Effect of upper body venoarterial ECMO on systemic hemodynamics and oxygenation: A computational study

Abstract

Background:

This study seeks to quantify the effects of upper body veno-arterial extracorporeal membrane oxygenation (VA ECMO) on the anatomical distribution of oxygen delivery in the setting of hypoxic respiratory failure and provide new insights that will guide clinical use of this support strategy to bridge patients to lung transplant.

Methods:

Employing a patient-specific vascular geometry and a quantitative model of oxygen transport, computational simulations were performed to determine hemodynamics and oxygen delivery in the ascending and descending aorta, left and right coronary arteries, and great vessels during upper body VA ECMO support. Oxygen content in ECMO circuit blood flow was varied while considering different degrees of lung failure severity. Using lumped parameter models to dynamically apply perfusion boundary conditions, hemodynamic parameters and oxygen content were analyzed to assess the effect of ECMO supply titration.

Results:

The results emphasize the importance of anatomical distribution for tissue oxygen delivery in severe lung failure, with ECMO-derived flow primarily augmenting oxygen content in specific vascular beds. They also demonstrate that although cannulating the subclavian artery can enhance cerebral oxygen delivery, its ability to ensure sufficient oxygen delivery to the coronary circulation seems to be comparatively restricted.

Conclusions:

The oxygen delivery to a specific vascular area is primarily determined by the oxygen content in the source of perfusion. Caution is advised with upper body VA ECMO for patients with hypoxic respiratory failure and right ventricle dysfunction, due to potential coronary ischemia. Management of these patients is challenging due to disease progression and organ availability uncertainties.

Graphical Abstract

1. Introduction

Extracorporeal membrane oxygenation (ECMO) is increasingly relied on to sustain patients with pulmonary disease undergoing evaluation for lung transplantation [1–4]. The growing adoption of ECMO as a means to bridge end-stage lung disease patients to transplant is driven partially by the current system for allocating donor lungs which prioritizes recipients based on medical urgency [5]. This system motivates centers to maintain patients with deteriorating health as likelihood of potentially life-saving transplants increases with worsening decline in clinical condition. Unlike conventional therapies for respiratory failure, which rely on mechanical ventilation and often require deep sedation and paralysis, ECMO provides gas exchange independent of lung function. Even in severe disease, ECMO may allow for avoidance of intubation while enabling ongoing participation in physical therapy to promote patient vitality before undergoing the rigors of transplant surgery [6].

The capability of ECMO to maintain transplant candidates is dependent on both the underlying lung disease and the specific implementation of extracorporeal support. End-stage disease can manifest with varying degrees of overlapping hypoxic and hypercapnic respiratory failure and is often associated with pulmonary hypertension and right ventricular (RV) dysfunction. While isolated respiratory failure may be adequately supported by venovenous (VV) ECMO cannulation methods, in which venous blood is withdrawn and passed through the extracorporeal circuit to return oxygenated blood to the venous system and forward through the cardiopulmonary circulation, the optimal strategy to maintain patients with concomitant RV failure is unknown [7,8].

The novel upper body venoarterial (VA) ECMO cannulation strategy has been described as a method to provide circulatory support to bridge patients with primary pulmonary hypertension complicated by RV failure to transplant [9,10]. This approach provides oxygenated blood from the ECMO circuit to the systemic arterial circulation via a graft sutured to the axillary or subclavian artery [10]. More recently, this strategy has been advocated as a means of supporting transplant candidates suffering from hypoxic respiratory failure compounded by RV dysfunction – a common manifestation of end-stage interstitial lung disease (ILD) [11,12]. The purported advantage is that it avoids the complication of differential hypoxia associated with femoral artery cannulation and retrograde delivery of oxygenated blood to the distal aorta in the setting of underlying lung disease. By supplying perfusion to the aortic arch via the axillary or subclavian artery, upper body VA ECMO aims to simultaneously [1] offload the failing RV to improve cardiac function and [2] provide adequate systemic oxygenation to sustain end-organ homeostasis.

