In Vivo Testing of an Ambient Air Based, Portable, and Automated CO₂ Removal Controller for Artificial Lungs

Abstract

Portable artificial lung systems are under development, but there are few technologies available that adjust the carbon dioxide (CO₂) removal in response to changes in patient metabolic needs. Our work describes the second generation of a CO₂ based portable servoregulation system that automatically adjusts CO₂ removal in artificial lungs (ALs).

Four adult sheep (68±14.3 kg) were used to test the servoregulator. The servoregulator controlled air sweep flow through the lung to meet a target exhaust gas CO₂ (tEGCO₂) level in normocapnic and hypercapnic (PaCO₂ >60mmHg) conditions at varying flow rates (0.5–1.5 L/min) & at tEGCO₂ levels of 10, 20 & 40 mmHg.

In hypercapnic sheep, average post-AL blood pCO₂ values were 22.4±3.6 mmHg for tEGCO₂ of 10 mmHg, 28.0±4.1 mmHg for tEGCO₂ of 20 mmHg and 40.6±4.8 mmHg for tEGCO₂ of 40 mmHg. The controller successfully and automatically adjusted the sweep gas flow to rapidly (<10 min) meet the target tEGCO₂ level when challenged with changes in inlet blood flow or target EGCO₂ levels for all animals.

These in vivo data demonstrate an important step toward portable artificial lungs that can automatically modulate CO₂ removal and allow for substantial changes in patient activity or disease status in ambulatory applications.

Introduction

Both end-stage lung disease (ESLD) and acute respiratory distress syndrome (ARDS) have a profound impact on the healthcare system, each affecting more than 100,000 Americans per year.^1–3 Artificial lung (AL) and extracorporeal membrane oxygenation (ECMO) based CO₂ removal promptly relieves the symptoms of ESLD and can serve as a bridge to recovery by enabling the lung to heal in acute cases and as a bridge to transplant in chronic cases.^4–7

Traditionally, femoral cannulation for ECMO immobilizes patients, but with the advent of dual lumen internal jugular vein cannulation for veno-venous (VV) ECMO, ambulation is possible.⁸ The central sport model can be used via innominate or subclavian artery cannulation. This provides the ability for physical therapy and ambulation while patients await lung transplantation.⁹ Ambulatory ECMO systems with the AL, blood pump, controller, battery, and oxygen cylinder on a wheeled pole enable limited mobility and exercise.^10–12 Because ambulatory ECMO provides a more active physiotherapy for patients, this has led to decreased length of stay in post-transplant ICU, leading to lower costs for the hospital and patient.¹³ Additionally, ambulatory ECMO has been shown to have improved outcomes in critically ill patients.¹⁴ Despite these promising results, many patients are still restricted to ICU care due to several logistical challenges that arise from developing an ambulatory ECMO program.¹⁵

A CO₂ removal system that automatically regulates CO₂ to meet dynamic physiological needs can simultaneously minimize logistical issues of an ambulatory ECMO program, while improving patient stability. Traditional ECMO systems risk removing insufficient CO₂ during periods of activity (resulting in dyspnea and increased minute ventilation) or excessive CO₂ during periods of inactivity (resulting in hypocapnia and blood alkalosis).^16,17 Without CO₂ regulation, patient comfort, activity, and quality of life will thus be severely limited when used for long-term ambulatory ECMO.¹⁸ The Quantum Ventilation Lite is a recently developed semi portable ECMO control system that can make automatic ventilatory changes but has not yet been tested on patients in the current literature.¹⁹

Our lab previously developed a subclavian arterio-venous (AV) access method for an ovine model. The published data show that varying sweep flow can drastically impact CO₂ removal and sweep flows up to 15 L/min can significantly reduce PaCO₂ from 74 to 49 mmHg using a blood flow of 1.0 L/min.²⁰ Based on this established principle, we developed and performed in vitro and in vivo testing of the first-generation benchtop servoregulation system (Gen 1) that automatically adjusts CO₂ removal in ALs under various metabolic conditions. It does this by modulating sweep gas flow rate based on exhaust gas CO₂ partial pressure (EGCO₂) as a representation for arterial partial pressure of arterial CO₂ (PaCO₂).²¹ The Gen 1 system rapidly and effectively adjusted to maintain the target EGCO₂, which was reflected in the PaCO₂.

