Investigation of the Effect of the Extraction Phase Geometry on the

May 4, 2009 - ... address: Professional Analytical Systems Technology, Richard-Wagner St. 10, 99441, Magdala, Germany. Cite this:Anal. Chem. 2009, 81,...
0 downloads 0 Views 1MB Size
Anal. Chem. 2009, 81, 4226–4232

Investigation of the Effect of the Extraction Phase Geometry on the Performance of Automated Solid-Phase Microextraction Erasmus Cudjoe, Dajana Vuckovic, Dietmar Hein,† and Janusz Pawliszyn* Department of Chemistry, University of Waterloo, 200 University Avenue W, Waterloo, Ontario N2L 3G1, Canada A new configuration of C18 thin film extraction phase designed for high sample throughput has been developed and applied to the analysis of benzodiazepines in spiked urine samples using high performance liquid chromatography coupled with tandem mass spectrometry. The high throughput analysis was achieved with the use of a robotic autosampler which enabled parallel analyte extraction in a 96-well plate format. Factors affecting data reproducibility, extraction kinetics, sample throughput, and reliability of the system were investigated and optimized. The intrawell reproducibility was 4.5-7.3%, while interwell reproducibility was 7.0-11% in urine and PBS samples. The limits of detection and quantitation were 0.05-0.15 ng/mL and 0.2-2.0 ng/ mL for all analytes, respectively. By comparison with optimized automated multifiber SPME relying on rod geometry, the C18 thin films showed higher extraction rates (approximate 2-fold increase) and hence higher sample throughput because of the improved configuration and more effective agitation/mass transfer. In addition, this new configuration provided an extraction phase with greater surface area to volume ratio and greater extraction phase volume, which resulted in approximately 2-fold increase in the extraction capacity for diazepam compared with the extractions with automated multifiber SPME rod geometry. The results of this investigation demonstrated the advantages of using thin films to improve extraction kinetics and sensitivity of automated SPME methods for high performance liquid chromatography. Since its introduction in 1990,1,2 solid phase microextraction (SPME) as a sample preparation method continues to attract the interest of researchers as a potential alternative method for drug analysis.3-15 Despite this continuing interest, most of the developed methods were characterized by low sample throughput * To whom correspondence should be addressed. E-mail: [email protected]. Phone: 1-519-888-4567, ext. 84641. Fax: 1-519-746-0453. † Present address: Professional Analytical Systems Technology, RichardWagner St. 10, 99441, Magdala, Germany. (1) Arthur, C. L.; Pawliszyn, J. Anal. Chem. 1990, 62, 2145–2148. (2) Zhang, Z.; Yang, M.; Pawliszyn, J. Anal. Chem. 1994, 66, 844A–853A. (3) Chia, K.; Huang, S. Anal. Chim. Acta 2005, 539, 49–54. (4) Chiarotti, M.; Marsili, R. J. Microcol. Sep. 1994, 6, 577–580. (5) Chou, C.; Lee, M. Anal. Chim. Acta 2005, 538, 49–56. (6) del Olmo, M.; Zafra, A.; Suarez, B.; Gonzalez-Casado, A.; Taoufiki, J.; Vılchez, J. L. J. Chromatogr. B 2005, 817, 167–172.

4226

Analytical Chemistry, Vol. 81, No. 11, June 1, 2009

because of the lack of appropriate extraction phases, as well as lack of automation which would enable performing multiple extractions in parallel. Recent work within our research group has made it possible to perform multiple extractions in parallel on a 96-well format with the aid of an automatic robotic unit.16,17 Possible applications of such automated SPME technique include high-throughput analysis of drugs in complex biofluids including whole blood,17 automated drug-protein binding studies,18 and high-throughput toxicological screening studies to monitor exposure to various mycotoxins such as Ochratoxin A.19 According to SPME fundamental principles, as shown below, the amount of analyte extracted by SPME is proportional to the volume of the extraction phase. n)

KesVeVsC0 KesVe + Vs

(1)

