Using HPLC-MS-MS to Detect Pharmaceutical Contaminants in Water

E-Separation Solutions

LCGC recently spoke with Edward T. Furlong of the Methods Research and Development Program at the National Water Quality Laboratory with the U.S. Geological Survey (USGS) about his group's work on developing new methods to detect pharmaceutical contaminants in waterways.

LCGC recently spoke with Edward T. Furlong of the Methods Research and Development Program at the National Water Quality Laboratory with the U.S. Geological Survey (USGS) about his group’s work on developing new methods to detect pharmaceutical contaminants in waterways.

What prompted you to develop a direct-injection high performance liquid chromatography (HPLC) tandem mass spectrometry (MS-MS) method to determine pharmaceuticals in filtered-water samples?
For more than a decade the presence of pharmaceuticals and personal-care products (PPCPs) in the aquatic environment has been a topic of increasing interest to the public and to scientists. A simple search of the terms “antibiotics, drugs, or pharmaceuticals” in the six environmental science journals that have published the bulk of papers on this topic would have resulted in 18 to 25 papers total in any one year between 1998 and 2002; in 2002 that number increased to over 40, and in 2012 the number of publications in those six journals was over 350. We expect this publication trend to continue.

Concurrently, mass spectrometers have become more and more sensitive, so that triple-quadrupole instruments can now be routinely used to detect many small molecules, including most pharmaceuticals, at subpicogram amounts (on column). This level of performance suggested that a direct-injection HPLC-MS-MS method was possible at the nanogram-per-liter concentrations typically observed in natural aquatic environments.

Finally, as ecotoxicologists and other environmental scientists have studied the effects of exposure to single pharmaceuticals and pharmaceutical mixtures on fish and other aquatic life, both in the laboratory and in the field, significant sublethal effects, often behavioral in nature, have been demonstrated at the ambient parts-per-trillion concentrations that are routinely observed. 

Thus, we saw that there was a compelling need to develop a single method that could comprehensively, sensitively, and specifically identify and quantify pharmaceuticals in environmental samples. In addition to sensitivity and specificity, we hoped to include the widest range of pharmaceuticals, particularly human-use pharmaceuticals that researchers at the USGS might expect to encounter in samples from across the range of water types and sources present in the United States. Our final list of analytes for the method was the result of our own understanding of prescribing trends and that information, our knowledge of the available scientific literature, and input and collaboration with our USGS colleagues, particularly the USGS’s Toxic Substances Hydrology Program-Emerging Contaminants Project and the National Water Quality Assessment.

In what ways is this technique an improvement over existing methods?
Our method separates, identifies, and quantifies 110 pharmaceuticals from 100 µL of a filtered-water sample using a small-particle-diameter (1.8 µm), reversed-phase HPLC column with a mobile phase consisting of formic acid–sodium formate aqueous buffer and methanol.  The eluted pharmaceuticals are ionized under positive electrospray ionization and qualitatively identified by multiple reaction monitoring (MRM) of two precursor–product transitions and by retention time. Just prior to analysis we fortify the sample with a suite of stable isotopically labeled pharmaceuticals and determine concentrations by isotope-dilution standard quantitation.

The major improvements of this method over many other current methods are the large number of pharmaceuticals identified and quantified, the absence of a sample preparation step other than filtration prior to analysis, and the small overall sample size. The large number of pharmaceuticals determined permits the USGS and its collaborators to comprehensively assess for the presence and distribution of pharmaceuticals that can derive from many sources and be present in a wide variety of water types. The absence of a sample preparation step other than filtration, which we typically do at the time of sample collection, reduces bias, variability or inadvertent error that may be introduced by sample preparation. 

Direct analysis of a filtered 100-µL sample aliquot reduces the potential for matrix effects, as the number of protons available to ionize sample constituents is more favorable for efficient ionization of the pharmaceuticals of interest and any interferences and other compounds also present in the sample. As a result it is less likely that there will be competition between the compounds of interest and the interferences for the hydrogen protons that ionize compounds in electrospray. This competition for charge can lead to matrix effects, particularly matrix suppression, which can result in overestimation of concentration.

Finally, an important improvement is the cost savings achieved. Our field scientists can now collect much smaller samples, using disposable filter cartridges and syringes, thus reducing effort in the field, and also reducing the overnight delivery costs for sample shipping. In the laboratory we reduce the consumables and solvents that require use and disposal. Effectively we are reducing the carbon footprint and financial costs for collecting ultratrace contaminant data, a win-win for the laboratory and the field, the USGS as a federal agency, and for the US taxpayer.

What results have you seen using this method?
We have validated our method by determining recoveries of the entire suite of pharmaceuticals at a minimum of four fortification concentrations in groundwater, surface water, drinking water, wastewater effluent, and wastewater influent. The wastewater matrixes were of particular interest in that they are particularly challenging to analyze using more standard solid-phase extraction sample preparation and HPLC–MS-MS. 

We have been very pleasantly surprised by how well our method has performed across this diverse span of matrixes, especially in wastewater effluent and influent. For these matrixes in particular, the presence of ambient concentrations of many compounds in the validation matrix has required correcting for background concentrations when determining recovery, which can introduce additional variability. This is particularly a problem where the ambient concentration is an appreciable fraction of the amended concentration or greater than the amended concentration, which is often the case for the wastewater matrixes. 

