Venoarterial extracorporeal membrane oxygenation (VA-ECMO) in high-risk (HR) percutaneous coronary intervention (PCI)

Abstract

Objective

Mechanical circulatory support is increasingly utilised in peri-procedural contexts. Herein we describe the characteristics and outcomes of patients undergoing high-risk (HR) percutaneous coronary intervention (PCI) with venoarterial extracorporeal membrane oxygenation (VA-ECMO) support.

Design

Single centre, retrospective cohort study conducted at a quaternary Australian ECMO retrieval hospital.

Methods

Consecutive patients between January 2019 and June 2024, who received peripheral VA-ECMO support for HR-PCI were identified from the institutional ECMO database. The primary outcome was the incidence of major adverse cardiovascular and cerebrovascular events (MACCE) to hospital discharge. Secondary endpoints included procedural success, incidence of vascular access and bleeding complications, acute kidney injury (AKI), and duration of mechanical circulatory support (MCS), ventilation, ICU and hospital length of stay.

Results

Seventeen patients received VA-ECMO supported HR-PCI during the study period. The median age was 70 years (IQR; 59–78). Fourteen (82 %) patients were male; fifteen (88 %) had moderate to severe left ventricular dysfunction pre-procedurally. Ten (59 %) patients underwent emergency VA-ECMO support and seven (41 %) patients were supported prophylactically, with mean pre-procedural SYNTAX scores of 39 ± 12 and 30 ± 10, respectively. Seven MACCE events occurred, all being death in the emergency group. Fifteen patients (88 %) were successfully revascularised. Significantly more patients in the emergency versus the prophylactic group developed AKI (80 % vs 14 %, p = 0.02). There were significantly more vascular and access-related complications in the emergency group (7 vs 1, p = 0.049).

Conclusion

In seventeen patients undergoing HR-PCI, prophylactic VA-ECMO support was feasible, safe and associated with good short-term outcomes, whilst emergency VA-ECMO supported HR-PCI was associated with greater burden of mortality and morbidity. Larger prospective studies, including cost analysis comparing other MCS devices are required.

1. Introduction

1.1. Background

Complex high-risk percutaneous coronary intervention (HR-PCI) is increasing with the age and frailty of the patient population undergoing percutaneous coronary intervention (PCI).¹ There is currently no uniform definition of HR-PCI,² however commonly accepted features include significant comorbidities, concomitant cardiogenic shock, complex coronary disease including: left main (or equivalent) disease, chronic total occlusions, large areas of at risk myocardium, and severe ventricular and/or severe valvular dysfunction (Fig. 1).³ In lieu of a clear consensus or objective criteria, multi-disciplinary teams play a key role in providing a robust framework to scrutinise case selection for HR-PCI.³ Mechanical circulatory support (MCS) is increasingly being used to provide peri-procedural circulatory support for HR-PCI,[4], [5], [6] despite conflicting international guidelines.[7], [8], [9], [10] To date MCS has largely been with intra-aortic balloon pump (IABP) and microaxial flow devices for PCI, with relatively limited data on VA-ECMO support.[11], [12], [13] Herein, we describe our processes for and experience of VA-ECMO to facilitate HR-PCI.

Fig. 1.

HR-PCI: high-risk percutaneous coronary intervention.

2. Methods

2.1. Study population

This single centre retrospective study included patients undergoing VA-ECMO supported HR-PCI between January 2019–June 2024 at Royal Prince Alfred Hospital, a quaternary ECMO retrieval hospital in Sydney, Australia. Ethical approval was obtained via Sydney Local Health District (SLHD) research ethics committee (2024/ETH01514) with waiver of consent.

Patients were categorised by whether they underwent emergency or prophylactic ECMO support for HR-PCI. Patients undergoing emergent VA-ECMO support included those with clinically urgent indications for PCI such as extracorporeal cardiopulmonary resuscitation (ECPR) for refractory cardiac arrest or cardiogenic shock. Amongst those undergoing ECPR, the term ‘salvage’ is used to delineate ECPR for cardiac arrest during PCI, compared to ECPR for cardiac arrest prior to PCI. Prophylactic ECMO support described planned, scheduled, elective ECMO implantation in patients with clinical characteristics that were not requiring emergent PCI based on clinical status.

