ARTICLE IN PRESS

Journal of Biomechanics 42 (2009) 722–730

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/jbiomech

Journal of Biomechanics

0021-92

doi:10.1

� Corr

Enginee

London

Kingdom

E-m

www.JBiomech.com

Manufacture of small calibre quadruple lamina vascular bypass graftsusing a novel automated extrusion-phase-inversion method andnanocomposite polymer

Sandip Sarkar a,d, Gaetano Burriesci b, Adam Wojcik b, Nicholas Aresti c,George Hamilton d, Alexander M. Seifalian a,d,�

a Centre for Nanotechnology, Biomaterials & Tissue Engineering, UCL Division of Surgery & Interventional Science, University College London, London, UKb Department of Mechanical Engineering, University College London, London, UKc University College Medical School, University College London, London, UKd Vascular Unit, Royal Free Hospital Hampstead NHS Trust, London, UK

a r t i c l e i n f o

Article history:

Accepted 9 January 2009Long-term patency of expanded polytetrafluoroethylene (ePTFE) small calibre cardiovascular bypass

prostheses (o6 mm) is poor because of thrombosis and intimal hyperplasia due to low compliance,

Keywords:

Bypass graft

Nanocomposite

Compliance

Viscoelastic

Polymer

Coronary artery

90/$ - see front matter & 2009 Elsevier Ltd. A

016/j.jbiomech.2009.01.003

esponding author at: Centre for Nanotechn

ring, UCL Division of Surgery & Interventiona

, Royal Free Hampstead NHS Trust Hospita

Tel.: +44 20 7830 2901; fax: +44 20 7472 644

ail address: A.Seifalian@UCL.AC.UK (A.M. Seifa

a b s t r a c t

stimulating the search for elastic alternatives. Wall porosity allows effective post-implantation graft

healing, encouraging endothelialisation and a measured fibrovascular response. We have developed a

novel poly (carbonate) urethane-based nanocomposite polymer incorporating polyhedral oligomeric

silsesquioxane (POSS) nanocages (UCL-NANOTM) which shows anti-thrombogenicity and biostability.

We report an extrusion-phase-inversion technique for manufacturing uniform-walled porous

conduits using UCL-NANOTM. Image analysis-aided wall measurement showed that two uniform wall-

thicknesses could be specified. Different coagulant conditions revealed the importance of low-

temperature phase-inversion for graft integrity. Although minor reduction of pore-size variation

resulted from the addition of ethanol or N,N-dimethylacetamide, high concentrations of ethanol as

coagulant did not provide uniform porosity throughout the wall. Tensile testing showed the grafts to be

elastic with strength being directly proportional to weight. The ultimate strengths achieved were above

those expected from haemodynamic conditions, with anisotropy due to the manufacturing process.

Elemental analysis by energy-dispersive X-ray analysis did not show a regional variation of POSS on the

lumen or outer surface. In conclusion, the automated vertical extrusion–phase-inversion device can

reproducibly fabricate uniform-walled small calibre conduits from UCL-NANOTM. These elastic

microporous grafts demonstrate favourable mechanical integrity for haemodynamic exposure and are

currently undergoing in-vivo evaluation of durability and healing properties.

& 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Since, expanded polytetrafluoroethylene (ePTFE’s) introductionin 1973 (Matsumoto et al., 1973), no better performing syntheticprosthesis o6 mm in diameter has emerged. However, itsmedium-term patency is compromised by intimal hyperplasia,with two-thirds of synthetic femoral-distal bypass grafts occlud-ing without surgical intervention (Sarkar et al., 2006b). The directlink between PTFE inelasticity and graft occlusion was made byKidson and Abbott (1978). Despite considerable research, an

ll rights reserved.

ology, Biomaterials & Tissue

l Science, University College

l, London, NW3 2PF, United

4.

lian).

elastic material with the blood compatibility and biostability ofPTFE is yet to be achieved.

Many methods have been utilised to manufacture porousgrafts for vascular bypass including salt leaching (Chen et al.,1998), electrospinning (Inoguchi et al., 2006), electrosprayingwith phase-inversion (Khorasani and Shorgashti, 2006a), excimerlaser ablation (Doi et al., 1996), dip-coating and extrusion(Zdrhala, 1996). Porosity is desired to provide graft elasticity(Moroni et al., 2006); also enhancing adherence of endothelialcells to the graft lumen; as well as encouraging transmuralsupport for successful cellular integration (Tsuchida et al., 1997).Neovascularisation through porous wall may have a role inlimiting neointimal hyperplasia and atherosclerosis, by prevent-ing hypoxia of cells attached to the lumen (George et al., 2001).However, very high porosity adversely affects handling properties,requiring graft pre-clotting before implantation; increasing

ARTICLE IN PRESS

S. Sarkar et al. / Journal of Biomechanics 42 (2009) 722–730 723

thrombogenicity and causing massive fibrovascular infiltration(Rosengren and Bjursten, 2003) which soon leads to a loss ofcompliance.

