Advancing Nontargeted Analysis of Water

Environmental analysis of water presents many challenges, one of which is the potential presence of unknown contaminants. Nontargeted analysis seeks to address that challenge, but has its own complications. Imma Ferrer, an associate research scientist at the University of Colorado, has spent the last 22 years developing methods for emerging contaminants using liquid chromatography–mass spectrometry (LC–MS) techniques, including nontargeted approaches. Here, she discusses some of her recent research on nontargeted analysis of water, including water from wastewater treatment plants, hydraulic fracturing wastewater, and environmental water samples.

In a recent study, you used nontargeted analysis of water samples from wastewater treatment plants (WWTPs) to assess whether the treatment methods were effectively removing not only the toxic substance N-Nitrosodimethylamine (NDMA), but also its precursors (1). Why did you undertake this study and why is it important?

NDMA is known to be a carcinogen, and many states have regulations in place for drinking water. Over the course of many years, efforts have been made to understand what causes the formation of NDMA in water treatment plants, and one of the established theories is that some precursor compounds (such as pharmaceuticals and pesticides) have the ability to form NDMA, after source water is treated with chloramines. Thus, to be able to screen for these potential precursors, mainly by high-resolution mass spectrometry (HRMS) techniques, is of essential importance.

In 2014, we had a project with the Water Research Foundation (WRF) where we partnered with colleagues from Arizona State University. We had screened for compounds that were potential precursors of NDMA, and we successfully identified a few of them using our HRMS techniques. Therefore, that same approach was used for a second project involving the drinking water treatment technologies used by Orange County in California. In this study, we were able to see the effect that chloramination had in the formation of NDMA, and also investigate the role of precursors present in water samples before and after treatment.

What approach did you use to qualify significant environmental analytes of importance?

The approach we used is based on nontargeted screening for a whole variety of compounds. Currently, we have a database of more than 200 compounds, including pharmaceuticals, pesticides, and surfactants. The data gathered by the high-resolution instrument (in this case an LC–quadrupole time-of-flight MS [LC–QTOF-MS]) system is then compared to the database, and possible identification hits are then scrutinized in terms of accurate mass and fragmentation patterns. In my opinion, there is no gold standard for this type of approach. Many papers in the literature have tried to come up with a universal and single way to carry out this step, but there are still a lot of variables that make each approach unique depending on the focus of the study. For example, experience analyzing hundreds of compounds and looking meticulously at their fragmentation data is no substitute for any large database or mass bank information on the web. I can look at mass spectra and quickly identify an analyte without the need for these tools; that saves a lot of time, and people usually do not account for this type of skill.

You also looked at neutral losses of dimethylamine using auto MS-MS. Can you briefly explain this approach?

The idea behind this process was that a precursor compound able to form NDMA will also have a common moiety related to chemical structure. In this case, the dimethyl-amine group, very common in pharmaceuticals and pesticides, would be the one susceptible to be fragmented from a precursor ion. So, by performing MS-MS fragmentation and identifying compounds that have this same group in the chemical structure, we would find out a list of potential NDMA precursors.

At the time of the study, commercial software for this type of screening was not available in the instrument. So, we imported the data to MatLab, and looked for the neutral losses that would indicate the presence of precursor compounds with that exact common moiety. This feature is now included in many software packages from instrument manufacturers, so it makes the process easier. That is why it is important that bench scientists like myself collaborate and talk to the big instrument companies to ask them to include those essential tools that make research much easier.

In what databases did you search for matches with your results? How important is the development and maintenance of such databases for nontargeted screening in environmental analysis? Are you involved in any database development yourself? Where might one find information about the available databases?

The development of databases for nontargeted screening is extremely important, in my opinion. I have used a relatively small database based on standards that we have analyzed throughout the years. But there are also commercial ones, such as Chemspider, Comptox-EPA, and PubChem. The main problem is to sort through these huge and extensive databases. Having a link (pathway) between instrument data and databases is the key. These two interfaces have to talk to each other. I just submitted a paper where we use a novel approach that combines MS-MS fragmentation data with a large database. The software we used is called MathSpec (www.mathspec.com), and takes into account not only accurate mass and formula generation, but also fragmentation data and fragmentation pathways. The idea behind it is that each fragment mass has to be interpreted in the context of other masses. What we learned is that good data will lead to successful identifications, even when no standards are available.

