Attachment 1 AOAC Chloramphenicol and Nitrofuran MLV.pdf

F O O D C H E M I C A L C O N T A M I N A N T S

Determination of Chloramphenicol and Nitrofuran

Metabolites in Cobia, Croaker, and Shrimp Using

Microwave-Assisted Derivatization, Automated SPE, and LC-MS/MS–Results from a U.S. Food and Drug

Administration Level Three Inter-Laboratory Study Brian T. Veach

U.S. Food and Drug Administration, Office of Regulatory Affairs, Arkansas Laboratory, 3900 NCTR Road, Jefferson, AK 72079

Corresponding author’s e-mail: brian.veach@fda.hhs.gov.

Collaborators: N. Bandara; B. Broadaway; C. Casey; A. Fong; A. Karmakar; M. Mayweather; A. Morara; N. Williams; B. Wright; L. Ye

Abstract

Background: An interlaboratory study was conducted to test a published, peer-reviewed manuscript in the Journal of AOAC INTERNATIONAL Vol 98, No. 3, 2015, “Quantitation of Chloramphenicol and Nitrofuran Metabolites in Aquaculture Products Using Microwave-Assisted Derivatization, Automated Solid-Phase Extraction, and LC-MS/MS.”

Objective: The purpose of this study was to demonstrate the performance of the method in shrimp, cobia, and croaker matrices.

Method: Three U.S. Food and Drug Administration laboratories participated in the collaborative study. The laboratories tested matrix blanks and laboratory-fortified matrix blanks at various levels in three separate matrices. The method evaluation included determination of the LOQ, accuracy, and precision.

Results: The reproducibility and repeatability of the RSD, % levels for matrix spikes fortified below the action level were <

10% for all residues in shrimp, < 13% for all residues in cobia except for 3-amino-2-oxazolidinone which was � 22%, and <

16% for croaker. The RSD, % levels for all other spikes in the study were < 14%. Average percent recoveries for all matrices ranged from 81.6% – 102%.

Conclusions: The study demonstrated that the method is acceptable for use for the combined determination of chloramphenicol and nitrofuran metabolites in the study matrices.

Highlights: The study showed acceptable quantitation for the high-throughput chloramphenicol and nitrofuran metabolites method.

Chloramphenicol and nitrofurans are broad-spectrum antimi-crobial drugs that have been used against gram-positive, gram-negative, and anaerobic bacteria in aquaculture products (1–4).

Nitrofurans are pro-drugs, which means they are administered in an inactive form, but through a metabolic process they are converted into a pharmacologically active drug. The nitrofurans are rapidly metabolized, with nitrofurantoin into 1-aminohy-dantoin (AHD), nitrofurazone into semicarbazide (SC), Received: 1 October 2019; Revised: 23 January 2020; Accepted: 6 February 2020 AOAC INTERNATIONAL 2020. This work is written by a US Government employee and is in the public domain in the US.

Journal of AOAC INTERNATIONAL, 103(4), 2020, 1043–1051 doi: 10.1093/jaoacint/qsaa019 Advance Access Publication Date: 4 June 2020 Article

D ow nloaded from https://academ ic.oup.com /jaoac/article/103/4/1043/5851561 by FD

A M edical Library - C orp Blvd user on 10 D ecem ber 2020 https://academic.oup.com/ furaltadone into 3-amino-5-morpholinomethyl-2-oxazolidinone (AMOZ), and furazolidone into 3-amino-2-oxazolidinone (AOZ) (5–7). The presence of these nitrofuran metabolites and chlor-amphenicol are of great concern to many countries including the United States and the European Union. This is due to their potential adverse health effects, which include aplastic anemia, carcinogenicity, and mutagenicity. Therefore, they are not ap-proved for veterinary drug applications in foodstuff.

There is a need to efficiently analyze the food supply for these residues to ensure they are not present in aquaculture products. Although international regulations have prohibited the use of these drugs in foodstuff, their affordability and avail-ability make them a continued concern for several countries.

In response to this concern, numerous LC-MS/MS methods for determination of these residues have been developed over the years (6–16). However, most of these methods are not inclusive of all residues, utilize a 16-hour acid hydrolysis and derivatiza-tion procedure, and are labor intensive.

The U.S. Food and Drug Administration’s (FDA) Office of Regulatory Affairs (ORA) recently made the decision that all ORA laboratories that currently analyze regulatory samples for chloramphenicol and nitrofuran metabolites are required to use the method previously published in the Journal of AOAC

INTERNATIONAL Vol 98, No. 3, 2015 which uses microwave-assisted derivatization and automated solid-phase extraction (SPE) to drastically improve throughput when compared to traditional methods (5). This method originally went through FDA Level Two, single laboratory validation for five different matrices in 2014.

To demonstrate the method capabilities, a multi-laboratory validation was performed with the three ORA laboratories that perform these analyses in aquaculture samples. This study determined method LOQ levels and method accuracy and precision in three distinctively different matrices.

Methods and Materials Apparatus

(a) LC-MS/MS.—AB Sciex (Framingham, MA) QTRAP 5500, AnalystVR software.

(b) HPLC.—Agilent (Santa Clara, CA) 1260 series.

(c) Chromatographic column.—Agilent Zorbax Eclipse XDB-C18,

1.8 mm, 4.6 � 50 mm.

(d) Centrifuge.—Capable of 3700 g.

(e) Microwave reaction system.—CEM (Matthews, NC) MARS 6.

