Detection of Trace Level Pharmaceuticals in Drinking Water by nline SPE Enrichment with the Agilent Infinity Series nline-spe Solution Application Note Environmental Author Edgar Naegele Agilent Technologies, Inc. Waldbronn, Germany Abstract This Application Note demonstrates the use of the Agilent Infi nity Series nline-spe solution combined with triple quadrupole mass spectrometric detection for the analysis of antibiotics at trace levels down to ppt in drinking water. In addition, performance data of the online SPE system for linearity, area and retention time precision, recovery, and concentration precision and accuracy in real water samples from different sources are shown and discussed. rmetoprim Trimethoprim Nalidixic acid Counts Flumequine Erythromycin rnidazole xolinic acid Sulfamethoxazole..5..5..5..5 4. 4.5 5. 5.5 6. 6.5 7. 7.5 8. 8.5 9. 9.5..5 Acquisition time (min)
Introduction Antibiotics are widely used in human and animal treatment of bacterial infections. After passing through water treatment plants, they can be found in surface water and drinking water. It is important to detect those compounds, even at trace levels, down to the lower single digit ppt (ng/l) level. There are nine typical classes of antibiotics in medical use. Thirteen compounds from these classes were used in this Application Note (Table ). This Application Note describes the Agilent Infinity Series nline-spe solution based on the Agilent 9 Infinity Flexible Cube for first, enriching different trace level antibiotics and second, separating them from each other. The online-spe system includes an Agilent 6 Infinity Quaternary Pump because commercially available SPE cartridges do not support backpressures above 6 bar. The full functionality of the online-spe system is achieved by using just one HPLC pump. The performance of the system is demonstrated by the mass spectrometric detection of a suite of antibiotics down to a limit of detection (LD) of less than ppt. To enhance the throughput of the system, a valve solution for the parallel use of two alternating SPE cartridges is presented. Experimental Instrumentation Agilent Infinity Series nline-spe solution system consisting of: Agilent 6 Infinity Quaternary Pump with internal degasser (GC) and LAN card (G69C) Agilent 6 Infinity Standard Autosampler (G9B) with 9 μl head (G-67), multidraw kit (G-687) and thermostat (GB) Table. Classes of antibiotics and individual compounds used in this study. Antibiotics class Compounds Fluoroquinolones Flumequine Quinolones Nalidixic acid xolinic acid Sulfonamides Sulfamethoxazole Imidazoles rnidazole Macrolides Erythromycin Tylosin Tetracyclines Tetracycline Chlortetracycline b-lactams Cefotaxim Diaminopyrimidines rmetoprim Trimetoprim Glycopeptides Vancomycin Agilent 9 Infinity Flexible Cube (G47A) to mount the -position/-port valve Agilent 9 Infinity Thermostatted Column Compartment (G6C) Agilent Infinity Series online SPE starter set G474A. Figure shows how the modules of the system are set up, and Table lists the capillaries and accessories needed to run the online-spe application. With the G474A nline SPE starter set, all necessary parts such as capillaries, cartridges, and the -position/-port valve head (6 bar) are ordered in one package. Agilent 9 Infinity Thermostatted Column Compartment Agilent 646 Triple Quadrupole MS Agilent 9 Infinity Flexible Cube Agilent 6 Infinity Standard Autosampler Agilent 6 Infinity Quaternary Pump Thermostat Figure. Setup of the Agilent Infinity Series nline-spe solution with MS detector (the solvent bottles in the center are for SPE loading, rinsing and conditioning).
