Anal. Chem. 2002, 74, 1081-1087
Direct Determination of Benzodiazepines in Biological Fluids by Restricted-Access Solid-Phase Microextraction Wayne M. Mullett and Janusz Pawliszyn*
Department of Chemistry, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada
A biocompatible solid-phase microextraction (SPME) fiber was prepared using an alkyl-diol-silica (ADS) restrictedaccess material as the SPME coating. The ADS-SPME fiber was able to simultaneously fractionate the protein component from a biological sample, while directly extracting several benzodiazepines, overcoming the present disadvantages of direct sampling in biological matrixes by SPME. The fiber was interfaced with an HPLC-UV system, and an isocratic mobile phase was used to desorb, separate, and quantify the extracted compounds. The calculated clonazepam, oxazepam, temazepam, nordazepam, and diazepam detection limits were 600, 750, 333, 100, and 46 ng/mL in urine, respectively. The method was confirmed to be linear over the range of 50050 000 ng/mL with an average linear coefficient (R2) value of 0.9918. The injection repeatability and intraassay precision of the method were evaluated over 10 injections, resulting in a RSD of ∼6%. The ADS-SPME fiber was robust and simple to use, providing many direct extractions and subsequent determination of benzodiazepines in biological fluids. Solid-phase microextraction (SPME) is a sampling and sample preparation technique invented by Pawliszyn for volatile organic compound analysis in environmental samples over 10 years ago.1 SPME provides many advantages over conventional sampling methods by integrating sample extraction, concentration, and introduction into a single step. SPME has been successfully coupled with gas chromatography (GC), high-performance liquid chromatograph (HPLC), and capillary electrophoresis and has found numerous applications in many disciplines.2 Most recently, SPME has been extended to various aspects of biological sample analysis and has been the subject of several reviews,3-5 However, with respect to biological sample analysis using commercially available fibers, SPME has met some difficulties. The preferred extraction mode for the SPME analysis of biological samples is headspace extraction; it produces cleaner extracts and longer fiber lifetimes because of minimal fiber fouling (1) Belardi, R. G.; Pawliszyn, J. J. Water Pollut. Res. Can. 1989, 224, 179-184. (2) Pawliszyn, J., Ed. Applications of Solid-Phase Microextraction; Royal Society of Chemistry: Cambridge, U.K., 1999. (3) Theodoridis, G.; Koster, E. H. M.; de Jong, G. J. J. Chromtogr., B 2000, 745, 49-82. (4) Lord, H.; Pawliszyn, J. J. Chromatogr., A 2000, 902, 17-63. (5) Snow, N. H. J. Chromatogr., A 2000, 885, 445-455. 10.1021/ac010747n CCC: $22.00 Published on Web 01/24/2002
© 2002 American Chemical Society
resulting from protein adsorption during direct extraction.6-8 Unfortunately, most drug compounds are semi- or nonvolatile organic compounds, making SPME headspace extraction of body fluids impossible to analyze. One class of promising biocompatible sample preparation material is a restricted-access material (RAM),9-11 such as alkyldiol-silica (ADS).12-14 The material fractionates a sample into the protein matrix and the analyte fraction. Simultaneous with this size exclusion process, low molecular weight compounds are extracted and enriched, via partition, into the phase’s interior. At present, these procedures require a column-switching apparatus,14 including solvents for extraction and desorption. Utilizing the ADS particles as a SPME coating can further simplify the extraction process and experimental setup, while completely eliminating the requirement of extraction solvents. The purpose of this study was to employ the ADS restrictedaccess material as a SPME coating for the determination of benzodiazepines in biological samples. Benzodiazepines represent a class of drug compounds administered for a wide range of clinical disorders,15 and the determination often requires sample pretreatment involving tedious and complex pretreatment protocols. The developed ADS-SPME fiber was interfaced to a HPLC system for the simple determination of benzodiazepines in urine, confirming the functionality of the coating material. EXPERIMENTAL SECTION Materials. All solvents were HPLC grade or better and purchased from Caledon (Georgetown, ON, Canada). The benzodiazepines, shown in Figure 1, were purchased from Radian International (Austin, TX) as 1 mg/mL methanol solutions and stored at 4 °C. 3H-Diazepam was purchased from NEN Life Science Products, Inc. (Boston, MA) as a 3.454 µg/mL ethanol solution. The specific activity was 82.5 Ci/mmol. Deionized water, from a (6) Koster, E. H. M.; Wemes, C.; Morsink, J. B.; de Jong, G. J. J. Chromatogr., B 2000, 739, 175-182. (7) Ulrich, S. J. Chromatogr., A 2000, 902, 167-194. (8) Kroll, C.; Borchert, H. H. Pharmazie 1998, 53, 172-177. (9) Mislanova, C.; Stefancova, A.; Oravcova, J.; Horecky, J.; Trnovec, T.; Lindner, W. J. Chromatogr., B 2000, 739, 151-161. (10) Lamprecht, G.; Kraushofer, T.; Stoschitzky, K.; Lindner, W. J. Chromatogr. B 2000, 740, 219-226. (11) Lauber, R.; Mosimann, M.; Buhrer, M.; Zbinden, A. M. J. Chromatogr., B 1994, 614, 69-78. (12) El Mahjoub, A.; Staub, C. J. Chromatogr., B 2000, 742, 381-390. (13) Boos, K. S.; Grimm, C. H. Trends Anal. Chem. 1999, 18, 175-180. (14) Yu, Z. X.; Westerlund, D.; Boos, K. S. J. Chromatogr., B 1997, 704, 53-62. (15) Drummer, O. H. J. Chromatogr., B 1998, 713, 201-225.
