.~l.lh/ ('ompul. ,~l.del/ing, Vol. 11, pp. 232 234, 1988 0895-7177/88 $3.00 + 0.

00
Printed in Great t~;ritain PergamorkPrcss plc *

MATHEMATICAL MODELLING IN CARDIOLOGY

MODELS AND CONCEPTS OF DIASTOLIC MECHANICS: PITFALLS IN THEIR

MISAPPLICATION

Ares Paslpoularides, M.D., Ph.D.*; Israel Mirsky, Ph.D.**

~(:ardiology Service, Department of Medicine, Brooke Army Medical Center,
Fort Sam Houston, Texas USA; **Department of Medicine, Harvard Medical
School, Brigham and Women's Hospital, Boston, Massachusetts USA

Abstract. Mathematical models have been remarkably effective in th~ study

of diastolic ventricular mechanics. However, significant pitfalls are
found in the misapplication of models and concepts, especially when proper
attention is not paid to all the variables that affect the range of their
applicability. Models incorporate two ideas: the measurement of variables
a~d parameters, and their interrelationships. With regard to measurement,
due attention is not always paid to the frequency response characteristics
required in signal acquisition and processing. This results in waveform
distortions and spurious time delays. These should be avoided or appropri-
ate corrections made. The inattention to the range of applicability of
w~rious concepts and models in the literature represents a source of
misleading conclusions and controversy. This is demonstrated by means of
several examples.

Keywords. Myocardial relaxation; active stress decay; diastolic mechanics;

myocardial stiffness; cardiac mechanics; ventrlcular function.

Both cardiow~scular mechanicists and clinical configuration" (Brecher, 1958).Subaequently

cardiologists have in recent years grown
increasingly aware of the important reciprocal
influence that exists between the diastolic
behavior of the left ventricle and its systolic
performance. Thus, considerable Interest in
modeling mechanisms underlying dynamic dias-
tolic pressure-volume relationships of the left
ventricle has developed. The subject has
remained controversial, however, to a large
part because diastolic models are used while
proper attention is not paid to all the vari-
ables and conditions that affect the range of
their applicability and validity.
Models incorporate two ideas: the measurement
of variables and parameters, and their inter-
relationships. With regard to measurement, due
attention is not always paid to the frequency
response characteristics required in signal
acquisition and processing. This results in
waveform distortions and spurious time delays.
These should be avoided or appropriate
corrections made.
The inattention to the range of applicabillty
of various concepts and models in the litera-
ture represents a source of misleading con-
clusions and controversy. We will consider
several notable examples in the context ~f a
comprehensive model for diastolic mechanics
that we have published recently (Pasipoularldes
et al., 1986).
In early diastole, after mitral valve opening,
left ventricular pressure continues to fall
although the chamber is expanding. The early
diastolic deviation from what is expected of a
purely passive distended elastic chamber had
precluded a better understanding of the mechan-
ical behavior of ventricular muscle throughout
the entire diastole. Some investigators
construed declining ventricular pressure and
wall stresses in the face of rising chamber
dimensions in the rapid filling phase as indi-
cating that the ventricle consistently con-
tracts down to end-systolic volumes beyond that
corresponding to its "elastic equilibrium
i t a c t i v e l y sucks up blood ~n t h e , e a r l y filling
p e r i o d , a s i t r e c o i l s back toward e l a s t i c
equilibrium.
Gauer and Henry (1964) were able to produce
evidence of suction only in animals with severe
hypovolemia due to exsanguination or negative
gravitational forces which induced e "patho-
logically small ventricular blood content".
Such low operating volumes ~o not apply~
ordinarily in subjects studied in recumbency
under basal cardiac catheterization conditions,
and thus suction is unlikely to be responsible
for the declining pressure during early
filling.
We have developed an alternate model for
diastolic dynamics, in which We account for
declining measured ventrlcuiar pressure and
total wall stresses in the face of rising
chamber dimensions in th~,rapld filling phase
by continuing contractile~wallstress decay
(Pasipoularides et al., 1986). Since wall
relaxation does not cease abruptly nor is it
complete at mltral valve opening, the net
passive diastolic filling pressure, P*, is
given by the difference between the total
measured pressure, PM and tl#e values of the
relaxation pressure~ P., theft would apply if
the ventricle had cont~nue6L~to relax isovolum-
ically. Moreover, the pass~Ye stress, G,
needed to assess passive stiffness levels over
the entire filling period Is obtained by
using the difference between measured and
relaxation pressures instead of the measured
pressure by itself in the general fbrmulas for
ventrlcular wall stresses (Paslpoularldes, et
al., 1986).
To obtain the relaxation pressure, PR(t), we
make use of the concept of a relaxation time
constant of isovolumlc pressure fall
(Weisfeldt, et al., 1978; C~alg, Murgo,
Paslpoularldes, 1987; Cralg,: Murgo, 1980).
The relaxation time constant, T, is determined
from the monoexponentlal POrtion of the
isovolumlc left ventrlcular pressure decay

