In vivo solid-phase microextraction for monitoring

protocol

In vivo solid-phase microextraction for monitoring intravenous concentrations of drugs and metabolites Heather L Lord1, Xu Zhang2, F Marcel Musteata3, Dajana Vuckovic1 & Janusz Pawliszyn1 1 3

Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada. 2Department of Biology, University of Waterloo, Waterloo, Ontario, Canada. Albany College of Pharmacy and Health Sciences, Albany, New York, USA. Correspondence should be addressed to J.P. ([email protected]).

© 2011 Nature America, Inc. All rights reserved.

Published online 2 June 2011; doi:10.1038/nprot.2011.329

This protocol for in vivo solid-phase microextraction (SPME) can be used to monitor and quantify intravenous concentrations of drugs and metabolites without the need to withdraw a blood sample for analysis. The SPME probe is inserted directly into a peripheral vein of a living animal through a standard medical catheter, and extraction occurs typically over 2–5 min. After extraction, the analytes are removed from the sorbent and analyzed by, for example, liquid chromatography–tandem mass spectrometry. It has been validated in comparison with conventional blood analysis, and we describe here the in vitro experiments typically conducted during method development. The new-generation biocompatible SPME probes are designed specifically for extraction of semi-volatiles and nonvolatiles directly from aqueous samples and can be steam sterilized. Sorbents are coated on fine-gauge surgical steel wire (200-mm diameter), which is more rugged and biocompatible than conventional fibers (100-mm fused silica fiber). They incorporate a binding agent that resists fouling by the biological matrix and does not cause an immune response in the experimental animal. The sorbents used (coating thickness of ~50 mm) are selected for their affinity for the types of small molecules of interest. The procedure is illustrated by the analysis of benzodiazepines with polypyrrole-coated wires inserted into peripheral blood vessels of beagles, although it can be adapted for use in smaller animals. The in vivo sampling can require as little as 1 min, in which case the entire procedure from sampling to instrumental analysis can take as little as 30 min.

INTRODUCTION The concentration of analytes in blood can be analyzed either by drawing blood and performing the entire analysis in a laboratory (in vitro analysis) or by inserting a suitable probe into the blood stream (in vivo analysis) and performing some or all of the ana lysis on-site. In vitro analysis is by far the most common of these two options. After blood collection, red blood cells are typically removed and blood plasma or serum is used for further analysis. Although the procedures are more cumbersome than in vivo ana lysis, in vitro analysis offers the advantages of flexibility in the type of analytical method used, sensitivity, and the ability to store plasma for an extended period of time for later analysis. In vivo analysis, however, has some important advantages over in vitro analysis. In the case of monitoring circulating drug concentrations directly from an animal, in vivo analysis greatly simplifies and shortens the analytical effort and time required, and limits the exposure of personnel to blood1. It also offers the potential of monitoring dynamic processes, providing faster results2 and capturing shortlived and/or unstable analytes. A simple and fast means of monitoring circulating drug concen trations is of interest for both research and therapeutic purposes. For research, the ability to conduct faster pharmacokinetic studies in experimental animals can reduce the time and expense of phar maceutical development. For therapeutic applications, it provides the possibility of quickly verifying circulating drug concentrations. This is of particular interest for drugs that have a very narrow thera peutic range, that is, compounds for which there is a very small difference between the dose that is required to produce the desired effect and the dose that will cause toxicity or some other negative consequence for the patient. It is also of interest in patients who have undergone surgery and in intensive care patients, in whom 896 | VOL.6 NO.6 | 2011 | nature protocols

