macroporous hydrogels as vascularizable soft tissue

Macroporous Hydrogels as Vascularizable Soft Tissue – Implant Interfaces:

Materials Characterization, In Vitro Evaluation, Computer Simulations, and

Applications in Implantable Drug Delivery Devices

A Thesis

Submitted to the Faculty

of

Drexel University

By

Thomas D. Dziubla

in partial fulfillment of the

requirements for the degree

of

Doctor of Philosophy

November 2002

ii

Dedications

This thesis is for my parents, Ray and Cathy. Mom and Dad, I am grateful for all you have done and continue to do for me.

I love you.

iii

Acknowledgements

I wish to give thanks first to the most important person in my life, my

wife. Justine has continually made sacrifices in her life to help me achieve my

career goals. In rough times she reminds me why I am in research, because I love

it. I am thankful for all she is, and all she helps me be.

I came to Drexel University for one reason, to work for Dr. Tony Lowman.

I am blessed to have both a role model and a good friend in my advisor. He has

given me opportunities that I never dreamed of. If I could achieve half of his 5

year success in my career, I will feel accomplished.

The in vivo section of this work would not be possible without the work

and guidance of Dr. Jeff Joseph and Dr. Marc Torjman. Special thanks go to the

professors who helped this research by allowing me to take up space in their lab,

Dr. Wheatley and Dr. Laurencin. And thanks to Dr. Abrams for his advice in the

computer simulations, and Dr. Dan whose career guidance helped me obtain a

post doctoral position.

I would also like to thank those who have supported me through their

friendship. The list is long but not exhaustive, which is only a testament to the

iv

quality of students at Drexel. Nikhil Dhoot, his wife Arti, and lovely daughter,

Napul, Dalia El Sherif, Bob Murray (even if he is at UVA now), Jon Thomas,

Xinyin Liu, Arvind Sivasubramanian, Greg Troup, Meredith Hans, Koji

Nakumara, Ravi Gudetti, and Pinar Ozkan. Without their input, both scientific

and supportive, I would still be floundering in the lab.

I would like to give special recognition to the Whitaker foundation for

providing financial support of this work.

v

Table of Contents

LIST OF TABLES.............................................................................................................. x

LIST OF FIGURES........................................................................................................... xi

ABSTRACT .................................................................................................................... xix

1. INTRODUCTION ................................................................................................ 1

2. BACKGROUND................................................................................................... 3 2.1 Controlled Drug Release......................................................................... 3

2.2 Implantable Controlled Drug Delivery ................................................ 5

2.3 Implant-Body Chemical Communication Through Diffusion .......... 7

2.4 Tissue Engineering................................................................................... 8

2.4.1 Blood Vessel Formation............................................................. 9

2.4.2 Angiogenesis ............................................................................. 11

2.4.3 Tissue-Implant Interactions .................................................... 14

2.5 Hydrogels................................................................................................ 19

2.5.1 Poly (2-hydroxyethyl methacrylate) ...................................... 21

2.5.2 Controlling Macroporous Structure of PHEMA Hydrogels.................................................................................... 21

2.5.3 Poly (ethylene glycol)............................................................... 26

2.5.4 PEG-grafted PHEMA Sponges as an Implant Material ...... 26

vi

2.6 Modeling of Angiogenesis.................................................................... 27

2.6.1 Continuous models of Angiogenesis..................................... 28

2.6.2 Cellular Automata.................................................................... 30

List of References ............................................................................................... 40

3. RESEARCH GOALS .......................................................................................... 50

4. SYNTHESIS AND CHARACTERIZATION OF PHEMA SCAFFOLDS ....................................................................................................... 52 4.1 Introduction ............................................................................................ 52

4.2 Experimental Section............................................................................. 53

4.2.1 Macroporous Hydrogel Synthesis ......................................... 53

4.2.2 PEGylation of PHEMA Sponges ............................................ 55

4.2.3 FTIR Spectroscopy.................................................................... 56

4.2.4 Pore Morphology Determination........................................... 56

4.2.5 Mechanical Analysis ................................................................ 58

4.3 Results and Discussion.......................................................................... 64

4.3.1 FTIR Analysis ............................................................................ 64

4.3.2 Pore Morphology Characterization ....................................... 68

4.3.3 Mechanical Analysis ................................................................ 75

4.4 Conclusions........................................................................................... 100

List of References ............................................................................................. 101

vii

5. IN VITRO VASCULARIZATION .................................................................. 104 5.1 Introduction .......................................................................................... 104

5.1.1 Cellular Techniques Used in Angiogenesis Research ....... 105

5.1.2 Biomaterial-Endothelial Cell Interaction Experiments ..... 106

5.2 Materials and Methods ....................................................................... 107

5.2.1 Cell Handling and Storage.................................................... 107

5.2.2 Cryogenic Freezing of Endothelial Cells............................. 109

5.2.3 In vitro Biomaterial Vascularization Studies....................... 109

5.2.4 Matrigel® Impregnated Sponges ......................................... 110

5.2.5 Sample Fixation and Sectioning ........................................... 110

5.2.6 Immunoflourescent Microscopy .......................................... 111

5.3 Results and Discussion........................................................................ 112

5.3.1 Positive Endothelial Tubule Formation Control................ 113

5.3.2 Analysis of Fluorescently-Labeled HMVEC Seeded Networks ................................................................................... 114

5.3.3 Matrigel® Loaded Polymer Samples................................... 120

5.4 Conclusions........................................................................................... 145


6. COMPUTER SIMULATIONS OF POROUS MATERIALS VASCULARIZATION..................................................................................... 150 6.1 Introduction .......................................................................................... 150

viii

6.1.1 Computer Simulations........................................................... 150

6.1.2 Random-Walk Model............................................................. 151

6.1.3 Angiogenesis Modeling......................................................... 152

6.1.4 Model Objectives .................................................................... 153

6.2 Simulation Methods ............................................................................ 154

6.2.1 Porous Polymer Network Formation .................................. 154

6.2.2 Porous Polymer Network Analysis ..................................... 155

6.2.3 Vessel Growth Simulations................................................... 155

6.2.4 Simulation Data Analysis...................................................... 156


6.3.1 Polymer Analysis.................................................................... 157

6.3.2 Vessel Growth Simulations................................................... 158

6.4 Conclusions........................................................................................... 179


7. IN VIVO IMPLANTABLE INSULIN DELIVERY ........................................ 182

7.1 Introduction .......................................................................................... 182

7.2 Materials and Methods ....................................................................... 184

7.2.1 Catheter Assembly ................................................................. 184

7.2.2 In Vivo Experiments................................................................ 184


ix

7.3.1 In Vivo Insulin Infusion Kinetics .......................................... 187

7.3.2 Histological Evaluation of Catheter Sponge Explants ...... 187

7.4 Conclusions........................................................................................... 193


8. RECOMMENDATIONS.................................................................................. 197

8.1 Network Synthesis. .............................................................................. 197

8.2 Protein Functionalization of Sponge Pore Surface.......................... 198 8.3 In Vitro Growth Factor Selection........................................................ 200

VITA ............................................................................................................................. 202

x

List of Tables

5.1 List of Supplements added to the EGM-2-MV Media .................................... 108

xi

List of Figures

2.1 Repeat dosing of a typical (blue) oral drug delivery scheme. Shaded area is equal to total drug delivered to patient. (red) Controlled release can deliver therapeutic levels of drugs for longer times longer with less drug......................................................................................... 33

2.2 An active delivery system would be able to dynamically control the amount of insulin delivered based upon demand........................................ 34

2.3 Schematic representation of the growth of a capillary during angiogenesis........................................................................................................ 35

2.4 The endothelial cell response to VEGF and ANG1/ANG2 during vasculogenesis and angiogenesis. .................................................................. 36

2.5 Classic foreign body response typically ends with the surrounding of an implant with a dense fibrous layer called the fibrous capsule. ......... 37

2.6 Summation of vascularized tissue response to implants with varying pore sizes. ............................................................................................................ 38

2.7 Schematic representation of macroporous PHEMA hydrogel sponges. Interstitial spaces between polymer droplets create a macroporous structure 1-20 µm in size, whereas the polymer network creates a 1-100nm mesh size in the polymer phase....................... 39

4.1 Representation of the two pore sizes present in the PHEMA

sponges. The networks possess the characteristic swollen mesh size of hydrogels and cellularly invasive macropores. ........................................ 60

4.2 Structures of monomers used in polymerization reactions. ........................ 61

4.3 Isocyanate linkage with the pendant hydroxyl group of PHEMA. This results in a urethane linkage of PEG to PHEMA. The remaining isocyanate group can be hydrolyzed in aqueous medium under acid or basic conditions or be used to immobilize protein and peptide sequences. ............................................................................................. 62

4.4 PTFE reaction mold for Implant studies and porosimetry data. ................ 63

xii

4.5 FTIR spectrograms of PHEMA and PEG monomethyl ether. The key absorbencies are the ester linkage absorbance of PHEMA at 1730 cm-1, and PEG’s aliphatic ether absorbance at 1110 cm-1. .......................... 79

4.6 Peak height ratio of PEG:PHEMA as a function of PEG mole fraction. Calibration is based upon blend of the two polymers and not of physically linked copolymers. .............................................................. 80

4.7 Varying blends of PEG and PHEMA to determine if FTIR can be used to calculate the relative concentration of PEG and PHEMA.............. 81

4.8 Subtractions of sponges from PEG-isocyanate reaction with 90 vol% water PHEMA sponges with 0.3% dibutyltin dilaurate in THF at 50ºC. This was the only PEG reaction that exhibited a moderate amount of PEGylation. ...................................................................................... 82

4.9 Subtraction result of (Blue) PHEMA with PEG minus PHEMA. (Red) FTIR of PEG monomethyl ether (350 MW).......................................... 83

4.10 Comparison of compression corrected and uncorrected porosimetry data. As shown there is negligible difference in the cumulative mercury intrusion volume of the uncorrected sample ( ) and corrected ( ). The only visible difference occurs at the larger pore sizes in the incremental intrusion volume for the uncorrected(-) and corrected (- -) data. ............................................................................................. 84

4.11 A porosimetry plot of the unsonicated 85 vol% diluted PHEMA sponge. The average pore diameters calculated are presented to demonstrate the relationship between the porosimetry data and the statistics that are calculated. ............................................................................. 85

4.12 Micrograph of 30vol% PHEMA polymer surfaces. Reduced pore sizes were evident in both (a) PTFE molds and (b) glass molds................. 86

4.13 Surface pore structure of PHEMA sponges reacted in a PTFE mold with sonication. .................................................................................................. 87

4.14 Volume average pore size as a function of reaction mixture dilution. PHEMA ( ) with and ( ) without sonication. (n=4 ± SE) ......................... 88

4.15 Porosity as a function of reaction mixture dilution. PHEMA ( ) with and ( ) without sonication. (n=4 ± SE)................................................. 89

xiii

4.16 Pore size dispersity index as a function of reaction mixture dilution. PHEMA ( ) with and ( ) without sonication. (n=4 ± SE) .......................... 90

4.17 Volume average pore size as a function of reaction mixture dilution. PHEMA ( ) with PEG grafts and ( ) without PEG grafts. (n=4 ± SE) 91

4.18 Porosity as a function of reaction mixture dilution. PHEMA ( ) with PEG grafts and ( ) without PEG grafts. (n=4 ± SE)............................ 92

4.19 Pore size dispersity as a function of reaction mixture dilution. PHEMA ( )with PEG grafts and ( ) without PEG grafts. (n=4 ± SE) ...... 93

4.20 SEM Micrographs of (a) PEG-grafted PHEMA and (b) pure PHEMA polymer sponges using 70 vol% water. .......................................................... 94



4.23 Stress-strain response of pure PHEMA sponges. Each curve is labeled based upon the vol% of water in the reaction solution. As the dilution increased, initial modulus decreased. ....................................... 97

4.24 Stress-strain response of 6.5 mol% PEG-grafted PHEMA sponges. Each curve is labeled based upon the vol% of water in the reaction solution. As the dilution increased, initial modulus decreased. ................ 98

4.25 Initial compressive modulus of ( ) pure and ( ) PEG-grafted PHEMA sponges as a function of solvent volume fraction in the reaction mixture. ................................................................................................ 99

5.1 Scale bar taken at 250X magnification and 1712X1368 resolution.

Under these settings, 175 pixels was equivalent to 100µm........................ 122

5.2 Matrigel® Positive control reference. Tubule formation was evident after 1 day. Tubule legths and diameter s varied greatly. Scale bar is equal to 100µm. ............................................................................................ 123

xiv

5.3 This sample was not fully covered by the Matrigel® basement. This resulted in a hybrid expression; confluent EC that turns into tubules at the Matrigel® TCPS interface. ................................................................... 124

5.4 Scale bar taken with a fluorescent microscope at 100X. ............................. 125

5.5 Negative control staining of 90vol% (a) PEG-grafted and (b) pure PHEMA -hydrogel sponges. From this result, it can be assumed that all brightly fluorescing structures are positively stained HMVEC. ......... 126

5.6 Surface of 60 vol% PEG-grafted PHEMA sponges. Bright spots represent endothelial cells. The lack of cell spreading and tube formation is indicative of pore sizes too small for penetration as well as a surface with no adhesive properties...................................................... 127

5.7 Surface adhesion of HMVEC-ad onto 60vol% pure PHEMA hydrogels. After 2 weeks culture, cells were spread onto the surface into elongated structures. These structures are more similar to the attachment of EC onto TCPS than the tubule formation............................ 128

5.8 Cross-section of 60 vol% pure PHEMA hydrogel. As shown, no endothelial cells penetrated into the small pores of these networks. However, endothelial cells were evident on the surface of the polymer sponge. In this photo, cells have detached from the surface in a thread shape. It is not clear whether these cells are in tubule formation, or a slice of a confluent layer. ..................................................... 129

5.9 Surface image of 100X 70 vol% PEG grafted PHEMA network. Many of the MVEC present possess a slightly diffused glow. This is due their penetration into the samples. In the top left corner. There is some evidence of surface adhesion, but this was minimal compared to the sample penetration. ........................................................... 130

5.10 Cross-section of 70% PEG grafted PHEMA networks (200X). There is extensive evidence of tubule formation and EC elongation. The sizes of the tubules are smaller than the Matrigel® control, due to size limitations within the polymer network............................................... 131

xv

5.11 70vol% PEG grafted PHEMA sponges at 100X. Another image depicting longer tubules. Bottom edge of the sponge was the surface that was seeded. The greater density of tubules near the outer rim of the sponge is most likely due to greater nutrient exchange with the media. ............................................................................... 132

5.12 Surface image of 70 vol% pure PHEMA network. No endothelialization was evident on this sample. .......................................... 133

5.13 Cross sections of 70vol% pure PHEMA. In these networks vascularization of the surface was evident. Little to no penetration was evident in these networks due to the small pore size and low porosity.............................................................................................................. 134

5.14 Surface of 80 vol% PEG-grafted PHEMA sponges. Some HMVECs are evident on the surface. There was not extensive evidence from this analysis of HMVEC attachment and penetration. ............................... 135

5.15 Surface staining of 80 vol% pure PHEMA sponges. Extensive endothelialization is evident. There is also evidence of HMVEC penetration from this analysis as well. No information about tubule formation was obtained. ................................................................................. 136

5.16 Cross sectional view of 80 vol% PEG-grafted PHEMA networks. Due to the interconnected structure of these polymers, there was an abundance of tubule formation. The greater porosity of these samples also allowed for greater nutrient transfer, which helped increase cellular density. ................................................................................. 137

5.17 Cross section of 80 vol% pure PHEMA sponge. Penetration of HMVECs was superficial; only 100-200µm deep. In this layer, the HMVEC density was extremely high, and fluorescence was too great to determine any tubule formation. These cross sections also revealed a large pore size disparity that was not evident in porosimetry and SEM...................................................................................... 138

5.18 Surface image of 90vol% PEG-grafted PHEMA Network. HMVEC were not spread onto the surface, but had penetrated into the polymer network.............................................................................................. 139

xvi

5.19 Surface image of 90 vol% pure PHEMA Network. HMVEC were not spread onto the surface, but had penetrated into the polymer network.............................................................................................................. 140

5.20 90 vol% PEG grafted PHEMA cross section. The irregular shape is a result of sectioning errors. 90 vol% samples contained a high degree of vascularization, and morphologically resembled the Matrigel® control. ............................................................................................................... 141

5.21 Cross section of 90 vol% pure PHEMA cross section. Tubule formation is abundant. Tubules formed along the pore surfaces............ 142

5.22 Matrigel® coated 60 vol% PEG-grafted PHEMA sponges. The HMVEC layer had detached from the edge of the section. This is probably due to the reduced adhesion of protein layer and PEG surface................................................................................................................ 143

5.23 Penetration of HMVEC tubule into Matrigel® loaded 80vol% PEG grafted polymer networks. This image depicts the polymer’s ability for vascularization. (250X Magnification) .................................................... 144

6.1 Pore size vs. porosity for simulated polymer networks. Average

pore size was large due to the high variability of pore sizes present. ( ) 1unit, ( ) 2 units, ( ) 4 units, and ( ) 8 units. .................................... 164

6.2 Histogram of Polymer Gap Size for 50% porosity 1 unit pore size polymer network.............................................................................................. 165

6.3 Histogram of Polymer Gap Size for 50% porosity, 3 unit pore size polymer network.............................................................................................. 166





xvii



6.10 “Parking Lot” plot of 5000 random points obtained by the RAND function. There is some evidence of local trends which is a result of the non-randomness of the generator. However, these orientations were not global through the domain, and considered not significant for the purposes of this study. ....................................................................... 173

6.11 The rate of change (slope) of the square mean displacement as a function of porosity at long time steps for pore sizes ( ) 1, ( ) 3, ( ) 5, and ( ) 9. No surface gap was present during these simulations. Line represents the rate of change of the moving particle in unhindered conditions. All error bars and line thickness represent 99.99% confidence limits. ................................................................................ 174

6.12 The rate of change (slope) of the square mean displacement as a function of porosity at long time steps for pore sizes ( ) 1, ( ) 3, ( ) 5, and ( ) 9. A surface gap was present during these simulations. Line represents the rate of change of the moving particle in unhindered conditions. All error bars and line thickness represent 99.99% confidence limits. ................................................................................ 175

6.13 Number of free moving particles vs. time step for all simulations performed. The important point to note is that the majority of the simulations possessed a linear rate of entrapment. The deviation of the skewed lines is thought to be a result of the rate being too rapid for the number of simulations performed to adequately represent. ........ 176

6.14 The rate of entrapment as a function of porosity for pore sizes ( ) 1, ( ) 3, ( ) 5, and ( ) 9. No surface gap was present during these simulations. The solid represents the rate of entrapment with no polymer present. .............................................................................................. 177

xviii

6.15 The rate of entrapment as a function of porosity for pore sizes ( ) 1, ( ) 3, ( ) 5, and ( ) 9. A surface gap was present during these simulations. The solid and dashed lines represent the rate of entrapment with no polymer present and a solid polymer with no porosity, respectively. ..................................................................................... 178

7.1 Depiction of perforated catheter tubing inserted axially into the hydrogel sponge. A silicone adhesive was used to permanently fix tubing assembly. .............................................................................................. 186

7.2 Systemic glucose response following infusion of human insulin from an external pump 5 months post implantation. ....................................... 189

7.3 Systemic human insulin concentration following infusion of human insulin from an external pump 5 months post implantation. ................ 190

7.4 Histological slides of mesenteric implant, (a) 100X, (b) 200X.................... 191

7.5 Histological slides of subcutaneous implant, (c) 100X, (d) 200X. ............. 192

xix

Abstract Macroporous Hydrogels as Vascularizable Soft Tissue – Implant Interfaces: Materials Characterization, In Vitro Evaluation, Computer Simulations, and

Applications in Implantable Drug Delivery Devices Thomas D. Dziubla

Anthony M. Lowman, Ph.D.

Implantable medical devices, such as biosensors and implantable drug delivery systems, function

optimally when rapid solute exchange can occur between implant and surrounding tissue.

However, almost all materials implanted into the body are encapsulated in a fibrous layer that

prevents this rapid communication. Macroporous materials are known to change this response

by allowing vascularized tissue ingrowth, however many questions still exist as to the role

material properties play. In this work, macroporous hydrogels are presented as an ideal interface

between implant and tissue due to there mechanical properties which are similar to soft tissue.

These materials were synthesized with varying degrees of porosity, pore size, and surface

hydrophilicity. It was found from that when the hydrogel’s pore sizes were 10 µm or larger, they

became highly vascularized in vitro, regardless of surface hydrophilicity. This response was

different from previous literature where larger pores sizes (~60 µm) were necessary. It was

thought that the lack of a secondary infiltrating cell (macrophages) during the in vitro studies was

the cause for this discrepancy. Computer simulations verified the in vitro results presented.

From in vivo studies, this high degree of vascularity was found to not only lengthen the life span

of an implanted drug delivery device, but also improve the associated uptake response.

1

CHAPTER 1: INTRODUCTION

There have been many exciting advances made in the field of medical

implants. Concepts such as biosensors and implantable controlled drug

delivery have great promise, but cannot be realized without a clear

understanding and control of the biological response. The first implants

ever created were bone and joint replacements [1]. Under optimal conditions,

there would be minimal scar tissue surrounding these structures. And

since these devices were predominantly physical/mechanical in function, the scar

tissue never posed a significant problem. This scar tissue and implant-body

interaction are respectively called the fibrous capsule and the foreign body

response, and were once considered the mark of a biocompatible material [1].

However, this is no longer acceptable for the newer, more sophisticated implant

designs. While fibrous encapsulation mattered little with the physical

devices, this process disables biosensors and drug delivery devices after a few

weeks or months by acting as a barrier which greatly impedes electrical and

chemical transmission [2-4].

As a way of controlling the foreign body response, it may be possible to

specially design materials as tissue-implant interfaces. These materials would

ideally allow for a permanent, highly vascular tissue to surround the implant.

2

This highly vascular tissue would allow for the rapid exchange of chemical

signals, such as drugs and nutrients. To develop this interface, a detailed

understanding of both the biology of the tissue response and blood vessel

formation is required.

3

CHAPTER 2: BACKGROUND

2.1 Controlled Drug Release The goal of controlled drug delivery is to provide a specified drug

concentration within the body for an extended period of time [5-7]. A device that

provides a sustained release of drug can maintain desired drug concentrations in

the blood with reduced number of doses, while also minimizing the concern of

undesirable, sometimes toxic, side effects. Controlled release is, also, a more

cost-effective way of delivering expensive medications. For example, Figure 2.1

depicts the systemic drug concentration of typical repeated doses as a function of

time. The amount of total drug delivered is equal to the gray shaded region on

the graph. A significant portion of the time, drug is systemically present but not

in therapeutic amounts. The red line shown depicts an ideal drug concentration

profile, where the same amount of drug is delivered but in therapeutic

concentrations for a longer time. With less drug wasted, costs can be reduced.

The first design concepts for controlled release were passive delivery

systems. In passive delivery, unassisted diffusion of solvent and solute is the

only means of modulating the rate of drug delivery. Typically, there is a depot

of drug contained within a polymer matrix which releases over time. A

4

convenient way to evaluate the release profile from these passive systems is by

the following power law

ntM ktM∞

= (2.1)

where Mt is the amount of drug release at a specific time, M the total

amount of drug released at infinite time, and k and n are both weighting

constants that best fit experimental data. While this equation is inherently curve

fitting, there is a theoretical basis for its existence. A solution to Fick’s second

law on a slab with diffusion across both edges results in the following short time

approximation [8],

12tM 4 Dt

M∞

= δ π (2.2)

which is analogous to equation (2.1) with n = ½. When n is equal to 1,

this is known as Case II transport. Continuous release occurs with a time-

independent delivery scheme, most commonly called zero-order release kinetics.

