Recent Advances and Future Directions in Extracorporeal Carbon Dioxide Removal

1. Introduction

Extracorporeal carbon dioxide removal (ECCO₂R) is a technique designed to remove carbon dioxide from the venous blood, thereby reducing the intensity of mechanical ventilation. Unlike high-flow extracorporeal membrane oxygenation (ECMO), ECCO₂R does not provide oxygenation, as it operates at lower blood flows ranging from 0.4 up to 1.5 L/min, depending on the device. The lower blood flow reduces the procedure's invasiveness, allowing smaller cannulae and simpler extracorporeal devices. To refine this lung-support technology, various ECCO₂R configurations and techniques have been explored in benchtop and animal models.

There is a specific interest in using ECCO₂R to improve the management of acute respiratory distress syndrome (ARDS) and chronic obstructive pulmonary disease (COPD). ECCO₂R shows potential for other applications, including serving as a bridge to lung transplantation and enabling long-term CO₂ management in outpatient settings. ECCO2R faces barriers such as hemolysis, thrombosis and bleeding, which limit its clinical adoption. This review highlights recent advancements in ECCO₂R technology, including device configurations, CO₂ removal efficiency, and optimised anticoagulation strategies. We explore the physiological impacts of ECCO₂R systems studied through animal models and mock circulation setups to evaluate safety and performance. We also explore key research priorities, including improving gas exchange membranes, refining blood flow and anticoagulation protocols to reduce complications, and investigating innovative approaches such as membrane acidification and hybrid devices.

2. Clinical Evidence for ECCO₂R

ECCO₂R has been investigated for its role in managing acute respiratory distress syndrome (ARDS) and acute hypercapnic respiratory failure, particularly acute exacerbations of chronic obstructive pulmonary disease (AECOPD). The earliest model of ECCO₂R utilised a pumpless arterio-venous (AV) approach, diverting a portion of the patient’s cardiac output through the device [1] via a femoral arterio-venous circuit. Several case reports and case series documented the use of pumpless AV ECCO₂R [2,3,4,5,6,7,8,9,10,11,12] and demonstrated their ability to decrease arterial CO₂ and resolve respiratory acidosis. However, complications associated with the use of AV ECCO₂R were considerable, including bleeding (18-47%), thrombosis (0-20%) and limb ischaemia (4.5-22%)[2,4].

Pumped ECCO₂R systems were later developed, enabling a veno-venous (VV) approach, typically using a double-lumen cannula placed in the femoral or jugular vein[13]. Animal studies consistently show that VV ECCO₂R can effectively control respiratory acidosis [14,15,16,17,18]. Similarly, uncontrolled clinical studies demonstrated similar results in patients with ARDS and AECOPD[4,16,19,20,21,22,23,24]. However, a retrospective propensity-matched case-control study using VV ECCO₂R in COPD found no difference in outcome with ECCO₂R use [25]. Recently, VV ECCO₂R has been safely integrated with renal replacement therapy platforms in patients with acute respiratory distress syndrome [26,27] and AECOPD [28]. Case series have also reported reduced intubation rate in AECOPD and decreased respiratory rate when used alone or with non-invasive mechanical ventilation[29,30,31].

In ARDS management, the delivery of lung protective ventilation - reducing driving pressure, optimising PEEP, limiting plateau pressure and tidal volume – is hindered by significant respiratory acidosis and acute cor pulmonale. Trials, such as the SUPERNOVA study, investigated using ECCO₂R to achieve ultra-low tidal volumes of 3-4 mL/kg in mechanically ventilated ARDS patients, demonstrating feasibility and safety in controlling arterial CO2 [26]. The UK-based REST trial, the first randomised controlled trial in this area, compared standard tidal volume ventilation (6-8 mL/kg) to ultra-low tidal volume ventilation (3-4 mL/kg) using ECCO₂R. While the trial achieved significant separation between groups in tidal volume and maintained comparable PaCO2 and pH, it showed no difference in 90-day mortality. The ECCO₂R group experienced significantly higher complication rates, including bleeding with intracranial haemorrhage, leading to early trial termination [32].