Though functional and effective, the previous system lacked a unitary enclosure, was wall-powered, utilized various large components, used desiccant dehumidifiers which required frequent replacements, and incorporated a loud volumetric pump. These characteristics preclude its use in portable applications. However, it successfully proved the sweep gas volume and EGCO₂ servoregulation method which the portable “Gen 2” model described in this paper uses. We have recently developed a portable Gen 2 version of this control system that weighs less than 2 kg and can operate for up to 12 hours on a single battery charge.²² This manuscript describes the use of this portable Gen 2 model in vivo with 4 sheep to assess its ability to successfully modulate EGCO₂ in vivo under hypercapnic conditions.

Methods

Servoregulation Principle

A diagram of the CO₂ based sweep gas servoregulation controller is shown in Figure 1. Using a Proportional-Integral-Derivative (PID) feedback controller, the system modulates pump-driven sweep air flow through an AL in order to attain a desired EGCO₂. The PID feedback controller operates in a continuous cycle, reading the current value of EGCO₂ from the CO₂ sensor and comparing it to the user-specified target EGCO₂ value (tEGCO₂). If the tEGCO₂ is less than the current EGCO₂, power to the air blower is increased, which increases sweep gas flow through the AL. If the tEGCO₂ is greater than the current EGCO₂, the controller decreases power to the air pump, thus reducing the sweep gas flow through the AL. The proportional, integrative, and differential parameters determine exact pump power based on target and measured EGCO₂ values. Properly tuned PID controllers are capable of responding rapidly to changes in equilibrium and can do so with minimal oscillation.

Figure 1. Flow path diagram of servoregulator system and components.

Blue arrow is exhaust and gray arrow is sampled gas to ambient air.

Flow Path and System Design

The components of the servoregulator system were selected to minimize size, power consumption, and noise with a view to facilitating integration into future ambulatory and wearable AL systems (Figure 1).

A WM7040–24V blower (Haoson, China) takes in ambient air, providing up to 20 L/min against the resistance of the AL and circuit. The blowers individually measure 70×70×40 mm in size and offer a quiet operation (~45 dBA). Only one blower is used at a time, with check valves to ensure unidirectional air flow. The blower-driven air passes through a flow sensor (Sensirion, Germany SFM3400-AW) before being routed into the AL. A differential pressure sensor (Honeywell, Charlotte, NC ABP2) is connected across the AL to monitor gas-side pressure. The exhaust gases leaving the AL are routed through a water trap and then expire into the atmosphere.

Side-stream capnography was used to measure EGCO₂. The sampling circuit includes a TCS Micropump, (England, D250BLZ-V) to sample the exhaust gases. To remove moisture, the sampled gases are passed through a moisture trap (CO₂ Meter Inc., Ormand beach, FL CM-0103), and 24 inches of tubing (Nafion, Lakewood, NJ, ME-110–24BB). The dried air then enters the CO₂ sensor assembly (heated to increase dew point) containing a CO₂ sensor (SprintIR-W20%, Gas Sensing Solutions Ltd., Scotland) before being exhausted into the ambient atmosphere.

Hollow fiber ALs typically accumulate moisture at the outlet of fiber bundles when moisture from plasma diffuses across the microporous membrane. Some of that moisture condenses and if enough condensation forms it can affect the CO₂ clearance ability of the lung over time.²³ To address this issue, the Gen 2 servoregulation system continuously monitors the fluidic resistance of the gas side of the AL. When the resistance has significantly increased, indicating moisture accumulation on the gas side of the AL, the system automatically generates a high flow pulse of sweep gas to flush out any accumulated moisture. Prior to the described animal study, the Gen 2 servoregulation system was tested in vitro with blood and water to verify operation, functionality, and reliability.²²

Animal Preparation

In vivo experiments were conducted on 4 adult sheep (68±14.3 kg; Figure 2). All sheep were fasted the night prior to surgery. All experiments were conducted in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals. This study was approved by the University of Michigan Institutional Animal Care and Use Committee (IACUC). All sheep were sedated with propofol 6 mg/kg through a hoof vein catheter prior to endotracheal tube intubation, followed with general inhalational anesthesia with isoflurane (1–3%) and ventilated under volume-controlled ventilation. The sheep were paralyzed with vecuronium 0.01–0.04 mg/kg/hr as minute ventilation was modified during the study. The sheep were kept under anesthesia for 10–12 hours per procedure for completion of the study. The sheep were cannulated directly into the subclavian artery and vein to create an arterio-venous (AV) shunt to attach the Novalung iLA oxygenator (Novalung iLA Membrane Ventilator, Xenios AG, Heilbronn Germany, 0.5–7.0 L/min). Once it was attached to the lung, the servoregulator controller was connected to the gas side of the artificial lung and mounted on a cart adjacent to the animal. An arterial line inserted in the femoral artery and a Swan-Ganz catheter in the jugular vein allowed for hemodynamic monitoring, arterial blood sampling, and ease of fluid administration intraoperatively. Heparin was titrated to achieve an activated clotting time (ACT) of 275–375 s (Hemochron, Netherlands).

Figure 2. In vivo servoregulator testing.

Schematic diagram of circuit set up with cannula placement on animal. Gas circuit is marked in light blue and blood circuit marked in red.

Ventilation and Servoregulation

The servoregulator was initially tested in these sheep at 0.5 and 1.0 L/min AL blood flow rates with tEGCO₂ levels of 10, 20, & 40 mmHg while ventilated physiologically (7 mL/kg) and at a rate of 12 breaths/minute to maintain normocapnia (goal PaCO₂ of 35–45 mmHg). The minute ventilation was then reduced to 30–40% of baseline to induce hypercapnia (goal PaCO₂ 60–90 mmHg) while maintaining appropriate oxygenation. The blood flow rate through the AL circuit was set at 0.5, 1.0 & 1.5 L/min, which was controlled using an adjustable Hoffman clamp. The servoregulation system was subsequently tested for tEGCO₂ levels of 10, 20, & 40 mmHg at these blood flow rates and the response of the PID controller was measured.

Figure 3 depicts how the hypercapnic conditions were created and blood gases were measured throughout the animal experiments. For each test setting (“therapy” in Figure 3), the servoregulator remained on until PaCO₂ stabilized (~20 min), then steady-state measurements were taken for that setting. Then the servoregulator was turned off and the animal was permitted to return to hypercapnia (“washout” in Figure 3) before the next condition was tested.

Figure 3. Schematic for hypercapnic conditions.

X axis is time duration of prep, blue areas indicate when therapy was administered, with time duration above. Note 60 minutes of baseline conditions before initiation of therapy.

Data Collection

Animal hemodynamics were recorded throughout the duration of the prep including heart rate (HR), central venous pressure (CVP), cardiac output (CO), and temperature. Every 15 minutes, tidal volume, respiratory rate, and minute ventilation were obtained as well as arterial blood gases (pH, partial pressure of CO₂ and O₂, and electrolytes). Artificial lung data included: pre (inlet) and post (outlet) blood gases, exhaust CO₂, blood flow, and pre(inlet) and post (outlet) blood pressure. ACT was obtained hourly with our ACT goal as stated above.

Results

All 4 surgeries and post-surgical evaluations occurred without any major incidents. No studies were required to terminate early due to lack of animal stability. In all studies, the servoregulator performed appropriately without any significant issues.

When the controller was challenged with modifications to inlet blood flow or tEGCO₂ levels, it was able to adjust the sweep gas flow in all animals to rapidly (<10 min) achieve the specified EGCO₂ level. Figure 4 shows operation of the servoregulator system during three washout/therapy sessions in a single hypercapnic sheep for tEGCO₂ values of 10 (left), 20 (middle), and 40 (right) mmHg and an AL blood flow of 1 L/min. Data for all four hypercapnic animals were combined in Figures 5 and 6. Increasing the tEGCO₂ resulted in a decrease in sweep gas flow rate, CO₂ removal, and electrical power consumption of the system (Figure 5). The CO₂ removal rate was 46.89±9.6 versus 57.77±8.4 mL/min at .5L/min with a tEGCO₂ of 20 mmHg and 58.3±3.52 versus 106.7±11.37 mL/min at1.0 L/min with a tEGCO2 of 20 mmHg.

Figure 4. Servoregulator function in single sheep.

This figure represents servoregulator system function during three washout/therapy sessions in a single hypercapnic sheep for tEGCO₂ values of 10 (left), 20 (middle), and 40 (right) mmHg and an AL blood flow of 1 L/min. X axis = Time, blue line = exhaust CO₂, dotted line = target exhaust CO₂, Red line = pump flow rate, red dot = blood CO₂ pre oxygenator, green star = blood CO₂ post oxygenator.

Figure 5. Combined results n=4 in hypercapnic conditions.

X axis = target exhaust CO₂, Y axis = Outlet blood PCO₂, CO₂ removal rate, Sweep flow rate, Power consumption (clockwise). Each color represents a different blood flow rate (.5, 1, 1.5 L/min).

