Basic Research in

Cardiology Basic Res Cardio183:510-524 (1988)

Model of the coronary circulation based on pressure

dependence of coronary resistance and compliance*)
P. Bruinsma, T. Arts*), J. D a n k e l m a n * * ) , and J. A. E. Spaan

D e p a r t m e n t of Medical Physics, University of A m s t e r d a m ,

*) D e p a r t m e n t of Biophysics, University of Limburg, Maastricht, and
**) D e p a r t m e n t of Mechanical Engineering, Delft University of T e c h n o l o g y , Delft, T h e
Netherlands

Summary: The effect of pressure-dependent changes in vascular volume, resistance and capacitance
in the coronary micro-circulation, has been studied by a distributed mathematical model of the coronary
micro-vasculature in the left ventricular wall. The model does not include regulation of coronary blood
flow and is evaluated only for the fully dilated coronary vasculature. The left ventricular wall was
thought to consist of eight parallel layers, each of them with an arteriolar, capillary and venular
compartment. The resistance of each vessel was thought to depend on the inverse of squared volume,
according to Poiseuille's Law for tubes with constant length. Tissue pressure has been assumed to be
equal to left ventricular cavity pressure at the endocardium and to decrease linearly to atmospheric level
at the epicardium. The pressure-volume relation of the vessel compartments were assumed to be
sigmoidal. There is a rest volume at transmura[ pressure zero and AV/AP decreases with increasing
transmural pressure. Simulation of experimental protocols described by other authors yielded results
which were similar to the experimental outcomes, illustrated by: (1) a parallel shift to the flow axis of
the pressure-flow curves due to cardiac arrest (2) steady-state endo/epi ratio of flow as a function of
heart rate. It is concluded that interpretation of transients in coronary flow and/or pressure by models
containing fixed resistance and capacitance may seriously underestimate intramyocardial capacitative
effects and characteristic time constants for pressure-induced resistance changes.

Key words: coronary circulation model; _intramyocardial compliance; diastolic pressure-flow curves;
coronary resistance

Introduction
The realisation that myocardial micro-vessels are compliant is b e c o m i n g m o r e common
(e.g. 2, 10, 24, 27, 29). Because of this compliance, the v o l u m e of the micro-vessels will vary
depending on the difference b e t w e e n luminal pressure and tissue pressure: transmural
pressure. As a result, microvascular resistance and compliance will also be d e p e n d e n t on
these pressures (4, 8, 12). The purpose of this study is to evaluate the consequences of
pressure-dependent distributed resistance and compliance for the coronary circulation
mechanics.
The first premise of this study is that the transmural p r e s s u r e - v o l u m e relation of the
vessels is sigmoid, as is generally true for a biological c o m p a r t m e n t (e.g. 29). This sigmoidal

*) Supported by grants from the Foundation for Medical Research Medigon (grant 13-22-63) and
Biophysics (grant 810-406-014).
508
Bminsma et al., Model of the coronary circulation 511

shape implies that (a) there is a finite vessel lumen at transmural pressure zero, (b) the
vessels collapse if the external pressure becomes sufficiently higher than internal pressure
(e.g. Fig. 1), and (c) the compartment is stiffer at higher vascular transmural pressures.
The present model differs essentially from earlier ones of coronary circulation. The
waterfall model of Downey and Kirk (15) assumes an open or closed behavior of the vessels.
The shape of the applied pressure-volume relation, then, is different: a step shape vs.
smooth sigmoidal. In both models, microvessels can collapse. The difference is that in the
waterfall model the vessel collapse occurs instantaneously and very locally, in theory at an
infinitely small spot. In the present study, we assume that such a local collapse will not occur
because of the structural connectedness of the micro-vessel with myocardial tissue (7).
Hence, the volume variations of vessels will be more homogeneously spread, A consequence
is that a larger vascular space will vary more or less homogeneously due to the myocardial
squeeze. As a result, time is required for these volume variations to occur. Dole et al. (12)
also suggested a model with pressure-dependent resistances. However, in that model, the
resistance-pressure relation was chosen to fit a specific set of experimental data. The model
study of Beyar et al. (6) is based on the assumption of a Starling resistor effect in the
arterioles and consequently the transmural pressure in the arteriolar compartment is
prescribed. In the present study, the pressure dependence of the resistance, as well as the
compliance effects, result from a pressure-volume relationship. The transmural pressure in
any of the compartments is prescribed (with the exception of a reference condition) but
rather follows as a result of the solutions to the model equations.

