Extracorporeal Carbon Dioxide Removal to De-escalate Venovenous Extracorporeal Membrane Oxygenation in Severe COVID-19 Acute Respiratory Distress Syndrome

Abstract

Objectives:

In a subset of patients with severe COVID acute respiratory distress syndrome (ARDS) there was a need for extracorporeal membrane oxygenation (ECMO) for pulmonary support. The primary extracorporeal support tool for severe COVID ARDS is venovenous (VV) ECMO, however after hypoxemic respiratory failure resolves many patients experience refractory residual hypercarbic respiratory failure. Extracorporeal Carbon Dioxide Removal (ECCO₂R) for isolated hypercarbic type II respiratory failure can be used in select cases to deescalate patients from VV ECMO while the lung recovers the ability to exchange CO₂. The objective of our study is to describe our experience in using ECCO₂R as a bridge from VV ECMO.

Design:

Hemolung RAS, is one of the commercially available (ECCO₂R) device, and the Food and Drug Administration in the United States accelerated its use under its Emergency Use Authorization for the treatment of refractory hypercarbic respiratory failure in COVID-19 induced ARDS. This created an environment in which selected and targeted mechanical circulatory support therapy for refractory hypercarbic respiratory failure could be addressed. This retrospective study describes the application of Hemolung RAS as a VV-ECMO de-escalation platform to treat refractory hypercarbic respiratory failure after resolution of hypoxemic COVID ARDS.

Setting:

Quaternary Care Academic Medical Center, Single Institution

Participants:

Patients with refractory hypercarbic respiratory failure after COVID ARDS who were previously supported with VV ECMO

Measurements and Main Results:

21 patients were placed on ECCO₂R after VV ECMO for COVID ARDS. 17 patients were successfully transitioned to ECCO₂R and then decannulated, 3 patients required re-escalation to VV ECMO secondary to hypercapnic respiratory failure, and one patient died while on ECCO₂R. 5 (23.8%) of the 21 patients were transitioned off of VV ECMO to ECCO₂R with a compliance of < 20 (mL/cmH2O). 3 of these patients with low compliance reesclated to VV ECMO.

Conclusions:

ECCO₂R can be used to continue supportive methods for patients with refractory type 2 hypercarbic respiratory failure after COVID ARDS for patients previously on VVECMO. Patients with low compliance have a higher rate of reescalation to VV ECMO.

Introduction

Supportive management for acute respiratory distress syndrome (ARDS) may require venovenous extracorporeal membrane oxygénation (VV ECMO) when the patient’s gas exchange requirements cannot be met by conventional optimized medical management. Clinically ARDS can present as hypoxemia (Type 1) and/or hypercarbic (Type 2) respiratory failure; with or without associated ventilator induced lung injury. There are different mechanical circulatory support (MCS) tools to address each aspect of lung failure. Hypercarbia and hypoxemia can be treated separately with two distinct MCS tools, ECMO which addresses both issues, and extracorporeal carbon dioxide removal (ECCO₂R) which addresses CO₂ removal only. During the pandemic, the Food and Drug Administration (FDA) granted Emergency Use Authorization (EUA) for Hemolung Respiratory Assist System (RAS). Hemolung RAS functions as an ECCO₂R device.

Typically, vascular access for VV ECMO and ECCO₂R is obtained percutaneously. Both modalities are venous system series-circuit MCS tools in which blood from the central venous circulation is diverted to an extracorporeal membrane where gas exchange occurs, and returned though a cannula to central venous circulation.^1,2 Both platforms use a sweep gas, that is introduced into a hollow fiber tubule which is in contact with blood and results in removal through diffusion of CO₂ from the blood stream. (Figure 1) Differing concentration of gases, create a gradient and facilitate gas exchange with the blood, and the resultant gas byproduct is expired though the gas outlet.^1,2 The greatest difference between the VV ECMO and ECCO₂R is in the total blood flow required. ECCO₂R operates at lower blood flow rates in comparison to VV ECMO. VV ECMO targets systemic oxygen delivery goals of at least 300 mL/m^{^2}/min, or at least 60% of native cardiac output, thereby requiring 3-6 L/min of extracorporeal blood flow.² ECCO₂R platform blood flow rate ranges from 350 - 500mL/min.¹ Due to the lower blood flow requirement for ECCO₂R, one of the advantages of ECCO₂R is that it allows for the use of smaller bore central venous catheters. Furthermore, the access cannula for ECCO2R is a dual lumen single site central line similar to a dialysis catheter and therefore less access sites are needed. This results less associated vascular complications like bleeding, DVT, and infection.

