Interlaboratory Comparison of a General Method To Screen Foods for

Dec 9, 2013 - An interlaboratory comparison of a multipesticide residue analytical method is reported. The goal of the comparison was to evaluate the ...
4 downloads 0 Views 653KB Size
Article pubs.acs.org/JAFC

Interlaboratory Comparison of a General Method To Screen Foods for Pesticides Using QuEChERs Extraction with High Performance Liquid Chromatography and High Resolution Mass Spectrometry Brian D. Eitzer,*,† Walter Hammack,‡ and Michael Filigenzi§ †

Department of Analytical Chemistry, The Connecticut Agricultural Experiment Station, 123 Huntington Street, New Haven, Connecticut 06511, United States ‡ Florida Department of Agriculture, 3125 Conner Blvd, Tallahassee, Florida 32399-1650, United States § California Animal Health and Food Safety Laboratory, University of California at Davis, West Health Sciences Drive, Davis, California 95616, United States S Supporting Information *

ABSTRACT: An interlaboratory comparison of a multipesticide residue analytical method is reported. The goal of the comparison was to evaluate the potential for liquid chromatography/high resolution mass spectrometry along with a specific automated screening procedure to allow the determination of the presence or absence of a set of targeted compounds without additional manual review. The method utilized an off the shelf QuEChERs based extraction followed by analysis with an orbitrap mass spectrometer with the data evaluated by ToxID. The method was tested at three laboratories, with three produce matrices (spinach, carrots, and oranges), and three levels of spiked pesticides with all analyses in triplicate. A series of 247 compounds were tested, and it was found that the three laboratories produced consistent data; however, manual review was still necessary. The data was shown to have no false negatives for 211 compounds in the three produce matrixes at 200 ppb. Of these 211 compounds, 189 had no false negatives at 50 ppb, and 129 had no false negatives at 10 ppb. The HRMS method was shown to be robust with similar data being achieved by all three laboratories and detectable concentrations only slightly above the range shown for triple quadrupole MS/MS. KEYWORDS: QuEChERs, pesticide residue, interlaboratory comparison, high resolution mass spectrometry



INTRODUCTION

The extracts produced in these multiresidue techniques have been commonly used along with liquid chromatography tandem mass spectrometry. These combined techniques lead to reports of the determination of pesticide residues in produce, wine, mango, watermelon, tea, and many other foods.5−10 These quantitative methods require that the mass spectrometric procedures be targeted for a specific set of pesticides prior to analysis, so that the scan functions can be optimized for the selected reaction monitoring of each individual pesticide. One drawback of this optimization can be a limitation on the number of pesticides monitored at similar retention times due to the duty cycle of the instrument (the number of different scan functions that can be cycled through while still getting sufficient points across each chromatographic peak) Untargeted compounds will not be determined. More recently the use of high resolution mass spectrometry (HRMS) rather than tandem mass spectrometry has been explored. Various instrumental configurations have been used: an orbitrap for the analysis of 350 pesticides in honey;11 pesticide residues in fruits and vegetables with an orbitrap12,13 and a quadrupole orbitrap;14 organic contaminants in foods using a quadrupole time-of-flight MS;15 a report of pesticide residues in ginseng and spinach using several different HRMS

It has become evident that there are a multitude of potential contaminants in foods. The vast numbers of compounds include hundreds of different pesticides, mycotoxins, industrial compounds, and their byproducts. Insuring food safety therefore increasingly requires the use of methodologies that can monitor for hundreds of chemicals at the same time. These multiresidue methods combine a single extraction technique with analytical instrumentation capable of determining numerous compounds. Screening methods that can relatively quickly examine hundreds of compounds at relevant concentrations (i.e., below tolerance levels) have their place along with more rigorous quantitative methods. It has been noted that for screening to be effective for hundreds of compounds automated library based procedures will need to be implemented.1 Extraction protocols for multiresidue methods have been investigated by many researchers, with much of the recent efforts focused on modifications to the QuEChERS procedure as originally described by Anastassiades et al.2 Versions with various buffers have been compared by Lehotay et al.3 and a modified QuEChERS version was compared with aqueous acetonitrile extraction and pure acetonitrile extraction by Lacina.4 Both groups report that the various different versions of QuEChERS perform reasonably well for most pesticides with the differences between the various procedures most pronounced for a couple of pH dependent pesticides. © 2013 American Chemical Society