standard dosing regimens may not produce the expected circulating concentration of the drug3,4. In vivo analyses have historically been conducted using biosensors; the intravenous glucose sensor is a prime example5. Electrochemical biosensors, often based on carbon microfiber electrodes, have seen widespread use, particularly for monitoring electroactive neuro chemicals6 either intravenously or from other living tissues. Surface modifications7–9 and the incorporation of selective membranes10 have opened up the field for the analysis of a broader range of com pounds. The sensors described to date depend on electrochemical measurements and incorporate transduction and detection mecha nisms within the sensor. Everything reacting at the surface of the sensor will produce a signal to some degree, although they are typically designed such that the target analyte produces a stronger signal than non-target analytes. However, detection selectivity can never be absolute. By definition, analytes must typically be redox active. The most common analytes are electroactive neurotrans mitters11, urea12 and nitric oxide13. Optical detection has also been investigated to extend biosensor application to non-redox active compounds14. Despite these advances, sensors with sufficient selec tivity for drug analysis are not generally available. Solid-phase microextraction (SPME) is an ideal option for such analyses, as the requisite selectivity is provided by the off-line detection device rather than by the sensor itself, and there is no requirement for electroactivity. In SPME, an extraction phase is placed directly in contact with the system under study. Analytes of interest concentrate in the SPME extraction phase because of their higher affinity for the extraction phase relative to the sample matrix. Subsequently, the SPME device is removed from the system under study, the extracted analytes desorbed, then separated and


protocol finally measured in an analytical instrument—typically by liquid chromatography–mass spectrometry (LC-MS). SPME-based sen sors may be tailored for the target analyte by selecting a sorbent with appropriate affinity for the analyte. Microdialysis (MD) techniques are also commonly used for in vivo monitoring15, including monitoring of blood16. In this tech nique, an MD probe is placed into the target tissue. The probe typically consists of a semipermeable hollow fiber membrane that is sealed at one end and affixed onto a small tube at the other17. The device has another smaller tube located inside, as well as inlet and outlet ports at the distal end. After the probe is implanted into tis sue, a fluid (perfusate) mimicking the composition of the extracel lular fluid is delivered to the interior of the hollow fiber membrane. Thus, the inner surface of the membrane is continuously swept with fresh perfusate. Small molecules are free to permeate across the membrane into the lumen and are carried away in the perfusate, whereas large biomolecules are excluded from the perfusate. Analyte concentrations in the perfusate may be measured either online18 or off-line19. Perfusate concentrations can then be related to sample concentrations by calibration. For online monitoring, the perfusate line exiting the probe is directly connected to an analytical system, typically either an HPLC or a capillary electrophoresis sys tem. This is beneficial for automated analysis and for the analysis of very small sample volumes; however, in some cases, particularly when analyte levels are very low, the data are limited by sensitivity or time resolution. This is because the requirement of directly con necting the perfusate line to the analytical system limits the options for optimization. In these cases, analysis may be conducted off-line, by connecting the perfusate to a sample collection system instead. This can improve method sensitivity, as the analyst has options for preconcentrating the sample before analysis. It can also improve time resolution if, for example, the analytical method is lengthy, limiting the number of samples that can be collected and analyzed per hour. With off-line sample collection, samples may be collected rapidly, improving time resolution, and analyzed later as instru mental availability permits. However, off-line analysis has a limita tion in processing very small samples because of sample loss. MD is most appropriate for polar molecules, as less-polar molecules can be retained in the hollow fiber membrane. This can sometimes be problematic for drug analysis, as parent compounds are normally less polar and hence may preferentially adsorb to the membrane, rendering them less available for uptake by the perfusate. By contrast, in vivo SPME can be used for molecules with a range of polarities, depending on the nature of the sorbent selected. Sorbents may also be selected to extract biomolecules if that is the analytical target. Finally, MD techniques require that the experi mental animal be ‘tethered’ to a sampling system during the experi mental procedure. In vivo SPME eliminates this necessity. In vivo SPME is well suited for drug, endogenous metabolite and biomarker analysis and determination. In our experience, phase I drug metabolites, i.e., oxidation products of parent drugs, produced by the activity of cytochrome P450 enzymes in the liver, are well extracted by the same sorbents as are appropriate for the parent molecules. Indeed, in ANTICIPATED RESULTS, chromatographic evidence is provided of the analysis of drug metabolites, in addition to the parent drug, when only the parent was dosed. We have also observed that the technique is useful for non-targeted metabolite analysis, particularly when more polar sorbents are selected. We successfully monitored endogenous

