 

 

 

 

 

 

mm“ 9””76670

 

 

 

 

 

 

 

 

 

 

\|\\\\\\l|\\\\||\\llllllllllllllllll

3 1293 006 6050

 

LIBRARY A
Michigan State
University .

 

 

 

This is to certify that the
dissertation entitled
Development of Sensitive Assays Based on Chemical
Oxidation and Electron Capture Negative Ionization

Mass Spectrometry for the Determination of
Corticosteroids in Biological Fluids

presented by

Kathleen Ann Kayganich

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in Chemistry

  
 
 

 

Major professor

Date 8-1 1—89

MSU is an Afﬁrmative Action/Equal Opportunity Institution 0-12771

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

 

 

 

l—r: 1 —-——
DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

“—7

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative ActiotVEqual Opportunity Institution

 

 

 

 

DEVELOPMENT OF SENSITIVE ASSAYS BASED ON CHEMICAL
OXIDATION AND ELECTRON CAPTURE NEGATIVE IONIZATION MASS
SPECTROMETRY FOR THE DETERMINATION OF CORTICOSTEROIDS IN

BIOLOGICAL FLUIDS

by

Kathleen Ann Kayganich

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry

1989

 

CMOM‘I

ABSTRACT

DEVELOPMENT OF SENSITIVE ASSAYS BASED ON CHEMICAL
OXIDATION AND ELECTRON CAPTURE NEGATIVE IONIZATION MASS
SPECTROMETRY FOR THE DETERMINATION OF CORTICOSTEROIDS IN

BIOLOGICAL FLUIDS

By

Kathleen Ann Kayganich

The sensitivity and selectivity of electron capture negative ionization mass
spectrometry (ECNI/MS) make this a valuable technique for the determination of low
levels of analytes in complex matrices. However, most compounds of biological interest
do not have a high electron capture response. Derivatization of functional groups with
halogenated moieties is usually employed to enhance the electron capture responses of
compounds having poor electron capture response. Often, derivatization causes an
increase in the chemical background, so that the increase in the signal-to—background
ratio for the analyte is not greatly enhanced.

As an alternative to the use of derivatization, chemical oxidation has been
explored as a more selective means to enhance electron capture response. The responses
of some corticosteroid drugs are increased markedly upon oxidation. The responses of
most endogenous steroids are not greatly enhanced by oxidation. Dexamethasone is a
corticosteroid drug that is oxidized to a highly electrophilic analog. Sensitive
methodology is needed to determine low levels of dexamethasone in human plasma.
Procedures for the isolation and oxidation of dexamethasone were investigated to
optimize the production and recovery of the oxidized analog. A sensitive assay was
developed for dexamethasone in plasma using oxidation and ECNI-MS detection. The
results obtained from the analyses of several plasma samples using this method were

comparable to the results obtained by a conventional GC—MS method and by RIA in an

interlaboratory study.

The oxidation product of the major metabolite of dexamethasone, 6B-
hydroxydexamethasone, was also detected by ECNI/MS in the analysis of plasma
samples.

Tandem mass spectrometry (MS/MS) was investigated for the detection of
oxidized dexamethasone in the plasma matirx. The direct inlet probe (DIP) could be
used to introduce the sample for analysis by ECNI/MS/MS due to the additional
selectivity provided by this technique.

An assay for 6B-hydroxycortisol, based on oxidation and ECNI/MS analysis, was
also investigated. 6B-Hydroxycortisol is a minor component of urine that is useful as an
index of microsomal enzyme activity. The results obtained for the determination of 6B-
hydroxycortisol in human urine by this method were comparable to those obtained using
a conventional GC-MS method. The oxidation method with GC-ECNI/MS analysis was
more selective than the conventional GC-MS method. This allowed the use of less

rigorous GC conditions and provided a shorter analysis.

 

 

ACKNOWLEDGEMENTS

I would like to acknowledge Doug Gage, Mike Davenport, Mel Micke, Melinda
Beming, Betty Baltzer, and Bev Charnberlin, the staff of the Michigan State University
Mass Spectrometry Facility, for the expert assistance and encouragement they have given
me during my years at Michigan State. I especially appreciated their willingness to help
at a moment’s notice. I would also like to thank the Watson group, past and present, for
their help and encouragement. A thank-you to Linda Doherty, Dan Kassel and Brian
Musselman for teaching me to boldly disassemble mass spectrometers and for teaching
me many other things. Thanks especially to Tim Heath for helpful discussions and his
willingness to work with me on the MS/MS experiments. I would also like to thank
Mark McMills for all the valuable advice he gave me concerning organic chemistry and
Blake Rees for the use of his computer. A special thanks to Curt Heine for his
thoughtfulness and encouragement, and to Ellen Yurek, Helen Mayer, and Dave Wagner
for their friendship and support. Thanks also to Ben Gardner for plumbing expertise and
humor, to Gary Schultz for some good meals, to Karen Light for her thoughtfulness, for
help with the HP5985, and for making ASMS hotel reservations year after year. I also
want to thank my friends Lauren and Mark McMills, Deb de Vries, Liann Short, Pat
Redmon, Helen Palmer, Dennis Rich, Mark and Sherre Whalon, Joan Stark, and my
parents for listening and being their for me and helping me to grow. And, ﬁnally, I
would like to thank Dr. Watson for his helpfulness - especially in preparation for ASMS
presentations, for research support, for his encouragement and advice, the freedom he

allowed me in the lab, and for salmon steaks and a good auto mechanic!

iv

 

 

 

TABLE OF CONTENTS

List of Figures ........................................... viii
List of Tables ............................................ xi
Key to Abbreviations ....................................... xiii
CHAPTER 1. INTRODUCTION ................................ 1
1. Mass Spectrometry in Biomedical Research ...................... 1
A. Advantages of Mass Spectrometry .......................... 1

1. Specificity ...................................... 1

2. Sensitivity ...................................... 4

B. Derivatization of Corticosteroids for Analysis by GC-MS ............ 4

1. Qualitative Gas Chromatography-Mass Spectrometry ............ 4

2. Quantitative Mass Spectrometry ......................... 5

3. Mass Spectrometry as a Reference Method .................. 6

H. Conventional Mass Spectrometry for Corticosteroids Analysis ........... 7
A. Corticosteroids ..................................... 8

B. Derivatization of Corticosteroids for Analysis by GC—MS ........... 10

C. Other Mass Spectrometric Techniques for Corticosteroids Analysis ..... 13

III. Electron Capture Negative Ion Mass Spectrometry .................. 14
A. History and Theory of Electron-Capture Negative Ionization ......... 14

B. Limitations of Electron Capture Negative Ionization .............. 15

C. Advantages of ECNI ................................. l7

1. Sensitivity ..................................... 17

2. Selectivity ..................................... 18

D. Derivatization to Enhance ECNI Response .................... 19

IV. An Alternative Approach for Enhancing ECNI Response of Corticosteroids . . .22
A. Previous work ..................................... 23

B. Majors goals of this research ............................ 26
References .............................................. 27

V

 

 

CHAPTER 2. DEVELOPMENT OF OPTIMAL CONDITIONS FOR ISOLATION

AND CHEMICAL OXIDATION OF DEXAMETHASONE ...... 30
I. Introduction ......................................... 3 0
11. Method Development ................................... 32
A. Extraction of Dexamethasone and 6B—Hydroxydexamethasone
from Plasma ...................................... 32
1. Solvent Extraction vs. Solid Phase Extraction ..... , ........... 32
2. Solid-Phase, Reverse—Phase (C18) Extraction ................ 36
3. Inﬂuence of Com osition Column Wash (CH3OHzH20)
on Recovery of 6 -Hydro>~. ,. dexamethasone ................. 38
B. Oxidation ........................................ 40
1. Choice of Oxidation Reagent .......................... 4O
2. Optimization of FCC Oxidation Conditions ................. 51
3. Determination of the Yield of 11,17-Keto Dexamethasone ........ 54
4. The Effect of Extracted Plasma Matrix on the Oxidation of
Dexamethasone .................................. 55
5. Oxidation of 6B-Hydroxydexarnethasone ................... 57
C. Removal of Excess Reagent and Isolation of Oxidized
Dexamethasone and 6B-Hydroxydexamethasone ................. 61
1. Silica vs. Solvent Extraction .......................... 61
2. Comparison of Sephadex LII-20 and Silica .................. 6 1
3. Optimization of silica solid phase extraction of 11,17-keto
dexamethasone and 6,11,17—keto dexamethasone .............. 62
111. Conclusion ......................................... 66
References .............................................. 67

CHAPTER 3. COMPARISON OF GC-MS AND MS/MS TECHNIQUES IN

II.

THE ANALYSIS OF PLASMA FOR DEXAMETHASONE ...... 69
Introduction ......................................... 69
Sample Preparation and Analysis of Plasma for Dexamethasone Using SIM . . .70
A. Experimental ...................................... 7 O

1. Materials and Sample Preparations ....................... 7 0
2. Adjustment of Instrumental Parameters .................... 74
B. Results and Discussion ................................ 77
C. Conclusion ....................................... 9O
MS/MS Analysis of Plasma for Dexamethasone ................... 91

vi

A. Experimental ...................................... 9 2

B. Results and Discussion ................................ 93
C. Conclusion ...................................... 101
References ............................................. 102

CHAPTER 4. INVESTIGATION OF AN OXIDATION BASED-ASSAY FOR

6B-HYDROXYCORTISOL IN URINE; COMPARISON WITH

A CONVENTIONAL GC-MS METHOD ................. 103
I. Introduction ........................................ 103
A. 6l3-Hydroxycortisol and Enzyme Induction ................... 103
B. Methods for Determination of Urinary 6B-Hydroxycortisol .......... 104
C. Proposed ECNI-MS Method for Determination of 6B-Hydroxycortisol . . .105
H. Experimental and Results ................................ 107
A. Preliminary Study .................................. 108
l. Extraction of Urine ............................... 108
2. Oxidation ..................................... 108
3. GC-ECNI/MS and GC-EI/MS Analysis of Oxidized Urine Extract . . .109
4. Investigation of Sample Introduction by Direct Inlet Probe

for ECNI/MS ................................... 110
5. Conclusion .................................... 112
B. Determination of 6B-Hydroxycortisol in Guinea Pig Urine ........... 114
1. Sample Preparation ............................... 114
2. Instrumental Analysis ............................. 116
3. Results and Discussion ............................. 119

C. Determination of 6B—Hydroxycortisol in Human Urine by GC-EI/MS,
GC—ECNI/MS, and DIP-ECNI/MS ........................ 121
1. Sample Preparations .............................. 121
2. Deuteration Experiments ............................ 122
3. Instrumental Analysis ............................. 124
4. Results and Discussion ............................. 125
III. Conclusions ........................................ 126
References ............................................. 127
CHAPTER 5. SUMMARY ................................... 129
References ............................................. 133

vii

 

LIST OF FIGURES

Figure 1.1 Numbering system for steroid carbon skeleton ................. 9

Figure 1.2 ECNI mass spectrum of PFB-oxime, TMS-ether of testosterone.
The base peak at, m/z 181, represents the pentafluorobenzyl anion.
This ﬁgure was taken from S. J. Gaskell, "Analysis of Steroids" in
Methods of Biochemical Analysis, vol. 29, D. Glick, ed., Wiley
Interscience, New York, pp. 385-434, 1983 .................. 21

Figure 1.3 Structures of steroids having high electron capture responses. I

and II result from the oxidation of the naturally occuring

c,ompounds cholesterol and 6B- -hydroxycortisol. III, IV, and V

result from the oxidation of the corticosteroid drugs prednisolone,
fludrocortisone, and dexamethasone, respectively .............. 24

Figure 2.1 Oxidative conversion of dexamethasone to 11,17—keto
dexamethasone ................................... 33

Figure 2.2 Reconstructed selected ion current proﬁles from GC-ECNI/MS
analysis of oxidized dexamethasone, 13C6-2H3- dexamethasone,
and 6[3_-hydroxydexamethasone isolated from plasma by a) solvent
extraction and b) C18 solid- -phase extraction. Arrow indicates
retention time of oxidized dexamethasone ................... 35

Figure 2.3 Structures of some chromium (VI) oxidants; chromic acid (1),
chromyl acetate (2), chromyl chloride (3), tert-butyl chromate (4),
chromium trioxide-pyridine complex (5), dipyridine chromium
(VI) oxide (6), pyridinium chlorochromate (7), pyridinium
dichromate (8), and 2,2’-bipyridinium chlorochromate (9).
Adaptedfromreference(15)..........................42

Figure 2.4 Mechanism for the oxidation of alcohols with chromium (VI)
oxidants ....................................... 44

Figure 2.5 a) Mechanism for the oxidative cleavage of the CC bond of a 1,2-
diol and b) proposed mechanism for the oxidative cleavage of the
dexamethasone C-l7 sidechain ......................... 46

Figure 2.6 Oxidation of 6B-Hydroxydexamethasone to 6,11,17-keto
dexamethasone ................................... 58

Figure 2.7 Mass spectra of oxidized 6B—hydroxydexamethasone obtained by
a) electron impact ionization and b) electron capture negative
ionization ...................................... 59

viii

Figure 3.1 Selected ion current proﬁles for detection of the TMS-enol-TMS
derivative of dexamethasone within plasma matrix,
by GC—EI/MS .................................... 7 1

Figure 3.2 Selected ion current proﬁle for detection of oxidized
dexamethasone in plasma matrix, by GC-ECNI/MS ............. 72

Figure 3.3 ECNI mass spectrum of 11,17-keto dexamethasone and structure
of 11,17-keto dexamethasone .......................... 78

Figure 3.4 GC-ECNI/MS selected ion current proﬁle from 160 fg of 11,17-
keto dexamethasone standard .......................... 79

Figure 3.5 Calibration curves for determination of dexamethasone in plasma
b GC-ECNI/MS with SIM using a) 35 ng, and b) 3.5 ng of
1 C5-2H3-dexamethasone internal standard .................. 81

Figure 3.6 Selected ion current proﬁles from GC-ECNI/MS analysis of
plasma sample P5 from patient undergoing the DST ............. 85

Figure 3.7 GC—ECNI/MS selected ion current proﬁles at m/z 310, 319, and
324, representing the (M-HF)- ions of oxidized dexamethasone,
13C6-2H3-dexamethasone, and 6B-hydroxydexamethasone,
respectively, in plasma sample P5 ........................ 86

Figure 3.8 Daughter ion spectrum of the molecular anion of 11,17-keto
dexamethasone obtained by ECNI/MS/MS with CAD ............ 94

Figure 3.9 The abundance of the (M-HF)- ion, m/z 310, from CAD of the
molecular anion of 11,17-keto dexamethasone at different collision
offset energies and collision gas pressures ................... 96

Figure 3.10 Comparison of selected ion current proﬁles obtained by a) GC-
ECNI/MS using SIM, with those obtained by b) GC-ECNI/MS/MS
with SRM, for the detemination of dexamethasone in a real plasma
sample ........................................ 98

Figure 3.11 Calibration curves for determination of dexamethasone in plasma
b GC-ECNI/MS/MS with SRM using a) 35 ng, and b) 3.5 ng of
1 C6-2H3-dexamethasone internal standard .................. 99

ix

Figure 4.1 a) Oxidative conversion of 6B-hydroxycortisol to electrophilic
product. b) ECNI mass spectrum of oxidized 6B-
hydroxycortisol .................................. 106

Figure 4.2 Reconstructed total ion current chromatogram and mass
chromatogram at m/z 314 from a) GC-EI/MS and b) GC-
ECNI/MS analyses of an oxidized urine extract ............... 111

Figure 4.3 Selected ion current proﬁle obtained by DIP-ECNI/MS analysis of
an oxidized urine extract. The molecular anion of oxidized 6B-
hydroxycortisol corresponds to m/z 314 ................... 113

Figure 4.4 Sample preparation scheme for the determination of 6B-
hydroxycortisol in urine by MO-TMS derivatization and by
oxidation ...................................... 115

Figure 4.5 Selected ion current proﬁles obtained from a) GC-EI/MS analysis
of MO-TMS derivatized guinea pig urine sample and b) GC-
ECNl/MS analysis of oxidized guinea pig urine sample .......... 117

 

Table 2.1

Table 2.2

Table 2.3

Table 2.4

Table 2.5

Table 2.6
Table 2.7
Table 2.8
Table 2.9

Table 2.10
Table 2.11

Table 2.12

Table 2.13

Table 2.14

Table 3.1

LIST OF TABLES

Extraction recoveries of dexamethasone (dex) and 6B:
hydroxydexamethasone (6B-OH dex) vs. percentage of methanol
in solid phase extraction wash ......................... 3 9

Relative yields of 11,17-keto dexamethasone obtained with three
different oxidation reagents ........................... 4 3

Reaction time vs. yield of 11,17-keto dexamethasone by PVPDC
oxidation of dexamethasone at 70'C, in cyclohexane ............ 47

Relative yield of 11,17-keto dexamethasone by PVPDC oxidation
vs. presence of excess H20 and heating and stirring. 26 ug of
dexamethasone was used for each oxidation .................. 49

Relative yield of 11,17-keto dexamethasone by PDC oxidation vs.

time ......................................... 50
The effect of a base on the yield of 11,17-keto dexamethasone ....... 52
The effect of heating on the yield of 11,17-keto dexamethasone ...... 53
Yield of 11,17-keto dexamethasone by PCC oxidation vs. time ...... 55

Ratio of GC-FID responses for 11,17-keto dexamethasone and the

internal standard, 4-androstene-3,6,17-one, for oxidations

conducted with and without presence of plasma matrix extract.

8: std. deviation of 3 injections ......................... 56

Relative yield of 6,11,17-keto dexamethasone vs. time ........... 60

Recovery of 11,17-keto dexamethasone by Sephadex LH-20 and
by silica SPE column ............................... 62

HPLC analysis of fractions from silica column used for removal of
FCC following oxidation of dexamethasone .................. 63

DIP-ECNI/MS analysis of silica SPE column for recovery of
11,17-keto dexamethasone and 6,11,17-keto dexamethasone ........ 64

GC-ECD analysis of silica fractions for isolation of 6,11,17-keto
dexamethasone ................................... 65

Determination of dexamethasone in 1 ml blank plasma "spiked"
with 2.09 ng of dexamethasone by GC-ECNI-MS with SIM ........ 84

xi

Table 3.2

Table 3.3

Table 4.1

Table 4.2

Comprison of concentrations of dexamethasone in ng/ml of plasma
as determined by GC-ECNI-MS with SIM and by conventional
GC-EI/MS of the TMS-enol-TMS derivative ................. 87

Comparison of concentrations of dexamethasone obtained by GC-
ECNI-SIM, GC-ECNI-SRM, and DIP-ECNI-SRM ............. 100

Comparison of GC-EI/MS of the MO-TMS derivatived urine with
GC-ECNI/MS, and DIP—ECNI/MS, of the oxidized urine, for the
determination of 6B-hydroxycortisol excretion by control and

indu- *d guinea pigs ............................... 119

Comparison of GC-EI/MS of MO-TMS derivatives with GC-

ECNl/MS and DIP-ECNI/MS of oxidized urine samples for the
determination of the total daily excretion of 6B—hydroxycortisol by

a 49-year old male ................................ 125

xii

 

SIM
SRM
MS
MS/MS
CAD
TQMS
Q1

Q2

Q3

EF
EI
CI
NCI
EC
ECNI
DIP
TIC
APC
S/N
s/n
s/b
LOD
GC

KEY TO ABBREVIATIONS

selected ion monitoring

selected reaction monitoring

mass spectrometry

tandem mass spectrometry
collisionally activated dissociation
triplequadrupole mass spectometer
ﬁrst quadrupole

second quadrupole

third quadrupole

high resolution

electric ﬁeld

electron impact

chemical ionization

negative chemical ionization
electron capture

electron capture negative ionization
direct inlet probe

total ion current

automatic pressure controller
signal-to—noise

signal-to-noise
signal-to-background

limit of detection

gas chromatography

xiii

LC
HPLC
TLC

ECD

UV

C1 8

SPE

RIA

PCC
PDC
PVPDC
PyrzCr03
N aOAc
BSTFA
TMS
MO-TMS
TBDMS
PFB
DHEA
LTB4
DST

liquid chromatography

high performance liquid chromatography
thin layer chromatography

ﬂame ionization detector

electron capture detector

ultra-violet

octadecylsilyl

solid-phase extraction
radioimmunoassay

pyridinium chlorochromate
pyridinium dichromate

polyvinyl pyridinium dichromate
pyridine chromium trioxide complex
sodium acetate
bis-(trimethylsilyl)-triﬂuoroacetamide
trimethylsilyl

methoxime trimethylsilyl ether
tert-butyl-dimethylsilyl
pentaﬂuorobenzyl
dehydroepiandrosterone

leukotriene B4

dexamethasone suppression test

xiv

 

CHAPTER 1. INTRODUCTION

1. Mass Spectrometry in Biomedical Research

It may be said that progress in science is limited by the development of analytical
methodology by which the system of interest may be measured. For-instance, in the ﬁeld
of biochemistry, the chemical details of many biological processes have been discovered
through the determination of the molecules involved in these processes. The application
of mass spectrometry to this ﬁeld has contributed greatly to the elucidation and
quantitation of many compounds involved in biochemical systems. In order to use mass
spectrometry for new applications, new methodology must often be developed. This
thesis project involves the development of new methodology for the mass spectrometric

determination of corticosteroids in biological samples.
A. Advantages of Mass Spectrometry
1. Speciﬁcity

Mass spectrometry is a very sensitive and an inherently speciﬁc technique. The
speciﬁcity arises from the fact that the masses of ions produced from a molecule are
measured directly. The basis for many other analytical techniques is the measurement of
a physical pr0perty, such as the absorption of radiation. The structural moiety
responsible for an absorption of radiation may be common to many different molecules,
whereas the masses of the ions produced from a compound, and the relative abundances
of these ions, are very speciﬁc for the compound.

In mass spectrometry, a molecule is introduced into the vapor phase and
bombarded with a relatively high energy (70 eV) beam of electrons. The collision

l

2

between the molecule, M, and the electron imparts so much internal energy to M that in
order to dissipate this energy, the molecule ﬁrst ejects an electron, resulting in the
formation of a molecular ion, M+. Molecular weight information is obtained by
observing the mass-to-charge ratio at which the molecular ion appears. The molecular
ion often fragments further to relieve itself of the remaining excess internal energy. The
fragment ions produced are indicative of the structural features of the molecule. A mass
spectrum is simply a plot of the mass-to-charge (m/z) ratio of the ions against their
relative intensities. Except in the case of some isomeric compounds which produce
similar to identical mass spectra, a mass spectrum is a highly speciﬁc ’ﬁngerprint’ for a
compound, providing a means by which compounds can be positively identiﬁed.

There are several types of mass spectrometers commercially available. Currently
the most popular types of mass spectrometers are the quadrupoles and the double
focusing mass spectrometers.

Quadrupole mass spectrometers consist of an ion source, focusing lenses, a
quadrupole mass analyzer, and a detector. The mass analyzer consists of four cylindrical
rods. DC potentials of opposite sign are applied to the two sets of adjacent rods. An RF
potential is superimposed on the DC potentials. The ratio of the DC and RF potentials
determines the mass resolution of the quadrupole mass analyzer. The magnitudes of the
DC and RF potentials can be adjusted such that only ions of a speciﬁed m/z will have a
stable trajectory through the quadrupole. The other ions are ﬁltered out. The DC and RF
potentials can be varied so that ions of each m/z value are sequentially allowed to pass
through the quadrupole and be detected, resulting in a mass spectrum.

A variation on quadrupole instruments is the triple quadrupole mass spectrometer
(TQMS). The TQMS is identical to a single quadrupole mass spectrometer except that
its mass analyzer consists of three quadrupole mass ﬁlters in tandem. With this type of
instrumental arrangement, two stages of mass analysis can be performed; hence, the

TQMS performs a tandem mass spectrometry experiment, often called by the acronym

3
MS/MS. In MS/MS, the ﬁrst quadrupole can ﬁlter out all the ions produced from a
sample except the selected parent ion. The parent ion is passed into the second
quadrupole, which serves as a high transmission collision cell, where the ion may
undergo collisionally activated dissociation (CAD). The ions produced upon CAD are
called daughter ions. All the daughter ions are transmitted into the third quadrupole,
which is then scanned to obtain the mass analysis of the daughter ions.

Double-focusing mass spectrometers consist of an electrostatic analyzer and a
magnetic analyzer. Ions are accelerated to high energy (3-10 kV) and are directed into
the electrostatic analyzer which acts as an energy ﬁlter by allowing all ions of a given
kinetic energy to pass into the magnetic sector which then transmits the ions to the
detector based on their momenta. The resolution of double-focusing mass spectrometers
may be controlled by adjusting the width of the slits through which the ions must pass.
As the slits at the exit of the ion source and the exit of the electric sector are narrowed,
the energy spread of the ions which are allowed to pass is decreased, resulting in much
higher resolution.

The type of mass spectrometer chosen for a particular analysis depends on the
required speciﬁcity. TQMS adds an extra dimension of selectivity to mass spectrometric
analysis by differentiating ions of a given nominal mass on the basis of their daughter ion
spectra. Double-focusing mass spectrometers operated at high resolution are capable of
differentiating ions of the same nominal mass by more accurately measuring the exact
rn/z values of the ions. Both of these types of mass spectrometric methods have been
used to provide more speciﬁcity for the analysis of analytes in complex biological

matrices

 

2. Sensitivity

Mass spectrometry is useful for biomedical research not only due to its high
speciﬁcity, but also to its high sensitivity which allows trace determinations to be
performed. Often, the analyte to be measured in a biological sample may have a very
low or ’trace’ (ng/g) concentration. A mass spectrum having good signal-to—background
ratios may be obtained from as little as 1 ng of a pure material. In the presence of high
background (chemical "noise") from the sample matrix, more of the analyte must be
present to obtain a mass spectrum having sufﬁcient signal-to-background ratio to provide
interpretable information for structural elucidation.

If greater sensitivity is required for a quantitative mass spectrometric assay, the
technique of selected ion monitoring (SIM) may be used. In this technique only the ion ’
currents from selected mass-to-charge ratios that are indicative of the analyte of interest
are monitored. Since a longer time is spent monitoring these ion currents, the signal-to-
background (s/b) ratios for these ions are greater than they would be if the whole mass
range were scanned. Because the s/b ratios are greater, the detection limit is lowered.
Typical detection limits of EI-MS assays using SIM are in the low ng/ml range (1-100 pg

absolute quantities introduced) ( l ).
B. Some Applications of Mass Spectrometry

1. Qualitative Gas Chromatography-Mass Spectrometry

The use of mass spectrometry has become increasingly important in biomedical
research over the last two to three decades, especially since the development of gas

chromatography-mass spectrometry (GC-MS) in the 1960’s. The coupling of the gas

chromatograph with the mass spectrometer allowed the identiﬁcation of the components

5

of complex mixtures. Even after several isolation procedures, samples of biological
origin are likely to contain many compounds other than the compound of interest to the
investigator. Gas chromatography separates mixture components based on their
differential boiling points and afﬁnity for the stationary phase of the chromatographic
column. Ideally, the components enter the mass spectrometer sequentially, separate from
all interfering substances which were isolated along with the component(s) of interest. In
this way, a mass spectrum representing the pure component may be gathered. Mass
spectrometry has been used for the identiﬁcation of precursors and metabolites of
biochemical pathways. For instance, leukotrienes (2) and their metabolites (3) have been
both conﬁrmed and discovered through the use of mass spectromety. Other examples of
biologically important molecules identiﬁed by mass spectrometry include metabolites of
biogenic amines (4), peptides (5), and xenobiotically modiﬁed analogs of nucleic acids
(6).

