<%BANNER%>

Drug Overdose Treatment with Liposomes

Permanent Link: http://ufdc.ufl.edu/UFE0041374/00001

Material Information

Title: Drug Overdose Treatment with Liposomes
Physical Description: 1 online resource (233 p.)
Language: english
Creator: Howell, Brett
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: amitriptyline, antidepressants, bupivacaine, colloids, detoxification, dopg, liposomes, overdose, pbpk, phospholipids, tca, toxicity
Chemical Engineering -- Dissertations, Academic -- UF
Genre: Chemical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Many drugs such as tricyclic antidepressants and local anesthetics cause severe toxicity and/or death when taken at excessive and sometimes normal dosage levels. The subject of this work was the development of a drug overdose treatment to counteract toxicity. Liposomes were chosen as the primary vehicle for toxicity reversal. The binding of the tricyclic antidepressants amitriptyline, nortriptyline, imipramine, dosulepin, and opipramol to several variations of liposomes was measured. The medium of measurement was also varied from buffer to serum to assess in vivo affects. The effects of lipid type and loading, liposome size, polyethylene glycol inclusion and chain length, protein interaction, and storage were considered. Pegylated, anionic liposomes exhibited high affinity binding to all of the tricyclic antidepressants studied, despite the presence of serum proteins. Liposome size and polyethylene glycol chain length were inconsequential, while the proportion of polyethylene glycol incorporated into liposomes was optimal at about 5%. Liposome-drug binding was also found to occur to a significant extent for the local amide anesthetic drug bupivacaine. Additionally, interactions between cationic drugs and anionic liposomes were studied by measuring binding of drugs and the effect of binding on liposome permeability. Experiments and modeling indicated that, although electrostatic interactions were important, the fraction of drug sequestered in the double-layer was negligible. The majority of the drug enters the bilayer with the charged regions interacting with the charged lipid head groups and the lipophilic regions associated with the bilayer. Bupivacaine binds significantly less compared to tricyclic antidepressants because its structure is such that the charged region has minimal interactions with the lipid heads once the bupivacaine molecule partitions inside the bilayer. Conversely, the tricyclic antidepressants are linear with distinct hydrophilic and lipophilic regions, allowing the lipophilic regions to lie inside the bilayer and the hydrophilic regions to protrude out. This conformation maximizes the permeability of the bilayer, which leads to an increased release of a hydrophilic fluorescent dye from liposomes. Lastly, physiologically based pharmacokinetic models were developed for the design and optimization of liposome therapy for overdoses. The in vitro drug-binding data for pegylated, anionic liposomes and published mechanistic equations for partition coefficients were used to develop the models. The models were proven reliable through comparisons to intravenous data. Drug overdoses were simulated for various drug and liposome doses, elapsed time between drug intake and liposome treatment, and patient specific input parameters. The liposomes were predicted to be highly effective at treating amitriptyline overdoses. Although liposomes could potentially treat an anesthetic overdose, the drug redistribution was less effective. Published data on local cardiac function was used to relate the predicted concentrations in the body to local pharmacodynamic effects in the heart.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Brett Howell.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Chauhan, Anuj.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041374:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041374/00001

Material Information

Title: Drug Overdose Treatment with Liposomes
Physical Description: 1 online resource (233 p.)
Language: english
Creator: Howell, Brett
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: amitriptyline, antidepressants, bupivacaine, colloids, detoxification, dopg, liposomes, overdose, pbpk, phospholipids, tca, toxicity
Chemical Engineering -- Dissertations, Academic -- UF
Genre: Chemical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Many drugs such as tricyclic antidepressants and local anesthetics cause severe toxicity and/or death when taken at excessive and sometimes normal dosage levels. The subject of this work was the development of a drug overdose treatment to counteract toxicity. Liposomes were chosen as the primary vehicle for toxicity reversal. The binding of the tricyclic antidepressants amitriptyline, nortriptyline, imipramine, dosulepin, and opipramol to several variations of liposomes was measured. The medium of measurement was also varied from buffer to serum to assess in vivo affects. The effects of lipid type and loading, liposome size, polyethylene glycol inclusion and chain length, protein interaction, and storage were considered. Pegylated, anionic liposomes exhibited high affinity binding to all of the tricyclic antidepressants studied, despite the presence of serum proteins. Liposome size and polyethylene glycol chain length were inconsequential, while the proportion of polyethylene glycol incorporated into liposomes was optimal at about 5%. Liposome-drug binding was also found to occur to a significant extent for the local amide anesthetic drug bupivacaine. Additionally, interactions between cationic drugs and anionic liposomes were studied by measuring binding of drugs and the effect of binding on liposome permeability. Experiments and modeling indicated that, although electrostatic interactions were important, the fraction of drug sequestered in the double-layer was negligible. The majority of the drug enters the bilayer with the charged regions interacting with the charged lipid head groups and the lipophilic regions associated with the bilayer. Bupivacaine binds significantly less compared to tricyclic antidepressants because its structure is such that the charged region has minimal interactions with the lipid heads once the bupivacaine molecule partitions inside the bilayer. Conversely, the tricyclic antidepressants are linear with distinct hydrophilic and lipophilic regions, allowing the lipophilic regions to lie inside the bilayer and the hydrophilic regions to protrude out. This conformation maximizes the permeability of the bilayer, which leads to an increased release of a hydrophilic fluorescent dye from liposomes. Lastly, physiologically based pharmacokinetic models were developed for the design and optimization of liposome therapy for overdoses. The in vitro drug-binding data for pegylated, anionic liposomes and published mechanistic equations for partition coefficients were used to develop the models. The models were proven reliable through comparisons to intravenous data. Drug overdoses were simulated for various drug and liposome doses, elapsed time between drug intake and liposome treatment, and patient specific input parameters. The liposomes were predicted to be highly effective at treating amitriptyline overdoses. Although liposomes could potentially treat an anesthetic overdose, the drug redistribution was less effective. Published data on local cardiac function was used to relate the predicted concentrations in the body to local pharmacodynamic effects in the heart.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Brett Howell.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Chauhan, Anuj.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041374:00001


This item has the following downloads:


Full Text

PAGE 1

DRUG OVERDOS E TREATMENT WITH LIPOSOMES By BRETT A. HOWELL A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORID A IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010 1

PAGE 2

2010 Brett A. Howell 2

PAGE 3

To my lovely wife, Michele, and my parents, Dwight and Carla 3

PAGE 4

ACKNOWLEDGMENTS First and foremost, I would like to expr ess my deepest gratitude to my doctoral advisor, Dr. Anuj Chauhan. Throughout my ti me at the University of Florida, Dr. Chauhan has challenged me on every level. He has directed me down the right path countless times, while also ensuring my growth as an independent scientist and engineer. He has been both an excellent advisor and friend, and I am deeply blessed to have spent my time as a doc toral student under his guidance. I also wish to extend many thanks to my other doctoral committee members, Dr. Tanmay Lele, Dr. Yiider Tseng, and Dr. Sihong Song, for their insightful viewpoints and willingness to participate in my doctoral review process. In addition, I must thank Dr. Spyros Svoronos and Dr. Oscar Crisalle fo r the opportunity to serve as a teaching assistant under their leadership. They we re both gracious and kind leaders. Dr. Crisalle was also instrumental in persuadi ng me to choose the Department of Chemical Engineering at the University of Florida, for which I am extremely grateful. As a graduate student in Dr. Chauhans lab, I have had the privilege of working with many fine colleagues. Dr. Jinah Kim was especially helpful during my first days in the lab, as she taught me m any important lab procedures and techniques, as well as the consequence of efficiency. Dr. Yash Kapoor was an excellent m entor and friend, with whom I had the opportunity to do collaborative research on assessing ocular toxicity. Not only was Yash exceptionally talented in the laboratory, but he also never failed to brighten everyones day with his constant smile and positive attitude. Many others were also very helpful lab mates and good friends, including Dr. Chi-Chung Li, Dr. Heng Zhu, Chhavi Gupta, Cheng-Chun Peng, Hyun Jung Jung, and Lokendrakumar Bengani. 4

PAGE 5

Many current and former staff member s of the Department of Chemical Engineering were also very a ccommodating during my time in Gainesville. I would like to thank Shirley Kelly, Deborah Aldrich, Deborah Sandov al, Donna Roberts, Melissa Fox, and Sean Poole for their invaluable assistance. Next, I must thank my wonder ful parents, Dwight and Carla Howell, for a lifetime of support and encouragement. Without their constant uplifting words and trusted direction, I would certainly not have comple ted this work. They are and always will be precious people in my life. My gr andparents, Jim and Carolyn Howell, and Bill and Rachel Gaskey, are also due many thanks for t heir love and support ov er the years. My father-in-law and mother-in-law, David and Linda Popple, have also done much to encourage Michele and me during our stay in Gainesville. Although all the indi viduals listed above have done much to aid in my completion of my doctoral degree, none more so than my loving wife, Michele. She has been extremely understanding, patient, and helpful since the moment we decided to relocate to Gainesville. Words cannot express my af fection and gratitude for her. She is an angel in my eyes, and has been by my side through both the good times and the bad. I am forever indebted to her fo r all her love and support. 5

PAGE 6

TABL E OF CONTENTS page ACKNOWLEDG MENTS..................................................................................................4 LIST OF TABLES..........................................................................................................10 LIST OF FI GURES ........................................................................................................11 LIST OF ABBR EVIATIONS...........................................................................................14 ABSTRACT ...................................................................................................................16 CHAPTER 1 INTRODUC TION....................................................................................................18 2 UPTAKE OF AMITRIPTYLINE AND NO RTRIPTYLINE WITH LIPOSOMES, PROTEINS, A ND SERUM ......................................................................................30 2.1 Introduc tion....................................................................................................30 2.2 Materials and Met hods...................................................................................31 2.2.1 Mate rials..............................................................................................31 2.2.2 Liposome Preparatio n.........................................................................31 2.2.3 Amitriptyline Up take Exper iments........................................................32 2.2.4 Amitriptyline Up take with Li posomes...................................................33 2.2.5 Amitriptyline Up take with Pr oteins .......................................................34 2.2.6 Amitriptyline Uptake with Mixt ures of Liposomes and Proteins............34 2.2.7 Amitriptyline Upta ke with Hum an Seru m.............................................34 2.2.8 Reversibility of Bi nding (Diluti on Met hod)............................................35 2.2.9 Time Dependency of Amitriptylin e Upta ke...........................................35 2.2.10 Nortriptyline Uptake with Liposomes and Albumi n...............................36 2.3 Results an d Discussi on..................................................................................36 2.3.1 Amitriptyline Up take with Li posomes...................................................36 2.3.1.1 Effect of lipid load ing...............................................................37 2.3.1.2 Effect of charge.......................................................................38 2.3.2 Amitriptyline Up take with Pr oteins .......................................................39 2.3.2.1 Uptake with album in...............................................................39 2.3.2.2 Uptake wit h fibri nogen ............................................................43 2.3.2.3 Uptake with globu lins..............................................................44 2.3.3 Amitriptyline Uptake with Mixt ures of Liposomes and Proteins............44 2.3.3.1 Albumin and lipos omes...........................................................45 2.3.3.2 Fibrino gen and lipos omes.......................................................46 2.3.3.3 Globulins and lipos omes.........................................................47 2.3.3.4 Albumin, fibrinogen, globulins, and liposomes........................48 2.3.3.5 Effects of lipos ome-protein binding.........................................49 2.3.4 Uptake of Amitriptyline wit h Liposomes in Human Se rum...................51 6

PAGE 7

2.3.5 System Char acterization.....................................................................52 2.3.5.1 Reversibilit y of bi nding ............................................................52 2.3.5.2 Time dependency of amitriptylin e upta ke...............................53 2.3.6 Nortripty line Upta ke.............................................................................53 2.3.6.1 Nortriptyline uptake wi th 50:50 DMPC:DOPG liposomes.......54 2.3.6.2 Nortriptyline uptak e with DOPG liposom es.............................54 2.3.6.3 Nortriptyline uptake with albumin ............................................54 2.4 Conclu sions....................................................................................................55 3 AMITRIPTYLINE BINDING TO PEGYLATED, ANIONIC LIPOSOMES..................69 3.1 Introduc tion....................................................................................................69 3.2 Materials and Met hods...................................................................................69 3.2.1 Mate rials..............................................................................................69 3.2.2 Liposome Preparatio n via Sonication..................................................70 3.2.3 Liposome Preparatio n via Extr usion....................................................71 3.2.4 Liposome Char acteriza tion..................................................................71 3.2.5 Measurement of Am itriptyline Uptake by Liposomes in Buffer and Human Se rum.....................................................................................71 3.2.6 Storage Tests......................................................................................72 3.2.7 Data Analys is.......................................................................................73 3.3 Results an d Discussi on..................................................................................73 3.3.1 Liposome Char acteriza tion..................................................................73 3.3.2 Increased Lipid Loading for 50:50 DMPC:DOP G Liposomes..............74 3.3.3 Effect of Vesicle Size on Amitriptyline Se questration..........................75 3.3.4 Liposomes Incorporated wit h Polyethylene Gl ycol (PEG )...................76 3.3.4.1 Amitriptyline removal from human serum by various liposome formu lations .............................................................78 3.3.4.2 Assessment of protei n-liposome inte ractions .........................82 3.3.4.3 Increased lipid loadi ng for 95:5 DOPG:DPPE-mPEG-2000 liposom es...............................................................................84 3.3.5 Drug Uptake by Stored Li posomes ......................................................86 3.4 Conclu sions....................................................................................................87 4 BINDING OF IMIPRAMINE, DOSULEPI N, AND OPIPRAMOL TO LIPOSOMES FOR OVERDOSE TREATMENT............................................................................98 4.1 Introduc tion....................................................................................................98 4.2 Materials and Methods .................................................................................100 4.2.1 Materi als............................................................................................100 4.2.2 Liposome Preparatio n via Extrus ion..................................................100 4.2.3 Liposome Char acterizati on................................................................101 4.2.4 Drug Uptake With Liposomes in Human Seru m or PBS....................101 4.2.5 Data A nalysis .....................................................................................102 4.3 Results and Discussion ................................................................................102 4.3.1 Liposome Char acterizati on................................................................102 4.3.2 Imiprami ne Uptake ............................................................................103 7

PAGE 8

4.3.3 Dosulepi n Uptake .............................................................................. 105 4.3.4 Opipramol Uptake and the Importanc e of Electrostatic Interactions..106 4.3.5 In Vivo Considerations: Co mpetition and Pharma cokinetics.............108 4.3.6 Comparisons wit h Prior St udies......................................................... 111 4.4 Conclusions..................................................................................................113 5 BUPIVACAINE BINDING TO PEGYLATED LIPO SOMES...................................121 5.1 Introduc tion..................................................................................................121 5.2 Met hods....................................................................................................... 121 5.2.1 Liposome Pr eparation .......................................................................121 5.2.2 In Vitro Drug Bi nding Measurements in Buffer or Human Serum......122 5.2.3 Statistical Analys is.............................................................................124 5.3 Result s.........................................................................................................124 5.3.1 Bupivacaine Extrac ted from Buffer....................................................124 5.3.2 Bupivacaine Extracte d from Human Serum.......................................124 5.4 Discuss ion....................................................................................................125 5.5 Conclusions..................................................................................................127 6 THE INTERACTION OF CATIONIC DRUGS WITH LI POSOMES.......................131 6.1 Introduc tion..................................................................................................131 6.2 Materials and Methods .................................................................................132 6.2.1 Materi als............................................................................................132 6.2.2 Liposome Preparat ion for Drug Binding and Zeta Potential Measurement s...................................................................................132 6.2.3 Liposome Preparation fo r Calcein Leakage Studies ..........................133 6.2.4 Preparation of Poly(methacry lic acid) and Poly(acrylic acid) Microparticl es....................................................................................134 6.2.5 Liposome Char acterizati on................................................................134 6.2.6 Drug Binding to Liposom es in Buffer Solutions ..................................135 6.2.7 Zeta Potentia l Measurem ents............................................................136 6.2.8 Liposome Leak age Studies ...............................................................136 6.2.9 Statistical Analys is.............................................................................137 6.3 Results and Discussion ................................................................................137 6.3.1 Electrostatic Contribut ion to Drug Bi nding.........................................138 6.3.1.1 Effect of ionic strength..........................................................138 6.3.1.2 Drug binding to neutral lipos omes........................................139 6.3.1.3 Drug binding to ani onic micropart icles..................................140 6.3.2 Lipophilic Contribut ion to Drug Binding ..............................................141 6.3.2.1 Drug binding wit h DMPG liposomes: the effect of bilayer fluidity ...................................................................................141 6.3.2.2 Drug binding with multilamellar lip osomes (MLL).................142 6.3.2.3 Liposome leakage induced by drugs....................................143 6.3.3 Mechanism Validation th rough Continuum Modeling.........................144 6.3.3.1 Model development ..............................................................144 6.3.3.2 Model fits to data for antidepressa nt drugs...........................147 8

PAGE 9

6.3.3.3 Model fits to dat a for bupiva caine.........................................148 6.3.3.4 Effect of surface char ge on isotherm par ameters.................149 6.3.3.5 Model validation via salt effects............................................151 6.3.4 Summary of the Proposed Mechanism of Drug Sequestration..........153 6.4 Conclusions..................................................................................................154 7 PREDICTING THE EFFICACY OF AMITRIPTYLINE AND BUPIVACAINE OVERDOSE TREATMENT WITH LIPOSOMES IN MAN WITH PHYSIOLOGICALLY BASED PHAR MACOKINETIC MODELS...........................163 7.1 Introduc tion..................................................................................................163 7.2 Met hods....................................................................................................... 164 7.2.1 Obtaining Tissue Part ition Coeffi cients..............................................164 7.2.2 PBPK Model Structur e.......................................................................167 7.2.3 Model Input Paramete rs....................................................................169 7.2.4 Validation of Model Parameters by Predicting Intravenous Data.......169 7.2.5 Determining Absorption Rate Constants by Fitting to nonIntravenous Data...............................................................................170 7.2.6 Drug Overdose Treat ment Simula tions.............................................172 7.2.7 Liposome and Dr ug Clearanc e..........................................................172 7.2.8 Metabo lites........................................................................................174 7.2.9 Drug Dose in Overdose Simulati ons..................................................175 7.2.10 Validation of the Numerical Ca lculati ons...........................................176 7.2.11 Sensitivit y Analys is............................................................................177 7.3 Result s.........................................................................................................178 7.3.1 Tricyclic Antidepressant Model Validation and Fits for Absorption Coefficient s........................................................................................178 7.3.2 Bupivacaine Model Va lidation and Fits for Absorption Coefficients...178 7.3.3 Tricyclic Antidepressant Overdose Simu lations.................................179 7.3.4 Bupivacaine Over dose Simulati ons...................................................180 7.3.5 Sensitivity Analysis Results...............................................................181 7.4 Discuss ion....................................................................................................181 7.4.1 Model Va lidation ................................................................................181 7.4.2 Evidence for Drug Redist ribution with Li posomes.............................181 7.4.3 Pharmacodynamics...........................................................................183 7.4.4 Sensitivit y Analys is............................................................................187 7.4.5 Liposome Dose Optimizati on.............................................................188 7.4.6 Time Lapse Between Drug Administ ration and Liposome Treatment189 7.4.7 Limita tions.........................................................................................189 7.5 Conclusions..................................................................................................190 8 CONCLUSION S...................................................................................................210 LIST OF REFE RENCES.............................................................................................216 BIOGRAPHICAL SKETCH ..........................................................................................233 9

PAGE 10

LIST OF TABLES Table page 4-1 Association constant comparison for liposomes and r ed blood ce lls................119 4-2 Time between ingestion and treatment for antidepressant overdoses..............119 4-3 Pharmacokinetic properties of various organs and imipramine partition coefficient s........................................................................................................119 4-4 Comparison of drug deto xification v ehicles ......................................................120 6-1 Values for constants and parameters used for drug binding predictions..........162 7-1 Drug specific parameters for Kpu calculations..................................................200 7-2 Calculated drug to tissue partiti on coefficient (K pu) values..............................201 7-3 Partition coefficients for drug bindi ng to proteins, liposomes, and red blood cells ..................................................................................................................201 7-4 Organ blood flow and volume fractions fo r humans..........................................202 7-5 Input parameters for PBPK model.................................................................... 203 7-6 Drug specific inputs fo r overdose simu lations ...................................................203 7-7 Sensitivity anal ysis setup..................................................................................204 7-8 AMI overdose simu lation resu lts.......................................................................205 7-9 BUP overdose simu lation resu lts......................................................................207 7-10 AMI sensitivity analysis results.........................................................................208 7-11 BUP sensitivity anal ysis resu lts........................................................................209 10

PAGE 11

LIST OF FIGURES Figure page 2-1 Structures of drugs and lipids used for drug uptake studies with liposomes.......57 2-2 Measured and predicted percent AMI uptake for pure DOPG liposomes and 50:50 DMPC:DOPG liposom es in buffer............................................................58 2-3 Measured and modeled percent AMI uptake for 2% and 4% albumin in buffer..59 2-4 Measured and modeled percent AM I uptake for 2% fibrinogen and 1% globulins in buffer; predicted percent AMI uptake for 1% fibrinogen and 2% globulins in buffer...............................................................................................60 2-5 Measured and predicted percent AMI upt ake for mixtures of 4% albumin and 50:50 DMPC:DOPG or DOPG liposomes in buffer.............................................61 2-6 Measured and predicted percent AMI upt ake for mixtures of 2% fibrinogen and 50:50 DMPC:DOPG lipos omes in buffer......................................................62 2-7 Measured and predicted percent AMI upt ake for mixtures of 1% globulins and 50:50 DMPC:DOPG lipos omes in buffer......................................................63 2-8 Measured and predicted uptake of AMI with 7% proteins and 50:50 DMPC:DOPG liposomes; measured uptake of AMI with 7% proteins................64 2-9 Measured uptake of AMI in human serum with and without 50:50 DMPC:DOPG li posomes....................................................................................65 2-10 Percent AMI uptake for mixtures of 4% albumin and 50:50 DMPC:DOPG liposomes in buffer where standard and d ilution test methods were used..........65 2-11 Measured and predicted percent AMI upt ake for mixtures of 2% fibrinogen and 50:50 DMPC:DOPG liposomes in buffe r filtered shortly after mixing, 24 hours later, and 48 hours late r............................................................................66 2-12 Percent AMI and nortriptyline uptake for 50:50 DMPC:DOPG liposomes in buffer..................................................................................................................67 2-13 Percent AMI and nortriptyline uptake fo r pure DOPG liposomes in buffer..........68 3-1 Size distributi ons of lip osomes ...........................................................................90 3-2 Percent AMI uptake from human serum by 50:50 DMPC:DOPG liposomes; prediction for 50:50 liposomes in human serum.................................................91 3-3 Percent AMI uptake from buffer by 50:50 DMPC:DOPG liposomes of various sizes ...................................................................................................................91 11

PAGE 12

3-4 Percent AMI uptake from human se rum samples without liposomes and with various lipos ome ty pes.......................................................................................92 3-5 Percent AMI uptake from human serum by 95:5 DOPG:DPPE-mPEG-2000 liposomes and 50:50 DMPC:DOPG lipos omes..................................................92 3-6 Percent AMI uptake from human serum by 95:5 and 85:15 DOPG:DPPEmPEG-2000 li posomes.......................................................................................93 3-7 Percent AMI uptake from buffe r by 95:5 and 85:15 DOPG:DPPE-mPEG2000 liposom es..................................................................................................93 3-8 Percent AMI uptake by human seru m and by mixtures of human serum and 95:5 DOPG:DPPE-mPEG -2000 lipos omes........................................................94 3-9 Percent AMI uptake from human serum by 95:5 DOPG:DPPE-mPEG-2000 liposomes and pure DO PG lipos omes................................................................94 3-10 Percent AMI uptake from human serum and buffer by 95:5 DOPG:DPPEmPEG-2000 li posomes.......................................................................................95 3-11 Percent AMI uptake from human seru m and buffer by pure DOPG liposomes...95 3-12 Percent AMI uptake by human seru m and by mixtures of human serum and 95:5 DOPG:DPPE-mPEG -2000 lipos omes........................................................96 3-13 Free drug concentration reduction of AMI in human serum by 95:5 DOPG:DPPE-mPEG-2000 liposomes at various lipid loadi ngs..........................96 3-14 Percent AMI uptake from human seru m by pure DOPG liposomes stored for one mont h..........................................................................................................97 3-15 Percent AMI uptake from human serum by 95:5 DOPG:DPPE-mPEG-2000 liposomes stored for one m onth.........................................................................97 4-1 Percent IMI uptake by human seru m and by mixtures of human serum and 95:5 DOPG:DPPE-mPEG-2000 li posomes liposomes.....................................115 4-2 Percent DOS uptake by human seru m and by mixtures of human serum and 95:5 DOPG:DPPE-mPEG2000 liposom es......................................................116 4-3 Percent opipramol uptake by human se rum and by mixtures of human serum and 95:5 DOPG:DPPE-mPEG -2000 liposom es...............................................117 4-4 Free drug concentration reductions fo r AMI, IMI, DOS, and opipramol in human serum by 95:5 DOPG:D PPE-mPEG-2000 li posomes...........................118 5-1 Percent of total BUP bound to 95:5 DOPG:DPPE-mPEG-2000 liposomes......128 12

PAGE 13

5-2 Percent of total BUP bound in hu man serum and in a mixture of human serum and unilamella r liposom es.....................................................................129 5-3 Free BUP versus total BUP in hu man serum samples in the absence and presence of 95:5 DOPG:DPPE-mPEG-2000 lipos omes...................................130 6-1 Structures of the drugs and lipids used for studying drug-liposome interacti ons.......................................................................................................156 6-2 Percent of IMI, AMI, or BUP bound to various liposome types at various ionic strengths ...........................................................................................................157 6-3 Percent of entrapped calcein re leased by 95:5 DOPG:DPPE-mPEG-2000 liposomes after exposure to AMI, IMI, or BUP..................................................158 6-4 Concentration of AMI, IMI, DOS, or BUP bound to anionic, pegylated liposomes with model pr edictions s hown..........................................................159 6-5 Concentration of IMI or BUP bound to DMPC or DOPG liposomes at various ionic strengths with model predictions shown ...................................................160 6-6 Plausible mechanism for TCA and BUP binding to pegylated, anionic liposomes .........................................................................................................161 7-1 Organ to blood partition coeffi cients for AMI and BUP .....................................192 7-2 PBPK model structure used for AM I and BUP overdose si mulations...............193 7-3 IMI, AMI, or BUP concentrations ve rsus time predicted using PBPK models or measured afte r IV dos age............................................................................194 7-4 IMI, AMI, or BUP concentrations versus time for PBPK models where ka was allowed to vary to obtain best fits to measured non-IV dosage dat a.................195 7-5 Summary of AMI overdose simulation results...................................................196 7-6 Simulated AMI concentrations versus time for various lag times and organs...197 7-7 Simulated BUP concentrations versus time for various lipid loadings and organs.............................................................................................................. 198 7-8 Pharmacodynamics in the hear t as a function of time......................................199 13

PAGE 14

LIST OF ABBREVIATIONS AA Acrylic Acid AMI Amitriptyline APL Acidic Phospholipids AUC Area Under the Drug Concentration Versus Time Curve B:P Blood to Plasma Drug Concentration Ratio BSA Bovine Serum Albumin BUP Bupivacaine CH Cholesterol CI Confidence Intervals DI Deionized DMPC 1,2-dimyristoyl-sn-glycero-3-phosphocholine DOPG 1,2-dioleoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] DOS Dosulepin DPPE 1,2-dipalmitoyl sn -glycero-3-phosphoethanolamine EGDMA Ethylene Glycol Dimethacrylate F Bioavailability H Hematocrit IMI Imipramine IV Intravenous MAA Methacrylic Acid MLL Multilamellar Liposomes PBPK Physiologically Based Pharmacokinetic PBS Phosphate Buffered Saline PD Pharmacodynamic 14

PAGE 15

PEG Polyethylene Glycol PK Pharmacokinetic RBC Red Blood Cell RES Reticuloendothelial System SUV Small Unilamellar Vesicles TCA Tricyclic Antidepressant ULL Unilamellar Liposomes UV Ultraviolet 15

PAGE 16

Abstract of Dissertation Pr esented to the Graduate School of the University of Fl orida in Partial Fulf illment of the Requirements for t he Degree of Doctor of Philosophy DRUG OVERDOSE TREATMENT WITH LIPOSOMES By Brett A. Howell May 2010 Chair: Anuj Chauhan Major: Chemical Engineering Many drugs such as tricyclic antidepre ssants and local anesthetics cause severe toxicity and/or death when taken at excessive and sometimes normal dosage levels. The subject of this work was the devel opment of a drug overdose treatment to counteract toxicity. Liposomes were chosen as the primary vehicle fo r toxicity reversal. The binding of the tricyclic antidepressants amitriptyline, nortriptyline, imipramine, dosulepin, and opipramol to several variat ions of liposomes was measured. The medium of measurement was also varied from buffer to serum to assess in vivo affects. The effects of lipid type and loading, li posome size, polyethylene glycol inclusion and chain length, protein intera ction, and storage were consid ered. Pegylated, anionic liposomes exhibited high affinity binding to all of the tricyclic antidepressants studied, despite the presence of serum proteins. Liposome size and polyethylene glycol chain length were inconsequential, while the proportion of polyeth ylene glycol incorporated into liposomes was optimal at about 5%. Liposome-drug binding was also found to occur to a significant ex tent for the local amide anesthetic drug bupivacaine. Additionally, interactions between ca tionic drugs and anionic liposomes were studied by measuring binding of drugs and the effect of binding on liposome 16

PAGE 17

17 permeability. Experiments and modeling indicated that, although electrostatic interactions were important, the fraction of drug sequestered in the double-layer was negligible. The majority of the drug ent ers the bilayer with the charged regions interacting with the charged lipid head group s and the lipophilic regions associated with the bilayer. Bupivacaine binds significant ly less compared to tricyclic antidepressants because its structure is such that the charged region has mi nimal interactions with the lipid heads once the bupivacaine mo lecule partitions inside the bilayer. Conversely, the tricyclic antidepressants are linear with distinct hydrophilic and lipophilic regions, allowing the lipophilic regions to lie inside the bilayer and the hydrophilic regions to protrude out. This conformati on maximizes the permeability of the bilayer, which leads to an increased release of a hydrophilic fluorescent dye from liposomes. Lastly, physiologically based pharmacoki netic models were developed for the design and optimization of liposome therapy fo r overdoses. The in vitro drug-binding data for pegylated, anionic liposomes and published mechanistic equations for partition coefficients were used to develop the models. The models were proven reliable through comparisons to intravenous data. Drug over doses were simulated for various drug and liposome doses, elapsed time between drug intake and liposome treatment, and patient specific input parameters. The liposomes were predicted to be highly effective at treating amitriptyline overdos es. Although liposomes c ould potentially treat an anesthetic overdose, the drug r edistribution was less effective. Published data on local cardiac function was used to relate the predicted concentrations in the body to local pharmacodynamic effects in the heart.

PAGE 18

CHA PTER 1 INTRODUCTION Over the past century, the development of new drugs and therapies has advanced at a rapid pace, providing those who st ruggle with sickness and dis ease exciting new treatment options. In many cases, however the same drugs designed to improve the quality and/or length of life fo r individuals are accidentally or purposefully abused. Thus, prescription drug poisonings now represent a significant public health problem in the United States and around the world. A publication by Diane Wysowski from the Food and Drug Administration pointed to ov er 25,000 deaths in 2003 in the United States due to toxicity from prescription drugs [1]. Furthermore, the number of poisonings increased by 55% from 1999 to 2003, and Wysowski concluded, deaths due to overdoses are the most prominent c ause of drug related mortality in death certificate data, and that preventive strategies should be considered. Tricyclic antidepressant (TCA) drugs are frequently the cause of many of those poisonings. TCA poisoning is a leading cause of self-poisoning in the world [2], causing extensive hospital stays [3,4] and many deat hs [5,6]. In fact deaths from such overdoses represent the third most reported to poison cont rol centers in the United States [7], and approximately 268 people die from TCA overdos es every year in Britain [8]. The primary and typically lethal effect of TCA drugs is the impact on the cardiac system [9,10]. Tachycardia, vasodilation, myocardial depression, and cardiac conduction disturbances are some of the seri ous cardiovascular effects caused by TCA overdose [11]. As opposed to selective serotonin re-uptake inhibitors (SSRI), which are antidepressant medications developed in more re cent years with milder side effects and higher toxic dosage levels, TCAs are often toxic at low dosages. This is especially true 18

PAGE 19

when taken by young children, where one or two pills can cause acute toxicity [2]. Furthermore, overdose cases involving TCAs can be fatal, and have been statistically shown to result in longer hospital stays than overdose cases with other antidepressant drugs [3]. Still, a larg e number of indivi duals continue to use TCAs to treat common depression disorders. Mor gan et al. have recently repor ted a rise in overall TCA prescriptions in England from 1993 to 2002 [6]. Local anesthetics are less likely to be taken at elevated levels than TCAs, since they are administered at medical faciliti es, but adverse reactions during routine treatment remain a threat [12-15]. They present dangers to both t he cardiac [16,17] and central nervous systems (CNS) [17]. Although most patients recover from such reactions as a result of being in medical facilities at the time of local anesthetic administration, many face extended recovery periods and long term injuries, while a select few do not survive. This is especiall y tragic since many patients arrive for routine surgeries requiring local anest hetics in fairly good health. Traditional methods of treating drug or substance toxicity include specific antidotes, such as anti-venoms developed to counteract snake bites, gastric lavage, charcoal administration fo r toxin absorption, and pH manipulation through sodium bicarbonate administration. These and other similar methods of tr eatment are marked by key deficiencies. Specific antidotes ar e difficult to develop, produce, and store for every possible toxin. In addition, they oft en lack the ability to circulate in the blood compartment for adequate periods of time due to the immune system responses they invoke. Charcoal administration is aimed at removing the toxin from the victims stomach prior to absorption, but is only effe ctive if given shortly after overdose or for 19

PAGE 20

toxins slowly absorbed. Other methods of tr eatment typi cally focus on restoring patient well-being while the toxin is naturally re moved from the body. The major disadvantage to this approach is the extended period of ti me required for the clearance of some drugs or substances, resulting in days or even wee ks of often painful and expensive recovery. One possible method of drug overdose treatm ent is the use of small nano or micro carriers designed to sequester excess drug from the body. This idea was first proposed in 1973, but has recently garnered more at tention due to burgeoning research in the general area of nanotechnology [18]. Nanoparti cles can induce detoxification through a variety of mechanisms. Toxins may be redistri buted from the site of toxicity, which often includes the heart and/or brain, into the blood compartment as a result of specific or non-specific drug-particle binding. The nanoparticles may also carry some antidote or deactivating enzyme, combining traditional treatment ideas of the past with improved delivery methods. Finally, the particles ma y be composed of a material or substance that acts directly on the injured area to rapidly improve the effected organs functions, rather than interacting with the toxin or drug compound directly. Several studies have been done in recent years involving drug detoxification with nano or microparticles. Varshney et al. used eight pluronic surfactants in combination with ethyl butyrate to find the best surfactant for drug binding to mi croemulsions [19]. Pluronic F-127 was deemed the most promising surf actant, and an optimized microemulsion formulation was capable of ex tracting 60% of the local anesthetic bupivacaine (BUP) from a buffered solution at an oil content of 1 mg oil/mL across a wide range of drug concentrati ons. In an effort to understand the binding process between the TCA amitriptyline (AMI) and similar emulsions, James-Smith et al. used 20

PAGE 21

turbidity analysis to determine that approx im ately 12 AMI molecules were bound per F127 molecule [20]. The importance of the oil phase in their systems was also demonstrated based on its ability to induce a surface charge and attract AMI. Underhill et al. studied oil-filled nanocapsules stabiliz ed by Brij 97 surf actant and thereafter coated with polysilicate/polysiloxane shells [21]. Their presumed mechanism for drug affinity was hydrophobic interactions between lipophilic drugs and oil phases. They were able to remove greater than 99% of BUP from drug solutions at concentrations below 200 M. Quinoline was also studied fo r proof of concept purposes, and 97% was removed from solutions after 15 minutes with 1.4% w/v oil content nanocapsules. Lee and Baney studied yet another type of detoxific ation system where a different attraction mechanism was proposed [22]. They r eacted chitosan polymers with dinitrofluorobenzene to form dinitrophenyl chitosans. The resulting chitosans were only partially soluble in water, attributed to the reduced protonation of amino groups [22]. The insoluble chitosans were used for dr ug uptake studies and removed 90% of AMI from saline solutions at a relatively high polymer concentration of 0.4%. Chakraborty and Somasundaran used poly(acrylic acid) microparticles to sequester AMI and BUP from normal saline [23]. Perhaps the most relevant work to the research presented herein was done by Dhanikula et al. They r eported results from experimen ts using nanoparticles for the sequestration of AMI and other drugs [24,25]. In one study, they use oil-filled lipid nanocapsules for haloperidol, docetaxel, and paclitaxel sequestration [24]. They hypothesized that the oil-filled nanocapsules bind to drug based on oil-drug affinities. In a second study, they used spherulites and nanocapsules to do in vitro experiments 21

PAGE 22

involv ing AMI, as well as ex vivo experiments using amitri ptyline-intoxicated rat hearts [25]. Spherulites are similar to nanocapsul es and liposomes, with one key difference. They have numerous concentric bilayers su rrounding their core, ra ther than a single layer. The work of these and other res earchers on drug binding to nanoparticles is further discussed throughout Chapters 2-7. Despite t he numerous studies conducted on developing nanoparticles for drug detoxificati on, no effective, widely useful solution has been introduced into the market. Accordingly, the overall goal of the pr esent work was to design nanoparticles capable of treating TCA and BUP overdoses. Liposomes were chosen as the platform for development. Liposomes are spherical nanoparticles composed of a bilayer of phospholipids encapsulating an aqueous core. Their unique structures make it possible for them to carry water so luble compounds, which has led to applications in drug delivery [26]. Unlike micelles, liposomes ra rely form spontaneously and typically require some form of energy for comple te formation [26]. This is especially true if unilamellar, uniformly sized liposomes are desired. Co mmon methods of making liposomes include the reverse phase evaporation (RES) method [27], probe or bath sonication [27], and hydration followed by extrusi on [28]. One major advantage of using liposomes is their proven biocompatibility for m any phospholipids used. To accomplish the goal of designing liposomes for the treatment of TCA or BUP overdose, liposomes were optimized (Chapters 2-5), t he binding process was characterized (Chapter 6), and physiologically based pharmacokinetic (PBPK) models were used to predict in vivo efficacy based on the in vitro data (Chapter 7). 22

PAGE 23

AMI is a pri mary TCA of concern. In C hapter 2, we tested t he AMI binding ability of charged liposomes in the presence of seru m proteins. Serum proteins are present throughout the blood stream and c ould therefore interact wit h liposomes during the drug sequestration process, renderi ng the liposomes less effectiv e. The primary goal of Chapter 2 was to quantify the protein e ffects on liposome treatment, both from a competition and interference stand point. Liposome charge and loading were varied, drug binding was measured to serum prot eins individually and via human serum samples, and the primary metabo lite of AMI, nortrip tyline, was also tested. A two-site binding model was proposed and fit to the drug-protein binding data. The low drug concentration regimes in which most of the data was measured allowed protein-drug binding to be uniquely characterized. Based on the results from Chapter 2, wher e proteins were shown to interact with charged liposomes, lipids altered to incl ude covalently attached polyethylene glycol (PEG) were added to the liposome formulations. Coverage with surface polymers such as PEG is the most widely known method of counteracting opsonization and protein binding. The resulting liposomes have a surface layer of PEG th at is approximately 5 nm thick [29]. Using this te chnique, Awasthi et al. have shown significant increases in circulation times for pegylated versus conv entional liposomes, with less accumulation in the liver as well [30]. This phenom enon has also been observed for nanocapsules fabricated with and without PEG [31]. The degree of su rface coverage, achieved through polymer chain length and percent inclusion, is al so noteworthy, as liver and spleen accumulation are minimal at PEG inclusions of around 10% by mole, but increase beyond such concentrations [31,32 ]. By choosing an optimum proportion of 23

PAGE 24

PEG, the correct chain length, and the bes t preparation technique, one can greatly enhance the pharmacokinetic properties of most nanoparticles. AMI binding to the PEG-altered liposomes was report ed in Chapter 3. Additiona lly, the effect of increased lipid loading and various liposom e sizes on AMI binding was investigated. To simulate the packaging effects that liposomes would u ndergo during clinical si tuations, liposomes stored for a period of one month were tested for their drug binding ability. In Chapters 2 and 3, liposomes were optimized for maximum AMI uptake. A system capable of treating multiple overdoses is significantly superior to a therapy targeted at only one drug. AMI is one of three key antidepressants related to overdose fatalities, along with imipramine (IMI) and dosul epin (DOS), which is also known as dothiepin. Henry et al. [5] and Morgan et al. [6] reported th at these three drugs produce the highest death rates among all antidepress ants used in the United Kingdom. The first goal of Chapter 4 was to therefore extend the use of li posomes to IMI and DOS. In addition, opipramol is a drug simi lar in structure to tricyclic antidepressants but used for the treatment of anxiety diso rders. It has also been cit ed as heavily involved in drug overdose cases in Germany and Turkey, where it is primarily used [3]. Including opipramol into the study allowed us to test liposome-drug binding fo r an important drug, and the effect of drug char ge on binding, as opipramol is only 83% protonated compared to 99% for the TCAs. Chapter 4 also included experiments aimed at quantifying the effect of c hanging the PEG chain length, if any, on the drug-liposome interactions. PEG length is an important parameter for liposome in vivo circulation time. Local anesthetic drugs are useful for pr oviding regional anesthesia or analgesia with limited systemic exposure. However, t heir ability to cause toxicity has been pointed 24

PAGE 25

out above. Such instances can occur wit hout warning and require rapid action to ensure patient survival [17]. A rapid treatment protocol for tox icity reversal would drastically reduce the risks of using such drugs. Several in vivo animal models and emerging case studies show that lipid emulsions, such as Intralipid 20%, composed of 20% soybean oil, 1.2% egg yolk phospholipid s, 2.25% glycerin, and water, through rapid drug binding or improv ed oxidative metabolism, may be one useful modality for treating overdoses [12-14,17,33-40]. Such emulsions have traditionally been used to supply patients with calories and essential fatty acids during illness and other times when consuming adequate amounts of fat is difficult. Weinber g et al. first studied the pre-treatment effects of lipid emulsions on rat toxicity induced with BUP [41]. They showed an in vitro sequestration in plasma of 75.3% of BUP at fina l lipid concentrations of 15%. More significantly, they showed me dian lethal doses of 82 mg/kg for rats treated with lipid emulsions prior to drug ex posure, which was about 5 fold higher than the dose of 17.8 mg/kg in the control, unt reated group. Their study was followed by additional work by Weinberg et al., wher e dogs were treated with lipid emulsions 10 minutes after drug induced toxicity [42]. Six treated dogs survived, while the six untreated dogs did not. Furthermo re, they reported recovery of myocardial tissue pH and oxygen pressure for lipid em ulsion treatment versus the control. Since the results of their studies were published, several cases of successful detox ification with similar lipid emulsions have been reported [12-14,33,34]. Thus, lipid emulsions have been shown to be an effective detoxification therapy under critical conditions in clinical settings. 25

PAGE 26

Still, questions remain about the use of lipid emulsions for det oxific ation. First, the mechanism of toxicity reversal is still unc lear, as Weinberg et al. and many others suggest [41-46]. Possible mechanisms in clude the redistribution mechanism, where drug is relocated to the blood compartment, as well as restored energy production in the myocardium. BUP is known to hinder fa tty acid transport at the mitochondrial membrane, causing disruption to energy demanding processes. The possibility of lipid emulsions improving fatty acid oxidation and subsequent improved adenosine triphosphate production has been widely sugges ted and even supported through clinical results [14]. Yet another mechanism has al so been mentioned, in which increased nitric oxide production induced by lipid emulsions c ould reduce bupivacaine toxicity [41]. The mode of toxicity reversal is not t he only issue surrounding the use of lipid macroemulsions for detoxification therapy. Concerns over the abandonment of fervent toxicity prevention efforts and the possible side effects of administering such large amounts of lipids to patients have also been raised [45,46]. Chapter 5 represents our attempts to both improve the current treatment protocol for local anesthetic adverse reactions thr ough superior liposome therapies, and better understand the mode of action of emulsions currently being used. Liposomes have already been explored as drug delivery vehi cles for local anesthetics, which suggest high-affinity binding between liposomes and lo cal anesthetics [47-53]. The PEG coated, anionic liposomes explored in Chapters 2-4 were used to meas ure BUP binding in Chapter 5. Multilamellar liposomes were also utilized in drug binding experiments, which led to some very interesting obser vations about multilamellar versus unilamellar liposomes in the context of charged liposomedrug interactions. Data reported for BUP 26

PAGE 27

binding to macroemulsions in literature was compared to in vitro data for liposomes to determine which formulation had the higher drug affinity. In Chapter 6, we move from the optimi zation phase to a more fundamental look at liposome-drug interactions. The association of small molecules with lipid and/or cell membranes has been a subject of several studies, primarily due to the pivotal role such interactions play in determining how drugs nutrients, and other xenobiotics introduced into the body affect the function and vitalit y of the cell membrane [54-61]. These interactions have also been explored due to t heir relevance in applications, such as drug loading and release from liposomes [6268] and, within the context of this work, drug overdose treatment with lip osomes [24,25,69-72]. The interactions of molecules with liposomes and lipid bilayers in general impact drug delivery and drug overdose treatment through several mechanisms. Fi rst, the equilibrium binding of drugs to bilayers controls the release rate when the re lease is limited by equilibrium partitioning, which is true for drugs that exhibit a high bi nding affinity for liposomes. Second, the bilayer permeability of water-soluble drugs loaded into liposomes has a profound effect on their rate of release unless liposome des truction is triggered by some alternative mechanism. Finally, the interaction of drugs with lipid bilayers of the endothelial cells lining the capillaries can impact the transport of drugs and other solutes into tissues. Chapter 6 involves additional liposome-drug binding studies wher e various parameters such as surface charge, bilayer fluidity, number of lamellae, and ionic strength were altered. Moreover, leakage of a water so luble fluorescent dye from the core of liposomes was studied to give some insights into differences in drug-bilayer interactions 27

PAGE 28

for TCAs and BUP. A continuum model in corporating electrostatics and a Langmuir binding isot herm was used to help anal yze the experimental results. In Chapters 2-6, we identified anionic, pegylated liposomes capable of sequestering TCAs and BUP from human se rum solutions with the goal of using liposome therapy to reduce toxicity through dr ug redistribution from vital organs to the blood compartment. While our in vitro drug binding data suggested great potential for liposome based overdose therapy, the in vivo efficacy could not be directly assessed. As a next step in our efforts to prove the effectiveness of the liposomal systems, we attempted to utilize physiologically based pharmacokinetic (PBPK) models and our in vitro data to predict the in vivo behavior. Predictions of in vivo behavior based on in vitro data repres ent an exciting and growing research area that can lead to reductions in time and costs associated with developing new drug therapies, as well as t he amount of animal testing required. PBPK models are exceptional tools for this purpose [73-76], offe ring a versatile method for predicting the pharmacokinetic profiles, optimal doses, and pharmacodynamic (PD) effects of new therapies early in the drug discovery process. PBPK models are essentially block compartment models mo re closely representing physiology than traditional pharmacokinetic models [77], and have been utilized for a wide array of purposes [78-82]. We combined partiti on coefficients calculated from published mechanistic equations with our in vitro data fo r protein and liposome binding to estimate drug distribution throughout the human body. T he models were validated for reliability using published intravenous data. The models allowed for variations in drug dose, liposome dose, time lapse between dr ug and liposome dose, hepatic clearance, 28

PAGE 29

29 absorption times, partition coefficients, and liposome clearance to be simulated without any additional in vivo experi ments. Lastly, reported local pharmacodynamic changes in cardiac tissue and cell components as a func tion of AMI or BUP concentration were correlated to concentration versus time profil es generated by our models to allow for an estimate of the extent to which liposom es would ultimately induce cardiac tissue recovery to basal levels.

