Anal. Chem. 2003, 75, 4335-4340

Detection and Time Course of Cocaine N-Oxide and Other Cocaine Metabolites in Human Plasma by Liquid Chromatography/Tandem Mass Spectrometry Shen-Nan Lin,*,† Sharon L. Walsh,‡ David E. Moody,† and Rodger L. Foltz†

University of Utah Center for Human Toxicology, 20S 2030E, Room 490, Salt Lake City, Utah 84112-9457, and Behavioral Pharmacology Research Unit, Department of Psychiatry and Behavioral Sciences, Johns Hopkins University School of Medicine, 5510 Nathan Shock Drive, Baltimore, Maryland 21224

Gas chromatography/mass spectrometry (GC/MS) is often used for detection and measurement of cocaine metabolites in biological specimens. However, cocaine N-oxide, a recently identified metabolite of cocaine, is thermally degraded when introduced into a GC/MS. The major degradation products are cocaine and norcocaine. When cocaine N-oxide was measured in rat plasma using liquid chromatography in combination with electrospray ionization-mass spectrometry (LC/ESI-MS), the cocaine N-oxide concentrations in the rat plasma were reported to be as high as 30% of the cocaine concentrations. However, in our study involving LC/ESI-MS/MS analysis of plasma collected from human subjects following administration of oral cocaine, we determined that the concentrations of cocaine N-oxide relative to the cocaine concentrations never exceeded 3%. This suggests that determination of cocaine concentration in human plasma by GC/MS analysis will not significantly distort the actual cocaine concentrations due to thermal conversion of cocaine N-oxide to cocaine. In the work reported here, we compared results obtained using GC/MS, LC/ESI-MS/MS, and liquid chromatography/atmospheric pressure chemical ionization-tandem mass spectrometry (LC/APCI-MS/MS) to determine thermal degradation of cocaine N-oxide. LC/ ESI-MS/MS was selected to determine cocaine, benzoylecgonine, and cocaine N-oxide, and LC/APCI-MS/MS was selected to determine ecgonine methyl ester and norcocaine in plasma collected from three human subjects participating in a clinical study. The resulting time course data provide additional information into kinetic interrelationships between cocaine N-oxidation and cocaine hydrolysis. Cocaine is capable of producing severe hepatocellular necrosis in laboratory animals and in humans. The mechanism of cocaine hepatotoxicity is not well understood but appears to be associated * Corresponding author. Phone: (801) 581-5117. Fax (801) 581-5034. E-mail: [email protected]. † University of Utah Center for Human Toxicology. ‡ Johns Hopkins University School of Medicine. 10.1021/ac030037c CCC: $25.00 Published on Web 07/19/2003

© 2003 American Chemical Society

with actions of one or more N-oxidative metabolites of cocaine.1-3 Norcocaine, the most prominent N-oxidative metabolite, is reported to be formed by two alternative routes: (1) direct N-demethylation by cytochrome P450 and (2) formation of cocaine N-oxide by FAD-containing monooxygenase, followed by cyctochrome P450-mediated conversion to norcocaine and formaldehyde.4 Cocaine N-oxide has been detected in rat plasma5 and in human meconium.6 Liquid chromatography/electrospray ionization-mass spectrometry (LC/ESI-MS) was required for those studies because cocaine N-oxide is converted back to cocaine when samples are analyzed by gas chromatography/mass spectrometry (GC/MS). Cocaine N-oxide concentrations in rat plasma were reported as high as 30% of the cocaine concentration measured by LC/MS.5 Those observations have raised concerns regarding the accuracy of cocaine measurements in human plasma that are based on GC/MS analysis, because such a high percentage of cocaine N-oxide would cause cocaine concentrations to be overestimated. Cocaine N-oxide has not been previously reported in human plasma. In this study, we compared thermal stability of cocaine N-oxide under GC/MS, liquid chromatography/electrospray ionization tandem mass spectrometry (LC/ESI-MS/MS), and liquid chromatography/atmospheric pressure chemical ionization-tandem mass spectrometry (LC/APCI-MS/MS) conditions. Less than 0.5% of cocaine N-oxide decomposition occurred during LC/ESI-MS/ MS analysis. Therefore, LC/ESI-MS/MS was selected for simultaneous determinations of cocaine N-oxide, cocaine, and benzoylecgonine in human plasma samples collected at multiple time points after oral administration of cocaine. Concentrations of norcocaine and ecgonine methyl ester in the same plasma samples (1) Ndikum-Moffor, F. M.; Schoeb, T. R.; Roberts, S. M. J. Pharmacol. Exp. Ther. 1998, 284, 413-419. (2) Boess, F.; Ndikum-Moffor, F. M.; Boelsterli, U. A.; Roberts, S. M. Biochem. Pharmacol. 2000, 60, 615-623. (3) Lloyd, R. V.; Shuster, L.; Mason, R. P. Mol. Pharmacol. 1993, 43, 645648. (4) Kloss, M. W.; Rosen, G. M.; Rauckman, E. J. Mol. Pharmacol. 1983, 23, 482-485. (5) Wang, P. P.; Bartlett, M. G. J. Anal. Toxicol. 1999, 23, 62-66. (6) Xia, Y.; Wang, P. P.; Bartlett, M. G.; Solomon, H. M.; Busch, K. L. Anal. Chem. 2000, 72, 764-771.

