Introduction to applying dose-response modeling from epidemiological data to derive toxicity values

Title: Introduction to applying dose-response modeling from epidemiological data to derive toxicity values Background and Aim: EPA derives toxicity values (e.g., cancer unit risks and non-cancer reference concentrations) based on exposure-response functions. EPA rarely has individual-level epidemiology data to model and therefore uses slopes extracted from published studies. Methods: EPA selects a benchmark response (BMR) based on biological and statistical considerations to identify points-of-departure (PODs) from the exposure-response functions.  For cancer, the POD is often the exposure concentration associated with 1% extra risk of cancer (the BMR) and is identified using the upper-bound of published exposure-response functions (e.g., hazard ratio) and vital statistics data to account for competing causes of death.  For an inhaled exposure, the toxicity value is the inhalation unit risk derived as the BMR/POD when the expectation that extra cancer risk is linear at low doses is appropriate.  Noncancer effects are assumed to occur above a threshold level of exposure.  For non-cancer effects, the POD can be estimated as the lower bound of the benchmark dose (i.e., the BMDL) using the upper-bound on the exposure-response functions for a specific BMR. A reference concentration (RfC) is derived as the POD divided by the product of uncertainty factors. Results: Challenges may arise over questions of the best-fitting shape of exposure-response functions and unintended consequences regarding low-dose extrapolations of log-transformed exposures.  Many published exposure-response functions are only assessed for fit at the mean exposure while toxicity values are derived at the low end of the distribution.  Sometimes publications do not provide information on the low end of the exposure distribution to allow EPA to see where estimated PODs may fall within the distribution. Conclusions: A better general understanding of how EPA applies exposure-response functions from the epidemiological literature may allow investigators to anticipate some challenges EPA can face when deriving toxicity values, and thus make those publications more impactful for assessing health risks. Keywords: Epidemiology, risk assessment, toxicity values, methods

Impact/Purpose

This presentation will be part of a symposium at the annual conference of the International Society for Environmental Epidemiology. Topic Overview:  This session will be an introduction to how international agencies perform dose-response analysis and derive toxicity values based on epidemiological data, with the goal of helping epidemiologists understand the downstream applications of their research. The session will begin with an overview of the challenges that arise when using epidemiological data in dose-response modeling (Tom Bateson, EPA). Then, a series of presentations will provide examples where academic scientists and international agencies have conducted such modeling with epidemiological data, including promising advancements for the field in the areas of benchmark dose (BMD) modeling and meta-regression. Drs. Budtz-Jorgensen, Grandjean, and Wheeler will discuss the use of epidemiologic data in benchmark (BMD) modeling and approaches to improve evaluation of and control for confounding. Dr. Larsen will discuss challenges and advancements with analyzing heterogenous epidemiologic data and proposed solutions in the context of meta-regression to evaluate health effects of inorganic arsenic. The final presentation, from Dr. Vermeulen, will present applications of Bayesian meta-regression to multiple data streams (human, animal, in vitro).  Attendees will gain an improved understanding of the challenges and innovations with applying epidemiologic data in dose-response modeling to inform public health policies.

Citation

Bateson, T. Introduction to applying dose-response modeling from epidemiological data to derive toxicity values. International Society for Environmental Epidemiology, Athens, GREECE, September 18 - 21, 2022.