Development and Evaluation of a Database of Dietary Bioaccumulation Test Data for Organic Chemicals in Fish

Abstract

Dietary bioaccumulation tests for fish have been conducted for about 40 years. Standardized test guidance has recently been developed. Test metrics of primary scientific and regulatory interest are the whole body depuration rate constant (kT), whole body growth corrected depuration rate constant (kTg), and corresponding chemical half-lives (t1/2 and t1/2g), dietary chemical absorption efficiency (AE), and biomagnification factor (BMF). A database of 3032 measurement end points for 477 discrete organic chemicals including 964 half-lives, 1199 AEs and 869 BMFs from 19 species (primarily trout and carp) was developed from the literature. Biological properties (e.g., organism weight, lipid content) and exposure conditions (e.g., temperature, feeding rate, dietary lipid content, exposure duration) are documented. Test chemicals range in molar mass from 120 to 1423 g·mol(-1) with log octanol-water partition coefficients (KOW) ranging from 0.8 to 14.3; 50% of the database entries are for polychlorinated biphenyls. The measured end points are derived from various protocols and sources of variability are described. The data are evaluated and categorized using proposed data quality (confidence) criteria derived from the standardized test protocol providing initial guidance for data users. Half-lives range from 0.13 to 2600 days; however, approximately 54% have an identifiable source of uncertainty. The data suggest that chemicals absorbed from the gastrointestinal tract with a log KOW ≥ ∼5 and at least as high as ∼9 have biomagnification potential in fish. A mechanistic bioaccumulation model is compared to the measured data and used to illustrate the influence of growth and biotransformation rates on the BMF.

Acceptable‐by‐design QSARs to predict the dietary biomagnification of organic chemicals in fish

“…The affinity with lipids is, in fact, a surrogate for water‐organism partitioning. As already reported (Arnot and Quinn ), the logBMF tends to increase when increasing the K OW . In the absence of significant loss processes (biotransformation or gill elimination), the biomagnification of chemicals is due to a decreased fugacity of digesta in the gastrointestinal tract (GIT), leading to an increase in the chemical fugacity in the GIT relative to the organism generating a fugacity gradient for chemical diffusion from GIT to body and, ultimately, a chemical fugacity in the predator larger than the fugacity in its prey (Gobas et al ).…”

“…One compound (PCB78, Figure ) has large prediction residuals when considering both M1 and M2 (logBMF pred – logBMF exp > 2 log units), and, consequently, also when considering the consensus. The experimental BMF value of PCB78 (logBMF = −2.52 on Gasterosteus aculeatus , Andersson et al []) was labeled as of medium reliability in the original database (Arnot and Quinn ), because the growth was not addressed in the experimental study. As the fish growth may lead to a dilution of the chemical concentration (resulting in a potentially lower measured BMF), the reported experimental BMF may be downward biased and responsible for the large prediction residuals.…”

“…A data set of fish laboratory‐based BMF values for 214 compounds was used (Grisoni et al ). The experimental data were extracted from a recently published database of aquatic endpoints (Arnot and Quinn ), which was already curated by the authors according to current OECD 305 test guidelines and contained annotations on the data reliability. In the work of Grisoni et al (), the data set was curated from the structural point of view to detect errors in the annotated structures and was statistically examined from the experimental point of view to obtain median values for each compound; compounds with only one lowly reliable annotation were discarded.…”

“…In the work of Grisoni et al (), the data set was curated from the structural point of view to detect errors in the annotated structures and was statistically examined from the experimental point of view to obtain median values for each compound; compounds with only one lowly reliable annotation were discarded. Each compound BMF value was then scored according to the reliability of the individual data points used to calculate the median (as annotated by Arnot and Quinn []), obtaining 15 (7%) very high‐, 75 (35%) high‐, 81 (38%) medium‐, and 44 (20%) low‐reliability compounds (see Supplemental Data for a summary of the analysis steps). Because every measurement is affected by experimental error, using multiple sources of information reduces the uncertainty compared to using single measurements.…”

“…The proposed QSAR approach was developed on a data set of laboratory‐based fish BMF of 214 chemicals deriving from a recently published database of aquatic endpoints collected and curated for quality according to OECD 305 test guidelines (Arnot and Quinn ). Attention was paid to the selection of simple modeling approaches and to the interpretation of molecular descriptors.…”