A novel ‘shunt fraction’ method to derive native cardiac output during liberation from central VA ECMO

Abstract

The Fick principle is an established method to quantify intracardiac shunts. The Fick principle has also found utility in the practice of extracorporeal membrane oxygenation (ECMO). This report describes a novel ‘shunt fraction’ method to calculate intrinsic cardiac output in central (right atrial‐to‐aorta) ECMO. The physiological basis of this ‘shunt fraction’ method is described, followed by the case presentation that details the clinical application of this method of quantifying intrinsic cardiac output to guide weaning and liberation from central VA ECMO.

Introduction

The assessment of cardiac function/output (CO) is central to the process of weaning/liberation from veno‐arterial extracorporeal membrane oxygenator (VA ECMO). Clinicians have employed surrogates such as ejection fraction or velocity time integral of left ventricular outflow tract Doppler due to the challenges of assessing CO on VA ECMO support. ¹ This case report describes a novel ‘shunt fraction’ method to calculate native CO to guide weaning/liberation from central (right atrium‐ascending aorta) VA ECMO. The theoretical background for this method is described.

Theoretical background – The ‘shunt fraction’ method

The Fick principle is used to calculate the size or flow across the shunt between the pulmonary (Qp) and systemic (Qs) circulations using the differences in oxygen content at two points of Qp and Qs. In practice, oxygen saturation (O2sat) is used instead of oxygen content, as the other determinants of oxygen content are assumed to be constant across both circulations. This is the basis for the shunt fraction equation familiar to clinicians (Equation 1):

$$\frac{\mathrm{Qp}}{\mathrm{Qs}}=\frac{\left(\mathrm{O}2\text{satA}-\mathrm{O}2\text{satMV}\right)}{\left(\mathrm{O}2\text{satPV}-\mathrm{O}2\text{satPA}\right)}=k$$ (1)

where Qp = pulmonary blood flow; Qs = systemic blood flow; k is the ratio of Qp/Qs; O2satA = systemic arterial blood O2sat; O2satMV = mixed venous blood O2sat; O2satPV = pulmonary venous blood O2sat; O2satPA = pulmonary arterial blood O2sat.

In the absence of an intracardiac left‐to‐right shunt, O2satMV=O2satPA and Equation (1) can be rewritten as Equation (2):

$$\frac{\mathrm{Qp}}{\mathrm{Qs}}=\frac{\left(\mathrm{O}2\text{satA}-\mathrm{O}2\text{satMV}\right)}{\left(\mathrm{O}2\text{satPV}-\mathrm{O}2\text{satMV}\right)}=k$$ (2)

In right‐to‐left shunting, Qs is the sum of Qp (or CO) and shunt flow. The mathematical inference from Equation (2) is that Qp can be calculated if O2sats at the different sites and shunt flow are known. In central VA ECMO, we can produce an iatrogenic right‐to‐left (right atrial‐to‐aorta) shunt by turning off the sweep gas/oxygen flow (Figure 1); and as this shunt flow, that is, VA ECMO flow (Qe) is known, Qp, which is the native CO under steady state conditions, can then be determined.

Figure 1.

(A) Chest radiograph of right atrial and aorta cannulation in central VA ECMO. With sweep gas and oxygen turned off, this right atrial‐aorta VA ECMO flow, Qe becomes a right‐to‐left shunt. (B) The calculation of Qp/Qs from measured O2satA, O2satMV, O2satPA and O2satPV (Equation 1). O2satMV can be assumed to be the same as O2satPA. Initial weaning of central VA ECMO flow to 800 mL/min and turning off sweep/oxygen created a right‐to‐left shunt with resultant O2satA‐O2satMV of 28% and O2sat PV‐O2satPA of 34% and Qp/Qs of ∼0.8 (white dot). Titration of inotropes at the same VA ECMO flow of 800 mL/min reduced O2satA‐O2satMV to 25% and O2sat PV‐O2satPA to 30%, increasing Qp/Qs to 0.84 (grey dot). Further weaning of central VA ECMO flow to 400 mL/min resulted in O2satA‐O2satMV of 30% and O2sat PV‐O2satPA of 32% and Qp/Qs of ∼0.9 (black dot). (C) Qp or native CO can be calculated with known Qe and the calculated Qp/Qs (Equation 3). At the initial Qp/Qs of ∼0.8 at Qe of 800 mL/min, the calculated Qp or native CO was 3.2 L/min (white dot). With up‐titration of inotropes at the same VA ECMO flow, Qp/Qs increased to 0.84 and CO to ∼4.1 L/min (grey dot). Further weaning of central VA ECMO flow to 400 mL/min resulted in Qp/Qs of ∼0.9 and calculated native CO of 4.0 L/min (black dot).

