Tailored ECMO for Respiratory Failure: Moving From Off-the-Rack to Bespoke

In a study of the Extracorporeal Life Support Organization (ELSO) registry in the current issue of ASAIO Journal,1 Nguyen et al.1 performed a comparative analysis of between patients treated for refractory coronavirus disease (COVID)-19 related acute respiratory distress syndrome (ARDS) with either veno-venous (V-V) ECMO, and patients treated with V-V ECMO that subsequently required conversion to veno-arterial (V-A) ECMO or another hybrid variant. The subsequent rates of complications and in-hospital mortality between those two groups were examined. Markedly higher rates of morbidity and mortality in the conversion group were identified, and are consistent with prior observations in other case series and registry data sets comparing outcomes between V-V ECMO and other ECMO platforms. It stands to reason from an actuarial perspective that patients with refractory respiratory failure but preserved cardiac function and their counterparts with mixed cardiorespiratory failure would have different risks for death, all other things being equal. But to what mechanism(s) can we ascribe the sizable separation in mortality risk among the patients observed in this study who were converted from V-V ECMO to a more invasive strategy? Complications from the mechanical differences in support themselves? Progression of disease severity or even the presence of two altogether different disease states? Or does the mortality difference owe at least in part to delayed recognition of an opportunity to make necessary adaptations to the support platform? Put another way, was the initial treatment strategy less well-suited for the group who was just a sicker or more advanced form of the other, which then contributed to or perhaps even outright led to the disparity in outcomes between them? The answers are of course not knowable from a retrospective analysis, and that remains as a primary challenge to determining how these data might impact our practice moving forward. Veno-venous ECMO is by many standards, including the ELSO guidelines and the American Association for Thoracic Surgery (AATS) guidelines for the use of mechanical circulatory support in lung transplantation, the recommended initial mechanical support platform for refractory respiratory failure.2,3 Limitations to V-V ECMO include recirculation and inadequate support in both high (insufficient volumetric flow rate relative to cardiac output) and low cardiac output (lack of required cardiac support) states. Alternatively, V-A ECMO offers membrane lung support as well as mechanical support of acute right ventricular (RV) failure and shock, which has been used to treat combined respiratory and RV failure. However, it may also contribute to 1) maladaptive cardiopulmonary effects4 such as a) left ventricular (LV) distension (occurring due to inadequate LV volume unloading in the setting of LV afterload mismatch—low LV contractility relative to high LV afterload) and b) differential hypoxemia (occurring due to poor intrapulmonary gas exchange with inadequate LV volume unloading, in the absence of LV afterload mismatch); and 2) higher risks of both bleeding and thrombotic/thromboembolic complications. These and other limitations have prompted broader utilization of hybrid systems. These include VV-A and V-AV ECMO, which carry the disadvantages of V-A ECMO, but from which subsequent conversion to V-V ECMO may be undertaken easily if/when cardiac function improves. Other hybrid approaches provide RV mechanical assistance and have similar risk profiles as V-V ECMO and are increasingly used to treat respiratory failure. These so-called “oxy-RVAD” platforms most notably include percutaneous veno-pulmonary (V-PA) ECMO. Although no randomized head-to-head comparisons of these various configurations are available, V-PA ECMO may confer distinct advantages in some instances of refractory respiratory failure with isolated RV compromise (realized or anticipated) including decreased RV wall tension and reduced recirculation fraction with drainage and return channels separated at least in principle by the tricuspid and pulmonary valves, and reduced rates of AKI, presumably due to augmented venous decongestion.5 There are also distinct risks using oxy-RVAD systems, including high rates of continuous and only extrinsic pump-controlled blood flow into the pulmonary arterial system, which can augment pulmonary capillary and venous pressures excessively, thereby generating additional extravascular lung water and even pulmonary hemorrhage, particularly in the fragile state of ARDS. To offset some of that risk, a novel approach to V-PA ECMO for the treatment of COVID-related ARDS was recently described using the Protek Duo (Livanova, London, UK) dual lumen single cannula, in conjunction with a separate drainage cannula, as a single return limb with split flow through both the RA and PA channels of the device, thus allowing for modulation of the fraction of flow seen directly by the PA.6 Resultant or exacerbated tricuspid valve (TV) regurgitation has also been described as a rate-limiting factor to the net impact on forward flow when these cannulae are in place traversing the valve apparatus, and must be evaluated particularly when the measured flow rates through them are not commensurate with the expected degree of venous decongestion that should be afforded by them. In the current study, the link between pre-ECMO cardiac arrest, the use of certain inotropic and vasoactive support agents with the eventual conversion from V-V ECMO to hybrid or VA ECMO, and the indication for renal replacement therapy, strongly suggest that a substantial fraction of those patients studied were already manifesting features of concomitant cardiac failure. Indeed, depending on the definitions used, a 10–25% coincidence of RV failure in ARDS has been reported.7 However, without discrete physiologic measurements from echocardiography or catheterization, physiologic profiling is challenged. The incidence of COVID-19 related LV dysfunction is well-described, and could reasonably be expected to be encountered with some frequency in a survey of patients this large. In the setting of LV failure, mechanical RV support with ECMO could exacerbate LV volume overload, pulmonary edema, poor lung mechanics, and intrapulmonary gas exchange. In these instances, other strategies including, but not limited to those above, may be used.8 We broadly agree with the authors that precannulation cardiac arrest is rarely if ever best supported with V-V ECMO as the mode of mechanical support. It is our view that the concept of a standard approach to using V-V ECMO for severe refractory respiratory failure, COVID-19 related ARDS or otherwise, is inherently limited—something like a suit bought “off the rack” with standard rather than specific measurements. Dividing heart and lung function can be reasonable for the purpose of organizing our understanding of the physiology and relevant anatomy in the abstract, but the intimate interplay between the two systems cannot be overstated in clinical decision-making, particularly and perhaps most urgently in states of refractory system failure. In addition to addressing what is often the most forefront issue of restoring adequate gas exchange, developing an initial strategy for mechanical support of severe respiratory failure must include evaluations of 1) biventricular morphology and function, and trends of any used vasoactive or inotropic support infusions, 2) the presence of any other physiologically important structural heart disease, 3) the acuity of the disease state and the potential for RV adaptation (i.e,. has the RV been strained by abnormal loading conditions for many hours, weeks, or years?), which can impact the likelihood of response to standard measures for RV support without a device, 4) the development of substantial venous congestion manifesting as climbing liver function test values or acute kidney injury, which in advanced states of dysfunction can themselves become precluding factors to organ transplantation as a destination option, and 5) opportunities or pathways to earlier intermediate goals such as airway management (ideally lifting sedation and extubation) and early mobilization (avoiding or transitioning away from femoral access points). Furthermore, those evaluations should be continued closely to ensure that changes are promptly recognized and that appropriate adaptations are made. Perhaps most importantly, Nguyen et al.1 suggest—even if they cannot clearly demonstrate it—that making the best choice regarding initial cannulation strategy is important, and that upgrading support may not be without costs. We recommend, as much as possible, an up-front bespoke approach, rather than “off-the-rack” with subsequent alterations.