For AECOPD, non-invasive ventilation (NIV) remains the gold standard. However, ECCO₂R has been explored as an option for patients who fail or cannot tolerate NIV. Trials have shown faster resolution of respiratory acidosis and work of breathing with ECCO₂R, though they were not powered to assess mortality. Other studies have evaluated ECCO₂R for preventing intubation or enabling early extubation, finding no differences in ventilator-free days or mortality, with higher mortality in the NIV group when ECCO₂R was added[33].

Despite mixed results, interest in ECCO₂R persists, particularly regarding the optimal blood flow rate for clinical benefit. Evidence suggests that higher-flow devices (500–1,000 mL/min) provide superior CO₂ clearance compared to lower-flow systems (<500 mL/min)[34,35]. All the RCTs to date have used devices with a maximum blood flow below 500mL/minute. Future trials should investigate the potential of higher blood flow ECCO₂R devices to improve clinical outcomes.

3. Key Areas of Future Research

The future of ECCO₂R remains challenging due to the clinical outcomes observed so far and the significant complications associated with current devices. Priority should be given to addressing technical issues such as haemolysis, thrombosis, and bleeding. Research must also define optimal parameters for blood flow, sweep gas efficiency, and monitoring while developing methods to compare different devices for practical application. Experimental studies using appropriate circuits and biological models are essential to translate findings into clinical practice. Ultimately, well-designed clinical trials addressing patient-centred outcomes will be crucial in determining this technology's role in critically ill patients.

3.1 Hemolysis

Hemolysis is a significant complication of extracorporeal support and is independently associated with mortality [36,36,37]. A key contributor to hemolysis is shear stress caused by blood circulating through artificial circuits. This shear rate and stress arise from artificial surfaces and continuous flow patterns [38]. Factors such as catheter size and type, flow rates, pump design (roller or centrifugal), lung membrane characteristics, anticoagulation methods (coating and systemic), and ECCO₂R settings all influence haemolysis rates and system performance.

Haemolysis is best detected by measuring plasma free Hb (PFHb), the most reliable marker of RBC injury and breakdown [39,40]. Evidence suggests that haemolysis increases non-linearly with reducing blood flows, as prolonged exposure to artificial surfaces amplifies the effective stress on blood components [41,34].

When the haemoglobin scavenging capacity of haptoglobin is exceeded, free haemoglobin dimers circulate in the bloodstream, releasing free heme [42]. This free haemoglobin depletes nitric oxide (NO) by converting it into nitrate, impairing NO’s regulatory role in vascular smooth muscle tone. The resulting NO depletion leads to potent vasoconstriction and increased systemic and pulmonary vascular resistance [43,44,45]. Additionally, NO depletion disrupts platelet and endothelial function, enhancing platelet aggregation and thrombus formation mediated by the von Willebrand factor [46,47,48].

Free haemoglobin initiates a cycle of inflammation that culminates in a procoagulant state and thrombus formation. This process exacerbates haemolysis, depletes clotting factors, and leads to thrombocytopenia and impaired platelet function [49], which may result in an acquired von Willebrand syndrome [50,51].

To address this frequent complication, we require improved devices and a greater clinical understanding of how device usage impacts haemolysis.

3.2 Thrombosis and Bleeding

Anticoagulation is a critical concern in extracorporeal circulation technology, given that anticoagulation and the impact of shear stress on coagulation proteins may lead to increased bleeding risk [21,32,52,53,54,55,56]. On the other hand, thrombosis is responsible for locally impaired flow conditions in cannulas, pumps and membrane lungs, which lead to mechanical stress and hemolysis. The coagulation cascade is activated when whole blood comes into contact with an artificial surface, exacerbated by slow blood flow. Relative slow blood flow occurs in low-shear circuit parts, such as tubing connectors, with increased thrombin generation[57,58] and minimal clot initiators required to trigger clot formation [59]. The artificial membrane oxygenator and the centrifugal head pump are the two main circuit components affected by higher mechanical stress and risk of thrombus formation. Centrifugal head pump thrombosis is usually associated with acute and severe hemolysis [37]. Membrane thrombosis reduces gas exchange impacting device efficiency.