Figure 1.

A) ECCO2R console. B) Membrane gas exchanger. C) Cannulation location

Once patients are on VV ECMO for COVID ARDS with improvement in oxygenation as the lungs recover, there is a high percentage of patients that have refractory hypercarbic respiratory failure with resolution of hypoxemia.³ There are a limited number of prediction measures and variables that can ensure the success of decannulating from VV ECMO without the need for reescalation. Factors that may predict successful VV ECMO decannulation include tolerating sweep off for a period of time, high lung compliance, and resolution of the underlying cause of ARDS. Furthermore, weaning off VV ECMO without a secondary MCS tool can be made more challenging due to attempted simultaneous ventilator weaning resulting in tachypnea, ventilator desynchrony, introduction of increased metabolic demand (physical therapy, wakefulness, as well as increased nutrition) all of which can result in increased CO₂ production, in the backdrop of recently recovered fragile ARDS lungs. At the time of decannulation the question becomes, can the fragile recently recovered lungs cope with the increased demands of CO₂ removal without VV ECMO support? Deescalation from VV ECMO, and MCS in general, in patients with COVID ARDS, may be more successful if a stepwise approach is performed. ECCO2R can be a useful tool in refractory type II respiratory failure, for patients who are on VV ECMO with resolution of hypoxemia, now entering the decannulation phase. This study is a retrospective review of using ECCO₂R as a de-platforming strategy from VV-ECMO in COVID-19 ARDS patients at a single institution.

Methods

This study is a retrospective case series. The patients were treated at the University of Pennsylvania, which is a quaternary care medical center with ECMO and ECCO₂R capabilities from May 2020 to March 2022. One hundred and seventeen patients were placed on ECMO support (107 VV ECMO and 10 VA ECMO) for COVID ARDS.

VV ECMO management

Basic parameters for weaning VV ECMO were defined as follows: A stepwise reduction in sweep gas flow rate by 0.5 - 1L/min, while targeting pH goal of > 7.3 and P_aCO₂ < 65 mmHg. Initial management of the ventilator at the onset of VV ECMO support included tidal volume < 6 cc/kg (ideal body weight), respiratory rate < 18 breaths per minute, and plateau pressure of < 30 cm H20. At our institution, the primary goal for VV ECMO support in COVID ARDS was ventilator weaning first and then ECMO weaning. Patients were only removed from ECMO when sedation and paralysis were discontinued as well as mechanical ventilation goals reduced to pressure support ventilation trials or tracheostomy collar trials. At the outset of VV ECMO support, initial primary mode of ventilator support was pressure control ventilation. All patients were cycled through prone positioning after initiating VV ECMO. At the time of ECMO de-escalation, sweep gas was reduced to 1 L/min or less and F_iO₂ requirements of 40% or less. If the patients at this point were off infusion sedation, defined as continuous infusion of propofol, midazolam, fentanyl or hydromorphone as well as off paralytics, and on pressure support ventilation trials or tracheostomy collar trials then they were deemed ready for minimal to sweep off trial. In the sweep off trial the sweep gas was set to 0.5 or 0 L/min. Thirty minutes into the sweep off trial, the arterial blood gas (ABG) was obtained. The ABG will be subsequently checked every 6 hours for a minimum total duration of 24 hours of sweep off trial. The sweep off trial was considered a failure if the pH dropped below 7.3 and/or P_aCO₂ exceeded 65 mmHg at any point in the first 24 hours after sweep off. Furthermore, the sweep off trial was considered a failure if the respiratory rate exceeded 24 breaths per minute or the minute ventilation exceeded 12 L/min. If patients failed sweep off trial 3 times without escalation in F_iO₂ requirements, they were considered a candidate for ECCO₂R therapy. All VV ECMO circuits were maintained on bivalirudin. (Table 2)

Table 2:

ECMO data for type of VV ECMO circuit and pump head, average time on ECMO and complication events while on ECMO. Anticoagulation medication used and PTT target. * PTT = Partial Thromboplastin Time