Patients in the emergency group had cardiothoracic surgery considered where feasible, whereas electively supported patients had declined, or were deemed inappropriate for cardiothoracic surgery through multi-disciplinary team (MDT) consultation. All elective cases were selected by a MDT which included members from intensive care, the ECMO service (intensivist cannulator and cardiothoracic anaesthetist), cardiothoracic surgery, interventional and heart failure cardiology teams as per our local process (Fig. 2).¹⁴ Vascular surgical colleagues were also consulted on a discretionary basis. This MDT determined candidacy for salvage surgical intervention and discussed the appropriateness of extended MCS if required following coronary revascularisation. Informed patient consent was obtained from elective patients with documentation of medical goals of care.

Fig. 2.

Extracorporeal membrane oxygenation (ECMO) multidisciplinary consultation by modality at Royal Prince Alfred Hospital.

VA-ECMO: venoarterial ECMO, PERT: pulmonary embolism response team, VV-ECMO: venovenous ECMO, ECPR: ECMO cardiopulmonary resuscitation, MDT: multi-disciplinary team.

Whilst there are no strict pre-defined criteria to indicate the need for prophylactic VA-ECMO, patient, lesion and procedural factors guide consideration. Factors include severe left ventricular dysfunction, evolving or potential cardiogenic shock, left main coronary artery occlusion or equivalent, bifurcation lesions, severe triple-vessel disease, chronic total occlusions, or operative factors including the use of rotational atherectomy and prolonged procedural time. Patient age, clinical frailty and comorbidities were also considered in determining both eligibility and appropriateness.

Patient characteristics including age, comorbidities and severity of illness at presentation were collected. Available transthoracic or transoesophageal echocardiography was reviewed with ventricular function and pulmonary pressures assessed as per established guidelines.¹⁵ Angiographic characteristics reported included number of stents inserted and use of rotablation (rotational atherectomy). Left main equivalent lesions were defined as ≥ 70 % stenosis of the proximal left anterior descending coronary artery and of the proximal circumflex coronary artery.¹⁶ The cohort was assessed with the Synergy between PCI with Taxus and Cardiac Surgery (SYNTAX) score, which grades coronary artery disease complexity and helps guide revascularisation with coronary artery bypass graft or PCI,^9,17,18 the residual SYNTAX score, which quantifies vessel stenosis post PCI¹⁹ and the European System for Cardiac Operative Risk Evaluation II (EuroSCORE II), which helps predict risk of in-hospital mortality post major cardiac surgery.^20,21 Incidence of successful re-vascularisation was also evaluated, and defined as a Thrombolysis in Myocardial Infarction (TIMI) flow grade of III and a final vessel stenosis < 30 %.^22,23

ECMO characteristics assessed included configuration, cannula size, extracorporeal blood flow, ECMO duration, method of anaesthesia, methods of closure and complications of vascular access. Post procedural vasoactive-inotropic score (VIS) was also assessed, with higher scores shown to be associated with greater morbidity and mortality.^24,25

2.2. Procedure

Percutaneous ultrasound guided cannulation for ECMO supported PCI was performed in the emergency department (emergent PCI/ECPR) utilising transoesophageal echocardiography (TOE), or in cardiac catheterisation laboratory utilising fluoroscopy and TOE as described previously.¹⁴ Whilst it is routine to perform ipsilateral cannulation in our service, in case of preexisting medical devices or contraindications, a bifemoral configuration is occasionally used. A distal perfusion cannula (DPC) was placed prophylactically in the ipsilateral superficial femoral artery in cases requiring post-procedural support. Both Cardiohelp® and Rotaflow® (Maquet, Rastatt, Germany) consoles were utilised during this study period, and extracorporeal blood flow was generally titrated to a cardiac index of > 2.2L/min/m², and to maintain a mean arterial pressure > 65 mmHg, and/or an SvO₂ > 65 %.

