Screening of Pesticides QuEChERS, LC-MS and Beyond

Transcription

1 Pushing the Limits of Target Analysis and Non-target Screening of Pesticides QuEChERS, LC-MS and Beyond Paul Yang 1, Jon W. Wong 2, James S. Chang 3, and Jennifer Sutton 3 1. Ontario Ministry of the Environment, Etobicoke, Ontario, Canada 2. FDA-CFSAN, College Park, Maryland, USA 3. Thermo Fisher Scientific, San Jose, CA, USA 16 th CVG Annual Meeting, October 5, 2010

2 Outline Targeted analysis and non-targeted screening Limits of using liquid chromatography and mass spectrometry detector (LC-MS) for targeted analysis and non-targeted screening QuEChERS (Quick, Effective, Cheap, Easy, Rugged & Safe) Method performance of targeted analysis using high resolution mass spectrometry (HRMS) detector Using SIEVE as a tool for non-targeted screening Future work 2

3 Targeted and Non-targeted Pesticides Analysis Pesticides Known Knowns Known Unknowns Unknown Unknowns Targeted analysis (routine and non-routine analysis) Non-targeted screening 3

4 What Are Identification Points? (ID Points) Derived dfrom European Commission i Council Directive 96/23/EC, eu/food/food/chemicalsafety/residues/council directive pdf Detailed examples given for its applications in LC-MS analysis ( Identification of residues by LC-MS. The application of new EU guidelines, A.A.M. A Stolker, R.W. Stephany and L.A. van Ginkel, ANALUSIS, 2000, 28, #10, p.947) 4

5 Identification Points (ID points) Legal compounds require 3 ID points while illegal compound require 4 ID points to be qualified as true-positively identified MS technique ID Points earned per ion Low resolution mass spectrometry (LRMS) 1 LRMS n Precursor ion 1 LRMS n Transition products (tandem MS) 1.5 High resolution mass spectrometry (HRMS) 2 HRMS n Precursor ion 2 HRMS n Transition products (tandem MS) 2.5 5

6 Identification of Pesticides Using Liquid Chromatography- Tandem Mass Spectrometry (LC-MS/MS) More than 1,000 pesticides used worldwide, along with their metabolites and degradation products, might be present in food and the environment LC-MS/MS is well established for target multi-pesticide screening with ihh the highest hih selectivity ii and sensitivity, iii and wide linearity range for quantitative results. Two LC-MS/MS multiple reaction monitoring (MRM) transitions can deliver four ID points and has been considered as an effective and rugged analytical approach to detect low level pesticides 6

7 LC-MS and LC-MS/MS in Non-targeted Screening The Limits it from the Outset t Require analytical standard to carry out non-targeted screening An (educated) trial and error process that requires multiple LC-MS or MS/MS runs Will require the use of library spectra or HRMS data to confirm the identity Difficult to obtain quantitative data Analysis of unknown practically not possible The limits has been a moving target since the late 1990s and changes with every new technology! 7

8 Limits in the Late 1990s LC/MS Analysis For example: Hewlett-Packard 1100 single quadrupole LC/MS system Inferior ID points provided d for the analysis Single ion monitoring is vulnerable to interference Limited horse power for multiresidue analysis 8

9 LC/MS Analysis Glyphosate and Glufosinate Glufosinate Glufosinate AMPA AMPA Glyphosate Glyphosate C C 15 N Glyphosate Glyphosate HP 1100 LC-MS system C18 column, from 95:5 to 25:75 A: 5mM CH 3 COONH 4 in H 2 O; B: CH 3 CN Electrospray ionization (ESI) Negative, single ion monitoring of target and qualifier ions for glyphosate (390, 168), glufosinate (402, 180), AMPA (332, 110) and 13 C 15 N-glyphosate (392, 170) Identification Point of 2 AMPA: aminomethylphosphonic h h i acid L. Grey, B. Nguyen, and P. Yang, J. AOAC, 84, #6, 1770 (2001). 9

