Harlequin effect

In peripheral veno-arterial extracorporeal membrane oxygenation, the oxygenated blood via the femoral cannula flows in a direction opposite to that of the native blood flow. As cardiac function recovers, the blood ejected from the left ventricle converges with the oxygenated, retrograde pump flow. The location of this flow convergence (watershed) zone is dependent on the amount of extracorporeal life support and myocardial function. If the cardiac function recovers but the pulmonary function continues to lag behind (examples: acute cardiogenic shock with pulmonary oedema, pneumonia, acute respiratory distress syndrome etc), or if ventilation is inadequate, the watershed zone migrates distally, resulting in upper body hypoxaemia (desaturation in coronary and cerebral circulation). ^{Reference Mossadegh, Faulkner, Brogan, Lequier, Lorusso, MacLaren and Peek2} This may impede weaning from extracorporeal life support and could delay myocardial recovery.

Multiple manoeuvres have been described to troubleshoot this problem to preserve coronary and cerebral oxygenation, including treatment of pulmonary pathology with optimisation of ventilatory settings, left ventricle venting in cases of cardiogenic pulmonary oedema, increasing the extracorporeal flow in cases of non-cardiogenic pulmonary oedema, using the axillary artery or subclavian artery or right common carotid as the inflow vessel, using veno-arterial-venous extracorporeal membrane oxygenation instead of traditional femoral veno-arterial extracorporeal membrane oxygenation, and placement of the femoral artery inflow cannula high in the thoracic aorta. ^{Reference Rao, Khalpey, Smith, Burkhoff and Kociol3}

To better identify the watershed area in patients receiving peripheral extracorporeal life support via femoral cannulation, it is prudent to place a right radial arterial line to monitor both the pulse pressure (narrow – watershed in the aortic root, wide – watershed distal to the brachiocephalic artery) and obtain an arterial blood gas sample that is representative of both the cerebral and myocardial oxygenation status. ^{Reference Rao, Khalpey, Smith, Burkhoff and Kociol3}

Can Harlequin effect occur in central veno-arterial extracorporeal membrane oxygenation?

Though this effect is classically explained in patients receiving peripheral veno-arterial extracorporeal membrane oxygenation via femoral cannulation, this phenomenon can potentially manifest in central veno-arterial extracorporeal membrane oxygenation for cardiac indications (and theoretically can also occur in central veno-arterial extracorporeal membrane oxygenation for respiratory indications).

Watershed phenomenon in post-cardiotomy central veno-arterial extracorporeal membrane oxygenation is seen in cases of severe right ventricular dysfunction in which there is an interatrial communication created for offloading the right heart, as in Ebstein’s anomaly and tetralogy of Fallot repair. Figure 1 depicts arterial blood gas analysis, sampled simultaneously from right radial and femoral arterial lines in a patient with central veno-arterial extracorporeal membrane oxygenation post-cone repair and atrial septal defect creation.

Figure 1. Arterial blood gas (ABG) analysis from a post-operative patient with Ebstein’s anomaly, who underwent cone repair with creation of an atrial septal defect to decompress the right heart, while being weaned from central veno-arterial extracorporeal membrane oxygenation (VA-ECMO) instituted for right ventricular (RV) dysfunction (( a ) Femoral artery sample, ( b ) Right radial artery sample). Lower saturation and partial pressure of oxygen (pO2) in the right radial arterial line indicate RV dysfunction. Improvement in the RV function decreases the pO2 gradient in the ABG.

Unlike the Harlequin effect of peripheral extracorporeal life support, this phenomenon is most probably caused by the tendency of the fluid jet to stay attached to a convex surface (Coanda effect). In human ontogeny, the blood which is ejected from the left ventricle preferentially perfuses the coronary and cerebral circulation and then other parts of the body through the aortic arch.

In post-cardiotomy central veno-arterial extracorporeal membrane oxygenation scenarios, the same aortic inflow cannula which is used for running cardiopulmonary bypass during surgery is usually used for extracorporeal life support (the site being typically opposite to the base of the brachiocephalic artery, with the tip directed towards the aortic arch). With full flow, the right atrium is maximally emptied and trans-pulmonary flow is minimal. The arterial inflow cannula supplies oxygenated blood to the systemic circulation. When at basal ventilatory settings and full extracorporeal flows, pulmonary venous return to the left atrium is minimal, but is subsequently ejected by the left ventricle into the aorta, resulting in near similar partial pressure of oxygen or saturations in the radial and femoral arterial samples.