2.3. Data collection

Study data was reviewed and collected from patient electronic medical records stored in Philips IntelliSpace Critical Care and Anesthesia (ICCA®, Philips Medical Systems) and Cerner® PowerChart® (Cerner Corporation) and managed using Research Electronic Data Capture (REDCap).²⁶

2.4. Outcomes

The primary outcome was major adverse coronary and cerebrovascular events (MACCE) to hospital discharge defined as a composite of death, myocardial infarction, transient ischaemic attack or stroke, or target vessel revascularisation.⁷ Secondary outcomes were successful revascularisation, incidence of dialysis and acute kidney injury (AKI) assessed as per the Kidney Disease Improving Global Outcomes (KDIGO) staging criteria,²⁷ sepsis (defined by the presence of bacteraemia, clinical suspicion ± radiological evidence of sepsis), bleeding complications according to the Bleeding Academic Research Consortium (BARC) scale,²⁸ vascular and access-related complications according to the Valve Academic Research Consortium 3 (VARC-3) definitions,²⁹ duration of MCS and ventilation, ICU utilisation and hospital length of stay. Sequential Organ Failure Assessment (SOFA) score was calculated using a modified respiratory component where FiO₂ and PaO₂ were not available, as previously described.³⁰

2.5. Statistical analysis

Data analysis was performed with R statistical software (version 4.3.0, R Core Team, Vienna, Austria).³¹ Numerical variables were analysed using Mann–Whitney U tests and presented with mean ± standard deviation (SD), and categorical or binary variables analysed with Fisher exact tests and presented with median and inter-quartile range (IQR; Q1-Q3). All tests were two-tailed, and a p value of < 0.05 was considered significant.

3. Results

3.1. Baseline clinical characteristics

Seventeen patients underwent ECMO supported PCI during the study period (Table 1). Sixteen (94 %) patients presented with acute coronary syndrome (ACS) and one (6 %) patient presented with heart failure (Fig. 3). Fourteen (82 %) patients were male, and median age was non-significantly different between the emergency (70 years old, IQR 58–73) and the elective groups (78 years old, IQR 62–82) (p = 0.14). Body mass index (BMI) was higher in the emergency (31 kg/m², IQR 30–36) than the elective group (25 kg/m², IQR 22–28) (p = 0.03).

Table 1.

Baseline patient characteristics.

	Emergency VA-ECMO (n = 10)	Elective VA-ECMO (n = 7)	p value
Age, median (IQR)	70 (58–73)	78 (62–82)	0.14
Gender - # (%)			1.00
Male	8 (80.0 %)	6 (85.7 %)
Female	2 (20.0 %)	1 (14.3 %)
BMI - kg/m2, median (IQR)	31 (30–36)	25 (22–28)	0.03
Risk factors - # (%)
Hypertension	9 (90.0 %)	5 (71.4 %)	0.54
Hypercholesterolaemia	2 (20.0 %)	3 (42.9 %)	0.59
Current smoker	0 (0.0 %)	2 (28.6 %)	0.15
Diabetes	6 (60.0 %)	3 (42.9 %)	0.64
Peripheral artery disease	1 (10.0 %)	2 (28.6 %)	0.54
Chronic kidney disease	3 (30.0 %)	3 (42.9 %)	0.64
Intermittent haemodialysis	0 (0.0 %)	1 (14.3 %)	0.41
Known coronary artery disease	2 (20.0 %)	3 (42.9 %)	0.59
Previous PCI	2 (20.0 %)	2 (28.6 %)	1.00
Coronary artery bypass graft surgery	1 (10.0 %)	1 (14.3 %)	1.00
Presentation			0.41
Acute coronary syndrome - # (%)	10 (100.0 %)	6 (85.7 %)
Heart failure - # (%)	0 (0.0 %)	1 (14.3 %)
Left ventricular ejection fraction - # (%)			0.61
Mild	2 (20.0 %)	0 (0.0 %)
Moderate	2 (20.0 %)	2 (28.6 %)
Severe	6 (60.0 %)	5 (71.4 %)
Right ventricular dysfunction - # (%)	3 (30.0 %)	5 (71.4 %)	0.15
Pulmonary hypertension - # (%)	2 (20.0 %)	2 (28.6 %)	1.00
SCAI grade - # (%)			0.001
A	0 (0.0 %)	5 (71.4 %)
B	1 (10.0 %)	1 (14.3 %)
C	1 (10.0 %)	1 (14.3 %) a
D	2 (20.0 %)	0 (0.0 %)
E	6 (60.0 %)	0 (0.0 %)
Pre ECMO cardiac arrest - # (%)	7 (70.0 %)	1 (14.3 %)	0.05
ROSC prior to ECMO	1 (10.0 %)	1 (14.3 %)
ECPR	6 (60.0 %)	0 (0.0 %)	0.035
Pre PCI	3 (30.0 %)	0 (0.0 %)
During PCI	3 (30.0 %)	0 (0.0 %)
SOFA, mean (±SD)	4.6 (3.5)	2.5 (1.7)	0.29
EuroSCORE II, mean (±SD)	19.5 (20.7)	8.8 (6.3)	0.17

