Approaching a 0% False Positive Rate for PFAS Determination Leveraging Only MS1 Data

Liquid chromatography high-resolution tandem mass spectrometry
(LC-HRMS/MS) is commonly used for the analysis of per- and polyfluoroalkyl
substances (PFAS). Targeted approaches with LC-HRMS/MS often cover
less than 30% of PFAS across various matrices and hence nontargeted
strategies are necessary to enhance identification coverage. We expanded
FluoroMatch Suite, a nontargeted PFAS data-processing software, to
leverage full-scan (MS1) data for highly accurate formula prediction
and Kaufmann analysis. Software features include Kaufmann analysis
with isoline cutoffs determined using kernel density based on an EPA
PFAS data set and an 11-step formula prediction algorithm. Application
of the FluoroMatch Suite with the MS1 extension to AFFF contaminated
soil revealed 179 PFAS-confirmed features. Kauffman 95% isoline cutoffs
captured 94% of the confirmed PFAS and removed 96% of features assigned
as likely non-PFAS. The PFAS-formula prediction introduced in this
manuscript had a false positive rate of 26% and a false negative rate
(no predicted formula) of 30%. Using a novel homologous series voting
algorithm, where the predominant subclass from formula prediction
were used to predict all formulas for the homologous series, we achieved
a 0% false positive rate and 6% false negative rate in formula prediction.
The novel nontargeted algorithms developed in this study proved to
be highly accurate and by leveraging MS1 data enhances the capacity
to identify unknown PFAS in complex environmental matrices.