Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 2

Liquid Chromatography – Mass Spectrometry Analysis of Short-lived Tracers in Biological Matrices

MARTIN LAVÉN

Exploration of Radiotracer Chemistry as an Analytical Tool

ISSN 1651-6214ISBN 91-554-6126-3urn:nbn:se:uu:diva-4727

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2005

Kabhi Khushi Kabhie Gham

Sometimes Happiness, Sometimes Sorrow

-Karan Johar

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals:

I Determination of flumazenil in human plasma by liquid chromatography - electrospray ionisation tandem mass spec-trometry Martin Lavén, Lieuwe Appel, Robert Moulder, Niklas Tyrefors, Karin Markides and Bengt Långström. Journal of Chromato-graphy B, 2004, 808, 221-227.

II Analysis of microsomal metabolic stability using high-flow-rate extraction coupled to capillary liquid chromatography–mass spectrometry Martin Lavén, Karin Markides and Bengt Långström. Journal of Chromatography B, 2004, 806, 119-126.

III Determination of metabolic stability of positron emission tomography tracers by LC-MS/MS: an example in WAY-100635 and two analoguesMartin Lavén, Oleksiy Itsenko, Karin Markides and Bengt Lång-ström. Submitted to Journal of Pharmaceutical and Biomedical Analysis.

IV Radionuclide Imaging of Miniaturized Chemical Analysis SystemsMartin Lavén, Susanne Wallenborg, Irina Velikyan, Sara Bergström, Majda Djodjic, Jenny Ljung, Oskar Berglund, Niklas Edenwall, Karin E. Markides and Bengt Långström. Analytical Chemistry, 2004, 76, 7102-7108.

V Imaging of peptide adsorption to microfluidic channels in a plastic compact disc using a positron emitting radionuclide Martin Lavén, Irina Velikyan, Majda Djodjic, Jenny Ljung, Oskar Berglund, Karin Markides, Bengt Långström and Susanne Wallenborg. Submitted to Lab on a chip.

Reprints were made with kind permission from the publishers.

The author’s contribution to the papers:

Paper I: Wrote the paper, in close discussion with Lieuwe Appel. Planned and performed the experiments, except for the LC-MS analysis, which was headed by Niklas Tyrefors. Paper II: Planned, performed the experiments and wrote the paper. Paper III: Planned, performed the experiments, except synthesis of the compounds, and wrote the article. Paper IV: Wrote the paper, planned and performed the extraction column experiments. Supervised and took part in planning student projects on CD imaging with Susanne Wallenborg and took part in initial planning of PDMS experiments. Paper V: Wrote the major part of the paper. Supervised and took part in planning student projects with Susanne Wallenborg. Planned and per-formed quantification experiments and some of the pH tests.

Bengt Långström and Karin Markides acted as supervisors in all papers.

Contents

1 Introduction................................................................................................11

2 The tracer technique...................................................................................132.1 Positron emission tomography (PET) ................................................132.2 PET tracers .........................................................................................152.3 Determination of tracers in plasma ....................................................15

3 Analysis of tracers using liquid chromatography – mass spectrometry.....183.1 Liquid chromatography ......................................................................18

3.1.1 Efficiency, speed and resolution.................................................193.1.2 Sensitivity and detection limits...................................................21

3.2 Mass spectrometry..............................................................................223.2.1 Electrospray ionisation ...............................................................233.2.2 Collision induced dissociation ....................................................24

3.3 Liquid chromatography – mass spectrometry ....................................263.3.1 Striving for maximum signal intensity: an example ...................263.3.2 Ionisation suppression ................................................................273.3.3 Quantification .............................................................................283.3.4 Validation ...................................................................................29

3.4 Implications of using LC-MS in PET tracer analysis.........................29

4 Sample preparation ....................................................................................314.1 High flow rate extraction....................................................................324.2 A fast and simple sample preparation step .........................................35

5 Metabolism ................................................................................................365.1 WAY-100635 and analogues .............................................................38

6 Radiotracer chemistry applied ...................................................................416.1 Detection of annihilation radiation.....................................................416.2 Radionuclide imaging ........................................................................42

6.2.1 Extraction column imaging.........................................................456.2.2 Imaging of peptide-surface interactions within microchannels ..46

7 Concluding remarks and future outlook.....................................................48

8 Acknowledgements....................................................................................50

9 Summary in Swedish .................................................................................52

10 References................................................................................................54

Abbreviations

APCI atmospheric pressure chemical ioni-sation

Bq Becquerel CID collision induced dissociation DOTA 1,4,7,10-tetraazacyclododecane-

1,4,7,10-tetraacetic acid ESI electrospray ionisation HLB hydrophilic lipophilic balance 5-HT1A 5-hydroxytryptamine-1AIS internal standard LC liquid chromatography MALDI matrix assisted laser desorp-

tion/ionisationMRM multiple reaction monitoring MS mass spectrometry m/z mass-to-charge ratio PDMS poly(dimethylsiloxane) PET positron emission tomography PFP pentafluorophenyl Re Reynolds number RSD relative standard deviation SIM selected ion monitoring S/N signal-to-noise ratio SPE solid phase extraction SPR surface plasmon resonance SRA specific radioactivity

+ positrondp particle diameter DM diffusion coefficient of the analyte in

the mobile phase H height equivalent of a theoretical

plateN plate number

P pressure drop u linear velocity Vi injection volume

11

1 Introduction

In tracer chemistry, isotopically substituted molecules can be used to trace the behaviour of the corresponding unmodified molecules. The label may be a stable isotope, such as 2H or 13C, or a radionuclide, such as 3H or 11C. Ap-plying radiotracer chemistry to biological systems permits the visualisation and quantification of a number of biochemical processes. Positron emission tomography (PET)1, where short-lived positron emitting radionuclides like 15O, 13N, 11C and 18F are used, provides the possibility to follow the uptake of a labelled substance, the radiotracer, in living species. One can for instance investigate whether a drug candidate reaches the intended receptors and quantify the amount of drug. PET is used clinically for cancer localisation2,in neurology3 and cardiology4, but also in basic biochemistry1, 5 and drug development6, 7.

Radiotracer chemistry can also be applied to areas other than medicine and biochemistry. Analytical chemistry typically involves qualitative and quantitative determinations that are derived from a chain of events that in-clude sampling, sample preparation, separation and detection. These steps may be discrete or integrated in an analysis system. By using a radiotracer it is possible to monitor the analyte along its path through the analytical chain or system.5, 8, 9 Processes such as adsorption and analyte losses may for ex-ample be studied. Applying radiotracer chemistry to an integrated system can be of particular value since information can be obtained from selected parts of the system without physically removing these parts. Radiotracer chemistry can thus be applied as a valuable tool in analytical method devel-opment and in fundamental studies of chemical analysis systems.

High mass sensitivity may be obtained in the analysis of PET radio-tracers, due to efficient labelling methods with short-lived radionuclides.6This sensitivity decreases however rapidly with time, because of the short half-lives, and limits the time frame for radiotracer analysis. This can com-plicate for instance radiotracer analysis of blood in a PET study. Since a radiolabelled substance not only consists of isotopically substituted mole-cules, but also of unmodified molecules, other techniques that do not rely upon radioactivity measurements may be used in the analysis of radiotracers. Due to the very low concentrations of the PET tracer in for instance plasma (typically sub nM), this must be a truly sensitive technique. Liquid chroma-tography – mass spectrometry (LC-MS) may provide very low detection

12

limits in the analysis of biological samples.10 However, to meet such sensi-tivity requirements, calls for rigorous method development.

In the development of new PET tracers, pharmacokinetic properties such as body distribution, receptor binding affinity and metabolism need to be studied. These can be examined using radioactivity based methods such as autoradiography. LC in conjunction with radio-detection may be used for metabolism elucidation. In the latter case, again, LC-MS could be a tech-nique of great value. By analysing molecules composed of stable nuclides, it is possible to perform repeated sampling and measurements without the need of a new radiosynthesis for each experiment.

This thesis deals with a combination of LC-MS, radiotracer chemistry and analytical methodology. LC-MS methods were developed to improve PET and radiotracer technology, while radiotracers were used in the development of LC-MS methods and techniques. Also, microfluidic channels designed for biochemical and analytical applications were studied by peptide-tracer imag-ing.

The aims of the study were:

To develop methods for the determination of radiotracers in biological matrices which are independent of radioactivity measurements. These concerned quantification of a tracer in plasma from PET subjects (paper I) and metabolic stability of tracers in microsome preparations (papersII-III) by LC-MS analysis.

To explore the use of radiotracer chemistry as an analytical tool, evalu-ated by imaging molecular interactions in miniaturised chemical analysis systems (papers IV-V) and by supporting LC-MS method development (papers I-II).

13

2 The tracer technique

In 1911 George de Hevesy was assigned the task of separating the radioac-tive radium D (what was later found to be 210Pb) from lead.11 Hevesy worked with the project for almost two years, but “failed completely”. However, Hevesy instead made use of the inseparability of radium D from lead and transformed the failure into a very useful concept; the tracer principle. Ra-dioactive lead could thus be used as an “indicator” of lead, to trace the stable isotope and study its chemical and biological behaviour.11, 12 The uptake of lead into plants could in this way be studied by incubating the roots in a so-lution containing a mixture of lead and radioactive lead, followed by radio-activity measurements of different parts of the plant.13 Another, anecdotal, tracer experiment is described below.I, 14

The tracer principle is today used in biomedical science and forms the ba-sis of imaging techniques such as PET and single photon emission computed tomography (SPECT).1 Other medical imaging techniques include magnetic resonance imaging (MRI)15 and computed tomography (CT)16. More recently introduced is tissue imaging by matrix assisted laser desorption/ionisation (MALDI) MS.17 These latter imaging modalities are, however, not based on the tracer principle.

2.1 Positron emission tomography (PET) PET is used in biomedical research, drug development and for clinical appli-cations, e.g. in oncology.1, 2, 5-7 The technique is non-invasive and can pro-vide functional imaging of the brain and is therefore increasingly used in research on neurological disorders such as Parkinson’s disease, Alzheimer’s disease and epilepsy.3 Using PET enables quantification of receptor occu-pancy, cerebral blood flow and metabolism. Due to the high sensitivity, very

I The first tracer experiment?During his stay in Manchester in 1911 Hevesy was lodged in a boarding house. He suspected that the landlady only served freshly prepared meat on Sundays and recycled the meat into hash, goulash and other dishes for the rest of the week. When he brought up the subject with the landlady she answered with indignation that she served fresh meat every day. Hevesy, however, spiked some leftover meat on his plate with a small amount of radioactive material and brought an electroscope to the dining room a few days later. He was then able to detect radioactivity in the food served that day! This experiment was however never published…

14

low amounts of tracer (a few µg) can be administered to humans, which has led to the PET microdosing concept, for early clinical drug development.6

With PET, short-lived +-emitting radionuclides, such as 15O, 13N, 11C,68Ga, 18F and 76Br, are used for tracer purposes (Table 1). These are typically produced using accelerated charged particles for nuclear reactions. 68Ga is obtained from a 68Ge/68Ga generator.18 This radionuclide was used in papersIV and V, whereas 11C was used in papers I, II and IV.

Table 1. Radionuclides typically used in positron emission tomography5, 6

Radionuclide Half-life Production modeaTheoretical SRAb

(GBq/µmol)

Typical SRAb

(GBq/µmol)

15O 2 min 14N(d,n)15O 3.4 x 106 - 11C 20 min 14N(p, )11C 3.4 x 105 10-20068Ga 68 min 68Ge 68Ga 1 x 105 -18F 110 min 18O(p,n)18F 6.3 x 104 50-40076Br 16 h 76Se(p,n)76Br 7.1 x 103 50-200aShowing nuclear reactions where d = deuteron, n = neutron, p = proton and = alfa-particle bSpecific radioactivity, SRA, radioactivity per unit mass.

The radionuclides used in PET are all neutron deficient, which causes in-stability. During a decay event, a proton is converted into a neutron, with the release of a positron, a neutrino and kinetic energy, which is divided be-tween the two particles. The positron will loose its kinetic energy as it trav-els through and interacts with the surrounding environment. The range is dependent on the initial kinetic energy and the medium which is traversedII,

19. When most of the kinetic energy has dissipated, the positron will interact with an electron, resulting in annihilation of the two particles. The masses of the two particles are converted into two high-energy (511 keV) photons (an-nihilation radiation), travelling in nearly opposite directions. These last two characteristics are particularly useful for PET imaging. The energy of the photons is of such quantity that a human body may be traversed without total absorption. A positron emitting radionuclide can therefore be traced in a living body by means of external detection. The second feature, opposite directions of the simultaneously emitted annihilation photons, facilitates localisation of the radiation source, when a ring array of detectors (the PET scanner) is used.20

IIA 11C positron displays a maximum energy of 0.96 MeV, with a maximum and average range of 3.9 mm and 0.9 mm, respectively, in water.

15

2.2 PET tracers With PET, the radionuclides are rarely used in an ionic form, but they are incorporated into molecules. The building blocks of living organisms, or-ganic molecules, consist of carbon and hydrogen, and often oxygen and ni-trogen. Substituting one of these stable isotopes with a positron emitting isotope will yield a molecule that differs from its stable counterpart by only one neutron. Such a substituted molecule will in principle display similar chemical and biological behaviour as the unmodified molecule. A significant number of labelling methods have been developed for the production of both small molecule and large biomolecule tracers.5, 6, 21, 22

An important feature in radiotracer chemistry is the specific radioactivity (SRA) of the radiolabelled compound. The SRA is defined as the amount of radioactivity per amount of substance, expressed in mass or mol substance. The SRA is inversely proportional to the half-life of the radionuclide.III This means that very high theoretical SRA values may be obtained with short-lived radionuclides (see Table 1). Detection of such a tracer can be very mass sensitive, since a small mass may display a high radioactivity. How-ever, the short half-life also leads to a reduced sensitivity in the radioactivity measurements with time. Additionally, due to isotopic dilution in the pro-duction of the tracer, the obtained SRA is lower than the theoretical value. From the figures in Table 1, it can be calculated that 1 molecule out of 1700 to 34000 is typically labelled in a 11C-labelling synthesis. The SRA is usu-ally measured with radioactivity measurements of the labelled molecules and LC separation coupled to ultraviolet absorption (UV) detection for quantifi-cation of the total fraction. Alternatively, MS can be used in the determina-tion of the unmodified fraction.23 In this thesis, radiotracer is defined as the entire system of molecules, covering both labelled and unmodified mole-cules.

2.3 Determination of tracers in plasma In a PET study, the concentration of the tracer generally needs to be estab-lished. This can readily be performed if the specific radioactivity is known, whereby radioactivity concentration is converted into mass of substance. However, the PET camera detects only a radioactive signal, it cannot distin-guish between a signal from the tracer and radiolabelled metabolites. If a tracer displays substantial metabolism other techniques must therefore be used to determine the amount of intact tracer. To overcome this limitation plasma samples are withdrawn from the subject during the PET investiga-

III The theoretical specific radioactivity can be calculated from: A = ln2N/T1/2, where A repre-sents the radioactivity, T1/2 the half-life and N the number of atoms.

