Wikoff D, Goodrum P, Haws L, Budinsky R. Application of quantitative approaches to assess uncertainties in the development of toxicity values: A case study involving the reference dose (RfD) for 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). Society of Toxicology Annual Meeting. March 11–15, 2018. San Antonio, TX.
Recent efforts to improve risk assessments include recommendations from The National Academy of Sciences (NAS) to incorporate quantitative uncertainty analysis to better inform risk management decisions. However, there are relatively few examples where such has been applied in development of toxicity values. Using the RfD for TCDD as an example, we investigated the impact of uncertainty related to selection of critical studies, points of departure (PODs) from key datasets, and application of uncertainty factors (UFs) on the RfD. The current TCDD RfD is based on equivalent PODs (20 pg/kg-day) for sperm concentration in males and neonatal TSH screening levels from two human studies, with a 30x UF applied to arrive at an RfD=0.7 pg/kg-day. Significant limitations in internal validity (e.g., high risk of bias in key domains including exposure and outcome assessment) of each of the critical studies were identified but could not be evaluated quantitatively (though the potential impact is described qualitatively). With respect to identification of a POD, multiple aspects of uncertainty were assessed quantitatively: addition of TEQ, kinetic model uncertainties (estimation of daily intake associated with serum concentration in children), clinically relevant thresholds (e.g., TSH screening levels of 10 µU/ml) and designations of such as no- or low-effect levels, and dose-response information for neonatal TSH and risk of congenital hypothyroidism from multiple study populations was combined using Bayesian statistics to develop PODs for specified risk levels. Due to lack of statistical significance and/or clinical significance, alternative UFs were also applied (e.g., elimination of 10x for LOAEL to NOAEL). This assessment, including consideration of study quality, dose-response, magnitude, and clinical relevance, supports RfD values ranging from 70 pg/kg-day. This range of plausible RfD values informs risk management decisions and highlights key uncertainties, as well as demonstrates the utility of evidence-based approaches in deriving health-based guidance values.