Anal. Chem. 2010, 82, 9492–9499
Temporal Resolution of Solid-Phase Microextraction: Measurement of Real-Time Concentrations within a Dynamic System Xu Zhang,† Ken D. Oakes,† Di Luong,† John Z. Wen,‡ Chris D. Metcalfe,§ Janusz Pawliszyn,| and Mark R. Servos*,† Departments of Biology, of Mechanical & Mechatronics Engineering, and of Chemistry, University of Waterloo, Ontario N2L 3G1, and Worsfold Water Quality Center, Trent University, Peterborough, Ontario K9J 7B8, Canada To address the challenge of measuring real-time analyte concentrations within dynamic systems, the temporal resolution of the solid-phase microextraction (SPME) approach has been investigated. A mass-uptake model for SPME within a dynamic system was developed and validated, with experimental factors affecting the temporal resolution (sampling time, agitation, SPME fiber dimensions, sample concentration and change rate, and instrument sensitivity) characterized. Calibration methods for time-resolved sampling in a dynamic system were compared. To demonstrate the efficacy of time-resolved SPME, this approach was successfully applied to investigate the binding kinetics between plasma proteins and pharmaceuticals, which verified a decrease in free pharmaceutical concentrations over time in the presence of bovine serum albumin. The current study provides the theoretical and logistical framework for applying SPME to the real-time measurement of dynamic systems, facilitating future SPME applications such as in vivo metabolomic studies. Solid-phase microextraction (SPME) has been widely employed for quantitative analysis in various fields ranging from environmental studies to pharmacokinetics due to its operational time and cost efficiencies.1-12 Advantages of SPME relative to * Corresponding author. Phone: (519) 885-1211, ext. 36034. E-mail:
[email protected]. † Department of Biology, University of Waterloo. ‡ Department of Mechanical & Mechatronics Engineering, University of Waterloo. § Trent University. | Department of Chemistry, University of Waterloo. (1) Gorecki, T.; Martos, P.; Pawliszyn, J. Anal. Chem. 1998, 70, 19–27. (2) Oomen, A. G.; Mayer, P.; Tolls, J. Anal. Chem. 2000, 72, 2802–2808. (3) Jahnke, A.; Mayer, P.; Broman, D.; McLachlan, M. S. Chemosphere 2009, 77, 764–770. (4) Heringa, M. B.; Hermens, J. L. M. TrAC, Trends Anal. Chem. 2003, 22, 575–587. (5) Pino, V.; Ayala, J. H.; Gonzalez, V.; Afonso, A. M. Anal. Chem. 2004, 76, 4572–4578. (6) Louch, D.; Motlagh, S.; Pawliszyn, J. Anal. Chem. 1992, 64, 1187–1199. (7) Pawliszyn, J. Solid Phase Microextraction: Theory and Practice; Wiley-VCH: New York; 1997. (8) Vaes, W. H. J.; Ramos, E. U.; Verhaar, H. J. M.; Seinen, W.; Hermens, J. L. M. Anal. Chem. 1996, 68, 4463–4467. (9) van Eijkeren, J. C. H.; Heringa, M. B.; Hermens, J. L. M. Analyst 2004, 129, 1137–1142. (10) Lord, H.; Grant, R.; Walles, M.; Incledon, B.; Fahie, B.; Pawliszyn, J. Anal. Chem. 2003, 75, 5103–5115.