Figure 1. Example of different blood washouts depending on high and low blood flows in membrane oxygenator with impact on anticoagulation requirements to prevent membrane clotting. Pictures kindly provided by F. Hesselmann and R. Borchardt, Aachen, Germany.

Figure 1. Example of different blood washouts depending on high and low blood flows in membrane oxygenator with impact on anticoagulation requirements to prevent membrane clotting. Pictures kindly provided by F. Hesselmann and R. Borchardt, Aachen, Germany.

Contrary to expectations, low blood flow rates (1–1.5 L/min) are associated with a significantly higher risk of haemolysis (3.2 to 6.6 times greater Haemolysis Index) compared to higher blood flow (4 L/min). In low-flow systems (<1.5 L/min), citrate anticoagulation has been shown to outperform heparin by causing less haemolysis and platelet loss [60]. Citrate better preserves red blood cell density, membrane stability, and deformability, reducing haemolysis over three days of storage [61]. In animal studies, Cardenas et al. demonstrated no thrombus formation and hemolysis at different low flow levels (500ml/min, 800ml/min, 1000ml/min) using citrate anticoagulation in 24 hours. Although higher blood flows may reduce calcium chelation efficiency due to elevated ionised calcium levels, this did not correlate with thrombus formation. When comparing citrate to heparin anticoagulation at electronic microscopy, analysis of membrane oxygenator fibres revealed cellular and fibrin adhesion on heparin anticoagulation even after 6 hours of anticoagulation [62]. Despite these theoretical advantages, heparin remains the most commonly used anticoagulant in clinical ECCO₂R due to concerns about citrate toxicity.

A significant complication of citrate anticoagulation in animal studies was progressive hypocalcaemia, even with calcium supplementation, due to the absence of a hemofilter to remove excess citrate. This increased systemic citrate load, heightening the risk of citrate toxicity [62,63,64]. Further research is needed to enhance the safety profile of citrate anticoagulation for clinical ECCO₂R.

An emerging alternative to sodium citrate is citric acid anticoagulation, which provides a triple effect: calcium chelation, platelet inhibition, and a regional anticoagulant effect from an acidic environment [70]. Citric acid has also enhanced CO₂ removal by blood acidification at the artificial membrane [65]. However, hepatic clearance of citrate in humans is limited [66], which reduces the maximum blood flow of ECCO₂R to 150ml/min, which can be offset by an increase in CO₂ removal efficiency driven by the acidification of the blood [67,68,69,70].

Other alternatives to heparin, such as nafamostat mesylate, bivalirudin, and argatroban, are being considered, especially for patients at high risk of bleeding or heparin-induced thrombocytopenia (HITT). These alternatives offer advantages regarding bleeding risks and drug monitoring but are limited by high costs and short half-lives.

4. CO₂ Removal Rate Performance

4.1 Cannula

ECCO₂R typically employs smaller cannulas (13–14 Fr) than the larger cannulas used in higher-flow ECMO. Venous access is generally achieved via the right internal jugular or femoral veins, with vessel puncture performed under ultrasound guidance. While the smaller cannulas offer advantages such as reduced invasiveness, their use presents several challenges that can impact both efficacy and patient safety. A key issue is maintaining adequate blood flow through the smaller cannulas. A smaller diameter cannula with higher flow rates increases shear stress, which may reduce CO₂ removal efficiency due to insufficient flow, ultimately limiting the therapy’s effectiveness. Multiple single-lumen catheters can replace double-lumen catheters. This involves using separate access points in the jugular and femoral veins to achieve uninterrupted blood flow of approximately 450 mL/min with minimal recirculation [71]. Another relevant factor for ECCO₂R efficiency is the minimisation of blood recirculation related to catheters. Catheter recirculation was related to catheter type and brand, site of placement, catheter length, time on dialysis, and time on the current catheter and was measured via ultrasound dilution technique [72]. Catheters placed in the internal jugular or subclavian veins, with the tip near the right atrium, minimise recirculation, making them more effective for ECCO₂R. In contrast, femoral catheters have higher recirculation rates and are less suitable for this therapy.