Thus, native CO can be determined in two steps:

Firstly, Qp/Qs can be calculated from Equation (2) by measuring O2satA, O2satMV and O2satPV. O2satPV can be measured by sampling blood gas from ‘wedged’ pulmonary artery catheter (which reflects pulmonary venous or left atrial oxygen saturation) ² (Figure 1).

Secondly, as $\frac{\mathrm{Qp}}{\mathrm{Qs}}=k$ and Qs = (Qp + Qe).

$$\mathrm{Qp}=k\mathrm{Qp}+k\mathrm{Qe}$$

By rearranging the equation, Qp can be calculated as a function of VA ECMO flow (Qe) and the ratio of Qp/Qs (Equation 3) (Figure 1B):

$$\mathrm{Qp}=\frac{k\mathrm{Qe}}{\left(1-k\right)}$$ (3)

Of note, O2satPV can be taken either from a ‘wedged’ pulmonary artery catheter or assumed to be 100%. A ‘wedged’ pulmonary artery catheter, by occluding the vessel creates a continuous column of blood from the end of the catheter to the left atrium, essentially a manometer for the measurement of the pulmonary venous and left atrial pressure. By aspirating from a ‘wedged’ pulmonary artery catheter and discarding the initial sample of pulmonary arterial blood, highly oxygenated pulmonary venous blood that approximates left atrial and systemic arterial blood oxygen saturation can be sampled. This sampling of ‘wedged’ blood has been used to confirm appropriate ‘wedging’ of the catheter. ³

The assumption of 100% O2satPV may be valid with fractional inspired oxygen increased to 100% in the absence of significant intrapulmonary shunt or lung injury. However, this assumption should be made with caution in critically ill patients on mechanical ventilation. Native CO or Qp would be underestimated if O2satPV is erroneously assumed to be 100% (i.e. if true O2satPV is <100%), because Qp/Qs is lowest with O2satPV at 100% (Equations 1 to 3). In other words, in the presence of a right‐to‐left shunt, the lower the O2satPV, the higher the native CO (i.e. assuming O2satPV of 100% can only underestimate, and not overestimate Qp). The practical implication is that, pulmonary venous blood gas sampling may not be necessary if calculated Qp is satisfactory at the assumed O2satPV of 100%.

Case Report

A 44‐year‐old female with double outlet left ventricle underwent orthotopic heart transplantation, from a 49‐year‐old donor from circulatory death. She required high doses of inotropes/vasopressors after separation from cardiopulmonary bypass and was established on central VA ECMO support for severe biventricular graft dysfunction. After 6 days of support, with echocardiographic recovery of left ventricular function, we proceeded with weaning/liberation from central VA ECMO.

In preparation, a pulmonary artery catheter is inserted prior to weaning. In theatre, heparin was administered to maintain activated clotting time (ACT) of >200 s, and ventilator fractional inspired oxygen increased to 100%. The weaning of central VA ECMO flow was then performed in two stages.

Firstly, VA ECMO flow was reduced to ∼20% of her predicted ‘full flow’ (predicted ‘full flow’ is VA ECMO flow that provides cardiac index of 2.4 L/min/m²; for example, for a patient with body surface area of 2.0m², the predicted full flow would be 4.8 L/min/m²). In this case, VA ECMO flow was reduced to ∼800 mL/min, based on her predicted ‘full flow’ of 4.0 L/min. Inotropes/vasopressors were initially titrated to achieve mean arterial‐central venous pressure difference of 55 mmHg, with concurrent pump speed adjustments to maintain constant flow of ∼800 mL/min. A period of at least 10 min was allowed to achieve steady state.

