Water pollution is still a significant concern worldwide. A broad range of measures is being imposed to prevent such pollution in EU countries and set limits on permissible levels of certain substances in water sources. The detection and quantification of known pollutants such as pharmaceutical compounds in water is widely carried out using liquid chromatography coupled with mass spectrometry (LC-MS) based techniques. Introduction

LC-MS has become increasingly important in recent years following significant improvements in method development and technologies, which now offer very high sensitivity, reliability, and speed, as well as the ability to analyse numerous compounds from a single run. Detecting and quantifying unknown contaminants has remained a challenge, however. This article describes a new strategy for the analysis of pharmaceuticals and personal care compounds in water.

The method uses LC-MS combined with full scan time-of-flight (TOF) mass spectrometry for simultaneous quantitative and qualitative screening of numerous pharmaceutical compounds in river water. The method achieved satisfactory sensitivity with limits of detection for selected targeted compounds – including analgesics, opioids and psychostimulants – ranging from 0.003 to 0.01 µg/L and 0.01 to 0.5 µg/L in MS and MS/MS modes respectively.

Relevant legislation

Various EU measures have been announced in recent years to prevent pollution of water sources, including three major directives to control emissions (briefly reviewed in1): the Directive on Integrated Pollution Prevention and Control (IPPC); the REACH regulation for reforming chemical policy, which lists 100,000 regulated substances; and the Urban Waste Water Treatment (UWWT) Directive that will reduce the discharge of priority substances.

The prevention of water pollution and management of water quality remains challenging for several reasons, however. Firstly, these Directives have yet to be fully implemented, and it has not been possible to evaluate the Directives’ impacts on reducing priority substance emissions. Secondly, it is broadly believed that many other contaminants of concern have yet to be identified.

Thirdly, monitoring concentrations of the very large number of chemicals, along with their diverse chemical properties, is complicated. Finally, screening water samples for unknown (non-target) pollutants, as well as screening for known target pollutants and their metabolites and degradation products, remains challenging.

Causes for concern

The presence of pharmaceuticals and personal care products (PPCPs) in water sources is a growing concern. According to the United States Environmental Protection Agency2, PPCPs can include any product used by individuals for personal health or cosmetic reasons, or by agribusiness to enhance the growth or health of livestock. PPCPs therefore comprise a diverse collection of thousands of chemical substances, including prescription and over-the-counter therapeutic drugs, veterinary drugs, fragrances, and cosmetics.

PPCPs can enter the environment through several different routes, including bathing, swimming, consuming prescription drugs, manufacturing and consuming illicit drugs, veterinary drugs, agribusiness, residues from hospitals and residues from pharmaceutical manufacturing – although this industry is well defined and highly controlled.

From 1999-2000, the US Geological Society provided the first nationwide reconnaissance of pharmaceuticals, hormones and other organic wastewater contaminants3 and found contaminants in 80% of the sampled water sources. A variety of studies have since demonstrated the prevalence of PPCPs in water and soil, and further research has indicated certain pharmaceuticals could cause ecological harm2. To date, no evidence has been found of adverse affects on human health caused by PPCPs in the environment, but the full risks and potential effects on humans and aquatic organisms are currently unknown.

PPCPs comprise many drug classes of concern, including antibiotics, antimicrobials, oestrogenic steroids, antidepressants, calcium-channel blockers, anti-epileptics, and genotoxic drugs. It is already known that some of these substances can have damaging effects; for example, a class of antidepressants – selective serotonin reuptake inhibitors (SSRIs) – has been shown to have profound effects on spawning and other behaviours in shellfish2. Calcium-channel blockers can cause inhibition of sperm activity in certain aquatic organisms, and antiepileptic drugs such as phenytoin, valproate and carbamazepine can trigger extensive apoptosis in the developing brain, leading to neurodegeneration. In order to fully understand the effects of such compounds on the environment, scientists need technologies that can accurately identify, measure and monitor the presence of all known and unknown contaminants in water samples.

State of the art analysis

Liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) is widely used for analysing polar, semi-volatile, and thermally-labile compounds across a wide molecular weight range, including pesticides, antibiotics, drugs of abuse, X-ray contrast agents, drinking water disinfection byproducts, and so on.

High performance LC-MS/MS instruments in selective multiple reaction monitoring (MRM) mode can offer unmatched selectivity and sensitivity for quantifying PPCPs at trace levels, in reproducible fashion and without time consuming sample preparation, using either triple quadrupole or hybrid quadrupole/linear ion trap technology with pre-concentration steps.

