“Cardiac output cannot be accurately measured during veno-venous ECMO using traditional pulmonary artery catheter thermodilution because recirculation in the ECMO circuit tends to overestimate true cardiac output.”

Veno-venous extracorporeal membrane oxygenation (ECMO) has become a widely accepted therapy for adults with severe acute respiratory failure. According to the extracorporeal life support organization registry, veno-venous ECMO has been used to treat greater than 51,000 patients worldwide since 1990, and nearly 17,000 patients were treated worldwide during the COVID-19 pandemic. Although veno-venous ECMO allows for adequate arterial oxygenation to be achieved in most patients, in some cases, hypoxemia persists despite institution of ECMO.

The effectiveness of veno-venous ECMO for improving blood oxygenation is contingent upon several factors including the relative matching of ECMO circuit blood flow with the patient’s cardiac output. The amount of recirculation within the ECMO circuit also affects blood oxygenation. In a study by Schmidt et al. 10 adults with acute respiratory distress syndrome and constant mechanical ventilation settings had cardiac output serially measured using echocardiography during veno-venous ECMO. ECMO circuit blood flow was decreased progressively to as low as 40% of its maximum.  The authors also adjusted sweep gas fractional inspired oxygen tension and flow, while ECMO circuit blood flow was maintained at a constant level.  Important findings from the study include the following: (1) when ECMO circuit blood flow was reduced by 60%, Po2 fell by almost 50%; (2) when ECMO circuit blood flow was reduced by 60%, arterial oxygen content fell by 20%; and (3) all patients who had an ECMO circuit blood flow to cardiac output ratio exceeding 0.6 had an arterial oxygen saturation greater than 90%.  These findings highlight the important relationship between ECMO circuit blood flow and cardiac output in veno-venous ECMO patients and also why accurate measurement of cardiac output could be clinically useful.

Recirculation occurs when oxygenated blood that is returned from the ECMO circuit is drained back into the venous drainage cannula rather than entering the patient’s circulation. Recirculation reduces the amount of blood that is oxygenated and effectively delivered to the patient. Although frank recirculation is straightforward to recognize because blood in the venous return cannula becomes bright red (like arterial blood) and its oxygen saturation is high (e.g., greater than 80%), lesser degrees of recirculation are often more subtle and may not be clinically apparent. In a cohort study of 19 veno-venous ECMO patients who had recirculation serially measured using ultrasound-based flow dilution, median recirculation fraction was 14 to 16%, with 10% of measurements demonstrating recirculation of more than 35%.  In the same study, higher ECMO circuit blood flow was associated with more recirculation in a linear fashion, while higher cardiac output was associated with less recirculation.  Interestingly, higher ECMO circuit blood flow also increased arterial oxygenation saturation despite increasing recirculation. The explanation for this is that oxygen delivery from the ECMO circuit increased by approximately 130 ml O2 × min−1 for each 1 l/min increase in ECMO circuit blood flow. 

In this issue of Anesthesiology, Berger et al. describe an innovative series of experiments in which they utilized a mock ECMO circuit with an adjustable recirculation loop and mock patient circulation (with adjustable shunt) to simultaneously measure cardiac output and recirculation in an ECMO circuit using thermodilution. To accomplish this, the authors used traditional temperature-based thermodilution with three thermistors on pulmonary artery catheters placed into the ECMO circuit. The thermistors were located in the following locations: (1) ECMO circuit inflow limb, (2) ECMO circuit outflow limb, and (3) mock patient circulation. Room temperature injectate was administered into the ECMO circuit just after the oxygenator and 65 cm proximal to the outflow limb thermistor. The recirculation fraction in the ECMO circuit was calculated using the area under the temperature change curve (AUC) measured at the inflow limb and outflow limb with the following formula:

Recirculation fraction (%) = AUCECMO In/ AUCECMO Out

In other words, the authors anticipated that injectate volumes would split according to the ratio of their respective blood flows (recirculation vs. patient circulation), and AUCs would reflect relative blood flows in each limb. Cardiac output was calculated by determining how much injectate volume was recirculated versus how much entered the mock patient circulation and then applying standard thermodilution cardiac output calculations.

