Annu. Rev. Biomed. Eng. 1999.

01:299–329
Copyright q 1999 by Annual Reviews. All rights reserved

Fluid Mechanics of Vascular

Systems, Diseases, and Thrombosis
David M. Wootton1 and David N. Ku
G.W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology,
Atlanta, Georgia 30332–0405; e-mail: [email protected],
[email protected]

Key Words platelets, shear, arteriosclerosis, stenosis, intimal hyperplasia

Abstract The cardiovascular system is an internal flow loop with multiple
branches circulating a complex liquid. The hallmarks of blood flow in arteries are
pulsatility and branches, which cause wall stresses to be cyclical and nonuniform.
Normal arterial flow is laminar, with secondary flows generated at curves and
branches. Arteries can adapt to and modify hemodynamic conditions, and unusual
hemodynamic conditions may cause an abnormal biological response. Velocity profile
skewing can create pockets in which the wall shear stress is low and oscillates in
direction. Atherosclerosis tends to localize to these sites and creates a narrowing of
the artery lumen—a stenosis. Plaque rupture or endothelial injury can stimulate throm-
bosis, which can block blood flow to heart or brain tissues, causing a heart attack or
stroke. The small lumen and elevated shear rate in a stenosis create conditions that
accelerate platelet accumulation and occlusion. The relationship between thrombosis
and fluid mechanics is complex, especially in the post-stenotic flow field. New con-
vection models have been developed to predict clinical occlusion from platelet throm-
bosis in diseased arteries. Future hemodynamic studies should address the complex
mechanics of flow-induced, large-scale wall motion and convection of semisolid par-
ticles and cells in flowing blood.

CONTENTS
Introduction ..................................................................................... 300
Physiologic Environment.................................................................... 300
Flows in Specific Arteries................................................................... 302
The Carotid Arteries .........................................................................
302
The Aorta ......................................................................................
303
Flow at the Left Coronary Artery Bifurcation ...........................................
304
Flows in the Heart and Great Vessels.....................................................
305
Biological Responses to Hemodynamics ............................................... 305
Hemodynamics of Stenoses ................................................................ 308

1
Department of Biomedical Engineering, Johns Hopkins University School of Medicine,
Baltimore, Maryland 21205

1523–9829/99/0820–0299$08.00 299
300 WOOTTON n KU

Diagnosis of Disease......................................................................... 309

Shear-Dependent Thrombosis .............................................................. 310
Arterial Thrombosis .......................................................................... 310
Cellular and Molecular Mechanisms of Thrombosis.................................... 311
Hemodynamics and Thrombosis .......................................................... 312
Hemodynamics in Advanced Atherosclerosis............................................. 313
Hemodynamics and Thrombus Composition ............................................. 313
Shear and Platelet Accumulation .......................................................... 313
Shear-Linked Mechanisms .................................................................. 314
Modeling Clinical Thrombosis ............................................................ 317
Model Based on Ex Vivo Experiments .................................................... 318
Model Development .......................................................................... 318
A Model of Occlusion ........................................................................ 320
Future Directions............................................................................. 322
Conclusions ..................................................................................... 322

INTRODUCTION

Nutrient and waste transport throughout the body is the primary function of the
cardiovascular system. The heart serves to pump blood through a sophisticated
network of branching tubes. The flow is not steady but pulsatile. The blood vessels
distribute blood to different organs while maintaining vessel integrity. The arteries
are not inert tubes but adapt to varying flow and pressure conditions by growing
or shrinking to meet changing hemodynamic demands.
It is important to study blood flows during disease as well as under normal
physiologic conditions. The majority of deaths in developed countries are from
cardiovascular diseases. Most cardiovascular diseases are associated with some
form of abnormal blood flow in arteries. This review focuses on some selected
areas of importance to cardiology.

PHYSIOLOGIC ENVIRONMENT

The fluid blood is a complex mixture of semisolid and liquid material. Blood is
composed of cells, proteins, lipoproteins, and ions by which nutrients and wastes
are transported. Red blood cells (RBCs) typically comprise ;40% of blood by
volume. In most arteries, blood behaves in a Newtonian fashion, and the viscosity
can be taken as a constant 4 centipoise (cP) for a normal hematocrit. The non-
Newtonian viscosity is extensively studied in the field of biorheology and has
been reviewed by others (e.g. 21, 89).
Blood flow and pressure are unsteady. The cyclic nature of the heart pump
creates pulsatile conditions in all arteries. The aorta serves as a compliance cham-
ber that provides a reservoir of high pressure during diastole as well as systole.
Flow is zero or even reversed during diastole in some arteries such as the external
 FLUID MECHANICS AND THROMBOSIS 301

carotid, brachial, and femoral arteries. These arteries have a high distal resistance
during rest, and flow is on/off with each cycle. Flow during diastole can also be
high if the downstream resistance is low, as in the internal carotid or the renal
arteries.
Pulsatile flows dominate many of the problems in the cardiovascular system.
The existence of unsteady flow forces the inclusion of a local acceleration term
in most analyses. In contrast to unsteadiness, several features of biological flows
may often be neglected as being of secondary importance for particular situations.
These include vessel wall elasticity, non-Newtonian viscosity, slurry particles in
the fluid, body forces, and temperature. Although each of these factors is present
in physiology, the analysis is greatly simplified if they can be justifiably neglected,
which is the case in most arterial flows.
Biologists are often concerned with the local hemodynamic conditions in a
particular artery or branch. The important fluid mechanic parameter is often a de-
tailed local description of the fluid-wall shear stress in a blood vessel for a given
pulsatile flow situation. The three-dimensional nature of many of these unsteady
flows has provided an important challenge to computational methods, because
the computational time required is enormous.
The arterial system is tortuous and must branch many times to reach an end
organ. The cross-sectional area along the axis may enlarge at branch points,
sinuses, and aneurysms. However, if the area diverges, the flow must decelerate,
and an adverse pressure gradient can exist. In this situation, flow separation is
possible and typically occurs along the walls of the sinus.
As blood flows across the endothelium, a shear stress is generated to retard
the flow. The wall shear stress is proportional to the shear rate ċ (velocity gradient)
at the wall, and the fluid dynamic viscosity l: s 4 lċ. Shear stress for laminar
steady flow in a straight tube is
s 4 32lqp11D13,
where q is volume flow rate, and D is tube diameter. This approximation is a
reasonable estimate of the mean wall shear stress in arteries. For situations in
which the lumen is not circular or the blood flow is highly skewed, as it is at
branch points, shear stress must be determined by detailed measurements of veloc-
ity near the wall. Shear stress is not easily measured for pulsatile flows. The
velocity and velocity gradient must be measured very close to a wall, which is
technically difficult. The gradient will depend highly on the shape of the velocity
profile and the accurate measurement of distance from the wall. For blood flow,
the viscosity very near a wall is not precisely known because the red cell con-
centration is reduced. Thus, arterial wall shear stress measurements are estimates
and may have errors of 20%–50%.
At the lumenal surface, shear stress can be sensed directly as a force on an
endothelial cell. In contrast, cells cannot sense flow rates directly. Determination
of the flow rate would require knowledge of blood velocities far away from cells
in the artery wall, as well as some way to integrate the velocities to give the
302 WOOTTON n KU

volume flow rate. Thus, it is natural for endothelial cells to sense and respond to
shear stress.
Arteries will typically adapt to maintain a wall shear stress of ;15 dyn/cm2
(41). This appears to be true for different arteries within an animal, between
animal species, as well as after large changes within a single artery. The blood-
wall shear stress modulates diameter adaptive responses, intimal thickening, and
platelet thrombosis. The wall shear stress is thus central to the vascular response
to hemodynamics.
The other major hemodynamic force on an artery is the transmural pressure
across the thickness of the wall. Arteries have a mean pressure of ;100 mmHg,
whereas veins have pressures of ;10 mmHg. The hoop stress can be estimated
by Laplace’s Law as
r 4 0.5PDt11,
where t is wall thickness, D is vessel inner diameter, and P is transmural pressure,
for vessels with circular lumens that are not too thick (38). It is possible that the
primary determinant of smooth muscle cell response is the local strain of these
cells. The arterial wall may remodel in response to both static and cyclical loading
conditions by secretion and organization of collagen and elastin, respectively (88).

FLOWS IN SPECIFIC ARTERIES

There are four major arteries that are subject to the most clinical disease. These
include the carotid bifurcation, the abdominal aorta, the left coronary artery, and
the heart and proximal aorta.

The Carotid Arteries

The carotid arteries are located along the sides of the neck. These arteries supply
the brain and face with blood. Atherosclerosis, which develops right at the bifur-
cation, causes the majority of strokes in patients. The branch is unique in that
there is an anatomic sinus or expansion at the origin of the internal carotid. The
mean Reynolds number is ;300, and the Womersley parameter is ;4. The daugh-
ter branches are ;258 off-axis of the parent artery, on average.
Measurements of velocity have been made in machined plastic models of this
bifurcation by laser Doppler anemometry (65). Secondary flows are produced
downstream of the bifurcation (Figure 1). Velocity profiles obtained by laser
Doppler anemometry and computational fluid dynamics quantify the extent of
reverse velocities at the outer wall of the internal carotid sinus (Figure 2). A
region of transient flow separation is created along the posterior wall of the carotid
sinus, which is prominent during the downstroke of systole. The artery wall in
the sinus region would experience oscillations in near-wall velocity and a low
mean wall shear stress. Atherosclerotic plaque is highly localized to a small area
 FLUID MECHANICS AND THROMBOSIS 303

FIGURE 1 Hydrogen bubble visualiza-

tion of flow through a model carotid bifur-
cation illustrating the laminar flow at the
flow divider and separation of flow at the
posterior wall of the internal carotid sinus.
The separation region of transient reverse
velocities is also the site of secondary vor-
tex patterns. (Reprinted with permission of
the American Heart Association, Inc.)

within this sinus region and correlates with low wall shear stress with coefficients
greater than 0.9, p , 0.001. Comparison of the unsteady, three-dimensional in
vitro results against in vivo measurements with Doppler ultrasound confirms that
the assumptions of the modeling are valid (66). Several groups have recently used
computational fluid dynamics to study the effects of wall elasticity and non-
Newtonian viscosity (4, 86). These effects are small in comparison with the ana-
tomic and flow variations between patients (79, 83).

