R T Schoephoerster, F Oynes, G Nunez, M Kapadvanjwala and M K Dewanjee

surfaces.
Effects of local geometry and fluid dynamics on regional platelet deposition on artificial
Print ISSN: 1079-5642. Online ISSN: 1524-4636
Copyright 1993 American Heart Association, Inc. All rights reserved.
Avenue, Dallas, TX 75231
is published by the American Heart Association, 7272 Greenville Arteriosclerosis, Thrombosis, and Vascular Biology
doi: 10.1161/01.ATV.13.12.1806
1993;13:1806-1813 Arterioscler Thromb Vasc Biol.
http://atvb.ahajournals.org/content/13/12/1806
World Wide Web at:
The online version of this article, along with updated information and services, is located on the

http://atvb.ahajournals.org//subscriptions/
at:
is online Arteriosclerosis, Thrombosis, and Vascular Biology Information about subscribing to Subscriptions:

http://www.lww.com/reprints
Information about reprints can be found online at: Reprints:

document. Permissions and Rights Question and Answer Further information about this process is available in the
is being requested is located, click Request Permissions in the middle column of the Web page under Services.
Clearance Center, not the Editorial Office. Once the online version of the published article for which permission
can be obtained via RightsLink, a service of the Copyright Arteriosclerosis, Thrombosis, and Vascular Biology
Requests for permissions to reproduce figures, tables, or portions of articles originally published in Permissions:
by guest on August 14, 2014 http://atvb.ahajournals.org/ Downloaded from by guest on August 14, 2014 http://atvb.ahajournals.org/ Downloaded from
1806
Effects of Local Geometry and Fluid Dynamics
on Regional Platelet Deposition
on Artificial Surfaces
Richard T. Schoephoerster, Finn Oynes, German Nunez,
Monsoor Kapadvanjwala, Mrinal K. Dewanjee
An important aspect of blood-material interactions is the activation, adhesion, and subsequent aggrega-
tion of blood platelets on the artificial surface, all of which are directly affected by local fluid dynamics.
The objective of this work was to directly correlate changing local fluid dynamic conditions produced by
various vessel geometries, including stenosis, aneurysm, and separate contraction and expansion
geometries, with quantitative in vitro measurements of regional platelet deposition. We directly measured
platelet deposition as a function of axial position along four Lexan flow chambers with axisymmetric
models of these geometries using '"In-labeled platelets. Platelet deposition was maximum in observed
areas of flow recirculation and reattachment and minimum in locations of high shear and separation. For
the stenosis geometry, two distinct regions of increased platelet deposition were apparent, one proximal
to and one distal to the stenosis throat. An approximately linear increase in platelet densities was
produced in the aneurysm region, increasing in the direction of flow. Through a comparison of platelet
deposition with local fluid streamline orientation, we have shown that platelet deposition is increased in
certain areas due to the enhanced convective transport of platelets and blood cells to the vessel wall along
locally curved streamlines with velocity components perpendicular to the vessel wall. (Arterioscler Thromb.
1993;13:1806-1813.)
KEY WORDS fluid dynamics regional platelet deposition
stenosis aneurysm flow streamlines '"In platelet labeling
blood-materials interaction
T
hromboembolic complications still constitute a
significant drawback in the use of synthetic
biomaterials in cardiovascular devices such as
artificial heart valves,
12
vascular grafts,
3
and ventricular-
assist devices.
4
It has long been known that the process
of thrombosis may be affected by a series of rheological
and fluid dynamic parameters, including high rates of
shear, areas of flow stagnation or recirculation, and
turbulence.
5
Stein and Sabbah
6
correlate turbulence
with thrombus formation on artificial surfaces, and
Yoganathan et al
7
report thrombus formation and tissue
overgrowth on explanted Bjork-Shiley heart valves in
areas of low shear and stagnation in the minor outflow
region of the valve.
An important aspect of blood-material interactions is
the activation, adhesion, and subsequent aggregation of
blood platelets on the artificial surface. The genesis of
thrombus formation is a cascade phenomenon of bio-
chemical events induced by several clotting factors,
ultimately resulting in the conversion of the soluble
plasma protein fibrinogen to insoluble fibrin. The fibrin
Received June 24, 1993; revision accepted September 10, 1993.
From the Departments of Mechanical Engineering (R.T.S.,
F.O.) and Industrial Engineering (G.N.), Florida International
University, and the Departments of Biomedical Engineering
(M.K.) and Radiology (M.K.D.), Division of Nuclear Medicine,
University of Miami, Miami, Fla.
Correspondence to Richard T. Schoephoerster, PhD, Mechan-
ical Engineering Department, Florida International University,
University Park Campus, Miami, FL 33199.
threads that are formed enmesh the aggregated platelets
and clotting factors to form the thrombus, part of which
may dislodge into the vasculature, inducing ischemic
organ damage by the blockade of an adequate blood
supply. Therefore, there has been much effort in the last
two decades to study the effects of fluid dynamics on
platelet kinetics.
