The Analyst

Evaluation of bio-compatible poly(ethylene glycol)-based solid-phase microextraction fiber for in vivo pharmacokinetic studies of diazepam in dogs

Ali Es-haghi, Xu Zhang, Florin Marcel Musteata, Habib Bagheri and Janusz Pawliszyn*
Department of Chemistry, University of Waterloo, Ontario, Canada N2L 3G1. E-mail: janusz@uwaterloo.ca; Fax: +1-519-746-0435; Tel: +1-519-888-4641

Solid-phase microextraction probes based on poly(ethylene glycol)/C18-bonded silica were used for in vivo monitoring of drugs from circulating blood of beagles, over a period of 8 h. After sampling, the extracted drugs were subsequently quantified by liquid chromatography coupled with tandem mass spectrometry. External calibrations in whole blood and phosphate-buffered saline were used to correlate the amount of analytes extracted in regard to the total and free concentrations in blood respectively. The probe provided sufficient sensitivity for the drugs in the blood matrix, while the need for drawing blood was eliminated. The limit of detections of the method from whole blood were 1.7, 1.4 and 2.8 ng mL^–1 for the analysis of diazepam, nordiazepam and oxazepam respectively, and the linear range was from 4 ng mL^–1 to 2 µg mL^–1. The method was applied for the monitoring of pharmacokinetic profiles of intravenous administration of diazepam and its two main metabolites in dogs, and the results were compared with profiles determined by conventional methods. This approach offered increased sensitivity and accuracy, short extraction time, and convenient calibration for in vivo sampling for dynamic monitoring.

There are several motivations for doing in vivo analysis. Primarily, studying chemical processes in association with the normal biochemical environment of a living system is desirable, especially when removing suitable samples for study from the living system is impossible. In pharmacokinetic studies with rodents (mice in particular), because the blood volume of the animals being used is limited, a large number of animals is needed to obtain profiles with sufficient data points. If blood were not removed for analysis, smaller numbers of animals would be required and the data generated would be improved by reduced inter-animal variation. Since compounds of interest are not exhaustively extracted from the investigated system, extraction can be performed in such a way that a small proportion of the total free compounds is removed. Thus, the disturbance of the normal balance of chemical components is avoided. If significant depletion of the free fraction occurs, the depletion in free concentration results in the release of some of the bound fraction until a new binding equilibrium is established. This could be beneficial in the non-disruptive analysis of very small tissue sites or samples.^1,2

Currently, microdialysis and ultrafiltration are the main approaches for in vivo sampling.^2–6 However, pumps and other appliances are required, which makes these approaches more suitable for laboratory use than for field sampling. In addition, pre-concentration is poor and the sampling process may affect the local dissociation equilibrium between the bound and free analytes, thus interfering with the biological system under study.

Solid-phase microextraction (SPME) is a sampling and sample preparation technique which was first developed over ten years ago for the analysis of volatile and semi-volatile organic compounds in environmental samples.^7,8 SPME provides many advantages over conventional sampling methods by integrating analyte extraction, concentration, and sample introduction into a single step. SPME is a non-exhaustive extraction technique where extracted analytes come in equilibrium with the extracting phase and sample matrix. According to SPME theory, above a certain sample size, sample volume does not impact the results,^9,10 therefore, it is not necessary to define a specific sample size for the analysis, which is very desirable for on-site sampling. Additionally, SPME directly extracts a small fraction of free analyte, so a negligible depletion of the free fraction is achieved after extraction. Finally, the technology is easy to be miniaturized, and thus suitable for both small living systems and microanalytical instruments. These advantages make SPME a promising tool for directly assaying in vivo chemical concentrations.^11–13