The Aorta
The aorta is the large vessel from the heart that traverses the middle of the abdo-
men and bifurcates into two arteries supplying the legs with blood. The renal
arteries have a low resistance so that two-thirds of the entering flow leaves the
abdominal aorta through these branches at the diaphragm. Curiously, atheroscle-
rotic disease extends along the posterior wall of the relatively straight abdominal
aorta downstream of the renal arteries in all people. Little disease is ever present
in the upstream thoracic aorta.
In vitro measurements in a glass-blown aorta model show that outflow con-
ditions combine with curvature to create an oscillation in velocity direction at the
posterior wall of the aorta, with a corresponding low average wall shear stress
(77). The area of low wall shear stress correlates very well with the location of
304 WOOTTON n KU

FIGURE 2 a. Axial velocity

profiles in the sinus region of a
three-dimensional model of the
carotid bifurcation, using laser
Doppler anemometry (LDA) and
computational fluid dynamics
(CFD). b. Flow in the carotid
sinus is unsteady with a transient
reverse flow at the outer wall
shown in this three-dimensional
plot of velocity vs diameter posi-
tion and time. (Reprinted from 65
with permission from Elsevier
Science, Ltd.)

atherosclerotic plaque measured in autopsy specimens, p , 0.001 (35, 77). As

verification, measurements of in vivo flow in humans exhibit the same skewing
and time-varying velocity profiles as are produced in the model (76).

Flow at the Left Coronary Artery Bifurcation

Flow at the left coronary artery bifurcation is complicated by several important
features (10). First, the left main coronary artery is quite short, leading to an
entrance type flow at a small Womersley parameter of 3. Second, the flow wave-
form in the left coronary artery is reversed in comparison with that in most arter-
 FLUID MECHANICS AND THROMBOSIS 305

ies, having more flow during diastole. Flow can be reversed during systole. The
high pressure in the myocardium during systolic contraction causes the blood
flow to reverse direction in the coronary arteries. Third, the bifurcation does not
lie in a single plane but curves around the heart while branching. The curvatures
likely set up secondary flows during part of the cardiac cycle. The actual fluid
dynamics have been characterized with large-scale experimental models (103)
and spectral-element computational modeling (50, 51). Comparison of the flow
field with maps of atherosclerotic disease locations yields a strong correlation
between oscillations in shear stress and probability of plaque (r . 0.85, p ,
0.001) (51). Surprisingly, variations in branch angle do not alter the overall flow
field regimes in a dramatic way (50). However, changes in the coronary flow
waveform affect the magnitudes of oscillation significantly (50).

Flows in the Heart and Great Vessels

Flows in the heart and great vessels are dominated by inertial forces over viscous
forces. Reynolds numbers at peak systole are ;4000. The flow in the aorta and
pulmonary trunk is similar to an entrance-type flow, which is not developed.
Consequently, the core of the flow can be considered as an inviscid region away
from a developing boundary layer at the wall. The pressure and velocity patterns
in a complex chamber of the heart can be modeled in three dimensions, even
including a moving boundary condition that develops tension (80, 113).
The analysis of hemodynamics in this representative set of arteries enables
one to develop a general understanding of the fluid mechanics in the normal
cardiovascular system. It should be remembered that arteries are not fixed tubes.
They are biological organs, which remodel over time.

BIOLOGICAL RESPONSES TO HEMODYNAMICS

The artery reacts to the dynamic changes in mechanical stress. Several physiologic
responses are essential to the maintenance of normal functioning of the circulatory
system. The responses of arteries to the hemodynamic environment may create
normal adaptation or pathological disease.
Hemostasis is the arrest of bleeding. Trauma is a common occurrence, and the
body must be able to deal with this possibility. In this hemodynamic environment
of high shear stresses, hemostasis is maintained primarily by platelet adherence
and activation. Platelets pass quickly over the injury site, and adherence must
occur in milliseconds.
On a longer time scale, an artery can respond to minute-to-minute changes in
hemodynamics. The blood vessels must adapt to differing physiologic demands
and conditions from changes in blood pressure and flow. This response is typically
governed by the need to control systemic vascular resistance, venous pooling,
and intravascular blood volume.
306 WOOTTON n KU

Arteries adapt to long-term increases or decreases in wall shear stress. The

response to increased wall shear stress is to vasodilate and then remodel to a
larger diameter with the same arterial structure. This situation is commonly seen
with the creation of an arteriovenous fistula for hemodialysis access. Decreased
flow rates will cause a thickening of the intimal layer to reestablish a normal wall
shear stress (41). Eventually, the artery may maintain a thickened intima or
remodel to a normal artery of smaller diameter.
On an even longer time scale of weeks to months, arteries will remodel their
intima and media layers. The medial thickness is influenced by the local amount
of hoop stress and nutrition. As described above, as the blood pressure increases,
the hoop stress will proportionally increase (22). Because the formation of a
lamellar unit requires the proliferation of smooth muscle cells and the creation of
a highly organized extracellular structure, the process may take several days.
Alterations in the pulsatile pressure lead to changes in organization of the elastin
and collagen structure within the media (41, 88).
The effects of flow, shear stress, and stretch on arteries in vivo have been
studied by several groups. Flow can be augmented through an artery by the crea-
tion of an arteriovenous fistula. Such increased flow causes a dilation of the artery
until the wall shear stress reaches the baseline level of the artery (60, 116). This
baseline appears to be ;15–20 dyn/cm2 for most arteries in a wide range of
species (41). Conversely, restricted flow through an artery produces a smaller-
diameter vessel (68).
Several pathological states may arise from an excessive or uncontrolled
response to a hemodynamic stimulus. Long-term hypertension produces a gen-
eralized medial thickening of blood vessels. Studies of intimal hyperplasia in a
canine model clearly indicate that low shear stresses can accelerate intimal thick-
ening. Shear stress can also be varied in a single artery by using tapered vascular
grafts with differing diameters. In this case, intimal thickening follows from low
shear stresses even for a constant flow rate as depicted in Figure 3a (94).
Atherosclerotic disease forms over decades. Atherosclerosis is highly localized
to only a few places in the systemic vasculature. The primary locations of ath-
eroma are at the carotid artery sinus, the coronary arteries, the abdominal aorta,
and the superficial femoral arteries. In each of these arteries, there are localized
sites where the mean wall shear stress is very low and oscillates between positive
and negative directions during the cardiac cycle. Comparisons of the sites of
disease with the local hemodynamic conditions reveal a consistent curve where
low wall shear stress is strongly correlated with atherosclerotic intimal thickening
(Figure 3b) (51, 67, 77). Typically, most intimal thickening is found where the
average wall shear stress is , 10 dyn/cm2 and follows the curve shape shown for
intimal hyperplasia and arterial adaptation. Thus the biological pattern of arterial
response to shear stress appears to be consistent and preprogrammed.
Currently, a field of cellular and tissue engineering is developing that attempts
to subject cultured cells and tissues to well-defined stresses in an in vitro envi-
 FLUID MECHANICS AND THROMBOSIS 307

FIGURE 3 a. Neointi-
mal hyperplasia thickening
vs wall shear stress in a
dog arterial graft. The
inverse relationship indi-
cates more thickening at
low shear stresses. b. Ath-
erosclerotic intimal thick-
ening vs wall shear stress
in human carotid arteries.
The reciprocal relationship
holds for mean and maxi-
mum wall shear stresses
and correlates directly
with oscillatory shear
stress.

ronment. The creation of flow chambers that recreate physiologically realistic in

vivo stresses is an important area of research (53, 75).
The effects of hemodynamics on convective mass transfer should not be
neglected. Most biologically active molecules are convected from one site to
another. These molecules may be nutrients, wastes, growth factors, or vasoactive
compounds. Systemic hormones reach an artery by convection and then may
diffuse through the wall, with the intima as a major barrier. However, convective
mass transport may be a limiting factor for small molecules such as nitrous oxide
and oxygen, which diffuse rapidly through the wall. Such convection would be
impaired in areas of flow separation or reversing wall shear at sites prone to
atherosclerosis (70, 71). Alternatively, biologically active molecules released by
endothelial cells may have an effect downstream if the molecules are trapped in
a boundary layer near the wall.
308 WOOTTON n KU

HEMODYNAMICS OF STENOSES
When arteries become severely diseased, the arterial lumen becomes restricted
over a short distance of about 1 cm. This constriction is commonly referred to as
a stenosis. An example of an atherosclerotic carotid artery stenosis is depicted in
Figure 4.
In clinical medicine, percent stenosis is commonly defined as percent occlusion
by diameter, as follows:

% stenosis 4 (D11D2)/D1 2 100%,

where D1 is upstream diameter and D2 is the minimum diameter in the stenosis.

As disease advances, the percent stenosis increases.
Stenotic flows have been well characterized by a number of investigators.
Some important summary features are that flow separation (Figure 5b) occurs in
the expansion region at Reynolds numbers of .10 for a 70% stenosis, a strong
shear layer develops between the central jet and the recirculation region, the
critical upstream Reynolds number for turbulence is ;300 (114), turbulence
intensity levels reach up to 100% of the upstream velocity values, and the tur-
bulence is high for ;1.5–6 diameters downstream (69).
For stenoses .75%, flow is limited severely by two mechanisms. Intense tur-
bulence downstream of the stenosis creates large pressure losses. In addition, low
pressure at the stenosis throat, owing to a Bernoulli-type pressure drop, can cause
local collapse in severe stenoses.