PAGE 30

CHA PTER 2 UPTAKE OF AMITRIPTYLINE AND NO RT RIPTYLINE WITH LIPOSOMES, PROTEINS, AND SERUM 2.1 Introduction The goal of this chapter was to test t he suitability of charged liposomal systems at treating amitriptyline (AMI) overdose under physiological conditions. The major difference between PBS and the blood is the pr esence of plasma proteins, which also bind a significant amount of drug. Thus, in this chapter we focused on the effect of plasma proteins on the sequestration, or temporary removal from solution by complexation, of AMI by liposomes. Additionally, we sought to understand the role of liposome-protein interactions in drug overdose treatment, and to apply that knowledge to further optimize our liposomal systems. Uptake of nortriptyline, the majo r metabolite of AMI, was also investigated with similar liposomal suspensions. A primary characteristic of each species considered in this study was net charge. At the physi ological pH of 7.4, both AMI and nortriptyline exist predominantly in their positively c harged forms. DMPC has no net charge, whereas DOPG bears a negative charge. Previous studies showed that liposomes composed of more negatively charged lipids sequestered more AMI than those with neutral lipids [71]. Accordi ngly, we studied liposomes composed of a 50:50 molar ratio of DMPC and DOPG, as well as pure DOPG lipids. The st ructures of both drugs and both lipids are shown in Figure 2-1. In addition to investigating the suitability of liposomes for overdose treatment, we have in vestigated binding isotherms for AMI with albumin, fibrinogen, and globulins. These stud ies have revealed some very interesting behavior which is particularly evident at lo w drug concentrations, a regime which has not been explored in detail by previous investigators. 30

PAGE 31

2.2 Materials and Methods 2.2.1 Materials Methanol, chloroform, Dulbecco s phos phate buffered saline (PBS) without calcium chloride and magnesium chloride, bov ine serum albumin (BSA), fibrinogen from bovine plasma, -globulins from human blood, hum an serum from male plasma, nortriptyline hydrochloride, and amitriptyline hydrochlori de were purchased from Sigma Aldrich. 0.45 m nylon syringe filters, YM30 centri fugation filters (30,000 molecular weight cut-off), and YM10 centri fugation filters (10,000 molecu lar weight cut-off) were purchased from Fisher Scientific The lipids 1,2-dimyristoylsn -glycero-3phosphocholine (DMPC), in powder form, and 1,2-dioleoylsn -glycero-3-[phosphorac(1-glycerol)] (sodium salt) (DOPG), in powder form, were purchased from Avanti Polar Lipids, Inc. 2.2.2 Liposome Preparation Liposomes containing a mixt ure of DMPC and DOPG lipids as well as liposom es containing DOPG lipids were prepared using an ultrasoni cation procedure. For preparing DOPG liposomes, 20 mg of lipid was dissolved in a 9:1 mixture (by volume) of chloroform:methanol such that a 10 mg / mL concentration of lipid s was obtained. The organic solvent was then evaporated under a st ream of nitrogen. After an even and uniformly dried lipid film was obtained, the dr ied lipid layer was hydrated with PBS, such that the lipid concentration was 40 mg / mL, and the hydrated lipid was sonicated in a bath sonicator (G112SP1 Specia l Ultrasonic Cleaner, Avanti Po lar Lipids, Inc.) at room temperature for 20 minutes to form lipid ve sicles. More PBS was then added, such that the lipid concentration became 4 mg / mL, and the lipid suspension was sonicated using a probe sonicator (Fisher Scientific Sonic Dismembrator Model 100) for 40 minutes at 31

PAGE 32

room temperature to reduce the vesicle siz e. The suspension was surrounded by a cool water bath during the sonication to avoid excessive heat buildup. The liposomal dispersion was filtered using a 0.45 m filter. Effective di ameter was previously determined to be 40-45 nm using a Brookhaven particle size analyzer [71]. Liposomes containing both DMPC and DOPG lipids in 50:50 molar ratios were prepared by following the same procedure as described above, except that the two lipids were mixed in a 50:50 molar ratio before dissolving the lipids in the organic liquid. In some experiments the lipid loading was doubled at t he dissolution step, and this resulted in a loading of 8 mg / mL in the final liposomal suspension. 2.2.3 Amitriptyline Uptake Experiments To quantify AMI uptake by liposomes, prot ei ns, and mixtures of both liposomes and proteins, filtration and HPLC analysis were used. Percentage of drug uptake was calculated by subtracting the final drug concentration from the initial drug concentration measured from control solutions and dividing by the initial drug c oncentration. This method could measure the bindi ng of AMI, but it provided no information regarding the time scale on which binding occurred, due to t he long time required for filtration. The time scale for binding between AMI and lipos omes was previously investigated, however, and found to be extremely rapid [71]. To determine the initial drug concentrations used for the percent uptake ca lculations, control drug solutions were made and a volume of PBS equal to the volume of the liposome dispersion added to the samples was added to the control solutions to eliminate error due to dilution. 32

PAGE 33

2.2.4 Amitriptyline Uptake with Liposomes AMI uptake was expected to depend on lipid and drug loading. The initial AMI concentration was varied from around 1 to 100 M, and two different lipid loadings, 0.36 and 0.72 mg lipid / mL were used in the experiments described below. Liposomes were added to solutions of AMI, which were all made with PBS (pH 7.4), such that the volume of the liposomal dispersi on (containing 4 or 8 mg lipid / mL) was 9% of the total solution volume giving a final lipid concentration of 0.36 or 0.72 mg lipid / mL. After being stirred for 10 minutes, the solutions were ultracentrifuged at 3000 rpm for 15-20 minutes in a vial that cont ained a YM30 filter (30, 000 molecular weight cut-off). Filter sizes were chosen to ensur e that most of the dr ug that was not bound passed through the filters, while liposomedrug complexes and liposomes did not pass through the filters, as verifi ed in [71]. To ensure that all unbound AMI was accounted for, solutions of AMI at various concent rations were passed through YM30 and YM10 filters in a separate test. Small amounts of AMI were taken up by the filters, and a linear correction curve was made and used to corr ect for AMI adsorbed by the membranes. To minimize the effect of any leaching component s from the filter, the filters were rinsed with de-ionized water and then PBS at 3000 rpm for 20 minutes prior to their use in these experiments. T he concentration of AMI in the filtrate (free drug concentration) was measured using HPLC analysis. AM I was detected at an absorbance of 215 nm after passing through a C18 column (Wat ers) using an acetonitrile/50 mM NH4H2PO4 solvent mixture in a 35/65 ratio. The calib ration curve for concentration versus area under the curve was linear with R2 > 0 99. 33

PAGE 34

2.2.5 Amitriptyline Up take with Proteins Albumin from bovine serum, fibrinogen from bovine plasma, and globulins from human blood were each tested separately for their ability to bind AMI. Human blood is made up of approximately 7% pr otein, with 4.5% albumin, 0.3% fibrinogen, and 2.5% globulins [83]. We conduct ed experiments with 2% and 4% (w/w) albumin, 2% (w/w) fibrinogen, and 1% (w/w) globulins. Also, a mixture of 4% albumin, 2% fibrinogen, and 1% globulins was tested for AMI uptake. Ea ch protein was weighed and AMI solutions were then added to the proteins. The procedure described above for uptake with liposomes was then followed, with the exce ption of filtration times, which were increased to a period ranging from 45 minutes to 5 hours. 2.2.6 Amitriptyline Uptake with Mi xtures of Liposomes and Pro teins For the mixtures tested, the proteins were first weighed and combined with AMI solutions. Liposomes were then added to the solutions such that the volume of the liposomal dispersion (containing 4 or 8 mg lipid / mL) was 9% of the total solution volume (giving a final concentration of 0.36 or 0.72 mg lipid / mL). Mixtures with protein concentrations of 4% albumin (w/w), 2% fibrinogen (w/w), and 1% globulins (w/w) were tested individually. To simulate more rea listic serum conditions, a mixture of 4% albumin (w/w), 2% fibrinogen (w/w), and 1% globulins (w/w) was tested. 2.2.7 Amitriptyline Up take w ith Human Serum 50:50 DMPC:DOPG liposomes were added to solutions of AMI in human serum from male plasma, such that the volume of the liposomal di spersion was 9% of the total solution volume, giving a final lipid concentration of 0.72 mg lipid / mL. Control solutions of AMI in PBS and AMI in serum without liposom es were also made to allow for uptake quantification and comparis on. After being stirred, the solu tions were ultracentrifuged at 34

PAGE 35

5000 rpm for 15 minutes in a vial that cont ained a YM10 filter (10,000 molec ular weight cut-off). YM10 filters were used for serum, rather than YM 30 filters, due to the presence of many components present in serum that could pass through the membrane. To minimize the effect of any l eaching components from the f ilter, the filters were rinsed first with DI water and then with PBS at 5000 rp m for 10 minutes prior to their use in these experiments. 2.2.8 Reversibility of Binding (Dilution Method) To ensure that the drug binding to lipos omes and proteins is reversible, two different paths were followed to obtain the same final composition, and uptake was measured for both cases. Specifically, tw o samples wer e tested with identical AMI concentrations at a 4% albumin (w/w) c oncentration. The drug and albumin were combined as described above for the c ontrol sample, and then 50:50 DMPC:DOPG liposomes were added to produce a final lipid concentration of 0.72 mg lipid / mL. For the second sample, a more concentrated drug solution was first combined with the proteins, followed by the 50:50 DMPC:DOPG liposomes. After 10 minutes, the second sample was diluted to the same final AM I and liposome concentrations as the control sample. Both samples were filtered and the AMI concentrations were measured via HPLC. 2.2.9 Time Dependency of Amitriptyline Uptake To gain ins ight into the effects of time on the liposome-drug binding and the liposome-drug-protein interactions, experiments were conducted in which liposomes were added to AMI solutions as described in Section 2.2.4 and uptake was measured after approximately 24 and 48 hours. This ex periment was carried out for mixtures with 35

PAGE 36

2% fibrinogen (w/w) and 50:50 DMPC:DOPG li posomes. The final lipid conc entration was again 0.72 mg lipid/ mL. 2.2.10 Nortriptyline Uptake w ith Liposomes and Albumin Once inside the body, the major metabolit e of AMI is nortriptyline. Since an overdose of nortriptyline may also cause toxicity, we have measured the uptake of nortriptyline with pure DOPG and 50:50 DMPC:DOPG liposomal systems [84]. Uptake measurements were done in PBS and in the pr esence of 4% albumin. The procedures were the same as described in Sections 2.2. 4, 2.2.5, and 2.2.6, with the exception of the protocols for HPLC analysis. A solvent mixture of acetonitrile/50 mM NH4H2PO4 at a ratio of 32.5/67.5 was used for nortriptyli ne detection. Nortriptyline was detected by measuring absorbance at 215 nm. The calibra tion curve for concentration versus area under the curve was linear with R2 > 0.99. 2.3 Results and Discussion 2.3.1 Amitriptyline Uptake w ith Liposomes In all of the results reported below, the percentage of drug uptake was determined by measuring both the AMI sample and a contro l AMI solution. The control solution was diluted with PBS to match the dilution effe ct from the liposome solution added to the AMI sample. Percentage of AMI uptake wa s plotted as a function of final AMI concentration. Typically, two experiment s were done with the same starting drug concentration, and results from each of these are included in the figures. The close proximity of the results from the repeat runs indicates the reproducibility of AMI uptake with liposomes. 36

PAGE 37

2.3.1.1 Effect of lipid loading The effect of lipid loading was in vestigated for both the 50:50 DMPC:DOPG system and the pure DOPG system Figure 2-2 shows the AMI uptake as a function of final AMI concentration for lipid loadi ng levels of 0.36 and 0.72 mg lipid / mL for both systems. For pure DOPG, uptake values varied from around 95% at 0.36 mg lipid / mL and final AMI concentrations of around 2.5 M, to nearly 100% at 0.72 mg lipid / mL and final AMI concentrations on the order of 0.01 M. The 50:50 DMPC:DOPG system seemed to be more sensitive to loading, wit h uptake values ranging from 90 to 99% in the same concentration range. In both ca ses, uptake increased as lipid loading increased, which was expected. The frac tional uptake decreases on increasing the concentration, which is expected because t here are only a limited number of binding sites on liposomes, and as the bulk drug concentra tion gets larger, a majority of these sites are occupied (saturated) leading to a r eduction in fractional drug uptake on further increase in concentration. Both lipid load ing levels produced relatively high uptake values for both the 50:50 DMPC:DOPG syst em and the pure DOPG system. Also important to note is that the maximum lipid loading level used in this study of 0.72 mg lipid / mL is still low compared to lipid loading le vels in other studies [25]. Thus, the results presented here are the lower bounds of the uptake capabilities of these systems and may be improved with higher lipid loading. If one neglects interaction between liposom es, the uptake data for the system in which the lipid loading is doubled ( double) can be predicted by scaling up the uptake data at the original lipid loading ( ). Here refers to the fracti on of drug bound to liposomes divided by the total amount of drug initially in solution, and so the ratio of drug on liposomes to fr ee drug is as follows: 37

PAGE 38

1 (2-1) This ratio depends only on the drug concentration and the amount of lipids added to the system. At a given free drug concentrati on, the amount of bound drug should double on doubling the loading, and so the ratio of bound to free drug should also double, i.e., 11 1double double. (2-2) The above equation can be simplified to give 1 2double. (2-3) However, one has to ensure that and double correspond to the same final drug concentration. The solid and dashed lines in Figure 2-2 are tr end lines from the predicted data based on the above equation. A good match between the predictions and the experimental data suggest s that the system is dilute in lipid and so interliposome interaction is negligible. If such interaction were present, liposomes could possibly fuse, which may reduce the number of sites available for drug adsorption. 2.3.1.2 Effect of charge At a pH of 7.4, DMPC has a net neutral charge, DOPG carries a ( 1) charge, and AMI is predominately pres ent in its charged form (AH+). To clearly illust rate the effect of charge on uptake, DOPG and 50:50 DOPG:DMP C should be compared at equal lipid loadings of 0.72 mg / mL. The data plotted in Figure 22 confirms that charge, and hence electrostatic interactions, play an important role in sequestration. Both systems sequestered AMI extremely well, but the pur e DOPG system, which carried twice as much net charge as the 50:50 system, appr oached 99.99% uptake at very low AMI 38

PAGE 39

concentrations. The DOPG lipid s form an exterior bilay er surface that is negatively charged, which attracts the positively char ged drug. Although t he DMPC lipid also contains a negative charge, its positively charged group repels the positively charged drug and likely also causes inter-lipid intera ction, which further reduces the available negative charges for binding. Also, it is inte resting to note that pure DOPG liposomes at 0.36 mg / mL sequestered more drug than the 50:50 DMPC:DOPG liposomes at 0.72 mg / mL. This suggests that the dependence of uptake on charge is non-linear, and probably depends on other factor s, such as bilayer stru cture and the presence of positive charge on DMPC lipids. 2.3.2 Amitriptyline Up take w ith Proteins The amount of AMI bound to albumin from bovine serum, fibri nogen from bovine plasma, and -globulins from human blood was measur ed, because it is well known that AMI is approximately 95% bound to serum protei ns while inside the blood stream [85]. Any particles developed for overdose treatment will be competing with proteins for AMI, as well as interacting with proteins themselv es. Our goal was to quantify the in vitro binding between AMI and protei ns without liposomes using our uptake procedures, and then compare this data with AM I uptake from mixtures of proteins and liposomes to determine if significant interactions bet ween liposomes and proteins occurred. 2.3.2.1 Uptake with albumin AMI uptake by albumin at 2% and 4% (w/w) is plotted as a function of final AMI concentration in Figur e 2-3. Similarly to lip id loading, increased prot ein loading resulted in increased uptake. The fractional uptak e also decreased as the concentration was increased, which was probably due to saturation effects. Ho wever, the fractional uptake leveled off above an equilibriu m concentration of about 20 M. This behavior was also 39

PAGE 40

evident in other results shown below. T he 4% albumin data, which is approximately equal to the amount of albumin in blood, s hows binding levels of around 90% at low concentrations. This data shows that on a weight basis, albumin has a much lower affinity for AMI compared to the liposom es because liposomes have a maximum AMI uptake of 99.9% at concentrations of 0.072% (w /w). The solid line in Figure 2-3 is the trend line predicted by Equation 2-3 for 4% albumin based on the uptake data for 2% albumin, assuming no protein-protein intera ction. Again, a good match between the prediction and the experim ental data suggests that the system is dilute in protein and so inter-protein interaction is negligible. While inter-protein interaction appeared to be negligible, the AMI binding curve for albumin showed an unexpected behavior. The albumin concentrations of 4% and 2% (w/w) were equivalent to around 600 and 300 M, respectively. At AMI concentrations below 20 M, one would expect AMI uptake to be constant across a range of drug concentrations because the binding sites on pr oteins are far more numerous than the number of drug molecules in solution. However, the dat a shows that the fractional uptake decreases significantly, even at very low concentrations. To ensure that time dependent kinetic effects were not responsible for this behavior, the fractional uptake was measured for an initial AMI concentration of 25 M after allowing the system to equilibrate for a period of time varying from one to three days. These results in Figure 2-3 show that the uptake values after days one, two, and three were consistent with the previous data. This suggests that the upt ake values reported her e represent equilibrium uptake. 40

PAGE 41

The sharp decay in fractional upt ake followed by leveling off at concentrations above 20 m suggests that there are two different binding sites on the proteins. Below, we propose a simple two site binding model to quantitatively understand the drug binding. The detailed interactions at t hese binding sites cannot be explored by the techniques used in this paper, but the model still sheds light on the mechanisms of uptake. We propose that the drug can bind to two different sites on t he protein to form complexes denoted by DP and DP: 11 K D + P DP (2-4) 2K D + P DP* (2-5) where D is unbound AMI, P is unbound protei n, DP is the drug-protein complex resulting from one mechanism, and DP is the drug-protein complex resulting from a second mechanism. We have further assumed that the number of sites are much larger than the number of drug molecules and so the reaction resulting in DP is zero order with respect to drug concentration. Th is assumption is based on the fact that the fractional uptake levels off beyond the 20 m drug concentration. Accordingly, the equilibrium for the binding can be represented by the following equation: DDP*, (2-6) where is the equilibrium constant for the binding under the assu mption of very large protein concentration. A mass balance for the drug yields 41

PAGE 42

totalDDPDD)( (2-7) where D and Dtotal are the free and total AMI concentrations, respectively, and DP is the concentration of the DP complex. A mass balance on the protein yields totalPPDP (2-8) where P and Ptotal are the free and total concentrations of the protein that contains a site that could be used for formation of the DP complex. The binding sites for the two mechanisms are assumed to be completely independent, and so the mass balance for DP is unaffected by DP. Based on Equations 2-4 through 2-8, Equation 2-9 was derived for the drug-protein complex DP, 2 )PD(4)PDK (K)PDK (K DPtotal total 2 total total11 total total11 (2-9) which was then used in Equation 2-10 to solve for the fraction of AMI bound (f): total totalD DP) (D 1 DP f (2-10) The model was used to find the values of K1 and Ptotal that best described the experimental data. The value of was determined by equating the fractional uptake value in the plateau region to /(1+ ). For the case of 2% albumin where a value of 1.78 was used, the best fit K1 value was 2.78 M 1, and the effective Ptotal for albumin was 5.13 M. For 4% albumin ( = 3.35), the best fit values of K1 and Ptotal were 1.83 M 1 and 9.24 M, respectively. The fitted values of Ptotal were much smaller than the total protein concentration, which were 600 and 300 M for the 4 and 2% protein solutions, respectively. Thus, the data sugges t that a significant number of albumin molecules were unavailable for binding. This could be due to inhibition of one 42

PAGE 43

mechanism by the other or conformati onal changes upon binding. Aggregation was unlikely, as lyophilized protein in the presenc e of salt was used to ensure solubility. Additionally, the close correlation between the predictions from Equation 2-3 and the 4% albumin data further support the lack of aggregation. Another potential reason for the difference between the fi tted and the true value of Ptotal could be that the binding site is not available in the native st ate of the protein, but slight conformational fluctuations around the native state allow the bindi ng. In this case, the fitted Ptotal would be the concentration of the protein molecules that possess the conformations that allow binding. These results suggest that the concentration dependent nature of AMI binding to serum proteins at therapeutic concentrations could be important considerations for physicians, particularly in the case of overdose treatment. 2.3.2.2 Uptake with fibrinogen Figure 2-4 shows AMI uptake with 2% fibrinogen (w/w), which is about double the fibrinogen concentration in plasm a. Since the system is dilute for 4% albumin, it can be assumed that 2% fibrinogen is also dilute and so Equation 2-3 can be used to predict the uptake data for 1% fibrinogen (w/w). Fibr inogen at 2% (w/w) binds about the same amount of drug as 2% (w/w) albumin, and so the model presented in Section 2.3.2.1 was used to predict K1 and the effective Ptotal value. A value of 0.67 was used for in these calculations. A K1 value of 2.90 M 1 suggested a similar affinity for AMI as albumin, while a Ptotal value of 1.68 M versus an actual fibrinogen concentration of around 60 M again pointed to a small percentage of protein molecules available for binding. 43

PAGE 44

2.3.2.3 Uptake with globulins Figure 2-4 also shows the drug uptake dat a for globulins at a concentration of 1% (w/w), and the scaled up predict ion for uptake by 2% (w/w). The scaled up data shows that at the same loading, the globulins ta ke up slightly less drug than albumin or fibrinogen. Again, Equation 2-10 was applied to the globul in case to examine the affinity and availability of the protein. The Ptotal of 2.08 M versus an actual globulin concentration of 67 M was consistent with albumin a nd fibrinogen, once more pointing to a lack of available protein molecules for AMI binding. The K1 value of 7.10 M 1 confirmed the lower affinity of globulins for AMI than albumin or fi brinogen. A value of 0.025 was used for in these calculations. 2.3.3 Amitriptyline Uptake with Mi xtures of Liposomes and Pro teins The results from Sections 2.3.1 and 2.3. 2 provided information on the AMI binding properties of liposomes and proteins independently. The liposomes were shown to sequester large amounts of AMI in PBS, while the proteins also bound to the drug, although to a lesser extent. Below we repor t the drug uptake in the presence of both liposomes and proteins. These experiments we re conducted to better simulate in vivo conditions and also to determine whether t he proteins bind to the liposomes. To accomplish the latter objective, a simple mass balance was utilized to predict the drug uptake by a mixture of liposomes and proteins, based on the assumption that these components do not interact with each other, i.e. proteins do not bind to liposomes. The mass balance leads to the following equation: n n mix1 1 1i i i i (2-11) 44

PAGE 45

where i and mix are the uptakes in the ith component and the mixture, respectively. If the measured uptake for the mixture is less than that predicted by the above equation, one may expect that the proteins are adsorbing on the liposomes, leading to a reduction in the number of available sites on liposomes for drug adsorption. 2.3.3.1 Albumin and liposomes In the experiments described below, 4% albumin was combined with 50:50 DMPC:DOPG liposomes at lipid concentrations of 0.36 and 0.72 mg lipid / mL as well as pure DOPG liposomes at a lipid concentration of 0.36 mg lipid / mL. Figure 2-5 shows the AMI uptake predictions and experimental results for 50:50 DMPC:DOPG liposomes and 4% albumin as a function of final AMI concentration. At the lower lipid loading level of 0.36 mg lipid/ mL, the system behaved as if no interaction took place between the liposomes and albumi n. The experimental values closely corresponded to the trend line predicted by the mass balance. This would seem to suggest that the effectiveness of the liposomes at sequestering AMI is uninhibited by the presence of albumin. However, at the higher lipid loading le vel of 0.72 mg lipid / mL, the experimental values were consistently lower than the predicted values. Increasing the lipid loading level seemed to increase binding between the lipos omes and proteins, causing the overall system uptake of AMI to be reduced. At the lowest final AMI concentrations of around 0.1 M, the uptake values were around 97%, as opposed to predictions approaching 99%. The system of liposomes and albumin was still a marked improvement over the 4% albumin (w/w ) system alone, which only bound around 90% of the drug at similar concentrations. Doubl ing the amount of lipid in the system did increase the uptake in the low concentration regime from around 94-95% to around 97%. Thus, increasing lipid loading appears to increase uptake, but the effect is less 45

PAGE 46

than expec ted in the presence of albumin due to compet ing interactions with the proteins. AMI uptake values for mixtures of pure DO PG liposomes and 4% albumin at a lipid loading level of 0.36 mg lipid / mL are also plotted in Figure 2-5. The data shows that significantly less AMI was bound in the mixt ure of liposomes and proteins than the predicted values. This was particularly true in the low concentration region. Additionally, the measurement s have more variation compared to that in results reported earlier. There also seems to be a minimum in the drug uptake that occurs at a concentration of 0.35 M. The drug uptake by the mi xture of 4% albumin and pure DOPG liposomes at 0.36 mg / mL was comparable to the upt ake by the mixture of 4% albumin and 50:50 DOPG:DMPC liposomes at 0.36 mg lipid / mL. This is in contrast to the performance of the liposomes in the absence of albumin, where the pure DOPG liposomes sequestered significantly more AMI than the 50:50 liposomes. The pure DOPG liposomes carry twice as much c harge as the 50:50 DMPC:DOPG system, and this may be leading to increased protein bi nding, causing a greater reduction in uptake than for the 50:50 DMPC:DOPG system. T hese results suggest that pure DOPG liposomes may not be the opt imal system to use for drug detoxification due to high levels of protein interaction. Due to this reason, the 50:50 DMPC:DOPG system at a lipid concentration of 0.72 mg lipid / mL was used for subsequent mixture experiments involving fibrinogen and globulins, rather than pure DOPG liposomes. 2.3.3.2 Fibrinogen and liposomes Mixtures of 2% fibrinogen and 50: 50 DMPC:DOPG liposomes at lipid concentrations of 0.36 and 0.72 mg lipid / mL were tested for AMI uptake, and the measured and predict ed uptake values are plott ed in Figure 2-6. For the lower lipid 46

PAGE 47

loading of 0.36 mg lipid / mL, the measured uptake valu es were around 88 to 89%, falling significantly below the predicted values. This was especially true in the lowest concentration regime, where the uptake was roughly 5% less than expected at a final concentration of around 0.1 M. A lipid loading of 0.72 mg lipid / mL produced a mixture with similar behavior. Maximum uptakes were between 95% and 96%. The largest deviation from the predicted values was again in the low concentration area of the plot, with the measured uptake falling almo st 4% below the prediction. These results show that 50:50 DMPC:DOPG liposomes sequester significantly more AMI than 2% fibrinogen alone. 2% fibrinogen only bound 75 % of the AMI in solution at a final AMI concentration of nearly 0.10 M. The mixture of both liposomes and fibrinogen bound 95-96% in the same concent ration region. However, the results also point to interactions between liposomes and fibrinogen that hinder the ability of liposomes to sequester free AMI from soluti on, similar to the effect observed with albumin and pure DOPG liposom es. The albumin also affected the 50:50 DMPC:DOPG system, but to a lesser extent. It is important to note that fibrinogen has an inhibitory effect on drug uptake by liposomes at 2% (w/w), which is half the concentration of albumin (4% (w/w)). Therefore, it seems that fibrinogen has a higher inhibitory effect, presumably due to a larger binding to the liposomes. With t he approximate protein concentration of 7% (w/w) in plasma, it is apparent that liposome-protein interactions will play an important role in overdose treatment, and t hat fibrinogen is one of the key proteins to consider when studying those interactions. 2.3.3.3 Globulins and liposomes Figure 2-7 shows the predicted trend line an d the measured value s for AMI uptake behavior for a mixture of 1% globulins (w /w) and 50:50 DMPC:DOPG liposomes at a 47

PAGE 48

lipid loading of 0.72 mg lipid / mL. The uptake values fo r the system were around 9596%, as compared to 30-40% in the sa me drug concentration range for pure 1% globulins. Thus, the liposome-protein mixture sequestered a substantially larger amount of AMI from solutions than globu lins alone. An inhibition was once again observed at low concentrations as the uptake seemed to level off at 95-96%, rather than approaching 99%. Based solely on these preliminary results, it se ems that all three proteins reduce the uptake of AMI by liposomes at low AMI concentrations, with fibrinogen and globulins having more drasti c effects than albumin. To check for additional effects arising from interact ions between the proteins themselves, experiments were conducted where mixtures of all three proteins were used. These results are presented below. 2.3.3.4 Albumin, fibrinogen, globulins, and liposomes AMI uptake was first measured for the 7% (4% albumin (w/w), 2% fibrinog en (w/w), 1% globulins (w/w)) pr otein mixture without the additi on of liposomes. The data from measurements of protein binding for each individual protein were used to predict the uptake for the mixture by using Equation 2-11. Nex t, mixtures containing 4% albumin (w/w), 2% fibrinogen, 1% globulin s (w/w), and 50:50 DMPC:DOPG liposomes with a lipid loading of 0.72 mg lipid / mL were tested. The pr edicted trend line of AMI binding in the presence of 7% proteins plus liposomes, measurements from 7% proteins, and measurements from 7% proteins plus liposomes are plotted in Figure 2-8. Several noteworthy observations can be made from the experimenta l results. As previously observed, the pr otein-liposome mixture did no t sequester as much AMI as the simple mass balance equation predicted Rather than appr oaching 98-99% uptake at low concentrations, the system only bound ar ound 96% of the drug. This result again 48

PAGE 49

suggests that under the experimental condi tions, proteins are interacting with lipos omes, and are perhaps adsorbing on the surface. However the presence of liposomes results in a significant reduction in the free drug concentration. For instance, at an initial AMI concent ration of nearly 1.70 M, the proteins bou nd around 90% of the AMI, bringing the final concentration to roughly 0.17 M. With an almost identical initial drug concentration, the protein-liposom e mixture bound around 96% of the drug, resulting in a final AMI concentration of around 0.07 M. Therefore, we have succeeded in reducing the free AMI concentration by nearly 60% in the presence of 7% proteins with 50:50 DMPC:DOPG liposomes at a lipid loading of 0.72 mg lipid / mL. The initial AMI c oncentration of 1.70 M was tested based on the reported relevant values for overdose treatment of 1-3 M [9]. Other groups, incl uding Dhanikula et al. [25], have tested in vitro AMI uptake at even lower concentrations. Uptake increases with decreasing concentration, and the report ed therapeutic dose range for AMI is 0.3 to 0.8 M [9]. Thus, any uptake values found in or below this concentration range may not be as relevant to overdose treatment. While the 7% protein mixture does not contain all the components found in human plasma, the ex perimental uptake values found in this study still highlight the potent ial for DMPC:DOPG liposome systems to be used for AMI overdose treatment. 2.3.3.5 Effects of liposome-protein binding All of the results presented thus far suggest that proteins interact w ith liposomes and perhaps adsorb on the surface. This hypothesis is supported by extensive literature in this field. Chonn, Semple, and Cullis [ 86,87] have done numerous studies and concluded that significant amounts of proteins bind to liposomes. The amount of protein bound depends largely on the composition and charge of the liposomes. They 49

PAGE 50

have also found that in vivo half-lives of lip osomes tend to be longer for those that bind less protein. One of thei r studies involved the use of 35:45:20 phos phatidylcholine (PC):cholesterol (CH):phosphatidylglycerol (PG) liposomes [86]. While these liposomes were slightly different than the ones used in our experiments, they did contain both the PC and PG components. Protei n binding data taken from t hat study suggested that roughly 6 10 7 moles of protein were bound to lipos omes per gram of lipid. Assuming an AMI uptake of 98% at 1 M, 1 40 6 moles of AMI would be bound to liposomes per gram of lipid. These rough estimates show that at small drug concentrations, the number of drug molecules that bind to t he liposome surface is comparable to the number of bound protein molecules, and so it is expected that the fractional drug uptake will decrease. However at higher drug concentrations, the bound drug molecules are significantly more numerous than the bound protein molecules leading to a small change in fractional uptake. In spite of the detrimental effect of protein binding, the liposomes were able to reduce the free AMI concentration by 50-60% While encouraging, this in vitro result failed to account for possible in vivo effects. One of the mo st important effects to consider is the response by the bodys i mmune system and its effect on the lifetime of liposomes in the circulatory system. While we cannot make any assumptions about the potential half-life of our li posomes without in vivo ex periments or pharmacokinetic modeling, we can review the reported values from Chonn, Semple, and Cullis. In their studies, they report 50% of the 35:45:20 PC:CH:PG liposomes to be recovered from mice after 30 minutes, and around 30% to be recovered after 120 minutes [86]. For some applications, including drug delivery, t hese circulation times may be completely 50

PAGE 51

unacceptable. However, these may be adequat e for drug detoxification as AMI uptake has been shown to be rapid. A q uick remova l of the liposomes by the immune system from the blood stream to t he liver may even be desirable in the case of overdose treatment. But, if increasing residence time in the circulatory system is necessary, it can be done by increasing the lip id loading [87]. As the lipid loading used in our study was very low compared to other studies, we could increase our lipid loading, and this may have the dual benefits of increases in drug uptake and circulatory half lives. Cholesterol addition is another possibility, as it adds stability to liposomes and increases in vivo circulation times in some cases [87]. 2.3.4 Uptake of Amitriptyline w ith Liposomes in Human Serum The results presented above show that the three main proteins present in blood bind to liposomes, leading to a reduction in dr ug uptake efficiencies. In reality, human blood also contains other proteins which are present in trace amount s and are difficult to isolate. Thus, to better gauge the effectivene ss of liposomes at sequestering drug under in vivo conditions, drug uptake experiments were conducted in human serum. AMI uptake in human serum is plotted as a function of final AMI concentration in Figure 2-9. In serum, AMI was roughly 90-92% bound wit hout liposomes, which is comparable to the experimental results found in 7% proteins. When liposomes were added, the uptake rose to 94-98%. Thus, AMI uptake values with liposomes fell from 99% in PBS to 96 in 7% proteins, and increased to 98% in human serum. As opposed to the 50-60% free drug concentration reduction seen in t he presence of 7% proteins, the 50:50 DMPC:DOPG liposomes produced a 35-70% reduc tion in serum. Clearly, the proteinliposome interactions have increased the va riability in drug uptake by liposomes, while the overall effectiveness of t he liposomes in serum with re spect to AMI detoxification 51

PAGE 52

appears to be comparable to 7% protein mixtures. A reduction of free drug concentration by about 35-70% could have a si gnificant beneficial effect in cases of drug overdose [88]. Furthermore, it may be possible to give the overdose patient multiple doses of liposomes, which could lead to a larger reduction. It may also be possible to increase the lipid loading, which could lead to higher drug uptake values and longer circ ulation times in the bloodstream. 2.3.5 System Characterization The results presented in the previous sect ion inherently assume that the binding of AMI to proteins and liposomes is reversib le, and that the meas ured values correspond to thermodynamic equilibrium. To test these hypotheses, t he experiments described below were conducted. 2.3.5.1 Reversibility of binding Albumin and 50:50 DMPC:DOPG liposomes wit h a final lipid concentration of 0.72 mg lipid / mL were used to test for bi nd ing reversibility. As de scribed in Section 2.2.8, two samples with the same final compositi on were prepared by fo llowing two different paths, and the drug uptake was measured for eac h case. If the AMI was bound to the albumin or liposomes irreversibly, the uptak e for the two samples would not match, as they were allowed to reach an initial equilibri um at different concent rations. The results of the experiment are shown in Figure 2-10. At a final AMI concentration of 0.50 M, the uptake values for both the diluted sample and standard sample were around 96.4%. At a final concentration closer to 0.12 M, the uptake values for the two samples differed by less than 1%, which is within the range of experimental error. It is noted that the experimental error is likely to be larger at smaller concentrations because a small amount of mass loss from the systems 52

PAGE 53

could caus e appreciable errors in the fractional uptake. Based on this data, the binding of AMI appears to be reversible. 2.3.5.2 Time dependency of amitripty line uptake To determine whether the measur ements reported above correspond to equilibrium, it was decided to conduct uptake measurements at various times. A system with 2% fibrinogen (w/w) and 50:50 DMPC :DOPG liposomes with a lipid loading of 0.72 mg lipid / mL was used for these experiments. AMI uptake was measured 24 and 48 hours after mixing all the components, and these results, along with those measured immediately after mixing, are plotted in Figure 2-11. The uptake values of AMI after 24 and 48 hours were very similar to the sample filtered shortly after mixing. This result suggests that the binding is quick and that the measured values correspond to equilibrium binding. 2.3.6 Nortriptyline Uptake Based on the fact that nortrip tyline is the well-known prim ary metabolite of AMI, it can be anticipated that nortripty line may be important in AMI overdose cases [89]. This is supported by the results of a case study by Franssen et al. [89], which showed an initial concentration of nortriptyline of around 700 g / L for a patient who had overdosed on AMI. Nortriptyline could be of concern for AMI overdose patie nts, as many people who overdose on AMI have been using the drug for depression related illnesses prior to overdosing. Thus, nortriptyline will undoubtedly be present in their system at the time of overdose if they have recently taken AMI re gularly. The study conducted by Franssen et al. also demonstrated that nortriptyline c an cause many of the same serious health problems associated with AMI. For this reason, nortriptyli ne uptake was investigated 53

PAGE 54

with 50:50 DMPC:DOPG liposomes at lipid loading lev els of 0.36 and 0.72 mg lipid / mL and DOPG at lipid loading le vels of 0.36 and 0.72 mg lipid / mL. 2.3.6.1 Nortriptyline uptake wi th 50:50 DMPC:DOPG liposomes Figure 2-12 shows nortriptyline and AM I uptake by 50:50 DMPC:DOPG liposomes versus final drug concentration. At final nortriptyline concentrati ons of nearly 0.025 M, the maximum uptake was slight ly higher than 96% at a lipid loading of 0.36 mg lipid / mL. For a lipid loading of 0.72 mg lipid / mL, the nortriptyline uptake reached values slightly greater than 99%. The AMI data plotted in the same figure shows that the liposomes sequester nortriptyline more efficiently compared to AMI. These results show that any system designed for the treatment of AMI overdose will likely result in reduction of free drug concentration of nortripty line as well. This was expected, as both compounds are predominantly charged at pH 7.4 and have similar structures. 2.3.6.2 Nortriptyline upt ake w ith DOPG liposomes Nortriptyline uptake with pur e DOPG liposomes is plotted as a function of final concentration in Figure 2-13. Both lipid loadings for both compounds showed uptakes greater than 99%. As is the case for AM I, the pure DOPG liposomes sequester more nortriptyline than the 50:50 DMPC:DOPG liposomes. A gain, this was expected because the binding of nortriptyline to lipos omes is also driven by electrostatic interactions. 2.3.6.3 Nortriptylin e uptake w ith albumin To determine the effect of protein-liposom e interaction on nortriptyline binding, the uptake of nortriptyline with 50:50 DMPC:DOPG liposomes at a lipid concentration of 0.72 mg lipid / mL was tested in the presence of 4% albumin. In the presence of liposomes and albumin, around 98% of the nortriptyline was bound. Without liposomes, 54

