Closing the Organofluorine Mass Balance in Marine Mammals Using Suspect Screening and Machine Learning-Based Quantification

Abstract

High-resolution mass spectrometry (HRMS)-based suspect and nontarget screening has identified a growing number of novel per-and polyfluoroalkyl substances (PFASs) in the environment. However, without analytical standards, the fraction of overall PFAS exposure accounted for by these suspects remains ambiguous. Fortunately, recent developments in ionization efficiency (IE) prediction using machine learning offer the possibility to quantify suspects lacking analytical standards. In the present work, a gradient boosted tree-based model for predicting log IE in negative mode was trained and then validated using 33 PFAS standards. The root-mean-square errors were 0.79 (for the entire test set) and 0.29 (for the 7 PFASs in the test set) log IE units. Thereafter, the model was applied to samples of liver from pilot whales (n = 5; East Greenland) and white beaked dolphins (n = 5, West Greenland; n = 3, Sweden) which contained a significant fraction (up to 70%) of unidentified organofluorine and 35 unquantified suspect PFASs (confidence level 2−4). IE-based quantification reduced the fraction of unidentified extractable organofluorine to 0−27%, demonstrating the utility of the method for closing the fluorine mass balance in the absence of analytical standards.

Combining Advanced Analytical Methodologies to Uncover Suspect PFAS and Fluorinated Pharmaceutical Contributions to Extractable Organic Fluorine in Human Serum (Tromsø Study)

“…In the second step, the LC-Orbitrap-HRMS full scan data were screened for the suspect hits found by DI-FT-ICR-MS and for 209 additional PFAS masses. These additional masses were obtained from a list of 324 PFAS previously reported in human serum and biota samples compiled by Lauria et al 35 from which we removed 31 PFAS included in our previous target analyses 22 and 84 PFAS that were already included in the first suspect screening step. The LC-Orbitrap-HRMS full scan data were also screened for a list of 342 fluorinated pharmaceuticals ( Table S12 ), including 340 fluorinated pharmaceuticals part of the WHO ATC (Anatomical Therapeutic Chemical) classification compiled by Inoue et al 36 and 2 additional pharmaceuticals used to treat diabetes (ATC = A10B).…”

A brief review on the assessment of potential joint effects of complex mixtures of contaminants in the environment

“…109 Recently, machine learning models combining molecular and eluent descriptors showed the potential to predict IE. 110 Lauria, et al 110 used a gradient-boosted tree-based model to train the absolute response factors in negative mode for prediction of IE with the validation of 33 per- and polyfluoroalkyl substances (PFAS). This method was first applied to the liver of marine mammals (pilot whales and white-beaked dolphins) to estimate the concentrations of unidentified suspect PFAS.…”

Can Qualitative Nontarget Data Be Indicative of PFAS Contamination? First Evidence by Correlation with EOF in Environmental Samples

“…To unravel the unknown gaps often observed in PFAS mass balances while preserving structural information, NTS approaches can be utilized . Depending on the PFAS input and methods used, in different environmental media such as in contaminated soils, water, and sludge, significant unknown EOF fractions were observed. ,, NTS based on high resolution-mass spectrometry (HRMS) is very powerful to identify unknown PFAS in various samples .…”