CUN. CHEM. 40/2, (1994) #{149} Drug Monitoring and Toxicology

Transcription

1 CUN. CHEM. 40/2, (994) #{49} Drug Monitoring and Toxicology Characterization of Drug Interferences Caused by Coelution of Substances in Gas Chromatography/Mass Spectrometry Confirmation of Targeted Drugs in Full-Scan and Selected-Ion Monitoring Modes Alan H. B. Wu, 3 Donna Ostheimer, Michael Creniese, Elaine Forte, and Dennis Hill2 Interference by substances coeluting with targeted drugs is a general problem for gas chromatographic/mass spectrometric analysis of urine. To characterize these interferences, we examined human urine samples containing benzoylecgonine and fluconazole, and other drug combinations including deuterated internal standards that coelute (lsdc) with target drugs, by selected-ion monitoring (SIM) and full-scan mass spectrometry. We show that, by SIM analysis, detecting the presence of an interferent is dependent on the specific IS used for the assay. When an lsdcis used, the presence of another coeluting substance (interferent) suggests that the intensity of IS ions is substantially diminished, because the interferent affects both the lsdc and target drug. When a noncoeluting IS (IS,,.,Jis used, the interferent cannot be discerned unless it coincidently contains one or more of the ions monitored for either the target drug or ls. Under full-scan analysis, a coeluting interferent is directly discernable by examining the total ion gas chromatogram. Indexing Terms: fluconazole/cocaine/workplace drug testing/ urine/abused drugs Gas chromatography/mass spectrometry (GC/MS) is considered the gold standard for definitive confirmation analysis for drugs of abuse in urine.4 However, unknown substances in urine at high relative concentrations can interfere with targeted drug analysis. Optimum quantitation requires extraction and separation of all potentially interfering drugs and substances from the assayed drug before injection into a GC/MS analyzer. When an interfering substance that coelutes with the target drug is present at low concentrations, the ions from the target drug can be extracted from the full mass spectrum, and ion ratios can be used to confirm the presence of the drug of interest. If, however, the coeluting compound has ions in common with the target drug, ion ratios will not fall within the expected range. In addition, if a coeluting drug is present at high concen- Forensic Toxicology Laboratory, Hartford Hospital, Hartford, CT Fax Mia-ochemiy Laboratory, U-93, University of Connecticut, Storrs, CT Author for correspondence. 4Nonstandard abbreviations: GC/MS, gas chromatography! mass spectrometry; FLUC, fluconazole; BE, benzoylecgornne; TMS, trimethylsilyl; BSTFA, N,O-bis(trimethylsilyl)trifluoroacetamide; TMCS, trimethylchlorosilane; SIM, selected-ion momtorin, TLC, thin-layer chromatography; 5d,c, internal standard, deuterated, coeluting, and IS,,,, internal standard, noncoeluting. Received February 8, 993; accepted September 29, 993. trations, saturation of the ion source can lead to false detection (). Therefore, when new GC/MS drug assays are developed, drug analogs, metabolites, and other substances should be tested for their interference potential. Interfering drugs and substances have led to falsepositive results for methamphetamine (2), and falsenegative results for the marijuana metabolite z#{76}-carboxytetrahydrocannabinol (3). Because it is not possible to test all combinations of drugs, metabolites, and endogenous compounds, it is incumbent on the analyst to review carefully all data to determine the invalidity, particularly when results from screening tests (e.g., immunoassay) do not match GCIMS confirmations. The National Laboratory Certification Program requires active review of all GC/MS results by a qualified CertilS - ing Scientist for all federally mandated drug-testing programs (4). Review of data should include an examination for the presence of potentially interfering substances. We recently reported on the interference of fluconazole (FLUC) in the GC/MS confirmation analysis of benzoylecgonine (BE) following derivatization to trimethylsilyl (TMS) analogs (5). FLUC is an antifungal drug approved for use in 990 for treatment of patients with acquired immune deficiency syndrome (6). We tested four urine samples from patients on FLUC who were also positive for BE by both the EMIT (Syva Co., Palo Alto, CA) and TDx (Abbott Labs., Abbott Park, IL) immunoassays. Each of these samples was negative for BE by GC/MS confirmation after der-ivatization to the TMS analog. The FLUC derivative under full-scan mass spectral analysis coeluted with the BE derivative and was present in urine at much higher concentrations than the cutoff used for BE (300 p.gfl). We