812
This work has included analytic
models of convective-diffusive transport to explain the
migration of red blood cells away from the wall and
increased concentration of platelets near the wall of tube
flow. To date, the vast majority of experimental observa-
tions includes flow visualization using latex beads; con-
centration profiles using freeze-capture methods; or bulk
flow-type effects of high rates of fluid shear, areas of
recirculation, and flow separation and reattachment on
platelet aggregation and adhesion in flow chamber ge-
ometries that deviate greatly from those found in the
circulation. The only direct correlations of local fluid-
dynamic events with platelet adhesion and aggregation
on surfaces
9
" were obtained at extremely low Reynolds
numbers to simulate flow in the microcirculation.
Schmid-Schoenbein et al
13
summarize these observa-
tions and the local fluid dynamic phenomena leading to
the thrombotic process by using flow through a stenosis
as a model. In this model, local increases in shear in the
throat region cause damage to and activation of plate-
lets. Immediately downstream from the stenosis, flow
separating from the throat creates a region of slowly
recirculating flow conducive to aggregation of platelets
by guest on August 14, 2014 http://atvb.ahajournals.org/ Downloaded from
Schoephoerster et al Fluid Dynamics and Platelet Deposition 1807
(a)
0 38 cm
25 cm
(d)
0 95 cm / 0 91
Hi
FIG 1. Diagrams showing geometries of the flow cham-
bers machined from Lexan to simulate the various flow
phenomena, a, Stenosis geometry; b, aneurysm geome-
try; c, contraction/expansion geometry; and d, circular
cylinder geometry (control). These geometries resulted in
a reduction in area (or percent stenosis) for the stenosis
model of 84% and an increase in area for the aneurysm
model of 156%. r(z) indicates the local radius of the tube.
and platelet-activating factors, and further downstream
the main flow reattaches to the tube wall, creating a
point of primary adhesion and subsequent aggregation
of the activated platelets. In this study, we directly
tested this theory by measuring regional platelet depo-
sition as a function of axial position along a model
axisymmetric stenosis using '"In-labeled platelets. We
also measured platelet deposition along an aneurysm
geometry and a flow chamber containing separate con-
traction and expansion geometries. All measurements
were obtained at flow rates ranging from 0.5 to 6.0
L/min to secure the full range of flow phenomena that
implanted cardiovascular devices would create, from
low shear to fully turbulent flow. The overall objective
of the work was to directly correlate changing local
fluid-dynamic conditions produced by these various
vessel geometries and flows with quantitative measure-
ments of regional platelet deposition.
Methods
Regional platelet deposition was determined as a
function of axial distance along the four Lexan flow
chambers shown in Fig 1. A streamlined axisymmetric
stenosis geometry with the form shown was chosen so
comparisons could be made with previous experimental
and numerical flow studies.
1415
The streamlined con-
tractions and expansions in the flow geometry follow a
sinusoidal function of the form
TABLE 1.
Flow
Rate,
L/min
0.5
1.5
3.0
6.0
Flow Parameters
Upstream
Reynolds
Number
300
900
1800
3600
Upstream
Wall Shear
Rate, s"
1
84
252
504
1008
Stenosis Throat
Wall Shear
Rate, s"'
1312
3926
7872
15744
r
(z)= 0.47 0.14 1+cos -
where r(z) is the local radius of the tube and z is
measured in centimeters in the axial direction of flow
from the beginning of the contraction or expansion
region of the tube. The plus sign is used in the expan-
sion region and the minus sign in the contraction region.
A similarly streamlined axisymmetric aneurysm geome-
try served as a model of recirculating flow without the
upstream increases in shear stress, and the contraction/
expansion geometry provided a separation of the com-
bined effects of a gradual contraction and expansion
that form the stenosis geometry. This resulted in a
reduction in area (or percent stenosis) for the stenosis
model of 84% and an increase in area for the aneurysm
model of 156%. The straight-pipe flow chamber served
as a control and allowed the separation of all other
effects on the platelet deposition from those determined
by local geometry and flow conditions.
For each experiment, 100 mL blood was obtained
from the jugular vein of one of two conditioned female
beagle dogs (10 to 12 kg) with two syringes (50 mL
each), each preloaded with 10 mL acid-citrate-dextrose
(ACD) solution in saline for anticoagulation. The plate-
lets were labeled with '"In tropolone by using proce-
dures described in detail by Dewanjee et al
16
and are
only briefly outlined here. Platelet-rich plasma was
separated from blood by centrifugation for 10 minutes
at 180g. The platelet pellet was obtained by centrifuga-
tion of the platelet-rich plasma at lOOOg for 10 minutes.
Next, 250 to 300 ^tCi '"In tropolone was added to the
platelet pellet, which was suspended in 2 mL ACD-
saline. After incubation at room temperature for 30
minutes, unbound '"In was removed by washing with 2
mL ACD-saline (1000g for 10 minutes). In this way, an
efficiency of platelet labeling of 90% was routinely
achieved.