For large sample volumes, when KesVe , Vs, eq 1 simplifies to n ) KesVeC0

(2)

where n is the amount of analyte extracted at equilibrium, Kes is the partition coefficient between the extraction phase and the sample matrix, Ve is the volume of extraction phase, Vs is the volume of sample, and C0 is the original concentration of the analyte. (7) de Toledo, F. C. P.; Yonamine, M.; de Moraes Moreau, R. L.; Silva, O. A. J. Chromatogr. B 2003, 798, 361–365. (8) Fan, Y.; Feng, Y.; Zhang, J.; Da, S.; Zhang, M. J. Chromatogr. A 2005, 1074, 9–16. (9) Fan, Y.; Feng, Y.; Da, S.; Shi, Z. Anal. Chim. Acta 2004, 523, 251–258. (10) Kataoka, H.; Lord, H. L.; Pawliszyn, J. J. Chromatogr. B 1999, 731, 353– 359. (11) Koster, E. H. M.; Wemes, C.; Morsink, J. B.; de Jong, G. J. J. Chromatogr. B 2000, 739, 175–182. (12) Lokhnauth, J. K.; Snow, N. H. J. Sep. Sci. 2005, 28, 612–618. (13) Mullett, W. M.; Martin, P.; Pawliszyn, J. Anal. Chem. 2001, 73, 2383– 2389. (14) Paradis, C.; Dufresne, C.; Bolon, M.; Boulieu, R. Ther. Drug Monit. 2002, 24, 768–774. (15) Theodoridis, G.; Lontou, M. A.; Michopoulos, F.; Sucha, M.; Gondova, T. Anal. Chim. Acta 2004, 516, 197–204. (16) Cudjoe, E.; Pawliszyn, J.; J. Pharm. Biomed. Anal. 2008 [Online early access]. DOI: 10.1016/j.jpba.2008.07.014. (17) Vuckovic, D.; Cudjoe, E.; Hein, D.; Pawliszyn, J. Anal. Chem. 2008, 80, 6870–6880. (18) Vuckovic, D.; Pawliszyn, J. J. Pharm. Biomed. Anal. 2008 [Online early access]. DOI: 10.1016/j.jpba.2008.08.023. (19) Vatinno, R.; Vuckovic, D.; Zambonin, C. G.; Pawliszyn, J. J. Chromatogr. A 2008, 1201, 215–221. 10.1021/ac802524w CCC: $40.75  2009 American Chemical Society Published on Web 05/04/2009

An increase in the volume of the extraction phase can be achieved by increasing the thickness of the extraction phase, as is accomplished in stir-bar sorptive extraction.20,21 However, this is characterized by longer equilibration times which obviously lead to lower sample throughput. An alternative way to improve the sensitivity of the SPME method is to use an extraction phase with larger surface area to volume ratio, a configuration known as thinfilm microextraction.22 To date, this approach has only been applied to the analyses of volatile organic compounds using gas chromatography (GC)23-25 probably because of the ease with which the adapted method is compatible with the GC injector. The inherent advantage of this approach is that the increase in surface area helps improve the extraction kinetics. Another important factor that contributes to the advantage of thin film microextraction over traditional SPME is the faster rate of extraction and thus shorter analysis time. Theoretically, this is described by the equation shown below,26 (dn/dt) ) (DsA/δ)Cst

(3)

where dn/dt is the rate of extraction, Ds is the diffusion coefficient of the analyte in the sample matrix, A is the surface area of the extraction phase, δ is the thickness of the boundary layer surrounding the extraction phase, Cs is the concentration of analyte, and t is the extraction time. Hitherto, two different configurations of the extraction phase (Empore solid phase extraction disks supported on stainless steel pins16 and coated silica particle-based extraction phase immobilized on stainless steel fibers17) were evaluated for drug analysis applications. Empore disks had better extraction efficiency, but the movement of extraction phase during extraction was found to be problematic. To overcome this limitation and further improve sensitivity over fiber geometry, a new configuration of the extraction phase is proposed in the current work such that a thin film of extraction phase with higher surface to volume ratio versus fiber geometry is used. Although the sensitivity of fiber geometry was sufficient for applications developed to date, further enhancement in sensitivity would enable future applications to more poorly extracted drugs (more polar drug compounds), highly bound drugs, and drugs for which MS instrumental sensitivity is poor. The objective of this project was to study the effect of the new thin-film configuration of the extraction phase on the performance of automated SPME on 96-well plate format. Octadecyl silica coated particles (C18) were immobilized on thin rectangular stainless steel films, and the performance of the new device was evaluated using benzodiazepines as model compounds. The main parameters investigated included extraction efficiency and extraction kinetics. The analysis of benzodiazepines (diazepam, lorazepam, oxazepam, and nordiazepam) was per(20) Kawaguchi, M.; Ito, R.; Saito, K.; Nakazawa, H. J. Pharm. Biomed. Anal. 2006, 40, 500–508. (21) David, F.; Tienpont, B.; Sandra, P. LCGC North Am. 2003, 21, 108–118. (22) Bruheim, I.; Liu, X.; Pawliszyn, J. Anal. Chem. 2003, 75 (4), 1002–1010. (23) Qin, Z.; Bragg, L.; Ouyang, G.; Pawliszyn, J. J. Chromatogr. A 2008, 11961197, 89–95. (24) Bragg, L.; Qin, Z.; Alaee, M.; Pawliszyn, J. J. Chromatogr. Sci. 2006, 44, 317–323. (25) Hu, Y.; Yang, Y.; Huang, J.; Li, G. Anal. Chim. Acta 2005, 543, 17–24. (26) Koziel, J.; Jia, M.; Pawliszyn, J. Anal. Chem. 2000, 72 (21), 5178–5186.