16

tion. These samples are typically analysed using techniques such as LC, thin layer chromatography (TLC) or solid phase extraction (SPE).24 The intact tracer is separated from metabolites and the ratio between unchanged tracer and the total amount of radioactivity in plasma is determined with radio-detection.24 The sensitivity obtained from these methods can be high and they are relatively simple. However, the mass sensitivity decreases rapidly with time, as a result of the short half-life of the radionuclides used. This may result in data of low precision and accuracy. Additionally, it is only possible to perform analysis within a limited time frame.

0

1000

2000

3000

4000

5000

6000

7000

5 10 20 40 60Time of plasma withdrawal (min)

Cou

nts

0

1

2

3

4

5

6

7

8

9

RS

D (%

)

Figure 1. Scintillation well counter measurements of intact radiotracer in plasma. Samples were obtained from a [11C]flumazenil PET investigation, performed at Uppsala Imanet, where 272±6 MBq had been injected in a human subject, in three consecutive scans. Separation from metabolites was performed by LC. Number of counts vs. time of blood withdrawal from start of scan is plotted, with actual nr of counts ( ) and decay corrected nr of counts ( ). The estimated relative standard deviation (RSD) of non-decay corrected measurements ( ), calculated from the square root of the number of counts, is also plotted. Note that the scale on the x-axis is non-linear.

The errors associated with the measurement of radioactive decay (stan-dard deviation), can be approximated to the square root of the number of counts.25 At a high number of counts a high precision may be obtained, but as time passes the relative error increases. In Fig. 1 data obtained from ra-dioactivity measurements of intact tracer in plasma from a [11C]flumazenil PET study are plotted. The actual number of counts, as well as decay cor-rected data, are given in the graph. Additionally, the estimated relative stan-dard deviation (RSD) of the actual number of counts has been added. It can

17

clearly be seen that the precision is decreasing with time, as the number of counts becomes lower and lower. Another source of error, not added to the graph, is fluctuations in the background, which may become significant at a low number of counts. The decrease in radioactivity is a result of both the fast decay process of 11C and biological removal, through urine and bile ex-cretion, and by metabolism. The decay corrected data give an estimate of the rate of biological removal. The difference in rates between the actual number of counts and decay corrected data should thus point to the effect of the de-cay. In the presented case, the data indicate that the decay process was a highly contributing factor to the low number of counts and reduced precision at late time points.

Additionally, the short half-life of 11C restricts the analysis of a large number of plasma samples, unless these can be analysed in parallel, which would increase the quality of the plasma data. It would therefore be benefi-cial with a complementary technique for determination of the quantity of unchanged tracer in plasma. Considering that only a fraction of the mole-cules in a tracer batch is labelled with the radionuclide, a technique measur-ing the unmodified, non-labelled, molecules could be used. This would re-move the sometimes hampering time restrictions associated with radio-analysis of short-lived radionuclides. The technique must however be of high sensitivity, since the typical concentration of a PET tracer in blood is in the sub nM region. Liquid chromatography coupled to mass spectrometry is such a technique, which can provide very high sensitivity together with mo-lecular information.

18

3 Analysis of tracers using liquid chromatography – mass spectrometry

As discussed in chapter 2, a complementary technique that does not rely on radioactivity measurements for plasma analysis in PET studies would be valuable. Previously, LC-MS has been used to determine the plasma concen-tration of raclopride in a human PET investigation.26 The sensitivity obtained permitted quantification of 0.2 nM raclopride. This was sufficient to monitor the tracer plasma concentration during 60 min and the study thus demon-strated the feasibility of LC-MS quantification in a PET study. In paper I,the concept of using LC-MS for obtaining supportive metabolic information was extended to cover multiple PET scans for the tracer flumazenil, where plasma concentrations ranging from 0.07 to 0.21 nM were quantified. LC-MS may of course be used in other PET related contexts, such as in preclini-cal tracer development. In papers II-III, LC-MS was used in the analysis of in vitro metabolic stability of tracers.

3.1 Liquid chromatography When analysing a complex sample a separation of its components is neces-sary. A number of techniques have been developed for separation in liquid, gas and supercritical fluid phases, such as LC, gas chromatography (GC), supercritical fluid chromatography (SFC), capillary electrophoresis (CE) and capillary electrochromatography (CEC). Of these, LC is the most practically useful on account of its simplicity and wide versatility. In the opimisation of an LC method, parameters such as stationary phase, particle structure, col-umn dimensions, flow rate, mobile phase composition and temperature need to be considered. A current trend has been towards the use of shorter col-umns (typically 50 mm) packed with smaller particles ( 3.5 µm) for fast and efficient analysis.27 Additionally, columns of reduced internal diameter ( 2.1 mm) are typically used with electrospray ionisation (ESI) MS, to minimise sample dilution and obtain flow rates compatible with the detec-tion technique28.

19

3.1.1 Efficiency, speed and resolution

The efficiency of a column is a description of the broadening of a peak as it passes through the column. It can be quantified by the plate number (N) or the height equivalent of a theoretical plate (H), where a low value of H indi-cates a high efficiency. The resulting plate height can be described by differ-ent band broadening contributions, summarised in the following equation29-31

M

pM

M

pp D

udC

uD

BDud

AdH233.0

(1)

where A, B and C are coefficients, dp is particle diameter, u linear velocity and DM the diffusion coefficient of the analyte in the mobile phase.IV It is possible to reduce the A coefficient by using homogeneous particles that are efficiently packed.32 The use of non-porous particles33 can reduce the C coef-ficient, however at the cost of reduced sample loading capacity and analyte retention due to lower surface area34. Fig. 2, displaying a theoretical H/u curve calculated from equation (1), clearly illustrates the dependence of plate height on linear velocity and particle size. The theoretical minimum plate that can be reached is significantly reduced with particles of smaller diame-ters. In addition, the slope of the curve is markedly reduced for a column packed with small particles. This means that fast linear velocities may be used for rapid analyses.

05

1015202530354045

0 2 4 6 8 10Linear velocity (mm/s)

Plat

e he

ight

(µm

)

5 µm

3.5 µm

1.5 µm

Figure 2. A theoretical H/u curve, plotted for particle diameters of 1.5, 3.5 and 5 µm, using equation (1) with DM = 10-5 cm2/s, A = 1, B = 2, C = 0.1.

IV The first term describes bandspreading due to flow variations caused by eddy diffusion and laminar flow. The second term describes longitudinal diffusion, which is only of significance at reduced flow rates not often used in LC. The third term describes bandbroadening due to resistance to mass transfer, mainly caused by stagnant mobile phase in the pores of the col-umn particles.

20

The velocity can however only be increased to a certain point, since the pressure drop ( P) is proportional to the linear velocity and inversely pro-portional to the square of the particle diameter.V, 35 As an example, P for a 100 x 2.1 mm column with dp = 1.5 µm, operated at optimal flow rate and normal conditionsVI, would theoretically be 485 bar. Standard commercially available LC pumps today can tolerate pressures up to approximately 400 bar. To reduce the pressure shorter columns may be used, however at the cost of efficiency. It is also possible to perform the chromatography at ele-vated temperatures, thereby decreasing the viscosity and thus the back pres-sure.36, 37 Additionally, an increase in temperature, and reduction in viscosity, leads to a higher analyte diffusion, which can result in a higher efficiency. Another alternative to deal with a high P, is to construct systems that can cope with higher pressures. This has led to the development of ultra-high pressure liquid chromatography (UHPLC), where pumps, connections and columns can withstand pressures of up to 5000 bar.38 Non-porous particles of sizes around 1-1.5 µm, down to 0.67 µm39, have been used to generate up to 300 000 plates in UHPLC.38, 40 Recently Mellors and Jorgenson34 described the use of commercially available 1.5 µm porous particles in a UHPLC con-text.

With LC, one cannot only consider the efficiency, but one must also take into account the resolution (Rs)VII of a separation. Interestingly, Rs is propor-tional to N1/2, which implies that an increase in speed can be gained without a loss of resolution at the same rate. A reduction of column length from 100 to 50 mm will thus yield an analysis that is double as fast, but only accom-panied by a loss in resolution of 30%. Therefore, a current strategy has been to use columns of shorter lengths packed with smaller particles, for faster analysis without significant losses in resolution.27 In paper I a 50 x 2.1 mm column was used with dp = 3 µm. If a column packed with 5 µm particles had been used, while operated under optimum flow and maintaining the same efficiency and resolution, the theoretical analysis time would have been three times higher.

The discussion above has been based on isocratic conditions. A gradient elution (used in papers I-III) offers improved separation capability and peaks are generally more compressed than in isocratic elution, which can result in an increased sensitivity.41

V The pressure drop, P = Lu/dp2, where, L is the length of the column, is the flow resis-

tance parameter, related to the structure of the column packing, and the viscosity of the mobile phase. VI DM = 10-5 cm2 s = 600, = 1 mPas.

VII

k

kNRs 1

1

4 where is the selectivity and k the capacity or retention

factor of the second peak.

21

3.1.2 Sensitivity and detection limits Sensitivity is defined as “the slope of the calibration curve”42. Consequently, if a method displays a high sensitivity a small change in analyte concentra-tion or amount will yield a great change in detector signal. Generally, a high sensitivity also entails the ability to detect low analyte concentrations. The term sensitivity has therefore often been associated with the ability to detect a certain concentration or amount of analyte. Accordingly, a high sensitivity indicates that low concentrations of analyte can be measured with the method.

In this thesis work ESI-MS was used for detection (paper I-III), a tech-nique generally regarded as concentration sensitive43. To obtain low detec-tion limits, the concentration of the analyte in the peak (Cp) should thus be maximised. This may be achieved by increasing the injection volume (Vi) as indicated by the following equation:

2R

iip

V

NVCC (2)

where Ci is the analyte concentration in the injected sample and VR the reten-tion volume.44 Also, bandbroadening in the chromatographic system should be minimised. One must, however, consider the increased bandbroadening contribution due to an increased Vi. The peak broadening, due to a large Vi,can be reduced by using a sample solvent which is of lower elution strength than the mobile phase, thereby concentrating the analytes on top of the col-umn. This effect was sought for in paper I and III, where samples were dissolved in a low amount of organic modifier. When using such an on-column focusing technique, one should be observant that the reduction in organic solvent content does not cause solubility problems and/or analyte adsorption to sample vials and injector surfaces.45 If the sample volume is limited, detection limits may be improved by increasing N (by lowering H) and by using a column with a small diameter to minimise dilution in the column. In paper II, a large injection volume (200 µl) was used together with miniaturised columns (internal diameter 0.5 mm) to concentrate the analyte and minimise dilution. In paper II, the analyte flumazenil (Fig. 3) was measured at a signal-to-noise ratio (S/N) of approximately 20 at a con-centration of 1 nM or 300 pg/ml (200 fmol injected on column) in a micro-some solution. The same analyte was analysed in human plasma, in paper I,with a S/N of approximately 30 at a plasma concentration of 0.05 nM or 15 pg/ml. WAY-100635 (Fig. 3) was measured at a concentration of 0.01 nM or 4 pg/ml (0.15 fmol injected on column) in microsome solution with a S/N of approximately 30, in paper III. The significantly different sensitivities of

22

the methods were mainly due to the fact that different mass spectrometers were used.

OCH3

N N N

N

OR2

2H2H

2H

2H2H

O I

F

NN

NO R1

O

O

(1) R1 = CH3(2) R1 = C2H3(3) R1 = 11CH3

(4) R2 =

(6) R2 =

(5) R2 =

(7) R2 =

Figure 3. Structures of analytes. (1) flumazenil, (2) flumazenil-d3, (3) [11C]flumazenil, (4) WAY-100635, (5) WAY-d5, (6) MeO-WAY and (7) I-WAY.

3.2 Mass spectrometry Mass spectrometry is a powerful detection technique that enables separation and characterisation of substances according to their mass-to-charge ratio (m/z). By coupling chromatography to MS, a second separation step, or sev-eral in the case of MSn, is thus added to the initial separation in time.

A mass spectrometer analyses gas-phase ions in vacuum, whereas with LC, the compounds, both charged and neutral, are separated in a liquid. When coupling LC to MS the analytes therefore need to be transferred from solution into the gas-phase, and if not already in a charged state, to be ion-ised. The three main ionisation techniques in use today for LC-MS, namely electrospray ionisation46, 47, atmospheric pressure chemical ionisation (APCI)48 and atmospheric pressure photoionisation (APPI)49, generate gas-phase ions under atmospheric pressure. Of these, ESI is the most commonly used. It is characterised as a soft ionisation technique, capable of forming pseudo-molecular ions, typically protonated molecules, [M+H]+, or de-protonated ones, [M-H]-, and is suitable for relatively polar substances. Ad-ditionally, adduct ions such as [M+Na]+, [M+K]+ and [M+NH4]+ can be formed. ESI has been used for analysis of both large biomolecules50 and small molecules51.

The gas-phase ions generated during ionisation can be separated using different analysers. In this work quadrupole analysers52 were used. Other mass analysers include time-of-flight (TOF), ion trap, Fourier transform ion cyclotron resonance (FTICR) and sector instruments.53 These mass analysers

23

display different characteristics that can be exploited for different analytical needs. To increase the performance and flexibility, mass analysers can be coupled and fitted into one instrument, yielding for instance triple quadru-pole and quadrupole-TOF instruments.

Another mass spectrometric technique is accelerator mass spectrometry (AMS).54, 55 It can provide ultra sensitive analysis (10-18 – 10-21 mol) of for instance 14C-labelled compounds, and is being explored in human microdos-ing studies, alongside with PET, for drug development.6, 56

3.2.1 Electrospray ionisation A schematic representation of an electrospray ion source is depicted in

Fig. 4. The LC effluent is pumped through a capillary held at a potential of typically 2- 5 kV, relative a counter plate in the ion source. This potential gradient may be either positive (positive mode) or negative (negative mode), depending on the substances to be analysed. The resulting electric field causes charge separation of ions in the solution, with positive ions moving towards the surface liquid and negative ions migrating back into the liquid, in positive mode. As a result, the liquid protrudes from the capillary tip to-wards the counter plate. As charge of the same polarity accumulates at the surface, the Coulombic forces will at a certain point be greater than the sur-face tension. This will result in a disruption of the liquid surface and drop-lets, of a positive excess charge in positive mode, will be emitted. As the charged droplets move towards the inlet of the mass spectrometer solvent evaporates and the droplets fission into multiple and highly charged droplets. Eventually gas-phase ions are formed for entry into vacuum regions of the mass spectrometer.57

Figure 4. Schematic representation of the electrospray ionisation process, in positive mode. Illustration by Andreas Dahlin.

