Citation
Pharmacokinetics and Tissue Distribution of 5,7-Dimethoxyflavone in Mice

Material Information

Title:
Pharmacokinetics and Tissue Distribution of 5,7-Dimethoxyflavone in Mice
Creator:
Bei, Di
Publisher:
University of Florida
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Pharmaceutical Sciences
Pharmaceutics
Committee Chair:
DERENDORF,HARTMUT C
Committee Co-Chair:
HOCHHAUS,GUENTHER
Committee Members:
SONG,SIHONG
HUIGENS,ROBERT W III

Subjects

Subjects / Keywords:
adme
lc-msms
pbpk
pharmacokinetics
phytochemical

Notes

General Note:
The purpose of this study was to investigate the PK profile of 5,7-DMF in vivo. 5,7-DMF is a natural flavonoid with many beneficial pharmacological effects. However, to date it was evaluated mainly in vitro and its PK in vivo remains largely unknown. In addition, current available quantification methods of 5,7-DMF all lack sufficient sensitivity. Therefore in our study, we established a sensitive quantification method of 5,7-DMF using LCMSMS. It was fully validated and accuracy, precision, sensitivity, selectivity, recovery and stability were evaluated thoroughly in mouse plasma. This quantification method was also validated in mouse tissues. In the dose escalation study in mice, AUC was demonstrated dose proportional within the dose range 10 to 50 mg/kg 5,7-DMF. And in the following study of 10 mg/kg oral dosing, the calculated bioavailability was 13.8%. The maximal 5,7-DMF concentrations in plasma and tissues were reached within 30 min. Plasma Cmax was 1870 ng/mL, and AUCt was 532 hr*ng/mL and terminal half-life was 3.40 hr. The volume of distribution was 90.1 L/kg. Clearance was 20.2 L/hr/kg. Except for muscle and adipose, other tissues had higher Cmax than plasma, ranging from 1.75- to 9.96-fold. After oral administration, Kp of these tissues were 0.65 to 12.9. Two-compartmental model without Tlag model best described the oral plasma data. The fitted PK parameter Ka was 6.04 1/h, V1 was 0.05 L, V2 was 0.56 L, CL was 0.55 L/h, and CL2 was 0.04 L/h. This model was extrapolated to human of average body weight of 70 kg with V1, V2, CL and CL2 scaled up, which may provide insight for 5,7-DMF concentration prediction in human plasma. In conclusion, we reported for the first time the PK, tissue distribution and modeling and simulation of 5,7-DMF in mice. These results will be critical in evaluating if those beneficial in vitro effects can be translated in vivo. Ultimately, the plasma and tissue levels of 5,7-DMF can be used for PBPK modeling and can extrapolated in predicting 5,7-DMF concentrations in human.

Record Information

Source Institution:
UFRGP
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
12/31/2017

Downloads

This item has the following downloads:


Full Text

PAGE 1

PHARMACOKINETICS AND TISSUE DISTRIBUTION OF 5,7 DIMETHOXYFLAVONE IN MICE By DI BEI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2015

PAGE 2

2015 Di Bei

PAGE 3

T o the people who lov e me and the people that I love

PAGE 4

4 ACKNOWLEDGMENTS First and foremost, I would give special thanks to the most amazing advisor Dr. Hartmut Derendorf for wisely mentoring me, unconditionally supporting me and continuously guiding me through difficult times with incredible patience. He is so knowledgeable fair minded and considerate that I can never stop learning from him. I would also like to thank my supervisory committee, Dr. Guenther Hochhaus, Dr. Sihong Song and Dr. Robert W Huigens for their time and effort on advisi ng me and helping me. I want to thank Dr. Guohua An, my previous advisor back in Orlando campus, for kindly giving me opportunity to work on PBPK project and publishing those works. Without her, the animal experiments would not be accomplished in such an e fficient way. PhD study to me is like a marathon. There have been a lot of frustrations which almost made me qui t half way. Thanks to my faith as a Christ ian and my loving families, I could stick to my dream and able to finally reach the finish line. So I want to take this opportunity to genuinely appreciate the love and support from them. Finally, I want to extend my thanks to all the graduate students, post docs, and staff in our program for being so nice and helpful. I want to share with you a poem enti tl ed The Road Not Taken And sorry I could not travel both, And be one traveler, long I stood, And looked down one as far as I could, To where it bent in the undergrowth; I shall be telling this with a sigh, Somewhere ages and ages hence: Two roads d iverged in a wood, and I I took

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURE S ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ........................... 10 ABSTRACT ................................ ................................ ................................ ................... 13 1 INTRODUCTION ................................ ................................ ................................ .... 15 Specific Aim 1 ................................ ................................ ................................ ......... 18 Specific Aim 2 ................................ ................................ ................................ ......... 18 Specific Aim 3 ................................ ................................ ................................ ......... 19 2 DEVELOPMENT AND METHOD VALIDATION OF BIOANALYTICAL METHODS ................................ ................................ ................................ .............. 20 Background ................................ ................................ ................................ ............. 20 Materials and Methods ................................ ................................ ............................ 21 Reagents and Chemicals ................................ ................................ ................. 21 Instrument Conditions ................................ ................................ ...................... 21 HPLC c onditions ................................ ................................ ........................ 21 MS/MS conditions ................................ ................................ ...................... 22 Sample Preparation of Calibration Standards and Quality Control for LC MS/MS ................................ ................................ ................................ .......... 22 Assay Validations in Mouse Plasma ................................ ................................ 23 Matrix Effect and Recovery Yield in Mouse Plasma ................................ ......... 24 Stability of 5,7 DMF in Mouse Plasma ................................ ............................. 24 Stability of 5,7 DMF in stock solution ................................ ......................... 24 Freeze thaw stability of 5,7 DMF in mouse plasma ................................ .. 25 Long term stability of 5,7 DMF in mouse plasma ................................ ....... 25 Short term temperature stability of 5,7 DMF in mouse plasma .................. 25 Post preparative stability of 5,7 DMF ................................ ......................... 25 Validation of Quantification Method in Tissues ................................ ................. 26 Results ................................ ................................ ................................ .................... 26 Determination of 5,7 DMF and I.S by LC MS/MS ................................ ............ 26 Method Validation ................................ ................................ ............................. 32 Matrix Effect and Recovery Yield ................................ ................................ ..... 33 Stability of 5,7 DMF ................................ ................................ .......................... 34 Stability of 5,7 DMF in stock ................................ ................................ ...... 34 Freeze thaw stability of 5,7 DMF in mouse plasma ................................ .. 34

PAGE 6

6 Long term stability of 5,7 DMF in mouse plasma ................................ ....... 35 Short term temperature stability of 5,7 DMF in mouse plasma .................. 35 Post preparative stability of 5,7 DMF ................................ ......................... 36 Validation of quantification method in tissues ................................ ............ 36 Discussions ................................ ................................ ................................ ............. 45 Summary ................................ ................................ ................................ ................ 48 3 PHARMACOKINETICS AND TISSUE DISTRIBUTION OF 5,7 DMF IN MICE ....... 49 Background ................................ ................................ ................................ ............. 49 Material and Methods ................................ ................................ ............................. 50 Chemical s and Reagents ................................ ................................ ................. 50 Animals and In Vivo Study ................................ ................................ ............... 51 Sample Preparations and LC MS/MS Assays ................................ .................. 52 Pharmacokinetic Analysis ................................ ................................ ................. 53 Resu lts ................................ ................................ ................................ .................... 53 LC MS/MS Method Development and Validation ................................ ............. 53 Dose Linearity of 5,7 DMF in Mice ................................ ................................ ... 54 Bioavailability of 5,7 DMF in Mice ................................ ................................ .... 55 Pharmacokinetic Analysis and Tissue Distribution in Mice ............................... 55 Discussions ................................ ................................ ................................ ............. 60 Summary ................................ ................................ ................................ ................ 64 4 PHARMACOKINETIC MODELING AND SIMULATION ON 5,7 DMF .................... 66 Background ................................ ................................ ................................ ............. 66 Methods ................................ ................................ ................................ .................. 67 Assumptions and Model Building ................................ ................................ ..... 67 Population Modeling Analysis ................................ ................................ ........... 69 Model V alidation ................................ ................................ ............................... 71 Allometric S calling ................................ ................................ ............................ 71 Model S imulation and Prediction ................................ ................................ ...... 71 Results ................................ ................................ ................................ .................... 72 Population Modeling Analysis ................................ ................................ ........... 72 Model V alidation ................................ ................................ ............................... 82 Allometric Scalling ................................ ................................ ............................ 85 Model Si mulation and Prediction ................................ ................................ ...... 86 Discussions ................................ ................................ ................................ ............. 88 Summary ................................ ................................ ................................ ................ 88 5 CONCLUSIONS ................................ ................................ ................................ ..... 90 Summary ................................ ................................ ................................ ................ 90 Perspective ................................ ................................ ................................ ............. 91 LIST OF REFERENCES ................................ ................................ ............................... 96 BIOGRAPHIC AL SKETCH ................................ ................................ .......................... 104

PAGE 7

7 LIST OF TABLES Table page 2 1 Intra day (within day) validation of 5,7 DMF in mouse plasma. .......................... 32 2 2 Inter day (between day) validation of 5,7 DMF in mouse plasma. ...................... 33 2 3 Matrix effect and recovery yield of 5,7 DMF in mouse plasma. .......................... 33 2 4 Freeze thaw stability of 5,7 DMF in mouse plasma. ................................ ........... 34 2 5 Long term stability of 5,7 DMF in mouse plasma. ................................ ............... 35 2 6 Short term temperature stability of 5,7 DMF in mouse plasma. .......................... 36 2 7 Post preparative stability of 5,7 DMF. ................................ ................................ 36 2 8 Method validations of 5,7 DMF in mouse plasma and tissues. ........................... 44 2 9 Comparison among different sample preparation methods on quantification of liver samples from a pilot study. ................................ ................................ ..... 48 3 1 Dose linearity of 10 50 mg/kg IV dose in mice ................................ .................. 55 3 2 Bioavailability of 5,7 DMF following 10 mg/kg IV and oral dose ........................ 55 3 3 Non compartmental parameters of 5,7 DMF in mouse plasma after oral administration of 10 mg/kg 5,7 DMF. ................................ ................................ .. 56 3 4 PK parameters of 5,7 DMF in mouse tissues after oral administration of 10 mg/kg 5,7 DMF ................................ ................................ ................................ .. 59 4 1 Model comparison under PopPK approach in Phoenix. ................................ ..... 73 4 2 PopPK parameters for the final model. ................................ ............................... 75 4 3 Scaled up PK parameters in two compartment model for human beings. .......... 85

PAGE 8

8 LIST OF FIGURES Figure page 2 1 Chemical structure of 5,7 DMF and I.S ................................ .............................. 27 2 2 Mass spectra of 5,7 DMF and I.S. ................................ ................................ ...... 28 2 3 XIC of the 5, 7 DMF and I.S in mouse plasma. ................................ ................... 30 2 4 Calibration curve of 5,7 DMF in mouse plasma. ................................ ................. 31 2 5 Calibration curve of 5,7 DMF in mouse liver. ................................ ...................... 31 2 6 Calibration curve of 5,7 DMF in mouse kidney. ................................ .................. 38 2 7 Calibration curve of 5,7 DMF in mouse intestine. ................................ ............... 39 2 8 Calibration curve of 5,7 DMF in mouse brain. ................................ .................... 39 2 9 Calibration curve of 5,7 DMF in mouse spleen. ................................ .................. 40 2 10 Calibration curve of 5,7 DMF in mouse heart. ................................ .................... 41 2 11 Calibration curve of 5,7 DMF in mouse lun g. ................................ ..................... 41 2 12 Calibration curve of 5,7 DMF in mouse muscle. ................................ ................. 42 2 13 Calibration curve of 5,7 DMF in mouse fat. ................................ ........................ 43 2 14 XIC of the 5,7 DMF and I.S in mouse plasma and tissues ................................ 47 3 1 Dose proportionality in mice. Plasma PK profile of 10, 25 and 50 mg/kg IV dose of 5,7 DMF. ................................ ................................ ................................ 57 3 2 Plasma PK profile of IV and oral dose of 10 mg/kg 5,7 DMF in mice. ............... 59 3 3 Concentration time p rofiles of 5,7 DMF in mice following oral administration of 10 mg/kg in mouse plasma, liver, kidney, small intestine and brain. .............. 61 3 4 Mean plasma and tissue PK profiles of oral dose of 10 mg/kg 5,7 DMF in mice. ................................ ................................ ................................ ................... 63 4 1 Structural of one two and three compartmental model. ................................ ... 69 4 2 Overall g oodness of fit plots. ................................ ................................ .............. 76 4 3 Individual goodness of fit plot. ................................ ................................ ............ 77

PAGE 9

9 4 4 Diagnostic plots for the final model. ................................ ................................ .... 80 4 5 Visual predictive check for two compartmental final PopPK model. ................... 83 4 6 Simulated mouse plasma PK profile following oral dose of 10 mg/kg 5,7 DMF. ................................ ................................ ................................ ................... 84 4 7 Simulated mouse plasma PK profile following oral dose of 25 mg/kg 5,7 DMF. ................................ ................................ ................................ ................... 84 4 8 Simulated mouse plasma PK profile following oral dose of 25 mg/kg 5,7 DMF. ................................ ................................ ................................ ................... 84 4 9 Simulation of 5, 7DMF concentration time profile in 70 kg human being after oral dose of 10 mg/kg 5,7 DMF. ................................ ................................ ......... 87 5 1 Exemplary whole body PBPK model for oral dosing. ................................ ......... 94 5 2 Exemplary local PBPK model for oral dosing. ................................ .................... 95

PAGE 10

10 LIST OF ABBREVIATIONS 2LL 5,7 DMF ACN ADME Log likelihood. 5,7 Dimethoxyflavone. Elimination rate constant in terminal phase. Acetonitrile. Absorption, distribution, metabolism, and excretion. AIC Akaike Information Criterion. AUC Area under the curve. Bap Benzo[a]pyrene. BCRP Breast cancer resistance protein. BIC Bayesian Information Criterion. CID C L Collision induced dissociation. Clearance. C L 2 Intercompartmental clearance. Cmax CV CWRES Maximal concentration. Coefficient of variance. Conditional weighted residual. CYP Cytochrome P450. DDI DV Drug drug interaction Dependent variable. ESI F Electro spray ionization. Bioavailability. GC Gas chromatography. HPLC High performance liquid chromatography. IACUC Institutional Animal Care and Use Committee.

