Access and return cannula devices for ECMO

ECMO is half of "Mechanical supports: ECMO/IABP", an L1 topic of Section 2.1.4 in the second edition of the CICM Syllabus for the Second Part Examination. And access is half of ECMO, in the sense that the membrane oxygenation would not be extracorporeal without them.  A significant proportion of the problems (and their solutions) in ECMO originate from the access and return devices. What follows is an attempt to discuss these in a way that does not digress into palaeontology or forensics.

In summary:

ECMO cannula design:

• Wide (25-29 Fr) access, multi-stage (many sideholes) to achieve high rates of venous drainage from several places in the venous circulation
• Narrow (17-19Fr) return cannulae to minimise arterial damage
• Steel spiral wire keeps them kink-free (and MRI incompatible)
• Biologically inert material (PVC)
• Able to conduct flows of 6-8 L/min with a pressure drop of no more than ~100 mmHg (for 25 and 29Fr drainage cannulas)

In terms of the best single paper to read for some peer-reviewed material, Strunina et al (2019) or Kohler et al (2013) are probably the most detailed, and the Alfred guidelines are the the most practical.

 Broad principles of ECMO cannula design

A tubular device like this has some basic requirements:

• A sufficiently flexible body that it conforms to body shape and does not damage vessels
• A sufficiently rigid body that it remains kink-resistant, and does not collapse under suction
• Large enough to conduct huge volumes of blood with low resistance
• Small enough not to completely destroy the vessel during insertion
• Bevelled and smooth to permit easier percutaneous cannulation
• Biologically inert, so as not to anger the immune system or invite thrombosis
• Designed in a way that defends the tip against occlusion

These requirements were probably already largely met by the fenestrated polyethylene tubes used by the early pioneers of ECMO, and resemble the expectations one may fairly have of vas caths and other large lines. ECMO cannulae do not differ from these very much, except in terms of size and survival importance (making some of the requirements slightly more important than others). This is notable in papers like Kohler et al (2013),who offer a thorough list of peformance characteristics for ECMO cannulae followed by a discussion of how contemporary devices met them, and who refer mostly to dialysis catheter data in their technical discussion.

Like other lines in the ICU, these are inserted using Seldinger technique, using serial dilation to make the tissue passage larger. The kit would not be completely unfamiliar to even a very junior ICU staff member, as all of the components resemble CVC and arterial line insertion kits in everything but size (eg. the guidewire is 150cm long). Here is a representative kit from Getinge:

These catheters are made from one-piece molded silicone polyurethane polymer, often with additives that have Tradmarked Names™ and are therefore chemically unknowable but which are purported to decrease platelet aggregation and protein deposition (Strunina et al, 2019). They are usually reinforced with a stainless steel spiral wire, making them kink-resistant (or at least it seems like  non-reinforced cannulae are less and less available). The main disadvantage of this design is the fact that these cannulae are not MRI-compatible; and if by some chance they do become kinked, then no force can subsequently unkink them. 

The multi-stage access cannula for ECMO

"Multi-stage" here really just means "many holes". This design feature helps distribute the negative suction pressure of the blood pump among many openings, preventing the cannula from becoming occluded by vessel walls. The term comes from an archaic description of drainage of venous blood during bypass, such as where it is described by Riley et al (1986) or Arom et al (1981). Here, "dual-stage" cannulae were so named because they had two main drainage regions, one from the RA and the other from the inferior vena cava, as seen here from the photograph by the authors:

This is bizarre, as the use of large fenestrated cannulae dates back to even the fifties and sixties, but in those days the writers referred to them as "fenestrated" (eg. Helmsworth et al, in 1952, wrote of some mysterious "long fenestrated koroseal® tubes" ), and there was no talk of "stages". The "stage" terminology arose later (perhaps the early 1980s) and appears to be assoicated with the marketing of  "dual stage" or "triple stage" cannulae as upgrades for single-stage lines, but nowere is there any explanation for why the word "stage" was used instead of anything else.

Anyway: to misappropriate this excellent image from Broman et al:

The total length of such a device is usually 55-60cm, which corresponds to a set of normal human height and weight variables (as there is usually about 45 cm from the puncture site to the right atrium). The tip may be somewhere close to the RA, where the flow is greatest, but clever modelling suggests that the holes that aspirate the maximum blood flow are actually the proximal ones (i.e. those about 10cm from the tip of the cannula, which aspirate blood from the vicinity of the hepatic vein and renal veins when inserted femorally).

The obsession with cannula tip design is understandable when one considers that rate of blood drainage from the venous circulation is the main determinant of blood flow in the whole circuit. 

The return cannula for ECMO

No less important, the return cannula is responsible for politely returning the oxygenated blood without knocking over all the furniture. These are often said to be significantly smaller than the access cannulae for this exact reason, i.e the arteries are much smaller than veins, which means the maximum cannula diameter is somewhat limited; and the tip tends to be designed to "fan-disperse" the arterial blood so that it does not direct a high-pressure jet at the fragile fatty atheroma of the aortic arch.

Does this "jet" translate into an increased risk of stroke? According to Kohler et al (2013), yes; but the paper they cite was  Muehrcke et al (1995), which is mostly a description of a unqiue cardiopulmonary bypass cannula design not used anywhere else. So it not clear whether all ECMO return cannulae in use worldwide now follow this pattern, but it looks like an inexpensive modification, and one which is sufficiently desirable that one might use it as a specific marketing instrument, eg. in this Getinge promotional literature:

The size difference and the tip design has implications. Where an IJ cannula was inserted to return oxygenated blood and where it is subsequently needed to act as a drainage line to transform the circuit into V-VA, the flow rate will be much less than from a "proper" large multistage line.

Selection of ECMO access cannula sizes

Size selection is largely determined by the expected flow rate. For adults expecting to have all of their haemodynamic needs met by an ECMO circuit, that equates to something like 60-80ml/kg/min, or 4-6L/min for a normal size adult. This is difficult to model because blood has mystical pseudoplastic non-Newtonian properties and defies crude simplifications, which means that when each manufacturer produces the (often water or glycerin-derived) data for their product, it is often not representative of clinical experience.

In graphopictorial form, and specifically for Getinge cannulas: 

Assuming we flog the pump and accept cava-shredding access pressures of -100 (the point at which the machine starts to alarm), the maximum flow rate through a 23Fr access cannula ends up coming out as something like 7L/min. That means this cannula, at the ragged edge of its performance, can supply enough oxygenated blood for the survival of a 87kg person.