and Pharmacodynamics of the selective reuptake inhibitors, fluoxetinean d , during and the nursing period.

by

John Kim

M.Sc.(Pharm.), The University of British Columbia, Vancouver, Canada, 1995 B.Sc. (Biochem), Simon Fraser University, Burnaby, Canada, 1991

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

in

THE FACULTY OF GRADUATE STUDIES

(Faculty of Pharmaceutical Sciences) (Division of Pharmaceutics and Biopharmaceutics)

We accept this thesis as conforming to the required standard

THE UNIVERSITY OF BRITISH COLUMBIA November, 2000

©John Kim, 2000 In presenting this thesis in partial fulfilment of the requirements for an advanced

degree at the University of British Columbia, I agree that the Library shall make it

freely available for reference and study. I further agree that permission for extensive

copying of this thesis for scholarly purposes may be granted by the head of my

department or by his or her representatives. It is understood that copying or

publication of this thesis for financial gain shall not be allowed without my written

permission.

Department of r^hfartrMMJ-ioHsf (SdU^j

The University of British Columbia Vancouver, Canada

Date 2T) fJl>S p p

DE-6 (2/88) 11

Abstract

The prevalence of depressive disorders during pregnancy and the postpartum period and the need for continuous pharmacological intervention necessitate a better understanding of disposition via placental transfer and breast-feeding. However, there is relatively limited information available for pharmacokinetics of these drugs. In the present studies, the pharmacokinetics of and paroxetine were examined and compared in humans and sheep. In order to characterize drug disposition, several sensitive analytical methods for quantitative determination of fluoxetine and norfluoxetine isomers and paroxetine were developed using GC/MS and LC/MS/MS.

In humans and sheep, fluoxetine and norfluoxetine cross the placenta extensively. A relatively lower fetal-to-maternal ratio of paroxetine compared to fluoxetine was observed in humans. In adults, fluoxetine is extensively metabolized; however, minimal metabolic capacity was observed in human and ovine fetus. In the fetal lamb, no detectable concentrations of norfluoxetine isomers were observed in fetal plasma and amniotic fluid following fetal fluoxetine administration. Limited accumulation of fluoxetine in amniotic fluids was observed in fetal lamb unlike other basic amine drugs. In humans, serum neonatal fluoxetine and norfluoxetine concentrations were remained elevated following birth and slowly declined. These data were consistent with an study in sheep, which indicated lack of fetal N-demethylation in contrast to adult microsomes. In contrast, neonatal paroxetine concentration declined rapidly following birth. In fetal lambs, moderate transient changes in blood gas status were observed following fluoxetine administration.

However, in both fluoxetine- and paroxetine-exposed human gravida, neither significant changes in birth-outcome nor perinatal complications were observed. Ill

Fluoxetine, norfluoxetine and paroxetine are excreted in human breast milk with the milk-to- serum ratio higher for fluoxetine compared to paroxetine, which resulted in relatively higher exposure to fluoxetine in combination with relatively lower metabolic capacity in the neonate. Serum drug concentrations were also measured in nursing infants, and detectable levels of fluoxetine and norfluoxetine were observed in infants younger than 2 months.

Compared to adult ewes, significant changes in total body clearance, half-life and steady- state volume of distribution were observed in pregnant ewes. In addition, the stereoselective disposition of fluoxetine isomers was observed in both humans and sheep. In sheep, stereoselective pharmacokinetics following a single dose are mainly mediated by stereoselective . However, during chronic therapy, stereoselective along with fluoxetine-mediated inhibition of hepatic enzymes is responsible for stereoselectivity. Clearance- and exposure time-dependency of stereoselective disposition was observed in both species. Therefore, stereoselectivity during chronic dosing may be mediated by the inhibition of CYP2D6 by fluoxetine and norfluoxetine and stereoselective metabolism of fluoxetine by CYP2C9/19.

In conclusion, the present studies present the first detailed pharmacokinetics of fluoxetine and paroxetine during pregnancy and the nursing period, which suggest relatively lower exposure of paroxetine compared to fluoxetine. Furthermore, stereoselective disposition of fluoxetine was examined during both acute and chronic administration, and potential mechanisms were proposed. iv

Table of Contents

Abstract ii Table of Contents iv List of Tables xii List of Figures xv List of Abbreviations xix Acknowledgments xxiii

Chapter 1 Introduction 1

1.1 Depression during pregnancy and the postpartum period 1

1.2 Association and impact of maternal depression on infant/child development 2

1.3 Pharmacotherapy during pregnancy and the postpartum period 3

1.4 Pharmacological during pregnancy

and the nursing period 6

1.5 Teratogenicity of SSRIs in experimental animals 7

1.6 Teratogenicity and perinatal complications of SSRIs in humans 9

1.7 Clinical use of SSRIs 14

1.8 Pharmacology of fluoxetine, norfluoxetine and paroxetine 15

1.9 Pharmacokinetics of fluoxetine and norfluoxetine 18

1.10 Pharmacokinetics of paroxetine 25

1.11 Placental transfer of the therapeutic agents 27

1.12 Methods of studying placental transfer and fetal exposure 28

1.13 Pharmacokinetics of fluoxetine and paroxetine during pregnancy 31

1.14 Breast-feeding and pharmacotherapy 32

1.15 Lactation and of medication in breast milk 34

1.16 Excretion of fluoxetine, norfluoxetine and paroxetine in breast milk 35

1.17 Neonatal drug disposition 36 V

1.18 Rationales 39

1.19 Objectives and specific aims 41

Chapter 2 Analytical Method Development 43

2.1 Materials 45

2.2 Instrumentation 48

2.2.1 Gas chromatograph-mass selective detector (GC/MSD) 48

2.2.2 Gas chromatograph-mass spectrometer (GC/MS) 49

2.2.3 Liquid chromatograph-tandem mass spectrometer (LC/MS/MS) 49

2.2.4 Other equipment 50

2.3 Stereoselective GC/MS/EI method for fluoxetine and norfluoxetine isomers 50

2.3.1 Methods 50

2.3.1.1 Separation of fluoxetine and norfluoxetine using differential re-crystallization 50

2.3.1.2 Standard stock solution preparation 52

2.3.1.3 Sample extraction 53

2.3.1.4 Gas chromatography/mass spectrometry in electron impact mode 55

2.3.1.5 Calibration curve and regression model 55

2.3.1.6 Extraction recovery 5 6

2.3.1.7 Method validation 56

2.3.1.8 Analyte stability 5 7

2.3.2 Results and discussion 59

2.3.2.1 Mass spectrometric detection 59

2.3.2.2 Chiral separation of fluoxetine and norfluoxetine isomers 61

2.3.2.3 Extraction, recovery and stability 65 vi

2.3.2.4 Method validation 66

2.4 Synthesis and purification of deuterium-labeled fluoxetine hydrochloride 70

2.4.1 The synthesis of 2-benzoyl-l-(N-benzyl-N-methyl)-ethyl-ethylamine hydrochloride: B-acetylation 70

2.4.2 The synthesis of 1 -D5-phenyl-3-(N-methyl)aminopropan-1 -ol: hydrogenation 71

2.4.3 The synthesis of deuterium-labeled fluoxetine 72

2.5 Stereoselective LC/MS/MS method for fluoxetine and norfluoxetine isomers 73

2.5.1 Methods 74

2.5.1.1 Standard stock solution preparation 74

2.5.1.2 Sample extraction 74

2.5.1.3 Liquid chromatography/electrospray tandem mass spectrometry 75

2.5.1.4 Calibration curve and regression model 75

2.5.1.5 Extraction recovery 76

2.5.1.6 Method validation 77

2.5.1.7 Analyte stability 78

2.5.2 Results and Discussion 78

2.5.2.1 Mass spectrometric detection 78

2.5.2.2 Chiral separation of fluoxetine and norfluoxetine isomers 81

2.5.2.3 Extraction, recovery and stability 83

2.5.2.4 Method validation 83

2.6 GC/MS/EI method for paroxetine 88

2.6.1 Methods 88

2.6.1.1 Standard stock solution preparation 8 8 vu

2.6.1.2 Sample extraction 89

2.6.1.3 Gas chromatography/mass spectrometry 91

2.6.1.4 Calibration curve and regression model 92

2.6.1.5 Extraction recovery 92

2.6.1.6 Method validation 93

2.6.1.7 Analyte stability 93

2.6.2 Results and Discussion 94

2.6.2.1 Mass spectrometric detection and chromatograms 94

2.6.2.2 Extraction, recovery and stability 98

2.6.2.3 Method validation 100

2.7 GC/MS/NCI method for paroxetine 102

2.7.1 Methods 102

2.7.1.1 Standard stock solution preparation 102

2.7.1.2 Sample extraction 102

2.7.1.3 Gas chromatography/mass spectrometry 102

2.7.1.4 Calibration curve and regression model 103

2.7.1.5 Method validation 104

2.7.1.6 Recovery and analyte stability 105

2.7.2 Results and Discussion 105

2.7.2.1 Mass spectrometric detection and chromatograms 105

2.7.2.2 Method validation 108

Chapter 3 Stereoselective pharmacokinetics of fluoxetine and norfluoxetine in non-pregnant and pregnant ewes 112 Vlll

3.1 Materials and supplies 112

3.2 Study methods 113

3.2.1 Animals and surgical preparation 113

3.2.1.1 Adult (non-pregnant) sheep 113

3.2.1.2 Pregnant sheep 115

3.2.2 Pharmacokinetic experimental protocols 117

3.2.2.1 Intravenous administration of racemic fluoxetine for characterization of stereoselective pharmacokinetics in non-pregnant sheep 117

3.2.2.2 Paired intravenous administration of racemic fluoxetine for characterization of stereoselective maternal and fetal pharmacokinetics 118

3.2.3 Physiological monitoring 119

3.2.4 Plasma protein binding studies 120

3.2.4.1 Preparation of reagent solutions and spiked plasma samples 120

3.2.4.2 Ultrafiltration procedure 121

3.2.4.3 Equilibrium dialysis procedure 121

3.2.5 Preliminary in vitro metabolism studies of fluoxetine in ovine microsomes 123

3.2.5.1 Preparation of ovine hepatic microsomes 123

3.2.5.2 Preliminary maternal and fetal fluoxetine N-demethylation

in ovine hepatic microsomes 124

3.2.6 Analysis of biological samples 124

3.2.7 Data Analysis 125

3.3 Results 128

3.3.1 Pharmacokinetics of fluoxetine and norfluoxetine isomers in non-pregnant ewes following intravenous bolus fluoxetine administration 128 3.3.1.1 Plasma pharmacokinetics 128 ix

3.3.1.2 Urine pharmacokinetics 133

3.3.2 Pharmacokinetics of fluoxetine and norfluoxetine isomers in pregnant ewes following maternal intravenous fluoxetine administration (10 min infusion) 137

3.3.2.1 Plasma pharmacokinetics 137

3.3.2.2 Amniotic and fetal tracheal fluids 145

3.3.2.3 Maternal urine pharmacokinetics 148

3.3.2.4 Fetal physiological parameters following maternal fluoxetine administration 151

3.3.3 Pharmacokinetics of fluoxetine and norfluoxetine isomers in pregnant ewes following fetal intravenous fluoxetine administration (10 min infusion) 154

3.3.3.1 Plasma pharmacokinetics 154

3.3.3.2 Amniotic and fetal tracheal fluids 160

3.3.3.3 Maternal urine pharmacokinetics following fetal

fluoxetine administration 163

3.3.3.4 Fetal physiological parameters 166

3.3.4 Plasma protein binding of fluoxetine and norfluoxetine 169

3.3.4.1 Stability of fluoxetine and norfluoxetine in plasma 169

3.3.4.2 Free fraction determination of fluoxetine and norfluoxetine using ultrafiltration 170 3.3.4.3 Optimization of equilibrium time and assessment of volume shift and pH changes of equilibrium dialysis samples 170

3.3.4.4 Free fraction determination of fluoxetine and norfluoxetine

using equilibrium dialysis 171

3.3.5 Stereoselective in vitro drug metabolism of fluoxetine in hepatic microsomes 176

3.4 Discussion 177 3.4.1 Pharmacokinetics of fluoxetine and norfluoxetine in adult and fetal sheep following intravenous fluoxetine administration 177

3.4.2 Placental transfer of fluoxetine and norfluoxetine 180 X

3.4.3 Fluoxetine effects on fetal blood gas and acid-base status 182

3.4.4 Fluoxetine metabolism in adult and fetal sheep 183

3.4.5 Stereoselective pharmacokinetics of fluoxetine 186

Chapter 4 Human clinical and in vitro studies of fluoxetine and paroxetine

during pregnancy and the postpartum period 193

4.1 Study design 194

4.2 Patient recruitment 196

4.3 Inclusion and exclusion criteria 197

4.4 Informed patient consent and patient interview 198

4.5 Study protocol 200

4.5.1 Sample collection and APGAR scoring at birth and the PKU test 200

4.5.2 Postpartum sample collection 201

4.6 Analysis of perinatal complication 202

4.7 Materials and supplies 203

4.8 Analysis of clinical study samples 205

4.8.1 Determination of fluoxetine and norfluoxetine in human serum and milk 205

4.8.2 Determination of paroxetine in human serum and milk 206

4.9 Plasma protein binding study 206

4.9.1 Preparation of reagent solutions and spiked plasma samples 206

4.9.2 Equilibrium dialysis procedure 207

4.10 Preliminary in vitro studies of fluoxetine metabolism in human microsomes 208

4.11 Data analysis 209

4.12 Results 211 xi

4.12.1 Determination of fluoxetine/norfluoxetine and paroxetine in clinical samples 211

4.12.2 Demography and medical history of the study population 213

4.12.3 Fluoxetine and paroxetine disposition during the perinatal period 224

4.12.4 Fluoxetine and paroxetine disposition during the nursing period 231

4.12.5 Plasma protein binding of fluoxetine and norfluoxetine 240

4.12.6 Stereoselective in vitro drug metabolism of fluoxetine in hepatic microsomes 245

4.13 Discussion 248

4.13.1 Determination of fluoxetine/norfluoxetine and paroxetine in clinical samples 248

4.13.2 Demography and medical history of the study population 249

4.13.3 Perinatal complication for fluoxetine and paroxetine-exposed pregnancy 251

4.13.4 Fetal and neonatal exposure to fluoxetine and paroxetine

during the perinatal period 254

4.13.5 Infant exposure to fluoxetine and paroxetine during the nursing period 261

4.14 Stereoselective disposition of fluoxetine in human chronic therapy 266

Chapter 5 Summary and conclusions 279

References 287 Xll

List of Tables

Table 1.1 Comparison of uptake inhibition of commonly used 16 SSRI and .

Table 2.1 Intra-batch precision and accuracy of the fluoxetine and norfluoxetine 68 isomers using the GC/MS/EI method in ovine plasma (n=6).

Table 2.2 Inter-batch precision and accuracy of the fluoxetine and norfluoxetine 69 isomers using the GC/MS/EI method in ovine plasma (n=5).

Table 2.3 Intra-batch precision and accuracy of the fluoxetine and norfluoxetine 86 isomers using the LC/MS/MS method in human plasma (n=6).

Table 2.4 Inter-batch precision and accuracy of the fluoxetine and norfluoxetine 87 isomers using the LC/MS/MS method in human plasma (n=5).

Table 2.5 Intra-batch precision and accuracy of paroxetine using the GC/MS/EI 101 method in human plasma (n=6).

Table 2.6 Inter-batch precision and accuracy of paroxetine using the GC/MS/EI 101 method in ovine plasma (n=5).

Table 2.7 Intra-batch precision and accuracy of paroxetine using the 109 GC/MS/NCI method in human plasma (n=6).

Table 2.8 Inter-batch precision and accuracy of paroxetine using the 109 GC/MS/NCI method in human plasma (n=5).

Table 3.1 Plasma pharmacokinetic parameters following an iv bolus 131 administration of racemic fluoxetine (100 mg equivalent) in adult non-pregnant ewes (n=6).

Table 3.2 Urine pharmacokinetic parameters following an iv bolus 135 administration of racemic fluoxetine (100 mg equivalent) in non• pregnant ewes (n=6).

Table 3.3 Maternal plasma pharmacokinetic parameters following a maternal iv 143 administration of racemic fluoxetine (50 mg equivalent) in pregnant ewes (n=5).

Table 3.4 Fetal plasma pharmacokinetic parameters following a maternal iv 144 administration of racemic fluoxetine (50 mg equivalent) in pregnant ewes (n=5).

Table 3.5 Maternal urine pharmacokinetic parameters following a maternal iv 149 administration of racemic fluoxetine (50 mg equivalent) in pregnant ewes (n=5). Xlll

Table 3.6 Fetal plasma pharmacokinetics parameters following a fetal iv 157 administration of racemic fluoxetine (10 mg equivalent) (n=5).

Table 3.7 Maternal plasma pharmacokinetics parameters following a fetal iv 159 administration of racemic fluoxetine (50 mg equivalent) (n=5).

Table 3.8 Maternal urine pharmacokinetics parameters following a fetal iv 165 administration of racemic fluoxetine (10 mg equivalent) (n=5).

Table 3.9 In vitro plasma protein binding of fluoxetine and norfluoxetine in 174 ovine adult plasma (heparinized) determined separately for the fluoxetine and norfluoxetine isomers.

Table 3.10 In vitro plasma protein binding of fluoxetine and norfluoxetine in 174 ovine adult plasma (EDTA-treated) determined separately for the fluoxetine and norfluoxetine isomers.

Table 3.11 In vitro plasma protein binding of fluoxetine and norfluoxetine in 175 ovine fetal plasma (heparinized) determined separately for the fluoxetine and norfluoxetine isomers.

Table 3.12 plasma protein binding of fluoxetine and norfluoxetine in 175 pooled ovine plasma (heparinized) (n=3).

Table 3.13 Formation rate of norfluoxetine isomers in pooled ovine microsomes 176 from pregnant ewes and fetal sheep (n=2).

Table 4.1 Comparison and stability of fluoxetine and norfluoxetine in human 212 whole blood, plasma and serum (n=3).

Table 4.2 Comparison and stability of paroxetine in human whole blood, 212 plasma and serum (n=3).

Table 4.3 Demographic information for subjects on fluoxetine therapy during 214 the perinatal period.

Table 4.4 Demographic information for subjects on paroxetine therapy during 216 the perinatal period.

Table 4.5 Demographic information for subjects on fluoxetine therapy during 222 the nursing period.

Table 4.6 Demographic information for the subjects on paroxetine therapy 223 during the nursing period.

Table 4.7 Summary of maternal and fetal (umbilical cord) serum concentrations 226 of fluoxetine and norfluoxetine isomers at birth (ng/mL).

Table 4.8 Summary of fetal (umbilical cord) and neonatal serum concentrations 227 of fluoxetine and norfluoxetine isomers at birth and PKU testing . xiv

Table 4.9 Summary of maternal, fetal (umbilical cord) and neonatal serum 229 concentrations of paroxetine at birth and PKU testing (ng/mL)

Table 4.10 Summary of maternal and infant serum and breast milk 233/4 concentrations of fluoxetine and norfluoxetine isomers (ng/mL)

Table 4.11 Summary of maternal and infant serum and breast milk 238 1 concentrations of paroxetine (ng/mL)

Table 4.12 In vitro plasma protein binding of fluoxetine and norfluoxetine in 243 human plasma determined separately between fluoxetine and norfluoxetine isomers

Table 4.13 In vitro plasma protein binding of fluoxetine and norfluoxetine in 243 human plasma determined simultaneously

Table 4.14 In vivo plasma protein binding of fluoxetine and norfluoxetine in 244 pooled human plasma (n=3)

Table 4.15 Formation rate of norfluoxetine isomers in pooled microsomal 246 preparations from humans and rats (n=2)

Table 4.16 Formation rate of norfluoxetine isomers in human cDNA-expressed 247 single CYP isozyme microsomal preparation (n=2) XV

List of Figures

Figure 1.1 Chemical structures of selective serotonin reuptake inhibitors. 17

Figure 1.2 Proposed metabolic pathways of fluoxetine. 24

Figure 1.3 Proposed metabolic pathways of paroxetine. 27

Figure 2.1 Mass spectra and proposed fragment ions of the derivatives of 59 fluoxetine, norfluoxetine and the internal standard in the electron impact ionization mode.

Figure 2.2 Representative chromatograms of fluoxetine and norfluoxetine 64 isomers using the GC/MS/EI method.

Figure 2.3 Representative calibration standard curve for fluoxetine and 67 norfluoxetine isomers in human plasma using the GC/MS/EI method.

Figure 2.4 Positive ion electrospray daughter (product) ion mass spectra of 80 2 fluoxetine, norfluoxetine and [ H5]-fluoxetine (internal standard).

2 Figure 2.5 Representative LC/MS/MS ion chromatograms of [ H5]-fluoxetine 82 (internal standard), fluoxetine and norfluoxetine.

Figure 2.6 Representative calibration standard curves for fluoxetine and 84 norfluoxetine isomers using the LC/MS/MS method.

Figure 2.7 Cross-validation of plasma (S)-fluoxetine isomers in human plasma 85> using LC/MS/MS and GC/MS/EI method.

Figure 2.8 Mass spectra and proposed fragmentation ions of the HFBA- 96 derivatives of paroxetine and the internal standard in the electron impact ionization mode.

Figure 2.9 Representative chromatogram of paroxetine and the internal standard 97 in the GC/MS/E method.

Figure 2.10 Representative calibration standard curve of paroxetine using the 100 GC/MS/EI method.

Figure 2.11 Mass spectra and proposed fragmentation ions of the HFBA- 106 derivatives of paroxetine and the internal standard maprotiline in the negative chemical ionization mode.

Figure 2.12 Representative chromatogram of paroxetine and the internal standard 107 in the GC/MS/NCI method.

Figure 2.13 Representative calibration standard curve of paroxetine using the 108 xvi

GC/MS/NCI method.

Figure 2.14 Cross-validation of paroxetine in human serum using GC/MS/EI and 110 GC/MS/NCI method.

Figure 3.1 Plasma concentration versus time profile of fluoxetine and 132 norfluoxetine isomers in femoral arterial plasma following an iv bolus administration of racemic fluoxetine to non-pregnant ewes (100 mg equivalent) (n=6; mean ± SD).

Figure 3.2 Representative urinary accumulation of fluoxetine and norfluoxetine 136 isomers following an iv bolus administration of racemic fluoxetine to an adult non-pregnant ewe (100 mg equivalent).

Figure 3.3 Representative urinary accumulation of the glucuronide conjugates 136 of fluoxetine and norfluoxetine isomers following an iv bolus administration of racemic fluoxetine to an adult non-pregnant ewes (100 mg equivalent)

Figure 3.4 Plasma concentration versus time profile of fluoxetine and 141 norfluoxetine isomers in maternal femoral arterial plasma following a 10-min iv infusion of racemic fluoxetine to pregnant ewes (50 mg equivalent) (n=5; mean ± SD).

Figure 3.5 Plasma concentration versus time profile of fluoxetine and 142 norfluoxetine isomers in fetal femoral arterial plasma following a 10-min iv infusion of racemic fluoxetine to pregnant ewes (50 mg equivalent) (n=5; mean ± SD).

Figure 3.6 Amniotic fluid concentration versus time profile of fluoxetine and 146 norfluoxetine isomers following a 10-min iv infusion of racemic fluoxetine to pregnant ewes (50 mg equivalent) (n=5; mean ± SD)

Figure 3.7 Fetal tracheal fluid concentration versus time profile of fluoxetine 147 and norfluoxetine isomers following a 10-min iv infusion of racemic fluoxetine to pregnant ewes (50 mg equivalent) (n=5; mean ± SD)

Figure 3.8 Representative urinary accumulation of fluoxetine andmorfluoxetine 150 isomers following a 10 min iv infusion of racemic fluoxetine to a pregnant ewes (50 mg equivalent)

Figure 3.9 Representative urinary accumulation of the glucuronide conjugates 150 of fluoxetine and norfluoxetine isomers following a 10 min iv infusion of racemic fluoxetine to a pregnant ewes (50 mg equivalent)

Figure 3.10 Fetal arterial blood gas status (p02 and pC02) versus time profile 152 following a 10 min iv infusion of racemic fluoxetine to a pregnant ewes (50 mg equivalent) (n=5; mean ± SEM) XVII

Figure 3.11 Fetal arterial blood glucose and lactate concentration and pH versus 153 time profile following a 10 min iv infusion of racemic fluoxetine to a pregnant ewes (50 mg equivalent) (n=5; mean ± SEM)

Figure 3.12 Plasma concentration versus time profile of fluoxetine and 156 norfluoxetine isomers in fetal femoral arterial plasma following a 10-min iv infusion of racemic fluoxetine (10 mg equivalent) (n=5; mean ± SD).

Figure 3.13 Plasma concentration versus time profile of fluoxetine and 158 norfluoxetine isomers in maternal femoral arterial plasma following a 10-min iv infusion of racemic fluoxetine (50 mg equivalent) (n=5; mean ± SD; n=2 for the NFX isomers).

Figure 3.14 Amniotic fluid concentration versus time profile of fluoxetine and 161 norfluoxetine isomers following a 10-min fetal iv infusion of racemic fluoxetine (10 mg equivalent) (n=5; mean± SD)

Figure 3.15 Fetal tracheal fluid concentration versus time profile of fluoxetine 162 and norfluoxetine isomers following a 10-min fetal iv infusion of racemic fluoxetine (10 mg equivalent) (n=5; mean ± SD)

Figure 3.16 Representative maternal accumulation of fluoxetine and 164 norfluoxetine isomers following a 10 min fetal iv infusion of racemic fluoxetine (10 mg equivalent)

Figure 3.17 Representative maternal urinary accumulation of the glucuronide 164 conjugates of fluoxetine and norfluoxetine isomers following a 10 min fetal iv infusion of racemic fluoxetine (50 mg equivalent)

Figure 3.18 Fetal arterial blood gas status (p02, pC02 and 02 saturation) versus 167 time profile following a fetal 10 min iv infusion of racemic

fluoxetine to a pregnant ewes (10 mg equivalent) (n=5 for p02 and

pC02; n=3 for 02 saturation; mean ± SEM)

Figure 3.19 Fetal arterial blood glucose and lactate concentration and pH versus 168 time profile following a fetal 10 min iv infusion of racemic fluoxetine (10 mg equivalent) (n=5; mean + SEM)

Figure 3.20 Optimization of equilibrium dialysis time for fluoxetine and 171 norfluoxetine isomers.

Figure 3.21 Correlation of AUC S/R ratio of fluoxetine isomers and total body 191 clearance of racemic fluoxetine in non-pregnant and pregnant sheep (n=10). xvm

Figure 4.1 Correlation of maternal and fetal (umbilical cord) serum 230 concentrations of FX, NFX and PX at birth (n=8 for FX and NFX, and n=20 for PX).

Figure 4.2 Mean (±SD) fluoxetine and norfluoxetine concentrations in maternal 235 serum and breast milk at each dosing level [10 mg/d (n=5), 20 mg/d (n=17), 30 mg/d (n=3) and 50 mg/d (n=l)].

Figure 4.3 Correlation of maternal serum and breast milk concentrations of 235 fluoxetine and norfluoxetine in nursing mothers (n=27).

Figure 4.4 Infant-to-maternal serum concentration ratio of fluoxetine and 236 norfluoxetine versus time profile in infants exposed via breast• feeding (n=5 for fluoxetine and n=l 1 for norfluoxetine).

Figure 4.5 Mean (±SD) paroxetine concentrations in maternal serum and breast 239 milk at each dosing level [10 mg/d (n=10, 15 mg/d (n=2), 20 mg/d (n=12), 30 mg/d (n=8) and 50 mg/d (n=l)).

Figure 4.6 Correlation of maternal serum and breast milk concentrations of 239 paroxetine in nursing mothers (n=33).

Figure 4.7 Correlation of the oral clearance of fluoxetine (total and individual 269 isomers) (L/h) in humans (n=39).

Figure 4.8 Correlation of the S/R ratio of serum concentrations of fluoxetine 270 isomers and exposure times in the patients in acute and chronic fluoxetine therapy (n=40)

Figure 4.9 Correlation of the S/R ratio of steady-state serum concentrations of 271 fluoxetine isomers and total fluoxetine oral clearance (L/h) in pregnant and postpartum women (n=35)

Figure 4.10 Correlation of the S/R ratio of steady-state serum concentrations of 271 fluoxetine isomers and NFX/FX ratio in pregnant and postpartum women (n=35)

Figure 4.11 Correlation of apparent oral clearance of paroxetine (L/h) vs. 272 exposure time (month) in postpartum women (n=25). xix

List of Abbreviations p Micron

°C Degree Celsius

Lig Microgram urn Micrometer pM Micromolar

~ Approximately

A AG a 1-acid glycoprotein

ANOVA Analysis of Variance

AUC Area under the plasma concentration vs. time curve

AUC0.W Area under the plasma concentration vs. time curve from zero to infinity

AUMC0.„ Area under the first moment curve

BCCWH BC Children's and Women's Hospital

BRU Biobehavioral Research Unit

CLH Hepatic clearance

CLint Intrinsic clearance

CLR Renal Clearance of the total drug

CLJB Total body clearance based on total drug concentrations

Cmax Maximal plasma concentration

Cp Plasma concentrations of the total drug

Cu Plasma concentrations of the unbound drug

CYP Cytochrome (P450) enzyme d Day

D, 2H Deuterium

Da Dalton

DPHM XX

ED Eating disorder

EDTA Ethylenediaminetetraacetic acid

EI Electron Impact (Ionization)

EM Extensive metabolizer

F/M Fetal-to-maternal ratio

fub Unbound fraction

FX Fluoxetine (Prozac®) g Gram

GC Gas chromatography

GFR Glomerular filtration rate h Hour

HPLC High Performance Liquid Chromatography i.d. Internal diameter i.e. id est; that is

I/F Neonatal (at PKU test)-to-fetal (umbilical cord) ratio

I/M Infant-to-maternal ratio

IQ Intelligence quotient

IV Intravenous

KCI chloride

IC,,, A Michaelis-Menten parameter for enzymatic reactions; substrate concentration at which the reaction velocity is at half-maximal.

LC Liquid chromatograph log P Logarithm of partition coefficient (octanol-to-water)

LOQ Limit of quantitation of the assay

M Molar (moles/litre)

M/P Milk-to-plasma(serum) ratio

MDD Major depressive disorder mg milligram xxi

min Minute ml Milliliter mm Millimeter mM Millimolar mo months

MPvM Multiple reaction monitoring

MRT Mean residence time of the total drug

MRT Mean residence time

MS Mass spectrometry

MS/MS Tandem mass spectrometry n Number of subjects or animals

NADPH Reduced B-- dinucleotide tetrasodium salt

NCI Negative (Ion) Chemical Ionization

NFX Norfluoxetine ng Nanogram

NMR Nuclear magnetic resonance o.d. Outer diameter

OCD Obsessive compulsive disorder

PBS Phosphate-buffered saline

pC02 Partial gas pressure of carbon dioxide

PD pH Negative logarithm of hydrogen ion concentration pKa Negative logarithm of acid association constant

PKU Phenylketonurea

PM Poor metabolizer pmol picomole

p02 Partial gas pressure of oxygen XXII

PX Paroxetine (Paxil®)

QC Quality control

QH Hepatic blood flow r2 Coefficient of determination

RE Relative error

RFX (R)-fluoxetine

RNFX (R)-norfluoxetine

RPP Reproductive Psychiatry Program

RSD Relative standard deviation

S/R (S)-to-(R) ratio

SD Standard deviation

SEM Standard error of mean

SFX (S)-fluoxetine

SIM Single ion monitoring

SIR Single ion recording

SNFX (S)-norfluoxetine

SSRI Selective serotonin

Apparent terminal half-life

TEA Triethylamine

TFAP (S)-trifluoroacetylprolyl chloride tmax Time of occurrence of maximal plasma concentration TRIS 2-amino-2-hydroxymethyl-1,3 -propanediol

UDPGT UDP-glucuronosyltransferase

UTI Urinary tract infection

Vd Volume of distribution

Vdss Apparent steady-state volume of distribution of the total drug x Dosing interval XXlll

Acknowledgments

I would like to acknowledge my research advisors, Drs. K. Wayne Riggs and Dan W. Rurak, for their encouragement, support, friendship, and patience throughout my graduate training. Also, I would like to thank members of my graduate research committee (Drs. Donald Lyster, Frank S. Abbott, James E. Axelson and Joanne Douglas) and examining committee (Drs. David V. Godin and Keith McErlane) for their valuable time and suggestions. Further I would like to thank Dr. Shaila Misri, Dr. Deirdre Ryan, Dr. Diana Carter, Ms. Anne Kuan, Ms. Christen Wilson and Ms. Xanthoula Korstaras from the Reproductive Psychiatry Program at BC Children's and Women's Hospital and St. Paul's Hospital and Ms. Colleen Fitzgerald, RN at BC Children's and Women's Hospital. I also thank Drs. Nancy Kent and many obstetricians and nurses at the Centre for Perinatal Diagnostics and Treatment, BC Children's and Women's Hospital. In addition, I would like to thank staff members and phlebotomists at Outpatient Laboratories, BC Children's and Women's Hospital, St. Paul's Hospital and other regional hospitals in the Lower Mainland.

A special thanks to Drs. Sanjeev Kumar and Havey Wong for his deep insight into science from our many discussions and friendship to survive all those days in the graduate school. The assistance of Ms. Caroline Hall, Ms. Nancy Gruber and Mr. Eddie Kwan with sheep studies and Mr. Roland Burton for LC/MS/MS is greatly appreciated. Furthermore, I would like to thank present and past fellow graduate/undergraduate students and postdoctoral fellows in the laboratory for their help, Dr. Reza Anari, Ms. Yoonhee Kim, Mr. Caly Chien, and Ms. Janna Morrison.

Furthermore, I thank Andrew Jayaraj, Wendy Jones, David Pendergast at Biogen for their support and for helping me to manage both academic and industry responsibility. I also thank members of Bioanalytical Chemistry and Metabolism Department of Biogen, Jianbo Zhang and Dale Mongeon. Comments, help and friendship from Dr. Abhijit Chakraborty at Johnson & Johnson were greatly appreciated. I would like to acknowledge the financial support received from BC Medical Service Foundation and the University of British Columbia during the course of my graduate studies. These studies were funded by the Medical Research Council of Canada and BC Medical Service Foundation.

Finally, I sincerely thank all the mothers and infants participated in the clinical study. Many of these patients encouraged me to continue the study to provide better information on fetal and infant drug exposure. I wish the best for these mothers and infants and hope that the information here will be used to provide better treatment for these mothers afflicted with depression. CHAPTER 1

INTRODUCTION

1.1 Depression during pregnancy and the postpartum period.

Epidemiological research has shown that depressive disorders of a non-psychotic

nature are prevalent, especially among women of childbearing age (Weissman 1987).

The incidences of depressive disorders during pregnancy and the postpartum period

are as high as 10 and 15%, respectively (O'Hara et al, 1984, 1988; Hopkins et al,

1984; Brockington et al, 1981; Appleby et al, 1997). In addition, Cox et al, (1993)

reported a threefold increase in the incidence of depression in the first month after

delivery. The symptoms of pre- and postpartum depression include tearfulness,

despondency, emotional lability, guilt, , sleep disturbance and feelings of

inadequacy in coping with pregnancy and the infant (Kumar and Robinson 1978).

Hypochondriasis, irritability, impaired concentration, poor memory and undue fatigue

are also common (Robinson and Stewart 1986). Postpartum depression is distinct

from postpartum blues, which occur in about 50-70% of puerperal women (O'Hara et

al, 1984, Hopkins et al, 1984, Pitt 1973). Unlike postpartum blues, postpartum

depression requires psychotherapeutic interventions (both psychosocial and

pharmacological treatments). The onset of depressive symptoms requiring

antidepressant therapy usually occurs within 1 to 12 weeks postpartum (Mortola

1989), and coincides with the period of breast-feeding (O'Hara et al, 1988). In the

Province of British Columbia, the Reproductive Psychiatry Program at BC Women's

and St. Paul's Hospitals, which specializes in psychiatric care during and following

Page 1 pregnancy, has provided patient care to over 3,500 pregnant and postpartum depressed women and their families annually (Misri et al, 1996).

Association and impact of maternal depression on infant/child development.

Untreated depression during pregnancy and the postpartum period can adversely affect a woman's health. Misri and Sivertz (1991) compared 20 postpartum women with depression who chose (TCA) treatment and achieved symptom remission, with a group of 5 postpartum depressed women who refused

TCAs. All 5 mothers in the latter group had an exacerbation of depressive symptoms, with 2 of the women needing hospitalization and the other 3 requiring family assistance secondary to . In a study involving 23 pregnant women with a history of at least one prior episode of postpartum depression, Wisner and

Wheeler (1994) also observed that women without active treatment had a recurrence rate of 62.5%, while those taking medication had a recurrence rate of 6.7%.

Untreated maternal depression also has considerable adverse impact on infant and child psychological development. Deleterious outcomes include insecure attachments, mild behavioural difficulties, delays in cognitive development, and poorer quality of interaction between the mother and child (Cogill et al, 1986;

Lyons-Ruth et al, 1986; Murray 1992; Stein et al, 1991). Evidence has also established that the negative pattern of mother-infant interaction established during the early postnatal period does not necessarily improve along with the mother's recovery from the depressive episode, even when the depressive episode is relatively short (Murray 1992; Sinclair 1994; Stein et al, 1991). in

Page 2 infants is also observed in depressed mother-infant pairs (Field et al, 1985). These effects are indicative of the importance of the quality of the early mother-infant relationship in the continuity of normal infant development. Thus, early psychiatric intervention in these periods is important to ensure normal child development.

In addition to the immediate impact on neonatal care and other complications, maternal affective disorders have been identified as an important variable influencing development throughout childhood (Cogill et al, 1986). The majority of these studies have focused on the impact of maternal affective disorders on mental and behavioral development of school-age children. Five-year-old children who had been exposed to maternal depression during pregnancy have found to exhibit social and behavioural difficulties in school (Alpern, Lyons-Ruth 1993; Sinclair 1994). These studies have consistently revealed adverse effects in the offspring of depressed mothers such as psychiatric disturbances (McKnew et al, 1979), cognitive and attention deficits (Orvaschel et al, 1988; Worland et al, 1984), conduct disorders

(Beardslee et al, 1987) and maladaptive patterns of aggression (Zhan-Waxier et al,

1984).

Thus, management of affective disorders during these periods will not only reduce maternal psychiatric morbidity, but also decrease potential adverse effects on infant neurological/psychological development.

Pharmacotherapy during pregnancy and the postpartum period.

No statistical data are available for the use of antidepressants during pregnancy and the postpartum period. In general the use of medications during these periods is not

Page 3 recommended; however, many surveys have indicated that it does occur in a majority of as well as in the postpartum period. Rurak et al, (1991) reviewed the epidemiological data on medication use during pregnancy prior to 1990 and concluded that the incidence of medication use by pregnant women ranged from 35-

100%. Averages of 2-4 drugs were taken by each woman during this period (Rurak et al, 1991). More recent surveys as well as reviews by other investigators also indicate similar rates (Bonati et al, 1990; Collaborative group on drug use in pregnancy 1992; de Jong-van der Berg et al, 1993; Bologa etal, 1994; Iri et al, 1997). Similar trends have also been observed in postpartum women (Passmore et al, 1984; Matheson

1989; Uppal 1993). These epidemiological data from different regions of the world indicate that a significant number of human fetuses and infants are exposed to medications at some points during gestation and the nursing period. Most commonly used agents include antibiotics, antiseptics/antibiotics for vaginal and urinary tract infections, analgesics, cough/cold remedies, laxatives, medications for /vomiting and anti-allergic medications. Medications for pregnancy-related complications such as pre-eclampsia and preterm labour (Kyle and Redman 1992;

Hauth et al, 1993) are also used commonly. In addition, medications for pre-existing medical conditions such as epilepsy, cardiovascular and affective disorders (Yerby et al, 1992; Mitani et al, 1987; Kulin et al, 1998; Goldstein et al, 1997; Pastuszak et al, 1993; Chambers et al, 1996) are usually continued through pregnancy unless well-known teratogenicity exists. The use of illicit drugs during pregnancy and the postpartum period also exposes fetuses and infants to these agents, and may have significant effects on fetal/neonatal neurological, cardiovascular and metabolic

Page 4 functions (Rurak, 1992; Chiriboga 1993). In spite of efforts to minimize the use of these agents during and following pregnancy, human fetuses and infants continue to be exposed to a wide range of over-the-counter, and prescription medications.

In 1994, the American Academy of Pediatrics (AAP) undertook an initiative to provide pharmacotherapy recommendations for breast-feeding. In its policy statement, the following guideline is provided: "The following question and options should be considered when prescribing drug therapy to lactating women. (1) Is the drug therapy really necessary? Consultation between the pediatrician and the mother's physician can be most useful. (2) Use the safest drug, for example, acetaminophen rather than for analgesia. (3) If there is a possibility that a drug may present a risk to the infant, consideration should be given to measurement of blood concentrations in the nursing infant. (4) Drug exposure to the nursing infant may be minimized by having the mother take the medication just after she has breast-fed the infant and/or just before the infant is due to have a lengthy sleep period" (Committee on Drugs 1994). The AAP (Committee on Drugs 1994) has categorized fluoxetine and other antidepressants as "drugs whose effect on nursing infants is unknown but may be of concern" in its policy statement on "The transfer of Drugs and Other

Chemicals into Human Milk." No policy statement or guideline on medication use during pregnancy has been stated by this or any other organization, to the best of my knowledge. However, a portion of this specific guideline provided by the AAP could be equally applied to drug therapy during pregnancy. The first three questions (with the third question modified to refer to the fetus) from the AAP guidelines would be directly applicable to the pregnant patient as well.

Page 5 Pharmacological management of depression during pregnancy and the nursing period.

Treatment of depression during pregnancy and lactation poses a difficult therapeutic problem, due to the need to consider the potential fetal and neonatal behavioral and developmental effects of medications taken by the mother, in addition to the primary concern for the health and safety of the mother. It is further complicated by the definite need for uninterrupted antidepressant treatment for the remission of the depressive episode. As described in the Sections 1.1 and 1.2, the substantial risks of untreated maternal depression on the wellbeing of the mother and infant should be considered.

Traditional antidepressant therapy with tricylics (TCA) and monoamine oxidase inhibitors (MAOI) is discouraged due to a relatively high incidence of adverse anticholinergic (e.g. blurred vision, dry mouth, urinary retention) and hypotensive effects, in the mother and possibly the fetus and infant (Mortola 1989; Kerns 1986).

However, a new class of antidepressant drugs, the selective serotonin reuptake inhibitors (SSRIs) including fluoxetine (Prozac®), paroxetine (Paxil®) and

(Zoloft®), with fewer adverse reactions, appears to provide an alternative treatment option. While specific data are not available on the frequency of SSRI use during pregnancy and the postpartum period, the use of these drugs has increased over the past few years. Several prospective and retrospective studies on the exposure to FX and other SSRIs during pregnancy have been reported (Kulin et al, 1998; Goldstein et al, 1997; Nulman et al, 1997; Chambers et al, 1996). Discussions with many psychiatrists from different regions of North America (British Columbia, Alberta,

Page 6 Ontario, New York, Washington, California, Georgia and Massachusetts) also indicate that the use of SSRI has increased significantly over the last few years (1993-

1997) (APA annual meeting 1997, personal communication). Fluoxetine, paroxetine and sertraline were the most commonly prescribed SSRIs among these psychiatric practices. The choice of specific SSRI antidepressants by a psychiatrist is largely based on observed adverse effects in the general patient population, as the pharmacological effects of these agents are generally similar (APA annual meeting

1997, personal communication). The continuation of current SSRI antidepressant therapy (i.e. medication taken prior to pregnancy or delivery) has been suggested

(Stowe 1997, personal communication; Stowe et al, 1997) and is generally accepted by many psychiatric practitioners.

Teratogenicity of SSRIs in experimental animals

Several reports in humans (Goldstein 1995) and experimental animals (Cabrera-vera and Battaglia 1998; Cabrera-vera et al, 1997; Montero et al, 1990; Romero et al,

1994) suggest that SSRIs like FX do not appear to cause morphological teratogenicity at therapeutic concentrations. Exposure of mouse embryos in whole embryo culture to sertraline and FX at 10 pM did not cause general embrotoxicity, but did result in craniofacial malformations (Shuey et al, 1992). Shuey et al, (1992) suggest that the inhibition of serotonin uptake into craniofacial epithelia may interfere with regulation of epithelial-mesenchymal interactions which are important for normal craniofacial morphogenesis. However, these concentrations (10 pM = 3.1 pg/mL for FX) are approximately 10 to 100 fold higher than those observed in human

Page 7 studies. Furthermore, these serotonin-related craniofacial malformations were not reported in SSPJ-exposed pregnancies.

In experimental animals such as rats and mice, prenatal exposure to psychotropic drugs is associated with abnormalities of behavioural and neurochemical development in the absence of morphological abnormalities (Montero et al, 1990;

Romero et al, 1994; Cabrera-vera et al, 1997; Cabrera-vera and Battaglia 1998).

Changes in neurotransmitter receptors appear to be more marked and prolonged in offspring exposed to FX during pregnancy compared to controls exposed to FX at later time points following birth. For example, a 30% decrease in the density of 3H- binding sites was observed in the cortex of rat offspring exposed in utero to FX (maternal 2.5 mg/kg p.o.) compared to controls (Montero et al, 1990), but with no apparent morphological abnormality. Constant fetal exposure to FX in the rat (d

6-21 of gestation) led to a reduced response to serotonin-stimulated phosphate (PI) hydrolysis in 25 and 90 day-old rats, while chronic FX exposure to adults rats did not (Romero et al, 1994). Prenatal exposure to FX in the rat also resulted in an increased density of serotonin transporters (Cabrera-vera and Battaglia

1998) and serotonergic (Cabrera-vera et al, 1997) in the frontal cortex in prepubescent male rats.

Paroxetine does not appear to cause teratogenicity in humans (Kulin et al, 1998) and experimental animals (Baldwin et al, 1994; Coleman et al, 1999). In mice, prenatal exposure to a clinically relevant dose of PX (30 mg/kg/d for 2 weeks to give plasma

PX concentration similar to the upper therapeutic range in human) was associated

Page 8 with an increase in in infant mice (i.e. more separation vocalization) on the third postpartum day (Coleman et al, 1999).

These data suggest that there is a potential for long-term functional teratogenicity (i.e neurological and behavioural) of FX and other SSRIs in the absence of morphological abnormalities. The implications of these animal studies to the human infant and child remain to be determined; however, they certainly illustrate the need to examine behavioural and psychosocial consequences of these drugs on infants exposed during the pre- and post-natal periods.

In addition to these morphological and behavioural abnormalities, maternal intake of

FX may cause perinatal complications. Increased frequency of bruising at birth was observed following maternal intake of FX in rats (Stanford and Patton 1992). This finding is consistent with adverse vascular effects of FX in adult humans, which include bruises, hemoptysis, melena, hematemesis and hematuria (Pai and Kelly

1996; Yaryura-Tobias etal, 1991).

Teratogenicity and perinatal complications of SSRIs in humans.

There are a few case reports suggesting an association between exposure to FX and other SSRIs during pregnancy and morphological malformations (e.g., lipomeningocele, macrostomia, atrial septal defect, urethral disorders) (Goldstein

1990; 1991; Goldstein etal, 1992; Vendittelli 1995). However, several prospective studies examining birth outcome from SSRI-exposed pregnancies have reported that there is no significant increase in morphological teratogenic risks associated with maternal therapy compared to those from the control group. (Eli Lilly & Co. database

Page 9 (Goldstein and Marvel 1993); Motherisk program (Pastuszak et al, 1993); California

Teratogen Information Service (Chambers et al, 1996); Eli Lilly & Co. database

(Goldstein 1995); Motherisk program (Nulman et al, 1997); Motherisk multi- centered study (Kulin et al, 1998); meta-analytical review (Addis and Koren, 2000).

An increased frequency of miscarriage has been reported in FX-exposed pregnancies

(Pastuszak et al, 1993). However, this finding should be taken with caution, as stated by the authors, since the information was obtained from the patient and the patient may report miscarriage even though she may have chosen to terminate her pregnancy. Later studies (Goldstein 1995; Chambers et al, 1996) found that there was no statistically significant difference in the rate of spontaneous pregnancy loss or congenital malformation between the FX-exposed group and the control group. The incidence of minor structural abnormalities was reported to be higher in the first- trimester FX-exposed cohort (Chambers et al, 1996); however, the definition or criteria of minor abnormalities was not given. Based on clinical information currently available, FX appears to have minimal potential to cause major fetal morphological anomalies during first-trimester exposure.

In contrast, there are several reports suggesting an association between third-trimester

FX exposure and prenatal complications. There is a case report on FX toxicity in a newborn of 38 week's gestation, whose mother was receiving a daily dose of 20 mg during most of her pregnancy (Spencer 1993). Marked acrocyanosis, tachypnea with a respiratory rate of 70-76, temperature instability, continuous crying, irritability, and an elevated rate of 140 bpm were observed. A FX level of 26 ng/ml

(below the adult therapeutic range of 40-250 ng/ml (Krogh 1995)), and norfluoxetine

Page 10 concentration of 56 ng/ml (therapeutic range 30-325 ng/ml (Krogh 1995) were determined in a cord blood sample collected at delivery. In another case, jitteriness and hypertone were reported following birth in an infant whose mother was taking 60 mg of FX daily; petechiae and cephalo-hematoma were also observed. In this instance, infant serum concentrations of FX and NFX were 129 and 227 ng/mL, respectively, on the second day of life (Mhanna et al, 1997)

Goldstein (1995) prospectively evaluated 112 pregnancies (115 infants) in which the mother received FX during the third trimester until delivery. Postnatal complications unrelated to a malformation (e.g., irritability, hyper-bilirubinemia, sleepiness) were reported in 13% of infants. Goldstein (1995) concluded that a comparison with a general population database did not support any apparent pattern of findings, dose- relationships or increased frequencies of complications in the exposed newborns than expected in a non-exposed population. Unlike the findings of Goldstein (1995),

Chambers et al, (1996) reported a lower mean birth weight and shorter mean birth length in full-term infants exposed to FX in late gestation. In addition, the frequency of poor neonatal adaptation defined as jitteriness, tachypnea, hypoglycemia, hypothermia, poor tone, respiratory distress, weak or absence of cry, or de-saturation on feeding and special nursery admission were higher in the third-trimester exposed group (Chambers et al, 1996). Questions have been raised on the methodology and the definition of neonatal adaptation used in the study by other groups [Massachusetts

General Hospital group (Cohen and Rosenbaum 1996), Motherisk Program (Nulman et al, 1997) and the Eli Lilly Group (Goldstein et al, 1996)]. However, these findings along with other observations of apparent neonatal FX toxicity (Spencer

Page 11 1993; Mhanna et al, 1997) have raised concern about the safety of FX use during this period.

Compared to FX, information on PX and other SSRIs is more limited. Data on PX exposure during pregnancy are limited to a case report (Dahl et al, 1997). An increased respiratory rate, jitteriness, increased muscle tone and tremor were observed in the neonate whose mother was taking 30 mg daily during the last trimester. The serum PX concentration in the neonate was 68 nmol/L (22 ng/ml) at the age of one day, increasing slightly to 75 nmol/L (25 ng/ml) at two days and then decreasing to

23 nmol/L (7.6 ng/mL) on the third day. These adverse effects are similar to those in

FX-exposed pregnancies. In a prospective study on pregnancy outcome of maternal

SSRI use (paroxetine, sertraline and ), exposure was not associated with increased risk of major malformation or birth outcomes such as miscarriage, stillbirth or prematurity (Kulin et al, 1997). In addition, Kulin et al, (1997) stated that there was no difference in birth outcome between first-trimester and whole-pregnancy

(thus, third-trimester)-exposed groups. Recently, no significant increase in withdrawal symptoms in newborns following third trimester PX exposure in 52 women was reported (Ho et al, 2000).

In addition to morphological teratogenicity, there remains the possibility of more subtle effects of the drug on fetal neurological development. The developing central nervous (CNS) appears to be particularly susceptible to toxic drug effects, partly because of its extended period of development. While serious CNS morphological anomalies are usually the result of drug use in early pregnancy during the period of organogenesis (Dobbing 1976), the use of drugs during the second and third

Page 12 trimesters can result in subtle alterations in CNS function (Hutchings 1978).

Neurotransmitters are important in controlling cell proliferation in the nervous system. There is, therefore, concern that compounds which affect neurotransmitter activity in the CNS (e.g., tricyclic antidepressants and SSRIs) may interfere with cell differentiation and proliferation, resulting in a permanent change in later functional capacity but without obvious structural malformations (Beeley 1986;

Chasnoff 1988). These subtle abnormalities are termed 'behavioural teratogenic effects', which involve deficits in CNS activities such as motor function, learning and memory.

Studies in experimental animals have shown that prenatal FX exposure could potentially cause long-term neurochemical alteration (Montero et al, 1990; Cabrera- ver et al, 1997; 1998). The doses (weight-adjusted) used, however, were much higher compared to normal human doses. The Motherisk program at the Hospital for

Sick Children (Nulman et al, 1997) has examined the effects of FX and tricyclic antidepressants on infant/children's global IQ and language development. The

Bayley Scale of Infant Development or the McCarthy Scales of Children's Abilities and the Reynell Developmental Language Scales were used for the measurement of these parameters. These authors concluded that in utero exposure to FX does not affect global IQ, language or behavioural development in infants and preschool children.

Page 13 Clinical use of SSRIs.

Fluoxetine has become one of the most frequently prescribed antidepressant drugs since its approval for clinical use in late 1987. It is currently marketed under the trade name of Prozac® by the Eli Lilly and Co. A recent estimate suggests there are more than 11 million people taking Prozac® worldwide and more than 200,000 in

Canada (Nichols, 1994). Fluoxetine is effective in the treatment of major depressive disorders (MDD) and obsessive-compulsive disorders (OCD). Most cjinical studies have shown that the antidepressant effect of recommended doses ranging from 20-80 mg daily in patients with moderate to severe depression is greater than that of and comparable to that of the (TCAs), maprotiline or (Bremner

1984). Similarly, PX is a selective serotonin reuptake inhibitor marketed under the trade name of Paxil® by SmithKline Beecham. The recommended dose of PX ranges from 10-30 mg daily for the indications of MDD, OCD and panic disorders.

The profiles among SSRIs (FX and PX) and TCAs are very different.

Due to less frequent anticholinergic (e.g. dry mouth, , constipation) and cardiovascular effects (e.g. and conduction disturbances) which are usually associated with TCAs, FX and PX may be preferred in patients in whom such effects are not tolerated or are of potential concern. However, antidepressants other than the SSRIs may be an alternative in the patients in whom certain adverse Gl effects (e.g. nausea, anorexia) or nervous system effects (e.g. anxiety, nervousness, , weight loss) are of concern. The most frequent adverse effect associated with FX and PX therapy is nausea which occurs in about

20% of patients (AHFS 1994). Nausea generally is mild, occurs early in therapy, and

Page 14 usually subsides after a few weeks of continued therapy. While some clinicians reserve FX for patients who do not respond to TCAs or monoamine oxidase inhibitors

(MAOIs), other clinicians consider FX and other SSRIs to be first-line agents for the management of major depression because of their favorable adverse effect profile

(Leonardo 993). . -

Pharmacology of fluoxetine, norfluoxetine and paroxetine.

Fluoxetine and paroxetine are potent and selective inhibitors of serotonin (5-HT) reuptake (van Harten 1993; Wong et al, 191'4; Wong et al, 1975), and, as mentioned previously, are currently used clinically in the treatment of depressive disorders

(AHFS 1994). Fluoxetine facilitates serotoninergic neurotransmission through selective inhibition of the pre-synaptic uptake of serotonin (Fuller et al, 1978) resulting in increased synaptic serotonin concentrations, while the uptake of (Wong et al, 1974; Wong et al, 1975) and (Wong et al,

1974) is not affected at usual therapeutic dosages. Similarly, PX selectively inhibits reuptake of serotonin (Magnussen et al, 1982; Thomas et al, 1987). In vitro inhibition of tritiated serotonin ([ H]-5-hydroytryptamine), norepinephrine and dopamine uptake on rat hypothalamic synpatosomes by several antidepressants is shown in the Table 1.1 (summarized from Thomas et al, 1987, Nemeroff, 1994 and

Riesenman, 1995). The chemical structures of several SSRIs are also provided in

Figure 1.1 for comparative purposes.

Page 15 The SSRIs (FX, PX, sertraline, fluvoxamine and ) have a higher selectivity

for serotonin reuptake inhibition (lower 5-HT Kj than for epinephrine and dopamine)

than traditional tricyclics such as , imipramine, and . The

TCA, has selectivity for serotonin comparable to the SSRIs; however,

its is a potent inhibitor of norepinephrine reuptake, thus retaining

tricyclic characteristics.

Table 1.1 Comparison of neurotransmitter uptake inhibition of commonly used SSRI and tricyclic antidepressants.

Mean uptake inhibition constant (Ki) Selectivity of reuptake inhibition (nmol/L) of serotonin vs. others Serotonin Norepinephrine Dopamine Serotonin/ Serotonin/ (5-HT) norepinephrine dopamine

Paroxetine 1.1 350 2000 318 1818

Citalopram 2.6 3900 1500 Fluvoxamine 6.2 110 18 Sertraline 7.3 1400 230 191 32

Fluoxetine 25 500 4200 20 168 Clomipramine 7.4 96 9100 13 98 Amitriptyline 87 79 4300 0.9 49 Imipramine 100 79 0.8 Desipramine 1400 65 0.05 Summarized from Thomas et al., 1987, Nemeroff, 1994, and Riesenman, 1995

Page 16 Fluoxetine (Prozac) N Paroxetine (Paxil)

F

Figure 1.1 Chemical structures of selective serotonin reuptake inhibitors.

Fluoxetine is a bicyclic derivative of phenylpropylamine that is different structurally

and pharmacologically from other current antidepressants (Goodnick 1991). The p-

trifluroromethyl substituent on the molecule appears to contribute to the drug's high

selectivity and for serotonin reuptake, possibly as a result of its electron-

withdrawing effect and lipophilicity (AHFS 1994). Fluoxetine is a

(50:50) of two optical isomers, with each isomer exhibiting about equal potencies

both in vitro and in vivo (Wong et al, 1991). However, the SFX isomer is reported to

have a slightly longer duration of action than RFX in vivo (Lemberger et al, 1985).

Fluoxetine is N-demethylated to form pharmacologically active NFX in humans and

Page 17 experimental animals, and the plasma levels of NFX are comparable to those of FX at steady-state (Bergstrom et al, 1993). Norfluoxetine is also a potent serotonin reuptake inhibitor. Although SNFX is essentially equally effective as SFX, RNFX is relatively weak as a serotonin uptake inhibitor (Wong et al, 1991; Fuller et al, 1991;

Fuller et al, 1992). The formation of pharmacologically active SNFX likely contributes to the longer duration of action of SFX observed in vivo.

Currently, SFX and RFX are being developed by Sepracor Co. as single isomer drugs in collaboration with Eli Lilly & Co. SFX is in phase II clinical trials for the prevention of migraine headache (Sepracor 1999). In addition, RFX is in preclinical testing because of potential therapeutic advantages associated with its shorter half-life

(Sepracor 1999; Miningco.com 1998). Unlike FX, PX is a single isomer drug, and has no pharmacologically active metabolites.

Fluoxetine, NFX, PX and other SSRIs exhibit similar pharmacological characteristics in spite of their different chemical structures (see Figure 1.1). Therefore, the selection of a particular SSRI should be based on the difference in pharmacokinetics

(i.e. half-life and volume of distribution), metabolism (i.e. pharmacologically active metabolite) and adverse effect profile observed in the particular patient population.

Pharmacokinetics of fluoxetine and norfluoxetine.

