Veno-Pulmonary Arterial Extracorporeal Membrane Oxygenation in Severe Acute Respiratory Distress Syndrome: Should We Consider Mechanical Support of the Pulmonary Circulation From the Outset?

Severe acute respiratory distress syndrome (ARDS) confers significant in-hospital mortality despite implementation of evidence-based ventilatory strategies.1–3 Acute right ventricular (RV) injury characterized by abnormal RV and pulmonary vascular biomechanics is one of the major determinants of short-term mortality in patients with severe ARDS.4,5 The reported prevalence of RV injury in ARDS (defined as RV dysfunction, acute cor-pulmonale (ACP), RV dysfunction with hemodynamic compromise, or RV failure) is 21%.4 A universal RV injury definition is however lacking and a spectrum of different phenotypes and subphenotypes (from RV diastolic dysfunction to advanced RV injury/failure) have been linked to mortality in patient populations with ARDS.6–9 The main mechanisms of pulmonary vascular dysfunction and RV injury in ARDS include: obstruction due to thrombosis and remodeling; compression due to elevated extravascular lung water and injurious invasive ventilation; and constriction due to hypercapnia, hypoxemia, and acidemia. The resultant acute pulmonary hypertension and negative diastolic interaction between the right and left ventricle (LV) may lead to RV dilatation with or without systemic venous congestion, uncoupling between the RV and the pulmonary circulation, reduced LV filling and shock.7,10 Veno-venous extracorporeal membrane oxygenation (V-V ECMO) is increasingly being used as part of the algorithm for the management of patients with severe ARDS in whom conventional ventilatory strategies (including prone positioning) fail to preserve gas exchange without causing an additional lung insult.11 During V-V ECMO support, the high extracorporeal blood flow rate and diffusion of gases between central venous blood and the “sweep gas” flowing through the artificial membrane’s fiber bundles provide oxygenation and removal of excess CO2.12,13 It would therefore stand to reason that V-V ECMO unloads the RV by mitigating or reversing the aforementioned mechanisms: correction of hypoxemia and hypercapnic acidemia, and reduction in the intensity of mechanical ventilation by facilitating a decrease in mechanical power while maintaining alveolar recruitment.14–19 However, in a recent systematic review and meta-analysis including seven studies and 452 patients with severe ARDS receiving V-V ECMO, our group demonstrated that RV injury may persist or even develop despite application of V-V ECMO and is strongly associated with significant mortality.20 In this editorial, the authors discuss the potential role of veno-pulmonary arterial (V-PA) ECMO for additional RV support in patients with severe ARDS requiring V-V ECMO support. Veno-Pulmonary Arterial ECMO Physiology In patients with severe ARDS requiring respiratory extracorporeal support, V-V ECMO can be applied through different strategies: a) dual-site cannulation of two separate central veins (femoral vein - femoral vein or femoral vein - internal jugular vein) where deoxygenated blood is drained by one central venous cannula into the membrane lung and following oxygenation and removal of excess CO2 it is returned to another central venous site by another cannula; and b) single-site (internal jugular vein) dual lumen bicaval cannula with one lumen draining blood from the subhepatic inferior vena cava (IVC) and superior vena cava (SVC) and a second lumen returning oxygenated blood to the right atrium (RA).21 In severe ARDS, assuming there is minimal or no native lung function the systemic arterial content is the result of the mixture of two flows: the oxygenated blood flow from the membrane lung (usually 60–80% of the venous return) and the native venous blood flow passing into the RV.22 Oxygen supply is considered adequate when the ratio of oxygen delivery (DO2) to oxygen consumption (VO2) is greater than 3.22 Despite adequate systemic oxygenation and CO2 clearance during V-V ECMO support however, RV injury and occasionally failure may still ensue.6,20 In V-PA ECMO, there is less mixing of deoxygenated blood since the ECMO inflow and outflow