Vol. 84, Issue 3, 868-876, March 1998
Physiological control of a total artificial heart: conductance-
and arterial pressure-based control
Y.
Abe1,
T.
Chinzei1,
K.
Mabuchi2,
A. J.
Snyder3,
T.
Isoyama2,
K.
Imanishi2,
T.
Yonezawa2,
H.
Matsuura2,
A.
Kouno1,
T.
Ono1,
K.
Atsumi1,
I.
Fujimasa2, and
K.
Imachi1
1 Institute of Medical
Electronics, Faculty of Medicine, and
2 Research Center for Advance
Science and Technology, University of Tokyo, Tokyo 113, Japan; and
3 Department of Surgery, College
of Medicine, The Pennsylvania State University, Hershey, Pennsylvania
17033
 |
ABSTRACT |
To obtain a physiological response by a total artificial heart
(TAH), while eliminating the hemodynamic abnormalities commonly observed with its use, we proposed the use of a conductance- and arterial pressure-based method (1/R control) to determine TAH cardiac
output. In this study, we endeavored to make use of a variable more
closely tied to central nervous system (CNS) efferents, systemic
conductance, to provide the CNS with more direct control over the
output of the TAH. The control equation that calculates the target
cardiac output of the TAH was constructed on the basis of measurement
of blood pressures and TAH flow. The 1/R control method was tested in
TAH-recipient goats with an automatic method by using a microcomputer.
In 1/R control animals, the typical TAH pathologies, such as mild
arterial hypertension and substantial systemic venous hypertension, did
not occur. Cardiac output varied according to daily activity level and
exercise in a manner similar to that observed in natural heart goats.
These results indicate that we have determined a control method for the
TAH that avoids hemodynamic abnormalities exhibited by other TAH
control systems and that exhibits physiological responses to exercise
and daily activities under the conditions tested. The stability of the
control and the complete lack of inappropriate excursions in cardiac
output is suggestive of CNS involvement in stabilizing the system.
central nervous system; 1/R control method
| |
INTRODUCTION |
NO ORTHOTOPIC TOTAL ARTIFICIAL HEART (TAH) developed to
date is sensitive to either central nervous system (CNS) or endocrine signals. How a TAH should be controlled in the absence of these signals
has been the subject of great debate since the beginning of artificial
heart research.
Early TAH developers most commonly applied Starling's law to their
designs (9). The Starling concept is important in postsurgical patient
care and thus was attractive to the cardiac surgeons who typically led
the development teams. Cardiac output (CO) dependence on central venous
pressure (CVP) alone was easily implemented in artificial heart
systems, often without the use of pressure transducers or electronic
controls. Long-term survival of experimental animals was achieved by
using pumps controlled according to venous pressure alone (8, 19, 20).
The animals generally suffered from hemodynamic abnormalities,
including both systemic venous hypertension sufficient to cause marked
hepatic congestion and slight arterial hypertension (8, 19,
20). Some believed that these problems were related to hypoperfusion or insufficient Starling sensitivity (19), cardiac receptor dysfunction (20, 11), or simply poor fit of the devices within
the chest. Imachi et al. (12), however, showed that in adult goats COs
necessary to maintain normal CVP were at times twice the normal level
and that the animals in which the TAH output was elevated to maintain
low venous pressures succumbed within 2 wk to peripheral
circulatory insufficiency. Furthermore, they showed that long-term
survival could be obtained by using normal CO regardless of CVP (7).
Pierce et al. (16) introduced peripheral vascular resistance as a
correlate of metabolic demand and obtained the long-term survival by
changing the TAH output according to changes in the aortic pressure
(AoP). In this afterload-based system, CO was increased as the AoP
decreased below the normal value (17). The elevation of CVP, however,
was not prevented (5, 18).
In exercise, good results were obtained by using predictive methods in
which CO was controlled according to the time profile of the natural
heart's output in a normal animal undergoing the same level of
exercise (10) or by using body motion as measured by acceleration
sensors as an indication of exercise level (14). Neither of these
methods, however, accounted for changes in CO that naturally occur
independent of the exercise state.
Thus, a means of controlling the output of a TAH, in which the neurally
and hormonally isolated device responds appropriately to metabolic
demands and in a manner that results in the absence of hemodynamic
abnormalities over the long term, has not been demonstrated in the
past. In this study, we describe a method of controlling a TAH by
monitoring pressure signals and TAH flow but in a manner that appears
to provide a sufficient link between CNS efferents and CO to permit CNS
control of the heart, resulting in elimination of hemodynamic
abnormalities from and allowing a physiological response by the
TAH.
| |
METHODS |
Control method (1/R control).
