Liberation From Mechanical Ventilation Before Decannulation From Venovenous Extracorporeal Life Support in Severe COVID-19 Acute Respiratory Distress Syndrome

Abstract

The coronavirus disease 2019 (COVID-19) pandemic has been associated with the significant use of venovenous extracorporeal membrane oxygenation (VVECMO) globally. Identifying strategies to optimize care is essential to improving patient important outcomes. By liberation from mechanical ventilation (MV) before VVECMO to provide awake-ECMO, complications related to MV could be minimized, leading to improved outcomes. Between March 2020 and October 2021, we conducted a prospective observational study at the Kuwait Extracorporeal Life Support Program, of patients admitted for COVID-19 acute respiratory distress syndrome (ARDS), with recording baseline characteristics, respiratory support, and ECMO parameters. Of the 207 patients who underwent VVECMO for COVID-19 ARDS during this period, only 5 patients were successfully liberated from MV before decannulation to provide awake-ECMO. Four were female with a median age of 38. Before VVECMO, all patients received corticosteroids and lung-protective ventilation with four receiving prone positioning. The median duration of MV use was 4 days, whereas the median duration of VVECMO use was 12 days, with early mobility, and all survived until hospital discharge. The safety and feasibility of liberation from MV before ECMO decannulation to provide awake-ECMO were demonstrated, but further studies are warranted to identify factors associated with this success.

Coronavirus disease of 2019 (COVID-19) resulted in significant morbidity and mortality worldwide.¹ As a viral pneumonia, the most feared complication is acute respiratory distress syndrome (ARDS) and critical illness.² Although therapeutics and vaccination were being developed and with the failure of patients to improve with conventional support, the different waves of the pandemic saw a significant adoption of extracorporeal life support (ECLS) as a rescue therapy for respiratory or cardiac support, with more than 10,000 confirmed COVID-19 cases reported to the Extracorporeal Life Support Organization registry to date (https://www.elso.org/COVID19.aspx).

Currently, the use of venovenous extracorporeal membrane oxygenation (VVECMO) is advised when lung-protective mechanical ventilation (MV) and prone ventilation fail to improve severe hypoxemia and respiratory acidosis; therefore, VVECMO is used while the lungs are allowed to heal and recover.³ Nonetheless, MV, deep sedation, and immobility to facilitate prone ventilation are not risk-free strategies, they might be at the expense of lung and diaphragm protection.⁴ Ideally, with VVECMO, patients should be awake, interactive, and participating in physical rehabilitation to speed up lung recovery,⁵ but significant respiratory drive, risk of patient self-inflicted lung injury,^6,7 and infection control demands of patients with COVID-19 might not allow this to happen.

Therefore, in this case series, we sought to report our experience with COVID-19 severe ARDS patients who have been successfully liberated from MV before decannulation for VVECMO to provide awake-ECMO and describe their respiratory support, ECLS management, and radiological findings.

Methods

We conducted a prospective analysis of all severe COVID-19 ARDS patients who required rescue therapy with VVECMO, managed by the Kuwait Extracorporeal Life Support Program and admitted to two COVID-19 centers at Al-Amiri Hospital and Jaber Al-Ahmed Hospital, Ministry of Health (MOH), Kuwait, from March 2020 until October 2021. Approval from MOH ethics committee was granted (1405/2020). Patients who were successfully liberated from MV before VVECMO decannulation were identified. Patients’ demographics, clinical characteristics, respiratory support including MV settings (before VVECMO, 24 hours after VVECMO, and just before extubation), VVECMO characteristics, and clinical management of patients were obtained from patients’ charts or electronic medical records. Data were expressed as proportions, mean (standard deviations [SD]), or median (interquartile range [IQR]), as appropriate. Furthermore, Patients’ timeline charts were created to outline the respiratory support course during their ICU admission.