Detecting unknown targets, however, such as metabolites of target compounds or even completely unexpected pollutants, requires a different approach. Non-targeted screening workflows do not use a target analyte list, and compound detection is not based on any a priori knowledge, e.g. retention times, or information on possible fragment ions. The acquired chromatograms can therefore easily contain thousands of ions from any compounds present in the sample, as well as from the sample matrix, requiring powerful software tools to explore such data and to identify the unexpected compounds.

Full scan accurate mass analysers have become increasingly popular among environmental scientists in recent years because these offer improved confidence for screening non-targeted compounds, and can offer retrospective analyses. Such systems have high mass accuracy for both precursor and product ions, and can enhance identification of isobaric compounds by distinguishing between compounds with identical nominal masses.

In particular, time-of-flight (TOF) MS has been applied for structure elucidation or identification of unknown compounds. When compared to triple quads, however, TOF MS systems tend to have the drawbacks of lower detection sensitivity, and lower efficiency in obtaining quantitative information.

New analytical approaches

In an innovative recent development, a new LC-MS system has become available that combines the high resolution and mass accuracy required for qualitative analyses, with quantitative workflows in targeted analyses, on a single platform. This new technology provides improved sensitivity for full scan MS, and uses information dependent acquisition (IDA) to enable the collection of MS/MS spectra for further identification of compounds. This means that accurate MS and MS/MS information can be used for simultaneous identification of both target and non-target organic compounds.

This technology was used to develop a new analytical strategy for simultaneous quantitative and qualitative screening of PPCPs in a single analysis. Accurate full scan mass spectrometry is combined with high MS/MS spectral acquisition rates, through IDA. Non-targeted compounds can be identified based on library searching, following data processing of accurate mass measurements in MS and MS/MS mode. Other non-target compounds, not included in the library, can also be identified simultaneously by screening the highest intensity peaks detected in the samples and by analysis of the full scan TOF MS, isotope pattern and MS/MS spectra.

Methods and materials

LC-MS The full experimental details are described elsewhere. In brief, LC-MS analysis was performed with a TOF system utilising an electrospray interface (ESI), using the following ion source-dependent optimised parameters: ion spray voltage floating, 5500 V; temperature 500° C; curtain gas 20; and ion source gas (GS1 and GS2) at 55 psi.

Flow injection analysis compound optimisation was performed in positive ionisation mode to obtain maximum sensitivity for identification and detection by means of the following parameters: declustering potential (DP), collision energy (CE) and collision energy spread (CES). An HPLC binary solvent delivery system equipped with a reversed-phase C8 analytical column, 150 mm length, 4.6 mm ID and 5 µm particle size was used.

Mobile phases A and B were acetonitrile and HPLC-grade water with 0.1% formic acid in both phases. The flow rate was 500 µL/min. A linear gradient was set from 10% A to 100% A in five minutes, then maintained at 100% A for five minutes; re-equilibration time was five minutes and injection volume was 100 µL.

The MS was operated in full scan TOF MS and MS/MS mode with IDA in a single run analysis, for targeted and non-targeted analysis screening as described previously. The IDA method was programmed for a TOF MS survey of 250 ms and up to five dependent MS/MS scans per cycle. Total cycle time was fixed to 0.6750 s. Mass calibration, resolution adjustments and optimisation experiments were performed as described, the acquisition method was developed for identification of as many putative compounds as possible in one run, with sufficient sensitivity. This was done via IDA, simultaneously combining a TOF MS survey scan and several corresponding MS/MS events.

Samples River samples were collected from Henares River, Madrid, Spain. Sampling points included the treated effluents from a very industrial and densely populated area; the mouth of an important wastewater treatment plant; positions downstream from the sewage plant discharge area; and from a heavily populated area with large contributions from urban and industrial zones (20 m and 1.15 km downstream from the effluent point, respectively).

Grab samples (1 L) were collected during a sampling campaign carried out in June 2011. No conventional treatment, such as solid phase extraction (SPE), was carried out on water samples. Before analysis, all samples were spiked with the mixture of surrogate standards, nicotine-d3 and C3-Caffeine, in acetonitrile, and then all extracts were filtered with a 0.45 µm PTFE syringe filter to remove suspended solids and particulate matter and to check the analysis.

Results and discussion

Qualitative and quantitative screening of target species via QTOF Eleven target compounds in solvent were successfully identified and quantified using the new method. Mass accuracy was <5 ppm in full scan acquisition and purity score values higher than 65% were obtained in MS/MS mode, leading to an appropriate identification. Mass errors and purity score values were reported in the worst cases at 0.01 µg/L and 1 µg/L from injection of standards in solvent (n = 3).