performed FLUC addition studies to characterize the nature of the interference by FLUC on the analysis of BE. In a similar manner, we studied the effect of high concentrations of propoxyphene and methaqualone on the GC/MS response of their coeluting deuterated analogs, which are often used as internal standards (ISa.,). We discuss the general impact of coeluting drugs and IS on quantitative GC/MS analysis and devise strategies for detecting these interferents in urine when either selected-ion monitoring (SIM) or full-scan mass spectral analysis modes are used. MaterIals and Methods Reagents. All reagents and solvents were of analytical grade or better. BE ( g/l in methanol), methaqualone ( g/l), and the 5dc BE-d3 tetrahydrate (00 mg/l) were 26 CUNICAL CHEMISTRY, Vol. 40, No. 2, 994

2 from Ailtech (Deerfield, IL). Difluoro-BE (00 mg/l) was from ElSohly Labs. (Oxford. MS). D-Propoxyphene ( g L), propoxyphene-d7 (00 mgfl), and methaqualone-d4 (00 mg/l were from Sigma Chemical Co. (St. Louis, MO). N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA) with 0 mill triinethylchlorosilane (ThICS) was from Pierce Chemical Co. (Rockford, IL). FLUC was obtained in purified form as a gift from the manufacturer (Pfizer, New York, NY). We prepared a working solution of FLUC (250 mg L) by dissolving 25 mg in 00 ml of methanol Solidphase extraction columns (Clean Screen DAU, 200 mg) were from Worldwide Monitoring Corp. (Horsham, PA). Patients samples. This study was conducted in accordance with the Helsinki Declaration of 975, as revised in 983. Urine was collected without preservatives from four patients on FLUC who were positive by immunoassay screening for BE and negative by GC/MS confirmation. Urine was collected from three other patients who were taking FLUC but were negative for BE by immunoassay screening. All samples were tested at Hartford Hospital, Hartford, CT, either for clinical toxicology purposes or compliance drug testing. Urines were screened within 24 h after collection and stored at 4#{76}C until analysis by GCMS. Assays. We followed the manufacturers recommendations and screened urine for BE with the EMIT assay on the ETS analyzer. (Syva) and with the fluorescence polarization immunoassay on the TDX analyzer (Abbott). The cutoff concentration in both assays was 300,u.g/L. For GC/MS confirmation analysis, urine samples positive for BE were extracted from copolymeric cationexchange/hydrophobic chromatographic columns. The column was conditioned with methanol, deionized water, and 00 mmol/l phosphate buffer. A 5-mL sample and 50 L of IS (50 gfl) were diluted with phosphate buffer (ph 6.0) and applied to the column. The column was washed with deionized water, 00 mmol/l HC, and methanol. BE was eluted with a mixture of methylene chloride, isopropanol, and ammonium hydroxide (78: 20:2 by vol). The solvent was evaporated to dryness, and the residue was derivatized at 65#{76}Cfor 5 miii with 75 L of BSTFA (with 0 mlil TMCS) and 25 L of ethyl acetate. The derivatized extracts were injected onto the GC/MS analyzer (5890/5970B MSD; Hewlett-Packard, Palo Alto, CA) with an HP Ultra capillary column [2 m x 0.2 mm (i.d.) x 0.33 m film thickness] and electron impact ionization. The injector was operated in the splitless mode at 275#{76}C. Because of differences in sensitivity with different GCiMS modes, the sample volume was 2 pl for full scan and jtl for SIM. The oven temperature was ramped from 50 to 280#{76}C at a rate of 30#{76}C per minute. The carrier gas was helium at a flow rate of mllmin. Targeted drugs were identified by full scan analysis (mass range amu, scan rate.36 scan/s) and SIM (dwell time 00 ms). The quantitative ions for both full scan and SIM were m/z 240,243, and 397 for BE-TMS, BE-TMS-d3, and difluoro-be- TMS, respectively. With SIM, ion ratios were calculated from the m/z 240, 256, and 36 ions for BE-TMS, and m/z 243, 259, and 364 for BE-TMS-d3. We also analyzed urine from one of the patients who was positive for both FLUC and BE by GC/MS after liquid-liquid extraction and isolation from a thin-layer chromatography (TLC) system (Toxi-Lab A; Toxi-Lab, Irvine, CA). We followed the manufacturer s recommendations up to the detection procedure, at which point we cut a portion of the chromatogram at the R value expected for BE, eluted the drug off the chromatogram with the BE elution solvent used for the solid-phase extraction, derivatized with BSVrA, and analyzed for BE-TMS by GCIMS, as described above. Experimental procedures. We prepared a methanolic BE standard at 500 g/l and added various amounts of FLUC to produce FLUC-TMSIBE-TMS ratios ranging from 0 to 00. These samples also contained two different IS: BE-d3 ( 5d,c), which coelutes with BE and FLUC when derivatized to TMS analogs in our assay, and noncoeluting difluoro-be (IS,,,), which elutes several seconds earlier than these drugs. Each was present at a concentration of 50 pgfl. These samples were injected into the GC/MS analyzer and monitored by full-scan mass spectral analysis, then reinjected into the GCIMS analyzer and monitored in the SIM mode. To determine whether coelution is a generic problem, we reversed the above procedure by adding to a 500 j.gfl FLUC standard various amounts of BE to produce BE- TMS/FLUC-TMS ratios ranging from 0 to 00. We also studied the effect of other drugs with their ISdC. We tested mixtures of drug standards in methanol containing propoxyphene-d7 at 500.