Despite proposed benefits of upper body VA ECMO, there is insufficient understanding of the physiological effects of this approach to optimize its use and determine appropriate patient selection. ILD patients can suffer from fulminant disease, making adequate support clinically challenging. Improved understanding of the anatomical distribution of oxygenated blood in upper body VA ECMO and how titration of support affects oxygen delivery is urgently needed to determine whether this modality is feasible for this critical patient population, despite the inherent risks. While rigorous in vivo assessment of blood flow distribution and oxygen delivery requires highly invasive methods that are impractical in critically ill patients requiring extracorporeal support [13], advanced modeling and computational fluid dynamics (CFD) are capable of generating new insights to guide clinical use [14–16].

CFD has successfully provided a new understanding of the hybrid circulation produced by interactions between the failing heart and the ECMO circuit. Numerous studies have examined the size and behavior of the watershed region in ECMO flow [17,18]. Ahmed et al. [19] employed patient-specific computational modeling to evaluate different cannulation strategies for ECMO. Khamooshi et al. [20] used CFD simulations to investigate the complications associated with VA-ECMO, particularly focusing on how support levels and return cannula sizes affect blood flow and emboli transport. The current study builds on the foundation of prior work in which advanced computational methods were developed to investigate the effect of varying vascular loading conditions on the efficacy of extracorporeal circulatory support [21,22], and uses that platform to examine how altering the placement of the return cannula alters systemic hemodynamics and oxygen delivery in the setting of underlying lung disease. The broader goal of this work is to create powerful analytic tools to further understanding of ECMO and improve clinical use of advanced mechanical support.

2. Methods

2.1. Patient-derived anatomical geometry and computational domain

De-identified DICOM images were obtained from clinical computer tomography performed on a patient receiving upper body VA ECMO as a bridge to lung transplantation. The images were obtained and processed in accordance with Institutional Review Board (IRB) approval and specified guidelines (IRB Protocol #: 2017P002655; Partners Human Research Committee; Brigham and Women’s Hospital, Boston, MA). Images were segmented using the 3D Slicer package to generate vascular geometry [23]. The patient-specific vascular model extended from the aortic valve to the mid-thoracic descending aorta and included brachiocephalic artery (BCA), left common carotid artery (LCCA), and left subclavian artery (LSA) with some major immediate bifurcations (Fig. 1). Of note, this patient had a ‘bovine’ aortic arch, a common anatomical variant in which the LCCA originates from the trunk of the BCA instead of the aortic arch. The aortic tree also included the left and right coronary arteries with major bifurcations. The suturing of a Dacron graft to the LSA as source of ECMO-supplied perfusion was geometrically replicated using CAD software. After conducting a mesh-independency test for the model, the computational grid, generated using COMSOL (COMSOL Multiphysics, Burlington, MA), divided the domain into 1.9 million tetrahedral volumetric elements. The big size of computational mesh suggested by the mesh-independency test is due to the high velocity of ECMO jet requiring considerable mesh refinement in its vicinity to accurately capture complex flow features. The average quality of mesh was approximately 0.68 and, to ensure desired Y-plus values to properly capture near-wall turbulent effects, sizing of neighboring wall elements was carefully adjusted.

Fig. 1.

Patient-specific model for the cannulated case. The inset provides a closer view of the computational grid, which has been refined in proximity to the luminal wall to ensure accuracy. Dynamic boundary conditions are applied using lumped parameter models, and the native heart and ECMO inlets are identified for reference by arrows.

2.2. Computational method

Blood flow was modeled as an incompressible, homogenous, and Newtonian fluid with dynamic viscosity of 3.5 cP and density of 1065 kg/m^{^}3 [24]. Turbulent regime settings were applied using the k-w model to describe the velocity field and pressure due to the high velocity of the ECMO jet and its interaction with native flow [25]. The CFD framework was set up with two distinct inlet flow sources from the left ventricle into the ascending aorta and through the axillary artery from the sutured ECMO graft. A physiologically scaled waveform was applied as heart output and a steady flow rate was set for ECMO flow [25]. Simulations were performed at ECMO flow rates of 2, 3, and 4 L per minute (LPM) to evaluate the effects across a broad range of support. Although 4 LPM via an axillary graft is not clinically feasible in our experience, it was chosen as an extreme boundary to quantify the theoretical upper limit of extracorporeal oxygen delivery. The heart flow waveform was scaled to conserve total perfusion at 5 LPM to enable comparative analysis to published studies [17,18,25]. Due to the variable proportion of inlet flow from native and support sources, dynamic boundary conditions were employed to simulate physiological conditions and ensure unbiased distribution of blood to different aortic outlets. To this end, a 3-element Windkessel model, comprising two resistance elements and one compliance element (Fig. 1), was used at the proximal outlets and descending aorta [21–23,26] as below:

$$\frac{\partial p}{\partial t}+\frac{p}{C{R}_d}=\frac{Q}{C}\left(1+\frac{R_p}{R_d}\right)+R_p\frac{\partial Q}{\partial t}$$ (1)

Where $R_p,R_d$, and $C$ are the proximal resistance, distal resistance, and compliance of the lumped elements, respectively. To account for the phase difference between coronary branches and the root of the aorta, a 7-element windkessel model was employed to properly consider the additional effect of myocardial compliance and intramyocardial pressure [27,28]. This model is depicted in Fig. 1, and the corresponding ODE is as follows:

$$A_0\frac{{\partial }^{2}p}{\partial t^2}+A_1\frac{\partial p}{\partial t}+A_{2}p=B_0\frac{\partial \left(P_{im}\right)}{\partial t}+C_0\frac{{\partial }^{2}Q}{\partial t^2}+C_1\frac{\partial Q}{\partial t}+C_{2}Q$$ (2)

with the following coefficients:

$$A_0=C_a{C}_{im}{R}_{a-micro}\left(R_V+R_{V-micro}\right),$$

$$A_1=C_a{R}_{a-micro}+\left(R_V+R_{V-micro}\right)\left(C_a+C_{im}\right),$$

$$A_2=1,$$

$$B_0=C_{im}\left(R_V+R_{V-micro}\right),$$

$$C_0=C_a{C}_{im}{R}_a{R}_{a-micro}\left(R_V+R_{V-micro}\right),$$

$$C_1=R_a{C}_a\left(R_{a-micro}+R_V+R_{V-micro}\right)+C_{im}\left(R_a+R_{a-micro}\right)\left(R_V+R_{V-micro}\right),$$

$$C_2=\left(R_a+R_{a-micro}\right)+\left(R_V+R_{V-micro}\right)$$

In equation (2), $P_{im}$ is the intramyocardial pressure, which induces a phase difference between flow and pressure in the coronary arteries during the systole and diastole. Moreover, $R_a,R_{a-micro},R_V$, and $R_{V-micro}$ represent coronary arterial resistance, resistance in the coronary arterial microcirculation, coronary venous resistance, and resistance in the coronary venous microcirculation, respectively. Also, $C_a$ is the coronary arterial compliance, and $C_{im}$ represents myocardial compliance [27].

In blood, up to 97 % of the oxygen transport occurs through binding to hemoglobin within red blood cells. The process of oxygen transport in blood can be mathematically described by a transient advection-diffusion equation, given by Refs. [29,30]:

$$\frac{\partial P{O}_2}{\partial t}+\overrightarrow{u}\cdot \nabla P{O}_2=\nabla \cdot \left(D_b\nabla P{O}_2\right)$$ (3)

where $P{O}_2$ is the partial pressure of oxygen, $\overrightarrow{u}$ is the blood flow velocity field, and $D_b$ is the diffusivity of oxygen in blood, set to 1.2e⁻⁹ m^{^}2/s [31]. This formulation captures the critical aspects of oxygen transport, integrating the effects of both advection due to the movement of blood and diffusion resulting from the concentration gradients of oxygen molecules. By employing this equation, one can model the transient behavior of oxygen distribution in the bloodstream, providing insights into the dynamics of oxygen delivery to tissues. Simulations were performed for partial pressure oxygen levels of 30 mm Hg and 60 mm Hg in the arterial blood ejected by the left ventricle to model both severe hypoxic lung disease and borderline hypoxia, respectively. The level of oxygenation provided by the ECMO circuit was simulated at three different levels to correspond to variations in both oxygenator performance as well as titration of sweep gas oxygen fraction (150 mmHg, 300 mmHg, and 500 mmHg).