Mass spectrometry is also used extensively in drug metabolism studies. The main
objectives of drug metabolism studies are to identify the pathways by which drugs are
modiﬁed in the body and to determine the quantitative importance of each intermediate
and pathway (7). GC-MS is particularly suited for drug metabolism studies because it
allows reliable determination of many similar compounds within a single analysis. Drug
metabolism studies by less speciﬁc methods are often unsuccessful because the similarity

of a drug and its metabolites makes differentiation of the compounds difﬁcult.

2. Quantitative Mass Spectrometry

Used as a quantitative technique, mass spectrometry has, in addition to high
sensitivity, another important advantage over many other quantitative methods in that
isotopically-labelled internal standards can be used. The main advantages of using

isotopically-labelled internal standards is the improved accuracy and precision of the

6
assay because the behavior of the internal standard is identical to that of the analyte
throughout isolation and analysis procedures (8).

Some examples of the uses of quantitative mass spectrometry are in the
determination of the pharmacokinetics and bioavailability of drugs. Bioavailability,
deﬁned as the amount of administered drug that becomes available for pharmacological
effect, differs considerably depending on the formulation of the drug. Bioavailability is
best determined by measuring the plasma levels of the drug. The complexity of the
sample and the low concentration of the drug warrant the use of mass spectrometry (9).
Drug pharmacokinetics, the rate of metabolism and/or disposition of a drug, also can be
studied using mass spectrometry (7). The circulating concentrations of drugs and/or their
metabolites are monitored over time. Such information is important for determining
proper dosage. Also, drugs may be administered and monitored to test and study a

biological function, for instance, enzyme induction or endocrine function.

3. Mass Spectrometry as a Reference Method

Another function of mass spectometry in biomedical research is to serve as a
standard against which the reliability and accuracy of other new methods, such as
immunoassays, may be tested (10). Immunoassays are based on the speciﬁc binding of
an antigen, which is the analyte of interest, to an antibody. Antibodies are proteins that
are formed as an immunologic response to a foreign substance. To obtain the antibodies
for an immunoassay, either an animal or a cell culture is exposed to the substance for
which the immunoassay is to be developed. Antibodies are produced by the animal or
cell culture as a response to this substance. If a cell culture is used to produce the
antibodies, they are termed monoclonal antibodies because they result from a single kind
of cell. After isolation and puriﬁcation of the antibodies, an assay for the antigen may be

developed based on competitive binding of the antigen and a labelled analog of the

7
antigen. For instance, in radioimmunoassay a known amount of radiolabelled antigen is
incubated with an amount of antibody. This preparation is then re-incubated with the
sample which contains the unlabelled antigen to be assayed. A portion of radiolabelled
antigen will be displaced by the unlabelled antigen present in the sample. The antigen-
antibody complex is then separated from the non-bound matrix. The amount of
radiolabelled antigen displaced into the non-bound matrix is proportional to the amount
of unlabelled antigen that was present in the sample.

The reliability of RIA and other types of imrnunoassays is dependent upon the
speciﬁcity of the antibody. Interaction of the antibody with components of the sample
matrix that are similar to the analyte, such as its metabolites or isomeric compounds or
analogs, is called cross-reactivity. Cross-reactivity obviously leads to false results.
However, immunoassays have the advantages of very high sensitivity, simple or no
sample preparation. Also, analyses can be done on very small amounts of sample, and
once a speciﬁc antibody is raised, multiple samples can be analyzed very rapidly using
devices such as scintillation spectrometers. Because of these advantages, it is desirable
to be able to use these methods provided they can be validated. Mass spectrometry is
often used to validate immunoassay methods. Some recent examples include
comparisons of SIM and RIA for the measurement of l7l3-estradiol (11) and the

quantitation of 5-hydroxy eicosatetraenoic acid (12).

11. Conventional Mass Spectrometry for Corticosteroids Analysis

Mass spectrometry has been used for the structural elucidation and quantitative
determination of many classes of biological molecules. This dissertation is concerned
with the deve10pment of methodology for the determination of corticosteroids by mass
spectrometry. This section contains background information on the corticosteroids and

the conventional methods available for their determination.

 

A. Corticosteroids

Corticosteroids are a class of steroid hormones secreted by the adrenal gland.
These steroids have 21 carbon atoms and a 4-ene-3-ketone moiety. The basic structure of
a steroid is shown in Figure 1.1. The types of steroids differ by the number of carbons
within the basic steroid skeleton. Nomenclature rules are also summarized in Figure 1.1.
The corticosteroids are responsible for regulating many biological phenomena including
carbohydrate, protein, and lipid metabolism, and water and electrolyte balance.
Corticosteroids also inﬂuence the functions of the kidney, skeletal muscle, the nervous
system, and the cardiovascular system. The corticosteroids can be further categorized as
mineralocorticoids, which are responsible primarily for maintaining electrolyte balance,
and glucocorticoids, which are primarily involved in carbohydrate metabolism, although
there is some overlap of activity between these two categories. The glucocorticoids are
also potent anti-inﬂammatory agents. In addition to cortisol (4-pregnen-11j3, 17a ,21-
triol-3, 20-dione), which is the most potent naturally occurring glucocorticoid, synthetic
analogs of cortisol having greater potency have been developed for clinical use. For
instance, dexamethasone (90t-F, 16a—CH3, 1-ene analog of cortisol) has 25 times the anti-
inflammatory effect of cortisol and no effect on electrolyte metabolism. The isolation,
structural elucidation, and economical synthesis of corticosteroids has made the clinical
use of these compounds possible.

Identiﬁcations of the structures of corticosteroids were postulated based on
observation of physical properties such as ultra-violet absorption, molecular rotation, and
infra-red absorption, and chemical reactivity toward oxidation and reduction or organic
reagents selective for speciﬁc types of functional groups. Conﬁrmations of the
postulated structures were done by comparing the properties of the unknown with the
properties of a synthesized version of the postulated structure. Conﬁrmation of

postulated structures was made much simpler with the development of GC—MS.

 

21 22 24 26
18 20 3 25
12
17
11 13 16
19
1
14 ’15
2 8
3 7
5
4 6

A = androstane, C19 skeleton
P = pregnane, C21 skeleton
C = cholestane, C27 skeleton

ene = double bond

01 = hydroxyl group

one = carbonyl group

or = substituents extending into the plane of the page
B = substituents extending out of the plane of the page

Figure 1.1. Numbering system for steroid carbon skeleton.

 

10

After the structures of the corticosteroids had been determined, there developed
more concern for quantitiative measurement of corticosteroids. The quantitative
determination of the naturally occurring corticosteroids is done primarily to assess the
function of the adrenal cortex. Quantiﬁcation of the synthetic analogs used as drugs is
done primarily to assess proper therapeutic levels. A variety of methods are available for
the quantiﬁcation of both the endogenous and synthetic corticosteroids. These include
colorimetric, ﬂuorimetric, and gas and liquid chomatographic, as well as mass
specuometric methods. Colorimetric and ﬂuorimetric methods have been in use the
longest, but due to the prevalence of non-speciﬁc interferences in these types of methods
and the development of more speciﬁc chromatographic methods, more laboratories are
using GC and LC based methods. Chromatographic based methods have greater
speciﬁcity than the colorimetric and ﬂuorimetric procedures because identiﬁcation is
based not only on detector response of separated components, but also on retention times
of the separated components. Chromatographic methods also have the advantage that
identiﬁcation and quantiﬁcation of multiple sample components may be accomplished
within a single analysis. Although preliminary identiﬁcation of chromatographic peaks
can be made based on retention time data, conﬁmation of peak identity must often be
done with GC-MS, especially when the component(s) in question is minor or
incompletely resolved from other components. It is possible to make positive
identiﬁcations as well as quantitative detemrinations of both co—eluting components and

minor components using GC-MS due to the speciﬁcity of the mass spectrometer.

B. Derivatization of Corticosteroids for Analysis by GC-MS

In order to determine corticosteroids by GC-MS, the polar groups on the molecule

must be modiﬁed to render these groups non-polar, allowing the compound to vaporize

at a temperature lower than that at which the compound would decompose. The presence

 

1 l
of a Nor-hydroxy on the C-17 sidechain of corticosteroids results in thermal lability (13).
Derivatization stabilizes reactive structural arrangements and prevents thermal
decompostion. The vapor phase properties of cortisteroids are also improved by
derivatization of the polar groups, resulting in good gas chromatographic behavior.

Criteria for a good derivative include the following (14):

l) completeness of derivative formation

2) selectivity of derivative formation

3) ease and reliability of derivative formation

4) stability of the derivative that is formed

5) GC retention time and chomatographic quality (symmetry) of the peaks

6) the stability and inertness of the derivative during GC

Also, when SIM is used, it is important to employ a derivative that provides high
abundance of the molecular ion or other structurally-informative ions in the mass
spectrum.

Different derivatives have been developed to try to meet these criteria; however,
the most commonly used derivative for general steroid analysis is the methoxime (MO) -
trimethylsilyl (TMS) ether (MO-TMS). Reaction with methoxyamine hydrochloride
converts ketones to methoxirnes. In this protected form, the ketones will not enolize
during the subsequent reaction with a reagent such as bis(trimethylsilyl)-
triﬂuoroacetamide (BSTFA) which converts the hydroxy groups to TMS ethers. MO-
TMS derivatives have a number of advantages and disadvantages. The mass spectra of
MO-TMS derivatives of steroids are characterized by losses of 90 amu (TMSOH) and 31
amu (OCH3 radical). These characteristic losses aid in interpreting the spectrum making
it relatively simple. The abundance of the molecular ion and the high mass ions is

relatively low, however, decreasing the sensitivity available for SIM assays. Because

l 2
MO-TMS formation is a two step process, a mixture of products can occur. Also,
because of the steric hindrance of the 17a-hydroxy of corticosteroids, longer reaction
conditions are required to fully derivatize these steroids. The corticosteroid drug,
dexamethasone, possesses a 16a-methyl group which makes MO-TMS formation
particularly difﬁcult. Three hours are required for methoxime formation and six hours
for TMS ether formation at elevated temperatures (15).

The trimethylsilyl ether (TMS)-enol-TMS ether (16) has been found to be most
useful for the analysis of adrenocortical steroids. Enolization of the C—20 ketone
followed by TMS ether formation allows a one step reaction for the analysis of
corticosteroids having different sidechain types. This reaction takes less time than MO-
TMS formation.

Some other derivatives available for corticosteroids include various types of silyl
ethers such as the t-butyl dimethyl silyl (T BDMS) ether, and various cylic derivatives
such as the methyl boronate (17), dimethylsiliconide (18), and the diethylhydrogensilyl
cyclic diethylsilylene (19) derivatives, which are speciﬁc for 1,2- or 1,3-diols. The
TBDMS derivative has an abundant high mass fragment in its mass spectrum which is
advantageous for SIM assays. However, this type of derivative is not really practical for
corticosteroids because the molecular weight of the derivatized compound will be very
high if the multiple hydroxy groups are derivatized. Also, MO formation is still required
and the TBDMS group is even more bulky than a TMS group making derivatization of
sterically hindered groups even more difﬁcult. The methyl boronate derivatives afford
abundant molecular ions of relatively low mass. Also the GC retention times of these
corticosteroid speciﬁc derivatives are relatively short compared to MO-TMS derivatives.
However, subsequent trimethylsilylation of unprotected hydroxy groups can lead to
partial cleavage of the boronate (20).

One disadvantage common to all these derivatives is that they are all subject to

hydrolysis. Therefore, storage conditions must be moisture free.

13

C. Other Mass Spectrometric Techniques for Corticosteroids Analysis

The determination of very low levels of speciﬁc corticosteroids within a complex
biological matrix sometimes requires even higher speciﬁcity and sensitivity than can be
obtained through the use of GC-MS with SIM. Some instrumental methods have been
used to increase the sensitivity and/or selectivity for corticosteroid determination. These
include high resolution selected ion monitoring (HR-SIM) and alternate ionization
techniques such as positive chemical ionization (CI), negative chemical ionization
(NCI), and electron-capture negative ionization (ECNI).

High resolution mass spectrometry (HRMS) is not practical for many
investigators due to the expense of a high resolution mass spectrometer. However, there
are some groups which have made use of this technique and achieved high sensitivity due
to the increased speciﬁcity of the technique. For example, the selectivity for 1713-
estradiol in plasma is shown to be considerably increased through the use of HR-SIM
(21). Also the improved selectivity of HR-SIM allowed 2 pg/ml level detection of
testosterone in the saliva of female subjects (22).

Positive CI may offer a sensitivity advantage over electron-impact ionization in
selected ion monitoring assays since the relative intensity of the molecular species to be
monitored is increased. Upon chemical ionization, little energy is imparted to the
molecule, therefore little or no fragmentation of the molecular species (M+H)+ occurs.
For certain molecules, NCI and ECNI offer both sensitivity and selectivity advantages.

This dissertation emphasizes ECNI-MS as a technique for enhancing the

selectivity and sensitivity of assays for corticosteroids.

14

III. Electron Capture Negative Ion Mass Spectrometry
A. History and Theory of Electron-Capture Negative Ionization

Much of the early investigation of electron-capture negative ionization (ECNI) of
organic compounds was done in the laboratory of von Ardenne (23, 24). His experiments
were conducted with an argon discharge source in which a primary electron beam was
slowed by collisions with a high pressure of argon atoms resulting in the formation of a
low energy plasma of electrons and argon ions. Molecules having positive electron
afﬁnities are able to absorb a low energy or ’thermal’ electron resulting in the formation
of a molecular anion. After absorption of the electron, the molecular anion is in an
excited state, having internal energy at least equal to its electron afﬁnity. The excess 9
internal energy of the excited molecular anion may be dissipated through collisions with
inert species or by radiative emission. If the molecule anion has a sufﬁcient number of
internal degrees of freedom over which to dissipate the excess energy, the anion will be
sufﬁciently long-lived for stabilization to occur. In small molecules with few internal
degrees of freedom, ejection of the electron is more likely to occur before stabilization
can occur.

Molecular anions are formed by electron capture of electrons having energies in
the 0 to 2 eV range. If the energy of the captured electron is larger, dissociative electron
capture can occur. In a collision with a higher energy electron (0-15 eV) more internal
energy must be dissipated in order to stabilize the anion. The internal energy absorbed
upon electron capture is dissipated by breaking one or more bonds, resulting in a
stabilized fragment anion. The "driving force" for an electron capture reaction may be
the formation a resonance stabilized fragment ion, such as (M-H)- (52).

The argon discharge source used by von Ardenne was an in-house modiﬁcation.

When it was found that the high pressure source used for positive CI could also be used

1 5
for ECNI more laboratories began to investigate this mode of ionization. The mechanism
of the formation of the low energy electron plasma under CI conditions is shown by the

equation
CH4 + e' (100 eV) --> CH4+ + e-(t) + e' (<100 eV)

where e-(t) represents an electron having only thermal energy. Upon each ionizing
collsion, the primary electron loses 30 eV (25). After several collsions, the primary
electrons are thermalized. In the high pressure CI source, a plasma of thermalized
electrons is readily produced.

Dougherty (26) was the ﬁrst to suggest the use of the conventional CI source for
obtaining electron-capture negative ionization spectra. He used this technique to obtain
ECNI spectra of chlorinated environmental contaminants (49). Hunt et al (27) were
among the ﬁrst groups to anticipate the analytical potential of ECNI for biological
applications. A detection limit of 25 fg (s/n = 4) of a pure standard of the
pentaﬂuorobenzoyl derivative of amphetamine was obtained by ECNI-MS with SIM.
Since then, applications of ECNI have increased. It is commonly used for detection of
many halogenated environmental contaminants as well as for the detection of compounds
of biological interest such as prostaglandins and leukotrienes derivatized as
pentaﬂuorobenzyl esters (28). Methods have been developed for many other compounds

such as neurotransmitters (29) and drugs (30).

B. Limitations of Electron Capture Negative Ionization

Limitations of ECNI include the acute dependence of the extent of ionization on
the ionization parameters, such as temperature and pressure, as well as the limited
applicability of the technique. Because the mean energy of the electrons in the CI source

is highly dependent on the temperature and pressure of the CI gas, and the extent of

1 6

ionization and fragmentation of an analyte is dependent upon the energy of the electrons,
it is important to carefully regulate both temperature and pressure. This is particularly
important when conducting quantitative analyses. Slight pressure changes from run to
run decrease precision. Even pressure changes during a single analysis can decrease
precision if a structural analog is used as internal standard. Better precision is maintained
when isotopically labelled internal standards are used since any ﬂucuations in conditions
are experienced simultaneously by the analyte and internal standard as they co-elute.

The purity of modifying gas is also of utmost importance for ECNI. The presence
of trace amounts of 02 or H20 within the mass spectrometer can result in ion/molecule
reactions with 02, 02', 0-, or OH- which may be more facile than the electron/molecule
reactions (31). For example, trace amounts of oxygen within the modifying gas have
been demonstrated to completely change the ECNI spectrum of the plant hormone methyl
abscissate (32). Normally, the CH4-ECNI mass spectrum of methyl abscissate contains
m/z 278 (M') as the base peak and a few minor fragment ions. If a trace of oxygen is
introduced, an ion/molecule reaction occurs to form an (M+02)- adduct ion which then
undergoes unique fragmentation pathways. In this case, the spectrum consists of the ions
resulting from electron capture as well as the ions resulting from the ion/molecule
reaction, but the peaks dominating the spectrum are fragments resulting from the
ion/molecule reaction.

The cleanliness of the ionization chamber may also affect ECNI spectra. Halogen
impurities on the chamber walls can cause changes in spectra such as the appearance of
M+Cl adducts (33). Such surface reactions with species adsorbed to the source walls can
also distort the chromatographic peak shape producing broadening and tailing of peaks
(34, 35).

The care that must be taken to control ion source ionization parameters makes the
use of ECNI more difﬁcult than EI ionization, which has become rather routine. Still,

the reproducibility of ECNI can be reasonably good as long as care is taken concerning

 

 

l 7
the factors mentioned above. Stemmler and Hites (36) published a study of the
parameters which affect the reproducibility of ECNI spectra which indicated thatthese
spectra are more reproducible than has been commonly perceived. They found that with
careful control of instrumental parameters, such as ion source pressure, emission current,
ion source temperature, and ion focus potential, reproducible ECNI spectra could be
obtained throughout a one-year period. An inter-laboratory study by the same authors
supported these conclusions (37). Although exact duplication of the relative intensities
of the ions in the ECNI spectra would require assessment and control of the lens and ion
source parameters, and of the presence of impurities, the general features of the ECNI

spectra of several compounds were reproduced on different instruments.
C. Advantages of ECNI

The two notable advantages of ECNI are the increase in sensitivity and the

selectivity of the ionization.
1. Sensitivity

For certain types of compounds, electron capture negative ionization can be
orders of magnitude more efﬁcient than positive ionization. For instance, the total ion
current obtained from 2 n g of the PFB-TBDMS derivative of LTB4 is 50 times greater by
ECNI than by El ionization and 20 times greater than by positive CI (38). Hunt used
pulsed positive negative ion chemical ionization mass spectrometry to demonstrate that
negative ion currents were 100 to 1000 times greater than the corresponding positive ion
currents for compounds such as anthraquinone, phthalic anhydride, and trinitrotoluene
(23). The greater negative ion currents can be explained on the basis of the rate constants

for positive ion formation and negative ion formation by electron capture. As an

l 8

approximation, the ratio of negative ion current (d(M’)/dt) to positive ion current is
proportional to the ratio of the rate constant for an electron/molecule reaction to the rate
constant of a positive reagent ion/molecule reaction (50). The maximum values for the
rate constant of reactions between reagent ions and molecules are on the order of 3-4 x
10-9 cm3 mol-l sec-1. In contrast, electron/molecule reaction rate constants are on the
order of 4 x 10'7 cm3 mol-l sec-1 (51). The higher mobility of the electrons in the
chemical ionization plasma may account for the difference in rate constants. Based on
these values, negative ion current could be two orders of magnitude greater than positive
ion current.

Even if the actual ionization of the analyte is not enhanced over its positive
ionization efﬁciency, sensitivity of an assay may still be enhanced by ECNI. The
effective signal-to-background ratio is normally enhanced for a compound with good
electron capture response because the ionization of most sample matrix components will
have poor or no electron capture response. Also, because ECNI is a ’soft’ ionization
technique, little fragmentation occurs, allowing the ion current to be concentrated in a

few ionic species. This effect maximizes the sensitivity available with SIM.

2. Selectivity

ECNI is called a selective mode of ionization because most compounds do not
give an appreciable response under ECNI conditions. Among the few classes of
compounds which have good EC properties are those that have a halogen or a nitro group
attached to a pi-bonded carbon system, a highly conjugated carbon system, or a weak
electron capturing group, such as a carbonyl, attached to a conjugated system (30). For
compounds falling into these classes, the fact that few other compounds respond under
these conditions can be used to advantage for the determination of these compounds in

complex sample matrices. For example, Dougherty shows that dieldrin at 0.1 ppm

1 9
concentration in chicken tissue extract is easily identiﬁed by ECNI. The most prominent
peak in the spectrum represents the molecular ion of the chlorinated pesticide dieldrin.
However, under positive ionization conditions interferences from co-extracted lipids
dominate the spectrum obscuring identiﬁcation of the dieldrin (39). Other examples of
compounds which have been detected in complex matrices by ECNI are the
benzodiazepines. This class of drugs contains a halogen atom on a conjugated ring
system and, thus, has good electron capture response. The ECNI assay developed for
ﬂuazepam, a particularly potent benzodiazepine hypnotic drug, was the ﬁrst chemically
based method that was sensitive enough to detect ﬂuazepam to provide a

pharmacokinetic proﬁle in man (40).

D. Derivatization to Enhance ECNI Response

Because determination of compounds of biological interest are normally found in
complex matrices, it would be an advantage to be able to selectively ionize the '
component(s) of interest without ionizing the considerable matrix components. Because
most compounds of biological interest are neither highly conjugated nor halogenated, the
approach of derivatization of functional groups with ﬂuorinated moieties has been
applied to enhance sensitivity and selectivity of detection of analytes within biological
matrices. Although the electron capture responses of the halogens increase from ﬂuorine
to iodine, the use of ﬂuorine is the best compromise for a derivative because the increase
in retention time is least when ﬂuorines are used (41). Some examples of derivatives
include pentaﬂuorobenzyl esters, perﬂuoroacyls, bis(triﬂuoromethy1) benzoyls, and
pentaﬂuorobenzyloximes. Assays for fatty acids, amino acids, steroids, and other types
of molecules based on derivatization with the appropriate reagent have been published.
Recently, a 2 fg limit of detection (LOD) was reported for PFB ester-dimethylethylsilyl

ether derivatives of bile acids (42).

2 0

Despite the success with PFB esters, limits of detection for many ECNI assays
based on derivatization represent only modest gains over those of the corresponding EI-
MS based methods. The expected gains in sensitivity and selectivity may not be realized
in many cases without extensive analyte puriﬁcation because of two reasons. The ﬁrst is
that these types of derivatization reactions are not very selective. Any concomitant in the
sample matrix having a reactive functional group will produce a highly responding
derivative. For instance, the perﬂuoroacyl derivatives can be formed with a variety of
functional groups including hydroxyls, amines, phenols, and the enol form of ketones.
Thus, the effective s/b is not increased without extensive puriﬁcation of the analyte.

The second reason for lower than expected sensitivity in the use of derivatives is
that many derivatives used to enhance electron capture result in the production of a mass
spectrum that is more representative of the derivatization moiety than of the intact
derivatized molecule. An example of this type of mass spectrum is shown in Figure 1.2,
which was obtained from the PFB oxime-TMS ether of testosterone. Low abundance of
the molecular ion or high mass fragments will decrease the sensitivity that can be realized
in a quantitative assay based on SIM of these ions. The most intense ions which are
unrepresentative of the intact derivatized analyte would not be suitable for a SIM assay

because every derivatized molecule in the matrix will produce these same ions.

100*

50‘

relative abundance

 

100

Figure 1.2.

21

 

 

 

 

181
OSHCHag
C5F5CH2EO-N
m/z 181 <3
167 197
jl .l 1 11. .1 1‘ ' L LJ-l . 1
200 300 1.00 500 600
m/z

ECNI mass spectrum of PFB-oxime, TMS-ether of testosterone.
The base peak at, m/z 181, represents the pentaﬂuorobenzyl anion.
This ﬁgure was taken from S. J. Gaskell, "Analysis of Steroids" in
Methods of Biochemical Analysis, vol. 29, D. Glick, ed., Wiley
Interscience, New York, pp. 385-434, 1983.

22

Corticosteroid determinations by ECNI have been very limited because the
derivatization of this class of steroids with electrophilic moieties has been unsatisfactory
due to the multiple hydroxy and ketone groups present in'these types of molecules. For
example, the formation of perﬂuoroacyl derivatives of corticosteroids, such as the
heptaﬂuorobutyryl, results in complex mixtures of products due to enolization of ketone
groups, making these types of derivatives unsuitable for quantitative analyses. Another
group of electron capturing derivatives for corticosteroids, the electron capturing
boronates, such as the 3,5-bis(triﬂuoromethyl)benzene boronate, were investigated by
Poole et a1 (20). Multiple products were obtained upon derivatization of cortisol and
other related corticosteroids with 3,5-bis(triﬂuoromethyl)benzene boronic acid. Also,
subsequent trimethylsilylation of the remaining hydroxy groups resulted in partial
cleavage of the boronate. Therefore, these derivatives were not suitable for quantitative
determination of corticosteroids.

An ongoing goal of some researchers in analytical methodology is to develop new
derivatives which will meet the needs for speciﬁc types of determinations, such as
derviatives which may produce stable molecular anions and/or high mass fragments
under ECNI conditions. A goal in this laboratory is to deve10p alternative approaches to
analyte modiﬁcation which will be more selective based on chemistry, and which will
allow the advantages of sensitivity and selectivity of ECNI to be realized without the

requirement of extensive analyte puriﬁcation.

IV. An Alternative Approach for Enhancing ECNI Response of Corticosteroids

Of the derivatives commonly used to increase electron capture response, none
seem to be suitable for the derivatization of corticosteroids. Therefore, an alternative
approach, involving the chemical oxidation of corticosteroids, is being studied in this

laboratory as a means for enhancing the electron capture response of these compounds.