In typical SPME, a fused-silica fiber, which is coated with a thin layer of polymeric stationary phase, is used to extract analytes from the sample. The extracted analytes are then thermally desorbed in the injector of a gas chromatograph or in a suitable solvent for further analysis. However, SPME with polymer-coated silica fibers still exhibits two serious disadvantages: (1) the fused-silica fiber is very fragile, and, therefore, extra care must be taken during use; and (2) SPME with polymer coatings requires thick films to achieve the desired sensitivity. Increasing the coating thickness can improve the sensitivity of the method, but the time to reach equilibrium becomes longer because extraction is a process usually controlled by diffusion in the coating.^14,15 The absorption rate of a thick coating is higher than a thin coating because the thick coating has a larger coating volume and accordingly a higher loading capacity. Depending on the analytical instrument used for detection, a film with sufficient capacity is needed to extract enough analytes to be detectable by the instrument.

Polypyrrole (PPY)-coated fiber has been used as an SPME probe for in vivo pharmacokinetic studies.^11,12 This fiber was usually prepared using electrodeposition, but the procedure is not only time consuming but also involves hazardous chemicals such as pyrrole.^16–20 As polypyrrole is a porous coating, it extracts analytes mainly by adsorption processes. According to the current theory of analyte extraction by porous SPME coatings, the number of effective surface sites where adsorption can take place is limited.^9,10,21 When all such sites are occupied, no more analyte can be extracted. This suggests that analyte extraction is a competitive process in which a molecule with a higher affinity for the surface can replace a molecule with a lower affinity. Consequently, the linear range of the probe is low and depends on the concentration of other compounds. This problem is significant in complicated matrices such as whole blood or plasma where many endogenous compounds exist.

Poly(ethylene glycol) (PEG) has been known as bio-compatible polymer for in vivo studies,^12,22 and the procedure for preparation is much simpler and does not need any hazardous materials. In order to achieve higher sensitivity for analytes to be detectable, for instance at the last time point of a pharmacokinetic study, C18-bonded silica particles were introduced in PEG glue to increase its loading capacity. In comparison to commercial fibers with a silica core, SPME fibers prepared on a metal surface by dip coating have sufficient robustness for exposure to blood flow inside a vessel. The coating is deposited as a thin film, and the reduced thickness relative to the commercial SPME fibers allows for shorter extraction times and better desorption efficiencies, with sufficient sensitivity for analysis of drugs during a pharmacokinetic study.

The aim of the present study was the evaluation of C18-bonded silica/PEG fiber as an in vivo SPME probe for monitoring drugs and metabolites in living systems. Diazepam and its two main metabolites were chosen as a model drugs and the investigation was performed on beagles.

PEG and C18-bonded silica particles were obtained from Suppelco (Bellfonte, PA, USA). Benzodiazepine standards (1 mg mL^–1 in methanol) were purchased from Cerilliant (Austin, TX), and were diluted in methanol or phosphate-buffered saline (PBS), pH 7.4. Dog whole blood (EDTA as anticoagulant) was obtained from Biological Specialties Corp. (Colmar, PA, USA) and plasma was prepared from it. Plasma was stored frozen at –20 °C until use, and whole blood was stored at 4 °C for a maximum of two weeks. HPLC grade acetonitrile for HPLC mobile phase, methanol for standards preparation and desorption solution, formaldehyde, ethanol and acetic acid (glacial) were from Fisher Scientific (Unionville, ON, Canada). Water was obtained from a Barnstead Nanopure water system.

SPME fibers were prepared as described in the literature.¹² Before in vivo experiments, SPME probes were evaluated for their extraction time profile in various conditions such as different agitations, temperatures and matrices, desorption efficiency, linear range and limit of detection, in a series of in vitro experiments. The influence of flow rate on extraction characteristics was examined with a flow system. In this system a peristaltic pump provided the power for fluid delivery at a precisely controlled linear velocity from the matrix reservoir to the tubing (4.8 mm inner diameter). The SPME sampler was placed into the tubing with the aid of an in-dwelling intravenous catheter assembly. The flow rates were set at 75 and 320 mL min^–1.

