Anal. Chem. 2000, 72, 2711-2716
Liquid Chromatography/Mass Spectrometry for Timely Response in Regulatory Analyses: Identification of Pentobarbital in Dog Food David N. Heller
U.S. Food and Drug Administration, Center for Veterinary Medicine, 8401 Muirkirk Rd., Laurel, Maryland 20708
A limited liquid chromatography/mass spectrometry (LC/ MS) data set was acquired under conditions which called for timely response without benefit of a fully developed method. Quality assurance elements verified that an LC/ MS procedure developed in a short time was sufficiently under control to meet its purpose. LC/MS was used to rule out a potential problem with a gas chromatography (GC)/MS method that had been developed for regulatory purposes. The LC/MS data set showed that signals identified by GC/MS as diagnostic of pentobarbital (PB) were not artifacts of derivatization or GC analysis. Samples of dry dog food identified by GC/MS as containing PB were also shown by LC/MS to contain PB. The LC/MS method would not be recommended as a substitute for GC/MS, primarily because of poorer sensitivity. Although the data set is limited, and justifiably represents only the starting point for conventional method development, the purpose at hand was served adequately. This work demonstrates the utility of LC/MS for rapid regulatory response, provided there is a framework of quality assurance checks. The capability of LC/MS to address regulatory issues involving drugs and antibiotics has led to the development of many methods for such applications.1,2 In many of these cases, LC/MS facilitated rapid method development. This attribute can be of particular value in circumstances which call for timely response without benefit of a previously developed method.3,4,5 A recent case in the Center for Veterinary Medicine provided an opportunity to explore conditions for providing useful data with LC/MS in such circumstances. Our laboratory previously reported a procedure for identifying pentobarbital (PB) residues in dog food using GC/MS.6 Other regulatory analysts have also developed methods for euthanizing agents, including pentobarbital (PB) to investigate whether such residues occur in animal feed ingredients that incorporate some * Corresponding author: (E-mail)
[email protected]; (Tel.) 301-827-8156; (Fax) 301-827-8170. (1) Niessen, W. M. A. J. Chromatogr., A 1998, 812, 53-75. (2) Careri, M.; Mangia, A.; Musci, M. J. Chromatogr., A 1996, 727, 153-184. (3) Heller, D. N.; Ngoh, M. A. Rapid Commun. Mass Spectrom. 1998, 12, 20312040. (4) Biancotto, G.; Angeletti, R.; Piro, R. D. M.; Favretto, D.; Traldi, P. J. Mass Spectrom. 1997, 32, 781-784. (5) Saltron, F.; Berthoz, Y.; Rues, R.; Auguin, N.; Belhade, L. J. Mass Spectrom. 1996, 31, 810-818. (6) Adam, L. A.; Reeves, V. B. J. AOAC Int. 1998, 81, 359-367. 10.1021/ac9913053 CCC: $19.00 Published on Web 05/10/2000
© 2000 American Chemical Society
rendered euthanized animals.7,8 While applying our method in a survey of dry dog food for PB residues, a question was raised about the validity of a derivatization step. Since the derivatization transformed the target analyte, there was some concern that signals interpreted as PB might be artifactual instead. Although the GC/MS method had met a conventional in-house validation, the analytical situation could tolerate no uncertainty. Having identified a potential flaw in the GC/MS results which could leave the survey’s conclusions in doubt, it was necessary to address this concern in an appropriate manner. The nature of our response was typical of situations where time is of the essence, resources are limited, and suitable methods are not already available. It was not considered necessary to produce a second validated method. Rather, the GC/MS results could be corroborated by a limited study which firmly identified underivatized PB in extracts. LC/MS was chosen for this task not only because derivatization was not needed for analyte volatilization, but also for speed and flexibility. Unfortunately, the literature on LC/MS of barbiturates was not promising. Sensitivity by electrospray,9 thermospray,10 or particle beam11 LC/MS was found to be limited. For this work, pentobarbital was extracted from feed using the published procedure.6 Sample preparation differed at the point where extracts would have been derivatized for GC/MS analysis. Instead, extracts were dissolved in water/methanol, separated by reversed-phase gradient chromatography, ionized by negative-ionatmospheric pressure chemical ionization (NI-APCI), and detected by ion trap tandem mass spectrometry (MS/MS). EXPERIMENTAL SECTION Apparatus. All necessary materials were used as described by Adam and Reeves6 with the following substitutions and additions: The mass spectrometer was a model LCQ equipped with an APCI interface and controlled by Navigator software version 1.2 (Finnigan). The LC column was a PLRP-S, 150 × 2.1 mm, with 5-micrometer particles (Polymer Labs). The liquid chromatograph was a Series 1050 LC pump with a Series 1100 autosampler (Hewlett-Packard). Solid-phase extraction (SPE) was (7) O’Connor, J. J.; Stowe, C. M.; Robinson, R. R. Am. J. Vet. Res. 1985, 46, 1721-1724. (8) Hooijerink, D.; Schilt, R.; Brouwer, B.; van Bennekom, E. Analyst (Cambridge, U.K.) 1998, 123, 2513-2516. (9) Spell, J. C.; Srinivasan, K.; Stewart, J. T.; Bartlett, M. G. Rapid Commun. Mass Spectrom. 1998, 12, 890-894. (10) Lurie, I.; Cooper, D.; Krull, I. J. Chromatogr. 1993, 629, 143. (11) Ryan, T. J. Liq. Chromatogr. 1994, 17, 867.