The blood was then warmed to 37C in an incubator
(model 81, Precision Scientific) and introduced into the
flow loop, taking special care to purge the system of any
air bubbles, and was run at a constant flow rate for 30
minutes. Table 1 contains the flow rates and corre-
sponding Reynolds numbers and wall shear rates (based
on Poiseuille flow) used in this study. The flow loop
consisted of the particular flow chamber, a water bath
(model 4061, Buchi) with the temperature adjusted to
maintain blood temperature at 37C, a steady-flow
peristaltic pump (model 7592, Masterflex), and approx-
imately 2 ft of silicone rubber tubing (Fig 2). After each
run the blood was drained from the flow loop and stored
in the water bath for use in later runs. The flow chamber
was removed from the flow loop, drained in the direction
by guest on August 14, 2014 http://atvb.ahajournals.org/ Downloaded from
1808 Arteriosclerosis and Thrombosis Vol 13, No 12 December 1993
37C
Blood Reservoir
Lexan
Flow
Chamber
8000
FIG 2. Schematic of the flow loop, consisting of the
interchangeable Lexan flow chamber, a blood reservoir,
peristaltic pump, water bath adjusted to maintain blood at
37C, and approximately 2 ft of silicone rubber tubing.
of flow for approximately 8 minutes, and gently washed
with 50 mL saline until no visible traces of blood were
observed on the interior of the flow chamber. The
chamber was then disassembled (into two halves) for
wiping and measurement of radioactivity. The interior
surface of each chamber half was wiped along predeter-
mined lengths of sections (1 cm for the constant radius
sections, 0.5 cm for the variable radius sections) with two
saline-dampened cotton swabs per section. The two
swabs were placed in a test tube and their total
radioactivity in ' "In counts per minute was measured
in an automatic gamma-well counter (model 5003,
Cobra II, Packard, Inc) for 1 to 2 minutes. After each
run the flow chambers were cleaned with a biodeter-
gent solution, rinsed thoroughly with distilled water,
and allowed to dry.
Quantification of Platelet Deposition
To compute platelet densities (number of platelets
per unit surface area), the amount of platelet-bound
'"In in whole blood must first be determined. The total
radioactivity in the entire volume of blood was obtained
by averaging the radioactivity as measured with an
ionization chamber before and after each run. A 0.5-mL
blood sample was centrifuged to remove plasma, and
the radioactivity in the plasma was used to determine
the percentage of free '"In. By using that, the percent-
age of radioactivity bound to platelets was determined,
and the platelet-bound '"In cpm per milliliter was
determined by
(2)
Platelet-Bound '"In cpm
mL Whole Blood "
Whole-Blood "'In cpm
mL Whole Blood
x(100-%Free
The platelet count was determined with a Coulter
counter (Coulter S+, Hialeah, Fla), and, from '"In
cpm/mL blood and the platelet count, the number of
platelets per "'In cpm was calculated.
(3)
No. Platelets No. Platelets
1 2
Number of Wipes
FIG 3. Bar graph showing typical platelet densities mea-
sured from individual successive swab wipes along a spec-
ified region of the interior surface of the flow chambers.
mL Whole Blood
Platelet-Bound In cpm
The area of the interior of the test chamber segment
was determined by the diameter at the regional location
and the length of the section. Thus, the number of
platelets deposited per unit area of surface of each
segment of the test chamber along its axis was
calculated.
(4)
No. Platelets 'In cpm
Unit Area (cm
2
) Section Area (cm
2
No. Platelets
'In cpm mL Whole Blood
'"In cpm
These methods and equations have been successfully
applied previously.
31617
-
23
Flow Visualization
In addition to the platelet deposition measurements,
qualitative flow visualization was performed under each
experimental condition by using 400- to 450-/xm-diam-
eter neutrally buoyant resin beads (Amberlite IRA-904,
Rohm and Haas, Philadelphia, Pa) suspended in a
solution of 36% glycerine in saline (to match the density
and viscosity of blood) and illuminated with a sheet of
light that was produced by passing a laser beam (8 mW,
model 1105P, Uniphase He-Ne) through a glass rod to
visualize the axisymmetric flow in a single plane.
Results
Verification of Methods
The effectiveness of the cotton swab wipe method in
obtaining accurate and reliable estimates of the platelet
deposition on the flow chamber walls was verified in the
following manner. Typical platelet densities measured
from successive individual cotton swab wipes (Fig 3)
indicate that the number of platelets obtained from the
third swab wipe was generally 2% to 3% of the first and
second swab wipes combined. However, some platelets
may be lost during the wiping procedure. To estimate
the number of platelets lost, it was necessary to obtain
local radioactivity levels from sections without wiping.
These measurements were obtained in a geometry
similar to the stenosis geometry (so there would be a
variation in platelet density along the axial direction of
the tube) using a 10-in section of silicone tubing made
by guest on August 14, 2014 http://atvb.ahajournals.org/ Downloaded from
Schoephoerster et al Fluid Dynamics and Platelet Deposition 1809
Fl ow .