The goal of the new control method is to provide the CNS, specifically
the cardiovascular center, with control over the output of the TAH.
Direct access to nerve signals was judged to be impractical at this
time. As an alternative, we attempted to derive a signal, based only on
measurement of blood pressures and knowledge of TAH flow, that would be
closely tied to CNS efferents. It was our expectation that, by using
such a signal to control TAH output, a control loop as described in
Fig. 1 could be constructed and the CNS
would become capable of controlling the artificial device.
It has been suggested (6) that significant plasticity in the autonomic
nervous system exists and may even be important in tuning physiological
regulatory loops. If this is the case in the most general sense, one
might expect that merely tying CO as directly as possible to an
autonomic efferent will afford the CNS the ability to control the TAH
as necessary. Given 1) the
controversial nature of such claims,
2) the need for the TAH to provide
acceptable flows immediately on implantation, without a waiting period
for adaptation to occur, and 3) the
fact that major changes in autonomic efferents to properly control the
heart replacement might cause intolerable disruptions in vascular tone, we chose to avoid undue reliance on such principles. We endeavored to
tie TAH output to CNS efferents in as similar a manner as possible to
the natural control of the heart, to the extent this could be done in a
reasonably simple system.
We determined that peripheral vascular resistance should be a good
control signal. Total peripheral resistance changes markedly with
changing CO, and these changes are mediated by the cardiovascular center, thus providing the link we required between artificial heart
output and the relevant CNS center. In the intact natural heart system,
vascular conductance remains roughly proportional to CO, so it is
natural to use conductance (1/R; i.e., the reciprocal of the
resistance), as the variable that determines CO in the artificial
heart. Finally, it is possible, if challenging, to obtain arterial
pressure, right atrial pressure (RAP), and flow measurements needed to
compute the resistance.
Preliminary studies with pure conductance control were encouraging but
revealed anomalous operations related to nonmetabolically mediated
changes in peripheral resistance. Specifically, in animals that were
frightened, e.g., when approached from the rear or with a hypodermic
needle and, occasionally, in those that stood after a period of rest,
peripheral resistance would rise and CO would fall accordingly. This
was thought to be due to independent vasoconstrictive and
vasodilatative effects (e.g., 
- and 
-adrenergic effects).
To avoid undue reliance on autonomic plasticity, and avoid, as
discussed above, creating a situation in which the chosen
CO was obtained at the expense of anomalous vascular tone, we attempted to isolate the vasodilatative effects by using a parallel circuit conductance model. We express the total peripheral vascular conductance (1/TPR) as
|
(Eq. 1)
|
where
(1/R)0 is baseline conductance and
(1/R)
and
(1/R)
are changes in
conductance due to vasoconstrictive and vasodilatative efferents,
respectively. Then, the requirement that
|
(Eq. 2)
|
is
satisfied by
|
(Eq. 3)
|
where
COTAH is the output required for
the TAH,
f(p,t)
is a gain, dependent on pressures and time, to be determined later, and
f() is a correction term
( correction) intended to cancel the effect of vasoconstriction on
total conductance as given in Eq. 1.
Substituting measured AoP, RAP, and flow (CO) for TPR
|
(Eq. 4)
|
For the gain function,
f(p,t),
we chose a set-point perfusion pressure
|
(Eq. 5)
|
so
that Eq. 4 becomes
|
(Eq. 6)
|
AoPset
was taken to be the set point of baroreceptors and was determined by
heavily filtering the measured AoP: at each measurement interval
(T)
|
(Eq. 7)
|
where
the time constant 
was set to 12 h in an attempt to capture the
circadian rhythm in arterial pressure changes.
AoPset was initialized to a
nominal value (100 mmHg) when the control system was started.
RAPset was taken to be the
aggregate set point of the autonomic control loops involving atrial
afferents and the atrial secretion of atrial natriuretic polypeptide
and was determined at the time the control system was set up.
By using the approximation that vasoconstriction mainly affects
arterial pressure, compensation for outflow was made proportional to departure from AoPset
|
(Eq. 8)
|
where
BW is body weight. Substituting into Eq. 6 gives the final expression for the output of the TAH
|
(Eq. 9)
|
where
AoPset is determined according to
Eq. 7, and
RAPset and CP are free parameters
to be determined experimentally.