9492
Analytical Chemistry, Vol. 82, No. 22, November 15, 2010
other in vivo sampling approaches have been well demonstrated, but from a temporal perspective include the ability to extract shortlived intermediate metabolites from a sample matrix. Since these short-lived species are absorbed into the polymer coating, they are physically prevented from interacting with enzymes in the sample matrix. A further advantage is SPME extracts exhibit less matrix effect than those yielded from microdialysis or ultrafiltration, both of which can introduce small quantities of salt buffer or biological fluid into the analytical instrument. To date, the calibration and quantitation of SPME has been based on theories and approaches developed within static systems where analyte concentrations were considered constant during sampling.6-9 Accordingly, most SPME applications have been restricted to relatively static systems, although some applications in highly dynamic systems have been reported.10-12 Temporal resolution is an essential in vivo sampling consideration as internal analyte concentrations may change rapidly with fluctuating environmental conditions,13,14 and inducible enzyme systems responsible for biotransformation can be up-regulated over time with exposure to many xenobiotics. Consequently, it is critical to evaluate the relevance of established SPME theory to dynamic systems to ensure the accuracy of the quantitative analysis. A thorough evaluation of the temporal resolution of the SPME approach is an important issue to ensure that real-time or short-term concentrations can be measured with acceptable accuracy and precision. The temporal resolution of SPME is defined as its capacity to accurately determine sample concentrations at specific points on a time continuum (referred to as “real-time concentrations”) and to clearly resolve two different concentrations in rapid succession. Defining the temporal resolution of a sampling approach is important as, to a certain extent, almost all biological and environmental systems are dynamic with fluctuations in analyte concentration over time owing to biological or physical processes (i.e., metabolism or sorption). Heterogeneous sample systems, as has been previously discussed, are inherently dynamic due to the spontaneous tendency of free molecules to achieve a uniform distribution.12 Defining the temporal resolution of sampling (11) Zhang, X.; Eshaghi, Al.; Cai, J.; Pawliszyn, J. J. Chromatogr., A 2009, 1216, 7664–7669. (12) Zhang, X.; Cai, J.; Oakes, K.; Breton, F.; Servos, M.; Pawliszyn, J. Anal. Chem. 2009, 81, 7349–7356. (13) Wang, M.; Roman, G. T.; Schultz, K.; Jennings, C.; Kennedy, R. T. Anal. Chem. 2008, 80, 5607–5615. (14) Schultz, K. N.; Kennedy, R. T. Annu. Rev. Anal. Chem. 2008, 1, 627–661. 10.1021/ac102186u 2010 American Chemical Society Published on Web 10/18/2010
approaches such as SPME is important as short-term analyte concentrations cannot be measured unless an appropriate sampling time is chosen. For example, resolving drug concentrations in a pharmacokinetic study at 5 and 10 min postadministration would of necessity dictate sufficient approach sensitivity to utilize a sampling time shorter than the interval between the time points (i.e., 0.13 min, tmax < 2.5 min) would be suitable to determine real-time FLX concentrations (Table 2). In a static 25 ng/mL CBZ system, the RSD for 2 min sampling durations (by five SPME fibers from the same batch) was 11.5%, while that of five sequential 2 min sampling trials by the same fiber was 6.7%, indicating interfiber variability can substantially reduce the precision. The RSDs did not change significantly from those of the 2 min trial when the sampling time was increased to 4 h, an interval greater than the equilibrium extraction time for CBZ (1.5 h) or FLX (3 h) under agitation. Consequently, a 2 min sampling is deemed reproducible. The effect of the sampling time on the TR-SPME accuracy was evaluated by comparing the relative recovery (measured concentration/real concentration) as determined with 2, 5, and 10 min sampling durations at the 60 min time point (during which the instantaneous concentration was 85 ng/mL); the relative recoveries were 102 ± 5%, 106 ± 7%, 112 ± 7%, respectively, for both CBZ and FLX. Here, the “measured concentration” is an integrated concentration over the sampled duration, while the “real concentration” is the theoretical instantaneous concentration (the equations for evaluation of the accuracy of TR-SPME are detailed in the Supporting Information). The relative recovery increased to 120 ± 6% when we employed a 10 min sampling duration at the 10 min time point (35 ng/mL after 10 min of infusion of the 1 (ng/mL)/min standard into the sample). These phenomena can be well explained by eq 5. At the 60 min time point, C0 was 85 ng/mL and b was 1 (ng/mL)/min. To determine the real-time concentration, a tmax of 8.5 min is required. Conversely, at the 10 min time point, b did not change, but C0 was 35 ng/mL, resulting in a tmax of 3.5 min. As a result, the 10 min sampling duration produced a greater violation