Figure 6. pH and CO₂ level at blood flow .5, 1 and 1.5 L/min.

X axis depicts target exhaust CO₂, Left Y axis represents blood CO₂ and Right Y axis represents blood pH levels. Blue and red are baseline and steady state respectively.

Increasing blood flow through the AL increased sweep gas flow rate, CO₂ removal, and electrical power consumption.

The servoregulator further decreased animal arterial blood PaCO₂ and increased pH in hypercapnic animals under all test conditions (Figure 6). The average pre-device pH was 7.14±.007 while post-device was 7.43±0.08 in our hypercapnic models. This demonstrates normalization of post-device pH correlating with adequate CO₂ removal.

In normocapnic sheep, CO₂ removal was lower than in the hypercapnic animals. At 1.0 L/min and tEGCO₂ of 20 mmHg, blood PaCO₂ decreased from 106.7±11.4 mmHg to 58.3±3.52 mmHg (p<0.01). This trend was consistent at 0.5 L/min with a target exhaust gas of 20 mmHg, where blood PaCO₂ decreased from 57.8 ±8.4 to 46.9±4.7 (p=0.14) mmHg in normocapnic sheep.

Discussion

The advancement in this servoregulator system provides a crucial step in the advancement of an ambulatory artificial lung that can make real-time adjustments to changes in blood CO₂ levels (patient metabolic demands). As in the previous generation of this system, this servoregulator was able to effectively adjust CO₂ removal rapidly to meet the tEGCO₂ value. In a normocapnic animal, the servoregulator functioned appropriately and did not induce a state of hypocapnia for tEGCO₂ values of 20 and 40 mmHg. In addition, in hypercapnic animals, the servoregulator was able to remove higher levels of CO₂ successfully. Both the arterial pCO₂ as well as the pH levels reflected this function.

Membrane lungs are very efficient at removing CO₂ which can lead to hypocapnia. Not only does this device aid in regulating hypercapnia but it can also prevent hypocapnic events using servoregulation. Our primary target population consists of COPD patients using this therapy as a bridge to transplantation. The ECCOR study discussed (20) was specifically for use in acutely decompensated COPD patients whereas we are looking to use this device in chronic patients. In a systematic review and meta-analysis in 2022 of the ECCOR device, there was no major difference in the prognosis of patients with and without this treatment, but it did reduce PaCO2, increased pH and improved the PaO2/FiO2 ratio in both ARDS and COPD patients.²⁴ This is important in caring for chronic patients as a bridge to transplantation. We do not foresee our device changing prognosis but improving quality of life, reducing metabolic changes seen by hypo and hypercapnia and increasing ease of breathing. Additionally, some of the complications seen with ECCOR were related to bleeding due to vascular access issues. We plan to place these devices in a more controlled setting and use a vascular graft which could potentially reduce the risk of access complications. Work by Combes et al. describes that when ECCOR was initially studied, there was no statistically significant difference in mortality, but an increased number of patients died which led to the decreased usage of the device.²⁵ Since then, there have been changes to these devices and they are more frequently used to reduce the potential damage from mechanical ventilation and can facilitate lower tidal volumes.^26,27 The SUPERNOVA pilot trial also demonstrated feasibility and safety of this device in reducing tidal volume without serious adverse events.²⁸ ECCOR has also been studied in end stage cystic fibrosis patients with 1 year survival at 80%.²⁹

This model of servoregulator is improved from its previous model in several different parameters. It is a more compact system with an integrated battery pack allowing for improved portability. Additionally, this model has an automated sweep gas flushing mechanism to reduce condensation which can decrease CO₂ clearance efficiency.

We envision that this servoregulator will function in the clinical setting to allow for moment-to-moment adjustments in CO₂ removal based off the patient condition and current clinical status. If a patient is active, naturally their CO₂ will rise and require an increase in the CO₂ removal from the servoregulator. This system has demonstrated its ability to function in that application by increasing sweep gas and CO₂ clearance to sustain the appropriate target exhaust gas CO₂ level.

Conclusion

These in vivo data demonstrate an important step toward the development of ambulatory and wearable artificial lungs that can effectively modulate CO₂ removal and thereby allow for significant fluctuations in patient activity or disease status. In future work, we plan to further miniaturize the system so that it is wearable and aim to perform longer term animal studies. Additionally, to achieve approval for clinical use, we will use standard clinical device protocol for the hardware and software development of this device.