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003 4335

were determined in separate runs using a modification of our validated LC/APCI-MS/MS assay.7 The data allow kinetic analysis of the metabolic interrelationships of cocaine N-oxide with cocaine and other oxidative and hydrolytic metabolites. EXPERIMENTAL SECTION Materials. Cocaine, cocaine-d3, benzoylecgonine, benzoylecgonine-d3, norcocaine, norcocaine-d3, ecgonine methyl ester, and ecgonine methyl ester-d3 were purchased from Radian International (now Cerilliant, Austin, TX). Cocaine N-oxide was provided by the National Institute on Drug Abuse (Bethesda, MD). Clinical Specimens. Blood was collected from three subjects taking part in a larger study concerning the time course, nature, and magnitude of withdrawal from repeated cocaine use. All subjects were healthy individuals with prior history of cocaine use. At the time of the study, they were residing in the Behavioral Psychology Research Unit at the Johns Hopkins University for a period of at least 40 days. Samples were collected after the first of a series of oral administrations of cocaine. Within a session, the dosing and blood collection protocols were similar to those previously described by Jufer et al.8 from a separate dose-ranging study. Cocaine (175 mg, po) was administered at 0, 1, 2, 3, and 4 h (total 875 mg, po, over 5 h). Blood was collected by venipuncture into gray-topped Vacutainer tubes just before the initial administration and then at 0.5, 1, 2, 3, 4, 5, 6, 7, 8, 12, and 24 h. Each blood sample was centrifuged soon after collection, and the plasma fraction was transferred to a separate container and stored at -20 °C. Samples were shipped on dry ice to the University of Utah and retained at -20 °C until analysis. Extraction. Deuterium-labeled isomers (25 ng each of cocained3, benzoylecgonine-d3, norcocaine-d3, and ecgonine methyl esterd3) were added as the internal standards to 1-mL aliquots of plasma specimens and calibrators. Calibrator samples for LC/ESI-MS/ MS analysis were freshly prepared just prior to extraction by spiking a series of 1-mL blank plasma samples with cocaine, benzoylecgonine, and cocaine N-oxide to give concentrations of 2.5, 5, 10, 30, 90, 350, and 750 ng/mL. Calibrator samples for LC/ APCI-MS/MS analyses were prepared by replacing the cocaine N-oxide with norcocaine and ecgonine methyl ester. Plasma samples were then extracted using our previously described solidphase extraction method for cocaine and benzoylecgonine.7 GC/MS. To investigate thermal degradation of cocaine N-oxide under GC/MS conditions, cocaine N-oxide was dissolved in ethyl acetate and analyzed using a HP5973 MSD with a HP6890 series GC. The capillary column was a DB5, 30 m × 0.25 mm i.d. with 0.25-µm film thickness (J&W, Folsom, CA). Helium was the carrier gas. The oven temperature was programmed starting from 140 °C for 1 min and then increased to 250 °C at 18 °C/min and held at 250 °C for 2 min. The injector temperature was 250 °C (unless stated otherwise). Sample molecules were subjected to positive ion chemical ionization (PCI) using methane as the reagent gas, and ions were detected by selective ion monitoring (SIM). LC/ESI-MS/MS and LC/APCI-MS/MS. Injections were made using a CTC-A200S autosampler (Finnigan MAT, San Jose, CA) into a Waters 626 liquid chromatograph (Milford, MA) with (7) Lin, S.-N.; Moody, D. E.; Bigelow, G. E.; Foltz, R. L J. Anal. Toxicol. 2001, 25, 497-503. (8) Jufer, R. A.; Walsh, S. L.; Cone, E. J. J. Anal. Toxicol. 1998, 22, 435-444.