However, this is just a subset of the actual goal of controlled release. The

primary aim of controlled drug delivery is complete optimization therapeutic

delivery; that is the ability to deliver to the desired location, a precise dose for a

finite period of time [9-11]. With this ideal system, one could achieve high

bioavailability with minimal side effects and drug exposure. To achieve this

5

idealization, systems must be responsive to fluctuations in the patient’s needs.

The advantage to implantable drug delivery devices is that they can be designed

to meet these aims by providing a means of continually monitored and

administered drug delivery.

2.2 Implantable Controlled Drug Delivery

Even when continual monitoring is not needed, there are some instances

where just sustained release is not the ideal delivery mechanism. For example

with a gonadotropin-releasing hormone, it has been shown that a pulsatile

delivery scheme is most effective at stimulating the pituitary gland [12, 13]. For

such a demand, a passive diffusion controlled drug delivery device is not the

best alternative. For this reason, active systems that can allow for servo or

responsive delivery schemes have been an area of increasing interest in drug

delivery. The advantages of active delivery systems can be seen in Figure 2.2.

As shown, drug is only delivered at times of need, and is turned off instantly

when the demand has been met. One type of active system that is currently

being developed is the drug array implant [14]. This device is a silicon chip with

many tiny reservoirs filled with drug or a microporous membrane where the

drug is held. In one system, the reservoirs are coated in a thin nonporous metal

layer. When voltage is applied, the metal layer breaks and delivers its reservoir

contents [15, 16]. This design holds great promise, as it is capable of rapid on/off

6

delivery. Also, the reservoirs can be filled with many different types of drugs,

allowing for complex drug delivery regimes. Sustained release can be achieved

through the sequential rupturing of wells containing the same drug.

Another type of active delivery device is the drug delivery micro pump.

Currently, some diabetic patients use an external pump connected to a

subcutaneous catheter [17, 18]. The pump is set to deliver basal levels of insulin,

and can give bolus injections to meet demands during meal times. With

advances in electronic miniaturization, these pumps are continually being made

smaller and more reliable allowing them to be implantable.

One design concept that uses the implantable pump is the artificial

pancreas. This device can be broken down into three components; the glucose

sensor which monitors blood-glucose levels, the control mechanism which

determines rates of delivery based upon the physiological data obtained from the

control mechanism, and the delivery pump and catheter which is the active

system that delivers the insulin into the body. Determined through several

clinical trials, the most common cause of device failure was due to tissue

inclusion at the catheter port of delivery caused by the foreign body response [2,

3, 17-21]. When catheters were flushed with saline solution to remove blockages,

30% still occluded after 1 year. This number increased to 50% after 2 years, and

70% after 3. For the artificial pancreas to be functional, a material needs to be

7

designed that can prevent this encapsulation. A layer that would allow vascular

tissue ingrowth rather than fibrotic tissue inclusion would be solution to this

problem.

2.3 Implant-Body Chemical Communication through Diffusion

For biological systems, chemical communication is the exchange of solutes

between cells, tissues, organs, and implanted devices. These solutes can either be

nutrients/waste for cellular metabolism or chemical signals that elicit a specific

biological response, such as drugs and hormones. In biological systems, there is

some point at which the process is diffusional. Hence, an understanding of the

native diffusion barriers that are found in localized tissue is required to

understand what variables are important in the control of the transport rate.

To describe the diffusion of a solute to the circulatory system, it is

beneficial to divide the process into two parts, diffusion in the bulk tissue and

diffusion through the vessel wall [4, 22]. Tissue diffusion is usually modeled as

the diffusion of a porous media. The density of the extra-cellular matrix (ECM)

proteins, cellular bodies and their orientation regulates the diffusivity. These

bodies can act in two main ways. First, they can take-up the diffusing solute,

either degrading it or imparting their own diffusional limitations which will

result in decreased release. Or, these cells act to block diffusion and increase the

8

path tortuosity. As a result, the diffusivity of the tissue decreases as the tissue

proteins and cell bodies become more tightly packed.

Once a solute reaches the blood vessel, the transport into the blood stream

is dictated by the permeability of the vessel wall[23-25]. Primarily the total

surface area of the vessels within the tissue and the permeability of the vessel

wall regulate this transport. The total surface area is a function of the diameter

and the density of the vessels within the tissue. Vessel permeability is dynamic

and determined by the balance of signaling proteins in the vicinity. For instance,

an increase in vascular endothelial growth factor (VEGF) has shown to increase

permeability while an increase in Angiostatin-1 (ANG-1) decreases vessel

permeability [26].

Based upon this description of solute transport from implant to

circulatory system, a loose connective tissue with high vascularity and vessel

permeability would provide the fastest route for systemic delivery. It may be

possible to remodel the tissue surrounding the implant by applying tissue

engineering techniques. This work may have implications which can extend to

key difficulties being faced in tissue engineering, as discussed in the next section.

2.4 Tissue Engineering

The goal of tissue engineering is to repair an existing tissue/organ or

completely regenerate a tissue/organ that has failed to function [2]. In order to

9

achieve this, there are two main strategies currently being pursued. One method

is the in vitro regeneration of a tissue/organ from primary cells obtained by the

patient, and the subsequent reimplantation of the newly generated tissue [27, 28].

The other technique is to implant a device that would temporarily provide or

assist the functions of the organ/tissue being replaced, while simultaneously

allowing the in situ formation of a new organ/tissue [29]. Both of these strategies

require a biomaterial scaffold, which organizes the growth of cells into the

proper configuration to form the desired tissue [30].

Both of these strategies have been limited by the depth of cellular

penetration into the porous networks. It is believed that this limitation is directly

related to the depth of penetration of the vascular which penetrates the

scaffolding [31]. Without capillaries being fully extended throughout the

scaffold, deeper cells will not be able to achieve the required nutrient/waste

exchange rates. In order to specifically select vessel growth, an understanding

the physiological pathways of capillary growth is needed. In the next section, an

overview of the two interrelated methods of vessel formation is provided.

2.4.1 Blood Vessel Formation

Blood vessel formation is usually considered to progress through two

distinct yet related processes; vasculogenesis and angiogenesis. Angiogenesis is

the formation of new blood vessels by the growth of “sprouts” from existing

10

vasculature (See Figure 2.3) [32]. This self-limiting process is seen in

reproduction, wound repair, and placental development.

Vasculogenesis is the developmental formation of vasculature. There

are still many holes in the knowledge base of this process, but the general

behavior of the development is understood. Vasculogenesis occurs through a

sequence of time dependant events, where each link in the chain must occur for

the formation of a healthy, functional vasculature. The following sequence

represents the currently understood steps in vasculogenesis [33, 34].

First a signal, VEGF, is released into the embryonic environment. This

signal will target the VEGF receptor, VEGF-R1 located on angioprogenitor and

endothelial cells. When this receptor is activated, the progenitor cells will

differentiate into endothelial cells, and the resulting cells will proliferate. When

VEGF activate the VEGF-R2 receptor, the endothelial cells will start to organize

into tube-like vascular structures. At this point, these tubule structures lack the

secondary support cells, pericytes and smooth muscle cells. These vessel

structures also lack branching networks, and the organization of larger to smaller

vessels characteristic of a mature circulatory system. This mature formation is

dependant upon the signals of ANG1 and ANG2. These receptors target the TIE2

receptor. ANG1 signals the formation of the branching structures, and allows

the endothelial cells to recruit the pericytes to form a mature vessel. ANG2 is

11

almost completely analogous in structure to ANG1, but when bound to TIE2

receptor, it elicits no signal cascade. As such, it is a competitive inhibitor to

ANG1, and is believed to be instrumental in the vasculature’s ability to remodel

itself. There exists one more known receptor, TIE1, which plays an important

role in vasculogenesis, TIE1. While its complement signal is still not known, it is

known through TIE1 gene knockout studies, that the receptor/signal function to

control fluid exchange across the capillary walls, and plays a part in modulating

homodynamic stress resistance.

2.4.2 Angiogenesis

Many of the factors involved in vasculogenesis still play a crucial role in

angiogenesis [34]. There are a host of signals/factors that seem to initiate the

angiogenic response, however not all of these signal cascades are understood [33,

35, 36]. It is believed that VEGF and ANG1/ANG2 play a part in most cases of

vascular remodeling, and is depicted in Figure 2.4 [34]. A start signal is released

into the ECM when an area in the body needs to remodel its vasculature. This

need can arise in situations such as wound healing, hypoxic tissue, or a tumor-

induced event. This start signal is either VEGF or ANG2 directly, or signals that

induce the release of VEGF/ANG2 [37]. When ANG2 hits the TIE2 receptor, it

inhibits ANG1 ability to maintain vessel integrity. Hence, the vessel becomes

12

locally unstable. The basement membrane surrounding the blood vessel is

digested, the pericytes recede, and if no other signal is present the local

endothelial cells will undergo apoptosis. This is believed to be the way the body

will digest unneeded vasculature [34]. However, if VEGF is present during this

time, the endothelial cells will start migrating chemotaxicly toward increasing

VEGF. These leading cells, do not usually proliferate, rather the endothelial cells

that follow will divide and align along the space created by the leading cells to

form a lumen. The cells form tube like structures, which resemble budding

blood vessels. These sprouts, the budding vessels, continue to grow until they

reach another sprout, and the link to form a functioning capillary. This linking

behavior is termed anastamosis. Over time as the ANG2 signal is diminished,

the greater concentration of ANG1 allows for the reactivation of the TIE2

receptor, which allows the endothelial cells to call for the support of the pericytes

to stabilize these newly formed vessels. It is believed that it is this continual

balance of signals, which controls the maintenance, and remodeling of adult

vasculature.

2.4.2.1 Effects of Extracellular Matrix Ligands in Angiogenesis

While not discussed in most descriptions of angiogenesis, adhesion

proteins play a crucial role in the formation of new blood vessels. The reason for

this omission is due to the extensive availability of adhesion proteins in normal

13

extra cellular matrix. The basement membrane that surrounds blood vessels is

comprised primarily of collagen IV and laminin. There have been many studies

that evaluate the in vitro and in vivo ability of the endothelial cells to form tubules

in and on different membrane proteins. The results of these studies were highly

dependant upon variables such as cell type, whether it was a 2D or 3D matrix

study, or if the studies were handled in vivo. For example, Dvorak demonstrated

that collagen I implanted subcutaneous did not induce vascularization, while

Hoying et al. showed that the vascular fragments seeded onto collagen I matrices

provided vascular growth in 1 week. In spite of these irregularities, one general

trend observed is tubule formation occurred most rapidly when in the presence

of collagen IV and laminin. It is believed that observed complex behavior is a

result of cross-talk that exists between adhesion integrins and growth factor

receptors expressed on the endothelial cell surface. Integrins are cell receptor

proteins comprised of two subunits, alpha and beta. There are currently 8

known adhesion integrins that are expressed on most endothelial cells,

α1β1, α2β1, α3β1, α5β1, α6β1, αvβ3, and αvβ5. It was found that in in vitro settings,

α2β1 interaction was crucial in the tubule formation in collagen matrices, where

as the αvβ3, and α5β1 integrins were necessary in fibrin matrices. Moreover in

studies where αvβ3 was ligated, migration on fibronectin (a process mediated

by α5β1) was inhibited. The converse effect was also true. Further evidence of

14

cross-talk exists in the work of Friedlander, who demonstrated that when

αvβ3 was blocked, fibroblast growth factor induced angiogenesis was inhibited,

but not VEGF angiogenesis. Where as when αvβ5 was blocked, the reverse was

true.

2.4.3 Tissue-Implant Interactions 2.4.3.1 Classic Foreign Body Response

Implants are foreign bodies that will invoke the natural defense mechanism

against such intrusions; the inflammatory response. This process is outlined in

Figure 2.5. Typically the inflammatory response is split into two categories,

acute and chronic inflammation [38, 39]. During the acute phase, an influx of

fluid, plasma proteins, and neutrophils enter the wound/implant site [40]. These

neutrophils accumulate at the site of implantation and start to phagocytize any

small debris/bacteria that are present. Phagocytosis is activated when the

neutrophils comes into contact with activating factors called opsonins [38]. If an

implant surface absorbs opsonins, such as the antibody immunoglobulin G (IgG),

the neutrophil will try to engulf the implant. But since there is a large size

disparity between the implant and neutrophils, phagocytosis cannot occur. This

leads to an event known as frustrated phagocytosis, where the neutrophils dump

the contents of lysosomes into the ECM [41]. This process is highly unfavorable

since it is very irritating to the surrounding tissue and leads to chronic

15

inflammation. After the neutrophils have entered the area and cleared away any

debris, granulation tissue (highly vascularized tissue) begins to form, and the

natural wound healing response continues. At this point the response can split

into either a chronic inflammatory response or a foreign body reaction of the

acute type [39]. If there is a constant chemical or physical irritation (as in free

movement of the implant), the chronic inflammatory response will occur [42]. If

there are no negative chemical or physical signals then classic foreign body

response occurs. Typically, the foreign body response results in 3 characteristic

layers [39]. A primary layer of macrophages and/or foreign body giant cell

formations surrounds the implant. These cells secrete the second layer

composed of dense fibrous tissue 30-100 µm in thickness. A third layer of

granulation tissue surrounds this fibrous wall. This response is indefinitely

stable except for a decrease in cellularity of the primary layer. The dense nature

of the fibrous layer greatly impedes the diffusion of most chemical species, as a

result prevents any implanted drug delivery device from functioning effectively

[43].

2.4.3.2 Tissue Response to Porous Materials

The tissue response changes greatly when the implanted material has a

porous morphology. Brauker et al. published a paper demonstrating the ability

of porous materials to remodel the tissue response. They subcutaneously

16

implanted several hydrophobic materials (PTFE, cellulose acetate, cellulose

esters, and acrylic copolymers) with pore sizes ranging from 0. 2 -15 µm. It was

found that materials with pores greater 5µ were surrounded by highly vascular

loose connective tissue. When the pore sizes further increased, evidence of

vascular penetration was evident. This is the result depicted in Figure 2.6. The

astounding part of there study was that this vasculature persisted for the entire

duration of the study, 1 year. Shwarkawy et al. studied acetylized PVA with

pore sizes 5, 60, and 200 µm in size [4, 23, 25]. Their 5-micron pore size

corroborated the results obtained by Brauker et al. However, they noted a very

high degree of vascularization of implants with the 60 µm pore size, and when

this pore size increased beyond 100 µm, the vascularity of the materials actually

decreased.

Shwarkawy also demonstrated that changes in pore size not only effected

vascular density but also the response to systemic uptake of drug through a

vascularized implant. It was demonstrated that the 60 µm pore material

delivered the drug in almost half the time it took for a subcutaneous injection to

be taken up systemically. This is due to the increased vascular density as well as

increased vascular permeability at these pore sizes [4, 23, 25].

There are two main theories that have been proposed to describe the

dependence of vascular penetration on implant pore size. Padera and Colton

17

have suggested that it is the macrophages degree of attachment onto the material

surface that dictates the signals that they send out [44]. When the macrophages

are able to spread onto the surface of the material, they release signals that call

for the deposition of the tight collagen layer. When these macrophages penetrate

into a porous sample, and cannot spread fully on the surface, this signal is not

released or released to a reduced extent. However, due to the macrophages

being further from a nutrient source, they release signals that initiate

angiogenesis. When the macrophages penetrate into the very large pores, they

are able to once again release the collagen deposition signals, and the pores

become filled with the avascular collagen layer that typically surrounds a non-

porous implant.

Rosengren has suggested that it may be implant mobility that controls

the degree of implant vascularity [45]. They suggest that smooth implants are

capable of high relative motion. This motion shears the adjacent cells inducing

necrosis. The degree of necrosis is the cause of the severity of the inflammatory

response, hence the thickness of the fibrous capsule. They further suggest that

porous materials possess little to no fibrous capsule, because the tissue that

penetrates works to stabilize the relative motion. While it is still not known

whether or not these hypotheses are correct or to what degree they are

18

important, it is evident that simple morphological changes have a great effect

upon the vascularization of implants.

2.4.3.3 Chemical vs. Physical Effects

Many of the porous implant studies compared the results of materials

with varying surface chemistries. These studies looked at materials of varied

hydrophilicity, such as hydrophobic PTFE, and acetylized PVA, to the more

hydrophilic cellulose esters and acetates and poly(vinyl alcohol)s [4, 23, 25, 46-

49]. It was found that the ingrowth of vascularized and loose connective tissue

was dictated primarily by the pore size rather than chemical properties of the

material. However, it would be wrong to assume that no control could be

obtained through modifications of the implant surface chemistry.

Endothelial cells interact with the ECM through adhesion moieties called

integrins [50]. It is believed that cells attach onto synthetic materials through

intermediary proteins, such as fibrin, which absorb onto polymer surfaces.

Hence, by changing the protein absorption properties of surfaces, it is possible to

alter the adhesion of endothelial cells. Moreover, it is also possible to bind

specific adhesion ligands onto surfaces for a more direct control of the cellular

attachment [51, 52]. Endothelial cells are able to adhere to the common

attachment sequences that are found on fibrin, such as RGD and YISGR. It was

found, however, that another adhesion peptide sequence, the RDEV ligand,

19

preferentially bound endothelial cells over fibroblasts, smooth muscle cells, or

activated platelets [51]. Through this ligand, it may be possible to explicitly

control the formation of capillaries into the implant.

Tube formation of the endothelial cells is an essential characteristic for the

formation of capillaries, and is controlled by both chemical and physical

properties of the material. There has been a significant lack of in vitro research

showing the effects of synthetic biomaterials on endothelial cell’s ability for tube

formation [46]. One study coated fibronectin in 10 and 30µm stripes. They noted

that tube formation occurred on the 10 µm stripes but not the 30. This study

demonstrates the general trend of tube formation that the more adherent the cells

are to a surface, the more they spread and are less likely to express tube

formation. Also, that cells with greater spreading (attachment) exhibited

increased proliferation, yet a decrease in cellular mobility. Moreover, tube

formation was most prominent in surfaces that exhibited moderate adhesive

characteristics [53]. There is also evidence that material stiffness also plays a part

on tube formation. Ingber et al. showed that softer, more malleable materials

exhibited an increase in cell tube formation [54].

2.5 Hydrogels

Hydrogels are three-dimensional, water-swollen structures composed of

mainly hydrophilic homopolymers or copolymers [55, 56]. They are rendered

20

insoluble due to the presence of chemical or physical crosslinks. The physical

crosslinks can be entanglements, crystallites or weak associations such as van der

Waals forces or hydrogen bonds. The crosslinks provide the network structure

and physical integrity. It is possible to design hydrogels with swelling behavior

and mechanical modulus that is dependent on the external environmental

factors.

Hydrogels are classified in a number of ways. [55, 58] They can be neutral

or ionic based on the nature of the side groups. They can also be classified based

on the network morphology as amorphous, semi crystalline, hydrogen-bonded

structures, supermolecular structures and hydrocolloidal aggregates.

Additionally in terms of their network structures, hydrogels can be classified as

macroporous, microporous, or nonporous. [55, 56, 59] Since the early 1960s,

hydrogels have been considered for use in a wide range of applications. Most

notably these materials are considered ideal for biomedical and pharmaceutical

devices, mainly due to their high water content and rubbery nature which

resembles natural living soft tissue more than any other class of synthetic

biomaterials [55, 56, 59]. Furthermore, the high water content allows these

materials to exhibit excellent biocompatibility. Since softer materials seem better

suited to supporting endothelial tube formation, it is believed that hydrogels will

make excellent candidates for vascularizable implant materials [60].

21

2.5.1 Poly (2-hydroxyethyl methacrylate)

One of the first hydrogels studied for biomedical applications was poly (2-

hydroxyethyl methacrylate) (PHEMA) [55, 61]. It is a non-ionic hydrogel, and as

such exhibits no pH swelling dependence. It was used as one of the first soft

contact lenses. Unlike other hydrogels, the monomer is infinitely soluble in

water while the polymer exhibits a limited solubility. This phase behavior allows

for the formation of a macroporous sponge structure when reacted in dilute

monomer solutions. In the late 1960s, these porous forms of PHEMA were

studied for the potential applications of soft tissue replacement, such as breast

augmentation and nasal cartilage replacement [61-63]. However, complications

with long-term calcification hindered further development. Then in the 1980s,

work was done with pancreatic islet sequestering using PHEMA sponges [64,

65]. While the hydrogels sponge performed well as an immunoisolation device,

long-term viability of the islets was not achieved.

2.5.2 Controlling Macroporous Structure of PHEMA Hydrogels PHEMA hydrogel sponge formation is controlled by the thermodynamic

phase behavior between the polymer-rich phase, and the aqueous-rich phase

during polymerization. Chirila noted that the formation of the porous structure

is dependant upon a kinetic competition between gel point and phase separation

[66]. If gelation occurs first, the resulting material is a hydrogel with little to no

22

macropores, but will still contain the typical hydrogel mesh size on the angstrom

level. If phase separation occurs first, the resulting material contains water filled

spaces that can vary in size from sub micron up to 20 microns in size. The

presence of the two different pore sizes present in macroporous PHEMA sponges

is schematically shown in Figure 2.7. Since the sponge formation is dependant

upon both polymerization kinetics and solution thermodynamics, there are many

variables that can be altered in order to control the pore morphology of the

resulting hydrogel sponge. The following is a selection of methods that can be

used to tailor PHEMA porous networks.

2.5.2.1 Water Content

The amount of water added to the reaction mixture produces the most

dramatic effect upon the size of the pores in a PHEMA sponge. When the water

content is below 45-50%, the PHEMA polymer chains remain soluble and do not

form a 2 phase system. When the reaction solution’s water content is increased,

phase separation occurs with excess water acting as the pore forming agent.

Hence, as we further increase the water content, the number of water molecules

excluded from the polymer phase increases creating larger voids between the

polymer droplets. It is well established that networks containing 85% water or

greater possess pore sizes that are large enough for cellular invasion.

23

Unfortunately, these high water solutions result in materials with

characteristically weak mechanical properties and large pore size distributions.

2.5.2.2 Crosslinker, Crosslinking Density and Comonomers

Since different crosslinking agents possess different solubilities in water, it

was hypothesized that by altering the crosslinking agent used it should be

possible to alter the networks pore morphology. Chirlia et al. performed a rather

extensive evaluation of crosslinkers to determine there relative impact upon the

networks ability to form large macropores [67-69]. They determined that using

typical concentrations of crosslinker content (0.1-2 mol %) had very little effect of

the ultimate morphology and mechanical strength of the networks formed.

While many studies on crosslinker selection have been performed, little

work has been done on the effect of more/less hydrophilic comonomers on the

formation of the macropores. The comonomers that have been attempted were

more hydrophobic monomers such as methyl methacrylate [70]. This is most

likely due to the commonly used hydrophilic comonomers, acrylic acid and 2,2-

diethylaminomethacrylate result in transparent gels.

2.5.2.3 Nonreactive Components

The presence of non-reacting, inert, components can also affect the pore

size of the resulting polymer sponge. One of the first methods pursued was that

of porogens. A porogen is a space filling particulate that prevents

24

polymerization in specific locations through physical hindrances [71]. Sucrose,

glucose, and ice crystals have all been used as void fillers to create macroporous

PHEMA hydrogels [72, 73]. The porogen must be selected based on its ability to

remain suspended in the reaction mixture, and provide some mechanism of

being leached from the next work after the sponge is formed [74].

Another technique is to control the solubility of PHEMA by addition of a

tertiary component. For example, PHEMA solubility decreases with an increase

in ion content. As a result, Mikos et al. used salt solutions of varying ionic

strength to dilute the reaction mixtures [75]. It was noted that increasing the ion

content of the aqueous solution to 0.7 molar, interconnected macropores were

obtained at 60 vol% water. Surfactants may also be used to control the network

pore structure. However, not much work has been done in this area, since

surfactants typically work to reduce the surface repulsions between the two

phases and form smaller droplets. These smaller droplets when gelled are

expected to possess a smaller pore size. However this is still a promising area of

exploration, since it may be possible to form alternate phase structures such as

bicontinous phases, which would be ideal for cellular invasion.

