymphoproliferative malignancies (p = 0.01) and lung cancer (p = 0.01) were increased among workers with greater cumulative exposures (Hayes et al., 1996). This rate was similar to that seen for Italian workers exposed to up to 600 ppm benzene in shoe and rotogravure factories (Vigliani, 1976). Li et al. (1994) reported similar findings for 74,828 benzene-exposed workers compared with 35,805 unexposed workers from 12 cities in China; in addition, no significant differences in the relative risks for total mortality and cancer mortality were found between female and male benzene-exposed workers. These results corroborated their previous findings (Yin et al., 1994). Leukemogenic effects were also reported in the Pilofilm (rubber) worker population (Infante et al., 1977; Rinsky et al., 1981, 1987; Paustenbach et al., 1992; Utterback and Rinsky, 1995; Crump, 1996; Paxton, 1996). It should be noted that there are reports of the occurrence in young children of acute monoblastic leukemia that appears to be associated with parental exposure to pesticides, solvents, and petroleum products (Odom et al., 1990; Robinson et al., 1989; Shu et al., 1988).

    In an ecological study in which both gasoline consumption data and data on leukemia mortality and incidence were collected for 19 European countries, Swaen and Slangen (1995) found a weak inverse association between temporal trends in gasoline consumption and temporal trends in leukemia mortality and a weak positive association between the age-adjusted myeloid leukemia incidence in 14 areas and the gasoline consumption per square kilometer. The authors cautioned that their findings could be explained by other factors, such as differences in leukemia case ascertainment. Van Raalte (1982) reported similar findings. Wolff (1992), however, reported a statistically significant association between automobile ownership and acute myeloid leukemia, as well as between all lymphoproliferative diseases and automobile ownership. His findings are in agreement with those obtained by Robinson (1982) who showed a strong relationship between Australian leukemia mortality and vehicle usage. A study of childhood leukemia incidence in Denver, Colorado, indicated that rates of leukemia were higher in areas of higher traffic density (Savitz and Feingold, 1989); this finding also potentially supported Wolff's findings.

    In a retrospective follow-up study of 594 chemical workers occupationally exposed to benzene, which was expanded and updated in 1986 (Bond et al., 1986), Ott et al. (1978) reported an incidence rate ratio of 4.4 (1.2-11) for myelocytic leukemia for benzene-exposed workers relative to the general population. A cohort mortality study of 4,172 chemical plant workers reported elevated rates of leukemia (SMR = 2.3; 95%CI = 0.7, 5.3) and multiple myeloma (SMR = 2.3; 95%CI = 0.7, 9.4) among production workers with benzene exposure, predominantly among those whose exposures were 20 or more years prior to the onset of disease (Ireland et al., 1997). Other worker studies (McMichael et al., 1975; Rushton and Alderson, 1981a,b; Wolf et al., 1981; Decouflé et al., 1983; Wen et al., 1983; Wong, 1983; Divine et al., 1985; McCraw et al., 1985; Wong and Raabe, 1989; Marsh et al., 1991; Mele et al., 1995; Fu et al., 1996) generally confirmed these findings.

    Unfortunately, there are few data from which dose-response relationships can be established because of the deficiencies in the clinical and epidemiologic studies-such as the lack of appropriate sampling techniques, incomplete exposure determinations, lack of followup, and weak experimental designs or methodology-as well as the intermittent exposures to benzene, which make it difficult to assume that the average concentrations of benzene measured in a workplace actually indicate the true exposure experienced by each worker (Goldstein, 1985). Indeed, there are reports of a negative relationship for benzene and leukemia. For example, Hurley et al. (1991) reported that male industrial workers employed in 1967 in self-contained coke works or coke departments had no excess mortality from leukemia. Another epidemiologic study of Texas refinery workers showed no leukemia deaths from benzene exposures that were <1 ppm (Tsai et al., 1983). Finally, in an analysis of published case series and epidemiological studies through mid-1995, Bezabeh et al. (1996) found that the literature indicated benzene exposure was not a likely causal factor for multiple myeloma.