Model of the coronary circulation

Model equations
The coronary circulation within the wall of the left ventricle is represented by a network of
resistances and compliances. The wall is subdivided into eight layers from the endocardium
to the epicardium. The choice of eight is arbitrary. It allows us to take a distribution of
parameters from subendocardium to subepicardium, but it is still acceptable in terms of the
computer time needed for model calculations. The vasculature of each layer is represented
by an arteriolar, a capillary and a venular compartment in series. Each compartment is
compliant and hydrodynamically resistive. The blood volume of a compartment, V~,
depends on the vascular transmural pressure, Ptr, by a non-linear relation (Fig. 1, upper
panel). It is assumed that the resistance of a compartment is related to its volume as if the
compartment was a cylindrical conduit. According to Poiseuille, the resistance is then
proportional to 1/r 4 where r is the radius of the conduit. Assuming that vessel length is
constant, the volume is proportional to r 2 yielding, as the relation between resistance (R)
and volume:
Sr
R = ~-~ (1)

where Sr is a constant, specific for that compartment, and

Vc = Vc(Prr) = volume of the compartment as a function of vascular transmural pressure.
The vascular transmural pressure is the difference between intravascular pressure Pc, and
hydrostatic pressure in the surrounding tissue, Pim
Pt~ = Pc - Pi,,. (2)
The lower panels of Fig. 1 depict the schematic representation of one compartment. The
resistance of the compartment, 2R, has been divided into equal parts. Their junction is
512 Basic Research in Cardiology, Vol. 83, No. 5 (1988)

v, 9" ~ Cb=tg~

', C , = t g a

V
o P, ~o P, 2o
a Tranamural
pressure (mmHg)

i ,1 i o.. o-,.
,..7 ......... : . . . . . . , P, P,

P, Po
. 9 , , R =f(P~-Pim) ['~c~ R

aV ~ I C=glP9
=~V. l A p.
b R - l/V: C p,.

Fig. 1. Schematics related to a compliant vessel compartment.

(a) Relation between transmural pressure, P~,, and vascular volume, Vc. At each point on the curve the
compliance, dVc/dP,, equals the tangent at that point. (Pa, V~,a) and (Pb, Vc.b) are two points on the
transmural pressure-volume curve.
(b) The effect of a change in volume Vc on resistance, R. Pp, Pc and Pd are the proximal, compartmental
and distal pressures, respectively. Flows into and out of the compartment are determined by the
pressure differences, Pp - Pc and Pc - Pd, respectively, and the resistance which has an Poiseuille-law
behavior. The transmural pressure is the difference between the compartmental pressure, Pc, and
pressure outside, P~.. A change in volume, Vc, can be caused by a change in distending pressure, Pc, or
in compressive pressure of the surrounding tissue, Pim.
(c) Electric analog of a vessel compartment. Resistances and the compliance depend on volume, which
in turn depends on transmural pressure. Q~. and Qo,, are the flows into and out of the compartment,
respectively. For the capacitative flow: Qcap = C dP,r/dt.

connected to tissue pressure by a capacitor, representing vascular compliance. The flows

into, Qi., and out of a c o m p a r t m e n t , Qout, are d e t e r m i n e d by the distribution of hydrostatic
pressures:
Oin ~-- (Vp - Pc)/R (3)
and Qo.t = (Pc - Pd)/R (4)

in which Pp, Pd and Pc are the hydrostatic pressures at the proximal end, at the distal end,
and within the c o m p a r t m e n t , respectively.
Because of the conservation of mass, the change of v o l u m e of a c o m p a r t m e n t is related to
both its inflow and outflow according to:

dVc
dt = (Oin - Qout) (5)

where compartmental v o l u m e and transmural pressure are mutually related, as shown by

Fig. 1 (upper panel).
Bruinsma et al., Model o f the coronary circulation 513

Vascular transmural p r e s s u r e - v o l u m e relations

The numerical solution method applied to our equations requires a mathematical formula-
tion of the pressure-volume relationship without discontinuities in both the function and its
derivatives.
The three conditions for the pressure-volume relation mentioned above; possible total
collapse, rest volume at Ptr = 0, and decreasing dV/dPtr with increasing Ptr, are met by the
following equation:
Ptr = a. exp(eV) + b . ln(cV) - P0 (6)

The first term dominates at high and the second term at low volume values.
Experimental data to estimate the value of the parameters in the equation are scarce. One
way to evaluate the shape of the pressure-volume relation is by the dependence of vessel
distensibility, (AV/APtr)/V , on transmural pressure.
The coefficients a, b and P0 influence the shape of the curve, whereas the constant c acts as
a scaling factor for the volume axis. Its determination will be discussed below. The constants
a, b and P0 for the different compartments were estimated such that the pressure distensibil-
ity curve agreed over a range of positive transmural pressure with a recent survey (29). For
the capillary compartment, this is shown in Fig. 2. The pressure-volume curve for the
arteriolar compartment was made somewhat stiffer than the capillary and venular compart-
ments at higher transmural pressure to match the diastolic pressure flow curve of Downey
and Kirk (15). The curves, used as a standard for this study, are shown in Fig. 3.
P0 = 5 mm Hg. For values of a, b and e see Table 1. The effect of the pressure-volume
relation on the results will be discussed below.

Definition o f r e f e r e n c e distribution o f pressure, f l o w a n d v o l u m e

The reference condition was the non-beating heart with zero tissue pressure in all layers
(zero, left ventricular pressure) and with steady flow through all compartments. This
reference condition is defined by the data in Table 1.

EG a~1
20,

9 ~ ..... ~
| |

0 5O 100

Transmural pressure
mmHg
Fig. 2. Distensibility as a function of transmura[ pressure ( - - ) related to pressure-volume relation for
capillary compartment (Fig. 3), (-.-.-) compromise curve of Spaan (29), ( 0 ) arteriolar data of
Wiederhielm (33) and (11) capillary data of Smaje (26).
514 Basic Research in Cardiology, Vol. 83, No. 5 (1988)

Table 1. Value variables and parameters in the reference condition.

