and medical records of workers who developed hematologic neoplasms and related disorders. Yin and colleagues (1996a) reported nine cases of aplastic anemia in the benzene-exposed cohort as compared with no cases in the nonexposed population. Because the study used such a large population, it was possible to detect differences in relatively infrequent outcomes. Furthermore, there was a careful review of medical records to confirm the diagnoses. However, there is a possibility that the results were confounded by other occupational exposures.
Potential risk factors for aplastic anemia were examined in a case-control study in Baltimore, Maryland. Linet and colleagues (1989) compared 59 cases of aplastic anemia diagnosed in 1975–1982 with 59 controls matched for age, sex, race, and geographic area and selected by random-digit dialing. An increased risk of aplastic anemia was associated with self-reported benzene exposure (OR=3.1, 95% CI=1.0–9.2). However, for the purposes of this review, the inferences from the study are limited by the fact that 41% of the patients with aplastic anemia were under 20 years old at diagnosis and would not have had substantial occupational exposures.
Two other studies provide information on the relationship between benzene and aplastic anemia. In a case series from Turkey, Aksoy and colleagues (1984) reported that about 23% of patients with aplastic anemia had reported exposure to benzene. The study examined potential risk factors but did not have a comparison population. Ott and colleagues (1978) reviewed the deaths (1938–1970) of 594 workers chronically exposed to benzene at concentrations of 1 ppm to over 30 ppm at a Dow Chemical plant. One death from aplastic anemia was reported, whereas only 0.1 would have been expected.
The relationship between changes in hematologic parameters and exposure to low concentrations of benzene has been extensively studied. However, the studies generally are cross-sectional and do not provide substantial information about persistent or long-term effects of interest in this review, and they generally have had inconsistent findings. When changes were seen in hematologic measures at low exposure, the differences (such as in hematocrit, hemoglobin, white-cell count and platelet count) often were not internally consistent with other measures (for example, an elevated mean red-cell volume would be expected but decreased mean corpuscular hemoglobin concentration would not). Changes in hematologic or immunologic parameters are not necessarily stages in the development of a pathologic process. For example, the finding of gradually lower numbers of blood cells does not mean that continued exposure would lead to the development of aplastic anemia.
Examples of studies of hematologic parameters include the studies by Kipen and colleagues (Cody et al., 1993; Kipen et al., 1988, 1990) who followed a cohort of rubber workers exposed to varying concentrations of benzene. Studies of this cohort are described in Chapter 6 regarding cancer outcomes, particularly leukemia (Rinsky et al., 1981, 1987). During the period from 1940–1948, as benzene exposures gradually dropped from 137 ppm to 32 ppm, white-cell counts rose (from 6200 to 9591), red-cell counts rose (from 4.67 to 5.13), and hemoglobin rose (from 97.0 to 108.0) (Kipen et al., 1988). Those findings indicate that exposure to relatively high concentrations of benzene depressed the production of blood cells. In the years after 1948, when benzene exposure was much reduced, the values for white- and red-cell counts were within the normal ranges (although a nonexposed comparison group was not studied in the same period).
A study by Collins and colleagues (1991) compared hematologic parameters of 200 workers at a chemical factory who were exposed to low concentrations of benzene (0.01–1.4