10 Limits in the Early 2000s (circa 2000~2003) For example: HP 1100-Sciex API 2000 LC-MS/MS system Does require a standard to create a method Gave the opportunity to explore mega methods! Can provide an ID point of three using two Multiple or selected reaction monitoring i (MRM or SRM) transitions MRM/SRM provided good selectivity Limited horse power for multiresidue analysis and may not provide enough data points to define chromatographic peaks correctly (< 100 analytes) 10

11 LC-MS/MS Analysis of Phenyl Ureas, Carbamates and Organophosphates Reconstructed MRM chromatograms of 39 target analytes in a 10-min LC-MS/MS analysis Area Co ounts, Ar rbitrary 12 Phenyl ureas 12 Carbamates 15 Organophosphates p Time (Minutes) Chunyan Hao, Bick Nguyen, Xiaoming Zhao, Ernie Chen, and Paul Yang, J. of AOAC, 93(2) ,

12 Limits in the mid-2000s (circa 2004~2007) Hybrid LC-Tandem MS/Linear Iontrap (LC-qqqLIT) Analysis For example: HP 1100-Sciex API 4000q LC-MS/MS system Does require a standard to create a method Can provide an ID point of three using two multiple or selected reaction monitoring (MRM or SRM) transitions Allows for the collection of spectral library and library search (more ID points but inferior duty cycle time) MRM/SRM provided good selectivity and superb sensitivity Improved horse power for multiresidue analysis to provide enough data points to define chromatographic peaks for about 250 analytes 12

13 Typical MRM/SRM Chromatograms 57 Target Pharmaceutical and Personal Care Products 13

14 QuEChERS Sample Preparation Step 1: Weigh 5-gram of homogenized sample; add 5.0-mL of CH 3 CN 14

15 QuEChERS Sample Preparation Step 2: Add method surrogates, MgSO 4 and NaCl; shake and centrifuge CH 3 CN Extract Plant Matrix Aqueous 15

16 QuEChERS Sample Cleanup Dispersive Solid Phase Extraction ti Step 3: Add primary-secondary amine and MgSO 4 to extract 16

17 LC-MS/MS Chromatogram and Time Program (200 pesticides and Scheduled MRM ~400 transitions) XIC of +MRM (436 pairs): 238.1/163.0 amu Expected RT: 5.7 ID: 3-Hydroxycarbofuran.1 from Sample 13 (5.0 ppb-1) of Data blue b... Max. 1.4e4 cps. Flow rate 0.45 ml/min 2.6e5 Injection volume 20 μl 2.5e5 95% B A: 5 mm NH 4 HCO 2 in 2.4e5 2.3e5 0.1% HCOOH/H 2 O 2.2e5 2.1e5 B: 5 mm NH 4 HCO 2 in 2.0e5 0.1% HCOOH/MeOH OH 1.9e5 1.8e5 1.7e5 1.6e5 1.5e5 14e5 1.4e5 LC time 1.3e5 1.2e5 program 1.1e Time, min Column: RESTEK Ultra Aqueous C18 (100 mm x mm x μm) ) 1.0e5 MS source temp: 500 o C 9.0e4 80e4 8.0e4 Mode: ESI positive 7.0e4 6.0e4 Spray voltage: 5000 volts 5.0e4 Curtain gas: 30 PSI 4.0e4 3.0e4 Collision gas: medium 2.0e4 1.0e % B Gas 1: 50 PSI 0.0 Gas 2: 50 PSI 17

18 Using Information Dependent Acquisition-Enhanced Product Ion (IDA-EPI) Spectrum to Collect Library Spectra Threshold value Data/Information triggered EPI Scan MRM Scan Courtesy of fthermofisher h 18

19 IDA-EPI Data Processing 19

20 Typical Library Search Result 20

21 Limits in Known Unknown Analysis Using LC-qqqLIT MS Is time consuming and must have a standard to start with Requires the use of many injections to identify unknowns in a sample May run out of sample before the identification Faster LC, like fast gas chromatography (GC), will produce narrower peakwidths Still needs at least 10 data points to define a chromatographic peak for quantitation Increasingly difficult for the mass spectrometer t hardware to handle the overhead required for MRM, EPI spectrum and correctly define peaks Retention times shift Available analytical standards to carry out the analysis, especially standard mixes in mega-method, becomes the biggest challenge 21