What results did you find in terms of the presence of NDMA precursors in treated wastewater treated with reverse osmosis or more advanced treatments? Can those results help us understand how to better manage treatment?

We found out that reverse osmosis and advance treatment processes, such as UV, are very effective to remove NDMA precursors. If no NDMA precursors are present, the formation of NDMA itself will be diminished. So, in general, these are good treatment techniques that can be used to minimize the presence of NDMA in our drinking water, and these results will help to determine what courses of action to take and better manage treatment.

What did you find in terms of the ability to detect NDMA precursors, either using auto MS-MS analysis or through nontargeted analysis using databases?

The approaches we used were successful in identifying a few NDMA precursors. But many of these precursors do not have the ability to form NDMA, or the yield is very small. However, in a previous study, we found the presence of methadone as an important precursor for NDMA. Prior to that date, the focus had been on ranitidine (a common pharmaceutical), so that finding broadened the focus to include other compounds.

How important is the role of nontargeted analysis is water quality investigations? How widely is it used?

Nontargeted analysis is widely used by everyone using HRMS techniques. I would say that is the main reason behind using this type of instrumentation, to be able to see beyond a targeted list of contaminants. As I mentioned above, there are lots of papers dealing with workflows, or universal ways to carry out this approach. Each one is unique, and focused on the type of application the researcher is conducting, and sometimes even on the instrument used. We still have a long way to go to make this approach universal and successful for everybody.

I personally do not like other terms, such as suspect analysis. Your analysis is either targeted (you know what you are looking for), or nontargeted (you are not looking for it, but the instrument can see it). With newer and more accurate instruments, the databases, including accurate mass and fragment ions, will definitely continue to grow and expand.

In another paper, you studied the identification of proprietary compounds, including poly(ethylene glycol)s, amino-poly(ethylene glycol) carboxylates, and amino-poly(ethylene glycol) amines, in hydraulic fracturing wastewater using LC–QTOF-MS with solid-phase extraction (SPE) (2). Why did you choose these analytical techniques?

SPE is one of the most important steps in sample preparation, and yet sometimes gets forgotten. We have been studying hydraulic fracturing waters for over eight years now, and all we had to do is directly inject the sample, because usually the main additives were in very high concentrations. However, there are always those compounds present at very trace levels that are invisible even to the most sensitive instrument, and those compounds can be enhanced by SPE techniques. In this case, we were able to identify new proprietary compounds that we had not seen before. Another issue is the high salt content usually present in this type of samples. This salt usually masks the detection of compounds by mass spectrometry. By using SPE, salt can be removed and detection can be enhanced, and instrument life will also be lengthened.

The reason for using LC-QTOF-MS is to be able to see beyond the list of targeted contaminants, as explained above. Moreover, these compounds were not in any database, so we used our knowledge in accurate mass and fragmentation to be able to identify them. The instrument sees everything that is ionizable; it is the human eye that has to do the ultimate work
of identification.

How is this study different from others that you or others have done on the analysis of hydraulic fracturing compounds in environmental samples?

This study combined the use of SPE and TOF for the analysis of hydraulic fracturing waters, which to date, was one of the first to combine these two techniques for this type of
water samples.

What compounds were you able to identify, and with what level of confidence, in terms of their identification or link to hydraulic fracturing? Had these compounds been detected in other studies? Overall, what does this study tell us? What was your most surprising discovery of the compounds you detected, with masses in the range of m/z 120−986?

The level of confidence in the identification of these compounds was 100%, as we also purchased pure standards and confirmed retention time, accurate mass, and fragment ions (which is consider the gold standard in identification). These compounds had never been detected in other studies related to hydraulic fracturing. In fact, there are no known reports of the addition of these compounds to hydraulic fracturing fluids. The terms the oil and gas industry uses to refer to these compounds are general and do not include specific chemical names for additives. Therefore, their identification is an important step in understanding the chemistry, treatment, and possible toxicity of hydraulic fracturing wastewater.

You recently published a review paper on the occurrence of opioids in environmental water samples (3). For how many years have researchers been conducting such analyses? How many different opioids and their metabolites have been studied? Are there any important gaps in the data so far?