Preparation.
Adult female goats weighing 38-54 kg were used for the
experiments. The TAH was implanted under extracorporeal circulation. The natural heart was resected at the atrioventricular groove. Pump
inflow cannulas were attached via atrial cuffs sutured to the remnant
atria. The pulmonary outflow cannula was inserted into the pulmonary
artery and ligated. The systemic outflow cannula was anastomosed end to
side with the descending aorta; the ascending aorta was closed by using
an arterial clamp. Pneumatically driven blood pumps (4) with a sac
volume of 60 ml (Nihon Zeon, Tokyo, Japan) were connected to the
cannulas and set outside the body on the chest wall (Fig.
2). A computer-controllable artificial heart drive unit (Corart 103C; Aisin Seiki, Aichi, Japan) was used.
View larger version (61K):
[in this window]
[in a new window]
|
Fig. 2.
Schematic diagram of TAH system used in present study. LAH and RAH,
left and right artificial heart ventricles, respectively; EMF,
electromagnetic flowmeter; Ao, PA, LA, and RA: aorta, pulmonary artery,
and left and right atrium, respectively; AoP, CO, LAP, and RAP: aortic
pressure, cardiac output, left atrial pressure, and right atrial
pressure, respectively.
|
|
Hemodynamic data were collected from three groups of goats.
1) In four goats (identified by
their experimental nos. as goats 9107,
9111,
9206, and
9209), the TAH was operated by using
1/R control, as described above. We refer to this as the 1/R TAH group. 2) In three goats (goats
8811, 9005, and
9010), the TAH was set for a
constant rate and pneumatic drive pressures. In these animals, CO
necessary to obtain a pulmonary flow of 80-100
ml · min
1 · kg1
at the time of implantation was computed, and the rate and drive pressures were set to obtain that flow rate. Thereafter, passive changes in output with changing preload and afterload were possible, but only over a limited range; CO is nearly constant. For simplicity, we refer to this as the constant-CO TAH group.
3) In three goats (goats
9002, 9007, and
9210), fluid-filled catheters and
electromagnetic flow probes were placed, but the natural heart and its
innervation were left intact. We refer to this as the natural heart
group.
Measurement techniques.
In the TAH recipients, CO was measured at the outlet cannula of the
left ventricle by using an electromagnetic flowmeter (Nihon Koden,
Tokyo, Japan), and the AoP, left atrial pressure (LAP), and RAP were
measured by using pressure transducers (Nihon Koden) through
fluid-filled side catheters built into the outlet port of the left pump
and the atrial cuffs (Fig. 2). In the intact natural heart goats, CO
was determined by electromagnetic flow measurement (Nihon Koden) at the
ascending aorta, and pressures were measured via fluid-filled catheters
at the descending aorta, left atrium and right atrium, and
by pressure transducers (Nihon Koden). Heart rate was determined
via an electrode at the right atrial appendage. In all animals, all
pressure lines received slow, continuous infusions of heparinized
saline to prevent clotting. Pressure transducers were affixed
externally to the chest wall so that they remained near the level of
the heart as the animal moved about.
Hemodynamic data were low-pass filtered ( = 3 s) to cancel the
vibration caused by the movement of the animal or water hammer effects
of the mechanical valves. The signals were then collected through an
analog-to-digital converter into the microcomputer (PC9801VM2, NEC,
Tokyo, Japan), and a mean of each was computed for each
beat period. The averaged data were sampled every 2 s for analysis,
implementation of the automatic control, and calculation of
AoPset.
Collection of hemodynamic data for comparison among groups began after
at least 2 wk of recovery from surgery. Data were collected during
normal daily activities and during exercise sessions. The hemodynamic
response with exercise was examined by using 3 min of treadmill
exercise with treadmill speeds of 1.33, 2.88, and 4.6 km/h.
Calculation of CP.
The correction, CP, was calculated by using data obtained from the
three natural heart goats. Equation 9
was recast as follows
|
(Eq. 10)
|
where
and 
are the
mean AoP and RAP over the entire data-recording session,
respectively. COcalc is the CO
predicted by the control equation.
All data were averaged over 30-s intervals to remove any fluctuations
due to the respiratory cycle. The 30-s average data were picked up
every 3 min for the calculation of CP. By numerical means, we
determined the value of CP that minimized the discrepancy between
COcalc and CO.