24

The electrospray source can be viewed as an electrolytic cell with the cap-illary as working electrode and the inlet plate in the mass spectrometer as counter electrode.58 In positive mode the current can be followed from the positive end of the power supply, to the solution/capillary contact where oxidation reactions take place, through the space occupied by the spray, to reduction at the counter plate and back to the negative terminal of the power supply (Fig. 4). Oxidation reactants can be the iron in the stainless steel cap-illary, solvent molecules, electrolytes in the solvent or analytes.

The flow rate used in ESI is in the order of 0.1 – 10 µl/min. However, in order to accommodate for higher flow rates, pneumatically assisted ESI may be used.59 By using a coaxial gas flow, typically of nitrogen gas, nebulisation is aided and flow rates up to the ml/min region can be achieved. In addition, it is possible to heat the electrospray probe, as well as the whole ion source, and drying gas can be applied to facilitate solvent evaporation. All mass spectrometric measurements described in this thesis work were performed using pneumatically assisted ESI at flow rates from 15 to 300 µl/min.

Formation of gas-phase ions in ESI can occur by transfer of ions already existing in solution to the gas-phase. This is the primary route for molecules with acidic or basic functional groups. It should be noted that the pH of the droplets in the spray may be significantly different from the pH in the LC mobile phase.60, 61 Adduct formation is another way of ionisation, giving adducts with for instance sodium and potassium in the positive mode. Ionisa-tion through gas-phase reactions, e.g. gas-phase proton transfer reactions, and electrochemical red-ox reactions are other routes.57

3.2.2 Collision induced dissociation In a triple quadrupole instrument the middle quadrupole focuses and trans-mits almost all ions and may be used as a collision cell for fragmentation reactions. When a single quadrupole instrument is used, fragmentation reac-tions can be induced in the source region. The type of fragmentation occur-ring in quadrupole instruments is referred to as collision induced dissociation (CID). CID is achieved by accelerating the ions from the source for subse-quent collision with gas molecules, either in a vacuum region up-front the analyser, as in in-source fragmentation62, or in the collision cell. Some of the initial ion kinetic energy is converted into internal energy, bringing the ion into an excited state, which may be followed by fragmentation. The degree of fragmentation occurring can be altered by changing parameters such as ion accelerating potential, collision gas pressure and type of collision gas.52

A triple quadrupole instrument offers the possibility of using different scan modes with CID. In papers I and III, multiple reaction monitoring (MRM) scans were employed, for which quadrupole one and three were both locked onto specific m/z ratios. This yields ultimate selectivity and sensitiv-ity and is therefore often used in quantitative analysis. The signal is less than

25

in selected ion monitoring (SIM)VIII, used in paper II, but the noise is con-siderably reduced, yielding generally better signal-to-noise ratios than in SIM. Additional signal-to-noise improvements may be gained in MRM by reducing the resolution and by increasing the dwell time. In papers I and III, product ion scans were additionally performed, where the first quadru-pole is locked to a specific m/z ratio and the third is scanning the daughter ions produced in the collision cell. This was carried out for setting up MRM scans, but also for structure elucidation.

The power of using MS in metabolite structure elucidation was demon-strated in paper III, where the compound WAY-100635 and analogues (Fig. 3) were analysed. In Fig. 5, product ion spectra are given of WAY-100635 and two metabolites of the same m/z ratios. The M+16 values of the metabo-lites suggested hydroxylation reactions. By using fragmental information, proposed metabolite structures could be given, with a hydroxy group at-tached to two different positions in the molecule.

219

219111

111

127

235

[M+H]+

[M+H]+

[M+H]+

OCH3

N N N

N

O

247

121

313

219 OH

127

OCH3

N N N

N

O

231

121

313

219

111

OCH3

N N N

N

O

231

121

329

235

HO

111

Figure 5. MS spectra of product ion scans of WAY-100635 (top) and two of its metabolites (middle and bottom).

VIII SIM can be used in single or triple quadrupole instruments, whereby the quadrupole is locked to one or a limited number of m/z ratios, without using CID.

26

3.3 Liquid chromatography – mass spectrometry When coupling LC to MS, one must sometimes balance the performance of chromatography with that of ionisation efficiency. The mobile phase compo-sition not only affects resolution and efficiency in the LC separation, but also influences the spray performance and the generation and identities of the gas-phase ions that are formed. Non-volatile salts that have been commonly used for buffers in LC, such as phosphate, adversely affect the ESI process due to the build-up of salt on metallic surfaces in the ion source and also due to ionisation suppression (see chapter 3.3.2). As an alternative, volatile addi-tives such as formic or acetic acid, should be used with volatile solvents, like methanol and acetonitrile, for efficient solvent evaporation. In the analysis of amines a low pH (<pKa) is preferable for ESI since this will increase the concentration of the protonated and charged analyte form in solution. This may, however, be detrimental to the chromatography, since interactions be-tween positively charged amine groups and residual silanols can lead to peak tailing. The use of a strong acid, such as trifluoroacetic acid (TFA), for ion-pairing purposes, and thereby a reduction in peak tailing, may lead to a re-duction in signal. This is due to a strong pairing with the analyte, resulting in neutralisation.57

3.3.1 Striving for maximum signal intensity: an example For a high analyte signal in ESI a high organic solvent content is generally preferable. The lower surface tension in a solvent such as methanol com-pared to water reduces the activation energy barrier in transferring ions from the spray droplets into the gas-phase.63 This effect has however been re-ported to be less pronounced in pneumatically assisted ESI, most likely as a result of improved spray stability when using a nebulising gas.28, 64 The op-timised organic solvent content is most likely dependent on instrumental settings57 and type of instrument.

To determine the amount of the tracer flumazenil in human plasma re-quires a method of high sensitivity. In the initial development of such a method (paper I) it was therefore investigated whether a higher organic solvent content in the mobile phase would yield an increased analyte MS signal. In direct infusion experiments, the response was improved by a factor of 2.9 by increasing the acetonitrile content from 30% to 80%. At higher acetonitrile contents (>80%), the signal started to decrease, however; a phe-nomenon which has been observed previously.64

In order to exploit the flumazenil signal increase with a higher acetonitrile content, an LC packing material with a higher retentive power was investi-gated (Allure pentafluorophenyl (PFP) propyl65). However, in-house packed capillary column experiments indicated no significant improvement. This was due to poor efficiency in the PFP column, maybe caused by an ineffec-

27

tive packing procedure. The development of a packing procedure more suit-able for the PFP material may have resulted in a better efficiency, but was not explored within this work.

3.3.2 Ionisation suppression With ESI ionic species compete for the limited excess charge or space on the droplet surface.57 A certain electrolyte concentration is needed for carrying the charge in the electrospray process, but at high concentrations of addi-tives, for instance, the analytes may be out competed, resulting in a de-creased ion signal64. The concentration of the additive therefore has to be balanced between acceptable LC performance and ESI response. Ionisation suppression of the analyte can also be caused by other analytes present at high concentrations. Additionally, a limited linear dynamic range at higher concentrations (usually around 10-5 M)43 can be seen as the result of “self-suppression”.

In the ideal situation, all electrolyte concentrations are known and regu-lated for optimum performance. In the analysis of biological samples this, however, may not be the case. Components from the sample matrix, for ex-ample salts, amines, proteins, fatty acids, triglycerides and phospholipids from plasma, may cause both signal suppression and enhancement, i.e. ma-trix effects.66 Changes in the droplet environment affecting e.g. boiling point and surface tension, caused by nonvolatile solutes, have been proposed as the main cause of ionisation suppression in the analysis of biological sam-ples by ESI-MS.67 Some form of pretreatment is therefore usually carried out in order to remove abundant and interfering components (see chapter 4), prior to further separation by LC-MS. In the case of signal suppression, higher (i.e. poorer) detection limits will follow. Matrix effects may also vary from samples taken from different individuals, thereby causing poor accu-racy and precision in the analytical method. The variability may be less pro-nounced for in vitro studies, since as with, microsome preparations for in-stance, these are generally from the same source, thereby minimising bio-logical differences.68 Matrix effects are typically more pronounced for meth-ods using a non-selective extraction step, such as protein precipitation, in combination with a fast LC gradient. Suppression can also be highly analyte specific. For example, Bonfiglio et al. reported that the highest suppression was detected for the most polar compound analysed.69

Solutions for reducing matrix effects include cleaner samples through a more extensive sample pretreatment procedure in combination with a slower gradient and longer LC run times. Switching ionisation technique, to for instance APCI, may sometimes be helpful.66 By using an isotopically la-belled analogue as an internal standard (IS), matrix effects on precision and accuracy may be alleviated, if the IS behaves in a similar way as the analyte. In paper I, deuterium labelled flumazenil (Fig. 3) was used as an IS, solid

28

phase extraction was used for sample clean-up and a relatively slow gradient was used in combination with a long run time. This resulted in a high preci-sion (inter-assay RSD <7%) and accuracy (bias 4%) of the method. In pa-per II, the matrix effect was estimated to reduce the signal by 15%. By us-ing the same IS as in paper I, a very high precision (inter-assay RSD

1.5%) and accuracy (bias 2%) could be obtained. In paper III, the matrix effect was below 10%, and again precision (inter-assay RSD <7%) and accu-racy (bias 8%) well within accepted limitsIX were obtained.

3.3.3 Quantification Quantification of drugs in biological samples by LC-MS is typically per-formed using a calibration curve prepared in sample matrix to relate instru-ment signal intensity to analyte concentration.70 Additionally, an IS is often added to samples and standards prior to sample preparation to improve pre-cision and accuracy. The IS can be an isotopically labelled analogue or a structurally similar compound. The chosen compound should ideally display nearly identical physical and chemical properties with the analyte, to com-pensate for losses, ionisation effects and random errors.52 When an isotopi-cally labelled analogue is used, there is practically no chromatographic sepa-ration between the IS and analyte. A mass difference of three between ana-lyte and IS is usually sufficient to achieve quadrupole separation between the two species, when low molecular weight compounds composed of elements such as C, N, O and H are analysed. The purity of the IS should preferably be high to avoid IS contribution to the analyte signal. This contribution may be particularly pronounced in the lower concentration range of the calibra-tion curve where the IS to analyte ratio may be relatively high. Consider an isotopically labelled IS with a purity of 99% with a concentration of 30 nM relative an analyte concentration of 1 nM. This would increase the analyte signal with 30%. Given that the same amount of IS is added to samples and calibration standards, this need not cause inaccuracies, but the intercept of the regression line will not be zero.

In papers I and II, [2H3]flumazenil (Fig. 3) was used as an IS. The purity was determined to >99.9% and the ratio between concentrations of IS and lowest calibration standard was 6 and 11.5 in papers I and II, respectively. This ensured minimum IS interference. In paper III, the mass difference between the IS (WAY-d5, Fig. 3) and the analyte WAY-100635 was only one. 13C contributions from the IS caused a signal in the MRM channel of WAY 100635. Since full baseline separation was achieved this did not cause inaccurate measurements.

IX US Food and Drug Administration (FDA) recommend bias and RSD to be less or equal to 15%, except at the lower limit of quantification where it should not exceed 20%, see ref. 70.

29

3.3.4 Validation The reason for performing a validation of a developed method is to ensure that reliable data can be obtained from the method. Parameters such as accu-racy, precision, selectivity, sensitivity, reproducibility and stability may be examined in a validation.70 How thorough the validation needs to be, and which of these parameters should be investigated, depends on the application of the method. Validations are typically performed using spiked blank sam-ple matrix for model samples. These are in most cases accurate models for real samples obtained from e.g. a subject dosed with the drug. One should be aware however, that in certain cases the model fails. This can occur when a metabolite, such as the drug conjugated with for instance glucoronic acid, reverts back into the parent compound, before, during or after sample prepa-ration. Additionally, mass spectrometric in-source degradation and transfor-mation to the parent drug may lead to inaccurate results if the chroma-tographic separation is insufficient. By analysing samples obtained from the real application, thus containing possibly interfering metabolites, such ef-fects may be investigated.71 Knowledge of metabolic pathways can exclude such extra validation schemes. In papers I-III, parameters such as accuracy, precision, reproducibility, linearity, selectivity, ionisation suppression, re-covery, extraction efficiency, sensitivity and carryover were studied for method development and validation purposes.

3.4 Implications of using LC-MS in PET tracer analysis Implementing an LC-MS method in the analysis of tracers removes the time constraints associated with radiodetection of short-lived radionuclides. In the case of plasma analysis for determining the fraction of intact tracer, as dis-cussed in chapter 2 and paper I, LC-MS analysis of the non-labelled fraction can offer several advantages. The sensitivity will thus, theoretically, be con-stant with timeX, which makes it possible to analyse an increased number of samples. This means that it is possible to withdraw additional blood samples from the subject and analyse these without large drifts in sensitivity. The technique should be particularly beneficial in the analysis of plasma from the later time points in a PET study, where low amounts of radioactivity may make conventional radio-analysis particularly troublesome. A logistic asset is the ability to store samples in the freezer for later analysis. In the case of instrument failure of a radioactivity detection method during a study, the result may be a complete loss of plasma data due to the fast decay process. The overall implication of LC-MS analysis in tracer plasma analysis should be better quality of PET quantification data in certain tracer studies. This

X In practice, instrumental drifts and build-up of contaminations in the ion source can cause a change in sensitivity with time.

30

statement should however be tested by a proper and thorough comparison of plasma analysis methods. As discussed in paper I, a comparison could be performed in a large-scale PET study, by investigation of results such as the binding potential of the tracer to the benzodiazepine receptor.

LC-MS should at the current stage be regarded as a complementary tech-nique to radiodetection methods in tracer plasma analysis. Method develop-ment is generally more time consuming for LC-MS analysis, with a greater need for thorough sample pretreatment protocols for optimised preconcentra-tion and clean extracts. Considering the very low tracer concentrations, from 0.1 – 0.2 nM in paper I, it is recommended to use a stable isotopically la-belled analogue as an internal standard to achieve sufficient accuracy and precision. This will entail the synthesis of a new internal standard for each tracer analysed. Additionally, MS instrumentation is considerably more ex-pensive than radiodetection equipment. Method development using radio-detection may on the other hand be expensive, considering the costs in-volved in the production of the tracer. Implementing LC-MS methods should therefore be most beneficial in those cases where very low amounts of radio-activity in plasma constitute a cause for poor precision and accuracy of the plasma data.

It is not only possible to use LC-MS in plasma analysis, but also in pre-clinical tracer development. LC-MS methods were developed for the deter-mination of the metabolic stability of tracers in liver microsome solutions, in papers II-III. The advantages of the technique for in vitro experiments were evident. Instead of relying on tracer production for method development and incubation experiments, cold, non-labelled substance could be used for re-peated experiments. The non-labelled analyte was either obtained from a cold chemistry synthesis or from one synthesis of labelled analyte with all radioactivity dissipated by time. Expensive cyclotron time was therefore minimised and repeated analyses of an increased number of samples could be performed.

