European Journal of Cardio-thoracic Surgery 17 (2000) 440±448

www.elsevier.com/locate/ejcts

Polyurethane: material for the next generation of heart valve prostheses? q

D.J. Wheatley a,*, L. Raco a, G.M. Bernacca a, I. Sim b, P.R. Belcher a, J.S. Boyd c
a
University Department of Cardiac Surgery, Glasgow Royal In®rmary NHS Trust, 10 Alexandra Parade, Glasgow G31 2ER, UK
b
University Department of Medicine and Therapeutics, Western In®rmary, Glasgow, UK
c
Department of Veterinary Anatomy, University of Glasgow, Glasgow, UK
Received 7 September 1999; received in revised form 25 January 2000; accepted 8 February 2000

Abstract
Objectives: The prospects for a durable, athrombogenic, synthetic, ¯exible lea¯et heart valve are enhanced by the recent availability of
novel, biostable polyurethanes. As a forerunner to evaluation of such biostable valves, a prototype trilea¯et polyurethane valve (utilising
conventional material of known in vitro behaviour) was compared with mechanical and bioprosthetic valves for assessment of in vivo
function, durability, thromboembolic potential and calci®cation. Methods: Polyurethane (PU), ATS bilea¯et mechanical, and Carpentier±
Edwards porcine (CE) valves were implanted in the mitral position of growing sheep. Counting of high-intensity transient signals (HITS) in
the carotid arteries, echocardiographic assessment of valve function, and examination of blood smears for platelet aggregates were under-
taken during the 6-month anticoagulant-free survival period. Valve structure and hydrodynamic performance were assessed following
elective sacri®ce. Results: Twenty-eight animals survived surgery (ten ATS; ten CE; eight PU). At 6 months the mechanical valve group
(n ˆ 9) showed highest numbers of HITS (mean 40/h, P ˆ 0:01 cf. porcine valves), and platelet aggregates (mean 62.22/standard ®eld), but
no thromboembolism, and no structural or functional change. The bioprosthetic group (n ˆ 6) showed low HITS (1/h) and fewer aggregates
(41.67, P ˆ 1:00, not signi®cant), calci®cation and severe pannus overgrowth with progressive stenosis. The PU valves (n ˆ 8) showed a
small degree of ®brin attachment to lea¯et surfaces, no pannus overgrowth, little change in haemodynamic performance, low levels of HITS
(5/h) and platelet aggregates (17.50, P , 0:01 cf. mechanical valves, P ˆ 0:23 cf. porcine valves), and no evidence of thromboembolism.-
Conclusions: In the absence of valve-related death and morbidity, and retention of good haemodynamic function, the PU valve was superior
to the bioprosthesis; lower HITS and aggregate counts in the PU valve imply lower thrombogenicity compared with the mechanical valve. A
biostable polyurethane valve could offer clinical advantage with the promise of improved durability (cf. bioprostheses) and low thrombo-
genicity (cf. mechanical valves). q 2000 Elsevier Science B.V. All rights reserved.
Keywords: Polyurethane; Prosthetic heart valve; In vivo; Thrombogenicity

1. Introduction valves have a limited lifetime due to calci®cation (particu-

Over 5000 prosthetic heart valves are implanted annually
in the UK [1]. In spite of more than 30 years of research into
valve development, none of the prostheses currently avail-
able is ideally suited for any clinical situation. While the
mechanical valve is relatively durable, it demands contin-
uous anticoagulation therapy for the patient with the asso-
ciated risks of thromboembolic complication or
anticoagulant-related bleeding episodes. This is a particular
problem where medical services are not always readily
available for patient follow-up, and when patient compli-
ance with anticoagulant regimes is poor. Bioprosthetic
q
Presented at the 13th Annual Meeting of the European Association of
Cardio-thoracic Surgery, Glasgow, Scotland, UK, September 5±8, 1999.
* Corresponding author. Tel.: 1 44-141-211-4730; fax: 1 44-141-552-
0987.
E-mail address: d.j.wheatley@clinmed.gla.ac.uk (D.J. Wheatley)
larly in the young), pannus overgrowth and tissue failure,
although recipients do not usually require long-term antic-
oagulation [2,3].
In haemodynamic terms, the trilea¯et central ¯ow design
of the natural aortic valve and the majority of pericardial
and porcine aortic valves currently available is favoured [4].
Synthetic elastomeric materials allow the possibility of
engineering a material to provide the best qualities of both
mechanical valves and bioprostheses in a new ¯exible
synthetic lea¯et valve prosthesis. Historically, several
synthetic polymers have been tested as lea¯et materials
[5±8]. Silicone and polyole®n rubbers had inadequate
durability [9,10]. A polytetra¯uoroethylene valve was
seriously compromised by thrombosis and calci®cation
[8,11]. Much research has focused on polyurethanes: the
segmented nature of these materials permits alteration of
the polymer chemistry to achieve both ¯exibility and