8. 4 cm 8.4 cm
!N 0.42 cm
silicone collar
tubing
0.95 cm
FIG 4. Diagram showing geometry configura-
tion for a model stenosis, made by placing a
collar over silicone tubing, used for validation of
the methods. Platelet densities were measured
at each of the locations numbered 1 through 6.
At each location platelet densities were esti-
mated from cotton swab radioactivity subse-
quent to double wiping of region, and actual
platelet densities were determined from radioac-
tivity measurements of direct insertion of mea-
surement section in the gamma-well counter.
stenotic by placing a collar, with a smaller inner diam-
eter than the tube, around a portion of the silicone tube
(Fig 4). This section was introduced into the flow loop in
the same location as the flow chambers, and blood with
radiolabeled platelets was allowed to flow through the
tube by using methods described above. After the usual
draining and washing, six 1-cm sections (locations
shown in Fig 4) were carefully cut from the silicone tube
using a razor blade. These sections were then sliced in
half along the axial direction. One half was directly
inserted into a test tube for radioactivity measurement
in the gamma-well counter. The other half was wiped
with two cotton swabs, and their radioactivity was
measured as described above.
As indicated in Table 2, the platelet densities deter-
mined from the cotton swab measurements were on
average 82% of the densities measured from direct
insertion of the specimen. The difference in platelet
densities measured from the wipe compared with the
whole specimen ranged from 7% to 28% in this data set
and can be attributed to platelets lost during the wiping
procedures. If the data are normalized with respect to
the average platelet density over the entire flow section
(Fig 5), the distribution of platelet densities as a func-
tion of axial distance along the tube displays the same
general tendencies for the unwiped whole segments and
those segments measured with the cotton-swab wipes.
The magnitude of platelet densities was higher for the
silicone rubber (Table 2) compared with the Lexan (Fig
3), and these values are slightly lower than those
obtained by other investigators.
1118
However, direct
comparison is misleading due to differences in species,
surface materials, surface preparation, and exposure
time. Since the primary intent of this study was to
TABLE 2. Platelet Densities for the Silicone
Rubber Stenosis
delineate the relative effects of local geometry and flow
conditions on platelet deposition on the wall, the results
for the Lexan flow chambers are shown, with platelet
densities normalized by the spatially averaged platelet
density in the entire flow chamber. The results, then,
are given as normalized platelet density (NPD).
(5) NPD =
Regional Platelet Density Measurement
Mean Platelet Density Over the Entire Flow Chamber
Effects of Geometry on Platelet Deposition
The effects of the local geometric configuration and
the resulting dynamics of the flow on the variation of
NPD is evident in Fig 6. The NPD curves shown are
best-fit lines using data collected from five separate
measurements obtained at 1.5 L/min. No significant
variation in platelet density occurred in the control
(straight tube) chamber (Fig 6a), where the geometry
(radius) was uniform, and thus the local velocity profile
remained relatively undisturbed along the length of the
flow chamber. In general, platelet deposition was max-
imum just distal to the stenosis (Fig 6b), on the down-
stream side of the aneurysm cavity (Fig 6c), and in the
expansion region (Fig 6d), corresponding to observed
areas of flow recirculation and reattachment. NPD was
at a minimum in locations of high shear and separation,
ie, the throat of the stenosis, the leading edge of the
aneurysm, and downstream from the contraction.
A sharp downward spike in relative platelet density
occurred at the throat of the stenosis, where both high
shear stresses and flow separation occurred. The NPD
gradually increased toward the throat in the direction of
1 5
Position
1
2
3
4
5
6
Estimated
Platelet
Density
34 025
23 493
27 322
31 857
20 600
31 520
Actual
Platelet
Density
36 600
27 747
34 100
37 891
28 642
40 985
Estimated/
Actual, %
93
85
80
84
72
77
e
l
e
t

D
e
n
s
i
t
P
l
a
t
i
z
e
d
r
m
a
c
H Estimated platelet density from cotton swab radioactivity
^ Actual platelet density from total specimen radioactivity
Platelet densities are expressed as platelets per centimeter
squared. Position refers to location of measurement as shown in
Fig 4. The techniques used to acquire estimated and actual
platelet densities are described in "Methods."
t 1.0-
3 05-
1 2 3 4 5
Posi t i on
FIG 5. Bar graph showing platelet densities in the sili-
cone stenosis model at each location (position number)
shown in Fig 4 normalized with respect to the average
platelet density over the entire flow section.
by guest on August 14, 2014 http://atvb.ahajournals.org/ Downloaded from
1810 Arteriosclerosis and Thrombosis Vol 13, No 12 December 1993
a
a.
z
2.0
1.5
1.0
0.5
0.0
flow
(a)
2.0
1.5
1.0
0.5
0.0
flow
(c)
Q
z
flow, and it gradually decreased distal to the throat. This
phenomenon of two distinct regions of increased NPD
appeared to be flow dependent, as explained in more
detail below. A slight increase in NPD was also dis-
played upstream from the contraction region (Fig 6d).