BMI: body mass index, PCI: percutaneous coronary intervention, SCAI: Society for Cardiovascular Angiography and Interventions, ROSC: return of spontaneous circulation, ECMO: extracorporeal membrane oxygenation, ECPR: extracorporeal cardiopulmonary resuscitation, SOFA: sequential organ failure assessment, EuroSCORE II: European System for Cardiac Operative Risk Evaluation II.

^aOne patient in the elective group suffered a brief cardiac arrest. Further details in Supplementary Table 2.

Fig. 3.

Cohort composition. HR-PCI: high risk percutaneous coronary intervention, ECMO: extracorporeal membrane oxygenation, ACS: acute coronary syndrome, ECPR: Extracorporeal cardiopulmonary resuscitation.

^a One patient in the elective ECMO group with ACS had a STEMI, with revascularisation delayed until MDT discussion.

There were no significant differences in comorbidities, mean SOFA, median APACHE III or mean EuroSCORE II scores, severity of left ventricular dysfunction, incidence of right ventricular dysfunction or pulmonary hypertension at baseline (Table 1).

Of the ten patients in the emergency group, there were six (60 %) ECPRs including three during PCI (salvage ECPRs), three (30 %) ST-segment elevation myocardial infarctions (STEMIs) and one (10 %) non-ST-segment elevation myocardial infarction (NSTEMI) (Supplementary Table 1). Of the seven patients in the elective group, there were five (71 %) NSTEMIs, one (14 %) STEMI, and one (14 %) elective PCI for heart failure. 80 % of the emergency cohort suffered advanced cardiogenic shock (SCAI D-E) (Table 1 and Supplementary Table 1). Seven (70 %) patients in the emergency group and one (14 %) patient in the elective group suffered pre-procedural cardiac arrest (p = 0.05) (Table 1).

3.2. ECMO characteristics

All patients underwent general anaesthesia for ECMO procedural support (Table 2). Two (12 %) patients were already mechanically ventilated prior to arrival in cardiac catheterisation laboratory. Six patients received ECPR with a mean time from arrest to ECMO flow of 27.7 ± 9 min (Supplementary Table 2).

Table 2.

ECMO characteristics.

	Emergency VA-ECMO (n = 10)	Elective VA-ECMO (n = 7)	p value
Intubation - # (%)			0.42
Not intubated	0 (0.0 %)	0 (0.0 %)
Already intubated	1 (10.0 %)	0 (0.0 %)
Peri-procedural	5 (50.0 %)	6 (85.7 %)
Due to cardiac arrest	4 (40.0 %)	1 (14.3 %)
Cannulation - # (%)			0.23
Unilateral	7 (70.0 %)	7 (100.0 %)
Bilateral	3 (30.0 %)	0 (0.0 %)
Access - # (%)			0.028
21 Fr	0 (0.0 %)	1 (14.3 %)
23 Fr	2 (20.0 %)	5 (71.4 %)
24 Fr	1 (10.0 %)	0 (0.0 %)
25 Fr	7 (70.0 %)	1 (14.3 %)
Return - # (%)			0.003
15 Fr	2 (20.0 %)	7 (100.0 %)
16 Fr	1 (10.0 %)	0 (0.0 %)
17 Fr	6 (60.0 %)	0 (0.0 %)
19 Fr	1 (10.0 %)	0 (0.0 %)
Distal perfusion cannula (superficial femoral artery) - # (%)			0.05
None	3 (30.0 %)	6 (85.7 %)
7 Fr	4 (40.0 %)	1 (14.3 %)
8 Fr	3 (30.0 %)	0 (0.0 %)
ECMO blood flow - L/min, mean (±SD)	2.7 (0.8)	2.4 (0.4)	0.13
Post procedural VIS - mean (±SD)	16 (22)	6 (5)	0.24
Decannulated at end of procedure - # (%)	2 (20.0 %)	6 (85.7 %)	0.015
Lactate – median (IQR)	3 (1.6–4.5)	1 (0.9–1.2)	0.024
Haemoglobin decrease >20 g/L - # (%)	5 (50.0 %)	2 (28.6 %)	0.62
Access site closure technique - # (%)			0.001
Simple suture	5 (50.0 %)	1 (14.3 %)
Perclose ProGlide™	0 (0.0 %)	6 (85.7 %)
Palliative decannulation	5 (50.0 %)	0 (0.0 %)
Return site closure technique - # (%)			0.009
Perclose ProGlide™	3 (30.0 %)	7 (100.0 %)
Pre-closure	2 (20.0 %)	7 (100.0 %)
Post-closure	1 (10.0 %)	0 (0.0 %)
Open repair	3 (30.0 %)	0 (0.0 %)
Palliative decannulation	4 (40.0 %)	0 (0.0 %)
Adjunct mechanical support - # (%)
Intra-aortic balloon pump	1 (10.0 %)	0 (0.0 %)

