No delayed temporal response to sample concentration changes during enhanced microdialysis sampling using cyclodextrins and antibody-immobilized microspheres

Abstract

The temporal response to concentration changes external to a microdialysis probe containing trapping agents in the perfusion fluid was studied. Native β-cyclodextrin and a water-soluble β-cyclodextrin polymer were used as trapping agents in the microdialysis perfusion fluid to study the temporal concentration response to carbamazepine, a hydrophobic analyte. The temporal response of microdialysis probes containing antibody-immobilized microspheres against five different cytokines (tumor necrosis factor-α (TNF-α), interferon-γ (IFN-γ), interleukin-2 (IL-2), IL-4, and IL-5) to concentration changes outside of the probe was also determined. In both cases, no delayed temporal response of enhanced microdialysis was observed for either carbamazepine or the cytokines as compared to standard microdialysis sampling procedures.

---

Introduction

Microdialysis sampling is a technique commonly used in complex biological systems to obtain site-specific chemical information. Microdialysis sampling has become an important tool for in vivo neurochemistry, metabolism, pharmacology, and pharmacokinetics studies.^1–3 The conventional microdialysis sampling approach uses a syringe pump to pass a perfusion fluid through the inner fiber lumen of a semi-permeable hollow fiber membrane at microlitre per minute flow rates. In this approach, the chemical sampling occurs due to the diffusion-based separation process across the semi-permeable hollow fiber membrane. The resulting dialysate analyte concentration is often a relative amount of the actual unbound analyte concentration in the matrix into which the microdialysis probe is immersed. Microdialysis sampling probes are typically calibrated to determine the recovery ratio between the analyte concentration in the dialysate versus the external sample solution. When the analyte concentration entering the microdialysis probe is zero, the relative recovery (RR) is the ratio between the dialysate concentration (C_dialysate) and the external sample concentration (C_sample), RR = C_dialysate/C_sample. The microdialysis RR is a complex function of the analyte mass transport properties through the dialysate, membrane and sample as well as probe factors including surface area and length.⁴

Conventional microdialysis sampling can pose difficulties for in vivo collection of hydrophobic drugs, peptides, and proteins.^5–7 Hydrophobic analytes sometimes do not dialyze well. This is typically due to nonspecific adsorption onto the probe polymeric materials combined with high protein binding resulting in low free analyte concentrations in the sample medium.⁸ For peptides and proteins, their larger sizes and thus smaller aqueous diffusion coefficients cause mass transport limitations through the probe resulting in a lower microdialysis RR.⁷ Nonspecific adsorption of peptides or proteins onto the microdialysis apparatus including the membrane and polymeric tubing can also cause difficulties during microdialysis sampling. Oftentimes microdialysis RR values for 10 kDa or larger proteins across 100 kDa molecular weight cutoff (MWCO) membranes range between 1 and 5% at a flow rate of 1.0 µL min⁻¹. This problem coupled with the often low in vivo concentrations external to the microdialysis device, causes great difficulty with analyte identification and quantitation, thus precluding in vivo studies. To overcome this significant low microdialysis recovery problem of these analytes, different methods for enhancing mass transport across microdialysis probes have been reported and include addition of proteins to prevent nonspecific adsorption (e.g., albumin),^9–11 lipids,^9,12 polymeric microspheres,¹³ antibody-immobilized microspheres¹⁴ and different cyclodextrins^15–19 to the microdialysis perfusion fluid (Fig. 1).

Fig. 1 Schematic of the enhanced microdialysis sampling process. Target molecules diffuse into the microdialysis membrane and are trapped by trapping agents via a complexation or binding process. The inclusion of the trapping agent causes analyte mass transport to increase significantly.