1010-7940/00/$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
PII: S10 10-7940(00)0038 1-X

Downloaded from https://academic.oup.com/ejcts/article-abstract/17/4/440/439167

by The Technical University of Bucuresti user
on 09 March 2018
 D.J. Wheatley et al. / European Journal of Cardio-thoracic Surgery 17 (2000) 440±448
mechanical strength within the same material. They are bio-
and blood-compatible, with applications in a wide range of
medical devices [12,13]. Several research groups have
investigated a variety of polyurethanes for application in
heart valves [8,9] and have reported problems with calci®-
cation and thrombosis. Long-term in vitro durability of
polyurethane valves has been achieved [14] and, in our
own experience, polyurethane valves manufactured from a
commercially available textile polyurethane were capable
of achieving more than 800 million cycles in laboratory
fatigue testing (equivalent to more than 20 years of `normal'
function) [15]. Calci®cation localised only to tears and
wear-induced defects in the material, during in vitro fatigue
testing [16,17]. Our previous research highlighted the need
to modify the valve lea¯et [18] and frame design (unpub-
lished data) to minimise stenosis of the valve by reducing
the pressure gradient across the valve, especially critical in
smaller sizes. A prototype polyurethane valve has resulted
with lea¯et geometries as previously de®ned [19,20]. The
frame was machined from polyetheretherketone (PEEK),
coated with a thin layer of lea¯et polyurethane. Lea¯ets of
a commercially available polyetherurethane suitable for
animal implantation (Estane 58315, BF Goodrich,
Westerlo-Oevel, Belgium) were dip-coated onto the frame
as previously described [19,20]. This valve design has
achieved durabilities in excess of 400 million cycles during
in vitro fatigue testing.
The present study compares this prototype polyurethane
valve with well-established mechanical and bioprosthetic
valves in a sheep implant model, in terms of in vivo func-
tion, durability, thromboembolic potential and calci®cation
risk.
2. Methods
2.1. Valve implantation
Polyurethane valves (24 mm) were implanted in growing
Texel sheep. ATS bilea¯et mechanical valves (24 mm) and
Carpentier±Edwards supra-annular porcine aortic valves (25
mm) were implanted for comparison. All valves were
placed in the mitral position. The procedures were carried
out in conformance with the Home Of®ce code of practice
and standards, and with the appropriate Home Of®ce project
and personal licences granted under the Animals (Scienti®c
Procedure) Act 1986.
Anaesthesia was induced with alfaxalone/alfadolone
(0.333 ml kg 21) following intravenous diazepam premedi-
cation (0.2 mg kg 21) and maintained with 3% inhaled
iso¯urane. A single bolus of atracurium (0.5 mg kg 21)
was given intravenously.
A left thoracotomy was performed through the bed of the
resected ®fth rib. The animal was heparinised (300 units
kg 21). The descending aorta was cannulated for arterial
return and venous blood drained from the right ventricle
Downloaded from https://academic.oup.com/ejcts/article-abstract/17/4/440/439167
by The Technical University of Bucuresti user
on 09 March 2018
441
via the main pulmonary artery. Normothermic cardiopul-
monary bypass was set up and the heart was electrically
®brillated. The left atrial appendage was opened and the
anterior lea¯et of the mitral valve was resected leaving a
1±2 mm cuff of lea¯et base. The posterior lea¯et was
partially resected with preservation of secondary chordae.
The valve was sutured into the mitral position using 12
interrupted 2/0 Ethibond pledgeted sutures, placed with
the pledgets on the ventricular side and tied on the atrial
side of the mitral annulus. No anticoagulation was used
postoperatively.
Investigative studies were carried out at 6 weeks, 3
months and 6 months after surgery. The sheep were killed
humanely by intravenous overdose of barbiturate. They
were exsanguinated and full post-mortem examinations
performed. The hearts were removed and the valves excised
for further examination. Samples of myocardium from the
free wall of the left ventricle, caudal lobe of the lung, spleen,
kidney, cerebral cortex and right lobe of the liver were
collected into 10% neutral buffered formalin for histological
examination.
2.2. Follow-up investigations
Carotid Doppler studies for detection of high-intensity
transient signals (HITS), echocardiographic assessment of
valve function and examination of blood smears for platelet
aggregation were performed at follow-up. Valve structure
and hydrodynamic performance were assessed following
elective sacri®ce at 6 months.
To collect the Doppler ultrasound data, the animal was
sedated with an intravenous bolus dose of 0.4 mg kg 21
diazepam. Additional boluses of 0.2 mg kg 21 were adminis-
tered every 20 min when required. Thirty minutes of the
Doppler spectra and audio signal were then recorded from
the thoracic outlet of the carotid artery at the base of the
neck, using an EME TC 4040 multichannel transcranial
Doppler ultrasound machine (Eden Medizinische Elektro-
nik, Uberlingen, Germany) equipped with a dual-depth
probe. HITS were counted according to standard criteria