FIGURE 4 X-ray contrast angiogram of a dis-

eased carotid bifurcation illustrating the focal
nature of a stenosis. The stenosis (arrow) will
reduce blood flow and pressure to the brain. (From
Strandness DE and van Breda A, 1994. Vascular
Diseases: Surgical and Interventional Therapy.
Reprinted with permission of Churchill Living-
stone Inc.)
 FLUID MECHANICS AND THROMBOSIS 309

FIGURE 5 Steady flow

through a moderate steno-
sis (50% diameter reduc-
tion, 1.2 cm long, 4-mm
diameter, Re 4 160) (98).
a. Stenosis configuration.
b. Streamlines show sepa-
ration distal to the throat.
c. Wall shear rate. Shear
rate increases sharply in
the converging stenosis,
reaching a peak just
upstream of the throat.
Shear rate is negative and
low in the post-stenotic
recirculation region.

Two important clinical consequences arise from the collapse of stenoses. One
is that the flow rate can be limited by choking, beyond that of purely turbulent
losses. This flow limitation or critical flow rate has long been observed by phys-
iologists and described as the coronary flow reserve that is limited even with
decreases in distal resistance. Estimates of coronary flow reserve should include
this choking flow limitation as well as other forms of viscous losses (46, 95). A
second consequence is that of the imposed loading conditions on an atheroscle-
rotic plaque. Stenotic flow collapse creates a compressive stress that may buckle
the structure. The oscillations in compressive loading may induce a fracture
fatigue in the surface of the atheroma, causing a rupture of the plaque cap.
Because plaque cap rupture is the precipitating event in most heart attacks and
strokes, the fluid-solid mechanical interactions present in high-grade stenosis may
contribute to the catastrophic material failure (74).

Diagnosis of Disease
There are a wide variety of clinical applications for hemodynamic studies of
stenoses. One area of investigation revolves around the diagnosis of severe ste-
nosis. The most accepted clinical predictors of impending heart attack, stroke,
and lower-limb ischemia are based on the presence of hemodynamically signifi-
cant stenoses. Currently, the best indicator for surgical treatment of arterioscle-
rosis is the degree of stenosis. Although X-ray angiography is currently the
standard, cost and morbidity are distinct disadvantages.
Doppler ultrasound can be used to measure the increased velocities in the
stenotic jet and back out a percent stenosis. This technique is widely used to
determine levels of stenosis in carotid artery disease, with an accuracy of 90%.
Doppler ultrasound can also be used to measure the flow waveform in the leg
310 WOOTTON n KU

arteries. Normal arteries have a characteristic triphasic pattern, whereas diseased

arteries with a stenosis exhibit a blunted monophasic pattern.
Recently, magnetic resonance imaging (MRI) has been proposed as a less
expensive, less morbid alternative to X-ray angiography (115). In contrast to
Doppler techniques, which require an acoustic or optical window, MRI uses an
electromagnetic window that does not contact the flow. Thus, much more of the
body can be studied.

Shear-Dependent Thrombosis
Stenotic flows become critical to clinical medicine in the acute symptoms of
atherosclerosis. After the plaque cap ruptures, the revealed contents of the ath-
eroma stimulate a blood-clotting reaction called thrombosis. For the arterial sys-
tem, thrombosis is initiated by the adherence of platelets at the surface with rapid
accumulation of additional platelets. Although a number of confusing in vitro
experiments are described in the literature, studies with nonanticoagulated blood
through stenoses indicate that platelets stick at the throat of the stenosis. The
adherence and accumulation of these platelets are shear dependent—with more
accumulation at higher shear rates. The time scale of adhesion is on the order of
milliseconds. Likewise, the adhesion strength must be enormous because the shear
stresses on the platelet are large and increasing as the throat fills with clot. The
following sections explore some of the relationships between thrombosis and
hemodynamics and how these relationships may be used to understand the risk
of clinical thrombosis.

ARTERIAL THROMBOSIS
Thrombosis is the formation of a blood clot, called a thrombus, inside a living
blood vessel. The mechanisms of thrombosis are identical to the mechanisms of
hemostasis, the clotting system that protects the body from excessive blood loss.
A thrombus is composed primarily of two blood cell types, platelets and RBCs.
The cells are bound together by molecules in the cell membrane of the platelets,
called membrane glycoproteins (GPs), by a variety of plasma proteins, and by a
network of polymerized plasma protein called fibrin.
Arterial thrombosis is an extremely significant health problem because it is
linked to the onset of acute clinical symptoms in atherosclerosis. Thrombus super-
imposed on ruptured atherosclerotic plaque is commonly found in autopsy studies
of heart disease (24–27). Thrombosis is also associated with carotid artery plaque
rupture in stroke and transient ischemic attack (24, 85). Platelets and fibrin emboli
are frequently found in the myocardium (heart muscle) of victims of heart disease
(28, 37). Clinical studies confirm the link between thrombosis and atherosclero-
sis—antithrombotic drugs significantly reduce the risk of clinical ischemia (40,
81, 108).
 FLUID MECHANICS AND THROMBOSIS 311

Cellular and Molecular Mechanisms of Thrombosis

Thrombosis is a complicated interaction of platelets and plasma proteins. At a
functional level acute thrombosis is described by three platelet functions (adhe-
sion, activation, and aggregation) and the coagulation cascade (Figure 6). These
mechanisms can occur simultaneously and have multiple interactions, with the
enzyme thrombin playing a central role.

Adhesion Thrombosis is triggered when a thrombogenic surface is exposed to

blood (Figure 6a). Thrombogenic surfaces include most artificial surfaces, the
subendothelial and medial layers of blood vessels, and subendothelial components
of atherosclerotic lesion such as fibrous plaque cap and atheromatous core (30).
Platelets adhere to proteins in the surface via platelet membrane GPs (62).
Subendothelial tissue and atheroma contain collagen, to which platelets bind via
glycoprotein GPIa/IIa (91), and von Willebrand factor (vWF), to which platelets
bind via two GPs. GPIb mediates a rapid but transient binding to vWF, whereas
GPIIb/IIIa mediates more permanent binding (96). GPIIb/IIIa can bind to many
other plasma and vessel wall proteins, including fibrinogen, fibrin, fibronectin,
thrombospondin, and vitronectin (62).

Activation Activation refers to platelet functions triggered by chemical or physi-

cal agonists (stimuli). Chemical agonists include ADP, thrombin, thromboxane
A2 (TxA2), fibrillar collagen, platelet-activating factor, and serotonin (23). Shear
stress (in the presence of vWF, ADP, and Ca2`) can activate platelets (52). Plate-
lets may also be activated by biomaterials via the complement system (39, 59).
Perhaps the most important activation function is a conformational change in
GPIIb/IIIa that allows it to bind to plasma proteins. GPIIb/IIIa activation has been
estimated to occur within 0.1 s (84) and is required for aggregation and permanent
adhesion to vWF (96).

FIGURE 6 Thrombosis in late-stage ath-

erosclerosis. a. Plaque rupture exposes sub-
endothelium to the blood, causing platelet
adhesion. b. Platelet aggregation forms a
platelet plug. c. Coagulation and platelet
aggregation may cause occlusion.
312 WOOTTON n KU

Activation causes shape change with pseudopod extension, which increases

the strength of adhesion and may decrease the resistance of platelets to aggre-
gation. Activation triggers upregulation and local clustering of GPIIb/IIIa, which
may also strengthen adhesion. Activated platelets contract, consolidating loose
cells and fibrin into compacted thrombus, and release granular contents (23).
Several activation functions are positive feedback mechanisms for activation
of other platelets (23). Activated platelets synthesize platelet agonist TxA2 and
release ADP from dense granules; inhibition of either the ADP (87) or TxA2 (64)
activation pathway significantly reduces thrombus growth. Activated platelets
also catalyze thrombin production (23).

Aggregation Aggregation is essential to formation of a platelet plug (Figure

6b). Aggregation can proceed via several mechanisms. At low to moderate shear
rates, activated platelets can bind to other activated platelets via fibrinogen or
fibrin and GPIIb/IIIa. At higher shear rates, platelets aggregate primarily via vWF
(3, 58, 73). It is not clear whether activation occurs before or after initial vWF
binding. Incoming platelets may be activated by passing through an agonist
‘‘cloud’’ of thrombin, TxA2, or ADP before interacting with an adherent throm-
bus (57, 84). Alternatively, vWF on the surface of aggregated and activated plate-
lets could support binding of unactivated platelets via GPIb, followed by
activation and permanent binding via GPIIb/IIIa.

Coagulation Exposure of a thrombogenic surface is also likely to trigger the

coagulation cascade, leading to fibrin coagulation (Figure 6c). In normal hemo-
stasis, injury exposes tissue factor, which rapidly leads to thrombin generation.
Tissue factor is found at high concentrations in the necrotic core of the atheroma
(110) and may be exposed by plaque rupture. Coagulation may also be triggered
by exposure of collagen or an artificial surface (23).
The ultimate reaction in the coagulation cascade converts prothrombin to
thrombin. Thrombin cleaves fibrinogen so that it can polymerize to form fibrin,
which traps red cells in the clot and supports platelet adhesion. Thrombin is also
one of the most potent platelet agonists, causing activation, release of granular
contents, and irreversible aggregation (23). This interaction between platelets and
thrombin is important in thrombosis lasting longer than 10 or 15 min (48, 49,
61).