polar metabolites using both the C18 sorbent recently introduced by Supelco and polar sorbents under development20,21. The protocol described here focuses on the in vivo determina tion of benzodiazepines in beagle peripheral blood vessels, using polypyrrole (PPY)-coated wires, as this was the first comprehen sive description of the process22. Additional publications have since appeared, outlining the fact that, with minor modification, the described procedure can be extended to smaller animals such as rodents23,24. New biocompatible sorbents and commercially avail able in vivo SPME probes have recently been introduced25. In the original description of the technique22, a fine surgicalgrade wire, coated with an appropriate extraction sorbent, is exposed to the flowing blood of an animal for a predetermined length of time through a conventional medical catheter. After selective preconcentration in vivo (that is, target analyte is con centrated in the extraction phase at a higher degree compared with non-target matrix components), the probe is removed and the extracted analytes desorbed, separated and quantified using conventional LC-MS equipment. More recently, we have expanded on the original concepts and investigated alternative calibration strategies1,26–28. The in vivo SPME sampling procedure described here is a major departure from the conventional meaning of sam pling. Conventionally, a portion of the system under study—in this case, blood—is removed from its natural environment, and the compounds of interest are extracted and analyzed in a labora tory environment. For SPME, there is no requirement to remove a blood sample from a living system. All the publications to date have confirmed that the data obtained are comparable with that obtained from conventional analysis. There are two main motivations for exploring in vivo configu rations, with either conventional or SPME techniques. The first is the desire to study chemical processes in association with the normal biochemical milieu of a living system, and the second is the impracticality frequently associated with removing suitably sized samples from living systems. As with any microextraction technique, SPME does not exhaustively remove compounds of interest from the investigated system, thus allowing one to avoid disturbing the normal balance of chemical components in the system being studied. It is also possible to study the distribution of a drug of interest among the blood compartments (plasma, cells) and assess drug-protein binding interactions. This method can also facilitate the non-destructive analysis of very small tissue sites or samples23,29,30. Here we present a method for monitoring parent drugs and their phase I metabolites from intravenous blood in beagles. In practice, with the introduction of commercially available probes having a selection of sorbent chemistries, the method could be expanded to a broad range of small-molecule analytes. We have studied intra venous pharmacokinetic profiles in pigs using essentially the same method. With slight variation, the method can also be applied to rodents; we have studied rats and mice24,31. In addition, sampling sites other than peripheral intravenous blood are feasible. We have used the probes to monitor the central venous compartment as well as various tissue sites. The application of kinetic calibration strategies has facilitated the in vivo monitoring of soft tissues (mus cle, adipose and brain) without the requirement of knowing sam ple volume or controlling temperature, agitation or diffusion in the sample. Moreover, the flexibility in calibration has facilitated the speciation of analytes in vivo, in terms of bound versus free nature protocols | VOL.6 NO.6 | 2011 | 897