[21]. HITS data for all valves were collected at 3 months
and 6 months of follow-up.
Blood smears were made from femoral arterial samples
collected into EDTA/formalin at 6 weeks, 3 months and 6
months follow-up, and stained with May±Grunwald±
Giemsa stain. Platelet aggregate counts were averaged
over nine standard ®elds under light microscopy at £100
magni®cation.
Haemodynamic function of the valves was assessed by
transthoracic echocardiography, performed after sedation of
the animal with a single dose of 0.2 ml kg 21 of intramus-
cular ketamine. Pressure gradient and ¯ow velocity across
the prosthesis was measured at 6 weeks, 3 months and 6
months.
Mean pressure gradients across the explant valves were
measured in vitro using a pulse duplicator, described
 442 D.J. Wheatley et al. / European Journal of Cardio-thoracic Surgery 17 (2000) 440±448
Table 1
Summary of valve implantation and survival
Valve Number Age at implant (weeks,
implanted mean ^ SD, range)
ATS bilea¯et mechanical 13 19 ^ 1; 17±20
Carpentier±Edwards 12 19 ^ 1; 18±20
Polyurethane 14 21 ^ 4; 17±25
previously [19]. The valves were tested in the mitral posi-
tion at ®ve pulsatile ¯ow rates, ranging from 3.6 to 9.6 l
min 21, at a mean aortic pressure of 95 mmHg.
Organ sections were stained with haematoxylin±eosin. In
addition, kidney sections were stained with Sirius Red to
demonstrate collagen and splenic sections by the Perl's
method for iron. Thin sections of explant polyurethane
valve lea¯ets were prepared from wax-embedded portions
of lea¯et. The sections were stained for calcium using
Alizarin Red S or von Kossa: von Kossa-stained sections
were counter-stained with haematoxylin±eosin, Gomori or
Von Gieson stains for examination of ®brous material
attached to the explanted lea¯ets. Following sputter-coating
with gold, segments of lea¯et surface were mounted for
scanning electron microscopy and examined at 15 kV accel-
erating voltage.
2.3. Statistical analysis
Results are expressed as mean (SD) unless otherwise
stated. Statistical analyses were performed using Minitab
for Windows, release 12 (Minitab Inc.), assuming a signi®-
cance level of 5%. HITS data were incremented by 1, to
remove zero values, followed by logarithmic transforma-
tion, to approximate normality among groups. The trans-
formed data were then analysed using a general linear
model (ANOVA), incorporating the three follow-up times
as a repeated-measures variable, with Bonferroni-corrected
multiple post hoc comparisons (using a Minitab macro
supplied by Glasgow University Department of Statistics).
Logarithmic transformation was applied to aggregate data
prior to similar data analysis. Velocity data from echocar-
diographic studies were compared by similar methods with-
out data transformation. Pressure gradient data could not be
transformed to approximate normality and equal variance,
and hence were analysed using Kruskal±Wallis one-way
analysis of variance.
3. Results
Details of the numbers of sheep implanted for each valve
type and post-operative survivors are displayed in Table 1.
Of ten post-operative survivors in the mechanical valve
Downloaded from https://academic.oup.com/ejcts/article-abstract/17/4/440/439167
by The Technical University of Bucuresti user
on 09 March 2018
Post-op. Survivors to 6 Age of 6-month Valve-related
survivors months survivors at implant deaths
(weeks, mean ^ SD, range)
10 9 19 ^ 1; 18±20 Possibly one cause
unknown
10 6 19 ^ 1; 18±20 Two stenosis due
to pannus
overgrowth
8 8 22 ^ 4; 17±25 None
group, one death occurred at 6.6 weeks but no de®nitive
pathology was discovered at post-mortem examination. A
total of nine sheep survived to 6 months with a mechanical
valve implant. Two out of ten post-operative survivors in the
porcine valve group died, at 2 weeks and 15.14 weeks, from
valve-related causes: both valves were heavily stenosed,
associated with pannus overgrowth of the lea¯ets. A total
of six sheep survived to 6 months with a porcine valve
implant. No animals in the polyurethane valve group
(eight post-operative survivors) died from valve-related
causes. All eight post-operative survivors with a polyur-
ethane valve implant survived to 6 months. The mechanical
valves were unchanged in appearance at explant with no
signi®cant pannus overgrowth of the valve lea¯ets (Fig.
1). All the porcine valve explants were affected by extensive
pannus overgrowth with minor focal calci®cation (Fig. 2).
One polyurethane valve had two lea¯ets restrained by a
chordal loop. All polyurethane valve lea¯ets showed a
degree of ®brin attachment with no signi®cant pannus over-
growth and calci®cation that was exclusively associated
with ®brin material attached to the lea¯ets (Fig. 3).
HITS per hour are detailed in Table 2. Overall, there was
a signi®cant effect of valve type, with the mechanical valve
group producing greater numbers of HITS than either the
porcine (P , 0:01) or the PU (P , 0:01) valve types. At 3
months, mechanical valves produced signi®cantly greater
Fig. 1. Out¯ow surface of mechanical valve viewed in situ immediately
prior to explant.
 D.J. Wheatley et al. / European Journal of Cardio-thoracic Surgery 17 (2000) 440±448 443

