Search Strategy

Two authors (J-CL and CM) independently searched for eligible studies in the PubMed, Embase, Cochrane Library database, China National Knowledge Infrastructure, and Wanfang Database before Feb 30, 2021, which was the last search. We limited our language to English and Chinese. Details in the literature search terms were summarized in Additional File 2. The search strategy was restricted to RCTs and observational studies with matched groups (cohort studies with two-arms or case-control studies). We also evaluated the reference lists of relevant studies to ensure the inclusion of all potential studies.

Study Selection

Studies were assessed for eligibility if they fulfilled the following criteria: (1) comparing levosimendan to control (i.e., placebo, any other drug or no drug) in patients undergoing MV/ECMO; (2) reporting data on the successful weaning rate from MV/ECMO. We excluded studies conducted in pregnant women and studies conducted in review, case reports, or case series.

Data Extraction and Outcomes

The two authors (CM and J-CL) extracted the data independently on the first author's name, study design (retrospective/prospective, RCT/cohort/case-control), year of publication, inclusion criteria, characteristics (age, male or female, and disease severity), levosimendan and control regimens as well as predefined outcomes. The primary outcome was the ECMO or MV weaning. Secondary outcomes included MV duration, length of stay in ICU, overall mortality at the longest following-up available, and adverse events. Discrepancies were identified and resolved through discussion.

Quality Assessment

CM and J-CL independently evaluated the methodological quality of the individual studies using the Cochrane risk of bias tool for RCTs (19) and the Newcastle-Ottawa Quality Assessment Scale (20) for case-control/cohort studies. We evaluated publication bias by visually inspecting funnel plots when at least 10 studies were included in this meta-analysis.

Studies Characteristics and Quality Assessment

The main characteristics of included studies and predefined outcome measures are shown in Table 1 and Additional File 3. Fourteen observational studies (9–11, 15–17, 22–26, 30–32) and four RCTs (12, 27–29) were included, which were conducted between 2009 and 2021. All but two studies (12, 28) were single-center studies. Eight (12, 16, 23, 26–29, 31) of the 18 included trials focused on MV weaning, with or without low LVEF of the recruited patients. The remaining ten studies (9–11, 15, 17, 22, 24, 25, 30, 32) focused on ECMO weaning, with five enrolling patients after cardiac surgery (11, 17, 24, 30, 32) and six enrolling patients suffering from cardiogenic shock (9, 10, 15, 22, 25, 32). Most included studies reported the detail of the levosimendan therapy regimen. As to the control group, two studies (24, 27) used milrinone while others used a placebo or no use.

We evaluated the included studies' risk of bias using the Newcastle-Ottawa Quality Assessment Scale for the 14 observational studies (9–11, 15–17, 22–26, 30–32) and the Cochrane risk-of-bias tool for the four RCTs (12, 27–29) (Additional File 4). The quality of case-control/cohort studies was moderate to high, and the risk of bias in RCTs was low in all critical domains. Assessment of publication bias using visually inspecting funnel plots showed no potential publication bias among the included studies (Additional File 5).

Primary Outcome

All 18 studies reported MV/ECMO weaning rates. The pooled analysis showed that, compared with control, levosimendan improved MV/ECMO weaning (n = 2,274; OR = 2.32; 95%CI, 1.60 to 3.36; P < 0.00001), with high heterogeneity (I² = 68%) among the studies. In the sensitivity analysis, excluding any single trial did not significantly alter the overall combined OR (P-value ranging from <0.00001 to <0.0001). Similarly, subgroup analyses confirmed the higher successful weaning rates in patients with low LVEF and MV (27–29, 31), patients with ECMO after cardiac surgery (11, 17, 24, 30), and patients with ECMO and cardiogenic shock (9, 10, 15, 22, 25) (Figure 2). However, levosimendan showed no beneficial effect on patients with MV weaning difficulty than the control group (12, 16, 23, 26) (Figure 2).

Open in a separate window

Figure 2

Forest plots of the levosimendan therapy on weaning rates from cardiopulmonary support.

Levosimendan in Weaning From ECMO

We found levosimendan facilitates the weaning from ECMO and reduces mortality rate, which is consistent with the findings of two previous meta-analyses (33, 34). Both meta-analyses reported levosimendan could improve weaning from ECMO based on five (N = 557) (34) and seven (N = 630) (33) studies, respectively. Our study added several newly published studies based on the previous meta-analyses with a large sample size of 1,336 patients, which allowed for better statistical efficacy and allowed subgroup analyses to verify our results' robustness.