Pressure Volume Flow a b c

mm Hg ml/100 g ml/s/100 g mm Hg mm Hg 1/ml

Arteriolar 35 1.8 4 0.01 4.13 4.5

Capillary 15 4.0 4 0.17 7.57 1.0
Venular 7.5 3.6 4 0.14 5.89 1.0
Endo/epi ratio 1 1.14 1.56 1 1 1/1.4

As we are aware of the lack of experimental data and the restrictions of our model, we
assumed that the pressure distribution from arteries to veins was the same for all layers in
the reference condition, i.e. intracavitary pressure = zero. This distribution was taken to be
50, 35, 15, 7.5 and 5 m m H g for the arterial, arteriolar, capillary, venular and venous
pressures, respectively.
In the reference situation, total coronary flow was assumed to be 4 ml/s/100g LV.
However, the flow distribution from endocardium to epicardium is not uniform in the
arrested heart. We assumed an endo/epi flow ratio of 1.56. The flow values in the
intermediate layers were determined by linear interpolation. The assumed endo/epi ratio is
within the range of experimentally found ratios in the non-beating heart: 1.6 (34), and
1.52 (14).
We assumed a total coronary blood volume of 9.4 ml/100g LV. Using the ratio for the
three compartmental volumes, as given by Spaan (29), one arrives at a distribution of
volume of 1.8, 4.0 and 3.6 ml/100g LV for the arteriolar, capillary and venular compart-
ments, respectively. The coronary volume is also not uniformly distributed from suben-
docardium to subepicardium. We assumed an endo/epi volume ratio of 1.14 (32) for all three
compartments in the reference condition.
The combination of pressure and flow distribution through each layer allows the calcula-
tion of all the resistances in the reference condition using Eqs. (3) and (4). Then, by using
the volumes in this condition, the parameter S~ in Eq. (1) can be calculated for each
compartment. In the reference condition, transmural pressure Ptr equals Pc since Pim equals
zero (Eq. 2). Knowing the distribution of volume and pressure, the parameter c in Eq. (6)
can be determined for each compartment.

Method o f solution o f model equations

The behavior of each layer is described by three coupled first order non-linear differential
equations of the form:
dVi Pp,i -- 2 Pc i -t- Pd,i
W ~- Ri' (7)

where index i (= 1, 2, 3) refers to the arteriolar, capillary and venular compartment,

respectively.
Pc is a function of both Pim for the layer and V~.
R i is a function of Vi.
Pp,1 = Pa, coronary arterial pressure and
Pa,3 = Pv, coronary venous pressure.
The solution can be obtained from any numerical scheme of integration when Pa, Pv and P~m
are defined as a function of time. A t the start of the calculations, Pc,l, Pc,2 and Pc,3 are
defined.
Bruinsma et al., Model o f the coronary circulation 515

150,

100 art. cap,

E
E

9 50

_,=
o
9~ 4.0 alo
Endocardial V o l u m e

(ml/100gr)

50

Fig. 3. Endocardial volume-transmural pressure relations over the relevant upper range of pressure.
The solid lines apply to the artcriolar, capillary and venular compartments as used in the standard
model. The working area lies between < [] > and < < [] > > (for perfusion pressure of 50 mm Hg and
100ram Hg, respectively); ( - - - ) is an alternative venular pressure-volume relation Ptr = 0.045 V~)
differing essentially from the standard relation in that volume approaches zero at transmural pressure
zero. This curve represents venular space that cannot withstand negative pressures.

At t -- 0, dVi/dt is known for i = 1-3 and hence estimates of Vi can be made for a little
time, At, later. Then, from the new values of Vi(At), a new transmural pressure and
resistance distribution in the layer can be calculated. From this and tissue pressure, new
values of dVi/dt at t = At can be predicted. The procedure can then be repeated. We used a
4th order Runga-Kutta method. Where P,~, Pv and/or Pim are periodic with a certain
frequency, the solution of the integration procedure will become periodic at the same
frequency after a certain time, regardless of the definition of the variables at t = 0. From the
calculated pressure and resistance distribution as a function of time, the flows into and out of
each compartment can be calculated.
In the present form of the model, the solutions of the layers are independent of each
other, since inlet pressure (Pp,1) and outlet pressure (P~.3) are equal for all layers. The total
coronary arterial flow was obtained simply by summation of the arterial flows into each
layer.
The solutions for the layers are different mainly because Pin, differs from layer to layer.
Theoretically, from cardiac mechanics, intramyocardial pressure can taken to be equal to
the negative of radial stress in the wall (3). Intramyocardial pressure was equal to left
ventricular pressure at the endocardium and was assumed to decrease linearly to zero at the
epicardium. This is in agreement with recent results (20). Moreover, it is consistent with the
observation that in the vasodilated heart, heart rate does not effect subepicardial flow but
only subendocardial flow (5).
 P )err = 50mmHg p )err = 100mrnHg
20'
2D

~ ~ d
0

" illllllll -"

o, " ~ o
;' s

~ ~
. . . _ I
-2oi -2o~

Time Is) i "

Time
"-"
te~
10

8
""-. 8 "=
0 "" . . . . . . . . . . . . 9 0

$ t I0
Tlml (sl

8'
g

k,~,,: ....... _a_......