Figure 2. Fiber arrangement in parallel plate and circular Membrane Lungs (ML).In the circular ML, blood flow is nearly antiparallel to the gas fibres and flow. At the lower part of the ML, where blood enters, it encounters sweep gas with the highest CO₂ concentration. Since CO₂ removal depends on the diffusion gradient between the gas and blood—small in this region—CO₂ clearance is inefficient here. In the parallel-plate MLs, blood flows perpendicularly to the gas fibres. Similarly, at the lower part of the ML, where blood interacts with gas containing the highest CO₂ levels, CO₂ removal is less effective. With permission from Schwärzel et al. [73].

Figure 2. Fiber arrangement in parallel plate and circular Membrane Lungs (ML).In the circular ML, blood flow is nearly antiparallel to the gas fibres and flow. At the lower part of the ML, where blood enters, it encounters sweep gas with the highest CO₂ concentration. Since CO₂ removal depends on the diffusion gradient between the gas and blood—small in this region—CO₂ clearance is inefficient here. In the parallel-plate MLs, blood flows perpendicularly to the gas fibres. Similarly, at the lower part of the ML, where blood interacts with gas containing the highest CO₂ levels, CO₂ removal is less effective. With permission from Schwärzel et al. [73].

4.2 Pumps, Membranes and Circuits

4.2.1. Pumps

Conventionally, ECCO₂R uses roller, centrifugal or diagonal, electric or electromagnetic pumps. The roller pumps used in ECCO₂R systems were developed for renal replacement therapy (RRT) [74,75,76] or centrifugal pumps used for high-flow extracorporeal membrane oxygenation (ECMO) devices, with few systems explicitly designed for ECCO₂R [14,29,77]. However, RRT devices driven by roller pumps are limited in blood flow rates (usually up to 500 mL/min), limiting the CO₂ removal performance. Standard ECMO centrifugal pumps running below 2L/min cause increasing shear stress, leading to haemolysis [41].

Potential new pump technologies are on the horizon, for example, an innovative membrane and pump integrated to operate at very low flow (250ml/min) using six rotating impellers inside a cylindrical membrane configuration with a total surface area of 0.42m2. Blood moves from inside to outside through membrane fibres, allowing up to 75ml/min CO₂ removal rate with a hemolysis level comparable to standard CO₂ removal devices [78].

4.2.2. Membranes

Three major factors determine the amount of gas crossing membranes: the diffusion gradient, the membrane-blood contact time, and the characteristics of the membrane diffusion. The CO₂ diffusion gradient is determined by the CO₂ content of the blood and the air passing through the membrane lung, as well as the speed of the airflow. Membrane-blood contact time is determined by membrane geometry. Modern membrane lungs achieve adequate gas exchange with 1 to 3 m2 surface areas (Table 1).

Table 1. Comparison of available ECCO2R in the market based on the pump (pumpless, roller and centrifugal) and membrane function (gas and/or fluid CO₂ removal). The table summarises the key characteristics of roller and centrifugal pumps, focusing on parameters such as blood flow limits, priming volume, membrane surface area, membrane material, CO₂ extraction efficiency, VCO₂ monitor and circuit life. [19,27,79–83]

Table 1. Comparison of available ECCO2R in the market based on the pump (pumpless, roller and centrifugal) and membrane function (gas and/or fluid CO₂ removal). The table summarises the key characteristics of roller and centrifugal pumps, focusing on parameters such as blood flow limits, priming volume, membrane surface area, membrane material, CO₂ extraction efficiency, VCO₂ monitor and circuit life. [19,27,79–83]

4.2.2.1 Materials

Early fibres were constructed with microporous polypropylene. Micropores create microscopic blood-gas interfaces, allowing efficient gas exchange and causing plasma leaks. Recently, non-microporous poly-4-methyl-1-pentene (PMP) has been used; it provides superior gas exchange, better biocompatibility, and is less susceptible to plasma leak [84,85,86]. The gas exchange has been improved by arranging fibres into a complex mat and running blood on the outside[87]. This arrangement allows perpendicular blood flow to the fibres, enhancing mass transfer by reducing the diffusion path length compared to parallel flow.

