A United Kingdom Register study of in-hospital outcomes of patients receiving extracorporeal carbon dioxide removal

Abstract

Introduction

Extracorporeal membrane carbon dioxide removal may have a role in treatment of patients with hypercapnic respiratory failure and refractory hypoxaemia and/or hypercapnia.

Methods

We report on the use, outcomes and complications in United Kingdom intensive care units reporting patients on the Extracorporal Life Support Organisation register.

Results

Of 60 patients, 42 (70%) had primarily hypoxic respiratory failure and 18 (30%) primarily hypercapnic respiratory failure. Use of veno-venous procedures increased compared to arterio-venous procedures. Following extracorporeal membrane carbon dioxide removal, ventilatory and blood gas parameters improved at 24 h. Twenty-seven (45%) of patients died before ICU discharge, while 27 (45%) of patients were discharged alive. The most common complications related to thrombosis or haemorrhage.

Discussion

There is limited use of extracorporeal membrane carbon dioxide removal in UK clinical practice and outcomes reflect variability in indications and the technology used. Usage is likely to increase with the availability of new, simpler, technology. Further high quality evidence is needed.

Introduction

Acute respiratory failure is a life-threatening condition where the respiratory system’s function of oxygenation and/or elimination of carbon dioxide is inadequate, resulting in abnormally low levels of oxygen in the blood (hypoxaemia, i.e. arterial oxygen partial pressure (PaO₂) < 8.00 kPa) and/or abnormally high carbon dioxide levels in the blood (hypercapnia, i.e. arterial carbon dioxide partial pressure (PaCO₂) > 6.00 kPa). Hypercapnic respiratory failure may be associated with mild hypoxaemia as found in exacerbations of severe chronic obstructive pulmonary disease (COPD) or with significant hypoxaemia in conditions such as acute respiratory distress syndrome (ARDS) resulting from pathology including pneumonia, sepsis and chest trauma. Patients are often treated with non-invasive ventilation (NIV), which delivers ventilatory support via the patient’s upper airway through a mask interface. For conditions such as COPD, NIV has been demonstrated to prevent the need for mechanical ventilation involving endotracheal intubation, and its complications including lung injury.¹ Rescue therapy has traditionally been invasive mechanical ventilation: this can be a life saving intervention for acute respiratory failure, but despite optimal lung protective ventilation, refractory hypoxaemia and/or hypercapnia may result in some patients. There is a small body of literature describing a role for extracorporeal membrane carbon dioxide removal (ECCO₂R) in this circumstance.² Other developing roles for ECCO₂R include the facilitation of better lung protective ventilation in patients with ARDS as an adjunct to mechanical ventilation.³

Extracorporeal membrane oxygenation (ECMO) and more recently ECCO₂R are techniques that can provide additional and alternative respiratory support for patients using specifically designed membrane gas exchangers derived from cardiac bypass technology for extracorporeal blood flow. When used for respiratory failure, veno-venous (VV) ECMO requires high blood flow rates (3–6 L/min) to provide full respiratory support for patients to enable both adequate oxygenation and carbon dioxide (CO₂) clearance. However when CO₂ clearance alone is required, much lower blood flows are required. CO₂ clearance is determined by multiple factors, including the carriage of CO₂ in the blood (dissolved, as bicarbonate and as carbamino compounds), the gradient between the venous partial pressure of CO₂ and the sweep gas CO₂, sweep gas flow rate, pH, haemoglobin and the efficiency of the gas exchange membrane. From a clinical perspective, CO₂ removal will occur and can be manipulated by alteration in the sweep gas flow rate, which is analogous to minute ventilation in the native lung.⁴ As long as a concentration gradient across the membrane is maintained with adequate fresh gas flow, ECCO₂R may be performed with much lower blood flows which enable limited oxygenation but can achieve substantial CO₂ clearance. This supplementary, or partial, CO₂ clearance effectively allows for reduced minute ventilation by the native lungs. It is possible that this approach may benefit patients with COPD by avoiding mechanical ventilation and allow improved lung protective ventilation in patients with ARDS. The aim is to supplement native pulmonary CO₂ clearance, either to allow a reduction in mechanical ventilator settings with the intention of limiting ventilator induced lung injury in conditions such as ARDS^3,5 or to avoid intubation/facilitate early extubation in patients with conditions such as COPD.

