ARTICLE IN PRESS

Biomaterials 28 (2007) 5009–5027

www.elsevier.com/locate/biomaterials

Leading Opinion

Prosthetic vascular grafts: Wrong models,

wrong questions and no healing$
Peter Zilla, Deon Bezuidenhout, Paul Human
Christian Barnard Department of Cardiothoracic Surgery/Cardiovascular Research Unit, University of Cape Town Medical School, Cape Town, South Africa
Received 5 June 2007; accepted 6 July 2007
Available online 3 August 2007

Abstract

In humans, prosthetic vascular grafts remain largely without an endothelium, even after decades of implantation. While this
shortcoming does not affect the clinical performance of large bore prostheses in aortic or iliac position, it contributes significantly to the
high failure rate of small- to medium-sized grafts (SMGs). For decades intensive but largely futile research efforts have been under way
to address this issue. In spite of the abundance of previous studies, a broad analysis of biological events dominating the incorporation of
vascular grafts was hitherto lacking. By focusing on the three main contemporary graft types, expanded polytetrafluoroethylene
(ePTFE), Dacron and Polyurethane (PU), accumulated clinical and experimental experience of almost half a century was available. The
main outcome of this broad analysis—supported by our own experience in a senescent non-human primate model—was twofold: Firstly,
inappropriate animal models, which addressed scientific questions that missed the point of clinical relevance, were largely used. This led
to a situation where the vast majority of investigators unintentionally studied transanastomotic rather than transmural or blood-borne
endothelialization. Given the fact that in patients transanastomotic endothelialization (TAE) covers only the immediate perianastomotic
region of sometimes very long prostheses, TAE is rather irrelevant in the clinical context. Secondly, transmural endothelialization seems
to have a time window of opportunity before a build-up of an adverse microenvironment. In selecting animal models that prematurely
terminate this build-up through the early presence of an endothelium, the most significant ‘impairment factor’ for physiological tissue
regeneration in vascular grafts remained ignored.
By providing insight into mechanisms and experimental designs which obscured the purpose and scope of several decades of vascular
graft studies, future research may better address clinical relevance.
r 2007 Elsevier Ltd. All rights reserved.

Keywords: Vascular grafts; Endothelialization; Polytetrafluoroethylene; Polyethylene terephthalate; Polyurethane; Foreign body response

1. Introduction and iliac surgery, smaller diameter grafts became the

For almost half a century synthetic vascular grafts have
been an integral tool of vascular surgery. However, while
large bore prostheses added an impressive scope to aortic
$
Note: Leading Opinions: This paper provides evidence-based scientific
opinions on topical and important issues in biomaterials science. They
have some features of an invited editorial but are based on scientific facts,
and some features of a review paper, without attempting to be
comprehensive. These papers have been reviewed for factual, scientific
content.
Corresponding author. Cape Heart Centre, University of Cape Town/
Faculty of Health Services, Observatory 7925, Cape Town, South Africa.
E-mail address: peter.Zilla@uct.ac.za (P. Zilla).
nemesis of research and a symbol for the limitations of
modern biotechnology. The main reasons for the poor
performance of small- (p4 mm) to medium-sized (p7 mm)
grafts (SMGs) are anastomotic intimal hyperplasia and
ongoing surface thrombogenicity. In spite of these perpe-
tual shortcomings of contemporary vascular prostheses, no
alternative concept has yet emerged that promises to
replace the current generation of synthetic grafts soon.
While a small spearhead of researchers pursues sophisti-
cated tissue engineering approaches [1–4] the majority of
surgeons continue to implant the well-established products
of the past decades. Similarly, commercial grafts continue
to be manufactured on the basis of material choices,

0142-9612/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biomaterials.2007.07.017
 ARTICLE IN PRESS
5010 P. Zilla et al. / Biomaterials 28 (2007) 5009–5027

mechanical strength, regulatory compliance and surgical

preferences rather than with a view towards biological
integration and functional tissue regeneration. Neverthe-
less, regenerative medicine has made significant inroads in
recent years and it seems only a matter of time until
synthetic vascular grafts will also benefit from this
development. Before embarking on concepts stimulating
the in situ generation of functional vascular tissue,
however, it seems paramount to understand why such a
physiological tissue formation would not spontaneously
occur in contemporary grafts. For reasons still incomple-
tely understood, neither transmural ingrowth through the
graft wall nor transanastomotic ingrowth from the
adjacent artery seem to be capable of endothelializing
more than a narrow zone confined to the immediate
anastomotic area in humans [5,6]. Given the high number
of annual implants, it is surprising how poorly investigated
these limitations have remained throughout the years.
While the initial enthusiasm for the availability of polyester
grafts prompted several major studies in the 1960s and
1970s [7–10], relatively little has been added to our
understanding since then. Typical for subsequent prosthe-
tic vascular graft research, intensive efforts went into ever
new permutations of prostheses without a sound baseline Fig. 1. Schematic representation of rapid trans-anastomotic endothelia-
lization (TAE) in animal models (bottom) compared to limited TEA in
knowledge of pathological events behind the healing humans (top). Typically, the transanastomotic ingrowth stoppage in
impairment of grafts which have been clinically implanted humans results in permanently non-endothelialized midgraft sections. By
for decades. Furthermore, by choosing inadequate animal prematurely reaching confluence in experimental grafts through transa-
models and too short graft lengths, an involuntary nastomotic ingrowth, neither blood-born endothelial cells nor ingrowing
emphasis of the majority of these studies was on capillaries from outside can unfold their potential to populate the blood
surface.
transanastomotic endothelial ingrowth—a biological phe-
nomenon of utter irrelevance for the clinical set-up.
the time of retrieval (Fig. 1) [11]. Until 10 years ago,
2. Transanastomotic endothelialization (TAE): barking up approximately 90% of prostheses in large animal models
the wrong tree were as short as 5.571.2 cm (Dacron and expended
polytetrafluoroethylene (ePTFE)) and 4.571.7 cm polyur-
Given the key role endothelium plays in preventing a ethanes (PU), with a median implantation periods of
blood vessel from occluding, it is understandable that the 91.8717.3 days for Dacron grafts, 60.079.36 days for
main focus of vascular graft studies was on endothelializa- polytetrafluoroethylene (PTFE) grafts [11] and 907127.1
tion. It is therefore even more surprising that the majority days for PU grafts. In more recent years, slightly longer
of investigators chose largely inadequate models for grafts were implanted (64% were still shorter than 7 cm)
assessing midgraft endothelialization although the clinical and observation periods became shorter (medians:
background to all these studies has been unambiguous 28.078.9 and 14.0718.7 for ePTFE and Dacron, respec-
throughout the years. It has been known for more than tively) [12–29]. In the small proportion of studies where
four decades that in humans, transanastomotic endothelial truly long prostheses were implanted, they were either
ingrowth does not exceed more than 1–2 cm [6] even after aorto–aortic [30–35], aorto–iliac [36–39] or aorto–femoral
years of implantation. At the same time peripheral bypass [40]. As much as these grafts allowed one to distinguish
grafts are usually between 40 and 60 cm long. Thus, TAE is transanastomotic from transmural endothelium, their
obviously an irrelevant mechanism for achieving confluent surrounding environment differed so profoundly from that
surface endothelialization on clinically implanted grafts. of clinically implanted grafts that deductions regarding
Yet, TAE was the very phenomenon that was inadvertently transmural endothelialization in humans need to be seen
studied in a majority of animal implants. By using animal with caution.
models that characteristically show rapid and extensive Against this background, it seems logical that a prime
TAE, ‘‘spontaneous’’ endothelialization was studied on paradigm for future vascular graft studies should be the
grafts that would never have endothelialized in humans. separation of TAE from other sources of endothelializa-
Moreover, by choosing particularly short graft lengths and tion. A prerequisite for this is a better insight into TAE.
long implantation periods, it does not come as a surprise In principle, there are four factors influencing
that the prostheses were frequently fully endothelialized at TAE: species, senescence, anatomical dimensions and graft
 ARTICLE IN PRESS
P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
surfaces. Amongst these four main determinants, species is
certainly most influential. The same ingrowth distance, for
instance, for which humans need 56 weeks [6] is reached
after approximately 4 weeks in young yellow baboons [41]
and 3.5 weeks in dogs [33,34,42–44]. The second most
influential determinant of TAE is whether an animal is
juvenile or senescent. Even in the non-human primate model
senescence leads to a distinctly slower rate of endothelial
outgrowth [45] than infancy [39]. Senescence also needs to be
considered in cross-species comparisons. Pigs [32], sheep
[46–49] and calves [50], for instance, are usually large
enough for vascular implants at young age but dog puppies
are too small. Therefore, dogs are usually senescent at
implantation [51] whereas porcine, ovine and bovine models
tend to be juvenile. To a certain degree, senescence may even
outweigh species differences. After several months, the
senescent dog model covers a comparable ingrowth distance
[33,34,42–44,52] to the juvenile yellow baboon [41,53]. To
further complicate the issue of interpreting TAE, anatomical
dimensions may make it challenging to compare results. It is
impossible, for example, to relate the transanastomotic
ingrowth distance of 2.2 mm after 6 weeks in the rat
[41,54–56] and 1–2 mm (2 weeks) in the rabbit model [57]
(ePTFE/30 mm internodal distance (IND)) to the 7.8–
13.6 mm seen on the same grafts in large animal models
[11]. Last, but not least, the type of prosthetic graft
implanted does seem to influence TAE. On low-porosity
ePTFE, for example, TAE is twice as pronounced
[33,34,42–44] as on Dacron [6,58]. Nevertheless, if one puts
this ‘augmentation’ of TAE by certain graft surfaces into
clinical perspective, the futility of TAE as a means of
achieving confluent graft endothelialization is highlighted
once again: even a doubling of TAE is a far cry from the
30-fold increase required under clinical conditions (Fig. 2).
3. Anastomotic intimal hyperplasia: clinical villain,
experimental void
Intimal hyperplasia occurs in vein grafts throughout
their length. In contrast, intimal hyperplasia is confined to
Fig. 2. Schematic comparison of graft lengths typically used for clinical
bypass grafts (60 cm, top) and experimental grafts in animals (in the
majority of studies o6 cm; bottom). In most animal models, transanas-
tomotic endothelial outgrowth (red) leads to complete surface endothe-
lialization within weeks making it impossible to study transmural or
blood-born endothelialization. In clinically implanted bypass grafts, in
contrast, ingrowth stoppage permanently leaves over 90% of the graft
non-endothelialized (blue). Therefore, animal studies generally investigate
a mode of surface endothelialization that is of utter irrelevance to the
human situation.
5011
the peri-anastomotic region of prosthetic grafts in humans.
The simple reason for this lies in the fact that intimal
proliferation can only occur on the back of existing tissue
and in man, most of the blood surface of a synthetic
vascular graft remains uncovered by tissue. This also
explains the clinical failure mode of prosthetic grafts: it is
either midgraft thrombosis for a lack of endothelium or
intimal narrowing at the anastomosis, which may prompt
occlusion. Aetiology and pathogenesis of this often-termed
‘pannus’ tissue are multi-factorial, from surgical trauma to
lack of endothelium, from compliance mismatch to flow
conditions. Of these, flow conditions and shear stress seem
to be overriding forces. However, as much as Alexander
Clowes’ group [38,59] and others showed that variations in
flow [60,61] and shear stress [61] result in an adaptive
reduction or increase in the thickness of the subendothelial
tissue, almost all the studies of the past four decades used
models where the shear stress conditions did not in the least
resemble those in clinical bypass grafts (Fig. 3). In animal
models where distal arteriosclerotic narrowings are not
present, the main determinant of shear stress in vascular
grafts is the cross-sectional quotient (Qc) between prosthe-
sis and run-off vessel. In a typical clinical small-graft
situation, Qc is estimated to lie somewhere in the range of
0.25–0.35 [62]. From Qc40.5 onwards (cross-sectional area
of the graft less than twice that of the run-off artery), shear
forces seem to become inhibitory for the development of
intimal hyperplasia [62]. Typically, almost all the vascular
graft studies in the past chose graft diameters that were
largely size matched or even smaller than the artery and
thus did not undercut the critical cross-sectional ratio from
which point intimal hyperplasia develops. For example, the
estimated mean Qcs for interposition grafts ranged from
0.63 to 42.0 in dogs [14–17,23,42,63–66], 0.56 to 42.0 in
Fig. 3. Representation of anastomotic pannus formation in clinical (left)
and experimental grafts (right). Since anastomotic intimal hyperplasia
(AIH; red) is augmented under low flow conditions, the typical clinical
scenario of a graft (blue) larger than the artery (yellow) leads to flow
conditions that are conducive for AIH. In contrast, the overwhelming
majority of experimental grafts have a smaller luminal diameter than the
artery and thus experience increased flow. Therefore, experimental
conditions are largely ignorant of a phenomenon that is behind a
significant proportion of clinical graft occlusions.
 ARTICLE IN PRESS
5012 P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
rats [14–17,23,42,63–66], 1.65 in pigs [32] and 0.77 to 1.00
in baboons [45,67,68]. Assuming normal anatomy for each
species and weight group, the estimated average Qc for all
studies considered in this review (n ¼ 70) was as high as
1.5270.11(SEM) (Fig. 4). The resulting shear forces were
therefore approximately 1.8 times that of the artery (as
opposed to 1/8th for a Qc of 0.25), far above the threshold
level below which intimal thickening occurs [69] (Fig. 5).
Some of the studies even used aorto–aortic or aorto–iliac
end-to-end or end-to-side models, where an increased
inflow most likely produced even more pronounced
supra-normal flow conditions [30–34,36–40,70–75]. In
some studies, additional down-stream arterio–venous
fistulae were created to further increase the flow through
the graft [38]. The fact that even in these grafts relative
differences in flow velocity did have an impact on diffuse
Fig. 4. Relative vessel (yellow) and graft (blue) diameters used in clinical
applications and in experimental animal models. The cross-sectional
quotient (Qc) between run-off artery and graft is typically in the range of
Qc0.25–0.35 in clinical small-diameter grafts (left) as opposed by
Qc1.4–1.6 in the vast majority of experimentally implanted grafts (right).
Fig. 5. Schematic representation of longitudinal cross-sections of end-to-
end (top) and end-to-side (bottom) anastomoses showing shear force
differences resulting from the typical diameter mismatch between graft
(blue) and artery (yellow) under clinical circumstances (left) and in
experimental conditions (right).
intimal hyperplasia [38] only underlines the importance of
getting the flow-dynamics right. Considering the overall
design of the studies, however, it should not come as a
surprise that flow-restricting anastomotic intimal hyper-
plasia (AIH) hardly developed in any vascular graft study.
This does not mean that intimal tissue did not reach across
the anastomosis or that other mechanisms triggering and
augmenting AIH were not at work. One such mechanism is
an almost template function of the initial fibrinous inner
capsule. Two weeks after implantation for instance, the
thickness of the surface thrombus on ePTFE grafts
measured about 18 to 14 of that observed on Dacron grafts
[33,34,76]. Similarly, the thickness of the ‘pannus’ tissue
replacing this thrombus in the anastomotic region not only
measured 18 to 14 on ePTFE grafts compared with Dacron
prostheses [34,70] but also actually had the same thickness
as the pre-existing fibrin layer [34,60,77]. Another deter-
mining factor for the thickness of anastomotic pannus
tissue seems to be the existence or the lack of an
endothelium. Similar to post-PTCA- and stent-injuries,
the absence of a mature and confluent endothelium at the
time of subintimal tissue development seems to augment an
overshooting hyperplastic proliferation of poorly differ-
entiated connective tissue. As early as 1975, Mansfield et al.
[50] described the prevention of undifferentiated pannus
ingrowth in calves through pre-endothelialization of the
grafts. In the mid-1980s, clinical in vitro endothelialization
showed a similar phenomenon [45].
4. Graft-structure-related limitations to tissue regeneration
Initial attempts to replace arteries with solid tubes of
synthetic material [78] soon made it clear that porosity is a
prerequisite for graft patency [79,80]. Therefore, ranking
structural porosity above material properties emerged as
the early creed of synthetic vascular graft research. To date,
this dogma is still unchallenged. Efforts to determine
porosity requirements for graft healing, however, are
complicated by material specific characteristics, structural
uniqueness and the patho-physiological microenvironment.
The complex three-dimensional structure of contemporary
grafts results in a wide range of voids between the synthetic
components, hardly allowing a precise definition of
ingrowth spaces. For manufacturers, a way out of this
dilemma was the introduction of water permeability as a
means of defining porosity. For example, materials such as
Dacron were not only defined on the basis of being
knitted or woven, but also further characterized as being
of high porosity (1500–4000 ml/cm2/min) or low porosity
(200–1000 ml/cm2/min) [81]. The same principle applied to
ePTFE on the basis of distances between the nodes of the
expanded material. However, although these rough para-
meters served as useful means to characterize the materials,
they were only of limited value in determining structural
limitations of healing. Researchers were long puzzled as to
why interstitial graft spaces of particular prostheses were
sufficient to allow macrophages and other inflammatory
 ARTICLE IN PRESS
P. Zilla et al. / Biomaterials 28 (2007) 5009–5027 5013

