Estimation of Cardiac Output Under Venovenous Extracorporeal Membrane Oxygenation: Comparing Thermodilution Methods to 3D Echocardiography

To the Editor: We have read with interest the study by Linden et al.1 comparing thermodilution methods to three-dimensional (3D) echocardiography for cardiac output estimation under venovenous extracorporeal membrane oxygenation (VV-ECMO).1 Their conjecture of indicator loss and the consecutive findings of overestimating cardiac output by thermodilution methods have been confirmed in our previous work.2–4 Our research has demonstrated that standard thermodilution methods are unreliable during ECMO support, whether venovenous or venoarterial, because of indicator loss into the extracorporeal circuit. In our porcine model of VA-ECMO, we observed a negative mean bias and negative limits of agreement between cardiac output measured by an ultrasonic flow probe placed around the pulmonary artery and transcardiac thermodilution,4 with indicator loss occurring both with injections into the right atrium and the right ventricle.4 This indicator loss resulted in an overestimation of pulmonary blood flow. Higher ECMO flows resulted in more pronounced overestimation because of higher indicator loss into the circuit, consistent with Linden et al.’s1 findings. To circumvent these limitations, we have developed and assessed modified thermodilution techniques for both VA-ECMO and VV-ECMO.3,4 In VA-ECMO, our approach uses temperature measurements in the ECMO and pulmonary circuits. This allows for accurate pulmonary blood flow and right ventricular function assessment by directly estimating indicator loss in the extracorporeal circuit.4 For VV-ECMO, we have proposed a similar method to measure recirculation fraction and cardiac output simultaneously using a single cold bolus injection.3 How the indicator is lost differs between VA- and VV-ECMO: In VA-ECMO, the extracorporeal circuit acts as a parallel circulation, whereas in VV-ECMO, recirculation within the venous system is the primary concern.3,5 Despite these differences, adapting traditional thermodilution techniques in both scenarios may account for indicator distribution between native and extracorporeal circulations and thus allow accurate cardiac output estimations. Such modified thermodilution needs special considerations: Although catheter position significantly impacts signal properties such as peak temperature and exponential decay, the area under the temperature–time curve remains constant.2 Changes in the exponential decay as a function of catheter position may impact estimations of right ventricular ejection fraction.2 The work by Linden et al.1 provides valuable clinical validation of the discrepancies between thermodilution and echocardiographic cardiac output measurements in VV-ECMO. However, we believe that modified thermodilution techniques, as demonstrated in our research, offer a promising solution to this problem. These methods could provide continuous, accurate assessment of cardiac output and, in the setting of VV-ECMO, recirculation without the limitations of echocardiography, such as operator dependency and intermittent measurements. Still, our methods lack validation in a clinical scenario. In conclusion, although conventional thermodilution is unreliable during ECMO support, adapted techniques that account for the unique circulatory dynamics of extracorporeal circuits can overcome these limitations. We encourage further clinical validation of these modified approaches to improve hemodynamic monitoring and optimize patient management in VV-ECMO and VA-ECMO settings.