Appendix B Evaluation of the Molecular Connectivity Index

PROJECT SUMMARY Under a previous work assignment with NCEA CTAC and OSW, RTI evaluated the Travis and Arms (1988) methodology used to predict biotransfer factors (BTF) of chemicals into beef and milk. The methodology based on the relationship between log K ow and the chemical concentrations in beef tissue and milk. RTI investigated the basis for this relationship to evaluate the level of uncertainty associated with BTFs predicted using the Travis and Arms (1988) methodology. Despite the fact that this methodology has been the standard in risk assessment for more than a decade, our findings suggested that the BTFs were associated with considerable uncertainty (Birak et al., 2002). For example, our results for some chemicals indicated that the BTFs could vary over several orders of magnitude. Based on these findings, NCEA CTAC and OSW tasked RTI with investigating alternative approaches to calculating BTFs. Specifically, we have been investigating the use of molecular connectivity indices (MCIs) as an alternative predictor of BTFs. Of particular interest was the BTF methodology for beef and milk presented by Dowdy et al. (1996) and based on methods originally developed by Meylan et al. (1992) to predict organic carbon partition coefficients (K oc). Meylan et al. (1992) developed an approach for estimating K oc using a two-step process in which (1) nonpolar chemicals were plotted against K oc to examine the relationship between parameters (in this case, linear), and (2) correction factors were applied for polar chemicals based on the functional groups and deviations from their nonpolar regression line. Dowdy et al. used the polar correction factors from Meylan et al. to develop polar-adjusted molecular connectivity indices (1 P pc), and found a strong linear relationship between the BTFs and MCIs. To investigate this relationship further, RTI plotted the molecular connectivity indices for chemicals against BTFs, first for chemicals without polar correction factors and then for chemicals with polar correction factors as determined by Meylan et al. and Dowdy et al. Contrary to what other authors have found between K oc and 1 P, we found no relationship between the BTFs and 1 P for chemicals without polar correction factors. Using the method presented by Dowdy et al. to derive polar correction factors, we regressed the MCIs versus the BTFs and found that in this case a relationship does exist. Specifically, the R-square values for milk and beef are 0.55 and 0.54, respectively, as compared to …