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000 2711
Figure 1. Ion trap MS/MS fragmentation patterns for barbiturates, NI-APCI mode.
carried out with Bond Elut Certify II 10 cm3/200 mg cartridges (Varian). Reagents. Spectrophotometric-grade methanol, ethyl acetate, hexane, and acetonitrile were used (Burdick & Jackson). Water was purified through the Milli-Q system to a purity of >17 M-ohm/ cm (Millipore). Sodium acetate and glacial acetic acid were ACS certified grade (Fisher). Standards. Pentobarbital and phenobarbital (PH) were obtained from Sigma Chemical Co. as 1.0 mg/mL stock solutions in methanol. Standard solutions were prepared by diluting 100 uL of the 1.0 mg/mL standards to 10 mL with methanol to yield
Figure 2. Pentobarbital standard, equivalent to 50 ppb in feed. 2712
Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
a 10 ng/mL solution (10 ppm). One milliliter of each 10 ppm standard was then diluted to 100 mL with methanol (100 ppb). Appropriate volumes of diluted standards were further diluted with methanol and water to provide PB at 10, 20, 50, 100, 250, 500, 1000, and 2000 ppb and PH at 200 ppb. The extraction included a five-fold concentration, so these solutions were equivalent to PB at 2, 4, 10, 20, 50, 100, 200, and 400 ppb in feed and PH at 40 ppb in feed. The final diluent was water/methanol 80:20. Solutions. Other solutions were prepared as described in the GC/MS procedure.6 Extraction. The extraction was used as described6 except that the evaporation steps were carried out at 40 °C. (It had been shown that PB may be extracted by SFE at 40 °C.9) The dried extract was redissolved in 200 µL of methanol, the tubes were vortexed, and then 800 µL of water was added and the tubes were vortexed again. Extracts were filtered through a polypropylene syringe filter into autosampler vials. LC/MS Operating Conditions. The Finnigan LCQ was maintained and tuned according to the manufacturer’s operating specifications. The APCI source was operated in negative ion mode. Injection volume was 100 µL. The column was equilibrated in water/acetonitrile 80:20. The LC flow rate was 300 µL/min. The LC pump was programmed for gradient elution: linear ramp 0-4 min to 50:50 water/acetonitrile and hold 4-7 min; step to 10:90 acetonitrile and hold 7-9 min; ramp to 80:20 water/ acetonitrile 9-10 min; and equilibrate 10-15 min. LC flow was diverted to waste for the first three min after injection and again 9 min after injection. PB retention time was about 6.0 min in
Figure 3. Control corn meal fortified at 40 ppb (confirmed by LC/MS).
samples and 6.5 min in standards. PH retention time was about 7.1 min in samples and 7.5 min in standards. The column was equilibrated in initial mobile phase for 10 min at 0.3 mL/min before analyses were begun. The LC/MS operating parameters were optimized by using a Tee connector to infuse 10 ppm pentobarbital into the column effluent while pumping 40% ACN and 60% water at 300 µL/min. APCI conditions were optimized daily using the LCQ automatic tuning procedure. The LCQ was operated in MS/MS scan mode. Automatic Gain Control was on, and a 1000 ms maximum isolation time was used. Product ion spectra were acquired from PB [M H]- at m/z 225.2 and from PH [M - H]- at m/z 231.2. Isolation width was 2 amu, and 22% relative collision energy was used. The spectra were acquired in the range m/z 60-235. Selected ion chromatograms were generated for the base peak (m/z 182). The chromatogram was smoothed using a 9-point Gaussian smooth. Spectra were averaged across the time range on the ion chromatogram peak which was about 10% of full height or greater. The same time range was used for averaging control extracts or test samples where a clear peak did not appear. RESULTS AND DISCUSSION It was determined in advance that only a limited data set would be acquired. Acceptance of the LC/MS data set was conditional on the following elements, for the purpose of corroborating the GC/MS method: (1) Meet confirmation criteria that accord with Agency precedent (to ensure specificity with the same rigor as for fully developed methods).
(2) Carry out analyses on three different days (to demonstrate repeatability and analyst proficiency). (3) Require as a suitability test that an analytical standard meet confirmation criteria before continuing (to show the system was under control each day). (4) Require correct identification of a positive and negative control on each day (to show the procedure was under control, with no cross-contamination). (5) Employ solvent blanks (to show that carryover does not occur). (6) Provide for an analytical surrogate (PH) in all test samples (to show that each extraction performed acceptably.) (7) Samples had to be analyzed as soon as practical after extraction, pending a specific study of stability. (Standards were later found to be stable for several weeks.) (8) If a problem was documented during extraction or a sample gave unreasonable results for a specific, documented reason, the analysis could be be repeated. While these steps do not equate to development of a rugged, validated methodsnor would the outcome be considered as suchs they served the purpose at hand. The specificity of the data would be equal to that of fully validated methods, and checks were in place to show the procedure was under control during its application. Although full-scan MS/MS spectra were acquired, confirmation criteria were based on conventions for selected ion monitoring (SIM), i.e., there was a requirement for relative abundance matching. To be assigned confirmed status, a sample had to meet the following criteria for specificity: Analytical Chemistry, Vol. 72, No. 13, July 1, 2000
2713
Figure 4. Control corn meal (failed to confirm by LC/MS).
(9) A suspect peak had to show retention time ((2%) similar to average retention time of a fortified control analyzed that day. (Extracts showed different retention behavior than pure standards, and retention times were somewhat variable, probably due to different coextractants from various feeds.) (10) Product ions had to show coeluting behavior (to show they arose from a single source). (11) PB product ion chromatograms had to show signal-tonoise of at least 3:1 (to discriminate between random background and target analyte). (12) Suspect peaks had to show relative abundance (10% (arithmetic difference) of the relative abundance of standard analyzed that day (e.g., the acceptable range for a relative abundance of 50% would be 40-60%, not 45-55%). For peaks