of tmax at the 10 min, relative to the 60 min, infusion point, in turn generating less error at the 60 min time point (112%) than the 10 min time point (120%). The above-described numerical example is summarized in Table S1 in the Supporting Information. In contrast to the above results obtained with external calibration curves, the results with kinetic calibration yielded slightly reduced accuracy. At the 60 min time point, the kinetic calibration relative recoveries were 104 ± 7%, 109 ± 8%, and 117 ± 10% for the 2, 5, and 10 min sampling durations, respectively. The reduced accuracy may be attributable to additional error introduced during preloading of the deuterated analyte to the fiber. These results demonstrate both the potential and caveats associated with kinetic calibration at a relevant time range (2 min sampling duration) for monitoring real-time concentrations. Further in-depth research studying absorption/desorption within dynamic sample systems is ongoing and will be reported as available. As conceptualized, the TR-SPME theory can be important for developing methods to monitor dynamic systems, especially with respect to sampling duration. For example, upon re-evaluating our previous pharmacokinetic studies,22 we could be confident in the concentrations of diazepam determined using the 2 min, but not the 5 min, sampling duration as the concentration change rate for diazepam was 17 (ng/mL)/min (which according to eq 5 would permit a tmax of 4 min to measure the highest real-time concentraAnalytical Chemistry, Vol. 82, No. 22, November 15, 2010
9497
Table 2. Effects of the Rate of the Sample Concentration Change (b) on the Minimum (tmin) and Maximum (tmax) Fluoxetine Sampling Times
C0/(10b) 0.2/a tmax tmin temporal resolution
min min min min min-1
b ) 0.1 (ng/mL)/min
b ) 0.5 (ng/mL)/min
b)1 (ng/mL)/min
b)2 (ng/mL)/min
b)3 (ng/mL)/min
b ) 10 (ng/mL)/min
b ) 20 (ng/mL)/min
25.0 11.8 11.8 0.13 7.4
5.0 11.8 5.0 0.13 7.4
2.5 11.8 2.5 0.13 7.5
1.3 11.8 1.3 0.13 7.5
0.8 11.8 0.8 0.13 7.5
0.3 11.8 0.3 0.13 7.6
0.1 11.8 0.1 0.13 7.8
tion of 700 ng/mL). However, using data from the same study, 5 min was appropriate to measure the real-time concentrations of nordiazepam and oxazepam, the highest rates of concentration change of which being 5 and 2 (ng/mL)/min, respectively. Admittedly, there are circumstances when the measurement of real-time concentrations in a dynamic system is not possible, such as when tmin is longer than tmax, as shown in Figure S2 (Supporting Information). For example, when 10 min is needed for the SPME fiber to extract the minimum amount of analyte required for detection (potentially due to a small fiber coating volume, lower instrument sensitivity, or low sample concentrations) and if tmax from eq 4 is 2.5 min, we must increase the SPME fiber surface area and volume or use an instrument with greater sensitivity; otherwise, the initial sample concentration cannot be determined accurately. In addition, it is worth noting that when the concentration change rate is too slow to have a significant effect on the initial sample concentration, temporal resolution is of little importance in what is essentially a static system. Application: Studying the Protein Binding Kinetics of Pharmaceuticals. For drug discovery, an understanding of the drug-plasma protein binding kinetics is critical as plasma proteins, such as serum albumins, are important carriers of drugs and small bioactive molecules. SPME has been adopted as an effective experimental tool to determine protein binding constants and free concentrations of various pharmaceuticals;4,8,9,30-32 however, conventional SPME approaches have never been used to track rapid binding kinetics due to a lack of temporal resolution. In the current study, the dynamic binding process between BSA and CBZ/FLX was monitored by TR-SPME, as shown in Figure 4. The possible influence of albumin on SPME sorption kinetics discovered by J. L. M. Hermens and P. Mayer et al.2,4,31,32 was corrected using the kinetic calibration method with deuterated calibrants on fibers. For CBZ, the MD data were consistent with the results obtained by TR-SPME. However, MD data were not available for FLX (log Kow ) 4.64) as it is approximately 100 times more hydrophobic than CBZ (log Kow ) 2.40) and thus unsuitable for MD (which is not reliable for hydrophobic compounds which are sparingly soluble in the perfusion fluid, but can easily adsorb onto the polymeric materials of the microdialysis probe).33,34 The 2 min SPME analysis generated higher concentrations than the comparable 5 and 10 min duration SPME samples, especially during the early stages of the protein (30) Yuan, H.; Pawliszyn, J. Anal. Chem. 2001, 73, 4410–4416. (31) Heringa, M. B.; Hogevonder, C.; Busser, F.; Hermens, J. L. M. J. Chromatogr., B 2006, 834, 35–41. (32) Kramer, N. I.; van Eijkeren, J. C. H.; Hermens, J. L. M. Anal. Chem. 2007, 79, 6941–6948. (33) Groth, L.; Jorgensen, A. Anal. Chim. Acta 1997, 355, 75–83. (34) Davies, M. I.; Cooper, J. D.; Desmond, S. S.; Lunte, C. E.; Lunte, S. M. Adv. Drug Delivery Rev. 2000, 45, 169–188.