cells to infiltrate, whereas connective tissue cells of similar

dimensions were often conspicuously absent [34,82–84].
Today, we understand that certain of these connective
tissue cells, like smooth muscle cells, may follow ingrowing
endothelial cells [85,86]. Therefore, one important deter-
mining dimension for complete graft healing is that which
permits capillaries to sprout into the interstices of the
prosthesis and still leave some space for accompanying
cells. Since the average diameter of a capillary is 10 mm
[87–89], it would seem as if the minimum area required for
ingrowth of capillaries is at least 20–80 mm2. The diameter
of a functional arteriole, i.e. an endothelium and at least
one layer of smooth muscle cells, is approximately
23.1713.1 mm [11]. However, with future regenerative
vascular prostheses in mind, space orientation needs to
be additionally considered beyond mere dimensional
concerns. As the different orientation of smooth muscle Fig. 6. Scanning electron micrograph of a cross-section through a high-
cells in arteries and veins shows, biomechanical forces porosity ePTFE vascular graft with 150 mm internodal distance (IND). It
determine smooth muscle orientation. The only geometric is obvious that internodal and interfibrillar distances determine ingrowth
spaces. The non-elastic fibrils allow only a limited degree of stretching,
structure, which allows free orientation in response to a making disproportional increases in IND necessary to achieve moderate
particular biomechanical environment, is the open dode- increases in ingrowth dimensions. Furthermore, available ingrowth
cahedron of a foam [90]. channels diminish in size with increasing penetration depth due to the
random obstruction by adjacent fibrils.
4.1. ePTFE grafts

ePTFE grafts consist of circumferentially aligned, thin

and irregular-shaped solid membranes (the so-called
‘‘nodes’’) and a dense meshwork of fine fibrils stretching
between the nodes. Although the porosity of ePTFE grafts
is traditionally defined on the basis of the distance between
the nodes (IND), the actual dimension of voids within the
prosthetic wall is defined by the spaces between the fibrils.
Thus, the available ingrowth spaces in ePTFE grafts are
manifold smaller than the IND suggests. Although the
process of expanding Teflon material to the typical node-
and-fibre structure of ePTFE grafts allows for a wide range
of INDs, the resulting changes in ingrowth spaces are
moderate. With well-defined anchor points and a rigid
material like Teflon, the dispensability of the fibrils is
limited, thus making disproportional increases in IND
necessary to achieve moderate increases of ingrowth
dimensions of a few micrometers (Fig. 6). Apart from
differences in fibril length between equally defined grafts of Fig. 7. Demonstration of the dependence of ingrowing capillaries on
available ingrowth spaces. Dual-layer ePTFE graft with high-porosity
different diameters (e.g. 24.3 versus 20.7 mm for 6 and outer two thirds and low-porosity inner third. (Femo–femoral interposi-
10 mm grafts), there are distinct differences in the IND of tion graft; 4 mm ID; 4 weeks; senescent Chacma Baboon; CD 31).
‘30 mm’ ePTFE grafts between different manufacturers (e.g.
17.875.6 and 23.876.2 mm for two market leaders,
respectively) [11]. Furthermore, theoretically available, IND, in contrast, transmural capillarization would be
continual transmural ingrowth spaces in ‘low-porosity’ permitted by 100% of the transmural spaces and even some
(30 mm IND) grafts diminish in size over the first ten layers arterioles would be able to reach the blood surface [11].
of fibrils out of an overall number of approximately 2000 Even before these calculations were available, a similar cut-
layers/wall thickness [11]. This is a result of the random off point of porosity emerged empirically over the years
obstruction of interfibrillar spaces of one layer by fibrils of below which tissue ingrowth is impermissible (Fig. 7). In
the next. The available channel size would therefore not spite of manufacturer-related differences between contem-
accommodate capillaries from as early as layer 200 porary ePTFE grafts, this cut-off point was identified to lie
onwards, the equivalent of the outer 1/10th of wall within a relatively narrow range somewhere between 30
thickness. In ‘high- porosity’ ePTFE grafts of 60 mm and 45 mm of IND. Therefore, it became customary to refer
 ARTICLE IN PRESS
5014 P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
to grafts with an IND of less than or equal to 30 mm as low
porosity ePTFE grafts and to those of 45 mm or more as
high- porosity ePTFE grafts. This distinction is of course
invalid in grafts where the manufacturer adds an impene-
trable external wrap.
4.2. Dacron grafts
In contrast to the two distinctly different structures of
ePTFE grafts (nodes and fibrils), Dacron prostheses consist
of one basic element, namely polyester fibres. However,
since the fibres are mostly bundled into multifilament yarns
that are either woven or knitted into a fabric, there are in
effect again two different structures determining the shape
and the dimensions of voids. Generally speaking, weaving
results in narrow spaces for tissue ingrowth whereas
knitting leads to looser patterns containing larger voids.
Woven material is often preferred by surgeons, as it is
known for its high bursting strengths, low permeability to
liquids and minimal tendency to deform under stress. The
almost complete impermeability of woven Dacron is
opposed by a range of limited ingrowth spaces in knitted
Dacron. Still, in the lower range of water permeability
capillaries are also space-restricted to the outer 23 of the wall
and spaces are generally too narrow for arterioles [11].
4.3. PU grafts
The enthusiasm with which PUs were initially adopted as
vascular graft materials was based on two main advantages
over polyester and Teflon: elasticity and ease of handling.
Porosity was an integral part of PU grafts from the
beginning, but was mainly understood as a principle of
patency rather than tissue ingrowth. As a result of this,
defined inter-connectivity of the pores was largely absent in
both fibrillar and foamy PU grafts. Although the pore size
was somehow changeable in either of the two structures it
was not defined around ingrowth dimensions. In fibrillar
designs porosity was varied through fibre thickness, wind-
ing angle, regularity of the deposition and extent of
coagulation of fibres before solidification. With fibre
thickness varying between 1 and 30 mm [72,91], for
instance, pore sizes could be achieved between 10 and
60 mm [72,91–93]. However, regardless of whether fibrillar
grafts were produced through weaving [94,95], knitting
[94], electrostatic spinning [91] or winding processes [72,73]
the presence of communicating transmural spaces permis-
sible for blood vessel ingrowth was coincidental rather than
intentional.
Similar to fibrillar grafts, foamed prostheses mostly
lacked spaces that reproducibly allowed blood vessels to
reach through the entire wall thickness. Phase inversion
alone, for example, led to micropores that were generally
poorly interconnected and smaller than 15 mm [92,96–98]
therefore not allowing more than a shallow, superficial
penetration by capillaries. Differences between various
phase/temperature inversion techniques [99] allowed some
degree of variability in this microporosity. Other techni-
ques to achieve porosity include foam flotation [100] and
dip-coating [96] (laminar structures with ill-defined and
poorly interconnected voids), gas expansion [101] (closed
porosity), and laser perforation [102] (radial channels
through a solid tube rather than three-dimensional pores).
Once the focus shifted from material choice, surface
structure and ‘water porosity’ to transmural tissue regen-
eration, continuity in void dimensions became a desirable
goal. Two methods that did result in open-porosity foam
structures capable of allowing tissue ingrowth were
replamineform [103] (where porous sea-urchin spikes were
used as a template to produce pores ranging from 18 to
180 mm), and reticulation (opening of closed-pored foams
to yield open porosity) [104]. The first method is limited by
issues pertaining to supply, machining and the porosity of
naturally occurring structures, while the second involved
the rolling of reticulated sheets into tubular structures;
both methods were not further pursued. The most obvious
practical way of achieving ingrowth-permissive porosity
was by superimposing macropores onto the microporous
structure obtained by phase inversion. The addition and
subsequent extraction of porogens to the phase inversion
technique was suggested [105–107] leading to pores of
30–75 [105], 40 [106,108] and 200 mm [109] as well as pore
size gradients (10–100 mm) [110]. However, the use of salts
and other irregular porogens, together with the pre-mixing
of the porogen and the polymer solution, generally resulted
in irregular-shaped pores, limiting connectivity between
pores while trapping residual porogens in the structure
[111]. In order to overcome these limitations, a variation of
the phase inversion/porogen extraction method has been
developed, in which a PU solution is infiltrated into a
tightly prepacked column of spherical porogens. Subse-
quent precipitation and extraction yields a well-defined,
open porosity comprising uniformly sized dodecahedrons
with even and equally well-defined interconnections
[90,112]. Such an ‘open-cell’ foam represents the ideal
ingrowth scaffold if alignment along stress and strain needs
[4] or nutrient and oxygen gradients is desired because it is
the only structure which allows cell and tissue forma-
tions to freely orientate themselves in any given direction
(Fig. 8).
5. Midgraft tissue response: no healing, no regeneration
Even after prolonged periods of implantation, a persis-
tent foreign body response dominates the interstices of
permanent (non-resorbable) synthetic arterial prostheses.
Simultaneously, thrombotic appositions build up on the
luminal surface. Thus, healing—defined as the end point of
a pacified repair process—does not occur. Moreover, since
not even traces of vascular tissue are being formed in the
interstices or on the surface of these grafts, regeneration
remains permanently absent, too. Yet, in spite of many
millions of clinical implants during the course of the past
decades the degree of misconception about the presumed
 ARTICLE IN PRESS
P. Zilla et al. / Biomaterials 28 (2007) 5009–5027 5015