The extraction characteristics of the probes were initially studied in 25 mL of PBS or dog plasma in 40 mL sample vials (Supelco). Finally, probe characterization was performed in whole dog blood using 1 mL samples in 2 mL deep well polypropylene 96-well plates (VWR Canada, Mississauga, ON, Canada). A solution of 50 ng mL^–1 in PBS was used for characterization. Extraction calibrations were performed from 0.1 ng mL^–1 to 2 µg mL^–1 in PBS, 1 ng mL^–1 to 5 µg mL^–1 in plasma, and 0.5 ng mL^–1 to 2 µg mL^–1 in whole blood. For the SPME probe extractions in the flow system, fibers were placed in a needle, and after piercing the tubing, fibers were exposed into the extraction solution. Extraction using magnetic stirring was conducted on a VWR Dylastir hot plate stirrer (VWR Scientific, Canada). After extraction from plasma and whole blood, probes were rinsed briefly with a stream of water, before being allowed to dry and desorbed in methanol as the desorption solution. Rinsing of the probes improved the reproducibility of the results, presumably by removing any sample adhering to the outside of the polymer by surface tension but not absorbed.

For in vitro experiments desorption was conducted in 500 µL of methanol in 96-well plates, while for in vivo experiments 300 µL polypropylene (Supelco) inserts were put into 96-well plate and 250 µL methanol was used as the desorbing solvent. Desorption was performed for 10 min. After desorption, methanol was evaporated with a flow of nitrogen using a device designed for solvent evaporation in the 96-well plate. The dried well was reconstituted by a 25 µL solution of 10 ng mL^–1 of lorazepam as internal standard in methanol–water (1 : 1). For in vivo experiments, 20 µL of reconstituted sample was injected in duplicate, so 50 µL of solution was used for reconstitution.

Analyses were performed on an LC–MS/MS system consisting of a Shimadzu 10AVP liquid chromatograph with a system controller and dual binary pumps interfaced to a CTC-PAL autosampler and an MDS Sciex API 3000 tandem mass spectrometer. The column was a Waters Symmetry Shield RP18, 2.1 × 50 mm, 5µm particle size (Millford, MA). Mobile phases were as follows: (A) acetonitrile–water (10 : 90) with 0.1% acetic acid; (B) acetonitrile–water (90 : 10) with 0.1% acetic acid. Mobile-phase flow was 0.5 mL min^–1, and the gradient used was 10% B for the first 0.5 min. This was ramped to 90% B over 2.0 min, held for 1.5 min, and finally returned to 10% B for 1 min. This provided a total 5 min run time including reconditioning. For experiments using plasma or whole blood, LC effluent was directed to waste for the first 1 min of the run time, to eliminate coextracted compounds from entering the MS. During this diversion time, a makeup flow was supplied to the MS via the quaternary pump. The HPLC effluent was analyzed after ESI in positive ion mode with selected reaction monitoring. Nebulizer (N₂) and curtain gas (N₂) flow settings were 8 and the turboheater was set to 500 °C. Nitrogen was used as the collision gas at a collision setting of 4. Gases were supplied from a tank of UHP grade nitrogen (Praxair, Toronto, ON, Canada). Transitions monitored were as follows: diazepam, m/z 285.0/154.1; oxazepam, m/z 286.9/241.0; nordiazepam, m/z 271.1/140.0; lorazepam, m/z 321.1/275.1.

The device for in vivo extraction was made as described before.¹² In order to decrease the variability between the probes, all of the fibers with the same extracted amount of diazepam in a 30 min extraction in a standard solution (50 ng mL^–1 of diazepam in PBS) were used for the in vivo experiments.

Because animal sampling is performed in a sterile environment, the SPME device must be sterilized prior to use. The most convenient method of sterilization is steam sterilization. As the SPME probe coating was not stable enough in autoclaving, sterilization was performed by soaking the device in a solution of 8% formaldehyde and 70% ethanol for 18 h, which was Approved by the Animal Care Committee at the University of Guelph on January 26, 2001.²³ The sterilized devices were put into sterile polypropylene tubes under sterile conditions.

