Share this article on:

Endogenous Carbon Monoxide Production: A Rare and Detrimental Complication of Extracorporeal Membrane Oxygenation

Hermans, Greet*; Wilmer, Alexander*; Knockaert, Daniel*; Meyns, Bart†

doi: 10.1097/MAT.0b013e318185e1e6
Case Report

Extracorporeal membrane oxygenation (ECMO) is a technique that can support gas-exchange and cardiac function in patients with acute respiratory or cardiac failure that is not responsive to conservative treatment. ECMO is a high-risk procedure in critically ill patients and both technical and patient-related complications frequently occur. We report on a patient with end-stage pulmonary fibrosis (histiocytosis ×), in whom ECMO was used as bridge to urgent lung transplantation. Respiratory insufficiency necessitating ECMO therapy was due to spontaneous pneumothorax with bronchopleural fistula. Access was initially made by femoral veno-venous canulation. Due to right heart failure, access was switched to veno-arterial using the two existing canulae as efferent system and implanting a third canula on the femoral artery using a graft. Despite marked hemodynamic improvement after this intervention, high flow rates creating high premembrane pressures were required to ensure oxygenation. High levels of carboxyhemoglobin (COHb) occurred, most likely due to massive mechanical hemolysis in the ECMO circuit. This may have had detrimental effects by further complicating tissue oxygenation. We recommend that COHb should routinely be checked in ECMO patients.

From the *Department of General Internal Medicine, Medical Intensive Care Unit; and †Department of Cardiac Surgery, University Hospitals, Leuven, Belgium.

Submitted for consideration April 2008; accepted for publication in revised form May 2008.

Correspondence: Greet Hermans, MD, Department of Internal Medicine, Medical Intensive Care Unit, University Hospitals Leuven, Herestraat 49, B-3000 Leuven, Belgium. Email:

Back to Top | Article Outline

Case Report

A 39-year-old patient with end-stage pulmonary fibrosis due to histiocytosis X, was already on the urgent transplantation list when he developed acute respiratory insufficiency due to a spontaneous pneumothorax, for which he was urgently intubated and a chest drain was placed. Subsequently, he was transferred to our Intensive Care Unit (ICU). Over the next 24 hours, the patient developed severe hypoxia (paO2/FIO2 ratio 39 mm Hg) with prominent air leakage despite good expansion of the lung. At this point, extracorporeal membrane oxygenation (ECMO) was started by femoral veno-venous access (using a vena femoralis inflow cannula of 29 French and an outflow vena femoralis canula 27 French, both Medtronic Biomedicus Carmeda coated cannulae), which markedly improved oxygenation (paO2/FIO2 ratio 119 mm Hg). We used a centrifugal pump; Biomedicus, Medtronic, MN, and oxygenator Medos Hilite7000LT. The whole circuit was Medtronic Carmedo coated. On day 5 after ICU admission, the patient developed a septic shock complicated by ischemic hepatitis, treated with fluids, vasopressor drugs and antibiotics. Over the next 24 hours, the patient’s condition improved and vasopressor therapy was stopped. Therapy with inhaled nitric oxide (NO) was started because of significant pulmonary hypertension which improved oxygenation (PaO2/FIO2 rose from 54 to 67 mm Hg with the same ventilatory and ECMO settings). On day 9 of ICU stay, the patient developed a cardiogenic shock requiring massive doses of inotropic and vasopressor drugs due to dilatation of the right ventricle with compression of the left ventricle. Therefore, the ECMO access was switched to a veno-arterial setting using the two existing canulae as the efferent system and implanting a third canula (21 French, Medtronic Biomedicus Carmeda coated) on the femoral artery using a graft. Inotropics and vasopressors were discontinued quickly after this intervention. However, during the following days, high flow rates creating high premembrane pressures were required to ensure adequate oxygenation, which resulted in significant hemolysis. Carboxyhemoglobin (COHb) levels rose up to 4.4% on day 13 of ICU stay, which urged us to reduce and eventually stop NO therapy. However, the COHb level continued to rise up to 9.5%. On day 18, the patient was transplanted, still on ECMO, but did not survive the operation due to massive pleural bleeding and hemorrhagic shock.