As discussed in chapter 2, PET radiodetection cannot differentiate be-tween different chemical species containing the radiolabel. This can poten-tially cause errors, if radiolabelled metabolites coelute with the parent com-pound. Using MS provides m/z information, of molecular and fragment ions, and thus improves identification capabilities and information on metabolisa-tion.

31

4 Sample preparation

Biological samples, such as plasma and microsome preparations, generally need to be processed prior to analysis by LC-MS. The sample preparation step should ideally result in both enrichment and selective extraction of the analytes of interest from interfering sample matrix. A concentration of the analytes will generally lead to improved sensitivity. Removal of interfering sample matrix substances is beneficial both for the separation and ionisation steps. A well executed sample clean-up will minimise the build-up of parti-cles and proteins on the column. This sort of aggregation may cause deterio-ration of separation efficiency and will eventually lead to pressure increases, shifting of retention times and may result in a total clogging of the column. Sample clean-up is also performed to minimise matrix effects in the ionisa-tion process, as discussed previously (chapter 3.3.2).

A number of sample preparation techniques have been developed for liq-uid samples that vary in degree of speed, selectivity, enrichment, robustness and cost. Protein precipitation is a fast and simple method, that may be per-formed with organic solvents, acids, salts and metals as precipitants.72 There is however no enrichment and the selectivity is low, since proteins are sim-ply precipitated and removed by centrifugation. A technique that may pro-vide good selectivity, is liquid-liquid extraction.10 It is however a rather time demanding method, though robotic pipetting stations can be used, and may require large solvent volumes.

A major extraction technique today is solid phase extraction (SPE).73 A wide array of sorbents have been developed, ranging from chemically bonded reversed phase silica, such as C18 silicas, to polymeric, porous graph-itic carbon and ion exchange sorbents.73 These sorbents can be used for off-line configurations, that are easily automated, or in on-line systems. Re-stricted access media (RAM) is another type of SPE material that is used in on-line fashion and permits direct injection of biological samples.74 In paper II, an on-line approach with SPE particles was used for rapid sample extrac-tion. There is thus a great potential for fast, completely automated analysis systems, based on extraction in the SPE mode.10

In paper I, SPE was employed for sample preparation purposes. A ro-botic system was used for pipetting in off-line mode and a C18 cartridge that could handle the direct application of plasma was chosen. Furthermore, a low amount of sorbent (15 mg) was used to enable elution in a low volume (0.5 ml). This was chosen in order to minimise sample losses and to speed

32

up the subsequent evaporation step. In the sample clean-up the sample vol-ume was reduced by a factor of 7.5, yielding a more concentrated sample for a lower detection limit.

4.1 High flow rate extraction For conventional LC separations a laminar flow is used. A laminar, or

steady, flow in a tube is characterised by streamlines running parallel to the walls and by a parabolic flow rate profile (Fig. 6). If the flow rate is suffi-ciently increased, the streamlines will no longer be parallel and eddies will be created. This turbulent flow regime is irregular, with local differences in flow rate and pressure. Also, the velocity profile is more plug shaped. The type of flow encountered in a packed column can be described by the Rey-nolds number, ReXI, where turbulence may be encountered at a Re number of approximately 5.32 In paper II, an elevated flow rate was used for rapid and efficient on-line extraction of microsome samples.

Figure 6. Schematic illustration of laminar (top) and turbulent (bottom) flow in a tube. Arrows indicate streamlines, the length of which represents the magnitude of the flow velocity. Illustration by Andreas Dahlin.

In the 1960’s the effect of turbulence was studied in columns packed with glass beads75 and in open tubular columns76. It was shown that the H/u curve (Fig. 2) flattened off and declined at very high velocities. This was attributed to the onset of turbulence. The beneficial effects of turbulence were postu-lated to arise primarily from an improved effective diffusion in the inter-particle regions, generated by increased convection, and to a lesser extent from the flattened velocity profile, which reduces flow inequalities.76 It was

XI The Reynolds number; Re = udp / , where is the mobile phase density.

33

also argued that turbulent flow does not effect intraparticle mass transfer (the C-term in equation (1)) and that turbulence can only be of practical use when that dispersion contribution is significantly reduced.75 It has been suggested that pellicular packings77, with nonporous cores and a thin film of stationary phase, can provide the fast intraparticle mass transfer needed.32 Alterna-tively, rapid gradients may be used to minimise interactions with the station-ary phase, thereby lowering the capacity factor and the C coefficient in equa-tion (1).

More recently, turbulent flow has been used in columns packed with chromatographically active particles.78 The curve in a reduced parameter H/u plotXII (as in Fig. 2) obtained from such a system stopped rising at a reduced velocity (v) of approximately 5000, and gradually declined until it reached a magnitude of the same order as in the optimum laminar flow, at v = 30000 – 50000.79 In order avoid high back pressures and to enhance the onset of tur-bulence, particles with large diameters (typically 30-50 µm) are used. The formation of a turbulent flow is further aided by the use of rough and irregu-lar particles together with a high interstitial volume fraction exceeding 45% of the total column volume.78 High efficiency separations are not attained, however, as a result of the use of large particles.79 The high speed analysis that may be obtained must therefore be considered as one of the chief merits of turbulent flow chromatography. Another advantage is the ability to inject biological fluids directly onto the column (although a centrifugation step prior is recommended). This can be performed due to increased convection in the turbulent flow regime, but also due to the use of large particles and screens of large porosities (10-20 µm). These factors facilitate the passage of large proteins without blockage of the column.80, 81

The use of high flow rate chromatography with large particles has been used in high-throughput contexts.10, 74, 80, 82, 83 Short columns are used in con-junction with fast gradients, generating basically no chromatographic separa-tion. Instead, separation has been accomplished by the coupled MS instru-ment. Indeed, one can argue that this is not true chromatography.32 In order to achieve higher separation efficiency, the high flow column can be coupled to an analytical column, by a column-switching set-up. In this configuration, the high flow column works as an extraction column, with elution often per-formed in a back-flush mode.10, 74, 81, 84-86 Significantly reduced back-pressure buildup and carryover have been reported, in such a set-up, when comparing turbulent and laminar flow.81

In paper II, a method was developed for the analysis of microsome samples, using a high flow extraction column coupled to an analytical column in the capillary scale (internal diameter 0.5 mm). Since extreme high-throughput was not an objective, but rather the attainment of high-quality data, a col-umn-switching configuration was chosen for the study. Apart from improved

XII Reduced plate height, h = H/dp, vs reduced velocity, v = udp/DM.

34

chromatographic efficiency, this set-up also permitted the injection of large sample volumes (200 µl). As seen from equation (2) this approach can result in higher sensitivity, compared to injections of low volumes. A large injec-tion volume may, however, result in very long sample loading times. The loading time in paper II, using the analytical column flow rate (15 µl/min), would have been approximately 13 min. Using a high flow rate (1 ml/min) instead yielded a very fast extraction step of 0.75 min. The extraction was not only extremely rapid, but also very efficient (>99%).

R2 = 0,9997

0

0,4

0,8

1,2

1,6

2

0 2 4 6P (MPa)

F (m

l/min

)

Figure 7. Flow rate vs. pressure drop. The correlation coefficient, R2, was obtained from linear regression of flow rates of 0.1 to 1 ml/min. Column: 40 x 0.5 mm, Oasis HLB, dp = 30 µm. Mobile phase: 5 mM formic acid in water.

The transition from a laminar to a turbulent flow in a packed column, as the flow rate is increased, does not occur instantaneously in all regions of the column. Rather, turbulence starts in a few regions containing large void spaces and gradually spreads throughout the column as the velocity is further increased.35 The question arouse whether some degree of turbulence was induced under the conditions described in paper II. The Re number was calculatedXIII as 4.5, which is in the region of turbulence in a packed col-umn32. The occurrence of turbulence may also be investigated using Poiseuille’s formulaXIV, 87, 88 At the onset of turbulence a linear dependence of flow on the pressure drop is no longer present. In Fig. 7 a graph with flow rate versus pressure drop is plotted, for data obtained from an extraction column used in paper II. In the region of up to 1 ml/min, which was the

XIII Using the following values; u = 15 cm/s, dp = 30 µm, = 1 g/ml and = 1 mPas. XIV Poiseuille’s formula: F = Pr4/8 L, where F is volumetric flow through an open tube and r is the radius of the tube. The formula is valid for laminar flow.

35

velocity used in the study, linearity is maintained, as indicated by a correla-tion coefficient of > 0.999. In the higher flow rate regions linearity is no longer present, indicating the onset of turbulence. From the estimation of Re number and the Poiseuille’s formula experiments it was not possible to ver-ify that turbulence was induced in the extraction column. Some form of tur-bulence may however exist in parts of the column. The form of extraction used in paper II was therefore termed as high flow rate extraction.

The ionisation suppression using in-house packed Oasis HLB (Hydro-philic Lipophilic Balance) extraction columns was estimated to 15% (paperII). Using SepPak C18 columns, no suppression was seen at all. This likely followed from the very different packing material compositions. SepPak C18is a silica based material with linked C18 chains, whereas Oasis HLB is a polymer composed of hydrophilic N-vinylpyrrolidone and lipophilic divi-nylbenzene monomers. The extended functionalities of the latter material, as used in the method described in paper II, may very well result in a higher degree of extracted sample matrix. Additionally, the SepPak material was composed of larger particles (50 – 105 µm), compared to Oasis HLB (30 µm). The estimated Re numbers were 8 and 16 for 50 and 105 µm particles, respectively. Visual microscopic inspection revealed a higher degree of ir-regularity in the SepPak material. These factors should contribute to a higher degree of turbulence and thus possibly also a higher degree of matrix re-moval. In spite of these positive features of the SepPak material, Oasis HLB was used in the developed method. This choice was made, firstly, since a generic extraction protocol was desired, and hence a material capable of extracting a great variety of compounds was required. From a theoretical perspective, considering the hydrophilic and lipophilic properties of Oasis HLB, this was the natural choice. Secondly, from a practical standpoint, as difficulties were encountered in packing the SepPak C18 material, with quite frequent clogging of the packing system, Oasis HLB was superior. The cause of the clogging was not established.

4.2 A fast and simple sample preparation step Simplicity in an analytical method is always preferable. In paper III, micro-some samples could be injected directly, after sample centrifugation, onto an analytical column, without the use of an extraction column. This resulted in shorter run-times and above all a set-up that was much less complex than the one described in paper II. This more straightforward approach was possible due to the fact that a microsome concentration that was 20 times less was used. Also, a different mass spectrometer was available with MS/MS capa-bilities and significantly improved sensitivity. This made it feasible to use a smaller injection volume (15 µl) and thus removed the necessity of an ex-traction column required for very large injection volumes.

36

5 Metabolism

When introducing a tracer in the human body the compound might be me-tabolised. It is desirable to use tracers that are not extensively metabolised. Metabolites containing the radiolabel may complicate PET imaging since the technique does not allow differentiation between parent compound and ra-dioactive metabolites, as discussed in chapter 2.3. This becomes particularly troublesome if the radioactive metabolites enter the target area, e.g. cross the blood-brain-barrier where they may contribute to both specific and non-specific receptor binding. Additionally, metabolism will reduce the quantity of the tracer (as illustrated in Fig. 1), thus decreasing the radioactive PET signal, causing potential loss of accuracy and precision of the data obtained.

In the development of new PET tracers, it is therefore important to inves-tigate metabolism at an early stage. The first step in elucidating the meta-bolic stability of the compound is often undertaken using research animals, before proceeding to human studies. In order to minimise in vivo studies and the sacrifice of animals, in vitro methods may initially be used. Although these do not entirely reflect the complexity of the in vivo situation, valuable information can still be obtained.

The main organ involved in drug metabolism is the liver, although other organs take part such as kidney, lung and brain, as well as blood.89 The liver can be prepared in many ways for in vitro metabolism studies, using for in-stance whole liver, liver slices, isolated liver cells, liver homogenate and isolated liver cell sub-fractions. These preparations all have advantages and disadvantages regarding complexity and proximity to the in vivo situation. Methods using liver microsomes are probably the most common. Advan-tages include ease of preparation, low cost, good reproducibility and the ability of long-term storage. The in vivo relevance may however be low since the whole cell content is not present.89-91

Liver microsomesXV, 92 contain enzymes involved in metabolism, notably the cytochrome P450 family. During the metabolic process the substrate is generally transformed into a more polar compound with increased water solubility and therefore into a more easily excretable form.89

XV Liver microsomes are prepared by initial homogenisation of liver, causing fragmentation of the endoplasmic reticulum and subsequent reclosure into vesicles (microsomes). The mi-crosomes are thereafter isolated by differential centrifugation.

37

Metabolic reactions are commonly divided into two stages or phases. Phase I reactions introduce or uncover a chemical functional group to the drug, thereby increasing the polarity. More importantly, the phase I func-tionalising of the drug can transform it to a suitable substrate for phase II conjugations, further increasing the water solubility and excretability. Phase I reactions typically include oxidation, reduction, hydrolysis and hydration. Phase II, or conjugation, reactions include glucoronidation, sulfation, methy-lation, acetylation, amino acid conjugation and glutathione conjugation.89

In paper II-III, analytical methods were developed for the determination of microsomal metabolic stability of PET tracers by LC-MS. In paper II, the tracer flumazenil (Fig. 3) was used as a model compound to demonstrate the applicability of the analytical method. Flumazenil is a benzodiazepine recep-tor antagonist that has been used for PET imaging in the human brain.93-96 Its main metabolite in human, the free carboxylic acid, is formed by phase I ester hydrolysis and its other metabolic reactions include demethylation and glucoronidation.97-99 Using the method developed in paper II, the metabolic stability of flumazenil in microsome incubation could be established (Fig. 8). The metabolic rate was relatively high with 52% of the substrate remaining intact after 20 min of incubation.

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60

Incubation time (min)

Inta

ct s

ubst

rate

(%)

Figure 8. Metabolic stability of flumazenil in a rat liver microsome solution (proteinconcentration 1 mg/ml).

38

5.1 WAY-100635 and analogues In paper III, the PET tracer WAY-100635 and two analogues (Fig. 3) were analysed. WAY-100635 is a selective antagonist for the 5-hydroxytryptamine-1A (serotonin, subtype 1A, 5-HT1A) receptor.100, 101 The 5-HT1A receptor is reported to be involved in psychiatric disorders such as anxiety, depression and schizophrenia. Research has therefore been con-ducted to image the receptor in the human brain with [11C ]WAY-100635.102-

104

The importance of the position of the radiolabel can clearly be illustrated in the stages of development of [11C ]WAY-100635. The compound was initially labelled with 11C in the methoxy position (Fig. 9), but a radio-labelled metabolite, [O-methyl-11C ]WAY-100634, was taken up by the brain, causing difficulties in PET quantification.105 Labelling in the carbonyl position, giving [carbonyl-11C]WAY-100635, on the other hand, yielded [11C]cyclohexanecarboxylic acid as the major radiolabelled metabolite, with a relatively low brain uptake.106 The result was a ten fold increase in PET signal contrast compared with [O-methyl-11C ]WAY-100635.106, 107

OCH3

N N N

N

O

OCH3

N N N

N

O

OCH3

N N NH

N

OHO

[O-methyl-11C]WAY-100635 [carbonyl-11C]WAY-100635

[O-methyl-11C]WAY-100634 [11C]cyclohexanecarboxylic acid

Figure 9. Structures of [O-methyl-11C ]WAY-100635, [carbonyl-11C]WAY-100635 and radiolabelled metabolites. The star represents position of 11C-label.