 

 

23

A. Previous work

It was ﬁrst noted by Lovelock (43) that steroids containing cab-unsaturated
ketones gave unexpectedly high ECD responses. The magnitude of the response of the
steroid containing the 0t,B-unsaturated ketone increased depending on the presence and
position of other ketones or double bonds within the molecule. Poole (20) conﬁrmed that
certain moieites, in speciﬁc geometries (especically the C-11 and C-17 ketones in
combination with the 4-ene-3-ketone) on the steroid nucleus results in high EC response.
Some of these structures are shown in Figure 1.3. The natural presence of electron
capture properties in certain steroids has been used to advantage in some cases for the
sensitive detection of these compounds in complex samples. For instance, an assay was
developed to detect low levels of the steroid drug melengestrol acetate in meat (44).
Also, ecdysteroids, a class of insect hormones that contains a 7-ene-6-one moiety, have
been detected in insects at low levels using the electron capture detector (45).

The use of chemical oxidation to incorporate electron capturing moieties into a
steroid was found to be useful for the detection of dehydroepiandrosterone (DHEA, 5—
androsten-3B-ol-l7-one) by de Jong and van der Molen (46). It was found that the CrO3
oxidation product of DHEA, in addition to having good gas chromatographic properties,
was also highly electrophilic due to the 4-ene-3,6-dione moiety incorporated upon
oxidation. This group developed an assay for DHEA in plasma based on oxidation of the
extract, followed by GC-ECD. Speciﬁcity for DHEA was ensured by separation of
DHEA from other 5-ene-3-ols prior to oxidation. The electron capture detector and
ECNI operate on the same principle. In the ECD, a radioactive (63Ni or 3H) foil emits [3
particles which ionize the carrier and make-up gas (nitrogen or argon) resulting in a
plasma of thermal electrons that are available for electron capture by compounds as they
elute from the GC and pass through the detector along with the make-up gas. Most
compounds having ECD response will also give a response to ECNI-MS.

24

 

Figure 1.3. Structures of steroids having high electron capture responses; I
and 11 result from the oxidation of the naturally occuring
compounds, cholesterol and 6B-hydroxycortisol. III, IV, and V
result from the oxidation of the corticosteroid drugs prednisolone,
ﬂudrocortisone, and dexamethasone, respectively.

25

Previous work in this laboratory has shown the feasibility of using oxidation to
incorporate C-11 and C-17 ketone moieties and enhance the EC properities of
corticosteroids drugs (47). Oxidation of dexamethasone, ﬂudrocortisone,
ﬂuormetholone, prednisolone, and ﬂumethasone with chromium trioxide di-pyridine
complex results in 11,17-keto analogs of these drugs that all have very high ECNI-MS
responses.

The feasibility of this approach as well as the advantages of sensitivity and
selectivity were demonstrated by an assay for dexamethasone in horse urine (48). Ng/ml
quantities were detected by ECNI-MS following oxidation of the solvent extract of the
urine. Dexamethasone was oxidized to a highly responding structure, whereas the other
co-extracted steroids were not. Most endogenous steroids, lacking any double bonds, are
not oxidized to highly responding products. Only the corticosteroids, progesterones and
some androgens have the required (LB-unsaturated ketone. And only a small percentage
of the androgens and progesterones have an oxygen at the C-11 position. Fewer still
have a C-l couble bond or are oxygenated at the 06 position. Thus oxidation provides a
way to selectively enhance the responses of only a few types of steroids, namely those
having the 4-ene-3-one-6-oxy moiety, the 1,4-diene-3-one-11-oxy moiety, or the 5-ene-

3-ol moiety.

26

B. Majors goals of this research

The three major goals of the research project described in this dissertation are: l)
the optimization of the oxidation methodology for the quantitative conversion of
corticosteroids to their electrophilic oxidized analogs; 2) to ascertain the extent to which
the enhanced selectivity of the oxidation and the ionization will allow a less selective
method of sample introduction, such as direct probe rather than GC; and 3) to apply this
oxidation methodology a) to the determination of dexamethasone and its major
metabolite, 6B-hydroxydexamethasone, in plasma from patients undergoing the
dexamethasone suppression test, and b) to the determination of 6B—hydroxycortisol in

urine for use as an index of enzyme induction.

10.

ll.

12.

13.
14.

15.
16.
17.

18.

REFERENCES

A. L. Burlingame, T. A. Baille, and P. J. Derrick, Anal. Chem, 58, 165R-211R,
1986.

R. C. Murphy, S. Hammerstrom, and B. Samuelsson, Proc. Natl. Acad. Sci.
USA, 76, 4275, 1979.

D. O. Stene and R. C. Murphy, J. Biol. Chem, 263, 2773-2778, 1988.
S. Higa and S. P. Markey, Anal. Biochem, 149, 492-500, 1985.

F. De Angelis, M. Botta, S. Ceccarelli, and R. Nicoletti, Biochem. J ., 236, 609-
612, 1986.

R. K. Mitchum, J. P. Freeman, f. A. Belank, and F. F. Kadlubar, in Mass
Spectrometry in the Health and Life Sciences, A. L. Burlin game and N.
Castagnoli, Jr., eds., Elsevier, Amsterdam, pp. 547-580, 1985.

A. Goldstein, L. Aronow, and S. M. Kalman, Principles of Drug Action: The
Basis of Pharmacology, John Wiley and Sons, New York, 2nd edition, p. 230,
1974.

J. Sjovall and M. Axelson, Vit. and Hormones, 39, 31-144, 1982.

W. A. Garland and M. L. Powell, J. Chromatogr. Sci, 19, 392-434, 1981.

H. Aldercreutz, "Biomedical Application of the Mass Spectrometry of Steroid
Hormones" in A. Quayle, ed., Advances in Mass Spectrometry, vol . 88, London,
Heyden, 1165-1179, 1980.

V. Papadopoulos, S. Carreau, E. Szerman-Joly, M. A. Drosdowsky, L. Dehennin,
and R. Scholler, J. Steroid Biochem, 24, 1211-1216, 1986.

J. Y. Wescott, S. Chang, M. Balazy, D. O. Stene, P. Pradelles, J. Maclouf, N. F.
Voeikei, and R. C. Murphy, Prostaglandins, 32, 857-873, 1986.

H. Gottfried, Steroids, 5, 385-397, 1965.

T. A. Baillie, C. J. W. Brooks, and B. S. Middleditch, Anal. Chem, 44, 30-37,
1972.

J-P. Thenot and E. C. Homing, Anal. Letters, 5, 21-33, 1972.
EM. Chambaz, G. Defaye, and C. Madani, Anal. Chem. 45, 1090-1098, 1973.

C. J. W. Brooks, W. J. Cole, T. D. V. Lawrie, J. MacLachlan, J. H. Borthwick,
and G. M. Barrett, J. Steroid Biochem, 19, 189-201, 1983.

R. W. Kelly, Steroids, I3, 507, 1969.

27

19.

20.

21.

22.

23.

24.

25.

26.
27.
28.

29.

30.

31.

32.

33.

34.

35.
36.

28

M. Ishibashi, M Itoh, K Yamashita, H Miyazaki, and H Nakata, Chem. Pharm.
Bull., 34, 3298-3305, 1986.

C. F. Poole, W. F. Sye, S. Singhawangcha, and E. D. Morgan, Lipids, 15, 734-
744, 1980.

S. J. Gaskell, C. J. Porter, and B. W. Green, Biomed. Mass Spectrom, 12, 139-
141, 1985.

V. J. Gould, A. O. Turkes, and S. J. Gaskell, J. Steroid Biochem, 24, 563-567,
1986. ’

D. F. Hunt, G.C. Stafford, Jr., F. W. Crow, and J. W. Russell, Anal. Chem, 48,
2098-2105, 1976.

H. Budzikiewicz, Angew. Chem. Int. Ed. Engl., 20, 624-637, 1981.

C. E. Klots, "Energy Deposition Processes" in Fundamental Processes in
Radiation Chemistry, P. Ausloos, ed., John Wiley and Sons, New York, p. 40,
1968.

R. C. Dougherty and C. R. Weisenberger, J. Am. Chem. Soc., 90, 6570, 1968.
D. F. Hunt and F. W. Crow, Anal. Chem, 50, 1781-1784, 1978.

I. A. Blair, S. E. Barrow, K. A. Waddell, and P. J. Lewis, Prostaglandins, 23,
579-589, 1983.

K. F. Faull and 0. Beck, "Quantiﬁcation of Neurotransmitters and Related
Compounds by Electron Capture Chemical Ionization GC-MS" in Mass
Spectrometry in Biomedical Research, 8. J. Gaskell, ed., John Wiley and Sons
Ltd., Chichester, pp. 403-421, 1986.

W. A. Garland and M. P. Barbalas, "Determination of Drugs in Biological
Matrices Using Gas Chromatography-Electron Capture Negative Chemical
Ionization Mass Spectrometry" in Mass Spectrometry in Biomedical Research, S.
J. Gaskell, ed., John Wiley and Sons Ltd., Chichester, pp. 379-401, 1986.

D. Stock] and H. Budzikiewicz, Org. Mass Spectrom, 17, 470-474, 1982.

T. G. Heath, D. A. Gage, J. Zeevart, and J. T. Watson, Proceedings of the 37th
Annual Conference on Mass Spectrometry and Allied Topics, Miami Beach, FL,
May 1989.

D. B. Kassel, K. A. Kayganich, J. T. Watson, and J. Allison, Anal. Chem, 60,
911-917, 1988.

L. J. Sears, J. A. Campbell, E. P. Grimsrud, Biomed. Environ. Mass Spectrom,
14, 401-415, 1987.

V. Burger and P. J. Pretorius, J. Chromatogr. Sci, 25, 118-121, 1987.

E. A. Stemmler and R. A. Hites, Biomed. Environ. Mass Spectrom, 15, 659-667,
1988.

 

 

37.

38.

39.
40. A
41.
42.

43.

45.
46.
47.
48.
49.
50.

51.
52.

2 9
E. A. Stemmler, R. A. Hites, B. Arbogast, W. L. Budde, M. L. Deinzer, R. C.
Dougherty, J. W. Eichelberger, R. L. Foltz, C. Grimm, E. P. Grimsrud, C.
Sakashita, and L. J. Sears, Anal. Chem, 60, 781-787, 1988.

M. Dawson, C. M. McGee, P. M. Brooks, J. H. Vine, and T. R. Watson, Biomed.
Environ. Mass Spectrom, 17, 205-211, 1988.

R. c. Dougherty, Anal. Chem, 53, 625A-636A, 1981.
B. J. Miwa and w. A. Garland, Anal. Chem, 53, 793-797, 1981.
CF. Poole and A. Zlatkis, Anal. Chem, 52, 1002A-1016A, 1980.

J. Goto, K Watanabe, H. Miura, T. Nambara, and T. Iida, J. Chromatrogr, 388,
379-387, 1987.

J. E. Lovelock, P. G. Simmonds, and W. J. A. Vandenheuvel, Nature, 197, 249-
251, 1963.

R. A. Davis, K. T. Koshy, M. L. Ogilvie, G. L. Stahl, and W. A. Callahan, J.
Assoc. 01?. Anal. Chem, 56, 103, 1973.

E. D. Morgan and C. F. Poole, J. Chromatogr, 116, 333, 1976.

F. H. de Jong and H. J. van der Molen, J. Endocr, 53, 461-474, 1972.

G. R. Her and J. T. Watson, Biomed. Environ. Mass Spectrom, 13, 57-63, 1986.
G. R. Her and J. T. Watson, Anal. Biochem, 151, 292-298, 1985.

R. C. Dougherty, J.’ Dalton, F. J. Biros, Org. Mass Spectrom, 6, 1171, 1972.

F. H. Field, Proceedings of the 28th Annual Conference on Mass Spectrometry
and Allied Topics, New York, NY, pp. 2-13, May 1980.

L. G. Christopherou, Chem. Rev, 76,409, 1976.
J. G. Dillard, Chem. Rev, 73, 589-643, 1973.

CHAPTER 2. DEVELOPMENT OF OPTIMAL CONDITIONS FOR ISOLATION
AND CHEMICAL OXIDATION OF DEXAMETHAS ONE

1. Introduction

As described in the previous chapter, dexamethasone is a synthetic,
corticosteroid, anti-inﬂammatory drug that, upon oxidation, yields a highly electrophilic
11,17-keto analog. Besides its use as an anti-inﬂammatory drug, dexamethasone has
been found to aid in the diagnosis of Cushing’s Syndrome, a hypercortisolemic condition.
Upon administration of dexamethasone, plasma cortisol levels of healthy people are
suppressed for a 24-hour period. In patients having Cushing’s Syndrome the
administration of dexamethasone has no effect upon the plasma cortisol levels. The
results of this test, coined the dexamethasone suppression test, has served as a basis for
diagnosing Cushing’s Syndrome for many years.

Recently (within the last decade) the dexamethasone suppression test (DST) has
been proposed as a tool for physicians and psychiatrists in the diagnosis of endogenously
depressed psychiatric patients. The DST has been shown to aid in differentiation
between melancholia (endogenous depression) and depression caused by neurotic,
reactive, or characterological factors (1). Because the treatments for these types of
depression differ, it is important to be able to obtain a proper diagnosis. The DST
typically consists of administration of 1 mg of dexamethasone with subsequent
monitoring of plasma cortisol levels at 8, 16, and 24 hours following administration. In
non-endogenously depressed subjects, cortisol levels are suppressed for a 24-hour period,
whereas in endogenously depressed patients, the plasma cortisol levels ’escape’ to pre-
drug levels within three to four hours. Some recent studies have indicated that the
bioavailability of dexamethasone may be a critical factor for determining the results of

the DST (2). Therefore, it appears necessary to measure dexamethasone plasma levels

30

3 1
along with cortisol levels in order to appropriately evaluate DST results.

Typical dexamethasone concentrations in plasma of patients undergoing the DST
diminish from 4 ng/ml to 0.1 ng/ml during the 24-hour period of the test. Due to the low
concentrations of dexamethasone and the complexity of the plasma sample matrix,
sensitive and speciﬁc methodology is needed for accurate determination of the
concentration of dexamethasone in plasma. There are adequately sensitive
radioirnmunoassays available to measure these levels of dexamethasone in plasma (3,4);
however, the possibility of cross-reactivity of the antibody with endogenous compounds
as well as dexamethasone metabolites may make RIA results dubious. HPLC methods
employing UV detection are also available for the determination of dexamethasone in
plasma. The sensitivity of these methods are limited, however, due to interferences from
co-extracted endogenous material. HPLC methods are suitable for measuring plasma
dexamethasone concentrations no lower than 3 to 5 ng/ml (5). Mass spectrometric
methods have also been reported. Some of these reported MS methods are based on the
formation of the tetra-trimethylsilyl ether (TMS) derivative of dexamethasone with
detection by gas chromatography-mass spectrometry with either chemical ionization or
electron impact ionization (6,7). Using selected ion monitoring GC-EI/MS Kasuya et a1
(6) were able to detect an injection of 100 pg of dexamethasone (S/N = 2.5) from 1 ml of
plasma spiked to a concentration of 300 pg/ml. Extensive isolation procedures,
consisting of extraction by C18-bonded phase reverse-phase cartridge followed by
normal phase HPLC puriﬁcation, were necessary to eliminate the plasma matrix
interferences. A method based on direct liquid introduction microbore HPLC-MS has
been reported for the detection of betamethasone in equine urine (9). This drug is
identical to dexamethasone except in the sterochemistry of the 16-methy1 group. The
reported sensitivity of 5 ng/ml would not be adequate for the determination of
dexamethasone in the plasma of patients undergoing the DST.

In addition to the measurement of plasma dexamethasone concentrations, the

32
measurement of 6B-hydroxydexamethasone, the major metabolite of dexamethasone (8),
may also be desirable since there is some indication that 6B-hydroxydexamethasone is
bioactive (10).

A goal of this research is to apply oxidation methodology for the detection of
dexamethasone and 6B-hydroxydexamethasone in plasma at levels expected to be present
in patients undergoing the DST. This chapter describes the optimization of plasma
extraction, extract oxidation, removal of excess oxidation reagent for the detection of
dexamethasone in plasma. Some preliminary work for the application of this

methodology to the detection of 6B-hydroxydexamethasone is also described.

II. Method Development

The oxidation method consists of four parts; plasma extraction, extract oxidation,
removal of excess oxidation reagent, and ﬁnally ECNI-MS analysis. The oxidation of
dexamethasone to its 11,17-keto analog is shown in Figure 2.1. This is the reaction that
serves as the basis for the ECNI-MS assay of dexamethasone. The following sections
describe the development and optimization of experimental conditions and brieﬂy discuss
the results for isolation procedures and chemical oxidation methodology for

dexamethasone and 6B-hydroxydexamethasone.

A. Extraction of Dexamethasone and 6B-Hydroxydexamethasone from Plasma

1. Solvent Extraction vs. Solid Phase Extraction

Solvent extraction was once the most extensively used means by which to isolate

steroids from physiological ﬂuids such as plasma and urine. The development of

 

1

 

dexamethasone

Figure 2.1.

Oxidative conversion
dexamethasone.

33

Oxidation

of dexamethasone

[0

11,17-keto

 

34
reverse-phase, solid-phase extraction technology has provided a more rapid means for the
isolation of steroids from these ﬂuids. Near quantitative recoveries of steroids from urine
through the use of C18 Sep-Pak cartridges have been reported (13). In addition to high
recovery, solid-phase extraction technology allows the possibility of isolating the analyte
from many interfering substances within the sample matrix.

Aliquots of "spiked" plasma were extracted by solvent extraction and C18 solid-
phase extraction to compare the efﬁciency of each method for the recovery of
dexamethasone, 13C6-2H3-dexamethasone (an isotopically labelled analog of
dexamethasone used as an internal standard), and 6B-hydroxydexamethasone from
human plasma. The plasma samples were prepared by adding dexamethasone (16.7 ng),
6B-hydroxydexamethasone (19.4 ng), and 13C6-2H3-dexamethasone (35 ng) to ten ml of
blood bank plasma. Aliquots of 2 ml were extracted by each method. Figure 2.2
illustrates the impact of each of the extraction procedures on the GC-ECNI/MS response
obtained by analysis of these test samples.

From Figure 2.2 it is apparent that the recovery of the dexamethasone is much
more efﬁcient by solid-phase extraction than by solvent extraction with ethyl acetate.
However, the intensity of the ion current from co—extracted matrix components is about
the same by solvent extraction and solid-phase extraction as can be seen by comparing
the ion current intensity numbers on the right border of Figure 2.2a and Figure 2.2b. By
varying the solvent strength of the column washes and the eluate, it would probably be
possible to eliminate some of these co-extracted compounds. This option would not be
possible in the case of solvent extraction.

The solvent extraction method consisted of extraction of the plasma three times
with equal volumes of ethyl acetate. The combined ethyl acetate extracts were then dried
with anhydrous sodium sulfate. The extract was evaporated under N2, reconstituted in

lml of CH2C12, and oxidized with pyridinium chlorochromate according to procedures

 

%100-

80'

40‘

20'

 

35

-*E+04
4.706

 

l

 

 

 

 

A
1'1‘11

10: 00

uI-Ivlu'ulvlv'w

7:00 8:00

1' '1-I . . . .
9:00 11:00 12:00
.*E+04
4.072

 

 

 

0 T

Figure 2.2.

ujulviv

'vleﬁl

7: 00 8:

ul‘lv1v

1., . .
00 9:00 10:00

Y'iIvlvl

' 1
11:00 12:00

time (min)

Reconstructed selected ion current proﬁles from GC-ECNI/MS
analysis of oxidized dexamethasone, 13C6-2H3-dexamethasone,
and 6B-hydroxydexamethasone isolated from plasma by a) solvent
extraction and b) C18 solid-phase extraction. Arrow indicates
retention time of oxidized dexamethasone.

 

 

 

3 6
described in IIB. of this chapter. Conditions for analysis of the sample by GC-ECNI/MS
with SIM are described in Chapter 3.

For the solid phase extraction, two milliliters of the spiked plasma was passed
through a 200 mg C18 solid phase extraction column that had been previously washed
with 5 ml of 8M urea, 5 ml of methanol, and 10 ml of distilled water. The purpose of the
urea wash is to derivatize any free silanol groups on the column packing. The methanol
wash elutes any contaminants sticking to the C18 packing. The water wash removes the
methanol and prepares the column to receive the sample. Following the application of
the plasma sample, the column was washed with 4 ml of distilled water to remove polar
components of the plasma sample. To elute the dexamethasone and 6B-
hydroxydexamethasone, four milliliters of 70% methanol-30% hexane was applied to the
column and the eluate collected. The eluate was evaporated under N2 and reconstituted

in 1 ml CH2C12 for oxidation followed by GC-ECNI/MS analysis.

2. Solid—Phase, Reverse-Phase (C18) Extraction

Two different solid-phase extraction methods were investigated for the isolation
of dexamethasone and 6B-hydroxydexamethasone in human plasma.

The method of Plezia and Berens (14) involved conditioning a C18 SPE column
in tandem with a silica SPE column with 2 ml methanol and then 2 ml water. After
passing the sample through the extraction system, the silica phase was removed and the
C18 column was washed with 3 ml 95% aqueous methanol and 4 ml of CHC13. The
dexamethasone was eluted with 2 ml of methanol-CHC13 (3:1). The authors reported
72% extraction recovery. Also, many of the matrix components that would interfere with
their analysis of dexamethasone by HPLC were removed by the washes.

A preliminary experiment using this method indicated that both dexamethasone

and 6B-hydroxydexamethasone were recovered from a real plasma sample taken from a

 

37
DST patient. Analysis of the oxidized plasma extract by GC-ECNI-MS with SIM
revealed peaks corresponding to the masses and retention times of oxidized
dexamethasone and oxidized 6B-hydroxydexamethasone.

Recovery of dexamethasone by this solid phase extraction method was not
determined rigorously. The estimated recovery of dexamethasone was 65%. This
estimate was based on comparison of the GC-FID peak area of the 17-keto
decomposition peak, resulting from the injection of a known amount of dexamethasone,
with the peak area of the 17-keto decomposition peak resulting from a known amount of
dexamethasone that had been extracted according to the Plezia and Berens method. This
recovery is reasonably comparable to that found by the authors.

Because the Plezia and Berens solid phase extraction method was deve10ped as
puriﬁcation step to be used prior to HPLC analysis, the elimination of interfering
substances was of utmost importance to these authors, as the selectivity of the UV
detector of the HPLC is limited in the analysis of complex samples. In fact, the
puriﬁcation of the dexamethasone from interfering matrix components was optimized at
the expense of a quantitative recovery of dexamethasone. Because ECNI mass
spectrometry with SIM is a highly selective means of detection, and because oxidation
selectively enhances the response of dexamethasone within the extracted plasma matrix,
removal of matrix components is not as necessary for ECNI-MS analysis. Instead, a
quantitative recovery of dexamethasone is more desirable to allow the lowest possible
detection limit. Therefore, less selective solid phase extraction methodology that would
maximize the recovery of dexamethasone and 6B-hydroxydexamethasone was
investigated.

A solid phase extraction method for dexamethasone was developed by Clinton
Kilts and Jim Ritchie at Duke University (10). As part of our collaboraton with the Duke
laboratory we received the protocol for their solid phase extraction method. The method

consisted of the following steps:

 

 

3 8

1) precondition a 200mg C18 SPE column with

a) 5 ml 8M urea

b) 5 ml methanol

c) 10 ml distilled water
2) pass sample through column, 1-2 drops/sec
3) wash column with

a) 4 ml distilled water

b) 4 ml 38% methanol-water

4) elute dexamethasone with 4 ml 70% methanol-30% hexane

Through the use of a radiolabelled analog of dexamethasone, Kilts and Ritchie
determined the recovery of dexamethasone by this method to be 85%.

Preliminary experiments with this extraction method were satisfactory for the
recovery of dexamethasone. However, the recovery of 6B-hydroxydexamethasone was
very poor. Since 6B-hydroxydexamethasone is more polar than dexamethasone, this
compound was probably eluted by the the 38% methanol-water wash, whereas the more
non-polar dexamethasone remained on the non-polar reverse-phase column.

3. Inﬂuence of Composition of Column Wash (CH3OHszO) on Recovery of 6B-
Hydroxydexamethasone

In order to ﬁnd conditions that would improve the recovery of 6B-
hydroxydexamethasone, an experiment was done in which the percentage of methanol in
the methanol-water wash was decreased. Four 0.5 ml aliquots of plasma, each spiked
with 230 ng of 6B-hydroxydexamethasone and 420 ng of dexamethasone, were loaded
onto four, 200 mg, C18 SPE columns. The extraction procedure outlined above was
followed except that the percentage of methanol in the methaonol-water wash was

different for each of the four extractions. In order to estimate the percent recoveries of

 

3 9
dexamethasone and 6B-hydroxydexamethasone by each of the extractions, the eluates
were reconstituted in 200 111 of CH2C12 and analyzed by normal phase HPLC. The peak
areas obtained for dexamethasone and 6B-hydroxydexamethasone were compared to the
peak areas obtained from a standard which consisted of 230 11g 613-
hydroxydexamethasone and 420 ng dexamethasone in 200 111 of CH2C12. Because the
eluates could be analyzed directly, HPLC provided a faster and simpler means to
determine the percent recoveries than would oxidation followed by GC-ECNI/MS
analysis. The results of this experiment are shown in Table 2.1. Conditions for the
HPLC separation were as follows: partisil silica gel column (25cm x 4.6 mm, 10 11
particle size), 7% ethanol-33% CH2C12-60% hexane mobile phase, 1.8 ml/min = ﬂow

rate, injection volume = 20 111, wavelength = 240 nm, 0.1 absorbance units full scale.

Table 2.1 Extraction recoveries ' of dexamethasone (dex) and 6B-
hydroxydexamethasone (6B-OH dex) vs. percentage of methanol
in solid phase extraction wash.

 

% methanol % recovered 6 -OH dex % recovered dex
38 40 0.0
25 1 13 25
15 104 98
5 98 84

Most of the dexamethasone and 6B-hydroxydexamethasone was recovered if the
methanol-water wash was 15% or less in methanol. The quantities to be recovered from
real plasma samples will be on the order of 100 times less. Also, when extracting real
plasma samples, the sample volume will be two to four times larger than those used in
this experiment. Therefore, it is possible that a 15% methanol-water wash may not
represent the optimum for analysis of realistic samples. However, it is likely to be close
to the optimum condition. Therefore, the 15% methanol-water wash will be used for all

subsequent extractions of dexamethasone and 6B-hydroxydexamethasone from plasma.

 

40

B. Oxidation

Previous work in this laboratory indicated a yield of about 30% 11,17-keto
dexamethasone upon 3— to 5-hour oxidation of dexamethasone with chromium trioxide-
pyridine complex (pyr2CrO3) in pyridine (11). After evaporation of the pyridine, the
product(s) was isolated from the reaction mixture by ﬁltration through a small sephadex
LH-20 column (3cm x 0.5 cm) with ethyl acetate. The yield could be increased slightly,
to about 45% through the use of NaBiO3, a reagent that is speciﬁc for oxidation of 1,2-
diols or 1,2—hydroxy ketones, to cleave off the C-17 side-chain followed by pyr2CrO3
oxidation to oxidize the ll-hydroxyl. However, this involves a two-step process which
requires more labor and could lead to losses of analyte during transfer.