Results

Baseline Characteristics and Management Before VVECMO

Between March 2020 and October 2021, 207 patients with critical COVID-19 pneumonia were supported with VVECMO in our program, of which 5 patients (2.4%) were successfully extubated before decannulation from VVECMO (Table 1). Four (80%) were female with a median age of 38 (IQR 16.5) years. Furthermore, the median body mass index was 29.4 (IQR 18.1) kg/m², with the vast majority being relatively healthy (Key median of 0 [IQR 0.5]). All patients were positive for SARS-CoV-2 by real-time reverse transcriptase-polymerase chain reaction and had symptoms before admission to ICU at a median of 7 (IQR 3.5) days. Upon admission to ICU, the median Acute Physiology and Chronic Health Evaluation II (APACHE II) score was 24 (IQR 9.5), while the median Sequential Organ Failure Assessment (SOFA) score was 12 (IQR 2) and the median Richmond Agitation and Sedation Scale (RASS) of 0 (IQR 3). All patients received corticosteroid (dexamethasone 6 mg for 10 days) and two patients (40%) were given monoclonal antibody against interleukin-6 receptor (Tocilizumab). All patients were mechanically ventilated when they developed severe hypoxemia denoting progressive severe ARDS and all had extensive bilateral pulmonary infiltrates (Figure 1). As rescue therapies before VVECMO, four patients (80%) received prone ventilation, two patients (40%) were given inhaled nitric oxide, and all patients received neuromuscular blockers. The median duration of MV before VVECMO was 18 hours (IQR 18).

Table 1. - Baseline Characteristics and Management Before VVECMO for Five Consecutive Patients with COVID-19 ARDS

Patient Demographics	1	2	3	4	5	Total/Median (IQR)
Age (years)	38	51	49	34	33	38 (16.5)
Gender	F	F	M	F	F	-
BMI (kg/m2)	57.2	26	29.4	24.7	29.7	29.4 (18)
Charlson comorbidity index	0	1	0	0	0	0 (0.5)
SARS-CoV-2 RT-PCR	Positive	Positive	Positive	Positive	Positive	—
Days of symptoms before ICU	7	3	7	7	10	7 (3.5)
APACHE II on admission	24	34	21	30	24	24 (9.5)
SOFA on admission	12	12	14	12	10	12 (2)
RASS on admission	1	0	-4	0	1	0 (3)
COVID-19–specific therapies
Dexamethasone	Yes	Yes	Yes	Yes	Yes	5/5
Tocilizumab	Yes	Yes	No	No	No	2/5
Rescue therapies before ECMO
Prone	No	Yes	Yes	Yes	Yes	4/5q
Inhaled nitric oxide	No	Yes	Yes	Ni	No	2/5
Neuromuscular blockade	Yes	Yes	Yes	Yes	Yes	5/5
Hours of MV before VVECMO	66	6	6	18	24	18 (18)

Results in absolute values or median (interquartile range).

APACHE II, acute physiology and chronic health evaluation II; ARDS, acute respiratory distress syndrome; BMI, body mass index; COVID-19, coronavirus disease 2019; ICU, intensive care unit; IQR, interquartile range; MV, mechanical ventilation; RASS, Richmond sedation and sedation scale; SARS-CoV-2 RT-PCR, severe acute respiratory syndrome coronavirus 2 real-time reverse transcriptase-polymerase chain reaction; SOFA, sequential organ failure assessment; VVECMO, venovenous extracorporeal membrane oxygenation.

Figure 1.: Radiological imaging on admission to intensive care unit, after mechanical ventilation and before venovenous extracorporeal membrane oxygenation (VVECMO), before extubation while on VVECMO and before hospital discharge for five consecutive patients with coronavirus disease 2019 acute respiratory distress syndrome.