Received: August 16, 2013 Accepted: December 9, 2013 Published: December 9, 2013 80

dx.doi.org/10.1021/jf405128y | J. Agric. Food Chem. 2014, 62, 80−87

Journal of Agricultural and Food Chemistry

Article

configurations.16 One focus has been on how much mass resolution is necessary. Recent efforts have postulated that complex matrices require resolution greater than 50000 (m/Δm),1 and in a comparison of HRMS to tandem MS, it was shown that above 50000 resolution, the selectivity of HRMS is greater than standard tandem MS.17 In a later report, Kaufmann et al. stated that while “sensitivity was slightly inferior ... HRMS has advantages that will most likely never be met by MS/MS”.18 Among the advantages is the nontargeted nature of full-scan data acquisition. Signals for virtually all ions within the selected mass range are acquired as opposed to only those targeted for acquisition as occurs with multiple reaction monitoring in MS/ MS. This makes it possible to analyze data retrospectively for analytes that were not considered at the time of acquisition. Potential drawbacks of the orbitrap based methods are some ion suppression issues that can occur in the c-cell of these instruments12,19 and the need for robust automated methods to handle the abundance of data produced.1,13 Many of these recent efforts utilizing both time-of-flight and orbitrap platforms have been reviewed with the conclusion that these systems are “attractive for residue analysis in food”.20 It is, however, important to show that these methods are robust and rugged with results that agree across different users. This robustness can be best shown by comparing results on the same samples from different laboratories. We report here on such an interlaboratory comparison using an orbitrap HRMS to screen for pesticide residues in three diverse food matrices: spinach, a leafy green; oranges, watery fruit; and carrots, a root crop. The goal of the project was not the development of a full quantitative method but rather an examination of the utility of HRMS along with an automated data processing program to relatively quickly screen produce for the presence of pesticide residues. Our intent was not to investigate the method but to evaluate current procedure. We therefore chose to only utilize the 100000 resolution achievable by the orbitrap. As mentioned previously there are many versions of the QuEChERs extraction protocol, all of which do reasonably well for most compounds. We therefore chose to mimic a recently published validation of a QuEChERS LC-tandem mass spectrometric procedure10 using the same matrices and extraction protocols. These choices would allow us to examine the utility of the HRMS screening procedure without going through optimization of the extraction or analysis protocols. To further examine the utility of the procedure, we chose to use a commercially available set of 247 pesticides, none of which were preselected to be amenable to the chosen QuEChERS procedure and LCMS. Although this set of pesticides was not the same as the procedure that we were mimicking there was a sufficient overlap that we could compare our HRMS results with their tandem MS results. All samples were prepared at a single laboratory, frozen, and shipped to the other two laboratories. At each laboratory, the samples were spiked with the set of standards, extracted using the modified version QuEChERs, and analyzed by LC/HRMS on an orbitrap mass spectrometer.



Materials. Prefilled QuEChERs extraction tubes containing 4 g of MgSO4 and 1 g of NaCl (part no. ECMSS50CT, UCT, Bristol, PA) and prefilled QuEChERs extraction tubes containing 50 mg of PSA and 150 mg of MgSO4 (part no. CUMPS2CT, UCT). Standards. A mixed pesticide standard set was obtained from Accustandard (part no. S-21974-set, New Haven, CT). This set consists of 13 1 mL ampules each containing 16−22 pesticides at 100 μg/mL, for a total of 247 compounds. The stock standard was prepared by combining 400 μL from each of the ampules in a single 10 mL volumetric flask, and the final volume was brought to 10 mL with acetonitrile for a final concentration of 4 μg/mL. A diluted stock solution was prepared by bringing 2.5 mL of the stock solution to 10 mL with acetonitrile for a final concentration of 1 μg/mL. A 5 ppm solution of triphenylphosphate and d-6 malathion in acetonitrile was prepared as an internal standard. Sample Preparation. Three different matrices were prepared. One pound or more of carrots, oranges, and spinach were chopped in a Hobart food chopper without peeling. The mixture was then thoroughly blended in a Robot Coupe, model R2 equipped with a 3 quart stainless steel bowl. The sample homogenate was frozen in 18 ounce Nasco whirl-pak bags with each bag containing about 250−300 g each. Bags of each matrix were shipped overnight to the other two laboratories for all subsequent work. At each laboratory, 12 10 g samples of each homogenized commodity were placed into 50 mL centrifuge tubes. Three tubes were left as unspiked matrix samples, three were spiked with 100 μL of the diluted stock standard (10 ppb in produce), three with 500 μL of the diluted stock standard (50 ppb in produce), three with 500 μL of the stock standard (200 ppb in produce). All were spiked with 100 μL of the internal standard solution. A reagent blank consisting of 10 mL water was also prepared. Acetonitrile was added to each of the tubes so that the total volume of acetonitrile added (spikes plus acetonitrile) was 10 mL. A ceramic stone was added to each tube, and the sample was vigorously shaken for 1 min. A UCT salt packet containing 4 g of MgSO4 and 1 g of NaCl was then added, and again the samples were shaken for 1 min. The sample was centrifuged for 5 min at ≥3000 rpm. A 1 mL aliquot was transferred to a tube containing 50 mg of PSA and 150 mg of MgSO4. The sample was shaken for 1 min. The sample was centrifuged for 5 min at ≥3000 rpm. A 0.5 mL aliquot was diluted with 0.5 mL of the aqueous mobile phase (water with 0.1% formic acid), and the sample was filtered into a microvial for LC-HRMS analysis. Instrumental Analysis. Samples were analyzed using an Agilent (Santa Clara, CA) 1200 series rapid resolution liquid chromatograph interfaced to a Thermo Fisher Scientific (West Palm Beach, FL) Exactive orbitrap mass spectrometer. The LC was operated using a Thermo Fisher Scientific Hypersil Gold aQ C-18, 100 mm × 2.1 mm, 1.9 μm column held at 40 °C. The mobile phase flow rate was 0.25 mL/min, going from 99% water with 0.1% formic acid to 95% acetonitrile with 0.1% formic acid from 1 to 10 min followed by a 5 min hold at 95%, with a 5 min re-equilibration step for a total run time of 21.5 min. Three microliters of extract was injected. HRMS data was acquired in the positive electrospray mode using a metal needle at 3.5 kV, a source capillary temperature of 300 °C and mass resolution of 100000 (as specified in the tune page), the balanced mass accuracy setting, and an injection time of 20 ms. Data was acquired over the mass range of m/z = 75−1500. Prior to analysis, the mass spectrometer was tuned by infusion of azoxystrobin into the system. The tube lens, skimmer, and other parameters were adjusted so that the relative abundance of m/z 372 compared with m/z 404 was less than 10%. This tuning step was performed in order to ensure that any fragmentation of analytes through the tube lens and skimmer was minimized. A single tune file from each laboratory was used for the entire set of analyses though a mass calibration was performed prior to the analysis of each sample matrix. It should be noted that the final extract concentration was 0.5 g sample/mL extract. At the 10 ppb concentration using a 3 μL injection, this would result in a total of 15 pg on column. While these low levels may help reduce ion suppression, they also can be a challenge to detect.