A goal of this research is to improve the sensitivity of the assay for
dexamethasone in order to be able to detect 100 pg/ml concentrations of dexamethasone
in 1 to 2 ml plasma samples. Improving the oxidation yield of the 11,17-keto
dexamethasone ought to improve the sensitivity of the assay. However, another goal is to
avoid complicating the procedure so as to provide a simple approach for the analysis of
plasma for dexamethasone. Therefore, rather than use a two-step oxidation procedure,
several parameters including choice of reagent, reaction time, reaction solvent, heat,
and the presence of a base were investigated as means for increasing the oxidative yield

of 11,17-keto dexamethasone.

1. Choice of Oxidation Reagent

The requirements for a reagent to be used for the oxidation of dexamethasone are:
(1), the reagent is capable of producing a high yield of the 11,17-keto analog of
dexamethasone in a reasonable amount of time; (2), losses due to over-oxidation (beyond

the ketone) are minimal; (3), separation of the oxidation reagent from the dexamethasone

4 l
product(s) is easily achieved; and (4), the reaction involves only one step.

There are a wide variety of chemical oxidation reagents available for the
oxidation of organic compounds. Chromium trioxide is a very strong oxidizing agent
that has been used for oxidation of dexamethasone to its 11,17-keto analog (12).
However, this reagent is capable of oxidizing a hydroxy group all the way to a carboxyl
group and may also react with double bonds and other functionalitie‘s. This would make
it difﬁcult to convert dexamethasone to only one product, namely 11,17-keto
dexamethasone. Also, the stronger the oxidation reagent, the more oxidized the sample
matrix will become. For instance, an alkene can be converted to an a,B-unsaturated
ketone through the oxidation of the allylic C-H bond by any of several chromium (VI)
oxidants (15). Increased oxidation of the matrix components could increase the electron
capture response of the matrix, thereby increasing the probability of interference from the
sample matrix. For these reasons, milder oxidation reagents were investigated. The
derivatives of chromium oxides that were investigated included pyridinium dichromate
(PDC), pyridinium chlorochromate (PCC), poly(4-viny1pyridinium dichromate)
(PVPDC) which is an immobilized reagent, and pyrzCrO3 complex. The structures of
these reagents are shown in Figure 2.3.

An initial experiment was done to compare the yields of 11,17-keto
dexamethasone that could be obtained with PCC and PVPDC with those obtained using
pyr2CrO3 under the conditions used previously in this laboratory. These results are
shown in Table 2.2. Dexamethasone (260 11g) was placed into three silanized vials. (All
glassware used in these procedures was silanized to prevent adsorptive losses of
dexamethasone or its oxidized products.) For the PCC oxidation, the dexamethasone was
dissolved in 1 ml CH2C12; excess PCC, that had been ﬁnely ground with celite, was then

added to the vial. The celite provides a surface for the reaction to occur and also aids the

 

 

~

AWN

42
O

Oicrfo CN+—Cr—O‘ [CN] Cro,
X X — (ll) ‘ 2
X=OH 5 6
X=OAc
X=Cl
X=O—t—Bu

:1 I crcro; a 0,0; M C100,-
\N ‘N FIN-

H 2
8

4:112

9

Figure 2.3. Structures of some chromium (VI) oxidants; chromic acid (1),

chromyl acetate (2), chromyl chloride (3), tert—butyl chromate (4),
chromium trioxide-pyridine complex (5), dipyridine chromium
(VI) oxide (6), pyridinium chlorochromate (7), pyridinium
dichromate (8), and 2,2’-bipyridinium chlorochromate (9).
Adapted from reference (15).

4 3
removal of the reagent by adsorbing the reduced chromium salts. Used without celite,
PCC is reported to deposit a black insoluble tar as the reagent becomes reduced (16). For
the pyr2CrO3 oxidation, the reagent was added as a solution in pyridine (0.5 ml of 7.5
mg/ml). To remove the excess reagent and isolate the products, the solvents were
evaporated and the residues reconstituted in 0.5 ml ethyl acetate. These were then
transferred to short Sephadex LH—20 columns. About 3 ml of ethyl acetate eluate was
collected. For the PVPDC oxidation the reagent was soaked in distilled water overnight
and ﬁltered in the morning to remove the water and the non-immobilized chromium salts.
The dexamethasone was dissolved in 1 ml of cyclohexane followed by addition of the
wet PVPDC. To remove the PVPDC, the reaction mixture was ﬁltered through a glass
wool plug in a pasteur pipette. Other details about the oxidations are included in Table

2.2.

Table 2.2 Relative yields of 11,17-keto dexamethasone obtained with three
different oxidation reagents.

 

reagent rel. %yld reaction conditions

pyrzCrO3 5.1 stirred, room temp., 1 2/3 hr., pyridine
PCC 66 stirred, room temp, 1 1/2 hr., CH2C12
PVPDC 100 70C, 21 hrs., cyclohexane

Oxidation of alcohols with Cr(VI) oxidants usually proceeds with the formation
of a chromate ester followed by abstraction of a hydrogen from the reaction substrate and
the elimination of the reduced chromate. This mechansim is shown in Figure 2.4. The
rate limiting step is apparently the abstaction of the hydrogen from the substrate.
Tertiary alcohols have no hydrogen atom on the same carbon atom to which the hydroxy
group is attached; therefore, these compounds are relatively inert to oxidation with

Cr(VI) oxidants. The cleavage of the C-17 sidechain of dexamethasone to form

44

(It) .
\ /o’—“cu:r—orr \ (I?
/ \P{’ HO/ \OH

Figure 2.4. Mechanism for the oxidation of alcohols with chromium (V I)
oxidants.

4 5

the 17-ketone involves the oxidation of a tertiary hydroxyl group. Therefore, a more
complicated oxidation mechanism may be involved in this reaction.

A mechanism suggested for the oxidative cleavage of 1,2-diols is illustrated in
Figure 2.5a. Chromium (VI) oxidants are capable of cleaving the C-C bond of a diol
which results in two ketone products. The sidechain of dexamethasone does not contain
a 1,2-diol moiety. However, the ketol present on the sidechain may be capable of the
tautomerization shown in Figure 2.5b that would provide a 1,2-diol for oxidative
cleavage by Cr(VI) which would result in the production of 17-keto dexamethasone.
This mechanism may not. be operative in the production of 17-keto dexamethasone,
however, no deﬁnite mechanism for the oxidation of these types of corticosteroid
sidechains by chromium reagents has been found in literature as this type of destructive

reaction is not normally considered to be synthetically useful.

 

 

 

45

the 17-ketone involves the oxidation of a tertiary hydroxyl group. Therefore, a more
complicated oxidation mechanism may be involved in this reaction.

A mechanism suggested for the oxidative cleavage of 1,2-diols is illustrated in
Figure 2.5a. Chromium (VI) oxidants are capable of cleaving the C—C bond of a diol
which results in two ketone products. The sidechain of dexamethasone does not contain
a 1,2-diol moiety. However, the ketol present on the sidechain may be capable of the
tautomerization shown in Figure 2.5b that would provide a 1,2-diol for oxidative
cleavage by Cr(VI) which would result in the production of 17-keto dexamethasone.
This mechanism may not. be operative in the production of 17-keto dexamethasone,
however, no deﬁnite mechanism for the oxidation of these types of corticosteroid
sidechains by chromium reagents has been found in literature as this type of destructive

reaction is not normally considered to be synthetically useful.

46

1'1 R
I

R—C—OH R—C—O ,o R

1 +H,cro,——- I \Cr’ -—»2 \C==O+Cr(lV)
R—C—OH R—C—O/ ‘0 /

1 1 R

R R

OH 0
OH O
--OH --OH Cr(VI)
- ~CH3 3:: --CH3 —» --CH3

Figure 2.5. a) Mechanism for the oxidative cleavage of the C-C bond of a 1,2-
diol and b) proposed mechanism for the oxidative cleavage of the
dexamethaspne C-17 sidechain.

47

Each reaction produced a single product as indicated by a single peak by GC with
ﬂame ionization detection (FID). The peak from each reaction eluted from the GC
column at the time predicted by the retention index of 11,17-keto dexamethasone. No
other peaks were observed in the chromatograms.

The results in Table 2.2 indicate that the yields achievable with PCC and PVPDC
are likely to be much higher than those with pyr2CrO3 under these oxidation conditions.
One possible reason for the relatively poor yield achieved with pyr2CrO3 is that the
reaction was conducted in pyridine. Reaction rates with pyr2CrO3 are reported to be
higher in CH2C12 than in pyridine (17). This is probably because the reagent is twice as
soluble in CH2C12 as it is in pyridine. Better yields might have been achieved with
pyr2CrO3 if CH2C12 had been used. However, due to the relatively high yields achieved
with the other reagents, further study of py12CrO3 was abandoned and studies of PCC
and PVPDC were pursued. Since 2/3 of the yield achieved with PVPDC was achieved in
1/10 of the time with PCC, PCC seemed the better candidate for the most efﬁcient
oxidizing agent. PVPDC, however, has an advantage of very simple clean-up because
the reagent is immobilized on polyvinyl beads. To remove the reagent, the reaction
mixture is simply ﬁltered through a glass wool plug, and the polyvinyl beads are
collected. Because of this advantage an experiment was done to determine if high yields
could be achieved with PVPDC at shorter reaction times. These results are shown in

Table 2.3.

Table 2.3 Reaction time vs. yield of 11,17-keto dexamethasone by PVPDC
oxidation of dexamethasone at 70°C, in cyclohexane.

 

reaction time rel. yield ca. %yield
1 hr 0.02 1.7
3 hr 0.23 19
4.5 hr 0.36 30
24 hr 1.00 83

4.5 , CH2C12-acetone 0.18 15

 

 

4 8

Percent yields of 11,17 -keto dexamethasone were approximated by comparison of
TIC peak areas to a known amount of 1,4-androstadiene-3,11,17-one obtained by GC-
EI/MS. Unfortunately, the results in Table 2.3 cannot be directly compared to those in
Table 2.2. However, if one assumes that the yields of the 21-hour PVPDC oxidation in
Table 2.2 and the 24-hour PVPDC oxidation in Table 2.3 are equal, then as an
approximation it appears that the yield of 11,17-keto dexamethasone by PCC oxidation
for 1.5 hours is still about two times higher than that achieved with PVPDC at a reaction
time of 4.5 hours. The yields at times greater than 4.5 hours were not investigated since
reaction times much greater than 5 hours would not allow the whole sample preparation
(extraction, oxidation, and removal of excess reagent) to be completed in a regular
working day. Instead, other factors were investigated which might increase the yields
available with PVPDC.

Because dexamethasone is only sparingly soluble in cyclohexane, CH2C12-
acetone was added to the reaction mixture to enhance the solubility of dexamethasone.
Rather than increase the yield, this decreased the yield of the reaction as can be seen in
the last entry of Table 2.3. Perhaps the more polar solvent solvated the dexamethasone
more strongly, leading to a slower rate of oxidation. Or, the conformation of the
polyvinyl polymer beads was changed in a way that slowed the reaction. When a
reaction with wet PVPDC is carried out in cylcohexane, the PVPDC sits on the bottom of
the vial inside the bead of water. When the reaction was carried out with CH2C12-
acetone, the reaction mixture was a cloudy emulsion. It is recommended in the original
journal article (18) that the PVPDC be washed with water to remove loose chromium
salts and then soaked in water and ﬁltered, but not dried, prior to use. The dry reagent
was found to be inactive. Perhaps the use of polar solvents displaces from the PVPDC
some of the water necessary for oxidation.

Because some water is necessary for PVPDC activity, another experiment was

designed to investigate whether the amount of water present during the reaction would

4 9
affect the yield of 11,17-keto dexamethasone achieved with PVPDC. The effects of
heating and stirring upon the yield also were compared in this experiment. These results

are shown in Table 2.4.

Table 2.4 Relative yield of 11,17-keto dexamethasone by PVPDC oxidation
vs. presence of excess H20 and heating and stirring. 26 ttg of
dexamethasone was used for each oxidation.

time reagent conditions relative ﬁeld
3 hr PCC stirred 100
3 hr PVPDC heated 40
3 hr PVPDC heated, + H20 46
4.5 PVPDC stirred 33
4.5 PVPDC heated 84
4.5 PVPDC heated and stirred 74

From the results in Table 2.4, a number of observations may be made. First,
stirring the PVPDC reaction does not increase the oxidation yield. Second, heating is
neccessary to promote a relatively high yield with PVPDC. Third, the presence of extra
water in the PVPDC reaction may slightly increase the yield. However, the 6%
difference between the yields achieved by PVPDC with and without excess water may
not be signiﬁcant because of experimental error; each oxidation was done only once.
Finally, the yield achieved with the 3 hour PCC oxidation is still higher than any of the
yields achieved using PVPDC under all the different conditions. Therefore, further
oxidations with PVPDC were abandoned and effort was concentrated on maximizing the
yield by PCC oxidation.

A related reagent, PDC, also was investigated. PDC is a neutral reagent, unlike
PCC which is slightly acidic. Therefore, PDC is used especially for oxidation of
compounds contained acid sensitive groups. Since dexamethasone contains an 01,13-
unstaurated ketone, it is possible that it would be acid sensitive. Also, because PDC is

more polar than PCC, it cannot be eluted from a silica column by ethyl acetate during

 

4 9
affect the yield of 11,17-keto dexamethasone achieved with PVPDC. The effects of
heating and stirring upon the yield also were compared in this experiment. These results

are shown in Table 2.4.

Table 2.4 Relative yield of 11,17-keto dexamethasone by PVPDC oxidation
vs. presence of excess H20 and heating and stirring. 26 pg of
dexamethasone was used for each oxidation. —

time reagent conditions relative yield
3 hr PCC stirred 100
3 hr PVPDC heated 40
3 hr PVPDC heated, + H20 46
4.5 PVPDC stirred 33
4.5 PVPDC heated 84
4.5 PVPDC heated and stirred 74

From the results in Table 2.4, a number of observations may be made. First,
stirring the PVPDC reaction does not increase the oxidation yield. Second, heating is
neccessary to promote a relatively high yield with PVPDC. Third, the presence of extra
water in the PVPDC reaction may slightly increase the yield. However, the 6%
difference between the yields achieved by PVPDC with and without excess water may
not be signiﬁcant because of experimental error; each oxidation was done only once.
Finally, the yield achieved with the 3 hour PCC oxidation is still higher than any of the
yields achieved using PVPDC under all the different conditions. Therefore, further
oxidations with PVPDC were abandoned and effort was concentrated on maximizing the
yield by PCC oxidation.

A related reagent, PDC, also was investigated. PDC is a neutral reagent, unlike
PCC which is slightly acidic. Therefore, PDC is used especially for oxidation of
compounds contained acid sensitive groups. Since dexamethasone contains an 01,13-
unstaurated ketone, it is possible that it would be acid sensitive. Also, because PDC is

more polar than PCC, it cannot be eluted from a silica column by ethyl acetate during

5 O
removal of the reagent following the oxidation. PCC begins to elute with ethyl acetate
and sometimes may not be completely separated from the oxidized dexamethasone. If,
after optimization of the silica clean-up step, it is found that the PCC eluted with the
oxidized dexamethasone, then PDC could possibly be used as an alternative to PCC. The
results from a preliminary study using PDC are shown in Table 2.5. For these reactions
excess PDC ﬁnely ground up with celite was added to 50 ug of dexamethasone that had
been dissolved in 1 ml CH2C12. The reactions were stirred at room temperature. Excess
PDC was removed by passing reaction mixture through a 500 mg silica column. Ethyl
acetate was added to elute the oxidized dexamethasone. Relative yields are reported plus
or minus (i) one standard deviation calculated from the GC-FID peak areas of at least

three separate injections.

Table 2.5 Relative yield of 11,17-keto dexamethasone by PDC oxidation vs.

time.
reaction time rel. yield
2 hrs. 77 i 8
3 hrs. 78
5 hrs. 100 :1: 1 (ca. yld. 50%)
7 hrs. 96 i 3
9 hrs. 78 i- 10

The yield achieved with the PDC oxidation is maximized between 5 and 7 hours.
The yield appears to decrease if the reaction is allowed to continue for 9 hours. Over-
oxidation could be starting to occur at the longer reaction times.

The yield of the 5-hour reaction was estimated to be 50% by comparing the peak
area of the 11,17-keto dexamethasone to the peak area of a known quantity of 170t-
methyl testosterone. Since Nor-methyl testosterone contains the same number of carbon
atoms as 11,17-keto dexamethasone it was initially assumed that the FID responses per
nanogram of the two steroids would be essentially equal. However, the FID response of

a compound is not simply proportional to the number of carbon atoms within a molecule,

51

but to the number of oxidizable carbon atoms within the molecule (19). Because 11,17-
keto dexamethasone is more highly oxidized than Nor-methyl testosterone, the response
per nanogram of 11,17-keto dexamethasone is less than the response per nanogram of
Nor-methyl testosterone. Therefore, yields of 11,17-keto dexamethasone calculated by
comparison to the response of a known amount of 1701-methyl testosterone will be
atiﬁcially low because a greater amount of 11,17-keto dexamethasone must be injected to
achieve the same peak area as produced by the given amount of Nor-methyl testosterone.

In order to determine the yield of 11,17-keto dexamethasone accurately from any
oxidation, it was necessary to Obtain a primary standard of 11,17-keto dexamethasone
which could be weighed on an analytical balance. Therefore, a large quantity of
dexamethasone was oxidized and recrystallized. The reaction was scaled up in order to
oxidize 18 mg of dexamethasone. The reaction was followed by thin layer
chromatography (TLC). After 1.5 hours only one spot was apparent. This spot had the
same Rf value as an authentic standard of 11,17-keto dexamethasone. After removal of
the excess oxidation reagent on a silica column and recrystallization of the product from
aqueous acetone, a quantity of the product was weighed. GC-FID analysis of a solution
of the product resulted in one peak having the retention time of authentic 11,17-keto
dexamethasone. Later, GC-EI/MS conﬁrmed the identity of the product as 11,17-keto
dexamethasone. Serial dilutions of this solution were made in order to construct standard
curves for the determination of unknown concentrations of 11,17-keto dexamethasone by

GC with FID and ECD.

2. Optimization of PCC Oxidation Conditions

In order to improve the yield of 11,17-keto dexamethasone achieved by PCC

oxidation, the effects of heating and the presence of a base (either sodium acetate or

pyridine) were studied. Sodium acetate has been used to buffer the acid liberated during

52
the PCC oxidation. A small percentage of pyridine in CH2C12 could also be used as a
buffer. However, the selectivity of the reaction has been observed to be altered with 2%
pyridine. Sodium acetate has no effect on the selectivity of PCC oxidation.

Tables 2.6 and 2.7 summarize the results of these studies of the effect of heat,
sodium acetate, and pyridine upon the yield of 11,17-keto dexamethasone by PCC
oxidation. All of the oxidations were carried out for 2 1/4 hours in 1.0 ml of CH2C12,
with an excess of ﬁnely ground pyridinium chlorochromate (PCC)/Celite mixture. Fifty
micrograms of dexamethasone were used in each oxidation. Removal of the excess
oxidation reagent and recovery of the oxidation products were done by passing the
reaction mixture through a 500 mg silica solid phase extraction (SPE) column. The

products were eluted from the column with 2.8 ml of ethyl acetate.

Table 2.6 The effect of a base on the yield of 11,17-keto dexamethasone.

vial no. treatment %yield
1a N aOAc 54
lb NaOAc 55
2 1% pyridine 59
3 none 27

The percent yields were calculated by comparison of the GC-FID peak areas to a
standard curve constructed with the serial dilutions of the known concentration of the
recrystallized 11,17-keto dexamethasone. From comparison of the percent yields for
each reaction in Table 2.6 it is apparent that the presence of a base in the oxidation
enhances the rate at which the 11,17-keto dexamethasone is produced. Also to be noted
from Table 2.6 is the closeness of the results for the reactions which contained a base
(vials 1a and 1b, which were duplicates, and vial 2). Since the relative standard deviation
(RSD) for GC analysis is 2.4%, it is difﬁcult to determine whether there is any difference

between the yields when pyridine is used as the base or when NaOAc is used.

53

The closeness of the results for the duplicate reactions, 1a and 1b, is encouraging
with regard to the reproducibility of the oxidation reaction. The difference between the
percent yields for these reactions is within the variation to be expected due to GC

injection and integration errors.
, It should also be mentioned that the percent yield calculated for the reaction in
which no base was used may be artiﬁcially low due to losses of sample during transfer
steps. To prevent this kind of ambiguity an internal standard, 4-androstene-3,6,17-one,

was used for all subsequent experiments.

Table 2.7 The effect of heating on the yield of 11,17-keto dexamethasone.

vial no. treatment ratio
1 1% pyridine, R.T."' 0.66
2 1% pyridine, heat 0.79
3 NaOAc, R.T.* 0.66
4 NaOAc, heat 0.78

I"R.T. = room temperature

The effect of heating upon the reaction rate is recorded in Table 2.7. The ratio in
the third column is the ratio of the peak area of 11,17-keto dexamethasone to the peak
area of the recovery standard (4-androstene-3,6,17-trione) that was added to each
oxidation mixture just prior to the clean-up step. Comparison of this ratio for each
reaction, rather than the absolute peak areas of 11,17-keto dexamethasone, ought to
compensate for losses incurred during transfer steps following the oxidation and for
variation in GC injection volume. An increase in the rate of oxidation is observed upon
heating both when pyridine is used as the base and when NaOAc is used. The increase in
rate is ca. 17%. Due to the boiling point of CH2C12, it is not practical to raise the
temperature above 60°C.

It should be mentioned that using pyridine causes broadening of the PCC band on

54
the silica column during the clean-up step. Therefore, the use of pyridine could be a
disadvantage if the pyridine causes the PCC and the 11,17-keto dexamethasone to co-
elute from the silica column. For this reason, and because there appears to be no reaction
yield advantage in using pyridine, sodium acetate will be used in all subsequent PCC

oxidations.

3. Determination of the Yield of 11,17-Keto Dexamethasone

The yield of 11,17-keto dexamethasone from a 5.5 hour oxidation of 50 1.1g of
dexamethasone was calculated as 71% (injection standard deviation = 2.4) from the FID
standard curve. After 5.5 hours no unoxidized dexamethasone was seen by HPLC or
TLC. However, the HPLC chromatogram did contain extra peaks one of which may
have been due to a partially oxidized intermediate. (HPLC preparative collection of the
peaks followed by direct probe EI-MS was done to determine if any were a partially
oxidized form of dexamethasone. However, the presence of contaminants, probably from
the concentration of the mobile phase, obscurred the spectra making interpretation
difﬁcult. To identify an intermediate oxidation product a larger quantity would be
needed.)

In order to determine if the reaction yield could be increased beyond the 70%
range without a prohibitively long reaction time, another experiment was done in which
167 ng of dexamethasone was oxidized for different lengths of time. The results of this

experiment are shown in Table 2.8.

55

Table 2.8 Yield of 11,17-keto dexamethasone by PCC oxidation vs. time.

reaction time relative yield
4’50" 74%
8’ 78%
9’ 10" 100%

The yield after nine hours was signiﬁcantly higher than the yield after both 8 and
5 hours. However, since the yield after 5.5 hours was found to be in the 70% range in the
earlier experiment, the extra four hours required to increase the yield another 25% were
not considered to be worth the extra time. Therefore, a reaction time of 5-6 hours was

judged to be sufﬁcient.

4. The Effect of Extracted Plasma Matrix on the Oxidation of Dexamethasone

It is possible that the components of the plasma which are extracted along with
dexamethasone during the initial isolation step of the method could have an effect on the
oxidation of the dexamethasone. Although a good yield may be achieved by a 5-hour
PCC oxidation of a pure standard of dexamethasone, the presence of the plasma matrix
could decrease the yield considerably. In order to determine the effect of the extracted
plasma matrix, triplicate oxidations of 50 ug of dexamethasone were done within the
plasma extract. Three milliliters of plasma was extracted using a C18 solid phase
extraction column as described in section IIA. The extract was divided into three
portions and 50 ug of dexamethasone was added to each portion. For comparison
triplicate oxidations of 50 pg of dexamethasone without the plasma matrix were done
concurrently. For all of these oxidations NaOAc was used as the base and the reactions
were heated to 57°C. After 21/4 hours the recovery standard (4-androstene-3,6,17-trione)

was added to each of the vials. The contents of each vial were placed on separate 500 mg

5 6
silica SPE columns. The steroids were recovered with ethyl acetate. Table 2.9 shows the

results obtained by GC-FID analysis of these samples.

Table 2.9 Ratio of GC-FID responses for 11,17-keto dexamethasone and the
internal standard, 4-androstene-3,6,l7-one, for oxidations
conducted with and without presence of plasma matrix extract. s=
std. deviation of 3 injections.

vial no. plasma area ratio mean area ratio
1 yes 0.536 (s=.002)
2 yes 0.59 (s=.01) 0.56 (s=.03)
3 yes 0.554 (s=.006)
4 no 0.614 (s=.002)
5 no 0.664 (s=.004) 0.65 (s=.03)
6 no 0.657 (s=.004)

The results of the experiment in which dexamethasone was oxidized in the
presence of plasma extract are recorded in Table 2.9. The last column in Table 2.9
contains the means of the two sets of triplicate reactions. These means are signiﬁcantly
different by the t-test at a conﬁdence level of 99%. In other words, there is only a
probability of 1% that the difference between the two means is due to random ﬂucuation.
This indicates that the production of 11,17—keto dexamethasone is slower in the presence
of the plasma matrix. The ratio of the mean ratios is 0.87 which indicates that in the
presence of plasma matrix the reaction is only 87% as efﬁcient. The effect of the plasma
matrix upon the oxidation of dexamethasone may be even more pronounced when the
concentration of the dexamethasone is much less than the concentrations of the matrix
components as it would be in real samples. Because of this, reaction times longer than 5
hours may be necessary. This experiment was repeated at the 5 ng/ml level; however, the
results by GC-ECD were inconclusive due to interferences from the plasma matrix
obscurring the 11,17-dexamethasone and internal standard peak areas. Unfortunately,
when this experiment was conducted no mass spectrometer having adequate sensitivity

was available to more selectively analyze these samples.

57

5. Oxidation of 6B-Hydroxydexamethasone

The compound 6B-Hydroxydexamethasone differs from dexamethasone only by
the introduction of the 6B-hydroxy group. It was expected that oxidation of 6B-
hydroxydexamethasone would also yield a highly electrophilic product which would
have the structure shown in Figure 2.6.

