Highlights
– In the randomized ECLS‑SHOCK cohort (n=417), VA‑ECMO was associated with higher rates of moderate–severe bleeding and vascular complications requiring intervention compared with medical therapy alone.
– Despite the higher complication rate, only four patients died primarily because of bleeding, and causal mediation analysis showed no significant mediation effect of these ECMO‑related complications on 30‑day mortality.
– The findings suggest that higher procedural morbidity alone does not explain the lack of mortality benefit of routine VA‑ECMO in infarct‑related cardiogenic shock; patient selection, timing, and noncomplication pathways likely drive outcomes.
Background: the clinical problem and evidence gap
Cardiogenic shock complicating acute myocardial infarction (AMI‑CS) remains a leading cause of early mortality in patients presenting with acute coronary syndromes. Mechanical circulatory support (MCS) devices, including venoarterial extracorporeal membrane oxygenation (VA‑ECMO), can provide immediate hemodynamic and respiratory support and are increasingly used in refractory shock. Observational series have reported survival with VA‑ECMO in selected patients, but the device carries substantial risks—most notably bleeding, limb ischemia from large arterial cannulae, and infection.
Recently, randomized data and pooled analyses have challenged the expectation that routine early VA‑ECMO reduces mortality in AMI‑CS. One hypothesized explanation for neutral mortality effects is that ECMO‑related complications offset any hemodynamic benefit. The ECLS‑SHOCK randomized trial tested routine early VA‑ECMO versus medical therapy in AMI‑CS. The present sub‑analysis (Thevathasan et al.) uses causal mediation techniques to ask whether VA‑ECMO–related complications are causally responsible for increased death, thereby explaining the lack of net mortality benefit.
Study design
Population and setting
This is a sub‑analysis of the multicenter, international ECLS‑SHOCK randomized trial (ClinicalTrials.gov NCT03637205). The parent trial enrolled adult patients with acute myocardial infarction complicated by cardiogenic shock across 44 specialized centers. The analysis dataset included 417 randomized patients.
Interventions and comparator
Patients were randomized to routine VA‑ECMO support plus standard care or to standard medical therapy alone (which could include inotropes, vasopressors, and revascularization per local practice). Cannulation strategies and anticoagulation were performed according to local protocols.
Definitions, endpoints and analytic approach
Investigators predefined a composite of possible VA‑ECMO–related complications as moderate or severe bleeding and peripheral vascular complications requiring intervention. The principal outcome in this mediation analysis was 30‑day all‑cause mortality. Timing and causes of death were adjudicated. The analysis used causal mediation techniques to estimate whether the increased frequency of VA‑ECMO–related complications mediated (explained) any effect of randomization to VA‑ECMO on 30‑day mortality. Logistic regression models were also used to assess associations between complications and mortality.
Key findings
Overall complication rates and timing
Of 417 patients, 88 (21.1%) experienced at least one possible VA‑ECMO–related complication. Most complications occurred early, predominantly within the first five days after randomization—consistent with a causal relationship to device use and cannulation procedures.
Complication frequencies differed significantly by treatment arm:
- Moderate to severe bleeding: 49/209 (23.4%) in the VA‑ECMO group vs 20/208 (9.6%) in control (p < 0.001).
- Peripheral vascular complications requiring intervention: 23/209 (11.0%) vs 8/208 (3.8%) (p = 0.008).
These differences confirm that VA‑ECMO increases early procedural morbidity—an expected and clinically important observation.
Causes and timing of death
Despite the higher complication burden in the ECMO arm, only four patients (across the whole cohort) had bleeding recorded as the primary cause of death within 30 days. The majority of deaths were attributed to refractory shock, multiorgan failure, or irreversible neurological injury—events frequently upstream of or independent from the bleeding and vascular complications captured as possible ECMO‑related events.
Mediation and regression analyses
Causal mediation analysis did not identify a significant mediation effect of the composite of possible VA‑ECMO‑related complications on 30‑day mortality. In other words, although ECMO increased the incidence of bleeding and vascular complications, these events did not statistically explain the absence of a survival benefit from routine VA‑ECMO in AMI‑CS.
Complementary logistic regression models evaluating the association between the occurrence of possible VA‑ECMO‑related complications and 30‑day mortality also failed to show a significant relationship after adjustment, further supporting the mediation results.
Interpretation and clinical implications
These analyses provide two related but distinct messages. First, VA‑ECMO is not benign: use was associated with a clinically meaningful increase in early bleeding and vascular complications that often require intervention and can prolong ICU stays, transfusion needs, and resource utilization. Clinicians should weigh these morbidities when considering device deployment and optimize cannulation technique, anticoagulation management, and vascular surveillance.
