362 Updated results on the risk of death from lung cancer by silica dust in German Wismut uranium miners, 1946–2008

Abstract

Objectives To investigate the exposure-response relationship between silica dust and lung cancer mortality within German uranium miners in the extended follow-up period from 1946 to 2008. The cohort consists of 58,672 miners including 2.2 Mio person-years and 3,477 lung cancer deaths. Methods Individual information on occupational exposure to crystalline silica in mg/m3-years and the potential confounders radon and arsenic is available for all cohort members based on a detailed job-exposure matrix. Internal Poisson regression with baseline-stratification by age and calendar year was used to estimate the excess relative risk (ERR) per mg/m3-year. Several spline functions were applied to investigate the exposure-response relationship. Detailed adjustment for cumulative radon and arsenic exposure was conducted. Effect modification by age, time and exposure-rate was tested. Additionally detailed risk analyses with specific focus on the low dose range have been performed. Results All miners were exposed to crystalline silica at any time. The mean cumulative silica exposure was 5,9 mg/m3-years with a maximum of 56 mg/m3-years. A piece-wise linear spline function with a knot at 10 mg/m3-years provided the best model fit. After full adjustment for radon (continuous variable with consideration of time since median exposure and dose rate) and arsenic no increase in risk below 10 mg/m3-years was observed. Fixing the parameter estimate of the ERR in this range at 0 provided the best model fit with an ERR of 0.065 (95% CI: 0.043, 0.087) above 10 mg/m3-years. The silica-induced lung cancer risk decreased with increasing attained age. Data of a nested case-control study did not indicate a major correlation between silica and smoking status. Conclusions The study confirms a positive exposure-response relationship between silica and lung cancer, particularly for high exposures. No clear evidence was observed in the low-dose range. Confounding by smoking is unlikely but residual confounding cannot be fully excluded.