    Exposure to benzene or mixtures containing benzene have also been implicated in causing other types of cancer in humans. These cancers include bladder, stomach, prostate, and lung neoplasms (McMichael et al., 1974; Fox and Collier, 1976; Monson and Nakano, 1976; Monson and Fine, 1978; Parker et al., 1982; IARC, 1982b; Bernadinelli et al., 1987), primarily observed in rubber workers. Stomach cancer was also reported for shoe workers (Fu et al., 1996). In 1989, Steineck et al. (1989) observed a small increase in the risk of urinary bladder cancer after exposure to benzene. These results were reported to be similar to those of Wong (1987b), who observed two cases of urinary bladder cancer among workers exposed from 5 to 14 years to benzene and two cases among workers exposed 15 years or more. In a historical cohort study of 2,008 Italian shoe workers, Fu et al. (1996) found some evidence for an excess of bladder cancer. Finally, in a population-based case referent study of urothelial cancer, Steineck et al. (1990) reported that exposure to benzene (any annual dose) gave a relative risk of 2.0 for development of urothelial cancer; however, the authors concluded that the determination of a cause-and-effect relationship was confounded by exposure to other chemicals in addition to benzene.

    In the study of Chinese workers exposed to benzene, Yin et al. (1989) reported significantly elevated SMRs for lung cancer and nonstatistically significantly elevated SMRs for primary hepatocarcinoma and stomach cancer. It should be noted, however, that the authors cautioned that the estimated average and cumulative lifetime benzene exposure levels were based on relatively few measurements. Among Turkish workers, benzene exposure was related to the development of other forms of cancer, such as lung cancer (Aksoy, 1985a,b). Dagg et al. (1992), however, found significantly lower rates of lung cancer among a cohort of 14,074 refinery workers.

    A few studies (Woods et al., 1987; Eriksson et al., 1990; Serraino et al., 1992) have reported nonsignificantly elevated risks of developing soft tissue sarcoma (STS) in benzene-exposed workers. It should be noted that both misclassification of STS and the small numbers of cases often are problematic for interpreting these studies.

    Feingold et al. (1992) examined the association between parental occupational exposure to benzene and childhood cancer. They found elevated odds ratios for maternal exposure and for paternal exposure to benzene in relation to total cancers in their offspring. Control for other potential childhood cancer risk factors did not alter the results substantially. The authors believed their findings of a suggested association with parental benzene and petroleum exposure corroborated similar observations by Shu et al. (1988).

    Reports of the risk of developing renal cancer following benzene exposure have been negative. Enterline and Viren (1985) reviewed the epidemiologic evidence for an association between petrol (gasoline) exposure and kidney cancer and concluded that there was little support for an etiologic link in the 12 cohort, 3 case referent, and 3 ecologic studies included in their review. Similar conclusions had been reached in a workshop on the subject a year earlier (Raabe, 1983; Higginson et al., 1984) and in a 1987 review of several cohort studies examining the relationship between organic solvents and renal cancer (Harrington, 1987). Three additional studies (Van de Laan, 1980; Divine and Barron, 1985; Harrington et al., 1989) also found no relationship between exposure to solvents and renal cancer. Only one study, Fu et al. (1996), indicated a potential for development of kidney cancer following exposure to benzene.

    Multiple studies (Maltoni and Scarnato, 1979; Snyder et al. 1980, 1988; Maltoni et al., 1982a,b,c, 1983a,b, 1985; Maltoni, 1983; Cronkite et al., 1985, 1984, 1989; NTP, 1986; Snyder, 1987; Ashby and Tennant, 1988; Huff et al., 1989; Ciranni et al., 1991; ATSDR, 1993) have demonstrated that benzene is a multisite carcinogen in animals. Cancer sites reported have included the nasal and oral cavity; lung; forestomach; liver; skin; zymbal, mammary, Harderian, and preputial glands; ovary; and uterus. Lymphoma, hemolymphoreticular neoplasia, and all types of leukemias have also been reported in animals. For leukemogenicity, as opposed to cytotoxicity, there is some evidence that lifetime exposure to benzene in animals might actually suppress the development of leukemia-presumably because of the cytotoxic effects that prevent potentially leukemic cells from surviving and expressing themselves-but a much shorter (16-week) exposure period dramatically increases the incidence of myelogenous and other neoplasms (Cronkite et al., 1989; Cox, 1991). Metabolites of benzene, including hydroquinone, catechol, and p-benzoquinone, have been reported to damage or suppress the activities and levels of various white blood cells in the immune system, leaving the exposed animal vulnerable to both bacterial pathogens and to transplanted tumor cells (Lewis et al., 1988a; Cox, 1991).