are separated by the RV.23 In addition, in V-PA ECMO, the extracorporeal contribution to DO2 is more assured as it is not dependent on RV contractility. Two potential V-PA ECMO cannulation strategies to be employed include: a) single-site, dual stage (through the right internal jugular vein), RA (proximal fenestrations, inflow to ECMO circuit/drainage) to PA (distal fenestrations, outflow/return) (Figures 1–4); and b) dual-site cannulation of one femoral vein (inflow to ECMO circuit/drainage) and PA (outflow/return) (Figures 3, 4). The return of oxygenated blood distal to the tricuspid and pulmonary valves into the PA (avoiding recirculation) theoretically protects and supports the RV.24 Right atrial drainage when using a dual-stage cannula and bypassing of the RV theoretically reduce systemic venous congestion and RV preload, decrease RV wall stress and oxygen consumption, and unload the injured ventricle.25 The V-PA ECMO configuration may therefore restore the relationship between RV contractility and RV afterload in patients with severe ARDS and RV injury or protect the RV and theoretically mitigate RV-PA uncoupling in patients without evidence of RV injury at the time of ECMO application. Figure 5 summarizes the main differences between conventional V-V ECMO and V-PA ECMO. Although V-PA ECMO provides excellent RV unloading and support in those with isolated RV injury, concerns remain over ability to safely provide higher blood flows (5–7 liters per minute) especially when using a dual lumen cannulae (Table 1). However, the need for such a high ECMO blood flow could be reduced by negligible recirculation achieved with V-PA ECMO. Table 1. - Studies Reporting on Outcomes Relating to Percutaneous Placement of V-PA in Patients With ARDS Source Study Type V-PA ECMO Configuration Comparator ARDS Etiology, Sample Size (Subgroups) V-PA ECMO From the Outset Outcomes Mustafa et al. 23 Retrospective propensity score matched analysis (2 major centers) Single-site, dual-lumen RA-to-PA cannula; EBF not available Maximum conventional IMV COVID-19, 80 matched pairs after propensity score matching Yes Survival to hospital discharge: 68% V-PA ECMO vs. 26% IMV aloneMajor complications (septic shock, ventilator associated pneumonia, inotropic requirements, acute liver and kidney injuries) less frequent in V-PA ECMO group Mustafa et al. 24,25 Single-center, retrospective observational Single-site, dual-lumen RA-to-PA cannula; EBF 3-4.5 LPM None COVID-19, 40 Yes Survival to hospital discharge: 82.5%Liberation from ECMO: 80%Mean (SE) time-to-extubation: 13 (2.6) daysMinimal cannula-associated complications or revisions El Banayosy et al. 26 Single-center, retrospective observational Single-site, dual-lumen RA-to-PA cannula used alone (V-PA) or in combination with V-V ECMO (VV-PA); EBF up to7 LPM with VV-PA configuration None COVID-19, 9 2 patients had empiric V-PA ECMO from the outset and 7 had either V-V or VA-ECMO that was reconfigured to V-PA at later stage Survival to hospital discharge: 67%V-PA and VV-PA configurations are feasible and without complications Saeed et al. 27 Multicenter, retrospective observational Single-site, dual-lumen RA-to-PA cannula; EBF not available a) Dual-site femoral vein-femoral vein or femoral vein-internal jugular vein; b) Single-site, dual-lumen cannula in IJV advanced through the SVC into the RA with tip positioned in the IVC COVID-19, 435 (99 V-PA ECMO; 247 dual-site V-V ECMO; 89 single-site V-V ECMO) Yes 90 day in-hospital mortality for entire cohort: 55%Unadjusted 90 day in-hospital mortality: V-PA ECMO 41%; dual-site V-V ECMO 60%; single-site dual-lumen V-V ECMO 61%, p = 0.06Adjusted 90 day in-hospital mortality was lower in V-PA ECMO (HR 0.52, p = 0.029) and similar in single-site V-V ECMO (HR 0.98, p = 0.86) compared with dual-site.V-PA ECMO: longer duration of ECMO support compared with other modes.V-PA ECMO: shorter duration of IMV and more commonly discharged home Cain et al. 28 Single-center, retrospective observational Single-site, dual-lumen RA-to-PA cannula; EBF 3-4.5 LPM Patients receiving IMV alone COVID-19, 39 (18 V-PA ECMO; 21 IMV alone) Yes In-hospital mortality for the entire cohort: 33%In-hospital mortality: V-PA ECMO 11.1%; IMV 