2.5.2.4 Temperature Effects Isotactic PHEMA was found to possess negative temperature dependence

in water [76]. While atactic PHEMA is not expected to have as strong of a

25

negative temperature dependence, the mechanisms of this behavior can still exist

over short ranges and may effect the phase behavior. As such, increased

temperatures may also function to control the pore morphology by allowing the

polymer to phase separate sooner in the reaction.

Temperature not only plays a critical role with the thermodynamics, but

also with the kinetics of the polymerization. Once phase separation occurs, the

polymer phase will start to settle out of solution since it is denser than the

aqueous phase. Chirlia noted this phenomenon by stating that in some reactions,

a water layer was evident over the polymer sponge layer [77]. Temperature can

reduce this settle out by speeding up the reaction kinetics, and forcing gelation to

occur sooner.

2.5.2.5 Mechanical Effects Since two phases are present, mechanical agitation can be used to control

the distribution of the phases. Dalton synthesized porous tubes of PHEMA

hydrogels by reacting the monomer solution in a radially rotating glass tube [70].

It was found that this rotation resulted in a dense outer layer of polymer (due to

centripetal force) and a more porous inner surface. Minor evidence of pore

organization under this radial agitation was noticed when HEMA was

copolymerized with PEG.

26

2.5.3 Poly (ethylene glycol)

Poly (ethylene glycol) (PEG) is another hydrophilic polymer that has

important biological properties. There has been an extensive amount of research

performed showing that the presence of a PEG layer on any surface (from metal

to plastic to ceramic) will greatly reduce the adsorption of protein and cellular

adhesion onto that surface [78-82]. Hence, a PEG-ylated layer prevents

interactions between the surface and protein. This has been thought to be a

result of the rapid molecular mobility of the highly hydrophilic PEG chains in the

presence of water, and the ability of these chains to exclude solutes [83]. These

grafts are moving so rapidly, that they do not allow the protein enough time to

interact with the ether groups. Also, PEG chains tend to interact with themselves

in such a way that any molecule other than water will be forced out of their

domain. This is known as the “salting out” effect [83]. Since proteins do not

adhere to PEG-grafted layers and cellular adhesion to materials is controlled

though protein-ligand interactions, PEG layers may also reduce cellular

adhesion.

2.5.4 PEG-grafted PHEMA Sponges as an Implant Material

From our current understanding of tissue-implant interactions, it should

be possible to tailor the vascularization of materials by changing the surface

hydrophilicity, pore size, and mechanical properties. Adding and controlling the

27

amount of PEG grafts in a PHEMA sponge can vary the degree of hydrophilicity

of the sponge surface. Moreover, PEG grafts make an excellent tether for the

attachment of proteins and peptide sequences [84]. The potential benefit of a

PEG-grafted PHEMA system is derived from the application of reduced

nonspecific adhesion and the conjugation of specific ligands. First, the

hydrophilicity of the material might suppress general cellular adhesion and

tissue ingrowth. Then, through the presence of cell specific adhesion ligands,

desired the cell lines could be selected for directed ingrowth [60].

2.6 Modeling of Angiogenesis

Angiogenesis is an orchestration of complex pro and anti angiogenic

regulators, growth kinetics, and adhesion proteins [32, 36, 37]. Events at the

molecular, cellular, and tissue level all play a part into the final structure of the

newly formed vasculature. For this reason, it is difficult to obtain a full

understanding of this process through experiment alone. Mathematical

modeling of angiogenesis can provide some useful insights into the viability of

vessel growth theories and what factors are most likely dominant. The

angiogenesis models that have been proposed can be grouped into two main

classes of models, continuous models and cellular automata.

28

2.6.1 Continuous Models of Angiogenesis In continuous models, contributing factors are expressed explicitly in a

series of non-linear PDEs in order to describe the movement and growth of

endothelial cells. One of the first descriptions of this type for angiogenesis was

by Edelstein where filament and sprout tip densities were described as

continuum variables [85]. Terms were also included to allow for branching,

anastamosis, and death. Baldwin and McElawin adopted this approach to look

at tumor induced angiogenesis [86]. This time, sprout tips chemotaxicly moved

toward a tumor angiogenesis factor (TAF). In this model, TAF consumption by

the migrating endothelial cells was ignored. The most recent model is that of

Anderson et al [87, 88].

2n D n ( (c)n c) ( n f )t

∂= ∇ − ∇ ⋅ χ ∇ − ∇ ⋅ ρ ∇

∂ (2.3)

f f (1 f )n nft

∂= β − − γ

∂ (2.4)

c nct

∂= −η

∂ (2.5)

0(c)1 ac

χχ =

+ (2.6)

29

In this series of equations, n is the endothelial cell density, D the

endothelial cell diffusivity, χ the chemotaxic function, c the TAF concentration, ρ

the hepatotaxic constant, and f the adhesion protein density. Β, γ, and η are

positive, scaled parameters. Equation (2.3) describes the change in endothelial

cell density by typical Fickian diffusion, chemotaxic directed and hepatotaxic

directed motion. Hepataxisis is the tendency of endothelial cells to move in the

direction of increasing adhesion protein concentrations. Equation (2.4) accounts

for the endothelial cells tendency to remodel the ECM by simultaneously

digesting and secreting adhesion proteins. Also, equation (2.5) is used to

describe growth factor consumption by endothelial cells. As most cells,

endothelial cells are limited in their sensitivity to growth factor concentrations.

Any additional amount of growth factor beyond a certain value will have no

increasing affect on the chemotaxis of the migrating cells. Equation (2.6)

mathematically describes this limiting behavior. This model is currently the

most extensive in its attempt to include many different aspects of angiogenesis.

This extensive nature leads to the inclusion of many curve fitting parameters that

bring into question the validity of the model

There are some problems inherent in using continuous models to describe

angiogenesis. Since endothelial cells are discrete entities, the use of continuum

variables to describe endothelial cells is highly suspect. The definition of the

30

derivative does not apply. Moreover, due to the non-linear nature of these

models, explicit solutions are difficult to obtain and finite element method or

other numerical solution techniques must be employed. Finally, continuous

models are only able to provide statistical trends in cell migration and growth

factor concentrations [89]. These models are not able to explicitly demonstrate

the growth of vascular networks.

2.6.2 Cellular Automata

Cellular automata, while not an explicit model, can reproduce many

complex phenomena shown by the use of simple rules. Cellular automata,

originally created by Von Neumann, are a grid of many cells that can possess

discrete values dictated by simple rules [90, 91]. With each time step, the state of

every cell is calculated, and the time course of development can be plotted. One

of the first and probably most popular cellular automata was developed by John

Conway, and is most commonly called “Conway’s Game of Life” [92]. In this

automaton, a cell is either alive or dead. If there are two or three live cells near a

neighboring a cell, then that cell stays alive, otherwise that cell becomes dead. If

three live cells surround a dead cell, then the dead cell becomes live. When

these rules are repeated over many iterations, complex patterns emerge that

resemble patterns of growth and migration seen in nature. By altering the rules

31

that control the automaton, it may be possible to elucidate the underling factors

that are involved in many biological processes.

Cellular automata can be divided into three main categories; eulerian,

lattice gases, and solidification models [89]. In eulerian models, every cell can

possess many discrete states, and the state of each cell is dependant upon its

previous state and the state of the neighboring cells. This is the type that was

evident in the game of life model. In lattice gases, solid particles move around

and interact with other particles. In this class, turbulent behavior of gases in

complex geometries have been described where more through Navie-Stokes

evaluations would have been time-limiting. Finally, solidification models are

used to describe events such as crystallization. Moving particles can be

irreversibly bound to a lattice point, or cells undergo irreversible changes.

Markus et al used this final class of models to describe vessel morphogenesis as a

sequential series of irreversible steps [89].

Cellular automata have been applied to simulate the formation of vessel

structures in angiogenesis [89, 93]. The rules governing these simulations have

been based on both geometric and biological mechanisms. For example, due to

the similarities between fractal structures and vessel networks, some groups

have based their vessel growth on events such as crystallizations [94]. Other

groups have confined the growth of vessel to the migration of the vessel tip

32

(since the forming blood vessel is dependant upon this leading cell) [88]. These

models use the descretized PDEs to describe probability fields for the

neighboring cells of sprout tips. The models work off an Eulerian based cellular

automata. At every time point, the change of each cell’s sprout tip density is

calculated. This change is used to create an array of probabilities that dictate

which simulation cell space the sprout tip will move to next(or if it will stay

stationary). Then a random number is generated, and the sprout tip moves

accordingly. While this method is highly dependant upon the scaled values

assumed by the PDE equations, and the time steps selected, these models are

capable of recreating the vessel growth, branching and brush tip disorganization

of vessels that is commonly seen in tumor-induced angiogenesis.

33

Figure 2.1 Repeat dosing of a typical (blue) oral drug delivery scheme. Shaded area is

equal to total drug delivered to patient. (red) Controlled release can deliver therapeutic levels of drugs for longer times longer with less drug.

34

Time

Dru

g R

elea

se

Basal levels

small meal small meal

large meal large meal

Figure 2.2 An active delivery system would be able to dynamically control the amount of

insulin delivered based upon demand.

35

Figure 2.3 Schematic representation of the growth of a capillary during angiogenesis.

36

Figure 2.4 The endothelial cell response to VEGF and ANG1/ANG2 during vasculogenesis and angiogenesis. This figure is reproduced from reference [34].

37

Fibrous Encapsulation

Capillaries

Dense Fibrous Capsule

Foreign Body Giant Cell

Macrophage

1. Implantation

Neutrophils enter clean loose debris

Plasma proteins, fluid enter area

Granulation Tissue Forms:Highly vascularmacrophages

Figure 2.5 Classic foreign body response typically ends with the surrounding of an implant with a dense fibrous layer called the fibrous capsule.

38

Figure 2.6 Summation of vascularized tissue response to implants with varying pore

sizes.

39

Figure 2.7 Schematic representation of macroporous PHEMA hydrogel sponges. Interstitial spaces between polymer droplets create a macroporous structure 1-20 µm in size, whereas the polymer network creates a 1-100nm mesh size in the polymer phase.

Mc, ξ

Mc, ξ

Mc, ξ

M c , ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

1-100 nm

1-20 µm

40

List of References [1] Park, J.B., Biomaterials: An Introduction. 1979, New York: Plenum Press. [2] Scavini, M., Galli, L., Reich, S., Eaton, R.P., Charles, M.A., and Dunn, F.L.

ʺCatheter Survival During Long-Term Insulin Therapy with an Implanted Programmable Pump.ʺ Diaetes Care, 1997, 20: p. 610-613.

[3] Selam, J.L., Micossi, P., Dunn, F.L., and Nathan, D.M. ʺClinical Trial of

Programmable Implantable Insulin Pump for Type I Diabetesʺ. Diabetes Care, 1992, 15: p. 877-885.

[4] Shwarkawy, A.A., Klitzman, B., Truskey, G.A., and Reichert, W.M.

ʺEngineering the Tissue Which Encapsulates Subcutaneous Implants I. Diffusion Propertiesʺ. J Biomed Mater Res, 1997, 37: p. 401-12.

[5] Fung, L.K. and Saltzman, W.M. ʺPolymeric Implants for Cancer

Chemotherapyʺ. Advanced Drug Delivery Reviews, 1997, 26: p. 209-230. [6] Siepmann, J. and Goepferich, A. ʺMathmatical Modeling of Bioerodible,

Polymeric Drug Delivery Systemsʺ. Advanced Drug Delivery Reviews, 2001, 48: p. 229-247.

[7] Dash, A.K. and Cudworth II, G.C. ʺTherapeudic Applications of

Implantable Drug Delivery Systemsʺ. J Pharm Tox Meth, 1998, 40: p. 1-12. [8] Crank, J. and Park, G.S., Diffusion in Polymers. 1968, London, New York,:

Academic Press. xii, 452. [9] Hertzog, B.A., Thanos, C., Sandor, M., Raman, V., and Edelman, E.R.,

Cardiovascular Drug Delivery Systems, in Encyclopedia of Controlled Drug Delivery, Mathoiwitz, E, Editor. 1999, John Wiley and Sons, Inc.: New York.

[10] Leach, K., Cancer, Drug Delivery to Treat - Local & Systemic, in

Encylcopedia of Controlled Drug Delivery, Mathoiwitz, E, Editor. 1999, John Wiley and Sons, Inc.: New York.

41

[11] Groves, M.J., Parenternal Drug Delivery Systems, in Encylcopdia of Controlled Drug Delivery, Mathoiwitz, E, Editor. 1999, John Wiley and Sons, Inc.: New York.

[12] Breimer, D.D. ʺPharmacokinetic and Pharmacodynamic Basis for Peptide

Drug Delivery System Designʺ. J Control Release, 1992, 21: p. 5-10. [13] Konobil, E. ʺThe Neuroendocrine Control of the Menstrual Cycleʺ. Recent

Progr Horm Res, 1980, 36: p. 53-88. [14] Low, L., Seetharaman, S., He, K., and Madou, M.J. ʺMicroactuators toward

Microvalue for Responsive Controlled Drug Deliveryʺ. Sensors and Actuators B, 2000, 67: p. 149-160.

[15] Tierney, M.J. and Martin, C.R. ʺNew Electrorelease Systems Based Upon

Microporous Membranesʺ. J Electrochem Soc, 1990, 137(12): p. 3789-3793. [16] Tierney, M.J. and Martin, C.R. ʺElectroreleaseing Composite Membranes

for Delivery of Insulin and Other Macromoleculesʺ. J Electrochem Soc, 1990, 137: p. 2005-2006.

[17] Saudek, C.D. ʺFuture Developments in Insulin Delivery Systemsʺ. Diabetes

Care, 1993, 16(Duppl 3): p. 122-132. [18] Hepp, K.D. ʺImplantable Insulin Pumps and Metabolic Controlʺ.

Diabetologia, 1994, 37(Suppl 2): p. S108-S111. [19] Saudek, C.D. ʺNovel Forms of Insulin Deliveryʺ. J Clin Endocrinol Metab

North Amer, 1997, 26: p. 599-610. [20] Santiago, J.V., White, N.H. and Skor, D.A., Mechanical Devices for Insulin

Delivery, in Recent Advanaces in Diabets, Nattrass, M and Santiago, JV, Editors. 1984, Churchhill Livingstone: Edinburgh. p. 145-63.

[21] Knatterud, G. and Fisher, M. ʺReport from the International Study Group

on Implanable Insulin Delivery Decivesʺ. ASAIO Transactions, 1988, 34(2): p. 148-149.

42

[22] Lieb, W.R. and Stein, W.D. ʺImplications of Two Different Types of Diffusion for Biological Membranesʺ. Nature New Biol, 1971, 234: p. 220-222.


ʺEngineering the Tissue Which Encapsulates Subcutaneous Implants Ii. Plasma-Tissue Exchange Propertiesʺ. J Biomed Mater Res, 1998, 40: p. 586-597.

[24] Curry, F.E., Huxley, V.H. and Adamson, R.H. ʺPermeability of Single

Cpillaries to Intermediate-Sized Colored Solutesʺ. Am J Physiol, 1983, 245: p. H495-H505.


ʺEngineering the Tissue Which Encapsulates Subcutaneous Implants Iii. Effective Tissue Response Timesʺ. J Biomed Mater Res, 1998, 40: p. 598-605.

[26] Thurston, G. ʺDistinct Actions of Vegf and Angiopoietin-1 on Blood Vessel

Permeabilityʺ. Annals of Biomedical Engineering, 2000, 28: p. S72-73. [27] Murphy, W.L., Kohn, D.H. and Mooney, D.J. ʺGrowth of Continuous

Bonelike Mineral within Porous Poly(Lactide-Co-Glycolide) Scaffolds in Vitroʺ. J Biomed Mater Res, 2000, 50(1): p. 50-58.

[28] Wang, Y., Yang, Z., Xie, H., and Li, S. ʺStress-Strain Properties of the

Tissue-Engineered Cartilageʺ. Journal of Biomedical Engineering, 2001, 285(1-3): p. 116-22.

[29] Thomson, R.C., Mikos, A.G., Beahm, E., Lemon, J.C., and Satterfield, W.C.

ʺGuided Tissue Fabrication from Periosteum Using Preformed Biodegradable Polymer Scaffoldsʺ. Biomaterials, 1999, 20(21): p. 2007-18.

[30] Freyman, T.M., Yannas, I.V. and Gibson, L.J. ʺCellular Materials as Porous

Scaffolds for Tissue Engineeringʺ. Progress in Materials Science, 2001, 46(3-4): p. 273-82.

[31] Griffith, L.G. and NAughton, G. ʺTissue Engineering-Current Challenges

and Expanding Opportunitiesʺ. Science, 2002, 295: p. 1009-14.

43

[32] Bertolini, F., Mancuso, P., Gobbi, A., and Pruneri, G. ʺThe Thin Red Line: Angiogenesis in Normal and Malignant Hematopoiesisʺ. Experimental Hematology, 2000, 28: p. 993-1000.

[33] Rossant, J. and Howard, L. ʺSignaling Pathways in Vascular

Developmentʺ. Ann Rev Cell Dev Biol, 2002, 18: p. 541-73. [34] Hanahan, D. ʺSignaling Vascular Morphogenesis and Maintenanceʺ.

Sience:Cell Biology, 1997, 277(5322): p. 48-50. [35] Nguyen, M., Akrell, J. and Jackson, C.J. ʺHuman Endothelial Gelatinases

and Angiogenesisʺ. Inter J Biochem cell Bio, 2001, 33: p. 960-970. [36] Gerwins, P., Skoeldenberg, E. and Claesson-Welsh, L. ʺFunction of

Fibroblast Growth Factors and Vascular Endothelial Growth Factors and Thier Receptors in Angiogenesisʺ. Critical Reviews in Oncology/hematology, 2000, 34: p. 185-194.

[37] Daniel, T.O. and Abrahamson, D. ʺEndothelial Signal Integration in

Vascular Assemblyʺ. Annu Rev Physiol, 2000, 62: p. 649-71. [38] Anderson, J.M. ʺMechanisms of Inflammation and Infection with

Implanted Devicesʺ. Cardiovasc Pathol, 1993, 2(3): p. 33-41. [39] Anderson, J.M. ʺInflammatory Responce to Implantsʺ. Trans Am Soc Artif

Intern Organs, 1988, 19: p. 101-107. [40] Malech, H. and Gallin, J. ʺCurrent Concepts: Immunology. Neutrophiles

in Human Diseases.ʺ N Engl J Med, 1987, 317: p. 687-694. [41] Henson, P. ʺMechanisms of Exocytosis in Phagcytic Inflammatory Cellsʺ.

AM J Pathol, 1980, 101: p. 494-511. [42] Gallin, J., Goldstein, I.M. and Snyderman, R., Inflammation: Basic Principles

and Clinical Correlates. 1988, New York: Raven Press. [43] Scharp, D.W., Mason, N.S. and Sparks, R.E. ʺIslet Immuno-Isolation: The

Use of Hybrid Artificial Organs to Prevent Islet Tissue Rejectionʺ. World J. Surg., 1984, 8: p. 221-229.

44

[44] Padera, R.F. and Colton, C.K. ʺTime Course of Membrane Microarchitecture-Driven Neovascularizationʺ. Biomaterials, 1996, 17: p. 277-284.

[45] Rosengren, A., Danielsen, N. and Bjursten, L.M. ʺReaective Capsule

Formation around Soft-Tissue Implants Is Related to Cell Necrosisʺ. J Bio Mat Res, 1999, 46(4): p. 458-464.

[46] Sieminski, A.L. and Gooch, K.J. ʺBiomaterial-Microvasculature

Interactionsʺ. Biomaterials, 2000, 21: p. 2233-2241. [47] Brauker, J.H., Carr-Brendel, V.E., Martinson, L.A., Crudele, J., and

Johnston, W.D. ʺNeovascularization of Synthetic Membranes Directed by Membrane Microarchitectureʺ. J Bio Mat Res, 1995, 29: p. 1517-1524.

[48] Lipsky, M.H., P ʺEstablishment of a Neovascular Bed in a Collagen-

Impregnated Polyurathane Spongeʺ. J Bio Mat Res, 1989, 23: p. 1441-1452. [49] Wake, M.C., Mikos, A.G., Sarakinos, G., Vacanti, J.P., and Langer, R.

ʺDynamics of Fibrovascular Tissue Ingrowth in Hydrogel Foamsʺ. Cell Transplantation, 1995, 4(3): p. 275-279.

[50] Saltzman, W.M., Cell Interactions with Polymers, in Principles of Tissue

Engineering, Lanza, R, Editor. 1997, RG Landes Company: Austin, TX. p. 228-46.

[51] Hubbel, J. ʺSurface Grafted Cell-Binding Peptides in Tissue Engineering of

the Vascular Graftʺ. Ann NY Acad Sci, 1992, 665: p. 253-8. [52] Cook, A. ʺCharacterization and Development of Rgd-Peptide-Modified

Plga as an Interactive, Resorbable Biomaterialʺ. J Biomed Mater Res, 1997, 35: p. 513-23.

[53] Matsuda, T. and Kurumantani, H. ʺSurface Induced in Vitro Angiogenesis:

Surface Property Is a Determinant of Angiogenesisʺ. ASAIO Trans, 1990, 36(M565-8).

45

[54] Ingber, D.E. and Folkman, J. ʺMechanochemical Switching between Growth and Differentiation During Fibroblast Growth Factor-Stimulated Angiogenesis in Vitro: Role of Extra Cellular Matrixʺ. J Cell Bio, 1989, 109: p. 317-330.

[55] Peppas, N., Hydrogels in Medicine and Pharmacy, Vol. I: Fundamentals. 1986,

Boca Raton: CRS Press. 180. [56] Lowman, A.M. and Peppas, N.A., Hydrogels, in Encyclopedia of Controlled

Drug Delivery, Mathoiwitz, E, Editor. 1999, John Wiley and Sons, Inc.: New York.

[57] Kost, J. and Langer, R. ʺResponsive Polymeric Delivery Systemsʺ.

Advanced Drug Delivery Reviews, 2001, 46: p. 125-148. [58] Langer, R. and Peppas, N.A. ʺChemical and Physical Structure of

Polymers as Carriers for Controlled Release of Bioactive Agets: A Reviewʺ. Rev Macromol Chem Phys, 1983, C23: p. 61-126.

[59] Langer, R. and Peppas, N.A. ʺChemical and Physical Structure of

Polymers as Carriers for Controlled Release of Bioactive Agents: A Reviewʺ. J Macromol Sci, Rev Macromol Chem Phys, 1983, C23(1): p. 61-126.

[60] Mann, B.K., Gobin, A.S., Tsai, A.T., Schmedlen, R.H., and West, J.L.

ʺSmooth Muscle Cell Growth in Photopolymerized Hydrogels with Cell Adhesive and Proteolytically Degradable Domains: Synthetic Ecm Analogs for Tissue Engineeringʺ. Biomaterials, 2001, 22(22): p. 3045-51.

[61] Kliment, K., Stol, M., Fahoun, K., and Stockar, B. ʺUse of Spongy Hydron

in Plastic Surgeryʺ. J Bio Mat Res, 1968, 2: p. 237-243. [62] Voldrich, Z., Tomanek, Z., Vacik, J., and Kopecek, J. ʺLong Term

Experience with Poly(Glycol Monomethacrylate) Gel in Plastic Operations of the Noseʺ. J Bio Mat Res, 1975, 9: p. 675-685.

[63] Simpson, B.J. ʺHydron: A Hydrophilic Polymerʺ. Biomed Eng, 1969, 4: p.

65-68.

46

[64] Klomp, G.F., Hashiguchi, H., ursell, P.C., Takeda, Y., Taguchi, T., and Dobelle, W.H. ʺMacroporous Hydrogel Membranes for a Hybrid Artificial Pancreas. Ii. Biocompatibilityʺ. J Bio Mat Res, 1983, 17: p. 865-871.

[65] Ronel, S.H., DʹAndrea, M.J., Hashiguchi, H., Klomp, G.F., and Dobelle,

W.H. ʺMacroporous Hydrogel Membranes for a Hybrid Artificial Pancreas. I. Synthesis and Chamber Fabricationʺ. J Bio Mat Res, 1983, 17: p. 855-864.