HEMODYNAMICS AND THROMBOSIS

Thrombosis is fundamentally linked to hemodynamics because blood transports
cells and proteins to the thrombus and applies stresses that may disrupt the throm-
bus. In this section, we review how blood flow conditions affect the rate and
localization of platelet accumulation, platelet activation, and fibrin coagulation.
 FLUID MECHANICS AND THROMBOSIS 313

Hemodynamics in Advanced Atherosclerosis

Most thrombosis experiments with controlled flow report the wall shear rate,
ċ 4 s/l. In major arteries subject to occlusive clinical thrombosis (Table 1), mean
shear rate normally ranges from 200 to 500 s11 and mean flow Reynolds numbers
range from 100 to 400. Where atherosclerosis creates a stenosis, shear rate
increases to a peak just upstream of the stenosis throat (Figure 5c). The wall shear
rate may be estimated by using scaling based on the Reynolds number and geom-
etry (99). Peak shear rate increases with Reynolds number and stenosis severity,
to ;10,000 s11 for moderate stenoses and ;100,000 s11 for severe stenoses.
Distal to the stenosis, a recirculation region may develop, with unusually high
residence time and low shear rate.

Hemodynamics and Thrombus Composition

The composition of a thrombus depends on local flow conditions. In static and
low-shear recirculating flow, the bulk of a thrombus consists of RBCs trapped in
fibrin. But in unidirectional flow at shear rates of 100 s11 and higher, the bulk
of an acute thrombus consists of platelets (18, 92, 101). At arterial and stenotic
shear rates, mechanisms of platelet adhesion, activation, and aggregation domi-
nate, and the thrombus size can be estimated by counting the number of platelets
that accumulate.

Shear and Platelet Accumulation

Platelet Accumulation Rate Increases with Shear Rate An increase in platelet
accumulation is directly related to the shear rate. This has been observed in vitro
for platelet deposition on subendothelium (106) and collagen-coated surfaces (3,
8, 93, 100). The effect of shear rate has also been demonstrated in human (11,
92), baboon (72), and porcine (8) ex vivo experiments, which are dominated by
the aggregation phase of thrombosis. The rate of platelet accumulation on fibrillar

TABLE 1 Mean blood flow parameters for human arteries commonly subject to occlusive
thrombosis in atherosclerosis

Mean Mean wall Mean wall

Diameter Average flow Reynolds shear shear stressa
Vessel (reference) (mm) rate (ml/s) number ratea (s!1) (dyne/cm2)
Femoral artery (44) 5.0 3.7 280 300 11
Common carotid (65) 5.9 5.1 330 250 8.9
Internal carotid (65) 6.1 5.0 220 220 8
Left main coronary (51) 4.0 2.9 240 460 16
Right coronary (50) 3.4 1.7 150 440 15
a
Mean wall shear rate and shear stress are estimated from the Poiseuille profile.
314 WOOTTON n KU

collagen increases for shear rates from 100 to at least 10,000 s11 (Figure 7). At
low shear rates, accumulation is roughly proportional to shear rate. At higher
shear rates, there may be a divergence from this trend. One experiment shows a
decrease in deposition rate between 10,000 and 32,000 s11 (11), whereas another
experiment shows an increase in deposition rate between 4,300 and 20,000 s11
(72).

Platelets Adhere Preferentially in High-Shear Regions Shear also appears to

affect where platelets are deposited. Platelet accumulation on collagen-containing
stenotic surfaces is highest at the stenosis throat, where shear rate is highest (8,
72), for peak shear rates ranging from 1,300 to .20,000 s11 (72). On smooth
artificial surfaces by contrast, platelet accumulation may be depressed in high
shear regions (15, 97).

Shear-Linked Mechanisms
The correlations between shear and platelet accumulation may be explained in
terms of several shear-linked mechanisms: platelet transport, platelet activation,
and embolization.

Transport Platelet transport is important in acute thrombosis because platelet

accumulation on highly thrombogenic surfaces in vivo may be transport limited
or transport modulated for shear rates up to at least 20,000 s11 (111). Transport

FIGURE 7 Average platelet accumulation rate in ex vivo baboon (72), pig (8), and
human (11, 92) experiments, as a function of peak wall shear rate. Platelet accumulation
rate on collagen I is averaged over 15 min, measured on tubes (●) and stenoses (2) (72),
and in U-channels (n) (8). Platelet accumulation rate on collagen III over 5 min (m,n)
(11, 92), estimated from thrombus volume by a linear correlation of data published for the
same system (93), platelets/thrombus volume 4 9 2 1010 platelets/ml.
 FLUID MECHANICS AND THROMBOSIS 315

in blood is still not completely understood, partly because there is no fundamental

theory to predict dispersion in a concentrated suspension like blood. But exper-
imental studies have identified two mechanisms that influence the rate of platelet
interaction with a thrombogenic surface: (a) RBC motion, the dominant mecha-
nism, increases small-scale transport by several orders of magnitude (44); (b)
nonuniform platelet concentration may increase platelet transport by a factor of
1–10. Both of these mechanisms increase platelet transport as shear rate increases.

RBC motion RBCs exhibit randomlike transverse motion and rotation in shear
flow (43), which displaces plasma and platelets and increases lateral transport.
The rate of platelet transport has often been quantified by an effective diffusivity,
derived by fitting experimental data to a species transport model of platelet adhe-
sion (e.g. 106). Power law correlations in the form De 4 D1 (ċ/ċ1)n, where ċ is
the shear rate and ċ 4 1 s11) give power n and coefficient D1 that are functions
of hematocrit and the stiffness and size of the RBCs. For platelets or chemical
solutes in anticoagulated human blood at 40% hematocrit, n ranges from 0.49 to
0.89, and D1 ranges from 1019 to 3 2 1018 cm2/s (1, 5, 107, 109). From these
correlations, De ranges from 2 2 1018 to 3 2 1017 cm2s11 at shear rate 4 100
s11 and from 5 2 1017 to 1 2 1015 cm2s11 at shear rate 4 10,000 s11, 1 to
4 orders of magnitude above the thermal diffusivity for platelets in plasma (1.6
2 1019 cm2s11).
Estimates of De vary by up to an order of magnitude between experiments.
One source of variation may be the variability of platelet adhesion rates with
differences in anticoagulation and platelet handling; the adhesion rate begins to
affect the deposition rate in vitro for shear rates . 300 s11 (107). This difficulty
can be avoided by using a global model of enhanced diffusivity in sheared con-
centrated suspensions (117). Assuming that RBC rotation is relatively unimpor-
tant, the effective diffusivity De is the sum of the RBC dispersion and the thermal
diffusivity:
De 4 DR ` Ds (1)
where DR is the RBC dispersion coefficient and Ds is the thermal diffusivity of
the solute or platelets in the stationary blood. The dispersion coefficient for RBCs
is correlated to experiments by
DR 4 a2ċcfp(11fp)m (2)
with c 4 0.15 5 0.03 and m 4 0.8 5 0.3, where a is the RBC major radius, ċ
is the shear rate, and up is the hematocrit. For platelets, De is essentially propor-
tional to ċ for ċ . 10 s11. The model is consistent with transport rates for a
variety of solutes in whole blood and was later confirmed for macromolecule
transport (63).

Nonuniform Concentration RBCs are concentrated in the center of a blood

vessel, and appear to force increased platelet concentration toward the vessel wall.
316 WOOTTON n KU

This effect has been studied most heavily in narrow vessels (e.g. 43, 104) but has
also been observed in 3-mm-diameter tubes at arterial and higher shear rates (2).
Platelet concentration at the wall increases with increasing hematocrit, shear rate,
and platelet concentration. For example, in a 3-mm tube with a 40% hematocrit
and a 0.25 billion/ml average platelet count, near-wall concentration is a factor
of 2 to 4 higher than the average platelet count as the shear rate increases from
240 s11 to 1200 s11 (2).
Both RBC motion and enhanced platelet concentration link high shear to
increased platelet deposition. As long as molecular mechanisms of adhesion and
aggregation are rapid enough to permit platelet incorporation into a thrombus,
increasing shear will drive more platelets into the thrombus, resulting in more
rapid thrombus growth.

Activation The role of shear stress activation in clinical thrombosis is not clear.
The threshold shear exposure required for platelet activation in vitro has been
measured for shear rate (in whole blood) ranging from ;103 to 107 s11 (52) and
fit to a platelet stimulation function, PSF (16), such that PSF 4 s t 0.452, where
s is shear stress in dynes per square centimeter and t is exposure time in seconds.
The threshold for shear-induced activation is PSF $ 1000 (16). High shear stress
activates platelets with short exposure, whereas lower shear stress activates plate-
lets over longer durations.
A platelet flowing through a stenosis in vivo is exposed to high shear stress,
but the exposure time is at least one order of magnitude lower than the threshold
for shear-induced platelet activation (16). Shear stress exposure may not be
directly responsible for platelet activation in most cases of relatively severe ath-
erosclerosis, if activation is required for the initial interaction between a circu-
lating platelet and growing thrombus. Shear stress exposure time will exceed the
activation threshold only if a platelet adheres to a stenosis.
Even if shear stress is not the sole activating agonist in vivo, the history of
shear stress exposure may change the threshold of platelet activation by chemical
agonists (42, 45). Compared with flow that has a gradually changing shear rate,
stenotic flow with a rapid increase in shear stress may significantly increase plate-
let activation and platelet deposition (54, 111, 112).

