Modern Nontargeted Screening (NTS) Methods for Environmental Analysis
LCGC Europe spoke to Selina Tisler and Jan H. Christensen from the University of Copenhagen in Denmark, about recent research projects focusing on nontargeted screening (NTS) approaches for important environmental monitoring applications, including the analysis of chemical leaching from plastic bottles using liquid chromatography tandem mass spectrometry (LC–MS/MS) and sediment analysis using comprehensive two-dimensional gas chromatography–high-resolution MS (GC×GC–HRMS).
Q. Your team recently investigated the leaching of chemicals from reusable plastic bottles into the drinking substances they hold (1). Why is this an issue and what are the possible health consequences for consumers?
Selina Tisler: Clean water is obviously important. There are high standards in place for drinking water to be as clean as possible. The same standard should be applied to the bottles used to store it. What is the point of having clean water if we add chemicals later? We do not currently understand the full extent of the health consequences for consumers of any leaching that may occur. We identified some chemicals that are known to have adverse effects to humans. With a nontargeted screening approach, we were not quantifying the concentrations, but this research is ongoing. It is not known if the concentration of these chemicals was high enough to cause any effects.
Q. What was novel in your approach for the analysis of this issue?
Jan Christensen: It was the first time that reusable PE plastic bottles were investigated using a nontargeted screening approach. Initially it was the strange smell of the bottles that inspired the research. Nobody had investigated it before, and yet the bottles are handed out every year to Danish children attending DBU (Danish Soccer Union) Football Schools—and they are also hugely popular in other sports. From the start we were not focused on specific compounds. In the production of plastic bottles, there are so many different compounds used, so-called intentionally added compounds, that target approaches cannot identify all of them. Furthermore, in the production process or afterwards, compounds can be transformed to other compounds. These non-intentionally added compounds are difficult to predict and analyze, and can only be tackled by nontarget screening approaches.
Q. What is novel about the nontargeted liquid chromatography tandem mass spectrometry (LC–MS/MS) method you have developed?
JC: The nontarget screening approach in general was not novel, but the sample preparation, the type of compounds investigated, and our data science workflow were. We investigated the migration of pure tap water, without the use of solvents or acids for an accelerated migration with higher concentrations of leaching compounds, which is typical in these kinds of studies. We up-concentrated the water 1000 times using a multi-layer solid-phase extraction (SPE) procedure after 24 h of storage in the plastic bottles. The main challenge was the filtering for blank compounds, which in this case meant compounds that are detected in the bottles by a nontargeted screening approach, but they originated from the water itself, the LC–MS/MS instrument, or the laboratory. This is a common challenge when you have low concentrations of the compounds of interest and increase the sensitivity of the analysis by up-concentration. Both your compounds of interest and also the blank compounds will be up-concentrated. We overcame this challenge by analyzing several control samples that followed the same process. These control samples consisted of the same water as in the bottles but stored in specially cleaned laboratory glass bottles. If compounds were also detected in these control bottles, they were not considered for further investigation.
Q. Do you have any general advice for analysts who are developing nontargeted LC–MS/MS methods?
ST: lt is important to always include enough controls and do “self-tests”, such as spiking your sample with target compounds to see if you can detect all your target compounds by your nontarget screening approach to avoid false-negative results. It is even more important to include enough control samples to avoid false-positive reported compounds.
The analysis of replicates is also essential. Make sure that you see the compound in three out of three analyses, to avoid reporting artefacts (false-positive compounds). We developed a workflow that describes the steps we consider to be important to produce reliable data with nontarget screening (2)
Q. What were the main findings of this research?
ST: The method was very effective and provided an overview of the compounds that were migrating from the plastic, which amounted to more than 400, and from the dishwasher soap, which was more than several thousand.The identification of the individual chemicals is a time-consuming process, therefore we only focused on the highest signals. We identified 45 chemicals, consisting mostly of plasticizers, antioxidants, slip agents, or photo initiators. Our results indicated that after cleaning with the dishwasher soap, the dominant group leaching out of the bottles were surfactants.
Q. One of the interesting findings of this study was the enhanced leaching of compounds from the bottles following a cycle in a dishwasher. Why do you think this occurs and what are the repercussions of this?
JC: Ingredients from dishwasher soap seem to stick or adsorb to plastic more than to glass. Even after additional flushing of the bottles with tap water, the surfactants, especially the more nonpolar ones, were adsorbing to the bottle surfaces.
Q. You also recently published a paper discussing the correction of matrix effects for nontargeted LC−electrospray ionization (ESI)−MS analysis of wastewater (3). Why are matrix effects a challenge in these types of studies?
ST: As a result of the competition for charge in ESI, in samples with a high matrix complexity, the ionization efficiency can be strongly influenced. We can see that in, for example, effluent wastewater samples that have been enriched 50 times, the signal intensity can reduce to 1/5 compared to the signal in pure solvent. This would lead to a large bias in the quantification. The method development in nontarget screening is built to cover a broad range of chemicals because the matrix effect for individual compounds are unknown. In a target screening approach you have internal standards to correct for it.
Therefore, the challenge was to compare peak intensities between different sample types, such as influent and effluent wastewater samples. If you want to evaluate a process such as degradation of micropollutants in wastewater treatment plants, it would be biased if matrix effects were not taken into consideration because the matrix suppression effects can be very different in different sample types.
Furthermore, correcting for the matrix effect is crucial to obtaining a reliable concentration estimate, especially during semi-quantification, which is a complex approach to quantifying unknown chemicals in nontarget screening.