Polyurethanes (PU) have attracted attention as possibleprostheses due to the wide range of properties achievable withinthis class of material (Tiwari et al., 2002). Their fundamentalproblem is long-term biodegradation. The earlier generationpolyester soft segments showed rapid hydrolysis, whereas thepolyether-based PU were susceptible to oxidative attack (Dempseyet al., 1998). One method of eliminating the ether groups was touse polycarbonate soft segment. This proved to be far moreresistant to biodegradation than either of its predecessors(Salacinski et al., 2002). However, despite the hard segment’srelative shielding of the soft segment from scission, there isgrowing evidence that there is slow degradation of polycarbonate.It is not completely clear whether this is via hydrolysis, oxidationor a combination of the two.

Our nanocomposite polymer based on incorporating polyhe-dral oligomeric silsesquioxane (POSS) into poly (carbonate-urea)urethane (PCU) has been awarded an international patent(UCL-NanoTM). (Seifalian et al., 2005) It was initially developedto improve the poor in-vivo stability of polyurethanes describedabove. It was based on the observation of high biostability ofsiloxane-based polyurethanes, and hoped to use the nanoscaleproperties of POSS to replicate this without the poor mechanicalproperties of siloxane. Previous work has demonstrated it’s long-term biostability in-vitro when exposed to degradative solutions(Kannan et al., 2006b).

Another exciting prospect is the impressive blood compat-ibility profile of the nanocomposite (Kannan et al., 2006a). ThePOSS confers lower surface tension properties than PTFE, withweak fibrin binding to the surface and poor platelet adherence.The combination of biostability and anti-thrombogenicity makesthe nanocomposite a potential candidate for cardiovascular graftmaterial (Kannan et al., 2006d).

The cast sheet material made from NC is unsuitable as a graftdue to its lack of porosity and its low elasticity in the physiologicalpressure range. However, the cast sheet demonstrates impressivemechanical strength and is biocompatible, with minimal fibrouscapsule formation around it after long-term implantation in-vivo.

Phase-inversion was used to form porous small calibreconduits. This relates to the exchange of solvent and non-solventbetween the UCL-NANOTM polymer in solution and coagulant,resulting in two separate phases – a polymer-rich and a polymer-poor phase (Khorasani and Shorgashti, 2006b). In order to achieveuniform wall thickness, a novel automated alignment system wasdeveloped to coat mandrels with the nanocomposite before

Fig. 1. (a) Extruder device showing vertical mechanical arm with mandrel attached, (

channel laterally and (c) under surface of polymer chamber showing adaptors enabling

phase-inversion. Quality control was performed on the rawmaterials by batch rheological testing of the nanocomposite insolution beforehand. The aim was to develop 5 mm internaldiameter elastic microporous grafts which were robust enough forimplantation into the haemodynamic circulation for subsequentin-vivo study.

2. Materials and methods

2.1. Polymer synthesis

The polymer development has been described in detail previously (Seifalian

et al., 2005; Kannan et al., 2006c, 2005; Seifalian et al., 2005). In brief, polycarbonate

polyol and trans-cyclohexanediolisobutyl-silsesquioxane were heated to dissolve

the nanocage. Methylene di-isocyanate was added and reacted at 70 1C for

90 min to form a pre-polymer. N, N-dimethylacetamide (DMAC) was added.

Chain extension of the pre-polymer was carried out by adding ethylenediamine

and diethylamine. Then, 1-butanol in DMAC was added to the mixture to form

a 2% polyhedral oligomeric silsesqioxane-poly (carbonate-urea) urethane

(UCL-NANOTM) solution. All reagents were purchased from Aldrich Limited,

Gillingham, UK.

2.2. Extrusion–coagulation

A mechanical arm was positioned such that it held a 5 mm diameter stainless

steel cylindrical mandrel within the exit aperture of a stainless steel polymer

chamber (Fig. 1). This latter structure comprised a circular 5 mm entry aperture

superiorly; a luer-lock syringe compatible polymer introduction channel laterally;

a 5.5 mm or alternatively 6 mm circular exit aperture inferiorly; and a PTFE sliding

cover for the exit aperture. The mandrel cross-section was aligned perfectly

concentric with respect to the exit aperture.

