Chrysotile Asbestos - Chrysotile Fibres, Chrysotile Exposure
7.1.2.2 Comparisons of lung cancer exposure-response - critical studies
The slopes of the relationship between cumulative exposure to chrysotile and the relative risk of lung cancer are summarized in Table 23 for those studies that reported this information. These studies all expressed this relationship using the following linear relative risk (RR) model:
RR = 1 + B x E
where B is the slope and E is the cumulative exposure to chrysotile asbestos expressed in f/ml-years. The slopes from the studies of the mining and milling industries (0.0006 to 0.0017), the latter having been estimated on a subset of the cohort on which the former was based, and the friction production industries (0.0005 to 0.0006) are reasonably similar. Hughes et al. (1987) in a study of cement workers (section 7.1.2.1b) reported a similar slope (0.0003) in one plant (plant 1) that only used chrysotile, and a nearly 20-fold higher slope (0.007) among workers only exposed to chrysotile in another plant (plant 2).
The slopes of 0.01 and 0.03 reported for the two studies of the chrysotile-exposed textile workers conducted on overlapping populations, as well as the slope of 0.007 from one of the two plants (plant 2) of cement workers in the study of Hughes et al. (1987), were an order of magnitude greater than those reported for the other cohorts. It should be noted that the two textile cohorts were identified from the same textile facility, but were based on different cohort definitions. Hence, it is not surprising that the slopes from these two studies were similar. The slopes in the studies of chrysotile-exposed textile workers are also remarkably similar to those reported in other studies of textile workers with mixed fibre exposures (Peto, 1980; McDonald et al., 1983b; Peto et al., 1985). This similarity in findings provides some support for the validity of the slopes reported in the chrysotile-exposed textile cohorts.
The reason for the much higher slopes observed in studies of textile workers is unknown, although several possible explanations have been suggested. The first is that these differences might be attributed to errors in the classification of exposures in these studies. Particular concern has been raised about errors in the exposure assessment related to conversions from mpcm (mpcf) to fibres/ml that were performed, particularly in the mining and milling studies (Peto, 1989). Sebastien et al. (1989) conducted a lung burden study specifically designed to examine whether the differences in lung cancer slopes observed in the Charleston chrysotile textile cohort and the Quebec mining industries could be explained by differences in errors in exposure estimates. Lung fibre concentrations were measured in: (a) 32 paired subjects that were matched on duration of exposure and time since last exposure; and (b) 136 subjects stratified on the same time variables. Both analyses indicated that the Quebec/Charleston ratios of chrysotile fibres in the lungs were even higher than the corresponding ratios of estimated exposures. This finding was interpreted by the author as being clearly inconsistent with the hypothesis that exposure misclassification could explain the large discrepancy in the lung exposure-response relationships observed in the two cohorts.
Sebastien et al. (1989) offered a second possible explanation for the differences, which was that observations in the Charleston textile cohort may have been confounded by exposure to mineral oils. Dement et al. (Dement, 1991; Dement et al., 1994) have conducted two nested case-control studies designed to evaluate the potential for confounding by exposure to mineral oils in the Charleston textile cohort. Cases and controls were assigned to a qualitative mineral exposure category as well as asbestos exposure. The relationship between chrysotile exposure and lung cancer risk was observed to be virtually unaffected by control for exposure to mineral oils in these analyses. The authors concluded that confounding by machining fluids was unlikely. It should also be noted that studies of other cohorts of workers exposed to machining fluids (including mineral oils) have failed to detect an increase in lung cancer risk (Tolbert et al., 1992).
Finally, it has been suggested that the higher lung cancer risk observed among textile workers might be explained by differences in fibre size distributions (Dement, 1991; McDonald et al., 1993; Dement et al., 1994). Textile operations have been shown to produce fibres that are longer in length than in mining and other operations using chrysotile asbestos (Dement & Wallingford, 1990). The study of Sebastien et al. (1989) also examined the hypothesis that differences in fibre size distribution could explain the discrepancy in lung cancer exposure-response relationships between the Quebec mining and Charleston textile cohorts. Although the authors concluded that differences in fibre size distributions were an unlikely explanation, it was noted that there was a slightly higher percentage of long chrysotile fibres (> 20.5 mm) in the lungs of workers from the Charleston textile facility than in the Quebec miners.