9498
Analytical Chemistry, Vol. 82, No. 22, November 15, 2010
binding trial (Figures 4 and S2, Supporting Information). Such a response is reasonable as the shorter 2 min sampling duration would integrate a time frame of relatively higher concentrations, while the 5 and 10 min sampling durations would incorporate additional 3 and 7 min intervals, respectively, during which free pharmaceutical concentrations would decline. BSA binding was a pseudo-first-order reaction up to t ) 30 min as the plots of ln [CBZ] and ln [FLX] versus time are linear over this interval (Figure S3, Supporting Information). Both the 2 and 5 min sampling durations generally met the minimum and maximum sampling time requirements specified by eqs 4 and 5. During the first 45 min of the binding reaction, the concentration change rate
Figure 4. Temporal changes in the free concentrations of CBZ and FLX as determined by TR-SPME after introduction of BSA protein. The SPME sampling from t ) 10 min to t ) 30 min and from t ) 90 min to t ) 240 min was validated by MD.
was about 16 (ng/mL)/min for CBZ and 10 (ng/mL)/min for FLX. Consequently, the maximum sampling time to measure the 800 ng/mL concentration (t ) 10 min) would be 5 and 7 min durations for CBZ and FLX, respectively, but a 2 min sampling interval exhibited better temporal resolution. In contrast, a 10 min sampling duration could not differentiate the sample concentrations in two successive time points, thus generating an overlapped concentration profile (Figures 4 and S2). According to eq S12 in the Supporting Information, the 10 min sampling duration underestimated the real-time CBZ concentration at the 40 min time point by 44% ((10 × 16)/(2 × 180) ) 0.44, where the sampling duration, t, is 10 min, with a C0 of 180 ng/min and b ) -16 (ng/mL)/ min), while the 5 and 2 min sampling durations generated less than 22% and 10% errors ((5 × 16)/(2 × 180) ) 0.22, (2 × 16)/ (2 × 180) ) 0.09), respectively. For both CBZ and FLX, the 10 min sampling duration in the early dynamic phase (t < 50 min) of the binding experiment biased the concentration-time profile by about 20 min (Figure S3). Effectively, the CBZ and FLX concentrations at t ) 45 min determined by 10 min SPME are comparable to those measured at t ) 65 min by 2 min SPME, illustrating the significant bias introduced by a longer than necessary sampling duration. The SPME data measured during the steady-state phase of the experiment (t > 90 min) were consistent with those derived by MD. CONCLUSIONS For the first time, the temporal resolution of SPME was investigated in detail and the mass-uptake model for a dynamic system proposed. In an application of the TR-SPME approach, the
protein binding kinetics of two pharmaceuticals were monitored, with real-time concentrations validated by MD during dynamic and static portions of the trial. The choice of TR-SPME sampling duration is critical, both from a temporal resolution and from a quantitative analysis perspective. Additional applications for measuring real-time concentrations in dynamic biological systems are being investigated. However, it is evident the TR-SPME approach can facilitate future work, such as metabolomic studies, where the real-time identification and quantification of transient metabolites in biochemical networks would be of significant interest. ACKNOWLEDGMENT This work has been financially supported by the Ontario Ministry of Research & Innovation Post-Doctoral Fellowship Program, the Natural Sciences and Engineering Research Council of Canada, the Canadian Water Network, and the Canada Research Chairs Program. We thank Ms. Claudia Lee and Ms. Shirin TaheriNia for their efforts in making the SPME probes, performing the experiments, and processing the instrumental data and the anonymous reviewers for their constructive comments, which improved the quality of the manuscript. SUPPORTING INFORMATION AVAILABLE Additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review August 19, 2010. Accepted October 1, 2010. AC102186U
Analytical Chemistry, Vol. 82, No. 22, November 15, 2010
9499