[66] Chirila, T.V., Higgins, B. and Dalton, P.D. ʺEffect of Synthesis Conditions

on the Properties of Poly(2-Hydroxyethyl Methacrylate) Spongesʺ. Cellular Polymers, 1998, 17(3): p. 141-162.

[67] Lou, X., Dalton, P.D. and Chirila, T.V. ʺHydrophilic Sponges Based on 2-

Hydroxy Ethyl Methacrylate. Part Vii: Modululation of Sponge Characteristics by Changes in Reactivity and Hydrophilicity of Crosslinking Agentsʺ. J Mat Sci Mat Med, 2000, 11(5): p. 319-325.

[68] Clayton, A.B., Chirila, T.V. and Dalton, P.D. ʺHydrophilic Sponges Based

on 2-Hydroxyethyl Methacrylate. Iii. Effect of Incorporating a Hydrophilic Crosslinking Agent on the Equilibrium Water Content and Pore Structureʺ. Polym Int, 1997, 42(1): p. 45-56.

[69] Clayton, A.B., Chirila, T.V. and Lou, X. ʺHydrophilic Sponges Based on 2-

Hydroxyethyl Methacrylate. V. Effect of Crosslinking Agent Reactivity on Mechanical Propertiesʺ. Polym Int, 1997, 44: p. 201-207.

[70] Dalton, P.D., Flynn, L. and Shoichet, M.S. ʺManufacture of Poly(2-

Hydroxyethyl Methacrylate-Co-Methyl Methacrylate) Hydrogel Tubes for Use as Nerve Uidance Channelsʺ. Biomaterials, 2002, 23(18): p. 3843-3851.

[71] Badiger, M.V., McNeill, M.E. and Graham, N.B. ʺPorogens in the

Preperation of Microporous Hydrogels Based on Poly(Ethylene Oxides)ʺ. Biomaterials, 1993, 1993(14): p. 14.

[72] Oxley, H., Corkhill, P.H., Fitton, J.H., and Tighe, B.J. ʺMacroporous

Hydrogels for Biomedical Applications:Methodology and Morphologyʺ. Biomaterials, 1993, 14: p. 1064-72.

47

[73] Kang, H.W., Tabata, Y. and Ikada, Y. ʺFabrication of Porous Gelatin Scaffolds for Tissue Engineeringʺ. Biomaterials, 1999, 20(14): p. 1339-1344.

[74] Carenza, M. and Veronese, F.M. ʺEntrapment of Biomolecules into

Hydrogels Obtained by Radiation-Induced Polymerizationʺ. J Control Rel, 1994, 29: p. 187-93.

[75] Liu, Q., Hedberg, E.L., Liu, Z., Bahulekar, R., Meszlenyi, R.K., and Mikos,

A.G. ʺPreperation of Macroporous Poly(2-Hydroxyethyl Methacrylate) Hydrogels by Enhanced Phase Seperationʺ. Biomaterials, 2000, 21(21): p. 2163-2169.

[76] Oh, S.H. and Jhon, M.S. ʺTemperature Dependence of Unperterbed

Dimensions for Isotactic Poly(2-Hydroxyethyl Methacrylate)ʺ. J Polym Sci Polym Chem, 1989, 27: p. 1731-1739.

[77] Chirila, T.V., Constable, I.J., Crawford, G.J., Vijayasekaran, S., Thompson,

D.E., Chen, Y.-C., Fletcher, W.A., and Griffen, B.J. ʺPoly(2-Hydroxyethyl Methacylate) Sponges as Implant Materials: In Vivo and in Vitro Evaluation of Cellular Invasionʺ. Biomaterials, 1993, 14: p. 26-38.

[78] Am Ende, M.T., Diffusion-Controlled Delivery of Proteins from Hydrogels

and Other Hydrophilic Systems, in Protein Delivery: Physical Systems, Sanders and Hendren, Editors. 1997, Plenum Press: New York.

[79] Apicella, A., Cappello, B. and al, e., Poly(Ethylene Oxide)-Based Delivery

Systems, Acs Syposium Series, in Polymeric Drugs and Drug Administration. 1994, ACS: Washington, DC.

[80] Stryer, L., Biochemistry 4th Ed. 1995, New York: W. H. Freeman and

Company. [81] Wang, W. ʺInstability, Stablilization, and Formulation of Liquid Protein

Pharmaceuticalsʺ. Inter J Pharm, 1999, 185: p. 129-188. [82] Peracchia, M.T., Harnisch, S., Pinto-Alphandary, H., Gulik, A., Dedieu,

J.C., Desmaele, D., dʹAngelo, J., Mueller, R.H., and Couvreur, P. ʺVisualization of in Vitro Protein-Rejecting Properties of Pegylated Stealth Polycyanoacrylate Nanoparticlesʺ. Biomaterials, 1999, 20: p. 1269-1275.

48

[83] Harris, J.M., Poly(Ethylene Glycol) Chemistry, Biotechnical and Biomedical Applications. 1992, New York: Plenum Press.

[84] Hermanson, G.T., Bioconjugate Techniques. 1st ed. 1996: Acedemic Press.

785. [85] Edelstein, L. ʺThe Propogation of Fungal Colonies: A Model for Tissue

Growthʺ. J Theor Biol, 1982, 18: p. 679. [86] Balding, D. and McElwain, D.L.S. ʺA Mathmatical Model of Tumor-

Induced Capillary Growthʺ. J Theor Biol, 1985, 114: p. 53. [87] Chaplain, M.A.J. and Anderson, A.R.A. ʺMathmatical Modelling,

Simulation and Prediction of Tumor-Induced Angiogenesisʺ. Invasion Metastasis, 1996, 16: p. 222-234.

[88] Anderson, A.R.A. and Chaplain, M.A.J. ʺContinous and Discrete

Mathmatical Models of Tumor-Induced Angiogenesisʺ. Bull Math Biol, 1998, 60: p. 857-899.

[89] Markus, M., Boehm, D. and Schmick, M. ʺSimulation of Vessel

Morphogenesis Using Cellular Automataʺ. Mathmatical Biosciences, 1999, 156: p. 191-206.

[90] Kier, L.B., Cheng, C.-K. and Testa, B. ʺCellular Automata Models of

Biochemical Phenomenaʺ. Future generation computer systems, 1999, 16(273-289).

[91] Talia, D. and Sloot, P. ʺCellular Automata:Promise and Prospects in

Computational Scienceʺ. Future generation computer systems, 1999, 16: p. v-vii.

[92] Bosch, R.A. ʺInterger Programming and Conwayʹs Game of Lifeʺ. SIAM

Review, 1999, 41(3): p. 594-604. [93] Wolfram, S., A New Kind of Science. 2002: Wolfram Media, Inc.

49

[94] Baish, J.W., Gazit, Y., Berk, D.A., Nozue, M., Baxter, L.T., and Jain, R.K. ʺRole of Tumor Vascular Architecture in Nutrient and Drug Delivery:An Invasion Percolation-Based Network Model.ʺ Microvascular Research, 1996, 51: p. 327-402.

50

CHAPTER 3: RESEARCH GOALS

In this work, it is hypothesized that macroporous hydrogels can be

designed to allow vascular penetration, and that these vascularized materials can

be used to improve the long term functioning of implants such as biosensors and

implantable drug delivery devices. The goals of this research are to synthesize

and characterize macroporous hydrogels by relating porous morphology,

mechanical properties, and swelling properties as a function of synthesis

conditions, and evaluate these networks ability to support a viable

microvasculature using in vitro experimental techniques. Also, computer

simulations are proposed as way of understanding the relationship between

cellular penetration and macroporous properties. Finally, in vivo experiments

are performed to demonstrate the ability of a macroporous hydrogel coating to

improve the long term uptake response of an implantable drug delivery device.

The following lists an outline of the specific aims of this research.

1. Synthesize and characterize macroporous networks of 2-hydroxyethyl

methacrylate with and without poly (ethylene glycol) grafts using free

radical solution polymerizations. Network’s porosity, pore size

distribution, structure, mechanical properties, and swelling behavior

are determined to evaluate material morphology on material

vascularization.

51

2. Develop an in vitro microvascular-biomaterial interaction analysis

technique based upon in vitro angiogenesis research using human

microvascular endothelial cells. Use the in vitro method developed to

determine effects of pore size and surface chemistry upon endothelial

cell penetration, proliferation, and tubule formation.

3. Simulate the ingrowth of endothelial cells using a discrete angiogenic

model and compare these results to the data obtained from the cellular

ingrowth studies and existing literature to determine if simple size

effects can explain vascularization’s dependence on pore size and

porosity.

4. Verify the significance of a vascularized tissue implant interface in

long-term implantable drug delivery devices such as the artificial

pancreas.

52

CHAPTER 4: SYNTHESIS AND CHARACTERIZATION OF PHEMA SCAFFOLDS

4.1 Introduction

PHEMA possesses a limited solubility in water, while HEMA monomer

and water are fully miscible. This results in the formation of 2 phases during

PHEMA synthesis in dilute aqueous systems; an aqueous phase and a polymer

rich phase. The resulting polymer scaffold possesses porous organization at two

distinct order of magnitude, a macroporous and a microporous level (see Figure

4.1). The macroporous network structure is dependent upon both

thermodynamic and kinetic effects. If phase separation occurs before gelation,

then the structure will contain macropores. If gelation occurs first, then the

ensuing polymer is a regular hydrogel containing only micropores [1, 2]. The

phase equilibrium of the reacting system also has a significant effect upon final

porous structure. For example as the phases become increasingly dissimilar, the

resulting pore morphology will be of beaded chains of polymer droplets

suspended in an aqueous medium. This is because the spherical shape results in

the least surface area between the two phases, hence the lowest energy state.

In this work, polymer pore morphology was tailored by the control of

several key reaction variables such as; temperature, mechanical agitation,

comonomer addition, crosslinking density, and aqueous dilution rate. Network

53

and polymer properties are both directly and indirectly measured using Fourier

Transform Infrared Spectroscopy (FTIR), mercury porosimetry, scanning electron

microscopy, and mechanical deformation analysis. The information obtained

from this study will be evaluated based upon the current understanding of

biological responses to materials.

4.2 Experimental Section 4.2.1 Macroporous Hydrogel Synthesis

PHEMA Sponges were prepared via free radical solution polymerization

of 2-hydroxyethyl methacrylate (HEMA, Aldrich Chem. Co., Milwaukee, WI).

HEMA is an ionically neutral hydrophilic methacrylic acid ester. All monomer

structures are shown in Figure 4.2. In order to remove the inhibitor, 4-

methoxyphenol (MEHQ), HEMA was passed through in exchange column

(DEHIBT 200, Polysciences, Inc., Warrington, PA). All other chemicals were

used as received. Tetra (ethylene glycol) dimethacrylate (TEGDMA,

Polysciences, Inc., Warrington, PA) was used as the crosslinking agent. For all

studies, the crosslinking agent was set at 1 mol%. 2,2 azo-bisisobutryonitrile

(AIBN, Aldrich Chem. Co., Milwaukee, WI) was used as the reaction initiator at

elevated temperatures. AIBN breaks down into two radicals under increasing

temperatures. As more radicals are formed, polymerization kinetics is increased.

54

All reactants were mixed including the initiator to alleviate solubility

concerns. This mixture was then diluted to 10-40 vol % with deionized water (18

MΩ, Barnstead E-pure). Nitrogen or argon was bubbled through the reaction

mixture for 10 minutes to remove any dissolved oxygen. Oxygen acts a free

radical scavenger and can result in unwanted peroxide formation within the

polymer backbone. Since removing radicals lowers the reaction rate, the

presence of oxygen during synthesis will result in a more random kinetic profile.

Two types of reaction molds were used for polymer synthesis. A PTFE

reaction mold (Figure 4.4) was selected for the formation of catheter tips for use

in animal studies, in vitro insulin infusion studies, and mercury porosimetry.

For compression studies and in vitro cell culturing, cylindrical glass molds were

used due to the useful geometry of the ensuing polymers provided.

Comparisons between samples prepared using different mold types were made

using scanning electron microscopy (Amray 1830, Amray Bedford, MA).

Synthesis proceeded in a water bath set at 70±3°C for 1 hour. Mechanical

agitation was added to some reactions by means of bath sonication (FS20 Bath

Sonicator 44-48Khz, Fisher Scientific). The resulting hydrogel sponges were then

placed into deionized (DI) water for several days to remove any sol fraction

present.

55

4.2.2 PEGylation of PHEMA Sponges

Two techniques were used to add poly (ethylene glycol) (PEG) grafts on

the surface of the PHEMA sponge. First, PEGylating technique involved the

incorporation of poly (ethylene glycol) 200 monomethyl ether monomethacrylate

(Polysciences, Inc., Warrington, PA) directly in the synthesis step of the PHEMA

sponge. PEGMA is methoxy-terminated methacrylic acid ester with the relative

double bond on one end of the PEG polymer chain. PEGMA was added at 6.5

mol% of total monomer content. Higher contents of PEGMA adversely affected

the phase behavior, and resulted in insolubilities with higher content.

Secondly, PEG-diisocyanate (shearwater chemical, Huntsville, AL) was

used to link PEG onto the pendant hydroxyl group of HEMA. The reaction

mechanism is shown in Figure 4.3. PHEMA sponges were lyophilized then

immersed into either THF containing 0.3% dibutyltin dilaurate or a basic

aqueous solution. PEG diisocyanate and PHEMA were added to the aqueous

solution such that the PEG to PHEMA weight ratio was 2:1. This was to ensure

that the isocyanate was kept in excess to minimize the occurrence of loop

formations on the polymer surface. Reactions were carried out at 25 and 50 ºC

overnight under a nitrogen atmosphere. Following the reaction, the polymer

sponges were rinsed in deionized water for several days to remove unreacted

PEG and residual solvent.

56

4.2.3 FTIR Spectroscopy

FTIR was used to verify the incorporation PEG into the PHEMA samples.

Freeze dried samples were ground into a powder, and mixed with KBr. The KBr

was freeze dried to reduce water contamination in the spectra. The KBr to

polymer ratio was between 1:20 to 1:50. A Nicolet economy KBr sample press

was used to obtain optically clear pellets of KBr and sample. Pellets were

analyzed using transmission FTIR in a Mange IR560 (Nicolet, Madison, WI). Dry

air was used as the chamber purge stream for all analyses. The scanning

resolution was set at 4 nm with a total of 512 scans per sample. The background

was obtained against a pure KBr pellet, and was recalculated every 2 hours.

4.2.4 Pore Morphology Determination

4.2.4.1 Porosimetry

In order to obtain quantifiable information on pore size, pore size

distribution, and porosity of the polymer networks, mercury porosimetry was

used. Samples were flash frozen by immersing into a mixture of dry ice and

acetone and then freeze dried using a Virtis Bench top 3.3XL lyophilizer

(Gardiner, NY). The mercury porosimeter was a Micromeritics Autopore III

(Norcross, GA). The pressure was varied from 0.37 to 50 psi. Each pressure

point was allowed to equilibrate for 10 seconds, the suggested equilibration time

for the Autopore III. At each equilibration point, the amount of mercury that

57

penetrated into the sample was recorded. Pore diameters were calculated from

the Washburn equation:

( )1D 4 cosP

= γ ϕ

(2.1)

Where D is pore diameter, P is the applied pressure, γ is the surface

tension, and ϕ is the contact angle. This equation assumes cylindrical pores, and

that the liquid is nonwetting. The contact angle for mercury and dry PHEMA

was assumed to be 130º, the typical contact angle of mercury on most materials

in air. The surface tension used was that of pure mercury, 485 dynes/cm. Non-

porous PHEMA samples were evaluated to determine compressibility of the

polymer networks. When sponges are allowed to dry at ambient conditions, the

polymer chains are allowed to rearrange and relax. This relaxation results in the

collapsing of the sponge pores. Since the collapsed sponges possess extremely

small pore sizes, they were considered non-porous in the pressure range used for

porosimetry. These non-porous samples were used as reference samples for

measuring the compressibility of the polymer under test conditions. Any

intrusion volume measured under these conditions could be attributed directly

58

to the compressibility of the polymer and not to mercury penetrating the porous

structure.

4.2.4.2 Scanning Electron Microscopy SEM micrographs were obtained for both the internal and external

structure of all sponges that were analyzed. Freeze-dried samples, both freeze-

fractured and non freeze-fractured, were gold sputter coated (Denton Desk 1

Cold Sputter/etch, 35mA, Denton Vacuum, Inc. Cherry Hill, NJ) to obtain a

conductive coating visible by the SEM. All micrographs were taken using an

Amray 1830 scanning electron microscope (Amray, Inc., Bedford, MA). For all

samples, the accelerating potential was set at 20KHz to obtain greater contrast of

the porous surface, and the working distance was set at 25 mm. Most samples

were imaged at 3 different magnifications to obtain information at all orders of

magnitude. While every attempt was made to keep magnifications consistent

throughout all samples, deviations from the set values were necessary with some

samples to improve resolution and focus.

4.2.5 Mechanical Analysis The mechanical strength of the polymer sponges is dependent

upon the synthesis conditions such as comonomer selection and aqueous

dilution rate. An automated materials mechanical tester (Instron 4422, Canton,

MA) equipped with a 50 Lb load cell was used to obtain bulk compressive

59

modulii. Cylindrical samples were cut to maintain a height to diameter ratio of

3:1. The top and bottom edges of the samples were evaluated to insure only right

cylinder geometries were used. Prior to evaluation, all samples were

equilibrated in solutions of phosphate buffered saline (PBS, pH 7.4) at 37ºC.

Each sample was kept in the 37ºC water bath until the compression test was

performed. The strain rate was set at 50% compression/min, and force vs.

crosshead movement was recorded. The studies were conducted in atmosphere

at ambient conditions. Bulk polymer compressive response was also determined

in order to compare the effect of relative porosity on the compressive strength of

samples.

The compressive stress was calculated as the instantaneous force dived by

the initial cross sectional area

o

FA

σ = (2.2)

The strain was calculated as the difference between instantaneous heights to

initial height divided by the initial height.

t

o

hh

ε = (2.3)

Since the hydrogel samples used possess high deformations, it was assumed that

the crosshead travel distance is equal to the amount of deformation of the

sample.

60

Figure 4.1 Representation of the two pore sizes present in the PHEMA sponges. The networks possess the characteristic swollen mesh size of hydrogels and cellularly invasive macropores.

Mc, ξ

Mc, ξ

Mc, ξ

M c , ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

Mc, ξ

macropore

1-100 nm

1-20 µm

61

Figure 4.2 Structures of monomers used in polymerization reactions.

O

C

N

CH2-CH2-OH O C=O

CH2=C CH

(CH2-CH2-O-)4 CH3 OC=O

CH2=CCH

(CH2-CH2-O)4 O C=O

CH2=CCH

OC=O

CH2=CCH

2-Hydroxyethyl methacrylate Poly(ethylene glycol) (n=200) monomethylether monomethacrylate

Tetra(ethylene glycol) dimethacrylate

CH2-CH2-O-(CH2-CH2-O-)n CH2- CH2

PEG MW=3400 Diisocyanate

O

C

N

62

Figure 4.3 Isocyanate linkage with the pendant hydroxyl group of PHEMA. This results in a urethane linkage of PEG to PHEMA. The remaining isocyanate group can be hydrolyzed in aqueous medium under acid or basic conditions or be used to immobilize protein and peptide sequences.

C

CH2-CH2-OH O C=O

-[CH2-C]- CH O

C

N

CH2-CH2-O-(CH2-CH2-O-)n CH2- CH2

O

C

N

+

0.3% DBTLD in THF or

pH 10 aqueous CH2-CH2-O-C-NH-(CH2-CH2-O-)n-CH2-CH2

OC=O

-[CH2-C]-CH O

N O

63

Figure 4.4 PTFE reaction mold for Implant studies and porosimetry data.

64

4.3 Results and Discussion

The synthesis of PHEMA hydrogel sponges has been well documented [1-

5]. Sponge formation is dependent upon the interaction between two phases,

polymer phase and the aqueous phase [1]. Since these phases are not the same

density, longer reaction times would allow for settling of the two phases. This

would result in a porous network that could have an inhomogeneous porous

structure. Prior investigations demonstrated that a water layer periodically

formed on top of the hydrogel, indicative of this settling polymer phase [4, 5].

For this reason, reactions were performed at higher temperatures than previous

work and also used sonication to act as a mixing aide during the reaction [6].

Sonication adds agitation to the system by means of cavitations. It has been used

in microsphere and microemulsion preparations to aide in the homogeneity as

well as uniformity of microsphere size distribution [7, 8]. To avoid the potential

problem of focusing wave propagation, a low frequency bath sonicator was used

(44-48 kHz).

4.3.1 FTIR Analysis

In order to obtain information about the success of incorporating PEG

grafts into the PHEMA hydrogel sponges, FTIR studies were performed. In the

infrared spectrum, chemical bonds absorb incident energy at the wavelength that

is resonant for each particular bond type. Each type of bond possesses several

65

different absorptions based upon the type of resonances that can be induced;

such as wobbles, oscillations, and stretching [9]. As such, FTIR is limited in its

ability to explicitly describe what bonds are present without a priori information.

A comparison of the spectra for both pure PHEMA and pure PEG monomethyl

ether is presented in Figure 4.5. As shown, the two most important bonds are the

carbonyl stretching of the ester linkage at 1730 cm-1 and the aliphatic ether

absorbance at 1110 cm-1 [9]. The ester bond is present in the PEG methacrylate,

since it is the linkage that connects the PEG to the methacrylate end cap.

However, this presence is assumed to be negligible compared to the size of the

PEG group. It was thought that a ratio between these two peaks should give the

relative fractions of each polymer. In Figure 4.7, the absorbencies of several

different blends of PHEMA and PEG are shown. Below a concentration of

6.5mol% PEG, the ether bond becomes masked by the neighboring peaks. If it is

assumed that the height of the curve at 1110 cm-1 is solely from the aliphatic

ether, then this height can be used to calculate a calibration curve of peak height

ratio to PEG mole fraction (Figure 4.6). While this assumption is not wholly

true, it is useful for helping quantitate the relative success of incorporating PEG

onto the PHEMA networks. Equation (2.4) shows the formula used to calculate

the peak height ratio.

66

ether baseline

ester baseline

A A PeakRatioA A

−=

− (2.4)

Here, Aether and Aester are the absorbencies of the ether peak and ester peak,

respectively. The Abaseline is the absorbance of the baseline of the individual

spectrum. This reference was used to account for variations in pellet thickness

and opacity.

Two techniques were used to incorporate PEG grafts onto the PHEMA

porous surface. It was originally assumed that PEG could not be incorporated

into the initial HEMA reaction mixture. Since the formation of the sponge

network is dependent upon the phase behavior of the polymer, it was thought

that PEG would increase the overall monomer solubility. As a result, the

hydrogel would not possess a macroporous structure. For this reason, a linkage

reaction of PEG diisocynate was used. Diisocyanate was used for two reasons.

First, an increase in functional groups was thought to increase the graft yield.

Also, pendant isocyanate groups could be used to immobilize proteins, such as

growth factors and adhesion ligands. For all reactions, the resulting spectra were

indistinguishable from the PHEMA reference. Even weighted subtractions for

almost all post reaction sponges possessed no ether peak. However, the organic

phase reaction with dibutyltin dilaurate at 50ºC did exhibit comparatively strong

ether absorption (Figure 4.8). Based upon the PEG:PHEMA blend calibration,

67

PEG only accounted for 1.5 wt% of this sample, which is only 0.04 mol% of the

PEG 5000 chains. This low yield is most likely due to the short shelf life of the

PEG isocyanate. Aqueous reactions were attempted, but the hydrolysis of the

isocyanate group occurred too rapidly for the linking of PEG onto the polymer

surface to occur.