    Cancer is caused by genotoxic events in somatic cells (Ciranni et al., 1991); benzene is also a known genotoxin. Cytogenetic studies in benzene-exposed workers have shown that exposure to benzene is associated with frequencies of both structural and numerical chromosomal aberrations (Funes-Cravioto et al., 1977; IARC, 1982a,b; Aksoy, 1988; Sasiadek et al., 1989; Popp et al., 1992; Sasiadek, 1992; ATSDR, 1993; Eastmond, 1993; Carere et al., 1995; Kara_i_ et al., 1995; Rothman et al., 1995; Silva and Santosmello, 1996). The limitations of occupational studies, including a lack of accurate exposure data, possible coexposure to other chemicals, and selection of appropriate control groups must be considered when interpreting the data from such studies. Currently many investigators believe that two or more benzene metabolites might be responsible for the myelotoxic and genotoxic effects of benzene (Goldstein, 1989b; Snyder et al., 1989; Guy et al., 1990, 1991; Chen and Eastmond, 1995a,b; Eastmond, 1993; Anderson et al., 1995; Hedli et al., 1996; Ross, 1996).

    Benzene crosses the placenta and is found in cord blood in amounts equal to or greater than those in the maternal blood (Dowty et al., 1976). Funes-Cravioto et al. (1977) examined genetic outcomes in children of female workers exposed by inhalation to benzene and other organic solvents during pregnancy. They found increased frequency of chromatid and isochromatid breaks and sister chromatid exchange (SCE) in the lymphocytes of these children.

    Negative results have been reported concerning chromosomal aberrations following benzene exposure in humans (Pitarque et al., 1996). Yardley-Jones et al. (1988b) and Seiji et al. (1990) investigated SCEs in humans and found negative results. Sarto et al. (1984) and Watanabe et al. (1980) also did not find raised SCE frequencies following exposure to benzene; Clare et al. (1984) reported an insignificantly raised SCE frequency after acute exposure to high benzene concentrations during one shift. Hallberg et al. (1996) found a lack of significant differences in deoxyribonucleic acid (DNA) repair capacity between benzene-exposed and control workers; however, the authors believed this finding to be due to an extremely low exposure to benzene (<0.3 ppm), the small population size, or a lack of benzene genotoxicity at these concentrations.

    Data from animal studies provide convincing evidence that benzene and its metabolites are genotoxic (Meyne and Legator, 1980; Siou et al., 1981; Harper et al., 1984; Smith et al., 1989; Snyder et al., 1989; Suzuki et al., 1989; Guy et al., 1990, 1991; ATSDR, 1993; Eastmond, 1993; Chen and Eastmond, 1995a). Positive results have been reported from analyses of chromosomal aberrations, SCE, and micronuclei in the bone marrow, lymphocytes, and erythrocytes of rats, mice, and rabbits (Meyne and Legator, 1980; Siou et al., 1981; Tice et al. 1982, 1989, 1980; Toft et al., 1982; IARC, 1982a,b; Tice and Ivett, 1985; Luke et al., 1988a,b; Harper et al., 1989; Suzuki et al., 1989; Cox, 1991; Ning et al., 1991; Xing et al., 1992; ATSDR, 1993). Other evidence of genotoxicity induced by benzene includes inhibition of DNA synthesis (Lee et al., 1988) and delayed cell cycle in mouse bone marrow and DNA adducts with benzene metabolites in mouse and rat hemoglobin or rat liver cells (Tice et al., 1980, 1982; Sabourin et al., 1990; ATSDR, 1993; Pathak et al., 1995; Levay and Bodell, 1996; Levay et al., 1996).

