JOURNAL OF MASS SPECTROMETRY, VOL. 31, 1033-1038 (1996)

Determination of Alwazolam and a-Hydroxyalprazohm in Human Plasma by Gas Chromatography/Negative=ion Chemical Ionization Mass Spec trornetr y

Karin M. Hold? Utrecht Institute of Pharmaceutical Sciences (UIPS), Utrecht University, Department of Analysis and Toxicology, Sorbonnelaan 16,3584 CA Utrecht, The Netherlands

Dennis J. Crouch, Douglas E. Rollins and Diana G. Wilkins Center for Human Toxicology, Department of Pharmacology and Toxicology, University of Utah, 490 Biomedical Polymers Building, Salt Lake City, Utah 84112, USA

Dennis V. Canfield Toxicology and Accident Research Laboratory, FAA Aeronautical Center, Oklahoma City, Oklahoma 73125, USA

Robert A. M a e Utrecht Institute of Pharmaceutical Sciences (UIPS), Utrecht University, Department of Analysis and Toxicology, Sorbomelaan 16,3584 CA Utrecht, The Netherlands

A sensitive and specific method was developed for the determination of alprazolam and its major metabolite a-hydroxyalprazolam in plasma. After the addition of deuterium-labeled internal standards, plasma samples were buffered to pH 9 with 1 ml of saturated sodium borate buffer, extracted with toluen+methylene chloride (7 : 3) and evaporated to dryness. The residues were treated with N,O-bis(trimethylsily1)tnfluoroacetamide containing 1% of trimethylchlorosilane and analyzed on a Finnigan-MAT mass spectrometer operated in the negative-ion chemical ionization mode with methane as the reagent gas. Chromatographic separation was achieved on a Restek-200 capillary column using hydrogen as the carrier gas. The assay was linear from 0.25 to 50 ng m1-I for both compounds. The intra-assay precision for alprazolam was 16.1% at 0.5 ng ml-' and 4.6% at 50 ng ml-' and that for a-hydroxyalprazolam was 15.8% at 0.5 ng ml-I and 4.2% at 50 ng m1-l. The method was used to determine alprazolam and a-hydroxyalprazolam in human plasma samples collected after a single 2 mg oral dose of alprazo- lam. A peak concentration of 32.9 ng ml-' of alprazolam was detected at 1 h following the dose.

KEYWORDS: benzodiazepines; plasma; gas chromatography/mass spectrometry; metabolites; phar- macokinetics

INTRODUCTION

Alprazolam (Xanax) (AL) is a popular tri- azolobenzodiazepine. Since its introduction into clinical practice in the early 1980s, AL has become extensively used for the control of panic attacks and in the manage- ment of anxiety disorders.' It is prescribed in daily doses of up to 4 mg and is readily absorbed following oral doses.2

Many toxicology laboratories supporting clinical, medical examiner, drugs-and-driving and workplace drug testing analyze specimens for benzodiazepines (BZPs). Clinically, analytical testing may be used to confirm suspicious abuse of BZPs. Medical examiner investigations often focus on drug toxicity as a cause of death, and although death is rarely attributable solely to benzodiazepine ingestion, they are often taken in

t Author to whom correspondence should be addressed.

combination with prescribed or illegal drugs and may contribute to drug-related fatalities. BZPs can affect complex behavioral tasks such as driving. Specimens collected from impaired, injured and fatally injured drivers are often tested for BZPs. Screening tests may be performed on the urine or other biological speci- m e n ~ , ~ - ~ but quantitative blood values are needed for the interpretation.

Compared with the older BZPs, AL has a shorter half-life, a smaller dose is required for therapeutic effi- cacy and lower plasma concentrations are found. These factors result in the requirement for increased analytical sensitivity. Today, capillary gas chromatography (GC) with mass spectrometric (MS) detection is the method of choice for confirmation in both forensic and clinical applications. GC/MS offers a reliable, selective and sen- sitive technique capable of determining BZPs in a variety of biological matrices.'-' Published methods recommend extraction of the BZPs using an organic solvent or solid-phase extraction cartridge, forming tri- methylsilyl (TMS) derivatives of the polar metabolites

CCC 1076-5 174/96/09 1033-06 0 1996 by John Wiley & Sons, Ltd.