4336

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

an Inertsil 5-µm particle size ODS-AQ (100 mm × 2 mm) LC column (YMC Waters, Milford, MA). The mobile phase consisted of 55% water containing 0.1% formic and 45% methanol with a flow rate of 0.15 mL/min. Mass spectrometry was performed using a Finnigan model TSQ-7000 triple quadrupole mass spectrometer with an ESI or an APCI interface to the liquid chromatograph. Conditions for ESI operations included a heated capillary temperature of 225 °C (unless stated otherwise), an electrospray voltage of 4.5 kV with the sheath gas pressure set at 60 psi, and auxiliary gas of nitrogen gas at the arbitrary unit of 12 on the flow meter associated with the TSQ-7000 mass spectrometer. Conditions for APCI operations included a heated capillary temperature of 150 °C, a thermal evaporator temperature of 375 °C (unless stated otherwise), and a corona source current of 4.5 µA with the nitrogen sheath gas pressure set at 30 psi. Cocaine, cocaine-d3, benzoylecgonine, benzoylecognine-d3, cocaine N-oxide, norcocaine, norcocaine-d3, ecgonine methyl ester, and ecgonine methyl ester-d3 were detected by selective reaction monitoring (SRM) with a collision energy of 20 eV and a collision gas (Ar) pressure of 2.6-3.0 mTorr. Calibration curves were established from the peak area ratios (cocaine/cocaine-d3, benzoylecgonine/benzoylecgonine-d3, cocaine N-oxide/cocaine-d3, norcocaine/norcocaine-d3, or ecgonine methyl ester/ecgonine methyl ester-d3) of calibrators from 2.5 to 750 ng/ mL, using linear regression with a 1/Y2 weighting. RESULTS AND DISCUSSION GC/MS of Cocaine N-Oxide. Wang and Bartlett have demonstrated that cocaine N-oxide in 1 µL of a 1 µg/mL concentration was 100% thermally degraded during GC/MS analysis, and cocaine was the only reported degradation product.5 In our study, we also observed that cocaine N-oxide in 1 µL of a 5 µg/mL concentration was undetectable by GC/MS analysis. Furthermore, in addition to cocaine as degradation product, we found that norcocaine was formed in substantial quantities at lower injector temperatures (<280 °C). We used a more sensitive and yet highly specific SIM GC/MS method to repeat the Wang and Bartlett experiments. We first established the GC retention times of cocaine (7.8 min) and norcocaine (7.6 min) and then obtained standard methane PCI spectra of cocaine and norcocaine. The cocaine spectrum had a major fragment ion (m/z 182) with an intensity of ∼45% of the protonated molecule ion (MH+, m/z 304), while norcocaine had a major fragment ion (m/z 168) with approximately the same intensity of the protonated molecule ion (MH+, m/z 290). The similarity in the PCI spectra between cocaine and norcocaine, both losing benzoic acid to form the prominent fragment ions, enabled us to predict the major ions in the PCI mass spectrum for cocaine N-oxide. In addition to a protonated molecule ion at m/z 320, loss of benzoic acid would lead to a fragment ion at m/z 198. Since cocaine N-oxide is completely degraded under these conditions, no actual full spectrum could be obtained.5 Therefore, the predicted ion fragments of m/z 320 and 198 were used in the SIM scan mode to test whether a trace of cocaine N-oxide could still be detected after injection of 1 µL of a 5 µg/mL cocaine N-oxide standard solution. The answer is negative. No cocaine N-oxide was detected (data not shown). The four SIM ions (m/z 304, 182, 290, 168) were then used for specific detection of cocaine and norcocaine after cocaine N-oxide injection.