All simulations were performed using COMSOL Multiphysics, utilizing fluid flow and specific transport modules. The computations were executed on a local processing cluster with 48 computational nodes, with each case requiring an average of 100 h of computational time. An implicit backward differentiation formula (BDF) was used for time discretization, with a maximum timestep of 0.01 s. All models were solved using a fully coupled approach, and simulations were considered converged when the maximum residuals dropped below 1e-4. To ensure accurate convergence to the periodic state, simulations were extended to span fifteen cardiac cycles, with data from the final cycle extracted for presentation.

3. Results

The anatomical distribution of ECMO-circuit-derived blood flow depends on both the relative fraction of total perfusion provided by extracorporeal support and the phase of the cardiac cycle. At peak systole, pulsatile antegrade blood flow from the left ventricle (LV) impedes the proximal extent of perfusion provided by the ECMO circuit. Upon completion of LV ejection and the onset of diastole, continuous ECMO flow enters the aortic arch, supplying perfusion to the descending aorta (Fig. 2). As the fraction of total perfusion delivered by extracorporeal support increases, the anatomical distribution of ECMO-circuit-derived blood shifts proximally, entering the ascending aorta. Compared to the healthy control, in which all perfusion is provided by the pulsatile heart, the institution of ECMO support reduces the systolic contribution of perfusion while proportionally increasing perfusion during diastole (within a range of [0–4] Lpm), depending on the ECMO flow rate (Fig. 3). Unlike the large vessels, coronary artery perfusion primarily occurs during diastole, when the myocardium relaxes, and extrinsic compression on cardiac vasculature dissipates, allowing flow. This dependence of the heart on the diastolic phase for perfusion minimizes the observed differences in coronary blood flow following initiation of ECMO support.

Fig. 2.

Hemodynamic patterns and interaction of blood flow from different sources, i.e. native heart and ECMO, at selected time points of cardiac cycle for varying titration of extracorporeal support.

Fig. 3.

Relative perfusion of different vascular outlets over the cardiac cycle for varying ECMO input compared with the physiological state.

To evaluate the effect of upper body VA ECMO on oxygen delivery, simulations were performed at three different oxygen content levels in ECMO-circuit-derived blood (150 mmHg, 300 mmHg, and 500 mmHg)—selected to represent a range from moderately supraphysiological levels to the upper extreme of oxygenator performance. To model both decompensated and compensated hypoxic respiratory failure, the partial pressure of oxygen in the blood ejected from the left ventricle was set at either 30 or 60 mm Hg. The peak oxygen content in any specific vascular bed was found to be predominantly dependent on the source of perfusion, with oxygen diffusion from ECMO-derived flow providing only a negligible increase. These findings highlight the importance of the anatomical distribution of extracorporeal perfusion as the primary means of augmenting tissue oxygen delivery in the setting of severe lung failure.

The watershed region, the anatomical location within the aorta where extracorporeal perfusion collides with antegrade perfusion ejected by the LV, differs markedly between upper body VA ECMO and femoral artery cannulation. ECMO-circuit-derived blood enters the aortic arch and is directed retrograde towards the right aortic sinus of Valsalva (Supplemental Video 1), where it mixes with heart-generated flow. This directed jet of extracorporeal perfusion has a significant impact on the oxygen content of blood flow to the right coronary artery (RCA) (Fig. 4). As the ECMO flow rate increases, the proportion of ECMO-circuit-derived flow to the RCA rises, in contrast to the left coronary artery (LCA) which remains predominantly perfused by the heart. Only when the ECMO flow rate reaches 80 % of total perfusion does the oxygen content of LCA blood flow increase substantially, and even then, this increase is observed primarily during systole, with diastolic flow continues to be provided by the heart.

Fig. 4.

Variation in oxygen partial pressure across different vascular outlets throughout the cardiac cycle with titrated extracorporeal support: 2 LPM (left), 3 LPM (middle), and 4 LPM (right).