232
 Proc. 6th Int. Conf: on Mathematical
following maximum negative dP/dt, and after the
ventricular pressure transients associatred with
aortic valve closure (Pasipoularides et al.,
1987). The lower cut-off pressure data point
used in the regression should exceed the
subsequently attained end-diastolic pressure
level by at least 3 mmHg, to allow for the
possibility that the isovolumic relaxation
period might be shorter than is generally
accepted.
From time t=O, corresponding to PO, the first
point on the exponential portion of the iso-
volumic pressure curve, up to the time of
mitral valve opening, the left ventricular
pressure decay conforms to the equation:
-t/T + p
PR(t)=PO e B (1)
where the subscript R identifies the decaying
variable as "relaxation pressure" and PB is an
asymptote (Craig, Murgo, Pasipoularides, 1987;
Craig, Murgo, 1980). Beginning with mitral
opening, the rate of decline of the pressure
that is actually measured, P toward its
diastolic minimum falls shorV' of the rate given
by the above equation. This follows because PM
values in the filling ventricle reflect not
only the ongoing relaxation process, but also
the simultaneous passive pressure buildup
associated with filling. As time of mitral
opening, we thus take the nominal instant at
which the measured pressure P begins to
diverge from the decaying relixation pressure
pR'
Experimental validation for the preceding
fundamental ideas has been provided by Yellin
and his associates. Using the elegant mitral
inflow occlusion technique in the intact beat-
ing ventricle of the dog, these investigators
(Yellin et al., 1981) demonstrated that when
pressure traces from filling and nonfilling
beats are superimposed, with the onset of
filling, the observed early diastolic pressure
decay slows down markedly compared to its rate
in the absence of filling. This observation
validates the idea that the deviation from the
relaxation pressure (P ) in the presence of
filling is due to the 3evelopment of the
passive filling pressure component. After
subtracting the decaying active pressure or
stress component from the total, passive fill-
ing pressures and stresses are shown by our
model to increase continuously throughout fill-
ing, as required by a purely passive process.
Suction needs not to be evoked to account for
decaying measured pressure in the face of
chamber expansion.
Another common problem is the misapplication
of the monoexponential isovolumic pressure
decay model, even where it is inadmissible, to
assess a time constant of relaxation. Recent
publications have addressed this problem (Craig,
Pasipoularides, 1986; Schuurbiers,
Hugenholtz, Brewer, 1984). Briefly, the
pressure transients associated with aortic
closure (Pasipoularides et al., 1987) should be
avoided. Moreover, the monoexponential decay
model should not be applied when the isovolumic
fall follows a more complicated time course
(Pasipoularides et al., 1985).
Although in principle the relaxation model of
Pasipoularides et al., (1986) is applicable
irrespective of the form of the function to
which the relaxation pressure is found to
Modelling 233
conform, in practice the relaxation pressure
function must be modified appropriately before
the model is applied to situations in which the
isovolumic pressure decay does not conform to a
monoexponential. The pressure fall during
isovolumic relaxation does not conform ade-
quately to a monoexponential decay in a number
of clinically important disease states, such as
regional ischemia associated with segmental
coronary disease and hypertrophic cardio-
myopathy. Evidence that lack of a monoexpo-
nential pressure decay may be directly
connected to asynchronous relaxation has been
provided both in a canine preparation (Craig,
Pasipoularides, 1986) and in patients
undergoing percutaneous transluminal
angioplasty during transient coronary
occlusion, when there are striking patterns of
asynergic segmental wall motion during
isovolumic relaxation (Brewer, Meij, Serruys,
1983). The mathematical model developed by
Brewer and co-workers (1983) and Serruys et al.
(1984) shows that a biexponential pressure
decay is produced by a strongly asynchronous
left ventricular relaxation. If the two-time
constant model they described is correct, in
the presence of a biexponential isovolumic
pressure decay resulting from asynchronous left
ventricular relaxation, it is the second
(faster) exponential decay process that should
be used to obtain the relaxation pressure and
stresses of our comprehensive model
(Pasipoularides et al., 1986).
Consider now the question of load-dependence
of relaxation after mitral valve opening
(Brutsaert, Housmans, Goethals, 1980). Load-
dependence of relaxation could cast doubt on
the validity of our method of extrapolating the
isovolumic pressure decay after mitral opening,
in order to obtain the passive filling pressure
by subtracting the relaxation pressure from the
measured pressure (Pasipoularides et al.,
1986). However, consider the applying
conditions: A load-independent relaxation at
or after mitral opening could be associated
with the then applying low rates of activation
decay. Extrapolations from findings at high
(systolic) pressures to events at or after
mitral opening fail to recognize that the decay
rate of an exponential is proportional to its
instantaneous values. Relaxation pressure
values at or afer mitral opening are low,
leading to correspondingly low rates of
activation decay. That load-dependence of
relaxation may be absent when activation decay
rate is low (after caffeine) has been
established (Poggesi et al., 1983). Thus load-
dependence may be manifest when elongation of a
muscle occurs at or soon after the peak of the
contraction, when the applying rates of active
tension decay are high, and absent later on
when tension levels and their decay rates are
low, as was shown by Poggesi et al. (1983).
Final illustration of problems which accrue
when the range over which a variable applies is
not considered is the following: we have shown
that when high late diastolic stress levels are
attained, passive stiffness-stress plots are
bilinear (Pasipoularides et al., 1986). The
bilinearity of the passive stiffness-stress
plots when high late-diastolic stress levels
are attained brings forth the need to always
specify the stress range over which a particu-
lar elastic stiffness constant applies, even in
a given ventricle. This bilinearity is
probably an expression of the ensuing strong
recruitment of stiff fibrocollagenous composite
234 Proc. 6th Int. Conf. on Mathematical Modelling