    Several studies (Swenberg et al., 1976; Sina et al., 1983; Garberg et al., 1988; Lee et al., 1989) have reported no increased strand breakage in vitro related to benzene exposure; however, metabolic activation of benzene or treatment with benzene metabolites has led to increased DNA strand breakage (Tice et al., 1980; Pellack-Walker and Blumer, 1986; Garberg et al., 1988; Lewis et al., 1988b; Glatt et al., 1989; Kawanishi et al., 1989; Witz et al., 1990; Yager et al., 1990; Popp et al., 1992).

    Benzene was not mutagenic in a range of bacterial and yeast systems (Lyon, 1976; Tanooka, 1977; Shahin and Fournier, 1978; Kaden et al., 1979; Lebowitz et al., 1979; Ning et al., 1991), including the conventional Ames test (De Flora et al., 1984). It was, however, found to be mutagenic in a sensitive microsuspension assay using the Salmonella strain TA 100 (McCarroll et al., 1980) and to produce positive responses in repair-deficient strains of Escherichia coli and Bacillus subtilis (McCarroll et al., 1981a,b).

    Results of animal studies investigating benzene as a transplacental genotoxic agent have been varied. Xing et al. (1992) reported that benzene caused a significant increase in micronuclei and SCE in both maternal bone marrow and fetal liver cells of mice. This result was consistent with that reported by Sharma et al. (1985) and Ning et al. (1991). Contrary to the positive findings, Harper et al. (1989) reported that benzene caused almost no response in either pregnant female or fetal mice at a single oral or injected dose.

    Although excess cancer risks have been reported in part of the Benzene Subregistry population, the cancers reported, with the exception of cancers of lymphatic tissues and leukemia, were not those most commonly associated with exposure to benzene; the cancers reported were cancers of the skin, genital organs, digestive system, breast, and urinary organs, as well as the previously mentioned cancers of lymphatic tissues and leukemia. It should be noted, however, that the route of exposure (via household water) was different than that for most occupational exposures (typically, the air route). Further studies with the Benzene Subregistry population should attempt to clarify these issues.

    CONCLUSIONS

    The Benzene Subregistry population reported more adverse health outcomes compared with a national sample. The adverse health outcomes reported in excess of those reported by the national sample, for all or specific age groups, were diabetes, urinary tract disorders, skin rashes, kidney disease, and respiratory allergies. Hearing impairment, asthma and emphysema, arthritis, and speech impairment were significantly reduced in the Benzene Subregistry population. Many of the effects seen in the Benzene Subregistry population have also been reported in the literature; however, the limitations of both the Benzene Subregistry data and many of the studies cited should be taken into account when assessing the value of the findings. Epidemiologic studies are frequently limited by multiple or mixed exposures to many chemicals, or both; inadequate latency periods; small cohorts; and the “healthy worker effect.” Additional limitations of the Benzene Subregistry data include the comparability between the questions posed by the Subregistry and those posed by the NHIS, recall bias, and frequency of health care utilization. Also, because of the many comparisons made, some of the positive results might have been chance occurrences, or might have occurred because the true causal factors (confounders) were not identified.

    Cause-and-effect relationships cannot be determined from simple analyses of the Subregistry-based information; however, information obtained from this database can and will be used to determine appropriate future activities and research. Considerations for further research using this database are modification of the current data collection procedures and methods; exploration of recognized sources of bias and reduction or elimination of these biases; acquisition of additional definitive information on, and confirmation of, selected outcomes that appear to be in excess; and substance-specific research with specific hypotheses clearly identified.

    REFERENCES

    Agency for Toxic Substances and Disease Registry (ATSDR). (1988). VIEW database. U.S. Department of Health and Human Services, Public Health Service, Atlanta, GA.