MS-Detection Agilent 646 Triple Quadrupole LC/MS with Agilent Jet Stream Technology Analytical Column Agilent ZRBAX Eclipse Plus C8,. 5 mm,.5 μm (p/n 95976-9) Trapping Columns x Guard Column Hardware Kit (p/n 8999-9) Agilent PLRP-S Cartridges,..5 mm, 5- µm (p/n 598-7) Software Agilent MassHunter data acquisition for triple quadruple mass spectrometer, Version 6. Agilent MassHunter ptimizer software, Version 6. Agilent MassHunter Qualitative software, Version 6. Agilent MassHunter Quantitative software, Version 5. Table A. Content of the Agilent Infinity Series online SPE solution starter set G474A. (SST = stainless steel). Parts Qty Description rder No. -position/-port valve head Valve head to be mounted in Flexible Cube 567-445 Guard column hardware kit To insert SPE cartridge 8999-9 nline SPE capillary kit Contains required capillaries 567-578 Table B. Content of the online SPE capillary kit which is part of G474A, Agilent Infinity Series online SPE solution starter set. (SST = stainless steel). Parts and Capillaries Qty Description rder No. mm,. mm id, SST 5 Valve to guard column hardware and back to 567-465 valve; and one valve crossing BondElut online PLRP-S 5- µm SPE cartridges 598-7..5 mm 4 mm,. mm id, SST Valve to column, valve to autosampler 567-4647 Waste line m Valve to waste 89-7 5 mm,.5 mm id, SST Flexible Cube pump to autosampler 567-57 7 mm,.7 mm id, SST LC pump to valve 567-4648 Finger tight fitting For waste line -56 Table. Detailed HPLC method according to the individual modules. HPLC Method Agilent 6 Infinity Quaternary Pump Solvent A Water, 5 mm ammonium formate Solvent B ACN +5 % water, 5 mm ammonium formate Flow rate.4 ml/min Gradient minutes 5% B, 5 minutes 5% B, minutes 98% B Stop time 5 minutes Post time minutes Agilent 9 Infinity Thermostatted Column Compartment Column temperature 4 C Agilent 9 Infinity Flexible Cube Right valve -position/-port Quick-Change valve head Pump.5 ml/min Solvents A: Water, B: ACN minutes Pump seconds, Solvent A 5 minutes right valve change position 7 minutes Pump 8 seconds, Solvent B minutes Pump seconds, Solvent A Agilent 6 Infinity Standard Autosampler Injection volume,8 µl (automated multidraw of two times 9 µl) Needle wash In vial (MeH) Draw and eject speed, µl/min Sample temperature C Two trays with 5 positions each (G-445) 6-mL screw cap vials (glass, p/n 9-77), screw caps (p/n 9-79), preslit septa for 6-mL screw cap vials (p/n 588-758)
In the setup of the online-spe LC system, the 9 Infinity Flexible Cube (Figure ) has a -position/-port valve with two trapping columns next to the piston pump and the solvent selection valve for flushing the sample on the trapping columns and the re-equilibration of those columns (Figures A and B). The piston pump inside the Flexible Cube is connected to the autosampler to flush the sample directly onto one trapping column (SPE ). The other trapping column (SPE ) is connected to the analytical column and to the LC pump (Figure A). Rail for additional valves Solvent selection valve (three wash solvents) Quick-Change valves and Piston pump (4 ml/min, 6 bar) Figure. The Agilent 9 Infinity Flexible Cube is an additional module for the Agilent 9/6 Infinity LC system, using one or two Agilent Infinity Series Quick-Change valves. Analytical column Agilent 646 Triple Quadrupole MS Flexible cube SPE (Elute) Solvent selection valve 9 8 Piston pump 4 5 6 7 SPE (Load) Standard Autosampler (9 µl Head) LC Pump Waste Figure A. Valve positions for loading the sample on trapping column SPE while trapping column SPE is being eluted. 4