Received 3 M a y 1996 Accepted 1 1 June 1996

1034 K. M. HOLD ET AL.

and analysis by MS.3712 Javaid and Liskevych" showed that negative-ion chemical ionization (NCI) MS offers much higher sensitivity than positive-ion chemical ionization (PCI) or electron impact (EI) ionization for the determination of AL in plasma. Cairns et aL9 used GCWCI-MS to determine AL in blood and liver digest.

AL is extensively metabolized in man, the principal metabolites being a-hydroxyalprazolam (OH-AL), ben- zophenone and 4-hydroxyalprazolam,' of which only OH-AL has pharmacological activity.' Phar- rnacokinetic studies indicate that the a- and 4-hydroxy metabolites of AL can be detected in plasma, but that their concentrations (in unconjugated form) are con- siderably lower (less than 10%) than those of the parent drug.' However, there is increasing interest in determin- ing active metabolitis of psychoactive drugs in phar- macokinetic studies. A major objective of this study was to develop and validate a method that was sensitive enough to determine both AL and its active metabolite OH-AL in plasma after a single-dose administration.

EXPERIMENTAL

Reagents

AL, OH-AL and the internal standards [12H,]triazolam (TR-d,) and [2H,]OH-AL (OH-AL-d,) were obtained from Radian (Austin, TX, USA). High-purity-grade toluene and capillary GC/MS grade methylene chloride, ethyl acetate and methanol were obtained from Baxter (McGraw Park, IL, USA). Analytical-grade sodium borate was obtained from Mallinckrodt (St Louis, MO, USA). N,O-Bis(trimethylsily1)trifluoroacetamide con- taining 1% of trimethylchlorosilane (BSTFA f 1% TMCS) was purchased from Pierce Chemical (Rockford, IL, USA). Human drug-free plasma was out- dated plasma from local blood banks.

GC/MS analysis

Analyses were performed on a Finnigan-MAT 4500 GC/MS system equipped with INCOS software (Finnigan-MAT, San Jose, CA, USA). The chromato- graphic column was a Restek-200 capillary column (1 5 m x 0.25 mm i.d., 0.25 pm film thickness) (Restek, Belle- fonte, PA, USA). The initial column temperature of 190°C was held for 1 min, then increased to 320°C at 20 "C min- ', the final temperature being held for 2 min. The carrier gas was hydrogen with a head pressure of approximately 7 psi (1 psi = 6.895 Pa). The tem- peratures of the injection port, interface, transfer line and ionizer were 275, 310, 300 and 130°C, respectively. The reagent gas was methane adjusted to a source pres- sure of 0.60 Torr (1 Torr = 133.3 Pa). The emission current, electron energy, conversion dynode and multi- plier were set at 0.15 mA, -70 eV, - 3 kV and - 1500 V, respectively. The mass spectrometer was operated in the NCT detection mode. A 1 pl volume of sample was injected on to the column in the splitless mode. The mass spectrometer was set to monitor ion currents at m/z 308 and 310 for AL and TR-d,, and m/z 396 and

401 for the derivatized OH-AL and OH-AL-d,. Peak- height ratios were calculated, and the concentration of each analyte was determined from least-squares equa- tions generated from peak-height ratios of the cali- brators.

Stock solutions and extraction

Drug reference solutions were diluted in methanol to obtain separate stock solutions containing 1.0, 0.1 and 0.01 pg ml-' of AL and OH-AL. A similar procedure was used to make stock solutions, which were prepared from separate lot numbers of drug, for the preparation of quality control samples. Batches of controls were pre- pared at three different concentrations in plasma. Inter- nal standard solutions were combined and diluted in methanol to achieve a final concentration of 200 ng ml-' TR-d, and 100 ng ml-' OH-AL-dS. All solutions and controls were stored at - 20 "C until used.