PAGE 11

11 IC50 Concentration to achieve 50% of maximal inhibition. IPRED Individual prediction. I.S IV K0 Ka Kp LC MS/MS Internal standard. Intravenous. Zero order absorption rate constant. First order absorption rate constant. Tissue partition coefficient. Liquid chromatography coupled with tandem mass spectrometry. LLOQ Lower limit of quantification. MDR Multi drug resi stance. MRP Multidrug resistance associated proteins. NCA PBPK Noncompartmental analysis. Physiology based pharmacokinetic. PBS PD PEG P gp PK Phosphate buffer saline. Pharmacodynamic. Poly(ethylene glycol). P glycoprotein. Pharmacokinetic. PopPK PRED Population pharmacokinetics. Prediction. Q C QQ SE TAD Tlag Quality control. Data quantile theoretical quantile. Standard error. Time after dose. Lag time in absorption.

PAGE 12

12 TRAIL Tumor necrosis factor related apoptosis inducing ligand. UDPGT VPC XIC Uridine diplosphate glucuronyl transferase. Visual predictive check. Extracted ion chromatogram.

PAGE 13

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PHARMACOKINETICS AND TISSUE DISTRIBUTION OF 5,7 DIMETHOXYFLAVONE IN MICE By Di Bei December 2015 Chair: Hartmut Derendorf Major: Pharmaceutic al Science s The purpose of this study was to investigate the PK profile of 5,7 DMF in vivo. 5,7 DMF is a natural flavonoid with many beneficial pharmacological effects. However, to date it was evaluated mainly in vitro and its PK in vivo remains largely unknown. In a ddition, current available quantification methods of 5,7 DMF all lack sufficient sensitivity. Therefore in our study, we established a sensitive quantification method of 5,7 DMF using LC MS/MS. It was fully validated and accuracy, precision, sensitivity, s electivity, recovery and stability were evaluated thoroughly in mouse plasma. This quantification method was also validated in mouse tissues. In the dose escalation study in mice, AUC was demonstrated dose proportional within the dose range 10 to 50 mg/kg 5,7 DMF. And in the following study of 10 mg/kg oral dosing, the calculated bioavailability was 13.8%. The maximal 5,7 DMF concentrations in plasma and tissues were reached within 30 min. Plasma C max was 1870 1190 ng/mL, and AUC t was 532 165 hr*ng/mL and terminal half life was 3.40 2.80 hr. The volume of distribution was 90.1 62.0 L/kg. Clearance was 20.2 7.49 L/hr/kg. Except for muscle and adipose, other tissues had higher C max than plasma,

PAGE 14

14 ranging from 1.75 to 9.96 fold. After oral administrat ion, K p of these tissues were 0.65 to 12.9. Two compartmental model without Tlag model best described the oral plasm a data. The fitted PK parameter K a was 6.04 1/h, V 1 was 0.05 L, V 2 was 0.56 L, CL was 0.55 L/h, and CL 2 was 0.04 L/h. This model was extrapolated to human of average body weight of 70 kg with V 1 V 2 CL and CL 2 scaled up, which may provide insight for 5,7 DMF concentration prediction in human plasma. In conclusion, we reported for the first time the PK, ti ssue distribution and modeling and simulation of 5,7 DMF in mice. These results will be critical in evaluating if those beneficial in vitro effects can be translated in vivo. Ultimately, the plasma and tissue levels of 5,7 DMF can be used for PBPK modeling and can extrapolated in predicting 5,7 DMF concentrations in human.

PAGE 15

15 CHAPTER 1 INTRODUCTION Phytochemicals are more and more popular because of the potential as alternative or adjunct medicine [ 1 2 ]. Herb drug inter action of clinically importance has gained more interest during the past years [ 2 5 ]. Especially for flavonoids abundant in fruit, vegetables and wine, they are polyphenols that can impact on cancer proliferation, inflammation, angiogenesis and metastasis [ 6 10 ]. Its pharmacological evidence of chemoprevention ha s been reported concerning cancer in GI tract [ 11 14 ] and liver [ 15 ] prostate [ 16 17 ] ovarian [ 18 ] lung [ 19 ] endometrial [ 20 ] and breast cancers [ 21 26 ]. 5,7 Dimethoxyflavone (5,7 DMF) is a n aturally occuring flavonoid that is abundant in tropical plant sources such as Kaempferial paviflora (also known as Thai Genseng) Piper caninum and Leptospermum scoparium [ 27 29 ], all of which have been used as folk medicine to treat gastrointestinal disorder, infections and hypertension [ 27 2 8 ]. 5,7 DMF has been evaluated extensively in vitro and the results showed that 5,7 DMF has many beneficial pharmacological activities. For example, it has anti inflammatory activity [ 30 ], vasorelaxation and car dioprotection effect by increasing potassium efflux and inhibiting calcium influx [ 31 ], and selectively inhibitory effect against butyrylcholinesterase versus acetylcholinesterase [ 32 conducted a series of in vitro studies to eval uate the chemotherapeutic and chemopreventive potential of this compound [ 33 35 ]. They found that 5,7 DMF can prevent hepatic carcinogenesis by inhibiting Cytochrome P450 (CYP) 1A1 activity,

PAGE 16

16 which subsequently decreased the carcinogenbenzo[a]pyrene (BaP) i nduced DNA adduct formation [ 35 ]. The chemopreventive effect of 5,7 DMF was also observed in human esophageal cancer cells and human oral carcinoma cells by regulating CYP1B1 activity [ 34 35 ]. 5,7 DMF was also reported to have chemosensitizing effect in h etapocellular carcinoma cells and human leukemic cells through tumor necrosis factor related apoptosis inducing ligand (TRAIL) mediated apoptosis [ 36 37 ]. In addition, 5,7 DMF was shown to improve the accumulation of rhodamine 123 into the cells with overe xpression of P glycoprotein (P gp), an efflux transporter which is present in various cancer cells and plays an important role in multi drug resistance (MDR) [ 27 ]. In addition to P gp inhibition, 5,7 DMF also has inhibitory effect on other two efflux trans porter members multidrug resistance associated proteins (MRPs) and Breast Cancer Resistance Protein (BCRP). For example, 5,7 DMF can increase the accumulation of doxorubicin in A549 cells as a result of suppression of MRPs [ 38 ]. 5,7 DMF demonstrated ver y potent BCRP inhibition both in vitro and in vivo even at low micro molar concentrations with the in vitro IC 50 for BCRP inhibition is 1.41 M [ 3 9 4 0 ]. The plasma and tissue concentrations of mitoxantrone, an anticancer drug with high affinity to BCRP, we re significantly increased in vivo with the co administration of 5,7 DMF [ 39 ]. As is known, MDR is the major cause for the failure of cancer chemotherapy treatment. An important mechanism for MDR is the decreased intracellular concentration of anticancer drugs due to the overexpression of efflux transporters, including P gp, BCRP, and MRPs [ 41 ]. The broad inhibitory effect of 5,7 DMF on P gp,

PAGE 17

17 BCRP, and MRPs makes it a very promising chemosensiting agent. Although 5,7 DMF has numerous beneficial pharmacol ogical activities as mentioned above, it should be noted that most of the results were obtained in vitro. To evaluate if those in vitro effects can be translated in vivo, it is important to know the PK information of 5,7 DMF, especially the in vivo concent rations of 5,7 DMFat the target site(s) (i.e. tissues). The ADME property of 5,7 DMF is promising in five major aspects. First of all, it has low toxicity against normal healthy cells with IC 50 around 100 M in lung BEAS 2B cells and the esophageal HET 1A cells [4 2 ] while its analog chrysin exhibited a n IC 50 of 2 M. secondly, 5,7 DMF is a potent BCRP inhibitor as demonstrated in the literature that in vitro IC 50 for BCRP inhibition is 1.41 M and under 25 mg/kg doing, 5,7 DMF significantly increased the mitoxantrone levels in liver and kidney [ 39 40 ]; thirdly, it led to high tissue accumulation after doing [ 43 44 ] which makes tissues interesting site for research ing 5,7 Fourthly, it had improved intestinal absorption and metabolic stability [ 44 ]. Compared to other methylated and unmethylated flavonoids, 5,7 DMF demonstrated the highest stability when exposed to human liver hepatocyte, human liver microsome and human liver S9 fraction [4 5 4 7 ]; Besides, methylated flavones showed approximately 5 to 8 fold higher apparent permeability. The high metabolic stability in addition to increased intestinal absorption of the methylated polyphenols make them more promising and interesting than the unmethylated polyphenols as potential chemopreventive compounds. [ 4 7 ] However, to date there still have been large knowledge gap about this compound. Primarily, the information related to the in vivo PK of 5,7 DMF is very limited [ 43 44 ]. In addition, to the best of our knowledge, there are only two quantification

PAGE 18

18 methods of 5,7 DMF that have been published, both of which lack sufficient sensitivity, with one of them reported lower limit of quantification (LLOQ) of 1860 ng/mL using HPLC and the other method reported LLOQ of 800 ng/mL in non biological matrices using gas chromatography (GC) [ 44, 48 ]. In order to better understand the pharmacokinetics of 5,7 DMF in vivo, especially to capture its terminal phase, a more sensitive quantification method is needed. The purpose of our study was to establish and validate a sensitive quantification method of 5,7 DMF in biological matrices such as mouse plasma and mouse tissues using LC MS/MS assay, to apply this quantification method evaluate the PK profile of 5,7 DMF in mouse and finally to construct PK model based on it. And the objectives for 5,7 DMF project is listed as followed: Specific Aim 1 The first aim is to develop and validate a new bioassay to sensitively and accur ately quantify 5,7 DMF in mice matrices such as plasma, and other tissues. This aim will be accomplished in Chapter 2. The significance and innovations of this chapter will be, (i). Analytical methodology using liquid chromatography tandem mass (LC MS/MS) to quantify 5,7 DMF is new. Once established, the sensitivity will be greatly improved compared to previously reported methods using HPLC and GC; (ii). The matrix effect and recovery yield will be first reported; and (iii). The assessment of 5,7 DMF stabil ity in biological matrix is novel. Our assay will provide basis for future PK sample analysis. Specific Aim 2 The second aim is to establish PK profiles and tissue distribution of 5,7 DMF in mice for the first time. This aim will be accomplished in Chapter 3. The significance and innovations of this chapter will be, (i). Dose linearity of 5,7 DMF has not been

PAGE 19

19 investigated; (ii). Bioavailability in mice has not been reported; and (iii). Tissue distribution profile remained unknown. Specific Aim 3 The third a im is to construct the PK model based on the mouse plasma PK profile and extrapolate this model to human species in order to predict the human plasma level of 5,7 DMF. This aim will be accomplished in Chapter 4. The significance and innovations of this cha pter will be, (i). Up to date, the literature on PK model of 5,7 DMF is missing so our work will be the first to initiate PK modeling on 5,7 DMF; (ii). Our model will be the first to explain and predict 5,7 DMF in vivo; and (iii). This will be further util ized in building PBPK modeling for clinical relevance. As learnt from previous literature, 5,7 DMF could reverse BCRP mediated MDR in mouse tissues [13 14], but in clinical trials, it is fairly difficult to measure tissue concentrations. In this regard, PB PK is advantageous over traditional PK method in that it consists of physiologically meaningful compartments so that the tissue concentrations can be accurately predicted. The ultimate goal is to extrapolate the PBPK model in mice to human in order to simulate the 5,7 DMF levels in human plasma and most importantly, the accumulations in tissues where 5,7 DMF had the promising PD effect.

PAGE 20

20 CHAPTER 2 DEVELOPMENT AND METHOD VALIDATION OF BIOANALYTICAL METHODS Background For the beneficial pharmacological e ffects of 5,7 DMF mentioned above, it should be noted that most of the data were obtained from in vitro studies. The information of the bioavailability of 5,7 DMF, as well as its pharmacokinetics (PK), is critical for estimating whether the concentrations tested in vitro are physiologically relevant and whether those in vitro effects can be translated in vivo. However, before our study there are only two quantification methods of 5,7 DMF that have been published, both of which lack sufficient sensitivity, w ith one of them reported a lower limit of quantification (LLOQ) of 1860 ng/mL using HPLC and the other method reported a LLOQ of 800 ng/mL using gas chromatography (GC) [ 44, 48 ]. As is generally agreed, sensitive and accurate bioassay is the key to profile PK and PD [ 49 ]. In order to better understand the pharmacokinetics of 5,7 DMF in vivo, especially to capture its terminal phase, a more sensitive quantification method is needed. The purpose of our study in this chapter was to establish and validate a sen sitive quantification method of 5,7 DMF using LC MS/MS assay, and to evaluate the PK of 5,7 DMF in mouse. LC MS/MS is a state of art technique which coupled high performance liquid chromatogram with triple quadruple to separate analyte from other substance s based on different retention time in HPLC and different m/z after fragmentation inside of mass spectrometer This type of bioassay is widely used in preclinical and clinical sample analysis [ 50 ]. Since LC MS/MS has been reported as fast, sensitive and se lective method which can generate reliable and efficient data out of plasma, urine and tissues [ 51 53 ], it was chosen for this study as the main bioassay.