52.4%, p = 0.00830 day mortality: V-PA ECMO 5.6%; IMV 42.9%; p = 0.011V-PA ECMO: shorter duration of IMVAKI: V-PA ECMO 0%; IMV 71.4%; p ≤ 0.001 Smith et al. 29 Single-center, retrospective observational Single-site, dual-lumen RA-to-PA cannula; EBF not provided Conventional V-V ECMO[Early pandemic V-PA ECMO (Era 1) was compared to late pandemic V-PA ECMO (Era 2)] COVID-19, 54 (38 V-PA ECMO; 16 V-V ECMO) Yes In-hospital mortality for the entire cohort: 42.6%In-hospital mortality: V-PA ECMO 39.5%; V-V ECMO 50%120 day mortality: V-PA ECMO 40%; V-V ECMO 60.8%Duration of IMV after cannulation: V-PA ECMO 16 (1–80) days; V-V ECMO 30.5 (0–43) daysTime-to-liberation from ECMO: V-PA ECMO 26 (4–85) days; V-V ECMO 35 (3–78) daysAKI requiring RRT: V-PA ECMO 26.3%; V-V ECMO 42.9%Cannula-associated complications: V-PA 13.2%; V-V ECMO 15.4% Khorsandi et al. 30 Single-center, retrospective observational Single-site, dual-lumen RA-to-PA cannula; EBF not available None COVID-19, 33 (14 RV injury; 19 no RV injury; 3 with RV injury had V-PA ECMO) No V-PA ECMO-related outcomes not reported Ivins-O’Keefe et al. 31 Single-center, retrospective observational Dual-site, femoral vein-to-PA; EBF 3-6 LPM None COVID-19 (90%), other etiology (10%), 21 No(V-V ECMO or V-VA-ECMO was converted to V-PA ECMO in all 21 patients) Survival to hospital discharge: 25%Median time-to-liberation from ECMO: 32 [21–94] daysMedian time from V-V to V-PA ECMO: 12 [8.5–23.5] daysPre-V-PA hepatic dysfunction: 71%Pre-V-PA renal dysfunction: 76% AKI, acute kidney injury; ARDS, acute respiratory distress syndrome; COVID-19, coronavirus disease 2019; EBF, extracorporeal blood flow; ECMO, extracorporeal membrane oxygenation; IJV, internal jugular vein; IMV, invasive mechanical ventilation; IVC, inferior vena cava; LPM, liters per minute; PA, pulmonary artery; RA, right atrium SVC, superior vena cava; VA-ECMO, veno-arterial extracorporeal membrane oxygenation; V-PA ECMO, veno-pulmonary arterial extracorporeal membrane oxygenation; V-V ECMO, veno-venous extracorporeal membrane oxygenation; VV-PA ECMO, veno-venous pulmonary arterial extracorporeal membrane oxygenation. Figure 1.: Dual-lumen cannula for single-site V-PA ECMO. PA, pulmonary artery; RA, right atrium; V-PA ECMO, veno-pulmonary arterial extracorporeal membrane oxygenation.Figure 2.: Midesophageal RV inflow/outflow view in transesophageal echocardiography showing the course of a single-site dual-lumen RA-to-PA cannula (Protek-Duo TandemHeart cannula (CardiacAssist Inc., Pittsburgh, PA), blue arrows) through the RV and pulmonary valve entering the main PA. Ideally the tip of the cannula (not visible in this image) should lie before/at the PA bifurcation. L/N/R, aortic valve cusps, left, noncoronary, right respectively; LA, left atrium; PA, pulmonary artery; RA, right atrium; RVOT, right ventricular outflow tract; TEE, transesophageal echocardiography.Figure 3.: V-PA ECMO configurations. A: Single-site V-PA ECMO through dual-lumen RA-to-PA cannula; (B) Dual-site V-PA ECMO through femoral vein-to-PA cannulation. IVC, inferior vena cava; LA, left atrium; LV, left ventricle; ML, membrane lung; PA, pulmonary artery; RA, right atrium; RFV, right femoral vein; RIJV, right internal jugular vein; RV, right ventricle; SVC, superior vena cava; V-PA ECMO, veno-pulmonary arterial extracorporeal membrane oxygenation.Figure 4.: Chest radiographs showing different V-PA ECMO configurations. A, B: Single-site dual-lumen RA-to-PA cannula (ProtekDuo); the drainage lumens end in the RA (white arrows), with the tips and re-infusion lumens lying in the main PA (blue arrows). C, D: Dual-site cannulation through femoral drainage cannula and single-lumen pulmonary artery re-infusion cannula. DLPAC, dual-lumen pulmonary artery cannula; FC, femoral cannula; PA, pulmonary artery; PAC, pulmonary arterial cannula; RA, right atrium; V-PA ECMO, veno-pulmonary arterial extracorporeal membrane oxygenation.Figure 5.: Main features (advantages and disadvantages) of V-V and V-PA ECMO. ARDS, acute respiratory distress syndrome; COVID-19, coronavirus disease 2019; DLC, dual-lumen cannula; FV, femoral vein; IJV, internal jugular vein; IVC, inferior vena cava; LV, left ventricle; PA, pulmonary artery; RA, right