The NPD remained low throughout the extended ste-
notic region, and then increased sharply in the region of
expansion. A much sharper increase and decrease in
NPD was observed in comparison with the stenosis. The
aneurysm produced a slight decrease in NPD on the
proximal side, a dramatic increase along the centerline,
and a continued increase of NPD along the entire distal
length of the chamber (compared with the proximal-
side NPD).
Effects of Flow Rate on Platelet Deposition
The regional distribution of platelet deposition also
varied along the axial direction of the stenosis flow
chambers as a function of flow rate and the resulting
changes in shear rates and shear stresses in the fluid.
For the stenosis geometry, two distinct regions of in-
creased platelet deposition were apparent, one proxi-
mal to and one distal to the stenosis throat (Fig 7). The
distal region of increased NPD was very dependent on
the state of the wake region downstream from the
stenosis. For the 0.5-L/min flow rate, a laminar wake
was produced with an axisymmetric region of slowly
recirculating flow that extended approximately 3 diam-
2.0-
1.5-
1.0-
0.5-
nn -
/ <
\
J
(d)
FIG 6. Diagrams showing
normalized platelet density
(NPD) as a function of axial
distance along the flow cham-
ber for the a, circular cylinder
(control geometry); b, steno-
sis; c, aneurysm; and d, con-
traction/expansion geometry.
All measurements were ob-
tained at a flow rate of 1.5
L/min. Data for the control ge-
ometry are the averages and
standard deviations of five
measurements, while the re-
maining curves are best-fit
lines through measured data
points.
eters downstream from the stenosis. The distal region of
increased NPD peaked at about 2 diameters and ex-
tended to about 4 diameters from the throat, somewhat
past the reattachment point of the recirculation region.
At 1.5 L/min the jet was turbulent, so the region of
slowly recirculating flow was disturbed and confined to
approximately 2 diameters downstream. The distal re-
gion of increased NPD was likewise reduced in length,
confined to a region just beyond the region of slowly
recirculating flow. This trend continued for flow rates of
3.0 and 6.0 L/min. The wake became more disturbed,
the region of slowly recirculating flow was reduced, and
the region of increased NPD was likewise reduced. The
proximal region of elevated NPD increased in magni-
tude with flow rate relative to the decreasing number of
platelets in the distal region.
The aneurysm geometry produced less variation in
NPD with flow rate than the stenosis geometry (Fig 8).
In general, an approximately linear increase in NPD
was produced in the aneurysm region, increasing in the
direction of flow. For the 0.5-L/min flow rate, a very
slow moving laminar recirculation zone was confined to
the aneurysm, with the dividing streamline parallel to
the upstream chamber wall. At 1.5 and 3.0 L/min, the
flow in the annular region of the aneurysm was more
disturbed, and the dividing streamline bulged into the
main flow, causing disturbances downstream from the
aneurysm. This may be causative of the slight increase
Q
z
a
a.
Z
FIG 7. Diagrams showing
normalized platelet density
(NPD) as a function of axial
distance along the flow cham-
ber for the stenosis geometry
at a, 0.5 L/min; b, 1.5 L/min; c,
3.0 L/min; and d, 6.0 L/min.
The shaded regions denote
the visualized regions of
slowly recirculating flow for
each case. At 6.0 L/min no
region of slow recirculation
was created due to the estab-
lished turbulence in the flow.
by guest on August 14, 2014 http://atvb.ahajournals.org/ Downloaded from
Schoephoerster et al Fluid Dynamics and Platelet Deposition 1811
2.0
1.5 H
1.0
0.5
0.0
2.0
1.5
1.0
05
0.0
flow ^
y
/
(a)
flow ^^-
I
J
**
r
(b)
2.0-
1.5-
1.0
0.5
0.0
flow-
(d)
FIG 8. Diagrams showing
normalized platelet density
(NPD) as a function of axial
distance along the flow cham-
ber for the aneurysm geome-
try at a, 0.5 L/min; b, 1.5
L/min; c, 3.0 L/min; and d, 6.0
L/min. The aneurysm pro-
vided a suitable region for a
slowly rotating vortex that was
skewed toward the upstream
edge of the aneurysm.
in NPD distal to the aneurysm compared with the
proximal side. At 6 L/min, the flow in the aneurysm was
quite disturbed, which resulted in a dampening of the
effects of the geometry on NPD. At this flow rate, the
NPD slowly increased along the entire length of the
chamber in the direction of flow, with a slight increase
in the center of the aneurysm.
Discussion
NPD remained constant along the straight-tube flow
model, but for geometries that produce flows that
deviate from Poiseuille-type flow, marked variations in
NPD with local changes in geometry and flow condi-
tions were observed. As indicated in theory by Schmid-
Schoenbein et al,
13
NPD was elevated in areas of flow
recirculation and reattachment. Furthermore, the ex-
tent of elevated NPD was directly correlated with the
length of the recirculation zone as it varied with flow
rate for the stenosis geometry.