(f) Blender.—RobotCoupe (Ridgeland, MS).

(g) Automated solid-phase extraction system (ASPEC).—Gilson

(Middleton, WI) GX-274.

(h) Solid-phase extraction cartridges.—Waters Oasis (Milford, MA)

HLB 3 cc � 60 mg.

(i) Polyvinylidene difluoride (PVDF) centrifuge filtration tubes.—

Grace Davison Discovery Science (Deerfield, IL), part no.

8604751.

(j) PVDF syringe filters.—Restek (Bellefonte, PA) 13 mm, 0.45 mm, part no. 26151.

(k) 2mL Autosampler vials with inserts.—Agilent Technologies, part no. 5188-8594.

(l) Autosampler vial caps.—Agilent Technologies, part no. 5182- 3459.

Standards and Reagents

(a) Acetonitrile.—LCMS grade.

(b) Methanol.—LCMS grade.

(c) Water.—LCMS grade.

(d) Ammonium acetate.—LCMS grade.

(e) 0.1% formic acid in water.—LCMS grade.

(f) Ethyl acetate.—HPLC grade.

(g) 0.125 N hydrochloric acid.—Reagent grade.

(h) Di-potassium hydrogen phosphate.—Reagent grade.

(i) Semicarbazide hydrochloride as semicarbazide reference stan-dard.—Sigma-Aldrich (St. Louis, MO), part no. 33656.

(j) AMOZ reference standard.—Sigma-Aldrich, part no 33349.

(k) AOZ reference standard.—Sigma-Aldrich, part no. 33347.

(l) 1-Aminohydantoid hydrochloride as AHD reference standard.—

Sigma-Aldrich, part no. 33655.

(m) Chloramphenicol reference standard.—Sigma-Aldrich, part no.

C0378.

(n) Chloramphenicol-d5 internal standard.—Cambridge Isotopes

(Andover, MA), part no. DLM-1190-0.

(o) SCA-13C-15N2 internal standard.—Sigma-Aldrich, part no.

33882.

(p) AMOZ-d5 internal standard.—Sigma-Aldrich, part no. 33881.

(q) AHD-13C3 internal standard.—Santa Cruz Biotechnology

(Santa Cruz, CA), part no. 208596.

(r) 2-Nitrobenzaldehyde.—Fisher Scientific (Houston, TX), part

no. 552-89-6.

Standards Preparation

Nitrofuran analytical and internal standard stock solutions (100mg/mL) were prepared in an 80:20 methanol: water (v/v) so-lution. These solutions are stable for up to 6 months when stored at � 5

C. Chloramphenicol stock and internal standard solutions (20.0 mg/mL) were prepared in methanol. These solu-tions are stable for up to 1 year when stored at � 5

C. A mixed intermediate analytical standard was then prepared to a con-centration of 25.0 ng/g for each of the nitrofurans and 5.00 ng/g for chloramphenicol in an 80:20 methanol: water (v/v) solution.

Additionally, a mixed internal standard solution was prepared in an 80:20 methanol: water (v/v) solution at a concentration of

80.0 ng/g for each deuterated or isotopically labeled nitrofuran standard, and 24.0 ng/g for the deuterated chloramphenicol standard. The mixed analytical standard and the mixed internal standard are stable for up to 6 months when stored at � 5

C.

Five extracted calibration standards were prepared.

Approximately 2.00 g (þ/- 0.03) of control matrix was fortified with the mixed intermediate analytical standard (underivatized nitrofuran metabolites: 0.25, 0.50, 1.00, 2.00, and 5.00 ng/g; chlor-amphenicol: 0.050, 0.100, 0.200, 0.400, and 1.00 ng/g). The cali-bration standards were additionally fortified with the mixed intermediate internal standard solution (nitrofuran metabolites:

2.00 ng/g and chloramphenicol 0.60 ng/g).

Sample Preparation

Control cobia, croaker, and shrimp products were obtained from commercially available sources for the demonstration of nega-tive controls and spike recoveries. The edible portion of tissue was homogenized with dry ice and a blender until powdery con-sistencies were obtained. Samples should sublime for a mini-mum of 12 h in a refrigerator prior to use. A volume of 50.0 mL of mixed internal standard solution was added to 2.00 g (þ/- .03) of negative control matrix in a 50 mL centrifuge tube.

1044 | Veach: Journal of AOAC INTERNATIONAL Vol. 103, No. 4, 2020

D ow nloaded from https://academ ic.oup.com /jaoac/article/103/4/1043/5851561 by FD

A M edical Library - C orp Blvd user on 10 D ecem

Approximately 10 mL of 0.125 M HCl was added to each tube.

Additionally, 200 mL of 100 mM 2-nitrobenzaldehyde prepared fresh daily in methanol was added to each vessel. Each vessel was then vortexed (�20 s) and placed in the microwave for 6 min (5 min ramp from ambient temperature to 95�C, and hold 1 min).

