In this retrospective, multicenter, international registry of patients with non-ischaemic CS, lower LVEF was associated with higher CS severity, but was not associated with an increased risk of 30-day mortality. However, there was a significant interaction between MCS use and severely reduced LVEF, indicating a lower mortality risk in patients with a severely reduced LVEF and treated with MCS vs. those not treated with MCS. These results suggest that severely reduced LVEF, such as LVEF ≤ 20%, could potentially serve as an additional parameter to consider when guiding the use of MCS devices in non-ischaemic CS.

Recent studies have indicated that nearly half of all CS cases have a non-ischaemic aetiology [9, 26]. Whilst in CS caused by acute myocardial infarction, a relevant reduction in mortality risk can be achieved by early revascularization of the culprit artery, no such risk-reducing intervention exists for non-ischaemic CS [4,5,6, 27]. MCS devices target the haemodynamic culprit of non-ischaemic CS and could be beneficial treatments, but are also associated with a high risk of complications. Better identification of patients with non-ischaemic CS who might benefit from MCS devices is desirable to optimise any benefit–risk ratio. We sought to evaluate if LVEF, which can be easily and rapidly measured by TTE even in the acute setting of CS, could help to guide the use of MCS in non-ischaemic CS.

Interestingly, in this large retrospective, multicenter, international study cohort, lower LVEF was associated with higher CS severity, but, after adjustment for factors reflecting CS severity, not with 30-day mortality risk. This is in contrast to studies on CS caused by acute myocardial infarction. In a sub-study of the SHould we emergently revascularize Occluded Coronaries for cardiogenic shocK (SHOCK) trial, LVEF < 28% was an independent predictor of 30-day and 1-year mortality [28]. In the CardShock study, LVEF < 40% was independently associated with increased short-term mortality in patients irrespective of CS aetiology; however, the proportion of patients with non-ischaemic CS in this study was low, introducing a high risk of bias [29]. Several observations might explain this discrepancy. The difference between the results of the present study and the various AMI-CS studies could be due to less rigorous adjustment for shock severity in previous studies as compared to this study [28, 29]. In addition, the different CS subtypes (AMI-CS vs. non-ischaemic CS) differ in their pathogenesis, comorbidity burden and most importantly treatment. Revascularization of the culprit artery in patients with AMI-CS can improve not only the survival but also LV function as a surrogate for therapeutic success. Persistently severely reduced LVEF in AMI-CS patients may therefore indicate treatment failure or suboptimal revascularization outcome. On the other hand, non-ischaemic CS is a heterogeneous condition with various subtypes and underlying pathophysiologies (e.g. acute-on-chronic HF-related CS vs. de novo HF-related CS). In some patients, preexisting LVEF depression is common, and recovery may be limited to the pre-existing LVEF after treatment. Therefore, changes in LVEF may not necessarily correlate with treatment success and improved clinical outcomes in patients with non-ischaemic CS. It is also noticeable that previous AMI-CS studies frequently examined patients with an LVEF > 30%. In contrast, the median LVEF amongst patients with non-ischaemic CS included for this analysis was lower (20%). Lastly, our findings might extend those of a previous study assessing TTE for risk prediction in the cardiac intensive care unit. In this study, measures of LV function were more useful for mortality risk stratification in patients with lower SCAI shock stages (A to C), as compared to patients with higher SCAI shock stages as observed on our cohort [20].

