Society of Toxicology Annual Meeting 2017 & Expo
Baltimore, MD, USA
March 12, 2017 – March 16, 2017

Comparison and Analysis of Toxcast Data with In Vivo Data for Food-Relevant Compounds Using The Risk21 Approach
Presenting Author: Alexandra E Turley
14 March 2017
9:30 AM – 12:15 PM

Abstract Number: 1647
Platform Session Room Number: CC Room 310

Abstract: The ToxCast program has generated a wealth of in vitro high throughput screening (HTS) data for a large number of compounds, providing a unique resource of information on chemical bioactivity.  With analyses of these data ongoing, best approaches for the interpretation and use of the ToxCast data, including for safety assessment of food-related compounds, remain to be evaluated.  To fill this gap, we present case studies of two food-related compounds. The ToxCast and in vivo toxicity data were compared using the RISK21 approach to evaluate their potential use in chemical safety assessment. RISK21 is an exposure-driven flexible risk assessment framework that allows for the simultaneous evaluation of exposure and toxicity data. Prior work examined all food-relevant compounds in ToxCast and showed that some food contact substances have higher bioactivity in ToxCast assays relative to direct food additives. Two food contact substances, sodium pyrithione and dibutyltin dichloride, were selected for this study. Available exposure data, toxicity data, and model predictions were compiled and analyzed for both compounds. In-vitro to in-vivo extrapolation (IVIVE) was performed to determine oral equivalent doses for the ToxCast data.  For sodium pyrithione, concentrations that elicited bioactivity in ToxCast assays corresponded to low- and no- observed adverse effect doses in in vivo animal studies. For dibutyltin dichloride, the concentrations that elicited bioactivity in ToxCast were below the doses that demonstrated toxicity in animals; however, this was confounded by the conservative pharmacokinetic parameter estimates used in the IVIVE calculations due to a lack of pharmacokinetic data. For both compounds, no specific mode of action was identifiable from the ToxCast data, and the exposure data needed for risk assessment were limited. These studies highlighted the potential promise of the ToxCast approach while also demonstrating gaps in knowledge surrounding the ToxCast data. Key gaps in using the ToxCast data included the lack of coverage of many biological targets, and the lack of pharmacokinetic data.

Identification, Curation, and Prioritization of Food-Use Chemicals in ToxCast
Presenting Author: Agnes Karmaus
15 March 2017

Abstract Number/Poster Board number: 2559/P404
Session Title: Food Safety and Nutrition
CC Exhibit Hall or Hall A

Abstract: Evaluating chemicals that are directly added to or come in contact with food poses a challenge due to the time, cost, and data necessary to thoroughly conduct comprehensive toxicological testing. The diversity among food-use chemicals, and lack of hazard and exposure data, impede curation and prioritization of these chemicals. This study compiled a list of food-use chemicals in the United States (U.S.) and demonstrates approaches amenable to the evaluation of this large and diverse chemical inventory. 8,659 unique food-use chemicals were compiled from 8 public sources and mined against the ToxCast in vitro high-throughput screening inventory to identify 1,530 food-use chemicals with in vitro assay data. Each of these chemicals was then manually evaluated for current registration status and categorized based on exposure likelihood from food in the U.S. into four categories: direct food additive, indirect additive, pesticide/residue, or non-food. 319 chemicals were categorized as non-food and removed from the list, leaving 556 direct additives, 339 indirect additives, and 406 pesticides/residues. The cytotoxicity elicited by the curated list of food-use chemicals in ToxCast revealed that only 10% of direct additives elicited cytotoxicity, while 24% of indirect additives and 41% of pesticides/residues were cytotoxic. To address the need to prioritize chemical mixtures, we used frequent itemset mining (FIM) to identify which individual chemicals or combination of chemicals appear with the greatest frequency in the U.S. Food and Drug Administration’s Effective Food Contact Substances (EFCS) database. The EFCS database comprises 978 registrations, containing 715 unique chemicals; in total, 189 of the registrations contain mixtures of ≥2 chemicals and 110 registrations contain≥3 chemicals, up to a maximum of one registration containing 9 chemicals. The FIM approach identified acetic acid, hydrogen peroxide, and peracetic acid as the most abundant co-occurring chemicals in EFCS registrations, each occurring in 42 (or 4.3%) of 978 registrations. The combination of peracetic acid and hydrogen peroxide occurred most frequently, appearing in 40 (or 21%) of the 189 registered mixtures. The current inventory and analysis of ToxCast cytotoxicity and EFCS mixture prioritization represent the first evaluation of food-use chemicals and provides insight into this overlooked but critical chemical inventory. This work does not reflect EPA policy.