PAGE 55

the albumin bound only 85% to 86% of the drug. Both of these results were comparable to the binding seen by AMI in the presence of albumin and mixtures of albumin and lipos omes. Overall, the results provide additional evidence fo r the possible dual effectiveness of liposomes at seques tering both AMI and nortriptyline. 2.4 Conclusions Liposomes composed of DO PG a nd 50:50 DMPC:DOPG were tested for their ability to sequester AMI in PBS, in the pr esence of albumin, fibrinogen, and globulins, as well as in the presence of human se rum. Experiments we re conducted using filtration procedures where free AMI was allowed to pass through filters, while liposomes and proteins were largely removed. In PBS at a pH of 7.4, both the DOPG liposomes and the 50:50 DMPC:DOPG liposomes sequestered at least 99% of the AMI at low concentrations. In the presence of albumin, pure DOPG liposomes were shown to be greatly inhibited by the proteins. In the presence of 7% proteins (w/w) composed of 4% albumin, 2% fibrinogen, and 1% globulins, the 50:50 DMPC:DOPG liposomes at a lipid loading of 0.72 mg lipid / mL sequestered 95-96% of t he free drug, as opposed to a predicted 99%. The free AMI concentration was still reduced by 50-60%, however. In human serum, the 50:50 DMPC:DOPG liposom es took up about 94-98% of the drug and reduced the free drug concentration by 35-70%. These results suggest that liposome-protein interactions reduce the effectiveness of 50:50 DMPC:DOPG liposomes to sequester AMI. This is probably due to protein adsorption to the surface of liposomes, reducing the number of available charged sites for drug binding. Other aspects of liposome dispersions fo r drug detoxification were examined as well. Filtration techniques were again used to determine that AMI binding to liposomes in the presence of proteins is quick and reve rsible. Finally, uptake studies conducted 55

PAGE 56

with nortriptyline, the major meta bolite of AMI, suggests that systems developed for AMI overdose treatment may also be useful for reducing the free concentration of nortriptyline. While the presence of serum proteins reduces the drug binding capacity of liposomes, a 35-70% reduction in drug concen tration, which is attainable by the systems explored in this chapter, may have significant benefits fo r overdosed patients [88]. Additionally, even the hi ghest lipid loading used in this chapter is below the lipid loadings used in other related studies, and so it may be feasible to increase the lipid loading or give multiple liposome injections to the overdosed patient s. Still, the study demonstrates the ability of liposomes to remove free AMI from solutions in the presence of proteins and human serum, and supports the idea of usi ng a single detoxification treatment for both AMI and nortriptyline. 56

PAGE 57

A B C D Figure 2-1. Structures of drugs and lipids used for drug uptake studies with liposomes: (a) amitriptyline; (b) nortriptyline; (c) 1,2-dimyristoyl-sn-glycero-3phosphocholine (DMPC) lipid; (d) 1,2-di oleoyl-sn-glycero-3-[phospho-rac-(1glycerol)] (DOPG) lipid. 57

PAGE 58

Figure 2-2. Measured and pr edicted (Equation 2-3) percent amitriptyline uptake for pure DOPG liposomes and 50:50 DMPC:DOP G liposomes in pH 7.4 buffer at final lipid concentrations of 0.36 and 0.72 mg lipid/mL versus final amitriptyline concentration. Key: DOPG at 0.72 mg/mL ( ); DOPG at 0.36 mg/mL ( ); DOPG at 0.72 mg/mL predictio n (solid line); 50:50 DMPC:DOPG at 0.72 mg/mL ( ); 50:50 DMPC:DOPG at 0.36 mg/mL ( ); 50:50 DMPC:DOPG at 0.72 mg/mL prediction (-). 58

PAGE 59

Figure 2-3. Measured and model ed (Equation 2-10) percent am itri ptyline uptake for 2% albumin (w/w) in pH 7.4 buffer versus final amitriptyline concentration. Measured, modeled (Equation 2-11), and predicted (Equation 2-3) percent amitriptyline uptake for 4% albumin (w /w) in pH 7.4 buffer versus final amitriptyline concentration. Key: 4% albumin ( ); 4% albumin extended mixing time ( ); 4% albumin prediction (so lid line); 4% albumin model ( ); 2% albumin ( ); 2% albumin model (-). 59

PAGE 60

Figure 2-4. Measured and model ed (Equation 2-10) percent am itriptyline uptake for 2% fibrinogen and 1% globulins (w/w) in pH 7.4 buffer versus final amitriptyline concentration; predicted (Equation 2-3) percent amitriptyline uptake for 1% fibrinogen and 2% globulins (w/w) in pH 7.4 buffer versus final amitriptyline concentration. Key: 2% fibrinogen ( ); 2% fibrinogen modeled ( ); 1% fibrinogen prediction ( ); 1% globulins ( ); 1% globulins mo deled (-); 2% globulins prediction (solid line). 60

PAGE 61

Figure 2-5. Measured and pr edicted (Equation 2-11) percent amitriptyline uptake for mixtures of 4% albumin (w/w) and 50:50 DMPC:DOPG liposomes with lipid concentrations of 0.72 and 0.36 mg lipid/m L and mixtures of 4% albumin and DOPG liposomes with lipid concentrations of 0.36 mg lipid/mL in pH 7.4 buffer versus final amitriptyline concentration. Key: 4% albumin + 50:50 DMPC:DOPG at 0.36 mg lipid/mL ( ); 4% albumin + 50:50 DMPC:DOPG at 0.36 mg lipid/mL prediction (solid li ne); 4% albumin + 50:50 DMPC:DOPG at 0.72 mg lipid/mL ( ); 4% albumin + 50:50 DMPC :DOPG at 0.72 mg lipid/mL prediction (-); 4% albumin + DOPG at 0.36 mg lipid/mL ( ); 4% albumin + DOPG at 0.36 mg lipid/mL prediction ( ). 61

PAGE 62

Figure 2-6. Measured and pr edicted (Equation 2-11) percent amitriptyline uptake for mixtures of 2% fibrinogen (w/w) and 50:50 DMPC:DOPG liposomes with lipid concentrations of 0.36 and 0.72 mg lipid/mL in pH 7.4 buffer versus final amitriptyline concentration. Key: 2% fibrinogen + 50:50 DMPC:DOPG at 0.36 mg lipid/mL ( ); 2% fibrinogen + 50:50 DMPC:DOPG at 0.36 mg lipid/mL prediction (solid line); 2% fibri nogen + 50:50 DMPC:DOPG at 0.72 mg lipid/mL ( ); 2% fibrinogen + 50:50 DMPC:D OPG at 0.72 mg lipid/mL prediction (-). 62

PAGE 63

Figure 2-7. Measured and pr edicted (Equation 2-11) percent amitriptyline uptake for mixtures of 1% globulins (w/w) and 50:50 DMPC:DOPG liposomes with lipid concentrations of 0.72 mg lipid/mL in pH 7.4 buffer versus final amitriptyline concentration. Key: 1% globulins + 50:50 DMPC:DOPG at 0.72 mg lipid/mL ( ); 1% globulins + 50:50 DMPC:DOPG at 0.72 mg lipid/mL prediction (solid line). 63

PAGE 64

Figure 2-8. Measured and predi cted (Equation 2-11) uptake of amitriptyline with 7% proteins (4% albumin, 2% fibrinogen, 1% globulins) and 50:50 DMPC:DOPG liposomes (0.72 mg lipid/mL) versus final amitriptyline concentration; measured uptake of amitript yline with 7% proteins versus final amitriptyline concentration. Key: 7% proteins + 50:50 DMPC:DOPG at 0.72 mg lipid/mL ( ); 7% proteins + 50:50 DMPC:DOPG at 0.72 mg lipid/mL prediction (solid line); 7% proteins ( ). 64

PAGE 65

Figure 2-9. Measured uptake of amitriptyline in human serum with and without 50:50 DMPC:DOPG liposomes (0.72 mg lipid /mL) versus final amitriptyline concentration (the data corresponding to 98% uptake is two overlapping points). Key: human serum + 50:50 DMPC:DOPG at 0.72 mg lipid/mL ( ); human serum ( ). Figure 2-10. Percent amitriptyline uptake fo r mixtures of 4% albumin (w/w) and 50:50 DMPC:DOPG liposomes in pH 7.4 bu ffer versus final amitriptyline concentration where standard and dilution te st methods were used to test for reversible binding. Ke y: standard test method ( ); dilution test method ( ). 65

PAGE 66

Figure 2-11. Measured and pr edicted (Equation 2-11) percent amitriptyline uptake for mixtures of 2% fibrinogen (w/w) and 50:50 DMPC:DOPG liposomes with lipid concentrations of 0.72 mg lipid/mL in pH 7.4 buffer versus final amitriptyline concentration filtered shortly after mixi ng, 24 hours later, and 48 hours later. Key: shortly after mixing ( ); prediction (solid line); 24 hours later ( ); 48 hours later ( ). 66

PAGE 67

Figure 2-12. Percent amitriptyline and nortriptyline uptake for 50:50 DMPC:DOPG liposomes in pH 7.4 buffer at final lipid concentrations of 0.36 and 0.72 mg lipid/mL versus final drug concentrati on. Key: amitriptyline with 50:50 DMPC:DOPG at 0.36 mg/mL ( ); nortriptyline with 50:50 DMPC:DOPG at 0.36 mg/mL ( ); amitriptyline with 50:50 DMPC:DOPG at 0.72 mg/mL ( ); nortriptyline with 50:50 DMPC :DOPG at 0.72 mg/mL ( ). 67

PAGE 68

68 Figure 2-13. Percent amitriptyline and nortr iptyline uptake for pure DOPG liposomes in pH 7.4 buffer at final lipid concentrati ons of 0.36 and 0.72 mg lipid/mL versus final drug concentration. Key: amitri ptyline with DOPG at 0.36 mg/mL ( ); nortriptyline with DOPG at 0.36 mg/mL ( ); amitriptyline with DOPG at 0.72 mg/mL ( ); nortriptyline with DO PG at 0.72 mg/mL ().

PAGE 69

CHA PTER 3 AMITRIPTYLINE BINDING TO PE GYLAT ED, ANIONIC LIPOSOMES 3.1 Introduction In Chapter 2, we showed that although negat ively charged liposomes are good candidates for amitriptyline (AMI) overdose treatment, their effect is reduced due to interactions between liposomes and proteins present in human blood. The current chapter intends to prove that the inclusion of polyethylene glycol (PEG) chains on the liposomes could significantly improve the drug sequestration by liposomes. In addition to improving drug uptake characteristics, incl usion of polyethylene glycol (PEG) into the liposomes will presumably lead to longer in vivo circulation times of the vesicles, which is desirable. Additionally, we report the effect of increased lipid loading on drug uptake, as the loading of 0.72 mg lipid / mL explored in Chapter 2 was considerably lower than reported values in other studies [25]. Fu rthermore, we report the effects of liposome size on the drug binding characteristi cs, which to our knowledge has not been considered before. During storage, pharmaceutical products such as liposomes can undergo changes resulting in a lack of pot ency or undesired side effects. As a preliminary screening for packing effects, li posomes were also stored and checked periodically for their drug uptake properties. 3.2 Materials and Methods 3.2.1 Materials Human serum from male plasma, met hanol, chloroform, Dulbeccos phosphate buffered saline (PBS) without calcium chloride and magnesium chloride, cholesterol (CH), and amitriptyline hydr ochloride were purchased fr om Sigma-Aldrich. 0.45 m nylon syringe filters and Centri prep YM10 centrifugation filters (10,000 molecular weight 69

PAGE 70

cut-off) were purchased from Fisher Scientific. The lipids 1,2-dimyristoylsn -glycero-3phosphoc holine (DMPC), in powder form, 1,2-dioleoylsn -glycero-3-[phospho-rac-(1glycerol)] (sodium salt) (DOPG), dissolv ed in chloroform, and 1,2-dipalmitoyl sn -glycero3-phosphoethanolamineN -[methoxy(polyethylene glyc ol)-2000] (ammonium salt) (DPPE-mPEG-2000), in powder form, were purc hased from Avanti Polar Lipids, Inc. A Mini-Extruder kit for liposome preparation wa s also purchased from Avanti Polar Lipids, Inc. 3.2.2 Liposome Prepar ation via Sonication Liposomes composed of a molar ra tio of 50:50 DMPC:DOPG, 55:15:30 DMPC:DPPE-mPEG-2000:CH, 95:5 DO PG:DPPE-mPEG-2000, 85:15 DOPG:DPPEmPEG-2000, and pure DOPG lipid s were prepared using an ultrasonication procedure. Lipids were combined in their respective molar ratios and then dissolved in a 9:1 mixture (by volume) of chloroform:methanol such that a 20 mg / mL concentration of lipids was obtained. The organic solvent was then ev aporated under a stream of nitrogen. After an even and uniformly dried lipid film was obtained, the dried lipid layer was hydrated with PBS, such that the lipid concentration was 80 mg / mL, and the mixture was sonicated in a bath sonicator (G112SP1 Special Ultrasonic Cleaner, Avanti Polar Lipids, Inc.) at room temperature for 20 minutes to form lipid vesicles. More PBS was then added, such that the lipid concentration became 8 mg / mL, and the lipid suspension was sonicated using a probe sonicator (Fisher Scientific Sonic Dismembrator Model 100) for 40 minutes at room temperature to reduce the vesicle size. The suspension was surrounded by a cool water bath during the sonication to avoid excessive heat buildup. The liposom e dispersion was filtered using a 0.45 m 70

PAGE 71

filter. In some experiments, the lipid loading was doubled at the dissolution step, and this resulted in a loading of 16 mg lipid / mL in the final liposome suspension. 3.2.3 Liposome Prepar ation via Extrusion It has been reported that liposome diamet ers of around 100-200 nm are best for extending in vivo circulation time [26,90]. Accordingly, it was decided to explore the effect of liposome size on drug binding by preparing liposomes of controlled size by using the extrusion process. Specifically, liposomes composed of a molar ratio of 50:50 DMPC:DOPG, 95:5 DOPG:DPPE-mPEG-2000, and pure DOPG lipid s were prepared via extrusion. An even and uniformly dried lipid film was obtained with the appropriate lipid composition by following the same proc edures as described above. Next, the dried lipid layer was hydrated with PBS, such that the lipid concentration was 8 mg / mL. The lipid suspension was then heated to 30C and stirred for 45 minutes. After stirring, the lipid dispersion was extruded through a 100 nm membrane 13 times. The same procedures were used to prepare larger li posomes by using a membrane with a pore size of 400 nm. 3.2.4 Liposome Characterization The mean diameters and size distributions of the liposomes were analyzed using a Precision Detectors P DDLS/CoolBatch + 90T instrument. The data was analyzed with the Precision Deconvolve32 Program. The measurements were taken at 20C with a scattering angle of 90 using a 683 nm laser source. 3.2.5 Measurement of Amitriptyline Up take b y Liposomes in Buffer and Human Serum Liposomes were added to solutions of AMI in PBS or human serum from male plasma, such that the volume of the liposome dispersion was 9% of the total solution 71

PAGE 72

volume, giving a final lipid concent ration of 0.72 mg lipid / mL. In some cases the final lipid concentration in the drug-liposome mixture was doubled to 1.44 mg lipid / mL by either preparing liposomes with a higher lipid loading of 16 mg lipid / mL or by mixing a larger volume of liposomes with 8 mg / mL lipid loading. The initial AMI concentration varied from about 1 to 10 M, and in some cases 1 to 20 M, which is a physiologically relevant range for toxic plasma drug levels [9,91,92]. Control solutions of AMI in PBS and AMI in serum without liposomes were also made to allow for uptake quantification and comparison. After being stirred, the solu tions were ultracentri fuged at 5000 rpm for 15 minutes in a vial that contained a YM10 f ilter (10,000 molecular weight cutoff). To minimize the effect of any l eaching of components from the filter, the filters were rinsed first with DI water and then with PBS at 5000 rp m for 10 minutes prior to their use in these experiments. T he concentration of AMI in the filtrate (free drug concentration) and the control solution was detected by measuring UV absorbance at 215 nm after passing the samples through a C18 column using an acetonitrile/50 mM NH4H2PO4 solvent mixture in a 35/65 ratio. The ca libration curve for concentration versus area under the curve was linear with R2 > 0 99. To ensure that all unbound AMI was accounted for, solutions of AMI at various concentrations were passed through YM10 filters in a separate test. Small amounts of AMI were taken up by the filter, and a linear correction curve was made and used to correc t for AMI adsorbed by the membranes in subsequent tests. 3.2.6 Storage Tests The stabilit y and effectiveness of any pharma ceutical product after storage is an important issue, especially for complex dos age forms such as liposome suspensions. Experiments were therefore carried out to study the effects of storage on liposomes in 72

PAGE 73

the context of drug overdose treatment. The liposomes ch osen for this portion of the study, based on results from experiments discussed in Section 3.2.5, were pure DOPG lipos omes and 95:5 DOPG:DPPE-mPEG-2000 lip osomes. Lipid dispersions were extruded through 100 nm membranes as described in Section 3.2.3, with a lipid loading of 8 mg / mL. The liposomes were stored in the refr igerator at temperatures of 2-8C for 34 days. Periodically, liposome samples were withdrawn and tested to measure the drug binding characteristics at an AMI concentration of 2.7 M. This method of storage was chosen based on similar reported stor age methods for comm ercial drug loaded liposome based therapies such as CAELYX and DOXIL. 3.2.7 Data Analysis The data was analyze d using JMP software developed by SAS. Regression lines were fitted to the concentration dependent uptake data, and 95% confidence intervals (CI) are displayed to allow for both experim ental error analysis and comparison between two or more data sets. The confidence intervals were calculated for mean uptake values using the Student t distribution. For the ti me dependent storage data, data points were replicated and error bars are shown. 3.3 Results and Discussion 3.3.1 Liposome Characterization Sonication and extrusion methods were used in the present study to form lipos omes. Sonication was one of the firs t methods commonly used to form lipid vesicles and has been used by a number of researchers for preparing small unilamellar vesicles (SUV) [27,93,94]. The mean diameter s and size distributions of several of the liposomes explored here are shown in Figure 3-1. The 85:15 DOPG:DPPE-mPEG2000 liposomes and pure DOPG li posomes, which were both sonicated, showed similar 73

PAGE 74

size distributions with primary modes containing liposom es with mean diameters of 41 5 5 and 39 2 4 5 nm, respectively. In both cases, and particularly for the PEG-lipids, a small percentage of the liposomes had diamet ers in the range of about 400 nm. The 55:15:30 DMPC:DPPE-mPEG -2000:CH liposomes, which were also sonicated, were much larger with diameters of 77 7 10 4 nm. This trend of increasing vesicle size upon inclusion of CH was also observed by Lapinski et al. with DOPC lipids [94]. Extrusion, a more recently developed method of forming lipid vesicles, has become a widely accepted technique due to the ability to control vesicle size [94,95]. The 95:5 DOPG:DPPE-mPEG-2000 liposomes extruded with 100 nm membranes had a mean diameter of 118 15 nm. The 50:50 DM PC:DOPG liposomes made with 400 nm membranes had a vesicle size of 284 36 nm. 3.3.2 Increased Lipid Loading for 50:50 DMPC:DOPG Liposomes In Chapter 2, a concentration dependent free AMI reduction of 35-70% by 50:50 DMPC:DOPG liposomes was reported in hum an serum samples. While the magnitude of this reduction is clinically significant, the concentration depende nce of the uptake is undesirable. The decrease in fractional upt ake with increasing concentration likely arises due to saturation of the binding sites on liposomes, and so it may be speculated that an increased lipid loading would reduce this effect. Increasing the lipid loading in the drug-liposome mixtures c an be accomplished either by increasing the volume of liposome dispersions added to drug solutions, or by increasing the lipid loading in the liposome formulation. Both of these approac hes were utilized to increase the lipid loadings. Figure 3-2 shows AMI uptake values as a function of final drug concentration for the 50:50 DMPC:DOPG liposomes at 0.72 mg lipid / mL and data for loadings of 1.44 mg lipid / mL obtained by adding the liposome dispersion at either 9% by volume at 8 74

PAGE 75

mg / mL lipid loading or at 18% by volume wit h 16 mg / mL lipid loading. The data shows minimal improvement in drug removal in both cases despite doubling of lipid loading. A simple mass balance can be used to facilit ate quantitative comparisons between the expected increase due to double lipid loadi ng and the observed behavior. For any mixture of colloids in equilib rium with free drug, the partition coefficients between the bound and the free concentration can be obtained from uptake data for each colloid separately, and the uptake fo r the mixture can then be obt ained, assuming that no interactions are taking place within the system (discussed in Chapter 2). The uptake in the mixture is given by n n 2 2 1 1 mix mix1 ... 111 (3-1) where mix is the fraction of dr ug bound in the mixture and 1, 2, etc., are the fractions of drug bound in the components of the mixture measured s eparately. It must be ensured that all the uptakes in the above equation correspond to identical free drug concentration. This equation was solved to obtain the fraction of drug bound for 50:50 DMPC:DOPG liposomes at 0.72 mg lipid/m L in human serum samples, and the results are also shown in Figure 3-2. The dispar ity between the predicted behavior in a system lacking protein-liposome interactions and the experimental data is most likely due to saturation of binding sites on liposomes by pr oteins, which implies a need for shielding of those sites by inclusion of PEG into the liposomes. 3.3.3 Effect of Vesicle Size on Amitripty line Sequestration Any foreign particle introduced into the body will be naturally removed by the reticuloendothelial system (RES). For a most effective drug detoxification system, blood 75

PAGE 76

stream circulation times should be maximi zed. A number of fa ctors impact liposome circulation time, including lipos ome size, lip id type and composition, and modifications such as PEG incorporation. Literature i ndicates that liposomes with sizes ranging from roughly 100-200 nm may be best for in vivo applications [26,90]. In addition, commercial liposome formulations such as CAELYX and DOXIL r eport average vesicle sizes of 100 nm. In view of the importance of the liposome size on circulation times, we chose to investigate the uptake depende ncy on size using 50:50 DMPC:DOPG liposomes in PBS solutions of AMI. Drug upt ake values for sonicated (40-45 nm) and extruded (100 and 284 nm) liposomes are shown in Figure 3-3. All three liposome sizes showed maximum uptake values of approxim ately 99%, and the drug uptake seems to be independent of liposome size. This result shows that the curvature of the lipid bilayers does not impact drug sequestration. In the results reporte d below, both 100 nm and 40-45 nm liposomes were used. 3.3.4 Liposomes Incorporated wi th Polyethylene Glycol (PEG) A common method used to overcome the effe ct of protein-liposome interactions and liposome scavenging by the RES is incorpor ation of PEG into the lipid membrane. Such liposomes have been used for a variety of applications and are currently incorporated into commercial drug delivery formulations such as CAELYX and DOXIL. Dhanikula et al. have used PEG coated nanocapsules composed of triglycerides for detoxification purposes, alt hough their proposed methods of drug-particle interaction were pH and concentration gradients rather than charge-charge interactions, as in our case [24]. Pegylation of liposomes can be achieved by mixing lipids that have covalently attached PEG chains into the lipid mixture. Here, lipids modified with PEG were 76

PAGE 77

incorporated into neutral and pr edominantly anionic lipos omes at various molar ratios to assess the affect of PEG on in vitro drug remo val. Cholesterol (CH) was also included in the neutral liposome formulation for a sc reening of its potential effects on drug removal. To achieve incorporation of PEG, DPPE-mPEG-2000 was chosen due to its negative charge and PEG chain lengt h. It was hypothesized t hat lipids containing PEG chains of 2000 units would minimize protei n interactions with the liposomes without impacting drug binding. DPPE-mPEG-2000 was added at 5% by mole based on results from several studies indicating long circul ating properties for similar formulations [24,87,96-98]. In addition, 15% by mole inco rporation was also tested to explore the effect of larger proportions of polymer in the liposome bilayer. A number of researchers have explored th e configuration of the surface of pegylated liposomes. The approx imate thickness of the surface coating for PEG chains of 2000 kD is 5 nm, which is much less t han typical liposome diameters observed here [29]. The area of coverage for one PEG2000 chain on the surface of a liposome was approximated to be 2.6 nm2 by Rex et al. [99]. Addi tionally, they performed MonteCarlo simulations which suggested that a large portion of the liposome surface was uncovered by PEG-2000 polymers at 2-10% addition of PEG modified lipid (molar basis).Yoshida et al. also conducted simulati ons to estimate surface coverage and concluded that approximately 45% of the surface would be covered at 5% addition of PEG-2000 [95]. As far as the conformation of the PEG chains is concerned, Moghimi and Szebeni suggest that at very low covera ges (below 5%), t he polymers exhibit a mushroom-like behavior with a lack of extens ion into the aqueous bulk phase [100]. At moderate to high coverages (5 % or more), a brush-like configuration is assumed. 77

PAGE 78

Gbadamosi et al. showed that liposomes with polymers in the brus h-like configuration were much less susceptible to phagocytosis by macrophage type cells than liposomes with polym ers in the less extended mushroom configuration [101]. Thus, the inhibition of phagocytosis by the addition of PEG is not si mply a function of the surface coverage, but is more directly affected by the PEG confi guration, which in turn is directly affected by surface coverage. The studies cited above clearly prove that 5-15% pegylation significantly reduces protein binding, but t he effect of the pegylation on drug binding needs investigation, and this issue is discussed below. In addition to DPPE-mPEG-2000, DOPG wa s used as the primary component in the 95:5 and 85:15 DOPG:DPPE-mPEG-2000 lipos ome suspensions in view of its affinity for AMI due to electr ostatic interactions (see Chapt er 2 and [71]). Also, drug uptakes were measured for pure DOPG liposomes to g auge the effect of polymer inclusion on drug binding. Additionally, predominantly neutral pegylated liposomes were prepared (55:15:30 DMPC :DPPE-mPEG-2000:CH) to demon strate the effect of lack of charge on drug binding pr operties. All reported final lipid concentrations refer to phospholipid concentrations and do not include CH. 3.3.4.1 Amitriptyline removal from human serum by various liposome formulations To compare the drug uptake performance of several liposome formulations in human serum, liposomes were made via soni cation (see Section 3.2.2) with lipid concentrations of 0.72 mg lipid/ mL. AMI uptake for pure human serum without liposomes, 95:5 and 85:15 DOPG:DPPE-m PEG-2000 liposomes, 50:50 DMPC:DOPG liposomes, and 55:15:30 DMPC :DPPE-mPEG-2000:CH liposomes is plotted as a function of final AMI concentration in Figure 3-4. Clearly, the liposomes including 78

PAGE 79

DMPC and CH have a much lower binding c apac ity than the other th ree formulations. By comparing the binding capacity of a mixture of the 55:15: 30 DMPC:DPPE-mPEG2000:CH liposomes and serum proteins with serum pr oteins alone, it is clear that they fail to reduce the free drug concentration by any appreciable amount. The results also suggest that the 95:5 liposomes are slightly superior to the other two types. Figures 3-5 and 3-6 show the 50:50 DMPC:DOP G and the 85:15 DOPG:DPPE-mPEG-2000 liposome uptake values, respectively, com pared to the 95:5 DOPG:DPPE-mPEG-2000 values, along with the 95% CI for each dat a set. Based on the CI, the 95:5 and 50:50 liposomes have significantly different uptake behaviors at most AMI concentrations, but behave similarly for final concentrations of less than 0.10 M. The fractional uptake reduces with increasing concentration for both 50:50 and the 95:5 liposomes, but the reduction is less for the 95:5 liposomes. Th e most plausible explanation for this behavior is protein-liposome interactions in both cases, which saturate the liposome surface and limit the number of adsorption sites available to the drug. The PEG modification seems to decrease this effe ct, however, causing a more gradual decrease in drug binding in the case of the 95:5 liposomes. In the case of the 85:15 liposomes, a less dr astic decrease in drug affinity is again seen, but the uptake values as a whole are lower than the 95:5 liposomes. The polymer inclusion has decreased the protein-liposom e interactions as hoped, but has also affected the drug-liposome affinity. The ex tra 10% by mole of DPPE-mPEG-2000 is actually a much larger increase by weight, as the ratio of formula weights for DOPG and DPPE-mPEG-2000 is roughly 4:1. The amount of negative charge in the system has therefore been decreased substantially when comparing the 95:5 and 85:15 79

PAGE 80

DOPG:DPPE-mPEG-2000 liposomes. Clearly, the increase in negative charge for the 95:5 liposomes has led to an increase in drug uptake. It is, however, unclear whether the presence of the P EG affe cts the drug uptake directly in any manner. To address this issue, AMI uptake was also m easured for 95:5 and 85:15 DOPG:DPPE-mPEG2000 liposomes at 0.72 mg lipid / mL in buffer solutions with out proteins present. The goal was to use the uptake values for one syst em to scale up to the other system, as discussed in Section 3.3.2. In this case, the method was slight ly adapted to probe for the effect of changing the net charge of the system. The ratio of net charge for 95:5 liposomes compared to the 85:15 liposomes (also the ratio of the total moles of lipid in the systems) is 1.217. Based on the assumptions that increased drug uptake would be directly proportional to increased charge and that no other factors played a role in binding, the following equation was derived for the partition coefficient for the drug in liposomes with increased charge: 15:85 15:85 5:95 5:951 217.1 1 (3-2) The fraction of drug bound to the 95:5 liposomes was obtained from the above equation, and the results, al ong with the uptake data for both types of liposomes in buffer, are shown in Figure 3-7. The 95:5 DOPG:DPPE-m PEG-2000 liposomes took up more drug than the scale up equation predict ed, suggesting a direct effect of PEG on drug binding. The 5% PEG liposomes sequestered more drug than predicted, suggesting a negative correlation between in creased PEG-lipid percentage and drug binding. This effect could be due to steric interactions which limit the number of sites on liposomes that are available for drug bindi ng. Alternatively, the high PEG percentage 80

PAGE 81

could be im pacting the structure of the lipid bilayer leading to a reduction in drug sequestration in the bilayer, which is another potential mechanism for drug uptake by liposomes. In either case, a factor other than electrostatics seems to be affecting the binding between the liposomes and the drug, and a 5% inclus ion of lipids modified with PEG was nearly optimal for minimization of liposome-protein interactions while maximizing drug-liposome affinity. The average difference in fractional drug uptake between the 85:15 and 95:5 DOPG:DPPE-mPEG-2000 formulations in Figure 3-6 is roughly 2-3%, which may seem insignificant. However, it is important to not e that the goal of lipos ome injections will be to reduce the free drug concentration as mu ch as possible because the free drug will bind to the receptors in the body (sodium ion channels for AMI) leading to toxic effects. Since the fractional uptake values are very close to 100%, a difference in 2-3% in fractional uptake could lead to a very large percentage decrease in the free drug concentration. To better illustrate the effe ct of liposomes on overdose treatment, the fractional drug uptake by serum is com pared with that for 95:5 DOPG:DPPE-mPEG2000 liposomes in serum in Figure 3-8. Se rum binds to about 92% of the drug, leading to 8% free drug, while liposomes and seru m together bind to about 98% of the drug, leading to 2% free drug. Thus, liposome a ddition reduces the free drug from 8 to 2%, which represents a 75% reduction. Similarly, the reduction in total uptake by 2-3% for the 85:15 liposomes results in a free drug reduction of around 30-50%, as opposed to 75%. The AMI uptake by 95:5 DOPG:DPPE-mPEG -2000 liposomes is thus significantly superior to the other system s studied here and all previous ly published data. More importantly, the variability over the relev ant initial concentration range of 1-10 M has 81

PAGE 82

been drastically reduc ed with the inclusion of PEG chains, making these treatments more suitable clinically. Predicted upt ake values for 95:5 DOPG:DPPE-mPEG-2000 liposomes in human serum were again calcul ated using Equation 31, and are also shown in Figure 3-8. Although the drug concentration dependenc y of the uptake values has been reduced, the failure to reach the predicted values still shows evidence for site occupation by proteins. However, the di fference between the predictions for the pegylated liposomes is much less than the difference for the unpegylated liposomes seen in Figure 3-2. Clearly, strong evidence exist to conclu de that the 95:5 DOPG:DPPE-mPEG-2000 vesicles are superior to the 50:50 DM PC:DOPG, 85:15 DOPG:DPPE-mPEG-2000, and 55:15:30 DMPC:DPPE-mPEG-2000: CH vesicles. To determi ne if the effect was actually a function of polymer inclusio n and increased charge or simply increased charge, pure DOPG liposomes were test ed and compared to 95:5 DOPG:DPPE-mPEG2000 liposomes in Figure 3-9. As shown in the figure, the uptake values for both systems in human serum were almost equal. This shows that perhaps the reduction in protein binding is balanced by the reduction in charge leading to similar drug uptake by these systems. To determine the effect of pegylation, additional experiments were conducted to measure drug uptakes by both pure DOPG and 95:5 DOPG:DPPEmPEG-2000 liposomes in PBS (without any pr oteins) and results were compared to those reported in Figure 3-9 for uptake in seru m (with proteins) in the following section. 3.3.4.2 Assessment of protei n-liposome interactions To assess the affect of proteins on dr ug binding to lipos omes with and without PEG, the summing of partition coefficient s was again used (Equation 3-1). The predictions were made for drug uptake in mixt ures of proteins and liposomes, and were 82

PAGE 83

based on uptake experiments done with eac h component. Figure 3-10 shows the drug uptake for 95:5 DOPG:DPPE-mPEG-2000 liposomes in serum and PBS, along with the predictions for a mixture of liposomes and se rum. Figure 3-11 shows the same for pure DOPG liposomes. Both liposome types were made via sonication (Section 3.2.2). For both liposomes, the drug uptake was lower than the predicted values. The fact that the fractional uptake was lower in mixtures of liposomes and serum than predicted suggests that proteins must be bindi ng directly to the liposomes leading to a reduction in drug binding in serum. To dete rmine the effect of pegylation on protein binding, it is instructive to compare the magnitude of differences betw een the predictions and the experimental results for both pegylated and unpegylated liposomes. This difference is smaller for the pegylated liposomes (Figure 3-10), with overlappin g behavior at small concentrations. Interestingly, the drug upt ake by the pegylated liposomes in PBS is almost concentration independent, whereas t he pure DOPG liposomes show a clear reduction in drug uptake with increasing concent ration. It is possible that the PEG chains provide additional sites for drug bindi ng through non specific adsorption and this eliminates the saturation effects responsible for the decrease in fractional uptake with increasing concentration evident in all unpegylated systems. Regardless, this effect sharply contrasts the results in Section 3.3.4.1, which suggest a negative effect of PEG on drug uptake. We speculate that the presence of PEG does provide additional binding sites but 15% PEG reduces binding of the drug to liposomes due to effects discussed above, as well as to possible bila yer disruptions, and this reduction offsets the binding of the drug to the PE G. Again, this is a speculat ion and more investigations are needed to validate this hypothesis. 83

PAGE 84

Based on the in vitro drug bindin g data reported above, the 95:5 DOPG:DPPEmPEG-2000 liposomes seem best suited for overdose treatment but the pure DOPG systems are almost equally effective. Consequently, it is vital to consider in vivo effects as well, as they may be much more important in determining whether or not polymer inclusion is necessary. Studies in literature suggest that liposome removal by the RES to the liver is imminent without polymer shielding of proteins [30]. In the absence of pegylation, the liposomes along with the bound drug may enter the liver and spleen very quickly and fragment, leading to the possible release of the dr ug back into circulation. It is necessary to ensure that the drug is re leased back into the circulatory system at a rate that does not cause toxicity, and so the addition of PEG will most likely be necessary for in vivo applications. 3.3.4.3 Increased lipid loading fo r 95:5 DOPG:DPPEmPEG-2000 liposomes In Section 3.3.2, data was presented to show that in creased lipid concentrations were ineffective at substantially increas ing drug sequestration for 50:50 DMPC:DOPG liposomes in human serum. The 50:50 lipos omes quickly became ineffective as the AMI concentration increased, making it very difficult to increase drug binding over a wide range of AMI concentrations. On the contrary, 95:5 DOPG:DPPE-mPEG-2000 liposomes appear to be less affected by prot ein interactions, and were subsequently tested for AMI binding at double the or iginal lipid loading of 0.72 mg / mL. The liposomes were made via extrusion with 100 nm memb ranes. Figure 3-12 shows drug uptake as a function of final AMI concentration for human serum without liposomes and for 95:5 DOPG:DPPE-mPEG-2000 liposomes loaded at 1.44 mg lipid / mL in human serum. The lipid concentration in the liposome dispersion was 8 mg lipid / mL, and the dispersion was then added to the serum samples at 18% of the final soluti on volume. Also shown are 84

PAGE 85

the predicted values for the liposome-serum mi xture based on Equation 3-1. The higher loading resulted in consistent uptake values of nearly 99% across the drug concentrations tested. Additionally, t he uptake values were nearly equal to the predicted values, point ing to a lack of binding inhibition by the proteins. By comparing Figures 3-2, 3-8, and 3-12, one can clearly see that the pegylated liposomes at higher lipid loadings are binding to t he drug as effectively as they do in buffer solutions with no proteins present. Protein binding to the liposome surface has therefore been reduced to a level that is insignificant with respect to drug removal. Although the data appears to suggest that a small drug concentration r ange was tested, the actual initial AMI concentrations ranged from 1 to 20 M. The liposomes sequeste red high proportions of the drug and made the final drug concent ration range except ionally small. The results are encouraging when consi dering the percentage reduction in free drug concentration. Figure 3-13 compar es the percentage reduction achieved with three liposomal systems at initial AM I concentrations of 3, 6, and 9 M. The magnitude of the free drug reduction has risen to almost 90% with 95:5 DOPG:DPPE-mPEG-2000 liposomes at 1.44 mg lipid / mL for total concentrations as high as 9 M. To achieve a comparable lipid concentration in the bloodstr eam, roughly 5 g of lipid would need to be administered to a patient, which is half of t he amount of spherulites composed of similar materials that are claimed to be safe for intravenous admini stration [25]. A 90% free drug reduction implies a subst antial drug-liposome affinity and a high degree of binding selectivity despite the presence of serum proteins. While these values may not correlate directly to in vivo applications due to the presence of the immune system and other factors, a free drug reduction in the blood stream of even 20% could be significant 85

PAGE 86

[88]. Thus, the systems tested he re have a high likelihood of reducing toxicity in an overdosed patient. The magnitude of the free drug reduction is not the only factor that should be considered from Figure 3-13. The free dr ug reduction dropped from 55% to 45% for the 50:50 DMPC:DOPG liposomes, and from 75% to 68% for the 95:5 DOPG:DPPEmPEG-2000 liposomes at 0.72 mg lipid / mL. Such declines in reduction show evidence for site saturation at physiologically relev ant concentrations. The 95:5 liposomes with more lipid showed no decline over the entire range tested. For the purposes of drug overdose treatment, the concentration indep endent behavior is enormously beneficial in clinical applications so that the same treatment would be equally effective in all overdosed patients. The 95:5 DOPG:DPPE-mPEG-2000 lip osomes at 1.44 mg lipid / mL show promise as a detoxifying agent. In addition to high binding efficacy, these systems are likely to circulate in the body for an extended period of at least a day. The liposomes will then accumulate in the liver, spleen, lungs, and other organs, and be eventually broken down by the liver into their phos pholipid components. The breakdown of the liposomes will possibly release some drug back into the circulation but this amount will likely be significantly less than that during overdose be cause of the metabolization of the drug during the 24 hour period in which the liposom es circulate in the body. These issues are explored through pharmacokinet ic modeling in Chapter 7. 3.3.5 Drug Uptake by Stored Liposomes AmBisome is an anti-fungal liposome formulation supplied as lyophilized powder, which is reconstituted with sterile wate r prior to i njection. Although this method increases storage time and could possibly be used with the liposomes tested here for 86

PAGE 87

drug overdose treatment, a significant amount of lag ti me would be necessary for pharmacy reconstitution. Time is a key i ssue during drug overdose, and so the storage method for CAELYX and DOXIL s eemed to be the most promisi ng method to test. Both are liposomes for drug delivery to tumors, a nd both include PEG in their formulations for increased circulation time, as do the liposomes tested in the present study. They are stored as concentrated liposom e dispersions at temperatur es of 2-8C for extended periods of time (greater t han one month), although their exac t shelf lives are unknown. Just prior to injection, both products are d iluted with 5% dextrose solutions. To explore the effect of shelf life on the binding c apacity of liposomes, 95:5 DOPG:DPPE-mPEG2000 and pure DOPG liposomes were prepa red via extrusion through 100 nm membranes at concentrations of 0.72 mg lipid / mL. AMI uptake by the liposomes was measured from the day of preparation to day 34. Most readings were taken in 7 day increments. Figures 3-14 and 3-15 show drug uptake as a function of time for pure DOPG and 95:5 DOPG:DPPE-mPEG-2000 liposomes, respec tively. The results show that liposome dispersions, especially those including PEG, conti nued to effectively remove AMI from human serum after being stor ed for several weeks. Although these experiments differed from actual clinical conditions and provided only a small sample, the preliminary assumption is that thes e liposome dispersions could be prepared and stored at 2-8C for at least one month for drug overdose treatment. 3.4 Conclusions In this chapter, liposomes compos ed of DOPG, 50:50 DMPC:DOPG, 95:5 and 85:15 DOPG:DPPE-mPEG-2000, and 55:15:30 DMPC:DPPE-mPEG-2000:CH were tested for their ability to sequester AMI in PBS and in the presence of human serum. 87

PAGE 88

Factors such as lipos ome size, lipid load ing, PEG inclusion, and storage were analyzed for their effects on drug bi nding. Uptake of drug by 50:50 DMPC:DOPG liposomes was not enhanced by increased lipid loading and remained at 35-70% in human serum. Liposome size was shown to have no effect on drug-liposome affinity, which implies that size can be chosen solely for in vivo considerations. When considering liposomes incor porated with PEG, those composed of predominantly anionic lipids with a small am ount of PEG-modified lipids were shown to sequester the drug from human serum most effectively. 95:5 DOPG:DPPE-mPEG2000 liposomes at 0.72 mg lipid / mL reduced the free drug c oncentration of AMI in human serum by 65-75%. 85:15 DOPG:D PPEmPEG-2000 liposomes at 0.72 mg lipid / mL appeared to have too much PEG shield ing and not enough negative charge, as they reduced the free drug conc entration by only 30-50%. Thus, the optimal amount of PEG-modified lipid to be incorporated into liposom es to effectively shield proteins while also allowing diffusion and binding of the drug to the charged lipid membrane appeared to be 5%. The best results were obtained for 95:5 DOPG:DPPE-mPEG-2000 liposomes loaded at 1.44 mg lipid / mL, which reduced the free AMI concentration in human serum by nearly 90% across a wide range of init ial drug concentratio ns. Stability and continued effectiveness during st orage are also important considerations for clinical applications, and preliminary storage te sts performed on pure DOPG and 95:5 DOPG:DPPE-mPEG-2000 liposomes at 0.72 mg lipid / mL provided evidence for sustained drug binding effectiveness after being stored at 2-8C for at least 34 days. 88

PAGE 89

Overall, the results of th is chapter indicate that liposomes composed of 95:5 DOPG:DPPE-mPEG-2000 lipids at a concentration of 1.44 mg lipid/ mL sequester extremely large amounts of fr ee AMI from h uman serum. Furthermore, these systems may be stored at 2-8C for extended periods of time with little to no effect on drug uptake properties. Thus, these systems seem highly suitable for conducting animal and human trials for overdose treatment of AMI. 89

PAGE 90

A B CD E Figure 3-1. Size distributions of liposom es from dynamic light scattering for (a) sonicated 85:15 DOPG:DPPE-mPEG2000, (b) sonicated DOPG, (c) sonicated 55:15:30 DMPC:DPPE-m PEG-2000:CH, (d) extruded 95:5 DOPG:DPPE-mPEG-2000, and (e) extruded 50:50 DMPC:DOPG. 90

PAGE 91

Figure 3-2. Percent amitriptyline uptak e from human serum by 50:50 DMPC:DOPG liposomes at lipid loadings of 0. 72 mg lipid/mL and 1.44 mg lipid/mL; prediction for 50:50 liposomes in human serum based on 50:50 uptake in buffer and serum uptake using Equation 3-1. Key: 50:50 DMPC:DOPG at 0.72 mg/mL ( ); 50:50 DMPC:DOPG at 1.44 mg/mL added at 9% by volume ( ); 50:50 DMPC:DOPG at 1.44 mg/m L added at 18% by volume ( ); prediction for 50:50 DMPC:DOPG at 1.44 mg lipid/mL in human serum ( ). Figure 3-3. Percent amitriptyline uptake from buffer by 50:50 DMPC:DOPG liposomes of various sizes (40-45 nm, 100 nm, 284 nm) at a lipid loading of 0.72 mg lipid/mL. Key: 284 nm ( ); 100 nm ( ); 40-45 nm ( ). 91

PAGE 92

Figure 3-4. Percent amitriptyline uptake from human serum samples without liposomes and from human serum samples with 95:5 and 85:15 DOPG:DPPE-mPEG2000 liposomes, 50:50 DMPC:DOPG li posomes, and 55:15:30 DMPC:DPPEmPEG-2000:CH liposomes at lipid loadings of 0.72 mg lipid/mL as a function of final amitriptyline concentrati on. Key: human serum (+); 95:5 DOPG:DPPE-mPEG-2000 (X); 85 :15 DOPG:DPPE-mPEG-2000 ( ); 50:50 DMPC:DOPG ( ); 55:15:30 DMPC:DPPE-mPEG-2000:CH ( ). Figure 3-5. Percent amitriptyline upt ake from human serum by 95:5 DOPG:DPPEmPEG-2000 liposomes ( ) and 50:50 DMPC:DOPG liposomes ( ) at lipid loadings of 0.72 mg lipid/mL versus fi nal amitriptyline concentration with 95% CI (dashed lines). 92

PAGE 93

Figure 3-6. Percent amitriptyline upt ake from human serum by 95:5 and 85:15 DOPG:DPPE-mPEG-2000 liposomes at lipid loadings of 0.72 mg lipid/mL versus final amitriptyline concentration with 95% CI (dashed lines). Key: 95:5 DOPG:DPPE-mPEG-2000 ( ); 85:15 DOPG:DPPE-mPEG-2000 ( ). 93 Figure 3-7. Percent amitriptyline uptak e from buffer by 95:5 and 85:15 DOPG:DPPEmPEG-2000 liposomes at lipid loadings of 0.72 mg lipid/mL versus final amitriptyline concentration with 95% CI (dashed lines); prediction for 95:5 uptake based on 85:15 uptake using Equ a tion 3-2. Key: 95:5 DOPG:DPPEmPEG-2000 ( ); 85:15 DOPG:DPPE-mPEG-2000 ( ); prediction for 95:5 DOPG:DPPE-mPEG-2000 ( ).

