2.1. Industry-based studies
Austin and Schnatter (1983) performed a nested case-control study of brain cancer within a cohort of employees at a petrochemical plant in Texas (United States). Twenty-one deceased brain tumour patients and two control groups (80 deceased ex-employees in each) were selected. Job history records were assessed by industrial hygienists for the purpose of assigning potential for exposure to each of 42 substances, one of which was toluene. Results were expressed as percentages of cases and controls exposed. Cases had lower exposure prevalence than controls (36% versus 45-53%) [leading to an apparent approximate odds ratio of 0.6, 95% CI, 0.2-2.2]. [The Working Group had some difficulty understanding the constitution of the control groups.]
In a nested case-control study among rubber workers in the United States (Wilcosky et al., 1984), described in more detail in the monograph on dichloromethane (see this volume), one of the substances evaluated was toluene and another was ‘solvent A’ (a proprietary mixture containing mostly toluene). For toluene itself, the numbers of exposed cases were very low (less than three for each case series). For lung cancer, the odds ratio was 0.6 based on three exposed cases. For lymphatic leukaemia, there were two cases exposed to toluene (odds ratio, 3.0; p > 05).There were somewhat higher numbers exposed to ‘solvent A’, with increased relative risks for stomach cancer (odds ratio, 1.4; n = 15), lymphosarcoma (odds ratio, 2.6; n = 6) and lymphatic leukaemia (odds ratio, 2.8; n = 7). [The Working Group noted that the numbers of cases exposed to pure toluene was small and the odds ratio estimates imprecise. Workers were typically exposed to multiple exposures and positive associations were found for many of the other substances analysed in this study, indicating a lack of specificity in the toluene or ‘solvent A’ associations].
Carpenter et al. (1988) carried out a nested case-control study of cancer of the central nervous system among workers at two nuclear facilities located in Tennessee (United States). They identified 89 cases (72 males and 17 females) who had died between 1943 and 1979. Four controls, living at the time the case was diagnosed, were matched to each case. Job history records were scrutinized by an industrial hygienist to assess potential exposure to each of 26 chemicals or chemical groups. Toluene, xylene (see this volume) and 2-butanone (methyl ethyl ketone) were evaluated as one chemical group; the matched relative risk was 2.0 (95% confidence interval (CI), 0.7-5.5; n = 28) in comparison with unexposed workers. Almost all cases had had low exposure, according to the classification used and there was no dose-response trend. The authors stated that the relative risks were adjusted for internal and external exposure to radiation.
Svensson et al. (1990) studied a cohort of 1020 Swedish rotogravure printers exposed primarily to toluene and employed for a minimum period of three months in eight plants during 1925-85. Data were available on air levels of toluene since 1943 in one plant and since 1969 in most. Based on these measurements and on present concentrations of toluene in blood and subcutaneous fat, the yearly average air levels in each plant were estimated. They reached a maximum of about 450 ppm [1700 mg/m3] in the 1940s and 1950s but were only 30 ppm [113 mg/m3] by the mid-1980s. Exposure to benzene had occurred up to the beginning of the 1960s, but not since then. Records of employment were combined with these retrospectively estimated plant-specific exposure levels to derive cumulative exposure estimates. The mortality experience of the cohort, during the follow-up period of 1952-86, was compared with that of the geographical region in which the plants were located, and cancer incidence, during the follow-up period of 1958-85, was analogously compared with regional incidence rates. The ‘all causes’ standardized mortality ratio (SMR) was 1.0 (129 observed deaths). There was no increase in mortality from non-malignant respiratory diseases (SMR, 0.8; 95% CI, 0.3-1.9; n = 5). For all cancers combined, there was some overall excess of mortality (SMR, 1.4; 95% CI, 1.0-1.9; n = 41) and morbidity (standardized incidence ratio (SIR), 1.3; 95% CI, 1.0-1.6). Among specific cancers, there were no excess risks for urinary cancers or leukaemias, lymphomas and myelomas. There were indications of excess risk for respiratory tract cancer (SMR, 1.4; 95% CI, 0.7-2.5; n = 11; SIR, 1.8; 95% CI, 1.0-2.9; n = 16) , for stomach cancer (SMR, 2.7; 95% CI, 1.1-5.6; n = 7; SIR, 2.3; 95% CI, 0.9-4.8; n = 7) and colo-rectal cancer (SMR, 2.2; 95% CI, 0.9-4.5; n = 7; SIR, 1.5; 95% CI, 0.7-2.8; n = 9) . Restricting analysis to those with at least five years of exposure did not lead to higher relative risk estimates. Further, there was no dose-response relationship with cumulative toluene dose (ppm years). [The Working Group noted that this study population had the ‘purest’ exposure to toluene of the groups evaluated in this monograph. This study had the best exposure assessment. Although the absence of an excess risk of nonmalignant respiratory disease is reassuring, it was based on very small numbers and thus does not prove that this cohort had ‘normal’ smoking habits].
