Preoperative initiation of peripheral veno-arterial extracorporeal membrane oxygenation for a complex case of cardiac tamponade

Abstract

In this case report, we present an alternative approach to the anaesthetic management of patients presenting with delayed postoperative cardiac tamponade physiology. Given that pericardiocentesis was deemed unsafe, and a protracted surgical dissection was anticipated, peripheral veno-arterial extracorporeal membrane oxygenation (VA-ECMO) support was established prior to induction of anaesthesia to prevent catastrophic circulatory failure. To the best of our knowledge, this is the first reported case of planned preoperative commencement of peripheral VA-ECMO in a complex case of cardiac tamponade. We discuss the challenges associated with this case and the process for selecting this strategy. We also describe the role of transoesophageal echocardiography in planning the surgical approach. This report is completed by a discussion on the topic of delayed postoperative pericardial effusion and tamponade.

Background

The occurrence of postoperative pericardial effusion (PPE) after cardiac surgery is well documented. Although small and inconsequential effusions are common, late PPE (occurring after postoperative day 15) may be present in up to 22% of patients at postoperative day 20.¹ Only a minority progress to cardiac tamponade (CT), with an estimated incidence of 4% at postoperative day 30.¹ Surgical intervention may be required in up to 6.2% of late PPE cases, of which in 4.1% the main indication is tamponade.²

Preoperative initiation of extracorporeal membrane oxygenation (ECMO) for circulatory support prior to induction of anaesthesia is uncommon but is occasionally employed in certain circumstances where induction poses a high risk of precipitating cardiac arrest/hypoxia.^{3 4} Notably, very little evidence exists regarding the preoperative initiation of peripheral veno-arterial ECMO (VA-ECMO) in patients prior to surgical treatment of tamponade.

Case presentation

A man in his 60s (body mass index 26.4, body surface area 1.95 m²) was transferred from another hospital for the management of loculated CT. His cardiac history was relevant for coronary artery disease, moderate calcific aortic valve stenosis and persistent foramen ovale (PFO) for which he underwent coronary artery bypass grafting (CABG) (saphenous vein graft to the posterior descending artery, radial artery graft to the first obtuse marginal artery and left internal mammary artery graft to the left anterior descending artery), 23 mm bioprosthetic aortic valve replacement and PFO closure 18 weeks prior. Surgery and initial perioperative course were uneventful.

Fourteen weeks after the initial surgery, he presented to his local hospital with progressive dyspnoea and fatigue. Transthoracic echocardiography (TTE) demonstrated satisfactory placement and function of the bioprosthetic aortic valve with no paravalvular leak, preserved biventricular function and no evidence of pericardial effusion. CT pulmonary angiogram imaging demonstrated bilateral, subsegmental pulmonary emboli (PE). Therapeutic anticoagulation with apixaban was initiated and he was discharged home after 3 days.

Four weeks later, he re-presented with progressive dyspnoea and a single syncopal episode. On admission, his examination and vitals were unremarkable. However, TTE imaging demonstrated a large posterior pericardial effusion (measuring 5×7 cm) and compromised diastolic left ventricle filling. On discussion with the authors’ institution, apixaban was discontinued and he was transferred (approximately 48 hours after the initial presentation) for consideration of drainage of the effusion. On arrival, he was haemodynamically stable, vitals were heart rate (HR) 101 beats/min, blood pressure (BP) 100/72, respiratory rate (RR) 18 breaths/min, SpO₂ 98% on 4 L oxygen via nasal cannula.

Investigations

Repeat TTE on arrival confirmed a posterior pericardial effusion with the largest dimension of 5.2 cm and CT imaging showed bilateral pleural effusions and pulmonary oedema.

Treatment

Following consultation with cardiology and interventional radiology, percutaneous drainage was deemed unsafe due to the posterior location of the effusion; therefore, the patient was listed for an elective surgical pericardial window via left thoracotomy the following day. Within hours, the patient’s symptoms and haemodynamic status deteriorated significantly: HR 113 beats/min, BP 82/55 mm Hg, RR 33 breaths/min and SpO₂ 87–92% maintained with high-flow nasal cannula delivering 50 L/min (HFNO) (fractional inspired oxygen (FiO₂) 0.6). This clinical deterioration prompted immediate transfer to the operating room (OR) for emergency drainage.

On arrival to the OR, an obstructive shock state was apparent. The patient was sitting upright and unable to tolerate recumbent positioning due to respiratory distress. Significant pulsus paradoxus was evident on arterial pressure monitoring as well as transient loss of ventricular ejection when the patient coughed. Due to the patient’s evolving severe tamponade physiology, it appeared unlikely that he would tolerate induction of anaesthesia, positive pressure ventilation and prolonged one-lung ventilation without cardiovascular collapse occurring. Considering this, it was decided to initiate peripheral VA-ECMO pre-induction to maintain adequate end-organ perfusion and to facilitate safe performance of the procedure.