Fig. 8. Comparitive schematic representation of available ingrowth spaces in ePTFE, Dacron and well-defined PU scaffolds. While both ePTFE and
Dacron grafts have structures that predetermine ingrowth directions and pose some degree of spatial limitation, an open dodecahedron structure allows
cells to freely orientate themselves in any given direction (Insert: single cell of polyurethane graft with well-defined, spherical 157 mm spaces and equally
well-defined interconnecting windows of 76 mm). In a graft concept that includes compliance, the cell orientation follows biomechanical strain
requirements. This principle is not only realized in the different alignment-angle of smooth muscle cells (SMCs) in arteries and veins, but was also
demonstrated in tissue engineering constructs where SMC were embedded in a gel and where strain resulted in the artery-like orientation of the cells (upper
right corner) [4].

degree of ‘healing’ of vascular grafts amongst surgeons is
often surprising.
5.1. Low-porosity ePTFE grafts (o30 mm IND)
There is hardly any difference in the midgraft response of
low-porosity ePTFE grafts between most recipient models.
At the blood surface, a thin, 10–20 mm thick fibrin layer
develops during the initial 2 weeks of implantation [33,34].
Over the subsequent weeks the thickness of this layer stays
constant [113] but its composition becomes more diverse
ranging from loose [45,113] or compacted acellular fibrin
[114] to platelet carpets with interspersed granulocytes
[33,45]. Even after prolonged observation periods, midgraft
regions lack all forms of cellular coverage in low-porosity
ePTFE [5,6,36,57].
Similarly, there is no transmural tissue ingrowth. During
the initial 2 weeks of implantation, most of the interstices
of the prosthetic wall are devoid of any recognizable
cellular material [33]. Only a few macrophages begin to
migrate into the micropores—almost exclusively from
outside [54]. After 3 weeks we find a typical triple layered
picture with macrophages invading from the blood—and
to a much larger extent—the adventitial surface into an
otherwise almost acellular central fibrin matrix. Eventually,
after 6 weeks of implantation, connective tissue cells begin
to sporadically invade the interstices of non-wrapped
prostheses from outside [54,82,114–116]. Even after 6
months, however, this connective tissue ingrowth has
hardly increased at all, remaining mostly limited to the
outer part of the graft wall [41,63,64,83,84,117] with only
scanty capillaries [117,118]. In wrapped grafts, the ePTFE
membrane reinforcing the prosthesis acts as an effective
barrier against cellular infiltration. These grafts show an
almost acellular fibrin matrix extending from the centre of
the prosthetic wall all the way to the outside surface while
eliciting a much stronger foreign body giant cell reaction
on the outside than unwrapped grafts.
5.2. High-porosity ePTFE grafts (445 mm IND)
In contrast to low-porosity ePTFE grafts, there is a
distinct difference between the healing responses of various
animal models in high-porosity ePTFE grafts particularly
with regards to the timing of events.
Initially, a thin fibrin or amorphous protein layer
covering a significant proportion of the graft coexists next
to exposed areas, where the nodal structure of the ePTFE
remains visible. Over time, the fibrin attracts white blood
cells and platelets [119] on its surface. In the depth, it
continues to be widely acellular except for a mild rim of
macrophages lining the ePTFE surface. Underneath these
macrophages, the fibrin filling the interstices of the ePTFE
is initially also acellular but over time gets progressively
 ARTICLE IN PRESS
5016 P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
popularized with macrophages and polymorphnuclear
leukocytes [120,121] particularly in the outer third of the
graft wall . In more senescent animal models like the adult
dog or chacma baboon this stage persists into the 4th to 6th
week of implantation [119] as opposed to less than 2 weeks
in juvenile models such as the yellow baboon [121]. At that
time, sprouting capillaries begin to reach into the outer one
third to one half of the graft wall [119,120]. Although
fibroblasts from the peri-graft region soon follow the
capillaries [55], the majority of the interstices remain
dominated by macrophages [121], while foreign body giant
cells (FBGCs) are conspicuously absent. Distinctly differ-
ent from the widely dilated, sinus-like endothelial cell
formations found in Dacron grafts (see below), however,
the microvessels soon resemble mature arterioles. They are
small in diameter (23.06713.11 mm) [11] and often
accompanied by at least one layer of smooth muscle cells
(Fig. 9). In contrast to senescent animal models where it
may take months for these microvessels to reach the blood
Fig. 9. High-porosity (90 mm IND) ePTFE grafts implanted into the
femoral arteries of senescent Chacma baboons (4 weeks). A significant
number of factor-VIII-positive vessels (A) is supported by an accompany-
ing layer of actin-positive SMC (B), indicating that continual ingrowth
spaces of more than 23 mm are available.
surface, this is accomplished in the 2nd week of implanta-
tion in the juvenile yellow baboon. In the latter, patches of
endothelial cells [68] and capillary orifices—approximately
100–500 mm apart [59]—appear on the blood surface and
confluence is often reached before the 3rd or 4th week of
implantation [59,68,70,121–123]. Soon thereafter a layer of
actin positive cells develops underneath the endothelium
[59,68,70,121–123]. These cells—which exhibit the ultra-
structural characteristics of arterial smooth muscle cells
[68]—only begin to appear and proliferate after the
endothelium has formed, corresponding with the sequence
of events in angiogenesis, where endothelial-derived PDGF
chemotactically attracts pericytes to follow and proliferate
[124]. The actual discovery of PDGF–A mRNA in the
endothelium of high-porosity ePTFE grafts [121] increases
the likelihood of this scenario. It would also explain why
the density of actin-positive cells is highest close to the
endothelium [121]. PDGF is not only one of the strongest
growth factors for smooth muscle cells but also is known to
be secreted by endothelial cells into the abluminal basal
compartment [125]. It’s strong presence in the endothelium
covering high-porosity ePTFE grafts may also explain the
5–100 times higher proliferative activity of SMCs and
10–100 times higher mitotic activity of endothelial cells
[126] in grafts than in normal arteries. Although it remains
open whether this is an adaptive response or a primary
phenomenon, the manifold thicker neo-intima found on
high-porosity ePTFE grafts exposed to low flow conditions
as compared to high flow conditions [127] points at least
partly to an adaptive process.
5.3. Woven Dacron grafts
Immediately after implantation, a thin layer of fibrin,
erythrocytes, white blood cells and platelets is deposited on
the blood surface of the prosthesis [10,58]. During the first
few hours to days this thrombus layer slowly starts
thickening until it reaches a stable equilibrium of com-
pacted fibrin [128]. In dogs this equilibrium stage is reached
within 6 months [128] whereas in humans it is only
observed after 18 months. In the depth of the graft, fibrin
also instantly fills the narrow interstices of the graft wall
and the space between the graft and the surrounding tissue
[129]. At the interface between this fibrous capsule and the
Dacron yarns, a variable degree of FBGCs begins to build
up [129]. Subsequently, a few capillaries and fibroblasts
start growing into the tight interstitial spaces of the woven
Dacron [58,129]. This process is quite variable in its extent
and time course. In dogs it can be seen as early as 2–3
weeks after implantation [129], whereas in humans it may
be observed after 5 months [58] or not at all [10,130,131].
Exceptionally, the tissue reaches the inner fibrin capsule
and begins to organize the basis of the fibrin capsule in a
few scanty areas of the graft [130]. In the rare event that
such tissue forms, it was found to occur after 4–8 weeks in
dogs [58,128] and after 18 months in humans [130]. In even
fewer cases, these tissue areas replace the inner fibrin
 ARTICLE IN PRESS
P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
capsule in its entire thickness and form a pseudo-neointima
[58,130,131]. Both the pseudo-neointima and the com-
pacted fibrin usually reach an equilibrium at a thickness of
approximately 400–500 mm [130]. In humans these rare and
localized neointimal spots extend over less than 1 mm and
are collagen rich, with increasing cellularity towards the
graft surface [10]. If these islands of connective tissue are
rare, areas that are covered by an endothelium are even
rarer. Usually no endothelial cells are found at all within
the first 1–2 years [58,130,131]. After prolonged implanta-
tion periods of up to 11 years—which are naturally
restricted to humans—small islands of endothelial cells
may occasionally appear [130,131]. These endothelial
islands always rest on the compacted fibrin rather than
the scanty islands of connective tissue.
In summary, tissue ingrowth is very limited in woven
Dacron grafts, seldom penetrating the entire wall thickness.
Most importantly, even if it extends to the inner capsule, it
hardly manages to break through the compacted fibrin to
reach the blood surface [132]—even after decades of
implantation.
5.4. Knitted Dacron grafts
In knitted Dacron grafts, the initial surface meshwork of
fibrin, white blood cells, erythrocytes and platelets [7]
increases to a thickness of 100–120 mm in the course of the
1st week of implantation [76]. During the subsequent 5
weeks, the inner fibrin capsule has not reached an
equilibrium yet holding a wide range of thicknesses:
100–500 mm in humans [76], 30-210 mm in dogs [34,66,77,
133–135] and 290-300 mm in the yellow baboon [67,70].
After a further insignificant increase thereafter [66,76,133],
the thickness of the internal capsule eventually stabilizes
before the 6th month of implantation [66,133]. Although a
superficial screening of the literature suggests that it is only
a matter of time until this entire layer of compacted fibrin
gets organized by ingrowing tissue, deeper analysis makes
it likely that this may well be a misinterpretation based on
the confusion of midgraft healing with anastomotic
ingrowth. It is true that the vast majority of studies describe
a mature endothelium and a well-developed intimal smooth
muscle layer, resting directly on the graft surface [136–138].
However, the graft length and implantation period in these
studies make it almost certain that this mature ‘‘neointi-
mal’’ tissue represents nothing but extended anastomotic
ingrowth [34,70,82,122,137–139]. This suspicion is con-
firmed by descriptions of true midgraft regions in particu-
larly long grafts in which the inner fibrin capsule appears to
be both impenetrable for the ingrowing interstitial graft
tissue [34,70,82,122,137–139] and non-endothelialized
[76,140,141]. This lack of surface healing occurs
[6,7,76,131,132,140,142] although transmural tissue reaches
the compacted fibrin of the blood surface within 3–4 weeks
in the calf [76], the yellow baboon [70] and the dog [34,143]
and 3–6 months in humans [7,76]. Very occasionally,
though, even complete endothelialization of very long
5017
prostheses, [65,126] comparable to that in highly porous
ePTFE grafts, is seen. However, in contrast to high-
porosity ePTFE grafts it still remains unclear whether this
endothelium is derived from transmural tissue ingrowth,
facilitated transanastomotic ingrowth or fall-out healing
[144,145]. The latter is a phenomenon predominantly
observed in Dacron grafts where—apparently independent
from transmural tissue ingrowth and with some delay—
endothelial islands emerge on the surface of the compacted
fibrin. These islands are well-separated from the transmural
tissue ingrowth by a distinct layer of compacted fibrin
[6,7,76,131]. They either rest directly on the fibrin [6,131,
140] or on a few layers of isolated actin-positive cells that
have no other tissue connection [131,142]. Such endothelial
islands can be observed in dogs after 2–6 months [140] and
in humans perhaps in every 4th graft after 1–11 years
[6,131,142]. Nevertheless, although sporadic endotheliali-
zation occasionally occurs in limited areas of clinically
implanted grafts, the majority of the surface remains
covered by fibrin only.
If one focuses on the transmural tissue ingrowth itself, it
also begins with the infiltration of the surrounding outer
fibrin matrix by granulation tissue [7,34]. Initially it is
primarily macrophages that begin to invade the interstitial
fibrin [115], popularizing the entire depth of the graft at the
end of the 1st month of implantation [76]. Early giant cell
formation surrounding the yarns commences after 7 days.
During the following months a distinct foreign body
reaction occurs predominantly in the outer half of the
graft wall [66,133]. Between this outer part of the prosthetic
wall that is packed with FBGC, and the surrounding
connective tissue capsule with its longitudinally aligned
collagen bundles [76,131,146], hugely dilated capillary
sinuses are usually present during the early weeks of
implantation. These sinuses typically consist of a single
endothelial cell layer that contains no smooth muscle cells,
but does contain macrophages and FBGC that tightly
snuggle against the basal lamina from outside. The few
capillaries that sometimes extend through the entire graft
wall and reach the inner capsule also remain free of smooth
muscle cells.
In summary, the porosity of knitted and to some extent
even woven Dacron prostheses eventually allows a certain
degree of tissue ingrowth from outside. At the same time,
however, it seems again that the compacted inner fibrin
capsule on the blood surface of Dacron grafts represents an
ingrowth barrier for transmural connective tissue regard-
less of whether the grafts are knitted or woven. Although
this barrier may eventually be overcome by ingrowth of
undifferentiated fibrous tissue, capillaries remain uncon-
nected to the often poorly endothelialized blood surface
[132]. Moreover, it is also interesting that the fibrin
coverage on Dacron grafts seems to have a higher
propensity to capture blood borne cells than that on
ePTFE grafts [145]. In this context it is noteworthy that the
resulting isolated endothelial islands often rest on equally
isolated smooth muscle layers [6,131]. However, in spite of
 ARTICLE IN PRESS
5018 P. Zilla et al. / Biomaterials 28 (2007) 5009–5027