Upon microwave completion, 5 mL of 1.0 M dipotassium hy-drogen phosphate was added to each vessel to adjust the pH to

7.3 (þ/- 0.3). If the desired pH is not achieved, the extract can be further adjusted by the addition of dilute sodium hydroxide or hydrochloric acid. Contents were then centrifuged for 10 min at 3700 g, and the supernatant was subsequently transferred into a

0.45 mm PVDF 50 mL filtration tube where it was centrifuged for 5 min at 2500 g. Utilizing the ASPEC system, the HLB (3 cc � 60 mg), solid-phase extraction cartridges were conditioned with 3 mL of ethyl acetate, 3 mL of methanol, and 3 mL of water. The ASPEC system loaded the samples onto the conditioned SPE car-tridges and washed the cartridges with 2 mL of water and 2 mL of 30% methanol in water. Subsequently, the ASPEC system dried the SPE cartridges and eluted them with 3 mL of ethyl ace-tate. The eluent was evaporated to dryness and reconstituted with 250 mL of reconstitution solution (40% methanol and 60%

8.5 mM ammonium acetate in 0.1% formic acid). Reconstituted samples were vortexed (�10 s), sonicated (�5 min), and filtered through a 0.45 mm PVDF syringe filter into a 2 mL autosampler vial with insert.

LC-MS/MS

The prepared extracts were analyzed under two separate LC-MS/MS conditions (chromatography is illustrated in Figures 1–2).

Nitrofuran metabolite extracts have been shown to be stable for approximately 3 days when stored at 5

C. However, chloram-phenicol extracts have proven to be stable for up to 6 months when stored at 5

C. Considering the stability of the extracts, it was believed the sample vials should first be analyzed for nitro-furans, and then subsequently assayed for chloramphenicol.

Nitrofuran metabolites were assayed using atmospheric pressure chemical ionizatin (APCI). Samples were injected (10mL) onto a 50 mm � 4.6 mm � 1.8 mm XDB-C-18 reverse-phase column and eluted under isocratic conditions with an 8.5 mM ammonium acetate in 0.1% formic acid (mobile phase A) and methanol (mobile phase B) followed by column cleaning and re-generation. Mobile phase A was held at 60% (600 mL/min) for

2.3 min. At 2.4 min, mobile phase B (600mL/min) was ramped up to 80% for 1 min. At 3.5 min, mobile phase B was ramped down to 40% (750mL/min) and held for 1.5 min. Data from the mass spectrometer was acquired utilizing selected reaction monitor-ing (SRM) in positive APCI mode (precursor ions and MS/MS fragment ions are listed in Table 1).

Chloramphenicol was analyzed using ESI. Samples were injected (5 mL) onto a 50 mm � 4.6 mm � 1.8 mm XDB-C-18 re-verse-phase column. Chloramphenicol was eluted from the col-umn with a gradient flow (750mL/min) of water (mobile phase A)

Figure 1. LC-MS/MS nitrofuran metabolite quantitation ion chromatograms of a 0.25 ng/g extracted fortified control shrimp (left) and a matrix blank (right).

Veach: Journal of AOAC INTERNATIONAL Vol. 103, No. 4, 2020 | 1045

D ow nloaded from https://academ ic.oup.com /jaoac/article/103/4/1043/5851561 by FD

A M edical Library - C orp Blvd user on 10 D and acetonitrile (mobile phase B). Mobile phase A was held for an initial time of 0.5 min at 75%, then the proportion of mobile phase B was increased linearly to 50% in the following 3 min. At

3.6 min the proportion of mobile phase B was increased to 90%, followed by a hold of 1 min at 90%. At 4.7 min mobile phase B was returned to 25% and re-equilibrated for 1.3 min prior to sub-sequent injection. The mass spectrometer data were collected in negative ESI mode using SRMs (precursor and MS/MS frag-ment ions obtained are listed in Table 2).

Data Analysis

Quantitation was performed for each analyte of interest by cal-culating the ratio of the chromatographic area of the quantita-tion ion with respect to the chromatographic area of the internal standard (AMOZ/AMOZ d5, SC/SC 13C 15N2, AOZ/SC 13C 15N2, AHD/AHD 13C3, and CAP/CAP d5). Each representative ratio was plotted against the concentration of the corresponding ma-trix extracted calibration standard. The linear calibration curve fit obtained should yield a regression (R^2) of � 0.995. For posi-tive confirmation all product ions must be detected, the associ-ated chromatographic peak must exhibit a retention time within 5% of the average retention time of the calibration stand-ards, and the product ion ratios must be within 20% (absolute value) of the product ion ratios obtained from calibration stand-ards (17).

Collaborative Study

This study was designed to fulfill a Level Three multi-laboratory validation in accordance with FDA guidelines for method vali-dations (18). Three laboratories were selected by FDA ORA’s Office of Regulatory Science, based on available equipment and intended future use of the method for regulatory testing. The multi-laboratory validation was separated into three phases.

The first phase was to verify that the instrumentation used was sufficiently selective and sensitive for the analysis. This was conducted by shipping the laboratories extracted matrix calibration standards to be analyzed. Each lab was required to demonstrate linearity of R2 > 0.995 and a signal-to-noise ratio � 10:1 for each quantitation ion in the low-level calibration standard.

The second phase was designed to accurately assess if all extraction equipment was functioning uniformly and analysts were adequately trained. Each laboratory was provided with laboratory fortified calibration standards, laboratory fortified blanks, and matrix blanks for sample extraction.

Once each laboratory had completed the extraction and analysis of the quality control samples, they were required to demonstrate all analytical criteria required for phase one.

Study laboratories then provided their calculated value in ng/g for all the laboratory fortified blanks and matrix blanks.

Laboratory fortified blanks were required to be within 630% of the target value.

The third phase of the multi-laboratory validation was the analysis of the collaborative study samples. During this phase each participating laboratory was sent blind proficiency sam-ples of shrimp, cobia, and croaker matrices. All three laborato-ries completed and submitted their analytical findings for the collaborative portion within 4 weeks of receipt.