&g/Land added increasing amounts of propoxyphene, and to 500.gfL methnqualone-d4 standard we added increasing amounts of methaqualone. Ratios were varied from 0 to 00. For propoxyphene, propoxyphene-d7, methaqualone, and methaqualone-d.f, the mlz values were 250,257,250, and 254, respectively. Results Fig., A and B, shows a total ion gas chromatogram and mass spectrum for a urine sample that was found to be positive for BE by immunoassay and negative for BE by GC/MS. Similar results were obtained for the other three clinical samples containing FLUC and BE. The mass spectrum of BE-TMS is shown in Fig. C for comparison. On the basis of analyzed controls and standards, the expected retention time for BE-TMS and BE- TMS-d3 for this analysis was 3.6 min. In this chromatogram, however, a very broad interfering GC peak was present between 3.55 and 3.75 mm. The presence of BE-TMS in this region of the chromatogram could not be detected by either the background-corrected full-scan spectrum or the reconstructed ion-chromatographic ratios. When TLC was used to separate BE from the interferent on one of these urine samples, however, a positive GC/MS confirmation for BE was obtained (data not shown). A review of available medical records revealed that in three of these cases, FLUC was prescribed (the medical record on the fourth case was not available for review). CLINICALCHEMISTRY, Vol. 40, No. 2,

3 4:44!82 FLUC-TMS Time (mm) FLUC-TMS Results of BE analyses with various concentrations of FLUC under full scan analysis and SIM are shown in Fig. 2. The peak areas for the monitored ions under both measuring modes decreased as the concentration of FLUC and ratio of FLUC to BE increased. Note also that the ions for the 5d,c BE-TMS-d3, also decreased to the same extent as BE-TMS with increasing FLUC concentrations. In contrast, no changes in peak areas were seen with the IS,, (difluoro-be-tms). Table shows the ion ratios for BE-TMS obtained by SIM analysis. When we used criteria of 20%, these ratios began to fluctuate when FLUC was present in twofold excess over BE- TMS. At 20-fold excess FLUC, all ion ratios were significantly altered. These FLUC concentrations are well within the urinary concentration range that would be expected after therapeutic use (7). For BE-TMS-d3, ion ratios diminished at 0-fold FLUC concentrations. maccuracies for the IS began at a lower FLUC concentration because the m/z 243/364 and 259/364 ratios were also Mass/Charge 363 RelatIve response (%) 82 BE-TMS j Mass/Charg#{233} Fig.. (A) Total Ion gas chromatogram Of a urine extract from a patient on FLUC who was positive for BE by Immunoassay screening; (B) mass spectrum of FLUC-TMS from the urine of patient in A; (C) mass spectrum of a solventstandard for BE-TMS; retention time for BE,3.6 mm. To determine whether FLUC cross-reacts with BE antibodies used in screening immunoassays, we added FLUC from 2.5 to 250 mg/l to a drug-free urine sample and found negative results or BE by immunoassay. We also obtained urine from other patients who were given equivalent amounts of FLUC. Each was negative for BE both by inununoassay and GC/MS. When we added BE at 500,ugiL to one of the samples, it became positive by immunoassay; in the subsequent GC/MS analysis, however, the presence of BE could not be confirmed. The GC/MS analysis of a FLUC standard prodtlced a mass spectrum for the TMS derivative that was identical to the interferent observed in urine of patients that had received FLUC. We determined that the observed mass spectrum was consistent with the proposed structure of FLUC-TMS. The molecular ion was absent after electron impact ionization. The m/z 363 resulted from the loss of a methyl group. The m/z 296 resulted from the loss of one of the triazole groups with a hydrogen rearrangement. The m/z 73 resulted from the loss of TMS as often seen in these derivatives Fiuconazole Concentration (mg/l) Relative response (%) Fluconazoie Concentration (mg/l) Fig. 2. Effect of Increasing concentrations of FLUC-TMS on the GC/MS response relative to the zero FLUC reference sample. (#{49}) BE-TMS; (+) dlfluoro.be-tms; (s) BE-TMS-d3. (A) Full-scan mass spectral analysis. The amount of BE-TMS Injected into the GC/MS was held constant at 25 ng. (B) SIM mode. The amount of BE-TMS Injected was held constant at 2.5 ng. 28 CUNICAL CHEMISTRY, Vol. 40, No. 2, 994