Second, and critically, the higher rate of these complications does not appear to be the main reason why routine VA‑ECMO failed to reduce 30‑day mortality in ECLS‑SHOCK. Most 30‑day deaths were not directly caused by bleeding or limb ischemia but by progressive shock and end‑organ failure. This suggests that factors other than procedure‑related morbidity—such as heterogeneity in patient severity, timing of support, incomplete myocardial salvage, or other downstream pathophysiology—likely determine survival.
Operationally, these findings argue against the idea that simply reducing ECMO complications alone would convert a neutral mortality effect into a clear benefit. Instead, nuanced strategies are required: better patient selection to identify those in whom temporary full‑flow circulatory support will alter the disease trajectory; earlier or tailored timing of support; integration of adjunct MCS (e.g., left ventricular unloading) where indicated; and standardized anticoagulation and cannulation protocols to minimize harm.
Mechanistic considerations
VA‑ECMO provides systemic flow and oxygenation but increases left ventricular afterload, which may worsen myocardial wall stress and impede infarct recovery unless decompression strategies are used. If VA‑ECMO is deployed without addressing ventricular unloading or persistent ischemia, the physiological benefits may be blunted. Additionally, the inflammatory and coagulopathic milieu of AMI‑CS predisposes to bleeding and thrombosis; ECMO circuits and systemic anticoagulation further modulate this balance.
Therefore, the absence of mediation by procedural complications implies that the pathway linking VA‑ECMO to outcomes is complex and may involve hemodynamic, myocardial, and systemic biological effects beyond overt bleeding or limb ischemia.
Expert commentary and contextualization
From a clinical standpoint, the work by Thevathasan et al. provides clarity: device‑related morbidity is real and must be mitigated, but simply blaming these complications for the neutral mortality signal is overly simplistic. The findings align with increasing recognition that MCS in AMI‑CS demands individualized deployment and integration into comprehensive shock care systems (rapid revascularization, appropriate use of inotropes, and organ support).
Limitations worth noting include the prespecified definition of “possible VA‑ECMO‑related complications,” which captured bleeding and peripheral vascular events but may not have fully encompassed other ECMO‑linked harms (e.g., hemolysis, infection, cerebral hemorrhage) or subclinical contributors to organ dysfunction. Additionally, mediation analysis—while powerful—relies on assumptions (no unmeasured confounding between mediator and outcome) that are difficult to verify fully even in randomized datasets. Residual confounding or misclassification of causes of death could influence estimates.
Generalisability depends on participating centers’ expertise and cannulation practices; high‑volume ECMO centers may have different complication rates and outcomes. Importantly, the analysis pertains to routine early VA‑ECMO for AMI‑CS and does not address ECMO used selectively (e.g., refractory arrest, bridge to transplant) or combined strategies (e.g., ECMO plus venting devices).
Practical takeaways for clinicians
- Recognize that VA‑ECMO increases moderate–severe bleeding and vascular complications; implement best practices for cannulation, decannulation, vascular surveillance, and anticoagulation.
- Do not assume that reducing procedural complications alone will necessarily improve short‑term mortality; focus should also include optimal patient selection, timing of support, and strategies to mitigate LV afterload.
- Adopt multidisciplinary shock team approaches to standardize selection criteria and integrate ECMO within a pathway that prioritizes rapid revascularization and organ protection.
Conclusion
The ECLS‑SHOCK mediation analysis demonstrates that VA‑ECMO increases bleeding and peripheral vascular complications but that these events do not statistically mediate 30‑day mortality in AMI‑CS. This indicates that the neutral mortality effect of routine VA‑ECMO is not simply explained by higher procedural morbidity. Future progress requires refining indications, timing, and combined physiologic strategies for MCS rather than focusing solely on complication reduction.
Funding and trial registration
Trial registration: ClinicalTrials.gov NCT03637205 (ECLS‑SHOCK). Funding sources are reported in the parent trial publication.
References
1. Thevathasan T, Alyahoud M, Freund A, et al. Extracorporeal Membrane Oxygenation Complications Are Not Causal for Mortality in Patients With Infarct‑Related Cardiogenic Shock: A Mediation Analysis of the Extracorporeal Life Support in Infarct‑Related Cardiogenic Shock (ECLS‑SHOCK) Trial. Crit Care Med. 2025 Dec 1;53(12):e2607‑e2617. doi:10.1097/CCM.0000000000006893.
2. ClinicalTrials.gov. Extracorporeal Life Support in Infarct‑Related Cardiogenic Shock (ECLS‑SHOCK). Identifier: NCT03637205. https://clinicaltrials.gov/study/NCT03637205. Accessed 2025.
3. Imai K, Keele L, Tingley D. A general approach to causal mediation analysis. Psychological Methods. 2010;15(4):309‑334. doi:10.1037/a0020761.
Additional contemporary randomized data and systematic reviews on ECMO in cardiogenic shock should be consulted for comprehensive clinical decision making.