The metabolism of [carbonyl-11C]WAY-100635 is however rapid, with less than 10% of the radioactivity in plasma representing intact tracer, 10 min after tracer injection.104, 108 A number of analogues have therefore been developed in the hope of finding a tracer with increased metabolic stability but retained selectivity for the 5-HT1A receptor.109-113

In our laboratory a strategy has been adopted with the aim to develop new improved WAY-100635 analogues in a short time. The idea is to avoid in-vesting too much time in the early stages of development, such as optimisa-tion of synthetic methods for each specific analogue, quantification and

39

characterisation of each tracer, and costly in vivo PET imaging animal ex-periments. Instead a large number of analogues are synthesised, using one pharmacophore (WAY-100634, Fig. 9) and different structurally similar substrates, with the use of a common labelling method.114 The labelled ana-logues are subsequently screened with autoradiography and biodistribution methods for finding tracers with good receptor binding properties. Top can-didates from the binding and distribution screening are assessed for meta-bolic stability using liver microsomes. Analogues passing the metabolic test are used for subsequent in vivo animal and human PET studies. In the later stages, optimisation of labelling methods for each of the remaining ana-logues can be performed, if needed, as well as tracer characterisation by mass spectrometry and 13C-nuclear magnetic resonance (NMR) techniques, in addition to more thorough metabolite studies, such as identification with authentic metabolite standards.

For a fast and efficient development of WAY-100635 analogues, as out-lined above, the substrate for the metabolic stability assay should be col-lected from the labelling batch and not be synthesised separately using a cold chemistry method. This puts high demands on the sensitivity of the analyti-cal method since the amount of tracer in the labelling batch is limited (in the lower µg range). In paper III, an LC-MS/MS method was developed for the determination of metabolic stability of potential PET tracers, with sufficient sensitivity to comply with this demand, as exemplified in the analysis of WAY-100635 and analogues. The low amount of substrate used in incuba-tion experiments (5 nM) guaranteed that repeated experiments could be car-ried out, calibration curves of each specific analogue set up and, if necessary, validations could be performed. In addition, it is advantageous to use con-centrations in the same vicinity as in the intended imaging application, since the metabolism may display a concentration dependence.

Analogues synthesised with a generic and rapid 11C-labelling method in low nmol or µg quantities may however contain more impurities than those obtained from a compound dedicated cold chemistry synthesis with product in mg quantities. LC-MS analysis performed in paper III indicated that this was the case for WAY-100635, obtained in a quantity of 0.3 µg. Metabolic stability tests were performed to investigate whether different biological responses would be given dependent on the substrate source. The results, for the tested compounds, indicated that the measured metabolic rate was the same, irrespective of source. This suggested that a fast 11C-labelling method development could be used, without performing extensive purification op-timisations, for in vitro metabolic screening. The overall result of the work performed in paper III, was an efficient screening method of potential PET tracers.

Initial experiments, using the same protein concentration as in paper II (1 mg/ml), showed that the metabolic rate of WAY-100635 was rapid when incubated in rat liver microsome solutions, with less than 5% of the com-

40

pound intact after 5 min. For subsequent stability tests, the protein concen-tration was reduced (0.05 mg/ml) to obtain an experimental set-up with a more measurable metabolic rate. In Fig. 10, the metabolic stability of WAY-100635, MeO-WAY and I-WAY (Fig. 3) is plotted. Using the method de-veloped in paper III, it was thus possible to show that MeO-WAY was sig-nificantly more resistant to metabolic transformation than WAY-100635. After 50 min of incubation the amount of intact MeO-WAY was on average 2 times that of WAY-100635. For I-WAY, on the other hand, a concentra-tion of 50% of WAY-100635 was measured, after 50 min of incubation.

0

20

40

60

80

100

0 10 20 30 40 50

Incubation time (min)

Inta

ct s

ubst

rate

(%)

Figure 10. Rat liver microsomal metabolic stability of MeO-WAY ( ), WAY-100635 ( ) and I-WAY ( ), at a protein concentration of 0.05 mg/ml. Error bars represent one standard deviation, with n = 5 (WAY-100635) and n = 3 (MeO-WAY). For I-WAY, the error bars show max/min values, with n = 2.

Metabolites of WAY-100635 and MeO-WAY were also analysed by LC-MS for identification purposes, as illustrated in Fig. 5. Fragmentation data suggested hydroxylated, demethylated and dearylated metabolites of both WAY-100635 and MeO-WAY. Metabolites formed by amide hydrolysis, the main pathway in humans, were not detected. This was in accordance with previous rat metabolic studies on WAY-100635115 and fluoro-analogues of WAY-100635116, 117. To further increase the strength of the metabolite identi-ties, analysis of synthesised metabolite standards can be carried out. Since the work in paper II was performed in an early stage of tracer development this was not included.

41

6 Radiotracer chemistry applied

Radiotracer chemistry may also be applied to other fields than those con-cerned with biological systems. Applied to analytical chemistry contexts it can offer truly unique information. High sensitivity detection may be achieved for a tracer embedded in non-transparent materials or dissolved in solutions containing, what are for many other techniques, highly interfering matrices.

6.1 Detection of annihilation radiation The highly energetic annihilation radiation can be used to monitor an analyte in a chemical analysis process, since the radiation readily penetrates liquids, chromatography packing materials and steel surfaces and may be subse-quently detected with a gamma detector. A simple way of detection can be performed using a hand-held radiometer for locating the analyte in a chemi-cal analysis system where severe adsorption occurs. More precise and quan-titative measurements can be performed using a scintillation well counter118.Using such a detection scheme can be very useful in the development and validation of an analytical method. Recovery of L-[methyl-11C]methionine has been determined through SPE sample preparation and LC analysis using such an approach.8 Extraction efficiencies of 11C-labelled metabolites using supercritical ammonia have been investigated in a similar manner.9

In paper I, [11C]flumazenil (Fig. 3) in human plasma was traced through sample preparation using SPE. A recovery of 78% was established for 0.4 nM samples, with analyte losses in the washing liquid and onto the column packing material. Additional losses most likely occurred by adsorption to the tubing of the robotic sample handling system. When analysing samples of low concentrations (sub nM) small absolute analyte losses can become quite large relative losses. In this case, a loss of 22% represented 20 pg in analyte mass, which is a very low amount of substance. It is also possible to estimate the recovery in an SPE step without the use of a tracer. Samples can be ex-tracted and analysed by e.g. LC-MS and compared to blank samples proc-essed by SPE and spiked just prior to the LC-MS step. Using such a scheme, however, does not readily permit monitoring of the analyte to, for instance, waste liquids and chromatographic particles.

42

To determine the absolute recovery in an on-line sample extraction sys-tem can be complicated without the aid of a radiotracer. Zeng et al. estimated the relative extraction efficiency in a high flow rate column-switching LC-MS system by comparison with samples analysed by a conventional LC-MS method.85 In paper II, the extraction efficiency in an on-line high flow ex-traction LC-MS system could conveniently be determined using [11C]flumazenil as a tracer. Furthermore, using the tracer permitted establish-ing the total recovery by measuring the amount of radioactivity injected and the amount in liquids that were withdrawn from system end-points. It was also possible to investigate possible column overload at increased analyte concentrations.

Another example of how useful radiotracer chemistry can be in method development is illustrated by initial experiments performed in the context of paper II. By radioactivity measurements it could be shown that only small amounts of tracer was lost to the pellet, produced in the centrifugation step (3.1±0.1%). Additionally, determination of carryover in paper II was per-formed by detection of the annihilation radiation of [11C]flumazenil, since the MS sensitivity was not sufficient for such a determination.

These cases from paper I and II show the unique possibilities that detec-tion of the annihilation radiation of a tracer can bring to analytical method development and validation.

6.2 Radionuclide imaging The other constituent of positron decay, the +-particle, may also be used for detection of the tracer. Using this mode, quantitative images can be gener-ated without the use of an expensive ring detector and advanced reconstruc-tion algorithms, as in PET proper. The concept of radionuclide imaging, as used in this thesis, involves exposure of a storage phosphor plate to a short-lived positron emitting radionuclide. The latent image formed in the plate is extracted and formed into a quantitative and digital image.119 The perform-ance of storage phosphor plates in the imaging of 11C tracers has previously been investigated.120 Good linearity, sensitivity and adequate resolution and uniformity of the plates for in vitro characterisation of receptor binding pa-rameters were reported. Additionally, capillary LC columns have previously been imaged in a feasibility study using 11C tracer technology.5

In paper IV radionuclide imaging was proposed as a tool for studying molecular interactions in miniaturised chemical analysis systems. One of the advantages of the method lies in the ability to detect and quantify the tracer as it is still within the system. The information obtained using end-point detection, such as MS, will always be a reflection of the sum of all events the eluting peak has “experienced”. Using radionuclide imaging, on the other

43

hand, allows focusing on different parts of the system and thus a separation of the effects of these.

Another advantage is the high sensitivity that can be obtained. This is a direct consequence of the high specific radioactivity that may be obtained by labelling with short-lived positron emitting radionuclides (Table 1). It must however be stressed that the sensitivity of a method using short-lived ra-dionuclides is highly time dependent. The sensitivity obtained using 11C, for instance, is reduced by a factor of 8 after 1 hour.

Figure 11. Images of tracer distribution within microstructures. Top left image: extraction column injected with 200 µl of a microsome solution containing 100 nM [11C]flumazenil. The entry point of the column is to the left in the picture. Right image: distribution of 68Ga-DOTA-angiotensin II in a plastic CD channel system after incubation during 5 min at a concentration of 300 µM and a washing step with water. Bottom left: distribution of 68Ga-DOTA-angiotensin II in poly(dimethyl-siloxane) (PDMS) channels after incubation (300 µM) and washing.

In paper IV three types of microstructures were imaged (Fig. 11). The sensitivity varied with type of radionuclide and tracer used, as well as with the microstructure (Table 2). The results illustrate that very low amounts of tracer can be detected and also demonstrate the importance of high mass sensitivity in the analysis of sub-microliter volumes. In the case of micro-channels in the CD format, the mass sensitivity permitted detection of ap-proximately 200 times lower amounts compared to imaging of extraction columns. Despite this, the concentration sensitivity of extraction column imaging made it possible to detect concentrations that were approximately nine times lower in the extraction column. The reason for the relatively

44

lower concentration sensitivity of CD imaging was a result of sample vol-umes that were 2000 times smaller. When analysing nanoliter volumes, such as 100 nl in the CD microchannels, it is clearly important with a high mass sensitivity. The results also show that the sensitivity is decreased when im-aging structures embedded relatively deeply in a supporting material. The sensitivity (in radioactivity or mass) was thus decreased by a factor of ap-proximately 10 when comparing imaging of CD and poly(dimethylsiloxane) (PDMS) channels. The latter were moulded approximately 1 mm from the structure surface, compared to 40 µm in the former case. On the other hand, it was demonstrated that such structures could indeed be imaged with ade-quate sensitivity.

Table 2. Radionuclide imaging of three microstructures (paper IV). Detectable ra-dioactivity and molar amounts are given together with resolution

Microstructure Radio-nuclide

Detectableradioactivity (Bq)

Detectableamounta

(amol)

Detectable concentrationa

(pM)

FWHMb

(mm)

Extraction columnc 11C 4 500 2.4 1 CD channelsd 68Ga 2 2.3 22 0.9 PDMS channelse 68Ga 20 23 110 2.7 aCalculated from SRA values of 8 and 880 GBq/µmol for 11C and 68Ga tracers, respectively. bThe full width of the peak at half its maximum height (FWHM). cFused silica column with an internal diameter of 0.5 mm and outer diameter of 0.75 mm. dMoulded in plastic with dimensions 200 x 40 µm (width x depth) and with a lid 40 µm thick. eMoulded in PDMS with a diameter of 0.18 mm, embedded approximately 1 mm in the mate-rial.

Due to the relatively high penetrating power of +-particles, there is no need for the imaged material to be transparent, as is the case for optical tech-niques such as fluorescence detection. It is thus possible to image a tracer within for instance chromatography particles and polyimide coated fused silica. The resolution, on the other hand, is not particularly high (Table 2). This is due to the relatively long mean ranges of +-particles. Positrons from 11C decay travel on average 0.75 mm in Plexiglass, whereas the correspond-ing value for 68Ga is 2.0 mm.19 An optimum in resolution, and minimisation of the dispersion of the +-particles, can be achieved by keeping the distance between the radiating source and the detecting plate as low as possible. In other words, a thin object will yield a better resolution. It is also possible to improve the resolution by choosing the radionuclide with the lowest positron emission energy.

45

6.2.1 Extraction column imaging Paper IV describes imaging, using [11C]flumazenil, of on-line extraction columns from the LC-MS system developed in paper II. Using radionuclide imaging made it feasible to delineate and to obtain a relative quantification of tracer distribution within the column (Fig. 11).

Authentic microsome samples could be used in these imaging experi-ments, where peak heights were used as a measurement of the extraction performance. Peak heights could be calculated by transforming images into graphs (Fig. 12). The effect of analyte concentration on the extraction per-formance was thus studied. Increasing the concentration from 30 to 3000 nM gave no significant difference in peak height (Fig. 12) and it was concluded that the extraction was stable in the studied range. These results agreed with those obtained in paper II, where overloading effects at increased analyte concentrations were studied using annihilation radiation detection. Addition-ally, the effect of diluting the sample, in order to reduce possibly detrimental sample matrix effects on the extraction, was investigated. No positive effects could, however, be discerned by a ten-fold dilution of the sample.

0

500

1000

1500

2000

2500

3000

3500

0 200 400 600 800 1000pixel

PI s

igna

l

Figure 12. Imaging of extraction columns displayed in graph form, with phosphor imager (PI) signal on the y-axis and pixel number (pixel size 50 µm) on the x-axis. The different graphs represent extraction of 11C[flumazenil] in a microsome solution at concentrations of 30, 300 and 3000 nM. Injection was performed from left in the graph, where each line represents an average of two samples. Time was used to dissipate the amount of radioactivity of 300 and 3000 nM samples to reach ap-proximately equal amounts of the 30 nM samples. Additionally, pixel intensities were normalised to the column with the lowest pixel sum intensity, as described in paper IV.