ECMO: extracorporeal membrane oxygenation, VIS: vasoactive-inotropic score.

Fourteen (82 %) patients underwent cannulation of the ipsilateral common femoral artery and vein (Table 2). Both access (p = 0.028) and return (p = 0.003) cannula were significantly larger in emergency cannulations, with the elective group most commonly receiving 23 Fr multistage venous access and 15 Fr single stage return cannula, and the emergency group receiving 25 Fr multistage access and 17 Fr return cannula. Eight (47 %) patients received prophylactic DPCs for ongoing post procedural ECMO support, weighted towards the emergency cohort (70 % vs 14 %, p = 0.05). One (6 %) patient in the elective group had severe peripheral vascular disease necessitating on-table iliac artery angioplasty to facilitate ECMO cannulation (Supplementary Table 3).

There was no significant difference in procedural ECMO blood flow rate and post-procedural VIS was non-significantly higher in the emergency versus the elective group (16 ± 22 vs 6 ± 5, p = 0.24) (Table 2). Eight (47 %) patients were decannulated at the conclusion of PCI, more of whom were elective versus emergency cases (86 % vs 20 %, p = 0.015). The post-procedural lactate was higher in the emergency group (3.0 mmol/L, IQR 1.6–4.5) vs (1 mmol/L, IQR 0.9–1.2) (p = 0.024). There was no difference in the number of patients with a significant haemoglobin drop (>20 g/L).

For femoral vein decannulation, six (86 %) patients in the elective group underwent Perclose ProGlide™ closure, compared to the five (50 %) simple suture and five (50 %) palliative decannulations in the emergency group (p = 0.001). All femoral arterial decannulations in the elective group were performed with the ProGlide™ system, whilst in the emergency group three (30 %) underwent Perclose ProGlide™, three (30 %) open repair, and four (40 %) palliative decannulations (p = 0.009).

One patient (6 %) received adjunct mechanical circulatory support with an IABP, and 1 patient (6 %) had an extended admission with transitions between different ECMO configurations including hybrid VA-V ECMO. Neither patient survived to hospital discharge.

3.3. Coronary angiography characteristics

Twelve (71 %) patients had left main or equivalent disease and eleven (65 %) had multi-vessel disease (Table 3). Fifteen (88 %) patients underwent coronary artery stent insertion, of whom fourteen (82 %) achieved TIMI III reflow. Two (12 %) patients, both of whom were in the emergency group, did not undergo coronary stent insertion. One underwent suction thrombectomy for paroxysmal embolism to the coronary arteries whilst the other had pathology not amenable to coronary artery stenting (Supplementary Table 3). Thirteen (76 %) patients required multiple coronary stents. Four (57 %) patients in the elective group underwent rotablation (p = 0.015), with two (29 %) of these having chronic total occlusions. Differences between the groups in SYNTAX (emergency; 39 ± 12) vs (elective; 30 ± 10) (p = 0.23) and residual SYNTAX (emergency; 8.3 ± 10) vs (elective; 2.6 ± 3) (p = 0.28) scores did not reach statistical significance.

Table 3.

Coronary angiographic characteristics.