Due to the low yield of the isocyanate reactions and relative expense of

the diisocyanate PEG, incorporation of PEG into the sponge formation reaction

was attempted. PEGMA was added in sparing amounts so that the phase

separation would still occur. Short chain PEG grafts were used, to evenly

disperse the PEG chains throughout the resulting sponge. In the FTIR spectra of

these samples (Figure 4.9), the ether presence is obvious. From this result, it is

evident that PEG has been successfully incorporated into the polymer. Also,

when comparing the calculated relative peak height, a PEG concentration of 5.2

wt% is obtained. This translates into a 3.4 mol% of the 200 MW PEGMA. While

this value is not very accurate, it does demonstrate that PEG was incorporated

into the sponges to a greater extent than was possible with the isocyanate

reactions. This does not determine if PEG is present on the surface, however

since PEG is more hydrophilic than the PHEMA, it is highly probable that there

is at least equal if not higher density of PEG on the surface than in the bulk

polymer.

68

4.3.2 Pore Morphology Characterization

To characterize the macroporous properties of the synthesized scaffolds,

mercury porosimetry and SEM analysis were performed. In order for these

techniques to be employed, samples were freeze-dried. Although samples were

handled carefully to prevent gross alterations of porous structure, this process

has been shown to reduce the pore size and porosity of macroporous hydrogels

compared to their hydrated state. [10] For this reason, calculated pore sizes and

porosity are likely to be somewhat smaller than the native hydrated state.

Another concern of contrary effects is in using mercury porosimetry with

samples that may be compressible, such as rubbery polymers. If a sample

compresses during the study, then the volume of mercury entering the sample

would no longer equal to the volume of the pores, hence resulting in inaccurate

readings for the pore size distribution and porosity. Samples were tested for

compressibility by running a compression corrected blank using a non-porous

dried PHEMA hydrogel as described in the materials and methods section. A

comparison of corrected and uncorrected intrusion data from the 75 water vol%

samples are shown in Figure 4.10. As depicted in the graph, the pore

distribution varies only slightly, and the average pore sizes were altered only

1.2% or less for all samples, it was concluded that compressibility had a

negligible effect on the porosity analyses. As such, all porosimetry data

69

presented were not compression corrected due to the error that may be

introduced by the additional manipulation.

For all porosimetry data, three main values were calculated; the volume

average pore size, the porosity, and the pore size dispersity index. The volume

average pore size was calculated by the following equation

i iv

i

p Vp

V= ∑∑

(2.5)

where pv is the volume average pore volume, pi is the pore size at each

incremental pressure, and Vi is the volume of mercury that penetrating the

sample at each incremental pressure point. This equation is a volume average

weighted equation, which favors the larger pores, since larger pores will possess

a greater volume.

Porosity was determined from the mercury porosimetry data obtained

using the following equation

chamber o

final o

V VV V

−ε =

− (2.6)

where Vchamber is the volume of the sample chamber, Vo is the volume that entered

the chamber at very low pressure, and Vfinal is the total mercury that entered the

sample chamber at the end of the experiment. The sample chamber was filled

with mercury at 0.37 psia. From the Washburn equation this translates into a

70

pore size of 280µm. Since our pore sizes are much smaller than this value, it is

assumed that no mercury enters the samples at the start of the experiment. Also,

the porosity measured by this technique is limited to the pores that can be

reached by mercury penetration. Hence, any pores that do not have some form

of interconnection will not be measured. This is advantageous, since only the

internal pores that can be reached by infiltrating cells are of interest. As such, the

porosity obtained by mercury porosimetry is more meaningful when evaluating

vascularization potential.

In this work, the pore size dispersity index (PDI) was defined as the ratio

of the volume average and area average pore sizes.

v

a

pPDIp

= (2.7)

i ia

i

p Ap

A= ∑∑

(2.8)

where Ai is the pore area at each incremental pressure. Since the ratio of pore

area to pore volume increases with smaller pores, this average favors smaller

pore sizes. The relationship between these two pore averages to the overall pore

distribution is shown in Figure 4.11. While this PDI is not as sensitive at the

dispersity index used in polymer molecular weights, it is still useful to determine

how varied the pore sizes are overall.

71

4.3.2.1 Surface Pore Structure

While the use of sonication seemed to have little to no effect upon the pore

morphology of the polymer sponges, there seemed to be a significant effect upon

the surface morphology of the polymer samples. The formation of a porous

structure on the surface of the reacting polymer is dependant upon three phases;

the polymer phase, the aqueous phase, and the mold surface. If the surface

possesses a greater affinity for the polymer phase, the polymer will coat the mold

excluding water, and result in a much smaller pore structure and lower porosity

at the external sponge surface. This was evident in the reactions performed in

the PTFE molded and glass molded reaction (see Figure 4.12). However, when

sonication was introduced, this energy input seemed to disrupt the interactions

between the polymer and mold surface, resulting in a more porous surface at this

interface (see Figure 4.13). These surfaces possessed a pore structure that was

more equivalent to the structure of the internal pores. For this reason, sonication

was used for all PEG-grafted polymer reactions.

4.3.2.2 Effects of Sonication on Pore morphology

Sonication has been used in the preparation of many microemulsions [7,

8]. By the introduction of agitation through cavitations, a more homogeneous

distribution of particle sizes is achieved. It was hypothesized that this same

mechanism would have a similar effect upon the distribution of polymer

72

droplets and hence pore size of the ensuing polymer sponge. The volume

average pore sizes for the sonicated and unsonicated polymer samples are

compared in Figure 4.14 . As the monomer solution becomes more dilute, the

pore sizes of the ensuing polymer samples increase. Moreover, there is no

statistical difference between pore sizes for the sonicated and unsonicated

samples.

Porosity is the ratio of pore volume to total volume. It is useful parameter

to consider, since the greater the porosity the greater probability that an

interconnected pore structure exists. When the porosity is compared, the trend

follows a similar path to that of the pore sizes (Figure 4.15). As the water content

in the reaction mixture increases, the porosity increases. This is due to the fact

that the water phase acts at the pore forming agent when phase separation

occurs. A greater content of water results in a larger porosity. Here, it is

expected that the sonicated samples would possess an overall greater porosity,

since the agitation would reduce the settling of the polymer phase prior to

gelation. This effect was evident in the 75 vol% water samples, but at the higher

dilutions there was no noticeable effect present. Moreover, for the 60 vol%

samples, a decrease in porosity was noted in the sonicated samples.

From Figure 4.16, as the water content is increased the pore size

distribution is also increased. As there are more spaces between polymer

73

droplets, the rate at which individual droplets fuse into a lattice is reduced. This

means that more time is needed for a gel to form. As a result, more time is

allowed for polymer settling which will result in a greater dispersity of pore

sizes. From this data, the presence of sonication appears to have a negative

impact on the pore size distribution. However, it is believed that this is not an

accurate interpretation of the data. Since there is a shell present on the

unsonicated samples, the surface pore structure will dominate the porosimetry

data. The internal porous structure is on average expected to be larger than the

pores on the polymer shell. When a pressure is reached that allows for the

mercury to penetrate these surface pores, the mercury will flow into the larger

internal spaces. This creates an artificially narrow pore size distribution. As

such, the distributions from the sonicated samples are more likely to be an

accurate account of the actual porous distribution, since the porous structure of

the surface of these samples is similar to the internal structure.

4.3.2.3 Comparison of PEG Grafted to Non-grafted Sponges

Since the addition of PEGMA during the synthesis step will alter the

phase behavior of the polymer, the pore size and distribution of the ensuing

polymer will be altered. For this reason, porosimetry data and SEM micrographs

were taken of both PEG-grafted and non grafted polymer sponges to determine

the effects of PEG on the final porous structure.

74

The results of the volume average pore size for pure PHEMA and PEG-

grafted sponges is shown in Figure 4.17. As the reacting solution’s water content

increased from 60 vol% to 90 vol%, volume average pore sizes increased from 5

to 14 µm for pure PHEMA and 7 to 16 µm for PEG grafted networks. The slight

discrepancy to previous published literature for pure PHEMA is most likely a

result again of the elevated reaction temperature [11, 12]. From Figure 4.18, it

is clear that as the amount of water is increased, the porosity is again increased.

The PEG-grafted samples possessed a greater porosity at all dilutions compared

to the pure PHEMA samples. This alludes to PEG-grafted networks having a

more interconnected porous structure.

In Figure 4.19, the pore size distribution index is shown as a function of

solvent content in the reaction mixture. Both samples exhibited the same trend

for pore size dispersities. As the reaction mixture dilution increased above 80%,

the pore size distribution increased. This trend seemed to be more prevalent in

the PHEMA samples then in the PEG-grafted samples. It is speculated that this

is a function of the different densities of the polymer phase.

From Figure 4.20 to Figure 4.22, a good indication that there is a uniform,

interconnected porous structure for sponges with 80% water content or greater is

shown. Interestingly, there seems to be evidence of interconnected pore

morphology of the PEG-grafted networks at just 70% water content. Previous

75

work by Dalton and Charila required 90% water to exist [4, 5, 13]. This deviation

from literature is thought to be a result from the elevated temperature of the

reaction. PHEMA and PEG both have a negative temperature dependant

solubility [11, 14]. At elevated temperatures, phase separation should occur with

less water, hence larger porosity. It is evident from these micrographs, that the

PEG network’s phase behavior is different from the pure PHEMA. Pure PHEMA

possesses a sintered microsphere structure. These microspheres are a result of a

strong repulsion of the two phases. This strong repulsion results in a minimum

surface area per volume structure, which is a sphere. While the PEG-grafted

networks have a lattice type structure that resembles what could be a type of

bicontinuous phase. While PEG does possess a negative temperature behavior, it

is still thought to be more hydrophilic than the PHEMA polymer. For this

reason, the PEG grafts may act as a type of surfactant, in stabilizing the repulsive

energies at the surface of the polymers.

4.3.3 Mechanical Analysis

There is increasing evidence that the soft tissue response is dependent

upon the mechanical properties of the implanted material. [15] A biomimetic

approach to this statement would be that an implant that possesses a material

modulus similar to the surrounding tissue will be better tolerated. Also, the

more ductile the material is, the more readily endothelial cells express can show

76

the tubule phenotype [16]. For these reasons, the mechanical properties of the

PHEMA sponges have been evaluated.

Due to the porous nature of the implants and the edge effects imparted by

reaction molds, tensile experiments upon thin film samples were not possible.

For this reason, compression studies were used to obtain a relative comparison of

sample modulii. The compression studies for the PHEMA sponges and PEG-

grafted PHEMA sponges are shown in Figure 4.23 and Figure 4.24, respectively.

The samples of 40 vol% and 50 vol% water are non-macroporous samples. These

samples represent the compressive modulus of the bulk polymer without pores.

The plot shown is the engineering stress vs. negative strain. This convention was

used to depict the stress-strain response in the right hand quadrant. The 90 vol%

samples of both the pure and PEG-grafted networks were not able to support

their own weight in ambient conditions. In order to compare the 90 vol% sample

data to the other samples, the samples heights were measured first in water then

in atmosphere. This height was calculated and considered the initial strain point.

All 90 vol% samples were offset by this initial strain value. As shown, the

mechanical response of the “bulk” hydrogels varies only slightly. They exhibit

the classic compressive response of elastic materials. As the water content

increased, the stress-compression response exhibited two modulii; an initial

weak response followed by a dramatic increase in modulus. The reason for this

77

behavior is due to the porosity of the samples. As these materials compress, the

macropores collapse which allow for large deformation with little stress. When

these macropores are mostly closed, the bulk material starts to compress which

results in a modulii with the same order of magnitude as the 40 and 50 vol %

polymer samples. For the 60 and 70 vol% samples, these two compression

pathways appear to occur in tandem resulting in a stress-strain response

between these two extremes. It is interesting to note that the modulus increased

for the 80 vol% and 90 vol% samples roughly at the point where the strain is

equal to the porosity. This lends credence to the idea that the collapsing of the

pores is the dominating mechanical response.

In Figure 4.25, the initial compressive modulus is plotted as a function of

solvent dilution in the reaction. When the dilution is increased from 60 to 80

vol%, the modulus drops by an order of magnitude, and again another order of

magnitude when increased to 90 vol%. For most dilutions except 70 vol%, the

PEG grafted and pure PHEMA sponges had roughly comparable modulii. Pure

PHEMA was always slightly higher, since PEG-grafted networks are more

hydrophilic and have a greater equilibrium water content. PEG-grafted sponges

possessed a lower modulus at 70 vol% compared to the pure PHEMA sponges,

since PEG-grafted PHEMA possesses a larger porosity and pore size at 70 vol%.

78

This data provides further evidence that these materials may have an

interconnected porous structure with less water content in the reaction mixture.

79

Figure 4.5 FTIR spectrograms of PHEMA and PEG monomethyl ether. The key absorbencies are the ester linkage absorbance of PHEMA at 1730 cm-1, and PEG’s aliphatic ether absorbance at 1110 cm-1.

80

y = 3.0709x + 0.2893R2 = 0.9322

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

0 0.1 0.2 0.3 0.4 0.5 0.6

Mole fraction of PEG in PEG:HEMA blend

Rat

io o

f Est

er p

eak/

ethe

r pea

k

Figure 4.6 Peak height ratio of PEG:PHEMA as a function of PEG mole fraction.

Calibration is based upon blend of the two polymers and not of physically linked copolymers.

81

Figure 4.7 Varying blends of PEG and PHEMA to determine if FTIR can be used to

calculate the relative concentration of PEG and PHEMA.

82

Figure 4.8 Subtractions of sponges from PEG-isocyanate reaction with 90 vol% water

PHEMA sponges with 0.3% dibutyltin dilaurate in THF at 50ºC. This was the only PEG reaction that exhibited a moderate amount of PEGylation.

83

Figure 4.9 Subtraction result of (Blue) PHEMA with PEG minus PHEMA. (Red) FTIR of

PEG monomethyl ether (350 MW).

84

0

0.05

0.1

0.15

0.2

0.25

0.3

1 10 100 1000

Pore Diameter(µm)

Cum

mul

ate

Intr

usio

n V

olum

e(m

l/g)

0

0.0005

0.001

0.0015

0.002

0.0025

0.003

0.0035

0.004

0.0045

0.005

INcr

emen

tal I

ntru

sion

Vol

ume(

ml/g

)

Figure 4.10 Comparison of compression corrected and uncorrected porosimetry data. As

shown there is negligible difference in the cumulative mercury intrusion volume of the uncorrected sample ( ) and corrected ( ). The only visible difference occurs at the larger pore sizes in the incremental intrusion volume for the uncorrected(-) and corrected (- -) data.

85

0

0.001

0.002

0.003

0.004

0.005

0.006

1 10 100 1000

Pore Diameter(µm)

Incr

emen

tal I

ntru

sion

Vol

ume

(ml/g

)

Da=9.6µm

Dv=10.95µm

Figure 4.11 A porosimetry plot of the unsonicated 85 vol% diluted PHEMA sponge. The

average pore diameters calculated are presented to demonstrate the relationship between the porosimetry data and the statistics that are calculated.

86

(a)

(b)

Figure 4.12 Micrograph of 30vol% PHEMA polymer surfaces. Reduced pore sizes were evident in both (a) PTFE molds and (b) glass molds.

87

Figure 4.13 Surface pore structure of PHEMA sponges reacted in a PTFE mold with

sonication.

88

0

2

4

6

8

10

12

14

16

18

0.50 0.60 0.70 0.80 0.90 1.00

Vol

ume

Ave

rage

Por

e Si

ze (µ

m)

Fractional Solvent Content in Monomer Solution

Figure 4.14 Volume average pore size as a function of reaction mixture dilution. PHEMA ( ) with and ( ) without sonication. (n=4 ± SE)

89

50%

60%

70%

80%

90%

0.50 0.60 0.70 0.80 0.90 1.00

Poro

sity


Figure 4.15 Porosity as a function of reaction mixture dilution. PHEMA ( ) with and ( )

without sonication. (n=4 ± SE)

90

1

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.9

0.50 0.60 0.70 0.80 0.90 1.00

Pore

Siz

e D

ispe

rsity

Inde

x


Figure 4.16 Pore size dispersity index as a function of reaction mixture dilution. PHEMA ( ) with and ( ) without sonication. (n=4 ± SE)

91

Figure 4.17 Volume average pore size as a function of reaction mixture dilution. PHEMA

( ) with PEG grafts and ( ) without PEG grafts. (n=4 ± SE)

0

2

4

6

8

10

12

14

16

18

20

0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95Solvent Volume Fraction in Reaction Mixture

Vol

ume

Ave

rage

Por

e Si

ze (µ

m)

92

Figure 4.18 Porosity as a function of reaction mixture dilution. PHEMA ( ) with PEG

grafts and ( ) without PEG grafts. (n=4 ± SE)

40%

50%

60%

70%

80%

90%

100%

110%

0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95

Solvent Volume Fraction in Reaction Mixture

Poro

sity

93

Figure 4.19 Pore size dispersity as a function of reaction mixture dilution. PHEMA

( )with PEG grafts and ( ) without PEG grafts. (n=4 ± SE)

1

1.2

1.4

1.6

1.8

2

2.2

0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95

Solvent Volume Fraction in Reaction Mixture

Pore

Siz

e D

ispe

rsity

Inde

x

94

(a)

(b)

Figure 4.20 SEM Micrographs of (a) PEG-grafted PHEMA and (b) pure PHEMA polymer sponges using 70 vol% water.

95

(a)

(b)


96

(a)

(b)


97

0

20

40

60

80

100

120

140

160

180

200

0% 20% 40% 60% 80% 100%Strain (Percent Compressed)

Stre

ss (K

Pa)

40 vol %

50 vol %

60 vol %

70 vol %

80 vol %

90 vol %

Figure 4.23 Stress-strain response of pure PHEMA sponges. Each curve is labeled based

upon the vol% of water in the reaction solution. As the dilution increased, initial modulus decreased.

98

0

20

40

60

80

100

120

140

160

180

0% 20% 40% 60% 80% 100% 120%

Strain (Percent Compressed)

Stre

ss (K

Pa)

40 vol% 50 vol%

60 vol%

70 vol%

80 vol%

90 vol%

Figure 4.24 Stress-strain response of 6.5 mol% PEG-grafted PHEMA sponges. Each curve is

labeled based upon the vol% of water in the reaction solution. As the dilution increased, initial modulus decreased.

99

1

10

100

1000

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Solvent Volumen Fraction in Reaction Mixture

Mod

ulus

(KPa

)

Figure 4.25 Initial compressive modulus of ( ) pure and ( ) PEG-grafted PHEMA

sponges as a function of solvent volume fraction in the reaction mixture.

100

4.4 Conclusions

PHEMA sponges with and without PEG grafts can be synthesized with a

reproducible pore size, porosity, and mechanical properties. The water content

in the reaction mixture provided the greatest effect upon varying all these

properties. This was due to the water phase behaving as the pore forming agent

when phase separation occurs. The inclusion of PEG into the polymer synthesis

reaction resulted in a greatly altered phase behavior when compared to the pure

PHEMA sponges. Pure PHEMA morphology resembled typical oil/water

mixtures; polymer droplets dispersed in an aqueous medium. On the other

hand, the PEG grafted networks possessed a more cylindrical type lattice

structure. The 70 vol% water resembled a bicontinuous phase system. Such an

interconnected porous network would be ideal for cellular penetration. It was

possible to synthesize networks with pore sizes in the greater than 10µm range,

which is required for cellular penetration. The modulus of the more porous

networks, while fragile, was similar to that of soft tissues.

101

List of References

[1] Chirila, T.V., Constable, I.J., Crawford, G.J., Vijayasekaran, S., Thompson, D.E., Chen, Y.-C., Fletcher, W.A., and Griffen, B.J. ʺPoly(2-Hydroxyethyl Methacylate) Sponges as Implant Materials: In Vivo and in Vitro Evaluation of Cellular Invasion.ʺ Biomaterials, 1993, 14: p. 26-38.

[2] Chirila, T.V., Higgins, B. and Dalton, P.D. ʺEffect of Synthesis Conditions on the Properties of Poly(2-Hydroxyethyl Methacrylate) Spongesʺ. Cellular Polymers, 1998, 17(3): p. 141-162.

[3] Simpson, B.J. ʺHydron: A Hydrophilic Polymerʺ. Biomed Eng, 1969, 4: p. 65-68.

[4] Clayton, A.B., Chirila, T.V. and Lou, X. ʺHydrophilic Sponges Based on 2-Hydroxyethyl Methacrylate. V. Effect of Crosslinking Agent Reactivity on Mechanical Propertiesʺ. Polym Int, 1997, 44: p. 201-207.

[5] Clayton, A.B., Chirila, T.V. and Dalton, P.D. ʺHydrophilic Sponges Based on 2-Hydroxyethyl Methacrylate. Iii. Effect of Incorporating a Hydrophilic Crosslinking Agent on the Equilibrium Water Content and Pore Structureʺ. Polym Int, 1997, 42(1): p. 45-56.

[6] Dziubla, T.D., Torjman, M.C., Joseph, J.I., Murphy-Tatum, M., and Lowman, A.M. ʺEvaluation of Porous Networks of Poly(2-Hydroxyethyl Methacrylate) as Interfacial Drug Delivery Devicesʺ. Biomaterials, 2001, 22(21): p. 2893-2899.

[7] Scholes, P.D., Coombes, A.G.A., Illum, L., Davis, S.S., Vert, M., and Davies, M.C. ʺThe Preparation of Sub-200nm Poly (Lactide-Co-Glycolide) Microspheres for Site Specific Drug Deliveryʺ. J Control Rel, 1993, 25: p. 145-153.

102

[8] Reich, G. ʺUltrasound-Induced Degradation of Pla and Plga During Microsphere Processing: Influence of Formation Variablesʺ. Euro J Pharm Biopharm, 1998, 45: p. 165-171.

[9] Bellamy, L.J., The Infrared Spectra of Complex Molecules. 2 ed. Vol. 2. 1980, London and New York: Chapman and Hall.

[10] Liu, Q., Hedberg, E.L., Liu, Z., Bahulekar, R., Meszlenyi, R.K., and Mikos, A.G. ʺPreperation of Macroporous Poly(2-Hydroxyethyl Methacrylate) Hydrogels by Enhanced Phase Seperationʺ. Biomaterials, 2000, 21(21): p. 2163-2169.

[11] Oh, S.H. and Jhon, M.S. ʺTemperature Dependence of Unperterbed Dimensions for Isotactic Poly(2-Hydroxyethyl Methacrylate)ʺ. J Polym Sci Polym Chem, 1989, 27: p. 1731-1739.

[12] Šprincl, L., Kopecek, J. and Lim, D. ʺEffect of the Structure of Poly (Glycol Monomethacrylate) on the Calcification of Implantsʺ. Calc Tiss Res, 1973, 13: p. 63-72.

[13] Lou, X., Dalton, P.D. and Chirila, T.V. ʺHydrophilic Sponges Based on 2-Hydroxy Ethyl Methacrylate. Part Vii: Modululation of Sponge Characteristics by Changes in Reactivity and Hydrophilicity of Crosslinking Agentsʺ. J Mat Sci Mat Med, 2000, 11(5): p. 319-325.

[14] Harris, J.M., Poly(Ethylene Glycol) Chemistry, Biotechnical and Biomedical Applications. 1992, New York: Plenum Press.

[15] Sieminski, A.L. and Gooch, K.J. ʺBiomaterial-Microvasculature Interactionsʺ. Biomaterials, 2000, 21: p. 2233-2241.

103

[16] Ingber, D.E. and Folkman, J. ʺMechanochemical Switching between Growth and Differentiation During Fibroblast Growth Factor-Stimulated Angiogenesis in Vitro: Role of Extra Cellular Matrixʺ. J Cell Bio, 1989, 109: p. 317-330.

104

CHAPTER 5: IN VITRO VASCULARIZATION 5.1 Introduction

Based upon in vivo studies, there are several trends known to exist for

porous implants [1-6]. Implants with small pores are surrounded by vascular

tissue with only macrophages and fibroblasts penetrating the implants. As pore

sizes increase, the density of vascular penetration increases up to a maximum at

60µm. When the pore sizes become too large, the density of vascular penetration

decreases dramatically. While it is not fully known to what extent material

properties play in this response, it is believed that this general trend exists in all

porous implants [7].