MATERIALS AND METHODS

Reagents. The three laboratories in this study may have had different suppliers for some of the reagents so these are listed by specification. Water was supplied as 18.2 MΩ water from a Millipore Milli-Q source or purchased as HPLC grade. Acetonitrile was HPLC grade. When used for the mobile phases, the water and acetonitrile were either fortified at 0.1% with ACS reagent grade formic acid or purchased as prefortified HPLC grade reagents. 81

dx.doi.org/10.1021/jf405128y | J. Agric. Food Chem. 2014, 62, 80−87

Journal of Agricultural and Food Chemistry

Article

Figure 1. Mass profile for an infusion of thiobendazole and carbaryl on the orbitrap mass spectrometer at a resolving power of 100000. The theoretical m/z of the (M + H)+ ions are 202.04334 and 202.08626, respectively. Data Analysis. The raw pesticide data from the instrument was analyzed using the ToxID software package (Thermo Scientific). This program required an input file listing the molecular formula for each pesticide of interest, along with its expected retention time. During the setup of the method, the stock solution of 247 standards was injected on the instrument to determine retention times. Five compounds had distinctively split peaks and were entered into the input file twice (with separate retention times), while six compounds were not detected in the positive electrospray mode, thus in total there were 246 entries in the input file used for these tests. During operation of the software, the raw exact mass data for each entry was compared with the theoretically expected data (as computed from the entries molecular formula) for each of the individual pesticides in the input file. User parameters were set for detection at an elution time of ±0.25 min with a mass accuracy of ±5 ppm. The response threshold was set by each laboratory and had a slight variability depending on instrument but was in the range of 1000- 5000 counts. The ToxID data output for samples, spiked samples, and method blanks were reviewed to determine if analytes were found correctly and identified in amounts significantly greater (3x) than that observed in the method blanks. Safety. Pesticides are considered hazardous materials. In accordance, all laboratory procedures were conducted with good chemical handling practices, and waste was disposed of as per the laboratory’s hazardous waste plan.

and thiabendazole (C10H7N3S) have the same nominal mass of 201, the theoretical exact m/z’s of their (M + H)+ ion are 202.08626 and 202.04334, respectively. While these ions would be indistinguishable on unit resolution mass spectrometers, they are easily differentiated under high resolution conditions with a required resolution (m/Δm) of only 4700. This is clearly demonstrated in Figure 1, which shows the infusion (no chromatography) of these two compounds into the orbitrap system used in the current work. The protonated molecular ions of the two compounds are quite distinct. The work herein described was performed at a mass resolution of 100000. Isomeric compounds, having identical formulas, cannot be distinguished by differences in the exact mass of the molecular ion. However, these compounds can often still be distinguished by differences in relative retention time through the chromatographic system. Alternatively they can be differentiated by collisional fragments generated in the HCD cell of the instrument. These experiments could be conducted on the described instrumentation, but that is the subject of future work. Interlaboratory Comparison. Data was reviewed for each individual pesticide at each level in each matrix. The number of detections (out of three samples) for each combination (matrix−pesticide−level) was recorded. All data from the three laboratories is included in the Supporting Information. The number of detections for the same combination was then compared across the three laboratories and is shown in Table 1. As can be seen in the table, the three different laboratories were very consistent in their ability to detect the pesticides. At the lowest level over 70% of the time either all three laboratories found a pesticide the same number of times in a particular matrix or did not see it at all in that matrix. By the highest concentration level the consistency in the data was almost at 90%. Clearly the entire method showed good consistency across the three laboratories.