46

6.2.2 Imaging of peptide-surface interactions within microchannels Peptide adsorption to surfaces can be studied by labelling techniques with radionuclides such as 125I121, 122 and 3H123, or fluorescent molecules124. Other techniques include colorimetry121, 125, LC121, 125, MALDI-MS126 and ellip-sometry127. These techniques have typically been applied to flat macro sur-faces and not to measurements of processes occurring within microfluidic channels. Additionally, the surfaces generally need to meet certain require-ments, such as transparency for ellipsometry. Protein adsorption to micro-channels has been studied with 125I labelling128 and fluorescence129 tech-niques and surface plasmon resonance (SPR) microscopy130. In the case of the two latter techniques, certain surface characteristics have to be consid-ered. SPR requires a gold surface, while background signals and lack of sur-face clarity may complicate fluorescence detection. Measurements of ra-diation from a 125I labelled protein using a scintillation counter do not con-tain any spatial information. Radionuclide imaging, on the other hand, pro-vides the possibility to obtain quantitative images of retained peptide tracer within selected microchannels without the need of surface or structure modi-fications.

Radionuclide and fluorescence labelling techniques are generally more sensitive than other techniques, but include additional steps for the labelling procedure. Additionally, the labelling may change the chemical properties of the molecule and hence give a substance with different characteristics than the non-labelled analogue.121 From this perspective, labelling of an organic molecule is best performed using isotopes of native elements such as 3H and 11C. However, labelling a peptide with 11C requires a very fast synthesis, due to the short half-life, and can thus be inconvenient. A different approach that can be used is to attach a chelator, such as DOTAXVI, to the studied peptide, which enables labelling with 68Ga (T1/2 = 68 min).131 By this procedure, the conjugated and purified peptide can be stored in aliquots for a fast labelling step prior to the imaging experiment. Very high specific radioactivity values can be obtained131 and furthermore, the half-life permits evaporation after labelling and solvation in a buffer of choice.

In papers IV and V, DOTA was conjugated to two peptides (angiotensin II and vasoactive intestinal peptide) and labelled with 68Ga. The imaging strategy was applied to measure peptide adsorption in microfluidic channels in a plastic CD (paper V) and also briefly exemplified in paper IV in PDMS microchannels, further explored by Bergström et al.132. Changes in adsorption in the CD channels were clearly measured with the method when varying parameters that typically affect peptide adsorption. These parameters were peptide-tracer concentration, pH, ionic strength and surface modifica-

XVI 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid, DOTA.

47

tions. Additionally, a quantification method was set up for measuring the amount of retained tracer. Considering that the chelator is fairly large (DOTA, Mw = 404) it is not unlikely that the adsorption characteristics of the peptide are somewhat changed upon conjugation. This can be investigated in a competition experiment between the non-labelled peptide and the tracer. Additionally, determinations of pI values and hydrophobicity can be useful. These experiments were not performed within the study, but the pH and surface modification experiments performed in paper V suggested that the pI values were not changed dramatically. Furthermore, preliminary molecu-lar modelling of angiotensin II and DOTA indicated that the attachment of DOTA to the peptide did not alter the configuration of the peptide to a large extent.XVII

XVII Rasmussen Torben, personal communication.

48

7 Concluding remarks and future outlook

LC-MS methods were developed for improving the quality of data used in a PET context. The tracer flumazenil was determined in human plasma, with high precision and accuracy, in a multiple scan PET study. The developed method provides the opportunity of analysis of an increased number of sam-ples, compared with radio-detection methods, together with the possibility of sample storage for later analysis. The results clearly suggest that LC-MS analysis can serve as an advantageous complement to conventional methods. As a next step, a large scale [11C]flumazenil PET study should preferably be performed, with plasma analysis carried out using the developed LC-MS method and a conventional LC radio-detection method. This would permit a comparison of the accuracy and precision of the ensuing PET quantification data, such as the binding potential, which is usually calculated using plasma data.

The advantages of using LC-MS in in vitro tracer analyses were also clearly demonstrated. Metabolic stability tests could be carried out using the substrate obtained from a single labelling experiment, thereby considerably reducing expensive labelling and cyclotron time. The result was an efficient screening method of potential PET tracers. Additionally, an on-line high flow extraction technique was developed for rapid and efficient sample ex-traction. Molecular weight information also strengthened identification of the tracer, thus avoiding potential errors caused by coeluting radiolabelled metabolites. Finally, mass spectrometric detection proved to be powerful in metabolic profiling of WAY-100635 and an analogue less susceptible to microsomal degradation (MeO-WAY). Authentic metabolite standards should, however, be used for a more thorough identification. As a next step, MeO-WAY will be used in in vivo PET experiments, evaluating receptor binding properties, biodistribution and metabolism. Potentially, the analogue can serve as a more potent tracer than WAY-100635 and thus play an impor-tant role in the elucidation and treatment of psychiatric disorders such as anxiety, schizophrenia and depression.

Radiotracer chemistry was shown to be very valuable in analytical tech-nique and method development. Exploiting the highly energetic annihilation radiation for detection allows tracing molecules located in for instance waste liquids and chromatography particles. The radionuclide imaging approach is well suited for the study of molecular interactions in miniaturised chemical analysis systems. A great advantage of the developed imaging method is the

49

ability to study events directly within microchannels, without removal of the studied analytes or by the use of model surfaces. The effect of attaching a label to the native molecule should preferably be evaluated as a next step. The radionuclide 68Ga is of particular interest, since it can be obtained from a generator, and thus be used by research groups lacking expensive accelerator equipment. Radionuclide imaging may therefore be used in the future by other groups in the development and evaluation of for instance miniaturised biotechnology platforms.

50

8 Acknowledgements

Tack alla som hjälpt mig och som jag fått samarbeta med under den här ti-den! Särskilt vill jag tacka:

Bengt Långström och Karin Markides för att jag fick ta del av ert spännande samarbete mellan olika kemidiscipliner. För att ni haft en sådan tillit till min kapacitet och därmed den stora frihet jag fått under de här åren.

Lieuwe Appel för ett enormt tålamod, noggrannhet och engagemang. Jag lärde mig mycket under alla de där ROB-diskussionerna!

Susanne Wallenborg för ett spännande samarbete i ett projekt som verkli-gen växte under åren. Tack för ditt stöd när det verkligen behövdes!

Irina Velikyan för ett långt och lyckat samarbete kring peptid-imaging. Skönt att vi var två uppe i labbet som jobbade långa timmar!

Robert Moulder for all the help when I was a rookie in the lab and for al-ways taking time to discuss problems when I was stuck.

Oleksiy Itsenko för ett bra och lyckat samarbete i WAY-projektet.

Niklas Tyrefors för alla små specialtrick och hjälp med LC-MS analys.

Jenny Ljung, Oskar Berglund, och Majda Djodjic, för er stora insats i imaging-projektet, som utfördes med stor drivkraft, även om det inte var så lätt alla gånger.

Sara Bergström och Niklas Edenwall för ett roligt samarbete med PDMS-imaging.

Rolf Danielsson för stort tålamod med att öppna upp och förklara sin stati-stikvärld.

Barbro Nelson för att du tagit hand om alla oss doktorander!

51

För värdefulla kommentarer till avhandlingen: Pernilla Frändberg, Susan-ne Wallenborg, Sara Bergström, Robert Moulder och Per Sjöberg.

Torben Rasmussen för tiden som du gav till peptidmodellering.

Andreas Dahlin för hjälp med illustrationer.

Ardeshir Amirkhani för givande diskussioner kring LC-MS.

Alla på Analysen, i grupp BLå och på Uppsala Imanet!

Stort tack till Knut och Alice Wallenbergs fond, Liljewalchs resestipendie-fond, Kungliga vetenskapsakademien, Svenska kemistsamfundet och Svenska masspektrometrisällskapet, för ekonomiskt stöd som gjort att jag kunnat delta i vetenskapliga konferenser.

Mina vänner – utan er hade jag inte orkat med all kemi...

Maria, Gösta och David, och farmor.

Sofia – you’re number one!

52

9 Summary in Swedish

Vätskekromatografi – masspektrometrianalys av kortlivade spårsubstanser i biologiska matriser

Utforskande av radiospårkemi som ett analytiskt verktyg

Inom positronemissionstomografi (PET) används radioaktivt märkta sub-stanser, eller spårmolekyler, för att visualisera och mäta olika biologiska skeenden hos människan. Neurologiska sjukdomar som Alzheimers och Parkinsons sjukdom kan studeras med PET, men även cancertumörer kan spåras. PET-tekniken används också inom läkemedelsutveckling för att t.ex. fastställa huruvida det potentiella läkemedlet når det önskade målområdet i kroppen.

Mycket låga mängder spårsubstans kan uppmätas tack vare effektiva märkningsmetoder och korta halveringstider hos de använda radionuklider-na. De positronemitterande nuklider som vanligen användas inom PET är 15O, 13N, 11C, 68Ga och 18F, med halveringstider på 2, 10, 20, 68 respektive 110 minuter. Dessa korta tider gör dock att mätningar endast kan utföras inom en högst begränsad tidsrymd: en PET-scan med 11C utförs vanligen under 60 - 90 minuter. Då radioaktivitet mäts i plasma taget från en patient, eller försöksperson, i en PET-undersökning, kan det i vissa fall röra sig om ett ytterst få antal sönderfall som mätningen baseras på. Detta kan ge data med låg precision och riktighet. I dessa situationer skulle en kompletterande teknik som inte omfattas av sådana tidsrestriktioner vara värdefull.

Vätskekromatografi kopplat till masspektrometri (LC-MS) är en teknik med vilken de molekyler som innehåller stabila isotoper i PET-spårsubstansen kan uppmätas. Analys med LC-MS kan med andra ord utfö-ras utan att beakta spårämnets halveringstid. Vidare är det en teknik som kan ge hög känslighet vid analys av biologiska prover och med vilken ämnets molekylvikt, samt viss strukturinformation, kan erhållas.

I denna avhandling har LC-MS-metoder utvecklats för analys av PET-spårsubstanser i biologiska provmatriser. Spårsubstansen flumazenil (vilken bl.a. används vid epilepsiundersökningar) kunde således bestämmas med hög precision och riktighet i humanplasma. Applicerbarheten hos metoden demonstrerades genom analys av plasma från en PET-undersökning, där flumazenil förekom i låga koncentrationer (0.07 – 0.21 nM). Genom att an-vända metoden vid framtida PET-undersökningar med flumazenil kan even-

53

tuellt plasmadata fås med större precision och riktighet än vid analys med konventionell radioaktivitetsmätning. Detta kan resultera i att kvaliteten hos PET-data som t.ex. receptorbindningspotential förbättras.

En PET-spårsubstans bör vara relativt biologiskt stabil eftersom en hög grad av metabolism kan komplicera tolkning av PET-data. Vid utveckling av nya spårsubstanser är det därför viktigt att vid ett tidigt skede undersöka metabolismen. I detta arbete utvecklades LC-MS-metoder för att bestämma grad av metabolism hos potentiella spårsubstanser i provrörsförsök baserade på inkubering i levermikrosomer. Med de nya metoderna kunde upprepade metabolismförsök utföras under olika dagar utan att upprepade märknings-synteser behövde genomföras, vilket inte är möjligt med radioanalys av kort-livade nuklider. En metod baserades på snabb direktkopplad hög-flödes-extraktion, vilket minimerade provupparbetning, som ofta kräver en hög grad av manuellt arbete. Vid analys av spårsubstansen WAY-100635 och två analoger till denna, uppnåddes en sådan känslighet att substrat till mikroso-minkubering kunde hämtas direkt från en märkningssyntes. Detta gjorde metoden till ett effektivt screeningverktyg av nya potentiella PET spårsub-stanser.

PET-substanser kan även användas till att spåra molekyler i andra miljöer än i en människa. I avhandlingen visas hur en 11C-märkt substans kunde användas för att ta fram värdefull information vid LC-MS metodutveckling. Det var på så sätt möjligt att spåra det analyserade ämnet under provuppar-betning och analys och därmed beräkna förluster av substansen. En metod för att ta fram kvantitativa bilder (se figur 11) av spårsubstans i miniatyrise-rade kemiska analyssytem togs även fram. Användbarheten hos metoden demonstrerades bl.a. genom avbildning av en kapillärkolonn för hög-flödes-extraktion, för att studera t.ex. hur utspädning av provet påverkade extrak-tionen. Med metoden kunde även ytinteraktioner i mikrokanaler studeras. 68Ga-märkta peptider spårades på så sätt i CD-skivor utvecklade för provana-lys i mikroskala, enligt idéen om ett s.k. ”Lab-on-a-chip”, och i kanaler gjut-na i poly(dimethylsiloxane). Denna användning kan vara till stor nytta vid utveckling av protein- och peptidanalyssystem i mikroskala för att minimera provförluster p.g.a. ospecifik adsorption.

54

10 References

1 Wagner, H. N., Szabo, Z., Buchanan, J.W., Ed. Principles of Nuclear Medicine,2 ed.; W.B. Saunders Company, USA: Philadelphia, 1995.

2 Gambhir, S. S. Molecular imaging of cancer with positron emission tomography Nature Rev. Cancer 2002, 2, 683-693.

3 Tai, Y. F.; Piccini, P. Applications of positron emission tomography (PET) in neurology J. Neuro. Neuros. Psych. 2004, 75, 669-676.

4 Rimoldi, O. E.; Camici, P. G. Positron emission tomography for quantitation of myocardial perfusion J. Nucl. Card. 2004, 11, 482-490.

5 Långström, B.; Kihlberg, T.; Bergström, M.; Antoni, G.; Björkman, M.; Forn-gren, B. H.; Forngren, T.; Hartvig, P.; Markides, K.; Yngve, U.; Ögren, M. Compounds labelled with short-lived beta(+)-emitting radionuclides and some applications in life sciences. The importance of time as a parameter Acta Chem. Scand. 1999, 53, 651-669.

6 Bergström, M.; Grahnén, A.; Långström, B. Positron emission tomography microdosing: a new concept with application in tracer and early clinical drug development Eur. J. Clin. Pharmacol. 2003, 59, 357-366.

7 Cherry, S. R. Fundamentals of positron emission tomography and applications in preclinical drug development J. Clin. Pharmacol. 2001, 41, 482-491.

8 Lindner, K. J.; Hartvig, P.; Åkesson, C.; Tyrefors, N.; Sundin, A.; Långström, B. Analysis of L- methyl-C-11 methionine and metabolites in human plasma by an automated solid-phase extraction and a high- performance liquid chroma-tographic procedure J. Chromatogr. B-Biomed. Appl. 1996, 679, 13-19.

9 Jacobson, G. B.; Moulder, R.; Lu, L.; Bergström, M.; Markides, K. E.; Långström, B. Supercritical fluid extraction of C-11-labeled metabolites in tis-sue using supercritical ammonia Anal. Chem. 1997, 69, 275-280.

10 Hopfgartner, G.; Bourgogne, E. Quantitative high-throughput analysis of drugs in biological matrices by mass spectrometry Mass Spec. Rev. 2003, 22, 195-214.

