Chapter 4: Animal Experiments with the Passive-Filling Heart Replacement Pumps

Using the mechanical pumps that have four characteristics of the heart (passive filling, pulsatile outflow, uninterrupted inflow, excess flow capacity) to replace heart function in dogs, it is possible to demonstrate the significance of those characteristics in determining the following circulation phenomena:

  • Blood volume balancing between the systemic and pulmonary circuits
  • Blood volume's effect on mean cardiovascular pressure and thus on cardiac output
  • Circulation rate control by the peripheral vascular effects of chemical, humeral, and neural stimuli
  • The atrial effect in providing uninterrupted venous flow to the heart
  • Pulsatile blood flow causing diffuse distribution of circulation

The Dog-Pump Model:

The following demonstrations were obtained during cardiac replacement of both right and left ventricles with passive filling pumps in greyhound dogs. One pump was hooked up with a cannula connected from the right atrium to the pulmonary artery; the other was connected from the left atrium to the aorta (Fig. 7). After the pumps were started, the hearts were fibrillated with a low voltage alternating electric current, thereby diverting all of the blood flow to the pumps. The pump rate was set arbitrarily at 80 beats per minute. The ventricles had a maximum capacity of 80 cc. (Figs. 14 & 15).

Blood Volume Balancing Between the
Pulmonary and Systemic Circuits

The model with non-sucking pumps gives an excellent opportunity to demonstrate the automatic blood volume balancing between the two circuits by passive filling pumps. The balancing can be demonstrated by putting shunts between the two circuits, in various locations, that would cause imbalance with other types of pumps.

Figure 16 shows a biopsy of normal appearing lung, obtained after bypassing the heart with the non-sucking pumps for a short time. Figure 17 shows the same lung after 2 hours, where a shunt of 1 liter per minute — a total of 160 liters of blood — was shunted from the aorta to the pulmonary artery. The shunt was a 1/4 inch tube placed between the two arteries. An adjustable roller pump, incorporated in the tubing, produced the shunt. During this time, without any adjustment, the left heart replacement pump automatically put out approximately one liter more blood per minute than the right pump, thereby maintaining the blood volume balance between the two circuits. The function of the lungs remained normal in the presence of great disparity in circulation rate between the systemic and pulmonary circuits.

Contrasted with this experiment was one where the heart was replaced with two roller pumps, adjusted so that the output of each was identical. When a 1 liter per minute right-to-left shunt was placed in the circuit, within 15 seconds the lungs became so engorged and hemorrhagic that they bled from the bronchi and appeared beefy red and consolidated like liver (Fig. 18 and Fig. 19).

With the passive filling pumps, reversing the great vessel shunt and introducing shunts at other locations all failed to create a disequilibrium between the blood volumes in the systemic and pulmonary circuits.

Figure 20 is a record of the effect of adding fluid to the systemic venous system on blood flow from the pumps. The two lower superimposed tracings are the flow rates of the right and left passive filling pumps. At the beginning of the graph, the flow to the pulmonary artery is slightly more than the flow to the aorta. Shortly thereafter, the pulmonary artery flow is slightly less. At point A of Figure 20, 200 cc. of blood is rapidly injected into a systemic vein. At point B, the right pump puts out more volume than the left, adding the injected fluid to the previous blood flow. When the bolus of fluid reaches the left pump (Fig. 20, from B to C), the right passive filling pump automatically has a greater output. At point C, the increased blood volume caused by the fluid addition is redistributed back to a normal equilibrium, with the pulmonary and systemic vascular beds each having received their representative volume expansion by the transfused fluid.

This is a demonstration of how intravenous infusions and transfusions, even though introduced into only one circuit, do not cause an imbalance in the blood volumes between the lungs and the rest of the body. The new equilibrium occurs automatically, without any sensing devices or neuro-chemical feedback mechanisms, as the non-sucking pumps have none. With passive filling pumps, it makes no difference where the fluid is introduced, no disequilibrium can occur.

Single Ventricle Replacement:

The right ventricle was bypassed with a pump from the right atrium to the pulmonary artery. The pulmonary artery was clamped between the pulmonary valve and the pulmonary artery cannula, thereby diverting all of returning systemic blood to the pump.

With the right ventricle replaced, circulation continued in a normal fashion, with the systemic-pulmonary blood equilibrium staying unaltered. The lung's function and appearance remained normal in all but one case.

A complication occurred the second time this experiment was done on a dog, which prevented the disaster that might have resulted if it had been used in a human heart surgery case. The single right heart bypass, with the left heart still functioning, was anticipated to be used in cases of pulmonary valvular stenosis. Five minutes into that experiment, the heart developed ventricular fibrillation; the pump didn't. The fibrillating ventricle of course quit its pumping, so no blood left the lungs thereafter. The pump, in the next three to four beats, overloaded the lungs with enough blood to give them the appearance of liver. The result was total loss of pulmonary function with no chance of recovery of the animal.