These studies were performed in vitro by using canine
blood as a model for human blood. Dewanjee et al,
19
who compared dog, pig, and human blood, observed
that dog platelets were five times more thrombogenic
than porcine or human platelets for flow through a
hemodialyzer. However, this property of dog blood
served our experiments well due to the nature of the
methods and the low material-surface contact area with
blood in the flow chambers. The method of using
labeled platelets to quantify NPD did not allow differ-
entiation between adhered single platelets and platelet
aggregates within the cumulative deposit, but it did
provide an accurate and reliable measure of the total
number of adhered platelets. Methods such as electron
microscopy allow visualization of platelet deposits, but it
may not be reliable in situations with large deposits due
to the buildup of several layers of platelet aggregates.
Flow rates were chosen to cover the full range of
physiological flow that would be caused by a variety of
cardiovascular devices. Steady flow through axisymmet-
ric geometries was modeled to reduce the complexities
of the problems that would arise with unsteady flow and
nonsymmetric geometry (as would be the case in vivo)
thus allowing a more fundamental understanding of the
underlying mechanisms involved. Previous studies with
stenotic geometries using vascular substrates
2021
have
indicated increases in platelet deposition in the throat
of the stenosis, contrary to the data presented here.
This is not only most likely due to differences in
adhesive forces between platelets and Lexan compared
with those of platelets and the vascular wall but is also
likely affected to some extent by differences in geometry
and flow conditions as well as elasticity of the wall. For
this study, the Lexan chambers were machined with a
uniform and consistent surface and geometry that pro-
vided an appropriate environment with which to visual-
ize the details of the flow and to correlate local flow
dynamics with platelet deposition on thrombogenic ar-
tificial surfaces.
The distribution of NPD for our stenosis geometry
compared favorably with results obtained by Karino and
Goldsmith,
11
who used low-Reynolds-number flow
through an abrupt expansion. They measured elevated
NPD in the recirculation zone immediately downstream
from the expansion and reported a sharp decrease in
NPD at the point of reattachment followed by a slight
increase and slow decay thereafter until the fully devel-
oped value of NPD was obtained. Our measurements
were unable to capture this sharp decrease at the
reattachment point. This could be attributed to the
relative sensitivities of each method to axial location.
Whereas our method provided average NPD over a
5-mm length, they obtained NPD at 1-mm intervals, and
the length of decrease in NPD at the reattachment point
was on the order of 2 to 4 mm. However, as the
Reynolds number was increased (>100) in their exper-
iments, with convective transport becoming more dom-
inant and the vortex being stretched in the direction of
flow, the region of decreased levels of NPD at the
reattachment point was less sharp (more spread out),
and the magnitude of decrease was attenuated. The
stretching of the vortex causes the angle made by the
dividing streamline with the wall to become acute, as
was the case in our study. Therefore, this local decrease
at the point of reattachment may be dampened by
decreasing the angle that the dividing streamline makes
with the wall.
Karino and Goldsmith
11
attribute this observed in-
crease in NPD to the geometry of the local streamlines.
In their view, NPD is increased in these areas due to the
enhanced convective transport of platelets and blood
cells to the vessel wall along the locally curved stream-
lines. Streamlines with components perpendicular to
by guest on August 14, 2014 http://atvb.ahajournals.org/ Downloaded from
1812 Arteriosclerosis and Thrombosis Vol 13, No 12 December 1993
c B A
Area of Interest
-Row
(a)
Row
(0
Fie 9. Schematic diagram of the streamlines for flow a,
into an expansion; b, into a contraction; and c, past an
aneurysm. The spheres A, B, and C represent platelets
being carried along the streamlines labeled 1, 2, and 3. In
all three cases, platelets are being carried to within the
critical distance for collision with the wall in increasing
numbers from sphere A on streamline 1 through sphere
C on streamline 3.
the wall will cause the flowing platelets to collide with
the wall at higher rates than streamlines parallel to the
wall. This explanation applies directly to the recircula-
tion region downstream from the stenosis and in the
expansion region of the contraction/expansion chamber
(see Fig 9a). Here we see that platelets carried by
streamlines 1 through 3 all come within the critical
distance for collision with the wall. The number of
platelets carried within the adhesion region of the wall
increases from points A through C. From point C
through point D, the collision frequency should be
similar, and this would be the location for maximum
deposition of platelets. As the Reynolds number in-
creases and the recirculation region decreases, the
region from points C through point D shrinks in length,
resulting in an NPD distribution with a stronger peak at
the center of the vortex (see Fig 7).