the scientific excitement arising from this observation, the
rare and late occurrence of the phenomenon make it rather
irrelevant in the healing process of contemporary Dacron
grafts.
5.5. Fibrillar PU grafts
With very rare exceptions fibrillar PU grafts are largely
impenetrable for transmural tissue ingrowth [91,147,148].
Only particularly large inter-fibrillar spaces allow complete
fibro-connective tissue penetration throughout the wall
thickness [49,73,93]. At the same time, events on the blood
surface resemble more those on woven Dacron than on
ePTFE. Particularly, anastomotic pannus tissue tends to be
thicker on PU [149] than on expanded Teflon with capillary
orifices opening onto the anastomotic endothelium
[148,150]. From day one the midgraft region is continu-
ously covered by a fibrin layer [151] hardly diminishing in
thickness during the course of the 1st year [71,72,91,
147,148,150–152]. The fibrin layer itself is not as mighty as
on knitted Dacron, some times giving a ‘ghost impression’
of the underlying prosthetic fibres. The outside capsule and
scar formation is usually thin and dense [147,148,150] with
occasional FBGC wrapping around PU fibres [73]. At 2
weeks the FBGC become more prevalent increasing until
week 6 [153]. The inflammatory reaction does not further
increase between week 6 and 12 [153].
determine that interdependence of inflammation and
vascularization was less pronounced than assumed, [90]
as well as a reproducible chronology of the tissue
responses. Most importantly, the inflammatory response
and the FBGC index could be reduced by 56% and 21%,
respectively, when the pore size was increased from 66 to
157 mm, while vascularization did not diminish. The fast
and massive early vessel penetration of this well-defined
graft structure [112] from outside (Fig. 10) highlighted the
basic dilemma of transmural endothelialization: while
blood vessels rush into the graft wall, their crossing of a
distance of as short as 0.6 mm towards the blood surface
was distinctly protracted in spite of widely open traverse-
spaces. The build-up of an adverse microenvironment was
best studied in these foam grafts when femorally implanted
in senescent Chacma baboons. Immediately upon implan-
tation, the entire graft wall was filled with a loose fibrin
matrix allowing sprouting capillaries from outside to reach
as deep as half into the graft wall after 7 days. After 2
weeks, transmural vessel ingrowth had hardly progressed

5.6. Foamy PU grafts

In microporous PU foams with less than 15 mm pore size,

there is relatively little ingrowth, even over an extended
period of time [108,109,154–156]. Below this cut-off point,
only plasma-like insudation with some erythrocytes and a
few clusters of white blood cells are found in the depth of
the wall [47,155]. Macrophages, giant cells and a few
fibroblasts may penetrate a short distance into the implant
[156] if the pore size is borderline. The foreign body giant
cell response along the outer graft interface tends to get
augmented until week 6–9 [153]. From an interconnecting
pore size of 25–40 mm onwards macrophages, giant cells,
fibroblasts and capillaries rapidly populate the wall during
the initial 3–4 weeks [153,156,157]. The degree of ingrowth
is initially more pronounced in larger porosity, but
equalizes for all porosities after 3 months [156]. Compar-
able to fibrillar PU grafts, the blood surface also shows a
significantly more pronounced fibrin layer than on ePTFE
[47,158]. Similarly, anastomotic pannus tissue is strongly
developed with capillary openings to the blood surface.
This anastomotic intimal hyperplasia was distinctly miti-
gated when compliant PU grafts were compared with non-
compliant ones [159]. Due to a fast build-up of collagen
within the pores of cell-populated PU grafts [74,157], Fig. 10. Corrosion casts of PU grafts showing vessel ingrowth into well-
defined pores. The vigorous vessel-penetration of the graft from outside
however, compliance decreases by 50% over the initial 6
demonstrates how attracted capillaries initially are by the scaffold and
weeks [159]. Going beyond ingrowth permissible porosity how increasingly adverse the micro-environment of a vascular graft must
by introducing well-defined, equally sized spherical pores become to make the crossing of the remaining 0.6 mm distance to the
and large, equally well defined interconnectivity, we could blood surface such a protracted and challenging affair.
 ARTICLE IN PRESS
P. Zilla et al. / Biomaterials 28 (2007) 5009–5027 5019

further towards the blood surface, which showed early content of fibrinogen. Since TGF-b is also liberated from
signs of a dense fibrin layer. Another week later, the a-granules and its incorporation into fibrin matrices is
capillaries had reached the inner third of the wall, but the proportional, a platelet-augmented, fibrinogen-rich fibrin
fibrin layer had become even denser and ‘squeezed’ into matrix is rich in TGF-b. High concentrations of TGF-b are
the pores. While it took 4–6 weeks for a few capillaries to known to be inhibitory for angiogenesis [161]. Since it is
reach the surface, open up through a microorifice and known that the fibrin structure influences cell growth, the
begin to proliferate on top of a very dense fibrinous surface main question is whether high concentrations of fibrinogen
layer, the entire distance of 0.6 mm was rapidly crossed if and other a-granule and dense granule release products
grafts occluded in the first few days. such as calcium affect the structure and/or composition of
fibrin in a way that it becomes hostile to tissue ingrowth.
6. Biological events: adversity rather than facilitation Although none of the investigators has yet analysed the
composition of fibrin on vascular grafts, there is evidence
The complexity of an unabated chronic foreign body that the slow growing fibrin layer on blood surfaces does
reaction at the blood–tissue interface is certainly beyond indeed differ from rapidly generated fibrin found in wound
the scope of a short review. Yet, the two seemingly trivial injuries or acute vessel occlusion (Fig. 11). It is recognized
main components of the tissue incorporation response— that an increase in fibrinogen concentration results in both
fibrin deposition and macrophage infiltration—are suffi-
ciently contentious to offer themselves as potential keys to
understanding main mechanisms behind the mitigated
tissue incorporation of prosthetic vascular grafts. Almost
from the time of implantation, fibrin becomes an integral
part of prosthetic vascular grafts both in the interstices and
on the blood surface. Fibrin is known to be an ideal
attachment and ingrowth matrix for endothelial cells. Yet,
circulating endothelial and progenitor cells hardly populate
the blood surface of grafts, making ‘fall out healing’ a late
and very scarce event. Similarly, transmurally ingrowing
capillaries hardly ever manage to fully cross a stretch of less
than 1 mm of a fibrin-containing graft wall, even if
ingrowth spaces are permissible and grafts were implanted
for years. Similarly, macrophages—which persist in their
presence from the earliest time of implantation—have been
the subject of much speculation with the spectrum ranging
from being quintessential for vascularization to being the
main culprits behind ingrowth inhibition.