formed using high performance liquid chromatography coupled to tandem mass spectrometer (LC-MS/MS) according to method published elsewhere.16,17 EXPERIMENTAL SECTION Reagents and Materials. Methanol and acetonitrile solvents (HPLC grades) used in the experiments were obtained from EMD Chemicals Inc. (Darmstadt, Germany). Acetic acid was obtained from BDH Inc. (Toronto, Ontario). The benzodiazepine drugs (diazepam; nordiazepam; oxazepam; and lorazepam) were obtained from Radian International (Austin, TX, U.S.) as 1 mg/mL standard in methanol with the exception of lorazepam, which was in acetonitrile, and stored at 4 °C in a refrigerator. A mixed standard (100 ng/mL) of the benzodiazepines was prepared in 1:1 (v/v) methanol-water mixture, stored in the fridge, and used as the stock solution for subsequent experiments. Deionized water used for dilution of stock solutions was from a Barnstead/ Thermodyne NANO-pure ultra water system (Dubuque, IA, U.S.). Phosphate buffer solutions (PBS) were prepared in the laboratory using analytical grade chemicals by mixing 8.0 g of NaCl, 0.2 g of KCl, 144 g of Na2HPO4, and 0.24 g of KH2PO4 in deionized water and the pH adjusted to 7.4. Drug-free urine samples were obtained from a healthy volunteer. Discovery C18 particles (5 µm) immobilized on stainless steel blades were obtained from Supelco (Bellefonte, PA, U.S.). Stainless steel tubes, approximately 1.4 mm in diameter, were obtained from Small Parts Inc. (Miami, Florida, U.S.). HPLC and Mass Spectrometry Conditions. A Shimadzu (LC-10 AD) HPLC coupled with a triple quadrupole mass spectrometer (MS) (API 3000) from Applied Biosystems, Toronto, Ontario, Canada was used for separation and quantitative analyses of the benzodiazepines. The CTC PAL autoinjector from Leap Technologies (CTC Analytics, Carrboro, NC, U.S.) was used to inject all extracted analytes and standards into LC-MS/MS. Chromatographic separation was achieved with gradient elution using two Shimadzu SCL 10A high pressure liquid chromatographic pumps. Mobile phase A consisted of 10:90 (v/v) and B of 90:10 (v/v) acetonitrile-water mixture with 0.1% acetic acid in both mobile phases to enhance ionization. A sample volume of 20 µL was injected, with a flow rate of 0.5 mL/min maintained during the 5 min chromatographic separation. In the case of the spiked urine samples, a TosoHass TSK precision bypass pump ran in isocratic mode (flow rate 0.5 mL/min) and Waters switching valve were used to direct column effluent to waste in the first minute of chromatographic run. The API 3000 triple quadrupole MS, equipped with a TurboIonSpray Source and heated pneumatic nebulizer interface was operated in the positive mode under multiple reaction monitoring (MRM) conditions. The transitions monitored were 271.1f140.0, 285.0 f153.9, 287.1f241.1, 321.1f275.1 for nordiazepam, diazepam, oxazepam, and lorazepam, respectively. The source temperature and voltage were set at 250 °C and 4500 V, respectively. The curtain, nebulizer, and collision gases were set at 10, 6, and 12, respectively. Further details of the MS acquisition method can be obtained elsewhere.16 Data analysis was performed using Analyst 1.4.1 software which is integrated with the API 3000 instrument. Preparation and Conditioning of Extraction Phases. The extraction phase material consisted of a single layer of 5 µm C18Analytical Chemistry, Vol. 81, No. 11, June 1, 2009

