Biological and Statistical Approaches to Predicting Human Lung Cancer Risk from Silica

Abstract

Chronic inflammation is a key step in the pathogenesis of particle-elicited fibrosis and lung cancer in rats, and possibly in humans. In this study, we compute the excess risk estimates for lung cancer in humans with occupational exposure to crystalline silica, using both rat and human data, and using both a threshold approach and linear models. From a toxicokinetic/dynamic model fit to lung burden and pulmonary response data from a subchronic inhalation study in rats, we estimated the minimum critical quartz lung burden (Mcrit) associated with reduced pulmonary clearance and increased neutrophilic inflammation. A chronic study in rats was also used to predict the human excess risk of lung cancer at various quartz burdens, including mean Mcrit (0.39 mg/g lung). We used a human kinetic lung model to link the equivalent lung burdens to external exposures in humans. We then computed the excess risk of lung cancer at these external exposures, using data of workers exposed to respirable crystalline silica and using Poisson regression and lifetable analyses. Finally, we compared the lung cancer excess risks estimated from male rat and human data. We found that the rat-based linear model estimates were approximately three times higher than those based on human data (e.g., 2.8% in rats vs. 0.9-1% in humans, at mean Mcrit lung burden or associated mean working lifetime exposure of 0.036 mg/m3). Accounting for variability and uncertainty resulted in 100-1000 times lower estimates of human critical lung burden and airborne exposure. This study illustrates that assumptions about the relevant biological mechanism, animal model, and statistical approach can all influence the magnitude of lung cancer risk estimates in humans exposed to crystalline silica.