PAGE 94

Figure 3-8. Percent amitriptyline uptake by human serum and by mixtures of human serum and 95:5 DOPG:DPPE-mPEG-2000 lipos omes at a lipid loading of 0.72 mg lipid/mL versus final amitriptyline concentration with 95% CI (dashed lines); prediction for 95:5 uptake in se rum based on 95:5 uptake in buffer and serum uptake using Equation 3-1. Key: 95:5 DOPG:DPPE-mPEG-2000 ( ); human serum ( ); prediction for 95:5 DOPG:DPPE-mPEG-2000 ( ). Figure 3-9. Percent amitriptyline upt ake from human serum by 95:5 DOPG:DPPEmPEG-2000 liposomes and pure DOPG lipos omes at lipid loadings of 0.72 mg lipid/mL versus final amitriptyli ne concentration. Key: 95:5 DOPG:DPPEmPEG-2000 ( ); DOPG ( ). 94

PAGE 95

Figure 3-10. Percent amitriptyline uptake from human serum and buffer by 95:5 DOPG:DPPE-mPEG-2000 liposomes at lipid loadings of 0.72 mg lipid/mL versus final amitriptyline concentration with 95% CI (dashed lines); prediction for 95:5 uptake in serum based on 95:5 uptake in buffer and serum uptake using Equation 3-1. Key: human serum ( ); buffer ( ); prediction for 95:5 DOPG:DPPE-mPEG-2000 in serum ( ). 95 Figure 3-11. Percent amitriptyline uptak e from human serum and buffer by pure DOPG lipos omes at lipid loadings of 0.72 mg lipid/mL versus final amitriptyline concentration with 95% CI (dashed lines ); prediction for DOPG uptake in serum based on DOPG uptake in buffe r and serum uptake using Equation 31. Key: human serum ( ); buffer ( ); prediction for DOPG in serum ( ).

PAGE 96

Figure 3-12. Percent amitriptyline uptake by human serum and by mixtures of human serum and 95:5 DOPG:DPPE-mPEG-2000 lipos omes at a lipid loading of 1.44 mg lipid/mL versus final amitriptyline concentration with 95% CI (dashed lines); prediction for 95:5 uptake in se rum based on 95:5 uptake in buffer and serum uptake using Equation 3-1. Key: 95:5 DOPG:DPPE-mPEG-2000 ( ); human serum ( ); prediction for 95:5 DOPG:DPPE-mPEG-2000 ( ). Figure 3-13. Free drug concentration reduction of amitriptyline in human serum by 95:5 DOPG:DPPE-mPEG-2000 liposomes at loadings of 1.44 ( ) and 0.72 mg lipid/mL ( ) and 50:50 DMPC:DOPG liposomes at 0.72 mg lipid/mL ( ). Free drug reduction was calculated bas ed on the difference between free drug in human serum samples and free dr ug in human serum samples mixed with liposomes. 96

PAGE 97

97 Figure 3-14. Percent amitriptyline uptake from human serum by pure DOPG liposomes at a lipid concentration of 0.72 mg lipid/mL and an in itial drug concentration of around 2.7 M (corrected for dilution) ve rsus days stored. Data are reported as mean SD with n = 2. Figure 3-15 Percent amitriptyline uptake from human serum by 95:5 DOPG:DPPEmPEG-2000 liposomes at a lipid conc entration of 0.72 mg lipid/mL and an initial drug concentration of around 2.7 M (corrected fo r dilution) versus days stored. Data are reported as mean SD with n = 2.

PAGE 98

CHA PTER 4 BINDING OF IMIPRAMINE, DOSULEPIN, AND OPIPRAMOL TO LIPOSOMES FOR OVERDOSE TREATM ENT 4.1 Introduction In Chapters 2 and 3, pegylat ed, anionic liposomes were proven to be effective binding agents for amitriptyline (AMI) and no rtriptyline. Three mai n goals were addressed in this chapter. Fi rst, we explored binding of other tricyclic antidepressants commonly involved in overdoses to determi ne whether pegylated liposomes could be useful as a general treatment strategy for drug overdose. Imipramine (IMI) and dosulepin (DOS), both weak bases existing in the predominantly charged state at the physiological pH of 7.4, were studied. Opipramol, a diprotic compound with pKa values of around 4 and 8.1 [102-104], was also studi ed. At a pH of 7.4, one opiprmaol protonation site is 83% protonated while the other site is uncharged. Opipramol thus gave us the opportunity to test the effect iveness of liposomes at binding an important overdose drug and the chance to c onfirm the importance of electrostatic effects for drug binding to pegylated, charged liposomes. Next, we explored the mechanisms of binding, particularly focusing on the importance of electrostatic effects, and also investigating the role of the PEG chain length on drug binding. It is im portant to understand the role of electrostatic interactions in drug binding because several overdos e drugs are charged at physiological conditions. Electrostatics were studied using opipramol, as mentioned above. The role of PEG chain length is also very import ant because PEG chain lengths have been shown to affect the biodistribution and circulat ion time of polymer coated liposomes [31]. Thus, it was important to determine whether PEG chain length impacted drug binding so 98

PAGE 99

that an optimum PEG chain length based on both pharmacoki netics (circulating times) and equilibrium drug binding could be determined. Finally, we wished to compare the binding efficacy of liposomes to other targets inside the body to determine the binding potential of the liposomes in vivo. It is common knowledge that most tricyclic antidepressant drugs are hi ghly bound to proteins once inside the blood stream [25, 85], and in our prior work, we have shown that liposomes can compete favorably with plas ma protei ns for drug uptake ([71] and Chapters 2 and 3). However, there are acidic lipids in the body similar in structure to lipids considered here and so it is extremely important to co mpare the binding affinity of liposomes with other acidic lipid containing targets in side the human body. A brief discussion on the time scales relevant for drug redistribution fr om the organs to the blood compartment is also presented as evidence for the clinic al feasibility of overdose treatment with liposomes. A more detailed pharmacokinet ic analysis is presented in Chapter 7. All of the binding experiments performed in this chapter were done in vitro in human serum samples or PBS at a pH of 7.4. It is misleading to measure the absolute binding of drug molecules to liposomes in the presence of proteins and assume the effective free drug concentration has been reduced by the same margin as in PBS (without proteins). For that reason, we have measured the bindi ng of drugs to serum proteins in human serum, followed by mixtures of human serum and liposomes. Protein binding has already been accounted for w hen free drug concentrati on reductions are reported. 99

PAGE 100

4.2 Materials and Methods 4.2.1 Materials Methanol, chloroform, Dulbeccos P hosphat e Buffered Saline (PBS) without calcium chloride and magnesium chloride, am itriptyline hydrochloride, imipramine hydrochloride, and opipramol dihydrochloride were purchas ed from Sigma Aldrich (St. Louis, MO). Dosulepin (dot hiepin) hydrochloride was pu rchased from LGC Promochem (Teddington, UK). Human serum from male plasma and Centriprep YM10 centrifugation filters (10,000 molecular weight cut-off) were purchased from Fisher Scientific (Pittsburg, PA). T he lipids 1,2-dioleoyl-sn-glycero -3-[phospho-rac-(1-glycerol)] (sodium salt) (DOPG), disso lved in chloroform, 1,2-dipalmitoyl-sn-glycero-3phosphoethanolamine-N-[m ethoxy(polyethyleneglycol)-2000] (ammonium salt) (DPPEmPEG-2000), in powder form, and 1,2-dipal mitoylsn-glycero-3-phosphoethanolamineN-[methoxy-(polyethylene glycol)-5000] (ammonium salt) (DPPE-mPEG-5000), in powder form, were purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). A MiniExtruder kit for liposome preparation was also purchased from Avanti Polar Lipids, Inc. 4.2.2 Liposome Prepar ation via Extrusion Liposomes composed of a molar ratio of 95:5 DOPG/DPPE-mPEG-2000 and 95:5 DOPG/DPPEmPEG-5000 were prepared via extrusion. Lipids were combined in their respective molar ratios and then dissolved in a 9:1 mixture (by volume) of chloroform/methanol such that a 12 mg/mL concentration of lipids was obtained. The organic solvent was then evaporated under a st ream of nitrogen. After an even and uniformly dried lipid film was obtained, the dr ied lipid layer was hydrated with PBS. The lipid concentration after hydration was 8 mg lipid/mL PBS for 95:5 DOPG/DPPE-mPEG2000 liposomes or 9.33 mg lipid/mL PBS for 95:5 DOPG/DPPE-mPEG-5000 liposomes. 100

PAGE 101

The mass loading was slightly higher for the PEG-5000 liposomes so that the DOPG content would be identical for both types of liposomes. The lipid suspension was then heated to 30C and stirred for 45 minutes. A fter stirring, the liposome solution was extruded through a 100 nm membrane 13 times. 4.2.3 Liposome Characterization The mean diameters and size distributions of the liposomes were analyzed using a Precision Detectors P DDLS/CoolBatch+90T instrument. The data were analyzed with the Precision Deconvolve32 Program. The measurements were taken at 20C with a scattering angle of 90 using a 683 nm laser source. 4.2.4 Drug Uptake With Liposomes in Human Serum or PBS 95:5 DOPG/DPPE-mPEG-2000 lipos omes were adde d to solutions of IMI, DOS, or opipramol in human serum such that the volume of the liposome dispersion was 9% or 18% of the total solution volu me, giving a final lipid concentration of 0.72 or 1.44 mg lipid/mL. 95:5 DOPG/DPPE -mPEG-5000 liposomes were added to solutions of IMI, DOS, or opipramol in human serum or PBS such that the volume of the liposome dispersion was 18% of the total solution volu me, giving a final lipid concentration of 1.68 mg lipid/mL. The initial drug concentrations for all three drugs varied from around 3-25 M, due to the significance of this range for serum drug concentrations during drug overdose [105]. Control solutions of dr ug in PBS and drug in serum without liposomes were also made to allow for uptake quantific ation and comparison. After being stirred for 15 minutes, the solutions were ultracentrifuged at 5000 rp m for 15 minutes in a vial that contained a YM10 filter (10,000 molecular weight cutoff). Experiments were done in Chapter 2 to ensure that the systems were at equilibri um before filtration. To minimize the effect of any l eaching components from the f ilter, the filters were rinsed 101

PAGE 102

first with DI water and then with PBS at 5000 rp m for 10 minutes prior to their use in these experiments. The c oncentration of the drugs in the filtrate (free drug concentration) and the control solution was detected by measuring UV absorbance at 215 nm after passing the samples through a C 18 column (Waters Co rp., Milford, MA) using an acetonitrile/50 mM NH4H2PO4 solvent mixture in a 35/65 ratio for IMI and DOS and a 30/70 ratio for opipramol. The calibration curves for concentration versus area under the curve were linear with R2>0.99. To ensure t hat all unbound drug was accounted for, solutions of the drugs at va rious concentrations were passed through YM10 filters in a separate test. Small amounts of drug were taken up by the filter, and a linear correction curve was made and used to correct for drug adsorbed to the membranes in subsequent tests. 4.2.5 Data Analysis The data were analyz ed using JMP software developed by SAS. Best fit regression lines were fit to the concentration-dependent uptake data, and 95% confidence intervals (CIs) were displayed to allow for both experimental error analysis and comparison between two or more data sets. The CIs were calculated for mean uptake values using the Students t distribution. 4.3 Results and Discussion 4.3.1 Liposome Characterization The 95:5 DOPG/DPPE-mPEG-2000 and DOPG/DPPE-mPEG-5000 lip osomes extruded with 100 nm membranes had mean di ameters of 118 15 and 112 15 nm, respectively. 102

PAGE 103

4.3.2 Imipramine Uptake The binding of IMI to liposomes was explored for initial drug concentrations ranging from 3 to 22 M, which are relevant to overdose cases. Musshoff et al. [105] reported toxic values in human plasma to be 1.58 to 4.73 M, but observed levels in dead victims from 1.58 to 79 M. IMI percent uptake values for human serum and mixtures of human serum and liposomes ar e plotted as a function of final drug concentration in Figure 4-1. In Figures 4-1 through 4-3, t he y-axis represents x 100%, where (CT-CF)/CT, CT is the total drug concentration, and CF is the free concentration assayed by HPLC. The final or equilibrium, drug concentrations (CF) were used on the x-axis rather than the init ial drug concentrations so that the curves would represent equilibrium binding isotherms. IMI binding to proteins in serum sample s was roughly 80-84%, or equivalently, 1620% of the IMI remained unbound in the solution. These values agree well with the free fractions of 24% and 8-15% reported by R odgers et al. [106] and Kristensen [107], respectively. The total drug binding in the presence of liposomes and proteins rose to 96-98% for 95:5 DOPG/DPPE-mPEG-2000 liposom es at 0.72 mg lipid/mL. Doubling the lipid loading to 1.44 mg lipid/mL caused the drug uptake values to increase to around 98%, while also eliminating t he uptake reduction with increasing drug concentration. The lack of overlap in the CIs for upt akes for the 0.72 and 1.44 mg/mL lipid loadings demonstrates the significanc e of the reduction. The 95:5 DOPG/DPPEmPEG-5000 liposomes at 1.68 mg lipid/mL also reduced the overall drug concentration by 98%. Other than a slight drop in drug uptake at low drug concentrations, the PEG5000 liposomes behaved similarly to the PEG-2000 liposomes (CI overlap). This may 103

PAGE 104

be an important result when final formulations are developed for in vivo use, as formulators could choose polymer chain lengt hs based on pharmacokinetic considerations without the risk of affect ing the drug uptake c apacity of the system. To analyze the suitability of liposomes for overdose treatment, it is instructive to calculate the percentage change in the free drug concentrat ion after liposomes are added to serum. The free drug concentration in serum is about 16-20%, and it reduces to 2% after liposome addition. This imp lies that liposomes reduce the free drug concentration in serum by about 88%. Poisoni ngs occur due to partitioning of the drug into the cardiovascular tissue and central nervous system, which induces ion channel disruption and other effects [2]. Thus, a r eduction in the free drug concentration could shift equilibrium towards the blood stream, drivi ng some of the drug out of the most vital tissues and into the bloodstream and/or other less vital tissues. This is discussed in more detail in Section 4.3.5 and Chapter 7. Pegylated liposomes are likely to be more suitable for overdose treatment because of a reduction in protein binding due to pegylation. The degree of protein binding to liposomes can be indirectly es timated by combining the uptake data from serum with the data from liposomes in buffer, and comparing the result with the case in which liposomes and serum are mixed. If a mi xture contains several types of binding targets, the following mass balan ce can be used to obtain uptake values in mixtures if the uptakes are known for individual components, n n 2 2 1 1 mix mix1 ... 111 (4-1) 104

PAGE 105

where mix is the fraction of dr ug bound in the mixture and 1, 2, etc., are the fractions of drug bound in the components of the mixture measured s eparately, with each being measured at the same final (equilibrium) c oncentration (see Chapters 2 and 3). For instance, 1 and 2 could be the bindings in serum (without liposomes) and liposomes in PBS, respectively, and the above equation c ould be used to predict the uptake in serum after the addition of liposomes, mix. Figure 4-1 shows the projected uptake values for a 1.44 mg/mL loading of 95: 5 DOPG/DPPE-mPEG-2000 liposomes in serum along with the measured values in the magnifi ed region of the figur e. The predicted values are higher than those obtained experimentally, although only slightly. This difference is likely due to pr otein binding to liposomes, and its magnitude is much less than that for unpegylated liposomes, which is expected as pegylation was shown to reduce protein binding to liposomes in Chapter 3. 4.3.3 Dosulepin Uptake DOS uptake versus final DOS concentration is shown in Figure 4-2. Again, the initial drug concentrations te sted ranged from 3 to 25 M, wh ich is the physiologically relevant range. The results in Figure 4-2 show that the serum proteins sequester 8892% of the drug. These values are similar to those for A MI and slightly higher than the values observed for IMI. This is reas onable as DOS and AMI are more similar in structure than AMI and IMI. At a lipid loading of 0.72 mg lipid/mL, the 95:5 DOPG/DPPEmPEG-2000 formulation sequestered almost all of the drug at low concentrations and just above 96% at higher concentrations. When the lipid loading was doubled, the values ranged from almost 100% to about 98.5%. The lack of overlap between the CIs for serum, serum and a 0.72 mg/mL loading of lipids, and serum with 1.44 mg/mL lipids points to a significant effe ct of liposomes on drug uptake. The free 105

PAGE 106

drug concentration reduction from 8-12% to a bout 1.5% after addition of liposo mes at 1.44 mg lipid/mL to serum implies an aver age free drug reduction of about 93%. The 95:5 DOPG/DPPE-mPEG-5000 liposomes at 1. 68 mg lipid/mL behaved exactly as the PEG-2000 liposomes, as was observed for IMI. These results for binding of DOS again point to the effectiveness of the polymer coated liposomes made of a predominantly anionic lipid for minimizing protein-liposome interactions and sequestering multiple weak bases from serum samples. These resu lts also show a lack of impact of the PEG chain length on drug uptake, as the uptakes are similar for the PEG chain lengths of 2000 and 5000 tested here. By again using the uptake data from seru m and liposomes in PBS in combination with Equation 4-1, we predicted the uptake values for 95:5 DOPG/DPPE-mPEG-2000 liposomes at 1.44 mg/mL in serum, and the re sults are shown in the magnified portion of Figure 4-2. The measured values are slightly less than the predicted ones, again demonstrating that while pegylation reduces protein binding to liposomes, it does not completely eliminate the binding. 4.3.4 Opipramol Uptake and the Importa nce of Electrostatic Interactions Unlike IMI and DOS, opipramol is a diprotic drug with pKa values of around 8 and 4, and thus, about 80% of the drug molecules have one site protonat ed at a pH of 7.4 [102-104]. A comparison between the uptake of opipramol with the uptakes of IMI and DOS illustrate the importance of the electros tatic interactions in drug binding to the anionic, pegylated liposomes. An earlier study by Austin et al. [55] involving the net neutral, zwitterionic lipid DMPC pointed to a more favorable thermodynamic interaction between the lipid bilayer and charged drugs versus neutral ones. Here, changes in 106

PAGE 107

partitioning are compared for charged versus partially uncharged drugs for liposomes composed of a net negative lipid, DOPG. Opipramol uptake as a function of final drug concentration is displayed in Figure 43. The total drug concentrations in these experiments ranged from 3 to 23 M. The serum proteins sequestered about 76-82% of the drug from solution, which is lower in comparison to binding for IMI and DOS to se rum proteins. This difference is likely due to reduced electrostatic interactions bet ween the drug and albumin and other serum proteins that possess multiple char ged sites. The 95:5 DOPG/DPPE-mPEG-2000 liposomes added to the serum at 0.72 and 1.44 mg lipid/mL increased uptakes to approximately 88-92% and 92.5-95.5%, respectively The CIs show that the liposomes significantly increase the amount of drug bound in the system. Again, the CIs for both PEG-2000 and PEG-5000 liposomes overlap and imply no drug binding differences between the different PEG chain lengths. The uptake values for the serum and mixtures of serum and liposomes were all lower than those observed for IMI and DOS, where over 99% of the drug molecules were in the protonated form. This confirm ed the hypothesis that charge was playing a role in drug binding. The change in free dr ug concentration remained relatively high at about 76%. A reduced presence of charge-char ge interactions in the system by 20% reduced the free drug concentration reduction by 12% compared to the values seen for IMI and DOS. This proves that electrostatic interactions play an important role in drug sequestration. Furthermore, the result s show that the 95:5 DOPG/DPPE-mPEG2000/5000 liposomes are capable of sequestering drugs that are only partially charged. 107

PAGE 108

4.3.5 In Vivo Considerations: Competition and Pharmacokinetics A nanoparticulate system can only reduce toxic effects from a dr ug overdose if it has a higher affinity for the drug in compar ison to other target sites in the body. Two major binding targets in the body for drugs considered here are the acidic phospholipids and serum proteins. Our data r eported above show that the par tition coefficient for drug binding to liposomes is significantly higher than that for drug-pr otein binding, as evidenced by the fact that a liposome addi tion of 1.44 mg/mL r educes the free drug concentration in serum, which contains about 7% protein by weight, by as much as 90%. Below, we compare the binding capacity of the lipos omes with another important binding target in the body, acidic phospholipids. Acidic phospholipids (APL) are present inside red blood cells and within the cells of all major tissues. Here, we utilize t he approach developed by Rodger s et al. [106] to estimate the equilibrium constant (Ka) for drug-APL binding inside red blood cells. The estimated Ka for drug-APL binding inside red blood ce lls is representat ive of equilibrium constants for drug-APL binding for all ot her tissues in the body [106]. The Ka value for IMI binding to APL in red blood cells c an be calculated from the following equation (Equation 20 in [106]), (4-2) 108

PAGE 109

This equation is based on using the known pa rtition coefficient for the drug in the RBC and t he known composition of the RBC to estimate the partition coefficient for binding to the APL. For details on t he equation and the parameters used for the calculation, see [106], alt hough it should be noted that a bl ood to plasma ratio of 1.1 was used here for IMI [ 108], rather than the value of 1.67 used by Rodgers et al. [106]. The resulting Ka value for IMI is 3.9 g/mg. The a ssociation constant between liposomes and IMI can be calculated based on the data r eported for drug binding to liposomes in PBS at 0.72 mg lipid/mL by using the fo llowing equation (Equation 10 from [106]), Pf f Ku b a, (4-3) where fb is the fraction of bound drug, fu the fraction of unbound drug, and [P] the concentration of APL in mg /g. The results for Ka are shown as a function of concentration in Table 4-1. The mean Ka value for IMI-liposome binding is about 77 g/mg over the relevant therapeutic concentra tion range, which is about 20 times larger than the value of 3.9 g/mg for APL in the r ed blood cells. As the approximate amount of lipids added to an overdosed patient could approach 10 g wit hout toxic effects, the amount of lipids added via liposomes w ould substantially increase the total APL concentration in the blood compartment, whic h has about 1.2 g of APL [25,106]. A human body contains about 90 g of APL, and thus the total drug bound to 10 g of liposomes would far exceed the drug bound to all the APL in the body [106]. It should also be noted that a large fracti on of the APL in the body ar e present in organs with low blood flow and so these APL are not truly av ailable for binding within the time frame relevant to overdoses. 109

PAGE 110

Based on the above data, it is clear that liposome administration into the blood can reduce the free drug concentration in the blood by a factor of about 8. Howev er, overdose treatment requires a s ubstantial reduction in drug conc entration in vital organs such as the heart and central nervous system. If the overdose occurred a long period of time prior to treatment, all tissues would be in equilibrium at t he instance of liposome administration. In this scenario, the tota l change in drug concentration in any tissue would be negligible after the body achieved t he new steady state concentrations due to the large volume of distributions for the dr ugs of interest. Howe ver, most overdose cases do not correspond to this scenario. As shown in Table 4-2, the average times between ingestion and treatment observed in several overdose cases was typically between 2 and 4 hours. The drugs considered in this chapter are consumed orally, in which case the maximum blood levels ar e reached about 2-8 hours after ingestion [2,109]. During overdose, this time will increas e substantially [2,8]. Additionally, several tissues in the body equilibrate rather slowly with the blood compartment. A good first approximation for equilibration times is the product of the partition coefficient of an organ and its volume, divided by the blood flow rate to the organ. In Table 4-3, these quantities have been estimated for many of the vital organs of the bo dy. Many of the organs equilibrate within an hour or so of an IV dose. The muscle, skin, and adipose tissues, which account for almost 80% of the volume of distribution, equilibrate on a slower time scale than most other organs, requiring 4-10 times longe r. If the time required for significant oral absorption and fo r the slow tissues to reach equilibrium are roughly combined, these large tissues will have little drug in them at the time of liposome administration, a ssuming that time lapses between overdose and treatment 110

PAGE 111

are similar to those reported elsewhere [7 ,110-112]. The lipos omes will then remove drug exclusively from vital tissues, which would include the brain an d heart. In fact, in such cases, the large organs could act as sinks and remove excess drug from the blood stream as liposomes redistribute the drug from faster equilibrating organs to slower ones. The drug partitioning into the slowly equilibrating organs could actually speed up after liposome administration due to the decreas e in the partition coefficient between the organs and the whole blood including liposomes. Effectively, the liposomes could act as transporters, removing the drug from the rapidly equilibrating organs and r edistributing it to the slowly equilibrating ones. This method of redistribution is likely the me chanism for overdose treatment by Fab fragments for digoxin and antidepressants, where the amount of Fab fragments administered is far less than the amount needed to bind the amount of drug present in the entire body [4,113-116]. Evidence for improved pharmacodynamic responses due to redistribution was shown in several animal studies, while human studies revealed an increase in tricyclic antidepressant seru m concentrations after protein fragment administration. These pharmacokinetic issu es have previously been included in a detailed mathematical model that is applicable to overdose treatment by any type of particles [117], as well as more quantitativel y addressed for liposomes in Chapter 7. It is also noted that liposome administrati on could increase the hepatic drug clearance due to drug concentration increases in the blood, which could in turn lead to additional benefits for overdose treatment. 4.3.6 Comparisons with Prior Studies Bind ing measurements in the presenc e of human serum for 95:5 DOPG/DPPEmPEG-2000 and 95:5 DOPG/DPPE-mPEG-5000 liposomes demonstrated their 111

PAGE 112

effectiveness at sequestering the weak bas es DOS and IMI, and the diprotic drug opipramol. Figure 4-4 summarizes the result s from the present and previous chapters by showing the average free drug concentra tion reductions over the concentration ranges tested for AMI, IMI, DOS, and opipramol. The liposomes were able to reduce the free drug concentration relativ e to protein samples without liposomes by 84-93% for the charged species and 76% for the partia lly charged opipramol. In Chapter 2, nortriptyline showed similar uptake results to that of AMI in PBS solutions by similar liposomes, and so its uptake in serum is also expected to be similar to that of AMI. Thus, the liposomes explored in this study have been shown to be effective at sequestering a large fraction of drug under ph ysiological conditions, which includes protein binding. Such result s demonstrate the feasibility of using a single detoxification system in a wide variety of circumstances. In Chapter 5, the use of such liposomes is expanded to other classes of drugs as well. Table 4-4 compares the uptake efficacy of the formulations described here with prior studies on drug sequestr ation by nanoparticles. Dhanikula et al. [24,25,70] reported significant in vitro uptake of AMI by spherulites composed of soy phosphatidylcholine and cholesterol with a 100 mM citrate buffer internal phase. They reported an AMI sequestration of 97.3 0.4% in the presenc e of 3% albumin, which is similar to the results obtained with the li posomes tested here. However, only 3% albumin was used, rather than human serum samples containing all serum proteins. Furthermore, their initial drug concentrati on was only 0.20 M, which is within the therapeutic range for AMI but fa r below the toxic level of 1-3 M. Also, the mass loadings in all of their studies were almost twice as high as those used here. The in 112

PAGE 113

vitro studies of Dhanikula et al. were s upplemented with in vivo studies on is olated rat hearts. They showed that infusion of the na noparticles leads to a significant reduction in toxicity in isolated rat hearts that had been previously injected with relatively high amounts of AMI. Varshney et al. [19] also used nanocarriers in the form of nanoemulsions composed of ethyl butyrate, fatty acids, and poloxamer to sequester bupivacaine from buffer solutions. They r eported maximum uptakes of about 90% in PBS, but the effect of serum proteins wa s not considered and their total mass loadings (surfactants and oil) were fairly high. As summarized in Table 4-4, several types of particles have shown high drug sequestrations However, pegylated, anionic liposomes seem to have the widest possible spectrum of use and one of the highest binding efficacies for drugs that are mostly pos itively charged under physiological conditions [21-23]. It should be noted that several factors must be taken into account when evaluating different drug detoxification vehicles. Sequestering high amounts of drug at therapeutically feasible mass loadings, causing minimal toxicity, remaining effective in the presence of serum proteins and the reticuloendothelial system (RES), and remaining in circulation for some period of time after admin istration are all of utmost importance. Based on the comparisons shown in Table 4-4 and the well-known biocompatibility characteristics of pegylated liposomes, we believe liposomes to be the optimal vehicle for drug overdose treatment. Additional studies on biodetoxification have been reviewed by Leroux [69]. 4.4 Conclusions In this chapter, polymer shielded liposomes composed o f 95:5 DOPG/DPPEmPEG-2000 and 95:5 DOPG/DPPE-m PEG-5000 lipids were tested as nanocarriers for 113

PAGE 114

the treatment of IMI, D OS, and opipramol overdose. The liposomes proved to be capable of sequestering about 98% or mo re of IMI and DOS and 92.5-95.5% of opipramol. Additionally, the liposomes r educed the free drug concentrations in human plasma by 88-93% in the case of the weak bases IMI and DOS, and 76% for the diprotic drug opipramol (1.44 mg lipid/mL loading). Comparison of the drug binding in PBS with that in human serum showed that the pres ence of the PEG on the liposome surface reduced protein binding to liposomes, but di d not completely eliminate it. These experiments also showed that the length of the PEG chain did not make any detectable difference in uptake. Thus, PEG chain lengt hs of up to 5000 and possibly higher may be chosen if pharmacokinetic parameters dictate their necessity. Calculations suggested that liposomes are approximately 20 times more effective at binding antidepressants than acidic phospholipids already present in the body. Esti mates of the times required for tissues to reach equilibrium also demons trated that liposomes could significantly alter the drug concentrations in vital organs after a drug overdose. Such free drug reductions and calculations lead to the conc lusion that the liposomes tested are prime candidates for treating drug overdoses from a variety of tricyclic antidepressant drugs. 114

PAGE 115

Figure 4-1. Percent imipramine uptake by human serum and by mixtures of human serum and 95:5 DOPG:DPPE-mPEG-2000 liposom es at lipid loadings of 0.72 and 1.44 mg lipid/mL (predicted via E quation 4-1) and mi xtures of human serum and 95:5 DOPG:DPPE-mPEG-5000 lipos omes at a lipid loading of 1.68 mg lipid/mL versus final imipra mine concentration with 95% CI (dashed lines). Key: human serum (X); 95 :5 DOPG:DPPE-mPEG-2000 at 0.72 mg lipid/mL ( ); 95:5 DOPG:DPPE-mPEG-2000 at 1.44 mg lipid/mL ( ); 95:5 DOPG:DPPE-mPEG-5000 at 1.68 mg lip id/mL (O); prediction for 95:5 DOPG:DPPE-mPEG-2000 at 1.44 mg lipid /mL in human serum via Equation 4-1 ( ). 115

PAGE 116

Figure 4-2. Percent dosulepin uptake by human serum and by mixtures of human serum and 95:5 DOPG:DPPE-mPEG-2000 liposom es at lipid loadings of 0.72 and 1.44 mg lipid/mL (predicted via E quation 4-1) and mi xtures of human serum and 95:5 DOPG:DPPE-mPEG-5000 lipos omes at a lipid loading of 1.68 mg lipid/mL versus final dosulepin concentration with 95% CI (dashed lines). Key: human serum (X); 95 :5 DOPG:DPPE-mPEG-2000 at 0.72 mg lipid/mL ( ); 95:5 DOPG:DPPE-mPEG-2000 at 1.44 mg lipid/mL ( ); 95:5 DOPG:DPPE-mPEG-5000 at 1.68 mg lipid/mL (O); prediction for 95:5 DOPG:DPPE-mPEG-2000 at 1.44 mg lipid /mL in human serum via Equation 4-1 ( ). 116

PAGE 117

Figure 4-3. Percent opipramol uptake by human serum and by mixtures of human serum and 95:5 DOPG:DPPE-mPEG-2000 liposom es at lipid loadings of 0.72 and 1.44 mg lipid/mL and mixtures of human serum and 95:5 DOPG:DPPEmPEG-5000 liposomes at a lipid loading of 1.68 mg lipid/mL versus final opipramol concentration with 95% CI (dashed lines). Key: human serum (X); 95:5 DOPG:DPPE-mPEG-2000 at 0.72 mg lipid/mL ( ); 95:5 DOPG:DPPEmPEG-2000 at 1.44 mg lipid/mL ( ); 95:5 DOPG:DPPE-mPEG-5000 at 1.68 mg lipid/mL (O). 117

PAGE 118

Figure 4-4. Free drug concentra tion reductions for amitriptyline, imipramine, dosulepin, and opipramol in human serum by 95:5 DOPG:DPPE-mPEG-2000 and 95:5 DOPG:DPPE-mPEG-5000 liposomes at loadings of 1.44 and 1.68 mg lipid/mL, respectively. Free drug r eductions were calculated based on the differences between free drug concentrations in human serum samples and free drug concentrations in human se rum samples mixed with liposomes. 118

PAGE 119

Table 4-1. Association constant co mparison for liposomes and red blood cells Imipramine Concentration ( M) Ka 2.0 112.7 3.5 40.2 6.5 85.7 9.3 75.3 11.7 79.5 16.1 67.3 Mean a 76.8 KaRBC b 3.9 aAssociation constant for liposomes and imipramine from Equation 4-3. bAssociation constant for red blood cells and imipramine from Equation 4-2. Table 4-2. Time between ingestion and tr eatment for antidepressant overdoses Source Subjects (N) Avg Time (h) [7] 1 4.00 [110] 21 3.05 [111] 3 2.30 [112] 4 4.00 Table 4-3. Pharmacokinetic properties of various organs and imipramine partition coefficients Organ V (L) a Q (L/h) a Kimipramine b KV / Q (h) Type Bone 2.9 41.7 6.6 0.46 Fast Brain 0.4 6.8 7.8 0.45 Fast Gut 1.9 44.8 20.8 0.88 Fast Heart 0.2 16.8 16.9 0.23 Fast Kidney 0.5 48.2 35.0 0.37 Fast Liver 2.6 59.9 32.2 1.38 Fast Lung 0.4 314.6 29.0 0.03 Fast Spleen 0.1 6.8 22.2 0.45 Fast Muscle 28.3 95.1 11.7 3.47 Slow Skin 13.3 19.8 16.0 10.73 Slow Adipose 5.3 23.9 18.8 4.18 Slow aOrgan volume (V) and blood flow rate (Q) from Rodgers et al. [106]. bBlood/tissue partition coefficients (K) calculated by Rodg ers et al. [106] a ssuming an imipramine unbound fraction of 0.24. 119

PAGE 120

120 Table 4-4. Comparison of drug detoxification vehicles Source Vehicle Drug(s) Mass Loading (mg/mL) Maximum Uptake (buffer) Medium Free Drug Reduction in Mediuma Ch. 2, 3, 4 PEG Liposomes Amitriptyline Imipramine Dosulepin 1.68 99.50% Human Serum 88% Ch. 2, 3, 4 PEG Liposomes Opipramol 1.68 96.00% Human Serum 76% [24] Nanocapsules Haloperidol Docetaxel Paclitaxel 3 46.2% 75% 75% 3% Albumin 0% 60.8% 48.6% [25] Spherulites Amit riptyline 2.5 98.10% 3% Albumin 97.30% [70] Spherulites Haloperidol Docetaxel Paclitaxel 3 75.2% 94.4% 91.5% Bovine Serum 38.3% No Data No Data [19] Nanoemulsions Bupivacaine 11 90% No Data No Data [21] Nanocapsules Quinoline Bupivacaine 14 >97% >99% No Data No Data [22] Oligochitosans Amitripty line 4 90.70% No Data No Data [23] Nanoparticles A mitriptyline 1 90% No Data No Data aRefers to the difference between the free drug concentration in the presence of proteins and the free drug c oncentration in the presence of proteins and detoxifying agents.