VV ECMO Data
ECMO Circuit, # (%)	ECMO run time (mins)	Adverse Events, # (%); total 28	Anticoagulation, #, %	PTT goal, #(%)
Cardiohelp, 5 (23.8)	67,374.45 (average)	Venothromboembolism, 6 (17.6)	Bivalirudin 21 (100)	40-50, 16 (76.2)
Rotoflow, 14 (66.7)	69,435.5 (median)	Hematologic Circuit Obstruction, 3 (13.6)	Heparin 0 (0)	45-55, 1 (4.8)
Spectrum, 2 (9.5)		Parenchymal bleeding, 3 (13.6)		50-60, 4 (19)
		Interventional site bleeding, 4 (18.2)
		Console Error/Malfunction, 2 (5.9)
		Cannula dislodgment, 1 (2.9)
		Chest tube insertion (pleural eff/PTX), 8 (23.5)
		Terminal delivery, 1 (2.9)
		No Events, 7 patients

ECCO₂R

Access for ECCO₂R was obtained through the right internal jugular vein in all patients. When patients were selected for ECCO₂R, first VV ECMO was decannulated at the bedside in the intensive care unit. Subsequently, after obtaining hemostasis after decannulation, new access was obtained for the Hemolung RAS. There were some patients who originally passed sweep off trial and were decannulated from VV ECMO however required ECCO₂R at a time point not at the time of ECMO decannulation. Initially ECCO₂R sweep was set at 10 L/min and flow at 500 cc/min. For this study the primary endpoint collected was weaning ECCO₂R to recovery versus reescalation to VV ECMO. Secondary data end point included dynamic lung compliance, which was calculated in setting of invasive mechanical ventilation via delta volume divided by delta pressure. Lung mechanics were expressed as a 6-hour average. Delta pressure was calculated as plateau pressure (P_plateau) minus positive end expiratory pressure (PEEP) in volume control, inspiratory pressure (IP) in pressure control, and peak pressure (P_peak) minus PEEP in pressure support.⁴ Three data points for ventilatory mechanics were collected for each patient while undergoing treatment with ECCO₂R. At least one of these data points was within the first 24 hours of ECCO₂R cannulation. Additional data collected includes, demographics, ventilatory mechanics, ECMO circuit type, extracorporeal run time, adverse events, and anticoagulation strategy.

Adverse events were defined as complications related to bleeding, clot formation, circuit malfunction and thoracic space occupying complication which limited lung compliance. Circuit malfunction included console errors, cannula dislodgment, and obstructive intra-circuit blood clots. Bleeding complications were classified as access site bleeding and parenchymal bleeding. Access site bleeding included bleeding associated with invasive procedure requiring multiple transfusions ( > 1 U PRBC within 24 hours). Parenchymal bleeding was defined as any bleeding of an un-instrumented organ system. Clot forming complications included left ventricular thrombus, deep vein thrombosis and pulmonary embolism. Patients on ECCO2R were maintained on Bivalirudin or Heparin with typical goals of 40 – 50 mseconds unless there was ongoing clotting in which the target was increased. (Table 3)

Table 3:

ECCO₂R data for type of VV ECMO circuit and pump head, average time on ECCO₂R, and complication events while on ECCO₂R. Anticoagulation medication used and PTT target. * PTT = Partial Thromboplastin Time

ECCO2R data
Hemolung run time (mins)	Adverse Events, # (%); total 15	Anticoagulation, # (%)	PTT goal, # (%)
14,267 (average)	Death, 1(4.5)	Bivalirudin, 12 (57.2)	40-50, 15 (71.4)
28,779.5 (median)	VTE, 0	Bivalirudin + Aspirin 81 mg, 5 (23.8)	45-55, 0
	Hematologic Circuit Obstruction, 8 (36.4)	Unfractionated heparin, 4 (19)	50-60, 4 (19)
	Parenchymal bleeding, 1 (4.5)		55-65, 2 (9.5)
	Interventional site bleeding, 0		60-70, 1 (4.7)
	Console Error/Malfunction, 1 (4.5)
	Cannula Dislodgement, 0
	Chest Tube insertion, 1 ( 4.5)
	Inadequate CO2 removal resulting in respiratory acidosis, 3 (13.6)
	None 7 patients

Results

In a period from May 2020 to March 2022, a total of 21 patients were placed on ECCO₂R after VV ECMO for COVID ARDS. Eighteen patients (85.7%) were male, and the mean age for all patients was 45 ± 10.7 years. (Table 1)

Table 1:

Demographic data for patients deescalated with ECCO2R from VV ECMO.