MV, Illness Acuity, and Sedation Prior and During VVECMO

Before VVECMO, while all patients were on controlled MV, the following MV parameters were noted (mean [SD]), tidal volume 4.4 [1.5] ml/kg per predicted body weight (PBW), plateau pressure (Pplat) of 37.9 [6.1] cmH₂O, positive end expiratory pressure (PEEP) of 13.4 [2.3] cmH₂O, fraction of inspired oxygen (FiO₂) of 0.92 [0.18], respiratory rate (RR) of 32 [4] breaths/min, and oxygen saturation (SaO₂) of 65 [35.5]%. Two patients required handbag ventilation to sustain oxygenation before VVECMO cannulation. Furthermore, they had a driving pressure (ΔP) of 23.5 [7.5] cmH₂O, minute ventilation of 9.8 [3.2] L/min, ventilatory ratio (VR) of 4.3 [2.6], and airway pressure during the first 100 ms (P0.1) of 2.5 [1.0] cmH₂O with arterial blood gases reflecting refractory respiratory acidosis and hypoxemia (pH 7.2 [0.1], paCO₂ 82.9 [24.9] mmHg, paO₂ 66.8 [52.4] mmHg, and a PaO₂/FiO₂ ratio of 69.4 [49.7] mm Hg/%). Their SOFA score was 8.2 [1.3], heart rate (HR) of 129 [14] beats per minute, and they were all deeply sedated with multiple agents and paralyzed with RASS score of –4 [0.0] (Table 2).

Table 2. - Mechanical Ventilation, Illness Acuity, and Sedation Before, 24 Hours After and Before Extubation for Patients on VVECMO for Five Consecutive Patients with COVID-19 ARDS

Settings	Before VVECMO	24 Hours After	Before Extubation
Mode	PCV	PSV	PSV
Vt, mean (SD), ml/kg per PBW	4.4 (1.5)	5.7 (4.2)	7.1 (4.3)
Plateau pressure, mean (SD), cmH2O	37.9 (6.1)	18.8 (7.6)	17.6 (8.3)
PEEP, mean (SD), cmH2O	13.4 (2.3)	9.6(0.9)	8.2 (2.0)
FiO2, mean (SD), %	92 (18)	38 (4)	41 (2)
RR, mean (SD), breaths/min	32 (4)	12 (3)	20 (13)
SaO2, mean (SD), %	65 (35.5)	98 (2.1)	95 (6.4)
PaO2/FiO2 ratio, mean (SD)	69.4 (49.7)	194.6 (47.9)	197 (36.2)
Driving pressure, mean (SD), cmH2O	23.5 (7.5)	9.2 (7.3)	9.4 (6.6)
Minute Ventilation, mean (SD), L/min	9.8 (3.2)	4.9(4.8)	7.6 (3.2)
Ventilatory ratio, mean (SD),	4.3 (2.6)	1.3 (1.2)	2.2 (1.2)
P 0.1, mean (SD), cmH2O	2.5 (1.0)	1.2 (0.5)	6.9 (1.3)
Arterial pH, mean (SD)	7.20 (0.1)	7.37 (0.0)	7.39 (0.0)
PaCO2, mean (SD), mmHg	82.9 (24.9)	54.7(13.3)	55.3 (12.8)
PaO2, mean (SD), mmHg	66.8(52.4)	74.2(22.2)	92.1 (21.5)
SOFA score, mean (SD)	8.2 (1.3)	3.6(1.95)	2.2 (0.84)
Heart rate, mean (SD), beats/min	129 (14)	104 (22)	91 (18)
RASS score, mean (SD)	–4.0 (0.0)	–3.8 (0.5)	–0.4 (1.1)
Sedation and marcotics
Propofol	5/5	4/5	1/5
Dexmedetomidine	3/5	2/5	1/5
Remifentanil/fentanyl	5/5	5/5	4/5

Results in mean (SD).

ARDS, acute respiratory distress syndrome; cmH₂O, centimeter of water; COVID-19, coronavirus disease 2019; FiO₂, fractional inspired oxygen ratio; mmHg, millimeter of mercury; P0.1, airway pressure during the first 100 ms; PaCO₂, partial pressure of carbon dioxide; PaO₂/FiO₂, arterial oxygen partial pressure to fractional inspired oxygen ratio; PaO2, partial pressure of oxygen; PBW, predicted body weight; PCV, pressure-controlled ventilation; PEEP, positive end expiratory pressure; PSV, pressure support ventilation; RASS, Richmond sedation and sedation scale; RR, respiratory rate; SaO₂, oxygen saturation; SOFA, Sequential Organ Failure Assessment; Vt, tidal volume; VVECMO, venovenous extracorporeal membrane oxygenation.