Approximately forty micrograms of 6B-hydroxydexamethasone was oxidized in
CH2C12 by PCC/celite with NaOAc, at 60°C for about 12 hours. Gas chromatography
with ECD yielded a large peak having a retention time very close to that of 11,17-keto
dexamethasone. The identity of the peak was conﬁrmed to be 6,11,17—keto
dexamethasone by GC-EI/MS. The EI mass spectrum is shown in Figure 2.7a. The
ECNI mass spectrum in shown in Figure 2.7b. The base peak, at m/z 324, in the ECNI
spectrum represents the loss of HF from the intact molecule. No peak representing the
molecular anion is present. Also, the 6,11,17-keto dexamethasone apparently undergoes
a reaction under ECNI conditions to produce an ion of mass 326. Reaction of the (M-
HF)' with hydrogen radicals could account for the peak at m/z 326. These types of

reactions have been reported for aromatic compounds under ECNI conditions (20).

58

Oxidation

 

Figure 2.6. Oxidation of 6B-Hydroxydexamethasone to 6,11,17-keto
dexamethasone.

 

59

 

 

 

 

 

 

 

 

 

 

145
tom
R l 69 .
f ‘ 135 '
‘ 106 4 -
a 50‘ 57 115125 17 -
t . 81 95 153 .
i j 13 185 199210221i
V i
e 0‘
A 50 too 150 200 _
b I
U D
n .
d A
a ,
n .
g 245 259 279 288296
250 300 350 400
m/z
' 324. 4 TxE+o4
100-
.—5
89-
326. 4 -4
r__
60-1
-3
40- __2
20- 1.4
309.3 328.4
1 t—
0 11. L A
1 i m ' 1 m ' 1
280 300 320 340 360 380
Figure 2.7. Mass spectra of oxidized 6B-hydroxydexamethasone obtained by

a) electron impact ionization and b) electron capture negative

ionization.

6 0
An experiment was conducted to determine the relative yield of 6,11,17-keto
dexamethasone vs. time. These results are shown in Table 2.10. For these reactions 194
ng of 6B-hydroxydexamethasone was used. The reaction conditions used were the same
as those determined to produce a good yield of 11,17-keto dexamethasone from
dexamethasone. The compound 4-Androstene-3,6,17-one was added to the reactions

prior to the silica step to serve as an internal standard to correct for transfer losses.

Table 2.10 Relative yield of 6,11,17-keto dexamethasone vs time.

reaction time rel. %yield
4.5 hr. 86
6.0 hr. 93
7.5 hr. 100

The highest relative yield of 6,11,17-keto dexamethasone was achieved after 7.5
hours. The actual percent yield of 6,11,17-keto dexamethasone could not be determined
as no standard of 6,11,17-keto dexamethasone was available. Such a standard could not
be synthesized in large quantity as only 2mg of 6B-hydroxydexamethasone was available
(a gift from C. Kilts of Duke University). The percent yield for the 7.5 hour oxidation
may be estimated to be on the order of 15% by assuming that the ECD responses of
6,11,7-keto dexamethasone and 11,17-keto dexamethasone are approximately 3 to 1.
This assumption is based on the observation (21) that the relative ECNI responses of 1,4-
androstadiene-3,6,11,17-one and 1,4-androstadiene-3,11,17-one are 3 to 1. The only
difference between these molecules and the oxidation products of 6B-
hydroxydexamethasone and dexamethasone is that the dexamethasones have 9a—F and
l6oc-methyl groups.

No other major peaks were seen in the GC-ECD chromatogram. No peak was
seen at the retention time corrresponding to the 17-keto analog of 6B-

hydroxydexamethasone (a standard was prepared by NaBiO3 oxidation of

 

61
6B-hydroxydexmethasone). If unoxidized 6B-hydroxydexamethasone were present in the
samples it would be expected that the 17-keto analog would be observed in the
chromatogram as a result of thermal decomposition of 6B-hydroxydexamethasone in the

GC injection port.

C. Removal of Excess Reagent and Isolation of Oxidized Dexamethasone and 613-
Hydroxydexamethasone

It is important to remove the oxidant completely when isolating the reaction
products so that the solution of products will not degrade over time. Complete removal
of oxidant also prevents damage to the GC column. This section discusses methods for
PCC removal and optimal recovery of 11,17-keto dexamethasone and 6,11,17-keto

dexamethasone.

1. Silica vs. Solvent Extraction

An experiment was done to compare the efﬁciency of the removal of PCC and the
recovery of 11,17-keto dexamethasone by ethyl acetate solvent extraction with the
efﬁciency by silica solid phase extraction, employing ethyl acetate as the eluent. The
recoveries of 11,17-keto dexamethasone obtained by the two methods were not
signiﬁcantly different (both were ca. 95%). Because solvent extraction involves at least
three extraction steps and emulsions can form at each step, it is a more time consuming

method than solid phase extraction. Therefore, solvent extraction was not used.

2. Comparison of Sephadex LH-20 and Silica

Sephadex LH-20 had been used previously in this laboratory for removal of the

excess pyrzCr03 and isolation of oxidized dexamethasone. Removal of this oxidation

62
reagent was efﬁcient using this procedure, but the recovery of oxidized dexamethasone
was never assessed. Sephadex LH-20 was found to effectively remove PCC as well. In
order to determine the most efﬁcient method for the recovery of 11,17-keto
dexamethasone, the efﬁciency of recovery using Sephadex LH-20 was compared to the
efﬁciency achieved using silica columns. The results of the comparison of Sephadex

LH-20 and silica are shown in Table 2.11.

Table 2.11 Recovery of 11,17 -keto dexamethasone by Sephadex LH-20 and
by silica SPE column.

column type eluent % recovegl
Sephadex 3.5 ml ethyl acetate 96%
silica 3.5 ml ethyl acetate 104%
silica 3.5 ml ethyl acetate-CH2C12 (3:1) 80%

The results in Table 2.11 indicate that both silica and Sephadex give quantitative
recovery of 11,17-keto dexamethasone. The recovery of 11,17-keto dexarrrethasone is
somewhat less if an equal volume of a less polar eluent is used, as seen in the third entry
of Table 2.11.

Since the silica SPE columns are commercially made, the reproducibility of the
recovery ought to be better with these columns than with the home-made Sephadex LH-
20 columns. Because of this and because the use of Sephadex LH-20 has no advantage
for recovery of 11,17-keto dexamethasone, silica SPE columns were used for all

subsequent experiments.

3. Optimization of silica solid phase extraction of 11,17-keto dexamethasone and
6,1 1,17-keto dexamethasone

Although much selectivity for the detection of 11,17-keto dexamethasone and

6,11,17-keto dexamethasone is provided by the chemical oxidation and the

 

63
GC-ECNI/MS (SIM) analysis, the silica step may be used to provide another dimension
of selectivity. Isolation of 11,17-keto dexamethasone and 6,11,17-keto dexamethasone
from other potentially interfering components in oxidized plasma matrix may allow a less
selective means of introducing the sample into the mass spectrometer, such as the direct
inlet probe (DIP). The use of DIP would simplify and shorten the mass spectrometric
analysis.

In order to isolate 11,17-keto dexamethasone and 6,11,17-keto dexamethasone
more selectively, the volume and the polarity of the eluent was varied and fractions were
collected. The following experiment was done to determine the volume of ethyl acetate
necessary to elute all of the 11,17-keto dex from the silica column. After a 5 1/2 hour
PCC oxidation of 50 ug of dexamethasone, the reaction mixture was transferred to a 500
mg silica Bond-Elut column. Fractions of the eluate were collected and analysed by
normal phase HPLC to determine the presence of product. These results are shown in

Table 2.12.

Table 2.12 HPLC analysis of fractions from silica column used for removal of
PCC following oxidation of dexamethasone.

 

 

fraction eluent 11.17-keto dex
1 1 ml CH2C12 0%
2 2.5 ml ethyl acetate 97.5%
3 2.5 ml ethyl acetate 2.5%
4 80% MeOH in ethyl acetate 0%

The HPLC analysis of the silica fractions shown in Table 2.12 indicate that nearly
all the 11,17-dex is eluted from the 500 mg silica column with 2.5 ml of ethyl acetate.
Previous to this experiment, 4-8 ml of ethyl acetate had been used to recover 11,17 -keto
dexamethasone from reaction mixtures. The use of a smaller volume of ethyl acetate will

eliminate the elution of more polar matrix components from the column.

64

The use of an eluent less polar than ethyl acetate was investigated for the recovery
of both 11,17-keto dexamethasone and 6,11,17-keto dexamethasone. The use of a
CH2C12 wash was implemented, as well, to elute matrix components of lower polarity
than 11,17-keto dexamethasone and 6,11,17-keto dexamethasone. These results are
shown in Table 2.13. For this experiment 3 ml of plasma were spiked with 105 ng of
dexamethasone and 78 ng of 6B-hydroxydexamethasone. The dexamethasone and 6B-
hydroxydexamethasone were extracted from the plasma and oxidized with PCC
according to the procedures described previously in this chapter. After the oxidation the
reaction mixture was divided into three approximately equal portions. Each portion was
placed on a separate silica column. Different eluents and volumes of eluents were
applied to each column. Only the results from analysis of fractions collected from one of
the columns are shown in Table 2.13. Due to the low levels of dexamethasone and 6B-
hydroxydexamethasone added to the plasma, mass spectrometry was used to provide

more selectivity for the analysis of the fractions than is provided by the GC-ECD.

Table 2.13 DIP-ECNI/MS analysis of silica SPE column for recovery of
11,17-keto dexamethasone and 6,11,17-keto dexamethasone.

 

 

 

fraction eluent 11.1ﬂceto dex 6,11,17-keto dex
1 2 ml CH2C12 0% 0%
2 0.7 ml 5% acetone-CH2C12 0% 0%
3 2.8 ml 5% acetone-CH2C12 100% 100%

The results in Table 2.13 indicate that the CH2C12 wash does not elute 11,17-keto
dexamethasone or 6,11,17-keto dexamethasone and that 5% acetone-CH2C12 is polar
enough to elute both of these compounds. The percentages in Table 2.13 represent the
portion eluted in the fraction relative to the total amount eluted in all three fractions. It is
possible that some of the 11,17-keto dexamethasone and 6,11,17-keto dexamethasone

still remained on the column after the collection of the third fraction.

65

Another experiment was done to determine the volume of 5% acetone-CH2C12
necessary to recover all of the oxidized 6B-hydroxydexamethasone. Since 6,11,17-keto
dexamethasone is more polar than 11,17-keto dexamethasone, the volume necessary to
recover 6,11,17-keto dexamethasone will most likely be adequate to recover all of the
11,17-keto dexamethasone as well. Approximately 40 mg of 6B-hydroxydexamethasone
was oxidized with PCC. The results of the GC-ECD analysis of the silica fractions is
shown in Table 2.14.

Table 2.14 GC-ECD analysis of silica fractions for isolation of 6,11,17-keto

 

 

dexamethasone.
fraction eluent 6,11,17-keto dex
1 1.5 ml CH2C12 4%
2 4 ml 5% acetone-CH2C12 89%
3 4 ml ethyl acetate 7%

A volume of 4 ml of 5% acetone-CH2C12 recovered 89% of the 6,11,17-keto
dexamethasone. A small percentage was eluted by the CH2C12 wash. In the previous
experiment, none of the 6,11,17-keto dexamethasone eluted in the 2 ml wash. The most
likely reason for this discrepancy is that a much greater amount of the sample (ca. 40 mg)
was applied to the column in the second experiment than in the frrst experiment (ca. 26
ng). The greater amount of sample would have a much broader elution band than the
small amount of sample.

Four milliliters of 5% acetone-CHzClz were used to elute 11,17-keto
dexamethasone and 6,11,17-keto dexamethasone from silica. This ensures that nearly all
the 11,17-keto dexamethasone and 6,11,17-keto dexamethasone are eluted from the silica
column, and minimizes the elution of components more polar than these analytes. Only
one and one-half milliliters of CH2C12 were used to minimize the possible elution of

11,17-keto dexamethasone and 6,11,17-keto dexamethasone in the CH2C12 wash.

66

III. Conclusion

Conditions for the isolation and oxidation of dexamethasone have been studied
extensively to establish optimal recovery of 11,17-keto dexamethasone. Some
preliminary work for the isolation and oxidation of 6B-hydroxydexamethasone has also
been completed.

Extraction of both dexamethasone and 6B-hydroxydexamethasone from plasma
was most efﬁcient using the C18 solid phase extraction method of Kilts and Ritchie
modiﬁed by the implementation of a 15% methanol-water wash, rather than the 38%
methanol-water wash. Recovery of dexamethasone and 6B-hydroxydexamethasone (at
the 200-400 11 g level) was approximately complete using the modiﬁed method.

The best conditions of those investigated for the conversion of dexamethasone to
11,17-keto dexamethasone were oxidation with PCC ground up with celite, with NaOAc
buffer, in CH2C12, stirred at 60°C. A yield of 70% 11,17-keto dexamethasone could be
achieved after 5.5 hours using these conditions. Higher yields could be obtained at
longer reaction times. The yield of 6,11,17-keto dexamethasone from the PCC oxidation
of 6B-hydroxydexamethasone was not optimized, however, 6,11,7-keto dexamethasone
was produced under the conditions used for the oxidation of dexamethasone.

Recovery of 11,17-keto dexamethasone and 6,11,17-keto dexamethasone from
the reaction mixture may be accomplished through the use of a silica solid phase
extraction column. Complete recovery can be achieved using ethyl acetate as the eluent.
To achieve a nearly quantitative recovery of 11,17-keto dexamethasone and 6,11,17-keto
dexamethasone, while avoiding co-elution of some of the matrix components, a 1.5 ml

CH2C12 wash followed by collection of 4 ml of 5% acetone-CH2C12 is recommended.

10.

ll.

12.
13.
14.
15.

l6.
l7.
18.

REFERENCES

B. J. Carroll, M. Feinberg, J. F. Greden, J. Tarika, A. A. Arbala, R. F. Haskett, N.
M. James, Z. Kronfol, N Lohr, M. Steiner, J. P. de Bigne, and E. Young, Arch.
Gen. Psychiatry, 38, 5-22, 1981.

M. T. Lowy and H. Y. Meltzer, Biol. Psychiatry, 22, 373-385, 1987.

G. F. Johnson, G. Hunt, K. Kerr, and I. Caterson, Psychiatry Res, 13, 305-313,
1984.

G. W. Arana, R. J. Workman, and R. J. Baldessarini, Am. J. Psychiatry, 141,
1619-1621, 1984.

B. S. Cham, B. Sadowski, J. W. O’Hagon, C. N. de Wytt, F. Bochner, and M. J.
Eadie, Ther. Drug Monitoring, 2, 373-377, 1980.

Y. Kasuya, J. R. Althaus, J. P. Freeman, R. K. Mitchum, and J. P. Skelly, J.
Pharm. Sci, 73, 446-451, 1984.

K. Minagawa, Y. Kasuya, S. Baba, G. Knapp, and J. P. Skelly, J. Chromatogr,
343, 231-237, 1985.

Minagawa, Y. Kasuya, S. Baba, G. Knapp, and J. P. Skelly, Steroids, 47, 175-
188, 1986.

D. S. Skrabalak, K. S. Cuddy, and J. D. Henion, J. Chromatogr, 341 , 261-269,
1985.

personal communication with Clinton Kilts of Duke University Medical Center,
Departments of Pharmacology and Psychiatry, Durham, NC.

Guor Ron g Her, Mass Spectrometric Studies of Molecules Modiﬁed for Analysis
by Electron Capture Negative Chemical Ionization, Ph.D. Dissertation, Michigan
State University, Department of Chemistry, 1985.

K. S. A. Razzak and K. A. Hamid, Biomed. Mass Spectrom, 7, 505-510, 1980.
C. H. L. Shackleton and J. 0. Whitney, Clin. Chim. Acta, 107, 231-243, 1980.

P. M. Plezia and P. L. Berens, Clin. Chem, 31, 1870-1872, 1985.

F. Freeman, "Oxidation by Oxochromium (VI) Compounds" in Organic
Syntheses By Oxidation With Metal Compounds, M. J. Mijs, and C. R. H. I. de
Jonge, eds, p.44, Plenum Press, New York, NY, 1986.

E. J. Corey and J. W. Suggs, Tetrahedron Lett, 2647-2650, 1975.

J .C. Collins, W. W. Hess, and F. J. Frank, Tetrahedron Lett, 3363-3366, 1968.

J. M. J. Frechet, P. Darling, and M. J. Farrall, J. Org. Chem, 46,1728-1730, 1981.

67

 

19.

20.

21.

 

6 8
H. H. Willard, L. L. Merritt, J. A. Dean, and F. A. Settle, Instrumental Methods of
Analysis, Sixth edition, Wadsworth Publishing Company, Belmont, CA, pp. 468-
470, 1981.

P. S. Callery, W. A. Garland, and E. K. Fukuda, Org. Mass Spectrom, 24, 385-
390, 1989.

G. R. Her and J. T. Watson, Biomed. Environ. Mass Spectrom, 13, 57-63, 1986.

 

CHAPTER 3. COMPARISON OF GC-MS AND MS/MS TECHNIQUES IN THE
ANALYSIS OF PLASMA FOR DEXAMETHASONE

I. Introduction

Described in this chapter is the adjustment of ECNI-MS conditions and MS/MS
conditions for the detection of oxidized dexamethasone. Concentrations of
dexamethasone in plasma samples from patients undergoing the dexamethasone
suppression test (DST) have been determined by these mass spectrometric techniques
using the sample preparation methodology described in Chapter 2. These results are
compared with those obtained by GC-EI/MS and RIA methods in other laboratories,
where aliquots of the same plasma samples were analyzed.

The GC-EI/MS method for the determination of dexamethasone is based on the
formation of the TMS-enol-TMS ether derivative (1). This method represents the best
conventional GC-MS method available for the determination of dexamethasone (2). The
structure of this derivative and selected ion current proﬁles of the (M-CHzOTMS)+ ion at
m/z 577 and the molecular ion at m/z 680 are shown in Figure 3.1. The peak at m/z 577
represents the most intense high mass fragment ion in the EI mass spectrum of this
derivative. The sensitivity of an assay based on this derivative is limited by the low
intensity of the peaks representing these ions.

Figures 3.1 and 3.2 compare the sensitivity available through this GC-EI/MS
method with the sensitivity available through oxidation and GC-ECNI/MS analysis. The
data in Figure 3.1 were obtained from 2 ml of blank plasma that was spiked with 5 ng of
dexamethasone. The plasma was extracted using a C18 solid-phase extraction column.
The extract was divided into two equal portions. One portion was derivatized to make
the TMS-enol-TMS ether and the other portion was oxidized to produce 11,17-keto

dexamethasone. The GC-EI/MS selected ion current proﬁles in Figure 3.1 result from a

69

 

7 0
6% injection of the derivatized portion; this represents at most 150 pg of the
dexamethasone added to the blank plasma. Figure 3.2 shows the selected ion current
proﬁle obtained by GC-ECNI/MS analysis of a 5% aliquot of the oxidized portion of the
extracted plasma; this represents no more than 125 pg of the original dexamethasone. By
comparison of the s/b ratios of the selected ion current proﬁles by the two methods it is
apparent that greater sensitivity ought to be available using oxidation with GC-ECNI/MS

analysis.

II. Sample Preparation andAnalysis of Plasma for Dexamethasone Using SIM

A. Experimental

This section describes the experimental details for the analysis of plasma for
dexamethasone. Included here is the protocol for sample preparation, as well as the gas
chromatographic conditions and the selected ion monitoring parameters that were used.

The adjustment of the ECNI parameters is also discussed.

1. Materials and Sample Preparations

Chemicals used for the oxidation procedure include pyridinium chlorochromate
(Aldrich Chemical Company, Inc.) and anhydrous sodium acetate (J. T. Baker Chemical
Company). All solvents used were either analytical or HPLC grade. They were
purchased from Burdick and Jackson Division of Baxter Healthcare Corporation, J. T.
Baker Chemical Company, EM Science, and Mallinckrodt, Inc. CeliteTM analytical ﬁlter
aid was purchased from Johns-Manville Products Corporation. Analytical reagent grade
urea was purchased from Mallinckrodt, Inc. Solid-phase extraction columns (200 mg

C18 PrepSepTM) were purchased from Fisher Scientiﬁc. Silica, solid-phase extraction

 

71

CHOTMS

TMSO

— _ CH3
0

El

1‘1

MMNJWMWWMMQW WWMML MW

m/z 680

J. x l -

WI‘WWNIW‘VWR'MWR‘MtWKWW‘M.W#~

 

 

 

1'4 is is 1'1 18

Time (minutes)

Figure 3.1 Selected ion current proﬁles for detection of the TMS-enol-TMS
derivative of dexamethasone within plasma mauix, by GC-EI/MS.

72

 

EC-Nl
- m/z 330 plus m/z 310

 

 

 

 

 

Time (minutes)

Figure 3.2 Selected ion current proﬁle for detection of oxidized
dexamethasone in plasma matrix, by GC-ECNI/MS.

7 3
columns (500 mg) were purchased from Burdick and Jackson Division of Baxter
Healthcare Corporation. Dexamethasone was purchased from Steraloids, Inc. l3C6-2H3
dexamethasone was a gift from Clinton Kilts of Duke University. All materials were
used without further puriﬁcation. Silanized glassware was used throughout the
procedures.

The preferred conditions for sample isolation and oxidation'were established by
the studies described in Chapter 2. This sample preparation consists of three steps:
plasma extraction, chemical oxidation, and removal of excess oxidation.

Before extraction, the internal standard (35 ng of 13C5-2H3 dexamethasone in 50
u] of methanol) was added to each sample. Samples consisted of 1.5 ml of plasma taken
from patients at various times following administration of 1 mg of dexamethasone.
PrepSepTM solid phase extraction columns containing 200 mg of C18 packing were pre-
conditioned with 5 ml of an 8 M urea solution to deactivate any remaining active sites,
followed by washes with 5 m1 of methanol and 10 m1 of distilled water. The plasma
samples were then applied to the columns followed by washes with 4 ml distilled water
and 4 ml of 15% methanol in water. Dexamethasone was eluted with 4 ml of 70%
methanol in hexane. All solvents were pulled through the solid-phase extraction columns
at ﬂow rates of approximately 1-2 drops/sec through the use of a vacuum manifold,
except the ﬁnal elution which was done by gravity ﬂow to avoid contamination from the
vacuum manifold.

Following removal of the eluting solvent under nitrogen, the dried residues were
each reconstituted in 1 ml of CH2C12 while stirring with a magnetic stirrer to promote
dissolution of the extracted material. Approximately ten milligrams of anhydrous sodium
acetate were added to the reconstituted extracts. Sodium acetate is added to buffer acid
liberated during the oxidation. An excess of pyridinium chlorochromate, ﬁnely ground
with celite (ca. 1:3 w/w) by using a mortar and pestle, was then added to the solutions.

The reaction mixtures were stirred and heated to 50-60‘C for 6 1/2 hours.

 

74

After completion of the oxidation, the reaction mixtures were directly transferred
to 500 mg silica solid phase extraction columns that had been prewashed with several
column volumes of CH2C12. The columns were then washed with 1.5 ml of CH2C12.
The oxidation product of dexamethasone was eluted with 4 ml of 5% acetone in CH2C12.

After removal of the 5% acetone-CH2C12 under nitrogen, the residue was
reconstituted in 100 [.11 of ethyl acetate. The GC-ECNI/MS analysis required 1 to 3 pl of
sample.

For preparation of the calibration curve, 13C5-2H3-dexamethasone (35 ng) was
added to each of several 1 ml aliquots of blank plasma obtained from the Red Cross
blood bank. Following this, additions of 0.0, 0.209, 0.418, 1.67, 4.18, 8.36, and 16.7 ng
of dexamethasone, in 25 to 100 pl of methanol, were made to the 1 ml aliquots of
plasma. . Each standard was prepared in duplicate. The extraction, oxidation, isolation,
and analysis procedures for these standards were identical to the procedures used for the
clinical plasma samples.

Another standard was made to be used in the assessment of the precision and
accuracy of the methodology. For this standard, one milliliter of blood bank plasma was
spiked with 35 ng of 13C6-2H3-dexamethasone, followed by addition of 2.09 ng of
dexamethasone. This standard was prepared in duplicate and analysed in triplicate by

GC-ECNI/MS with SIM.

2. Adjustment of Instrumental Parameters

All GC-ECNI/MS data were acquired with a Finnigan TSQ70 directly coupled
with a Varian GC. For selected ion monitoring, windows of 0.8 u about the ion currents
at m/z 339.2, 319.2, 330.2 and 310.2 were monitored for a dwell time of 50 milliseconds

each. Ion current at m/z 324.2 was also monitored to detect the oxidation product of

75
6B-hydroxydexamethasone. This 0.8 u window was chosen to maximize the detection of
the ion current at these masses without detection of ion current at adjacent masses.

Ammonia (Matheson, 99.99% purity) was used as the reagent gas at a source
pressure that corresponded to a foreline pressure of 9300-9500 mtorr. This pressure was
obtained from the convectron gauge that is located on the foreline before ion source and
the turbomolecular pumps. The pressure reading at this point is considerably higher than
the actual pressure inside the ion source chemical ionization volume. There was no
available means by which to measure the actual ion source pressure. The operator’s
manual for the TSQ70 recommends a pressure of about 0.5 torr be used for CI conditions.
Because good sensitivity was obtained at convectron readings of 9000-9900 mtorr, it was
assumed that this foreline pressure corresponds to an ion source pressure on the order of
0.5 torr. Both CH4 and NH3 were investigated as the modifying gas. The sensitivity for .
the detection of 11,17-keto dexamethasone appeared to be same regardless of the gas that
was used; however, when CH4 was used higher background was experienced. Intense
peaks at m/z 233 and m/z 235 in a ratio of 1:3 were observed, particularly when CH4 was
used. The masses of these ions and the isotope pattern correspond to ReO3-. Others have
noted a high background due to these isotopes under CH4 ECNI conditions (3). The
source of the rhenium is the ﬁlament used to generate the primary electron beam.

The ion source pressure was roughly optimized by comparing the ion current
resulting from 3.6 ng of 11,17-keto dexamethasone obtained at three different pressures
of NH3. The pressures investigated corresponded to readings of 9000, 9500, and 9900
mtorr on the convectron gauge. The intermediate pressure corresponding to a reading of
9500 mtorr on the convectron gauge resulted in the most intense signal. Therefore, each
time the instrument was used, the pressure was adjusted to operate in this region. No
rigorous attempt was made to optimize the pressure because the sensitivity of the
instrument seemed to be adequate at each of these three pressures.

The ion source temperature used for the analyses was 120°C. As the ion source

 

7 6

temperature increases, the relative abundance of the fragments increase and the
abundance of the molecular ion decreases until none is present. As the temperature of the
ion source rises, the total ion current increases also. The source temperature of 120°C
was chosen in order to balance the tradeoff between the absolute abundance of the ion
current and the abundance of the molecular anion. At this temperature the relative
abundance of the molecular ion was about 10% of the (M-HF)' fragment ion. More
selectivity could be provided by monitoring both the molecular anion and this major
fragment ion.

The ionization voltage used was 100 eV. The ion current resulting from 3.6 ng of
11,17-keto dexamethasone was compared at ionization voltages of 70, 100, 120, and 150
eV. The greatest ion current was obtained at 100 eV, although the differences in ion
current at the different ionization voltages were not very signiﬁcant. The difference in
ion current obtained upon changing the ionization voltage seemed to be more pronounced
at the lower ion source temperature (100°C) than at the higher ion source temperature
(140°C).