Blair et al. (1998) updated a cohort mortality study reported by Spirtas et al. (1991) on 14 457 workers who had been employed as civilians for at least one year during the interval 1952 to 1956 in an aircraft maintenance facility located in Utah (United States). The study methods are described in the monograph on dichloromethane (see this volume). About 13% of the cohort were deemed to be exposed to toluene (Stewart et al., 1991). Using Poisson regression analysis, rate ratios were estimated for each of three types of cancer, multiple myeloma, non-Hodgkin lymphoma and breast cancer. Among toluene-exposed workers, there was an indication of an excess of multiple myeloma among women (RR, 5.0; 95% CI, 1.1-23.1; n = 4) but not among men (RR, 0.9; 95% CI, 0.2-4.8; n = 2). There was no meaningful excess risk of non-Hodgkin lymphoma among men (RR, 1.0; 95% CI, 0.1-4.2; n = 3) or among women (RR, 2.2; 95% CI, 0.4-13.1; n = 2). There was a slight excess of breast cancer (RR, 2.0; 95% CI, 0.9-4.2; n = 10). [The Working Group noted that the numbers on which these associations were based were very small and that workers typically had multiple exposures.]
Walker et al. (1993) conducted a cohort mortality study among 7814 shoe-manufacturing workers (2529 males and 5285 females) from two plants in Ohio (United States) that have been in operation since the 1930s. The workers, men and women, were potentially exposed to solvents and solvent-based adhesives. It was thought that toluene may have been a predominant exposure, but a hygiene survey in 1977-79 showed that, in addition to toluene (10 measurements ranged from 10 ppm to 72 ppm [38-270 mg/m3]), there were also 2-butanone (methyl ethyl ketone), acetone, hexane and several other solvents in concentrations as high as or higher than that of toluene. It is not clear whether these substances were present in earlier years. Benzene (IARC, 1987) may have been present as an impurity of toluene. Mortality follow-up was from 1940 to 1982. Relative risk estimates (SMRs) for white workers were derived by comparison with the general population of the United States. Among men, the SMR for all causes of death combined was close to 1.0, as was the SMR for all cancers combined. This cohort had no excess of lymphatic and haematopoietic cancer as a whole (SMR; 0.9; 95% CI, 0.6-1.3; n = 29) nor for any subtype. There were excess risks of lung cancer among men (SMR, 1.6; 95% CI, 1.2-2.0; n = 68) and among women (SMR, 1.3; 95% CI, 0.9-1.9; n = 31). Relative risk of lung cancer did not increase with increasing duration of employment. Mortality from chronic non-malignant respiratory disease was significantly elevated among men (SMR, 1.6; 95% CI, 1.1-2.2) but was less than expected among women (SMR, 0.8; 95% CI, 0.4-1.3), a finding suggesting a possible contribution of smoking to the male mortality from respiratory cancer. Adjustment for the potential effects of smoking by Axelson’s (1978) method reduced the relative risk estimate for lung cancer to 1.4 (95% CI, 1.1-1.8). There were slight excess risks for colon cancer among men (SMR, 1.3; 95% CI, 0.8-2.1; n = 18) and among women (SMR, 1.2; 95% CI, 0.8-1.8; n = 28). Other cancers showed no excess risk. [The Working Group noted that there was sparse information on what substances were historically present in this workplace. The procedure for adjustment of smoking is imperfect and could leave a confounded estimate.]