This experience not only prevented the use of this procedure in human heart surgery, but gave one other significant benefit: The exposed hazard made it clear what the order of cannulation should be in two ventricle bypass. The left heart should be cannulated first and the pump started. If the heart should happen to fibrillate before the right cannulation was completed, pumping all of the pulmonary blood into the systemic circuit would not be disastrous because of the lung's comparative low blood volume. Then the right cannulation can be done, and total bypass then initiated.

The Relationship of Blood Volume to Mean Cardiovascular
Pressure and Cardiac Output

Figure 21 is a baseline graph obtained from a dog with his heart function replaced by two passive filling pumps. His vascular volume is normal. The pressures were taken without the presence of any vascular stimulants or depressants, except for a pentothal anesthetic. In order to determine the mean cardiovascular pressure, the pumps were turned off for 30 seconds. In 10 seconds the venous pressure had risen to 16 cm. of water pressure, and the arterial pressure had dropped from 125/55 to 16 cm. as well. This pressure — 16 cm. of water — that equalized in the entire pump and vascular system is the mean cardiovascular pressure. The mean cardiovascular pressure can be read more accurately on the venous recording than the arterial because of the difference in the scale.


Low blood volume in a dog, which had moderate hypertension, was produced by draining 600 cc. of blood from the venous cannula. Figure 22 was taken after withdrawal of the 600 cc. This produced a drop in pump output from the control of 2 liters per minute to less than 1 liter per minute, and a drop in the mean cardiovascular pressure from 16 cm. water pressure to approximately 10 cm., and a drop in arterial pressure to 120/50 from an original pressure of 160/90. Figure 23 was obtained after restoring the original blood volume by transfusion of the 600 cc. of blood. The mean cardiovascular pressure went back to the original 16 cm., and the flow rate returned to 2 liters per minute. Note that the arterial blood pressure overshot the original pressure. It then took several minutes to come back to normal. These findings parallel those seen in the intact animal, thus illustrating the relationship of mean cardiovascular pressure, resulting from blood volume and cardiovascular compliance, to cardiac output.

The Effects of Peripheral Vascular Response from
Chemical Stimuli on Cardiac Output

With a large number of stimuli that can affect either cardiac or peripheral-vascular funtion, or both, it is important to know which effector site is responsible for the resulting circulation change. Post hoc ergo propter hoc reasoning ("following this, therefore because of this") has misled thinking in the past. One example of an erroneous conclusion by this reasoning is that variations in heart rate from stimuli are responsible for variations in cardiac output. Heart rate, myocardial contractility, ejection fraction, onset to peak time, end diastolic pressure, arteriolar resistance, and venous pressure are all responses to stimuli from baroreceptors, neurologic stimuli, humoral vascular stimuli, arteriolar vasopressors*, vasorelaxers*, and stretch receptors. The pump-animal preparation allows visualization of cause and effect of many cardiovascular phenomena in a way not possible in the intact animal. Knowing that the mechanical pump cannot react to neurological or chemical stimuli, we can separate changes that are due to a peripheral vascular response from those that are due to a cardiac response. If a stimulus causes the same response as when the heart is in the circuit, we know that the response is from the peripheral vascular effect alone. Then, conversely, any change found with the heart functioning, which is absent with the pumps in the vascular system, can be interpolated as a change that results from a cardiac response. Thus, the animal-pump model allows us to demonstrate whether the effect of a specific stimulus on circulation is due to a cardiac or a peripheral vascular response.


The use of terms "vasoconstrictors" and "vasodilators" are conventionally used for stimuli that would better be referred to as "vasopressors" and "vasorelaxers," as blood is non-compressible. For every increment that vessels constrict in one place, there must be a simultaneous dilation somewhere else in the vascular system. When one segment dilates, there must be a corresponding constriction elsewhere, unless additional fluid enters the vascular space. Unless referring to a localized area of the vascular bed, such as the arterioles, it is more accurate to refer to vasopressors and vasorelaxers.


Figures 24 and 25 document responses to epinephrine, with the heart function replaced with the passive filling pumps. Figure 24 shows an increase in the circulation rate going from 4 1/2 liters per minute to 7 liters per minute after intravenous injection of 1 cc. of 1/10,000 epinephrine. Figure 25 shows the mean vascular pressure, from a normal of 15 cm. water, going to 19 cm. after the epinephrine. Any slowing effect on pump output that might be expected because of increased resistance from spasm of the vascular system is obviously offset by the increased circulation rate caused by the increase in mean cardiovascular pressure. The increase in arterial blood pressure, in this case, is from a combination of increased arteriolar resistance and elevation in pump output from the increased mean cardiovascular pressure.

