ECMO post iconECMO can fix anything. But, surprisingly, it has some limitations. One of them is differential oxygenation in VA-ECMO, also known as watershed. A recent paper looks into new ECMO set-ups in an experimental animal model that seeks to solve the differential oxygenation problem, and it seems they’re on to something. Now, this is a rather long post, but hopefully it’s worth it.

Watershed and differential oxygenation
Standard VA-ECMOIn VA-ECMO, the ECMO circuit competes in parallel with the native circulation/oxygenation. This gives rise to some challenges. One of them being differential oxygenation, which happens if the lungs are really sick, but the heart delivers significant cardiac output. This means the native circulation delivers poorly oxygenated blood to the aorta, competing with the well saturated blood from the ECMO circuit.

The ECMO arterial return cannula is usually inserted into the femoral artery, and ends somewhere in the descending aorta. So blood flows in the normal direction towards the lower body, but retrograde up the descending aorta to try to deliver oxygenated blood to the upper body. Somewhere in the upper parts of the aorta, the retrograde ECMO blood will meet the blood flow from the patient’s heart. This is the watershed. The more cardiac output the patient generates, the further into the aorta it’ll push the point of watershed. Meaning more of the aortic arch will be filled with badly oxygenated blood. And this is a problem, as we really, really want the brain and the coronary arteries to receive well oxygenated blood. On the other side, the feet and most of the lower body are not that important to oxygenate well.

Fixing the watershed
So ECMO guys hate differential oxygenation. Mostly because the differential balance tips the wrong way. So what to do about it? Some try to attenuate the native heart’s cardiac output with beta-blockers, and this can help. Other, technical, strategies we’ve written about before, are hybrid VAV-ECMO (or VVA-ECMO) configurations. This paper describes similar technical approaches, but some of them are easier to achieve than the full hybrid set-ups. And all of these set-ups shed light on what differential oxygenation is and how the human circulation works.

Some strategies depend on improving SvO2. Now, why would that help oxygenate the upper body on the arterial side? Well, as the patient’s lungs aren’t working, we will have to provide oxygenated blood to the left ventricle (LV) somehow. Now, if the SvO2 is 100%, the hemoglobin oxygen saturation will probably be 100% in the LV too. If the heart puts out 100% saturated blood, we don’t have to care about the watershed. If both the natively circulated blood and the ECMO circuit blood is well saturated, we don’t have a problem anymore. So trying to improve SvO2, or improving blood saturation close to the left ventricle are strategies here. The first one looks at improving SvO2 through simple physiology.

ECMO and the Superior Vena Cava
SVC VA-ECMOSo, in standard VA-ECMO, we drain blood from the IVC and pump it back through the femoral artery into the descending aorta. And the heart receives blood mostly from the SVC. Why is this a problem in differential oxygenation? Well, we create a circulation of well oxygenated blood, and one circulation of poorly oxygenated blood.

Because of the watershed, the lower body is well oxygenated by the ECMO circuit, and the still well oxygenated venous blood return from the lower body into the IVC. The upper body receives poorly oxygenated blood from the heart, and drains this even worse oxygenated venous blood into the SVC.

Now, the well oxygenated blood from the IVC is sucked straight into the ECMO circuit and goes back into the lower body. A good circle, but in the wrong place. On the other hand, the now extremely poorly oxygenated blood returning through the SVC from the upper body, enters the right atrium and goes into the sick lungs where it hardly receives any oxygen, and is pumped from the LV back into the upper body. A vicious circle.

The simple and elegant fix: push the ECMO drainage cannula further in, just past the right atrium and into the SVC! Now, the ECMO circuit will suck in poorly oxygenated blood from the SVC and oxygenate it properly, while the already rather well oxygenated blood from the IVC will pass into the right atrium and get pumped into the upper body by the LV. Instead of two almost separate circulations for the upper body and lower body, you’ll create a “figure of eight” circulation, supplying the whole body with better oxygenated blood.

ECMO and the Carotid Artery
CA VA-ECMOIf you keep the drainage cannula in the IVC, you could rahter try and do something creative with the arterial return cannula.

Here, they placed it in the right carotid artery, thereby supplying oxygenated blood closer to the native circulation start-point. You’ll get oxygenated blood to all of the aortic arch, and that way oxygenate the upper body. However, cannulating the carotid artery might not be a risk free long term. Any dissection or clotting here might be catastrophic for the patient’s brain. Others have used the right axillary artery, and this might be a safer bet? I have also seen these set-ups as one standard cannula into the femoral artery, and a side cannula into the axillary artery.