The circuit of the ECCO₂R systems can use either arterio-venous (AV) or VV configuration. AV-ECCO₂R drains blood from the patient’s arterial system through a femoral arterial line and returns it to the femoral vein, effectively creating an artificial AV shunt. Consequently, AV-ECCO₂R relies upon the patient’s circulation and does not require a pump as arterial blood pressure maintains blood flow continuously through the circuit and is returned through the vein. VV-ECCO₂R uses a dual lumen cannula to access blood and a centrifugal pump to drive the blood through the gas exchange membrane and is conceptually similar to renal replacement therapy widely used in critical care. AV-ECCO₂R has an intrinsic risk of arterial injury which may be avoided using the VV technique. Systemic anticoagulation is preferred and recommended but not essential for both AV and VV ECCO₂R.

A high quality systematic review⁶ on the use of ECCO₂R for acute respiratory failure secondary to ARDS included two randomised control trials (RCTs)^7,8 and 12 observational studies with AV and VV ECCO₂R used in seven studies each. Neither RCT showed a significant difference with respect to mortality although in the more recent RCT of low-flow VV ECCO₂R ventilator free days at 28 and 60 days were increased only in the more hypoxaemic subgroup with partial pressure of inspired oxygen (PaO₂)/fraction of inspired oxygen (FiO₂) < 20 kPa. Complication rates ranged from 0 to 25%, the most common complication with AV ECCO₂R being lower limb ischaemia secondary to arterial cannulation. Other adverse events included compartment syndrome and one lower limb amputation. In VV ECCO₂R, clotting within the circuit was the most common complication. A further systematic review of the use of ECCO₂R in exacerbations of COPD included 10 studies at high risk of biased reporting on 87 patients. In 65 out of 70 patients, intubation was avoided with the use of ECCO₂R. Eight out of 11 major complications were bleeding episodes with one venous perforation at the catheter site, one pneumothorax and a retroperitoneal bleed secondary to iliac artery perforation.⁹

National Institute for Health and Care Excellence (NICE) guidance¹⁰ recommends that ECCO₂R should only be used in patients with potentially reversible hypercapnic respiratory failure or those being considered for lung transplantation; the procedure should only be undertaken with special arrangements for clinical governance, consent and audit or research, because the evidence on the safety of ECCO₂R showed a number of well-recognised complications, and evidence on its efficacy was limited in quality and quantity.

A UK survey of Intensive Care Units (ICUs) was carried out with the support of the Intensive Care Society, yielding information on UK clinical practice.¹¹ Out of 141 ICUs (57% response rate), 47 (33%) had used ECCO₂R, although many had used it infrequently (median two patients). The most common indications were pneumonia and asthma, but ECCO₂R had also been used for a number of other indications including in patients with ARDS from non-respiratory sepsis and trauma, in COPD, and as a bridge to lung transplant. AV-ECCO₂R was more frequently used but more complications were reported. The use of VV- ECCO₂R as a newer technology was reported to have increased. The survey indicated ECCO₂R uptake in UK ICUs is characterised by sporadic use for a range of indications.

NICE commissioned the Birmingham and Brunel Consortium External Assessment Centre to facilitate clinical data collection by clinicians caring for patients receiving ECCO₂R in the UK by working with the Extracorporeal Life Support Organisation (ELSO) registry. The collection of UK ECCO₂R patient data in the ELSO register provides an opportunity to consider outcomes including safety in a UK cohort of patients.

Methods

Aim

To report on the use, outcomes and complications of ECCO₂R.

Design

Observational study of ICU patients undergoing ECCO₂R.

Setting

UK ICUs.