Twenty-four hours after VVECMO, significant lung-protective strategy was achieved (tidal volume 5.7 [4.2] ml/kg per PBW, Pplat of 18.8 [7.6] cmH₂O, PEEP of 9.6 [0.9] cmH₂O, FiO₂ of 0.38 [0.04], RR of 12 [3] breaths/min, and SaO₂ of 98 [2.1]%, ΔP 9.2 [7.3] cmH₂O, minute ventilation of 4.9 [4.8] L/min, and VR of 1.3 [1.2], P0.1 of 1.2 [0.5] cmH₂O) with significant improvement in oxygenation and gas exchange (pH 7.37 [0.0], paCO₂ 54.7 [13.3] mmHg, paO₂ 74.7 [22.2] mmHg and a PaO₂/FiO₂ ratio of 194.6 [47.9] mm Hg/%) while SOFA score improved (3.6 [1.95]), HR reduced to 104 [22] beats per minute but not RASS score (–3.8 [0.5]) with some reduction of sedation infusions. Before extubation the situation was similar, and the patients were awake and cooperative (tidal volume 7.1 [4.3] ml/kg per PBW, ΔP 9.4 [6.6] cmH₂O, VR 2.2 [1.2], P0.1 of 6.9 [1.3] cmH₂O, SOFA score 2.2 [0.84], HR 91 [18] beats per minute, RASS score –0.4 [1.1]). Furthermore, gradual improvement was noted in radiological imaging (Figure 1) albeit not completely cleared. Moreover, following extubated, patients were supported with either high-flow oxygen therapy or low-flow oxygen therapy for a variable amount of time (Figure 2).

Figure 2.: Timelines of respiratory support and venovenous extracorporeal membrane oxygenation (VVECMO) for five consecutive patients with coronavirus disease 2019 acute respiratory distress syndrome.

ECMO Configurations, Settings upon Initiation, Complications, and Short-Term Outcomes

All five patients, following obtaining informed consent from the patient substitute decision maker, underwent two-site cannulation draining from the inferior vena cava via a multistaged cannula in the femoral vein (all right femoral vein size 25 Fr cannula), and reinfusion to internal jugular vein and superior vena cava (all right internal jugular vein 21 Fr–23 Fr cannula). Unfractionated Heparin bolus was given during cannulation then infusion was initiated following initiation of VVECMO, as per institutional protocol. Maximally during ECMO support, the median blood flow was 4 (IQR 1.85) L/min, and median gas sweep was 4 (IQR 3.5) L/min, while fraction of delivered oxygen remained constant at 100% (Table 3).

Table 3. - Configuration and ECMO Maximal Settings During Support, Complications, and Short-Term Outcomes of VVECMO for Five Consecutive Patients with COVID-19 ARDS

	1	2	3	4	5	Total/median (IQR)
ECMO Configuration	25RFV–21RIJ	25RFV–21RIJ	25RFV–23RIJ	25RFV–21RIJ	25RFV–22RIJ	—
RPM	6600	2405	2570	2568	3200	—
Blood flow (L/min)	6	3.5	4	3.6	4.8	4 (1.85)
Sweep gas (L/min)	2	4	4	7	6	4 (3.5)
FDO2 (%)	100	100	100	100	100	100 (0)
Use of heparin	Yes	Yes	Yes	Yes	Yes	5/5
Hemorrhage	No	Yes *	No	Yes †	No	2/5
Thrombosis	Yes ‡	Yes §	No	No	No	2/5
Circuit change	No	No	No	No	No	0/5
Infection  Bacteremia  Pneumonia	No Yes	No Yes	No No	No No	Yes Yes	1/5 3/5
Pneumothorax	No	No	No	No	No	0/5
Reintubation	No	No	No	No	No	0/5
Tracheostomy	No	No	No	No	No	0/5
Dialysis	No	No	No	No	No	0/5
Duration of MV	4	1	3	7	13	4 (8)
Duration of ECMO	3	12	5	23	21	12 (18)
ICU LOS	8	25	9	28	38	25 (24.5)
Hospital LOS	14	56	19	51	52	51 (37.5)

Results in absolute values or median (interquartile range).