Figure 1. Thermal conversion of cocaine N-oxide to cocaine and norcocaine during GC/MS analysis. (A) Total ion chromatogram with injection temperature at 225 °C. (B) Change in production of cocaine ([) and norcocaine (O) with change in injection temperature; no cocaine N-oxide (0) was detected.

Figure 2. (A-C) Comparison of SRM profiles resulting from injection of cocaine N-oxide into an LC/MS/MS using ESI. (A) Cocaine N-oxide (CONO), (B) cocaine (CO) either from interference (i) or decomposition (d), (C) norcocaine (NorCO) and BE, and (D) change in production of cocaine ([) and production of norcocaine (O) from cocaine N-oxide (0) with changes in heated capillary temperature.

At an injector port temperature of 225 °C, both cocaine and norcocaine appeared as the major chromatographic peaks (Figure 1A). The relative amounts of cocaine and norcocaine formed over a temperature range of 200-300 °C are shown in Figure 1B. When the injector temperature was raised to 300 °C, cocaine was the major degradation product and norcocaine was undetectable, an observation similar to that made by Wang and Bartlett.5 At an injector temperature of 250 °C, peak heights of norcocaine and cocaine were about equal. LC/ESI-MS/MS. To determine whether cocaine N-oxide undergoes any thermal degradation when subjected to ESI, a standard solution of 1 µg/mL cocaine N-oxide in a methanol/ water mixture (50:50) was infused into the ESI ion source at 10

Figure 3. Comparison of SRM profiles resulting from injection of cocaine N-oxide into an LC/MS/MS using APCI. (A) Cocaine N-oxide (CONO), (B) cocaine (CO-d) from decomposition, (C) norcocaine (NorCO-d) from decomposition, and (D) change in production of cocaine ([) and norcocaine (O) from cocaine N-oxide (0) with changes in heated evaporator temperature

µL/min. The heated capillary temperature was set at 225 °C. Mass spectra were recorded by scanning from m/z 285 to 325. The most intense ion in the spectrum (data not shown) occurred at m/z Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

4337

Figure 4. Analysis of the following plasma extracts by LC/ESI-MS/MS using SRM: (A) a blank plasma sample spiked with 2.5 ng/mL each of cocaine N-oxide (CONO), cocaine (CO), and benzoylecgonine (BE); (B) a plasma sample collected 15 min before oral administration of cocaine, and (C) a plasma sample collected 30 min after oral administration of cocaine.

320, consistent with the protonated molecule ion of cocaine N-oxide. The spectrum also contained an ion at m/z 304, consistent with the MH+ of cocaine. The intensity of the ion peak at m/z 304 was less than 5% of the intensity normalized to the ion peak of the m/z 320. No ions were detected at m/z 290, the MH+ of norcocaine. LC/ESI-MS/MS was then used to determine whether cocaine was an impurity in the cocaine N-oxide reference material or was formed by thermal degradation of cocaine N-oxide in the heated capillary of the ESI source. Ten microliters of cocaine N-oxide in 1 µg/mL concentration were injected. The mass spectrometer was set to monitor the SRM transitions: m/z 320 to 198 (Figure 2A, cocaine N-oxide); m/z 304 to 182 (Figure 2B, cocaine); and m/z 290 to 168 (Figure 2C, either norcocaine or benzoylecgonine). The heated capillary was set at 300 °C to enhance any possible thermal degradation of cocaine N-oxide. The intense cocaine N-oxide peak in Figure 2A shows a retention time of 2.77 min and a peak area of 1 140 306. Figure 2B shows two peaks due to cocaine; one has a retention time consistent with that of cocaine (2.40 min) and a peak area of 4735; the other has a retention time (2.81 min) nearly the same as that of cocaine N-oxide and a peak area of 5463. We conclude that the earlier-eluting peak, labeled CO-i, was due to cocaine that was present as an impurity in the cocaine N-oxide standard and that the later-eluting peak, labeled CO-d, resulted from thermal degradation of cocaine N-oxide. Two peaks in Figure 2C, labeled NorCO-i and BE-i, have retention times consistent with norcocaine (RT ) 2.58 min, peak area 13 578) and benzoylecgonine (RT ) 3.24 min, peak area 7725) and therefore also represented impurities in the cocaine N-oxide. Based on the peak area counts, the cocaine, norcocaine, and benzoylecgonine impurities appeared to account for less than 2.3% of the cocaine N-oxide preparation. The heated capillary temperature of the LC/ESI-MS/MS was optimized between 225 (peak area of 1 313 831) and 250 °C (peak area of 1 810 102) for detection of cocaine N-oxide. When the temperature was lowered to 200 °C, the cocaine peak resulting 4338

