University of Groningen

Biomedical polyurethane networksBruin, Peter

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:2006

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Bruin, P. (2006). Biomedical polyurethane networks. s.n.

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 15-02-2021

Rijksuniversiteit Groningen

BIOMEDICAL POLYURETHANE NETWORKS

ter verkrijging van het doctoraat in de

Wiskunde en Natuurwetenschappen

aan de Rijksuniversiteit Groningen

op gezag van de

Rector Magnificus Dr. S.K. Kuipers

in het openbaar te verdedigen op

vrijdag 6 november 1992

des namiddags te 4.00 uur

door

PETER BRUIN

geboren op 22 maart 1963

te Hoorn

Promotor: Prof. Dr. A.J. Pennings

Dear Sir or Madam will you read my book

it took me years to write will you take a look

(from "Paperback writer" by the Beatles)

Dankwoord

Iedereen die op enigerlei wijze heeft bijgedragen aan de totstandkoming

van dit proefschrift wil ik hiervoor bedanken, met name:

mijn promotor prof. dr. A.J. Pennings voor de geboden mogelijkheid om

onder zijn deskundige toezicht dit promotieonderzoek uit te voeren,

prof. dr. G. Challa, prof. dr. J.H. Teuben en prof. dr. B. Witholt voor de

bereidwilligheid om zitting te willen nemen in mijn promotiecommissie en

het manuscript te beoordelen,

Annemarie Brummelhuis, Andries Hanzen, Henk Hoppen, Koen Knol, Hendrik

Luttikhedde, Edwin Meeuwsen, Henk-Jaap Meijer, Joke Smedinga, Gert-Jan

Veenstra, Norbert Wolberink, Geartsje Zondervan voor de waardevolle

bijdragen aan het experimentele werk,

Adams Verweij, Harry Nijland (electronen microscopie en fotografie), Anne

Appeldoorn (instrumentmakerij), Henk Knol (glasblazerij), Harm Draaijer &

Jan Ebels (microanalyse afdeling) voor de practische assistentie en

technische ondersteuning,

dr. Jan Willem Leenslag zonder wiens werk dit proefschrift er heel anders

uitgezien zou hebben,

dr. Marcel Jonkman en drs. Jean Coenen voor de prettige samenwerking op

het terrein van de wondbedekking,

dr. Berend van der Lei voor zijn bijdrage aan het vaatprothese onderzoek,

drs. Pek van Andel, drs. Gerard van der Veen en prof. dr. J.G.F. Worst

voor het enthousiasmeren voor de oogheelkundige toepassing van polymere

materialen,

en verder (zonder anderen te kort te willen doen) Machiel Bos, Jacqueline

de Groot, Atze Nijenhuis, Jan "Les" Paul Penning , alle "anonieme"

collega's, studenten, secretaresses, andere medewerkers en staf van de

vakgroep polymeerchemie voor de hulp, discussies en sfeer tijdens mijn

verblijf in jullie midden.

CONTENTS

Chapter 1 Introduction 1

Chapter 2 Design and synthesis of biodegradable 17

poly(ester-urethane) elastomer networks composed of

non-toxic building blocks

Chapter 3 Biodegradable lysine diisocyanate-based

poly(glyco1ide-co-E-capro1actone)urethane network

in artificial skin

Chapter 4 A new porous polyetherurethane wound covering

Chapter 5 A two-ply artificial blood vessel of polyurethane and 53

poly(L-lactide)

Chapter 6 Autoclavable highly cross-linked polyurethane networks 77

in ophtalmology

Summary

Samenvatting

Chapter 1

Introduction

Polymer networks

Real cross-linked polymer networks always deviate from ideal, perfect

networks, which are defined as random, homogeneous collections of

(Gaussian) chains between multifunctional junction points (cross-links)

under the condition that all functionalities of the junction points have

reacted with the ends of all and different chains (1,2). In other words,

the ideal network entirely consists of elastically effective chains,

meaning chains connecting two neighbouring cross-links in the network,

able to transfer a retractive force throughout the material if subjected

to an elastic deformation.

Chompff stated that real polymer networks consist of inhomogeneous

structures because if polymers form homogeneous continua their strengths

would theoretically be about one hundred times higher than is observed

experimentally ( 3 ) . The degree of inhomogeneity of polymer networks

depends on the way in which the network has been formed and the ultimate

strength, for example, is sensitively dependent on such defects (4).

Besides inhomogeneity in cross-link distribution there are other network

imperfections which may be introduced upon network formation: network

defects, like dangling or pendant chain, i.e., a chain attached to the

network at only one of its ends (unreacted functional endgroups, for

instance), elastically inactive loops (as a result of intramolecular

cross-linking), chain entanglements (51, and heterogeneity due to phase

separation ( 1 1.

Polymeric, chemically cross-linked networks can be formed in three

different ways:

-cross-linking (vulcanization) of existing linear polymers.

-chain cross-linking (co)polymerization.

-stepreaction of small molecules (multifunctional monomers and/or

prepolymers), all of which are reactive at the same time.

Polymer networks need not to be formed exclusively by chemical pathways

leading to permanent networks. Cross-linking by physical aggregation of

polymer chains also results in network structures. Examples of polymer

gels containing physical cross-links include microcrystalline polymers,

ionomers, chelation polymers, blockcopolymers (thermoplastic elastomers,

like polystyrene-butadiene triblockcopolymers and segmented

polyurethanes), stereocomplexes. In the case of physically cross-linked

polymer networks the cross-links are not permanent; and the physical

aggregation is often (thermo)reversible (2.10).

Cross-linking of linear polymers

Polymer chains can be cross-linked by chemically joining different

primary, linear polymer molecules. The techniques generally used to

introduce cross-links are peroxide decomposition, high-energy irradiation

and sulfur addition to skeletal or side-chain double bonds (2.6.7). All of

these cross-link methods are statistical processes. Cross-links are

introduced in a highly random manner resulting in polymeric networks

having a rather undefined network topology (2,s).

The minimum number of dangling ends is inversely proportional to the

number-average molecular weight of the starting polymer ( 7 , s ) . Due to the

finite molecular weight, dangling ends will always be present in the final

network. Networks cross-linked by means of high-energy radiation may

contain even higher concentrations of dangling chain ends arising from

chain-scission occurring during the cross-linking process.

In the case of peroxide curing, especially when carried out in solution.

peroxide radical fragments may chemically contaminate the initial polymer

chains ( 5 ,9 ) . Networks cross-linked in solution usually contain a lower

concentration of trapped entanglements (compared to cross-linking in the

melt), but here loop formation becomes important (2,10,111.

Chain cross-linking (co)polymerization

The free radical initiated chain (co)polymerization of monovinyl and

polyvinyl monomers leads to the formation of polymeric networks. Typical

examples are the bulk photopolymerization of diacrylates resulting in

glassy, densely cross-linked networks (12,131; the copolymerization of

styrene and divinylbenzene in an inert solvent leading to phase separated,

microporous polystyrene gels (14); and the copolymerization of acrylarnide

and N,N'-methylene bisacrylamide in aqueous solution leading to highly

swollen polyacrylamide gels (151.

Boots used the kinetic gelation model for the simulation of the free

radical chain cross-linking polymerization and showed that network

formation by chain reactions is an intrinsically inhomogeneous process, in

contrast to network formation by stepreactions (16-20). Snapshots taken

during a simulation of polymerization showed an inhomogeneous spatial

distribution of polymer. The homogeneity of the network created could be

improved by decreasing the kinetic chain lengths (In practice this means

increasing the initiator concentration or the temperature). Only at

unrealistically high initiation rates the network formed by the chain

reaction process becomes as homogeneous as the one formed by a step

mechanism. It should be mentioned that for chain cross-linking

polymerizations in the bulk the model implies homogenization towards the

end of the reaction. Complete monomer conversion in the bulk

polymerization is an unlikely event as a result of premature

vitrification.

According to DuSek cyclization plays a dominant role in the chain reaction

network formation. Primarily internally cross-linked microgel particles

are formed, which are linked through peripheral double bonds to form a

heterogeneous gel (21,221. In an inert solvent this leads to the formation

of macroporous, inhomogeneous structures; in bulk reactions

re-homogenization seems likely since polymers cross-linked to high

conversion appear, optically for example, to be quite homogeneous in

agreement with the model described by Boots.

Experimental evidence for the heterogeneity of polyadrylamide gels,

representative for chain cross-linked polymer networks, comes from several

independent studies, such as swelling measurements (231, scattering

studies (41, kinetics (241, electron microscopy (251, and permeability

studies (26-29).

The latter will be discussed in some more detail. Silberberg and coworkers

measured the permeability of polyacrylamide gels (made by the free radical

copolymerization of acrylamide and N,N'-methylene bisacrylamide in water)

and aqueous solutions of linear polyacrylamide. First of all, they

observed that when the total monomer concentration was kept constant, but

the percentage of the cross-linking monomer was increased, the

permeability of the gel rose markedly. It was also found that, at a

comparable concentration, a system of uncross-linked polymer possessed the

lowest permeability. This could only be accounted for by an inhomogeneous

(nonuniform) distribution of the monomer in the gel. In the gel two

regions are formed: one, containing a high proportion of the gel substance

(having a high cross-link density), is essentially non-draining and the

other, containing a very low fraction of the gel substance, is "freely"

draining. Microgel particles, formed in the early stage of the

polymerization, are weakly linked to form a macroscopic gel. The effective

concentration of polymer in the freely draining region is well below the

total monomer concentration as a result of the clustering. Fluid can thus

move faster through the heterogeneous gel than through a polymer solution

of corresponding overall, but uniform concentration. The permeability is

increased when more gel substance is incorporated into the non-draining

region, which happens when the cross-linker concentration is increased.

Another salient aspect is that the faster the initiation, the less

permeable a gel becomes (291. In other words, a high rate of initiation

leads to a more homogeneous network, which again is in full agreement with

the results of the computer simulations done by Boots.

Stepgrowth polymerization

Network structures may also be formed by the random stepgrowth

polymerization of monomers (or prepolymers), at least one type of which

has a functionality of 3 or greater. A classic example is the

polycondensation of dicarboxylic acids with glycerol leading to a

cross-linked polyester (7). Other polymer networks synthesized by

stepreactions include polyurethanes, epoxies, formaldehyde-based

thermosets (Bakelite. Novolac, etc. 1 , polydimethylsiloxane elastomers

(30,311.

By using stepreactions for the formation of polymer networks the topology

of the resulting polymeric network is well defined. Networks having known

values of cross-link functionality and the molecular weight between

cross-links (with known distibution of Mc) are called model polymer

networks (2,6,31-35). These are usually prepared by selectively

end-linking bifunctionally terminated chains (telechelic prepolymers)

with a multifunctional cross-linker. The resulting networks thus have

cross-links of functionality equal to that of the cross-linking agent, and

values of Mc corresponding to the values of the number-average molecular

weight (and its distribution) of the prepolymer prior to cross-linking.

Model elastomeric polydimethylsiloxane (PDMS) networks were prepared by

reaction of hydroxylterminated PDMS chains with tetraethylorthosilicate;

polyurethane elastomers were synthesized by end-linking of

dihydroxylterminated PPO, PEO or polycaprolactone prepolymers with a

triisocyanate, or a polytriol with a diisocyanate (38). These model

elastomeric networks were used to gain some more insight into the

relationships between the structure of a network and its (ultimate)

properties.

One of the imperfections known to be present in network structures is the

dangling chain. The incidence of dangling-end network imperfections in

model networks is very small when the end-linking reaction is carried out

stoichiometrically and to high conversion of functional groups. By

unbalancing the stoichiometry in the end-linking reaction or by

incorporation of monofunctional prepolymers known numbers of dangling

chains were introduced into the network structure (8). As expected an

increase in the concentration of dangling chains had a negative effect on

the ultimate properties, both ultimate strength and maximum extensibility,

of the elastomer.

Another interesting result obtained from these model networks concerns the

network chain length distribution. Properties of bimodal networks,

consisting of very short chains (Mc around a few hundred gmol-i) and

relatively long chains (Mc around 20000 gmol-l), were compared with

unimodal elastomeric networks. It was found that increasing the number of

very short chains in bimodal networks did not show any decreases in

ultimate properties. The strain, even at high elongations, is apparently

reapportioned (nonaffinely) within the network so as to ignore as long as

possible the difficultly deformable short chains (35,361. According to the

"weakest-link" theory rupture of an elastomer is caused by failure of the

shortest network chains (37). This theory is based on the assumption of

affine deformation, which does not seem to be correct. Bimodal networks

containing very large concentrations of short chains (90-95 mol %) were

found to have both high ultimate strength and high maximum extensibility

and as a result of that high toughness, in contrast to both unimodal

networks containing either only short chains or only long chains. This

result is rather surprising, since usually an elastomer will have good

ultimate properties only when reinforced with some mineral filler or hard,

glassy domains in the case of a multiphase polymer, or when it can

generate its own reinforcement through strain-induced crystallization

(39). Since the bimodal elastomeric (PDMS) networks studied have a low T 0

(-40 C) they can not crystallize upon stretching at room temperature.

These bimodal networks show upturns in the modulus at high elongations

that are diminished by neither increase in temperature nor swelling,

unlike networks that can undergo strain-induced crystallization. Any

intermolecular reinforcing effects can thus be ignored. Due to their

limited extensibility of the short chains in the bimodal network the

modulus and ultimate strength are high. The long chains present in the

network somehow delay the growth of the rupture nuclei required for

catastrophic failure. Beyond 95 mol % short chains properties decline,

because of increasing brittleness (21. The unusual ultimate properties of

bimodal networks are due to limited chain extensibilities (non-Gaussian

behaviour) rather than to reinforcing effects (40).

The extent of cyclization in model networks, prepared in the absence of a

diluent, is rather weak when compared to networks formed by chain

cross-linking polymerization (22,411. Dilution may considerably increase

the cyclization and may cause the formation of inhomogeneities, or even

phase separation (11,41). The extent of cyclization depends predominantly

on the polymerization mechanism (22). Stepto stated that the formation of

a perfect network from an end-linking polymerization requires that pre-gel

intramolecular reaction is negligible and that post-gel intramolecular

reaction always leads to elastically ac-tive chains. These requirements are

unlikely to be met. He showed that inelastic loops will arise from both

pre-gel and post-gel intramolecular reactions, which will always occur in

non-linear stepgrowth polymerizations (42-45).

Summarizing we may state that in principal network formation by

stepreactions is a homogeneous process as shown by Boots (18,191 and model

networks with a known structure can be obtained, having small numbers of

dangling-end network imperfections. However, perfect networks free of any

imperfections are not accessible experimentally (44,461.

Polymer networks in medicine

Polymer networks, especially elastomers, have been used in biomedical

applications for several decades (47). Silicone rubber (medical grade) for

example is used in numerous applications including mammary and facial

prostheses, contact and intraocular lenses (55-571, prosthetic heart

valves (48-50). Silicone elastomers are (usually peroxide) cross-linked

polysiloxanes, predominantly PDMS, that are mostly reinforced with

ultrafine particle silica filler to improve the mechanical properties.

Silicone rubbers are highly hydrophobic materials known for their

inertness, high oxygen permeability and relatively good

bloodcompatibility. Thermoplastic polyurethane elastomers, which will be

discussed in the next section, have also been used extensively for medical

applications.

Another class of biomaterials concerns the hydrogels, water swollen

polymers, usually polymer networks (51-53). Hydrogels are very interesting

materials since they superficially resemble living soft tissue in their

physical properties. Hydrogels have relatively high water contents and a

soft, rubbery consistency, causing no mechanical, frictional irritation to

the surrounding tissue, and allow the permeation and diffusion of small

molecules, metabolites just like living tissue. Hydrogels have a low

interfacial free energy in aqueous surroundings. The higher the water

content of the gel, the lower the interfacial free energy becomes. This

low interfacial tension should reduce the tendency of proteins in body

fluids to adsorb and to unfold upon adsorption and also minimize the

driving force for cell adhesion. Minimal protein interaction may be

important for the acceptance of any material when implanted. The

denaturation of proteins by surfaces of implant materials may serve as a

trigger mechanism for the initlation of thrombosis or for other biological

rejection mechanisms. Hydrogels are considered (soft tissue) biocompatible

and bloodcompatible, but suffer from poor mechanical properties. The

higher the water content of the gel, the poorer the mechanical properties

become. The inferior mechanical properties severely limit the potential

applicability of hydrogels. Nevertheless, hydrogels have found many

biomedical applications, like soft contact lenses (57). wound coverings

(541, drug delivery systems, foldable intraocular lenses (55,561,

encapsulation of living cells (58). There are ways to overcome these

mechanical limitations: surface grafting of hydrophilic polymers onto

mechanically strong (hydrophobic) polymers, for example grafted

polyacrylamide or polyethyleneoxide onto polymers for vascular prostheses

to improve the bloodcompatibility (59,601, copolymerization of a

hydrophilic monomer with a more hydrophobic one to form a linear

blockcopolymer, or formation of interpenetrating polymer networks (52).

Wichterle and Lim developed the concept of synthetic, polymeric hydrogels

designed for biomedical applications (53). They described the synthesis of

,lightly cross-linked poly(2-hydroxyethyl methacrylate) (PHEMA) gels. Other

types of hydrogels are prepared from (meth)acrylamide, vinylalcohol and

N-vinylpyrrolidone monomers (51,521. Also hydrophilic networks with

polyethyleneoxide have been synthesized by cross-linking (high molecular

weight) PEO by g-irradiation or by peroxide curing (60.61). Better defined

PEO containing polyurethane networks were obtained by reaction of low

molecular weight PEO diols with triisocyanates or by mixing diisocyanates

with di- and triols (62,631.

Finally, dental restoration materials based on photocurable (aromatic)

dimethacrylates should be mentioned. Concerns over the toxicity of dental

amalgam and an increased emphasis on aesthetics have popularized the

development and clinical use of dental composite restorative materials.

These dental composite resins are composed of a photopolymerizable

dimethacrylate matrix filled with fine (treated) silica particles to

increase the hardness and to lower the overall shrinkage upon

polymerization of the composite material (64).

Polyurethanes

Polyurethanes are a class of polymers having only in common the presence

of urethane bonds somewhere in their chains. The name polyurethane given

to a polymeric material does not tell anything about its chemical and

physical characteristics. Polyurethanes may be lightly or highly

cross-linked or uncross-linked and be highly crystalline, elastomeric or

amorphous and glassy. In the biomedical field polyurethane usually stands

for thermoplastic polyurethane elastomer. Thermoplastic polyurethanes,

also named segmented polyurethanes, are linear blockcopolymers composed of

chainextended diisocyanate hard segments dispersed in a soft segment

polyol matrix (65,661. All commercially available biomedical

polyurethanes, like Biomer, Estane, Cardiothane, Pellethane, Tecoflex, are

composed of an aromatic diisocyanate MDI (4.4' -methylenediphenyl

diisocyanate), except for Tecoflex which contains hydrogenated MDI, a

cycloaliphatic diisocyanate. The soft segment in commercial polyurethanes

is mostly a polyethermacrodiol (polytetramethyleneoxide) having a

molecular weight of 1000-2000 gmol-'. The chainextender is either

ethylenediamine (in the case of a polyurethaneurea) or l,4-butanediol

(formation of a polyurethane). Linear segmented polyurethanes are

preferrably synthesized in a two-step process, where diisocyanate and

polyol are reacted first, and then chainextended with a diol or diamine.

This method of preparation, in contrast to the one-step process where all

reactants are mixed together simultaneously, leads to blockcopolymers with

better defined structures and better properties. These polyurethane

blockcopolymers exhibit microphase separation due to the incompatibility

of the hard and soft chain segments. Elastomeric behaviour is observed

because the hard domains, having a high glass transition temperature or a

high melting temperature, act as multifunctional physical (hydrogen

bonding) cross-links and as a reinforcing filler in the soft (low T

matrix.

Since these polyurethane elastomers are rnultiphase elastomers they show

very good ultimate mechanical properties compared to one-phase

non-crystallizable elastomeric polymer networks (39,671. In such one-phase

elastomers microcracks, once formed, encounter little resistance to growth

because the network chains are highly mobile. High strength and toughness

result from mechanisms that impede crack growth. An elastomeric material

can only exhibit good ultimate mechanical properties when it consists of

two phases. These two phases normally consist of a rubbery amorphous

matrix containing glassy or crystalline domains or reinforcing filler

particles. In the case of crystallizable elastomers, the second phase is

generated upon stretching. Crystallites formed act as reinforcing domains

in the network and thus increase the ultimate properties. Strain-induced

crystallization provides plastic domains which block, retard or deflect

growing cracks. Filler particles or plastic, hard domains (in

blockcopolymers) may also deform plasticly in high-stress regions, thereby

relieving stress concentrations and dissipating energy.

Elastomeric polyurethanes due to their good mechanical properties (high

tensile strength, good flex life, good tear strength, high toughness),

reasonable bloodcompatibility and biocompatibility, have been used in many

medical applications, such as total artificial heart, heart valves,

vascular prostheses, wound dressings etc. (65).

By mixing segmented polyurethanes with (5-20 wt%) high molecular weight

poly(L-lactide) (PLLA), Gogolewski, Leenslag and Pennings developed

elastomeric, biodegradable mixtures with remarkable in vivo performance.

Quenched physical polyurethane/PLLA mixtures, in porous form, were

successfully applied as a small-caliber vascular prosthesis, artificial

skin, meniscus lesion repair material, nerve guide (68-74).

Since these polyurethanes are not cross-linked through covalent chemical

bonds, but physically through hydrogen bonding, they show stress softening

(stress hysteresis) when subjected to multiple stretching, which is a

serious limitation of these elastomers. This phenomenon is attributed to a

disruption of hard segments with strain, leading to a decrease in their

ability to reinforce the rubbery phase upon strain cycling. This problem

can be overcome by chemically cross-linking the linear polyurethane chains

(75). Another disadvantage of commercial biomedical polyurethanes concerns

their composition. As mentioned before, most polyurethanes are composed of

the aromatic diisocyanate MDI since aromatic polyurethanes show better

microphase separation than those based on aliphatic diisocyanates,

resulting in polyurethanes with superior mechanical properties (65.66).

Degradation (through hydrolysis) of the polymer may result in the

formation of the toxic, carcinogenic, mutagenic MDA, methylenedianiline.

Although it has not been shown unambiguously that MDI-based polyurethanes

induce the formation of cancer, it would be more elegant and safer to seek

for a replacement for this component in the polyurethane formulation. The

use of cycloaliphatic diisocyanates, for instance hydrogenated MDI

(Tecoflex) or 1,4-trans cyclohexanediisocyanate, also leads to segmented

polyurethanes having good ultimate properties. Aliphatic diisocyanates,

which are not that suited for the synthesis of thermoplastic polyurethane

elastomers, may be used for the formation of chemically cross-linked

polyurethanes. Especially aliphatic diisocyanates, producing non-toxic

diarnines (eg. lysine, 1,4-diarninobutane ) after eventual degradation, seem

the ultimate choice for the synthesis of biomedical polyurethanes.

Aim and survey of this thesis

As its title implies, the aim of this thesis is to investigate the

possibilities of polyurethanes, especially cross-linked ones, as

biomaterials. The present thesis can be considered an extension of

previous work from our laboratories on elastomeric

polyurethane/poly(L-lactide) mixtures as biomaterials, with the emphasis

here on polyurethanes. In this thesis the preparation, properties and

medical applications of some new polyurethane networks will be discussed.

In chapter 2 the design and synthesis of biodegradable lysine

diisocyanate-based elastomeric polyurethane networks are discussed. The

polyurethane networks, designed to release only non-toxic degradation

products, are prepared from hexafunctional hydroxy terminated starshaped

copolyesters, synthesized by ring-opening copolymerization of L-lactide or

glycolide and &-caprolactone, initiated by myo-inositol; and cross-linked

with ethyl 2,6-diisocyanatohexanoate (lysine diisocyanatel (761.

Chapter 3 is concerned with the evaluation of a lysine diisocyanate-based

polyurethane network (described in previous chapter) as a material for the

construction of a macroporous bottom-layer (dermal analogue) in a

two-layer artificial skin. In vitro and in vivo degradation studies are

presented (77 1.

Chapter 4 deals with the preparation and evaluation of a microporous

polyetherurethane wound covering having a high water vapour permeability.

The elastic, very thin (15-20 pm) wound covering, prepared by means of a

phase inversion process, has been tested on partial-thickness wounds in

guinea pigs and on donor sites in the clinical situation as well. It is

shown that an accelerated wound healing (enhanced reepithelialization)

under this polyurethane membrane is very likely caused by its high water

vapour permeability (78-85).