Recruitment of centres

The ELSO register dataset was edited to allow the recording of ECCO₂R procedures and outcomes including procedure related adverse events by ELSO members. ICUs that were not already full members of ELSO were identified in 2013 through a survey.¹¹ An associate membership category was introduced to enable UK ICUs that were not full members (these are centres providing ECMO or cardiac extracorporeal support) to register patients receiving ECCO₂R. Hospital Episode Statistics (HES) data, manufacturer data and reported use were cross-tabulated to inform the process of contacting centres while recognising that HES data had some limitations in that data were available only retrospectively and, although the ECCO₂R procedure was recorded, relevant procedures may have been misclassified. ICUs were contacted a second time via email and if needed were telephoned in the summer of 2014 when registration and input of patient data were once again encouraged as appropriate. A further round of targeted email and telephone follow-up of centres that may have had cases was undertaken at the beginning of 2015. A request for anonymised data for research purposes was made to ELSO and a final anonymous dataset was produced in June 2015. The study was carried out under the pre-existing arrangements for clinical governance and agreements between contributing centres and the ELSO register.

Statistical considerations

The co-primary efficacy and safety outcomes of interest were: discharged home or transferred alive from the hospital offering ECCO₂R and adverse events including procedure related complications. Pre-specified adverse events of interest were lower limb ischaemia (including compartment syndrome and amputation), arterial, venous and device thrombus formation, plasma leakage, vascular access damage and bleeding complications. Other outcomes of interest were pre-ECCO₂R blood gases and ventilator settings compared with those 24-h post application.

AV and VV ECCO₂R were predefined subgroups. Although diagnostic information was included in the ELSO register, the indication for using ECCO₂R was not explicitly stated in the ELSO registry. Therefore based on the limited dataset available and taking into account ventilator settings, blood gases and haemodynamics prior to ECCO₂R to derive a consensus opinion, four independent ICU clinicians retrospectively assessed if patients primarily received ECCO₂R to manage hypercapnia or to manage the consequences of hypoxia by enabling lung protective ventilation.

Pre-specified subgroups by indication were ARDS, asthma and COPD. The rationale for subgroup selection was as follows: ARDS is the indication most reported; however use of ECCO₂R in asthma has been reported in the UK; and there are trials registered to evaluate its use in severe exacerbations of COPD. In practice, however, the reported indications did not fit easily into unique subgroups or reporting was constrained by group size and therefore these subgroups have not been reported below.

Given that this was an observational register, no formal sample size calculation was carried out prior to data collection. Statistical analysis is descriptive, with statistical tests having only been performed for pre-specified subgroups where data were sufficient. Tests were appropriate to the population distribution of the relevant variables and where appropriate exact methods and non-parametric tests were used.

Ethical approval and consent to participate

The purpose of the ELSO registry is to provide member institutions data to improve quality of care to patients but data may also be requested by members for research purposes. Data are submitted as a limited de-identified dataset. Given purpose of the register is to collect data for quality improvement and anonymous data are collected, individual patient consent for data entry into the register is not sought. Approval for this study using anonymised data was obtained from the Science, Technology, Engineering and Mathematics Ethical Review Committee of the University of Birmingham (reference ERN_15-0556) prior to the request for research data from ELSO.

Results

ELSO supplied an anonymised download of data on 60 patients registered as having received ECCO₂R in UK hospitals on 11 June 2015.

Patient characteristics and clinical course pre-ECCO₂R

Patient characteristics and pre-ECCO₂R support are described in Table 1. In all, 78% of patients were specified as having received conventional ventilator support, while two (3%) patients received high frequency oscillatory ventilation with the mode of ventilation not specified for the remainder. The median time from admission to ECCO₂R treatment was 96 h (interquartile range 18 to 30 h, n = 58), and the median time from intubation to ECCO₂R was 48 h (interquartile range 24 to 202 h, n = 51).

Table 1.

Patient characteristics and pre-ECCO₂R support.