For hydrophilic analytes, dialysate concentrations obtained during conventional microdialysis sampling have been shown to quickly reflect rapid concentration changes external to the dialysis probe.^20,21 A possible concern during enhanced microdialysis sampling approaches for hydrophobic analytes is that trapping agents may interact with the membrane resulting in analyte dialysate concentrations that do not reflect sample concentration changes. For example, if a large static layer of the analyte–trapping agent complex (e.g., cyclodextrin–analyte or antibody–analyte) gradually builds on the surface of the inner membrane wall along the concentration boundary layer, this could potentially be a source of falsely high analyte recoveries over time. Alternatively, this may cause a delay in observed analyte dialysate concentrations after significant sample concentration changes are introduced external to the microdialysis probe. The purpose of this study is to address this potential concern and to show that during enhanced microdialysis procedures, the dialysate reflects chemical concentration changes as they occur external to the dialysis probe. Since in vivo microdialysis sampling for most drugs is applied to determine either pharmacokinetic parameters or for metabolism studies, it makes sense to ensure that analyte concentrations obtained in the dialysate are not affected by including trapping agents into the microdialysis sampling perfusion fluid.

Experimental procedures

Chemicals

Native β-CD was obtained as a gift from Wacker Chemical (Norwalk, CT, USA) and used as received. Carbamazepine, NaCl, and water-soluble epichlorohydrin-based β-CD polymer (50 to 55 m/v% β-CD) were purchased from Sigma (St. Louis, MO, USA). A mouse Th1/Th2 Cytokine Cytometric Bead Array (CBA) kit that includes bead-based immunoassays for five cytokines, tumor necrosis factor-α (TNF-α), interferon-γ (IFN-γ), interleukin-2 (IL-2), IL-4, and IL-5, was obtained from BD Pharmingen (San Diego, CA, USA).

Microdialysis sampling and analysis of carbamazepine

Commercially available CMA-12 (4 mm) microdialysis probes (CMA Microdialysis, North Chelmsford, MA) with a 20 kDa molecular weight cutoff (MWCO) polycarbonate–polyether membrane (PC) were prepared for first use as per the manufacturer instructions. The perfusion fluid flow rate was manually controlled with a BAS Bee microdialysis infusion pump (Bioanalytical Systems, Inc., West Lafayette, IN). The microdialysis perfusion fluid contained either saline solution (control, 0.9% m/v, g per 100 mL, NaCl) or saline supplemented with 1% m/v native β-CD or β-CD polymer (enhanced). The analyte sample was a well-stirred 100 µmol dm⁻³ carbamazepine solution prepared in saline and placed into a 10 mL beaker at room temperature. The blank control sample medium was a well-stirred saline solution in a 10 mL beaker. Two identical CMA-12 PC 4 mm microdialysis probes (control and enhanced) were simultaneously immersed into the blank stirred sample medium. Dialysates were collected every 5 min at 1.0 µL min⁻¹ from both probes. Three samples were obtained from this medium prior to simultaneous immersion of both microdialysis probes into the stirred 100 µmol dm⁻³ carbamazepine solution.

The dialysates were analyzed using a Shimadzu HPLC-UV system (Shimadzu Corporation, Kyoto, Japan) with a spherex 3 µm C-18 column (2 × 150 mm) (Phenomenex, Torrance, CA, USA) at a flow rate of 0.2 mL min⁻¹. The mobile phase consisted of 50/50 (v/v%) of 0.1 mol dm⁻³ sodium phosphate buffer, pH 2.6 and acetonitrile. The UV detector was set to 220 nm and the injection volume was 4 µL. The microdialysis RR was calculated based on the carbamazepine peak area ratios between the dialysates and the average peak area of three repeated injections from 100 µmol dm⁻³ sample medium before and after microdialysis sampling.

Microdialysis sampling and analysis of cytokines

Microdialysis sampling of the cytokines (TNF-α, IFN-γ, IL-2, IL-4, and IL-5) was performed in a microcentrifuge tube containing a 1.5 mL quiescent solution (1500 pg mL⁻¹) of all five cytokine standards spiked into the assay diluent obtained from the CBA kit. The assay diluent solution was used as the blank control sample medium. The BAS microdialysis pump was fixed onto a rotator made in-house. This rotator is controlled by a computer program that rotates the microdialysis pump by 180° in both a clockwise and counterclockwise direction. The rotator completes a cycle of clockwise rotation to 180° in 22.5 seconds and then 180° counterclockwise back to the starting position in the next 22.5 seconds. The first complete rotation takes 45 seconds. The rotator then continues the counterclockwise movement through another 180° in 22.5 seconds and returns clockwise back to the starting position in 22.5 seconds. These two separate cycles are repeated in 90-second intervals. This rotation serves to keep the cytokine antibody-immobilized microspheres in suspension within the microdialysis syringe (CMA/Microdialysis). Preventing bead settling is important since it causes difficulties with the flow cytometry analysis.