Chapter 5 describes a two-ply biodegradable microporous small-caliber

vascular prosthesis composed of polyurethane and poly(L-lactide). The

microporous innerlayer, which is supposed to be highly antithrombogenic,

has been made by cross-linking of a mixture of linoleic acid and a

cycloaliphatic polyetherurethane with dicumylperoxide. The outer ply,

containing much larger pores, has been constructed from

polyurethane/poly(L-lactide) mixtures. The biological performance of the

artificial blood vessels and the effects of cross-linking the innerlayer

with peroxides in the presence of linoleic acid on the antithrornbogenicity

of the prosthesis, the prevention of aneurysm formation and the rate of

degradation are discussed (75).

In chapter 6 the synthesis and properties of densely cross-linked

polyurethane networks, and their potential use in ophtalmology

(intraocular lens, keratoprosthesis) are described. Glassy polyurethane

networks are obtained from the bulk reaction of low molecular weight

polyols and (cyc1o)aliphatic diisocyanates. It is shown that these

optically transparent materials, which can be sterilized by autoclaving,

are rather well tolerated in rabbit eyes (86).

References

1. K. DuSek, W. Prins, Adv. Polym. Sci., 6, 1 (1969)

2. J . E . Mark, B. Erman, "Rubberlike elasticity, a molecular primer",

Wiley-Interscience, 1988

3. S.S. Labana, S. Newman, A. J. Chompff, in: "Polymer networks, structure

and mechanical properties", Eds. A. J. Chompff and S. Newman, Plenum

Press, NY-London, 1971, p. 453

4. A.M. Hecht, R. Duplessix, E. Geissler, Macromolecules, 18, 2167 (1985)

5. A. Posthuma de Boer. "Polyethylene networks cross-linked in solution",

Ph.D. thesis, University of Groningen, The Netherlands (1980)

6. J . E . Mark, Acc. Chem. Res., 18, 202 (1985)

7. P.J. Flory, "Principles of Polymer Chemistry", Cornell University

Press, Ithaca, NY, 1953

8. A.L. Andrady, M.A. Llorente, M.A. Sharaf, R.R. Rahalkar, J . E . Mark,

J.L. Sullivan, C.U. Yu, J . R . Falender, J . Appl. Polym. Sci., 26, 1829

(1981 )

A. Posthuma de Boer, A. J. Pennings, J. Polym. Sci. Polym. Phys. , 14,

187 (1976)

P.-G de Gennes. "Scaling Concepts in Polymer Physics". Cornell

University Press, Ithaca, 1979

K.DuSek, M. Ilavsky, in: "Biological and synthetic networks", Ed. 0.

Kramer, Elsevier Applied Science, London & New York, 1988, p. 233

J.G. Kloosterboer, Adv. Polym. Sci., 84, 1 (19801

J.G. Kloosterboer, G.M.M. van de Hei, G. F.C.M. Lijten, in:

"Integration of Fundamental Polymer Science and Technology", Eds.

L. A. Kleintjes and P.L. Lemstra, Elsevier, London, 1986, p. 198

K. DuSek, in: "Polymer networks, structures and mechanical

properties", Eds. A. J. Chompff and S. Newman, Plenum Press, NY-London,

1971. p. 245

J. Baselga, I. Hernandez-Fuentes, I. F. PiBrola, M. A. Llorente,

Macromolecules, 20, 3060 (1987)

J.G. Kloosterboer, G.M.M. van de Hei, H.M.J. Boots, Polymer Commun.,

25 , 354 (1984)

H.M. J. Boots, J.G. Kloosterboer, G.M.M. van de Hei, R.B. Pandey, Br.

Polym. J., 17, 219 (1985)

H.M. J. Boots, in: "Biological and synthetic networks", Ed. 0. Kramer,

Elsevier Applied Science, London & New York, 1988, p. 267

H.M. J. Boots, Physica, 147A, 90 (1987)

H.M.J. Boots, in: "Integration of Fundamental Polymer Science and

Technology", Eds. L. A. Kleintjes and P. J. Lemstra, Elsevier, London,

1986, p. 204

K. DuSek, H. Galina, J. Mike?, Polym. Bull., 3, 19 (1980)

K. DuSek, in : "Developments in Polymerisation Vol. 3", Ed. R. N.

Haward, Applied Science London, 1982, chapter 4

E.G. Richards, C. J. Temple, Nature Phys. Sci., 230, 92 (1971)

J. Baselga, M.A. Llorente, J.L. Nieto, I. Hernandez-Fuentes, I.F.

PiBrola, Eur. Polym. J., 24, 161 (1981)

T.-P. Hsu, C. Cohen, Polymer, 25, 1419 (1984)

N. Weiss, A. Silberberg, Polym. Prepr, ACS. Div. Polym. Chem., 16(21,

289 (1975)

27. N. Weiss, T. van Vliet, A. Silberberg, J. Polym. Sci. Polym. Phys.,

17, 2229 (1979)

28. N. Weiss, A. Silberberg, in: "Hydrogels for medical and related

applications", Ed. J.D. Andrade, Am. Chem. Soc. Symp. Ser. 31, ACS.

Washington DC, 1976, p. 69

29. N. Weiss, A. Silberberg, J. Polym. Sci. Polym. Phys., 19, 1505 (1981)

30. W.D. Cooke, G.B. Guise, "Polymer Update: Science and Engineering",

Australian Polym. Sci. Ser. Vol. 2, Polym. Div. Royal Australian Chem.

Inst.

31. J.E. Mark, Adv. Polym. Sci., 44, 1 (1982)

32. J.E. Mark, J. Chem. Ed., 58, 898 (1981)

33. J.E. Mark, Pure Appl. Chem., 53, 1495 (1981)

34. J.E. Mark, Makromol. Chem. Suppl., 2, 87 (1979)

35. M.A. Llorente, A.L. Andrady, J.E. Mark, J. Polym. Sci. Polym. Phys.,

19, 621 (1981

36. A.L. Andrady, M.A. Llorente, J.E. Mark, J. Chem. Phys., 72, 2282

( 1980 1

37. F. Bueche. "Physical properties of Polymers", Wiley-Interscience, NY,

1962

38. M. Ilavsky, K. DuHek, Polymer, 24, 981 (1983)

39. T.L. Smith, Polym. Eng. Sci., 17, 129 (1977)

40. J.E. Mark, A.L. Andrady, Rubber Chem. Technol., 54, 366 (1981)

41. M. Ilavsky, K. Dusek, Macromolecules, 19, 2139 (1986)

42. R.F.T. Stepto, Polymer, 20, 1324 (1979)

43. R. F. T. Stepto, in: "Advances in Elastomers and Rubber Elasticity",

Eds. J. La1 and J.E. Mark, Plenum Publ. Corp. NY, 1986, p. 329

44. R.F.T. Stepto, in: "Biological and synthetic networks", Ed. 0. Kramer,

Elsevier Applied Science, London & New York, 1988, p. 153

45. R.F. T. Stepto, in: "Developments in Polymerisation Vol. 3", Ed. R.N.

Haward, Applied Science London, 1982, p. 81

46. K. DuHek, Rubber Chem. Technol., 55, 1 (1982)

47. H.M. Leeper, R.M. Wright, Rubber Chem. Technol., 56, 523 (1983)

48. K. J. Quinn, J. M. Courtney, Br. Polym. J. , 20, 25 (1988) 49. M.B. Habal, Arch. Surg., 119, 843 (19841

50. E.E. Frisch, in: "Polymeric Materials and Artificial Organs", Ed. C.G.

Gebelein, ACS Symp. Ser. 256, ACS Washington DC, 1984, p. 63

51. B.D. Ratner. A.S. Hoffman, in: "Hydrogels for medical and related

applications", Ed. J.D. Andrade, Am. Chem. Soc. Symp. Ser. 31, ACS

Washington DC, 1976, p. 1

52. B.D. Ratner, in: "Biocompatibility of Clinical Implant Materials", Ed.

D.F. Williams, CRC Press, Boca Raton, Florida, USA, 1981, p. 145

53. 0. Wichterle, D. Lim, Nature, 185, 117 (1960)

54. P.H. Corkhill, C. J. Hamilton, B. J. Tighe, Biomaterials, 10, 3 (1989)

55. L. Allarakhia, R.L.Knol1, R.L. Lindstrom, J. Cataract Surg., 13, 607

( 1987)

56. A.C. Neumann. B. Cobb, J. Cataract Surg., 15, 257 (1989)

57. A.J. Phillips, J. Stone, "Contact lenses, a textbook for practioner

and student", 3rd ed., Butterworths, 1989

58. R. M. Dawson, R. L. Broughton, W. T. H. Stevenson, M. V. Sef ton,

Biomaterials, 8, 360 (1987)

59. Y. Ikada, Proc. of Polymers in Med. and Surg., 5:6/1 (1986)

60. J. G. Bots. "Polyethers as biomaterials", Ph. D. thesis. University of

Twente, The Netherlands (1988)

61. E.W. Merrill, E.W. Salzman, J. Am. Soc. Art. Int. Org., 6, 60 (1983)

62. N.B. Graham, M.E. McNeill, Biomaterials, 4, 27 (1984)

63. Y. Gnanou, G. Hild, P. Rempp, Macromolecules, 17, 945 (1984)

64. D.C. Watts, in: "Materials Science and Technology, Vol. 14: Medical

and Dental Materials", Volume Ed. D.F. Williams, page 209 (19911

65. M.D. Lelah, S. L. Cooper, "Polyurethanes in medicine", CRC Press, Boca

Raton, Florida, USA, 1986

66. S. Gogolewski, Colloid Polym. Sci., 267, 757 (1989)

67. T.A. Speckhard, S.L. Cooper, Rubber Chem. Technol., 59, 405 (1986)

68. S. Gogolewski, A. J. Pennings, Makromol. Chem., Rapid Commun., 3, 839

( 1982)

69. S. Gogolewski, A.J. Pennings, Colloid Polym. Sci., 261, 477 (1983)

70. S. Gogolewski, A. J. Pennings, Makromol. Chem., Rapid Commun., 4, 675

(1983)

71. J. W. Leenslag, A. J. Pennings, R.P. H. Veth, H.K.L. Nielsen, H.W.B.

Jansen, Makromol. Chem., Rapid Commun., 5, 815 (1984)

72. J. W. Leenslag. M. T. Kroes, A. J. Pennings, B. van der Lei, New

Polymeric Materials, 1, 111 (1988)

J.W. leenslag, "Poly(L-lactide) and its biomedical applications",

Ph.D. thesis, University of Groningen, The Netherlands (1987)

H. J. Hoppen, J.W. Leenslag, B. van der Lei, P.H. Robinson, A. J.

Pennings, Biomaterials, 11, 286 (1990)

A. J. Pennings, K.E. Knol, H. J. Hoppen, J.W. Leenslag, B. van der Lei,

Colloid Polym. Sci.. 268, 2 (1990)

P. Bruin, G. J. Veenstra, A. J. Nijenhuis, A. J. Pennings, Makromol.

Chem. , Rapid Commun. , 9, 589 (1988) P. Bruin, J. Smedinga, M.F. Jonkman, A. J. Pennings, Biomaterials, 11,

291 (1990)

P. Bruin, M.F. Jonkman, H.J. Meijer, A. J. Pennings, J. Biomed. Mater.

Res., 24, 217 (1990)

M. F. Jonkman, H. J. Mei jer, J. W. leenslag, A . J. Pennings, P.

Nieuwenhuis, I. Molenaar, in: "Biomaterials and Clinical

Applications", eds. A. Pizzeferrato, P.G. Marchetti, A. Ravaglioli,

A. J. C. Lee, Elsevier Science Publishers, Amsterdam (19871, p. 361

M.F. Jonkman, I. Molenaar, P. Nieuwenhuis, P. Bruin, A. J. Pennings,

Biomaterials, 9, 263 (1988)

M.F. Jonkman, P. Bruin, E.A. Hoeksma, P. Nieuwenhuis, H. J. Klasen,

A. J. Pennings, I. Molenaar, Surgery, 104, 537 (1988)

M.F. Jonkman. I. Molenaar, P. Bruin, A. J. Pennings, P. Nieuwenhuis,

Transactions 3rd World Biomaterials Congress, Volume XI, Kyoto:

Business Center for Academic Societies Japan (19881, p. 218

M.F. Jonkman, J.M.F. H. Coenen, P. Bruin, A. J. Pennings, H. J. Klasen,

Burns, 15, 211 (1989)

M.F. Jonkman, P. Bruin, J. Biomaterials Appl., 5 , 3 (1990)

M.F. Jonkman, "Epidermal wound healing between moist and dry",

Ph.D. Thesis, University of Groningen, The Netherlands (1989)

P. Bruin, E. Meeuwsen, P. van Andel, J. Worst, A. J. Pennings, to be

pub1 ished

Chapter 2

Design and synthesis of biodegradable poly(ester-urethane) elastomer

networks composed of non-toxic building blocks

Summary

Biodegradable poly(ester-urethane) networks, designed to release only

non-toxic degradation products, were prepared from hydroxy terminated

starshaped prepolymers, synthesized by ring-opening copolymerization of

L-lactide or glycolide and E-caprolactone initiated by myo-inositol, and

ethyl 2,6-diisocyanatohexanoate. The poly(ester-urethane) networks, having

T s in the range 0-10 OC and gel contents up to 95 %, showed rubber-like 4 behaviour and after extraction relatively high. tensile strength (30-40

MPa) .

Introduction

Polyurethanes are considered "excellent" biomedical materials possessing

good mechanical and physical properties and showing relatively good bio-

and bloodcompatibility (1). For these reasons segmented polyurethane

elastomers have been used in biodegradable polyurethane/poly(L-lactide)

(PLLA) mixtures for application as vascular prosthesis (2,6), meniscus

prosthesis (31, artificial skin (4) and nerve guide (5). which have been

developed in our laboratory.

The in vivo rate of degradation, after initially observed fragmentation,

of the polyurethane/poly(L-lactide) mixtures appeared to be very low (6).

A further complication was the observation of creep failure upon dynamical

(cyclic) loading as a consequence of stress softening, always associated

with thermoplastic elastomers, which led to aneurysms of the artificial

blood vessels (6).

However, the major drawback of so called biomedical grade polyurethane

elastomers, like Biomer or Estane, used for biodegradable applications, is

their chemical composition. These polyurethanes contain the aromatic

diisocyanate 4.4'-methylenediphenyl diisocyanate (MDI), which is converted

to the toxic, mutagenic, carcinogenic diamine 4,4'-methylenedianiline

(MDA) after degradation ( 7 , s ) . This problem has been overcome by using

Stashaped polyester prepolymer

HO OH

Llnear plyester prepd ymer

(cyc1o)aliphatic diisocyanates like hydrogenated MDI, 4,4'-methylene-

dicyclohexyl diisocyanate HlzMDI in Tecoflex (71 or hexamethylene

diisocyanate (HDI) (91, but the corresponding diamines are still more or

less toxic.

Therefore, it is more elegant to use L-lysine based di-(or tri)isocyanates

for the synthesis of biodegradable polyurethanes. In scheme 1 two

approaches to the design of such degradable poly(ester-urethane) elastomer

networks, which are designed to release only non-toxic degradation

products, are depicted.

Schindler et al. have already reported on the alcohol initiated

ring-opening polymerization of &-caprolactone. In this manner starshaped

polycaprolactone polymers wcre obtained by using sugar alcohols, like

sorbitol, xylitol or ribitol (10). Lately, Pitt et al. have reported on

the synthesis of biodegradable polyurethanes composed of trihydroxy

terminated prepolymers, made by glycerol initiated ring-opening

copolymerization of a 1: 1 mixture of 6-valerolactone and E-caprolactone,

cross-linked with HDI (11). Lipatova et al. have investigated the

(enzymatic) hydrolysis of poly(ester-urethane) networks, containing 20 wtX

sugar as a filler (12). From the literature, degradable copolyesters of

L-lactide or glycolide and &-caprolactone are well known (13, 14).

The polyurethane networks in scheme 1 ( A ) are built up from hexafunctional

hydroxy terminated starshaped polyester prepolymers, synthesized by

ring-opening copolymerization of L-lactide or glycolide and c-caprolactone

initiated by myo-inositol, which can be cross-linked with ethyl

2,6-diisocyanatohexanoate (i.e. lysine diisocyanate). The degradation

products, myo-inositol, a vitamin widely distributed in the human body

(151, L-lactic acid or glycolic acid, 6-hydroxyhexanoic acid, L-lysine and

ethanol, which are set free upon biodegradation of this polyurethane

network, are all non-toxic which is very essential for the use as a

degradable biomedical material. This chapter reports in more detail on

these networks. The other polyurethane network (scheme 1 (B)) consists of

linear polyester prepolymers, dihydroxy terminated copolyesters of

L-lactide or glycolide and e-caprolactone, using 2-isocyanatoethyl

2,6-diisocyanatohexanoate as a cross-linking agent (16).

Experimental part

Prepolymer synthesis

L-lactide (C. C. A. Gorinchem, The Netherlands; recrystallized from dry

toluene) or glycolide (DuPont) and e-caprolactone (Janssen Chemical,

Belgium; distilled) and myo-inositol (Merck) were dissolved in dry DMF at 0 140 C. Stannous octoate (Sigma Chem. Corp. USA; 0,5 wt%) was added and

0 polymerization was carried out for 20 h at 120-130 C under nitrogen

atmosphere. After removal of the solvent i. vac. a tacky, yellowish

prepolymer resulted, which could be precipitated in ethanol (-70 O C ) from

chloroform solution and subsequently dried i. vac. at ambient temperature.

Ethyl 2,6-diisocyanatohexanoate (17)

L-lysine monohydrochloride (Janssen Chemical, Belgium) was first converted

to L-lysine ethyl ester dihydrochloride by refluxing in ethanol while

passing HC1 gas through the solution. The dihydrochloride was phosgenated

in o-dichlorobenzene or dioxane at 100-110 OC for ca. 8 h. The crude

diisocyanate was purified by vacuum distillation (bp. 125 OC/O,I mmHg).

Network formation

Prepolymers poly(L-lactide-co-E-caprolactone)~ were cross-linked by

treatment with ethyl 2,6-diisocyanatohexanoate ([OHI/[NCOl = 1) in

toluene, and poly(glyco1ide-co-E-caprolactone) prepolymers in CH C1 . Thin 2 2

films were obtained by reaction in a Petri-dish at room temperature under

nitrogen for one day and post-curing at 100-110 OC for 3 h. The elastic, 0 transparent films were dried at 50 C i. vac. Porous materials were

synthesized by curing a viscous slurry of the prepolymer, diisocyanate,

solvent and an amount of dried NaCl powder of variable particle size by

the method described previously. Afterwards the salt was removed by

washing the NaC1/ polymer mixture with water.

Characterization

Gel contents (in wt%) were determined by extraction of the networks with

chloroform. The extracted networks were first carefully air-dried and then

dried i. vac. at 50 OC to constant weight.

Swelling measurements were carried out on extracted networks in chloroform

at room temperature. The degree of swelling was calculated from the weight

increase after swelling using the densities of chloroform (p=1,48 g/cm3)

and the dry extracted networks (p=0,90-0,95 g/crn3).

Thermal analysis of the networks was performed by means of a Perkin-Elmer

DSC-7, calibrated with I. C. T. A. (International Confederation of Thermal

Analysis) certified reference materials and operated at a scan speed of 10

'chin.

Mechanical properties were determined at room temperature using an Instron

(4301) tensile tester equipped with a 10 N load-cell, at a cross-head

speed of 12 mm/min. Specimens (15 x ca. 0,75 x ca. 0,25 mm) were cut from

(unlextracted thin films.

An I. S. I.-DS 130 scanning electron microscope was used to study the

microstructure of the porous materials.

Results and discussion

Poly(ester-urethane) networks were formed by treatment of hexahydroxy

terminated starshaped prepolymers with ethyl 2.6-diisocyanatohexanoate.

Prepolymers were synthesized by ring-opening copolymerization of L-lactide

or glycolide with c-caprolactone in a 1: 1 mole ratio, initiated by

myo-inositol (hexahydroxycyclohexane) using stannous octoate as a

catalyst.

A rather unique aspect of these polyurethane networks is the use of ethyl

2,6-diisocyanatohexanoate, which has not been reported that of ten in the

literature (21-23). Besides the fact that L-lysine is the degradation

product of the incorporated diisocyanate, hydrolysis first of the ethyl

ester results in the introduction of a carboxylic group into the network.

The choice of prepolymers poly(L-lactide(or glyco1ide)-co-c-caprolactone)~

was based on the idea to obtain elastomeric polyurethanes exhibiting a

high rate of degradation (see next chapter). Polyurethane networks

composed of prepolymers poly(L-lactide (or glyco1ide)-co-myo-inositol)~

have glass transition temperatures above room temperature. Therefore,

copolyester prepolymers containing L-lactide (or glycolide) and

E-caprolactone in a 1:l mole ratio were used in order to obtain

poly(ester-urethane) elastomer networks, having T values far below room 9

temperature. The other extreme, polyurethane networks composed of only

poly(&-caprolactone) prepolymers are expected to degrade more slowly than

the co-poly(ester-urethane) networks. For some applications, however, a

low rate of biodegradation seems desirable ( 3 ) .

In table 1 some relevant data of the poly(ester-urethane) networks are 0 collected. The T values of the networks were in the range of 0-10 C,

g

depending on the branch length. The T can be lowered by increasing the B

branch length of the copolyesterprepolymers. The T of a network was '3

raised after extraction with chloroform. Remnants of unreacted monomers

(and oligomers), which were removed by the extraction procedure had a

plasticization effect on the networks. These remnants also had to be held

partially responsible for the observed gel content. Polyurethane networks

with the highest gel contents (95%) were obtained by using precipitated

prepolymers. Even higher gel contents can be expected by using a slight

excess of isocyanate groups ([OHI/[NCOI < 1). Besides urethane bond

formation excess cross-linking can take place through formation of

allophanate groups (18).

Table 1. Poly(ester-urethane) network data

~ 0 1 ~ - prepolymer T gel elongation tensile degree g

urethane branch content at break strength of

networka) length b, (OC) ( % I ( % I (MPa) swelling c

a#l=~repolymer poly(myo-inositol-co-glycolide-co-c-caprolactone + ethyl 2,6-diisocyanatohexanoate; #2=extracted network 1; #3=precipitated

prepolymer poly(myo-inositol-co-glycolide-co-e-caprolactone + ethyl

2,6-diisocyanatohexanoate; #4=extracted network 3; #5=prepolymer

poly(myo-inositol-co-L-lactide-co-c-capro1actone) + ethyl 2.6-diiso-

cyanatohexanoate; #6=extracted network 5.

b)~ranch length: number of lactone molecules (L-lactide, glycolide, c-

caprolactone) per OH group of myo-inositol, calculated from the initial

proportions of starting materials employed.

"1n chloroform at 20 OC.

Fig 1 . Stress-strain curves of poly~glycolide-co-~-capr01actone~-

urethane networks before (-----I and a i t c r (- 1 ~xlraction with

chloroform (respectively networks 3 and 4 i n t a b l e 11.

Fig. 2 . Scannrng t : l t . c t r u n mzcruzraph L,: J F ~ I c u s p . ~ l y ( e s l p r - ~ r e thane

matrix.

The networks were also characterized by their degree of swelling in

chloroform, which ranged from ca. 3,O for the glycolide based networks to

4,75 for the L-lactide based networks.

Fig. 1 shows typical stress-strain curves of poly(glyco1ide-co-c-

caprolactone) networks before and after extraction with chloroform. All

the polyurethane networks showed rubber-like behaviour, but from table 1

and fig. 1 it is clear that the extracted polyurethane networks exhibit

better tensile properties, increased elongation at break and higher

tensile strength (30-40 MPa). Only the extracted networks exhibit

pronounced strain-induced crystallization. Crystallites thus formed have a

reinforcing effect within the network, and thus increase its ultimate

strength and maximum extensibility. The presence of diluent (plasticizer)

suppresses the strain-induced crystallization and thus diminishes the

ultimate properties (20).

Fig. 2 shows a scanning electron micrograph of a porous

poly(ester-urethane) matrix, which was obtained by curing of prepolymer

poly(L-lactide-co-e-caprolactone) with ethyl 2,6-diisocyanatohexanoate in

the presence of an amount of salt (pore volume ca. 85%). In a very

straight forward way (salt casting method) porous materials of these

polyurethane networks for degradable biomedical applications can be

constructed. Preliminary experiments in guinea pigs have shown that the

poly(ester-urethane) networks biodegrade when implanted subcutaneously

(19). Concluding, we state that degradable poly(ester-urethane) networks,

designed to produce only non-toxic degradation products, as described

here, are very promising biodegradable materials. Further work and

especially in vitro and in vivo degradation studies are in progress (19).