Non-ischaemic CS can be caused by a variety of triggers which either cause or aggravate a pre-existing ventricular dysfunction [9]. Most prior RCTs have excluded patients with non-ischaemic CS. Therefore there is currently no evidence-based therapy for non-ischaemic CS [12, 13, 30]. Catecholamines are frequently used to support cardiac function, but their effects are limited, and they may be associated with worse outcome. A recent study reported that increasing vasopressor requirements in patients with CS was independently associated with mortality risk [31]. Furthermore, a RCT comparing dopamine and norepinephrine in the management of CS showed no significant differences in mortality risk between the two study arms [32]. This study also demonstrated an increased risk of arrhythmias with dopamine use [32]. Even when comparing milrinone and dobutamine in treatment of CS, no significant advantage of milrinone over dobutamine in terms of efficacy and safety was found [41]. Despite an absence of compelling evidence for their use, and some association with harm, the use of inotropes including short-acting catecholamines continues to form part of international guidelines as a bridge to MCS in unstable or deteriorating patients [1, 33, 34]. Aside from catecholamines, MCS may be used for the treatment of non-ischaemic CS, but the evidence supporting this is yet scarce [1]. Also, MCS use is associated with a high risk of complications, so that selecting the right patients for this approach is crucial to optimise the benefit-risk-ratio. [3, 7, 15,16,17,18,19, 35]

In this study, LVEF was not associated with 30-day mortality risk, but a significant interaction was observed between MCS use and lower LVEF, indicating a lower mortality risk in patients with a LVEF ≤ 20% treated with MCS. This could imply that use of MCS offers a net-benefit (e.g. expected benefit higher than risk of complications) in patients with a severely reduced LVEF. Two factors might contribute to explain this finding: First, in a severely reduced LVEF, the cardiac component is most likely the main factor driving outcome. Hence, MCS, which specifically addresses this issue by providing cardiac output support until native heart recovery or durable replacement therapy, is relatively more effective. Second, the more efficient and effective MCS is in supporting organ perfusion and bridging to cardiac recovery (and MCS explant) the less relevant any impact of MCS complications are likely to become. Thus, in this subpopulation of patients with non-ischaemic CS, the potential benefits of MCS usage may outweigh the associated risks of complications. These hypotheses warrant more in-depth exploration in the future research endeavours within this field.

These findings may inform the clinical decision on when to use MCS in patients presenting with non-ischaemic CS, especially when embedded within the case-based discussions of a well-organised “Shock Team” [36, 37]. Aside from potentially guiding the use of MCS in clinical practice, these results yield a rationale for using severely reduced LVEF as an inclusion criterion for randomised MCS trials. In the current RCTs testing MCS devices, LVEF is only used in the DanGer trial as an inclusion criterion [12], but not in others [13, 30, 38]. Future trials may opt to also include an enrolment criterion which reflects LV function.

The main strengths of this study are the use of a large, contemporary, international, multi-centre registry dedicated to the enrolment of patients with non-ischaemic CS. The main limitation of this study is the non-randomised retrospective design, so that a causal relation between risk predictors and outcome cannot be concluded.

Furthermore, the assessment of LVEF is susceptible to examiner- and centre-dependent subjectivity, especially when evaluated under clinically challenging conditions in intensive care units, emergency departments or catheterization laboratories with patients in a supine position. Whilst a centre-specific adjustment was conducted as a sensitivity analysis, revealing once more that there was no significant association between LVEF and 30-day mortality, but again a significant interaction between MCS use and LVEF ≤ 20%, it is crucial to underscore the following: this adjustment can reduce inter-observer variability, but it does not replace the valuable comparison of additional TTE measurements and invasively obtained haemodynamic data. It has been demonstrated that parameters, such as Cardiac Power Index, Cardiac Output, Cardiac Index and Stroke Volume, could potentially hold prognostic relevance in the context of cardiogenic shock [39]. Likewise, evaluating the prediction of afterload-related cardiac performance would be of interest [40]. However, these parameters are not adequately represented in this registry and should be the focus of future research endeavours dedicated to LVEF in non-ischaemic cardiogenic shock. Additionally, the lack of relevant data on co-existing valvular diseases, which could influence both baseline LVEF and subsequent treatment decisions, was not adequately represented in this registry.

Although the data were generated from different hospitals in different countries, all hospitals are large tertiary care centres with experience in treating patients with CS and with using MCS devices and all centres operate as part of a hub-and-spoke model. This may result in a higher prevalence of more severe CS and also higher use of MCS per se. In addition, the use of MCS in practice is a selective process in which patients with a higher physiological reserve are more likely to be treated with MCS devices, resulting in selection bias which might have influenced our results. Therefore, the generalizability of these data may be limited.