Age	45 +/− 10.7
Sex (%)	18 male 3 Female	85.7%/14.3%
Co-morbidity, #(%)
Diabetes	4	19 %
Hypertension	5	23.8 %
Chronic Pain	1	4.8 %
Arrhythmia	1	4.8 %
Heart Failure with Reduced Ejection Fraction	1	4.8 %
History Thromboembolism	5	23.8 %
Obesity	11	52.4 %
ADHD	1	4.8 %
Reactive Airway Disease	2	9.5 %
Anxiety	1	4.8 %
Coronary Artery Disease	3	13.6 %
Pregnancy	2	9.5 %

The average VV ECMO run time was 67,374 ± 31,260 mins (46.8± 21.7 days). Average ECCO₂R run time was 14,257 ± 11,537 mins (9.9 days ± 8 days). (Table 2) Seventeen (80.9%) patients were placed on ECCO₂R at the time of ECMO decannulation while 4 (19.1%) were placed after ECMO decannulation. The 4 patients who were placed on ECCO₂R remotely all passed standard 24 hour sweep off trial at F_iO₂ less than 40%, and were cannulated on ECCO₂R 1, 2, 6, and 29 days post VV ECMO decannulation. Twenty-eight adverse events occur while patients were on VV-ECMO, and 15 adverse events while on ECCO₂R . (Table 2 and 3)

Thirteen patients underwent 0 L/min sweep trials, 3 patients underwent 0.5 L/min sweep trials and 5 patient underwent 1 L/min sweep trials. Seventeen patients were successfully transitioned to ECCO₂R and then decannulated, 3 patients required re-escalation to VV ECMO secondary to recrudesce of hypoxic respiratory failure, and one patient died while on ECCO₂R. Of the three patients who were transitioned back to VV ECMO, one had a zero sweep trial, and the other two had a 1 L/min sweep trial. In these three cases, the compliance at the time of decannulation for the first patient was 20 ml/cmH₂0 on pressure support ventilation through a tracheostomy after 62 days of VV ECMO support. The patient had tolerated a 24 hour 0 LPM sweep trial. Two days after decannulation, the patient suffered an asystolic cardiac arrest (<1 min) with ROSC. The patient underwent CT chest noting increased fluid burden of bilateral hemopneumothoracies and associated consolidations. The blood gas revealed hypercarbic respiratory failure and was the patient was placed on ECCO₂R for support 2 days after VV ECMO decannulation. Two days after ECCO₂R placement her compliance decreased to 11 ml/cmH20 and the patient had progressive hypoxemia. VV ECMO was reinitiated without further lung recovery and she was subsequently bridged to bilateral orthotopic lung transplantation. The second patient was on VV ECMO for 28 days. The patient tolerated a 1 LPM sweep trial with a compliance of 31 ml/cmH20 with ongoing pressure support trials on day 21 and was decannulated from VV ECMO. On ECCO₂R day 21 the patient had a tracheostomy failure with balloon rupture and loss of pressure and derecruitment of lung volumes. The compliance had reduced to 8 ml/cmH₂0 and ensuing hypoxemia required reescalation to VV ECMO. The patient was decannulated from VV ECMO a second time 2 weeks later with a compliance of 50 ml/cmH₂0 on pressure support. This time he did not require deescalation with ECCO₂R. The final patient who reesclated to VV ECMO had COVID ARDS and was placed on VV ECMO for 41 days. She was decannulated from VV ECMO with a compliance of 11 ml/cmH₂0 and a sweep trial of 1 LPM. The patient tolerated hemolung for 24 hours and was required to be placed back on VV ECMO for progressive hypoxia. Ultimately the patient required another 30 days of VV ECMO and was eventually decannulated without ECCO₂R deesclation after a sweep off trial and compliance of 22 ml/cmH₂0.

Twelve out of 13 patients (92.3%) who underwent a zero-sweep trial were successfully decannulated from ECCO₂R, all patients with a 0.5 L/min sweep trial were successfully decannulated from ECCO₂R, and 2 out of 5 patients with a 1 L/min sweep trial were successfully decannulated from ECCO₂R. The three patients who were not decannulated from ECCO₂R from this group, 2 were placed back on VV ECMO and one died. Ultimately, the three patients who were re-escalated to VV ECMO were then bridged to recovery without requiring re-initiation of ECCO₂R. (Table 4)

Table 4:

ECCO₂R weaning data from VV ECMO to success decannulation, categorized by sweep trial and compliance at the time of VV ECMO decannulation. * LPM = Liters per minute.

	0 LPM Sweep Trial	0.5 LPM Sweep Trial	1 LPM Sweep Trial	Total
Total patients categorized by sweep trial	13	3	5	21
Patients with hypercarbic respiratory failure recovery and decannulated from ECCO2R	12	3	2	17
Patients with rebound hypoxemic respiratory failure recannulated to VV ECMOx	1	0	2	3
Patients who deesclated to ECCO2R but died while on the device	0	0	1	1
Data by compliance and sweep trial
Compliance ≤ 20 mL/cmH20	3	0	2 Both reescalated to VV ECMO	5
Compliance > 20 mL/cmH20	10	3	3	16

Withdrawal of advanced MCS care was performed on one patient while on ECCO₂R in the setting of multisystem organ failure, septic shock secondary to recurrence of methicillin sensitive staphylococcus aureus bacteremia, with associated new bibasilar lung consolidations and difficulty maintaining ventilatory goals despite maximal ECCO₂R support. This patient had a 1 LPM sweep off trial prior to decannulation from VV ECMO and initiation of ECCO₂R . Table (4) provides further description of sweep flow, frequency of re-escalation, and dynamic compliance data at the time of initiating ECCO₂R.

When categorizing the data by compliance at the time of decannulation, in total five (23.8%) of the 21 patients were transitioned off of VV ECMO to ECCO₂R with a compliance of < 20 (mL/cmH2O). Two of these patients with low compliance reesclated to VV ECMO. Dynamic compliance for the duration of the ECCO2R run, was subdivided into categories of ECCO₂R success and failure, which was represented in box plot in figure (1). Failure was considered re-initiation of VV ECMO secondary to refractory hypercapnic respiratory failure. Compliance data that was excluded from this analysis include one patient whom was cannulated on ECCO₂R 29 days after VV ECMO de-cannulation.

Discussion

The COVID-19 pandemic stressed hospital resources globally. VV ECMO has been traditionally utilized in cases of hypercarbic/hypoxic respiratory failure refractory to optimal medical management. Clinically, we observed a large percentage of COVID ARDS patients on VV ECMO, over time, resolve their hypoxemia and improve oxygenation, however; these patients tended to have residual hypercarbic respiratory failure, such that they required de-escalation with ECCO₂R. Hemolung RAS, a ECCO₂R device, was made available during the pandemic which allows for a select group of patients with hypercarbic type II respiratory failure to have a mechanical circulatory support option with continued lung rest. We used the ECCO₂R device to successfully deescalate VV ECMO patients with COVID ARDS to recovery without precipitating further lung damage.

There is a significant debate and lack of clarity surrounding VV ECMO weaning strategy⁵. The Extracorporeal Life Support Organization provides general guidelines for weaning strategies for VV ECMO. Criteria for VV ECMO patients to undergo weaning trials contained in these guidelines include: Ventilator and ECMO F_iO₂ ≤ 60%, positive end expiratory pressure ≤ 10 cm H₂0, P_aO₂ ≥ 70 mmHg, tidal volume ≤ 6 cc/kg ideal body weight, P_plateau ≤ 28 cm H₂O, and respiratory rate ≤ 28 breaths per minute. Patient’s arterial blood gas should have acceptable pH and P_aCO₂ based on clinical condition and improvement in chest radiograph appearance.² The underlying cause of ARDS should be resolved, such as treatment of pneumonia or resolution of trauma induced ARDS, etc. The ECMO wean consists of incrementally reducing the sweep gas flow and ECMO and ventilator F_iO₂ while meeting the weaning requirements.² Once minimal extracorporeal support is reached, a sweep off gas challenge can be performed for at least 2-3 hours.²

Likewise, there has been limited publications both in the utility of ECCO₂R as well as the weaning strategies for ECCO₂R.⁶ Theoretical ECCO₂R weaning parameters are similar to the ECMO weaning guidelines mentioned above. Criteria included: ECCO₂R therapy applied for at least 48 hours, P/F ratio > 200 mm Hg prior to weaning, tidal volume 6 cc/kg IBW, positive end expiratory pressure ≤ 10 cm H₂0. Furthermore, in particular, attention to be paid to minute ventilation and driving pressure for CO₂ weaning. Goal driving pressure of < 14 cm H₂O, respiratory rate < 25 breaths/min, minute ventilation < 10 L/min, and pH > 7.30.⁶ The strategy for weaning ECCO₂R support is a gradual reduction in sweep gas flow in incremental steps, and once minimal support reached, to trial sweep off after a minimum of 12 hours of stability.⁶

There are minimal publications pertaining to the transition between VV ECMO and ECCO₂R support. The strategy we applied consisted of weaning the VV-ECMO via the aforementioned guidelines, reaching a sweep gas flow of 1L/min or less. If all other goals were achieved at that time, we decided to transition the patient over to ECCO₂R support. The purpose of this transition being to facilitate lung protective strategies, achieve gas exchange goals, and liberate VV-ECMO pumps for more acutely/severely ill patients.

In review of our 21 patients, it appeared that a lung compliance of < 20 correlated with a higher probability of failure to wean from ECCO₂R and required re-cannulation of VV-ECMO. When we examined the cases of failure more closely, we noted that 2 of the 3 patients had acute events unrelated to the ECCO₂R, that significantly affected lung compliance, and subsequently CO₂ removal requirements to maintain a normal pH.

It is difficult to assess the relationship of adverse events, anticoagulation strategy, and extracorporeal device type applied in this review. Despite the much longer run time of VV ECMO, we did not see a proportionally increased number of adverse events. This may be due to our limited sample size, but also the fact that ECCO₂R was applied in a setting post VV-ECMO run. It is noted in the literature that blood flow at both higher rates (tested at 4L/min) and lower rates (tested at 1L/min) during extracorporeal support both contribute to adverse events such as increased hemolysis, platelet activation and bleeding complications.⁷ Centrifugal pumps show increased blood damage when increasing motor speed; however this does not mean that low flow pumps rates lead to less damage.⁷ The authors of this study extrapolate that low-flow rates are plausible explanation for increased hemolysis due to increased re-circulation within the pump, leading to longer pump contact time with the blood.⁷ Another study looking at the use of ECCO₂R specifically noted that thrombocytopenia, hemolysis, factor XIII deficiency and acquired von Willebrand syndrome were typical findings, that were reversible when the circuit was discontinued.⁸ These findings are similar hemolytic factors that typically are present in VV-ECMO.⁹

Conceptually, ECCO₂R allowed us to avoid over sedation and thus we did not have to increase the minute ventilation when patients transitioned from VV ECMO to ECCO2R. We know that wakefulness produces more CO₂ due to increased metabolic production. We know that lightening sedation also produces increased metabolic rate and resultant higher CO₂. In this context, the fragile, recently recovered COVID ARDS lung, is unable to handle this level of gas exchange. Second, it is precisely at this stage that nutritional intake is increased such that patients had a basal metabolic rate of 1.5x. We found that the COVID ARIDS recovering lungs are not able to handle this new production. ECCO₂R provided a bridge to recover for that gap in CO₂ clearance. In conclusion it appears plausible to use ECCO2R as a means to de-escalate from VV-ECMO in the setting of isolated hypercarbic respiratory failure. It may provide a safety zone from complete decannulation from VV ECMO mechanical support as a transition from VV-ECMO to ECCO₂R if the patient is tolerating sweep off trial. And it may be a mechanism to transition patients who are on minimal sweep settings of.5 or 1 L/min. Patients who had a dynamic compliance of > 20 (mL/cmH2O) were found to have more likely to successfully wean from ECCO₂R and may be used as a cutoff for those who may be ideal candidates for a successful ECCO₂R transition.

Limits to this analysis include small sample size, un-evenly distributed groups for transitional sweep, and novelty of use of both device and disease. We are unable to draw conclusions from our adverse event data to provide insight into comparative hemolytic profile between devices. Further studies will be needed to assess comparative safety between devices, and thereby contribute to the usefulness of transition between devices.