In an article published on August 31, 2017 in the peer-reviewed journal Environment International, Derya Biryol and colleagues from the U.S. Environmental Protection Agency (EPA) reported on the development of a framework to predict dietary exposure for food contact substances (FCSs) in a high-throughput (HT) approach. The approach can serve as tool for prioritizing those chemicals with highest exposure estimates for subsequent detailed risk assessment. The work was carried out as part of the EPA’s ExpoCast program which aims to develop approaches for computational prediction of aggregate exposures for multiple chemicals. Aggregate exposure is defined as exposure to a single chemical via all possible exposure routes, for example through skin, diet, or inhalation.

In the presented workflow, a linear regression-based empirical model of chemical migration was combined with estimates of daily population food intakes derived from food diaries available from the National Health and Nutrition Examination Survey (NHANES) data. 1,009 chemicals were identified via publicly-available data sources as FCSs being present in synthetic polymer materials, and migration-caused concentrations of these chemicals in foods at equilibrium were then predicted for twelve food groups (derived as pairwise combinations of four food types (fatty, aqueous, acidic, and alcoholic) and three storage temperatures (0 °C, 4 °C, and 27 °C)). The estimates of the initial chemical concentration in a given food contact material (FCM) were predicted based on the functional role of chemicals in these materials, following a previously developed approach for predicting the function and weight fraction of a chemical in a product based on that chemical’s structure. Finally, population exposure through ingestion (in mg/kg bodyweight/day (mg/kg bw/day)) was estimated by combining the predicted migration values with NHANES data on daily intakes for particular food groups.

The thus-predicted population exposure estimates spanned nine orders of magnitude, with the population median intakes ranging from 0.8 ng/kg bw/d to 34 mg/kg bw/d. Plasticizers exhibited the highest dietary exposure estimates. The largest exposure was predicted for dihexyl azelate (CAS 109-31-9), a compound used as plasticizer. Among the age groups, the largest exposures were predicted for children of 0-5 years old.

Next, calibrated aggregate exposures (from FCS- and consumer product-derived exposure sources) were estimated for 1,931 chemicals. These estimates were then compared to exposures previously inferred from NHANES biomonitoring data. It was found that both pathways – i.e. from FCSs and from consumer products – were significantly predictive of exposures inferred based on NHANES data. Further, NHANES chemicals that possessed FCS exposure sources exhibited in general higher exposure levels compared to those with only a consumer product exposure pathway. Thus, in many cases the contribution from food packaging and other food contact applications appeared to dominate the overall exposure. Of note, the FCS prediction often overestimated the exposures inferred from NHANES data by up to three orders of magnitude. This finding was, according to the authors, “not unexpected given the conservative nature of the equilibrium concentration estimates, consumption factors, and FCS chemical prevalence,” implying that due to the lack of actual data on FCS migration, food consumption and the use of chemicals in specific food packaging uses, the model likely overestimates actual exposures.

When discussing their framework, the authors emphasized that the presented approach has large uncertainties and is therefore “not appropriate for assessments of single chemicals.” However, these methods “can provide critical refinement to aggregate exposure predictions used in risk-based chemical priority-setting.” They further noted that improvement of exposure predictions shall be possible once “additional data on concentrations of chemicals in different types of packaging, market penetration of individual chemicals in packaging, and improved estimates of packaging-specific food consumption” become available.


Biryol, D., et al. (2017). “High-throughput dietary exposure predictions for chemical migrants from food contact substances for use in chemical prioritization.Environment International, 108:185-194.

Isaacs, K., et al. (2016). “Characterization and prediction of chemical functions and weight fractions in consumer products.Toxicology Reports, 3:723-732.

Wambaugh, J., et al. (2014). “High throughput heuristics for prioritizing human exposure to environmental chemicals.Environmental Science & Technology, 48:12760-12767.