While weaning the patient from post-cardiotomy central veno-arterial extracorporeal membrane oxygenation, in the presence of interatrial communication, and recovered right ventricular function, right to left shunting across the interatrial communication will be minimal, with good trans-pulmonary flow (Fig 2). With optimum pulmonary mechanics, oxygenated blood returning to the left atrium gets ejected by the left ventricle into the aorta along with competitive oxygenated flow from arterial inflow cannula, resulting in near similar saturations or partial pressure of oxygen in the radial and femoral samples.

Figure 2. Representation of a post-cardiotomy central veno-arterial extracorporeal membrane oxygenation (VA-ECMO) for severe right ventricular (RV) dysfunction in the presence of interatrial communication created to decompress the right heart. ( a ) Patient on full extracorporeal life support. ( b ) Patient in weaning mode in the presence of improving RV function. There is good trans-pulmonary flow. In the presence of adequate ventilation and pulmonary function, pulmonary venous return is oxygenated and is subsequently ejected by the left ventricle (LV) into the aorta. This blood mixes with the oxygenated blood of aortic inflow cannula, causing minimal or no difference in pO2 or saturation between brachiocephalic artery (BCA) territory and rest of the body. ( c ) Patient in weaning mode in the presence of RV dysfunction leads to right to left shunting at the atrial level. This in combination with decreased pulmonary venous returns to the LA and accentuates the amount of desaturated blood that is subsequently ejected by the LV into the aorta. The BCA territory therefore receives desaturated blood due to Coanda effect, and the rest of the body receives this desaturated blood mixed with the oxygenated blood from the aortic inflow cannula, causing a zone of differential saturation at the level of the BCA origin.

However, if the right ventricular dysfunction persists while weaning extracorporeal life support, there will be right to left shunting across the interatrial communication depending on the degree of right ventricular dysfunction. Concurrently, the trans-pulmonary flow will be suboptimal, in spite of adequate ventilation and pulmonary function. The combination of right to left shunting at the atrial level and decreased pulmonary venous return to the left atrium leads to deoxygenated blood entering the left ventricle, to be ejected into the ascending aorta, brachiocephalic artery, and the aortic arch (Fig 2). In the arch, the deoxygenated blood mixes with the oxygenated blood from the arterial inflow cannula resulting in differential saturation between brachiocephalic artery territory and rest of the body.

Case report

A 51-year-old female (50 kg and 155 cm) presented with NYHA class III dyspnoea without cyanosis and was diagnosed on echocardiography as Carpentier’s type-B Ebstein’s anomaly, with severe tricuspid regurgitation, smallish functional right ventricle, large secundum atrial septal defect, and preserved biventricular function. The patient was in atrial fibrillation with a controlled ventricular rate. Preoperative coronary angiogram demonstrated both left main and right coronary artery arising from left coronary sinus, with separate ostia and left dominant circulation. Pulmonary artery pressure was 22/10 mmHg (mean of 14 mmHg), left ventricular end-diastolic pressure was 10 mmHg. Calculated Qp/Qs was 1.4, and indexed pulmonary vascular resistance was 1.6 wood unit/m². Arterial blood gas analysis at room air from an aortic sample showed a partial pressure of oxygen of 53.8 mmHg and saturation of 90%. The patient underwent cone reconstruction of tricuspid valve, leaving a small (6 mm) atrial septal defect (for anticipated right ventricular dysfunction) and right atrial volume reduction plasty with a bypass time of 140 minutes and cardioplegic arrest of 110 minutes. Post-operative transesophageal echocardiogram demonstrated trivial to mild tricuspid regurgitation and biventricular dysfunction. In view of echocardiographic findings and persistent atrial arrhythmia (flutter and fibrillation), the chest was left open, and the patient was shifted to intensive care unit with amiodarone, adrenaline, and dobutamine infusions.