Analytical Chemistry, Vol. 74, No. 5, March 1, 2002 1081
Figure 1. Structures of benzodiazepines used in this study.
Barnstead/Thermodyne NANOpure ultrapure water system (Dubuque, IA), was used for dilution of the standards. Fused-silica optical fibers (various diameters) were purchased from Polymicro Technologies Inc (Pheonix, AZ). LiChrospher RP-18 ADS, 25-µm ADS particles was supplied by Merck KGaA (Darmstadt, Germany). Preparation of ADS-SPME Fibers. The silica fibers were cut into 38-mm lengths and cleaned with a 30:70 mixture of 30% hydrogen peroxide and concentrated sulfuric acid by ultrasonic wave for 1 h. They were thoroughly rinsed by sonification in water, pure ethanol, and water, respectively. This cleaning procedure was sufficient to remove any coating and buffer from the optical fibers. The ADS particles were immobilized on the silica fiber with Locktite 349 adhesive (Rocky Hill, CT). After a thin and uniform layer of the adhesive (binding agent) was applied, the silica fiber was carefully dipped into a 1.0-mL plastic Eppendorf microcentrifuge tube containing the 25-µm ADS particles. The excess particles were removed from the fiber by gentle tapping. The binding agent was cured using a Locktite Zeta 7500 portable UV lamp for 30 min. A Hitachi model S-570 (San Jose, CA) scanning electron microscope was used to image the prepared surface of the base silica and the ADS-SPME fibers. Conditioning of ADS-SPME Fibers and Extraction of 3HDiazepam. The prepared fibers were initially conditioned by successively shaking the submerged fibers in 2-propranol, methanol, and water for 20 min. The fibers were then stored in a water/ methanol (95:5 v/v) mixture until ready to use. The ADS-SPME and blank silica fibers were placed in 1.5-mL Eppendorf (Brinkmann Instruments, Mississauga, ON, Canada) plastic microcentrifuge tubes containing 1.0 mL of 3H-diazepam standard solution (prepared in water) over a range of concentrations and salt conditions followed by agitation on a shaker table for 3 h. The fiber was removed, rinsed twice by total immersion in water, and placed into scintillation vials containing 20 mL of Ecolume scintillation cocktail (a proprietary mixture of linear alkybenzene, phenylxylylethane, nonionic surfactants, 2,5-diphenyloxazole, and p-bis(o-methylstyryl)benzene). The vials were vigorously shaken and counted in a Beckman-Coulter (Fullerton, 1082
Analytical Chemistry, Vol. 74, No. 5, March 1, 2002
CA) model LS1701 scintillation counter for 5 min. This completely removed the labeled diazepam from the fiber coating as determined by subsequent recounting of the fiber in fresh scintillation cocktail. A 3H-diazepam standard mass calibration curve was constructed for conversion of the DPM values to an absolute mass of diazepam. The C18 extraction process is noncompetitive (in comparison to adsorption), and the amount of analyte extracted from a sample is independent of the matrix composition. Once equilibrium is reached, the extracted amount is constant and is independent of further increases in extraction time. For a sufficiently large sample volume (>100 µL), in relation to the fiber coating volume (Vf), and a constant fiber coating/sample partition constant (Kfs), the amount of analyte extracted (n) is directly proportional to the concentration sample (Co), as represented by eq 1.16 From eq 1,
n ) KfsVfC0
(1)
the calibration curve is expected to be linear and the sensitivity of the extraction was related to the partition coefficient of the analyte in the sample for the fiber coating. Instrumentation and Analytical Conditions. A HewlettPackard (Palo Alto, CA) HPLC system (model 1050) complete with autosampler and multiple wavelength UV detector (λ ) 230 nm) was used with a Supelcosil C18 column (5.0 cm × 4.6 mm i.d.; 5-µm particle size) from Supelco (Bellefonte, PA). A LiChrosorb RP-18 guard column (1 cm × 4.6 mm) from Supelco was installed at the inlet of the chromatographic column for protection of the analytical column. A SPME-HPLC interface was constructed as previously described17 using a Valco zero-volume tee from Chromatographic Specialties (Brockville, ON, Canada). However, the thru-hole of the tee was enlarged to facilitate a largediameter fiber. The volume of the empty desorption chamber was (16) Pawliszyn, J. Solid-Phase Microextraction: Theory and Practice; Wiley-VCH: New York, 1997; pp 15-16. (17) Chen, J.; Pawliszyn, J. B. Anal. Chem. 1995, 65, 2530-2539.
