A New Way to Look at Old Questions of Silica Carcinogenicity

Abstract

Although results from some epidemiologic studies indicate an association between silicosis and lung cancer, other investigations have not confirmed these findings. The discrepancies may be related to methodologic problems. These include the lack of autopsy confirmation for the presence of silicosis or of information on the histologic lung cancer type, and unavailable or uncertain estimates of exposures to silica and to known carcinogens that may interact with silica or silicosis. This study will be designed to avoid the above weaknesses and reconcile observations in humans with results from animal studies, which show a carcinogenic effect for silica. We have formulated a series of testable hypotheses based on plausible mechanisms of carcinogenesis. These hypotheses suggest that silica interacts with known carcinogens (e.g., radon or cigarette smoke), and that only specific types of lung cancer are associated with silica or silicosis: (1) if silica or silicosis of lung parenchyma causes proliferation of adjacent epithelial cells previously damaged by known carcinogens (e.g., radon or cigarette smoke), then the percentage of peripheral adenocarcinomas will be increased; and (2) if silicosis of hilar lymph nodes increases residence time of known carcinogens such as radon and cigarette smoke, then the percentage of centrally located squamous and small cell carcinomas will be increased. To test these hypotheses, we plan to compare the histologic type and location of lung cancers in subjects with silicosis to the histologic type and location of lung cancer in subjects without silicosis. Access has been obtained to an autopsy archive comprising 30,000 cases, including some 5000 lung cancers among former East German uranium miners. Information on smoking habits is available for those who died after 1971, and detailed employment histories for all decedents have been computerized. Increasingly detailed records of radon and dust levels underground, from the late 1950s onward, provide an opportunity to develop jobexposure matrices that may enable estimation of individuals' exposures to these and other agents. The efficiency of these procedures and of the proposed laboratory analyses is being tested by sampling 250 autopsy cases based on year of birth, year of death, type of employment in uranium mining (underground, surface, none), gender, and pathologic diagnosis (primary lung cancer, no cancer). This should generate an informative distribution of exposures to radon and to dust and will permit refinement of provisional power estimates. The population and data identified are suitable in principle for testing the research hypotheses. Results from the current pilot study will deter mine whether the longer-term work plan is feasible in practice.