4.2.2.2 Novel Surfaces for ECCO₂R

New materials and surface coatings are designed to reduce the inflammatory and coagulation responses triggered by artificial surfaces. Bioactive coatings, such as heparin-based and nitric oxide (NO) releasing materials, mimic natural endothelium and prevent blood clotting. Membranes may include a nitric oxide (NO)-eluting NO-eluting ECCO₂R system, tubing with NO-catalyzing surface coating or NO gas (80ppm) delivered into the membrane that inhibits platelet activation and aggregation to minimise thrombosis during extracorporeal CO₂ removal (ECCO₂R) [88].

Endothelialisation of surfaces is an emerging approach, with efforts to create artificial materials that mimic the human endothelial layer, potentially reducing the need for systemic anticoagulation and preventing thrombosis. Further innovations focus on developing fully biocompatible materials to avoid bleeding and clotting complications without compromising oxygenation efficiency, raising the possibility for long-term respiratory support in chronic lung failure​[89]. These include bioactive hollow fibre membranes (HFMs) coated with carbonic anhydrase (CA). CA immobilisation on HFMs increases the conversion of bicarbonate into CO₂, improving removal efficiency by creating a steeper diffusion gradient at the membrane surface. The CA-coated membranes also improved hemocompatibility, reducing platelet adhesion by 95% [90]. Acidic sweep gas containing sulfur dioxide (SO₂), in combination with carbonic anhydrase (CA) coated hollow fibre membranes (HFMs), can significantly enhance CO₂ removal from blood. This technique offers the potential for developing more efficient ECCO₂R devices. The findings could lead to smaller, less invasive respiratory support systems for patients with acute respiratory failure​. ​[91].

Innovative therapies, such as intravascular gas exchange devices, hold promise for lung support in acute respiratory conditions both in and outside the hospital. They also have long-term potential for managing chronic lung diseases while preserving patient mobility through continuous ECCO₂R. Despite their appeal, this technology has yet to reach clinical use. Only one device, the IVOX catheter, has advanced to human clinical trials but has yet to receive FDA approval. Technical challenges have hindered their progress, including optimising gas exchange within the vascular space and ensuring safety. Advancements in design may eventually offer a less invasive alternative to ECMO for managing acute respiratory failure​ [92].

4.2.2.3 Measuring device VCO₂ to assess membrane performance

When comparing devices and evaluating efficiency, it is crucial to understand how CO₂ is measured to estimate carbon dioxide transfer across the artificial membrane. This can be done by assessing trans-membrane CO₂ content differences using whole blood CO₂ content using the Douglas equation [93] or measuring the partial pressure of CO₂ in the effluent gas using infrared CO₂ sensors.[94] In both approaches, normalising to a standardised inlet PCO₂ is essential to define the device’s operating range and enable meaningful efficiency comparisons.

One of the concerns with effluent gas VCO₂ measurement is the need for more independent validation of VCO₂. However, this has been done with one device, little is known about the accuracy of the devices in which VCO₂ is measured with an infrared sensor measurement (Table 1)[95,96]. To improve comparative evaluations, the VCO₂ of each membrane and device should be standardised to inlet CO₂ and blood flow for consistent comparison and performance assessment. Manufacturers should ideally report their device’s index as VCO₂ per membrane surface area (e.g., ml/min/m²).[78]

4.3. Combined CO₂ Removal (“Lung Dialysis”) with renal support

Other strategies for enhancing CO₂ removal from the blood focus on methods targeting bicarbonate, as around 90% of CO₂ in the blood is transported as bicarbonate (Figure 2). The gas exchange membrane can be isolated or combined with a “renal” membrane (hemofilter) to achieve this. Research has shown that combining extracorporeal CO₂ removal (ECCO₂R) with CRRT effectively reduces arterial CO₂ (PaCO₂), improves pH, and stabilises hemodynamics in patients with acute respiratory distress syndrome (ARDS) and renal failure. Despite these benefits, mortality rates in critically ill patients remain high, especially in the most severe cases [97]. Integrating a hollow-fibre gas exchanger into CRRT platforms is relatively simple, and the gas exchange membrane can be placed before or after the haemofilter. Some data suggest that the CO₂ removal rate tends to be higher when the membrane is before the hemofilter. However, the clinical impact of this remains unknown [74]. An alternative strategy is integrating gas and fluid fibres in parallel using a shared circuit​ [98].