The reason behind publishing a review paper on the occurrence of opioids in environmental waters is the importance of the topic: Each year, thousands of deaths occur as a result of the consumption of these drugs. I also realized that not all opioids were accounted for, and, in fact, only a small group of them had been monitored over the years. One exchange student from abroad came along to spend a few months in our laboratory, and I thought that would be the perfect topic to work on. From what we found in the literature, it seemed that only 20% of the total number of opioids (>200 total) and their metabolites had been studied. One of the reasons for this is the difficulty to find pure standards, so scientists usually choose only a few targeted analytes to study. Our approach was to take advantage of the screening capability of accurate mass analysis to be able to see beyond that reduced group of analytes. In a second study we carried out, we detected the presence of some opioid metabolites that had not been reported to date.

What methods-including both sample preparation and analytical techniques-are most researchers using for these studies, and why? Are the chosen methods the most suitable ones? Has the thinking about which approaches are best changed over time?

The methods most researchers are using for these studies involve SPE and LC–MS techniques (both targeted and nontargeted approaches). Depending on what you are looking for, you might choose one approach or another. For example, if looking only at a relatively reduced number of compounds (say, 20–30 compounds), then a targeted approach with tandem mass spectrometry techniques would be the best in terms of sensitivity (one can see down to the low ng/L most of the time). If a researcher is looking to expand the scope and look also for metabolites or degradation products, then a nontargeted approach using accurate mass techniques would be the ideal method. Lately, in the literature, there is a slight preference for the latter method, because instruments that do accurate mass have become more and more sensitive, so you see your chosen analytes of interest but also other contaminants that might be present in the sample, so that is an added bonus. Also, do not forget, there is always the possibility to perform retrospective data analysis with accurate mass techniques; the data gathered with the instrument will always be there.

What do the data so far show about the occurrence of opioids or transformation products of concern in different classes of water-wastewater, surface water, and drinking water?

The data showed that wastewater is the source of these contaminants in water, as we already knew from past studies. It also revealed that the concentrations in surface water were much lower, but it also showed that some compounds and their metabolites persist in water, and have the potential to end up in drinking water if not treated properly.

What is known so far about the presence of opioid metabolites and transformation products in environmental waters? You have mentioned that a few nontargeted studies have used high resolution accurate mass to determine the transformation products or metabolites. Why is this an important topic for future research?

Pharmaceutical compounds such as opioids are metabolized in the human body and excreted as metabolites. Once they reach surface waters, other processes, such as degradation, might occur and transformation products can be formed as well. These metabolites and transformation products are often omitted from targeted studies, and here is when high resolution mass spectrometry can aid to the detection of these compounds, by doing a full-spectrum scan of the whole sample. Not much is known about the toxicity of metabolites or degradation products, so that is an important topic for future research.

References

1. S.L. Roback, I. Ferrer, E.M. Thurman, K.P. Ishida, M.H. Plumlee, A. Poustie, P. Westerhoff, and D. Hanigan, Environ. Sci. Water Res. Technol. 4, 1944 (2018).
2. K.A. Sitterly, K.G. Linden, I. Ferrer, and E.M. Thurman, Anal. Chem. 90, 10927–10934 (2018).
3. M.C. Campos-Mañas, I. Ferrer, E.M. Thurman, and A. Agüera, Trends Environ. Anal. Chem. 20, e00059 (2018).

Imma Ferrer is an Associate Research Scientist at the University of Colorado, in Boulder, Colorado. Her PhD at the University of Barcelona (Catalonia, Spain) focused on liquid chromatography–mass spectrometry (LC–MS) techniques to detect pesticides in environmental samples. She then conducted post-doctoral research with the U.S. Geological Survey at the National Water Quality Laboratory in Denver, Colorado, where she worked on tandem MS techniques to analyze pharmaceuticals in the environment. She then spent five years as an Assistant Professor at the University of Almeria (Almería, Spain), where her main focus was developing advanced LC–MS methodologies to analyze pesticides in food. In 2008, she moved back to the United States, and is currently an Associate Research Scientist and a co-founder of the Center for Environmental Mass Spectrometry at the University of Colorado in Boulder.