4 Table. Ion ratios of peak areas for BE by GC/MS SIM analysis In the presence of IncreasIng FLUC concentrations. FLUC-TMS/ BE -TMS ratios BE-TMS-d3 ratios BE-TMS ratloa 240/ /36 256/36 243/ / / b 0947b 0#{49}54b b 4b 0b 002b 20 7#{49}4b 427b 0jjb b 0490b 0#{23}jj,b 40 3b 382L 7b 8b 0309b oc7b 60 3#{49}2L 359b 4b 699b ofllb 0034b 80 3o3 73b 09b 74b 0b #{23}jb 00 2pb 330b 6b b 0.26 a Total amount of BE-TMS Injected into the GC/MS analyzer was held constant at 2.5 ng. b Ion ratio outside 20% of BE ratio of the reference standard (BE in the absence of FLUC). influenced by the increasing amounts of the M+ (m/z 364) ion for FLUC. Fig. 3 shows the reconstructed ion chromatograms for BE-TMS in the presence of increasing ratios of FLUC-TMSIBE-TMS. Although the peak heights are diminished, no distortions of peak shapes are evident in these curves. On the basis of these chromatograms, it would be difficult to determine if an interfering substance was present. Similar results were obtained when the FLUC concentration was kept constant and increasing concentrations of BE were added. As shown in Fig. 4, the apparent FLUC concentration decreased when the BE-TMS/ FLUC-TMS ratio exceeded 40 to. Other drug combinations also showed direct interferences when present at saturating concentrations. Fig. 4 shows the decrease in response in the presence of increasing amounts of the coeluting nondeuterated drug. These data suggest that the false-negative results for BE observed for the clinical samples were not due to interactions specific to FLUC itself, but that high relative concentrations of other coeluting compounds can influence the detection a, 0 C 0 C -o FLUC/BE-TMS Ratio Time Fig. 3. Reconstructed Ion chromatograms for mlz 240 under SIM mode for BE-TMS with increasingratios of FLUC. The amount of BE-TMSinjectedwasheldconstant at 2.5 ng.interference by FLUC-TMScannot be dcerned by qualitative examination of peak shapes at,n z240. Relative response (%) interferent/target Drug Ratio Fig. 4. The effect of increasing concentrations of (U) BE, (4.) propoxyphene-d7, and (*) methaqualone-d4 on the GC/MS response relative to the zero reference sample for FLUC, propoxyphene, and methaqualone, respectively. and linearity of GC/MS analysis. At low interferent/ target drug ratios, the response relative to the reference (zero interferent) sample was >00%. This was unexpected, and we are not certain as to the cause of these observations. The ionization efficiency may be better at these drug concentrations. Discussion For workplace drug testing, a major effort is placed on ensuring against false-positive results so that individuals who have undergone testing are not falsely accused. Federal guidelines for drug testing state that any laboratory returning a false-positive result on a proficiency survey will partially or completely lose its certification. Because of problems in false-positive reporting of methamphetamine, the US Department of Health and Human Services has adopted more stringent criteria for reporting methamphetamine confirmation (8). It is also important that false-negative results be kept at a minimum so that individuals with evidence of drug abuse do not enter or remain in the workplace. False-negative results compromise the integrity of the work force in safety-sensitive occupations. The interference in the GC/MS analysis of BE by FLUC occurred because our routine extraction procedure was not able to separate BE from FLUC. We also found that other drug combinations have the potential to produce interferences if they are present in high concentrations in urine and coelute with the targeted drug. To prevent false-negative results, it is important for laboratories to develop a strategy to detect the presence of coeluting compounds. However, identifying the existence of a potentially interfering drug or substance deponds on the ionization and detection mode of the mass spectrometric measurement and proper interpretation of results. With SIM analysis, detection of a coeluting interferent depends on the IS used for the assay. Use of ISdC is CLINICAL CHEMISTRY, Vol. 40, No. 2,