22 Two Viable Alternatives? LC using a quadrupole tof-ms (qtof) or Orbitrap MS as a Detector Scanning speed tof-ms can scan 20 < 30,000 resolution, practical for most routine LC experiments Orbitrap MS can scan 2 50,000 resolution, practical for a 2.1 mm LC column Accurate mass measurement will give one more level of confirmation of results, better data quality Product ion scan followed by full spectral library search will provide unambiguous identification of a suspicious component 22

23 Reconstructed LC/qTOF-MS Total Ion Chromatogram 23 qtof: quadrupole time-of-flight flight

24 LC/qTOF-MS Analysis of 39 Pesticides 32 of the 39 targets confirmed with mass accuracy within ±10 ppm 24

25 Un-identifiable Ions Root Cause Analysis van der Heeft E, Bolck YJ, Beumer B, Nijrolder AW, Stolker AA, Nielen MW. J Am Soc Mass Spectrom. (2009) 20(3):

26 Exactive Benchtop LC-MS An Orbitrap MS Based LC-MS Mass Resolution (R) 10,000 to 100,000, variable Mass accuracy Sub ppm Sensitivity 500 fg Buspirone with S/N >10:1 Dynamic range >10, within a spectrum Scan speed Up to 10 scans per second Mass range m/z Polarity switching One positive and one negative scan < 1 second (25K Resolution) 26

27 Can Orbitrap MS Provides a Viable Solution for Non-targeted t dknown Unknown Screening? Validation of Method Parameters 181 pesticides concentrations of 25, 100 and 250 ppb (mg/kg) were prepared in three matrices (oranges, spinach and apples) Target analytes introduced using ultra high pressure liquid chromatography (UHPLC) Replicates (4) analysis at mass resolution (R) of 10,000 (10k), 25k, 50k and 100k 27

28 Recovery Studies of Spinach, Oranges and Apples 25 µg/kg % Recovery Spinach Orange Apple 100 µg/kg % Recov very Spinach Orange Apple 250 µg/kg % Recover ry Pesticide ID Number Spinach Orange Apple 28

29 Relative Standard Deviations (RSD %) of Spinach, Orange and Apple (N = 7) 25 µg/kg 100 µg/kg 250 µg/kg % RSD % RSD % RSD Pesticide ID Number Spinach Orange Apple Spinach Orange Apple Spinach Orange Apple 29

30 Method Detection Limits, n = 99% confidence (Spinach, Orange, Peach, Apple) 25 MDL (ng/g) Spinach Orange Peach Apple Pesticide ID Number Federal Register, U.S. Code of Federal Regulations, Part 136, Appendix B, 49 FR 43430, Oct 26, 1984; 50 FR 694, 696, Jan 4, 1985, as amended at 51 FR 23703, June 30,

31 Resolving a Pesticide Mixture of Similar Masses Molecular Isobars Pymetrozine C 10 H 11 N 5 O Pyracarbolid C 13 H 15 NO Ethoxyquin C 14 H 19 NO R = 10,000 Pymetrozine Pyracarbolid Ethoxyquin Estimated R (R est ) required for baseline separation 62,388 23,868 R = 25,000 R = 50, R = 100,000 31

32 Resolving [M + H] + of Isazophos ( ) and Triazophos ( ) (R est ~53,690) Isazophos Ti Triazophos R s = 10, R s = 25,000 R s = 50,000 R s = 100, Isazophos C 9 H 17 ClN 3 O 3 PS Triazophos C 12 H 16 N 3 O 3 PS

33 Aldicarb sulfone C 7 H 14 N 2 O 4 S, MW = Accurate Mass EIC 1ppb in Spinach R= 10,000 R= 10,000 R= 25,000 R= 25,000 R= 50,000 R= 50,000 R= 100,000 R= 100,000 Stick Plot Point to Point Plot Stick Plot Point to Point Plot 33

34 Bitertanol C 20 H 23 N 3 O 2, MW = ppb in Spinach 1ppb Standard Matrix interference! R= 10,000 R= 25,000 R= 50,000 R= 100,000 Stick Plot Point to Point Plot Stick Plot Point to Point Plot 34

