Extra-corporeal Carbon Dioxide Removal as an adjunct to Non-Invasive Ventilation in Acute Exacerbations of Chronic Obstructive Pulmonary Disease

The second element of the study was to compare NIV and ECCO₂R in patients with an acute exacerbation of COPD (AECOPD) at risk of failing NIV. NIV failure is complex and relates to severity of illness, patient factors including agitation, delirium, tendency to claustrophobia and distress as well as NIV device factors, especially relating to device settings and mask fit/comfort. If patients fail NIV, they commonly require tracheal intubation and mechanical ventilation. In AECOPD, this can result in prolonged periods of mechanical ventilation leading to a tracheostomy and is often punctuated by recurrent episodes of hospital acquired infection. Patients receiving mechanical ventilation are at significant risk of muscular deconditioning and ICU acquired weakness as well as delirium. All these potential sequelae have significant impacts on long term outcomes. At present studies using ECCO₂R in AECOPD have been limited to observational studies without a contemporaneous control group. Hence there is a need for a randomised, controlled trial comparing outcomes for patients with AECOPD receiving NIV or ECCO₂R.

There has been minimal work on the impact of ECCO₂R on respiratory physiology, especially in patients with AECOPD.

1. to assess the CO₂ removal by the device;

2. to assess whether ECCO₂R is of benefit to patients compared with NIV;

3. and to assess the impact of ECCO₂R on work of breathing.

The in vitro component tested the device in accordance with regulatory requirements for extracorporeal devices. This required the development of a circuit with several membranes in series and one in parallel. The circuit was arranged to grossly mimic human physiology with blood flow (provided by an extracorporeal membrane oxygenation pump and providing the surrogate for cardiac output), CO₂ production (provided by CO₂ instillation via a dedicated membrane) and ventilation (CO₂ clearance provided by the study device). The study was based on the physiology seen in AECOPD and created the conditions of a hypercapnic respiratory acidosis using expired human packed red cells. CO₂ clearance was calculated using the transmembrane CO₂ content difference and the CO₂ content in the gases from the exhalation port at each change in sweep gas flow rates. This data was then compared with the CO₂ removal reported by the device.

The second phase of the basic investigation of the device and the kinetics of CO₂ clearance was an in vivo study. In this study, subjects commenced on ECCO₂R had serial trans-membrane samples measured as sweep gas flow was increased and transmembrane CO₂ clearance was calculated at each step. The CO₂ reported by the device was also recorded. The two methods were directly compared to ascertain the accuracy of the device. The relationship between sweep gas flow rate and CO₂ clearance in vivo was also explored.

The second component of the thesis was the performance of a prospective, randomised controlled trial of NIV alone compared with NIV and ECCO₂R. Patients were included if they were over 18 years of age, had a history of COPD presenting with an acute exacerbation and with a persisting pH <7.30 due to hypercapnia after initial medical therapy and at least one hour of NIV. Patients were excluded if they had acute multiple organ failure, intolerance, allergy or contraindication to heparin, a contraindication to NIV or were receiving chronic domiciliary NIV.

The study was designed for a primary endpoint of time to cessation of NIV with a power calculation performed based on reduction in NIV use of at least 12 hours. Secondary outcomes included physiological measurements, intensive care and hospital length of stay and outcomes at 90-days including quality of life. Adverse outcomes included incidence of major haemorrhage, thrombosis, haemolysis, mechanical complications, need for invasive mechanical ventilation and mortality. Subjective discomfort and dyspnoea were measured using a visual analogue scale. ECCO₂R was delivered using the Hemolung device. All care not directly relating to ECCO₂R was in accordance with standard clinical protocols.

Three different methods were explored simultaneously in the physiological study which included a subset of patients who consented for these elements. The first was oesophageal pressure measurement using an oesophageal balloon. This was used to both measure oesophageal pressure and calculate muscular pressure and transpulmonary pressure. Work of breathing and the pressure time product of each breath was calculated. The second method was electrical impedance tomography where a small current is injected across the thorax using a dedicated device and a belt placed at the 4th intercostal space anteriorly which measures the changes in impedance with each breath and over time. The impedance change is used to generate changes in tidal impedance change, surface volume, inspiratory time, homogeneity and end expiratory lung impedance. These measurements allow insight into the distribution and timing of aeration both globally and regionally within the lungs. The final method was parasternal electromyography which measures the electrical signal in the intercostal muscles in the upper thorax and is a surrogate for work of breathing. This is achieved through electrodes placed in the second intercostal space anteriorly.

The in vitro study demonstrated that firstly the membrane was capable of clearing CO₂ and secondly that the CO₂ clearance reported by the device accurately reflected the CO₂ clearance using the other methods. Finally, there is a clear and consistent relationship between sweep flow rate and CO₂ clearance. CO₂ clearance increases rapidly as sweep flow increases from 0 to 2 L/min. Thereafter the rate of rise of clearance decreases with a plateau between 4 and 6L/minute giving a maximum ventilation/perfusion ratio of approximately 10-15:1 as the limit of efficiency for this artificial membrane lung. These results were replicated by the in vivo study.