PAGE 21

21 The innovations and significance therefore include but not limited to, (i). Analytical methodology us ing liquid chromatography tandem mass (LC MS/MS) to quantify 5,7 DMF is novel; Once established, the sensitivity will be greatly improved compared to previously reported methods using HPLC and GC; (ii). The matrix effect and recovery yield will be first re ported; (iii). The assessment of 5,7 DMF stability in biological matrix is novel. Our assay will provide basis for future PK sample analysis. Materials and Methods Reagents and Chemicals 5,7 DMF (>99%) and 5,7,4 trimethoxyflavone (TMF, >99%) were purchased from Indofine Chemical Company, Inc. (Hillsborough, New Jersey, US). Ammonium formate (99%) was obtained from ACROS Organics (New Jersey, US). Analytical HPLC grade acetonitrile (ACN), water and formic acid were purchased Fisher Scientific (Pittsburg, Pen nsylvania, US). Analytical spectrophotometric grade ethyl acetate (99.5%) was obtained from Alfa Aesar (Ward Hill, Massachusetts, US). Heparin treated mouse plasma was purchased from BioreclamationIVT (East Meadow, New York, US). (2 Hydroxypropyl) cyclod extrin and poly(ethylene glycol) Mn 400 (PEG400) were purchased through Sigma Aldrich (St. Louis, Missouri, US). Heparin injectable (1000 U/mL) was purchased from Patterson Vet Generics (Devens, Massachusetts, US). Instrument Conditions HPLC conditions Hig h performance liquid chromatogram (HPLC) was performed with Agilent 1200 series systems (Agilent, California, US) that include a 1260 degasser, a 1260 binary pump, a 1260 autosampler and a 1290 thermostat. 5,7 DMF and TMF were separated by Agilent XDB C18

PAGE 22

22 Agilent, California, US) under an isocratic elution at a flow rate of 0.4 mL/min. The mobile phase is 50/ 50 (v/v) of 20 mM ammonium formate in water (containing 0.1% min for each injection. MS/MS conditions 5,7 DMF and I.S was monitored under positive mode by Agilent 6 460 triple quadruple (Agilent, California, US) equipped with Agilent Jet Stream electrospray ionization (AJS ESI). In multiple reaction monitor (MRM) mode, transition ion pair for 5,7 DMF is m/z 283 > m/z 239 while the transition ion pair for I.S is m/z 31 3 > m/z 298. To obtain the optimal sensitivity, the fragmentor is 120 V for both 5,7 DMF and I.S, and the collision energy are 33 V and 25 V for 5,7 DMF and I.S, respectively. The mass spectrometer was operated at the gas temperature of 300 C, gas flow of 8 L/min, nebulizer of 35 psi, sheath gas temperature of 400 C, sheath gas flow of 11 L/min, capillary voltage of 4500 V and cell accelerator voltage of 7 V. The response acquisition was performed using MassHunter Data Acquisition software and the qualita tive analysis was done with MassHunter Qualitative Analysis software (Version B 6.0). Then the quantitative analysis was performed by MassHunter QQQ Qualitative Analysis software (Version B 6.0). Sample P reparation of C alibration S tandards and Q uality C ont rol for LC MS/MS TMF was chosen as the internal standard (I.S) because it has similar chemical structure as that of 5,7 DMF. Stock solutions of 5,7 DMF and I.S at concentration of 1 rmediate working solutions of 5,7 DMF were prepared by serial dilution with mobile phase (neat

PAGE 23

23 solution). Prepared 10 fold intermediate working solutions of analyte by serial dilution with mobile phase from 10 g/mL working solution ( range: 10, 20, 50, 10 0, 200, 500, 1000, 2000, 5000 and 10000 ng/mL). Calibration curve in mobile phase was prepared by 10 L of each intermediate working solutions and 10 L IS together with 80L mobile phase. After calibration curve in mobile phase (not shown) indicated linea rity, c alibration of ethyl acetate was added for liquid liquid extraction on a mechanical shaker for 5 min. the liquid liquid extraction agent was ethyl acetate which can form two immiscible layers with water phase. Then the supernatant layer was transferred to clean tubes after the centrifuge at 17,000 rpm under 4 C for 4 min, and placed unde r nitrogen for centrifuged at 17,000 rpm under 4 C for 4 min to get rid of precipitation. After MS/MS. Standards were prep ared at concentrations 2 1000 ng/mL. Quality control (QC) samples were prepared the same way as the standards and at the concentrations of 2, 5, 50 and 500 ng/mL. Data were quantified with MassHunter QQQ Quantitative Analysis (Version B 6.0). Assay V alidations in M ouse P lasma The intra run/within run validation was performed on concentrations of 2, 5, 50 and 500 ng/mL with six replicates for each concentration. And the inter run/between run validations were evaluated at 3 separate days with triplicate s during each day on each of the concentration at 2, 5, 50 and 500 ng/mL. For separate validations, separate standard curves were freshly prepared. The standard curves were fitted by a linear regression and the validation samples were calculated back by th e calibration curve of

PAGE 24

24 that day. The mean and the coefficient of variance (CV) were calculated accordingly. Accuracy was calculated by comparing calculated concentrations to corresponding nominal ones. Matrix E ffect and R ecovery Y ield in M ouse P lasma Stand ards in mobile phase/neat solution, pre extracted spiked samples and post extracted spiked samples were prepared in triplicate at concentrations of 2, 5, 50 and 500 ng/mL to analyze the matrix effect and recovery yield of the 5,7 DMF in mouse plasma follow ing liquid liquid extraction with ethyl acetate. The blank mouse plasma used was the heparin treated mouse plasma was purchased from BioreclamationIVT (East Meadow, New York, US). Group A: 5,7 DMF and I.S were prepared directly with mobile phase without an y matrix or extraction. Group B: 5,7 DMF and I.S were spiked into blank mouse plasma followed by liquid liquid extraction, centrifugation, and reconstitution process (as described in Section 2.3). Group C: 5,7 DMF and I.S were spiked in the extracted matri x of blank mouse plasma to prepare the final concentrations of 2, 5, 50 and 500 ng/mL. Matrix effect (%) = C 100/A and recovery (%) = B 100/C, where A = absolute response obtained from standards mobile phase; B = absolute response obtained from pre ext racted spiked samples; C = absolute response obtained from post extracted spiked samples. Stability of 5,7 DMF in Mouse Plasma Stability of 5,7 DMF in stock solution Freshly prepared 5,7 DMF solution (1 mg/mL) and the stored DMF solution (1 mg/mL) were dil uted to 500 ng/mL with mobile phase with six replicates in each group. The final concentration of the I.S is 1000 ng/mL. 5,7 DMF concentrations were compared by calculating the ratio of 5,7 DMF to I.S response.

PAGE 25

25 Freeze thaw stability of 5,7 DMF in mouse pla sma 5,7 DMF solution was spiked into mouse plasma to prepare final concentrations of one, two or three sequential freeze thaw cycles respectively prior to sample analyses using LC MS/MS (as described in Section 2.3). During each cycle, the samples were thawed under room temperature (25 C) for 60 min before re C, and the next cycle happened after 24 h interval. Samples were prepared in triplicates for each concentration of each cycle set. Long term stability of 5,7 DMF in mouse plasma 5,7 DMF solution was spiked into mouse plasma to prepare final concentrations of and 14, samples in triplicates were analyzed and compared with freshly prepared plasma samples. Short term temperature stability of 5,7 DMF in mouse plasma The short term stability was conducted by measuring triplicate freshly prepared QC samples at conce ntrations 2, 5, 50 and 500 ng/mL in mouse plasma versus those QC samples at the same concentrations placed on bench top for 24 h. Post preparative stability of 5,7 DMF On Day 1, standards in mouse plasma and QC samples in triplicates were prepared and plac ed in autosampler for quantification. The autosampler remained the temperature control of 4 C all the time. After 24 h sitting inside the autosampler, the same QC samples were re analyzed on Day 2 by the instrument and the concentration was re determined by the standard of Day 1. The calculated concentrations in Day 2 were compared in parallel with those in Day 1.

PAGE 26

26 Validation of Quantification Method in Tissues For tissue samples, tissue was weighed and added PBS buffer (pH=7.4) based on 10 fold volume of e ach tissue weight. They were cut into smaller pieces before homogenization. Then 10 L of I.S was spiked into 100 L of tissue homogenate. Other steps were the same as plasma sample preparation. When validating the LC MS/MS methods in tissues, QC sample pr eparation followed the same procedure of calibration curve. And the QC concentrations were selected at 2 and 500 ng/mL, which represented the lower and higher end of the tissue PK samples Results Determination of 5,7 DMF and I.S by LC MS/MS To acquire t he mass spectra of analyte and I.S, 1000 ng/mL of 5,7 DMF and I.S MS/MS and infused with 0.4 mL/min flow rate of the mobile phase. The analyte and I.S were monitored under M RM in positive mode for good ionization. The chemical structures of analyte and I.S were shown in Fig. 2 1. The full scan and the proposed ion fragmentations of 5,7 DMF and I.S was shown in Fig. 2 2 The analyte exhibited the precursor ion of m/z 283 (Fig. 2 2A) corresponding to the protonated form [M+H]+ of the compound with the molecular weight of 282. The precursor ion of 283 was chosen over 305 is because of higher intensity. This form of ion is quite typical in positive mode on the compounds prone to g et a proton and become protonated. After collision induced dissociation (CID), the resulting product ion has a unimodal peak and a predominant m/z of 239 ( Fig. 2 2B). The 44 Da l o ss compared with the precursor ion wa s possibly due to the loss of CO2 group from the molecule. But it wa s also possible some other groups were lost during the dissociation process. The CO2 group proposed here is only

PAGE 27

27 putative hypothesis waiting to be confirmed with future experiment. The I.S exhibited the precursor ion of m/z 313 ( Fig. 2 2C) corresponding to the protonated form of the compound with the molecular weight of 312. Again, I.S precursor ion of 313 was chosen over 335 because of more abundance which led to possibly higher intensity of product ion. After CID, the resultin g product ion of the I.S has a fragment ion with m/z of 298 ( Fig. 2 2D), with a loss of 15 Da likel y because of dissociation of radical group CH3 radical dot. Based on the full scan of Thus for the transition channel, the ion pairs chosen for the analyte a Figure 2 1. Chemical structure of 5,7 DMF and I.S

PAGE 28

28 Figure 2 2. Mass spectra of 5,7 DMF and I.S.

PAGE 29

29 Liquid liquid extraction method was chosen because of following three benefits it offers. First of all, it excludes the water soluble sodium and potassium which can be detrimental to the instrument due to involatile property. Secondly, it results in higher extraction and lower LLOQ than the protein precipitation method for 5,7 DMF. Thirdly, samples can be condensed through liquid liquid extraction, which is very beneficial to those samples with concentration range between LOD and LLOQ and cannot be detected otherwise. The mobile phase of 50/50 (v/v) of 20 mM ammonium formate in water (containing 0.1% formic acid)/acetonitrile (containing 0.1% formic acid) was chosen based on good peak shape and retention time of the analyte and I.S. The mass spectrometer was operated at the gas temperature of 300 C, gas flow of 8 L/min, nebulizer of 35 psi, sheath gas temperature of 400 C, sheath gas flow of 11 L/min, capillary voltage of 4500 V and cell accelerator voltage of 7 V. These parameters were optimized by the bui lt in software MassHunter Optimizer. The parameters that led to highest ion pair abundance and signal response were recorded and chosen. The Agilent phase column, was used to separate the ana lyte from the endogenous interferences in mouse plasma. As shown in Fig. 3 reconstituted residues of 5,7 DMF in mouse plasma demonstrated the specificity of this method to detect and measure 5,7 DMF. The extracted ion chromatogram (XIC) of 5,7 DMF (Fig. 2 3 A) has a single peak with the retention time of 3.7 min in blank mouse plasma through chan Fig. 2 3 B) has a single peak with the retention time of 3.6 min in blank mouse plasma through me is consistent with the hydrophobicity of the compounds. Although the analyte and I.S in this case have similar retention time, their

PAGE 30

30 mass chromatograms clearly distinguish from each other as they have very different transition ions. Thus there were no s houlder peaks or split peaks observed in any XIC of the analyte or I.S, which means there was no interference from other compounds or matrix during the quantification of either 5,7 DMF or the I.S. In addition, there was no carryover effect observed within the concentration range of the calibration curve in mouse plasma. The LC MS/MS method established in mouse plasma has good linearity of greater than 0.99, as shown in Fig. 2 4, with weighted factor of 1/x 2 Figure 2 3. XIC of the 5,7 DMF (2 ng/mL) and I.S (1000 ng/mL) in mouse plasma. A1 was XIC signal of analyte in blank mouse plasma; A2 was XIC signal of analyte at 2 ng/mL in mouse plasma. B1 was XIC signal of I.S in blank mouse plasma; B2 was XIC signal of analyte at 1000 ng/mL in mouse plasma.

PAGE 31

31 Fi gure 2 4. Calibration curve of 5,7 DMF in mouse plasma. Figure 2 5. C alibration curve of 5,7 DMF in mouse liver y = 5.177E 004x + 2.902E 004 R = 0.9949 Weighted factor 1/x 2 0.0 0.2 0.4 0.6 0 200 400 600 800 1000 Analyte/I.S area ratio Concentration (ng/mL) Calibration curve of 5,7 DMF in mouse plasma y = 9.268E 004x + 0.002 R = 0.9970 Weighted factor 1/x 0.0 0.5 1.0 0 200 400 600 800 1000 Analyte/I.S area ratio Concentration (ng/mL) Calibration curve of 5,7 DMF in mouse liver

PAGE 32

32 Method V alidation Based on the method described above, the calibration curve has a linearity of 2 1000 ng/mL. The limit of detection for t his method is 1 ng/mL with signal to noise (S/N) ratio of greater than 3, and the lower limit of quantification for this method is 2 ng/mL with S/N ratio greater than 5. It was fully validated for concentrations of 2, 5, 50 and 500 ng/mL by six replicates for each concentration during intra run validation and triplicate for each concentration during three inter run validations. The accuracy and precision were presented in Table 2 1 and Table 2 2. The accuracy for intra run validation was from 94.4% to 106% with CV% ranging from 0.50% to 4.60%. The accuracy for inter run validation was from 100% to 107% with CV% ranging from 2.71% to 9.07%. As shown in Table 2 1 and Table 2 2, the means and CVs for all concentrations were within 15% difference, including LLO Q. According to FDA guideline [ 54 ], the accuracy and precision should be within 15% of the nominal concentrations except for the LLOQ, which should be within 20%. Table 2 1.Intra day (within day) validation of 5,7 DMF in mouse plasma (N=6).

PAGE 33

33 Table 2 2. Inter day (between day) validation of 5,7 DMF in mouse plasma (N=3*3). Matrix Effect and Recovery Yield Standards in mobile phase/neat solution, pre extracted spiked samples and post extracted spiked samples were prepared in triplicate analyze the matrix effect and recovery yield. Based on the data shown in Table 2 3, the matrix effect and recovery yield of 5,7 DMF in mouse plasma were 86.8 99.8% and 87.4 99.8%, respectively. Currently, there are no guidelines for the criteria of matrix effect and recovery yield. The matrix effect within 15% is generally accepted. In this case, the quantification method had a consistent matrix effect throughout low, medium and high concentrations of the calibration curve, which indicated that this method has little or negligible ion suppression. The recovery yield was proved consistent (87.4 99.8%) through different concentra tions, indicating the high extraction efficiency if this method. Table 2 3. Matrix effect and recovery yield of 5,7 DMF in mouse plasma (N=3).