With the mandrel’s base level with the exit aperture, the PTFE sliding cover

was closed before 3.5 ml of nanocomposite that was injected slowly into the

polymer chamber. For this step it was imperative that the syringe contained no air

bubbles.

A vertical column of coagulant solution was placed directly below the sliding

cover leaving a 5 mm gap between the two. The cover was opened and the mandrel

driven vertically into the coagulant at 10 mm/s immediately afterwards. The

alignment was checked during this extrusion by visually comparing the layers of

nanocomposite coating on either side of the mandrel within the coagulant solution

in the moments before coagulation occurred. This was undertaken in two planes

perpendicular to each other about the longitudinal axis of the mandrel, and was

facilitated by viewing against a natural light background.

The column was undisturbed for 20 min then placed in the fridge at 5 1C for a

minimum of 48 h. Thereafter, the mandrel was removed from the coagulant and

dried, before transferring to an absolute ethanol bath for degassing over 10 min,

thereby facilitating the removal of the formed tube from the mandrel.

The process was repeated for different concentrations of DMAC and ethanol in

de-ionised water and coagulant temperature range from 0 to 50 1C, using scanning

electron microscopy to visualise the details of the wall cross-section and pores.

Conduits were extruded with 5.5 and 6 mm exit apertures. Image analysis

software (Carl Zeiss KS 400 version 3.0) was used to measure the uniformity of

b) polymer chamber with mandrel entering superiorly and polymer introduction

control of exit aperture size

ARTICLE IN PRESS

S. Sarkar et al. / Journal of Biomechanics 42 (2009) 722–730724

wall thickness at 72 points which was distributed equally around the circumfer-

ence of 5 thin- and 5 thick-walled grafts.

The nanocomposite was also cast by evaporation of DMAC in an oven at 60 1C

into a 0.5-mm-thick sheet for comparative tensile testing of the non-porous

material.

2.3. Energy-dispersive X-ray analysis (EDXA)

Serial dehydration by using graded ethyl alcohols and finally hexamethyldi-

silazane (TAAB Ltd, UK) was performed on grafts. The dried grafts were sputter-

coated with gold/palladium (Polaron E5000, Quorum Technology, UK). Random

areas on the lumen and outer aspects were examined via energy-dispersive X-ray

analysis.

2.4. Tensile testing

A Hounsfield H5K-S Universal Testing Machine with an environmental

chamber (water at 37 1C) was used to characterise stress–strain profiles of

dogbone and ring specimens. For comparison, a 4 mm PTFE graft and the

nanocomposite cast sheet were also tested. ‘Dogbone’ specimens were prepared

using a sharp cutter and mechanical press in accordance with ISO 37 with

specimen dimensions conforming to type 2 (ISO, 2005). Ring specimens were

4 mm in width corresponding with type A. The tensile strengths for 6 thin- and 6

thick-walled grafts were found using a tension rate of 60 mm/min after a pre-test

rate of 5 mm/min until a threshold force of 0.05 N was achieved.

2.5. Burst pressure testing

Our previously described validated technique was used (Sarkar et al., 2006a). A

high-pressure syringe pump (Harvard Apparatus PHD 2000 Programmable)

containing freshly de-ionised water (pH 7.0) was connected via a pre-calibrated

transducer (Honeywell Component No. 22PCCFB6G) to 8 cm lengths of graft with

non-porous latex lining. This was vertically suspended with distal clamping and

weighted to ensure 10% longitudinal stretch. A personal computer with bespoke

software for recording the pressure at a sample rate of 10 Hz was used. Infusion at

0.2 ml/min was employed until the graft burst. 6 grafts were subjected to each

method.

2.6. Compliance testing

2.6.1. Flow circuit

The compliance circuit used has been validated and detailed previously (Vara

et al., 2006; Rashid et al., 2008). Briefly, 8 cm lengths of graft were individually

Speed(RPM)

Motor Drive Unit

Haemofilter Heat exchan

Bypa

7.5 MHz Probe

Fluid reservoir

Phantom pulse generator

Fig. 2. Flow circuit for measuring

connected with a 10% longitudinal stretch to a phantom pulse generator (Harvard

Apparatus Model 1421) set at 60 pulses/min and a pulse pressure of 40 mmHg. A

circuit was completed to include a reservoir at variable height to control mean

pressure in the system and a Millar probe sensor within the graft (Fig. 2). Standard

stored leucocyte-depleted blood with SAG-M additive, warmed to 37 1C was used.

