Passive leg raising test predicts fluid needs in ECMO patients

Patients on veno-arterial extracorporeal membrane oxygenation, a heart-lung support machine used increasingly for cardiogenic shock, often receive intravenous fluid boluses to boost the output of their own failing heart. Deciding how much fluid to give is difficult, because excess fluid can be dangerous, yet a simple bedside test to guide the decision has been lacking. A new prospective, interventional study published in Annals of Intensive Care now reports that a passive leg raising maneuver, combined with a routine carbon dioxide monitor, can identify which patients will benefit from fluid expansion, particularly those whose native cardiac output is very low.

The research, conducted at Montpellier University Hospital between February 2020 and April 2022, enrolled 41 adults receiving VA-ECMO support who required volume expansion according to the institution’s protocol. From these patients the team recorded 58 paired tests, each consisting of three sequential steps: a baseline measurement in a semi-recumbent position, a passive leg raising in which the lower limbs were elevated to a 30 to 45 degree angle while the trunk was lowered, and finally the administration of 500 milliliters of normal saline over 15 minutes after the patient returned to the baseline position. Hemodynamic parameters, including blood pressures, heart rate, echocardiographic velocity-time integral at the left ventricular outflow tract, end-tidal carbon dioxide and pump output, were recorded after one minute of stabilization at each step.

Fluid responsiveness was defined as an increase of 15 percent or more in the velocity-time integral after volume expansion. Using this criterion, 38 of the 58 measurements, or 65 percent, were classified as responsive. The passive leg raising test proved to be a faithful miniature version of the real fluid challenge: changes in velocity-time integral induced by the maneuver correlated closely with those induced by the actual infusion, with an intraclass correlation coefficient of 0.83, with a 95 percent confidence interval of 0.73 to 0.90. No adverse events occurred during any of the tests, and all were completed, supporting the safety of the procedure in this fragile population.

The diagnostic performance of the test was strong. Across all 58 measurements, the area under the receiver operating characteristics curve for passive leg raising induced changes in velocity-time integral was 0.89, with a confidence interval of 0.79 to 0.99. At a threshold of 15 percent, sensitivity was 92 percent and specificity was 80 percent, with confidence intervals of 85 to 100 percent and 69 to 90 percent respectively. Notably, this accuracy held regardless of the patient’s baseline native cardiac output, whether it was at or below 1 liter per minute, between 1 and 1.5 liters per minute, or above 1.5 liters per minute.

The most striking finding concerned end-tidal carbon dioxide, the amount of carbon dioxide measured in exhaled breath. Because this measurement requires only a ventilator carbon dioxide analyzer rather than an ultrasound probe, it would be far easier to obtain at the bedside than repeated echocardiographic measurements, which can be difficult and time consuming in patients with very low pulsatility. The study found that passive leg raising induced changes in end-tidal carbon dioxide predicted fluid responsiveness with an area under the curve of 0.86 across all measurements, statistically comparable to the velocity-time integral approach. However, the performance depended strongly on the patient’s baseline native cardiac output.

When native cardiac output was at or below 1 liter per minute, the area under the curve for the carbon dioxide signal rose to 0.98, with a confidence interval of 0.95 to 1. When native output exceeded 1.5 liters per minute, the same signal performed no better than chance, with an area under the curve of 0.50. This pattern makes physiological sense. VA-ECMO diverts blood from the right atrium to the aorta, creating a pump outflow while leaving a residual native cardiac output that flows through the lungs. Gas exchange is therefore partitioned between the lungs and the membrane oxygenator. In the study, pump output and sweep gas flow were held constant during the tests, so any rise in end-tidal carbon dioxide in responsive patients most likely reflected a reduction in alveolar dead space as pulmonary blood flow increased, an effect that is proportionally greater when baseline pulmonary flow is very low.

These results align with the investigators’ earlier observation that end-tidal carbon dioxide correlates strongly with native cardiac output when that output is below 2 liters per minute during VA-ECMO, and with an experimental model in which lung carbon dioxide elimination tracked increases in pulmonary blood flow between 0 and 2 liters per minute as ECMO flow was reduced. The authors note that the reliability of the carbon dioxide signal depends on maintaining a normal arterial carbon dioxide level and a ventilation-to-perfusion ratio of the membrane oxygenator around 1, conditions that were standardized in the protocol. The measurement also requires an intubated, mechanically ventilated patient, since it depends on access to exhaled gas.

The clinical stakes are considerable. In this cohort, pump output accounted for the majority of total systemic output, with a median of 3.2 liters per minute representing 78 percent of the total, and native cardiac output was low in most patients, at or below 1.5 liters per minute in 78 percent of measurements and at or below 1 liter per minute in 55 percent. Volume expansion is given under VA-ECMO either to raise total systemic output when perfusion is inadequate or to increase native output and prevent blood stasis in the cardiac chambers and aortic root, which can lead to thromboembolic complications. But inappropriate fluid loading carries an increased risk of mortality and may cause left ventricular distension and pulmonary edema, a recognized complication of VA-ECMO support. The authors calculate that a fluid strategy guided by a passive leg raising induced velocity-time integral change above 15 percent would have avoided 16 of 20 inadequate volume expansions in this study, at the cost of missing 3 of 38 adequate ones.

The study has limitations that the authors acknowledge. It was conducted at a single center, which may limit generalizability. The same baseline measurement served both the passive leg raising and the volume expansion comparisons, introducing a risk of regression to the mean and of overlapping effects that could artificially inflate the apparent correlation between the two interventions and overestimate diagnostic accuracy. Fluid responsiveness was assessed only at the end of the infusion, so the transient and persistent effects of fluid in responders were not characterized. The carbon dioxide approach cannot be used in patients who are not ventilated, and the definition of an inadequate expansion in cases where the goal was total output rather than native output could be debated in clinical practice.

Even with those caveats, the findings offer a practical tool for a setting where validated dynamic tests are often unusable. Low arterial pulsatility, low tidal volume ventilation and cardiac arrhythmias, all common in VA-ECMO patients, undermine conventional indices of fluid responsiveness, and a previous study of the Trendelenburg maneuver excluded patients with a pulse pressure below 15 millimeters of mercury, limiting its applicability to the low-pulsatility state frequently seen early in ECMO support. The passive leg raising test, by contrast, was feasible and safe in this population, and the carbon dioxide variant requires equipment already present in most intensive care units.

Subject of Research: Predicting fluid responsiveness in VA-ECMO patients using passive leg raising induced changes in end-tidal carbon dioxide and velocity-time integral

Article Title: End-tidal carbon dioxide changes induced by passive leg raising can predict fluid responsiveness in patients on veno-arterial extracorporeal membrane oxygenation: a prospective, interventional study

Article References: Hannequin, T., Ughetto, A., Eliet, J., Blin, C., Canon, A., Saour, M., Gaudard, P., Vidal, C., Molinari, N., Colson, P., & Mourad, M. (2025). End-tidal carbon dioxide changes induced by passive leg raising can predict fluid responsiveness in patients on veno-arterial extracorporeal membrane oxygenation: a prospective, interventional study. Annals of Intensive Care, 15(1), Article 185. https://doi.org/10.1186/s13613-025-01604-2

Image Credits: AI Generated

DOI: 10.1186/s13613-025-01604-2

Keywords: VA-ECMO, cardiogenic shock, passive leg raising, end-tidal carbon dioxide, fluid responsiveness, volume expansion, native cardiac output, velocity-time integral, echocardiography, intensive care, hemodynamic monitoring, fluid overload