While vascular response is highly dependent on the pore morphology, the

exact mechanisms for this dependency are still unknown. For instance, can

smaller pores be designed to allow for increased vasculature? Can vessels

penetrate the implants without the initial invasion of macrophages and

fibroblasts? To answer these questions, in vitro techniques developed in

angiogenesis research may prove useful. Angiogenesis is the formation of new

blood vessels from existing vasculature [8-11]. This self-limiting process occurs

naturally in reproduction, wound repair, placental development, and the foreign

body response. One crucial step of this process, tubule formation, can be

replicated in vitro. When endothelial cells are grown on a 3 dimensional

105

collagen, fibrin matrix in the presence of specific growth factors, these cells curl,

associate, and elongate into tubule structures that resemble immature capillaries

[12, 13]. While the focus of these studies have been on pro and anti angiogenic

factors for cancer treatments, it is thought that this model may serve as an

excellent tool for evaluating the vascularization of implant biomaterials, in vitro.

One additional advantage gained in performing in vitro experiments, is that

endothelial cells derived from human origin can be used. This allows for the

direct evaluation of implant design on the human microvascular endothelial cells

(HMVEC), which are the exact cells involved in implant vascularization in

humans.

5.1.1 Cellular Techniques used in Angiogenesis Research

The original human endothelial cell line isolated for in vitro research was

human umbilical vein endothelial cells (HUVEC) [14, 15]. This was due to the

readily available nature of the cells, as well as the ease of harvesting and

isolating for cell culture work. However, these cells are macrovascular and are of

an embryonic origin. Macrovascular cells are cells obtained from the larger

vessels such as the aorta or superior vena cava. These cells are typically larger

than the microvascular cells, and not the endothelial cells that are involved in

typical adult angiogenesis [15]. Hence, data obtained may not be fully applicable

toward angiogenesis as seen in the microvasculature in the adult human. For

106

example, shiga toxins are acutely toxic toward HMVEC but possess little toxicity

toward HUVEC [16]. To address these concerns, researchers have started to use

HMVEC obtained from either the adult dermis or lung as model systems. Due

to the smaller nature of these cells, it is believed that HMVEC in vitro porous

penetration data would be more indicative to the pore morphology limitations of

implant vascularization.

5.1.2 Biomaterial-Endothelial Cell Interaction Experiments

The bulk of in vitro biomaterials research has centered on issues such as

cytotoxicity and cell seeding technologies. [17, 18] For endothelial cells, these

studies focused on how to improve cell adherence onto vascular grafts. For

vascular grafts, increased proliferation and cellular attachment results in a

decreased probability of thrombogenesis, blood clot formation [19]. While these

studies are useful for evaluating vascular grafts, little useful information is

obtained about a material’s ability to support capillary ingrowth. For this reason,

most research analyzing vascularization has been performed in in vivo models

[7]. For more explicit analysis of vascular growth into biomaterials, an in vitro

endothelial cell growth into 3-dimensional biomaterial scaffolds is needed.

In this section, in vitro angiogenesis studies on the 3 dimensional

macroporous soft polymers synthesized in chapter 4 are performed. The

ultimate effects of pore size and porosity are then compared to tubule

107

parameters, such as average tubule length and tubule diameter. These values are

then evaluated against standard in vitro angiogenesis conditions (Matrigel®). It

is hoped to obtain information about the limits of pore size and porosity on

implant vascularization without the presence of cells lines that compete in

porous implant infiltration.

5.2 Materials and Methods 5.2.1 Cell Handling and Storage

HMVEC-adult dermal (HMVEC-ad) (Biowhittaker, Inc. Wakersville, MD)

were received in a cryogenic frozen ampoule of 500,000 cells in 1 ml, and were

stored in a liquid nitrogen cryogenic chamber until use. Cells were in 3rd passage

upon arrival. Endothelial Growth Media (EGM-2MV, Biowhittaker)

supplemented with 5% fetal bovine serum (FBS) and standard list of HMVEC

growth factors (see Table 5.1). Gentamicin and Amphotericin-B were added as

antibiotic and antimicrobial agents. Prior to all cell manipulations, a sterile field

was established in a laminar flow hood by wiping down the area with 70 vol%

isopropanol in water solution. All items used were autoclaved unless stated

otherwise.

108

Table 5.1 List of Supplements added to the EGM-2-MV Media Supplement Human Recombinant Epidermal Growth Factor (hEGF) Human Fibroblast Growth Factor-Basic with Heparin (hFGF-B) Vascular Endothelial Growth Factor (VEGF) Ascorbic Acid Hydrocortisone Human Recombinant Insulin-like Growth Factor Gentamicin Amphphotericin-B

Cell maintenance procedures followed the recommend guidelines

established by Biowhittaker. For first propagation, 15 ml media was equilibrated

in a T-75 Flask (75cm2) in an incubator set at 37 and 5% CO2 for 20 minutes. This

was to ensure that media contained the desired temperature and dissolved CO2

concentration to minimize cell shock. Cryovials were warmed in a water bath at

37ºC until all ice crystals were melted. Immediately following the melting of the

last ice crystal, the cells were pipetted out of the vial and into the equilibrated T-

flask. Cells were inspected ½ hour after incubation to ensure cell viability.

Media was replaced every two days. When cells reached 50% confluency, 20ml

of media were used. At 70-80% confluency, cells were trypsinized (0.025%

trypsin in EDTA solution, Biowhittaker) and split into more T-75 flasks. Cells

were counted using a hemocytometer, and flasks were seeded with 500,000 cells

each.

109

5.2.2 Cryogenic Freezing of Endothelial Cells

Freezing media was prepared by mixing 60 vol% EGM-2MV with

supplements, 30vol% FBS, and 10vol% sterile filtered dimethyl sulfoxide (Cell

culture grade, Sigma, Milwaukee, WI). At passage 5 or 6, endothelial cells were

harvested. Counted cells were suspended in the freezing media at a

concentration of 1.5million cells/ml. The suspension was the pipetted into sterile

cryovials at 1ml/vial. These vials were then placed in a Naglene® controlled rate

freezer (Fisher Scientific, Hanover Park, IL) and frozen to -80ºC over 12 hours.

Afterwards, cells were transferred into a cryogenic freezer for storage.

5.2.3 In vitro Biomaterial Vascularization Studies

Sponges were synthesized as mentioned in Section 4.2.1. Discs with

thicknesses varying from 1 to 5 mm were cut using a PTFE coated razor blade.

Ethanol was used as a machining lubricant to reduce sample tearing. These

samples were then immersed in a 70vol% ethanol/water solution for storage and

sterilization. Prior to seeding experiments, ethanol was removed from the

samples by a serial dilution of 70/30 ethanol to pure sterile deionized water using

5 steps. Samples were then equilibrated in media for an additional 2 steps. 60-90

vol% of both PHEMA and PEG-grafted PHEMA samples were placed in 24-well

plates. Four samples per formulation were studied. These samples were

incubated at 37ºC and 5% CO2 for 20 minutes prior to seed. Endothelial cells

110

from passage 8 or 9 were collected for the biomaterial experiments. Each sample

was seeded with 250,000 cells/cm2 based upon disc surface area. The porous

surface area was not considered when calculating the number of cells required.

Cell concentrations were established such that seeded of samples would occur

with 0.5 ml media. The seeded samples were then allowed to equilibrate for 0.5

hour to allow for attachment/settling. After this time, an additional 1 ml of

warmed media was added to each sample well. Media was replaced every other

day until the end of the experiment, 2 weeks. Wells containing 0.5 ml Matrigel®

were used as a positive control for the experiment. Positive controls were

evaluated daily using light microscopy.

5.2.4 Matrigel® Impregnated Sponges

Matrigel® is a liquid at 4ºC and gels at 37ºC. In order to infuse Matrigel®

into the polymer samples, samples and pipette tips were first chilled in a

refrigerator. Then, 0.5ml of Matrigel® was infused into each type of polymer

formulation. These Matrigel® loaded samples were then placed into the

incubator and allowed to equilibrate for 1 hour. After this time, cells were

seeded according to the method previously mentioned.

5.2.5 Sample Fixation and Sectioning

When the experiments were finished, samples were fixed in a neutral

buffered 10% formalin solution (Sigma Chemical Co, Saint Louis, MO). 50 ml

111

solution was used for 4 samples. The samples were kept overnight at room

temperature, and then stored in a refrigerator for 1 week. After this time,

samples were prepared for sectioning by emersion into a 30% sucrose solution in

7.4 pH phosphate buffered saline.

A cryostat (CM 3000 Microtome, Leica Microsystems, Bannockburn, IL)

was used for sample sectioning. Samples were oriented so that sections were

perpendicular to the seeding surface, and were frozen in tissue imbedding media

(Tissuetek, Sakura Finetechnical Co. Ltd, Japan). Frozen samples were then

mounted onto the cryosectioner and cut into 60µm sections. These sections were

mounted onto glass slides treated with a gelatin/polylysine coating for enhanced

adhesion. Glass slides were stored in a freezer until staining and viewing.

5.2.6 Immunoflourescent Microscopy

In order to view HMVEC-ad within the polymer network, a secondary

immunoflourescent staining procedure was employed. Slides were coated with a

1 wt% bovine serum albumin (BSA) in PBS solution for 0.5 hour at 37ºC to block

nonspecific protein absorption. The slides were then rinsed twice with PBS to

remove unbound BSA. Polyclonal Rabbit antihuman Von Willenbrand factor

(DAKO Corporation, Carpenteria, CA) was diluted to 1:200 and added to the

slide (300µl/slide). This primary staining step was allowed to proceed for 0.5

hour at 37ºC. Slides were subsequently washed with PBS and allowed to dry to

112

prevent detachment of sections from the glass slide. Following the drying step,

samples were then wetted with a 3µg/ml solution of the secondary antibody,

goat-anti rabbit IgG labeled with Alexafluor® 488 (Molecular Probes, Inc.

Eugene, OR). This binding step was allowed to proceed for 1 hour at 37ºC. After

the fluorescent label was added, slides were kept covered with aluminum foil to

reduce photo bleaching. Samples were then visualized with using an Axioskop2

plus fluorescent microscope (Carl Zeiss, Göttingen, Germany).

5.3 Results and Discussion Ethanol/water was used to sterilize the hydrogels, because of the

harshness of available alternatives. While some reports have shown PHEMA

sponges are stable under autoclave temperatures and conditions. It is known that

PEG’s ether linkages are unstable at temperatures greater than 80ºC [20, 21]. As

such, thermal sterilization was not possible. UV sterilization was not reliable

since the sponge networks are opaque in the uv-vis range. It was found though

the course of this study that the ethanol/water was an adequate means of

sterilization for the in vitro experiments.

Due to the increased swelling of the polymer networks in ethanol/water, a

serial dilution was necessary to remove all residual ethanol. If samples were

placed directly into water, the outside edge of the polymer samples would

rapidly deswell. This resulted in a layer that greatly impeded the remainder of

113

ethanol from diffusing out. Serial diffusion prevented this “shell” from forming,

which allowed for adequate solvent exchange.

5.3.1 Positive Endothelial Tubule Formation Control

High passages of HMVEC lose the ability to express the tubule formation

phenotype. For this reason, low numbered passages were used for all

experiments. To ensure the cells used in the seeding experiments were still

capable of tubule formation, a Matrigel® positive control was used. Tubule

formation in the controls samples was evident after 1 day. Further organization

continued up to 3 days, and expressed this phenotype for the entire study, 2

weeks. Using a scale bar (Figure 5.1) and keeping a constant image resolution, it

was possible to obtain approximations of the tubules sizes. Using imaging

software, 4 points were selected to establish a rectangle around each tubule. The

size of this rectangle was calculated in pixels, and the values were then converted

into µms based upon the pixel to length ratio. The lengths of tubules measured

varied from 100 up to 550 µm. The average tubule length was calculated at

340µm with a standard deviation of 150µm (Figure 5.2). The tubule diameters

also varied greatly from 5 to 35 µm, with an average around 11µms. In these

tubules, nodules were evident. These nodules allowed for a grid type

arrangement of the tubules, the organization that is seen in highly vascularized

membranes such as the mesentery [22]. These findings are consistent with

114

MATRIGEL® and HMVEC using the EGM2-MV media, and verify the ability for

cells to express the desired phenotype [23].

Interesting to note was one cell where Matrigel® did not fully coat the

bottom of the surface well. In this case, the endothelial cells would attach as a

confluent layer to the bottom of the tissue culture polystyrene (TCPS) bottom

(Figure 5.3). A transition interface was evident, where the endothelial cells were

started to express tube formation. The tubule formation would occur along the

interface, and then travel into the bulk of the Matrigel® basement.

5.3.2 Analysis of Fluorescently-Labeled HMVEC Seeded Networks

Originally, gluteraldyhyde was used as the fixative. However, the

destabilization of the protein’s secondary structures (alpha helices and beta-

sheets) from gluteraldyhyde fixation resulted in a low antibody binding.

Formalin was shown to not significantly reduce the antibody sensing of the

target proteins.

Negative immunoassaying controls were performed on the 90 vol% PEG-

grafted and non PEG-grafted PHEMA samples to obtain information about the

non-specific staining that occurs under the experimental conditions. From

Figure 5.5, the haze of green is a result of backscattering of light from the

polymer sample. The random points of fluorescence shown are examples of non-

specific staining. This can happen in varying degrees from sample to sample.

115

However, it can be assumed that the irregular, brightly fluorescing structures are

the positively stained HMVEC. Due to errors in sample sectioning, it was not

possible to obtain information on depth of penetration. Instead, information

about the size and length of penetrating tubules were obtained.

5.3.2.1 Immunohistology of 60 vol% PHEMA Hydrogel Sponges

60 vol% samples possessed the smallest pore size with the lowest porosity.

As such, it is expected that these samples will possess very little cellular

penetration and vascularization. This was the case for both the grafted and the

non grafted samples. It was found that the PEG-grafted samples possessed little

to no endothelialization (Figure 5.6). This reduced attachment and spreading is

believed to be a result of the PEG-grafts ability to resist protein deposition.

Reduced protein adhesion results in lower amount of fibronectin and collagen on

the material surface, without which cells can not attach and spread.

There was some evidence of attachment and spreading on the pure

PHEMA sponge samples (Figure 5.7). In order to obtain size information from

these images, a scale bar was imaged to determine the pixel to µm ratio for all

images taken on the fluorescent microscope (Figure 5.4). PHEMA does not

possess the same protein resistance as PEG. This results in the attachment of the

HMVEC leading to some cell spreading and growth. In a pure sponge cross-

section (Figure 5.8), cells were evident only on the surface. This layer was

116

slightly detached, most likely a result of sectioning technique. The thickness of

this detached structure was 12-16µms.


The 70 vol% samples possessed a significant change in the HMVEC-ad

response. PEG-grafted networks at this dilution possess a pore size of 10-12µm.

It is expected that these networks will be conducive to some vascular

penetration. In Figure 5.9, HMVEC presence is obvious at the surface. Few of

the HMVEC were evident on the surface, while the majority seemed to have

penetrated into the porous structure. This was confirmed in a cross-sectional

photo taken at 200X (Figure 5.10). This figure was taken near the surface of edge

of the sample, and illustrates the penetration of the HMVEC. The sizes of

tubules averaged 5.2 µm with a standard deviation of 3.2 µm. This was

calculated based upon the random selection of 11 locations and measuring the

tubules present at each location. The average length was only 26.5 µm with a

standard deviation of 7 µms. These vessels were significantly smaller than those

in Matrigel®. This is thought to be due to the pore diameter being too small for

larger vessels to organize and form. There was some evidence of larger tubule

structures with in some sections of the hydrogels. Figure 5.11, depicts several

vessels with a length of 98µm and diameter of 7µm. These vessels are probably

following channeled defects within the porous sponge. Moreover, this figure

117

depicts a greater concentration of EC at outer edge of the polymer sample. This

is most likely due to the greater nutrient exchange possible at this edge.

The results of the pure PHEMA samples were dramatically different.

These networks still possess a small pore size and porosity (approximately 7µm

and 70%). This results in a network that is unlikely to have any cellular

penetration. In fact, the surface analyzed sample possessed almost no evidence

of endothelial cells (Figure 5.12). This is most likely due to the lower staining

efficacy of these samples. These networks were originally stained with an

unoptimized version of the reported staining procedure. However the same type

of endothelial cell layer seen in the 60 vol% pure PHEMA samples was evident

when cross sections were taken. This leads to the conclusion that while the 70

vol% pure PHEMA networks can support endothelialization of the surface,

tubule formation and penetration does not occur (Figure 5.13).


At the 80 vol% concentration, there was extensive evidence of HMVEC

penetration and tubule formation. According to the porosimetry data, the pore

size is about 10-11 µms for both networks, while the PEG grafted and non-

grafted sponge porosity is 85% and 75%, respectively. The surface staining of the

80 vol% PEG grafted and ungrafted networks both exhibited endothelial

penetration, and little surface attachment and spreading (Figure 5.14 and Figure

118

5.15). This is due to the reduced regularity of the surface. The porous surface

does not have large enough flat domains allowing for cellular spreading. From

the surface analysis, the pure PHEMA sponges seem better suited in supporting

HMVEC. However in the cross sectional analysis, a more accurate view of the

HMVEC response is obtained. The PEG grafted samples contained many

endothelial tubules (Figure 5.16). These tubules possessed a diameter similar to

the Matrigel® reference (10.6µm, 4.6µm standard deviation), with a highly

variable tubule length (77 µm, 97µm standard deviation). These values were

calculated by randomly selecting 17 imaging locations and evaluating the tubule

sizes present at each location. The variability of the tubule lengths is most likely

due to the 3-dimensional system. A single cross section will contain tubules that

are oriented along different planes. The size that is measured in this method is

the projection of each tubule onto the cross sectioned plane.

The cross sections of the 80 vol% pure PHEMA samples reveal two

interesting facts. First, these samples seem to possess a greater inhomogeneity of

pore size than what was evident from porosimetry and SEM imagery. In Figure

5.17, it is evident that large 50 microns pockets exist. This resulted in a non-

interconnected structure. Fewer large pores are needed to equal the same

porosity of a material with smaller pores. The fewer number of pores is then

119

spread farther apart, and is not connected to each other. This is the reason why

little evidence of HMVECs was found beyond a depth of 180 microns.


At 90 vol%, both the PEG-grafted and non PEG-grafted samples are

extremely fragile. Due to this fragility, it was not possible to maintain a

consistent orientation of the sectioned samples. Hence, comparisons made

between these samples and lower dilution samples must be qualified with

processing errors that may be present. In spite of this difficulty, there are still

some valuable insights that can be obtained. These networks both have

approximately 15µm pores with porosities of 90% for the PEG-grafted and 80%

for the non PEG-grafted samples. Hence, both networks should be capable of

supporting HMVEC tubules.

Cellular penetration is shown in the surface images, Figure 5.18 and

Figure 5.19. These samples both have a high surface roughness. This resulted

into multiple plans of focus that could not be simultaneously resolved. This is

the reason why only localized areas of EC are visible. These areas, however, are

representative of all surfaces imaged. The surface samples show a similar

cellular response as that in the 80 vol% PEG grafted sponges. The lack of cellular

spreading on the polymer surface is evident. From the cross sections of the PEG

grafted sponges, it can be seen that the tubule network is very similar in

120

structure to the Matrigel® reference tubules. From a sampling of 15 locations,

the diameter was calculated as 7.85µm with a standard deviation of 3.5µm. This

is only slightly smaller than the tubules in the Matrigel®. The tubule lengths

were, again, highly variable with an 88µm average. One tubule in Figure 5.20

measured 450µm in length. The highly open structure of these polymers allowed

for much longer tubule lengths. This was almost identical to the cross sectional

results of the pure PHEMA networks (Figure 5.21). These samples possessed a

tubule diameter of 7.5+-2.6 µm, and a tubule length of 102µm with a maximum

length of 680µms. This evidence supports the claim that it is physical

morphology that dominates the vascularization behavior of materials.

5.3.3 Matrigel® Loaded Polymer Samples

Due to Matrigel®’s relative expense, only one sample of each polymer

types was analyzed. Matrigel® is obtained from immortalized carcinoma cells.

As such, they contain not only the fibronectin and laminin necessary for EC

cellular attachment, but also contain the vascularization signals that are released

from these tumor cells. It is expected that these loaded samples will be more

conducive to tubule formation, and penetration.

For the majority of samples analyzed, the response did not vary much

from the samples without MATRIGEL®. This is most likely due to the ability of

EC to secrete its own basement membrane, and then attach to this secreted layer.

121

Matrigel® might have aided the time it takes for tubule formation, however

shorter timed experiments would be needed to verify this hypothesis. After 2

weeks, however, there was little difference between loaded and unloaded

samples, except in the 60 vol% PEG-grafted polymers (Figure 5.22). The

Matrigel® was able to form a layer onto the PEG surface. It is not clear whether

this layer was attached to the PEG, but it did allow for tubule formation on the

surface of these PEG grafted networks.

A HMVEC tubule penetrating into the side of an 80 vol% PEG-grafted

polymer is seen in Figure 5.23. In this sample, not all of the Matrigel®

penetrated into the polymer. On one edge of the sponge, the Matrigel®

accumulated, and allowed for tubule formation to occur outside the polymer.

This allowed for the visualization of tubule growth into the porous network.

122

Figure 5.1 Scale bar taken at 250X magnification and 1712X1368 resolution. Under these

settings, 175 pixels was equivalent to 100µm.

123

Figure 5.2 Matrigel® Positive control reference. Tubule formation was evident after 1

day. Tubule legths and diameter s varied greatly. Scale bar is equal to 100µm.

124

Figure 5.3 This sample was not fully covered by the Matrigel® basement. This resulted

in a hybrid expression; confluent EC that turns into tubules at the Matrigel® TCPS interface.

125

Figure 5.4 Scale bar taken with a fluorescent microscope at 100X.

126

(a)

(b)

Figure 5.5 Negative control staining of 90vol% (a) PEG-grafted and (b) pure PHEMA -hydrogel sponges. From this result, it can be assumed that all brightly fluorescing structures are positively stained HMVEC.

127

Figure 5.6 Surface of 60 vol% PEG-grafted PHEMA sponges. Bright spots represent

endothelial cells. The lack of cell spreading and tube formation is indicative of pore sizes too small for penetration as well as a surface with no adhesive properties.

128

Figure 5.7 Surface adhesion of HMVEC-ad onto 60vol% pure PHEMA hydrogels. After 2

weeks culture, cells were spread onto the surface into elongated structures. These structures are more similar to the attachment of EC onto TCPS than the tubule formation.

129

Figure 5.8 Cross-section of 60 vol% pure PHEMA hydrogel. As shown, no endothelial

cells penetrated into the small pores of these networks. However, endothelial cells were evident on the surface of the polymer sponge. In this photo, cells have detached from the surface in a thread shape. It is not clear whether these cells are in tubule formation, or a slice of a confluent layer.

130

Figure 5.9 Surface image of 100X 70 vol% PEG grafted PHEMA network. Many of the

MVEC present possess a slightly diffused glow. This is due their penetration into the samples. In the top left corner. There is some evidence of surface adhesion, but this was minimal compared to the sample penetration.

131

Figure 5.10 Cross-section of 70% PEG grafted PHEMA networks (200X). There is extensive

evidence of tubule formation and EC elongation. The sizes of the tubules are smaller than the Matrigel® control, due to size limitations within the polymer network.

132

Figure 5.11 70vol% PEG grafted PHEMA sponges at 100X. Another image depicting

longer tubules. Bottom edge of the sponge was the surface that was seeded. The greater density of tubules near the outer rim of the sponge is most likely due to greater nutrient exchange with the media.

133

Figure 5.12 Surface image of 70 vol% pure PHEMA network. No endothelialization was

evident on this sample.