Supplementary video related to this article can be found at https://doi.org/10.1016/j.compbiomed.2024.109124

The perfusion of the descending aorta consists of a mixture of ECMO- and heart-derived blood flow. At a moderate ECMO flow rate of 3 L/min (Fig. 5), the oxygen content in the descending aorta is maintained at supraphysiological levels, even in cases of severe lung failure. In contrast, oxygen levels in the coronary arteries remain markedly lower than what is observed in the great vessels and distal to the aortic arch (Fig. 6). The partial pressure of oxygen in the LCA does not substantively rise until ECMO flow rates reach 80 % of total perfusion. Even with a starting PO₂ of 60 mm Hg, the oxygen content in the LCA remains below 100 mm Hg until the ECMO circuit delivers 4 LPM of blood flow with an oxygen level of 300 mm Hg.

Fig. 5.

Comparing the effect of change in ECMO and native heart input partial pressure of oxygen on oxygenation of different vascular outlets throughout the cardiac cycle with moderate ECMO flow rate (3Lpm). Left: increasing the input partial pressure of oxygen for ECMO to 500 mmHg; Right: inducing pulmunary insufficiency reducing the native blood partial pressure of oxygen down to 30 mmHg.

Fig. 6.

Mapping oxygenation across different vascular outlets with changing support flow rate and oxygen saturation.

4. Discussion

Upper body VA ECMO has been proposed as a method to provide both systemic oxygenation and circulatory support for patients with end-stage disease and concomitant RV failure despite limited understanding of its efficacy and impact on end-organ oxygen delivery. Rigorous in vivo assessments of upper body VA ECMO are restricted by the severity of patient’s disease and the need for highly invasive diagnostic methods. To enhance our understanding of the physiological effects of upper body VA ECMO and guide its clinical application, we developed an advanced computational platform to investigate the anatomical distribution of blood flow and oxygen delivery using a patient-derived vascular model.

4.1. End-organ oxygenation is determined by extracorporeal perfusion

Unlike femoral arterial cannulation, in which continuous perfusion from the extracorporeal circuit advances retrograde up the distal aorta to collide with pulsatile antegrade blood flow from LV, upper body VA ECMO provides perfusion proximally into the aortic arch via the subclavian artery. While femoral cannulation creates a distinct watershed region where end organs receive perfusion either from the extracorporeal circuit or the heart, upper body VA ECMO induces mixing of these two sources of perfusion within the aortic arch, leading to variability in the predominant source of blood flow to specific vascular beds. At low ECMO flow rates, extracorporeal blood flow is insufficient to overcome pulsatile flow ejected by the LV and is directed distally into the descending aorta. As ECMO flow increases, it advances retrograde into the arch and can also provide perfusion to the contralateral side of upper body.

A key finding of this work is that the effect of oxygen diffusion on total oxygen delivery to a specific anatomical region is negligible. Instead, oxygen delivery to a given vascular bed is almost entirely determined by the source of perfusion. The rate of flow into the aortic arch minimizes the time available for oxygen to diffuse from the high-oxygen-content blood of the ECMO circuit to lower-oxygen-content blood ejected by the heart. Our simulations qualitatively and quantitatively demonstrate that increased ECMO flow alters flow patterns throughout the cardiac cycle, affecting coronary artery oxygenation. Even with higher ECMO blood oxygen partial pressures, minimal increases in oxygen delivery were observed, emphasizing the dominance of flow patterns. Therefore, it is critical to model both oxygen transport and blood flow to understand oxygen delivery accurately. Even when the oxygen partial pressure of ECMO blood was increased to 500 mm Hg to maximize the diffusion gradient, while modeling severe hypoxic respiratory failure with an oxygen partial pressure of 30 mm Hg, minimal increase in oxygen delivery was observed. This result affirms that the efficacy of upper body VA ECMO in achieving oxygenation is determined by the anatomical distribution of extracorporeal blood flow.

4.2. Oxygen delivery and coronary artery ischemia

Published reports on differential hypoxemia primarily focus on cerebral perfusion and the associated risk of ischemic brain injury due to insufficient oxygenation of blood ejected by the LV [32,33]. This focus is mirrored in published computational models of VA ECMO and femoral artery cannulation which typically restrict their simulations to the effects on the great vessels [17,18,21,22,25]. A key limitation of these approaches is the lack of consideration given to the coronary vasculature, which is anatomically the first to arise from the aorta. The current study seeks to advance the understanding of extracorporeal circulatory support by examining blood flow distribution within the proximal and descending aorta. To achieve this, we expand prior anatomical geometries to include physiologically validated models of the coronary vasculature. This innovation is crucial for studying upper body VA ECMO as this modality has been proposed to support patients with severe lung disease who are at high risk of coronary ischemia induced by insufficient oxygenation of blood passing through the lungs.