Embolization Another feature of thrombosis is embolization, the removal of

parts of the thrombus owing to fluid mechanical stress. A theoretical model has
been developed for embolization in steady and pulsatile flows (14), based on
models of drag on a protrusion into steady (12) and pulsatile (13) flow. The stress
on the thrombus depends on the particle Reynolds number, Rep 4 ċHp2/t, where
Hp is the thrombus height and t is the kinematic viscosity. For small thrombi
(Rep , 1), stress on an isolated thrombus is four- to fivefold the wall shear stress
of the approaching flow. For larger thrombi, stress becomes a function of throm-
bus height, and stress increases rapidly as the thrombus grows.
 FLUID MECHANICS AND THROMBOSIS 317

The missing part of the model is quantitative data on the stress required for
platelet removal from a surface. Mechanical properties of platelets are the subject
of ongoing study (47), so the critical stress for embolization may soon be within
reach, using a combination of modeling and experiments.
Differences in platelet embolization stress may explain the difference between
platelet accumulation patterns on collagen (72) or damaged artery (7) and accu-
mulation on Lexan (97). Platelets probably adhere more strongly to collagen in
the natural surfaces than to the smooth Lexan surface and can support larger
thrombi without embolization. Ultrasound measurements of embolization from
knitted Dacron or collagen surfaces in ex vivo experiments show low emboliza-
tion rates (111).

Recirculation and Residence Time The effect of hemodynamics on thrombosis

is well documented in uniform or unidirectional shear flow. But separated flow
occurs at bifurcations, and downstream of stenoses that occur in atherosclerosis.
In regions of complex flow, the relationships between flow and thrombosis are
not very clear.
Convection patterns and high residence times may modulate thrombosis in
separated flows. In vitro experiments show increased platelet accumulation near
the reattachment point in Lexan stenoses, presumably caused by increased con-
vection (15). Platelets may recirculate in the separated region long enough to
become activated and form small aggregates. Residence time and convection
patterns have also been related to fibrin polymerization in shear flow (36, 82).
Based on steady-flow experiments, residence time on the order of at least 10
s is required for significant shear-induced aggregation (56) or fibrin polymer
accumulation (82). In physiologic pulsatile flow, 10-s residence is quite long; for
example, .99% of particles are washed out of the recirculation zone of a 75%
or 95% area reduction stenosis within 10 s (19). One potential location for phys-
iologic high residence time would be along the trailing edge of a sharp geometric
flow separator, which could be created by a tear or flap following plaque rupture,
or by a poorly designed prosthetic valve. A sudden expansion, which has a geo-
metric flow separator, creates an environment favoring a fibrin-rich red thrombus
(18). High residence time could also occur distal to a flow-limiting acute platelet
plug, in which case occlusion becomes the primary cause of high residence time
and fibrin coagulation.

MODELING CLINICAL THROMBOSIS

Despite the well-documented role of thrombosis in clinical ischemia, thrombosis
risk is not used as a surgical indicator in atherosclerosis. Stenosis severity is used
to identify patients who are good candidates for surgical treatment because it
correlates with risk of ischemia (6, 20, 78). The relationships between shear rate,
stenotic flow, and thrombus growth rate are probably reflected in the clinical
318 WOOTTON n KU

statistics. However, using only stenosis severity misses patients with moderate or
mild stenoses (,50% diameter reduction), who have a significant risk of ischemic
attack and death (17, 25, 105).
Clinically it is important to know the likelihood that thrombosis will lead to
occlusion following plaque rupture or ulceration. A model of occlusion risk could
be combined with a model of plaque rupture risk to decide which patients are
good candidates for surgical treatment and which patients can be managed med-
ically. A clinical thrombosis model has not been developed yet, owing to the
complexity of thrombosis and the wide range of data produced by different throm-
bosis experiments. But there is enough experimental data available to begin build-
ing a model, based on a theoretical mechanistic thrombosis model. The model
can estimate the time required for a thrombus to occlude a vessel, based on
hemodynamics and geometry, and the occlusion time can be used as a measure
of risk of occlusion in the event of plaque rupture.

Model Based on Ex Vivo Experiments

Only a few experiments are representative of clinical occlusive thrombosis. In
vivo and ex vivo experiments, which use nonanticoagulated blood or mildly hep-
arinized blood, include all of the major mechanisms of thrombosis and can occur
over time periods long enough for occlusion of a major artery. In contrast, in
vitro experiments require anticoagulation and often involve extensive blood sam-
ple manipulation, and platelet deposition rates are much lower in in vitro than in
ex vivo experiments (9, 30, 92). For model development, ex vivo experiments
have an advantage over in vivo experiments; precise control of flow rate and
thrombogenic surface geometry allows the shear rate to be calculated.
Relevant thrombosis experiments need to match the hemodynamic environ-
ment of stenotic arteries subject to clinical occlusive thrombosis. The controlling
hemodynamic variable is the shear rate on the thrombogenic surface, influenced
by the shear rate history of platelets flowing over the thrombogenic surface in
stenotic flow. The Reynolds number does not appear to be significant for forward
flow, apart from its direct effect on the shear rate.
Several experiments approximate the hemodynamic and hematologic environ-
ment expected in atherosclerosis in vivo. Baboon (72) and porcine (7, 8, 30) ex
vivo experiments span physiologic shear rates and approach a physiologic Rey-
nolds number, using little or no anticoagulation; the baboon experiments also
have well-characterized stenotic flow (72). Human ex vivo experiments (11, 92)
match physiologic shear rates without anticoagulation, although the duration of
the experiments is limited. These experiments can be used to guide and test a
model of occlusive thrombosis. Well-characterized in vivo canine experiments
can also be used to test an occlusion model (101).

Model Development
Several experiments provide insight into occlusion when platelets may adhere to
the entire lumen surface, a relatively severe injury. In stenotic geometry, the
 FLUID MECHANICS AND THROMBOSIS 319

stenosis throat is the location of most rapid platelet accumulation and of occlusion
(72, 101). For 4-mm–inside-diameter stenoses at a 100-ml/min flow rate, occlu-
sion occurs for smaller lumen sizes (,2.7 to 3 mm) and for higher shear rates
(.600 s11) (72). Occlusion occurs consistently and rapidly for narrower lumens
and higher shear rates (33, 101).
A theoretical model can help scale experimental data to other flow conditions.
Occlusion can be estimated by predicting the size of the thrombus, which is
proportional to the number of accumulated platelets, because platelets comprise
the bulk of the thrombus. The time course of platelet accumulation in ex vivo
experiments (72) is dominated by an acute phase, which eventually decelerates
to a slow phase (Figure 8). In some experiments, a platelet plug occludes the
lumen, slowing flow and platelet accumulation, but in other experiments, the rate
of accumulation is limited by a drop in the aggregation rate. To first order, the
final size of a thrombus is proportional to the acute rate of platelet accumulation
and the duration of the acute phase of platelet deposition.
The first objective of the model is to estimate the acute rate of platelet accu-
mulation, as a function of hematologic and hemodynamic variables. Several good
theoretical thrombosis models have been developed to understand thrombosis
experiments. Some models treat platelets as discrete particles (31, 55, 84, 90,
102); this approach has the potential to be more accurate as molecular models of
adhesion are developed, but can become complicated. Current particle models
idealize or ignore the particle-fluid interactions and thrombus shape or are explor-

FIGURE 8 The characteristic time course of platelet accumulation on collagen-coated

tubes, in ex vivo baboon experiments (m) (72). The experiment can be divided into three
phases: (I) an accelerating phase, lasting for about 5 min, (II) an acute phase, lasting ;50
min, and (III) a slow phase, which extends to the end of the experiment, characterized by
a lower accumulation rate than the acute phase. The total platelet accumulation can be
approximated by a 5-min delay, constant accumulation at the acute rate, and no accumu-
lation during the slow phase (solid line).
320 WOOTTON n KU

atory tools. Other models treat platelets as a continuous chemical species (29, 32,
34, 106, 112) reacting with a reactive surface. Species transport models are quan-
titative and relatively simple but have not been applied to clinical thrombosis and
occlusion.
A modified species transport model has been developed to compute platelet
accumulation rates based on hemodynamics, geometry, platelet count, and aggre-
gation rate (111, 112). Unactivated platelets in the blood are treated as a chemical
species, which is transported by convection and shear-enhanced diffusion (117).
Near-wall platelet concentration is enhanced by a factor of two above average
platelet concentration, consistent with experiments in similar-sized tubes (2).
Platelets at the surface are incorporated into the thrombus by a first-order reaction
step that includes both aggregation and activation. Flow and transport equations
can be solved analytically in tubular flow. For a stenosis, a commercial compu-
tational fluid dynamics package is used to compute the flow field and platelet
accumulation rate. This approach predicts the acute platelet accumulation rate on
collagen-coated tubes and in the upstream, converging, and throat sections of
collagen-coated stenoses (111, 112) of differing stenosis severity (Figure 9c). The
platelet accumulation rate is highest at the stenosis throat and increases with
increasing percent stenosis (Figure 9b). The model is less successful in recircu-
lating post-stenotic flow, but in experiments the maximum platelet deposition rate
is located in the throat section (7, 72), where occlusion occurs (101), so the model
is applicable to predicting occlusion in stenotic flow.