4227

Table 1. Optimized Experimental Conditions for Analysis of Benzodiazepines

Figure 1. Flat stainless steel metal thin films coated with Discovery C18 particles from Supelco supported by a custom-made 96-multi fiber holder.

coated porous silica particles immobilized unto the surface of a stainless steel thin metal film. Scanning electron microscope (SEM) image confirming single layer of the particles immobilized on a stainless steel rod has been presented elsewhere.17 Thin metal films were obtained by an initial flattening of one end of a 5 cm stainless steel tube of approximately 0.061′′ diameter. To ensure effective coating of the C18 particles unto the stainless steel metal, the flattened surface was physically etched by sanding with aluminum oxide sand paper and rinsed thoroughly with water to remove traces of metal pieces that might be on the surface. The stainless steel metal was then sonicated with acetone, for about 20 min, rinsed thoroughly with water, and then sonicated with 8 M nitric acid followed by large volumes of water again. The metal was allowed to airdry thoroughly and a thin layer of Impruv Loctite glue purchased from R.S. Hughes Company (Plymouth, MI) was then applied to the flattened end of the tube. Finally, the metal surface was thinly coated by dipping the gluecoated metal surface into a vial containing C18 particles (5 µm particle size) to form a thin film of extraction phase. The total area of the extraction phase on each stainless steel metal was 102 mm2. To enable the use of these thin films with PAS Concept 96 automated robotic unit, an in-house multifiber support as shown in Figure 1 designed with the help of the University of Waterloo machine shop, was used to secure the thin films in place such that each stainless steel metal fits into individual wells of a 96well plate format. After the coating process, the thin film extraction phase was preconditioned by treating it with methanolic solution (1:1 v/v methanol-water) for at least 24 h without agitation. Prior to each set of experiments, the membranes were reconditioned with 1 mL of 1:1 (v/v) methanol: water and then agitated at 850 rpm for 15 min. Automated Thin Film Method for LC-MS/MS on a 96-Well Plate Format. Simultaneous extractions of 96 samples were performed using an automated robotic unit (Concept 96) which was obtained from Professional Analytical System Technology (PAS, Magdala, Germany). The completely automated robotic unit consisted of three integrated parts and two separate orbital agitators which were controlled by the Concept software. One part was designed to hold, move, and place the thin films supported on the in-house multifiber holder into the 96 wells and was used primarily for the extraction and desorption processes. During the extraction and desorption processes, the 96-well plate was agitated 4228

Analytical Chemistry, Vol. 81, No. 11, June 1, 2009

parameter

C18 thin films

agitation speed desorption solvent desorption volume equilibration time desorption time

850 rpm 80% methanol solution 1.0 mL 25 min 30 min

at a particular speed. The second part was tailored for solvent reconstitution and/or analyte preconcentration steps specifically in situations were enhanced sensitivity was critical. To facilitate easy solvent evaporation, an evaporation device that allows the flow through of nitrogen gas was used. With the use of the Concept 96, the device could be placed directly above the center of each well for direct flow of nitrogen gas into each well for solvent evaporation. The third part of the Concept 96 played a dual function of dispensing precise volumes of solvents into the individual wells of the 96-well plate and also for injection of samples into the HPLC port for chromatographic separation. The dispensed solvents could be used as desorption, reconstitution, and/or internal standard solvents. However, for the sake of this project this feature was not used since all samples and solvents were manually pipetted. Preparation of Calibration Standards. Initial stock solutions of the four analytes originally kept in the refrigerator at 4 °C were serially diluted in 1:1 (v/v) methanol/water to a known concentration and used as stock for further analysis. Specific amounts of this stock were added to each matrix while maintaining the percentage of organic solvent constant (1%) in each case to generate a set of calibration standards (0.05-500 ng/mL). Blank samples were also prepared to check for any interferences and instrument performance. Experimental Procedure. Because of the volume capacity of the well, 1 mL sample volumes were used for all experiments. The experimental procedure involved an initial method development and optimization in phosphate buffer solution (PBS) maintained at pH 7.4 and then application of the developed method to extraction of the analytes in spiked urine samples. For the extraction process, fixed sample volume (1 mL) of benzodiazepines in PBS buffer was placed inside the wells of the 96-well plate and the C18 thin film extraction phase was immersed into the samples until equilibrium was reached to allow the partitioning of the analytes between the matrix and the extraction phase. Equilibrium SPME was employed to ensure that the maximum amount of the analytes was extracted to improve method sensitivity. After the extraction process, the analytes were desorbed from the C18 thin film extraction phase by using the optimized solvent system for which the analytes had the greatest affinity. The method optimization processes included monitoring the effect of the agitation speed to gain insight of the kinetics of analyte transfer in each well, the effect of the position of the thin-film extraction phase inside the well, comparison of the extraction rate and capacity of the C18 thinfilm extraction phase coated on the flat stainless steel metals and cylindrical rods. Table 1 shows the optimized sample preparation conditions used for the analysis of benzodiazepines in PBS buffer and urine.