Typical calibration solutions contained concentra- tions of AL and OH-AL of 0, 0.25, 0.50, 1.0, 2.5, 5.0, 7.5, 10.0, 25.0 and 50.0 ng ml-' and were prepared in 1 ml of drug-free plasma. To each tube was added 50 pl (10 ng of TR-8, and 5 ng of OH-AL-d,) of the internal standards solution and the tubes were mixed briefly and allowed to equilibrate for at least 30 min. A 1 ml volume of saturated sodium borate buffer was added to each tube. The tubes were mixed and 7 ml of toluene- methylene chloride (7 : 3) were added. All tubes were capped, rocked gently for 30 min and centrifuged for 10 min at 2500 rpm. The organic phase was transferred into clean silanized tubes and the solvent was evapor- ated at < 40 "C under a stream of air. A 25 p1 volume of ethyl acetate followed by 25 p1 of BSTFA + 1% TMCS were added to the tubes. All tubes were heated at 80°C for at least 30 min. The tubes were removed from the heating block and allowed to cool to room temperature. The liquid was transferred to labeled autosampler vials for injection into the GC/MS system.

Recovery

Recovery studies were performed by preparing two sets of samples (A and B). Set A consisted of samples with n = 10 at two different concentrations, 1 and 50 ng ml-'. This sample set was extracted using the method as presented. Samples in set B (n = 10) were extracted as described, except that the AL and OH-AL were added to the organic solvent just prior to evaporation. The ratio of the analyte peak height to the correspond- ing deuterated internal standard scale height was obtained and mean ratios were calculated for samples in sets A and B. The mean ratio at each of the two concen- trations for set A was divided by the mean ratio for the corresponding concentration in set B to obtain the recovery ratio. The recovery ratio was multiplied by 100 to calculate the percentage recovery.

Application of the method

The method was used to analyze plasma samples from seven male subjects receiving a single 2 mg oral dose of AL. Blood samples were collected at 0,0.5, 1,2, 3, 5 and

DETERMINATION OF a-HYDROXYALPRAZOLAM 1035

.- tl rn 0, LT -

50 -

Q-.

50

loo/ A

relationship between the peak-height ratio and the con- centration of each analyte up to 50 ng ml-'. However, to enhance the quantitative accuracy at low concentra- tions, the five closest calibration graph values were used. Tabe 1 shows the mean values of the slope and

OH-AL-TMS

, , intercept obtained from four calibration graphs. l ~ = ~ - i - - ~ ~ l - - - ~ ,

mlz 308

I

50 -

0

Figure 2 shows the GC separation of AL and its metabolite obtained with the column and temperature conditions given above. Selected ion chromatograms obtained after injection of a 2.5 ng ml-' extracted and derivatized calibrator are shown.

The accuracy and precision of the assay at the limit of quantitation (LOQ) and two additional concentrations in the linear range were established from intra- and inter-assay replicate analyses (Table 2). For intra-assay determination, each concentration studied was analyzed

OH-ALIS-TMS

I , , , , , , , , ,

o- LOO 2 50 300 350 400 450 Recovery

' O 0 1

m/z 310 The extraction recoveries for AL and OH-AL were 108.0% and 75.5%, respectively, at 1 ng ml-I in plasma and 100.8% and 99.4%, respectively, at 50 ng ml-'.

Figure 1. NCI mass spectra of alprazolam (AL), triazolam-d, (TR-d,), derivatized a-hydroxyalprazolarn (OH-AL-TMS) and derivatized GL- hydroxyalprazolam-d, (OH-AL-d,-TMS).

then used to determihe the inter-assay accuracy and precision. The mean, standard deviation and relative standard deviation (RSD) were calculated at each con- centration for AL and OH-AL. The results showed that the inter- and intra-assay accuracy and precision were in agreement with accepted validation procedures.14 The LOQ for the analysis of 1 ml of plasma was 1.0 ng ml-' for both AL and OH-AL.

10 h after the dose. Blood was centrifuged after collec- tion and plasma was stored frozen at - 20 "C until used.

RESULTS

GC/MS

Methane NCI mass spectra of AL (m/z 308), derivatized OH-AL (m/z 396), TR-d, (m/z 310) and derivatized OH-AL-d, (m/z 401) are shown in Fig. 1.