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

from thermal degradation of cocaine N-oxide was no longer detectable. At temperatures between 225 and 300 °C (Figure 2D), cocaine formed by thermal degradation of the injected cocaine N-oxide ranged from 4439 to 5463, corresponding to a thermal degradation of cocaine N-oxide of less than 0.5%. LC/APCI-MS/MS. The optimal temperature of the heated capillary for APCI operation for analysis of cocaine and benzoylecgonine is 150 °C.7 One would not anticipate decomposition of cocaine N-oxide at this temperature, judging from the results of our LC/ESI-MS/MS studies. However, another potential source of thermal degradation of cocaine N-oxide in the APCI source is the thermal evaporator, where heat is supplied to aid evaporation of the LC effluent before ionization by corona discharge. The optimal temperature of the thermal evaporator for cocaine and benzoylecgonine analysis is 375 °C.7 We investigated thermal degradation of cocaine N-oxide in the APCI source with the same set of SRM transitions as those used for the LC/ESI-MS/MS method; the same amount of cocaine N-oxide (10 µL of a 1 µg/ mL concentration) was injected to permit direct comparison between the two LC/MS/MS methods. Marked differences were evident in the measured area ion counts for cocaine N-oxide: 14 916 by APCI (Figure 3A) versus 1 140 306 by ESI (Figure 2A). Cocaine N-oxide was extensively decomposed during LC/APCIMS/MS analysis. Based on the relative area ion counts, ∼50% of the injected cocaine N-oxide was detected as cocaine (peak area of CO-d, 29 451, Figure 3B) and 24% as norcocaine (peak area of NorCO-d, 13 810, Figure 3C). Temperature dependence of this degradation is shown in Figure 3D. LC/ESI-MS/MS Analyses of Cocaine N-Oxide in Human Plasma Samples. LC/ESI-MS/MS was used to determine the time course of cocaine N-oxide in human plasma. Figure 4 compares LC/ESI-MS/MS chromatograms from analysis of the following samples: (1) a calibrator sample spiked with cocaine, cocaine N-oxide, and benzoylecgonine at the lower limit of quantitation, 2.5 ng/mL (Figure 4A); (2) a plasma sample collected 15 min before oral cocaine dosing (Figure 4B); and (3) a plasma

Table 1. Pharmacokinetics of Cocaine and Metabolites after Five Hourly Oral Doses of Cocainea analyte

Tmax (h)

Cmax (ng/mL)

AUC0-24 (h)(ng/mL)

cocaine benzoylecgonine ecgonine methyl ester norcocaine cocaine N-oxide

5.0 ( 0.0 5.7 ( 0.7 5.0 ( 0.0 5.0 ( 0.0 4.3 ( 0.3

620 ( 84 2970 ( 140 1440 ( 330 46.9 ( 12.5 18.0 ( 2.4

3080 ( 570 35600 ( 2800 12000 ( 2100 281 ( 33 76.7 ( 2.9

a Values are the mean ( SEM for three individuals. T max is the time to maximal plasma concentration and reflects time after the initial dose. the last dose was given at 4 h. Cmax is the maximal plasma concentration. AUC0-24 is the area under the plasma time concentration curve from 0 to 24 h.

