Benzene, an experimental multipotential carcinogen: results of the long-term bioassays performed at the Bologna Institute of Oncology.

In 1976, a systematic and integrated project of long-term carcinogenicity bioassays began at the Bentivoglio Experimental Unit of the Bologna Institute of Oncology. The Bologna experiments proved for the first time that benzene is an experimental carcinogen. These experiments demonstrated that benzene is carcinogenic when administered by ingestion and by inhalation and that it cause tumors in the various tested animal models (Sprague-Dawley rats, Wistar rats, Swiss mice, and RF/J mice). They also showed that benzene is a multipotential carcinogen, as it produces a variety of neoplasias in one or more of the tested animal models, including Zymbal gland carcinomas, carcinomas of the oral cavity, nasal cavities, skin, forestomach, and mammary glands, as well as angiosarcomas of the liver, hemolymphoreticular neoplasias, tumors of the lung, and possibly hepatomas. The Bologna experiments also indicated a clear-cut dose-response relationship in benzene carcinogenesis. This report presents the up-to-date results of the Bologna project. The need for more experimental research aimed at assessing the carcinogenic effects of low doses of benzene, of chemical mixtures containing benzene, and of benzene substitutes is emphasized. Also recommended are more comprehensive epidemiological investigations, extended to all types of malignancies, with particular regard to lung carcinomas.


Introduction
Benzene has been produced industrially from coal since 1849 and from petroleum since 1941. At present the major source of benzene is petroleum. Benzene is one of the largely diffused and produced industrial compounds. It is a constituent of crude oil, it is present in gasoline and other fossil fuels, and it is currently produced at the rate of about 15 million tons per year (the major producers are the U.S., Japan, and Western Europe). The total global annual cycle of benzene is estimated to be 32 million tons per year (1).
The major use of benzene in past was in blends with gasoline. Although this use has been reduced in the United States, benzene is still extensively employed in many countries for the production of commercial gasoline. The benzene content in gasoline varies from country to country, and its range is estimated to be from 1 to 15%. Currently, benzene is used as a chemical intermediate for the production of many important industrial compounds, such as ethylbenzene (used in the production of styrene), phenol, cyclohexane, maleic anhydride, aniline, dichlorobenzenes, etc., which, in turn, supply numerous sectors of the chemical industry, especially those producing plastics, resins, elastomers, dyes, and pesticides. In the past, benzene was also used as a solvent for paints and rubber, in the production of rubber cement (widely used in the shoe and garment industries), and in the manufacture of artiflcial leather. It has also been used in medicine in the treatment of hemoblastomas (leukemias, polycytemia, and malignant lymphomas) and in veterinary medicine of disinfecting wounds.
Of the 32 million tons of benzene circulated globally per year, 4 million tons are estimated to be lost to the environment (1). The major source is motor vehicle emission and evaporation losses during handling, distribution, and storage of petrol (2). Burning wood and organic material also results in an appreciable release of benzene. Tobacco smoke contains benzene at levels of 47 to 64 ppm (3). It is believed that plant and animal matter also release benzene into the environment (4). Population groups that may be exposed to benzene in-clude workers engaged in its production; workers in chemical industries using benzene as an intermediate; workers in industries producing materials containing benzene as a constituent (gasoline), as a solvent (rubber cement), or as an impurity (i.e., industry toluene); people living near factories producing or using benzene, or compounds containing it; tobacco smokers; and the general population (particularly in industrialized towns), as benzene is contained in gasoline, drinking water, and many other goods and is highly volatile. Prolonged exposure to benzene causes toxic effects on bone marrow, both in animals and humans. The toxic effects on the hematopoietic system in humans have been known for about 90 years and are well documented in the literature. The classical clinical finding in benzene hematotoxicity is a decrease in the various formed elements of circulating blood (pancytopenia) as a consequence of the decrease in identifiable granulocyte, erythrocyte, and platelet precursors within the bone marrow. The association between long-term exposure to benzene and the occurrence of leukemia was suggested as early as 1928 by Delore and Borgomano (5), who reported a lymphoblastic leukemia in a worker who had been exposed to benzene for 5 years.