1/R control protocol.
The pneumatically driven sac-type blood pumps were actuated by
alternate application of positive air pressure to effect systole, followed by negative pressure to effect diastole. Changing the positive
and negative drive air pressures changes the rates at which the sacs
collapse and refill. As the beat rate is varied, so is the amount of
time spent in systole and diastole, and thus the stroke volumes of the
pumps are affected. As the beat rate was varied, the stroke volume was
controlled by titrating the pneumatic drive pressures. The percent
systole was fixed in each animal at 45-50. This narrow range was
sufficient to compensate for the small variations in surgical placement
of the inflow and outflow cannulas, which caused some variation among
subjects in the relative ease of ejection and filling of the pumps.
Left and right pumps were run at identical rates. The pump rate and
pneumatic drive pressures were controlled by the same microcomputer
used to collect the hemodynamic data (Fig.
3).
View larger version (32K):
[in this window]
[in a new window]
|
Fig. 3.
Schematic diagram of automatic control system of TAH. L, left; R,
right; 1/R, conductance- and arterial pressure-based method.
|
|
The left blood pump was maintained at a constant stroke volume,
SVset. Unlike the natural heart,
the artificial heart requires active control over left-right balance.
Balance was enforced by manipulation of the right pump stroke volume to
maintain a fixed relationship between LAP and RAP
|
(Eq. 11)
|
where
a and
b are constants determined in each
animal to keep LAP below 10 mmHg. In the absence of abnormally high
RAP, a was set to one and
b was set according to the difference
in the height of the two atria in the individual animal, to keep the filling pressures of the left and right pumps equal.
Both enforcement of the constant left-pump stroke volume and control of
right-pump stroke volume to maintain the left-right balance
relationship of Eq. 11 were achieved
by using the following control logic
where
SV refers to the measured stroke volume of the left pump. In essence,
the control addresses a discrepancy in LAP by adjusting the left or
right pump, depending on whether the left-pump stroke volume is above
or below the desired value. The balance and stroke volume control
algorithm was executed every 2 s.
With the left-pump stroke volume kept constant at
SVset, the CO was set by
calculating the required CO according to Eq. 9 and setting the pulse rate to
|
(Eq. 12)
|
where
PR is pulse rate. A new CO value was calculated and a new pulse rate
was set in the artificial heart drive unit every 6 s.
| |
RESULTS |
Determination of CP.
Figure 4 shows correlation of measured with
COcalc for various values of CP.
As Fig. 4 shows, CP values from 0.8 to 1.1 resulted in the best
regression results (from 0.912 to 0.955) and the closest agreement
between calculated and observed values (slope close to 1 and intercept
close to 0).
View larger version (35K):
[in this window]
[in a new window]
|
Fig. 4.
Results of calculations by using data of natural heart goats:
goats 9002 (left),
9007 (middle), and
9210 (right).
A-C: results with correction (CP) = 0; D-F: results with CP = 0.8-1.1.
|
|
Control stability.
In any control system, particularly one involving multiple control
loops and implemented in a sampled fashion, control stability is a
potential problem. In the four 1/R control experiments, control system
function was eventually lost due to obstruction of one or more pressure
measurement catheters. In these 1/R TAH animals, stable operation of
the control system was maintained for 65, 21, 41, and 340 days, as
described herein.
Stable control required CP values between 0.4 and 0.8. When CP was
larger than 0.8, large excursions in CO, both hyper- and hypoperfusion,
were seen in response to the small changes in pressure transducer
offset that occurred with postural changes. When CP was <0.4, marked
hypoperfusion occurred in response to the sharp rises in AoP that
sometimes occurred during urination, receipt of injections, or at the
onset of treadmill exercise. CP was typically fixed at 0.6 because
stable operation was obtained in all animals at this setting.
Figure 5 shows the correlation of 24-h mean
RAP with RAPset. As indicated at
the beginning of this study, aberrant regulation of extracellular
volume, characterized by steadily increasing CVP, is regularly observed
in TAH-recipient animals. In the 1/R control subjects, the RAP reliably
stabilized at a value related to
RAPset, as indicated by the narrow
error bars in Fig. 5. RAPset values of 4-12 mmHg resulted in RAP values consistently below 15 mmHg. With RAPset above 12 mmHg,
excursions in measured RAP above 20 mmHg were observed. With
RAPset below 4 mmHg, the
excursions in measured RAP below zero were encountered, resulting in
insufficient filling pressures to maintain the required CO. Typically,
RAPset was fixed between 6 and 8 mmHg. At this setting, the right-pump filling pressure was sufficient
to obtain the required CO.
