Pham LL, Shefﬁeld T, Pradeep P, Brown J, Haggard D, Wambaugh J, Judson R, Friedman KP. 2019. Estimating uncertainty in the context of new approach methodologies for potential use in chemical safety evaluation. Curr Opin Toxicol (in press).
Acceptance of new approach methodologies (NAMs) for use in characterizing chemical hazard and risk requires informed expectations regarding the minimum precision and maximum accuracy of their results. Uncertainty in NAMs derived from variability in the traditional reference data used to train or validate performance of the NAM, and uncertainty in the modeling procedures themselves, limit NAM performance. Herein we review current approaches to characterizing uncertainty in NAMs. We discuss variability in in vivo data used as a reference for NAM development and validation; the quantitative uncertainty in concentration-response modeling for high-throughput in vitrobioactivity screening; the uncertainties associated with in vitro to in vivoextrapolation using toxicokinetic information; and the quantitative uncertainty in the experimental inputs and modeled outputs from QSAR models for prediction of point of departure doses. Communication of the amount of uncertainty, both in the input and output for NAMs, often involves a confidence or prediction interval around a given potency estimate, derived from an understanding of the variability in the data modeled. Tuning expectations of NAM performance to an understanding of the reproducibility and variability, both of traditional approaches and NAM approaches, provides a path for the adoption of NAMs as alternatives in screening chemicals for risk.