Hypothesis-driven weight of evidence framework for evaluating data within the US EPA's Endocrine Disruptor Screening Program

“Weight of Evidence” (WoE) approaches are often used to critically examine, prioritize, and integrate results from different types of studies to reach general conclusions. For assessing hormonally active agents, WoE evaluations are necessary to assess screening assays that identify potential interactions with components of the endocrine system, long-term reproductive and developmental toxicity tests that define adverse effects, mode of action studies aimed at identifying toxicological pathways underlying adverse effects, and toxicity, exposure and pharmacokinetic data to characterize potential risks. We describe a hypothesis-driven WoE approach for hormonally active agents and illustrate the approach by constructing hypotheses for testing the premise that a substance interacts as an agonist or antagonist with components of estrogen, androgen, or thyroid pathways or with components of the aromatase or steroidogenic enzyme systems for evaluating data within the US EPA's Endocrine Disruptor Screening Program. Published recommendations are used to evaluate data validity for testing each hypothesis and quantitative weightings are proposed to reflect two data parameters. Relevance weightings should be derived for each endpoint to reflect the degree to which it probes each specific hypothesis. Response weightings should be derived based on assay results from the test substance compared to the range of responses produced in the assay by the appropriate prototype hormone and positive and negative controls. Overall WoE scores should be derived based on response and relevance weightings and a WoE narrative developed to clearly describe the final determinations.

► A framework was developed for weight of evidence evaluations of endocrine activity. ► The framework is tailored to the US EPA's Endocrine Disruptor Screening Program. ► The framework is hypothesis-driven and seeks methodological transparency. ► Data validity and relevance are evaluated according to published literature. ► Procedures are outlined for assigning quantitative or rank ordered weights to data.