In spite of its industrial importance, widespread use, ubiquitous diffusion, the large number of people potentially exposed, and the early reports of occupational leukemias, there were no adequate epidemiological investigations nor adequate experimental research on benzene until the mid-1970s.
Knowledge of Benzene Carcinogenicity until the Mid-1970s Human data are based, almost exclusively, on reports of a series of clinical cases of leukemia (generally in individuals with a history of benzene myelotoxicity) and on more indirect epidemiological investigation on the correlation between benzene exposure and the incidence of leukemias.
Since the first report of Delore and Borgomano (5), many leukemias in people exposed to benzene were the subject of case reports. Vigliani in 1976 (6) stated that it... an approximate estimation of the available literature, including some unpublished North Italian cases of which we have knowledge, puts the number of known cases of leukemia attributed to benzene at, at least, 150." These cases were collected and reviewed by Goldstein (7) ( Table 1).
Acute myelogenous leukemia has been the most frequent form of leukemia associated with benzene exposure. Other forms of leukemia that have also been associated with benzene are erythroleukemia, acute monocytic leukemia, chronic myelogenous leukemia, myelofibrosis and myeloid metaplasia, thrombocytemia, acute lymphoblastic leukemia, chronic lymphocytic leukemia, and lymphomas of various types.
The majority of the cases have been found in Italy (6,8), France (9)(10)(11)(12)(13)(14), and Turkey (15,16). Most of the cases oc- Table 1. Case reports of hemolymphoreticular neoplasia and correlated diseases observed in individuals exposed to benzene reported in scientific literature until the mid-1970s (7 curred among shoemakers and garment industry workers handling rubber cement. The variety in type and distribution of the described leukemias in the major case reports (with a more or less pronounced prevalence of acute myelogenous leukemia) may be due to several fattors: a) difference in level of exposure; b) exposure in association with other agents, with toxic effects on hematopoietic tissues; c) individual responsiveness; and d) differences in the morphological interpretation of the hemopathological pictures, which vary with time, country, and training.
All of the previously mentioned reports deal with a limited series of cases, and therefore the cases are fragmented into a large number of reports (Table 1). In most of these reports, the causal relation between benzene exposure and leukemia is developed more from the history of the single cases, showing characteristic sequence, benzene exposure-myelotoxicity-leukemia, than from statistical and biological evaluation.
Moreover, the indirect results of one epidemiological investigation performed in Japan, were reported. Ishimaru and co-workers (17) examined 303 cases of leukemia that occurred among adult survivors of the atomic bomb explosions at Hiroshima and Nagasaki and compared them with 303 matched controls exposed to the same amount of ionizing radiation from the bomb. They found that the risk of leukemia was approximately 2.5 times higher among those with the history of a probable exposure to benzene or its derivatives and to medical X-rays.
Benzene was shown to be genotoxic to human blood cells. Increased rates of both stable and unstable chromosome changes have been described both in lymphocytes and bone marrow cells of patients with benzene hemopathy and in lymphocytes of workers with past exposure to benzene but without signs of poisoning (18)(19)(20)(21)(22)(23)(24). Chromosome damage from benzene can persist for years in longlived lymphocytes and may result in the formation of abnormal cell clones in the absence of any sign of disease (21,22). Increased chromosome aberrations of blood lymphocytes have also been reported in workers exposed to less than 25 ppm of benzene (25)(26)(27). There have been a number of reports of cases of benzene pancytopenia in which the observed bone marrow cellular atypias were progressing toward acute leukemias.
The available experimental data prior to 1976, summarized in Table 2, are scanty and insubstantial (28)(29)(30)(31). In IARC Monograph No. 7 (32), it was concluded that "benzene has been tested only in mice by subcutaneous injection and skin application. The data reported do not permit the conclusion that carcinogenic activity has been demonstrated." Knowledge of Benzene Carcinogenicity from the Mid-1970s to the Present A systematic integrated experimental project (the largest up to present) of long-term carcinogenicity bioassays on benzene began in April 1976 at the Bentivoglio (BT) Experimental Unit of the Bologna Institute of Oncology. The project was aimed at studying the car-cinogenic effects of benzene administrated by different routes (ingestion and inhalation) at different daily doses/concentrations on animals of different species and strains (Sprague-Dawley and Wistar rats and Swiss and RF/J mice) and of different ages at the start of the treatment. As early as November-December 1977, preliminary results were published (33) showing that benzene was producing in Sprague-Dawley rats Zymbal gland carcinomas, an increase of other solid tumors and a marginal increase of malignant hemolymphoreticular neoplasias. Since then, the data of this project, have shown that benzene has carcinogenic effects when given by inhalation and by ingestion, it causes tumors in all tested animal species and strains, it is a multipotential carcinogen, as it produces a large variety of neoplasias, and there is a clear-cut dose-response relationship in benzene carcinogenesis. The data have been summarized, from 1977 to 1987, in various publications (34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45).