	Median	Interquartile range	Minimum	Maximum
Age (n = 60)	58	46–68	24	78
	n	%
Male	35	58.3
Female	25	41.7
White	54	90.0
Asian	3	5.8
Black	1	1.7
Other	2	3.4
Hypoxic	42	70.0
Hypercapnic	18	30.0
Ventilatory support
Conventional	47	78.3
High frequency oscillatory ventilation	2	3.3
Unknown	11	18.3
Vasopressor/inotropic drugs	25	41.7
Norepinephrine	13	22.8
Neuromuscular blockers	11	19.3
Steroids	8	13.3
Narcotics	5	8.3
Nitric oxide	2	3.5
Epinephrine	1	1.8
Milrinone	1	1.8

ECCO₂R: extracorporeal membrane carbon dioxide removal.

Of the 42 (70.0%) patients considered to have a primarily hypoxic presentation, 24 (57.1%) had pneumonia, 5 patients septic shock, 2 were specified as having acute respiratory failure without further qualification, 2 influenza and 8 other underlying diagnoses. Of the 18 patients considered to be primarily hypercapnic, 5 had COPD, 3 pneumonia, 2 pneumothorax, 2 asthma and 6 other underlying diagnoses.

Pre-ECCO₂R history

Twenty-two patients (36.7%) were not recorded as receiving any specific organ support prior to ECCO₂R, while 18 had one organ support coded, 7 had 2 and 13 more than 2. The most common organ support recorded was vasopressor/inotropic drugs (Table 1).

Three patients suffered a cardiac arrest prior to being commenced on ECCO₂R. One had asthma requiring manual hand ventilation and was eventually discharged alive. One patient had a nutritional/metabolic cardiomyopathy and asthma, another parainfluenza virus pneumonia and both subsequently died despite recovery from respiratory failure. A further patient with bronchiectasis had ECCO₂R as a bridge to lung transplant also died despite recovery from respiratory failure. One patient with cerebral oedema also required manual hand ventilation.

ECCO₂R treatment and outcomes

ECCO₂R treatment characteristics are described in Table 2 below. AV ECCO₂R predated VV ECCO₂R, with VV-ECCO₂R becoming the most prevalent once this technology became available (Figure 1).

Table 2.

Ventilator settings, blood gases and haemodynamics pre-ECCO₂R and at 24 h of ECCO₂R.

		Worst values in 6-h pre-ECCO2R				Best values at 24 h of ECCO2R
	n	Median	Interquartile range	Minimum	Maximum	n	Median	Interquartile range	Minimum	Maximum	p (Wilcoxon)
Rate (Hz)	37	22	17.0–28.0	4	42	36	18	14–24	8	38	.002
Mean airway pressure	21	16	9–27	5	35	21	15	10–23	5	37	.033
FiO2	54	.70	.50–.90	.21	1.00	.53	.55	.38–.70	.21	1.00	<.000
Peak inspiratory pressure/amplitude	49	30	26.0–33.2	18	72	45	24	20–28	8	74	<.000
Positive end-expiratory pressure	46	8	5.0–12.0	0	20	43	10	5–12	1	20	0.032
pH	55	7.1	7.1–7.3	6.8	7.6	55	7.4	7.3–7.4	7.2	7.5	<.001
PaCO2 (kPa)	55	11.4	9.0–14.0	3.9	17.0	55	7.0	6.1–8.0	4.3	11.0	<.001
PaO2 (kPa)	54	10.5	9.0–13.0	3.6	20.1	55	9.3	8.1–10.7	3.2	17.0	<.004
PaO2 (kPa)/FiO2	52	.17	.12–.23	.05	.48	52	.17	.14–.24	.05	.57	0.555
Systolic blood pressure	39	110	96 -130	64	207	39	123	110–135	83	168	.043
Diastolic blood pressure	39	57	52–73	42	125	39	60	55–70	40	95	.980
Mean arterial pressure	35	72	67–95	52	141	35	84	73–95	61	122	.301

ECCO₂R: extracorporeal membrane carbon dioxide removal.

Figure 1.

Mode of ECCO₂R by treatment year. ECCO₂R: extracorporeal membrane carbon dioxide removal.