5 highly desirable because it mimics the extraction characteristics and recovery of the target drug. It also enables detection of a coeluting compound, because saturation of the ionization chamber in the presence of a coeluting substance at high concentration will also decrease the peak area of the IS. But to detect this situation, it is incumbent on laboratories that use SIM with coeluting IS to either compare the intensity of IS ions relative to that of a calibration standard, or to calculate ion ratios and apply similar acceptance criteria, as for the target drug. A low relative response of the 5d,c for a given sample and ion ratios outside expected ranges may indicate the presence of a coeluting compound. In the example in Fig. 2B, a 60% decrease in the response of BE-TMS-d3 and low ion ratios for m/z 243/364 and 259/364 (observed when BE-TMS/FLUC ratio was 0) implicated the presence of an interfering substance. Many laboratories use IS that do not coelute with the target drug (9,0). However, when SIM analysis is used in this situation, there are no data to indicate when an interfering substance might be present, unless the mass fragmentation pattern coincidentally contains one or more ions that are being monitored for the target drug. In Fig. 2B, increasing concentrations of FLUC had no effect on the response of the IS,, and a false-negative result for BE would have been produced if this interferent were present at high concentrations. Furthermore, reconstructed ion chromatograms are not sufficiently distorted to indicate a problem (Fig. 3). On the basis of these findings, we recommend use of an IS if the SIM mode is used, because this provides a mechanism for detecting coeluting interferents. Other studies have shown that use of ISd,C might also be helpful in determining the presence of interferents that produce falsenegative results by consuming the derivatization reagent (3). The most obvious manner by which an interfering substance can be identified is with full-scan MS analysis. Not only is the presence of a large coeluting peak directly discernable by examination of the total ion gas chromatogram (see, e.g., Fig. ), but identification of the interfering substance might be made with the mass spectrum. In addition, it might be possible to extract the ions expected for the targeted drug from the mixed mass spectra by background subtraction, although the relative ion ratios would probably be altered beyond acceptable limits. Such alteration would be most likely to occur if the coeluting compound contained any significant ions in common with the targeted drug. In conclusion, these results indicate that when GC/MS assays are being developed, strict attention must be given to effective urine extraction procedures to minimize the incidence of coeluting substances. Irrespective of the GC/MS analytical procedure, when interfering substances are present, correction of the problem requires reanalysis of the urine under alternative extraction or gas-chromatographic conditions. We have described the use of TLC to further puri urine samples containing FLUC. Changing other gas-chromatographic conditions such as using different column lengths or stationary phases, or changing the temperature program, may be effective in separating targeted drugs from coeluting interferents. References. Macbury GD, Turlington JM, Thschall JR, Kantor EJ. Unusual non-linearity effects observed in the analysis of 2,3,7,8-TCDD: implications for quantitative accuracy. Chemosphere 989;8: Wu AHB, Wong SS, Johnson KG, Ballatore A, Seifert WE. The conversion of ephedrine to methamphetamine and methainphetamine-like compounds during and prior to gas chromatographic/ mass spectrometric analysis of CB and HFB derivatives. Biol Mass Spectrom 992;2: Brunk SD. False negative CC/MS assay for carboxy THC due to ibuprofen interference. J Anal Toxicol 988;2: Department of Health and Human Services. Mandatory guidelines for federal workplace drug testing programs; final guidelines notice. Fed Regist 988;53: & Anderson D, Cremese M, Forte E, Wu A, Hill D. False negative results for benzoylecgonine in urine by GC/MS due to fluconazole [Abstract]. Clin Chem 993;39: Byrne WR, Wajszczuk CP. Crytpcoccal meningitis in the acquired immunodeficiency syndrome (AIDS): successful treatment with fluconazole after failure of amphotericin B. Ann Intern Med 988;08: Tucker EM, Wiinm PL, Arathoon EG, Levine BE, Hartatein Al, Hanson LH, Stevens DA. Pharmacokinetics of fluconazole in cerebrospinal fluid and serum in human coccidiodial meningitis. Antimicrob Agents Chemother 988;32: Department of Health and Human Services. Notice to DHHS/ NIDA certified laboratories. Dec 9, 990:-2. 9 Joern WA. Marijuana testing in urine: use of a hexadeuterated internal standard for extended linearity, and ion trap vs mass selective gas chromatograph/mass spectrometer systems. Clin Chem 992;38: Needleman SB, Romberg RW. Limits of linearity and detection for some drugs of abuse. J Anal Toxicol 990;4: CLINICAL CHEMISTRY, Vol. 40, No. 2, 994