PAGE 34

34 Stability of 5,7 DMF Stability of 5,7 DMF in stock To investigate the stability of 5,7 freshly prepared 5,7 DMF solution was compared with stock solution that had been stored in the freezer for 2 months. The mean ratio of stored solutions to freshly prepared 5,7 DMF was 0.86 (or 86%). The analysis result indicated that the degradation of the stock solutions in the freezer is less than 15% compared to freshly prepared 5,7 DMF solutions. Based on the result, the stored stock 5,7 DMF solutions were stable. Freeze thaw stability of 5,7 DMF in mouse plasma The results in Table 2 4 showed that different concentrations of 5,7 DMF in mouse plasma going thro ugh up to three freeze thaw cycles, the degradation of the analyte was all within 15%. Therefore, the results were reliable since the analyte was stable for three freeze thaw cycles. Table 2 4. Freeze thaw stability of 5,7 DMF in mouse plasma (N=3). Freeze thaw cycle 5,7 DMF Concentration (ng/mL) Ratio to freshly prepared samples 1 2 105 1 5 105 1 50 94 1 500 96 2 2 86 2 5 86 2 50 107 2 500 112 3 2 93 3 5 92 3 50 103 3 500 99

PAGE 35

35 Long term stability of 5,7 DMF in mouse plasma 5,7 DMF at concentrations of 2, 5, 50 and 500 ng/mL in mouse plasma were results in Table 2 5 indicated that the changes in the analyte concentration were within 15% compared with fre shly prepared mouse plasma samples of same nominal concentrations. Thus the plasma samples were stable in the storage conditions for up to 14 days. Since this period exceeded the time interval between the time of first plasma sample collected and the time of last sample analyzed, the PK study results were reliable. Table 2 5. Long term stability of 5,7 DMF in mouse plasma (N=3). Storage duration (Day) 5,7 DMF Concentration (ng/mL) Ratio to freshly prepared samples 4 2 105 4 5 88 4 50 93 4 500 108 7 2 97 7 5 96 7 50 90 7 500 104 14 2 103 14 5 98 14 50 88 14 500 97 Short term temperature stability of 5,7 DMF in mouse plasma The results shown in Table 2 6 indicated that the ratio of analyte to I.S in prepared samples that were exposed to room temperature for 24 h to the ratio in freshly prepared samples were between 94 and 102, indicating the degradation was less than

PAGE 36

36 5%. With in 24 h on the bench top, plasma samples preparation was finished. Therefore, the samples were considered stable during preparation. Table 2 6. Short term temperature stability of 5,7 DMF in mouse plasma (N=3). Post preparative stability of 5,7 DMF The results in Table 2 7 demonstrated that the ratios of QC concentrations of Day 2 were between 103 and 113 compared with those calculated of Day 1. Since the whole batch of PK samples were finished running within 24 h, the measured were considered stable within 24 h during the resident time inside the autosampler after preparation. Table 2 7. Post preparative stability of 5,7 DMF (N=3). Validation of quantification method in tissues As reported, our LC MS/MS quantification method on mouse plasma was proven to be accurate, reliable, fast, sensitive and selective [ 55 ]. It was fully validated and the precision and accuracy of various QC samples all passed the criteria based on FDA guideline [ 54, 56 ]. In this stud y, it was applied and validated in tissue matrices. As indicated by the FDA guideline and the newest draft guidance [ 54, 56 ], the change in

PAGE 37

37 the biological matrices within same species is the typical method modification and only partial validation is needed Further, it is indicated in these guidance that, the total number of QC samples should be either 5% of the total samples to be analyzed or at least 6 QC samples in total, whichever number is greater [ 54, 56 ]. Therefore, in this case, for each tissue anal ysis, we performed intra day validations on various concentrations with each concentration in triplicates (N=3) and the method validation results were shown in Table 2 8 The LC MS/MS methods in mouse tissues had calibration curves with good linearity, as indicated by high R 2 values of greater than 0.99 Fig. 2 5 to Fig. 2 13 represented mouse liver, kidney, intestine, brain, spleen, heart, lung, muscle and fat. Except mouse spleen, muscle and fat, most calibration curves were captured in one linear relati onship to describe concentrations between 2 and 1000 ng/mL. For spleen, muscle and fat, the calibration curves were split into 2 segments on the concentration of 100 ng/mL. Panels A represented lower calibration curves from 2 to 100 ng/mL whereas Panels B represented higher calibration curves from 100 to 1000 ng/mL. Our results indicated QC samples at LLOQ were within 20% of the nominal concentrations and QC samples at higher concentrations were within 15% of the nominal concentrations. Thus these quantific ation methods were successfully validated and sample analyses using these assays were reliable. The XIC of 5,7 DMF in tissue samples were shown in Fig 2 14 The sample preparation method for tissue samples was optimized based on the comparison of four tis sue preparation methods from a pilot study. In method A, 70% ice cold methanol was used as homogenate buffer then ethyl acetate was added prior to

PAGE 38

38 the liquid liquid extraction. In method B, PBS buffer was used as the homogenate buffer. 500 L of acidified acetonitrile (acetonitrile with 0.2% formic acid) was added for further protein precipitation. The samples were placed in 20C freezer for 5 min before centrifugation. The supernatant layer was dried and reconstituted prior to injection. In method C, aci dified methanol was used as the protein precipitation agent. Other steps were the same as that in method C. In method D, PBS buffer was used as the homogenate buffer and ethyl acetate was added for extraction. The detailed comparison of these four extracti on methods was listed in Table 2 9. Figure 2 6. Calibration curve of 5,7 DMF in mouse kidney. y = 10.05E 004x + 3.387E 004 R = 0.9992 Weighted factor 1/x 0.0 0.5 1.0 0 200 400 600 800 1000 Analyte/I.S area ratio Concentration (ng/mL) Calibration curve of 5,7 DMF in mouse kidney

PAGE 39

39 Figure 2 7. Calibration curve of 5,7 DMF in mouse intestine. Figure 2 8. Calibration curve of 5,7 DMF in mouse brain. y = 9.398E 004x + 0.002 R = 0.9988 Weighted factor 1/x 0.0 0.5 1.0 0 200 400 600 800 1000 Analyte/I.S area ratio Concentration (ng/mL) Calibration curve of 5,7 DMF in mouse intestine y = 8.564E 004x + 0.007 R = 0.9990 Weighted factor 1/x 0.0 0.1 0.2 0 50 100 150 200 Analyte/I.S area ratio Concentration (ng/mL) Calibration curve of 5,7 DMF in mouse brain

PAGE 40

40 Figure 2 9 Calibration curve of 5,7 DMF in mouse spleen.

PAGE 41

41 Figure 2 10. Calibration curve of 5,7 DMF in mouse heart. Figure 2 11. Calibration curve of 5,7 DMF in mouse lung. y = 9.744E 004x + 3.505E 004 R = 0.9985 Weighted factor 1/x 2 0.0 0.5 1.0 0 200 400 600 800 1000 Analyte/I.S area ratio Concentration (ng/mL) Calibration curve of 5,7 DMF in mouse heart y = 9.288E 004x + 0.014 R = 0.9937 Weighted factor 1/x 0.0 0.1 0.2 0 50 100 150 200 Analyte/I.S area ratio Concentration (ng/mL) Calibration curve of 5,7 DMF in mouse lung

PAGE 42

42 Figure 2 12. Calibration curve of 5,7 DMF in mouse muscle.

PAGE 43

43 Figure 2 13. Calibration curve of 5,7 DMF in mouse fat.

PAGE 44

44 Table 2 8. Method validations of 5,7 DMF in mouse plasma and tissues (N=3). 110 0.92 2 500 105 1.50 Kidney 107 1.58 106 5.10 (*denotes one determination)

PAGE 45

45 Discussions Compared to two assays on 5,7 DMF published in the past [44, 48] our assay is simpler and more sensitive. Sutthanut et al. reported a LLOQ of 5,7 DMF at 800 ng/mL using GC whereas Mekjaruskul et al. reported a LLOQ of 5,7 HPLC UV. Our result of LLOQ of 2 ng/mL demonstrates to be a significant improvement in the assay sensitivity, which is useful in quantification in terminal phase of plasma samples. Furthermore, in terms of sample preparation, our study used ethyl acetate to extract sample only once, compared to Mekjaruskul's method of using et hyl acetate three times to extract analytes. In addition, our assay is much faster than the other two assays since the retention time for 5,7 DMF in our assay is 3.7 min compared with 9 min and 26 min for GC method and HPLC method, respectively. Our quanti fication methods in mouse plasma can be applied in clinical trials and adapted to measure 5,7 DMF concentration in human plasma and human liver, kidney, brain, intestine, spleen, heart, lung, muscle and fat tissues as well. One interesting finding about t he matrix effect was that, in matrix blank such as 2 3 A, ouse plasma was not due to error or contamination in the samples. It was constantly showing up in the XIC which could negatively affect the LLOQ. But it was observed that with increase in the analyte concentration, the noise onally. But even with the strongest cleaning organic analyte getting stuck inside certain part of the column due to the physic chemical interaction between the chemica l and the column stuff. More often than not, such

PAGE 46

46 interference appeared in LC MS/MS. The bottom line is to ensure such signal to noise ratio is greater than 5. And in our case, the LLOQ of 2 ng/mL had a signal to noise ratio of above 7. So the ghost peak w than that.

PAGE 47

47 Figure 2 14 XIC of the 5,7 DMF and I.S in mouse plasma and tissues

PAGE 48

48 Table 2 9. Comparison among different sample preparation methods on quantification of liver samples from a pilot s tudy. 2h 3 21.8 2h 3 17.0 C 2h 3 16.5 2h 3 Summary A fast, accurate, sensitive and selective quantification method of DMF was established to warrant further PK study in m ic e. The LLOQ of this method is 2 ng/mL. This quantification method was validated and the matrix effect and recovery yield were consistent and reproducible. The stability of 5,7 DMF in stock solution, in 80 C freezer for up to 2 weeks and the stability in up to three freeze thaw cycles were also proved. The developed method was successfully adapted to and validated on mouse tissues, including liver, kidney, heart, lung, spleen, brain, intestine, muscle a nd fat. This warrants further PK study of 5,7 DMF in mice.

PAGE 49

49 CHAPTER 3 PHARMACOKINETICS AND TISSUE DISTRIBUTION OF 5,7 DMF IN MICE Background 5,7 DMF has been evaluated extensively in vitro and the results showed that 5,7 DMF has many beneficial pharmacological activities. For example, it has anti inflammatory activity [30], vasorelaxation and cardioprotection effect by increasing potassium eff lux and inhibiting calcium influx [31], and selectively inhibitory effect conducted a series of in vitro studies to evaluate the chemotherapeutic and chemopreventive po tential of this compound [33 35]. They found that 5,7 DMF can prevent hepatic carcinogenesis by inhibiting Cytochrome P450 (CYP) 1A1 activity, which subsequently decreased the carcinogenbenzo[a]pyrene (BaP) induced DNA adduct formation [35]. The chemoprev entive effect of 5,7 DMF was also observed in human esophageal cancer cells and human oral carcinoma cells by regulating CYP1B1 activity [34 35]. 5,7 DMF was also reported to have chemosensitizing effect in hetapocellular carcinoma cells and human leukemic cells through tumor necrosis factor related apoptosis inducing ligand (TRAIL) mediated apoptosis [36 37]. In addition, 5,7 DMF was shown to improve the accumulation of rhodamine 123 into the cells with overexpression of P glycoprotein (P gp), an efflux tr ansporter which is present in various cancer cells and plays an important role in multi drug resistance (MDR) [27]. In addition to P gp inhibition, 5,7 DMF also has inhibitory effect on other two efflux transporter members multidrug resistance associated proteins (MRPs) and Breast Cancer Resistance Protein (BCRP).

PAGE 50

50 It is noteworthy however, that most of the literature reported the beneficial efficacy of 5,7 DMF in vitro, but the report on its in vivo PK profile is very limited. Currently, there is one pa per reporting PK in killifish, and two papers reporting PK in rats. The PK in mice and other animal species are still missing. Therefore, in this chapter, we will explore the PK of 5,7 DMF in mice to study its dose proportionality, bioavailability and tiss ue distribution. Previously, the analytical methods available can only quantify 5,7 DMF with the lower limit of quantification (LLOQ) of more than 800 ng/mL [ 44 48] In our recent study, we established a fast, accurate and sensitive LC MS/MS method to quantify 5,7 DMF in mouse plasma with the LLOQ of 2 ng/mL [ 55 ]. It was applied to this study to investigate the PK profile and tissue distribution of 5,7 DMF in mice in this chapter. Materia l and M ethods Chemicals and R eagents 5,7 t rimethoxyflavone (TMF, >99%), shown in Fig 1 were purchased from Indofine Chemical Company, Inc (Hillsborough, New Jersey, US). Ammonium formate (99%) was obtained from ACROS Organics (New Jersey, US). Analytical HPLC grade acetonitrile (ACN), isopropanol, water, ethanol, phosphate buffered saline (PBS) 10X solution, Triton lysis buffer (pH 8.0) and formic acid were purchased Fisher Scientific (Pittsburg, Pennsylvania, US). Analytical spectrophotometric grade ethyl acetate (99.5%) was obtained from Alfa Aes ar (Ward Hill, Massachusetts, US). Cyclohexane Hexahydrobenzene (99.5%) was purchased in VWR International LLC (Philadelphia, Pennsylvania, US). Heparin treated mouse plasma was purchased from BioreclamationIVT (East Meadow, New York, US). Poly(ethylene gl ycol) Mn 400 (PEG400) were purchased through Sigma Aldrich (St. Louis, Missouri, US). Heparin

PAGE 51

51 Injectable (1,000 U/mL) was purchased from Patterson Vet Generics (Devens, Massachusetts, US). Isoflurane was provided in Animal Facility surgery room under IACUC protocol. Animals and In Vivo S tudy 4 weight of 26. 2 g were obtained from Harlan Laboratories. Before experiment, all mice were settled and housed in the University of Florida (UF) Ani mal Research Facility for a week following a 12 hr light/dark cycle. All mice had access to normal/standard diet and water and this in vivo study was carried out in accordance with IACUC protocol evaluated and approved by review board of UF Animal Care and National Institutes of Health. In dose linearity study, 90 mice were randomly separated into 3 dosing groups with each group of 30 mice assigned randomly into 10 time points (up to 48 h) triplicate mice at each time point (N=3). 10, 25 or 50 mg/kg 5,7 DMF was administered through penile vein injection (IV) to the mice. The mice had free access to food and water before and after oral administration. Mouse plasma was collected after immediately centrifuged at 2,000 g. Supernatant layer of the blood was th us collected as plasma and all the plasma and tissue samples were stored at 80C until analysis. In bioavailability study, 60 mice were randomly separated into either IV or Oral group with each group of 30 mice assigned randomly into 10 time point s tripli cate mice at each time point (N=3). 5,7 DMF was first dissolved in dimethyl sulfoxide at the concentration of 20 mg/mL and prior to dosing, it was further diluted as solution in the vehicle consist of 35% PEG400, 2% ethanol and 63% deionized water at the f inal concentration of 2 mg/mL. This solution containing 2 mg/mL 5.7 DMF was stable in the