Table 1. MS/MS acquisition parameters for nitrofuran metabolites using APCI

Nitrofuran Precursor Product Retention Metabolite Ion, m/z Ions, m/z Collision Enbergy (CE), V Time, min

NBA-AHD 249.0 134.0, 104.0, 178.1 17, 30, 22 2.1

NBA-AOZ 236.0 134.0, 104.0, 149.0 18, 29, 20 2.2

NBA-AMOZ 335.1 291.2 262.2, 128.0 16, 23, 30 1.7

NBA-SC 209.1 166.1, 134.0, 192.1 14, 15, 15 2.3

NBA-AHD 13C3 252.0 134.0 16 2.1

NBA-AMOZ d5 340.1 296.2 16 1.7

NBA-SC 13C 15N2 212.1 168.1 14 2.3

Table 2. MS/MS acquisition parameter for chloramphenicol analysis using ESI

Precursor Product Retention Analyte Ion, m/z Ions, m/z Collision Enbergy (CE), VTime, min

CAP 321.1 151.8, 193.8, 257.2 �23, �16, �16 3.1

CAP d5 326.1 262.2 �16 3.1

Figure 2. LC-MS/MS chloramphenicol quantitation ion chromatograms of a 0.05 ng/g extracted fortified control shrimp (left) and a matrix blank (right).

1046 | Veach: Journal of AOAC INTERNATIONAL Vol. 103, No. 4, 2020

D ow nloaded from https://academ ic.oup.com /jaoac/article/103/4/1043/5851561 by FD

A M edical Library - C orp Blvd user on 10 D

Study Materials

As part of the collaborative study, control shrimp, cobia, and croaker products were obtained from commercially available sources for the demonstration of negative controls. These prod-ucts were assayed to ensure the absence of the target residues.

Each of the participating labs were shipped frozen shrimp, co-bia, and croaker matrices for use as matrix matched extracted calibration standards. Along with the negative controls, each lab received 12 blind test samples of each matrix. Three of the blind samples provided were negative controls, and the remain-ing nine were laboratory fortified matrices, three at less than the regulatory action level, three were at or near the regulatory action level, and three were greater than the regulatory action level.

Protocol

All participating laboratories were instructed not to share results with other laboratories. Furthermore, all labs were also instructed that the method was to be followed as written, and that any deviations from the method had to be pre-approved by the study coordinator even if the deviations appear trivial in na-ture. Slight modifications were permitted to source parameters, collision energies, and other voltages necessary as these are all tunable parameters and should be optimized on each instru-ment used in the study.

As part of the study, each lab was required to fortify their own matrix matched calibration standards with both analytical standards and internal standards and fortify all blind samples with internal standards as outlined in the method. The partici-pating laboratories were to provide the study coordinator with all analytical findings including chromatograms, signal-to-noise calculations for the quantitation and confirmation ions, ion ratios, and correlation coefficients for the calibration curves.

A spreadsheet was also provided to each lab to report their ana-lytical findings for each blind sample received. The analytical results were required to be reported in ng/g and to a minimum of three significant figures.

Data Evaluation

Results from each laboratory were submitted to the study coordinator. Prior to conducting any statistical evaluations of the data, the study coordinator reviewed all chromatograms, ion ratios, calibration curve coefficients, and signal-to-noise cal-culations to verify that method specifications were met. If the acceptable criteria were not directly specified in the method, then FDA CVM Guide 118 was used for assessing data (17).

The results for the blind samples provided in the spreadsheets were used to calculate accuracy, reproducibility, LOQs, and to identify any statistical outliers using the Grubbs test.

In addition to the statistical evaluations referenced above, z-scores were calculated to evaluate the individual measure-ments relative to the overall mean values for the unknown samples. For a two-tailed Grubbs outlier test, 95% confidence al-pha is 0.005, and the critical value of the test statistic (G) is 2.21 when n¼ 9. The z-scores were calculated using the following equation:

Z ¼ ðXmeas– Xref Þ=SD;

where Xmeas ¼ the measured result, Xref ¼ the overall mean of the data for unknown samples, and SD is the standard deviation.

Results and Discussion

Although extenuating circumstances can occur, there was not any preconceived notion at the start of the study that any method modifications would be needed by any of the labs. In fact, it was believed that if changes were allowed they would have been solely due to analyst’s preferences. These changes could introduce variability in the data provided and negate the collaborative study. However, one condition which was never taken into consideration during the initial planning of the col-laborative study, is that the laboratory which originally devel-oped and validated the method was at an elevation of approximately 100 m. The original published method specifies a microwave temperature ramp to 95�C. The maximum temper-ature was chosen to help achieve an optimal extraction and derivatization conditions, without volatizing the aqueous solution.

Prior to phase three, one of the participating laboratories reached out to the study coordinator because they were losing a portion of their samples during the microwave digestion.

It was realized that the issue was due to the laboratory’s eleva-tion which was more than 1500 m. The study coordinator allowed this laboratory to reduce their temperature ramp on the microwave digestion procedure to a maximum tempera-ture of 90�C.