Fig. 3. Explant polyurethane valve viewed in situ immediately prior to

Fig. 2. Porcine valve viewed in situ immediately prior to explant. (a) In¯ow explant. (a) In¯ow surface; (b) out¯ow surface; (c) radiograph of explant
surface; (b) out¯ow surface; (c) radiograph of explant porcine valve. polyurethane valve.

numbers of HITS than the porcine valves (P , 0:01). No

polyurethane valve produced HITS at 3 months, thus,
Table 2
although by de®nition this group produced signi®cantly HITS data at 3 and 6 months follow-up
fewer HITS than any other group, it could not be included
in a formal statistical analysis. At 6 months, mechanical Valve type n Follow-up time Mean HITS/h SD
valves produced signi®cantly more HITS than did porcine Mechanical 9 3 73 81.7
valves (P ˆ 0:01), but neither the difference between 9 6 40 41.1
mechanical and polyurethane (P ˆ 0:60) nor between Porcine 8 3 7 12.2
porcine and polyurethane (P ˆ 1:00) reached signi®cance. 6 6 1 2.5
There were no signi®cant differences within any valve type Polyurethane 8 3 0 0
8 6 5 4.0
over time.

Downloaded from https://academic.oup.com/ejcts/article-abstract/17/4/440/439167

by The Technical University of Bucuresti user
on 09 March 2018
 444 D.J. Wheatley et al. / European Journal of Cardio-thoracic Surgery 17 (2000) 440±448

Table 3 cantly lower velocities than the porcine valve group

Platelet aggregates at 6 weeks, 3 months and 6 months follow-up
Valve type n Follow-up time Mean aggregates
Mechanical 9 6 weeks 14.14
9 3 months 50.67
9 6 months 62.44
Porcine 8 6 weeks 5.29
8 3 months 18.78
6 6 months 41.67
Polyurethane 8 6 weeks 15.62
8 3 months 16.63
8 6 months 17.50
Aggregate counts for each valve type are detailed in
Table 3. At 6 weeks, both mechanical (P , 0:01) and poly-
urethane (P , 0:01) valves had greater numbers of aggre-
gates than porcine valves. At 3 months, mechanical valves
were associated with signi®cantly greater numbers of aggre-
gates than either porcine (P , 0:01) or polyurethane
(P , 0:01) valves: there was no signi®cant difference
between porcine and polyurethane valves. At 6 months,
mechanical valves were associated with signi®cantly more
aggregates than polyurethane valves (P , 0:01): there was
no signi®cant difference between mechanical and porcine
valves (P ˆ 1:00). Both mechanical and porcine valves
were associated with signi®cantly increased aggregate
numbers with time, comparing 6 weeks with 3 months
(P , 0:01 for both valve types) and 6 months (P , 0:01
for both valve types). Differences between 3 and 6 months
did not reach signi®cance although there remained a trend
towards increased aggregation with time, particularly in the
porcine valve group (P ˆ 0:09). Platelet aggregation
remained constant with time in animals with polyurethane
valves (P ˆ 1:00).
Examination of the correlation between HITS and platelet
aggregate numbers, using a two-tailed test on the trans-
formed data described above, con®rms a signi®cant positive
correlation (r ˆ 0:287, P ˆ 0:032).
Echocardiographic data are presented in Table 4. At 6
weeks and 3 months, there were no signi®cant differences
in blood velocity across the three valve types. At 6 months,
the mechanical valve and polyurethane valve groups were
not signi®cantly different and both groups produced signi®-
(P , 0:01 for both valve types). Neither the mechanical
SD nor the polyurethane valve groups indicated any increase
in velocity with time. The porcine valve group produced
8.07 increasing velocities between 6 weeks and 3 months
20.48
(P ˆ 0:06), with differences becoming signi®cant compar-
27.03
4.39 ing 6 weeks with 6 months (P , 0:01). Signi®cant increases
9.96 in valvular pressure gradients were also associated with the
25.77 porcine valve group at 3 months and 6 months follow-up.
2.77 No other signi®cant differences in pressure gradients were
2.13
present among valve types.
2.27
In vitro hydrodynamic testing of the explant valves
con®rms the haemodynamic ®ndings of the echocardiogra-
phy (Fig. 4). The mechanical valve group produces the
lowest group mean pressure gradient pro®le. Data is
presented here for the mean of the eight PU explant valves
(Fig. 4). The polyurethane explant group is more stenotic at
6 months than the mechanical valve group. The porcine
valve group is extremely stenotic, and, indeed, it was possi-
ble to measure the mean pressure gradient of the six CE
valves at only the four lowest ¯ow rates.
No pathological ®ndings in the animal organs were attri-
butable to the presence of any of the prosthetic valves. Some
pathology present was associated with the death of the
animal and with the implant operation itself, for example,
minor myocardial necrosis associated with de®brillation.
Some minor degree of renal infarction was present in four
animals implanted with the porcine bioprosthesis. These
were old infarcts of similar age, probably associated with
the absence of anti-coagulation therapy during the initial
post-operative period while the sewing ring and frame are
exposed to the blood, rather than attributable to any pros-
thetic valve per se.
4. Discussion
There has long been an interest in developing a prosthetic
heart valve that combines the durability of a mechanical
valve with the ¯ow pro®le and blood compatibility of a
bioprosthesis, while overcoming the disadvantages of each
of these types of valve. Polyurethanes have been a popular
choice of material, given their relatively good blood- and