ARDS, acute respiratory distress syndrome; COVID-19, coronavirus disease 2019; FDO₂, fraction of delivered oxygen; ICU, intensive care unit; IQR, interquartile range; L/min, liters per minute; LIJ, left internal jugular vein; LOS, length of stay; MV, mechanical ventilation; RFV, right femoral vein; RIJ, right internal jugular vein; RPM, revolutions per minute; VVECMO, venovenous extracorporeal membrane oxygenation.

^*Left femoral arterial-line site bleeding requiring surgical repair.

^†Retroperitoneal hematoma.

^‡Postdecannulation right internal jugular deep vein thrombosis.

^§Postdecannulation right internal jugular and femoral vein deep vein thrombosis.

During the entire ECMO duration, two patients had significant hemorrhagic events (femoral arterial site bleeding requiring surgical repair and retroperitoneal hemorrhage, both unrelated to cannulation site), two patients developed postdecannulation deep vein thrombosis, and three patients had nosocomial infection (two patients with pneumonia and one with pneumonia and bacteremia). None of the patients required circuit change, developed pneumothorax, required reintubation or tracheostomy, or required dialysis during their ECMO course.

With regard to outcomes, the median duration of MV was 4 (IQR 8) days, the median duration of ECMO was 12 (IQR 18) days, and the median ICU length of stay 25 (IQR 24.5) days. All patients survived hospital discharge with a median hospital length of stay of 51 (IQR 37.5) days.

Discussion

In our case series, we report our experience with awake-ECMO by successfully weaning MV before VVECMO decannulation without the risk of reintubation in patients with severe COVID-19 pneumonia-associated ARDS. In our center, only five (2.4%) patients were successfully extubated to achieve awake-ECMO. All of them were critically ill on ICU admission with severe ARDS and all had failed to improve despite lung-protective strategies and prone ventilation (in fact two patients required handbag ventilation). All patients received two-site cannulation for VVECMO because of severe hypoxemia and intractable respiratory acidosis with the maintenance of lung-protective ventilation. Sedation was minimized and all patients were awake and cooperative allowing for successful extubation. All patients were supported with low-flow or high-flow oxygen therapy after extubation and did not require reintubation. Complications were minimal, with nosocomial infections being the most frequent (present in three out of five patients). All patients were discharged alive from the hospital.

Previously, multiple single case reports and some case series have been published in support of maintaining patients with severe COVID-19 pneumonia and ARDS awake while providing VVECMO support. The use of VVECMO as a salvage therapy for failed MV has been reported for a variable duration with successful extubation while ECMO support continued until full recovery.^8,9 Whereas, Schmidt et al. reported their successful single case experience with complete avoidance of MV albeit a prolonged ECLS duration of 41 days.¹⁰ Similarly, Azzam et al. instituted awake-ECMO for a patient who developed severe surgical emphysema to avoid worsening barotrauma and hemodynamic compromise.¹¹ Furthermore, Aziz et al. reported awake-ECMO for 54 days in a patient who refused intubation and MV.¹²

Earlier during the pandemic, Mustafa et al. reported their experience in 40 patients on VVECMO with the use of single-access, dual-stage right atrium-to-pulmonary artery cannula and extubation on ECMO which appeared to be safe and associated with a favorable outcome for patients with severe COVID-19 ARDS,¹³ thought this cannulation strategy is not widely used. More recently, Assanangkornchai et al. reported the provision of awake-ECMO with conventional cannulation, which failed in four out of seven patients who required intubation because of worsening hypoxemia and delirium leading to poor secretion clearance, nonetheless, six of seven patients survived till hospital discharge.¹⁴ Furthermore, Mang et al. reported the largest cohort to date of an attempt of awake-ECMO in 18 patients which failed in 14 patients (78%) requiring them to be intubated due to either worsening delirium and need for sedation or worsening hypoxemia despite maximal support.¹⁵ They further report that those with successful awake-ECMO had lower BMI and shortest duration between admission to ICU and ECMO canulation, but those who got intubated after starting ECMO had a higher mortality than those managed conventionally with MV and VVECMO. Finally, Gurnani et al. reported on their experience with extubation of 62 VVECMO patients for both COVID-19 and non–COVID-19 patients with favorable outcomes in both, while those 34 patients with COVID-19 being cannulated with dual-stage right atrium-to-pulmonary artery cannula providing right ventricular support.¹⁶