All experimental procedures on dogs were approved by the Animal Care Committees at the Universities of Guelph and Waterloo. Generally, the experiments were performed as previously described.^11,12 Blood concentrations of diazepam and its metabolites were monitored for 8 h after intravenous dosing with diazepam. For each time point, a blood draw was performed first, and then the probe was introduced in place for 5 min immediately after blood drawing. During in vivo sampling, the sterile SPME device was placed through the catheter into the cephalic vein so that only the coated portion of the wire was exposed to the venous blood. After sampling, probes were slowly rinsed with a small portion of water and then placed in their corresponding tubes on dry ice. All probes were of single use. In order to compare and validate the data obtained with SPME, 1 mL of blood was withdrawn from the same catheter and placed in a plasma separator tube (Becton Dickinson, Franklin Lakes, NJ, USA). Plasma was prepared by centrifugation of half the volume of the blood sample and frozen at –20 °C in 2 mL cryovials (Wheaton Science Products, Millville, NJ, USA) until analysis. For plasma and whole blood analysis, 100 µL of sample was mixed with 500 µL of acetonitrile in a microtube. After vortexing and centrifugation at 5000 rpm for 10 min, 400 µL of supernatant was transferred to a 96-well plate and was evaporated to dryness under nitrogen. Samples were reconstituted in 50 µL of methanol–water (1 : 1) and shook to ensure dissolution. A 20 µL aliquot was injected into the LC–MS/MS for analysis and the same chromatographic conditions as were used for the SPME probe analysis. The analytical range was 1–2000 ng mL^–1 in plasma and 5–2000 ng mL^–1 in whole blood.

Poly(ethylene glycol) and C18-bonded silica particles are well-known stationary phases in gas-and liquid-chromatography respectively. Poly(ethylene glycol) is used as a coating in some types of commercial SPME fibers containing carbowax, and its biocompatibility has been proved.^22,24 The polymer is attractive due to its low toxicity, general ruggedness, insolubility in common solvents, tolerance to elevated temperatures, and applicability in many chemical environments. In order to increase the fiber capacity and consequently the sensitivity, C18-bonded silica particles were immobilized on the metal fiber using PEG as glue. The loading capacity of the C18-bonded silica/PEG coating was significantly increased when compared with PEG-coated fibers, mainly because of the stronger absorption capability of C18.^25–27Fig. 1 shows a scanning electron micrograph of an SPME fiber coated with C18-bonded silica/PEG. As it can be seen, the structure of the coatings is a monolayer of particles bonded to the surface of the stainless steel wire and the particles were bonded to each other with PEG. Both the particles and PEG contribute to solute partitioning and to the extraction/desorption characteristics of the probes. The mechanical strength of the fibers was greatly improved when stainless steel wires were used as support. As probes were prepared manually, the variability between probes made in the same batch ranged in 15–25%. Further improvements in the coating technique are expected to reduce this variability.

	Fig. 1 Scanning electron micrograph of an SPME fiber coated with C18-bonded silica/PEG fiber with different magnifications, surface view.

Extraction using C18-bonded silica/PEG fibers was accomplished in 20 min in 50 ng mL^–1 of analyte solution in PBS and fibers were desorbed in 250 µL at times varying from 1 to 15 min. Each fiber was desorbed again in another 250 µL of desorbing solvent to investigate the carryover. Data showed that after 5 min desorption, carryover was less than 1%. When a deep well 96-well plate was used, 500 µL of desorbing solvent was used to ensure that all the coating was immersed in solvent, as the length of the coating was 1.5 cm. The solvent volume could be reduced to 100 µL when 300 µL polypropylene inserts were put into each well of the plate.