Figure 2. Dependence of the amount of analyte carried over at various desorption times. SPME was performed from 200 ng/mL benzodiazepine standard solution dissolved in PBS.

All samples used were separately made to 50 ng/mL of benzodiazepines by spiking the sample matrix with known amount of each drug while maintaining the percentage of organic solvent fixed at 1%. The spiked drug samples were allowed to equilibrate within the urine matrix for 12 h. Inter- and intrawell variations for the 96-well plate were evaluated including repeatability and absolute recovery from each matrix. RESULTS AND DISCUSSION Method Optimization with Immobilized C18-coated Thin Film. SPME method optimization was performed by varying one parameter at a time. In SPME, maximum method sensitivity is achieved if the extraction is performed at equilibrium because the amount of the analyte extracted is the highest. In this work, extraction time required to reach equilibrium was 25 min and was used for all studies. Second, to ensure accurate quantitation of the amount of each analyte extracted, analyte carryover was also significantly reduced by selecting the suitable solvent/solution system for the C18 thin films. To achieve optimum desorption conditions, the desorption efficiencies of two different solvent (50% and 80% methanol-water) systems for all four analytes were compared. This was done by extracting PBS samples spiked with 50 ng/ mL of benzodiazepines at constant extraction time (25 min). The analytes were later desorbed in 1 mL of 50% methanolwater solution at varying times (5, 10, 20, 30, 40, 60 min). The experiments were repeated using 80% methanol solution for desorption of the extracted analytes. Results showed that using 50% methanol-water solution would require a minimum of 45 min to give a relative recovery of 70% while 80% methanol-water solution required 30 min desorption and gave relative recovery of 95%. The carryover experiments were performed using 200 ng/mL of each of the 4 analytes. After the initial desorption, a second desorption was carried out using a fresh portion of desorption solution. Figure 2 shows the dependence of the carryover amount with desorption time using 80% methanol-water solution as the desorption solvent. When desorption time was extended to 25 min (Table 1), the carryover of all analytes was reduced to e0.3% which can be considered negligible. Effect of Agitation Speed. One of the important factors to consider during parallel extraction is the uniformity and effectiveness of the agitation speed and analyte kinetics in each well. This is important because poor agitation in any of the wells of the 96well plate could significantly affect the amount of analyte extracted and thus lead to higher relative standard deviations (RSDs) within