Furthermore, this may also explain the increases in
NPD observed in the entrance to the throat of the
stenosis, the entrance to the contraction, and the distal
side of the aneurysm. As flow enters a contraction, the
flow is convectively accelerated, resulting in a contrac-
tion of the streamlines in direct proportion to the
geometric contraction (see Fig 9b). We see here again,
as we proceed from point A through point C, that the
number of streamlines carrying platelets within the
critical collision distance increases and then remains
constant from this region through point D. This would
lead to the prediction that NPD should increase from A
through C and remain constant from C through D, as in
the recirculation region. What we observed (Figs 6 and
7), however, is that NPD was elevated only in the
entrance region (points A through C), followed by a
dramatic decrease at point D. This phenomenon con-
firms the theoretical predictions of Basmadjian
22
and
others that increased platelet deposition on an artificial
surface is dependent on the ability of the local geometry
to enhance transport of platelets to the vessel wall as
well as a diminished capacity to embolize these deposits
through decreased wall shear stresses. In this particular
case, wall shear stress will increase from point A
through point D. It can be deduced from the NPD
distributions (Figs 6 and 7) that at some point between
C and D, the wall shear stress increases past the
magnitude required to embolize the platelets that might
otherwise have been deposited. From Fig 7, we see that
as the Reynolds number increased, there was a relative
increase in NPD in the throat entrance and an upstream
movement of the peak NPD. This follows intuitively
from the fact that as the Reynolds number grows there
will be more contraction of the streamlines (resulting in
more platelet collisions with the wall), and as the flow
becomes turbulent the velocity profile will flatten and
cause a higher increase in wall shear stress (and thus an
increase in embolizing force) in the entrance region
than at lower flow rates.
Folie and Mclntire
9
also show increased platelet
aggregation in regions between adjacent mural thrombi
(creating flow past a cavity similar to flow through an
aneurysm). We have shown here that the distribution of
platelet deposition is not necessarily uniform across the
aneurysm, but is again dependent on the dynamics of
the flow within the aneurysm, as was the case down-
stream from the stenosis. A region of recirculation
similar to that observed downstream from the stenosis
existed in the region of the aneurysm (see Fig 9c). In
this case, the vortex was skewed toward the downstream
edge of the aneurysm, causing a contraction of the
vortex streamlines on the distal edge of the aneurysm
and an expansion of the streamlines on the proximal
edge. As the flow rotated counterclockwise from point
C through point A, the number of streamlines carrying
platelets within the critical adhesion distance from the
wall decreased. This implies that the number of plate-
lets colliding with the wall decreased from point C
through point A in the direction of the local flow in the
recirculation region, or, from a different frame of ref-
erence, increased along the wall of the aneurysm in the
direction of the bulk flow. For flow conditions of
Reynolds numbers 300, 900, and 1800, vortex regions
similar to that shown in Fig 9c were observed, thus
creating the approximately linear increases in NPD in
the aneurysm in the direction of the bulk flow (Fig 8).
At 6 L/min, the turbulence intensities were effective in
disturbing the flow within the aneurysm to the point of
dominating the laminar convection mechanisms de-
scribed above.
Relevance of Platelet Deposition to the Configuration
of Cardiovascular Devices
Thromboembolism and anticoagulant-related hemor-
rhage still represent the major causes of valve-related
morbidity and mortality in patients with implanted
heart valve prostheses and left ventricular-assist devic-
by guest on August 14, 2014 http://atvb.ahajournals.org/ Downloaded from
Schoephoerster et al Fluid Dynamics and Platelet Deposition 1813
es.
1
Dewanjee
23
summarizes the results of a series of
systematic studies on the imaging of platelet deposition
and quantification of platelet thrombus in a variety of
cardiovascular devices, including heart valve prostheses,
synthetic vascular grafts, oxygenators, and hemodialyz-
ers. Also, since local disturbances in the flow of blood
contribute significantly to the initiation of platelet dep-
osition and aggregation, as we have shown here, many
studies have been performed to determine the effects of
valve geometry on the dynamics of the flow downstream
from the valves.
24
-
26
Flow through mechanical heart
valve prostheses produces contraction and curvature of
the streamlines similar in nature to those depicted in
Fig 9, and a detailed analysis of the local fluid mechan-
ics in the close vicinity of the valve may provide infor-
mation as to the location and magnitude of platelet
deposition on the valve structure. However, more stud-
ies such as this one are necessary to elucidate the
interaction of the various fluid-mechanical and mass-
transport mechanisms and their relative influence on
platelet aggregation and deposition. We are currently in
the process of extending the qualitative correlations
between fluid dynamics and platelet deposition deter-
mined in this study with quantitative correlations by
using experimental and numerical simulations of blood
flow in identical geometries, in nonsymmetric geome-
tries, and under pulsatile flow conditions. We anticipate
these correlations may aid in elucidating the underlying
mechanisms involved in the process of flow-induced
platelet aggregation and deposition.
Ack nowledgments
Supported by grants from the National Institutes of Health
(HL46444 and HL47201) and the American Heart Associa-
tion, Florida Affiliate (Initial Investigatorship 91 11/10).
References
1. Addonizio VP, Edmunds ZH. Thromboembolic complications of
prosthetic valves. Cardiovasc Clin. 1985;3:431-435.
2. Chandran KB. Heart valve prostheses: in vitro flow dynamics. In:
Webster JG, ed. Encyclopedia for Medical Instrumentation, 111. New
York, NY: John Wiley & Sons; 1988:1475-1483.