View larger version (14K):
[in this window]
[in a new window]
|
Fig. 5.
Relationship between 24-h mean RAP  and
aggregate set point of autonomic control loops involving atrial
afferents and atrial secretion of atrial natriuretic polypeptide
determined when control system was set up
(RAPset).
n, No. of 24-h mean RAP.
|
|
No pulmonary edema was observed in any of the 1/R control animals. LAP
was well controlled below 10 mmHg with the left-right balance control
strategy described above.
Hemodynamic changes.
Figure 6 shows typical hemodynamic changes
over 24 h in the natural heart, constant-CO TAH, and 1/R TAH goats.
Each data point represents a 30-s average. Points are plotted at 3-min
intervals. The most striking differences among the groups occurred in
CO. In the constant-CO group, changes in CO were very small over a 24-h
period. In the natural heart and 1/R groups, CO tended to be lower when
the goats were at rest and higher when they were standing or eating.
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 6.
Hemodynamic changes in natural heart
(A; goat
9002), constant-CO TAH
(B; goat
9005), and 1/R TAH
(C; goat
9209) goats.
|
|
Figure 7 shows 24 h (means ± SD) of CO,
AoP, and RAP in all goats. There were no significant differences among
groups with regard to mean CO. Significant elevations in AoP and RAP
(both P < 0.01) were observed in the
constant-CO group when compared with the natural heart group. The 1/R
group showed no significant difference from the natural heart group.
In treadmill exercise, goats in the natural heart and 1/R groups
exhibited increases in CO. Figures 8 and
9 show exercise data in a 1/R goat
(goat 9209) and a natural heart goat
(goat 9002). These animals had similar
physiques and body weights (1/R: 48 kg, natural heart: 43 kg). Figure 8
shows the comparative changes in CO, percent increase in heart rate,
and percent change in stroke volume in both goats during the exercise
session. Percent change in heart rate was computed relative to the last
30-s average before the onset of treadmill exercise. Percent change in
stroke volume was computed relative to the last 30-s average before the
onset of treadmill exercise for the natural heart and relative to the SVset control setting in the 1/R
TAH goats.
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 8.
Comparative changes in CO, %increase in heart rate (pulse rate), and
%stroke volume between a 1/R goat (goat
9209) and a natural heart goat (goat
9002) with treadmill exercise. Treadmill speed was
2.88 km/h.
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 9.
Comparative increase in CO (A) and
comparative elevation in AoP (B) and
RAP (C) between an 1/R goat
(goat 9209) and a natural heart goat
(goat 9002) with treadmill exercise.
|
|
CO changes with the 1/R method are strikingly similar to those observed
with the natural heart, although a delay of ~30 s is apparent in the
1/R system relative to the natural heart. No such delay is seen in the
heart rate, but the delay is observed in the stroke volume data. Thus
it appears that the speed of the stroke volume control loop is
responsible for a lag in the 1/R TAH control.
Figure 9 shows relative changes in AoP and RAP, calculated relative to
the last preexercise 30-s average in each animal. At 2.88 km/h, there
was no significant difference between the two goats. At 4.6 km/h, the
TAH had reached its maximum pump output of 8 l/min (167 ml · kg1 · min1
in this goat), less than that provided by the natural heart. Accordingly, there was an increase in RAP in the 1/R goat
(P < 0.05).
| |
DISCUSSION |
It is not presently feasible to provide an artificial heart with the
complete repertoire of natural heart responses because 1) continuous, long-term measurement
of the natural heart's control signals, particularly neural and
endocrine influences, has not been realized, and
2) even if such measurements were
available, we have not yet uncovered a suitably detailed description of
the natural heart's responses to all possible combinations of these signals (i.e., we do not have an accepted set of control equations for
the natural heart). Previously, TAH developers have attempted to
control their devices by using hemodynamic parameters that are thought
to represent or correlate with metabolic requirements, such as RAP
(venous return) and changes in AoP (roughly, peripheral resistance). It
has proved difficult, however, to find a single simply measured
parameter that correlates well with the natural CO over the range of
conditions and states encountered.
