Improved multivariate quantification of plastic particles in human blood using non-targeted pyrolysis GC-MS

Abstract

Accurate analytical methods are crucial to assess human exposure to micro- and nanoplastics (MNPs). Quantitative pyrolysis-gas chromatography coupled with mass spectrometry (Py-GC-MS) has recently been used for quantifying MNPs in human blood. However, pyrolysis introduces complex effects such as secondary reactions between matrix compounds and polymers. This work introduces a non-targeted and multivariate approach to improve the identification and quantification of polyethylene (PE), poly(vinyl chloride) (PVC) and polyethylene terephthalate (PET). After spiking of extracted blood samples, PARADISe was used for componentization and integration of 417 features detected with Py-GC-MS. Quantification based on multivariate calibration models demonstrated a superior performance when compared to univariate regression. Feature selection approaches were used to identify optimal feature subsets, which reduced quantification errors by 30 % for PE, 10 % for PVC and 38 % for PET. In addition, chemical insight into pyrolysis processes was obtained by studying the matrix effects (MEs) of blood. The pyrolysis of PE and PVC appeared to be minimally affected (MEs = 81-154 %), while PET exhibited complex interactions with the matrix (MEs = 40-9031 %), impacting its quantification accuracy. In conclusion, this research highlights the importance of accounting for secondary effects during pyrolysis and introduces a multivariate approach for more accurate and robust quantification of MNPs in blood.

Keywords: Human whole blood; Machine learning; Matrix effects; Micro- and nanoplastics; Multivariate calibration; Pyrolysis-GC-MS.