3. Dewanjee MK. In-111 platelets in bypass grafts: experimental and
clinical applications. In: Thakur ML, Hardeman M, Ezekowitz
MD, eds. Radiolabeled Cellular Elements of Blood. New York, NY:
Plenum Press; 1985229-263
4 Muneretto C, Solis E, Pavie A, Leger P, Gandjbakhch I, Szefner J,
Bors V, Piazza C, Cabrol A, Cabrol C. Total artificial heart:
survival and complications. Ann Thorac Surg. 1989;47:151-157.
5. Dintenfass L. Rheological approach to thrombosis and athero-
sclerosis. Angiology. 1964;15:333-343.
6. Stein PD, Sabbah HN. Measured turbulence and its effect on
thrombus formation. Ore Res. 1974;35:608-614
7. Yoganathan AP, Corcoran NH, Harrison EC, Carl JR. The Bjork-
Shiley aortic prosthesis: flow characteristics, thrombus formation
and tissue overgrowth. Circulation. 1978;53:70-75.
8. Eckstein EC, Tilles AW, Millero FJ. Conditions for the occurrence
of large near-wall excesses of small particles during blood flow.
Microvasc Res. 1988;36:31-39.
9. Folie BJ, Mclntire LV. Mathematical analysis of mural thrombo-
genesis: concentration profiles of platelet-activating agents and
effects of viscous shear flow. Biophys J. 1989;56:1121-1141.
10. Goldsmith HL, Turitto VT. Rheological aspects of thrombosis and
haemostasis: basic principles and applications. Thromb Haemost.
1986;55:415-435.
11. Karino T, Goldsmith HL. Adhesion of human platelets to collagen
on the walls distal to a tubular expansion. Microvasc Res. 1979;17:
238-262.
12. Wang SK, Hwang NHC. On transport of suspended particulants in
tube flow. Biorheology. 1992,29:353-377.
13. Schmid-Schoenbein H, Born GVR, Richardson PD, Cusack N,
Rieger H, Forst R, Rohling-Winkel 1, Blasburg P, Wehmeyer A.
Rheology of thrombotic processes in flow: the interaction of eryth-
rocytes and thrombocytes subjected to high flow forces. Biorheology
1981;18:415-444.
14. Young DF, Tsai FY. Flow characteristics in models of arterial
stenosis, I: steady flow. J Biomech. 1973;6:395-410.
15. Deshpande MD, Giddens DP. Computation of turbulent flow
through constrictions. Eng Meek 1983; 109:466-478.
16. Dewanjee MK, Rao SA, Didisheim P. Indium-Ill tropolone, a
new platelet label: preparation and evaluation of labeling
parameters. J Nud Med. 1981;22:981-987.
17. Pumphrey CW, Chesebro JH, Dewanjee MK, Wahner HW,
Hollier PC, Pairolero PC, Fuster V. In vivo quantitation of platelet
deposition on human peripheral arterial bypass grafts using
Indium-Ill labeled platelets: effect of dipyridamole and aspirin.
AmJCardiol. 1983;51:796-801.
18. Brash JL, Brophy JM, Feuerstein IA. Adhesion of platelets to
artificial surfaces: effect of red cells. J Biomed Mater Res. 1976;10:
429-443.
19. Dewanjee MK, Kapadvanjwala M, Sanchez A, Elson R, Serafini
AN, Zilleruelo GE, Sfakianakis GN. Quantitation of comparative
thrombogenicity of dog, pig, and human platelets in a hemodi-
alyzer. ASA1O Trans. 1992:38:88-90.
20. Badimon L, Badimon JJ. Mechanisms of arterial thrombosis in
nonparallel streamlines: platelet thrombi grow on the apex of
stenotic severely injured vessel wall. J Clin Invest. 1989;84:
1134-1144.
21. Lassila R, Badimon JJ, Vallabhajosula S, Badimon L. Dynamic
monitoring of platelet deposition on severely damaged vessel wall
in flowing blood. Arteriosclerosis. 1990;10:306-315.
22. Basmadjian D. The effect of flow and mass transport in thrombo-
genesis. Ann Biomed Eng. 1990;18:685-709.
23. Dewanjee MK. Platelet thrombosis in cardiovascular prostheses,
the role of platelet scintigraphy and quantitation of regional
platelet density. In: Platelets and Atherosclerosis. Berlin, FRG:
Springer-Verlag; 1990:71-86.
24. Woo YR, Yoganathan AP. In vitro pulsatile flow velocity and shear
stress measurements in the vicinity of mechanical mitral heart
valve prosthesis. J Biomech. 1986; 19:39-51.
25. Baldwin JT, Tarbell JM, Deutsch S, Geselowitz DB. Mean flow
velocity patterns within a ventricular assist device. ASA1O Trans.
1989;35:429-433.
26. Schoephoerster RT, Chandran KB. Velocity and turbulence mea-
surements past mitral valve prostheses in a model left ventricle.
J Biomech. 1991;24:549-562.
by guest on August 14, 2014 http://atvb.ahajournals.org/ Downloaded from