References

1. M.D. Lelah, S. L. Cooper, "Polyurethanes in Medicine", CRC Press,

Boca Raton, Florida. 1986

2. S. Gogolewski, A. J. Pennings, Makromol. Chem., Rapid Commun., 3, 839

( 1982)

3. J.W. Leenslag, A. J. Nijenhuis, A. J. Pennings, R.P.H. Veth, H.K. L.

Nielsen, H. W. B. Jansen, Proc. PIMS V, Noordwi jkerhout, The

Netherlands, Sept. 10 - 12, p. 10/1 - 10/9 (1986) 4. S. Gogolewski, A. J. Pennings, Makromol. Chem. , Rapid Commun. , 4, 675

( 1983

5. H.J. Hoppen, J.W. Leenslag, B. van der Lei, P.H Robinson, A . J.

Pennings, Biomaterials, 11, 286 (1990)

6. J.W. Leenslag, M.T. Kroes, A. J. Pennings, B. van der Lei, New

Polymeric Mater. , 1, 111 (1988)

7. M. Szycher, V.L. Poirier, D. J. Dempsey, J. Elastomers Plast., 15, 81

(19831

8. S. Gogolewski, Colloid Polym. Sci., 267, 757 (1989)

9. S. Gogolewski, G. Galletti, Colloid Polym. Sci., 264, 854 (1986)

10. A. Schindler, Y.M. Hibionada, C.G. Pitt, J. Polym. Sci., 20, 319

(1982)

11. C.G. Pitt, 2. W. Gu, P. Imgram, R. Wayne Mendren, J. Polym. Sci., 25,

955 (1987)

12. T.E. Lipatova, S.M. Loos, N.N. Mombuzhai, Vysokomol. Soedin., Ser. A

12, 2051 (1970)

13. A. Schindler, R. Jeffcoat, G.L. Kimmel, C.G. Pitt, M.E. Wall. R.

Zweidinger, in: "Cont. Topics in Polymer Sci. ", ed. by E.M. Pearce,

J.R. Schaefgen, Plenum Press N. Y. ,USA, 1977, vol. 2, p. 251

14. H. R. Kricheldorf, T. Mang, J. M. Jonte, Macromolecules, 17, 2173 (1984)

15. "The Vitamins", ed. by W. H. Sebrell, R. S. Harnis, Academic Press Inc.

N.Y., USA, 1954, vol. 2, p. 321

16. P. Bruin, A. J. Pennings, unpublished results

17. Fr. 1351 368 (19641, Merck & Co. Inc., invs.: J.D. Garber, R.A.

Gasser, D. Wassermann; Chem. Abstr. , 60, P15740d (1964)

18. M. Ilavsky, K. DuSek, Polymer, 24 , 981 (1983)

19. P. Bruin, J. Smedinga, M.F. Jonkman, A. J . Pennings, Biomaterials, 11,

291 (1990)

20. J.E. Mark, Polym. Eng. & Sci., 19, 409 (1979)

21. R. D. Katsarava. T. M. Kartvelishvili, M. M. Zaalishvili, Dokl. Akad.

SSSR 281(3), 591 (1985); Chem. Abstr., 103, 124020~

22. R.D. Katsarava, Kompoz. Polim. Mater., 29, 77 (1986); Chem. Abstr.,

105, 134427q

23. S. J. Huang, K. -W. Leong, ACS Polymer Preprints, 20 (2). 552 (1979)

Chapter 3

Biodegradable lysine diisocyanate-based poly(glyco1ide-co-c-capro1actone)-

urethane network in artificial skin

s-Y A biodegradable lysine diisocyanate-based poly(glyco1ide-co-c-

capro1actone)urethane network has been evaluated as a material for the

construction of a macroporous bottom-layer (dermal analogue) in a

two-layer artificial skin.

High rates of in vitro degradation were observed; degradation of the

porous poly(glyco1ide-co-E-capro1actone)urethane networks was faster in

vivo than in vitro.

Subcutaneous implantation in guinea pigs showed that the porous

polyurethane networks allowed rapid cell ingrowth, degraded almost

completely between 4 and 8 weeks after implantation and evoked no adverse

tissue reaction.

Introduction

Recently we showed that epidermal wound healing of partial-thickness

wounds was accelerated when covered with a microporous polyetherurethane

membrane of high water vapour permeability 1 2 The healing of

full-thickness wounds is much more complicated because there are almost no

epidermal islands left in the wound-bed where skin regeneration may

commence. These wounds heal primarily by wound contraction, resulting in

scar formation ( 3 ) . Therefore a two-layer artificial skin is needed,

comprising a macroporous, biodegradable bottom-layer functioning as a

scaffold for skin regeneration, which enables fibrovascular ingrowth and

which should be resorbed when cell ingrowth is complete; and a

non-degradable top-layer, providing a barrier against infection and

optimal water vapour permeability, which can be peeled off the wound after

healing. This may be combined with seeding epidermal cells in the

bottom-layer (stage 2 artificial skin). This concept for covering

full-thickness wounds was originated by Yannas and Burke ( 4 , s ) . It has

also been described by Gogolewski and Pennings who used polyurethane/

poly(L-lactide) mixtures to construct a two-layer biodegradable artificial

skin ( 6 ) . However, the elastomeric polyurethanes used do not seem to be

ideal for biodegradable applications for two reasons. First, the rate of

degradation is too low, which is especially a problem in case of

applications like biodegradable artificial skin when a high rate of

degradation is desirable. Second, the segmented elastomeric polyurethane

is capable of releasing the toxic, carcinogenic methylenedianiline upon

degradation, as a result of the incorporated aromatic diisocyanate MDI

(7.81.

To overcome these problems we have developed new lysine diisocyanate-based

polyesterurethane elastomer networks, designed to degrade rapidly, thereby

releasing only non-toxic degradation products as outlined before ( 9 ) .

OH

HO

OH

Glycolide + c-caprolactone + myo-inositol ----+

Poly(glyco1ide-co-c-caprolactone) prepolymer

1 HO OH

0 + QCN-FH-$ lysine diisocyanate I$H21r OEt NCO

Poly(glyco1ide-co-s-caprolactonelurethane network

Figure 1. Synthesis of polyesterurethane networks.

Figure 1 shows how these polyesterurethane networks are built up. In

short, hexahydroxyterminated starshaped poly(g1ycolide-co-E-caprolactone)

prepolymers are synthesized by ring-opening copolymerization of glycolide

and E-caprolactone initiated by myo-inositol using stannous octoate as a

transesterification catalyst (9,101. These prepolymers are cross-linked

with 2,6-diisocyanato ethylhexanoate (referred to here as lysine

diisocyanate) to form poly(glyco1ide-co-E-capro1actone)urethane networks.

This chapter reports on the preparation, the physical characteristics and

biological performance after subcutaneous implantation of a porous

bottom-layer of a two-layer artificial skin, composed of this lysine

diisocyanate-based polyesterurethane network; and also how this can be

combined with the previously described polyetherurethane top-layer having

high water vapour permeability to form a stage 1 artificial skin (4.5).

Experimental part

Synthesis of prepolymer poly(glyco1ide-co-&-caprolactone)

Poly(glyco1ide-co-E-caprolactone) prepolymers were synthesized as

described elsewhere (9) .

Two prepolymers differing in glyco1ide:c-caprolactone ratio were

synthesized. Prepolymer A contained glycolide and z-caprolactone in a 1:l

mole ratio, with a calculated branch length of 6 lactone units (i.e.

glycolide or E-caprolactone) per OH group of myo-inositol. Prepolymer B

was synthesized from a 1:1.7 glyco1ide:c-caprolactone feed mole ratio. The

calculated branch length was 7.6 lactones per OH group of myo-inositol.

Porous poly(glyco1ide-co-E-capro1actone)urethane network (bottom-layer)

Poly(glyco1ide-co-E-caprolactone) prepolymer was dissolved in

dichloromethane and freshly distilled lysine diisocyanate (synthesized as

described elsewhere (9)) was added ([OHI/[NCOI = 1). This solution was

mixed with an amount of dry NaCl particles, resulting in a very viscous

slurry. The volatile solvent was allowed to evaporate during this process.

The mixture was then poured into a Petri-dish and extra salt was sprinkled

on top to avoid skin formation. Cross-linking reaction was carried out at

room temperature for one day while all of the solvent was allowed to

0 evaporate and post-curing at 100-120 C for at least 5 h. under nitrogen

atmosphere. Afterwards, the salt was leached out with water and a porous,

sponge-like sheet resulted after subsequent air drying. The porous

networks were extracted with chloroform and dried carefully to constant

weight, from which gel contents (in wt%) were calculated. The porevolume

of the porous materials was calculated from the weight ratio of the

(prepolymer + diisocyanate) and salt.

Two-layer artificial skin

The method for the construction of the polyetherurethane top-layer has

been described earlier (1). This porous PEU top-layer could be glued to

the porous bottom-layer (thickness ca. 2 mm) by using a viscous PEU

solution in THF which was cast in a thin layer onto the top-layer and

subsequently glued to the bottom-layer, dried and placed in water. Thus a

two-layer membrane was constructed.

In vitro degradation

Porous, extracted poly(glyco1ide-co-E-capro1actone)urethane network

samples, pore size 90-250 pm, porevolume 80 % (1 x 1 cm x 2 mm) were

subjected to degradation at 37 + 1 OC in phosphate buffer, pH=6.9.

Degradation was monitored by determination of the weight change.

An I.S.1.-DS 130 scanning electron microscope was used to study the

structure of the porous materials.

In vivo degradation and cell ingrowth

Strips (2 x 2 x 10 mm) of porous lysine diisocyanate-based

poly(glyco1ide-co-e-capro1actone)urethane (pore size 90-250 pm, pore

volume 80 %) were subcutaneously implanted in the dorsum of guinea pigs

(n=4), weighing between 300 and 400 grams. Each animal received six

strip-implants, three based on prepolymer A and three based on prepolymer

B. Every strip was implanted via a separate incision in a surgically

created pocket underneath the panniculus carnosus. The cutaneous incisions

were closed by interrupted 6-0 polyglycolic acid sutures.

The animals were sacrificed 2, 4 , 8 and 12 weeks after implantation and

the location of the implants was identified by blunt dissection of the

complete dorsal skin from the underlying fascia of the paravertebral

muscles. The implants were harvested by wide excision with scalpel and

immersion fixed in 10 % formalin.

The specimens were histologically processed as described previously (17).

Briefly, the specimens were embedded in glycol methacrylate resin, cut

perpendicularly to the axis of the strip-implant in 2 pm thick sections

(thus visualizing the initial 2 x 2 mm cross-surface area), and stained

with Sudan black B and hematoxylin. The Sudan black B stains the polymer

material dark green. Sections were photographed with a Zeiss

Photomicroscope 111.

Results and discussion

All lysine diisocyanate-based poly(glyco1ide-co-c-capro1actone)urethane

networks synthesized according to figure 1, having gel contents 92-95 %,

were extracted with chloroform to remove unreacted monomers (and

unreactive oligomers), which function as swelling agents. The extracted

networks show significantly improved ultimate mechanical properties

(tensile strength, elongation at break) in comparison with the unextracted

networks. Strain-hardening only displayed by the extracted networks is

apparently due to strain-induced crystallization, which is hindered by

swelling agents, even if present in small amounts (the sol fraction

comprises only a few percentages) (9,181. Another reason for extracting

the networks, besides improvement of the mechanical properties, is the

removal of the residual monomers like c-caprolactone, which might give

rise to undesired tissue reactions when implanted.

Porous, sponge-like materials, in sheet form, were obtained by "in-situ"

cross-linking of the prepolymers with lysine diisocyanate in the presence

of an amount of NaCl particles (saltcasting method1 and afterwards

leaching out the salt with water. Fig. 2 shows a scanning electron

micrograph of a porous poly(glyco1ide-co-c-capro1actone)urethane network,

with a mean pore size of 90-250 pm and a pore volume of 80 %, prepared by

saltcasting. It can be seen that by using this saltcasting method an open

porous structure was obtained. As stated in the introductory part, this

porous poly(glyco1ide-co-E-capro1actone)urethane network should function

as a bottom-layer of a two-layer artificial skin allowing fibrovascular

ingrowth and thus has to exhibit an open pore structure.

Another important characteristic of such a bottom-layer is its

degradability. Once the cell ingrowth is complete the porous scaffold has

no function anymore and ideally be resorbed from this moment on, i.e.

after ca. 3-4 weeks. So the material used, should exhibit a high rate of

degradation. For this reason a prepolymer composed of glycolide building

blocks was chosen for the formation of a polyesterurethane network, since

it is known that polyglycolide and its copolymers show a high rate of

degradation (11,12) as compared with poly(L-lactide), for instance.

Hydrolysis of semicrystalline polyesters first takes place in the

amorphous regions, followed by degradation in the crystalline phase (12).

Therefore, it is concluded that purely amorphous polyesters will show a

high rate of degradation as confirmed by the work of Gilding and Reed on

amorphous poly(L-lactide-co-glycolide) (11) and Schindler and Pitt on

amorphous, cross-linked elastomeric poly(valero1actone-co-E-caprolactone)

(13.14,15).

To obtain elastomeric, amorphous polyesterurethane networks, prepolymers

had to be built up from glycolide and E-caprolactone, since

polyesterurethanes from only polyglycolide prepolymers have a too high T .

From the literature, linear copolyesters of glycolide and c-caprolactone

are known (16). Incorporation of &-caprolactone into the branches of the

starshaped prepolymers lowered the T as compared with pure 4

polyglycolide-based branches, so that elastomeric polyurethane networks

with T far below roomtemperature resulted. Besides lowering the '3

glasstransition temperature, crystallization of polyglycolide (which may

happen when the branchlength is long) will be suppressed. All

poly(glyco1ide-co-c-capro1actone)urethane networks were amorphous as

observed by DSC.

Figure 2 . Scanning eIectron micrn~raph n f a porous lysine

diisocyanatc-baecd po!y(glycolide-co-r-caprolsctone)tlretha~~c n~twcrk.

wi tll a mean pore size of 9C-250 pm and a pore volume of 80 %,

prepared by salt-castlng.

Figure 3. Scarrnirlg r , lecL~ cj11 [ I I ~ C I ugt dph oi t h ~ . po rous polyurethane

network ( A ) depxctcd in f i g u r e 2 degraded i n v l t r o f o r 4 wk.

Table 1. In vitro degradation of porous lysine diisocyanate-based poly-

(glycolide-co-E-capro1actone)urethane networks A and B

Time (weeks) X weight loss

Table 1 summarizes the weight loss observed under in vitro conditions for

two elastomeric, porous lysine diisocyanate-based poly(glyco1ide-co-

c-capro1actone)urethane networks A and B with prepolymer

glyco1ide:c-caprolactone feed mole ratio of 1: 1 (prepolymer A) and 1: 1 . 7

(prepolymer B ) , respectively. Significant weight loss occurred after 2

weeks already for A, whereas B showed a comparatively delayed degradation

pattern, because B was built up from the E-caprolactone-richer prepolymer.

Thus by varying the feed mole ratio of glyco1ide:c-caprolactone in the

prepolymer, the rate of degradation of the resulting polyurethane network

can be controlled, due to the greater hydrophobicity of c-caprolactone

units compared with the relatively hydrophilic glycolide units.

Figure 3 shows a scanning electron micrograph of the same porous network A

as in fig. 2 degraded in vitro for 4 weeks. Besides 15 % weight loss (see

table 11, the porous network had also degraded visually. The porous

structure had partially collapsed and the sharp edges had been smoothed.

Two weeks later the porous structure had completely collapsed and the

network had turned into a tacky, chloroform-soluble polymer. Again,

network B showed a delayed degradation in comparison with A. High rates of

degradation in vitro were observed, owing to the amorphous nature of the

polyesterurethane networks (13). Random hydrolytic chain cleavage

apparently caused immediate weight loss, because of the formation of

water-soluble degradation products.

In contrast to the in vitro degradation results, the in vivo results

showed no difference in rate of biodegradation between samples made from

prepolymer A and prepolymer B. All porous implants showed signs of

degradation after four weeks implantation, reflected by distortion of the

outer dimensions, erosion and "foaming" of interporous walls (Figure

4a,b). After eight weeks the implants were almost completely degraded

(Figure 4c,d). At that time polymer remnants had lost their affinity for

Sudan black B and appeared as transparent particles which had been

engulfed by multinuclear giant cells. No polymer particle could be

detected after twelve weeks implantation, nor could the implantation sites

be identified. Histological evaluation of "blindly" taken twelve-week

tissue samples did not show polymer material or scar tissue.

Cell ingrowth was already seen in the two-week samples. Cells had filled

the complete labyrinth of micropores and consisted of macrophages,

epithelioid cells, fibroblasts and endothelial cells. The endothelial

cells had a lumen by that time, thus forming capillaries deep in the pores

of the implant. These capillaries had grown to 40 pm wide vascular

structures by the eigth week (figure 4d). In the course of weeks,

epithelioid cells predominated the infiltrate fusing into multinuclear

giant cells.

The poly(glyco1ide-co-c-capro1actone)urethane material can be considered

biocompatible, since no adverse tissue reactions developed. The implants

did not evoke any granulocyte-mediated inflammatory reaction. A thin

fibroblast layer initially encapsulated the polymer strip, but merged in

the surrounding loose connective tissue by the eigth week (figure 4c).

Connective tissue fibers, identified as being type I11 collagen using

Herovici's connective tissue stain, had been deposited, probably by

fibroblast, into the pores of the implant by the fourth week.

Figure 4.

Histology of cross-section of porous biodegradable lysine

diisocyanate-based poly(glyco1ide-co-e-capro1actone)urethane networks, 4

(a,b) and 8 (c,d) weeks after subcutaneous implantation in the guinea pig.

The implant was originally 2 x 2 mm in cross-section. (Sudan black B and

hemotoxylin; skin is to the top; bars represent 100 pm).

a. Four weeks after implantation: the polymer strip has colllapsed to 0.6

x 1.9 mm and is encapsulated by a thin fibroblast capsule. Polymer

fragments of interporous walls (PI are separated by ingrowing cells and

blood vessels (BV).

b. High power view of a. Epithelioid cells (E) and multinuclear giant

cells (MNG) engulf the polymer particles. The two polymer fragments

present at the bottom of the figure ( P I show "foaming" as a result of

resorption.

c. Eight weeks after implantation (note the same magnification as in a. 1:

The polymer strip has now collapsed to 0.09 x 1.9 mm and is penetrated by

numerous blood vessels (BV). The fibrous capsule has been absorbed by the

surrounding loose connective tissue.

d. High power view of c. Polymer particles (PI have been engulfed or

phagocytosed by multinuclear giant cells (MNG) and do not stain anymore

with Sudan black B. Note the numerous blood vessels (BV) involved in the

process of degradation.

Degradation of the porous poly(glyco1ide-co-E-capro1actone)urethane

samples was faster in vivo than in vitro. This faster biodegradation might

well be explained by the mechanical strain of ingrowing cells, the

additional effect of biologically available enzymes, and the subsequent

intracellular degradation of small polymer particles.

In conclusion, this new biodegradable poly(glyco1ide-co-c-caprolactone)

urethane seems promising as a material for the construction of a

macroporous bottom-layer, with a mean pore size of 90-250 pm (dermal

analogue) in a two-layer artifical skin, since it evokes no adverse tissue

reactions, allows rapid cell ingrowth and degrades almost competely

between 4 and 8 weeks after implantation. Futher studies are planned to

examine the efficacy of the two-layer artificial skin in a full-thickness

wound model in the Yorkshire pig.

References

1. P. Bruin, M. F. Jonkman, H. J. Meijer, A. J. Pennings, J. Biomed. Mater.

Res., 24, 217 (1990)

2. M.F. Jonkman, P. Bruin, E . A . Hoeksrna, P. Nieuwenhuis, H. J. Klasen,

A. J. Pennings, I. Molenaar, Surgery, 104, 537 (1988)

3. W. van Winkle, Surg. Gynec. Obstet., 124, 369 (1967)

4. I.V. Yannas, J.F. Burke, J. Biomed. Mater. Res., 14, 65 (1980)

5. I.V. Yannas, D.P. Orgill, in: "Polymeric Biomaterials" (Ed. E . Piskin

and A. S. Hoffman), Ni jhoff Publishers, The Netherlands (19861, p. 221

6. S. Gogolewski, A. J. Pennings, Makromol. Chem., Rapid Commun., 4, 675

(1983)

7. M. Szycher, V.L. Poirier, D. J. Dempsey, Elastomers and Plastics, 15,

81 (1983)

8. R.E. Marchant, Q. Zhao, J.M. Anderson, A . Hiltner, Polymer, 28, 2032

( 1987 )

9. P. Bruin, G. J. Veenstra, A. J. Nijenhuis, A. J. Pennings, Makromol.

Chem. , Rapid Commun. , 9, 589 (1988)

10. A. Schindler, Y.M. Hibionada, C.C. Pitt, J. Polym. Sci., 20, 319

(1982)

11. A.M. Reed, D.K. Gilding, Polymer, 22, 494 (1981)

12. D.F. Williams, J. Mater. Sci., 17, 1233 (1982)

13. C.G. Pitt, R. W. Hendren, A. Schindler, S. C. Woodward, J. Contr.

Release, I, 3 (1984)

14. A. Schindler, C.G. Pitt, Polymer Preprints, 23(2), 111 (1982)

15. C. G. Pitt, A. E. Schindler, "Biodegradable polymers of lactones". U. S.

Patent 4.379.138.

16. H. R. Kricheldorf, T. Mang, J. M. Jonte, Macromolecules, 17, 2173 (1984)

17. E.A. Hoeksma, B. van der Lei, M.F. Jonkman, Biomaterials, 9, 463

(1988)

18. J. E. Mark, Polymer Eng. & Sci. , 19, 409 (1979

Chapter 4

A new porous polyetherurethane wound covering

Summary

A polyetherurethane (PEU) wound covering with non-interconnected

micropores up to approximately 5 pm has been prepared by means of a phase

inversion process. This highly elastic, very thin (15-20 pml, pliable

wound covering showed good, immediate adherence to wet wound surfaces and

high water vapour permeability, but was impermeable to bacteria.

In guinea pigs epidermal wound healing of partial-thickness wounds under

PEU wound coverings was accelerated as compared with uncovered controls

and an occlusive wound covering, OpSite. Water in liquid form or wound

exudate could not leak through the PEU covering, but its high water vapour

permeability induced concentration of the wound exudate into a jellylike

clot layer, which apparently accelerated reepithelialization.

The main conclusion from a clinical study on 20 donor sites was that the

use of the PEU covering reduces pain, besides prevention of fluid

retention and enhanced reepithelialization.

Introduction

The concept of a two-layer artificial skin for covering full-thickness

(burn) wounds, consisting of a biodegradable, porous bottom-layer

functioning as a temporary template for skin regeneration and a protective

non-degradable top-layer was first described by Yannas and Burke (1).

Gogolewski and Pennings constructed an artificial skin based on

polyurethane/poly(L-lactide) mixtures using the above concept (2).

Our current concept of the synthetic skin substitute consists of a

microporous vapour permeable polyetherurethane (PEU) top-layer and a

separate bottom-layer, composed of a biodegradable polyesterurethane

elastomer network, which is designed to degrade rapidly to non-toxic

degradation products, not having the disadvantages like low rate of

degradation and release of the toxic, carcinogenic, aromatic diamine

4.4'-methylenedianiline upon degradation of the segmented polyurethane,

associated with the PU/PLLA mixture (3).

The PEU top-layer in itself can also serve as a wound covering, acting as

a temporary covering of donor sites and second degree burns

(partial-thickness wounds).

It is commonly accepted that such a wound covering should be adherent,

elastic, pliable, impermeable to bacteria, easy to handle, non-toxic,

hemostatic and also allow the proper water vapour transport through the

covering (4,5). However, no data concerning the optimal water vapour

permeability of wound coverings, to prevent desiccation of the wound and

avoid simultaneously fluid accumulation under the covering, are available

(5.6). Despite the fact that many wound coverings are commercially

available (e.g., OpSite, Biobrane, Omiderm, Epigard) many surgeons still

prefer treatment of donor sites and second degree burns with conventional

methods like tulle gras dressing, which may indicate that the ideal.

synthetic wound covering has yet to be developed (5.8). This chapter

reports on the preparation and characteristics of this new

polyetherurethane wound covering.