RESULTS AND DISCUSSION Specificity. Individual pesticide residues were identified in samples through two fundamental characteristics of the pesticide. The first was that the retention time for the pesticide must have been within 0.25 min of the time observed for an authentic standard of that pesticide. The second and more important characteristic was that the exact m/z of the pesticide must have been within 5 ppm of the calculated theoretical value. As an example of the specificity that is provided by an exact mass instrument consider the case of the two pesticides, carbaryl and thiabendazole. Although carbaryl (C12H11NO2) 82

dx.doi.org/10.1021/jf405128y | J. Agric. Food Chem. 2014, 62, 80−87

Journal of Agricultural and Food Chemistry

Article

Reagent Blanks. With each set of samples, a reagent blank (using clean water and all other reagents including the internal standards) was analyzed. These samples were processed using the same ToxID protocol as used for all samples. A total of 43 of the compounds on the entry list were reported as a positive

hit in at least one reagent blank sample by the ToxID software. Upon further inspection almost all of these “detections” appeared to be the result of software integrating noise. This points out the need for user accessible data integration parameters (other than intensity threshold) in the ToxID software. In addition, this also indicates that there needs to be a manual review of data. Unspiked Matrix. The analyzed samples were purchased at the supermarket with no a prioi knowledge of pesticide use. Therefore, the unspiked matrix samples were examined for the presence of any of the pesticides in the spiked mixture. Each laboratory extracted a set of three samples of each matrix without any spikes other than the internal standards. Three incurred residues were identified in the oranges (imazalil, pyriproxifen, and thiadendazole); two residues in the spinach (imidacloprid and spinosad), and none in the carrots. For these residues, in all cases there were good distinct peaks that appeared to be real in all nine samples analyzed of the unspiked matrix. There were a number of additional compounds tentatively identified by ToxID in the unspiked matrix that on closer inspection did not look like real peaks. These pesticides would be considered as false positves and are summarized in Table 2. There are two ways to consider the false positive rate: either (number of pesticides seen at least

Table 1. Comparison of the Percentage of Times That the Different Laboratories Agree on the Detection (Or Nondetection) At the Specified Concentration Level concn level

% 0 of 3 agree

% 2 of 3 agree

% 3 of 3 agree

10 50 200

5.65 2.0 1.6

22.6 14.6 9.5

71.8 83.3 88.8

Table 2. The Number of Pesticides with the Specified Number of False Positive Hits in Each Matrixa matrix

1 hit

2 hits

3 hits

4 hits

5 hits

6 hits

7 hits

8 hits

9 hits

oranges carrots spinach

13 5 11

6 2 6

11 10 13

3 3 3

1 0 1

4 0 2

0 0 0

1 0 0

1 1 1

a

Each unspiked matrix was analyzed three times by each laboratory so the maximum number of hits is nine.

Figure 2. Selected ion chromatogram (5 ppm window) around (M + H)+ ion for aldicarb sulfoxide spiked into oranges at 10 ppb (top) and 50 ppb (bottom). ToxID identifies a peak in both samples; however, the compound was not considered to be identified in the upper sample upon review. 83

dx.doi.org/10.1021/jf405128y | J. Agric. Food Chem. 2014, 62, 80−87

Journal of Agricultural and Food Chemistry

Article

once)/(the total number of pesticides, 247) or (the total number of hits)/(total possible hits, 2223). Thus, the false positive rates for the three matrixes are as follow: oranges 16% and 5.2%, carrots 9% and 2.7%, spinach 15% and 4.5%. Although these rates are low, they again point to the need for additional data review beyond the library program. False Negatives. A false negative occurs if a residue that should be reported was missed. To give a true assessment of the false negative rates, the data needed to be adjusted for positive detections based on the integration of noise. To do this, any pesticide that had a ToxID reported detection in either an unspiked matrix or reagent blank was carefully examined in the lowest spiked samples. In a few cases, the ToxID program integrated a noise peak at the 10 ppb level and then identified the compound as having been found. An example of this noise integration is shown in Figure 2 (top); however, in all of these cases, when the compound was examined at the 50 ppb level, the peak was present at sufficient signal-to-noise ratio to be correctly identified, Figure 2 (bottom). In these cases, the detections for the 10 ppb sample were manually changed to nondetects prior to the analysis of false negatives. The false negative rate for each pesticide at each concentration level is defined by eq 1, where F is the false negative rate, N is the number of missed detections at a given concentration, and D is the number of possible detections at a given concentration. For most pesticides, D is 27, but for those pesticides for which there was an incurred residue, the data was adjusted to only use the matrixes without that incurred residue (D = 18 in those cases). F = 100 × N /D