A Model of Occlusion
A model of acute platelet accumulation rate can be used to estimate thrombus
size and occlusion risk if the duration of the acute phase can be predicted. Unfor-
tunately, the mechanisms that are responsible for reducing the accumulation rate
are not well studied. Embolization has been assumed an important limiting mech-
anism, but embolization loss is difficult to measure, and large emboli appear to
be infrequent in ex vivo experiments (111). A model of embolization has been
derived (14), but the embolization stress is unknown. In addition, systemic
changes may reduce the rate of platelet activation, or the concentration of platelet
activation agonists may decrease locally as the thrombus size increases.
Absent a clear mechanism to limit thrombus growth, the occlusion time can
be estimated from the acute accumulation rate and lumen diameter, assuming that
the acute phase does not end. This extrapolated occlusion time can be used as a
risk indicator; a short occlusion time indicates a higher risk of occlusion when
there is plaque rupture.
For a fully reactive surface, occlusion occurs when the thrombus height reaches
the lumen radius. The occlusion time is

DlumenCth
Tocclusion 4 ` td , (3)
2fjlumen
 FLUID MECHANICS AND THROMBOSIS 321

FIGURE 9 Model of platelet accumulation on 4-mm-inner-diameter collagen-coated

stenoses (111). a. Collagen-coated surface consists of straight segment from x 4 11.2
cm to 10.6 cm, a cosine-shaped stenosis from 10.6 cm to 0.6 cm, and a straight segment
from 0.6 cm to 1.2 cm. b. Acute platelet accumulation rate (j*) vs axial location (x) in
50%, 75%, and 90% area reduction stenoses. j* peaks just upstream of the throat. Peak
accumulation rate increases with increasing stenosis severity. c. Average platelet accu-
mulation in the stenosis throat section (x 4 10.48 cm to x 4 0.48 cm), for 50% (m),
75% (n), and 90% (l) area reduction (72), compared with model (lines).
322 WOOTTON n KU

where Dlumen is the diameter of the vessel lumen (throat diameter in a stenosis),
Cth is the concentration of platelets in arterial thrombus [estimated to be 75 billion/
ml from ex vivo experiments (111)], and f is the ratio of thrombus height to
thrombus cross-section area, which accounts for the roughness of the thrombus
(estimated to be ;2 based on experimental occlusion times in stenoses). A 5-min
delay (td) is used to model the effect of the accelerating phase of thrombosis. The
acute rate of platelet accumulation jlumen is computed by using the species trans-
port model. Equation 3 estimates occlusion times of 16, 28, and 66 min for 90%,
75%, and 50% area reduction stenoses, respectively. In experiments, occlusion
times were 18–25 min for the 90% stenosis and 25–35 min for the 75% stenoses
(72), relatively close to the model. The 50% stenosis does not occlude, so an
occlusion time somewhere between 30 and 60 min indicates a low risk of occlu-
sive thrombosis.
The model predicts increased risk of occlusion (decreasing occlusion time)
with increasing shear rate, decreasing lumen diameter, and increasing platelet
count. Because shear rate increases and lumen diameter decreases with increasing
percent stenosis, the correlation is consistent with clinical studies linking risk of
ischemia and benefit of surgery with percent stenosis. Platelet count is a hema-
tologic parameter that should also have a strong influence on risk of occlusion,
based on this model.

Future Directions
The knowledge that shear affects platelets is already being applied to the design
of cardiovascular devices, to minimize shear stress and residence time in blood
pumps, cardiopulmonary bypass devices, and prosthetic valves.
Clinical application of an occlusive thrombosis model depends on a better
understanding of mechanisms that limit thrombus growth after the acute aggre-
gation phase that is typically observed. Embolization and systemic negative feed-
back may contribute to subocclusive thrombus under some flow conditions. A
second requirement for an occlusive thrombosis model is a risk model for plaque
rupture. Combined understanding of plaque rupture and thrombosis, along with
measurements of degree of stenosis, could increase the accuracy of screening
patients for surgical treatment of atherosclerosis.

CONCLUSIONS

The study of hemodynamics is a rich field that allows one to characterize the
biological responses to mechanical forces. Specific arteries exhibit flow charac-
teristics that are three-dimensional and developing. Diseased arteries can create
high levels of turbulence, head loss, and a choked flow condition in tubes that
can collapse. The pulsatile nature of the flow creates a dynamic environment with
many interesting fundamental fluid mechanics questions. The fundamental knowl-
edge can be used to predict and change blood flow to alter the course of disease.
 FLUID MECHANICS AND THROMBOSIS 323

Shear stress and shear rate have emerged as important parameters that modulate
both chronic and acute biological responses.
The relationships between thrombosis and fluid mechanics are complicated. A
species transport model can be used to estimate clinical thrombosis risk based on
the hemodynamic environment. Future studies will be driven by the need to
understand the complex effect of hemodynamics on cells and the design of new
devices to modulate this effect.

Visit the Annual Reviews home page at http://www.AnnualReviews.org.

LITERATURE CITED
1. Aarts PAMM, Steendijk P, Sixma JJ, vessel wall. Experimental study in the
Heethaar RM. 1986. Fluid shear as a pig model. J. Clin. Invest. 84(4):1134–
possible mechanism for platelet diffusiv- 44
ity in flowing blood. J. Biomech. 8. Badimon L, Badimon JJ, Galvez A,
19(10):799–805 Chesebro JH, Fuster V. 1986. Influence
2. Aarts PAMM, van den Broek SA, Prins of arterial damage and wall shear rate on
GW, Kuiken GDC, Sixma JJ, Heethaar platelet deposition. Ex vivo study in a
RM. 1988. Blood platelets are concen- swine model. Arteriosclerosis 6:312–20
trated near the wall and red blood cells, 9. Badimon L, Badimon JJ, Lassila R,
in the center in flowing blood. Arterio- Heras M, Chesebro JH, et al. 1991.
sclerosis 8(6):819–24 Thrombin regulation of platelet interac-
3. Alevriadou BR, Moake JL, Turner NA, tion with damaged vessel wall and iso-
Ruggeri ZM, Folie BJ, et al. 1993. Real- lated collagen type I at arterial flow
time analysis of shear-dependent throm- conditions in a porcine model: effects of
bus formation and its blockade by hirudins, heparin, and calcium chelation.
inhibitors of von Willebrand factor bind- Blood 78(2):423–34
ing to platelets. Blood 81(5):1263–76 10. Bargeron CB, Deters OJ, Mark FF,
4. Anayiotos AS, Jones SA, Giddens DP, Friedman MH. 1988. Effect of flow
Glagov S, Zarins CK. 1994. Shear stress partition on wall shear in a cast of a
at a compliant model of the human human coronary artery. Cardiovasc. Res.
carotid bifurcation. J. Biomech. Eng. 22(5):340–44
116:98–106 11. Barstad RM, Kierulf P, Sakariassen KS.
5. Antonini G, Guiffant G, Quemada D, 1996. Collagen induced thrombus for-
Dosne AM. 1978. Estimation of platelet mation at the apex of eccentric sten-
diffusivity in flowing blood. Biorheology oses—a time course study with
15:111–17 non-anticoagulated human blood.
6. Asymptomatic Carotid Atherosclerosis Thromb. Haemost. 75(4):685–92
Study (ACAS). 1995. Endarterectomy 12. Basmadjian D. 1984. The hemodynamic
for asymptomatic carotid artery stenosis. and embolizing forces acting on thrombi,
JAMA 273(18):1421–28 from incipient attachment of single cells
7. Badimon L, Badimon JJ. 1989. Mecha- to maturity and embolization. J. Bio-
nisms of arterial thrombosis in nonpar- mech. 17(4):287–98
allel streamlines: Platelet thrombi grow 13. Basmadjian D. 1986. The hemodynamic
on the apex of stenotic severely injured and embolizing forces acting on
324 WOOTTON n KU

thrombi—II. The effect of pulsatile lamellar unit revisited. Arteriosclerosis

blood flow. J. Biomech. 19(10):837–45 5:19–24
14. Basmadjian D. 1989. Embolization: criti- 23. Colman RW. 1993. Mechanisms of
cal thrombus height, shear rates, and pul- thrombus formation and dissolution.
satility. Patency of blood vessels. J. Cardiovasc. Pathol. 2(3):S23–32
Biomed. Mater. Res. 23(11):1315–26 24. Constantinides P. 1990. Cause of throm-
15. Bluestein D, Niu L, Schoephoerster RT, bosis in human atherosclerotic arteries.
Dewanjee MK. 1997. Fluid mechanics of Am. J. Cardiol. 66(16):G37G–40
arterial stenosis: relationship to the 25. Davies MJ. 1990. A macro and micro
development of mural thrombus. Ann. view of coronary vascular insult in ische-
Biomed. Eng. 25:344–56 mic heart disease. Circulation 82(Suppl.
16. Boreda R, Fatemi RS, Rittgers SE. 1995. 3):II38–46
Potential for platelet stimulation in criti- 26. Davies MJ, Thomas A. 1984. Thrombo-
cally stenosed carotid and coronary arter- sis and acute coronary-artery lesions in
ies. J. Vasc. Invest. 1(1):26–37 sudden cardiac ischemic death. N. Engl.
17. Brown BG, Gallery CA, Badger RS, J. Med. 310(18):1137–40
Kennedy JW, Mathey D, et al. 1986. 27. Davies MJ, Thomas AC. 1985. Plaque
Incomplete lysis of thrombus in the mod- fissuring—the cause of acute myocar-
erate underlying atherosclerotic lesion dial infarction, sudden ischaemic death,
and crescendo angina. Br. Heart J.
during intracoronary infusion of strepto-
53(4):363–73
kinase for acute myocardial infarction:
28. Davies MJ, Thomas AC, Knapman PA,
quantitative angiographic observations.
Hangartner JR. 1986. Intramyocardial
Circulation 73(4):653–61
platelet aggregation in patients with
18. Cadroy Y, Horbett TA, Hanson SR.
unstable angina suffering sudden ische-
1989. Discrimination between platelet-
mic cardiac death. Circulation 73(3):
mediated and coagulation-mediated
418–27
mechanisms in a model of complex
29. Eckstein EC, Belgacem F. 1991. Model
thrombus formation in vivo. J. Lab. Clin.
of platelet transport in flowing blood
Med. 113(4):436–48 with drift and diffusion terms. Biophys.
19. Cao J, Rittgers SE. 1998. Particle motion J. 60(1):53–69
within in vitro models of stenosed inter- 30. Fernandez-Ortiz A, Badimon JJ, Falk E,
nal carotid and left anterior descending Fuster V, Meyer B, et al. 1994. Charac-
coronary arteries. Ann. Biomed. Eng. terization of the relative thrombogenic-
26(2):190–99 ity of atherosclerotic plaque compo-
20. Chaitman BR, Fisher LD, Bourassa MG, nents: implications for consequences of
Davis K, Rogers WJ, et al. 1981. Effect plaque rupture. J. Am. Coll. Cardiol.
of coronary bypass surgery on survival 23(7):1562–69
patterns in subsets of patients with left 31. Fiechter J, Ku DN. 1998. Numerical
main coronary artery disease. Report study of platelet transport in flowing
of Collaborative Study in Coronary blood. Proc. ASME Biomed. Eng. Div.,
Artery Surgery (CASS). Am. J. Cardiol. NYC. Adv. Bioeng. 39:11–12, Anaheim,
48(4):765–77 Ca
21. Chien S. 1970. Shear dependence of 32. Fogelson AL. 1992. Continuum models
effective cell volume as a determinant of of platelet aggregation: formulation and
blood viscosity. Science 168:977 mechanical properties. SIAM J. Appl.
22. Clark JM, Glagov S. 1985. Transmural Math. 52(4):1089–110
organization of the arterial wall: the 33. Folts JD, Crowell EB Jr, Rowe GG.
 FLUID MECHANICS AND THROMBOSIS 325