Standard monitoring, bispectral index and cerebral near-infrared spectroscopy (NIRS) were employed. A central venous catheter and a pulmonary catheter introducer sheath were placed through the right internal jugular vein under local anaesthesia with the patient in semirecumbent position (approximately 30°) receiving HFNO (60 L/min FiO₂ 1). Norepinephrine (maximum dose 0.2 µg/kg/min) was used to maintain systemic BP perioperatively. As pulmonary oedema was already present, no pre-induction fluid was administered. VA-ECMO cannulation of the left femoral artery and vein was achieved via direct cut down technique under local anaesthesia and minimal sedation with the patient in semirecumbent position. The drainage cannula was advanced into the inferior vena cava (IVC) and the return cannula advanced into the descending aorta (appropriate placement was subsequently confirmed with transoesophageal echocardiography (TOE)). ECMO flows were gradually increased up to 3 L/min (1.54 L/min/m²) leading to an increase in systemic pressure, a reduction in tachycardia and subjective improvement of the patient’s dyspnoea.

Anaesthesia was induced in semirecumbent position using titrated doses of fentanyl 200 µg, midazolam 4 mg, ketamine 50 mg and rocuronium 100 mg. A mean arterial pressure of 60 mm Hg and above was maintained throughout. A 39 Fr left-sided double-lumen endotracheal tube was placed and single-lung ventilation was tolerated with no cardiorespiratory compromise. Cerebral oximetry values remained above baseline values throughout the procedure. Post-induction TOE revealed a large loculated effusion overlying the entire free wall of the left ventricle, leading to compression of the left ventricle, significantly reducing left intraventricular volume and causing left ventricular diastolic collapse. The resulting displacement of the left ventricle also indirectly compressed the right ventricle, with significant reduction in right intraventricular volume (figure 1 and online supplemental videos 1–7).

Figure 1.

(A) Mid-oesophageal four-chamber view, pre-drainage. (B) Transgastric short-axis view, pre-drainage. Both images depict left ventricular diastolic collapse and indirect right ventricular compression through the displacement of the left ventricle and the interventricular septum. (C) Mid-oesophageal long-axis view, pre-drainage. There is left ventricular diastolic collapse. (D) Mid-oesophageal aortic valve long-axis view, pre-drainage. There is systolic anterior motion of the anterior mitral valve leaflet and a functional bioprosthetic aortic valve. The effusion is marked by a star.

An anterolateral thoracotomy was performed, and a pericardial window was created; the pericardial entry point was identified using TOE and was more posterior than was originally anticipated (figure 2 and online supplemental video 8). Dissection was difficult and prolonged due to adherence of the left lung to the pericardium. Sanguineous pericardial fluid 600 mL was drained. Post-drainage, transient hypokinesia of the right ventricle and of the inferior segments of the left ventricle were apparent, along with ST-segment depression on ECG, which was diagnosed as pericardial decompression syndrome (PDS) with a differential diagnosis of acute kinking of previous coronary graft (figure 3 and online supplemental videos 9–11). Epinephrine (0.05 µg/kg/min) was added to improve contractility and inhaled nitric oxide added at 10 ppm to reduce the right ventricular afterload. Red cell concentrate (one unit) and solvent/detergent plasma (four units) were transfused as required to maintain haemoglobin concentration and to optimise post-drainage preload. VA-ECMO was discontinued uneventfully after a total duration of 2 hours and 26 min. A left pleural drain and a pericardial drain were left in situ.

Figure 2.

Mid-oesophageal 0° view, centred on the effusion, pre-drainage. The yellow arrow marks the site for the pericardial window, identified by palpation.

Figure 3.

(A) Mid-oesophageal four-chamber view, post-drainage. (B) Transgastric short-axis view, post-drainage. The left ventricular diastolic filling is improved. Of note, there is ST-segment depression on the ECG suggestive of pericardial decompression syndrome or graft kinking.

Outcome and follow-up

The patient was transferred to the intensive care unit where he spent 72 hours. He was weaned from sedation and successfully extubated 17 hours later. Minimal vasopressor support was required during the first 12 hours of critical care stay. Repeat TTE demonstrated recovery of right and left ventricular function, left ventricular ejection fraction of >55% and no pericardial effusion. The pleural and pericardial drains were removed on postoperative days 7 and 9, respectively, when daily drainage volume was minimal. Oral anticoagulation was restarted, and he was clinically well at time of discharge home after a 10-day hospital stay.