and between wells. Thus, to examine the effect of the agitation speed, 50 ng/mL benzodiazepines in PBS buffer were extracted at varying times (5, 10, 15, and 20 min) at two different agitation speeds (500 and 850 rpm) and the analytes extracted using C18 thin films were desorbed using the optimized desorption method. The average amount extracted and overall RSDs (interwell variation) at the various extraction times for all 96 wells were then monitored. As expected, the mean amount of analytes extracted increased as the extraction time increased because the higher amounts of analytes were extracted until the system attains equilibrium. Although lower agitation speed could have been used, equilibration of the analytes in the matrix with the extraction phase would be much longer and thus would cause a corresponding lower sample throughput. Furthermore, Figures 3a and 3b show the dependence of the interwell RSDs on the extraction time and agitation speed. The decrease in RSDs with increasing extraction time during initial points of extraction suggests that thorough mixing in all of the wells has not been achieved at any extraction time using lower agitation speed tested (500 rpm). In contrast, the use of higher agitation speed (850 rpm) resulted in a more uniform agitation in all wells at extraction times g10 min. Similar results were obtained with polydimethylsiloxane coating where high RSD was observed with short extraction times (15 min) and 850 rpm agitation speed.17 Therefore, on the basis of these results, it is recommended to use higher agitation speeds (g850 rpm) and extraction times which are sufficiently long to establish uniform agitation in all wells. In the current study, the use of agitation speed >850 rpm was not feasible because of possible spilling of the sample solution from the wells which was observed when using 1.0 mL sample volumes and very high agitation speeds. Effect of Position of C18 Thin Films within Wells. From our previous work using solid phase extraction disks, it was important to maintain a fixed position of the extraction phase inside the well during extraction to obtain lower inter- and intrawell RSDs.16 Thus, to investigate the effect of position of the C18 thin film on the amount of analyte extracted, three sets of 20 C18 thin films were prepared as described earlier and arranged at various well positions to give fair representation of the 96-well plate. The length of each thin film coating was decreased to 1 cm to ensure that different positions could be accommodated inside the well. Each of the three sets of thin film coatings were placed at three different positions inside the wells as follows: (i) thin films positioned immediately below the sample solution surface, (ii) thin films positioned in the middle of sample solution, and (iii) thin films positioned near the bottom of the well. Using the optimized method (Table 1) the amount extracted and interwell variations (RSDs) were determined for three independent experiments as shown Table 2. The results showed that for each set of the C18 thin films placed just below the sample solution, lower amounts of the benzodiazepines were extracted with respect to those at the bottom and middle positions. This could be due to the fact that at higher agitation speeds, the sample solution formed a vortex and thus exposed part of the extraction phase from the sample solution. Analytical Chemistry, Vol. 81, No. 11, June 1, 2009

4229

Figure 3. (a) Monitoring % RSD of the amount extracted (n ) 96) between the wells with increasing extraction time at 500 rpm and 30 min desorption time using C18 coated thin film on stainless steel SPME blades; (b) Monitoring % RSD of the amount extracted (n ) 96) between the wells with increasing extraction time at 850 rpm and 30 min desorption time using C18 coated thin film on stainless steel SPME blades.

Figure 4. Intrawell variations for five successive extractions of diazepam using C18 coated thin films.

Figure 5. Distribution of the nordiazepam amount extracted from all the 96 wells for 50 ng/mL spiked urine samples using C18 coated thin films as extraction phase.

However, in the case of the other two positions (middle and bottom), the surfaces of the C18 thin films were well beneath 4230

Analytical Chemistry, Vol. 81, No. 11, June 1, 2009

the sample solution maintaining constant and uniform contact with the sample for effective mass transfer. Therefore, higher

Table 2. Investigating the Effect of the C18 Thin Film Position on the Amount Extracted and the Inter-Well Variations of Benzodiazepines for PBS Extractions (n ) 20 Thin Films)a experiment I thin film position

a

experiment II

experiment III

compound

amount extracted (ng)

RSD (%)

amount extracted (ng)

RSD (%)

amount extracted (ng)

RSD (%)

TOP

Diazepam Nordiazepam Oxazepam Lorazepam

8.7 (1.8) 6.0 (1.0) 2.5 (0.6) 3.9 (0.8)

21 16 23 19

8.5 (2.0) 6.2 (1.9) 2.1 (0.4) 3.0 (0.6)

24 30 19 21

8.9 (1.3) 6.1 (1.2) 2.4 (0.3) 3.5 (0.7)

14 19 13 21

MIDDLE

Diazepam Nordiazepam Oxazepam Lorazepam

13 (0.9) 13 (0.9) 4.2 (0.4) 5.1 (0.5)

7.1 6.9 8.4 10

13 (1.0) 12 (0.9) 3.1 (0.2) 5.4 (0.6)

7.6 7.4 7.3 11

13 (0.9) 12 (0.9) 4.0 (0.3) 4.9 (0.5)

7.1 7.2 7.9 10

BOTTOM

Diazepam Nordiazepam Oxazepam Lorazepam

13 (1.0) 13 (1.0) 3.9 (0.3) 5.1 (0.5)

7.5 7.4 7.9 10

13 (1.0) 13 (1.0) 3.9 (0.3) 4.9 (0.5)

7.2 7.2 8.1 11

13 (1.0) 13 (0.9) 3.5 (0.3) 4.8 (0.5)

7.2 7.2 7.7 10

All experiments were performed in triplicate so Experiment I, II, and III denote three successive independent extractions using 20 thin films.