134

Figure 5.13 Cross sections of 70vol% pure PHEMA. In these networks vascularization of

the surface was evident. Little to no penetration was evident in these networks due to the small pore size and low porosity.

135

Figure 5.14 Surface of 80 vol% PEG-grafted PHEMA sponges. Some HMVECs are evident

on the surface. There was not extensive evidence from this analysis of HMVEC attachment and penetration.

136

Figure 5.15 Surface staining of 80 vol% pure PHEMA sponges. Extensive

endothelialization is evident. There is also evidence of HMVEC penetration from this analysis as well. No information about tubule formation was obtained.

137

Figure 5.16 Cross sectional view of 80 vol% PEG-grafted PHEMA networks. Due to the

interconnected structure of these polymers, there was an abundance of tubule formation. The greater porosity of these samples also allowed for greater nutrient transfer, which helped increase cellular density.

138

Figure 5.17 Cross section of 80 vol% pure PHEMA sponge. Penetration of HMVECs was

superficial; only 100-200µm deep. In this layer, the HMVEC density was extremely high, and fluorescence was too great to determine any tubule formation. These cross sections also revealed a large pore size disparity that was not evident in porosimetry and SEM.

139

Figure 5.18 Surface image of 90vol% PEG-grafted PHEMA Network. HMVEC were not

spread onto the surface, but had penetrated into the polymer network.

140

Figure 5.19 Surface image of 90 vol% pure PHEMA Network. HMVEC were not spread

onto the surface, but had penetrated into the polymer network.

141

Figure 5.20 90 vol% PEG grafted PHEMA cross section. The irregular shape is a result of

sectioning errors. 90 vol% samples contained a high degree of vascularization, and morphologically resembled the Matrigel® control.

142

Figure 5.21 Cross section of 90 vol% pure PHEMA cross section. Tubule formation is

abundant. Tubules formed along the pore surfaces.

143

Figure 5.22 Matrigel® coated 60 vol% PEG-grafted PHEMA sponges. The HMVEC layer

had detached from the edge of the section. This is probably due to the reduced adhesion of protein layer and PEG surface.

144

Figure 5.23 Penetration of HMVEC tubule into Matrigel® loaded 80vol% PEG grafted

polymer networks. This image depicts the polymer’s ability for vascularization. (250X Magnification)

145

5.4 Conclusions

It was possible to obtain information on the vascularization of porous

biomaterials from the in vitro studies. The physical properties of the polymer

networks (pore size, porosity) were the dominate factor in determining the extent

to which HMVECs can penetrate and express tubule formation. The PEG-grafted

samples seemed to reduce surface adhesion of the HMVECs on the samples that

were not porous enough for penetration. When the pore size and porosity was

larger enough, these pores could be occupied and filled with the basement

membrane secreted by the polymer networks. This demonstrates that the

basement does not need to strongly attach to the polymer pore surface for

HMVEC to migrate into hydrogel sponges and express tubule formation.

The unique lattice structure of the PEG grafted samples at 70 and 80 vol%

allowed for significant penetration of EC. The 70 vol% porosity was too small to

allow for long tubules. This may mean that these networks do not allow for

pathways for anastamosis to readily occur. Moreover, the tubule diameters were

typically smaller than those seen in the Matrigel®. At the 80 vol%, the tubules

were similar to the positive controls, and the length of the tubules was much

longer. The 80 vol% pure PHEMA samples analyzed possessed a highly bimodal

pore size distribution. This great variability resulted in a reduced interconnected

structure, which did not allow for tubule penetration. There was little difference

146

in the size and structure of the HMVEC tubule networks in the 90 vol% pure

PHEMA and PEG-grafted samples. The tubules formed in these networks were

almost identical to that of pure Matrigel®.

Hence, the hydrophilic materials presented here, when possessing

adequate pore size and porosity, are capable of inducing HMVEC to express the

tubule phenotype. When no competing cell lines exist, HMVEC can create

tubule networks in materials with pore sizes much smaller than seen in previous

in vivo studies.

147

List of References

[1] Anderson, J.M. ʺInflammatory Responce to Implantsʺ. Trans Am Soc Artif Intern Organs, 1988, 19: p. 101-107.


ʺEngineering the Tissue Whcih Encapsulates Sybcutaneous Implants I. Diffusion Propertiesʺ. J Biomed Mater Res, 1997, 37: p. 401-12.


ʺEngineering the Tissue Whcih Encapsulates Sybcutaneous Implants Ii. Plasma-Tissue Exchange Propertiesʺ. J Biomed Mater Res, 1998, 40: p. 586-597.

[4] Brauker, J.H., Carr-Brendel, V.E., Martinson, L.A., Crudele, J., and


[5] Mikos, A.G., Sarakinos, G., Lyman, M.D., Ingber, D.E., Vacanti, J.P., and

Langer, R. ʺPrevascularization of Porous Biodegradable Polymersʺ. Biotech Bioeng, 1993, 42: p. 716.

[6] Mooney, D.J. and Langer, R.S., Engineering Biomaterials for Tissue

Engineering: The 10-100 Micron Size Scale, in The Biomedical Engineering Handbook, Bronzino, JD, Editor. 1995, CRC Press: Boca Raton. p. 1609-1618.

[7] Sieminski, A.L. and Gooch, K.J. ʺBiomaterial-Microvasculature

Interactionsʺ. Biomaterials, 2000, 21: p. 2233-2241. [8] Bertolini, F., Mancuso, P., Gobbi, A., and Pruneri, G. ʺThe Thin Red Line:

Angiogenesis in Normal and Malignant Hematopoiesisʺ. Experimental Hematology, 2000, 28: p. 993-1000.

[9] Gerwins, P., Skoeldenberg, E. and Claesson-Welsh, L. ʺFunction of

Fibroblast Growth Factors and Vascular Endothelial Growth Factors and Thier Receptors in Angiogenesisʺ. Critical Reviews in Oncology/hematology, 2000, 34: p. 185-194.

148

[10] Nguyen, M., Akrell, J. and Jackson, C.J. ʺHuman Endothelial Gelatinases

and Angiogenesisʺ. Inter J Biochem cell Bio, 2001, 33: p. 960-970. [11] Thurston, G. ʺDistinct Actions of Vegf and Angiopoietin-1 on Blood Vessel

Permeabilityʺ. Annals of Biomedical Engineering, 2000, 28: p. S72-73. [12] Matsuda, T. and Kurumantani, H. ʺSurface Induced in Vitro Angiogenesis:

Surface Property Is a Determinant of Angiogenesisʺ. ASAIO Trans, 1990, 36(M565-8).

[13] Bachetti, T. and Morbidelli, L. ʺEndothelial Cells in Culture: A Model for

Studying Vascular Vunctionsʺ. Pharmacological Research, 2000, 42(1): p. 9-19.

[14] Jaffe, E.A., Nachman, R.L., Becker, C.G., and Minick, C.R. J Clin Invest,

1973, 52: p. 2745. [15] Kirkpatrick, C.J., Otto, M., Kooten, T.V., Krump, V., Kriegsmann, J., and

Bittenger, F. ʺEndothelial Cell Cultures as a Tool in Biomaterial Researchʺ. J Mat Sci, 1999, 10: p. 589-594.

[16] Ohmi, K., Kiyokawa, N., Takeda, T., and Fujimoto, J. ʺHuman

Microvascular Endothelial Cells Are Strongly Sensitive to Shiga Toxinsʺ. Biochem Biophys Res Com, 1998, 251: p. 137-141.

[17] Kirkpatrick, C.J., Wagner, M., Kohehler, H., Bittinger, F., Otto, M., and

Klein, C.L. J Mat Sci: Mat Med, 1997, 8: p. 131. [18] Kirkpatrick, C.J. and Dekker, A. in vitro Adv Biomater, 1992, 10: p. 31. [19] Weslowski, S.A., Fries, C.C., Karlson, K.E., Bakey, M.D., and Sawyer, P.N.

ʺPorosity: Primary Determinant of Ultimate Fate of Synthetic Vascular Grafts.ʺ Surgery, 1961, 50(1): p. 91.

[20] Harris, J.M., Poly(Ethylene Glycol) Chemistry, Biotechnical and Biomedical

Applications. 1992, New York: Plenum Press.

149

[21] Dalton, P.D., Flynn, L. and Shoichet, M.S. ʺManufacture of Poly(2-Hydroxyethyl Methacrylate-Co-Methyl Methacrylate) Hydrogel Tubes for Use as Nerve Uidance Channelsʺ. Biomaterials, 2002, 23(18): p. 3843-3851.

[22] Baish, J.W., Gazit, Y., Berk, D.A., Nozue, M., Baxter, L.T., and Jain, R.K. ʺRole of Tumor Vascular Architecture in Nutrient and Drug Delivery:An Invasion Percolation-Based Network Model.ʺ Microvascular Research, 1996, 51: p. 327-402.

[23] Fujiwara, M., Jin, E., Ghazizadeh, M., and Kawanami, O. ʺAn in Vitro

Model to Evaluate Regulatory Mechanisms of Antigen Expression by Normal Pulmonary Vessel Endothelial Cellsʺ. Microvascular Research, 2001, 61: p. 215-219.

150

CHAPTER 6: COMPUTER SIMULATIONS OF POROUS MATERIALS VASCULARIZATION

6.1 Introduction

From in vitro and in vivo experiments, it is known that pore size and

porosity has the greatest effect upon endothelial cell penetration and vascular

formation into porous materials [1, 2]. However, the exact mechanism of this

relationship is still unknown. It is not fully clear to what extent biological

signaling limits growth, or if sieving effects plays any part in endothelial cell

penetration [2]. Mathematical models and simulations can be used in

conjunction with experiments as a means of evaluating competing mechanistic

hypotheses. In this chapter, a Monte Carlo type cellular automaton is proposed

as a means for future evaluation of different mechanisms used to describe the

process of vascular growth into porous implants.

6.1.1 Computer Simulations

With the advent of the computer and continually increasing processing

power, two complementary mathematical simulation techniques, cellular

automata and Monte Carlo simulations, have become extremely useful in

describing complex systems [3, 4]. So much so, that Stephen Wolfram published

a book this year claiming that cellular automata will have the same effect on

science that Darwin’s Origin of Species and Newton’s Principia Mathamatica had

151

[5]. While this rather bold claim still remains to be determined, much can be

gained by their use.

In cellular automata, a sample space is divided into finite number of

steps [6]. Each cell within this space is then assigned its initial conditions from a

finite set of discrete possible states. For each subsequent time step, every cell’s

state is calculated based upon simple rules applicable to the system being

modeled.

For Monte Carlo simulations, a process is modeled by the inclusion of

random number generation [3]. The simplest of such schemes is the calculation

of the probability distribution of a coin. As consecutive coin tosses are tallied,

the probability distribution converges toward uniformity (assuming a fair coin).

When used in the field of statistical mechanics, random number generation in is

used in conjunction with cellular automata for the determination of state

changes.

6.1.2 Random-Walk Model

Random-walk describes the motion of a single particle in a sample space.

As the name suggests, the movement of the particle is unrestricted and

uniformly random. Hence in three dimensions, each cell neighboring the

moving particle has a 1 in 26 (3^3-1) of being occupied. Each simulation will

possess a unique highly varying path. However, when simulation times are

152

sufficiently long and with adequate number of replicates, the mean distance

traveled converges to the simple relationship

( )1

2 2R N N= (6.1)

where N is the number of time steps. This relationship directly depicts the

relationship of Brownian motion to that of classic diffusion [3].

An augmentation of random-walk is self-terminating random-walk.

Under this model, sites previously occupied by the moving particle will block

further movement. This model was initially used to describe the motion of

polymer chains in dilute media. Since back tracking is limited in this instance,

the relationship of distance traveled to that of mean end to end distance is

different. For 3 dimensions, the power term in equation (6.1) is 0.588.

6.1.3 Angiogenesis Modeling In order to consider the modeling of angiogenesis, it is important to

understand the key steps involved. For this reason, this section will provide a

summary of angiogenesis. A single endothelial cell starts to degrade the

basement membrane and moves in the direction of increasing growth factor

concentration [7, 8]. Its path is determined not only by the growth factor

concentration, but also the availability of adhesion proteins along the path. The

153

space behind the leading cell is subsequently filled by tailing endothelial cells

which form the tubule that will become a new capillary. When presented a

porous network, the vessels will either penetrate or surround the porous

implant. When materials have a small pore size, vessels will surround the

implant. As pore size and porosity increase, capillary penetration increases [9-

11].

6.1.4 Model Objectives The Chaplain and Anderson model mentioned in Section 2.6.1 explicitly

describes these features through a series of PDEs. While the descretized form of

this model can be used to explicitly simulate the motion of the lead endothelial

cell in an open system, the boundary conditions presented in porous material

results in a simulation box that is too computationally intensive for useful

analysis.

Hence, in this chapter we suggested the use of a self-terminating 3

dimensional random-walk as a first step in modeling the growth of capillaries

into the porous networks. The self avoiding form was selected to recreate the

existence of the tubule that is trailing this lead cell. Since pore size and porosity

are crucial to vessel growth, motion into randomly generated networks with

predetermined pore size and porosity was simulated. To compare surface

growth to surface penetration, a set simulations where surface sites are available

154

for movement was also performed. The rate of increase in the mean square end

to end distance and the rate of vessel entrapment are used as gauges for

comparing the ease of vessels to grow through the porous network.

6.2 Simulation Methods

All simulations were performed using Matlab Release 12 (Mathworks,

Natick, MA) on a 1.2Ghz Anthlon computer with 768Mb of DDR RAM. Matlab’s

default random number generator was used for all random number calls. Since

Matlab resets the random seed at each startup, a new seed based upon the

system clock was selected at the beginning of each simulation. This generator

has a theoretical limit of 21492 steps before the sequence repeats. The total number

of random number calls fell within 230. As such, error associated with random

number periodicity is not expected.

6.2.1 Porous Polymer Network Formation

The polymer network was represented as a 3 dimensional logical array

(130X130X130), where true represents the presence of polymer, and false no

polymer. To generate a porous network, spherical holes (diameter of 1, 3, 5, or

9), were sequentially removed from a solid (all true) polymer array. The location

of each hole was randomly selected, and all points described by the sphere were

set to zero. This process continued until the desired porosity (50, 60, 70, 75, 85,

and 90%) was achieved. Polymer spheres were allowed to overlap to allow for

155

pore size distribution and interconnection. Periodic boundary conditions were

used for the formation of the polymer networks. All indices within 2 units from

the edge of the polymer were set to true, in order to create “walled” boundary

conditions in the random walk simulations. Each network was saved with and

without a 3 unit gap between the wall and polymer. This gap was added to

represent external surface available for exterior vascularization. One polymer

network was generated for each pore size and porosity.

6.2.2 Porous Polymer Network Analysis

To obtain statistics on pore size distribution, a simple sequential array

scanning technique was used. Matlab’s FIND command was used to report all

array locations, excluding the wall and gap added, which contain a true value.

FIND’s output is the sequential indexing of “TRUE” array locations, where

indices are wrapped across row, column, and page. By subtracting each

sequential index reported by the previous index plus 1, the gap size between

each polymer is known. This process was repeated three times for each axis

(X,Y,Z), to determine if an orientation bias existed with this method.

6.2.3 Vessel Growth Simulations

Vessel growth was approximated as the random motion of a single

particle within a porous network. The size of the endothelial cell was set equal to

1 unit. Since vessel penetration was of primary concern, the starting point for

156

growth was set at location (3, 55, 65). Each of the 26 neighboring sites was

assigned equal probability. A random number is then used to determine which

neighboring site the vessel tip will move to. This new location is referenced to

the polymer array. If polymer does not exist, the move is successful; otherwise

the move is considered a failed attempt. Once a site is occupied, it is no longer

available. This was achieved by setting the polymer array index to a value of

true. Each simulation was performed for 10,000 time steps, and repeated 2000

times.

6.2.4 Simulation Data Analysis 6.2.4.1 End to End Distance

For every time step, the R2 distance from the starting point was calculated.

This distance was then averaged across all runs for each time step. The resulting

curve was plotted for each network. At extended time, the square of the mean

distance traveled vs. time becomes linear. The value of this rate of change was

determined for each simulation set to compare the effect of pore size and

porosity on the rate of penetration of growing vessels into porous networks.

These values were also compared to the case of no polymer present and polymer

with no pores and only gap space available.

157

6.2.4.2 Rate of Vessel Entrapment

Each simulation run was evaluated to determine if vessel entrapment

occurred. Entrapment was defined as the point at which no further movement of

the particle was possible. This occurs when all adjacent sites are occupied by

either polymer or the vessel. The time step at which entrapment occurred was

recorded for all runs of a specific simulation configuration. The cumulative

number of entrapped particles vs. time was then plotted to obtain an average

rate of vessel entrapment for all polymer networks.

6.3 Results and Discussion 6.3.1 Polymer Analysis

The average pore sizes for all polymer networks were larger than the pore

size desired, Figure 6.1. Also shown is the large standard deviations that were

present for all the networks. Both the pore size and standard deviation increased

as the polymer porosity increased. This increase in distribution and pore size

with increasing porosity is comparable to the experimental results in chapter 4.

A more informative view of the network structure is obtained by looking at

Figure 6.2 to Figure 6.9. Since the calculation of pore sizes was performed by a

sequential array addressing method, it was thought that there might be bias

based upon which axis the array indexed through. To address this concern, 3

array addressings per porous polymer were performed, one for each axis

158

orientation. Each of these orientations was plotted on the graphs, along with the

summation of the three biases. Since no significant difference existed for each

addressing method, it was concluded that the array addressing method

possessed no bias and the polymer networks were isotropic. The error in

average pore size is a result of the skewed nature of the pore size distribution.

Due to the pore generation technique, no pore sizes smaller than the desired pore

is possible. Since the most prominent peak for each polymer network was the

designated polymer pore size, the networks simulated were considered useful

representations of materials with known pore sizes.

6.3.2 Vessel Growth Simulations

Vessel growth was simulated as the random-walk of a single particle

within a porous matrix. The simulation conditions that were established were to

closely approximate the conditions that are found during typical porous implant

vascularization. For this reason, the leading particle (sprout tip) cannot reoccupy

locations. This is to simulate the forming capillary. Since vessel growth is

known to occur around and into porous networks, it was hypothesized that by

having exterior surface available for vessel growth, the relative amount of vessel

growth into and around the polymer block can be calculated. From simulations,

site occupancy was not significantly different than that of no polymer network.

Moreover, since the system is under confined boundaries this relative occupancy

159

would not be physically meaningful when compared to the unbounded iv vivo

experimental results. However, due to the presence of the gap, the moving

particle can enter the porous material at random locations across the network

surface. This results in data with a reduced likelihood of location specificity.

6.3.2.1 Simulation Limitations and Random Number Generation

In order to understand the relevance of the data obtained from these test

conditions, it is important to know the limitations presented by this simulation

scheme. Vessel growth occurs chemotaxicly toward an increase in growth factor.

The simulation presented here represents the case where there is a uniformly

distributed vessel growth factor, and only random motion determines growth.

While this is not physiologically meaningful, it has been shown in chapter 5 that

in vitro endothelial cell penetration can occur under such conditions.

The starting location of the moving particle was constant for all

simulations, and only one polymer network for each type of simulation was

evaluated. Further work should include the evaluation of several separately

generated polymer networks with the same pore size and porosity and with

multiple initial locations to further reduce location specific effects.

No computer random number generator is purely random. In order to

evaluate the usefulness of Matlab’s RAND function, two types of analyses were

performed. In these studies, a uniform distribution of 26 outcomes was desired.

160

For 10000 counts, a 20.3% deviation from uniformity existed. When the number

increased to 100000, the deviation dropped to only 6.33%. Both these values

possessed a standard error less than 0.01%. From a “parking lot” (Figure 6.10)

type of analysis for non randomness, it was found with 5000 points no global

trends were noticed. This lead to the conclusion that the RAND function was

adequate for the simulations presented.

6.3.2.2 R2 Distance Results

The slope of the R2 vs. time at long times is shown in Figure 6.11 and

Figure 6.12 for simulations with and without a surface gap present, respectively.

At low porosity, a hindered pathway exists, which results in a decrease in the

rate of displacement compared to the unhindered random-walk path (the solid

line). This effect was most prominent with a pore size of 3. This is most

probably due to the fact that the pores are too small for free movement within

each pore, but large enough to prevent interconnection at this porosity. Only a

pore size of 1 possessed a rate of change similar to the unbounded conditions.

This is due to the fact that at the same porosity, small pores possess more

interconnections than large pores. As the porosity increased, the rate of

displacement also increased. Pore size 1 possessed a rate of displacement greater

than the unhindered network at 70%. This is due to the pores directing particle

motion away from the starting point. This was evident at the intermediate

161

porosities for all pore sizes, with pore size of 9 being the smallest. This leads to

the conclusion that pore size has a great importance on the directing of vessel

growth. When the porosity increased to 90% the rate of change diminished to

values comparable to the unhindered case. The same trends were noted in the

simulations with a gap space. The dashed line in Figure 6.12 represents the rate

of change for a polymer block with no pores and only a surface space present,

which was shown to have little bearing on the results of the porous materials.

6.3.2.3 Rate of Entrapment The reduced rates at these porosities were theorized to be a result of vessel

entrapment in polymer networks. As mentioned, the 2000 replicates of each

simulation were evaluated to determine when and if entrapment of a vessel end

occurred. These numbers were serially summed, and then plotted vs. time step.

As shown in Figure 6.13, the rate of change was linear for almost all plots except

those with very high entrapment rates. It is believed that the deviation from

linearity is a result of not enough replicates to properly represent the rate of

entrapment. However, it may be that the large initial rate is due to local effects

in the simulation space. While it does not adequately describe all the curves

presented, a linear trend was calculated throughout the entire range of data as a

compromise when comparing the simulation data.

162

The entrapment rates for the simulations without and with surface space

are shown in Figure 6.14 and Figure 6.15, respectively. While the rate of

entrapment significantly decreases with increasing porosity, the smaller pores

possessed an overall greater tendency for entrapment compared to larger pores.

The exception is in simulations with a pore size of 1, where pore interconnection

and directed growth is at a maximum. It is believed that the larger pores allow

for more unhindered motion within the pore, which results in the lower

entrapment rates.

From the simulations with a surface space available, it was shown that at

50% porosity all large pore samples had a rate of entrapment greater than the

solid polymer block. This proves the hypothesis that the moving particle

becomes entrapped in the pores of the low porosity networks instead of moving

across the surface space. Hence this model does not currently adequately

describe the penetration vs. surface growth trends seen in vivo studies.

An interesting result is shown in the intermediate porosities and pore

sizes. The rate of entrapment and the rate of the mean square displacement were

not fully coupled. At 75% porosity, the 5 unit pore network had the greatest rate

of mean displacement, but only the second lowest rate of entrapment. This stems

from the fact that a network can both direct growth as well as entrap vessels. It

is believed that the rate of entrapment is a better indicator of vascular potential.

163

In order for a growing tubule to become a capillary it must encounter another

tubule. Networks that have to great of an entrapment will result in a lower

degree of vascularization, due to a decrease in probability for anastamosis to

occur.

164

0

10

20

30

40

50

60

70

80

90

45 55 65 75 85 95

Porosity (%)

Pore

Siz

e

Figure 6.1 Pore size vs. porosity for simulated polymer networks. Average pore size was

large due to the high variability of pore sizes present. () 1unit, ( ) 2 units, ( ) 4 units, and ( ) 8 units.

165

Figure 6.2 Histogram of Polymer Gap Size for 50% porosity 1 unit pore size polymer

network.

166

Figure 6.3 Histogram of Polymer Gap Size for 50% porosity, 3 unit pore size polymer network.

167

Figure 6.4 Histogram of Polymer Gap Size for 70% porosity 5 unit pore size polymer network.

168


169


170


network.

171


network.

172


network.

173

Figure 6.10 “Parking Lot” plot of 5000 random points obtained by the RAND function. There is

some evidence of local trends which is a result of the non-randomness of the generator. However, these orientations were not global through the domain, and considered not significant for the purposes of this study.

174

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

45 55 65 75 85 95

Porosity(%)

Rat

e of

Cha

nge

(R /

tim

e st

ep)

2

Figure 6.11 The rate of change (slope) of the square mean displacement as a function of porosity

at long time steps for pore sizes ( ) 1, ( ) 3, ( ) 5, and ( ) 9. No surface gap was present during these simulations. Line represents the rate of change of the moving particle in unhindered conditions. All error bars and line thickness represent 99.99% confidence limits.

175

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

45 55 65 75 85 95Porosity(%)

Rat

e of

Cha

nge

(R /

tim

e st

ep)

2

Figure 6.12 The rate of change (slope) of the square mean displacement as a function of porosity at long time steps for pore sizes ( ) 1, ( ) 3, ( ) 5, and ( ) 9. A surface gap was present during these simulations. Line represents the rate of change of the moving particle in unhindered conditions. All error bars and line thickness represent 99.99% confidence limits.

176

0

200

400

600

800

1000

1200

1400

1600

1800

2000

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000

Time Step

Free

Mov

ing

Part

icle

s

Figure 6.13 Number of free moving particles vs. time step for all simulations performed. The

important point to note is that the majority of the simulations possessed a linear rate of entrapment. The deviation of the skewed lines is thought to be a result of the rate being too rapid for the number of simulations performed to adequately represent.

177

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

45 55 65 75 85 95

Network Porosity(%)

Rat

e of

Ent

rapm

ent(P

artic

les/

time

step

)

XXX

Figure 6.14 The rate of entrapment as a function of porosity for pore sizes ( ) 1, ( ) 3, ( )

5, and ( ) 9. No surface gap was present during these simulations. The solid represents the rate of entrapment with no polymer present.

178

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

45 55 65 75 85 95

Network Porosity(%)

Rat

e of

Ent

rapm

ent(P

artic

les/

time

step

)

XXX

Figure 6.15 The rate of entrapment as a function of porosity for pore sizes ( ) 1, ( ) 3, ( )

5, and ( ) 9. A surface gap was present during these simulations. The solid and dashed lines represent the rate of entrapment with no polymer present and a solid polymer with no porosity, respectively.

179

6.4 Conclusions

Monte Carlo based cellular automata can be used to simulate the growth

of vessels into porous networks. In these initial studies, the effects of pore size

and porosity on the extent of implant vascularization have been replicated. This

leads to the conclusion that there are percolation effects that play a part in

implant vascularization. From the results shown, pore size and porosity control

both the rate of mean square displacement and the rate of entrapment. While

related, these two processes are shown to be somewhat decoupled, and networks

with the fastest rate of displacement do not always have the slowest rate of

entrapment. The networks with increased entrapment are thought to possess a

lower probability of vessel anastamosis. This would result in a decreased extent

of vascularization. It is hoped through the addition of chemotaxic attract factors,

material surface properties, and multiple propagating vessels, that this model

will be able to fully replicate the processes involved in implant vascularization.

Such studies would lead to a better understanding of the limitations of porous

networks to promote vascular penetration, and to better designs of

vascularizable implant interfaces.

180

List of References [1] Brauker, J.H., Carr-Brendel, V.E., Martinson, L.A., Crudele, J., and Johnston, W.D. ʺNeovascularization of Synthetic Membranes Directed by Membrane Microarchitectureʺ. J Bio Mat Res, 1995, 29: p. 1517- 1524. [2] Sieminski, A.L. and Gooch, K.J. ʺBiomaterial-Microvasculature

Interactionsʺ. Biomaterials, 2000, 21: p. 2233-2241. [3] Landau, D.P. and Binder, K., A Guide to Monte Carlo Simulations in

Statistical Physics. 2000, Cambridge, UK: Cambridge Press. [4] Talia, D. and Sloot, P. ʺCellular Automata:Promise and Prospects in

Computational Scienceʺ. Future generation computer systems, 1999, 16: p. v-vii.

[5] Wolfram, S., A New Kind of Science. 2002: Wolfram Media, Inc. [6] Markus, M., Boehm, D. and Schmick, M. ʺSimulation of Vessel

Morphogenesis Using Cellular Automataʺ. Mathmatical Biosciences, 1999, 156: p. 191-206.

[7] Bertolini, F., Mancuso, P., Gobbi, A., and Pruneri, G. ʺThe Thin Red Line:

Angiogenesis in Normal and Malignant Hematopoiesisʺ. Experimental Hematology, 2000, 28: p. 993-1000.

[8] Hanahan, D. ʺSignaling Vascular Morphogenesis and Maintenanceʺ.

Sience:Cell Biology, 1997, 277(5322): p. 48-50. [9] Shwarkawy, A.A., Klitzman, B., Truskey, G.A., and Reichert, W.M.

ʺEngineering the Tissue Whcih Encapsulates Sybcutaneous Implants I. Diffusion Propertiesʺ. J Biomed Mater Res, 1997, 37: p. 401-12.


ʺEngineering the Tissue Whcih Encapsulates Sybcutaneous Implants Iii. Effective Tissue Response Timesʺ. J Biomed Mater Res, 1998, 40: p. 598-605.

181

[11] Shwarkawy, A.A., Klitzman, B., Truskey, G.A., and Reichert, W.M. ʺEngineering the Tissue Whcih Encapsulates Sybcutaneous Implants Ii. Plasma-Tissue Exchange Propertiesʺ. J Biomed Mater Res, 1998, 40: p. 586-597.

182

CHAPTER 7: IN VIVO IMPLANTABLE INSULIN DELIVERY

7.1 Introduction Diabetes is a disease in which the pancreas secretes effectively no insulin

(TYPE I) or secretes marginal amounts of insulin and/or the body develops a

resistance to insulin (TYPE II). In both instances, diabetes is characterized by an

elevated level of glucose in the blood stream as a result of the lack of insulin

control. It was reported in 1997 that there are currently 20 million patients

diagnosed with Diabetes. Of these, 1-2 million of them are classified type I,

insulin dependant [1]. The most common treatment for this disease is daily

injections of insulin as well as diet regulation to help maintain normal blood

glucose levels. If supplementary insulin is not delivered and glucose control is

not maintained, many secondary complications can occur. These include

retinopathy (loss of vision), neuropathy (loss of nerve function), nephropathy

(loss of kidney function), and if treatment is postponed long enough, coma and

death [2-6].

In a landmark study called the Diabetes Control and Complications Trial,

it was found that with increasingly tighter control on blood glucose levels,

patients can almost completely avoid the occurrence of secondary complications

[2]. The primary method available to achieve tighter controls on blood glucose

levels is though multiple daily injections (3 or more). Due to the extreme

183

discomfort, patient compliance is highly variable. A better approach to insulin

delivery would be an implantable device that can monitor real time glucose

levels, and administer insulin as needed. This device is known as the artificial

pancreas [7, 8].

Most attempts in developing the artificial pancreas have focused on an

implantable insulin pump, which can deliver basal levels of insulin as well as

bolus delivery during meals. These devices have had reasonable success, but

their life spans are severely shortened (less than 1 year) due to three principle

problems; pump failure (2.5%), surgical technique (8%), and fibrous

encapsulation of the catheter port (13%) [6, 9-11]. Further studies demonstrated

that even frequent flushing of the catheter port did not prevent eventual tissue

encapsulation [9]. From the International Study Group on Implantable Insulin

Delivery Devices (ISGIID), it was stated that 56% of all problems associated with

the implantable insulin pump were catheter related [12]. All explanted non-

functional intraperitoneal catheters possessed varying degrees of tissue

encapsulation [13-16]. For these reason, it has been said that the catheter port is

the weakest link in the implantable insulin pump, and should be the focus of any

future research [17-19].

In this work, the effectiveness of a macroporous PHEMA hydrogel as a

tissue/implant interfacial drug delivery device was evaluated. The model system

184

presented is that of a catheter port coated in the macroporous PHEMA hydrogel.

By allowing permanent vascular ingrowth, it is suggested that this coating will

circumvent fibrous encapsulation. Furthermore, it is hypothesized that systemic

uptake rates will increase due to an increased presence of vascular tissue

surrounding the catheter port.

7.2 Materials and Methods 7.2.1 Catheter Assembly Sponges were synthesized as mention in Section 4.2.1. Only the 75 vol%

non PEG-grafted PHEMA sponges were used as catheter coatings. The assembly

of the catheter device is depicted in Figure 1b. Polypropylene catheter tubing

(20-gauge) was cut into 20 cm lengths. One end of the tubing was lanced with a

20-gauge needle to create 20 evenly spaced holes over a 1.5 centimeter length.

This end was inserted axially into the sponge. The assembly was dried under

vacuum to collapse the sponge tip onto the tubing. The dried hydrogel was

permanently fixed onto the tubing by the addition of a medical-grade silicone

adhesive (Medtronic, Inc. Mpls., MN) at the insertion point.

7.2.2 In Vivo Experiments

Sprague-Dawley rats (n=4) were anesthetized using isoflurane under

spontaneous ventilation. The abdomen of the rats were shaved and cleaned with

an iodine wash for laparotomy. The surgical procedures were performed under

185

sterile conditions. Each rat was implanted with 2 catheters, one subcutaneously

and one intraperitoneally. During the operation and recovery period, the rats

were administered 100% oxygen. Rats were kept 2 per cage, and allowed free

access to food and water. At 5 months, rats were anesthetized again and the end

of the catheter was exposed and connected to a syringe pump (Model 22,

Harvard Apparatus, Holliston, MA). Each rat was infused with human insulin

via one of the implants to determine the glucose response and insulin absorption

profile. Infusions were set at 10 milliU/kg/min with infusion rates of 60 µL/hr.

Blood glucose concentrations were determined using a Hemocue Analyzer

(Hemocue, Agelholm, Sweden), and human insulin levels were determined

using ELISA (ALPCO, Windham, NH).

After infusion studies, the animals were sacrificed and the catheter tips

were explanted with excision of a 1 cm section of surrounding tissue. The tissue

blocks were immediately submerged in 10% formalin for histological evaluation.

Tissue specimens were then embedded in paraffin and 7-10µm sections were

prepared for H and E stains.

186

Figure 7.1 Depiction of perforated catheter tubing inserted axially into the hydrogel sponge. A silicone adhesive was used to permanently fix tubing assembly.

187

7.3 Results and Discussion 7.3.1 In Vivo Insulin Infusion Kinetics For the animal studies pure PHEMA sonicated sponges with 75% water

content were used, which possesses a pore size of 8 µm. Hence, it does not

possess the ideal porous morphology for implant vascularization. Under

isoflurane general anesthesia the catheterʹs proximal end was exteriorized and

connected to a microinfusion pump. Blood insulin levels increased rapidly with

5 minutes post infusion and remained elevated for 30 min for both mesenteric

and subcutaneous infusions (Figure 7.3). The mean plasma insulin concentrations

rose to near 300 µIU/mL, which was beyond the standard range (3-200µIU/mL)

of the assay. Blood glucose concentrations decreased in proportion to increasing

insulin concentrations (Figure 7.2). The high baseline glucose levels are

attributable to a combination of the animals not having been fasted prior to the

experiment, and the effects of isoflurane anesthesia (suppression of insulin

production). Glucose infusions were not needed under this experimental

protocol, which was of short duration and involved non-survival surgery.

7.3.2 Histological Evaluation of Catheter Sponge Explants

On explants of both subcutaneous and intraperitoneal catheters, gross

examination of surrounding tissues appeared normal with no evidence of

inflammation or encapsulation. Histological sections of implants and

188

surrounding tissues (Figure 7.4 and Figure 7.5) supported these observations

revealing little to no lymphocyte infiltration peripheral to the hydrogel and only

a thin (10-35 µm) connective tissue boundary (CTB) adjacent to the hydrogel.

The CTB was richly vascularized but with the methods used it was not possible

to ascertain the extent of vascularization into the hydrogel scaffold. This

surrounding tissue is similar to other porous implants in the 5-10µm range [20].

The tissue immediately adjacent to the CTB therefore provided a large surface

area for insulin diffusion into the systemic (subcutaneous implant) or portal

(mesenteric implant) circulation. Numerous capillaries of varying sizes can be

seen as well as capillaries along the CTB.

189

Figure 7.2 Systemic glucose response following infusion of human insulin from an external pump 5 months post implantation.

190

Figure 7.3 Systemic human insulin concentration following infusion of human insulin from an external pump 5 months post implantation.

191

(a) (b)

Figure 7.4 Histological slides of mesenteric implant, (a) 100X, (b) 200X.

192

(a) (b)

Figure 7.5 Histological slides of subcutaneous implant, (c) 100X, (d) 200X.

193

7.4 Conclusions Catheters coated in the macroporous hydrogel exhibited no dense fibrous

encapsulation 5 months post implantation. The tissue surrounding the implant

was highly vascularized, and demonstrated the ability to rapidly respond to

changes in insulin infusion. This data proves the hypothesis that an increased

vasculature surrounding the devices improves the uptake response. The

coatings studied, while unoptimized, were still capable of enhancing the life span

of the implanted catheters. Based off of this evidence, it is believed that these

materials would make excellent coatings for implants, where long-term solute

exchange with the circulatory system is required. This is not limited to just

implantable drug delivery devices, but can extend to biosensors and tissue

engineering applications as well.

194

List of References [1] CDC ʺReport of the Expert Committee on the Diagnosis and Classification

of Diabetes Mellitusʺ. Diabetes Care, 1997, 20(7): p. 1183-97. [2] Diabetes Control and Complications Trial Research Group ʺThe Effect of

Intensive Treatment of Diabetes on the Development and Progression of Long-Term Complications in Insulin-Dependant Diabetes Mellitusʺ. N Engl J Med, 1993, 329: p. 977-986.

[3] Hanssen, K.F., Bangstad, H.J., Brinchmann-Hansen, D., and Dah

Joegensen, K. ʺBlood Glucose Control and Diabetic Microvascular Complications: Long-Term Effects of near-nor Moglycemiaʺ. Diabetic Med, 1992, 9: p. 697-705.

[4] Hepp, K.D. ʺImplantable Insulin Pumps and Metabolic Controlʺ.

Diabetologia, 1994, 37(Suppl 2): p. S108-S111. [5] Tamborlane, W.V., Sherwin, R.S., Genel, M., and Felig, P. ʺReduction to

Normal of Plasma Glucose in Juvenile Diabetes by Subcutaneous Administration of Insulin with a Portable Infusion Pumpʺ. N Engl J Med, 1979, 300(11): p. 574-580.

[6] Jaremko, J. and Rorstad, O. ʺAdvances toward the Implantable Artificial

Pancreas for the Treatment of Diabetesʺ. Diabetes Care, 1998, 21(3): p. 444-450.

[7] Ronel, S.H., DʹAndrea, M.J., Hashiguchi, H., Klomp, G.F., and Dobelle,

W.H. ʺMacroporous Hydrogel Membranes for a Hybrid Artificial Pancreas. I. Synthesis and Chamber Fabricationʺ. J Bio Mat Res, 1983, 17: p. 855-864.

[8] Klomp, G.F., Hashiguchi, H., ursell, P.C., Takeda, Y., Taguchi, T., and

Dobelle, W.H. ʺMacroporous Hydrogel Membranes for a Hybrid Artificial Pancreas. Ii. Biocompatibilityʺ. J Bio Mat Res, 1983, 17: p. 865-871.

195

[9] Scavini, M., Galli, L., Reich, S., Eaton, R.P., Charles, M.A., and Dunn, F.L. ʺCatheter Survival During Long-Term Insulin Therapy with an Implanted Programmable Pump.ʺ Diaetes Care, 1997, 20: p. 610-613.

[10] Selam, J.L., Micossi, P., Dunn, F.L., and Nathan, D.M. ʺClinical Trial of

Programmable Implantable Insulin Pump for Type I Diabetesʺ. Diaetes Care, 1992, 15: p. 877-885.

[11] Thompson, J.S., Duckworth, W.C. and Saudek, C.D. ʺSurgical Experience

with Implantable Insulin Pumpsʺ. Amer J Surg, 1998, 176: p. 622-626. [12] Knatterud, G. and Fisher, M. ʺReport from the International Study Group

on Implanable Insulin Delivery Decivesʺ. ASAIO Transactions, 1988, 34(2): p. 148-149.

[13] Saudek, C.D., Selam, J.L., Pitt, H.A., Waxman, K., Fischell, R.E., and

Carles, M.A. ʺA Preliminary Insulin Trial with the Programmable Implantable Medication System for Insulin Deliveryʺ. N Engl J Med, 1989, 321(574-79).

[14] Selam, J.L. Diabetic Med, 1988, 5(8): p. 724-733. [15] Kritz, H., najemnik, C., Hagmuller, G., Loddolter, S., Olbert, F., Mustbek,

A., Dench, H., and Irsigler, K., Long Term Results Using Different Routes of Infusion, in Diabetes Treatment with Implantable Insulin Infusion System, Kritz, H and Lovett, R, Editors. 1983, Urban and Schwartzenberg: Muenich. p. 82-102.

[16] Selam, J.L., Giraud, P., Mirouze, J., and Saeidi, S. Diabetes Care, 1985, 8: p.

34-38. [17] Selam, J.L. and Charles, M.A. Diabetes Care, 1990, 13: p. 9. [18] Renard, E., Balder, P., Picot, M.C., Jacques-Apostol, D., Lauton, D.,

Costalat, G., Bringer, J., and Jaffiol, C. ʺCatheter Complications Associated with Implantable Systems for Peritoneal Insulin Deliveryʺ. Diabetes Care, 1995, 18(3): p. 300-306.

196

[19] Von Recum, R.H. ʺApplications and Failure Modes of Percutaneous Devices:A Reviewʺ. J Bio Mat Res, 1984, 18(323-336).

[20] Brauker, J.H., Carr-Brendel, V.E., Martinson, L.A., Crudele, J., and


197

CHAPTER 8: RECOMMENDATIONS

As mentioned in chapter 3, the overall goal of this work can be divided

into 4 main components. In the materials synthesis and characterization section,

PHEMA sponges were synthesized with varying pore size and pore morphology.

By adopting established in vitro angiogenesis models for biomaterial studies, the

PHEMA sponges were assessed for their ability to support neovascularization.

As a means of understanding the dependence of vascularization on pore size, a

random walk in a porous network model was developed. Finally, to test the

thesis’ hypothesis, long term in vivo drug delivery studies were performed. As

such, it was verified that macroporous hydrogels prevent the fibrous

encapsulation of an implant, and improve the long term drug delivery response

of a implanted catheter. While the goals that were originally set were met, there

are still some questions surrounding the ultimate extent of sponge

vascularization and long term functioning. The remainder of this section

includes a set of recommendations for future research in helping overcome some

of the existing limitations.

8.1 Network Synthesis

While the sponges generated were capable of cellular invasion, it would

be beneficial to perform a more detailed analysis of reaction conditions upon the

198

sponge structure. Ideally, synthesis conditions which can increase pore-

interconnection and pore size, while maintaining soft tissue-like mechanical

properties, should be found. Toward this goal, a ternary phase diagram

depicting the scaffold’s pore size and structure as a function of water, PEG, and

HEMA would be of keen interest. From this data, it should be possible to

determine at what exact PEG content the structure changes from a sintered

microsphere to the lattice arrangements shown in Chapter 4.

Due to the inverse temperature behavior on PEG and PHEMA’s water

solubility, a study of different reaction temperature can be performed to

determine if decreased solubility will result in a denser reacting polymer phase.

This increase in density will translate into a larger pore size and porosity with

increased mechanical strength. To this end, increased salt content at these

elevated reaction temperatures may play a synergistic role in sponge formation.

8.2 Protein Functionalization of Sponge Pore Surface

Current implants of PHEMA sponges have relied upon secondary

signaling of vascular ingrowth. The primary source of this signaling is the initial

infiltration of macrophages into the scaffold which release pro-angiogenic

growth factors as a result of hypoxia. As a way of directly regulating the vessel

199

ingrowth, signals such as VEGF and ANG2 can be attached to the sponge surface

across the terminal hydroxyl ends of PEG.

The two most probable schemes for protein attachment are either the

conversion of terminal hydroxyl to a succinate carbonate (SC) or a maleimide

(MAL) group. The SC will form a permanent bond with any free nucleophilic

amine on a protein. However, it is also readily hydrolyzed when in the presence

of water. For this reason, the SC group will most likely not survive the synthesis

of the sponge, and this functionalization of the polymer must occur post network

formation. The MAL group, on the other hand, is fairly stable in water, but

reacts with free sulfhydryls on proteins. Unfortunately, the most proteins

cysteine groups are not freely available. It is recommended here that both

techniques be attempted to determine which is best suited for this application.

The functionalization step can be performed at either pre or post

polymerization. Both techniques offer advantages and disadvantages, however

it is the belief of the author that post polymerization functionalization will be the

more successful approach. Pre-polymerization reactions will most likely result

in high functionalization yields. However, these functional groups will most

likely be lost during the severe conditions of polymerization (free radicals, high

temperature, water, etc). Also not all of the PEG chains are present on the

200

polymer pore surface, further reducing functional group availability for protein

attachment.

With post polymerization functionalization, only surface hydroxyl groups

participate in the reaction, hence higher effective yields are likely. Also, since the

reaction is on a solid substrate, purification is greatly simplified. The difficulties

with this technique are with proper solvent selection and flow of reacting solvent

through the polymer sponge. The solvent must be selected, such that it is soluble

for the reactants but does not readily swell the sponge. Also, sponges must be

lyophilized (unless a proper solvent exchange pair is selected) prior to

immersion in the non-swelling solvent to help maintain pore liquid exposure.

Finally, a reactor that allows for fluid flow through the polymer scaffold is

crucial to maintaining high enough functionalization yields, as well as

attachment yields in the subsequent protein binding steps.

8.3 In Vitro Growth Factor Selection

Using the in vitro techniques established in this work, a detailed

evaluation in what growth factor signals should be attached to the sponges can

be performed. The best method to evaluate this is by using a growth factor

reduced medium, and install depots of growth factor combinations into the

sponges. A depot of growth factor can also be used in comparison with a

201

homogenous growth factor concentration (soaking sponges in growth factor

solution). Here, a depot is hypothesized to possess a growth factor gradient

which should provide a more directed growth of tubules within the polymer

sponge.

While many growth factor combinations can be attempted, it is the belief

of the author that the simplest strategy is to use VEGF in conjunction with

ANG2. As mentioned in Chapter 2, these pairs are known to be instrumental in

turning on angiogenesis. This process will continue until ANG2 is eliminated

and ANG1 replaces it. At this point, the vessels mature into fully formed

capillaries. This information may lead to a concept the author feels is most

intriguing, programmed factor release. In this system, ANG2 would be attached

through an ester linkage which can be hydrolyzed, but VEGF is bound with the

more permanent carbonate linkage. In this way, over time ANG2 is lost from the

system, resulting in the programmed maturation of the budding tubules. The

long term presence of VEGF is desirable since it helps increase vessel

permeability, which is an ideal circumstance in chemical communication.

202

Vita

Thomas D. Dziubla was born at Mercy hospital in Chicago, IL on August 3,

1975. He graduated with honors from Merrillville High School in Merrillville, IN

in 1993. After graduation, he enrolled in the Chemical Engineering program at

Purdue University. Since the age of eight, he had aspirations to perform medical

research. After several co-op experiences in the environmental engineering field,

he followed these dreams and joined the research group of Dr. Nicholas Peppas

as an undergraduate researcher. After receiving his honors B.S. in Chemical

Engineering in 1998, he enrolled in the graduate Chemical Engineering program

at Drexel University to work this Dr. Anthony Lowman. During his time there,

he received several graduate student research awards, including Sigma Xi

research award. He has coauthored numerous papers, book chapters, and

abstracts and has presented papers at national and international scientific

meetings. He received his Ph.D. in Chemical Engineering from Drexel

University in November of 2002. Receiving a National Research Service Award

from the National Institute of Health, he accepted a post doctoral position at the

University of Pennsylvania School of Medicine.

macroporous hydrogels as vascularizable soft tissue

Documents