Q. The study describes a three-step method to evaluate and compensate for matrix effects. What are these steps and how effective are they?
ST: First, for individual compounds, the matrix effect can be evaluated by a dilution series. Without matrix effects, the dilution factor and response of the compound would generally show a linear behaviour. The deviation from linearity at your desired dilution factor indicates the size of the matrix effect. It is the easiest way to compensate for your matrix effect. Using a dilution factor in the linear range (no matrix effect) but with low signals can then fall below the detection limit because of the dilution.
Second, we showed in the study that the total ion chromatogram (TIC) gives an indication about the matrix effect in the respective chromatographic area. The higher the TIC signal, the more compounds elute at the same time, the more compounds compete for ionization, and the matrix effect increases. Using the TIC profile for correcting the matrix effect could compensate for around 85% of the matrix effect in our analysis of wastewater extracts.
Third, in case the samples with higher matrix effect needs a better matrix effect correction than 85%, a quantitative structure-property relationship (QSPR) model can be used. We showed that the compound specific matrix effect can be partly predicted by the characteristic structures of the compounds.
Q. Your team also worked on a comprehensive two-dimensional gas chromatography–high-resolution MS (GC×GC–HRMS) method for the nontargeted screening of water sediments (4). Why is the analysis of sediments important and what can they tell us about the environment they are taken from?
JC: The analysis of sediments is important because sediments are sinks for chemicals in the environment, especially hydrophobic micropollutants. Using nontargeted screening (NTS) to measure micropollutants in sediments can tell us about the sources of pollutants. In the study from 2020, we identified organic micropollutants of high abundance and relevance in the urban sediments. Using multivariate statistics, we were able to isolate some distinct sources of chemicals as a natural input, namely a high relative abundance of mono-, di-, and sesquiterpenes, and a weathered oil fingerprint, namely alkanes, naphthenes, and alkylated polycyclic aromatic hydrocarbons. A dilution effect of the weathered oil fingerprint was observed in lake samples that were close to a channel. Several benzothiazole-like structures were identified in lake samples close to a high-traffic road, which could indicate a significant input from asphalt or tyre wear particles.
Q. How effective was the developed method and does it have any limitations?
JC: The method was able to identify the main sources but many chemicals remain unidentified, and due to the relatively low number of samples we were not able to identify more distinct sources in the sediment or to model the distribution of the micropollutants. The data analysis method in this study was based on pixel-based analysis that treats the two-dimensional (2D-)GC chromatograms as images. With this approach, it was possible to obtain an overall comparison of samples, but a full identification and quantification of all chemicals was not obtained. We are currently working on a strategy to quantify in GC×GC–HRMS with the use of a selection of standards spread in the 2D-space.
Q. Is GC×GC becoming more commonly used in environmental analysis and do you think it will become more widely adopted?
JC: Yes it is, but only slowly. I think it has the potential of becoming more widely adopted as the data processing workflows become standardized. The current software are still not optimal when it comes to suspect screening and NTS workflows. However, the technique has great potential, even for aqueous samples, and especially for the identification and quantification of compounds.
The analysis of compounds in aqueous samples would benefit from derivatization to increase the range of compounds detectable with GC×GC, for example, to lower the boiling point and decrease the polarity of compounds.
Q. Are there any general trends in NTS? Is NTS likely to become more commonly used?
ST: There is an increase in the requests from authorities for NTS. Strategies for the prioritization of chemicals and regulation using NTS is strongly needed. Another trend is to automize as many tools as possible for the identification of the micropollutants. Furthermore, estimating concentration of NTS compounds by different semi-quantification approaches is quite a new research field, which can help to increase the use of NTS.
Q. What is your group currently working on?
JC: We are developing a semi-quantification method to estimate the concentration of the identified compounds, and a nontargeted screening of drinking water itself. We are also working on the application of complementary methods, such as supercritical fluid chromatography (SFC–) HRMS—methods that are especially relevant to detect persistent and mobile organic compounds (PMOCs). PMOCs are man-made, highly polar organic chemicals that only degrade very slowly, if at all, in the environment and that show a low tendency to sorb to surfaces or to organic matter in soil and sediments.
References
- S. Tisler and J.H. Christensen, J. Hazard Mater. 429, 128331 (2022). doi: 10.1016/j.jhazmat.2022.128331.
- S. Tisler et al., (2022) (Under revision)
- S. Tisler, D.I. Pattison, and J.H. Christensen, Anal. Chem. 93(24), 8432–8441 (2021). doi: 10.1021/acs.analchem.1c00357
- J.S. Lübeck, G.L. Alexandrino, and J.H. Christensen, Environ. Sci. Eur. 32, 78 (2020).
Selina Tisler has been a postdoc in the Environmental Analytical Chemistry Group at the University of Copenhagen since 2020. She received her Ph.D. in 2019 from the University of Tuebingen, Germany. Her research focus is the advanced analysis of water, including nontarget screening of compounds of emerging concerns (CECs) and the identification of transformation products in the aquatic environment.
Jan H. Christensen is a professor in the Environmental Analytical Chemistry Group at the University of Copenhagen. He is leader of the group and of the Research Center for Advanced Analytical Chemistry (RAACE). He has pioneered analytical and chemometric methods for oil hydrocarbon fingerprinting and now works with all aspects of chemical fingerprinting. His group develop analytical platforms, new tools to process complex data, and apply this (contamin)-omics concept for analysis of complex mixtures of mainly organic contaminants.