The concept of 1/R control grew out of the hypothesis that if we can
provide a means by which the cardiovascular center can control the TAH,
then metabolic requirements will be met, regardless of whether our
control signal is precisely predictive of natural CO. We found vascular
conductance to be a viable candidate for such a control signal
(1-3), when modified in an attempt to respond to metabolically
mediated vasodilation while ignoring transient vasoconstriction related
solely to environmental or physiological factors. Significantly,
although preload-based control systems have been found consistently to
result in CO rising inevitably to the maximum available from the pump
(12), use of 1/R control yielded stable CO and arterial pressures,
which varied according to exercise and daily activity level in a manner
similar to that observed in control animals having intact natural
hearts.
We should emphasize the necessity of CP for the function of 1/R
control. It is this arterial pressure-based correcting term that
isolates vasodilatative responses and distinguishes 1/R control clearly
from previously attempted afterload controls (17). In the control
system previously described (17), sensitivity to afterload changes is
intentionally kept low to avoid unwanted excursions in CO at the onset
or cessation of activity or on changes in posture. Similarly, we
observed unacceptable transient increases or decreases in CO when the
correcting term was absent (or was far too small). Unwanted
transients that occurred at the CP gain predicted by curve fitting to
natural heart data were triggered by positional shifts in pressure
transducers relative to the heart. We expect that, with pressure
sensors located more intimately to the heart, the curve-fit CP value
could have been used. Nonetheless, stable control was obtained with
somewhat lower CP value.
The 1/R system made use of the simplification that there was a fixed
RAPset. It is expected in any TAH
implantation that mechanical disruption of the right atrium can disturb
any natural afferents from that area. Vasku (21) reported receptor
abnormalities in the right atrium after TAH placement in calves. Olsen
et al. (15) and Mabuchi et al. (13) reported abnormalities in atrial
natriuretic polypeptide secretion in TAH-recipient animals. Acceptable
function was obtained by assuming a constant
RAPset. The instantaneous RAP
varied over the course of each day and according to activity, but the
24-h mean remained close to the
RAPset value without the need for
medical intervention. This should be contrasted to the constant CO
results and with prior experience with other systems (5, 7, 8, 11, 12,
18-21), in which RAP was observed to rise to abnormally high
levels and medical intervention was typically necessary to prevent
fluid overload. This is an interesting observation, because, whereas in
the conception of the control system we thought of
RAPset as the aggregate set point
of the natural loop that incorporates the right atrial stretch
receptors, the control system does not explicitly control RAP. Proper
function of the system required that RAP be sufficient to
provide sufficient preload to the TAH so that required CO were
available.
The linear relationship that was used between RAP and LAP resulted in
sufficiently low LAP values that no clinical evidence of pulmonary
edema was detected. The linear relationship was intended to approximate
Starling's law's control over left-right balance that occurs in the
natural heart. This appears to be a sufficiently effective balance
control for the TAH.
It is well known that local autoregulation has an important influence
on blood flow distribution and that, when it involves a sufficiently
large tissue bed, TPR can be affected as well. In attempting to isolate
vasodilatation from the peripheral resistance signal, we have ignored
the influence of local phenomena. This appears not to have adversely
affected the stability of the control system or the appropriateness of
the CO that resulted.
Of greatest significance is the lack of observation of typical TAH
pathologies (mild arterial hypertension, substantial systemic venous
hypertension) in the 1/R TAH recipients, whereas the fixed-rate subjects did exhibit these pathologies despite the lack of any significant deficiency in cardiac index (Fig. 7). The similarity between 1/R and intact heart responses to exercise are striking and
also without precedent in TAH devices controlled by hemodynamic measurements. The limitation of the TAH output to 8 l/min, less than
the capacity of the intact hearts, limited the scope of the exercise
measurements, but the 1/R TAH did, appropriately, remain at its maximum
output under conditions for which the natural heart output was greater.
We are interested in determining to what extent autonomic learning or
adaptation phenomena occurred in these animals vs. our having merely
discovered a more appropriate heuristic than has been used previously.
Because a reasonable CO during the period of postoperative recovery and
adaptation to the TAH is important to the animal's survival, and
because the efferents by which we expected the CNS to gain control over
the TAH influence other important aspects of hemodynamics, we required
that our control signal be reasonably predictive of natural CO. Thus,
although we succeeded in our effort to devise a good physiological
control system for the TAH, we cannot claim to have shown that
autonomic learning or adaptation was involved in these studies.