	Emergency VA-ECMO (n = 10)	Elective VA-ECMO (n = 7)	p value
Left main or equivalent - # (%)	7 (70.0 %)	5 (71.4 %)	1.00
Multi-vessel PCI - # (%)	5 (50.0 %)	6 (85.7 %)	0.30
Neither left main or multi-vessel PCI - # (%)	1 (10.0 %)	1 (14.3 %)
Coronary artery stent insertion - # (%)	8 (80.0 %)	7 (100.0 %)	0.49
Suction thrombectomy - # (%)	1 (10.0 %)	0 (0.0 %)
TIMI grade flow - # (%)			1.00
3	8 (80.0 %)	7 (100.0 %)
2	1 (10.0 %)	0 (0.0 %)
No revascularisation	1 (10.0 %)	0 (0.0 %)
Successful revascularisation - # (%)	8 (80.0 %)	7 (100.0 %)	1.00
Stents - # (%)			0.29
0	2 (20.0 %)	0 (0.0 %)
1	2 (20.0 %)	0 (0.0 %)
2	0 (0.0 %)	2 (28.6 %)
3	3 (30.0 %)	4 (57.1 %)
4	0 (0.0 %)	0 (0.0 %)
5	2 (20.0 %)	1 (14.3 %)
6	1 (10.0 %)	0 (0.0 %)
Rotablation - # (%)	0 (0.0 %)	4 (57.1 %)	0.015
Chronic total occlusion (CTO) - # (%)	0 (0.0 %)	2 (28.6 %)	0.15
Average J-CTO score - #	0	2.5
SYNTAX, mean (±SD)	38.5 (11.6)	30.4 (10.3)	0.23
Residual SYNTAX, mean (±SD)	8.3 (10.1)	2.6 (3.0)	0.28

PCI: percutaneous coronary intervention, TIMI: thrombolysis in myocardial infarction, SYNTAX: Synergy between PCI with Taxus and Cardiac Surgery.

3.4. Primary outcome

Seven MACCE events occurred in total, all due to deaths prior to hospital discharge (one as late as 60 days) (Table 4). These occurred exclusively in the emergency group (Fig. 4) (7 vs 0, p = 0.01), with five of the six (83 %) ECPR cases dying prior to hospital discharge (Supplementary Table 2). Further details including causes of death are outlined in Supplementary Table 1.

Table 4.

Outcomes.

	Emergency VA-ECMO (n = 10)	Elective VA-ECMO (n = 7)	p value
In hospital mortality – # (%)	7 (70.0 %)a	0 (0.0 %)	0.01
Death despite mechanical circulatory support - # (%)	1 (10.0 %)	0 (0.0 %)
ECPR	5 (50.0 %)	0 (0.0 %)
Palliative withdrawal - # (%)	6 (60.0 %)b	0 (0.0 %)	0.035
ICU admission – # (%)	10 (100.0 %)	6 (85.7 %)	0.41
CCU admission – (# (%)	0 (0.0 %)	1 (14.3 %)
Duration of ECMO support – hours, median (IQR)	70.1 (9–132)	4.4 (3–6)	0.09
Duration of ventilation – hours, median (IQR)	103.6 (31–166)	19.5 (10–22)	0.07
Acute kidney injury – # (%)	8 (80.0 %)	1 (14.3 %)	0.02
Stage 1	2 (20.0 %)	0 (0.0 %)
Stage 2	0 (0.0 %)	0 (0.0 %)
Stage 3	6 (60.0 %)	1 (14.3 %)	0.13
Dialysis – # (%)	5 (50.0 %)	2 (28.6 %)c	0.62
Sepsis – # (%)	3 (30.0 %)	3 (42.9 %)	0.64
Vascular and access-related complications (VARC-3)	7	1	0.049
Nil	3	6
Major	1	0
Minor	6	1
Bleeding complications (BARC)	5	1	0.30
0	5	6
1	0	0
2	3	0
3a	0	1
3b	1	0
4	0	0
5a	1	0
APACHE III score, median (IQR)	72 (47–101)	62 (40–77)	0.36
Duration of ICU admission – days, median (IQR)	6.5 (3–10)	3 (2–8)	0.43
Duration of hospital admission – days, median (IQR)	15 (8–23)	13 (11–21)	0.70

ECPR: extracorporeal cardiopulmonary resuscitation, ICU: intensive care unit, CCU: coronary care unit, VA-ECMO: venoarterial extracorporeal membrane oxygenation, VARC-3: Valve Academic Research Consortium, BARC: Bleeding Academic Research Consortium, APACHE III: Acute Physiology, Age and Chronic Health Evaluation III.

