Benzene-associated leukemia and its risk assessment

Abstract

There is sufficient epidemiological evidence indicating that benzene is a human carcinogen. Although some reports suggest an association of benzene with multiple myeloma, malignant lymphoma and lymphatic leukemia, the closest association is found with the development of acute myeloleukemia. In contrast to epidemiological findings, benzene has induced a variety of tumors in tissues other than bone marrow in experimental animals. Attempts to induce acute myeloleukemia in rodents by benzene have not been fully successful until recently. While relatively low doses of benzene have been confirmed to induce chromosomal aberrations, sister chromatid exchanges and micronuclei in rats and mice, it has not been proven that similar changes occur in humans. For evaluation of the dose-response relation of benzene-induced leukemia, the epidemiological studies with a Pliofilm (rubber hydrochloride) cohort are considered most reliable and useful. According to recent re-examinations of the Pliofilm cohort, it is highly possible that the past epidemiological studies with this cohort underestimated the benzene exposure level, and thus overestimated the risk of benzene toward leukemia. Many extrapolation models including a linear model, a linearized multistage model, a conditioned log- logistic model, a quadratic model and a proportional hazards model have been used in the calculation of benzene risk. The risk values calculated so far with the Pliofilm cohort greatly differ according to the extrapolation models employed, and thus it cannot be determined which model is superior. While benzene exposure in the workplace is decreasing, petroleum containing a few percent of benzene has become an important source of benzene exposure for both industrial workers and the general population. Some reports suggest a positive association between gasoline exposure and the development of acute myeloleukemia.