Back to Top | Article Outline


Carbon monoxide (CO) is an exogenous poison that binds hemoglobin with a much higher affinity than oxygen and, therefore, displaces oxygen from hemoglobin. At the same time, it increases the affinity of the remaining sites for bound oxygen. This results in a reduced oxygen transport capacity and decreased release of oxygen to the tissues. It is, however, less known that CO is produced endogenously during the normal catabolism of hemoglobin. Hemoxygenase (HO) enzymes break down heme molecules into free iron, biliverdine and CO. HO enzymes exist in different isoforms: inducible HO-1, and constitutive HO-2 and HO-3. HO-1 is induced by its substrate heme and a number of other agents that cause oxidative stress (e.g., hypoxia, hyperoxia, acute lung injury, sepsis, and NO).1–3

The levels of COHb of up to 9.5% in this patient clearly exceeded the slightly increased levels (0.8%–2%) that have been reported in critically ill patients.1,4,5 Therapy with NO has been associated with increased levels of COHb due to induction of HO by NO and production of methemoglobin, releasing heme.2 Therefore, therapy with NO was reduced and subsequently stopped. Despite this intervention, COHb levels kept rising. Importantly, significant hemolysis became apparent after switching the ECMO access route from veno-venous to veno-arterial, reflected by increased levels of bilirubin, lactate dehydrogenase (LDH) and free hemoglobin (Figure 1). As evidenced by an excess of fragmentocytes, hemolysis was due to mechanical trauma, a well-known complication of ECMO,6 that is determined by shear forces, turbulence and blood flow velocity. Blood clots in the circuit can also cause hemolysis.7 This was ruled out by daily checks of the circuit. The patient was 1.84 m and weighed 76 kg. The cannulae were appropriately sized for this patient, and therefore unlikely to have been a cause of hemolysis. The gradual increase in ECMO flow needed to maintain adequate oxygenation resulted in very high premembrane pressures and worsened hemolysis. COHb levels paralleled the increase in hemolysis parameters. As heme is a potent inductor of HO, we concluded that the massive hemolysis most likely caused the elevated COHb levels. Moreover, the slight decrease in ECMO flow noticed during day 17 of ICU stay, coincided with lower mean premembrane pressures and a small decrease in COHb on day 18.

Other factors such as sepsis and oxidative lung stress may also contribute to COHb production. However, it seems unlikely that these were major determinants in this case, as the first episode of sepsis was quickly controlled, and a second inflammatory episode occurred on day 15 after ICU admission, after a significant rise in COHb had already been documented. Oxidative stress in the lung was already present long before the rise in COHb. Transfusion of packed cells is another potential direct source of COHb.8 A random check of administered packed cells revealed the absence of detectable COHb levels. Therefore, the high level of 9.5% in this patient was unlikely the result of transfusion.

The presence of up to 9.5% COHb might have been detrimental in this patient as it obviously reduced oxygen transport capacity and tissue oxygenation.2 The cause of the massive hemolysis remains somewhat elusive. High postmembrane pressures and a normal pressure drop over the membrane indicate that there was no oxygenator problem. Postmortem evaluation of this case by the ECMO team suggested that the increasing resistance in the afferent system might have resulted from the implantation of the cannula at the graft and reintervention might have been beneficial.

Back to Top | Article Outline


We report on a patient who developed significant levels of COHb, most likely caused by massive mechanical hemolysis due to ECMO. This may be detrimental in patients with an already limited tissue oxygen supply. We suggest that COHb should be routinely checked in patients treated with ECMO. If a rise is detected and mechanical hemolysis is obvious, careful analysis of the pressures at different sites of the circuit should enable physicians to locate the problem and act accordingly.

Back to Top | Article Outline


1. Scharte M, Bone HG, Van Aken H, Meyer J: Increased carbon monoxide in exhaled air of critically ill patients. Biochem Biophys Res Commun 267: 423–426, 2000.
2. Rusca M, Oddo M, Schaller MD, Liaudet L: Carboxyhemoglobin formation as an unexpected side effect of inhaled nitric oxide therapy in severe acute respiratory distress syndrome. Crit Care Med 32: 2537–2539, 2004.
3. Morse D, Choi AMK: Heme oxygenase-1—The “emerging molecule” has arrived. Am J Respir Cell Mol Biol 27: 8–16, 2002.
4. Meyer J, Prien T, Van Aken H, et al: Arterio-venous carboxyhemoglobin difference suggests carbon monoxide production by human lungs. Biochem Biophys Res Commun 244: 230–232, 1998.
5. Morimatsu H, Takahashi T, Maeshima K, et al: Increased heme catabolism in critically ill patients: Correlation among exhaled carbon monoxide, arterial carboxyhemoglobin and serum bilirubin IX{alpha} concentrations. Am J Physiol Lung Cell Mol Physiol 290: 114–119, 2005.
6. Paul R, Apel J, Klaus S, et al: Shear stress related blood damage in laminar couette flow. Artif Organs 27: 517–529, 2003.
7. Steinhorn RH, Isham-Schopf B, Smith C, Green TP: Hemolysis during long-term extracorporeal membrane oxygenation. J Pediatr 115: 625–630, 1989.
8. Ehlers M, McCloskey D, Devejian NS: Alarming levels of carboxyhemoglobin in a unit of banked blood. Anesth Analg 97: 289–290, 2003.
Copyright © 2008 by the American Society for Artificial Internal Organs