How I wean patients from veno-venous extra-corporeal membrane oxygenation

Identifying patients who are ready for weaning and liberation from veno-venous extracorporeal membrane oxygenation (ECMO) is challenging in clinical practice. Compared to the several trials addressing the safety and efficacy of ECMO in severe ARDS [1–4], the body of literature regarding ECMO weaning is remarkably scarce. Therefore, this essential component of the management of patients on ECMO is highly variable and often lacks of a systematic approach [5], analogously to the weaning protocols and spontaneous breathing trials used for liberation from mechanical ventilation [6]. 
 
The trajectory from ECMO cannulation to lung recovery and ECMO decannulation consists in the transition from a phase in which ECMO is essential to meet the patient’s metabolic needs (i.e. metabolic oxygen consumption and CO2 production) to a phase in which the native lung function has recovered to satisfy completely the metabolic demands, even if with a degree of ventilatory support considered “safe”. In between these two phases is a continuum of lung healing, during which lung function becomes sufficient to maintain a gas exchange compatible with life, but at the expenses of a high respiratory drive and large swings in transpulmonary pressures. In these conditions, ECMO has the role of maintaining lung protection partially contributing to the patient’s gas exchange [5]. In the effort to track the progress of an individual patient along this imaginary line, it is necessary to measure the relative contribution of the membrane and native lungs in terms of gas exchange, as well as the response of the patient’s respiratory drive and mechanics to the variation in ECMO settings. 
 
We propose a physiology-based assessment protocol, which combines an objective assessment of the native and artificial lung function and quantifies the patient’s response to a standardised weaning trial.


Introduction
Identifying patients who are ready for weaning and liberation from veno-venous extracorporeal membrane oxygenation (ECMO) is challenging in clinical practice. Compared to the several trials addressing the safety and efficacy of ECMO in severe ARDS [1][2][3][4], the body of literature regarding ECMO weaning is remarkably scarce. Therefore, this essential component of the management of patients on ECMO is highly variable and often lacks of a systematic approach [5], analogously to the weaning protocols and spontaneous breathing trials used for liberation from mechanical ventilation [6].
The trajectory from ECMO cannulation to lung recovery and ECMO decannulation consists in the transition from a phase in which ECMO is essential to meet the patient's metabolic needs (i.e. metabolic oxygen consumption and CO 2 production) to a phase in which the native lung function has recovered to satisfy completely the metabolic demands, even if with a degree of ventilatory support considered "safe". In between these two phases is a continuum of lung healing, during which lung function becomes sufficient to maintain a gas exchange compatible with life, but at the expenses of a high respiratory drive and large swings in transpulmonary pressures. In these conditions, ECMO has the role of maintaining lung protection partially contributing to the patient's gas exchange [5]. In the effort to track the progress of an individual patient along this imaginary line, it is necessary to measure the relative contribution of the membrane and native lungs in terms of gas exchange, as well as the response of the patient's respiratory drive and mechanics to the variation in ECMO settings.
We propose a physiology-based assessment protocol, which combines an objective assessment of the native and artificial lung function and quantifies the patient's response to a standardised weaning trial.

Patients' selection
The prerequisites for a weaning assessment is that patients are comfortable, haemodynamically stable and on spontaneous or assisted mechanical ventilation (e.g. CPAP/PS).

Baseline measurements
On a daily assessment of ECMO patients, we routinely measure the following: -Gas exchange variables, including the 100% test; post-membrane PO 2 , VO 2 and VCO 2 of the membrane lung (VCO 2ML ) and VO 2ML ; and VCO 2 of the natural lung (VCO 2NL ) using volumetric capnometry. This allows calculating the total VCO 2 (VCO 2tot = VCO 2ML + VCO 2NL ) and the proportion of VCO 2tot eliminated by the natural lung (VCO 2NL / VCO 2tot ). -Ventilatory variables, including the negative inspiratory force (NIF), the P0.1 (as estimate of the inspiratory drive) and the ratio of P0.1/NIF, and swings in oesophageal pressure (if available).
In our protocol (see Additional file 1 and Fig. 1), we assess separately the patient's dependence on ECMO-in terms of both oxygenation and decarboxylation, or decarboxylation alone-by conducting two sequential steps. The first step consists in the progressive reduction of the ECMO FdO 2 (fraction of oxygen in the sweep gas flow), while the second entails the stepwise reduction of the SGF. The ECBF remains unchanged throughout the procedure.

Step 1: ECMO Deoxy Challenge Test (EDCT)
Before commencing the assessment, the FiO 2 on the ventilator is increased to 60% in order to prevent the transient hypoxaemia that may happen during ECMO weaning. This frequently observed hypoxaemia is partly explained by the abolition of hypoxic vasoconstriction during ECMO support (due to elevated SvO 2 ). The reduction of ECMO VO 2 will reduce the SvO 2 , which will allow the return of the physiologic hypoxic vasoconstriction. This pulmonary vascular response, however, takes time (minutes), and the slow adaptation can lead to a transiently worsened functional shunt (venous admixture). The role of the EDCT is allow time to restore the physiological hypoxic vasoconstriction and optimise ventilation perfusion matching (V/Q). If V/Q is not restored, the functional dead space due to shunt increases and the ability of the natural lung to eliminate CO 2 can be impaired. FdO 2 is decreased from 100 to 60%, 30% and 21% in 5-min steps. A peripheral oxygen saturation (SpO 2 ) > 88% and P0.1 < 10 cmH 2 O will need to be maintained throughout the test. If a patient meets stopping criteria, the test is suspended and the patient is declared "ECMO-dependent".
If EDCT is successful, FdO 2 can be kept at 21% and the test can proceed to step 2 (Fig. 1).
Step 2: ECMO CO2 Challenge Test (ECCT) During this phase of the weaning trial, the SGF is reduced by 30% every 5-10 min, while measuring the patient's response (see Additional file 1 and Fig. 1). Weaning failure and test interruption are indicated by SpO 2 < 88%, RR > 35 bpm and P0.1 > 10 cmH 2 O; VCO 2NL /VE fall by 20% from baseline; and the negative swings of oesophageal pressure are < 15 cmH 2 O, or if any signs of distress/instability are evident (Fig. 1).
If the patient's response remains within set limits at 0 SGF, the weaning test is successful, and the clinical team will consider whether to remain off ECMO or reintroduce a variable degree of extracorporeal support pending decannulation.
We believe that a standardised approach to ECMO assessment and weaning is essential to identify patients who are no longer ECMO dependent. In addition, it provides clinicians with a reproducible protocol to ECMO liberation and researchers with a tool to compare duration on ECMO in clinical trials where ECMO duration is an outcome measure.