ECPR evidence – a historical journey

The first recorded attempts to resuscitate patients were as early as 1530 where Swiss physician Paracelsus used fireplace bellows to reinflate the lungs of those who had stopped breathing. Several more centuries passed before techniques more akin to modern day cardio-pulmonary resuscitation were reported.  Outside of reports on animals, it was a German surgeon, Fredrich Maass, who first successfully used external compressions on two patients in 1891; and in 1947 the first successful electrical cardioversion was reported by Claude Beck, a cardiothoracic surgeon. It wasn’t until 1960 however, when the combination of chest compressions and mouth-to-mouth breathing was advocated¹. This is shown in Figure 1. Only six years later, the use of cardio-pulmonary bypass in cardiac arrest was first suggested as a potential management strategy, where 7/8 refractory cardiac arrest cases were “successfully resuscitated”².

The Extracorporeal Life Support Organisation (ELSO) database is the largest registry of all Extracorporeal Membrane Oxygenation (ECMO) episodes. Whilst acknowledging not all centres who provide ECMO are registered with ELSO³, an analysis of this dataset showed that in 2003 there were < 50 ECPR runs reported in total⁴.  However, as practice has been refined and populations of patients who may benefit have been identified⁵, the use of extracorporeal CPR (ECPR) has been rapidly growing. The last ELSO Registry Report cumulatively reports over 12,000 ECPR runs with a 30% survival rate through to the end of 2021⁶.

The early evidence base for ECPR in humans was limited to case-reports and case-series. A Japanese meta-analysis that analysed studies from 1983 to 2008 found the rate of survival to discharge was approximately 27%, with a large proportion having good neurological recovery⁷.

Subsequently, the SAVE J II study⁸, was published. This involved thirty-six centres in Japan between 2013 and 2018. In this retrospective database analysis over 2300 patients received ECPR of which 1644 were included in the analysis. This showed a similar rate of survival to discharge (27%), of whom approximately 50% had a favourable neurological outcome on discharge (CPC 1 or 2). Complication rates were 37%. Approximately two thirds of these were procedure related (unsuccessful cannulation, cannula malposition and cannulation related bleeding) as opposed to ECMO related (bleeding or ischaemia).

Multiple meta-analyses^9,10 found that ECPR, when compared to conventional CPR (CCPR), was associated with improved 30-day mortality and neurological outcomes. However, there was much heterogeneity between studies making direct comparisons challenging¹¹. In particular there were wide ranging differences in inclusion and exclusion criteria¹². As prognostic factors have been identified that are associated with a favourable functional outcome, these differences in inclusion and exclusion criteria may have impacted the results of these trials. Prognostic factors associated with a good neurological outcome include age, shockable rhythm, bystander CPR, pre cannulation ROSC and time to cannulation¹³.

The last few years has seen some notable RCTs with patient important outcomes published (14-16). More detailed summaries are available at “The Bottom Line” (ARREST, Prague OHCA, and INCEPTION) for each of the following, however some of the key findings have been summarised in the table below. A meta-analysis of these trials favoured the use of ECPR (OR 1.72 [95% CI 1.09 – 2.70])¹⁷. This included an additional randomised feasibility study to three trials in the table. There were only 15 patients in the study, in which the primary outcome was to assess the proportion of patients arriving in ED within 30 minutes of 911 call, and arrival in ED to ECPR flow of < 30 minutes. Only 5 patients received ECPR of which none survived¹⁸.

There are a lot of other factors that will influence outcomes of ECPR trials outside the provision of ECMO itself. Only a small number of cardiac arrest patients are suitable for ECPR, and therefore it is imperative that the chain of survival is conducted in an efficient manner as these interventions are key in maximising the survival for all patients¹⁹. Other logistical factors that will influence outcomes of ECPR trials outside the provision of ECMO itself include time to ambulance arrival and transport times to hospital. These all need to be considered when evaluating the results of ECPR trials.

Even with the advent of large RCTs in this area, there are important questions addressing the provision of ECPR that still need to be answered.

Optimal timing

The rationale for ECPR is to restore adequate organ perfusion and gas exchange. This then enables the safe provision of any interventions necessary to restore intrinsic circulation. Defining the optimal time is a challenge – too early and patients may not have required ECPR but are still exposed to the risks and too late and any benefit may be negated. The ASAIO guidelines recommend preparing to cannulate after 20 minutes of failed resuscitation, with the goal to establish VA ECMO within 60 minutes of cardiac arrest²⁰. This, however, is not supported by high quality randomised controlled trials.

Pre-hospital initiation vs transport to ECMO centre

The optimal location of where ECMO should be initiated is yet to be determined. Unsurprisingly, the time to initiation is important, with longer times being associated with worse outcomes. Initiation in ED (as compared to the Cath Lab) shortens on-pump time and has an association with increased survival to discharge²¹. The use of pre-hospital ECPR will shorten this time further, but at the expense of logistical and technical challenges alongside increased cost.  Additionally, pre-hospital ECPR may negate the need to transport patients with CPR ongoing. Observational data has shown an association with poorer survival to hospital discharge with those transported with CPR compared to resuscitation on site²².  

It is likely that the individual geography of each metropolitan area will mean this is a question that will need to be assessed on a local level. In Australia, the CHEER3 pilot trial²³ has recently been published. This showed that pre-hospital ECMO in the Melbourne metropolitan area is feasible . Ten cases were analysed with a median time of EMS call to ECMO blood flow > 3L being 50 minutes. 40% survived, all with good neurological outcomes (a CPC of 1 at 6 months). In Sydney, modelling has shown that the use of a single pre-hospital team would add a significant amount of geographical coverage when considering the locations of current ECPR capable hospitals and a time limit of 60 minutes to implement support²⁴. Outside Australia, a survival rate of just under 30% with the initiation of pre-hospital ECMO was shown in Paris²⁵.

Design of ECPR trials

Cross-over is an important factor to consider in trial design. This has not been a significant issue in the ECPR trials so far (~2% in INCEPTION and ~8% in the Prague Study received ECPR in the standard group), but it has been an issue in other ECMO trials, such as EOLIA²⁶ where 28% crossed over from conventional treatment to VV-ECMO.

Ethical Considerations

Justice is an important ethical consideration both with respect to cost, and resource use. Few studies have evaluated the cost-effectiveness of ECPR²⁷. With respect to ECPR patients the mean cost of ECPR per patient was €50,535. However where the optimal cost-benefit ratio lies is unknown and also whether these costs should include start-up costs at de-novo ECMO centres²⁸.