The laboratory that originally developed the method ana-lyzed incurred residues of nitrofurans in shrimp obtained from the Australian Government National Measurement Institute prior to the collaborative study. The analysis of incurred resi-dues was to demonstrate the effectiveness of the microwave for the acid hydrolysis and derivatization process as written in the method (5). The recoveries of the incurred residues for semi-carbazide and AOZ ranged from 91.3–119%, with RSD, % levels �6% (see Tables 3 and 4). To determine if this method differs significantly from the method utilized by the Australian Government National Measurement Institute, a two-sided F test was performed. The F test calculated value for

Table 3. Incurred residue analysis recovery results

Theoretical Calculated Semicarbazide Theoretical Calculated AOZ Semicarbazide Semicarbazide Recovery, % AOZ AOZ Recovery, % Concn, ng/g Concn, ng/g Concn, ng/g Concn, ng/g

4.51 5.28 117 1.94 1.85 95.5

4.56 5.20 114 1.96 1.79 91.3

4.55 5.43 119 1.95 2.02 103

Veach: Journal of AOAC INTERNATIONAL Vol. 103, No. 4, 2020 | 1047

D ow nloaded from https://academ ic.oup.com /jaoac/article/103/4/1043/5851561 by FD

A M edical Library - C orp Blvd user on 10 D semicarbazide and AOZ was 17.4 and 5.10 respectively, which is significantly less than the F critical value of 39.4 for both compounds. Therefore, it can be concluded that there is no significant difference between the two variances at the 95% confidence level.

The participating laboratory which had to drop their micro-wave temperature also successfully demonstrated that they could recover nitrofuran metabolites in incurred residues with a temperature setting of 90�C. It is evident that this participating laboratory produced equivalent results during the validation.

This can be seen in the statistical evaluations shown in

Tables 5–17. However, efforts should be made by all labs near sea level to use the recommended 95�C microwave temperature.

All the participating laboratories utilized the same model tandem mass spectrometer, and the HPLC systems used, al-though slightly different, were all essentially equivalent. Other than these variations, no other notable deviations or modifica-tions were noted by the study participants. Laboratories reported analytical findings back to the study coordinator for each phase of the study in less than four weeks after receipt of samples.

Statistical Evaluation

Since the analytical method utilizes matrix matched extracted calibration standards, all statistical evaluations were separated based on the specific matrix analyzed. The LOQ was calculated from the standard deviation of the quantitative values for all participating laboratories’ low-level fortified blanks as deter-mined by the equation below:

Table 4. SD and RSD, % from incurred residues

Semicarbazide AOZ (Degrees freedom ¼ 2) (Degrees freedom ¼ 2)

Average, ng/g 5.30 1.89 SD, ng/g 0.117 0.119

RSD, % 2.20 6.32

Table 5. Collaborative study shrimp matrix results for Lab 1

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.487 0.520 0.474 0.457 0.102 Average recovery, % 97.3 104 94.8 91.3 102 SDr 0.00702 0.00436 0.00854 0.0142 0.00115 RSDr, % 1.44 0.838 1.80 3.11 1.14 LOQ, ng/g 0.0702 0.0436 0.0854 0.142 0.0115 Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.928 1.00 1.01 0.973 0.207 Average recovery, % 92.8 100 101 97.3 104 SDr 0.0246 0.0121 0.0195 0.0321 0.00721 RSDr, % 2.65 1.21 1.93 3.30 3.48 Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 1.84 1.98 2.06 1.98 0.397 Average recovery, % 92.0 98.8 103 99.2 99.2 SDr 0.0400 0.0321 0.0600 0.0924 0.00651 RSDr, % 2.17 1.63 2.91 4.66 1.64

Table 6. Collaborative study shrimp matrix results for Lab 2

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.435 0.451 0.409 0.438 0.0934 Average recovery, % 86.9 90.2 81.7 87.6 93.4 SDr 0.0211 0.0112 0.0307 0.0416 0.0071 RSDr, % 4.86 2.49 7.52 9.49 7.56

LOQ 0.211 0.112 0.307 0.416 0.0706

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.734 0.847 0.852 0.951 0.180 Average recovery, % 73.4 84.7 85.2 95.1 90.2 SDr 0.0823 0.0700 0.0687 0.0490 0.0116 RSDr, % 11.2 8.27 8.06 5.15 6.43 Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 1.93 1.97 1.91 2.14 0.4213 Average recovery, % 96.7 98.4 95.4 107 105 SDr 0.0259 0.0526 0.0391 0.0312 0.0146 RSDr, % 1.34 2.67 2.05 1.46 3.47

Table 7. Collaborative study shrimp matrix results for Lab 3

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.442 0.488 0.419 0.492 0.102 Average recovery, % 88.5 97.5 83.8 98.5 102 SDr 0.0295 0.0100 0.0234 0.0140 0.0070 RSDr, % 6.67 2.05 5.58 2.85 6.91

LOQ 0.295 0.100 0.234 0.140 0.0702

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.901 1.02 0.960 1.01 0.198 Average recovery, % 90.1 102 96.0 101 99.0 SDr 0.0370 0.0105 0.0364 0.0252 0.00458 RSDr, % 4.11 1.03 3.79 2.50 2.31 Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 2.00 2.01 2.02 1.98 0.387 Average recovery, % 100 101 101 98.9 96.8 SDr 0.0275 0.0441 0.102 0.0757 0.00643 RSDr, % 1.37 2.19 5.03 3.83 1.66

Table 8. Collaborative statistical evaluation for shrimp matrix

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.455 0.486 0.434 0.462 0.0989 Average recovery, % 90.9 97.2 86.8 92.5 98.9

SDR 0.0305 0.0309 0.0363 0.0333 0.00651

RSDR, % 6.72 6.36 8.36 7.19 6.58

LOQ 0.305 0.309 0.363 0.333 0.0651

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.854 0.954 0.941 0.977 0.195 Average recovery, % 85.4 95.4 94.1 97.7 97.6