Figure 5. Plasma concentration versus time (A) of cocaine, benzoylecgonine, and ecgonine methyl ester and (B) of norcocaine and cocaine N-oxide. Cocaine at 175 mg was orally administered at 0, 1, 2, 3, and 4 h. Each data point was the average of plasma samples from three subjects. LC/ESI-MS/MS method was used for cocaine, benzoylecgonine, and cocaine N-oxide measurements while LC/APCI-MS/MS was used for ecgonine methyl ester and norcocaine. The time course data were separated into two panels so that the low concentrations of norcocaine and cocaine N-oxide can be seen more clearly.

sample collected 30 min after oral cocaine dosing (Figure 4C). A cocaine N-oxide peak was not present at the retention time of 2.77 min in the sample collected 15 min before cocaine dosing (top panel of Figure 4B). The cocaine N-oxide peak was detectable in human plasma collected 30 min after cocaine dosing (top panel of Figure 4C). As expected, the cocaine and benzoylecgonine peaks are much larger in the SRM profiles from the 30-min plasma sample (second and fourth panels from the top, Figure 4C). Kinetics of Cocaine N-Oxide and Other Metabolites. With five successive hourly oral doses of cocaine, concentrations of cocaine, benzoylecgonine, ecgonine methyl ester, norcocaine, and cocaine N-oxide increased during the period of administration (Figure 5). Peak concentrations were reached at 5 h after the initial dose (1 h after the last dose) for all analytes except cocaine N-oxide, which peaked at 4 h for two subjects, and benzoylecgonine, which peaked at 6 h for two subjects (Figure 5 and Table 1). Highest concentrations were found for benzoylecgonine, followed by ecgonine methyl ester, cocaine, norcocaine, and cocaine N-oxide (Figure 5).

The areas under the curve (AUC) for 0-24 h showed a similar order (Table 1). When expressed as percentage of cocaine, the Cmax and AUC of cocaine N-oxide were 2.9 and 2.5%, respectively. A number of studies have examined the pharmacokinetics of cocaine and metabolites after administration by intravenous or nasal routes; few have followed both cocaine and metabolites after oral administration. An earlier study used a similar protocol for administration of oral cocaine,8 but the doses used were different from those administered in the present study. However, the rank order of cocaine and metabolite concentrations (excluding cocaine N-oxide) after five successive oral doses of 300 mg were the same. Furthermore, curves reported in the earlier study8 for AUC versus dose were extrapolated to the total dose of 875 mg used in our study, approximate values of 2800, 37 000, 11000, and 300 (h) × (ng/mL) were found for cocaine, benzoylecgonine, ecgonine methyl ester, and norcocaine, respectively, in good agreement with our results. Wang and Bartlett5 reported a very high level of 740 ng/mL cocaine N-oxide in rat plasma collected after 2 h of continuous IV infusion of cocaine (5.25 mg/mL) at 20 µL/min. Compared to our Cmax of 18 ng/mL in the plasma of orally dosed humans, the difference is remarkable, suggesting that the interspecies differences or the route of administration could have significant impact on the production of cocaine N-oxide. Since oral administration is normally associated with enhanced oxidative metabolism, we suspected that interspecies differences were the primary source of the difference in cocaine N-oxide concentrations seen between this and Wang and Bartlett’s study. CONCLUSIONS Cocaine N-oxide is thermally decomposed under conditions required for GC/MS determination of cocaine and metabolites, which explains, in part, why cocaine N-oxide had not until recently been demonstrated in human plasma. As a result, the physiological relevance of cocaine N-oxides in humans is not known. We have now detected cocaine N-oxide in human plasma using LC/MS methods and compared its concentration to cocaine and other metabolites. LC/APCI-MS/MS is also not suitable for measurement of cocaine N-oxide. We have demonstrated here that thermal degradation of cocaine N-oxide by LC/APCI-MS/MS is extensive, although not to the extreme of 100% when injected into a GC and that LC/ESI-MS/MS is the method of choice for detection and quantitation of cocaine N-oxide in biological samples. This study Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

4339

has shown that the conversion of cocaine N-oxide to cocaine at an electrospray heated capillary temperature of 300 °C is less than 0.5%.

Grant R01 AG12548-04A1. Portions of these findings were presented at the 48th annual conference of the American Society of Mass Spectrometry.

ACKNOWLEDGMENT The authors are grateful to Ruth Foltz for critical reading of the manuscript. This work was supported in part by U.S. Public Health Service Contract N01-DA-6-7052, Grant R29 DA10029, and

Received for review January 21, 2003. Accepted May 22, 2003.

4340

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

AC030037C