PAGE 121

CHA PTER 5 BUPIVACAINE BINDING TO PEGYLATED LIPOSOMES 5.1 Introduction In Chapters 2, 3, and 4, pegylated, anioni c liposomes were proven capable of sequestering tricyclic antidepres sants. Local anesthetics represent another class of charged drugs for which adverse reactions are a major concern. The development of an overdose treatment capable of treating mult iple drug classes would greatly increase its utility. We hypothesized that an IV delivery of anionic liposomes after a local anesthetic overdose could reduc e the free drug concentration of the drug in the blood stream and lead to the redistribution of the drug from vital organs to the blood compartment and possibly other sink organs as well. Accordingly, the goal of this chapter was to again use in vitro experi ments to assess the ability of uni and multilamellar, polymer coated, anionic lipos omes to sequester the local anesthetic bupivacaine (BUP) from buffer solutions and human serum. In addition to testing liposome-drug binding within a new drug cla ss, this chapter also includes another strategy for increasing liposome-drug affi nity: the use of multilamellar liposomes. 5.2 Methods 5.2.1 Liposome Preparation Liposomes composed of a molar rati o of 95:5 DOPG:DPPE-mPEG-2000 were prepared via extrusion (Avanti Polar Lipids, Al abaster, AL). Lipids were combined in their respective molar ratios and then dissolv ed in a 9:1 mixture (by volume) of 99.9% chloroform/99.9% methanol (S igma-Aldrich, St. Louis, MO ) such that a 21 mg/mL concentration of lipids was obtained. T he organic solv ent was then evaporated under a stream of nitrogen. After an even and uniformly dried lipid film wa s obtained, the dried 121

PAGE 122

lip id layer was hydrated with PBS (Sigma-A ldrich, St. Louis, MO). The lipid concentration after hydration was 16 or 25 mg lipid/mL PBS, depending on the final lipid loading desired. The lipid suspension was t hen mixed for 1-2 minutes on level 5 using a Mini Vortexer mixer (Fisher Scientific, Pi ttsburg, PA). Unilamellar liposomes were then bath sonicated at room temper ature for 20 minutes using a G112SP1 Special Ultrasonic Cleaner (Avanti Polar Lipids, Alabaster, AL), whereas multilamellar liposomes were not immediately sonicated. The lipid suspensions were then heated to 30C and stirred overnight for approximately 20 hours. After st irring, the unilamellar liposomes were extruded through a 100 nm membrane 15 times using an Avanti Mini-Extruder (Avanti Polar Lipids, Alabaster, AL). The size dist ributions of the extr uded liposomes measured by dynamic light scattering are reported in Chapter 3. The liposomes had a mean diameter of 118 15 nm. To make multila mellar liposomes, the lipid dispersion was bath-sonicated for 2 minutes after the overni ght stirring. Such short bath sonication times produce multilamellar liposomes [27]. 5.2.2 In Vitro Drug Binding Measure ments in Buffer or Human Serum Liposome s olutions at loadings of 16 or 25 mg/mL were added to solutions of bupivacaine hydrochloride (Sigma-Aldrich St. Louis, MO) in PBS or human serum (Fisher Scientific, Pittsburg, PA) such that the volume of the liposome dispersion added was 9% or 11.6% of the final solution volume giving final lipid l oadings of 1.45 or 2.9 mg lipid/mL. These mass loadings were chosen based on results from Chapters 2-4, where similar loadings allowed liposomes to compete with serum proteins for antidepressant binding. Multilamellar liposom es were added at a lipid concentration of 1.45 mg lipid/mL, and a higher binding affini ty was observed compared to unilamellar liposomes at the same loading. Based on t he increased binding, it would seem logical 122

PAGE 123

to focus on multilamellar liposom es. Howe ver, large liposomes will be cleared from systemic circulation into the liver and spleen faster than smaller liposomes, which may not be desirable. Such issues can only be conclusively addressed through in vivo experiments, but, because of the slower clear ance, we believe unilamellar liposomes to be more promising for drug overdose treatmen t. Thus, only unilamellar liposomes were measured at both 1.45 and 2.9 mg lipid/mL. Serum concentrations for BUP during infusion can range from 1 to 15 M [118,119]. Circulatory collapse in ewes has been observed at concentrations as high as 27. 7 M [120]. Accordingly, the total BUP concentrations tested in our studies ranged from 5 to 50 M. Control solutions of BUP in PBS and serum without liposomes were also prepared to allow for uptake quantification and comparison. After stirring for 15 minut es, the solutions were ultracentrifuged at 5000 rpm for 15-25 minutes in a 15 mL Cent riprep Centrifugal Filter Device (Millipore Corp., Billerica, MA) c ontaining an Ultracel YM10 filter (10,000 molecular weight cutoff). The concentrati on of BUP in the filtrate (free drug concentration) and the control solution wa s detected by measuring ultraviolet absorbance at 215 nm after passing the sample s through a C18 column (Waters Corp., Milford, MA) using a 99.9% acetonitrile/50 mM KH2PO4 solvent mixture in a 25/75 ratio (Sigma-Aldrich, St. Louis, MO ). The calibration curves for concentration versus area under the curve were linear with R2 > 0.99. To ensure that all free BUP unbound to liposomes or proteins was accounted for, solutions of BUP at various concentrations were also filtered through YM10 filters and a calibration curve (R2 > 0.98) was made and used to correct for the dr ug adsorbed to the membranes. 123

PAGE 124

5.2.3 Statistical Analysis Drug binding results were compared using the Student t distribution. Comparisons between drug uptake values or the am ount of drug unbound to proteins and/or liposomes were made at each drug conc entration because of the presence of interactions between drug concentration and the desired effect of study, which was the liposome loading or the presence or absence of liposomes. In most cases, drug binding was measured for each initial drug concentration at least twice (n = 2). Results were considered significant when P < 0.05. 5.3 Results 5.3.1 Bupivacaine Extracted from Buffer The results for BUP extraction fr om PBS with 95:5 DO PG:DPPEmPEG-2000 liposomes are shown in Figure 5-1. T he unilamellar liposomes sequestered roughly 60%-65% and 77%-85% of BUP from PBS at 1.45 and 2.9 mg lipid/mL loading, respectively. The increased lipid loading increased the percent BUP uptake substantially at all BUP concentrations measured ( P = 0.001, 0.002, <0.001, and <0.003 for 5, 20, 35, and 50 M, respectively). Al so shown in the plot are the drug uptake values measured for multilamellar liposomes at 1.45 mg lipid/mL. As Figure 5-1 shows, the drug binding increased from 60%-65% for unilamellar liposomes to 71%-90% for multilamellar liposomes at the same mass loading ( P = 0.002, 0.001, 0.001, and 0.08 for 5, 20, 35, and 50 M, respectively). 5.3.2 Bupivacaine Extracted from Human Serum The amount of BUP bound to serum proteins decreased as drug concentration increased (Figure 5-2). At low BUP c oncentrations, the per centage bound approached 90%, which is consistent with reported valu es [121]. When liposomes were added to 124

PAGE 125

serum, the total bound fraction increased. The free BUP unbound to liposomes and/or proteins before and after liposom e addition is shown in Figure 5-3. The liposomes reduced the free drug in solution by 36% (P = 0.037), 56% ( P = 0.022), 47% ( P = 0.042), and 50% ( P = 0.018) for BUP concentrati ons of 5, 20, 35, and 50 M, respectively. 5.4 Discussion The treatment of local anesthetic-induced toxi city has rapidly evolved over the past 10 years with the emergence of lipid emulsion therapy [41,42,122]. Many case studies continue to provide evidence for the effectivene ss of lipid emulsions at toxic ity reversal in actual clinical settings [12-14,33-36,40]. However, two major issues surrounding the use of lipid emulsions are currently at the forefront. First, the doses of lipid emulsions required for toxicity reversal are potentially high. The resulting hyperlipidaemia could pose considerable risks to patients, many of which are unknown or not fully understood [43,45]. Second, the mechanism of toxi city reversal remains a major question associated with the use of lipid emulsions. The results from this chapter may offer some insights and potential for ex ploring these critical issues. Our results for drug binding support the pos sibility of achieving significant free drug concentration reductions with liposom es, perhaps with doses lower than those used with lipid emulsions. In vitro BUP binding by lipid emulsions (Intralipid 20%, Fresenius Kabi, Bad Homburg, Germany) and pluronic microemulsions in buffer was explored by Varshney et al. [19]. They tailored the pluronic microemulsions for maximum drug binding per unit ma ss by testing several pluronic surfactants, varying the oil to surfactant ratio, and adjusting the fatt y acid chain length and concentration. The final concentration of lipid emulsion in the in vitro binding experiments was 125

PAGE 126

approximately 1 mg/mL, which is about 30% lower than the lipid loadings cons idered in this chapter. At total BUP concentrations similar to those tested here, nearly 30% of BUP was bound to the lipid emulsions, whic h is about one-third of the drug bound to the liposomes. The total loading of pluronic mi croemulsions used (surfactant and oil) was 7.7 mg/mL, which is about 5 times the lip id loadings considered here. The maximum drug binding to the microemulsions was 60%, which is substantially less than the drug bound to the liposomes. Thus, comparison of in vitro measurem ents reported here with the data reported by Varshney et al. [19] suggest that p egylated, anionic liposomes composed of DOPG lipids have more BUP bound per unit mass than lipid emulsions or pluronic microemulsions. Liposomes sequester such large amounts of BUP and other drugs because of their unique structures and compositions (see Chapters 2,3, and 6). They can be tailored to more specifically target toxins because of surface c harge and/or lipophilicity, and can be pharmacokinetically enhanced to ci rculate for long times through the use of polymers, such as polyethylene glycol. A pproximately 83% of BUP, which has a pKa value of 8.1, is in the c harged state at physiological conditions [121,123]. Previous chapters have shown that anioni c liposomes preferentially bind to cationic drugs. As drug delivery vehicles, liposomes have also b een proven biocompatible. The results in serum confirm that liposomes can effectiv ely compete with serum proteins for drug binding and could rapidly reduc e the free drug in the blood co mpartment. This could, in turn, cause a redistribution of the drug from vital organs to the blood and/or other slowly equilibrating organs. Thus, we propos e that liposomes, based primarily on redistribution, could be a more effective detoxification treatment than lipid emulsions 126

PAGE 127

and could be applied to a wider array of drugs, in cluding local anesthetics. The in vitro drug-binding data presented here provide the rati onale for in vivo studies to test this hypothesis. It has been proposed that lipid emulsions ma y reduce toxicity by either inducing drug redistribution from vital organs to the blood compartment and/or slowly equilibrating organs or the enhanc ement of oxidative metabolis m [44]. The difficulty in discerning between these mechanisms stems from a lack of knowledge as to the destination of lipid emulsions and the time scales for transport to those destinations. Lipid emulsions most likely enter tissues, while polymer-coated liposomes of the type used in this chapter circulate in the bloodstr eam for some period of time. Our results strongly infer that liposomes bind more BUP per unit mass than lipid emulsions. Therefore, if liposome administration after an overdose r educes BUP toxicity, the mode of reversal will be related to redistribution. Furthermore, if redistribution is the dominant mechanism, then liposomes would be the better detoxification candidate. 5.5 Conclusions In conclusion, pegylated, anionic lipos omes were shown to be highly effective at sequestering the local anesthetic drug BUP from in vitro solutions with a removal efficiency of about 80% in buffer and a free-drug reduction of about 50% in human serum. The promising results present a rationale for further investigation into the treatment of local anesthetic toxicity with liposomes. Such liposomes may offer an alternative therapy to lipid emulsions or microemulsions at treat ing local anesthetic toxicity. Furthermore, in vivo comparisons of liposomes and lipid emulsions at treating drug toxicity may lead to an improved underst anding of the mode of action of lipid emulsions. 127

PAGE 128

Figure 5-1. Percent of total bupivaca ine bound to 95:5 DOPG:DPPE-mPEG-2000 liposomes in buffer solutions for unilamella r liposomes at lipid loadings of 1.45 (O) and 2.9 mg lipid/mL ( ) and multilamellar liposomes at 1.45 mg lipid/mL (); n = 2 at all concentrations. 128

PAGE 129

Figure 5-2. Percent of tota l bupivacaine bound in human se rum (O) and in a mixture of human serum and unilamellar lipos omes at 2.9 mg lipid/mL ( ); n = 2 for all measurements except serum bindi ng at 35 and 50 M, where n = 3. 129

PAGE 130

130 Figure 5-3. Free bupivacaine (unbound to pr oteins and/or liposomes) versus total bupivacaine concentration in human serum samples in the absence ( ) and presence ( ) of unilamellar, 95:5 DOP G:DPPE-mPEG-2000 liposomes at 2.9 mg lipid/mL calculated from data in Figure 5-2. Differences were significant at all concentrations tested (P = 0.037, 0.022, 0.042 and 0.018 for 5, 20, 35, and 50 M, respectively).

PAGE 131

CHA PTER 6 THE INTERACTION OF CATIONIC DRUGS WITH LIPOSOMES 6.1 Introduction In the past several chapters, we have shown that anionic liposomes can sequester significant amounts of antidepressant and anest hetic drugs. It is imperative that the interactions responsible for such behaviors ar e well characterized. Such interactions are important for developing better drug overdose treatments, as well as for many other important membrane rela ted applications related to drug delivery, cell viability, and so on. In this chapter, we focus on char acterizing liposome-drug interactions via equilibrium binding and its effect on liposome permeability. Anionic and zwitterionic liposomes were used, along with bupivaca ine (BUP), amitriptyline (AMI), and imipramine (IMI). The varying physical prop erties and structures of the drug molecules allowed a detailed understanding of the ef fects of these factors on binding and permeability. Additionally, binding was measured to charged poly(methacrylic acid) and poly(acrylic acid) microparticles with total charge comparable to liposomes to investigate the importance of the charge distribution (i.e ., surface versus bulk) on binding. To explore the interactions, li posome properties such as surface charge, bilayer fluidity, and number of lamellae were va ried, as well as the ionic strength of the bulk medium. The experimental results were supplemented by a continuum model that incorporated electrostatic interactions and a Langmuir binding isotherm. The model was validated by comparing predictions with experiments at high ionic strengths for unilamellar liposomes (ULL). 131

PAGE 132

6.2 Materials and Methods 6.2.1 Materials Methanol, chloroform, Dulbeccos P hosphat e Buffered Saline (PBS) without calcium chloride and magnesium chloride (0 .165 M ionic strength), ethylene glycol dimethacrylate (EGDMA), acrylic acid ( AA), amitriptyline hy drochloride (AMI), imipramine hydrochloride (IMI), bupivacai ne hydrochloride (BUP), and Sephadex G-50 (fine) were purchased from Si gma Aldrich (St. Louis, MO, U SA). Dosulepin (dothiepin) hydrochloride (DOS) was purchased from LGC Promochem (Teddington, Middlesex, U.K.). Methacrylic acid (MAA) was purchased from Polysciences, Inc. (Warrington, PA). Ciba Darocur TPO, a photoinitiator, was a gift from Ciba Specialty Chemicals (Tarrytown, NY). Sodium chloride, Centri prep Centrifugal Filter Devices (Millipore Corp., Billerica, MA) containing Ultracel YM 10 filters (10 000 molecular weight cutoff), Whatman GF/B glass microfi ber filters, and calcein dye (fluorexon) were purchased from Fisher Scientific (Pittsburgh, PA). The lipids DOPG, dissolved in chloroform, DPPE-mPEG-2000, in powder form, DMPG, in powder form, and DMPC, in powder form, as well as a Mini-Extruder kit for liposome preparation, we re purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). 6.2.2 Liposome Preparation for Drug Bi nding and Zeta Potentia l Measurements Liposomes composed of pure DOPG, pure DMPG, pure DMPC, and molar ratios of 50:50 DMPC:DOPG, 95:5 DOPG:DPPEmPEG-2000, and 95:5 DMPG:DPPEmPEG2000 were prepared via extrusion. Lipids were combined in their resp ective molar ratios and then dissolved in a 9:1 mixture (by volume ) of chloroform:methanol such that a 1220 mg/mL concentration of lipids was obt ained. The organic solvent was then evaporated under a stream of nitrogen. The dried lipid layer was hydrated with PBS or 132

PAGE 133

PBS that had an incre ased ionic strength of 1.5 M via sodium chloride. The lipid concentration ranged from 8 to 25 mg lipid/mL buffer. The lipid suspension was then bath sonicated for 20 minutes, heated to 30C, and stirred over night. After stirring, the liposome solution was extruded through a 100 nm membrane13-15 times to produce liposomes with diameters slightly great er than 100 nm. DOPG liposomes with diameters close to 40 nm were prepared vi a probe sonication as described in Chapter 2. To prepare DOPG and 95:5 DOPG:DPPE-mPEG-2000 MLL, a similar procedure to the one described above was followed to obtain a hydrated lip id film. The dispersion was then vortex mixed for 2 mi nutes at medium speed, heated to 30C, stirred for 24 hours, and bath sonicated for 2 minutes. Shor t bath sonication times are well-known to produce MLL [27]. 6.2.3 Liposome Preparation for Calcein Leakage St udies Liposomes composed of pure DOPG and a molar ratio of 95:5 DOPG:DPPEmPEG-2000 were prepared as described above in Section 6.2.2, except the dried lipid layer was hydrated with 100 mM calcein dye dissolved in PBS such that the lipid concentration was 25 mg lipid/mL soluti on for pure DOPG and 28 mg/mL for 95:5 DOPG:DPPE-mPEG-2000. The mass loading was slightly higher for the PEG-2000 liposomes so that the num ber of charged head groups in the liposomes would be identical for both types of liposomes. T he lipid suspension was then mixed using a Fisher vortex mixer for 1-2 minutes, followed by bath sonication for 20 minutes. The vesicles were then gently stirred overni ght at 30C for approx imately 20 hours to increase the entrapped aqueous volume of liposom es [27]. After stirring, the liposome solution was extruded through a 100 nm membr ane 15 times. To remove excess dye 133

PAGE 134

from the bulk, the liposome solution was passed through a mini-column of Sephade x G50 (fine) using the centrifugation method to ensure that almost all ( 90%) of the lipids added to the Sephadex bed were recovered [124 ]. The resulting liposome solution was diluted by a factor of 1001 based on the obs ervation that the calc ein release at the subsequent liposome loading was in t he linear detection range of 0 to 10 M even upon 100% calcein release after bilayer disruption. 6.2.4 Preparation of Poly(met h acrylic acid) and Poly(acr ylic acid) Microparticles Porous poly(methacrylic acid) and poly(ac rylic acid) microparticles were studied for their ability to remove AMI from soluti on. For particle synthesis, MAA or AA was combined with the cross-linker EGDMA. After purging with nitrogen to remove excess oxygen, the photoinitiator Ciba Darocur TPO was added, and 1 mL of the resulting mixture was added to 5 mL of et hyl acetate. The solution was then placed under a UV light source to induce polymerization. Afte r the formation of microparticles, the ethyl acetate was boiled off to yield a gel of micro particles. The microparticles were added to PBS such that the concentration of microparti cles was around 9 mg/mL, similar to the concentration of lipids in liposome dispersion s previously described. The microparticles were then resuspended in the buffer solution with the use of bath and probe sonicators. The resulting microparticle suspension was then tested for its ability to remove AMI from buffer solutions as described in Section 6.2.6. The final concentration of microparticles was around 0.85 mg/mL and the AMI concen trations measured ranged from 1 to 50 M. 6.2.5 Liposome Characterization Mean diameters and size distributions for th e liposomes were reported in Chapt ers 2 and 3 and elsewhere [71]. Briefly, t he extruded ULL had diameters of 118 15 nm and the probe sonicated ULL had di ameters of 39.2 4.5 nm. The bath sonicated MLL 134

PAGE 135

produced widely varying size dis tributions wit h liposomes ranging from roughly 50 nm to 3 m in size. The zeta potentials of t he liposomes are reported in Table 6-1. 6.2.6 Drug Binding to Liposomes in Buffer Solutions DOPG, DMPG, DMPC, 95:5 DOP G:D PPE-mPEG-2000, and 95:5 DMPG:DPPEmPEG-2000 liposomes were added to buffer so lutions containing drug such that the final lipid concentration ranged from 0.72 to 1.44 mg lipid/mL. Buffer solutions with ionic strengths of 0.165 M, which wa s the original ionic strength of the buffer, and 1.5 M were used. The increased ionic strength was achi eved with the addition of sodium chloride. Control solutions of drug in PBS were also made to allow for uptake quantification and comparison. After being stirred for 15 mi nutes, the drug-liposome solutions were ultracentrifuged at 5000 rpm for 15 minutes in a vial that contained an Ultracel YM10 filter (10,000 molecular weight cutoff). Experiments were previously done to ensure that the systems were at equilibrium before filtrati on in Chapter 2. The filters were rinsed with DI water and then with PBS at 5000 rpm for 10 minutes prior to their use in the experiments. The concentration of the drugs in the filtrate (free drug concentration) and the control solution was detec ted by measuring UV absorbanc e at 215 nm after passing the samples through a C18 column (Water s Corp., Milford, MA USA) using an acetonitrile/50 mM NH4H2PO4 solvent mixture in a 35/65 ratio for AMI, IMI, and DOS, and an acetonitrile/50 mM KH2PO4 solvent mixture in a 25/75 ratio for BUP. The calibration curves for concentration vers us area under the curve were linear with R2 > 0.99. To ensure that all of the unbound drug was accounted for, solutions of the drugs at various concentrations were passed thr ough YM10 filters in a separate test. Small amounts of drug were taken up by the filter, and correction curves were made and used to correct for drug adsorbed to the membranes in subsequent tests. 135

PAGE 136

6.2.7 Zeta Potential Measurements Zeta potential measurements were made for pure DOPG, pure DMPC, and 50:50 DMPC:DOPG liposomes for use in the model developed in Section 6.3.3. The measurements were made in PBS (0.165 M i onic strength) using a Brookhaven Zeta Plus machine with a voltage of 2.5 V at a field frequency of 2 Hz. At least 10 runs were done for each liposome type. Measurements were also done in PBS at AMI and BUP concentrations of 100 M and 50 M, respectively. The pres ence of drug molecules did not change the zeta potential values appreciably. Results are shown in Table 6-1. Measurements could not be made in 1.5 M i onic strength solutions due to extremely high conductance values. 6.2.8 Liposome Leakage Studies The lipos omes prepared by the methods described in Section 6.2.3 contain 100 mM calcein solution in the liposome core. This concentration is sufficiently high so that the fluorescence is quenched, resulting in n egligible signal, except from the dye that diffuses out from the liposomes into the bul k [124]. The baseline fluorescence of the liposome solution prior to leakage was first measured using a Quant ech Digital Filter Fluorometer with excitation and emission f ilters at 490 and 515 nm, respectively. A concentrated drug solution or PBS was then added, and the fluorescence was measured after 10 minutes. Drug concent rations of 0 (PBS), 8.6, 38, and 150 M were studied for 95:5 DOPG:DPPE-mPEG -2000 liposomes. To compute the percent release, the following formula was used, 100x FF FF Release%ototal ot (6-1) 136

PAGE 137

where Ft was the fluorescence m easurement at time t, F0 was the fluorescence at time zero, and Ftotal was the total calcein released, which was determined by breaking the liposomes with 100 L of 20% (v/v) Triton X-100. Corr ections were made to account for the dilution upon addition of t he drug, PBS, protein, or Tr iton X-100 solutions. All release experiments were carried out at least twice. 6.2.9 Statistical Analysis The data was analyzed using R, a free stat istical software package developed by the R Proje ct (http://www.r-project.org/). Fo r the drug uptake expe riments, ANOVA was used. In such cases, P values were reported and a si gnificance level of 0.05 was assumed. When interacti ons were significant ( P < 0.05), drug binding was analyzed for each drug concentration indepen dently using the Students t distribution. Error bars are presented in most of the results but are not always evident due to the small standard errors observed in some cases. The dat a are plotted and report ed as mean standard error of the mean. Comparisons were made using the Students t distribution. The 95% confidence intervals discussed in Section 6. 3.2.2 were computed using JMP (SAS, Cary NC). The standard errors reporte d for the fitted parameters in Section 6.3.3 (a and K) were computed with R using a nonlin ear least-squares (nls) model. 6.3 Results and Discussion To facilitate the discussion of the results in this chapter, the structures of AMI, BUP, and the lipids used are shown in Figure 6-1. The structure of AMI is similar to all other TCAs. They have two distinct regions a hydrophilic tail and a lipophilic, aromatic region. Their pKa values are around 9.5, so that 99% of their molecules are in the charged form at pH 7.4 [106]. BUP has a mo re centrally located charge, is only 86% protonated at pH 7.4 (pKa 8.2 at 25C [123]), and has less surface activity compared to 137

PAGE 138

TCAs. DOPG is anionic and pos sesses olei c acid carbon chains with cis double bonds at its ninth carbon, causing it s gel to fluid phase transition temperature to be very low at -18C. DMPC is zwitterionic and has saturated myristic acid chains, resulting in a gel to fluid phase transition temperatur e of 23C. DMPG is anion ic like DOPG but also has saturated myristic acid chains. Thus, the nonpolar interiors of DMPC and DMPG are most likely in the gel phas e at room temperature. 6.3.1 Electrostatic Cont ribution to Drug Binding To probe the importance of electrostati cs in bind ing between cationic drugs and liposomes, various types of binding experim ents were performed. First, anionic DOPG and 95:5 DOPG:DPPE-mPEG-2000 liposomes were tested in solutions of high ionic strength to probe the effect of electrostatics. Next, net n eutral DMPC lipids were used to conclusively prove that charge was instru mental in achieving high drug sequestration. Lastly, porous microparticles with negative charge comparable to liposomes but dispersed uniformly throughout the particles we re also tested for their drug binding ability to help determine whether or not the charge distribution and lip ophilic nature of liposomes were important factors in binding. 6.3.1.1 Effect of ionic strength If drug binding to lipos omes is dominated by electrostatic effects, increasing the ionic strength should reduce drug binding due to shortening of the Debye length and screening of charge-charge interactions. In this case, buffer was used with an ionic strength, I, of 0.165 M, as well as buffer with NaCl added su ch that the ionic strength was 1.5 M. As described in Se ction 6.2.2, lipid bilayers we re typically hydrated with the 1.5 M ionic strength solution during liposome preparat ion. In some cases, hydration was carried out with 0.165 M buffer and the i onic strength of the uptake medium was 138

PAGE 139

adjusted to give a final value of 1.5 M. No differences were detected in binding for the two methods with ULL ( P = 0.114, data not shown), demon strating that sodium ions permeated through the bilayers and eliminat ed ionic strength gradients across the bilay ers prior to filtration. The Debye l engths, calculated with the following equation for aqueous solutions at room temperature: 0.50 DI 3045.0 l (6-2) were 0.75 and 0.25 nm for the lower and higher i onic strength solutions, respectively. In addition, Equation 6-12 (see below) predicts the zeta potential change on the liposome surface to be approximately 50 mV for DOPG liposomes when transitioning from 0.165 to 1.5 M ionic strength, in agreement with the experimental results of Egorova [125] and McLaughlin [126]. Results from IMI and BUP uptake studies in both mediums wit h anionic ULL are shown in Figure 6-2. Clear ly, the increased ionic strength reduced the drug uptake for IMI ( P = 0.003 for 2.9 M, P < 0.001 for 24.2 M) and BUP ( P < 0.001). The free fraction of drug was increased by a factor of roughly 2-5 for IMI and 2 for BUP, proving that electrostatics was a significant factor for drug binding in both cases. Furthermore, the amount of BUP bound per unit mass of lipid wa s less than the amount of IMI bound. This likely resulted from the lower proporti on of charged molecules at pH 7.4 for BUP versus IMI, as well as for other reas ons discussed later in Section 6.3.4. 6.3.1.2 Drug binding to neutral liposomes To further explore the importance of electrostatics, drug binding to DMPC lipos omes was measured, which have a mu ch less negative surface charge than DOPG. This point is clear from the measured zeta potentia l values for both liposomes 139

PAGE 140

shown in T able 6-1. The results in Figur e 6-2 comparing binding to anionic and DMPC liposomes show that reducing the net negative charge drastically reduces the drug uptake for both IMI ( P < 0.001) and BUP ( P < 0.001). This, al ong with the reduced binding at higher ionic strengt hs, proves that the binding is dominated by electrostatic interactions between the cationic drug and the negatively char ged surface of the liposomes. 6.3.1.3 Drug binding to anionic microparticles To compare drug binding for micr oparticles and lipos omes, AA and MAA microparticles were prepared and drug bindi ng was measured. Similar experiments have been conducted elsewhere at AM I concentrations greater than 300 M, which are significantly higher than concentrations expl ored here and also well beyond the relevant concentrations for drug overdose treatment [ 23]. At similar mass loadings to liposomes, both the MAA and AA microparticles were unabl e to sequester any significant amount of AMI from PBS. At 0.85 mg/mL, MAA and AA microparticl es possess charges of -0.95 and -1.14 C/mL, respectively, compared to -0.18 C/mL for DOPG liposomes at 1.45 mg/mL. Thus, the ability for liposomes to bind large amounts of drug must be attributed to both their charged surfaces and lipophilic bilayers, as opposed to the charged microparticles, which lack a lipophilic component. Chakraborty and Somasundaran observed relatively large bindings to anionic microparticles at higher drug concentrations but also observed a reduc tion in binding upon lowering the drug concentration. Upon extrapolation of t heir results to the low physiological concentrations explored here, the results agree reasonably well [23]. 140

PAGE 141

6.3.2 Lipophilic Contribution to Drug Binding Section 6.3.1 suggests that TCAs and BU P are preferentially attracted to negatively charged liposomes, whereupon they are bound and retained. This observation provides no information about the conformation of the drugs after binding. Drugs could reside in the charged double layer surrounding the lipos omes or enter the lipid bilayer to some degree. This issue was investigated through drug binding comparisons between liposomes with c ontrasting nonpolar, inner bilayers and liposomes with single or multiple concentric bilayers. Additionally, leakage of a water soluble fluorescent dye from the aqueous core of liposomes was monitored while in the presence of AMI, IMI, or BUP. The dye leakage exper iments, when viewed in light of the liposome-drug binding results, afforded addi tional insight into t he specific nature of the drug-bilayer interactions. 6.3.2.1 Drug binding with DM PG liposomes: the effect of bila yer fluidity The difference in bilayer fluidity betw een DOPG and DMPG is a useful probe for drug-bilayer interactions, since any corre sponding change in drug bound would imply that drug-bilayer interactions are import ant, as opposed to drug-surface interactions alone. Drug binding for both lipid types is plotted as a function of initial drug concentration in Figure 6-2. The DOPG liposomes bound more drug than the DMPG liposomes for IMI ( P < 0.001) and BUP ( P < 0.001). The magnitude of the difference was about 2% and 10% of the tota l IMI and BUP present, respectively. This statistically significant difference in binding leads to the conclusion that both drug types are interacting with the lipid bilayers and suggest that drugs enter fluid bilayers more easily than gel bilayers. Also note the gap in binding for DMPG and DMPC lipids, again confirming the importance of electrostatic interactions. 141

PAGE 142

6.3.2.2 Drug binding with mult ilamellar liposomes (MLL) To confirm the hypothesis of drug-bilayer interaction from the bilayer fluidity tests, drug binding was measured with MLL. At equivalent mass loadings, MLL and ULL should bind equal drug amounts if the drugs readily cross lipid bilayers. Figure 6-2 shows the outcomes of AMI and BUP binding experiments using DOPG MLL. MLL bound more drug than ULL at equivalent mass loadings. The disparity is small for AMI due to the high proportion of total drug bound in both cases, but 95% confidence intervals (magnified portion of Figure 6-2) confirm this point. Considerably larger binding increases were observed for BUP at three of the four c oncentrations measured ( P = 0.002, 0.001, 0.001, and 0.08 for 5, 20, 35, and 50 M, respectively). The results prove that both drug types c an cross the bilayers and that MLL bind more drug than ULL for the same lipid loading. We believe this enhancement to be the resu lt of electrostatic interactions or energy barrier reductions. For MLL without PE G, such as the ones shown in Figure 6-2 for AMI binding, the concentric bilayers we re separated by a water layer approximately 1-3 nm thick [127,128]. This proximity a llowed adjacent charged layers to interact, thereby increasing the effective potential experienced by the drug bound to the inner layers. BUP binding with PEG coated M LL must have been enhanced in a different way, since adjacent charged layers were s eparated by about 10 nm, making interaction negligible [129]. In this ca se, the dielectric constants of the aqueous layers between bilayers were decreased due to the presenc e of PEG. Aqueous PEG solutions have been shown to have lower dielectric constants than pure water, and this effect is mentioned for liposomes elsewhere as well [ 130,131]. Lower dielectr ic constants would reduce the surface potentials of the bilayers [131], which would act to reduce drug 142

PAGE 143

binding bas ed on the results presented thus far. However, there is a competing effect which we believe is overcoming the reduced surface potentials and actually increasing the amount of drug bound. As a result of the reduced dielectric constant, the energy barrier for the cationic drugs to enter the aliphatic bilayers is reduced. The chargecharge interactions are still key, but binding is more energetically favorable. This effect of reduced dielectric constants on i on transport across membranes has been extensively studied [132-134]. To demonstrate that binding was still driven by electrostatics in spite of the reduced energy barrier for membrane intera ction, BUP binding to MLL prepared using PBS with salt added to increase the ionic str ength to 1.5 M was measured (Figure 6-2). High ionic strength buffer was used to hydrat e the liposomes to avoid ionic strength gradients across the bilayers, as such gradien ts could take long periods of time to disappear due to limited sodium permeability th rough the multiple layers of the MLL. The BUP binding was much lower than the MLL at the original ionic strength ( P < 0.001). Additionally, the MLL in 1.5 M ioni c strength solutions bou nd similar amounts of drug to ULL at 1.5 M ionic strength ( P = 0.935). Thus, whil e enhanced drug binding stemming from energy barrier reductions fo r charge transfer across the membrane was important, electrostatic attractions still domi nated. The effect of increasing the ionic strength of the medium overcame the lower dielectric constant effect of the aqueous layers. 6.3.2.3 Liposome leakage induced by drugs In Figure 6-3, the percent of calcein re lease from pegylated, anionic liposomes 10 minutes after drug application is displayed. For AMI ( P < 0.001 and P < 0.001) and IMI ( P = 0.001 and P < 0.001), leakage at 38 M and 150 M was significantly greater than 143

PAGE 144

the control. For BUP, significant leakage was only observed at 150 M (P = 0.016). Even at 150 M, BUP provoked less leakag e than the TCAs at 38 M. Despite the higher perc entage of drug bound for AMI and IM I compared to BUP, this factor alone cannot account for the differences in l eakage because the number of drug molecules bound to the liposomes is larger for BUP at 150 M compared to TCAs at 38 M. Results with DMPG and MLL showed that both TCAs and BUP traverse the lipid bilayers easily and likely reside there to so me extent. Thus, it is clear that the conformation of TCAs in the bilayer must be such that they enhance the bilayer permeability more than BUP. This seem s plausible upon exam ination of the drug structures. The two contrasting regions of TCAs are structurally separated, which would allow the lipophilic porti on to associate with the b ilayers and the charged region to extend into the bulk phase. This dual association would disrupt the bilayer significantly. BUP, by contrast, probably rema ins inside the nonpolar region to a greater degree and alters the bilayer st ructure to a lesser extent. 6.3.3 Mechanism Validation through Continuum Modeling 6.3.3.1 Model development All of the experimental findings pr esent ed in Chapter 6 thus far point to a combinatorial drug uptake mechanism that includes both electrostatics and drug-bilayer interactions. To further va lidate that observation, a continuum model was developed and validated by matching model predictions to experiments in solutions of increased ionic strength. If the Poisson-Boltzmann equatio n for the potential distribution around spherical surfaces, such as liposomes, is solved un der the Debye-Huckel approximation, the following equation is obtained: 144

PAGE 145

R)(r e r R (6-3) where is the zeta potential, r is the radial coor dinate with its origin at the center of the liposome, R is the liposome radius, and is the inverse of the Debye length [135]. To determine the amount of drug sequestered in the double layer, one can solve the species conservation equati on for the drug to obtain kT eC(r)Ce (6-4) where C(r) is the drug concent ration in the double layer as a function of the radial coordinate, C is the bulk drug concentration far from the liposome surface, e is the elementary charge, k is the Bo ltzmann constant, and T is the temperature. An overall mass balance on the drug gives ip 0 2 LC SCdrrCC(r)N 4 (6-5) where Ci is the total drug concentration, NL is the number of liposomes per volume, is the concentration of the dr ug specifically bound to t he bilayer surface, and Sp is the surface area of the liposomes per volume, which is given by MW AfN S0 Al p, (6-6) where MW is the average mole cular weight of the lipids present (excluding PEG), NA is Avogadros number, l is the lipid loading in the so lution (excluding PEG), f is the fraction of lipid molecules that participate in drug binding, and A0 is the average area per lipid molecule on the surface. In the limit of double layer thickness (typically a few 145

PAGE 146

nm) much smaller than the liposome radius (>50 nm), the concentration profile simplifies to )(e kT -y eCCe, (6-7) where y is the distance from the liposome surface. The overall mass balance also simplifies to iP 0 PC Sdy1y)) exp( kT exp(S1C e. (6-8) The specific drug binding of drugs to lipid bilayers can be modeled as a Langmuir isotherm [57]: Tk Tk eCK eCa R)C(rK R)C(ra e e (6-9) where a and K are the paramet ers of the isotherm. T he zeta potentials used for modeling drug binding at 0.165 M ionic str engths were measured, whereas changes due to an increased ionic strength were calcul ated as detailed below (Table 6-1). Zeta potential values were also measured in the presence of AMI (100 M) and BUP (50 M), but the values were relatively unaffe cted by drug binding. PEG chains have a negligible effect on the surfac e potential of liposomes but move the slipping plane away from the liposome surface, leading to changes in zeta potential. As a result, zeta potential measurements for pegylated liposom es are less representative of the true surface potential characteristics when compared to unpegylated liposomes. 146

PAGE 147

Measurements were therefore made for unpegylated liposomes and used for pegylated lipos omes with similar lipids [136]. In Equat ion 6-6, the value used for f was 1, since experiments have already shown that all drugs tested can pass through MLL and bind to both sides of the bilayers. Also, the mole cular weights and lipid loadings used were corrected to exclude the PEG, as the bulk of the PEG is extended into the aqueous phase on either side of the liposome b ilayer and should have little bearing on drug binding. All data used for model fitting or model validation was derived using ULL extruded with 100 nm membranes. The values used for the constants in Equations 6-6 through 6-12 are listed in Table 6-1, along with ranges of previously reported values for similar systems for comparison. 6.3.3.2 Model fits to da ta for antidepressant drugs Initially, the amount of AM I bound exc lusively within the double layer of DOPG liposomes (1.44 mg/mL) at an AMI concentration of 25 M was estimated by calculating the first term on the left-hand side of Equation 6-8. The model predicted that a zeta potential value of -50 mV woul d yield double layer binding of less than 1%. Based on our observations reported in Section 6.3. 1 and previous chapters, the drug uptake actually approaches 99% in some cases, pr oviding strong evidence for significant drugbilayer interactions [71]. Accordingly, Equation 6-8 was used in combination with AMI uptake data by 40 nm, pure DOPG liposomes at a lipid loadi ng of 0.72 mg/mL reported in Chapters 2 and 3 to estimate the parameters a and K under the assumption that the drug molecules were located both in the double layer and the lipid bilayer. T he data used ranged from an initial AMI concentration of 1 M to around 110 M. Figure 6-4a shows the experimental data and the fitted data based on Equation 6-8. The va lues of the fitting 147

PAGE 148

parameters, a and K, were 5.36 x 10-7 3.3 x 10-8 mol/m2 and 1.88 x 10-2 2.1 x 10-3 mol/m3, respectively. The fit was very good, indicating that a combination of electrostatic and lipophilic interactions is a more plausible explanation for the results obtained, rather than simple double layer electrostatics. Based on the value of a obtained from the fit, the area per drug molecule at maximum packing is about 310 2, which is about 5 times the area pe r molecule for the lipid heads. In Figures 6-4b through 6-4d, the values of a and K for AMI binding to the DOPG liposomes were used to estimate AMI, IMI, and DOS binding by pegylated, anionic liposomes. The fits were all reasonable, suggesting that pegylated and unpegylated liposomes exhibit similar interactions with t he drugs, and that the different TCAs exhibit similar interactions with the liposomes due to similarities in their molecular structures. 6.3.3.3 Model fits to data for bupivacain e The overall mass balance for BUP is iP 0 PpC Sdy1y)) exp( kT exp(SCC e, (6-10) where Cp = 0.86C is the bulk concentration of t he protonated form of the drug [123]. The specific drug binding of BUP to lipid bilayers can be modeled as a sum of two isotherms, one each for the protonated and t he unprotonated forms. The concentration of the protonated fo rm in the bulk, Cp, is about 6 times that of the unprotonated form, C, and the negative zeta potential further enhances the concentration of the protonated form near the liposome surface by a factor of ee / kT. Thus, the binding contribution from the unprotonated form can be neglected, and the specific binding is given by the following equation: 148

PAGE 149

Tk p Tk p p peCK eCa R)(rCK R)(rCa e e (6-11) where a and K are the param eters of the isotherms fo r the protonated form. BUP binding to anionic, pegylated liposomes wa s significantly lower than for TCAs, and fitting the binding data to the model (Figur e 6-4e) yielded an a/ K ratio of 3.68 x 10-7 6.9 x 10-9 m. The concentration range tested for BUP was within the linear range, and so determination of individual values of a and K was not possible. The a/K ratio for BUP was about 2 orders of magnit ude lower than for TCAs, proving that BUP has a significantly lower affinity for liposomes, lik ely stemming from the reduced proportion of charged BUP compared to TCAs, and its inabili ty to associate with the charged lipid heads once inside the bilayer. Note that the pK shift for local anesthetics inside lipid bilayers is minimal, leading to a similar proportion of charged drug in the bilayer and the bulk [137]. 6.3.3.4 Effect of surface charge on isotherm parameters The presence of charge on the liposome surface likely contributes to increased drug binding through two mechanisms. First, the negative surface potential leads to an elevated dr ug concentration near the surface, and thus an increased concentration in the bilayer. This effect is evident in the exponential factor in Equation 6-9. Second, the adsorbed molecules could still interact wit h the charged surface, resulting in an increased partitioning reflected in the bi nding isotherm parameter s a and K. If the increased binding is purely due to the first effect, i.e., the enhanced surface 149

PAGE 150

concentration, the binding param eters a and K should be relatively similar for both charged and uncharged liposomes. In Chapter 2, we observed up to 99% of total AMI in so lutions bound to liposomes composed of 50:50 DMPC:DOPG. The DOPG liposomes carry a negative zeta potential (-50.9 mV) which is significantly larger in magnitude compared to DMPC liposomes (-6.6 mV) at 0.165 M ionic strength, whereas the 50:50 DMPC:DOPG liposomes more closely resemble the DOPG system (-42.7 mV). The drug uptake data for 50:50 DMPC:DOPG liposomes was fitted with Equation 6-8 (Figure 6-4f) and new values for a and K were obtained (a = 6.43 x 10-7 4.2 x 10-8 mol/m2, K = 5.90 x 10-2 5.5 x 10-3 mol/m3). The area per molecule for t he 50:50 DMPC:DOPG liposomes was assumed to be the mean of the area per molecule of pure DMPC and pure DOPG liposomes. The values of a for both liposom e types were similar, whereas K was three times larger for the mixed system, impl ying that the 50:50 DMPC:DOPG liposomes have a lesser affinity for the drug in the dilu te regime. The value of maximum uptake is similar for both systems, but the saturation occurs at a higher concentration for the 50:50 DMPC:DOPG liposomes due to the lower affinity. The value of a was most likely similar due to the fact that DOPG and DMPC lipids are well-mixed in these systems [138], and the maximum drug to lipid ratio obse rved in these experiments was only 0.12. Additionally, the value of a could be limited by the size of the drug molecules and interactions between bound drug molecules. The binding of IMI and BUP to DMPC li posomes was simulated using the a and K values obtained with purely anionic liposom es previously mentioned and the measured value of zeta potential (Table 6-1). A good fi t to the DMPC data with the same a and K 150

PAGE 151

values would imply that increased drug affinity for anionic liposomes stems purely from increased c oncentrations near the liposomes, and that bound molecules do not interact with the charged surface. Conver sely, if the fitted values fail ed to accurately predict the drug uptake, it could be viewed as strong evidence for direct interaction between charged drug sites and lipid head groups. The drug uptake results and predicted values for TCAs are compared in Figure 6-5a. Clearly, the predictions were not good, confirming that cationic TCAs interact with DMPC and DOPG differ ently. In Figure 65b, the a/K ratio previously obtained for BUP and anionic liposomes was used to predict BUP binding to DMPC vesicles. Again, t he model significantly over predicted the amount of drug bound, and points to intera ctions between bound BUP molecules and lipid head groups. However, the discrepancy between the model and the data for BUP is slightly less than that fo r IMI, supporting the previous hypothesis that BUP interacts with the charged lipid head groups to a lesser extent than TCAs. 6.3.3.5 Model validation via salt effects The model proposed above can be used to pr edict the effect of increased ionic strength on binding without introducing any ne w parameter. The fitted a and K values for TCAs and BUP from above were used to predi ct drug binding with anionic vesicles under such conditions. The liposome zeta po tential change (Table 6-1) was estimated via surface potential calculations using the following equation [135]: ])nTk (8*[ sinh Tk2 2 1 1 e, (6-12) where is the surface charge density expressed as a function of A0, the surface area per charge, in the following way: 151

PAGE 152

oe *, (6-13) and n is the number of ions per cubic mete r in the bulk, calc ulated by the equation below: IN1000nA, (6-14) where I is the ionic strength of the m edium. Based on Equation 6-12 and published experimental results [125,126], the zeta potential of the liposomes decreases by about 50 mV at an ionic strength of 1.5 M. In Figures 65c and 6-5d, experimental and predicted binding is compared for IMI and BUP. The calculated values for the amount of each drug bound agreed well with the experi mental values. The model demonstrates that increasing the ionic str engths of the solutions reduc ed the amount of drug bound by reducing the bulk drug concentration in t he immediate vicinity of the liposomes. Fisar et al. also fitted binding data for IMI to both charged and uncharged liposomes and showed that neutral phosphatidylcholine (PC) liposomes were less effective at binding to IM I than anionic phosphatidylserine (PS) vesicles [57]. Somewhat similar values fo r the parameters denot ed herein as a and K (a = 9.4 x 10-7 mol/m2, K = 0.12 mol/m3) were reported for PS vesicles and IMI in their study. They attributed high IMI binding to electrostatic interactions in concurrence with van der Waals and hydrophobic interactions. Despite t hese similarities, our systematic variance of lipid charge and bilayer fluidity and structur e, in combination with the inclusion of PEG-modified lipids, as well as our direct comparison between BUP and TCAs and the use of dye leakage to augment the binding and modeling results, allowed us to gain a deeper understanding of the drug-liposome interaction details. 152

PAGE 153

6.3.4 Summary of the Proposed Mechanism of Drug Sequestration To summarize the findings from this chapter, TCAs are attracted to anionic liposomes by electrostatic interactions and as sociate with lipid bilayers so that their charged regions interact with the char ged phospholipid head groups, while their lipophilic, aromatic regions asso ciate with the lipid bilayers. This is illustrated in Figure 6-6a (bilayer thickness and drug size is exa ggerated for clarity). Because of their structures, which include di stinct hydrophilic and lipophilic regions, they are able to associate with the bulk aqueous phase, the charged phospholipid head groups, and the lipid bilayers. BUP, however, with a more ce ntrally located charge, sits within the lipid bilayers as depicted in Figur e 6-6b. It is again attrac ted to the vesicle through electrostatics and is also positioned so that charge-charge interactions are maximized, but interacts with the bulk phase and the phos pholipid head groups to a lesser extent. The increased drug-liposome affinity for TCAs over BUP stems from tw o factors. First, the lower pKa of BUP when compared to TCAs results in a lower proportion of BUP molecules in the charged state. Second, t he unique structures of TCAs afford them very distinct lipophilic and hydr ophilic regions, allowing t hem to reach a lower energy state than BUP within lipid bilayers. Austin et al. and Deo et al. proposed similar mechanisms for cationic drug uptake by liposomes [55,72]. In both cases, c harge was identified as a key aspect for high affinity binding. Sanganahalli et al. comple ted fluorescence spectroscopic studies using phosphatidylcholine liposomes and confirm ed that both IMI and AMI penetrate lipid bilayers [58]. While studying the effect of BUP on model membranes composed of PC and phosphatidylethanolamin e (PE), Suwalsky et al. [59] found that BUP interacted with the membranes to a significant degree and proposed that BUP partitioned into lipid 153