Figure 1 shows the identification of nicotine from a river water sample. Good mass accuracy and purity score values were obtained, suggesting correct compound identification. The TOF MS spectrum (left of Fig. 1) shows the measured mass of nicotine at m/z 163.1229, which matches the calculated mass 163.1222 with an error of −4.5 ppm.

In order to provide the highest possible confidence in compound identification, a MS/MS library spectrum was performed to contain the complete molecular fingerprint of each molecule. As can be seen in the TOF MS/MS spectrum on the right of Fig. 1, the fragment ion masses agree with the MS/MS spectra recorded at 30 eV, enabling precise structural assignments and increasing result confidence (purity score = 68.5).

The method was applied to the analysis of several river water samples. As can be seen in Table 1, most of the selected compounds included in the study were present in the river samples at concentration levels ranging from 14 ng/L (EDDP) to 677 ng/L (BE, benzoylecgonine). Morphine was not detected in any sample, and methadone only once, but at levels below its quantification limit (7 ng/L). The ability to confirm compounds using MS/MS library searching ensured that no false positives were reported.

Identification of non-target compounds In order to identify non-target compounds, river samples were reprocessed using our software’s non-targeted peak finding algorithm. Fourteen drugs tested positive in the river water samples analysed (n = 7), based on library searching, and taking into account the following criteria: a purity score value >65, and an error mass <5 ppm, as summarised in Table 2. The most frequently detected non-target drugs were metformin (anti-diabetic), venlafaxine, tradozone (antidepressant), and ketorolac (nonsteroidal anti-inflammatory). The greatest number of chemicals was identified at a sampling point that corresponded with the discharge zone from an important wastewater treatment plant.

The approach was limited in that it was only possible to confirm compounds that were included in the MS/MS library. Despite finding more than 44 peaks with relatively high intensities (>10.000 cps), only an average of five peaks were confirmed in the library search by accurate MS and MS/MS. As a consequence, an enlargement of the libraries is necessary to ensure adequate screening.

Expanding non-target identification The use of accurate mass full scan MS, MS/MS data and isotope patterns can be further used in retrospective analysis to identify non-targeted compounds1. For example, loratidine, which was not included a priori in the library, was found in a river water sample. Its presence was confirmed based on the mass accuracy of the molecular ion (0.1 ppm), the characteristic fragment ions in MS/MS mode and on the isotopic profile. In this case, the measured mass matches the calculated mass within <5 ppm for MS as well as for MS/MS spectra.

Analytical performance QTOF provides sufficient sensitivity to measure contaminants in river water at concentrations in the ng/L range (ppt). The limits of detection (LOD) for the selected compounds range from 0.003 µg/L to 0.01 µg/L, and from 0.01 µg/L to 0.5 µg/L, in MS and MS/MS, respectively. Therefore, with 100 µL injected, QTOF sensitivity and quantification capabilities can be considered similar to triple quadrupole MS, where the injection volume is typically 10–20 µL1.

While triple quadrupole MS delivers high selectivity by double mass filtering in Q1 and Q3, full scan TOF MS sensitivity is only based on the mass accuracy of molecular ions and high resolution is used to remove interference. Additionally, QTOF gives the benefit of quantifying over the ng/L to µg/L range with correlation coefficients superior to 0.997 (results determined from solvent standard and spiked matrix calibration curves, data not shown).

Conclusions

The new method for simultaneous screening of targets and non-targets by QTOF-MS/MS with IDA achieved high confidence results through accurate mass scoring, isotope pattern and the use of accurate mass MS/MS information. For targeted compounds, the combination of TOF MS and high sensitivity MS/MS was shown to be a powerful tool for screening and quantification. All of the target drugs were shown to be present in most of the river water samples at ng/L concentration levels, with the exception of morphine. By injecting large sample volumes (100 µL) into the QTOF-MS, it was possible to achieve similar sensitivity and quantification capabilities to those typically achieved using triple quadrupole MS.

Identification of non-target compounds could be performed through manual sample analysis, by means of data processing in order to find the best degree of confidence based on mass accuracy in the full scan mode; the degree of purity score with spectra from a MS/MS library; and structural elucidation and mass errors of fragment ions.

This new method has the potential to provide valuable new information about known and unknown PPCP contaminants in water sources. The simultaneous provision of qualitative and quantitative data on multiple known and unknown compounds from a single run will enable environmental scientists to analyse and monitor water samples with much higher confidence and higher throughput.

Published: 07th Mar 2013 in AWE International