    Agency for Toxic Substances and Disease Registry (ATSDR). (1993). Toxicological Profile for Benzene (Update). U.S. Department of Health and Human Services, Public Health Service, Atlanta, GA.

    Agency for Toxic Substances and Disease Registry (ATSDR). (1995). National Exposure Registry, Benzene Subregistry Baseline Technical Report. U.S. Department of Health and Human Services, Public Health Service, Atlanta, GA.

    Agency for Toxic Substances and Disease Registry (ATSDR). (1996). Toxicological Profile for Benzene (Update). U.S. Department of Health and Human Services, Public Health Service, Atlanta, GA.

    AKEFELDT, A., ANVRET, M., GRANDELL, U., NORDLINDER, R., and GILLBERG, C. (1995). “Parental exposure to hydrocarbons in Prader-Willi syndrome.” Dev. Med. Child Neurol. 37(12):1101-1109.

    AKSOY, M. (1980). “Different types of malignancies due to occupational exposure to benzene: A review of recent observations in Turkey.” Environ. Res. 23:181-190.

    AKSOY, M. (1985a). “Benzene as a leukemogenic and carcinogenic agent.” Am. J. Ind. Med. 8:9-20.

    AKSOY, M. (1985b). “Malignancies due to occupational exposure to benzene.” Am. J. Ind. Med. 7:395-402.

    AKSOY, M. (1988). “Benzene carcinogenicity.” In: Benzene Carcinogenicity (M. Aksoy, ed.). CRC Press, Boca Raton, FL. pp. 113-151.

    AKSOY, M. (1989a). “Hematotoxicity and carcinogenicity of benzene.” Environ. Health Perspect. 82:193-198.

    AKSOY, M. (1989b). “Leukemogenic and carcinogenic effects of benzene.” In: Benzene: Occupational and Environmental Hazards-Scientific Update (M. Mehlman, ed.). Princeton Scientific Publishing Co., Princeton, NJ.

    AKSOY, M., DINCOL, K., AKGUN, T., ERDEM, S., and DINCOL, G. (1971). “Hematological effects of chronic benzene poisoning in 217 workers.” Br. J. Ind. Med. 28:296-302.

    AKSOY, M. and ERDEM, S. (1978). “Follow-up study on the mortality and the development of leukemia in 44 pancytopenic patients with chronic benzene exposure.” Blood 52(2):285-292.

    AKSOY, M., ERDEM, S., and DINCOL, G. (1974). “Leukemia in shoe-workers exposed chronically to benzene.” Blood 44:837-841.

    AKSOY, M., ERDEM, S., and DINCOL G. (1975). “Two rare complications of chronic benzene poisoning: myeloid metaplasia and paroxysmal nocturnal hemoglobinuria. Report of two cases.” Blut 30:255-260.

    AKSOY, M., OZERIS, S., SABUNCU, H., INANICI, Y., and YANARDAG, R. (1987). “Exposure to benzene in Turkey between 1983 and 1985: A hematological study on 231 workers.” Br. J. Ind. Med. 44:785-787.

    ALEKSEEVA, O.G. and ZORINA, L.A. (1969). “Autoantibodies in experimental chronic exposure to small doses of benzene.” Gigiena Truda i Professionalnye Zabolevaniia 13(6):30-33.

    ANDERSON, D., YU, T.W., and SCHMEZER, P. (1995). “An investigation of the DNA-damaging ability of benzene and its metabolites in human lymphocytes, using the comet assay.” Environ. Mol. Mutagen. 26(4):305-314.

    ANTOINE, S.R., DELEON, I.R., and O'DELL-SMITH, R.M. (1986). “Environmentally significant volatile organic pollutants in human blood.” Bull. Environ. Contam. Toxicol. 36:364-371.

    AOYAMA, K. (1986). “Effects of benzene inhalation on lymphocyte subpopulations and immune response in mice.” Toxicol. Appl. Pharmacol. 85:92-101.

    ARANYI, C., O'SHEA, W.J., GRAHAM, J.A., and MILLER, F.J. (1986). “The effects of inhalation of organic chemical air contaminants on murine lung host defenses.” Fundam. Appl. Toxicol. 6:718-720.