The ionization current used for all analyses was 300 uA. The electron multiplier
voltage was 1.2 kV.

The direct capillary inlet was heated to 280°C. Splitless injection was used to
introduce samples. The injector temperature was 280°C. A 30 m x 0.25 mm x 0.25 pm
ﬁhn thickness DB-S fused silica capillary column was used for all analyses. The
temperature program employed was 60°C-260°C at 40°C/min, then to 280°C at 4°C/min.
After each analysis, the oven was ramped up to 300°C for 3-5 minutes to bake out any

remaining sample components. The GC column head pressure was 10 psi g.

77

B. Results and Discussion

The electron capture negative ion mass spectrum of 11,17-keto dexamethasone is
shown in Figure 3.3. The peak at m/z 330 represents the molecular anion, the peak at
m/z 310 represents the neutral loss of HF from the molecular species, and the peak at m/z
295 represents the loss of a methyl radical from the (M-HF)- ion.

The chemical oxidation product of dexamethasone, 11,17-keto dexamethasone, is
shown in the inset in Figure 3.3. Both the 16a- and 16B-methyl epimers of 11,17-keto
dexamethasone are observed. About 20% epirnerization occurs after formation of the 17-
keto group due to the acidity of the 16B-hydrogen alpha to the 17-ketone. The presence
of pyridine (from PCC), and possibly the sodium acetate, promote enolization of the
ketone, leading to epimerization. Without the addition of sodium acetate, however, the
extent of epirnerization is about the same. These epimers can be separated by capillary
gas chromatography. The presence of the two peaks (a well resolved doublet) can be
seen as an advantage because the pattern is easily recognizable in the presence of other
chromatographic peaks.

The limit of detection of a pure standard of 11,17-keto dexamethasone was about
160 fg with s/n of 7 (n = noise calculated at the top of the peak) by selected ion
monitoring. The ion current proﬁle at m/z 310 is illustrated in Figure 3.4. A blank
injection of solvent prior to the injection of 160 fg of 11,17-keto dexamethasone was
monitored to ensure that the signal observed for 160 fg of 11,17-keto dexamethasone was
not the result of residual 11,17-keto dexamethasone being "washed off" the injector or
septum. The detection limit is improved by about 50 times over that reported previously
by Her and Watson (4) with the HP5985 modiﬁed in-house for negative ion detection.
Also contributing to the improved sensitivity observed with the TSQ70 is the fact that the
GC column is directly interfaced. The HPS985 had an open split GC-MS interface which

decreased the amount of analyte entering the ion source.

 

 

.ocommﬁoamxow 982%.: .«o
8309.5 93 25350838 98.-S.S «o 883on 9.38 20m m.m 2:me

 

 

 

 

 

 

 

NE
3m 98 an . o 5 8m 8m 0
.._.. pb...r...-..p....._p. ..._....__...»_._.....
_

[mom 0 mam low

m -A IQHEE
18
r8
.2 0mm Ex: 18
roe

o8

 

79

‘00 m/z 310 s/n i 7

80 ' I 1

40"

20-

 

 

TT‘IIUUUITU'Ith'IIIVUUI'TTT‘U'IIIUUUIIII

11 11:30 12 12:30

time

Figure 3.4 GC-ECNI/MS selected ion current proﬁle from 160 fg of 11,17-
keto dexamethasone standard.

 

8 O

The calibration curve, shown in Figure 3.5a, was obtained by plotting the ratio of
the peak height at m/z 310 to that at m/z 319 ((M-HF)- from the M+9 internal standard)
for the 16-0t epimer of 11,17-keto dexamethasone against ng of dexamethasone added to
the calibration curve standards. Peak heights rather than peak areas were used to
construct the curve because of occasional difﬁculty in deﬁning the baseline due to co-
eluting interfering substances that produced ion current at either m/z 319 or m/z 310.
Linear regression of the data provided the equation to be used for quantiﬁcation of
dexamethasone in real plasma samples (y = 0.0259 + 0.0360x). The correlation
coefﬁcient for the curve was 0.998. The plot shown in Figure 3.5a conﬁrms the linear
relationship of the data throughout this range of dexamethasone concentrations.

The high y-intercept value is partly due to the presence of unlabelled
dexamethasone in the labelled internal standard, 13C6-2H3 dexamethasone. This internal
standard was received as a gift and, unfortunately, it was contaminated with unlabelled
dexamethasone upon arrival. Although the reagents available for synthesis of 13C-
labelled compounds are less than 100% enriched, the source of the unlabelled
dexamethasone in the 13C5-2H3 dexamethasone was probably not due to the synthetic
process. If this were so, molecular ions at m/z 331-338 would be observed. Molecular
ions at these masses are not observed, except for naturally occurring isotopic peaks at
m/z 331 and m/z 332.

The limit of detection for dexamethasone (LOD) in plasma may be calculated
from the y intercept of the calibration curve and the standard deviation of the y intercept.
A generally accepted criterion for the limit of detection is YIod = yb + 381) (5), where yb
is the value of the y intercept of the calibration curve or the calibration curve blank, sb is
the standard deviation associated with the measurement of the blank, and YIod is the
lowest y value (ratio of m/z 310 to m/z 319 in this case) that will reliably represent the
presence of the analyte. Duplicate SIM analyses of duplicate blanks (plasma samples

containing only internal standard), gave a standard deviation of 0.003 (%RSD = 11).

81

 

.o .o
O\ 00

I (m/z 310) /I (m/z 319)
o
.b

 

 

 

 

20

 

 

 

 

 

0.2
0.0 d ‘
0 10
ng of dexamethasone
_ b)
0.3
a y = 7.55 e-2 + 0.175 x R42 = 0.997
F;
E 0.2
’9?
m 0.1
N
E
0.0 I? v I v I r I V I
0.0 0.2 0.4 0.6 0.8

ng of dexamethasone

Figure 3.5 Calibration curves for determination of dexamethasone in plasma
b GC-ECNI/MS with SIM using a) 35 ng, and b) 3.5 ng of
1 C6-2H3-dexamethasone internal standard.

1.0

82

From the y intercept of the calibration curve and this standard deviation, the ylod is
calculated as 0.0350. This y value corresponds to an x value of 0.25 ng
dexamethasone/35 n g internal standard. This detection limit will be adequate to detect
dexamethasone reliably in all but the least concentrated plasma samples, since the
expected concentration range is from 0.1 to 4 ng/ml. This detection limit does not
approach the actual detection limit of the instrument (160 fg pure standard gives s/n = 7).
Rather, the limitation comes from the poor precision experienced in measurement of the
peak height ratio of the blank, and the low slope of the calibration curve. A rather large
excess of internal standardwas added in order to serve as a carrier as well as an internal
standard. The sensitivity of the assay might be increased if a structurally-similar
compound was used as the internal standard or a smaller amount of the labelled internal
standard was used. The smaller amount of internal standard would increase the slope of
the calibration curve, leading to a lower limit of detection and higher sensitivity. This
would allow more reliable detection of the lowest levels of dexamethasone in the plasma
samples. To demonstrate this, another set of standards containing 0.0, 0.209, 0.418, and
0.836 ng of dexamethasone was processed. The procedures used were identical to those
described above, except that only one-tenth the amount of internal standard (3.5 ng) was
added. No other compound was added to act as a carrier. The calibration curve obtained
from the GC-ECNI-SIM analyses of these standards is shown in Figure 3.5b. The
equation obtained upon linear regression of the data is inset in Figure 3.5b. The standard
deviation associated with the measurement of the peak height ratio of the blank was
0.009 (12% RSD). The limit of detection was calculated as 0.15 ng/ml from the equation
for the ylod that was stated above and the y-intercept of the calibration curve shown in
Figure 3.5b. This represents a 40% decrease in the LOD.

Another way to decrease the LOD would be to use a structural analog for the
internal standard, such as flumethasone. Flumethasone is identical to dexamethasone

except that this drug contains an additional ﬂuorine at C6. The use of a structurally

8 3

similar internal standard ought to result in a much lower blank value for the calibration
curve. When a structurally similar internal standard is used, the standard deviation of the
blank will be lower because the blank value is lower. This will be true as long as the
%RSD for the measurement of the blank is not higher when using a structural analog as
the internal standard. If the slopes of the calibration curves obtained using the labelled
internal standard and the structurally similar internal standard are equal, the decrease in
the LOD will be equal to the decrease in the standard deviation of the blank. And, the
decrease in the standard deviation of the blank ought to parallel the decrease in the value
of the blank, or y-intercept.

Unfortunately, for determinations of dexamethasone by oxidation and GC-ECNI-
MS, ﬁumethasone would not be a good choice for the internal standard. The ECNI mass
spectrum of the oxidized 11,17-keto ﬁumethasone contains a peak corresponding to m/z
310 from the (M-2HF)- ion. Because the oxidized products of dexamethasone and
ﬂumethasone cannot be completely separated by the GC capillary column, the response
at m/z 310 from the internal standard would overlap with the response at m/z 310 from
11,17-keto dexamethasone. This would result in a high blank value. Thus, no advantage
would be realized by using ﬂumethasone as the internal standard. If the molecular anions
were used instead of the fragment ions for the determination it would be possible to use
flumethasone. However, this would decrease the sensitivity of the analysis as the
abundances of the molecular anions are lower than the abundances of the fragment ions.

For future analyses it is advisable to use the smaller amount of isotopically-
labelled internal standard. However, it was not possible to repeat the analyses of these
DST plasma samples as all of the sample was consumed.

The accuracy of the assay for dexamethasone in plasma was determined by
triplicate analyses of duplicate samples "spiked" to a concentration of 2.09 ng/rnl. For
the determination of dexamethasone in real plasma samples, these data were included in

the calibration curve.

84

Table 3.1 Determination of dexamethasone in 1 ml blank plasma "spik "
with 2.09 ng of dexamethasone by GC-ECNI-MS with SIM.

sam le trial 1 trial 2 trial 3 mean s %RSD %error

7A 2.01 2.21 2.05 2.1 0.1 5.1 0.0

7B 1.97 1.79 1.72 1.8 0.1 7.1 12.4
overall 2.0 0.2 9.1 6.2

Table 3.1 indicates that the accuracy of the assay is reasonably good at the 2
ng/ml level. The mean of all six determinations was within 6.2% of true value. The
precision of the instrumental method is 5 to 7 %RSD. The precision of the overall
procedure is on the order of 10 %RSD. The variation between 7A and 7B is also cause
by the variation experienced in the additions of the dexamethasone and the 13C5-2H3
dexamthasone to the two aliquots of blank plasma.

Results from the SIM analysis of a real plasma sample (P5) are shown in Figure
3.6. Selected ion current proﬁles for the M- and (M-I-IF)- ions produced from
dexamethasone and the internal standard are shown. Substances present in the plasma
matrix produced very little ion current that interfered with the detection of 11,17-keto
dexamethasone. The peak height ratio of m/z 310 to rn/z 319 of the ﬁrst epimer was used
to calculate the dexamethasone concentrations of the ten DST plasma samples that were
analyzed. The concentrations are shown in Table 3.2. The peaks at m/z 330 and m/z
339, representing the molecular anions, were also monitored to provide more speciﬁcity
for the analyses.

The selected ion current proﬁle at m/z 324 for detection of oxidized 6B-
hydroxydexamethasone in plasma sample P5 is shown in Figure 3.7 along with the
proﬁles for m/z 310 and m/z 319. The arrows indicate the retention time of the 160t- and
l6B-methyl epimers of 6,11,17-keto dexamethasone. Because no calibration curve for
6B—hydroxydexamethasone was available, the concentration of 6B-hydroxy-

dexamethasone was not determined. However, these data show that it is possible to

 

 

85

Selected ion current profiles

 

 

 

 

 

 

 

m/231O
: m/2330
.é‘ -
(D
C .1
2
.5 '
9
f; ‘ m/z 319 1306:2113 internal standard '
72 - -
. (M-HF)
- _ A _ A _
1 m/2339 M'
*1!“r'rfTrTrrfT‘:r'rrr'r'r‘I‘r‘rrjfrﬁrﬁT
9 10 11 12 13 14 15

time

Figure 3.6 Selected ion current proﬁles from GC-ECNI/MS analysis of
plasma sample P5 from patient undergoing the DST.

 

 

86

 

 

.mm 0383. «Emma 5 53:89.8
.oaomanmeoaonEéo Ea .ocomaﬁoﬁaxoe-mm~bvm_
658525er 3528 .«o 2.2 -Eméé 05 wcucomoaou .vmm

23.3 .on NE a $an 2:88 :2 8823 $628.00

00mm ommm oomm omkm 80km omen 000m

3 28$

onwm Gawm

l—rbrrPPbbLbP—bupb-PblbbFPnbb_bbhlP—thLlFP~blF~®

 

 

 

:35.
mo+mx

eNmu~\e

 

 

 

 

 

 

Gamun\a

ION
16+.
r80

row

 

roaa X

 

87

detect 6B-hydroxydexamethasone in human plasma through oxidation and GC—ECNI-MS

analysis.
Table 3.2 Comprison of concentrations of dexamethasone in n g/ml of plasma
as determined by GC-ECNI-MS with SIM and by conventional
GC-EI/MS of the TMS-enol-TMS derivative.
Sample ECNI-SIMa EI-SIMb 31AM
(n=9)
P2 0.63(s=0.07, n=2) 0.83 0.66(s=0.01)
P5 0.62(s=0.09, n=3) 0.30 0.4 (s=0.1)
P11 3.2 (3:04, n=2) 0.50 0.9 (s=0.3)
P18 30 (S: 4, n=2) 28.01 21 (5:8)
P3 0.18(s=0.01, n=4) 0.24 0.3 (s=0.1)
P4 0.06(s=0.04, n=2) 0.47 0.2 (s=0.1)
P16 0.10(s=0.12, n=2) 0.15(LOD) 0.10(s=0.04)
P14 0.04(s=0.01, n=2) 0.15(LOD) 0.12(s=0.04)
P15 0.06(s=0.06, n=2) <LOD 0.10(s=0.06)
P17 (blank) <0 <LOD 0.08(s=0.05)

a standard deviation from replicate injections, representing instrumental precision

b The EI—SIM and RIA data were obtained from C. Kilts of Duke University as
part of an interlaboratory comparison.

C standard deviation represents variation among nine different laboratories

The LOD for the GC-EI/MS method was not determined by the criterion used to
calculate the LOD for the GC-ECNI/MS method. The LOD of the GC-EI/MS method is
reported as 0.15 ng/ml because the s/b ratio for the molecular ion of derivatized
dexamethasone by SIM was less than or equal to two at this concentration. For these
determinations, ﬂumethasone (6a-ﬂuoro-dexamethasone) was used as the internal
standard, which avoids the problem of a high blank value. The relative standard
deviation of the GC-EI/MS method was determined by processingten 5 ml aliquots 6f
plasma spiked to a concentration of 0.588 ng/ml. After derivatization, the standards were
reconstituted in 5 [.11 of hexane. All of this volume was used for one GC-MS analysis.
The RSD was 16% for the ten replicate determinations. Each of the real samples (5 ml of

plasma) was also analyzed only once by GC-EI/MS with SIM.

 

88

The limits of detection for the RIA methods varied between 0.02 ng/rnl and 0.3
ng/ml.

The ﬁrst four samples shown in Table 3.2 have concentrations that are above the
limit of detection by all of these methods of analysis. The concentration found for
plasma P3 is below the LOD for the ECNI-SIM method, but above the LOD for the other
methods. The concentrations found for the last four samples entered are near or below
the LCD for all the methods. The concentrations listed that are below the LOD are not
considered to be reliable. Nevertheless, they are included to compare the results of the
three methods at this level.

Sample P17 was a blank so it is expected that the concentration found should be
below the limit of detection. Sample P18 was taken two hours after drug administration
to provide a high level sample. The concentration found by RIA for P18 is low because
the upper limit of the RIA is 20 ng/ml. Concentrations above 20 ng/ml cannot be reliably
determined by the RIA. The other samples were taken either 9, 16, or 24 hours following
drug administration. The ranges of expected plasma concentrations 9, 16, and 24 hours
following drug administration are 1-5 ng/rnl, 0.1—2 ng/ml, and 0.1-1 ng/ml, respectively.
These ranges have been established in the literature by RIA. The column corresponding
to RIA in Table 3.2 represents the average of the RIA measurements obtained in nine
different laboratories.

The concentrations of dexmethasone found by these three methods agree well for
the ﬁrst ﬁve samples shown in Table 3.2, with the exception of sample P11. The value
obtained by the ECNI-SIM assay is signiﬁcantly higher than the values obtained by both
the EI-SIM and RIA assays. Possible reasons for the discrepancy could include a
determinate error such as addition of less internal standard or poor equilibration of the
internal standard with the plasma leading to greater loss of internal standard during the
extraction process. Co-eluting material giving a response at m/z 310 is probably not

responsible for the high value because the ratio of m/z 330 to m/z 310 is not abnormally

8 9
low as would occur if there were co-elutin g interference. The ratio of the intensity at m/z
330 to that at m/z 310 for sample P11 is comparable to the ratios obtained for the other
samples analyzed.

For the least concentrated samples, the results found by the three methods agree
fairly well considering that these concentrations fall below the LODs of the methods. An
exception is sample P4. The concentrations found by EI-SIM' and RIA are both
considerably higher than the concentration found by ECNI-SIM. It is possible that co-
eluting matrix components interfered with the determination of dexamethasone by the E1-
SIM method or the ECNI-SIM method. The high concentration found by RIA could be
due to cross-reactivity with matrix components. A low concentration by the ECNI-SIM
method would result from co-eluting interferences giving a response at m/z 319. In the
absence of co-eluting interference, the ratio of the ion current at m/z 339 to that at m/z
319 ought to be the same as the ratio of the ion current at m/z 330 to that at m/z 310. For
sample P4, these ratios are nearly identical. Therefore, co-elutin g interferences in the
ECNI-SIM analysis are probably not responsible for the difference between the methods.

A potential disadvantage of the oxidation method is that other molecules within
the sample matrix could be oxidized to the same product to which the analyte is
converted. However, the possibility of endogenous co-extracted material being oxidized
to an electrophilic substance having the same ECNI mass spectrum as 11,17-keto
dexamethasone is unlikely. The presence of the ﬂuorine atom in the dexamethasone
leads to a unique loss of HF. Few, if any, endogenous materials possess ﬂuorine. There
are no other possiblities for the loss of 20 mass units other than the loss of HF. It is
possible that other co-extracted non-endogenous materials, such as drugs taken by the
DST patients, could present an interference since many drugs do possess ﬂuorine and
may undergo the loss of HF. However, in the analysis of these plasma samples,
components from the sample matrix did not seem to interfere with the determination of

dexamethasone.

90

The other possibility for interference is the oxidation of dexamethasone
metabolites to the same structure to which dexamethasone is converted. The major
metabolite of dexamethasone is 6B-hydroxydexamethasone. Sixty percent of
administered dexamethasone is excreted as the 6B-hydroxy metabolite (6). The
oxidation product of 6B-hydroxy dexamethasone, 6,11,17-keto dexamethasone, does not
have the same mass spectrum as 11,17-keto dexamethasone; therefore, no interference is
expected. The other pathways for corticosteroids metabolism such as A-ring reduction,
llB-OH oxidation and reduction of the C-20 ketone are inhibited by the structural
features incorporated into this drug, the C1 double bond, the 9a-ﬂuorine atom, and the
l6a-methyl group. An additional 5-10% of dexamethasone is excreted as the 6B-
hydroxy, 20-dihydro metabolite. The oxidation product for this metabolite would be the
same as that for 6B—hydroxydexamethasone. The only metabolite which can be oxidized
to the same product as is dexamethasone is the ’20-dihyro metabolite. However, since

this is a minor metabolite, its interference is not expected to severly limit this technique.
C. Conclusion

In conclusion, the results obtained by the oxidation method with ECNI-SIM
detection do not differ greatly from the results obtained by the reference GC-EI-SIM
method (with the exceptions of two of the ten samples analyzed). The concentrations
reported at or below the LOD of the GC-MS methods are not sufﬁciently reliable to
make a comparison. To determine whether the RIA is valid at the lowest levels, the LOD
of the SIM methods must be decreased. As demonstrated above, the addition of less
isotopically-labelled internal standard ought to lower the LOD of the GC-ECNI-MS

assay.

9 1
III. MS/MS Analysis of Plasma for Dexamethasone

Although the matrix components did not obscure the GC-ECNI-MS determination
of dexamethasone, at the lowest levels of dexamethasone in plasma, interferences from
the sample matrix may become more prominent in the SIM assay.

The additional selectivity available through tandem mass spectrometry has been
demonstrated in the determination of compounds present in complex biological mixtures
(7). In triple quadrupole mass spectrometry, the ﬁrst mass analyzer, Q1 selects an ion
from the desired compound and ﬁlters out all others. This selected parent ion passes into
the second quadrupole, Q2, where fragmentation is induced by collisionally activated
dissociation (CAD). The second mass analyzer (Q3) can be scanned to provide the
daughter ion spectrum of the selected precursor parent ion and yield structural
information. To maximize sensitivity: in a quantitative analysis, the selected reaction
monitoring technique (SRM) is used. In SRM both mass analyzers are set to pass a
constant mass, offset by the neutral loss that the parent ion undergoes in the CAD
process. Often, positive chemical ionization, a soft ionization technique is employed
generating an abundant protonated molecule species which is selected as the parent mass.
The most abundant stucturally informative fragment is chosen as the daughter mass. This
results in greater speciﬁcity than is achieved in the analogous SIM experiment used in
GC-MS. Generally, the gains in selectivity associated with MS/MS markedly improve
the detection of analytes in complex matrices. There have been a few reports of using
ECNI GC-MS/MS for quanititative purposes (8). Recently, a quantitative assay for
Leukotriene B4 (LTB4) in synovial ﬂuid was developed based on negative ion GC-
MS/MS detection of the pentaﬂuorobenzyl ester di-t-butyldimethylsilyl ether derivative
of LTB4. The selectivity of the ionization method (ECNI) in combination with MS/MS
allowed quantiﬁcation of lower levels of LTB4 in the matrix than would have been

possible using ECNI-SIM.

92
In this section, MS/MS, speciﬁcally selective reaction monitoring, is evaluated
and compared with selected ion monitoring for the determination of dexamethasone in
plasma of patients undergoing the dexamethasone suppression test. In addition, the
feasibility of eliminating the gas chromatographic step of the MS/MS analysis and
introducing the sample by the direct inlet probe (DIP) is investigated. If the
dexamethasone can be adequately determined with DIP sample introduction, the analysis

would be simpliﬁed and the time required for sample throughput would be reduced.

A. Experimental

The samples used for MS/MS analyses are the same oxidized plasma samples
prepared for the SIM analyses described in the previous section.

All data were collected on the Finnigan triple stage quadrupole (TSQ 70) mass
spectrometer equipped with the Varian 3400 gas chromatograph. The ion source
conditions were chosen to maximize the abundance of the parent ion for the MS/MS
analyses. The molecular anion of the oxidized dexamethasone was at the greatest
absolute abundance when the ion source temperature was maintained at 100°C and the
electron energy was between 70 and 100 eV (70 eV was used for the SRM analyses). As
for the SIM analyses, ammonia was used as the modifying gas at pressure of about 1 torr,
and the the ionization current was 300 mA. An electron multiplier voltage of 1.4 kV was
used.

Samples were introduced into the ion source by gas chromatography in which a
DB-S capillary column (30m x 0.25mm) was directly interfaced to the mass
spectrometer. Helium was used as the carrier gas with a column head pressure of '12
psig. Following splitless injection, the GC oven was ramped from 120°C to 260°C
degrees at 40°C/min and 260°C to 280°C at 4°C/min. For the previous SIM analyses an

initial temperature of 60°C had been used. It was found that using an initial temperature

 

 

 

 

93
of 120°C shortened the retention time of the 11,17-keto dexamethasone while
maintaining good chromatographic resolution of the or and [3 epimers. Also, memory
effects from injection to injection appeared to be lessened by starting the temperature
program at the higher temperature.

Samples were also introduced into the ion source via the direct probe inlet system
available with the TSQ70 instrument. After insertion of the probe, the temperature was
ramped from 35°C-200’C at a rate of 70°C/min and held at 200°C for two minutes.

The conditions for collisionally activated dissociation of the molecular anion of
11,17-keto dexamethasone were optimized by varying both the collision gas pressure and
collision energy. Optimum conditions for which the daughter ion, (M-HF)-, was at its
greatest absolute abundance occurred when argon was introduced into the second
quadrupole at a pressure of 1.4 mtorr, and the collision energy was set to 3 eV.

For the SRM studies the ion current from a neutral loss of 20 11 (HF) from mass
330.2 and 339.2 were monitored with a 1.0 u window and a dwell time of 50

milliseconds.

B. Results and Discussion

When the molecular anion of 11,17-keto dexamethasone is chosen as the parent
and subjected to collision with argon in Q2, the predominant fragment is, as expected,
(M-I-IF)-. The CAD spectrum of the molecular ion is shown in Figure 3.8. Minor peaks
are observed at m/z 295 and m/z 280. These represent (M-l-IF-CH3)- and (M-HF-
2CH3)-, respectively. Conditions for CAD were sought that would maximize the
detection of m/z 310.

The inﬂuence that the collision energy had on the absolute abundance of the (M-
I-IF)- fragment ion was examined at three pressures of argon in the Q2 collision cell.

Source parameters had been previously adjusted to yield the greatest amount of the M-

 

94

d<0 £5 wngSZUm B 8:850 ocommﬁoﬁmxoe
98:1: me some“ 3:822: 05 .«0 85.58% com Seam—EC w.m 053m

 

 

 

 

NE.
.omm . om...” . 05 0mm 8w 8.0.
» pbbp >>.L >>>.-.p.- .pr....>> >.-—»>P»> .p.>»r.>l«_>>..~» O
mam 0mm
row
Wow
L2
0mm
18
his
12:

 

 

05

95

parent ion of oxidized dexamethasone. During CAD, only a few electron volts of energy
are required to induce fragmentation to the (M-HF)' species. This suggests that the (M-
I-IF)' species is more stable than the molecular parent ion. Multiple collision conditions,
generated with higher collision gas pressure is favorable, presumably to ensure that all
the parent ions do undergo collision with argon. A plot of the absolute abundance of m/z
310 at each energy and pressure that was investigated is shown in Figure 3.9. From
Figure 3.9, it is evident that the optimum detectability of the daughter fragment occurs at
a collision gas pressure of 1.4 mtorr argon and a collision energy of 3 eV.