^aAll deaths occurred in ICU.

^bPatient who died after 30 days.

^cPatient with CKD regularly dialysed.

Fig. 4.

Incidence of major adverse coronary and cerebrovascular events (MACCE).

3.5. Secondary outcomes

Sixteen (94 %) patients were admitted to the ICU post-procedurally, and one (6 %) to the coronary care unit (Table 4). The median duration of ECMO support (p = 0.09) or mechanical ventilation (p = 0.07) was not significantly different between the groups. One (6 %) emergency patient underwent ECMO support for more than 25 days. Nine (53 %) patients developed AKI, with significantly more in the emergency versus the elective group (80 % vs 14 %, p = 0.02) (Fig. 5). Seven (41 %) patients required dialysis (one of whom was dialysis dependent prior to PCI) with no significant difference between groups. Six (35 %) patients developed sepsis secondary to pneumonia or the urinary tract, with no significant difference between groups (Supplementary Table 5).

Fig. 5.

Incidence of acute kidney injury and vascular and access-related complications.

There were eight vascular and access-related complications in total, with significantly more occurring in the emergency group (7 vs 1, p = 0.049) (Fig. 5). There were six BARC bleeding complications with no significant difference between groups (p = 0.3). One ECPR patient (emergency group) experienced a major vascular and access-related complication – with misplacement of a multistage venous cannula in the common femoral artery causing acute limb ischaemia which may have contributed to their death. One ECPR patient died from oesophageal trauma and massive haemorrhage related to TOE probe insertion, a major BARC complication. Vascular and access-related complications, and bleeding complications, are further detailed in Supplementary Table 3. There was no difference in the median ICU or hospital length of stay. When the emergency group was adjusted to exclude three patients who underwent non-salvage ECPR (i.e. ECPR prior to PCI), mortality (p = 0.02) and AKI (p = 0.03) remained significantly higher compared to the elective group (Supplementary Table 4). Additionally, the duration of ventilation in the emergency group became significantly longer (p = 0.04), and the significantly higher rate of VARC-3 complications became non-significant (p = 0.27), compared to the elective group.

4. Discussion

4.1. Key findings and complications

Patients undergoing HR-PCI are at risk of suffering significant haemometabolic insult due to impaired perfusion, which may be protected against with MCS.[32], [33], [34] Evidence to date in the use of prophylactic VA-ECMO supported HR-PCI has been limited.^12,13,35,36 We report our experience of ECMO supported HR-PCI in a small cohort of seventeen predominantly male patients from an established Australian ECMO service and found elective VA-ECMO supported HR-PCI to be feasible and safe, with an acceptably low incidence of complications.

In our study, successful coronary revascularisation was achieved in 80 % of the emergency VA-ECMO group and 100 % of the elective VA-ECMO group, with both groups having similar coronary lesion characteristics. Importantly, 86 % of the patients in the elective group were able to be decannulated at the end of the procedure. Emergently cannulated patients however bore the burden of MACCE (in-hospital mortality) despite rapid ECMO implantation and commencement of flow.³⁷ They suffered greater rates of renal and vascular complications including two complications that likely contributed to the patients’ deaths. These vascular complications, in part, may be related to the larger cannulae implanted. This represents the complexities and challenges of initiating emergent VA-ECMO support, in keeping with existing ECPR[38], [39], [40] and broader VA-ECMO literature.[41], [42], [43] Interestingly, three patients had salvage ECPR during PCI. Whilst individual decisions cannot be scrutinized in retrospect, two of these cases may have been appropriate for prophylactic ECMO based on patient, clinical and angiographic factors.

Recently, the first randomised controlled trial (RCT) comparing prophylactic with rescue VA-ECMO in HR-PCI patients was published.⁵ They reported a low MACCE rate but noted a 19.4 % rate of life-threatening intra-procedural complications in the control group necessitating rescue MCS. Interestingly this was not necessarily identified by the preoperative LVEF, emphasising the critical nexus of patient/lesion and procedural factors.⁵ Our study differs notably in its inclusion of patients with recent ACS, more severe baseline characteristics with a greater proportion of patients having severe left ventricular dysfunction (64 % vs 7 %) and left main or equivalent disease (70.5 % vs 26.3 %).⁵ Duration of support in our elective cohort was lower; (4.4 hr vs 25 hr in Pan et al.),⁵ perhaps influenced by logistical factors in part, as we avoided the need for operating theatre availability to decannulate.