wall elements at high diastolic stress levels,
and is not discernible if a single stiffness
constant (unique stiffness-stress relation) is
assumed to cover both the lower and the high
filling stress ranges. Thus, different
conclusions may be derived if evaluation and
comparison of stiffness constants are made with
due attention to large differences in applying
filling stress ranges than if such differences
are overlooked.
We conclude that mathematical models can be
effective in the study of diastolic ventricular
mechanics. However, significant pitfalls are
found in the misapplication of models and
concepts, especially when proper attention is
not paid to all the variables and operating
conditions that affect the range of their
applicability and validity.
REFERENCES
Brecher, G.A. (1958). Critical review of
recent work on ventricular diastolic
suction as a mechanism of ventricular
filling. c. I&., 5, 554.
Brewer, R.W., S. Meij, and P.W. Serruys (1983).
A model of asynchronous left ventricular
relaxation predicting the bi-exponential
pressure decay. Cardiovasc. w., 17, 482.
Brutsaert, D.L., P.R. Housmans, and M.A.
Goethals (1980). Dual control of
relaxation: its role in the ventricular
function ii the mammalian heart. -*
Circ
=., 47, 637.
Craig, W.E., and J.P. Murgo (1980). Evaluation
of isovolumic relaxation in normal ma"
during rest, exercise, and isoproterenol
infusion. Circulation 62 (Suppl II), $?._
inactivation dynamics to the impairment of
relaxation in hypoxic cat papillary muscle.
&n. 2. Physiol.; Regulatory Integrative
Camp. Physiol., 248, R54-R62.
Poggesi, D., C. Reggiani, R. Bottinelli, L.
Ricciardi, and R. Minelli (1983).
Relaxation in atria1 and ventricular
myocardium: activation decay and different
Basic Res. Cardiol., 2,
load sensitivity. --~
256.
Schurbiers, J.C.H., P.G. Hugenholtz, and R.W.
Brewer (1984). Left ventricular function
during transluminal angioplasty: a
hemodynamic and angiographic study. Acta
Med. Stand. (Suppl), 694, 197.
--
Serruys, P.W., W. Wijns, M. van den Brand, S.
Meij, C. Slager, J.C.H. Schuurblers, P.G.
Hugenholtz, and R.W. Brewer (1984). Left
ventricular performance, regional blood
wall motion, and lactate metabolism during
transluminal angiography. Circulation,
2, 25.
Weisfeldt, M.L., J.W. Frederiksen, F.C.P. Yin,
and J.L. Weiss (1978). Evidence of
incomplete left ventricular relaxation in
the dog: prediction from the time constant
for isovolumic pressure fall. --
J. Clin.
62 1296.
Invest., _,
Yellin, E.L., C. Yoran, M. Hori, S. Gabbay,
E.H. Sonnenblick, and R.W.M. Frater (1981).
Analysis of left ventricular relaxation in
the filling and transiently non-filling
(completely isovolumic) intact canine
heart. Fed. w., 4Q, 486.

Craig, W.E., J.P. Murgo, and A. Pasipoularides

(1987). Calculation of the time constant
of relaxation. In: W. Grossman (ed.),
Proceedings of International Symposiums The
Physiology ofDiastole in Health and -
Disease. Section III: Evaluation of
Relaxationandmznce in the Intact
--~
Heart. -

Craig, W.E., and A. Pasipoularides (1986).

Ventricular diastolic dynamics: Effects of
wall asynchrony on global relaxation
indices; In: Proceedings ---
of the 21st
Annual Meeting --
of the Association --
for the
Advancement -~
of Medical Instrumentation.
Chicago.

Gauer, O.H., and J.P. Henry (1964). Negative

acceleration in relation to arterial oxygen
saturation. Aerospace Med., 21, 533.

Pasipoularides, A., I. Mirsky, O.M. Hess, J.

Grimm, and J.P. Krayenbuehl (1986). Muscle
relaxation and passive diastolic properties
in man. Circulation, 76, 991-1001.

Pasipoularides, A., J.P. Murgo, J.W. Miller,

and W.E. Craig (1987). Nonobstructive left
ventricular ejection pressure gradients in
ma". w. *. (in press).

Pasipoularides, A., I. Palacios, W. Frist, S.

Rosenthal, J.B. Newell, and W.J. Powell,
Jr. (1985). Contribution of activation-