In order to determine the extent to which the collision gas present in Q2
attenuated the total ion beam reaching the detector, the abundance of the total ion current
resulting from MS/MS of m/z 330 was measured with and without collision gas in Q2.
This was repeated at a series of collision energies, between 2 and 20 eV. The peak areas .
of the chromatographic proﬁles were measured and compared to determine the collection
efﬁciency at each energy, with and without collision gas. The collection efﬁciency is
deﬁned as the percentage of the ion current transmitted, either as unreacted parent ion or
as daughter ions, through the second quadrupole collision cell. The ratio of the ion
current detected at a given collision energy with and without gas in Q2 is the collection
efﬁciency. At collision energies less than 5 eV, the collection efﬁciency was near 100%,
however at higher energies it was diminished. This was true for the three pressures that
were examined. This decreased efﬁciency may be attributed to poorer transmission
properties of the rf-only second quadrupole when operated at high collision energy since
the scattering angle of a collision increases with collision energy. In addition, at higher
collision energy electron detachment may occur and result in the formation of
undetectable neutral species. To avoid poor collection efﬁciency due to these effects, the
SRM experiments were conducted at low collision energy conditions.

Figure 3.10 illustrates the improved selectivity of SRM over SIM. These data

were obtained by the extraction and oxidation of a 2 ml plasma sample containing 40 n g

Figure 3.9

96

 

 

 

 

 

 

 

 

 

800000 I —I f I ' I I I
- (l ' E 0.6 mtorr argon
600000 " ‘
§ .- ° 0.4 mtorr argon
I,” ’ . 1.1 mtorr argon
7:: 400000 - -
:1 p
200000 " ‘
I o \E‘. i :
0 . r . ‘9
0 15 20

 

collision offset energy (eV)

The abundance of the (M-I-IF)- ion, m/z 310, from CAD of the
molecular anion of 11,17-keto dexamethasone at different collision
offset energies and collision gas pressures.

 

97

of internal standard and a dexamethasone concentration of 0.6 ng/ml. About 20 pg of
oxidized dexamethasone extracted from the plasma sample and 0.8 ng of the oxidized
isotopically labelled dexamethasone were injected onto the column for each analysis.
The sum of the selected ion currents, due to the M- and (M-HF)- of 11,17-keto
dexamethasone and the 11,17-keto analog of the 13C6-2H3-dexamethasone, is shown in
Figure 3.10a. The sum of the selected reaction current obtained by SRM of the M- of
11,17-keto dexamethasone and of the 11,17-keto internal standard is shown in Figure
3.10b. In Figure 3.10b, all the interfering ion current is eliminated. The only signal
present is that from the 11,17-keto dexamethasone and its 16B-CH3 epimer. The absence
of background makes the determination of the peak height or the integration of the peak
area straightforward and increases the precision of replicate analyses. Determination of
the baseline position for peak height or area measurement from the SIM data in Figure
3.10a is more arbitrary which could lead to poorer precision than in SRM.

The calibration curve obtained by SRM, shown in Figure 3.1 la is very similar to
the calibration curve obtained by SIM that was shown in Figure 3.5a. The limits of
detection of dexamethasone in plasma by the two techniques appear to be similar. The
LOD of dexamethasone in plasma by SRM was calculated as 0.31 ng/ml from the y—
intercept of the calibration curve and the standard deviation of the blank. It is likely that
the LCD can be decreased by decreasing the amount of internal standard that is added.
As found for SIM, the precision with which the blank can be measured, in addition to the
small slope of the calibration curve, limit the sensitivity of the SRM assay. An additional
experiment was done to demonstrate how the LOD for dexamethasone in plasma can be
decreased by the addition of less internal standard. Preparation of these standards, in
which ten times less internal standard was added, is described in part II of this chapter.
The calibration curve obtained from these standards by SRM is shown in Figure 3.11b.
Based on this calibration curve and the standard deviation obtained upon measuring the

blank (0.005, 11 %RSD), an LOD of 0.05 ng/ml was calculated! This is three times

Figure 3.10

98

 

 

 

 

 

 

 

 

 

 

 

 

8)
m/z 310+330+319+339 SIM
100‘ '
80
60-
40‘
20-
Q If:'I'I'TWﬁwrﬁiI'I'I'I‘I*T‘ﬁ'I'I‘I'I‘I‘
t) 11 12 13 14 _15 16
b)
100- m/z 310+319 SRM r
80‘
M:->(M-HF)=
60-1
40,
20- w
0*1* 'U'IfI‘V'T‘I'V‘T'I‘U‘ ' ‘U'I'U'V'l' 'r
10 11 12 13 t 13 15 1‘6

Comparison of selected ion current proﬁles obtained by a) GC-
ECNI/MS using SIM, with those obtained by b) GC-ECNI/MS/MS
with SRM, for' the detemination of dexamethasone in a real plasma

sample.

 

99

 

 

 

 

 

 

 

 

 

0.7 W I ' l ' F +
A 0.6 -
‘33 ’ 1
m 0.5 - '
N I
E 0.4 - -
: I- a I:
§ 0.3 - '1
m
N 0.2 - ‘
E 01 _ y=2.9le-2+3.43e-2x RA2=0.997 _
0 0 1 I .L l 1 I 1 q
0 5 10 15 20
ng dexamethasone
A 0.3 r
O\
'_I
m
E
:1 0.2 - '
\
§ 1
m
N
E 0.1 - '
y = 5.25e-2 + 0.293;: RA2 = 0.985
0 0 _ , . r . . l . . I . . r r .
0.0 0.2 0.4 0.6 0.8 1.0

ng dexamethasone

Figure 3.11 Calibration curves for determination of dexamethasone in plasma
b GC-ECNI/MS/MS with SRM using a) 35 ng, and b) 3.5 ng of
1 C6-2H3-dexamethasone internal standard.

1 0 0
lower than the LOD calculated from the calibration curve obtained by SIM of these
standards. The improved LOD is the result of a lower standard deviation and a steeper
sloped calibration curve by SRM.

Some of the same plasma samples from patients undergoing the dexamethasone
suppresion test analyzed by SIM were analyzed by SRM. These results are shown in
Table 3.3. The concentrations found using the direct probe sample introduction are also
shown in Table 3.3. The concentrations determined for these ﬁve samples are similar by
all three modes of analysis. Each sample (1.5 ml of plasma) was processed once. Also
shown in Table 3.3 are the concentrations determined for the duplicate 1 ml plasma
standards that were "spiked" with 2.09 ng of dexamethasone. The concentrations are
reported in ng/ml of plasma +/- one standard deviation. The number of replicate
instrumental analyses (n) is three unless indicated otherwise. The standard deviations

shown represent the precision of the instrumental analysis.

Table 3.3 Comparison of concentrations of dexamethasone obtained by GC-
ECNI-SIM, GC-ECNT-SRM, and DIP-ECNI-SRM.
Sample GC-SIM GC-SRM DIP-SRM
P2 0.63 :1: 0.07 (n=2) 0.76 i 0.01 0.68 i- 0.03 (n=4)
P5 0.62 :l: 0.09 0.69 :l: 0.02 0.35 i 0.01
P11 3.2 i 0.4 (n=2) 4.0 i 0.4 4.4 i 0.4
P18 3014 (n=2) 36:1: 3 35 i 1.0
P3 0.18 i 0.01 (n=4) 0.29 i 0.01 0.15 i 0.01
7A* 2.1iO.1 1.9i0.1 2.11:0.05
78* 1.8 i- 0.1 1.82 i 0.04

*duplicate spiked plasma sample (2.09 ng/ml), measured in triplicate

The calibration curve obtained using the DIP with SRM was similar in
appearance to the curves obtained by GC-SIM and GC-SRM. The concentrations
determined by DIP-SRM are comparable to those found by GC-SIM and GC-SRM. This
indicates that the added selectivity of MS/MS can allow the elimination of the

chromatographic inlet. This will greatly reduce analysis time. For instance, to obtain the

1 O 1
results shown in Table 3.3, required ﬁve hours using the DIP. To obtain the GC-SRM
results, 15 hours was required.

Other researchers have noted that the chromatographic step is necessary in
negative ion SRM when a general derivatization method is used to prepare an
electrophilic derivative of the sample. This is because the electron capture responses of
the matrix components are enhanced as well. Without sufﬁcient chromatographic
separation, the low level of analyte present in the ion source is unable to effectively
compete with the matrix for the pool of thermal electrons. However, oxidation
selectively converts the analyte to an electrophilic analog without greatly enhancing the
response of the matrix. Therefore, introduction of an oxidized sample by direct probe is
viable because the matrix is more transparent to electron capture and does not hinder the

electron capture ionization of the analyte.

C. Conclusion

The use of selected reaction monitoring for the detection of oxidized
dexamethasone in a plasma matrix yields similar results to those obtained through the use
of selected ion monitoring. The additional selectivity of SRM allows the direct probe
sample inlet to be used. However, when the GC inlet is used, SRM offers no signiﬁcant
advantage over SIM. This is due to the high level of selectivity already achieved through
the oxidation and the electron capture negative ionization, as well as the chromatographic

inlet.

SAP?!"

REFERENCES

K. Minagawa, Y. Kasuya, S Baba, G. Knapp, and J. Skelly, J. Chromatogr., 343,
231-237, 1985.

M. T. Lowy and H. Y. Meltzer, Biol. Psychiatry, 22, 373-385, 1987.

M. Oehme, D. Stock], and H. Knoppel, Anal. Chem, 58, 554-558, 1986.

G. R. Her and J. T. Watson, Anal. Biochem, 151, 292-298, 1985.

American Chemical Society Committee on Environmental Improvement and
Subcommittee on Environmental Analytical Chemistry, Anal. Chem, 52, 2242-
2249, 1980.

M. J. Green and B. N. Lutsky, "Anti-Inﬂammatory Steroids" in CRC Handbook of
Cardiovascular and Anti-Inﬂammatory Agents, M. Verderame, ed., CRC Press,
Inc., Boca Raton, FL, pp. 1-26, 1986.

F. W. McLafferty, ed., Tandem Mass Spectrometry, Wiley, New York, 1983.

M. Dawson, C. M. McGee, P. M. Brooks, J. H. Vine, and T. R. Watson, Biomed.
Environ. Mass Spectrom, I 7, 205-211, 1988.

102

CHAPTER 4. INVESTIGATION OF AN OXIDATION BASED ASSAY
FOR 6B-HYDROXYCORTISOL IN URINE;
COMPARISON WITH A CONVENTIONAL GC-MS METHOD

1. Introduction

This chapter describes the investigation of an oxidation-based, GC-ECNI/MS
assay for urinary 6B—hydroxycortisol. Results from this assay are compared to results
obtained by a conventional GC-MS method based on formation of the methoxime-
trimethylsilyl ether derivative of 6B-hydroxycortisol. The use of the direct inlet probe
(DIP) is investigated for the ECNI/MS analysis to determine whether the overall
selectivity provided by the oxidative treatment of the sample and the mode of ionization

is sufﬁcient to allow elimination of the chromatographic inlet.

A. 6B-Hydroxycortisol and Enzyme Induction

The compound 6B-hydroxycortisol (4-pregnene-6B,1lB,17a,20-tetrol-3,20-dione)
is a minor metabolite of cortisol that is excreted in urine in free form (not conjugated
with sulfate or glucuronate). Normal adults excrete between 200 and 400 ug of 6B-
hydroxycortisol per day. This amount represents only 2% of the total weight of cortisol
metabolites excreted in urine each day. Therefore, selective methodology is needed to
detect the relatively small quantity of 6B-hydroxycortisol within the urine matrix.

Unlike the other pathways for the metabolism of cortisol that involve reduction of
the 4-ene-3-one moiety followed by conjugation with glucuronic acid, the formation of
6B-hydroxycortisol (6B-OHC) occurs through oxygenation by hepatic cytochrome P-450
microsomal enzymes. Upon induction of the P-450 enzymes, the excretion of 6B-OHC

increases markedly. Enzyme induction'is deﬁned as a state in which the rate of enzyme

103

l 04
protein synthesis is stimulated to a level above the normal rate of synthesis. Enzyme
induction can be triggered by exposure to a variety of xenobiotic substances. Some
examples include drugs, such as phenobarbital, and pesticides, such’as DDT (1). PBBs
(2), polycyclic aromatic hydrocarbons (1), and styrene (3) also have been reported to
induce microsomal enzymes.

The measurement of urinary 6B-OHC has been proposed as anon-invasive means
to monitor microsomal enzyme induction (1). The measurement of 6B-OHC has also
been used to assess adrenocortical function in health and disease (4), and to assess the
abnormality of cortisol metabolism in cancer patients (5), and as a diagnostic test for
hypercortisolemic states (6).

The major microsomal steroid hormone 6B-hydroxylase enzyme was recently
identiﬁed as P-450N1: (7). These authors suggest that the level of urinary 613-
hydroxylated steroids is likely to reﬂect the activity of only one or a limited number of
closely related P-450 enzymes. The 6B-hydroxylated steroids are likely to be useful
indicators of the metabolism of xenobiotic substances that are principally metabolized by
P-450NF. These would include many compounds such as dihydropyridine calcium

channel blockers, quinidine, erythromycin, benzphetamine, and aldrin.

B. Methods for Determination of Urinary 6B-Hydroxycortisol

There are many analytical methods available for the determination of 6B-
hydroxycortisol in urine. The older methods, based on rigorous isolation of 6B-OHC
with detection by a colorimetric technique (8), have been replaced by HPLC and RIA
methods (9,10). A study comparing results obtained by an HPLC method and by RIA
found a good correlation between the two methods (11). A line with a slope of 0.89 and
a y-intercept of 22.03 was obtained by plotting the amount of 6B-OHC found by RIA on
the x-axis against the amount found by HPLC. At higher levels of 6B-OHC, the RIA

 

 

1 05

values tended to be higher than the HPLC values. The higher levels of 6B-OHC were
found in samples taken from subjects that had taken the drugs rifampicin and antipyrine
in order to cause enzyme induction. The higher values found by RIA could be an
indication of cross-reactivity with other substances that also increase upon enzyme
induction. At the lower levels of 6B-OHC, the HPLC values were generally higher than
the RIA values. This could be an indication of a co-eluting interference in the HPLC
assay. At lower levels of 6B-OHC, interference from co-eluting material would become
more pronounced.

Other methods mentioned in the literature for the determination of 6B-OHC
include GC-FID and GC-MS with the formation of the MO-TMS derivative (12). A GC-
ECD assay based on oxidation of 6B-OHC (13) is also mentioned. These methods have
not been used extensively.

The GC-ECD assay has not found extensive use. Criticism of the method
included the possibility that other components of the urine matrix could be oxidized to
the same product as 6B-hydroxycortisol. Also, the poor chromatographic resolution '
available with the packed column might not separate the oxidized 6B-hydroxycortisol

from other similar oxidized components that have an electron capture response.

C. Proposed ECNI-MS Method for Determination of 6B-Hydroxycortisol

The basis for our proposed assay for 6B-hydroxycortisol is the chemical oxidation
of 6B-hydroxycortisol to an electrophilic product. Figure 4.1 shows the oxidative
conversion of 6B—hydroxycortisol to 4-androstene-3,6,l1,17-one. The 4-ene-3,6-dione
moiety incorporated into the molecule is primarily responsible for the high electron
capture response. Cleavage of the C17 sidechain makes the 6B-hydroxycortisol analog
thermally stable during GC analysis.

 

 

 

 

1 o 6
CHz-OH
c=o
HO "OH
oxidation
OH

 

 

 

 

 

I IIIIIIIUIFTFIIITIIIIIT—T‘ITIIUIIITTI
2802852932953D3530315320325330335340345350

Figure 4.1 a) Oxidative conversion of 6B—hydroxycortisol to electrophilic
product. b) ECNI mass spectrum of oxidized 6B-hydroxycortisol.

 

 

 

 

 

l 0 7

The ECNI mass spectrum of oxidized 6B-OHC is also shown in Figure 4.1. The
only peak in the spectrum, at m/z 314, represents the molecular anion.

The additional selectivity of the mass information provided in ECNI-MS analysis
eliminates the GC-ECD problem of interference from co-eluting compounds having
electron capture response. However, as in the GC-ECD assay, the criticism of an ECNI-
MS method lies in the question of whether the amount of 4-androstene-3,6,11,17-one
formed upon oxidation of an urine extract truly represents the amount of 6B-
hydroxycortisol in that extract. A literature search was done to help answer this question.
According to the literature that was surveyed, 6a-hydroxycortisol is the only compound,
of those normally excreted in human urine in amounts comparable to 6B-hydroxycortisol,
that may be oxidized to 4-androstene-3,6,l1,17-one. The compound 6a-hydroxycortisol
normally comprises about 5-10% of the total 6-hydroxycortisol that is excreted (8). The
measurement of the total 6-hydroxycortisols has also been used as an index of

microsomal enzyme induction (14).

11. Experimental and Results

This section contains the details of the method development and the results
obtained for the determination of 6B-hydroxycortisol in urine. Three studies are
discussed here. These include a preliminary study, a study using urine from control and
phenobarbital treated guinea pigs, and a study using human urine from a volunteer before

and after receiving niacin treatment to lower cholesterol.

 

108

A. Preliminary Study

This study was done to assess the feasibility of oxidation and ECNI-MS for the
detection of 6B-hydroxycortisol in human urine. The ECNI mass spectra of other
oxidized components of urine could produce a response at m/z 314. This preliminary
study was also done to determine the extent of interference at m/z 314 from the oxidized
urine matrix. The sample preparation, the ECNI parameters, and the results are

discussed.

1. Extraction of Urine

A large volume (120 ml) of human urine was used for this study to ensure
detection of 6B-hydroxycortisol as well as other components of the urine matrix.

The steroids were extracted from the urine by reverse-phase, solid-phase
extraction. After passing the sample through the C18 solid-phase extraction (SPE)
column, the column was washed with 5 ml of distilled water. The steroids were eluted
with 3 ml of methanol. Methanol elutes both conjugated and free steroids. The methanol
was evaporated and the residue reconstituted in 1 ml of water. The water was extracted
three times with 1-2 ml of ethyl acetate to isolate the free steroids from the conjugated
steroids. The combined ethyl acetate extracts were divided into three portions. One of

the portions was used for the following oxidation.

2. Oxidation

The ethyl acetate extract was reconstituted in 1 ml of CH2C12 for the oxidation.

An excess of the oxidation reagent, pyridinium chlorochromate (PCC) ﬁnely ground up

with celite, was added to the solution. Sodium acetate was added to buffer acid liberated

1 O 9
during the oxidation. The reaction was stirred at room temperature for 4.5 hours. The
excess oxidation reagent was removed by passing the reaction mixture through a silica
(500 mg) SPE column. Four milliliters of ethyl acetate were used to elute the oxidized
6B-hydroxycortisol. The ethyl acetate was evaporated and the residue reconstituted in

100 111 ethyl acetate.

3. GC-ECNI/MS and GC-EI/MS Analysis of Oxidized Urine Extract

The oxidized urine extract was analyzed by GC-ECNI/MS and by GC-EI/MS.
The instrument used for these analyses was an I-IP5985 quadrupole mass spectrometer
that had been modiﬁed in this laboratory to allow negative ion detection. Preliminary
work with standards of oxidized 6B-OHC indicated that sensitivity increased with
increasing source temperature. The total ion current (TIC) from oxidized 6B-
hydroxycortisol was measured at source temperatures of 100°C, 150°C, and 200°C. The
total ion current at a source temperature of 200°C was forty times greater than the TIC at
a temperature of 100°C. A source temperature of 200°C was used for this study.
Methane was used as the modifying gas. The pressure was adjusted using an automatic
pressure controller (APC) valve that was feedback-regulated by the ionization gauge. A
reading of 2.7 x 10 - 4 on the ionization gauge situated below the ion source housing was
maintained by the APC valve. This ion gauge reading corresponded to an ion source
pressure of about 0.3 torr.

The GC-MS was equipped with an on-column injector and a "megabore" DB-5
capillary column (25 m x 0.53 mm). The column efﬂuent was introduced to the mass
spectrometer via an open-split interface. About one-half to one-ﬁfth of the column
efﬂuent was transferred to the MS by this interface; the exact amount depended on the
ﬂow rate of the carrier gas through the column. At higher column ﬂow rates, a greater

proportion of the efﬂuent is split out because the ﬂow through the transfer line into the

 

 

l 1 0
mass spectrometer is limited by the narrow inner diameter of the tubing (0.11 mm). The
column was operated at a head pressure of 5 psi. This corresponded to a ﬂow rate of
about 5 ml/min.

For the GC—EI/MS analysis, conventional mass spectrometer conditions were
used (ion source temperature = 200°C, 70 eV). The results of the analysis of the oxidized
urine extract by GC-EI/MS are shown is Figure 4.2a. The use of the non-selective
electron impact ionization gives an indication of the complexity of the sample because all
of the sample components are ionized. The peak corresponding to oxidized 6B-
hydroxycortisol (indicated with an arrow) indicates that this is only a minor component
of the sample.

The results of the analysis of the oxidized urine extract by GC—ECNI/MS are
shown in Figure 4.2b. Under ECNI conditions, oxidized 6B-hydroxycortisol is
respresented by one of the major peaks in the reconstructed total ion current (TIC)
chromatogram. Very few of the other components in the matrix give an appreciable
response.

Comparison of the reconstructed total ion current chromatograms in Figure 4.2
demonstrates the high degree of selectivity for 6B-hydroxycortisol that is imparted by the
oxidative treatment of the sample and subsequent analysis by ECNI. The increase in
absolute sensitivity of detection is also evident by comparison of the 100% TIC for the
molecular ion of oxidized 6B-hydroxycortisol by El and by ECNI, shown in Figure 4.23

and 4.2b, respectively. .

4. Investigation of Sample Introduction by Direct Inlet Probe for ECNI/MS

The reconstructed mass chromatogram at m/z 314 from the GC-ECNI/MS
analysis of the oxidized urine extract is also shown in Figure 4.2b. This mass

chromatogram indicates that the only major response at m/z 314 is from the molecular

 

 

 

Figure 4.2

111

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

) 100°/.=4328 “0
"'I'I‘I“'1"‘I"'ITTTI"'1"'17"I"‘I'I'IH'IH‘ITTT
4 5 6 7 6 9 10 11 12 13 14 15 16 17
100%=11 m/z314
ITIV‘"I'1_"T"‘I'JTT“‘I“‘Iﬁ'1**‘7“‘l"'I"‘I"'1"
4 5 6 7 6 9 10 11 12 13 14 15 16 17
T1me(minutes)
) J100%=10384 TIC
A l LMAA A; A
T‘TTT'TI‘ITTT' 17"I'Iﬁn'l‘7‘l7fﬁjﬁ'.‘T'In'lh7TT‘
4 5 6 7 6 O 10 11 12 13 14 15 16 17
1 100%-3370 111/2314
11 -. 1 1- 1 _1' 1
I I‘ I ‘1 'IT' I "I "1"‘Iﬁ71ﬂ'17‘l‘ ‘ITT
4 5 6 7 8 9 10 11 12 13 14 15 16 17
Time (minutes)
Reconstructed total ron current chromato

chromatogram at m/z

gram and mass

314 from a) GC-El/MS and b) GC-ECNI/MS

analyses of an oxidized urine extract.

 

 

l 12
anion of oxidized 6l3-hydroxycortisol. Because no other compounds in the sample matrix
produce an appreciable response at m/z 314, it may be possible to use the direct inlet
probe (DIP) and selected ion monitoring to determine oxidized 6B-hydroxycortisol in
urine. This would simplify the analysis and reduce the time required.

It was found that the best signal-to-background (s/b) ratio could be achieved by
rapidly heating the DIP. Rapid heating of the DIP also reduces the possibility of sample
decomposition. When the probe is heated rapidly, the rate of desorption is greater than
the rate of decomposition (15). The maximum heating rate allowed by the data system
was 30°C/min. To achieve a faster heating rate, the heating cable was disconnected from
the probe and the initial probe temperature was set to 280°C through the data system.
Then, after insertion of the probe into the ion source, the cable was reconnected. A
heating rate of about 200°C/min can be achieved by this procedure. The desorption
proﬁle shown in Figure 4.3 was obtained for the analysis of the oxidized urine extract by
ECNI with SIM, using rapid heating. The use of the slower heating rates available
through the data system resulted in much broader and slower desorption proﬁles. A
sharp, non-tailing desorption proﬁle is much easier to integrate, leading to better

precision in quantitative analyses.

5. Conclusion

This study indicated that GC-ECNI/MS provides good selectivity and sensitivty
for the detection of the oxidation product of 6B-hydroxycortisol within the urine matrix.
The feasibility of using this methodology to determine urinary 6B-hydroxycortisol was

demonstrated.

 

 

 

113

 

SIM at m/z 314

  

Area = 76400

 

 

 

Time (minutes)

Figure 4.3 Selected ion current proﬁle obtained by DIP-ECNI/MS analysis of
an oxidized urine extract. The molecular anion of oxidized 6B-
hydroxycortisol corresponds to m/z 314.

 

 

114

B. Determination of 6|3-Hydroxycortisol in Guinea Pig Urine.

A study was done in which the urines of a control and an "induced" guinea pig
were analyzed for 6B-hydroxycortisol. The "induced" guinea pig was given
phenobarbital, which is known to cause P-450 enzyme induction. There was indication
in literature that guinea pigs excrete 6B-hydroxycortisol and that the amount excreted
increases upon P-450 enzyme induction with phenobarbital (16). This study was
undertaken to determine whether the oxidation methodology for the determination of 613-
hydroxycortisol would indicate that enzyme induction had occurred in the phenobarbital
treated animals. The guinea pig urines were also analyzed by the conventional GC-MS
method for steroid analysis, in which the methoxime-trimethtyl silyl ether derivatives are
made. The results of the conventional method were compared to the results obtained by

oxidation and ECNI-MS.

1. Sample Preparation

The protocol followed for the sample preparation is illustrated by the scheme
shown in Figure 4.4. Three-quarters of the extract was used for the MO-TMS procedure
because it was anticipated that the EI/MS detection might not be as sensitive as the ECNI
detection. The control and the induced samples were processed in triplicate.

The internal standard, 4-pregnene-6B-ol-3,l1,20-one, was chosen because: 1) this
compound is oxidized to an electrophilic product that was not found in oxidized guinea
pig urine by a preliminary analysis; 2) the compound undergoes the reactions to form an
MO-TMS derivative; 3) the compound is structurally similar to 6B-hydroxycortisol and
ought to behave similarly throughout the isolation procedures, although it is somewhat
less polar than 6B-hydroxycortiso1; and 4) it was not found in guinea pig urine by a

preliminary analysis using the MO-TMS method.

 

 

 

115

SAMPLE PREPARATION

2% of total daily urine

+ 680' ng internal standard

1

C18 Sep-PakTN extraction

solvent extraction

I 34

75% aliquot ' 25% aliquot
Methoxyamine, 60°C, 1hour PCC oxidation, 5 hours
Trimethylsilylimidazole, 100°C, 4 hours silica Bond-Elutm

1

Sephadex LH-20 column

Sep-Pakm is a trade name of Waters Associates
Bond-Elutw is a trade name of Analytichem International

Figure 4.4 Sample preparation scheme for the determination of 6B-
hydroxycortisol in urine by MO-TMS derivatization and by
oxidation.