6.1. The potential role of fibrin

The phenomenon of a seemingly impenetrable inner

fibrin capsule has been reported for more than 20 years in
porous Dacron grafts [6,7,51,131,132,140,142]. Likewise,
the ‘rubbery’ inner fibrin layer in aortic aneurysms [160] is
equally cell-poor to acellular, as are intra-atrial clots. It
seems reasonable to consider the possibility that ingrowth
inhibition may be linked to the particular nature of fibrin
forming on a blood surface of synthetic implants. In
contrast to fibrin plugs forming in wound spaces or Fig. 11. Schematic presentation of the hypothetical build-up of an adverse
occluding arteries, the continual apposition of fibrin on ingrowth environment over time: The fibrin clot initially filling the
interstices of the graft and the surgical wound on the adventitial side of the
the blood surface of vascular grafts occurs in an environ-
graft is formed on the basis of the limited fibrinogen supply present at the
ment of undiminishing supply. Ongoing platelet activation time of implantation and the surgical trauma. Typically, lower fibrinogen
provides high concentrations of a-granule products like concentrations lead to thicker and fewer fibrin strands that strongly
platelet factor 3, fibrinogen and von Willebrand factor as stimulate angiogenesis and cell infiltration. In contrast, the continual
well as dense granule contents such as calcium. At the same replenishment of clotting factors and calcium from the blood surface
create conditions where thinner and much denser fibres, which are
time, graft surfaces are continually rinsed by blood with its
inhibitory for vessel ingrowth, are being formed, Furthermore, continual
undiminishing concentrations of fibrinogen and other platelet replenishment and degranulation makes this fibrin rich in TGF-b,
clotting factors. Thus, the fibrin generated on the inner which at high release-concentrations is additionally inhibitory to capillary
surface of vascular prostheses has a particularly high formation.
 ARTICLE IN PRESS
5020 P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
a decrease in fibre bundle thickness and an increase in fibre
bundle density [162]. While a fibrin matrix strongly
stimulates angiogenesis at lower fibre density [163], it
inhibits capillary formation at higher fibre density.
A similar inverse correlation between cell growth and fibre
density was found on an ultrastructural level [162]. The
higher the density of fibrin fibres and the lower the average
thickness of these fibre bundles was, the lower was its
penetrability for cells [162]. The better healing of graft
scaffolds in subcutaneous than vascular position seems to
support this explanation.
6.2. Potential role of macrophages
In attempting to answer the often recurring question of
whether macrophages are friends or foes in prosthetic graft
healing, it seems as if they are both: friends at the beginning
and foes at the end. Their active secretion of cytokines acts
as a key chemotactic and mitogenic factor for capillary and
connective tissue ingrowth in the early phases of wound
healing. Their continual presence, however, increasingly
derails the delicate chronology of cytokines required for
the successful accomplishment of healing. During the
initial infiltration of the fresh fibrin matrix, macrophage
migration is facilitated by a low level of basic uPA activity.
This is sufficient to liberate and activate platelet-derived
TGF-b from the fibrin [164]. Since TGF-b upregulates the
expression of integrins on the surface of the blood
monocytes thereby promoting adhesion and migration
[165], it acts as a strong chemotactic agent to recruit
macrophages to the inflammatory site [165]. Moreover, it
also upregulates the uPA activity of macrophages [166] to
further facilitate migration. Macrophages are also capable
of binding and internalizing fibrin through the Mac-1
receptor, which augments fibrinolysis by a plasmin-
independent mechanism [167]. The degradation products
of fibrinolysis in return also act as chemoattractants
recruiting still more macrophages into the implant site
[168]. In addition, routing of monocytes to inflammatory
sites also involves the superfamily of chemoattractant
cytokines (chemokines ) and their receptors [169]. Specifi-
cally the C–C chemokine MCP-1 has been demonstrated to
be expressed by macrophages residing at the implant site as
early as 48 h post-implantation [170]. This sequence of
events illustrates how this initial phase of the so-called
healing is dominated by self-augmentation of macrophage
chemotaxis. Until macrophages actually contact the
material surface, the recruitment mode into the wound
fibrin of the prosthetic graft does not deviate from that
taking place in normal wound healing. As the macrophage
population migrates into the depth of the graft structure it
must eventually overstep the maximum distance a cell can
migrate away from a capillary before getting mildly
hypoxic. Macrophages exposed to hypoxia synthesize and
release increased amounts of PDGF as well as aFGF and
bFGF [171]. Medium conditioned by hypoxic macrophages
in culture is mitogenic for hypoxic endothelial cells through
the release of the FGFs. This suggests a paracrine
mechanism in which hypoxia stimulates macrophages to
release mitogenic factors for hypoxically preconditioned
endothelial cells thereby promoting angiogenesis. Addi-
tionally, TNF-a, which is also upregulated in macrophages
by hypoxia [172], is known to play a role in macrophage-
mediated angiogenesis presumably through the activation
and recruitment of further monocytes [173]. The role of
TNF-a in angiogenesis, however, is controversial and
seems to depend on the biological context [173]. These
effects are further potentiated by upregulation of MIP-1
during oxygen deprivation [172], a potent macrophage
chemoattractant. Therefore, macrophages residing in the
outer half of synthetic grafts together with VEGF-releasing
infiltrating neutrophils [174] are likely to continue to
contribute to a pro-angiogenic milieu at this stage. The
low concentration of TGF-b released with migration-
associated degradation of the adventitial ‘wound’ fibrin
should not be inhibitory for endothelial cell proliferation
[161]. However, once the dense inner fibrin capsule is
reached, released TGF-b concentrations would most likely
be inhibitory.
As the organization of the macrophages within the graft
wall changes and transmural endothelial cell ingrowth
begins, the situation where the outer wall of the graft is
dominated by single macrophages also slowly begins to
change by the increasing appearance of FBGCs [41,54,77].
These may persist for the life-time of the implant. Giant
cell formation depends on the material surface as well as
structure. It has been shown that relatively hydrophilic
material surfaces induce more fusion than relatively
hydrophobic ones [175–178]. Theoretical analyses show
that the time-dependent foreign body giant cell formation
is also influenced by the initial density of adherent
macrophages [179]. Therefore, any surface characteristic
that initially influences the number of adherent cells will
eventually also influence the extent of giant cell formation.
The most conspicuous deviation from this sequence is the
absence of FBGCs inside the wall structure of ePTFE
grafts compared to Dacron and PU, in spite of a distinct
presence of macrophages. The absence of giant cells on the
inner nodal surfaces which are subdivided by numerous
fibrils indicates that giant cell formation may be suppressed
by geometrical space limitations. However, given (i) the
in vitro evidence that fewer monocytes adhere to ePTFE,
(ii) that ePTFE is less inflammatory than Dacron and PU
[180–183], and (iii) that a more hydrophobic surface
decreases the propensity of single macrophages to fuse
[175,177–179], the inflammatory potential of the surface
may also play a role in giant cell formation insofar as a
more chemically inert, more hydrophobic ePTFE surface
reduces fusion. Not only do mononuclear cells gradually
fuse to form more giant cells in a material-dependent
manner but further complexity is added by the changing
nature of the mononuclear macrophage population over
time. Macrophages found in intimate contact with
synthetic materials in situ were ED1 and MHCII positive,
 ARTICLE IN PRESS
P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
whereas ED2 positive macrophages were observed lying
behind the ED1 positive cells [184]. This indicates a pattern
of distribution of not only mono versus multinuclear
macrophages, but also immature (ED1 positive) versus
mature (ED2 positive) and activated (MHCII positive)
versus non-activated macrophages within the implant site.
Theoretical analysis shows that up to 5 weeks of
implantation, FBGCs are in fact formed from the fusion
of a small percentage of the adherent macrophages which
are already present on the third day of implantation
[175,179]. This raises the issue of the fusogenic potential of
different subsets of cells. Expression of cell surface
receptors may be a prerequisite. For example, inhibitors
of mannose receptor activity have been shown to inhibit
IL-4-induced fusion in vitro suggesting that this receptor
may play a role in giant cell formation [185].
If it is difficult to decide whether macrophages are
beneficial for graft healing or not, it is even more difficult
to decide whether the presence of FBGCs represents a first
step towards pacification or a particularly detrimental
variant of chronic inflammation. One indication for the
FBGC’s role as a mediator of pacification of inflammation
arises from the fact that their fusion can be induced by IL-4
[186,187]. IL-4 is secreted by Th2 lymphocytes and is
known to downregulate several macrophage inflammatory
functions [188]. In this context it has been proposed that
giant cells may represent an attempt to ‘‘mop-up’’ the
deleterious effect of a persistent chronic foreign body
reaction. In contrast, a detrimental role for the giant cell is
suggested by the fact that material surface cracking occurs
directly beneath adherent FBGCs, implicating the secre-
tory products of these cells in the degradation of
biosynthetics [189]. It is also peculiar that the accumulation
of giant cells in Dacron regularly coincides with the
development of huge convolutes of capillary sinuses in
the vicinity. These widely dilated and irregularly shaped
endothelial sacks are principally devoid of any second cell
layer. In situ hybridization of human explants, however,
has demonstrated that interstitial macrophages and
FBGCs are still secreting IL-1b even after numbers of
years [190]. Although IL-1b is known to stimulate both
smooth muscle cell proliferation and fibronectin deposition
in atherosclerotic plaque formation [191], it has been
suggested that this interleukin can also inhibit vascular
smooth muscle proliferation in an autocrine fashion
through nitric oxide upregulation [192]. The giant cells,
therefore, may be responsible for not only affecting
capillarization but also inhibiting smooth muscle cell
ingrowth through persistent IL-1b secretion.
In summary, the true character of the FBGC remains
elusive. On the one hand, their persistence at the implant
site, their destructive role in biodegradation as well as their
possible role in the mitigation of vessel formation seems to
label them as villains of incomplete graft healing. Yet, the
evidence supporting a role for a typically anti-inflamma-
tory cytokine-IL-4 in stimulating giant cell formation is
fairly convincing both in vitro and in vivo [186,187]. Given
5021
that inflammation associated with normal wound healing is
resolved because of the transient nature of the macrophage
response, it would appear that the persistence of the
inflammatory stimulus in the form of a non-degradable
biomaterial confounds even the best intentions of these
cells to mediate healing.
What is becoming increasingly obvious is that a
macrophage is not always a macrophage, and that the
contribution of each of these subsets of macrophage
populations to the lack of healing of the vascular graft
remains to be investigated. Overall, the macrophage
remains the Jekyll and Hyde of graft healing but as our
insight into the mechanisms of healing grows so does our
understanding of which critical questions remain to be
answered in the elucidation of the role of this enigmatic
cell.
7. Half a century of vascular graft implantation: the essence
The key to new and emerging concepts for replacement
arteries lies in understanding the obstacles of contemporary
vascular prostheses. By analyzing studies of the past
half century, certain principles emerged which confirm or
defy prevailing stereotypes—some of them more than
others:
 The vast majority of contemporary synthetic vascular
grafts are so impervious that transmural tissue ingrowth
is impossible.
 None of the clinically used ePTFE, Dacron or PU
prostheses spontaneously develop a neo-intima, except
for sporadically observed small islands of endothelium.
 The overwhelming majority of animal experiments
studied TAE, a rather irrelevant mechanism for the
clinical situation.
 The cross-sectional quotients of the vast majority of
experimentally implanted grafts exceeded the clinical
situation by more than a factor 3, thereby seriously
distorting both the shear stress-linked potential for
developing significant anastomotic intimal hyperplasia
as well as endothelial cell function and metabolism.
 ‘‘Fall-out healing’’ leading to the formation of isolated
endothelial islands with no connection to either transa-
nastomotic or transmural tissue formations occurs
more often on Dacron than ePTFE grafts. All together,
however, this is a scanty and late-occurring phenomen-
on that does not play any significant role in contem-
porary vascular prostheses.
 If interstitial spaces are ingrowth-permissible, dimensions