"''- . . . . . . . . . . . 4

8
i " i
i

~ . , = Q

(.1 f " " 'vo

Time (lit Time (e)

h
i , d

> o, ~ : - - -- ~ o.
5 ~ IC 8 1 10
T~Q ~e) Time (e)

Fig. 4. Instantaneous subendocardial flow and vascular compartmental volumes as predicted by the
model. Perfusion pressures are 50 m m H g (left panels) or 100 mm Hg (fight panels). After two heart
beats, both a prolonged systole (broken lines marked with "s") and a prolonged diastole (broken lines
marked with "d") were simulated. (a) and (e) show the arlefial inflow signals; (b) and (f) the arteriolar
volumes; (c) and (g) capillary volumes and (d) and (h) venular volumes at the different per~usion
pressures, respectively. The arrows point to the beginning of a systole. Venous pressure was 5 mmHg.
Diastolic and systolic left ventficular pressure were 0 and 100 mm Hg, respectively.
Bruinsrna et al., Model o f the coronary circulation 517

Model evaluation
Firstly, the m o d e l is evaluated for the case that intramyocardial pressure varies periodi-
cally with a frequency of 100/min (Figs. 4 and 5). W e used a square wave form so that in the
simulated systolic and diastolic signals the transients are accentuated and d e t e r m i n e d by the

Perfusion pressure

50mmHg 100 mmHg

I.I. 201 a 201 d

m

-- o o

I- E -'tot
t
, . I5
U.
20. 9 20,
0
"~

m
o
o
'o

"~
o)
o
o
10.

0.

-10.

-20,
~ -

9
10,

9-10,

"-20
0'

t
c

50, 9 200 50. 200

e- oo
25, ", ,' a ", I00 25. 100'
.m_
9 E
a: E 0
- - i r ,

t
i~ b I
Time 0.5s

Fig. 5. Instantaneous total flow (a) and (d), subendocardial flow (b) and (e), and subendocardial
resistance variations (c) and (f). Left and right panels depict these variations at low and high perfusion
pressure, respectively9 In the four top panels the flow variation calculated from the model are
represented by the heavy lines, and the variations in flow due to resistance effects alone by the thin
lines. At a low level of perfusion pressure, the phasic behavior of flow is solely due to intramyocardial
pumping, whereas at higher values resistive effects are dominant at end-diastole. Bottom panels show
the resistance variations of the arteriolar (a), capillary (c), and venular (v) subendocardial compart-
ment. Scalings for the arteriolar and capillary resistance are given at the left-hand side. Venular scaling
is given at the right-hand side. The arrow denotes the beginning of a systole (redrawn from [30] with
permission).
518 Basic Research in Cardiology, Vol. 83, No. 5 (1988)

pressure-dependent capacitance and resistance effects. The simulated left ventricular square
wave had a minimal value ("diastolic") of 0 mm Hg and a maximal value ("systolic") of
100 mm Hg. The transients from a steady beating state to a permanent systole and diastole
(dashed lines) are presented as well, because these conditions are often considered as a
reference to evaluate perfusion of the beating heart. Moreover, these transients elucidate
the characteristics of the model. Next, it is assessed whether the model is able to predict the
shift in mean arterial pressure-flow relations in the arrested and beating heart. Then the
predictive value of the model for the microsphere distribution in the beating heart is
evaluated. For the last two cases, the left ventricular systolic pressure wave was approxi-
mated more realistically, using a fourth order polynome.

Dynamic characteristics o f the model

The dynamic behavior of the model depends on the level of coronary perfusion pressure,
as illustrated in Fig. 4, for the subendocardial layer, applying a square wave form to the left
ventricular pressure curve. The first 5 s of the simulation were allowed for the model
solutions to become stationary. Figure 4e-h depicts the time course of arteriolar inflow and
compartmental volumes of the subendocardial compartments at a perfusion pressure of
100 mm Hg. End-systolic and end-diastolic inflows are close to the values for the permanent
systole and diastole, respectively. However, at lower values of perfusion pressure
( 5 0 m m H g , Fig. 4a-d), only the diastolic steady state flow is reached quite closely within
each cycle. Averaged compartmental volumes are lower at the lower perfusion pressure,
resulting in higher resistances (Fig. 5), leading to significantly lower rates of change of most
variables, especially in the transients to permanent systole. Generally, transients are found
to slow down by either a lower perfusion pressure or a higher tissue pressure. Note that the
predicted volume variations in the capillary compartment during "beating" are small at both
perfusion pressures, but that the mean level is higher at the higher perfusion pressure
(Fig. 4). Also note that the venous compartment collapses if systole is long enough, but that
the rate at which this collapse occurs is finite and decreasing with decreasing volume.
Arterial flow in the subepicardial layer is almost steady because the variations in intramy-
ocardial pressure are low. As is shown, systolic retrograde flow in the subendocardium layer
is compensated for by the more continuous flow in the outer layers, resulting in only a small
retrograde flow in the summed signal and even in an antegrade systolic flow at a perfusion
pressure of 100 m m H g (Fig. 5). Obviously the retrograde inflow can only be due to
capacitative effects denoted as intramyocardial pumping (28).
The subendocardial resistance variations (Fig. 5, c and f) are largest in the venular
compartment. At a perfusion pressure of 100 m m H g , the relative variation in venular
resistance throughout the cardiac cycle is much larger than at 50 m m H g . A t the lower
perfusion pressure, diastole is not long enough to reopen the venules completely. Moreover,
at these lower perfusion pressures, the arterioles are squeezed as well, resulting in a time-
varying arteriolar resistance. However, its relative contribution to variations in overall
resistance remains limited. At the lower perfusion pressures, the predicted resistance in the
beating heart is predominantly in the venular compartment throughout the cycle. This seems
contrary to the generally accepted principle that the major part of the resistance is located in
the arterioles. However, note that due to the compression of the venular compartment,
capillary volume (see Fig. 4g) and hence transcapillary pressure are preserved, not-
withstanding the compression of capillary space. In this way the pressure dependence of the
venular compartment provides a regulatory mechanism for transcapillary water flux (31).
The relative contribution of capacitative flow can now be assessed by comparing the
calculated subendocardial flow in the presence and absence of capacitative flow effects. In
Bruinsma et al., Model o f the coronary circulation 519