PAGE 52

52 vehicle for at least 24 hr under room temperature. The target dose of 10 mg/kg 5,7 DMF was administered through either penile vein injection (IV) or through oral gava ge to the mice. The mice had free access to food and water before and after oral administration. Blood, heart, lung, gut, brain, liver, kidney, spleen, muscle and adipose tissues were collected at time points 5, 15 and 30 min and 1, 2, 3, 4, 8, 12 and 21 h r after oral dosing. Mice were euthanized by isoflurane followed by cervical dislocation. The heparinized blood was immediately centrifuged at 2,000 g for 8 min with a mini centrifuge. Supernatant layer of the blood was thus collected as plasma and all the plasma and tissue samples were stored at 80C until analysis. Sample Preparations and LC MS / MS Assays The LC MS/MS condition was the same as previously reported [20]. Briefly, under positive mode of Agilent triple quadruple 6460, 5,7 DMF (m/z 283 > m/z 239) and I.S (m/z 313 > m/z 298) were simultaneously monitored with isocratic elution at a flow rate of 0.4 mL/min. The mobile phase is 50/50 (v/v) of 20 mM ammonium formate in water and acetonitrile. Plasma sample preparation method was completed followi ng previous reported method [ 55 ]. For tissue samples, tissue was weighed and added PBS buffer (pH=7.4) based on 10 fold volume of each tissue weight. They were cut into smaller pieces before homogenization. Then 10 L of I.S was spiked into 100 L of tissu e homogenate. Other steps were the same as plasma sample preparation. After vortex mixing for 1 min, 600 L of ethyl acetate was added for liquid liquid extraction on a mechanical shaker for 5 min. Then the supernatant layer was transferred to clean tubes after the centrifuge at 17,000 rpm under 4 C for 4 min, and placed under nitrogen for evaporation. The residue was reconstituted with 100 L of mobile phase and centrifuged at 17,000 rpm under 4 C for 4 min to get rid of precipitation. 5 L of the

PAGE 53

53 supern atant aliquot was injected to the LC MS/MS. Calibration curve were prepared at concentrations 2, 5, 10, 20, 50, 100, 200, 500 and 1000 ng/mL in blank mouse tissues. Pharmacokinetic A nalysis The non compartmental analysis (NCA) was employed to calculate t he PK z ) is the elimination rate constant at the terminal phase, and it was calculated as the slope of the linear regression on the terminal data points. C max is the maximal plasma con centration based on the experimental data. The area under the plasma concentration time curve from time zero to the last time point (AUC t ) was calculated by linear trapezoidal rule. AUC infinity (AUC inf ) was determined by AUCt plus extrapolated portion C t / z Terminal half Dose/AUC inf The apparent volume of distribution (V z z The partition coefficient in tissue (K p ) was calculated by AUC t (area unde r the curve from 0 to the last time point that the analyte can be measured) of tissue divided by AUC t of plasma. Results LC MS / MS Method Development and Validation As reported, our LC MS/MS quantification method on mouse plasma was proven to be accurate, r eliable, fast, sensitive and selective [ 55 ]. It was fully validated and the precision and accuracy of various QC samples all passed the criteria based on FDA guideline [ c ]. In this study, it was applied and validated in tissue matrices. As indicated by the FDA guideline and the newest draft guidance [ 54, 56 ], the change in the biological matrices within same species is the typical method modification and only partial validation is needed. Further, it is indicated in these guidance that, the total

PAGE 54

54 number of QC samples should be either 5% of the total samples to be analyzed or at least 6 QC samples in total, whichever number is greater [ 54, 56 ]. Therefore, in this case, for each tissue analysis, we performed intra day validations on various concentrations with each concentration in triplicates (N=3) and the method validation results were shown in Table 2 8 Our results indicated QC samples at LLOQ were within 20% of the nominal concentrations and QC samples at higher concentrations were within 15% of the nomina l concentrations. Thus these quantification methods were successfully validated and sample analyses using these assays were reliable. The XIC of 5,7 DMF in tissue samples were shown in Fig 2 1 4 The sample preparation method for tissue samples was optimized based on the comparison of four tissue preparation methods from a pilot study. Among these four methods, the measured extraction efficiency was the highest in method D. The detailed comparison of these four extraction methods was listed in Table 2 9. Method D exhibited constant higher extraction efficiency for different sample ID. Besides, the procedures of method D were relatively simple and less time consuming. Due to the simple and efficient procedure of method D, it was applied for tissue preparations in this study. Dose Linearity of 5,7 DMF in Mice 5,7 dmf solution within the dosing range of 10 to 50 mg/kg resulted in linear PK profile, as shown in Fig 3 1. Time courses of 10 50 mg/kg admini stration were almost parallel to each other. The NCA parameters were listed in Table 3 1 The area under the curve (AUCinf) and peak plasma concentration (Cmax) of each dose was proportional with dose.

PAGE 55

55 Table 3 1. Dose linearity of 10 50 mg/kg IV dose in mi ce (N=3) Bioavailability of 5,7 DMF in M ice When compared the NCA results between IV and oral dose of 10 mg/kg 5,7 DMF solution, as shown in Table 3 2, the mean AUC of IV and oral routes were 3886 and 537, respectively, which made the calculated bioavailability at 13.8%, based on the equation where AUC po is the AUC after oral dose and AUC iv is the AUC after IV dose. The time course of both dosing routes was shown in Fig. 3 2. The curve shape at the terminal phase seemed similar. There was no sign of flip flop phenomenon. The oral absorption was fast. The IV dosing route, plasma concentration was detectable up to 4 h while in oral route, the plasma concentration was detectable up to 12 h. Table 3 2. Bioavailability of 5,7 DMF following 10 mg/kg IV and oral dose (N=3) V_F (L/kg) 2.52 Pharmacokinetic Analysis and Tissue Distribution in Mice The NCA result in Table 3 3 showed that the maximal plasma concentration following oral dose of 10 mg/kg 5,7 DMF was ng/mL. The terminal half life t 1/2 was hr. The volume of distribution by the area method V z _F was

PAGE 56

56 L/kg. Clearance (CL_F) was L/hr/kg. The drug exposure AUC t and AUC inf was hr*ng/mL, respectively. Table 3 3 Non compartmental parameters of 5,7 DMF in mouse plasma after oral administration of 10 mg/kg 5,7 DMF (N=3). The data were presented as mean and standard deviation. The concentration time profiles of plasma, liver, kidney, gut and brain were presented in Fig 3 3 Through the oral gavage, the C max of 5,7 DMF was quickly reached among plasma and tissues within 0.5 hr after dosing due to fast absorption, with mean T max ranging from 0.14 to 0.36 hr. The semi log scale plots indicated a bi exponential disposition pattern of 5,7 DMF. The plasma concentr ation after T max dropped drastically due to fast and vast distribution into tissues and then it followed a long terminal phase probably because of the contribution of the 5,7 DMF in the tissue back to plasma. The analyte was detectable up to 21 hr in mouse plasma, gut, liver, kidney, spleen and adipose. The time course of mouse liver, kidney, gut, spleen and adipose followed the same pattern as that of mouse plasma.

PAGE 57

57 Figure 3 1. Dose proportionality in mice. Plasma PK profile (mean SD) of 10 (A), 25 (B) and 50 (C) mg/kg IV dose of 5,7 DMF(N=3).

PAGE 58

58 Figure 3 1. Continued As shown in Fig 3 4 the 5,7 DMF concentration in mouse tissues were much higher than that in mouse plasma. Except for muscle and adipose, other tissues had the C max higher than that in the plasma, ranging from 1.75 to 9.96 fold. The highest tissue C max was in mouse gut, followed by liver, brain, kidney, heart, spleen, lung, muscle and adipose. As indicated by the NCA analysis results listed in Table 3 4, AUC t of gut, liver, kidney, brain and spleen were , and hr*ng/g, respectively. Thus, 5,7 DMF exposure in these tissues were 2.35 to 12.9 fold higher than that of mouse plasma. It is shown that 5,7 DMF through oral dose was most abundant in gut, followed by liver, kidney, brain, spleen, heart, lung, adipose and lastly, muscle.

PAGE 59

59 Figure 3 2. Plasma PK profile (mean SD) of IV and oral dose of 10 mg/kg 5,7 DMF in mice (N=3) Table 3 4 PK parameters of 5,7 DMF in mouse tissues after oral administration of 10 mg/kg 5,7 DMF (N=3). The data were presented as mean and standard deviation AUC t (hr*ng/g) K p Intestine 12.9 8.71 3.94 2.86 2.31 1.47 0.99 0.65 The PK profile of plasma and nine other tissues was shown in Fig. 3 4. The calculated K p values for major tissues were listed in Table 3 4. K p values for intestine,

PAGE 60

60 liver, kidney, brain, spleen, heart, lung, fat and muscle were 12.9, 8.71, 3.94, 2.86, 2.35, 2.31, 1.47, 0.99 and 0.65, respe ctively. They will be further evaluated in our future investigation on the physiologically based PK modeling of 5,7 DMF. Discussions Breast Cancer Resistance Protein (BCRP) is one of the main efflux transporters associated with MDR. 5,7 DMF has been repor ted to be a potent inhibitor of BCRP, with an IC 50 39 40 ]. 5,7 DMF can significantly increase the accumulation of mitoxatrone (an anticancer agent) in BCRP overexpressed breast cancer cells and correspondingly enhance d the cytotoxicity of mitoxatrone in those cells at nano molar to low micro molar concentration range. In this case, under the dose of our study (10 mg/kg), 5,7 which can effectively inhibit BCRP at the cellular level and may represent clinical relevance in cancer treatment.

PAGE 61

61 Figure 3 3. Concentration time profiles of 5,7 DMF in mice following oral administration of 10 mg/kg 5,7 DMF (N=3). A, B, C, D and E represented the pharmacokinetic profile of 5,7 DMF in m ouse plasma, liver, kidney, small intestine and brain.

PAGE 62

62 The pharmacokinetic parameters of 5,7 DMF after 10 mg/kg oral dose in mice were determined in this study. Based on the previous literature, 10 and 30 mg /kg equivalent 5,7 DMF oral doses were proven safe in rats [ 44 ]. In addition to 10 mg/kg oral dose, we also evaluated 10, 25 and 50 mg/kg intravenous dose in our pilot study and no toxicity was observed at any dose tested, indicating the favorable safety profile of 5,7 DMF. In addition, our pilot stu dy had demonstrated the linear relationship between dose and concentration on 10, 25 and 50 mg/kg intravenous dose of 5,7 DMF. Therefore the C max and AUC can be reasonably estimated within that dosing range. It has been reported that the IC 50 value of 5,7 DMF for BCRP inhibition in vitro is 1.41 M and at this concentration 5,7 DMF can significantly increase the intracellular accumulation of mitoxantrone in the BCRP overexpressing cells [ 39 40 ]. Our result revealed that under 10 mg/kg oral dose, 5,7 DMF co uld easily reach the concentration of 1.41 M (equivalent to 398 ng/mL) in plasma and various tissues. Based on our results, it is reasonably to speculate that the BCRP inhibitory effect of 5,7 DMF observed in vitro can be extrapolated to in vivo. Similarl y, 5,7 DMF was reported to increase the doxorubicin accumulation into P gp overexpressed cell line in a concentration dependent manner within 1 300 M (282 84600 ng/mL) [ 27 ], and this in vitro P gp inhibitor effect should be able to be translated to in viv o as well since our PK data showed that the C max of 5,7 DMF in plasma and different tissues range fr om 1870 to 18600 ng/ml (Tables 3 1 and 3 2 ).

PAGE 63

63 Figure 3 4. Mean plasma and tissue PK profiles of oral dose of 10 mg/kg 5,7 DMF in mice (N=3). Terminal half life of 5,7 DMF calculated in plasma was 3.40 2.80 hr, so it might take 17 hr (5 half lives) to clear 5,7 DMF completely (>95%). T max for mouse gut, plasma, liver and kidney were 0.083 0.00, 0.14 0.10, 0.14 0.10 and 0.19 0.10 hr, respective ly, indicating rapid distribution of the 5,7 DMF from plasma to tissues [ 44 ]. The peak concentration in major tissues such as liver, brain, kidney, heart, and lung were higher than plasma peak concentration, with the C max ratio to plasma of 5.86, 3.88, 3.8 5, 2.67 and 1.62, respectively. The high tissue accumulation was also in accordance DMF was 7 fold higher than that of plasma level and the peak lung and kidney 5,7 DMF were almost 2 fold high er than plasma concentration [ 44 ].

PAGE 64

64 The large value of indicated extensive accumulation in [ 43 44 ]. This may indicate tissues as the favorable sites for 5,7 DMF to exert the greatest pharmacological action. Among different studies and different species, 5,7 DMF may have different tissue distribution profile. In our study, we found that in mice, the compound after oral dosing was most concentrated in g ut > liver > spleen > kidney DMF most accumulated in rats liver > lung > kidney > plasma [ 44 ] whereas > lung [ 43 ]. 5,7 DMF in killifish was reported to have highest concentration in liver > brain > gut > gill > skin [ 4 8]. But in general, liver and kidney had higher 5,7 DMF than other tissues and plasma probably because they are well perfused organs. It is noteworthy that liver was the major tissue for 5,7 DMF accumulation. 2 20 M (564 5640 ng/mL) of 5,7 DMF was achievable through this 10 mg/kg oral dose which was reported to inhibit benzo[a]pyrene induced DNA binding in liver [ 33 ]. Also, 2.5 10 M (705 28 20 ng/mL) of 5,7 DMF was achievable in vivo to chemosensitize Trail induced apoptosis in liver carcinoma cells [ 36 ]. Interestingly, 5,7 DMF could penetrate blood brain barrier (BBB) and blood testicular barrier, which was demonstrated by our study in mouse brains and 44 ]. This may imply promising medical use of 5,7 32 ]. Summary In summary, this study described the PK of 5,7 DMF in mice follo wing 10 mg/kg oral dose for the first time. It established the tissue distribution profile and PK parameters in vivo and provided further basis for physiologically based PK modeling,

PAGE 65

65 drug drug interaction studies, dosing regimen and prediction in clinical trials in the future.

