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
List of References ............................................................................................. 147
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 Results and Discussion........................................................................ 157
6.3.1 Polymer Analysis.................................................................... 157
6.3.2 Vessel Growth Simulations................................................... 158
6.4 Conclusions........................................................................................... 179
List of References ............................................................................................. 180
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
7.3 Results and Discussion........................................................................ 187
ix
7.3.1 In Vivo Insulin Infusion Kinetics .......................................... 187
7.3.2 Histological Evaluation of Catheter Sponge Explants ...... 187
7.4 Conclusions........................................................................................... 193
List of References ............................................................................................. 194
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
4.21 SEM Micrographs of (a) PEG-grafted PHEMA and (b) pure PHEMA polymer sponges using 80 vol% water. .......................................................... 95
4.22 SEM Micrographs of (a) PEG-grafted PHEMA and (b) pure PHEMA polymer sponges using 90 vol% water. .......................................................... 96
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
6.4 Histogram of Polymer Gap Size for 70% porosity 5 unit pore size polymer network.............................................................................................. 167
6.5 Histogram of Polymer Gap Size for 50% porosity 9 unit pore size polymer network.............................................................................................. 168
6.6 Histogram of Polymer Gap Size for 90% porosity 1 unit pore size polymer network.............................................................................................. 169
6.7 Histogram of Polymer Gap Size for 90% porosity 3 unit pore size polymer network.............................................................................................. 170
xvii
6.8 Histogram of Polymer Gap Size for 90% porosity 5 unit pore size polymer network.............................................................................................. 171
6.9 Histogram of Polymer Gap Size for 90% porosity 9 unit pore size polymer network.............................................................................................. 172
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.
[23] Shwarkawy, A.A., Klitzman, B., Truskey, G.A., and Reichert, W.M.
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.
[25] Shwarkawy, A.A., Klitzman, B., Truskey, G.A., and Reichert, W.M.
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
Fractional Solvent Content in Monomer Solution
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
Fractional Solvent Content in Monomer Solution
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)
Figure 4.21 SEM Micrographs of (a) PEG-grafted PHEMA and (b) pure PHEMA polymer sponges using 80 vol% water.
96
(a)
(b)
Figure 4.22 SEM Micrographs of (a) PEG-grafted PHEMA and (b) pure PHEMA polymer sponges using 90 vol% water.
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.
5.3.2.2 Immunohistology of 70 vol% PHEMA Hydrogel Sponges
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).
5.3.2.3 Immunohistology of 80 vol% PHEMA Hydrogel Sponges
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.
5.3.2.4 Immunohistology of 90 vol% PHEMA Hydrogel Sponges
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.
[2] 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.
[3] 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.
[4] 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.
[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
Figure 6.5 Histogram of Polymer Gap Size for 50% porosity 9 unit pore size polymer network.
169
Figure 6.6 Histogram of Polymer Gap Size for 90% porosity 1 unit pore size polymer network.
170
Figure 6.7 Histogram of Polymer Gap Size for 90% porosity 3 unit pore size polymer
network.
171
Figure 6.8 Histogram of Polymer Gap Size for 90% porosity 5 unit pore size polymer
network.
172
Figure 6.9 Histogram of Polymer Gap Size for 90% porosity 9 unit pore size polymer
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.
[10] Shwarkawy, A.A., Klitzman, B., Truskey, G.A., and Reichert, W.M.
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
Johnston, W.D. ʺNeovascularization of Synthetic Membranes Directed by Membrane Microarchitectureʺ. J Bio Mat Res, 1995, 29: p. 1517-1524.
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.