Leveraging the MS1 Dimension and Formula Prediction in Non-Targeted Analysis of PFAS using New FluoroMatch Algorithms: Assessing Confidence and Coverage

Importance of the Topic

Per- and polyfluoroalkyl substances (PFAS) are ubiquitous environmental contaminants with complex chemistries and low detection levels in biological matrices. Accurate identification and annotation of diverse PFAS structures is critical for exposure assessment, regulatory compliance, and understanding potential health impacts. Integrating MS1 spectral data with advanced formula prediction enhances non-targeted workflows, improving confidence in compound discovery and expanding coverage beyond targeted screening.

Objectives and Study Overview

This work presents an evaluation of FluoroMatch Flow, a comprehensive software suite that automates non-targeted PFAS analysis. The primary objectives are:

• To implement MS1-based formula prediction alongside traditional spectral evidence.
• To integrate homologous series detection and class-specific rules for PFAS classification.
• To benchmark coverage and annotation confidence against targeted QqQ screening in dried blood spot samples.

The study assesses discovery rates, formula assignment accuracy, and the impact of blank filtering on feature retention.

Methodology and Instrumentation

The analytical workflow comprised the following steps:

• Sample Preparation: Extraction of PFAS from dried blood spot cards (paper matrix).
• Data Acquisition: LC-HRMS and MS/MS data were acquired using an Agilent 6546 LC/Q-TOF instrument. Data-dependent acquisition (DDA) or internal-standard-assisted DDA (IE-DDA) modes captured MS1 and MS/MS spectra.
• Preprocessing: Peak picking performed with Agilent Profinder and MZMine 4, followed by blank filtration and alignment using MassProfiler Professional.
• FluoroMatch Flow Processing: Automated file conversion, feature detection, annotation, confidence scoring, homologous series grouping, formula prediction, and interactive visualization.

Main Results and Discussion

Application of FluoroMatch Flow to dried blood spots yielded 112 PFAS features before blank filtering, which reduced to 18 after removing background signals. Key findings include:

• Unique Discoveries: FluoroMatch identified 13 PFAS not detected by targeted QqQ screening.
• Formula Prediction Performance: Of 30 representative PFAS, 34% achieved correct top-rank formula assignments (score A), and 66% appeared within the top 10 candidates.
• Homologous Series Approach: Four homologous PFAS series were predicted with 100% accuracy by leveraging repeating CF2 units and common structural motifs.

Formula prediction integrates mass defect filtering, isotopic pattern fidelity, atomic constraints (C, H, O, N, S, F, Br, Cl, P), nitrogen rule, and senior rule validation. Candidate formulas are ranked by m/z and intensity similarity, supplemented by MS/MS evidence when available.

Benefits and Practical Applications

This workflow offers:

• Automated End-to-End Processing: Streamlines non-targeted PFAS analysis from raw data to structured reports.
• Enhanced Confidence: Multi-layered scoring and homologous series grouping reduce false positives and improve annotation reliability.
• Interactive Visualizations: Cross-filterable plots (Kendrick mass defect, retention time versus m/z, isotopic distributions, fragmentation tables) facilitate data exploration and collaborative review via web links.

Future Trends and Opportunities

The integration of MS1-centric algorithms and formula prediction is expected to evolve with:

• Machine Learning: Automated recognition of novel PFAS classes and improved scoring models.
• Expanded Libraries: Community-driven spectral and homologous series databases to support emerging fluorinated compounds.
• Cloud-Based Platforms: Real-time collaboration and scalable processing for high-throughput environmental and clinical studies.

Conclusion

FluoroMatch Flow demonstrates a powerful, automated solution for non-targeted PFAS profiling by leveraging MS1 formula prediction, homologous series detection, and interactive visualization. The approach enhances annotation coverage, identifies compounds missed by targeted methods, and supports high-confidence discovery in complex matrices.

References

• Lin EZ, Koelmel J, Bowden J, et al. FluoroMatch: A Software Suite for Non-Targeted PFAS Annotation and Visualization. Anal Bioanal Chem. 2021; DOI:10.1007/s00216-021-03392-7.
• Ashoori S, et al. Senior Rule and Isotopic Filtering in Formula Prediction. BMC Bioinformatics. 2008; DOI:10.1186/1471-2105-8-105.