AND structures seem to determine the tissue response.
While minimal dimensions are required to allow tissue
ingrowth in all, shape and sub-structure of the voids
appear to significantly affect the type of healing reaction.
 In senescent animal models and humans vessel,
ingrowth shows progressive transmural inhibition. In
contrast, juvenile models often allow complete trans-
mural endothelialization, suggesting that a competitive
 ARTICLE IN PRESS
5022 P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
situation between the alacrity of capillary ingrowth and
the build-up of an adverse biological environment exists.
 The inner fibrinous capsule—particularly pronounced on
Dacron prostheses—seems to be the main barrier against
tissue ingrowth. Although both connective tissue and
capillaries reach this capsule from underneath, they
hardly ever connect to the blood surface, even after
years of implantation.
 The mere fact that transmural endothelialization did
occur in some juvenile animal models refutes the
possibility of explaining ingrowth inhibition with a
reverse oxygen gradient from outside to inside.
8. Conclusions for vascular graft research
Many of the experimental concepts for the vascular
grafts of tomorrow promise to overcome the main
obstacles plaguing those prostheses currently available for
clinical implantation. At the same time, resolving one
problem may introduce another. By eliminating the
underlying cause of the chronic foreign body response by
using resorbable materials, for example, the resulting scar
tissue may deprive us of that very compliance which we
tried to introduce through new elastomers. Similarly, by
implanting dense, tissue-engineered tubes of smooth
muscle cells, we may deprive ourselves of the possibility
of allowing transmural endothelialization—should the
engineered endothelium undergo the initial detachment
often seen in veingrafts. However, as much as all these new
problems will need to be addressed in a dynamic way as
they come up, the overall questions asked and models used
need to be of relevance. Moreover, understanding where
we were misled in the past will be as helpful as under-
standing the obstacles which made half a century of SMG
research so frustrating and unrewarding.
 The main issue seems to be the need to clearly separate
trans-anastomotic from transmural or blood-borne
endothelialization. Isolation models, where an experi-
mental graft was separated from the artery by an
interposition segment, have been suggested before [16].
The externally sealed segments, however, would add
to the already high occlusion rate of small diameter
graft. In our hands the use of standard low-porosity
ePTFE instead sufficiently prevented transmural in-
growth without worsening overall thrombogenicity [112]
(Fig. 12).
 In as much as a smaller graft diameter means a higher
occlusion rate, vascular graft studies need to choose
animal models that emulate flow conditions of clinical
bypass grafts.
 The surrounding environment needs to be representative
for the clinical situation of SMGs. In order to avoid
abdominal implantations because of the need for too
long grafts, the length of the ‘isolation segments’ will
need to be optimized for each model and each question.
Fig. 12. Schematic representation of an isolation graft showing artery
(yellow), isolation segments (blue), and test segment (red). In order to
avoid the confusion of transanastomotic with transmural endothelializa-
tion, experimental grafts are isolated from the adjacent artery through
low-porosity isolation segments (ePTFE o30 mm IND). The length of the
isolation segments can be adjusted to the transanastomotic outgrowth
situation of the respective animal model as well as the planned
implantation period.
A prerequisite for this is a profound knowledge of the
TAE rates of a particular model.
 The build-up of adverse ingrowth conditions does not
occur in the presence of an endothelium. Therefore,
rapidly endothelializing animal models pre-empt this
build-up and thus fail to mimic a clinically relevant
situation. Senescent animal models are therefore prefer-
able to juvenile ones.
In summary, without a serious focus on developing new
animal models which address the real issues which have
plagued half a century of clinical graft implantation, l’art
pour l’art laboratory research will continue to chase virtual
rather than the real problems that led to the failure of
millions of bypass grafts in patients.
References
[1] Weinberg C, Bell E. A blood vessel model constructed from collagen
and cultured vascular cells. Science 1986;231:397–400.
[2] Niklason L. Techview: medical technology. Replacement arteries
made to order. Science 1999;286(5444):1493–4.
[3] Lheureux N, Germain L, Labbe R, Auger F. Construction of a
human blood vessel from cultured vascular cells; a morphologic
study. J Vasc Surg 1993;17:499–509.
[4] Hirai J, Matsuda T. Self-organized, tubular hybrid vascular tissue
composed of vascular cells and collagen for low-pressure-loaded
venous system. Cell Transplant 1995;4(6):597–608.
[5] Crombez M, Chevallier P, Gaudreault RC, Petitclerc E, Mantovani
D, Laroche G. Improving arterial prosthesis neo-endothelialization:
application of a proactive VEGF construct onto PTFE surfaces.
Biomaterials 2005;26(35):7402–9.
[6] Berger K, Sauvage L, Rao A, Wood S. Healing of arterial prostheses
in man: its incompleteness. Ann Surg 1972;175(1):118–27.
[7] Wesolowski S, Fries C, Gennigar G, Fox L, Sawyer P, Sauvage L.
Factors contributing to long-term failures in human vascular
prosthetic grafts. Cardiovas Surg 1964;38:544–67.
[8] Takebayashi J, Kamatani M, Namba S, Katagami Y, Hayashi K.
Early reparative process of implanted arterial prosthesis: a
morphological and hematological study. J Surg Res 1974;17(2):
102–10.
[9] Ghidoni J, Liotta D, Hall C, Adams J, Lechter A, Barrionueva M,
et al. Healing of pseudointimas in velour-lined, impermeable arterial
prostheses. Am J Pathol 1968:375–83.
 ARTICLE IN PRESS
P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
[10] Szilagyi D, Smith R, Elliott J, Allen H. Long-term behavior of a
Dacron arterial substitute: clinical, roentgenologic and histologic
correlations. Ann Surg 1965;162(3):453–77.
[11] Davids L, Dower T, Zilla P. The lack of healing in convenitonal
vascular grafts. In: Zilla P, Greisler H, editors. Tissue engineering of
prosthetic vascular grafts. Austin: R G Landes; 1999. p. 3–44.
[12] Lin P, Chen C, Bush R, Yao Q, Lumsden A, Hanson S. Small-
caliber heparin-coated ePTFE grafts reduce platelet deposition and
neointimal hyperplasia in a baboon model. J Vasc Surg 2004;39(6):
1322–8.
[13] Murray-Wijelath J, Lyman DJ, Wijelath ES. Vascular graft healing.
III. FTIR analysis of ePTFE graft samples from implanted bigrafts.
J Biomed Mater Res B Appl Biomater 2004;70(2):223–32.
[14] Pfeiffer T, Kever M, Grabitz K, Reiher L, Muller KM, Hildebrand
A, et al. Healing characteristics of small-calibre vascular prostheses
coated with plasmin-treated fibrin—an experimental study. Vasa
2000;29(2):117–24.
[15] Nishibe T, Okuda Y, Kumada T, Tanabe T, Yasuda K. Enhanced
graft healing of high-porosity expanded polytetrafluoroethylene
grafts by covalent bonding of fibronectin. Surg Today 2000;30(5):
426–31.
[16] Bhattacharya V, McSweeney PA, Shi Q, Bruno B, Ishida A, Nash
R, et al. Enhanced endothelialization and microvessel formation in
polyester grafts seeded with CD34(+) bone marrow cells. Blood
2000;95(2):581–5.
[17] Lyman DJ, Murray-Wijelath J, Ambrad-Chalela E, Wijelath ES.
Vascular graft healing. II. FTIR analysis of polyester graft samples
from implanted bi-grafts. J Biomed Mater Res 2001;58(3):221–37.
[18] Kuzuya A, Matsushita M, Oda K, Kobayashi M, Nishikimi N,
Sakurai T, et al. Healing of implanted expanded polytetrafluor-
oethylene vascular access grafts with different INDs: a histologic
study in dogs. Eur J Vasc Endovasc Surg 2004;28(4):404–9.
[19] Shimada T, Nishibe T, Miura H, Hazama K, Kato H, Kudo F, et al.
Improved healing of small-caliber, long-fibril expanded polytetra-
fluoroethylene vascular grafts by covalent bonding of fibronectin.
Surg Today 2004;34(12):1025–30.
[20] Ueberrueck T, Meyer L, Zippel R, Nestler G, Wahlers T, Gastinger
I. Healing characteristics of a new silver-coated, gelatine impreg-
nated vascular prosthesis in the porcine model. Zentralbl Chir
2005;130(1):71–6.
[21] Kenagy RD, Fischer JW, Lara S, Sandy JD, Clowes AW, Wight
TN. Accumulation and loss of extracellular matrix during shear
stress-mediated intimal growth and regression in baboon vascular
grafts. J Histochem Cytochem 2005;53(1):131–40.
[22] Phaneuf MD, Dempsey DJ, Bide MJ, Quist WC, LoGerfo FW.
Coating of Dacron vascular grafts with an ionic polyurethane: a
novel sealant with protein binding properties. England, 2001.
[23] Miura H, Nishibe T, Yasuda K, Shimada T, Hazama K, Katoh H,
et al. The influence of node-fibril morphology on healing of high-
porosity expanded polytetrafluoroethylene grafts. Eur Surg Res
2002;34(3):224–31.
[24] Begovac PC, Thomson RC, Fisher JL, Hughson A, Gallhagen A.
Improvements in GORE-TEX vascular graft performance by
Carmeda BioActive surface heparin immobilization. England, 2003.
[25] Hazama K, Nishibe T, Shimada T, Miura H, Watanabe S, Kondoh
S, et al. Effects of omental wrap on performance of small-caliber
high-porosity expanded polytetrafluoroethylene grafts. J Surg Res
1999;81(2):174–80.
[26] Lin PH, Bush RL, Yao Q, Lumsden AB, Chen C. Evaluation of
platelet deposition and neointimal hyperplasia of heparin-coated
small-caliber ePTFE grafts in a canine femoral artery bypass model.
J Surg Res 2004;118(1):45–52.
[27] Cagiannos C, Abul-Khoudoud OR, DeRijk W, Shell DHt, Jennings
LK, Tolley EA, et al. Rapamycin-coated expanded polytetrafluor-
oethylene bypass grafts exhibit decreased anastomotic neointimal
hyperplasia in a porcine model. J Vasc Surg 2005;42(5):980–8.
[28] Heise M, Schmidmaier G, Husmann I, Heidenhain C, Schmidt J,
Neuhaus P, et al. PEG-hirudin/iloprost coating of small diameter
5023
ePTFE grafts effectively prevents pseudointima and intimal
hyperplasia development. Eur J Vasc Endovasc Surg 2006;32(4):
418–24.
[29] Sun LB, Utoh J, Moriyama S, Tagami H, Okamoto K, Kitamura N.
Pretreatment of a Dacron graft with tissue factor pathway inhibitor
decreases thrombogenicity and neointimal thickness: a preliminary
animal study. ASAIO J 2001;47(4):325–8.
[30] Sterpetti A, Hunter W, Schultz R, Farina C. Healing of high-
porosity polytetrafluoroethylene arterial grafts is influenced by the
nature of the surrounding tissue. Surgery 1992;111(6):677–82.
[31] Kusaba A, Fischer CD, Matulewski T, Matsumoto T. Experimental
study of the influence of porosity on development of neointima in
Gore-Tex grafts: a method to increase long-term patency rate. Am
Surg 1981;47(8):347–54.
[32] Hollier L, Fowl R, Pennell R, Heck C, Winter K, Fass D, et al. Are
seeded endothelial cells the origin of neointima on prosthetic
vascular grafts? J Vasc Surg 1986;3(1):65–73.
[33] Graham L, Burkel W, Ford J, Vinter D, Kahn R, Stanley J.
Expanded polytetrafluoroethylene vascular prostheses seeded with
enzymatically derived and cultured canine endothelial cells. Surgery
1982;91(5):550–9.
[34] Herring M, Baughman S, Glover J, Kesler K, Jesseph J, Campbell J,
et al. Endothelial seeding of Dacron and polytetrafluoro-
ethylene grafts: the cellular events of healing. Surgery 1984;96(4):
745–55.
[35] Whitehouse WJ, Wakefield T, Vinter D, Ford J, Swanson D, Thrall
J, et al. Indium-111-oxine labeled platelet imaging of endothelial
seeded dacron thoracoabdominal vascular prostheses in a canine
model. Trans Am Soc Artif Intern Organs 1983;29:183–7.
[36] Golden M, Hanson S, Kirkman T, Schneider P, Clowes A. Healing
of polytetrafluoroethylene arterial grafts is influenced by graft
porosity. J Vasc Surg 1990;11(6):838–44 [discussion 845].
[37] Hussain S, Glover J, Augelli N, Bendick P, Maupin D, McKain M.
Host response to autologous endothelial seeding. J Vasc Surg
1989;9(5):656–63 [discussion 663-654].
[38] Geary R, Kohler T, Vergel S, Kirkman T, Clowes A. Time course of
flow-induced smooth muscle cell proliferation and intimal thicken-
ing in endothelialized baboon vascular grafts. Circ Res 1994;74(1):
14–23.
[39] Clowes A, Kirkman T, Reidy M. Mechanisms of arterial graft
healing. Rapid transmural capillary ingrowth provides a source of
intimal endothelium and smooth muscle in porous PTFE prostheses.
Am J Pathol 1986;123(2):220–30.
[40] Zamora J, Navarro L, Ives C, Weilbaecher D, Gao Z, Noon G.
Seeding of arteriovenous prostheses with homologous endothelium.
A preliminary report. J Vasc Surg 1986;3(6):860–6.
[41] Clowes A, Gown A, Hanson S, Reidy M. Mechanisms of arterial
graft failure. 1. Role of cellular proliferation in early healing of
PTFE prostheses. Am J Pathol 1985;118(1):43–54.
[42] Hanel K, McCabe C, Abbott W, Fallon J, Megerman J. Current
PTFE grafts: a biomechanical, scanning electron, and light
microscopic evaluation. Ann Surg 1982;195(4):456–63.
[43] Koveker G, Burkel W, Graham L, Wakefield T, Stanley J.
Endothelial cell seeding of expanded polytetrafluoroethylene vena
cava conduits: effects on luminal production of prostacyclin, platelet
adherence, and fibrinogen accumulation. J Vasc Surg 1988;7(4):
600–5.
[44] Ombrellaro M, Stevens S, Kerstetter K, Freeman M, Goldman M.
Healing characteristics of intra-arterial stented grafts: effect of intra-
luminal position on prosthetic graft healing. Surgery 1996;120(1):
60–70.
[45] Zilla P, Preiss P, Groscurth P, Rosemeier F, Deutsch M, Odell J,
et al. In vitro-lined endothelium: initial integrity and ultrastructural
events. Surgery 1994;116(3):524–34.
[46] Christenson J, Thulesius O, Owunwanne A, Nazzal M. Forskolin
impregnation of small calibre PTFE grafts lowers early platelet graft
sequestration and improves patency in a sheep model. Eur J Vasc
Surg 1991;5(3):271–5.
 ARTICLE IN PRESS
5024 P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