A systematic investigation was conducted to study the effect of different agitation, matrix, and temperature conditions on the extraction behaviors of the probes including extraction time profile, desorption time profile, and dynamic range. By doing these in vitro experiments, the experimental conditions were optimized for the in vivo experiments.

Firstly, in order to study the effect of different agitation conditions, the extraction time profile was obtained using stirrer bar agitation and a flow system. Extraction was conducted in a 50 ng mL^–1 solution of diazepam, nordiazepam and oxazepam in PBS and plasma. The extraction time was varied from 1 to 20 min. The extraction vial was kept in a water bath to keep the temperature at 37 °C. Extraction with the flow system was conducted at two different flow rates, corresponding to the minimum and maximum capability of the pump of the system. In order to compare the results from different agitation techniques, extraction was also performed using a magnetic stirrer at two different stirring speeds related to the corresponding flow rates in the flow system. Flow rates of 75 and 320 mL min^–1 were used with the flow system. These flow rates provided linear velocities of 6.9 and 29.5 cm s^–1 respectively through the 4.8 mm inner diameter tubing. To calculate the stirring rate in the vial to obtain the same velocity as in the tubing, the following formula was applied:

V = 2r

(1)

	Fig. 2 Extraction time profile in PBS for diazepam, nordiazepam and oxazepam using the flow system at different flow rates: (a) 75 mL min–1; (b) 320 mL min–1.

Furthermore, extraction was performed in plasma and whole blood to investigate the matrix effect on the extraction behavior of probe. Extraction in plasma was conducted using the flow system with different agitation efficiencies, but for whole blood static extraction was performed due to risk of cell breakage with flow and stir bar agitations. Plasma is more viscous than PBS so the diffusion coefficient in plasma is lower than in PBS, which results in a longer equilibration time. As Fig. 3 shows, compared with extraction from PBS, the extracted amount of analytes from plasma are about ten-fold lower. In plasma, most portions of drugs are bound to protein so the free fractions of analytes, which can be extracted by SPME, are decreased. The binding constant of the drug to plasma protein and the distribution coefficient of the free drug between fiber and sample determine the amount of drug extracted by SPME fibers. Although a longer equilibration time was needed in extraction from plasma and whole blood, due to a lower mass transfer rate of analytes, results showed that 5 min was enough for equilibrium extraction. A short extraction time is desirable for pharmacokinetic studies because more time points can be acquired and thus more detailed and accurate information about the drug metabolism can be obtained. In addition, it is much more practicable and convenient for animal experiments.

	Fig. 3 Extraction time profile in plasma solution using the flow system at two different flow rates: (a) 75 mL min–1; (b) 320 mL min–1.

To examine the linear range of the fiber, extraction was conducted in buffer, plasma, and whole blood. In preliminary experiments, extractions from buffer and plasma were tested over the range of 0.01–5000 ng mL^–1 to get a general idea about the linear range and limit of detection in the two sample matrices. Calibration in PBS was linear from 1 ng mL^–1 to 2 µg mL^–1, while in plasma a linear range was obtained from 3 ng mL^–1 to 5 µg mL^–1. A similar linear range was obtained for extraction calibration from whole blood, 4 ng mL^–1 to 2 µg mL^–1, for the three compounds tested. Limits of detection were calculated based on signal/noise ratios (s/n = 3) and were determined to be about 0.4, 0.1, 0.1 ng mL^–1 in buffer, 1.2, 1.2, 1.5 ng mL^–1 in plasma and 1.7, 1.4, 2.8 ng mL^–1 in whole blood for diazepam, nordiazepam and oxazepam respectively. Because of the high level of protein binding (>90%), sensitivity in plasma and whole blood were lower than for buffer.