Even with that qualifier, we found that individual recoveries of the suite of pharmaceuticals fortified in water samples and determined by this method typically were greater than 90% in reagent water, groundwater, drinking water, and surface water. Correction for ambient environmental concentrations of pharmaceuticals hampered the determination of absolute recoveries and method sensitivity of some compounds in some water types, particularly for wastewater influent and effluent, but for those pharmaceuticals where ambient concentrations were not substantial, acceptable recoveries were obtained. 

The calibration range for each compound typically spanned three orders of magnitude of concentration. Absolute sensitivity for some compounds, as determined by the US Environmental Protection Agency’s (US EPA) method detection limit (MDL) method, was as low as 0.449 ng/L, and ranged as high as 94.1 ng/L, primarily as a result of the inherent ionization efficiency of each pharmaceutical in the electrospray ionization process.  However, for most pharmaceuticals in our new method the MDL was less than 50 ng/L.

We have subsequently used the method for the analysis of surface water, ground water, wastewater effluents, and leachates from landfills and have found a wide range of pharmaceuticals; in some sample types greater than 60 individual pharmaceuticals were determined in individual samples. We are in the process of preparing the results of these studies for publication, and it appears that that the pharmaceuticals and the concentrations we observe appear to reflect the uses and behaviors of our population at large. Ultimately, however, there are many factors that govern the presence and distribution of pharmaceuticals in water resources, including the inherent metabolic recalcitrance of many pharmaceuticals, the multiple source pathways by which they are introduced into water resources including wastewater treatment and discharge, and the subsequent chemical and biological processes that act to transform or remineralize them after introduction.

What kind of environmental impact do you expect this method to have? Will it eventually help reduce the presence of pharmaceuticals in filtered-water samples?

The method itself we hope is environmentally friendly, since it reduces consumables and disposal costs associated with sample preparation, along with the carbon footprint associated with sample collection and transport. 

More broadly, I think the impact of having this method available to our USGS and other federal, state, and university collaborators will allow scientists to more comprehensively “map” the distributions, compositions, and concentrations of pharmaceuticals in US water resources. This method will become incorporated into national-scale monitoring, such as what has already been initiated by the USGS Toxics Program, and is now being incorporated into the USGS’s National Water Quality Assessment Program and in more regionally or locally focused studies conducted by the USGS’s state-based Water Science Centers.

The method also will have a major impact on the quality and depth of the applied research being conducted to elucidate the sources, fates, and ultimate effects of pharmaceuticals and other emerging contaminants, particularly that undertaken by the USGS Toxics Emerging Contaminants Project and its many collaborators. That project will apply the method at some key long-term research sites and in specific projects focused on providing more focused, hydrologically grounded understanding of the effects of these compounds on ecosystem and human health. 

One such project where that has already occurred is a joint study conducted by USGS and the US EPA in which we have sampled source and treated waters for 25 municipal drinking water treatment plants (DWTPs) across the United States for pharmaceuticals and other compounds that the US EPA classifies as contaminants of emerging concern (CECs). This study, which is being prepared for publication, will provide insight into the compositions and concentrations of pharmaceuticals and many other CECs entering DWTPs and their subsequent removal or reduction during treatment.

Was your research based on water samples from several different locations or was it localized to one area? 

We anticipated that this method would be used in a variety of water studies and contexts, so we developed a method that could be applied to a wide range of water types. When we validated it we looked at major water matrixes from ultrapure reagent water to wastewater influent. We found it impractical to assay multiple samples from within each matrix type and chose to focus on one representative sample of each matrix type and analyze it at a minimum of four fortifications from 40 to 2000 ng/L. Subsequently, we included comprehensive matrix replicate and matrix spike samples for both source and finished water as part of our experimental design of our joint USGS-US EPA drinking water study, the first project that used this method extensively. As we have continued to expand the matrixes and sample types in additional studies we have added random matrix replicate and matrix spikes as part of our own internal method quality control evaluation. These data are under evaluation, but our preliminary assessment is that it confirms the good method performance in many matrixes that we observed in our initial method validation.

What are the next steps in your research?

Right now we are finishing the documentation of our method for publication and use as an official USGS method. We also plan to publish specific aspects and observations from our method development in peer-reviewed journals.

After that we plan to develop a second analogous direct-injection HPLC–MS-MS method for pharmaceuticals that are more optimally ionized under negative electrospray ionization conditions. We have begun the optimization for that effort.

The USGS also began deploying our method in 2013 as part of the National Water Quality Assessment. It is being used to evaluate the presence and distribution of pharmaceuticals in about 700 groundwater samples being collected across the United States.

Finally, I would like to point out that a project of this complexity and depth is not something any one person can do alone. I am extremely fortunate to work with several talented chemists at the USGS National Water Quality Laboratory, particularly Mary Noriega, Laura Coffey, Chris Kanagy, and Mark Burkhardt, now at the U.S. Environmental Protection Agency, all talented, dedicated scientists and with whom it is my greatest pleasure to call my collaborators. This method could not have been developed without them or the financial, intellectual, and logistic support we received from the management of our laboratory and from the USGS Programs and Projects mentioned above.

Related content:

Analysis of Pharmaceuticals and Personal Care Products in River Water Samples by UHPLC–TOF-MS

Screening of Pollutants in Water Samples and Extracts from Passive Samplers using LC–MS and GC–MS