[47] Aldenhoff Y, van DVF, ter WJ, Habets J, Poole - WL, Koole L.
Performance of a polyurethane vascular prosthesis carrying a
dipyridamole (Persantin) coating on its lumenal surface. J Biomed
Mater Res 2001;54(2):224–33.
[48] Fleser P, Nuthakki V, Malinzak L, Callahan R, Seymour M,
Reynolds M, et al. Nitric oxide-releasing biopolymers inhibit
thrombus formation in a sheep model of arterio-venous bridge
grafts. J Vasc Surg 2004;40(4):803–11.
[49] Poole - WL, Schindhelm K, Graham A, Slowiaczek P, Noble K.
Performance of small diameter synthetic vascular prostheses with
confluent autologous endothelial cell linings. J Biomed Mater Res
1996;30(2):221–9.
[50] Mansfield P, Wechezak A, Sauvage L. Preventing thrombus on
artificial vascular surfaces: true endothelial cell linings. Trans Am
Soc Artif Intern Organs 1975;21:264–72.
[51] DeBakey M, Jordan GJ, Beall A, O’Neal R, Abbott J, Halpert B.
Basic biologic reactions to vascular grafts and prostheses. Surg Clin
North Am 1965;45:477.
[52] Herring M, Gardner A, Peigh P, Madison D, Baughman S, Brown
J, et al. Patency in canine inferior vena cava grafting: effects of graft
material, size, and endothelial seeding. J Vasc Surg 1984;1(6):
877–87.
[53] Keough E, Callow A, Connolly R, Weinberg K, Aalberg J, O DTJ.
Healing pattern of small caliber Dacron grafts in the baboon: an
animal model for the study of vascular prostheses. J Biomed Mater
Res 1984;18(3):281–92.
[54] van der Lei B, Wildevuur C. Microvascular polytetrafluoroethylene
prostheses: the cellular events of healing and prostacyclin produc-
tion. Plast Reconstr Surg 1988;81(5):735–41.
[55] Stronck J, van dLB, Wildevuur C. J Thorac Cardiovasc Surg
1992;103(1):146–52.
[56] van der Lei B, Wildevuur C. Improved healing of microvascular
PTFE prostheses by induction of a clot layer: an experimental study
in rats. Plast Reconstr Surg 1989;84(6):960–8.
[57] Friedman E, Hamilton A. Polytetrafluoroethylene grafts in the
peripheral venous circulation of rabbits. Am J Surg 1983;146(3):355–9.
[58] Stewart G, Essa N, Chang K, Reichle F. A scanning and
transmission electron microscope study of the luminal coating on
Dacron prostheses in the canine thoracic aorta. J Lab Clin Med
1975;85(2):208–26.
[59] Kohler T, Kirkman T, Kraiss L, Zierler B, Clowes A. Increased
blood flow inhibits neointimal hyperplasia in endothelialized
vascular grafts. Circ Res 1991;69(6):1557–65.
[60] Binns R, Ku D, Stewart M, Ansley J, Coyle K. Optimal graft
diameter: effect of wall shear stress on vascular healing. J Vasc Surg
1989;10(3):326–37.
[61] Kraiss L, Geary R, Mattsson E, Vergel S, Au Y, Clowes A. Acute
reductions in blood flow and shear stress induce platelet-derived
growth factor-a expression in baboon prosthetic grafts. Circ Res
1996;79(1):45–53.
[62] Zilla P, Human P, Wolf M, Lichtenberg W, Raffie N, Bezuidenhout
D, et al. Clinically relevant downsizing of veingrafts through
external nitinol stenting in non-human primates. 2007. Submitted
for publication
[63] Florian A, Cohn L, Dammin G, Collins JJ. Small vessel replacement
with gore-tex (expanded polytetrafluoroethylene). Arch Surg 1976;
111(3):267–70.
[64] Matsumoto H, Hasegawa T, Fuse K, Yamamoto M, Saigusa M. A
new vascular prosthesis for a small caliber artery. Surgery 1973;
74(4):519–23.
[65] Shi Q, Wu M, Hayashida N, Wechezak A, Clowes A, Sauvage L.
Proof of fallout endothelialization of impervious Dacron grafts in
the aorta and inferior vena cava of the dog. J Vasc Surg 1994;
20(4):546–56 [discussion 556-547].
[66] Baitella - EG, Groscurth P, Zilla P, Lachat M, Muller - GW,
Schneider J, et al. Long-term results of tissue development and cell
differentiation on Dacron prostheses seeded with microvascular cells
in dogs. J Vasc Surg 1993;18(6):1019–28.
[67] Shepard A, Eldrup - JJ, Keough E, Foxall T, Ramberg K, Connolly
R, et al. Endothelial cell seeding of small-caliber synthetic grafts in
the baboon. Surgery 1986;99(3):318–26.
[68] Clowes A, Kirkman T, Clowes M. Mechanisms of arterial graft
failure. II. Chronic endothelial and smooth muscle cell proliferation
in healing polytetrafluoroethylene prostheses. J Vasc Surg 1986;3(6):
877–84.
[69] Sho E, Nanjo H, Sho M, Kobayashi M, Komatsu M, Kawamura K,
et al. Arterial enlargement, tortuosity, and intimal thickening in
response to sequential exposure to high and low wall shear stress.
J Vasc Surg 2004;39(3):601–12.
[70] Zacharias R, Kirkman T, Clowes A. Mechanisms of healing in
synthetic grafts. J Vasc Surg 1987;6(5):429–36.
[71] Therrien M, Guidoin R, Adnot A, Paynter R. Hydrophobic and
fibrillar microporous polyetherurethane urea prosthesis: an ESCA
study on the internal and external surfaces of explanted grafts.
Biomaterials 1989;10(8):517–20.
[72] Leidner J, Wong E. A novel process for the manufacturing of
porous grafts: Process description and product evaluation. J Biomed
Mater Res 1983;17:229–47.
[73] Wilson GJ, MacGregor DC, Klement P, Lee JM, del Nido PJ,
Wong EW, et al. Anisotropic polyurethane nonwoven conduits: a
new approach to the design of a vascular prosthesis. Trans Am Soc
Artif Intern Organs 1983;29:260–8.
[74] Seifalian AM, Salacinski HJ, Tiwari A, Edwards A, Bowald S,
Hamilton G. In vivo biostability of a poly(carbonate–urea)urethane
graft. Biomaterials 2003;24(14):2549–57.
[75] Berkowitz HD, Perloff LJ, Roberts B. Pseudointimal development
on microporous polyurethane lattices. Trans Am Soc Artif Intern
Organs 1972;18(0):25–9.
[76] Sauvage L, Berger K, Wood S, Nakagawa Y, Mansfield P. An
external velour surface for porous arterial prostheses. Surgery 1971;
70(6):940–53.
[77] Margolin D, Kaufman B, DeLuca D, Fox P, Graham L. Increased
platelet-derived growth factor production and intimal thickening
during healing of Dacron grafts in a canine model. J Vasc Surg 1993;
17(5):858–66 [discussion 866-857].
[78] Haimovici H. History of vascular surgery. In: Haimovici H, editor.
Vascular surgery principles and techniques. 2nd ed. Norwark,
Connecticut: Appleton-Century-Crofts; 1984. p. 3–18.
[79] Voorhees A, Jaretzki A, Blakemore A. The use of tubes constructed
from vinyon ‘‘N’’ cloth in bridging arterial defects. A preliminary
report. Ann Surg 1952:332–6.
[80] Wesolowski S. Foundations of modern vascular grafts. In: Sawyer
P, Kaplitt M, editors. Vascular grafts. New York: Appleton Century
Croft; 1978.
[81] Tabbara M, White R. Biologic and prosthetic materials for vascular
conduits. In: Veith F, Hobson R, Williams R, Wilson S, editors.
Vascular surgery: principles and practice. USA: McGraw-Hill, Inc;
1994. p. 523–35.
[82] Mathisen S, Wu H, Sauvage L, Usui Y, Walker M. An experimental
study of eight current arterial prostheses. J Vasc Surg 1986;4(1):
33–41.
[83] Campbell C, Goldfarb D, Roe R. A small arterial substitute:
expanded microporous polytetrafluoroethylene: patency versus
porosity. Ann Surg 1975;182(2):138–43.
[84] Bellon J, Bujan J, Contreras L, Hernando A, Jurado F. Simi-
larity in behavior of polytetrafluoroethylene (ePTFE) prostheses im-
planted into different interfaces. J Biomed Mater Res 1996;31(1):
1–9.
[85] Hirschi K, Rohovsky S, D AP. Cell–cell interactions in vessel
assembly: a model for the fundamentals of vascular remodelling.
Transpl Immunol 1997;5(3):177–8.
[86] Beck LJ, D AP. Vascular development: cellular and molecular
regulation. FASEB J 1997;11(5):365–73.
[87] Cheresh D, Berliner S, Vicente V, Ruggeri Z. Recognition of distinct
adhesive sites on fibrinogen by related integrins on platelets and
endothelial cells. Cell 1989;58(5):945–53.
 ARTICLE IN PRESS
P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
[88] Dvorak H, Nagy J, Berse B, Brown L, Yeo K, Yeo T, et al. Vascular
permeability factor, fibrin, and the pathogenesis of tumor stroma
formation. Ann N Y Acad Sci 1992;667:101–11.
[89] Gamble J, Matthias L, Meyer G, Kaur P, Russ G, Faull R, et al.
Regulation of in vitro capillary tube formation by anti-integrin
antibodies. J Cell Biol 1993;121(4):931–43.
[90] Bezuidenhout D, Davies N, Zilla P. Effect of well defined
dodecahedral porosity on inflammation and angiogenesis. ASAIO
J 2002;48(5):465–71.
[91] Annis D, Bornat A, Edwards RO, Higham A, Loveday B, Wilson J.
An elastomeric vascular prosthesis. Trans Am Soc Artif Intern
Organs 1978;24:209–14.
[92] Eberhart A, Zhang Z, Guidoin R, Laroche G, Guay L, De LFD,
et al. A new generation of polyurethane vascular prostheses: rara
avis or ignis fatuus? J Biomed Mater Res 1999;48(4):546–58.
[93] Wilson GJ, MacGregor DC, Klement P, Dereume JP, Weber BA,
Binnington AG, et al. The composite Corethane/Dacron vascular
prosthesis, canine in vivo evaluation of, 4, mm diameter grafts with
1 year follow-up. ASAIO Trans 1991;37(3, M475–476).
[94] Wintermantel E, Mayer J, Blum J, Eckert K, Luscher P, Mathey M.
Tissue engineering scaffolds using superstructures. Biomaterials
1996;17(2):83–91.
[95] Gupta B, Kasyanov V. Biomechanics of human common carotid
artery and design of novel hybrid textile compliant vascular grafts. J
Biomed Mater Res 1997;34(3):341–9.
[96] Hayashi K, Takamizawa K, Saito T, Kira K, Hiramatsu K, Kondo
K. Elastic properties and strength of a novel small-diameter,
compliant polyurethane vascular graft. J Biomed Mater Res
1989;23(A2 Suppl):229–44.
[97] Chen J, Laiw R, Jiang S, Lee Y. Microporous segmented
polyetherurethane vascular graft: I. Dependency of graft morphol-
ogy and mechanical properties on compositions and fabrication
conditions. J Biomed Mater Res 1999;48(3):235–45.
[98] Martz H, Beaudoin G, Paynter R, King M, Marceau D, Guidoin R.
Physicochemical characterization of a hydrophilic microporous
polyurethane vascular graft. J Biomed Mater Res 1987;21(3):
399–412.
[99] Zdrahala R. Small caliber vascular grafts. Part II: polyurethanes
revisited. J Biomater Appl 1996;11(1):37–61.
[100] Kowligi RR, von Maltzahn WW, Eberhart RC. Synthetic vascular
graft fabrication by a precipitation-flotation method. ASAIO Trans
1988;34(3):800–4.
[101] Kowligi RR, von Maltzahn WW, Eberhart RC. Fabrication and
characterization of small-diameter vascular prostheses. J Biomed
Mater Res 1988;22(3 Suppl):245–56.
[102] Doi K, Nakayama Y, Matsuda T. Novel compliant and tissue-
permeable microporous polyurethane vascular prosthesis fabricated
using an excimer laser ablation technique. J Biomed Mater Res
1996;31(1):27–33.
[103] Hiratzka LF, Goeken JA, White RA, Wright CB. In vivo
comparison of replamineform, silastic, and bioelectric polyurethane
arterial grafts. Arch Surg 1979;114(6):698–702.
[104] Berkowitz H, Perloff L, Roberts B. Pseudointimal development on
microporous polyurethane lattices. Surgery 1972;72(6):888–96.
[105] Hinrichs WL, Kuit J, Feil H, Wildevuur CR, Feijen J. In vivo
fragmentation of microporous polyurethane- and copolyesterether
elastomer-based vascular prostheses. Biomaterials 1992;13(9):585–93.
[106] Murabayashi S, Kambic H, Harasaki H, Morimoto T, Yozu R,
Nose Y. Fabrication and long-term implantation of semi-compliant
small vascular prosthesis. Trans Am Soc Artif Intern Organs
1985;31:50–4.
[107] Uchida N, Kambic H, Emoto H, Chen JF, Hsu S, Murabayshi S,
et al. Compliance effects on small diameter polyurethane graft
patency. J Biomed Mater Res 1993;27(10):1269–79.
[108] Kambic H, Murabayashi S, Yozu R, Morimoto T, Furuse M,
Harasaki H, et al. Small vessel replacement with elastomeric protein
composite materials: preliminary studies. Trans Am Soc Artif Intern
Organs 1984;30:406–10.
5025
[109] Williams SK, Carter T, Park PK, Rose DG, Schneider T, Jarrell BE.
Formation of a multilayer cellular lining on a polyurethane vascular
graft following endothelial cell sodding. J Biomed Mater Res
1992;26(1):103–17.
[110] van der Lei B, Wildevuur C, Dijk F, Blaauw E, Molenaar I,
Nieuwenhuis P. Sequential studies of arterial wall regeneration in
microporous, compliant, biodegradable small-caliber vascular grafts
in rats. J Thorac Cardiovasc Surg 1987;93(5):695–707.
[111] Nam Y, Park T. Porous biodegradable polymeric scaffolds prepared
by thermally induced phase separation. J Biomed Mater Res 1999;
47(1):8–17.
[112] Bezuidenhout D, Vici M, Davies N, Schilling L, Fittkau M, Zilla P.
Small diameter vascular grafts: a Meaningful model at last. Nineth
biennial meeting of the ISACB: Savannah, GA; 2004.
[113] Boyd K, Schmidt S, Pippert T, Sharp W. Endothelial cell seeding of
ULTI carbon-coated small-diameter PTFE vascular grafts. ASAIO
Trans 1987;33(3):631–5.
[114] Sterpetti A, Lepidi S, Cucino A, Patrizi A, Palumbo R, Taranta A,
et al. Growth factor production after polytetrafluoroethylene and
vein arterial grafting: an experimental study. J Vasc Surg 1996;
23(3):452–60.
[115] Bartels H, van dLB, Robinson P. Contrib Title: Prosthetic
microvenous grafting in the rat femoral vein. Lab Anim 1993;27(1):
47–54.
[116] Van der Lei B, Stronck J, Wildevuur C. Enhanced healing of 30
microns Gore-Tex PTFE microarterial prostheses by alcohol-
pretreatment. Br J Plast Surg 1991;44(6):428–33.
[117] Kenney D, Tu R, Peterson R. Evaluation of compliant and
noncompliant PTFE vascular prostheses. ASAIO Trans 1988;34(3):
661–3.
[118] Herring M, Gardner A, Glover J. Endothelial seeding on vascular
prostheses. Arch Surg 1979;114:679.
[119] Douville E, Kempczinski R, Birinyi L, Ramalanjaona G. Impact of
endothelial cell seeding on long-term patency and subendothelial
proliferation in a small-caliber highly porous polytetrafluoroethy-
lene graft. J Vasc Surg 1987;5(4):544–50.
[120] Bull D, Hunter G, Holubec H, Aguirre M, Rappaport W, Putnam
C. Cellular origin and rate of endothelial cell coverage of PTFE
grafts. J Surg Res 1995;58(1):58–68.
[121] Golden M, Au Y, Kirkman T, Wilcox J, Raines E, Ross R, et al.
Platelet-derived growth factor activity and mRNA expression in
healing vascular grafts in baboons. Association in vivo of platelet-
derived growth factor mRNA and protein with cellular prolifera-
tion. J Clin Invest 1991;87(2):406–14.
[122] Clowes A, Zacharias R, Kirkman T. Early endothelial coverage of
