Investigation of the impact of pharmacokinetic variability and uncertainty on risks predicted with a pharmacokinetic model for chloroform

A sensitivity and uncertainty analysis was performed on the Reitz et al. (Toxicol. Appl. Pharmacol., 1990: 105, 443) physiologically based pharmacokinetic (PBPK) risk assessment model for chloroform. The analytical approach attempted to separately consider the impacts of interindividual variability and parameter uncertainty on the predicted values of the dose metrics in the model, as well as on liver cancer risk estimates obtained with the model. An important feature of the analytical approach was that an attempt was made to incorporate information on correlation between important parameters, for example, the observed correlation between total blood how and alveolar ventilation rate. Using the published PBPK model for chloroform, the best estimate of the average population risk based on the preferred pharmacodynamic dose metric (PTDEAD), representing cell death, is 9.2 x 10(-7); this estimate is more than 500-fold lower than the risk estimate of 5.3 x 10(-4) based on an alternative pharmacokinetic dose metric (AVEMMB), which represents tissue adduct formation. However, when interindividual variability was considered the range of individual risks (from the 5th to the 95th percentile of the population) predicted with PTDEAD was extremely broad (from 3.0 x 10(-13) to 3.2 x 10(-4)), while individual risks predicted with AVEMMB only varied over a factor of four (from 1.9 x 10(-4) to 7.4 x 10(-4)). As a result, the upper 95th percentile of the distribution of individual risk estimates based on the preferred cell death metric were within a factor of three of the 95th percentile for the pharmacokinetic alternative. The crucial factor with respect to the much greater variability of chloroform risk estimates based on cell death is that the dose metric, PTDEAD, is exquisitely sensitive to variation of the parameters in the model defining the response of cells to the cytotoxicity of chloroform. Unfortunately, these key parameters are also highly uncertain, as well as strongly correlated. As a result it proved impossible to accurately quantify the additional impact of parameter uncertainty on the dose metrics and risk estimates for chloroform. In general, however, the approach used in this study should be useful for differentiating the impact of interindividual variability and parameter uncertainty on PBPK-based risk assessments of other chemicals where the sensitivity, uncertainty, and correlation of the key parameters are more limited.