Similar to our study, Ungureanu et al. also reported the feasibility and safety of VA-ECMO supported PCI, with eight (88 %) patients with recent AMI or episode of acute heart failure surviving for 30 days.¹² However, they reported higher rates, than our study, of vascular and bleeding complications in the context of surgical cannulations employing a crossover arterial graft in contrast to our performance of percutaneous cannulations, a technique which is associated with lower infective and bleeding complications.⁴⁴

Our elective group experienced few vascular, bleeding, or renal complications and compares favourably to that of elective implantation of alternative MCS devices (Impella and IABP) reported in an industry sponsored RCT.⁴⁵ Of note, our elective group experienced a relatively high rate of sepsis (42.9 %), albeit with small patient numbers, even accounting for differences in event definition compared with Pan et al.⁵ Even brief intubation in our elective group may have carried infective risk, with data suggesting risk of ventilator associated pneumonia is highest in the early period.⁴⁶ Furthermore, agreement between clinical suspicion and culture proven infection can be poor,⁴⁷ with diagnosis confounded by poor interobserver agreement.⁴⁸ Notably, a high rate of pulmonary sepsis was also observed in a large retrospective Australian study with a broader spectrum of pathology in their VA-ECMO and ECPR cohorts.⁴⁹ Awake VA-ECMO initiation may help ameliorate the risk posed by ventilation and is actively being reviewed by our group, with greater use of fluoroscopy to guide cannulation.^11,50

Our experience adds to the observational literature supporting the feasibility and safety of ECMO supported HR-PCI, with favourable outcomes in those electively supported with MCS, and mortality and morbidity in the emergent setting reflective of the shock severity at implantation and high rates of cardiac arrest in this latter cohort.

The lack of a consensus definition for HR-PCI and high-quality evidence in this space dictates that local processes use collaborative MDTs and regularly review performance and outcomes. We differ slightly in describing our comprehensive MDT process for defining patient candidacy and avenues of therapy, and for our close liaison with our vascular surgical colleagues facilitating high rates of non-operative ECMO decannulation.

4.2. Strengths and limitations

Our study is limited by its nature as a retrospective study and its small cohort size. We report on short-term outcomes identified through electronic medical record review, as we did not conduct patient follow-up post discharge from our facility. The small cohort size may overstate significant results, demonstrated by the effect that group adjustment for non-salvage ECPR cases had on significant results. Differences in characteristics were not accounted for as confounders in the statistical analysis of our outcomes due to the small cohort size. Multiple comparisons may have impacted the secondary outcomes in our study, as all secondary outcomes (including adjustment for salvage ECPR only emergency patients) did not reach significance with post-hoc Bonferroni correction.

Selection bias is a key issue in the emergency cohort, and is reflected by the negative weighting conferred by parameters such as pre-ECMO cardiac arrest in the SAVE score⁵¹ and demonstrated by mortality rates of 29.4 % and 59.1 % for VA-ECMO and ECPR cohorts, respectively, in a recent binational ECMO registry (EXCEL) report.⁵² In our cohort, selection bias is represented by multiple confounding factors including patient deaths, particularly with a high number of palliative withdrawals, that are difficult to account for with small patient numbers. Rather than suggesting superiority, we seek to draw attention to the similar pattern of poorer outcomes for those requiring emergent rather than elective support. Despite these limitations, we describe a multidisciplinary consultation model, with percutaneous cannulation and predominantly non-operative decannulation to support complex HR-PCI. Improvements to local working definitions of HR-PCI, formalisation of the MDT process, and larger Australasian multicentre studies are warranted to elucidate the optimal benefit, safety and cost effectiveness of MCS modality in supporting HR-PCI.

4.3. Conclusion

Our study provides further insights into outcomes following emergent and elective VA-ECMO supported HR-PCI and to our knowledge, is the first published experience in an Australian cohort. Our findings corroborate recent observations that prophylactic VA-ECMO can facilitate complex coronary intervention.

Credit authorship contributions statement

LU, MD, EW conceived of the study and drafted the manuscript. LU was project manager. LU, MT, EW analysed the data. All authors read and edited the final manuscript and contributed to the final version.

Conflict of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.

Appendix A Supplementary data

The following is the Supplementary data to this article: