Artificial Gas Exchange

Proceedings of the 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference Shanghai, China, September 1-4, 2005

Artificial Gas Exchange Dongfang Wang1, Joseph B. Zwischenberger1, SD Chambers2 1: Cardiothoracic Surgery, University of Texas medical Branch; 2: MC3 Inc. Ann Arbor, MI Introduction: Artificial gas exchange can achieve total oxygen transfer and/or CO2 removal to allow various degrees of lung rest during severe respiratory failure. I. Extracorporeal membrane oxygenation (ECMO) Extracorporeal membrane oxygenation (ECMO) is the term used to describe prolonged extracorporeal cardiopulmonary bypass achieved by extrathoracic Fig. 1 Extracorporeal membrane oxygenation circuit vascular cannulation. A modified heart-lung machine is used, most often consisting of a distensible venous blood drainage reservoir, a servoregulated roller pump, a membrane lung to exchange oxygen and carbon dioxide (Fig. 1). ECMO was first attempted in the 1960s [1] and in newborns with severe respiratory failure (80% mortality) achieve an overall survival rate of 75% to 95% [2-4]. In ARDS patients, progressive mechanical ventilation with high peak airway pressures will cause further injury to lung parenchyma by Baro/Volume trauma. ECMO perform total gas exchange, avoid further baro-trauma and allow native sick lung recover. But, ECMO is very complicated, labor intensive Fig.4 AVCO2R circuit in smoke and burn induced sheep model

and blood traumatic with daily blood transfusion. II: AVCO2R: ARTERIOVENOUS CARBON DIOXIDE REMOVAL AVCO2R has been developed using a simple arteriovenous shunt with a gas exchange. It is a simplified ECMO by eliminating ECMO-related components and reduce the foreign surface area, and priming fluid and blood transfusion volume(Figure 2). A pumpless AVCO2R functions with lower flow rates, and is expected to produce less hemodynamic or Figure 3. Simulation of apnea and total CO2 removal via AVCO2R

260

251

Steady State PaCO2 (tor r )

240 220 200 180 160

142

140 120

93

100

70

80

58

60

49

44

40 0.05

0.10

0.15

Shunt flow (fr action of CO)

0-7803-8740-6/05/$20.00 ©2005 IEEE.

0.20

hematologic disturbances. Mathematical model and experimental evaluation: We developed a mathematical model to simulate AVCO2R and predict the feasibility and necessary conditions for total CO2 removal[5]. The mathematical model incorporated compartments representing blood, pulmonary alveoli, pulmonary capillaries, peripheral tissues and capillaries, and an extracorporeal gas exchange device and has been validated against an animal model which consisted of anesthetized and mechanically ventilated piglets with an AVCO2R device placed by cannulation of a femoral artery and vein. Dynamic and steady state measurements of CO2 transfer were made and compared with simulations using the mathematical

413

model. There was good agreement between experimental and simulated data, validating the mathematical model under a variety of conditions. The mathematical model was used to determine operating parameters for total CO2 removal. Relationships between extracorporeal blood flow, device diffusing capacity, and device gas sweep flow were established for CO2 removal at various levels of CO2 production. These simulations indicate that it is possible to achieve total CO2 removal using a AV shunt of 10%-15% of cardiac output (Fig 3), a device diffusing capacity of 0.5 ml x min(-1) x torr(-1) (kg body weight)(-1), and a gas:blood flow of 5 or greater. During AVCO2R, carbon dioxide removal and oxygen transfer are uncoupled (carbon dioxide is secreted through the membrane gas exchanger whereas oxygen diffuses through the native lungs with extremely low tidal volumes and respiratory rates called apneic oxygenation). This method can be used to supplement low tidal volume mechanical ventilation for ARDS patients, allowing reduced peak airway pressure and therefore decreasing hemodynamic compromise and barotrauma while promoting lung rest and healing. A 10F arterial 12F venous percutaneous cannula will allow arteriovenous shunt flow greater than 500 mL/min and provide lung rest with permissive hypercapnia. Our data confirmed, despite a 20% to 26% cardiac shunt, that AVCO2R can be used for total carbon dioxide removal for up to 7 days, without hemodynamic compromise or instability, in an adult sheep model of severe respiratory failure[6]. Investigating the performance characteristics of a low-resistance membrane gas exchanger, reported that normal PCO2 could be maintained with minimal ventilator support at blood flows of 500 mL/min or higher[7]. In our ovine model of smoke inhalation injury, AVCO2R achieved total carbon dioxide removal and significantly reduced minute ventilation and peak inspiratory pressure (PIP) while maintaining normocapnia. In our new, clinically relevant, large animal model of ARDS secondary to smoke inhalation and cutaneous flame burn injury, percutaneous AVCO2R achieved near-total carbon dioxide removal and allowed significant reductions in minute ventilation, tidal volume, and PEEP[8,9]. AVCO2R flows of 800 to 900 mL/min (11% to 13% cardiac output) achieved 77 to 104 mL/min of carbon dioxide removal (95% to 97% total carbon dioxide production) while maintaining normocapnia. Significant reductions in ventilator settings were tidal volume from 450 to 270 mL; PIP from 25 to 14 cm H2O; minute ventilation from 13 to 6/L min; respiratory rate from 25 to 16 breaths/min; and FiO2, from 0.86 to 0.34[9]. Our initial patients required percutaneous 12F arterial and 16F venous cannulus to provide adequate flow. They were all safely cannulated and connected to the AVCO2R device and received good support. No complications have been noted[5].We are currently conducting prospective, randomized, controlled, unblinded, multicenter outcomes studies to compare the effect of percutaneous extracorporeal arteriovenous carbon dioxide removal (AVCO2R) to standardized pressure controlled mechanical ventilation (SMV) on all cause mortality and ventilator free days in children with acute, severe respiratory failure secondary to burn injury with or without severe smoke inhalation and in adults, with ARDS. The concept of percutaneous arteriovenous extracorporeaI gas exchange continues to expand and gain acceptance [7-10]. The group at the University of Regensburg has recently completed an experience of "Percutaneous Extracorporeal Lung Assist (PECLA)" in 75 patients with severe ARDS who were perceived to have a predicted 100% mortality or exceeded adult ECMO criteria. PECLA achieved near total removal Of CO2 production with 30% survival! Cardiac output improved. And they required only mild anticoagulation (ACTs 140-160 sec) [11-12]. Ambulatory AVCO2R: We are developing an ultracompact gas exchanger to allow patient ambulation during AVCO2R. Chronic carbon dioxide retention syndromes such as COPD and emphysema may increase work of breathing from 2% to 50%. We propose that extracorporeal CO2 removal will decrease the work of breathing to decrease muscle wasting and allow rehabilitation or support during acute exacerbations. Ambulatory AVCO2R gas exchanger (MC3, Inc.) has a total volume of 340 ml and a fiber surface of 1.3 m2. The hollow fibers are OxyPlus PMP 90/200 (Membrana GmbH) with an outer skin to mitigate plasma leakage. The device CO2 removal was measured at a constant flow rate of 1 L/min, varying sweep gas from 1 to 15 L/min, to determine the optimal sweep gas:blood flow ratio. Gas exchanger resistance remained stable at 2.3 +/-0.53 mmHg/min. CO2 removal with 1 l/min blood flow and 5 L/min sweep gas was 110 +/- 12 then stabilized at 85+/-14 ml/min to 6 hours. We conclude AVCO2R is capable of providing near total extracorporeal removal of carbon dioxide production during respiratory failure, while maintaining CO2 and pH homeostasis. III: Paracorporeal Artificial Lung (PAL) A paracorporeal artificial lung (PAL) will allow long-term total respiratory support and may allow lung recovery or bridge to lung transplant. We have described the challenges and design requirements (paracorporeal), provide total oxygen and CO2.[13] Paracorporeal placement allows easy access for blood sampling and device change out, gas flow venting to atmosphere, and ambulation. Two basic configurations of

414

artificial lung implants are in series (pulmonary artery-to-pulmonary artery [PA-PA]) or in parallel (pulmonary artery-to-left atrium [PA-LA]) with the pulmonary circulation, Fig.3 Paracorporeal artificial lung prototype without splitting blood flow between the distal pulmonary artery (top) and with (bottom) the inflow passive balloon compliance chamber. (native lung) and the artificial lung. In collaboration with MC3 (Ann Arbor, MI), our group has been developing a low resistance, low impedance PAL prototype designed for weeks to months of respiratory support (Figure 3).[14-16] To date, we have implanted variations of the MC3/UTMB PAL in 34 sheep with the PAL implanted in-series with the pulmonary artery circulation. Blood flows from the right ventricle, into a cannula anastomosed to the proximal pulmonary artery (snared to divert all cardiac output into the PAL), into a compliance chamber, and then into the PAL housing where blood is oxygenated by radial flow over microporous fibers. Blood exits through an outflow cannula anastomosed to the distal pulmonary artery, then through the native lungs. Prior to the first large animal series, the PAL underwent surface modifications (port removal, collet-nut connection improved housing) to meet the physical demands of being attached to an awake, standing sheep for up to 7 days.[14] Unfortunately, right heart failure was 50% due to the added resistance and impedance of the PAL. To reduce right heart failure, three modifications decreased PAL impedance and resistance. First, an inflow passive balloon, made to receive the full right ventricular stroke volume, was added to reduce right ventricular outflow impedance (Figure 3, bottom). Second, an inlet blood flow separator was added to convert the inflowing blood from an axial direction, to radial flow to decrease resistance and improve blood/gas interface. Lastly, the PAL outlet diameter was increased to decrease resistance. These three major internal modifications markedly improved cardiac function. In our next series, the modified PAL was implanted in 7 healthy sheep for up to 72 hours.15 Cardiac output improved from 2.8 to 4.2 liters/min with average CVP 6-7 mmHg. Device resistance decreased from 2.5 to 0.79 Wood units (less than the native lung resistance of ~1 Wood unit!).[15] Finally, we used our LD100 smoke/burn ARDS sheep model to compare PAL (n=8) to volume-controlled mechanical ventilation (VCMV; n=6) in a prospective, randomized, controlled, unblinded outcomes study.4 Six of 8 PAL sheep and 1/6 VCMV sheep survived the 5-day study period. Ventilator settings in the PAL sheep 48 hours after lung injury were significantly lower compared to the VCMV sheep. Similarly, the PaO2/FiO2 ratio was normalized in the PAL sheep, but still met ARDS criteria in VCMV sheep. At autopsy, the lung wet-to-dry ratio was significantly improved in the PAL group. There were no PAL device failures or evidence of right heart failure. The PAL provided total gas exchange, significantly decreased ventilator support and decreased ventilator induced lung injury in a LD100 ARDS model, improving 5-day survival compared to VCMV. Optional Active Compliance Chamber (OACC) and Mathmetical Model: Although PAL design modifications lowered impedance and resistance, implantation in patients with end-stage lung disease and pulmonary hypertension may be problematic. Utilizing a PA-LA configuration reduces resistance as some right ventricular output flows through the native lungs and some flows through the PAL in parallel[17] . However, the PA-LA configuration provides only partial gas exchange and has several disadvantages (risks of systemic embolism, the PAL cannot be removed from the circuit due to massive right to left shunting). Therefore, we favor the addition of an optional active compliance chamber in a PA-PA configuration. Our optional active compliance chamber (OACC) is completely enclosed, allowing optional counterpulsion (low pressure in systole and high pressure in diastole against a closed pulmonic valve) augmenting blood flow through the PAL and native lung. The OACC is a polyurethane compliance balloon enclosed in a rigid ABS housing connected to an intraaortic balloon pump (IABP) console. The IABP console pneumatically drives the OACC (timed to early diastole by the EKG) to push blood out of the compliance balloon, through the PAL, into the pulmonary circulation. Immediately prior to systole, the IAPB evacuates the housing to expand the compliance balloon and assist RV ejection. A valved conduit added to the PAL outflow prevents blood backflow from the native lung before right ventricular contraction. Eleven sheep (30-50 kg) were divided into a Non-valve group (6) and Valve (5) group. A left lateral thoracotomy exposed the PA trunk. Two 18 mm grafts were anastomosed end-to-side to the PA. The OACC-PAL was connected and the PA was

415

snared diverting total cardiac output (CO) through the OACC-PAL and native lungs. A C-clamp placed distal to the OACC-PAL until a 40% decrease in CO was achieved. The OACC was activated, and RVP, PAP, ABP and CO were recorded by data acquisition system (National Instruments, Austin, TX).With activation of the OACC, CO was increased 26% in Non-valve group (p