Figure 2. Schematic representation of ADS-SPME HPLC interface in the inject position (desorption).
estimated to be 5 µL. The ADS-SPME fiber was connected to the HPLC interface as shown in Figure 2. The 1/16-in. PEEK tubings and nuts were received from Upchruch Scientific (Oak Harbor, WA). Elution of the extracted compounds from the ADS-SPME fiber and separation by the reversed-phase HPLC column was accomplished by switching the six-port injection valve to redirect the water/methanol (52:48 v/v) mobile phase over the fiber surface at a flow rate of 1.0 mL/min. Preparation of Urine Samples. Urine samples were collected from a drug-free healthy volunteer. As recommended by the manufacturer of the ADS material, any precipitated material was removed by centrifuging the sample at 10000g for 10 min.18 The five benzodiazepines were directly spiked into the supernatant of the biological samples over a range of 0.50-50 µg/mL. The ADSSPME fiber was submerged into 1.5 mL of the urine, contained in a 2.0-mL amber sample vial. Magnetic stirring with a 0.60-cmlong Teflon-coated stir bar was used to agitate the sample at 800 rpm for direct extraction over 60 min. The ADS-SPME fiber was rinsed twice by total immersion in water before interfacing to the HPLC system for desorption and separation of the extracted analytes. RESULTS AND DISCUSSION Immobilization of ADS Material. The chemical and physical properties of the ADS material have been previously discussed in the literature.13 In summary, the porous ADS particles possess a hydrophilic electroneutral diol exterior surface to prevent protein adsorption and an inner surface of C18 alkyl hydrophobic bonded phase that is responsible for extraction of the target compounds. Immobilization of the material onto a silica fiber provided a SPME coating whereby the inert outer layer protected the coating from contamination by proteins, while allowing direct and multiple extractions of biological fluids. The immobilization of the ADS particles was accomplished by adhering the particles on to a cleaned silica fiber. Several binding agents were evaluated and optimized for their ability to physically (18) LiChrospher ADS Application Guide; Merck KGaA: Darmstadt, Germany, 1999.
Figure 3. Scanning electron micrographs of bare silica fiber (A) and ADS-SPME fiber coating (B). Gold coating overlayer, 30 nm; accelerator voltage, 15 kV.
maintain the ADS coating throughout the SPME experiments. The binding agent must also be chemically stabile in the various organic solvents and pHs of the mobile phase since the presence of background peaks, resulting from the breakdown of the binding agent, could interfere with the analyte’s determination. It was also necessary to apply a thin overlayer of the agent to ensure the extraction coating was not blocked. The Locktite 349 adhesive provided the most uniform and robust bonding of the ADS particles to the silica fiber. As shown in Figure 3, scanning electron micrographs (SEM) of blank silica (A) and ADS-SPME (B) fibers were recorded for comparison purposes. The confirmation of the ADS particles immobilized on the fiber over a fairly uniform coating was obvious. The SEM data were used to estimate an average coating thickness of 40 µm. The extraction mechanism of the ADS-SPME coating was absorption as the analytes partition into the C18 stationary phase of the inner pores. Extraction of analytes by C18 is a well-utilized chemistry as indicated by the common use of C18 analytical Analytical Chemistry, Vol. 74, No. 5, March 1, 2002
1083
Table 1. 3H-Diazepam Extraction Performance of ADS-SPME Fibers DPM counts (mass extracted) blank
3.43 ng/mL 3H-diazepam
trial no.