panels b and e of Fig. 5 the subendocardial inflow, including capacitative effects, is

compared with the resistance-determined inflow. The latter has been calculated as the ratio
between arterial-venous pressure difference and instantaneous sum of subendocardial
resistances. At a perfusion pressure of 50 mm Hg the pulsatility of the subendocardial flow
signal is mainly due to intramyocardial pumping in the compliant coronary vasculature,
while at 100 mm Hg the effect of resistance variations become more pronounced.
The same conclusions hold for inflow of the heart as a whole: i.e. the sum of inflow of all
layers. This is shown by Fig. 5a and d. Although diastolic flow reaches its end-diastolic value
quickly, coronary resistance is changing during the whole of diastole. These resistance
variations are not apparent in the flow signal because of the compliance effects. The
interaction between the capacitative flow component and resistance-determined transients
will be analyzed in more detail in the discussion.
Obviously, the numerical results are sensitive to the assumed reference condition and
pressure-volume relationships. When we used different pressure-volume curves, the exact
shapes of the transients in the signals were curve-dependent. However, the basic elements
described above were always present.

Transmural f l o w distribution
From Fig. 4, top panels, it is clear that the model predicts a decrease of endocardial flow
due to cardiac contraction. In this section we will analyze how well this prediction is in
agreement with published experimental data and how sensitive this prediction is for model
parameters. For this we selected experimental studies, done in the fully dilated coronary bed
and under steady conditions. Although many studies on the transmural myocardial flow
distribution have been published (e.g. [17]), very few met these conditions. The selected
papers are referred to in Fig. 6.
For the simulations, the systolic left ventricular pressure wave was approximated by a
fourth order polynome. Diastolic pressure was assumed to be zero. The results of the
simulations were very sensitive to diastolic time fraction of the heart cycle. Using the values
given by Bache and Cobb (5) the predictions are close to the experimental data (Fig. 6).

I
1.5 ~

e3
o
O 1.0 ~4 I5

r [] 0 9 3 <'~
=
| 0.5
O []

o 1~o 2~o
Heart rate (mill I)

Fig. 6. Simulated and experimentally determined ( 0 ) transmural distribution of flow as a function of

heart rate. In the model, flow through the eight layers in the reference situation was chosen in such a
way that the flow ratio between the two inner layers and the two outer layers was 1.53. (C)) and (D)
were calculated using diastolic time fractions of 0.53, 0.47 and 0.4 at H R = 100, 150 and 200,
respectively. ( ~ ) was calculated using diastolic time fractions of 0.65, 0.59 and 0.53 at the same heart
rates (6); (U) the alternative pressure-volume curve was used. Experimental results are obtaincd from:
(1), ref. [34]; (2), ref. [14]; (3), ref. [13]; (4), ref. [5]; and (5), ref. [16].
520 Basic Research in Cardiology, Vol. 83, No. 5 (1988)

However, Bache and Cobb determined the diastolic time from the time from the aortic
dicrotic notch to aortic pressure rise. The time for which the ventricles are relaxed is
somewhat shorter. The diastolic time fraction was reduced by about 20 % to compensate for
this. In this case, the model predicted lower values, as shown in Fig. 6.
Using the shorter diastolic time fractions, the effect of the shape of the pressure volume
curve on the endo/epi distribution was studied. The results were not very sensitive to a shift
in the pressure-volume relations of both the capillary and venous compartment over the
pressure axis. In order to achieve an increase in endo/epi ratio at H R between 100 and 200
beats/min comparable to the effect of increase in diastolic-time fraction of 20%, the
pressure-volume relations had to be shifted 30 m m H g to the left. The effect of an
alternative shape of the venous pressure-volume relation (broken line in Fig. 3) was studied
as well. With this curve, blood volume approaches zero at zero vascular transmural pressure,
and then the predicted effect of heart rate on the endocardial flow is larger (see Fig. 6, open
squares). The relative importance of this effect is larger at higher heart rates and lower
perfusion pressure. At a perfusion pressure of 70 mm Hg the predicted endo/epi ratio of flow
is 0.6 with the standard model but 0.3 with the use of the alternative venous
pressure-volume relation.
The sensitivity of the model for the reference volume of the venular compartment was
studied. Endo/epi ratio at an arterial pressure of 100 mm Hg and H R of 100 beats/min varied
from 0.88 to 0.817 (9 %) if venular reference volume was altered from 2.4 to 4.5 ml (187 %).
In conclusion, calculated mean endocardial flow is almost uninfluenced by the assumed
distribution of volume, but it depends strongly on whether the vasculature can maintain