Effect of storage and freezethaw cycles

To determine the stability of AL and OH-AL, three ali- quots of the 5.0 ng ml-' control were analyzed after

Table I. Linearity of low and high concentration calibration graphs for alprazolam (AL) and a-hydroxyalprazolam (OH-AL) in plasma"

AL OH-AL

Linear range (ng ml-') y-l ntercept Siope y-Intercept Slope

0.255.0 -0.22 f 0.1 2 1 3.88 * 3.1 1 -0.1 k 0.05 5.76 * 0.36 2.550 1.52*0.76 10.18i1.09 -0.7k0.09 6.18*0.34

a Calibration graph equations were calculated from plots of concentration in ng ml-' on the ordinate and peak-height ratio on the abscissa. The intercepts and slopes are expressed as the mean *standard deviation for four graphs.

1036

310

K. M. HOLD ET AL.

-

396

Table 2. Inter- and intra-assay precision (RSD) and accuracy (n = 5) for alprazolam (AL) and a-hydroxyalprazolam (OH-AL)

AL OH-AL

Target concentration Accuracy RSD Accuracy RSD (ng ml-') (% of target) (%.) (YO of target) (W

0.5 95.6 16.1 102.4 15.8 5.0 1 14.4 5.2 104.0 6.9

50.0 99.0 4.6 109.6 4.2

Inter -assa y

Intra -assay

1 .o 105.6 16.2 108.4 5.9 5.0 106.4 14.7 103.8 8.9

50.0 101 .o 9.5 101 .o 9.0

401

storage at room temperature for 0, 3, 6 and 24 h (Table point. In addition, three control samples at 5.0 ng ml-' 3). When the mean concentrations were compared with were analyzed after two freeze-thaw cycles and com- the zero-time controls, neither compound showed a sig- pared with the unfrozen controls. The decrease in con- nificant (>lo%) decrease in concentration at any time centration after two freeze-thaw cycles for both

compounds was similar to that in the 3, 6 and 24 h experiments (< 10%).

Table 3. Stability of alprazolam (AL) and a-hydroxyalprazolam (OH-AL) in plasma

Time at ambient temperature' % of zero-time control

(h) AL OH-AL

0 100.0 100.0 3 92.7 102.3 6 100.1 98.7

24 100.1 101 .o *Quality control samples of 5.0 ng ml-' were placed at room temperature.

Application of the method

The pharmacokinetic disposition of AL of one represen- tative subject out of the seven subjects is shown in Fig. 3. AL was present after 1 h following an oral dose and in subsequent samples collected up to 10 h. The maximum concentration found was 32.9 ng ml- ' in plasma. No OH-AL could be detected in the samples.

DISCUSSION

BZPs are important to toxicologists because of their fre-

1037 DETERMINATION OF a-HYDROXYALPRAZOLAM

401

0 2 4 6 8 10 1 2

Time (h)

Figure 3. Pharmacokinetic disposition of alprazolam (AL) in a sample from one male subject.

quent use and their ability to affect the performance of tasks such as driving a motor vehicle or flying an air- craft. Moskowitz and Burns" showed that a 5 mg dose of diazepam was essentially equivalent to a 0.06% blood ethanol concentration in adversely affecting pursuit tracking, divided attention and information processing tasks. Diazepam also impaired subjects' ability to follow lanes, maintain a constant speed, make emergency deci- sions and perform effective visual searches in driving simulator studies. '

A major problem facing toxicologists is how to detect and determine AL and OH-AL in biological samples. Screening methods have been developed to detect metabolities of AL in However, these lack specificity and do not have sufficient sensitivity to detect the concentrations expected in blood or plasma follow- ing therapeutic administration. Studies have shown that peak plasma concentrations following oral or sub- lingual AL administration are - 10-20 ng ml-' per mg administered.' To attain this sensitivity, the extraction efficiency and the chromatographic and MS conditions need to be optimized.

A series of experiments were performed to assess the extraction efficiency and chromatographic cleanliness of various liquid-liquid and solid-phase extraction (SPE) methods. Solvents and extraction buffers were selected from those found in the literature and those unpub- lished but that predictably would be e f f e c t i ~ e . ~ - ' ~ ~ ' ~ , ' ~ A mixture of toluene and methylene chloride (7: 3 ) was the most efficient solvent and that gave chromato- graphically acceptable results. Bond Elut Certified and Worldwide Monitoring Clean Screen showed the most promise as potential SPE columns. However, solvent extraction with toluene-methylene chloride (7 : 3) was selected because it provided a good recovery and was far less time consuming than SPE protocols.