Patients receiving VV ECCO₂R had relatively fewer hours of ECCO₂R. The median duration and interquartile range of hours on ECCO₂R treatment were 192 (108–324) for AV (n = 17) and 120 (95–208) for VV (n = 38) (Mann–Whitney test not statistically significant).

Blood flow rate at four and 24 h was only recorded for three patients receiving AV ECCO₂R. Where the flow rate unit was recorded in ml/min, this was converted to L/min except for one patient where the incorrect unit of measurement seemed to have been recorded. The median and interquartile range at 4 h (n = 40) for VV-ECCO₂R was 0.47 L/min (0.40 to 1.15) and at 24 h (n = 38) was also 0.47 L/min (0.42 to 1.13) (Wilcoxon test not statistically significant). Cannulation is described in Table 3.

Table 3.

Cannulation.

	n
Mode
AV	22
Cannulation (where specified)
Left femoral vein/left femoral artery	3
Left femoral vein/right femoral artery	4
Right femoral vein/left femoral artery	14
Unspecified	1
VV	38
Cannulation (where specified)
Left internal jugular vein	2
Right femoral vein	10
Right internal jugular vein	10
Unspecified	16

AV: arterio-venous; VV: veno-venous.

When ventilation parameter settings were compared, taking those during the worst ABG values in the 6 h prior to ECCO₂R and those at the best ABG value at 24 h following application of ECCO₂R, all ventilation settings were shown to have been reduced (Table 2). Median PaCO₂ was reduced with a concomitant improvement in median pH.

In all, 27/60 (45%) patients were discharged from the hospital alive. Of the 27/60 (45%) patients who died in the course of the procedure, 21 had multi-organ failure, four were described as having diagnosis incompatible with life and two died following family requests. Thirty-three (55%) patients survived ECCO₂R but a further 6/60 (10%) died prior to discharge from the ECCO₂R centre. Overall survival to discharge rate of 45%. Thirteen patients, 48.1% of those discharged alive, were discharged home. Nine out of 22 patients receiving AV ECCO₂R and 18 out of 38 patients receiving VV ECCO₂R were discharged alive (p = .907). In all, 20/42 (47.6%) of hypoxic patients compared with 13/18 (72.2%) hypercapnic patients survived the procedure (p = 0.428) and 17/42 (40.5%) compared with 10/18 (55.6%) were discharged alive (p = 0.141). Age, hours of ECCO₂R treatment, time from intubation to ECCO₂R, worst PaCO₂ in the 6 h before ECCO₂R, worst PaO₂ in the 6 h before discharge, worst pH in the 6 h before discharge and time from admission to ECCO₂R were not associated with death before discharge (Mann–Whitney test). Ten women (40.0%) compared with 23 men (65.7%) died before discharge (p = 0.087).

Nineteen patients (31.7%) experienced complications (Table 4), with 11 having 1 complication, 2 having 2, 3 having 3, 2 having 4 and 1 having 7. Fifteen patients (39.5%) of those receiving VV ECCO₂R and four (23.5%) of those receiving AV ECCO₂R had one or more complication. Three patients receiving AV ECCO₂R and four receiving VV ECCO₂R had cannulation site bleeding.

Table 4.

Complications.

	n	% of patients
Mechanical
Gas exchange membrane failure	1	1.7
Pump malfunction	2	3.3
Clots: oxygenator	1	1.7
Clots: other	2	3.3
Cannula problems	1	1.7
	7	11.7
Haemorrhagic
GI haemorrhage	1	1.7
Cannulation site bleeding	7	11.7
Haemolysis (plasma  free Hb > 50 mg/dl)	1	1.7
Surgical site bleeding	1	1.7
	10	16.7
Neurologic
Seizures: EEG determined	1	1.7
CNS haemorrhage by US/CT	1	1.7
	2	3.3
Renal
Creatinine 1.5–3.0	1	1.7
Haemofiltration required	5	8.3
	6	10.0
Cardiovascular
Inotropes	4	6.7
Cardiac arrhythmia	2	3.3
	6	10.0
Other
Pneumothorax requiring treatment	1	1.7
Culture proven infection	5	8.3
pH < 7.20	1	1.7
Hyperbilirubinemia  (>2 direct or >15 total)	1	1.7

CNS: central nervous system; CT: computed tomography; EEG: electroencephalogram; GI: gastrointestinal; US: ultrasound.

