Abstract

The International Agency for Research on Cancer has deemed beryllium a Group 1 lung carcinogen, based in part on findings from a pooled cohort of U.S. beryllium processing workers. The U.S. Occupational Safety and Health Administration (OSHA) recently proposed a tenfold-decreased permissible exposure limit for beryllium, based partly on estimates of lung cancer risk from recent follow-up of this pooled cohort (the only cohort available with quantitative exposure information). Criticisms in the literature and public comments hypothesise that workers hired after 1954 (when exposures were lower) or exposed only to insoluble forms of beryllium do not exhibit increased risk of lung cancer. We evaluated this hypothesis by conducting Cox proportional hazards regression analyses in age-based risk sets within two (of three) plants in the pooled cohort. 98% of workers at these plants were hired 1955–1969. We used categorical and power models to evaluate exposure-response patterns for mean and cumulative beryllium exposure in the two-plant cohort, comparing findings to published estimates from the full pooled cohort. We also evaluated the distribution of exposure-years in each cohort by solubility class (soluble, insoluble, mixed). Mean beryllium exposure averaged <2 µg/m³ and the predominant form was insoluble, among the two-plant cohort. We observed 75 lung cancer cases (through 2005). Adjusting for confounders related to smoking and the healthy worker effect, we observed a monotonic increase in lung cancer mortality across exposure categories. The exposure-response coefficients (per unit of exposure) were 0.242 (p = 0.087) for mean exposure and 0.151 (p = 0.049) for cumulative exposure, compared to 0.155 and 0.094 (respectively) in the full cohort. The low exposure levels at these two plants and the predominance of insoluble beryllium suggest that the overall pooled cohort findings upon which OSHA’s lung cancer risk assessment is based are highly relevant for current workers exposed to all forms of beryllium.