Materials and Methods

PEU wound covering

In this study Tecoflex EG-80A (9) (Thermedics Inc. a so called second

generation medical-grade segmented cycloaliphatic elastomeric

polyetherurethane (PEU) was used as supplied.

A 5% (w/w) PEU solution in tetrahydrofuran (THF), containing 1% (w/w)

lithiumchloride (LiC1) was refluxed for 2 h under nitrogen atmosphere and

subsequently stirred for 2 h at room temperature. A film of this polymer

solution was cast on a glass plate, using a 500 pm cast-iron. The glass

plate was placed just above a water layer in an open box. Within some

minutes the clear film turned white. Half an hour later on, when most of

the solvent had evaporated, the non-translucent film was dried in vacuo at

50 OC for 2 h and subsequently placed in water to detach the film from the glass plate and to remove the salt. Finally the white elastic film was

dried. The PEU covering was sterilized by means of gamma radiation (25

kGy) .

Porous PU membrane

A Petri-dish containing a 7% (w/w) polyurethane (Estane, Goodrich U.S.A. )

solution in 1,4-dioxane was placed above a 1:l v/v water/l,4-dioxane

mixture in a closed box. After 2 days the porous membrane was placed in

water and subsequently dried.

Water vapour permeability

The wound covering was stretched in a screwed open cap onto a glass cup,

which was partially filled with water and inverted so that the wound

covering was in direct contact with water.

A ServoMed evaporimeter (Model EP-lC, ServoMed AB, Vallingby, Sweden) was

used for measuring the water vapour transmission rate (WVTR in gm-2h-1) at

various water vapour pressure differences across the wound covering. The

measurements were performed in a closed cabinet to prevent disruptive air

currents as described elsewhere in detail by Erasmus and Jonkman (6,7).

Stress-strain measurements were performed on cut specimens ( 5 x 20 mm)

from the above described dry PEU film and the commercial wound covering

OpSite (10) (thickness 28 pm, Smith & Nephew Ltd., Hull, U.K. at room

temperature using an Instron (4301) tensile tester equipped with a 10 N

load-cell, at a cross-head speed of 50 mm/min.

An I.S.1.-DS 130 scanning electron microscope was used to study the

microstructure of the membranes.

Animal and clinical studies

Under sterile conditions, at the back of each of 61 guinea pigs two

partial-thickness wounds (2 x 2 cm) with a mean depth of 0,32 mm were made

with a dermatome. The wounds were either covered with a dry, sterilized

PEU membrane or with OpSite or were left uncovered. The coverings were not

changed during the experiment. All wounds were evaluated histologically 1

to 14 days after excision as described in detail previously (12)

Twenty adult burn wound patients undergoing split-skin graft procedures at

the Burns Centre of the Roman Catholic Hospital of Groningen were

candidates for the clinical study. Skin grafts with a mean thickness of

0,30 mm were taken with a dermatome. Half pf the donor site was covered

w i t h the PZU cuverinq, tlre crther h a l f w l t t ~ = s l n p l e laycr of tuIle eras

drrs9lne (paraffin gauze) . The complete d a ! l ~ r rile w , ~ s cover~d w i t h f o u r

absorptive co t t on pads, comprising 32 s Lng!r l a y ~ r s f i n r rnrmsh ~ a i ~ p , hn'd

i n place with a crepe bandage The c g t t o n ~ d d s a-d crepe handage above t h e

PEU cover in^ uere cut abay hr twccn 1 and 5 days a f t e r npet-ation, thus

allowing free ventilation. T h e FEU cuvur i r ) ~ w , l s peeled o f f th i l bound

between 5 and 21 d a y s a f t e r operatlun '[ tw crepe bsndnjie and Lau7t.s or. t o g

of t h e t u l l e eras were removcd b v t w ~ ~ r , 7 a n d 10 d a y s a f t c t nperat i n n

B i n p s l e s were t a k ~ n between 3 days to 3 m o n t h s af t u r o p c r z l i u r ~ and sl~died

histologically as drscritcd e l s c h h e r e in dctall (15)

Results and discussion

Figure 1 presents a s c a n n i n e electrotj micrograph of a PtJ membrane,

possessing a w r y rc.gular pore q t s t i c t u r ~ I t car1 be s e m t h ~ t t h e large

pores are interconnected with c m a l l e r o n e s . l h i s pornus mrmbrane was

obtained by s l o ~ cvapuratiurt u f k l i c - sc7!vur~? (1,4-dioxane] and

slmu1:aneously s l o w diffuszon n f water vapou1- ( r w n s r> lven t ) i n t o t h e

polymer solution.

F i w r e 1. Scann lne e l e c t r o n rnIcrr,gi,?ptl o: a PLI rnerrbranr prt-pared f rom

a 7 utX polymer solutie? i n 1.4-dinxzne us in€ a I : 1 v/v 1.4-dinvan?/

w a t ~ r snlvent/nonsol v ~ n t mixtur-e. N o t e t h r I - P R U ~ ~ I - pure s t r u c t u r e .

I ' lgure 2. Scannine e l e c t r o n rnicrazral>h of a PEU mr:mbr-arw pr cparrzd by

c a s t i n g a fiPn from, a 5 w t % polymer solution i n TIIF, containing i w t X

LIC1. The rough topslde of t l e r n ~ m b r a n ~ i s t h e bo t to rns idc o f the FEU

wound covering, which faces the wound.

Figure 3 . Scanning electron micrograph u f a crcss-section of t h e PFU

wound covering.

By using a more volatile solvent (THF) combined with the presence of the

hygroscopic lithiumchloride in the polymer solution, which might attract

water vapour, the process of phase inversion was accelerated, which

resulted in a different, less regular membrane structure.

Figure 2, 3 and 4 show scanning electron micrographs of the dry PEU wound

covering, prepared by casting a film of a 5 wt% polymer solution in THF,

in the presence of 1 wt% LiCl. The topside of the PEU membrane, which

actually is the bottomside of the wound covering facing the wound, showed

a rough surface, composed of pits and small pores up to approximately 5 pm

(fig. 2). The membrane of thickness 15-20 pm contained micropores up to

ca. 5 pm, which were not interconnected (fig. 3 ) . The bottomface of the

membrane (i.e. the topside of the wound covering), which was originally

stuck to the glass plate, showed a smoother surface with pores up to ca. 5

Mm (fig. 4). As a result of this structure water in liquid form (or wound

exudate) could not leak through the PEU membrane, but water vapour

diffused at a high rate (see below). Due to the porous structure and its

hydrophilic characteristics the PEU membrane showed good adherence to wet

surfaces. The dry PEU wound covering immediately adhered well to

partial-thickness wound beds, thereby sealing the wound and forming a

barrier against infection, since it was shown that the covering was

impermeable to bacteria (12).

Figure 5 presents the water vapour transmission rate (WVTR) vs. the water

vapour pressure difference across two wound coverings, namely the PEU

wound covering and OpSite, measured in vitro, "upside down" (6.7) . The

slopes of the lines empirically found are designated water vapour

permeance (WVP) . The water vapour permeability of the PEU wound covering, expressed as

water vapour permeance was 20,l gm-2h-'k~a-', which was much higher than

the corresponding value found for OpSite (5.3 gm-2h-1k~a-'), which is

considered an occlusive wound covering ( l o ) , but less than Omiderm (24.6 gm-2h-1k~a-' ) , a commercially available polyurethane wound covering which has been grafted with polyacrylamide and which is known to be a high

vapour permeable wound covering (14). It is known that the use of

occlusive wound coverings in partial-thickness wounds accelerates

reepithelialization. stimulates wound healing, prevents wound desiccation

P N

ca.

n t I 3 L 5 r; Wnler vopor pressure d,Hermce (k%l

F l y r e 5. Wntar vapour p 6 r t r o a b i l i t y o f wound coverin~s: Water vapokir

transmission rate (WVTR) a s a f u n c t i ~ n of the water vapour pressure

diffcrcnce across t hc wound covering. The slopes of the individual -2 -1

lines are designated water vapour per rrlt,ance Igln 11 k ~ a - I ) .

and body heat loss, but has the disadvantage of accumulation of wound

exudate under the covering, which may lead to infection, especially in

case of occlusive film dressings like OpSite (10,181. It is suggested that

a more water vapour permeable wound covering, like the constructed PEU

membrane, will avoid the latter.

Figure 6 shows typical force-strain curves of the dry PEU wound covering

and for comparison OpSite, a commercial, clinically used wound covering.

It is clear that the dry PEU wound covering was far more elastic than

OpSite, having a lower modulus, but exhibiting lower strength at break.

A wound covering should be elastic if it has to be applied over bending

surfaces like over joints, to facilitate an intimate cover of the wound.

The thickness of the PEW membrane (15-20 pm) together with the elasticity

made the PEU membrane very pliable, a property which is needed to enclose

the wound surf ace very near (conf ormabi 1 i ty ( 16 1 1. The PEU covering is

elastic both in the dry and wet state, in contrast to some hydrogel-based

wound coverings, like Omiderm for instance, which are only elastic in the

wet state and contract the wound when losing water (vapour) during the

healing process (17).

The force-strain behaviour of a PEU membrane determined directly after

preparation differed from an aged PEU membrane as can be seen in fig. 6.

The P N membrane was prepared by casting of a LiCl containing refluxed

polymer solution. Heat treatment of a polyurethane solution will result in

complete dissolving of the physically cross-linked PU chains by breaking

up the existing supermolecular structure in solution. The LiCl keeps the

chains from aggregation upon cooling down to room temperature by

complexation to the urethane bonds, thus enhancing the solubility of the

PU and leading to the formation of a new super molecular structure (13).

A membrane cast from this PEU solution showed relatively high elasticity

(low modulus). However, after 1 to 3 months the elasticity had decreased

slightly. The at room temperature aged PEU membranes also exhibited

decreased elongation at break.

Preliminary studies on wound healing of partial-thickness wounds in rats,

using the PEU wound covering were encouraging (11 1. Extensive studies in

guinea pigs (121, in which wound healing of 122 partial-thickness wounds

under PEU coverings was compared with wounds covered with OpSite, an

Fieure 5 . Force - s t r - a in bchaviour of. ( a ) PEll wuur~d covcrine 3 mcnths

after preparation, IkE PEL1 woirrid covrring r l i r r r t l y aft-r prppmration

and CpSite

-

F i ~ i i r e 7 . F r e s h split-thickn~ss d r ~ n c s s i t e t~:>i!- dres srd wi t.h t.he PW

membrane Iriehtl and hali with a pararrin gauze ( l e f t ) . N u t e t h a t the

PEIJ membrane hecunles I r k ~ ~ s ~ i s [ t . r ~ l wt!e11 v~ic-ker i t.rr t he woll lrt lberl .

occlusive film dressing, and uncovered controls, showed that

reepithelialization and keratinization were enhanced in wounds covered

with the high water vapour permeable, porous PEU membrane compared with

wounds covered with OpSite or uncovered air exposed controls. The

percentage of reepithelialization on day 2 after operation was 85 % in

wounds covered with the PEU membrane, whereas it was 66 % and 35 %,

respectively in wounds covered with OpSite or exposed to air. In PEU

covered wounds 100 % reepithelialization was attained by day 3, one day

earlier than in the other wounds. Under the PEU covering the wound exudate

had turned into a jellylike clot layer by day 1 as a result of the high

water vapour permeablity of the porous covering. The high water vapour

permeability of the PEU covering prevented fluid retention (as was

observed in the case of OpSite covering) as well as complete wound

desiccation (uncovered controls). The jellified clot layer underneath the

PEU covering apparently provided an ideal matrix for epidermal wound

healing.

In order to evaluate the clinical efficacy of the PEU wound covering, it

was compared with the conventional treatment of tulle gras dressing (see

fig. 7) plus absorptive gauzes and crepe bandage on split-thickness skin

graft donor sites of 20 burn wound patients (15). The, initially fluid,

wound exudate under PEU coverings concentrated into a jellylike clot layer

after the extra gauzes had been cut away, which is needed to allow free

ventilation. After 5 days already the PEU covering could be peeled off the

wound without pain or epithelial damage. Clinically and histologically no

significant difference was observed in the rate of healing between the PEU

and tulle gras covered wounds, which may be explained by the fact that

tulle gras packed in a thick layer of gauzes and bandage prevented wound

desiccation, as a result of a sultry effect caused by the gauzes. Both

treatments enhanced reepithelialization at a similar rate. Further it was

observed that the use of the PEU wound covering reduces pain completely

compared with the rather painful treatment with tulle gras. One point to

note, finally, is that the PEU wound covering is not hemostatic in itself,

which, however did not turn out to be a problem in the clinical situation.

Acknowledgements

The authors would like to thank Dr. J.W. Leenslag for his contribution to

this work.

References

1. I.V. Yannas, J .F . Burke, J. Biomed. Mater. Res., 14, 65 (1980)

2. S. Gogolewski, A. J. Pennings, Makromol. Chem., Rapid Commun., 4, 675

(1983)

3. P. Bruin, G. J. Veenstra, A. J. Nijenhuis. A. J. Pennings, Makromol.

Chem. . Rapid Commun. , 9 . 589 (1988) 4. M. J. Tavis, J. Thronton, R. Danet, R . H . Barlett, Surg. Clin. North

Am., 58, 1233 (1978)

5. K. J. Qu~M, J.M. Courtney, J . H . Evans, J. D.S. Gaylor, W.H. Read,

Biomaterials, 6, 369 (19851

6. M.F. Jonkman. I. Molenaar, P. Nieuwenhuis, P. Bruin, A. J. Pennings,

Biomaterials, 9, 263 (1988)

7. M. E. Erasmus, M. F. Jonkman, Burns, 15, 371 (1989 )

8. C. P. Artz, D. R. Yarbrough, in: "Textbook of Surgery ed. ll",

W.B. Saunders Company, Philadelphia, 295 (1977)

9. M. Szycher, V.L. Poirier, D. J. Dempsey, J. Elastomers and Plastics,

15, 81 (1983)

10. S. R. May, in: "Burn Wound Coverings" (Ed. D. L. Wise), CRC Press, Boca

Raton, Florida, 53 (1984)

11. M. F. Jonkman, H. J. Mei jer, J. W. Leenslag, A. J. Pennings, P.

Nieuwenhuis, I. Molenaar, in: "Biomaterials and Clinical Appl.",

Elsevier Science Publishers, Amsterdam, 361 (1987)

12. M.F. Jonkman, P. Bruin, E.A. Hoeksma, P. Nieuwenhuis, H. J. Klasen,

A. J. Pennings, I. Molenaar, Surgery, 104, 537 (1988)

13. A.G. Zhigotskii, Z.N. Pazenko, T. I. Zhila, A.A. Panasevich, A.G.

Yakovenko, International Polymer Science and Technology, 3, 28 (19761

14. D. Behar, M. Jaszynshi, N. Ben Hur, J. Golan, A. Eldad, Y. Tuchman, N.

Stevenberg, B. Rudensky, J. Biomed. Mater.Res., 20, 731 (1986)

15. M.F. Jonkman, J.M. F. H. Coenen, P. Bruin, A. J. Pennings, H. J. Klasen,

Burns, 15, 211 (1989)

16. D. Queen. J.D.S. Gaylor. J.H. Evans, J.M. Courtney, W.H. Reid,

Biomaterials, 8, 372 (1987)

17. C. Cristofoli, M. Lorenzini, S. Furlan, Burns, 12, 587 (1986)

18. V. Falanga, Arch. Dermatol., 124, 872 (1988)

Chapter 5

A two-ply artificial blood vessel of polyurethane and poly(L-lactidel

S-Y

A biodegradable microporous small-caliber vascular prosthesis has been

developed that consists of two layers. The inner layer has been made

highly antithrombogenic by cross-linking of a mixture of linoleic acid and

a cycloaliphatic polyetherurethane with dicumylperoxide. Microporosity was

introduced by adding sodiumfluoride crystals of about 5 pm in diameter

prior to cross-linking and leaching them out afterwards. The outer ply has

been constructed by precipitating a (95/5) physical mixture of

polyesterurethane and poly(L-lactide) from solution in the presence of

sugar crystals with dimensions in the range 30-90 pm which were removed by

exposing the prosthesis to water.

The two-ply prostheses were tested in vivo by replacing 1 cm of the

abdominal aorta of rats. All the prostheses remained patent at least up to

1 year and did not exhibit any aneurysmal formation. The inner layer was

covered with endothelial cells and several layers of smooth muscle cells.

Introduction

Since the majority of deaths in the western countries are caused by

malfunctioning of diseased arteries, there has been a virtually unlimited

clinical need for arterial prostheses. This has brought about a

considerable research effort at constructing polymeric conduits for blood

and testing their biological performance. Especially small-caliber

artificial blood vessels with inner diameters of less than 6 mm seem to be

difficult to design and often exhibit poor patency as a result of blood

clotting.

Our approach to solving this problem has been to fabricate compliant,

microporous tubes of biodegradable polymer mixtures of polyurethanes and

poly(L-lactide), which act as temporary scaffolds for the ingrowth and

overgrowth of cells so that a neo-artery could develop 11.21. It was

indeed found that our vascular prosthesis when implanted in rats to

replace 1-cm length of the abdominal aorta induced the formation of a new

arterial wall consisting of an inner endothelial lining (neo-intima),

several layers of smooth muscle cells connected by elastin and collagen

(neo-media) and an outerlayer of fibrohistiocytic tissue constituting a

neo-adventitia [31. At one year after implantation some neo-arteries were

found to have newly formed smooth muscle cells that were

circumferentially arranged and elastic fibers formed concentric layers as

in the natural aorta. The development of the new artery wall is to be

attributed to a very well-matched rate of desintegration of the vascular

prosthesis and the rate of tissue ingrowth and the mechanical stimulation

on cell growth and orientation by the arterial pulsation of the blood. The

rate of fragmentation of this compliant prosthesis has been enhanced by

the rapid precipitation of the physical mixture of the polyesterurethane

and the high molecular weight poly(L-lactide) as used in the applied

dip-coating technique [4,181. The polymers solidify far from equilibrium

conditions thereby generating a substantial amount of residual stress in

the porous material. These residual stresses, as well as those arising

from the blood pulses and the large internal surface make the structural

breakdown by the hydrolytic and enzymatic environment quite unpredictable

[5-71. This has also led to the formation of aneurysms, a catastrophic

dilation of the artery wall, in several cases.

If one wishes to ever apply these artificial blood vessels to patients for

the replacement of damaged veins and arteries due to atherosclerosis it

obviously is a basic requirement that the rate of fragmentation of the

implants does not exceed the rate of tissue formation. But the major

problem arises if the prosthesis measures more than one cm in length

because the smooth muscle cells and endothelial lining have to grow mainly

from the anastornotic sites and require much more time to cover the entire

luminal surface of longer prostheses 181. This difficulty might be

circumvented, for instance by seeding of endothelial cells [91 and smooth

muscle cells all along the inner wall of the prosthesis as investigated by

Wildevuur et al. [lo]. But also improving the implant construction may be

sufficient as demonstrated by Gogolewski et al. [ I l l . They implanted

prostheses that measured 6 cm in length in healthy pigs and observed the

development of a neoartery which also grew in length reaching 12 cm after

one year. The key feature is that blood should not coagulate when it comes

in contact with the polymer surface in vivo which is initiated by the

adsorption of platelets. This may be achieved by special polyurethanes

which preferentially adsorb albumin and therefore almost no platelets, as

developed by Lyman et al. 1121. Ikada [I31 improved the

antithrombogenicity by grafting polyacrylamide on polymer surfaces. Bots,

van der Does and Bantjes 1141 made use of polyethers as biomaterial for

vascular prostheses and Klopper 1151 made them of collagen cross-linked

with glutaraldehyde. Bamford I161 has been able to prevent platelet

aggregation by attaching a synthetic analogue of prostacyclin to a variety

of polymer surfaces. Grafting of heparin [17,39,401 onto the luminal side

of our prosthesis decreased the platelet deposition but did not improve

the patency rate [171.

In the present study an attempt has been made to increase the

antithrombogenicity and lifetime of the vascular prosthesis by

cross-linking a thin inner layer of microporous polyurethane in the

presence of linoleic acid with dicumylperoxide. The negative charges [211

of the carboxyl groups [221 on the inner wall repel the platelets which

are also negatively charged. In addition, carboxyl groups appear to favour

adherence and spreading of cells 119-201 which therefore may promote the

attachment of layers of smooth muscle cells and endothelial cells. This

thin antithrombic inner layer was covered with a compliant microporous

material composed of a physical mixture of polyurethane and high molecular

weight poly(L-lactide) that was used successfully in previous animal

experiments.

This chapter describes the preparation of the two-ply artificial blood

vessel and its biological behaviour in rats. The prostheses that were

implanted were all patent for at least one year and did not exhibit any

aneurysm formation.

Experimental

Materials

The poly(L-lactide) (Mv=9.6. lo5) used throughout this study was

synthesized by ring-opening polymerization of L-lactide that was catalyzed

by Sn(I1)-2-ethylhexanoate. A medical grade segmented polyesterurethane

(Estane 5701 F1, Goodrich USA) was used after being purified by

precipitation from solution in N,N-dimethylformamide (DMF). The solution

was poured into ice-water. The polymer was washed with ethanol (96%) and

ether and dried in vacuo at 40'~. A second medical grade polyetherurethane

(Tecoflex EG 80A, Thermedics Inc., USA) was used as supplied.

The solvents, DMF, tetrahydrofuran (THF) and 1,4-dioxane (Janssen Chimica,

Belgium) were purified prior to use according to standard procedures.

Chloroform, linoleic acid (9,12-octadecadienoic acid, cis, cis) (no.5353

Merck, Germany), sugar crystals (Suiker Unie, The Netherlands) and NaF

(Baker, UK) were used as supplied. Dicumylperoxide (Dicup R, Hercules,

USA) was purified by repeated recrystallization from methanol.

Prostheses preparation

The inner layer was made of a suspension that was prepared by mixing 0.05

g of linoleic acid and 2 g of small NaF crystals (size 5-15 pin) and 0.033

g of dicumylperoxide with 6.5 g of a solution of Tecoflex (3.4 wt %) in

chloroform. A glass mandrel (diameter 1.4 mm) was coated with the inner

layer by immersing it in the suspension and pulling it out slowly in order

to obtain a film of uniform thickness. The coated mandrel was put in an

oven at a temperature of 150 OC under a dry nitrogen atmosphere for 30

minutes.

For the preparation of the outer layer a 32.5 wt % solution was made of

Estane 570l/poly(L-lactide) (95/5(w/wl) in a mixture of 1.4-dioxane/DMF

(1/3(v/v)). At first the high molecular weight poly(L-lactide) (PLLA) was

dissolved in the refluxing solvent mixture. After cooling the solution

down to room temperature, the purified polyurethane was added and the

temperature was raised to 150 OC. Subsequently, the solution was quickly

cooled down to room temperature and 2 g of sugar crystals (size 30-90 wm)

were added to 3 g of this solution to obtain a porous outer layer.

This viscous suspension was put on the mandrel already covered with the

inner layer. Next the mandrel was rolled over a glass plate that was

covered with a thin layer of small NaF crystals in order to prevent the

formation of a skin and to obtain a pore structure on the outside of the

prosthesis. Two small rings on the mandrel, at a distance of 5 cm, took

care of achieving a wall thickness of 0.7 mm. The polymer mixture was

precipitated in a coagulation bath containing a mixture of ethanol and

water (8/l(v/v) at 20'~. After 30 minutes the prosthesis was stripped off

the mandrel and submerged in water at room temperature in order to leach

out the sugar and NaF crystals. The prostheses were left for several days

in the distilled water which was renewed several times before

sterilization and implantation.

Characterization of the prostheses

A scanning electron microscope (ISI-DS-130) was used for examining the

microstructure of the prostheses prior to, as well as after implantation.

Stress-strain measurements were carried out using an Instron 4301 tensile

tester, equipped with a 1 N load cell at a cross-head speed of 12 mm/min.

Cross-sections of the prostheses were determined using a Profile Projector

(Nikon, model 6C, Japan) with an accuracy of 0.005 mm.

Wet samples were subjected to mechanical deformation in their longitudinal

direction and the distance between the clamps was 15 mm. Extraction of the

sol fraction was performed in an excess of tetrahydrofuran at room

temperature for 72 hours. The porosity of the prostheses was determined by

density measurements. The radial strength of the prostheses was tested by

clamping one side and filling it with water to which a pressure of 2 m

water was applied. In animal studies only those prostheses were used that

could withstand this pressure and did not exhibit leakage as a result of

structural defects.