PAGE 66

66 CHAPTER 4 PHARMACOKINETIC MODELING AND SIMULATION ON 5,7 DMF Background Pharmacokinetic models are relatively simple mathematical schemes that represent complex physiologic spaces or processes. Accurate PK modeling is important for precise determination of elimination rate. In recent years, progress has been made in PK modeling to better understand the absorption, distribution, metabolism, and excretion (ADME) of medications in vivo [ 57 ]. However, the PK model or parameters to describe the ADME of 5,7 DMF in vivo is missing to date, which hinders the further exploration or development of this promising compound. Therefore, in our study we explore d the appropriateness of using population PK as a tool to explain and predict 5,7 DM F in vivo. Empirical PK models normally take the form of one, two, or three compartment, depending on the drug disposition pattern in blood or plasma [ 58 ] And the most commonly used are one and two compartment model. The construction of compartmental mod el takes into account of the drug properties such as lipophilicity, permeability and solubility. One compartment model assumes that the whole body is like a well stirred container with drug rapidly reaching equilibrium inside it; whereas two compartment mo del assumes that the central compartment is like well perfused layer that exchange drug with tissue layer with permeability rate limited manner. Extrapolation of pharmacokinetics from animal model to human model usually depends on either allometric scaling or physiologically based pharmacokinetic modelling (PBPK) [ 57 ] Allometric scaling is a rather empirical approach, in which volume of distribution wi ll be scaled up in linear fashion while clearance will be scaled up in

PAGE 67

67 empirical power of 0.75. PBPK is ad vantageous in semi mechanistic based to integrate compound dependent properties and physiology dependent characteristics. But their development tends to be resource demanding and time consuming Thus in this chapter, we focused mainly on the population pha rmacokinetic modeling tools to explain the ADME of 5,7 DMF in mice and then to predict the plasma concentration of 5,7 DMF in human of 70 kg body weight. Methods Assumptions and Model Building In our study, we mainly focused on classi cal compartmental mod el on oral plasma PK data and the models are shown in Fig. 4 1. One two and three compartmental models will be explored and compared with and without Tlag, with zero or first order absorption. A one compartmental model assumes linear pharmacokinetics a nd immediate distribution and equilibrium of the drug throughout the body. A two compartment model consists of two parts : The first (central) compartment X 1 is identified with the blood and organs that are well perfused with the rate limiting step of blood perfusion The second (peripheral) compartment X 2 describe d for tissues that are less perfused with the rate limiting step of pe r meability A three compartmental model takes into account the drug distribution into deep tissue and relatively shallower tiss ue.

PAGE 68

68 Figure 4 1. Structural of one two and three compartmental model and corresponding differential equations.

PAGE 69

69 Figure 4 1. Continued The structural models of one two and three compartment were compared based on fitting plots, diagnostic plots and statistical criterion such as 2LL value, Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). On top of the base model with first order absorption K a zero order absorption was also considered into the model to compare wi th the fitting, precision of fitted parameters and AIC or BIC to make the final decision. As for oral absorption, the lag time (Tlag) in absorption phase was also taken into consideration. Models with or without Tlag were compared as well. Population Model ing Analysis The population PK (PopPK) approach was used in this study to cover the random variation into the PK model. Phoenix version 1.3 was used for this purpose. Since it was constructed on mouse plasma data with triplicate (N=3) mice sacrificed at ea ch of ten time points, the total of 30 data points were serving as individual s pooled together.

PAGE 70

70 Therefore, the nave pool population method was chosen in terms of modeling. In this case, there was no differentiation on between subject variability or intra subject variability, because all sort of vari ability is lumped into residual/ error model. In this study, we explored the residual model in additive, log additive and multiplicative manner. Their equations are listed below. Additive model Log additive model Multiplicative model Comparisons among different residual models were based on goodness of fit diagnostic plots and statistical criterion such as 2LL value, AIC and BIC. C ovariates, however, were not investigated in our study, because the plasma data were collected from mice instead of humans. So less confounding factors have been reported or collected on 5,7 DMF in mice before.

PAGE 71

71 Model V alidation After the final model was se lected, it was internally validated by population predictive check (VPC). Basically, it is a comparison between observed data and simulated data. VPC compared a different kind of prediction other than diagnostic plots. They are based on simulations of mode l predictions including random effects. Summary measures of the distribution of predictions and observations were compared visually. The results included 5% (lower), 50% (mean) and 95% (upper) percentiles. Allometric S calling Scaling of a dose is expected when predicting first in human doses for clinical trials, dose extrapolation in veterinary practice and dose extrapolation for experimental purposes [ 59 ]. Allometric scalling was conducted on V 1 V 2 C L and C L 2 The average body weight of the experimental mice was 26.2 g, which was equivalent to 0.0262 kg. The V 1 and V 2 in mice model were scaled up to 70 kg body weight human in linear fashion; so that the power b for V 1 and V 2 was fixed at 1. The Cl and Cl2 were scaled up to 70 kg body weight human in less than proportion fashion; which means the power b for C L and C L 2 was fixed at 0.75 empirical value [ 60 ]. Absorption rate K a and bioavailability F of 5,7 DMF were fixed to the values in mice. Model S imulation and Prediction With the final model of two compartmental first order absorption without Tlag model in log additive residual model, and the scaled up PK parameters were plugged in simulation mode of Phoenix the model was extrapolated to predict human of 70 kg in body weight.

PAGE 72

72 Resu lts Population Modeling Analysis The comparisons among different models were listed in Table 4 1. Based on the results shown in Table 4 1, one compartmental model was not good in terms of fitting and AIC/BIC. The standard error and the precision (CV%) of fitted PK parameters were huge (>100%) for one compartmental model. Two compartmental and three compartmental base model s (without Tlag) both led to good fitting. So statistical criterion such as 2LL, AIC and BIC values played a role in decision making. G enerally, AIC and BIC is balancing between goodness of fit and overparameterization So the preferred model is the one with the minimum AIC/BIC value since it rewards goodness of fit (as assessed by the likelihood function), and includes a penalty for incr easing number of estimated parameters. AIC takes the function of where k is the total number of parameters in the model and L is the maximal likelihood. BIC takes the function of where L is maximal likelihoo d, n is the number of observed data and k is the number of parameters.

PAGE 73

73 Table 4 1. Model comparison under PopPK approach in Phoenix.

PAGE 74

74 in negative value, log additive model provides smaller AIC/BIC with better fitting in two and three compartmental model. So in this case, we focused on log additive error model. Compared wi th three compartm ent model, two compartment al model had similar residual, QQ (data quantile theoretical quantile) and DV (dependent variable) vs PRED (prediction) plots. Further, two compartmental model has smaller value in AIC/BIC than three compartment model. So the two models had AIC values of 44.2 versus 47.4, respectively. In this case, two compartmental model was chosen as the base model. When further compare two compartmental model and that with Tlag, the fitting generally look similar; but AIC and BIC punish the inc rease in total number of parameters. So the two models had AIC values of 44.2 versus 46.0, respectively. Based on minimizing AIC/BIC principle, we chose two compartmental model without Tlag. Since in absorption modeling, zero order absorption rate constan t could be considered instead of first order rate constant, we compared two compartmental model with either zero order (K 0 ) or first order (K a ) absorption. It is interesting that the AIC/BIC values for these two models are the same; although there was a dr astic increase when K 0 replace K a in three compartmental model, cha ng ing from 47.4/57.5 to 176/186. When looking at the fitting figure and parameter estimation table closely, we found that the fitting is better for two compartmental model with first order absorption; furthermore, the fitted parameters in two compartmental model with first order absorption had smaller standard error and CV% than two compartmental model with the zero order absorption

PAGE 75

75 rate thus we chos e two compartmental model with first order absorption as our final model for further simulation and prediction. The fitting plots for the final model are shown in Fig. 4 2. Fig. 4 2A showed goodness of fit plot in log transformation in DV and PRED versus time after dose. Fig. 4 2B showed goodness of fit plot without log transformation in DV and PRED versus time after dose. In either plot, the model fit the observed data in a decent way with reasonable variation. And the fitted parameters were listed in Tab le 4 2. Table 4 2. PopPK parameters for the final model in mouse Parameter Population mean estimate SE CV% K a (/h) 6.04 1.07 17.8 V 1 (L) 0.05 0.03 56.1 V 2 (L) 0.56 0.41 73.8 C L (L/h) 0.55 0.08 14.3 C L 2 (L/h) 0.04 0.02 40.2

PAGE 76

76 Figure 4 2. Overall g oodness of fit plots with log transformation (A) and without log transformation (B).

PAGE 77

77 And the individual fi tting plot was shown in Fig 4 3 with A, B, and C panels representing the goodness of fit plots of ID 1, 2 and 3. Figure 4 3. Individual goodness of fit plot.

PAGE 78

78 Figure 4 3. Continued Diagnostic plots in our study included the DV vs PRED, DV vs IPRED, CWRES (conditional weighted residual) vs PRED, CWRES vs TAD (time after dose) and QQ plots. These results were shown in Fig. 4 4 Fig. 4 4 A indicated that the predicted concentration were in line with the observed concentration, which means good fit; same for Fig. 4 4 B, dependent variable versus individual prediction. In the CWRES vs PRED plot, there is no trend between residual and prediction, which m eans the assumptions for modeling were valid; and there was no bias in this model. The random pattern was also observed in Fig. 4 4 D, CWRES vs time after dose, which means there was no trend going with time. So there was no time trend to lead to any bias i n the model. In both Fig. 4 4 C and Fig. 4 4 D plots, most of the dots were within 2 to 2 line so that only a few outliers fell out of this interval, which was a good sign indicating the goodness of fit. Fig. 4 4 E, QQ plot showed a dissection line of y=x li near relationship, which means

PAGE 79

79 the weighted residual met the expectation of standard normal quantile. There was no obvious deviation or screwness, in which normality assumption was met. The major goal for diagnostic plots wa s to check the assumptions of i ndependence, equal variance and normal distribution for the model. Residual plot can check independence and equal variance as well. By plotting residual against time or prediction, the random scattering indicated the independence of time or prediction and the assumption of equal variance were met. The robustness was demonstrated when no departure from the model assumptions were observed.

PAGE 80

80 Figure 4 4 Diagnostic plots for the final model.

PAGE 81

81 Figure 4 4. Continued

PAGE 82

82 Figure 4 4. Continued. Model V alidation Model validation was performed by VPC, as shown in Fig. 4 5. It showed that VPC had good agreement with observations. And observation intervals were within prediction intervals; there was no sign of model mis specification, under prediction or ov er prediction.

PAGE 83

83 Figure 4 5. Visual predictive check for two compartmental final PopPK model. With this final model and its PK parameters, the simulated mouse plasma PK profile for oral administration of 10, 25 and 50 mg/kg were shown in Fig. 4 6 to 4 8 In these plots, only mean concentrations were shown. 1 10 100 1000 10000 0 3 6 9 12 5,7 DMF concentration (ng/mL) Time after dose (h)

PAGE 84

84 Figure 4 6. Simulated mouse plasma PK profile following oral dose of 10 mg/kg 5,7 DMF. Figure 4 7. Simulated mouse plasma PK profile following oral dose of 25 mg/kg 5,7 DMF. Figure 4 8. Simulated mouse plasma PK profile following oral dose of 25 mg/kg 5,7 DMF. 1 10 100 1000 10000 0 3 6 9 12 5,7 DMF concentration (ng/mL) Time after dose (h) 10 100 1000 10000 100000 0 3 6 9 12 5,7 DMF concentration (ng/mL) Time after dose (h)

PAGE 85

85 Allometric S calling Allometric scalling to human of average body weight of 70 kg resulted in parameters was shown in table 4 3 So the mean Ka was 6.04 h 1 central volume of distribution was 0.05 L, peripheral volume of distribution was 0.55 L, clearance was 0.55 L/h and the intercompartmental clearance was 0.04 L/h. Table 4 3 Scaled up PK parameters in two compartment model for human species Parameter Mean (SE) in mice Scaled for humans K a (/h) 6.04 (1.07) 6.04 V 1 (L) 0.05 (0.03) 131 V 2 (L) 0.56 (0.41) 1484 C L (L/h) 0.55 (0.08) 204 C L 2 (L/h) 0.04 (0.02) 16.1

PAGE 86

86 Model S imulation and Prediction With the final model of scaled up parameters, the simulation of human plasma concentration time profile fol lowing oral dose of 10 mg/kg 5, 7 DMF solution were shown in Fig. 4 9 Panel A showed the simulation figure generated by the software under log scale whereas panel B was the figure after transformation back to linear scale. Both graphs indicated the 5, 50 and 95 percentiles. Predicted human plasma concentration following oral dose of 10 mg/kg 5,7 DMF. The predicted half life is 3.45 h, T max is 0.25 h, C max is 9860 ng/mL, AUC inf is 21200 h*ng/mL. Currently, however, there has not been any clinical data to externally validate this model. But our simulation could serve in the ne ar future as the starting point for prediction of human plasma concentration for future clinical trials possibly. Besides, our report will be the first to describe the modeling and simulation on 5,7 DMF, which may shed light on further model refinement.

PAGE 87

87 Figure 4 9 Simulation of 5, 7DMF concentration time profile in 70 kg human being after oral dose of 10 mg/kg 5,7 DMF (N=3).

PAGE 88

88 Discussions In the model selection part, first order absorption rate was more reasonable than zero order absorption rate becaus e usually, the controlled release or IV infusion enhibit constant doing rate into the body. Allometric scaling across different species can help understand the implication from preclinical data to clinical trials and it is rational tool despite its limita tions [6 1 6 2 ]. Usually the extrapolation among species is based on parameter estimation on data of at least three species [ 63 ]. The single point extrapolation based on one animal species to corresponding human being could not ensure accuracy [ 64 ] Besides this single point method, there are other extrapolation methods on clearance using brain weight longevity, GFR, bile flow or UDPGT [6 1 ]. Depending on different properties of the drug, the predictability of each method varies. For this study, there is too limited animal species available. Furthermore, the physic chemical characteristics of 5,7 DMF have not been fully captured yet. Based on the PK profile, the parent compound seems to undergo fast clearance. With historic knowledge that the excretion of 5,7 DMF through urine and feces was less than 5% of the total dosing, we assume that the majority of the parent compound went through metabol ic pathway, so that this simple allometry method should be a good estimate. First order absorption rate constant Ka was not scaled up to predict human concentration. It was not well discussed by literature probably because of the large difference in absorp tion among species [ 59, 65, 66 ]. Summary PK model was formulated for the first time to describe 5,7 DMF in vivo.

