Short and long-term outcomes of patients with COVID-19-associated acute respiratory distress syndrome and difficult veno-venous-ECMO weaning

To the editor,

The 2019 coronavirus pandemic induced a massive influx of patients with acute respiratory distress syndrome [1], a part of them requiring veno-venous (VV)-extra-corporeal membrane oxygenation (ECMO) support [2]. A consensus of experts has recently published recommendations on VV-ECMO weaning [3, 4], derived from the EOLIA trial [5]. VV-ECMO weaning should be tested when native lung function has sufficiently recovered, allowing for adequate oxygenation and protective mechanical ventilation [e.g., ventilator FiO₂ ≤ 60%, tidal volume ≥ 6 mL/kg of predicted body weight (PBW), respiratory rate ≤ 28/min, plateau pressure (P_plat) ≤ 28 cmH₂O)]. Success criteria of a weaning test (with the membrane ventilation decreased to 0 L/min) for safe decannulation from ECMO are typically as follows: PaO₂ ≥ 60 mmHg and PaCO₂ ≤ 50 mmHg or pH ≥ 7.36 with ventilator FiO₂ ≤ 60% and protective mechanical ventilation. However, some patients may undergo ECMO decannulation without meeting readiness to wean criteria and/or succeeding the weaning test.

The aim of this monocentre retrospective cohort study was to report the outcome of patients who underwent a conventional ECMO weaning (withdrawal after readiness to wean and successful weaning test as per EOLIA criteria) [5] to that of patients who underwent an unconventional facilitative weaning (because of a serious complication of VV-ECMO or lack of respiratory system mechanics improvement despite prolonged support (i.e., ≥ 10 days) in patients who have recovered a satisfactory native lung oxygenation, which justifies withdrawal despite no readiness to wean and/or unsuccessful weaning test). No other treatment was discontinued after ECMO weaning. Fifty-one COVID-19 patients admitted between March 2020 and June 2021 in our French tertiary center who required VV-ECMO support were included in the study. Seventeen patients (33%) died on VV-ECMO, whereas 34 (67%) were weaned off VV-ECMO, including 30 who were discharged alive from our ICU (three patients died and one is still in our ICU). Eighteen patients presented the criteria for facilitative weaning while 16 underwent conventional weaning. VV-ECMO weaning was justified in the facilitative group by one or more of the following: major bleeding (n = 5), infection (n = 2), severe hemolysis (n = 2), no respiratory function improvement despite prolonged duration of VV-ECMO support (n = 12, median [interquartile range 25–75] duration: 24 days [13–43]). Patients of the facilitative weaning group had more complications before VV-ECMO weaning, more often required prone position after VV-ECMO withdrawal, and had longer mechanical ventilation support and ICU length of stay than their counterparts (Table 1). Only two patients with facilitative weaning and one patient with conventional weaning died in the ICU. Strikingly, respiratory system mechanics, gas exchanges and CT-scan were more impaired at the time VV-ECMO was weaned off with facilitative versus conventional strategy (Table 1), consistent with a lung fibrosing process in the former group. Notably, the high plateau and driving pressure levels measured in this group were observed while ventilating patients with low tidal volumes as 75% of these were receiving less than 6 mL/kg PBW. Interestingly, no differences were observed regarding echocardiography, pulmonary function tests and chest CT-scan patterns of lung fibrosis in a subgroup of patients followed-up until 3–6 months of hospital discharge, except for more traction bronchiectasis in patients who underwent facilitative weaning (Table 2).

Table 1 Patients’ characteristics and outcomes in the intensive care unit of patients with conventional or facilitative ECMO weaning

Table 2 Long-term outcomes (three to six months after hospital discharge) of patients with conventional or facilitative weaning

Despite they did not meet the classical weaning criteria [3, 4], patients with facilitative weaning had a low ICU mortality. At long-term follow-up, they also showed good recovery on pulmonary function tests and chest CT imaging. These data illustrate that VV-ECMO withdrawal criteria could be less restrictive, especially in patients developing life-threatening complications under VV-ECMO support or with reasonable recovery of native lung oxygenation function but no improvement of respiratory system mechanics. Our results need to be confirmed and the best ventilator settings to be applied after ECMO weaning to be further studied.