11 Hevesy, G. In Nobel Lectures, Chemistry 1942-1962; Elsevier Publishing Com-pany: Amsterdam, 1964.

12 Hevesy, G.; Paneth, F. Die Löslichkeit des Bleisulfids und Bleichromats. Z. Anorg. Chem. 1913, 82, 322.

13 Hevesy, G. The absorption and translocation of lead by plants. A contribution to the application of the method of radioactive indicators in the investigation of the change of substance in plants. Biochem. J. 1923, 17, 439-445.

14 Myers, W. Georg Charles de Hevesy: The father of nuclear medicine J. Nucl. Med. 1979, 20, 590-594.

15 Jacobs, R. E.; Cherry, S. R. Complementary emerging techniques: high-resolution PET and MRI Curr. Opin. Neurobiol. 2001, 11, 621-629.

16 Townsend, D. W.; Carney, J. P. J.; Yap, J. T.; Hall, N. C. PET/CT today and tomorrow J. Nucl. Med. 2004, 45, 4S-14S.

55

17 Chaurand, P.; Schwartz, S. A.; Caprioli, R. M. Profiling and imaging proteins in tissue sections by MS Anal. Chem. 2004, 76, 86A-93A.

18 Welch, M. J.; McCarthy, T. J. The potential role of generator-produced radio-pharmaceuticals in clinical PET J. Nucl. Med. 2000, 41, 315-317.

19 Broseus In Principles of Nuclear Medicine, 2 ed.; Wagner, H. N., Szabo, Z., Buchanan, J.W., Ed.; W.B. Saunders Company, USA: Philadelphia, 1995, pp 1177.

20 Ter-Pogossian, M. In Principles of Nuclear Medicine, 2 ed.; Wagner, H. N., Szabo, Z., Buchanan, J.W., Ed.; W.B. Saunders Company, USA: Philadelphia, 1995, pp 342-346.

21 Welch, M. J.; Redvanly, C. S., Eds. Handbook of Radiopharmaceuticals; John Wiley & Sons, Ltd, 2003.

22 Iwata, R. Reference book for radiopharmaceuticals, 4th ed.; Cyric Tohoku Uni-versity, 2004.

23 Hyllbrant, B.; Tyrefors, N.; Markides, K. E.; Langstrom, B. On the use of liquid chromatography with radio- and ultraviolet absorbance detection coupled to mass spectrometry for improved sensitivity and selectivity in determination of specific radioactivity of radiopharmaceuticaIs J. Pharm. Biomed. Anal. 1999,20, 493-501.

24 Mazière, B.; Cantineau, R.; Coenen, H.; Guillaume, M.; Halldin, C.; Luxen, A.; Loc'h, C.; Luthra, S. In Radiopharmaceuticals for Positron Emission Tomogra-phy; W., S. G. a. P. V., Ed.; Kluwer Academic Publishers, 1993, pp 151-178.

25 Lundqvist, H.; Carlsson, J. Nuclide technique; Uppsala university, 1999. 26 Forngren, B. H.; Tyrefors, N.; Markides, K. B.; Långström, B. Determination of

raclopride in human plasma by on-column focusing packed capillary liquid chromatography-electrospray ionisation mass spectrometry J. Chromatogr. B2000, 748, 189-195.

27 Majors, R. E. Advances in column packing design LCGC Eur. 2003, 16, 8-13. 28 Voyksner, R. In Electrospray ionization mass spectrometry: fundamentals,

instrumentaion and applications; Cole, R., Ed.; John Wiley: New York, 1997, pp 323-341.

29 van Deemter, J. J.; Zuiderweg, F. J.; Klinkenberg, A. Longitudinal diffusion and resistance to mass transfer as causes of nonideality in chromatography Chem. Eng. Sci. 1956, 5, 271-289.

30 Knox, J. H.; Bristow, P. A. Standardization of test conditions for high perform-ance liquid chromatography columns Chromatographia 1977, 10, 279.

31 Cunico, R. L.; Gooding, K. M.; Wehr, T. Basic HPLC and CE of biomolecules;Bay Bioanalytical Laboratory, Inc., 1998.

32 Knox, J. H. Band dispersion in chromatography - a new view of A-term disper-sion J. Chromatr. A 1999, 831, 3-15.

33 Unger, K. K.; Jilge, O.; Kinkel, J. N.; Hearn, M. T. W. Evaluation of advanced silica packings for the separation of biopolymers by high-performance liquid chromatography II. Performance of non-porous monodisperse 1.5-µm Silica beads in the separation of proteins by reversed-phase gradient elution high-performance liquid chromatography J. Chromatr. A 1986, 359, 61-72.

34 Mellors, S. J.; Jorgenson, J. W. Use of 1.5-µm porous ethyl-bridged hybrid particles as a stationary-phase support for reversed-phase ultrahigh-pressure liq-uid chromatography Anal. Chem. 2004, 76, 5441-5450.

35 Giddings, J. C. Unified Separation Science; John Wiley & Sons, 1991. 36 Meyer, V. R. Practical high-performance liquid chromatography, third ed.;

John Wiley & Sons, 1998.

56

37 Zhu, C.; Goddall, D.; Wren, S. Elevated temperature HPLC: principles and applications to small molecules and biomolecules LCGC Eur. 2004, 17, 530-540.

38 MacNair, J. E.; Lewis, K. C.; Jorgenson, J. W. Ultrahigh pressure reversed-phase liquid chromatography in packed capillary columns Anal. Chem. 1997,69, 983-989.

39 Cintron, J. M.; Colon, L. A. Organo-silica nano-particles used in ultrahigh-pressure liquid chromatography Analyst 2002, 127, 701-704.

40 MacNair, J. E.; Patel, K. D.; Jorgenson, J. W. Ultrahigh pressure reversed-phase capillary liquid chromatography: Isocratic and gradient elution using columns packed with 1.0-µm particles Anal. Chem. 1999, 71, 700-708.

41 Snyder, L. R.; Dolan, J. W.; Gant, J. R. Gradient elution in high-performance liquid chromatography : I. Theoretical basis for reversed-phase systems J. Chromatr. A 1979, 165, 3-30.

42 IUPAC IUPAC Compendium of Chemical Terminology, 1997. 43 Bruins, A. P. In Electrospray ionization mass spectrometry: fundamentals, in-

strumentaion and applications; Cole, R., Ed.; John Wiley: New York, 1997, pp 107-136.

44 Karger, B. L.; Martin, M.; Guiochon, G. Role of column parameters and injec-tion volume on detection limits in liquid chromatography Anal. Chem. 1974, 46,1640-1647.

45 Mills, M. J.; Maltas, J.; Lough, J. W. Assessment of injection volume limits when using on-column focusing with microbore liquid chromatography J. Chromatr. A 1997, 759, 1-11.

46 Dole, M.; Mack, L.; Hines, R.; Mobley, R.; Ferguson, L.; Alice, M. Molecular beams of macroions J. Chem. Phys. 1968, 49, 2240-2249.

47 Whitehouse, C.; Dreyer, R.; Yamashita, M.; Fenn, J. Electrospray interface for liquid chromatographs and mass spectrometers Anal. Chem. 1985, 57, 675-679.

48 Horning, E. C.; Carroll, D. I.; Dzidic, R. N.; Haegele, K. D.; Horning, M. G.; Stillwell, R. N. Liquid chromatograph-mass spectrometer-computer analytical systems. A continous-flow system based on atmospheric pressure ionization mass spectrometry J. Chromatogr. 1974, 99, 13-21.

49 Robb, D. B.; Covey, T. R.; Bruins, A. P. Atmospheric pressure photoionisation: An ionization method for liquid chromatography-mass spectrometry Anal.Chem. 2000, 72, 3653-3659.

50 Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Electros-pray Ionization for Mass-Spectrometry of Large Biomolecules Science 1989,246, 64-71.

51 Voyksner, R. Electrospray LC/MS - can it be used to determine lower molecular weight molecules? Nature 1992, 356, 86-87.

52 Hoffmann, E.; Stroobant, V. Mass spectrometry, principles and applications,Second ed.; John Wiley & Sons, 2002.

53 Cole, R. Electrospray ionization mass spectrometry: fundamentals, instrumen-taion and applications; John Wiley: New York, 1997.

54 Barker, J.; Garner, R. C. Biomedical applications of accelerator mass spec-trometry isotope measurements at the level of the atom Rapid Comm. Mass Spec. 1999, 13, 285-293.

55 Lappin, G.; Garner, R. C. Current perspectives of C-14-isotope measurement in biomedical accelerator mass spectrometry Anal. Bioanal. Chem. 2004, 378, 356-364.

57

56 Lappin, G.; Garner, R. C. Big physics, small doses: the use of AMS and PET in human microdosing of development drugs Nature Rev. Drug Disc. 2003, 2, 233-240.

57 Cech, N. B.; Enke, C. G. Practical implications of some recent studies in elec-trospray ionization fundamentals Mass Spec. Rev. 2001, 20, 362-387.

58 Van Berkel, G. J.; Zhou, F. M. Characterization of an Electrospray Ion-Source as a Controlled-Current Electrolytic Cell Anal. Chem. 1995, 67, 2916-2923.

59 Bruins, A. P.; Covey, T. R.; Henion, J. D. Ion Spray Interface for Combined Liquid Chromatography/Atmospheric Pressure Ionization Mass-Spectrometry Anal. Chem. 1987, 59, 2642-2646.

60 Van Berkel, G. J.; Zhou, F. M.; Aronson, J. T. Changes in bulk solution pH caused by the inherent controlled-current electrolytic process of an electrospray ion source International Journal of Mass Spectrometry and Ion Processes 1997,162, 55-67.

61 Kebarle, P.; Ho, Y. In Electrospray ionization mass spectrometry: fundamen-tals, instrumentaion and applications; Cole, R., Ed.; John Wiley & Sons: New York, 1997, pp 3-63.

62 Johnstone, R.; Rose, M. Mass spectrometry for chemists and biochemists, Sec-ond ed.; Cambridge University press, 1996.

63 Ikonomou, M. G.; Blades, A. T.; Kebarle, P. Electrospray Ion Spray - a Com-parison of Mechanisms and Performance Anal. Chem. 1991, 63, 1989-1998.

64 Zhou, S. L.; Hamburger, M. Effects of solvent composition on molecular ion response in electrospray mass spectrometry: Investigation of the ionization processes Rapid Comm. Mass Spec. 1995, 9, 1516-1521.

65 Needham, S. R.; Brown, P. R.; Duff, K.; Bell, D. Optimized stationary phases for the high-performance liquid chromatography-electrospray ionization mass spectrometric analysis of basic pharmaceuticals J. Chromatr. A 2000, 869, 159-170.

66 Matuszewski, B. K.; Constanzer, M. L.; Chavez-Eng, C. M. Matrix effect in quantitative LC/MS/MS analyses of biological fluids: A method for determina-tion of finasteride in human plasma at picogram per milliliter concentrations Anal. Chem. 1998, 70, 882-889.

67 King, R.; Bonfiglio, R.; Fernandez-Metzler, C.; Miller-Stein, C.; Olah, T. Mechanistic investigation of ionization suppression in electrospray ionization J.Am. Soc. Mass Spec. 2000, 11, 942-950.

68 Zheng, J. J.; Lynch, E. D.; Unger, S. E. Comparison of SPE and fast LC to eliminate mass spectrometric matrix effects from microsomal incubation prod-ucts J. Pharm. Biomed. Anal. 2002, 28, 279-285.

69 Bonfiglio, R.; King, R. C.; Olah, T. V.; Merkle, K. The effects of sample prepa-ration methods on the variability of the electrospray ionization response for model drug compounds Rapid Comm. Mass Spec. 1999, 13, 1175-1185.

70 Guidance for Industry: Bioanalytical Method Validation; U.S. Department of Health and Human Services, Food and Drug Administration, 2001.

71 Jemal, M.; Zheng, O. Y.; Powell, M. L. A strategy for a post-method-validation use of incurred biological samples for establishing the acceptability of a liquid chromatography/tandem mass-spectrometric method for quantitation of drugs in biological samples Rapid Comm. Mass Spec. 2002, 16, 1538-1547.

72 Polson, C.; Sarkar, P.; Incledon, B.; Raguvaran, V.; Grant, R. Optimization of protein precipitation based upon effectiveness of protein removal and ionization effect in liquid chromatography-tandem mass spectrometry J. Chromatogr. B2003, 785, 263-275.

58

73 Hennion, M. C. Solid-phase extraction: method development, sorbents, and coupling with liquid chromatography J. Chromatr. A 1999, 856, 3-54.

74 Souverain, S.; Rudaz, S.; Veuthey, J.-L. Restricted access materials and large particle supports for on-line sample preparation: an attractive approach for bio-logical fluids analysis J. Chromatogr. B 2004, 801, 141-156.

75 Knox, J. H. Evidence for turbulence and coupling in chromatographic columns Anal. Chem. 1966, 38, 253-261.

76 Pretorius, V.; Smuts, T. Turbulent flow chromatography: a new approach to faster analysis Anal. Chem. 1966, 38, 274-281.

77 Kirkland, J. High-speed liquid chromatography with controlled surface porosity supports J. Chromatogr. Sci. 1969, 7, 7-12.

78 Quinn, H. M.; Takarewski, J. J. In International Patent No. WO 97/16724, 1997. 79 Oberhauser, C. J.; Niggebrugge, A. E.; Lachance, D.; Takarewski, J. J.; Pegram,

M. M.; Quinn, H. M. Turbulent-flow LC for LC-MS and LC-MS-MS bioanaly-sis LCGC N. Am. 2000, 18, 716-725.

80 Ayrton, J.; Dear, G. J.; Leavens, W. J.; Mallett, D. N.; Plumb, R. S. The use of turbulent flow chromatography mass spectrometry for the rapid, direct analysis of a novel pharmaceutical compound in plasma Rapid Comm. Mass Spec. 1997,11, 1953-1958.

81 Wu, J. T.; Zeng, H.; Qian, M. X.; Brogdon, B. L.; Unger, S. E. Direct plasma sample injection in multiple-component LC-MS-MS assays for high-throughput pharmacokinetic screening Anal. Chem. 2000, 72, 61-67.

82 Zimmer, D.; Pickard, V.; Czembor, W.; Muller, C. Comparison of turbulent-flow chromatography with automated solid-phase extraction in 96-well plates and liquid-liquid extraction used as plasma sample preparation techniques for liquid chromatography-tandem mass spectrometry J. Chromatr. A 1999, 854,23-35.

83 Ding, J. M.; Neue, U. D. A new approach to the effective preparation of plasma samples for rapid drug quantitation using on-line solid phase extraction mass spectrometry Rapid Comm. Mass Spec. 1999, 13, 2151-2159.

84 Lim, H. K.; Chan, K. W.; Sisenwine, S.; Scatina, J. A. Simultaneous screen for microsomal stability and metabolite profile by direct injection turbulent-laminar flow LC-LC and automated tandem mass spectrometry Anal. Chem. 2001, 73,2140-2146.