atrium; RCT, randomized controlled trial; RV, right ventricle; V-PA ECMO, veno-pulmonary arterial extracorporeal membrane oxygenation; V-V ECMO, veno-venous extracorporeal membrane oxygenation.Evidence for Veno-Pulmonary Arterial ECMO in ARDS There is a lack of direct comparison randomized studies of V-PA ECMO versus other ECMO modalities. Current V-PA ECMO literature is limited to observational data (Table 1).23–32 A recent systematic review of five observational studies including 194 patients with coronavirus disease 2019 (COVID-19)–related ARDS receiving V-PA ECMO showed that application of V-PA ECMO, using single-site dual-lumen RA-to-PA cannula was associated with high survival rates.32 Veno-pulmonary arterial ECMO was also associated with reduced incidence of acute kidney injury (AKI) and need for renal replacement therapy (RRT).32 In all studies included in the review, V-PA ECMO was applied empirically irrespective of the presence of RV injury or not at the time of ECMO initiation, and the ECMO configuration used was mostly single-site dual lumen RA-to-PA.24–29 In the largest V-PA ECMO case series (n = 40), Mustafa et al.24,25 reported clinically important outcomes of patients with severe COVID-19–related ARDS requiring ECMO. The entire cohort received V-PA ECMO at the time of cannulation utilizing a single-site dual lumen (RA-to-PA) cannula.24,25 A bundle of interventions implemented in addition to V-PA ECMO included: spontaneous breathing, early liberation from invasive mechanical ventilation and rehabilitation, early corticosteroid therapy, inhaled pulmonary vasodilators, anticoagulants, and diuretics.24,25 The authors reported high survival rate with 88% of the cohort being extubated within 13 days of ECMO initiation.24,25 Each of the treatment bundle components (in particular the V-PA ECMO configuration and early extubation) have a theoretical RV- and pulmonary vascular-protective effect but it is unclear whether the observed outcome benefit can be attributed to prophylactic mechanical support of the pulmonary circulation or the bundle of mechanical circulatory, ventilator, and pharmacological measures.18,24,25 In a propensity score match analysis of patients with severe COVID-19–related ARDS, the same group of investigators found that empiric V-PA ECMO (initiated using a single-site, dual-lumen RA-to-PA cannula) was associated with a three-fold improvement in survival compared with maximum conventional ventilatory support alone.23,33 Saeed et al.27 compared V-PA (single-site, dual-lumen RA-to-PA cannula) versus V-V ECMO (dual-site cannulation or single-site dual lumen cannula with tip in IVC) in a multicenter retrospective cohort of patients with COVID-19-ARDS receiving ECMO (n = 435). The authors reported a reduced 90 day in-hospital mortality in the V-PA group compared with dual-site V-V ECMO cannulation after adjusting for clinical and center-related confounders.27 The V-PA group had longer time-to-liberation from ECMO but shorter time-to-liberation from invasive ventilation. The latter adds to the RV- and pulmonary vasculature-protective effect of this strategy and may explain the observed differences in survival rates. Similar results were observed in other retrospective cohorts.29 It should be noted that the role of V-PA ECMO has not been studied in non-COVID-19 ARDS patient populations. However, the prevalence of RV and pulmonary vascular injuries and their effects on outcomes are the same in COVID-19 and non–COVID-19-associated ARDS, and unless high quality emerging data indicate otherwise, management of ARDS of any etiology should follow current evidence-based practices.4,34,35 Veno-Pulmonary Arterial ECMO and ARDS Cardiorespiratory Phenotypes In the majority of V-PA ECMO studies, there is a lack of cardiovascular phenotypic characterization of the study population before application of ECMO.23–31 Patients who met ECMO criteria and required extracorporeal support received V-PA ECMO as a prophylactic measure and not based on their echocardiographic or pulmonary hemodynamic profile. Given the observational nature of available data and despite a signal of outcome benefit, there is a need to enhance the precision in individual delivery of V-PA ECMO in ARDS. It