Extracorporeal membrane oxygenation (ECMO) is a standard treatment for refractory hypoxaemia (veno-venous ECMO, VV-ECMO) and cardiogenic shock (veno-arterial ECMO, VA-ECMO). Severe hypoxaemia may persist despite ECMO. Prone positioning (PP) can improve outcomes of acute respiratory distress syndrome (ARDS) [1, 2]. However, few data exist on PP in hypoxaemic patients receiving VV-ECMO or VA-ECMO, particularly after cardiothoracic surgery. Here, we evaluated oxygenation and complications seen with PP during ECMO.

We retrospectively studied consecutive patients managed with PP and ECMO between August 2014 and December 2020. PP was used in patients with either refractory hypoxaemia (PaO2/FiO2 < 80 despite 100% FiO2 on ECMO) or persistent hypoxaemia (FiO2 requirement ≥ 80% with ECMO and lung condensations by CT). PP was chosen in patients on VA-ECMO because an additional venous cannula would have decreased arterial flow, potentially causing intolerance and, in the event of posterior basal pulmonary condensation, inducing adverse effects. We recorded ventilation and ECMO parameters, reason for PP, and complications. FiO2 ECMO, FiO2ventilator, and PaO2 were collected before, during, and 6–12 h after PP.

Of 556 patients managed with ECMO, 34 (6.1%) (25 VV-ECMO, 9 VA-ECMO) received PP during ECMO (Table 1). PP significantly improved oxygenation (Fig. 1). Of the 87 PP sessions, six (6.9%) were followed by severe complications requiring emergent treatment. No patient experienced ECMO decannulation. Grade 3 or 4 pressure sores developed on the face or trunk in six (18%) patients. Of the 34 patients, nine (26%) died in the ICU. No patient died after ICU discharge. Of the 522 patients who received ECMO without PP, 237 (45.4%) died in the ICU, and median ECMO duration was 7 days [4–12].

Table 1 Characteristics and outcomes of the 34 patients managed with prone positioning during extracorporeal membrane oxygenation
Full size table
Fig. 1
figure 1

Oxygenation parameters before, during, and after prone positioning (PP) during veno-venous or veno-arterial extra-corporeal membrane oxygenation (ECMO). The grey bars and grey circles represent the PaO2/FiO2 ECMO ratio and ventilator FiO2 values in patients receiving veno-venous ECMO. The open bars and open circles represent the PaO2/FiO2 ECMO ratio and ventilator FiO2 values in patients receiving veno-arterial ECMO. Repeatedly measured quantitative variables were analysed by ANOVA. The PaO2/FiO2 ECMO ratio changed significantly across time points in both the VA-ECMO group (p = 0.007) and the VV-ECMO group (p < 0.001). * VV-ECMO: PaO2/FiO2 ECMO before PP/during PP, p = 0.007. ‡ VV-ECMO: PaO2/FiO2ECMO before PP/after PP, p = 0.001. † VA-ECMO: PaO2/FiO2ECMO before PP/during PP, p = 0.007. VA-ECMO: PaO2/FiO2ECMO before PP/after PP, p = 0.148. ** VV-ECMO: FiO2 ventilator before PP/during PP, p < 0.001. ‡‡ VV-ECMO: FiO2 ventilator before PP/after PP, p < 0.001. †† VA-ECMO: FiO2 ventilator before PP/during PP, p < 0.001. # VA-ECMO: FiO2 ventilator during PP/after PP, p = 0.04

Full size image

In patients receiving VV or VA-ECMO, PP improved oxygenation. Maintenance of the benefits after PP was most obvious in the VV-ECMO group. With VV-ECMO, the benefits of PP can be ascribed to well-documented mechanisms including a ventral-to-dorsal shift of tidal-volume distribution [2] and a decrease in the atelectasis very often seen after protective ventilation. With VA-ECMO, PP may be less likely to improve oxygenation, as gas exchange reflects the combined effect of VA-ECMO and of the native-lung ventilation/perfusion ratio, which is influenced by hypoxic vasoconstriction, shunting, alveolar collapse, and the dead space [3]. Hypoxaemia may worsen due to reduced pulmonary-artery flow during alveolar recruitment. We noted that the flow provided by the ECMO device remained constant during PP. As previously reported, cardiac output can increase, decrease or remain unchanged, depending on preload [4]. Finally, the beneficial effect of PP on the lung parenchyma outweighs the systemic hemodynamic effect even when cardiac output decreases.

In our study, ECMO duration before PP was 7 days, compared to 2 days in another study [2]. One quarter of our patients were successfully weaned off ECMO three days after the last PP session. Thus, PP may break the vicious circle of hypoxaemia, possibly allowing faster weaning off ECMO.

Another important result is the low frequency of complications, in keeping with earlier studies of VV-ECMO for ARDS [5, 6].

The main limitations are the retrospective design and single-centre recruitment of patients who underwent highly specific procedures such as lung transplantation or pulmonary endarterectomy.

Given the low frequency of severe complications, PP in patients under prolonged VA- or VV-ECMO may deserve consideration as a means of improving hypoxaemia and, perhaps, expediting weaning off ECMO.