SDR 0.102 0.0884 0.0808 0.0396 0.0138

RSDR, % 11.9 9.27 8.59 4.05 7.06

Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 1.92 1.99 2.00 2.03 0.402 Average recovery, % 96.2 99.3 99.9 102 100

SDR 0.0746 0.0434 0.0926 0.101 0.0175

RSDR, % 3.88 2.19 4.64 4.95 4.35

1048 | Veach: Journal of AOAC INTERNATIONAL Vol. 103, No. 4, 2020

D ow nloaded from https://academ ic.oup.com /jaoac/article/103/4/1043/5851561 by FD

A M edical Library - C orp Blvd user on 10 D

Table 9. Collaborative study cobia matrix results for Lab 1

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.431 0.449 0.460 0.468 0.0932 Average recovery, % 86.3 89.9 92.1 93.6 93.2 SDr 0.0162 0.0166 0.0208 0.0181 0.00373 RSDr, % 3.76 3.70 4.53 3.86 4.00

LOQ 0.162 0.166 0.208 0.181 0.0373

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.917 0.930 0.905 0.932 0.200 Average recovery, % 91.7 93.0 90.5 93.2 100 SDr 0.0223 0.0315 0.0260 0.00643 0.0197 RSDr, % 2.44 3.38 2.88 0.690 9.81 Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 1.90 1.87 1.90 1.86 0.379 Average recovery, % 95.0 93.3 95.0 93.0 94.8 SDr 0.0500 0.0208 0.0458 0.0265 0.0105 RSDr, % 2.63 1.12 2.41 1.42 2.77

Table 10. Collaborative study cobia matrix results for Lab 2

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.448 0.449 0.295 0.487 0.0897 Average recovery, % 89.6 89.7 58.9 97.5 89.7 SDr 0.0226 0.0406 0.0521 0.0406 0.0112 RSDr, % 5.05 9.06 17.7 8.33 12.5

LOQ 0.226 0.406 0.521 0.406 0.112

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.989 1.02 0.758 0.883 0.209 Average recovery, % 98.9 102 75.8 88.3 104 SDr 0.0712 0.0223 0.164 0.164 0.0138 RSDr, % 7.20 2.18 21.6 18.5 6.60 Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 2.06 1.97 1.56 1.93 0.390 Average recovery, % 103 98.3 78.0 96.3 97.5 SDr 0.0489 0.0207 0.400 0.0231 0.0131 RSDr, % 2.37 1.05 25.6 1.20 3.36

Table 11. Collaborative study cobia matrix results for Lab 3

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.429 0.438 0.469 0.479 0.0920 Average recovery, % 85.7 87.7 93.7 95.8 92.0 SDr 0.0169 0.00252 0.0107 0.0235 0.00361 RSDr, % 3.93 0.574 2.28 4.91 3.92

LOQ 0.169 0.0252 0.107 0.235 0.0361

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.941 0.937 0.970 0.943 0.187 Average recovery, % 94.1 93.7 97.0 94.3 93.3 SDr 0.0400 0.00200 0.0452 0.0452 0.00473 RSDr, % 4.25 0.213 4.66 4.79 2.53 Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 1.85 1.88 1.84 1.82 0.367 Average recovery, % 92.7 94.2 91.8 91.2 91.7 SDr 0.00896 0.0529 0.0726 0.101 0.00586 RSDr, % 0.483 2.81 3.95 5.56 1.60

Table 12. Collaborative statistical evaluation for cobia matrix

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.436 0.445 0.408 0.478 0.0916 Average recovery, % 87.2 89.1 81.6 95.6 91.6

SDR 0.0186 0.0226 0.0897 0.0265 0.00637

RSDR, % 4.27 5.08 22.0 5.54 6.95

LOQ 0.186 0.226 0.897 0.265 0.0637

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.949 0.963 0.878 0.919 0.199 Average recovery, % 94.9 96.3 87.8 91.9 99.3

SDR 0.0530 0.0478 0.127 0.0893 0.0155

RSDR, % 5.58 4.97 14.5 9.72 7.82

Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 1.94 1.90 1.77 1.87 0.379 Average recovery, % 96.9 95.2 88.3 93.5 94.7

SDR 0.0996 0.0548 0.257 0.0701 0.0134

RSDR, % 5.14 2.87 14.6 3.75 3.54

Table 13. Collaborative study croaker matrix results for Lab 1

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.465 0.437 0.424 0.452 0.0901 Average recovery, % 93.0 87.5 84.7 90.5 90.1 SDr 0.0106 0.00666 0.00153 0.00874 0.000781 RSDr, % 2.28 1.52 0.361 1.93 0.867

LOQ 0.106 0.0666 0.0153 0.0874 0.00781

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.910 0.886 0.861 0.900 0.189 Average recovery, % 91.0 88.6 86.1 90.0 94.7 SDr 0.0468 0.0346 0.0350 0.0301 0.0095 RSDr, % 5.14 3.91 4.07 3.35 5.02 Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 1.88 1.90 1.85 1.89 0.417 Average recovery, % 93.8 94.8 92.5 94.7 104 SDr 0.0611 0.0321 0.100 0.0416 0.00200 RSDr, % 3.26 1.69 5.41 2.20 0.480

Table 14. Collaborative study croaker matrix results for Lab 2

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.348 0.372 0.377 0.416 0.0845 Average recovery, % 69.5 74.4 75.3 83.1 84.5 SDr 0.0340 0.00844 0.0165 0.0589 0.00247 RSDr, % 9.79 2.27 4.37 14.2 2.92

LOQ 0.340 0.0844 0.165 0.589 0.0247

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.916 0.937 0.943 0.970 0.193 Average recovery, % 91.6 93.7 94.3 97.0 96.6 SDr 0.00398 0.0262 0.0317 0.0200 0.0237 RSDr, % 0.435 2.80 3.36 2.07 12.2 Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 2.01 1.93 1.84 1.99 0.40 Average recovery, % 101 96.6 91.9 99.5 100 SDr 0.181 0.0180 0.0622 0.0446 0.0702 RSDr, % 8.98 0.931 3.38 2.24 17.5

Veach: Journal of AOAC INTERNATIONAL Vol. 103, No. 4, 2020 | 1049

D ow nloaded from https://academ ic.oup.com /jaoac/article/103/4/1043/5851561 by FD

A M edical Library - C orp Blvd user on 10 D

LOQ ¼ SD � 10

The LOQs for chloramphenicol were very consistent when comparing the three matrices. This was also mostly true with respect to the nitrofuran metabolites in shrimp and croaker;

however, greater variation was seen in the cobia matrix. This is partly due to one of the laboratories having a much higher RSD, % than other participating labs.

The Grubbs outlier test showed no outliers for the shrimp matrix, with the absolute value of all the z-scores being � 2.08.

When the outlier test was applied to the cobia matrix, three out-liers were found: two AOZ outliers at 2.41 and 2.58, and one out-lier for AMOZ at 2.46. In the croaker matrix, there was one outlier for semicarbazide at 2.34, and one outlier for AMOZ at

2.29. Although a total of five outliers were determined, all the outliers were only slightly above the 2.21 critical value; thus, the outliers were not removed from the statistical data. The statistical outliers, repeatability SDs (SDr), reproducibility SDs (SDR), repeatability RSDs (RSDr), and reproducibility RSDs (RSDR) are all listed in Tables 5–17.

The quantitative results provided by all three participating laboratories were excellent. Trueness ranged from 81.6% to

102% recovery for all analytes in all matrices. RSDr and RSDR were generally � 15%, except for AOZ in cobia.

Conclusions

The method has provided acceptable quantitative results at and below the level of interest for nitrofuran metabolites (1 ng/g) and chloramphenicol (0.3 ng/g) in all the matrices evaluated.

The method accuracy was deemed excellent with most recoveries being � 90%, and precision generally being � 15%RSDr. Therefore, the study coordinator recommends this method for acceptance as Final Action status.

Acknowledgments

The study coordinator is extremely appreciative to all laboratories that participated in this study, which are listed in alphabetical order below:

• U.S. Food and Drug Administration, Office of Regulatory Affairs, Arkansas Laboratory, Jefferson, AR (7 collaborators)

Table 16. Collaborative statistical evaluation for croaker matrix

SC AHD AOZ AMOZ CAP

Concn, ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.434 0.418 0.427 0.441 0.0900 Average recovery, % 86.8 83.6 85.4 88.3 90.0

SDR 0.0690 0.0353 0.0476 0.0357 0.00504

RSDR, % 15.9 8.45 11.1 8.10 5.60

LOQ 0.690 0.353 0.476 0.357 0.0504

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 0.941 0.904 0.930 0.923 0.190 Average recovery, % 94.1 90.4 93.0 92.3 94.8

SDR 0.0485 0.0331 0.0635 0.0405 0.0141

RSDR, % 5.16 3.66 6.83 4.39 7.45

Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 1.95 1.88 1.88 1.90 0.404 Average recovery, % 97.3 93.9 94.0 95.1 101

SDR 0.112 0.0602 0.100 0.0801 0.0385

RSDR, % 5.78 3.20 5.32 4.21 9.54

Table 15. Collaborative study croaker matrix results for Lab 3

SC AHD AOZ AMOZ CAP

Concn ng/g 0.500 0.500 0.500 0.500 0.100 Average recovery 0.490 0.445 0.481 0.456 0.0953 Average recovery, % 98.0 89.0 96.2 91.2 95.3 SDr 0.0200 0.00819 0.0242 0.00819 0.00252 RSDr, % 4.08 1.84 5.04 1.80 2.64

LOQ 0.200 0.0819 0.242 0.0819 0.0252

Concn, ng/g 1.00 1.00 1.00 1.00 0.200 Average recovery 1.00 0.889 0.987 0.898 0.186 Average recovery, % 100 88.9 98.7 89.8 93.2 SDr 0.0208 0.00500 0.0402 0.0136 0.0107 RSDr, % 2.09 0.562 4.08 1.52 5.74 Concn, ng/g 2.00 2.00 2.00 2.00 0.400 Average recovery 1.95 1.80 1.95 1.82 0.394 Average recovery, % 97.4 90.2 97.7 91.0 98.4 SDr 0.0228 0.0117 0.119 0.0142 0.0238 RSDr, % 1.17 0.647 6.09 0.781 6.05

Table 17. z-scores (values listed are absolute values)