Table 4. Comparison of the Minimum Concentration with Zero False Negatives with the Tandem MS MDL as Listed by Sack et al.10 for the Same Compounda

Table 3. Number of Pesticides with the Specified False Negative Rate Using the ToxID Software at Each Concentration 0%

0−25%

25−50%

50−75%

75−100%

129 189 211

34 18 9

30 16 13

16 14 11

37 9 2

acephate

50

4

acetamiprid

50

5

acibenzolar S-methyl

50

8

b

8

aldicarb

50

6

aldicarb sulfoxide

50

4

aminocarb

10

4

azoxystrobin

10

4

benalaxyl

10

5

bendiocarb

50

4

benfuracarb

50

3

benzoximate

10

6

bifenazate

200

5

bitertanol

10

6

boscalid

50

6

bromuconazole-cis

50

7

bromuconazole-trans

50

7

bupirimate

10

7

buprofezin

10

5

butafenacil

50

5

butocarboxim

50

9

b

6

carbaryl

50

5

carbendazim

10

5

carbetamide

10

4

carbofuran

10

4

carbofuran-3OH

50

5

200

5

chlorfluazuron

50

6

chloroxuron

10

7

clethodim

50

5

clofentezine

200

5

clothianidin

200

7

cyazofamid

50

8

cycluron

10

5

cyflufenamid

10

5

butoxycarboxim

The false negative rate for each pesticide at each concentration is included in the Supporting Information. The data is summarized in Table 3 showing the number of pesticides

10 50 200

tandem MS MDL (ppb)

alanycarb

(1)

concn

minimum concentration with zero false negatives

compound

chlorantraniliprole

within a specified false negative percentage range. The method was tested with a mixed pesticide standard containing 247 different pesticides along with two internal standards. Seven compounds did not respond on the LC/MS system while five others gave split peaks creating a total of 246 distinct ToxID entries to examine for false negatives. At the 10 ppb spiking level, just over half (129) of the pesticides were seen in all spiked samples, with this number increasing to about 75% (189) at 50 ppb, and 85% (211) at 200 ppb. This table also shows that there were only a couple of truly problematic compounds that had false negative rates of over 50% at the higher concentration level. Several of these compounds also had poor (though) detectable response in the standards so likely lacked sensitivity to be detected at lower concentrations in matrix. It is also possible that a couple of these pesticides may have had problems with the chosen extraction procedure, though this was not evaluated. It should be noted that this version of QuEChERS has an overall dilution of the sample by a factor of 2. Some of the other versions of QuEChERs result in

b

5

cyproconazole

10

7

cyprodinil

10

6

cyromazine

b

4

desmedipham

50

5

diclobutrazol

10

7

dicrotophos

10

4

diethofencarb

10

5

difenoconazole

10

4

diflubenzuron

b

5

dimethoate

10

4

dimethomorph

50

6

dimoxystrobin

10

5

dinotefuran

50

5

b

6

cymoxanil

dioxacarb diuron doramectin 84

50

6

200

13

dx.doi.org/10.1021/jf405128y | J. Agric. Food Chem. 2014, 62, 80−87

Journal of Agricultural and Food Chemistry

Article

Table 4. continued

Table 4. continued minimum concentration with zero false negatives

tandem MS MDL (ppb)

emamectin benzoate B1a

10

6

monolinuron

emamectin benzoate B1b

200

6

moxidectin

ethiofencarb

10

4

ethiprole

50

ethirimol

10

compound

compound

minimum concentration with zero false negatives

tandem MS MDL (ppb)

10

4

200

8

myclobutanil

10

7

7

neburon

10

5

3

nitenpyram

50

5

etoxazole

10

4

novaluron

b

7

fenamidone

10

6

nuarimol

10

7

fenazaquin

10

5

omethoate (dimethoate oxon)