20

.,J

0
Q
15

10
0

=
o
5,
r

I-"
0 ..~ t .
0 100 200

Perfusion p r e s s u r e (mrnHg)

Fig. 7. Coronary blood flow as a function of perfusion pressure in the beating and in the arrested heart.
Left and right solid lines represent standard model predictions for the arrested and beating heart (heart
rate 150 beats/rain), respectively. (- - -) represents the linear extrapolation of the solid curves towards
the pressure-axis. This results in apparent zero flow pressure of 20 and 45 mm Hg, respectively. (Q) and
(11) represent one typical experiment of Downey and Kirk ([15], their Fig. 7) in the arrested and beating
heart, respectively. (O) represents a typical experiment of Klocke et al. ([23], Fig. 1) in the arrested
heart only. Points of scaling are (| resp. (9 (redrawn from [30] with permission),
Bruinsma et al., Model of the coronary circulation 521

volume at zero transmural pressure. The diastolic time fraction seems to be the major
determinant of the models in predicting endocardial flow.

Pressure-flow relations
In this section we analyze the shape of the pressure-flow relation in the arrested and
beating heart, as predicted by the model and depicted in Fig. 7. As a test of these
predictions, the experimental data from two other studies (15, 23) are provided. The model
parameters were according to their reference values. For the left ventricular pressure curve,
the same fourth order polynome was used as applied when predicting the flow distribution.
The flow data on the arrested and beating heart of the study of Downey and Kirk (15)
were scaled such that the diastolic flow at a pressure of 130 m m H g coincided with the
predicted curve. This was done at a pressure of 75 m m H g with the data of the study of
Klocke et al. (23). Heart rate was not given in the paper of Downey and Kirk, but a value of
150 can be deduced as a minimal value from Fig. 6 of their paper.
With a diastolic time ratio of 0.47, but without any further change in parameters, the
standard model predicted the pressure-flow relation of the beating heart according to the
heavy curve on the right. The agreement between experiment and theory is fair. The
average shift of the parallel curves for higher perfusion pressures amounted 31 mm Hg in the
experiments and is 26 mm Hg for the theoretical curves. Of course the magnitude of this shift
depends on the value of diastolic time fraction.
The deviation between model and experiment is largest at lower flow rates. Basically, the
model curve for the arrested heart approaches venous pressure asymptotically and the
experiments show an intercept. The differences will be discussed in the following.
The most important conclusion to be drawn, however, is not the quantitative agreement
between model and experiment. Figure 7 shows that the curvature of the pressure-flow line
in the beating heart can be explained on the basis of the pressure dependence of resistance of
the coronary circulation. The assumption of a gradual dropping out of systolic perfusion
from endo- to epicardium is not a prerequisite.

Discussion
Sensitivity o f the model for model parameters
The reference condition was altered by changing total intramyocardial blood volume, as
well as the endo/epi ratio of blood volume. If, in the reference condition, the blood volume
of all compartments is increased by 100%, the steady state flow in the beating heart
(HR = 100) at perfusion pressure of 100 mm Hg is affected by less than 2 %. At a lower
value of perfusion pressure, 50 mm Hg, the decrease in coronary flow in the beating heart is
less than 4 %. When in the model, at constant total volume, the endo/epi ratio of the volume
is changed from 1.14 to 1.42, which is the experimental range (11, 25, 32, 33), the endo/epi
ratio of flow is only affected at perfusion pressures lower than 50 mm Hg. The effect of a
change of venular volume fraction was also minor, as discussed above.
The shape of the pressure-volume relation of the venous compartment is important to the
model results, particularly the volume at zero transmural pressure. If this rest volume exists,
the model predictions of transmural flow distribution are more realistic than when this rest
volume is disregarded. This rest volume is in agreement with the experimental findings of
Kajiya et al. (22). Because the venous compartment acts as a regulator to maintain
transmural capillary pressure (Fig. 4), the pressure distribution in the reference condition is
522 Basic Research in Cardiology, Vol. 83, No. 5 (1988)

not of major importance. The effect of a different pressure-volume relation on microsphere

distribution was discussed above.