Avoidance of MV during VVECMO support is of interest because it might reduce exposure to prolonged sedation, reduce risk of ventilator-associated pneumonia and diaphragm dysfunction, allow active physical therapy, and allow optimal psychosocial support.¹⁷ This has been successfully reported in bridge to lung transplantation,¹⁸ postthoracic surgery,¹⁹ in immunocompromised patients with pneumonia mostly of undetermined pathogen²⁰ or pneumocystis jirovecii pneumonia²¹ but with a significant risk of intubation (nearly 60%) in the latter due to delirium and agitation. Several factors might make awake-ECMO challenging. Strong spontaneous efforts and associated large transpulmonary pressure swings in awake-ECMO patients might result in harm due to “self-inflicted lung injury.”⁶ Furthermore, sustaining spontaneous breathing might be challenging. Crotti et al. demonstrated while patients bridged to lung transplant and those with chronic obstructive lung disease had more successful rate of extubation on ECMO, only 8 out of 30 ARDS patients were able to maintain spontaneous breathing for some time and only 6 were extubated while ECMO support is continued.²² This might be because of the feasibility of modulation of inspiratory effort and work of breathing by carbon dioxide removal during the recovery phase of ARDS but not during the earlier acute phase.²³ Control of inspiratory efforts remains the main challenge for weaning patients MV and ECMO,⁷ where Al-Fares et al. demonstrated that the patients who failed weaning from VVECMO for ARDS have significantly higher esophageal swing and subsequently larger transpulmonary pressures. Another reason for failed awake-ECMO might be the development of acute complications like bleeding, pneumothorax, bacterial superinfection, sepsis, and disease progression leading to respiratory exhaustion and intubation.¹⁵

Our case series uniquely highlights the consideration physicians should keep in mind when electing to undertake such a strategy. Awake-ECMO is feasible but might only be possible in a very small proportion of patients. Support with conventional MV allows physicians to use sedation and paralysis if needed for safe cannulation for VVECMO without the risk of injury or uncontrolled bleeding. All our patients demonstrated failure of conventional MV within a median of 24 hours of intubation, therefore had minimal risk of ventilator-induced lung injury. Furthermore, when patients were allowed to breathe spontaneously, they had signs of acceptable inspiratory effort to allow extubation on ECMO. None of the patients required reintubation or other significant respiratory support or tracheostomy. All our patients were able to be started on oral diet soon after extubation without the need for nasogastric feeding. Furthermore, all our patients were able to participate in physical therapy, with no requirement of sedation infusion but occasional use of anxiolytics. Finally, the rate of complication was small, with nosocomial infection being most common which is similar to other patients with COVID-19.

Some limitations to our case series exist. Although it is a two-center experience, it contains only a small number of patients, which might be partly due to the nature of the disease with a significant increase in respiratory drive and infection control precautions. In addition, we could not report on those patients extubation was attempted but failed because this was not recorded prospectively. Furthermore, matching comparison to those who were only intubated or failed awake-ECMO was beyond the scope of this study. While long-term radiological and quality of life follow-up is desired, this could not be achieved for this cohort. Finally, daily management and interprofessional team support is program specific and might not be generalizable to other programs.

Conclusions

While liberation from MV first while maintaining VVECMO to provide awake-ECMO for COVID-19 severe ARDS patients is feasible, it can only be achieved in a very small proportion of patients. Advantages of MV liberation before decannulation from VVECMO might out way outweigh any negative aspects that might be faced. It is unclear from our case series and others what factors are associated with the success or failure of this strategy. All these questions are warranted to be tested in further studies especially randomized controlled trials for ARDS of all etiologies.