PAGE 89

89 Two compartmental model best fit the observed plasma data. Its validity was evaluated through diagnostic plots, fitting figures and statistical criteria. The final model incorporated first order absorption without lag time. This model was helpful to describe current observation in mice and may predict the plasma levels of 5,7 DMF in human. For example, t he fitted parameters could also be useful for future PBPK modeling. K a can be used as initial estimate in coding. The predicted volume of distribution can be also estimate as tissues for perfusion rate limited model.

PAGE 90

90 CHAPTER 5 CONCLUSIONS Summary W e established for the first time a sensitive, fast and selective quantification method using LC MS/MS to quantify 5,7 DMF in mouse plasma and tissues. Further we reported for the first time the dose linearity, PK and tissue distribution of 5,7 DMF in mice. Last but not the least, we constructed for the first time PK modeling and simulation of 5,7 DMF in vivo. Our bioanalytical method was fully validated and all of the fundamental parameters in method validation, including accuracy, precision, sensitivity, selectivity, recovery and sta bility were evaluated thoroughly in mouse plasma. The calibration curve covered 2 1000 ng/mL with the lower limit of quantification (LLOQ) of 2 ng/mL. The inter run and intra run precision and accuracy were less than 15% of nominal concentrations. The matr ix effect and recovery yield were within 15% of nominal concentrations. This sensitive and specific quantification method was also adapted to quantify mouse tissues for further PK studies in vivo. To investigate the PK properties and tissue distribution, 5,7 DMF was first dosed intravenously (IV) in a parallel study of 10 to 50 mg/kg 5,7 DMF to mice. The AUC was demonstrated dose proportional within the dose range. And then in the following study of 10 mg/kg oral dosing, mouse plasma, heart, lung, liver, kidney, intestine, brain, spleen, muscle and fat tissues were collected and analyzed using LC MS/MS. The calculated bioavailability was 13.8%. Following oral dosing, the maximal 5,7 DMF concentrations in plasma and tissues were reached within 30 min. The peak plasma

PAGE 91

91 concentration (C max ) was 1870 1190 ng/mL, and area under the curve (AUC t ) was 532 165 hr*ng/mL and terminal half life was 3.40 2.80 hr. The volume of distribution was 90.1 62.0 L/kg. Clearance was 20.2 7.49 L/hr/kg. Except for muscle and adipose, other tissues had higher C max than plasma, ranging from 1.75 to 9.96 fold. After oral administration, 5,7 DMF was most abundant in gut, followed by liver, kidney, brain, spleen, heart, lung, adipose and muscle. The partition coefficient (K p ) of these tissues were 0.65 to 12.9. We further constructed compartmental modeling on mouse plasma and two compartmental model with first order absorption and without Tlag model best described the oral plasm a data. The fitted PK parameter K a was 6.04 1/h, V 1 was 0.05 L, V 2 was 0.56 L, CL was 0.55 L/h, and CL 2 was 0.04 L/h. This final model was extrapolated to human of average body weight of 70 kg with V 1 V 2 CL and CL 2 scaled up, which may provide insight for 5,7 DMF concentration prediction in human plasm a. PK model was formulated for the first time to describe 5,7 DMF in vivo. Two compartmental model best fit the observed plasma data. Its validity was evaluated through diagnostic plots, fitting figures and statistical criteria. The final model incorporat ed first order absorption without lag time. This model was helpful to describe current observation in mice and may predict the plasma levels of 5,7 DMF in human. The fitted parameters could also be useful for future PBPK modeling. Perspective Given the c hallenges in the complex scenario, PBPK tools are popular for orally administered compounds as the absorption depends on drug related factors and also physiology of GI tract [ 67 ]. incorporated measured p arameters that can impact drug ADME [ 68 ] The major

PAGE 92

92 advancement and wide application nowadays are due to refined characterization of human body [ 69 70 ], progressive understanding of total body water [ 71 ] and development of a variety of in silico software [ 67 ] including Simcyp, Gastroplus, Matlab and so on. Other software such as ADAPT is less user friendly compared with Simcyp and Gastroplus. But either way, the essence for PBPK modeling utilizes differential equations and in vitro measured and preclinical data to gain mechanistic insight. It is especially useful to extrapolate across dosing routes and across different animal species and to understand dose selection in clinical trials [ 72 ] special populations such as pediatric patients and pregnant women [ 7 3 75 ] and to investigate DDI studies [ 76 ] Its application does not only exist in drug discovery and development, but also helps risk assessment and environmental protection [ 77 79 ]. With the fitted empirical model, our ultimate goal is to construct a PBP K model using mouse plasma and tissue PK data, and then extrapolate to human species to predict human plasma and tissue levels of 5,7 DMF. The physiologically based pharmacokinetic (PBPK) model was introduced in 1924 [ 8 0] which incorporated physiologically meaningful organs into a mass balance differential equation connected by body fluid fluxes to describe fate of the substance (Fig 5 1) and local model could be perfusion rate limited or permeability rate limited, as shown in Fig 5 2. It generally has thr ee different types: whole body PBPK, partial PBPK and metabolic PBPK [ 8 1 8 3]. The mathematical description of mass balance for each compartment also depends on the mechanisms of drug ADME. But compared to conventional approach, PBPK has advantages in that, first of all, it is an useful tool to extrapolate model across species; secondly, it allow the extrapolation among different doses and dosing routes; thirdly,

PAGE 93

93 since each compartment in the model has physiological meaning, the model enables us to predict t he tissue concentration in human without actually collect human tissues; lastly, it is widely applied nowadays since it includes both extrinsic and intrinsic factors [ 8 4]. The PBPK model consists of drug specific parameters and system specific ones. Typic al drug specific parameters can be estimated by either in vitro and in vivo experiments [ 8 5 8 6] or quantitative relationships [ 8 7] or allometric scaling [ 8 8]. The physiological parameters usually consist of blood flow and volumes of blood/plasma and differ ent tissues with measured values reported by previous literature [ 8 9 9 0]. The PBPK models have been developed for many anticancer drugs such as mitoxantrone [ 60 ], and cisplatin [ 91 ]. However, PBPK model for 5,7 DMF has not been reported before. In our st udy, we have measured 5,7 DMF concentrations in mouse plasma and nine tissues, which provided valuable information to build whole body PBPK model in the near future. Once this PBPK model is successfully developed and validated, it can be then extrapolated to predict the corresponding plasma and tissue concentrations in human species. It will be clinically relevant to use that model for dosing regimen and other adjustment as well. The research on 5,7 DMF as model compound of methylated flavonoids could shed light on future investigation on analogue with similar stability to liver metabolism yet demonstrating promising anti proliferative, and chemopreventive traits. The bioanalytical condition can be applied to establish the quantification methods; and the PK methodology can be helpful to study similar compounds in vivo; the modeling

PAGE 94

94 approaches in the empirical and PBPK construction may provide insight for other scientists researching on phytochemicals or natural flavonoids. Figure 5 1. Exemplary whole body PBPK model for oral dosing

PAGE 95

95 Figure 5 2. Exemplary local PBPK model for oral dosing.

PAGE 96

96 LIST OF REFERENCES [1] Y.Y. Zhao, Traditional uses, phytochemistry, pharmacology, pharmacokinetics and quality control of Polyporus umbellatus (Pers.) Fries: a review, J. Ethnopharmacol. 149 (2013) 35 48. [2] X. Wang, H. Zhang, L. Chen, L. Shan, G. Fan, X. Gao, Liquorice, a unique "guide drug" of traditional Chinese medicine: a review of its role in drug interactions, J. Ethnopharmacol. 150 (2013) 781 90. [3] B.J. Gurley, Pharmacokinetic herb drug interactions (part 1): origins, mechanisms, and the impact of botanical dieta ry supplements, Planta. Med. 78 (2012) 1478 89. [4] B.J. Gurley, E.K. Fifer, Z. Gardner, Pharmacokinetic herb drug interactions (part 2): drug interactions involving popular botanical dietary supplements and their clinical relevance, Planta. Med. 78 (2012) 1490 514. [5] A. Nahrstedt, V. Butterweck, Lessons learned from herbal medicinal products: the example of St. John's Wort (perpendicular), J. Nat. Prod. 73 (2010) 1015 21. [6] E.J. Park, J.M. Pezzuto, Flavonoids in cancer prevention, Anticancer. Agents. M ed. Chem. 12 (2012) 836 51. [7] L.H. Yao, Y.M. Jiang, J. Shi, F.A. Tomas Barberan, N. Datta, R. Singanusong, S.S. Chen. Flavonoids in food and their health benefits, Plant. Foods. Hum. Nutr. 59 (2004) 113 22. [8] Y.J. Surh. Cancer chemoprevention with d ietary phytochemicals, Nat. Rev. Cancer. (2003) 768 80. [9] C. Heiss, C.L. Keen, M. Kelm, Flavanols and cardiovascular disease prevention, Eur. Heart. J. (2010) 2583 92. [10] P.M. Kris Etherton, M. Lefevre, G.R. Beecher, M.D. Gross, C.L. Keen, T.D. Etherto n, Bioactive compounds in nutrition and health research methodologies for establishing biological function: the antioxidant and anti inflammatory effects of flavonoids on atherosclerosis, Annu. Rev. Nutr. 24 (2004) 511 38. [11] M. Rossi, C. Bosetti, E. Neg ri, P. Lagiou, C. La Vecchia, Flavonoids, proanthocyanidins, and cancer risk: a network of case control studies from Italy, Nutr. Cancer. 62 (2010) 871 7. [12] W. Garavello, M. Rossi, J.K. McLaughlin, C. Bosetti, E. Negri, P. Lagiou, Flavonoids and larynge al cancer risk in Italy, Ann. Oncol. 18 (2007) 1104 9. [13] A.M. Ekstro¨m, M. Serafini, O. Nyren, A. Wolk, C. Bosetti, R. Bellocco, Dietary quercetin intake and risk of gastric cancer: results from a population based study in Sweden, Ann. Oncol. 22 (2011) 438 43.

PAGE 97

97 [14] M. Rossi, A. Lugo, P. Lagiou, A. Zucchetto, J. Polesel, D. Serraino, Proanthocyanidins and other flavonoids in relation to pancreatic cancer: a case control study in Italy, Ann. Oncol. 23 (2012) 1488 93. [15] P. Lagiou, M. Rossi, A. Lagiou, A Tzonou, C. La Vecchia, D. Trichopoulos, Flavonoid intake and liver cancer: a case control study in Greece, Cancer. Causes. Control. 19 (2008) 813 8. [16] C.L. Heald, M.R. Ritchie, C. Bolton Smith, M.S. Morton, F.E. Alexander, Phytooestrogens and risk of prostate cancer in Scottish men, Br. J. Nutr. 98 (2007) 388 96. [17] M.D. Jackson, N.D. McFarlane Anderson, G.A. Simon, F.I. Bennett, S.P. Walker, Urinary phytoestrogens and risk of prostate cancer in Jamaican men, Cancer. Causes. Control. (2010) 2249 57. [18] M. Rossi, E. Negri, P. Lagiou, R. Talamini, L. Dal Maso, M. Montella, Flavonoids and ovarian cancer risk: a case control study in Italy, Int. J. Cancer. 123 (2008) 895 8. [19] Y. Cui, H. Morgenstern, S. Greenland, D.P. Tashkin, J.T. Mao, L. Cai, Die tary flavonoid intake and lung cancer a population based case control study, Cancer. 112 (2008) 2241 8. [20] E.V. Bandera, M.G. Williams, C. Sima, S. Bayuga, K. Pulick, H. Wilcox, Phytoestrogen consumption and endometrial cancer risk: a population based ca secontrol study in New Jersey, Cancer. Causes. Control. 20 (2009) 1117 27. [21] Q. Wang, H. Li, P. Tao, Y.P. Wang, P. Yuan, C.X. Yang, Soy isoflavones, CYP1A1, CYP1B1, and COMT polymorphisms, and breast cancer: a case control study in southwestern China, D NA. Cell. Biol. 30 (2011) 585 95. [22] Y.A. Cho, J. Kim, K.S. Park, S.Y. Lim, A. Shin, M.K. Sung, J. Ro, Effect of dietary soy intake on breast cancer risk according to menopause and hormone receptor status, Eur J. Clin. Nutr. 64 (2010) 924 32. [23] M. Iw asaki, G.S. Hamada, I.N. Nishimoto, M.M. Netto, J.J. Motola, F.M. Laginha, Isoflavone, polymorphisms in estrogen receptor genes and breast cancer risk in case control studies in Japanese, Japanese Brazilians and non Japanese Brazilians, Cancer. Sci. 100 (2 009) 927 33. [24] M. Iwasaki, G.S. Hamada, I.N. Nishimoto, M.M. Netto MM, Motola, F.M. Laginha, Dietary isoflavone intake and breast cancer risk in case control studies in Japanese, Japanese Brazilians, and non Japanese Brazilians, Breast. Cancer. Res. Tre at. 116 (2009) 401 11. [25] B.N. Fink, S.E. Steck, M.S. Wolff, J.A. Britton, G.C. Kabat, J.C. Schroeder, Dietary flavonoid intake and breast cancer risk among women on Long Island, Am. J. Epidemiol. 165 (2007) 514 23.

