Do the risks of extracorporeal membrane oxygenation in cardiogenic shock outweigh the potential benefits?

Cardiogenic shock (CS) remains a highly morbid condition with in-hospital mortality rates around 50%.^1,2 Despite advancements in medical technology, the increasing use of temporary mechanical circulatory support (tMCS), and the implementation of CS systems of care, mortality rates have not improved significantly in recent decades.² Large randomized controlled trials of tMCS in CS have not shown improved survival with tMCS, but have shown an association with a higher risk of complications.^1,3,4 Despite these disappointing results, it is possible that select patient populations may still benefit from tMCS, but identifying and enrolling these patients into clinical trials is challenging, given the haemodynamic and phenotypic heterogeneity of CS and ethical barriers to enrolling patients in randomized trials involving potentially time-sensitive life-saving therapies.

Veno-arterial extracorporeal membrane oxygenation (VA-ECMO) provides more robust circulatory support than other forms of tMCS and, therefore, offers promise in improving the care of patients with CS. This support, however, requires larger cannula sizes and is associated with an increased risk of complications. The Extracorporeal Life Support Organization recommends consideration of VA-ECMO when the rate of mortality from heart failure, despite conventional therapies, exceeds 50%.⁵ Since the mortality for most patients with CS is in this range, VA-ECMO is a reasonable therapeutic consideration in most patients with CS. Unfortunately, large-scale randomized controlled trials assessing the early use of VA-ECMO in CS and routine use of VA-ECMO in acute myocardial infarction (MI)-related CS have failed to show a benefit.^1,4 However, the negative findings in these trials should be interpreted in the context of the clinical trial population studied, which may not address the risk/benefit relationship in all patient subgroups. Notably, in the trial of VA-ECMO for MI-related CS, ECLS-SHOCK, 50% of patients enrolled were society of cardiovascular angiography & intervention (SCAI) shock Stage C¹; the mortality rate of patients with SCAI shock Stage C is <50%, and in some studies, it is as low as 12%.^6,7 As a result, the negative findings in this trial may be related to enrolling a large number of patients with SCAI shock Stage C instead of enrolling patients with a higher SCAI shock stage.

Despite these negative randomized controlled trials of VA-ECMO for CS, most experts believe that VA-ECMO may still be potentially life-saving in appropriately selected higher-risk patients with advanced SCAI shock Stage D or early-Stage E CS. Importantly, identifying the CS population most likely to benefit from VA-ECMO requires an understanding of the risks of VA-ECMO and identifying strategies to mitigate these risks to optimize outcomes.

In this issue of the European Heart Journal – Acute Cardiovascular Care, Beer et al. quantify VA-ECMO complications and assess the associations of complications with survival and favourable neurological outcomes.⁸ The authors identified 501 unique patients using a cohort of CS patients supported with VA-ECMO from 16 tertiary care centres in 4 countries.⁹ They reported a 30-day survival rate of 40%, and the authors report a sobering device–related complication rate of 52%. The most common adverse events in the overall population were renal failure requiring renal replacement therapy (53%), bleeding (33%), sepsis (25%), and vascular complications requiring an intervention (18%). Notably, 22% of patients accounted for 50% of all adverse events. The authors note that certain risk calculators, such as the survival after VA-ECMO (SAVE) score and the Simplified Acute Physiology Score (SAPS II), may help identify the patients who are more likely to experience multiple adverse events.^10,11

All adverse events were associated with a worse prognosis. Bleeding reduced the chance of 30-day survival by 45%, neurologic events reduced survival by 62%, and specific device-related complications reduced survival by 38%. The authors also explored baseline predictors of adverse neurological events, which included a lower pH, higher lactate, higher neuron-specific enolase (or an increase in neuron-specific enolase), the presence of extracorporeal cardiopulmonary resuscitation, a lower SAVE score, and a higher SAPS-II score. The only variable independently associated with device-related complications was female gender.

The authors should be lauded for their undertaking to better define the incidence and outcomes associated with VA-ECMO complications in a real-world population. While other studies have assessed complications of VA-ECMO in CS, these studies are often in the context of a randomized controlled trial or lack the granularity of this study.^12–14 The strengths of this study also include its multi-centre and multi-national design, a modestly large VA-ECMO population, and a dataset with granular demographics, haemodynamic values, and laboratory data. This study provides a window into the frequency of complications and neurologic outcomes for a high-risk group of patients with CS supported with VA-ECMO.

While this study is overall well done, it is disappointing that the ability to predict which patients are most likely to suffer device-related complications (outside of neurologic events) is poor. There are a number of important limitations to the present study. First, the rate of adverse events and device-related complications was not adjusted for shock severity in the present analysis. With the emergence of standardized classification tools, such as the SCAI shock classification, it would be important to understand whether complications vary by shock stage.^6,15 Second, it is possible that device-related complications may be much higher than those reported in this study, as many non-device-related adverse events are likely exacerbated by VA-ECMO, yet these are not coded as device-related events in the present study. Third, there are no standardized cannulation criteria, and care processes including withdrawal of care were not standardized. Fourth, the authors do not discuss what proportion of patients were bridged to recovery vs. progressed to a durable MCS device or transplantation. Finally, the predictors found in the current study have not yet been validated with an external cohort.

This manuscript does, however, highlight a number of key principles, including that baseline patient variables are associated with both outcomes and complications, and most survivors of VA-ECMO have a good neurologic outcome. The current challenge with VA-ECMO cannulation decisions is that post-cannulation outcomes are influenced by both pre-cannulation clinical variables and post-cannulation complications and clinical care. Our inability to accurately predict post-cannulation complications, as highlighted in this manuscript, suggests that we need dynamic risk predictor models that can be updated to incorporate post-cannulation variables such as changes in haemodynamics and support, laboratory variables, candidacy for bridging or destination therapies, and complications. Given the strong association between adverse events and mortality, identifying quantitative individualized risks for post-cannulation complications would be of tremendous value. These variables could also help identify an enriched population at high risk for device-related complications that could be enrolled in randomized controlled trials evaluating risk mitigation strategies and could be used as a quality benchmark.¹⁶ Future studies should seek to validate and build upon these findings by further identifying specific markers of patients most likely to experience complications.

The authors should be congratulated on their valuable contribution to the field of CS and VA-ECMO by starting to identify factors associated with complications, reaffirming that complications are associated with worse prognosis, and that most survivors of CS supported with VA-ECMO have good neurologic outcomes. This manuscript reinforces the myriad of complications associated with VA-ECMO and that these complications have a significant impact on survival. Because of these complications, the risks of the routine use of VA-ECMO do not outweigh the benefits in all comers with CS. However, it is likely that in carefully selected groups of patients with CS, the benefits of VA-ECMO do, in fact, outweigh the risks. Understanding the groups of patients most likely to have complications will help to further define the population of patients with CS that will have the greatest benefit from VA-ECMO.

Funding

None declared.

Data availability

No new data were generated or analysed in support of this research.

References

Author notes