1976. Platelet aggregation in partially dependent effects. Thromb. Haemost.

obstructed vessels and its elimination 71(1):78–90
with aspirin. Circulation 54(3):365–70 43. Goldsmith HL, Marlow JC. 1979. Flow
34. Friedman LI, Liem H, Grabowski EF, behavior of erythrocytes. II. Particle
Leonard EF, McCord CW. 1970. Incon- motions in concentrated suspensions of
sequentiality of surface properties for ini- ghost cells. J. Colloid Interface Sci.
tial platelet adhesion. Trans. Am. Soc. 71(2):383–407
Artif. Intern. Organs 16:63–73 44. Goldsmith HL, Turitto VT. 1986. Rheo-
35. Friedman MH, Bargeron CB, Hutchin- logical aspects of thrombosis and hae-
son GM, Mark FF, Deters OJ. 1980. mostasis: basic principles and
Hemodynamic measurements in human applications. ICTH-Report—Subcom-
arterial casts and their correlation with mittee on Rheology of the International
histology and luminal area. J. Biomech. Committee on Thrombosis and Haemos-
Eng. 102:247 tasis. Thromb. Haemost. 55(3):415–35
36. Friedrich P, Reininger AJ. 1995. Occlu- 45. Goldsmith HL, Yu SS, Marlow J. 1975.
sive thrombus formation on indwelling Fluid mechanical stress and the platelet.
catheters: in vitro investigation and com- Thromb. Diath. Haemorrh. 34(1):32–41
putational analysis. Thromb. Haemost. 46. Gould KL. 1978. Pressure-flow charac-
73(1):66–72 teristics of coronary stenoses in unse-
37. Frink RJ, Rooney PA Jr, Trowbridge JO, dated dogs at rest and during coronary
Rose JP. 1988. Coronary thrombosis and vasodilation, Circ. Res. 43(2):242–53
platelet/fibrin microemboli in death asso- 47. Haga JH, Beaudoin AJ, White JG, Strony
ciated with acute myocardial infarction. J. 1998. Quantification of the passive
Br. Heart J. 59(2):196–200 mechanical properties of the resting
38. Fung YC. 1984. Biodynamics: Circula- platelet. Ann. Biomed. Eng. 26(2):268–
tion. New York: Springer-Verlag. 404 77
pp. 48. Hanson SR, Harker LA. 1988. Interrup-
39. Gemmell CH, Black JP, Yeo EL, Sefton tion of acute platelet-dependent throm-
MV. 1996. Material-induced up-regula- bosis by the synthetic antithrombin
tion of leukocyte CD11b during whole D-phenylalanyl-L-prolyl-L-arginyl chlo-
blood contact: material differences and a romethyl ketone. Proc. Natl. Acad. Sci.
role for complement. J. Biomed. Mater. USA 85(9):3184–88
Res. 32(1):29–35 49. Hanson SR, Markou C, Lindahl AK, Ku
40. Gent M, Blakely JA, Easton JD, Ellis DN, Scarborough RM, et al. 1993. Inhi-
DJ, Hachinski VC, et al. 1989. The bition of in vivo thrombus formation in
Canadian American Ticlopidine Study an arterial stenosis model. Thromb. Hae-
(CATS) in thromboembolic stroke. Lan- most. 69:540 (Abstr.)
cet 1(8649):1215–20 50. He X. 1993. Numerical simulations of
41. Glagov S, Zarins C, Giddens DP, Ku blood flow in human coronary arteries.
DN. 1988. Hemodynamics and athero- PhD thesis. Georgia Inst. Technol.,
sclerosis: insights and perspectives Atlanta. 249 pp.
gained from studies of human arteries. 51. He X, Ku DN. 1996. Pulsatile flow in the
Arch. Pathol. Lab. Med. 112:1018–31 human left coronary artery bifurcation:
42. Goldsmith HL, Frojmovic MM, Braovac average conditions. J. Biomech. Eng.
S, McIntosh F, Wong T. 1994. Adeno- 118:74–82
sine diphosphate-induced aggregation of 52. Hellums JD. 1994. 1993 Whitaker lec-
human platelets in flow through tubes. ture: biorheology in thrombosis research.
III. Shear and extrinsic fibrinogen- Ann. Biomed. Eng. 22:445–55
326 WOOTTON n KU

53. Helmlinger G, Geiger RV, Schreck S, cellular interactions. Annu. Rev. Cell
Nerem RM. 1991. Effects of pulsatile Biol. 6:329–57
flow on cultured vascular endothelial cell 63. Kim D, Beissinger RL. 1993. Aug-
morphology. J. Biomech. Eng. 113:123– mented mass transport of macromole-
31 cules in sheared suspensions to surfaces.
54. Holme PA, Orvim U, Hamers MJ, J. Colloid Interface Sci. 159:9–20
Solum, NO, Brosstad FR, et al. 1997. 64. Kotze HF, Lamprecht S, van Badenhorst
Shear-induced platelet activation and PN, vanWyk V, Roodt JP, Alexander K.
platelet microparticle formation at blood 1993. In vivo inhibition of acute platelet-
flow conditions as in arteries with a dependent thrombosis in a baboon model
severe stenosis. Arterioscler. Thromb. by Bay U3405, a thromboxane A2-
Vasc. Biol. 17(4):646–53 receptor antagonist. Thromb. Haemost.
55. Huang PY, Hellums JD. 1993. Aggre- 70(4):672–75
gation and disaggregation kinetics of 65. Ku DN, Giddens DP. 1987. Laser Dopp-
human blood platelets: Part I. Develop- ler anemometer measurements of pulsa-
ment and validation of a population bal- tile flow in a model carotid bifurcation.
ance model. Biophys. J. 65(1):334–43 J. Biomech. Eng. 20:407
56. Huang PY, Hellums JD. 1993. Aggre- 66. Ku DN, Phillips DJ, Giddens DP, Strand-
gation and disaggregation kinetics of ness DE. 1985. Hemodynamics of the
human blood platelets: Part II. Shear- normal human carotid bifurcation: in
vitro and in vivo studies. Ultrasound
induced platelet aggregation. Biophys. J.
Med. Biol. 11:13–26
65(1):344–53
67. Ku DN, Zarins CK, Giddens DP, Glagov
57. Hubbell JA, McIntire LV. 1986. Platelet
S. 1985. Pulsatile flow and atheroscle-
active concentration profiles near grow-
rosis in the human carotid bifurcation:
ing thrombi—a mathematical considera-
positive correlation between plaque
tion. Biophys. J. 50:937–45
localization and low and oscillating shear
58. Ikeda T, Handa M, Kawano K, Kamata
stress. Arteriosclerosis 5:292–302
T, Murata M, et al. 1991. The role of von
68. Langille BL, O’Donnell F. 1986. Reduc-
Willebrand factor and fibrinogen in tion in arterial diameter produced by
platelet aggregation under varying shear chronic decreases in blood flow are endo-
stress. J. Clin. Invest. 87:1234–40 thelium-dependent. Science 231:405–7
59. Kalman PG, McCullough DA, Ward CA. 69. Lieber BB, Giddens DP. 1990. Post-ste-
1990. Evacuation of microscopic air bub- notic core flow and its effects on wall
bles from Dacron reduces complement shear stress. J Biomech. 23(6):597–605
activation and platelet aggregation. J. 70. Lutostansky EM. 1996. The role of con-
Vasc. Surg. 11(4):591–98 vective mass transfer in atherosclerosis.
60. Kamiya A, Togawa T. 1980. Adaptive PhD thesis. Georgia Inst. Technol.
regulation of wall shear stress to flow Atlanta. 195 pp.
change in the canine carotid artery. Am. 71. Ma P, Li X, Ku DN. 1994. Heat and mass
J. Physiol. 239:H14–21 transfer in a separated flow region for
61. Kelly AB, Marzec UM, Krupski W, Bass high Prandtl and Schmidt numbers under
A, Cadroy Y, et al. 1991. Hirudin inter- pulsatile conditions. Int. J. Heat Mass
ruption of heparin-resistant arterial Transf. 37(17):2723–36
thrombus formation in baboons. Blood 72. Markou CP, Hanson SR, Siegel JM, Ku
77(5):1006–12 DN. 1993. The role of high wall shear
62. Kieffer N, Phillips DR. 1990. Platelet rate of thrombus formation in stenoses.
membrane glycoproteins: functions in Proc. ASME Biomed. Eng. Div., NYC.
 FLUID MECHANICS AND THROMBOSIS 327

