Differences between the data bases, statistical analyses, and interpretations of lung tumors of the 19-dust study: two controversial views.

The article presents a comparison of two opposing evaluations and interpretations of the carcinogenicity study with 19 granular dusts. It is one of the two responses to a Letter to the Editor of Exp Toxicol Pathol by Morfeld and Borm(1), who had analysed selected data from the primary source of our study [Borm et al., Inhal Toxicol 2000;12(Suppl 3):225-31; Borm et al., Int J Cancer 2004;110:3-14; Morfeld et al., Inhal Toxicol 2006:18:215-25]. In contrast, our statistical analyses are based on the authentic and complete results of the study published with a detailed description of materials, methods and results [Pott and Roller, Eur J Oncol 2005,10:249-81; Mohr et al. Exp Toxicol Pathol 2006;58:13-20; Roller and Pott, Ann NY Acad Sci 2006,1076:266-80]. Analysis of the tumor incidences from non-specifically toxic dusts in terms of a usual multistage model leads to reasonable dose-response curves and to the finding that tumor risk depends on the retained dust volume and on mean particle size. The model is adequate because survival of the groups included in the analysis was not reduced. In contrast to this analysis, Morfeld et al. (2006) calculated an implausible threshold of 10mg instilled total dust per rat and a saturation dose of 20mg. These values are not compatible with the data because three dose groups with highly significantly increased tumor incidences (up to 67%) lie within the 95% confidence interval of the threshold dose of Morfeld et al. (2006). The claim of Morfeld [Zbl Arbeitsmed 2004;54(7):246-58] to consider as "of interest" only carcinogenicity tests with doses similar to current occupational exposure limit values would make it impossible to identify the carcinogenicity of non-specifically toxic dusts even if the excess risk is higher than 1 in 100.