amounts of the drugs were extracted with lower RSDs for all three replicates. Two-factor ANOVA with replication (factor 1 ) position, factor 2 ) analyte) was used on the above data set to compare the results from top and bottom positions. At 95% confidence level, the effect of position (top or bottom) was found to be significant with F ) 1598, Fcrit ) 4.49 and the probability that the observed F value was due to random variation only was 1.8 × 10-17. Two-factor ANOVA was also performed to compare top versus middle position. At 95% confidence level, the effect of position (top or middle) was found to be significant with F ) 537.9, Fcrit ) 4.49, and the probability that the observed F value was due to random variation only was 9.7 × 10-14. In contrast, full immersion and effective agitation and mass transfer were achieved in the middle and lower parts of the well. This was confirmed with nearly equal amounts of benzodiazepines extracted using the middle and lower placement of the C18 thin films inside the well. This was verified using two-factor ANOVA with replication (factor 1 ) position, factor 2 ) analyte) on the above data set to compare the results from middle and bottom positions. At 95% confidence level, the effect of position (middle or bottom) was found to be not statistically significant with F ) 0.8, Fcrit ) 4.49, and the probability that the observed F value was due to random variation only was 0.38. Reproducibility of Extractions From Spiked Urine Samples. After optimization of the agitation speed and position, C18 thin films with 1.5 cm length of coating (102 mm2 surface area) were prepared and used for all subsequent experiments. Prior to the reproducibility tests, a large set of thin films was prepared as described in the Experimental Section, and 96 thin films with similar extraction efficiency were preselected for further use. Using the optimized method and the selected 96 thin films, the overall reproducibility of the thin films was determined for five independent extractions (n ) 5) for all 96 wells from the standard benzodiazepine mixture in PBS buffer and urine samples. Intra- and interwell reproducibility was monitored. RSDs between wells for each set of extraction of the benzodiazepines ranged from 8% to 11% as shown in Table 3 and were very good for “in-house” manufactured thin film extraction phase.

Table 3. Summary of Results Obtained for Inter-Well RSD % and Absolute Recoveries for the All Analytes for Extractions in PBS and Urine Samples for All 96 Thin Films Used for Five Independent Extractionsa C18 thin films absolute recovery (%)

interwell RSD (%)

sample matrix

D

L

O

N

D

L

O

N

Urine

48

15

18

39

8

11

11

9

a

D ) Diazepam; L ) Lorazepam; O ) Oxazepam; N ) Nordiazepam.

RSDs obtained within wells were calculated for the five successive extractions, and it ranged from 4.5 to 7.3%. In comparison to intrawell RSDs obtained from our previous work with the SPE disks,16 the C18-coated thin films gave more reproducible data with a narrower range in variations (Figure 4). This confirmed the fact that the immobilization of the extraction phase inside the well was very crucial for data reproducibility and precision of the method. The excellent results obtained with C18 thin films indicate that the coating can be used successfully in biological matrixes such as urine without adverse effects on method precision. As shown in Table 3, the absolute recovery for all analytes was g15% for all analytes. Recoveries obtained for nordiazepam and diazepam were in the range 39%-48%, respectively, while lorazepam and oxazepam recorded relatively lower (e18%) recoveries. The typical amount of nordiazepam extracted from urine samples for all 96 thin films is shown in Figure 5. Detection and Quantitation Limits and Linear Range. Limit of detection (LOD) was determined based on the signal-tonoise ratio (S/N) method as 3× S/N in urine. The LODs ranged from approximately 0.05-0.15 ng/mL for all the benzodiazepines in spiked urine samples, with diazepam and lorazepam recording the least and the highest LODs, respectively. The limit of quantitation (LOQ) for the extraction of the benzodiazepines, calculated based on 10× S/N, ranged from 0.2-2.0 ng/mL for all four analytes in urine. Comparison of Extractions from PBS buffer using C18 Thin Films and SPME Fibers. Results obtained with the C18 thin films were compared with results obtained using fiber Analytical Chemistry, Vol. 81, No. 11, June 1, 2009

4231

Table 4. Comparison of the Optimized C18 Thin Film and SPME Fiber Extraction Method for Diazepam in PBS Buffer parameters

RPA SPME fiber (0.061′′)

C18 thin film

surface area equilibration time desorption time mean inter-well reproducibility (n ) 96 fibers) mean intra-well reproducibility (5 extractions; n ) 96 fibers) linear range absolute recoveries carryover

102 mm2 25 min 30 min 7.0%

73 mm2 30 min 30 min 7.1%

5.0% 0.2-500 ng/mL 51%