Aims: Cardiogenic shock (CS) remains a condition with high mortality. Mechanical circulatory support devices (MCS) are clinically used as a ‘bridge to recovery’. MCS provides cardiac output and oxygenation, but can also lead to complications. This study analysed a large routine care data set to quantify MCS complications and their impact on outcomes.  

Methods: Patients with CS unselected for its aetiology treated with a veno-arterial extracorporeal membrane oxygenation device (VA-ECMO) with and without active left ventricular unloading (ECMELLA) between 2005 and 2019 were retrospectively enrolled from 16 centres in 4 countries. The following adverse events (AEs) were considered in this analysis: neurological ones (hypoxic brain damage or intracerebral bleeding or haemorrhagic stroke or ischaemic stroke), bleeding events (intracerebral bleeding or haemorrhagic stroke or life-threatening bleeding or surgical or radiographic intervention due to bleeding) and ischaemic events (peripheral vascular complication requiring an intervention or mesenteric ischaemia with the need of laparotomy or ischaemic stroke). Of these AEs, typical MCS complications (contrary to those based solely on CS) were identified and analysed separately as device-related complications (intracerebral bleeding or haemorrhagic stroke or ischaemic stroke or peripheral vascular complication requiring an intervention or mesenteric ischaemia with the need of laparotomy or life-threatening bleeding or surgical or radiographic intervention due to bleeding). Aside from their frequency, the association with survival with good neurological outcome (CPC 1-2) and overall survival was determined by fitting adjusted logistic regression models.

Results: N = 501 patients were included in this study. 118 were female (24 %), mean age was 56.0 years (1^st quartile 48.7, 3^rd 65.0) years, mean lactate 8.1 mmol/l (1^st quartile 4.0, 3^rd 12.9). Acute myocardial infarction was the cause of CS in 289 patients (58 %), 301 patients (60 %) presented after or in ongoing cardiac arrest. Neurological AEs occurred in 108/469 (23 %) of patients, bleedings in 192/480 (40 %), ischaemic AEs in 123/478 (26 %) and 252/486 (52 %) of all patients experienced at least one device-related complication (specific AEs/complications and their frequencies: Table 1). Of note, AEs were not equally distributed among patients. 22 % of all patients did not suffer from any AE, whereas the upper 22 % of patients accounted for 50% all AEs (Figure 1A). Importantly, all types of AEs were independently associated with a worse prognosis, e.g. the likelihood of survival with a good neurologic outcome as well as overall survival was ~30% lower in patients with at least one AE, irrespective of its type (Figure 1B). Parameters such as baseline lactate, pH, neuron-specific enolase as well as increase in neuron-specific enolase allowed to identify patients at risk of neurologic AEs, but no other single parameter allowed to reliably identify patients at risk of ischaemic, bleeding or device-related AEs/complications (data not shown).

Conclusion:  Adverse events and device-related complications occur frequently in patients treated with MCS, seem to accumulate in some patients and are associated with a worse prognosis. Aside from neurologic events, identification of patients at risk of other complications is poor, highlighting the need to establish new (circulating bio-) markers to guide MCS treatment in CS.