For whole blood calibrations, which were to be used to quantify the probe data from the animal experiments, a six-point calibration was used (n = 3). Calibration details for buffer, plasma and whole blood are shown in Table 1. The results for calibration were more satisfactory in comparison with polypyrrole fibers for the pharmacokinetic evaluation of diazepam metabolism.^11,12 The dynamic range covers the expected concentration ranges for in vivo experiments, including the first and last parts of pharmacokinetic studies where concentrations are very high or low. Since the competition effect is common in complicated sample matrices such as whole blood, in vitro calibration experiments were conducted to investigate the competition between similar analytes, including diazepam, nordiazepam, and oxazepam.^9,10 No competition and displacement effect was observed during equilibrium extraction even at high analyte concentrations. This is a prominent advantage of C18-bonded silica/PEG fiber in complicated matrices such as whole blood. However, when the polypyrrole fibers were evaluated in a parallel study, it was found that the competition was significant, and dependent on the extraction time and the concentration of the competing molecules.

Compound	Linear range/ng mL–1	Calibration equation	Correlation coefficient, r2
SPME probe calibration from buffer
Diazepam	1.0–2.0 × 103	y = 8.7 × 10–2x + 0.70	0.998
Nordiazepam	1.0–2.0 × 103	y = 7.3 × 10–2x + 0.50	0.999
Oxazepam	1.0–2.0 × 103	y = 3.4 × 10–2x + 0.15	0.999
SPME probe calibration from plasma
Diazepam	3.0–5.0 × 103	y = 4.0 × 10–5x + 8.6 × 10–3	0.998
Nordiazepam	3.0–5.0 × 103	y = 1.0 × 10–4x + 6.1 × 10–3	0.998
Oxazepam	3.0–5.0 × 103	y = 1.0 × 10–4x – 7.6 × 10–3	0.998
SPME probe calibration from whole blood
Diazepam	4.0–2.0 × 103	y = 9.0 × 10–5x + 6.2 × 10–3	0.991
Nordiazepam	4.0–2.0 × 103	y = 1.0 × 10–4x + 7.6 × 10–3	0.991
Oxazepam	4.0–2.0 × 103	y = 1.0 × 10–4x + 6.6 × 10–3	0.991
a Calibration results were recorded as peak area ratios. For calibration, each point was repeated three times.

Comparison of the probe response before and after extraction revealed a small decrease in sensitivity, so for each in vivo experiment new fibers were used. Separate in vitro calibrations in buffer and whole blood were conducted with each in vivo extraction experiment, and final quantification was based on the ratio of analyte to internal standard (IS) peak areas to control for any variability in actual injection volume.

The comparison of extraction calibration between whole blood and buffer shown in Table 1 highlights the potential of the method for easy monitoring of protein binding affinity to the drug, which has been described previously using commercial SPME fibers.^28,29

The most established approach for SPME is equilibrium extraction, when the fiber is in contact with the sample until the partitioning equilibrium is reached. As equilibrium extraction is independent of blood flow when equilibrium has been reached, it offers reproducible results and maximum sensitivity, especially at low concentrations. Therefore it is applicable for in vivo pharmacokinetics when the equilibration time is short enough to allow monitoring of rapid changes in the concentrations of target analytes. The newly developed C18-bonded silica/PEG fiber showed a short enough equilibrium extraction time with sufficient sensitivity, so diazepam pharmacokinetics in dogs was investigated to evaluate the performance of the SPME probes. Nordiazepam and oxazepam, two of the main metabolites of diazepam, were monitored as well. In Fig. 4, the results of the probe and conventional analyses in whole blood are compared.

	Fig. 4 Averaged pharmacokinetic profiles for diazepam and its main metabolites from three studies on three dogs after intravenous dosing with diazepam (n = 6 for the last point, and n = 9 for all the other points); , in vivo SPME; , conventional analysis with whole blood; (a) diazepam; (b) nordiazepam; (c) oxazepam.

In previous works,^11,12 the results of the probe analysis were validated by comparison with the plasma concentrations obtained by a conventional analysis, involving a blood draw, plasma preparation, and analysis of plasma concentrations. Data showed that plasma concentrations based on conventional analysis provided an overestimation of the corresponding blood concentrations obtained with the SPME method, particularly at the first duration of pharmacokinetic studies. Depending on the blood cell/plasma partition coefficient (blood/plasma concentration ratio), a difference in concentration is observed between the two matrices.^30,31 In this research, the partitioning of the drug analytes between blood cells and plasma was determined. It was found that 32% of diazepam, 37% of nordiazepam, and 41%, of oxazepam were partitioned into the blood cells, so the total concentrations obtained from conventional analysis with whole blood were used to validate the total concentrations from SPME experiments. The procedure of the conventional assay is described in the Experimental section, and calibration curves based on that were employed to calibrate the concentrations in blood. The linear correlation coefficients (r² = 0.998) for calibration in conventional analysis of whole blood demonstrate the validity of the chemical assay over the linear range (5–2000 µg mL^–1). As it can be seen in Fig. 4, all three compounds showed excellent correlation between the values obtained with the SPME and conventional analysis.

The data presented in Fig. 4 sufficiently demonstrates the validity of the SPME technique. The blood draw data and the SPME probe data both showed a fast decay in diazepam concentrations over the 8 h period of the study. For nordiazepam and oxazepam, an initial increase or plateau in concentration was observed up to 1.5 h post-dose of diazepam, followed by a decline in concentration over the course of the study.

In many clinical studies, measurement of the free concentrations of drugs, which is the pharmacodynamic active portion, provides much more meaningful information; the present technique addresses this issue. With SPME, the concentration of free analyte is determined by calibration against buffer, while the total concentration can be determined as well by in vitro calibration against whole blood. The results for free concentration of the investigated compounds are shown in Fig. 5. Compared to Fig. 4, it was found that the free concentrations agreed with the total concentrations very well (r  1) due to the basic assumption of the approach: the ratio of the bonded drugs over free drugs will keep constant during the experimental course. Consequently, the apparent binding affinity of the drugs to the blood matrix, which is defined as all the components in blood that can bind drugs including blood cells and macromolecules such as proteins, can be calculated out by dividing the concentration of bonded drugs with the concentration of free drugs. In this experiment, the binding affinity constants to the blood matrix are about 166, 180, and 432 for diazepam, nordiazepam, and oxazepam respectively. Moreover, only the small free fractions of analytes are available for extraction and detection with SPME, because diazepam and its metabolites are rather highly bound to plasma proteins that cannot be absorbed into the fiber coating. The probe illustrated enough high sensitivity to extract the low free fraction of target compounds, particularly during the last part of the pharmacokinetic study.

	Fig. 5 Comparative pharmacokinetic profiles (free concentrations) from three studies on three dogs after intravenous dosing with diazepam (n = 6 for the last point, and n = 9 for all the other points); , diazepam; , nordiazepam; , oxazepam.

In this study, the effectiveness of the bio-compatible PEG-based SPME probe for the in vivo analysis of intravenous drug concentrations in a living animal was demonstrated. The pharmacokinetic profiles of a sample drug, diazepam and its two main metabolites, were obtained using in vivo SPME followed by LC–MS/MS determination. The technology simplifies sample preparation, reduces blood handling and blood draw stress on the animals, which results in a lower number of animals required for pharmacokinetic studies, therefore inter-animal variation in the results can be reduced.

The new probe has shown enhanced sensitivity compared with polypyrrole; therefore it can be employed for monitoring drugs with higher affinity to proteins and lower circulating concentrations. Compared with the previous studies,^5,6 this work provides more detailed information in the pharmacokinetic profiles due to the accuracy and high sensitivity resulting from the large linear range and high capacity of the probes. Application of 96-well plates for desorption of probes after extraction, together with automated injection, facilitated high-throughput analysis.

Footnote

Permanent address: Department of Chemistry, Sharif University of Technology, Azadi Av., P.O. Box 11365-9516, Tehran, Iran.