35 Clofentezine C 14 H 8 C l2 N 4, MW = ppb in Spinach 1ppb Standard R= 10,000 R= 25,000 R= 50,000 R= 100,000 Stick Plot Point to Point Plot Stick Plot Point to Point Plot 35

36 10 ng/g Mass Ac ccuracy (ppm ) Mass Accuracy of Pesticides in Different Matrices at 10 and 100 ng/g Short Term Mass Axis Stability: ± ppm, N=2,495, 10 days ng/g Pesticide ID Number Hazelnut Raisin Wheat flour Orange Spinach Mass Acc curacy (ppm) Short Term Mass Axis Stability: ± ppm, N=2,670, 10 days Pesticide ID Number Hazelnut Raisin Wheat flour Orange Spinach 36

37 The Limits? The method can be good for targeted and non-targeted analysis if we know what we are looking for (known unknown) Ionization source of Orbitrap has good universality The pro: any organic acids/bases will be fragmented and appear as a chromatographic peak in the extracted ion chromatogram The con: One too many peaks to examine and data interpretation has become the barrier How can we do an unknown unknown analysis? 37

38 What is SIEVE? SIEVE TM, XCMS TM, Masstrix TM, MZMINE TM, MetID TM, to name a few, are post data processing software developed by various vendors for metabolomic studies 38

39 Unalign vs Align Unaligned Aligned 39

40 Frame Selection (Time window with Accurate Mass) 40

41 Frame Report, RT4.205 m/z Mthi Methiocarb Monoiotopic+H Aligned Chromatogram 41

42 Gernerate Formula from Monoisotopic Mass ppm Mass Accuracy 10 ppb Right click on mass Then select Generate formula form mass 100 ppb 1000 ppb 42 42

43 Multilevel Methiocarb Spikes in Spinach Matrix 10, 100, 1000 ppb 43

44 Methiocarb Spiked Spinach Frames and dchemspider Search Report Matrix Blank 44

45 Principle Component Analysis (PCA): Spinach Methiocarb Outlier identified by PCA 45

46 Methiocarb ChemSpider Search Report: Corn 46

47 PCA Analysis: Corn Methiocarb Matrix Blank 47

48 ChemSpider Search: Orange Matrix Blank 48

49 PCA Analysis: Orange 49

50 Known Unknown Analysis (Test #1) Incurred sample Navel 50

51 Incurred Orange Sample Navel Imaverol 51

52 Incurred Orange Sample Navel, PCA Incurred Sample Matrix Blank 52

53 Isotope Ratio Actual vs. Simulation Mass Accuracy ppm Imaverol Actual 53 Simulation 53

54 Known Unknown Analysis (Test #2) Incurred samples, Clementine Incurred clementine samples Pesticide (ng/g) Carbendazim 1.0±0.2 Imazili 365±57 Thiabendazole 64±6 54

55 Matrix Blank vs. Navel vs. Clementine 55

56 Workflow of Unknown Unknown Analysis True outlier: Outlier in the training i set Unknown: Outlier not in the training set Outlier identified by PCA 56

57 Conclusions and Future Work Initial studies showed that Orbitrap and Sieve can do target and non-target identification automatically at a concentration at above 100 ppb and a mass resolution of 50, or greater Pending on data exchange ability, parallel studies with new generation LC-tof- MS Development of a good/extensive training set: A library (training set) of LC-Orbitrap MS runs of identical parameter (user contributed or in house) Replicate sample analysis to achieve over-determination for the training set Explore the ability of Orbitrap and Sieve in unknown identification Mathematically, principal component(s) identified by frame, confirmed by PCA bt but not tin the training ii set can be extracted td Characterization of the high score outliers shall result in the identification of unknowns 57

58 Acknowledgements Ginseng Board of Wisconsin U.S. Environmental Protection Agency National Pesticide Standard Repository US U.S. Food and ddrug Administration i i Field laboratories National Institutes of Health Office of Dietary Supplements IAG Y1-OD OD

59 Teşekkürler Thank You! Dziekuje 谢谢 감사합니다 Köszonom For Your Attention! 59