Interpretation of coronary arterial pressure-flow relations

In many studies, systole and diastole are treated as unrelated episodes in the cardiac cycle
(1, 9, 15, 18, 19). In the present analysis, the events in systole and diastole are coupled.
However, discontinuous behavior of especially venular resistance becomes apparent in our
model results when the volume variations are able to occur rapidly enough. In the beating
heart, this is the case at higher perfusion pressures, as shown in Figs. 4 and 5. This probably
causes the reasonable level of agreement at high perfusion pressure between the waterfall
interpretation and our simulated shift in pressure-flow relations due to cardiac arrest.
Indeed, the predicted coronary flow signal at a perfusion pressure of 100 m m H g exhibits
almost an on-off characteristic (Fig. 4d). However, at lower perfusion pressures, the changes
in vascular volumes become too slow to justify an on-off approximation of coronary
resistance. Rate of volume changes became even slower when the alternative venous
pressure-volume relation was used, since venous resistance became even higher. This effect
was also the cause of the reduced endo/epi ratio of flow, discussed above when applying this
curve. The interpretation of time-varying arterial flow signals should incorporate the effect
of variation of resistances and of flow due to volume changes of the compartment; c.q.
pumped flow. The intramyocardial pumping mechanism does decrease systolic arterial flow
but increases diastolic flow. As shown in Fig. 5, the pumped diastolic flow conceals the
diastolic resistance variation. If the pumped diastolic flow component were absent, flow
would increase during diastole. However, one also may turn this reasoning around, and
formulate that diastolic resistance variation conceals the diastolic capacitative (pumped)
flow. As is clear from Fig. 5, panels a and d, diastolic arterial flow reaches its late diastolic
value quickly, implying a short time constant characterizing the flow transient in diastole.
Fitting such a flow transient by a parallel resistance-capacitance network with constant
network elements, as done by Lee et al. (24), results in low RC values of the order of 0.2 s.
The conclusion, however, that the capacitance effect on flow is consequently small is
premature, since the quick transient in flow is due to the combined effect of capacitance
flow, decreasing in diastole and resistive flow, increasing in diastole. The influence of time-
varying resistance on the transients in the simulated flow signal is clear in the panels a, b, d
and e of Fig. 5. The intramyocardial compliance effects are over when the flow simulated by
the model reaches the flow simulated by the time-varying resistance components alone. The
time needed for this equilibration depends on vascular transmural pressure, and hence on
tissue pressure and perfusion pressure. Obviously, the same variable resistance effects play a
role in step function changes of perfusion pressure, as applied in the so-called compliance
free method to determine diastolic pressure-flow relations.

Subendocardial perfusion
The subendocardium is most vulnerable to ischemia (e.g. (21)). It is the area in the heart
where flow is compromised first. As is clear from the microsphere studies referred to in
Fig. 6, the effect of cardiac contraction is an important factor in this phenomenon and was
described fairly well by our model. The effect of cardiac contraction on endocardial flow is
not due simply to the decrease in diastolic time. The impediment by extravascular compres-
sion of mean subendocardial flow is mainly due to its effects on the resistance of the vessels.
Resistance of the different compartments varies during the cardiac cycle. Hence, those
factors that lead to reduction of the resistance in diastole and systole enhance subendocar-
dial flow. Diastolic decrease of resistance accelerates progressively with time, since the
Bruinsma et al., Model of the coronary circulation 523

increasing microvascular volume increases the rate of filling the intramyocardial vessels (e.g.
Figs. 4 and 5). Fortunately, the systolic increase of resistance slows down when systole
progresses. The rate of resistance (volume) variations are determined by the instantaneous
resistance and compliance values of the respective compartments. Via the pressure-volume
relations the instantaneous speed of changes in resistance depends on vascular transmural
pressures. The role of perfusion pressure is clear from Fig. 5, panels c and f. At a perfusion
pressure of 50 mm Hg, venular and capillary resistance are higher during the heart cycle than
at 100 m m H g , but the relative variations of resistance are much larger at the higher
perfusion pressure. Obviously, lowering tissue pressure will have the same effect as
increasing perfusion pressure. Because of the accelerated decrease of resistance in diastole,
diastolic duration might be of more importance than hitherto assumed.
Very little is known about wall mechanics of the coronary micro-circulation. It is not
impossible that pathological changes in the wall may lead to an increased effect of cardiac
contraction on subendoeardial flow in some circumstances. As is clear from this paper, an
understanding of the time-varying phenomena is necessary to describe the impediment of
subendocardial flow by cardiac contraction.

