An Introduction to Extracorporeal Membrane Oxygenation (ECMO)

Introduction

Extracorporeal Membrane Oxygenation (ECMO) is a form of partial cardiopulmonary bypass used for long-term support of respiratory and/or cardiac function. This technology arose from cardiopulmonary bypass used for cardiac surgery. Initial systems used bubble oxygenators which were poorly suited for prolonged use because of their tendency to hemolyze blood. Membrane oxygenators made long-term use of ECMO possible. The first report of successful ECMO support of an adult was published by Hill in 1972.

Rationale

ECMO is primarily indicated for patients with such severe ventilation and/or oxygenation problems that they are unlikely to survive conventional mechanical ventilation. Examples of such patients would include those with the adult respiratory distress syndrome (ARDS) without major non-pulmonary organ failure who are failing mechanical ventilation or who are suffering from major barotrauma that makes adequate ventilation impossible. ECMO is only useful in cases where the primary lung insult is reversible in the absence of the usual oxygen toxicity and barotrauma caused by usual mechanical support. Since most patients who die of ARDS die of multi-system organ failure or sepsis, it would follow that most ARDS patients are not good candidates for ECMO. Another criteria often cited for qualifying a patient for ECMO treatment is a pre-treatment predicted mortality threshold, typically above 80-90%. Such a threshold makes ECMO, a new and as yet non-mainstream therapy, an ethically appealing option for otherwise hopeless cases. Unfortunately, the use of such a criteria preselects patients that may already be too sick to benefit from ECMO. If the pre-treatment mortality predictions are over-estimated, as they frequently are, then ECMO, as a treatment, may get more credit than it deserves in saving these patients’ lives.

Techniques

ECMO currently comes in two varieties: venoarterial (VA), and venovenous (VV). VA ECMO takes deoxygenated blood from a central vein or the right atrium, pumps it past the oxygenator, and then returns the oxygenated blood, under pressure, to the arterial side of the circulation (typically to the aorta). This form of ECMO partially supports the cardiac output as the flow through the ECMO circuit is in addition to the normal cardiac output. VV ECMO takes blood from a large vein and returns oxygenated blood back to a large vein. VV ECMO does not support the circulation. VA ECMO helps support the cardiac output and delivers higher levels of oxygenation support than does VV ECMO. VA ECMO carries a higher risk of systemic emboli than does VV. VV ECMO systems may actually recirculate previously oxygenated blood depending on the placement of the inflow and outflow catheters. Another variant of VV ECMO is extracorporeal CO2 removal (ECCO2R). With this mode of support, oxygenation is provided by slow ventilation of the native lungs while CO2 removal is accomplished by the ECMO circuit. In all forms of ECMO, CO2 removal is more efficient than O2 addition because of the solubility and diffusion properties of CO2 relative to O2. In fact, CO2 normally has to be added to ECMO circuits to offset this efficiency at CO2 removal. The flow through the ECMO circuit is typically on the order of 100 mL/kg/minute. This would be from 25-75% of the cardiac output. This high flow requires the placement of large catheters into the circulation. For adults, the intravascular catheters may be 20 Fr or larger while 14 Fr catheters might be used in infants. These are generally placed by cut-down. Current circuits require the use of systemic anticoagulation with heparin to keep the systems patent. This anticoagulation is, in large part, responsible for bleeding complications that can be seen with the use of ECMO. Bleeding may occur either at the site of catheter insertion or at remote sites such as intracranial or the gastrointestinal tract. Future catheters with inpregnated heparin may obviate the need for systemic anticoagulation and would be expected to reduce bleeding complications. Once blood leaves the patient, it comes in contact with a gas-permeable membrane that allows gas exchange to occur between the blood and the gasses (oxygen and carbon dioxide) that are run into the oxygenator. Carbon dioxide levels leaving the oxygenator can be adjusted about as low as the physician wants while oxygen levels typically reach the 400-500 mm Hg level. Note that CO2 content varies fairly directly with partial pressure while O2 content follows the shape of the oxyhemoglobin dissociation curve. Under normobaric conditions, this limits the amount of oxygen that can be loaded into blood. If you were to try to predict the final mixed arterial oxygen content or partial pressure during VA ECMO, you would need to know the pulmonary venous oxygen content, the ECMO effluent oxygen content, and the % of cardiac output flowing through the ECMO circuit:

CaO2 = (EO x EF + NO x NF) / (EF + NF)

Where      EO: ECMO oxygen content

                EF: ECMO flow

                NO: non-ECMO oxygen content

                NF: non-ECMO flow

Predicting final arterial content during VV ECMO is more complex and cannot be easily calculated. During VV ECMO, the lungs may actually excrete oxygen rather than remove it from the inspired gas. Once ECMO has been started, the ventilator is typically reduced to minimal settings to prevent atelectasis and barotrauma. The FIO2 is lowered as much as possible, perhaps to room air. The ECMO flow rate is adjusted to the minimum needed to provide for adequate gas exchange and/or circulation support in the case of VA ECMO. ECMO is typically used for 7-12 days and is weaned when there is evidence of pulmonary improvement as indicated by compliance or radiologic appearance. ECMO can also be stopped if there are major complications or if the patient’s condition is determined to be hopeless.