Further experimental bioassays were then performed in other laboratories. Bone marrow hyperplasia, thymic lymphoma (6/40), plasmacytoma (1/40), and leukemia (1/40) were reported in C57BL/6J mice exposed to air contain- ing 300 ppm benzene for 6 hr/day, 5 days/week, for 488 days as compared with an incidence of 2/40 lymphomas (nonthymic) in the controls (46). Myelogenous leukemia occurred in 2/40 CD-1 mice exposed to air containing 300 ppm benzene for 6 hr/day, 5 days/week, for life (47). The results of these two studies, because of the small number of animals used and the marginal increase of the observed hemolymphoreticular neoplasias, do not allow establishment of a positive association between the onset of the neoplasias and the benzene exposure.
NTP performed a long-term carcinogenicity bioassay on F344/N rats and B6C3F1 mice (50). Fifty male and fifty female/dose level were gavaged 5 days/week for 103 weeks. Doses of 0, 50, 100, or 200 mg/kg body weight benzene in corn oil (5 mL/kg) were administered to male rats. Doses of 0, 25, 50, or 100 mg/kg benzene in corn oil were administered to female rats and to male and female mice. The results of this experiment confirm the findings of Bologna-BT project, i.e., that benzene is a multipotential carcinogen. A definite or marginally increased incidence of the following tumors was found in benzeneexposed animals of one or both sexes: tumors of Zymbal glands, oral cavity, and skin in rats; tumors of Zymbal glands, hemolymphoreticular tissues, lungs, Harderian glands, mammary glands, preputial glands, forestomach, ovary, and liver in mice.
Recent epidemiological studies of small cohorts exposed to benzene have demonstrated a causal association with leukemia (51-58). Infante et al. (51) reported an increased risk of leukemias among workers at three rubber hydrochloride plants (in two Ohio locations) who were exposed to benzene in the years 1940 to 1949. In that investigation, the vital status of 75% of the population was ascertained. The level of exposure was estimated to be not greater than the standards at that time would have allowed. Rinsky et al. (53) reported the findings of a more extensive investigation on the same rubber workers. The major findings of this important study may be summarized as follows: a) the workers were exposed to only one agent that has been associated with blood dyscrasias, i.e., benzene; b) exposure data, uncommonly complete throughout the study period , indicated that the exposures of the workers were, for the most part, within limits permissible at the time (these limits are not greatly higher than the current legal standards); and c) the vital status (alive, dead, and cause of death) for 98% of the study population was ascertained. There were seven deaths from leukemia (myelocytic or monocytic) among 748 workers; this rate is 5.6-fold greater than would be expected in a comparable population. For workers exposed 5 years or more, there was a 21-fold increased risk of death from leukemia. In 1986, Rinsky et al. (57,58) reexamined the updated mortality on the same cohorts and calculated a cumulative benzene exposure index (ppm x years) for each cohort member. These authors found that the standard mortality ratio (SMR) for leukemia was 328 and for multiple myeloma was 398. With stratification of the cohort by cumulative exposure, the SMRs for leukemia increased from 105 in workers with less than 40-ppm years exposure, to 314 in workers with 40to 199-ppm years, to 1757 in those with from 200to 399-ppm years, and to 4535 in those with 400-ppm years or more.
In 1985, Maltoni et al. reported the first experimental evidence of the carcinogenicity of benzene-correlated compounds, namely toluene and xylene (44). These results has been recently confirmed by the same authors by further experiments whose results are now in publication.
The history of benzene carcinogenicity is summarized in Table 3.