Implantation

22 prostheses were gas-sterilized with ethylene oxide according to

standard procedures. The in vivo experiments were performed by resecting 1

cm of the abdominal aorta of rats and replacing it by the two-ply

artificial blood vessels.

Results and discussion

General design considerations

This section describes our approach to the evaluation of the basic design

parameters for an artificial blood vessel. The development of such a

prosthesis is primarily an interfacial problem [23] because the luminal

side of the conduit is in contact with streaming blood and the outside

interferes with surrounding tissue which initiates vascularization and

penetration of cells as well as biofragmentation. Hench and Ethridge [231

developed the theory that the properties of an implant should match as

closely as possible those of the organ or tissue that has to be replaced

in the body of human beings or animals. This "natural approach" implies in

the case of an artificial blood vessel that their mechanical properties

should correspond to those of the arteries to be substituted and that the

inside should not cause any thrombus formation. In the natural artery this

is ideally effected by the endothelial lining at the bloodstream

interface. If the vascular prosthesis is not completely covered by

endothelial cells, the risk remains of thrombus formation with ultimate

occlusion. Endothelial cells grow from the anastomotic sites and are

deposited from the flowing blood if sufficient sites for anchoring are

available 181. Particularly in the case of human beings whose diseased

blood vessels measure more than 1 cm in length, quite some time will be

required before the entire luminal side is covered by smooth muscle cells

and endothelium and additional precaution has to be taken concerning

antithrombogenicity and cell adhesion. No material is completely

thromboresistant and also the mechanism of thrombosis prevention is

unknown. We have chosen for introducing COOH groups by cross-linking

polyurethanes with dicumylperoxide in the presence of linoleic acid

because negative charges in general increase the thromboresistance due to

the fact that blood platelets are also negatively charged. Another

advantage of the COOH groups on the luminal side of the prosthesis is that

they strongly promote the adherence and spreading of cells. The

antithrombogenicity may be further enhanced by adsorption of albumin

133,341, the protein that occurs so abundantly in blood. This protein is

known to bind strongly to fatty acids by interaction of the COOH groups

with the free NH groups of the lysine residues. Evidence for this 2

interaction has been obtained from studies of NMR relaxation times [291,

although it is believed that hydrophobic interactions at certain binding

sites may be dominating factors 1301.

Another advantage of the carboxyl groups on the luminal side is that they

promote strongly the adherence of cells which often only grow and

proliferate when attached to a suitable surface [26,271. This is the

reason why this cross-linked inner layer has also been made microporous,

thereby increasing the adherence and spreading of cells as well as

permitting transport of nutrients.

A second reason for putting chemical cross-links into polyurethanes is to

eliminate the serious limitation of the stress softening or hysteresis

which occurs especially in cyclic loading. It is to be attributed to the

deformation and rupture of the hard segment domains. This cyclic

creep-failure, invoked by the arterial pulsation of the blood, may lead to

the formation of aneurysms, the catastrophic dilation of the prosthesis.

Chemical cross-links also impede the structurization of the hard segment

domains on which the platelets seem to adsorb preferentially L24.251,

thereby inducing blood clotting. Furthermore, the cross-links improve the

fatigue life of the prosthesis and diminish the possibility of the

occurrence of environmental stress cracking.

A third reason for selecting the chemical network structure of the inner

layer is that the biofragmentation might be retarded. The implantation of

a foreign device as well as cellular damage will elicit an inflammatory

response and the release of hydrolases which are enzymes from lysosomes

that cause proteins, lipids and other biomolecules to fragment [281. These

enzymes may be actively involved in the phagocytosis of the artificial

blood vessel. But acids are also highly efficient catalysts in cleaving

the urethane linkages t5-7,35, 411.

In summary, the microporous inner layer of our proshesis is supposed to be

highly antithrombogenic and has an improved creep resistance that is aimed

atpreventing aneurysmal dilation, thereby resembling the intima and media

of a natural artery. A second outer layer is made of

polyurethane/poly[L-lactide) mixtures with rather large pores for

vascularization and cell ingrowth so that fibrohistiocytic tissue may

develop, as in the adventitia of natural blood vessels. This two-ply

prosthesis will have better anisotropic elastic properties matching closer

those of arteries in vivo.

Construction and properties of the inner layer

The inner layer of the prosthesis was prepared by immersing a smooth glass

mancirel i n a 7 . 4 %. s o l u l i on r l f 1 he cycl.o;il i phdt i c 1 yu[ c l l l ; l r~c , 1-<!cot i cx ,

I n chlnrofor-m c c ~ n : a l n i n ~ 30 % s l , , l l l NaF crystals I*l; ivlny, ;lirnrnsior.s in tbc

Irsngc 5-15 p m l , dic~irnylp(+t-ozlidf, 2nd 1 i ~ l o l e i c ;#:id.

T l w l a y e r thickr!css and u n i f o r ; r ~ i t v was a c h l e v c d by pulline th*= mandrel

slow1 y out n f t h ~ pfl l ymeric sulpr 'nsior l arir l t l l r - r1ilorr)i { r r fn W;~S 1 l r v w u ~ ~ l t c )

e v d [ ~ u i d t ~ Ily p u t t i l l g the ~ u v ~ r t - r I rnantlrcr i n an o ~ c f i a t 1 5 0 " ~ f o r 313

r n i n u t ~ s . I n t h i s w a y th? polyrrrr ct.3.n:: w c r : c~-r~~-.s-l i1tkc.c i n ;I staye o f

r'rlndomnrss, i . c=. . wi thuut the formation o t I.hr d r m ~ i r ~ c ; n f the hard

s~mgrnen ts .

A f t c r the synt .hcs i s of thc chcrniual n ~ t u c l r k I lhe s11;lll FIaF c r y s t a : ~ wprp

l ~ a r h r r l n11t w i t h H O and a micrclmrous s t r - u c t r ~ r ? r*mll i !br.d a s is shown by

the s c a n n i r . ~ electr-an micru,,l'ny~li of' 1 i g . 1 [-?l ,JI l 1111s microcraph a l s n

reveals C h d L L t i ~ ? surracc i s n o t rnmplrl = ! y : < m o o t h t n ~ t ha: a certain

r o u ~ h n ~ s s w t ? i r l r h a s a i a v o u r a b l n pffpr t rmii t h~ rt'l l ; I ~ I I I I T i o r l and possibly

also or1 t t ~ c flow beh~viuur of t h ~ uIood

Fie: I SFM mi crop rapt^ o f t l ~ r ml I:!-aporous skrur tun; of the inner layer of

t n ~ two-ply v n s c u l a r prozt hrs:,; p r r p a r r d 1 1 om ,I 3 . 4 X solut ~ n n of

cyclnal ipha t ic. pol y~lrr t l i : r lh? Tet <milex i n c : ~ 1t11.ofc)1 411 an11 30% ?ma 1 1 NaF

C I - y s t a l s whj ch were Ieiir1i.zd o l ~ : w l t ! ~ w n t r r

I

Fig. 2. Stress-strain dependance 6 0 0 -

of the cycloaliphatic

polyurethane Tecoflex (A) and of

the microporou~ polymer (porosity

35%) (B). Sample (C1 has a

porosity of 55%.

0 100 2 0 0 300 LOO

straln E l o l o I

0 0 200 LOO 500 800 1000

Fig. 3. Stress-strain dependance of the microporous ( 5 5 % ) aliphatic

polyurethane Tecoflex (A) illustrating the effect of cross-linking with

dicumylperoxide (DCP) at 150 OC for 30 minutes (B) cross-linked with 7.5

W/W % (DCP); (C) cross-linked with 15.0 w/w% DCP; (D) cross-linked with

20.0 w/w % DCP.

Fig. 4. The gel content of

the microporous (55%)

polyurethane Tecoflex

cross-linked with 15 w/w %

dicumylperoxide as a

function of the linoleic

acid concentration.

These small protuberances may just suppress the formation of eddying

vortices that swirl up from the surface in areas of turbulent flow.

The overall porosity of the inner layer is of the order of 50% and

diameter of the interconnected pores are indeed in the range 5-15 pm. As a

result of the pores there is a considerable decrease in modulus of the

polyurethane as illustrated by the stress-strain curves for specimen with

different degrees of porosity in fig. 2. The elastic modulus scales +2

approximately as 4 where 4 is the volume fraction of polymer [36-381. P

Since the polyurethanes and especially the ones based on aliphatic

diisocyanates exhibit a pronounced stress softening, which is bound to

lead to the development of aneurysms, we have attempted to diminish this

disadvantageous property by cross-linking with dicumylperoxide for 30

minutes at 150 OC. This reaction period is six-times the half-life of the

peroxide at this temperature which could not be increased because of the

thermal degradation of the polyurethane. The effect ofcross-linking of the

porous polyurethane is a pronounced reduction in Young's modulus as well

as in tensile strength and elongation at break, as manifested by the

stress-strain curves of fig. 3. Chemical cross-linking of randomized

polyurethane molecules is likely to lower the number of hydrogen bonds

that can be formed and will impede structural organization as hard

segments domains which will therefore lead to a lowering of the Young's

modulus. The decrease in tensile strength may be due to network

inhomogeneities but also degradation may have contributed to this marked

fall. In order to verify whether chain scissioning did occur, sol-gel

analysis were performed by swelling the networks in tetrahydrofuran. For

the polyurethane samples that were cross-linked with 15% by weight of

dicumylperoxide a gel content of 95% was attained, suggesting that, in

view of the rather broad molecular weight distribution, molecular

degradation does not seem to be a predominant factor. The cross-linking of

the polymer chains has been found to take place by recombination of free

radicals on the nitrogen and the carbon atom adjacent to the chain oxygen

l43.451. Both are losing a hydrogen atom by the dicumylperoxide fragments.

The cross-linking efficiency could be enhanced by adding linoleic acid

which contains two double bonds and mainly served to introduce carboxyl

groups for improving the antithrombogenicity. Fig. 4 shows that adding the

linoleic acid resulted in an increase of the gel content and at 20 %

linoleic acid a gel content of 100% was achieved, indicating that chain

scissioning did not occur. Supplying linoleic acid also had a favourable 2

effect on the stress at a strain of 300 % and increased from 50 N/cm to 2 70 N/cm as shown in fig. 5. In order to acquire some indication of the

creep and fatigue resistance the uncross-linked porous polyurethane sample

and one that was cross-linked under optimum conditions with 15%

dicumylperoxide and 20% linoleic acid, the testing-samples were stretched

20-times to a strain of 50% and allowed to recover for 2 hours.

Subsequently the stress-strain measurements on both samples showed a

permanent deformation of 16% for the porous polyurethane (fig. 61, whereas

the network exhibited still some hysteresis but no creep at all (fig. 7).

The most pertinent result of the cross-linking of the polyurethane in the

presence of linoleic acid is the marked increase in tensile strength which

is equal to that of the uncross-linked porous polyurethane as illustrated

by the stress-strain curves of fig. 8. Another salient feature is that the

stress-strain curve of the cross-linked porous polyurethane is concave and

therefore much closer approaches that of the aorta than the pure

polyurethane.

Fig.5. The stress at a

strdin of 300% for the

microporous (55%) polyurethane

Tecoflex cross-linked

with 15 w/w % dicumylperoxide as

a function of the linoleic acid

concentration.

. z -

J

Fig. 6. Stress-strain curves

for the uncross-linked

microporous (55%)

cycloaliphatic polyurethane

Tecoflex illustrating a permanent

deformation of 16% after being

loaded 20-times up to a strain of

50%.

0 2 0 40 6 0

strain E ( ' 1 0 )

I 12 -

I 1 Fig. 7 . Stress-strain curves for

the cross-linked microporous (55%)

1 cycloaliphatic polyurethane

Tecoflex after being loaded

20-times up to a strain of 50%.

There is no permanent deformation

and only a little hysteresis.

0 20 40 6 0

s t r a ~ n E t o l o )

Fig. 8. Stress-strain behaviour of

the microporous (55%) polyurethane

Tecoflex cross-linked with 15% DCP

and 20% linoleic acid ( A ) compared

with that of the uncross-linked

microporous polymer (B).

0 200 4 00

strain E ( ' l o )

Fig. 9. S M rnic:-ograpll ol t , lc . twcl-ply a r t i T i r i - + I L > I o , d w ? s ~ c I cornposed of a

cross-linked m i u r u p u r - o u s pulyui -et l~ane i n n e r layer and a- ouLer layer- wi t l ~

larger pores made af a physical mIxtur c o f a polyesturethane and

polytl-lactide) I95/51.

F i g . 10. SEM I : ~ ~ L I U ~ I clpli uf d c ush-tre~ - i o ~ i UI t 1 l r dl t i f i l i a 1 1,lood vessel

showing small pores 111 t h e polyurethane/poly(L-lactide) mixture I n

addi t i n n t o l a r g e p o r t s which o r i g i n a t e d irvrn leaching o u t the suRar

crystals.

Flg. 11. StM micrograph of t ~ c a r t l f l c la l blood vr:ssel I 1 l u s t r a t ing t h e

firm connection between the IOU-pm-thick i n n e r l a v c I- and t h r o u t e r layer

uith the large pcres.

50 - F i g . 12. " t r e s s - 5 t r a i n behnv i v u r uT

t l ~ e u u l e r l a y r - : . The cor,cavr cu rve A - LO - repses-nts t h c hehavicur 01' t h r

physical mixture of pn 1 y u r ~ t h a n r and

pnly(L-lactide) (q5/5) dissolved i n a

dioxane/DMF mixturc at 150 "C and

guencf i~d t o rocn I c m p e r i i t l l ~ ~ . C u r v y &

describes the strr%s+straln f o r t h ~

sanr polvmer m l x l u r e dissulvcd A ? 70'~

0 20 i D 6 0

5tra1n E ( ' 1 0 )

Construction of the outer layer

The optimal dimensions of the pores in the outer layer should be in the

range of approximately 25 to 150 pm. Previous studies have shown that a

homogeneous pore structure of this size facilitates cell ingrowth and

vascularization. Instead of applying a dip-coating technique which had to

be repeated many times in order to acquire a layer of sufficient thickness

a salt leaching method was developed. At first a rather viscous solution

was prepared of polyurethane and poly(L-lactide) in a mixture of

1,4-dioxane and dimethylformamide (1/3 ratio) to which sugar crystals

(30-90 pm) were added. The mandrel covered with the under layer was rolled

in this suspension and subsequently the polymer mixture was precipitated

in a coagulation bath of ethanol /water (8/1). The sugar crystals were

dissolved in water leaving the required pore structure. Fig. 9 presents a

scanning electron micrograph (SEM) of the two-ply artificial blood vessel

and fig. 10 gives a scanning electron micrograph at higher magnification,

which reveals that in addition to the large pores originated from the

leaching of the sugar crystals the polymeric material also contains very

small pores which are of importance for the fragmentation and gradual

degradation of the prosthesis. The connectedness between the inner layer

and the outer one is clearly demonstrated by the scanning electron

micrograph of fig. 11. The significance of the 5% of high molecular weight

poly(L-lactide) in the mixture with the polyurethane is essentially that

both polymers are highly entangled in these concentrated solution which

hinders the formation of hard segment domains and consequently increases

the rate at which the material can fragmentate. This effect is also

manifested in the stress-strain curve which turns out to be somewhat

concave for the polymer mixture dissolved in dioxane/DMF effectively at

150 OC and subsequently quenched at room temperature, whereas the curve

for the polymers dissolved at room temperature is convex (fig. 12).

Therefore, it is very essential that the polymers are dissolved at high

temperature in order to make the urethane linkages more accessible for

hydrolytic and enzymatic fragmentation which basically determines the rate

of cell ingrowth and development of the fibrohistiocytic tissue.

Biological ~valuat icra

TIE a r t i f i c::ll ; I T - l PI-ier; wcrc i r r . ~ l 2 n t u d i n 7 2 I-at ,s \ ) y rrhquvt-t in^ 1 cm of t h e

abdominal a c r t x artd - i : i l t y t t ~ t i v hy weans of

m i c r o s u r g l c n l t r>r -hnlques Up t o s>np yr:br ~ 1 - thc ?.I t i f i c i a l blood vvssels

wcrr patcnt [ I , ' ] .

R p ~ a f - e n t l y , t h e i nt:nrpi,raL i o n of t h e ! !n:>l P i I- a c i d 1nt.c thr i r m ~ 1 l ~ l l ayer

p r o v i d ~ d good a n t u +hi o n i ~ n g e n i c i t y . Two J ) I - { I ? ~ h ~ s e s w h i c h were ~ n a r ~ u f ' a c t u r ~ d

w1l.t) a lhinner o u t e r !dyrt 'xV. i lr i tc~d 6 slicht dilation as biell as nncu!-ysm

for-nl:~t ion af tcr 4 ucck:;. Ttic s:rc:%-7t.r ,!. 11 cu t ves oi' I t , rsr? t w r ~ p r u ~ l llebes

pr i c l r to implantntion a r e ccm~arcd with '.h,qt. cnf Lhu dhriominal aorta nt~ri of

a g~ a f t , wi t h Ills? r o r r ? c t lzyr-l- l . l i i c k t ~ ~ c ~ in I - i ~ . 1 3 . Arlci:rysm format 1011

can be avoided if the s t ress of ttu- als2arnirt;il L I I J I ~ L I illld Ih:tt the

proslhesis a r e ~ r q 1 1 ; . + 1 at. a s t , r ; . l~~ u i ?,(:'%.. TI , < h n ~ j l d tbc re:narkt?rl t 1 1 ; l i t h e s e

arleut ysrr.;~ 1 r j i l a l l o n s wcrc wlly o t r e r v c d v i ::lual l y and :!LC l,t!mn i n i r l ~

prosthesr:s wcrc found t o 1 ~ v s Lhc original rilan~el.nr- ii!; is i . l l ( ~ s t ~ - a t e d by

the ~ h n l u r n i c ~ i l p r ~ p h of t h ~ two-p ly val;c;llar ~ r n s t . h ~ s i s u n e year a r t e r

implantation ( f i g . 1 4 ) .

F i g . 15. SEM 1nicrograpl.1 of- tl;t-: l~rntlrial su~-f:ict- somr:wi~vr-c j n thr! m i ~ l r j l e of

the two-ply vascular prosthesis a t I n c yrar a l t e r implantatlon i n t n I h r

abdominal x o r t a Q T a ra i . l l . 1 1 ~ the ~ m n o t h r l > d o t t 1 ~ 1 i a l J i n ing .

M~vnification 1500 X I .

The luminal surface oi the p r o s t l ~ c s ~ s t u rned o u t t o be cornplrtcly covered

with endothelial cells, as demonstrated by t h e scanning e lect rnn

micrograph prescntrd i n Tig. 15.

I ~ i s t o l o g i c a l s t u d l e s of longitudinal and transverse cross-sections of the

prosthesis 1 year a f t e r i r n p l a n i a t i v ~ ~ rrvcalcd t h a t i r ~ a ~ l d i l i r j n t v the

complete endothelial l i n i n g making u~ a neo-int ima also smooth muscle

cells and elastine were formed. r a t h e r firmly attached t o the inncr l a y e r

nf t h e prosthesis [ f i g . 16) .

Thc light micrograph of a transverse cross-section presen ted i n rig. 17

~110~5, i n addition t o the neo-intirna 2nd neo-media, a l s o the gen~ratlon o f

flbrohystiocytlc t issue in the oute r l a y e r . Furthermore. it discloses the

presence of cap i l l a r ips and f r a ~ n e n C a l i o n of the polymeric prosthesis

material. T h i s ]pads to the conc l i l s i on t h a t one year after implantation

the deve loprn~nt of a neo-adventitia hes a l s o s tar ted .

F i g . 16. LighC micrograph of a l o n g i t u d i n a l scction D T t h r two-ply vascular

prosthesis at one year a i t r r implantatiun i n t u the ai~rlurnilbal a u r t d o f a

r a t . The inner d e n s e Xaycr, P i , artd bht. n u t e r loose layer, Po, have been

made v i s i b l e by staining u i t h Sudan black B 1441. Arrow marks weak spot

in the inncr layer The prosthesis is lined w i t h a neo-intima 111 and

neo-media (MI magnification 100 x.

F i c . 1 7 . L i c b t m l c r o ~ r d p i ui .j I r a r ~ 5 v p r ~ ; r r ~ n = ; - ~ : ~ c ' i o n n f the two-ply

vascular p r o s t h x i ~ a t OIIC j c l r a f t r r i~pl?ntltiun i n t o t h e ahdomlnal

a o r t a of a r a t . The p r o s t h e s ~ ? ha7 I hv i u r m clt a r1r.u-..r tcr y i n which t h r e e

s e p ~ r a t e layprs can h~ rrcngnizrd. At7 I t i t ~ r r I l l 1 layer, I .

[ l ieu- int i m a Subintima 1 l ay ; r? of r m o o l h nusclp cel 1s. M, In~o-mccl i a l

c 0 n t a i n i r . g r l n s t i n , F Fi Prnhyst iocytic t iss-1s l k u ~ h a s orrani r ~ d t h r

Aeslntegrating pros t,hesis, A , (neo-advent 1 t l a). Nrl t r - t i l ~ ~ T C T P T I L C ~ l f

cnpillarirs. C . anrl ~ ~ r n ~ t l ~ r t j r matcrial, I \ i i r l t l c neo-advcntltla

Hn~nificat ton 7(J r ) x

Conclusions

The two-ply microporous, biodegradable, and compliant artificial blood

vessel described in this chapter has proved to function adequately as a

scaffold for the generation of a neo-artery in rats. The cross-linking of

the microporous polyurethane inner layer with dicumylperoxide in the

presence of linoleic acid has substantially improved the

antithrombogenicity of the prosthesis. This may be due to the fact that

the fatty acid moieties extending into the lumen may tightly bind to

albumin which prevents the platelets from being adsorbed.

Carboxyl groups also promote the adherence of cells which resulted in a

firm connection between the smooth muscle cells of the neo-media and the

inner layer of the prosthesis.

Another salient feature is the pronounced increase in tensile strength of

the porous polyurethane when the cross-linking efficiency is enhanced by

adding linoleic acid. This strength-increasing factor has substantially

diminished the risk of ameurysm formation.

References

1. S. Gogolewski, A.J. Pennings , Advances in Biomaterials, 5, 313 (1984)

2. B. van der Lei, E. Lommen, C.R.H. Wildevuur, J.W. Leenslag, S.

Gogolewski, A. J. Pennings, H.L. Bartels, P. Nieuwenhuis, C.E.

Hulstaert, E.H. Blaauw, F. Dijk, I. Molenaar, Proc.of Polymers in

Medicine and Surgery, 5, 19/1 (1986)

3. B. van der Lei (1986) Ph.D. Thesis, Groningen

4. J. W. Leenslag (1987) Ph.D. Thesis, Groningen

5. M.D. Lelah. S.L. Cooper, "Polyurethanes in Medicine", CRC press Inc.,

Boca Raton, Florida (1986)

6. R.E. Marchant, Q. Zhao, J.M. Anderson, A. Hiltner, Polymer, 28, 2032

(1987)

7. Q Zhao, R.E. Marchant, J.M. Anderson, A. Hiltner, Polymer, 28, 2040

(1987)

8. J. F. Schoen (19861, in: "Handbook of Biomaterials evaluation", Ed.

A. F. von Recum, Macmillan Publ. Comp. , New York S.G. Eskin, Biomed. Mater. Res. Symp. Trans., 1, 37 (19771

C.R.H. Wildevuur, B. van der Lei, Y. Xue, G.H. van Oene, J.M.