However, the stability of the control, and the absence of inappropriate excursions in CO once the CP value was adjusted, is suggestive of
active involvement of the CNS in stabilizing the system. This result is
of direct relevance to those interested in devising control systems for
artificial hearts and may also be of significance to those interested
in further elucidating the nature of CNS control over the circulatory
system.
Future studies in development of this TAH control system should
concentrate on obtaining a better understanding of the ranges of
controller parameters for which desirable behavior occurs, in both the
short and long term. Studies of response to more severe exercise,
exceeding the maximum capacity of the artificial heart, will be useful
in verifying the stable behavior of the system at its limits. Studies
involving responses to vasoactive drugs may aid in determining the
degree of duplication of natural cardiac responses, beyond those
situations routinely encountered in normal internally mediated
physiology and routine interactions with the environment. Such studies
must make use of drugs having purely vascular effects so that cardiac
effects in the natural heart control group do not cloud their
interpretation.
| |
FOOTNOTES |
Address for reprint requests: Y. Abe, Institute of Medical
Electronics, Faculty of Medicine, Univ. of Tokyo., 7-3-1 Hongo,
Bunkyo-ku, Tokyo 113, Japan (E-mail: abe{at}bme.rcast.u-tokyo.ac.jp).
Received 30 September 1996; accepted in final form 30 October
1997.
| |
REFERENCES |
1.
Abe, Y.,
T. Chinzei,
K. Imachi,
K. Mabuchi,
K. Imanishi,
T. Isoyama,
H. Matsuura,
G. Senih,
H. Nozawa,
A. Kouno,
T. Ono,
K. Atsumi,
and
I. Fujimasa.
Can total artificial heart animals control their TAH by themselves: one year survival of a TAH goat using a new automatic control method (1/R Control).
ASAIO J.
40:
M506-M509,
1994[Medline].
2.
Abe, Y., T. Chinzei, K. Imachi, K. Mabuchi, K. Maeda, K. Imanishi, T. Yonezawa, K. Atsumi, and I. Fujimasa. Separate circulation with an artificial heart for the study of artificial heart
control. Artif. Organs 4, Suppl. 4: 76-78, 1991.
3.
Abe, Y.,
K. Imachi,
T. Chinzei,
K. Mabuchi,
K. Imanishi,
T. Isoyama,
T. Yonezawa,
A. Kouno,
T. Ono,
K. Atsumi,
and
I. Fujimasa.
Reciprocal of peripheral vascular resistance (1/R) control method for total artificial heart.
In: Heart Replacement. Artificial Heart, edited by T. Akutsu,
and H. Koyanagi. Tokyo: Springer-Verlag, 1992, vol. 4, p. 349-351.
4.
Atsumi, K.,
I. Fujimasa,
K. Imachi,
M. Nakajima,
S. Tsukagoshi,
K. Mabuchi,
K. Motomura,
A. Kouno,
T. Ono,
A. Miyamoto,
N. Takido,
and
N. Inou.
Long-term heart substitution with an artificial heart in goats.
ASAIO J.
8:
155-165,
1985.
5.
Aufiero, T. X.,
J. A. Magovern,
G. Rosenberg,
W. E. Pae,
J. H. Donachy,
and
W. S. Pierce.
Long-term survival with a pneumatic artificial heart.
ASAIO Trans.
33:
157-161,
1987[Medline].
6.
Dworkin, B.
Learning and Physiologic Regulation. Chicago, IL: Univ. of Chicago Press, 1993, p. 130-161.
7.
Fujimasa, I.,
K. Imachi,
M. Nakajima,
K. Mabuchi,
S. Tsukagoshi,
A. Kouno,
T. Ono,
N. Takido,
K. Motomura,
T. Chinzei,
Y. Abe,
and
K. Atsumi.
Pathophysiological study of a total artificial heart in a goat that survived for 344 days.
In: Progress in Artificial Organs-1985, edited by Y. Nose,
C. Kjellstrand,
and P. Ivanovich. Cleveland, OH: ISAO Press, 1986, p. 345-353.
8.
Hennig, E.,
C. Grosse-Siestrup,
W. Krautzberger,
H. Kless,
and
E. S. Bucherl.
The relationship of cardiac output and venous pressure in long surviving calves with total artificial heart.
ASAIO Trans.
24:
616-623,
1978.
9.
Hiller, K. W.,
W. Seidel,
and
W. J. A. Kolff.
Servo mechanism to drive an artificial heart inside the chest.
ASAIO Trans.
8:
125-130,
1962.
10.