VA-ECMO is increasingly being used in the short-term management of refractory circulatory failure. The main indications are myocarditis, cardiac arrest, refractory cardiogenic shock, and post-cardiotomy cardiac failure in high-risk patients with reduced LVEF (35, 36). However, this patient population still has high weaning failures. Our results show that ~40% of patients cannot successfully wean from VA-ECMO treatment under the conventional weaning process. Therefore, clinical research concerning improvement in the weaning rate of ECMO is rising (10, 15). After all, a successful weaning is a prerequisite for patient survival.

In the present study, we found that the successful weaning from VA-ECMO based on levosimendan was 80%, significantly higher than that of 60% in the control group. Some properties of levosimendan may explain its benefit for weaning from VA-ECMO. The most important thing is the sensitization of calcium ions, the positive inotropic effect without a significant increase in oxygen consumption in the myocardium (13). Second, levosimendan is an effective vasodilator. By opening ATP-dependent potassium channels in vascular smooth muscle, levosimendan has various protective effects against ischemic myocardium (preconditioning, post-processing, anti-coma, and anti-apoptotic effects) (37). Compared with other cardiotonic drugs, the effect of levosimendan is not affected by the combined use of β-blockers, and it lacks an arrhythmia-promoting effect. In addition, the long-lasting effects of its circulating active metabolites (up to 8–9 days) allow it to allow gradual weaning and provide continuous support during the critical period after ECMO (38).

Levosimendan in Weaning From MV

About 26–42% of ICU intubated patients have difficulty weaning from MV, increasing morbidity, mortality, and healthcare costs (5, 12, 16, 29). Diaphragm dysfunction is one of the critical factors contributing to weaning failure in such a patient population (7). Currently, no explicit drugs help restore diaphragm function. Therefore, whether levosimendan can improve diaphragm function, as it does in the myocardium, has aroused widespread interest. Some published studies supported such a hypothesis (14, 39). An in vitro study (39) showed that by increasing calcium sensitivity, levosimendan could enhance the contractility of diaphragm muscle fibers in patients with or without COPD. In the RCT by Doorduin et al. (14), the authors recruited 30 healthy volunteers who underwent an inspiratory loading task and found that levosimendan significantly improved neuromechanical efficiency and contractile function (P < 0.05) than placebo.

However, the results of our meta-analysis of clinical studies did not fully confirm this hypothesis. Patients who gained benefits from levosimendan during MV weaning are still those with concurrent low LVEF (27–29, 31). For such a patient population, cardiac function is the most critical factor of weaning. During the weaning process, abrupt transfer from MV to spontaneous breathing may significantly increase left ventricular filling pressure and pulmonary artery pressure (8). Simultaneously, sympathetic excitation induces the release of catecholamines, leading to peripheral vasoconstriction and increased cardiac workload. These mechanisms can cause heart failure and weaning failure, especially in patients with previous cardiac or pulmonary comorbidities.

In contrast, pooled studies focusing on patients without low LVEF showed no benefits of levosimendan on weaning from MV (12). Some explanations might help understand this failure. Firstly, most studies enrolling patients who already met the criteria for difficult weaning from MV. Besides heart function, the reasons for weaning difficulty in the ICU setting are respiratory, psychological, and psychomotor nutritional, while these factors are often combined to complicate the weaning process (5). Secondly, the patient's disease can influence the effect of levosimendan. As shown in the leoPARD study (12), the authors enrolled patients with sepsis/septic shock without septic myocardial suppression and found levosimendan was associated with a lower weaning rate from MV (HR 0.77, 95%CI 0.60–0.97, P = 0.03). The reason may be that levosimendan can cause peripheral vasodilation, resulting in the need for more norepinephrine to maintain blood pressure and cause increased adequate arterial elasticity to increased afterload. Therefore, the mechanicals might diminish the benefit of enhanced myocardial contraction from levosimendan. Moreover, the increased incidence of side effects of levosimendan, such as rapid supraventricular arrhythmias, might also contribute to the weaning failure (12).

Additionally, we found no levosimendan benefit in the length of stay in ICU or hospital. This may be because, for critically ill patients, ICU or hospital discharge was not always determined by the condition of the patients. The hospital policy to accept or refuse critically ill patients in general wards and the availability of beds for patients requiring long-term rehabilitation therapy may affect the length of ICU or hospital stay.