Table 4
Echocardiographic data at 6 weeks, 3 months and 6 months follow-up

Valve type n Follow-up time Mean velocity (m s 21) SD Mean pressure gradient (mmHg) SD

Mechanical 9 6 weeks 0.83 0.24 3.1 1.73

9 3 months 0.81 0.22 3.2 1.55
9 6 months 0.81 0.22 3.3 1.58
Porcine 8 6 weeks 0.81 0.20 2.7 1.34
8 3 months 1.14 0.26 5.8 1.92
6 6 months 1.34 0.17 7.8 1.84
Polyurethane 8 6 weeks 0.77 0.14 2.3 0.89
8 3 months 0.82 0.17 2.7 1.48
8 6 months 0.87 0.26 3.0 0.95

Downloaded from https://academic.oup.com/ejcts/article-abstract/17/4/440/439167

by The Technical University of Bucuresti user
on 09 March 2018
 D.J. Wheatley et al. / European Journal of Cardio-thoracic Surgery 17 (2000) 440±448
Fig. 4. Mean pressure gradient versus RMS ¯ow (mean value ^ SEM
plotted for nine mechanical valve explants, all eight polyurethane valve
explants and ®ve polyurethane valve explants at ®ve pulsatile ¯ow rates,
and six porcine valve explants at four pulsatile ¯ow rates).
bio-compatibility. However, there have been problems asso-
ciated with long-term implant, with material degradation
causing premature failure of devices. Thus, the development
of heart valves made of such materials has been slower than
anticipated. Recent developments in polymer science have
resulted in the synthesis of a number of reportedly biostable
polyurethanes. The availability of such materials has
con®rmed the usefulness of polyurethanes in their ability
to be chemically engineered to ®t a speci®c purpose and
has rekindled interest in the use of these materials in pros-
thetic heart valves. As a prelude to developing research
incorporating biostable polyurethanes into new valve
designs, this study utilised a well-characterised, non-
biostable polyurethane to investigate the ability of an altered
polyurethane valve design to perform satisfactorily in a
short-term implant environment. Major concerns in respect
of valve design are the haemodynamic performance and the
propensity of the design and/or material to induce a thom-
botic blood response, either on the valve itself or in the
circulation.
The haemodynamic function of the valves was assessed
by echocardiography and by in vitro hydrodynamic
measurements on the valves. The thrombotic response was
assessed by counting platelet microaggregates and HITS.
The exact nature of HITS is controversial. Current
opinion is that such HITS result from circulating gaseous
microbubbles, although there have been suggestions that a
contribution to HITS formation comes from circulating
microthrombi. HITS frequency is higher in recipients of
mechanical valve prostheses, which require systemic antic-
oagulation, in contrast to bioprosthetic valves, which are
relatively non-thrombogenic [22,23]. HITS have also been
detected, with this technique, in patients with clinically
evident embolic sources, e.g. atrial ®brillation or carotid
stenosis without the presence of a prosthetic valve, although
Downloaded from https://academic.oup.com/ejcts/article-abstract/17/4/440/439167
by The Technical University of Bucuresti user
on 09 March 2018
445
in these cases the nature and aetiology of the HITS may be
different [24]. In our study, HITS were also much more
frequent in the mechanical valve group than in either the
porcine or the polyurethane valve groups, and correlated
signi®cantly with platelet aggregate counts at 3 and 6
months. As expected, the mechanical valves produced the
greatest numbers of aggregates re¯ecting their higher
thrombogenicity. As stenosis progressed, the porcine valves
produced larger numbers of aggregates suggesting an
increasing tendency to thrombogenicity, although they did
not produce an increased frequency of HITS with time.
Regardless of whether HITS are directly related to throm-
bogenesis, the correlation between HITS and platelet aggre-
gation suggests that these phenomena are related, so that
valve factors that tend to increase thrombogenicity also
act to increase HITS. Therefore, the low HITS and aggre-
gate numbers associated with the polyurethane valve group
suggest that the polyurethane valve, like the bioprosthesis, is
relatively non-thrombogenic. Moreover, the aggregate
numbers suggest that aggregate counts are the more sensi-
tive means of assessing thrombogenic potential and that a
stenosed bioprosthesis develops an increased thrombogenic
potential. The sheep model tends to be less thrombogenic
than human recipients of valves, hence the success of
mechanical valve implants without the use of long-term
anti-coagulation. However, the pattern of HITS production
is generally similar to that found in our clinical experience
and suggests a similar pattern of behaviour. The current
study makes use of mechanical and bioprosthetic valve
controls over the polyurethane valve implants, and hence
the data produced are comparative and might reasonably
be expected to re¯ect a similar pattern of behaviour in the
human recipient.
In terms of haemodynamic function, as assessed by echo-
cardiography, the polyurethane valves performed as well as
the mechanical valve, with the porcine valve function
becoming compromised with time. More sensitive labora-
tory-based hydrodynamic testing detected larger differences
between the mechanical valve and the polyurethane valve,
but the difference was not clinically signi®cant at any
follow-up time. The polyurethane valves were less stenotic
after 6 months than the porcine valves. Three polyurethane
valves were seriously affected by thrombus and ®brous
attachment to the lea¯ets: one of these also had two lea¯ets
restrained by a chordal loop around one of the stent posts.
The ®brous material was also calci®ed, and these valves
were considerably more stenotic than the remaining ®ve
polyurethane valves. The variability of the group mean pres-
sure gradient data is greatly increased by the inclusion of the
three stenosed polyurethane valves: the remaining ®ve
valves produced a more consistent performance (Fig. 4).
The three most stenotic PU valves were implanted in the
youngest sheep, between 16.86 and 18 weeks: mechanical
and porcine valves had one sheep each implanted at 17.91
weeks, with all others implanted after 18.20 weeks and the
remaining polyurethane valves were implanted in sheep
 446 D.J. Wheatley et al. / European Journal of Cardio-thoracic Surgery 17 (2000) 440±448
Fig. 5. Thin section of polyurethane valve lea¯et, stained for calcium, with von Kossa (£312).
between 24 and 25 weeks. No other factors were different
among the valves. The differences in the three younger
sheep implanted with polyurethane valves may have been
partly related to the age at implant of the sheep or to physi-
cal factors associated with the individual implants.
However, further implant experience with various ages of
animals would be required to clarify this.
Overall, the polyurethane valve performed better than the
porcine bioprosthesis. In terms of blood ¯ow parameters,
the mechanical valve performed best, but the risk of throm-
boembolic events remained and may have caused the
premature death of one animal in this group. Only the
porcine valve function was affected by pannus overgrowth,
resulting in severe stenosis of the valve. There was only
minor focal calci®cation detected (Fig. 2c), which was
insuf®cient to affect lea¯et function. There were no failures
of the polyurethane valve lea¯ets and the slight increase in
valve stenosis was due to the attachment of ®brous and
thrombotic material to the lea¯et surface and the subsequent
calci®cation of this extrinsically attached material. In
general, the ®brous, thrombosed network on the lea¯et
Fig. 6. Scanning electron microscopy image of polyurethane valve lea¯et,
showing area of degradation surrounded by intact polymer.
Downloaded from https://academic.oup.com/ejcts/article-abstract/17/4/440/439167
by The Technical University of Bucuresti user
on 09 March 2018
was a surface layer, clearly separate from the underlying
lea¯et polymer. A number of points on the lea¯et surface
indicated a concentration of this thrombus, with associated
calci®cation, penetrating into the body of the polyurethane
(Fig. 5). This is suggestive of a degradative process affect-
ing the polyurethane surface structure, which is likely to
accelerate as calci®ed host material penetrates the polymer.
A few areas of the most severely ®brosed/thrombosed poly-
urethane valve lea¯ets appeared to be undergoing a degra-
dative process, viewed by scanning electron microscopy
(Fig. 6). Calci®cation of the lea¯ets was entirely associated
with extrinsically calci®ed material attached to the lea¯et
with some early penetration from the surface (Fig. 5), which
appeared to be associated with polymer degradation. No
intrinsic polyurethane calci®cation was found associated
with intact areas of polymer. It is likely that such ingressions
would cause stress concentrations and/or material thinning,
leading eventually to material failure. The problems of poly-
urethane biostability are well known [12] and such polymers
are especially vulnerable when used in demanding ¯exing
applications in which fatigue-induced damage seems to
accelerate material biodegradation [25]. This problem has
been the subject of much research interest, and a variety of
approaches to its solution has been attempted. Many of these
approaches have improved the biostability of polyurethanes
and it is well recognised that a safe, durable and effective
polymeric prosthetic valve is imminent [12].
As part of a multi-disciplinary programme 1 to develop a
polyurethane valve design with good biological durability,
fatigue resistance and haemodynamics, we are now working
with a new generation of biostable polyurethanes which
have proven themselves of superior biostability in a
demanding 6-month, strained, rat implant model. We antici-
1
Current MedLINK programme, in association with the University of
Liverpool, Department of Clinical Engineering, the University of Leeds,
Department of Mechanical Engineering, and AorTech International, with
funding from the U.K. Department of Health and the Engineering and
Physical Sciences Research Council.
 D.J. Wheatley et al. / European Journal of Cardio-thoracic Surgery 17 (2000) 440±448
pate early development of a polyurethane valve which has
good haemodynamic function maintained during long-term
implant, and which neither fails from biological degradation
nor from fatigue-induced material failure, while maintain-
ing a low thrombogenic surface.
Acknowledgements
This research was funded by the British Heart Founda-
tion.
References
[1] UK Department of Health. The United Kingdom Heart Valve Regis-
try, UK: Department of Health, 1992.
[2] Giddens DP, Yoganathan AP, Schoen FJ. Prosthetic cardiac valves.
Cardiovasc Pathol 1993;2:167S±177S.
[3] Bernacca GM, Mackay TG, Wheatley DJ. In vitro calci®cation of
bioprosthetic heart valves: report of a novel method and review of
the biochemical factors involved. J Heart Valve Dis 1992;1:115±130.
[4] Chandran KB, Fatemi R, Schoephoerster R, Wurzel D, Hansen G,
Pantalos G, Yu L-S, Kolff WJ. In vitro comparison of velocity pro®les
and turbulent shear distal to polyurethane trilea¯et and pericardial
prosthetic valves. Artif Organs 1989;13:148±154.
[5] Hilbert SL, Ferrans VJ, Tomaita Y, Eidbo EE, Jones MM. Evaluation
of explanted valves: performance and blood compatibility. Life
Support Systems 1987;4:130±132.
[6] Lo HB, Herold M, Reul H, Muckter H, Taguchi K, Surmann M,
Hildinger KH, Lambertz H, De Haan H, Handt S, Hollweg G, Mess-
mer BJ, Rau G. A tricuspid polyurethane heart valve as an alternative
to mechanical prostheses or bioprostheses. Trans Am Soc Artif Int
Organs 1988;34:839±844.
[7] Kolff WJ, Yu LS. The return of elastomer valves. Ann Thorac Surg
1989;48:S98±S99.
[8] Nistal F, Garcia-Martinez V, Arbe E, Fernandez D, Mazzora F, Gallo
I. In vivo experimental assessment of polytetra¯uoroethylene trileaf-
let heart valve prosthesis. J Thorac Cardiovasc Surg 1990;99:1074±
1081.
[9] Chetta GE, Lloyd JR. The design, fabrication and evaluation of a tri-
lea¯et prosthetic heart valve. J Biomech Eng 1980;102:34±41.
[10] Kiraly R, Yozu R, Hillegass D, Harasaki H, Murabayashi S, Snow J,
Nose Y. Hexsyn trilea¯et valve: application to temporary blood
pumps. Artif Organs 1982;6:190±197.
[11] Imamura E, Kaye MP. Function of expanded polytetra¯uoroethylene
laminated trilea¯et valves in animals. Mayo Clin Proc 1977;52:770±
775.
[12] Tanzi MC, Ambrosio L, Nicolais L, Iannace S, Ghislanzoni L,
Mambrito B. Comparative physical tests on segmented polyurethanes
for cardiovascular applications. Clin Mater 1991;8:57±64.
[13] Zdrahala RJ, Zdrahala IJ. Biomedical applications of polyurethanes: a
review of past promises, present realities and a vibrant future. J
Biomater Appl 1999;14:67±90.
[14] Jansen J, Reul H. A synthetic three-lea¯et valve. J Med Eng Technol
1992;16:27±33.
[15] Bernacca GM, Mackay TG, Wheatley DJ. In vitro function and
durability of a polyurethane heart valve: material considerations. J
Heart Valve Dis 1996;5:538±542.
[16] Bernacca GM, Mackay TG, Wilkinson R, Wheatley DJ. Calci®cation
and fatigue failure in a polyurethane heart valve. Biomaterials
1995;16:279±285.
[17] Bernacca GM, Mackay TG, Wilkinson R, Wheatley DJ. Polyurethane
heart valves: fatigue failure, calci®cation and polyurethane structure.
J Biomed Mater Res 1997;34:371±379.
Downloaded from https://academic.oup.com/ejcts/article-abstract/17/4/440/439167
by The Technical University of Bucuresti user
on 09 March 2018
447
[18] Bernacca GM, Mackay TG, Gulbransen MJ, Donn AW, Wheatley DJ.
Polyurethane heart valve durability: effects of lea¯et thickness. Int J
Artif Organs 1997;20:327±331.
[19] Mackay TG, Wheatley DJ, Bernacca GM, Hindle CS, Fisher AC. A
new polyurethane heart valve prosthesis: design, manufacture and
evaluation. Biomaterials 1996;17:1857±1863.
[20] Mackay TG, Bernacca GM, Wheatley DJ, Fisher AC, Hindle CS. In
vitro function and durability assessment of a polyurethane heart valve
prosthesis. Arti®cial Organs 1996;20:1017±1025.
[21] Ringelstein EB, Droste DW, Babikian VL, Evans DH, Grosset DG,
Kaps M, Markus HS, Russell D, Siebler M. Consensus on microem-
bolus detection by TCD. International Consensus Group on Micro-
