Normalizing the Abnormal: Hypoxemia in Venovenous ECMO

To the Editor:

We read with great interest a recent letter to the editor in the Journal of Cardiothoracic and Vascular Anesthesia.¹ The authors brought to light the frequent disconnect in the understanding of arterial oxygen saturation, oxygen delivery (DO₂), and oxygen consumption (VO₂). The authors succinctly referenced the Extracorporeal Life Support Organization (ELSO) guidelines, which caution against the practice of using the ventilator to increase oxygen saturation to improve an arbitrary physiologic threshold in the absence of symptomatic hypoxemia (hypoxia).² The authors also addressed the risks of using cardiac output modifying agents (such as beta-blockers) to modulate oxygen saturation in patients on venovenous extracorporeal membrane oxygenation (VV ECMO). This article addressed a very important topic in VV ECMO that should be explored further, particularly given the globally rapidly increasing number of ECMO centers. Our center's experience very much agrees with the points made by the authors.

ELSO specifically makes a comment in the updated 2021 VV ECMO management guidelines on using the ECMO circuit over the ventilator, targeting plateau pressures ≤25 mmH₂O and accepting a low oxygen saturation in the interest of lung recovery, as long as metabolic demands are met.² Evidence suggests that minimizing plateau pressures and driving pressures may improve survival in the VV ECMO population.^{3 , 4} If the goal of VV ECMO is lung recovery, then reducing secondary ventilator-induced lung injury and self-inflicted lung injury must be the main priority. We routinely accept abnormal physiologic parameters in the presence of critical illness to improve survival, with well-known examples including permissive hypotension, permissive hypercapnia and hypoxemia, low hemoglobin transfusion thresholds, and blood glucose management.5, 6, 7, 8 We agree that hypoxemia on VV ECMO in the absence of lactate elevation, organ dysfunction, escalating doses of vasoactive medications, or myocardial suppression should not prompt a drastic intervention without a thoughtful investigation into the cause. That being said, we also believe that the etiology of reductions in oxygen saturation (such as from 90%-82%) should be elucidated to prevent clinically impactful declines (eg, mucus plugging, superimposed pneumonia, pneumothorax, and pulmonary embolism).

Maneuvers to decrease cardiac output, such as using phenylephrine as a first-line vasoconstrictor or beta-blockade in VV ECMO to increase oxygen saturation, are performed at numerous centers. Evidence of benefit for this practice is lacking, with the largest examination into this practice being performed by Bunge et al in 33 patients at 2 centers prior to the global COVID-19 pandemic.⁹ In this work, the authors found that with the initiation of beta-blockers, oxygen saturation increased, but 2 (approximately 7%) had to have the therapy reduced or withheld due to bradycardia.⁹ The authors of this retrospective review concluded that the practice of beta-blockade use was safe. However, a 7% incidence of adverse reactions can have a serious impact on outcomes in a large number of patients. Additionally, the fact that Bunge et al used arterial oxygen saturation rather than DO₂ as the efficacy measure of beta-blocker intervention further highlighted the conceptual significance of the article by the authors. As the authors astutely noted, the use of beta-blockades (which affects contractility prior to heart rate) in patients supported by VV ECMO, may worsen subacute right ventricular dysfunction and cause hemodynamic decompensation. Right ventricular dysfunction is common in acute respiratory distress syndrome (22% in one study by Boissier et al, and potentially even higher in COVID-19 acute respiratory distress syndrome), and beta-blockade in patients with cor pulmonale can precipitate significant hemodynamic compromise.^{10 , 11}

At our center, VV ECMO is managed with a multidisciplinary team of ECMO providers, and interventions to increase oxygen saturation in the absence of physiologic decline are not performed. During the lung rest period of the VV ECMO run, we use rest setting of a driving pressure of 10, a rate of 10, a positive end-expiratory pressure of 10, and a fraction of inspired oxygen of 30% in the majority of patients. Our VV ECMO cannulation strategy typically is either with a 21-Fr multistage venous drainage cannula from one femoral vein, terminating in the mid-to-upper right atrium, and a 17-Fr return cannula from the right internal jugular vein (femoral-internal jugular), or a 21- or 25-Fr multistage drainage cannula deployed from one femoral vein terminating in the mid-to-upper right atrium, along with a 20- or 24-Fr long cannula terminating in the superior vena cava (femoral-femoral). Our usual ECMO circuit flows generally are in the range of 4.0-to-5.0 L/min, and after ensuring circuit function and the absence of new pathology, we do not consider using beta-blockade, ventilator maneuvers, or changes in cannulation strategy so long as the patient is supported and the oxygen saturations are 80% or greater. We follow serial lactate as a marker of inadequate DO₂:VO₂ ratio, given that VO₂ calculation is not practical. We believe that avoiding both pharmacologic cardiac output modulation and aggressive mechanical ventilation settings during the VV ECMO lung rest period have contributed to the essentially nonexistent rate of right ventricular failure in our cohort. Although there may be certain patient populations, such as those with pulmonary hypertension, who require higher oxygen saturation goals during VV ECMO support, we fully agree with the position of the authors on avoiding aggressive ventilator settings and beta-blockade to augment arterial oxygen saturation at the potential expense of ventilator-induced lung injury and decreased oxygen delivery. We believe the principle of “tolerating a new normal” should be applied to hypoxemia in patients supported by VV ECMO.

Conflicts of Interest

None.