Adv. Bioeng. 26:555–58, New Orleans, Physicians’ Health Study. Steering Com-
La mittee of the Physicians’ Health Study
73. Markou CP, Lindahl AK, Siegel JM, Ku Research Group. N. Engl. J. Med.
DN, Hanson SR. 1993. Effect of block- 321(3):129–35
ing the platelet GPIb interaction with von 82. Reininger AJ, Reininger CB, Heinzmann
Willebrand factor under a range of shear- U, Wurzinger LJ. 1995. Residence time
ing forces. Ann. Biomed. Eng. 21:220 in niches of stagnant flow determines
(Abstr.) fibrin clot formation in an arterial
74. McKinsey J, McCord BN, Aoki T, Ku branching model—detailed flow analysis
DN. 1991. Can mechanical stress cause and experimental results. Thromb. Hae-
fatigue of the atherosclerotic plaque? most. 74(3):916–22
Surg. Forum 42:319 83. Reuderink PJ. 1991. Analysis of the flow
75. Moore JE Jr, Burki E, Suciu A, Zhao in a 3D distensible model of the carotid
SM, Burnier M, et al. 1994. Device for artery bifurcation. PhD thesis. Tech-
subjecting vascular endothelial cells to nische Univ. Eindhoven, Netherlands
both fluid shear stress and circumferen- 84. Richardson PD. 1973. Effect of blood
tial cyclic stretch. Ann. Biomed. Eng. flow velocity on growth rate of platelet
22:416–22 thrombi. Nature 245(5420):103–4
76. Moore JE Jr, Maier SE, Ku DN, Boesiger 85. Ricotta JJ, Schenk EA, Ekholm SE,
P. 1994. Hemodynamics in the abdomi- DeWeese JA. 1986. Angiographic and
nal aorta: a comparison of in vitro and pathologic correlates in carotid artery
in vivo measurements. J. Appl. Physiol. disease. Surgery 99(3):284–92
76:1520–27 86. Rindt CC, van Steenhoven AA, Janssen
77. Moore JE Jr, Xu C, Glagov S, Zarins CK, JD, Reneman RS, Segal A. 1990.
Ku DN. 1994. Fluid wall shear stress A numerical analysis of steady flow
measurements in a model of the human in a three-dimensional model of the
abdominal aorta: oscillatory behavior carotid artery bifurcation. J. Biomech.
and relationship to atherosclerosis. Ath- 23(5):461–73
erosclerosis 110(2):225–40 87. Roald HE, Barstad RM, Kierulf P, Skjor-
78. North American Symptomatic Carotid ten F, Dickinson JP, et al. 1994. Clopi-
Endarterectomy Trial Collaborators dogrel—a platelet inhibitor which
(NASCET). 1991. Beneficial effect of inhibits thrombogenesis in non-anticoa-
carotid endarterectomy in symptomatic gulated human blood independently of
patients with high-grade carotid stenosis. the blood flow conditions. Thromb. Hae-
N. Engl. J. Med. 325:445–53 most. 71(5):655–62
79. Perktold K Rappitch G. 1995. Computer 88. Rodbard S. 1970. Negative feedback
simulation of local blood flow and vessel mechanisms in the architecture and func-
mechanics in a compliant carotid artery tion of the connective and cardiovascular
bifurcation model. J. Biomech. 28:845– tissues. Perspect. Biol. Med. 13:507–27
56 89. Rodkiewicz CM, Sinha P, Kennedy JS.
80. Peskin CS McQueen DM. 1989. A three- 1990. On the application of a constitutive
dimensional computational method for equation for whole blood. J. Biomech.
blood flow in the heart. I. Immersed elas- Eng. 112:198–206
tic fibers in a viscous incompressible 90. Ruckenstein E, Marmur A, Gill WN.
fluid. J. Comput. Phys. 81:372–405 1977. Growth kinetics of platelet
81. Physicians’ Health Study Research thrombi. J. Theor. Biol. 66(1):147–68
Group (PHSRG). 1989. Final report on 91. Saelman EU, Nieuwenhuis HK, Hese
the aspirin component of the ongoing KM, de Groot PG, Heijnen HFG, et al.
328 WOOTTON n KU

1994. Platelet adhesion to collagen types man B. 1993. Analysis of shear stress
I through VIII under conditions of stasis and hemodynamic factors in a model of
and flow is mediated by GPIa/IIa (alpha coronary artery stenosis and thrombosis.
2 beta 1-integrin). Blood 83(5):1244–50 Am. J. Physiol. 265(5):H1787–96
92. Sakariassen KS, Joss R, Muggli R, Kuhn 102. Tandon P, Diamond SL. 1997. Hydro-
H, Tschopp TB, Sage H, Baumgartner dynamic effects and receptor interactions
HR. 1990. Collagen type III induced ex of platelets and their aggregates in linear
vivo thrombogenesis in humans: role of shear flow. Biophys. J. 73(5):2819–35
platelets and leukocytes in deposition of 103. Tang TD. 1990. Periodic flow in a bifur-
fibrin. Arteriosclerosis 10(2):276–84 cating tube at moderate Reynolds num-
93. Sakariassen KS, Kuhn H, Muggli R, ber. PhD thesis. Georgia Inst. Technol.,
Baumgartner HR. 1988. Growth and sta- Atlanta. 211 pp.
bility of thrombi in flowing citrated 104. Tilles AW, Eckstein EC. 1987. The near-
blood: assessment of platelet-surface wall excess of platelet-sized particles in
interactions with computer-assisted mor- blood flow: its dependence on hematocrit
phometry. Thromb. Haemost. 60(3):392– and wall shear rate. Microvasc. Res.
98 33(2):211–23
94. Salam TA, Lumsden AB, Suggs WD, Ku 105. Torvik A, Svindland A, Lindboe CF.
DN. 1996. Low shear stress promotes 1989. Pathogenesis of carotid thrombo-
intimal hyperplasia thickening. J. Vasc. sis. Stroke 20(11):1477–83
Invest. 2:12–22 106. Turitto VT, Baumgartner HR. 1975.
95. Santamore WP, Bove AA, Carey RA. Platelet deposition on subendothelium
1982. Tachycardia induced reduction in exposed to flowing blood: mathematical
coronary blood flow distal to a stenosis. analysis of physical parameters. Trans.
Int. J. Cardiol. 2:23–27 Am. Soc. Artif. Intern. Organs 21:593–
96. Savage B, Saldivar E, Ruggeri ZM. 601
1996. Initiation of platelet adhesion by 107. Turitto VT, Weiss HJ, Baumgartner HR.
arrest onto fibrinogen or translocation on 1980. The effect of shear rate on platelet
von Willebrand factor. Cell 84(2):289– interaction with subendothelium exposed
97 to citrated human blood. Microvasc. Res.
97. Schoephoerster RT, Oynes F, Nunez G, 19(3):352–65
Kapadvanjwala M, Dewanjee MK. 1993. 108. United Kingdom Transient Ischaemic
Effects of local geometry and fluid Attack Study Group (UK-TIA). 1988.
dynamics on regional platelet deposition United Kingdom transient ischaemic
on artificial surfaces. Arterioscler. attack (UK-TIA) aspirin trial: interim
Thromb. 13(12):1806–13 results. UK-TIA Study Group. Br. Med.
98. Siegel JM. 1992. Wall shear stress J. Clin. Res. Ed. 296(6618):316–20
through an arterial stenosis and its impli- 109. Wang NHL, Keller KH. 1985. Aug-
cations to thrombosis. MS thesis. Geor- mented transport of extracellular solutes
gia Inst. Technol., Atlanta. 88 pp. in concentrated erythrocyte suspensions
99. Siegel JM, Markou CP, Ku DN, Hanson in Couette flow. J. Colloid Interface Sci.
SR. 1994. A scaling law for wall shear 103(1):210–25
stress through an arterial stenosis. J. Bio- 110. Wilcox JN, Smith KM, Schwartz SM,
mech. Eng. 116:446–51 Gordon D. 1989. Localization of tissue
100. Sixma JJ, de Groot PG. 1994. Regulation factor in the normal vessel wall and in
of platelet adhesion to the vessel wall. the atherosclerotic plaque. Proc. Natl.
Ann. NY Acad. Sci. 714:190–99 Acad. Sci. USA 86(8):2839–43
101. Strony J, Beaudoin A, Brands D, Adel- 111. Wootton DM. 1998. Mechanistic mod-
 FLUID MECHANICS AND THROMBOSIS 329

eling of occlusive arterial thrombosis. arterial stenoses. J. Biomech. Eng. 101:

PhD thesis. Georgia Inst. Technol., 157–75
Atlantic. 421 pp. 115. Yucel EK. 1994. Magnetic Resonance
112. Wootton DM, Markou CP, Hanson SR, Angiography in Vascular Diseases: Sur-
Ku DN. 1998. Mechanistic model of gical and Interventional Therapy, ed DE
arterial thrombosis in collagen-coated Strandness Jr, pp. 289–302. New York:
stenoses. Proc. ASME Biomed. Eng. Churchill Livingstone
Div., NYC. Adv. Bioeng. 39:113–14, 116. Zarins CK, Zatina MA, Giddens DP, Ku
Anaheim, Ca DN, Glagov S. 1987. Shear stress regu-
113. Yoganathan AP, Lemmon JD Jr, Kim lation of artery lumen diameter in ex-
YH, Walker PG, Levine RA, Vesier CC. perimental atherogenesis. J. Vasc. Surg.
1994. A computational study of a thin- 5:413–20
walled three-dimensional left ventricle 117. Zydney AL, Colton CK. 1988. Aug-
during early systole. J. Biomech. Eng. mented solute transport in the shear flow
116:307–17 of a concentrated suspension. Physico-
114. Young DF. 1979. Fluid mechanics of Chem. Hydrodyn. 10:77–96