Cardiac output curves and vascular function curves

This chapter is relevant to Section G3(i)  of the 2017 CICM Primary Syllabus, which asks the exam candidate to "describe and explain cardiac output curves, vascular function curves and their correlation".  This element of cardiovascular physiology which is a very attractive candidate for vivas and short answer questions, and it is surprising that it has only ever appeared once, as Question 13 from the second paper of 2011. Though the question asked us to "discuss the regulation of cardiac output", what the examiners clearly meant was "explain how the Guyton model describes the relationship of cardiac output and venous return".

The cardiac function and vascular function curves (and their "correlation", which is where you basically just superimpose them on top of each other) is an attempt to answer a series of questions.  What factors are responsible for the matching of cardiac output and venous return? Are the same factors acting on both, or are they independent? And what role does right atrial pressure (CVP to the rest of us) play in determining cardiac output? 

In summary:

Because this whole thing is Arthur Guyton's fault, it would be reasonable to reference his seminal 1955 paper here, but that is not the same as a recommendation to read it. Most normal people should start with watching the lecture series by Jon-Emile Kenny which, over 111.4 minutes, will bring about a feeling of peaceful cosmic oneness with this subject matter. In contrast, reading the reference list at the end of this chapter for 111.4 minutes will have a very different effect. A much better resource for the trainee who still insists on text would probably be this section on doctorlib.info, from what appears to be a free online version of Linda Costanzo's Physiology. Venous return and vascular function curves are handled best by the Venous Return chapter from Young's Control of Cardiac Output (2010).

The cardiac output curve

Otherwise known as the cardiac function curve, this thing resembles the Frank-Starling relationship, except right atrial pressure is used as the x coordinate instead of the end-diastolic volume, and the y-axis is cardiac output instead of stroke volume. Though those parameters are hardly interchangeable, it is possible to continue the discussion without breaking off into a digression about it because they are closely related. Right atrial pressure is an important determinant of right ventricular end-diastolic volume, and is therefore a good-enough surrogate for it. Moreover, it allows the venous return to be plotted on the same graph as the cardiac output, as now both are being measured with the same units.

When reproducing this diagram, the trainee needs to keep in mind that every textbook everywhere has made a completely arbitrary choice with regards to the labelling and scale. It seems one can put any combination of numbers on either axis. Some authors don't even bother. For some semblance of authenticity, the diagram above has been modelled on the original Guyton diagram from 1955, but - judging by the values and what we know about the author - these data may well have been collected from a particularly large dog. 

It is important to suppress these suspicions in order to function at the exam. For a quick sketch, the essential elements of such a diagram are:

• A plateau: cardiac output generally stops increasing beyond a certain right atrial pressure value, which for some reason seems to be 4 mmHg in the vast majority of publications. This is probably the effect of the experiment: most authors who report on this are working with a denervated canine heart, which means the Frank-Starling mechanism is the only autoregulatory process being observed. With an intact sympathetic nervous system, a larger mammal, and an adrenaline infusion, this plateau will be much higher.
• A linear portion: most of these graphs tend to plot a linear relationship across the lower range of right atrial pressures. This portion of the curve represents the range of preload values which the ventricle finds acceptable, i.e. the range of pressures and preloads for which the length-tension relationship is optimised. The slope of this linear portion represents different states of contractility.
• The x-intercept.  This curve is usually drawn extended to cross the x-axis, i.e. where a cardiac output of zero L/min is expected. In most (not all!) textbooks and in the original Guyton paper, this surreal point is achieved at a right atrial pressure of -2. This can't be a real measurement from human data, given that most investigators would politely stop the experiment before the cardiac arrest occurs. Guyton and colleagues collected this information by emptying the blood volume of experimental animals.

The vascular function curve

This is a graphical description of what happens to venous return, in L/min, when right atrial pressure changes.  As you can see, this is an inverse relationship. Like all godfearing fluids, blood flows along a gradient from an area of higher pressure (veins and venules) to an area of lower pressure (the atrium). Thus, as right atrial pressure decreases, the venous return flow increases, until it has reached a plateau. Vice versa, in this model the right atrial pressure forms a "back-pressure" to  the A stylised version of this relationship is offered here:

There is usually a sharp elbow to these diagrams, but as the original graphs below demonstrate, the in-vivo relationship is much smoother than that.  Again, all of this comes from experiments by Guyton, in 1950s. Instead of a pulsatile heart, the cardiac output was provided by a pump which sucked blood out of the right atrium and pushed it directly into the aorta. Mean systemic filling pressure was varied by adding fluid, and venous vascular resistance was stable over the course of the experiments.