Complications

ECMO complications are those associated with cannulation (pneumothorax, vascular disruptions, bleeding, infection, emboli), those associated with systemic anticoagulation (GI bleeding, intracranial bleeding etc), and exsanguination resulting from circuit disruptions. These potential complications require that a trained ECMO technician be present at the bedside 24 hours per day in addition to the patient’s usual nursing presence. As many as 30% of ECMO treated infants reported to the Extracorporeal Life Support Organization (ELSO) registry suffer some degree of CNS injury.

Patient Selection

As mentioned previously, usual ECMO criteria include patients with a severe reversible process that would result in a very high predicted mortality with conventional ventilatory support. The best candidates are those without multi-system organ failure that are otherwise considered to be salvageable. It would be expected, that once ECMO becomes safer and more accepted by physicians, that these criteria would be revised to eliminate the requirement for a very high predicted pre-treatment mortality. There are several ways to evaluate patients to determine if they are ‘sick enough’ for ECMO. APACHE and other scoring systems could be used to predict mortality. More frequently, indexes to assess refractory hypoxemia are used. The a:A ratio or A-a gradient can be used, but the oxygenation index (OI) is currently in more common use: OI = (FIO2) (MAP) / PaO2 An OI greater than 0.40-0.55 is thought to predict an 80% predicted mortality. Unfortunately, mortality predictors have been notoriously pessimistic since they are derived by looking at retrospective data and tend to do a poor job of predicting prospective results.

Evaluation of Results

ECMO is a difficult therapy to study and compare to conventional means of support. Ideally, a therapy like ECMO should be studied with a prospective randomized clinical trial where patients that meet an inclusion criteria are randomized to receive either ECMO or conventional therapy. Unfortunately, in such a study, the typical end-point would be the death of the patient and this is an end-point that most investigators and institutional review boards are uncomfortable with. As a result, studies comparing ECMO to conventional therapy typically include an escape clause of sorts that allows patients who are determined to have failed conventional therapy can get a trial of ‘rescue’ ECMO. The other main problem with studies of ECMO for adult patients are that there are very few patients who are sick enough to need ECMO that have reversible disease that ECMO might allow time for reversal of the process. Less than 20% of ARDS deaths are caused by respiratory failure while higher numbers of pediatric and neonatal deaths are caused by primary respiratory failure which would explain why ECMO would be more successful in pediatrics than it has proven to be in adult medicine.

In 1979 an NIH sponsored study compared ECMO with conventional support of patients with acute respiratory failure. Survival with ECMO was 9.5% while survival with conventional therapy was 8.3%. In this study, however, many of the included patients had end stage irreversible lung disease. Until recently, the above study all but stopped the use of ECMO in adults.

Today, there are about 85 ECMO centers worldwide and about 4,000 neonates have been treated (as of September 1990). All of these patients had predicted pre-ECMO mortality estimates greater than 80% (remember the problems with predicted mortality figures) and 83% survived with ECMO treatment. Infants with meconium aspiration syndrome had a 93% survival rate while patients with congenital diaphragmatic hernia had the lowest survival rate at 62%. In August 1986, Luciano Gattinoni reported the results of a study of 43 patients with an expected mortality of greater than 90%. He used ECCO2R (VV ECMO with low frequency ventilation) and had 21 survivors (48.8%). Lung function improved during ECMO in 72.8% of cases. His study was not controlled and again has the problem of the questionable validity of the 90% predicted mortality figure.

Conclusions

From the neonatal experience, it would appear that the key to making successful use of ECMO in adults is proper patient selection. The following indications and criteria are probably appropriate:

1. Primary reversible respiratory failure

VV ECMO: The patients should be failing conventional ventilatory support in that they have poor oxygenation and/or ventilation in spite of high ventilator settings that carry a high risk of lung damage from barotrauma and/or oxygen toxicity. These patients are rare but they do exist. ARDS with severe barotrauma might be an example of this indication if there is no multi-system organ failure.

2. Reversible cardiogenic shock

VA ECMO: This is a potential indication but reversible cardiogenic shock is a rare condition hence the fall in popularity of intra-aortic balloon pumps perhaps with the exception of being a bridge to transplant.

3. No multi-system organ failure

Patients with MSOF have a very high mortality rate even when ventilation and oxygenation are more than adequate. ECMO would not be expected to do more than prolong death.

4. No contraindications for anticoagulation

This may be obviated when heparin inpregnated ECMO circuits are available.