Analyses for AL, TR-d,, OH-AL and OH-AL-d,, underivatized or derivatized, on standard methylsilicone or phenyl-methylsilicone capillary columns resulted in unacceptable chromatographic peak shapes at low analyte concentrations. The Restek-200 column selected

is intermediate in polarity between these two standard columns because it contains trifluoropropyl groups in place of the phenyl groups on the stationary phase. This column produced good peak shapes and sensitivity without the excessive temperatures required with more polar columns. The chromatography was further improved through the use of hydrogen as the carrier gas. Figure 2 shows that the peak width at the baseline for AL is < 8 s and that for OH-AL is < 6 s. Using helium as the carrier gas, typical peak widths at the baseline were -50% greater for the analytes. The decreased peak width resulted in a greater signal-to- noise ratio and improved sensitivity.

Derivatization with BSTFA + 1% TMCS improved the peak shapes for all drugs. MS analysis demon- strated that TMS derivatives were formed with the hydroxylated metabolite and not with the parent drug and its internal standard (Fig. 1). Theoretically, only OH-AL and OH-AL-d, should form O-TMS deriv- atives and, therefore, demonstrate improved chromato- graphic characteristics. However, we believe that TMS may enhance the chromatography of AL and TR-d, by either associating with the drug or by deactivating the GC column. A disadvantage of TMS derivatives is that they are extremely prone to hydrolysis. Moisture intro- duced during the extraction or from the laboratory environment may produce hydrolysis. In a similar ana- lytical problem, we found that the tert- butyldimethylsilyl derivatives were more stable, less moisture sensitive and had higher m/z values than TMS derivative^.'^ Further, other workers have advocated the use of tert-butyldimethylsilyl derivatives for determination of OH-AL and a-hydroxytriazo1am.l Repeated experiments demonstrated that the TMS derivative was more easily and predictably formed and more stable than the tert-butyldimethylsilyl derivatives for OH-AL and OH-AL-d,.

Figure 1 shows the methane NCI mass spectra for the analytes. These spectra are simplistic compared with conventional EI mass spectra. The ion current is con- centrated in one major and two or three minor frag- ments representing the molecular anion, ' 3C and halogen isotope ratios. Only TR-d, fragmented by the loss of HCl. Since there is limited fragmentation, improved sensitivity is obtained over conventional EI-MS analysis. NCI has an additional inherent advan- tage over both PCI-MS and EI-MS detection since only compounds with electronegative components will be detected. This selectivity reduces the chemical noise encountered from other drugs or endogeneous sub- stances in biological specimens and results in improved signal-to-noise ratios. Fitzgerald et al." demonstrated that NCI may be 200 times more sensitive than EI or PCI for the detection of OH-AL. As with any analysis using deuterium-labeled internal standards, the poten- tial contribution of the internal standard to the ions monitored for the analyte being tested must be evalu- ated. TR-d, was selected as the internal standard for AL because the AL-d, evaluated contributed significantly to m/z 308. The AL-dS contributed 4.2% to AL-do and effectively limited the sensitivity to -0.5 ng ml-'. TR-8, contributed <1% to m/z 308 and was separated chro- matographically from AL, and therefore a sensitivity of 0.25 ng ml-' was routinely obtained.

1038 K. M. HOLD ET AL.

Table 2 shows that the control sample of 5.0 ng ml-' of AL gives an accuracy of 114.4%. Since separate solutions were used for calibration solutions and

CONCLUSION

control samples, one would expect an experimental error. Given the reproducibility of the assay (calibrators and controls) and accuracy at the other two concentra- tions, one would suspect that the 5.0 ng ml-' control might have been fortified at >5.0 ng ml-'. Also, the inter-assay value is the highest of the three. Neverthe- less, this value is within the limits for quantitation.