When the authors compared calculated recirculation against that measured from an ultrasonic flow probe, they found a high level of agreement with a systematic bias (mean difference) of −0.9%, which became more negatively biased (greater underestimation of flow) as set recirculation fraction increased or set cardiac output decreased. Similarly, low bias was observed for cardiac output measurements that were obtained using the authors’ methodology. Notably, when the set recirculation fraction exceeded 40%, bias increased significantly, and precision became low, suggesting that at very high recirculation fractions, accuracy would be poor.

Although the modified ECMO circuit used in the study probably cannot be employed immediately in most ECMO centers, the authors’ demonstrated its feasibility, measurement accuracy, and some potential limitations of using thermodilution. Other groups have also described thermodilution as an accurate method to calculate recirculation in veno-venous ECMO patients with similarly low bias.  Recirculation can be measured using other techniques such as ultrasound dilution, but this method also has limitations, and the most commonly used bedside method remains trending oxygen saturation in the ECMO circuit return blood in conjunction with arterial oxygen saturation.  When arterial oxygen saturation is low and venous return blood saturation is higher than anticipated, there is almost certainly significant recirculation present in the ECMO circuit.

Cardiac output cannot be accurately measured during veno-venous ECMO using traditional pulmonary artery catheter thermodilution because recirculation in the ECMO circuit tends to overestimate true cardiac output.  One study suggested that cardiac output measurement with traditional pulmonary artery catheter thermodilution overestimated true cardiac output by as much as 300% during veno-venous ECMO.  Hence, there are only a few available methods to measure cardiac output during veno-venous ECMO, including echocardiogram-derived cardiac output measurement and use of noninvasive cardiac output monitors that have not been validated in veno-venous ECMO patients.

Considering the impact that both matching ECMO circuit blood flow with cardiac output and recirculation have on arterial oxygenation during veno-venous ECMO, should thermodilution-derived measurement of these parameters be incorporated at the beside? Perhaps, but measurement would have to lead to clinical interventions that would improve ECMO gas exchange, facilitate protective mechanical ventilation, or improve cardiac output and organ perfusion. The fact that traditional cardiac output monitoring with a pulmonary artery catheter has not been shown to improve outcomes in patients with shock is one reason for skepticism.  Veno-venous ECMO patients represent a different patient population, however, and in cases in which oxygenation remains poor, use of a modified ECMO circuit that can measure recirculation and cardiac output frequently may allow clinicians to determine whether more ECMO circuit blood flow is needed, whether a different cannulation scheme or modified positioning is needed, and/or whether inotropic support might be beneficial. If these interventions help to optimize mechanical ventilation, reduce ventilator-induced lung injury, or improve organ perfusion, they could affect patient recovery during ECMO.

Another important consideration is the cost associated with adding additional technology to a veno-venous ECMO circuit. The current diagnosis-related group payments for ECMO remain high, but adding cost without demonstrating value will not be acceptable, as there is increasing emphasis on value-based healthcare in the United States. The cost of ECMO circuits has rapidly increased over the past decade, as new components have been introduced. If cannulae or tubing are engineered to perform thermodilution or to have implanted oxygen sensors, these features will have to substantially improve clinical outcome or at least maintain a similar cost profile. Companies that manufacture ECMO circuits continue to evolve their products so that they provide more data for the bedside clinician regarding flow and various pressures within the circuit. However, it remains unclear whether any of these additional data have added value or improved clinical outcomes. Additional “bells and whistles” that become a routine part of ECMO circuits should be clinically impactful, especially if they add significant cost.

In summary, accurate quantification of recirculation fraction and cardiac output remain significant bedside challenges for veno-venous ECMO patients with contemporary ECMO circuits. The modified ECMO circuit described by Berger et al. offers an approach for accurately measuring “typical” recirculation, as well as cardiac output, which could be clinically impactful in some patients. Nevertheless, critical care medicine has frequently demonstrated that more patient data do not automatically translate into improved patient outcomes. Clinical studies will be needed to confirm whether adjustment of ECMO parameters based on recirculation fraction or cardiac output measurement can improve meaningful patient outcomes before widespread adoption occurs.