    ASHBY, J. and TENNANT, R.W. (1988). “Chemical structure, Salmonella mutagenicity and extent of carcinogenicity as indicators of genotoxic carcinogenesis among 222 chemicals tested in rodents by the U.S. NCI/NTP.” Mutat. Res. 204:17-115.

    AUSTIN, H., DELZELL, E., and COLE, P. (1988). “Benzene and leukemia: A review of the literature and a risk assessment.” Am. J. Epidemiol. 127:419-439.

    AVIS, S.P. and HUTTON, C.J. (1993). “Acute benzene poisoning: A report of three fatalities.” J. Forensic Sci. 38:599-602.

    AXELROD, J. and REISINE, T.D. (1984). “Stress hormones: Their interaction and regulation. ” Science 224:452-459.

    BAARSON, K.A. and SNYDER, C.A. (1991). “Evidence for the disruption of the bone marrow microenvironment by combined exposures to inhaled benzene and ingested ethanol.” Arch. Toxicol. 65:414-420.

    BAARSON, K.A., SNYDER, C.A., and ALBERT, R.E. (1984). “Repeated exposures of C57B6 mice to 10 ppm inhaled benzene markedly depressed erythropoiesis colony formation.” Toxicol. Lett. 20:492-503.

    BAARSON, K.A., SNYDER, C.A., GREEN, J.D., SELLAKUMAR, A., GOLDSTEIN, B.D., and ALBERT, R.E. (1982). “The hematotoxic effects of inhaled benzene on peripheral blood, bone marrow, and spleen cells are increased by ingested ethanol.” Toxicol. Appl. Pharmacol. 64:393-404.

    BASLO, A. and AKSOY, M. (1982). “Neurological abnormalities in chronic benzene poisoning. A study of six patients with aplastic anemia and two with preleukemia.” Environ. Res. 27:457-465.

    BERN, M.M. (1982). “Hematologic problems in diabetes.” In: Clinical Diabetes Mellitus (G.P. Kozak, ed.). W.B. Saunders Co., Philadelphia, PA. pp. 401-405.

    BENNETT, P.H. (1990). “Epidemiology of diabetes mellitus.” In: Diabetes Mellitus, Theory and Practice (H. Rifkin and D. Porte Jr., eds.). Elsevier, New York, NY. pp. 357-377.

    BERNADELLI, L., DEMARCO, R., and TINELLI, C. (1987). “Cancer mortality in an Italian rubber factory.” Br. J. Ind. Med. 44:187-191.

    BEZABEH, S., ENGEL, A., MORRIS, C.B., and LAMM, S.H. (1996). “Does benzene cause multiple myeloma? An analysis of the published case-control literature.” Environ. Health Perspect. 104(Suppl 6):1393-1398.

    BIRNBAUM, L.S. (1994). “Advances in estimating and predicting health effects from exposure to environmental toxicants. In Proceedings of the International Congress on the Health Effects of Hazardous Waste. JS Andrews, H Frumkin, BL Johnson, MA Mehlman, C Xintaras, and JA Bucsela (eds.). U.S. Department of Health and Human Services, Public Health Service, Atlanta, Georgia. 435-447.

    BLECHA, F. and BAKER, P.E. (1986). “Effect of cortisol in vitro and in vivo production of bovine interleukin 2.” Am. J. Vet. Res. 47:841-845.

    BLOCH, P., KULIG, A., PARADOWSKI, M., and WYBRZAK-WROBEL, T. (1990). “The toxicodynamics of benzene, ethanol, and benzene plus ethanol based on the histopathological examination of selected organs in the rats.” Polish Journal of Occupational Medicine 3:69-82.

    BOIS, F.Y. and PAXMAN, D.G. (1992). “An analysis of exposure rate effects for benzene using a physiologically based pharmacokinetic model.” Regul. Toxicol. Pharmacol. 15:122-136.

    BOIS, F.Y., SMITH, M.T., and SPEAR, R.C. (1991a). “Mechanisms of benzene