PAGE 98

98 [26] J.W. Lampe, Y. Nishino, R.M. Ray, C. Wu, W. Li, M.G. Lin, Plasma isoflavones and fibrocystic breast conditions and breast cancer among women in Shanghai, China, Cancer. Epidemiol. Biomarkers. Prev. 16 (2007) 2579 86. [27] D. Patanasethanont, J. Nagai, R. Yumoto, T. Murakami, K. Sutthanut, Effects of Kaempferia parviflora extracts and their flavone constituents on P glycoprotein function, J. Pharm. Sci. 96 (2007) 223 33. [28] D. Wu, M.G. Nair, D.L. DeWitt, Novel compounds from Piper methysticum Forst (Kava Kava) roots and their effect on cy clooxygenase enzyme, J. Agric. Food. Chem. 50 (2002) 701 5. [29] H. Haberlein, K.P. Tschiersch, H.L. Schafer, Flavonoids from Leptospermum scoparium with affinity to the benzodiazepine receptor characterized by structure activity relationships and in vivo studies of a plant extract, Pharmazie. 49 (1994) 912 22. [30] C. Sae Wong, H. Matsuda, S. Tewtrakul, P. Tansakul, S. Nakamura, Y. Nomura, M. Yoshikawa, Suppressive effects of methoxyflavonoids isolated from Kaempferia parviflora on inducible nitric oxide s ynthase (iNOS) expression in RAW 264.7 cells, J. Ethnopharmacol. 136 (2011) 488 95. [31] P. Tep Areenan, P. Sawasdee, M. Randall, Possible mechanisms of vasorelaxation for 5,7 dimethoxyflavone from Kaempferia parviflora in the rat aorta, Phytother. Res. 24 (2010) 1520 5. [32] P. Sawasdee, C. Sabphon, D. Sitthiwongwanit, U. Kokpol, Anticholinesterase Activity of 7 Methoxyflavones Isolated from Kaempferia parviflora, Phytother. Res. 23 (2009) 1792 4. [33] X. Wen, U.K. Walle, T. Walle, 5, 7 Dimethoxyflavone do wnregulates CYP1A1 expression and benzo[a]pyrene induced DNA binding in Hep G2 cells, Carcinogenesis. 26 (2005) 803 9. [34] X. Wen, T. Walle, Cytochrome P450 1B1, a novel chemopreventive target for benzo[a]pyrene initiated human esophageal cancer, Cancer. Lett. 246 (2007) 109 14. [35] X. Wen, T. Walle, Preferential induction of CYP1B1 by benzo[a]pyrene in human oral epithelial cells: impact on DNA adduct formation and prevention by polyphenols, Carcinogenesis. 26 (2005) 1774 81. [36] J.F. Yang, J.G. Cao, L. Tian, F. Liu, 5, 7 Dimethoxyflavone sensitizes TRAIL induced apoptosis through DR5 upregulation in hepatocellular carcinoma cells, Cancer. Chemother. Pharmacol. 69 (2012) 195 206. [37] B. Wudtiwai, B. Sripanidkulchai, P. Kongtawelert, R. Banjerdpongchai, Methoxyflavone derivatives modulate the effect of TRAIL induced apoptosis in human leukemic cell lines, J. Hematol. Oncol. 4 (2011) 52.

PAGE 99

99 [38] D. Patanasethanont, J. Nagai, C. Matsuura, K. Fukui, K. Sutthanut, B.O. Sripanidkulchai, R. Yumoto, M. Takano, Mod ulation of function of multidrug resistance associated proteins by Kaempferia parviflora extracts and their components, Eur. J. Pharmacol. 566 (2007) 67 74. [39] G. An, F. Wu, M.E. Morris, 5,7 Dimethoxyflavone and multiple flavonoids in combination alter t he ABCG2 mediated tissue distribution of mitoxantrone in mice, Pharm. Res. 28 (2011) 1090 9. [40] G. An, M.E. Morris, Effects of single and multiple flavonoids on BCRP mediated accumulation, cytotoxicity and transport of mitoxantrone in vitro, Pharm. Res. 27 (2010) 1296 1308. [41] G. Szakcs, J.K. Paterson, J.A. Ludwig, C. Booth Genthe, M.M. Gottesman, Targeting multidrug resistance in cancer, Nat. Rev. Drug. Discov. 5 (2006) 219 34. [42] P.A. Tsuji, T. Walle, Inhibition of benzo[a]pyrene activating enzymes and DNA binding in human bronchial epithelial BEAS 2B cells by methoxylatedflavonoids, Carcinogenesis 27 (2006) 1579 1585. [43] T. Walle, N. Ta, T. Kawamori, X. Wen, P.A. Tsuji, U.K. Walle, Cancer chemo preventive properties of orally bioavailable flavono ids methylated versusunmethylated flavones, Biochem. Pharmacol. 73 (2007) 1288 1296. [44] C. Mekjaruskul, M. Jay, B. Sripanidkulchai, Pharmacokinetics, bioavail ability, tissue distribution, excretion, and metabolite identification ofmethoxyflavones in Kaempferia parviflora extract in rats, Drug Metab. Dispos.40 (2012) 2342 2353. [45] X. Wen, T. Walle, Methylation protects dietary flavonoids from rapid hepatic metabolism, Xenobiotica. 36 (2006):387 97. [46] T. Walle, Absorption and metabolism of flavonoi ds, Free Radicals in Biology and Medicine. 36 (2004):829 837. [47] X. Wen, T. Walle, Methylated flavonoids have greatly improved intestinal absorption and metabolic stability, Drug Metab Dispos. 34 (2006):1786 92. [48] K. Sutthanut, B. Sripanidkulchai, C. Yenjai, M. Jay, Simultaneous identificationand quantitation of 11 flavonoid constituents in Kaempferia parviflora by gaschromatography, J. Chromatogr. A 1143 (2007) 227 233. [49] S. Ponnayyan Sulochana, K. Sharma, R. Mullangi, S.K. Sukumaran, Review of the validated HPLC and LC MS/MS methods for determination of drugs used in clinical practice for Alzheimer's disease, Biomed. Chromatogr. 28 (2014) 1431 90.

PAGE 100

100 [50] K.S. Leung, B.M. Fong, LC MS/MS in the routine clinical laboratory: has its time come, Anal. Bioa nal. Chem. 406 (2014) 2289 301. [51] M.M Kushnir, A.L. Rockwood, J. Bergquist, Mass. Spectrom. Rev. 29 (2010) 480 502. [52] J.M.W. van den Ouweland, I.P. Kema, J. Chromatogr. B. 883 (2012) 18 32. [53] M. Himmelsbach, J. Chromatogr. B. 883 884 (2012) 3 17. [54] FDA Guidance for Industry Bioanalytical Method Validation, 2001,http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/ Guidances/ucm070107.pdf. [55] D. Bei, G. An, Quantification of 5,7 dimethoxyflavone in mouse plasma using liquid chromatography tandem mass spectrometry (LC MS/MS) and its application to a pharmacokinetic study, J. Chromatogr. B. Analyt. Technol. Biomed. Life. Sci. 978 979 (2015) 11 7. [56] FDA Draft Guidance for Industry Bioanalytical Method Validation, 2013, http: //www.fda.gov/downloads/drugs/guidancecomplianceregulatoryinformation/guidanc es/ucm368107.pdf. [57] C.A. Knibbe, K.P. Zuideveld, L.P. Aarts, P.F. Kuks, M. Danhof, Allometric relationships between the pharmacokinetics of propofol in rats, children and adu lts, Br. J. Clin. Pharmacol. 59 (2015):705 11. [58] F. Seker, J. Hesser, M.A. Brockmann, E. Neumaier Probst, C. Groden, R. Schubert, C. Brockmann, Pharmacokinetic Modeling of Intra arterial Nimodipine Therapy for Subarachnoid Hemorrhage Related Cerebral Va sospasm, Clin. Neuroradiol. (2015). [59] V. Sharma, J.H. McNeill, To scale or not to scale: the principles of dose extrapolation, Br. J. Pharmacol. 157(2009):907 21. [60] G. An, M.E. Morris, A physiologically based pharmacokinetic model of mitoxantrone in mice and scale up to humans: a semi mechanistic model incorporating DNA and protein binding, AAPS. 14 (2012):352 64. [61] Q. Huang, J.E. Riviere, The application of allometric scaling principles to predict pharmacokinetic parameters across species, Expert. Opin. Drug. Metab. Toxicol. 10 (2014) 1241 53. [62] H. Tang, M. Mayersohn, Controversies in allometric scaling for predicting human drug clearance: an historical problem and reflections on what works and what does not, Curr. Top. Med. Chem. 11 (2011) 340 50.

PAGE 101

101 [63] B. Clancy, B.L. Finlay, R.B. Darlington, K.J. Anand, Extrapolating brain development from experimental species to humans, Neurotoxicology. 28 (2007):931 7. [64] G.S. Duthu, Interspecies correlation of the pharmacokinetics of erythromycin, oleando mycin, and tylosin, J. Pharm. Sci. 74 (1985):943 6. [65] B. Clark, D.A. Smith, Pharmacokinetics and toxicity testing, Crit. Rev. Toxicol. 12(1984):343 385. [66] J.H. Lin, Applications and limitations of interspecies scaling and in vitro extrapolation in ph armacokinetics, Drug. Metab. Dispos. 26 (1998):1202 1212. [67] E.S. Kostewicz, L. Aarons, M. Bergstrand, M.B. Bolger, A. Galetin, O. Hatley, M. Jamei, R. Lloyd, X. Pepin, A. Rostami Hodjegan, E. Sjgren, C. Tannergren, D.B. Turner, C. Wagner, W. Weitschies J. Dressman, PBPK models for the prediction of in vivo performance of oral dosage forms, Eur. J. Pharm. Sci. 57 (2014) 300 21. [68] M. Rowland, C. Peck, G. Tucker, Physiologically based pharmacokinetics in drug development and regulatory science, Annu. R ev. Pharmacol. Toxicol. 51 (2011) 45 73. [69] T. Teorell, Kinetics of distribution of substances administered to the body. I The extravascular modes of administration, Arch. Int. Pharmacodyn. Ther. 57 (1937) 205 225. [70] T. Teorell, Kinetics of distributi on of substances administered to the body. I The intravascular modes of administration, Arch. Int. Pharmacodyn. Ther. 57 (1937) 226 240. [71] I.S. Edelmann, J. Liebmann, Anatomy of body water and electrolytes, Am. J. Med. 27 (1959) 256 277. [72] H.M. Jones K. Mayawala, P. Poulin, Dose selection based on physiologically based pharmacokinetic (PBPK) approaches, AAPS. J. 15 (2013) 377 87. [73] J.S. Barrett, O. Della Casa Alberighi, B. Ler S, Meibohm, Physiologically based pharmacokinetic (PBPK) modeling in children, Clin. Pharmacol. Ther. 92 (2012) 40 9. [74] G. Lu, K. Abduljalil, M. Jamei, T.N. Johnson, H. Soltani, A. Rostami Hodjegan, Physiologically based pharmacokinetic (PBPK) models for assessing the kinetics of xenobiotics during pregnancy: achievement s and shortcomings, Curr. Drug. Metab. 13 (2012) 695 720. [75] C.M. Thompson, D.O. Johns, B. Sonawane, H.A. Barton, D. Hattis, R. Tardif, K. Krishnan, Database for physiologically based pharmacokinetic (PBPK) modeling: physiological data for healthy and he alth impaired elderly, J. Toxicol. Environ. Health. B. Crit. Rev. 12 (2009) 1 24.

PAGE 102

102 [76] M. Chenel, F. Bouzom, F. Cazade, K. Ogungbenro, L. Aarons, F. Mentr, Drug drug interaction predictions with PBPK models and optimal multiresponse sampling time designs: application to midazolam and a phase I compound. Part 2: clinical trial results, J. Pharmacokinet. Pharmacodyn. 35 (2008) 661 81. [77] J.C. Lipscomb, S. Haddad, T. Poet, K. Krishnan, Physiologically based pharmacokinetic (PBPK) models in toxicity testing and risk assessment, Adv. Exp. Med. Biol. 745 (2012) 76 95. [78] P. Ruiz, M. Ray, J. Fisher, M. Mumtaz, Development of a human Physiologically Based Pharmacokinetic (PBPK) Toolkit for environmental pollutants, Int. J. Mol. Sci. 12 (2011) 7469 80. [79] J.B. Knaak, C.C. Dary, F. Power, C.B. Thompson, J.N. Blancato, Physicochemical and biological data for the development of predictive organophosphorus pesticide QSARs and PBPK/PD models for human risk assessment, Crit. Rev. Toxicol. 34 (2004) 143 207. [80] Haag ard, H. W. The absorption, distribution and elimination of ethyl ether, J. Biol. Chem. 59(1924):753 770. [81] I. Nestorov, Whole body pharmacokinetic models, Clin. Pharmacokinet. 42(2003):883 908. [82] W. Schmitt, S. Willmann, Physiology based pharmacokine tic modeling: ready to be used, Drug. Discov. Today. Technol. 2 (2005):125 132. [83] L. Liu, K. S. Pang, An integrated approach to model hepatic drug clearance, Eur. J. Pharm. Sci. 29(2006):215 230. [84] P. Zhao, L. Zhang, J.A. Grillo, Q. Liu, J.M. Bullock Y.J. Moon, P. Song, S.S. Brar, R. Madabushi, T.C. Wu, B.P. Booth, N.A. Rahman, K.S. Reynolds, E. Gil Berglund, L.J. Lesko, S.M. Huang, Applications of physiologically based pharmacokinetic (PBPK) modeling and simulation during regulatory review, Clin. Ph armacol. Ther. 89 (2011):259 67. [85] J. H. Lin, Y. Sugiyama, In vitro and in vivo evaluation of the tissue to blood partition coefficient for physiological pharmacokinetic models, J. Pharmacokinet. Biopharm. 10 (1982):637 647. [86] Y. Naritomi, S. Terashi ta, Prediction of human hepatic clearance from in vivo animal experiments and in vitro metabolic studies with liver microsomes from animals and humans, Drug. Metab. Dispos. 29 (2001):1316 24. [87] M.O. Fouchecourt, M. Beliveau, K. Krishnan, Quantitative structure pharmacokinetic relationship modelling, Sci. Total. Environ. 274 (2001):125 135.

PAGE 103

103 [88] H. Tang, M. Mayersohn, A novel model for prediction of human drug clearance by allometric scaling, Drug. Metab. Dispos. 33 (2005):1297 1303. [89] L.R. William s, R.W. Leggett, Reference values for resting blood flow to organs of man, Clin. Phys. Physiol. Meas. 10 (1989):187 217. [90] R. Price, R. Conolly, Modeling interindividual variation in physiological factors used in PBPK models of humans, Crit. Rev. Toxico l. 33 (2003):469 503. [91] F.F. Farris, R.L. Dedrick, F.G. King, Cisplatin pharmacokinetics: applications of a physiological model, Toxicol Lett. 43 (1988):117 37.

PAGE 104

104 BIOGRAPHICAL SKETCH Di Bei was born in November 1984, in Chengdu, China. She earned her b p harmacy in Sichuan University in 2007. She further accomplished the m p harmaceutical s ciences in 2009 from University of Missouri Kansas City with the thesis on cubosomes. Before she joined PhD program in University of Florida, she acquired the Statistical Minor in Data Analysis. She accomplished her PhD study from Philosophy degree in pharmaceutical sciences in December 2015.