Blood from different major antibody groups was not mixed in the circuit.

2.6.2. Distension measurement

The conduits were placed in a de-ionised water reservoir and subjected to

pulsatile flow. A 50 mm 7.5 MHz linear array probe with Doppler ultrasound (Picus,

Pye Medical) was clamped perpendicular to each graft along its longitudinal axis.

Wall tracking (Pie Medical Wall Track System II) was used to monitor distension of

the inner wall for 5 mmHg increments of mean pressures in the range

20–100 mmHg. At each pressure, the distension was measured for six random

cycles once steady state was achieved. Real-time pressure and distension data with

pulsatile flow was used to calculate the compliance

C ¼ðdsys � ddiasÞ

ddiasðpsys � pdiasÞ

C—compliance; dsys—systolic diameter; ddias—diastolic diameter; psys—systolic

pressure; pdias—diastolic diameter.

These measurements were undertaken on 6 thick- and 6 thin-walled speci-

mens.

3. Results

3.1. Grafts

All grafts were 18 cm long with a minimum of 10 cm useablelength. Grafts extruded into warm coagulant solutions weresusceptible to shrivelling on desiccation (Fig. 3a), made especiallyobvious on the systematic dehydration process involved inscanning electron microscopy (Fig. 3b). Only conduits formed incoagulant at 0 1C showed no distortion after storing in dryconditions (Fig. 3c). The wall was composed of four distinct layers– Fig. 3d – comprising from lumen outwards: a lumen skin withsmall pores; large interconnecting pores in high density formingthe majority of the cross-section; small pores in high density nearthe outer edge; a non-porous outer skin with microscopic ridgesthroughout.

ger

US Scanner

PC

ss graft

VOR

Wall track software

dynamic compliance of grafts.

ARTICLE IN PRESS

Fig. 3. (a) Comparison of grafts extruded into deionised ice–water mixture coagulant (i) vs. deionised water at 50 1C (ii) after drying at room temperature for 168 h showing

the shrivelling on dessication associated with the warmer manufacture conditions; (b) coagulant is deionised water at 50 1C; (c) coagulant is deionised water at 0 1C (scale

bar 2 mm) and (d) four-layered structure of the graft wall (scale bar 250mm).

Table 1Effect of different coagulants on composition of the graft wall. Keys: DMAC, N,N-

dimethylacetamide; EtOH, ethanol; LSPS, lumen surface pore size; MPS, maximum

pore size.

Additive to coagulant MPS (mm) % Large pores in wall cross-section LSPS (mm)

5% DMAC 236 89 0–4.5

10% DMAC 389 85 0–13.5

15% DMAC 278 87 0–13.5

20% DMAC 347 89 0–18

5% EtOH 250 60 4.5

10% EtOH 390 88 6.7

15% EtOH 305 92 8.9

20% EtOH 305 85 22.3

Thin 1 Thin 2 Thin 3 Thin 4 Thin 50.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Graft

Wal

l thi

ckne

ss (m

m)

0.0Thick 1 Thick 2 Thick 3 Thick 4 Thick 5

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Graft

Wal

l thi

ckne

ss (m

m)

Thickness of thick-walled grafts (with standard deviation)

Thickness of thin walled grafts (with standard deviation)

Fig. 4. (a) The intra- and inter-graft variation in wall thickness for the thin-walled

grafts and (b) the intra- and inter-graft variation in wall thickness for the thick-

walled grafts.

S. Sarkar et al. / Journal of Biomechanics 42 (2009) 722–730 725

Increasing concentrations of DMAC in the coagulant solutionvisibly slowed the coagulative process. This resulted in an increasein maximum pore size at the luminal aspect, but did notnoticeably affect the other layers. Twenty percent of ethanolcaused some shrivelling on drying, but below this concentration,ethanol’s effect was to increase the luminal surface pore size aswell as increase the uniformity of pore size at the lumen. Theseresults are summarized in Table 1.

Although there are four distinct laminae, they are not separateentities, as they have been produced en-bloc in a single-stepextrusion process, thereby minimising a weaknesses developingbetween the different laminae due to repetitive relative shear stress.

Fig. 4 shows the uniformity of wall thickness across the wholecircumference of each graft as well as the reproducibility of themanufacturing process for thin- and thick-walled specimens,respectively.

3.2. Energy-dispersive X-ray analysis

Fig. 5 shows the typical distribution of elements on the outerand luminal surface of one graft. The uniform distribution ofatoms was replicated at random points along the graft length.