10

4

fenbuconazole

10

6

oxadixyl

50

6

fenhexamid

50

6

oxamyl

50

4

fenobucarb

10

4

paclobutrazol

10

8

fenoxycarb

10

4

phenmedipham

50

5

fenpyroximate

10

4

picoxystrobin

10

6

fenuron

10

5

piperonyl butoxide

10

5

flonicamid

50

6

pirimicarb

10

5

flubendiamide

50

9

prochloraz

10

4

flufenoxuron

200

5

promecarb

10

5

fluometuron

10

5

propamocarb

10

4

fluoxastrobin

10

5

propargite

b

4

flusilazole

10

5

propiconazole

10

6

flutolanil

10

6

propoxur

10

3

flutriafol

10

6

pyracarbolid

10

4

forchlorfenuron

10

6

pyraclostrobin

10

4

formetanate

10

5

pyridaben

10

5

fuberidazole

10

4

pyrimethanil

10

6

furathiocarb

10

5

pyriproxyfen

10

5

halofenozide

50

5

rotenone

10

6

hexaflumuron

b

9

siduron

10

9

10

4

hexythiazox

50

4

spinetoram

hydramethylnon

50

5

spirodiclofen

50

17

5

spiromefisen

50

7

6

spirotetramat

50

6

200

7

imazalil imidacloprid

10 50

indoxacarb

50

8

spiroxamine

ipconazole

10

7

sulfentrazone

b

10

4

tebuconazole

10

7

3

tebufenozide

10

6

tebuthiuron

10

4

b

9

iprovalicarb isoprocarb

10 10

isoproturon

10

5

isoxaflutole

b

30

teflubenzuron

ivermectin

temephos

200

10

b

6

kresoxim-methyl

50

6

thiabendazole

10

5

linuron

50

6

thiacloprid

50

5

malathion

50

7

thiamethoxam

50

6

mandipropamid

10

6

thidiazuron

10

7

mepanipyrim

10

6

thiophanate-methyl

metaflumizone

b

7

metalaxyl

10

metconazole

10

200

5

triadimefon

10

6

5

triadimenol

10

9

6

tricyclazole

10

3

200

4

trifloxystrobin

10

4

methiocarb

50

5

triflumizole

10

5

methomyl

50

5

triflumuron

200

6

methoxyfenozide

10

6

triticonazole

10

7

metobromuron

50

4

vamidothion

10

5

mevinphos

50

6

zoxamide

50

7

mexacarbate

10

6

monocrotophos

10

4

methamidophos

a

Average of two transitions used. bNever reached, there was always at least one false negative.

85

dx.doi.org/10.1021/jf405128y | J. Agric. Food Chem. 2014, 62, 80−87

Journal of Agricultural and Food Chemistry

Article

Table 5. List of Compounds Tested for Which a Tolerance Has Been Established for One of the Matrixes Studied along with the Minimum Concentration with Zero False Negativesa compound acetamiprid acibenzolar S-methyl aldicarb aldicarb sulfone aldicarb sulfoxide azoxystrobin boscalid buprofezin carbaryl carfentrazone-ethyl chlorantraniliprole clethodim clothianidin cyazofamid cyprodinil desmedipham diazinon difenoconazole dimethoate dinotefuran diuron emamectin benzoate B1a emamectin benzoate B1b EPTC (eptam) ethofumesate fenamidone fenazaquin fenbuconazole fenhexamid fenpyroximate flonicamid flubendiamide flufenoxuron a b

minimum zero false negative concn (ppb)

carrots tolerance (ppb)

50 50 50 50 50 10 50 10 50 b 200 50 200 50 10 50 10 10 10 50 50 10

500 1000 2000

1000 800 90 750 750

oranges tolerance (ppb)

3000 1000

500

1000 35 000 22 000 100 13 000 2000 3000 9000 30 000 6000 700

300 300 300 10 000 1600 2500 10 000 100 1400

8000

600 2000 5000 50 100

200 10 b 10 10 10 50 10 50 50 200

spinach tolerance (ppb)

100 100 7000 150 500 1000 30 000 600 600

9000 11 000 300

compound

minimum zero false negative concn (ppb)

hexythiazox imazalil2 imidacloprid3 indoxacarb linuron malathion mandipropamid metalaxyl methidation methomyl methoxyfenozide metribuzin mevinphos myclobutanil norflurazon oxamyl phenmedipham piperonyl butoxide prometryn propiconazole pymetrozine pyraclostrobin pyridaben pyrimethanil pyriproxyfen spinetoram spinosad5 spirodiclofen spiromefisen spirotetramat tebufenozide thiabendazole thiamethoxam trifloxystrobin triflumizole

50 10 50 50 50 50 10 10 b 50 10 10 50 10 50 50 50 10 10 10 200 10 10 10 10 10 10 50 50 50 10 10 50 10 10

carrots tolerance (ppb)

400 1000 8000 500 200 100 300 30

spinach tolerance (ppb)

3500 14 000 8000 20 000 10 000 20030 000

oranges tolerance (ppb) 350 10 000 700

8000 10 000 4000 2000 10 000

1000 30 200 3000

100 4000

8000 450 250 400

5000 600 29 000

150 100 100

3000 8000 8000 12 000 9000 10 000

50 100

4000

2000 500 10 000 300 300 300 500 600 800 10 000 400 600

3500

21

Tolerance values from Pesticide Chemical News Guide, Arlington VA; also available from Code of Federal Regulations Title 40, 24, Part 180.22 Never reached, there was always at least one false negative.