ECMO and the Internal Jugular
IJ VAV-ECMOThis set-up looks a lot like the hybrid VVA-ECMO set-up we wrote about a while back. Here, the split the return cannula into one standard going into the femoral artery, and one going into the internal jugular vein.

That way, the ECMO circuit delivers blood directly to the lower body, and indirectly to the upper body by pumping well oxygenated blood into the internal jugular, going into the SVC, mixing with fairly well oxygenated blood from the IVC and then getting pumped into the upper body by the LV. So this is a version of the VAV-ECMO set-up.

So are these set-ups just fun and games by mad scientists, or are they doing any good? Well, judging from the oxygen saturation in different parts of the animals, all these set-ups significantly improved upper body oxygenation and attenuated the watershed problem. They were tested in anaesthetised animals, and respiratory failure induced by stopping mechanical ventilation. The graphs show the huge improvement in upper body oxygenation going from standard VA-ECMO set-ups and to the different new set-ups.

Before and after ECMO setups

In these experiments, the SVC set-up and the Carotid Artery set-up provides the best upper body oxygenation, improving oxygen saturation from 35-40% to 65-75%. This is a huge difference, although exaggerated in this experiment, as the animals have normal cardiac function, increasing the differential oxygenation problem. Real patients on VA-ECMO will have some degree of heart failure. However, heart failure patients might improve gradually, and approach normal cardiac output, and through this impose a gradually increasing watershed problem, while still needing circualtory support.

VA-ECMO might give you some challenges, one common challenge is the watershed, or differential oxygenation. ECMO DIY docs and researchers are trying to literally bypass this problem by creating new ways to hook up the ECMO circuits. By far the easiest and most elegant one here seems to be pushing the IVC cannula up into the SVC, however, pushing the cannula further in after initial IVC positioning will pose an infection problem. Also, the venous cannula used might impact the result. But they used multistage cannulas (with multiple side ports) in these experiments, and it still worked well.

In this paper they tried three different ECMO set-ups in animals with induced respiratory failure and severe differential oxygenation, and all of these ECMO set-ups provided better oxygenation to the upper body than standard VA-ECMO. Now, this is an experimental animal model, we’ll still have to see how these strategies might play out in real patients. It will be very interesting to see how ECMO set-ups will evolve over the next years. In that setting this is a very interesting paper to read for anyone into ECMO, as it provides possible solutions to the differential oxygenation problem as well as an insight into the physiology at play.

Go read the free full article here:

Superior vena cava drainage improves upper body oxygenation during veno-arterial extracorporeal membrane oxygenation in sheep, Critical Care, 2015.

Edit November 29th, 2015:
Another very good overview article on ECMO cannulation strategies:
Cannulation strategies for percutaneous extracorporeal membrane oxygenation in adults, Clin Res Cardiol, Nov 2015.

This entry was posted in ECMO. Bookmark the permalink.


  1. Vegard T D. says:

    That is indeed a very interesting article that perfectly underlines the importance of a thorough understanding of physiology in general and ECMO physiology in particular. The point of advancing the IVC cannula to the SVC is tempting, but probably not the best idea for two reasons:
    1. Advancing a previously placed cannula is not a perfectly safe practice from a line infection standpoint.
    2. Assuming that you are using a multistage cannula, simply advancing the tip to the SVC is by far much less potent than adding a IJV cannula and probably removing the femoral one (unless you plan to convert to a VAV setup)

    I really appreciate the blog so keep up the good work guys.

    • Thomas D says:

      Those are indeed good points. Regarding line infections, advancing the cannula already at the insertion time would solve that (but not the potential dangers of doing this maneuver). For multistage cannulas (cannulas with suction holes some way up the distal end, in addition to the distal opening), that would lessen the impact, and most cannulas used are multistage. In this article, advancing the cannula to the SVC makes a huge impact on oxygenation, and they did use multistage cannulas. So it might not impact as much in practice as theory might suggest. Thanks a lot for the input. Will add that to the post to make it clearer.

  2. Viking One says:

    The above mentioned clearly shows when one set up for inappropriate VA ecmo for respiratory failure. In some cases VV ecmo is a much better solution.

Leave a Reply

Your email address will not be published. Required fields are marked *