Figure 3 shows the plasma time course curve for a representative subject who had received a single 2 mg oral dose of AL. The analysis of this specimen was accomplished using the described method. The peak plasma concentration was reached at 1 h post-dose and was 32.9 ng m1-l. The concentration at 11 h post-dose was 12 ng ml-'. Given the average half-life of 12 h,* one would predict that AL could be detected in plasma for 2 days following a single oral dose with our method. Despite predictions that OH-AL would be in the plasma at a concentration of -10% of that of AL, it was not found.' This might be explained by excretion of the polar metabolite or by conjugation and excretion as theorized by Cairns.' Further, there are conflicting reports in the literature concerning the expected con- centrations of OH-AL and 4-hydroxy-AL. However, OH-AL has been identified in hydrolyzed urine from subjects by this method.

The described NCI-MS method for determination of AL and OH-AL is sensitive to 0.25 ng ml-I for both analytes and linear to 50 ng rn1-I. The method uses a simple extraction solvent and extraction of both AL and OH-AL is accomplished in one step, yet all recov- eries exceed 75%. BSTFA + 1% TMCS is used to derivatize OH-AL and OH-AL-dS, and improve the chromatography for AL and TR-d,. The method has been used for the analysis of plasma specimens collected from subjects who had been administered a single 2 mg oral dose of AL. A peak concentration of 32.9 ng ml-' of A1 was detected at 1 h following the dose. Because of the extremely low concentrations (< 1.0 ng rnl-'), OH-AL could not be determined. However, this method can be applied to measure the steady-state kinetics of AL and OH-AL in plasma.

Acknowledgements

Financial support for the development of this assay was provided by the Federal Aviation Administration, United States Department of Transportation (DTFA-0293A72B27) and the National Institutes of Health and National Institute on Drug Abuse (5R01-DA07820).

REFERENCES

1. D. J. Greenblatt and D. E. Wright, Clin. Pharmacokinet. 24,

2. G. W. Dawson, S. G. Jue and R . N. Brogden, Drugs 27, 132

3. L. E. Edinboro and A, Poklis, J. Anal. Toxicol. 18,312 (1 994). 4. A. 0. Fraser, J. Anal. Toxicol. 11,263 (1 987). 5. A. D. Fraser, J. Anal. Toxicol. 15,25 (1 991 ). 6. T. R . Koch, R. L. Raglin, S. Kirk and J. F. Bruni, J. Anal.

7. R. Meatherall,J.Ana/. Toxicoi, 18, 385 (1994). 8. W. Huang, D. E. Moody, D. M. Andrenyak and 0. E. Rollins,

J. Anal. Toxicol. 17, 365 (1 993). 9. E. R. Cairns, 6. R. Dent, J. C. Ouwerkerk and L. J. Porter, J.

Anal. Toxicol. 18, 1 (1 994). 10. J. I. Javaid and U. Liskevych, Biomed. Environ. Mass

Spectrom . 1 3, 1 29 (1 986). 11. W. A. Joern, J. Anal. Toxicol. 16, 363 (1 992). 12. D. A. Black, G. D. Clark, V. M. Haver, J. A. Garbin and A. J.

Saxon.J.Ana/. Toxicol. 18,185 (1994).

453 (1 993).

(1 984).

Toxicol. 18, 1 68 (1 994).

13. P. H. Dickson, W. Markus, J. McKernan and H. C. Nipper, J. Anal. Toxicol. 16,67 (1 992).

14. V. P. Shah, K. K. Midha, S. Dighe, I. J. McGilveray, J. P. Skelly, A. Yacobi, T. Layloff, 6. T. Viswanathan, C. E. Cook, R . D. McDowall, K. A. Pittman and S. Spector, J . Pharm. Sci. 81, 309 (1 992).

15. H. Moskowitz and M. Burns, in Proceedings of the 21st Con- ference of the American Association for Automotive Medicine, edited by D. Hulke pp. 266-240, AAAM (1 977).

16. J. M. Walsh. Effects of Drugs on Driving. Driving Simulator Tests of Secorbarbital, Diazepam, MarQuana, and Alcohol. National Institute of Drug Abuse, Rockville, MD (1985).

17. R. Meatherall,J.Anal. Toxicol. 18, 369 (1994). 18. R. L. Fitzgerald, D. A. Rexin and D. A. Herold, J. Anal. Toxicol.

19. D. J. Crouch, M. E. Alburges, A. C. Spanbauer, D. E. Rollins 17, 342 (1993).

and D. E. Moody, J. Anal. Toxicol. 19,352 (1 995).