3.3. Tensile testing

Fig. 6a shows the stress–strain relationship for expanded PTFE.Although, the highly crystalline structure endows it with an

extremely high tensile strength, the compliance in the physiolo-gical range is very low. The cast sheet shows even greater strengthand also great distensibility, demonstrated by its ultimate strain

ARTICLE IN PRESS

Fig. 5. (a) The elemental distribution across the outer aspect of the graft with specific atomic mapping for carbon (C Ka1_2), silicon (Si Ka1) and oxygen (O Ka1) and (b) the

elemental distribution across the lumen of the graft with specific atomic mapping for carbon (C Ka1_2), silicon (Si Ka1) and oxygen (O Ka1).

S. Sarkar et al. / Journal of Biomechanics 42 (2009) 722–730726

of nearly 10. However, at low stress, the elasticity is low (Fig. 6b).Testing of the phase-inversion porous graft material in long-itudinal and circumferential directions revealed far greatercompliance at low stress (Fig. 6c and d). By comparing with thecast sheet tensile results, it is evident that porosity reducedstrength greatly, but this still remained above physiological levels.Anisotropy was clearly demonstrated with testing in the circum-ferential direction revealing greater compliance and lowerultimate strength than longitudinal testing. Fig. 7 shows therelationship between wall thickness or dry weight of the testedspecimens and tensile strength for samples of different wall-thicknesses. There was a stronger linear correlation between thedry weight and tensile strength than wall thickness for bothlongitudinal (R2 value 0.9561 vs. 0.856) and circumferential(R2 value 0.9059 vs. 0.6165) test specimens. This may be due tothe relative influence of the different layers of pore size within thewall cross-section.

3.4. Burst pressure

The thin grafts had a mean burst pressure of 221 mmHg(S.D. 19.3 mmHg), for thick-walled grafts this was 357.5(15.41) mmHg.

3.5. Compliance

The compliance characteristics within pulsatile flow aredemonstrated for thick- and thin-walled conduits in Fig. 8. Thethin-walled grafts were more compliant than their thickercounterparts. In addition, the thin-walled grafts became moredistensible at high pressures, whereas distensibility was constantthroughout the physiological range of pressures measured forthick-walled tubes.

ARTICLE IN PRESS

Fig. 5. (Continued)

S. Sarkar et al. / Journal of Biomechanics 42 (2009) 722–730 727

4. Discussion

4.1. Grafts

Ethanol’s effect of reducing variation in pore size has beenshown previously by Chen et al. (1998), albeit at much higherconcentration than here. Khorasani and Shorgashti, (2006b) havefound a similar pore-size standardising effect with alcohols.However, at very high alcohol concentrations, the expectedreduction in size and number of the largest pores (which theyhave termed ‘‘macrovoids’’) was not seen with our polymer. Veryhigh ethanol concentrations rendered our developing graftunstable at the phase-inversion stage.

This extrusion process standardises the traditional dip-coatingtechnique for creating hollow conduits. Charlesworth’s horizontalextruder (Charlesworth et al., 1992) ensured wall thicknessuniformity by introducing a constant rotatory motion duringcoagulation. Our extruder’s vertical approach eliminates theproblems of unequal stresses on the sides of the forming tubedue to gravity. The difference in wall thickness uniformity for the

33mm wall thickness grafts and 50mm conduits was insignificant(mean coefficient of variation 0.1316886 vs. 0.1242805). Theseoverall values are certainly adequate to reliably produce vasculargrafts for research, removing the large expense of industrialmanufacture. Interestingly, the grafts were extruded on 5 mmmandrels but once removed; the neutral internal diameter was4.74–4.88 mm, suggesting that their initial extruded state isslightly stretched.

4.2. Energy-dispersive X-ray analysis

This technique was used for qualitative phase analysis of thegraft material at the atomic level to establish whether the phase-inversion/extrusion process resulted in a specific redistribution ofthe POSS moieties. The atomic spectra for the lumen aspect andouter walls were similar, although Na, K and Cl contaminantswere detected in the outer wall. These contaminants were likelyto be from handling of the grafts. When considering the specificdistribution maps of the selected atoms—C, Si and O, whatappears at first to be a concentration of silicon atoms at the

ARTICLE IN PRESS

PTFE

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.70.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Strain

Stre

ss (M

Pa)

Cast (sheet) nanocomposite

0.0 2.5 5.0 7.5 10.00

5

10

15

20

25

Strain

Stre

ss (M

Pa)

Stre

ss (M

Pa)

Stre

ss (M

Pa)

0.75

0.5

0.25

00.25 0.75 10.5

Strain Strain

Circumferential Graft Specimens Longitudinal Graft Specimens1.25

0.75

0.5

0.25

0.25 0.750.5 1.00

0

1

ThickThinThick

Thin

Fig. 6. . (a) Tensile test of expanded PTFE – ultimate tensile strength was 2.88 MPa, (b) tensile test of cast sheet 2% POSS nanocomposite, (c) tensile test of thin- and thick-

walled porous grafts in the circumferential direction and (d) tensile test of thin- and thick-walled porous grafts in the longitudinal direction.