PAGE 154

bilay ers. More recently, Mizogami et al. studied the effects of BUP enantiomers on PC membranes and found that BUP disorders lipid bilayers in cholesterol free membranes and selectively associates with membranes w hen cholesterol is present [60]. These results clearly support the drugliposome interaction details put forth in this chapter for both TCAs and BUP. In addition, our us e of a common drug binding model and liposome leakage experiments allowed for a direct comparison of the drugs bilayer conformations. Also, the use of MLL and ULL also provided new insight into the effect of PEG aqueous layers on inner bilayer surface charge. 6.4 Conclusions In this chapter, drug binding exper iments performed using filtration and centrifugation, calcein l eakage from liposomes, and a Langmuir binding model were utilize d to explore the mechanisms of drug binding to liposomes and anionic microgels. In spite of higher charge dens ities, anionic microgel particles bound significantly less drug than liposomes. The bind ing of tricyclic antidepressa nts such as AMI, IMI, and DOS was compared to each other and to the binding by the local amide anesthetic bupivacaine. The mechanisms, as validat ed with binding isotherm parameters, are similar for all TCAs but are si gnificantly different for BUP. The experiments and modeling in dicate that the amount of drug molecules in the electrical double layer is negligible but the electrostatic effects pl ay a major role in binding. The electrostatic interactions are responsible for the initial association between antidepressants and liposomes, whereupon the dr ug enters the bilayer with its charged region closely associated with the charged lipid head groups and its lipophilic region closely associated with the lipid bilayers. BU P, which is predominantly in the protonated state at pH 7.4 and 25C as well (86%), is also preferentially attracted to the charged 154

PAGE 155

vesicle [123]. Once bound, it appears to be located within the bilayer to a greater extent than the antidepressants with minimal inter a ctions with the charged surface. The structure of BUP makes it less able to access the bulk aqueous phase surrounding or encompassed by the lipid bilayers. The conclusions regarding the differences in conformations between antidepressants and the anesthetic bupivacaine are supported by the fact that bupivacai ne binding leads to a much lower increase in liposome permeability compared to antidepressant binding. Drug binding is similar for both pegylated and unpegylated liposomes because the PEG layer is sufficiently porous to allow rapid drug diffusion. For both the pegylated and unpegylated cases, as well as both drug types, more drug is bound to multilamellar rather than unilamellar liposomes. This difference vanishes at high ionic strengths. We attribute the increased binding for multilamellar liposomes without PEG to enhanced electrostatic interactions between adjacent charged layers. For pegylated liposomes, the reduced dielectric constant in the aqueous-PEG layer between adjacent bilayers results in a lower energy barrier for cati onic drug transport across the aliphatic tail region [130,132-134]. 155

PAGE 156

Figure 6-1. Structures of the drugs a) amitriptyline and b) bupivacaine and the phospholipids c) DOPG, d) DMPC, and e) DMPG used for studying drugliposome interactions. 156

PAGE 157

Figure 6-2. Percent of IM I bound to DOPG (anionic) UL L at 1.44 mg lipid/mL in solutions of 0.165 M ( ) and 1.5 M () ionic strength and to DMPG (anionic) ULL at 1.44 mg lipid/mL in solutions of 0.165 M ionic strength (). Percent of AMI bound to DOPG (anion ic) ULL () and MLL ( ) at 0.72 mg lipid/mL in PBS with 95% confidence intervals show n. Percent of BUP bound to 95:5 DOPG:DPPE-mPEG-2000 (anionic) ULL at 1.44 mg lipid/mL in solutions of 0.165 M () and 1.5 M () ionic strength. Percent of IMI bound to DMPC ULL at 1.44 mg lipid/mL in PBS (). Percent of BUP bound to DMPC ULL at 1.44 mg lipid/mL in PBS (). Percent of BUP bound to 95:5 DMPG:DPPEmPEG-2000 (anionic) ULL at 1.44 mg lip id/mL in PBS (). Percent of BUP bound to 95:5 DOPG:DPPE-mPEG-2000 (anionic) MLL at 1.44 mg lipid/mL in solutions of 0.165 M ionic strength () and 1.5 M ionic strength (). Data are reported as mean standard error of t he mean with n = 2, except for AMI binding data where all data points are shown and 95% CI displayed. 157

PAGE 158

Figure 6-3. Percent of entrapped calcein released by 95:5 DOPG:DPPE-mPEG-2000 ULL 10 minutes after exposure to AMI, IM I, or BUP at drug concentrations of 0 ( ), 8.6 (), 38 (), and 150 () M. Means are show n with n = 2-4. Marker (*) denotes P < 0.05. 158

PAGE 159

Figure 6-4. Concentration of a) AMI bound to DOPG ULL at 0.72 mg lipid/mL versus unbound AMI, b) AMI, c) IMI, and d) DOS bound to 95:5 DOPG:DPPEmPEG-2000 ULL at 0.72 mg lipid/mL versus unbound drug determined experimentally ( ) or estimated from Equation 6-8 () with a = 5.36 x 10-7 moles/m2 and K = 1.88 x 10-2 moles/m3, e) BUP bound to 95:5 DOPG:DPPEmPEG-2000 ULL at 1.44 mg lipid/mL versus unbound BUP determined experimentally ( ) or estimated from Equation 6-8 () with a/K = 3.68 x 10-7 m, f) AMI bound to 50:50 DOPG:DMPC ULL at 0.72 mg lipid/mL versus unbound AMI determined experimentally ( ) or estimated from Equation 6-8 () with a = 6.43 x 10-7 moles/m2 and K = 5.90 x 10-2 moles/m3. 159

PAGE 160

Figure 6-5. Concentration of a) IMI or b) BUP bound to DMPC ULL at 1.44 mg lipid/mL in 0.165 M ( ) ionic strength solutions determi ned experimentally and c) IMI or d) BUP bound to DOPG ULL at 1.44 mg lipid/mL determined experimentally in 0.165 M () or 1.5 M ( ) ionic strength solutions. Predictions were made using Eq uation 6-8 () with a = 5.36 x 10-7 moles/m2 and K = 1.88 x 10-2 moles/m3 for IMI and a/K = 3.68 x 10-7 m for BUP. See Table 6-1 for corresponding changes in zeta potential. 160

PAGE 161

A B Figure 6-6. Plausible mechanism for a) TCA and b) BUP binding to pegylated, anionic liposomes. For TCAs, charged li pid head groups associate with the protonated portion of the drug (depicted in gray, molecular structure also shown) and the bulk aqueous phase, whil e the uncharged portion of the drug associates with the hydrophobic lipid bilayer. For BUP, the drug is predominantly located within the lipid b ilayer with its charged region oriented towards the charged lipid head groups. 161

PAGE 162

162 Table 6-1. Values for cons tants and parameters used for dr ug binding predictions from Equations 6-6 through 6-12 ( lipid loadings are specified in figure legends) Constant Value Range of Observed Values Units A a 6.86 x 10-19 m2 / lipid A b 6.22 x 10-19 5.96 6.52 x 10-19m2 / lipid A c 6.54 x 10-19 m2 / lipid MW (DOPG) 797.04 g / mole MW (95:5 DOPG:DPPE-mPEG-2000) d 794.66 g / mole MW (DMPC) 677.94 g / mole MW (50:50 DMPC:DOPG) 737.49 g / mole (0.165 M) e 1.33 x 109 1/m (1.5 M) e 3.33 x 109 1/m e 1.6 x 10-19 C k 1.38 x 10-23 m2 kg s-2 K-1 T 298 K f 6.951 x 10-10 C / ( V m ) (DOPG 0.165 M) g -50.9 2.63 -60 to -65 mV (DOPG 1.5 M) h 0 mV (50:50 DMPC:DOPG 0.165 M) i -42.7 4.82 -40 mV (DMPC 0.165 M) j -6.6 7.56 0 to -20 mV aAverage area per DOPG lipid molecule on the liposome surface at 30C. The same value was used for 95:5 DOPG:DPPE-mPEG-2000 liposomes [139]. bAverage area per DMPC lipid molecule on the liposome surface at 27-30C [127,128,140]. cAverage area per lipid molecule on the liposome surface for 50:50 DMPC:DOPG li posomes at nearly 30C, calculated as the average of the pure values for DMPC and DOPG. dThe molecular weight of pegylated liposomes was calculated excluding the PEG chains. eInverse of the Debye length at 0.165 M and 1.5 M ionic strengths [135]. fProduct of the dielectric constant and the relative static permittivity of water (78.54) used for surface potential calculations for ULL. gMeasured zeta potential (mean standard error) of DOPG liposomes in PBS (0.165 M ionic strength). Observed values were measured at ionic strengths of 0.1 M [126] and 0.165 M [125]. hThe zeta potential change from about -50 mV to 0 mV when increasing the ionic strength to 1.5 M was estimated from the surface potential change calculated with Equation 6-12 and has also been noted in literature [125,126]. iMeasured zeta potential (mean standard error) of 50:50 DMPC:DOPG liposomes in PBS (0.165 M ionic strength). The observed value was measured at an ionic strength of 0.1 M [126]. jMeasured zeta potential (mean standard error) of DMPC liposomes in PBS (0.165 M ionic strength). Obse rved values were measured at ionic strengths of 0.1 M [126] and 0.165 M [95].

PAGE 163

CHA PTER 7 PREDICTING THE EFFICACY OF AMITRI PTYLINE AND BUPIVACAINE OVERDOSE TREATMENT WITH LIPOSOME S IN MA N WITH PHYSIOLOG ICALLY BASED PHARMACOKINETIC MODELS 7.1 Introduction Chapters 1-6 focused on developing and understanding high affinity binding between liposomes and drugs. In this chapt er, physiologically based pharmacokinetic (PBPK) models were constructed to simulate the treatment of AMI and BUP overdoses with liposomes to assess the effect of lipos ome administration on dr ug redistribution and possible toxicity reversal. A preliminar y model framework and discussion concerning requirements for nanoparticles capable of treating overdoses was formerly published, but lacked the details or in vitro data requir ed to extract specific information [117]. To develop our PBPK models, mechanistically based equations for tissue to blood partition coefficients developed by Rodgers et al. [106] were combined with in vitro drug binding data from previous chapters and drug specific input parameters from literature. The partition coefficients were validated by compar ing model predictions to intravenous (IV) data, followed by best fits to non-IV data to ob tain first-order absorption rate constants. AMI and BUP overdoses and the efficacy of lipos omes at reversing the overdoses were then simulated by including liposomes into the mass balances of the PBPK models and utilizing the in vitro drug-liposome binding da ta. Simulations were also conducted to evaluate the model sensitivity. Finally, published in vitro data on the function of cardiac ion channels [9,16,141] and atrial contractility [10] was utiliz ed to estimate, to a first approximation, the molecular level pharmacody namic effects in the heart resulting from liposome therapy. 163

PAGE 164

7.2 Methods 7.2.1 Obtaining Tissue Partition Coefficients A precursor to developing a useful PBPK m odel is obtaining reasonable values for the organ to blood par tition coefficients for the drugs or compounds of interest. This is often done experimentally using animal or post-mortem data. However, the values reported in such studies can vary widely from study to study and/or species to species, and many organs are typically omitted due to practical limitations. Furthermore, in cases of new compounds, studies to measur e the partition coefficients could be prohibitively expensive. Poulin et al. began the process of overcoming this problem by developing methods for predi cting partition coefficients based on compound properties [73,142-144]. Rodgers et al. subsequently deve loped additional me chanistic equations for acidic and basic compounds [106,145]. Their equations for moderate-to-strong bases were utilized here to predict the partition coefficients for AMI, IMI, and BUP [106]. The mechanistic equations were developed for partition coefficient predictions in rats [106], and we have assumed the values are reasonably transferrable from rats to humans. The drug partition coefficients between the tissues and the concentrations unbound in plasma, Kpu, were calculated ut ilizing the association constant for blood cells, KaBC, as detailed by Rodgers et al. [106]. The compound specific parameters necessary for the calculations are shown in Table 7-1 [106-108,11 9,121,123,146-158]. Table 7-2 [85,150,151,158,159] shows the calculated Kpu values for the drugs modeled. Note that TCAs ar e highly lipophilic and thus ve ry likely to partition into organs. The integration of Kpu values into our model is outlined bel ow. However, for a 164

PAGE 165

detailed review of the equati ons and parameters necessary fo r the Kpu calc ulations, the reader is directed to the wo rk of Rodgers et al. [106]. For the PBPK model developed herein, we hav e assumed steady state conditions between tissue and blood concentrations (i.e. bl ood-flow limited transpo rt). Kpu values can be related to Cf, the free concentration in the blood, and H, the hematocrit, in the following way: f t pu, tC H)(1C C C Kpu (7-1) where Ct is the concentration in the tissues and Cu,p is the concentration unbound in the plasma. The (1-H) factor is added to co rrect for the discrepancy in volume between blood and plasma [160]. The PBPK equations de scribed in the next section require the partition coefficients between the tissue and t he whole blood, including proteins and red blood cells (RBC). Additionally if the blood contains any other component that can bind substantial amounts of drug such as liposomes, the effect on the partition coefficient also needs to be accounted for. To descr ibe the binding between drugs and proteins, liposomes, and red blood cells (RBC), we define the partition coefficients K1, K2, and K3 as f RBCb, 3 f lipb, 2 f pb, 1C C K, C C K, C C K (7-2) where Cb,p, Cb,lip, and Cb,RBC are effective drug concentrations in blood bound to proteins, liposomes, and RBCs, respectively, based on the blood volume. A new parameter, Keff, is defined as follows: 3f2f1ff tKCKCKCC C bloodwholeinion concentrat drug tissuesinion concentrat drug Keff (7-3) 165

PAGE 166

By examining Equations 7-1 and 73, Keff can also be defined as )KKKH)(1(1321. Kpu Keff (7-4) The values of K1 and K2 shown in Table 7-3 were obtained using experimentally measured drug binding to liposome in ion must be s that lues ious s and proteins reported in Chapters 2-5. The values of K1 and K2 depend linearly on the amount of protei n and liposomes, respectively the blood. The concentration of protein in the blood is relatively constant, and so a constant K1 that corresponds to physiological conc entrations of protein can be used in the simulations. However, liposomes introduced into the blood through IV administration are cleared, and so the concentration of liposomes in blood is time dependent. Accordingly, the dependence of K2 on the liposome concentrat included in the simulations. Experimental data on drug binding to liposomes show K2 depends linearly on the liposome concent ration, which is expected based on mass balance considerations. Accordingly, a linear dependence of K2 on the liposome concentration is utilized (Table 7-3). T he drug binding to RBCs that is needed to determine K3 was not directly measured, but was obtained by utilizing reported va of blood-to-plasma drug concentration ratios (B :P), which can be related to the var partition coefficients through a mass balanc e to yield the following equation: H)(1 CKC CKCKCKCf1f f3f2f1f P:B (7-5) The K3 values shown in Table 7-3 were ca lculated by solving Equation 7-5 using reported B:P ratios (Table 7-1) with K2 in the cited studies in which B:P wa s measured. Table 7-3 shows that K1 was taken to going to zero, since no liposomes were present 166

PAGE 167

be independent of drug concentrations fo r TCAs and BUP. This is a reasonable assumption for TCAs across a fairly broad range of drug concentrations (see Chapters 2-4) and for BUP (Chapter 5) within the low concentration range observed during adverse reactions (< 1 g/mL). BUP concen trations are often low because ov erdo not intentional, as it occurs in a clinical setting. To validate the Keff values obtained from the methods illustrated above, K1 a values for AMI and BUP were used to calculate Keff values in the absence of lipos omes. Comparisons were then made to tis se is nd K3 sue to blood partition coefficients repor e 7-1 used om The PBPK model included venous and ar teri al blood compartments and 13 additional organs (Figure 7-2). A local equilibrium between tissue and blood ted from various animal or post-mortem human studies [161-167]. Comparisons were only made for the organs for which data was available in literature. Figur compares the experimental and calculated va lues for AMI and BUP, the two drugs for the overdose simulations. The partition c oefficients agree reasonably well for most of the organs. Experimental error is not shown due to the lack of error estimation in some of the studies with small sample sizes. However, when taking into consideration the large interand intra-species variations as sociated with tissue to blood partition coefficients, as well as the level of difficu lty of such experim ents, the experimental values clearly support the validity of the mechanistic equations developed by Rodgers et al. [106] for AMI and BUP. Moreover, t he larger than predicted values observed in the experimental studies for organs such as the lungs and liver could have arisen fr post-mortem drug redistribution, an occurr ence cited in several published reports [163,168]. 7.2.2 PBPK Model Structure 167

PAGE 168

concentrations (i.e. blood-flow limited trans port) was assumed for all organs. All equations were solved using MATLAB sof twa re (The Mathworks, Natick, MA, USA). rate s to organs (denoted by i), organ volum es, and drug concentrations of organs, respectively. Subscripts denote organs as follows: venous eart (ht), le (m), ) Qi, Vi, and Ci refer to blood flow whole blood (vb), arterial whole blood (ab), l ungs (l), adipose (ad), brain (br), h pancreas (p), gut (g), spleen (sp), liver (h), thymus (th), bone (bo) skin (sk), musc and kidneys (k). Hepatic and renal clearance values are denoted as CLh and CLr, respectively. The arterial blood flow directly to t he liver (ha) is denoted as Qha. The mass balance equation for non-eliminating tissue (i = ad, br, ht, p, g, sp, th, bo, sk, mu was as follows: ) Keff C (C V Q dt dCi i i ii. Liver concentrations were obtained from ab (7-6) h h h sp g p hC Q)E)(1 C Q C Q C QC(Q dt dC hh sp sp g g p pabhaV Keff Keff Keff Keff, (7-7) where the hepatic extraction ratio, Eh, was defined as [73,108,169] h h h hCLQ CL E (7-8) For the kidneys, Equation 7-9 was solved: kk kr k k ab k kkKeffV CCL ) Keff C (C V Q dt dC (7-9) For the lungs, Equation 7-10 was solved: ) Cl Keff (C V Q dt dCl vb l ll. (7-10) 168

PAGE 169

The venous blood compartment was represented by Equation 7-11: vbV dtkm,sk,bo,th,h,ht,br,ad, i i vbCQ Ke C Q dC (7-11) while the arterial blood was represented by Equation 7-12: vbl iff )C KeffVdtlab. C ( Q dCab ll ab (7-12) 7.2.3 Model Input Parameters Organ blood flow rates, organ volumes, and total cardiac output for humans were obtained from various sources [78,79,144,170-172] 7-9 and bioavailable drug fractions after oral dosages (F) were obt ained from human studies and are shown in Table 7-5 [73,108,148150,169,173-185]. H ere weighted according to the number of subjects for parameter values pooled from multiple studie s. 7.2.4 Validation of Model Paramete rs by Predicting Intravenous Data and dosage protocols (i.e. were then predicted using the PBPK model described above. The PBPK model results were then and are shown in Table 7-4. Renal and hepatic clearance values used in Equations 7-7 through ematocrit values were taken to be 0.45 for all simulations [181]. Averages w To prove the reliability of the parameters utilized here, PBPK simulations were conducted to simulate various experim ental studies where AMI [149,175,176], IMI [173,174], or BUP [151,177-179,186] was given intravenously. IMI was used for model validation but was omitted from drug overdose simulations due to the very similar nature of AMI and IMI. Reported subject weights, drug doses infusion time, etc.) were utilized in the si mulations, and drug concentrations compared to the corresponding data sets. Note that no fitting parameters were utilized 169

PAGE 170

for model validation, as all PBPK parameters were obt ained by procedures described above. 7.2.5 Determining Absorption Rate Constants b y Fitting to non-Intravenous D IV data afforded PBPK model and human dat a comparisons in the absence of fitting parameters. TCAs are, however, almost exclusively prescribed and overdosed on orally. To account for oral administr ation, an additional compartment was added to the PBPK model which fed into the venous blood compartment to simulate drug absorpti ata on from the gut [169,187]. The mass of drug re maining in this additional -13) where D is the drug dose. The rate of drug absorption from this additional compartment into the IV compartment was calculated as kaFX, where F is the bioavailability and ka is the first-order absorption rate cons tant. To obtain values for ka, least squares fits were done for several oral data sets for AMI [ 149,188-191] and IMI [109,173 ]. IMI oral data sets were used solely for model validation, since IMI overdose simulations were not conducted. The avera f F was assumed to be 1 for all BUP models, since first compartment X(t) is given by tkaDet)(X, (7 ge of the ka values obtained from the AM I fits was used for the AMI overdose simulations. Unlike TCAs, BUP is typically given through non-oral, non-IV routes that include injections or infusions into the dural intraperitoneal, or intercostal spaces, among others [192-198]. In many cases, short infusions are administered that can be modeled as bolus doses into an additional compartment that feeds into the venous blood, as described by Equation 7-13. This was done for several cases to obtain ka values for BUP [192-196]. The value o 170

PAGE 171

pass ce was to elimination is largely circumvented with non-oral dosage procedures. More lengthy or complex infusion pr oc edures [197,198] were also simulated for least squares fits using a single absorption rate constant, ka, where an additional compartment was again added that released drug into the venous blood. The following mass balan integrated from time zero (to) to the end of the first infusion (t1) for an infusion rate I1 in a non-IV compartment cont aining mass X at time t, XkI dX (7-14) to yield Equation 7-15, the amount of drug remaining in the compartment between to an t1: dta1d )e(1 k I X(t)tk 1aa. (7-15) For the ith additional infusion rate given after I1, the mass balance was solved to produce ) I )(X(t e I X(t)i 1i t)(tk i1ia k ka a, (7-16) for ti-1 to ti, where ti-1 and ti represent the time points at which the previous and ith infusions stop, respectiv (7-17) The above equations were solved to obtain X(t) and the rate of drug absorption into the IV compartment was again calc ulated as kaX. Although many BUP data sets were used for least squares fits to dem onstrate the models ability to predict non-oral, non-IV data with a single absorption rate constant (ka), the dosage methods varied widely ely. Finally, the am ount of drug in this compartment after the final ith infusion was obtained from Equation 7-17 for ti to t: X(t)t)(tk iiae)X(t 171

PAGE 172

throughout the non-oral data sets us ed, making an overall average ka value meaningless. Consequently, ka values obtained from fits to three studies with dural space dosages [192,194,197] were averaged to simulate BUP overdose treatment. 7.2.6 Drug Overdose Tr eatment Simulations AMI [7,110,111,115] and BUP [192-196] dos es for simul ating overdoses were based on values reported in literature for ov erdoses and/or adverse reactions. AMI overdose was modeled as an oral dose and BUP administration was modeled as a bolus injection into the dural space. Liposome doses were based on in vitro data of drug binding to liposomes (Chapters 2-6) and were modeled as bolus IV doses. Bolus administration is a reasonabl e approximation because the lipid concentrations in liposome formulations [199] can be as high as 50 mg/mL, allowing rapid delivery of the required amount of lipids in small fluid volu mes [12,14,15,40]. The total liposome doses simulated were deemed clinically As explained earlier, the value of K depends linearly on the liposome concentration in the blood (Table 7-3). It is thus necess ary to determine the liposome safe based on reported studies [25,200,201]. The amount of time expiring between drug inges tion or administration and liposome dosing (tlag) was based on the time required for TCA patients to reach the hospital [7,110112,202] after ingestion for AMI, and the time required to administer lipid emulsion therapy during adverse reactions for BUP [12-15]. Drug doses, liposome doses, and tlagvalues are given in Table 7-6. The cardiac output and organ volumes for overdose simulations were based on a patient wei ght of 70 kg. Note that several tlag values and drug doses were simulated for AMI due to the large variability in lag times and drug ingested during TCA overdoses. 7.2.7 Liposome and Drug Clearance 2 172

PAGE 173

concentration in the blood, which decreases with time due to clearance. Several important assumptions were made about the lipos ome fate in the body. First, the size and pegylated nature of the liposomes studied in previ ous chapters and simulated h make them stable and unlikely to traverse capillaries [26,90 ]. Therefore, the liposomes are assumed to remai n well-mixed in the venous and arterial blood compartments un elimination. Finally, the elim inatio ere til n of the liposomes has been assumed to be similar to som es observed in rabbits [30]. The following equa ments as a (7-18) The time dependent value of K2 was calculated by multiplying the percent of liposomes remaining at time t from E quation 7-18 by the initial liposome concentration in the blood stream from Table 7-6 and pl ugging the corresponding liposome concentration into the equation for K2 shown in Table 7-3. Note that prio r to the liposome administration, i.e., for t < tlag, the value of K2 is zero and increases suddenly upon liposome administration at t = tlag. Rather than assuming a step increase in K2, we have assumed a linear increase in K2 over 1 minute from zero at t = tlag to the value corresponding to the equation shown in Table 7-3 for K2. Practically speaking, this assumes that the liposomes become well-mixed and reach equi librium d the eliminat ion of pegylated, anionic lipo tion specifying the percent of liposomes remaining in the blood compart function of time (in hours) over a 22 hour period was fit from data by Awasthi et al. [30],100 +7.8865t 0.2354t Remaining Liposomes %2rug binding within the blood compartments after 1 minute. The 1 minute duration is suffi cient to achieve equilibrium between liposome and free drug based on in vitro studies [71]. Liposomes cause the drug concentrati ons in the venous and arteri al blood compartments to increase, which causes hepatic extraction of the drug to increase. The 173

PAGE 174

presence of liposomes in the blood could potent ially alter the hepatic extraction ratio, but since no data is available for this effect we assumed that the hepatic extr action is unaltered by the liposome addition. Effe ctively, hepatic extraction in the model i based on the total drug concentration in the blood including drug bound to liposo as opposed to the free drug concentrati on. This assumption could potentially overestimate the hepatic drug clearance. On the other hand, estimating the clearance of the liposomes as described above (Equation 7-18) implicitly a ratio s mes, ssumes that the liposo them sociated hepatic on of rding many others varies from individual to i ndividual [178,185,203,204]. Furthermore, many mes simply disappear fr om the blood stream and rel ease the drug bound to over time. A large portion of the liposomes actually end up in the liver [30], thereby potentially accelerating drug metabolism due to metabolism of the bound drug. Neglecting this increased metabolism l eads to an underestimation of the drug clearance, which partially offsets the overestimation caused due to the neglect of changes in hepatic extraction ratio upon li posome addition. Although no quantitative analysis to prove the effects roughly offset one another is possible, the error as with neglecting such competing effects is at least reduced. While including these effects in the PBPK model is theoretically possi ble, it would require details about metabolism of drug bound to liposomes, wh ich is not available in literature. 7.2.8 Metabolites Both AMI and BUP ar e converted to active metabolites in the body. Inclusi metabolites in the model is in principal stra ightforward, but requires details rega transformation and clearance rates, tissue partition coefficients, and protein and liposome binding, which are not available. Additionally the extent to which AMI [150,182,188] or BUP [178,185,203] is converted to one particular metabolite versus 174

PAGE 175

metabolites such hydroxymetabolites for AMI [150,182,188] and pipe colyxylidine ( for BUP [185], are much less lipophilic than t heir parent drugs, making their effects on the body, their organ partitioning and elim ination, and their liposome binding PPX) t from the parent molecules. Due to the reasons listed above, meta lso, d into ug e se. r is was d n overdosed patients. While this issue could introduce some errors into the model, it is signific antly differen bolism is not included in the model. Neglecting metabolism may not be critical because metabolism is the bodys natural way of preparing drugs for renal elimination making metabolites almost exclusively less dangerous than their parent drugs. A the large volume of distribution for AMI dictat es that most of the drug is partitione tissues over the first 5 to 10 hours of over dose, which makes metabolite tracking less important during the most crit ical moments for treatment. Additionally, the protein and liposome binding of AMI and BUP in the PBPK model ar e drug and liposome concentration independent, which makes the model valid even as metabolites are present. Finally, BUP adverse reactions are treated within 5-15 minutes of dr administration. The short time lapse between drug and liposome dosing reduces th amount of time available for me tabolite formation, making it le ss significant in this ca 7.2.9 Drug Dose in Overdose Simulations The doses simulated varied from 100 mg to 2500 mg. 2500 mg is clearly outside of the therapeutic dose range fo r AMI. The parameters obtained from clinical studies fo AMI, such as clearanc e and bioavailability, are perhaps less accurate at drastically increased AMI concentrations. Ideally, data taken from patients undergoing AMI or TCA overdoses would be compared to model predicti ons to refute this position. Th practically impossible due to the inexact nature of reported drug doses ingested an time lapses occurring between ingestion and drug concentration measurements i 175

PAGE 176

likely not critical for the following reasons. First, Venkatakrishnan et al. [204] have reported in vitro data suppor ting linear AMI metabolism bel ow AMI concentrations of around 200 M or 55 g/mL, well below all concentrations reached in the AMI overdose simu ne lations Next, reduced F values owing to self-inhibited AMI absorption, gastric lavage, or charcoal administration would not reduce the efficacy of the treatment, as evidenced by the effectiveness of the treatment at both high and low drug concentrations (see below). Finally, the effect of changing parameters like ka, F, and others has been directly addre ssed in Section 7.2.11. 7.2.10 Validation of the Numerical Calculations As a check for numerical accuracy, a conservation of mass calculation was do for the overdose simulations. The drug concentrations in each organ at a final time point were multiplied by organ volumes to ob tain the drug retained within the o rgans The total drug metabolized was calculated as shown below, f ot t i iKeff C, (7 with Ci representing the drug concentrations in the gut (g), pancreas (p), and spleen (sp). Renal elimination wa s obtained from Equation 7-20: iabhahdtQCQEdMetabolize Drug-19) dt Keff Cf ot t k rClnEliminatio Renalk. (7-20) rbed from the stomach or the dural spac AMI lost to first pass met abolism and drug not abso e by the final time point was also incl uded in the mass balance. The total mass in the system was at least 99.99% of the original mass at all times in all simulations, proving the numerical accuracy of the calculations. 176

PAGE 177

7.2.11 Sensitivity Analysis The model predictions for the effect of liposom ra and inter-subject variation in humans. While a detailed population PK m odel is beyond the scope of this work, the effect of parameter variati ons on treatment efficacy was explored by varying the chosen d erage ka value calculated from best 88-191] and non-IV BUP data [192,194,197]. Liposomedrug s e the sensitivity analysis for hepatic extraction, absorption, liposome-drug binding, and es during an overdose obvious ly depend on the various model paramet ers, which are uncertain due to variability in the studies from which they were obtained and t he inherent int parameter in a range and explorin g its effect on the results of the simulations. Hepatic extraction variation was based on maximu m [149,178] and minimum [176,179] average values observed from single st udies. Maximum and minimum ka values for AMI an BUP were based on the standard deviation of the av fits from oral AMI data [149,1 binding was reduced by reducing K2 values. K2 values were not increased since this would only increase the efficacy of the treatment. St andard deviations for K2 value from data reported in prev ious chapters were low and would not have afforded a meaningful sensitivity anal ysis. Therefore, K2 values at each liposome concentration were reduced by 20% and new curves for K2 as a function of liposome concentration were generated. The percent of liposomes remaining in the blood stream reported by Awasthi et al. [30] for pegylated, anionic liposomes in rabbits was reduced by 30% at each time point and a new curve generated. Once more, the standard deviations reported in their study were much lower t han 30%, but a higher value was used to allow for a meaningful examination of how enhanc ed liposome elimination would affect overdose treatment with liposomes. Slower liposome elimination was not studied, sinc it would only improve the treat ment. A detailed list of the values and equations used for 177

PAGE 178

lipos ome elimination is presented in Table 7-7. Finally, the sensitivity of the mode Keff was examined by varying B:P [150,151] and fu [85,158,159] within clinically feasib ranges and computing Kpu and Keff with the modified values To be consistent, the modified values of B:P and fu were also utilized to compute K1 and K3. Refer to Table 7 2 for the exact B:P, fu, and Kpu values used for the analysis. 7.3 Results 7.3.1 Tricyclic Antidepressant Model Validation and Fits for A bsorption Model predictions and human data sets for IV TCA administration are compared in predictions. The PBPK models predicted the drug concentrations accurately. Error unavailable in many cases. Figures 7-4a through 7-4k show TCA oral data sets and PBPK best fits using k as a fitting parameter. Some predictions such as Figures 4c The k value obtained for AMI overdose simulati ons from Figures 7-4d through 7-4k was -1l to le Coefficients Figures 7-3a through 7-3j. No model paramet ers were allowed to vary to match the bars are shown for all studies for which they were origin ally reported but were aand 4d did not match the data well, but overa ll the model fitted the data successfully. a0.0930 0.0394 h (mean standard deviation). 7.3.2 Bupivacaine Model Validation and Figur d Fits for Absorption Coefficients Model predictions and human data sets for IV BUP administration are compared in es 7-3k through 7-3o. No model paramet ers were allowed to vary to match the predictions. The PBPK models predicted the BU P concentrations very well except at high BUP concentrations in Figures 7-3k and 7-3o. This discrepancy likely arose from reduced protein-BUP bindi ng at high BUP concentrations, as has been observe (Chapter 5, [158]). Over-pr edictions at high BUP concentrations during BUP infusions 178

PAGE 179

have also been observ ed elsewhere [179]. Fi gures 7-4l through 7-4s show BUP nonnon-oral data sets and PBPK best fits using ka as a fitting parameter. The fits were reasonably good. The average ka value obtained for BUP from Figures 7-4l, 7-4r, and 7-4s (dural spaces) was 0.1794 0.0790 h-1 (mea IV, n standard deviation). This value 7.3.3 e as heart, and the brain were and insignificant for tlag > 2 hours n se g of was used in the overdose simulations reported below. Tricyclic Antidepressant Overdose Simulations AMI overdoses were simulated according to the conditions specified in Table 7-6 and an average ka value of 0.0930 h-1. Note that all four lipos ome loadings and all thre lag times were simulated for the 2500 mg drug dose, whereas only one liposome dose and lag time was simulated for the other tw o drug doses. Table 7-8 shows the are under the concentration versus time curves (AUC) and peak drug concentrations for AMI for several key organs under a variety of different conditi ons. The percent reductions were calculated relative to ov erdose simulations without liposomes at the same drug doses. AMI AUC reductions in free venous blood, the heart, and the brain varied from 63.3% to 20.9%, 64.0% to 21.2%, and 64.0% to 21.2%, respectively. AMI peak concentration reductions in free venous blood, the 15.6%, 20.1% and 20.0%, respectively, for tlag = 2 hours since AMI peaks had basically been reac hed prior to liposome dosing. The reductions of AUC and peak concentration in the brain and heart were similar because both of these tissues are in equilibrium with the venous blood because of the high perfusion rates. Figure 7-5 shows AUC or peak AMI concentration values as a functio of liposome dose or tlag. Note that the liposome effica cy saturates as the liposome do increases. Figure 7-5d di splays the effect of tlag on AUC at a set liposome dosin 1.44 g/L. 179

PAGE 180

Figure 7-6 shows AMI concentrations vers us time for several key organs at a lipos ome dose of 1.44 g/L. In Figure 7-6, the control cu rve should be followed until the desired tlag is reached. The first alternate curve should then be followed for tlag = 2 hours, the second for tlag = 4 hours, and so on. Note that peak Cvb values increased due to liposome administration by three to four fold for all time lags. The muscle is included in Figure 7-6e to illustrate the effect of sink organs. Sink organs require such long periods of time to reach their peak dr ug concentrations that liposome doses w about 4 hours actually cause their drug conc entrations to momentarily increase above the control value before enhanced eliminat ion reduces the blood and therefore the organ concentration as well. Although this effe ct is clear in Figure 7-6e, its magnitude suggests i ithin t to be fairly insignificant compar ed to the amount of AMI held in the blood comp 7-6 ug he h lower than those for AMI, but could still be clinically artment by the liposomes. 7.3.4 Bupivacaine Overdose Simulations BUP overdoses were simulated according to the conditions specified in Table and an average ka value of 0.1794 h-1. Table 7-9 shows the AUC values and peak dr concentrations for BUP. BUP AUC reductions in free venous blood, the heart, and t brain varied from 15.5% to 8.5%, 15.4% to 8.5% and 15.3 % to 8.3%, respectively. BUP peak concentration reductions in free v enous blood, the heart, and the brain were 17.3% for the maximum li posome dose of 2.88 g/L. Figure 7-7 shows BUP concentrations versus time for several key organs at a liposome dose of 2.88 g/L. O wing to the much shorter tlag value of 15 minutes for BUP, the plots appear much different fr om those in Figure 7-6. Cvb increases and organ concentration decreases are muc 180

PAGE 181

signific ant. The muscle (Figure 7-7e) is again relatively unaffected by the liposomes within 7-3 f rect evidence to both the by Ro dgers et al. [106] and the significant oppo Best n w ith Liposomes Simulated liposome treatments MI concentrations. Table 7-8 and ases of up to four fold, immediate concentration drops upon lipos o e the first few hours due to the long equilibration times. 7.3.5 Sensitivity Analysis Results Tables 7-10 and 7-11 show the perc ent AUC and peak drug concentration reductions as a function of altered sensitivity analysis parameters for AMI and BUP, respectively. The values from simulations with the original param eters are also shown for comparison at the top of the tables. 7.4 Discussion 7.4.1 Model Validation The close c orrelations between IV data and PBPK model predictions in Figure demonstrate the validity of the model parameters. Experim ents for obtaining partition coefficients are often very difficult, expensiv e, and require sacrificing large numbers o animals to ensure accuracy. Figures 7-1 and 7-3 stand as di reliability of the equations developed rtunities for using PBPK modeling in the drug development process [73-76]. fits to non-IV data in Figure 7-4 confirm that using a single first-order absorption rate constant for non-IV dosing was an accept able approach for the cases considered here. 7.4.2 Evidence for Drug Redi stributio greatly altered A Figure 7-6 show Cvb incre me dosing in Cvbf and critical organs, and enhanced elimination resulting in reduced drug concentrations over the entire 24 hour period. The prolonged concentration reductions led to much lower AUC values, as Figure 7-5 suggests. Peak reductions were lower for tlag = 2 hours and nonexistent otherwise. Peak reductions ar 181

PAGE 182

not the pri mary concern for TCAs since most patients spend several hours rece treatment in the hospital, with the mean time spent reported as 8. 5 hours in one study [3]. Patients often face complications after prolonged drug exposure, making the AUC reductions and the reduced time spent at el evated concentrations most important. iving [4] where patients suffering mode at using since s, the is of r heart, lts n An interesting clinical study was done by Heard et al. rate but not severe TCA overdoses were treated with Fab fragments. Fab fragments are isolated, concent rated protein fragments that preferentially bind to a targeted drug, and are thus similar in princi ple to liposomes. This is an attractive strategy since TCAs are naturally highly pr otein bound. The study showed th redistribution methods such as the liposom e treatment examined herein is a safe practice, and that elevated serum TCA c oncentrations pose no obvious, additional dangers. The patients were treated with protein fragments three times after hospital admittance. Final serum concentrations in creased an average of six fold compared to the initial serum concentrations. The study mentioned that all patients fully recovered but the researchers were unable to fully attr ibute patient recoveries to the treatment overdoses were less than severe. Although drug binding to Fab fragments is different than binding to liposome PBPK modeling results for Cvb are validated qualitatively by the fact that drug concentrations increased as much or more in their study, supporting the hypothes drug redistribution through lipos omes. Heard et al. were unable to comment on whethe or not their protein fr agments had any direct effect on spec ific organs such as the brain, etc. Since Cvb values are remarkably similar fo r both studies, our model resu could be used to gauge the effectiveness of the Fab fragments on tissue concentratio 182

PAGE 183

reductions. Figure 7-6 suggests that the Cvb reductions observed by Heard et al. [4] ar strong evidence for significant TCA concen tration reductions in the heart, brain, and Cvbf. This is an excellent example of how e PBPK modeling and clinical studies can work in tan less n system through slowed actio nd depleted Ca2+ levels in the SR, which in turn reduces bindi dem to provide clarity and improved understanding. In contrast to AMI, BUP results show n in Figure 7-7 and Table 7-9 show drastic drug redistribution upon liposome dosing, even with a fo ur fold increase in the lipid loading. Obviously, treating TCA overdose with liposomes is more likely to be successful than treating an adverse reacti on to BUP, due to stronger TCA-liposome binding. However, with AUC and BUP peak reductions in key organs of roughly 15% and 17%, respectively, liposome administration, with single or multiple doses, could still reduce adverse effects and speed recovery times for patients with BUP overdose. 7.4.3 Pharmacodynamics Both AMI a nd BUP can bind to fast Na+ channels [9,16] and lethally impair cardiac function. Na+ channel binding alters the card iac conductio n potential propagation, often leading to elongated QRS intervals [2,17]. Hypotension resulting from reduced cardiac cont ractility is a major cause of death [2]. Compromised cardiac myocyte contractility arises from both Na+ channel blockage a other effects. AMI has been shown to directly interfere with the calcium-inducedcalcium-release mechanism of myocyte c ontraction [141]. The open probability of ryanodine receptor (RyR) channels connecting the myocyte plasma membrane to the sarcoplasmic reticulum (SR) increases in the presence of AMI in a dose dependent fashion [141]. This leads to ng bet ween actin and myosin filam ents. AMI also inhibits SR Ca2+ ATPase pumps (SERCA) from pumping Ca2+ back into the SR [141]. Ef fectively, the cardiac system 183

PAGE 184

works by constantly creati ng ion gradients that drive ener gy intensive processe AMI disturb s this cycle. BUP also impairs contractility, possibly through direct Ca2+ related effects or interference with mitochondrial activity, which starves ATPase pumps of ATP mol needed to create ion gradients [17]. Both AM I and BUP can disrupt the central nerv system (CNS) and cause seizures, delirium, and di sorientation [2,17]. This is more of primary concern for BUP. Many ex-vivo studies involving isolat ed hearts have attempted to observe dru induced cardiac toxicity. The PBPK m odels developed herein predict the drug concentration in human heart tissue as a func tion of time. Many factors determine the state of the heart at any given moment. Indi cators such as the QRS interval, beats per minute, and blood press s, and ecules ous a gure show changes resu lting from local drug-induced effects ment a have on the cardiac system. ing ioned above and the sympathet ic and parasympathetic nervous systems. As result, it is impractical to attempt to pr edict heart function on the macro-scale as a function of time. However, published studies di rectly linking localized or molecular level cardiac function with drug concentrations do offer opportunities to approximate the local effects that liposomes ma y Accordingly, in vitro published data re lating localized cardiac function to drug concentrations was utilized to project loca l cardiac function improvements follow liposome administration. Several invest igators have measured and reported various cardiac parameters as a functi on of drug concentration. Corre lation curves were fit to those data points over the concentration ranges encountered in the PBPK models. Drug concentration predictions as a function of time from the PBPK models were then 184

PAGE 185

inserted into the correl ation equations, result ing in time dependent predictions of lo cardiac changes during drug overdos e. Human cardiac cell (Hh1) Na+ channel function in the presence of AMI and BUP was studied by Nau et al. [9,16]. Changes in baseline contractility of human atrial tissue versus AM I concentrat cal ion were reported by Heard et al. [1 g t l as 0]. AMI effects on RyR channels were published by Zima et al. [141]. The dru concentrations measured in the in vitro expe riments in these three cases would mos accurately correspond to the unbound drug conc entration in the extracellular water in the heart, Cu,EW,ht, which is equivalent to the unbound drug concentration in the plasma in the heart, Cu,p,ht [106]. This concentration was calculated using Cht from our mode follows: )KKK(1H)(1Keff C CC321 ht ht (7-21) AMI effects on calcium movement by SERCA we re also published by Zima et al. [141]. Because the SERCA are located within the my ocyte plasma membranes on the surface of the SR, they are exposed to the unbound drug in the intracellular water in the heart Cu,IW,ht. Intracellular and extracellular water concentrations can be related via their pH values and drug pKa values by the following equation [106], htEW,u,htp,u, EW IWpHpKa101 101 CC. (7-22) EquationpHpKa htEW,u,htIW,u, 7-22 was utilized to predict Cu,IW,ht and thus predict SERCA function versus time during an AMI overdose. Figures 78a through 7-8d show that Na+ channels, baseline contractility, SERCA velocity, and RyR channels rapidly recover from an AMI overdose with liposomes. Figure 7-8e show s that Na+ channel current more closely 185