The following essential elements belong in this graph:

• The mean systemic filling pressure,  the mean pressure in the systemic circuit of the circulation measured in the absence of flow. This bizarre concept is sufficiently fascinating to merit a whole page of its own. In summary, it is determined by venous vascular resistance and total body fluid volume, and represents the pressure in the peripheral circulation which - while it is higher than right atrial pressure - creates the pressure gradient for venous return. 
• The x-intercept: the venous return, being dependent on the pressure gradient, should theoretically cease if there is no pressure gradient (i.e. when the right atrial pressure is equal to mean systemic filling pressure). This implies that the mean systemic filling pressure represents some sort of limit on what the right atrial pressure can be (i.e. without venous return there is no cardiac output). Guyton and colleagues produced this result in hundreds- perhaps thousands - of dogs. Even though it was the 1950s, the investigators stopped short of human experiments, remarking that "it is doubtful that it will ever be measured in its entirety in the human being" because of the "drastic procedures required to determine the normal venous return curve".
• The plateau: when right atrial pressure is 0 mmHg or negative, this does not apply any sort of "suction" to the central veins. The major veins begin to collapse instead; this was observed directly by Guyton and colleagues. Thus, decreasing the right atrial pressure to below atmospheric pressure (i.e. 0 mmHg) does nothing to increase cardiac output; all it does is increase the venous resistance (as it is harder for blood to flow through semicollapsed veins)-  hence the plateau seen in the experiments. Here is a demonstration of a series of such plateaus from experiments by Guyton and colleagues (1957):
  
  Now, one might raise the criticism that these experiments were recorded in dogs who has their chests open and the situation may be different wherever the pleura and thoracic cavity are intact. But Guyton et al also performed the same experiments with closed-chest animals, and found that the inflexion point was basically the same, give or take 1mm Hg.
  The reader will notice that these recordings demonstrate a relative smooth transition to a flat plateau; this is because all the veins did not collapse at once- the investigators mentioned that some narrowed while others remained open as the RA pressure approached 0 mmHg. Only if all the veins narrow simultaneously (and to the same degree) will you get the sharp cliff seen in medical textbooks.

The intersection of the cardiac output and vascular function curves

Now that we have both the cardiac and the vascular curves in hand, and knowing that they are plotted on the same axis, the next natural step is to combine them into one diagram, which is occasionally referred to as the Guyton diagram for some reason. 

Yes, you are right, the only additional element is the steady state operating point. At this right atrial pressure, the cardiac output and the venous return are equal. This is in fact a defining characteristic of the cardiovascular system, as - apart from momentary variations - the cardiac output and the venous return must be equal. 

Now, let us see what happens when the variables are altered.

Effect of volume on the cardiac output/venous return relationship

Guyton and colleagues' dog preparation allowed them to vary the mean systemic filling pressure by adding or subtracting fluid from the circulatory system while all the other variables remained fixed. In this way, they were able to record a series of venous return curves which ultimately made their way into all the textbooks:

Thus, the linear relationship between RA pressure and venous return remains unchanged (which makes sense, because that is a pressure gradient) but the maximum venous return value increases.

Now, let us plot this on the same coordinates as the cardiac output. Observe what happens to the steady state operating point:

So, with the increase in volume, the RA-venous return relationship shifts right, and the steady state point migrates up along the cardiac output curve. This is exactly what would happen if the venous compliance were to change. The increase in venous smooth muscle tone would increase the size of the "stressed" volume (see the preload chapter again) and shift the curve to the right, i.e. the effect would be identical to giving a fluid bolus. Again, this is something Guyton et al (1954) measured directly, using an infusion of adrenaline.  

Effect of contractility on the cardiac output/venous return relationship

If all the volume variables remain the same, but the contractility increases, the venous return and cardiac output both increase:

The mean systemic filling pressure remains the same, However, as the cardiac function curve has shifted to the right, so has the steady state point, which means that the increase in cardiac output is associated with a decrease in right atrial pressure.