Schakenraad, (1986) Life Support Syst. Suppl. ESAO Proceedings

S. Gogolewski, G. Galetti, G. Ussia, Colloid Polym. Sci., 265, 774

(1987)

D. J. Lyman, F. J. Fazzio. H. Voorhees, G. Robinson, D. Albo Jr., J.

Biomed. Mater. Res.. 12. 337 (1978)

Y. Ikada, Proc. of Polymers in Medicine and Surgery, 5 , 6/1 (1986)

J.G.F. Bots, L. van der Does, A. Bantjes, Proc. of Polymers in

Medicine and Surgery, 5, 10/1 (1986)

P. J. Klopper, Eur. Surg. Res., 18, 218 (1986)

C.H. Bamford, I.P. Middleton, Proc. of Polymers in Medicine and

Surgery, 5, 9/1 (1986)

B. van der Lei, H.L. Bartels, F. Dijk, C.R.H. Wildevuur, Trans. Am.

Soc. Artif. Int. Organs, 31, 107 (1985)

J. W. Leenslag, M. T. Kroes, A. J. Pennings, B .van der Lei, New

Polymeric Mater., 1, 111 (1988)

J. M. Schakenraad (1987) Ph. D. Thesis, Groningen

P.B. van Wachem (1987) Ph.D. Thesis, University of Twente, The

Netherlands

P. N. Sawyer, Z. Sophic, 0' Shaughnessy (19861, in: "Handbook of

Biomaterials testing", Ed. A.F. von Recum, Macmillan Publ Comp, New

York

C.L. Kampen, D.F. Gibbons, J. Biomed. Mater. Res., 18, 165 (1984)

L.L. Hench, E.C. Ethridge, "Biomaterials, an interfacial approach",

Academic Press, New York (1982)

G. J. Picha, D.F. Gibbons, R.A. Auerbach, J. Bioeng., 2, 301 (1978)

T. Matsuda, H. Tanaka, K. Hayashi, Y. Taenaka, S. Takaichi, M. Umezu,

T. Nakamura, H Iwata, T. Nakatani, T. Tanaka, S. Takatani, T. Akutsu,

Trans. Am. Soc. Artif. Intern. Organs, 30, 353 (1984)

E.A. Vogler, R.W. Bussian, to be published

W.S. Ramsey, W. Herth, E.D. Nowlan, N. J. Binkowski, Vitro Cell Dev.

Bio., 20, 802 (1984)

R. Sharma, in: "Proc. Conf. Rubber and Rubber-like Materials", Ed. S. K.

De, Jamshedpur p. 479 (1987

M. Bouvier, G. R. Brown, L. E. St-Pierre, Can. J. Chem. , 65, 1927 (1987)

U. Kooystra (1988) M.Sc. Thesis, Groningen

W.V. Sharp, A.F. Finelli, W.H. Falor, J.W. Ferraro, Suppl.

Circulation, 29, 165 (1964)

E. Pollock, E. J. Andrews, D. Lentz, K. Sheikh, Trans. Am. Soc. Artif.

Intern. Organs., 27, 405 (1981)

P.G. Koutsoukos, C.A. Mumme-Young, W. Norde, J. Lylkema, Colloid and

Surfactants, 5, 93 (1982)

W. Norde, F. MacRitchie, G. Nowicka, J. Lylkema, J. Coll. Interface

Sci, 112, 447 (1986)

T.E. Lipatova, S.M. Loos, N.N. Mombuzhai, Vysokomol. Soedin, A 1 2 , 2051

(1970)

D.R. Moore, K. H. Couzens, M. J. Iremonger, J. Cell. Plast., 10, 135

(19741

E. Wolsink (1984) Internal report, University of Groningen, The

Netherlands

C.W. Bert, J. Mater. Sci., 20, 2220 (1985)

N.A. Plate, L. I. Valuev, Adv. Polym. Sci., 79, 96 (1986)

L. I. Valuev, V.V. Chupov, N.N. Plate, Makromol. Chem. Macromol. Symp.,

4, 245 (1986)

K.B. Stokes, A.W. Frazer, E.A. Carter, Prepr. 42nd Am. Tech. Conf.

Soc. Plast. Eng. , Washington DC, 30, 1073 (1984)

P.H. Robinson, B. van der Lei, K.E. Knol, H. J. Hoppen, A. J. Pennings,

to be published

43. X. Feng (S. T. Voong), Y. H. Sun, K. Y. Qiu, Makromol. Chem., 186, 1533

(1985)

44. E.A. Hoeksma, M.F. Jonkman, B. van der Lei, Biomaterials, 9, 463

( 1988 1

45. G. Oertel, "Polyurethane Handbook", Hanser Publishers, 1985, page 98

Chapter 6

Autoclavable highly cross-linked polyurethane networks in ophtalmology

Summary

Highly cross-linked polyurethane networks have been prepared by the bulk

stepreaction of various low molecular weight polyols and (cyc1o)aliphatic

diisocyanates. All these polyurethane networks were optically transparent,

colourless, amorphous glassy thermosets. The properties of the glassy

polyurethane, obtained from the bulk reaction of a tetrafunctional

secondary aminoalcohol tetrakis(2-hydroxypropy1)ethylenediamine or Quadrol

(containing an internal tertiary amino group, that can catalyze the

urethane reaction) and hexamethylenediisocyanate (HDI) in stoichiometric

proportions, have been investigated in more detail. This glassy

polyurethane, with an ultimate glass transition temperature of 85 OC, and

a very low degree of swelling in chloroform (1,271, exhibited good

ultimate mechanical properties (tensile strength 80-85 MPa, elongation at

break ca. 15 %, modulus ca. 1,s GPal. Infra-red spectra of these

hydrophobic polyurethane networks (water uptake ca. 1 %) revealed the

absence of an isocyanate absorption, indicating that all isocyanates,

apparently, had reacted during the cross-linking reaction. Preliminary

experiments and suggestions to increase the hydrophilicity of the networks

have also been described.

In contrast to poly(methylmethacry1atel [PMMA), which has been used

successfully as an intraocular lens material the last 15 years, these

transparent cross-linked polyurethanes can be sterilized simply by

autoclaving. The possiblity of an autoclavable lens is especially

interesting for use in eye surgery in the developing world where the

majority of the blind people live. These highly cross-linked

Quadrol/HDI-based networks, after being autoclaved, were implanted in

rabbit eyes, either in the form of small circular disks or in the form of

a keratoprosthesis (artificial cornea). It was shown that the material was

well tolerated by the rabbit eyes. A serious opacification of the cornea,

a direct result of an adverse reaction to the implant, was never seen.

Even one year after implantation of a polyurethane keratoprosthesis the

eye was still "quiet" and these findings were comparable with the ones

obtained from implantations of keratoprostheses made of PMMA or stainless

steel/glass. These results show that the transparent highly cross-linked

polyurethane network seems suited for use in ophtalmic applications, like

autoclavable intraocular lenses or keratoprostheses.

Introduction

It is estimated that cataract, i.e. an opacification of the crystalline

lens of the eye and the main cause of blindness, is responsible for

approximately 20 million blind people world-wide; most of them live in

developing countries in Asia or Africa (1-4). It is now recognized that

cataracts are not only another sad consequence of ageing. Other possible

risk factors include malnutrition, sunlight exposure, smoking. This has

led to the hypothesis that oxidative damage plays a major role in

cataractogenesis (5,6). Surgical removal of the cataractous natural lens

is the only medical treatment available for cataract patients. Spectacles

or contact lenses used to be the conventional way of replacing the natural

lens, but turned out to be far from ideal. However, implantation of an

intraocular lens in the place of the removed cataractous lens is presently

the best way to correct aphakia and to visually rehabilitate the cataract

patient (7.8). Modern intraocular lens implantation, using artificial

lenses made of poly(methy1methacrylate) (PMMA), had its start in 1949

after the Second World War during which Ridley, an English eye-surgeon who

performed the first PMMA lens implantation, noticed that Perspex splinters

of canopies of airplanes caused no irritations in the eyes of pilots.

Since then many intraocular lens designs have been developed and implanted

in cataract patients (8,9,12). The number of PMMA lens implantations has

increased enormously during the last 15 years and nowadays lens

implantation is considered a routine operation (7,101.

It is generally accepted that the artificial lens should be optically

satisfactory, inert, non-toxic, biocompatible, lightweight (glass, for

example, has always been considered as a possible, autoclavable lens

material, but weight problems have limited its use (1211, structurally

sound, durable, ultraviolet light absorbing, resistant to laser treatment

(in case of secondary cataract formation), easily implanted and securely

fixated and of course be sterilized safely (8,111. PMMA meets almost all

of the requirements listed, but does have some disadvantages as a lens

implant material. Although non-toxic, PMMA is extremely damaging to the

corneal endothelium if in contact during implantation surgery (13,18,19).

Endothelial cells adhere to the hydrophobic PMMA and may be stripped off.

Surgical skill is needed to minimize contact adhesion. Due to its rigidity

PMMA may cause mechanical irritation of uveal tissue. Another serious

problem with PMMA lenses has been with regard to sterilization. PMMA can

not be sterilized simply by autoclaving due to its relatively low T (100

OC), and therefore has to be sterilized either with the toxic ethylene

oxide or by sodium hydroxide sterilization. The last method has been

prohibited in the United States by the FDA, but is still used in Europe

(7.9). Both sterilization methods are, unlike autoclaving, not without a

certain risk. According to eye-surgeon Worst there is a need for an

autoclavable intraocular lens, especially for cataract surgery in

developing countries (14). Research in this field has led to the

development of other lens materials that are autoclavable. On the one hand

amorphous, aromatic thermoplastics with a very high glasstransition

temperature, like polycarbonate, poly(ether)sulphone, polyimide, fulfil

this requirement (7,11,15,16). On the other hand, polymeric networks can

be used as autoclavable intraocular lens materials. Hydrophobic silicone

elastomers and hydrogels (polyHEMA), which have been used in other medical

applications, e.g. soft contact lenses (171, for a long time, are examples

of polymeric networks which have been considered and evaluated as

potential, clinical intraocular lens materials (11,18-23). Both materials

are elastomeric and can be folded, which means that artificial lenses made

of these materials can be inserted via a smaller incision into the eye

than in the case of glassy polymers. Hydrogels are hydrophilic materials

having a soft consistency and are known for their soft tissue

biocompatibility (24 ) . So, these materials are expected to be less

damaging to the eye, especially to the corneal endothelium.

Besides opacification of the eye-lens, the cornea may become

non-transparent, leading to so-called corneal blindness. Throughout the

world about ten million people, mainly living in the developing world,

suffer from corneal blindness, for example from trachoma (2). If a corneal

graft is not available, due to a lack of donor eyes, or not indicated,

implantation of a keratoprosthesis is the only possibility to restore

sight (25). A keratoprosthesis (an artificial corneal is usually made of

the same materials that are used for intraocular lenses. The requirements

for an intraocular lens material and a keratoprosthesis material are

virtually the same.

In this chapter we will describe the synthesis and properties of glassy,

highly cross-linked polyurethanes and their potential application in

ophtalrnology as an autoclavable ocular implant material. Polyurethanes,

which in general are relatively biocompatible and used in many medical

applications (261, have never been considered as materials that might be

used in ophtalmic applications. All this motivated us to synthesize a

series of new, densely cross-linked polyurethane networks by stepgrowth

polymerization of low molecular weight polyols and (cyc1o)aliphatic

diisocyanates.

Experimental

The polyols used in this study were:

tetrakis(2-hydroxypropyl)ethylenediamine (Quadroll,

triisopropanolamine (TIPA, mp. 48-52 O C ) ,

triethanolamine (TEA),

tetrakis(2-hydroxyethy1)ethylenediamine.

bis-N,N-(2-hydroxyethyl)isopropanolamine (BHEIPA),

tetrakis(2-hydroxyethyl)methylaminomethylmethane,

octakis(2-hydroxypropyl)pentaerythrityltetraamine ("octaol"),

trimethylolpropane (TMP, mp. 60-62 O C ) ,

pentaerythritol (mp. 260 OC),

glycerol,

2.2-bis(hydroxymethy1)-2,2', 2"-nitriloethanol (BIS-TRIS, mp. 104 OC).

The chemical structures of these polyols are shown in figure 1. All

polyols were liquids at room temperature, unless a melting point is

mentioned in brackets, and were purified, if possible, by distillation

under reduced pressure. Most polyols were commercially available. Two new

polyols were synthesized as described below.

Tetrakis(2-hydroxyethyl)methylaminornethylmethane

A stirred mixture of 1 eq. pentaerythrityltetrachloride and 9 eqs.

N-methyl-ethanolamine was refluxed for 8 days under a nitrogen atmosphere

at ca. 165'~. Excess N-methylethanolamine then was distilled off and to

the residue ethanol and 4 eqs. powdered potassium hydroxide were added.

After stirring, ethanol was removed and the residue was extracted with

chloroform. After removal of the solvent, fractional vacuum distillation

yielded the product, a viscous, colourless liquid, bp. 148-155 O~/0,007

mbar. Analysis calculated for C H N 0 C 56.04, H 10,99, N 15,38. 17 40 4 4'

Found: C 56,27, H 10,76, N 15,49.

Octakis(2-hydroxypropyl)pentaerythrityltetraamine was prepared in the same

way as the previous compound from pentaerythrityltetrachloride and

diisopropanolamine. The octafunctional polyol was isolated, in poor yield,

as a yellowish, very viscous liquid, which could be decolorized by using

activated carbon. Bp. 200 '~/0,005 mbar. Anal. calcd, for C H N 0 : C 29 64 4 8

58,39, H 10,74, N 9,39. Found: C 58,24, H 10,46, N 9,33.

The polyfunctional amines used were pentaerythri tyltetraamine (also known

as tetrakisaminomethylmethane). tetrakis(N-propylaminomethyl)methane,

tris(2-aminoethy1)amine. In figure 1 the structural formulas are depicted.

The syntheses of the two tetraamines are described below.

Pentaerythrityltetraamine was synthesized from pentaerythritol in 5 steps.

In the first step pentaerythritol was converted to the

tetrabenzenesulfonate according to a published method (27). Following a

patented method (281, the tetrabenzenesulfonate was reacted with

sodium-p-tosylamide in N-methylpyrrolidone solution at 200 OC for 20 hours

to yield the tetratosyl-amide, a compound also described earlier by

Litherland and Mann, who started from pentaerythrityltetrabromide (29,301.

In the third step the tetratosylamide was hydrolyzed with 80% sulfuric

acid resulting in the formation of pentaerythrityltetraamine disulphate.

The next step was the continuous extraction of the disulphate with sodium

hydroxide in benzene to give the tetrahydrate of

pentaerythrityltetraamine. Both last steps were described in the

literature (29,311. Finally, the tetrahydrate was converted to the pure,

hygroscopic tetraamine by azeotropic distillation with benzene. The

overall yield was ca. 60%. Anal. calcd. for C H N : C 45,40, H 12.19, N 5 16 4

O C N + C H ~ ~ ~ N C O

Hexamethylenedllsocyanate (HDI)

OCN CH,

OCN NNC0 trans 1.4qclohexanedtisayanate (I-CHDI)

OCN-CH-CyO (Ial \"*F*.

NCO

Lysine dilrocy~me (LDI)

F", FH, HO-CH-CH

\2 ,CHZ-CH-OH

N-CH2-CH2-N

no-CH-cn2 \cH2-cn-oH I I CHI CH,

FH, CH,-CH-OH

HO-CH-CHz-N

\cH,-cH-oH

CHI

CH.-OH I

HC-OH I cn2-OH

Glycerol

42.37. Found: C 45,37, H 12,19, N 42,17.

TetrakidN-propylaminomethyl)methane

1 eq. pentaerythrityltetrabenzenesulfonate and 8 eqs. N-propylamine were

refluxed in N-methylpyrrolidone during 80 hours. The solvent was then

removed and 8 eqs. powdered KOH and ethanol were added to the residue.

After stirring for some time, the ethanol was distilled off and the crude

product was extracted with diethylether, which was removed subsequently.

The product, a colourless liquid with a strong odour, was obtained by

fractional vacuum distillation, bp. 130~~/0,028 mbar. Anal. calcd. for

C17H40N4: C 68,0, H 13.33, N 18,66. Found: C 67,90, H 13,37, N 18,37.

The diisocyanates hexamethylenediisocyanate (HDI), isophoronediisocyanate

(IPDI), trans 1,4-cyclohexanediisocyanate (tCHDI) were commercially

available, and ethyl 2,6-diisocyanatohexanoate (lysine diisocyanate, LDI)

was synthesized according to a previously described procedure (32). In

fig. 1 the structural formulas are shown. All diisocyanates were vacuum

distilled prior to use.

Polyurethane network preparation

A polyol and a diisocyanate were thoroughly mixed in stoichiometric

proportions ([NCOI/[OH]=l) at roomtemperature under a nitrogen atmosphere.

The homogeneous, colourless mixture was degassed repeatedly and allowed to

gelate at room temperature. Post-curing at a temperature above T (i.e. gm

the glass transition temperature of the fully cured sample) yielded

optically transparent, glassy thermosets. Cure of the polyurethane

thermosets could also be achieved isothermally at a temperature above T 9'

Alternatively, polyurethane networks were formed very fast when the

mixture was polymerized by means of microwave heating. Within minutes

gelation was achieved. Microwave cure was conducted in an ordinary

domestic microwave oven (microwaves with 2,45 GHz frequency).

All polyaminoalcohols, or poly(hydroxyalkyl)amines, from figure 1 were

colourless, viscous liquids at room temperature, except TIPA which

crystallized very slowly to a waxy solid after melting or distillation.

Prior to mixing this trio1 had to be melted, just like TMP. The latter

compound was mixed with diisocyanates at temperatures 80-110 OC, and also

allowed to gelate at these temperatures. Gelation in these formulations

was rather fast (ca. 0,5 hr. 1 . All diisocyanates from fig. 1 were

colourless liquids at room temperature, except tCHDI, melting at 60 OC.

This diisocyanate had to be mixed with polyols above its melting point.

Gelation at these temperatures is very rapid. Triethanolamine was miscible

with liquid diisocyanates only at temperatures higher than 40-50 'c.

Polyisocyanurate network formation

Hexarnethylenediisocyanate. containing 0,25 wt.-% stannous octoate as a

catalyst, was allowed to polytrimerize at 125 OC for some days. Gelation

at this temperature took about three hours. The glassy, transparent solid

was post-cured at 180 OC.

Characterization

Glass transition temperatures of the polyurethane networks were determined

using a Perkin-Elmer DSC-7, calibrated with ICTA (International

Confederation for Thermal Analysis) certified reference materials, and

operated at a scan-speed of 10 'chin.

Tensile testing was performed on rectangular-shaped specimens (ca. 50x6~2

mm), machined from glassy polyurethane samples, at room temperature using

an Instron (4301) tensile tester, equipped with a 5 kN load-cell, at a

cross-head speed of 10 mm/min. The gauge length was 25 mm. The reported

tensile data are the mean values from at least six tests. Failure was

nearly always initiated at the clamps as would be expected for rectangular

shaped specimens.

Swelling measurements were carried out on polyurethane samples weighing

less than ca. 0,5 g. that were immersed in chloroform at room temperature

for 2 days. The volume degree of swelling was calculated from the weight 3 increase, using the densities of chloroform (p=1,48 g/cm and the

polyurethane networks. Both the Quadrol/HDI-based network and the

TIPA/HDI-based network had a density of 1,135 g/cm3, which were determined

by weighing samples in air and by weighing them submersed in water.

Infra-red measurements were carried out on ultra thin (ca. 10 pm)

polyurethane films with a Bruker IFS-88 FT-IR spectrometer.

UV/VIS transmission spectra of samples having 1 mm thickness were recorded

on a Pye Unicam SP 8-200 UV/VIS spectrophotometer.

The refractive index of a Quadrol/HDI-based polyurethane network was

measured on a thin film wetted with dichlorobenzene (having a higher

refractive index than the polymeric material) with an Abb6 refractometer

( ATAGO .

Implantations

Circular disks with smooth edges having 6 mm diameter and 1 mm thickness,

containing one big central hole and three smaller peripheral holes for

fixation, were constructed from a fully cured Quadrol/HDI-based

polyurethane network sample. After a cleaning procedure, the disks were

sterilized by autoclaving at 120 OC for 20 minutes. Three disks were

inserted into the anterior chamber of rabbit (Chinchilla) eyes through a

corneal incision. Prior to implantation, the eyes had been made aphakic

(lens less). The disks were hung up (fixated) in front of the pupil using

two stainless steel 70 pm wires, led through the peripheral holes. The 70

pm wires were knotted together on the sclera.

One "mushroom"-shaped keratoprosthesis (artificial cornea) was cut on a

lathe from a fully cured Quadrol/HDI-based polyurethane network sample

(see figure 2). After polishing, cleaning, and autoclaving the

keratoprosthesis for 20 minutes at 120 OC, it was implanted by Van Andel

as a "porthole" in a cornea of an aphakic (Chinchilla) rabbit eye and

fixed on the eye like a "champagne cork" on a bottle. The central "column"

of the keratoprosthesis (the "leg" of the "mushroom") perforates the

centre of the cornea through a 3 mm trephined hole. The "hat" of the

"mushroom" (with a diameter of 6 mm) lies on the cornea and is "anchored"

with two permanent 70 pm thin soft stainless steel (type: AINSI 316) wires

around the whole eyeball, in two planes perpendicular to each other. The

keratoprosthesis works now as a valve: the peribular fixation keeps the

valve on the trephined hole and the internal pressure of the eye pushes

the corneal rim around the trephined hole against the back of the "hat" of

the "mushroom". The resulting pressure on the interface between

keratoprosthesis and cornea prevents the leaking of aqueous humour, the

melting away of corneal tissue, and the epithelial downgrowth

(fistula-formation) (33 ) .

Figure 2. A "mushroom"-shaped keratoprosthesis.

Results and discussion

As outlined in the introduction, densely cross-linked polyurethanes might

be interesting materials with potential ophtalmic applications, like

autoclavable intraocular lenses, and keratoprostheses.

First, it will be described how such polyurethane networks can be

synthesized. From the literature numerous examples of elastomeric model

polyurethane networks, formed by the endlinking of hydroxylterminated

prepolymers with polyisocyanates, are known (34-40). In contrast to

elastomeric polyurethane networks, densely cross-linked polyurethane

networks are relatively unknown (41,421. These model networks are formed

by stepreactions, which according to Boots is an intrinsically homogeneous

process, unlike network formation by chain reactions (43.44). In this way

polymer networks with a well-defined topology and a minimal number of

dangling-end network imperfections can be obtained when the cross-linking

reaction is carried out stoichiometrically and to very high conversion of

the functional groups (45). DuSek and Stepto pointed out that the extent

of cyclization in model networks prepared in the absence of a diluent is

rather small, but never negligible. Dilution may considerably increase

intramolecular reactions, leading to elastically inactive loop structures

and may cause the formation of inhomogeneities, or even phase separation

(micro-, macrosyneresis) (39,461.

All this has led to the idea to prepare polyurethane networks from pure,

low molecular weight polyols and diisocyanafes in the bulk and to obtain

highly cross-linked networks that are in principle homogeneous and contain

a minimal concentration of network imperfections, which should result in

materials with good (ultimate) mechanical and optical properties.

It is obvious that a first requirement for a network-forming reaction, in

which more than one component participates, is that, in the absence of a

solvent, the reactive components have to be miscible. Applying this to the

polyurethane network formation this means that the polyol and the

diisocyanate have to miscible. It appeared that not all low molecular

weight polyols from figure 1 were miscible with the (cyc1o)aliphatic

diisocyanates listed. For instance, pentaerythritol or glycerol were not

miscible with any of the diisocyanates at any temperature. No sign of

reaction could be observed. However, in the presence of a solvent (e.g.,

DMF) at room temperature, transparent swollen gels were obtained.

A second requirement is that in the case of a miscible formulation the

reactivity of the components should be low, low enough to achieve complete

miscibility before the polymerization reaction takes place in a

controllable manner. It turned out that some polyols,

tetrakis(2-hydroxyethy1)ethylenediamine. tetrakis(2-hydroxyethy1)methyl-

aminomethylmethane, BIS-TRIS were too reactive. During mixing these

polyols with diisocyanates noticeable polymerization (gelation) took

already place, resulting in a very macroscopically heterogeneous network

formation. It was also seen that low molecular weight polyfunctional

amines, primary or secondary amines, tri- or tetrafunctional (see fig. 11,

which are known to be highly reactive towards isocyanate groups (resulting

in urea bonds instead of urethane bonds in the case of the reaction of an

alcohol with an isocyanate), reacted instantaneously with diisocyanates,

both in the absence and in the presence of a solvent (for instance, DMF,

toluene). So, macroscopically homogeneous polyurea networks could not be

obtained by the direct addition of polyfunctional amines and

diisocyanates, due to the fact that the very fast polymerization reaction

interfered with the molecular mixing. Any attempt to slow down the

reactivity of the amine group was unsuccessful.