Two CMA/20 (10 mm) microdialysis probes with 100 kDa MWCO polyethersulfone (PES) membranes were used for cytokine collection. The probes have 20 cm of inlet and outlet tubing each with a dead volume of 3.6 µL. The probe inlet internal volume is 1.4 µL and the outlet internal volume is 2.4 µL. During enhanced microdialysis, the perfusion fluid contained a 1 ∶ 1 dilution containing one part of the antibody-immobilized bead suspension solution and one part of the manufacturer wash buffer. This ratio was used to stay consistent with the assay protocol, which requires equal volumes of sample, bead suspension and detection reagents. The probe for control experiments was perfused with the BD wash buffer by another BAS pump. The dialysates from the probe used for controls were collected every 20 min while dialysate for the probes containing the antibody-immobilized beads were collected every 40 min at 1.0 µL min⁻¹. Collected samples were stored at 4 °C prior to sample analysis. All experiments were carried out at room temperature that was measured daily and ranged between 23.5 and 24.5 °C.

Nine cytokine standards ranging from 20 to 5000 pg mL⁻¹ plus one negative control (0 pg mL⁻¹) were used to construct the calibration curve. The standards and the control dialysates were prepared similarly prior to analysis. For each sample, 15 µL of sample was added to 15 µL of mixed cytokine bead solution and 15 µL of phycoerythrin (PE) detection reagent on a Millipore Multiscreen^® BV 1.2 µm clear non-sterile 96-well filter plate, and then incubated for 2 h at room temperature protected from light. The filter plate was presoaked with 100 µL BD assay diluent per well and vacuum drained on a Millipore Multiscreen vacuum manifold. Dialysates (30 µL) containing 1 ∶ 1 (v/v) dilution of beads and wash buffer were directly analyzed by adding 15 µL of PE detection reagent on the filter plate. Before and after incubation, the plate was shaken for 5 min on an IKA MTS2/4 digital stirrer (VWR). After incubation and shaking, the plate was drained for washing (200 µL wash buffer per well). After 2 min shaking, the plate was drained again for bead resuspension (200 µL wash buffer per well) and analyzed on a BD FACSArray Bioanalyzer^™ (BD Immunocytometry Systems, San Jose, CA), which determines the mean fluorescence intensity of each bead. Microdialysis RR of cytokines was calculated from the detected dialysate cytokine concentration and detected average sample medium cytokine concentration (duplicate quantification before and after collection of dialysates).

Results and discussion

Cyclodextrin enhanced microdialysis of carbamazepine

By including native β-CD, or water-soluble epichlorohydrin-based β-CD polymer (β-CD-EPS) in the perfusion fluid, the steady-state microdialysis relative recovery (RR) has previously been shown to significantly increase.¹⁸ A potential concern using the cyclodextrin enhancement approach is that RR enhancements may be due to a build-up of membrane/cyclodextrin concentrations rather than a simple analyte trapping mechanism within the inner lumen of the probe membrane. Such a large static layer of the host–guest complex may contribute to false apparent RR enhancements observed after chemical concentration changes external to the microdialysis probe. If the microdialysis process cannot respond to rapid fluctuations in concentration, then no benefit results by including trapping agents in the perfusion fluid during the microdialysis sampling process.