protocol c oncentrations of analyte. If during calibration the fiber constant is determined from the linear region of the extraction isotherm from whole blood, then total drug concentration in circulating blood is determined. By comparison, if the fiber constant is determined from a binding isotherm in which the analyte is 100% free (e.g., physiological buffer) the free concentration of analyte in circulat ing blood is obtained. An important limitation of the approach is that there is limited facility for reinjecting a sample if there is a problem with it. For manual online desorption, the manner in which a partial injection of extracted analytes may be possible is not straightforward. With the current description of that method, a single injection loop is filled with the desorbed analyte in solvent and the entire volume of the loop is injected at once. For off-line desorption, although typically only a portion of the well-mixed desorption solution is injected, there is a motivation to limit the volume of desorption solution to maximize sensitivity; however, this also limits the abil ity to reinject a sample. However, a minimum of one reinjection is feasible in most experimental designs. In the described method, the identified goal is the investigation of a pharmacokinetic time profile with which to determine the fate of a chemical once dosed to an animal. The method presented is not, however, limited to determining pharmacokinetic time profiles. We have also used the method for monitoring endogenous compounds during untargeted metabolomics studies20,21 or for drug monitoring in which the goal is to achieve steady state levels for a period of time. Method sensitivity has proven to be sufficient for a very wide range of target analytes, having polarities (logP) ranging from compounds with negative logP values to those with logP in excess of 4.0 (see ref. 20). As with any animal research method, statistical relevance of the data must be considered. We normally conduct all extractions at least three times. In practice, this means that at least three animals are included in all experiments. Ultimately, however, a determination of the optimum number of animals to be included in a single experiment in order to provide statistically valid data for a particular experiment will be deter mined mathematically from estimations of the expected experimental variability. Statistical power analysis is commonly used to determine optimal sample size32. The considerations for the design of pharmacokinetic timecourse experiments by in vivo SPME are not significantly differ ent than considerations for a conventional analysis. Consideration should be given to selecting a therapeutically relevant dose of par ent compound and route of administration. Early sampling time points should be spaced more closely to better capture the faster rates of change in drug and metabolite concentrations. The limit of quantification of the analytical method should be verified as appro priate on the basis of the anticipated circulating concentrations during the expected time course. Total length of the experiment should be considered relative to any data known on the clearance rate of the drug and the limits of quantification of the analytical method. Consideration should be given to ensure stability of the extracts post sampling. Stability of drugs and metabolites on the sorbent versus those in the desorption solvent should be evalu ated and an appropriate storage temperature between sampling and analysis should be determined. The foregoing are all required for any pharmacokinetic experimental design, regardless of the sampling methodology used. The most important new consideration is in the design of the SPME probe. The probe design is affected by the nature of the 898 | VOL.6 NO.6 | 2011 | nature protocols

venous access provided. The probe length and coated sorbent length should be matched to the catheter length and length of straight vein present past the tip of the catheter. The overall length of the probe should be ~5 cm longer than the total length of the catheter from the tip to the end of the needle injection closure (PRN adapter used here). This allows 1–2 cm to be exposed in the vein with 3–4 cm of wire remaining outside the catheter during sampling. Further, the outer diameter of the probe must be significantly smaller than the inner diameter of the vein, such that the probe does not impede blood flow. For rodents, an external sampling port has been used20,21,24,33 as it is not currently possible to produce probes with a small enough diameter to access the veins of rodents because of the difficulties in reproducibly coating very thin wires. Table 1 outlines the developments in new biocompatible sorb ents for in vivo analysis and the target analytes we have studied with them. Some of these new and interesting developments are also described in more detail below. Experimental design Much of the method development for in vivo SPME is conducted through in vitro experiments. In vitro extractions are conducted with samples prepared in the laboratory and spiked with known concentrations of pure drug standards. Optimization of general in vitro SPME procedures has been covered in detail elsewhere34–37. Those parameters that are of particular importance for in vivo applications are reviewed here. An important advantage of the SPME technique is its inde pendence from needing to have a defined sample volume when extraction is conducted under conditions of negligible deple tion; that is, only a very small (insignificant) amount of the total analyte is removed from the sample during extraction. As long as sample volume is significantly greater than the product of the fiber constant (KfsVf ), the conditions of negligible depletion are met, and in vivo SPME can be used for quantitative analysis. Because animal blood volume is variable and the magnitude of product KfsVf depends on the nature of analyte and the SPME probe selected for the experiments, it is important to verify the volume required for non-depletion during extraction for in vitro samples, such that in vivo extraction results will be comparable. This is done by spik ing a known amount of the analyte into a whole-blood sample of known volume, determining the free fraction from a literature reference to the extent of protein binding (i.e., total mass of analyte free in solution), performing an extraction and determining the mass extracted and finally verifying that the mass extracted is