The capping or blocking of isocyanates is a way to prevent premature

reaction of the polyurethane (or polyurea) components. Isocyanate groups

are reacted with compounds which form a thermally weak bond. The reactive

isocyanate can be liberated at elevated temperatures [ca. 150 OC or

higher). Examples of compounds used for the blocking of isocyanates are:

phenols, caprolactam, oximes and 6-dicarbonyl compounds (ethyl malonate)

(47-49 ) . Fortunately, there were several low molecular weight polyols, having a

relatively low reactivity, that were miscible with diisocyanates:

tetrakis(2-hydroxypropy1)ethylenediamine (Quadrol), triisopropanolamine

(TIPA), triethanolamine (TEA), octakis(2-hydroxyethy1)pentaerythrityl-

tetraamine ("octaol"), bis(2-hydroxyethyl)isopropanolamine (BHEIPA),

trimethylolpropane (TMP). At first sight it looks like these polyols have

nothing in common, but this is an overstatement. First of all, one can see

that the polyols containing only secondary alcohol groups (TIPA, Quadrol,

octaol) are miscible with diisocyanates. All three compounds are so-called

secondary polyaminoalcohols, or poly(hydroxya1kyl)amines which can be

regarded as the reaction products of the corresponding amines and

propylene oxide. It is noteworthy that triethanolamine can be mixed with

diisocyanates whereas the analogous tetrakis(2-hydroxyethy1)-

ethylenediamine was considered too reactive to be miscible with

diisocyanates to form a homogeneous mixture that can be polymerized

controllably. This may be explained by the fact that the gelation of the

trifunctional primary polyol TEA/diisocyanate formulation is much slower

than the gelation of the tetrafunctional primary polyol/diisocyanate

mixture. The same was observed in the corresponding secondary polyol

systems: TIPA/HDI gelates at room temperature in ca. two days, Quadrol/HDI

gelates at room temperature in ca. 6-10 hours. So in the case of the trio1

the time available for mixing (before gelation sets in) is much longer

than in the tetra01 case. Due to the fact that primary alcoholgroups are

more reactive towards NCO groups than secondary alcohol groups, the

gelation time becomes shorter when in the formulation TIPA/HDI TIPA is

replaced with TEA, or in the case Quadrol/HDI Quadrol is replaced with the

analogous primary tetraol. In the latter case the polymerization

(gelation) interferes with the mixing. The gel time is too short to permit

sufficient mixing. The compound BHEIPA can be considered a polyol with

properties lying between those of TEA and TIPA. Trimethylolpropane,

finally, which also contains three primary OH groups, just like TEA, was

also miscible with diisocyanates above its melting point (60'~) (From the

literature aromatic, highly cross-linked polyurethane networks made from

TMP and MDI are known (41)). In this context it is strange that glycerol,

containing two primary and one secondary hydroxyl group, can not be mixed

with any diisocyanate. From the above, it may be clear that it is rather

unpredictable whether or not a certain pair of polyol/diisocyanate is

miscible.

Figure 3. The mechanism of the tertiary arnine catalyzed urethane reaction.

Another interesting thing that the above mentioned polyols have in common,

except TMP, is the presence of tertiary amino groups in the molecular

structure. Tertiary amines are known catalysts for the urethane reaction

(26,47-49). The mechanism for tertiary amine catalysis involves the

formation of an intermediate active complex of the isocyanate and the

tertiary amine. After the OH group is added, the intermediate complex

rapidly rearranges to the urethane linkage (figure 3 ) . So, these

polyaminoalcohols contain an internal, built-in catalyst (48). No external

catalyst is needed for the polyurethane reaction, which is interesting

with regard to the potential biomedical application of the polyurethane

networks. Added catalysts, for instance toxic tin salts, may be

responsible for complications after implantation.

The whole problem of miscibilty can be avoided by using only one compound

in the formation of a polymer network. An example is the polytrimerization

of diisocyanates in the presence of a suitable catalyst (e.g. stannous

octoate), resulting in polyisocyanurate networks. At temperatures higher

than 120 OC HDI formed a glassy, transparent, densely cross-linked network

with a high ultimate glass transition temperature of 150 OC. This high T 4m

is likely due to the presence of rigid 6-membered isocyanurate ring

structures in the network.

Polyurethane networks were prepared by firstly mixing the polyol component

and the diisocyanate under a nitrogen atmosphere, to avoid side reactions

of the isocyanate group with water which gives rise to undesired bubble

formation (or foaming). After mixing, the homogeneous, colourless viscous

mixture was allowed to gelate (usually at temperatures below Tgm).

Directly after gelation, the glass transition temperature of the network

is equal to the gelation temperature. Due to vitrification the chemical

reactions in principle are quenched. The mobility of the functional groups

in the glassy network that have not reacted yet is very low, which means

that the network formation takes place very slowly. After vitrification,

the cross-linking reaction becomes diffusion-controlled ( 5 0 ) . The samples

were fully cured at temperatures above the ultimate glass transition

temperature of the polyurethane networks T In order to reach a maximum 4m'

conversion of the functional groups, the cure temperature has to be above

the T of the system. The glass transition temperature is a sensitive gm

measure of the functional group conversion (51).

Instead of the conventional thermal curing of the samples, the

polyurethane resins may be cured by means of electromagnetic waves in the

microwave frequency range (2,45 GHz 1. In contrast to thermal heating,

which involves heat conduction and thermal lag associated with it,

microwaves can generate heat directly within the sample and thus offer the

possibility of very fast, uniform curing which should result in improved

physical/mechanical properties of the final material (52). Microwave

heating has been used for the cure of epoxy resins and polyurethane foams

(53).

All polyurethane networks were optically transparent, colourless,

amorphous, glassy materials. The refractive index of a fully cured

Quadrol/HDI-based polyurethane network is 1,50, slightly higher than the

value (1,491 reported for PMMA 7 9 1 1 1 5 1 The networks were

macroscopically homogeneous; apparently no phase separation had occurred

during the process of network formation. In table 1 the ultimate T 's of 9

the densely cross-linked networks, obtained from the bulk reaction of a

polyol and a (cyc1o)aliphatic diisocyanate in stoichiometric proportions,

are collected.

Table 1. Ultimate glass transition temperatures (in OC) of polyurethane

networks obtained from the bulk polymerization of a polyol and a

diisocyanate in stoichiometric proportions

TI PA Quadrol octaol TEA BHEI PA TMP

HD I 75 85 106 33 45 83

LD I - 72 - 34 - - IPDI 160 165 - 133 - 184

tCHDI 178 190 - 125 - 227

The ultimate glass transition temperature of a HDI-based polyurethane, as

can be seen in table 1, increases when the functionality of the polyol

increases (going from TIPA to Quadrol to the octaol), which is due to an

increase in the cross-link density. The T is also markedly raised when gm

an aliphatic diisocyanate (HDI, LDI) in a particular formulation is

replaced with a cycloaliphatic one (t-CHDI, IPDI), which is a consequence

of the higher rigidity in the case of a cycloaliphatic polyurethane

network.

Since all these polyols, and all diisocyanates as well, are miscible, the

TP of a polyurethane network can be varied endlessly. The T of a

4m multi-component system then lies between the T 's of the networks

4m resulting from the individual pair of reactants (two-component system). An

interesting example is the polyurethane network made from a polyol mixture

of TEA and TIPA in a 2 : l mole ratio, and HDI in stoichiometric

proportions. This polyurethane had a T (44 OC) that was virtually equal '3m

to the one of the BHEIPA/HDI-based polyurethane ,as could be expected, and

inbetween the T values of the TEA/HDI-based and the TIPA/HDI-based gm

polyurethane network.

Another way of influencing the T is to carry out the network-forming gm

reaction using off-stoichiometric proportions of reactive groups. For

example, when 1 eq. Quadrol is reacted with 1,5 eqs. HDI, instead of the

stoichiometric 2 eqs, the T of the final network drops ca. 20 OC. In 4m

this case a considerable amount of unreacted alcohol groups (dangling

ends) are present in the network, which are responsible for this lowering

of the T One can also say that the cross-link density is much lower gm'

than in the case of a stoichiometric network formation. Another example is

the drop in T when the octaol is reacted with 2 eqs. HDI instead of the 4m

stoichiometric 4 eqs.. The ultimate glass transition temperature of the

network now drops from 106 OC to 54 OC.

From swelling measurements using chloroform, it could be established that

the polyurethane networks were definitely highly cross-linked, i.e. having

a low degree of swelling. The volume degrees of swelling for the

Quadrol/HDI-based and the TIPA/HDI-based polyurethane network were 1.27

and 1.60, respectively. The higher degree of swelling of the trifunctional

aliphatic polyurethane network compared to the corresponding

tetrafunctional network is ascribed to the lower cross-link density in the

trifunctional network. It should be mentioned that in all cases it was

found that the gel content was 100 % (The drying of the swollen networks

to constant weight took about 3 weeks in a vacuum oven at 100 OC). No sol

fraction could be detected in any cross-linked polyurethane network. It

has been said before that by extraction of tight polymer networks, and

subsequently drying, the gel content can not be determined accurately,

since the actual sol fraction may not be able to diffuse out of the dense

network.

Infra-red spectra of fully cured Quadrol/HDI-based or TIPA/HDI-based

polyurethane networks, both spectra were virtually identical, revealed the

absence of an isocyanate absorption at ca. 2250 cm-l, indicating the

completeness of the cross-linking reaction. Apparently, all isocyanate

groups have reacted, either with the hydroxyl groups (urethane bond

formation) or with urethane bonds to form allophanate linkages. The latter

should not be excluded since the curing was carried out at ca. 100-110 OC,

at temperatures where the allophanate formation becomes competitive with

the urethane formation (54). The isocyanate absorption at 2260 cm-',

however, was not absent in the infra-red spectrum of the HDI-based

polyisocyanurate network. This indicates that in the case of a

polytrimerization, where three NCO groups have to react simultaneously to

form ring-like isocyanurate structures, residual isocyanate groups can be

detected in the final network. Apparently, a complete conversion of the

reactive groups can not be attained.

The Quadrol/HDI-based and TIPA/HDI-based polyurethane networks were

subjected to tensile testing. Both glassy materials showed virtually

identical stress-strain behaviour. Figure 4 shows a typical stress-strain

curve of a Quadrol/HDI-based polyurethane network. The tensile strength of

the two aliphatic polyurethane thermosets was in the range 80-85 MPa. Both

formulations had comparable moduli (1,5 GPa) and comparable strains at

break (ca. 15 % I . The polyurethane glasses usually yielded prior to break.

a phenomenon also displayed by other glassy cross-linked networks, for

example epoxies (55-571, and by glassy thermoplastic polymers, like PMMA,

as well. The plastic deformation of glassy thermosets below T is similar

to that of amorphous glassy thermoplastics (58) . The mechanical properties

of these densely cross-linked polyurethane networks will be discussed in

more detail in a forthcoming paper (591, in which also results of dynamic

mechanical measurements will be presented.

5 10 15 20

Strain (%)

Figure 4. Stress-strain behaviour of a glassy polyurethane network

obtained from the bulk polymerization of Quadrol and HDI in stoichiometric

proportions.

Some densely cross-linked polyurethanes were immersed in water for about

two months to measure their water uptake by weighing. The water uptake of

Quadrol/HDI-based, octaol/HDI-based, TEA/HDI-based polyurethanes were 1.2

%, 1%, and 8 %, respectively. The water uptake of the Quadrol/HDI-based

polyurethane was increased when the cross-linking reaction was carried out

off-stoichiometrically. The residual, polar hydroxyl groups in the

network, obtained from the polymerization of Quadrol with 1,s eqs. HDI,

were responsible for a higher water uptake (3.1 %) of this network. The

same was observed for the octaol/HDI-based network. The water uptake was

increased to ca. 8 % when the octaol was reacted with 2 eqs. HDI instead

of the stoichiometric 4 eqs. In any case it was seen that the uptake of

water dramatically lowered the T of the network, due to a plasticizing g

effect. The T of the stoichiometric Quadrol/HDI network dropped from 85 g

to 48 OC, and the T of the TEA/HDI network dropped from 33 to -20 OC

(resulting in a rubberlike polymer). It is well-known that the presence of

water in polymer networks positively contributes to their

biocompatibility. Hydrogels, usually water swollen polymer networks, are

said to be biocompatible due to a low interfacial tension which may be

exhibited between the hydrogel surface and an aqueous solution (as in

living tissue). An attempt to make the polyurethane networks more

hydrophilic, so that the water uptake would increase, was the

incorporation of low molecular weight polyols containing besides OH

groups, carboxyl groups in their molecular structure:

N.N-bis(2-hydroxyethy1)glycine (bicine), bis(hydroxymethy1)propanoic acid.

These two compounds could be "dissolved" in the polyaminoalcohols from

figure 1 only at high temperatures (ca. 100-150~~). The resulting clear

polyol mixture was then reacted with the stoichiometric amount of HDI, but

the water uptake of the resulting network was not noticeably higher than

in the case of a normal polyurethane formulation. Apparently, the carboxyl

groups had reacted during the course of the experiments, either with the

hydroxyl groups (esterification) or with the NCO groups.

Besides low molecular compounds, linear polymers may be incorporated into

the polymer network. Thus, so-called semi-interpenetrating networks are

formed. For example, poly(N-vinylpyrollidone), a water-soluble polymer,

can be dissolved in Quadrol (at elevated temperatures), and the resulting

polymer solution can be cross-linked with a diisocyanate to form a glassy

thermoset.

PMMA intraocular lenses have been surface modified with covalently linked

heparin to increase the biocompatibility (60,611. A (PMMA) lens surrounded

by a solution of hydrophilic monomers can be 7-irradiated, resulting in a

thin coating of grafted hydrophilic polymer (62). Another method used to

modify the surface of a polymeric material is the glow-discharge technique

(53,631. In this way surfaces can be made more hydrophilic, which might

also work out nicely for the polyurethane networks described here.

Alternatively, the surface of the polyurethanes might be modified to

increase the hydrophilicity by usual grafting techniques or by the two

following methods. Since NCO groups can also react (slowly) with amide

groups, the liquid polyol/diisocyanate mixture may be poured onto a dry

polyacrylamide film, and subsequently cured at high temperatures (ca.

1 1 0 ~ ~ ) . In this very straight-forward manner polyacrylamide may be grafted

chemically to the polyurethane. Another possibility is the "surface

etching" of the polyurethane network with a dicarboxylic acidchloride

(e.g. succinyl chloride) for a short period of time, and then placing the

whole sample in water in order to create carboxyl groups at the surface.

This last method is in slight analogy with the sodium hydroxide

sterilization technique used for the sterilization of PMMA, for instance

(7,9). Local hydrolysis of the methyl ester at the surface through the

action of sodium hydroxide results in the formation of carboxyl groups at

the surface, making the surface more hydrophilic. Although not intended as

a surface modification method, this sodium hydroxide sterilization might

act this way in case of PMMA.

As a pilot experiment, three disks made of the Quadrol/HDI-based

polyurethane network were implanted in rabbit eyes to see how the material

was tolerated in the eye. All three implantations, basically, gave the

same results. A serious opacification of the whole cornea, indicating an

adverse reaction to the implant material, was never seen. The cornea was

nearly completely clear two weeks after the surgical procedure. Only near

the site of the corneal incision, the cornea was a little hazy, and

starting from that site some vascular ingrowth, which was likely due to

some surgical trauma, could be seen. Figure 5 shows a photo of a rabbit

eye c0ntainin.q a c i r cu l a r d l s k m ~ d e o r b h ~ 1 1 3 ~ h l y rross-linked

polyurethanp tun u c c k s af t r r ~ m p l q n t a t i o n . From t h e ~ c p r c l l r n i n a r y

experiments i t may LC voncl~d-d t h t t h e !~ighly cr uss-1 inked pwlyur e t h a n e

material was rattler well tolerated i r k [lie r a b b i t eye. The material d i d n o t

t u rn out Lo be acute toxic and seems su l led for use i n ophtalmic

applications. These encouraging results have l e d t o t h r addi l iona l

implantation nf a k~ratoprostheris l n a r a b b f t e y e .

The preliminary experiment with onc lathe-cut keratoprosthesis uT t h e

prradro:/IIDI-based p o l y u r e t h a n n nclwurk implantrd i n on? Chinrhllla-rabbit

eye also showed Chat t i le materlal was t o l ~ r a t c d very w ~ l l by t h e hcal thy

rabbit eyr. I n e cornea stayed clear, indicating t h a t t h ~ ~ n d n t h ~ l i u m ,

stroma and epitheljum n f t h ~ cnrnp: and t h p antericr cycrhamber d i d not

Figure 5 . Photograph of a rahbi t ~ y c with a clrrular Implant made of a

highly cross-linked Quadrol/lIDI-bnscd po:yure thane nclwork t w o weeks aC ter

implantation.

Figure 6 . Pho tugrdp i u f a I dbLlL cys5 W L t l r a kera toprosthesis made o f a

h i p ; h l y crnss-l i nk rd Quadrol / l lnI-based po lyure t l l ane network o n e year ar ter

i m p l a n t a t i u r ~ .

Figure i. Ha.bb i r e y e 3 : L I; . i n i sla.inL~!:;c. I. 'J/YI,+SS

keratopros t n e s i s

react on the implant and/or its material. Even one year after implantation

(see figure 61 the eye was still "clear" as it is called in the clinical

terms. The opacification seen around the keratoprosthesis and the

incision, made to extract the lens, are the result of

"Kammerwasserausbruch": a reaction of the stromal tissue to the leakage of

aqueous humour into the stroma of the sclera, unavoidably caused by the

surgical trauma. The overgrowth of the keratoprosthesis by the stroma and

epithelium of the sclera is not a reaction to the material as such. This

also happens after implanting a keratoprosthesis made of inert PMMA CQ

(clinical quality) or inert stainless steel/glass (see figure 7 )

(33,64,65). These results show that the transparent highly cross-linked

polyurethane network is suited to make inert, autoclavable

keratoprostheses.

The crystalline lens and cornea filter most of the solar ultraviolet

radiation, having wavelengths from approximately 285 to 400 nm and which

itself is damaging to the retina, to which the eye is exposed. Quanta with

wavelengths below 300 nm are almost completely absorbed by the cornea.

Ultraviolet radiation ranging from 300 to 400 nm is normally absorbed by

chromophores in the crystalline lens. After cataract extraction, the

posterior segment of the eye is therefore exposed to UV radiation not

normally encountered in the phakic state. The ideal intraocular lens

should be ultraviolet light absorbing, i.e. absorb UV light with

wavelengths below 400 nm. Nowadays nearly all intraocular lenses contain a

UV-absorbing chromophore, either in the form of a low molecular additive

or polymer bound (66-681. It appeared that Coumarin 102 (see figure 81, a

so-called laser dye, was soluble in the polyol component of the

polyurethane formulation. This compound could withstand thecuring process

without losing its UV light absorbing activity, unlike the conventional

hydroxyl containing UV absorbers, like hydroxybenzophenones or

hydroxybenzotriazoles (69) (these compounds contain hydroxyl groups that

can react with the diisocyanate used and consequently losing their

UV-absorbing ability). In figure 8 UV/VIS transmission spectra are shown

of a Quadrol/HDI-based polyurethane sample ( 1 mm thickness) with and

without the additive Coumarin 102, in low concentration. As can be seen

the polyurethane thermoset without the additive transmits light with

wavelengths above 275 nm (ca. 90 % transmission). A few promille Coumarin

102, which is a fluorescent compound with its absorption maximum at h=390

nm and its fluorescence maximum at h=468 nm, is enough to completely

absorb the UV light with wavelengths below 450 nrn.

Coumarin 102 FH'

200 300 400 500 600 700

Wavelength (nm)

Figure 8. UV/VIS transmission spectra of the additive-free

Quadrol/HDI-based polyurethane network ( A ) , and of the same material

containing 0,08% Coumarin 102 ( B ) , a UV-absorbing chromophore (sample

thickness 1 mm) .

Finally, there may be some concern about the long-term stability of the

highly cross-linked polyurethanes. As known from the literature (26,701,

polyurethanes in general do (bioldegrade, although nearly all

polyurethanes investigated are (uncross-linked) thermoplastic elastomers.

In vivo degradation is usually hydrolytic, although tissue enzymes may

also participate in the degradation process. Polyurethanes can be made

deliberately degradable (see ref. 32), but also hydrolytically stable.

Since the polyurethane networks described here are rather hydrophobic and

very densely cross-linked, pronounced degradation is not to be expected.

An indication for this is that the T of a Quadrol/HDI-based sample g

immersed in water for 20 months did not differ from the value of a sample

of the same material immersed in water for 2 months ( T =48 O C ) . Another 4

noticeable feature is that the rabbit eye with the implanted polyurethane

keratoprosthesis was quiet one year after the surgical procedure.

References

1. J. Harding, "Cataract: Biochemistry, Epidemiology and Pharmacology",

Chapman and Hall, 1991

2. A. Foster, G. J. Johnson, International Ophtalmology, 14, 135 (1990)

3. N.E. Christy, International Ophtalmology, 14, 141 (1990)

4. A. Foster, Ophtalmic Surgery, 18, 384 (11987)

5. S.K. West, Arch. Ophtalrnology, 109, 196 (1991)

6. M. C. Leske, L. T. Chylach Jr. , Suh-Yuh Wu, Arch. Ophtalmology, 109, 244

(1991 )

7. J. J. Alpar, P.U. Fechner, "Fechner's Intraocular lenses", Thieme Inc.

1986

8. R.H.S. Langston, Artificial Organs, 2, 55 (1978)

9. P.A.M. Leonard, J.G. Rommel, "Lens Implantation, 30 years of

progress", Ph. D. thesis, University of Leiden, The Netherlands, Dr.

W. Junk Publishers. 1982

10. W.J. Stark, Arch. Ophtalmology, 107, 1441 (1989)

11. K.W. Jacobi, M.R. Nowak, Fortschr. Ophtalmol. 86, 203 (1989)

12. D. J. Apple, N. Mamalis. K. Loftfield, J.M. Googe, L.C. Novak, D.

Kavka-Van Norman, S.E. Brady, R. J. Olson, Survey of Ophtalmology,

29, 1 (1984)

13. H.E. Kaufman, J. Katz, J. Valenti, J. W. Sheets, E.P. Goldberg,

Science, 198, 525 (1977)

14. J. Worst, personal communication

15. M. Yalon, E.P. Goldberg, D. Osborn, J. Stacholy, J.W. Sheets,

J. Cataract Refract. Surg., 14, 393 (1988)

16. D. P. Choyce, UK patent application GB 2.081.469 (1981)

17. A. J. Phillips, J. Stone, "Contact lenses, a textbook for practioner

and student". 3rd ed. Butterworths, 1989

18. L. Allarakhia, R.L. Knoll, R.L. Lindstrom, J. Cataract Refract. Surg.,

13, 607 (1987)

19. A.C. Neumann, B. Cobb, J. Cataract Refract. Surg., 15, 257 (1989)

20. F.R. Christ, D.A. Fencil, S. Van Gent, P.M. Knight, J. Cataract

Refract. Surg., 15, 176 (1989)

21. S.Y. Buchen, S.C. Richards, K. D. Solomon, D. J. Apple, P.M. Knight, R.

Christ, L. T. Pham, D.L. Nelson, H.M. Clayman, L.G. Karpinski, J.