therefore remains uncertain whether empiric V-PA ECMO and mitigation of RV injury and pulmonary vascular dysfunction at the time of ECMO application actually improves patient-centered outcomes. In favor of this approach may be the fact that development of acute RV injury after initiation of V-V ECMO support is associated with a significant increase in mortality.20 Prophylactic mechanical support of the pulmonary circulation may therefore confer theoretical benefit allowing for early rehabilitation, potentially early liberation from invasive ventilation and mitigation of myotrauma.20,24,25,36 It is unknown whether personalization of V-PA ECMO to certain phenotypes of abnormal RV biomechanics in ARDS (e.g., isolated RV dilatation, RV dilatation with impaired function when the RV still meets flow demand or advanced RV injury/failure with circulatory shock) would improve outcomes.6 In a recent study published in the journal, Ivins-O’Keefe et al.31 described a case series of 21 patients with ARDS who developed acute RV failure despite standard V-V ECMO, requiring conversion to a V-PA ECMO configuration (dual-site cannulation with PA return). The median time from V-V ECMO initiation to conversion to V-PA ECMO was 12 (8.5–23.5) days, and survival to hospital discharge was only 25%.31 These data indicate that application of V-PA ECMO in advanced RV injury and despite V-V ECMO support may not significantly alter outcomes. Perhaps pre-emptive V-PA ECMO or application of V-PA ECMO in ARDS patients with RV injury phenotypes where the RV meets flow demand, that is, before RV failure and secondary organ injuries develop, would make more physiologic sense; this notion, however, should be supported by prospective randomized data. Characterizing RV and pulmonary vascular dysfunction before ECMO initiation using a multimodal approach (clinical examination, echocardiography, invasive pulmonary hemodynamic monitoring) aids in phenotyping and risk stratifying these patients. Given the risk of developing RV injury in ARDS and despite V-V ECMO support, close monitoring, for example, daily echocardiography should be considered in all ARDS patients receiving V-V ECMO potentially prompting an early intervention. Future Directions Acute respiratory distress syndrome mortality remains high increasing uptake of ECMO especially in resource rich settings. There is also significant variability in outcomes between centers and there appears to be a volume–outcome relationship, which was particularly stressed during the COVID-19 pandemic.37 It is important to draw a distinction between suboptimal outcomes due to deficiencies in overall ECMO processes and quality of care as opposed to suboptimal support of the pulmonary or systemic circulation in ARDS patients supported with V-V ECMO. Undoubtedly, ARDS patients who need additional mechanical circulatory support represent a more unwell subset and timely institution of support is critical to prevent multiple organ failure and death. Given the ease of application and available evidence, V-V ECMO will likely remain first line option in patients with severe ARDS who fail conventional mechanical ventilation. Future research must address two key questions: 1) Should V-PA ECMO be applied proactively in all patients with severe ARDS who fail conventional mechanical ventilation? 2) Should V-PA ECMO application be applied preferentially in V-V ECMO supported patients with severe ARDS who develop signs of RV injury? There may not be universal equipoise for the former approach at this stage and the risk to benefit ratios of either approaches need rigorous testing in clinical trials. The enthusiasm for RV support should not be at the expense of suboptimal lung protection due to an inability to achieve higher blood flows with dual lumen cannulae. The safe and most efficient V-PA ECMO perfusion configuration that allows higher blood flow rates to better optimize DO2/VO2 ratios also need to be defined in future studies. Acknowledgments The authors would like to thank Dr Kelly M. IvIns-O’Keefe (Department of Anesthesiology, Brooke Army Medical Center, Fort Sam Houston, TX) for providing one of the V-PA ECMO chest radiographs (dual-site, femoral vein-to-PA configuration).