Prioritization Strategies in Non-Target Screening of Environmental Samples by Chromatography with High-Resolution Mass Spectrometry

Non-target screening using chromatography coupled to high-resolution mass spectrometry is a powerful tool in environmental analysis for detecting chemicals of emerging concern. The challenge lies in the large number of analytical features generated per sample, which requires prioritization to focus resources on the most relevant features. We present an overview of seven prioritization strategies, as shown in Figure 1: (1) target and suspect screening, (2) data quality filtering, (3) chemistry-driven prioritization, (4) process-driven prioritization, (5) effect-directed prioritization, (6) prediction-based prioritization, and (7) pixel- or tile-based chromatographic analyses. This article discusses how integrating these strategies can enhance compound identification, support environmental risk assessment, and accelerate decision-making.

As the anthropogenic environmental chemical space expands as a result of industrial activity and an increase in the diversity of consumer products, non-target screening (NTS) using chromatography–high-resolution mass spectrometry (HRMS) has become essential in environmental monitoring. However, the sheer number of detected features (mass-to-charge ratio [m/z], t_r pairs), often thousands for each sample, creates a bottleneck at the identification stage. Without an effective prioritization strategy, valuable time and resources are spent on irrelevant or redundant data.

This short article brings together the prioritization approaches that we presented in our recently published paper.¹ The central message is that usually no single strategy is sufficient. Instead, combining strategies allows identification efforts to be focused where they matter most—enabling faster and improved assessments of chemical risk.

Seven Prioritization Strategies

1. Target and Suspect Screening (P1):

This strategy is based on predefined databases of known or suspected contaminants, such as those from PubChemLite, the CompTox Dashboard, or the NORMAN Suspect List Exchange. It narrows candidates early by matching features to compounds of known environmental relevance using m/z, isotope patterns, retention times, and MS/MS spectra. Although able to reduce complexity, this approach is constrained by the completeness of databases and their quality.

2. Data Quality Filtering (P2):

Reliability-driven filtering removes artifacts and unreliable signals based on occurrence in blanks, replicate consistency, peak shape, or instrument drift. Though not sufficient for prioritization on its own, P2 is foundational. Without reliable data, identification becomes hypothetical. This step reduces false positives and improves accuracy and reproducibility.

3. Chemistry-Driven Prioritization (P3):

This strategy focuses on compound-specific properties useful for finding certain compound classes of interest. Mass defect filtering identifies halogenated compounds such as per- and polyfluoroalkyl substances (PFAS). Homologue series detection, isotope patterns, and diagnostic MS/MS fragments help detect transformation products or homologues series. Multidimensional chromatography or ion mobility separation can be used here to increase specificity.

4. Process-Driven Prioritization (P4):

Prioritization is guided by spatial, temporal, or technical processes. Comparing influent and effluent samples from treatment plants, or upstream versus downstream samples from river systems, highlights persistent or newly formed compounds. Correlation-based approaches link chemical signals to events such as rainfall, or operational changes, whereas source-supported prioritization starts with known source signatures (plant toxins) and traces them into receiving environments (a river downstream from a field).

5. Effect-Directed Prioritization (P5):

Effect-directed analysis (EDA) integrates biological response data with chemical compositional data (chemical fingerprints). In traditional EDA, bioactive fractions are isolated and chemically analyzed. Virtual EDA (vEDA) links features to endpoints using statistical models (for example, partial least squares discriminant analysis) across multiple samples. This strategy directly targets bioactive contaminants and is useful when regulatory action depends on effect data.

6. Prediction-Based Prioritization (P6):

Combining predicted concentrations and toxicities allows calculation of risk quotients (PEC/PNEC, Predicted Environmental Concentration versus Predicted No Effect Concentration). Models such as MS2Quant predict concentrations directly from MS/MS spectra; MS2Tox estimates LC50 from fragment patterns. These tools are particularly useful when full identification is incomplete, but prioritization for risk is needed. The key advantage is focusing on substances of highest concern without full structural elucidation.

7. Pixel- and Tile-Based Approaches (P7):

For complex data sets (especially two-dimensional [2D] chromatography), feature-based analysis can be impractical. Pixel-based (two-dimensional gas chromatography [GC×GC], two-dimensional liquid chromatography LC×LC) and tile-based (rectangular RT´ RT windows) prioritization localizes regions of high variance or diagnostic power before peak detection. These strategies highlight chemically relevant regions for more elaborate analysis (peak curve resolution and compound detection) and are especially valuable in early-stage exploration or for large-scale monitoring.

Integrated Prioritization Strategies

Each strategy serves a different purpose in the NTS prioritization workflow (for example, P1 filters knowns, P5 links to biological effects, and P6 ranks features by predicted risk). When combined, they enable stepwise reduction from thousands of features to a focused shortlist. Figure 1 groups these strategies into four domains—chemical, toxicological, external, and preprocessing—each addressing a specific aspect of feature reduction.

For example, P1 may initially flag 300 suspects. P2 and P3 reduce this to 100 by removing low-quality and chemically irrelevant features. P4 identifies 20 linked to poor removal in a treatment plant. P5 finds 10 of these features in a toxic fraction, and P6 prioritizes five based on predicted risk. P7 can be used early to focus data analysis on informative chromatographic regions. This cumulative filtering, using a few strategies in combination, narrows down a complex data set of features to a manageable number of compounds worth investigating further. It enables laboratories and regulators to focus on what matters: identification of high-risk chemicals of emerging concern, assessment of exposure and hazard, and ultimately guiding environmental policy and remediation.

Conclusion

Prioritization is central to efficient NTS workflows. It prevents resources from being spent on uninformative signals and directs attention toward features most likely to represent relevant contaminants. This article has outlined seven complementary strategies that operate at different levels of the NTS process.

Combining approaches based on chemical structure, data quality, biological response, study design, and predictive modeling can accelerate identification and strengthen environmental risk assessment. The next step is to integrate these tools into reproducible, transparent, and scalable workflows, moving NTS from exploratory screening toward actionable regulatory support.

Reference

1. Zweigle, J.; Tisler, S.; Bevilacqua, M.; et al. Prioritization Strategies for Non-Target Screening in Environmental Samples by Chromatography – High-Resolution Mass Spectrometry: A Tutorial. J Chrom A. 2025, 1751, 465944. DOI: 10.1016/j.chroma.2025.465944