Rationale: Probabilistic techniques and PBPK modeling were employed in the derivation of site-specific remedial objectives (SSROs) for soils at the Former Domtar Edmonton Wood Processing Plant in Edmonton, Alberta. Probabilistic methods were used to explicitly address variability and uncertainty about inputs to risk calculations. PBPK modeling was used to compare estimated exposures at the SSRO with levels reported in the literature to be associated with adverse effects.

Methods: Using Crystal Ball software, distributions were incorporated into the absorption, toxicity and exposure factors. Exposure by a toddler to PCDD/Fs in soil was assumed via direct contact and home-grown produce. PBPK modeling was used to convert the ingested dose to an internal dose. Using the Carrier-Aylward model, modeling was performed for 30 years, including intake from breastfeeding and a toddler pica event, assuming the residential toddler SSRO. The effect of seasonal amortization was also evaluated.

Results & Discussion: Based on probabilistic analysis a SSRO of 883 ppt corresponds to a 95% probability of a hazard quotient of 1 or less. Deterministic analysis with default assumptions resulted in a SSRO of 645 ppt, corresponding to the approximate 98th percentile. PBPK modeled lipid-adjusted serum concentrations (LASC) are compared with published values associated with adverse effects. One value lies within the range; all other NOAEL/LOAEL LASC values are above the range of modeled values.

Conclusions and Implications: The probabilistic assessment characterizes the level of confidence that SSROs are health protective in the face of inherent variabilities and uncertainties. PBPK results indicate that SSROs yield blood levels below levels causing adverse effects in humans. Incorporation of a pica event did not elicit substantial changes in PCDD/F blood levels. Moreover, the background dose an infant receives from breastfeeding is substantially larger than the dose passed to an infant by an exposed pregnant female.