Lab Semicarbazide AHD AOZ AMOZ CAP

# 1 (s) 0.592–1.66 0.0750–1.19 0.0273–1.35 0.249–1.02 0.0699–1.44 # 2 (s) 0.132–2.05 0.299–1.65 0.0604–2.08 0.157–2.05 0.0768–2.04 # 3 (s) 0.0457–1.49 0.0252–1.50 0.0248–1.17 0.0126–1.31 0.153–1.55 # 1 (cb) 0.122–1.08 0.0712–1.23 0.00331–0.717 0.00206–1.02 0.0895–1.80 # 2 (cb) 0.209–2.02 0.500–1.81 0.0711–2.58a 0.186–2.46b 0.0895–1.80 # 3 (cb) 0.0203–1.02 0.194–1.02 0.0945–1.04 0.0717–1.89 0.101–1.39 # 1 (ck) 0.138–1.52 0.0869–1.75 0.0436–1.66 0.103–1.20 0.0425–0.679 # 2 (ck) 0.239–2.34c 0.254–1.49 0.0556–1.45 0.467–2.29d 0.106–2.07 # 3 (ck) 0.156–1.62 0.298–1.43 0.187–1.75 0.159–1.13 0.144–1.59

(s) ¼ Shrimp matrix.

(cb) ¼ Cobia matrix.

(ck) ¼ Croaker matrix.

a Two statistical outliers were calculted for AOZ from laboratory two in the cobia matrix (2.42, and 2.58).

b One statistical outlier was calculated for AMOZ in the cobia matrix.

c One statistical outlier was calculated for semicarbazide in the croaker matrix.

d One statistical outlier was calculated for AMOZ in the croaker matrix.

1050 | Veach: Journal of AOAC INTERNATIONAL Vol. 103, No. 4, 2020

D ow nloaded from https://academ ic.oup.com /jaoac/article/103/4/1043/5851561 by FD

A M edical Library - C orp Blvd user on 10 D

• U.S. Food and Drug Administration, Office of Regulatory Affairs, Denver Laboratory, Denver, CO (2 collaborators)

• U.S. Food and Drug Administration, Office of Regulatory Affairs, Southeast Food and Feed Laboratory, Atlanta, GA (1 collaborator)

The study coordinator would also like to acknowledge Chris Baker from the U.S. Food and Drug Administration, Office of Regulatory Affairs, Arkansas Laboratory who served as the backup study coordinator.

The views expressed in this document are those of the researchers and should not be interpreted as the official opinion or policy of the U.S. Food and Drug Administration, Department of Health and Human Services, or any other agency or compo-nent of the U.S. government. The mention of trades names, commercial products, or organizations is for clarification of the methods used and should not be interpreted as an endorsement of a product or manufacturer.

References

1. Agriculture USDo (2012) in United States Honey Production down

16 Percent, U.S.D.A. (Ed), Washington DC. http://www.ars.

usda.gov/News/docs.htm?docid¼15572

2. McCalla, D.R. (1983) Environ. Mutagen. 5, 745–765

3. Laporte, J.-R., Vidal, X., Balları́n, E., & Ibá~nez, L. (1998) Br. J.

Clin. Pharmacol. 46, 181–184

4. Borton, C., West, G.A.D., Clement, J.C., & Williams, D.M.

(1988) Rev. Infect. Dis. 10, 1048–1051

5. Veach, B.T., Baker, C.A., Kibbey, J.H., Fong, A., Broadaway, B.J., & Drake, C.P. (2015) J. AOAC Int. 98, 588–594

6. Chu, P.-S., & Lopez, M.I. (2005) J. Agric. Food Chem. 53, 8934–8939

7. Chu, P.-S., Lopez, M.I., Abraham, A., El Said, K.R., & Plakas, S.M. (2008) J. Agric. Food Chem. 56, 8030–8034

8. Veach, B., Hawk, H., Baker, C., & Linder, S. (2011) Lab Inf. Bull.

27, 1–15

9. Veach, B., Baker, C., Drake, C., Clark, S., Storey, J., & Turnipseed, S. (2012) Lab Inf. Bull. 1–41

10. Veach, B., Johnson, S., Rimmer, D., Drake, C., Fong, A., Delgado, Q., et al. (2014) Lab Inf. Bull. US FDA 1–28

11. An, H., Cain, T., Henry, M., Tran, B., Paek, H.C., & Farley, D.

(2011) FDA/ORA/DFS Lab Inf. Bull. 27, 1–35

12. Sørensen, L.K., Elbaek, T.H., & Hansen, H. (2003) J. AOAC Int.

86, 703–706

13. McCracken, R.J., & Kennedy, D.G. (1997) J. Chromatogr.

B Biomed. Sci. Appl. 691, 87–94

14. McCracken, R.J., Spence, D.E., Floyd, S.D., & Kennedy, D.G.

(2001) Food Addit. Contam. 18, 954–959

15. Verdon, E.C.P., & Sanders, P. (2007) Anal. Chim. Acta 586, 336–347

16. Radovnikovic, A., Conroy, E.-R., Gibney, M., O’Mahony, J., & Danaher, M. (2013) Food Addit. Contam. 30, 1858–1869

17. (2003) Fed. Regist. 31938–31939

18. (2015) Guidelines for the Validation of Chemical Methods for the

FDA Foods Program, 2nd ed. FDA U (Ed), US FDA: Washington

DC

Veach: Journal of AOAC INTERNATIONAL Vol. 103, No. 4, 2020 | 1051

D ow nloaded from https://academ ic.oup.com /jaoac/article/103/4/1043/5851561 by FD

A M edical Library - C orp Blvd user on 10 D ecem ber 2020 http://www.ars.usda.gov/News/docs.htm?docid=15572 http://www.ars.usda.gov/News/docs.htm?docid=15572 http://www.ars.usda.gov/News/docs.htm?docid=15572