After loading the trapping column with sample, the -position/-port valve was switched and thus the positions of the trapping columns were exchanged (Figure B). Now, the LC pump delivers the gradient to elute the enriched analytes in backflush mode from the trapping column (SPE ) onto the analytical column. Simultaneously, the trapping column (SPE ), which had been loaded with sample in the previous run, is cleaned and reconditioned by a purging procedure. This cleaning procedure is done by the piston pump with the cleaning solvents selected by the solvent selection valve. Table shows a summary of the LC method for the main modules. Flexible cube Solvent selection valve Piston pump 4 5 SPE (Elute) Analytical column SPE (Load) 6 7 9 Agilent 646 Triple Quadrupole MS 8 Standard Autosampler (9 µl Head) LC Pump Waste Figure B. Valve positions for loading the sample on trapping column SPE while trapping column SPE is being eluted. Table. Summary of the LC method for the Agilent 6 Infinity Standard Autosampler, the Agilent 6 Infinity Quaternary Pump and the Agilent 9 Infinity Flexible Cube. Agilent 6 Infinity Standard Autosampler Agilent 6 Infinity Quaternary Pump Multidraw 8 µl sample Inject 5% Solvent B Gradient 5% B to 98% B 98% Solvent B post run Agilent 9 Infinity Flexible Cube Minutes Pump seconds Switch valve to next position Pump 8 seconds Pump seconds solvent 4 5 6 7 8 9 4 5 6 7 8 9 4 5 5
The MRM and dynamic MRM triple quadrupole MS method was developed by using the MassHunter optimizer software and direct injections of individual antibiotic standards ( ng/μl) by flow injection into the mass spectrometer. The optimization was done to find the optimum fragmentor voltage for each individual compound, and the optimum collision energies for the fragmentation to the quantifier and qualifier ions (Table 4). Chemicals All solvents used were LC/MS grade. Acetonitrile was purchased from J.T. Baker, Germany. Fresh ultrapure water was obtained from a Milli-Q Integral system equipped with LC-Pak Polisher and a. μm membrane point-of-use cartridge (Millipak). All antibiotics standards were purchased from Dr. Ehrenstorfer GmbH, Germany. Calibration Standards A stock solution containing all antibiotics was prepared from the purchased standards to ng/l ( ppt) each in water. The dilution series for determination of the LD, LQ and the calibration curve was, 5,,, 5,,,.5 ppt. The developed MRM method was applied to a ng/l ( ppt) mixture of all standards to identify the retention time of the individual compounds in the final online-spe LC method. From the resulting data file, the dynamic MRM method was developed with a retention time window of ± times the measured peak width around the retention time of each compound. Table 5. Detailed optimized MS parameter for Agilent Jet Stream thermal gradient focusing technology. Triple Quadruple MS method Agilent Jet Stream thermal gradient focusing technology Gas temperature 5 C Gas flow 9 L/min Nebulizer 5 psi Sheath gas temperature 5 C Sheath gas flow L Capillary 4, V Nozzle V Table 4. MRM and Dynamic MRM MS method showing the identified optimum fragmentor and collision energy values for the individual antibiotics as well as for the quantifier and qualifier ions. The retention time was used to develop the dynamic MRM method with a window of ± times the peak width around the compound retention time. Name RT [min] Molecular mass Molecular ion Fragmentor Quantfier ion CE Qualifier ion CE Vancomycin.5,447.4 74.8 44.. 4 Trimethoprim.7 9. 9.4 5.. 4 rmetoprim.88 74.4 75.5 5 59. 4. 4 Cefotaxime.7 455.5 456.6 5. 5 6. 44 Tetracycline.44 444.5 445.6 9 4. 6 54. 8 rnidazole. 9.4.4 8. 8. Chlorotetracycline 4. 478. 479. 5 444. 46. Sulfamethoxazole 5. 5.95 54.6 95 9. 8 8. xolinic acid 6.8 6.6 6.7 85 44. 6 6. 8 Erythromycin 6.79 7.46 74.5 58. 576. 6 Tylosin 7.55 95.5 96.5 55 74. 4 77.4 8 Nalidixic acid 8.9.8.9 85 5. 8 87. 4 Flumequine 8.66 6.8 6.8 75 44.. The molecular ion of vancomycin is [M+H] +, all others are [M+H] + Fragmentor = voltage (V) RT = retention time (min) CE = collision energy (ev) 6