Cataract Refract. Surg., 15, 545 (1989)

22. 0. Sveinsson, J. Seland, Acta Ophtalmologica, 68, 43 (1990)

23. K Smetana Jr., J. Sulc, 2. Krcova, S. Pitrova, J. Biomed. Mater. Res.,

21, 1247 (1987)

24. B. D. Ratner, A.S. Hoffman, in: "Hydrogels for medical and related

applications", Ed. J.D. Andrade, Am. Chem. Soc. Symp. Ser. 31, ACS

Washington DC, 1976, p. 1

25. S. N. Fyodorov, 2. I. Moroz, V. K. Zuev, "Keratoprostheses" , Churchill

Livingstone, 1987

26. M.D. Lelah, S. L. Cooper, "Polyurethanes in medicine", CRC Press, Boca

Raton, Florida, USA, 1986

27. Organic Syntheses Coll. Vol. IV, page 753 (Hershel L. Herzog), 1963

28. U.S. patent 3.994.972 (1976) (inventors: M. J. Seely, T.C. Peterson)

29. A. Litherland, F.G. Mann, J. Chem. Soc., 1588 (1938)

30. F.G. Mann, A. Litherland, Nature, 141, 789 (1938)

31. G.M. Gibson, J. Harley-Mason, A. Litherland, F. G. Mann, J. Chem. Soc.,

163 (1942)

P. Bruin, G. J. Veenstra, A. J. Nijenhuis, A. J. Pennings, Makromol.

Chem. Rapid Commun., 9, 589 (1988)

P. van Andel, I. Singh, J. Worst, G. Bartman, P. Nieuwenhuis, Fourth

World Biomaterials Congress April 24-28, 1992, Berlin

J. E. Mark, B. Erman, "Rubberlike elasticity, a molecular primer",

Wiley-Interscience, 1988

J.E. Mark, Adv. Polym. Sci., 44, 1 (1982)

J.E. Mark, P. -H. Sung, Eur. Polym. J., 16, 1223 (1980)

P.-H. Sung, J.E. Mark, J. Polym. Sci. Polym. Phys. Ed., 19, 507 (1981)

M. Ilavsky, K. DuSek, Polymer, 24, 981 (1981)

M. Ilavsky. K. DuSek. Macromolecules. 19. 2139 (1980)

K. Dusek, Rubber Chem. Technol. , 55, 1 (1982)

X. Hu, R. J. Young, Plastics and Rubber Processing and Applications,

12, 113-126 (1989)

S.S. Labana, S. Newman, A. J. Chompff, in: "Polymer networks, structure

and mechanical properties", Eds. A. J. Chompf f and S. Newman, Plenum

Press, NY-London, 1971, page 453

H.M. J. Boots, in: "Biological and synthetic networks", Ed. 0. Kramer,

Elsevier Applied Science, London & New York, 1988, page 267

H.M. J. Boots, Physica, 147A, 90 (1987)

A.L . Andrady, M. A. Llorente, M. A. Sharaf, R.R. Rahalkar, J.E. Mark,

J.L. Sullivan, C.U. Yu, J.R. Falender, J. Appl. Polym. Sci., 26, 1829

(1981 )

R. F. T. Stepto, in: "Biological and synthetic networks", Ed. 0. Kramer,

Elsevier Applied Science, London & New York, 1988, page 153

J.H. Saunders, K.C. Frisch, "High Polymers XVI, Polyurethanes

Chemistry and Technology, part 1 Chemistry", Interscience Publishers,

1962

Kunststoff-Handbuch VII: Polyurethane, Carl Hanser Verlag Miinchen,

1966

G. Oertel, Polyurethane Handbook, Hanser Publishers, 1985

M.T. Aronhime, J.K. Gillham, Adv. Polym. Sci., 78, 84 (1986)

G. Wisanrakkit, J.K. Gillham, J. Appl. Polym. Sci., 41, 2885 (1990)

J. Mijovic, J. Wijava, Macromolecules, 23, 3671 (1990)

Encyclopedia of Polymer Science & Technology, Editorial board: H.F.

Mark, N.G. Gaylord, N.M. Bikales, Interscience Publishers, 1966

K. Dubek, Polymer Bulletin, 17, 481 (1987)

J.A. Schroeder, P.A. Madsen, R.T. Foister, Polymer, 28, 929 (1987)

R. J. Morgan, F.-M. Kong, C.M. Walkup, Polymer, 25, 375 (1984)

J.D. Lemay, B.J. Swetlin, F.N. Kelley, in: "Characterization of highly

cross-linked polymers", Eds. R. A. Dickie and S. S. Labana, ACS

Washington D. C., page 165 (1984)

S. Yamini, R. J. Young, J. Mater. Sci., 15, 1814 (1980)

M. Bos, A. J. Pennings, to be published

D.F. Williams, in: "Materials Science and Technology, volume 14:

Medical and Dental Materials", volume ed. D.F. Williams, page 415

[ 1991 1

R. Larsson. G. Selen, H. Bjorklund, P. Fagerholm, Biomaterials, 10,

511 (1989)

European patent 0.286.433A2 (inventor E. P. Goldberg)

A.S. Chawla ( p . 231) and H. Chmiel, H. Bauser (p. 2451, in: "Polymeric

Biomaterials", eds. E. Piskin, A.S. Hoffman, Martinus Nijhoff

Publishers (1986)

P.L. Cuperus, W. L. Jongebloed, P. van Andel, J.G.F. Worst, Documenta

Ophtalmologica, 71, 29 (1989)

P. van Andel, P.L. Cuperus, M. Kolenbrander, Daljit Singh, J.G. F.

Worst, Transactions 3rd World Biomaterials Congress, Kyoto, Japan,

1988, Vol. XI, page 542

H.M. Clayman, Am. Intra-ocular Implant Soc. J., 10, 429 (1984)

H.M. Clayman, J. Cataract Refract. Surg., 12, 529 (1986)

J.S. Werner, L. Spillmann, Graefe's Arch. Clin. Exp. Ophtalmol., 227,

248 (1989)

B.Ranby, J.F. Rabek. "Photodegradation, photo-oxidation and

photostabilization of polymers", John Wiley and Sons (1975)

S. Gogolewski, Colloid Polym. Sci., 267, 757 (1989)

Summary

This thesis is concerned with the synthesis, physical properties and

potential biomedical applications of some polyurethanes, especially

cross-linked ones. Polyurethanes are a class of polymers having only in

common the presence of urethane bonds somewhere in their chains. The name

polyurethane given to a polymeric material does not tell anything about

its chemical and physical characteristics. Polyurethanes may be lightly or

highly cross-linked or uncross-linked and be highly crystalline,

elastomeric or amorphous and glassy. In the biomedical field polyurethane

usually stands for thermoplastic polyurethane elastomer. Thermoplastic

polyurethanes, also named segmented polyurethanes, are linear

blockcopolymers composed of chainextended diisocyanate hard segments

dispersed in a soft segment polyol matrix. Due to their good mechanical

properties (high tensile strength, good tear strength, high toughness,

good flex life), reasonable bloodcompatibility and biocompatibility,

elastomeric polyurethanes have been used in many medical applications,

like total artificial heart, heart valves, vascular prostheses, wound

dressings etc. By mixing segmented polyurethanes with (5-20 wt.%) high

molecular weight poly(L-lactide) (PLLA), Gogolewski, Leenslag and Pennings

in Groningen developed elastomeric, biodegradable mixtures with remarkable

in vivo performance. Quenched physical polyurethane/PLLA mixtures, in

porous form, were succesfully applied as a small-caliber vascular

prosthesis, artificial skin, meniscus lesion repair material, nerve guide.

Since these polyurethanes are not chemically cross-linked (formation of

permanent cross-links), they show stress softening (stress hysteresis)

when subjected to cyclic deformation. This problem can be overcome by

chemically cross-linking the linear polyurethane chains, for instance,

with peroxides (see chapter 5 ) . Another drawback of commercial biomedical

polyurethanes concerns their chemical composition. Nearly all these

polyurethanes (Biomer, Estane, Pellethane, etc.) are composed of an

aromatic diisocyanate MDI (4,4'-methylene diphenyl diisocyanate).

Degradation (through hydrolysis) of the polymer may result in the

formation of the toxic, carcinogenic, mutagenic MDA.

4,4'-methylenedianiline, the degradation product of the incorporated

aromatic diisocyanate. Although it has not been shown unambiguously that

MDI-based polyurethanes induce the formation of cancer, it would be more

elegant and safer to seek for a replacement for this component in the

polyurethane formulation. The use of cycloaliphatic diisocyanates, for

instance, hydrogenated MDI in Tecoflex, also leads to segmented

polyurethanes with good ultimate properties. Aliphatic diisocyanates,

which are not particularly suited for the synthesis of thermoplastic

polyurethane elastomers, may be used for the formation of chemically

cross-linked polyurethanes. Especially aliphatic diisocyanates, producing

non-toxic diamines (e.g., lysine, 1,4-diaminobutane) after eventual

degradation, seem the ultimate choice for the synthesis of biomedical

polyurethanes.

In chapter 2 such lysine diisocyanate-based elastomeric polyurethane

networks are described. These polyurethane networks, designed to release

only non-toxic degradation products, were prepared by cross-linking

hexafunctional starshaped prepolymers with ethyl 2,6-diisocyanatohexanoate

(i.e., lysine diisocyanate). The hydroxy terminated prepolymers were

synthesized by the ring-opening copolymerization of L-lactide or glycolide

and E-caprolactone initiated by myo-inositol, a vitamin. The

polyesterurethane networks, having T 's in the range 0-10 OC and gel

contents of 90-95 %, showed rubber-like behaviour. It is noteworthy that

the chloroform-extracted networks exhibited much better tensile properties

(tensile strength ca. 30-40 MPa) than the unextracted networks (tensile

strength ca. 10 MPa). Only the extracted networks exhibited pronounced

strain-induced crystallization. The presence of plasticizer (sol fraction)

suppressed the strain-induced crystallization.

In chapter 3 the polyurethane networks described in the previous chapter

are evaluated as potential materials for the construction of a macroporous

bottom-layer (dermal analogue) in a multi-layer artificial skin. An

amorphous, elastomeric, porous lysine diisocyanate-based poly(glyco1ide-

co-E-capro1actone)urethane network degraded fast in vitro. In vivo the

same material was degraded even faster. Subcutaneous implantation in

guinea pigs showed that the porous polyurethane networks degraded almost

completely between 4 and 8 weeks after implantation, allowed rapid cell

ingrowth and evoked no adverse tissue reaction. The lysine

diisocyanate-based polyurethane networks can be considered biocompatible.

Chapter 4 is concerned with the preparation and characteristics of a

porous polyurethane wound covering. A very thin, porous membrane (15-20

pm) was prepared by means of a phase inversion process. This elastic film,

made of a cycloaliphatic polyetherurethane (Tecoflex), contained

micropores up to approximately 5 pm. The porous wound covering was

impermeable to bacteria. The polyurethane membrane appeared to be very

permeable to water vapour, whereas water in liquid form or wound exudate

could not leak through the membrane. In guinea pigs epidermal wound

healing of partial-thickness wounds under polyurethane wound coverings was

accelerated as compared with uncovered controls and an occlusive wound

covering (Op-Site). The high water vapour permeability of the polyurethane

wound covering induced concentration of the wound exudate into a jellylike

clot layer, which apparently accelerated reepithelialization. The main

conclusion from a clinical study on 20 split-skin donor sites was that the

use of the polyurethane covering reduces pain (as compared with the

conventional treatment of tulle gras dressing), besides prevention of

fluid retention and enhanced reepithelialization.

Chapter 5 describes a two-ply biodegradable artificial blood vessel made

of polyurethane and poly(L-lactide). The microporous innerlayer of the

small-caliber vascular prosthesis was constructed from a cycloaliphatic

segmented polyurethane (Tecoflex) cross-linked with dicumylperoxide in the

presence of linoleic acid. The reason for introducing chemical (permanent)

cross-links into the polyurethanes is to eliminate the serious limitation

of stress softening which occurs especially in cyclic loading. Cyclic

creep-failure, resulting from the arterial pulsation of the blood, may

lead to the formation of aneurysms (catastrophic dilation of blood

vessels). Furthermore, carboxyl groups of the linoleic acid on the luminal

side of the prosthesis contribute positively to the antithrombogenicity of

the artificial blood vessel. It appeared that adding linoleic acid during

the peroxide vulcanization led to a maintenance of the tensile strength of

the prostheses. The outer ply was constructed by precipitating a (95/5)

physical mixture of a polyesterurethane and poly(L-lactide) from solution

in the presence of sugar crystals in the range 30-90 pm which were removed

by exposing the prosthesis to water. The two-ply vascular prostheses were

tested in vivo by replacing 1 cm of the abdominal aorta of rats. All the

prostheses remained patent at least up to one year and did not exhibit any

aneurysmal formation, which has to be ascribed to the improved

creep-resistance of the prosthesis as a result of the cross-linking. The

inner layer of the prosthesis was covered with endothelial cells and

several layers of smooth muscle cells. Essential components of a

neo-artery were regenerated.

Chapter 6 deals with the synthesis, properties and potential ophtalmic

applications (intraocular lenses, keratoprostheses) of highly cross-linked

polyurethane networks. Such polyurethanes were prepared by the bulk

stepreaction of various low molecular weight polyols and (cyc1o)aliphatic

diisocyanates. All these polyurethane networks were optically transparent,

colourless, amorphous glassy thermosets. The properties of one particular

glassy polyurethane, obtained from the bulk reaction of a tetrafunctional

aminoalcohol tetrakis(2-hydroxypropy1)ethylenediamine (Quadroll and

hexamethylenediisocyanate (HDI) in stoichiometric proportions, were

investigated in more detail. This glassy polyurethane, with an ultimate

glass transition temperature of 85 OC, and a very low degree of swelling

in chloroform (1,271, exhibited good ultimate mechanical properties

(tensile strength 80-85 MPa, elongation at break ca. 15 X , modulus ca. 1,5

GPa). Infra-red spectra of these polyurethane networks revealed the

absence of an isocyanate absorption, indicating that all isocyanates.

apparently, had reacted during the cross-linking reaction.

These transparent cross-linked polyurethanes can be sterilized simply by

autoclaving, in contrast to polymethylmethacrylate (PMMA) which has been

used successfully as an intraocular lens material the last 15 years. The

possibility of an autoclavable lens is especially interesting with respect

to eye surgery in the developing world where the majority of the blind

people live. These highly cross-linked Quadrol/HDI-based networks, after

being autoclaved, were implanted in rabbit eyes, either in the form of

small circular disks or in the form of a keratoprosthesis (artificial

corneal. It was shown that the material was well tolerated by the rabbit

eyes. A serious opacification of the cornea, indicating an adverse

reaction to the implant, was never seen. One year after implantation of a

polyurethane keratoprosthesis the eye was still "quiet". These results

show that the transparent highly cross-linked polyurethane network seems

suited for use in ophtalmic applications.

Samenvatting

Dit proefschrift beschrijft een onderzoek naar de synthese, eigenschappen

en mogelijke biomedische toepassingen van een aantal (hoofdzakelijk

vernette) polyurethanen. De term "polyurethaan" heeft betrekking op een

klasse polymeren die slechts de aanwezigheid van urethaan bindingen ergens

in de polymere ketens gemeen hebben. De naam polyurethaan gegeven aan een

polymeer materiaal zegt weinig over de chemische struktuur en de fysische

eigenschappen ervan. Polyurethanen kunnen licht of sterk vernet of

onvernet zijn, hoog kristallijn, rubberachtig of amorf en glasachtig zijn.

In de biomedische wereld heeft de naam polyurethaan doorgaans betrekking

op thermoplastische polyurethaan elastomeren. Deze zogenaamde

gesegmenteerde polyurethanen zijn lineaire blokcopolymeren bestaande uit

harde segmenten ("chainextended" diisocyanaten) die gedispergeerd zijn in

een matrix gevormd door zachte polyol segmenten. Door hun goede

mechanische eigenschappen (hoge treksterkte, taaiheid, goede

scheursterkte) en acceptabele bloed- en biocompatibiliteit, hebben

multifase elastomere polyurethanen succesvolle medische toepassing

gevonden, bijvoorbeeld als kunsthart, hartklep, kunstaders, wondbedekking.

In Groningen werden dergelijke polyurethanen door Gogolewski, Leenslag en

Pennings gemengd met 5-20 X hoog moleculair poly(L-melkzuur) (PLLA). De zo

verkregen elastomere, biodegradeerbare mengsels voldeden opmerkelijk goed

in een aantal degradeerbare biomedische toepassingen.

Aangezien deze elastomere polyurethanen niet chemisch vernet zijn, maar te

beschouwen zijn als fysische netwerken, vertonen ze zogenaamde "stress

softening" bij cyclische belasting. Dit probleem kan verholpen worden door

de lineaire ketens alsnog chemisch te verknopen, bijvoorbeeld met

peroxides (zie hoofdstuk 5). Een ander nadeel van de commercieel

verkrijgbare biomedische polyurethanen heeft betrekking op de chemische

samenstelling. Bijna a1 deze polyurethanen (Biomer, Estane, Pellethane,

enz.) zijn opgebouwd uit een aromatisch diisocyanaat 4,4'-methyleendifenyl

diisocyanaat (MDI). Degradatie, door hydrolyse, van het polymeer kan

uiteindelijk leiden tot de vorming van het toxische, carcinogene, mutagene

4,4' -methyleendianiline (MDA) , het degradatieprodukt van het

gePncorporeerde aromatische diisocyanaat. Alhoewel het gezegd dient te

worden dat de implantatie van een polyurethaan op basis van MDI niet

ondubbelzinning aantoonbaar geleid heeft tot de vorming van een vorm van

kanker, zou het toch eleganter en veiliger zijn om te zoeken naar een

vervanger voor MDI in de polyurethaan samenstelling. In plaats van

aromatische diisocyanaten zijn cycloalifatische diisocyanaten gebruikt

(bijvoorbeeld in Tecoflex), hetgeen ook gesegrnenteerde polyurethanen met

goede eigenschappen oplevert. Vooral alifatische diisocyanaten, die er

voor zorgen dat niet toxische diamines (lysine, l,4-diaminobutaan) als

degradatieprodukten kunnen worden gevormd, lijken een goede keuze voor de

synthese van biomedische (vernette) polyurethanen.

In hoofdstuk 2 worden dergelijke rubberachtige polyurethaan netwerken op

basis van ethyl 2,6-diisocyanaathexanoaat ("lysine diisocyanaat")

beschreven. Deze polyurethaan netwerken, die zodanig ontworpen zijn dat

bij degradatie uitsluitend verbindingen worden gevormd die niet toxisch

zijn, worden verkregen door hexafunctionele stervormige prepolyrneren te

vernetten met het diisocyanaat op basis van lysine, een aminozuur. De

prepolyrneren op hun beurt worden gesynthetiseerd via de

ringopeningscopolymerisatie van hetzij L-lactide of glycolide, en

c-caprolacton, geYnitieerd door myo-inositol, een vitamine. De zo gevormde

biodegradeerbare polyesterurethaan netwerken zijn rubberachtig (T 's in de 'J

orde van 0-10 OC). Opmerkelijk is dat de rubber netwerken na extractie met

chloroform (de netwerken hebben gel percentages in de orde van ca. 90-95

% I zeer goede mechanische eigenschappen bezitten in vergelijking met de

niet geextraheerde netwerken, hetgeen wordt toegeschreven aan

"strain-induced crystallization". Na extractie van de sol fractie, die als

weekmaker fungeert en daardoor strain-induced crystallization verhindert,

stijgt de treksterkte van deze rubber netwerken van ca. 10 MPa naar 30-40

MPa. In hoofdstuk 3 wordt de biomedische toepasbaarheid van de

polyurethaan netwerken zoals beschreven in het voorgaande hoofdstuk

geevalueerd, en dan met name de toepasbaarheid van deze materialen als

biodegradeerbare, poreuze onderlaag in een meerlagen kunsthuid. Een amorf,

rubberachtig, poreus, poly(glyco1ide-co-E-capro1acton)urethaan netwerk op

basis van lysine diisocyanaat bleek in vitro snel te degraderen. Hetzelfde

rnateriaal bleek in vivo (subcutane implantaties bi j cavia' s) nog sneller

afgebroken te worden. De poreuze implantaten waren na 4-8 weken bijna

volledig gedegradeerd. Ondertussen was in de poreuze materialen duidelijke

celingroei waarneembaar. Het polyurethaan op basis van lysine diisocyanaat

bleek geen ongunstige weefsel/ontstekingsreactie te veroorzaken en zou als

biocompatibel bestempeld mogen worden.

Hoofdstuk 4 beschrijft de vervaardiging en eigenschappen van een poreuze

polyurethaan wondbedekking. Door middel van een fase inversie proces werd

een zeer dunne, poreuze membraan (15-20 pm) verkregen. Deze elastische

film, gemaakt van een cycloalifatisch polyetherurethaan (Tecoflex), bevat

een groot aantal microgaatjes ter grootte van ca. 5 pm. De poreuze

polyurethaan wondbedekking bleek ondoorlaatbaar voor bacterien (en voor

water in vloeibare vorm, b.v. wondvocht), maar zeer permeabel voor

waterdamp. Bij cavia's bleek dat de epidermale wondgenezing van 0.3 mm

diepe schaafwonden bedekt met de polyurethaan wondbedekking sneller

verliep dan die van onbedekte wonden en wonden bedekt met een occlusieve

wondbedekking (Op-Site). Deze stimulerende werking op het

wondgenezingsproces komt waarschijnlijk door de hoge

waterdampdoorlaatbaarheid van de poreuze polyurethaan film die het

wondvocht concentreert tot een soort gelachtige substantie. De

belangrijkste conclusie van het klinisch gebruik op 20 split-skin

donorplaatsen was dat de polyurethaan wondbedekking de pijn duidelijk

vermindert, in vergelijking met een conventionele behandeling met

ingezwachteld paraffine gaas (tulle gras). De polyurethaan wondbedekking

zorgde ervoor dat wondvochtretentie werd voorkomen, maar de epithelisatie

niet meer werd versneld dan bij gebruik van paraffine gaas.

In hoofdstuk 5 wordt een tweelagen biodegradeerbare kunstader, gemaakt van

polyurethaan en poly(L-melkzuur) beschreven. De microporeuze binnenlaag

van de kleine-diameter vaatprothese werd vervaardigd van een

cycloalifatisch gesegmenteerd polyurethaan (Tecoflex), dat vernet werd met

behulp van dicumylperoxide in de aanwezigheid van linolzuur. Op deze wijze

werd de kruipweerstand van het polyurethaan sterk verbeterd. Bovendien

zorgden de carbonzuur groepen van het linolzuur voor een niet trombogeen

binnenoppervlak van de vaatprothese. Ten derde zorgt de toevoeging van

linolzuur bij het vernetten met peroxides dat de sterkte als gevolg van de

vernetting niet afneemt. De macroporeuze buitenlaag van de prothese werd

gemaakt van een snel uit oplossing neergeslagen polyurethaardPLLA (95/5)

fysisch mengsel. De tweelagen kunstaders bleken in vivo (vervanging van 1

cm van de buikslagader in ratten) goed te functioneren. Alle protheses

waren na 1 jaar nog functioneel en lieten geen aneurysma' s (catastrofale

verwijding van de ader) zien, hetgeen toegeschreven dient te worden aan de

sterk verbeterde weerstand tegen kruip van de protheses als gevolg van de

chemische vernetting. De essentiele componenten van een natuurlijke

vaatwand bleken te zijn geregenereerd: de binnenlaag van de protheses was

volledig bedekt met endotheelcellen en enkele lagen gladde spiercellen.

Hoofdstuk 6 beschrijft de synthese en eigenschappen van zeer dicht

vernette polyurethanen, alsmede de potentiele toepassing van deze

materialen in de oogheelkunde, bijvoorbeeld als intraoculaire lens (ter

vervanging van de door staar troebel geworden natuurlijke ooglens) of als

keratoprothese (kunsthoornvlies). Deze sterk vernette polyurethanen werden

gesynthetiseerd door verschillende laag moleculaire polyolen te laten

reageren met (cyc1o)alifatische diisocyanaten in de afwezigheid van een

oplosmiddel. De zo verkregen polyurethaan netwerken ziJn optisch

transparante, kleurloze en arnorfe, glasachtige materialen. De

eigenschappen van een bepaald polyurethaan thermoharder, representatief

voor de hier gesynthetiseerde netwerken, verkregen door de stapreactie in

de bulk van een tetrafunctioneel aminoalcohol tetrakis(2-hydroxypropy1)-

ethyleendiamine (Quadrol) en hexamethyleendiisocyanaat (HDI) in

stoichiornetrische hoeveelheden, werden nader onderzocht. Bij volledige

uitharding heeft dit glasachtig polyurethaan netwerk een maxirnale T van g

85 OC, een zeer lage zwelgraad in chloroform (1,271 en goede mechanische

eigenschappen (treksterkte 80-85 MPa, rek biJ breuk ca. 15 %, Young's

modulus 1,5 GPa). Uit het infrarood spectrum bleek dat alle (toxische)

isocyanaat groepen, bij volledige uitharding, weggereageerd waren.

In tegenstelling tot polymethylmethacrylaat (PMMA), dat de laatste 15 jaar

met veel succes als intraoculaire lens materiaal gebruikt is, kunnen deze

vernette polyurethanen gesteriliseerd worden door autoclavering. De

mogelijkheid om een kunstlens te kunnen autoclaveren is met name

interessant voor de oogchirurgie in ontwikkelingslanden, waar de

meerderheid van alle blinde mensen leven. Implantaties van het

polyurethaan netwerk op basis van Quadrol en HDI in konijneogen lieten

zien dat het materiaal goed getolereerd werd. Een jaar na de implantatie

van een polyurethaan keratoprothese was het oog nog steeds rustig. De

resultaten van de polyurethaan keratoprothese waren vergelijkbaar met

implantaties van PMMA of metaal/glas keratoprotheses. A1 deze resultaten

laten zien dat het dicht vernette polyurethaan netwerk geschikt lijkt voor

toepassing in de oogheelkunde.