Imachi, K.,
T. Chinzei,
K. Maeda,
K. Mabuchi,
Y. Abe,
T. Yonezawa,
M. Asano,
M. Suzukawa,
A. Kouno,
T. Ono,
and
I. Fujimasa.
Predictive control of total artificial heart during exercise.
In: Total Artificial Heart
Problems of Regulation, edited by J. Vasku. Brno, Czech Republic: J. E. Purkyne Univ., 1988, p. 101-110.
11.
Imachi, K.,
I. Fujimasa,
M. Nakajima,
K. Mabuchi,
S. Tsukagoshi,
K. Motomura,
A. Miyamoto,
N. Takido,
N. Inou,
A. Kouno,
T. Ono,
and
K. Atsumi.
Overall analysis of the cause of pathophysiological problem in total artificial heart in animals by cardiac receptor hypothesis.
ASAIO Trans.
30:
591-596,
1984.
12.
Imachi, K.,
I. Fujimasa,
H. Omichi,
I. Mano,
T. Nishisaka,
N. Iwai,
and
K. Atsumi.
Cardiac output control and blood circulation in artificial heart.
Jpn. J. Artif. Organs
5:
321-327,
1976.
13.
Mabuchi, K.,
H. Hayakawa,
Y. Hirata,
M. Iizuka,
K. Imachi,
T. Chinzei,
H. Nozawa,
Y. Abe,
T. Yonezawa,
M. Suzukawa,
K. Imanishi,
K. Atsumi,
T. Sugimoto,
and
I. Fujimasa.
Suppression of the natriuretic effects of exogenous atrial natriuretic peptide in animals with total artificial hearts.
ASAIO Trans.
37:
M214-M216,
1991[Medline].
14.
Maeda, K.,
T. Chinzei,
K. Imachi,
K. Mabuchi,
Y. Abe,
T. Yonezawa,
K. Imanishi,
I. Fujimasa,
and
K. Atsumi.
Predictive control by physical activity rate of a total artificial heart during exercise.
ASAIO Trans.
34:
480-484,
1988[Medline].
15.
Olsen, D. B.,
C. Westenfelder,
G. L. Burns,
C. Kablitz,
and
R. L. Baranowski.
Neurohormonal responses in total artificial heart recipients.
In: Progress in Artificial Organs1985, edited by Y. Nose,
C. Kjellstrand,
and P. Ivanovich. Cleveland, OH: ISAO Press, 1986, p. 112-118.
16.
Pierce, W. S.,
D. Landis,
W. O'Bannon,
J. H. Donachy,
W. White,
W. Phillips,
and
J. A. Brighton.
Automatic control of artificial heart.
ASAIO Trans.
22:
347-356,
1976.
17.
Snyder, A. J.,
G. Rosenberg,
and
W. S. Pierce.
Noninvasive control of cardiac output for alternately ejecting dual-pusherplate pumps.
Artif. Organs
16:
189-194,
1992[Medline].
18.
Snyder, A. J.,
G. Rosenberg,
J. Reibson,
J. H. Donachy,
G. A. Prophet,
J. Arena,
B. Daily,
S. McGary,
O. Kawaguchi,
R. Quinn,
and
W. S. Pierce.
An electrically powered total artificial heartover one year survival in the calf.
ASAIO J.
38:
M707-M712,
1992[Medline].
19.
Takatani, S.,
H. Harasaki,
S. Koike,
I. Yada,
R. Yozu,
L. Fujimoto,
S. Murabayashi,
G. Jacobs,
R. Kiraly,
and
Y. Nose.
Optimum control mode for a total artificial heart.
ASAIO Trans.
28:
148-153,
1982.
20.
Vasku, J.
Perspectives of total artificial heart research as a valuable modelling system for general physiology and pathophysiology.
In: Heart Replacement. Artificial Heart, edited by T. Akutsu,
and H. Koyanagi. Tokyo: Springer-Verlag, 1992, vol. 4, p. 161-171.
21.
Vasku, J.
Total artificial heart research in Czechoslovakia: pathophysiological evaluation of long-term experiments performed from 1979 to 1985.
In: Heart Replacement. Artificial Heart, edited by T. Akutsu,
and H. Koyanagi. Tokyo: Springer-Verlag, 1991, vol. 1, p. 161-179.
JAP 84(3):868-876
0161-7567/98 $5.00
Copyright © 1998 the American Physiological Society
Top
Abstract
Introduction