Samples Water samples were taken directly from the Rhine river, from tap water, and from a spring in the region of Karlsruhe, Germany. The water samples were spiked to a final concentration of ppt with a concentrated solution of the antibiotics, vortexed, filtered with Captiva Premium Syringe Filter (.45 µm, 5 mm, p/n 59-59), and injected without further sample prep. Results and Discussion At first, a calibration from ppt to ppt was done for all compounds in the mixture. Within the used group of antibiotics, there is large molecular variety from very polar to nonpolar, small molecules to very large macrocyclic, and polypetidic molecules. For the molecules with very low limits of quantification (LQ) down to ppt, all levels in the calibration curves were used. For compounds with higher LQs, fewer levels were used for the calibration curves (Figure 4). 9 Responses Responses 8 7 6 5 4 Vancomycin y = 8.84* x 8.75778 R =.999 5 5 5 5 4 45 5 55 6 65 7 75 8 85 9 95 Concentration (ng/l) 5 Erythromycin. y = 9.878* x 94.4896.9 R =.999.8.7.6.5.4... 5 5 5 5 4 45 5 55 6 65 7 75 8 85 9 95 Concentration (ng/l) 5 Flumequine y = 477.57* x + 8.9554 R =.9999 H C NH H H H H H C Cl H H N H Cl N H N H N H N H H HN H C H H H NH CH H N CH H C CH H H H C H C CH H N H CH H C H C CH H C H C CH CH H Responses F N CH H 5 5 5 5 4 45 5 55 6 65 7 75 8 85 9 95 Concentration (ng/l) 4 rnidazole. y = 8.97786 * x +.45 R =.9999 Responses.5. N + N H.5 N CH Cl 5 5 5 5 4 45 5 55 6 65 7 75 8 85 9 95 Concentration (ng/l) Figure 4. Calibration curves for selected antibiotics. All compounds were measure from ppt to ppt with seven levels. Depending on the molecule, the LQ was taken at S/N = and levels adjusted accordingly. 7
As LQ, a signal-to-noise (S/N) ratio of (S/N=) was used, and as LD, S/N= was used. Typically, the LD was found one level below the LQ. For instance, flumequine had a LQ of ppt and a LD of ppt. All compounds delivered calibration curves with good linearity coefficients (Table 5). For a statistical evaluation, multiple injections of the 5 ppt level were done. The relative standard deviations (RSD) of the retention time were all below.5%. The RSD of the peak areas were typically below 6%. The recovery of the trapping process was measured for a 5 ppt level injection of the trapping columns in comparison to a 5 ppt injection going directly onto the analytical column. The recovery was typically above 7%, except for the tetracycline compounds. They have lower recoveries, which might be due to weaker bonding on the trapping column material but deliver enough signal intensity for LQ at 5 ppt each. Table 6. verview of calibration data (LD and LQ), recovery of the trapping process as well as statistical evaluation (retention time RSD and peak area RSD) of antibiotics used in this study. Compound RT (min) RT RSD (%) Area RSD (%) LD (ppt) LQ (ppt) Linearity Recovery (%) Vancomycin.5. 4.5. 5..999 76 Trimethoprim.7.5..5..9995 97 rmetoprim.88..95.5..999 95 Cefotaxime.7.4 5.66. 5..9994 99 Tetracycline.44.7 5.64. 5..999 4 rnidazole..7.45...9998 94 Chlorotetracycline 4...6. 5..995 45 Sulfamethoxazole 5..7.. 5..9994 99 xolinic acid 6.8. 4.4. 5..999 8 Erythromycin 6.79.7.66.5..999 6 Tylosin 7.55.9 7.96 5...999 69 Nalidixic acid 8.9..99.5..9994 9 Flumequine 8.66.4....9998 8