Effect of peripheral vascular resistance on venous return

If the total peripheral vascular resistance (i.e. arteriolar smooth muscle tone) were to increase,  it would represent an obstruction to the flow of blood out of the arterial and into the venous circulation.  Venous blood volume would therefore decrease (more of the blood is now stuck in the arterial circulation). The MSFP would remain more or less the same, because its major determinants are largely unaffected: the total blood volume has not changed, and the vascular resistance in the arterioles only affects about 3% of the total blood volume. So the steady state point would shift down, as below:

This model, therefore, suggests that as arterial resistance increases, there is an increased amount of blood being sequestered on the arterial side of the circulation, i.e. it cannot make its way into the venous side, and venous return therefore decreases. Cardiac output drops, and the right atrial pressure ...decreases? To borrow a turn of phrase from Beard & Feigl (2011) "At this point the knowledgeable reader has probably become impatient. This is not how the cardiovascular system behaves". Which brings us to:

What the hell is the point of all this

If one had the education of future intensive care specialists as one's primary goal, one would have to say that the main objective of studying these relationships is to generate a simplified model of the cardiovascular system, which: 

• Decribes the relationship and interdependence of cardiac function and the peripheral circulation,
• Predicts changes in cardiac output resulting from changes in the peripheral circulatory variables (eg. volume or peripheral vascular resistance)
• Predicts changes to peripheral circulatory variables (eg. RA pressure) on the basis of changes to cardiac function (eg. contractility)

And you could then use this model to describe preload and contractility, the effects of giving a fluid bolus, or the effects of giving vasopressors and inotropes - in short, it becomes a convenient instrument of education.  

What is almost certainly not the point of this, is to indoctrinate the trainees with a peculiar Guytonian model of the circulatory system, which maintains that the venous return plays a central role in the regulation of the cardiac output, that the heart plays a "permissive" role, i.e. it pumps everything it is presented with, and where the right atrial pressure is a "back-pressure" which restricts venous return and decreases preload. Henderson et al (2010) are probably the best at describing the main premises of this model in an impartial manner.  Beard & Feigl (2011) do an excellent job of explaining its limitations, and Jon-Emile Kenny does an excellent job of defending it. If we had to reduce this whole argument to point form, it would look like this:

Guytons model poses that:

• Independently of heart rate and contractility, preload changes the cardiac output by the Frank-Starling relationship, and appears central to the regulation of cardiac output.
• Preload is determined largely by venous return, and venous return (Guyton argued) was mainly determined by the mean systemic filling pressure and right atrial pressure.
• Cardiac output, it is argued, plays little role in venous return. The bucket or bathtub analogy is used here: arterial pressure fills the tub like a tap, and it makes sure there's constant delivery of blood into the system, but it does not determine pressure or the rate of flow through the drain hole (that being the venous return). 
• Ergo, the heart is a passive participant here, and the mechanisms which determine preload are the main regulators of cardiac output
• That means the difference between MSFP and RA pressure drives the venous return, not the arterial pressure generated by the heart.
• The benefit of this model is its power to predict cardiac output according to any combination of functional variables (eg. MSFP, cardiac contractility, RA pressure).
• The variables which make up this model and their interaction has been confirmed in humans and in animals experimentally, supporting the model.

However:

• The elastic recoil pressure which generates the MSFP is supplied by the LV (that's what is supplying the flow of blood to refill and stretch those capacitance vessels)
• The model as a predictor of cardiovascular behaviour works very well when one variable is being altered at a time, but all of the variables interact, all the time, which makes it difficult to apply in real-life situations 
• The model is silent as to what happens on the flat volume-overloaded part of the cardiac function curve, where both right atrial pressure and mean systemic filling pressure can continue increasing without much of a change in cardiac output. 
• There are concerns that Guyton's models were flawed by his adjustment of the blood pump to produce the desired level of RA pressure, which refutes his central point (as the pump, a surrogate for the heart, was then the main determinant of RA pressure)
• Authors (eg. Berlin et al, 2018) argue that the model uses the analogy of direct current, whereas the circulation behaves more like an alternating current circuit.
• Most like the MSFP only plays a role while the heart is in arrest; during normal function the gradient promoting venous return is probably not between the RA and the MSFP, but between the RA and the mean arterial pressure (Levy, 1979)
• The model also fails to incorporate such contributing factors as arterial reflected pressure waves, inertia of the blood, and the arterial Windkessel effect.

This is of course a grossly oversimplified representation of a debate which still continues in eldritch academic dimensions where ancient entities lash each other with unimaginable energies.  Those of us at the bedside are thankfully insulated from this unpleasantness, and it is possible to carry on with a fruitful career in critical care medicine with only the faintest appreciation of these issues, safely using fluid and vasoactive substances. Therefore the main take home message, after almost three thousand words of this, appears to be that separating the circulatory system into these discrete components makes it easier to explain to novices, at the cost of some accuracy and, some might argue, your soul.