embolus Detection. Stroke 1998;29(3):725±729.
[22] Mackay TG, Georgiadis D, Grosset DG, Lees KR, Wheatley DJ. On
the origin of cerebrovascular microemboli associated with prosthetic
heart valves. Neurol Res 1995;17(5):349±352.
[23] Graf TF, Fischer H, Reul H, Rau G. Cavitation potential of mechan-
ical heart valve prostheses. Int J Artif Organs 1991;14(3):169±174.
[24] Gaunt MEB, Mackay TGG, Sitzer MM, Steinmetz H. Unstable caro-
tid plaques: preoperative identi®cation and association with intrao-
perative embolisation detected by transcranial Doppler. Eur J Vasc
Endovasc Surg 1996;11(1):78±82.
[25] Kaps M, Hansen J, Weiher M, Tiffert K, Kayser I, Droste DW. Clini-
cally silent microemboli in patients with arti®cial prosthetic aortic
valves are predominantly gaseous and not solid. Stroke
1997;28(2):322±325.
Appendix A. Conference discussion
Dr D. Cosgrove (Cleveland, OH, USA): Will you describe the electron
microscopic surface of this material? Is it extremely smooth or do you
expect in-growth?
Professor Wheatley: The new polyurethanes that we are testing have
been strained to 100% and implanted in rat models for 6 months, and the
appearance looks entirely homogeneous, absolutely amorphous. There is no
evidence of any pitting or irregularity. Most of the other polyurethanes
won't last 3 months, and they have craters and look like the surface of
the moon in this test.
Dr R. Frater (Bronxville, NY, USA): We tested an earlier generation of
polyurethane three-lea¯et valves for a year in sheep, and at a year they all
had intrinsic calci®cation, which was de®nitely not related to a surface
phenomenon. It was within the substance of the polyurethane. I know
polyurethane comes in many forms, and I am not sure whether the form
of polyurethane you are using is necessarily different from the form we
would have used some 10 years ago in terms of the potential for intrinsic
calci®cation over time.
Professor Wheatley: I am not sure that I can answer that question. I
really need my biochemical and polymer chemist colleagues to tell me that,
but the structure of the soft segment is such that they don't believe that this
will calcify. We have had this in a fairly demanding animal model for quite
a good period, 6 months, but we have also had it in a calcifying model in the
laboratory for equivalent periods of much longer than this, and there
doesn't seem to be any intrinsic calci®cation at all. The answer is clearly
a biochemical one and a chemical one, and I don't know that anybody really
understands that.
Mr A. Ritchie (Cambridge, UK): A question or two. The new crop of
stentless valves that is on the market now requires no warfarin therapy, as is
the prospect with the polyurethane valves, so they are equivalent in that sort
of a sense. The question which will remain over the stentless valves is their
durability, which the polyurethane valve has to answer also. We have seen
in three of your valves, perhaps as some may judge, early signs of failure.
Where do you think they might lie in competing with stentless valves if that
continues to show up?
Professor Wheatley: The changes in three of the valves were quite
clearly related to a surface phenomenon. It was the expected ®nding we
would have with a nonbiostable material. We want these valves to last at
 448 D.J. Wheatley et al. / European Journal of Cardio-thoracic Surgery 17 (2000) 440±448
least as well as the current bioprostheses would in adults, because I think
the future for this valve, at least initially, will be the parts of the world
where rheumatic fever occurs in young patients, where the bioprosthesis is
not an option as it behaves like the juvenile sheep, and where mechanical
valves are not feasible. Stentless valve implantation techniques are such
that the degree of skill needed for these makes them sometimes not the most
obvious option. We can certainly fatigue these in the laboratory to do better
than the current frame-mounted bioprostheses, but the answers are clearly
going to come from a lot more further testing.
Downloaded from https://academic.oup.com/ejcts/article-abstract/17/4/440/439167
by The Technical University of Bucuresti user
on 09 March 2018
Dr Cosgrove: Could you tell us the mode of failure on the accelerated
wear tester?
Professor Wheatley: Usually it is a fatigue-type fracture of the material,
a crack which propagates slowly. Crack propagation is one of the concerns;
we have to make sure there is good resistance to that. One of the arguments
that was put up for the homografts many years ago is that they fail slowly,
and we hope that such will be the same for these valves, not in the near
future but perhaps that is the way these will eventually fail.