References
1. Archie JP (1975) Intramyocardia[ pressure: effect of preload on transmural distribution of systolic
coronary blood flow. Am J Cardiol 35:904-911
2. Arts MGJ (1978) A mathematical model of the dynamics of the left ventricle and the coronary
circulation. Ph.D. Thesis, State University of Limburg, Maastricht, The Netherlands.
3. Arts T, Reneman RS, Veenstra PC (1979) A model of the mechanics of the left ventricle. Am
Biomed Eng 7:299-318
4. Arts T, Reneman RS (1985) Interaction between intramyocardial pressure and myocardial circula-
tion. J Biomech Eng 107:51-56
5, Bache RJ, Cobb FR (1977) Effect of maximal coronary vasodilation on transmura[ myocardial
perfusion during tachycardia in the awake dog. Circ Res 41:648-653
6. Beyar R, Sideman S (1987) Time-dependent coronary blood flow distribution in left ventricular
wall. Am J Physiol 252:H417-H433
7. Borg TK, Caulfleld JB (1981) The collagen matrix of the heart. Fed Proc 40:2037-2041
8. Bruinsma P, Arts T, Spaan JAE (1985) Coronary pressure-flow characteristics in relation to the
distensibility of the microvessels. Med Biol Eng and Comp 23 suppl II:1325-1326
9. Buckberg GD, Eixler DE, Archie JP, Henney RP, Hoffman JIE (1975) Variable effects of heart
rate on phasic and rcgional left ventricular muscle blood flow in anaesthetized dogs. Cardiovasc Res
9:1-11
10. Canty JM, Klocke FJ, Mates RE (1985) Pressure and tone dependence of coronary diastolic input
impcdance and capacitance. Am J Physiol 248:H700-H711
11. Crystal GJ, Downey HF, Bashour FA (198l) Small vessel and total coronary blood volume during
intracoronary adenosine. Am J Physiol 241:H194-H201
12. Dole WP, Alexander GM, Campbell AB, Hixson EL, Bishop VS (1984) Interpretation and
physiological significance of diastolic coronary artery pressure-flow relationships in the canine
coronary bed. Circ Res 55:215-226
13. Domcnech RJ, Goich J (1976) Effect of heart rate on regional coronary blood flow. Cardiovasc Res
10:224-231
14. Domenech ILl (1978) Regional diastolic coronary blood flow during diastolic ventricular hyperten-
sion. Cardiovasc Res 12:639-645
15. Downey JM, Kirk ES (1975) Inhibition of coronary blood flow by a vascular waterfall mechanism.
Circ Res 36:753-760
16. Downey HF, Bashour :FA, Boatwright RB, Parker PE, Keehejian SJ (1975) Uniformity of
transmural perfusion in anesthetized dogs with maximal dilated coronary circulations. Circ Res
37:111-117
524 Basic Research in Cardiology, Vol. 83, No. 5 (1988)

17. Feigl EO (1983) Coronary physiology. Physiol Rev 63:1-205

18. Gregg DE, Green HD, Wiggers CJ (1935) Phasic variations in peripheral coronary resistance and
their determinants. Am J Physiol 112:362-373
19. Gregg DE, Green HD (1940) Registration and interpretation of normal phasic inflow into a left
coronary artery by an improved differential manometric method. Am J Physiol 130:114-125
20. Heineman FW, Grayson J (1985) Transmural distribution of intramyocardial pressure measured by
micropipette technique. Am J Physiol 249:H1216-H1223
21. Hoffman JIE (1981) Why is myocardial ischaemia so commonly subendocardial? Clin Sci
61:657-662
22. Kajiya F, Tsujioka K, Goto M, Wada Y, Chen X-L, Nakai M, Tadaoka S, Hiramatsu O,
Ogasawara Y, Mito K, Tomonaga G (1986) Functional characteristics of intramyocardial capaci-
tance vessels during diastoles in the dog. Circ Res 58:476-485
23. Klocke FJ, Mates RE, Canty jr JM, Ellis AK (1985) Coronary pressure-flow relationships:
controversial issues and probable implications. Circ Res 56:310-323
24. Lee J, Chambers DE, Akizuki S, Downey JM (1984) The role of vascular capacitance in the
coronary arteries. Circ Res 55:751-762
25. Myers WW, Honig CR (1964) Number and distribution of capillaries as determinants of myocardial
oxygen tension. Am J Physiol 207:653-660
26. Smaje LA, Fraser PA, Clough G (1980) The distensibility of single capillaries and vessels in the cat
mesentery. Microvasc Res 20:358-370
27. Spaan JAE, Breuls NPW, Laird JD (1981) Diastolic-systolic coronary flow differences are caused
by intramyocardial pump action in the anesthetized dog. Circ Res 49:584-593
28. Spaan JAE, Breuls NPW, Laird JD (1981) Forward coronary flow normally seen in systole is the
result of both forward and concealed back flow. Basic Res Cardiol 76:582-586
29. Spaan JAE (1985) Coronary diastolic pressure-flow relation and zero flow pressure explained on
the basis of intramyocardial compliance. Circ Res 56:293-309
30. Spaan JAE, Bruinsma P, Vergroesen I, Dankelman J, Stassen HG (1987) Distensibility of
microvasculature and its consequence on coronary arterial and venous flow. In: Sideman S, Beyar
R (eds) Activation, metabolism and perfusion of the heart. Martinus Nijhoff Publishers, pp 389--407
31. Spaan JAE, Dankelman J (in press) Prediction of dynamic transcapiUary pressure difference in the
coronary circulation. In: Sideman S, Beyar R (eds) Analysis and simulation of the cardiac system-
ischemia. CRC Press, Inc
32. Weiss HR, Winbury MM (1974) Nitroglycerin and chromonar on small-vessel blood content of
ventricular wails. Am J Physiol 220:838-843
33. Wiederhielm CA (1965) Distensibility characteristics of small bloodvessels. Fed Proc 24:1075-1084
34. Wiisten B, Buss DD, Deist H, Schaper W (1977) Dilatory capacity of the coronary circulation and
its correlation to the arterial vasculature in the canine left ventricle. Basic Res Cardiol 72:636-650

Received May 20, 1988

Authors' address:
Prof. Dr. Ir. Jos A. E. Spaan, Department of Medical Physics, Academic Medical Center, Meiberg-
dreef 15, 1105 A Z Amsterdam, The Netherlands