S. Sarkar et al. / Journal of Biomechanics 42 (2009) 722–730728

microridges is repudiated by the increased signal for all the atoms,and particularly carbon at those points. Hence, this is simply themass effect of the ridges.

4.3. Mechanical testing

Extrusion processing is known to cause the pattern ofmechanical anisotropy (Kurtz et al., 2006) as seen here. The EDXAresults confirm that the anisotropy seen is not due to a molecularredistribution of the POSS nanocage, leaving the hypothesis ofpolymer chain alignment expected from extrusion, which ispreserved during phase-inversion. The initial stress–strain beha-viour shows very little difference in elasticity between the thin-and thick-walled grafts. Once appreciable strain has developed,however, the gradient for the thicker specimens are higher thantheir thinner-walled counterparts. The initial gradients are likelyto be artifactual due to the initial positioning of specimens in thetensile tester. In case of the longitudinal ‘dogbone’ specimens, thisartefact may be avoided by positioning the specimen in the testerwith the aid of a template which is then removed from thespecimen before testing.

The strength of foam structures is often related to the porosityrather than the physical size of the specimen being tested. Stressacross the whole cross-section becomes less meaningful as muchof this is pore cross-section rather than material. As in case with

the porous grafts, dry weight may be a more reliable indicator ofstrength. Furthermore, with a standardised manufacture processfor different thicknesses of graft wall, it was not necessarily thecase that all four laminae varied in proportion due to the variationin the dynamics of coagulant exchange during phase-inversion.

The J-shaped stress–strain relationship seen at high stress ofthe cast sheet tensile test replicates the behaviour of arteries andis highly desirable. The increased stiffness at high pressure is asafety mechanism which prevents wall weakening and aneurysmformation. Highly elastic elastin fibres taking the load at lowpressures with gradual recruitment of collagen at higher loads isresponsible for this in arteries. In case of polyurethanes, there is agradual linear alignment of the haphazard long chains resulting instress crystallisation at higher strain. The highly porous graft doesnot replicate this pattern. In the circumferential direction, there isan almost linear (Hookean) relationship between stress and strainfor grafts of both wall-thicknesses, until failure. This is empha-sised even further by exposure of the grafts to biomimeticpulsatile flow. The dynamic compliance of the thin-walledconduits increases with increasing mean pressure, which trans-lates to increasing strain at high pressure. Although the thin-walled grafts can withstand physiological pressures, there is littlemargin of safety and consideration of the constant fluctuation ofpressure within the haemodynamic cycle would make themunsuitable for the clinical use. The thick-walled specimens exhibituniform dynamic compliance, which combined with burst

ARTICLE IN PRESS

Longitudinal strength of different graft specimens

0.5 1.0 1.5 2.0 2.5 3.00.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

mean wall thicknessspecimen dry weight

0.00

0.01

0.02

0.03

0.04

0.05

0.06

Tensile strength (N)

Mea

n w

all t

hick

ness

(mm

)

Dry

wei

ght (

g)

Circumferential strength of different graft specimens

2 3 4 50.30

0.35

0.40

0.45

0.50

0.55

0.60

0.65

mean wall thicknessspecimen dry weight

0.005

0.006

0.007

0.008

0.009

0.010

0.011

0.012

Tensile Strength (N)

Mea

n w

all t

hick

ness

(mm

)

Dry

wei

ght (

g)

Fig. 7. (a) The linear relation between graft specimen dimensions and longitudinal

strength (dogbone specimens) and (b) the linear relation between graft specimen

dimensions and circumferential strength (ring specimens).

0 25 50 75 100 1250

5

10

Thin walled conduits15Thick walled conduits

Mean Pressure (mmHg)

Com

plia

nce

(%/m

mH

gx10

-2)

Fig. 8. The dynamic compliance of thin- and thick-walled grafts (with standard

deviations, n ¼ 6).