The randomised controlled trial demonstrated a significant reduction the duration of NIV with the addition of ECCO₂R (7 hours compared with 30 hours in the NIV group). Additionally, there was a more rapid reduction in respiratory rate and faster resolution of respiratory acidosis with ECCO₂R than with NIV alone, but there was a small increase in respiratory work after commencement of ECCO₂R and removal of NIV. Intensive care and hospital lengths of stay were both approximately 117 hours longer in the ECCO₂R group than with NIV. Symptomatic dyspnoea resolved rapidly with commencement of ECCO₂R. The ECCO₂R group had a higher incidence of haemolysis however overall, NIV-related complications were more common than ECCO₂R -related complications.

It was demonstrated that oesophageal pressure pressure indices, work of breathing and pressure time product were lower in the NIV group initially and paradoxically increased as the clinical condition improved. Work of breathing and pressure time product were discordant with NIV, especially on day 2 of the study indicating an elevated isometric work. Electrical impedance tomography demonstrated progressive improvements in aeration in the NIV group. The parasternal electromyogram showed an elevated neural respiratory drive in the NIV group which decreased over time. In the ECCO₂R group the oesophageal pressure measurements indicated that work of breathing and pressure time product remained elevated throughout the study but were concordant without evidence of isometric work. The neural respiratory drive remained elevated than ECCO₂R group throughout the study. Electrical impedance tomography demonstrated a more dorsal distribution of ventilation with ECCO₂R, and that aeration was more inhomogeneous than with NIV. The lowest work of breathing and most homogeneous ventilation was found with the combination of NIV and ECCO₂R.

The important clinical implication of the in vitro and in vivo testing is that the relationship between CO₂ clearance and sweep gas flow rate is non-linear. At the onset of therapy with only 1L/minute of sweep gas, a substantial amount of CO₂ amounting to approximately a third of CO₂ production is cleared (assuming that 3mL/kg/min CO₂ is produced). A plateau is reached at around 4L of sweep gas flow giving a ventilation/perfusion ratio of approximately 10:1 as the limit of efficiency for this artificial membrane lung. It is reasonable to assume that the relationship between sweep flow rate and clearance shows no hysteresis, and consequently during device weaning there remains a significant CO₂ clearance at low sweep low rates. Given this, it is important that a period of slow weaning of the last elements of sweep flow gas occurs and that a period of observation following cessation of sweep flow gas should be undertaken to prevent rebound respiratory failure. The study also showed that the CO₂ clearance reported by the device is accurate and can be used instead of recurrent sampling from the circuit.

The randomised, controlled trial demonstrated that there was earlier cessation of NIV by approximately 30 hours. The trial also demonstrated a physiological benefit associated with ECCO₂R with a more rapid improvement in respiratory acidosis and tachypnoea with minimal complications. Additionally, there was a patient symptomatic benefit with rapid improvement in discomfort and dyspnoea after the addition of ECCO₂R. However, this data also indicates that there was a longer ICU and hospital length of stay for patients commenced on ECCO₂R. The study was not powered to demonstrate a mortality benefit or a difference in the need for intubation and these need to be explored in future larger trials. Although many patients did cease NIV shortly after commencing ECCO₂R, given that there was a deterioration in gas exchange after removal of NIV, this suggests that there may be a benefit in the combination of ECCO₂R and NIV.

The data from oesophageal pressure measurements, electrical impedance tomography and parasternal electromyography demonstrate differential effects of NIV and ECCO₂R on work of breathing and the distribution of ventilation. NIV provides mechanical support for breathing with improvements in work of breathing, increases in tidal volume and more homogeneous ventilation but in the first 24-48 hours of severe exacerbations there is a persisting dynamic hyperinflation which results in neuromechanical dissociation and increased isometric work despite the use of NIV. ECCO₂R provides clearance of carbon dioxide from the blood which results in a lower minute ventilation, reducing dynamic hyperinflation which in turn reduces neuromuscular dissociation and isometric work but results in an increase in overall work of breathing. The combination of ECCO₂R and NIV allowed elimination of a proportion of the metabolic CO₂ and a reduction in the requirements of alveolar ventilation and was also associated with a lower work of breathing, better neuromechanical coupling and more homogenously distributed ventilation.

This thesis has demonstrated the nature of gas exchange across a specific ECCO₂R device and explored its efficiency of gas exchange.

The clinical and physiological data supports different but complementary impacts of NIV and ECCO₂R in patients with AECOPD. NIV provided direct respiratory support but during the early phase of the exacerbation patients remained at end expiratory lung volumes close to total lung capacity and were unable to generate significant pleural pressures or provide adequate aeration of the lungs resulting in isometric contraction and neuromechanical dissociation along with respiratory acidosis. Over time as the exacerbation started to resolve, the ongoing physical support with NIV resulted in improved aeration which was homogenously distributed across the lung. As hyperinflation reduced, and higher muscular pressures were able to be generated, acidosis was corrected and sensations of dyspnoea reduced. ECCO₂R removed CO₂ from the venous blood and allowed early reduction in dyspnoea with reduced respiratory rate, lower dynamic hyperinflation and improvement in neuromechanical dissociation and isometric work.

The combination of ECCO₂R and NIV provided particular benefit early in the course of an exacerbation with elimination of a significant proportion of the metabolic CO₂ and a reduction in the requirements of alveolar ventilation and was also associated with a lower work of breathing, lower neural drive and more homogenously distributed ventilation than ECCO₂R alone.