This report presents the up-to-date results of the experiments on benzene carcinogenicity performed by the Institute of Oncology of Bologna (BT experimental project).

Materials and Methods
The plan of the experiments is shown in Tables 4-10. The bioassay on RF/J Mice (Table 10) is part of a larger experiment aimed at studying the effect of 10% ethyl alcohol, administered instead of drinldng water, on benzene carcinogenesis (Table 11). (The results of the whole experiment are now being submitted for publication.) Data on test compounds and test animals are presented in Table 12. Details on the conduct of the experiments are as follows: *The animals were exposed by inhalation in air, 4 to 7 hr/day, 5 days/week, for 15 and 104 weeks. The chambers for inhalation exposure were stainless steel, with two glass doors, and measured 135 x 98 x 65 cm. The volume was 860 L. Continuous air flow provided 12 to 15 air changes per hour. Before introduction, air was filtered and the chamber arrangement was such that air flowed from one part of the chamber to the other with- Table 3. History of benzene carcinogenicity.
Year Report 1928 First report on an acute leukemia following benzene intoxication (5) 1960s Reports of cases of leukemias among workers (mainly shoemakers) heavily exposed to benzene, in Italy (8) 1970s Reports of cases of leukemias among workers (mainly shoemakers) heavily exposed to benzene, in Turkey (15,16) 1977 First evidence of carcinogenicity of benzene in experimental animals (rats) (33) 1977-1983 Experimental evikence showing that benzene is a multipotential carcinogen in rodents (rats and mice) (43,44) 1977-1986 Epidemiological evidence of benzene leukemogenicity among exposed workers in the U.S., also at low doses (51,53,57,58) 1983-1987 First experimental evidence of carcinogenicity of benzene correlated compounds (toluene and xylenes) (44) bOlive oil alone. bOlive oil alone. bOlive oil alone. out recirculation. The internal pressure was about 1 mm Hg less than that of the room where the chamber was situated, to avoid any possible contamination of the outside environment. Chambers were maintained at 210C ± 30C and at 50%± 10% relative humidity. Lighting was provided by room light. The chambers were cleaned at monthly intervals. Exposure chambers were provided with a fixed point matrix for checking the distribution of the test substance. During treatment the distribution was continuously monitored by gas chromatographs. *The animals were exposed by ingestion (stomach tube, made of stainless steel), once daily, 4 to 5 days/week, for 52 (Sprague-Dawley rats, RF/J mice), 78 (Swiss mice) and 104 weeks (Sprague-Dawley and Wistar rats). *All the animals were kept under observation until spontaneous death. *The status and behavior of the animals were examined 3 times daily. *The animals were submitted to clinical examination for gross changes every 2 weeks. *The animals were weighed every 2 weeks during treatment, and then every 8 weeks. *Full necropsy was performed on all the animals; see below. *The housing and the diet of the animals were the same adopted in the BT Experimental Unit during the last 15 years. *All the experiments were performed with the same highly standardized procedures in order to allow comparison. The tissues and organs submitted to histopathological examination were the following: subcutaneous lymph nodes, brain and cerebellum, pituitary, Zymbal glands, interscapular brown fat, salivary glands, Harderian glands, oral and nasal cavities (seven sections of the head), tongue, pharynx, thymus and mediastinal lymph nodes, lungs, diaphragm, liver, kidneys, adrenals, spleen, esophagus, mesenteric lymph nodes, stomach, various segments of the intestine, bladder, uterus, gonads, any other organs with pathological lesions, and, only for experiment BT 909, lachrymal and preputial glands.     The most frequent tumors in this strain of rats, on the basis of the literature and of the historical controls of the BT Experimental Unit, are mammary tumors, malignant hemolymphoreticular neoplasias (leukemias), pheochromocytomas, and pheochromoblastomas. Moreover a variety of other miscellaneous tumors were also observed (59).