S. Sarkar et al. / Journal of Biomechanics 42 (2009) 722–730 729

pressures comfortably in excess of physiological pressure, makethem a safe proposition for in-vivo and potentially clinical use.The magnitude of the compliance also matches that for nativearteries, with the potential to reduce distal anastomotic intimalhyperplasia from compliance mismatch as seen with PTFE.

5. Conclusion

A gravity-dependent extrusion-phase-inversion device hasbeen used under low-temperature conditions to manufacture

10-cm-length small-diameter vascular conduits in the laboratory,using an anti-thrombogenic polyurethane-based nanocompositepolymer. This method reliably reproduces 34 and 50mm wallthickness conduits with a gradient in porosity through the wall.Using coagulant solutions at 0 1C ensures structural integrity indry storage conditions. Although ethanol in the coagulant hassome effect on reducing pore-size variability, the difference is asmall one. Both ethanol and DMAC had the effect of increasing thelumen pore size. Mechanical characterisation of the graftsrevealed an anisotropic product which failed at stress-levelswhich were higher than those expected in haemodynamic flow.Molecular regional distribution of the POSS nanocage appearedhomogenous on both surfaces of the grafts, so this could notaccount for the anisotropy. We hypothesise that it is due to theextrusion-phase-inversion process, as the pattern of anisotropy isidentical to that found in pure extrusion techniques. The graftswere more distensible at low stress-levels than either PTFE or thenanocomposite cast sheet. However, the distensibility at very lowstress levels was not reliably shown due to technical difficulty inisolating all slack in the specimen at the beginning of the tensiletest. At high strains, the grafts did not stiffen, appearing to becomeperfectly elastic. Dynamic compliance testing highlighted thatrepetitive pulsatile flow cycles promote distensibility of the graftand revealed the thin-walled grafts to increase in dynamicelasticity within the physiological range. This increased elasticitymay herald introduction of weakness in the graft wall andsubsequent aneurysmal behaviour. The thicker-walled conduits,however, had constant compliance through the physiologicalrange. Thick-walled grafts manufactured by this method aremechanically safe for in-vivo investigation. This will enable us toshow whether the biostability and non-thrombogenicity demon-strated in-vitro is confirmed in-vivo. The graft is currently waitingto undergo a clinical trial as a renal access graft.

Conflict of interest statement

None .

Acknowledgment

This study was supported by a grant from the Engineering andPhysical Sciences Research Council (EPSRC).

References

Charlesworth, D., Chian, K.S., Underwood, C.J., 1992. US patent number: US5,132,066.

Chen, J.H., Laiw, R.F., Jiang, S.F., Lee, Y.D., 1998. Microporous segmentedpolyetherurethane vascular graft: dependency of graft morphology andmechanical properties on compositions and fabrication conditions. J. Biomed.Mater. Res. 48, 235–245.

Dempsey, D.J., Phaneuf, M.D., Bide, M.J., Szycher, M., Quist, W.C., LoGerfo, F.W.,1998. Synthesis of a novel small diameter polyurethane vascular graft withreactive binding sites. ASAIO J. 44, M506–M510.

Doi, K., Nakayama, Y., Matsuda, T., 1996. Novel compliant and tissue-permeablemicroporous polyurethane vascular prosthesis fabricated using an excimerlaser ablation technique. J. Biomed. Mater. Res. 31, 27–33.

George, S.J., Izzat, M.B., Gadsdon, P., Johnson, J.L., Yim, A.P., Wan, S., Newby, A.C.,Angelini, G.D., Jeremy, J.Y., 2001. Macro-porosity is necessary for the reductionof neointimal and medial thickening by external stenting of porcine saphenousvein bypass grafts. Atherosclerosis 155, 329–336.

Inoguchi, H., Kwon, I.K., Inoue, E., Takamizawa, K., Maehara, Y., Matsuda, T., 2006.Mechanical responses of a compliant electrospun poly (L-lactide-co-epsilon-caprolactone) small-diameter vascular graft. Biomaterials 27, 1470–1478.

ISO, 2005. Rubber, vulcanized or thermoplastic-determination of tensile stress–-strain properties. ISO 37:2005 (E).

Kannan, R.Y., Salacinski, H.J., Butler, P.E., Seifalian, A.M., 2005. Polyhedraloligomeric silsesquioxane nanocomposites: the next generation material forbiomedical applications. Acc. Chem. Res. 38, 879–884.

ARTICLE IN PRESS

S. Sarkar et al. / Journal of Biomechanics 42 (2009) 722–730730

Kannan, R.Y., Salacinski, H.J., De Groot, J., Clatworthy, I., Bozec, L., Horton, M., Butler,P.E., Seifalian, A.M., 2006a. The antithrombogenic potential of a polyhedraloligomeric silsesquioxane (POSS) nanocomposite. Biomacromolecules 7,215–223.