Absorption & Distribution: Fluoxetine is well absorbed after oral administration. The systemic is about 72-90% with peak plasma concentrations occurring at 6-8 h (Lemberger et al, 1985; Bergstrom et al, 1986; Aronoff et al, 1984). The rate of absorption of FX and the time to reach peak plasma concentration are delayed

Page 18 by food, but the area under the curve (AUC) is not affected (Lemberger et al, 1985).

A first-pass effect of 30% was found in dogs based on the comparison of AUC between oral and intravenous bolus administration (Bergstrom et al, 1986).

Fluoxetine and NFX are highly bound (95%) to plasma protein in humans (Aronoff et al, 1984) and rats (Caccia et al, 1990). The apparent volume of distribution of the drug is about 12-43 L/Kg in humans (Aronoff et al, 1984; Benfield et al, 1986), suggesting extensive tissue distribution. With long-term administration of FX to rats and dogs, the highest concentrations of drug and metabolites are observed in the lung

(rats) and (dogs) (Benfield et al, 1986). Fluoxetine and NFX have been reported to accumulate extensively in the brain of rats (Caccia and Fracasso 1992), mice (Fuller and Snoddy, 1993) and humans (Komoroski et al, 1994; Karson et al,

1993; Renshaw et al, 1992). The FX and NFX concentrations in the brain exceeded those in plasma by 2-3 fold (Renshaw et al, 1992) and up to 20-40 fold in other studies (Pohland et al, 1989; Fuller and Snooddy 1993; Benfield et al, 1986). In a preclinical toxicology study, rats fed FX as part of their daily diet for up to 90 days exhibited the formation of foam cells in the lung, a manifestation of drug-induced phospholipidosis (Wold et al, 1976). Significant accumulation of FX was observed in the lungs (Pohland et al, 1989), which are enriched with . Due to the acid nature inside the , FX would be expected to ionize and become trapped.

This may contribute then to the high volume of distribution that has been reported for this drug (Daniel & Wojcikowski, 1997a, b).

Elimination & Metabolism: Following the oral administration of 14C-FX (30 mg) in humans, plasma concentrations of total radioactivity and FX peaked between 4-8

Page 19

) hours and declined slowly, with a half-life of 1-3 days for FX and 7-15 days for NFX

(Lemberger et al, 1985). Approximately 65-80% of the administered dose of radioactivity (FX and its metabolites) was recovered in the urine over a 3 5-day period, and 12-15% in the feces. Of this, only 3-10% was recovered as free FX, 7-

10%o as free NFX, 5-7% as FX glucuronide, and 8-10% as NFX glucuronide

(Lemberger et al, 1985; Bergstrom et al, 1988). More than 20% of the radioactivity was recovered as hippuric acid, a glycine conjugate of benzoic acid (Bergstrom et al,

1985). The proposed metabolic pathway of FX is shown in Figure 1.2. The kinetics of FX are not significantly affected by renal impairment (Aronoff et al, 1994), but are altered by hepatic dysfunction (Schenker et al, 1988) with the half-life of FX being significantly longer (7.6 days vs. 2.8 days) in cirrhotic subjects. Thus, the elimination of the drug appears to depend primarily upon hepatic metabolism, with formation of the pharmacologically active metabolite, NFX, in laboratory animals

(Caccia et al. 1990) and in humans (Lemberger et al, 1995; Shenker et al, 1988).

There are several kinetic studies in humans, dogs, guinea-pigs, mice and rats

(Goodnick 1991; Sommi et al. 1987). In rats, the plasma half-life of FX is much shorter (~5 h) than that in humans (1-3 days) (De Vane 1991). For NFX, the half-life is much longer than that of FX, 7-15 days in humans and about 15 hours in rats. The large interindividual variability observed in the pharmacokinetics of FX (e.g., elimination half-life) in human subjects has been suggested to be the result of polymorphic oxidative drug metabolism mediated through the polymorphic 2D6 isoenzyme (De Vane 1991; Otton et al, 1993; Riesenman

1995).

Page 20 Steady-state concentrations ranged from 35-415 ng/ml for FX and 40-260 ng/ml for

NFX in depressed patients after taking 20 mg p.o. daily for 60 days (Benfield et al,

1986). With long-term administration, steady-state plasma FX levels were achieved within 2 to 4 weeks, and no further accumulation occurred after FX administration for more than 1 year (Benfield et al, 1986). The pharmacokinetics of FX have been reported to be non-linear, since a higher dose resulted in disproportionately higher plasma concentrations (Altumura et al, 1994; Bergstrom et al, 1985). Comparision of single dose versus steady-state showed that the half-life of FX was longer (5.7 vs.

1.9 d), and clearance lower (10.8 vs. 35.5 L/h) in male subjects after chronic therapy

(Bergstrom et al, 1985). The areas under the plasma concentration versus time curves (AUC) were larger following multiple dosing compared to those from the same dose given only once. Non-linearity was also observed in a steady-state study in depressed patients receiving the drug for more than 1 year (Bergstrom et al, 1986;

Pato et al, 1991). However, when steady-state concentrations are reached, after -2-6 weeks, they do not change with further drug administration (Bergstrom et al, 1986).

Non-linearity has also been observed in rats following both intravenous and oral dosing (Caccia et al. 1990).

It has been suggested that the dose dependency observed for FX may be due to partial saturation of first pass metabolism (Caccia et al, 1990; Pato et al, 1991). In the study by Pato et al, (1991), discontinuation of chronic FX therapy resulted in prolonged elimination half-lives for both FX and NFX compared with values following acute dosing. In this case, end-product inhibition, in which the metabolite,

NFX, may serve as a competitive substrate for further metabolism of FX, was

Page 21 suggested as a possible mechanism (DeVane 1994). Auto-inhibition of FX metabolism has also been proposed (Preskorn 1994). However, these hypotheses are based on indirect observation of relatively limited pharmacokinetic data. The interpretation of metabolism and pharmacokinetics of SSRIs is complicated by the fact that most SSRIs are both substrates and inhibitors of the hepatic cytochrome

P450 enzyme system. In vitro and in vivo studies indicate that the SSRIs are metabolized by several P450 isoenzymes including 1A2, 2C9, 2C19, 3A3/4 and 2D6

(Otton et al, 1993; Stevens and Wrigthon 1993; Riesenman 1995; Preskorn 1994;

Aspelet et al, 1994; Shen 1995; von Moltke et al, 1997). Different SSRIs competitively and reversibly inhibit isoenzymes of P450s to varying degrees. Of these isoenzymes, CYP2D6 has been the most extensively studied. In vitro studies show that both FX and NFX are potent inhibitors of this isoenzyme (Otton et al,

1993; Riesenman 1995). This observation has also been made indirectly from in vivo studies using desipramine as a probe for CYP2D6, in which case co-administration

resulted in significant increases in the Cmsx and AUC of desipramine (Riesenman

1995). More recently, Alfaro et al (1999, 2000) have reported that over an 8-day infusion of FX to human volunteers, there was significant inhibition of CYP2D6, using the urinary ratio of /dextorphan as a probe. Thus, the current opinion is that FX and NFX isomers are a potent inhibitor of the CYP2D6 and to lessor degree of 2C9, a moderate inhibitor of 3A3/4 and 2C19 and a low to minimal inhibitor of 1A2 (Riesenman 1995).

Page 22 The N-demethylation FX, the first step in the oxidative pathway, results in the formation of pharmacologically active NFX. From previously published reports

(Stevens and Wrighton 1993, von Moltke et al, 1997), the involvement of CYP2C9 and to a much lesser degree CYP2C19 and CYP2D6 was suggested. Steven and

Wrighton (1993) showed relatively higher levels of N-demethylation for RFX than that of SFX in human microsomal preparations. Von Moltke et al, (1997) have suggested that the N-demethylation is mainly mediated by CYP2C9. The contribution of CYP2D6 to N-demethylation was largely discounted. However, these studies have used a FX concentration (100 uM = 31000 ng/ml) for isozyme characterization which is much higher than the Kj of FX for CYP2D6. One of reasons for the use of such a high concentration was due to analytical limitations

(Greenblatt 1998, personal communication). At this high concentration, FX is likely to inhibit its own N-demethylation mediated by CYP2D6. Therefore, the involvement of CYP2D6 enzyme in vivo or in vitro at concentrations in the therapeutic range cannot be dismissed. Determination of FX N-demethylation at a lower concentration is needed for a more complete understanding of FX pharmacokinetics and metabolism.

Page 23 Figure 1.2 Proposed metabolic pathway of fluoxetine. (Proposed chemical structures are shown in brackets).

Page 24 CYP 2 D6 pharmacogenetics and FX pharmacokinetics:

In contrast to the published in vitro metabolism studies of FX discussed above

(Steven and Wrighton, 1993; von Moltke et al, 1997), Hamelin et al (1996) and

Fjorside et al (1999) have suggested that CYP2D6 is involved in FX disposition.

Compared to extensive metabolizers (EMs) of debrisoquine (CYP2D6), a

significantly higher AUC and terminal half-life of FX were observed in poor

metabolizers (PMs) following a single dose of racemic fluoxetine (20 mg) (Hamelin

et al, 1996). Furthemore, the AUC of NFX and total amount of NFX excreted in

urine was much smaller in the PMs compared to the EMs, while the total amount of

FX in urine was about 3 fold higher in PMs than EMs. Therefore, these data suggest

that CYP2D6 contributes significantly to the N-demethylation of FX (Hamelin et al.,

1996). Similarly, much higher systemic clearance of FX in the EMs compared to

PMs of CYP2D6 was also observed by Fjordside et al. (1999). In this study the

stereoselective pharmacokinetics of the FX isomers were examined, indicating no

significant difference in systemic clearance of RFX (40.1 L/h) and SFX (35.7 L/h) in

the EMs of sparteine (i.e. CYP2D6) after a single dose of racemic FX (60 mg). In

contrast, a significantly higher systemic clearance of RFX (17.3 L/h) compared to

SFX (3.0 L/h) was observed in PMs.

1.10 Pharmacokinetics of paroxetine

Absorption and distribution: Paroxetine is also well absorbed following oral

administration, with a systemic bioavailability of about 50% following a single dose

(Lund et al, 1979). The rate of absorption and time to reach peak plasma

Page 25 concentration (~ 3-8 h) are not affected by the presence of food, milk or antacids

(Kaye et al, 1989, De Vane 1992). Paroxetine also binds extensively to plasma proteins (-95%) (Dechant and Clissold 1991; Grimsley and Jann 1992). The apparent volume of distribution of PX is about 17 L/Kg in humans (Nemeroff 1994), which suggests extensive tissue distribution and plasma protein binding, but less than

FX or NFX. The half-life of these two drugs in humans is remarkably different, approximating -2-3 days for FX and -10-16 h for PX. These differences are more significant when the half-life of NFX (-7-15 days) is also considered.

Metabolism: Paroxetine undergoes extensive first-pass metabolism (Lund et al,

1979) and is metabolized extensively in the liver. Less than 1% of total dose of PX is excreted as unchanged drug in urine (Kaye et al, 1989). At least 85% of the dose is metabolized via oxidative cleavage of the methylenedioxy-bridge to an unstable catechol intermediate (Figure 1.2) by CYP isozymes (possibly CYP2D6) (Kaye et al,

1989; Bloomer et al, 1992). This intermediate undergoes subsequent methylation via catechol-O-methyltransferase (Sindrup et al, 1992a,b) and further conjugates to form pharmacologically inactive glucuronide and sulphate metabolites. Recently, the involvement of CYP3A4 in PX metabolism in addition to CYP2D6 was suggested

(Kuss and Hergerl, 1998).

Page 26 Metabolite M1 Metabolite M2 Metabolite M3

Figure 1.3 Proposed metabolic pathway of paroxetine.

1.11 Placental transfer of therapeutic agents.

Placental transfer of most therapeutic agents from mother to the fetus has been

recognized for many years, and there have been extensive studies on the factors

affecting placental transfer of these agents and subsequent fetal exposure. Although

there are significant anatomical and histological differences among the placentas of

different mammalian species, the barrier limiting the movement of therapeutic agents

and other compounds is essentially a single or multiple phospholipid membrane.

Therefore, substances can move across the placenta by the normal biological

membrane transport mechanisms (i.e. passive diffusion, carrier-mediated transfer

[facilitated diffusion], active transport). In addition, pinocytosis, phagocytosis and

Page 27 passage though the paracellular aqueous pores (Reynolds and Knott 1989) may also

play a role in placental drug transfer. Of these, passive diffusion appears to be the

major mechanism for placental transfer for most therapeutic agents (Rurak et al,

1991). Thus, the rate and extent of placental transfer is largely dependent on the

physicochemical characteristics (i.e. lipophilicity, degree of ionization, molecular

weight) of the particular therapeutic agent (Reynolds and Knott 1989). For lipophilic

compounds with a low molecular weight (<500) such as FX and PX, placental

transfer is almost unimpeded (Reynolds and Knott 1989; Rurak et al, 1991).

1.12 Methods of studying placental transfer and fetal exposure.

Several methods have been developed to quantitatively determine the extent of fetal

drug exposure in both human and experimental animal models.

Single-point determination at parturition: Detailed pharmacokinetic studies in the

human fetus are limited by technical feasibility and ethical concerns. Therefore, most

human data on placental transfer and fetal exposure to therapeutic agents are based on

the single-point determination of drug concentrations in maternal and fetal (i.e.

umbilical cord) blood at birth. While such data may provide the most clinically

relevant information on the fetal exposure near term, it should be interpreted with

caution. For drugs with a high clearance and short half-life, there are two

confounding factors, which make the interpretation of maternal-cord drug

concentration data difficult. Due to the limitation of a single time point, measured

drug concentrations would be dependent on the time of sampling, as well as on the

administration of single or multiple (i.e. attainment of steady-state) doses of the drug.

Page 28 In addition, the placenta undergoes considerable trauma during labour and the uterine and umbilical blood flow patterns are altered (Hamshaw-Thomas et al, 1984).

Therefore, drug concentration data collected at the time of birth may not be a good quantitative representation of fetal exposure during pregnancy. On the other hand, drugs with a low clearance (relatively long half-life), such as FX and other SSRIs, are at steady-state following chronic therapy. Also, alterations in uterine and umbilical blood flow during labour may not have as significant an effect on drug transfer in this case. Therefore, a single-point concentration determination at parturition may provide a better representation of fetal drug exposure at term for low clearance drugs.

Single-point determination during pregnancy (cordocentesis): Using advanced ultrasound techniques, fetal blood sampling can be performed during pregnancy

(Brown et al, 1990; Moise et al, 1990). However, the use of this technique for the routine clinical study of fetal drug concentrations, in the absence of obvious fetal abnormalities, is questionable due to relatively high risks on fetal outcome.

In vitro human placental perfusion: Several methods have been described in the literature using isolated human placenta (Bloom et al, 1997; Bourget et al, 1995;

Bassily et al, 1995). This is the most feasible method to obtain detailed drug transfer data in humans (Rurak et al, 1991). In addition, the effect of alteration in protein binding and perfusion rate can be evaluated in these preparations (Krishna et al,

1993; Bassily et al, 1995). As with any experimental situation, direct extrapolation to fetal in vivo drug exposure needs to be made cautiously.

Page 29 In vivo animal studies: Due to the limitations imposed by the methods described above, studies in appropriate animal models are necessary to obtain pharmacokinetic data. Pharmacokinetic studies during pregnancy are commonly conducted in small mammalian species such as mice, rats, guinea pigs and rabbits.

However the relatively small volume of blood available in these species and the requirement for serial blood sampling make these models less useful for detailed pharmacokinetic studies. These problems were circumvented by determining fetal pharmacokinetic profiles either by taking fetal samples from different animals or by removing different fetuses from a single animal at different time points (Laishley et al, 1989; Huang et al, 1996). However, collecting samples from different fetuses or animals may increase variability of the experiments (/'. e. need higher N numbers for statistics). Furthermore, these experiments were conducted in acute anesthetized animal preparations, which could potentially affect maternal and fetal drug disposition and placental transfer (Rurak et al, 1991). Therefore, several larger mammalian species are used for pharmacokinetic studies such as pregnant sheep, macaque monkeys (Macaca nemestrina and M. mulatto) and baboons (Szeto et al,

1978; Tuntland et al, 1998; Patterson et al, 1997; Garland et al, 1998).

The chronically instrumented pregnant sheep has been extremely useful in addressing both the effects and pharmacokinetics of drugs in the mammalian fetus in utero (Van

Petten et al, 1978). Both human and ovine fetuses show similar fetal behavioral states such as electrocortical activity towards the end of gestation (de Vries et al,

1982; 1985; Nijhuis et al, 1982) as well as similar fetal physiology (Comline and

Silver 1974; Szeto et al, 1978). The fetal sampling, both pharmacokinetic and

Page 30 pharmacological, that can be accomplished in this model cannot be duplicated in

humans, for the reasons stated earlier. Neither can it be performed in small animal

models due to limited blood volumes and the inability to chronically implant catheters

or monitoring devices in the mother and fetus. In spite of structural differences

between the human and ovine placenta, this chronically instrumented preparation

does provide valuable information on the pharmacokinetics and pharmacodynamics

of drugs in the ewe as well as the fetus in utero.

1.13 Pharmacokinetics of fluoxetine and paroxetine during pregnancy.

The dynamic physiological changes during pregnancy influence the maternal-fetal

pharmacokinetics. As the glomerular filtration rate usually increases during

pregnancy, renal drug elimination is generally enhanced (Krauer and Krauer, 1997),

whereas hepatic drug metabolism may increase, decrease or remain unchanged. A

mean increase of 8 L in total body water alters drug distribution and results in

decreased peak serum concentrations of many drugs (Loebstein et al, 1997).

Decreased steady-state concentrations have been documented for many agents as a

result of their increased clearance. In addition, changes in plasma protein binding

during pregnancy could affect the maternal pharmacokinetics (Perucca and Crema,

1982). The placental and fetal metabolic capacity with physiological factors, such as

differences acid-base equilibrium of the mother versus the fetus, determine the fetal

exposure to the drugs taken by the mother.

There is very limited information on the pharmacokinetics of FX and PX during

pregnancy. Case reports (Spencer 1993, Mhanna et al, 1997) on FX and NFX

Page 31 exposure at birth and the immediate postpartum period indicate that FX and NFX

cross the placenta easily. Similarly, a case report describing the accumulation of PX

in the neonate during the early perinatal period indicates that this drug also crosses

the placenta (Dahl et al, 1997)

In the rat, placental transfer and fetal distribution of 14C-fluoxetine occurs, based

upon tissue drug assay and whole-body autoradiography (Pohland et al, 1989). The

fetal brain and thymus had the highest levels of FX (and radioactivity) and these were

comparable to those in the maternal brain and thymus (Polhand et al, 1989). Fetal

cardiac output is greater and a higher proportion is distributed to the brain compared

to the adult. Combined with greater blood-brain permeability, this leads to more

rapid and more extensive drug exposure of the fetal brain (Lewis 1978). In addition,

plasma protein content and binding affinity are less in the fetus, thus leading to higher

tissue distribution (Rayburn and Andersen 1982). In the following chapters, the

extent of fetal FX exposure and detailed pharmacokinetics of FX and NFX in the

ovine maternal-fetal unit will be discussed. In combination with these in vivo animal

data, fetal/neonatal exposure in humans will also be examined.

No detailed information on the pharmacokinetics of PX during pregnancy is presently

available.

1.14 Breast-feeding and pharmacotherapy.'

A Canadian survey on breast-feeding found that the national breast-feeding rate has

dramatically increased from 38% in 1963 to 75% in 1982 (McNally et al, 1985), with

the highest rate (86%) observed in British Columbia. Over the same period, the

Page 32 duration of breast-feeding also increased; at the national level, 59% of nursing mothers continued for at least four months, and 41% continued for six months

(McNally et al, 1985). Similar trends have been noted in other developed countries; a number of surveys indicate that more than 50%) of babies are breast-fed, and that number is increasing in the US and Europe (Ford and Labbok 1990; Cunningham et al, 1991).

The Canadian Paediatric Society and American Academy of Pediatrics emphasize breast-feeding as the best nutritional mode for infants for the first 6 months of life

(Briggs et al, 1994; Canadian Paediatric Society Nutrition Committee 1978; 1979;

Committee on Nutrition on the American Academy of Pediatrics 1982, 1992). Lower rates of gastrointestinal and respiratory disease, allergies, otitis media and other diseases were observed in breast-fed infants compared to bottle-fed infants (Hamosh

1996; Chen et al, 1988; Wilson 1983; Riordan 1997; Arita et al, 1997). In addition, studies have suggested significant psychological benefits of breast-feeding for both the mother and the infant (Briggs et al, 1994). Breast-feeding also contributes to mother-infant bonding and cognitive development (Rogan and Gladen 1993).

Drug therapy for mothers with postpartum depression is complicated by the need to consider the effects of neonatal drug exposure. In many cases, abrupt weaning of the infant is prescribed so that drug therapy can begin, but this may precipitate conflicting feelings of guilt and cause hormonal upset in the mother, both of which may exacerbate the . The personal and socio-cultural importance placed on breast-feeding by mothers should also be considered.

Page 33 1.15 Lactation and excretion of medication in breast milk.

Lactation usually begins after delivery with a rapid decrease in levels

and an increase in prolactin levels. Infant suckling maintains lactation by maintaining

higher prolactin levels (Suri et al, 1998). Psychotropic medications including the

SSRI antidepressants are excreted into breast milk mainly by passive diffusion of the

unionized and unbound fraction (Schanker 1962; Murray and Seger 1994; Buist et al,

1990). The extent of drug accumulation in breast milk is dependent on the degree of

ionization of the drug (i.e. pKa), the pH difference between plasma (7.4) and milk

(6.8-7.2), protein binding, and lipid solubility (Vorherr 1974). Basic drugs will be

more highly ionized in breast milk than in plasma and hence will be 'trapped' in the

milk compartment, due to the relatively low pH (6.8-7.2) of milk compared to plasma

(7.4) (Atkins et al, 1988). There is therefore, the potential for higher accumulation

of basic drugs in the breast-milk.

Blood flow to the breast, drug metabolism in breast tissues and milk composition (i.e.

protein and lipid content) also influence drug excretion (Edward, 1981; Welch and

Findlay 1981). The proteins in milk are either synthesized de novo in the mammary

glands (casein, lactalbumin and lactoferrin) or derived from other sources such as

plasma. The protein content of mature milk shows little diurnal or feed-to-feed

variation (Hytten, 1954). The protein binding of drugs in human milk is less than in

plasma (Atkinson and Breff 1987), a feature consistent with the lower protein content

of milk (8-12 g/L in mature milk and 35 g/L in colostrum) compared to plasma (75

g/L) (Allen et al, 1991). The lipid content of milk, however, varies considerably

within a feed, between feeds and between individuals (Neville et al, 1984). Diurnal

Page 34 (maximum at 10 am) and seasonal variations have been observed (Hytten, 1954).

The average lipid level in milk is about 40 g/L (range 20-58 g/L). Foremilk is low in

fat, while hind-milk has 2 to 3 times more fat (Neville et al, 1984). Lipid-soluble

drugs are found concentrated in milk lipid (Syverson and Ratjke, 1985) in both

animal and human studies. Thus, the milk-to-maternal plasma concentration ratio

(M/P ratio) of lipid soluble drugs is dependent on the lipid content of the milk

(Rasmussen, 1973; Matherson and Skjaeraasen 1988).

1.16 Excretion of fluoxetine, norfluoxetine and paroxetine in breast milk.

There are several reports concerning SSRI use during the nursing period, which

suggest that SSRIs are excreted into breast milk. Prospective studies have been

reported for FX (Taddio et al, 1996; Kristensen et al, 1999) and PX (Ohman et al,

1999; Begg et al, 1999; Stowe et al, 2000) excretion in the milk. Like most

antidepressants, which are in general lipophilic and basic compounds, PX and FX

(and its metabolite NFX) are excreted in breast milk of nursing mothers. Considering

the immature renal and hepatic functions present in the infant, there is considerable

potential for the accumulation of all three of these compounds (including the active

metabolite, NFX), given their long elimination half-lives.

There are several case reports on FX and NFX disposition in human breast milk, with

M/P ratios ranging from 0.14-0.50 and 0.09-0.35, respectively (Isenberg 1990; Burch

and Wells 1992; Lester et al, 1993). Two of these reports concluded that exposure to

FX through breast milk had no apparent adverse effects on the infants (Isenber 1990;

Burch and Wells 1992). The other reported colic and neonatal toxicity (increased

Page 35 irritability, poor sleeping, vomiting, loose stool) in a suckling infant whose mother

was taking a 20 mg daily dose of FX (Lester et al, 1993). Neonatal plasma FX and

NFX concentrations comparable to the adult therapeutic range were observed in spite

of a relatively small amount of drug transferred in breast milk (Lester et al, 1993).

However, the concentrations reported by Lester et al, 1993 appear to be much higher

than the levels reported by others (Taddio et al, 1996, Kristensen et al, 1999). The

M/P ratios for FX and NFX reported by Taddio et al, 1996 (0.88 ± 0.44 for FX and

0.82 ± 0.3 for NFX) and Kristensen et al, 1999 (0.68 for FX and 0.56 for NFX) are

considerably higher than the values from the case reports.

Similarly, there are several recent reports on the excretion of PX in human breast

milk (Spigset et al, 1996; Begg et al, 1999; Ohman et al, 1999; Stowe et al, 2000).

A wider range in the M/P ratio estimate for PX was reported by Spigset et al, 1996

(0.09), Ohman et al, 1999 (0.69 ± 0.29) and Begg et al, 1999 (0.39 ± 0.1 by AUC

method, 0.96 ±1.0 estimated from pre- and post-feed samples).

1.17 Neonatal drug disposition.

Neonates and infants are exposed to various therapeutic agents either by direct

administration (e.g. antibiotics) or indirectly through breast-feeding. Exposure and

subsequent pharmacological effects of these therapeutic agents depend on the

capacity of neonatal and infant drug absorption, distribution, metabolism and

elimination processes.

Absorption: Compared to older infants and adults, neonates and infants younger than

6 months have a prolonged gastric emptying time and higher gastric pH. These

Page 36 factors could potentially increase absorption of weakly basic compounds such as FX and PX, since a higher portion of these compounds would be unionized in these conditions. This may be of less importance for these compounds because of their already very high absorption characteristics (i.e. bioavailability) in the adult.

Distribution: Plasma/serum protein binding in the neonate/infant also plays an important role in drug disposition. Neonatal plasma protein binding depends on the amount of binding protein available, the affinity constant of the drug and the number of binding sites. Protein binding occurs to a lesser extent, both quantitatively (less protein) and qualitatively (less binding per protein molecule), in the neonate than in the adult. Therefore, the unbound fraction of drug is generally higher in the neonate.

A different degree of protein binding has been reported between newborn and adult plasma,for various therapeutic agents (Belpaire et al, 1995; Kurz et al, 1977a,b). A higher free fraction of therapeutic agents in the neonate/infant could result in increased exposure, thus amplifying the intended as well as the adverse effects of the agent compared to that in adults.

In addition, the blood-brain-barrier in neonate and infants is not fully developed compared to adults or older infants. This incomplete blood brain barrier and relatively immature central nervous system (Nahas and Gourjard 1978) may contribute to a higher sensitivity of infants to many therapeutic agents. Similarly, higher permeability of small molecules is observed in fetal lambs compared to adult sheep or newborn lambs (Saunders and Mollgard, 1984; Stonestreet et al, 1996)

Page 37 Elimination: In infants, renal elimination capacity of therapeutic agents is also lower than that of adults. The surface-area adjusted glomerular filtration rate (GFR) is 30-40% lower than that of adults during the early neonatal period, approaching normal adult values at -2.5 to 5 months (Rane and Wilson 1976). However, for compounds such as FX and PX, renal elimination of intact drug contributes only a small portion to overall drug elimination. Therefore, the relatively low renal elimination capacity of infants is not likely be an important factor for the elimination of SSRIs.

Metabolism: A relative immaturity of neonatal hepatic drug metabolism mediated by

CYP P450 enzymes often results in a longer half-life and thus a higher degree of drug exposure in newborns compared to adults (Besunder et al, 1988). Most phase I oxidative metabolic pathways are markedly impaired in the neonate (Miles 1983) and different hepatic cytochrome P450 isozymes develop at different rates (Mannering

1985; Morselli et al, 1980). For example, and phenobarbitone are metabolized at adult rates within 1 to 2 weeks of birth, whereas theophylline metabolism remains slower than in adults for up to 1 year (Mannering 1985; Dutton

1978). Recent studies suggest that the expression of the CYP2C subfamily and

CYP2D6 is absent in the human fetus and is low in first few months of an infant's life

(Treluyer et al, 1997; Jacqz-Aigrain and Cresteil 1992). Moreover, phase II reactions, such as glucuronidation, are impaired in the neonate to an even greater extent than phase I oxidative pathways (Dutton 1978). Together, these factors can contribute to considerable drug accumulation in breast-fed infants even though they may be receiving a relatively small amount of drug via breast milk.

Page 38 1.18 Rationale

There is a significantly elevated incidence of depression (~10-15%) during pregnancy

and the postpartum period compared to the general population. The use of

psychotropic drugs such as antidepressants, and for that matter most drugs, is

discouraged during these periods because of a lack of detailed information concerning

their safety. However, antidepressants such as FX and PX are being increasingly

used to treat depressive disorders. Drug therapy during pregnancy, in particular,

poses a difficult therapeutic problem due to the need to consider potential

teratogenicity, and fetal behavioral and neonatal developmental effects of drugs taken

by mothers, in addition to the primary concern for the health and safety of the mother.

It is further complicated by a high rate of unplanned pregnancies coupled with the

definite need for uninterrupted psychotropic drug treatment. These risks must be

weighed against the risks of untreated depression, which include poor self-care and

nutrition, disturbed sleep, lack of prenatal care, increased exposure to and

illicit drugs, and the risk of /infanticide. Compared to other antidepressants,

the SSRIs generally have a reduced side effect profile and may, therefore, provide an

alternative treatment option. However, there are no detailed data on the

pharmacokinetics of FX and PX in pregnancy and postpartum. This increasing use

has not been accompanied by either human or animal studies on the maternal-fetal

disposition of the drug or on its direct or indirect effects on the fetus.

In addition, the developing central nervous system (CNS) appears to be particularly

susceptible to toxic drug effects, partly because of its extended period of

development. While serious CNS morphological abnormalities generally do not occur

Page 39 in the second and third trimesters, the use of drugs during this period can result in subtle alterations in CNS function. At present, adverse effect information associated with SSRI use in the infant is limited to a few case reports listing complications such as irritability, tremor, respiratory depression and hypotonia.

In the Reproductive Psychiatry Unit at the BC Women's and St. Paul's Hospitals alone, -3500 pregnant women and nursing mothers are treated for clinical depression each year (-90% with psychotropic medication). The current therapy for these cases involves use of the SSRIs (FX, PX and sertraline to a lesser degree). However, the paucity of pharmacokinetic information and the concerns raised by case reports of adverse neonatal effects and observations from animal studies highlight the need for a detailed examination of the consequences of the fetal/neonatal exposure to these drugs. Both the psychiatrists treating these women and the patients themselves are in urgent need of this information in order to determine the drug therapy that is most effective in relieving depressive symptoms while minimizing adverse effects on the fetus and newborn. The studies presented here provide the type of information needed, thus benefiting women and families, as well as likely affecting clinical practice elsewhere.

The clinical investigations also represent a logical extension of our detailed pharmacokinetic studies in the chronically instrumented pregnant sheep. Using this model, we can assess, in detail, maternal and fetal drug pharmacokinetics and physiological effects over the last 2-3 weeks of gestation following short-term acute drug administration. Pharmacokinetic studies in the human fetus are limited by technical feasibility and ethical concerns. Thus, studies in appropriate animal models

Page 40 are necessary to obtain pharmacokinetic data. The chronically instrumented pregnant

sheep has been useful in addressing both the effects and kinetics of drugs in the fetus

(van Petten et al, 1978). The fetal sampling, both pharmacokinetic and

pharmacological, that can be accomplished in this model cannot be duplicated in

either humans, for the reasons stated earlier, or small animals models due to limited

blood volumes and the inability to chronically implant catheters or monitoring

devices in the mother and fetus. In spite of structural differences between the human

and ovine placenta, this chronically instrumented preparation does provide valuable

information on the pharmacokinetics of drugs in the ewe as well as the fetus in utero

(Rurakera/., 1991).

1.19 Objectives and Specific Aims

The overall objectives of this Ph.D. dissertation were to examine and compare the

pharmacokinetics of the selective serotonin reuptake inhibitors (SSRIs), FX and PX,

during pregnancy and the nursing period. In order to accomplish these objectives by

examining the pharmacokinetics of FX and PX in both animal and clinical studies, the

following specific aims were identified:

• To establish stereoselective analytical methods for fluoxetine and norfluoxetine in

biological fluids

• To establish analytical methods for paroxetine in biological fluids

• To characterize stereoselective pharmacokinetics of fluoxetine and norfluoxetine

in non-pregnant adult sheep

Page 41 To characterize stereoselective pharmacokinetics of fluoxetine and norfluoxetine

in pregnant sheep

To elucidate mechanisms for stereoselective fluoxetine and norfluoxetine

disposition

To characterize and compare fluoxetine and paroxetine disposition during

pregnancy and the nursing period in humans

Overall, the hypotheses of this thesis are that the fetus and neonate will be exposed to the SSRIs, fluoxetine and paroxetine, following maternal dosing and the degree of fetal and neonatal exposure to these compounds will depend on their physicochemical and pharmacokinetic characteristics. Furthermore, in the case of fluoxetine and norfluoxetine, stereoselective protein binding and metabolism will result in differential accumulation of one isomer over the other.

Page 42 CHAPTER 2

ANALYTICAL METHOD DEVELOPMENT

Several analytical methods have been developed and validated to support the animal and clinical pharmacokinetic studies of selective serotonin reuptake inhibitors [fluoxetine (FX), norfluoxetine (NFX) and paroxetine (PX)] during the course of this Ph.D. program. The analysis of FX and NFX isomers was accomplished by stereoselective analytical methods utilizing the following methods:

• Gas chromatography/mass spectrometry employing electron impact ionization

(GC/MS/EI) (Kim et al, 1995)

• Liquid chromatography/tandem mass spectrometry (LC/MS/MS) (Kim et al, 1998).

Similarly, the analysis of PX was accomplished by:

• Gas chromatography/mass spectrometry with electron impact ionization (GC/MS/EI)

(Kim etal, 1997)

• Gas chromatography/mass spectrometry with negative chemical ionization

(GC/MS/NCI).

Various analytical methods for determining FX and NFX concentrations in biological samples have been reported using gas chromatography (GC) and high performance liquid chromatography (HPLC) with various detection methods. (GC: Dixit et al, 1991; Torok-

Both et al, 1992; Lantz et al, 1993; Goodnough et al, 1995; Eap et al, 1996; Crifasi et al,

1997; Fontanille et al, 1997; GC/MS: Addison et al, 1998; HPLC: Wong et al, 1990; Peyton et al, 1991; Suckow et al, 1992; Thomare et al, 1992; Ramaiya et al, 1997). Similarly, analytical methods for PX using GC and HPLC have been reported (GC: Petersen et al,

Page 43 1978; HPLC: Brett et al, 1987; Gupta 1994; Hartter et al, 1994; Ramaiya et al, 1997; Bagli etal, 1997).

However, most of the analytical methods referenced above require a relatively large volume of sample (e.g. ~1 mL or more) and/or have a limited sensitivity (i.e., LOQ -5-10 ng/mL).

Due to technical, practical and ethical reasons, only a limited sample volume of fetal and neonatal plasma and/or serum can be obtained in both animal and clinical pharmacokinetic studies conducted during pregnancy and the postpartum period. For example, the volume of infant serum samples collected by heel-prick for phenylketonurea (PKU) testing are typically less than 100 uL and drug concentrations in these samples are usually in the low ng/mL range. Therefore, the analytical methods used in these studies must be sensitive and selective enough to detect small amount of analytes in the biological sample. For these reasons, developing very sensitive analytical methods for the analytes in question was an essential requirement for the proposed studies.

In the case of FX and NFX, a stereoselective analytical method was required to measure the concentrations of different optical isomers, which may have different pharmacological and/or pharmacokinetic properties. The separation of optical isomers has been achieved by two different methods:

• The FX and NFX isomers were converted to diastereoisomers, which have different

physicochemical characteristics, by derivatization with an optically pure derivatizing

agent. The derivatives are subsequently separated using a regular (i.e. non-chiral) GC

column [Section 2.3].

Page 44 • The application of a chiral stationary phase (protein-based HPLC column) in the

LC/MS/MS method enabled us to separate these isomers directly without the need for

derivatization [Section 2.5].

Details on the development and validation of these analytical methods are presented.

2.1 Materials

The reference standards, chemicals, reagents and other material used during analytical

method development and validation and sample analysis, along with information on

their purity (where applicable) and source, are listed below.

The reference standard, racemic fluoxetine hydrochloride [(#,S)-N-methyl-3-(p-

trifluoromethylphenoxy)-3-phenylpropylamine (>99% purity)], racemic norfluoxetine

hydrochloride [(R,S)-3 -(p-trifluoromethylphenoxy)-3 -phenyl-propylamine (>99%

purity)] and (fl)-norfluoxetine hydrochloride [RNFX, (R)-3-(p-

trifluoromethylphenoxy)-3-phenylpropylamine (>99% purity)] were generously

provided by Eli Lilly and Co. (Indianapolis, IN). Racemic fluoxetine hydrochloride

[(#,S)-N-methyl-3 -(p-trifluoromethylphenoxy)-3 -phenylpropylamine (>99% purity)]

for pharmacokinetic study was obtained from Sigma Chemical Co. (St. Louis, MO).

The internal standard for the GC/MS/EI method [2-(diphenylmefhoxy)-N-

methylethylamine hydrochloride] was synthesized in our laboratory. The deuterated

analogue of racemic fluoxetine hydrochloride ([ H5]-FX) [(/?,5)-N-methyl-3-(p-

trifluoromethylphenoxy)-3-[ Fy-phenylpropylamine] was synthesized and purified as

described later [see Section 2.4]. The deuterated analogue of racemic fluoxetine

Page 45 2 hydrochloride ([ HsJ-fluoxetine) was also subsequently obtained later from Isotec Inc.

(Miamisburg, OH).

Optically pure FX isomers, (/?)-fluoxetine [RFX, (/?)-N-methyl-3-(p- trifluoromethylphenoxy)-3-phenylpropylamine] and (^-fluoxetine [SFX, (5)-N- methyl-3-(p-trifluoromethyl-phenoxy)-3-phenylpropylamine] were separated from racemic fluoxetine hydrochloride in our laboratory as described in section 2.3.1.1.

The reference standard, paroxetine hydrochloride hemihydrate [(3S-trans)-3-[(l,3)- benzodioxol-5-yl-oxy)-methyl]-4-(4-fluorophenyl)-piperidine (>99% purity)] was generously provided by SmithKline Beecham Co. (King of Prussia, PA). The internal standard, maprotiline, was obtained from Sigma Chemical Co. (St. Louis, MO).

L- and D-mandelic acid for stereoselective re-crystallization were obtained from

Sigma Co. (St. Louis, MO) and used for the separation of the optical isomers of FX.

2 Deuterated ([ H5]-acetophenone, 99% isotopic purity) for the synthesis of deuterated FX was obtained from Cambridge Isotope Laboratories (Andover, MA).

Acetophenone, benzylmethylamine hydrochloride, para-formaldehyde, 10% palladium on carbon, N-methylpyrrolidone, potassium t-butoxide were purchased from Aldrich Chemical Co. (Milwaukee, WI) and used for chemical synthesis of FX and deuterium-labeled FX.

Ammonium acetate, sodium bicarbonate, glacial acetic acid, di-sodium hydrogen orthophosphate (dibasic), potassium di-hydrogen orthophosphate (monobasic), potassium chloride, pellets, magnesium chloride, and hydrochloric

Page 46 acid were obtained from BDH Chemicals Co. (Toronto, ON) and/or Fisher Scientific

(Nepean, ON) and were of analytical reagent or ACS grade. Anhydrous aluminum chloride, anhydrous sodium sulphate and petroleum ether used during synthesis procedures were also purchased from BDH Chemicals Co. (Toronto, ON).

"Sequanal" grade triethylamine (TEA) and heptafluorobutyric acid anhydride

(HFAA) were purchased from Pierce Chemical Co. (Rockville, IL). (S)-(-)- trifluoroacetylprolyl chloride (97% purity) was obtained from Aldrich Chemical Co.

(Milwaukee, WI). Acetonitrile, ethyl acetate, , , , , ethanol, isopropanol, , and n-hexane were purchased from Caledon Laboratories (Georgetown, ON) and were of distilled in glass GC or

HPLC grade.

De-ionized high purity water was produced on-site by reverse osmosis and subsequent filtration using a Milli-Q® water system (Millipore, Bedford, MA). Ultra high purity (UHP) helium for GC/MS analysis, high purity hydrogen for synthesis, and filtered for LC/MS/MS analysis and sample concentration were obtained from Wesco Gas Co. (Vancouver, BC).

Blank human plasma, serum and breast milk were obtained from Transfusion and

Lactation Services at BC Children's and Women's Hospitals (Vancouver, BC).

Blank ovine plasma and serum were prepared on-site from the Department of Animal

Science and/or British Columbia Research Institute of Children's and Women's

Health, the University of British Columbia (UBC) (Vancouver, BC).

Page 47 Glassware for extraction and sample preparation (15 mL Pyrex disposable culture

tubes and polytetrafluoroethylene (PFTE) lined crew caps) were obtained from

Corning (Corning, NY) and/or VWR Canlab (Vancouver, BC). Glassware for

chemical synthesis was purchased from Aldrich Chemical Co. (Milwaukee, WI),

Fisher Scientific Co. (Nepean, ON) or custom-made in-house (Vancouver, BC).

2.2 Instrumentation

2.2.1 Gas chromatograph-mass selective detector (GC/MSD)

A Hewlett-Packard (HP) model 5890 Series II gas chromatograph equipped with a

split-splitless capillary inlet system, a HP model 7673 autoinjector, a HP model

5971A quadrupole mass selective detector, and a Vectra 486 25T computer with HP

Chemstation model G1030A workstation software was utilized for the GC/MSD

analysis of FX and PX in electron impact (EI) mode (Hewlett-Packard, Avondale,

PA). The gas chromatograph was equipped with a DB5MS (30m x 0.18mm ID.; film

thickness 0.18 pm; 5% phenylmethylsilicone; J&W Scientific, Folson, CA) cross-

linked fused silica capillary column for FX and NFX analysis. A HP Ultra-2 (25m x

0.25mm ID; film thickness 0.25 pm; 5% phenylmethylsilicone, Hewlett-Packard,

Avondale, CA) cross-linked fused silica capillary column was used for PX analysis.

Samples were injected onto the GC-MSD via a 4 x 78mm deactivated glass inlet liner

and a Thermogreen LC-2 silicone rubber septum (J&W Scientific, Folson, CA) in the

splitless mode.

Page 48 2.2.2 Gas chromatograph-mass spectrometer (GC/MS)

A Hewlett-Packard (HP) model 5890 Series II gas chromatograph equipped with a

split-splitless capillary inlet system, a HP model 7673 auto injector, a HP model

5989A Mass spectrometer engine, and a HP59960 MS ChemSystem (HP-UNIX) was

utilized for the GC-MS analysis of PX in negative chemical ionization (NCI) mode

(Hewlett-Packard, Avondale, PA). The gas chromatograph was equipped with a HP

Ultra-2 (25m x 0.25mm ID; film thickness 0.25 pm; 5% phenylmethylsilicone,

Hewlett-Packard, Avondale, CA) cross-linked fused silica capillary column for PX

analysis. Samples were injected onto the GC-MS via a 4 x 78mm deactivated glass

inlet liner and a Thermogreen LC-2 silicone rubber septum (J&W Scientific, Folson,

CA) in the splitless mode.

2.2.3 Liquid chromatograph-tandem mass spectrometer (LC/MS/MS)

The LC/MS/MS system consisted of a HP 1909 II liquid chromatograph (Hewlett-

Packard, Avondale, PA) interfaced to a Fisions VG Quattro I triple quadrupole

tandem mass spectrometer (Micromass, Cheshire, UK). The operation of both

instruments and the acquisition of mass spectrometric data were handled by

MassLynx software running on a Window NT-based Pentium PC. Chromatographic

separations were carried out on an CHIRAL-AGP protein column (100mm x 2.1 mm

ID) column (ChromTech, Sweden) at ambient temperature. The autoinjector syringe

and sample loop volumes were 25 and 250 pL, respectively.

Page 49 2.2.4 Other equipment.

In addition, the following equipment was used for sample preparation and analysis:

Incubation oven (Isotemp model 350) (Fisher Scientific, Springfield, MA), IEC

model 2K centrifuge (Damon/IEC, Needham, MA), rotating mixer (Labquake model

415-110, Lab Industries, Berkeley, CA) and polarimeter (241 MC polarimeter with

sodium lamp, Perkin Elmer Corp., Norwalk, CT).

2.3 Stereoselective GC/MS/EI method for fluoxetine and norfluoxetine isomers

A stereoselective GC/MS/EI method was developed and validated for the

simultaneous quantitative determination of FX and NFX isomers in human, ovine and

rodent biological fluids and tissues

2.3.1 Methods

2.3.1.1 Separation of fluoxetine isomers using differential re-crystallization

Optically pure FX isomers were prepared by the classical resolution method using

fractional re-crystallization of their D- and L-mandelic acid salts based on the

procedure described by Robertson et a/., 1988.

A racemic mixture of FX hydrochloride (10.0 g) was converted to free base by

extracting with 200 mL of ethyl acetate after adding 50 mL of 1.0 M sodium

hydroxide solution. The ethyl acetate layer was separated and evaporated under

vacuum to obtain slightly yellow oil. The extraction and evaporation step was

Page 50 repeated twice to obtain additional racemic FX free base. The total racemic FX free base obtained was 6.4 g.

Preparation of (R)-fluoxetine hydrochloride

The resulting racemic mixture of FX free base (3.1 g, 10 mmol) and an equimolar amount of D-(-)-mandelic acid were dissolved in 40 mL of hot benzene. The solution was allowed to cool to room temperature and after 3 days, the precipitate was collected by filtration. The precipitate was dissolved in 15 mL of hot benzene and allowed to precipitate a second time. It was then filtered and dried. The resulting white power (1.1 g) was dissolved in diethyl ether (50 mL) and mixed with IM sodium hydroxide to generate FX free base. The ether layer was separated and dried under vacuum. The resulting yellowish oil was dissolved in 15 mL of ether/hexane

(50/50, v/v) and its hydrochloride salt was generated using hydrogen chloride gas.

The resulting white crystals were collected by filtration. A 1% (w/v) aqueous solution was prepared by dissolving 20 mg of white crystal in 2 mL of HPLC-grade water and the optical rotation was measured in Perkin Elmer 24IMC polarimeter

23 equipped with a sodium lamp. The measured optical rotation was [a] D +9.95° (c 1%

H20).

Preparation of (S)-fluoxetine hydrochloride

Similarly, the resulting racemic mixture of FX free base (3.1 g, 10 mmol) and an equimolar amount of L-(+)-mandelic acid were dissolved in 40 mL of hot benzene.

After 3 days, the precipitate was collected by filtration. This precipitate was again dissolved in 15 mL of hot benzene and allowed to precipitate a second time,

Page 51 following which it was filtered and dried. The resulting white power (0.9 g) was

dissolved in diethyl ether (50 mL) and mixed with 1M sodium hydroxide to generate

FX free-base. The ether layer was separated and dried under vacuum. The resulting

yellowish oil was dissolved in 15 mL of ether/hexane (50/50, v/v) and its

hydrochloride salt was generated using hydrogen chloride gas. The resulting white

crystals were collected by filtration. A 1% (w/v) aqueous solution was prepared by

dissolving 20 mg of white crystal in 2 mL of HPLC-grade water and the optical

rotation was measured in Perkin Elmer 24IMC polarimeter with a sodium lamp. The

23 measured optical rotation was [a] D-10.1°(c 1%H20).

These compounds were used to assign the chromatographic peaks of RFX and SFX

for the GC/MS/EI assay method.

2 Standard stock solution preparation

Aqueous stock solutions of racemic FX (10 pg/mL equivalent to free base) and NFX

(10 pg/mL equivalent to free base) were prepared by dissolving an appropriate

amount of the analyte in de-ionized water. Working/spiking solutions containing both

FX and NFX (100 and 1.0 ng/mL equivalent to free base of each isomer) were

prepared by serial dilution. An aqueous solution of the internal standard [2-

(diphenoylmethoxy)-N-methylethylamine hydrochloride] (10 pg/mL equivalent to

free base) was similarly prepared and diluted to the final working solution (2 pg/mL

equivalent to free base). These stock and final working solutions were stored at 4°C

for 6 months and their stability evaluated.

Page 52 2.3.1.3 Sample extraction

The analytes were extracted from the biological fluid samples using either single or

double liquid-liquid extraction procedures. For plasma/serum, urine, saliva, amniotic

and tracheal fluids, a single liquid-liquid extraction procedure was used. For human

breast milk, whole blood and rat tissue samples, a double liquid-liquid extraction

method was employed.

For a single liquid-liquid extraction, the biological samples (up to 1.0 mL) or spiked

standard and QC samples were pipetted into clean borosilicate tubes with

polytetrafluoroethylene-lined caps. The sample volume was adjusted to 1.0 mL with

de-ionized water, and 50 pL of the internal standard solution was added to each

sample except for the blank (control) sample. The samples were alkalinized by

adding 500 pL of 1.0 M sodium hydroxide solution. Extracting solvent (0.05 M TEA

in hexane: isopropanol 98:2 v/v, 7 mL) was added and the tubes capped. The samples

were vortex-mixed briefly and extracted with a slow rotary motion on a rotary shaker

for 20 min, after which they were cooled to -20°C for 10 min (to break any emulsion

formed during mixing). Following centrifugation at 3000 g for 10 min, the top

organic layer was separated and transferred to a clean borosilicate tube. The organic

layer was then evaporated to dryness under a gentle stream (5 p.s.i.) of nitrogen at

35°C using a Zymak Turbo Vap® LV Evaporator (Zymak Co., Hopkinton, MA). The

dried residue was reconstituted in 100 uL of derivatizing solution ((S)-(-)-N-

trifluoroacetylprolyl chloride (TFAP) in toluene (2 pL/mL) and vortex-mixed for 30

sec. Following heating at 60°C for 1 hour, the samples were cooled to room

Page 53 temperature and transferred to GC autosampler vials containing 350 pL disposable glass inserts. A 2 pL aliquot was injected into the GC.

For a double liquid-liquid extraction, the biological samples (up to 1.0 mL) or spiked standard and quality control (QC) samples were pipetted into clean borosilicate tubes with polytetrafluoroethylene-lined caps. The sample volume was adjusted to 1.0 mL with de-ionized water, and 50 pL of the internal standard solution was added to each except for the blank sample. The samples were alkalinized by adding 500 pL of 1.0

M sodium hydroxide solution. Extraction solvent (0.05 M TEA in hexane: isopropanol 98:2 v/v, 7 mL) was added and the tube capped. Samples were vortex- mixed briefly and subsequently extracted with a slow rotary motion on rotary shaker for 20 min after which they were cooled to -20°C for 10 min (to break any emulsion formed during mixing). Following centrifugation at 3000 g for 10 min, the top organic layer was separated and transferred to a clean tube. Diluted hydrochloric acid

(2 mL of 0.5 M HC1 solution) was added and the samples mixed for 20 min with a slow rotary motion. Following centrifugation for 5 min, the organic layer was aspirated and discarded and 500 pL of 3.0 M sodium hydroxide solution was added.

The extracting solvent (7 mL) was added again and the samples mixed for 20 min.

Following centrifugation at 3000 g for 10 min, the top organic layer was separated and transferred to a clean tube. The organic layer was then evaporated to dryness under a gentle stream (5 p.s.i.) of nitrogen at 35°C using a Zymak Turbo Vap® LV

Evaporator (Zymak Co., Hopkinton, MA). The dried residue was reconstituted in 100 uL of derivatizing solution ((S')-(-)-N-trifluoroacetylprolyl chloride in toluene) and

Page 54 vortex-mixed for 30 sec. The samples were derivatized at 60°C for 1 hour, cooled to

room temperature and then transferred to GC autosampler vials containing 350 uL

disposable glass inserts. A 3 uL aliquot was injected into the GC.

2.3.1.4 Gas chromatography/ mass spectrometry in electron impact mode

The derivatives (diastereoisomers) of FX and NFX were separated on a DB5MS

column using helium as the carrier gas. The injection was made in splitless mode

with an inlet temperature of 260°C. The oven temperature was 145°C initially and

increased by 40°C/min to 215°C, 2°C/min to 245°C and finally by 40°C/min to

300°C where it was held for 3 min. The mass selective detector transfer line

temperature was 300°C. The analytes were ionized in positive electron impact (EI+)

mode with selective ion monitoring (SIM). The ions monitored were m/z 341 for the

FX isomers, m/z 327 for the NFX isomers, and m/z 252 for the internal standard. The

limit of quantitation was 1.0 ng/mL using 200 uL sample for both the FX and NFX

isomers and the linear range was 1.0 to 500 ng/mL.

2.3.1.5 Calibration curve and regression model

Standard curves (1.0, 2.5, 5.0, 10, 25, 50, 100, 250 and 500 ng/mL for FX and NFX

isomers) were prepared by adding appropriate volumes of the prepared standard stock

solutions (100 or 1 ng/mL) to 200 uL of blank biological fluid. The internal standard

(100 ng) was then added to each sample. These samples were extracted and analyzed

in the same manner as described above. Weighted linear regression (weighting factor

= 1/y2) was used between the peak area ratio of the analyte and the internal standard

Page 55 versus the nominal concentrations. Linearity of standard curves was determined by

calculating the relative error (RE) and relative standard deviation (RSD) at each

nominal concentration. A RE and RSD < 15% (20% for the limit of quantitation

[LOQ]) was used as the acceptance criteria.

2.3.1.6 Extraction recovery

Absolute recovery for the extraction procedure was determined at LOW (4 ng/mL),

MED (40 ng/mL) and HIGH (400 ng/mL) FX and NFX isomer concentrations. Two

sets of samples (Recovery group and Control group) were prepared. The recovery

group was prepared in triplicate by adding 200 uL of LOW, MED and HIGH QC

sample and the internal standard. The control group was prepared in a similar

manner except 200 pL of blank plasma was added instead. All the samples were then

extracted according to the procedure described previously. After drying the organic

layer with nitrogen, an appropriate amount of the analyte in methanol was added to

the Control group and the samples were again dried under nitrogen stream. All the

samples were then reconstituted for derivative formation and analyzed as described

above. The absolute recovery was calculated as the ratio of measured concentrations

of Recovery (extracted) versus Control (unextracted) groups at each concentration

level.

2.3.1.7 Method validation

Intra- and inter-batch precision and accuracy were determined in the QC samples.

QC samples at the limit of quantitation (LOQ) (1.0 ng/mL each for FX and NFX

isomers), LOW, MED and HIGH concentrations were prepared by adding the

Page 56 appropriate amount of FX and NFX isomers to blank plasma or urine. These QC

samples were then divided into smaller aliquots (~1 mL) and frozen at -20°C until

analysis. For each sample run (batch), triplicate QC extractions were prepared by

pipetting 200 pL of QC sample into each tube and adding the internal standard. For

the determination of intra-batch variability parameters, 6 QC extractions were

performed. These QC extractions were carried out as described earlier. Intra-batch

precision and accuracy were determined by calculating the RSD and RE of the

measured concentrations of the 6 QCs at each of the 4 concentrations. Similarly,

inter-batch precision and accuracy were determined by calculating the RSD and RE

of the measured concentrations of 5 separate QC batches.

2.3.1.8 Analyte stability

A number of tests were carried out in order to evaluate the stability of the analytes

under routine sample handing conditions in the laboratory.

Freeze-Thaw Stability: Prepared QC samples were subjected to 3 freeze-thaw cycles

(-20°C-ambient temperature) prior to analysis. These concentrations were then

compared with those of QC samples which did not undergo this additional treatment.

Autosampler Stability: Extracted samples were stored at ambient temperature on the

GC autosampler tray and injected three times over a 48 period. The peak area counts

and the area ratio were subsequently evaluated.

Page 57 Bench-top stability: Prepared QC samples were left on bench top for 4 h at ambient temperature and then analyzed. These concentrations were compared with those of

QC samples, which did not undergo this additional treatment.

In all stability tests, the concentrations of analytes were compared between treated and untreated (nominal) QC samples. The analytes were considered stable if the measured concentration after the treatment was within 15% of the nominal value.

Page 58 2.3.2 Results and discussion

2.3.2.1 Mass spectrometric detection

Selective ion monitoring was used for the quantitation of the FX and NFX isomers.

Figure 2.1 shows the parent ion mass spectra of the TFAP derivatives of FX, NFX

and the internal standard in the EI mode.

laooooo

1600000-

1400000- COCF, 12,00000 - TFAP daitvattv* of FkMiMlna

1000000-

800000-

600000-

400000- 145 354377 417 446 4B3 200000- 1 |, U/.l jl 277 306

Avarage of 20.676 PROZAC.D (+,*)

2000000-

1800000-

1600000 - 1400000 Y-9 1200000 - COCF, TFAP darfvittra of Nsrfluc**** 1000000-

800000 -

600000 -

400000-

200000- 223

| | | 251277 403 446 4B8 517 100 150 250 300 350 400

750000 -i

700000 -

650000 -

600000 -

550000 -

500000 -

450000 -

400000 -

350000 -

300000 -

250000 -

200000 -

150000 - 86 152 100000 -

50000 A .,,„,:,; I,.!.,.,;,.i.I,, I I,„l„ 0 20 40 60 80 100 120 140 160 160 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500

m/z

Figure 2.1 Mass spectra and proposed fragment ions of the derivatives of fluoxetine, norfluoxetine and the internal standard in the electron impact ionization mode.

Page 59 For both FX and NFX, the major mass fragment ions were m/z 117 and 166, which represent the ions formed by the proposed fragmentation patterns (Torok-Both et al,

1992; Kim et al, 1995; Eap et al, 1996) shown below:

Fluoxetine and norfluoxetine derivatives

.. _ .?41. ( FX). 327 (NIFX) 117 7 ,R

F3C—4 j— 0;;-CHCH2C|*^ R = H (NFX) CO- CH3 (FX) N I

COCF3 •166

m/z 166 fragment ion m/z 117 fragment ion

H2 CH2CH2CH2+ H2

C C H2C^ \ H2C^ \

I CH2 I CH2

HC- / +HC^N/ N internal standard derivative \ \ COCF3 COCF3

m/z 341 (FX) and 327 (NFX) fragmentation ions 252

+ ^—CH CH2CH2N ?\ O r C0CF3 F3COC

Since m/z 166 fragment ions are formed from the derivatizing agent itself, there is potential for interference from the derivatives of other basic compounds in biological fluids. Thus, this ion was not used for the assay. Initially, the m/z 117 ion, which was the base peak for both FX and NFX, was evaluated for the assay. However, interference from the sample extract, especially following a single liquid-liquid extraction, was observed in some of control (blank) plasma samples. Therefore, the m/z 341 (FX) and 327 (NFX) fragment ions were chosen, as there was no chromatographic interference using these ions. In addition, the signal-to-noise ratio

Page 60 obtained using these ions was higher than that calculated for the m/z 111 ion at the

LOQ. A similar GC/MS method was reported later (Eap et al, 1996), which

monitors m/z 117 for both FX and NFX isomers but with a sensitivity limit (LOQ) of

5 ng/mL per isomer using 1 mL of plasma sample. For the TFAP derivative of the

internal standard, the fragmentation ion at m/z 252 was monitored.