PAGE 186

resembles normal lev els for BUP reaction when liposomes are administered. The necessary data for predicting additi onal BUP effects was unavailable. Figure 7-8 provides evidence for cardiac to xicity reversal from liposome therapy after AMI or BUP overdose. The data suggest s that the effect on contractility is as important as the effect on conduction di e other hand, the mode 5] ner ected. gh stur bances. This is evident for AMI, where percent baseline contractility and Na+ channel recovery are both significant and rapid. The dual nature of the recovery is of special significance when comparing liposome therapy to the administration of sodium bicarbonate or lipid emulsions. Sodium bicarbonate has traditionally been used to treat TCA overdose as a means of counteracting conduction disturbances [2]. This does not treat reduced contractility from direct drug-cell interacti on, the main cause of hypotension. On th of action of lipid emulsi on therapy used to treat loca l anesthetic overdoses [12-1 has been hypothesized as counteracting the inhibi tion of fatty acid transport at the in mitochondrial membrane [44,205]. This me thod represents a direct effect on contractility, but in vitro BUP binding data [19] with lipid emulsions composed of mostly soy bean oil suggests that redistribution is unlikely to be accomplished. Thus, conduction disturbances caused by Na+ channel blockage would be largely unaff Based on this reasoning, a combination of liposomes designed to sequester hi amounts of TCAs and sodium bicar bonate would optimally treat conduction disturbances and contractility reductions duri ng TCA overdoses. In the case of BUP, the conduction disturbance improvements i nduced with liposomes would have to be more thoroughly compared with the contractility recovery caused by lipid emulsions. 186

PAGE 187

7.4.4 Sensitivity Analysis The sensitivity of the PBP K model to key parameters was probed by varyi parameters as stipulated in Tabl e 7-7. This analysis is val uable for identifying future pitfalls for treatment devel opment. Although increasing Eh resulted in slightly high AUC reductions due to faster clearance, neither AMI nor BUP simulations we especially sensitive to clinic ally relevant changes in Eh. Increased organ to blood partition coefficients (Keff), B:P, and fu had the biggest effect on liposome efficacy. T tissues were more able to compete with t he liposomes for drug binding, causing the ng the er re he ess effective at r educing the AUC for BUP. AUC reductions of slig s still r the treatment to be about 50% l htly less than 50% were still ach ieved for AMI due to the extremely high binding affinity between AMI and liposomes. 20% K2 reductions representing less effective liposome-drug binding caused AUC and peak r eductions to drop by a few percentage points in both cases. This was less notew orthy for AMI as the overall change wa very large. The change for BUP represent ed a much larger portion of the AUC and peak reductions and may be important. A reduction in ka increases tmax, i.e., the time at which the peak is reached. Accordingly, for a fixed tlag (=2 hours), a reduction in ka results in a reduction in the blood concentrati on at the time of liposome administration, leading to a larger reduction in the peak concentration. No effect of changing ka was seen for BUP since liposome dosing occurs shortly after drug dosing. More rapid liposome clearance altered AMI AUC reducti ons by about 9% compared to 30% for BUP, again illustrating the reduc ed room for error for BUP. Overall, the model results were not par ticularly sensitive to any one paramete and the liposomes would probably redistri bute a considerable am ount of AMI and a smaller but potentially signific ant amount of BUP in most pat ients. Kpu values have 187

PAGE 188

potential to modify the results the most. The reduced liposome affinity for BUP makes the parameter changes more significant for the anesthetic compared to AMI. A more thorough analysis of threshold dr ug concentration changes needed for effective tox reversal is required for drawing more concrete conclusions. 7.4.5 Liposome Dose Optimization icity s e e as liposome dose he liposome dose increases. Figure 7-5c demo I become arge f doses Dose optimization is a difficult challenge when developing new therapies. Beside ensuring a low lev el of toxicity, it is pr actically very difficult to predict what concentrations are optimal for therapeut ic effectiveness while also avoiding unnecessary patient risk and material waste. Even arriving at the correct doses for animal experiments can be time consumi ng and expensive. PBPK models can improv dose optimization by identifying effective dose ranges early [74]. In Figures 7-5a and 75b, heart, brain, and Cvbf AUC values for AMI rapidl y decreas increases, but this effect subsides as t nstrates that AMI peak reductions level of f prior to reaching 0. 72 g/L. The benefit to dose ratio for AMI decreases as major changes in the distribution of AM harder to achieve at higher Cvb concentrations. This is a direct result of the l proportion of AMI being extract ed from the tissues at liposom e doses of 0.72 g lipid/L Essentially, loading much more than 0.72 g lip id/L into the body to further increase Cvb and decrease Cht and Cbr may not be worth the potential si de effects of increased lipid dosing. Conversely, liposom e-BUP binding is much lower and a large proportion o BUP still remains inside tissues at 2.88 g lipid /L. As displayed in Table 7-9, lipid for BUP treatment should be as large as possi ble, limited solely by the maximum safe dose of phospholipids. 188

PAGE 189

7.4.6 Time Lapse Between Drug Ad ministration and Li posome Treatment AMI overdose treatment and BUP adverse re action treatment represent two very different scenarios in terms of the amount of time elapsing between drug and liposome dosing. This difference is quantitatively evident in the PBPK m odel results. The therapeutic effect for AMI patients dimini shes as the time between ingestion and treatment increases (Figure 7-5d). Regardle ss, the model points to large AUC reductions even with tlag = 8 hours. Therapeutic benefits even for large tlag are possible id conc here rage meters dded to the modeling result s. Equations used to estimate hepatic extra in which because of high affinity binding between AM I and the liposomes, resulting in rap entration reductions upon dosing. The absorpt ion rate constant is also a factor for AMI since rapid absorption coupled with large tlag values increases the patients exposure to the drug. These effects are al together avoided in the case of BUP, w patients are already at a medical facilit y and available for immediate treatment. 7.4.7 Limitations The central goal of the PBPK mo del developed here is to serve as a tool for evaluating the potential of liposome based therapy for overdose treatment in an ave patient, and to aid the design of animal and human studies. While the model predictions are encouraging, seve ral issues related to the assumptions and para utilized in the model must be considered. Uncertainties in Kpu calculations, drug and liposome specific input parameters, and the data used for validation and fitting a inherent variability ction and non-IV absorptio n ignored possible non-linear ities at elevated drug concentrations. Drug metabolites, which c ould independently contribute to toxicity, were not accounted for. Finally, differences between the in vitro environment liposome-drug binding and protein-drug binding were measur ed and the physiological 189

PAGE 190

environment (i.e. shear forces on liposomes, presence of enzymes and other trace molecular entities, immune system response) could alter the ultimate effect of the therapy. These issues represent opportunities for further investigation in the futu 7.5 Conclusions A physiologically bas ed pharmacokinetic (PBPK) model consisting of 15 compartments was developed by employin g published [106] equations for tissue to blood partition coefficients and par ameters pooled from numerous clinical studies. IV data affirmed the ability of the models to predict drug concentrations in the blood compartments with reasonable accuracy. Least squares fits of model results to no data were utilized to obtain firstorder absorption rate constants. re. n-IV Drug overdose simulations for AMI revealed that liposomes are capable of reduc s by ses ound ome therapy, espe /L as y ing brain and heart AUC values by over 60% and peak AMI concentration 20% if treatments are provided within 2 hours of ingestion. AMI concentration increa in venous blood for the PBPK simulations and clinical overdose treatments with protein fragments were similar [4]. BUP AUC and peak reductions were much lower at ar 15-17%. First approximations of localized cardiac pharmacodynamics suggested improved ion channel function and myocyt e contractility with lipos cially for AMI. Although the models su ggested between 0.72 and 1.44 g lipid an optimal starting dose for treating AMI overdoses, the maximum safe dose of liposomes should be given for adverse reacti ons to BUP. The modeling results were relatively insensitive to reasonable variat ions in model parameters. The modeling predictions agree, at least qualitatively, with reports on overdose treatments by Fab fragments, which like liposom e administration, is an approach based on high affinit binding between fragm ents and the drug. 190

PAGE 191

In conclusion, the PBPK models develo ped in this chapter suggests that lipos omes are capable of redistributing bot h AMI and BUP into the blood compar to significant degrees and may be effective at treating AMI overdoses and adverse reactions to BUP. Liposomes will, in general be more effective at treating overdos drugs with high volumes of distribution and slow elimination characteristics, such as TCAs. tments es of 191

PAGE 192

Figure 7-1. Organ to blood partition coefficien ts (Keff) calculated using Equation 7-4 () or determined experimentally () for a) AMI [161-163] or b) BUP [164167]. 192

PAGE 193

Figure 7-2. PBPK model structure used fo r AMI and BUP overdose simulations. 193

PAGE 194

Figure 7-3. Drug concentrations versus ti me predicted using PBPK models () or measured () after IV dosage for a) IMI [ 173] in venous plasma, b-c) IMI [174] in venous plasma, d) AMI [175] in venous plasma, e-h) AMI [176] in venous plasma, i-j) AMI [149] in venous whole bl ood, k) BUP [179] in arterial plasma, l) BUP [151] in arterial plasma, m) BU P [178] in venous wh ole blood, n) BUP [177] in venous plasma, and o) BUP [186] in arterial plasma. Error bars are shown for experimental results when reported. 194

PAGE 195

Figure 7-4. Drug concentrations versus time for PBPK models () where ka was allowed to vary to obtain best fits to measured non-IV dosage data () for ab) IMI [173], c) IMI [109], d) AMI [191], e) AMI [190 ], f-h) AMI [188], i) AMI [189], j-k) AMI [149], l) BU P [197] into the epidural space, m) BUP [195] into the intercostal space, n-o) BUP [196] into the leg, p) BUP [198] into the intercostal space, q) BUP [193] into the intraperitoneal space, r) BUP [192] into the epidural space, and s) BUP [194] into the epidural space. TCA dosages (a-k) were all oral. Error bars are shown for experimental results when reported. 195

PAGE 196

Figure 7-5. Summary of AMI overdose (2500 mg) simulation results. AMI AUC values versus liposome dose for a) venous blood and key organs with tlag = 2 hours; b) free AMI in venous blood (Cvbf) with tlag = 2 hours; c) peak AMI concentrations versus liposome dose with tlag = 2 hours; d) AMI AUC values versus tlag, the time between drug ingesti on and liposome treatment, with Clip = 1.44 g/L. 196

PAGE 197

Figure 7-6. AMI concentrations versus ti me for simulated overdose cases (2500 mg) without liposomes and with liposomes at 1.44 g/L for tlag = 2, 4, 6, and 8 hours for a) venous whole blood, b) free AMI in venous blood, c) heart, d) brain, and e) muscle. 197

PAGE 198

Figure 7-7. BUP concentrations versus time for simulated adverse reactions (100 mg) without liposomes and with liposomes at 1.44 or 2.88 g/L for tlag = 0.25 hours for a) venous whole blood, b) free BUP in venous blood, c) heart, d) brain, and e) muscle. 198

PAGE 199

Figure 7-8. Pharmacodynamics in the heart as a function of time. Correlations from drug concentrations in the heart pr edicted by our PBPK models to local pharmacodynamic effects were estimated through the use of published data. Controls were simulated overdoses with no liposomes; lipid loadings are specified for simulated overdoses wit h liposomes. The plots show a) normalized current through cardiac (hH1) Na+ channels [9] for AMI overdose, b) % baseline contractility of atrial tiss ue [10] for AMI over dose, c) the open probability of type 2 ryanodine receptors (RyR) [141] for AMI overdose, d) maximum Ca2+ uptake by sarcoplasmic reticulum (SR) Ca2+ ATPase pumps (SERCA) [141] for AMI overdose, e) normalized current through cardiac (hH1) Na+ channels [16] for BUP adverse reaction. Tlag was 2 hours for AMI and 0.25 hours for BUP; AMI and BUP doses were 2500 mg and 100 mg, respectively. 199

PAGE 200

Table 7-1. Drug specific par ameters for Kpu calculations Imipramine Amitriptyline Bupivacaine Value Refs. Value Refs. Value Refs. pKa 9.5 [106] 9.4 [146] 8.1 [123] B:P 1.1 [147,148] 0.86 a [149,150] 0.64 a [151,152] fu 0.18 b Ch. 4 0.08 c Ch. 3 0.10 d Ch. 5 LogPvo:w 4 [106] 4.1 e [153] 2.4 e [153] LogPoc:w 4.8 [106] 4.92 [154] 3.4 [155] KpuBC 6.8 f [106] 8.6 f [106] 2.0 f [106] KaBC 6.13 g [106] 7.45 g [106] 0.10 g [106] Key: pKa acid dissociation constant; B:P blood to plasma concentration ratio; fu fraction of drug unbound in plasma; LogPvo:w log of the vegetable oil to water partition coefficient; LogPoc:w log of the octanol to wa ter partition coefficient; KpuBC affinity for blood cells [106]; KaBC association constant of drugs to blood cells [106]. aWeighted average from two studies. bOur value originated from experimental data at physiological IMI concentrations (Chapter 4) and 0.1 [108,156], 0. 12 [157], and 0.11 [107] have been reported in literature. An fu of 0.24 was used for Kpu calculations by Rodgers et al. [106]. cOur value originated from experim ental data at physiological AMI concentrations (Chapter 3) and 0.08 [ 157], 0.05 [108], and 0.06 [156] have been reported in literature. dOur value originated from exper imental data at physiological BUP concentrations (Chapter 5) and 0.10 [121], 0.06 [119 ], and 0.05 [158] have been reported in literature. eLogPvo:w values were calculated from published methods [153] from reported LogPoc:w values. fKpuBC values were calculated [106] using fu and B:P values with H = 0.45. gKaBC values were calculated using drug specific parameters reported here and publis hed equations [106]. 200

PAGE 201

Table 7-2. Calculated drug to tissue partition coefficient (Kpu) values Organ IMI AMI AMI a BUP BUP b Adipose (ad) 73.5 114.0 131.5 38.2 46.1 Bone (bo) 19.9 27.8 57.1 8.3 21.7 Brain (br) 27.4 41.6 59.1 18.1 26.1 Gut (g) 59.0 80.5 185.8 19.3 67.3 Heart (ht) 44.4 57.4 155.8 9.1 53.9 Kidney (k) 87.9 110.6 330.4 10.5 110.7 Liver (h) 81.8 103.6 302.9 11.3 102.1 Lung (l) 74.0 95.2 266.1 13.0 90.8 Muscle (m) 31.1 40.1 106.9 7.0 37.4 Pancreas (p) 49.1 68.8 141.8 20.3 53.5 Skin (sk) 51.8 76.1 133.8 26.9 53.2 Spleen (sp) 55.7 69.7 208.7 6.8 70.2 Thymus (th) 46.7 60.6 161.1 10.3 56.1 aThese Kpu values were used for the AMI sens itivity analysis. To estimate the upper Kpu bounds for AMI, an fu of 0.05 [85,159] and a B:P rati o of 1.04 [150], the weighted B:P average from 1 of 2 studies used for the overall AMI B:P av erage, were used. bThese Kpu values were used for the BUP sens itivity analysis. To estimate the upper Kpu bounds for BUP, an fu of 0.05 [158] and a B:P ratio of 0.73 [151], the weighted B:P average from 1 of 2 studies used for the overall BUP B:P average, were used. Table 7-3. Partition coefficients for drug binding to proteins, liposomes, and red blood cells K1 a K2 b K3 c IMI 4.8 1.1 5.803 AMI 9.8 1.3 64.15Clip 6.106 BUP 9.0 0.75 1.39Clip 1.64 Key: K1 Concentration ratio for drug bound to proteins to free drug; K2 concentration ratio for drug bound to liposomes to free drug; K3 concentration ratio for drug bound to RBCs to free drug; Cf free drug concentration in ng/mL; Clip liposome concentration in venous blood in g/L. aBased on measured fractions bound to serum proteins of 0.83 (IMI Chapter 4) and 0.91 (AMI Chapters 2,3) at drug conc entrations from 1 to 10 M (mean standard deviation) and BUP protei n binding at low BUP concentrations extrapolated from our data (Chapter 5). bIMI overdose simulations were not done. AMI and BUP equations are based on measured, concentration dependent drug binding to liposomes (Chapters 2,3,5). T he equations are valid from Clip = 0 to 1.44 mg/mL for AMI and 0 to 2.88 mg/mL for BUP. cK3 values for IMI, AMI, and BUP were based on B:P ratios of 1.1, 0.86, and 0.64, respectively, and solved for as shown in the text. 201

PAGE 202

Table 7-4. Organ blood flow and volume fractions for humans Organ Fraction of Cardiac Output a Fraction of Body Volume b Adipose (ad) 0.0342 0.1196 Bone (bo) 0.1665 0.0856 Brain (br) 0.1001 0.0200 Gut (g) 0.1368 0.0171 Heart (ht) 0.0316 0.0047 Kidney (k) 0.1632 0.0044 Liver (h) 0.2080 0.0260 Lung (l) 1.0000 0.0076 Muscle (m) 0.0395 0.4000 Pancreas (p) 0.0313 0.0014 Skin (sk) 0.2567 0.0371 Spleen (sp) 0.0351 0.0026 Thymus (th) 0.0014 0.0004 aTotal cardiac output in L/min was calculat ed from the allometric (CO = 0.235Weight0.71) with weight in kg [170]. Blood flow was scal ed to match the report ed weights of patients in studies when reported (70 kg otherwise). Fractions of ca rdiac output were calculated from values reported by Igari et al. [78] for a 70 kg man for all organs except the thymus and bone. The proportion of liver blood supply coming from the pancreas, spleen, gut, and arterial blood were calculated from data reported by Benowitz et al. [171] for rhesus monkeys. Thymus and bone bl ood supplies were scaled by organ weight from blood supplies to rat thymus and bone reported by M eno-Tetang et al. [79]. The fractions of cardiac output were scaled so they added to exactly 1 after all were combined. bFractions of total body volume were taken from Poulin et al. [144] except for the pancreas and thymus. Pancreas and thymus volumes were estimated from weights of 100 g and 25 g, respectively [172]. Total body volume was estimated from body weight at a density of 1 L/kg. Body volumes were scaled to match reported weights or average weights of patients st udied when reported (70 kg otherwis e). Venous and arterial blood volumes were taken as 3.6 and 1.8 L [78], respectively, for a 70 kg subject and scaled linearly with weight. 202

PAGE 203

Table 7-5. Input par ameters for PBPK model Imipramine Amitriptyline Bupivacaine Value Refs. Value Refs. Value Refs. Cl h 709.4 a [173,174] 579.6 a [149,150,175,176] 498.4 b [177-180] Cl r 12.3 c [173,174] 10.0 c [149,150,175,176] 10.7 d [177-180] H 0.45 [181] 0.45 [ 181] 0.45 [181] F 0.42 e [148,173,174] 0.45 e [149,150,182,183] 1 f Key: Clh hepatic clearance in mL/(h*kg); Clr renal clearance in mL/(h*kg); H hematocrit; F bioavailability. aWeighted average from cited st udies used to calculate Eh as shown in the text [73,108,169] assumi ng 98.3% of elimination is hepatic for TCA's [184]. bWeighted average from cited studies used to calculate Eh as shown in the text [73,108,169] assuming 97.9% of elimination is hepat ic for BUP [178,185]. cWeighted average from cited studies assuming 1.7% of elimination is renal for TCA's [184]. dWeighted average from cited st udies assuming 2.1% of elimination is renal for BUP [178,185]. eWeighted average from cited studies. fBioavailability was assumed to be 1 for BUP since non-oral routes of administration were used. Table 7-6. Drug specific inputs for overdose simulations Amitriptyline Bupivacaine Value Refs. Value Refs. Ddrug 100, 500, 2500 a [7,110,111,115] 100 b [192-196] D lip 0.36, 0.72, 1.44 c Ch. 2,3 1.44, 2.88 c Ch. 5 t lag 2, 4, 6, 8 d [7,110-112,202] 0.25 e [12-15] Key: Ddrug drug dose(s) simulated with PBPK models in mg; Dlip liposome doses simulated with PBPK models in g lipid/L blood; tlag elapsed time between drug administration or ingestion and lip osome treatment in hours. aDoses of 300 mg [7], 1-2 g [110], and 3.7-9.2 g [111] were clinically observed in AM I overdose cases and 1.4-2.8 g has been suggested as common for AMI overdoses [115]. bBUP Doses of 150 mg [192,196], 100 mg [193,194], and 140 mg [195] have been clinically administered. cCorresponds to 27.8-111 mg/kg or 1.9-7.8 g of total phospholipids for AMI and 111-222 mg/kg or 7.8-15.6 g of total phospholipids fo r BUP. Liposome doses were modeled as bolus doses based on maximum possible lipid concentrations of about 50 mg/mL in liposome formulations [199] and fluid volumes given as bolus doses or rapid infusions in clinical studies [12,14,15,40]. Dhanikula et al. [25] sug gest that at least 10 g of liposomes could be administered safely; up to 250 g of lipids (phospholipids and oil) have been suggested as safe for anesthetic overdose treatment (http://lipidrescue.squarespace.com/), and 100 g has been used in a clinical overdose case [200]. Up to 190 mg/kg of liposomes have been given to mice in animal studies [201]. dTime lags between overdose and treatm ent have been observed clinically from 2-8 hours for TCA's. eTime between drug administratio n and lipid emulsion treatment have been observed clinically from 0.1-0.25 hours for anesthetics. 203

PAGE 204

Table 7-7. Sensitiv ity analysis setup Amitriptyline Bupivacaine Simulation Sensitivity Simulation Sensitivity Eh 0.41 0.33, 0.45 a 0.37 0.33, 0.45 a k a 0.093 0.054, 0.132 b 0.179 0.100, 0.258 c K2 63.0 84.0 d 50.4 67.2 d 1.64 4.19 d 1.31 3.35 d Cl lip 0.24t2 7.89t + 100 e -9.1ln(t +0.01) + 59.4 0.24t2 7.89t + 100 e -9.1ln(t + 0.01) + 59.4 Key: Eh hepatic extraction factor (Equation 7-8); ka first-order absorption rate constant in h-1, K2 concentration ratio for drug bound to liposomes to free drug; Cllip equation defining the percent of the initial liposome dose rema ining in the blood stream as a function of time, used to simulate liposome clearance. aUpper [149,178] and lower [176,179] bounds were taken from maximum and minimum average values observed in single studies that were part of the studies used to obtain hepatic drug clearance (see Table 7-5). bUpper and lower bounds were calculat ed using the standard deviations of the averaged ka value which was averaged from best fi ts to data from oral AMI studies [149,188-191]. cUpper and lower bounds were calculated using the standard deviations of the averaged ka value which was averaged from bes t fits to data from non-IV BUP studies [192,194,197]. dRange of liposome concentration dependent K2 values used to calculate the equation for K2 as a function of Clip. The values were reduced by 20%, although this was larger than the observed standard dev iations, to investigate significantly reduced binding. The equations became (51.32Clip) for AMI and (1.11Clip) for BUP. See Table 7-3 for comparison. eA new liposome elimination curve as a function of time was generated by reducing t he percent of liposomes left at each time point by 30%. This was much larger than th e error bars shown for the data [30] but was done to investigate large variation in elimination. Extended liposome circulation times were not investigated as they only increase the efficacy of the treatmen t. Note that 0.01 was added to t to avoid values greater than 100% when logarithms were used. 204

PAGE 205

Table 7-8. AMI overdose simulation results AMI Dlip tlag AUCvb Peakvb AUCht Peakht AUCbr PeakbrAUCvbf Peakvbf 2500 0.00 0 28985 1652 177560 10175 128650 7372 1711 97.5 2500 0.36 2 32325 2665 118390 8131 85792 5900 1146 82.3 2500 0.72 2 34344 3721 91187 8131 66086 5900 887 82.3 2500 1.44 2 36483 5445 63874 8131 46302 5900 628 82.2 2500 0.36 4 32389 3229 124790 10039 90430 7275 1210 96.9 2500 0.72 4 34427 4574 100460 10040 72814 7275 980 97.0 2500 1.44 4 36546 6782 75998 10038 55093 7274 747 97.0 2500 0.36 6 32448 3239 132360 10175 95920 7372 1283 97.5 2500 0.72 6 34519 4616 111300 10175 80668 7372 1084 97.5 2500 1.44 6 36649 6886 89930 10175 65187 7372 881 97.5 2500 0.36 8 32486 3071 139860 10175 101350 7372 1354 97.5 2500 0.72 8 34594 4393 122060 10175 88458 7372 1186 97.5 2500 1.44 8 36761 6584 103750 10175 75194 7372 1012 97.5 500a 0.00 0 5797 330.3 35512 2035 25731 1474 342 19.5 500 1.44 2 7297 1089 12775 1626 9260 1180 125.5 16.4 100 a 0.00 0 1159 66.1 7102 407 5146 294.9 68.4 3.9 100 1.44 2 1459 217.8 2555 325.3 1852 236 25.1 3.3 Key: Dlip liposome doses simulated with PBPK models in g lipid/L blood; tlag elapsed time between drug administration or ingesti on and liposome treatment in hours; AUC area under the AMI concentration versus time curve in (ng*h)/mL; Peak maximum AMI concentration reached in the specified organ in ng/mL; vb v enous blood; ht heart; br brain; vbf free AMI in venous blood. aDrug doses of 500 and 100 mg are only shown for the most effective liposome dose of 1.44 mg/mL. 205

PAGE 206

Table 7-8 Continued. AMI overdose simulation results AMI Dlip tlag % AUCvbf % Peakvbf% AUCht % Peakht % AUCbr % Peakbr 2500 0.00 0 2500 0.36 2 33.0 a 15.6 33.3 20.1 33.3 20.0 2500 0.72 2 48.1 15.6 48.6 20.1 48.6 20.0 2500 1.44 2 63.3 15.6 64.0 20.1 64.0 20.0 2500 0.36 4 29.3 0.6 29.7 1.3 29.7 1.3 2500 0.72 4 42.8 0.5 43.4 1.3 43.4 1.3 2500 1.44 4 56.3 0.5 57.2 1.3 57.2 1.3 2500 0.36 6 25.0 0.0 25.5 0.0 25.4 0.0 2500 0.72 6 36.7 0.0 37.3 0.0 37.3 0.0 2500 1.44 6 48.5 0.0 49.4 0.0 49.3 0.0 2500 0.36 8 20.9 0.0 21.2 0.0 21.2 0.0 2500 0.72 8 30.7 0.0 31.3 0.0 31.2 0.0 2500 1.44 8 40.8 0.0 41.6 0.0 41.6 0.0 500 0.00 0 500 1.44 2 63.3 a 15.7 64.0 20.1 64.0 20.0 100 0.00 0 100 1.44 2 63.3 a 15.7 64.0 20.1 64.0 19.9 Key: Dlip liposome doses simulated with PBPK models in g lipid/L blood; tlag elapsed time between drug administration or ingesti on and liposome treatment in hours; AUC area under the AMI concentration versus time curve in (ng*h)/mL; Peak maximum AMI concentration reached in the specified organ in ng/mL; % AUC percent reduction in AUC compared to case without liposomes; % Peak percent reduction in drug concentration peak compared to case without liposomes; vb venous blood; ht heart; br brain; vbf free AMI in venous blood. aBlanks are cases with no liposomes and % AUC and peak changes with liposomes are calculated based on the no liposome cases at equivalent AMI doses. 206

PAGE 207

Table 7-9. BUP overdose simulation results Dlip a AUCvb Peakvb AUCht PeakhtAUCbrPeakbrAUCvbfPeakvbf 0 2399 323 3391 458 6695 911 206.1 27.7 1.44 2465 334 3105 414 6134 824 188.6 25.1 2.88 2526 344 2868 379 5669 754 174.2 22.9 Table 7-9 Continued. BUP overdose simulation results Dlip a % AUCvbf % Peakvbf % AUCht % Peakht % AUCbr % Peakbr 0 1.44 8.5 b 9.6 8.5 9.6 8.4 9.6 2.88 15.5 17.3 15.4 17.3 15.3 17.3 Key: Dlip liposome doses simulated with PBPK m odels in g lipid/L blood; AUC area under the BUP concentration versus time cu rve in (ng*h)/mL; Peak maximum BUP concentration reached in the specified organ in ng/mL; % AUC percent reduction in AUC compared to case without liposomes; % Peak percent reduction in drug concentration peak compared to case without liposomes; vb venous blood; ht heart; br brain; vbf free BUP in venous blood. aThe BUP dose was 100 mg and tlag = 0.25 hours. bBlanks are cases with no liposomes and % AUC and peak changes with liposomes are calculated based on the no liposome cases at equivalent BUP doses. 207

PAGE 208

Table 7-10. AMI sensitivity analysis results Change in Model AMI Dlip tlag % AUCvbf % Peakvbf% AUCht % Peakht % AUCbr % Peakbr None 2500 1.44 2 63.3 a 15.6 64.0 20.1 64.0 20.0 Eh (0.45) b 2500 1.44 2 63.8 13.9 64.6 18.2 64.5 18.1 Eh (0.33) 2500 1.44 2 61.9 19.6 62.5 24.2 62.5 24.1 Kpu c 2500 1.44 2 46.6 16.5 47.1 21.9 46.9 19.2 K2 d 2500 1.44 2 58.6 15.6 59.3 20.1 59.2 20.0 ka (0.132) e 2500 1.44 2 63.6 10.6 64.5 14.7 64.5 14.6 ka (0.054) 2500 1.44 2 62.7 23.4 63.2 27.8 63.1 27.7 Cl lip f 2500 1.44 2 57.7 15.6 58.4 20.1 58.4 20.0 Key: Eh hepatic extraction factor (Equation 7-8); Kpu organ to plasma partition coefficient with respect to the free drug concentration in plasma; K2 concentration ratio for drug bound to liposomes to free drug; ka first-order absorption rate constant in h-1; Cllip equation defining the percent of the init ial liposome dose remaining in the blood stream as a function of time, used to simu late liposome clearance; AMI AMI dose in mg; Dlip liposome doses simulated with PBPK models in g lipid/L blood; tlag elapsed time between drug administration or ingesti on and liposome treatment in hours; % AUC percent reduction in AUC compared to ca se without liposomes; % Peak percent reduction in drug concentration peak compared to case without liposomes; ht heart; br brain; vbf free AMI in venous blood. aAll % changes were calculated from the case of no liposomes with the same sensitivity analysis parameter altered as specified for that particular row. bThe value of Eh used for the overdose simulations was 0.41. cAn fu value of 0.05 and a B:P ratio of 1.04 was used to estimate the increased Kpu values; see Table 7-2 for values and details. dValues were decreased by 20% at each liposome concentration; see Table 7-7 for details. eThe absorption constant used for the AMI overdose simulations was 0.093. fLiposome clearance was incr eased by 30% at each time point to generate a new elimination function; see Table 7-7 for details. 208

PAGE 209

209 Table 7-11. BUP sensitivity analysis results Change in Model BUP Dlip tlag % AUCvbf % Peakvbf% AUCht % Peakht % AUCbr % Peakbr None 100 2.88 0.25 15.5 a 17.3 15.4 17.3 15.3 17.3 Eh (0.45) b 100 2.88 0.25 16.2 18. 2 16.2 18.1 16.1 18.1 Eh (0.33) 100 2.88 0.25 15. 0 16.8 15.0 16.8 14.9 16.7 Kpu c 100 2.88 0.25 8.2 8. 9 8.0 8.9 8.1 8.9 K2 d 100 2.88 0.25 12.8 14. 4 12.8 14.4 12.7 14.4 ka (0.258) e 100 2.88 0.25 15.8 17. 4 15.8 17.4 15.7 17.3 ka (0.100) 100 2.88 0.25 15. 1 15.9 15.0 15.9 14.9 15.9 Cl lip f 100 2.88 0.25 10.9 11. 7 10.9 11.7 10.8 11.7 Key: Eh hepatic extraction factor (Equation 7-8); K2 concentration ratio for drug bound to liposomes to free drug; ka first-order absorption rate constant; Cllip equation defining the percent of the initial liposome dose remaining in the blood stream as a function of time, used to simulate lipos ome clearance; BUP BUP dose in mg; Dlip liposome doses simulated with PBPK models in g lipid/L blood; tlag elapsed time between drug administration or ingestion and liposome treatment in hours; % AUC percent reduction in AUC compared to ca se without liposomes; % Peak percent reduction in drug concentration peak compared to case without liposomes; ht heart; br brain; vbf free BUP in venous blood. aAll % changes were calculated from the case of no liposomes with the same sensitivity an alysis parameter altered as specified for that particular row. bThe value of Eh used for the overdose simulations was 0.37. cAn fu value of 0.05 and a B:P ratio of 0.73 was used to estimate the increased Kpu values; see Table 7-2 for values and details. dValues were decreased by 20% at each liposome concentration; see Table 7-7 for details. eThe absorption constant used for the BUP overdose simulations was 0.1794. fLiposome clearance was increased by 30% at each time point to generate a new elimination function; see Table 7-7 for details.

PAGE 210

CHA PTER 8 CONCLUSIONS Treatments capable of reversing toxicity re sulting from prescription drug overdose or adverse reaction are needed. While several therapies are in development elsewhere, including pr otein fragments, lipid emulsions, polymeric particles, and microemulsions, no widely effective treatment has been introduced into the market. This work was aimed at developing a drug overdose treatment using liposomes. The primary drugs targeted for tr eatment included TCAs, such as AMI, IMI, and DOS, as well as the local anesthetic BUP. In Chapter 2, we demonstrated that DOPG liposomes and 50:50 DMPC:DOPG liposomes sequestered at least 99% of AMI at low drug concentrations. In the presence of albumin, pure DOPG liposomes were shown to be greatly inhibited by the proteins. In the presence of 7% proteins (w/w) composed of 4% albumin, 2% fibrinogen, and 1% globulins, the 50:50 DMPC:DOPG liposomes at a lipid loading of 0.72 mg lipid / mL sequestered 95-96% of the fr ee drug, as opposed to a predi cted 99%. The free AMI concentration was still reduced by 50-60%, however. In human serum, the 50:50 DMPC:DOPG liposomes took up about 94-98% of the drug and reduced the free drug concentration by 35-70%. Liposome-protein inte ractions were likely responsible for the reduced effectiveness of 50:50 DMPC:DOPG liposomes at sequestering AMI in serum versus PBS. We also proved that AMI binding to liposomes in the presence of proteins is quick and reversible. Uptake studies conducted with nortriptyline suggested that systems developed for AMI overdose treatment may also be useful for reducing the free concentration of nortriptyline. 210

PAGE 211

Chapter 3 revealed that AMI uptake by 50:50 DMPC:DOPG liposomes was not enhanced by increased lipid loading in the presence of serum protei ns. Drug binding was also shown to be independent of liposome size. Predominantly anionic lipids with a small amount of PEG-modified lipids were shown to sequester AMI from human serum most effectively. The optimal amount of PEG-modified lipid to be incorporated into liposomes to effectively shield proteins whil e also allowing diffusion and binding of the drug to the charged lipid memb rane appeared to be 5%. Increasing the lipid loading of the polymer shielded liposomes by a fact or of 2 resulted in a reduced free AMI concentration in human serum by nearly 90% across a wide range of initial drug concentrations. Liposomes were effective at drug binding after storage for about one month. The primary goal of Chapter 4 was to extend the use of pegylated, anionic liposomes from AMI binding to other TCA s and opipramol. In PBS, the liposomes proved capable of sequestering about 98% or more of IMI and DOS and 92.5-95.5% of opipramol. Additionally, the liposomes r educed the free drug concentrations in human plasma by 88-93% in the case of the weak bases IMI and DOS, and 76% for the diprotic drug opipramol (1.44 mg lipid/mL loading). The data supports liposome therapy as effective for a variety of TCAs. The length of the PEG chain did not make any detectable difference in uptake. Calc ulations suggested that liposomes are approximately 20 times more effective at binding antidepressants than acidic phospholipids already present in the body. Es timates of the times required for tissues to reach equilibrium also demonstrated that liposomes could significantly alter the drug concentrations in vital organs after a drug overdose. 211

PAGE 212

While Chapters 2-4 focused on optimizing pegylated, anion ic liposomes for TCA binding, BUP binding was the subject of Chapt er 5. Local anesthetics such as BUP are extremely useful clinically, but some patients suffer unexpected adverse reactions. Pegylated, anionic liposomes were shown to be highly effective at sequestering BUP from in vitro solutions, with a removal efficiency of about 80% in buffer and a free-drug reduction of about 50% in hum an serum. Comparisons to published data for lipid emulsions and microemulsions suggested that liposomes are better binding agents than either of the two. The re sults of Chapter 5 are especially encouraging, since a drug overdose therapy capable of successfully treat ing patients for a variety of drug classes would be more clinically acceptable and easier to administer. Chapter 6 was intended to explore the detailed interactions between drug molecules and liposomes. Binding of TCAs such as AMI, IMI, a nd DOS was compared to binding of the local ami de anesthetic BUP. The expe riments and modeling indicated that the amount of drug mole cules in the electrical double layer is negligible but the electrostatic effects play a major role in binding. The electrostatic interactions are responsible for the initial associati on between antidepressants and liposomes, whereupon the drug enters the bilayer with its charged region closely associated with the charged lipid head groups and its lipophilic region closely associated with the lipid bilayers. BUP, which is predominantly in the protonated state at pH 7.4 and 25C as well (86%), is also preferentially attracted to the charged vesicle [123]. Once bound, it appears to be located within the bilayer to a greater extent t han the antidepressants with minimal interactions with the charged surface. The stru cture of BUP makes it less able to access the bulk aqueous phase surrounding or encompassed by the lipid 212

PAGE 213

bilay ers. The conclusions regarding the differences in conformations between antidepressants and the anesthetic bupivacai ne were supported by the fact that bupivacaine binding leads to a much lower increase in liposome permeability compared to antidepressant binding. Assuming all other variables were held constant, multilamellar liposomes bound more drug molecules than unilamellar liposomes The difference vanished at high ionic strengths. We attributed t he increased binding for multil amellar liposomes without PEG to enhanced electrostatic interactions bet ween adjacent charged layers. For pegylated liposomes, the reduced dielectric constant in the aqueous-PEG layer between adjacent bilayers results in a lower energy barrier fo r cationic drug transport across the aliphatic tail region [130,132-134]. Over all, Chapter 6 afforded a better understanding of the molecular level interactions responsible for the high affinity binding between the liposomes and the drugs. While in vitro experiments allow fo r deeper understanding and the control of extraneous variables, estimates of in vivo efficacy are also extremely important during the early development phase. In Chapter 7, PBPK models were developed for this purpose. IV data affirmed the ab ility of the models to predict drug concentrations in the blood compartments with r easonable accuracy. Least squares fits of model results to non-IV data were utilized to obtain first-order absorption rate constants. Drug overdose simulations for AMI revealed that liposomes are capable of reducing brain and heart AUC values by over 60% and peak AMI concentrations by 20% if treatments are provided within 2 hours of ingestion. AMI concentration increases in venous blood for the PBPK simulations and clinical overdose treatments with protein 213

PAGE 214

fragments were similar [4]. BUP AUC and peak reductions were much lower at around 15-17%. First approximations of localized cardiac pharmacodynamics suggested improved ion channel function and myocyt e contractility with liposome therapy, especially for AMI. The PBPK models developed suggested that liposomes are capable of redistributing both AMI and BUP into the blood compartments to significant degrees and may be effective at treati ng AMI overdoses and adverse reactions to BUP. The models also suggested that liposomes will, in general, be more effective at treating overdoses of drugs with high volumes of distribution and slow elimination characteristics, such as TCAs. The work comprised herein shows that pegylated, anionic liposomes composed of DOPG and PEG-modified phospholipids exhibit high affinity binding to TCAs and BUP. The time scales for binding, feasible liposome doses, and typical drug doses all point to a very strong likelihood of treatment efficacy in critical care situations. However, several additional concerns should be addressed in future work. Liposome-drug binding should be measured in whole blood to ensure that RBCs and other w hole blood components do not negatively impact liposome-drug affinity. A detailed understanding of the fate of liposome-drug complexes in the liver and spleen should also be explored. This could be done with in vitro and in vivo experiments, along with a detailed hepatic model including the distinctive natur e of liver tissue, genetic fa ctors, and disease states. Results from the hepatic model would hopefully allow for a better estimate of the clearance characteristics of the exact liposom es used. Of course, in vivo experiments examining liposome clearance, liposome toxicity, and treatment effectiveness in animal models would also be tremendously informative. 214

PAGE 215

215 Related but broader work in this area could also be done. Molecular dynamics simulations could be performed to study the drug-bilayer interactions in even greater detail. Liposome-drug binding could be measured for non-prescription drugs or molecules, such as drugs of abuse or toxic compounds encountered in military applications. Liposomes could be loaded with ant idotes or specific neu tralizers, so that a combination of specific and non-specific treatment could be achieved. Finally, liposome surfaces could be designed to di rect the liposome-dr ug complexes to one tissue over another to achieve the slowest and safest elimination route possible. Each of the above mentioned resear ch topics represents an exce llent opportunity to utilize liposomes in the context of drug overdose tr eatment to improve treatment options or uncover stirring scientific findings.