ADS fiber
silica fiber
ADS fiber
1 2 3 4 5 % RSD
69 (0) 73 (0) 72 (0) 68 (0) 68 (0) 3.35
127 134 123 128 130 3.14
796 690 (1.65 ng) 898 580 (1.86 ng) 825 393 (1.71 ng) 809 744 (1.68 ng) 860 808 (1.79 ng) 4.93
columns and solid-phase extraction (SPE) cartridges. Although, similar SPME fibers based on this chemistry have been prepared, they did not possess a biocompatible surface to prevent protein adsorption and were therefore restricted to much cleaner matrixes such as water samples.19,20 ADS-SPME Fiber Characterization. The extraction performance of the ADS-SPME fiber was validated using 3H-diazepam and liquid scintillation detection. Scintillation counting was chosen due to its simplicity, speed, and sensitivity. The reproducibility of the coating was tested with five independently prepared ADSSPME fibers. The fibers were submerged in a 3.43 ng/mL standard 3H-diazepam solution (in 95:5 water/methanol) for 3 h on a shaking bed. The fiber was then removed from the solution, washed twice by totally immersion in 95:5 water/methanol, and placed in a scintillation vial containing 20 mL of scintillation cocktail. The vials were vigorously shaken and counted in triplicate with the liquid scintillation counter for 5 min. The average value of the three counts was considered as the final counting result. In addition, a bare silica fiber (control) was evaluated in the 3.43 ng/mL standard 3H-diazepam solution and an ADS-SPME fiber evaluated in a blank 95:5 water/methanol solution. To correlate the DPM counts to a diazepam mass value, a standard 3Hdiazepam mass calibration curve was prepared by directly spiking 3H-diazepam into the scintillation cocktail (data not shown). Table 1 summarizes the scintillation results, corresponding mass of diazepam extracted by each fiber, and its reproducibility. As expected, the scintillation values of ADS in the blank solution were equal (with in experimental error) to the background value of the scintillation cocktail. Similarly, the amount of diazepam binding to the bare silica control fiber was determined to be negligible. However, the ADS-SPME coating on the fiber’s surface was successful in extracting a significant portion of the diazepam from the sample. The diazepam penetrated into the porous structure of the ADS and was absorbed by the C18 extraction phase. The preparation of the ADS-SPME fibers and the extraction procedure was determined to be very reproducible with a RSD value of 5% v/v) to the sample to ensure release of protein-associated drugs such as benzodiaz(32) Seydel, J. K.; Schaper, K. J. Pharmacol. Ther. 1982, 15, 131.
1086 Analytical Chemistry, Vol. 74, No. 5, March 1, 2002
R2 value
detection limit (ng/mL)
0.9955 0.9806 0.9875 0.9969 0.9987
600 750 333 100 46
Concentration range, 0.05-50 µg/mL; number of data points, 6.