over 50% of these 161 pesticides had a zero false positives at a concentration on the same order of magnitude as the reported MDL (i.e., minimum concentration of 10 vs an MDL of 5). There were clearly some pesticides that worked much better on the tandem MS system; further investigation is needed to see what is causing this to occur. Furthermore, the current methodology can easily be expanded to include many other pesticides (and other toxic organic contaminants). All that is necessary is the knowledge of the chemical formula of the compound; it can then be easily added to the screen (provided it extracts and is detectable by the LC/HRMS system). In fact, because the instrumental conditions do not change, items can be added to the list retroactively. This is one item that the triple quadrupole system could not do because it needs to have screened for a specific (and preset) transition. It is also useful to examine the validated level in terms of what detection level is actually needed. This can be done by comparing the validated level to the tolerance for a particular pesticide in a given matrix as determined by the EPA and listed in the Code of Federal Regulations.21 Table 5 compares the

concentration of the extract by a factor of 2−5. It is possible that using one of these other versions would further reduce the false negative rate and improve the overall sensitivity of the method at the possible cost of increasing false positives. It is useful to compare the overall performance of this HRMS method with that of the triple quadrupole MS/MS reported by Sack et al. for these same three matrixes using the same extraction scheme but with multiple reaction monitoring tandem mass spectrometry as the detection method.10 This comparison is shown in Table 4. It should be noted that the Sack et al. data were reported using operating conditions that were optimized for each individual pesticide, whereas the current data was collected using a single scan function on the mass spectrometer. Furthermore, the Sack data are actual reported MDLs, while the current data only report on the minimum concentration in this screening project that had no false negatives; for many compounds, the detection limit is likely lower because there were many pesticides with strong signals at the lowest tested level. Nevertheless, the current data show a robust response with the minimum detection limits; 86

dx.doi.org/10.1021/jf405128y | J. Agric. Food Chem. 2014, 62, 80−87

Journal of Agricultural and Food Chemistry

Article

raphy-tandem mass spectrometry. J. Agric. Food Chem. 2009, 57, 4068−78. (8) Park, S.; Lee, S. J.; Kim, H. G.; Jeong, W. Y.; Shim, J. H.; Abd ElAty, A. M.; Jeong, S. W.; Lee, W. S.; Kim, S. T.; Shin, S. C. Residue analysis of multi-class pesticides in watermelon by LC-MS/MS. J. Sep Sci. 2010, 33, 493−501. (9) Huang, Z.; Zhang, Y.; Wang, L.; Ding, L.; Wang, M.; Yan, H.; Li, Y.; Zhu, S. Simultaneous determination of 103 pesticide residues in tea samples by LC-MS/MS. J. Sep Sci. 2009, 32, 1294−1301. (10) Sack, C.; Smoker, M.; Chamkasem, N.; Thompson, R.; Satterfield, G.; Masse, C.; Mercer, G.; Neuhaus, B.; Cassias, I.; Chang, E.; Lin, Y.; MacMahon, S.; Wong, J.; Zhang, K.; Smith, R. E. Collaborative Validation of the QuEChERS Procedure for the Determination of Pesticides in Food by LC-MS/MS. J. Agric. Food Chem. 2011, 59, 6383−6411. (11) Gomez-Perez, M. L.; Plaza-Bolanos, P.; Romero-Gonzalez, R.; Martinez-Vidal, J. L.; Garrido-Frenich, A. Comprehensive Qualitative and Quantitative Determination of Pesticides and Veterinary Drugs in Honey Using Liquid Chromatography-Orbitrap High Resolution Mass Spectrometry. J. Chromatogr. A 2012, 1248, 130−138. (12) Adler, L.; Steinborn, A.; Bergelt, B. Suitability of an Orbitrap Mass Spectrometer for the Screening of Pesticide Residues in Extracts of Fruits and Vegetables. J. AOAC Int. 2011, 94 (6), 1661−1673. (13) Mol, H. G. J.; Zomer, P.; de Koning, M. Qualitative Aspects and Validation of a Screening Method for Pesticides in Vegetables and Fruits Based on Liquid Chromatography Coupled to Full Scan High Resolution (Orbitrap) Mass Spectrometry. Anal. Bioanal. Chem. 2012, 403, 2891−2908. (14) Wang, J.; Chow, W.; Leung, D.; Chang, J. Application of Ultrahigh-Performance Liquid chromatography and Electrospray Ionization Quadrupole Orbitrap High-Resolution Mass Spectrometry for Determination of 166 Pesticides in Fruits and Vegetables. J. Agric. Food Chem. 2012, 60, 12088−12104. (15) Diaz, R.; Ibanez, M.; Sancho, J. V.; Hernandez, F. Target and Non-Target Screening Stratagies for Organic Contaminants, Residues, and Illicit Substances in Food, Environmental and Human Biological Samples by UHPLC-QTOF-MS. Anal. Methods 2012, 4, 196−209. (16) Hayward, D. G.; Wong, J. W.; Zhang, K.; Chang, J.; Shi, F.; Banerjee, K.; Yang, P. Multiresidue Pesticide analysis in Ginseng and Spinach by Nontargeted and Targeted Screening Procedures. J. AOAC Int. 2011, 94 (6), 1741−1751. (17) Kaufmann, A.; Butcher, P.; Maden, K.; Walker, S.; Widmer, M. Comprehensive comparison of liquid chromatography selectivity as provided by two types of liquid chromatography detectors (high resolution mass spectrometry and tandem mass spectrometry): “Where is the crossover point? Anal. Chim. Acta 2010, 673, 60−72. (18) Kaufmann, A.; Dvorak, V.; Cruzer, C.; Butcher, P.; Maden, K.; Walker, S.; Widmer, M.; Schurmann, A. Study of high-resolution mass spectrometry as a replacement for tandem mass spectrometry in the field of quantitative pesticide analysis. J. AOAC Int. 2012, 528−548. (19) Kaufmann, A.; Widmer, M.; Maden, K. Post-Interface Signal Suppression, a Phenomenon Observed in a Single Stage Orbitrap Mass Spectrometer Coupled to an Electrospray Interfaced Liquid Chromatograph. Rapid Commun. Mass Spectrom. 2010, 24, 2162− 2170. (20) Gomez-Ramos, M. M.; Ferrer, C.; Malato, O.; Aguera, A.; Fernandez-Alba, A. R. Liquid chromatography-high-resolution-massspectrometry for pesticide residue analysis in fruit and vegetables: Screening and quantitative studies. J. Chromatogr. A 2013, 1287, 24− 37. (21) Code of Federal Regulations Title 40, 24, Part 180. (22) Pesticide Chemical News Guide, Arlington VA.