85 Zeng, H.; Wu, J. T.; Unger, S. E. The investigation and the use of high flow column-switching LC/MS/MS as a high-throughput approach for direct plasma sample analysis of single and multiple components in pharmacokinetic studies Journal of Pharmaceutical and Biomedical Analysis 2002, 27, 967-982.

86 Herman, J. L. Generic method for on-line extraction of drug substances in the presence of biological matrices using turbulent flow chromatography Rapid Comm. Mass Spec. 2002, 16, 421-426.

87 Duncan, T. Advanced physics. Volume 1: Materials and mechanics; Murray: London, 1987.

88 Ayrton, J.; Dear, G. J.; Leavens, W. J.; Mallett, D. N.; Plumb, R. S. Optimisa-tion and routine use of generic ultra-high flow-rate liquid chromatography with mass spectrometric detection for the direct on-line analysis of pharmaceuticals in plasma J. Chromatr. A 1998, 828, 199-207.

89 Gibson, G.; Skett, P. Introduction to drug metabolism, Third ed.; Nelson Thor-nes Publishers, 2001.

90 Ekins, S.; Ring, B. J.; Grace, J.; McRobie-Belle, D. J.; Wrighton, S. A. Present and future in vitro approaches for drug metabolism J. Pharmacol. Toxicol. Meth. 2000, 44, 313-324.

59

91 Plant, N. Strategies for using in vitro screens in drug metabolism Drug Discov-ery Today 2004, 9, 328-336.

92 Alberts, B.; Bray, D.; Lewis, J.; Raff, M.; Roberts, K.; Watson, J. Molecular biology of the cell; Garland Publishing Inc.: New York, 1994.

93 Samson, Y.; Hantraye, P.; Baron, J. C.; Soussaline, F.; Comar, D.; Maziere, M. Kinetics and displacement of [11C]RO 15-1788, a benzodiazepine antagonist, studied in human brain in vivo by positron tomography Eur. J. Pharmacol.1985, 110, 247-251.

94 Persson, A.; Ehrin, E.; Eriksson, L.; Farde, L.; Hedström, C.-G.; Litton, J.-E.; Mindus, P.; Sedvall, G. Imaging of [11C]-labelled RO 15-1788 binding to ben-zodiazepine receptors in the human brain by positron emission tomography J.Psych. Res. 1985, 19, 609-622.

95 Koeppe, R. A.; Holthoff, V. A.; Frey, K. A.; Kilbourn, M. R.; Kuhl, D. E. Com-partmental Analysis of C-11 Flumazenil Kinetics for the Estimation of Ligand Transport Rate and Receptor Distribution Using Positron Emission Tomography J. Cerebr. Bl. Fl. Metabol. 1991, 11, 735-744.

96 Lassen, N. A.; Bartenstein, P. A.; Lammertsma, A. A.; Prevett, M. C.; Turton, D. R.; Luthra, S. K.; Osman, S.; Bloomfield, P. M.; Jones, T.; Patsalos, P. N.; Oconnell, M. T.; Duncan, J. S.; Andersen, J. V. Benzodiazepine Receptor Quan-tification in-Vivo in Humans Using C-11 Flumazenil and Pet - Application of the Steady-State Principle J. Cerebr. Bl. Fl. Metabol. 1995, 15, 152-165.

97 Hunkeler, W. Preclinical Research Findings with Flumazenil (Ro-15-1788, Anexate) - Chemistry Eur. J. Anaest. 1988, 37-38.

98 Klotz, U. Drug-Interactions and Clinical Pharmacokinetics of Flumazenil Eur. J. Anaest. 1988, 103-108.

99 Halldin, C.; Stoneelander, S.; Thorell, J. O.; Persson, A.; Sedvall, G. C-11-Labelling of Ro 15-1788 in 2 Different Positions, and Also C-11-Labelling of Its Main Metabolite Ro 15-3890, for Pet Studies of Benzodiazepine Receptors Appl. Rad. Isot. 1988, 39, 993-997.

100 Fletcher, A.; Bill, D. J.; Cliffe, I. A.; Forster, E. A.; Jones, D.; Reilly, Y. A Pharmacological Profile of Way-100635, a Potent and Selective S-Ht1a Recep-tor Antagonist Brit. J. Pharmacol. 1994, 112, U48-U48.

101 Forster, E. A.; Cliffe, I. A.; Bill, D. J.; Dover, G. M.; Jones, D.; Reilly, Y.; Fletcher, A. A Pharmacological Profile of the Selective Silent 5-Ht1a Receptor Antagonist, Way-100635 Eur. J. Pharmacol. 1995, 281, 81-88.

102 Mathis, C. A.; Simpson, N. R.; Mahmood, K.; Kinahan, P. E.; Mintun, M. A. [C-11] Way-100635 - a Radioligand for Imaging 5-Ht1a Receptors with Posi-tron Emission Tomography Life Sci. 1994, 55, Pl403-Pl407.

103 Pike, V. W.; Mccarron, J. A.; Lammerstma, A. A.; Hume, S. P.; Poole, K.; Grasby, P. M.; Malizia, A.; Cliffe, I. A.; Fletcher, A.; Bench, C. First Delinea-tion of 5-Ht1a Receptor in Human Brain with Pet and [C-11]Way-100635 Eur. J. Pharmacol. 1995, 283, R1-R3.

104 Farde, L.; Ito, H.; Swahn, C. G.; Pike, V. W.; Halldin, C. Quantitative analyses of carbonyl-carbon-11-WAY-100635 binding to central 5-hydroxytryptamine-1A receptors in man J. Nucl. Med. 1998, 39, 1965-1971.

105 Osman, S.; Lundkvist, C.; Pike, V. W.; Halldin, C.; McCarron, J. A.; Swahn, C. G.; Ginovart, N.; Luthra, S. K.; Bench, C. J.; Grasby, P. M.; Wikstrom, H.; Barf, T.; Cliffe, I. A.; Fletcher, A.; Farde, L. Characterization of the radioactive me-tabolites of the 5-HT1A receptor radioligand, [O-methyl-C-11]WAY-100635, in monkey and human plasma by HPLC: Comparison of the behaviour of an iden-tified radioactive metabolite with parent radioligand in monkey using PET Nucl. Med. Biol. 1996, 23, 627-634.

60

106 Osman, S.; Lundkvist, C.; Pike, V. W.; Halldin, C.; McCarron, J. A.; Swahn, C. G.; Farde, L.; Ginovart, N.; Luthra, S. K.; Gunn, R. N.; Bench, C. J.; Sargent, P. A.; Grasby, P. M. Characterisation of the appearance of radioactive metabolites in monkey and human plasma from the 5-HT1A receptor radioligand, [carbonyl-C-11]WAY-100635 - Explanation of high signal contrast in PET and an aid to biomathematical modelling Nucl. Med. Biol. 1998, 25, 215-223.

107 Pike, V. W.; McCarron, J. A.; Lammertsma, A. A.; Osman, S.; Hume, S. P.; Sargent, P. A.; Bench, C. J.; Cliffe, I. A.; Fletcher, A.; Grasby, P. M. Exquisite delineation of 5-HT1A receptors in human brain with PET and [carbonyl-C-11]WAY-100635 Eur. J. Pharmacol. 1996, 301, R5-R7.

108 Andree, B.; Halldin, C.; Pike, V. W.; Gunn, R. N.; Olsson, H.; Farde, L. The PET radioligand [carbonyl-C-11]desmethyl-WAY-100635 binds to 5-HT1A re-ceptors and provides a higher radioactive signal than [carbonyl-C-11]WAY-100635 in the human brain J. Nucl. Med. 2002, 43, 292-303.

109 Wilson, A. A.; Inaba, T.; Fischer, N.; Dixon, L. M.; Nobrega, J.; DaSilva, J. N.; Houle, S. Derivatives of WAY 100635 as potential imaging agents for 5-HT1A receptors: Syntheses, radiosyntheses, and in vitro and in vivo evaluation Nucl. Med. Biol. 1998, 25, 769-776.

110 Cliffe, I. A. A retrospect on the discovery of WAY-100635 and the prospect for improved 5-HT1A receptor PET radioligands Nucl. Med. Biol. 2000, 27, 441-447.

111 Pike, V. W.; Halldin, C.; Wikstrom, H.; Marchais, S.; McCarron, J. A.; Sandell, J.; Nowicki, B.; Swahn, C. G.; Osman, S.; Hume, S. P.; Constantinou, M.; An-dree, B.; Farde, L. Radioligands for the study of brain 5-HT1A receptors in vivo - Development of some new analogues of WAY Nucl. Med. Biol. 2000, 27, 449-455.

112 Houle, S.; DaSilva, J. N.; Wilson, A. A. Imaging the 5-HT1A receptors with PET: WAY-100635 and analogues Nucl. Med. Biol. 2000, 27, 463-466.

113 Sandell, J.; Halldin, C.; Pike, V. W.; Chou, Y. H.; Varnas, K.; Hall, H.; Mar-chais, S.; Nowicki, B.; Wikstrom, H. V.; Swahn, C. G.; Farde, L. New halo-genated [C-11]WAY analogues, [C-11]6FPWAY and [C-11]6BPWAY - Radio-synthesis and assessment as radioligands for the study of brain 5-HT1A recep-tors in living monkey Nucl. Med. Biol. 2001, 28, 177-185.

114 Itsenko, O.; Långström, B.; Kihlberg, T. Use of Lithium Amides in Palladium Mediated Synthesis of [Carbonyl-11C]Amides 2004, Manuscript.

115 Pike, V.; McCarron, J.; Hume, S. P.; Ashworth, S.; Opacka-Juffry, J.; Osman, S.; Lammerstma, A. A.; Poole, K.; Fletcher, A.; White, C.; Cliffe, I. A. Pre-clinical development of a radioligand for studies of central 5-HT1A receptors in vivo -[11C]WAY-100635 Med. Chem. Res. 1994, 5, 208-227.

116 Ma, Y.; Lang, L. X.; Kiesewetter, D. O.; Jagoda, E.; Sassaman, M. B.; Der, M.; Eckelman, W. C. Liquid chromatography-tandem mass spectrometry identifica-tion of metabolites of two 5-HT1A antagonists, N-{2-[4-(2-methoxylphenyl)piperazino]ethyl}-N-(2-pyridyl) trans- and cis-4-fluorocyclohexanecarboxamide, produced by human and rat hepatocytes J.Chrom. B 2001, 755, 47-56.

117 Ma, Y.; Lang, L. X.; Kiesewetter, D. O.; Eckelman, W. C. Liquid chromatogra-phy-tandem mass spectrometry identification of metabolites of three phenylcar-boxyl derivatives of the 5-HT1A antagonist, N-(2-(4-(2-methoxyphenyl)-1-piperazinyl)ethyl)-N-(2-pyridyl) trans-4-fluorocyclohexanecarboxamide (FCWAY), produced by human and rat hepatocytes J. Chrom. B 2002, 780, 99-110.

61

118 Andersson, J. L. R., Schneider, H. Design, construction and six years' experi-ence of an integrated system for automated handling of discrete blood samples Eur. J. Nuc. Med. 1998, 25, 85-90.

119 Johnston, R.; Pickett, S.; Barker, D. Autoradiography using storage phosphor technology Electrophoresis 1990, 11, 355-360.

120 Sihver, W.; Sihver, S.; Bergstrom, M.; Murata, T.; Matsumura, K.; Onoe, H.; Andersson, Y.; Bjurling, P.; Fasth, K. J.; Westerberg, G.; Ogren, M.; Jacobsson, G.; Lundqvist, H.; Oreland, L.; Watanabe, Y.; Langstrom, B. Methodological aspects for in vitro characterization of receptor binding using C-11-labeled re-ceptor ligands: A detailed study with the benzodiazepine receptor antagonist C- 11 Ro 15-1788 Nucl. Med. Biol. 1997, 24, 723-731.

121 Duncan, M.; Gilbert, M.; Lee, J.; Warchol, M. Development and Comparison of Experimental Assays to Study Protein Peptide Adsorption onto Surfaces J. Col-loid Interf. Sci. 1994, 165, 341-345.

122 Chico, D. E.; Given, R. L.; Miller, B. T. Binding of cationic cell-permeable peptides to plastic and glass Peptides 2003, 24, 3-9.

123 Loomans, E. E. M. G.; Gribnau, T. C. J.; Bloemers, H. P. J.; Schielen, W. J. G. Adsorption studies of tritium-labeled peptides on polystyrene surfaces J. Immu-nol. Methods 1998, 221, 131-139.

124 Hu, S. W.; Ren, X. Q.; Bachman, M.; Sims, C. E.; Li, G. P.; Allbritton, N. Sur-face modification of poly(dimethylsiloxane) microfluidic devices by ultraviolet polymer grafting Anal. Chem. 2002, 74, 4117-4123.

125 Duncan, M. R.; Lee, J. M.; Warchol, M. P. Influence of Surfactants Upon Pro-tein Peptide Adsorption to Glass and Polypropylene Int. J. Pharm. 1995, 120,179-188.

126 Walker, A. K.; Land, C. M.; Kinsel, G. R.; Nelson, K. D. Quantitative determi-nation of the peptide retention of polymeric substrates using matrix-assisted la-ser desorption/ionization mass spectrometry J. Am. Soc. Mass Spec. 2000, 11,62-68.

127 Ruzgas, T. A.; Kazlauskas, A. V.; Razumas, V. J.; Kulys, J. J. Adsorption of Heme-Containing Peptides on Silicon Surfaces J. Colloid Interf. Sci. 1992, 154,97-103.

128 Rossier, J. S.; Gokulrangan, G.; Girault, H. H.; Svojanovsky, S.; Wilson, G. S. Characterization of protein adsorption and immunosorption kinetics in photoab-lated polymer microchannels Langmuir 2000, 16, 8489-8494.

129 Linder, V.; Verpoorte, E.; Thormann, W.; de Rooij, N. F.; Sigrist, M. Surface biopassivation of replicated poly(dimethylsiloxane) microfluidic channels and application to heterogeneous immunoreaction with on-chip fluorescence detec-tion Anal. Chem. 2001, 73, 4181-4189.

130 Munson, M.; Hasenbank, M.; Fu, E.; Yager, P. Suppression of non-specific adsorption using sheath flow Lab on a Chip 2004, 4, 438-445.

131 Velikyan, I.; Beyer, G.; Långström, B. Microwave-supported preparation of 68Ga bioconjugates with high specific radioactivity Biocon. Chem. 2004, 15,554-560.

132 Bergström, S.; Edenwall, N.; Lavén, M.; Velikyan, I.; Långström, B.; Markides, K. E. Polyamine deactivation of integrated poly(dimethylsiloxane) structures investigated by radionuclide imaging and capillary electrophoresis experiments Anal. Chem. 2004, accepted for publication.

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 2

Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science and Technology, Uppsala University, is usually a summary of a number of papers. A few copies of the complete dissertation are kept at major Swedish research libraries, while the summary alone is distributed internationally through the series Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology. (Prior to January, 2005, the series was published under the title "Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology".)

Distribution: publications.uu.seurn:nbn:se:uu:diva-4727

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2005