epines prior to extraction.18,24 Although the separation of the benzodiazepines was more than adequate for quantification purposes, when compared to the separation of benzodiazepine standards (prepared in water) on the C18 analytical column, some peak tailing was observed. The effect appeared to be more pronounced for the later eluting compounds. Compounds with long elution times indicated a high partition coefficient with the C18 stationary phase. Therefore, the elution efficacy of the mobile phase may not have been strong enough for rapid desorption of these compounds from the ADS fiber. It is important ensure the extracted compounds are desorbed in as narrow a band as possible to prevent peak broadening. Increasing the desorption efficacy of the mobile phase, by decreasing its polarity, was not an appropriate solution as the HPLC separation of the desorbed compounds would be sacrificed (see Figure 5). A step-gradient elution profile was evaluated to enhance desorption of the benzodiazepines from the fiber; however, only minimal improvement in the chromatographic performance was observed. Alternatively, a small volume of solvent different from the mobile phase may be introduced through the SPME-HPLC interface for improved desorption efficacy16 or experimental parameters such as increasing the interface temperature can be employed to enhance the elution efficiency. More fundamentally, changing the chemical nature of the extraction phase located inside the pores of the ADS particles coated on the silica fiber can change the mobile phase’s elution efficacy. The chemical nature or polarity of the extraction phase can be adjusted to decrease the analytecoating binding affinity and will therefore provide an easy elution of the analyte from the fiber. Calibration curves were constructed over a range of 0.5-50 µg/mL for the five compounds. As shown in Table 2, excellent linearity was observed for all benzodiazepines in urine (average R2 ) 0.9918). The detection limit for each compound was determined at a concentration where the signal/noise ratio was equal to 3, and these calculated concentrations have also been included in Table 2. For a given sample concentration and fiber coating, the detection sensitivity is determined by the magnitude of the coating/sample partition constant (Kfs). Since the extraction phase of the ADS fiber’s coating was the same stationary-phase material of the analytical column, the retention times of the benzodiazepines can be used to estimate detection sensitivities. Therefore, analytes with long retention times and hence higher Kfs values, such as diazepam, will have improved detection sensitivities. In general, the poor sensitivity of UV detection resulted in relatively high detection limits. However, improved
detection systems, such as mass spectrometry, will provide enhanced detection sensitivities and enable trace analysis. The reproducibility of the developed method was determined with 10 injections of a 1.0 µg/mL urine sample. This injection repeatability was calculated as the RSD for each benzodiazepine HPLC peak area, and the average value for all compounds was determined to be 5.2%. The intra-assay precision was determined with repeated analysis of a sample that has been independently prepared, over 1 day, yielding an average RSD of 5.9% and the between-day variance was determined to be 6.4%. The ADS-SPME coating was based on a very robust material. The ADS material has been previously validated with biological samples for over 2000 injections.33 Although, the ADS-SPME fiber was not subjected to as many injections in this study, its stability was evaluated for over 50 analysis with minimal loss of performance and confirmed the suitability of the fiber for simple and direct extraction of benzodiazepines from a biological sample. CONCLUSIONS A novel ADS-SPME fiber was developed for the direct extraction, desorption, and HPLC determination of benzodiazepines in a biological sample. There was no requirement to precipitate proteins from the sample prior to extraction, therefore minimizing sample preparation time and eliminating potential sample preparation artifacts. The binding capacity, extraction efficiency, and reproducibility of the fiber were suitable for UV determination over a wide range of benzodiazepine concentrations in urine. The resultant biocompatible ADS-SPME fiber was reusable, was simple to use, and eliminated the requirement of extraction solvents. The utilization of the ADS material for other classes of drugs9,34 ensures the potential versatility and suitability of this approach. (33) Majors, R. E.; Boos, K. S.; Grmmm, C.-H.; Lubda, D.; Wieland, G. LC-GC 1996, 14, 554-559.
More fundamentally, the extraction phase located inside the pores of the coating can be designed toward the class of compounds under analysis. For example, phases with C4 or C8 functional groups will enable the extraction of analytes over a wide range of polarities. Enhanced selectivity for the extraction of target compounds is also possible with molecular recognition centers located in the pores, such as molecular imprinted polymers.35 The simplicity and compatibility of various formats of SPME with analytical systems such as HPLC, GC, and CE will also extend the versatility of the ADS-SPME fiber for proteinaceous biological samples.36 Studies are presently underway in our laboratories to further test the convenience, biocompatibility, and selectivity of various ADS-SPME coatings for the determination of drug concentrations in vivo (in the circulating blood of an animal) and thereby eliminating the need to draw blood. ACKNOWLEDGMENT The authors acknowledge the Natural Sciences and Engineering Research Council (NSERC) for partially funding this work and providing a postdoctoral fellowship to W.M.M. They also acknowledge Dieter Lubda at Merck KGaA (Darmstadt, Germany) for supplying the ADS material, Glaxo Wellcome (Mississauga, ON) for donating the HP-1050 HPLC system, and the von Humboldt Foundation (Bonn, Germany) for fellowship support. Received for review July 3, 2001. Accepted December 7, 2001. AC010747N (34) Baeyens, W. R.; Van der Weken, G.; Haustraete, J.; Smet, E. Biomed. Chromatogr. 2000, 14, 61-69. (35) Haginaka J.; Sanbe H. Anal. Chem. 2000, 72, 5206-5210. (36) Mullett, W. M.; Levsen, K.; Lubda, D.; Pawliszyn, J. J. Chromatogr., A, submitted.
Analytical Chemistry, Vol. 74, No. 5, March 1, 2002
1087