The results of this computational study demonstrate that while cannulation of the subclavian artery may increase cerebral oxygen delivery by augmenting oxygen content in at least one carotid artery feeding into the Circle of Willis, its effectiveness in ensuring oxygen delivery to the coronary circulation is more limited. Flow to the left coronary artery was generated almost entirely by the native heart when ECMO flow was at 40 % and 60 % of total perfusion. At ECMO flows of 80 % of total perfusion, a flow rate difficult to achieve in this configuration due to excessive flow to the ipsilateral arm, the LCA experiences increased oxygenation during systole which then rapidly gives way to decreased oxygenation as native flow is returned during diastole. Oxygen delivery to the RCA is higher throughout the cardiac cycle as ECMO flow rates increase which due in part to directionality of the flow jet originating from the subclavian artery and directed towards the RCA ostium. Another factor is the model itself which presumes physiologically normal RCA impedance which likely is not the case in the setting of severe lung disease and RV hypertrophy. As RV pressures increase, RCA impedance could be expected to mirror that of the LCA with resultant flow regimes likewise mirroring the left side. Thus, the improved RCA perfusion with increasing levels of ECMO flow may represent a best-case scenario, with actual RCA flow potentially being reduced in comparison.

4.3. Clinical translation and extracorporeal support strategies

Clinical management of end-stage lung disease patients with concomitant RV failure is challenging, with the optimal support strategy currently unknown. While upper body VA ECMO has been proposed as a support method for patients with hypoxic respiratory failure complicated by RV dysfunction, the results of the current study question its efficacy in the setting of severe hypoxia. Patients with severe impairment are at risk of potentially life-threatening coronary ischemia due to the LV ejecting low oxygen content blood. Additionally, the location of the watershed region within the aortic arch in upper body VA ECMO impedes the ability to monitor the oxygen content in the blood ejected by the LV as it mixes with blood from the ECMO circuit before reaching clinically accessible anatomical sites. This limitation makes bedside management of oxygen delivery to the lungs more uncertain and increases the risk of inducing cardiac dysfunction or injury from inadvertent mismanagement of support.

The ability of upper body VA ECMO to offload the impaired RV is its primary advantage as a cannulation technique and suggests that its use should be targeted towards patients with isolated RV failure. While providing oxygenated blood to the aortic arch, results of this study suggest that it primarily acts to increase oxygen delivery to the descending aorta rather than to the coronary circulation and ascending aorta. In the setting of severe hypoxic failure or in patients at risk of rapidly progressive disease, such as acute exacerbations of interstitial pulmonary fibrosis, hybrid cannulation methods will likely be required to ensure adequate oxygenation of blood ejected by the LV. Such approaches may also act to decrease pulmonary vascular resistance and improve RV function, although this has yet to be proven in restrictive lung disease.

4.4. Limitations

While our study provides valuable insights into the effects of upper body VA ECMO on oxygen delivery, it also has limitations that could introduce potential sources of error and uncertainty. First, our simulations, driven by therapeutic complications encountered in a specific patient treated at our institution, were based on a single patient-specific vascular geometry. Variations in geometry among different patients, particularly concerning size and sex, can influence flow patterns and mixing, potentially affecting the generalizability of our findings. Nevertheless, this study provides invaluable insights into the effects of support flow titration and coronary perfusion, with a primary focus on hemodynamic patterns and oxygen transport. These findings enhance our understanding of patient-specific cardiovascular dynamics and can guide future research and clinical practices. Second, we used a scaled physiological waveform as the domain inlet flow originated from the native heart instead of a detailed, patient-specific heart model or a catheter-based measurements. This was due to the ethical limitations associated with conducting invasive measurement, outside clinical routines, to extract simulation inputs. Consequently, this assumption limits our ability to capture the full interaction between ECMO flow and cardiac function, including the critical effect of varying afterload and consequent cardiac response. Other limitations include assuming rigid aorta, ignoring the vascular tone and vasomotion in peripheral arteries in response to cardiac output and support titration, and overlooking biological changes and comorbidities in the studied case. These simplifying assumptions are commonly made in similar CFD works, primarily due to limitations in measurement tools, the ethical challenges of in vivo measurements in clinical settings, and the complications and variability inherent in biological substrates. Despite these limitations, we are confident that our key insights into oxygen delivery distribution and anatomical variations remain solid. Though it is impossible to remove all these limitations, future research should include diverse patient geometries and more advanced heart models to deepen our understanding and make the simulation cases more physiologically accurate.

