Use of bi-caval cannulae for veno-venous ECMO in neonates and children

Veno-arterial ECMO remains the most common mode of extracorporeal support in infants and children, and despite increases in overall ECMO numbers the utilisation of veno-venous ECMO in neonates appears to be decreasing. We report here outcomes of neonatal and paediatric patients managed with veno-venous ECMO via bi-caval cannulae over a 10-year period in a tertiary referral ICU. Retrospective single-centre case series of veno-venous ECMO cases using dual lumen cannulae over a 10-year period at a tertiary referral paediatric hospital with a low volume ECMO program. In the 10-year period 2013–2022, 33 patients required ECMO with 23 receiving veno-arterial ECMO and 10 managed with veno-venous cannulation - 8 with bi-caval cannulae and 2 with multi-site cannulation. Overall survival was 23/33 (69.6%) and in the veno-venous group survival was 7/10 (70%). Median oxygenation index prior to veno-venous cannulation in the 8 patients undergoing bi-caval cannulation was 48 (range 34–54) and median PaO2 was 42 mmHg (range 34–59 mmHg). Duration of ECMO ranged from 7 to 14 days (median 9 days). Complications included migration of the cannula into the hepatic vein, minor and major bleeding, and compromised blood flow secondary to pneumomediastinum. Veno-venous ECMO can be reliably established via a single bi-caval cannula in the majority of patients. Outcomes in this small series from a low volume centre are broadly comparable to those reported from the ELSO database.

veno-venous (VV) ECMO appears to be declining, with the most frequently cited barriers being availability of suitable dual lumen cannulae and perceived difficulties in placement [2].Although multi-site cannulation is feasible in neonates the small calibre of femoral veins makes cannulation technically challenging.Placement of a dual lumen venous cannula (e.g., Avalon bi-caval cannula) via the right internal jugular vein allows for appropriate flows with minimal recirculation, however precise placement and close surveillance of cannula positioning is required.Mal-placement of the cannula tip has been associated with complications including pericardial tamponade [3] and flow disruption secondary to migration of the cannula tip into the hepatic veins [4].We undertook this study to review our experience, complications and outcomes with VV ECMO using bi-caval cannulae in a low volume paediatric ECMO centre.

Methods
ICU patients receiving ECMO at the Sydney Children's Hospital over a 10-year period from January 1 2013 to December 31 2022 were identified from a comprehensive ICU database that is maintained prospectively.Medical records for each patient were reviewed by one of the authors (JA).Sydney Children's Hospital is a universityaffiliated paediatric tertiary referral centre with all medical and surgical subspecialties represented.The ICU admits approximately 1100 patients per annum aged from birth to 16 years and is one of three tertiary paediatric centres in the state of NSW serving a total population of 8.2 million, including 1.5 million children aged 14 years or less [5].A separate neonatal ICU on campus cares for premature infants, however infants born with complex surgical conditions (e.g.congenital diaphragmatic hernia) are managed in the paediatric ICU.Paediatric cardiac surgeons undertake surgical cannulation for commencement of all VA ECMO, VV cannulation requiring surgical cut-down, and to assist the paediatric intensivists with percutaneous cannulation for VV ECMO.Perfusionists with both paediatric and adult experience are available on campus.Medical and nursing skills are maintained with regular ECMO wet-lab training as well as high fidelity simulation sessions.
Cannulation for VV ECMO using the Avalon bi-caval cannula was undertaken by a paediatric cardiac surgeon and paediatric intensivist, both scrubbed.Cannulation occurred in the ICU, avoiding the risks of patient transfer to operating theatres, and minimising delay.The patient is positioned with the neck extended.Both the neck and sternum are prepped and accessible, allowing for an urgent sternotomy if needed.Using ultrasound, the right internal jugular vein is accessed 1-2 cm above the clavicle.The vein is cannulated with a 5F sheath, allowing manipulation of a guidewire into the inferior vena cava (IVC).A sterile echocardiography probe positioned in the epigastrium provides excellent views of the right atrium and ventricle, tricuspid valve, IVC, and hepatic veins.This allows the proceduralist to both guide the wire and simultaneously steer the ultrasound images.Sequential dilation and catheter placement is performed with continuous echo imaging.All patients were anticoagulated with heparin infusions, targeting an activated clotting time (ACT) of 160-180 s.Standard circuits for patients weighing less than 13 kg included ¼ inch access and return tubing with either a Medos Hi-Lite 2400 LT Oxygenator (Xenios AG, Heibronn, Germany) or a Quadrox iD oxygenator (Maquet Getinge Group, Rastatt, Germany) for gas exchange.ECMO hardware of choice was the 2 nd generation Cen-triMag ® console coupled with the PediVAS ® blood pump (Thoratec Corporation, Pleasanton, CA, USA).For larger patients a 3/8 inch circuit was used, namely the HLS 5.0 or HLS 7.0 Advanced Cardiohelp circuit with integrated centrifuge pump head and oxygenator (Maquet Getinge Group, Rastatt, Germany).Pump flow rates were set to target S p O 2 of 88-94%.

Results
In the 10-year period from 2013-2022, 33 patients required ECMO support (Table 1).Overall survival was 23/33 (69.6%).VA ECMO was used in 23 children (16/23 = 69.5% survived) and VV ECMO was used in 10 children (7/10 = 70% survived).Paediatric Avalon bicaval cannulas were placed in 8 of the 10 VV ECMO patients, including 4 neonatal cannulations (Table 2).Cannula sizes were 13 Fr in the four neonates and 16 Fr in the older children.All patients supported with VV ECMO had hypoxic respiratory failure although one child had predominately ventilatory failure and a large air leak.The median P a O 2 prior to ECMO was 42 mmHg (range 34-59 mmHg), and the median Oxygenation Index was 48 (range 34-54) (Table 2).All patients had already received muscle relaxants, IV steroids and a trial of inhaled nitric oxide prior to cannulation.Ages ranged from 1 day to 4.8 years with weights 3.1 to 18 kg.Duration of ECMO ranged from 7 to 14 days (median 9 days).
Vascular access was achieved percutaneously using an ultrasound guided Seldinger technique in four patients.A further four patients were accessed via surgical cutdown with a semi-Seldinger technique.Cardiac surgeons attended all insertions other than one urgent case where a surgeon was not immediately available.There were no complications resulting from Avalon cannula insertion, and no intracardiac mal-placement at any time.
Circuit blood flows in the four neonates via 13F cannulae were sufficient to achieve adequate oxygenation.Median ECMO circuit blood flow in these patients was 98 ml/kg/min, with a range of 76-106 ml/kg/min.In the older children with 16F cannulae, median blood flow rate was 71 ml/kg/min, with a range of 57-90 ml/ kg/min.A number of complications were encountered during the ECMO runs (Table 3).Cannulas in two patients, both neonates, migrated into the hepatic vein with an associated reduction in ECMO flows.Both cases were easily managed by extension of the neck under ultrasound guidance, until the cannula tip was seen to sit just above the IVC/hepatic vein junction, with restoration of flows.One patient with H 1 N 1 influenza and significant air leak with mediastinal emphysema experienced a substantial decrease in blood flow rate on day three of ECMO.Cannula position could not be identified with echocardiograpy because of the mediastinal emphysema.The circuit was converted to a two cannula VV ECMO configuration, using jugular and femoral veins with good results.One patient aged 4.8 years had a preexisting Hickmann line in the right internal jugular vein.The long term line was removed during the ECMO cannulation, however scarring of the vein limited the cannula size to 16F.Despite only 66 ml/kg/min of blood flow adequate oxygenation was achieved.One patient suffered an intracranial haemorrhage which was fatal.This patient had acute lymphoblastic leukaemia and was thrombocytopaenic.Two patients had minor bleeding around the cannula site, which required further suturing.A single patient had a brief cardiac arrest secondary to a tension pneumo-mediastinum, requiring urgent decompression.

Discussion
Overall survival following ECMO support for severe acute respiratory failure is 74% in neonates and 58% in older children [1].Survival rates are largely dependent on the aetiology of the respiratory failure, ranging from 94% in meconium aspiration syndrome [1] to 28% in pertussis [6] and 11% in neonatal adenovirus [7].In contrast to adult practice, the majority of infants requiring extracorporeal support for severe acute respiratory failure receive VA ECMO, commonly with carotid or aortic cannulation [1].This is despite evidence showing a higher incidence of neurological complications, and worse neurological outcomes with VA ECMO compared to VV ECMO when used for respiratory indications [8].
Multi-site cannulation (usually femoral and jugular veins) for VV ECMO is often difficult in infants, and especially in neonates.A smaller femoral venous access cannula can be inserted, but blood flow rate will be reduced.Alternatively, reversing polarity-accessing the jugular and returning via the smaller femoral vein will afford more circuit flow than a dual lumen cannula,  however is associated with femoral vein occlusion [9].The right internal jugular cannula tip also needs careful placement to avoid recirculation.Double lumen venous cannulas have been used for paediatric ECMO for more than three decades [10].Problems experienced include recirculation at high flows, softening and kinking, collapsing, and poor drainage [11].The Avalon cannula design has addressed many of these issues.Being wire reinforced, it is resistant to kinking and collapse.When correctly positioned, the bi-caval drainage allows adequate blood flows to be achieved with minimal recirculation.
A number of Avalon cannula related complications have been reported in paediatric patients, especially in neonates.Pericardial tamponade from cardiac perforation has been described in up to 6.9% of neonates during VV ECMO [12].This complication generally occurs when the cannula has been pulled back into the right atrium to improve drainage; the Avalon cannula is not designed to have its tip positioned in an intracardiac location.
In neonates the 13F cannula tip tends to sit in the IVC just beyond the hepatic vein junction.In this series, the commonest cannula related complication we experienced was hepatic vein migration in two of eight patients (25%), both neonates.This was either detected incidentally on routine echocardiography, or in response to a more negative access pressure for venous drainage.Both cases resolved with simple neck extension under echo guidance, ensuring the cannula tip was in the IVC.
Positioning the cannula tip in the inferior vena cava is essential.We access the right internal jugular vein as low in the neck as is practical, with minimal tunnelling, in order to ensure an adequate length within the IVC.By placing a 5F sheath in the right internal jugular vein the wire is easily directed into the right atrium, facilitating manipulation into the IVC.On occasions a stiffer wire is needed to direct the cannula to the IVC.There are reports of wires being looped in the right ventricle, and not being visualised on trans-oesophageal echocardiography [13].Views of right sided cardiac structures and the IVC are excellent with an epigastric probe in neonates, which is our preferred method of visualising placement during neonatal cannulation (Fig. 1).We utilise transoesophageal echocardiography in older children.However, in neonates right atrial and IVC views with an epigastric probe are often superior to transoesophageal and allow the proceduralist to control the image.Whilst fluoroscopy has advantages and is preferred in some institutions, the transfer of a critically unwell child to the fluoroscopy suite is potentially hazardous and far too time consuming in a clinical situation that requires urgent intervention.If unable to place the Avalon cannula within the inferior vena cava, a two cannula approach should be used.The position of Avalon cannulae should be frequently monitored, ideally by intensivists familiar with bedside echocardiography.Patient repositioning needs to be undertaken with care, particularly with respect to head movement.Our approach is to maintain deep sedation and muscle relaxation during ECMO in order to minimise unnecessary and unexpected movement.
The relationship between case volume and outcome in ECMO is unclear.In a retrospective analysis of the ELSO registry higher ECMO case volume was associated with improved survival in adults, with centres undertaking more than 30 cases per year having lower risk-adjusted mortality than those with an annual volume of fewer than 6 cases [14].Paradoxically, superior outcomes in low (compared to high) volume centres have also been noted [15,16].Paediatric data variably suggests either no relationship between case volume and outcome [14,17] or an inverse relationship between case volume and mortality for cardiac surgical cases [18][19][20] but not in non-cardiac cases [20].ECMO case volume does not appear to be a surrogate marker for quality [14].
Children's hospitals with a large medical referral base need to maintain a safe provision of extracorporeal support in acute respiratory failure.Depending on the clinical situation, programs need to have options in venovenous cannulation.This includes surgical cut-down and percutaneous cannulation, as well as dual lumen and two cannula configurations.As a tertiary children's hospital with a low volume ECMO program, we have described our use of the dual lumen bicaval cannula in neonates and older children, with acceptable outcomes in this small series.The challenge is to maintain the quality in service delivery when there is a low volume of patients.This can only be achieved with timely access to cardiac surgical support together with regular education for medical and nursing staff, including interactive high fidelity ECMO simulation.

Table 2
Patients with V-V bi-caval cannulation ALL Acute lymphoblastic leukaemia, GBS Group B Stretptococcus, RSV Respiratory syncytical virus, Perc Percutanous a Converted to two cannula VV configuration on day 3

Table 3
Complications encountered in 8 patients undergoing VV ECMO with bi-caval cannulation