Figures 26 and 27 show the circulation response to neo-synephrine in an animal preparation, with the heart replaced with a passive filling pump. The pump has no control or regulating devices. The circulatory changes are, therefore, without any cardiac effect and are purely from the effect of the drug on the vascular system. Note that the increased arteriolar resistance, having little upstream compliance, does not decrease the flow rate. The increased mean cardiovascular pressure, which increases the circulatory rate, more than offsets the outflow impediment of the arteriolar resistance. Note also that the pulmonary artery pressure rises as well as the systemic pressure in response to neo-synephrine. The venous pressure stays the same after the injection, showing that the pump is not in failure at the increased flow rate.

Dopamine (Figures 28 and 29) and Levophed (Fig. 30) cause the same circulation response with the pulsatile outflow, uninterrupted inflow, passive filling mechanical pump as when the heart is in the circuit. This finding supports the thesis that vasopressors alter circulation by their peripheral vascular effect, and that the cardiac response to the drugs has its effect in preventing heart failure during variation in circulation rates.


Figures 31 and 32 show responses from injection of nitroprusside intravenously in a dog whose heart function is replaced with the passive filling pumps. Figure 31 shows the reduction in arterial blood pressure and a reduction in circulation rate similar to that seen in the intact animal.

Figure 32 shows the circulation response to nitroprusside with cardiac function replaced by passive filling pumps. Twenty seconds after the nitroprusside injection took effect in lowering the arterial pressure, the pump was turned off. The venous pressure rose and the arterial pressure fell to the mean cardiovascular pressure of 8 to 10 cm. water (normal being 16 to 18 cm. water). The circulation rate change was inconsistent, sometimes falling slightly and sometimes staying about the same. This occurs as the nitroprusside decreases inlet impedance to the extent that it offsets any circulation rate decrease that otherwise would occur from lowering mean cardiovascular pressure alone.

Isoproterinol (Fig. 33) causes the same drop in blood pressure and slight increase in circulation rate as in the intact animal.

Other vasorelaxer drug effects are illustrated by Figures 34, 35, and 36.


The vasopressor and vasorelaxer findings illustrate that the circulation changes in response to these stimuli are the result of their peripheral vascular effect. Any cardiac change in rate or strength of contraction in the intact animal from those stimuli is responsible for the heart staying out of failure concomitant with the peripheral vascular circulatory rate control changes. The pump-animal preparation becomes a useful study method of separating the cardiac from the peripheral vascular effect of drugs on circulation.

The Atrial Effect Experiment

The atrial portion of the pump was eliminated by blocking the atrial impeller in the down position, thereby rendering the atrium a fixed diameter tube, with its cross-section the same as that of the vena cava drainage tubing. In effect, this modification created a passive filling pump with the veins going directly to the ventricles without an intervening atrium. It was then possible to abruptly go back and forth from a pumping circuit, with or without an atrium, while making no other change. At a normal mean cardiovascular pressure and a pump rate of 80, with no atrial effect, and the inertia of stopped venous flow at each pump beat, the pump output was only 1/4 of that seen with the atria functioning. In the absence of atria, when the pump rate was progressively increased, the output was further decreased. At a pump rate of 120, which provided normal flow with the atria functioning, the circulation stopped completely. At this rate, whenever the stopped venous blood would start to flow again toward the ventricles, the inlet valve would close again so no flow would occur. With no atrial function, progressively lowering the flow rate below 80 caused some increase in flow over the 25%, as there was enough time between beats for the venous flow to accelerate toward the ventricles. At a rate of 30 or below, the atria provided no output benefit, as there was plenty of time for the interrupted venous blood to accelerate toward the ventricles. However, at that slow rate the ventricles were in failure (limiting flow by being maximally filled at end diastole). The atrial effect is the major factor in lessening the inlet impedance to passive filling pumps. The atrial effect is most important at a pump rate in the range of normal heart function.

Pulsatile Blood Flow Experiment

In order to observe circulation with and without pulsatile flow, the omentum of a dog was placed on the stage of a dissection microscope. Two parallel tubes, joined at the end with a "Y" connector, were interposed in the omental artery. Blood to the omentum could be diverted through one tube or the other by cross-clamping one of them. One of the tubes contained an elastic section with a slightly constricted outlet, which would damp out the pulse wave and make uninterrupted flow to the omentum when the other tube was clamped. Using this setup, the omental circulation could be observed during either pulsatile or non-pulsatile flow. During pulsatile flow, blood flow remained unchanged and stable through the many vessels. With non-pulsatile flow at the same rate, a marked change in distribution of blood flow progressively occurred. More and more of the flow went to fewer and fewer channels. After 15 minutes, a few vessels had dilated with most of the blood going through them, while other vessels had become empty with no flow. There were small islands of omentum with no blood flow next to others with higher than normal flow. When pulsatile flow was re-initiated, it took two minutes for the original flow distribution to be established. Parenthetically, the easiest way to produce a diffuse hydraulic distribution system is to make it pulsatile like the vascular system.

Go to Chapter 5: Hydraulic Model of the Cardiovascular System ➡    

© 2012 Estate of Robert M. Anderson, MD.  Site use is subject to these Terms & Conditions.