4.5. Future directions

Advanced computational and modeling methods enable rigorous analysis of blood flow dynamics and the precise control of input variables, which are not feasible in the in vivo setting. The computational platform detailed in this study provides the basis for further analysis of cannulation methods and approach strategies that can be employed for patients undergoing transplant evaluation in advance of clinical deterioration. These methods have the potential to both optimize the use of extracorporeal support and enhance the ultimate benefit of lung transplantation.

5. Summary & conclusion

ECMO is increasingly relied upon to bridge patients with end-stage pulmonary disease to lung transplantation. While many patients can be sustained with VV cannulation strategies, where venous blood is withdrawn via vascular cannula, passed through the ECMO circuit to return oxygenated blood to the venous system, and then into the cardiopulmonary circulation, the optimal support strategy for patients with lung failure and concomitant circulatory failure due to right ventricular dysfunction is unknown. Upper body VA ECMO has been described as a potential approach in which oxygenated blood is returned to the arterial system via a vascular graft to the axillary or subclavian artery. This cannulation strategy seeks to simultaneously unload the failing right ventricle while also improving systemic oxygenation. While upper body VA ECMO has been clinically described as a means of supporting patients with respiratory failure complicated by severe right ventricular impairment, as commonly occurs in end-stage interstitial lung disease, its effects on systemic oxygen delivery have yet to be fully investigated.

Using patient-derived anatomy, CFD, and a model of intravascular oxygen transport, this work evaluated the effect of increasing extracorporeal blood flow on systemic oxygen delivery in an upper body VA ECMO cannulation strategy. The patient anatomy included coronary arteries to specifically investigate the effect of this cannulation approach on the distribution of ECMO-derived blood to the heart. Continued forward flow through the cardiopulmonary circulation is vital to avoid stasis and the risk of lethal thrombus formation in the LV. In practice, this results in the ejection of blood from the LV that relies on the diseased lungs to maintain sufficient gas exchange. Blood ejected antegrade from the LV then collides with ECMO-derived blood advanced retrograde via the axillary or subclavian artery vascular graft into the aortic arch. The proximal extent of the ECMO-circuit-derived blood is unknown and, in the setting of profound lung disease with inadequate oxygen transfer, the patient risks coronary ischemia. The results of the CFD model demonstrate that even with supraphysiological flow through the ECMO circuit, there is insufficient delivery of oxygenated blood to prevent coronary ischemia in severe lung disease. These findings suggest that caution should be exercised when using upper body VA ECMO as a supportive strategy for patients with concomitant lung and right ventricular failure. Notably, while the ECMO circuit enhances cerebral oxygen delivery, this further complicates detection of coronary ischemia as oxygen saturation in the upper extremities and body may appear sufficient. These results underscore the importance of ongoing research into the optimal support strategy for transplant candidates with end-stage lung disease complicated by right ventricular failure.

Glossary of abbreviations:

BCA

Brachiocephalic Artery

CAD

Computer-aided Design

CFD

Computational Fluid Dynamics

ECMO

Extracorporeal Membrane Oxygenation

IRB

Institutional Review Board

LCCA

Left Common Carotid Artery

LCA

Left Coronary Artery

LSA

Left Subclavian Artery

LV

Left Ventricle

LPM

Liters per Minute

RCA

Right Coronary Artery

RV

Right Ventricle

VA

Veno-Arterial

VV

Veno-Venous