Finally, an experiment for the determination of carryover of the most intense polar (hydrophilic) and nonpolar (hydrophobic) compounds was done. An injection of the ppt level of all compounds was done, followed by a blank injection (Figure 5). 4 DMRM of most polar and nonpolar antibiotics at ppt with quantifier and qualifier A ppt rmetoprim Counts ppt Nalidixic acid 4 5 6 7 8 9 Acquisition time (min) MRM of rmetoprim and Nalidixic acid at ppt, only quantifier is shown B ppt rmetoprim ppt Nalidixic acid Counts 4 5 6 7 8 9 Acquisition time (min) Blank after ppt, MRM of quantifiers only, showing carryover of rmetoprim and Nalidixic acid.% Nalidixic acid C.8% rmetoprim Counts 4 5 6 7 8 9 Acquisition time (min) Figure 5. Determination of carryover of the two most intense hydrophilic and hydrophobic compounds. The carryover was determined from an injection of ppt by a following blank. The ppt injection is shown for comparisons. 9
The most intense polar compound was ormethoprim, and the most intense nonpolar compound was nalidixic acid. In the following blank, ormethoprim showed.8% carryover and nalidixic acid showed.%. Both compounds had an LQ of ppt and the carryover was approximately 5% of that (Figure 6). To evaluate the performance in real samples, the antibiotics were spiked in water samples from tap, the Rhine river, and spring water at a concentration of ppt (Figure 6). The responses of the individual signal were in the same order for each compound in the different matrices. The evaluation of the data showed that the area precision RSD is typically below % for all compounds in all matrices (Table 6). The accuracy was typically above 7%. nly one compound, erythromycin, was a little below 5%. rmetoprim Trimethoprim Nalidixic acid Counts Flumequine Erythromycin rnidazole xolinic acid Sulfamethoxazole..5..5..5..5 4. 4.5 5. 5.5 6. 6.5 7. 7.5 8. 8.5 9. 9.5..5 Acquisition time (min) Figure 6. Selected antibiotics from a spiking experiment in Rhine river water, tap water, and spring water. The spiked concentration level of all antibiotics was ppt. Table 7. Statistical evaluation of compounds selected from the spiking experiment shown in Figure 6. Rhine river water Tap water Spring water Average RSD Accuracy Average RSD Accuracy Average RSD Accuracy Compound (ppt) SD (%) (%) (ppt) SD (%) (%) (ppt) SD (%) (%) Trimethoprim 8.9.66.4 94.5 9.4.645. 96.98 7.6.8.6 88.5 rmetoprim 4.9.485. 74.5 5.9.4.9 79.58 4.7.4667.7 7.55 rnidazole..44.9 65. 4.45.77.49 7.5.45.77.5 67.5 Sulfamethoxazole 8.75.7.4 9.7.85.555 4.65 59.5 8.6.6. 9.8 xolinic acid 7.65.. 88.5 9..44.7 96.5 8.5.77.9 9.5 Erythromycin 9.46.495 5. 47. 9.78.485.5 48.88 9.5.99.4 47.55 Nalidixic acid 5.57.4.9 77.85 6.8.4596.7 84. 5.8..4 79.5 Flumequine 9..768.9 95..86.485.7 4.8 9.6.495.5 98.8
Conclusion This Application Note demonstrates the use of the Agilent Infinity Series nline-spe solution for enrichment, separation, and detection of antibiotic residues in trace level analysis of water samples by HPLC using triple quadrupole MS detection. It was demonstrated that LQs as low as ppt could be achieved. The methodology showed a high sample to sample reproducibility with area deviation of less than 6%. The efficient online SPE trapping process allowed detection in real drinking water samples with high precision and accuracy.
www.agilent.com/chem/9 This information is subject to change without notice. Agilent Technologies, Inc., Published in the USA, June, 599-5EN