1 1 6
Standards containing internal standard and 6B-hydroxycortisol were prepared in
duplicate for the construction of a calibration curve. The procedures applied for the

analysis of the guinea pig urine were also used to process the standards.
2. Instrumental Analysis

These samples and standards were analyzed with three different instruments.
Analysis of the MO-TMS samples by GC-EI/MS with SIM was initially performed on
the HP5985 mass spectrometer. Selected ion monitoring was used to monitor the ion
current from the molecular ion and the (M-OCH3)+ ion of 6B—hydroxycortisol MO-TMS,
and the molecular ion and the (M-CH3)+ ion of the internal standard. These ions
corresponded to m/z values of 724, 693, 474, and 459, respectively. The resolution of the
GC capillary column (12 m x 0.2 mm methyl silicone, SP2100) was not sufﬁcient to
resolve the 6B-hydroxycortisol MO-TMS derivative completely from sample constituents
that gave response at the same m/z values as the molecular ion and (M-OCH3)+ ion of
6B-hydroxycortisol MO-TMS. The results from the analysis of a derivatized guinea pig—
urine obtained with the I-IP5985 are shown in Figure 4.5a. The retention time of the 6B-
hydroxycortisol MO-TMS derivative is indicated with an arrow.

The guinea pig urine samples were re-analyzed later using a JEOL AX505 double
focussing mass spectrometer. A HP5990 gas chromatograph was directly interfaced with
this mass spectrometer. Better chromatograpic resolution was obtained using this system
and a new 30 m x 0.32 mm DB-l capillary column. Splitless injection was used to
introduce the samples. The temperature program was started at 50°C to condense the
solvent and the sample components. The oven was ramped up to 260°C at 35°C/min. and
then ramped up to 290°C at 4°C/min. The ion currents at m/z 724, 693, 474, and 459
were monitored using magnetic ﬁeld switching selected ion monitoring. The collector

slit (the slit just prior to the electron multiplier) was opened up to 200 um to maximize

 

E |

ECNI

Figure 4.5

117

 

 

 

 

 

 

 

 

 

1
_ l m/2693
l I l T T l ‘ I l l l l

: m/z459,474
- internal
- standard

‘T"'1"7T"‘l"‘f“'l"‘l'TT
12 13 14 15 16 17 18 19

 

m/z 314

 

 

 

rn/z 342

internal
standard

 

 

 

 

 

Time (minutes)

Selected ion current proﬁles obtained from a) GC-EI/MS analysis
of MO-TMS derivatized guinea pig urine sample and b) GC-
ECNI/MS analysis of oxidized guinea pig urine sample.

 

 

118

the detection of these ions. An accelerating voltage of 3 kV and post accelerating voltage
of 10 kV were used.

Due to sensitivity problems experienced with negative ion detection by the
HP5985, the analyses of the oxidized samples by GC-ECNI/MS were performed on the
JEOL HXl 10 mass spectrometer. At the time of these analyses, the HXllO was
interfaced with the HP5990 GC via a single stage jet separator. The samples entered the
separator via a "megabore" DB-l capillary column (15 m x 0.53 mm) with a carrier gas
ﬂow rate of 10-12 ml/min. Make-up gas was fed into the separator for a total ﬂow rate of
about 30 ml/min. An ion source temperature of 200°C was used for these analyses.
Methane was used as the modifying gas. The HXllO was not capable of performing
magnetic ﬁeld switching, so electric ﬁeld switching was used for the selected ion
monitoring. Magnetic ﬁeld switching SIM is preferred over electric ﬁeld (EF) switching
SIM because the latter discriminates against the ions of higher masses that are monitored
within an analysis. This is because the accelerating voltage is decreased to pass the
higher mass ions during EF switching SIM. The data shown in Figure 4.5b were
obtained during EF switching SIM analysis of an oxidized guinea pig urine.

DIP-ECNI/MS analyses of the oxidized guinea pig urines were performed on the
HP5985. (The negative ion sensitivty had been restored to an acceptable level by various
maintenance procedures.) The molecular anions of oxidized 6B-hydroxycortisol and the
oxidized internal standard, at m/z 314 and m/z 342, respectively, were monitored. The
ion source parameters were the same as those used for the preliminary study of oxidized
human urine. The direct probe was heated rapidly to 250°C. A good desorption proﬁle
was obtained by rapidly heating the probe to 250°C.

 

 

119

3. Results and Discussion

The results of the GC-EI/MS, GC-ECNI/MS, and the DIP-ECNI/MS analyses of
the control and guinea pig urines are shown in Table 4.1. The values in Table 4.1
represent the total daily excretion of 6B-hydroxycortisol (pg/day). The number of

analyses (n) for each sample was one unless it is indicated as otherwise.

Table 4.1 Comparison of GC-EI/MS of the MO-TMS derivatived urine with
GC—ECNI/MS, and DIP-ECNI/MS, of the oxidized urine, for the
determination of 6B-hydroxycortisol excretion by control and
induced guinea pigs.

 

 

Sample GC-EI GC-ECNI DIP-ECNI
control 1a 74 62 1 19

control 1b 61 48 95 i- 1 (n=2)
control 1c 55 50 101 i 4 (n=2)
mean 63 i 10 53 i 7 102 i 10
induced la 20 77 1 l 1

induced 1b 16 70 92

induced 1c 17 64 127

mean 18:2 701-6 110:18

The paired t-test was used to determine whether the three methods produced
signiﬁcantly different results for the determination of 6B-OHC in the control and the
induced samples.

The analysis of the control sample by GC-EI/MS and by GC-ECNI/MS did not
yield signiﬁcantly different results. The results found by DIP-ECNI/MS were
signiﬁcantly different from the results found by both GC-EI/MS and GC-ECNI/MS. This
indicates that the oxidation method can provide results that are comparable to the
conventional GC-MS method for the analysis of normal guinea pig urine, provided that

the GC inlet is used. Unfortunately, the analyses using the DIP gave erroneously high

 

120
results. This would indicate that other compounds within the oxidized urine matrix are
capable of producing a response at m/z 314 under ECNI conditions.

For the induced sample, the results from the GC-EI/MS analysis were
signiﬁcantly different from the results of the ECNI/MS analyses. The GC- and DIP—
ECNI/MS results were not signiﬁcantly different by the paired t-test. This indicates that,
in the induced state, the guinea pig may excrete compounds other than 613-
hydroxycortisol that are capable of being oxidized to 4-androstene-3,6,l1,17-one.

The regular t-test was applied to test for a signiﬁcant difference between the
control and induced samples. The difference between the control and induced samples
by the GC-ECNI/MS method was signiﬁcant at the 95% conﬁdence level. The control
and induced samples were also signiﬁcantly different by the GC-EI/MS method even at
the 99% conﬁdence level. By DIP-ECNI/MS analysis, the difference between the control
and induced samples was not signiﬁcant.

It is interesting that the results found 1;; the induced sample by the GC-EI/MS
method are signiﬁcantly lower than those for the control sample. This contradicts the
report that the amount of 6B-hydroxycortisol excreted by guinea pigs increases upon P-
450 enzyme induction with phenobarbital. Possible reasons for this disagreement could
be due to the methodology used by the researchers who reported these results. The
method used in their determinations was based on extraction and paper chromatography
and the Porter—Silber reaction for colorimetric determination. A more recent study (17)
has indicated that "6B-hydroxycortisol is not a useful index of hepatic microsomal
enzyme induction in the guinea pig because 6B-OHC itself undergoes extensive
metabolism in this species." The method for 6B-OHC determination used in this study
was RIA. The metabolites of 6B-OHC were not identiﬁed, but it was suggested that the
metabolites might contain a C21 carboxyl group, as they were acid extractable. It is very
likely that 21—carboxy 6l3-OHC could be converted to 4-androstene—3,6,1 1,17-one by the

PCC oxidation. This may explain the why the results found by the oxidation method for

1 2 l
the induced urine were higher than the results obtained by the MO-TMS method.

The relative standard deviations were 13-15% for the GC-EI/MS analyses, 8-13%
for the GC-ECNI/MS analyses, and 15% for the DIP-ECNI/MS analyses. With an
internal standard it should be possible to achieve better precision. The use of an internal
standard that more closely resembles 6B-hydroxycortisol may help to improve the
precision of the determination of 6B-OHC by these methods. Ideally, an isotopically-
labelled analog should be used for the internal standard.

C. Determination of 6B-Hydroxycortisol in Human Urine by GC-EI/MS, GC-ECNI/MS,
and DIP-ECNI/MS

The results of the study described above indicated that the guinea pig was not a
good model for studying 6B-hydroxylation of cortisol. The goal of this proposed
methodology is to provide a reliable and relatively simple means to determine 6B-OHC
in human urine. Another study was undertaken to determine the reliability of the
oxidation method for the determination of 6B-hydroxycortisol in human urine. The
excretion rate of 6B-OHC in normal human subjects is 200-400 ug/day. The samples
used in For this study twenty four-hour urine samples were collected before and after the
volunteer subject underwent niacin treatment to lower blood cholesterol levels. These
urine samples were analyzed for 6B-hydroxycortisol by the oxidation method with
ECNI/MS detection and the conventional MO-TMS method with GC-EI/MS analysis.

The results obtained by the different methods are compared.

1. Sample Preparations

The preparation of these samples was the same as that for the guinea pig urines,

with a few changes. The volume of the human urine used for these analyses was 2 ml.

This represented about 0.1 to 0.2% of the total amount of the 24-hour urine samples. The

122
compound 6B-hydroxyprednisolone was used as the internal standard. This compound is
non-endogenous and is identical to 6B-OHC except for the presence of a double bond in
the C1 position. Attempts to synthesize a deuterated analog of 6B-OHC for use as the
internal standard were not successful. (A description of these experiments and the results
are included below.) The ethyl acetate extraction following the C18 solid-phase
extraction of the urine was eliminated. The oxidation was conducted at 60°C instead of
room temperature. Experiments described in Chapter 2 indicated that the rate of
oxidation can be increased by heating the reaction. The reaction time used was 4.5 hours.
A comparison of the yield of 4-androstene—3,6,ll,l7-one from urinary 6B-
hydroxycortisol indicated that the yield was about the same after 2, 3, and 5 hours of
oxidation. Removal of the excess PCC for this study was accomplished with a 200 mg
silica SPE column. Previously, 500 mg columns had been used. After loading the
sample onto the column, the column was washed with 3 ml of CH2C12, and the oxidized
6B-OHC was eluted with 2 ml of ethyl acetate. Nearly all of the oxidized 6B-OHC is
eluted in this fraction by 2 ml of ethyl acetate. With larger volumes of ethyl acetate, the

PCC starts to elute from the column.

2. Deuteration Experiments

The goal of these experiments was to incorporate three deuterons by exchange of
the alpha hydrogens of the C3 ketone. In the ﬁrst experiment, excess solid sodium
methoxide was added to a solution of 6B-OHC in CH3OD. The reaction mixture was
stirred at room temperature in a capped vial for seven days. The mixture was then
diluted with CH2C12. The CH2C12 layer was isolated and washed with water to remove
the sodium methoxide. The CH2C12 was dried with anhydrous sodium sulfate to remove
traces of water. Analysis of the reaction mixture by GC-EI/MS revealed the presence of

a compound that could have been hexadeutero l7-keto 6B-OHC.

123

In the second experiment, 6B-OHC was dissolved in CH3OD and a piece of
sodium metal was added directly to the solution. The reduction of the Na(s) proceeded
spontaneously resulting in the in situ production of sodium methoxide and deuterium gas.
The methoxide ion is a strong base and is capable of pulling off acidic hydrogens,
including alpha-keto hydrogens. After 1 hour, an aliquot of the reaction mixture was
removed and diluted with CH2C12. The CH2C12 layer was isolated, washed with water,
and dried with anhydrous sodium sulfate. After 7 hours another aliquot of the reaction
was extracted, washed, and dried. No intact 6B—OHC was recovered in either of these
aliquots. Apparently, the compound is degraded in the presence of such a strong base.

Another method that was tried involved the use of basic alumina (18). Basic
alumina (8.2 g) having Brockmann I activity was incubated with D20 (0.25 ml) for 2 1/2
hours. About one quarter of the alumina was poured into a pasteur pipette plugged at the
bottom with glass wool. The column was wetted with acetonitrile and 0.5 ml of a
solution of 6B-OHC in CH3OD was loaded onto the column. The exchangeable
hydrogens were to have been replaced with deuterium by the time the compound eluted
from the column. Acetonitrile was used as the eluent. Several fractions of 3.5 ml were
collected from the column. The fractions were analyzed by HPLC to see if they
contained any 6B-OHC. The ﬁrst fraction contained a degradation product of 6B-OHC,
17-keto-6B-OHC. No 6B-hydroxycortisol was recovered from the column. The identity
of the degradation product was determined by GC-EI/MS of its MO-TMS derivative. No
deuterium was incorporated into this molecule. If a less polar eluent had been used, the
6B-OHC would have remained on the column for a longer time. This might have allowed
more time for the exchange to take place.

The labile C17 sidechain must be protected to prevent degradation during the
exchange reaction. This could be accomplished by the formation of the bis-

methylenedioxy derivative (19).

 

 

124

3. Instrumental Analysis

Two different mass spectrometers were used for the analysis of these urine
samples. The JEOL AX505 was used for the GC-EI/MS analysis of the MO-TMS
samples. The mass spectrometer conditions were the same as those described for the
analysis of the MO-TMS of the guinea pig urine samples. The AX505 was also used for
the DIP-ECNI/MS analysis of the oxidized urines. With this instrument, very high ion
source temperature is required to efﬁciently ionize oxidized 6l3-OHC. Methane was used
as the modifying gas at a pressure reading of 1.2 x 10-5 on the penning gauge. Selected
ion monitoring by magnetic ﬁeld switching was used to detect the molecular anions of
oxidized 6B-OHC and oxidized 6B-hydroxyprednisolone, at m/z 314 and m/z 312,
respectively. A good desorption proﬁle was obtained on this instrument by heating the
DIP from 30°C to 100°C at 32°C/min.

The Finnigan TSQ70 triplequadrupole mass spectrometer was used for the GC-
ECNI/MS analysis of the oxidized urine samples. The TSQ70 was operated as a
conventional scanning mass spectrometer for these analysis. The negative ion sensitivity
of this instrument is superior to that available with any of the other instruments.
Parameters for the ECNI analysis were roughly optimized. Rigorous optimization was
not necessary because the sensitivity was found to be adequate under the commonly used

conditions for ECNI on this instrument.

 

 

125

4. Results and Discussion

The results for the determination of 680HC in these urine samples by the MO-
TMS and oxidation methods are compared in Table 4.2. The number of replicate
analyses, n, was three, unless indicated otherwise. The values are reported i the standard

deviation of the triplicate analyses.

Table 4.2 Comparison of GC-EI/MS of MO-TMS derivatives with GC-
ECNI/MS and DIP-ECNI/MS of oxidized urine samples for the
determination of the total daily excretion of 6B-hydroxycortisol by
a 49-year old male.

Sam le GC-EI GC-ECNI DIP-ECNI
control 1a 340 i 10 290 i- 10 920 i 190 (n=4)
control 1b 380 i 30 345 i 20

niacin 1a 345 :l: 7 _ 280 i 20

niacin 1b 300 i 8

niacin 2a 377 (n=l)

niacin 2b 371 (n=1)

The paired t-test was applied to test for a signiﬁcant difference between the
oxidation GC-ECNI/MS method and the MO-TMS GC-EI/MS method for the
determination of 6B-hydroxycortisol. The difference between the two methods was not
signiﬁcant even at the 98% conﬁdence level. As found in the analysis of the guinea pig
urines, the values determined using the DIP were much higher than the values found
using the GC inlet. More rigorous separation of the 6B-hydroxycortisol from the other
components in the urine samples appears to be necessary in order to use the DIP for
sample introduction.

Treatment with niacin does not appear to alter the amount of 6B-hydroxycortisol
excreted per day. The amount of 6B-hydroxycortisol excreted per day as determined by
the oxidation method and the MO-TMS method fell within the expected normal range of
ZOO-400 ttg/day.

126

III. Conclusions

The results described in this chapter have demonstrated that oxidation can be used
to selectively enhance the ECNI-MS detection of 68-hydoxycortisol in human urine. The
results obtained by this oxidation method do not differ signiﬁcantly from the results
obtained by the GC-MS reference method based on the formation of the MO-T‘MS
derivative. The use of the DIP for the ECNI-MS analysis tended to give errroneosly high
results. The procedure used to isolate the 6B-hydroxycortisol from the urine samples was
not very selective for 6B-hydroxycortisol. Implementation of more selective isolation of
68-hydroxycortisol might remove matrix components that interfere with the analysis by
DIP-ECNI.

The use of oxidation and ECNI-MS for the determination of urinary 6B-
hydroxycortisol offers some advantages over the MO-TMS method. The good selectivity
of the methodology allows the use of simpler and faster GC separation. A "megabore"
capillary GC column can be used to introduce the oxidized urine samples to a mass
spectrometer. Also, only one product is produced by the one-step oxidation of 6B-
hydroxycortisol. The MO-TMS derivatization requires more rigorous and lengthy GC
separation. In addition to interference from sample components, syn and anti isomers of
6B—hydroxycortisol, produced in the formation of the methoximes, complicate the
chromatographic separation. The MO-TMS derivatization is a two-step reaction. The
use of a two-step reaction lengthens the sample preparation time and can lead to a
mixture of products. The product of the oxidation reaction is quite stable and requires no
special storage conditions. The MO—TMS derivative must be kept in a moisture free

environment to avoid hydrolysis of the TMS ethers.

 

 

 

10.

ll

12.

l3.
14.

15.
16.

17.

18.

REFERENCES

A. H. Conney, Pharmacol. Rev., 19, 317-366, 1967.

J. J. Vrabanac, Jr., Eﬁ’ects of Exposure to Polybrominated Biphenyls on Urinary
Steroid Metabolic Proﬁles in Man and Rat as Determined by Capillary GC and
GC/MS/DS, Ph. D. Dissertation, Department of Pharmacology and Toxicology,
Michigan State University, 1983.

P. Dolara, M. Lodovici, M. Salvadori, G. Santoni, and G. Cademi, Ann. Occup.
Hyg., 27, 183-188, 1983.

S. B. Pal, Metabolism, 27, 1003-1011, 1978.
E. M. Werk, Jr., J. MacGee, and L. J. Sholiton, Metabolism, 13,1425-1438, 1964.

E. Voccia, P. Saenger, R. E. Peterson, W. Rauh, K. Gottesdiener, L. S. Levine,
and M. I. New, J. C lin. Endocrinol. Metab., 48, 467-471, 1979.

D. J. Waxman, C. Attisano, F. P. Guengerich, and G. P. Lapenson, Arch.
Biochem. Biophys., 263, 424-436, 1988.

A. G. Frantz, F. H. Katz, and J. W. Jailer, J. Clin. Endocrinol. Metab., 21, 1290-
1303, 1961.

T. Ono, K. Tanida, H. Shibata, H. Shirnakawa, Chem. Pharm. Bull, 34, 2522-
2527, 1986.

B. K. Park, J. Steroid Biochem, 9, 963-966, 1978.

E. Gerber-Taras, B. K. Park, and E. E. Ohnhaus, J. Clin. Chem. Clin. Biochem,
19, 525-527, 1981.

M. G. Homing, S. S. Lau, A. Hung, W. G. Stillwell, and R. M. Hill, J. Steroid
Biochem, 5, 362 (abstract), 1974.

J. Chamberlain, Clin. Chim. Acta, 34, 269-271, 1971.

K. Thrasher, E. E. Werk, Jr., Y. Choi, L. J. Sholiton, W. Meyer, and C. Olinger,
Steroids, 14, 455-468, 1969.

R. J. Cotter, Anal. Chem, 52, 1589A-1606A, 1980.

R. Kuntzman, M. Jacobson, W. Levin, and A. H. Conney, Biochem. Pharmacol.,
17, 565-571, 1968.

B. K. Park, M. R. Challiner, and S. Newby, J. Steroid Biochem, 18, 453-457,
1983.

A. F. Hofmann, P. A. Szczepanik, and P. D. Klein, J. Lipid Res., 9, 707-713,
1968.

127

1 2 8
19. J. A. Edwards, M. C. Calzada, and A. Bowers, J. Med. Chem, 7, 528-530, 1964.

CHAPTER 5. SUMMARY

This research project involved development of new methodology for the mass
spectrometric determination of corticosteroids in biological samples. Because of the
selectivity and sensitivity associated with electron capture negative ionization (ECNI)
mass spectrometry, this technique was investigated for enhancing- the detectability of
corticosteroids. Corticosteroid determinations by ECNI have been very limited because
the derivatization of this class of steroids with electrophilic moieties has been
unsatisfactory due to the multiple hydroxy and ketone groups present in these types of
molecules.

A few types of steroids can be converted to highly electrophilic analogs by
oxidation. Examples include steroids having the 4-ene-3-one-6-oxy moiety, the 1,4- '
diene-3-one-l l-oxy moiety, or the 5-ene-3-ol moiety. Previous work in this laboratory
demonstrated the feasibility of using oxidation to enhance the electron capturing
properties of corticosteroids drugs by incorporating C-11 and C-17 ketone moieties into
these molecules.

The major goals of this research project were: 1) the optimization of oxidation
methodology for quantitative conversion of corticosteroids to their electrophilic oxidized
analogs; 2) to validate this oxidation methodology for a) the determination of
dexamethasone in plasma from patients undergoing the dexamethasone suppression test,
and b) the determination of 6B-hydroxycortisol in urine; and 3) to investigate the
feasibility of a faster, less selective method, such as direct probe, for introducing the
sample into the mass spectrometer.

Several Cr(VI) oxidants and various conditions for the oxidation of
dexamethasone were studied to establish optimal production of 11,17-keto
dexamethasone. A yield of 70% 11,17-keto dexamethasone could be achieved after 5.5

hours by PCC oxidation in CH2C12, with sodium acetate, at 60°C. Complete recovery of

129

1 3 0
the product from the reaction mixture could. be achieved using silica solid-phase
extraction with ethyl acetate or 5% acetone-CH2C12 as the eluent. Some preliminary
work indicated that the oxidation product of 6B-hydroxydexamethasone, the major
metabolite of dexamethasone, also gave a high electron capture response.

The oxidation methodology with ECNI-MS analysis for the determination of
dexamethasone in plasma was validated by comparison with other methods.
Concentrations of dexamethasone in plasma samples from patients undergoing the
dexamethasone suppression test (DST) were determined using oxidation and GC-
ECNI/MS with SIM. Aliquots of the same plasma samples were analyzed by GC-EI/MS
and RIA methods in other laboratories.

The results obtained by the oxidation method with ECNI-SIM detection did not
differ greatly from the results obtained by the reference GC-EI-SIM method. The EI-MS
method was based on the formation of the TMS-enol-TMS ether derivative of
dexamethasone. All of the dexamethasone extracted from a 5 ml plasma sample was
derivatized and concentrated to a volume of 5 111. All of the 5 111 was required for a
single analysis by EI-MS. An aliquot rcpresenting only 2% of a 1.5 ml plasma sample
was adequate for the ECNI-MS analyses. The relative standard deviations of the EI-MS
method and the oxidation method with ECNI-MS were 16% and 10%, respectively.

The concentrations of dexamethasone in the DST samples were also determined
by RIA in nine different laboratories. The results obtained by the oxidation method
generally fell within one standard deviation of the mean of the nine RIA determinations.
The relative standard deviations associated with the nine RIA determinations were
between 2% and 50%

The use of GC-ECNI/MS/MS was investigated for the detection of oxidized
dexamethasone in the plasma matrix. Selected reaction monitoring (SRM) was used to
detect the (M-HF)- daughter ion of the molecular anion of 11,17-keto dexamethasone.

This technique yielded similar results to those obtained through the use of selected ion

Eh,

131

monitoring. If the GC inlet is used, SRM offers no signiﬁcant advantage over SIM. This
is due to the high level of selectivity already achieved through the oxidation and the
electron capture negative ionization, as well as through the use of the chromatographic
inlet. However, with the additional selectivity of SRM, the direct probe sample inlet can
be used in place of the GC inlet. The plasma dexamethasone concentrations determined
using the direct inlet probe were comparable to those obtained using the GC inlet with
SIM and with SRM. Use of the direct probe can decrease analysis time to 3-5 minutes
per sample. Typical GC analysis time is 15-20 minutes per sample.

The only disadvantage associated with the oxidation method is the possibility that
other components in the sample matrix could be oxidized to the same product as
dexamethasone. The characteristic loss of HF from the molecular anion of oxidized
dexamethasone differentiates oxidized dexamethasone from endogenous sample
components because endogenous compounds do not contain ﬂuorine. Interference from
the metabolites of dexamethasone is not expected to be signiﬁcant because the major
metabolite of dexamethasone is 6B-hydroxydexamethasone. The oxidation product of
6B-hydroxy dexamethasone, 6,11,17-keto dexamethasone, does not have the same mass
spectrum as 11,17-keto dexamethasone; therefore, no interference is expected. The
oxidation product for the 20-dihydro metabolite of dexamethasone would be the same as
that for dexamethasone. However, since this is a minor metabolite, its interference is not
expected to severly limit this technique. The results that were obtained by the oxidation
method for the determination of dexamethasone in the plasma samples were similar to
the GC-EI/MS results. Therefore, components from the plasma matrix and
dexamethasone metabolites do not appear to interfere with the determination of
dexamethasone by the oxidation method with GC-ECNI/MS detection.

The feasibility of using oxidation methodology to determine urinary 6B-
hydroxycortisol was also demonstrated. The studies indicated that GC-ECNI/MS

provides good selectivity and sensitivity for the detection of the oxidation product of 6B-

 

 

132
hydroxycortisol within the urine matrix. Results from the oxidation-based assay were
compared to results obtained by a conventional GC-MS method based on formation of
the methoxime-trimethylsilyl ether derivative of 6B-hydroxycortisol. The determination
of 6B-hydroxycortisol in human urine was comparable by the two methods.

The results of this research project indicate that low levels of dexamethasone and
6B—hydroxycortisol can be determined accurately using oxidation to convert the steroids
to electrophilic analogs suitable for analysis by ECNI-MS.

This methodology could be applied to other corticosteroid drugs and steroids that
possess the 1,4-diene-3-one-11—oxy moiety. Also, steroids such as cholesterol and
dehydroepiandrosterone, that possess a 5-ene-3-ol moiety, can be oxidatively converted
to highly electrophilic 4-ene-3,6-dione analogs for analysis by ECNI-MS. A few of the
oxidized plasma and urine matrix components produced large responses under ECNI-MS
conditions. Identiﬁcation of these components might lead to the development of
sensitive assays for these compounds.

In addition to oxidation, there are many selective organic reactions that could be
utilized by the analytical chemist. For example, incorporation of the electrophilic 1,4-
diene-3-one moiety into 3-hydroxy bile acids has been accomplished in a one-step
synthesis using iodoxybenzene and benzeneselenic anhydride (1). This reaction could
possibly be used to selectively enhance the electron capture response of 3-hydroxy

steroids.

 

 

 

1.

REFERENCES

T. Iida, T. Shinohara, J. Goto, T. Nambara, and F. C. Chang, J. Lipid Res., 29,
1097—1 101, 1988.

133

 

 

 

"Ililllllllllllllllilllli