Predicting bioconcentration factors (BCFs) for per- and polyfluoroalkyl substances (PFAS)

• Published in: Chemosphere (Oxford) 2024-09, Vol.364, p.143146, Article 143146
• Main Authors: Kowalska, Dominika, Sosnowska, Anita, Zdybel, Szymon, Stepnik, Maciej, Puzyn, Tomasz
• Format: Article
• Language: English
• Subjects: BCF; Bioconcentration; In silico; MLR; Perfluoroalkyl compounds; PFAS; QSPR model
• ISSN: 0045-6535; 1879-1298; 1879-1298
• DOI: 10.1016/j.chemosphere.2024.143146

The bioconcentration factor (BCF) is an important parameter that gives information regarding the ability of a contaminant to be taken up by organisms from the water. Per- and polyfluoroalkyl substances (PFAS) are widespread in the environment, causing concern regarding their impact on human health. Due to the lack of available bioaccumulation data for most compounds in the PFAS group, we developed a quantitative structure-property relationship (QSPR) model to predict the log BCF for fish (taxonomic class Teleostei), based on experimental data available for the most studied 33 representatives of this group of compounds. Furthermore, we implemented the developed model to predict log BCF for an external dataset of 2209 PFAS. Consequently, 1045 PFAS were found not to be bioaccumulative, 208 were classified as bioaccumulative, and 956 were predicted to be very bioaccumulative. Finally, we obtained the high correlation (R2 = 0.844) between the log BCFs obtained in laboratory and field studies for 13 PFAS. In silico analyses indicate that PFAS bioconcentration depends on the size (chain length – number of CF2 groups in alkyl tail/chain) of a molecule, as well as on the atomic distribution properties. In general, long-chain PFAS - above 8 and 6 carbon atoms for perfluorinated carboxylic acids (PFCAs)and perfluorinated sulfonic acids (PFSAs), respectively - tend to bioconcentrate more compared to the short-chain ones. In conclusion, predicting BCF on fish is possible for a wide range of fluorinated compounds, which can be further used for estimating PFAS behavior in the environment. [Display omitted] •Novel approach of examining PFAS bioconcentration factor in Teleostei taxonomic class.•Multilinear regression model (MLR) with remarkable predictive abilities (R2 = 0.92; Q2 = 0.88).•Bioaccumulative classes determined for 2522 PFAS from the NORMAN database.•1045 PFAS are not Bioaccumulative (nB) and 1164 PFAS are B (Bioaccumulative) or vB (very Bioaccumulative).•Predicted log BCF could be used to estimate the log BAF.