Discussion

The definition of delayed PPE±CT is not standardised. It is generally regarded that acute CT occurs within 48–72 hours postoperatively and subacute/late CT after 72 hours.⁵ Some authors have considered delayed PPE if occurring after postoperative day 10⁶ or 15,¹ or by the time when treatment is required (eg, after the seventh postoperative day).² Severe delayed pericardial effusion had a 30-day mortality rate of 3% in one study,⁷ and the development of CT increases mortality to 11%.⁵

The progression of post-cardiotomy syndrome and inflammation seems to be the main cause for delayed PPE±CT. This is supported by the association with longer cross-clamp times, long pump run and higher volumes of transfused cells saved blood.^{2 8} Other surgical risk factors include elective surgery, surgery other than CABG, reoperation within 48 hours, elevated postoperative creatinine and tranexamic acid use.^{2 5 6} Patient risk factors include younger age, mild or moderate coronary artery disease, less beta-blocker usage and preserved preoperative ejection fraction. Routine deep vein thrombosis (DVT) prophylaxis or the timing of this was not associated with late CT.⁵ Preoperative anticoagulation therapy is associated with late CT, as is valve surgery and postoperative atrial fibrillation, possibly related to anticoagulant therapy in these cohorts.² Recent myocardial infarction (MI) (in the previous 30 days) and aspirin use are protective probably due to anti-inflammatory effect or because aspirin is indicated in MI which requires CABG—a surgery with lower CT association.⁵ The lower incidence in CABG may be explained also by the inadvertent opening of the left pleura which drains any source of bleeding that could lead to PPE±CT.² Early chest drain tube removal and low drainage volumes are associated with PPE±CT, although it is unclear from published studies whether low volumes may indicate tube blockage or malposition resulting in effusions developing unnoticed.^{2 9} Decreased frequency of postoperative echocardiographic evaluation is also associated with incidence of PPE±CT.²

In this case, it appears that commencing anticoagulation for PE initiated the effusion. The slow clinical onset, over the following 4 weeks, suggests a limited bleeding source such as pericardial inflammation or a small vessel, although no active source was identified during exploration to confirm this hypothesis. Preoperative apixaban reversal was considered unnecessary as surgery occurred after 48 hours since last dose and the patient had no renal dysfunction.

Treatment of pericardial effusions is recommended if the effusion size on echocardiography is >10 mm and is associated with haemodynamic instability, echocardiographic evidence of chamber compression, progressive enlargement or subjective symptoms.² Anterior effusions are generally amenable to drainage via pericardiocentesis or a subxiphoid incision, with more lateral and posterior effusions necessitating surgical drainage via thoracotomy, thoracoscopy or resternotomy.^{10 11}

The presence of CT at the time of surgery presents unique challenges to the anaesthetist, including a significant risk of loss of cardiac output on induction of anaesthesia and initiation of positive pressure ventilation. The mantra of anaesthesia induction in CT has been ‘fast, full and squeeze tight’, which translates to maintaining preload, avoiding bradycardia, maintaining systemic resistance and optimal contractility.¹² Percutaneous drainage of the effusion should be performed where possible prior to induction of anaesthesia to reduce the risk of haemodynamic compromise. However, pericardiocentesis may not be possible due to clot presence or inaccessible location as seen in this case and such limitations justify the use of ECMO in cardiac tamponade.¹³ When the effusion is accessible through a sternotomy approach, the patient is cautiously induced on the operating table with the chest prepared in standard sterile fashion in preparation for emergency chest opening, possibly while spontaneous ventilation is maintained. Immediate sternotomy and tamponade relief can then be performed expeditiously in the event of loss of cardiac output. Several challenges were considered when planning the anaesthetic management of this patient:

• The rapid development of obstructive shock due to an evolving pericardial effusion.

• Subxiphoid pericardiocentesis, pericardial window, thoracoscopic drainage or radiologically guided drainage were considered potentially injurious to the bypass grafts or not technically feasible due to the posterior location of the effusion or likely presence of postoperative adhesions, leaving left lateral thoracotomy under general anaesthesia as the only option.

• Emergency resternotomy in case of cardiovascular collapse on induction was not a favourable option due to a high risk of injuring coronary grafts given their location and the likelihood of significant adhesion burden over 4 months after initial surgery.

• The risk of loss of cardiac output on induction of anaesthesia with no prospect of immediate surgical tamponade relief and likely ineffective chest compressions in the event of cardiac arrest.

• Expected difficulty in accessing the pericardial sac, requiring prolonged dissection via a thoracotomy incision and requiring TOE guidance as well as one-lung ventilation—which itself carried a risk of hypoxaemia given the presence of pulmonary oedema and low cardiac output.

Although VA-ECMO has been used as a rescue tool for the emergency management of cardiac tamponade, we believe this case report is the first one to describe the planned preoperative commencement of peripheral VA-ECMO in a complex case of delayed PPE with CT, to facilitate safe induction of anaesthesia.^14–16 Although femoral cannulation for VA-ECMO is technically challenging in the semirecumbent position, this case shows it is possible, particularly in patients of a slim build. Data from the use of peripheral VA-ECMO for the emergency management of CT associated with post-MI-associated left ventricular free wall rupture provide evidence of limitations of this technique when significant haemodynamic compromise/arrest is present.^17–19 A high brain death rate was noted among patients in whom VA-ECMO was initiated prior to tamponade drainage, although it should be noted that VA-ECMO was established in all these patients following cardiac arrest and during cardiopulmonary resuscitation, so hypoxic brain injury may already have occurred.¹⁸ It has been suggested that IVC drainage alone may not allow for enough decompression of the superior vena cava (SVC) (and in turn the cerebral venous system) to provide a sufficient cerebral perfusion gradient when associated with decreased cardiac output in this cohort.¹⁹ Advancing the drainage cannula into the SVC is a potential solution.¹³ Additionally, the retrograde aortic flow of the return cannula may increase the afterload on the failing heart causing increased left ventricular wall tension/myocardial stress and increased oxygen demand that cannot be met leading to worsening myocardial ischaemia.

This case suggests that the elective use of VA-ECMO in CT cases at risk of cardiac arrest on induction of anaesthesia may be a safer option, rather than to be used as a rescue technique once cardiac arrest has occurred. How exactly the patient’s haemodynamic and organ perfusion status would respond to initiation of VA-ECMO was uncertain at the time. It was suspected that the reduction of right ventricular preload once ECMO drainage commenced would lead to underfilling and collapse of the right ventricle and left ventricle and effective cessation of native cardiac output resulting in the patient being entirely dependent on ECMO support until the tamponade was relieved. Inadequate cerebral perfusion during this period (as detected via NIRS monitoring and potentially related to failure of SVC drainage) would have precipitated a change to more direct sternotomy approach which would potentially have compromised coronary grafts. Fortunately, this was not the case. After ECMO initiation, left ventricular ejection continued (and arterial pulse pressure increased actually), central vein pressure (CVP) decreased, cerebral oximetry values were unchanged and the patient’s dyspnoea subjectively improved. Why this was the case is unclear, but what is obvious is that the behaviour of tamponade physiology once ECMO flows commence may be unpredictable and changes to management may need to be expedited if an adverse response occurs.

A potential complication of tamponade drainage (if rapid and of large volume) is the PDS.¹⁶ It has an incidence of up to 11% and is defined by the presence of uni/biventricular failure occurring immediately or within hours of drainage,^{20 21} and is associated with a mortality rate of up to 80%.¹⁶ A number of explanations have been proposed, including: excessive right ventricular preload leading to dilatation and impairment of the ventricular interdependence phenomenon and failure following application of intrapericardial negative pressure²¹; a sudden decrease in circulating catecholamines following drainage²² and myocardial stunning driven by alteration in the flow of epicardial vessel coronary flow due to changes in the intrapericardial pressure.²³ In this case, there was transient right ventricular failure and hypokinesia of the left ventricle (inferior segments) after drainage of effusion requiring temporary pulmonary vasodilator and inotropic support.

In conclusion, the use of VA-ECMO may provide a more stable pre-induction haemodynamic state, allowing for a measured, meticulous and safer surgical approach, greatly reducing the risk of anaesthesia-related morbidity or mortality in patients with tamponade physiology and limited/challenging surgical access.

Learning points.

• Initiation of anticoagulation after cardiac surgery may contribute to the development of pericardiac effusion and cardiac tamponade even when this occurs a few months postoperation.

• Although delayed postoperative pericardial effusion may be present in up to 20% of patients, only a minority progress to cardiac tamponade.

• Induction of general anaesthesia in patients with cardiac tamponade can result in catastrophic circulatory failure and death.

• Pericardiocentesis or subxiphoid surgical drainage under local anaesthesia should be considered where possible.

• Initiating peripheral veno-arterial extracorporeal membrane oxygenation (even with low flows as in this case) prior to induction of anaesthesia provides a means of supporting the circulation and preventing anaesthesia-related cardiovascular collapse in critically ill patients with cardiac tamponade and effusions which are difficult to access surgically.

Ethics statements

Patient consent for publication

Obtained.