Abstract:
Lung cancer risk associated with cotton dust and endotoxin exposure was examined in a case-cohort study of female workers in Shanghai. Among 26,400 women, 628 cases and 3148 non-cases were identified from 503 factories. Information describing factory operations, dust measurements, and subjects’ work histories were abstracted from industry records. The historical dust measurements were collected using a standard Chinese Dust Sampler (CDS). The results of a field survey at three Shanghai textile factories were used to compare the performance of the CDS o the American standard vertical elutriator (VE). The CDS, a high-volume total dust area sampler, collects 2 to 10 times more dust than the VE, a size- selective method, depending on the manufacturing process. In the textile environment, the CDS over-samples small particles and is prone to collecting large and biologically irrelevant particles.
A predictive exposure model was developed using 2413 historical cotton dust measurements and the abstracted factory information. The estimates of exposure were compared to independently collected measurements gathered over a 14-year period at two of the participating factories. The validated exposures were assigned to 4581 jobs in cotton factories of the 7242 jobs held among 3812 subjects. The remaining 2661 non-cotton jobs were assigned no exposure. Retrospective assessments of endotoxin exposure were assigned based on the estimated cotton dust concentrations, endotoxin levels measured during the sampling survey, and measurements from previously published studies.
Risk estimates, determined using a Cox proportional hazards model, indicated a negative exposure-response relationship. At the end of follow-up, women in the highest categories of exposure either to cotton dust or endotoxin possessed a statistically significant reduction in their hazards ratio of 0.7 (SE=0.12). When disease latency was considered, the hazard ratios among the highest exposure groups with 20 years of exposure discounted were 0.6 (SE=0.10) for cotton dust and 0.5 (SE=0.09) for endotoxin as compared to the lowest exposure category. The trends among the exposed categories were significant. Calculation of risk considering combined exposure was equivocal. Based on these and previously published results, and the existence of a plausible biological mechanism, long-term and high-level exposure to endotoxin reduces risk of lung cancer.