Discussion

This observational study of patients of patients undergoing ECCO₂R treatment in the UK reported to the ELSO registry displays considerable clinical heterogeneity, which may be reflected in the mixed outcomes at discharge. While ICUs were actively encouraged to register patients, complete population coverage was not achieved, in part reflecting unanticipated use of ECCO₂R for a single patient in many centres. Findings reflect changes in practice over time. Use of ECCO₂R in a VV configuration is increasing in respect to AV, advances in the technology and its perceived relative safety. In this series, it appears that the procedure is used in some patient groups to manage hypercapnia and in others, primarily hypoxic patients, to facilitate lung protective ventilation. ECCO₂R showed some success in reducing PaCO₂ and ventilator settings with the potential for lung protection, though only limited evidence of efficacy can be provided by a register study. ECCO₂R has been used in a relatively sick patient cohort: only 45% left hospital alive. As only a small number of patients received high frequency oscillatory ventilation, it was not possible to specifically look at this subgroup.

Complication rates were higher than previously reported in a systematic review⁵ but mechanical complications were relatively infrequent, with cannulation site bleeding being the most frequent procedure related adverse event. Assessment of whether other reported complications are procedure related is problematic, given the complexity of the included patients’ condition. Interpretation is difficult, given changing technology and experience: ECMO requires larger cannulae and no arterial puncture is required for VV ECCO₂R. In contrast to earlier studies, lower limb ischaemia, compartment syndrome and amputation were not reported: this may be related to increasing experience and use of VV technology. Given the sporadic use of the procedure in many centres and patient heterogeneity, further multicentre observational studies to capture procedure related adverse events and patient outcome, ideally linked to national audit data, would be of value.

Survival in this relatively sick cohort is within the range reported for observational studies of ECCO₂R in ARDS. There are insufficient patients in the study to confirm a difference in outcomes based on hypoxaemia vs hypercapnia though the data might be considered to support the conclusion that ECCO₂R enables a reduction in mechanical ventilation requirement in this group of patients. Evidence on comparative effectiveness from RCTs in well-characterised patient cohorts is needed. Ongoing trials such as the National Institute of Health Research Health Technology Assessment funded trial of protective ventilation with VV lung assist in patients with acute hypoxic respiratory failure (REST trial; ISRCTN31262122) and an ongoing trial in exacerbations of COPD (NCT02086084) should inform the effectiveness of ECCO₂R.

Availability of supporting data

The data that support the findings of this study are available from the Extra Corporeal Life Support Organisation (ELSO) but restrictions apply to the availability of these data, which were used under license for the current study in line with ELSO Data Use Agreements, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Extra Corporeal Life Support Organisation.

Authors’ contributions

CC contributed to the conception and design of the study, performed statistical analyses and contributed to the interpretation of data, and drafted the manuscript. NB, an ELSO steering committee member, requested anonymised data for research purposes from ELSO. AB, NB, D McA, J McN and HP contributed to the conception and design of the study, the interpretation of data and revision of the manuscript. AB, NB, DMcA and JMcN assessed the primary purpose of ECCOR in each patient, providing a consensus opinion. All authors have read and approved the final version of the manuscript.

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Daniel F McAuley and James McNamee: NIHR HTA funding (13/143/02) and NI HSC RDD funding to undertake a clinical trial of ECCO₂R with in kind contribution from ALung Technologies, Inc. Nicholas A Barrett: No personal financial conflicts of interest. Educational and research funding from Maquet, Drager, Fisher & Paykel, Mitsubishi Tanabe Pharmaceuticals, Corpak, ALung Technologies.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: University of Birmingham staff working on the study received funding support from NICE. NHS England provided funding to support ELSO register database modifications and ICU registration.