The administration of benzene by ingestion is associ-ated with an increase of total malignant tumors and carcinomas of the Zymbal glands (with sebaceous and squamous patterns), oral cavity, nasal cavities, skin (of different histotypes), forestomach (together with an increase of acanthomas and dysplasias), and with liver angiosarcomas, and a marginal increase of carcinomas of the mammary glands, hepatomas, and leukemias. The administration of benzene by inhalation is associated with an increase of total malignant tumors and carcinomas of the Zymbal glands and oral cavity, and with a marginal increase of carcinomas of the nasal cavities, mammary glands, and hepatomas.           (Tables 26-28) The most frequently expected tumors in this strain of rats, on the basis of the literature and of the historical controls of the BT Experimental Unit, are mammary tumors, leukemias, pheochromocytomas, and pheochrom-oblastomas. Moreover, a variety of other miscellaneous tumors were also observed (59). The administration of benzene by ingestion is associated with an increase of total malignant tumors and carcinomas of Zymbal glands, oral cavity, and nasal cavities. Table 26. Experiment BT 907: incidence of total tumors in Wistar rats exposed to benzene by ingestion (stomach tube) for 104 weeks.   The most frequently expected tumors in this strain of mice, on the basis of the literature and of the historical controls of the BT Experimental Unit, are mammary carcinomas (in females), lung tumors, leukemias, and hepatomas. Moreover, a variety of other miscellaneous tumors were also observed (59).
The administration of benzene by ingestion is associated with an increase of total malignant tumors, carcinomas of the mammary glands, lung tumors (adenomas, adenomas in deviation, and adenocarcinomas), and carcinomas of the Zymbal glands (together with an increase of dysplasias). Table 29. Experiment BT 908: incidence of total tumors in Swiss mice exposed to benzene by ingestion (stomach tube) for 78 weeks.   (Tables 34-37) The most frequently expected tumors of this strain of mice, on the basis of the literature and of our experimental experience, are mammary carcinomas, pulmonary tumors, and leukemias. The most frequent histotype of hemolymphoreticular malignant neoplasias are lymphoblastic lymphosarcomas (much greater percentage) and lymphoblastic lymphosarcomas with histocytic component.
The administration of benzene by ingestion is associated with an increase of total malignant tumors, mammary carcinomas, lung tumors (adenomas, adenomas in deviation, and adenocarcinomas), and leukemias. In the treated animals the number of pulmonary tumors per tumor-bearing animal is greatly enhanced (Table 37).  Multipotential Carcinogenicity, particularly marked in the case of carcinomas of Zymbal Dose-Response, and Effect of Age glands, oral cavity, and nasal cavities when considered either singularly or together (Table 39).
A varilety of tumors is associated to benzene exposure.
An enhanced carcinogenic effect of benzene was obin the animals of all tested speCieS and strainS (Table 3. served in animals on which treatment was started dur-A dose-response relationship was seen in the experiing embryonal life (Table 24-28). ment with Sprague-Dawley rats. This relation appears

Conclusions
The experiments performed at the Bologna Institute of Oncology on benzene carcinogenesis have shown that benzene is a strong carcinogen on experimental animals; it is carcinogenic on four different types of experimental animals, i.e., Sprague-Dawley and Wistar rats, and Swiss and RF/J mice. Exposure to benzene is associated with an enhanced incidence of a variety of tumors; therefore, benzene must be considered a multipotential carcinogen. The neoplastic response associated with benzene exposure varies in the different types of tested animals. Benzene has carcinogenic effects when given both by inhalation and by ingestion. The carcinogenic effects of benzene increase by increasing the doses (daily dose, length of treatment). There is a high response when treatment is started during embryonal life.
The experimental research may still improve our knowledge of benzene carcinogenicity and correlated problems. At present, in our opinion, the following experimental research deserves full priority: a) studies on the carcinogenic effects of minimal doses of benzene (in the range of the present allowable levels), delivered by inhalation, to large groups of animals (megaexperiments); b) carcinogenicity studies on chemical mixtures containing benzene (fuels); c) carcinogenicity studies on chemically correlated and/or alternative compounds, i.e., toluene, xylenes, trimethylbenzenes, ethylbenzene, etc. Studies on these three fields of research are now ongoing or planned at the Bologna Institute of Oncology. Moreover, comprehensive epidemiological investigations on population groups exposed to benzene extended to all types of malignancies, with particular regard to lung carcinomas (on the basis of our recent experimental results) must be undertaken without delay.
This work was supported in part by EEC contract 323-79-4-ENV-I.