Sweep gas/oxygen on VA ECMO was turned off when hemodynamic steady state is achieved to create a right‐to‐left shunt of 800 mL/min or a 20% shunt if her native cardiac index was 2.4 L/min/m². Over about 10 min, her O2satA dropped from 100% to a plateau of 93%, and O2satPA was 65%. O2satPV from ‘wedged’ pulmonary artery catheter was 99%. From Equation (2), Qp/Qs was calculated as ∼0.8 (Figure 1B, white dot).

From Equation (3), at Qe = 800 mL/min, Qp ≡ native CO was calculated as 3.2 L/min (or cardiac index of 1.93 L/min/m²) (Figure 1C, white dot). Escalation of inotropes/vasopressors (dopamine 7.5 mcg/mL/min, milrinone 0.33 mcg/kg/min and norepinephrine 0.075 mcg/kg/min) increased O2satA and O2satMV to 94% and 69%, respectively (no change in O2satPV), increasing Qp/Qs to 0.84 and CO to ∼4.1 L/min (∼2.38 L/min/m²) (Figure 1B,C, grey dots).

At this new steady state, we proceeded with the second stage of weaning. A further bolus of heparin was administered with resultant ACT of 256 s. VA ECMO flow was further reduced to 10% of ‘full flow’ to ∼400 mL/min. At VA ECMO flow of ∼400 mL/min (i.e. right‐to‐left shunt of 400 mL/min), O2satA and O2satMV were 97% and 67%. O2satPV from ‘wedged’ pulmonary artery catheter was unchanged at 99%. Calculated Qp/Qs and CO were ∼0.91 and ∼4.0 L/min (cardiac index ∼2.28 L/min/m²), respectively (Figure 1B,C, black dot). At this cardiac index of 2.28 L/min/m², on the same doses of vasoactive drugs, mean arterial and central venous pressures were 75 and 14 mmHg, respectively. This steady state was maintained for at least 10 min.

Thus, calculated oxygen delivery was ∼329 mL/min/m² (haemoglobin 106 g/L), indicative of adequate, albeit marginal oxygen delivery by our criteria. We proceeded with central VA ECMO liberation. Decannulation was uncomplicated. Serial thermodilution measurements post‐decannulation were consistent with ‘shunt fraction’‐derived cardiac index at 2.38 L/min/m².

Discussion

Clinicians have employed a range of parameters to estimate native CO to guide weaning/liberation from VA ECMO. This report showed that the ‘shunt fraction’ method, with its theoretical basis rooted in the Fick principle ⁴ can be used to calculate native CO by inducing a right‐to‐left shunt on central VA ECMO. In turn, cardiac reserve or response to escalation of inotropes and reduction in VA ECMO support can be determined.

Practical considerations for the application of this ‘shunt fraction’ method include: (i) this method specifically relates to central VA ECMO, and should not be applied to peripheral (e.g. femoral) VA ECMO (ii) monitoring of anticoagulation ACT is vital during periods of low VA ECMO flow and we aim for ACT of at least 200 s; (iii) we used this ‘shunt fraction’‐derived cardiac index to calculate oxygen delivery, in keeping with our broader practice in cardiogenic shock ⁵ ; but individual centres may incorporate this ‘shunt fraction’ method‐derived CO to complement their own institutional protocol for VA ECMO weaning/liberation.

In conclusion, this report describes a novel ‘shunt fraction’ method to derive native CO to guide weaning/liberation from central VA ECMO. Future studies should evaluate this method in larger series of patients.

Conflict of interest

None declared.

Lim, H. S. (2024) A novel ‘shunt fraction’ method to derive native cardiac output during liberation from central VA ECMO. ESC Heart Failure, 11: 570–573. 10.1002/ehf2.14441.