PAGE 216

LIST OF REFERENCES [1] Wysowski DK. Surveillance of Prescr iption Drug-Related Mortality Using Death Certificate Data. Drug Safety 2007;30(6):533-40. [2] Liebelt EL, Francis PD. Cyclic Anti depressants. In: Goldfrank LR, Flomenbaum NE, Lewin NA, Howland MA, Hoffman RS Nelson LS, editors. Goldfranks Toxicologic Emergencies, 7th ed. New York: McGraw -Hill, 2002. p. 847-64. [3] Unverir P, Atilla R, Karcioglu O, T opacoglu H, Demiral Y, Tuncok Y. A retrospective analysis of antidepressant poisonings in the emergency department: 11-year experience. Hu m Exp Toxicol 2006;25:605-12. [4] Heard K, Dart RC, Bogdan G, OMalle y GF, Burkhart KK, Donovan JW, Ward SB. A Preliminary Study of Tricyclic An tidepressant (TCA) Ovine FAB for TCA Toxicity. Clin Toxi col 2006;44:275-81. [5] Henry JA, Alexander CA, Sener EK. Re lative mortality from overdose of antidepressants. Brit Med J 1995;310:221-4. [6] Morgan O, Griffiths C, Baker A, Majeed A. Fatal toxicity of antidepressants in England and Wales, 1993. Health Statistics Quarterly 2004;23:18-24. [7] Kiyan S, Aksay E, Yanturali S, Atilla R, Ersel M. Acute Myocardial Infarction Associated with Amitriptyline Overdose. Basic Clin Pharmacol 2006;98:462-66. [8] Kerr GW, McGuffie AC, Wilkie S. Tri cyclic antidepressant overdose: a review. Emerg Med J 2001;18:236-41. [9] Nau C, Seaver M, Wang S, Wang GK. Block of Human Heart hH1 Sodium Channels by Amitriptyline. J Pharmacol Exp Ther 2000;292:1015-23. [10] Heard K, Cain BS, Dart RC, Cairns CB. Tricyclic Antidepressants Directly Depress Human Myocardial Mechanical Function Independent of Effects on the Conduction System. Acad Em erg Med 2001;8(12):1122-7. [11] Olson KR. Poisoning & Drug Over dose. New York: McGrawHill, 2007. [12] Ludot H, Tharin JY, Belooudah M, Mazoit JX, Malinovsky JM. Successful resuscitation after ropivacaine and lidoc aine-induced ventricular arrhythmia following posterior lumbar plexus blo ck in a child. Anesth Analg 2008;106:15724. [13] Warren J, Brian T, G eorgescu A, Shah S. Intravenous lipid infusion in the successful resuscitation of local anesthet ic-induced cardiovascular collapse after supraclavicular brachial plexus bl ock. Anesth Analg 2008;106:1578-80. 216

PAGE 217

[14] Litz RJ, Roessel T, Heller AR, Stehr SN. Reversal of central nervous system and cardiac toxicity after local anesthetic in toxication by lipid emulsion injection. Anesth Analg 2008;106:1575-7. [15] Whiteside J. Reversal of local anaesthetic induced CNS toxicity with lipid emulsion. Anaesthesia 2008;63:202-13. [16] Nau C, Wang S, Strichartz GR, W ang GK. Block of Human Heart hH1 Sodium Channels by the Enantiomers of Bupivacaine. Anesthesiology 2000;93:1022-33. [17] Kaufman B, Wahlander S. Local Anesthetics. In: Go ldfrank LR, Flomenbaum NE, Lewin NA, Howland MA, Hoffman RS, Nelson LS, editors. Goldfranks Toxicologic Emergencies, 7th ed. New York: McGraw -Hill, 2002. p. 824-34. [18] Rahman YE, Rosenthal MW, Cerny EA. Intracellular plutonium: Removal by liposome encapsulated chelati ng agent. Science 1973;180:300-2. [19] Varshney M, Morey TE, Shah DO, Flint JA, Moudgil BM, Seubert CN, Dennis DM. Pluronic Microemulsions as Nanoreser voirs for Extraction of Bupivacaine from Normal Saline. J Am Chem Soc 2004;126:5108-12. [20] James-Smith MA, Shekhawat D, Moudgil BM, Shah DO. Determination of Drug and Fatty Acid Binding Capacity to Pl uronic F127 in Microemulsions. Langmuir 2007;23:1640-4. [21] Underhill RS, Jovanovic AV, Carino SR, Varshney M, Shah DO, Dennis DM, Morey TE, Duran RS. Oil-Filled Silica Na nocapsules for Lipophilic Drug Uptake: Implications for Drug Detoxificati on Therapy. Chem Mater 2002;14:4919-25. [22] Lee D, Baney R. Detoxification of amit riptyline by oligochitosan derivatives. Biotechnol Lett 2004;26:713-6. [23] Chakraborty S, Somasundaran P. Sequestr ation of drugs using poly(acrylic acid) and alkyl modified poly(acrylic acid ) nanoparticles. Soft Matter 2006;2:850-4. [24] Dhanikula AB, Khalid NM, Lee SD, Yeung R, Risovic V, Wasan KM, Leroux J. Long circulating lipid nanocapsules for drug detoxification. Biomaterials 2007;28:1248-57. [25] Dhanikula AB, Lamontagne D, Leroux J. Rescue of am itriptyline-intoxicated hearts with nanosized vesicles. Cardiovasc Res 2007;74:480-86. [26] Lasic DD. Novel applications of liposomes. Trends Biotechnol 1998;16:307-21. 217

PAGE 218

[27] Szoka F, Papahadjopoulos D. Co mparative Properties and Methods of Preparation of Lipid Vesicles (Liposom es). Ann Rev Biophys Bioeng 1980;9:467508. [28] MacDonald RC, MacDonald RI, Menco BP, Takeshita K, Subbarao NK, Hu LR. Small-volume extrusion apparatus for preparation of large, unilamellar vesicles. Biochim Biophys Acta 1991;1061:297-303. [29] Woodle MC. Controlling liposome blood clearanc e by surface-grafted polymers. Adv Drug Delivery Rev 1998;32:139-52. [30] Awasthi VD, Garcia D, Klipper R, Goins BA, Phillips WT. Neutral and Anionic Liposome-Encapsulated Hemoglobin: E ffect of Postinserted Poly(ethylene glycol)-distearoylphosphatidylethanolam ine on Distribution and Circulation Kinetics. J Pharmacol Exp Ther 2004;309:241-8. [31] Hoarau D, Delmas P, David S, Roux E, Leroux JC. Novel Long-Circulating Lipid Nanocapsules. Pharm Res 2004;21:1783-89. [32] Lee CM, Choi Y, Huh EJ, Lee KY, Song HC, Sun MJ, Jeong HJ, Cho CS, Bom HS. Polyethylene Glycol (PEG) Modifi ed 99mTcHMPAO Liposome for Improving Blood Circulation and Biodistribution: The Effect of the Ext ent of PEGylation. Cancer Biother R adio 2005;20(6):620-8. [33] Rosenblatt MA, Abel M, Fischer GW, Itzkovich CJ, Eisenkraft JB. Successful Use of a 20% Lipid Emulsion to Resuscitate a Patient after a Presumed Bupivacainerelated Cardiac Arrest. Anesthesiology 2006;105:217-8. [34] Litz RJ, Popp M, Stehr SN, Koch T. Successful resuscitation of a patient with ropivacaine induced asystole after axillary plexus block using lipid infusion. Anaesthesia 2006;61:800-1. [35] Foxall G, McCahon R, Lamb J, Hard man JG, Bedforth NM. Levobupivacaineinduced seizures and cardiovascular colla pse treated with Intralipid. Anaesthesia 2007;62:516-8. [36] Zimmer C, Piepenbr ink K, Riest G, Peters J. Ca rdiotoxic and neurotoxic effects after accidental intravascular bupivacai ne administration: therapy with lidocaine propofol and lipid emulsion. Anaesthesist 2007;56:449-53. [37] Harvey M, Cave G. Intralipid outperfo rms sodium bicarbonate in a rabbit model of clomipramine toxicity. Ann Emerg Med 2007;49:178-85. [38] Tebbutt S, Harvey M, Nicho lson T, Cave G. Intralipid prolongs survival in a rat model of verapamil toxicity. Acad Emerg Med 2006;13:134-9. 218

PAGE 219

[39] Harvey M, Cave G. Intralipid infusion ameliorates propranolol induced hypotension in rabbits. J Med Toxicol 2008;4:71-6. [40] Sirianni AJ, Osterhoudt KC, Calello DP, Muller AA, Waterhouse MR, Goodkin MB, Weinberg GL, Henretig FM. Use of Li pid Emulsion in the Resuscitation of a Patient With Prolonged Cardiovascular Collapse After Overdose of Bupropion and Lamotrigine. Ann Emerg Med 2008;51:412-5. [41] Weinberg GL, VadeBoncouer T, Rama raju GA, Garcia-A maro MF, Cwik MJ. Pretreatment or Resuscitation with a Lipid Infusion Shifts the Dose Response to Bupivacaine-induced Asystole in Ra ts. Anesthesiology 1998;88:1071-5. [42] Weinberg G, Ripper R, Feinstein DL, Hoffman W. Lipid Emulsion Infusion Rescues Dogs From Bupiva caine-Induced Cardiac Toxicity. Reg Anesth Pain Med 2003;28:198-202. [43] Rowlingson JC. Lipid Rescue: A Step Forward in Patient Safety? Likely So! Anesth Analg 2008;106:1333-6. [44] Weinberg GL. Lipid Infusion Therapy: Translation to Clinical Practice. Anesth Analg 2008;106:1340-2. [45] Mathieu S, Cranshaw J. Treatment of severe local anaesthetic toxicity. Anaesthesia 2008;63:202-3. [46] Wildsmith JAW. Treatm ent of severe local anaest hetic toxicity. Anaethesia 2008;63:778-89. [47] Boogaerts J, Declercq A, Lafont N, Benameur H, Akodad EM, Dupont JC, Legros FJ. Toxicity of bupivacaine encapsulated into liposomes and injected intravenously: comparison with plain so lutions. Anesth Analg 1993;76:553-5. [48] Boogaerts JG, Lafont ND, Declercq AG, Luo HC, Gravet ET, Bianchi JA, Legros FJ. Epidural administration of liposome associated bupivacaine for the management of postsurgical pain: a firs t study. J Clin Anesth 1994;6:315-20. [49] Singh R, Vyas SP. Topical liposom al system for localized and controlled drug delivery. J Dermato l Sci 1996;13:107-11. [50] Fisher R, Hung O, Mezei M, Stewart R. Topical anaesthesia of intact skin: liposome-encapsulated tetracaine vs EMLA. Br J Anaesth 1998;81:972-3. [51] Eichenfield LF, Funk A, Fallon-Friedlander S, Cunning ham BB. A clinical study to evaluate the efficacy of ELA-Max (4% liposomal lidocaine) as compared with eutectic mixture of local anesthetics cream for pain r eduction of venipuncture in children. Pediatri cs 2002;109:1093-9. 219

PAGE 220

[52] Taddio A, Soin HK, Schuh S, Koren G, Scolnik D. Liposomal lidocaine to improve procedural success rates and reduce pr ocedural pain among children: a randomized controlled trial. Can Med Assoc J 2005;172:1691-5. [53] Franz-Montan M, Silva ALR, Cogo K, Bergamaschi CC, Volpato MC, Ranali J, Paula E, Groppo FC. Liposome-Encapsulated Ropivacaine for Topical Anesthesia of Human Oral Muco sa. Anesth Analg 2007;104:1528-31. [54] Casarotto MG, Craik DJ. Solution Dy namics of Imipramine and Amitriptyline in Phospholipid Vesicles as Studied by 1H and 13C NMR. J Colloid Interf Sci 1993;158:326-32. [55] Austin RP, Barton P, Davis AM, Fe ssey RE, Wenlock MC. The Thermodynamics of the Partitioning of Ionizing Mo lecules Between Aqueous Buffers and Phospholipid Membranes. Pharm Res 2005;22:1649-57. [56] Ahyayauch H, Goni FM, Bennouna M. In teraction of electrically neutral and cationic forms of imipramine with lipos ome and erythrocyte membranes. Int J Pharm 2004;279:51-8. [57] Fisar Z, Fuksova K, Velenovska M. Binding of Imipramine to Phospholipid Bilayers Using Radioligand Binding Assa y. Gen Physiol Bi ophys 2004;23:77-99. [58] Sanganahalli BG, Joshi PG, Joshi NB. Differential effects of tricyclic antidepressant drugs on membrane dynam ics-a fluorescence spectroscopic study. Life Sci 2000;68:81-90. [59] Suwalsky M, Schneider C, Villena F, Norris B, Cardenas H, Cuevas F, Sotomayor CP. Structural Effects of the Local Anesthetic Bupivacaine Hydrochloride on the Hu man Erythrocyte Membrane and Molecular Models. Blood Cell Mol Dis 2002;29:14-23. [60] Mizogami M, Tsuchiya H, Ueno T, Ka shimata M, Takakura K. Stereospecific Interaction of Bupivacaine En antiomers with Lipid Membranes. Region Anesth Pain M 2008;33:304-11. [61] Bryl K, Kedzierska S, Taylor A. Mechanism of the Leakage Induced on Lipid Model Membranes by Rz1 Lipoprotein, the Bacteriophage Rz1 Gene Product: A Fluorescence Study. J Fluoresc 200;10:209-16. [62] Nikolova AN, Jones MN. Effect of grafted PEG-2000 on the size and permeability of vesicles. Biochim Biophys Acta 1996;1304:120-8. 220

PAGE 221

[63] Hashizaki K, Taguchi H, Sakai H, Abe M, Saito Y, Ogawa N. Carboxyfluorescein Leakage from Poly(ethylene glycol)-G rafted Liposomes Induced by the Interaction with Serum. Chem Pharm Bull 2006;54:80-4. [64] Yokouchi Y, Tsunoda T, Imura T, Yam auchi H, Yokoyama S, Sakai H, Abe M. Effect of adsorption of bovine se rum albumin on liposomal membrane characteristics. Colloid Surface B 2001;20:95-103. [65] Yokoyama S, Takeda T, Tsunoda T, Ohta Y, Imura T, Abe M. Membrane properties of mixed dipalmitoylphosphat idylglycerol/ganglioside GM3 liposomes in the presence of bovine serum al bumin. Colloid Surface B 2002;27:141-6. [66] Sadzuka Y, Nakade A, Tsuruda T, S onobe T. Study on the characterization of mixed polyethyleneglycol modified liposom es containing doxorubicin. J Control Release 2003;91:271-80. [67] Monkkonen J, Liukkonen J, Taskinen M, Heath TD, Urtti A. Studies on liposome formulations for intra-articular deliv ery of clodronate. J Control Release 1995;35:145-54. [68] Soni V, Kohli DV, Jain SK. Trans ferrin coupled liposomes as drug delivery carriers for brain targeting of 5-florouracil. J Drug Ta rget 2005;13:245-50. [69] Leroux JC. Injectable nanocarriers for biodetoxification. Nat Nanotechnol 2007;2:679-84. [70] Dhanikula AB, Lafleur M, Leroux J. Characterization and in vitro evaluation of spherulites as sequestering vesicles with potential application in drug detoxification. Biochim Biophys Acta 2006;1758:1787-96. [71] Fallon M, Chauhan A. Sequestration of amitriptyline by liposomes. J Colloid Interf Sci 2006;300:7-19. [72] Deo N, Somasundaran T, Somasundaran P. Solution properties of amitriptyline and its partitioning into lipid bilayers. Colloids Surf B 2004;34:155-9. [73] Poulin P, Theil FP. Prediction of Pha rmacokinetics prior to In Vivo Studies. II. Generic Physiologically Based Pharmaco kinetic Models of Drug Disposition. J Pharm Sci 2002;91:1358-70. [74] Lav T, Parrott N, Grimm HP, Fleury A, Reddy M. Challenges and opportunities with modelling and simulation in drug discovery and drug development. Xenobiotica 2007;37(10):1295-310. [75] Parrott N, Lave T. Applications of Physiologically Based Absorption Models in Drug Discovery and Developmen t. Mol Pharm 2008;5(5):760-75. 221

PAGE 222

[76] Thomas S. Clinical relevance of predictive physiologi cally based pharmacokinetic methods. Expert Opin Dr ug Dis 2008;3(7):725-32. [77] Nestorov I. Whole Body Pharma cokinetic Models. Clin Pharmacokinet 2003;42(10):883-908. [78] Igari Y, Sugiyama Y, Sawada Y, Iga T, Hanano M. Prediction of diazepam disposition in the rat and man by a physi ologically based pharmacokinetic model. J Pharmacokinet Biopharm 1983;11:577-93. [79] Meno-Tetang GML, Li H, Mis S, Pyszczynski N, Heining P, Lowe P, Jusko WJ. 2006. Physiologically Based Pharmacoki netic Modeling of FTY720 (2-Amino-2[2(-4-octylphenyl)ethyl]propane-1,3-diol hydr ochloride) in Rats After Oral and Intravenous Doses. Drug Me tab Dispos 2006;34:1480-7. [80] Garg A, Balthasar JP. Physiologica lly-based pharmacokinetic (PBPK) model to predict IgG tissue kinetics in wild-type and FcRn-knockout mice. J Pharmacokinet Phar 2007;34:687-709. [81] Bonate PL, Swann A, Silverman PB. Preliminary Physiologically Based Pharmacokinetic Model for Cocaine in the Rat: Model Development and ScaleUp to Humans. J Pharm Sci 1996;85(8):878-83. [82] Peters SA. Identificat ion of Intestinal Loss of a Drug through Physiologically Based Pharmacokinetic Simulation of Pl asma Concentration-Time Profiles. Clin Pharmacokinet 2008;47(4):245-59. [83] Guyton AC. Textbook of Medical Ph ysiology. Philadelphia: Saunders, 1986. [84] Franssen EJF, Kunst PWA, Bet PM, Strack van Schijndel RJM, van Loenen AC. Toxicokinetics of Nortripty line and Amitriptyline: Two Ca se Reports. Ther Drug Monit 2003;25:248-51. [85] Benet LZ, Hoener B. Changes in plas ma protein binding have little clinical relevance. Clin Pharmacol Ther 2002;71:115-21. [86] Chonn A, Semple SC, Cullis PR. Association of Blood Proteins with Large Unilamellar Liposomes in Vivo J Biol Chem 1992;267:18759-65. [87] Semple SC, Chonn A, Cullis PR. Intera ctions of liposomes and lipid-based carrier systems with blood proteins: Relation to clearance behavior in vivo. Adv Drug Deliver Rev 1998;32:3-17. 222

PAGE 223

[88] Levitt MA, Sullivan JB, Owens SM, Bur nham L, Finley PR. Amitriptyline plasma protein binding: Effect of plasma pH and relevance to clinical overdose. Am J Emerg Med 1986;4:121-5. [89] Franssen EJF, Kunst PWA, Bet PM, Strack van Schijndel RJM, van Loenen AC. Toxicokinetics of Nortripty line and Amitriptyline: Two Ca se Reports. Ther Drug Monit 2003;25:248-51. [90] Awasthi VD, Garcia D, Goins BA, Phillips WT. Circulation and biodistribution profiles of long-circulating PEG-liposomes of various sizes in rabbits. Int J Pharm 2003;253:121-32. [91] Coune A, Sculier JP, Fruhling J, Stryckmans P, Brassi nne C, Ghanem G, Laduron C, Atassi G, Ruyssc haert JM, Hildebrand J. IV Administration of a Water-Insoluble Anti-Mitotic Compund En trapped in Liposomes Preliminary Report on Infusion of Large Volumes of Li posomes To Man. Cancer Treat Rep 1983;67:1031-33. [92] Sculier JP, Coune A, Brassinne C, Laduron C, Atassi G, Ruysschaert JM, Fruhling J. Intravenous Infusion of High Doses of Liposomes Containing NSC251635, A Water-Insoluble Cytostatic Agent A Pilot Study with Pharmacokinetic Data. J Clin Oncol 1986;4:789-97. [93] Weissig V, Babich J, Torchilin V. Long-circulating gadolinium-loaded liposomes: potential use for magnetic resonance imagi ng of the blood pool. Colloids Surf B 2000;18:293-99. [94] Lapinski MM, Castro-For ero A, Greiner AJ, Ofoli RY Blanchard GJ. Comparison of Liposomes Formed by Sonication and Ex trusion: Rotational and Translational Diffusion of an Embedded Chromophor e. Langmuir 2007;23:11677-83. [95] Yoshida A, Hashizaki K, Yamauchi H, Sakai H, Yoko yama S, Abe M. Effect of Lipid with Covalently Attached Poly(ethylen e glycol) on the Surface Properties of Liposomal Bilayer Membranes. Langmuir 1999;15:2333-7. [96] Newman MS, Colbern GT, Working PK, Engbers C, Amantea MA. Comparative pharmacokinetics, tissue distribution, and therapeutic effectiveness of cisplatin encapsulated in long-circulating, pegylated liposomes (SPI-077) in tumor-bearing mice. Cancer Chemoth Pharm 1999;43:1-7. [97] Nicolazzi C, Mignet N, Figuera N, Cadet M, Ibad RT, Seguin J, Scherman D, Bessodes M. Anionic polyethyleneglycol lipids added to cationic lipoplexes increase their plasmatic circulation time. J Control Release 2003;88:429-43. 223

PAGE 224

[98] Zhigaltsev IV, Maurer N, Wong KF, Cullis PR. Triggered release of doxorubicin following mixing of cationic and anioni c liposomes. Biochim Biophys Acta 2002;1565:129-35. [99] Rex S, Zuckermann MJ, Lafleur M, Silvius JR. Ex perimental and Monte Carlo Simulation Studies of the Thermodynamics of Polyethyleneglycol Chains Grafted to Lipid Bilayers. Biophys J 1998;75:2900-14. [100] Moghimi SM, Szebeni J. Stealth lipos omes and long circulating nanoparticles: critical issues in pharmacokinetics, opsonization and protein-binding properties. Prog Lipid Res 2003;42:463-78. [101] Gbadamosi JK, Hunter AC, Moghimi SM. PEGylation of microspheres generates a heterogeneous population of par ticles with differential surface characteristics and biological performance. FEBS Lett 2002;532:338-44. [102] Attwood D, Natarajan R. Effect of pH on the micellar properti es of amphiphilic drugs in aqueous solution. J Pharm Pharmacol 1981;33:136-40. [103] Wiczling P, Markuszewski MJ, Kali szan M, Galer K, Kaliszan R. Combined pH/organic solvent gradient HPLC in analysi s of forensic material. J Pharmaceut Biomed 2005;37:871-75. [104] Brittain H, Prankerd RJ. Profiles of Drug Substances, Excipients, and Related Methodology, Volume 33: Critical Compilation of Pka Values for Pharmaceutical Substances. London: Academic Press, 2007. p. 155. [105] Musshoff F, Padosch S, Stei nborn S, Madea B. Fatal blood and tissue concentrations of more than 200 drugs Forensic Sci Int 2004;142:161-210. [106] Rodgers T, Leahy D, Rowland M. Ph ysiologically Based Pharmacokinetic Modeling 1: Predicting the Tissue Distribution of Moder ate-to-Strong Bases. J Pharm Sci 2005;94:1259-76. [107] Kristensen CB. Imipramine serum pr otein binding in healthy subjects. Clin Pharmacol Ther 1983;34:689-94. [108] Obach RS. Prediction of Human Clear ance of Twenty-nine Drugs from Hepatic Microsomal Intrinsic Clearance Data: An Examination of In Vitro Half-Life Approach and Nonspecific Binding to Microsomes. Drug Metab Dispos 1999;27:1350-59. [109] Callaghan JT, Cerimele BJ, Kassahun KJ, Nyhart Jr EH, Hoyes-Beehler PJ, Kondraske GV. Olanzapine: inte raction study with imipramine. J Clin Pharmacol 1997;37:971-78. 224

PAGE 225

[110] Pedersen OL, Gram LF, Kristensen CB, Moller M, Thayssen P, Bjerre M, KraghSorensen P, Klitgaard NA, Sindrup E, Ho le P, Brinklov M. Overdosage of Antidepressants: Clinical and Pharmacokinetic Aspects. Eur J Clin Pharmacol 1982;23:513-21. [111] Swartz CM, Sherman A. The Treatment of Tricyclic Antidepressant Overdose with Repeated Charcoal. J Clin Psychopharmacol 1984;4:336-40. [112] Ilett KF, Hackett LP, Dusci LJ, Pate rson JW. Disposition of Dothiepin After Overdose: Effects of Repeated-Dose Ac tivated Charcoal. Ther Drug Monit 1991;13:485-9. [113] Ragusi C, Boschi G, Ris ede P, Rips R, Harrison K, Scherrmann J. Influence of various combinations of specific antibody dose and affinity on tissue imipramine redistribution. Brit J Pharmacol 1998;125:35-40. [114] Ragusi C, Scherrmann J, Harrison K, Sm ith DS, Rips R, Boschi G. Redistribution of Imipramine from Regions of the Br ain Under the Influence of Circulating Specific Antibodies. J Neurochem 1998;70:2099-105. [115] Pentel PR, Scarlett W, Ross CA, Landon J, Sidki A, Keyler DE. Reduction of Desipramine Cardiotoxicity and Prolongation of Survival in Rats With the Use of Polyclonal Drug-Specific Antibody Fab Fragments. Ann Emerg Med 1995;26:334-41. [116] Shelver WL, Keyler DE, Lin G, Murt augh MP, Flickinger MC, Ross CA, Pentel PR. Effects of Recombinant Drug-Specif ic Single Chain Antibody Fv Fragment on [3H]-Desipramine Distribution in Rats. Biochem Pharmacol 1996;51:531-7. [117] Fallon MS, Varshney M, Dennis DM, Chauhan A. A Physiologically-Based Pharmacokinetic Model of Drug Detoxifica tion by Nanoparticl es. J Pharmacokinet Phar 2004;31:381-400. [118] Dauphin A, Gupta RN, Young JEM, Morton WD. Serum bupivacaine concentrations during continuous extr apleural infusion. Can J Anaesth 1997;44:367-70. [119] Mazoit JX, Cao LS, Samii K. Binding of Bupivacaine to Human Serum Proteins, Isolated Albumin and Isolat ed Alpha-1-Acid Glycoprotein. Differences between the Two Enantiomers are Partly Due to Cooperativity. J Pharmacol Exp Ther 1996;276:109-15. [120] Morishima HO, Pedersen H, Finster M, Hiraoka H, Tsuji A, Feldman HS, Arthur GR, Covino BG. Bupivacaine Toxicity in Pregnant and Nonpregnant Ewes. Anesthesiology 1985;63:134-9. 225

PAGE 226

[121] Martinez-Gomez MA, Villanueva-Ca manas RS, Sagrado S, Medina-Hernandez MJ. Evaluation of enantioselective binding of basic drugs to plasma by ACE. Electrophoresis 2007;28:3056-63. [122] Weinberg GL, Ripper R, Murphy P, Edelman LB, Hoffman W, Strichartz G, Feinstein DL. Lipid infusion accelerates removal of bupivacaine and recovery from bupivacaine toxicity in the is olated rat heart. Reg Anesth Pain Med 2006;31:296-303. [123] Strichartz GR, Sanchez V, Arthur GR, Chafetz R, Martin D. Fundamental Properties of Local Anesthetics. II. Measured Octanol:Buffer Partition Coefficients and pKa Values of Clinically Us ed Drugs. Anesth Analg 1990;71:158-70. [124] Zuidam NJ, Vrueh R, Crommelin DJA. Characterization of Liposomes. In: Torchilin VP, Weissig V, editors. Liposomes, 2nd ed. New York: Oxford University Press, 2003. p. 31-61. [125] Egorova EM. Dissociation constants of lipid ionizable groups II. Changes in surface pK at low ionic strengths Colloid Surface A 1998;131:19-31. [126] McLaughlin S. The Electrostatic Proper ties of Membranes. Annu Rev Biophys Bio 1989;18:113-36. [127] Lis LJ, McAlister M, Fuller N, Ra nd RP, Parsegian VA. Interactions Between Neutral Phospholipid Bilayer Me mbranes. Biophys J 1982;37:657-65. [128] Rand RP, Parsegian VA. Hydration forces between phospholipid bilayers. Biochim Biophys Acta 1989;988:351-76. [129] Kenworthy AK, Hristova K, Needham D, McIntosh TJ. Range and Magnitude of the Steric Pressure Between Bilayers Containing Phospholipid s with Covalently Attached Poly(ethylene glyco l). Biophys J 1995;68:1921-36. [130] Arnold K, Herrmann A, Pr atsch L, Gawrisch K. The dielectric properties of aqueous solutions of poly(ethylene glyco l) and their influence on membrane structure. Biochim Bio phys Acta 1985;815: 515-8. [131] Kuhl TL, Leckband DE, Lasic DD, Israelachvili JN. Modulation of interaction forces between bilayers exposing s hort-chained ethylene oxide headgroups. Biophys J 1994;66:1479-88. [132] Blow AMJ, Botham GM, Lucy JA. Ca lcium ions and cell fusion. Effects of chemical fusogens on the permeability of er ythrocytes to calcium and other ions. Biochem J 1979;182:555-63. 226

PAGE 227

[133] Elworthy PH, McIntosh DS. The effect of solvent dielectric constant on micellisation by lecithin. Co lloid Polym Sci 1964;195:27-34. [134] Parsegian A. Energy of an Ion crossi ng a Low Dielectric Membrane: Solutions to Four Relevant Electrostatic Problems. Nature 1969;221:844-6. [135] Hiemenz PC, Rajagopalan R. Principles of Colloid and Surface Chemistry, 3rd ed. New York: Marcel Dekker, 1997. p. 499-533. [136] Silvander M, Hansson P, Edwards K. Liposomal Surface Potential and Bilayer Packing As Affected by PEG-Lipid Inclusion. Langmuir 2000;16:3696-702. [137] Malheiros SVP, Pinto LM A, Gottardo L, Yokaichiya DK, Fraceto LF, Meirelles NC, de Paula E. A new look at the hem olytic effect of local anesthetics, considering their real membrane/water partitioning at pH 7.4. Biophys Chem 2004;110:213-21. [138] Balali-Mood KT, Harroun A, Bradshaw JP. Molecular dynamics simulations of a mixed DOPC/DOPG bilayer. Eur Phys J E 2003;12:S135-40. [139] Kim JH, Kim MW. Tem perature effect on the trans port dynamics of a small molecule through a liposome bila yer. Eur Phys J E 2007;23:313-7. [140] Balgavy P, Dubnickova M, Kucerka N, Kiselev MA, Yaradaikin SP, Uhrikova D. Bilayer thickness and lipid interf ace area in unilamellar extruded 1,2diacylphosphatidylcholine liposomes: a sm all-angle neutron scattering study. Biochim Biophys Acta 2001;1512:40-52. [141] Zima AV, Qin J, Fill M, Blatter LA. Tr icyclic antidepressant amitriptyline alters sarcoplasmic reticulum calcium handling in ventricular myocytes. Am J PhysiolHeart 2008;295:H2008-H2016. [142] Poulin P, Theil FP. A pr iori prediction of tissue: pl asma partition coefficients of drugs to facilitate the use of physiol ogically-based pharmacokinetic models in drug discovery. J Pharm Sci 2000;89(1):16-35. [143] Poulin P, Schoenlein K, Theil FP. Prediction of adipose tissue: plasma partition coefficients for structurally unrelated drugs. J Pharm Sci 2001;90(4):436-47. [144] Poulin P, Theil FP. Prediction of pha rmacokinetics prior to in vivo studies. I. Mechanism-based prediction of volume of distribution. J Pharm Sci 2002;91(1):129-56. [145] Rodgers T, Rowland M. Physiologically Based Pharmacokinetic Modelling 2: Predicting the Tissue Distribution of Acids, Very Weak Bases, Neutrals and Zwitterions. J Pharm Sci 2006;95(6):1238-57. 227

PAGE 228

[146] Banerjee T, Singh SK, Kishore N. Bi nding of Naproxen and Amitriptyline to Bovine Serum Albumin: Biophysical Aspects. J Phys Chem B 2006;110:2414756. [147] Sallee FR, Pollack BG. Clinical pharma cokinetics of imipramine and desipramine. Clin Pharmacokinet 1990;18:346-64. [148] Abernathy DR, Greenblatt DJ, Shader RI. Imipramine and desipramine disposition in the elderly. J Pharmacol Exp Ther 1985;232:183-8. [149] Schulz P, Turner-Tamiyasu K, Smith G, Giacomini KM, Blaschke TF. Amitriptyline disposition in young and elderly normal men. Clin Pharmacol Ther 1983;33:360-6. [150] Rollins DE, Alvan G, Bertilsson L, Gille tte JR, Mellstrom B, Sjoqvist F, Traskman L. Interindividual differences in amitriptyline demethylation. Clin Pharmacol Ther 1980;28:121-9. [151] Tucker GT, Mather LE. Pharmacokineti cs of Local Anaesthetic Agents. Brit J Anaesth 1975;47:213-24. [152] Tucker GT, Boyes RN, Bridenbaugh PO, Moore DC. Binding of Anilide-type Local Anesthetics in Human Plasma: II Implications In Vivo, with Special Reference to Transplacental Distribut ion. Anesthesiology 1970;33:304-13. [153] Leo A, Hansch C, Elkins D. Partit ion Coefficients and Their Uses. Chem Rev 1971;71:525-616. [154] Verschueren K. Handbook of Environm ental Data on Organic Chemicals, 4th ed. New York: John Wiley and Sons, 2001. [155] Nava-Ocampo AA, Bello-Ramirez AM. Lipophilicity Affects the Pharmacokinetics and Toxicity of Local Anaesthetic Agents Ad ministered by Caudal Block. Clin Exp Pharmacol P 2004;31:116-8. [156] Martensson E, Axelsson R, Nyberg G, Svensson C. Pharmacokinetic properties of the antidepressant drugs amitriptyline, clomipramine, and imipramine: a clinical study. Curr Ther Res 1984;36:228-38. [157] Brinkschulte M, Breyer-Pfaff U. Bindi ng of tricyclic antidepr essants and perazine to human plasma. Methodology and findings in normals. Naunyn-Schmiedeberg's Archives of Pharmacology 1979;308:1-7. 228

PAGE 229

[158] Tucker GT, Boyes RN, Bridenbaugh P O, Moore DC. Binding of Anilide-type Local Anesthetics in Human Plasma : I. Relationships between Binding, Physiochemical Properties, and Anesthetic Activity. Anesthesiology 1970;33:287303. [159] Baumann P, Jonzier-Perey M, Koeb L, Le PK, Tinguely D, Schopf J. Amitriptyline Pharmacokinetics and Clinica l Response: I. Free and Tota l Plasma Amitriptyline and Nortriptyline. Int Clin Psychopharm 1986;1:89-101. [160] Rowland M, Tozer TN. Clinical pharmacokinetics: concepts and applications, 3rd ed. New York: Lippincott Williams & Wilkins, 1995. p. 502. [161] Bynum ND, Poklis JL, Gaffney-K raft M, Garside D, Ropero-Miller JD. Postmortem Distribution of Tramadol, Amitriptyline, and Their Metabolites in a Suicidal Overdose. J A nal Toxicol 2005;29:401-6. [162] Tracqui A, Kintz P, Ritter-Lohner S, Mangin P, Lugnier A, Chaumont A. Toxicological findings after fatal amitriptyline self poisoning. Hum Exp Toxicol 1990;9:257-61. [163] Hilberg T, Ripel A, Sm ith AJ, Slordal L, Morland J, Bjorneboe A. Postmortem Amitriptyline Pharmacokinetics in Pigs after Oral and Intravenous Routes of Administration. J Forens ic Sci 1998;43(2):380-7. [164] Shi B, Heavner JE. Nitric Oxide Modul ation Affects the Tissue Distribution and Toxicity of Bupivacaine. Pha rmacol Biochem Be 2000;66(3):623-9. [165] Feldman HS, Hartvig P, Wiklund L, Doucette AM, Antoni G, Gee A, Ulin J, Langstrom B. Regional Distribution of 11C-Labeled Lidocaine, Bupivacaine, and Ropivacaine in the Heart, Lungs, and Sk eletal Muscle of Pigs Studied with Positron Emission Tomogr aphy. Biopharm Drug Dispo s 1997;18(2):151-64. [166] Sjostrand U, Widman B. Distribution of Bupivacaine in the Rabbit under Normal and Acidotic Conditions. Acta Anaesth Scand 1973;17(s50):5-24. [167] Goehl TJ, Davenport JB, Stanley MJ. Distribution, Biotransformation and Excretion of Bupivacaine in t he Rat and the Monkey. Xenobiotica 1973;3(12):761-72. [168] Pounder DJ, Jones GR. Postmortem drug redistribution-a toxico-logical nightmare. Forensic Sci Int 1990;45:253-63. [169] Shargel L, Wu-Pong S, Yu ABC. Applied Biopharmaceutics and Pharmacokinetics, 5th ed. New York: McGraw-Hill, 2005. p. 164,339. 229

PAGE 230

[170] Simone de G, Devereux RB, Daniel s SR, Mureddu G, Roman MJ, Kimball TR, Greco R, Witt S, Contaldo F. Stroke Vo lume and Cardiac Output in Normotensive Children and Adults Assessment of Rela tions With Body Size and Impact of Overweight. Circulation 1997;95:1837-43. [171] Benowitz N, Forsyth RP Melmon KL, Rowland M. Lidocaine disposition kinetics in monkey and man I: Prediction by a perfusion model. Clin Pharamcol Ther 1974;16:87-98. [172] Gray H. Anatomy of The Human Body, 20th ed. New York: Lea and Febiger, 1918. p. 1199-273. [173] Abernethyl DR, Divoll M, Greenblatt DJ, Harmatz JS, Shader RI. Absolute bioavailability of imipramine: In fluence of food. Psychopharmacology 1984;83:104-6. [174] Brosen K, Gram LF. First-pass me tabolism of imipramine and desipramine: Impact of the sparteine oxidation phenot ype. Clin Pharmacol Ther 1988;43:4006. [175] Filser JG, Kaumeier S, Brand T, Schanz H, Terlinden R, Muller WE. Pharmacokinetics of Amitriptyline and Amitri ptylineoxide after Intravenous or Oral Administration in Humans. Pharmacopsychiatry 1988;21:381-93. [176] Jorgensen A, Hansen V. Pharmacokineti cs of Amitriptyline Infused Intravenously in Man. Eur J Clin Pharmacol 1976;10:337-41. [177] Burm AGL, De Boer AG, Van Kl eef JW, Vermeulen NP E, De Leede LGJ, Spierdijk J, Breimer DD. Pharmacokinet ics of Lidocaine and Bupivacaine and Stable Isotope Labelled Anal ogues: A Study In Healthy Volunteers. Biopharm Drug Dispos 1988;9:85-95. [178] Reynolds F. Metabolism and Excreti on of Bupivacaine in Man: A Comparison with Mepivacaine. Brit J Anaesth 1971;43:33-6. [179] Tucker GT, Wiklund L, Berlin-Wahlen A, Mather LE. Hepatic Clearance of Local Anesthetics in Man. J Pha rmacokinet Biop 1977;5:111-21. [180] Wiklund L. Human Hepatic Blood Flow and its Relation to Systemic Circulation during Intravenous Infusion of Bupivacai ne or Etidocaine. Acta Anaesth Scand 1976;21:189-99. [181] Fox SI. Human Physiology, 9th ed. New York: McGraw-Hill, 2006. p. 383. [182] Burch JE, Herries DG. The Demethylati on of Amitriptyline Administered by Oral and Intramuscular Routes. Psychopharmacology 1983;80:249-53. 230

PAGE 231

[183] Wong SL, Cavanaugh J, Shi H, Awni WM, Granneman GR. Effects of divalproex sodium on amitriptyline and nortriptyline pharmacokinetics. Clin Pharmacol Ther 1996;60:48-53. [184] Sjoqvist F, Berglund F, Borga O, Hammer W, Andersson S, Thorstrand C. The pH-dependent excretion of monomethylated tricyclic antidepressants. Clin Pharmacol Ther 1969;10:826-33. [185] Friedman GA, Rowlingson JC, DiFazio CA, Donegan MF. Evaluation of the Analgesic Effect and Urinary Excretion of Systemic Bupivacaine in Man. Anesth Analg 1982;61:23-7. [186] Tucker GT. Pharmacokinetics of Local Anaesthetics. Brit J Anaesth 1986;58:71731. [187] Welling PG. Pharmacokinetics Proce sses, Mathematics, and Applications, 2nd ed. Washington DC: American Chem ical Society, 1997. p. 224. [188] Burch JE, Hullin RP. Amitriptylin e Pharmacokinetics A Crossover Study with Single Doses of Amitriptyline and Nortriptyline. Psychopharmacology 1981;74:35-42. [189] Burch JE, Hullin RP. Amitriptyline Pharmacokinetics Single Doses of Lentizol Compared with Ordinary Amit riptyline Tablets. Psychopharmacology 1981;74:4350. [190] Gupta SK, Shah JC Hwang SS. Pharmacokinetic and pharmacodynamic characterization of OROS and immediate-release amitriptyline. J Clin Pharmacol 1999;48:71-8. [191] Venkatakrishnan K, Schmider J, Harmatz JS, Ehrenberg BL, von Moltke LL, Graf JA, Mertzanis P, Corbett KE, Rodriguez MC, Shader RI, Greenbl att DJ. Relative contribution of CYP3A to am itriptyline clearance in humans: in vitro and in vivo studies. J Clin Pharmacol 2001;41:1043-54. [192] Tucker GT. Determination of Bupiva caine (Marcaine) and Other Anilide-type Local Anesthetics in Human Blood and Plasma by Gas Chromatography. Anesthesiology 1970;32:255-60. [193] Narchi P, Benhamou D, Bouaziz H, Fernandez H, Mazoit JX. Serum concentrations of local anaesthetics fo llowing intraperitoneal administration during laparoscopy. Eur J Clin Pharmacol 1992;42:223-5. 231

PAGE 232

232 [194] Morrison LMM, Emanuelsson BM, Mc Clure JH, Pollok AJ, McKeown DW, Brockway M, Jozwiak H, Wildsmith JA W. Efficacy and kinetics of extradural ropivacaine: comparison with bupivac aine. Brit J Anaesth 1994;72:164-9. [195] Kopacz DJ, Emanuelsson BM, Thompson GE, Carpenter RL, Stephenson CA. Pharmacokinetics of Ropivacaine and Bu pivacaine for Bilateral Intercostal Blockade in Healthy Male Volunteer s. Anesthesiology 1994;81:1139-48. [196] Kopacz DJ, Bernards CM, Allen HW, Landau C, Nandy P, Wu D, Lacouture PG. A Model to Evaluate the Pharmacokinetic and Pharmacodynamic Variables of Extended-Release Products Using In Vi vo Tissue Microdialysis in Humans: Bupivacaine-Loaded Microcapsules. Anesth Analg 2003;97:124-31. [197] Emanuelsson BMK, Zaric D, Nydahl PA, Axelsson KH. Pharmacokinetics of Ropivacaine and Bupivacaine During 21 Hours of Continuous Epidural Infusion in Healthy Male Volunteers. Anesth Analg 1995;81:1163-8. [198] Schug SA, Payne JP, Baker P, Holfor d NHG. Non-stationary pharmacokinetics of bupivacaine during continuous interpleural infusion. Acute Pa in 1997;1:15-20. [199] New RRC. Liposomes: a practical approach. New York: Oxford University Press, 1990. p. 55. [200] Finn SDH, Uncles DR, Willers J, Sabl e N. Early treatment of a quetiapine and sertraline overdose with Intra lipid. Anaesthesia 2009;64:191-4. [201] Gabizon A, Horowitz AT, Goren D, Tz emach D, Shmeeda H, Zalipsky S. In Vivo Fate of Folate-Targeted Polyethylene-Gl ycol Liposomes in Tumor-Bearing Mice. Clin Cancer Res 2003;9:6551-9. [202] Spina E, Henthorn TK, Eleborg L, Nordin C, Sawe J. Desmethylimipramine Overdose: Nonlinear Kinet ics in a Slow Hydroxylator. Ther Drug Monit 1985;7:239-41. [203] Tucker GT, Mather LE. Clinical Pha rmacokinetics of Local Anaesthetics. Clin Pharmacokinet 1979;4:241-78. [204] Venkatakrishnan K, Grenblatt DJ, von Moltke LL, Schmider J, Harmatz JS, Shader RI. Five Distinct Human Cyto chromes Mediate Amitriptyline NDemethylation In Vitro: Dominance of CYP 2C19 and 3A4. J Clin Pharmacol 1998;38:112-21. [205] Tetzlaff JE. Clinical Pharmacology of Local Anesthetics. Boston: ButterworthHeinemann, 2000. p. 41-5, 115-23.

PAGE 233

BIOGRAPHICAL SKETCH Brett A. Howell was born in Salisbury, North Carolina in 1983. After graduating from South Rowan High School (China Grov e, North Carolina) in May of 2002, he began his undergraduate studi es at North Carolina State Un iversity in Raleigh, North Carolina in August of 2002. He received hi s Bachelor of Science degrees in both chemical and textile engine ering in May of 2006, and joined the Department of Chemical Engineering at the University of Florida in the Fall of 2006. Shortly thereafter, in January of 2007, he joined Dr. Anuj Chauhans research group, where he has since worked to complete his doctoral research on drug overdose treatm ent with liposomes. 233