During the course of the sample analysis, the mass spectrometric parameters were

modified to improve sensitivity of the assay. This is mainly achieved by maximizing

sensitivity in the m/z 300-350 range by manually adjusting tune parameters and

increasing electron multiplier voltage. Initially, the limit of quantitation was 0.5

ng/mL per isomer using a 1 mL sample volume (Kim et al, 1995). Subsequently, the

limit of quantitation was increased 2.5-fold to 1.0 ng/mL per isomer using a 0.2 mL

sample volume. The method was re-validated within this new concentration range

and data from this re-validation are presented in section 2.3.2.4, which follows. The

sensitivity of this assay (1.0 ng/mL using 200 pL of sample) represents a >50-fold

increase over the LOQ of 10 ng/mL using 1 mL of sample reported by Torok-Both et

al, 1992 and a >25-fold increase over the LOQ of 5 ng/mL using 1 mL sample

reported by Eap et al, 1996. In addition, the use of a single extraction procedure for

plasma and the absence of a drying step following derivatization greatly increased

recovery (85-95%) over these other reported methods.

.2.2 Chiral separation of fluoxetine and norfluoxetine isomers

Separation of FX and NFX isomers was achieved by the formation of their

diastereoisomers using the optically pure derivatizing agent, (5)-(-)-

Page 61 trifluoroacetylprolyl chloride (TFAP). This derivatization procedure performs two functions. First, diastereoisomers can be separated using conventional non-chiral GC columns such as the DB5MS column used in this assay since they have different physicochemical properties. Second, the derivatives have increased volatility, which improves their gas chromatographic characteristics compared to injections of underivatized analytes.

GC columns with different stationary phases (DB1, DB-5/Ultra-2, DB5MS and

DB1701) and dimensions (20 to 30 m length and 0.18 to 0.33 mm id) were evaluated for maximal separation (measured as resolution) for FX and NFX isomers. Among these columns, a mini-bore (0.18 mm id) column with a modified 5%- phenylmethylsilicone phase (DB5MS) provided the best resolution. The use of a smaller bore column (0.18 mm id) also provided better resolution of the NFX isomers, in particular, compared to that of larger bore columns (0.25 mm or 0.33 mm) which are used more conventionally.

Figure 2.2 shows a representative chromatogram of the FX and NFX isomers and the internal standard. Acceptable resolution of the FX and NFX isomers was achieved for reliable quantitative analysis. Complete baseline resolution was achieved (R>1.5) for the FX isomers, whereas the resolution was slightly less complete (R-1.2) for the

NFX isomers. No interference from other commonly prescribed medications was observed. The identity of the chromatographic peaks was determined by injecting the

TFAP derivative of the pure optical isomers of FX and NFX, which were either prepared as described in section 2.3.1.1 (for FX isomers) or provided by Eli Lilly &

Co (for NFX isomers). Data from these injections were also used to determine the

Page 62 degree of potential racemization of the analytes. Following separate injections of the

TFAP derivatives of SFX, RFX, and RNFX, approximately 2.3, 1.8 and 1.4 % of the relative peak area of the other optical isomer was observed, respectively. The identification of SNFX peak was inferred from the positive identification of the

RNFX peak, since pure SNFX was not available. Initially, the source of the impurity

(i.e. peaks for the other optical isomer) could not be definitely determined. Either the optical impurity was present in the reference material or racemization of the optically pure analyte occurred resulting in the observed impurity peaks. In either case, the proportion of these impurity peaks was considered to be negligible for quantitative analysis. Later, development of an LC/MS/MS assay method (see section 2.5), which does not require the derivatization of the analytes, proved that these peaks indeed came from an optical impurity in the single isomer preparations, not from racemization.

Page 63 25.8S6 08 Abundance (S)-Fluoxetine

90000 26.78 24.92 (R)-Fluoxetino 24.79 80000 (S)-Norfluoxetine IS

70000 H

(R)-Norfluoxetine 60000

50000

40000 -\

30000

20000

IOOOO H Ul Blank 0 1 1 1 1 1 • • • • i i 1 I " 1 I ' ' 1 ' I ' 24.50 25.00 25.50 26.00 26.50 27.00 27.50

Figure 2.2 Representative chromatograms of fluoxetine and norfluoxetine isomers using the GC/MS/EI method (blank ovine plasma and 250 ng/ml per isomer in ovine plasma). The retention order from left to right is SNFX, RNFX, RFX, SFX and the internal standard (IS).

Page 64 2.3.2.3 Extraction, recovery and stability

Several organic solvents such as ethyl acetate, hexane, toluene and dichloromethane

were evaluated with or without modifier (2-5% isopropanol) for maximal recovery of

the analytes. Triethylamine (0.05 M) was added to all solvent systems used in the

evaluation to reduce non-specific binding of the analytes to glassware.

Of the solvents tested, a hexane: isopropanol mixture containing 0.05 M TEA

provided the best extraction efficiency and the cleanest chromatograms. No

chromatographic interference from endogenous components in plasma or urine was

observed, using the SIM of the fragment ions m/z 327, 341, and 252 for NFX, FX and

the internal standard, respectively. Initially, a hexane: isopropanol (95:5 v/v) mixture

was used in the extraction. However, the proportion of isopropanol was reduced to

2% in order to eliminate interference observed in tissue and whole blood extracts.

The effect of this change in the overall recovery of FX and NFX was not significant.

The mean recoveries of SFX, RFX, SNFX and RNFX in human plasma were 91.7 ±

6.4%, 92.1 ± 7.1%, 84.5 ± 7.1% and 83.8 ± 6.9%, respectively, following single

liquid-liquid extraction. Similarly, the mean recoveries of the corresponding

compounds in ovine plasma were 92.3 ± 3.3%, 95.1 ± 6.8%, 82.9 ± 5.6% and 81.7 ±

4.7%, respectively. Similar recoveries were obtained for human urine and ovine

biological fluid samples (i.e. urine, tracheal and amniotic fluids). Mean recoveries of

SFX, RFX, SNFX and RNFX in human breast milk were 85.2 ± 7.2%, 82.5 ± 8.0%,

71.4 ± 2.1% and 72.8 ± 5.6%, respectively following double liquid-liquid extraction.

Page 65 A number of studies were conducted to evaluate the stability of the analytes in

biological fluids and prepared sample extracts under condition similar to routine

sample handling. The mean analyte concentrations measured in samples subjected to

3 freeze-thaw cycles, 4 h bench-top stability and 48 h autosampler stability were

found to be within ±10% of the nominal concentrations with an acceptable relative

standard deviation of <10%. Aqueous stock solutions of FX, NFX and the internal

standard were found to be stable at 4°C for up to 6 months with no noticeable

decrease (<5%) in analyte concentrations.

2.3.2.4 Method validation

Intra- and inter-batch variability and accuracy were calculated for the assay in human

and ovine biological samples. The summarized results for human plasma validation

are presented in Tables 2.1 and 2.2. Similar results were also obtained for the FX and

NFX isomers in other biological samples such as ovine plasma, urine and amniotic

fluid. The intra-and inter-batch variability (RSD) and accuracy (RE) were < 20% for

the QC LOQ and < 15% for the other QCs.

The standard curves for all the analytes [Figure 2.3] showed good linearity over the

concentration range tested (1.0 - 500 ng/mL). Weighted linear regression (weighting

factor = 1/y ) was used to reduce bias at the lower concentration range. A LOQ of 1.0

ng/mL was established for the FX and NFX isomers, based on the signal-to-noise

(S/N) ratio of > 10 with acceptable levels of variability and accuracy of less than

20%.

Page 66 The method validation protocol and acceptance criteria and the subsequent batch

acceptance criteria were adapted from the guidelines established by the American

Association of Pharmaceutical Scientists, Health Protection Branch and Food and

Drug Administration which is described in Shah et al, (1992).

In summary, this assay provides a sensitive and selective analytical method for the

stereoselective quantitation of FX and NFX isomers. It has been applied extensively

to stereoselective pharmacokinetic studies in the ovine maternal-fetal unit and clinical

studies during the ante- and postpartum periods, the results of which are presented in

subsequent chapters of this thesis.

concentration (ng/mL)

ure 2.3 Representative calibration standard curves for fluoxetine and norfluoxetine isomers in human plasma using the GC/MS/EI method. RFX: y = -0.0040 + 0.002757 x, r2 = 0.9994 SFX: y = -0.0083 + 0.002755 x, r2 = 0.9995 RNFX: y = 0.0078 + 0.002742 x, r2 = 0.9997 SNFX: y = 0.0002 + 0.002749 x, r2 = 0.9997

Page 67 Table 2.1 Intra-batch precision (RSD) and accuracy (RE) of fluoxetine and norfluoxetine isomers using the GC/MS/EI method in human plasma (n=6)

Analyte QC LOQ QC LOW QC MED QC HIGH (S)-fluoxetine Nominal cone, (ng/ml) 1 4 40 400 Measured cone, (ng/ml) 1.2 4.1 39.9 402.6 SD 0.1 0.3 2.6 9.5 RSD (%) 6.9 7.8 6.4 2.4 RE (%) 15.2 3.0 -0.4 0.7 (R)-fluoxetine Nominal cone, (ng/ml) 1 4 40 400 Measured cone, (ng/ml) 1.1 4.2 40.2 427.1 SD 0.1 0.2 3.5 10.2 RSD (%) 7.7 5.0 8.8 2.4 RE (%) 12.4 5.3 0.5 6.8 (S)-norfluoxetine Nominal cone, (ng/ml) 1 4 40 400 Measured cone, (ng/ml) 1.0 3.9 38.6 399.7 SD 0.1 0.2 2.4 6.7 RSD (%) 8.1 5.4 6.2 1.7 RE (%) -1.5 -2.3 -3.6 -0.1 (R)-norfl uoxetine Nominal cone, (ng/ml) 1 4 40 400 Measured cone, (ng/ml) 0.9 4.0 39.2 401.2 SD 0.1 0.2 1.5 5.5 RSD (%) 7.9 5.2 3.8 1.4 RE (%) -5.5 0.4 -2.0 0.3

Page 68 Table 2.2 Inter-batch precision (RSD) and accuracy (RE) of fluoxetine and norfluoxetine isomers using the GC/MS/EI method in human plasma (n=5)

Analyte QC LOQ QC LOW QC MED QC HIGH (S)-fluoxetine Nominal cone, (ng/ml) 1 4 40 400 Measured cone, (ng/ml) 1.1 4.2 40.7 399.3 SD 0.1 0.4 3.4 15.2 RSD (%) 11.0 8.7 8.4 3.8 RE (%) 10.5 6.0 1.8 -0.2 (R)-fluoxetine Nominal cone, (ng/ml) 1 4 40 400 Measured cone, (ng/ml) 1.0 4.3 41.2 415.3 SD 0.2 0.4 5.7 15.6 RSD (%) 14.7 9.7 13.8 3.8 RE (%) 4.5 8.0 3.1 3.8 (S)-norfluoxetine Nominal cone, (ng/ml) 1 4 40 400 Measured cone, (ng/ml) 1.0 4.0 39.3 395.7 SD 0.1 0.4 2.8 10.2 RSD (%) 13.0 10.5 7.1 2.6 RE (%) -2.5 0.3 -1.9 -1.1 (R)-norfluoxetine Nominal cone, (ng/ml) 1 4 40 400 Measured cone, (ng/ml) 1.0 4.2 40.1 398.4 SD 0.1 0.4 2.8 8.7 RSD (%) 10.1 9.5 6.9 2.2 RE (%) -2.5 3.8 0.3 -0.4

Page 69 2.4 Synthesis and purification of deuterium-labeled fluoxetine hydrochloride

2.4.1 The synthesis of 2-benzoyl-l-(N-benzyl-N-methyl)-ethylamine hydrochloride: Mannich reaction.

D5.acetophenone Para-formaldehyde N-methylbenzylamine

Material: Deuterium labeled acetophenone (D5-ring, 98%), N-methylbenzylamine hydrochloride, para-formaldehyde, absolute alcohol, and concentrated hydrochloric acid solution.

Procedure:

The desired product, 2-D5-benzoyl-l-(N-benzyl-N-methyl)-ethylamine hydrochloride

was synthesized by a Mannich ketonic base reaction between stable-isotope labeled

acetophenone (D5-acetophenone) and N-methylbenzylamine based upon the

procedure of Kuliev et al, (1971). Ds-acetophenone (5 mL) and

N-methylbenzylamine hydrochloride (8 g) were mixed with 1 mL of concentrated

hydrochloric acid solution. This mixture was vortex-mixed and refluxed at 80°C for

10 hours. The reaction was slowly cooled to 0°C, and the precipitated product was

filtered as white crystals with a yield of 9.46 g of 2-Ds-benzoyl-l-(N-benzyl-N-

methyl)-ethylamine hydrochloride.

Page 70 -.2 The synthesis of 1-Ds-phenyl-3-(N-methyl)aminopropan-l-ol: hydrogenation

Material: UHP hydrogen gas, 10% Pd on carbon, ethyl acetate, 2-D5-benzoy 1-1 -(N-benzyl-N-methyl)-ethylamine hydrochloride, hexane and 1 M sodium hydroxide solution

* denotes a chiral carbon y Procedure:

Nine grams of 2-D5-benzoyl-l-(N-benzyl-N-methyl)-ethylamine hydrochloride was

converted to free base using ethyl acetate extraction following alkalization with 50

mL of 1.0 M sodium hydroxide solution. The free base was hydrogenated (H2 gas)

with 150 mg of palladium on carbon (10%) using ethyl acetate as a solvent. The

reaction was allowed to proceed for 2.5 hours at a gas pressure of 75 p.s.i. at 50°C.

Following the hydrogenation step, the catalyst was removed by filtration and the ethyl

acetate was evaporated under vacuum at 50°C to form yellow-colored oil. Twenty mL

of hexane was added and the mixture was cooled to -5°C for 2 hours. The precipitate

was filtered and washed twice with 10 mL of cold hexane. The resulting precipitate

was dried under vacuum to a yield of 0.9 g of l-D5-phenyl-3-(N-

methyl)aminopropan-1 -ol.

Page 71 2.4.3 The synthesis of deuterium labeled fluoxetine

OH

I Etherificatlon

I (N-methylpyrrolidone, potassium t-butoxide)

T

Material: l-chloro-4-trifluoromethyl benzene, N-methylpyrrolidone, potassium t-butoxide, toluene, nitrogen gas, sodium sulfate, diethyl ether, and hydrogen chloride gas.

* denotes a chiral carbon Procedure:

Ten mL of N-methylpyrrolidone and 0.005 mole of potassium t-butoxide were

transferred into a nitrogen gas secured reaction vessel and mixed for 15 min. 0.005

mole of l-D5-phenyl-3-(N-methyl)aminopropan-l-ol and 0.01 mole of l-chloro-4-

trifluoromethylbenzene were added and mixed for 6 hours at 80°C. The solution was

extracted twice with 40 mL of toluene. After cooling to the room temperature, 60 mL

of water was added, and the solution extracted twice with 40 mL of toluene. The

toluene layers were combined and washed five times with 20 mL of water. The

toluene solution was further washed with 20 mL of saturated sodium bicarbonate

solution and brine and dried over sodium sulfate. The N-methyl-3-(p-trifluoromethyl-

Page 72 D5-phenoxy)-3-phenylpropylamine base was obtained as a brownish yellow oil following flash evaporation.

The oil was dissolved in diethyl ether, and subsequently acidified with hydrogen chloride gas at 0°C (to give an acidic ethereal solution). The ethereal solution was kept at -20°C for 5 days and a yellowish precipitate collected. This solid was re-

crystallized from acetonitrile at -20°C and washed with ether to provide 0.2 g of D5-

FX (white powder). The overall yield of deuterium-labeled FX was 3.5%.

Stereoselective LC/MS/MS method for fluoxetine and norfluoxetine isomers

The GC/MS/EI method described in section 2.3 has been extensively used in pharmacokinetic studies of FX and NFX isomers during this Ph.D. program. This method is sensitive and robust for routine sample analysis; however, the run time for the assay was relatively long (-25-30 min including the oven cool-down period between injections). This limited the total daily sample output to approximately 48-

50 including quality control and calibration curve samples. Therefore, an LC/MS/MS method which has a much shorter run time (-10 min) was developed and validated for the analysis of FX and NFX isomers. This technique has increased potential sample output to about 120-130/day. In addition, deuterium-labeled FX was used as the internal standard as opposed to the structurally quite dissimilar compound used in the GC/MS/EI method

Page 73 2.5.1 Method

2.5.1.1 Standard stock solution preparation

Aqueous stock solutions of racemic FX (10 pg/mL equivalent to free base) and NFX

(10 pg/mL equivalent to free base) were prepared by dissolving an appropriate

amount of the analyte in de-ionized water. Working/spiking solutions containing both

FX and NFX (100 and 1.0 ng/mL equivalent to free base of each isomer) were

prepared by serial dilution. A methanol solution of the internal standard (D5-

fluoxetine hydrochloride, 100 pg/mL equivalent to free base) was similarly prepared

and diluted to the working/spiking solution (100 ng/mL equivalent to free base).

These stock and spiking solutions were stored at 4°C for 6 months and stability of

aqueous stock and spiking solutions was evaluated over this interval.

2.5.1.2 Sample extraction

The analytes were extracted from plasma/serum by a single liquid-liquid extraction

procedure. Human or ovine plasma samples (200 pL) were pipetted into clean

borosilicate glass tubes with polytetrafluoroethylene (PTFE) lined caps. A 50 pL

aliquot of the IS solution was added to each sample and alkalinized by adding 0.5 mL

of 1.0 M sodium hydroxide solution. The sample volume was adjusted to 2.0 mL

with de-ionized water. Extracting solvent (0.05 M TEA in hexane: isopropanol 98:2

v/v, 7 mL) was added and the tubes capped. Samples were briefly vortex-mixed and

extracted with a slow rotary motion on rotary shaker for 20 min following which they

were cooled to -20°C for 10 min to break any emulsion formed during mixing.

Page 74 Following centrifugation at 3000 g for 10 min, the top organic layer was separated

and transferred to a set of clean tubes. The organic layer was then evaporated to

dryness under a gentle stream (5 p.s.i.) of nitrogen at 35°C using a Zymak Turbo

Vap® LV Evaporator (Zymak Co., Hopkinton, MA). The dried residue was

reconstituted in 100 pi of 2 mM ammonium acetate (pH 4.5) and 25 ul was injected

into the LC/MS/MS.

2.5.1.3 Liquid chromatography/ electrospray tandem mass spectrometry

The of FX and NFX were separated on a micro AGP-CHIRAL column

(100 x 2.0 mm, 5 pm) with a mobile phase of 5% acetonitrile in 2 mM ammonium

acetate buffer at pH 4.5. The flow rate was 0.25 mL/min and the column temperature

was ambient. The effluent from LC was directly introduced into the MS/MS through

an electrospray interface. The analytes were ionized in positive ion electrospray

(ESP+) mode with nitrogen as the nebulizing and bath gas. Argon was used for the

collision gas. The multiple reaction monitoring (MRM) mode was used for detection.

The ion transitions monitored were m/z 310 to 44 for the FX isomers, m/z 296 to 30

for the NFX isomers and m/z 315 to 44 for the internal standard ([2Ff5]-fluoxetine

isomers). These ion transitions were chosen based on the predominant fragmentation

patterns of the compounds in their daughter ion spectra.

2.5.1.4 Calibration curve and regression model

Standard curves were prepared by adding an appropriate volume of the prepared

standard stock solution to 200 pL of blank human or ovine plasma. (0.1, 0.25, 0.5,

Page 75 1.0, 2.5, 5.0, 10, 25, 50, 100, 250 and 500 ng/mL for FX and 0.5, 1.0, 2.5, 5.0, 10, 25,

50, 100, 250 and 500 ng/mL for NFX). The internal standard (5 ng per isomer of

[ H5]-fluoxetine) was then added to each sample. These samples were extracted and

analyzed in the same manner as described previously (Section 2.5.1.2). Weighted

linear regression (weighting factor = 1/y2) was used between the peak area ratio of the

analyte and the corresponding internal standard versus the nominal concentrations.

Linearity of the standard curves was determined by calculating the relative error (RE)

and relative standard deviation (RSD) at each nominal concentration. RE and RSD

values less than 15% (20% for the limit of quantitation) were used as acceptance

criteria for assay variability (Shah et al, 1992).

2.5.1.5 Extraction recovery

Absolute recovery of the extraction procedure was determined at LOW (4.0 ng/mL),

MED (40 ng/mL) and HIGH (400 ng/mL) QC concentration levels. Two sets of

samples (Recovery group and Control group) were prepared. The recovery group was

prepared in triplicate at each concentration by adding 200 pL of QC sample and the

internal standard. The control group was prepared in a similar manner except 200 pL

of blank plasma was added instead. All samples were then extracted according to the

procedure described earlier. After drying the organic layer, appropriate amounts of

analyte in methanol were added to the Control group and the samples further dried

under a nitrogen stream. All of the samples were then reconstituted and analyzed as

described in Section 2.5.1.2. Absolute recovery was calculated as the ratio of

Page 76 measured concentrations of Recovery versus Control groups at each concentration

level.

2.5.1.6 Method validation

Intra- and inter-batch precision and accuracy were determined for all QC samples.

QC samples at LOQ (0.1 ng/mL for FX and 0.5 ng/mL for NFX isomers), LOW (4.0

ng/mL), MED (40 ng/mL) and HIGH (400 ng/mL) levels were prepared by adding

the appropriate amount of FX and NFX isomers to blank plasma. These QC samples

were then divided into small aliquots (~ 1 mL) and frozen at -20°C until analysis.

For each batch analysis, triplicate extractions were prepared at each of the four

concentrations by pipetting 200 uL of QC sample into each set of tubes and adding

the internal standard (50 uL). For the determination of intra-batch variability, 6 QC

extractions were performed. The QC samples were extracted as described earlier in

Section 2.5.1.2. Intra-batch precision and accuracy were determined by calculating

the RSD and RE of the measured concentrations of the 6 QCs at each of the 4

concentrations. Inter-batch precision and accuracy were similarly determined by

calculating the RSD and RE of the measured concentrations of 5 separate batches.

The assay method was also cross-validated with our previous GC/MS/EI method

described in Section 2.3. Six QCs at each of the 4 concentrations (1.0, 4.0, 40 and

400 ng/mL) were analyzed by both LC/MS/MS and GC/MS/EI and the measured

concentrations compared.

Page 77 2.5.1.7 Analyte stability

Freeze-Thaw Stability: Prepared QC samples were frozen (-20°C) and thawed

(ambient) for an additional three cycles and then analyzed. These concentrations

were compared with those of QC samples which did not undergo this additional

treatment.

Autosampler Stability: The extracted samples were left on the GC autosampler tray

(ambient) and injected three times over a 48 h period. The peak area counts and the

area ratio were evaluated.

Bench-top stability: Prepared QC samples were left on the bench top for 4 h at

ambient temperature and then analyzed. These concentrations were compared with

those of QC samples, which did not undergo this additional treatment.

In all stability tests, comparison of the concentrations of analyte in the sample were

made between treated and untreated (nominal) QC samples. The analytes were

considered stable if the differences after treatment were within 15% of the nominal

value.

2.5.2 Results and Discussion

2.5.2.1 Mass spectrometric detection

Initially, single ion recording (SIR) by monitoring the molecular [M+H]+ ion for each

analyte was used for detection. However, the use of SIR resulted in a relatively

unstable baseline possibly due to interference from sample matrix components.

Page 78 The application of multiple reaction monitoring mode (MRM) provided a much cleaner baseline compared to the SIR mode, by utilizing the increased selectivity of

MS/MS detection. The molecular-to-daughter ion transition mass spectra of each analyte are shown in Figure 2.4. The daughter ions formed by fragmentation and formation of a resonance-stabilized dimethylamine ion (for FX) and a methylamine ion (for NFX) were the predominant ions. Based on this fragmentation, mass transitions of m/z 310 to 44 for the FX isomers, m/z 296 to 30 for the NFX isomers and m/z 315 to 44 for the internal standard ([ thjJ-FX isomers) were selected.

Page 79 JUN16012 63 (2.230) Daughters of 310ES+ 1.10e8 100 44

Fluoxetine

% / H2C=N+: \

310 04 1 1 M M 1 1 1 1 1 1 1 1 1 1 1 1 1 1 i i i i i i JUN16013 65 (2.300) Daughters of 296ES+ 100. 30 3.03e7

1 1 1 1 1 1 1 1 1 1 1 JUN16011 47 (1.670) " " ' ' ' " " ' " " " " ' " ba'Ug^tere of 315ES+ 44 P, ^spy

100-, i Internal standard (D6-fluoxetine) O.OOB/

,CH3 ,CH,

H2C=N^

315 1 1 1 1 m/z 1 1 ' so ' '166'"'" 'iio''' ' 260' 1 300 350 250

;ure 2.4 Positive ion electrospray daughter (product) mass spectra of fluoxetine (m/z 310 to 44), norfluoxetine (m/z 296 to 30) and [ HsJ-fluoxetine (internal standard, m/z 315 to 44). Corresponding (proposed) fragmentation ions are shown on each mass spectrum.

Page 80 The MRM transition for FX was more efficient than that of NFX, resulting in higher

sensitivity for FX. Mass spectrometric parameters such as collision cell pressure,

collision energy, ion source temperature, cone voltage, and ion source temperature

were optimized for maximal sensitivity. The dwell time for each transition was 300

msec, with an inter-channel delay of 20 msec to provide optimal sampling of each

peak (10-15 scans/peak). Collision-induced dissociation (CID) was achieved with

argon at a pressure of 2 x 10"4 m bar in the collision cell. The collision energy, cone

voltage and ion source temperature were optimized at 30 eV, 15 eV and 100°C,

respectively.

2.5.2.2. Chiral separation of fluoxetine and norfluoxetine isomers

Separation of FX and NFX isomers was achieved using an AGP-CHIRAL column

that contains an a-glycoprotein-based stationary phase. Figure 2.5 shows

2 representative chromatograms of FX and NFX isomers with [ H5]-FX as an internal

standard. This separation is based on differential interaction of the FX and NFX

isomers with the chiral stationary phase. Various organic modifiers (methanol,

acetonitrile and isopropanol), ionic strength and pH were evaluated for maximal

separation (as measured as resolution). The use of lower concentrations (0 to 2.5%)

organic modifiers such as methanol, acetonitrile and isopropanol gave better

separation of the isomers; however, this resulted in considerable band broadening.

Thus, a mobile phase of 5% acetonitrile in 2 mM ammonium acetate buffer at pH 4.5

was selected for acceptable baseline separation and run time of less than 10 min. The

identities of the FX and NFX isomers were determined by injecting optically pure

Page 81 isomers either obtained from Eli Lilly & Co. or prepared in our laboratory by

fractional re-crystallization as described earlier.

MRU of 3 Channels ES+ 315.00 > 44.001

(S^Hj-Fluoxet'ne •» *• (K>- H-Fluoxeone

MRM'of 3 Channels ES+ MRM of 3 Channels ES* 310.00 > 44.001 310.00 > 44.00| Fluoxetine

4- (R)-Ftuoxetine

' MRMof 3 Channels ES+ 296.00 > 30.

o.bb "'" 2,66 "'" 4.00 "'" e.6oa.bb'"'' ioloo ''' i2!w '1' '2.bd " ''' 4.66 '' " 'e.bo

Figure 2.5 Representative LC/MS/MS ion chromatograms of [ H5]-fluoxetine (the internal standard), fluoxetine and norfluoxetine isomers (Left panel: QC MED [40 ng/mL], Right panel: corresponding blank human plasma chromatograms at each ion transition). The retention order (left to right) is (S)- and (R)-isomer for all of the analytes.

Page 82 2.5.2.3 Extraction, recovery and stability

The mean recoveries of SFX, RFX, SNFX and RNFX in human plasma were 92.4 ±

7.3%, 90.8 ± 8.2%, 82.7 ± 6.8% and 84.1 ± 7.2%, respectively. Similar recoveries

were observed in ovine plasma samples.

A number of studies were conducted to evaluate the stability of the analytes in

biological fluids and in processed samples for injection into the HPLC under

conditions similar to routine laboratory sample handling. The mean analyte

concentrations measured in samples subjected to 3 freeze-thaw cycle stability, 4 h

bench-top stability and 48 h autosampler stability were found to be within ±15% for

the LOQ of the nominal concentrations with an acceptable relative standard deviation

of <15% (Shah era/., 1992).

2.5.2.4 Method validation

Intra- and inter-batch variability and accuracy were calculated for the method in

human and ovine plasma samples. The summarized results for the human plasma

validation are presented in Tables 2.3 and 2.4. Similar results were obtained for the

ovine plasma samples but are not presented here. The intra-and inter-batch variability

(RSD) and accuracy (RE) were < 20% for the QC LOQ and < 15% for all other QC

concentrations.

The standard curves for all the analytes show linearity over the concentration range

tested (0.1 - 500 ng/mL for FX isomers and 0.5 - 500 ng/mL for NFX isomers)

[Figure 2.6]. Weighted linear regression (weighting factor = 1/y2) was used to reduce

bias at the lower concentration range. The LOQ of 0.1 ng/mL and 0.5 ng/mL were

Page 83 established for FX and NFX isomers, respectively, based on the signal-to-noise ratio

of > 15 with acceptable levels of variability in accuracy of less than 20%. These

LOQs are a 10- and 2-fold improvement, respectively, over our previous GC/MS/EI

method.

14.0 -.

600 concentration (ng/mL)

Figure 2.6 Representative calibration standard curves for fluoxetine and norfluoxetine isomers in human plasma using the LC/MS/MS method. RFX: y = -0.0051 + 0.0249 x, r2 = 0.9980 SFX: y = -0.0083 + 0.0252 x, r2 = 0.9992 RNFX: y = -0.0046 + 0.0068 x, r2 = 0.9978 SNFX: y = -0.0046 + 0.0066 x, r2 = 0.9974

The cross validation of the LC/MS/MS assay and our previous GC/MS assay resulted

in good agreement between the two methods [Figure 2.7]. The measured

concentration for each compound (nominal concentration: 1.0, 4.0, 40 and 400 ng/mL

for each isomers) by both GC/MS and LC/MS/MS were highly correlated (r > 0.99)

and were not significantly different (p > 0.05, paired t-test). This indicates that both

methods can be used for stereoselective quantitation of FX and NFX in human and

Page 84 ovine plasma with a high degree of confidence. However, the current LC/MS/MS

method has the advantage of a lower LOQ (0.1 ng/mL vs. 1.0 ng/mL for FX isomers,

0.5 ng/mL vs. 1.0 ng/mL for NFX isomers). In addition, this method has a shorter

run time (10 min vs. 25 min) and no requirement for derivative formation.

450

y = 1.003x-0.1895 R2 = 0.9981 *

50 100 150 200 250 300 350 400 450 Concentration from GC/MS method

Figure 2.7 Cross-validation of plasma (S)-fluoxetine isomer in human plasma using LC/MS/MS and GC/MS method. Similar results are observed for the other isomers.

In summary, this assay provides a rapid, sensitive and selective analytical method for

stereoselective quantitation of FX and NFX without derivatization. It was applied to

stereoselective pharmacokinetic studies in the ovine maternal-fetal unit as well as in

clinical studies during the ante- and postpartum periods, the results of which are

presented in Chapters 3 and 4.

Page 85 Table 2.3 Intra-batch precision (RSD) and accuracy (RE) of fluoxetine and norfluoxetine isomers using the LC/MS/MS method in human plasma (n=6)

Analyte QC LOQ QC LOW QC MED QC HIGH (S)-fluoxetine Nominal cone, (ng/ml) 0.1 4.0 40 400 Measured cone, (ng/ml) 0.09 4.3 42.9 386.7 SD 0.01 0.25 2.1 10.3 RSD (%) 13.2 5.8 4.9 2.7 RE (%) -8.8 7.5 7.3 -3.3 (R)-fluoxetine Nominal cone, (ng/ml) 0.1 4.0 40 400 Measured cone, (ng/ml) 0.09 4.3 43.3 375.3 SD 0.01 0.31 4.2 21.1 RSD (%) 15.1 7.3 9.7 5.6 RE (%) -11.0 6.5 8.2 -6.2 (S)-norfluoxetine Nominal cone, (ng/ml) 0.5 4.0 40 400 Measured cone, (ng/ml) 0.53 3.9 41.8 395.7 SD 0.06 0.32 2.8 18.5 RSD (%) 11.5 8.3 6.6 4.7 RE (%) 6.4 -3.3 4.5 -1.1 (R)-norfluoxetine Nominal cone, (ng/ml) 0.5 4.0 40 400 Measured cone, (ng/ml) 0.52 3.9 42.1 391.8 SD 0.05 0.35 3.7 27.8 RSD (%) 10.2 9.1 8.8 7.1 RE (%) 4.2 -3.5 5.3 -2.1

Page 86 Table 2.4 Inter-batch precision (RSD) and accuracy (RE) of fluoxetine and norfluoxetine isomers using the LC/MS/MS method in human plasma (n=5)

Analyte QC LOQ QC LOW QC MED QC HIGH (S)-fluoxetine Nominal cone, (ng/ml) 0.1 4.0 40 400 Measured cone, (ng/ml) 0.11 4.1 42.2 394.5 SD 0.01 0.31 1.5 25.8 RSD (%) 13.4 7.7 3.7 6.5 RE (%) 8.7 2.7 5.4 -1.4 (R)-fluoxetine Nominal cone, (ng/ml) 0.1 4.0 40 400 Measured cone, (ng/ml) 0.11 4.2 44.1 388.7 SD 0.01 0.28 1.2 3.3 RSD (%) 13.5 6.7 2.7 0.8 RE (%) 8.8 5.3 10.3 -2.8 (S)-norfluoxetine Nominal cone, (ng/ml) 0.5 4.0 40 400 Measured cone, (ng/ml) 0.50 3.8 43.2 401.3 SD 0.05 0.23 4.1 20.4 RSD (%) 10.4 6.1 9.5 5.1 RE (%) 0.5 -4.1 8.0 0.3 (R) -norfluoxetine Nominal cone, (ng/ml) 0.5 4.0 40 400 Measured cone, (ng/ml) 0.53 3.9 41.4 398.2 SD 0.03 0.21 1.2 19.2 RSD (%) 5.0 5.4 3.0 4.8 RE (%) 6.2 -2.2 3.5 -0.5

Page 87 2.6 GC/MS/EI method for paroxetine determination in biological fluids

An analytical method for quantitative determination of PX in plasma and breast milk

was developed in connection with a clinical study of PX during pregnancy and the

postnatal period (see Chapter 4). This GC/MS/EI method provides an improved

sensitivity with a smaller volume requirement (0.5 ng/mL using 200 pL of sample)

over previously reported methods (1-5 ng/mL using 1-3 mL of sample) (Petersen et

al, 1978; Brett etal, 1987; Gupta 1994; Hartter etal, 1994).

2.6.1 Methods

2.6.1.1 Standard stock solution preparation

The stock solutions of PX (100 pg/mL equivalent to free base) and the internal

standard maprotiline (100 pg/mL equivalent to free base) were separately prepared in

methanol by dissolving appropriate amounts of the analyte, and stored at -20°C for

upto 3 months. The working/spiking solutions of PX (100 and 1.0 ng/mL equivalent

to free base of each isomer) were prepared by serial dilution with de-ionized water.

The internal standard stock solution was diluted to a working/spiking solution of 100

ng/mL equivalent to free base. The PX spiking solutions were stored at 4°C for up to

4 days and the stability of the aqueous stock and spiking solutions were evaluated.

Maprotiline spiking solution stability was also assessed following storage at 4° C for

up to 1 month.

Page 88 2.6.1.2 Sample extraction

The analytes were extracted from human and ovine serum and human breast milk

samples using single or double liquid-liquid extraction procedures. For human and

ovine serum, a single liquid-liquid extraction procedure was used. For human breast

milk, a double liquid-liquid extraction method was necessary to remove

chromatographic interferences, due to the high lipid content.

For a single liquid-liquid extraction, the serum samples (25 to 200 pL) or spiked

standard and QC samples were pipetted into clean borosilicate tubes with PTFE lined

caps. The sample volume was adjusted to 1.0 mL with de-ionized water and 50 uL of

the internal standard solution was added to each sample except for the blank sample.

The samples were alkalinized by adding 500 pL of saturated sodium carbonate

solution. Extracting solvent (0.05 M TEA in hexane: isopropanol 98:2 v/v, 7 mL)

was added and the tubes capped. Samples were vortex-mixed briefly and extracted

with a slow rotary motion on a rotary shaker for 20 min and then cooled to -20°C for

10 min (to break any emulsion formed during mixing). Following centrifugation at

3000 g for 10 min, the top organic layer was separated and transferred to a set of

clean tubes. The organic layer was then evaporated to dryness under a gentle stream

(5 p.s.i.) of nitrogen at 35°C using a Zymak Turbo Vap® LV Evaporator (Zymak Co.,

Hopkinton, MA). The dried residue was reconstituted in 100 pL of toluene and 10

pL of heptafluorobutyric acid anhydride (HFBA) and derivative formation was

allowed to proceed for 30 min at 60°C. After derivatization, the samples were mixed

with 2 mL of 2.5% aqueous NH3 solution containing 0.05 M TEA to remove excess

Page 89 derivatizing reagent. Following centrifugation for 2 min, the toluene layer was separated and transferred into GC autosampler vials with 350 uL disposable glass inserts and 2 uL was injected into the GC.

For a double liquid-liquid extraction, the breast milk (100-500 uL) or spiked standard and QC samples were pipetted into clean borosilicate tubes with PTFE lined caps.

The sample volume was adjusted to 1.0 mL with de-ionized water and 50 uL of the internal standard solution was added to each sample except for the blank sample. The sample was alkalinized by adding 500 uL of saturated sodium carbonate solution.

Extracting solvent (0.05 M TEA in hexane: isopropanol 98:2 v/v, 7 mL) was added to each tube and capped. Samples were vortex-mixed briefly and extracted with a slow rotary motion on rotary shaker for 20 min and then cooled to -20°C for 10 min.

Following centrifugation at 3000 g for 10 min, the top organic layer was separated and transferred to a set of clean tubes. Diluted hydrochloric acid (2 mL of 0.2 M HC1 solution) was added and mixed for 20 min with a slow rotary motion. Following centrifugation for 5 min, the organic layer was aspirated and discarded, and 500 uL of saturated sodium carbonated solution was added to the remaining layer. The extraction solvent (7 mL) was added again and the samples mixed for 20 min.

Following centrifugation at 3000 g for 10 min, the top organic layer was separated and transferred to a set of clean tubes. The organic layer was then evaporated to dryness under a gentle stream (5 p.s.i.) of nitrogen at 35°C using a Zymak Turbo

Vap® LV Evaporator (Zymak Co., Hopkinton, MA). The dried residue was reconstituted in 100 uL of toluene and 10 uL of heptafluorobutyric acid anhydride

Page 90 and derivatized for 30 min at 60°C. After derivatization, the samples were vortex-

mixed with 2 mL of 2.5% aqueous NH3 solution containing 0.05 M TEA to remove

excess derivatizing reagent. Following centrifugation for 2 min, the toluene layer was

separated and transferred into GC autosampler vials with 350 uL disposable glass

insert and 2 pL was injected into the GC/MS.

2.6.1.3 Gas chromatography/ mass spectrometry

The GC/MS system consisted of a HP5890 Series II GC, HP5971A Mass Selective

Detector, HP7673A autosampler and HP Vectra 486/25 (Hewlett Packard, Palo Alto,

CA) with a DB5MS column (20 m x 0.18 mm ID, 0.18 pm film thickness) (J&W

Scientific, Folsom, CA). The operating temperatures for the injector and transfer line

were 260°C and 300°C, respectively. Helium was used as the carrier gas (0.3

mL/min) with a column head pressure of 15 p.s.i. The oven temperature

programming was as follows: initial temperature of 145°C for 0.5 min, 40°C/min to

280°C and hold for 2 min., 40°C/min to 300°C and hold for 2.5 min. The analyte (2

pl) was injected in splitless fast injection mode. The analytes were detected using

electron impact (EI) ionization with selective ion monitoring (SIM). Ions monitored

were m/z 525 for PX and m/z 445 for the internal standard. The area counts were

integrated from the extracted ion chromatogram using a Chemstation integrator

(Hewlett Packard, Palo Alto, CA). Quantitation was based on the area ratios of

PX/internal standard versus calibration curve sample concentrations.

Page 91 2.6.1.4 Calibration curve and regression model

Standard curves were prepared daily by adding an appropriate volume of the prepared

standard stock solution to 200 pL of blank human or ovine serum. (0.5, 1.0, 2.5, 5.0,

10, 25, 50, 100, 250 and 500 ng/mL). The internal standard (50 ng of maprotiline)

was then added to each sample. These samples were extracted and analyzed using the

extraction procedure described above. Weighted linear regression (weighting factor =

1/y2) was used between the peak area ratio of the analyte and the corresponding

internal standard versus the nominal concentrations to reduce bias in the lower range.

Linearity of the standard curves was determined by calculating the relative error (RE)

and relative standard deviation (RSD) at each nominal concentration. A RE and RSD

< 15% (<20% for the limit of quantitation), at each concentration, were used as the

acceptance criteria for calibration curve linearity (Shah et al, 1992).

2.6.1.5 Extraction recovery

Recovery samples were prepared at a low (1.5 ng/mL), medium (15 ng/mL) and high

(400 ng/mL) concentration to determine the absolute recovery of PX in the liquid-

liquid extraction method. Two sets of samples (Recovery group and Control group)

were prepared. Recovery group samples were prepared in triplicate by adding 200 pL

of QC sample and the internal standard. Control group samples were prepared in a

similar manner except 200 uL of blank plasma was added instead. All the samples

were then extracted according to the procedure described previously. After drying

the organic layer under a gentle stream of nitrogen, an appropriate amount of the

analyte in methanol was added to the Control group and further dried under a

Page 92 nitrogen stream. All the samples were then reconstituted, as described above, and

analyzed. The absolute recovery was calculated as the ratio of measured

concentrations of the Recovery versus the Control group at each concentration level.

A similar recovery study was conducted for the extraction of PX from breast milk,

except that the double extraction procedure previously described was used.

2.6.1.6 Method volidation

Intra- and inter-batch precision and accuracy were determined in the quantitation of

the QC samples. QC samples at LOQ (0.5 ng/mL), LOW (1.5 ng/mL), MED (15

ng/mL) and HIGH (400 ng/mL) levels were prepared by adding an appropriate

amount of PX to blank human plasma. These QC samples were divided into smaller

aliquots (~1 mL) and frozen at -20°C until analysis. At each batch analysis, triplicate

QC extractions were prepared by pipetting 200 pL of QC sample into each tube and

adding the internal standard. For the determination of intra-batch variability, 6

extractions were performed at each QC concentration. These QC extractions were

processed as described earlier. Intra-batch precision and accuracy were determined

by calculating the RSD and RE of the measured concentrations of the 6 QC samples

at each of the 4 concentrations. Similarly, inter-batch precision and accuracy were

determined by calculating the RSD and RE of the measured concentrations of 5

separate QC batches.

2.6.1.7 Analyte stability

Freeze-Thaw Stability: Prepared QC samples were frozen (-20°C) and thawed

(ambient) for an additional three cycles and then analyzed. These concentrations

Page 93 were compared with those of QC samples which did not undergo this additional

treatment.

Autosampler Stability: Extracted samples on the autosampler tray (ambient) were

injected twice over a 24-hour period. The peak area counts and the area ratios were

evaluated.

Bench-top stability: Prepared QC samples were left on the bench top for 4 h at

ambient temperature and then extracted and analyzed. These concentrations were

compared with those of QC samples which did not undergo this additional treatment.

In all stability tests, the concentration of analyte in the sample was compared between

treated and untreated (nominal) QC samples. The analyte was considered stable if the

concentration measured after the given treatment was within 15% of the nominal

value.

2.6.2 Results and discussion

The development of a sensitive and rapid analytical method capable of analyzing low

concentrations (e.g., 0.1 ng/mL or lower) was an essential prerequisite for the clinical

study and future pharmacokinetic studies in pregnant ewes, due to limited blood

sample volumes available from the human infants and ovine fetus.

2.6.2.1 Mass spectrometric detection and chromatograms

The selective ion monitoring (SIM) mode of mass spectrometric detection was used

for the quantitative determination of PX in human and ovine serum as well as in

Page 94 human breast milk. Mass spectra of the HFBA derivatives of PX and the internal standard, maprotiline, are shown in Figure 2.8. Based on the fragmentation pattern obtained, the molecular ion (M+) m/z 525 for PX and fragment ion m/z 445 (loss of bicyclic ring and formation of double bond) for the internal standard were chosen for selective ion monitoring in the EI mode. The base peak ion (m/z 138) was initially used for PX quantitation (Kim et al. 1997). However, the analytical method was subsequently modified and re-validated using the molecular ion m/z 525, since the signal-to noise ratio obtained using this higher mass ion m/z 525 (S/N ratio > 25 at the

LOQ) was greater than that for m/z 138 (S/N ratio >10 at the LOQ). This alteration resulted in improved analysis performance.

Representative extracted ion chromatograms of human plasma spiked with PX and the internal standard and a human blank plasma sample are shown in Figure 2.9.

Extraction and subsequent chromatographic analysis of blank plasma demonstrated that no endogenous peaks co-eluted at the retention time of PX or the internal standard.

Page 95 Abundance Average of I .1 14 to 8.225 mln.:PAXSCAN2.D (+,*)

2 0 0 0 0 0

2 0 7 2 ,3 1 1 3 3 9 3 6 7 j 4 2 4 6 4 7 9

M/Z -> 100 150 200 250 300 350 400 450 5 0 0 550

Abundance A v e ra ge of 7.198 to 7.246 min.:PAXSCAN2.D (+,*)

1 91 1 4 4 5 2 4 0 0 0 0 0

2 0 0 0 0 0 0

16 0 0 0 0 0

12 0 0 0 0 0

8 0 0 0 0 0

4 0 0 0 0 0 1 6j Si, 2 4 0 0 "1 Q^'ii'lVy'v' , ,1ii n ~ i 4 7 4 5 In 5 ' 67 3 , 6 83 9 3 4 2 3 1 M/Z -> 2^ '33OV' 1'5 0' ' 4 0 0 4 5 0 5T0

Fragment ion of maprotiline derivative

Fragment ion of paroxetine derivative (molecular ion)

Figure 2.8 Mass spectra and proposed fragment ions of the HFBA-derivatives of paroxetine (Top panel) and the internal standard maprotiline (Bottom panel) in the electron impact ionization mode

Page 96 Paroxetine (100 ng/mL)

7000 Maprotiline (IS)

4000

Blank r • i' • i ' 7.80_ 5.40 6.80

Figure 2.9 Representative chromatograms of paroxetine and the internal standard in GC/MS/EI mode (blank human plasma (without internal standard) and paroxetine (100 ng/mL) with internal standard).

Page 97 2.2 Extraction, recovery and stability

Similar to the fluoxetine method development, several organic solvents such as ethyl

acetate, hexane, toluene and dichloromethane were evaluated with or without

modifier (2% isopropanol) for maximal recovery of the analytes. Triethylamine (0.05

M) was added to all solvent systems used in the evaluation to reduce non-specific

binding of the analytes to glassware. Of the solvents tested, a hexane: isopropanol

mixture containing 0.05 M TEA and dichloromethane with 0.05 M TEA provided the

best extraction efficiency and cleanest chromatograms. However, a hexane:

isopropanol mixture was chosen for safety concerns associated with the use of

dichloromethane. No chromatographic interference from endogenous components in

plasma or breast milk was observed, using the SIM of m/z 525 and 445 for PX and the

internal standard, respectively. A double liquid-liquid extraction method was

required for breast milk samples due to the relatively high content of lipid. Following

a single liquid-liquid extraction, a small amount of lipid was observed at the bottom

of tubes after nitrogen drying. Initially, this residue from breast milk was derivatized

and processed in the same manner as the serum samples. However, the sensitivity

declined rapidly after 10-15 injections of breast milk samples and the accumulation of

non-volatile lipid in the GC glass injector port liner was observed. These problems

were resolved following the implementation of the double liquid-liquid extraction

procedure.

Absolute extraction recovery of PX was determined by comparing the back-

calculated PX levels of standards injected directly to that of similar levels spiked in

Page 98 plasma and extracted. The recovery was determined at three different concentration levels (1.5, 15 and 400 ng/ml) in human plasma, and calculated to be 98.6 ± 6.1, 91.7

± 3.5 and 97.9 ± 2.5%, respectively, with an overall mean recovery of 96.1 ± 3.8%.

Similar recoveries were obtained for the extraction of ovine plasma samples but are not presented here. The mean recovery of PX from human breast milk averaged 79.6

± 5.6%) following the double liquid-liquid extraction.

A number of studies were conducted to evaluate the stability of analytes in biological and extracted samples under conditions similar to the routine sample handling in the laboratory. The mean PX concentration measured in QC samples subjected to 3 freeze-thaw cycles, 4 h bench-top stability and 24 h autosampler stability were found to be within ± 10% of the nominal concentrations with an acceptable standard deviation of <10%.

The aqueous spiking solution of PX was found to be stable at 4°C for up to 4 days with no significant changes in the analyte concentration during this period (<5%).

However, there was a significant decrease in PX concentration (>20%) in spiking solutions stored for 1 month at 4°C. Therefore, the aqueous spiking solution was prepared freshly from the methanol stock solution and discarded after 4 days.

Methanol PX stock solutions on the other hand were found to be stable for 3 months when stored at -20°C. For the internal standard maprotiline, both methanol stock and aqueous spiking solution were stable for more than 3 months at -20°C and 4°C.

Page 99 2.6.2.3 Method validation

The assay method was linear over the range of 0.5-500 ng/ml using 200 uL of human

plasma and breast milk [Figure 2.10]. A weighting factor of 1/y2 was used for linear

regression analysis. The coefficient of determination (r2) of the slope of the

calibration was consistently > 0.99, indicating good linearity. The intra- and inter-

batch precision and accuracy of the method at four different QC sample

concentrations in human plasma are shown in Tables 2.5 and 2.6. The intra- and

inter-day precision (RSD) was better than or equal to 11.8 for the entire range. The

intra- and inter-day accuracy (RE) ranged from -5.8% to +10.1% over the calibration

curve range. Similar results were observed in ovine serum and human breast milk

samples. Overall, these data show low relative standard deviation (RSD) and relative

error (RE) for the assay method suggesting it is suitable for quantitative PX analysis.

600 concentration (ng/mL)

Figure 2.10 Representative calibration standard curve of paroxetine in human plasma using the GC/MS/EI method

Page 100 Table 2.5 Intra-batch variability (RSD) and accuracy (RE) of paroxetine using the GC/MS/EI method in human plasma (n=6)

Analyte QC LOQ QC LOW QC MED QC HIGH paroxetine Nominal cone, (ng/ml) 0.50 1.5 15 400 Measured cone, (ng/ml) 0.47 1.6 14.5 386.8 SD 0.02 0.1 0.8 5.5 RSD (%) 4.9 3.4 5.8 1.4 RE (%) -6.6 4.5 -3.3 -3.3

Table 2.6 Inter-batch variability (RSD) and accuracy (RE) of paroxetine using the GC/MS/EI method in human plasma (n=5)

Analyte QC LOQ QC LOW QC MED QC HIGH paroxetine Nominal cone, (ng/ml) 0.50 1.5 15 400 Measured cone, (ng/ml) 0.47 1.5 15.2 378.5 SD 0.03 0.1 1.8 21.5 RSD (%) 6.8 6.7 11.8 5.7 RE (%) -5.2 1.3 1.4 -5.4

In summary, a sensitive and selective GC/MS/EI assay method has been developed

and validated for the quantitative determination of PX in human plasma and breast

milk. The method involves liquid-liquid extraction of the plasma and breast milk

samples and GC/MS analysis with the selective ion monitoring of PX and the internal

standard maprotiline. The LOQ was sufficiently low to determine plasma PX levels

following administration of normal therapeutic doses (10 to 50 mg/day) with a

minimal volume requirement (25-200 pL for plasma and 100-500 pL for breast milk).

The GC/MS assay method was applied to the quantitation of trace levels of PX (low

ng/mL range) in samples from human clinical studies, the results of which are

presented in Chapter 4 of this thesis.

Page 101 2.7 GC/MS/NCI method for paroxetine determination in biological fluids

The GC/MS/EI method for PX (section 2.6) was used for clinical studies of PX

during pregnancy and the postnatal period (Chapter 4). The sensitivity achieved in

the GC/MS/EI method was sufficient to determine PX levels in maternal and cord

blood samples with volumes of ~0.5 to 1 mL serum. However, the sample volumes

from the PKU and some neonatal serum samples were much smaller (<50-100 uL

serum) and with lower drug concentrations compared to maternal and cord serum

samples. Therefore, the development of a more sensitive analytical method having a

smaller sample volume requirement was necessary to monitor PX in neonates.

2.7.1 Method

2.7.1.1 Standard stock solution preparation

The same standard stock solutions were used as described in section 2.6.1.1.

2.7.1.2 Sample extraction

The same sample preparation and extraction procedure was used as in section 2.6.1.2,

except a 100 uL biological fluid sample volume was used for the analysis instead of

the 200 uL needed for the GC/MS/EI method.

2.7.1.3 Gas chromatography/ mass spectrometry

The GC/MS system consisted of a HP5890 Series II GC, HP5989A Mass

Spectrometer Engine and a HP7673A autosampler with an Ultra-2 column (25 m x

0.25 mm ID, 0.25 pm film thickness) (Hewlett Packard, Palo Alto, CA). The

Page 102 operating temperatures for the injector and transfer line were 260°C and 300°C,

respectively. Helium was used as the carrier gas (0.3 mL/min) with a column head

pressure of 15 p.s.i. Methane gas (1.2xl0"3 mm Hg) was used for chemical

ionization. The oven temperature programming was as follows: initial temperature of

145°C for 0.5 min, 40°C/min to 280°C and hold for 2 min., 40°C/min to 300°C and

hold for 2.5 min. The analyte (3 pi) was injected in splitless fast injection mode. The

analytes were detected using negative ion chemical ionization (NCI) with selective

ion monitoring (SIM). Ions monitored were m/z 453 for PX and m/z 485 for the

internal standard maprotiline. The area counts were integrated from the extracted ion

chromatogram using ChemStation software (Hewlett Packard, Palo Alto, CA).

Quantitation was based on the area ratios of PX/internal standard versus nominal

calibration curve concentrations.

2.7.1.4 Calibration curve and regression model

Standard curves were prepared daily by adding an appropriate volume of the prepared

working/spiking stock solution to 100 uL of blank human plasma or breast milk. (0.1,

0.25, 0.5, 1.0, 2.5, 5.0, 10, 25, 50, 100, 250 and 500 ng/mL). The internal standard

(50 ng of maprotiline) was then added to each sample. These samples were extracted

and analyzed in the same manner as described for the GCMS/EI method. Weighted

linear regression (weighting factor = 1/y ) was used between the peak area ratio of the

analyte and the corresponding internal standard versus the nominal concentrations to

minimize bias in the lower sample concentration range. Linearity of standard curves

was determined by calculating the relative error (RE) and relative standard deviation

Page 103 (RSD) at each nominal concentration. RE and RSD values less than 15% (<20% for

the limit of quantitation) were used as acceptance criteria for calibration curve

linearity (Shah et al, 1992).

1.5 Method validation

Intra- and inter-batch precision and accuracy were determined in the quantitation of

QC samples. QC samples at LOQ (0.1 ng/mL), LOW (0.75 ng/mL), MED (15

ng/mL) and HIGH (400 ng/mL) concentration levels were prepared by adding

appropriate amounts of PX to blank human plasma. These QC samples were divided

into smaller aliquots (~ 1 mL) and frozen at -20°C until analysis. At each batch

analysis, triplicate QC extractions were prepared for each of the 4 QC concentrations

by pipetting 100 pL of QC sample into each tube and adding 50 pL of the internal

standard. For the determination of intra-batch variability parameters, 6 QC

extractions were performed at each concentration. These QC extractions were

processed as described earlier in Section 2.6.1.2. Intra-batch precision and accuracy

were determined by calculating the RSD and RE of measured concentrations of the 6

QC samples at each of the 4 concentration levels. Similarly, inter-batch precision and

accuracy were determined by calculating the RSD and RE of the measured

concentrations of 5 separate batches. The GC/MS/NCI method was cross-validated

with the previously developed GC/MS/EI technique.

Page 104 2.7.1.6 Recovery and analyte stability

Analyte recovery and stability data were not determined separately for this method

since sample preparation and extraction procedures were identical to those used for

the GC/MS/EI assay method.

2.7.2 Results and discussion

2.7.2.1 Mass spectrometric detection and chromatograms

As in the GC/MS/EI method, selective ion monitoring was used for the quantitative

determination of PX. Mass spectra of the HFBA derivatives of PX and the internal

standard maprotiline in the negative chemical ionization mode are shown in Figure

2.11. Based on highest abundance, the fragment ion m/z 485 for PX and m/z 453 for

the internal standard were chosen for selective ion monitoring in the NCI mode. The

use of negative chemical ionization resulted in 10-fold higher sensitivity (0.1 ng/mL

using 100 uL of sample) for PX compared to that obtained using electron impact

ionization (0.5 ng/mL using 200 pL of sample). Unlike electron impact ionization,

the molecular ion of PX was not observed using negative chemical ionization. The

loss of 40 mass units from the molecular ion resulted in the highly abundant m/z 485

ion with negative charge. One can surmise that one HF group comes from the HFBA

group (as seen for maprotiline) and the second HF is from the fluorinated aromatic

ring (F Abbott, personal communication). The proposed fragment ions for PX and

the internal standard are shown in Figure 2.11.

Page 105 Abundance 400000- Maprotiline (IS) 453 300000-

200000- 413

100000 H 393 H 39 160 178 / 93 239 295 320 \ -i 1 r .r..C —I— H | 1 r 300 400 100 200 Mass/Charge

Abundance 600000H 466485 Paroxetine 327 500000 4 400000 300000 137 258 200000 178 93 662 100000 369 564 \ \ —1—i—r J.,JL. 1—r- I 200 300 100 500 600 Mass/Charge

Figure 2.11 Mass spectra and proposed fragment ions of the HFBA-derivatives of maprotiline (top panel) and paroxetine (bottom panel) in the negative chemical ionization mode.

Page 106 Representative extracted ion chromatograms of plasma spiked with PX and the

internal standard and human blank plasma sample are shown in Figure 2.12.

Extraction and subsequent chromatographic analysis of blank plasma demonstrated

that no endogenous peaks co-eluted at the retention time of PX.

TIC of 3301033.(1 100X = 42878

120-j]

100

80

80

40

20^

0 I r- I l l l l I 8.0 9.0 10.0

TIC of 1301013.(1

too* = 54393

120^ (D 100 ^

z <- Internal Standard 80 _ •i- paroxetine 60^

40-

20 •• , j .A. 0- I. l l 1 1 1 1 8.0 1 1 1 -7 1 I 1 1 11.0 i i i i 10.0 9.0

;ure 2.12 Representative chromatograms of paroxetine and the internal standard in GC/MS/NCI mode (Top panel: extracted blank plasma sample with internal standard. Bottom panel: extracted QC MED)

Page 107 2.7.2.2 Method val idation

The assay method was found to be linear over the range of 0.1-500 ng/ml using 100

pL of human plasma and breast milk. The limit of quantitation was 0.1 ng/mL using

100 pL of samples, with the S/N ratio larger than 10. Use of a weighting factor of

1/y2 for linear regression analysis resulted in reduced bias at lower sample

concentrations. The coefficient of determination (r2) of the slope of the calibration

was consistently larger than 0.99 [Figure 2.13].

600 concentration (ng/mL)

Figure 2.13 Representative calibration standard curve of paroxetine using the GC/MS/NCI method

The intra- and inter-batch precision and accuracy of the method at the four different

QC sample concentrations in human plasma are shown in Tables 2.7 and 2.8. The

Page 108 intra- and inter-day precision (RSD) was better than or equal to 9.4% for the entire

QC range except at the LOQ. The intra- and inter-day accuracy (RE) ranged from

-3.6% to +12.8% over the entire calibration curve range. Again, these data show

overall low and acceptable relative standard deviation (RSD) and relative error (RE)

(Shah et al, 1992). Similar results were observed in breast milk samples, but are not

presented here.

Table 2.7 Intra-batch variability (RSD) and accuracy (RE) of paroxetine using the GC/MS/NCI method in human plasma (n=6)

Analyte QC LOQ QC LOW QC MED QC HIGH Paroxetine Nominal cone, (ng/ml) 0.10 0.75 15 400 Measured cone, (ng/ml) 0.11 0.80 14.8 385.4 SD 0.01 0.05 1.0 5.2 RSD (%) 9.0 6.7 6.8 1.3 RE (%) 12.8 6.9 -1.2 -3.6

Table 2.8 Inter-batch variability (RSD) and accuracy (RE) of paroxetine using the GC/MS/NCI method in human plasma (n=5)

Analyte QC LOQ QC LOW QC MED QC HIGH paroxetine Nominal cone, (ng/ml) 0.10 0.75 15 400 Measured cone, (ng/ml) 0.11 0.79 15.1 395.1 SD 0.02 0.07 1.4 10.5 RSD (%) 14.4 9.2 9.4 2.7 RE (%) 9.5 4.7 0.8 -1.2

Page 109 This GC/MS/NCI technique was cross-validated with our earlier GC/MS/EI method

by determining the variability in PX concentrations at 4 levels (1.0, 4.0, 40, 400

ng/mL) using both methods. There was very good agreement between results

obtained from two methods (r2 = 0.9981, p> 0.05, unpaired t-test) [Figure 2.14].

These results indicate that plasma and breast milk PX concentrations can be measured

with a high degree of confidence with either assay method.

450

400 - y= 1.003x- 0.1895

R2 = 0.9981 TO3 JZ 350 - IS E o 300 - z w 250 - o o E o 200 -

c o ra 150 - c o c"i 100 - o CJ 50 -

0 , T 1 1 1 1 1 1 1 1 1 0 50 100 150 200 250 300 350 400 450 Concentration from GC/MS/EI method

Figure 2.14 Cross-validation of paroxetine concentration in human plasma using the GC/MS/EI and GC/MS/NCI methods. Similar results were observed in human milk samples.

In summary, a sensitive and selective GC/MS assay method has been developed and

validated for the quantitative determination of PX in human plasma and breast milk.

The sensitivity of the assay was improved 10-fold compared to the previous

GC/MS/EI method (0.1 ng/mL using 100 pL sample compared to 0.5 ng/mL using

Page 110 200 uL). This improvement in sensitivity of the analytical method allowed us to determine plasma PX levels in the limited plasma sample volumes (less than 100 pL) available for newborns (e.g. at the time of PKU testing (~2 day postpartum)) and nursing infants whose mothers were taking normal daily doses of PX. This improved sensitivity would also be beneficial for small animal studies (e.g. fetal lambs and mice) where biological fluid volumes are also limited. The method involves liquid- liquid extraction of the plasma and breast milk samples and GC/MS/NCI analysis with the selective ion monitoring of PX and its internal standard.

In conclusion, several analytical methods were developed and validated to support the animal and clinical pharmacokinetic studies of FX/NFX and PX in various biological matrices. These analytical methods were applied to the pharmacokinetic and clinical studies presented in the following chapters.

Page 111 CHAPTER 3

STEREOSELECTIVE PHARMACOKINETICS OF FLUOXETINE AND NORFLUOXETINE IN NON-PREGNANT AND PREGNANT EWES

Pharmacokinetic studies in non-pregnant and pregnant ewes were conducted to

characterize stereoselectivity in fluoxetine and norfluoxetine disposition during

pregnancy. As described in the Chapter 1, there are no detailed stereospecific

pharmacokinetic data during pregnancy available in the literature at this time.

Initially, intravenous bolus studies of racemic FX (100 mg equivalent to free base)

were conducted in 6 non-pregnant ewes to determine the stereoselective

pharmacokinetics of FX and its pharmacologically active metabolite NFX. Based on

the pharmacokinetic data from the non-pregnant ewes as well as those from

preliminary experiments in pregnant ewes, paired maternal and fetal pharmacokinetic

studies were conducted in 5 pregnant ewes. The results obtained from these

pharmacokinetic studies are described in this Chapter.

3.1 Materials and supplies

Materials and suppliers used during the animal experiments were as follows: Veramix

sheep sponges (Tuco Products Co., Orangerville, ON); Pregnant mares' serum

gonadotropin (Ayerst Laboratories, Montreal, QC); Thiopental sodium injectable (1

g/vial), sodium chloride for injection USP (Abbott Laboratories, Montreal, QC);

injectable ampicillin (250 mg/vial) (Novopharm, Toronto, ON); injectable atropine

sulfate (0.6 mg/ml) (Glaxo Laboratories, Montreal, QC); Heparin (1000 units/mL)

(Organon Canada Ltd., West Hill, ON); (Ayerst Laboratories, Montreal,

QC); 2% (Astra Pharma Inc., Mississauga, ON). All injectable

Page 112 formulations were obtained from BC Children's and Women's Hospital Pharmacy,

Vancouver, BC.

Syringe needles and plastic disposable Luer-Lok® syringes for drug administration

and sample collection were purchased from Becton-Dickinson US or Canada

(Franklin-Lake, NJ or Mississagua, ON); heparinized blood gas syringes (Marquest

Medical Products Inc., Englewood, CO); green-top heparinized Vacutainer® tubes,

purple-top K2 EDTA Vacutainer® tubes, red-top Vacutainer® tubes without additives

(Vacutainer systems, Rutherford, NJ) were also purchased as required.

Polytetrafluoroethylene-lined screw caps (Canlab, Vancouver, BC) and 15 mL

Pyrex® disposable culture tubes (Corning glass works, Corning, NY) were used for

the sample storage.

3.2 Study methods

3.2.1 Animals and surgical preparation

All studies described in this document were approved by the University of British

Columbia Animal Care Committee. The procedures performed on the animals

conform to the guidelines of the Canadian Council on Animal Care.

3.2.1.1 Adult (non-pregnant) sheep

Animals were brought into the research facility located either at the British Columbia

Research Institute for Children's and Women's Health or the Animal Care Unit at the

University of British Columbia. The ewes were allowed to acclimatize in large pens

Page 113 with other sheep for a minimum of 3-4 days prior to surgery. Standard diets and free access to water were provided.

Six non-pregnant Dorset-Suffolk cross-bred ewes were surgically prepared using aseptic techniques following acclimatization. Food was withheld for approximately

14-18 hours prior to surgery. Atropine (0.6 mg) was administered intravenously via the jugular vein pre-operatively to control salivation during surgery. Anesthesia was induced by intravenous administration of 1 g of sodium pentothal via the jugular vein.

The ewe was immediately intubated and maintained on halothane (1-2%) and (70%>) in oxygen anesthesia. A 5% dextrose solution (500mL) containing 500 mg of ampicillin was administered over 60 min via the jugular vein. After shaving, cleansing and disinfecting the flank and groin areas with 10%) providone-iodine solution, silicone rubber (Dow Corning, Midland, NJ) or polyvinyl catheters were implanted in a maternal femoral artery and vein.

In all animals, the catheters were tunneled subcutaneously and exteriorized via a small incision on the flank of the ewe, where they were stored in a denim pouch when not in use. All catheters were flushed daily with approximately 3 ml of sterile 0.9% sodium chloride solution containing 12 units of heparin per ml to maintain their patency. Intramuscular injections of ampicillin 500 mg were given to the ewe on the day of surgery and for 3 days post-operatively. The sheep were allowed to recover for

3-8 days prior to experimentation. Following the recovery period, the ewes were moved to a monitoring pen adjacent to and in full view of the holding pen for experimentation purposes. In experiments requiring collection of urine, a Foley®

Page 114 bladder catheter was inserted via the urethra on the morning of the experimental day

and attached to a sterile polyvinyl bag for cumulative urine collection.

3.2.1.2 Pregnant sheep

In order to obtain time-dated pregnant sheep, the estrus cycle of the ewes was

synchronized by inhibition of spontaneous ovulation. This was accomplished by the

administration of medroxyprogesterone acetate for two weeks via an intra-vaginal

pessary (Veramix Sheep Sponge). At the end of the two-week period, ovulation was

induced by removal of the pessary and intramuscular injection of 500 I.U. pregnant

mare's serum gonadotropin. The ewes were then placed with a ram for 1-2 days to

result in time-dated pregnancies. Pregnancy was subsequently confirmed at -70-120

days gestation by ultrasound observation of the fetus and placenta. At approximately

115-120 days of gestation, animals were brought into the research facility located

either at the British Columbia Research Institute for Children's and Women's Health

or Animal Care Unit at the University of British Columbia. The ewes were allowed

to acclimatize in large pens with other sheep for a minimum of 3-4 days prior to

surgery. Standard diets and free access to water were provided.

Pregnant Dorset-Suffolk cross-bred ewes were surgically prepared using aseptic

techniques between 118 and 128 days of gestation (term -145 days). Food was

withheld for approximately 14-18 hours prior to surgery. Pre-operatively, 0.6 mg of

atropine was administered intravenously via the jugular vein to control salivation

during surgery. Anesthesia was induced by intravenous administration of 1 g of

sodium pentothal via the jugular vein. The ewe was immediately intubated and

Page 115 maintained on halothane (1-2%) and nitrous oxide (70%) in oxygen anesthesia. A 5% dextrose solution (500mL) containing 500 mg of ampicillin was administered over 60 min via the jugular vein. A midline abdominal incision was made after shaving, cleansing and disinfecting the abdomen, groin and flank areas with 10% providone- iodine solution, and covering the animal with sterile surgical drapes. The uterus was exposed and access to the head and hind quarters of the fetal lamb was gained via two separate uterine incisions made carefully in areas lacking major blood vessels and placental cotyledons. Polyvinyl or silicone rubber catheters (Dow Corning, Midland,

NJ) were implanted in both fetal femoral arteries and lateral tarsal veins, fetal trachea and the amniotic cavity. The amniotic fluid lost during surgery was replaced with warm sterile irrigation saline and the uterine and abdominal incisions were closed in layers. Catheters were also implanted in a maternal femoral artery and vein.

In all animals, the catheters were tunneled subcutaneously and exteriorized via a small incision on the flank of the ewe, where they were stored in a denim pouch when not in use. All catheters were flushed daily with approximately 3 ml of sterile 0.9% sodium chloride solution containing 12 units of heparin per ml to maintain their patency. Intramuscular injections of ampicillin 500 mg were given to the ewe on the day of surgery and for 3 days post-operatively. Ampicillin (500 mg) was also injected into the amniotic cavity immediately following surgery and daily thereafter for the duration of the preparation. Following surgery, animals were kept in holding pens with other sheep and were given free access to food and water. The sheep were allowed to recover for 3-8 days prior to experimentation. Following the recovery period, the ewes were moved to a monitoring pen adjacent to and in full view of the

Page 116 holding pen for experimentation purposes. In experiments requiring collection of

maternal urine, a Foley® bladder catheter was inserted via the urethra on the morning

of the experiment and attached to a sterile polyvinyl bag for cumulative urine

collection.

3.2.2 Pharmacokinetic Experimental protocols

3.2.2.1 Intravenous administration of racemic fluoxetine for characterization of stereoselective pharmacokinetics in non-pregnant ewes.

For the 6 non-pregnant ewes, racemic FX hydrochloride (100 mg equivalent to free

base) was dissolved in isotonic saline (10 mL) and given as an intravenous bolus via

the femoral vein catheter over 1 min. Arterial blood samples (~3 mL) were collected

from the femoral arterial catheter at the following time points: -5 min, 2.5 min, 5

min, 10 min, 15 min, 30 min, 45 min, 1 h, 1.5 h, 2 h, 3 h, 4 h, 6 h, 8 h, 10 h, 12 h, 18

h, 24 h, 36 h, 48 h, 60 h and 72 h. Collected blood samples were transferred to

Vacutainers® containing heparin and centrifuged at 3000 g for 10 min to

separate the plasma. The plasma was transferred to clean borosilicate glass tubes

with PFTE-lined caps and immediately frozen at -20°C until analysis. Additional

blood samples (-0.5 mL) were also collected at the selected time points for blood gas

analysis and measurement of glucose and lactate concentrations. Cumulative urine

samples were also collected via the Foley bladder catheter, inserted prior to the

administration of FX at the following time intervals: up to 0 h, 0-0.5 h, 0.5-1 h, 1-2 h,

2-3 h, 3-4 h, 4-6 h, 6-8 h, 8-10 h, 10-12 h, 12-18h, 18-24 h, 24-36 h, 36-48 h, 48-60 h

and 60-72 h.

Page 117 3.2.2.2 Paired intravenous administration of racemic fluoxetine for characterization of stereoselective maternal and fetal pharmacokinetics.

Experiments were conducted at 124-135 d gestation (term -145 d). Randomized,

paired maternal and fetal administrations of racemic FX, separated by a suitable

washout period, were performed in five pregnant ewes. For maternal drug

administration, racemic FX hydrochloride (50 mg equivalent to free base) was

dissolved in isotonic saline (10 mL) and given via the maternal femoral vein catheter

over 10 min. using an infusion pump (Harvard model 944, Harvard Apparatus, Millis,

MA). Similarly for fetal administration, racemic FX hydrochloride (10 mg equivalent

to free base) was dissolved in isotonic saline (5 mL) and given via the fetal lateral

tarsal vein over 10 min. The order of maternal and fetal administration of FX was

randomized. There were approximately 5-8 days between the two doses. Prior to the

administration of FX, approximately 50 mL of replacement whole blood free of FX or

its metabolite from donor fetuses and/or ewes was collected aseptically and stored at

4°C.

In both sets of experiments, simultaneous blood samples were collected from the

maternal (~5 mL) and fetal (~1.5 mL) femoral arterial catheters at the following time

points: - 5 min, 2.5 min, 5 min, 10 min, 15 min, 30 min, 45 min, 1 h, 1.5 h, 2 h, 3 h, 4

h, 6 h, 8 h, 10 h, 12 h, 18 h, 24 h, 36 h, 48 h, 60 h and 72 h. Collected blood samples

were transferred to Vacutainers,® containing lithium heparin and centrifuged at 3000

g for 10 min to separate plasma. Plasma was transferred to clean borosilicate glass

tubes with PFTE-lined caps and immediately frozen at -20°C until analysis.

Page 118 Additional blood samples (-0.5 mL) were also collected from the fetal femoral

arterial catheter at the selected time points for blood gas analysis and measurement of

glucose and lactate concentration. After taking 2 to 3 fetal blood samples, an equal

amount of replacement whole blood was given to the fetus (via the femoral vein

catheter) after warming to 37°C.

Serial amniotic (-2 mL) and tracheal (-2 mL) fluid samples were collected via

amniotic and tracheal catheters implanted during surgery at the following time points:

0 h, 0.5 h, 1 h, 2 h, 3 h, 4 h, 6 h, 8 h, 10 h, 12 h, 18h, 24 h, 36 h, 48 h, 60 h and 72 h.

An equal volume of 0.9% sterile saline was injected into each catheter following

sample collection.

Cumulative maternal urine samples were also collected via a Foley bladder catheter,

inserted prior to the administration of the drug dose, at the following time intervals:

up to 0 h, 0-1 h, 1-2 h, 2-3 h, 3-4 h, 4-6 h, 6-8 h, 8-10 h, 10-12 h, 12-18h, 18-24 h, 24-

36 h, 36-48 h, 48-60 h and 60-72 h.

'.3 Physiological monitoring

Blood pH, pG"2, pCC>2 and other blood gas status were measured using an IL 1306

pH/blood gas analyzer (Allied Instrumentation Laboratory, Milan, Italy). Blood

oxygen saturation (O2 saturation) and hemoglobin concentration were determined

using Hemoximeter (Radiometer, Copenhagen, Denmark). Due to a malfunction in

the Hemoximeter, blood O2 saturation could not be determined in some animals.

Therefore, it was not used for statistical analysis. Blood glucose and lactate

Page 119 concentrations were measured using a 2300 STAT plus glucose/lactate analyzer

(Y.S.I. Inc., Yellow Springs, OH).

3.2.4 Plasma protein binding studies

3.2.4.1 Preparation of reagent solutions and spiked plasma samples

Isotonic phosphate-buffered saline (PBS) (pH 7.4, 0.067 M) was prepared by

dissolving accurately weighed 1.8 g of mono-potassium phosphate (KH2PO4),

7.4 g of di-sodium phosphate (Na2HP04) and 4.2 g of sodium chloride in 1.0 L of

deionized water. The pH of the solution was adjusted to pH 7.4 by adding a small

amount of sodium hydroxide solution (1.0 M) drop-wise. The FX and NFX spiking

solutions used for the protein binding studies were prepared in PBS.

Blank maternal ovine plasma was prepared from blood obtained by venipucture of the

jugular veins of 4 drug-free ewes. The whole blood was collected in both green-top

Vacutainer® tubes containing lithium heparin and purple-top Vacutainer® tubes

containing potassium EDTA. These samples were centrifuged immediately at 3000 g

for 10 min. Heparinized plasma samples (from green-top tubes) were separated and

pooled. Similarly, EDTA-treated plasma samples (from purple-top tubes) were also

separated and pooled. In addition, blank fetal ovine plasma (heparinized) samples

were obtained by pooling pre-dose fetal arterial plasma samples free of study drugs.

Blank maternal ovine plasma samples (both heparinized and EDTA-treated) were

spiked with three levels of FX and/or NFX (100 [low], 1000 [medium] and 10000

[high] ng/mL each, respectively). Similarly, heparinized fetal ovine samples were

Page 120 spiked with the same levels of FX and/or NFX. These samples were incubated at

39°C and 39.5°C for 1 hour for maternal and fetal ovine plasma, respectively. The

stability of FX and/or NFX in plasma during the incubation period was evaluated

prior to the plasma protein binding study.

3.2.4.2 Ultrafiltration procedure

Determination of unbound plasma concentration (free fraction) of FX and NFX

isomer by ultrafiltration using Centrifree® micropartition devices (Amicon Inc.,

Canver, MA) was evaluated. An aliquot of 0.5 mL of ovine or human plasma spiked

with the FX and/or NFX isomers was placed into the device and centrifuged at 1000 g

for 30 min at a temperature of 4°C (in order to minimize lipolysis and the release of

free fatty acids).

3.2.4.3 Equilibrium dialysis procedure

For the in vitro protein binding study, batches of blank maternal and fetal ovine

plasma were spiked with appropriate amounts of racemic FX and/or NFX (100, 1000

and 10000 ng/mL) as previously mentioned. The total volume of spiking solution (in

PBS) was less than 5% of the total plasma volume. These spiked plasma samples

were incubated for 1 hour at 39°C and 39.5°C for maternal and fetal ovine plasma,

respectively. Furthermore, spiked plasma samples were incubated in equilibrium cell

for 1, 2, 3, 4, 6 and 12 hours to determined the stability of FX and NFX over the

potential equilibrium period.

Page 121 The cellulose dialysis membrane (12KDa molecular weight cutoff, Sigma Chemical

Co., St. Louis, MO) was boiled in distilled water for 30 min and then soaked in PBS

(pH 7.4) for a minimum of 1 hour. The membrane was subsequently mounted into the dialysis cells. Care was taken not to touch the surface of the membrane that would be in contact with the test fluids (PBS, plasma). Equal volumes (0.9 mL) of the physiological buffer and study fluids (plasma or control containing FX and/or NFX) were added to the respective sides of the dialysis cells, separated by the cellulose membrane.

To determine the optimal equilibration time, the cells were dialyzed in a shaking water bath at 15 rpm, for 1, 2, 3, 4, 6 and 12 hours at 39°C and 39.5°C for maternal ovine and fetal ovine plasma, respectively. Before and after equilibrium dialysis, both pH and sample volume were measured in order to determine any changes. The free fraction was calculated using the following equation:

Free fraction = Cu/Cp (Equation 3.1)

where Cu and Cp are drug/metabolite concentration in buffer and plasma after dialysis, respectively. Aliquots of buffer and plasma were then extracted and analyzed using the developed GC-MS assay (see Section 2.3).

Preliminary plasma protein binding of FX and NFX isomers in pooled animal studies was measured by pooling equal volume of samples with high FX and NFX concentration. Plasma binding of FX and NFX isomers in maternal and fetal sheep and pregnant and postpartum women was determined.

Page 122 Aliquots of the plasma and dialysate samples were extracted using single liquid-liquid

extraction and analyzed by the GC/MS method (see Section 2.3).

3.2.5 Preliminary in vitro metabolism studies of fluoxetine in ovine microsomes

3.2.5.1 Preparation of ovine hepatic microsomes

Ovine hepatic microsome were prepared in-house in collaboration with Drs. Sanjeev

Kumar and Harvey Wong. In brief, maternal and fetal ovine hepatic microsomes

were prepared from 2 pregnant ewes (135 and 139 days gestation) following

euthanasia with iv sodium (120 mg/kg). A single slice of the maternal

liver and the whole fetal liver were rapidly obtained from the animals and thoroughly

washed free of blood with ice-cold 0.05 M TRIS buffer (pH 7.4) containing 1.15%

KC1. Microsomes were prepared using a standard differential ultracentrifugation

technique at 4°C (Lu and Levin, 1972). Liver slices were minced and then

homogenized in ice-cold TRIS/KC1 buffer using a Potter-Elvehjem glass mortar and a

motor driven pestle. The homogenate was centrifuged twice at 10,000 g for 20 min.

The supernatant was then centrifuged at 100,000 g for 60 min. The resulting pellet

was re-suspended in 10 mM EDTA, TRIS/KC1 buffer (pH 7.4), and centrifuged again

at 100,000 g for 60 min. The microsomal pellet was then re-suspended in 0.25 M

sucrose and aliquots were stored in cryogenic vials at -70°C until use. Microsomal

protein concentrations were measured using the Bradford method (Bradford 1976;

Bradford Protein Assay Kit).

Page 123 3.2.5.2 Preliminary maternal and fetal fluoxetine N-demethylation in ovine hepatic

microsomes

For assessing FX demethylation by the microsomal preparations, 3 mg/ml

microsomal protein was incubated at 39°C in 100 mM Tris buffer (pH 7.4) containing

10 mM MgCl2 and 1 mM NADPH. After a 3 min incubation period, the reaction was

started by adding 100 ng/ml racemic FX, with the final incubation volume being 1

mL. Following 30 min, the reaction was terminated by the addition of 100 uL

acetonitrile and the incubation mixture was stored at -20°C until drug and metabolite

assay.

3.2.6 Analysis of biological samples

Fluoxetine and NFX isomers were measured using GC/MS/EI method developed in

our laboratory (see section 2.3). In addition, the presence of glucuronide and sulphate

conjugate of fluoxetine and norfluoxetine isomers was determined by enzymatic

hydrolysis. The hydrolysis procedure used in the present study was based on the

procedures described for and ritodrine conjugates in urine (Kim,

1995; Brashear et al, 1988). Urine samples were divided into 3 aliquots of 0.1 mL to

determine the concentration of the analytes in non-conjugate form [Set A],

glucuronide-conjugate form [Set B] and sulphate-conjugate form [Set C].

Page 124 Non-conjugate [Set A]: a. volume of 0.9 mL of de-ionized water was added to each aliquot.

Glucuronide-conjugate [Set BJ: volumes of 0.4 mL of 0.2 M sodium acetate buffer (pH 5.0) and 0.5 mL of Glucurase® ((3-glucuronidase solution from bovine liver -5000 Sigma units/mL, Sigma Chemical Co., St. Louis, MO) were added to each aliquot to provide a glucuronidase activity of 2500 U/mL.

Sulphate-conjugate [Set CJ: a volume of 0.9 mL of diluted sulphatase (Type VI from Aerobacter aerogenes partially purified enzyme 19 units/mL in 50% glycerol-0.01 M TRIS solution, pH 7.5, Sigma Chemical Co., St. Louis, MO) solution in 0.05 M TRIS buffer (pH 7.5) was added to each aliquot to give a final sulphatase activity of 1.0 unit/mL.

All the aliquots were incubated for 6 hours at 37°C in a shaking water bath.

Following incubation, the samples were cooled to room temperature and extracted as

described in section 2.3. Non-conjugated (free) FX and NFX isomers were measured

from the samples in Set A. The concentrations of glucuronide conjugates of FX and

NFX isomers were calculated by deducting the corresponding concentrations [Set A]

from the values from the samples in Set B. The concentrations of sulphate-conjugates

were calculated similarly from the samples in Set C.

. 7 Data Analysis

The systemic total body clearance (CLTB(FXO) of the FX isomers was separately

calculated as:

CLTB(FXI)= 1^ (12)

Page 125 where, AUC0"00 is the area under the curve under the FX concentration-time curve from time zero to infinity (or the area under the zero moment curve).

The mean residence time (MRTFX) of the FX isomers in plasma was separately calculated as

X wn„ AUMC°- MRTFX = „ for iv bolus administration (3.3) x Auc°r:x and

AUMC0-"0 MRTFX = oif— t/2 for iv infusion administration (3.4) A UCr -FX where, AUMC and T are the area under the first moment curve and the infusion duration, respectively. (Note: Hereafter all AUC0"00 and AUMC0"00 values are reported simply as AUC and AUMC for ease of presentation, unless otherwise specified.)

The mean residence time (MRT) for the NFX isomers was also separately calculated and adjusted as follows:

AUL-NFX

MRTNFXadjusted = MRTNFX- MRTFX (3.6)

Similarly, the mean residence times of FX and NFX isomers in the fetal circulation following maternal FX administration and in the maternal circulation following fetal

FX administration were adjusted as follows:

MRTfetal(adjusted) = MRTfetal - MRTmaternal(d0sed) (3- 7)

= MR Tmaternal(adjusted) MR Tmaternal ~ MR Tfetai(d0Sed) (3.8)

Page 126 where MRTfetai and MRTmaternai(dosed) are the MRT's in the fetal and maternal circulation following maternal drug administration, respectively, and

MRTmaternai(adjusted) and MRTfetai(dosed) are the MRT's in the maternal and fetal circulation following fetal drug administration, respectively.

Steady-state volume of distribution (Vdss) of the parent drug was calculated as:

(VdsJFX = (CLTB)FX * MRTFX (3.9)

The terminal elimination half-life (t'/2p) of the parent drug as well as that of the metabolite was obtained from a non-compartment model fitting of the data using the nonlinear least-squares regression. Renal clearance (CLR) values for the parent drug isomers, in pregnant and non-pregnant ewes, were calculated by dividing cumulative renal excretion (Xu0-00) by the area under the plasma concentration-time curve (AUC):

XU CLR (FX) = ™ (3.10)

Similarly, the renal clearance of the NFX isomers was calculated as follows:

CLR (NFX) = "'ol (3.11) NFX

Statistical analysis was performed on various pharmacokinetic parameters using the

F-test, Student's t-test (paired and unpaired), ANOVA and Pearson correlation tests.

The level of significance was chosen as p<0.05. The theory and formulae used for pharmacokinetic analysis were obtained from Gibaldi and Perrier (1982), respectively. Microsoft Excel 97 for Windows® with Analysis Tools® (Microsoft

Corp., Redmond, WA) and WinNonlin Professional Edition (Pharsight Corp., Palo

Alto, CA) were used for data processing and plasma concentration-time analysis.

Page 127 Sigma Plot version 4.01 (SPSS Inc., San Rafael, CA) and Graphpad Prisom 3.0

(Graphpad Software Inc., San Diego, CA) was used for graphic presentation and

statistical analysis. Data values in the text and tables are presented as the mean ±

standard deviation (SD), unless otherwise specified.

3.3 Results

3.3.1 Pharmacokinetics of fluoxetine and norfluoxetine isomers in non-pregnant ewes

following intravenous bolus fluoxetine administration

The average body weight in 6 adult non-pregnant ewes used in the study was 73.8 ±

11.6 kg. Serial plasma samples collected over a 72-hour period were used to describe

the pharmacokinetic profile of FX and NFX. Cumulative urine samples collected

over this same period were used to determine the pharmacokinetics of FX and NFX

and their glucuronide conjugates. These results are presented below. Arterial blood

gas status (pH, PO2, PCO2, O2 saturation) and blood glucose and lactate concentration

were monitored during control (i.e. prior to drug administration) and experimental

periods. There were no significant changes in any of these physiological parameters

from control values.

3.3.1.1 Plasma Pharmacokinetics

The mean plasma concentrations of FX and NFX isomers versus time profiles in the

six ewes are shown in Figure 3.1. Plasma concentrations of both FX isomers declined

rapidly tri-exponentially after intravenous administration of the dose. Detectable

Page 128 levels of serum FX and NFX isomers were observed beyond the time point plotted in

Figure 3.1; however, they are not plotted since serum concentrations were below the limit of quantitation in some animals. The plasma concentrations of the (SJ-fluoxetine

(SFX) were consistently higher than those of ^-fluoxetine (RFX) and there were statistically significant differences in plasma concentrations of the two isomers throughout the experimental period (p<0.05, two-way ANOVA for repeated measurements). Interestingly, this difference was observed even at the first sampling time (2.5-min) after drug administration. The average plasma concentrations of SFX and RFX were 359.3 ± 85.8 and 274.3 ± 70.3 ng/mL, respectively, at 2.5 min.

The pharmacokinetic parameters obtained for the FX and NFX isomers in plasma are listed in Table 3.1. The S/R plasma FX AUC ratio was 1.42 ± 0.22, thus resulting in a higher total body clearance (CLTB) of RFX compared to that of SFX (S/R CLTB ratio of 0.72 ± 0.10). Interestingly, the apparent terminal elimination half-life of RFX

(21.1 ± 6.0 h) is significantly longer than that of SFX (14.8 ± 5.0 h). Similarly, the

MRT of RFX (10.0 ± 3.7 h) was significantly longer than that of SFX (8.8 ± 3.2 h).

Thus, the isomer with a higher total body clearance (i.e. RFX with CLTB 1.97 ± 0.48

L/h/kg) has a longer half-life than the isomer (i.e. SFX) with a lower CLTB (1.44 ±

0.51 L/h/kg). However, this appears to be consistent with the differential volume of

distribution (Vdss) for the FX isomers (S/R ratio of 0.63 ± 0.05, p < 0.05).

For the NFX isomers, the plasma concentrations increased gradually over first 8 hours and then declined slowly (Figure 3.1) with a significantly longer apparent terminal elimination half-life compared to the parent drug [Table 3.1]. Unlike the FX isomers, there was no statistically significant difference between the AUC of RNFX

Page 129 and SNFX. However, the plasma concentration of RNFX remained higher for SNFX during the terminal phase, thus resulting in a significantly longer apparent terminal elimination half-life for RNFX compared to SNFX.

Page 130 * O * * * OJ CM •2 w CM ^"i To— o d O d d d d o +1 +1 +l +1 +1 +i +1 a> CC c5 CM CN CO CD s —t s CD m 55 J I- 00 CD I- •"~ d d d d d CD

T- "3 Q 00 m CN a> o CM CM cq co CD CM d 00 evi iri T— iri CO > +i +1 +1 +1 +i +1 +i +i +1 +i +1 c s CN CD 00 I- oo o> o CM CM '3 s CD CD 00 00 O I-: CO d CD 5 00 T CM OD O O Q s s d oo o s oo CM CO I- I- I- CD •sr CO co iri o +1 d +i +1 +1 +i a> +l CO +s i +1 +1 _c CM m CM +1 I- o in oo tu iri d od co 00 ns CO d CO 00 X T- T- CM od OJ r-. O) N" o co CO s m .<_ O oo I- CM X r-- CO 5 CM a> d d d CM "c CD •st- d 7— 7— CM a O 00 CO i— s o CD CO CM I- ,- CM X cn CD CM CD I-I d .<— CX) iri T— CO CM' "c O) 00 T—7 — 7— CM £2 CM O IO cc m a IO 00 o CO m >< m CM p CO CM T— c T— CM CM .— T— oo ^— r— o CD co m s s I- IO CD I- CO >< in CO s in I-: CD ai CO T— "c CM T— T— CM CM T— o ^— 2 T— cc s m oo I- ,- X 00 CM in 00 d co in r-: ai y- d CO 00 CM CM a> >— T— r— T CO ~

s CD uo 00 CM I- X •<*• CO •vf CM °? iri T— d y- CO T— s CM T— .— T— I- T— CM or CO s o I- CO O) X 00 CD o CO CD CD d CO y- CM iri d o CM T— 7— r- T— CO T_ CO co CO CD CM T- T— o X co CM r- Is-: x— cn CM r-: CN CM CD s in 7— T- I- CO cc CO CD o O oq CM X CO CO C0 a. CM iri CO csi ^— CM 00 o T— T— n CM T i— 7— CO ^— lo • X cq o ai d Is- CO oo d 00 d 00 CN CM ^— CO T— CM o 2 ^— CM CM cc CM |T3 C CS * 3- IS 5 00 .c 3 S a> E as in co. I— I— ca 1 " co CM O CC CC CC u. < < H o ft.

(~iw/6u) uoijejjusouoo 6rup eiuseid

s- 3 3.3.1.2. Urine Pharmacokinetics

The renal elimination pharmacokinetic parameters for FX, NFX and their conjugates

were measured in five animals. In one animal, total urine collection was interrupted

prematurely due to detachment of the Foley catheter during the early experimental

period, and thus was not included in the study. The urinary pharmacokinetic

parameters were calculated based on the cumulative (£) FX and NFX isomers in

ovine urine collected over the experimental period. Total urine samples were

collected over the 72-hour period with the volume of urine recorded at each interval.

Both FX and NFX isomers and their glucuronide conjugates were excreted in ovine

urine following the iv administration of racemic FX. However, the contribution of

renal excretion to overall drug elimination is minimal (total contribution of all free

FX, NFX and their glucuronide conjugates is 3.6%) as shown in Table 3.2.

Representative plots for urinary accumulation of FX, NFX and their glucuronide

conjugate are shown in Figures 3.2 and 3.3.

The cumulative amount versus time plots of the FX isomers reaches a plateau about

48 hours after iv administration (i.e. amount of FX isomers excreted in the urine was

minimal after 48 hours, Figure 3.2). This corresponds with the plasma concentration

vs. time profile of the FX isomers during this period (i.e. below the limit of

quantitation, Figure 3.1). In contrast, the concentrations of the NFX isomers had not

reached a plateau by the end of the experimental period (72 hours, Figure 3.2).

Additional urine collection may have provided a more accurate estimation of the

urinary excretion data, particularly for NFX. However, further urine collection was

not performed for both practical reasons as well as humane concerns regarding the

Page 133 prolonged use of the Foley catheter. Since the rate of NFX excretion in urine was relatively low in the 60-72 hour samples, the calculated renal excretion data likely provide adequate estimations for the NFX isomers.

Similar to the pharmacokinetic data from plasma, there was a significant difference in the renal clearance of SFX and RFX following iv bolus drug administration (Table

3.2), with the (Sj-isomer being greater. In all five animals, the cumulative amount of

SFX in urine was higher than that of RFX. The S/R ratio of the FX isomers in urine was 2.53 ± 0.24, which was much higher than their S/R AUC ratio in plasma (1.42 ±

0.22, Table 3.1). This results in a significantly higher renal clearance value for SFX

(4.1 ± 1.8 mL/h/kg) compared to RFX (2.2 ± 1.2 mL/h/kg) with the S/R CLR ratio of

1.91 ± 0.34. Compared to the FX isomers, higher levels of NFX isomers were excreted in ovine urine (approximately 5 fold). Similar to FX, the cumulative amount and renal clearance of SNFX were significantly higher compared to those of the

RNFX isomer, but to lesser degree (S/R ratio of 1.54 ± 0.17 and 1.49 ± 0.72, respectively). The amount of glucuronide conjugate of the FX and NFX isomers in urine was determined by the difference between FX and NFX concentrations in urine before and after the enzymatic cleavage (Section 3.2.6). Stereoselectivity in the renal excretion of the FX isomer glucuronide conjugates was not observed (S/R ratio of

1.05 ± 0.31, p>0.05) compared to that of the parent (non-conjugated) compounds

(S/R ratio of 2.53 ± 0.24). Unlike the free NFX isomers, a much smaller amount of the glucuronide conjugate of SNFX was excreted compared to that of RNFX, with the

S/R ratio averaging 0.42 ±0.19 (p< 0.05).

Page 134 * * Q •tf •tf CO UO g CO CM CO CO a, "co d i +1 d ci d o o c +1 +1 +1 +1 +1 or T— +1 CO CO in •tf CD CN co to cn r- •tf cu CM UO d o CD Q CO C3 T— CO CD CO UO 3 +l CD 3 •tf M CD d tf d CD 3 +1 tf o o d uo o c +1 +1 o co 2 +l +1 CO CO +1 +1 CM +1 +1 +1 o (I> CM ,_ r-- m O o o CO CO CO CN •tf CO d CN CD d (O cd CO r- CM a UO CD o Q •tf CO •tf o +i uo CD CM CO CO cd 3 CO CO CD •tf CN CD o CM O CM o CN d uo tf 3 -H 1X1 o o 4-1 d +1 +1 o d +1 "eb c +1 d o +1 CO d r- o CM 3 CO CO *~: d +1 o CD d +1 •tf o Csi co o CO "M cu uo CO CO CM r- •tf o o o 00 CD CJ >< CO •tf co d CM o o CO o X CD r-- a CO •tf co CO CD o •tf UO LL o CM CM CM tf C3 d CM d 2 o CN d co d oo o O UO q CO CN C3 LL CM CM CM T— 00 C CO O CO CO X UO 7— CN O cn CN uo •tf o CO ^~ CO LL r- 00 00 cc co d d r- UO d CC o •tf oo C3 >< CO CO o CO O CD CN LL r-: CO UO X CD •tf CN d CM co lo UO CO CM CM U. O CO CM CM o d 2 CO o o d CO — •tf CM CD T X CO CN o CO CM CD X d CM CD CO o o co •tf T— CO 00 u M O o CO CN fe T— CM LL CO T— T3 CO d 2 CO oo O LL CC CD d CC O X O UO CO CD CO CO CD LL -tf O CD CM o X CO U0 d •tf O CO T— CM CM -tf co a> d CO uo LL co Is d CM dr- UO CO o CD CN CD o CO O oN 00 co CO :LU tf CO X CO q CD CO CO CD o UO CD ^— CD CD UO CM CN o TC—D CO LL r- o CM ~ d d 2 CO d d sx LL CC to cu CM uo Si o co O CD CM LL CM d •tf CO CO CD d d CN •S •§ CO d 1^ co OJ UO O CO tf c> 00 vO CM LL CO o o •tf CN o CD CD CO X d CO o •tf CO 1- in — •tf CC CN o o w CT—O CN LL fT—- d o CO uo co o d d *~ d CC d V | •tf CO CO X co CO CO co CD ^ Cw •tf o X U0 o CD oo CM o ,— CN r- 00 o o co r-~ o CD CN LL uo o o d d CO d ^ c CO d CM X tf O uo U0 CM o CO u- tf CD uo "I o UO UO o X UO o 00 U0 tf CC CM o CM CN U- CD d CN o duo 2 o CM d d rv |1

f CU SI 5 .5 5 1 O C 5 ca 3 uro cu cu a ° in in CU o i. cu c s c o c in 3 c in w § o o 3 TJ 3 o o u O TO O) o "D E T3 o o o a m ro O) FX- W O E i * w CO 2 w CO _| W O 1200

0 20 40 60

Time (h)

Figure 3.2 Representative urinary accumulation of fluoxetine and norfluoxetine isomers in urine following iv bolus administration of racemic fluoxetine to an adult non-pregnant ewe (100 mg equivalent). Data are presented for ewe ID F3230.

400

Time (h)

Figure 3.3 Representative urinary accumulation of the glucuronide conjugates of fluoxetine and norfluoxetine following iv bolus administration of racemic fluoxetine to an adult non-pregnant ewe (100 mg equivalent). Data are presented for ewe ID F3230.

Page 136 3.3.2 Pharmacokinetics of fluoxetine and norfluoxetine isomers in pregnant ewes following maternal intravenous fluoxetine administration (10-min infusion).

The average maternal body weight was 69.8 ± 16.4 kg and estimated average fetal

body weight on the day of the maternal experiments was 2.85 ± 0.83 kg. Average

gestational age on the day of FX administration was 130.4 ± 2.6 days. Mean

gestational age and fetal weight at delivery were 141.4 ± 3.0 days and 3.73 ± 0.92 kg,

respectively.

3.3.2.1 Plasma pharmacokinetics

Figure 3.4 illustrates the maternal arterial plasma concentration versus time profiles

obtained for the FX and NFX isomers following maternal racemic FX administration.

Plasma concentrations of both FX isomers increased rapidly during the 10-min

infusion period and then declined tri-exponentially. The plasma concentration of

SFX was significantly higher than that of RFX throughout the experimental period

with the individual plasma S/R concentration ratios ranging from 1.42 to 2.24

(p<0.05, two-way ANOVA for repeated measures). Table 3.3 presents the calculated

maternal pharmacokinetic parameters for the FX and NFX isomers following

maternal drug administration. The mean S/R ratio of plasma AUC for FX was 1.65 ±

0.33, resulting in a higher total body clearance of RFX (3.1 ± 0.8 L/hr/kg) than SFX

(1.9 ± 0.2 L/hr/kg) (p<0.05, paired t-test). The steady-state volume of distribution of

RFX was -50% higher than that of the SFX. However, no significant

stereoselectivity was observed in either apparent terminal elimination half-life or

MRT for the FX isomers. Maternal plasma concentrations of the NFX isomers

Page 137 increased over the first 2 hours and then declined slowly (Figure 3.4). Unlike the parent compound, there was no significant difference in the AUC of SNFX and

RNFX with the S/R ratio averaging 0.94 ± 0.22. The maternal AUC ratios of the

NFX to FX isomer were 0.63 ± 0.23 and 1.07 ± 0.39 for (S)- and (R)-isomers, respectively. Unlike the FX isomers, the apparent maternal terminal half-life and

MRT values for RNFX were greater than those for SNFX in all experiments.

However, because of inter-animal variability, the differences were not significantly different.

The apparent terminal half-lives of the FX isomers were much shorter in pregnant ewes (6.7 ± 0.7 h for SFX and 6.6 ± 2.5 h for RFX) than those of non-pregnant ewes

(14.8 ± 5.0 h for SFX and 21.1 ± 6.0 h for RFX, Table 3.1). Similarly, the MRTs of the FX isomers were much shorter (3.4 ± 0.4 h for SFX and 3.2 ± 0.8 h for RFX) than those calculated in the non-pregnant ewes (8.8 ± 3.2.h for SFX and 10.0 ± 3.7 h for

RFX, Table 3.1) (p<0.05, t-test). Unlike the case in non-pregnant ewes (S/R ratio of

0.71 ± 0.14), there were no significant differences in the apparent terminal half-lives

(t>/2p) of the FX isomers (S/R ratio of 1.15 ± 0.49) in the pregnant ewes. The steady-

state volume of distribution (Vdss) was 6.3 ± 1.1 L/kg for SFX and 9.4 ± 1.3 L for

RFX (S/R ratio of 0.77 ± 0.28, p<0.05, paired t-test). Although similar

stereo specificity (i.e. a higher Vdss for RFX) was observed in the non-pregnant ewes

(Table 3.1), the values in pregnant ewes were significantly smaller (p<0.05, t-test).

Similar to non-pregnant ewes, there was no significant difference in the maternal

AUC of NFX isomers (131.7 ± 65.9 pg-h/L for SNFX and 135.2 ± 50.5 pg-h/L for

Page 138 RNFX). There was also no significant difference in the adjusted MRT of the NFX isomers (6.7 ± 2.8 h for SNFX and 9.4 ± 3.7 h for RNFX).

Similar to the maternal profile, fetal plasma concentrations of FX isomers also increased rapidly during the infusion period and then declined exponentially (Figure

3.5). The fetal plasma SFX levels remained consistently higher than the corresponding RFX concentrations throughout the experimental period. The calculated fetal pharmacokinetic parameters for FX and NFX are presented in Table

3.4. The fetal S/R FX AUC ratio averaged 1.73 ± 0.29, which was not significantly different from the corresponding maternal value. The fetal-to-maternal (F/M) AUC ratio was similar for the SFX (0.60 ±0.15) and RFX (0.58 ±0.19) isomers. Whereas the apparent fetal terminal elimination half-life of the FX isomers did not exhibit any significant stereoselectivity, the values measured were significantly longer than those calculated for the mother (9.3 ± 0.3 vs. 6.7 ± 0.7 for SFX and 10.8 ± 1.3 vs. 6.6 ± 2.5 h for RFX) (pO.OOl). The MRT ratios for the fetal FX isomers were also not significantly different from each other. The fetal AUC ratios of the NFX to FX isomers were calculated to be 0.71 ±0.18 and 0.98 ±0.15 for (S)- and (R)-isomers, respectively, which are similar to the maternal values. As observed in maternal plasma, there was no significant difference in the AUC values for SNFX and RNFX, i with the S/R ratio averaging 1.25 ± 0.29. The F/M AUC ratios were 0.53 ±0.13 for

SNFX and 0.76 ± 0.33 for RNFX, and were not significantly different. Although the mean F/M AUC ratio for NFX (0.65) was higher than the mean value for FX (0.59), this difference was also not statistically significant. The apparent fetal terminal elimination half-lives of the NFX isomers showed significant stereoselectivity (S/R

Page 139 ratio of 0.66 ± 0.05, pO.OOl); however, they were not significantly different from the maternal values. The fetal MRTs of NFX isomers were not significantly different from one another nor from the corresponding maternal estimates.

The apparent terminal half-lives of FX isomers in fetal lamb were significantly longer

(9.3 ± 0.3 h for SFX and 10.8 ± 1.3 h for RFX, Table 3.4) than those of maternal values (p< 0.05 using paired t-test for both isomers) (Table 3.3). There was no significant stereoselectivity in the fetal values (S/R ratio of 0.87 ± 0.09). The adjusted MRTs were was 3.7 ± 1.0 h for SFX and 4.2 ± 1.5 h for RFX, which are similar to the maternal values. The fetal-to-maternal (F/M) AUC was used to determine the extent of fetal drug exposure during pregnancy. F/M ratios of the FX isomers ranged from 0.45 to 0.91 [Table 3.4] and were not statistically different (0.60

±0.15 for SFX and 0.58 ±0.19 for RFX, p>0.05, paired t-test). Similar F/M AUC ratios were observed for the NFX isomers, again with no statistically significant difference between them (0.76 ± 0.33 for SNFX and 0.53 ±0.13 for RNFX, p>0.05, paired t-test).

Page 140

Q CO CO CT) Csl CM LO CO T— —- +1 dCO dCM d dCM d dCM dCM d c +l +1 +i +1 +i +1 +1 +l SP to s T— OJ cCOq CsO Is- IO lOs CMs E dI - dI - T— dC D dI - dI -

Q CO co CD CM LO co CM d CD CM r-: CO +1 d +1 +i +1 +1 +1 s +1 c CM +1 CO I- Is- •* CM CO to O CM d <° d CO CD CM ^ CO E

Q LO s CD CO CO co cn O I- LO CO 00 T— LO 00 +1 CO d d CM 00 c +i +1 +' +1 +1 +1 +1 CO +1 7— CM CO CM CD CD oi CM CT) 00 E CM CO CO CO iri co

CT> LO CT) O d CT) co T- O CO CO o> CO LO CM CT) CO T— X CT) O LO CM s CO CT) I -: CO CM CM £ - cri CM I 2 *- CD 03 LO CO T— T CM d - CM CD CO CD LO CM TT iri CM

CO 00 Is- CT> Is- CO d cc CM CO 00 CO d CM 0) £ 00 CM • diff e ^ >*

CcO I ? o CD .5 XT' .c 4C c o *si g O in O § Z3 -1 CC m d CC 3 < O < CJ) Q CO O) CL g CO o COM) ou o ou o +1 OCM d d d d d bo TO c +1 +i +1 +1 +i +1 Ice co CO CM uo CO CO CD r-; OJ oo CM 00 CO o co sr— d d d d

* < * CU Q uo CM O) « •tf < T— T— CO CO o < S— r-: O) 00 a o d T— cu +1 m CO CO d +1 d +1 u+o1 +1 +i X c CO +1 +i o o CM C+D1 CN CO co +1 O) tf If) tf CO d CD CO oCOi dC D CT—O CM E O) o CM co O) tf O a Q CO r-~ co UO CO d co CU Tu—o O) d O O +1 d d +1 eci c +1 +1 +1 +i CM +1 +1 0. +i oo uo 1-1 co o> csi CO C+H 00 CM 00 oo +1 d d tf d co CO CD uo oT_o o d E CD d fl CO 00 CD COD) CO 00 CM O fc tf oi tf eci d co uo uo '3t-i o•tof d T— d T— -t-» CO x~ CM '3 •tf uo oo CM tf i— 00 00 O CM a d ^—' d l->: T— O) UO tf T— CTM— CO If) CD 00 If) uo CD "> TJ tf; d •v 2 a CO CO 00 oo UO •tf O) OO d o CM CU CM 1 s ^ O) uo o CO CO 00 O) 00 CM •tf 00 ra c fc O d o UJ 60 00 uo UO d 00 OJ a. cc CO d CM d CN cn => tf C T— cn o 3 o OJ 00 CO CO uo CN o tf CD lO O° ^Q. . •tf fc d tf d d lo? CM d UO dC O CO O) co T— GO 00 in V < !-H o CU CD H—» CN oo CO o CN CO CO CO CO " O cu CD fc d tf d d q CO T— TIS— . CO i cc uo d UO CO d o PH CM o tf tf 00 co OO uo CO _o fc d d CM d d uo *H—» T— ^— uo d d dC O CU CO CO 00 CM 5 o C CN CN o 00 CO OJ CM CD o O) CM |8t o fc d T_ lo < UO CM dU O CO CM 00 OJ eci cc 0) ^ CO d o , VQ .^ « * aj ca a uo f~ CD m oo 00 uo CM OJ PH CO CM CM tf 00 o ni CO d d «4 CO CON 00 CD CM Ii I £ CN CO o CO CD UO CO o o ed -g CN cu cu fc uo CO d tf T— CM lO cq 51 cc uo d UO CO d || 73 TO ca ao >, >> c c CU i_ ra ra o n 3" 3£:" on *= "& ;C O) CJ) ;C_ CU O) s D> ' 10 CO .c o o o o Q> ra "M "ra o H *_ m v o 1- cs § CO. M R F/ M F/ M M R F/ M H a AL U F/ M a 3.3.2.2. Amniotic and fetal tracheal fluids

The amniotic fluid concentration of the FX and NFX isomers increased over the first

12-18 hours of the experiments and then decreased slowly (Figure 3.6). These

concentrations were similar to or slightly lower than their respective fetal plasma

concentrations and no significant accumulation of either the FX or NFX isomers was

observed. Unlike the fetal plasma concentration profiles (which show a higher

concentration of SFX over RFX), there were no significant differences between the

amniotic fluid concentrations of the FX or NFX isomers (p> 0.05, two-way ANOVA

for repeated measures). Fetal tracheal-fluid concentrations of the FX isomers were

slightly higher than those measured in amniotic fluid (Figure 3.7). Similar to the

amniotic fluid, there were no significant differences in the concentrations of the FX or

NFX isomers (p> 0.05, two-way ANOVA for repeated measures). Glucuronide or

sulfate conjugates of FX or NFX were not found in detectable amounts in either

amniotic or fetal tracheal fluid samples.

Page 145

3.3.2.3 Maternal Urine pharmacokinetics

The renal clearance, cumulative amount excreted and percent of administered dose

excreted as FX, NFX and their corresponding glucuronides in maternal urine are

provided in Table 3.5. While both parent drug and metabolite as well as their

conjugates were present in maternal urine, their total contribution to overall maternal

drug elimination was minimal (3.4%). This is similar to the values observed in the

non-pregnant ewes (Table 3.2). Representative cumulative urinary excretion plots of

the FX and NFX isomers and their corresponding glucuronide conjugates are shown

in Figures 3.8 and 3.9, respectively. The FX and NFX isomers did not reach a plateau

level by the end experimental period, resulting in some underestimation of their

cumulative excretion. This was also the case with the FX and NFX isomer

glucuronides. Neither the cumulative amount nor the renal clearance of the FX and

NFX isomers, and their glucuronide conjugates, exhibited any significant

stereoselectivity, unlike the non-pregnant ewes [Table 3.2]. Additional urine

collection may have provided a more accurate estimate of urinary pharmacokinetic

parameters. However, further urine collection was not performed due to ethical and

practical concerns regarding prolonged use of a Foley catheter in the ewes. Sulphate

conjugates could not be detected in quantifiable amounts in these urine samples.

Page 148 s •3" g I- co CM CM +1 cn CO d O d Ox co ci ci +l o +1 +1 T+—l +1 +i tw LO co Is- CM E LO co rr *~ cn co ci Q o LO to o CO cn OT T— o CO CO 00 co O O CO o CM CD o D) CM LO d o ci O OT o d d +c1 +1 +1 +1 +1 +1 X CM d +1 +1 +1 — +1 s to R 00 +l +i I - CO cso i r>- « OT co CO CD o LO OT LO CO 00 co co 0.0 0 ci CO o d CM 0) CM d ci CM o Q cn LO 00 LO s o CO OT CM 3 I - o CD LO o +1 LO d d o CD o o d +1 d d d +1 +i c +1 +1 s CO +l CO +i +l +1 I - OT CD Is- CO LO LO OT CO cri LO CM oo d d CM d d d T— m co CO X LO CM CO LO cn co co d •st LL d oo CO CO d s 0.0 0 CM I - d oo d d d d LO CO T— co oo o LO oo X OT 00 OO Is- LO o CM LL d CD CO d co d o d cn d co d 0.0 0 d T_ CC d d

s cq X I - CM r-- X 00 O o o co LO d 00 oo LO CM CO oo O O) d £ CM O d LO d CM d CO d

cq cn s LL x 00 CT—O 00 X CM CM I - o o N; CD d CD CM LO 00 CC CD CM LO LO o d CO d £ CM CM O d d d CM cc d

CO LO LO oo LO X o CO O CM X 00 CD OT o CM Tt co 00 CO d CO CD d P d CO d £2 r- cn o d d CD d o CO CO cu CM CD s I - CM s LL X CD T— 00 X I - CO CD o CO CD CO CM LO O o LO L2L To_q CM ct S> O O d o d OT O d Is- d d CM CC T— d X •* CD Is- CM OT X 1- CD CO O o r- CD CO ^ LL CD- r-J O CM CO o CM CO O d OT d £2 CD d o d OT O CO CO d CM

s s CO o o 00s LO CD I - CO X CO I - CT—O X Sfc CD LO I- CC CM co r-: CD 2 00 CO 00 d o d £ d T— CO d 0.0 1 d CM cc d d

CO X CM CO CN o 00 LO oOT CD LO CO r-: CO CO O d o d £2 OT T— d

d 0.0 2 O d CM CO CM d CM T- CM CD CD LL 00 T— CO X 00 00 o LO 00 LO CM CO CO d LO L2L •st oq d CO CO d o d d T- T— co d CM cc o do d

oi . . o a) . 3 3. a 3. £ c c 3 S 2 3 o o 0) o O) •o i5 ° "5 X £ O ^ wI O« H. to o ™ _i 2 Figure 3.8 Representative urinary accumulation of fluoxetine and norfluoxetine isomers following a 10 min iv infusion of racemic fluoxetine (50 mg equivalent). Data are presented for ewe ID F5141

250

0 12 24 36 48 60 72

Time (hr)

Figure 3.9 Representative urinary accumulation of the glucuronide conjugates of fluoxetine and norfluoxetine following a 10 min iv infusion of racemic fluoxetine (50 mg equivalent). Data are presented for ewe ID F5141

Page 150 3.3.2.4 Fetal Physiological parameters following maternal fluoxetine administration

Fetal femoral arterial blood gas status (P02 and Pco2) [Figure 3.10] and glucose and

lactate concentrations and pH [Figure 3.11] were monitored during control (i.e. prior

to drug administration) and experimental periods (n=5). Due to a malfunction of the

hemoximeter, O2 saturation could not be measured in ewes ID 6209 and 6220, thus

this parameter was not used for analysis. The values for physiological data are

presented as mean ± SEM. During the fetal experiments, the control period fetal

femoral arterial, pH, Pco2, P02, hemoglobin, lactate and glucose concentration were

7.362 ± 0.001, 49.3 ± 4.0 mm Hg, 22.5 ± 1.5 mm Hg, 10.9 ± 0.5 g/dL, 1.14 ± 0.03

and 0.46 ± 0.03 mM, respectively. There were no significant changes in PC02 and

glucose concentration during the experimental period. The fetal arterial pH decreased

from 7.362 to 7.323 at 30 min and returned to the control values at about 2-3 hours,

but this decrease was not statistically significant. However, fetal arterial P02

decreased significantly from 22.5 ±2.1 mm Hg (pre-dosing) to a minimum of 17.5 ±

0.5 mm Hg at 10 min into the infusion and thereafter returned to control levels at 6 h

(22.6 ± 0.7). Fetal arterial blood lactate concentration increased significantly from

1.14 ± 0.03 mmol/L (pre-dose) to a maximum of 1.36 ± 0.15 mmol/L at 2 hours, and

subsequently returned to control values at 3 hours (1.15 ± 0.09 mmol/L).

Page 151 60

75 15 \

10 —i , , , 1 0 6 12 18 24 Time (h)

Figure 3.10 Fetal arterial blood gas status (pC>2, and PCO2) versus time profiles following maternal iv administration of racemic fluoxetine by 10 min infusion to pregnant ewes (50 mg equivalent) (n = 5; mean ± SEM). Only values up to 24 hours were plotted, since there were no changes beyond this time point.

Page 152 Figure 3.11 Fetal arterial blood glucose and lactate concentration and pH versus time profiles following maternal iv administration of racemic fluoxetine by 10 min infusion to pregnant ewes (50 mg equivalent) (n = 5; mean ± SEM). Only values up to 24 hours were plotted, since there were no changes beyond this time point.

Page 153 3.3.3. Pharmacokinetics of fluoxetine and norfluoxetine isomers in pregnant ewes following fetal intravenous fluoxetine administration (10-min infusion).

Average maternal and estimated fetal body weights were 69.8 ± 16.4 kg and 2.75 ±

0.68 kg, respectively on the day of the fetal experiments. Average gestational age on

the day of FX administration was 129.4 ±3.6 days.

3.3.3.1 Plasma pharmacokinetics

The fetal femoral arterial plasma concentration versus time profiles for FX and NFX

isomers following fetal racemic FX administration are depicted in Figure 3.12.

Similar to the maternal profiles observed following maternal FX dosing, plasma

concentrations of both FX isomers increased rapidly during the infusion period and

then declined tri-exponentially with SFX concentrations being significantly higher

than those of RFX throughout the experimental period (p<0.05, two-way ANOVA for

repeated measures). Unlike maternal FX administration, the NFX isomers were not

detected in any of the fetal plasma samples. Table 3.6 presents the calculated fetal

pharmacokinetic parameters for the FX isomers following fetal drug administration.

The S/R ratio of plasma AUC of the FX isomers ranged from 2.08 to 2.33 (mean 2.20

± 0.11) in fetal plasma following fetal FX administration, which is much higher than

the ratio observed in maternal plasma following maternal dosing (range 1.42 to 1.53).

Thus, it resulted in a significantly higher total body clearance (based on estimated

fetal weight) of RFX (10.1 ± 2.8 L/hr/kg) than that of SFX (4.6 ± 1.3 L/hr/kg)

(p<0.05, paired t-test). These fetal values are significantly higher (p<0.01) than the

corresponding estimates of maternal RFX and SFX total body clearances obtained in

the maternal dosing experiments (3.1 ± 0.8 and 1.9 ± 0.2 L/hr/kg, respectively, Table

Page 154 3.3). The steady-state volume of distribution (Vdss) for RFX (38.7 ± 17.7 L/kg) was significantly higher than that of SFX (17.4 ± 8.5 L/kg) resulting in an S/R ratio of

0.45 ± 0.06 and again these values are higher than the corresponding maternal Vjss estimates (9.4 ± 1.3 and 6.3 ± 1.1 L/kg, Table 3.3). However, no significant stereoselectivity was observed in either the apparent terminal elimination half-life or

MRT for the FX isomers. The apparent terminal half-lives of S and RFX were 7.3 ±

3.2 and 7.4 ± 2.9 h, respectively in fetal plasma. These results are very similar to the maternal values obtained following maternal drug dosing (6.7 ± 0.7 and 6.6 ± 2.5 h for SFX and RFX, respectively, Table 3.3). Similarly, there was no significant difference between the fetal MRTs of S and RFX (3.9 ± 1.4 and 4.0 i 1.6 h, respectively). Following fetal dosing, the FX isomers rapidly crossed the placenta to the maternal circulation [Figure 3.13]. Similar to maternal FX administration, maternal plasma concentrations of the FX isomers, following fetal infusion increased rapidly initially and then declined exponentially (Figure 3.13). Maternal SFX concentrations were also observed to be significantly higher than those of RFX following fetal FX dosing (p<0.05, two-way ANOVA for repeated measures). Unlike maternal FX administration, measurable levels of the NFX isomers were only observed in 2 of the 5 ewes. Table 3.7 presents the calculated maternal pharmacokinetic parameters for FX and NFX. The maternal S/R FX AUC ratio was

1.73 ± 0.39 (Table 3.5), which is similar to the maternal S/R AUC ratio calculated after maternal dosing (1.65 ± 0.33, Table 3.3), but is significantly lower than the fetal

S/R ratio (2.20 ± 0.11, Table 3.6; p<0.05). No significant differences were observed for either FX or NFX isomer apparent elimination half-lives or MRTs.

Page 155 su 0, —- •* Q T— CN CO CD o CO o o +1 o d d d d cc c +1 +1 +1 +1 +1 o CD 00 UO 00 co CO CN •tf CD •tf 05 cu csi d d d d E

Q CSJ CO +1 uo CN +1 d CO CO +l +1 CO c CO +1 +1 CO co CM +1 O) cu tf CO co co" •tf 00 E tf Q uo r-- CO CO +1 CO rsj cq 1^ O) CD +l o ^ CSJ T— csi -t—-*—»« c fc d c\j +1 +1 +1 +1 +1 TD CO CD o •tf * cu CM X~ cd o tf cd E CM CO raCl UO uo o CM 00 cq o> fc c CO d tf uo d ^_ CO CD CM Cl CD LO O CO CO 00 uo tf d V d 1— d tf CO d 00 •tf CL CM tt' ' c o CD CD o CO 00 o

CO tf uo csi tf Td— oo CO CD ro 00 •*—>'CcO

CN CN CM 00 CO m i LL e d fc i_ •a d Csi uo uo d 7d— o•tf CM CO Q. Xro tf CD CM uo o CCDLL . fc CO rn Td— uo uo d oo CO co CO d CZ CD CM LO 0 cu O ro o CO CM uo CD O E i5 CN fc V CO uo d CM UO d 00 CL CD cc CN 00 r— CD * C i 5 CD_ o uo CO CM 00 00 CD E "o fc d UO uo csi tf •tf o CO T— CO CO 00 CO i_ CD CN CL) _> o CD CD o CO uo JC va l csi csi d CO csi tf o 0 uo 00 T— 00 CD JC JC _. E E CM CO o 00 CD o o 4= l_ CO uo 00 uo csi ,d— 00 M— O -i—• -*—• CN CM c c CN 0 0 CO CM 00 CO q o 1 ffe i iff e fc 00 csi CM •tf T3 tf T— CM T3 cc _>> _>s •*-< -t—' Cco o Cro ifi c o LC_ _7 O CL)

03 c« IP .s t= is §

S ^ ft g

.2 OT °l-i "-1 +-• rN

CD CO Is x ?< LL. LL X X cK £ ZZtl.lL 03 CO QC 00 CC C O U ™ CO TlH ^ Q S§ g 00 cu

•-1 — CU

£ S o CM O • * cn OT LO _ x ^ Z ^ t T3 •a£H cu "~ E Scd OS fe s (U C+H ^1) ss OT CM _cu 3 o OT 2 fe .g

X i - o

s cIy *-8l 3 > LM o c CU ? O

g ~OT <+H O -fa p o O £ c 5o rt (-|w/6u) uoi}BJ}ueouoo 6njp ewseid rt *1 ,

Cs-U S fe ON < Q OJ g CO CO "ci o d d CM CD CO SP +1 +1 CJ) O) CO or c Tt CO +i d d d CO CD CO q Is- E

Q CO cq CD +1 S 5 CM o +1 +1 Td— TO co +i o> CO CO CM ed CD CO CO E CO LO• CO o in O IO T— CO +1 in cn T- c +1 +1 +1 Tt T- co CM Tt O) (g CO T- CM CO T— CM CM d CD CO O >< CO CO T- Tt y- 00 d TC—O vo CO " d co it T- CM io s I - CM >< CO Tt CO CM CD TC—O T- Tt CM y- d CC *

CO CM Tt Tt CO od .Q CM CO CM 31 c O CO 00 CO -*-> I Tt 13 CM d cr cO Is- CO _g 00 Tt d r-: CO CD cu O CN CO CO CO CD D CT d d CM O c 0 U3 E » CO in co X Is- oo O -2 iri CO CM d t CO iri w co CM Z > CO CO £ ro CN JC

s 00 O) I -; X CD Is- CD +Z CD iri CM ^ d iri r-: O CM CD 0) t d CM d jD c c= CM o CO Q, 0) CM o CM c cu CD CO cq CO CM c od CM iri TJ CO >, >. O" c c co co o o *E £ c c DO DO 5 5 CO CO * < CD 3 => cc o cc < 3.3.3.2 Amniotic andfetal tracheal fluids

The amniotic concentrations versus time profiles of the FX isomers are shown in

Figure 3.14. Similar to the fetal plasma drug concentration profiles, NFX isomers

were not detected in either amniotic or tracheal fluid. The amniotic fluid

concentrations of the FX isomers increased over the first hour and decreased slowly

thereafter reaching the assay limit of quantitation at 36 h. Post-infusion amniotic

fluid FX concentrations were about 5 to 10-fold lower than fetal plasma levels and no

significant accumulation was observed. Unlike the fetal plasma concentration

profiles (which show higher concentrations of SFX over RFX), there was no

significant difference between the levels of the FX isomers in amniotic fluid (p> 0.05,

two-way ANOVA for repeated measures). As observed with maternal FX dosing,

glucuronide or sulfate conjugates were also not detectable in amniotic fluid samples

following fetal drug administration. FX isomer concentrations in tracheal fluid were

~10-fold higher than those observed in amniotic fluid (Figure 3.15). Tracheal fluid

FX isomer concentrations were similar to those measured in the fetal plasma;

however, unlike plasma, no significant stereoselectivity was observed (p> 0.05, two-

way ANOVA for repeated measures).

Page 160 IP a.

03 cl)

o d (~|w uo!}BJ}uaouoo oijojutuv /6u) 6njp <5H 03 .2 CJ 00 03

s bO C C O

£ '3 IP O *-« a. ^ S3 .O (U ^ c _cu o "5b cu -fc! « '"5

° I cr*

CO

cu 1—1

lo CO

?S

tu JU fi "3 r CU bo CU s o <+H r-H O ^—' —tu flcu i+H -a 0 ^ T3 rn <4H tU § ° -5

R £

S C+c2H X o ca C 3 '3 a O ca o ("|LU/6U) UO!+BJ}U80UOO 6mp leauoej} |B),aj CU ~"^ C+H ca o

0) s WD • wm fa 3.3.3.3 Maternal urine pharmacokinetics following fetal fluoxetine administration

Both FX and NFX isomers and their glucuronide conjugates were excreted in

maternal urine following fetal FX administration [Figures 3.16 and 3.17]. Maternal

renal excretion parameters are given in Table 3.8. Similar to maternal FX dosing, the

total urinary excretion of FX and NFX and their glucuronides was low and accounted

for only 0.75% of maternal drug elimination (data not shown). However, unlike

maternal FX dosing, the cumulative amount of the FX isomers exhibited significant

stereoselectivity (S>R), with the S/R ratio averaging 1.63 ± 0.23 (p<0.05, paired t-

test). There was, however, no difference in either the renal clearance of the FX

isomers (S/R ratio of 0.98 ± 0.27), which was normalized to maternal AUC, or the

cumulative amount of the NFX isomers (S/R ratio of 1.09 ± 0.57; Table 3.8). As

observed in the previous experiments, sulfate conjugates could not be detected.

Page 163 -•- SNFX

Time (hr)

Figure 3.16 Representative maternal urinary accumulation of fluoxetine and norfluoxetine isomers following a 10 min fetal iv infusion of racemic fluoxetine (10 mg equivalent). Data are presented for ewe ID 123.

Figure 3.17 Representative maternal urinary accumulation of the glucuronide conjugates of fluoxetine and norfluoxetine following a 10 min fetal iv infusion of racemic fluoxetine (10 mg equivalent). Data are presented for ewe ID 123.

Page 164 Q g CO —• CN m m ro +1 o ci d ci ci o c +1 +1 +1 +1 +1 +1 CC ro co oo 01 m co cu o q q q IP co E c3) a. o O

Q o CD O 3 CO o 3 O o ci co O CO +1 +1 o 3) O fc " cn S a +i T O * +1 ro o co +1 +1 co If) CO tf o E o ci

Q CO 3 CO co a & cn o n CN P 3) «> +1 ; a 9> 2 E "* o o +l o a tf +1 * -h +1 CC T CN I * CM o £ ° i CD Q: i< 2:

X tf m T- CM CD U. CQ- o co pO § CO o o O ci CO r<

co CM X f- O tf U- tf ci o O cc o 1^

CD o O x 2 O co ci O O ci

co •5 -5 o - - CO d o P 15 o cS o cc o

X CD co tf. X r- o O —- o —• CM fe * o' ci CO cu

X cd co o cc CN o ci

•tf X CM o otf tf o tf o O CO fe * o en n. CN CO CO r- o o CO i- o tf © TJ o v a ci CC

CO o CO X "J — d*" o CN co c cn o E ° O C o CO o o ci CM CM £ "a CD CO o o CO CM CO 1^ o o CD ro _> o 1= CM 40

Kl Kl f I CJ +J 1 § 5 5 as 5 IH o a ..

3 op a o o T> Ol TJ TJ co "S o

W O '. 3.4 Fetal Physiological parameters

Fetal femoral arterial blood gas status (PO2, PCO2, O2 saturation) [Figure 3.18] and

glucose and lactate concentrations and pH [Figure 3.19] were monitored during

control and experimental periods (n=5). The values for physiological data are

presented as mean ± SEM. During the fetal experiments, the control period fetal

femoral arterial, pH, Pco2, P02, O2 saturation, lactate and glucose concentration were

7.361 ± 0.017, 48.5 ± 0.4 mm Hg, 21.7 ± 0.7 mm Hg, 57.0 ± 3.9%, 10.9 ± 0.5 g/dL,

1.08 ± 0.16 and 0.68 ±0.17 mmol/L, respectively. There were no significant changes

in PCO2 during the experimental period. However, as with maternal drug

administration, fetal pH and P02 decreased to 7.316 ± 0.008 and 17.2 ± 1.7 mm Hg at

1 hour, and then returned to control levels at 6 hours (7.351 ± 0.007 and 22.5-± 0.6

mm Hg). Similarly, fetal O2 saturation decreased from 57.0 ± 6.7 % (pre-dose) to

32.7 ± 3.0% (minimum at 15 min), and returned to normal values at 6 hours (43.8 ±

6.5%>). Fetal arterial blood lactate concentration increased from 1.03 ± 0.21 mmol/L

(pre-dose) to reach a maximum of 2.53 ± 0.65 mmol/L at 15 min, and subsequently

returned to the control value at 9 hours (1.30 ± 0.16 mmol/L). Blood glucose

concentration also increased to a maximum of 1.21 ± 0.33 mmol/L at 45 min, and

thereafter returned to control levels at 4 hour (0.78 ±0.13 mmol/L).

Page 166 Figure 3.18 Fetal arterial blood gas status (jpOi, PCO2 and O2 saturation) versus time profiles following a fetal 10 min iv infusion of racemic fluoxetine (10 mg

equivalent (n = 5 for p02, pC02; n=3 for O2 saturation; mean ± SEM). Only the values up to 24 hours were plotted, since there were no changes beyond this time point.

Page 167 4

Glucose Lactate

1 -\

12 18 24

Time (h)

7.40

Figure 3.19 Fetal arterial blood glucose and lactate concentrations and pH versus time profiles following a fetal 10 min iv infusion of racemic fluoxetine (10 mg equivalent) (n = 5; mean ± SEM). Only the values up to 24 hours were plotted, since there were no changes beyond this time point.

Page 168 3.3.4 Plasma protein binding of fluoxetine and norfluoxetine

3.3.4.1 Stability of fluoxetine and norfluoxetine in plasma

Stability of FX and NFX in maternal ovine plasma at physiological temperature was

examined by measuring the concentration of the FX and NFX isomers before and

after incubation. Blank maternal ovine plasma (both heparinized and EDTA-treated)

and fetal (heparinized) plasma samples were spiked with appropriate amounts of

spiking solution of FX only, NFX only or a FX-NFX mixture prepared in isotonic

buffer (PBS). The volume of spiking solution was less than 5% of the total volume to

minimize artifacts introduced by dilution of plasma protein. Following spiking,

plasma samples were incubated for 1 hour at appropriate physiological temperatures

to achieve equilibrium of plasma protein binding of drug and/or metabolite (39°C for

maternal ovine plasma and 39.5°C for fetal ovine plasma). The concentrations of FX

and NFX isomers were measured before and after a 1 -hour incubation. There was no

significant change (<5%) in plasma concentration of either FX or NFX during these

incubation periods.

In addition, spiked plasma samples were incubated in the dialysis cells under the

dialysis condition to determine if FX and NFX would be stable in plasma over the

duration of the equilibrium dialysis period. Similarly, no significant change (<5%) in

the concentrations of FX and NFX isomers was observed over the 1 to 12 hour period

of study.

Page 169 3.3.4.2 Free fraction determination offluoxetine and norfluoxetine using ultrafiltration

The ultrafiltration method for the plasma protein binding of the FX and NFX isomers

was evaluated using physiological buffer (PBS) and ovine plasma samples spiked

with known concentrations of FX and NFX (100, 1000 and 10000 ng/mL). A high

degree (>60%) of non-specific binding of FX and NFX to the filter membrane and/or

apparatus was observed following the ultrafiltration procedure. Therefore, this

method was not used for the determination of the free fraction of the FX or NFX

isomers.

3.3.4.3 Optimization of equilibrium time and assessment of volume shift and pH change

of equilibrium dialysis samples

Further to the stability in plasma at physiological conditions (section 3.3.4.1), the

optimal time to obtain equilibrium of protein binding was evaluated prior to the

dialysis study. Human and ovine (maternal and fetal) plasma samples spiked with FX

and NFX were incubated in the equilibrium cells for 1, 2, 3, 4, 6 and 12 hours at the

appropriate temperatures. The FX and NFX concentrations in both plasma and buffer

side of dialysis membrane were measured at each time point over the incubation

period. The concentrations of FX and NFX in the buffer compartment increased over

one to three hours of incubation, with no further significant increases observed after

4-6 hours (Figure 5.1). Therefore, an optimal incubation time of 4 hours was chosen

for the equilibrium dialysis study.

Page 170 10 J •a

• Maternal —9— Fetal —a—Human •a

0 4 8 12 Time (h)

Figure 3.20 Optimization of equilibrium dialysis time for fluoxetine and norfluoxetine isomers. Free fraction (%) of RFX over the incubation range is presented. Similar results were obtained for the S isomer (Data not shown).

Both the volume and pH of samples on each side of the dialysis cell were measured

before and after dialysis. No significant volume shift or pH change was observed in

the dialysate (t-test, p > 0.05). The volume shift was within 5% of the original

volume (0.90 mL, on each side of the dialysis cell before dialysis) and the pH change

was within 2.8% of the original pH (7.4) following 4 hours of incubation.

3.3.4.4 Free fraction determination of fluoxetine and norfluoxetine using equilibrium

dialysis

The free fraction (unbound) of FX and NFX in maternal and fetal plasma was

determined by the equilibrium dialysis method following 4 hours of incubation at

39°C and 39.5°C, respectively (Tables 3.9 to 3.12). Ovine plasma samples prepared

by both heparin- and EDTA-treatments were used to determine whether heparin

treatment would affect plasma protein binding (Tables 3.9 and 3.10). For the in vitro

plasma protein binding study, an appropriate amount of FX and/or NFX spiking

Page 171 solution prepared in PBS (less than 5% of total volume) was added to drug-free plasma. Furthermore, they were incubated for 1 hour to allow for equilibrium to be achieved between the analytes and the binding protein(s).

The FX and NFX isomers bind extensively to plasma proteins (i.e. bound fraction

>95%) in adult (non-pregnant) and maternal ovine plasma. The free fractions of FX isomers in maternal plasma (both heparinized and EDTA-treated) exhibited significant stereoselectivity (p<0.01, paired t-test) with the S/R ratio averaging 0.46 ±

0.09 and 0.49 ± 0.04, respectively. There was no significant difference in plasma protein binding of FX between heparinized or EDTA-treated ovine adult plasma

(Tables 3.9 and 3.10). No significant concentration dependency (p> 0.05, one-way

ANOVA) was observed. In vitro, the free fraction of RNFX tended to be higher than the value for SNFX, but this difference was not statistically significant. Compared to human plasma (see Section 4.12.5), the free fractions of both NFX isomers were higher (p>0.01). In contrast, the free fraction of SFX was lower (p<0.01). The S/R ratio for FX was significantly lower (0.46 ± 0.09) for ovine plasma than that of human plasma (0.74 ± 0.17). However, the S/R ratios of the free fractions of NFX were similar in adult ovine (range 0.64-0.94) and human plasma (range 0.72-0.81).

The free fractions of both the FX and NFX isomers were significantly higher (p<0.01) in fetal ovine plasma (Table 3.11) compared to those in maternal plasma (Table 3.9).

This indicates that plasma protein binding of FX and NFX is lower in the fetus.

Similar to the maternal plasma, there was a significant difference between SFX and

RFX (pO.Ol), with an S/R ratio of 0.56 + 0.08. The SNFX and RNFX free fractions

Page 172 were also significantly different from each other (p<0.05), with the S/R ratio averaging 0.64 ±0.16. There was no significant difference in the S/R ratio of both FX and NFX isomers between maternal and fetal ovine plasma.

Plasma protein binding of FX and NFX was also measured in pooled in vivo samples from selected samples with high concentrations of FX and/or NFX (>100 ng/mL per isomers). Similar to the in vitro protein binding samples, there was significant stereoselectivity in the free factions of the FX isomers (Table 3.12). No significant differences were observed in the free fraction of the analytes in plasma samples from non-pregnant and pregnant ewes.

Fetal NFX binding could not be determined ex vivo, because of the lack of these metabolites in fetal plasma during fetal drug administration. There was no significant concentration dependency for the free fraction of any of the isomers over the observed maternal plasma concentration range, and this was also the case in the fetus.

Similar to the in vitro study, the free fractions of the FX isomers were higher in fetal plasma compared to maternal plasma.

Page 173 CD -a CU

CD

CU CN < Tt CD t CD IS- o -a CO in CD d H CD +1 CU CD co CD d d N +1 Q CN 00 CO W d rt 03 OT ft rt CU

03 CD 3 Tt OT O rt o rt CM O CD CD CD Tt d Tt co +1 cu 5 d +1 CD g CD 3 t n d O d

o (U g Tt CO CN CO d co •> d d T—" d %-•a» v CU OT cp +1 +1 +1 +1 l-l CO CD 00 CM X CU CD T— CD Tt o CO o o co csi T- +1 CO CN CO CO >/-> =3 © o g o • oOTt—i CO _g d l-l CU X V o n e * a •*-» S CU co" Tt Tf I - T3 o CU X l-l C) d T ' T- Tt rt Tt o +1 +1 +1 +1 Cl CU o LO Tt Tf CO csi +1 o S cu CO 00 CO %-> I - _CgU n o CU OTCl CO C T3 VS

CO S ,—I O X CO o I - CO cp CN +1 +1 +1 +1 OB'S csi csi CN T3 CN CD CD CD "<_ .s § CU csi T~ "CJ cD 3c 5=1 a -t-» .5 CD rH IH CD S-l C+H CO ^ cu Tf '53 CN CO Ct) +J ,o 2i - S s T—• T I - Tt rt x d CN d o "a3 CD 2 § +1 +1 +1 +1 J—i -*-» d d d d o e ct: CO CD CO +1 +1 +1 +1 o 0ft3 rt Tt •-, OT 03 C "ft ^3 ^ cu ft CS co" o a CO co" II IS o II sr '5 n CU C

m * * a. CD s tU II I - T3 fi O CO O) CT) o O O •tf CD ro +1 T3 +1 +1 c U o O O CU CO N •tf N i-~ CD r-- o d d kH CO OH ro ca i PH CU CU .fi, -fi, CO * < ca ca # CO C*D C*D B O a o O o h- CO CO CD CO ci Ci m -tf CD d "P< +1 ca +1 +1 +1 CD d o CD cu 00 CD 00 "3 CO C •tf •tf LO '> d d d tU o fi •73 o JU < '> d LO •tf •tf O o V PH ci ci o CO co OH^ o * in +1 +1 v tU LO cb +1 * o .5 ro .fi CO CO CD •tf CU o in %—» .fi CO 00 DH^ 3 fi ci ci w t« § § LO CM h- ca ci) 2 X d tfi e ,x +1 Q. C LL. LL +1 +1 +1 cd tf cd c S co ra tu o fi -a cu E o. c fi O tu •2 fc 1 § & cu c o < cu .fi X CM 5 « CD CO 00 tU tu fi B s '5 o ci O ci o

C s raS ra c g S « S3 «S o o CO ro tfi Ifi c ro ra ra > PH CO J2 c T3 II cn i LfiLHS rv ca O) 'S CD CD c '3 "HH ro S'S 2 PH _> cu .S X .£P op o CD ?.2P JC cu CO CO ro 'co co ra < * < •5 * < ~s CN

CU es H H 3.3.5 Stereoselective in vitro drug metabolism of fluoxetine in hepatic microsomes

The in vitro metabolism of FX in hepatic microsomes was measured as the formation

rate of N-demethylated metabolite, NFX, following 30-min incubations. An

incubation concentration of 100 ng/mL (0.32 pM) per FX isomer was used for the

maternal and fetal ovine microsomes. The results obtained for the pooled sheep

microsomes are presented in Table 5.7. All microsomal incubations of FX were

performed in duplicate. Norfluoxetine isomers were not detected in ovine fetal

microsomal incubates. The formation rates of SNFX and RNFX were similar in

ovine maternal microsomes, with the S/R ratios close to unity.

Table 3.13 Formation rate of norfluoxetine isomers in pooled ovine microsomes from pregnant ewes and fetal sheep (n=2).

Hepatic microsome Formation rate

(pmol/mg protein/30 min)

SNFX RNFX S/R ratio

Ovine (maternal) 26.8 28.4 0.94

21.3 23.5 0.91

Ovine (fetal) N/D N/D

N/D N/D

N/D: not determined as the NFX concentration was below the limit of quantitation

Page 176 3.4 DISCUSSION

3.4.1 Pharmacokinetics of fluoxetine and norfluoxetine in adult and fetal sheep following intravenous fluoxetine administration

In non-pregnant ewes, the systemic clearance, elimination half-life and volume of

distribution of total (i.e. racemic) FX were ~1.7 L/h/kg, 17.5 h and 15.1 L/kg,

respectively. The corresponding values in normal mice are 1.48 L/h/kg, 12.9 h and

27.6 L/kg (Holladay et al, 1998), whereas for rats they are -2.8 L/h/kg, 5.1 h and 18

L/kg (Caccia et al, 1990). Given the marked difference in size and basal metabolic

rate with this 3 species, the pharmacokinetic parameters are surprisingly similar,

although the rat appears to have a lower FX half-life compared to the other 2 species.

In the human, the oral clearance and half-life values differ in poor and extensive

metabolizers in terms of debrisoquine and sparteine, which are model substrates for

cytochrome P450 2D6 (CYP2D6) (Hamelin et al, 1996; Fjorside et al, 1999).

However, in both cases, the clearance estimates (-0.15 L/hr/kg, poor metabolizers;

-0.60 L/hr/kg, extensive metabolizers) are considerably lower than the value in the

non-pregnant ewes. Likewise, the half-life values for FX and NFX in humans, which

range from -1-4 d and -7-15 d, respectively (Hiemke and Hartter, 2000), are much

longer than the overall values of -17.5 and - 37.5 h obtained in non-pregnant sheep.

In contrast, the steady-state volume of distribution of FX in ewes (-15.1 L/kg) is at

the lower end of the range (14 - 100 L/kg) reported in the human (Hiemke and

Hartter, 2000). The clearance values for several other drugs are higher in sheep

compared to the human, whereas the half-life values are lower. These include

etidocaine (Pedersen et al, 1982), lidocaine (Bloedow et al, 1980), meperidine

Page 177 (Szeto et al, 1978), metoclopramide (Riggs et al, 1988), diphenhydramine (Yoo et al, 1990) and valproic acid (Gordon et al, 1995). FX appears to exhibit the same pattern.

The current study appears to be the first that has assessed the impact of pregnancy on

FX pharmacokinetics in any species. In comparison to non-pregnant ewes, FX pharmacokinetics in pregnant sheep exhibit a number of statistically significant differences. These include a higher total body clearance (2.5 vs 1.7 L/hr/kg), shorter half-life (6.5 vs 18.0 h) and lower steady-state volume of distribution (7.8 vs 15.1

L/kg). Two potential mechanisms could underlie these differences. The elevated total body clearance of FX in pregnant sheep could be due to induction of CYP2D6, one of the enzymes involved in FX demethylation (see below). Such an induction of

CYP2D6 occurs during human pregnancy, and may be responsible for the increased dose requirements of antidepressants during pregnancy (Wisner et al, 1993,

Wadelius et al, 1997). The second potential mechanism is a rise in the plasma concentration of ai-acid glycoprotein (AAG), the major plasma binding protein for

FX during pregnancy. This phenomenon has been observed in pregnant rhesus monkeys studied longitudinally; AAG levels increased ~4 fold from 13 weeks to term

(Golub et al, 1997). This was in contrast to the fall in plasma albumin concentration that occurs in this and other species (Golub et al, 1997). A similar increase in AAG in late pregnancy has been observed in a longitudinal human study (Haram et al,

1983). In transgenic mice, which overexpress AAG by 8.6 fold compared to normal mice, the volume of distribution and half-life of FX are reduced by 41% and 33%, respectively (Holladay et al, 1998). Thus, a rise in this protein during pregnancy in

Page 178 sheep could explain the lowered FX volume of distribution and half-life. However, a decrease in free fraction of the drug would also be expected and this was not seen

(Table 3.12). A similar phenomenon was observed by Holladay et al (1998) in the transgenic mice overexpressing AAG; FX free fraction was not significantly different from the value in normal mice. It was suggested that this finding was due to non• specific binding of FX to albumin or AAG with the equilibrium dialysis method used to assess binding (Holladay et al, 1998). Further determinations of plasma protein binding and AAG levels in non-pregnant and pregnant sheep are warranted to address this issue.

The CL (7.2 L/hr/kg) and Vdss (28.1 L/kg) estimates of FX in the fetus, obtained in the fetal drug administration experiments, are significantly higher than in the ewe.

This is very likely due to rapid fetal to maternal transfer of the drug, so that at least a portion of the maternal clearance and volume of distribution are included in the fetal estimates. In contrast, the FX half-life in the fetus (7.4 hr) was similar to that in the ewe. This suggests that the elimination of the drug in the fetus is largely the result of transplacental FX transfer of the drug to ewe and elimination via maternal processes, which is consistent with the lack of evidence for fetal FX metabolism (see section

3.5.2.4).

Previous studies in our laboratory have demonstrated that other amine drugs, such as diphenhydramine, metoclopramide, ritodrine and labetalol, accumulate in fetal lung and amniotic fluid to achieve concentrations considerably higher than in fetal plasma

(Rurak et al, 1991). Furthermore, with metoclopramide we have shown that the accumulation in fetal lung fluid is accompanied by high drug concentrations in fetal

Page 179 lung tissue (Riggs et al, 1987). This suggests that the fetal lungs take up these amine

drugs, as is also the case in the adult, but that some of the drug in the fetal lung is

transferred into the fluid secreted into the airways. However, in contrast to previous

findings, significant accumulation of FX and NFX in fetal lung and amniotic fluid in

the current study was not observed; the drug levels in these fluids were lower than or

equal to those in fetal plasma. A similar finding for amniotic fluid has been reported

in pregnant rats by Pohland et al. (1989). At present, an explanation for the

difference between FX and the other basic lipophilic drugs studied is lacking. There

is extensive accumulation of FX and other basic lipophilic psychotropic drugs in the

rat lung via lysosomal trapping and phospholipid binding (Daniel and Wojcikowski,

1997). The former process involves diffusion of the non-ionized form of the drug

into the acidic interior of lysosome where protonation of the drug occurs, preventing

efflux from the lysosome. As noted below (Section 3.5.2), FX is more lipophilic than

the other basic drugs that we have studied. Thus it may be that lysosomal trapping is

more important for FX, and if this process also occurs in the fetus, lung accumulation

of the drug may occur, with limited efflux into lung fluid. However, measurement of

FX concentrations in subcellular components of the lung in comparison to the other

drugs is necessary to assess this possibility.

3.4.2 Placental Transfer of Fluoxetine and Norfluoxetine

There was rapid and extensive maternal to fetal placental transfer of both FX and

NFX following maternal FX administration, with the F/M AUC ratios for the drug

and metabolite averaging -0.60 and 0.65, respectively. The extent of placental drug

transfer following maternal bolus i.v. injection seems largely dependent upon the

Page 180 lipophilicity of the compound (Rurak et al, 1991). The log octanol/water partition coefficient (log P) of FX is 4.05; thus it is highly lipophilic. Its F/M AUC ratio is substantially higher than that of the more polar drugs, labetalol (F/M AUC ratio 0.14, log P 0.08, Yeleswaram et al, 1992) and ritodrine (F/M AUC ratio 0.02, log P -2.0,

Rurak et al, 1991). However, the F/M AUC ratio of FX is not much higher than that

(0.41) of valproic acid (Gordon et al, 1995) with a log P value of 2.6 and is lower than the values for diphenhydramine (0.85) and metoclopramide (0.82), both of which have substantially lower log P values of 3.1 and 1.1, respectively (Riggs et al, 1988;

Yoo et al, 1986). It is likely that the reduced fetal exposure to FX, compared to the diphenhydramine and metoclopramide, is due to the high degree of plasma protein binding of the drug in the ewe, and the lower binding in the fetus, since this can also reduce the extent of maternal-fetal drug transfer (Hill and Abramson, 1988).

The current study appears to be the first in which FX placental transfer has been quantitatively assessed in any species. However, the ratio of drug concentrations in human umbilical cord blood and maternal blood at delivery is close to 1.0 (see

Chapter 4; Kim et al, 1999b), which is substantially higher than the value obtained in the sheep in the current study. This difference is unlikely to be solely due to the structural and functional differences between the sheep and human placenta, since a lower placental permeability in the former species is largely restricted to hydrophilic compounds (Faber and Thornburg, 1983). The higher degree of fetal exposure to FX in the human is likely at least in part due to long term drug administration, in contrast to the acute FX administration in the current study. When drugs are administered to steady state, fetal drug elimination becomes important in determining fetal drug

Page 181 exposure (Rurak et al, 1991). If the human fetus has a limited ability to metabolize

FX, as in the fetal lamb (see section 3.5.2.4), this could explain the high F/M plasma

ratio. Similar to this, there is a current study involving 8-day iv FX infusion in

pregnant sheep (Chien et al, 1999). The fetal/maternal steady-state drug

concentration ratio is 0.75, i.e. higher than the F/M AUC ratio in the current study.

3.4.3 Fluoxetine Effects on Fetal Blood Gas and Acid-Base Status

With both maternal and fetal FX administration, there were transient decreases in

fetal oxygenation and blood pH and increases in blood lactate level, and with fetal

drug administration, blood glucose concentration. We have recently observed similar

transient changes over the first ~6 hr of an 8-day maternal i.v. FX infusion protocol

and these were temporarily associated with an increase in blood serotonin level and a

decrease in uterine blood flow (J Morrison, C Chien, KW Riggs and DW Rurak,

unpublished data). A similar acute action of FX to increase plasma serotonin

concentration has been observed in rats (Ortiz and Artigas, 1992). Serotonin is a

potent uterine vasoconstrictor (Clark et al, 1980). Thus, with maternal FX

administration it is likely that FX-elicited inhibition of serotonin reuptake by maternal

platelets and perhaps other cellular blood constituents increased blood serotonin

levels leading to decreased uterine blood flow. This in turn reduced fetal oxygen

delivery and caused the transient hypoxemia and acidemia. Although maternal blood

gas status was not determined in this project, it is unlikely that alterations in maternal

blood gases and pH contributed to the fetal hypoxemia and acidemia, since in the 8-

day maternal FX infusion protocol mentioned above, no changes in maternal P02,

Pco2 or pH have been observed. With fetal FX administration, a decrease in uterine

Page 182 blood flow could have been involved in the fetal hypoxemia and acidemia, since as

illustrated in Figure 3.13, there is rapid fetal to maternal transfer of the drug.

However, serotonin is also a potent umbilical vasoconstrictor (Berman et al., 1978).

Thus, FX-elicited increases in fetal plasma serotonin concentration could have caused

a fall in umbilical blood flow, which resulted in the fall in P02 and pH. Given the

modest and transient nature of the changes, the impact on the fetus is likely minimal,

since similar changes regularly occur during gestation as a consequence of

antepartum uterine contractions and fetal somatic activity (Harding et al, 1983).

3.4.4 Fluoxetine Metabolism in the Adult and Fetal Sheep

In humans, FX (-2.5%), FX glucuronide (-5.2%), NFX (-10%), NFX glucuronide

(~9.5%>) and hippuric acid (-20%) were recovered in urine after a single dose of I4C-

FX (Lemberger et al, 1985; Begstrom et al, 1988). Thus in total these urinary

metabolites account for -50% of the dose in this species, and an additional -13% of

the radioactivity was recovered in feces. In addition, p-trifluoromethylphenol, a

dealkylated metabolite of FX has been detected in rat and human plasma and urine

following FX administration (Urichuk et al, 1997), although the quantitative

importance of this metabolite is not yet known. In the current study, FX, NFX and

their glucuronides were detected in adult sheep urine. However in both non-pregnant

and pregnant ewes, they accounted for only -3-4% of the dose. The maternal renal

clearance values for FX and NFX were correspondingly low, which is similar to the

situation in humans (Lemberger et al, 1985). The low renal clearance of FX is

consistent with the observation that renal impairment does not significantly affect the

disposition of FX or NFX, except in extreme cases such as in patients with creatinine

Page 183 clearance below 10 mL/min (Aronoff et al, 1984). In the current study, the cumulative urinary excretion curves of these compounds did not reach a plateau level.

Thus, we may have slightly underestimated their role in FX elimination. Nonetheless the data strongly suggest that the renal excretion of FX, NFX and their glucuronides contribute much less to overall FX elimination in sheep than in the human. Urinary hippuric acid excretion was not determined since estimation of the amount resulting from FX metabolism would require a labeled form of the drug. Recently, our laboratory has detected trifluoromethylphenol in urine from adult sheep during and following FX administration, but have not yet assessed its quantitative contribution to drug elimination (C Chien, DW Rurak, KW Riggs, unpublished data). Thus, the metabolic profile of FX in adult sheep appears qualitatively but perhaps not quantitatively similar to that in the human.

As noted above, the difference in FX pharmacokinetics between poor and extensive metabolizers of CYP2D6 suggests the involvement of this P450 isozyme in FX demethylation in humans. However, there are conflicting in vitro reports suggesting that FX N-demethylation is mediated mainly by CYP2C9 with a minor role for

CYP2D6 (Stevens and Wrighton, 1993; von Moltke et al 1997). The limited information available on adult sheep cytochrome P450s indicates the presence of isoforms related to CYP1A, 2B, and 3A families (Murray, 1992; Galtier and

Alvinerie, 1996; Nelson, 1999). Moreover, our results indicating a negative correlation between the S/R ratio of FX AUC (see pp. 190-191) and systemic clearance in sheep and a progressive decrease in FX clearance during an 8-day infusion protocol (Chien et al., 1999) are consistent with the presence of a CYP2D6-

Page 184 like isozyme that is involved in FX metabolism and is also inhibited by FX. Thus, the

P450 isozymes implicated in FX N-demethylation in humans appear to also be present in sheep. However, further research is needed to fully characterize them.

In contrast to the situation in the ewe, this study provides no evidence for FX metabolism in the fetus. Evidence for the lack of fetal NFX formation comes both from the fetal FX administration experiments, in which NFX was present in maternal, but not fetal plasma or amniotic and tracheal fluids, and in the in vitro hepatic microsomal incubations, where NFX production by fetal microsomes was not detected. Although the metabolite was present in fetal fluids during maternal FX administration, the fetal plasma levels were always lower than in maternal plasma.

Thus the NFX in the fetus was likely formed in the ewe and diffused across the placenta into the fetal circulation. There is virtually no information available in the literature for the types of cytochromes P450 present in the ovine fetus. However,

CYP isozymes such as 2C9, 2C19 and 2D6, which appear to be involved in the N- demethylation of FX (see Section 4.12.6) are not expressed in the human fetal liver and develop in the early neonatal period (Jacqz-Aigrain et al, 1992; Hakkola et al,

1994; Rich et al, 1997; Treuyer et al, 1997). Thus, the capacity of the human fetus for oxidative drug metabolism is not fully developed during pregnancy and the early neonatal period and our data suggest that this is also the case in the ovine fetus.

The fetal lamb also appears to lack the phase II metabolic pathways for FX, since glucuronide and sulfate conjugates of FX and NFX were not detected in amniotic fluid with either maternal or fetal drug administration. This finding is similar to the observation of Pohland et al. (1989) where much lower radiocarbon concentrations

Page 185 (including FX and its metabolites) were observed in amniotic fluid compared to fetal

rats following oral administration of [14C]-FX to pregnant rats. In the literature, the

fetal lamb can produce glucuronide conjugates of other drugs including

acetaminophen (Wang et al, 1985), morphine (Olsen et al, 1988), ritodrine (Wright

et al, 1991), labetalol (Yeleswaram et al, 1993) and valproic acid (Kumar et al,

2000). Since the glucuronide conjugates are much more hydrophilic than the parent

compounds, they do not readily cross the sheep placenta. Thus if formed in the fetus,

the glucuronides accumulate in fetal fluids, particularly in amniotic and allantoic

fluids (Rurak et al, 1991). Consequently, the failure to detect FX and NFX

glucuronides in amniotic fluid provides a strong evidence for the lack of

glucuronidation of FX and NFX in the fetal lamb. The reason for this deficiency is

not yet apparent. However, with the drugs listed above, the glucuronide conjugates

involve carboxylic acid or alcohol linkages. In contrast, FX glucuronidation involves

an amine linkage. In addition, although morphine-3-glucuronide is formed in the

fetal lamb, morphine-6-glucuronide is not (Olsen et al, 1988), whereas both

conjugates are synthesized in adult sheep (Milne et al, 1993). These findings suggest

that while some UDP-glucuronosyltransferase isoforms are present in fetal lamb,

others, including those involved in glucuronidation of FX and NFX, are not.

3.4.5 Stereoselective Pharmacokinetics of fluoxetine

The present study provides the first report of the stereoselective disposition of FX and

NFX during pregnancy in any species and also obtained detailed observations on the

phenomenon in non-pregnant ewes. In both the adult and fetal sheep, the RFX had a

lower AUC and higher plasma free fraction, CL and VdSS compared to SFX. In non•

age 186 pregnant ewes, RFX t\ap was higher than that for SFX. However, this was not the case in pregnant ewes with maternal drug administration, although there was a difference in the maternal values with fetal drug injection. There was also stereoselective urinary excretion of FX in non-pregnant sheep, with cumulative excretion and renal excretion of SFX being significantly higher. This was not observed in the pregnant ewes with maternal drug injection, whereas with fetal administration, maternal cumulative urinary excretion was higher for SFX compared to RFX. In the fetus, these stereoselective differences were present with both maternal and fetal FX administration, and the latter finding indicates that the stereoselective processes are operative in fetal as well as adult sheep.

For a drug, such as FX, whose elimination is mainly by the hepatic route with limited renal clearance (Table 3.2), the total body clearance depends mainly on its hepatic clearance. The relationship of protein binding, hepatic blood flow and hepatic intrinsic clearance on hepatic drug clearance, based on the well-stirred model

(Wilkinson and Shand, 1975) is as follows:

s~i T _ QH ' fub ' CLIV H ~ Qti+fub-CL'L (3.12)

where QH, CLH, CLINT and fub are the total hepatic blood flow, total hepatic clearance, hepatic intrinsic clearance of the free (unbound) drug and the free fraction of the drug, respectively.

As the hepatic flow is the same for both drug isomers, the equation can be simplified through division of both the numerator and denominator with QH as follows:

Page 187 f •CJ" CL„ = J lib ^^in (3.13)

f • Cl"'int 1 +J lib ^in QH

Since FX is relatively a low clearance drug with high protein binding, the effective

U hepatic intrinsic clearance (fub • CL M) is much smaller than hepatic blood flow (QH)-

Therefore, equation 3.13 can be simplified as follows:

'int (3.14)

As equation 3.14 illustrates, stereoselectivity of hepatic drug clearance for a low CLH drug is likely caused by differences in free fraction (i. e. protein binding) and/or the intrinsic clearance (i.e. metabolism). In the case of FX, this raises the question whether the observed stereoselectivity in total body clearance of FX isomers is, in fact, caused by differential capacity in drug elimination (/'. e. intrinsic clearance) or by differential protein binding (i.e. distribution).

Pharmacokinetic parameters such as total body clearance and subsequently volume of distribution are affected by the extent of protein binding. The dependency on the extent of protein binding is more noticeable in the drug with low clearance (a low extraction ratio), since clearance is proportional to the unbound fraction in plasma.

The extent of drug distribution is largely determined by partition coefficient and plasma/tissue protein binding (Jamali et al, 1989). The former, which is a physical property, is not likely to be stereoselective. However, the extent of protein binding of enantiomers of both acid and basic drugs may differ substantially (Jamali et al, 1989;

Oravcova et al, 1996; Lapicque et al. 1993; Pacifici et al, 1992).

Page 188 In the present study, there were significant differences in plasma SFX and RFX concentrations immediately following FX administration in all the injection protocols

(i.e. non-pregnant, pregnant, fetal). In addition, the apparent volume of distribution was different between the FX isomers. These data suggest that the difference in plasma protein binding of the FX isomers may contribute to stereoselective disposition. Thus, stereospecific plasma protein binding of FX isomers was further examined. The free fraction for RFX in both the adult and fetus is significantly higher than the values for SFX, i.e. the extent of protein binding is less for the RFX, as shown in Tables 3.12. However, with both isomers, and for NFX as well, the degree of plasma protein binding in the ewe is high and similar to that reported for the human (Lemberger et al, 1985). Binding in the fetus is significantly less, and this is likely due to the lower concentrations of albumin and other plasma proteins in the fetus, including AAG (Seta et al, 1991; Kwan et al, 1995).

The adult and fetal ex vivo free fraction values in Table 3.12 can be used with the total drug RFX and SFX CL values in Tables 3.1, 3.3 and 3.6 to estimate the free drug CL values. The resulting estimates in the non-pregnant ewes (40.2 vs 62.6

L/hr/kg), pregnant sheep (53.3 vs 70.7 L/hr/kg) and fetus (96.2 vs 70.5 L/hr/kg) are much closer compared to the total drug CL estimates. Moreover, in the adult sheep, the free clearance of RFX is no longer higher than the value for the S isomer. Thus, differential protein binding of the FX isomers appears sufficient to largely explain the stereoselective disposition of the drug, at least with bolus drug administration.

As described in Chapter 4, the stereoselective disposition of FX associated with long- term drug administration in the human appears to involve FX-mediated inhibition of

Page 189 the non-stereoselective metabolism of the drug by CYP2D6, with a resulting increase in stereoselective metabolism by CYP2C9/2C19. Although FX metabolism by adult sheep hepatic microsomes did not exhibit stereoselectivity (Table 3.13), our laboratory has recently obtained some other evidence that a situation similar to the human may occur in sheep. As was discussed above, there is a decrease in maternal

FX clearance in pregnant sheep during an 8-day i.v. drug infusion protocol (Chien et al, 1999). This is associated with a progressive rise in the maternal plasma S/R FX concentration ratio from 1.78 ± 0.79 at day 1 to 3.40 ± 1.23 at day 8 (C Chien, JE

Morrison, DW Rurak and KW Riggs, unpublished data). These observations are consistent with FX-elicited inhibition of CYP2D6, as occurs in humans over the same time (Alfaro et al, 2000), allowing for greater stereoselective metabolism by

CYP2C9/2C19. Moreover, when the data from the non-pregnant and pregnant ewes in the current study are analyzed together, it appears the inter-animal variability in FX clearance is associated with differences in the degree of stereoselective disposition.

Figure 3.21 plots the SFX/RFX AUC ratio against the racemic clearance value in individual non-pregnant and pregnant ewes. There is a significant, negative correlation between the two variables. Thus, a lower clearance value is associated with a higher S/R FX ratio. This is similar to the stereoselective difference in the oral clearance of FX isomers in poor metabolizers of sparteine in humans, i.e. with a deficiency in CYP2D6 (Fjordside et al, 1999). However, further work in sheep is required to determine the degree of similarity to the human in terms of stereoselective

FX metabolism. In particular, further characterization of the CYP450's in sheep in terms of FX metabolism and inhibition is needed.

Page 190 « Nonpregnant • Regnant

1.3 r =-0.7151 (Fearson) 1.2 p= 0.020

1.1 1.7 1.8 1.9 2.0 2.1 2.2 2.3 log CL (racemic FX)

Figure 3.21 Correlation of AUC S/R ratio of fluoxetine isomers and total body clearance of racemic fluoxetine in non-pregnant and pregnant sheep (n=10). One pregnant ewe (F5141) was eliminated from the analysis due to unusually high S/R ratio.

One additional mechanism has been proposed for the stereoselective disposition of

FX: a stereoselective pulmonary first-pass effect (Bergstrom 1998, personal communication). Drugs administered via an intravenous route will go through the pulmonary circulation prior to the systemic circulation in mammalian species.

Similar to the hepatic first-pass effect observed following oral drug administration, any significant extraction of a drug by the lungs will result in a reduction of its systemic availability. As was discussed above, pulmonary extraction of FX appears to be high with the greatest accumulation of FX and/or NFX observed in the lungs

(Wold et al, 1976; Pohland et al, 1989; Daniel and Wojcikowski, 1997). Similar to these studies, we observed the highest accumulation of both FX and NFX isomers

(with higher concentrations for SFX and RNFX) in the rat lung following

Page 191 intraperitoneal administration of racemic FX (data not shown). As was noted above, this pulmonary accumulation is likely due to lysosomal trapping and phospholipid binding (Daniel and Wojcikowski, 1997). However, these processes, which are mediated mainly by passive diffusion, are not likely to cause stereoselective pulmonary uptake of FX isomers by lungs. The experimental protocol used in our sheep experiments, however, was not designed to determine a pulmonary first-pass effect. This type of experiment would require paired intravenous and intra-arterial administration of the drug. These effects could be examined in future studies.

In summary, these studies demonstrate differences in FX pharmacokinetics between non-pregnant and pregnant ewes, which may be due to elevated AAG levels during pregnancy. There is rapid placental transfer of both FX and NFX to the fetal lamb following maternal drug (racemic FX) administration resulting in significant fetal exposure. FX undergoes extensive stereoselective disposition resulting in higher levels of the SFX isomer in both adult and fetal plasma and this appears to be largely due to differential plasma protein binding of the FX isomers. However, stereoselective metabolism may also be involved, particularly with longer-term drug administration. In contrast, stereoselective disposition was not observed for NFX. It also appears that both phase I and II metabolic pathways for FX are lacking in the fetus.

Page 192 CHAPTER 4

CLINICAL STUDY OF FLUOXETINE AND PAROXETINE DURING PREGNANCY AND THE POSTPARTUM PERIOD

As described in the Chapter 1, the use of antidepressants, especially the selective serotonin reuptake inhibitors, during pregnancy and the postpartum (nursing) period has increased over the past several years. This trend is partly due to more awareness and recognition of pre- and post-partum depressive disorders among clinicians and the general public. In addition, the relatively low incidence of adverse effects of this newer class of drugs compared to traditional tricyclic and other antidepressants has made clinicians more willing to prescribe these medications.

However, information regarding fetal and neonatal exposure to these medications is still limited. In the following study, the extent of fetal and neonatal exposure to the SSRI antidepressants was evaluated. Following an extensive literature search and discussion with several local psychiatrists on current prescribing trends for these drugs, fluoxetine

(FX) and paroxetine (PX) were chosen for the study. In collaboration with the

Reproductive Psychiatry Program (RPP) and Biobehavioral Research Unit (BRU), the

Centre for Community Health Research at BC Children's and Women's Hospital

(BCCWH) and St. Paul's Hospital, the following study was initiated and conducted.

In conjunction with the in vivo sheep pharmacokinetic and in vitro drug metabolism studies described in Chapter 3, the clinical data obtained from this study could provide important information applicable to the rational pharmacological treatment of depression using the SSRIs during pregnancy and the postpartum period.

Page 193 Study design

This is a single-center, outpatient, open-label comparative study of the selective serotonin reuptake inhibitors, FX and PX, during pregnancy and the postpartum period. Subjects received one of two possible antidepressant treatment sequences as a part of their regular psychiatric treatments for major depressive disorders.

Subjects received a chronic daily oral dose of either 10-80 mg of FX or 10-60 mg of PX as prescribed by the attending physician.

The study population consisted of the following five subgroups:

A. Fluoxetine-exposed group recruited during pregnancy

B. Paroxetine-exposed group recruited during pregnancy

C. Fluoxetine-exposed group recruited during the nursing period

D. Paroxetine-exposed group recruited during the nursing period.

E. Control group recruited during pregnancy

For subgroups A and B (patients recruited prior to delivery), during the perinatal period, blood and/or milk samples were collected as described and fetal/neonatal drug exposure was assessed:

1. Prenatal period: For patients who were treated and recruited from the RPP, general health and medication questions (see section 4.5) were asked following their regular psychiatric session with a staff psychiatrist. In cases of complicated pregnancy, fetal monitoring was performed by staff obstetricians at the Centre for

Page 194 Perinatal Diagnosis and Treatment, Maternal-Fetal Medicine, BC Children's and

Women's Hospital.

2. At birth: A cord blood and maternal blood sample were collected immediately after birth to estimate fetal exposure during pregnancy in cooperation with the

Department of Obstetrics and Gynaecology, BC Children's and Women's

Hospital (see section 4.6.2).

3. At the time of PKU testing: During the routine PKU test, an additional blood sample from the infant was collected either at the BRU or Outpatient Laboratory,

BCCWH.

4. Postpartum period: Simultaneous maternal and infant blood samples as well as a breast milk sample were collected at the Outpatient Laboratories of BCCWH or

St. Paul's Hospital according to the protocol described in section 4.5.2.

For subgroups C and D, blood and milk samples during the postpartum period were collected following a minimum of 3-4 week of exposure to a constant dose of FX or paroxetine as described in section 4.5.2.

The subjects in the control group E were pregnant mothers who were neither clinically depressed nor taking any antidepressant medications. Psychosocial and biobehavioral parameters were measured in these control patients as a part of our collaborative research program. These subjects were recruited separately from the study described in this chapter, and these data will be reported elsewhere.

Page 195 The study protocol was reviewed and approved by University of British Columbia

Human Subject Research Ethics Review and BC Women's and St. Paul's Hospital

Research Review Committees.

Patient recruitment

Patients were mainly recruited from the Reproductive Psychiatry Program (RPP) and the Department of Obstetrics and Gynaecology at BCCWH. Pregnant and nursing mothers who came to the RPP with a major depressive disorder and other affective disorders such as panic and obsessive compulsive disorders were offered the opportunity to be included in the study. Additional patients from the Lower

Mainland area of British Columbia were referred to the present study program by local obstetricians or psychiatrists. A majority of the pregnant patients delivered at the BCCWH. However, some patients delivered in local hospitals in the Lower

Mainland area (Lions Gate Hospital, Peace Arch Hospital and Richmond General

Hospital) and the blood samples were obtained in cooperation with collaborators in corresponding hospitals. Each subject was informed of the available treatment options and the risks involved with continuation of medication during pregnancy and the nursing period by staff psychiatrists at RPP (Drs. Shaila Misri, Deirdre

Ryan and Diana Carter) and the investigator (John Kim). As previously mentioned, an age-matched control group was recruited from BCCWH for

APGAR scoring and other infant behavioral evaluation (data not shown) as a part of this collaborative study.

Page 196 Inclusion and exclusion criteria

Inclusion criteria: To be eligible for entry and continuation in the study, patients were required to meet the following eligibility criteria:

1. Subject must be between the age of 18 and 45

2. Subject must be pregnant (or plan to be pregnant) or nursing.

3. Subject must be on chronic fluoxetine and/or paroxetine therapy.

4. Subject must be on a constant drug dose for a minimum for 3 weeks prior to

sample collection.

5. Subject must give written informed consent as approved by the Ethics

Committees at the University of British Columbia and the participating

hospitals.

Exclusion criteria: patients were excluded from the study if any of the following existed at the time of enrollment or during the study:

1. Gross morphological malformation of fetus/neonate.

2. Premature delivery (< 35 weeks of gestation).

3. Spontaneous or elective abortion.

4. Use of diphenhydramine or its prodrugs (e.g. Gravol®) for patients who are

on fluoxetine therapy within 7 days prior to blood sample collection (sample

Page 197 collection may be delayed for at least 7 days following acute dose(s) of

diphenhydramine or its prodrugs).*

5. Use of maprotiline for patients who are on paroxetine therapy.*

6. Discontinuation of breast-feeding.

7. Hospitalization for reasons other than pregnancy/delivery

* Patients on diphenhydramine and maprotiline were excluded from the study due to potential interference of these compounds and their metabolites in analytical methods used for quantitation of FX, NFX and PX. Maprotiline and a chemical analogue of diphenhydramine were used as internal standards for PX and FX/NFX assays, respectively.

Informed patient consent and patient interview

Written informed consent was obtained from the subjects, and the sample collection protocol was reviewed with subjects prior to the study. All subjects who were enrolled and signed informed consent were interviewed by an investigator (John Kim) or assigned designate prior to sample collection. During the interview, the following information was obtained from the patient:

• Demographic information (i. e. age, weight and gestation period)

• General health question (i.e. pre-existing medical conditions)

• Current medication dose and regimen

• Medication history

• Compliance with medication

• Concurrent medications

• Smoking habits

Page 198 • Alcohol consumption

• Caffeine consumption (coffee, tea, cola drinks, cocoa, chocolate-based

products and liqueurs)

• Recreational drug use

• Sleeping patterns

• Appetite

• Nausea and other potential adverse effects related to a medication and/or

pregnancy

(The following additional questions were asked to nursing mothers)

• Infant's demographic information (i.e. sex, weight and age)

• General health question of the infant

• Breast-feeding patterns

• Sleeping patterns of the infant

• Adverse effects related to a medication (e.g. vomiting, and colic)

In addition, the infant blood collection procedure was reviewed and any questions or concerns were addressed. Patients were given an option to participate in the study without the collection of an infant blood sample (i.e. maternal blood and breast milk collection only) if they did not wish to have blood sample collected from the infant.

Page 199 4.5 Study protocol

4.5.1 Sample collection and APGAR scoring at birth and the PKU test

Simultaneous maternal and umbilical cord blood samples were collected at birth

by the investigator (John Kim) or attending physician or nurse at the Delivery

Unit in BC Children's and Women's Hospital or other regional hospitals

participating in the study. The maternal blood sample (~5 mL) was collected by

venipuncture into a 10 mL "red-top" Vacutainer® without additives (Becton

Dickinson, Franklin Lakes, NJ). The umbilical blood sample (~5 mL) was

collected from the umbilical vein of the placenta immediately after separation

from the newborn into a 10 mL "red-top" Vacutainer. Blood samples were

allowed to clot for 30 min and the serum separated following centrifugation at

3000 g for 10 min. The serum samples were transferred to disposable glass tubes

and stored at -20°C until analysis. In addition, APGAR scores were measured at

1 and 5 min following birth as a part of routine clinical care at the delivery units

of the participating hospitals (see section 4.6 for details).

An additional neonatal blood sample (-0.5-1 mL) was collected from the

newborn at the time of phenylketonuria (PKU) testing, which is normally

performed about 1-2 days after birth as a part of routine infant medical care. The

PKU test was performed either at the Outpatient Laboratory of BC Children's and

Women's Hospital or the BRU during assessment of maternal-infant interaction,

which formed another component of this collaborative research project. The

blood sample was collected either by heel-prick or venipuncture during the PKU

Page 200 test sample collection. Again, the blood samples were allowed to clot for 30 min

and the serum separated following centrifugation at 3000 g for 10 min. The

serum sample was transferred to a disposable glass tube and stored at -20°C until

analysis.

4.5.2 Postpartum sample collection

Postpartum blood and breast milk sample collection was performed during routine

psychiatric clinical visits and/or regularly scheduled study sessions for psychiatric

assessment at 1 and 2 months postpartum. Following the patient interview, a

maternal (~5 mL) and infant blood (~1 to 1.5 mL) sample were collected by the

phlebotomist at BC Children's or St. Paul's Hospital. The maternal blood sample

was collected by venipuncture into a 10 mL red-top Vacutainer® without additives

(Becton Dickinson, Franklin Lakes, NJ). Similarly, the infant blood sample was

collected either by heel-prick or by venipuncture using a 26 gauge butterfly

needle and a 3 mL syringe (Becton Dickinson, Franklin Lakes, NJ), depending on

the discretion of the attending physician and/or phlebotomist. Following

collection, the blood sample was transferred to 3 mL red-top Vacutainer® without

additives. In most cases, the investigator (John Kim) or an assigned designate

was present and monitored the infant blood collection procedure. On rare

occasions, if the first attempt to obtain the infant blood sample was not successful,

a second attempt was made after consulting with the mother. The blood sample

was allowed to clot for 30 min and the serum separated following centrifugation

at 3000 g for 10 min. The serum sample was transferred to a disposable glass

tube and stored at -20°C until analysis. A breast milk (foremilk) sample (-10-20

Page 201 mL) was collected at the same time either by manual expression or by breast pump.

Analysis of perinatal complications

In order to assess the degree of postnatal complications at birth, the newborn

(APGAR) score was recorded at 1 min and 5 min as a part of clinical treatment at the delivery unit of the participating hospital. The score was based on the method described by Apgar (1953) and subsequent papers. The APGAR score is a measurement of a newborn's response to birth and the ratings are based on the following simplified factors:

Appearance (colour): 0 for blue or pale 1 for pink with blue extremities 2 for pink overall

Pulse (heart rate): 0 for absent 1 for below 100 bpm 2 for above 100 bpm

Grimace (reflex irritability): Ofor non-responsive 1 for grimace 2 for cough or sneeze

Activities (muscle tone): 0 for absent (limp) 1 for arms and legs flexed 2 for active movement

Respiration: 0 for absent 1 for slow, irregular 2 for good, crying

Page 202 Materials and supplies

Chemicals, reagents and other materials used during the drug metabolism and protein binding studies as well as for sample analysis, along with the information on their purity (where applicable) and source, are listed below.

Ammonium acetate, sodium bicarbonate, glacial acetic acid, disodium hydrogen orthophosphate (dibasic), potassium dihydrogen orthophosphate (monobasic), potassium chloride, sodium hydroxide pellets, magnesium chloride, and hydrochloric acid were obtained from BDH Chemicals Co. (Toronto, ON) and/or

Fisher Scientific (Nepean, ON) and were of analytical reagent and/or ACS grade.

Sequanal grade triethylamine (TEA) was purchased from Pierce Chemical Co.

(Rockville, IL). Acetonitrile, ethyl acetate, methanol, dichloromethane, toluene, benzene, ethanol, isopropanol, acetone, diethyl ether and n-hexane were purchased from Caledon Laboratories (Georgetown, ON) and were of distilled in glass GC or HPLC grade. De-ionized high purity water was produced on-site by reverse osmosis and subsequent filtration using a Milli-Q® water system

(Millipore, Bedford, MA).

Blank human plasma, and breast milk were obtained from Transfusion and

Lactation Services at BC Children's and Women's Hospital (Vancouver, BC).

Glassware for sample preparation, extraction and storage (15 mL Pyrex® disposable culture tubes and polytetrafluoroethylene (PFTE) lined screw caps)

Page 203 were obtained from Corning (Corning, NY) and/or VWR Canlab (Vancouver,

BC).

Disposable needles and plastic Luer-Lok® syringes (for drug administration and sample collection) were purchased from Becton-Dickinson (Franklin-Lake, NJ or

Mississauga, ON). Heparinized blood gas syringes were obtained from Marquest

Medical Products Inc., Englewood, CO; green-top heparinized Vacutainer® tubes, purple-top K2 EDTA Vacutainer® tubes, and red-top Vacutainer® tubes without additives were purchased from Vacutainer systems, Rutherford, NJ.

Equilibrium dialysis cells (1.0 mL capacity) were made in-house using Plexiglas® material. Cellulose dialysis membranes (molecular weight cutoff 12KDa) were obtained from Sigma Chemical Co. (St. Louis, MO).

Pooled human hepatic microsome and cDNA-expressed single human CYP isozyme microsome preparations were purchased from Gentest (Woburn, MA).

Normal and -induced rat hepatic microsome preparations (from male Sprague-Dawley rats) were obtained from In Vitro Technologies (Baltimore,

MD.

Page 204 4.8 Analysis of clinical study samples

4.8.1 Determination of fluoxetine and norfluoxetine in human serum and breast milk

Fluoxetine and norfluoxetine isomers were measured using the GC/MS/EI method

developed in our laboratory (see section 2.3). In order to address potential errors

in serum sample collection and processing in the delivery suite and outpatient

laboratory (plasma sample collection [i.e. use of purple- or green-top Vacutainer®

tube in place of a red-top tube] in place of serum, and prolonged clotting time in

excess of the desired 30 minutes at ambient temperature), the following

parameters were evaluated.

a. Comparison of serum and plasma concentrations of fluoxetine and

norfluoxetine isomers: whole blood samples (-20 mL) were collected from

two volunteers. These whole blood samples were pooled and spiked with FX

and NFX (final concentration 50 ng/mL for each isomer). The spiked whole

blood sample was left for 10 min to equilibrate at room temperature and

divided into six 3 mL aliquots: 2 in purple-top (K2 EDTA), 2 in green-top

(lithium heparin) and 2 in red-top (no additive) Vacutainer blood collection

tubes. One sample in each of the purple-top and green-top tubes was

centrifuged immediately, and the plasma separated and frozen at -20°C until

analyzed. One sample in the red-top Vacutainer® was allowed to clot for 30

min and the serum then processed in the same manner.

b. Prolonged sample exposure to ambient temperature: The duplicate samples of

whole blood samples in the purple-, green- and red-top tubes were exposed to

Page 205 ambient temperature for 2 hours. These samples were subsequently

centrifuged, and plasma/serum separated and stored at -20°C until analysis.

Fluoxetine and norfluoxetine concentrations in these samples (EDTA plasma,

heparin plasma and serum) were compared.

4.8.2 Determination of paroxetine in human serum and breast milk

Serum and breast milk paroxetine concentrations were determined using the

GC/MS/EI and/or GC/MS/NCI methods developed in our laboratory (see sections

2.6 and 2.7). Similar to section 4.8.1, PX concentrations (100 ng/mL) in serum

and plasma samples were compared. Stability of PX (100 ng/mL) in

serum/plasma following prolonged exposure to ambient temperature was also

assessed as described in section 4.8.1.

4.9 Plasma Protein Binding study

4.9.1 Preparation of reagent solutions and spiked plasma samples

Isotonic phosphate-buffered saline (PBS) (pH 7.4, 0.067 M) was prepared by

dissolving accurately weighed 1.8 g of mono-potassium phosphate (KH2PO4),

7.4 g of di-sodium phosphate (Na2HP04) and 4.2 g of sodium chloride in 1.0 L of

deionized water. The pH of the solution was adjusted to pH 7.4 by adding a small

amount of sodium hydroxide solution (1.0 M) drop-wise. The FX and NFX

spiking solutions used for the protein binding studies were prepared in PBS.

Blank human plasma was obtained from the Red Cross Transfusion Service

(Vancouver, BC) through the Transfusion Unit of BCCWH. All batches of blank

Page 206 human plasma (heparinized) were checked for potential chromatographic

interference in the GC/MS analysis method (Section 2.3). Five batches of human

plasma were pooled in equal volume and used for the protein binding study.

Blank plasma samples (heparinized) were spiked with three levels of FX and/or

NFX (100 [low], 1000 [medium] and 10000 [high] ng/mL each, respectively).

These samples were incubated at 37°C for 1 hour. The stability of FX and/or

NFX in plasma during the incubation period was evaluated prior to the plasma

protein binding study.

4.9.2 Equilibrium dialysis procedure

For the in vitro protein binding study, human plasma was spiked with appropriate

amounts of racemic FX and/or NFX (100, 1000 and 10000 ng/mL) as previously

mentioned. The total volume of spiking solution (in PBS) was less than 5% of the

total plasma volume. These spiked plasma samples were incubated for 1 hour at

37°C. Furthermore, spiked plasma samples were incubated in equilibrium cells

for 1, 2, 3, 4, 6 and 12 hours to determine the stability of FX and NFX over the

potential equilibrium period.

The cellulose dialysis membrane (12KDa molecular weight cutoff, Sigma

Chemical Co., St. Louis, MO) was boiled in distilled water for 30 min and then

soaked in PBS (pH 7.4) for a minimum of 1 hour. The membrane was

subsequently mounted into the dialysis cells. Care was taken not to touch the

surface of the membrane that would be in contact with the test fluids (PBS,

plasma). Equal volumes (0.9 mL) of the physiological buffer and study fluids

Page 207 (plasma or control containing FX and/or NFX) were added to the respective sides

of the dialysis cells, separated by the cellulose membrane.

To determine the optimal equilibration time, the cells were dialyzed, in a

circulating water bath at 15 rpm, for 1, 2, 3, 4, 6 and 12 hours at 37°C. Before

and after equilibrium dialysis, both pH and sample volume were measured in

order to determine any changes. The free fraction was calculated using the

following equation:

Free fraction = C„ / Cp (Equation 4.1)

where Cu and Cp are drug/metabolite concentration in buffer and plasma after

dialysis, respectively. Aliquots of the plasma and dialysate samples were

extracted using single liquid-liquid extraction and analyzed by the GC/MS

method (see Section 2.3).

4.10 Preliminary in vitro studies of fluoxetine metabolism in human microsomes

The in vitro metabolism of FX was also determined by following the

stereoselective N-demethylation of the drug (/'. e. formation of NFX) in pooled

human and rat microsomes (Gentest Corp., Worburn, MA). In addition, the

human CYP isozymes involved in the N-demethylation of FX were determined in

Supersome™ (microsomes produced from an insect cell line transfected with

Baculavirus containing a single human CYP isozyme with supplemental reductase

(Gentest, 1998)): CYP-1A1, 1A2, 1B1, 2A6, 3B6, 2C8, 2C9, 2C18, 2C19, 2D6,

2E1, 3A4, 3A5, and 4A11. Final 1 mL incubations contained 2 mg microsomal

Page 208 protein, 10 mM magnesium chloride and 1 mM NADPH in 100 mM phosphate or

Tris buffer (pH 7.4) were prepared according to the manufacture's instructions.

After a pre-incubation time of 3 min at 37°C, the reaction was started by the

addition of substrate (100 ng/mL of FX isomers equivalent to free base), and

allowed to proceed for 30 min thereafter. At the end of the incubation period, the

reaction was stopped by the addition of 100 pL acetonitrile, and immediately

frozen at -20°C.

4.11 Data Analysis

Maternal and umbilical serum concentrations of the analyte(s) at birth were used

to estimate fetal exposure during pregnancy. The ratio of umbilical-to-maternal

serum drug/metabolite concentration was calculated and used to estimate the

fetal-to-maternal serum (F/M) ratio. In the case of FX-exposed pregnancies, the

ratios for both the individual FX and NFX isomers and the racemic mixture were

calculated.

In addition, the ratio of neonatal (at the time of PKU testing)-to-umbilical

drug/metabolite concentrations (I/F ratio) was calculated to determine neonatal

exposure and drug elimination capacity immediately following birth.

During the nursing period, the milk-to-serum/plasma ratio was calculated by

dividing the breast milk analyte concentration by that measured in maternal

serum. The daily dose of FX, NFX and PX received by infants was estimated as

follows:

Page 209 Infant dose (weight-normalized) = CmukX VmiikX FBF

where Cmiik is the analyte concentration, Vmiik is the estimated volume of milk

ingested daily by the nursing infant (150 ml/kg/day) (Pons et al, 1994) and FBF is the percentage of breast-feeding reported by the mother. The dose was compared to the maternal dose, normalized to maternal weight, to calculate the relative infant dose as follows:

Relative infant dose (%)= Infant dose (jug/kg/d) /Maternal dose (/ug/kg/d)

Data are reported as mean ± SD unless otherwise indicated. Statistical analysis was performed using nonparametric tests (Wilcox paired signed rank, Mann-

Witney, Kruskal-Wallis and Spearman correlation) and parametric tests with appropriate post hoc tests. The level of significance was chosen as p< 0.05 (two- tailed). Theory and formulae used for statistical analysis were obtained from Zar

(1984). Prism®, version 3.0 (Graphpad Inc., San Diego, CA) and Excel 97 with

Analysis Tools® (Microsoft Corp., Redmond, WA) were used for data processing and analysis.

Page 210 4.12 Results

4.12.1 Determination of fluoxetine/norfluoxetine and paroxetine in clinical samples

Comparison of serum and plasma concentration of the analytes: The analyte

(fluoxetine, norfluoxetine and paroxetine) concentrations in serum and plasma

prepared by different methods were compared. Serum samples (prepared in the

red-top Vacutainer® tubes) and plasma samples (heparinized in the green-top and

EDTA-treated in the purple-top Vacutainer® tubes) were prepared according to

the procedure described in sections 4.8.1 and 4.8.2. There was no significant

difference in the FX and NFX concentrations among the whole blood, serum and

plasma samples (<10%) [Table 4.1]. Similarly, no difference in PX

concentrations was observed between the whole blood, serum and plasma samples

(<10%) [Table 4.2].

Stability of the analyte in clinical samples during hospital laboratory handling:

The stability of the analytes in whole blood was evaluated to address the potential

delay between sample blood collection and sample processing in the hospital

laboratory. Whole blood samples spiked with the analytes were stored at ambient

temperature in corresponding Vacutainers for 2 hours prior to appropriate sample

processing. No significant changes were observed in the concentration of FX and

NFX in the whole blood, serum and plasma samples (<10%) [Table 4.1].

Similarly, no significant change was observed in the whole blood, serum or

plasma concentration of PX (<10%) [Table 4.2].

Page 211 CN ~H CN IP a. CO CSI OJ N. X d T— •H +1 +1 •H < CO CM •tf o m d d d d 1 •tf •tf s to to §- t -H TJ 2 I*- cu P iq lO o O co d iri X d OJ OJ fi +1 •H fc +i •H CO OJ s CO li. cc d (V d od •tf lO •tf

^_ CO CD fi T— X CSI cu fi -H -H +i •H on 5 r- CO fc CM 5 CO 00 d d d a -H •tf •H I m CO O tf tf 03 & CO ca s 00 d CO OJ OJ £ tf CO r- a CO CO CM T-' CM CO CO P. X a. •H -H +1 +1 c OJ CM in O) cu c T— T— X T— T- CD m •H •H +1 M CM .c fi fc & c •tf r- 00 CO (r\ 1 CD 1 d d d d CO t CO CD & tf tf tf £ OJ cm CO CD c CO o CO 00 m 3 - 43 0- E CD T • X d +1 -H fc •H -H J3> CO < CM CO co rv cu d CO d 00 CD •tf m •tf .fi tt tf cu TJ X O CO tf O CM csi X T~ -Q •H ro +1 £ fi -H ro ro CL, CM "£ ~c fc. 00 c o TJ CO +H o O p CO C d O TrlO- m -S2 o CM KJ d X •r^ fi +1 •5 +1 JO •tf c CM c T3 d T— o 1 fiCCJ c in in 3C3 o CJ fi o CJ O o fc? 3e3x CO "ex ex • ine c o Q. CO 03 CO 2 fc? o '= CD O .c CJ i_ z> c o OH O CM O c O c o § o s: -C s CJ o o e a CD a o CM o CM 0. LL. a Oo c3

fN

CW cu 3« H H 4.12.2 Demography and medical history of the patient population

Perinatal period

Fluoxetine-exposed group during pregnancy

Simultaneous umbilical cord and maternal blood samples were collected from 8

patients exposed to FX during pregnancy. Demographic data for these subjects

are presented in Table 4.3. The average age of mothers was 32.6 ± 4.6 years with

an average weight of 74.1 ± 17.6 kg. Six subjects were taking a daily dose of 20

mg, one subject was taking a daily dose of 10 mg, and the other was on a 30 mg

dose. The indication for FX therapy was a major depressive disorder in all the

patients [DSM IV]. Two of the patients were taking concurrently

with FX and one was also receiving amoxicillin. One patient (F8) smoked

marijuana (THC) occasionally. The average maternal exposure period for FX was

8.9 ± 15.9 months (range 1 to 48 months) at the time of delivery. None of

patients in this group was smoking cigarettes or cigars. Similarly, none of

patients consumed alcoholic drinks during pregnancy. Five of the eight patients

drank caffeine-containing beverages such as coffee, tea and cola in a moderate

amount. Pregnancy-related diabetes and hyperglycemia were reported in two

patients (F2 and F6, respectively). The mean gestational age and birth weight of

the infants was 38.6 ± 1.3 weeks and 3.43 ± 0.28 kg, respectively. Two (F2 and

F8) of the 8 infants were admitted to the neonatal care unit for 1.5 and 4 days,

respectively, due to respiratory problems.

Page 213 CN o o CO in o o m o °> o CO - • x: a> 00 CM oo CD CD m CO CM Ti• Tf CO o CO T— CO CN CD 00 00 00 00 CO CO CO 00 a. CD ro CO ™ o 3 a" CO a> S O CO 00 T- CO 00 ' CO CO 00 CO Tf 00 CO Tf CO a> CO ^ CD

£ x cn cu 5 2 2 LL LL

T3 O T- O CM _o " T3 °G u o atu, O 'ea o Y o S OOOOO O OO in rt .s < oo u cu — o cu

- T3 O n Xl l- _ . f-r cu D _4 CU o> l_ CO » a. C TJ CU >. CD _g X CU '-4—» .82 > '= CU c a- O "O [/J

_ o C2 OOOOO o O O o co 3 OOOOO O O co =5 2 2 5 2 2 Q

CU 5.-0

cn o cu E E X3 E CO o c-— +-» CO a. Q. ~.2 c S-l 3 .9 co cu co cu CJ O X - N ' - • X— O.*; o c o CO c o E c E E o CJ CO o 3 o ° s e CJ ou o o

00 CM m CM 05 ^ rr, TJ- od « 00

£-3 SB «

OOOOO O OO q « « o CN CM T- CM CM CO CMCM ° iri 00 D 10 i- CM O 111! CO T- CO O) oo o oo CD in •S? 00 s .2> co co m co o) ^J * S I inr m o .s; Q c co " rt ° a. & 1) -~- CO O) T- N CO 00 Tt O rt ^ X) co CO CM CO CM 00 CM c 00 Tt ~ b rt Tt « c .c o c K 00 2 ^ -S '53 XI CM co -rt m at 3 C Q o O }*3 CO H Paroxetine-exposed group during pregnancy

Maternal and umbilical cord samples were obtained from 20 patients who were taking PX during pregnancy. Demographic information for these patients is contained in Table 4.4. The mean age of the mothers was 31.0 ± 3.9 years with an average weight of 73.9 ± 12.1 kg. Nine subjects were taking a daily dose of 20 mg, four subjects were taking a daily dose of 10 mg, and three were on a 30 mg dose. The other 4 patients were taking a 5, 15, 25, and 40 mg daily dose. Twelve patients were taking clonazepam concurrently with PX, which represents 60% of patient population in the group. The indications for PX therapy (with concurrent clonazepam in some patients) were a major depressive disorder (MDD) [DSM IV] in 15 patients, a MDD/panic disorder in 4 patients, a MDD/eating disorder in one patient, and a MDD/obsessive compulsive disorder/panic disorder in one patient.

The average maternal exposure period for PX was 6.3 ±10.1 months (range 1 to

48 months) at the time of delivery. Occasional use of acetaminophen (Tylenol®) or acetaminophen with (Tylenol® #3) was reported in two patients (P3 and P15) prior to delivery. Use of antibiotics (amoxicillin) was reported in one patient (P16) for the treatment of urinary tract infections. Diphenhydramine and

Materna® (multi-) were also reported in one patient (P20). Concurrent use of trazodone and clonazepam in one patient (P13) was also reported.

Symptoms of moderate migraine were reported in two patients (P14 and P17).

Unlike the FX patient group, 3 of 20 patients were smoking cigarettes (from 1 to

10 cigarettes a day). Similarly, 2 of 20 patients reported occasional consumption of alcoholic drinks (1-2 drinks per week).

Page 215 CN

LO CO O 00 * O lO O O CO a. o o o o o O CO cm in CO CO CO CO 00 h~ ^— cm CO t- CO co o CD O •tf 00 CO o CO o o 2 tf cm o 7— CO CNtfI O cm •tf CM CO CO CM r-- CM CO CO CO CO cm co tf co CM tf •tf CO CO CO -tf 22 ^- CM in tf tf 3 «

cm o r- x— cm 00 l~- O cm 00 O CO rr— o cm cm X 1 CO CO CO CO CO CO CO tf O CL, co •tf •tf CO •tf •tf co co 0) tf tf tf T- CN Q

2 LL LL 2 LL 2 2 2 LL 2 LL LL LL LL •2 o»o 2 LL LL 2

S •-

(L> O m in C to i-OOOOOOCMOOO o o CO CD '5 8 r' CL "+H >, o V D to cd OOOOOOO tfOOOOOOOOO JS S 8 t 60 Z. < •a LH

1= o 00 o I i ooooooooooo E in 00 in O cu co E E c g o JB co CD CD 3 i- .Q C co o c C cu CD £ 0. i- 2 •e O) CO Z3 cm CL 2 >. I Q Q CJ o c cr CL Q Q Q Q Q Q O Q LU Q Q; Q; Q Q O Q Q; Q Q Q Q Q Q Q Q Q Q O t— CL o Q Q Q Q Q Q Q W ^5 2 2 2 2 2 Q 2 Q — o CJ o Q Q 2 2 2 a Q •—' CO u —. Q £ "3 g ao ll o E cd c c 00 s " CL O CO CD E E E E E E E CO CO E E E CD CD CD CD CO CO CO CD CO CD >• „ is. Q. CO• i /Tyle Q. Q. Q. CL CO• °s CO- CD CO CD CD CD Xi t-T E CD CD CD CD CD CO CD &i P__ CU N N CD N N N N N N ~ T3 CD CD r> CO CD CO CO CO CO N 2 u ii I & C c C C c c c c H o 1/1 O 2 O c u ze [ CO J? o 2 - a. o o o o o o o D.H3 _ 03 ° I up «.2 cE= o o o o u —c 3 m a O O O O 1_ o c D." COO o o CJ o 13 o o u- E " .. „_ 3> , oCJ CD ^ SQ cd u "- 3 £ f lO m ,„ w m oo bh ti cd •"-•CDmCOI"-'tfC0CNCNC\ICOC0 o T— -tf -tf -tf is- is a ^ O CM

T?P -U > td 2P e ri. ^ ° 2 5 2 c o o oooooooooinoooooiniriooo CO oo CM /I O Q CMCMtf-T-CMCNCMCMCMCMCOT-CMCOT-T-1" T-COCM .l-E S.i s s « <*H -a O X) -° COlOi-OCMI^CmcOOOOh-OCOOCOO CO 00 cm i- _ <= "S c § •3 rH to 22 o -2 U L- cd U CJ CL CU COtf-inOCOCMCOI^-tfCMCmCOCMCOCOCmr--<-CO'<- cm o Cd CL~ CMCOCMCOCMCOCMCOCOCOCMCOCOCOCOCMCOCOCMCO CO CM id E I IS to g c 2 > s Id c g r-Nn^incosoroOjNnjmcoNcoBo CO Q § 5D 60 " tu Q.&.Q.Q.Q.Q.Q.Q.Q.Q.Q. | CO .Soc co •Sf 1 o o| In patient Pl 1, consistent alcohol consumption (~4 drinks per week) throughout the pregnancy was reported. Moderate caffeine intake (0.5 to 2 cups a day) in 5 patients and higher consumption (-3-4 cups a day) was reported in one patient

(P15). The mean gestational age and birth weight of the infants was 39.2 ± 1.4 weeks and 3.48 ± 0.52 kg, respectively. The birth weight of three of the infants was less than the 10th centile for gestational age (P7, P16 and P17). Three of 20 infants were admitted to the neonatal care unit for respiratory concerns (P3) and complications at birth (P10 and P15), with a length of stay of 10, 3 and 1 day(s), respectively.

Gestational age and birth weight among the FX- and PX-exposed groups [Tables

4.3 and 4.4] were not significantly different (p>0.05, Kruskal-Wallis) from those of the control group from the collaborative study (38.8 ± 1.22 weeks and 3.42 ±

0.44 kg; n=26), which were not receiving any antidepressant treatment.

Perinatal complications in fluoxetine and paroxetine-exposed pregnancy

The APGAR score was measured as a normal part of clinical care by the participating hospitals and data were collected for the present study. The mean

APGAR scores at 1 min were 6.8 ± 1.7, 7.0 ± 1.6, and 8.2 ± 0.9 for the FX- exposed, PX-exposed and control groups, respectively. Scores from both drug- exposed groups were significantly lower (p<0.05, Kurskal-Wallis and Dunn) than those from the control group. Similarly, the mean APGAR scores at 5 min were

8.5 ± 0.8, 8.6 ± 0.9 and 9.0 ± 0.2 for the FX-exposed, PX-exposed and control groups, respectively. In this case, there were no statistically significant

Page 217 differences among the 5 min scores for the FX-and PX-exposed and control groups (p>0.05, Kruskal-Wallis).

In the PX-exposed group, the effect of concurrent clonazepam therapy was evaluated. There was no significant difference in APGAR scores between clonazepam-positive and -negative patients (7.0 ± 1.6 and 7.0 ± 1.7, respectively) at 1 min. Similarly, no apparent difference in the APGAR score was observed at

5 min (8.33 ± 1.07 and 9.00 ± 0.00, respectively). No significant correlations were observed among the caffeine intake, smoking or alcohol consumption and birth outcomes and perinatal complications.

Postpartum period

Fluoxetine-exposed group during the nursing period

Twenty-six sets of serum and breast milk samples were collected from 21 FX- exposed mother-infant pairs [Table 4.5]; one subject declined the collection of an infant serum sample (FP); no maternal serum sample was collected in one mother- infant pair (FF). The average age and weight of the nursing mothers was 33.4 ±

4.6 years and 67 ± 12 kg. The average daily dose was 23.2 ± 9.5 mg (range 10 to

50 mg) with the majority of patients taking 20 mg (n=14 including patients whose dose was increased from 10 mg). The average maternal exposure period to FX was 7.0 ± 12.3 month (range 1 to 60 months). The indication for FX therapy was major depressive disorder (DSM IV) in all subjects. Six of 21 patients were also taking a (4 clonazepam; 2 ). One patient was taking trazodone in combination with FX. One patient (FN) was taking salsalate

Page 218 (Salflex*). The use of an oral contraceptive (Triphasil®) was reported in one patient (FS). Symptoms of arthritis were reported in two patients (FE and FJ).

Crohn's disease was reported in one patient (FN).

Two patients were smoking cigarettes moderately (1-3 per day). Six patients consumed alcoholic drinks during the week prior to sample collection (3 occasional social drinkers (1-2 drinks per week) and 3 moderate drinkers (4-6 drinks per week). Moderate caffeine intake was reported in 4 patients (1-3 cup a day). The average age of the nursing infants was 3.5 ± 2.7 months (range 6 days to 9 months) with weight averaging 6.7 ± 2.4 kg (range from 3.5 kg in 6-day-old to 9 kg in 11-month-old infants). Exclusive breast-feeding was observed in 15 mother-infant pairs and partial breast-feeding (range 50 to 90%) in the others.

Serum and breast milk samples were collected twice in 6 mother-infant pairs (FI,

FJ, FQ, FR, FS, and FT). In two of these patients (FI and FJ), the daily FX dose was increased to 20 mg from 10 mg between the two sample collection periods.

Paroxetine-exposed group during the nursing period

Thirty-one sets of postpartum serum samples from 24 patients were collected in

PX-exposed mother-infant pairs [Table 4.6]. The average age and weight of nursing mothers was 31.0 ± 4.1 years and 69 ± 14 kg. There was no significant difference in maternal age or weight between the FX- and PX-exposed groups (p>

0.05, Mann-Whitney). The average daily dose was 19.3 ± 9.3 mg (range 5 to 40 mg) with patients taking 5 mg (n=l), 10 mg (n=9), 20 mg (n=8) and 30 mg (n=8).

The average maternal exposure period was 5.2 ± 3.0 months (range 1 to 14

Page 219 months) at the time of sampling. Twelve patients were taking a benzodiazepine

(11 clonazepam and 1 lorazepam) concurrently with PX, which represents 50% of the patient population in the group.

The indications for PX therapy (with concurrent clonazepam in some patients) were a major depressive disorder (MDD) [DSM IV] in 20 patients, a MDD/panic disorder in 3 patients and a MDD/eating disorder in one patient [DSM IV].

Occasional use of acetaminophen (Tylenol®) was reported in 3 patients (PC, PF and PM). Use of an oral contraceptive (Triphasil®) was reported in one patient

(PQ). Symptoms of hyperthyroidism were reported in one patient (PD).

Moderate to heavy cigarette smoking was reported in 2 patients (5-15 cigarettes a day). Alcoholic consumption was reported in 7 patients (occasional social drinking to moderate drinking (3-4 drinks per week)). Caffeine intake (1-2 cups a day) was reported in 6 patients. There were no significant correlations among caffeine intake, smoking and alcohol consumption and maternal or infant drug disposition and infant adverse effects.

The average age of the .infants was 2.8 ± 2.3 months (range 15 days to 10 months) with a mean infant weight of 5.6 ± 2.3 kg (range 2 kg in 1-month-old to 10 kg in

11-month-old infants). Exclusive breast-feeding was observed in 18 mother- infant pairs and partial breast-feeding (range 50 to 80%) in the others. Serum and breast milk samples were collected twice in 7 mother-infant pairs (PJ, PN, PO,

PR, PV, PU and PY) and three times in 1 mother-infant pair (PT). In two patients, the daily dose of PX was increased between sample collection sessions

(30 mg to 40 mg in PN; and 20 mg to 30 mg in PO). Maternal and infant age and

Page 220 weight for both the FX- and PX-exposed groups were not significantly different

(p> 0.05, Mann-Whitney).

Page 221 CN CN CN

cu cu cu cu cu co cu cu cu cu cu > > > > > > > > > > > > 'in 'in 'in 'in ca ^ (/} (ft (A (A Ul Ul '(/> w 3 3 3 3 3 3 3 3 3 3 3 o O 3 3. 3 £ t _ o o o o CJ o O CJ O O CJ CJ X X X X X X o cj o «, ra co X X X X X X CO CD CD 01 cu cu cu cu cu cu cu 'in cu cu cu

xi 1*3 ai in in u OJ OJ io o> o 5. CO CO CO «!

XI XI Xt Xt

CO^O)COx>.^l30CIllf)COCM>. \S £ 5-3 C cOn

V. >, o V t/1 eg c/1 — •£ -s o -£ O O O O CO o C t- •3 '3 oi} ,

1> cc XJ o (D r- > O o o o o o o

E O.

io CD n to oo->oooooooo oi c c c

E

xi -a

C/l

QQQQQQQQQQQQQQCJCJCIQCIQQ QQQQQQQQDQDQDQQQQQQQQ oo e z s? w -6 3~ ! cl

I g | i T3 CJ

o .£ o. S It O 5 o \z o o cj — o o •o H i I o |> " o 2 >• ° .Q Is CO o 5 2 >. w u on c CO Q « Co" CD ca cj ea ^ S CoOp g -a n .a u % M o u- .£ oo c E OJ CN O o ooooooooo ^ to o S -o S o ° CMCMCMCNCNCNCNCNCM CN T-CNCM OJ OJ, (0 JJ § 4J 73 "P o O ° E is IS rg CO in CD un 2 S .2 lO CO ID I 'to* h-cocococomcoco •° g.i h~ in XI ' -5CO I 3 £ a x: i CJ cr> O CO CM CO CD T- CD o •. o. m w O) n m m CO CO CO CO 00 ' w -•-I" ca < U

OJ OJ OJ OJ 0) OJ CU OJ OJ OJ cu OJ cu cu cu cu cu 0) cu cu > > > > > > ' — 03 -rr — > > > > > > > > > > > > '(/> 'to *C0 '(/) 'to 'to t—o 03 » « s ss SS .H 'to 'to 'to 'to 'to 'CO 'co 'co .£5 CO 'cO 'to 'to -j n =j t C 3 3 3 3 3 3 3 "3 3 3 3 3 OOOOOOO S 5 o u o m co •0c3 O O X X X X X X X °- °- X3 .X3 JX3 O O ^ oX Xo Xo Xo Xo Xo Xo Xo CL X Xo X oX OJ OJ OJ OJ OJ OJ OJ OJ CD OJ cu cu cu OJ CU OJ cu cu cu cu CO cu

CD C\l CD o 5 O) oo jo _ro ra N w co ro ra CO ^ oi to to o S * Iti c c c CN C C CO C "E ra ra ra o ^! IL 5 IL 5 IL IL 5 LL IL LL LL LL LL LL LL SH§ -o

Xi o >! o s « LH •c cd O £ Q S CL. cu c C "H S~ o 8 ^ i=5 § J^ o tu

E E E E E E E E E E E E E c E 03 — 03 ro ro ra 03 03 C • o CL CL CL CL roCL C L roCL roC L C L O C CD CU izz I 2 CL OJ OJ ro cu CO o oj cu tu cu ^ a a? JU N N N N N = 8 c OJ t CL N N c N N J3 co N r>*ro ro t o P > ° ^ cu ~ ! ro tu cd LO Tt* S CO S _ ro « to ro CD 00 CO CD u"& c c o c ^j^i>,-gCNicoi~-coco»irooi-- c c V ., u '3 td cu co X " o cs ra orj — c o o o o o oooooooooooo ^ " i~- 2 ci o ° T-T-CSICMCOCOCOCOtf'CMCOCN 2 3 E CJ CO CM CD TI "O o. g. 1 - « S C3 cwn «2 c 15 LH 2 -o 00 CO CD §•! -a c S o •a e cu o .2 E i2" Si"! £ a c cu cn ^. M .« cd c £ E 3 Q • - CL •=8 8 oo_ u s i .Soc ro si ro JO i cd co -S )«> < CO C_> Q LU '5 LL CD _ -> S£ O Lt » I- D D > •> X > ° p, fa CL CL CL LL Q_ S CL CL CL CL CL Q. D. CL CL CL CL CL CL CL CL E cd 5zz66i co < O 05 CL Q. CL 0. CL CL H 4.12.3 Fluoxetine and paroxetine disposition during the perinatal period

Fluoxetine exposure during the perinatal period

Fluoxetine (FX) and norfluoxetine (NFX) concentrations in maternal and

umbilical cord serum sample collected at birth are shown in Table 4.7. The serum

concentration of SFX was higher than that of RFX (i.e. S/R ratio higher than

unity, p<0.05, Wilcox, n=7) in maternal, umbilical cord and infant samples except

for the maternal serum sample from one patient (F8). No detectable levels of the

FX isomers were observed in umbilical cord serum sample of newborn F8 whose

mother was receiving a 20 mg daily dose. Similarly, the maternal, fetal and

newborn S/R ratio of NFX was significantly higher than unity in all samples (p<

0.05, Wilcox, n=8), except for the maternal sample from patient FI. There

appears to be no correlation between maternal dose and serum FX concentration;

however, proper statistical analysis could not be performed due to an insufficient

number of patients with varying doses (i.e. only 2 patients were receiving other

than a 20 mg dose).

The mean ratios of umbilical cord-to-maternal serum drug concentrations (F/M)

were 0.87 ± 0.33, 1.11 ± 0.27, 0.82 ± 0.20 and 1.15 ± 0.15 for RFX, SFX, RNFX

and SNFX, respectively. Similarly, the mean F/M ratios of racemic FX and NFX

were 1.04 ± 0.29 and 1.04 ± 0.16, respectively. There was a significant

correlation between maternal and cord serum concentration of racemic FX

(r=0.96, p<0.01, Spearman) and NFX (r=0.93, pO.Ol, Spearman) as shown in

Figure 4.1.

Page 224 Cord serum and neonatal serum concentration of FX and NFX at the time of PKU testing are shown in Table 4.8. The PKU test was performed on the neonate at an average of 48.3 ± 12.1 hours (n=7) following birth. Similar to maternal and cord serum, there was a statistically significant difference between the respective serum concentrations of FX and NFX isomers in the neonatal PKU serum samples

(P< 0.05, Wilcox).

On average, infant serum FX and NFX concentrations at PKU testing (~2 days postpartum) remained similar or were slightly elevated from those at delivery.

The average ratios of infant serum at PKU testing to cord serum (I/F) were 0.99 ±

0.28, 1.01 ± 0.39, 1.24 ± 0.26 and 1.18 ± 0.30 for RFX, SFX, RNFX and SNFX, respectively. For racemic FX and NFX, these ratios were 1.18 ± 0.26 and 1.00 ±

0.34, respectively. There were no significant differences (p>0.05, Wilcox) between umbilical cord and neonatal serum concentrations of FX and NFX or their respective enantiomers.

Page 225 IP a.

CM CO CD O CO Tt o o oo CO in in o o T— in JJ •r- Tt T— co _1 iri CO q CD CD 00 X> CM Oi CD CM m •cr 00 T_ d x— d d

DO in •

o co in oo CM N- O t CD 1--- o m in CD CM CM CM xi iri cd CM CD T- CD _i 00 iri co CD 00 in in to oo t— T— T— CD d d d T~ d d d

o CM h- 00 i— OJ CM in o r-- oo t— OO CM o CO a> CM K) m oo CM CM oi iri iri to CM q CM cn CM CM q t— ri Tt CM N- oo N- CM CD in d

30 co N- h- h- eg co CO r> N- oo .o O in Tt CN oo co |CM O OO :N CO Tt CM 00 Tt co iri J0 CD CD 00 fO CM T- r- T— i— T— T~ CM CM CM co d d d d d d 13

T— CO o CJ) CM CO N- Tt Tt CO 00 X) CD o CM Tt CM cn Vn 0 t~- ca oo 35 00 LO CD iri d Tt CO T— oq in a> o T— EM «*? X> T— o CM m co CO Tt CM Tf Tt CM "5. CL r- CO CO 00 o o 00 Tt Tt 00 co O Tt CD N- OO T- in T O X) Tt ci ^-c CM Tf Tt d d ^ S CCMN oo - cn o Tt CD r-~- CM o> CO CO m CO T- CO CD CO CO c CTl •* o CO ,_ O o o oo O) T- CO CO CM Tt JO CO 00 Kmscoo'-on d T-C O CO Tt d T-C T— d d ScDcocqcMqcqq 00 T— Tt CM CO CO M^ri^^cvicNO « CL 9 *& » oq CO CD Tt 00 N- q 00 00 o o co Tt oo co in CO 00 xi cri 1— iri ai CM it E C N CO CO N N CO *~ CN T— i£ - 2 lO CM T-• T-• Tf CM CM v- a- o K co <4-i CC co 5 O w j> r- r~- in co 00 O Tt Tt 00 CD st CM •

CN ,_ Tt oo ,- 00 CM OO 00 CD oo pooooooo co CM iri O d CD iri Tt co ai CMCMT-CMCMOOCMCM I £ CM T— CO CO CO CM T— CM 8^ JS §

c c ca CD t— CM CO Tt in CD N- 00 a> cM co Tt in CD N oo CU Q c3 Q JL LL LL LL LL LL LL LL E aCO c t LL LL LL LL LL LL LL Eco c r-(N00Mf!OtOO) inT-oi-incot- co oisscncoNr--g iridirir-i'tfddiri CD 00 O cS oo CO CD T- -^j- CM CO CM ci co d d ci T-1 d nscMtNiocosin •tf in oo 00 CM CO tf" CN N CO COC O CD 00 CM oo*doirido) •tf ri CO tf OO CO T- T- in a> T— T— T— O O T— T—

STJ-iNOOONNO n co s lOCOOOtfCOOOCD CD OO (3)SinO)CONT-'g dstf'tfoioiri-i ^ CN co 00 CN r- CM -tf CO CM CO d T-^ d d d d ^

'tfCOCO'tf'tf'tfC»C& O CM 00 CO CM N co s s co ra OJ 00 a> m i- OO CD CO 00 CO 00 "£ 00 CM ci (5 cz> d inoomoooocjoco o -tf T- CN CO CO CD CD0 0 O 00 CM O) N CM CO O) ffldSNCOCONCD T- CO co -tf T- CO T-CDCMCOCMV-CMCM -ST T-T-T-OOT-I-O

COOlNCNCOOCOtN lO CN cn OJ CO CO •tf CD 00 CO CM CM ai N cb ri ^ N CN ri CM h- CN o CD co co oo T-T-T-CNCOCNCOT- T- T- d T- d

SCMCOCOOnNO tf o r- T- T- m O h- O CM * CO s OJ u* : : 00 cj>T- 'd h^r~-<-cci—i id CD T- O O CD CO O CN • C3 •tf co •tf CO CM CM ^ LO tf" co CM cn T— CD CM CQ Q. CL. cqio-tfioco'tfoqco •tfomojmcoooo * CM CO ra siosiririiN^^ Si? Otfcoooocooocn H—» CO CNWCOOJOOCNO'tf tf ^CO^CN'f-tfCOT^ o co in Q. 2 c ^ -is S h-tf-OOi-OOCNOO OO CO O CN -tf CN OJ CD c d tfioioiridddJ CO o COOCOT- oor~iooo"g CM • r~- T- i*. T- m CM to CO CO o tf" Csi •tf T— T— CM a,

•tf co r-- r-h-tf T-cootf-cn ocqmcocNT-No "2 fr 00 ai in T-T-incNcoiqT-;a) oo [o id LO CO CM CD ^— CD TJco-*-ridririi-: T- CQ co N eg co ^ e Cw c+3 OCNh-OOi-CMCMLO o LricNNCNlOlONS CO CN *H a T-fl;CMNNO)NCN co in p .E ,CJ U3 J4^

OOCOSSNtNCOCO in s co CO T- CDOT-CDN-IOOOT- cNcO'tfCNob^-'tfri cd in 00 CN 'pin'tf inco'tfcoh- T-T-T-CMV-CMCMT- If OJ o -o « c c CD §1 r-CMntinCDNCO CD CN co -tf in CD h- co CU Q 3 co LL.LL.LLLLU.LL.LLLL. CO Q LL LL LL. LL LL LL LL E co c E CO C Paroxetine exposure during the perinatal period

Paroxetine (PX) concentrations in maternal and umbilical cord serum at birth

(cord) and in neonatal serum at the time of PKU testing are shown in Table 4.9.

There was a significant correlation between maternal PX dose and serum concentration (r= 0.56, p<0.05, Spearman). The mean ratio of umbilical cord-to- maternal serum concentration (F/M) is 0.40 ±0.16 for PX, which is significantly lower than that of FX and NFX (p< 0.05, Kurskal-Wallis and Dunn), which are

1.04 ± 0.28 and 1.04 ±0.16, respectively (Table 4.7). Similar to FX and NFX, there was a significant correlation between maternal and cord serum PX concentration (r=0.95, pO.OOl, Spearman) (Figure 4.1).

Neonatal serum concentrations of PX in the PKU test samples were only measured in 5 mother-infant pairs. The PKU concentrations were not available in the remaining 15 infants due to insufficient sample volume or because samples were not collected due to a change in protocol by other researchers involved in the study. The PKU test was performed at an average of 31.2 ± 11.5 hours following birth (for the samples analyzed), which is shorter than that in FX-exposed group

(48.3 ±12.1 hours). The average ratio of infant serum concentration at the time of PKU testing to the umbilical cord (I/F) was 0.53 ± 0.03, with the PKU concentrations (I) being significantly lower (p<0.05, Wilcox) than those determined at birth (F). This value is also significantly lower (p< 0.05, Kurskal-

Wallis and Dunn) than the corresponding ratio for FX (1.0 ± 0.34) and NFX (1.18

± 0.26).

Page 228 CN •g tN

a. _g tu oX 53 a, o on 3

a CD T- CD T- CM re ro ro re nj TO TO re re tu o -5 •S5 -5 -5 o E ~ CM CM Tf -si- CM "c "c "c tz c ~C "c c "c c •J= CO iB

(U a o £3 O) CD m Tt CM jo re re re re ja TO re re re CO Tt in io m m o uo O "c ~c "£ c c tz c "c iz "c o ooooo o

CU CD CD CO CO CO N 00 o UO 00 oo 00 CD UO CO O) Tt oo CM TJ- CO Tt CO CM CM uo CM CM CM CO oo UO PO CD CO CM OO CD O d d didid d d d d o' d d d d d d d d E O > _CD ca ca ca re ca ca TO TO CO CO ca ca ca T- N c a, o c c c c c c c c c c c c E 13 a o CD tu CO UO Tt d V CO T- CO PO oo o CM 00 Tt CM J -i C\i oi CO CD Tt CO Tt UO 1,1 T d d d cri csi T- !*3 a Sc5 c3 d 5 cIu — o UO UO CM T3 CO IO IO oo cq o CO uo oo oo PO uo UO • y— CM UO OO CD CO d CM CM Tt CM T"~ CO crTt i Tu—o "o o ao

LL LL u. LL LL LL LL LL LL LL 1 s i2 cu U.IL LL 5 5 5 5 S5 CD © CL ,' CL S * tU TJ ca O UO O O UO O O CO O °) o o ooooo o o o o o ^ oo CM CM CM PO CM UO CO CM °> od CM O Q E CM CM co a o c CD ca .2> cn OO UO PO O O O O CD o PO o CO -5 CO 3 LM CO Ol OO CD 1— 00 1— 1— UO tz OO CO CM -a CD r-~ r- N T- I- JB X) tu ca CO CO UO O 00 CM N Tt CM OO co CM 00 CO oo 00 °> CT) o ca c CO PO CO CO CM PO PO PO CM PO 00 CM CO ° CO CM > ca < oo CO ca o

c o C s O CM PO t UO CD h- Ot T- CM PO -rt UO CD I - 00 OO oo o .SP CM r! If 9 CL ti• a. a. CL CL CL CL CL CL CL 0. a. ^ 'cn 3 C * C/O as "TT CU 0 20 40 60 80 100 120 Maternal [Drug] - ng/ml

Figure 4.1 Correlation of maternal and fetal (umbilical cord) serum concentrations of FX, NFX and PX at birth (n=8 for FX/NFX and n=20 for PX).

Page 230 4.12.4 Fluoxetine and paroxetine disposition during the nursing period

Fluoxetine disposition during the nursing period

Maternal and infant serum concentrations and breast milk concentrations of FX

and NFX isomers are presented in Tables 4.1 OA and 4.1 OB. Detectable

concentrations of both FX and NFX were present in all maternal serum and milk

samples analyzed. Figure 4.2 shows the mean maternal serum and breast milk

concentrations of FX and NFX at each dose level. The milk-to-serum (M/P) ratio

was 0.59 ± 0.33 for FX and 0.56 ± 0.25 for NFX (0.81 ± 0.47, 0.52 ± 0.28, 0.73 ±

0.31 and 0.50 ± 0.23 for RFX, SFX, RNFX and SNFX, respectively) (Table

4.10B). A maternal serum sample could not be collected from patient FF.

Increases in the FX daily dose from 10 mg to 20 mg in 2 patients (FI and FJ)

resulted in an increased concentration of FX and NFX in both maternal serum and

milk samples. As illustrated in Figure 4.2, there appears to be a significant

correlation between the maternal dose and the serum FX concentration (r=0.43,

p<0.05, Spearman), but not for NFX concentration (r=0.35, p>0.05).

There was a significant correlation between maternal serum and breast milk

concentrations of FX (r=0.672, pO.OOl) and NFX (r=0.546, pO.Ol) (Figure 4.3).

Significant stereoselectivity was observed in maternal serum samples (p<0.001)

with an S/R ratio of 2.91 ± 1.36 for FX and 1.98 ± 0.75 for NFX. Similarly, there

was significant stereoselectivity (pO.OOl) in milk samples with the S/R ratio of

FX and NFX averaging 2.91 ± 1.37 and 1.89 ± 0.98, respectively.

Page 231 Out of the 23 infant serum samples (postpartum) analyzed, FX and/or NFX isomers were observed in 11 samples. With the exception of two infant samples at 5 and 7 months (FN and FB, respectively), with a very low infant-to-maternal ratio (~3%), quantifiable serum levels of FX and/or NFX were observed in infants who were younger than 2 months old. Figure 4.4 shows the plots of infant-to- maternal (I/M) serum concentration ratios of FX and NFX versus infant age. The

I/M ratios of FX and NFX declined rapidly.

Assuming an infant milk consumption of 150 mL/kg/day, the estimated mean infant dose of total FX and NFX was 11.8 ± 9.5 u.g/kg/day (range, 2.2 - 36.8 u.g/kg/day) and 12.0 ± 8.2 (range, 2.8 - 32.2 u.g/kg/day), respectively. These values represent 2.9 ± 2.0 % (FX) and 3.6 ± 2.1 % (NFX) of the maternal daily dose (312.3 ± 116.1 u.g/kg/day, range 172.4 - 629.3 ug/kg/day) (Table 4.12B).

This results in a total infant FX (weight-adjusted) dose of 6.5 ± 3.5% of the maternal daily dose.

Page 232 on »o CN

oooooooo o O O a 03 tfU O 03 Q o o _l _i _i _i 1^

CQCQCQLllLuLULum ^ [D 11 01 CO O m C^CN^QQ C m OB L O H- IP a m ca CO ca CO PL, o O O a a 03 CO r- uo 03 O C? 00 _J _j _i _l c oo csi Cvi co^-coH-cococacotf _i CO CO OX) m ca CO m ca c m ig cn oi fi

OJ OOOOOOOOH, .ro o O ro UO CM O O CO a O a _aJ a _I_I_I_I_I_J_1_|J; _i _l _i "c ci 6 6^ COCQCOCQGOCQCacO™ OQ ca CD CM ai OJ X o fi oooaaoooco co a a a a o ro o Ht i i i i i i i i -r= i i i i "c CM O CO a O CQCQCQCQCQCQCOCO" c CO CO CO CD CD _l ca c OuoOtNoaaa a a a o a 03 tf a q c _l _i _l 05 c CD CO B L OJ CQ CO 00 CO CO ca CQ CQ CQ CM —i r- _fi DQ uo 'nfi OJ X O OcoO rea a O a o ro co o ^ o tf -i -i a a —i g co 3 ta tf CO o _i _i _i _J c oj oi ^ d co If) CO CO CD c CO CO CO ca CO 00 c£ ca

O O CO uo oooooooo o o o c uo o uo CMm O o o o o o O o IO N S 1 o 't N N O CO CN 1^ o csi ai CO ai cd ui H- CM CMC O

cn CO CN O) cq cp tf OJ CO o JO tfC M CM o O CO uo Cn CO 00 CO CD CM CO CM CD UO CD O C r- CN O CO CM CD o uo tftf o io o CO cn CO 00 CO CO CM CN tf tf oCl o CO 00 o CO CO CJ) o CN oq O r-. £A CN O rotf o 1^ Cn l-~ CO tf0 0 U0 12 - OO CO 1^ to ai o CO oo o CO 00 cb CN CN ib oi cb • 1—1 o "c o CO o otf C O CM <0 CO CM 00 CD tf r-- CD CO CD H— 1— H~ H^ — o CN CO tftfC N CO CO CO CN o B o V oj 1) CL co uo m 1 00 tf LH •'. co CO O UO CM r-^ cb ib ib co tf it •5 ,„ ai m CM' 00 d d d d 3 c oo CM CM CD CO to to " -a co « tf ^ CM CD CO uo §; fi cn 5 03 I - CN cp 00 CO 00 CO co CM CM T— T— o U5 V cn i-^ O a ro CM ub itfb d ib d tf cb d 00 CD 03 oCL •tf cn cb ai ci LH uo CO H~ CM uo CM CD CN CN CM tfC N CN tf, — CM CM c OJ tf tf tf CO o JL> XI O CD CN 1^- JO CO CD cn (DT-IOIOOOICOS V tf ra tf tf tf 9 |H- d cb CN CD CN CM 2 to - CO 9 '3 1 CN c ai ai ooo C O o CtfO 0 0 N fi tf rZ tf CN tftfC N CM CtfM > • rH ^— —' a « ? to fi to v O CO S CO CO CD cn co CN 00 CO CM CO f- U0 CM 00 CO CM • I 03 E £ 00 cb cb cb cb I--' cb cb oi oj 00 00 csi oS id o 13 "> ? tf •tf roCM CO 00 CM CO CN CD CN 00 CN tftf CN CN CN tf OJ it

t/1 O) oooooooooooooo CMCNCNCNCNCNCMCMCM T- CN CM CM 00 CM B E 8 £ 5 Q_ ci CO CO h- VD a. UJ Q LL LL LL LL LL LL LL LL1 L L 1LL LL LL L_L1 LSL zLL LoL LL LoL L6L LL LL LL LL —Lr L L L S co c O 1—I H "O CN If

p S? SP s° ^ so sP sP SP NP SP vfl ifl -.O vp ^P -vP vO •

s s s p IO co IO 0 e> 0 o^otNSNioiDo1" ps ps CP 0 tf" ^nsqioat os as ds D m o CM is^^inTtTtcoiricriiricDcocMiricdcd

s 0" O^" o** XT S« S" V? >fl \P VP NP SP SP SP vfl yO ^p ^p ^p *0 s.O vQ in TC co N- Cflc>c>e>c>c>c>c>c>c>c>c>c>c>c>c>c^c>c^c^ovc^ ^T-CI)C\|T-COCN()lCncOOCnO)NO^COCONSCOO kt o o o iri ^in^riJiririco N^T^Nriin'ricvVdT-rCN

sO s9 c> S? S? os Tt cq ^T-oo)qqpjO)pcNicoojcocoo>oiOT-Ncoco ^ CM i— CTO- 1— Tt T^csi^idcbNcocdcococvidTJ^TJ^csicNcoiocD tO CM

:o T-OOf- .CDNCOIOOO^COCOOIIOOIOOICNCO o o co o 31 CO ,„ omio^qqcotNqNvqqTfconifiNiri^nnificD O CO m b> o o o' o OOOOOOOOOOO ^'OOT^O CO'O'OO a o M

o Tf cn oo T-omoonO'-^NNNscocooicoiD^co^ r> m ,„ co co co in ^incONNioiOSSr-lOCNl^T-LOS^CONSS n CM jo • • CN fc> o o' o o < c c c c o'o'ooooooooooo^^^ ^ ' ' ' ' 3 O

co T- o o m N CO n om * cn TT n CM J£> O CD O O T— o o oooooooooooo o o o o DO" o o o o

fc> rs. co is. co mO)SfiT-sinT-No^cONtN*C O O CO CO S PJ CO 3*OOS

r> in o co m -.oicOT-intoocvtiointomocMT-in^-ocoNco ncNcoN^^ovioiDin^vincoT-TttNTroinsTfcor-Nto ooooo cooooooooooood^000000° o d V o. co Tt i o cocoomi-coconT-oiocoMDST- T- in TT oo o>NcoqcNico(OLOcocoo)tocoTfino) K io do'oOOOOoOOOOOOO ' CM o d V ui o co co ^ r T qrcocNcp'-!co ". 'tscN'tiNiOTrcoin ?Y coHdoi^^co^dcddcNNtdcddTr M^^rSM^N^ST-COCOSOlh-in^ 1 CO o c

O V ts- u 8" o. fs TTTrO ;ycOCO^COcOCJlC)lcOCOCONNCOIOO) CL . S OVO hcoinco^JJconiopsfio^sm* O tj o

N. "S •- o, in O o CO TT TT T- O t» Tt V- :—,

TH ' CO CM IS- co r-. T- co N r- r- T- (O CO T- od in co co CM T- co CM m i- is. m .So-13

.•a ja J=

CNo'^'*co^^^CO^^^^I--^™<°'r:f900r--CNcoinTt fn Tt o P.

S c o 0)CJO)TrcqcocoiOTro)coN incqinh-^pTfpto CD T- CM o ca « CO CM o ll^coroiriTjoiricor^cTJcpg (rj CN iri iri co T- o TT CD CM OO CM CO CM T- CM in o CM Iv, o 3 1 ° la S> = CO -Q CO XS CO .Q CO .Q CQ * c kC CQ O D LL O I -L _L --> ^jSzOaOOciLiwtoLL LL LL u. LL LL U_ LL LL U. LL LLLLLLLLLLLLLLLLL-LLLLLLILL 350 Serum FX Serum NFX Milk FX r^l Milk NFX

20 30 50

Daily FX dose (mg/d)

Figure 4.2. Mean (± SD) fluoxetine and norfluoxetine concentration in maternal serum and breast milk at each dosing level [10 mg/d (n=5), 20 mg/d (n=17), 30 mg/d (n=3), 50 mg/d (n=l)]

250.0

200.0 r = 0.672 (FX)

E 150.0

r 100.0

50.0

oo 0.0 100.0 200.0 300.0 400.0 500.0 Maternal serum [Drug] (ng/mL)

Figure 4.3. Correlation of maternal serum and breast milk concentrations of fluoxetine and norfluoxetine in nursing mothers (n=27).

Page 235 0.9 n

-i 1 1 1 1 0 2 4 6 8 Infant age (month)

Figure 4.4. Infant-to-maternal serum concentration ratio of fluoxetine and norfluoxetine vs. time profile in infant exposed via breast-feeding (n=5 for FXandn=ll for NFX)

Page 236 Paroxetine disposition during the nursing period

For PX, detectable levels of the drug were also observed in all maternal serum and breast milk samples (Table 4.11). Figure 4.5 shows the maternal serum and breast milk concentrations of PX at each dose level. Similar to FX, there was a significant correlation between maternal daily dose and maternal serum concentration (r=0.58, p<0.01, Spearman). Furthermore, there is also a significant correlation between maternal serum and breast milk PX concentrations

(r=0.89, pO.OOl, Spearman) [Figure 4.6], which is similar to those of FX

(r=0.43) and NFX (r=0.35). The milk-to-serum (M/P) ratio of PX concentration was 0.46 ± 0.21 which was significantly lower (p<0.05, Kruskal-Wallis and

Dunn) compared to that of FX (0.66 ± 0.47) and NFX (0.56 ± 0.25) (Table

4.12B). The estimated mean infant dose of PX was 2.4 ± 2.0 u.g/kg/day (range,

0.2 - 6.7 u.g/kg/day), which represents 0.8 ± 0.5% of the maternal dose (Table

4.11). This represents a relatively lower (p<0.05, Kruskal-Wallis) exposure to PX compared to FX and NFX (2.9 ± 2.0% and 3.6% ± 2.1%, respectively).

In most infant serum samples, no detectable levels of PX were observed. Out of the 30 infant serum samples analyzed, detectable concentration of PX were only observed in 6 infants (range 0.1 - 0.7 ng/mL). However, the infant-to-maternal

(I/M) ratios were much lower (0.1 to 1.2%) than those observed in FX-exposed infants (2.7 to 49.9%).

Page 237 fO CO tf CO CM ^ODtNr;tNqcq03qtNCO^T- cqincrjCNtpCN^COincOCNlO O) oo m PP 10 i 0 0 0 0 d

COtD^-OSCMlNCMn ro co r- to T- co in oo CN o O) m co w ro co bg r— co co co CN TJ- tf tf tf CO 0>pO)CS[Tj--^lO'^r-r-l oi in q to co co S ^ T- CN cq csj cn co cn co O O O O oc^cNoovnTr'coTr c\i T- o r r 6 to ^ n co d CN r-' C) N" r- IX) CDF N - m

tf CO tf Ol CN CM CO CO CM O h~ O) to d d ci d ooooooooooooooo dddddddddddd

300000000 O Cf I I I I I I I I I • aoaoaoooaco a o o ,- a a a 000 0 0 'I i 00 CO CD OQ CQ CQ CO CQ CO C "=co co cacom m cu ID co co co ro ro ro rn rn

m n sr- CM CM CO m r- ^ ^. <°. "1 "i . i*. ™. I*- c*i "i . ™. • 0)-«coocMSr- oi^-j d ib rsi in cb - tfC D CO ' x> cb co cb cb cb eg T c c

S E 0 -^cs]conmo»-«)^^ |po p p cp cq Csj h-; ^ ocsicsiT r- ir> o> tf— . , o 5 ^^rodr^csiibro^T^cricvir^CNi^SS?: c\i r~: cb tf W fN tf cb ai Nln i cDincsimN*csiT-cos-T-(oj5 CM CO CO tf OD

II-LLSLLLLLLSJLLLLLLU-SLLSE U- U- LL. Li.

s O CO p CN p o Oi Oi CO CO I - T— CO (0 CO CO CO CO CO CO CO co co co to co co co co ^ CO CO CO CO ^~ tf) ro N N ^- wc c "c "c "c "c c c ^ cccccccc

i_ to

p o o o o 3 co ro CO Q E = a>

tf o co i— a> i_ > - S CDcnr-cocoRCMco 5 w

•* CO r CO r 05r-0)0>OT-mc0CD COCOO COCNCMtfO>O>a>COC0O)a>T- i- CN CM CM CN CO CO CO CM CMCOCMCMCOtfCOCMCMCOCOCOCMCOCOCOCMCMCNCOCOCMCMCOCOCOCCM O

n n co n co S3 o to £i ra n a o_ D_ o_ o_ P?>>r?CL>->- co • a. CL a. a. a. a. a. n a. Figure 4.5 Mean (± SD) paroxetine concentrations in maternal serum and breast milk at each dosing level [10 mg/d (n=10), 15 mg/d (n=2), 20 mg/d (n=12), 30 mg/d (n=8), 50 mg/d (n=l)]

50.0 -,

0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 Serum PX cone. (ng/mL)

Figure 4.6 Correlation between maternal serum and breast milk concentrations of paroxetine in nursing mothers (n=33).

Page 239 4.12.5 Plasma protein binding of fluoxetine and norfluoxetine

Stability of fluoxetine and norfluoxetine in adult human plasma

Stability of FX and NFX in adult human plasma at physiological temperature was

examined by measuring the concentration of the FX and NFX isomers before and

after incubation. Human plasma (heparinized) was spiked with FX and/or NFX.

The volume of spiking solution was less than 5% of the total volume to minimize

artifacts introduced by dilution of plasma proteins. Following spiking, plasma

samples were incubated for 1 hour at 37°C to achieve equilibrium of plasma

protein binding of drug and/or metabolite. The concentrations of FX and NFX

isomers were measured before and after a 1-hour incubation. There was no

significant change (<5%) in the plasma concentration of either FX or NFX during

these periods.

In addition, spiked plasma samples were incubated in the dialysis cells under the

dialysis conditions to determine if FX and NFX would be stable in plasma over

the duration of the equilibrium dialysis period. Similarly, no significant change

(<5%) in the concentrations of the FX and NFX isomers was observed over the 1

to 12 hour period of study.

Page 240 Free fraction determination of fluoxetine and norfluoxetine using equilibrium dialysis

The free fraction (unbound) of FX and NFX in adult human plasma was determined by equilibrium dialysis following 4 hours of incubation at 37°C

(Tables 4.12 to 4.14). Due to the unavailability of EDTA-treated human serum in the laboratory, no comparison was made between heparin- and EDTA-treated human plasma.

For the in vitro plasma protein binding study, an appropriate amount of FX and/or

NFX spiking solution prepared in PBS (less than 5% of total volume) was added to drug-free plasma and incubated for 1 hour to allow equilibrium for protein binding of the analytes. In addition, plasma protein binding in human plasma was evaluated separately (Table 4.12) and simultaneously (Table 4.13) for FX and

NFX, in order to determine a potential binding interaction between the parent drug and its N-demethylated metabolite.

FX and NFX isomers are extensively bound (>95%) in human plasma. There was a statistically significant difference in the free fractions of SFX and RFX

(p<0.01), with SFX having a lower free fraction (2.9%, i.e. higher plasma protein binding) than the RFX (4.1%) (Table 4.12). The S/R ratio of FX free fraction was

0.74 ±0.17. Similarly, the S/R ratio of NFX free faction was 0.79 ± 0.07. No significant differences were observed among the free fractions of the 100 (low),

1000 (med), or 10000 (high) ng/mL concentrations. The free fractions of FX and

NFX determined in the absence of the other were significantly lower (p<0.01)

Page 241 than those determined in the presence of the other compound except for SFX

(Tables 4.12 and 4.13).

Plasma protein binding of FX and NFX was also measured in pooled in vivo samples from selected adult patient samples with high concentrations of FX and/or NFX (100 ng/mL per isomer). Similar to the in vitro protein binding samples, there was significant stereoselectivity in the free fractions of the FX isomers (Table 4.14).

Page 242 CO

CN

o LO « o CD T- oo d CD r-- co +1 %- d d cb ci CD oq +| o o o oo d -+—' CO CO 1_ o CD or CD co a CO LO ccj O 00 o d oo o IO CD +1 2 CD +1 o cS o a O ,-i Tf CO LO co o C d d V CU < Tf Tf Sf CD CN o CU CO CN CO < d d d d V d d d Tf CL +1 +1 +1 +1 cd +1 +1 +1 d is +i * o CO o oq CT> CD CO LO 00 * V CN CN CN CN Tf - CL O in < d CN o w V d O CL CL +r> < 6 LO CO CO LO 1^ CN CN < CD d d d d d d d CS • d +1 +1 +1 +1 +1 +1 +1 V « o +1 h~ CO CO Tf LO CD Ti• 0 1 cn — CD CD T3 "V CB CD 1 CD sr CD Tf CO LO .ia CD CN CO Tf CD d d d d CB X d d d d .a m CL fa LL. £ +1 +i +1 +1 LL £ +1 +1 +1 CL ^3 C+H CO O CD Tf 00 LO CT) CD O CO o CD CD CN CO CN CN CN E CO CN CO CO c/i .5 o 00 CD £ c 1 & cn >.>> cd 1 S - CB CB CO co" co" CJ CO CO CO CT) o o II II II CT) II II II c c II 'S c II — *C SP c S '—' /5 T3 'S "B c JC CD .SP SP Lo \

M e '<*> Hi g < i/5 * <

T*

CU d H H CN cf a.

o +1 CO l.o 1^ CO CD d +1 CO CO o

•st Tf o d +1 +1 CO T- LO CO CN o d V CL

CO CO d d +1 +1 T- T-r •sf CO

CJ CD .ia CD CO CO CS CD d d CL 01 +1 +1 CO f O CO CO

cu CL)

LO 00 CL) d d 01 +i +i I— LO Tf T3 _>> C E o •c 'S CD CL .2P -i—1 c/5 CO o CL

3 es H 4.12.6 Stereoselective in vitro drug metabolism of fluoxetine in hepatic microsomes

The in vitro metabolism of FX in hepatic microsomes was measured as the

formation rate of the N-demethylated metabolite, NFX, following 30-min

incubations. An incubation concentration of 100 ng/mL (0.32 pM) per FX isomer

was used for the human and rat microsomes. The results obtained for the pooled

human and rat microsomes are presented in Table 4.15. All the microsomal

incubations of FX were performed in duplicate. The formation rates of SNFX and

RNFX were similar in human pooled microsomes, with the S/R ratios of NFX

formation close to unity (the S/R ratios of 0.90 and 0.95). Similar to human

pooled microsomes, there was no significant difference in SNFX and RNFX

formation rates in the ovine microsomes (Chapter 3, Table 5.7). In contrast, the

formation rates of RNFX in both normal and phenobarbital-induced Sprague-

Dawley rat microsomes were higher than those of SNFX (the S/R ratios of 0.71

and 0.76). The formation rate for NFX in phenobarbital-induced rat microsomes

was approximately 3 times higher than that obtained from normal (non-induced)

rats.

In order to identify the CYP isozyme(s) involved in the N-demethylation of FX,

single CYP cDNA-expressed microsomes (Supersomes™) containing various

individual human CYP isozymes were used (Table 4.16). The NFX isomers

were detected in microsomes containing CYP2C9, CYP2C18, CYP2C19 and

CYP2D6 isozymes following a 30-min incubation. Similar to the human pooled

microsomes, no significant differences in the formation rates of SNFX and RNFX

were observed in CYP2D6-expressed microsomes. The formation of SNFX was

Page 245 slightly higher with CYP2D6 than RNFX, with the S/R ratio of 1.17. However,

the formation of SNFX was significantly lower (-5-7 fold) than the corresponding

values of RNFX in the microsomes containing CYP2C9, 2C18 and 2C19, with the

S/R ratio ranging from 0.14 to 0.19. No detectable concentration of NFX isomers

was detected in CYP2C8 isozyme. Therefore, it appears that FX is metabolized

to NFX by both the CYP2D6 and CYP2C subfamilies (except CYP2C8) at the

concentration examined. No detectable formation of NFX was observed in the

other CYP isozymes tested including CYP3A4.

Table 4.15 Formation rate of NFX isomers in pooled microsomal preparations from humans, sheep and rats (n=2).

Hepatic microsome Formation rate

(pmol/mg protein/30 min)

SNFX RNFX S/R ratio

Human (pooled) 11.2 11.8 0.95

11.8 13.1 0.90

Rat 33.5 48.2 0.70

(Sprague-Dawley) 38.9 51.2 0.76

Rat (Sprague-Dawley) 112.0 153.9 0.73

(Phenobarbital-induced) 102.2 143.2 0.71

Page 246 Table 4.16 Formation rates of NFX isomers in human cDNA-expressed single CYP isozyme microsome preparations (n=2).

Formation rate

(pmol/nmol CYP/30 min)

SNFX RNFX S/R ratio

CYP1A1 N/D N/D

CYP1A2 N/D N/D

CYP1B1 N/D N/D

CYP2A6 N/D N/D

CYP3B6 N/D N/D

CYP2C8 N/D N/D

CYP2C9 23.3 164.5 0.14

CYP2C18 17.2 116.6 0.15

CYP2C19 13.1 70.7 0.19

CYP2D6 150.7 128.9 1.17

CYP2E1 N/D N/D

CYP3A4 N/D N/D

CYP3A5 N/D N/D

CYP4A11 N/D N/D

Control N/D N/D

N/D: not determined as the NFX concentration was below the limit of quantitation

Page 247 4.13 Discussion

4.13.1 Determination of fluoxetine/norfluoxetine and paroxetine concentrations in

clinical study samples

In the present study, all necessary supplies for blood sample collection were

provided in both the delivery units and outpatient laboratories of the participating

hospitals. Clear instructions for the blood sample collection procedure were also

supplied. Furthermore, the nurses involved in the study were trained in order to

ensure proper sample collection. In the case of postnatal sample collection, the

investigator (John Kim) conducted patient interviews and was present for most

sample collections. However, due to the unpredictable nature of childbirth,

changes in nursing staff in the delivery units and the participation of several

different hospitals, it was not possible for the investigator to be present for all

deliveries. This resulted in less control and custody over sample preparation and

handling compared to the animal studies described in Chapter 3. Therefore, the

impact of potential errors in sample collection and/or processing in the hospital

setting was examined to ensure integrity of the data.

In this regard, the stability of the analyte(s) in biological samples during normal

hospital procedures was examined following discussions with hospital staff

(delivery suite nurses, phlebotomists and chemistry laboratory staff). First, an

error in selection of the blood collection tube (i.e. use of wrong type of

Vacutainer tubes) could result in a different type of serum or plasma sample (e.g.

heparinzed or non-heparinzed). As described in Section 4.12.1, the use of any of

Page 248 the three common Vacutainer tubes (K EDTA [purple], heparin [green] and non-

additive [red]) yields similar drug concentrations. Therefore, this error did not

likely affect analyte concentrations measured during the study. Similar to the

findings of Amitai et al. (1993), there were no significant differences in FX and

NFX concentrations among whole blood, plasma (both heparinized and EDTA-

treated) and serum samples. This was also the case for PX. Therefore, in the

present study, the serum and plasma concentrations of FX, NFX and PX could be

used interchangeably.

Furthermore, in many hospitals, there are often procedural and time delays in

sample transfer from the delivery unit and outpatient laboratory to the chemistry

laboratory, which usually handles the serum/plasma preparation following sample

collection. Thus, it is not uncommon to have a time delay (15 min up to 1-2

hours) in serum/plasma separation in many cases. Our studies indicated that there

was no significant degradation of the analytes (FX, NFX and PX) in treated

(heparinzed or EDTA) or clotted blood for up to 2 hours. Therefore, a normal

delay (up to 2 hours) would not result in any significant changes in analyte

concentrations.

4.13.2 Demography and medical history of the study population

In the present study, the maternal age, gestational age, infant birth weight in FX-

and PX-exposed groups are similar to the control group who were not on

antidepressant therapy and were recruited for the collaborative biobehavioral

studies. For the study, 8 FX- and 20 PX-exposed pregnancies were examined.

Page 249 The higher number of patients in PX therapy compared to patients in FX does not necessarily reflect the overall antidepressant prescribing trends among the participating psychiatrists in the Lower Mainland area. Rather, it is likely due to

1) preference of PX over FX in the Reproductive Psychiatry Program, and 2) wider indication for Paxil (MDD, OCD and PD; Prescribing information, 2000) than for Prozac (MDD and OCD).

In addition, more than 50% of patients in the PX group were also taking a benzodiazepine (mostly clonazepam) concurrently with Paxil®. Discussion with attending psychiatrists indicated that a benzodiazepine is used concurrently to manage the higher incidence of anxiety disorders and sleep disturbances observed in patients on PX therapy (personal communication, 1998).

Overall, the PX-exposed group had a relatively higher incidence of alcohol and cigarette consumption than the FX-exposed group during pregnancy and the nursing period. However, there appears to be no logical explanation for this higher rate. The small sample size in the study in combination with a relatively higher incidence of anxiety and obsessive-compulsive disorders in the PX group may have contributed to the higher alcohol and cigarette consumption. However, there was no significant difference in the birth outcomes between the two groups.

Page 250 4.13.3 Perinatal complications for fluoxetine and paroxetine-exposed pregnancies

From the literature, there appears to be sufficient data suggesting that the use of

SSRIs during pregnancy does not increase the incidence of congenital

malformation (i.e. SSRIs are not likely to be morphological teratogens at normal

therapeutic doses) (Goldstein, 1995; Chambers et al, 1996). Infant neurological

effects of low dose exposure to SSRIs resulting from maternal administration

during pregnancy and the early infant period are not presently clearly understood,

but there appear to be no significant adverse effects. From gross assessment of

the infants' neurological development by measurement of IQs following SSRI-

exposure during pregnancy and possibly during the nursing period (Nulman et al,

1997), it appears that the SSRIs do not affect neurological development.

In the present study, there was a significant difference in APGAR scores for FX-

and PX-exposed infants at 1 min following birth; however, the APGAR scores for

these groups recovered to normal values by 5 min. Therefore, it is unlikely that

these low APGAR scores would be of any clinical significance. Moreover, it is

not clear whether the observed low APGAR scores at 1 min is an indicator of fetal

complications immediately following birth. The physiological effects of FX

exposure in utero in the sheep studies (Chapter 3) and as well as those ongoing in

the laboratory (Morrison et al, 1999a,b) suggest that effects such as lowered

oxygen delivery and uterine and umbilical blood flow are transient following

either a single iv dose or 8-day intravenous infusion administration. However, FX

is typically administered as an oral formulation in humans on a chronic basis.

Due to this limitation and the lack of comparative human physiological data, it is

Page 251 difficult to correlate our animal results directly to the observations that have been made in humans (see below).

There are conflicting reports on the incidence of fetal growth restriction and other complications in FX-exposed pregnancies (Goldstein, 1997; Goldstein et al,

1996; Chambers et al, 1996; Nulman et al, 1997). Chambers et al. (1996) reported a lower mean birth weight and shorter mean birth length in full-term infants exposed to FX in late gestation. In contrast, no significant difference in birth weight of FX-exposed infants compared to a matched control group was reported by Goldstein (1995), Goldstein et al. (1996) and Nulman et al. (1997).

FX exposure during the third trimester and/or entire pregnancy may also be related to perinatal complications such as premature delivery, admission to the special care nursery, and poor neonatal adaptation (e.g. respiratory difficulties, cyanosis on feeding and jitteriness) (Chambers et al. 1996). There are also case reports of perinatal complications (colic, sleep disturbance, jitteriness, irritability, palpitations, increased respiratory rate) following FX and PX exposure during pregnancy and/or breast-feeding (Spencer 1993; Dahl et al,. 1997; Mhanna et al,

1997) . For PX, no significant difference in birth outcome (congenital malformations, stillbirth, miscarrage, prematurity) has been reported (Kulin et al,

1998) .

In the present study, birth weight of both FX- and PX-exposed babies was not significantly different from the control group or the average estimated fetal weight at 37-40 weeks of pregnancy (Grunebaum 1999). No incidence of fetal growth restriction (less than 10th centile for gestation) was observed in the FX-

Page 252 exposed group. However, 3 out of 20 newborns were below the 10 centile at birth in PX-exposed pregnancies (one of these three infants was also exposed to clonazepam). Due to the relatively small number of subjects involved in the current studies, it is very difficult to access the potential effects of FX or PX exposure on fetal growth restriction. Moreover, by definition, a gestational age- adjusted birth weight less than the 10th centile will include about 10% of infants and 3 out of 20 (i.e. 15%) is close to this. Overall, then, the result of this study suggests minimal, if any, effects of either drug on fetal growth or condition at birth. This is contrary to the results of Chambers et al. (1996) for FX.

In the case of PX-exposed pregnancies, the use of a benzodiazepine in combination with PX may potentially complicate the interpretation of the perinatal outcome data. It is well established that maternal benzodiazepine

(clonazepam) use can suppress the newborn's respiratory response and motor activities (floppy baby syndrome) (McElhatton, 1994). In the present study, there was no significant difference in APGAR scores between clonazepam-exposed

(n=12) and naive (n-8) infants from the PX-exposed group. However, due to the small sample size, it was not possible to identify clonazepam as a covariate in the study.

Page 253 4.13.4 Fetal/neonatal exposure to fluoxetine, norfluoxetine and paroxetine during

the perinatal period

Comparing FX and PX during pregnancy, fetal exposure to FX and NFX appears

to be higher than for PX (Tables 4.7 and 4.9). Fetal exposure to these

antidepressants was estimated from the ratio of cord-to-maternal serum

concentration of FX, NFX and PX. In the case of FX and NFX isomers, these

F/M ratios were close to the unity (1.04 ± 0.28 for FX and 1.04 ± 0.16 for NFX).

Therefore, human fetuses appear to be exposed to FX and NFX concentrations

similar to those of the mother at parturition and possibly during pregnancy

following long-term maternal drug administration.

Umbilical cord serum/plasma FX and NFX concentrations have been previously

reported in two case reports (Spencer 1993; Mhanna et al., 1997). Total FX and

NFX concentrations in umbilical cord were 26 ng/mL and 54 ng/mL, respectively,

at the time of delivery in a neonate whose mother was taking a daily dose of 20

mg (Spencer 1993). In the other report, serum concentrations of FX and NFX

were 129 and 227 ng/mL, respectively, in an infant on the second day of life

whose mother was taking a daily dose of 60 mg (Mhanna et al., 1997). These

values are similar to the concentration data obtained in the current study (Tables

4.7 and 4.9). In both case reports, no maternal serum/plasma FX or NFX

concentrations were reported, so a F/M ratio could not be determined. During the

course of our studies, there were no previously reported data available for fetal

(i.e. cord blood) serum/plasma concentrations of PX. Recently, data on maternal

and umbilical cord blood concentrations of several SSRIs including FX, NFX and

Page 254 PX at birth were presented at the American Psychiatric Association annual meeting (Stowe et al, 2000b). In the presentation, the F/M ratios of FX, NFX,

PX, sertraline and were reported to be 0.83 ± 0.27, 0.81 ±

0.24, 0.34 ± 0.20, 0.44 ± 0.24 and 0.47 ± 0.36, respectively (Stowe et al, 2000b).

Similar to the present study, the F/M ratio of PX was lower than the corresponding values of FX and NFX.

With both drugs in the current study, a statistically significant correlation between maternal and umbilical cord concentrations of FX, NFX and PX was observed as shown in Figure 4.1. This suggests that maternal serum or plasma concentration of these drugs could be used as a surrogate marker of fetal drug concentration during pregnancy, once the F/M ratio of the particular drug has been established.

The relatively high degree of fetal/neonatal exposure observed in this study (as indicated by a higher F/M ratio and relatively higher levels of FX and NFX during, immediately postpartum and early nursing period, in combination with the long half-life of FX and NFX) may raise concern regarding the safety of maternal

FX use during the perinatal period. Elevated levels of FX and NFX immediately postpartum would be of a particular concern since, as indicated in the previous section, there are reported cases of postnatal complications associated with maternal FX intake (Spencer et al, 1993; Chambers et al. 1996; Mhanna et al.

1997). While the results of these reports are conflicting, these complications may have been related to high levels of FX and NFX in the neonate during the first few days of life as observed in the current study.

Page 255 These observations of placental transfer of FX and NFX in human are also consistent with the pharmacokinetic data obtained in third trimester pregnant sheep (see section 3.3), indicating that FX and NFX also cross the ovine placenta readily. However, the serum cord-to-maternal (F/M) ratios of FX and NFX concentrations were higher in the human studies than those observed in the pregnant ewes, where fetal exposure was estimated as the ratio of fetal and maternal plasma AUC (0.57 ±0.17 for FX and 0.51 ± 0.13 for NFX) following a single FX dose to the ewe. As discussed in Chapter 3, this difference is likely related to the longer-term drug administration regimen in the human study compared to the bolus injection protocol in the sheep project.

Fetal/neonatal drug metabolic capacity for FX and PX is not known in humans at this time; however, the pharmacokinetic and in vitro drug metabolism data from the ovine fetus suggest that the capacity of the fetus to metabolize FX is very limited. As shown in Chapter 3 (Section 3.5.4), the formation of NFX in the fetal lamb following fetal administration of FX was not observed. Neither significant formation of NFX nor amniotic accumulation of either FX or NFX isomers was observed in the fetal lamb (Chapter 3). Furthermore, no significant formation of

NFX was observed from a preliminary in vitro drug metabolism study in fetal ovine microsomes (Chapter 3), suggesting that metabolic capacity for FX is limited in the ovine fetus. There is no direct evidence for a lack of FX metabolism in the human fetus to our knowledge (no in vitro metabolism study in fetal microsomes has been conducted for FX). However, there are several reports

Page 256 indicating that the CYP isozymes involved in FX metabolism are expressed minimally in the human fetus and neonate during the first few weeks of life.

Hakkola and coworkers (1994) reported that CYP2C subfamily and CYP2D6, which are involved in the metabolism of FX, are not expressed in human fetal liver using a RT-PCR method. Treluyer et al, (1997) have also demonstrated that the protein levels and enzyme activity of CYP2C9, which is one of the isozymes involved in the conversion of FX to NFX, were not detected in human fetal liver but increased in the first month after birth. Similarly, CYP2D6 protein and enzyme activity are absent until birth (Jacqz-Aigrain and Cresteil 1992). These observations are in contrast to CYP3A activity, which is detectable as early as the

17th week of gestation in human fetuses (Jacqz-Aigrain and Cresteil 1992).

Therefore, from these reports it would appear that the metabolic capacity of the

CYP2C subfamily and CYP2D6, and thus the N-demethylation of FX, in the human fetus and neonate during the first weeks of life would be minimal.

In addition, the contribution of renal excretion of FX, NFX and their glucuronide conjugates is likely low in the fetus and neonate. There is no detailed information on the renal elimination of FX, NFX and their conjugates in the fetus and neonate.

However, in human adults, the renal elimination of FX and NFX and their conjugates account for only about 10% of the administered dose (Lemberger et al

1985). In pregnant and non-pregnant ewes, the contribution of maternal renal elimination of FX, NFX and their glucuronide conjugates was not significant

(-2% of the dose, see Section 3.3). Renal elimination capacity for many drugs is generally low in the fetus and neonate (see Section 1.17), and the amniotic

Page 257 accumulation of FX and NFX appears to be minimal in animal studies (Pohland et al, 1989; Chapter 3). Therefore, it is very unlikely that the renal elimination of

FX, NFX and their conjugates contributes significantly to the total body clearance of FX in the neonate/fetus. In ovine and rodent fetuses, FX, NFX and their conjugates are not significantly excreted into amniotic fluids (see Chapter 3;

Pohland et al 1989). Using a spot urine test, the renal excretion of FX and NFX in human infants has been reported by Taddio et al, (1996). However, this spot urine test may not provide sufficient information to assess neonatal renal elimination of FX and NFX. The amount of drug and metabolite measured on a semi-quantitative basis indicates that renal elimination of FX and NFX in the infant is not significant (Taddio et al, 1996; Motherisk program 1997, personal communication). Thus, it appears that the elimination of FX and NFX depends primarily on hepatic biotransformation to subsequent metabolites and elimination of these metabolites, as in the case of adults.

The overall reduced capacity for drug elimination and metabolism in the fetus is likely to contribute to the elevated FX and NFX levels observed immediately postpartum. Another possible factor contributing, in part, to elevated FX and

NFX levels in the neonate may be related to the loss of body weight (water) at parturition. Between 4 to 15% of birth-weight is lost during 2 to 6 days following birth (Teng et al, 1993) as a result of the initial delay in lactation (1-2 days) and subsequent neonatal body water loss. Thus, this may in part account for the increased neonatal drug concentration. However, in our studies this elevated drug concentration was only observed in the FX-exposed group, whereas neonatal

Page 258 serum PX concentrations declined rapidly during this perinatal period in all PX- exposed neonates. Thus, postnatal weight loss is not likely responsible for the elevated FX and NFX concentrations observed at the time of PKU sampling.

The F/M ratio of PX is lower (0.40 ±0.16) than the corresponding values for FX and NFX. Similarly, the infant serum PX concentrations had also declined by

-40-50% by the time of the PKU test compared to those at birth. Thus, the extent of fetal and neonatal exposure to PX appears to be lower during the perinatal period compared to FX. However, in a case report, tachypnea, jitteriness and tremor were observed in a neonate exposed to PX during the third trimester via a daily dose of 30 mg to the mother (Dahl et al, 1997). The serum PX level was 68 nmol/L (22 ng/mL), 75 nmol/L (25 ng/mL) and 23 nmol/L (7.6 ng/mL) at one, two and three days of life, respectively. The mother's serum PX concentration was subsequently measured to be 195 nmol/L (61 ng/mL). These values appear to be in the higher range of the PX concentrations observed in the present study, and may be responsible for the observed complication.

At present, there is no clear explanation for the lower F/M ratio for PX compared to FX and NFX. Lipophilicity, which is measured as Log P (partition coefficient in octanol/water), of FX (4.5) is higher than that of PX (3.4). However, this difference is unlikely to be of great importance since both FX and PX are highly lipophilic and have similar basic pKa values (8.9 for FX and 9.1 for PX). It is more likely that the difference in fetal/neonatal exposure results from differences in the fetal ability to metabolize and eliminate drugs and differential plasma protein binding.

Page 259 As with FX, the metabolic capacity for PX in the fetus/neonate is not known at this time. Based on the human adult data from the literature, PX is metabolized via oxidative cleavage of the methylenedioxy bridge, resulting in an unstable catechol intermediate, likely by a CYP isozyme (possibly CYP2D6) (Kaye et al.,

1989; Bloomer et al, 1992). This intermediate is further methylated in either the meta- or para-position possibly by catechol-O-methyltransferase (COMT) (Kaye et al, 1989; Sindrup et al, 1992). Subsequently, these metabolites are conjugated by glucuronidation (Sindrup et al, 1992). Recently, the involvement of CYP3A4 isozyme in PX metabolism in addition to CYP2D6 was suggested (Kuss and

Hergerl, 1998). Unlike the isozymes involved in FX metabolism (CYP2C9/19 and CYP2D6), the enzymes involved in PX metabolism (CYP3A4 and COMT) are expressed in the human fetus (Jacqz-Aigrain and Cresteil 1992, Parvez et al,

1979) and in the placenta (Barnea et al, 1988; Tilgmann and Kalkkinen, 1991), except for CYP2D6. Therefore, the human fetus and neonate may to be able to metabolize PX during pregnancy and in the immediate postpartum period via these enzymes. This fetal/neonatal metabolic capacity may be responsible for the observed lower F/M ratio at birth and the I/F ratio at the time of PKU testing. No data on fetal plasma protein binding of PX are available to assess its effect on fetal drug exposure (i.e. F/M ratio).

A number of patients taking FX (~18%) and PX (-50%) were also concurrently receiving clonazepam for the control of anxiety and sleep disturbances related to

SSRI therapy. The concomitant use of this benzodiazepine raises concern not only for potential fetal adverse effects related directly to this drug, but also for its

Page 260 possible effects on FX and PX metabolism and subsequent fetal drug/metabolite

exposure. While the induction of the CYP2B subfamily by has

been reported (Nims et al, 1997), these isozymes do not appear to be involved in

the metabolism of either PX or FX and therefore are not likely an important factor

influencing fetal exposure to these drugs.

In conclusion, the results of the current study indicate that in case of FX therapy,

infant serum FX and NFX concentrations may be elevated during first few days of

the infant's life. Therefore, provided that patient remission can be maintained, a

reduction in maternal dose about 2-3 weeks prior to delivery and early in the

nursing period (~ 2-3 weeks) may help to reduce the risk of perinatal

complications. In contrast, with PX therapy, fetal and neonatal exposure was

comparatively less suggesting that this might be a better treatment option. One

must also consider, however, the similar was achieved at lower plasma

concentration of PX compared with FX (see Introduction, Section 1.8).

4.13.5 Infant exposure to fluoxetine and paroxetine during the nursing period

There are several prospective studies indicating that like most antidepressants,

which are in general lipophilic and basic compounds, the SSRI's FX (Taddio et

al. 1996, Kristensen et al, 1999), PX (Ohman et al, 1999; Begg et al, 1999;

Stowe et al, 2000) and sertraline (Altshuler et al, 1995; Stowe et al. 1997) are

excreted in human breast milk. Our data also demonstrate that PX and FX (and

its metabolite NFX) are excreted in breast milk of nursing mothers.

Page 261 Similar to our observations during the perinatal period, infant exposure to FX and

NFX via breast-feeding appears to be higher than that of PX (Tables 10 and 11).

Infant exposure to these drugs was estimated from the weight-adjusted infant to maternal dose. Using this calculation, the infant dose was 2.9 ± 2.0% for FX and

3.6 ± 2.1% for NFX. Thus the infants were exposed via breast feeding to a combined FX plus NFX dose of 6.5 ± 3.5% (range 1.6-17.7%). These values appear to be consistent with those reported for FX by Taddio et al. (1996) (10.8 ±

2.2%; i.e. sum of FX and NFX expressed as FX equivalents, weight-adjusted) and

Kristensen et al. (1999) (6.81%; range 2.15-12%). In contrast, the corresponding value for PX was 0.8 ± 0.5%, which is significantly lower than those of FX and

NFX. Our weight-adjusted infant PX dose estimate is also similar to the range of values reported by Spigset et al. (1996) (0.34%), Beggs et al. (1999) (1.13 ±

0.46%) and Ohman et al. (1999) (1.4 ± 0.79%).

The milk-to-serum (M/P) concentration ratio was also used to assess infant exposure. While not significantly so, the M/P ratio of PX (0.46 ±0.21) appears to be lower than that of the FX (0.66 ± 0.47) and NFX (0.56 ± 0.25) (Tables 4.7B and 4.8). Similar to the estimated infant dose, the M/P values for FX and NFX are consistent with those reported by Kristensen et al. (1999) (0.68 for FX and

0.56 for NFX), but slightly lower than the ones determined by Taddio et al.

(1996) (0.88 ± 0.44 for FX and 0.82 ± 0.3 for NFX). Unlike FX and NFX, a wider range in the PX M/P ratio estimate has been reported by Spigset et al., 1996

(0.09), Ohman et al, 1999 (0.69 ± 0.29) and Begg et al, 1999 (0.39 ± 0.1 by the

AUC method, 0.96 ±1.0 estimated from pre- and post-fed samples). However,

Page 262 Spigset et al, (1999) suggested that their estimate may represent a minimum value, since the serum sample was taken at Cmax. Furthermore, Begg et al,.

(1999) suggested that the value from their first study (0.39 ± 0.1) is a more reliable estimate. Therefore, the M/P ratio from our studies represents a median of these values and appears to be a reasonable estimate of the M/P ratio for PX.

Therefore, the data from the present study as well as those from the literature suggest lower infant exposure to PX compared to FX and NFX. As previously mentioned in Section 4.13.4, differences in the physicochemical characteristics of these compounds such as lipophilicity may account in part for the lower M/P ratio for PX compared to FX. These data are also in line with the lower volume of distribution of PX (-13 L/kg) than that of FX (-25 L/kg) in humans. In addition,

PX has a shorter half-life (-1 d)(Kaye et al, 1989) and lacks an active metabolite compared to FX (-3-5 d) and its active metabolite NFX, which has an even longer half-life (-7-15 d) (Benfield etal, 1986).

In the present study, the extent of neonatal exposure to FX and PX was also assessed via direct infant blood sampling, with infant serum samples being collected at the same time as the maternal serum and breast milk samples. For the

FX-exposed group, detectable concentrations of FX and/or NFX were observed in the majority of infants younger than 2 months of age (Table 4.1 OA). As shown in

Figure 4.4, there is a decline in the infant-to-maternal serum ratio of FX and NFX between 2 days to 7 months. In contrast, detectable concentrations of PX were observed in only 6 infants over the first 1-3 months following birth (Table 4.11)

Page 263 and their I/M ratios were much lower than those of FX or NFX. These postnatal exposure data appear to be consistent with data obtained during the perinatal period (cord blood and PKU samples), where neonatal exposure to PX was much lower compared to FX or NFX.

From our observations and those of others (Wisner et al. 1999), antidepressant therapy in nursing mothers of neonates and younger infants (less than 2 months) resulted in relatively higher serum drug concentration than those measured in older infants (more than 2 months), possibly due to lower drug metabolizing capability of the younger infants. Maternal antidepressant therapy in older infants

(>2 months), then, may pose a relatively lower risk than in the younger group. A reduction in the intake of breast milk with the supplementing of solid food and increased relative maturity of hepatic metabolic enzyme systems in the older infants may responsible for lower serum levels of the drugs studied in this group.

Overall, the data from our studies appear to support favorable pharmacokinetic characteristics of PX during the perinatal and nursing periods; however, these data should be interpreted cautiously. In the absence of detailed information on fetal/neonatal morphologic and behavioral teratogenecity, especially for PX, it is difficult to assess the overall potential risk of PX based solely upon the degree of exposure.

The use of psychotropic medications during the ante- and post-partum periods poses a therapeutic dilemma. The decision to treat pregnant and nursing mothers with antidepressants should be based on a risk-benefit assessment of the

Page 264 individual patient by the attending physician and clinical pharmacokineticist, if available. From our experience in the Reproductive Psychiatry Program at

BCCWH and from the literature (Cogill et al, 1986; Stein et al, 1991; Murray

1992; Misri et al, 1994), there are substantial risks of untreated depression on the well-being of both the mother and fetus/neonate. These may outweigh the relatively low risk of commonly prescribed antidepressants such as FX, PX and sertraline on fetal development, labour, and the newborn infant. Therefore, there appears to be a consensus in the medical community that pharmacological treatment of clinical depression with careful therapeutic monitoring in this population would be beneficial for the well-being of both the mother and her infant. Accepting the necessity of pharmacological intervention, the goal of these treatments should be focused on minimizing fetal and neonatal exposure while maximizing clinical benefit of the medication to the mother.

Questions are often raised by both practitioners and patients regarding the need for discontinuation of breast-feeding during medication therapy. The decision to breast-feed ultimately lies with the patient herself following consultation with her physician and/or clinical pharmacokineticist. Thus, it is the responsibility of health-care providers to give the best available information. Monitoring the infant serum/plasma concentration may provide valuable information to determine the degree of exposure. The most direct and relevant indicator of infant exposure would be obtained by measurement of medication levels in the infant. Single point determination of infant serum concentration appears to be the most practical method of determining secondary infant exposure. Pharmacokinetic profiling

Page 265 through multiple blood sampling would provide very useful information on infant drug exposure; however, without a clear indication of complication(s) this would be ethically unacceptable and may further aggravate maternal anxiety, depression and feelings of guilt. However, analytical services for these SSRIs may not be available for the majority of practitioners, and this is further complicated by the fact that only a limited serum/plasma volume can be collected from infants, thus requiring sophisticated analytical equipment with high sensitivity.

In order to reduce infant exposure, Stowe et al. (1997) have suggested discarding a portion of the breast milk when medication concentrations are at their maximum in milk. Merits of discarding a portion of milk at this time may not be applicable in the practical setting. First, determination of the period at which higher milk concentrations occur may not be feasible due to individual variations in pharmacokinetics (time to Cmax), especially for a compound like sertraline where absorption of the drug is significantly affected by food intake and other factors.

Second, the psychological aspects of this practice on the mother should be considered. By instituting a regimen of milk disposal, mothers may be concerned enough to decide to discontinue breast-feeding or antidepressant drug therapy.

4.14 Stereoselective disposition of fluoxetine in human chronic therapy

For a drug like FX which is eliminated mainly by hepatic pathways, total body clearance of the drug depends principally on its hepatic clearance. As described in Section 3.4.5, the stereoselectivity of drug clearance is likely caused by the

Page 266 difference in the free fraction (protein binding) and/or the intrinsic clearance

(metabolism).

Therefore, these two possible mechanisms for stereoselective drug disposition were examined in Chapters 3 and 4. Plasma protein binding of FX and NFX isomers was assessed in vitro and in vivo in a limited number of samples. In addition, stereoselective in vitro metabolism of FX by hepatic microsome preparations was also studied in a preliminary fashion to elucidate the possible mechanism of stereoselective disposition of FX and NFX and to suggest future directions for subsequent studies.

During the course of the clinical studies, several interesting observations were made. First, there appears to be an overall increase in FX steady-state concentrations during chronic therapy (i.e. decreased oral clearance). Therefore, oral clearance of FX was estimated as follows from the serum data in Tables 4.7 and 4.12:

CLFX = (FX dose/r) / Cpss (Equation 4.3) where r is dosing interval (hr) and Cpss is the apparent steady-state concentration of total FX. Since the FX concentration was measured during steady-state after 3-

4 weeks of multiple dosing at a fixed dose and the bioavailability of FX is high, it is reasonable to use the above equation to estimate the oral clearance value.

Figure 4.7 shows the correlation between the apparent oral clearance of racemic

FX, SFX and RFX during chronic dosing in humans. Data collected from the present study population on chronic FX therapy (n=34) and reported values from

Page 267 five EM volunteers (Fjorside et al, 1999) were included in the plots (Figure 4.7).

Patients who were exposed to FX for up to 12 months were included. There was a statistically significant decrease in the oral clearance of total FX (Figure 4.7 A) during chronic therapy. This appears to be consistent with other reported observations (Catterson and Preskorn 1996; Hiemke and Hartter, 2000).

However, to the best of our knowledge, our study is the first to report a quantitative analysis of time-dependency in FX clearance. In addition, the clearance of both SFX and RFX was significantly reduced (Figure 4.7B).

However, the decrease for SFX was much more than for RFX. Consequently, a higher oral clearance value for RFX compared to SFX was observed in patients who were exposed to FX for an extended period, unlike the findings following a single dose in EMs where the clearance values for SFX and RFX were similar

(Fjorside etal, 1999).

In addition, the S/R ratio appears to be higher in the patients exposed to FX for an extended period compared to the values obtained following a single dose or at the onset of FX therapy. Figure 4.8 shows the correlation between the S/R ratio of the FX isomers and exposure time to FX in the patient population. Data from additional plasma samples collected from the patients who were on FX therapy for a short period (prior to steady-state) were also included in the plots. There was a tendency toward a higher S/R ratio as FX exposure time (range 2 days to 12 months) increased (r=0.316, p<0.05) (Figure 4.8).

Page 268 A. Oral clearance of total fluoxetine

100.0 ,

B. Oral clearance of fluoxetine isomers

;ure 4.7 Correlation of the oral clearance of fluoxetine (total and individual isomers) (L/h) in humans (n=39). Oral clearance values for 5 subjects at day 1 (single oral dose) were obtained from Fjorside et al, 1999.

Page 269 •r=0.316 p< 0.05 (Spearman) 0) O 5.00

CD c • % 4.00 • X o 3 IT 3.00 o o ts >- 2.00 Ct to

-0.50 0.00 0.50 exposure time (log month)

Figure 4.8 Correlation of the S/R ratio of serum concentration of fluoxetine isomers and exposure time in the patients in acute and chronic fluoxetine therapy (n=40).

Looking at the data in yet another way, one observes a significant negative correlation (r=-0.727, p<0.0001) between the S/R ratio of FX and apparent oral clearance of racemic drug in those patients who had taken FX for more than 3-4 weeks at a fixed dosing regimen (Figure 4.9). This trend appears to be related to the formation of NFX as shown in Figure 4.10, with lower steady-state concentrations of NFX (i.e. a low NFX/FX ratio) corresponding to a higher S/R ratio of FX. In contrast, there was a positive correlation between the S/R ratio of

NFX and relative steady-state concentrations of NFX (i.e. a high NFX/FX ratio).

Therefore, these data suggest that the stereoselectivity observed in humans during chronic FX therapy is related to the metabolism of the FX isomers (including formation of NFX).

Page 270 7.0

6.0 J

0.0 1.0 10.0 100.0 1000.0 CLFX (L/hr)

Figure 4.9 Correlation of the S/R ratio of steady-state concentration of the FX isomers and total fluoxetine systemic clearance (L/h) in pregnant and postpartum women (n=35).

NFX/FX ratio

Figure 4.10 Correlation of the S/R ratio of steady-state concentrations (Cpss) of the FX and NFX isomers and the NFX/FX ratio in pregnant and postpartum women (n=34). One patient with an unusually high NFX/FX ratio (17.9) was not included in the analysis.

- Page 271 Similar to the results from the human studies, the stereoselective disposition of

SFX and RFX was dependent on total body clearance following a single bolus IV injection in non-pregnant and pregnant sheep (i.e. negative correlation between

AUC S/R ratio and clearance of racemic fluoxetine; Figure 3.21). These data thus suggest that there are both oral clearance- and time-dependencies in terms of FX and NFX pharmacokinetics.

Time-dependency of PX clearance was also observed in postpartum women following chronic dosing as shown in Figure 4.11.

Figure 4.11 Correlation of apparent oral clearance of paroxetine (L/h) vs. exposure time (month) in pregnant and postpartum women (n=47).

Page 272 From the observations from both the human and sheep FX pharmacokinetic studies, the following hypotheses were initially proposed:

1. Fluoxetine and norfluoxetine bind to plasma protein in a stereoselective manner.

2. Fluoxetine is metabolized by multiple hepatic cytochrome P450 isozymes with a different degree of stereoselectivity.

3. Metabolism of fluoxetine isomers is altered during chronic therapy due to the inhibition of its own metabolism in a stereoselective manner.

4. Stereoselective plasma protein binding and metabolism (and its inhibition) of fluoxetine and norfluoxetine result in a higher accumulation of a single isomer over the other.

Possible causes of this stereoselectivity of FX/NFX disposition such as differential protein binding and metabolism were examined in a limited manner as presented in both Chapters 3 and 4.

Stereoselective disposition of FX and NFX has been suggested from previous studies in several species including humans (Fuller and Snoddy, 1993; Kim et al,

1996; Kim et al,. 1999; Fjordside et al., 1999). From Chapter 3, the stereoselective disposition of the FX isomers following a single intravenous dose in sheep appears to be related to stereoselective protein binding of the isomers and possibly to stereoselective metabolism as well. However, the formation rates of the NFX isomers in pooled ovine hepatic microsomes were not significantly different. Therefore, stereoselective disposition of FX in sheep after a single dose is likely to be largely the result of stereoselective plasma protein binding.

Page 273 For humans, Fjordside et al. (1999) indicated that there was no significant difference in systemic clearance of RFX and SFX in extensive metabolizers

(EMs) of sparteine, a model substrate for CYP2D6, after a single dose of racemic

FX. In contrast, a significantly higher systemic clearance of RFX was observed in poor metabolizers (PMs). In the present study, we have observed that the stereoselective disposition of FX is both concentration- and time-dependent. As shown in Figure 4.9, the S/R ratio of the FX isomers was significantly higher than unity in patients with a lower systemic clearance of FX (< 50 L/hr), who appear to form the majority. This observation is similar to the profile reported in PMs of sparteine (Fjordside et al, 1999). In contrast, the S/R ratio in patients with a higher systemic clearance (> 50 L/hr) was close to unity, which is similar to the profile in EMs. In addition, there was a positive correlation between the S/R ratio of FX and exposure time to FX (i.e. a higher S/R ratio was observed in patients chronically exposed to FX compared to those who just started therapy) (Figure

4.8). This finding coincides with the decrease in oral clearance of FX (Figure 4.7) and the dependency of stereoselective disposition (i. e. Cpss) on the formation of

NFX (Figure 4.10).

The N-demethylation of FX, the first step in the oxidative pathway, results in the formation of pharmacologically active NFX. Correlation data from clinical studies suggest the involvement of multiple isozymes for RFX and SFX N- demethylation. From previously published reports (Stevens and Wrighton 1993, von Moltke et al, 1996), the involvement of CYP2C9 and to a much lesser degree

CYP2C19 and CYP2D6 has been suggested. Stevens and Wrighton (1993) also

Page 274 reported relatively higher levels of N-demethylation for RFX than SFX in human microsomes. Von Moltke et al, (1996) suggested that FX N-demethylation was mainly mediated by CYP2C9 while the contribution of CYP2D6 was largely discounted. However, both studies used FX concentrations (100 uM = 31000 ng/ml) for these enzyme kinetic studies that are much higher than the Ki (-50-400 ng/mL) of FX for CYP2D6 (Brosen et al, 1991; Stevens and Wrighton 1993;

Otton et al, 1993). One of the reasons for the use of this high concentration was the limitations of the analytical method available to detect and quantify the small amounts of NFX generated by the microsomal system (von Moltke et al, 1998, personal communication). At this high concentration, FX likely inhibits its own

N-demethylation mediated by CYP2D6. Our preliminary study examining FX N- demethylation by microsomes prepared from a single CYP isozyme cDNA- expressed insect cell line shows a very different picture. Unlike the previous studies, a much lower and physiologically realistic FX concentration (0.32 u.M =

100 ng/ml per isomer) was used. This concentration is within the linear range and well below the Km for FX N-demethylation (33 uM, von Moltke et al, 1997). In addition, it is below the Ki for CYP2C9 (measured from phenytoin p- hydroxylation) and close to the Ki for CYP2D6 (-50-400 ng/mL; measured from bufuralol 1-hydroxylation) (Brosen et al, 1991; Schmider et al, 1997). Thus, the incubation concentration of FX used in the current study was likely sufficiently high that the formation of norfluoxetine was measurable but low enough that the

CYP2D6 was not completely inhibited.

Page 275 The results from the current study suggest that the N-demethylation of FX is mediated by at least three different isozymes: CYP2C9, CYP2C18/19 and

CYP2D6. Similar to previous literature reports, stereoselective N-demethylation was observed with CYP2C9 and CYP2C18/19 with higher RNFX formation

(about 10 and 5 fold, respectively). However, CYP2D6 mediated N- demethylation was also observed and the rates of RNFX and SNFX formation were similar, unlike the CYP2C-subfamily-mediated processes. An increase in

FX and NFX concentration during chronic therapy may inhibit CYP2D6 metabolism (with lower Ki) including its own N-demethylation (non- stereoselective). Thus, this results in an overall reduction in FX N-demethylation, particularly for SFX, due to more of the drug dose being metabolized by the

CYP2C subfamily. Therefore, this process in combination with stereoselective protein binding may explain the results obtained from our clinical study.

As previously mentioned, there are conflicting reports on the involvement of

CYP2D6 in FX metabolism with in vitro metabolism studies (Steven and

Wrighton, 1993; von Moltke et al, 1996) suggesting that the CYP2D6 plays a minor role in FX metabolism compared to CYP2C9. In contrast, in vivo human pharmacokinetic studies (Hamelin et al, 1996; Fjorside et al, 1999) suggest that there are significant differences in the pharmacokinetics of FX after a single dose in PMs and EMs of CYP2D6. However, our preliminary in vitro metabolism data suggests that FX is metabolized by both CYP2D6 and CYP2C9/18/19 isozymes, with N-demethylation by CYP2D6 being non-stereoselective and that by

CYP2C9/18/19 being stereoselective (higher for the (^-isomer). Further, the

Page 276 formation rate of NFX was higher from CYP2D6 than that from CYP2C9 and

CYP2C19 at the concentration examined. Therefore, it appears that FX is mainly metabolized by CYP2D6 and to a lesser degree by the CYP2C subfamily in EMs following a single dose or a few initial FX doses, resulting in apparent non- stereoselective disposition of FX (i.e. the systemic clearance of FX isomers is similar). However, in CYP2D6 PMs, FX isomers are metabolized mainly by

CYP2C9/18/19 in a stereoselective manner. Moreover, both FX and NFX are potent inhibitors of CYP2D6 (Brosen et al, 1991; Stevens and Wrighton 1993;

Otton et al, 1993) and relatively weak inhibitors of CYP2C9 (Schmider et al,

1997). Therefore, CYP2D6 is inhibited by FX and NFX concentrations within the therapeutic range (Ki range 50-400 ng/mL). In contrast, CYP2C9 (Ki range 4000-

20000 ng/mL) is not inhibited at these drug concentrations. Thus, the previous in vitro studies, which used higher FX concentrations (-100 uM), may have inhibited CYP2D6 activity resulting in a low estimate of CYP2D6's contribution to FX N-demethylation. Therefore, our data from the clinical and in vitro metabolism studies along with previously reported in vitro and in vivo data, suggest that FX metabolism is mediated by the CYP2D6 and CYP2C subfamilies with an apparent "phenotype conversion" for CYP2D6 during chronic therapy.

Initially, FX is metabolized mainly by CYP2D6 in a non-stereoselective manner.

However, as FX and NFX concentrations increase over time during chronic therapy, CYP2D6 is inhibited by FX and NFX and CYP2C9/18/19 become the major metabolizing enzymes responsible for FX N-demethylation. Therefore, the inhibition of CYP2D6 during chronic therapy may result in a reduction in

Page 277 systemic clearance and prolongation of apparent terminal half-life of FX in addition to increased stereoselective disposition. "Phenotype conversion" of

CYP2D6 following chronic administration of SSRIs has been demonstrated by

Alfaro et al. (1999; 2000). In their studies, the CYP2D6 status (degree of inhibition) was estimated by the 8-hour urinary dextromethorphan/

(DM/DX) ratio. A significant increase in DM/DX ratios was observed following

8 days of FX (60 mg/day) and PX (20 mg/day) administration (Alfaro et al,

1999). As mentioned earlier in Chapter 3, time-dependent stereoselective disposition of FX was also observed in our on-going studies involving the 8-day continuous intravenous infusion of FX in pregnant sheep (see Section 3.4.5).

Collectively then, our results suggest that the inhibition of CYP2D6 by FX may be responsible for the alteration of FX pharmacokinetics during chronic therapy, lending further support to the hypothesis of apparent "phenotype conversion" of

CYP2D6 during chronic FX therapy.

Page 278 CHAPTER 5

SUMMARY AND CONCLUSION

The overall objectives and specific aims of the present Ph.D. dissertation were to examine and compare the pharmacokinetics of fluoxetine and paroxetine during pregnancy and the nursing period. The prevalence of depression during pregnancy and the postpartum period and the need for psychopharmacological intervention necessitate a better understanding of antidepressant disposition via placental transfer and breast-feeding. In the present studies, both qualitative characterization and quantitative determination of disposition of antidepressants in fetuses and neonates were conducted following maternal administration.

In order to understand the disposition of antidepressants in the maternal-fetal unit, pharmacokinetic studies were conducted in both chronically catheterized non-pregnant and late gestational age pregnant sheep to examine the maternal-fetal disposition of fluoxetine.

Furthermore, clinical investigations of fetal and neonatal exposure to fluoxetine and paroxetine were conducted in humans, which represented a logical extension of the detailed pharmacokinetic studies in sheep.

An essential, initial requirement for the project was the development and validation of analytical methods to support the animal and clinical pharmacokinetic studies. The analysis of FX and NFX isomers was accomplished by stereoselective analytical methods using

GC/MS/EI and LC/MS/MS. Similarly, the analysis of PX was accomplished by GC/MS/EI and GC/MS/NCI. Most of the analytical methods available prior to the study require a relatively large volume of sample or have a limited sensitivity. Due to technical, practical and ethical reasons, only a limited sample volume of fetal and neonatal plasma and/or serum can be obtained in both animal and clinical pharmacokinetic studies conducted during

Page 279 pregnancy and the postpartum period. Therefore, the analytical methods used in these studies had to be sufficiently sensitive and selective enough to detect small amount of analytes in the biological sample. For these reasons, developing very sensitive analytical methods for the analytes in question was an essential requirement for the proposed studies. In the case of FX and NFX, a stereoselective analytical method was required to measure the concentrations of optical isomers, which may have different pharmacological and pharmacokinetic properties.

The separation of optical isomers has been achieved by the formation of diastereoisomers via derivatization with an optically pure derivatizing agent and by application of a chiral stationary phase.

Based upon the results of the sheep and human studies, the following conclusions can be made:

1. In comparison to non-pregnant ewes, FX pharmacokinetics in pregnant sheep exhibit

a number of statistically significant differences, which include a higher total body

clearance, shorter half-life and lower steady-state volume of distribution. These

differences could be due to induction of CYP2D6 and/or a rise in the plasma

concentration of a 1-acid glycoprotein during pregnancy. A pregnancy-related

induction of CYP2D6 occurs in the human, which may be responsible for the

increased dose requirements of antidepressants during pregnancy. This suggests that

changes in FX pharmacokinetics similar to those in the sheep may occur during

pregnancy in the human. Further studies are warranted to assess this possibility and to

elucidate the mechanisms involved.

Page 280 2. Demethylation of FX to form NFX and glucuronidation of FX and NFX occur in both

pregnant and non-pregnant ewes, although the contribution of these metabolites to

overall FX elimination (3-4% of the dose) appears to be less than in the human.

However, these findings plus our recent identification of trifluoromethylphenol in

sheep urine following FX administration suggest that the metabolic profile of the drug

is at least qualitatively similar in the 2 species. Moreover, the apparent decrease in FX

clearance in sheep with longer-term drug administration suggest that a CYP2D6-like

enzyme, which is inhibited by the drug, is involved in FX metabolism in the sheep, as

in the human. However, further research is needed to characterize the metabolites of

FX in sheep and the biotransformation enzymes involved.

3. There is extensive placental transfer of FX in both the sheep and human. The F/M

AUC ratio in pregnant sheep following single dose is substantially higher than those

of the polar drugs, but lower than the values for other amine drugs with lower

lipophilicity. This suggests that the high degree of plasma protein binding in the

mother, relative to the significantly lower extent of binding in the fetus, reduces

maternal-fetal FX transfer. However, higher F/M ratio of FX was observed in humans

and sheep following long-term administration. Therefore, human fetuses appear to be

exposed to FX concentrations similar to those of the mother on chronic therapy

during pregnancy.

4. Non-placental elimination (metabolic) capacity of FX appears minimal in ovine and

human fetuses. From both in vivo and in vitro studies, there was no evidence of fetal

NFX formation. A minimal capacity of phase I metabolism for FX was also

supported by lack of expression of CYP isozymes involved in FX metabolism in

Page 281 human fetuses and newborns. Furthermore, in the fetal lamb, there also appeared to

be a lack of phase II metabolism, as suggested by the minimal accumulation of

glucuronide or sulphate conjugates in amniotic fluid. However, further studies would

be needed to characterize the expression and capacity of fetal enzymes involved in

FX metabolism.

5. Elevated levels of FX were observed in human neonates for several days following

birth. The relatively high degree of fetal/neonatal exposure for FX, in combination

with the long half-life of FX and NFX, may raise concern regarding the safety of

maternal FX use during the perinatal period. Thus, these elevated levels may be

related to the reported cases of postnatal complications associated with maternal FX

intake.

6. There is also extensive placental transfer of PX in the human, although the degree of

fetal exposure to this drug is less than for FX. Furthermore, unlike FX, neonatal PX

levels rapidly decrease following birth. Thus, the extent of fetal and neonatal

exposure to PX is lower during the perinatal period compared to FX. Unlike the

isozymes involved in FX metabolism, at least some of the enzymes involved in PX

metabolism (CYP3A4 and COMT) are expressed in the human fetus/placenta and

neonate. Therefore, the human fetus and neonate may to be able to metabolized PX

during pregnancy and in the immediate postpartum period. This fetal/neonatal

metabolic capacity may be responsible for the observed lower fetal and neonatal

exposure.

Page 282 7. There were transient moderate changes in fetal blood gas status and glucose/lactate

concentrations following maternal and fetal FX administration in sheep. More recent

studies in the laboratory indicate that these changes are associated with a transient

decrease in uterine blood flow, and with fetal FX administration there is perhaps also

a decrease in umbilical blood flow. However, due to their moderate and transient

nature, these are unlikely to affect fetal development. Similarly in humans, no

significant changes in birth outcomes were observed following maternal use of

fluoxetine or paroxetine during pregnancy. Overall, the results of the present study

and others suggest minimal effects of either drug on fetal growth or condition at birth.

8. FX, NFX and PX are excreted in human breast milk with a higher milk-to-serum ratio

for FX and NFX compared to PX. As a result of the relatively higher concentration

of FX and lower metabolic capacity for the drug in neonate, there is a higher neonatal

exposure to FX than PX. Detectable concentrations of FX and/or NFX were observed

in the majority of infants younger than 2 months old, but there is a rapid decline in the

infant-to-maternal serum ratio of FX and NFX between 2 days to 7 months. In

contrast, detectable concentrations of PX were observed in a small numbers of

infants, and the infant to maternal ratio was much lower than for FX or NFX.

9. From our observations and those of others, FX therapy in nursing mothers of neonates

and younger infants (less than 2 months) resulted in relatively higher plasma

concentration than those measured in older infants (more than 2 months), possibly

due to lower drug metabolizing capability of the younger infants. Maternal

antidepressant therapy in older infants (>2 months), then, may pose relatively lower

risk than in the younger group.

Page 283 10. Stereoselective plasma protein binding of FX isomers was observed in both humans

and sheep. Differences in the free fractions of the isomers are likely to account for

the differences in pharmacokinetic parameters, such as systemic clearance and

volume of distribution, between isomers following a single dose. In the present

study, there were also significant differences in plasma SFX and RFX concentrations

immediately following intravenous FX administration. Furthermore, the free drug

clearance estimates in the non-pregnant ewes, pregnant sheep and fetus are much

closer compared to the total drug clearance estimates. Thus, differential protein

binding of the FX isomers appears sufficient to explain the stereoselective disposition

of the drug, at least with single dose administration.

11. The N-demethylation of FX, the first step in oxidative metabolic pathway, results in

the formation of pharmacologically active NFX. The results from the present study

suggest that the N-demethylation of FX in humans is mediated by at least three

different isozymes: CYP2C9, CYP2C18/19 and CYP2D6. Similar to previous

published reports, stereoselective N-demethylation was observed with CYP2C9 and

CYP2C18/19 with higher RNFX formation. However, CYP2D6 mediated N-

demethylation was also observed and the rates of RNFX and SNFX formation were

similar, unlike the CYP2C-subfamily-mediated processes.

12. Stereoselectivity of FX disposition appears to be clearance-dependent, particularly

with longer-term drug administration. A significant correlation between the AUC

S/R ratio of FX isomers and the systemic clearance of racemic FX was observed in

both non-pregnant and pregnant maternal sheep. Similarly, a significant negative

Page 284 correlation of the S/R ratio of steady-state FX concentrations vs. oral clearance of FX

was observed in humans during chronic therapy.

13. There was a negative correlation between oral clearance and exposure time for both

FX and PX in humans. Similarly, recent studies in the laboratory have demonstrated

that systemic clearance of FX progressively decreases during long-term drug

administration in sheep. These data along with clearance-dependency suggest that the

metabolism of FX and PX is mediated by CYP2D6, and also that CYP2D6 is

inhibited at drug concentrations in the therapeutic range. Therefore, these data

support a proposed "phenotype conversion" of CYP2D6 from EM to PM during

chronic SSRI therapy.

14. Clearance- and exposure time- dependency of FX and PX clearance during chronic

therapy could result in an elevation of maternal drug concentrations at a constant

dosing regimen, which in turn may increase fetal and neonatal drug exposure.

15. There was a positive correlation between the S/R ratio of FX and exposure time to FX

in humans and sheep. Stereoselective disposition of FX associated with long term

drug administration in the human and sheep appears to involve FX-mediated

inhibition of the non-stereoselective metabolism of the drug by CYP2D6, which

results in a progressive decrease in FX clearance and an increase in stereoselective

metabolism by CYP2C9/2C19.

In conclusion, the present studies present the first detailed pharmacokinetics of fluoxetine and paroxetine during pregnancy and the nursing period, which suggest relatively lower exposure of paroxetine compared to fluoxetine. Furthermore, stereoselective disposition of

Page 285 fluoxetine was examined during both acute and chronic administration, and potential mechanisms for differences observed were proposed.

Page 286 REFERENCES

Addis A, Koren G. (2000) Safety of fluoxetine during the first trimester of pregnancy: a meta-analytical review of epidemiological studies. Psychol Med. 30:89-94.

Addison RS, Franklin ME, Hooper WD. (1998) Sensitive, high-throughput gas chromatographic-mass spectrometric assay for fluoxetine and norfluoxetine in human plasma and its application to pharmacokinetic studies. J Chromatogr B Biomed Sci Appl. 716:153-160.

Alfaro CL, Lam YWF, Simpson J, Ereshefsk L. (1999) CYP2D6 status of extensive metabolizers after multiple-dose fluoxetine, fluvoxamine, paroxetine or sertraline. J Clin Psychopharmacol. 19:155-163.

Alfaro CL, Lam YWF, Simpson J, Ershefsky L. (2000) CYP2D6 inhibition by fluoxetine, paroxetine, sertraline, and in a cross-over study: intraindividual variability and plasma concentration correlations. J Clin Pharmacol. 40:58-66.

Allen JC, Keller RP, Archer P, Neville MC. (1991) Studies in human lactation: milk composition and daily secretion rates of macronutrients in the first year of lactation. Am J ClinNutr. 54(l):69-80.