Figure 2. The proportion of carbamino form contribution from arterial CO₂ content to venous CO2 content is significantly higher than the other forms of CO₂ content due to the Haldane effect (increasing Hemoglobin affinity to CO₂ in lower oxygen concentration). The CO₂ removal of the gaseous phase of CO₂ represents only a small fraction of total CO₂ content. The bicarbonate removal using zero bicarbonate dialytic solution through hemofilter could contribute to significantly higher CO₂ removal of total CO₂ content, however, the blood buffer must be replaced by other natural buffers to maintain the acid-base equilibrium.

Figure 2. The proportion of carbamino form contribution from arterial CO₂ content to venous CO2 content is significantly higher than the other forms of CO₂ content due to the Haldane effect (increasing Hemoglobin affinity to CO₂ in lower oxygen concentration). The CO₂ removal of the gaseous phase of CO₂ represents only a small fraction of total CO₂ content. The bicarbonate removal using zero bicarbonate dialytic solution through hemofilter could contribute to significantly higher CO₂ removal of total CO₂ content, however, the blood buffer must be replaced by other natural buffers to maintain the acid-base equilibrium.

The optimal dialysis solution to use in combined devices remains to be discovered. Commonly available bicarbonate-based solutions increase total blood CO₂ content, which is problematic when high dialysis rates are required for solute clearance. Alternative solutions allow the simultaneous removal of H⁺ and bicarbonate, maintaining acid-base balance whilst allowing CO₂ removal rates of up to 160 mL/min [99]. Several approaches seek to manipulate regional pH to alter bicarbonate or CO2 clearance. Alkalinisation with bicarbonate free solutions, enhance bicarbonate removal are being investigated in animal models [100]. However, these methods may prove impractical for clinical use due to acid-base derangements, hemolysis, cardiac arrhythmias, and depletion of micronutrients, even though several approaches to replace bicarbonate have been attempted [100,101]. Respiratory electrodialysis, which increases CO₂ partial pressure in the blood through regional acidification, has been shown to remove approximately 50% of the total CO₂ metabolic production, offering a promising strategy for efficient CO₂ removal. These varied strategies illustrate ongoing efforts to develop more effective means of extracorporeal CO₂ removal, particularly for critically ill patients with multiple organ failure [102].

5. Conclusions

The future of ECCO₂R holds significant promise. However, two actions need to occur in parallel. First, the technology needs to be improved to reduce the impact on blood and coagulation whilst simultaneously increasing the efficiency of CO₂ removal. Second, the population who may benefit and the timing of ECCO₂R need to be better defined. ECCO₂R is being evaluated for patients with ARDS and COPD, but other groups, including patients bridging to transplant, may benefit.

Advances in membrane technology and gas exchange efficiency could further reduce the size and blood flow requirements of ECCO₂R systems, enhancing their safety and accessibility. These improvements could lead to the development of smaller, catheter-based systems that are less invasive and more practical for broader clinical use [100]. Developing new anticoagulation strategies for ECCO₂R is crucial, given the delicate balance between preventing clot formation within the circuit and minimising the risk of bleeding complications in patients. Artificial intelligence (AI) and machine learning could act similarly to the respiratory centres, allowing automated changes in CO₂ clearance based on changes in blood CO₂ and pH [103]. Ultimately, miniaturisation, development of closed feedback loops, reduced impact on blood/coagulation, and efficiency improvements will benefit the ICU population and may also allow the development of wearable devices that act as destination or prolonged bridging therapies [104].

Clinical trials are ongoing and will determine ECCO₂R 's efficacy, safety, and best practices. Overall, ECCO₂R 's future looks bright, with ongoing innovation and research likely to expand its clinical applications and affect critical care practice.