A series of experiments was designed to investigate if the dialysate reflects concentration changes external to the microdialysis probe. Carbamazepine is a hydrophobic analyte with an octanol–water partition coefficient (log P) of 2.45 and binds with β-cyclodextrin (log K_f = 2.61).²²Fig. 2a shows the microdialysis concentration of carbamazepine from dialysates collected from controls that contain no trapping agents in the perfusion fluid and a probe which contains 1 m/v% β-CD as the trapping agent in the perfusion fluid. When both probes were immersed into the blank saline solution, there is no detected carbamazepine in the dialysates. The probes were immersed into a 100 µmol dm⁻³ carbamazepine solution directly after collection of dialysate #3 (15 minutes after the perfusion fluids were started), and the dialysate carbamazepine concentrations increased in both probes. These initial dialysates collected are denoted as # 4 (4.4 and 9.6 µmol dm⁻³, control and enhanced), # 5 (26.0 and 61.7 µmol dm⁻³, control and enhanced), and #6 (27.6 and 83.6 µmol dm⁻³, control and enhanced). For the control experiments, a steady state concentration does not appear to be reached until dialysate #7 at a concentration of 36.4 µmol dm⁻³ and was achieved nearly 20 minutes after the initial concentration change. This is in contrast to the probes containing β-CD in the perfusion fluid, which appear to reach a steady-state value at dialysate #6, which is 15 minutes after the concentration change. Each dialysate concentration detected reflects the time-averaged concentration of the sample over the sampling interval. For any concentration change external to the probe, the sampling interval includes a concentration average over the collection interval and also includes probe dead volumes.²³ Therefore the first dialysate collected after a concentration change represents a volume of sample related to the initial concentration and a volume of sample related to the new concentration.^21,24

Fig. 2 Microdialysis outflow concentrations of carbamazepine during a concentration step change experiment. Two CMA/12 4 mm polycarbonate (MWCO 20 kDa) microdialysis probes perfused with a control perfusion fluid (0.9% NaCl solution, ■) and an enhanced perfusion fluid (1 m/v% β-cyclodextrin in 0.9% NaCl solution, ●) at 1.0 µL min⁻¹. The dashed lines denote when the probes were placed into the new concentration.

The CMA/12 4 mm PC probe has 15 cm pieces of Teflon FEP tubing 0.65 mm O.D. × 0.12 mm I.D. with 1.8 µL on each side used for connection between syringe needle/probe inlet and probe outlet/dialysate collection vial. In addition to this tubing, the microdialysis probe has an inlet internal volume that is negligible and an outlet internal volume that is 3 µL according to the manufacturer. The combination of 1.8 µL of the internal outlet tubing volume and the 3 µL dead volume from the probe outlet causes the first dialysate collected after the concentration change (#4) to be contain low concentrations of carbamazepine. The carbamazepine concentrations exiting the control microdialysis probe exhibited concentration increases between dialysates #4 to #5, but did not reach an approximate steady-state value until dialysate #7. The approach to steady-state for carbamazepine took longer than 5 minutes using the 1.0 µL min⁻¹ flow rate combined with 5 minute sample intervals. This is in contrast with a published report for dopamine which is much more hydrophilic compared to carbamazepine, which took only 2 min to reach a steady-state value at 1.6 µL min⁻¹ after a concentration change.²⁰ Others have reported much longer time periods (up to three hours) for hydrophobic drugs to reach steady-state values.²⁵ The microdialysis RR of carbamazepine for control dialysates (#7–#10) at steady state was 38.6 ± 3.0% while for native β-CD enhanced dialysates (#6–#10) was 82.3 ± 4.9%. This represents a roughly two-fold enhancement over control RR values, which is comparable to our previously published work.¹⁸

After collection of sample 10 (50 minutes), both dialysis probes were removed from the 100 µM carbamazepine solution and returned to the blank solution. Fig. 2a shows the return to baseline concentrations for both probe conditions. The carbamazepine concentration from the probe with just saline solution slowly returned to baseline. More than 30 minutes was required for the control concentrations to reach less than 10% (<4 µM) of the maximum carbamazepine concentration obtained at steady state. This is in contrast with the β-CD infusion data, which reaches baseline faster than the control. Most likely these slow kinetics exhibited after membrane contact with high concentrations are the result of carbamazepine partitioning into the polymeric dialysis membrane as well as potential non-specific adsorption with the dialysate tubing. Carbamazepine complexation with the β-CD most likely prevents nonspecific adsorption onto the tubing and may help drive any partitioned carbamazepine out of the dialysis polymeric membrane.

Fig. 2b shows an additional experiment with carbamazepine and β-CD as the trapping agent. In this graph, the sampling time was extended to illustrate the differences in the maximum plateaus obtained. As in Fig. 2a, the control concentrations for carbamazepine were slower to reach a maximum value than those for β-CD and were slower to return to baseline.

Fig. 3 shows the microdialysis concentrations of carbamazepine from another series of dialysates collected from two microdialysis probes where one probe contained 1 m/v% water-soluble β-CD polymer (β-CD-EPS) as the trapping agent in the perfusion fluid. The results obtained are quite similar to those shown in Fig. 2a and 2b. The microdialysis concentrations for both the control and β-CD-EPS containing perfusion fluid reach a steady state within 15 minutes after the concentration switch (by sample #6). Upon placing the probes back into the blank saline solution, the return to baseline values of carbamazepine was again delayed as exhibited in Fig. 2a and 2b. As with β-CD, the β-CD-EPS enhanced dialysates reached baseline faster than the control. Again, it is likely that inclusion of the trapping agent, β-CD-EPS, removes any carbamazepine that may be partitioned into the dialysis polymeric membrane. In Fig. 3, the microdialysis RR for control dialysates (#6–#10) at steady state was 44.2 ± 3.8% while for β-CD-EPS microdialysis RR was 84.3 ± 4.8%. The enhancement observed for the β-CD-EPS enhancement was similar to our previously published report of two-fold.¹⁸

Fig. 3 Microdialysis outflow concentrations of carbamazepine during a concentration step change experiment. Two CMA/12 4 mm polycarbonate (MWCO 20 kDa) microdialysis probes perfused with control perfusion fluid (0.9% NaCl solution, ■) and enhance perfusion fluid (1% β-cyclodextrin polymer in 0.9% NaCl solution, ●) at 1.0 µL min⁻¹. The dashed lines denote when the probes were placed into the new concentration.

The concentration data presented in Fig. 2a, 2b and 3 clearly show that there was no temporal microdialysis recovery delay between control and enhanced experiments due to adding native β-CD or β-CD-EPS in the perfusion fluid. It has been reported that microdialysis can be used for reliable estimation of protein unbound carbamazepine both in vitro and in vivo, provided that binding to the plastic tubing is taken into account.⁸ The dialysate temporal response to the chemical concentration changes outside the microdialysis probe is not affected by adding CDs as trapping agents when the concentrations are increasing. However, in both cases, the return to baseline after a rapid concentration change was improved with the addition of CDs into the perfusion fluid. The addition of CD into the perfusion fluid caused the baseline concentrations of carbamazepine to return to zero approximately 10 minutes (or two sample volumes) earlier than controls without CD (n = 3, and data not shown).

The experiments with the cyclodextrins both used a five-minute sampling time. Others have reported sampling times coupled with chromatographic systems of one minute or less during microdialysis experiments with step changes in the external sample concentration.^20,21 While it would be possible to directly connect the dialysate outflow to the analytical detector, it is important to note that cyclodextrins when complexed with an analyte can reduce the absorbance properties for the analyte.²⁶ The chromatographic separation serves two purposes for these studies. First, it provides a means to detect the low-volume dialysates using UV. Second, the mobile phase organic modifier, acetonitrile, is known to compete for the binding site within the cyclodextrin cavity resulting in analyte displacement from the cyclodextrin on-column.¹⁸

Enhanced microdialysis sampling of cytokines using cytokine antibody-immobilized microspheres

The cytokine cytometric bead array kit requires a minimum sample volume of 15 µL. To achieve this volume, 20 µL of dialysate was collected for each dialysate with the additional volume being sufficient to compensate for both fluid ultrafiltration loss (5%) due to large PES membrane pore size¹⁴ and sample handling loss. In order to keep the cytokine quantification method consistent during the incubation, 30 µL dialysate samples including mixed cytokine antibody-immobilized beads originally used as trapping agent in the perfusion fluid to enhance the microdialysis RR of cytokines were required and 40 µL dialysates were collected for the same reason.

Fig. 4 shows the microdialysis concentrations of sequentially collected dialysates where the sample media outside of the membrane were switched between assay diluent buffer (blank control) and a 1500 pg mL⁻¹ solution containing the five cytokines (TNF-α, IFN-γ, IL-2, IL-4 and IL-5). The concentration data shown in Fig. 4a were obtained from the probe perfused with wash buffer while dialysates in Fig. 4b were obtained from an identical microdialysis probe perfused with a 1 ∶ 1 (v/v) dilution of mixed cytokine antibody-immobilized beads and wash buffer at 1.0 µL min⁻¹. The sampling intervals for the cytokine collection using probes with or without antibody-immobilized beads in the perfusion fluid are 20 and 40 minutes, respectively. Forty minutes is needed for the enhanced dialysates to maintain an equivalent number of beads per mL for the analysis. The cytokine concentrations obtained in the first dialysate samples after the concentration change (#5 for control and #3 for enhanced dialysates) were much closer to the steady-state values when compared to the carbamazepine data enhancements. The dead volume and the microdialysis approach to steady-state influenced the enhanced microdialysis RR of cytokines (dialysate #3 and #8) compared with the following two dialysates in steady-state microdialysis diffusion. Cytokine concentrations in the enhanced dialysate #6 and #11 were close to each other due to the dead volume in the outlet tubing and the outlet internal probe volume, as well as the residue cytokine solution left on the microdialysis probe during sample media switching.

Fig. 4 Microdialysis outflow concentrations of mouse Th1/Th2 cytokines. Two CMA/20 polyethersulfone (MWCO 100 kDa) 10 mm microdialysis probes were perfused with wash buffer (control) and 1 ∶ 1 (v/v) mixed mouse cytokine antibody-immobilized beads and wash buffer (enhanced): TNF-α (■), INF-γ (●), IL-5 (▲), IL-4 (▼), and IL-2 (♦). (a) Control microdialysis concentrations of mouse Th1/Th2 cytokines. (b) Microdialysis concentrations of mouse Th1/Th2 cytokines with inclusion of the antibody-immobilized beads. The dashed lines denote when the probes were placed into the new concentration.

The dialysate cytokine concentration changes obtained from dialysates collected using the enhanced approach (Fig. 4b) were significantly greater than those obtained from controls (Fig. 4a) after a concentration change. The temporal resolution for enhanced microdialysis sampling of cytokines was 40 min, which is much longer than control microdialysis sampling of cytokines due to analytical measurement requirements. It is hard to estimate the exact time needed to reach steady-state diffusion during enhanced microdialysis sampling of cytokines. For macromolecules such as proteins, the time to reach steady state concentrations of microdialysis diffusion is much longer than for small molecules. The time to reach steady state was estimated to be 40 min for conventional microdialysis sampling of macromolecules.⁶

Table 1 compares the microdialysis RR between the controls and antibody-immobilized microsphere enhancements for the five cytokines. Control microdialysis RR range from 1% (IL-5) to 10% (TNF-α). Among these five cytokines, IFN-γ showed the highest microdialysis recovery enhancement, increasing from about 2% to 70% while TNF-α showed the lowest microdialysis recovery enhancement (2.5 fold) increasing from about 10% to 35%. IL-2, IL-4 and IL-5 exhibited similar RR enhancement, about 3.5–5 fold.

Table 1 Microdialysis relative recovery of mouse Th1/Th2 cytokines at 1.0 µL min⁻¹ using CMA/20 polyethersulfone microdialysis probes immersed into a cytokine standard solution (1500 pg mL⁻¹) (mean ± SD)

	Relative recovery (%)
	Control (n = 12)	Enhanced (n = 6)
Tumor necrosis factor-α	9.9 ± 0.9	34.1 ± 9.7
Interferon-γ	1.5 ± 0.3	70.0 ± 21.4
Interleukin-5	0.9 ± 0.1	4.3 ± 1.7
Interleukin-4	8.7 ± 0.9	38.6 ± 10.0
Interleukin-2	3.2 ± 0.3	19.2 ± 5.2

---

Conclusions

Facilitated mass transport combining diffusion with chemical reaction is a well-known separation process to increase analyte mass transport flux and has been successfully applied to microdialysis sampling. The temporal response to a concentration change between enhanced microdialysis recovery achieved using appropriate trapping agents in the perfusion fluid as compared to controls for carbamazepine and some cytokines was not different. For carbamazepine, the inclusion of the β-CD trapping agent improved the overall dialysis process. No delay in the relative concentrations within the microdialysis perfusion fluid was observed for both carbamazepine and mouse Th1/Th2 cytokines during enhanced microdialysis procedures.

Acknowledgements

We gratefully acknowledge the support from NIH EB001441 and NSF 99-84150. We thank the staff at the Albany Medical College Immunology Core Instrumentation Facility. Samuk Pimanpang in the Department of Physics, Applied Physics and Astronomy at Rensselaer Polytechnic Institute made the rotator and wrote a computer program to keep the cytokine antibody-immobilized microspheres in suspension. Heidi Fletcher is acknowledged for performing additional carbamazepine experiments.

References

1. J. A. Bourne, Clin. Exp. Pharmacol. Physiol., 2003, 30, 16–24.
2. W. F. Elmquist and R. J. Sawchuk, Pharm. Res., 1997, 14, 267–288.
3. M. I. Davis, Anal. Chim. Acta, 1999, 379, 227–249.
4. P. M. Bungay, P. F. Morrison and R. L. Dedrick, Life Sci., 1990, 46, 105–119.
5. S. Mary, P. Muret, S. Makki, M. Jourdant, J. P. Belon, J. P. Kantelip, J. C. Henry and P. Humbert, Int. J. Pharm., 1998, 161, 7–13.
6. R. J. Schutte, S. A. Oshodi and W. M. Reichert, Anal. Chem., 2004, 76, 6058–6063.
7. S. Kjellström, N. Appels, M. Ohlrogge, T. Laurell and G. Marko-Varga, Chromatographia, 1999, 50, 539–546.
8. M. Lindberger, T. Tomson and L. Ståhle, Pharmacol. Toxicol., 2002, 91, 158–165.
9. C. Carneheim and L. Ståhle, Pharmacol. Toxicol., 1991, 69, 378–380.
10. M. Muller, R. Schmid, O. Wagner, V. Osten, H. Shayganfar and H. G. Eichler, J. Controlled Release, 1995, 37, 49–57.
11. W. J. Trickler and D. W. Miller, J. Pharm. Sci., 2003, 92, 1419–1427.
12. Y. Kurosaki, S. Nakamura, Y. Shiojiri and H. Kawasaki, Biol. Pharm. Bull., 1998, 21, 194–196.
13. A. Petersson, A. Amirkhani, B. Arvidsson, K. Markides and J. Bergquist, Anal. Chem., 2004, 76, 1678–1682.
14. X. Ao, T. J. Sellati and J. A. Stenken, Anal. Chem., 2004, 76, 3777–3783.
15. S. Kjellström, J. Emnéus, T. Laurell, L. Heintz and G. Marko-Varga, J. Chromatogr., A, 1998, 823, 489–496.
16. A. N. Khramov and J. A. Stenken, Anal. Chem., 1999, 71, 1257–1264.
17. K. W. Ward, S. J. Medina, S. T. Portelli, K. M. M. Doan, M. D. Spengler, M. M. Ben, D. Lundberg, M. A. Levy and E. P. Chen, Biopharm. Drug Dispos., 2003, 24, 17–25.
18. X. Ao and J. A. Stenken, Analyst, 2003, 128, 1143–1149.
19. T. Fukushima, E. Naka-aki, X. Guo, F. Li, T. Vankeirsbilck, W. R. G. Baeyens, K. Imai and T. Toyo'oka, Anal. Chim. Acta, 2004, 522, 99–104.
20. A. P. Newton and J. B. Justice, Anal. Chem., 1994, 66, 1468–1472.
21. A. Chen and C. E. Lunte, J. Chromatogr., A., 1995, 691, 29–35.
22. J. A. Stenken, R. Chen and X. Yuan, Anal. Chim. Acta, 2001, 436, 21–29.
23. L. Stahle, Eur. J. Clin. Pharmacol., 1992, 43, 289–294.
24. B. Hogan, S. M. Lunte, J. F. Stobaugh and C. E. Lunte, Anal. Chem., 1994, 66, 596–602.
25. V. J. A. Schuck, I. Rinas and H. Derendorf, J. Pharm. Biomed. Anal., 2004, 36, 807–813.
26. S.-F. Lin and K. A. Connors, J. Pharm. Sci., 1983, 72, 1333–1338.

---

Footnote

† Present address: Department of Radiation Oncology, University of Michigan, 1331 E Ann Street, Ann Arbor, MI 48109.

---