Kannan, R.Y., Salacinski, H.J., Edirisinghe, M.J., Hamilton, G., Seifalian, A.M., 2006b.Polyhedral oligomeric silsesquioxane-polyurethane nanocomposite microves-sels for an aretificial capillary bed. Biomaterials 27, 4618–4626.

Kannan, R.Y., Salacinski, H.J., Odlyha, M., Butler, P.E., Seifalian, A.M., 2006c. Thedegradative resistance of polyhedral oligomeric silsesquioxane nanocoreintegrated polyurethanes: an in vitro study. Biomaterials 27, 1971–1979.

Kannan, R.Y., Salacinski, H.J., Sales, K.M., Butler, P.E., Seifalian, A.M., 2006d. Theendothelialization of polyhedral oligomeric silsesquioxane nanocomposites:an in vitro study. Cell Biochem. Biophys. 45, 129–136.

Khorasani, M.T., Shorgashti, S., 2006a. Fabrication of microporous polyurethane byspray phase inversion method as small diameter vascular grafts material.J. Biomed. Mater. Res. A 77, 253–260.

Khorasani, M.T., Shorgashti, S., 2006b. Fabrication of microporous thermoplasticpolyurethane for use as small-diameter vascular graft material. I. Phase-inversion method. J. Biomed. Mater. Res. B 76B, 41–48.

Kidson, I.G., Abbott, W.M., 1978. Low compliance and arterial graft occlusion.Circulation 58, I1–I4.

Kurtz, S.M., Mazzucco, D., Rimnac, C.M., Schroeder, D., 2006. Anisotropy andoxidative resistance of highly crosslinked UHMWPE after deformationprocessing by solid-state ram extrusion. Biomaterials 27, 24–34.

Matsumoto, H., Hasegawa, T., Fuse, K., Yamamoto, M., Saigusa, M., 1973. A newvascular prosthesis for a small caliber artery. Surgery 74, 519–523.

Moroni, L., de Wijn, J.R., van Blitterswijk, C.A., 2006. 3D fiber-depositedscaffolds for tissue engineering: influence of pores geometry and

architecture on dynamic mechanical properties. Biomaterials 27,974–985.

Rashid, S.T., Fuller, B., Hamilton, G., Seifalian, A.M., 2008. Tissue engineering of ahybrid bypass graft for coronary and lower limb bypass surgery. FASEB J. 22,2084–2089.

Rosengren, A., Bjursten, L.M., 2003. Pore size in implanted polypropylene filters iscritical for tissue organisation. J. Biomed. Mater. Res. A 67, 918–926.

Salacinski, H.J., Tiwari, A., Hamilton, G., Seifalian, A.M., 2002. Performance of apolyurethane vascular prosthesis carrying a dipyridamole (Persantin) coatingon its lumenal surface-letter. J. Biomed. Mater. Res. 61, 337–338.

Sarkar, S., Hillery, C., Seifalian, A., Hamilton, G., 2006a. Critical parameter of burstpressure measurement in development of bypass grafts is highly dependent onmethodology used. J. Vasc. Surg. 44, 846–852.

Sarkar, S., Salacinski, H.J., Hamilton, G., Seifalian, A.M., 2006b. The mechanicalproperties of infrainguinal vascular bypass grafts: their role in influencingpatency. Eur. J. Vasc. Endovasc. Surg. 31, 627–636.

Seifalian, A.M., Handcock, S., Salacinski, H.J., 2005. Polymer for use in conduits andmedical devices. Patent number: WO2005070998.

Tiwari, A., Salacinski, H.J., Seifalian, A.M., Hamilton, G., 2002. New prostheses foruse in bypass grafts with special emphasis on polyurethanes. Cardiovasc. Surg.10, 191–197.

Tsuchida, H., Wilson, S.E., Ishimaru, S., 1997. Healing mechanisms of high-porosityPTFE grafts: significance of transmural structure. J. Surg. Res. 71, 187–195.

Vara, D.S., Punshon, G., Sales, K.M., Hamilton, G., Seifalian, A.M., 2006. The effect ofshear stress on human endothelial cells seeded on cylindrical viscoelasticconduits: an investigation of gene expression. Biotechnol. Bioeng. 45, 119–130.

Zdrhala, R.J., 1996. Small caliber vascular grafts. Part II: polyurethanes revisited.J. Biomater. Appl. 11, 37–61.