minimum validated levels with the tolerances for those compounds for which a tolerance has been established. Examination of this table shows that in almost all cases the minimum validated level is below (and in most cases, well below) the United States tolerance and for many compounds meets the 10 ppb default detection limit of Europe. Thus, use of the reported screening method would succeed in observing the pesticides at the levels where they would need to be found. This being a screening method, follow-up would be required to determine whether they are below tolerance when found. Overall, there clearly is utility in the HRMS screening procedures, though improvements still need to be made in the software for dealing with these large multiresidue screens.



ASSOCIATED CONTENT

S Supporting Information *

All data from the three laboratories and the false negative rate for each pesticide at each concentration. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Funding

This work was supported by the Food and Drug Administration through the Food Emergency Response Networks Chemistry Cooperative Agreement Program. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS We thank the personnel of the FDA’s Forensic Chemistry Center for suggestions during the study. REFERENCES

(1) Kellmann, M.; Muenster, H.; Zomer, P.; Mol, H. Full Scan MS in Comprehesive Qualitative and Quantitative Residue Analysis in Food and Feed Matrices: How Much Resolving Power is Required? J. Am. Soc. Mass Spectrom. 2009, 20, 1464−1476. (2) Anastassiades, M.; Lehotay, S. J.; Stajnbaher, D.; Schenck, F. J. Fast easy multiresidue method employing acetonitrile extraction partitioning and ‘dispersive solid phase extraction’ for determination of pesticide residues in produce. J. AOAC Int. 2003, 86, 412−431. (3) Lehotay, S. J.; Son, K. A.; Kwon, H.; Koesukwiwat, U.; Fu, W.; Mastovska, K.; Hoh, E.; Leepipatpiboon, N. Comparison of QuEChERS Sample Preparation Methods for the Analysis of Pesticide Residues in Fruits and Vegetables. J. Chromatogr. A 2010, 1217, 2458− 2560. (4) Lacina, O.; Zachariasova, M.; Urbanova, J.; Vaclavikova, M.; Cajka, T.; Hajslova, J. Critical Assessment of Extraction Methods for the Simultaneous Determination of Pesticide Residues and Mycotoxins in Fruits, Cereals, Spices, and Oil Seeds Employing Ultra-High Performance Liquid Chromatography-Tandem Mass Spectrometry. J. Chromatogr. A 2012, 1262, 8−18. (5) Ortelli, D.; Edder, P.; Corvi, C. Multiresidue analysis of 74 pesticides in fruits and vegetables by liquid chromatography-electrospray-tandem mass spectrometry. Anal. Chim. Acta 2004, 520, 33−45. (6) Zhang, K.; Wong, J. W.; Hayward, D. G.; Sheladia, P.; Krynitsky, A. J.; Schenk, F. J.; Webster, M. G.; Ammann, J. A.; Ebeler, S. E. Multiresidue pesticide analysis of wines by dispersive solid-phase extraction and ultrahigh-performance liquid chromatography-tandem mass spectrometry. J. Agric. Food Chem. 2009, 57, 4019−29. (7) Banerjee, K.; Oulkar, D. P.; Patil, S. B.; Jadhav, J.; Dasgupta, S.; Patil, S. B.; Bal, S.; Adsule, P. G. Multiresidue determination and uncertainty analysis of 87 pesticides in mango by liquid chromatog87

dx.doi.org/10.1021/jf405128y | J. Agric. Food Chem. 2014, 62, 80−87