Altumura AC, Moro AR, Percudani M. (1994) Clinical pharmacokinetics of fluoxetine. Clin Pharmacokinetics. 26: 201-214.

American Academy of Pediatrics, Committee on Drugs. (1983) The transfer of drug and other chemicals into human milk. Pediatrics. 72:375-383.

American Academy of Pediatrics, Committee on Drugs. (1989) The transfer of drug and other chemicals into human milk. Pediatrics. 84:924-936.

American Academy of Pediatrics, Committee on Drugs. (1994) The transfer of drug and other chemicals into human milk. Pediatrics. 93:137-150.

Apgar V. (1953) A proposal for a new method of evaluation of newborn infant. Curr Res Anesthes Analges. 32:260-264.

Appleby L, Warner R, Whitton A, Faragher B. (1997) A controlled study of fluoxetine and cognitive-behavioural counselling in the treatment of postnatal depression. Brit Med J. 314: 932-936.

Arita M, Mikawa H, Shirataka M, Takahashi K, Hayasawa H, Tomita M. (1997) Epidemiological research on incidence of atopic disease in infants and children in relation to their nutrition in infancy. Arerugi. 46:354-369.

Page 287 Aronoff GR, Bergstrom RF, Pottratz ST, Sloan RS, Wolen RL, Lemberger L. (1984) Fluoxetine kinetics and protein binding in normal and impaired renal function. Clin Pharmacol Ther. 36:138-144.

Artigas F, Sarrias MJ, Martinez E, Gelpi E. (1985) Serotonin in body fluids: characterization of human plasmatic and cerebrospinal fluid pools by means of a new HPLC method. Life Sci. 37:441-447.

Aspeslet LJ, Baker GB, Courts RT, Torok-both GA. (1994). The effects of desipramine and on levels of enantiomers of fluoxetine in rat brain and urine. . 6:86-90.

Atkinson HC, Begg EJ (1988) Prediction of drug concentrations in human skim milk from plasma protein binding and acid-base characteristics. Br J Clin Pharmacol. 25:495-503.

Atkinson HC, Begg EJ, Darlow BA. (1988) Drugs in human milk: Clinical pharmacokinetic considerations. Clin Pharmacokinetics. 14:217-240.

Bagli M, Rao ML, Sobanski T, Laux G. (1997) Determination of fluovoxamine and paroxetine in human serum with high-performance liquid chromatography and ultraviolet detection. J Liq Chromatography. 20:283-295.

Baldwin JA, Davidson EJ, Pritchard AL, Ridings JE. (1994) The reproductive toxicology of paroxetine. Acta Psychiatr Scand. 350): 37-39.

Barnea ER, MacLusky NJ, DeCherney AH, Naftolin F. (1988) Catechol-O-methyl transferase activity in the human term placenta. Am J Perinatol. 5: 121-127.

Bassily M, Ghabrial H, Smallwood RA, Morgan DJ. (1995) Determinants of placental drug transfer: studies in the isolated perfused human placenta. J Pharm Sci. 84:1054-1060.

Beardslee WR, Schultz LH, Selman RL. (1987) Level of social-cognitive development, adaptive functioning, and DSM-III diagnoses in adolescent offspring of parents with affective disorders: implication of the development of the capacity of mutuality. Developmental Psychol. 23:807-815.

Beeley L. (1986) Adverse effects of drugs in later pregnancy. Clin Obstet Gynaecol. 13:197- 214.

Begg EJ, Duffull SB, Saunders DA, Buttimore RC, Ilett KF, Hackett LP, Yapp P, Wilson DA. (1999) Paroxetine in human milk. Br J Clin Pharmacol. 48:142-147.

Belpaire FM, Wynant P, Van Trappen P, Dhont M, Verstraete A, Bogaert M. (1995) Protein binding of and enantiomers in maternal and foetal serum. Br J Clin Pharmacol. 39: 190-193.

Benfield P, Heel RC, Lewis SP. (1986) Fluoxetine: A review of its pharmacodynamic and pharmacokinetic properties, and therapeutic efficacy in depressive illness. Drugs. 32:481- 508.

Page 288 Bergstrom RF, Beasley CM, Levy NB, Blumenfield M, Lemberger L. (1993) The effects of renal and hepatic disease on the pharmacokinetics, renal tolerance and risk-benefit profile of fluoxetine. Int Clin Psychopharmcol. 8:261-266.

Bergstrom RF, Lemberger L, Farid NA, Wolen RL. (1988) Clinical pharmacology and pharmacokinetics of fluoxetine: A review. Br J Psychiatry. 153(Suppl): 47-50.

Bergstrom RF, Wolen R, Lemberger L, Dhahir P, Barret JL. (1985) Fluoxetine single dose- multiple dose kinetics. Am Pharm Assoc Acad Pharm Sci. 15: 137 (Abstract).

Berman W Jr, Goodlin RC, Heymann MA, Rudolph AM. (1978) Effects of pharmacologic agents on umbilical blood flow in fetal lambs in utero. Biol Neonate 33:225-235.

Besunder JB, Reed MD, Blumer JL. (1988) Principles of biodisposition in the neonate. Clin Pharmacokinetics. 14:189-216.

Bloedow DC, Ralston DH, Hargrove JC. (1980) Lidocaine pharmacokinetics in pregnant and nonpregnant sheep. J Pharm Sci. 69:32-37.

Bloom SL, Dias KM, Bawdon RE, Gilstrap LC 3rd. (1997) The maternal-fetal transfer of lamivudine in the ex vivo human placenta. Am J Obstet Gynecol. 176:291-293.

Bloomer JC, Woods FR, Haddock RE, Lennard MS, Tucker GT. (1992) The role of cytochrome P4502D6 in the metabolism of paroxetine by human liver microsomes. Br J Clin Pharmacol. 33:521-523.

Bologa M, Koren G, Fassos FF, McGuigan M, Reider MJ. (1994) Drugs and chemicals most commonly used by pregnant women, In: Maternal-fetal toxicology- a clinician's guide (Koren G, Ed.), New York: Marcel Dekker. Chapter 7, pp. 89-113.

Bonati M, Bortolus R, Marchetti F, Romero M, Tognoni G. (1990) Drug use in pregnancy: an overview of epidemiological (drug utilization) studies. Eur J Clin Pharmacol. 38:325- 328.

Bourget P, Roulot C, Fernandez H. (1995) Models for placental transfer studies of drugs. Clin Pharmacokinet. 28:161-80.

Bradford MM. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72:248- 254.

Brashear WT, Kuhnert BR, Wei R. (1988) Maternal and neonatal urinary excretion of sulfate and glucuronide ritodrine conjugates. Clin Pharmacol Ther. 44:634-641.

Bremner JD. (1984) Fluoxetine in depressed patients: a comparison with imipramine. J Clin Psychiatry. 45:414-149.

Page 289 Brett MA, Dierdorf HD, Zussman BD, Coates PE. (1987) Determination of paroxetine in human plasma, using high-performance liquid chromatography with fluorescent detection. J Chromatogr. 419:438-444.

Briggs GG, Freeman RK, Yaffe SJ. (1994) Drugs in pregnancy and lactation; A reference guide to fetal and neonatal risk. 4th Edition, Williams and Wilkins, NY.

Briley MS, Langer SZ, Raisman R, Sechter D, Zarifian E. (1980) Tritiated-imipramine binding sites are decreased in platelets of untreated depressed patients. Science. 209:303- 305.

Brockington IF, Cernik KF, Schofield EM, Downing AR, Francis AF, Keelan C. (1981) Puerperal , phenomena and diagnosis. Arch Gen Psychiatry. 38:829-833.

Brosen K, Skjelbo E. (1991) Fluoxetine and norfluoxetine are potent inhibitors of P450IID6— the source of the sparteine/debrisoquine oxidation polymorphism. Br J Clin Pharmacol. 32:136-137.

Brown HL, Miller JM Jr, Neumann DE, Sarpong DF, Gabert HA. (1990) Umbilical cord blood gas assessment of twins. Obstet Gynecol. 75:826-829.

Buist A, Norman TR, Dennerstein L. (1990) and the use of psychotropic medication: A review. J Affect Disord 19:197-206.

Burch KJ, Wells BG. (1992) Fluoxetine/norfluoxetine concentrations in human milk. Pediatrics. 89:676-677.

Cabrera-Vera TM, Battaglia G. (1998) Prenatal exposure to fluoxetine (Prozac) produces site-specific and age-dependent alterations in brain serotonin transporters in rat progeny: evidence from autoradiographic studies. J Pharmacol Exp Ther. 286:1474-1481.

Cabrera-Vera TM, Garcia F, Pinto W, Battaglia G. (1997) Effect of prenatal fluoxetine (Prozac) exposure on brain serotonin neurons in prepubescent and adult male rat offspring. J Pharmacol Exp Ther. 280:138-145.

Caccia S, Cappi M, Fracasso C, Guiso G, Sarati S, Garattini S. (1990) Influence of dose and route of administration on the kinetics of fluoxetine and its metabolite norfluoxetine in the rat. (Berl). 100:509-514.

Caccia S, Fracasso C. (1992) Effects of short- and long-term administration of fluoxetine on monoamine content of rat brain. Neuropharmacology. 31:343-347.

Canadian Paediatric Society Nutrition Committee. (1978) Breast-feeding: What is left besides the poetry. Can J Public Health. 69:13-19.

Canadian Paediatric Society Nutrition Committee. (1979) Infant feeding. Can J Public Health. 70:376-385.

Page 290 Carpenter SP, Lasker JM, Raucy JL. (1996) Expression, induction, and catalytic activity of the ethanol-inducible cytochrome P450 (CYP2E1) in man fetal liver and hepatocytes. Mol Pharmacol. 49:260-268.

Catterson ML, Preskorn SH. (1996) Pharmacokinetics of selective serotonin reuptake inhibitors: clinical relevance. Pharmacol Toxicol. 78:203-208.

Chambers CD, Johnson KA, Dick LM, Felix RJ, Jones KL (1996) Birth outcomes in women taking to fluoxetine. N Eng J Med. 335:1010-1015.

Chasnoff IJ. (1988) Drug use in pregnancy: parameters of risk. Pediatric Clin North Am. 35:1403-1412.

Chen Y, Yu SZ, Li WX. (1988) Artificial feeding and hospitalization in the first 18 months of life. Pediatrics. 81:58-62.

Chien T, Morrison J, Gruber N, Rurak DW, Riggs KW (1999) Chronic exposure and pharmacokinetics of fluoxetine and norfluoxetine enantiomers in pregnant ewe and fetus. Pharm Res. 1: S224.

Chiriboga CA. (1993) Fetal effects. Neurol Clin. 11:707-28.

Clark KE, Mills EG, Otte TE, Stys SJ (1980) Effect of serotonin on uterine blood flow in pregnant and non-pregnant sheep. Life Sci 27:2655-2661.

Cogill SR, Caplan HL, Alexandra H, Robson KM, Kumar R. (1986) Impact of maternal postnatal depression on cognitive development of young children. Br Med J (Clin Res Ed) 292:1165-1167.

Cohen LS, Rosenbaum JF. (1996) Birth outcomes in pregnant women taking fluoxetine, letter to the editor. N Eng J Med. 336:872.

Coleman FH, Christensen HD, Gonzalez CL, Rayburn WF. (1999) Behavioral changes in developing mice after prenatal exposure to paroxetine (Paxil). Am J Obstet Gynecol. 181:1166-1171.

Collaborative Group on Drug Use in Pregnancy (C.G.D.U.P.). (1992) Medication during pregnancy: an intercontinental cooperative study. Int J Gynaecol Obstet. 39:185-196.

Comline RS, Silver M. (1974). Recent observations on the undisturbed feuts in utero and its delivery. In: Recent Advances in Physiology. (Linden RJ, Ed). Edinburg: Churchill Livingstone, p.405-427.

Committee on Nutrition of the American Academy of Pediatrics. (1992) The use of whole cow's milk in infancy. Pediatrics. 89:1105-1108.

Cox JL, Murray L, Chapman G. (1993) A controlled study of the onset, duration and prevalence of postnatal depression. Br J Psychiatry. 163:27-31.

Page 291 Crifasi JA, Le NX, Long C. (1997) Simultaneous identification and quantification of fluoxetine and its metabolite, norfluoxetine, in biological samples by GC-MS. J Anal Toxicol. 21:415-419.

Cruz MA, Gonzalez C, Gallardo V, Lagos M, Albornoz J. (1993) Venous placental reactivity to serotonin in normal and preeclamptic gestants. Gynecol Obstet Invest 36:148-152.

Cunningham AS, Jelliffe DB, Jelliffe EF. (1991) Breast-feeding and health in the 1980s: A global epidemiological review. J Pediatrics. 118: 659-666.

Dahl ML, Olhager E, Ahlner J. (1997) Paroxetine withdrawal syndrome in a neonate. Br J Psychiatry. 171:391-392.

Daniel WA, Wojcikowski J. (1997a) Contribution of lysosomal trapping to the total tissue uptake of psychotropic drugs. Pharmacol Toxicol. 80:62-68.

Daniel WA, Wojcikowski J. (1997b) Interactions between and antidepressants at the level of cellular distribution. Pharmacol Toxicol. 81:259-264.

Dawes GS. (1994). Fetal physiology: Historical perspectives. In: Textbook of Fetal Physiology, (eds Thorburn GD, Harding R), Oxford, Oxford University Press, pp 1-5. de Jong-van den Berg LT, Waardenburg CM, Haaijer-Ruskamp FM, Dukes MN, Wesseling H. (1993) Drug use in pregnancy: a comparative appraisal of data collecting methods. Eur J Clin Pharmacol. 45:9-14. de Vries JIP, Visser GHA, Prechtl HFR. (1982) The emergence of fetal behavior. I. Qualitative aspects. Early Hum Dev. 7:301-322. de Vries JIP, Visser GHA, Prechtl HFR. (1985) The emergence of fetal behavior. II. Quantitative aspects. Early Hum Dev. 12:99-120.

Dechant KL, Clissold SP. (1991) Paroxetine. A review of its pharmacodynamic and pharmacokinetic properties, and therapeutic potential in depressive illness. Drugs. 41:225-253.

DeVane CL. (1994) Pharmacokinetics of the newer antidepressants: clinical relevance. Am J Med. 97:13S-23S.

DeVane CL. (1992) Pharmacokinetics of the selective serotonin reuptake inhibitors. J Clin Psych. 53(Suppl): 13-20.

Dixit V, Nguyen H, Dixit VM. (1991) Solid-phase extraction of fluoxetine and norfluoxetine from serum with gas chromatography-electron capture detection. J Chromatogr B Biomed Sci Appl. 563:379-384.

Page 292 Dobbing J. (1976) Vulnerable periods in brain growth and somatic growth. In: The biology of fetal human growth. (Roberts DF, Thomson EM, Eds.) London: Taylor and Francis Ltd. pp. 137-147.

Dutton GJ. (1978) Developmental aspects of drug conjugations, with special reference to glucuronidation. Ann Rev Pharmacol Toxicol. 18:17-35.

Eap CB, GaillardN, Powell K, Baumann P. (1996) Simultaneous determination of plasma levels of fluvoxamine and of the enantiomers of fluoxetine and norfluoxetine by gas chromatography-mass spectrometry. J Chromatogr B Biomed Sci Appl. 682:265-272.

Edward A. (1981) Drugs in breast milk - a review of recent literature. Aust J Hosp Pharmacy. 11:27-38.

Faber JJ, Thornburg KL. (1983) Placental Physiology. New York: Raven Press, pp. 79-89.

Field T, Sandberg D, Garcia R, Vega Lhar N, Goldstein S, Guy L. (1985) Prenatal problems, postpartum depression and early mother-infant interactions. Developmental Psychol. 12:1152-1156.

Fjordside L, Jeppensen U, Eap CB, Powell K, Baumann P, Brosen K. (1999) The stereoselective metabolism of fluoxetine in poor and extensive metabolizers of sparteine. Pharmacogenetics 9:55-60.

Fontanille P, Jourdil N, Villier C, Bessard G. (1997) Direct analysis of fluoxetine and norfluoxetine in plasma by gas chromatography with nitrogen-phosphorus detection. J Chromatogr B Biomed Sci Appl. 692:337-343.

Ford K, Labbok M. (1990) Who is breast feeding? Implications of associated social and biomedical variables for research on the consequences of method of infant feeding. Am J ClinNutr. 52:451-456.

Fuller RW, Snoddy HD, Krushinski JH, Robertson DW. (1992) Comparison of norfluoxetine enantiomers as serotonin uptake inhibitors in vivo. Neuropharmacology. 31:997-1000.

Fuller RW, Snoddy HD, Perry KW, Bymaster FP, Wong DT. (1978) Importance of duration of drug action in the antagonism of p-chloroamphetamine depletion of brain serotonin - comparison of fluoxetine and chlorimipramine. Biochem Pharmacol. 27:193-198.

Fuller RW, Snoddy HD. (1993) Drug concentrations in mouse brain at pharmacologically active doses of fluoxetine enantiomers. Biochem Pharmacol. 45(11):2355-2358.

Fuller RW, Wong DT, Robertson DW. (1991) Fluoxetine, a selective inhibitor of serotonin uptake. Med Res Rev. 11:17-34.

Galtier P, Alvinerie M. (1996) Pharmacological basis for hepatic drug metabolism in sheep. Vet Res. 27:363-372.

Page 293 Garland M, Szeto HH, Daniel SS, Tropper PJ, Myers MM, Stark RI. (1998) Placental transfer and fetal metabolism of zidovudine in the baboon. Pediatr Res. 44: 47-53.

Gibaldi M, Perrier D. (1982) Pharmacokinetics, 2nd Ed. New York: Marcel-Dekker.

Goldstein DJ, Corbin LA, Sundell KL. (1997) Effects of first-trimester fluoxetine exposure on the newborn. Obstet Gynecol. 89:713-718.

Goldstein DJ, Marvel DE. (1993) Psychotropic medication during pregnancy: risk to the fetus. JAMA. 270:2177.

Goldstein DJ, Sundell KL, Corbin LA. (1996) Birth outcomes in pregnant women taking fluoxetine, letter to the editor. N Eng J Med. 336: 872-873.

Goldstein DJ, Williams ML. (1992) Fluoxetine exposed pregnancies. Clin Res. 40:168A (Abstract).

Goldstein DJ. (1990) Outcome of fluoxetine-exposed pregnancies. Am J Human Genetics. 47:A136 (Abstract).

Goldstein DJ. (1991) Fluoxetine-exposed pregnancies. Clin Res. 39:768A (Abstract).

Goldstein DJ. (1995) Effects of third trimester fluoxetine exposure on the newborn. J Clin Psychopharmacol. 15: 417-420.

Golub MS and Kaaekuahiwi MA. (1997) Changes in plasma a 1-acid glycoprotein and albumin concentrations during late pregnancy in rhesus monkeys. Clin Chim Acta. 262:29-37.

Goodnick PJ. (1991) Pharmacokinetics of second generation antidepressants: fluoxetine, Psychopharmacology Bulletin. 27:503-512.

Goodnough DB, Baker GB, Coutts RT. (1995) Simultaneous quantification of fluoxetine, norfluoxetine and desipramine using gas chromatography with nitrogen-phosphorous detection. J Pharmacol Toxicol Methods. 34:143-147.

Gordon JD, Riggs KW, Rurak DW, Kwan E, Hall C, Abbott FS. (1995) The pharmacokinetics of valproic acid in pregnant sheep after maternal and fetal intravenous bolus administration. Drug Metab Dispos. 23:1383-1389.

Grimsley SR, Jann MW. (1992) Paroxetine, sertraline and fluvoxamine: new selective serotonin reuptake inhibitors. Clin Pharm. 11:930-957.

Gupta RN. (1994) Column liquid chromatographic determination of paroxetine in human serum using solid-phase extraction. J Chromatogr B Biomed Sci Appl. 661:362-365.

Hakkola J, Pasanen M, Pelkonen O, Hukkanen J, Evisalmi S, Anttila S, Rane A, Mantyla M, Purkunen R, Saarikoski S, Tooming M, Raunio H. (1997) Expression of CYP1B1 in

Page 294 human adult and fetal tissues and differential inducibility of CYP1B1 and CYP1A1 by Ah receptor ligands in human placenta and cultured cells. Carcinogenesis. 18:391-397.

Hakkola J, Pasanen M, Purkunen R, Saarikoski S, Pelkonen O, Maenpaa J, Rane A, Raunio H. (1994) Expression of xenobiotic-metabolizing cytochrome P450 forms in human adult and fetal liver. Biochem Pharmacol. 48:59-64.

Hakkola J, Pelkonen O, Pasanen M, Raunio H. (1998) Xenobiotic-metabolizing cytochrome P450 enzymes in the human feto-placental unit: role in intrauterine toxicity. Crit Rev Toxicol. 28:35-72.

Hamelin BA, Turgeon J, Vallee F, Belander PM, Paquet F, LeBel M. (1996) The disposition of fluoxetine but not sertraline is altered in poor metabolizer of debrisoquin. Clin Pharmacol Ther. 60:512-521.

Hamosh M. (1996) Breast-feeding: Unraveling the mysteries of mother's milk. Medscape Womens Health. 1:4.

Hamshaw-Thomas A, Rogerson N, Reynolds F. (1984) Transfer of bupivacaine, lignocaine and across the rabbit placenta: influence of maternal protein binding and fetal flow. Placenta. 5:61-70.

Haram K, Augensen K and Eslayed S. (1983) Serum protein pattern in normal pregnancy with special reference to acute-phase reactants. Br J Obstet Gynaecol. 90:139-145.

Harding R, Sigger JN, Wickham PJ. (1983) Fetal and maternal influences on arterial oxygen levels in the sheep fetus. J Dev Physiol. 5:267-276.

Hartter S, Hermes B, Szegedi A, Hiemke C. (1994) Automated determination of paroxetine and its main metabolite by column switching and on-line high-performance liquid chromatography. Ther Drug Monit. 16:400-406.

Hauth JC, Goldenberg RL, Parker CR Jr, Philips JB 3d, Copper RL, DuBard MB, Cutter GR. (1993) Low-dose aspirin therapy to prevent preeclampsia. Am J Obstet Gynecol. 168:1083-1091.

Hiemke C, Hartter S. (2000) Pharmacokinetics of selective serotonin reuptake inhibitors. Pharmacol Ther. 85:11-28.

Hill MD, Abramson FP. (1988) The significance of plasma protein binding on the fetal/maternal distribution of drugs at steady-state. Clin Pharmacokinet 14:156-170.

Ho T, Costei A, Ito S. (2000) Maternal paroxetine use and neonatal withdrawal. Clin Pharmacol Ther. 67: 105A (Abstract).

Holladay JW, Dewey MJ, Yoo SD. (1998) Pharmacokinetics and antidepressant activity of fluoxetine in transgenic mice with elevated serum alpha-1-acid glycoprotein levels. Drug Metab Dispos. 26:20-24.

Page 295 Hopkins J, Marcus M, Campbell SB. (1984) Postpartum depression: a critical review. Psychol Bull. 95:498-515.

Huang SCH, Boudinot FD, Feldman S. (1996) Maternal-fetal pharmacokinetics of zidovudine in rats. J Pharm Sci. 85:965-970.

Hutchings DE. (1978) Early influences, In: Studies on the development of behaviour and nervous system, vol. 4. (Gottleib G, Ed). New York: Academic press, pp. 7-34.

Hytten FE. (1954) Clinical and chemical studies in human lactation. Br Med J. 1:175-182.

Iri C, Hasford J. (1997) The PEGASUS project~a prospective cohort study for the investigation of drug use in pregnancy. PEGASUS Study Group. Int J Clin Pharmacol Ther. 35:572-576.

Isenberg KE. (1990) Excretion of fluoxetine in human breast milk. J Clin Psychiatry. 51(4): 169.

Jacqz-Aigrain E, Cresteil T. (1992) Cytochrome P450-dependent metabolism of dextromethorphan: fetal and adult studies. Dev Pharmacol Ther. 8:161-168.

Jamali F, Mehvar R, Pasutto FM. (1989). Enatioselective aspects of drug action and disposition: Therapeutic pitfalls. J Pharm Sci. 78:695-715.

Karson CN, Newton JE, Livingstone R, Jolly JB, Cooper TB, Sprigg K, Komoroski RA. (1993) Human brain fluoxetine concentrations. J Neuropsychiatry Clin Neurosci. 5:322- 329.

Kaye CM, Haddock RE, Langley PF, Mellows G, Tasker TC, Zussman BD, Greb WH. (1989) A review of the metabolism and pharmacokinetics of paroxetine in man. Acta Psychiatrica Scandiavica Suppl. 350:60-75.

Kim J, Axelson JE, Kearns GL, Yin W, Rurak DW. (1995) Stereoselective determination of fluoxetine and norfluoxetine by gas chromatography with mass selective detection (GC/MSD). Pharm Res. 12:S22 (Abstract).

Kim J, Burton R, Rurak DW, Riggs KW. (1998) Stereoselective quantitation of fluoxetine and norfluoxetine in plasma by LC/MS/MS. PharmSci. 1 :S185 (Abstract).

Kim J, Misri S, Kent N, Oberlander T, Rurak DW, Riggs KW (1999b) Comparison of fetal and neonatal exposure during the perinatal period in humans. AAPS PharmSci. 1:S96 (Abstract).

Kim J, Riggs KW, Axelson JE, Rurak DW. (1996) Stereoselective pharmacokinetics of fluoxetine and norfluoxetine in the sheep. Pharm Res. 13:S494 (Abstract).

Page 296 Kim J, Riggs KW, Kim Y, Rurak DW. (1997) Determination of paroxetine (Paxil®) in human serum and milk by gas chromatography/mass spectrometry (GC/MS). Pharm Res. 14:S683 (Abstract).

Kim J, Rurak DW, Riggs KW. (1999a) Stereoselective pharmacokinetics of fluoxetine and norfluoxetine during pregnancy in sheep. AAPS PharmSci. l:S223-224 (Abstract).

Komoroski RA, Newton JE, Cardwell D, Sprigg J, Pearce J, Karson CN. (1994) In vivo 19F spin relaxation and localized spectroscopy of fluoxetine in human brain. Magnetic Resonance Med. 31:204-211.

Komoroski RA, Newton JEO, Karson C, Cardwell D, Sprigg J. (1991) Detection of psychoactive drugs in vivo in human using 19F spectroscopy. Biol Psychiatry. 29:711- 714.

Krauer B, Dayer P, Anner R. (1984) Changes in serum albumin and alpha 1-acid glycoprotein concentrations during pregnancy: an analysis of fetal-maternal pairs. Br J Obstet Gynaecol. 91:875-881.

Krauer B, Krauer F. (1977) Drug kinetics in pregnancy. Clin Pharmacokinet. 2(3): 167-81.

Krishna RB, Levitz M, Dancis J. (1993) Transfer of by the perfused human placenta: the effect of binding to serum proteins. Am J Obstet Gynecol. 169:1418-1423.

Kristensen JH, Ilett KF, Hackett LP, Yapp P, Paech M, Begg EJ. (1999) Distribution and excretion of fluoxetine and norfluoxetine in human milk. Br J Clin Pharmacol. 48:521- 527.

Krogh CME. (1995) Compendium of Pharmaceuticals and Specialties 30th Ed. Ottawa, Ontario:Canadian Pharmaceutical Association.

Kulin NA, Pastuszak A, Sage SR, Schick-Boschetto B, Spivey G, Feldkamp M, Ormond K, Matsui D, Stein-Schechman AK, Cook L, Brochu J, Rieder M, Koren G. (1998) Pregnancy outcome following maternal use of the new selective serotonin reuptake inhibitors. JAMA. 279: 609-610.

Kumar R, Robinson K. (1978) Neurotic disturbance during pregnancy and the puerperium: preliminary report of a prospective survey of 119 primiparae. In: Mental illness in pregnancy and the puerperium. (Sander M, Ed.). LondomOxford Univ Press, pp. 40-51.

Kumar S, Wong H, Yeung SA, Riggs KW, Abbott FS, Rurak DW. (2000a) Disposition of valproic acid in maternal, fetal, and newborn sheep I: placental transfer, plasma protein binding, and clearance. Drug Metab Dispos. 28:845-856.

Kumar S, Wong H, Yeung SA, Riggs KW, Abbott FS, Rurak DW. (2000b) Disposition of valproic acid in maternal, fetal, and newborn sheep II: metabolism and renal elimination. Drug Metab Dispos. 28:857-864.

Page 297 Kurtz H, Mauser-Ganshorn A, Stickel HH. (1977a) Difference in the binding of drug plasma proteins from newborn and adult man Part I. Eur J Clin Pharmacol. 11.463-467.

Kurtz H, Mauser-Ganshorn A, Stickel HH. (1977b) Difference in the binding of drug plasma proteins from newborn and adult man Part II. Eur J Clin Pharmacol. 11.469-472.

Kuss HJ, Hegerl U. (1998) Serum concentration of paroxetine are influenced by CYP3A4. Naunyn Schmiedeberg Arch Pharmacol. 358: R782 (Abstract).

Kwan E, Rurak DW, Taylor SM. (1995a) Blood volume restriction and growth in fetal lambs after acute hemorrhage. Am J Physiol. 269:R749-757.

Kwan E, Rurak DW, Taylor SM. (1995b) Oxygen consumption, acid-base status, and behavior during and after acute, severe hemorrhage in fetal lambs. Am J Physiol 269:R758-766.

Kyle PM, Redman CW. (1992) Comparative risk-benefit assessment of drugs used in the management of hypertension in pregnancy. Drug Saf. 7:223-234.

Laishley RS, Carson RJ, Reynolds F. (1989) Effects of adrenaline on the distribution of bupivacaine in the rabbit fetus. Br J Anaesth 63:721-725.

Lantz RJ, Farid KZ, Koons J, Tenbardge JB, Bopp RJ. (1993) Determination of fluoxetine and norfluoxetine in human plasma by capillary gas chromatography with electron capture detection. J Chromatog. 614:175-179.

Lapicque F, Muller N, Payan E, Dubois N, Netter P. (1993) Protein binding and stereoselectivity of anti-inflammatory drugs. Clin Pharmacokinet. 25:115- 123.

Lemberger L, Bergstrom RF, Wolen RT, Farid NA, Enas GG, Aronoff GR. (1985) Fluoxetine: clinical pharmacology and physiological disposition. J Clin Psychiatry. 46:14-19.

Leonard BE. (1993) The comparative pharmacology of new antidepressants. J Clin Psychiatry Suppl. 54:3-15.

Lester BM, Cucca J, Andreozzi L, Flanangan P, Oh W. (1993) Possible association between fluoxetine hydrochloride and colic in an infant. J Am Acad Chil Adolesc Psychiatry. 32:1253-1255.

Lewis PJ. (1978) The effect of psychotropic drugs in the fetus. In: Mental illness in pregnancy and the puerperium. (Sandler M, Ed.). London: Oxford Univ Press.

Loebstein R, Lalkin A, Koren G. (1997) Pharmacokinetic changes during pregnancy and their clinical relevance. Clin Pharmacokinet. 33(5):328-43.

Page 298 Lund J, Lomholt B, Fabricius J, Christensen JA, Bechgaard E. (1979) Paroxetine: pharmacokinetics, tolerance and depletion of blood 5-HT in man. Acta Pharmacol Toxicol (Copenh). 44:289-95.

Lyons-Ruth K, Zoll D, Connell D, Grunebaum HU. (1986) The depressed mother and her one-year-old infant: environment, interaction, attachment and infant development. In: Maternal depression and infant disturbance (Tronick EZ, Field T, Eds.). San Francisco: Jossey-Bass. pp 61-82.

Magnussen I, Tonder K, Engbaek F. (1982) Paroxetine, a potent selective long-acting inhibitor of synaptosomal 5-HT uptake in mice. J. Neural Transmission. 55:217-226.

Mannering GJ. (1985) Drug metabolism in the newborn. Fed Proc. 44:2303-2308.

Mathesen I, Skajaeraasen J. (1988) Milk concentration of flupenthixol, and and between breast differences in two patients. Eur J Clin Pharmacol. 35:217-220.

McElhatton RP. (1994) The effect of benzodiazepine use during pregnancy and lactation. Reprod Toxicol. 8:461 -475.

McKnew DH. Cytryn L, Efron AM, Gershon ES, Bunney WE. (1979) Offspring of patients with affective disorders. Br J Psychiatry. 134:148-152.

McNally E, Hendricks S, Horowitz I. (1985) A look at breast-feeding trends in Canada (1963-1982). Can J Public Health. 76:101-107.

Mhanna MJ, Bennet JB, Izatt SD. (1997) Potential fluoxetine chloride (Prozac) toxicity in a newborn. Pediatrics. 100:158-159.

Miles MV. (1983) Pediatric pharmacokinetics. In: Applied clinical pharmacokinetics. (Mungall, Ed.). New York: Raven Press, pp.367-388.

Milne RW, Sloan PA, McLean CF, Mather LE, Nation RL, Runciman WB, Rutten AJ, Somogyi AA. (1993) Disposition of morphine and its 3- and 6-glucuronide metabolites during morphine infusion in the sheep. Drug Metab Dispos. 21:1151-1156.

Miningco.com (1998) Prozac® improved (R-fluoxetine) for next-generation antidepressants 12/07/98. http://pharmacologv.miningco.com/library/98news/blnl207a.htm.

Misri S, Ryan D, Carter D, Kuan A. (1996) Personal communication.

Misri S, Sivertz K. (1991) Tricyclic drugs in pregnancy and lactation: a preliminary report. Int J Psychiatry Med. 21:157-171.

Mitani GM, Steinberg I, Lien EJ, Harrison EC, Elkayam U. (1987) The pharmacokinetics of antiarrhythmic agents in pregnancy and lactation. Clin Pharmacokinet. 12:253-291.

Page 299 Moise KJ Jr, Kirshon B, Carpenter RJ Jr, Deter RL. (1990) Icteric serum as an indicator of fetal vascular access during intravascular transfusion for hemolytic disease. Am J Perinatol. 7:279-280.

Montero D, de Ceballos ML, Del Rio J. (1990) Down-regulation of 3H-immipramine binding sites in rat cerebral cortex after prenatal exposure to antidepressants. Life Sci. 46:1619- 1626.

Morrison J, Chien C, Gruber N, Riggs KW, Rurak DW. (1999) Acute effects of maternal fluoxetine administration on mothers and fetus in the sheep. Presented at Fetal and Neonatal Physiological Society, 26th Annual Meeting.

Morrison J, Chien C, Gruber N, Riggs KW, Rurak DW. (1999) IV maternal fluoxetine causes changes in sheep fetal behavioral state. Presented at World Federation of Sleep Research Societies. October 4-10, 1999.

Morselli PL, Franco-Morselli R, Bossi L. (1980) Clinical pharmacokinetics in newborns and infants: age-related differences and therapeutic implications. Clin Pharmacokinet. 5:487- 527.

Mortola JF. (1989) The use of psychotrophic agents in pregnancy and lactation. Psych Clinics N Am 12(1): 69-87.

Mote PL, Grizzle JM, Walford RL, Spindler SR. (1991) Influence of age and caloric restriction on expression of hepatic genes for xenobiotic and oxygen metabolizing enzymes in the mouse. J Gerontol. 46: B95-B100.

Murata Y. (1982) Advances on the horizon. Clinics Perinatol. 9:433-441.

Murray L, Seger D. (1994) Drug therapy during pregnancy and lactation. Emerg Med Clin North Am. 12:129-149.

Murray L. (1992) The impact of postnatal depression on infant development. J Child Psychol Psychiatry. 33: 543-561.

Nahas C, Goujard J. (1978) , benzodiazepines and the fetus. In Reviews in prenatal medicine (Scarpelli EM, Cosmi EV Eds.). New York: Raven Press, pp. 243-280.

Nelson D. (1999) Cytochrome P450 homepage. http://drnelson.utmem.edu/CytochromeP450.html

Nemeroff CB, Knight DL, Krishnan RR, Slotkin TA, Bissette G, Melville ML, Blazer DG. (1988) Marked reduction in the number of platelet-tritiated imipramine binding sites in geriatric depression. Arch Gen Psychiatry. 45:919-923.

Nemeroff CB. (1994) The clinical pharmacology and use of paroxetine, a new selective serotonin reuptake inhibitor. Pharmacotherapy. 14:127-138.

Page 300 Neville MC, Keller R, Seacat J, Lutes V, Neifert M, Casey C, Allen J, Archer P. (1988) Studies in human lactation: milk volumes in lactating women during the onset of lactation and full lactation. Am J Clin Nutr. 48:1375-1386.

Nichols M. (1994) Questioning Prozac. MacLean's - Canada's weekly newsmagazine, May 23, 1994, Toronto, pp. 36-40.

Nijhuis JG, Prechtl HFR, Martin CB Jr., Bots RSGM. (1982). Are there behavioural states in the human fetus? Early Hum Dev. 6:177-195.

Nijhuis JG. (1992) Fetal behavior: Developmental and perinatal aspect. Oxford Uuniv Press, Oxford, UK.

Nims RW, Prough RA, Jones CR, Stockus DL, Dragnev KH, Thomas PE, Lubet RA. (1997) In vivo induction and in vitro inhibition of hepatic cytochrome P450 activity by the benzodiazepine anticonvulsants clonezapam and . Drug Metabol Disp. 25:750- 756.

Nulman I, Koren G. (1996) The safety of fluoxetine during pregnancy and lactation. Teratrology. 53:304-308.

Nulman I, Rovet J, Steward DE, Wolpin J, Gardner HA, Theis JGW, Kulin N, Koren G. (1997) Neurodevelopment of children exposed in utero to antidepressant drugs. N Eng J Med. 336:258-262.

O'Hara MW, Zekoski EM. (1988) Postpartum depression: A comprehensive review. In: Motherhood and Mental Illness. (Kumar R, Brockington IF, Eds.) London: J Wright, pp. 17-63.

O'Hara MW, Neunober DJ, Zekoski EH. (1984) Prospective study of postpartum depression: prevalence and predictive factors. J Abnorm Psycol. 93:158-171.

Ohman R, Hagg S, Carleborg L, Spigset O. (1999) Excretion of paroxetine into breast milk. J Clin Psychiatry. 60:519-523.

Olsen GD, Sommer KM, Wheeler PL, Boyea SR, Michelson SP, Cheek DB. (1988) Accumulation and clearance of morphine-3-beta-D-glucuronide in fetal lambs. J Pharmacol Exp Ther. 247:576-584.

Oravcova J, Bohs B, Lindner W. (1996) Drug-protein binding sites. New trends in analytical and experimental methodology. J Chromatogr B Biomed Appl. 677:1-28.

Ortiz J, Aritigas F. (1992) Effects of monoamine uptake inhibitors on extracellular and platelet 5-hydroxytryptamine in rat blood: different effect of clomipramine and fluoxetine. Br J Pharmacol. 105:941-946.

Orvaschel H, Welsh-Allis G, Ye WJ. (1988) Psychopathology in children of parents with recurrent depression. J Abnormal Child Psychol. 1:17-28.

Page 301 Otton SV, Wu D, Joffe RT, Cheung SW, Sellers EM. (1993) Inhibition by fluoxetine of cytochrome P450 2D6 activity. Clin Pharmacol Ther. 53: 401-409.

Pacifici GM, Viani A. (1992) Methods of determining plasma and tissue binding of drugs. Pharmacokinetic consequences. Clin Pharmacokinet. 23:449-468.

Pai VB, Kelly MW. (1996) Bruising associated with the use of fluoxetine. Ann Pharmacother 30:786-788.

Parvez H, Ismahan G, Parvez S, Youdim MB. (1979) Developmental changes in the activity of catechol-O-methyl transferase in rat and rabbit fetuses. J Neural Transm. 44:65-75.

Passmore CM, McElnay JC, D'arcy PF. (1984) Drugs taken by mothers in the puerperium: in-patient surveys in Northern Ireland. Br Med J. 289:1593-1596.

Pastuszak A, Schick-Boschetto B, Zuber C, Feldkamp M, Pinelli M, Sihn S, Donnenfeld A, McCormack M, Leen-Mitchell M, Woodland C, Gardner A, Horn M, Koren G. (1993) Pregnancy outcome following first-trimester exposure to fluoxetine (Prozac®). JAMA. 269:2246-2248.

Pato MT, Murphy DL, DeVane CL. (1991) Sustained plasma concentration of fluoxetine and/or norfluoxetine four and eight weeks after fluoxetine discontinuation. J Clin Psychopharmacol. 11:224-225.

Patterson TA, Binienda ZK, Lipe GW, Gillam MP, Slikker W Jr, Sandberg JA. (1997) Transplacental pharmacokinetics and fetal distribution of azidothymidine, its glucuronide, and phosphorylated metabolites in late-term rhesus macaques after maternal infusion. Drug Metab Dispos. 25:453-459.

Pedersen H, Morishima HO, Finster M, Arthur GR, Covino BG. (1982) Pharmacokinetics of etidocaine in fetal and neonatal lambs and adult sheep. Anesth Analg. 61:104-108.

Perucca E, Crema A. (1982) Plasma protein binding of drugs in pregnancy. Clin Pharmacokinet. 7(4):336-52. Petersen EN, Bechgaard E, Sortwell RJ, Wetterberg L. (1978) Potent depletion of 5HT from monkey whole blood by a new 5HT uptake inhibitor, paroxetine (FG 7051). Eur J Pharmacol. 52:115-119.

Peyton AL, Carpenter R, Rutkowski K. (1991) The stereospecific determination of fluoxetine and norfluoxetine enantiomers in human plasma by high-pressure liquid chromatography (HPLC) with fluorescence detection. Pharm Res. 8:1528-1532.

Physicians Desk Reference. (1992) Oradell, NJ: Medical Economics.

Pitt B. (1973) Maternity blues. Br J Psychiatry. 122:431-433.

Pohland RC, Byrd TK, Hamilton M, Koons JR. (1989) Placental transfer and fetal distribution of fluoxetine in the rat. Toxicol Appl Pharmacol. 98:198-205.

Page 302 Preskorn SH, Magnus RD. (1994) Inhibition of hepatic P-450 isoenzmyes by serotonin selective reuptake inhibitors: In vitro and in vivo findings and their implications for patient care. Pyschopharmacol Bull 30:251-259.

Ramaiya A, Karnes HT. (1997) Simultaneous measurement of serotonin and paroxetine in rat brain microdialysis by a single-pump column-switching technique. J Chromatogr B Biomed Sci Appl. 691:119-129.

Ramaiya A, March C, Karnes HT. (1997) Single pump column switching technique employing a flow gradient and wavelength programmed fluorescence for simultaneous monitoring of serotonin, fluoxetine and norfluoxetine in rat brain microdialysate. J Pharm Biomed Analysis. 15:729-738.

Rane A, Wilson JT. (1976) Clinical pharmacokinetics in infants and children. Clin Pharmacokinetics. 1:2-24.

Rasmussen F (1973) The mechanism of drug secretion into milk. In: Gali Ed. Dietary lipids and postnatal development. New York: Raven Press, pp. 231-245.

Rayburn WF, Andersen BD. (1982) Principles of perinatal pharmacology. In: Drug therapy in obstetrics and gynecology. (Rayburn WF, Zuspan, Eds). Norwalk, CT: Appleton- Century-Crofts.

Renshaw PF, Guimaraes AR, Fava M, Rosenbaum JF, Pearlman JD, Flood JG, Puopolo PR, Clancy K, Gonzalez RG. (1992) Accumulation of fluoxetine and norfluoxetine in human brain during therapeutic administration. Am J Psychiatry. 149:1592-1594.

Reynolds LP, Knott C. (1989) Pharmacokinetics in pregnancy and placental drug transfer. In: Oxford Reviews of Reproductive Biology (Milligan SR, Ed), New York: Oxford University Press, pp. 389-449.

Rich KJ, Edwards RJ, Foster JR, Boobis AR. (1990) Ontogeny of expression and inducibility of hepatic cytochrome P450IA2 in the rat and rabbit. Hum. Toxicol. 9:330-333.

Riesenman C. (1995) Antidepressant drug interactions and the cytochrome P450 system: A critical appraisal. Pharmacotherapy. 15:84S-99S.

Riggs KW, Axelson JE, Gruber NC, McErlane BA, McMorland GH, Rurak DW. (1988) Metoclopramide pharmacokinetics in pregnant and nonpregnant sheep. J Pharm Sci. 77:373-379.

Riggs KW, Rurak DW, Yoo SD McErlane BA, Taylor SM, McMorland GH, Axelson JE. (1987) Drug accumulation in lung fluid of the fetal lamb after maternal and fetal administration. Am J Obstet Gynecol. 157:1286-1291.

Riordan JM. (1997) The cost of not breastfeeding: A commentary. J Hum Lact. 13:93-97.

Page 303 Robertson DW, Krushinski JH, Fuller RW, Leander JD. (1988) Absolute configurations and pharmacological activities of the optical isomers of fluoxetine, a selective serotonin- uptake inhibitor. J Med Chem. 31:1412-1417.

Robinson GE, Stewart DE. (1986) Postpartum psychiatric disorders. CMAJ. 134:31-37.

Rogan WJ, Gladen BC. (1993) Breast-feeding and cognitive development. Early Hum Dev. 31:181-193.

Romero G, Toscano E, Del Rio J. (1994) Effect of prenatal exposure to antidepressants on 5- HT-stimulated phosphoinositide hydrolysis and 5-HT2 receptors in rat brain. Gen Pharmacol. 25:851-856.

Roy A, DeJong J, Linnoila M. (1989) Cerebrospinal fluid monoamine metabolites and suicidal behaviour in depressed patients: a 5-year follow-up study. Arch Gen Psychiatry. 46:609-612.

Rudolph MA, Heymann MA. (1980). Methods for studying the circulation of the fetus in utero. In: Animal Models in Fetal Medicine I. (Nathanielsz PW, Ed.). Amsterdam: Elsvier. pp 1-57.

Rurak DW, Wright MR, Axelson JE. (1991) Drug disposition and effects in the fetus. J Develop Physiol. 15:33-44.

Rurak DW, Yoo SD, Kwan E, Taylor SM, Axelson JE. (1988) Effect of diphenhydramine in the fetal lamb after maternal or fetal administration. J Pharmacol Exp Ther. 247:271-278.

Rurak DW. (1992) Fetal behavioral states: pathological alterations with drug/alcohol abuse. Semin Perinatal. 16:239-251.

Rurak DW. (1994) Fetal oxygenation, carbon dioxide homeostasis and acid-base balance. In: Textbook of Fetal Physiology (Harding R, Thorburn GD, Eds.). Oxford: Oxford University Press, pp. 131-139.

Saunders NR, Mollgard K. (1984) Development of the blood-brain barrier. J Dev Physiol. 6:45-57.

Schanker LS. (1962) Passage of drugs across body membrane. Pharm Rev. 14:501-530.

Schenker S, Bergstrom RF, Wolen RL, Lemberger L. (1988) Fluoxetine disposition and elimination in cirrhosis. Clin Pharmacol Ther. 44:353-359.

Schmider J, Greenblatt DJ, von Moltke LL, Karsov D, Shader RI. (1997) Inhibition of CYP2C9 by selective serotonin reuptake inhibitors in vitro: studies of phenytoin p- hydroxylation. Br J Clin Pharmacol. 44:495-498.

Scott KG, Urbano JC, Bowsy CA. (1991) Long-term psychoeducational outcome of prenatal substance exposure. Sem Perinatol. 15:317-323.

Page 304 Sepracor (1999) Company website, http://www.sepracor.eom/html/2.ice/index.html.

Seta N, Tissol B, Forestier F, Feger J, Daffos F and Durand G. (1991) Changes in a 1-acid glycoprotein serum concentrations and glycoforms in the developing human fetus. Clin Chim Acta. 203:162-176.

Shah VP, Midha KK, Dighe S , McGilveray IJ, Skelly JP, Yacobi A, Layloff T, Viswanathan CT, Coo CE, McDowall RD, Pittman KA, Spector S. (1992) Association of American Pharmaceutical Scientists, United States FDA, Canadian HPB, and Association of Official Analytical Chemists Conference Report on: Analytical Methods Validation: Bioavailability, Bioequivalence and Pharmacokinetic Studies. Pharm Res. 9:588-592.

Shen WW. (1995) Cytochrome P450 monooxygenases and interactions of psychotropic drugs: A five-year update. Int J Psych Med. 25:277-290.

Shuey DL, Sadler TW, Lauder JM. (1992) Serotonin as a regulator of craniofacial morphogenesis: site specific malformations following exposure to serotonin uptake inhibitors. Teratology. 46:367-378.

Sinclair DA. (1994) The effects of postnatal depression on the social behaviour of children during the transition to school entry. Doctoral dissertation, University of Cambridge, UK.

Sindrup SH, Brosen K, Gram LF, Hallas J, Skjelbo E, Allen A, Allen GD, Cooper SM, Mellows G, Tasker TC, et al. (1992a) The relationship between paroxetine and the spartein oxidation polymorphism. Clin Pharmcol Ther. 51:278-287.

Sindrup SH, Brosen K, Gram LF. (1992b) Pharmacokinetics of the selective serotonin reuptake inhibitor paroxetine: nonlinearrity and relation to the spartein oxidation polymorphism. Clin Pharmacol Ther. 51:288-295.

Sommi RW, Crismon L, Bowden CL. (1987) Fluoxetine: a serotonin-specific second- generation antidepressant. Pharmacotheraphy. 7:1-15.

Sonnier M., Cresteil T. (1998) Delayed ontogenesis of CYP1A2 in the man liver. Eur J Biochem. 251:893-898.

Spencer MJ. (1993) Fluoxetine hydrochloride (Prozac) toxicity in a neonate. Pediatrics. 92:721-722.

Spigset O, Carleborg L, Norstom A, Sandlund M. (1996) Paroxetine level in breast milk. J Clin Psychiatry. 75:39.

Stanford MS, Patton JH. (1993) In utero exposure to fluoxetine HC1 increases hematoma frequency at birth. Pharmacol Biochem. 45:959-962.

Stein A, Gath DH, Bucher J, Bond A, Day A, Cooper PJ. (1991) The relationship between postnatal depression and mother-child interaction. Br J Psychiatry. 158:46-52.

Page 305 Stevens JC, Wrighton SA. (1993) Interaction of the enantiomers of fluoxetine and norfluoxetine with human liver cytochromes P450. J Pharmacol Exp Ther. 266:964-971.

Stonestreet BS, Patlak CS, Pettigrew KD, Reilly CB, Cserr HF. (1996) Ontogeny of blood- brain barrier function in ovine fetuses, lambs, and adults. Am J Physiol. 271: R1594- R1601.

Stowe ZN, Cohen LS, Hostetter A, Ritchie JC, Owens MJ, Nemeroff CB. (2000a) Paroxetine in human breast milk and nursing infants. Am J Psychiatry. 157:185-189.

Stowe ZN, Owens MJ, Landry JC, Kilts CD, Ely T, Llewellyn A, Nemeroff CB (1997) Sertraline and desmethylsertraline in human breast milk and nursing infants. Am J Psychiatry. 154:1255-1260.

Stowe ZN. (2000b) Fetal exposure of SSRIs at delivery. Presented at American Psychiatric Association annual meeting May 2000.

Suckow RF, Zhang MF, Cooper TB. (1992) Sensitive and selective liquid-chromatographic assay of fluoxetine and norfluoxetine in plasma with fluorescence detection after pre- column derivatization. Clin Chem. 38:1756-1761.

Suri RA, Altshuler LL, Burt VK, Hendrick VC. (1998) Managing psychiatric medications in the breast-feeding woman. Medscape Womens Health. 3:1.

Syversen GB, Rajtke SK. (1985) Drug distribution within human milk phases. J Pharm Sci. 74:1071-1074.

Szeto HH, Hinman DJ. (1985) Prenatal developement of sleep-wake patterns in sheep. Sleep. 8:347-355.

Szeto HH, Mann LI, Bhakthavathsalan A, Liu M, Inturrisi CE. (1978) Meperidine pharmacokinetics in the maternal-fetal unit. J Pharmacol Exp Ther. 206:448-459.

Taddio A, Ito S, Koren G. (1996) Excretion of fluoxetine its metabolite, norfluoxetine, in human breast milk. J Clin Pharmacol. 36:42-47.

Teng RJ, Tsou Yau KI, Ho MM, Hwang KC, Lee CY. (1993) Postnatal growth of low birthweight neonates in Taiwan. Chung Hua Min Kuo Hsiao Erh Ko I Hsueh Hui Tsa Chih. 34:443-450.

Thomare P, Wang K, van der Meersch-Mougeot V, Diquet B. (1992) Sensitive micromethod for column liquid chromatographic determination of fluoxetine and norfluoxetine in human plasma. J Chromatogr. 583:217-221.

Thomas DR, Nelson DR, Johnson AM. (1987) Biochemical effects of the antidepressant paroxetine, a specific 5-hydroxytryptamine uptake inhibitor. Psychopharmacology. 93:193-200.

Page 306 Tilgmann C, Kalkkinen N. (1991) Purification and partial sequence analysis of the soluble catecol-O-methyltransferase from human placenta: comparison to the rat liver enzyme. Biochem Biophys Res Commun. 174:995-1002.

Torok-Both GA, Baker GB, Coutts RT, McKenna KF, Aspeslet LJ. (1992) Simultaneous determination of fluoxetine and norfluoxetine enantiomers in biological samples by gas chromatography with electron-capture detection. J Chromatogr B Biomed Sci Appl. 579:66-106.

Treluyer JM, Gueret G, Cheron G, Sonnier M, Cresteil T. (1997) Developmental expression of CYP2C and CYP2C-dependent activities in the man liver: in-vivo/in-vitro correlation and inducibility. Pharmacogenetics. 7:441-452.

Tuntland T, Odinecs A, Nosbisch C, Unadkat JD. (1998) In vivo maternal-fetal-amniotic fluid pharmacokinetics of zidovudine in the pigtailed macaque: comparison of steady- state and single-dose regimens. J Pharmacol Exp Ther. 285: 54-62.

Uppal R, Karmakar S, Singh MM, Dhall Gl, Gupta I, Sharma PL. (1993) use in lactating mothers: an experience at a referral hospital and in a community in India. Int J Clin Pharmacol Ther Toxicol. 31:93-95.

Urichuk LJ, Aspeslet LJ, Holt A, Silverstone PH, Coutts RT, Baker GB. (1997) Determination of p-trifluoromethylphenol, a metabolite of fluoxetine, in tissues and body fluids using an electron-capture gas chromatographic procedure. J Chromatogr B Biomed Sci Appl. 698:103-109. van Harten J. (1993) Clinical pharmacokinetics of selective serotonin reuptake inhibitors. Clin Pharmacokinet. 24:203-220.

Van Petten GR, Mathison HJ, Harris WH, Mears GJ. (1978). Chronic preparation of the pregnant ewe and fetus for pharmacological research: The placental transfer and fetal effects of bunitrolol. J Pharmacol Meth. 1:45-60.

Vendittelli F, Alain J, Nouaille Y, Brosset A, Tabaste JL. (1995) A case of lipomeningocele with fluoxetine (and , vitamins Bl and B6, heptaminol) prescribed during pregnancy. Eur J Obstet Gynecol Reprod Biol. 58:85-86. von Moltke LL, Greenblatt DJ, Duan SX, Schmider J, Wright CE, Harmatz JS, Shader RI. (1997) Human cytochromes mediating N-demethylation of fluoxetine in vitro. Psychopharmacology (Berl). 132:402-407.

Vorherr H. (1974) Drug excretion in breast milk. Postgrad Med. 56:97-104.

Wadelius M, Darj E, Frenne G, Rane A. (1997) Induction of CYP2D6 in pregnancy. Clin Pharmacol Ther. 62:400-407.

Wang LH, Rudolph AM, Benet LZ. (1985) Distribution and fate of acetaminophen conjugates in fetal lambs in utero. J Pharmacol Exp Ther. 235:302-307.

Page 307 Weissman M. (1987) Advances in psychiatruc epidemiology: rates and risks of major depression. Am J Public Health 77:445-451.

Welch R, Findlay J. (1981) Excretion of drugs in human breast milk. Drug Metabol Rev. 12:261-277.

Wilkinson GR, Shand DG. (1975) Commentary: a physiological approach to hepatic drug clearance. Clin Pharmacol Ther. 18:377-390.

Wilson JT. (1983) Determinants and consequences of drug excretion in breast milk. Drug Metabol Rev. 14:619-652.

Wisner KL, Gelenberg AJ, Leonard H, Zarin D, Frank E. (1999) Pharmacological treatment of depression during pregnancy. JAMA. 282:1264-1269.

Wisner KL, Perel JM, Findling RL. (1996) Antidepressant treatment during breast-feeding. Am J Psychiatry. 153:1132-1137.

Wisner KL, Perel JM, Wheeler SB. (1993) Tricyclic dose requirements across pregnancy. Am J Psychiatry. 150:1541-1542.

Wisner KL, Wheeler SB. (1994) Prevention of recurrent postpartum major depression. Hosp Community Psychiatry. 45:1191-1196.

Wold JA, Joost R, Giffing WJ, Marroquin F, Harris PN. (1976). Phospholipid accumulation in rats produced by fluoxetine and . Toxicol Appl Pharmacol 37:118- 119.

Wong DT, Bymaster FP, Horng JS, Molloy BB. (1975) A new selective inhibitor for uptake of serotonin into synaptosomes of rat brain: 3-(p-trifluoromethylphenoxy)-N-methyl-3- phenylpropylamine. J Pharmacol Exp Ther. 193:804-811.

Wong DT, Bymaster FP, Reid LR, Mayle DA, Krushinski JH, Robertson DW. (1993) Norfluoxetine enantiomers as inhibitors of serotonin uptake in rat brain. Neuropsychopharmacology. 8:337-344.

Wong DT, Horng JS, Bymaster FP, Hauser KL, Molloy BB. (1974) A selective inhibitor of serotonin uptake: Lilly 110140, 3-(p-trifluoromethylphenoxy)-N-methyl-3-phenyl- propylamine. Life Sci. 15:471-479.

Wong SHY, Dellafera SS, Fernandes R, Kranzler H (1990) Determination of fluoxetine and norfluoxetine by high-performance liquid chromatography. J Chromatogr B Biomed Sci Appl. 499:601-608.

Worland J, Janes C, Anthony EJ, McGinnis M, Cass L. (1984) St. Louis risk research project: comprehensive progress report of experimental studies. In Watt N, Anthony EJ, Wynne L, Rolf J (Eds.) Children at risk for schizophrenia, New York: Cambridge Univ Press, pp. 105-147.

Page 308 Wright MR, Rurak DW, van der Weyde MP, Taylor SM, Axelson JE. (1991) Clearance and disposition of ritodrine in the fluid compartments of the fetal lamb during and after constant rate fetal intravenous infusion. J Pharmacol Exp Ther. 258: 897-902. Yaryura-Tobias JA, Kirschen H, Ninan P, Mosberg HJ (1991) Fluoxetine and bleeding in obsessive-compulsive disorder. Am J Psychiatry. 148(7): 949

Yeleswaram K, Rurak DW, Kwan E, Hall C, Doroudian A, Wright MR, Abbott FS, Axelson JE. (1993) Transplacental and nonplacental clearances, metabolism and pharmacodynamics of labetalol in the fetal lamb after direct intravenous administration. J Pharmacol Exp Ther. 267: 425-431.

Yerby MS, Friel PN, McCormick K. (1992) Antiepileptic drug disposition during pregnancy. Neurology. 42:12-16

Yoo SD, Axelson JE, Kwan E, Rurak DW. (1990) Pharmacokinetics of diphenhydramine after dose ranging in nonpregnant ewes. J Pharm Sci. 79:106-110.

Yoo SD, Axelson JE, Taylor'SM, Rurak DW. (1986) Placental transfer of diphenhydramine in chronically instrumented pregnant sheep. J Pharm Sci. 75:685-687.

Zhan-Waxier C, Chapman M, Cummings EM. (1984) Cognitive and social development in infants and toddlers with a bipolar parent. Child Psychiatry Human Develop. 15:75-85.

Page 309