synthetic arterial grafts: porosity revisited. Am J Surg 1987;153(5):
501–4.
[123] Lado M, Knighton D, Cavallini M, Fiegel V, Murray C, Phillips G.
Induction of neointima formation by platelet derived angiogenesis
fraction in a small diameter, wide pore, PTFE graft. Int J Artif
Organs 1992;15(12):727–36.
[124] Hirschi KK, Rohovsky SA, D’Amore PA. PDGF, TGF-beta, and
heterotypic cell–cell interactions mediate endothelial cell-induced
recruitment of 10T1/2 cells and their differentiation to a smooth
muscle fate. J Cell Biol 1998;141(3):805–14.
[125] Koyama N, Hart C, Clowes A. Different functions of the platelet-
derived growth factor-alpha and -beta receptors for the migration
and proliferation of cultured baboon smooth muscle cells. Circ Res
1994;75(4):682–91.
[126] Noishiki Y. Pattern of arrangement of smooth muscle cells in
neointimae of synthetic vascular prostheses. J Thorac Cardiovasc
Surg 1978;75(6):894–901.
[127] Bellon J, Bujan J, Hernando A, Honduvilla N, Jurado F. Arterial
autografts and PTFE vascular microprostheses: similarities in the
healing process. Eur J Vasc Surg 1994;8(6):694–702.
[128] Harrison H. Synthetic materials as vascular prostheses. Ia. A
comparative study in small vessels of Nylon, Dacron, Orlon, Ivalon
sponge and Teflon. Amer J Surg 1958;95:3–15.
 ARTICLE IN PRESS
5026 P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
[129] Weselow A. Biological behaviour of tissue and prosthetic grafts. In:
Haimovici H, editor. Vascular surgery: principles and techniques.
New York: Appleton-Century-Crofts; 1984. p. 93–118.
[130] Wu M, Shi Q, Wechezak A, Clowes A, Gordon I, Sauvage L.
Definitive proof of endothelialization of a Dacron arterial prosthesis
in a human being. J Vasc Surg 1995;21(5):862–7.
[131] Shi Q, Wu M, Onuki Y, Ghali R, Hunter G, Johansen K, et al.
Endothelium on the flow surface of human aortic Dacron vascular
grafts. J Vasc Surg 1997;25(4):736–42.
[132] Herring M, Dilley R, Jersild RJ, Boxer L, Gardner A, Glover J.
Seeding arterial prostheses with vascular endothelium. The nature of
the lining. Ann Surg 1979;190(1):84–90.
[133] Burkel W, Ford J, Vinter D, Kahn R, Graham L, Stanley J. Fate of
knitted dacron velour vascular grafts seeded with enzymatically
derived autologous canine endothelium. Trans Am Soc Artif Intern
Organs 1982;28:178–84.
[134] Schmidt S, Hunter T, Hirko M, Belden T, Evancho M, Sharp W,
et al. Small-diameter vascular prostheses: two designs of PTFE and
endothelial cell-seeded and nonseeded Dacron. J Vasc Surg 1985;
2(2):292–7.
[135] Schmidt S, Monajjem N, Evancho M, Pippert T, Sharp W.
Microvascular endothelial cell seeding of small-diameter Dacron
vascular grafts. J Invest Surg 1988;1(1):35–44.
[136] Criado E, Marston W, Reddick R, Woosley J. Contrib Title:
Endothelial coverage of endovascular Dacron grafts in dogs [letter;
comment]. J Vasc Surg 1996;23(4):736–7.
[137] Sottiurai V, Sue S, Rau D, Tran A. Comparative analysis of
pseudointima biogenesis in Gelseal coated Dacron knitted graft
versus crimped and noncrimped graft. J Cardiovasc Surg (Torino)
1989;30(6):902–9.
[138] Nomura Y. The ultra-structure of the pseudointima lining synthetic
arterial grafts in the canine aorta with special reference to the origin
of the endothelial cell. J Cardiovasc Surg (Torino) 1970;11(4):
282–91.
[139] Mesh C, Majors A, Mistele D, Graham L, Ehrhart L. Graft smooth
muscle cells specifically synthesize increased collagen. J Vasc Surg
1995;22(2):142–9.
[140] Hertzer N. Regeneration of endothelium in knitted and velour
dacron vascular grafts in dogs. J Cardiovasc Surg (Torino)
1981;22(3):223–30.
[141] Rosenfeld J, Savarese R, McCombs P, DeLaurentis D. Endothelial
infiltration and lining of knitted dacron arterial grafts. Surgical
forum 1981.
[142] DeBakey M, Jordan GJ, Abbott J, Halbert B, O’Neal R. The fate of
Dacron vascular grafts. Arch Surg 1964;89:757–82.
[143] Burkel W, Vinter D, Ford J, Kahn R, Graham L, Stanley J.
Sequential studies of healing in endothelial seeded vascular
prostheses: histologic and ultrastructure characteristics of graft
incorporation. J Surg Res 1981;30(4):305–24.
[144] Stumb M, Jordan G, DeBakey M. Endothelium growth from
circulating blood on isolated intravascular Dacron hub. Amer J
Path 1963;43:361–8.
[145] Hammond W. Surface population with blood-borne cells. In: Zilla
P, Greisler H, editors. Tissue engineering of prosthetic vascular
grafts. Austin: Landes Bioscience; 1998.
[146] Wu M, Shi Q, Kouchi Y, Onuki Y, Ghali R, Yoshida H, et al.
Implant site influence on arterial prosthesis healing: a comparative
study with a triple implantation model in the same dog. J Vasc Surg
1997;25(3):528–36.
[147] Hess F, Jerusalem C, Steeghs S, Reijnders O, Braun B, Grande P.
Development and long-term fate of a cellular lining in fibrous
polyurethane vascular prostheses implanted in the dog carotid and femo-
ral artery. A scanning and light microscopical study up to 53 months
after implantation. J Cardiovasc Surg (Torino) 1992;33(3):358–65.
[148] Kogel H, Vollmar JF, Cyba-Altunbay S, Mohr W, Frosch D,
Amselgruber W. New observations on the healing process in
prosthetic substitution of large veins by microporous grafts—animal
experiments. Thorac Cardiovasc Surg 1989;37(2):119–24.
[149] Walpoth B, Rogulenko R, Tikhvinskaia E, Gogolewski S, Schaffner
T, Hess O, et al. Improvement of patency rate in heparin-coated
small synthetic vascular grafts. Circulation 1998;98(19 Sup-
pl):II319–23 [discussion II324].
[150] Bull P, Denck H, Guidoin R, Gruber H. Preliminary clinical
experience with polyurethane vascular prostheses in femoro-
popliteal reconstruction. Eur J Vasc Surg 1992;6(2):217–24.
[151] Hess F, Jerusalem C, Braun B. The endothelialization process of a
fibrous polyurethane microvascular prosthesis after implantation in
the abdominal aorta of the rat. A scanning electron microscopic
study. J Cardiovasc Surg (Torino) 1983;24(5):516–24.
[152] Zhang Z, Marois Y, Guidoin R, Bull P, Marois M, How T, et al.
Vascugraft polyurethane arterial prosthesis as femoro-popliteal and
femoro-peroneal bypasses in humans: pathological, structural and
chemical analyses of four excised grafts. Biomaterials 1997;18(2):
113–24.
[153] Huang B, Marois Y, Roy R, Julien M, Guidoin R. Cellular reaction
to the Vascugraft polyesterurethane vascular prosthesis: in vivo
studies in rats. Biomaterials 1992;13(4):209–16.
[154] Ives CL, Zamora JL, Eskin SG, Weilbaecher DG, Gao ZR,
Noon GP, et al. In vivo investigation of a new elastomeric vascular
graft (mitrathane). Trans Am Soc Artif Intern Organs 1984;30:
587–90.
[155] Ota K, Sasaki Y, Nakagawa Y, Teraoka S. A completely new
poly(ether–urethane) graft ideal for hemodialysis blood access.
ASAIO Trans 1987;33(3):129–35.
[156] Pollock E, Andrews EJ, Lentz D, Sheikh K. Tissue ingrowth and
porosity of biomer. Trans Am Soc Artif Intern Organs 1981;27:
405–9.
[157] White RA. The effect of porosity and biomaterial on the healing and
long-term mechanical properties of vascular prostheses. ASAIO
Trans 1988;34(2):95–100.
[158] Brothers TE, Stanley JC, Burkel WE, Graham LM. Small-caliber
polyurethane and polytetrafluoroethylene grafts: a comparative
study in a canine aortoiliac model. J Biomed Mater Res 1990;24(6):
761–71.
[159] Uchida N, Emoto H, Kambic H, Harasaki H, Chen JF, Hsu SH,
et al. Compliance effect on patency of small diameter vascular
grafts. ASAIO Trans 1989;35(3):556–8.
[160] Crane C. Contrib Title: Arteriosclerotic aneurysm of the abdominal
aorta; some pathological and clinical correlations. N Engl J Med
1955;253(22):954–8.
[161] Pepper M, Vassalli J, Orci L, Montesano R. Biphasic effect of
transforming growth factor-beta 1 on in vitro angiogenesis. Exp Cell
Res 1993;204(2):356–63.
[162] Herbert C, Nagaswami C, Bittner G, Hubbell J, Weisel J. Effects of
fibrin micromorphology on neurite growth from dorsal root ganglia
cultured in three-dimensional fibrin gels. J Biomed Mater Res 1998;
40(4):551–9.
[163] Nehls V, Herrmann R. The configuration of fibrin clots determines
capillary morphogenesis and endothelial cell migration. Microvasc
Res 1996;51(3):347–64.
[164] Nunes I, Shapiro R, Rifkin D. Characterization of latent TGF-beta
activation by murine peritoneal macrophages. J Immunol 1995;
155(3):1450–9.
[165] Wahl S, Allen J, Weeks B, Wong H, Klotman P. Transforming
growth factor beta enhances integrin expression and type IV
collagenase secretion in human monocytes. Proc Natl Acad Sci U
S A 1993;90(10):4577–81.
[166] Falcone D, McCaffrey T, Haimovitz - FA, Garcia M. Transforming
growth factor-beta 1 stimulates macrophage urokinase expression
and release of matrix-bound basic fibroblast growth factor. J Cell
Physiol 1993;155(3):595–605.
[167] Loscalzo J. The macrophage and fibrinolysis. Semin Thromb
Hemost 1996;22(6):503–6.
[168] Gross T, Leavell K, Peterson M. CD11b/CD18 mediates the
neutrophil chemotactic activity of fibrin degradation product D
domain. Thromb Haemost 1997;77(5):894–900.
 ARTICLE IN PRESS
P. Zilla et al. / Biomaterials 28 (2007) 5009–5027
[169] Premack B, Schall T. Chemokine receptors: gateways to inflamma-
tion and infection. Nat Med 1996;2(11):1174–8.
[170] Rhodes N, Hunt J, Williams D. Macrophage subpopulation
differentiation by stimulation with biomaterials. J Biomed Mater
Res 1997;37(4):481–8.
[171] Kuwabara K, Ogawa S, Matsumoto M, Koga S, Clauss M, Pinsky
D, et al. Hypoxia-mediated induction of acidic/basic fibroblast
growth factor and platelet-derived growth factor in mononuclear
phagocytes stimulates growth of hypoxic endothelial cells. Proc Natl
Acad Sci U S A 1995;92(10):4606–10.
[172] Van Otteren G, Standiford T, Kunkel S, Danforth J, Strieter R.
Alterations of ambient oxygen tension modulate the expression of
tumor necrosis factor and macrophage inflammatory protein-1
alpha from murine alveolar macrophages. Am J Respir Cell Mol
Biol 1995;13(4):399–409.
[173] Leibovich S, Polverini P, Shepard H, Wiseman D, Shively V, Nuseir
N. Macrophage-induced angiogenesis is mediated by tumour
necrosis factor-alpha. Nature 1987;329(6140):630–2.
[174] Taichman N, Young S, Cruchley A, Taylor P, Paleolog E. Human
neutrophils secrete vascular endothelial growth factor. J Leukoc
Biol 1997;62(3):397–400.
[175] Kao W, Zhao Q, Hiltner A, Anderson J. Theoretical analysis of in
vivo macrophage adhesion and foreign body giant cell formation on
polydimethylsiloxane, low density polyethylene, and polyetherur-
ethanes. J Biomed Mater Res 1994;28(1):73–9 [published erratum
appears in J Biomed Mater Res 1994 Jun;28(6):761].
[176] Mathur A, Collier T, Kao W, Wiggins M, Schubert M, Hiltner A,
et al. In vivo biocompatibility and biostability of modified
polyurethanes. J Biomed Mater Res 1997;36(2):246–57.
[177] Defife K, Jenney C, Colton E, Anderson J. Confocal and light
microscopic evaluation of silane surface-dependent macrophage
development and IL-4 induced foreign body giant cell formation.
Fifth world biomaterials congress, 1996.
[178] McNally A, Anderson J. The lymphokine IL-4 induces very large
FBGC and syncytia from human macs in a material surface property
dependent manner in vitro. Fifth world biomaterials congress, 1996.
[179] Kao W, Hiltner A, Anderson J, Lodoen G. Theoretical analysis of
in vivo macrophage adhesion and foreign body giant cell formation
on strained poly(etherurethane urea) elastomers. J Biomed Mater
Res 1994;28(7):819–29.
[180] Petillo O, Peluso G, Ambrosio L, Nicolais L, Kao W, Anderson J.
In vivo induction of macrophage Ia antigen (MHC class II)
5027
expression by biomedical polymers in the cage implant system.
J Biomed Mater Res 1994;28(5):635–46.
[181] Swartbol P, Truedsson L, Parsson H, Norgren L. Surface adhesion
molecule expression on human blood cells induced by vascular graft
materials in vitro. J Biomed Mater Res 1996;32(4):669–76.
[182] Bonfield T, Colton E, Anderson J. Protein adsorption of biomedical
polymers influences activated monocytes to produce fibroblast
stimulating factors. J Biomed Mater Res 1992;26(4):457–65.
[183] Bonfield T, Colton E, Marchant R, Anderson J. Cytokine and
growth factor production by monocytes/macrophages on protein
preadsorbed polymers. J Biomed Mater Res 1992;26(7):837–50.
[184] Hunt J, Meijs G, Williams D. Hydrophilicity of polymers and soft
tissue responses: a quantitative analysis. J Biomed Mater Res 1997;
36(4):542–9.
[185] McNally A, DeFife K, Anderson J. Interleukin-4-induced macro-
phage fusion is prevented by inhibitors of mannose receptor activity.
Am J Pathol 1996;149(3):975–85.
[186] Kaufman B, DeLuca D, Folsom D, Mansell S, Gorman M, Fox P,
et al. Elevated platelet-derived growth factor production by aortic
grafts implanted on a long-term basis in a canine model. J Vasc Surg
1992;15(5):806–15 [discussion 815-806].
[187] McNally A, Anderson J. Interleukin-4 induces FBGCs from human
monocytes/macrophages. Differential lymphokine regulation of
macrophage fusion leads to morphological variants of multi-
nucleated giant cells. Am J Pathol 1995;147(5):1487–99.
[188] Essner R, Rhoades K, McBride W, Morton D, Economou J. IL-4
down-regulates IL-1 and TNF gene expression in human mono-
cytes. J Immunol 1989;142(11):3857–61.
[189] Zhao Q, Topham N, Anderson J, Hiltner A, Lodoen G, Payet C.
Foreign-body giant cells and polyurethane biostability: in vivo
correlation of cell adhesion and surface cracking. J Biomed Mater
Res 1991;25(2):177–83.
[190] Anderson J. Inflammatory reaction: the nemesis of implants. In:
Zilla P, Greisler H, editors. Tissue engineering of prosthetic vascular
grafts. Austin: Landes Bioscience; 1998.
[191] Forsyth E, Aly H, Neville R, Sidawy A. Proliferation and
extracellular matrix production by human infragenicular smooth
muscle cells in response to interleukin-1 beta. J Vasc Surg
1997;26(6):1002–7 [discussion 1007–1008].
[192] Makita S, Nakamura M, Yoshida H, Hiramori K. Autocrine growth
inhibition of IL-1 beta-treated cultured human aortic smooth muscle
cells: possible role of nitric oxide. Heart Vessels 1996;11(5):223–8.