The chest was closed within 24 hours, but she continued to be haemodynamically labile with persistent atrial flutter, needing escalation of amiodarone. She was extubated 54 hours post-operatively to high flow nasal cannula (20 L flow with Fio2 50%), but continued to deteriorate needing escalation of vasoactive agents. There were signs of sepsis and a blood culture grew Ralstonia mannitolilytica (order: Burkholderiales), which, combined with low cardiac output status, necessitated re-intubation and mechanical ventilation. On the 7^th post-operative day, due to rising vasoactive-inotropic score and increasing features of low cardiac output, an emergency bidirectional cavopulmonary shunt procedure was performed to offload the severely dysfunctional right ventricle, under appropriate antibiotic cover on cardiopulmonary bypass without a cardioplegic arrest. Elective central veno-arterial extracorporeal membrane oxygenation was initiated using a 20F inflow cannula in the ascending aorta (EOPA™, elongated one-piece arterial cannula, Medtronic, Minneapolis, MN, USA), and a 32F drainage cannula in the right atrium (DLP™, David, Lynda and Philip single stage venous cannula, Medtronic, Minneapolis, MN, USA) due to inability to separate from cardiopulmonary bypass.

An initial attempt to wean on day-2 of extracorporeal life support failed due to persistent severe right ventricular dysfunction. The arterial blood gas analysis in Figure 1 was during the initial failed weaning attempt, at around 40 hours of extracorporeal membrane oxygenation. We escalated the strategies to manage the post-operative right ventricular dysfunction, which included decreasing right ventricular pre-load, increasing coronary perfusion, increasing right ventricular function / contractility, decreasing right ventricular afterload, improving myocardial oxygen delivery and optimising cardiac rhythm. ^{Reference Varma, Srimurugan, Jose, Krishna, Valooran and Jayant4}

In our patient, right ventricular pre-load was decreased by the presence of an atrial septal defect (shunting right to left), creation of bidirectional cavo pulmonary connection, and aggressive diuresis (furosemide and metolazone). Coronary perfusion pressure was increased by raising the systemic pressure, aided by inotropic agents. To wean the patient from extracorporeal life support, right ventricular contractility was supported by levosimendan (improves right ventricular systolic and diastolic function and causes pulmonary and systemic vasodilatation) and inodilator (dobutamine). Heart rate was controlled by amiodarone. Adrenaline is inotropic, chronotropic, and proarrhythmogenic, and increases the myocardial oxygen demand. However, weaning from this agent was not clinically tolerated due to its effect on increasing systemic pressure and ventricular contractility.

Right ventricular afterload was reduced by optimising ventilation and use of pulmonary vasodilators (inhaled nitric oxide and intravenous sildenafil infusion).

With the above-mentioned strategy, the gap between radial and femoral artery oxygen saturation and partial pressure of oxygen became minimal, and a second attempt of weaning done on day-4 of extracorporeal life support was successful, with a total duration of extracorporeal membrane oxygenation being 84 hours and 26 minutes. Extracorporeal membrane oxygenation was discontinued with minimal inotropic support, and the patient was extubated 4 days after the weaning. She was discharged on the 30^th post-operative day. Due to persistent atrial flutter on follow-up, an electro-physiological study 2 months post-surgery revealed a counterclockwise cavotricuspid isthmus-dependent atrial flutter and scar re-entrant atrial flutter. Successful radio frequency ablation was performed via the inferior caval vein route, and she was discharged in a stable condition. The patient is on follow-up since then (18 months post-surgery) with improvement in symptoms and overall functional capacity, in sinus rhythm, with echocardiography demonstrating good left ventricular function, mild right ventricular dysfunction, and mild tricuspid regurgitation.

Clinical importance

1. 1. Harlequin effect can manifest in central veno-arterial extracorporeal membrane oxygenation, with the watershed area usually at the level of the brachiocephalic artery origin.

2. 2. Whenever post-cardiotomy veno-arterial extracorporeal membrane oxygenation is initiated for refractory low cardiac output syndrome secondary to right ventricular dysfunction, in a patient with an interatrial communication created for right-sided decompression, it is recommended to secure a right radial arterial line, during weaning from extracorporeal life support. Arterial blood gas analysis from radial arterial line mirrors cerebral and myocardial oxygenation, where the amount of desaturation is proportional to the degree of right ventricular dysfunction. The partial pressure of oxygen and saturation difference between the right radial and femoral arterial lines will aid in decision-making to wean the patient from extracorporeal life support.

3. 3. In this scenario, differential cyanosis is unlikely to cause profound cerebral hypoxia as the left carotid artery will perfuse the cerebral circulation with oxygenated blood, but can cause myocardial hypoxia, and can worsen myocardial function, ending in a vicious cycle.